Open Access

32nd Annual Meeting and Pre-Conference Programs of the Society for Immunotherapy of Cancer (SITC 2017): Part One

National Harbor, MD, USA. 8-12 November 2017
Journal for ImmunoTherapy of Cancer20175(Suppl 2):86

https://doi.org/10.1186/s40425-017-0289-3

Published: 7 November 2017

Oral presentations

O1 Identification of unique neoantigen qualities in long-term pancreatic cancer survivors

Vinod Balachandran1, Marta Luksza2, Julia N. Zhao1, Vladimir Makarov1, John Alec Moral1, Romain Remark3, Brian Herbst1, Gokce Askan1, Umeshkumar Bhanot1, Yasin Senbabaoglu1, Danny Wells4, Charles Ian Ormsby Cary4, Olivera Grbovic-Huezo1, Marc Attiyeh1, Benjamin Medina1, Jennifer Zhang1, Jennifer Loo1, Joseph Saglimbeni1, Mohsen Abu-Akeel1, Roberta Zappasodi1, Nadeem Riaz1, Martin Smoragiewicz5, Olca Basturk1, Mithat Gönen1, Arnold J. Levine2, Peter J. Allen1, Douglas T. Fearon6, Miriam Merad3, Sacha Gnjatic3, Christine Iacobuzio-Donahue1, Jedd Wolchok1, Ronald DeMatteo7, Benjamin Greenbaum8, Taha Merghoub1, Steven Leach1

1Memorial Sloan Kettering Cancer Center, New York, NY, USA; 2Institute for Advanced Study, Princeton, NJ, USA; 3Icahn School of Medicine at Mount Sinai, New York, NY, USA; 4Parker Institute for Cancer Immunotherapy, San Francisco, CA, USA; 5University of Cambridge, Cambridge, UK; 6Cold Spring Harbor, Cold Spring Harbor, NY, USA; 7University of Pennsylvania, Philadelphia, PA, USA; 8Tisch Cancer Institute, New York, NY, USA
Correspondence: Vinod Balachandran (balachav@mskcc.org)

Background

Pancreatic adenocarcinoma (PDAC) is a lethal cancer with <7% of patients surviving past 5 years. T cell immunity has been linked to the exceptional outcome of the rare long-term survivors but the antigens are unknown.

Methods

To identify T cell antigens in long-term PDAC survivors, we assembled the largest cohorts of stage-matched short [n=68, median overall survival (OS) 0.8y] and long-term PDAC survivors (n=82, median OS 6y). To enable antigen discovery, we performed a combined analysis of whole exome sequencing, neoantigen prediction, T cell receptor (TCR) Vβ-chain sequencing, 9-color multiplexed immunohistochemistry, and tumor transcriptomic profiling. To assess differential neoantigen immunogenicity, we performed neoantigen fitness modeling integrating clonal genealogy, epitope homology, and T cell receptor affinity. To examine in vivo T cell-neoantigen reactivity, we used functional assays in a subset of very long-term PDAC survivors (n=7, median OS 10.5 years).

Results

We found that tumors of long-term survivors displayed 12-fold greater cytolytic CD3+CD8+Granzyme-B+ cells, with >94% of intratumoral T cell clones unique to tumors and not shared with adjacent normal pancreatic tissue, suggesting intratumoral antigen recognition. Tumors of long-term survivors exhibited greater TCR repertoire diversity compared to tumors of short-term survivors, implying differential antigenic targets. In examining neoantigens as potential T cell targets, we found that patients with both the highest predicted neoantigen number and the greatest CD3+CD8+ infiltrates together but not either parameter alone, exhibited the longest survival (median OS not reached vs. 0.8y, P=0.004). In investigating possible unique neoantigen qualities in long-term survivors, a neoantigen quality model conferring greater immunogenicity to dominant neoantigens with homology to microbial epitopes identified long-term survivors independent of confounding variables and chemotherapy (Neoantigen QualityHi vs. Low - median OS 8.6y vs. 0.8y; P=0.002) whereas a model ascribing greater immunogenicity to increasing neoantigen number did not (Neoantigen QuantityHi vs. Low - median OS 0.8y vs. 1y; P=0.3) (Fig. 1). Similarly, neoantigen quality, but not quantity, was independently prognostic of survival in a larger, independent cohort unselected by survival (n =166; Neoantigen QualityHi vs. Low - median OS 30m vs. 14m; P<0.0001) (Fig. 1). Finally, we detected lasting circulating T cell reactivity to both high quality neoantigens and cross-reactive antigens in PDAC survivors, including identical intratumoral clones with specificity to both (n=5 of 7 patients tested).

Conclusions

Our results identify neoantigens with unique qualities as T cell targets in PDAC. More broadly, we identify neoantigen quality as a biomarker for immunogenic tumors that may facilitate rational application of immunotherapies.
Fig. 1 (abstract O1).

Neoantigen quality is prognostic of pancreatic cancer survival

O2 Predictive biomarkers for response to anti-CTLA-4 and anti-PD-1 immunotherapy in melanoma patients

Priyanka Subrahmanyam1, Zhiwan Dong2, Daniel Gusenleitner2, Anita Giobbie-Hurder2, Mariano Severgnini2, Jun Zhou2, Michael Manos2, Holden Maecker1, F. Stephen Hodi2

1Stanford University, Stanford, CA, USA; 2Dana-Farber Cancer Institute and Harvard Medical School, Boston, MA, USA
Correspondence: Priyanka Subrahmanyam (pri3@stanford.edu)

Background

Immune checkpoint blockade has greatly improved clinical outcomes in melanoma patients. However, there is an urgent need for predictive biomarkers to determine who will likely benefit most from which therapy. To date, most biomarkers of response have been identified in the tumors themselves. Biomarkers that could be assessed from peripheral blood would be even more desirable, because of ease of access and reproducibility of sampling.

Methods

We used mass cytometry (CyTOF), which is a technique similar to flow cytometry, but uses metal-ion tagged antibodies for high-dimensional immune profiling. We performed CyTOF on pre-treatment peripheral blood mononuclear cells from melanoma patients to identify potential biomarkers of response to anti-CTLA-4 or anti-PD-1 immunotherapy.

Results

Our studies revealed that anti-CTLA-4 and anti-PD-1 therapies have distinct sets of candidate biomarkers. Responders to anti-CTLA-4 therapy had higher memory CD4+ and CD8+ T cells compared to non-responders. Furthermore, responders showed higher frequencies of CD8+ effector memory T cells. In anti-PD-1 (but not anti-CTLA-4) treated patients, responders had higher frequencies of CD69 and MIP-1β expressing NK cells. Finally, using multivariate analysis, distinct models for the prediction of response to anti-CTLA-4 and anti-PD-1 were developed.

Conclusions

Our findings point to an important role for memory T cell subsets, as well as NK cells in response to immunotherapy, making them potential predictive biomarker candidates for response to checkpoint blockade.

O3 Combining immunophenomics with a gene expression panel for improved prostate cancer recurrence prediction

Nathalie Harder1, Harald Hessel2, Maria Athelogou1, Alexander Buchner2, Christian Stief2, Thomas Kirchner2, Günter Schmidt1, Ralf Huss1

1Definiens AG, Munich, Germany; 2Ludwig-Maximilians-University Munich, Munich, Germany
Correspondence: Nathalie Harder (nharder@definiens.com)

Background

Multi-omics data provide rich information on different levels such as genome, transcriptome, function or phenotype. Integrating such comprehensive measures offers great potential for identifying improved prognostic and predictive signatures in the context of precision medicine. Here we combine gene expression data with phenomic readouts extracted from immunohistochemically stained sections of resected prostate cancer (PCa) tissue. By integrating both data sets we identify multi-variate signatures with high prognostic value regarding prostate cancer recurrence after radical prostatectomy.

Methods

Data was collected from a cohort of 23 PCa patients (Gleason-score 6-9, pT2, age<=75y). Consecutive FFPE tissue sections were dual-stained for immune-related (CD3/CD8, CD68/CD163) and structural (CD34, CK18/p63) components. Using Tissue Phenomics, we extracted image-based features, e.g., densities, ratios, and distances of positive cells across differently stained sections after co-registration in specific regions-of-interest (tumor, TME, stroma) [1]. Gene data (NanoString Technologies, Seattle, nCounter PanCancer Immune Profiling Panel) was collected from sections containing tumor only and normalized w.r.t. housekeeping genes. We separately selected subsets of image-based and gene-based features by optimizing a cutoff regarding the uni-variate stratification performance (log-rank test, accuracy) on random data subsets using bootstrapping. Features from both sets were combined by Classification and Regression Trees (depth=2) and the best performing combinations were identified using cross validation. Clinical parameters (age, Gleason-score, PSA blood value) provided no prognostic value as shown by Cox regression analysis.

Results

Several tree models showed improved cross-validated stratification performance compared to all uni-variate stratifications using either feature set (Fig. 1). We found that (I) a low expression of C4B gene in combination with a low ratio of CD8(+) in stroma to CD3(+)CD8(-) in tumor (pcrossVal<0.05), and (II) a high expression of CD46 gene in combination with a low average distance of CD34(+) microvessels to neighboring CD168(+) M2 macrophages (pcrossVal<0.05) are strong prognostic factors for prostate cancer recurrence.

Conclusions

Our results suggest that an inhibited activity of the innate as well as the adaptive immune system provides prognostic value for prostate cancer recurrence. Both genes C4B and CD46 are related to the complement system where C4B positively correlates with complement activity while CD46 is an inhibiting complement regulator. Limited activation of CD8(+) cytotoxic T cells as well as increased activity of M2 macrophages fostering angiogenesis additionally contribute to the cancer promoting environment. Combining information from genome and phenome therefore helps gaining a broadened picture of factors promoting PCa progression.

Reference

1. Athelogou M, et al. Tumor-associated macrophages as prognostic marker for prostate cancer progression. SITC. 2016.

O4 Clinical activity of adenosine 2A receptor (A2AR) inhibitor CPI-444 is associated with tumor expression of adenosine pathway genes and tumor immune modulation

Andrew Hotson1, John Powderly2, Leisha Emens3, Patrick Forde3, Matthew Hellman4, Lawrence Fong5, Ben Markman6, Brett Hughes7, Jonathan Goldman8, Mario Sznol9, Daruka Mahadevan10, Shivaani Kummar11, Joshua Brody12, Philip Bonomi13, Jason Luke14, Matthew Riese15, Taofeek Owonikoko16, Sherene Loi17, Amy Wiese18, Robert Doebele19, James Lee20, Chunyan Gu21, Stephen Willingham21, Ginna Laport21, Richard Miller21, Ian McCaffery21

1Corvus Pharmaceuticals, Redwood City, CA, USA; 2Carolina BioOncology Institute, Huntersville, NC, USA; 3Johns Hopkins University, Baltimore, MD, USA; 4Memorial Sloan Kettering Cancer Center, New York City, NY, USA; 5University of California, San Francisco, San Francisco, CA, USA; 6Monash Medical Centre, Clayton, Australia; 7Royal Brisbane and Women’s Hospital, Herston, Australia; 8University of California, Los Angeles, Los Angeles, CA, USA; 9Yale University School of Medicine, New Haven, CT, USA; 10University of Arizona, Tucson, AZ, USA; 11Stanford University School of Medicine, Stanford, CA, USA; 12Icahn School of Medicine at Mount Sinai, New York City, NY, USA; 13Rush University Medical Center, Chicago, IL, USA; 14University of Chicago Medical Center, Chicago, IL, USA; 15Medical College of Wisconsin, Milwaukee, WI, USA; 16Emory University Hospital, Atlanta, GA, USA; 17Peter MacCallum Cancer Centre, Melbourne, Australia; 18Wayne State University, Detroit, MI, USA; 19University of Colorado Anschutz Medical Campus, Aurora, CO, USA; 20University of Pittsburgh, Pittsburgh, PA, USA; 21Corvus Pharmaceuticals, Burlingame, CA, USA
Correspondence: Ian McCaffery (imccaffery@corvuspharma.com)

Background

CPI-444, an A2AR antagonist, inhibits suppression of T cell function by adenosine and is active in multiple preclinical models. CPI-444 is being investigated as a single agent and in combination with atezolizumab in an ongoing Phase 1/1b clinical trial in patients with advanced cancers (NCT02655822). Biomarker investigations were conducted to explore immune modulation in serial tumor biopsies and peripheral blood as well associations between adenosine pathway genes and clinical activity.

Methods

Tumor biopsies and blood samples were analyzed in 80 patients with renal cell carcinoma (RCC: n=35), non-small cell lung cancer (NSCLC: n=26), triple negative breast cancer (TNBC: n=17), microsatellite instable colorectal cancer (MSI-CRC n=2) treated with CPI-444 100mg BID as single agent (‘SA’ n=18) or in combination with atezolizumab (‘combo’ n=62). Paired tumor biopsies were analyzed: gene expression profiles (Nanostring), CD8, PD-L1 and CD73 (IHC). T cell repertoires were examined by sequencing of the T cell receptor beta chain gene in PBMCs and tumors.

Results

Of 80 evaluable patients, 54 had progressed on prior anti-PD-(L) 1 therapy (‘IO-refractory’), 26 were IO-naïve. Most patients were PD-L1 negative in archival tumor (80% TNBC, 91% RCC, 69% NSCLC).

Six objective responses were observed: 2 partial responses (PR) in RCC (1 SA and 1 combo); 2 PRs in NSCLC (both combo); 1 PR MSI-CRC (combo) and 1 PR in TNBC (combo). Notably, 3 PRs were observed in IO-refractory patients (1 RCC, 2 NSCLC) and 2 PRs were PD-L1 negative (1 RCC, 1 NSCLC).

SA and combo treatment significantly increased CD8+ cell infiltration (mean 2-fold; 95%CI 1.18-2.91), T cell activation and interferon-gamma pathway gene expression in biopsies as well as T cell clone expansion in both tumor biopsies and periphery. Shared clonotypes expanded in matched post-dose PBMCs and tumor suggesting trafficking of T cells following CPI-444 treatment. This included IO-refractory and PD-L1-negative patients.

Patients with elevated CD73 and expression of adenosine genes in baseline tumor samples experienced significant tumor regression (best % change: high-CD73 -5.6% regression; low-CD73 17.9% growth; p-value=0.03, n=40). In IO-refractory patients, CD73 expression and adenosine pathway genes were higher than IO-naïve patients at baseline suggesting adenosine pathway is a mechanism of immune escape in IO-refractory patients.

Conclusions

CPI-444 anti-tumor activity was associated with immune-modulation of T cells in tumor and periphery, including IO-refractory and PD-L1 negative patients. Adenosine pathway is upregulated in IO-refractory patients and may be associated with clinical response to CPI-444 supporting a role for CPI-444 in treatment of IO-refractory tumors.

Trial Registration

ClinicalTrials.gov Identifier NCT02655822

O5 A dendritic cell targeting NY-ESO-1 vaccine significantly augments early and durable immune responses in melanoma patients pretreated with human Flt-3 Ligand

Nina Bhardwaj1, Philip Friedlander2, Anna Pavlick3, Marc Ernstoff4, Brian Gastman5, Brent Hanks6, Mark Albertini7, Jason Luke8, Michael Yellin9, Tibor Keler9, Thomas Davis9, Laura Vitale9, Ana Blazquez10, Chihiro Morishima11, Patrick Danaher12, Sarah Warren12, Bob Salim13, Bruce Hess11, Leonard D'Amico11, Lisa Lundgren14, Nora Disis11, Martin Cheever14, Steven Fling14

1Icahn School of Medicine at Mount Sinai, New York, NY, USA; 2Mount Sinai Beth Israel, New York, NY, USA; 3NYU Lagone Medical Center, New York, NY, USA; 4Roswell Park Cancer Institute, Buffalo, NY, USA; 5Cleveland Clinic, Cleveland, OH, USA; 6Duke, Durham, NC, USA; 7University of Wisconsin, Madison, WI, USA; 8University of Chicago Medicine, Chicago, IL, USA; 9Celldex Therapeutics, Needham, MA, USA; 10Mount Sinai, New York, NY, USA; 11University of Washington, Seattle, WA, USA; 12Nanostring Technologies, Seattle, WA, USA; 13Axio Research, Seattle, WA, USA; 14Fred Hutchinson Cancer Research Center, Seattle, WA, USA
Correspondence: Steven Fling (sfling@fhcrc.org)

Background

Patients with high-risk melanoma, (AJCC TNM stage II and stage III disease), have a 20-60% recurrence rate with 5-year overall survival (OS) between 45% and 70%. The adjuvant setting is an opportunity to test prevention vaccines that may have efficacy against disease recurrence. Vaccine therapy with CDX-1401 (a fusion protein consisting of human monoclonal IgG1 antibody targeting the dendritic cell (DC) receptor DEC-205 linked to the NY-ESO-1 tumor antigen) can safely lead to humoral and cellular immunity in cancer patients with advanced malignancies that express NY-ESO-1, including melanoma. CDX-301, a recombinant human Flt3 ligand (Flt3L), safely produces increases in DC in humans and may enhance vaccine responses through increased DC number and activity. We evaluated CDX-301 and CDX-1401 combination treatment in a phase II, open-label, multicenter, randomized study of subjects with resected melanoma, to determine whether immune responses to NY-ESO-1 elicited by vaccination with CDX-1401 + poly-ICLC are substantially increased by prior expansion of circulating DC with Flt3L therapy.

Methods

60 patients with resected melanoma were randomized to two cohorts: Cohort 1 received CDX-301 (Flt3L) pretreatment (25 ug/kg SC x 10 days) in two of four monthly cycles of vaccination with CDX-1401 (1mg IC) + poly-ICLC (2mg SC, days 1 and 2). Cohort 2 received 4 monthly cycles of vaccine with CDX-1401 and poly-ICLC without prior Flt3L. We also assessed immunogenicity to other melanoma-associated antigens and memory viral responses, character of PBMC subsets, and safety, tolerability and clinical efficacy of the regimens.

Results

Both treatments were well tolerated with grade 1-2 AEs of chills, injection site erythema and pain, fever and myalgias most common. A substantial increase of between ~15- to ~200-fold of innate immune cells (DC, monocytes and NK cells) was observed in subjects treated with CDX-301. Further, there was development of higher anti-NY-ESO-1 antibody titers and NY-ESO-1 specific T cells in cohort 1 vs. cohort 2, and evidence of induction of antigen-specific CD8+ T cells in cohort 1. Comparative immune cell gene expression profiling of PBMC are consistent with these differences and reveal a gene signature associated with Flt3L induction of an early and durable immune response.

Conclusions

DC mobilization with vaccines targeting DC is safe and significantly enhances vaccine responses to widely expressed tumor associated antigens.

Trial Registration

ClinicalTrials.gov Identifier NCT02129075

O6 Re-educating macrophage through glutamine metabolism-mediated metabolic and epigenetic reprogramming

Ping-Chih Ho (ping-chih.ho@unil.ch)

University of Lausanne, Epalinges, Switzerland

Background

The different regulatory mechanisms orchestrating macrophage activation, including signaling cascades and epigenetic programming, are increasingly appreciated. Among these regulatory mechanisms, differences in the bioenergetic demands of M1 and M2 macrophages are emerging as new regulatory circuits that adjust a macrophage’s behavior in response to the nutrient state in its habitation and the infected tissues. Glutamine metabolism provides synergistic support for macrophage activation and elicitation of desirable immune responses; however, the underlying mechanisms regulated by glutamine metabolism to orchestrate macrophage activation remain unclear.

Methods

We examine metabolic and epigenetic reprogramming in bone marrow-derived macrophages under M1 and M2 skewing condition. We also examine the impact of glutamine metabolism on supporting endotoxin tolerance format in bone marrow-derive macrophages in vitro and in the in vivo septic shock model. The anti-tumor responses of glutaminase 1 targeting approach was further evaluated in MC-38 and Braf/Pten melanoma-bearing mice in conjunction with or without adoptive tumor-specific T cell transfer.

Results

Here we show that the production of αα-ketoglutarate via glutaminolysis is important for alternative (M2) activation of macrophages, including engagement of fatty acid oxidation (FAO) and histone H3K27 demethylase Jmjd3-dependent epigenetic reprogramming of M2 genes. This M2-promoting mechanism is further modulated by a high αα-ketoglutarate/succinate ratio, while a low ratio strengthens the pro-inflammatory phenotype in classically (M1) activated macrophages. As such, α-ketoglutarate contributes to endotoxin tolerance following M1 activation. This study reveals new mechanistic regulations by which glutamine metabolism tailors the immune responses of macrophages through metabolic and epigenetic reprogramming. We further provide proof-of-concept evidence that targeting glutaminolysis can tailor the immune response of tumor-associated macrophages in vivo, and co-treating glutaminase inhibitor with adoptive tumor-specific T cell transfer can effectively suppress tumor growth.

Conclusions

This study highlights the unrecognized mechanisms controlled by glutaminolysis to fine-tune macrophage activities and advances modulation of glutamine metabolism as an attractive strategy for harnessing macrophage-mediated immune responses. Most importantly, targeting glutaminase is able to re-educate tumor-associated macrophages and provides synergistic anti-tumor responses with adoptive cell transfer therapy. Altogether, our work uncovers new regulatory circuits orchestrated by glutamine metabolism in macrophages and reveals the potential of these mechanisms on harnessing macrophage immune responses.

O7 Early FDG-PET response correlates with dose and efficacy in patients with microsatellite stable mCRC treated with carcinoembryonic antigen T cell bispecific (CEA-TCB) antibody plus atezolizumab

Said Bouseida1, Federico Sandoval1, Daniel Sabanes Bove2, Vaios Karanikas3, Abiraj Keelara1, Tapan Nayak1, Jose Saro3

1Roche Innovation Center Basel, Basel, Switzerland; 2F. Hoffmann-La Roche Ltd., Basel, Switzerland; 3Roche Innovation Center Zurich, Zurich, Switzerland
Correspondence: Said Bouseida (said.bouseida@roche.com)

Background

CEA-TCB (RG7802, RO6958688) is a novel T cell bispecific antibody targeting CEA on tumor cells and CD3 on T cells. An ongoing Phase Ib study (NCT02650713) is exploring the safety, tolerability and efficacy of CEA-TCB in combination with atezolizumab (anti–PD-L1). Here, we report preliminary results of 18F-fluorodeoxyglucose-positron emission tomography (FDG-PET) imaging as an early pharmacodynamic marker for this novel cancer immunotherapy combination in patients with microsatellite stable (MSS) metastatic colorectal cancer (mCRC).

Methods

In this study, CEA-TCB was given weekly in combination with atezolizumab 1200 mg every 3 weeks in patients with CEA-expressing solid tumors. On-treatment FDG-PET scans were performed at week 4 and compared with baseline. On-treatment changes in maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV) and total lesion glycolysis (TLG) were analyzed in up to 10 measurable lesions, identified at baseline by an independent reviewer, in each patient. Exploratory statistical analyses used semiparametric Gaussian regression models and progression-free survival (PFS) was assessed using Cox proportional hazard landmark analyses.

Results

As of March 3, 2017, a total of 35 patients with MSS mCRC were treated with CEA-TCB doses ranging from 5 mg to 160 mg; 17 patients were evaluable for PET image analysis. Of the 15 PET-evaluable patients, 8 (53%) had early (week 4) on-treatment reductions in SUVmax > 15% with CEA-TCB plus atezolizumab. Early decreases in SUVmax appeared to correlate with increasing CEA-TCB doses (P = 0.0081), with greater and consistent reductions seen at ≥ 80 mg. Partial metabolic response was observed in 6 of 15 patients, with all responders treated at ≥ 80 mg. Decreases in SUVmax appeared to be associated with reduction in tumor burden (best change from baseline per RECIST v1.1; P = 0.0061). Depth of reduction in MTV appeared to be associated with longer PFS (P = 0.013). In addition, reductions in MTV and TLG appeared to correlate with decreases in soluble CEA (sCEA) at week 6 (MTV, P = 0.013; TLG, P = 0.034). Data from an expanded patient population treated with CEA-TCB as monotherapy and in combination with atezolizumab will be presented.

Conclusions

In patients with MSS mCRC, changes in MTV and SUVmax correlated with longer PFS along with tumor shrinkage, and decreases in 2 FDG parameters, MTV and TLG, correlated with a decline in soluble tumor marker, sCEA. Thus, early on-treatment changes in FDG-PET can serve as a pharmacodynamic biomarker related to treatment efficacy with CEA-TCB.

Trial Registration

NCT02650713

O8 Lactic acid as a mediator of metabolic symbiosis between regulatory T cells and the tumor microenvironment

McLane Watson1, Ryan Whetstone1, Rahul Deshpande1, Ashely Menk1, Nicole Scharping1, Brett Morrison2, Stacy Gellhaus Wendell1, Greg Delgoffe1

1University of Pittsburgh, Pittsburgh, PA, USA; 2Johns Hopkins, Baltimore, MD, USA
Correspondence: Greg Delgoffe (delgoffeg@umpc.edu)

Background

The tumor microenvironment (TME) presents a major barrier to effective immunotherapy. Metabolically harsh conditions starve infiltrating effector T cells rendering them ineffective. In contrast, regulatory T (Treg) cells, a subset of CD4+ T cells that play a vital role in preventing autoimmunity and maintaining immune homeostasis, thrive in the TME. As a result, the TME becomes more immunosuppressive. It has been shown that Treg cells have a distinct metabolic profile compared to their effector T cell counterparts. Therefore, we hypothesized that Treg cells thrive because they are metabolically supported by the TME.

Methods

B16, a mouse model of melanoma, was used. From tumor bearing mice conventional effector T cells and Treg cells were transcriptionally and metabolically profiled using flow cytometry, proliferation assays, suppression assays, and isotopic flux analysis. A mouse with a Treg specific deletion of MCT1, a predominant lactate transporter was generated. A pharmacological inhibitor of MCT1 was used to prevent lactate uptake.

Results

Transcriptional and metabolic profiling revealed Treg cells from the tumor upregulate a distinct metabolic profile utilizing the glycolytic end-product, lactic acid. Lactic acid has been known to be immunosuppressive, and indeed, it curbed the proliferation of conventional effector T cells in vitro. However, lactic acid had the opposite effect on Treg cells, promoting their proliferation. Isotopic flux analysis revealed lactate is utilized by Treg cells to generate glycolytic intermediates, in part through gluconeogenic pathways. Preventing lactate transport in Treg cells through a conditional knockout of MCT1 resulted in mice with normal immune homeostasis, but superior anti-tumor immunity when implanted with melanoma. Similarly, in wild type tumor bearing mice, preventing lactate uptake through a pharmacological inhibitor resulted in decreased intratumoral Treg proliferation and increased efficacy of checkpoint blockade immunotherapy.

Conclusions

These data suggest that tumors evade the immune system, in part, by metabolically supporting suppressive Treg cells with lactic acid. These data also suggest that targeting lactate metabolism may increase the efficacy of checkpoint blockade immunotherapy.

O9 Transcriptional approach to understanding the role of tonic signaling and co-stimulation in CAR T cells

Angela Boroughs1, Nemanja D. Marjanovic2, Ana P. Castano3, Robert Morris3, Livnat Jerby2, Felipe Bedoya3, Aviv Regev2, Marcela V. Maus3

1Massachusetts General Hospital, Harvard University, Boston, MA, USA; 2Broad Institute of Harvard and MIT, Cambridge, MA, USA; 3Massachusetts General Hospital, Boston, MA, USA
Correspondence: Angela Boroughs (angelaboroughs@fas.harvard.edu)

Background

Chimeric antigen receptors (CARs) targeting CD19 have produced impressive outcomes for the treatment of B cell malignancies. Using RNA sequencing, we determined the transcriptional changes downstream of CARs with different co-stimulation domains pre-and post-antigen stimulation.

Methods

T cells were sorted from 3 donors and transduced with one of four CAR constructs. The constructs all contained a CD19-directed scFv fused to one of four different intracellular domains: 4-1BBζ, CD28ζ, ζ and Δζ. Cells were expanded with CD3/CD28 beads for 5 days and the rested. On day 12, the T cells were stimulated through their CAR or TCR for 0, 4 or 24 hours. T cells were flow-sorted on viable lymphocytes, CAR+, CD3+ and CD4+ or CD8+ and then sequenced using the SmartSeq2 protocol. Reads were aligned and quantified. Differential gene expression calculated using DESeq2. Significant differentially expressed (DE) genes had a FDR cutoff of <0.05.

Results

T cells containing CARs have clearly distinct transcriptional signatures both from each other and from untransduced T cells. We identified a gene signature for tonic signaling of all ζ-containing CAR T cells, which was identified in transduced resting T cells prior to antigen stimulation through the CAR. These genes were DE between untransduced or Δζ vs 4-1BBζ, CD28ζ, ζ CARs. Furthermore, we identified a set of genes that is DE between 41BBζ and CD28ζ CAR T cells. Some of these genes such as ENNP2 are DE before antigen stimulation through the CAR, while other genes were significantly DE only after antigen stimulation through the CAR. There was a clear upregulation of HLA Class II genes as well as CIITA, which controls MHCII upregulation, in activated 41BBζ vs 28ζ CAR T cells. Across all time points there was an enrichment for human TH1 signature genes in the 41BBζ compared to the CD28ζ CD4+ CAR T cells (p-value<0.0005). Cytokines, IL-15 and IL-21 were both significantly upregulated in 41BBζ vs CD28ζ CARs by 24 hours after CAR stimulation.

Conclusions

We have identified a transcriptional signature that reflects tonic signaling in CAR T cells bearing the CD3ζ signaling domain, independent of the co-stimulation domain. Differences between 41BBζ and 28ζ have been previously identified several days following CAR-antigen stimulation. Here we show many of these differences exist before CAR-mediated activation, indicating a transcriptional effect of tonic signaling. We also defined a transcriptional program in 41BBζ vs CD28ζ CAR T cells that polarizes CD4+ cells to a more TH1 like state pre-and post-signaling.

010 Gamma secretase inhibition increases recognition of multiple myeloma by BCMA-specific chimeric antigen receptor modified T cells

Tyler Hill, Margot Pont, Joe Abbott, Jessica Kelliher, Alexander Salter, Steven Liu, Damian Green, Stanley Riddell

Fred Hutchinson Cancer Research Center, Seattle, WA, USA
Correspondence: Margot Pont (mpont@fredhutch.org)

Background

B cell maturation antigen (BCMA) is expressed on most cases of multiple myeloma (MM) cases and can be targeted by T cells transduced with an anti-BCMA chimeric antigen receptor (CAR). The extracellular portion of BCMA is cleaved by gamma-secretase (γ-secretase), resulting in lower surface expression of BCMA on tumor cells and the presence of soluble BCMA (sBCMA) in the circulation, both of which could hamper BCMA CAR-T cell efficacy and enable escape of BCMAlow MM cells. We investigated whether γ-secretase inhibitor (GSI) treatment inhibited BCMA shedding, increased surface BCMA levels, and augmented BCMA-specific CAR-T cell antitumor functionality in vitro and in vivo.

Methods

BCMA expression on MM cell lines and primary CD138+ MM samples was measured by flow cytometry after treatment with the GSI RO4929097. sBCMA levels in patient sera and culture supernatants were measured by ELISA. BCMA-specific CAR-T cells were co-cultured with MM cell lines or primary MM, with or without GSI. Blocking effects of sBCMA on CAR-T cells were measured by titrating recombinant BCMA/Fc into co-cultures. CAR-T cell functionality was measured by cytokine ELISA, CFSE dye dilution, and chromium-release assays. To investigate in vivo CAR-T cell efficacy, MM1.Rffluc-tumor bearing NSG mice were treated with BCMA-specific CAR-T cells alone and in combination with GSI. Tumor burden was monitored by bioluminescence imaging.

Results

BCMA-specific CARs were optimized for functionality by spacer modifications. Patient serum levels of BCMA were found to often exceed 100 ng/mL and concentrations >33 ng/mL inhibited BCMA CAR-T cell function. Treatment with GSI RO4929097 increased BCMA levels on both MM cell lines and primary MM samples in a dose dependent and reversible fashion, and decreased sBCMA shedding. High doses of RO4929097 (>3 mM) inhibited CAR-T cell function, however low doses (0.1-1 mM) were sufficient to upregulate surface BCMA levels on primary MM up to 10-fold, leading to increased recognition by BCMA CAR-T cells. In a preclinical model of myeloma, treatment with RO4929097 increased BCMA on tumor cells in bone marrow and decreased sBCMA in peripheral blood. Treatment of myeloma bearing mice with BCMA CAR-T cells combined with intermittent doses of RO4929097 improved antitumor effects and increased survival as compared to mice that did not receive RO4929097.

Conclusions

Combining GSI with BCMA-specific CAR-T cells improves tumor recognition and therapeutic benefit in a preclinical model of myeloma. The data provide a rationale for translating this combination to the clinic.

O11 DNA vaccine with PD-1 blockade elicits anti-tumor responses in patients with metastatic, castration-resistant prostate cancer (mCRPC)

Douglas McNeel, Jens Eickhoff, Robert Jeraj, Mary Jane Staab, Jane Straus, Ellen Wargowski, Matthew Scarpelli, Glenn Liu

University of Wisconsin, Madison, WI, USA
Correspondence: Douglas McNeel (dm3@medicine.wisc.edu)

Background

We have previously investigated a DNA vaccine encoding prostatic acid phosphatase (PAP, pTVG-HP) in two trials of patients with PSA-recurrent prostate cancer. Favorable changes in PSA doubling time were associated with the development of PAP-specific immunity. A randomized phase 2 trial testing this vaccine is currently underway. In preclinical models, we have found that blockade of regulatory receptors, including PD-1, at the time of T cell activation with DNA vaccination produced anti-tumor responses in vivo. We have also found that patients immunized with this DNA vaccine developed PD-1-regulated T cells. These findings suggested that combined PD-1 blockade with vaccination might elicit superior anti-tumor responses in patients with prostate cancer.

Methods

A clinical trial was designed to evaluate the immunological and clinical efficacy of pTVG-HP when delivered in combination (over 12 weeks) or in sequence (over 24 weeks) with pembrolizumab, in patients with mCRPC. Correlative studies included effects on serial biopsies, different measures of systemic immunity, and exploratory FLT PET/CT imaging to evaluate proliferative changes in tumors and vaccine-draining lymph nodes.

Results

26 patients were randomized to treatment and evaluable for response. On-treatment adverse events > grade 2 included 1 episode of fatigue, 1 episode of diarrhea, and 1 episode of autoimmune hepatitis. No grade 4 events were observed. 8/13 subjects (61%) treated in combination had any decrease in PSA from baseline, with 4/13 (31%) having a PSA decline >25%. 1/13 (7%) treated in sequence had a decrease in PSA from baseline (p=0.013). Several patients treated with the combination experienced decreases in tumor volume by radiographic imaging at 12 weeks. Expansion of peripheral PAP-specific Th1-biased T cells was detected in some patients. Exploratory FLT PET/CT imaging has demonstrated proliferative responses in metastatic lesions and in vaccine-draining lymph nodes. Preliminary evaluation of biopsy specimens has suggested recruitment of activated T cells.

Conclusions

PD-1 pathway inhibitors have demonstrated little clinical activity to date when used as single agents for treating prostate cancer. Our findings demonstrate that combining this blockade with tumor-targeted T cell activation by a DNA vaccine is safe and can augment tumor-specific T cells, detectable within the peripheral blood and by imaging, and result in objective anti-tumor changes. Based on these results, the randomized portion of this trial was closed to accrual, and an expansion arm has been opened to evaluate the safety and clinical efficacy of combination treatment beyond 12 weeks.

Trial Registration

NCT02499835

O12 Immune and tumor responses to human IL-10 (AM0010, Pegilodecakin) alone or in combination with immune checkpoint blockade

Aung Naing1, Deborah J. Wong2, Jeffrey R. Infante3, Kyriakos Papadopoulos4, Raid Aljumaily5, W. Michael Korn6, Jeffrey G. Schneider7, Manish Patel8, Karen A Autio9, Gerald S. Falckook10, Nashat Y. Gabrail11, Bianca Rojo12, Navneet Ratti12, Scott McCauley12, Annie Hung12, Peter Van Vlasselaer12, Gail L. Brown12, Edward B. Garon2, Nizar M. Tannir1, Martin Oft12

1MDACC, Houston, TX, USA; 2UCLA, Los Angeles, CA, USA; 3SCRI, Nashville, TN, USA; 4START, San Antonio, TX, USA; 5OUHSC, OKLAHOMA CITY, OK, USA; 6UCSF, San Francisco, CA, USA; 7Winthrop University Hospital, Mineola, NY, USA; 8Florida Cancer Specialists, Sarasota, FL, USA; 9MSKCC, New York, NY, USA; 10Sarah Cannon at HealthOne, Denver, CO, USA; 11Gabrail Cancer Center, Canton,, OH, USA; 12ARMO BioSciences, Redwood City, CA, USA
Correspondence: Martin Oft (martinoft@gmail.com)

Background

IL-10 has anti-inflammatory functions and stimulates the cytotoxicity and proliferation of antigen activated CD8+ T cells. T cell receptor mediated activation of CD8+ T cells elevates IL-10 receptors and PD-1, providing the mechanistic rationale for combining AM0010 and anti-PD1 for the treatment of cancer pts. A phase 1 clinical trial investigated the tolerability and anti-tumor activity of AM0010 alone and in combination with anti-PD1 immune checkpoint inhibitors.

We previously reported partial responses in 4 of 16 RCC pts treated with AM0010 monotherapy (Naing et al JCO 2016) and 4 of 8 RCC pts (50% ORR) and 2 of 5 of NSCLC pts (ORR 40%) treated with AM0010 and pembrolizumab. We enrolled phase 1 cohorts of RCC and NSCLC pts on AM0010 + nivolumab as ≥ 2nd line of treatment.

Methods

A total of 34 NSCLC pts. were enrolled on AM0010 (10-20mg/kg QD, SC) and pembrolizumab (2mg/kg, q3wk IV; n=5) or nivolumab (3mg/kg, q2wk IV; n=29). Pts. had a median of 2 prior therapies (PT, range 0-5). 37 RCC patients were enrolled on AM0010 (10-20mg/kg QD, SC) and pembrolizumab (2mg/kg, q3wk IV; n=8) or nivolumab (3mg/kg, q2wk IV; n=29), with a median of 1 PT (1-3). Tumor responses were assessed by irRC. Serum cytokines, T cell activation (FACS) and peripheral T cell clonality (TCR sequencing) were analyzed. (Fig. 1).

Results

AM0010 + anti-PD-1 was well tolerated. TrAEs included anemia, thrombocytopenia and fatigue, and were reversible and transient. As of July 30 2017, partial responses (PRs) were observed in 10 of 26 evaluable NSCLC pts (38.5%), including 3 (of 11) with PD-L1+ <1% and 4 (of 5) with PD-L1+ >50% cancers. PRs were observed in 14 of 34 evaluable RCC pts (41%). An additional 15 RCC pts had stable disease (44%), 7 of those had a tumor reduction > 30%. The mPFS and mOS has not been reached, the mFU is 8.9 m (range 0.5-26.5). Updated response data including data on delayed responses and the durability of response will be available.

The induction of immune cytokines (IL-18) in the serum, invigoration of CD8+ T cells and the increase of newly expanding T cell clones correlated with objective tumor response to AM0010 monotherapy and AM0010 + anti-PD-1.

Conclusions

AM0010 in combination with anti-PD-1 is well-tolerated in RCC pts, the recommended phase 2 dose is 10ug/kg. The robust efficacy data and the observed CD8 T cell activation is promising and encourages the continued study of AM0010 in combination with nivolumab.
Fig. 1 (abstract O12).

RCC - AM0010 + Anti-PD-1. Anti-PD-1 naive RCC were treated with AM0010 in combination with nivolumab (black) or pembrolizumab (red). ORR 41%, DCR 85%

O13 A phase 1 study of TSR-022, an anti-TIM-3 monoclonal antibody, in patients (pts) with advanced solid tumors

Glen J. Weiss1, Jason J. Luke2, Gerald Falchook3, Zeynep Eroglu4, Judy Wang5, Erika Hamilton6, J. Randolph Hecht7, Patricia LoRusso8, Joseph Paul Eder8, Lorraine Hughes9, Jing Wang9, Kelli Running9, Kristen McEachern9, Dmitri Bobilev9, Antoni Ribas7

1Western Regional Medical Center Inc., Goodyear, AZ, USA; 2University of Chicago Medical Center, Chicago, IL, USA; 3Sarah Cannon Research Institute at HealthONE, Denver, CO, USA; 4H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA; 5Florida Cancer Specialists, Sarasota, FL, USA; 6Tennessee Oncology, PLLC, Nashville, TN, USA; 7University of California Los Angeles - Jonsson Comprehensive Cancer Center, Los Angeles, CA, USA; 8Smilow Cancer Hospital at Yale - New Haven, New Haven, CT, USA; 9TESARO Inc., Waltham, MA, USA
Correspondence: Antoni Ribas (aribas@mednet.ucla.edu)

Background

T cell immunoglobulin and mucin-domain containing-3 (TIM-3) is a key immune checkpoint protein that has been implicated in both exhaustion of T effector cells and immune suppression mediated by regulatory T cells and myeloid cells. Up-regulation of TIM-3 expression on PD-1-positive tumor-infiltrating T cells is associated with reduced proliferation and secretion of cytokines important for T cell mediated anti-tumor activity in several cancers. Blocking TIM-3 therefore has the potential to restore tumor recognition by T cells and to increase anti-tumor immunity. TSR-022 is a potent, selective, investigational anti-TIM-3 antibody that is in development for the treatment of solid tumors as both a monotherapy and in combination with PD-1 blockade.

Methods

TSR-022 is being investigated in a multicenter, open-label, first-in-human phase 1 trial enrolling pts with advanced or metastatic solid tumors who have progressed after, or are intolerant to, available or approved therapies. The primary objective of the current part of this study is to evaluate the safety and tolerability of TSR-022 as a monotherapy and to determine the recommended phase 2 dose of TSR-022 as a single agent. Pts received IV infusion of TSR-022 monotherapy every 14 days in six escalating dose levels.

Results

As of July 19, 2017, 31 pts have been treated with monotherapy: 3 pts at 0.03 mg/kg, 3 pts at 0.1 mg/kg, 3 pts at 0.3 mg/kg, 9 pts at 1 mg/kg (all, 43 days of median follow-up); 7 pts at 3 mg/kg (28 days of median follow-up), and 6 pts at 10 mg/kg (23 days of median follow up). Adverse events that occurred in >15% of patients were fatigue (8 pts, 26%), abdominal pain (7 pts, 23%), nausea (6 pts, 19%); elevated ALT, elevated AST, back pain, constipation, and vomiting was each reported in 5 pts (16%). One dose-limiting toxicity event occurred with TSR-022 monotherapy (immune-related Grade 3 lipase elevation), which did not require treatment modification. No treatment-related serious adverse events were observed. TSR-022 exposure and peripheral receptor occupancy increased in a dose proportional manner from 0.3 to 10 mg/kg. To date, the best response was disease stabilization, observed in patients with various cancer types.

Conclusions

TSR-022 monotherapy was well tolerated across multiple dose levels. Adverse events were manageable and consistent with the safety profiles of other checkpoint inhibitors. Dose-escalation of TSR-022 in combination with an anti-PD-1 antibody is currently ongoing.

Trial Registration

ClinicalTrials.gov identifier NCT02817633.

O14 Initial results from phase 1 trial of M7824 (MSB0011359C), a bifunctional fusion protein targeting PD-L1 and TGF-β, in patients with NSCLC refractory or resistant to prior anti–PD-1/anti–PD-L1 agents

Fabrice Barlesi1, Nicolas Isambert2, Enriqueta Felip3, Byoung Chul Cho4, Dae Ho Lee5, Julio Peguero6, Guy Jerusalem7, Nicolas Penel8, Esma Saada-Bouzid9, Pilar Garrido10, Christoph Helwig11, Isabelle Dussault12, Laureen Ojalvo12, James Gulley13

1Aix-Marseille University, Livon, France; 2Centre Georges-François Leclerc, Dijon Cedex, France; 3Vall D’Hebron University Hospital, Barcelona, Spain; 4Yonsei Cancer Center, Seoul, Republic of Korea; 5University of Ulsan College of Medicine, Asan Medical Center, Seoul, Republic of Korea; 6Memorial Hermann Texas Medical Center, Houston, TX, USA; 7CHU Liege and Liege University, Liege, Belgium; 8Centre Oscar Lambret, Lille, France; 9Centre Antoine Lacassagne, Nice, France; 10University Hospital Ramón y Cajal, Madrid, Spain; 11Merck KGaA, Darmstadt, Germany; 12EMD Serono, Billerica, MA, USA; 13National Cancer Institute, Bethesda, MD, USA
Correspondence: Fabrice Barlesi (fabrice.barlesi@ap-hm.fr)

Background

The transforming growth factor β (TGF-β) pathway plays an important role in tumor immune escape and may potentially contribute to the failure of programmed death 1/programmed death ligand 1 (PD-1/PD-L1) therapy. M7824 (MSB0011359C) is an innovative first-in-class bifunctional fusion protein composed of a human IgG1 monoclonal antibody against PD-L1 fused with 2 extracellular domains of TGF-β receptor II to function as a TGF-β “trap.” We report initial results on the efficacy and tolerability of M7824 in patients with non-small cell lung cancer (NSCLC) refractory or resistant to prior anti–PD-1/anti–PD-L1 agents.

Methods

NCT02517398 is an ongoing, phase 1, open-label trial. In this expansion cohort, patients with advanced NSCLC refractory (progression after treatment initiation) or with acquired resistance (initial disease control with subsequent relapse) to prior treatment with anti–PD-1/anti–PD-L1 therapy received M7824 1200 mg q2w until confirmed progressive disease, unacceptable toxicity, or trial withdrawal. The primary objective is best overall response (BOR) per RECIST v1.1; secondary objectives include safety/tolerability.

Results

To date, 83 heavily pretreated patients (74.7% received ≥3 prior therapies) have received M7824 for a median duration of 8 weeks (range: 2-30 weeks) and 23 patients remain on active treatment with a median follow-up of 22.6 weeks (range, 8-35.3 weeks). Preliminary data show initial clinical activity was observed in 19 patients (disease control rate=22.9%; 2 confirmed partial responses [ongoing at 4.5 and 7.5 months] and 17 patients with stable disease [15 ongoing at 3 months]) based on investigator-assessed BOR. Notably, initial clinical activity was observed both in patients refractory or resistant to prior anti–PD-1/anti–PD-L1 therapy. Fourteen patients (16.9%) experienced grade ≥3 treatment-related adverse events (AEs), most commonly rash/pruritus (n=4, 4.8%) and fatigue/asthenia (n=4, 4.8%). Cutaneous lesions occurred in 3 patients (3.6%), including keratoacanthoma and squamous cell carcinoma (similar to other TGF-β–inhibiting agents) and were well managed by surgical excision. One patient discontinued the study due to a treatment-related AE (bullous pemphigoid). One patient died from pneumonia assessed by the investigator as treatment related.

Conclusions

Based on preliminary data from this still-ongoing trial, treatment with M7824 results in initial clinical activity and a manageable safety profile in heavily pretreated patients with NSCLC refractory or resistant to prior anti–PD-1/anti–PD-L1 agents. Updated data, including results based on biomarkers such as tumor PD-L1 expression and circulating levels of TGF-β1 and cytokines, will be presented at the meeting.

Trial Registration

NCT02517398

O15 Neoadjuvant nivolumab versus combination ipilimumab and nivolumab followed by adjuvant nivolumab in patients with resectable stage III and oligometastatic stage IV melanoma: preliminary findings

Sangeetha M. Reddy, Rodabe N. Amaria, Christine N. Spencer, Michael Tetzlaff, Alexandre Reuben, Miles Andrews, Linghua Wang, Scott Woodman, Haifeng Zhu, Jorge Blando, Luis Vence, Shaojun Zhang, Hong Jiang, Vancheswaran Gopalakrishnan, Courtney Hudgens, Khalida Wani, Hussein Tawbi, Adi Diab, Isabella Glitza, Sapna Patel, Wen-Jen Hwu, Michael Wong, Patrick Hwu, Janice Cormier, Anthony Lucci, Richard Royal, Jeffrey E. Lee, Lauren Simpson, Elizabeth M. Burton, Jeffrey E. Gershenwald, Merrick Ross, James Allison, Padmanee Sharma, Michael Davies, Jennifer Wargo

The University of Texas MD Anderson Cancer Center, Houston, TX, USA
Correspondence: Jennifer Wargo (jwargo@mdanderson.org)

Background

Single agent nivolumab (nivo) and combination ipilimumab (ipi) + nivo are efficacious in advanced melanoma, but have not been extensively evaluated in patients with high-risk resectable disease. The optimal neoadjuvant immunotherapy regimen is unknown and development of biomarker specific neoadjuvant trials are critical in understanding mechanisms of response, resistance and informing clinical management. We designed a clinical trial to test these regimens in the neoadjuvant setting coupled with longitudinal sample collection designed to identify strategies to further improve outcomes.

Methods

Patients with resectable stage III or oligometastatic stage IV melanoma were randomized to nivo 3 mg/kg IV every 2 weeks for 4 doses or ipi 3 mg/kg + nivo 1 mg/kg IV every 3 weeks for 3 doses prior to surgical resection, with both arms receiving adjuvant nivo 3 mg/kg for 6 months (Fig. 1). Responses were measured at time of surgery as defined by RECIST 1.1 and percent tumor viability in surgical specimen. Blood and tumor samples were collected at baseline, on-treatment, and at surgery.

Results

23 out of a planned 40 patients were enrolled. The response rates (RR) were 33% for nivo and 67% for ipi+nivo in evaluable patients (n=18 total, 9 in each arm) (Fig. 2). Pathological complete response (pCR) rates were 22% for nivo and 56% for ipi+nivo. Importantly, 22% of patients in nivo arm experienced disease progression and were ineligible for surgical resection, prompting review by the data safety monitoring board and early trial closure. No grade 3 or higher toxicities were seen with nivo compared to 89% with ipi+nivo. All patients on ipi+nivo received surgical resection, but 33% had a surgical delay related to toxicity. RNA-seq analysis of longitudinal tumor samples revealed differential effects of each treatment on the tumor microenvironment as well as between responders (R) and non-responders (NR) within each treatment arm. Immune profiling demonstrated enriched CD8+ T cell infiltrates in on-treatment samples in R versus NR (p=0.04). T cell receptor sequencing showed a lower degree of remodeling from pre- to on-treatment tumor in R vs NR (as measured by Morisita’s index, p<0.01).

Conclusions

Treatment with neoadjuvant nivo or ipi+nivo has activity in resectable melanoma. Ipi+nivo possessed a higher RR but also higher grade III/IV toxicities whereas nivo led to a higher rate of disease progression and resultant unresectable disease. Ongoing correlative studies will provide insights into the mechanisms of response, toxicity and resistance.

Trial Registration

NCT02519322
Fig. 1 (abstract O15).

See text for description

Fig. 2 (abstract O15).

See text for description

O16 A phase 1/2 study of CB-839, a first-in-class glutaminase inhibitor, combined with nivolumab in patients with advanced melanoma (MEL), renal cell carcinoma (RCC), or non-small cell lung cancer (NSCLC)

Funda Meric-Bernstam1, Michael Gordon2, Scott Tykodi3, Elaine Lam4, Ulka Vaishampayan5, Jorge Chaves6, Petros Nikolinakos7, Alice Fan8, Richard Lee9, David McDermott10, Geoffrey Shapiro11, Leena Gandhi12, Hussein Tawbi1, Shailender Bhatia3, Lucas Muigai13, Yonchu Jenkins13, Sam Whiting13, Martin Voss14

1The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2Pinnacle Oncology Hematology, Scottsdale, AZ, USA; 3Seattle Cancer Care Alliance, Seattle, WA, USA; 4University of Colorado School of Medicine, Aurora, CO, USA; 5Karmanos Cancer Center, Detroit, MI, USA; 6Northwest Medical Specialties, Tacoma, WA, USA; 7University Cancer & Blood Center, Athens, GA, USA; 8Stanford University Medical Center, Stanford, CA, USA; 9Massachusetts General Hospitcal, Boston, MA, USA; 10Beth Israel Deaconess Medical Center, Boston, MA, USA; 11Medical Oncology, Boston, MA, USA; 12NYU Medical Oncology Associates, New York, NY, USA; 13Calithera Biosciences, Inc., South San Francisco, CA, USA; 14Memorial Sloan Kettering Cancer Center, New York, NY, USA
Correspondence: Funda Meric-Bernstam (fmeric@mdanderson.org)

Background

CB-839 is a first-in-class, oral, highly selective inhibitor of glutaminase that targets tumor glutamine metabolism. Competition between tumor cells and immune cells for nutrients, including glutamine, in the tumor microenvironment can create a metabolic checkpoint that induces local immune suppression. CB-839 inhibits tumor glutamine consumption and increases glutamine availability to support T-cell activity. In pre-clinical models, CB-839 increases intra-tumoral glutamine and enhances antitumor activity of PD-1/PD-L1 inhibitors. This study is evaluating the safety and efficacy of CB-839 in combination with the PD-1 inhibitor, nivolumab, in patients with advanced RCC, MEL, or NSCLC, including cohorts of patients refractory to anti-PD-1/PD-L1.

Methods

In dose escalation, patients received CB-839 600 mg or 800 mg PO BID in combination with standard-dose nivolumab. In expansion, patients are enrolling into one of five cohorts to receive CB-839 800 mg BID with standard-dose nivolumab: Naïve RCC – anti-PD-1/PD-L1 naïve, previously treated with anti-angiogenic therapy; Rescue RCC – on nivolumab with progressive disease (PD) or > 6 month stable disease without response; Other IO RCC – anti-PD-1/PD-L1 in any previous line of therapy with PD and no history of response; Rescue MEL – on anti-PD-1/PD-L1 with PD; Rescue NSCLC – on anti-PD-1/PD-L1 with PD or > 6 month SD without response. Tumor response assessment is every 8 weeks per RECIST 1.1. Tumor biopsy and blood are collected for biomarker analysis. Bristol-Myers Squibb is providing nivolumab for this study.

Results

The study is ongoing and 50 patients have enrolled to date (42 at 800 mg dose). A maximum tolerated dose of CB-839 was not reached. One DLT (Gr3 ALT increased) occurred at the 800 mg dose. The combination has been well tolerated; the most common treatment-related Gr≥3 AE has been ALT increased (4%). RECIST 1.1 tumor responses by cohort are shown below (Table 1).

There has been one complete response (CR) and two partial responses (PR) out of 7 response-evaluable MEL patients (ORR 43%). All three responding patients were progressing on anti-PD-1/PD-L1 at study entry and two had progressed on 3 prior immunotherapy regimens. Paired tumor biopsies were available from one MEL responder and analysis showed on-treatment increase in T-effector gene expression including granzyme and perforin.

Conclusions

Treatment with the CB-839+nivolumab is well-tolerated and has clinical activity in anti-PD-1/PD-L1 refractory patients. Notable responses have occurred in MEL patients progressing on nivolumab at study entry and refractory to multiple prior immunotherapy regimens. Updated safety, biomarker and response data will be presented.

Trial Registration

NCT02771626
Table 1 (abstract O16).

See text for description

O17 OX40 T-cell costimulatory agonist BMS-986178 alone or in combination with nivolumab in patients with advanced solid tumors: initial phase 1 results

Anthony J. Olszanski1, Ignacio Melero2, Michael Ong3, Anna Spreafico4, Kimberley Heinhuis5, Richard Carvajal6, Maria Ochoa de Olza7, Quincy Chu8, Anke Klippel9, Yun Shen9, Yan Feng9, Bryan Barnhart9, Alan Korman9, Michael Quigley9, Paula M. Fracasso9, Praveen Aanur9, Martin Gutierrez10

1Fox Chase Cancer Center, Philadelphia, PA, USA; 2Clinica Universidad de Navarra, Pamplona, Spain; 3The Ottawa Hospital Cancer Centre, Ottawa, ON, Canada; 4Princess Margaret Cancer Centre, Toronto, ON, Canada; 5The Netherlands Cancer Institute, Amsterdam, Netherlands; 6Columbia University Medical Center, New York, NY, USA; 7Vall d'Hebron University Hospital, Barcelona, Spain; 8Cross Cancer Institute, University of Alberta/Alberta Health Services, Edmonton, AB, Canada; 9Bristol-Myers Squibb, Princeton, NJ, USA; 10Hackensack University Medical Center, Hackensack, NJ, USA
Correspondence: Anthony J. Olszanski (anthony.olszanski@fccc.edu)

Background

With the clinical success of checkpoint blockade, focus has shifted to novel approaches to enhance the benefits of immunotherapy. Activation of the costimulatory tumor necrosis factor receptor family member OX40 may lead to T effector cell activation, inhibition of T regulatory cell–mediated suppression, and blockade of T regulatory cell generation. Preclinical data show enhanced antitumor T-cell activity when anti-OX40 is combined with checkpoint inhibitors (eg, anti–PD-1), supporting the potential benefits of this combination to provide deeper, more durable responses than checkpoint blockade alone. BMS-986178 is a fully human IgG1 agonist monoclonal antibody that binds with high affinity to OX40. Here we describe preliminary results of a phase 1/2a study of BMS-986178 ± nivolumab (anti–PD-1) in patients with advanced solid tumors (NCT02737475).

Methods

Patients with advanced/metastatic solid tumors with ≥1 prior therapy were treated in this open-label, dose-escalation and dose-expansion study. During escalation, patients received BMS-986178 (20-320 mg) IV Q2W or BMS-986178 (20-320 mg) + nivolumab 240 mg IV Q2W. Safety, immunogenicity, pharmacokinetics, pharmacodynamics, and preliminary antitumor activity were evaluated. Pharmacodynamic analyses included evaluation of OX40, PD-1, Ki67, FOXP3, and CD8 by immunohistochemistry in tumors and immunophenotyping in peripheral blood mononuclear cells.

Results

As of June 12, 2017, 59 patients were treated with BMS-986178 (n=20) or BMS-986178 + nivolumab (n=39). Approximately one-third of patients in the monotherapy (30%) and combination (28%) arms had received prior anti–PD-1/PD-L1–based therapy. Most treatment-related AEs were grade 1/2 (Table 1); 1 patient treated with BMS-986178 160 mg + nivolumab had treatment-related grade 3 pneumonitis that led to discontinuation. The maximum tolerated doses of BMS-986178 ± nivolumab have not yet been reached. The pharmacokinetics of BMS-986178 ± nivolumab were linear, and exposure increased proportionally with BMS-986178 doses of 20-320 mg. Preliminary pharmacodynamic activity showed an increase in proliferating T cells and a decrease in FOXP3 cells in tumors with BMS-986178 + nivolumab. Evidence of antitumor activity was observed in several patients treated with the combination, including a patient who had progressed on prior anti–PD-1 therapy; these data will be reported.

Conclusions

BMS-986178, an OX40 agonist, given alone and in combination with nivolumab, was well tolerated; the combination had a safety profile similar to that of nivolumab monotherapy. Treatment with BMS-986178 + nivolumab resulted in pharmacodynamic immune changes in tumors and preliminary antitumor activity. Further evaluation of this combination in patients with advanced solid tumors is ongoing.

Trial Registration

ClinicalTrials.gov, NCT02737475
Table 1 (abstract O17).

Treatment-related AEs with BMS-986178 and BMS-986178 + nivolumab

aGrade 3 treatment-related pneumonitis was reported in 1 patient treated with BMS-986178 160 mg + nivolumab 240 mg.

018 TLR9 agonist harnesses innate immunity to drive tumor-infiltrating T cell expansion in distant lesions in a phase 1/2 study of intratumoral IMO-2125+ipilimumab in anti-PD1 refractory melanoma patients

Cara Haymaker1, Marc Uemura1, Wen-Jen Hwu1, Ravi Murthy1, Marihella James1, Ankit Bhatta1, Salah Eddine Bentebibel1, Julie Brevard2, James Geib3, Kathryn Lipford2, Mark Cornfeld3, Srinivas Chunduru3, Cassian Yee1, Scott Woodman1, Rodabe Amaria1, Sapna Patel1, Hussein Tawbi1, Isabella Glitza Oliva1, Michael Davies1, Willem Overwijk1, Patrick Hwu1, Chantale Bernatchez1, Adi Diab1

1University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2Idera Pharmaceuticals Inc., Cambridge, MA, USA; 3Idera Pharmaceuticals, Inc., Exton, PA, USA
Correspondence: Adi Diab (adiab@mdanderson.org)

Background

While checkpoint inhibitors (CPI) have transformed melanoma treatment, many patients remain refractory. Pre-clinical models show combination intratumoral IMO-2125 (TLR9 agonist) with anti-CTLA-4 or anti-PD-1 antibody results in improved tumor control compared with either agent alone. We hypothesized that this combination would stimulate local antigen-presenting cells, resulting in activation of anti-tumor T cells and distant responses.

Methods

Adults with anti-PD-1 refractory, unresectable stage III/IV melanoma were enrolled. IMO-2125, escalating from 4 – 32 mg is administered under image guidance, intratumorally on weeks 1, 2, 3, 5, 8, and 11 with standard ipilimumab. Biopsies were obtained in both the injected and distant tumor at baseline, 1 day and 8 weeks (W8) post injection. Immune analyses included phenotypic, activation, and functional characterization of DC subsets and T cells. T-cell repertoire diversity was evaluated by high-throughput CDR3 sequencing and changes in gene expression signatures were assessed by NanoString.

Results

Biopsies were obtained from 17 of the 18 subjects enrolled on the IMO-2125-ipilimumab arm. Of 15 subjects with post-baseline disease evaluations, 9 (60%) showed shrinkage in uninjected tumors with an ORR of 33% by RECIST 1.1. Fresh tumor biopsies taken 24h post IMO-2125 injection demonstrated induction of an IFN-a-response gene signature (IRF7) in all patients (p<0.0001), as well as, maturation of the myeloid DC1 subset (CD1c+CD303-) by upregulation of MHC class II (6/12 patients) and upregulation of PD-L1 on malignant cells (10/14 patients). W8 biopsies of the uninjected tumor lesion reveal the induction of Ki67+ expression by flow cytometry in effector CD8+ T cells in 6 of 7 regressing lesions indicating an abscopal effect of this combination. In 6 patients with both tumor shrinkage and flow cytometry staining of PBMCs, 4 had higher CD8+ T-cell proliferation in the tumor than in blood on-treatment. CDR3 sequencing shows that the top 50 clones form a larger proportion of the repertoire at W8 suggesting local proliferation in both lesions, as well as, the specific expansion in the distant lesion of clones shared with the injected site in responders.

Conclusions

Combination IMO-2125 and ipilimumab therapy is a promising new treatment for anti-PD-1 refractory melanoma. We show this treatment strategy to result in a local IFN-a-gene signature coupled with mDC1 maturation. Additionally, the hallmark of tumor shrinkage appears to be the presence of Ki67+ CD8+ T cell effector cells in the uninjected tumor and expansion of clones shared between both lesions (abscopal effect).

Trial Registration

NCT02644967

019 ENCORE-601: Phase 1b/2 study of entinostat (ENT) in combination with pembrolizumab (PEMBRO) in patients with non-small cell lung cancer (NSCLC)

Leena Gandhi1, Melissa L. Johnson2, Mateusz Opyrchal3, Suresh Ramalingam4, Pasi Janne5, Abraham Chachoua1, Peter Ordentlich6, Susan Brouwer6, Serap Sankoh6, Emmett Schmidt7, Michael L. Meyers8, Matthew Hellmann9

1NYU Langone Medical Center, New York, NY, USA; 2Sarah Cannon Research Institute, Nashville, TN, USA; 3Roswell Park Cancer Institute, Buffalo, NY, USA; 4Winship Cancer Institute, Emory University, Atlanta, GA, USA; 5Dana Farber Cancer Institute, Boston, MA, USA; 6Syndax Pharmaceuticals, Inc., Waltham, MA, USA; 7Merck & Co., Inc., Kenilworth, NJ, USA; 8Syndax Pharmaceuticals, Inc., New York, NY, USA; 9Memorial Sloan Kettering Cancer Center, New York, NY, USA
Correspondence: Leena Gandhi (leena.gandhi@nyumc.org)

Background

Entinostat is an oral, class I selective histone deacetylase (HDAC) inhibitor that has been shown in pre-clinical models to enhance anti-PD-1 activity through inhibition of immune suppressor cells in the tumor microenvironment. ENCORE 601 is a Phase 1b/2 study evaluating safety and efficacy of ENT plus PEMBRO in NSCLC, melanoma and colorectal cancer patients. We previously reported the results of Phase 1b, which identified ENT 5 mg PO weekly plus PEMBRO 200 mg IV every 3 weeks as the dose to be further explored.

Methods

Efficacy of ENT 5 mg PO weekly plus PEMBRO 200 mg IV q3wks is being assessed in a Simon 2-stage Phase 2 study involving two cohorts of patients with advanced NSCLC: anti-PD-(L)1–naïve (Cohort 1), and progressed on anti-PD-(L)1 (Cohort 2). The primary endpoint is objective response rate. Based on investigator feedback to incorporate Phase 1b patients dosed at entinostat 5 mg into the Stage 2 go/no go assessment, revised criteria for advancement were ≥4 responses observed out of 17 evaluable patients in Cohort 1 and ≥3 out of 31 in Cohort 2. Tumor biopsies and blood samples for immune correlates were taken pre- and post-treatment.

Results

Enrollment has been completed for Stage 1 of both cohorts. In Cohort 1, 4 of 17 (24%, 95% CI: 7-50) evaluable patients achieved a partial response (PR; 2 confirmed, 2 unconfirmed). In Cohort 2, 3 of 31 (10%, 95% CI: 2-26) evaluable patients achieved a PR (2 confirmed, 1 unconfirmed). As of the data cutoff, the longest duration of response was 24 weeks (and ongoing). Baseline PD-L1 expression for responders was <1% (1 patient), 1-49% (2 patients) and not available (1 patient) for Cohort 1, and <1% (2 patients) and not available (1 patient) for Cohort 2. 31% of patients experienced a Grade 3/4 event deemed related to study drug. The most common of these events included hypophosphatemia and neutropenia in Cohort 1, and fatigue, anemia, and pneumonitis in Cohort 2. 13% of patients discontinued therapy due to an adverse event. Circulating myeloid derived suppressor cells were reduced in both cohorts at Cycle2/Day1 compared to pretreatment. Gene expression analysis of tumor biopsies and association with responses is in progress.

Conclusions

ENT plus PEMBRO demonstrates anti-tumor activity and acceptable safety in patients with NSCLC who are both naïve to and have progressed on prior PD-(L)1 blockade. Based on the responses seen, Cohort 2 has advanced to Stage 2 and is currently enrolling.

O20 Pivot-02: Preliminary safety, efficacy and biomarker results from the Phase 1/2 study of CD-122-biased agonist NKTR-214 plus nivolumab in patients with locally advanced/metastatic solid tumors

Adi Diab1, Nizar Tannir1, Daniel Cho2, Vali Papadimitrakopoulou1, Chantale Bernatchez1, Cara Haymaker1, Salah Eddine Bentebibel1, Brendan Curti3, Michael Wong1, Scott Tykodi4, Igor Puzanoff5, Ira Smalberg5, Ivan Gergel6, Mary Tagliaferri6, Jonathan Zalevsky6, Ute Hoch6, Sandra Aung6, Michael Imperiale6, Wendy Clemens7, Harriet Kluger8, Michael Hurwitz8, Patrick Hwu1, Mario Sznol8

1The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2NYU Medical Oncology Associates, New York, NY, USA; 3Providence Cancer Center and Earle A. Chiles Research Institute, Portland, OR, USA; 4University of Washington and Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 5Roswell Park Cancer Institute, Buffalo, NY, USA; 6Nektar Therapeutics, San Francisco, CA, USA; 7Bristol-Myers Squibb, Lawrence Township, NJ, USA; 8Yale School of Medicine, New Haven, CT, USA
Correspondence: Adi Diab (adiab@mdanderson.org)

Background

In patients with advanced solid tumors, NKTR-214, a CD-122-biased agonist, increased newly proliferative CD8+ T cells in the tumor and increased cell surface PD-1 expression, demonstrating a potentially synergistic mechanism with anti-PD-1 therapy. Preliminary safety, efficacy and biomarker results from dose-escalation of the Phase 1/2 study combining NKTR-214 with nivolumab are presented.

Methods

Enrolled patients in dose escalation were checkpoint inhibitor naïve with locally advanced/metastatic melanoma (mM), RCC or NSCLC. NKTR-214 and nivolumab were administered concurrently by IV infusion in the outpatient setting in Q2W or Q3W dosing schedules. Pre- and on-treatment blood and matched tumor samples were evaluated for changes in immune cell populations, gene expression, and T cell receptor repertoire. Responses were assessed Q8W per RECIST 1.1.

Results

At July 25, 2017, data cutoff, 31 patients were treated across one of 5 dose escalating cohorts of NKTR-214 (@0.003(n=3), @0.006(n=25) or @0.009(n=3) mg/kg) administered with nivolumab given either 240 mg Q2W or 360 mg Q3W. There were no Grade ≥3 treatment-related adverse events (TRAEs) reported at or below the NKTR-214 0.006 mg/kg plus nivolumab cohorts. There were no discontinuations from TRAEs. The most common TRAEs occurring in ≥50% of patients included grade 1-2 fatigue, flu-like symptoms and rash. At the highest dose (NKTR-214 0.009 mg/kg + nivolumab), two patients experienced DLTs including hypotension and metabolic acidosis. DLTs were short-lived and manageable with both patients continuing treatment at lower doses. 16 patients with Stage IV mM or Stage IV RCC were evaluable with ≥1 on treatment scan. In mM, 1 complete response was observed plus 4 partial responses (n=8). In RCC, 4 PRs were observed (n=8). 5/16 patients had SD, 2 patients had PD. In tumor tissue, there was an up to 450-fold increase in tumor-infiltrating CD8+ T cells from baseline with a limited (≤2-fold) increase in T regulatory cells. Infiltrating T cells were newly proliferative with elevated PD-1 expression. In blood, ICOS+ CD8+ T cells increased with treatment representing 13% of total CD8+ T cells. Updated data will be presented.

Conclusions

Combination of NKTR-214 and nivolumab was well-tolerated and manageable in outpatient setting. Nivolumab-related AEs were not increased by NKTR-214. Immune activation was observed in peripheral blood and tumor microenvironment. Preliminary efficacy with mM and RCC show encouraging response rates. A recommended Phase 2 dose schedule of Q3W was established (NKTR-214 0.006 mg/kg plus nivolumab 360 mg), and enrollment is underway in expansion cohorts in eight different indications (n=230).

Trial Registration

Trial Registration: NCT02983045

O21 Pooled 3-year overall survival data from phase II and phase III trials of nivolumab (NIVO) combined with ipilimumab (IPI) in advanced melanoma

Michael Postow1, James Larkin2, Jedd D. Wolchok1, Vanna Chiarion-Sileni3, F. Stephen Hodi4, Piotr Rutkowski5, Jean-Jacques Grob6, Kenneth Grossmann7, Dirk Schadendorf8, Jason Chesney9, Michael Smylie10, Andrew Hill11, Caroline Robert12, Anna C. Pavlick13, John Haanen14, Grant A. McArthur15, David McDermott16, Sheena Demelo17, Rafia Bhore17, Rene Gonzalez18

1Memorial Sloan Kettering Cancer Center, New York, NY, USA; 2Royal Marsden Hospital, London, UK; 3Oncology Institute of Veneto IRCCS, Padova, Italy; 4Dana-Farber Cancer Institute, Boston, MA, USA; 5Maria Sklodowska-Curie Memorial Cancer Center & Institute of Oncology, Warsaw, Poland; 6Hospital de la Timone, Marseille, France; 7Huntsman Cancer Institute, Salt Lake City, UT, USA; 8Department of Dermatology, University of Essen, Essen, Germany; 9James Graham Brown Cancer Center, Louisville, KY, USA; 10Cross Cancer Institute, Edmonton, AB, Canada; 11Tasman Oncology Research, Southport Gold Coast, Australia; 12Institut Gustave-Roussy, Villejuif, France; 13NYU Medical Oncology Associates, New York, NY, USA; 14Netherlands Cancer Institute, Amsterdam, Netherlands; 15Peter MacCallum Cancer Centre, Melbourne, Australia; 16Beth Israel Deaconess Medical Center, Boston, MA, USA; 17Bristol-Myers Squibb, Princeton, NJ, USA; 18University of Colorado Cancer Center, Denver, CO, USA
Correspondence: Michael Postow (postowm@mskcc.org)

Background

NIVO (anti-PD-1) plus IPI (anti-CTLA-4) improved progression-free survival (PFS) and objective response rate vs IPI alone in the phase II CheckMate 069 and phase III CheckMate 067 trials of treatment-naive patients with advanced melanoma. Recently, a 2-year overall survival (OS) rate of 64% was reported for patients treated with NIVO+IPI in CheckMate 067. Here, we report 3-year OS, PFS, and safety results in all randomized patients from CheckMate 069.

Methods

Patients were randomized 2:1 to NIVO 1 mg/kg + IPI 3 mg/kg (n=95) or IPI 3 mg/kg + placebo (n=47) Q3W for 4 doses followed by NIVO 3 mg/kg or placebo Q2W, respectively, until disease progression or unacceptable toxicity.

Results

55% of patients in the IPI group crossed over to receive NIVO upon progression. At a median follow-up of 37 months, median OS was not reached (NR) for NIVO+IPI and was 32.9 months for IPI (hazard ratio [HR]: 0.76; P=0.2829), with 3-year OS rates of 57% and 46%, respectively. Median PFS was NR for NIVO+IPI and was 3.0 months for IPI (HR: 0.35; P<0.0001). Median duration of response was NR in both groups. Time to subsequent therapy in the NIVO+IPI and IPI groups was NR and 5.4 months, where 63% (47/75) and 19% (7/37) remain free of subsequent therapy at the 3-year landmark, respectively. The safety profile remained similar to the earlier reports, with grade 3-4 treatment-related adverse events in the NIVO+IPI and IPI groups of 54% and 24%; 38% and 9% discontinued due to any grade adverse events, respectively. Three-year OS rates in patients who discontinued due to toxicity were 68% for NIVO+IPI and 25% for IPI; median duration of response was NR in both subgroups.

Conclusions

In summary, NIVO+IPI continues to show a significant improvement in PFS, with consistent safety results, and a favorable 3-year landmark OS rate. A pooled analysis, which will include updated 3-year OS data from CheckMate 067, will be presented as well as outcomes in patients off all therapy.

Trial Registration

NCT01927419 (CheckMate 069), NCT01844505 (CheckMate 067)

O22 Antibody-mediated neutralization of soluble MIC significantly enhances response to CTLA4 blockade and reduces therapy-induced colitis

Jennifer Wu1, Jinyu Zhang2

1Northwestern University, Chicago, IL, USA; 2Medical University of South Carolina, Charleston, SC, USA
Correspondence: Jennifer Wu (jennifer.wu@northwestern.edu)

Background

Antibody therapy targeting cytotoxic T lymphocyte-associated antigen 4 (CTLA4) elicited survival benefits in cancer patients; however, the overall response rate is limited. In addition, anti-CTLA4 antibody therapy induces a high rate of immune related adverse events (IRAEs). The underlying factors that may influence anti-CTLA4 antibody therapy are not well defined. Ligation of the human activation immune receptor natural killer group 2D (NKG2D) has been implicated in optimizing CD8 T cell co-stimulation and pathogenesis of colitis. Herein we investigated the impact of the human NKG2D ligand MIC, which was broadly expressed by human solid tumors, on the therapeutic outcome of anti-CTLA4 blockade.

Methods

We utilized the well-established MICnegative TRAMP and MICpositive TRAMP/MIC spontaneous pre-clinical tumor models to address our question. We treated cohorts of these animals with anti-CTLA4 antibody or anti-sMIC antibody monotherapy or combination therapy and assessed treatment response and related toxicity. We further corroborate our findings with various indications of sMICnegative and sMICpositive transplantable syngeneic tumor models.

Results

As expected, MICnegative tumors exhibited nominal response to anti-CTLA4 therapy and no response to anti-sMIC therapy. Unexpectedly, animals bearing MICpositive tumors and with elevated serum sMIC (sMIChi) responded poorly to anti-CTLA4 antibody therapy with significantly shortened survival (Fig. 1). Animals with high levels of serum sMIC had increased lung metastasis and developed colitis in response to anti-CTLA4 antibody therapy (Fig. 2). Co-administration of a sMIC-neutralizing monoclonal antibody B10G5 with the anti-CTLA4 antibody alleviated colitis in sMIChi animals and generated a cooperative anti-tumor therapeutic effect by synergistically augmenting innate and adoptive anti-tumor immune responses. Mechanistically, we show that sMIC downmodulation of CD3z on NK and T cells, which ultimately impairs TCR/CD3 signaling and NK cell receptor signaling and potentially CD4 T cell-mediated tolerance. We further demonstrate that the combination of anti-CTLA4 with anti-sMIC antibody cooperatively primes DC activation, overcomes CD8 T cell tolerance, enhances TCR/CD3 signaling capacity in CD8 T cells, and increases T cell clonality or repertoire complexity in tumor infiltrates.

Conclusions

Our findings imply that a new combination therapy could improve the clinical response to anti-CTLA4 antibody therapy. Our findings also suggest that pre-screening cancer patients for serum sMIC may help in selecting candidates who will elicit a better response to anti-CTLA4 antibody therapy.
Fig. 1 (abstract O22).

High soluble MIC 9sMIC) provokes poor response to anti-CTLA4 therapy

Fig. 2 (abstract O22).

High sMIC induces colitis in response to anti-CTLA4 therapy

023 Multiplex three-dimensional optical mapping of tumor immune microenvironment

Steve Seung-Young Lee, Vytautas P. Bindokas, Stephen J. Kron

University of Chicago, Chicago, IL, USA
Correspondence: Stephen J. Kron (skron@uchicago.edu)

Background

The tumor microenvironment is a three-dimensional (3D) system of diverse cellular and non-cellular components whose heterogeneous structure is typically defined by haphazard growth of cancer cells and a disordered microvasculature. Tumor-immune cell interactions occur within this context, providing a challenge for analysis of the infiltrate and anti-tumor immune responses by two-dimensional (2D) methods such as immunohistochemistry (IHC). Toward overcoming the limitations of 2D methods, we introduce Transparent Tissue Tomography (T3) as a tool for quantitative 3D imaging cytometry of the tumor immune microenvironment.

Methods

For T3 imaging cytometry, tumors were sliced into 400 μm macrosections to facilitate immunofluorescence staining, optical clearing, and confocal microscopic imaging. Macrosections were stained overnight with panels of four to six directly labeled fluorescent primary antibodies then cleared by D-fructose. The full volume of each macrosection was scanned in each channel using a confocal microscope and tumor images were tomographically reconstructed from the macrosection images. The tumor images were segmented to discriminate cell types, map biomarkers, and perform spatial analysis.

Results

As an application of T3 imaging cytometry, we examined the distribution of programmed death-ligand 1 (PD-L1) expression in spontaneous Her2+ mammary tumors formed in BALB-NeuT mice. T3 analysis of whole tumors determined PD-L1 expression by tumor cells at the periphery and CD31+ vascular endothelium in the core. For the first time, 3D tomographic projection disclosed vascular PD-L1 expression localized between the endothelium and inner layer of smooth muscle cells. In turn, T3 revealed a strong spatial correlation between CD45+ immune cell distribution and PD-L1 expression. Applying T3 to pharmacokinetic analysis, we observed uneven accumulation of anti-PD-L1 antibody in perfused regions and lack of delivery to hypoxic domains. In addition, 3D vascular mapping showed that the blood vessels wrapped in smooth muscle cell layers restrained anti-PD-L1 antibody extravasation into tumor. Toward translation, T3 was adapted to analyze whole core needle biopsies. Modeling on-treatment biopsy analysis, T3 confirmed a compound effect of radiation and anti-PD-L1 therapy on infiltration of effector cytotoxic lymphocytes such as granzymeB+ NK cells and T cells.

Conclusions

By assessing multiple tumor parameters simultaneously at cellular resolution, T3 provides a unique window into the heterogeneity of the tumor immune microenvironment. We anticipate that T3 can be applied broadly to facilitate preclinical studies of tumor biology and therapy. In particular, spatial, multiparameter T3 analysis may serve as a tool to improve diagnostic, prognostic and predictive testing of patient biopsies as part of evaluation for immune checkpoint blockade therapy.

024 First-in-human study with intratumoral administration of a CD40 agonistic antibody: preliminary results with ADC-1013/JNJ-64457107 in advanced solid malignancies

Peter Ellmark1, Sandra Irenaeus2, Dorte Nielsen3, Adnan Deronic1, Niina Veitonmäki1, Per Norlén1, Camilla Wennersten1, Jeffrey Yachnin4, Gustav Ullenhag2

1Alligator Bioscience, Lund, Sweden; 2Uppsala University, Uppsala, Sweden; 3University of Copenhagen, Herlev, Denmark; 4Karolinska Institutet and University Hospital, Stockholm, Sweden
Correspondence: Peter Ellmark (pek@alligatorbioscience.com)

Background

Agonistic antibodies targeting CD40 activate dendritic cells and can expand and activate tumor specific T cells. Preclinical studies have demonstrated that intratumoral administration of CD40 agonists has the potential to improve anti-tumor efficacy and reduce immune related adverse events compared to intravenous administration. Herein, we report the first study of a CD40 agonistic antibody administered intratumorally in cancer patients.

Methods

A Phase I, open label, multicentre study was conducted to evaluate the safety, pharmacokinetics and pharmacodynamics of ADC-1013 in patients with advanced solid tumors who had received established treatments. A 3+3 dose escalation was conducted with a 28-day window (bi-weekly dosing). Twenty-three patients were treated with ADC-1013 intratumorally (dosing from 22.5 μg/kg up to 400 μg/kg) or intravenously (dosing at 75 μg/kg). The pharmacodynamic effects observed in the patients were further verified in an hCD40tg mouse model.

Results

Adverse events were primarily fatigue, pyrexia, nausea and vomiting, and were mostly CTCAE Grade 1 or 2 and transient. Two subjects receiving intratumoral bolus injections of ADC-1013 into hepatic lesions at a dose of 400 μg/kg experienced Grade 3 abdominal pain assessed as DLTs. Elevations in cytokine levels (MCP-1, TNFα and IL-6) were observed in patients receiving 200 μg/kg and 400 μg/kg of ADC-1013. The systemic ADC-1013 exposure and cytokine release was more pronounced in patients receiving injections in a hepatic lesion than patients receiving injections in a non-hepatic lesion. Treatment with ADC-1013 resulted in a marked decrease in B cell levels in peripheral blood after 24 h, recovering to baseline levels within 8 days. Remaining B cells significantly increased their expression of the cell surface activation marker CD86 24 h after first dose, returning to baseline within 8 days.

The pharmacodynamic effects were further studied in hCD40tg mice demonstrating activation of antigen presenting cells such as dendritic cells and subsequent activation of T cells. Moreover, ADC-1013 treatment in this mouse model acts synergistically with a PD-1 inhibitor.

Conclusions

The results from the first-in-human study of ADC-1013 indicate that intratumoral administration of ADC-1013 into non-hepatic lesions (superficial metastases) is well tolerated at doses up to at least 400 μg/kg, while intratumoral administration of ADC-1013 into hepatic lesions resulted in toxicities associated with cytokine release and Grade 3 adverse events. The pharmacodynamic effects and the preclinical data support further clinical development of ADC-1013 as mono- or combination therapy with e.g. PD-1/PD-L1 targeting antibodies. Janssen is currently developing ADC-1013/JNJ-64457107 for intravenous administration in patients with solid tumors.

025 Phase I trial of IL-15 superagonist ALT-803 (IL-15N72D:IL-15RαSu/IgG1 Fc complex) in advanced solid tumors: tolerability and correlates of activity

Kim Margolin1, Chihiro Morishima2, Martin A. Cheever2, Andreanne Lacroix2, Steve Fling2, Shernan Holtan3, Monica Jones4, Judy Kaiser2, Sylvia Lee5, Jeffrey S. Miller3, Peter Rhode4, Amy Rock4, Ann W. Silk6, Vamsi Velcheti7, Hing C. Wong4, Marc S. Ernstoff8

1City of Hope, Duarte, CA, USA; 2Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 3University of Minnesota, Minneapolis, MN, USA; 4Altor BioScience, Miramar, FL, USA; 5University of Washington, Seattle, WA, USA; 6Rutgers Cancer Institute of New Jersey, New Brunswick, NJ, USA; 7Cleveland Clinic, Cleveland, OH, USA; 8Roswell Park Cancer Center, Buffalo, NY, USA
Correspondence: Kim Margolin (kmargolin@coh.org)

Background

IL-15, a common gamma chain (γc) cytokine, stimulates expansion and cytotoxic functions of NK and memory CD8+ T cells. Compared to unmodified IL-15, ALT-803 had superior antitumor activity, low toxicity and favorable pharmacokinetics in animal models. These promising data prompted this phase I trial of ALT-803 administered intravenously (i.v.) or subcutaneously (s.c.) to patients with advanced cancers.

Methods

ALT-803 was administered at doses of 0.3-20 mcg/kg weekly i.v. or s.c. for 4 consecutive weeks, every 6 weeks until toxicity or progression. Correlates included pharmacokinetics, immunogenicity, lymphocyte expansion and flow cytometric phenotyping. Clinical endpoints were toxicity and anti tumor activity.

Results

Eleven patients received i.v. ALT-803 (range 0.3 to 6 mcg/kg), and 13 patients received s.c. ALT-803 (range 6 to 20 mcg/kg) in 9 dose cohorts. Ten had melanoma, 7 renal, 3 head and neck, and 4 lung cancer. The most common dose-related toxicity was injection site reaction with s.c. ALT-803; biopsy showed inflammatory infiltrates. Fatigue, fever and nausea were the next most common adverse events. Maximal increases in white blood cell and lymphocyte counts during cycle 1 were modest across all doses, with means of 1.2-fold (1.2-fold at 20 mcg/kg s.c.) and 1.5-fold (1.4-fold at 20 mcg/kg s.c.) over baseline. NK cell expansion typically peaked on cycle 1 day 15 or 29, with a mean maximal 2.7-fold increase across all doses, and 3.3-fold for the highest (20 mcg/kg) s.c. dose cohort. The expansion of circulating CD8+ T cells was modest (mean maximal 1.4-fold increase), but the mean fold increase for the % positivity of activation marker HLA-DR on CD8+ T cells was 2.9-fold across all dose cohorts and 4.7-fold at 20 mcg/kg s.c., typically peaking on cycle 1 day 4. Immunogenicity assays demonstrated no reactivity in patients. Clinical activity was not observed in these heavily-pretreated patients, but combination therapy trials are ongoing. Toxicities were mild, non-dose-limiting and similar to IL-2 and other γc cytokines. One patient experienced symptomatic diffuse cardiomyopathy that resolved slowly after drug discontinuation and corticosteroid treatment.

Conclusions

The IL-15 superagonist ALT-803 was well-tolerated with minimal cytokine toxicities in patients with advanced cancer. Substantial increases in NK cells and rapid activation of CD8+ T cells as evidenced by increased HLA-DR expression are promising immunologic effects, providing a strong rationale for combination anticancer immunotherapy with ALT-803.

Trial Registration

NCT01946789

026 Metabolic adaptations establish immunotherapy resistance in melanoma

Ashvin R. Jaiswal1, Shivanand Pudakalakatti2, Prasanta Dutta2, Liu Arthur2, Todd Bartkowiak2, Casey Ager2, Cristina Ivan2, Michael A Davies2, James P Allison2, Richard Eric Davis2, Jennifer Wargo2, Pratip K Bhattacharya2, David Hong2, Michael A. Curran2

1UT MD Anderson Cancer Center, Houston, TX, USA; 2The University of Texas MD Anderson Cancer Center, Houston, TX, USA
Correspondence: Ashvin R. Jaiswal (arjaiswal@mdanderson.org); Michael A. Curran

Background

Despite the success of T cell checkpoint blockade antibodies in treating an array of cancers, the majority of patients still fail to respond to these therapies, or respond transiently and then relapse. The molecular mechanisms which drive lack of response to checkpoint blockade, whether pre-existing or evolved on therapy, remain unclear.

Methods

To address this critical gap in clinical knowledge, we established a mouse model of melanoma designed to elucidate the molecular mechanisms underlying immunotherapy resistance. Through multiple in vivo passages, we selected a B16 melanoma tumor line that evolved complete resistance to combination blockade of CTLA-4, PD-1, and PD-L1, which cures ~80% of mice of the parental tumor. Using gene expression analysis, proteomics, and immunogenomics, we determined the adaptations engaged by this melanoma to become completely immunotherapy resistant. NMR spectroscopy, Seahorse XF Analysis, flow cytometry, confocal microscopy and western blot analysis provided further insight into the mechanisms driving checkpoint blockade resistance.

Results

Acquisition of immunotherapy resistance by these melanomas was driven by coordinate upregulation of the glycolytic and aldose reductase pathways to create a metabolically hostile microenvironment in which T cell function is profoundly suppressed. When re-introduced into the parental tumor, the genes most closely associated with these metabolic adaptations confer enhanced immunotherapy resistance. We have validated upregulation of these pathways in a unique cohort of melanoma patients who failed dual checkpoint blockade. Additionally, we employed MRI imaging to visualize metabolic changes acquired by resistant tumors in live mice. Clinical application of this technique could provide a much-needed non-invasive tool to predict immunotherapeutic sensitivity of patients.

Conclusions

Upregulation of glycolytic metabolism and the aldose reductase pathway by melanoma tumor cells cripples T cells in the microenvironment and confers resistance to checkpoint blockade.

027 Single cell RNA sequencing reveals mechanisms of Merkel Cell Carcinoma escape from intense pressure of T cell immunotherapy

Kelly Paulson1, Valentin Voillet1, Maurizio Perdicchio1, Daniel Hunter1, William Valente1, Felecia Wagener1, Samson Koelle2, Candice Church3, Natalie Vandeven3, Hannah Thomas3, Aric Colunga3, David Koelle3, Cassian Yee4, Rima Kulikauskas3, Robert Pierce1, Philip Greenberg1, Shailender Bhatia1, Jason Bielas1, Raphael Gottardo1, Paul Nghiem3, Aude Chapuis1

1Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 2Fred Hutchinson Cancer Research Division, Seattle, WA, USA; 3University of Washington, Seattle, WA, USA; 4MD Anderson, Houston, TX, USA
Correspondence: Aude Chapuis (achapuis@fredhutch.org)

Background

80% of Merkel cell carcinomas (MCCs) are driven by viral (MCPyV) oncoproteins, rendering them ideal immunotherapy targets. Transfer of MCC-specific T cells alone is insufficient, and half of patients receiving anti-PD1 monotherapy progress. We hypothesized co-transferring MCPyV-specific T cells with checkpoint blockade may enhance efficacy and elucidate interactions between T cells, tumor and the microenvironment. A patient with a dramatic response to T cells plus checkpoint inhibition, whose MCC eventually progressed despite robust T cell persistence and MCPyV expression, was examined by single-cell RNA sequencing (scRNAseq) to identify the mechanisms of response then immunoevasion.

Methods

A 59-year-old man with heavily-pretreated, MCPyV-positive MCC (>5 prior therapies including pembrolizumab) received autologous ex-vivo expanded, HLA B*3502-restricted MCPyV-specific CD8+ T cells (CD8+s), with sequential addition of pembrolizumab and ipilimumab. A 90% tumor reduction was sustained for ~1 year, followed by progression. ScRNAseq was performed on PBMCs and tumor collected before treatment, during response and at progression. Results were compared to flow cytometry and multicolor-IHC.

Results

scRNAseq of PBMCs identified a cluster of intensely-activated CD8+s present exclusively at the time of immunotherapy response. Although flow cytometry detected MCPyV-specific CD8+s, it was unable to discriminate this proliferative subpopulation. Activated CD8+s in blood corresponded with HLA-DR+CD8+ T cell infiltration into the shrinking tumor (including infused cells) and supported a causative role of CD8+s in mediating MCC regression.

scRNAseq of tumor digests was performed to determine the mechanism of acquired immunotherapy resistance. Despite an input of only 10000 cells, the expression profiles of tumor, TIL, macrophages, and fibroblasts could be distinguished. TIL and macrophages expressed transcriptomes that were stable in the pre-immunotherapy and post-resistance samples, suggesting these cells were not mediating the acquired resistance. In contrast, tumor cells displayed strikingly different transcriptional profiles before and after development of resistance. Directed analysis revealed tumor-specific loss of only the targeted HLA-B, but not non-targeted HLA-A, at time of progression. This implies intense immunologic and selective pressure from the transferred HLA-B*3502-restricted CD8+s. Unbiased scRNAseq analyses identified significant upregulation of HLA-E, which is an inhibitory signal for NK cells, and helps explain NK avoidance in the context of HLA-loss.

Conclusions

With small cellular inputs, scRNAseq can provide mechanistic insights beyond traditional IHC or flow cytometry, here revealing allele-specific HLA downregulation and HLA-E induction by the escaping tumor. Such insights can guide design of improved cellular therapies, in this case suggesting the use of multiple specificities of T cells.

Trial Registration

NCT0175845

O28 Functional correlation of increased tumor intrinsic glycolytic activity with resistance to adoptive T cell therapy

Tina Cascone1, Jodi Alicia McKenzie1, Rina Mbofung1, Simone Punt1, Zhe Wang1, Chunyu Xu1, Leila Williams1, Zhiqiang Wang1, Christopher Bristow1, Alessandro Carugo1, Michael Peoples1, Lerong Li1, Tatiana Karpinets1, Shruti Malu1, Caitlin Creasy1, Sara Leahey1, Jiong Chen1, Chantale Bernatchez1, Y. N. Vashisht Gopal1, Timothy P. Heffernan1, Jianhua Hu1, Jing Wang1, Rodabe N. Amaria1, Levi A. Garraway2, Ignacio I. Wistuba1, Scott E. Woodman1, Jason Roszik1, R. Eric Davis1, Michael A. Davies1, John V. Heymach1, Patrick Hwu1, Weiyi Peng1

1The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2Dana-Farber Cancer Institute, Boston, MA, USA
Correspondence: Weiyi Peng (wpeng@mdanderson.org)

Background

Adoptive T cell therapy (ACT) has produced impressive responses in a subset of patients with advanced malignancies, particularly those with melanoma, where as many as 50% of patients may achieve an objective response. However, these responses tend to be transient and only present in a small portion of patients. Thus, there is an urgent need to understand the resistant mechanisms in non-responders and develop more effective ACT strategies that prevent emergence of resistance or reverse resistance once it occurs. Emerging evidence suggest that tumor intrinsic signaling pathways play a critical role in promoting resistance to T cell-mediated immunotherapy. For example, we recently reported that oncogenic activation of the PI3K pathway resulting from PTEN loss renders melanoma cells refractory to T cell-mediated therapies and that this could be reverted by targeted inhibition of PI3K signaling. However, work from our laboratory, and others, suggest that melanoma cells may have additional means to circumvent immunotherapy.

Methods

In this study, we employed two independent and unbiased approaches to identify novel molecular determinants of immune resistance. We generated gene expression profiles on PTEN-deficient melanoma cell lines from patients to identify alternative immunosuppressive mechanisms and then extended this finding to include non-small cell lung cancer. Then we utilized a new high-throughput shRNA screening platform developed by our group to functionally interrogate immune resistance in melanoma cells.

Results

Results from both analyses implicated tumor-associated glycolysis as a critical pathway that enables tumor cells to evade T cell-mediated antitumor activity. By using samples from melanoma and non-small cell lung cancer patients, we showed that increased expression of glycolysis-related genes is associated with poor T cell infiltration of tumors. In addition, we found that increased expression of ALDOA, which encodes a critical enzyme in the glycolysis pathway, was functionally correlated with reduced sensitivity of tumor cells to T cell-mediated killing. Overexpression of ALDOA impaired T cell killing of patient-derived melanoma cells, while inhibiting glycolysis restored T cell-mediated apoptosis of tumor cells. More importantly, when we characterized the patterns of gene expression and generated bioenergetic profiles on tumor samples from two non-overlapping ACT-treated patient cohorts, we discovered that tumor glycolytic activity in patients who experienced disease progression following ACT was significantly higher compared to those patients who were responsive to therapy.

Conclusions

Taken together, our results demonstrate that tumor glycolytic metabolism is associated with the efficacy of ACT and identify glycolysis as a candidate target for combinatorial therapeutic intervention.

O29 A dual and opposing role for tumor intrinsic type-II interferon sensing in tumor establishment and acquired resistance

Jason Williams, Shuyin Li, Thomas Gajewski, Haochu Huang1

University of Chicago, Chicago, IL, USA
Correspondence: Jason Williams (jaybwilliams1@uchicago.edu)

Background

Immunotherapy can augment the host anti-tumor response inducing tumor control and regression. However, in a subset of patients immune selective pressure can drive the outgrowth of resistant tumor cell clones. An understanding of the altered genetic pathways in resistant tumor cell populations is critical to develop therapies to target these tumors.

Methods

To this end, we utilized a whole genome CRISPR library screen to delete genes in B16.SIY cells and selected clones that gained resistance to T cell-mediated killing. One set of mutations was in genes involved in the type-II interferon pathway, namely IFNγR2 and Jak1. To test the role of these genes in subverting the host immune response we generated B16.SIY cells selectively lacking IFNγR2 or Jak1.

Results

In contrast to resistance to T cell-mediated killing in vitro, when implanted into mice, these mutant tumors paradoxically were better controlled in vivo. This phenotype was observed with multiple IFNγR2- and Jak1-deleted B16.SIY lines and using different guide-RNAs, and also with a second tumor cell line MC38. Re-introduction of IFNγR2 reverted the phenotype and restored tumor growth. Depletion of host CD8+ T cells eliminated regression of mutant tumors, and increased accumulation of antigen-specific CD8+ T cells was observed in the tumor microenvironment. It seemed likely that IFN-γ produced by CD8+ T cells in the tumor microenvironment might have a dominant negative effect through upregulation of inhibitory factors such as PD-L1 on tumor cells. Indeed, IFNγR2–/– and Jak1–/– tumor cells expressed markedly reduced PD-L1 and IDO in vivo compared to WT tumors. Further, transduction to express PD-L1 restored the ability of IFNγR2–/– tumors to subvert the host immune response and grow progressively.

Conclusions

In summary, we find that the type-II signaling in tumor cells in some models can be required for the initiation and stabilization of the suppressive tumor microenvironment. Combined with other published data, our results suggest that whether IFN-γ on tumor cells is predominantly a positive or negative factor for tumor control may depend on whether inhibitory factors like PD-L1 are dominantly expressed and functional through tumor cells versus host immune cells.

O30 Diversity and composition of the gut microbiome influences responses to anti-PD1-therapy through beneficial changes in innate and adaptive immunity

Vancheswaran Gopalakrishnan1, Christine Spencer1, Luigi Nezi1, Alexandre Reuben1, Miles Andrews1, Tatiana Karpinets1, Diego Vicente1, Peter Prieto2, Kristi Hoffman1, Spencer Wei1, Alexandria Cogdill1, Erick Riquelme Sanchez1, Florencia McAllister1, Michael Wong1, Michael Davies1, Alexander Lazar1, Jeffrey Gershenwald1, Nadim Ajami3, Zachary Cooper4, Joseph Petrosino3, Carrie Daniel-MacDougall1, Michael Tetzlaff1, James Allison1, Padmanee Sharma1, Robert Jenq1, Jennifer Wargo1

1The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2University of Rochester Medical Center, Rochester, NY, USA; 3Baylor College of Medicine, Houston, TX, USA; 4MedImmune, Gaithersburg, MD, USA
Correspondence: Jennifer Wargo (jwargo@mdanderson.org)

Background

Cancer therapy has benefitted greatly from immune checkpoint inhibitors that target negative regulators of T-cell activation, however responses are variable and not always sustained. Recent evidence in murine models suggests that the gut microbiome can be modulated to enhance responses to immune checkpoint blockade. However, this has not been deeply investigated in patients.

Methods

We collected buccal and stool samples from a cohort of anti-PD-1 treated metastatic melanoma patients (n=112). Patients were classified as either responders (R) or non-responders (NR) based on RECIST criteria (R=CR, PR, or SD>6 months, NR= PD or SD<6 months). 16S rRNA, and whole-genome shotgun sequencing were performed to characterize the diversity, composition and functional capabilities of the microbiome. Immune profiling (via immunohistochemistry and flow cytometry) was performed in available tumors and serum samples at baseline. Fecal microbiota transplants (FMT) from R versus NR patients were performed in germ-free (GF) mice, tumors were implanted and markers of systemic and anti-tumor immunity queried.

Results

We observed significant differences in the diversity and composition of the gut microbiome in R versus NR to anti-PD-1 therapy at baseline with no clear differences in the oral microbiome. Specifically, R had a higher alpha diversity compared to NR (11.2 versus 6.4, p<0.01), and an enrichment of the Faecalibacterium genus of the Ruminococcaceae family. In contrast, the Bacteroidales order was found to be enriched in NR. High abundance of Faecalibacterium was also associated with improved progression-free survival (HR=2.92. 95% C.I= 1.02 -12.52) Immune profiling demonstrated significantly increased immune infiltrates in baseline tumor samples of R versus NR (p=0.04), with a positive correlation between CD8+ T-cell density in baseline tumor and blood samples and abundance of beneficial Ruminococcaceae bacteria in the gut microbiome (r2=0.42, p<0.01). Distinct metabolic signatures were also observed in the gut microbiome with synthetic processes predominating in R and degradative processes predominating in NR. In FMT studies, GF mice transplanted with R-stool demonstrated delayed tumor outgrowth and enhanced responses to anti-PD-L1 therapy (p<0.01). Immune profiling in these mice revealed a higher density of innate and adaptive immune infiltrates in the tumor and gut, and upregulation of PD-L1 in the tumor microenvironment.

Conclusions

Diversity and composition of the gut microbiome was associated with differential responses to anti-PD-1 therapy, and enhanced systemic and anti-tumor immunity in metastatic melanoma patients. Our results have far-reaching implications and suggest that modifying the gut microbiome could potentially enhance therapeutic responses to immune checkpoint blockade.

O31 Preclinical characterization of a novel STING agonist, MK-1454

Saso Cemerski1, Michael Altman1, Madhavi Bandi1, Wonsuk Chang2, Yiping Chen1, Long Cui1, Jared Cumming1, Lauren Dorosh1, Lauren Flateland1, Yanhong Ma1, Sarah Javaid1, Thu Ho3, Johnny Kopinja1, Jason Laskey1, Ellen Minnihan1, Laura Price3, Robin Mogg3, Gregory O'Donnell3, Bo-Sheng Pan1, Samanthi Perera1, Jennifer Piesvaux1, Jeremy Presland1, Larissa Rakhilina1, Gottfried Schroeder1, Michael Slavonic1, Steven Smock1, Benjamin Wesley Trotter1, Daniel Wyss2, Qing Zhao4, Shuxia Zhao1, Svetlana Sadekova5, Ying Yu5, Anuradha Khilnani4, Archie Tse4

1Merck Research Laboratories, Boston, MA, USA; 2Merck Research Laboratories, Kenilworth, NJ, USA; 3Merck Research Laboratories, West Point, PA, USA; 4Merck Research Laboratories, Rahway, NJ, USA; 5Merck Research Laboratories, Palo Alto, CA, USA
Correspondence: Saso Cemerski (saso.cemerski@merck.com)

Background

Several innate immune danger-sensing pathways have emerged as promising targets for enhancing cancer immunotherapy. In recent years, cyclic dinucleotide agonists of STING (Stimulator of interferon genes) have demonstrated significant anti-tumor efficacy in several mouse syngeneic tumor models and set the stage for STING agonists to be evaluated in human clinical trials.

Methods

Binding of STING agonist MK-1454 to mouse and human STING was evaluated using Biacore. STING activity was monitored in human whole blood and PBMCs, in multiple primary mouse and human cells, cell lines and differentiated myeloid cell subsets using cytokine production and cell activation as readouts. In vivo efficacy studies of intratumorally-dosed MK-1454 were performed in the presence or absence of an anti-PD-1 antibody in MC38, CT26 and B16F10 syngeneic tumor models. Characterization of MK-1454 activity in primary human tumors was assessed in human histoculture studies. Gene profiling and STING gene signature studies were performed using custom-built Nanostring reagents.

Results

In vitro studies demonstrated potent binding of MK-1454 to human and mouse STING and rapid induction of type-I interferon and proinflammatory cytokine production in human and mouse cell lines and primary cells in a STING-dependent manner. Mechanistic studies were conducted to assess the effect of STING agonism across multiple myeloid cell subsets including monocytic MDSCs, M2 macrophages and TAMs, which have been previously suggested to represent the suppressive myeloid component of the TME. In addition, we describe important effects of STING agonism on mouse and human T cells and highlight some key species-related differences. Critically, we demonstrate that human primary tumors, stimulated ex vivo with MK-1454, respond robustly as indicated in cytokine production and gene profiling studies. Additional studies using in vitro-stimulated human whole blood were used to define a STING gene signature that could represent a valuable biomarker in the clinic. In multiple mouse syngeneic tumor models, MK-1454 induced strong anti-tumor responses, effecting full tumor eradication upon intratumoral dosing, either as a single agent or in combination with PD-1 blockade, depending on the tumor model and dose level. We also show that MK-1454 induced tumor-specific adaptive immune memory as demonstrated by the lack of tumor growth in tumor re-challenge studies.

Conclusions

MK-1454, a novel STING agonist, induces potent cytokine responses and immune cell activation in vitro and robust anti-tumor activity in vivo. The preclinical data presented here support the ongoing clinical evaluation of MK-1454 in cancer patients both as monotherapy and in combination with Keytruda.

O32 Phase 1 trial of CA-170, a first-in-class, orally available, small molecule immune checkpoint inhibitor (ICI) dually targeting PD-L1 and VISTA, in patients with advanced solid tumors or lymphomas

Yung Jue Bang1, John Powderly2, Manish Patel3, Kyu Pyo Kim4, James Lee5, Joshua Brody6, Javier Garcia-Corbacho7, Sun Young Rha8, Funda Meric-Bernstam9, Erika Hamilton10, Santiago Ponce Aix11, Radhakrishnan Ramchandren12, Jeffrey Sosman13, Myung Ju Ahn14, Mario Snozl15, Gerald Falchook16, Marta Gil Martin17, Hongwei Wang18, Lisa Adams18, Christopher Caldwell18, Adam Lazorchak18, Timothy Wyant18, Anna Ma18, David Tuck18, Adil Daud19

1Seoul National University Hospital, Seoul, Republic of Korea; 2Carolina BioOncology, Huntersville, NC, USA; 3Florida Cancer Specialists & Research Institute, Sarasota, FL, USA; 4Asan Medical Center, Seoul, Republic of Korea; 5University of Pittsburgh School of Medicine UPMC Cancer Pavilion, Pittsburgh, PA, USA; 6Icahn School of Medicine at Mount Sinai, New York, NY, USA; 7Hospital Clinic i Provincial, Barcelona, Spain; 8Yonsei University Health System - Severance Hospital, Seoul, Republic of Korea; 9UT MD Anderson Cancer Center, Houston, TX, USA; 10Sarah Cannon Research Institute - TN, Nashville, TN, USA; 11Hospital Universitario 12 de Octubre, Madrid, Spain; 12Karmanos Cancer Institute, Detroit, MI, USA; 13Northwestern University Research, Evanston, IL, USA; 14Samsung Medical Center, Seoul, Republic of Korea; 15Smilow Cancer Center, New Haven, CT, USA; 16Sarah Cannon Research Institute at HealthONE, Denver, CO, USA; 17Catalan Institute of Oncology, Bacelona, Spain; 18Curis Inc., Lexington, MA, USA; 19University of California San Francisco (UCSF), San Francisco, CA, USA

Background

Programmed-death 1 (PD-1) and V-domain Ig suppressor of T-cell activation (VISTA) are independent immune checkpoints that negatively regulate T-cell function and are implicated in various malignancies. Pre-clinical studies demonstrated that dual blockade of both checkpoints can be synergistic. CA-170 is a novel oral small molecule ICI that directly targets both PD-L1/L2 and VISTA and demonstrated significant anti-tumor activity in multiple preclinical animal models.

Methods

Enrollment initially followed accelerated titration and subsequently switched to 3+3 design. Cohorts of selected dose levels were expanded with additional patients. The expansion phase allows for enrollment of patients with selected indications known to be sensitive to ICI. Primary objectives: safety, maximum tolerated dose (MTD) and recommended Phase 2 dose. Secondary objectives: pharmacokinetics (PK) and anti-tumor activity. Exploratory endpoints: biomarkers and pharmacodynamic (PD) effects in peripheral blood and tumor tissues.

Results

A total of 34 patients have been treated across 6 dose levels (50 -800 mg) with 28 evaluable for safety and 4 still ongoing in cycle 1. Enrolled tumor types include lung, ovarian, lymphoma, head and neck, renal cell carcinoma, etc.

No DLTs or ≥ grade 3 drug-related adverse events (AEs) have been observed thus far. The most common treatment-emergent AEs (all grades) were fatigue (24%), gastritis (20%), nausea (20 %), vomiting (20%), chills (16%), constipation (16%) and headache (16%). These were predominantly grade 1 and self-limiting events.

Seven newly enrolled patients are pending restaging. Nineteen patients were evaluable for anti-tumor activity with 12 showing stable disease and 4 on study for > 6 cycles. An ICI naïve melanoma patient treated at 600 mg had lesion shrinkage by 14% per RECIST and 23% per immune-related response criteria by the end of cycle 2 and is currently ongoing on treatment.

CA-170 exhibits approximately dose proportional plasma exposure with T1/2 5 – 12 hours. Evidence of peripheral T cell activation was observed with an increased proportion of circulating CD8+ and CD4+ T cells expressing activation markers. Additionally, preliminary tumor biopsy data has shown increases in the number of immune cells post-treatment versus pre-treatment.

Conclusions

The data suggest CA-170 has an acceptable safety profile with preliminary signs of anti-tumor activity and peripheral immune modulation and approximately dose proportional PK profile. MTD has not been reached. These data warrant the continued clinical development of CA-170. Dose escalation is ongoing. Expansion cohorts in selected indications are planned.

Trial Registration

Clinical trial identifier: NCT02812875.

O33 Variance from evidence-based management of immune-related adverse events among healthcare providers: analysis of an online management decision tool

Krista Marcello1, Kevin L. Obholz1, Timothy A. Quill1, Jeffrey S. Weber2

1Clinical Care Options, LLC, Reston, VA, USA; 2NYU Langone Medical Center, New York, NY, USA
Correspondence: Krista Marcello (kmarcello@clinicaloptions.com)

Background

Immune checkpoint inhibitors (ICIs) are altering the treatment paradigm in oncology, impacting the care of a rapidly increasing number of patients; but many healthcare providers (HCPs) remain unfamiliar and inexperienced with managing the unique spectrum of immune-related adverse events (irAEs). An online decision support tool was developed to give HCPs easy access to fair-balanced guidance and management algorithms that are patient-specific and extend beyond standard recommendations found in the respective product inserts [1].

Methods

To use the tool, HCPs entered patient symptoms, organ system affected, and the grade or severity of the event along with their planned management strategy. Subsequently, the HCPs using the tool were provided an expert recommendation for management of that specific irAE. Recommendations in the tool comprised algorithms developed from peer-reviewed publications and the personal clinical experience of Jeffrey S. Weber, MD, PhD. Here, we report a comparison of the intended irAE management of HCPs for 2896 cases entered into the tool from 11/9/16 through 7/21/17 with the expert recommendations for those patient cases.

Results

Cases involving GI symptoms (n = 818; 28% of all cases) were most frequently entered into the tool with renal symptom cases (n = 153; 5% of all cases) having the fewest entries (Table 1). The planned irAE management strategy of HCPs differed from the expert recommendations for 49% of the cases overall with the greatest divergence in neurologic- and endocrine-related cases (56% and 66%, respectively). The proportion of cases in which the planned management strategies of the HCPs differed from the expert recommendations also varied by symptom grade/severity ranging from 84% differing for grade 3 adrenal insufficiency to 21% differing for grade 4 renal toxicity. Overall, the largest proportion of cases with HCP variance from optimal recommended practice occurred with those having intermediate to high symptom severity (grades 2/3). Overall, of 648 HCPs who responded to a survey about the impact of the tool on their practice, 93% indicated that the recommendations provided by the tool either changed or confirmed their management plan.

Conclusions

These data suggest that many HCPs are not optimally managing irAEs associated with ICI use and that an online tool can provide a resource to improve patient care and safety. A detailed analysis of the tool, including case entries and planned management vs best practice recommendations for each irAE and grade will be presented.

Reference

1. Clinical Care Options (US). Managing Immune-Related Adverse Events: An Interactive Algorithm Tool [Internet]. Clinical Care Options Oncology (US); 2017 May. Available from: www.clinicaloptions.com/immuneAEtool
Table 1 (abstract O33).

See text for description

O34 Cross-dressed dendritic cells drive anti-tumor immunity

Brendan MacNabb, Douglas Kline1, Justin Kline

The University of Chicago, Chicago, IL, USA
Correspondence: Brendan MacNabb (bwm@uchicago.edu)

Background

Dendritic cells (DCs) play a central role in the orchestration of anti-tumor T cell responses. The presentation of tumor-derived peptide antigens on class I major histocompatibility complex molecules (MHC I) by DCs is essential for the activation of antigen-specific CD8+ T cells. Emerging evidence has revealed that migratory Batf3-lineage DCs play a dominant role in the priming of anti-tumor CD8+ T cell responses against solid tumors. However, the cellular mechanisms of tumor antigen presentation by these DCs have not been formally demonstrated. Batf3-lineage DCs are adept at cross-presenting exogenous antigens on MHC I molecules, and it has been presumed that this antigen presentation pathway is critical in anti-tumor immunity.

Methods

To determine the contribution of cancer cell-derived MHC I on antigen-specific CD8+ T cell priming, we deleted H2-Kb in the murine acute myeloid leukemia cell line, C1498, using CRISPR/Cas9. Engraftment of Kb+/+ and Kb-/- C1498 cells into syngeneic C57BL/6 mice, as well as those lacking TAP-1 (a critical component of the classical cross-presentation pathway), and all MHC I (Kb/Db-/-), allowed us to interrogate the importance of different antigen presentation pathways to the anti-tumor CD8+ T cell response.

Results

Consistent with published data, antigen-specific CD8+ T cell priming against localized C1498 tumors was entirely dependent upon Batf3-lineage migratory DCs, revealing no role for direct priming by C1498 cells. Surprisingly, the activation of endogenous and adoptively-transferred tumor-specific CD8+ T cells was absent in tumor-draining lymph nodes of mice harboring localized C1498 Kb-/- tumors, suggesting that antigen cross-presentation by DCs was insufficient to activate anti-tumor CD8+ T cells. Conversely, cancer-derived MHC I was dispensable for antigen recognition by CD8+ T cells in spleens of mice with disseminated leukemia. Further, tumor-derived MHC I was abundantly observed on the surface of DCs isolated from the solid tumor micro-environment, and on migratory DCs in the tumor-draining lymph node. However, no cancer-derived MHC I was found on splenic DCs in leukemia-bearing mice, in which antigen-specific CD8+ T cell responses occurred in a TAP-dependent fashion.

Conclusions

Together, these observations support a model in which the CD8+ T cell response to an MHC I-expressing, localized cancer is driven by Batf3-lineage DCs cross-dressed with intact tumor-derived peptide-MHC I molecules. Furthermore, our data reveal that the processing and presentation of tumor antigens occurs in a fundamentally different manner for solid and hematologic malignancies. These findings could have broad implications both for basic immunology research, and for the development of DC-oriented tumor vaccines.

Poster presentations

Biomarkers and Immune Monitoring

P1 Characterization of the tumor microenvironment and immune profile in non-small cell lung cancer

Andreas Ackermann1, Hanny Musa1, Johannes Zimmermann2, Felix Segerer2, Moritz Widmaier2, Mindaugas Andrulis3
1Boehringer Ingelheim Pharma GmbH & Co. KG, 88397 Biberach an der Riss, Germany; 2Definiens AG, 80636 Munich, Germany; 3Klinikum Ludwigshafen, 67063 Ludwigshafen am Rhein, Germany
Correspondence: Andreas Ackermann (andreas.ackermann@boehringer-ingelheim.com)

Background

Immunotherapies with programmed death receptor-1/programmed death-ligand 1 (PD-1/PD-L1)-blocking antibodies have shown encouraging results in patients with advanced NSCLC, HNSCC, RCC, Melanoma, Hodgkin Lymphoma and other cancer types. Despite a subset of patients exhibiting durable and long lasting benefit, the majority of patients show no signs of efficacy. The identification of the right patient population is often challenging, and although there is an incremental understanding of tumor characteristics and biomarkers bearing limited predictive value, such as PD-L1, there is a clear need for an accelerated development of better predictive biomarkers. This knowledge may permit the guidance of therapy-relevant decisions and potentially increase the clinical benefit of CPI-related anti-cancer interventions.

Methods

To broaden our understanding of the spatial expression of relevant biomarkers in the tumor microenvironment and gain deeper insights into immune-stimulatory mechanisms, we performed histopathological analysis, image analysis, and gene expression analysis in a complementary approach. In thirty (n=30) NSCLC tissues the expression patterns of pre-selected biomarkers (CD8 and PD-L1) were conventionally evaluated by a board-certified pathologist and the obtained H-scores were compared with Image Analysis (IA)-derived i-scores. Subsequently, gene expression data for all 30 NSCLC samples were generated in order to complement the histopathological evaluations with mRNA expression patterns.

Results

Computer-based image analysis algorithms for the quantification of CD8 and PD-L1 have been established showing a very high accuracy and concordance compared to conventional histopathological evaluations performed by trained pathologists. The complementary analysis based on image analysis, conventional histopathological evaluation, and gene expression analysis revealed excellent/good agreement. Based on the attained knowledge from the multi-modal approach the image analysis tools were further optimized. Following optimization, preliminary data has demonstrated that the optimized image analysis algorithms and the application of virtual multiplexing approaches permits characterization of samples into four types of tumor microenvironment sensu Teng et al [1]. Subsequently, a further seventy (n=70) NSCLC samples will be interrogated using the optimized image analysis tools and correlated, where applicable, with gene expression analysis.

Conclusions

The insights gained during this complementary analysis will help us to design and implement advanced biomarker investigations in future clinical trials to support the co-development of valuable biomarkers in the field of cancer immunotherapy. The application of advanced image analysis and virtual multiplexing approaches may be suitable vehicles to better characterize the tumor microenvironment, honing precision medicine within the immune-oncology field.

References

1. Teng, et al. Classifying cancers based on T cell infiltration and PD-L1. Cancer Res. 2015; 75: 2139-45.

P2 ADXS-PSA immunotherapy increases the magnitude and quality of prostate cancer antigen-specific T cell responses in patients with metastatic castration-resistant prostate cancer

Sandra M. Hayes1, Robert G. Petit1, Lawrence Fong2, Mark Stein3, Ronald Tutrone4, Anthony Mega5, Naomi Haas6
1Advaxis Immunotherapies, Inc., Princeton, NJ, USA; 2UCSF University of California, San Francisco, CA, USA; 3The Cancer Institute of New Jersey CINJ Rutgers, Inc., New Brunswick, NJ, USA; 4Chesapeake Urology Research Associates, Towson, MD, USA; 5Lifespan Oncology Clinical Research, Providence, RI, USA; 6University of Pennsylvania Abramson Cancer Center, Philadelphia, PA, USA
Correspondence: Ronald Tutrone (ahmad@advaxis.com)

Background

Active immunotherapies, such as ADXS-PSA, are designed to generate tumor antigen-specific T cell effectors that recognize and kill tumor cells. ADXS-PSA, a highly attenuated Listeria monocytogenes-based immunotherapy that targets prostate-specific antigen (PSA), is currently being evaluated as a treatment for metastatic castration-resistant prostate cancer (mCRPC) in the phase 1/2 KEYNOTE-046 trial as a monotherapy (Part A) and in combination with KEYTRUDA® (Part B). Because tumor antigen-specific T cell responses may be linked to the clinical efficacy of active immunotherapies, we quantified the frequency of functional prostate cancer antigen-specific T cells in the peripheral blood of ADXS-PSA-treated mCRPC patients by ELISpot analysis.

Methods

ELISpot assays were performed on peripheral blood mononuclear cells (PBMCs) isolated at multiple time points from 9 mCRPC patients who received 3 doses of ADXS-PSA monotherapy in Part A, the ADXS-PSA dose-determining stage, of the KEYNOTE-046 trial (NCT02325557). In addition to peptides from PSA, CD4+ and CD8+ T cell reactivity was assayed to peptides from prostatic acid phosphatase (PAP), prostate-specific membrane antigen (PSMA), prostate stem cell antigen (PSCA), and prostein to determine the extent of antigen cascade/spreading after ADXS-PSA treatment. Secretion of IFNγ, TNFα, and the cytolytic granule granzyme B were assayed simultaneously to examine the quality (i.e. degree of multi-functionality) of the T cell responses.

Results

During a 9-week treatment course, which included 3 ADXS-PSA doses, 7/9 patients exhibited increases in the magnitude of the PSA-reactive T cell response, including a patient with undetectable PSA-reactive T cells at baseline. Notably, during the same time course, increases above baseline in the frequency of T cells reactive to one or more of the other prostate cancer antigens was observed in all 9 patients. In addition to increases in the frequency of prostate cancer antigen-specific T cells, all 9 patients exhibited increases in the quality of T cells reactive to at least 1 of the 5 prostate cancer antigens, indicating that ADXS-PSA induces multifunctional T cell responses to a broad range of prostate cancer antigens.

Conclusions

These results demonstrate that ADXS-PSA monotherapy increases the magnitude and quality of T cell responses specific not only to PSA, its target antigen, but also to other prostate cancer antigens, which is indicative of antigen cascade/spreading. Patient enrollment is ongoing for Part B, in combination with KEYTRUDA®.

Trial Registration

ClinicalTrials.gov Identifier NCT02325557

P3 Persistence of 6-thioguanine-resistant T cell clones in a melanoma patient with durable antitumor response following treatment with immune checkpoint blockade

Cindy Zuleger1, Michael Newton1, Irene Ong1, Mark Albertini1,2
1Carbone Cancer Center at the University of Wisconsin School of Medicine and Public Health, Madison, WI, USA; 2William S. Middleton Memorial Veterans Hospital, Madison, WI, USA
Correspondence: Mark Albertini

Background

Ex vivo selection of T cells resistant to 6-thioguanine can be used to identify T cells with in vivo mutation in the hypoxanthine-guanine phosphoribosyltransferase (HPRT) gene. We hypothesize that in vivo HPRT- mutant T cells (MT) will enrich for in vivo proliferating T cells and provide a predictive biomarker for melanoma patients treated with immune checkpoint blockade.

Methods

This study examined clonotypic diversity and T cell persistence in peripheral blood mononuclear cells (PBMC) freshly cryopreserved (T0) and in PBMC expanded as mass cultures with (MT) or without (wild-type (WT)) 6-thioguanine selection in a metastatic melanoma patient who achieved a durable antitumor response following treatment with intratumoral α-gal glycolipid injections (IT-AG) followed by Ipilimumab (Ipi). IT-AG (0.1 mg/injection every 4 weeks for 2 doses) was initially administered, and Ipi (3 mg/kg (capped at 125 kg)/dose every 3 weeks for 4 doses) was then administered 4 weeks later at a time of disease progression. Blood samples were obtained before and 4-weeks after IT-AG as well as 1-, 4-, and 13-months post-Ipi. TCR beta chain (TRB) repertoire of T0, MT and WT were examined via 5’ RACE and Illumina MiSeq sequencing. TRB repertoire of T cells from a tumor sample obtained ~20 months prior to IT-AG (T-T0) was also evaluated. TRB were collapsed to identical nucleotide sequences to enumerate T cell clones.

Results

Expanded or activated T cell clones, defined by sequences with identical TRB, were detected in T-T0 as well as in T0, WT, and MT at each of the 5 time points. The range of TRB sequences was as follows: T0: 47,602-58,588 (median 54,284); WT: 49,799-58,896 (median 52,277); MT: 35,527-58,217 (median 49,035), and 51,713 in T-T0. The range of distinct TRB amino acid sequences was as follows: T0: 4,682-9,644 (median 5,993); WT: 2,083-4,573 (median 2,702); MT: 84-182 (median 154), and 1850 in T-T0. Clonotypic diversity, measured as the number of unique TRB nucleotide sequences, was substantially reduced in MT compared to either WT or T0 at each time point. Numerous TRB matches were observed between MT and WT and T0, as well as between T-T0 and MT, WT and T0. The patient achieved a durable antitumor response, and several TRB persisted in this patient over a four-year span. Some TRB from this study matched TRB from the literature with defined melanoma specificity.

Conclusions

MT from melanoma patients are enriched for proliferating and/or activated T cells and merit additional study as a biomarker of response to immunotherapy.

P4 The prognostic and predictive roles of a CD8/PD-L1 signature in cancer

Sonja Althammer1, Keith Steele2, Moritz Widmaier1, Charles Brown2, Tze Heng Tan1, Lorenz Rognoni1, Hélène Kaplon3, David Friedrich1, Brandon W Higgs2, Günter Schmidt1, Koustubh Ranade2, Marie-Caroline Dieu-Nosjean3
1Definiens, Munich, Germany; 2Medimmune, Gaithersburg, MD, USA; 3INSERM, UMRS 1138, Paris, France
Correspondence: Sonja Althammer (ttan@definiens.com)

Background

Predictive signatures to a-PD1 and a-PD-L1 immunotherapies are active research areas. We previously showed that non-small cell lung cancer (NSCLC) patients with high densities of CD8(+) and PD-L1(+) cells have improved outcome compared to those with high CD8(+) or PD-L1(+) cell densities alone [1]. The applicability of this signature to other indications such as urothelial carcinoma (UC) is uncertain. The potential prognostic contribution of this signature is also unknown.

Methods

The predictive potential of the CD8/PD-L1 signature in UC, was tested in baseline tumor samples obtained from 43 durvalumab-treated patients in a phase 1/2 clinical trial (NCT01693562).

To assess the prognostic significance in NSCLC, we analyzed surgically-resected tumors of 134 NSCLC patients that later received chemotherapy (non-IO). Tumors were immunostained for CD8 (SP239) and PD-L1 (SP263) using single stains (UC) or a dual immunohistochemistry (IHC) assay (NSCLC). Densities of CD8(+) and PD-L1(+) cells from IHC stained digitized slides were calculated by the Definiens Tissue Phenomics Platform for a total of 183 patient samples.

Results

UC patients with CD8/PD-L1(+) profiles showed improved survival compared to those with CD8/PD-L1(-) profiles (p<0.06), which may help to identify patients with beneficial outcome under α-PD-L1 treatment.

Further, non-IO treated NSCLC patients with CD8/PD-L1(+) profiles did not show improved OS over those with CD8/PD-L1(-) profiles, suggesting that the CD8/PD-L1 signature is predictive of therapeutic response to α-PD-L1 in NSCLC patients (Table 1).

Conclusions

Our results demonstrate the value of profiling the tumor and its immune-contexture by image analysis for discovery of prognostic and predictive signatures in NSCLC and UC.

Trial Registration

ClinicalTrials.gov Identifier NCT01693562

Reference

1. Althammer S, et al. Combinatorial CD8+ and PD-L1+ cell densities correlate with response and improved survival in non-small cell lung cancer (NSCLC) patients treated with durvalumab. Immunother Cancer. 2016; 4(Suppl 2):91
Table 1 (abstract P4).

Clinical Outcomes

P5 Analytical comparison of methods used to assess mismatch repair deficiency

Steven Anderson1, Li Cai2, Lori Johnson3, Bryan McCune2
1Covance, Durham, NC, USA; 2Integrated Oncology, Research Triangle Park, NC, USA; 3Covance, Morrisville, NC, USA
Correspondence: Steven Anderson (sanderson@labcorp.com)

Background

Deficiency in mismatch repair occurs in approximately 15-20% of sporadic colorectal cancer, and to a lesser degree in a variety of other solid tumors. Mismatch repair deficiency (dMMR) has been associated with disease prognosis and recently has also been linked with potential response to the checkpoint inhibitors pembrolizumab and nivolumab. Defects in the mismatch repair process can be identified by a number of laboratory procedures including immunohistochemical (IHC) methods to assess expression of the mismatch repair proteins (MLH1, MSH2, MSH6 and PMS 2), or by using PCR methods to assess replication fidelity of specific microsatellite sequences in the genome. Tumor samples are indicated as dMMR when one or more the MMR proteins are not expressed, or if the tumor tissue has a high level of microsatellite instability (MSI-H).

Methods

In this study we compared the performance of the IHC and PCR methods, to identify colon cancer tumors that demonstrate dMMR. A series of archival formalin-fixed paraffin-embedded samples were used in this comparison study. The IHC staining was performed on the Ventana Benchmark XT system with standard detection chemistry. The specific monoclonal antibody clones were for MLH1 (G168-228, Cell Marque), MSH2 (G219-1129, Cell Marque), MSH6 (44, Ventana) and PMS2 (EPR3947, Cell Marque). For the MSI PCR assay the following microsatellites were evaluated BAT-25, BAT-26, TGFβRII, D5S346, D17S250 and D2S123. The panel contains both dinucleotide and mononucleotide repeat sequences, and includes the loci recommended by the NCI. PCR fragments were detected by capillary electrophoresis and fragment analysis was used to determined MSI status. If > 2 loci exhibited evidence of replication defects the sample was considered MSI-H.

Results

In this cohort approximately 40% of the samples showed absence of tumor staining in either the MLH1/PMS2 or MSH2/MSH6 pathways. All cases demonstrating the MSI-H phenotype also demonstrated lack of expression of one of more MMR proteins. The most common pattern of expression was the loss of MLH1 and/or PMS2, which has been reported as being the most common phenotype in sporadic tumors. For the cases with MMR protein expression, all were considered MSI-stable by PCR.

Conclusions

This comparison study shows an equivalent performance of the two commonly used methods for the detection of mismatch repair deficiency. These data are relevant for selecting the appropriate methodology to assess dMMR when considering the use of checkpoint inhibitors for the treatment of solid tumors.

P6 Characterization of AB928, a dual adenosine A2aR/A2bR antagonist that retains potency under conditions of high albumin and high receptor activation

Devika Ashok, Lisa Seitz, Joanne Tan, Adam Park, Tim Park, Ferdie Soriano, Terry Rosen, Ehesan Sharif, Dillon Miles, Manmohan Leleti, Juan Jaen, Jay Powers, Matthew Walters, Steve Young
Arcus Biosciences, Hayward, CA, USA
Correspondence: Devika Ashok (dashok@arcusbio.com)

Background

In many tumors, extracellular adenosine contributes to an immunosuppressed micro-environment (TME) via activation of the A2a receptor, expressed on lymphocytes, and the A2b receptor, expressed on myeloid cells. Relative to other tissues like the brain, adenosine concentrations in the TME are much higher. Tumors also contain higher levels (~3%) of albumin, to which many drugs bind non-specifically. These factors adversely affect anti-tumor efficacy of drugs previously designed as CNS drugs.

AB928 is a novel, selective, and highly potent small-molecule dual antagonist of A2aR and A2bR designed to minimize potency loss associated with non-specific albumin binding. Consequently, AB928 blocks the activation of the A2aR and A2bR receptors at high adenosine agonist concentrations, even in the presence of high levels of albumin.

Methods

The activation of A2aR and A2bR (Gs-coupled receptors) increases intracellular levels of cAMP and subsequent phosphorylation of the CREB protein, which can be used to determine receptor occupancy/activation. To assess the potency of AB928 under conditions with high non-specific protein binding and adenosine receptor signaling, 5mM of synthetic adenosine agonist (NECA) was used to stimulate CREB phosphorylation in human whole blood (WB). Flow cytometry was used to quantify AB928-mediated inhibition of CREB phosphorylation by NECA on CD8+ and CD4+ T cells, NK and B cells.

Results

AB928 inhibits NECA-induced phosphorylation of CREB with a potency of 88 nM (n=25) and is more potent than other adenosine receptor antagonists currently in clinical development (Table 1). Consistent with A2bR inhibition, AB928 inhibits CREB phosphorylation in monocytes. In the mouse, inhibition of CREB phosphorylation was evaluated using ex-vivo WB assays, allowing for correlation of compound plasma levels with inhibition of phosphorylated CREB, confirming 90%+ target inhibition between dose intervals.

Conclusions

In conclusion, AB928 was designed as a best-in-class dual antagonist of A2aR and A2bR with optimal properties for effective shut-down of the effects of adenosine in the TME. AB928 is expected to enter clinical trials in late 2017.
Table 1 (abstract P6).

Inhibition by clinical stage adenosine receptor antagonists of NECA (5 μM)-induced CREB phosphorylation in CD8+ T cells (human WB)

P7 Immuno-Oncology companion diagnostics development: a complex systems approach

Lourdes Barrera (lourdes.barrera@astrazeneca.com)
AstraZeneca, Gaithersburg, DC, USA

Background

Cancer immunotherapy is one of the most important medical advances of our time, and the first approach with the potential to generate long-lasting regressions for all types and stages of cancer. Development of companion diagnostics for emerging immunotherapies is more complicated because they are not dependent on driver mutations in the drug target. Consequently, we need to develop new strategies for the development of immunotherapies.

Methods

One emerging model is to review multiple immune checkpoints and other factors within the tumor that may be contributing to immune exclusion or susceptibility to immune attack. By getting a multifactorial assessment of what is happening within the tumor, investigators should be better able to derive indicators to rationally deliver individualized therapy, either monotherapy or combinatorial immunotherapy. During this review, we will give some examples of how complex systems approach is supporting the development of new biomarkers and potentially companion and complimentary diagnostic tests.

Results

We will review some examples of the complex systems approach such as: a cytokine profile determined by data-mining analysis to set into clusters non-small-cell lung cancer patients according to prognosis; a multiscale computational model for spatio-temporal tumor immune response and an IFN-g gene expression signature to correlate with clinical outcomes in PD-L1-treated advanced non-small-cell lung cancer and urothelial carcinoma patients.

Conclusions
  • The field of immune-biomarkers aims to characterize this ongoing interaction between the immune system and cancer

  • The interplay of these individual elements determines the balance of immune activation versus suppression

  • As components and regulators of the immune response, multiple immune-biomarkers, include cell surface proteins, secreted proteins or peptides, and tumor-infiltrating immune cells

  • We need to redouble our efforts to be more sophisticated in our use of the powerful technologies at our disposal

  • Combining drugs based on increasingly well-understood molecular interactions and attacking complementary cancer hallmarks or distinct cell populations in heterogeneous tumors, is now imperative

  • ​​​​​​​As a result, in the near future, we will develop multiple biomarkers on different samples using different platforms which will result in an integrative lab tests and serial testing which will provide added clinical but, also epidemiological value.

  • These next generation of immuno-oncology biomarkers will achieve different questions related with the outcome such as the treatment response, survival and toxicity, must probably at the same time

P8 T cell response profiling in colorectal carcinoma patients reveals an enrichment in responses to specific tumor-associated antigens

Wendy Broom1, Kyle Ferber1, Johanna Kaufmann1, Christine McCoy1, Judy Jacques1, Abdulmohammad Pezeshki2, Ning Wu1, Michael O'Keefe1, Crystal Cabral1, Yana Ostrovsky1, Jason Dobson1, Theresa Zhang1, Pamela Carroll1, Khashayarsha Khazaie2, Jessica Flechtner1
1Genocea Biosciences, Cambridge, MA, USA; 2Mayo Clinic, Rochester, MN, USA
Correspondence: Wendy Broom (wendy.broom@gmail.com)

Background

In colorectal cancer (CRC) the stage of the tumor at the time of diagnosis is considered to be the most important predictor of survival. Several non-invasive diagnostic tests for CRC are in development most focusing on detection of soluble macromolecules or circulating tumor cells. Here, we are exploiting the ATLAS™ technology, with the ability to identify an individual’s T cell recall responses to putative antigens, for profiling tumor-associated antigen (TAA)-specific T cell responses in CRC patients.

Methods

T cell responses to 23 common TAAs previously implicated in CRC were profiled using the ATLAS technology. Libraries composed of full-length open reading frames expressed in E. coli alone or together with listerolysin O, thereby targeting antigen presentation to the MHC II or I pathways, respectively, were generated. Peripheral blood mononuclear cells (PBMC) were isolated from 50+ subjects who were healthy or suffering from different stages of CRC or adenomatous polyps. Peripheral CD4+ and CD8+ T cells were non-specifically expanded using anti-CD3/CD28 beads and CD14+ monocytes differentiated into monocyte-derived dendritic cells (MDDCs). E. coli libraries were co-incubated with MDDCs for antigen processing and presentation before autologous T cells were added and incubated for 21h. Recall T cell responses were measured through detection of interferon-γ and TNF-α secretion using a Meso Scale Discovery assay. Normalized cytokine concentrations that differed by >3 median absolute deviations from the median concentration secreted in response to non-immunogenic control clones were considered antigen-specific T cell responses.

Results

Several TAAs were identified for which specific T cell responses were detected in CRC patients of all stages. The breadth of response varied between individuals, but a strong enrichment of recall responses was observed to a subset of TAAs in CRC patients, to which T cell responses were not detected in healthy individuals. These responses could be attributed to both CD4+ and CD8+ T cell subsets and appeared to be Th1-polarized based on the predominant secretion of IFN-γ.

Conclusions

ATLAS is a highly suitable platform to profile TAA-specific T cell responses in CRC patients despite its classification generally as a tumor type with relatively low immunogenicity. The potential emergence of a specific T cell response profile to a subset of TAAs may open the possibility for the development of a blood-based assay to support early detection and diagnosis of CRC. We are currently investigating the utility of the identified immune signature in detection of pre-malignant stages and whether it is associated with progression to malignancy.

P9 Using iPair-TCR™ and iPair-TCR+™ to track tumor infiltrating lymphocytes from peripheral blood in a longitudinal breast cancer patient study

Miranda Byrne-Steele, Wenjing Pan, Xiaohong Hou, Brittany Brown, Mary Eisenhower, Alexa Fort, Jian Han
iRepertoire Inc., Huntsville, AL, USA
Correspondence: Miranda Byrne-Steele (msteele@irepertoire.com)

Background

Historical tissue samples are typically stored in such a way that make it difficult to access single cell TCR or BCR receptor pairing sequence information, such as samples stored as FFPE slides or tissue stored in Trizol or RNAlater. However, interrogating the bulk TCR or BCR repertoire of these samples is possible by extracting total RNA from such tissue slices or appropriately stored tissue. However, the information regarding the cognate pairing of the alpha and beta chains is lost. We wanted to see if it was possible to identify the most likely paired alpha-beta receptor sequences which likely match the historical tumor infiltrating lymphocyte TCR beta repertoire using circulating PBMCs.

Methods

As part of a longitudinal study, we sequenced the human TCR beta CDR3 fragments from a patient’s breast cancer tissue six years ago. The patient’s peripheral blood TCR beta repertoire has been examined every 6-9 months since surgical resection by amplifying CD14-CD8-CD4+ T cells and various sorted T cell subsets using arm-PCR. Single cells from this patient were recently amplified using iPair-TCR™ technology directly from the sorted cells. Identified TCR-beta CDR3 regions were used to compare to the longitudinal data sets.

Results

Three matching TCR beta sequences to bulk TIL data were identified from 104 sorted single cells. Two of these were private (meaning not shared with any of our database of 75 million CDR3 beta sequences), and one of these contained a paired alpha chain. The third sequence was considered a very public CDR3, meaning that it is present in many of the TCR beta repertoires analyzed within our database.

Conclusions

Further single cell experiments are being performed to identify more potential pairs matching the TIL data, including additional detailed phenotyping of the circulating lymphocytes. The ability to match paired receptor chains to archived data contained within banked tissue is promising and has important implications for therapeutic development.

P10 Epithelial-mesenchymal transition (EMT) signature was inversely associated with activated CD8 infiltration in non-small cell lung cancer (NSCLC)

Young Kwang Chae1, Sangmin Chang1, Taeyeong Ko1, Jonathan Anker1, Sarita Agte2, Wade Iams1, Marcelo Cruz1
1Northwestern University Feinberg School of Medicine, Chicago, IL, USA; 2Robert H. Lurie Comprehensive Cancer Center of Northwestern University, Chicago, IL, USA
Correspondence: Young Kwang Chae (sangminchang@gmail.com)

Background

EMT is a dynamic process in which epithelial cells acquire traits of mesenchymal cells. It has been known to drive metastasis and drug resistance during progression of cancer, while it is also associated with poor prognosis in cancers such as NSCLC. Though it has been suggested that EMT may contribute to immune escape, a clear relationship is yet to be established [1].

Methods

Gene expression data was obtained from the TCGA database, which contained mRNA-seq expression data of lung adenocarcinoma (ADCC,n=515) and squamous cell carcinoma (SqCC,n=501) patients. The mRNA z-score of 812 immune metagene signatures from previous studies were evaluated using Gene Set Enrichment Analysis (GSEA) [2]. Any immune cell types with a false discovery rate (q-value) ≤10% were considered as positive infiltrations, and 31 distinct immune cells were analyzed in each tumor sample. The immune landscape of EMT in NSCLC was studied by calculating EMT scores by subtracting the average expression level of ‘epithelial’ genes (n=5) from ‘mesenchymal’ genes (n=24). The tumor samples were subsequently classified as EMT-high (EMT scores ≥highest1/3) or EMT-low groups (EMT scores ≤lowest1/3).

Results

There was a significantly lower infiltration of activated CD8 cells in EMT-high groups compared to EMT-low groups in both ADCC and SqCC (p=<0.001, 15.7%; 37.2%, p=<0.001, 15.5%, 40.5% respectively), as well as decreased infiltration of activated CD4 cells (p=0.161, 25.0%; 31.2%; p=<0.001, 26.8%; 48.2%) (Fig. 1). Effector memory CD4 cells showed significantly lower infiltration in both EMT-high groups of NSCLC (p=<0.001, 3.5%; 16.8%, p=<0.001, 3.6%; 16.7%). EMT-scores showed an inverse relationship with infiltrating immune cells, most notably by activated CD8 cells. Increased expression of mesenchymal genes (VIM, ZEB1, ITGB6, etc.) in both ADCC and SqCC resulted in significantly lower infiltration of activated CD4/CD8 cells. However, increased expression of epithelial genes (COL5A1, TJP1) also resulted in significantly lower infiltration of CD8 cells in NSCLC.

A significant positive correlation between immunosuppressive cytokine, IL-10 expression and EMT was found in both NSCLCs (p=<0.001). Other cytokines with immunosuppressive functions that showed a significant positive correlation with EMT among others were TGF-β and ADCC. EMT-score, or infiltration of CD8 cells among EMT-high/low groups did not show any significant differences in overall survival (OS).

Conclusions

We report for the first time, the association between EMT-score and decreased infiltration of activated CD8 cells in both NSCLCs. This may be related to increased expression of immunosuppressive cytokines such as IL-10 or TGF-β. Markers of immune exclusion such as EMT scores may be potential biomarkers that predict resistance to immunotherapy.

References

1. Terry S et al: Mol Oncol 2017; 11(7):824-846

2. Angelova M et al: Genome Biol. 2015; 16:64.doi:10.1186/s13059-015-0620-6.
Fig. 1 (abstract P10).

See text for description

P11 WNT/β-catenin pathway is inversely correlated with activated CD8 T cell infiltration in non-small cell lung cancer (NSCLC)

Young Kwang Chae1, Taeyeong Ko1, Sangmin Chang1, Jonathan Anker1, Sarita Agte2, Wade Iams1, Marcelo Cruz1
1Northwestern University Feinberg School of Medicine, Chicago, IL, USA; 2Robert H. Lurie Comprehensive Cancer Center of Northwestern University, Chicago, IL, USA
Correspondence: Young Kwang Chae (tyeongko@gmail.com)

Background

It is well known that the WNT/β-catenin pathway promotes cell proliferation which is associated with tumorigenesis. While many mechanisms of tumorigenesis have recently been elucidated, one which has particularly garnered attention is the immune escape mechanism [1]. The correlation between WNT/β-catenin pathway and immune escape has been investigated in several cancers including melanoma. However the relationship in lung cancer is still largely unknown.

Methods

In this study, TCGA data of 515 samples of lung adenocarcinoma (ADCC) and 501 samples of squamous cell lung carcinoma (SqCC) were utilized for analysis of immune landscape and gene expression scores. 17 promotor genes of WNT/β-catenin pathway (CTNNB1, DVL1, EP300, LRP6, etc.) and 8 suppressor genes of WNT/β-catenin pathway (AMER1, APC, AXIN1, etc.) were utilized for the analysis. The expression score of WNT/β-catenin pathway was calculated by subtracting the average of suppressor gene mRNA expression from the average of promotor gene mRNA expression. The tumor samples were then divided into ‘low WNT/β-catenin expression group’(WNT/β-catenin pathway expression scores≤lowest1/3), and 'high WNT/β-catenin expression group'(WNT/β-catenin pathway expression scores≥highest1/3) according to their level of gene expression scores. For immune landscape, Gene Set Enrichment Analysis (GSEA) was utilized to analyze the mRNA z-score of 812 immune metagene signatures from previous studies [2]. Positive infiltration was considered as any immune cell type with a false discovery rate (q-value) equal to or less than 10% and 31 types of immune cells were analyzed for each tumor sample.

Results

We found that in both ADCC and SqCC, the high WNT/β-catenin score group showed lower infiltration of activated CD8 T cell (p=0.006, <0.001 respectively) (Fig. 1). We also found a negative correlation between WNT/β-catenin pathway and the expression of immune stimulatory cytokines such as IL-2, IFN-gamma in ADCC (p=0.003, 0.029 respectively), whereas SqCC showed a negative correlation between WNT/β-catenin pathway and the expression of the immune stimulatory cytokines such as IL-2, IFN-gamma and IL-12(p=0.008, 0.041, 0.018 respectively). The expression of immune suppressive cytokine such as TGF-β showed positive correlation with WNT/β-catenin pathway (p=0.011 for ADCC, <0.001 for SqCC).

Conclusions

We report for the first time that immune exclusion by WNT/β-catenin pathway may also occur in NSCLC. The WNT/β-catenin pathway is associated with lower immune stimulatory cytokine secretion, which leads to lower infiltration of activated CD8 T cell. Such findings suggest that WNT/β-catenin pathway may serve as a potential biomarker for predicting resistance to immunotherapy.

References

1. Pai SG, et al. J Hematol Oncol. 2017; 10:101.doi:10.1186/s13045-017-0471-6

2. Angelova M, et al. Genome Biol. 2015; 16:64.doi:10.1186/s13059-015-0620-6
Fig. 1 (abstract P11).

Immune landscape of ADCC and SqCC

P12 Precision immune monitoring: new 30-parameter flow cytometry uniquely and comprehensively defines immune checkpoint expression and T cell phenotypes in tumor tissue

Pratip Chattopadhyay1, Amber Giles2, Leonard Nettey3
1NYU Medical School, New York, NY, USA; 2NIH, Bethesda, MD, USA; 3VRC, NIH, Bethesda, MD, USA
Correspondence: Pratip Chattopadhyay (pratip.chattopadhyay@nyumc.org)

Background

The cellular processes that govern T cell activity at the tumor-immune interface involve many different proteins. To better understand these processes, and identify biomarkers of efficacy and toxicity, comprehensive analysis will be needed, through the measurement of as many T cell related proteins as possible. In this abstract, we introduce state-of-the-art, 30-parameter flow cytometry technology for use in immunotherapy research and immune monitoring, and describe the first studies applying this technology to studies of peripheral blood and tumor tissue.

Methods

We developed the first 30-parameter flow cytometry panel focused on the expression of checkpoint molecules. The panel includes CD45, CD3, CD4, and CD8 (to identify major T cell lineages); CD45RO, CCR7, CD27, CD28, CD57, CD95, and CD127 (markers of T cell differentiation); PD1, ICOS, CTLA4, LAG3, TIM3, TIGIT, VISTA, GITR, 41BB, and 2B4 (immune checkpoint molecules, involved in T cell exhaustion); CXCR3, CXCR6, and CD103 (trafficking markers); CD25, CD69, and HLA-DR (immune activation markers); and a viability dye (to exclude dead cells from analysis). We also developed a 30-parameter panel that quantifies major cell lineages (T, B, NK, myeloid, dendritic) and their subsets (naive, central memory, effector, plasma, etc.). These panels were applied to peripheral blood and resected tumor tissue from glioblastoma multiformae (GBM) patients, in order to comprehensively define the types of cells present in patient peripheral blood and tumors.

Results

We could clearly identify populations of cells expressing and not expressing each marker. We developed new technical approaches to ensuring robust data analysis, and novel bioinformatics approaches that are easy to understand and accessible to researchers. This will allow broader adoption of this technology. Analysis of GBM tumor tissue revealed about 10 unique subsets of cells, some of which are not found in peripheral blood. Notably, we could easily identify cell types expressing checkpoint molecules that are current therapeutic targets, and define cells expressing combinations of those targets - or importantly - none of those targets (Fig. 1). Further analysis allowed comprehensive characterization of these cells, to identify their differentiation status, activation status, and trafficking markers.

Conclusions

The technology and approach we have developed will be broadly applicable to phenotyping and monitoring immunotherapy in cancer patients. The technical platform has a number of advantages over other approaches (such as mass cytometry and RNA sequencing), and can be a central technology in precision oncology efforts, which will depend on complete and comprehensive characterization of cells.
Fig. 1 (abstract P12).

T-sne analysis of T cells from GBM tumors

P13 Pre-existing T cell immunity, mutational burden and stromal TGF-β signaling drives clinical responses to atezolizumab in locally advanced or metastatic urothelial carcinoma (mUC)

Sanjeev Mariathasan1, Shannon J. Turley1, Dorothee Nickles1, Yulei Wang1, Edward E. Kadel III1, Hartmut Koeppen1, Kobe Yuen1, Allessandra Castiglioni1, Jillian L. Astarita1, Rafael Cubas1, Suchit Jhunjhunwala1, Romain Banchereau1, Yagai Yang1, Yinghui Guan1, Cecile Chalouni1, James Ziai1, Yasin Şenbabaoğlu1, Steve Lianoglou1, Michiel van der Heijden2, Yohann Loriot3, Johathan Rosenberg4, Lawrence Fong5, Ira Mellman1, Daniel S. Chen1, Marjorie Green1, Christina Derleth1, Gregg D. Fine1, Priti S. Hegde1, Richard Bourgon1, Thomas Powles6
1Genentech, Inc., South San Francisco, CA, USA; 2Netherlands Cancer Institute, Amsterdam, Netherlands; 3Gustave Roussy, Villejuif, France; 4Memorial Sloan Kettering Cancer Center, New York, NY, USA; 5University of California San Francisco, San Francisco, CA, USA; 6Barts Experimental Cancer Medicine Centre, London, UK
Correspondence: Sanjeev Mariathasan (genevieve.cohen@healthinteractions.com); Shannon J. Turley

Background

Checkpoint inhibitor blockade can result in robust and durable anti-tumor responses in many cancers. However, only a subset of patients experience long-term clinical remissions. Identifying determinants of response to cancer immunotherapy is critical for extending therapeutic benefit to more patients. Atezolizumab (anti–PD-L1) was approved in the US for the treatment of mUC based on the single-arm Phase II study IMvigor210 (NCT02108652). Here, we examined the biology underlying primary immune escape and responsiveness to anti–PD-L1 in patients from IMvigor210.

Methods

Cisplatin-ineligible patients previously untreated for mUC or patients who progressed post-platinum therapy received atezolizumab (1200 mg IV q3w). In both cohorts, RECIST v1.1 objective response was a primary endpoint evaluated in all patients and in PD-L1 expression subgroups on tumor-infiltrating immune cells (IC; VENTANA SP142 IHC assay). Exploratory analyzes in evaluable pre-treatment tissues included CD8 IHC, whole-transcriptome RNA sequencing, gene set enrichment analyzes and Lund subtyping. A FoundationOne panel was used to estimate tumor mutation burden (TMB), and whole-exome sequencing data were used for neoantigen prediction. EMT6-grafted BALB/c mice treated with anti–TGF-β and/or anti–PD-L1 antibodies were evaluated for tumor growth inhibition and characterization of tumor microenvironment phenotypes.

Results

PD-L1 IC expression and clinical response were positively associated with a CD8+ T-effector expression signature (P < 0.001 and P = 0.0087, respectively); patients with complete responses had higher T-effector signatures than did partial or non-responders (P = 0.002). Response was also strongly associated with high tumor mutation and neoantigen burden (P < 0.001 each). Likewise, response was positively associated with the genomically unstable molecular subtype (P < 0.001) and pathways involved in proliferation and DNA damage response. Further, non-responsiveness correlated with expression of a 2-gene TGF-β signature (P < 0.001), particularly in patients with CD8+ T cells predominantly in the collagen-rich matrix surrounding tumors (immune-excluded tumor phenotype; P < 0.001). Addition of anti–TGF-β to anti–PD-L1 in the mouse promoted T cell localization to immune-excluded tumors and reduced tumor growth in vivo.

Conclusions

Pre-existing T cell immunity and TMB are associated with response to atezolizumab in mUC, whereas TGF-β signaling in the stroma is a negative indicator of response, especially in immune-excluded tumors, a common phenotype of mUC. Integration of these 3 independent biological features provides a strong basis for understanding clinical outcomes in this setting and suggests that TGF-β can interact with the tumor microenvironment to restrain responsiveness to anti–PD-L1 and anti-tumor immunity.

Trial Registration

ClinicalTrials.gov Identifier NCT02108652

P14 Comprehensive characterization of solid tumor immune profiles for precision immunotherapy using Immune Report Card

Mary Nesline, Jeffrey Conroy, Sarabjot Pabla, Ji He, Blake Burgher, Vicent Giamo, Jonathan Andreas, Paul DePietro, Antonios Papanicolau-Sengos, Mark Gardner, Sean Glenn, Carl Morrison
OmniSeq, Inc., Buffalo, NY, USA
Correspondence: Carl Morrison (Jeffrey.conroy@omniseq.com)

Background

Monoclonal antibodies directed at the inhibitory immune receptors have emerged as successful treatment options for numerous tumor types using both mono and combination modalities. However, responses are not universal and better predictive biomarkers are needed to support precision immunotherapy. We present the findings from patients tested by Immune Report Card (IRC), a validated clinical assay that measures immunotherapeutic targets and immune response cycle markers to provide a complete profile for appropriate selection of immunotherapy.

Methods

300 FFPE cancer samples from patients with diverse histologies were evaluated by IRC using 5 test modes. RNA-seq measures transcript levels of genes related to T cell receptor signaling and tumor infiltrating lymphocytes, DNA-seq is used to estimate mutational burden, MSI-PCR to assess microsatellite instability, Fluorescent in situ hybridization (FISH) to detect PD-L1/2 copy number gain, and FDA-approved Immunohistochemistry to measure PD-L1 protein expression and pattern of expression for PD-L1, CD3, and CD8. Results are interpreted and summarized in an integrated report that comprehensively characterizes the tumor microenvironment to assess likelihood of response.

Results

At least one over expressed marker associated with either an FDA approved checkpoint inhibitor or combination immunotherapy clinical trial was identified for more than 80% of patients. Among PD-L1 IHC negative patients, the majority overexpressed at least one target with therapies in clinical development. Frequently overexpressed genes with targets in clinical development were identified for anti-inflammatory response (TGFB1, IL10), myeloid suppression (CSF1R), checkpoint blockade (PD-L1, LAG3, TIM3), T cell primed (GITR, OX40, CD40, CD137, ICOS), metabolic immune escape (IDO1) and pro-inflammatory response (IL1B, TNF) phenotypes. Conversely, tumors with low overall immune related expression (i.e. immune deserts), often exhibited high expression of a single marker with a pertinent clinical trial therapeutic opportunity.

Conclusions

Each of the 5 IRC test modes is necessary for determining the overall likelihood of patient response to FDA approved checkpoint inhibitors. Comprehensive profiling by IRC allows for the identification of patients with over expression of markers with therapeutic targets in clinical development, providing options for patients who are PD-L1 IHC negative. The results also highlight immune phenotypes with therapeutic targets beyond checkpoint blockade, suggesting a complex tumor microenvironment in many tumors. As immune therapy moves from treatment of last resort to first and second line treatment, IRC can provide actionable results for the total tumor immune microenvironment.

P15 Overexpression of immunotherapeutic targets in the immune desert phenotype

Sarabjot Pabla, Jeffrey Conroy, Sean Glenn, Ji He, Blake Burgher, Vincent Giamo, Jonathan Andreas, Maochun Qin, Mark Gardner, Antonios Papanicolau-Sengos, Mary Nesline Carl Morrison
OmniSeq, Inc., Buffalo, NY, USA
Correspondence: Carl Morrison (Jeffrey.conroy@omniseq.com)

Background

Immunotherapy using checkpoint blockade with monoclonal antibodies has gained increasingly high importance in treatment for cancer patients. However, a large proportion of cancer patients do not highly express primary cancer immune biomarkers such as PD-1/PD-L1, MSI, and mutational burden, which have been associated with response. Here we present immune-related expression signatures for patients that present with an immune desert phenotype, distinguished by lack of CD8 positive T cells and characteristic T cell receptor signaling expression levels. As part of our clinical immune cell analysis assay, Immune Report Card (IRC), we identified secondary immune biomarkers that are singularly expressed in the otherwise non-inflamed tumor microenvironment and are potential clinical immunotherapy targets.

Methods

167 formalin-fixed, paraffin-embedded (FFPE) cancer samples of diverse histologies were evaluated by the RNA-seq component of IRC to measure transcript levels of genes related to T cell receptor signaling and tumor infiltrating lymphocytes. Resultant data was QC filtered, normalized and ranked based on an assorted reference population of various tumor types. Gene signatures were determined using these ranked expression values with a rank value > 85th percentile considered high. Tumors are also defined as inflamed or non-inflamed based upon RNA-seq analysis of CD8, wherein, Tumors in upper 50th percentile of rank for CD8 are considered inflamed, while those in the lower 50th percentile are considered non-inflamed.

Results

The immune desert phenotype represented 28% (n=47) of the tested samples. 95% (n=45) tumor were non-inflamed (CD8 Rank < 50). 51% (n=24) of all samples did not express any gene at a high level. However, 19% (n=9) of the immune desert phenotype samples have potential for monotherapy targets and another 19% (n=9) have underlying biology indication for combination therapy based on the high expression of a singular immune-related gene.

Conclusions

Immune Report Card allows for the profiling of the tumor immune microenvironment to delineate underlying immune biology of solid tumor samples. With a significant number of non-inflamed tumors lacking high expression of any immune biomarker, IRC results suggests that an underlying biological immune ignorance state exists in the tumor microenvironment of many patients. However, IRC could identify so-called “oasis” targets that could be potentially targeted with mono or combination immune therapy in the immune desert phenotype. With the ever-increasing numbers of FDA-approved therapies and clinical trials, IRC offers a robust tool to identify patients that might benefit from these options.

P16 The immune activated phenotype: secondary immunotherapeutic targets in the primary biomarker negative inflamed tumor

Sean Glenn, Jeffrey Conroy, Sarabjot Pabla, Ji He, Blake Burgher, Vincent Giamo, Jonathan Andreas, Mark Gardner, Antonios Papanicolau-Sengos, Mary Nesline, Carl Morrison
OmniSeq, Inc., Buffalo, NY, USA
Correspondence: Carl Morrison (Jeffrey.conroy@omniseq.com)

Background

Immune checkpoint inhibitors are now used to treat many different types of cancer, with some patients demonstrating durable clinical responses. As expected, a significant number of responders express high levels of primary cancer immune biomarkers such as PD-1/PD-L1, MSI, or mutational burden. Here we present immune-related expression signatures for patients with an immune activated phenotype that overexpress several pro- and anti-inflammatory genes with or without primary biomarker detection. These immune signatures were identified as part of Immune Report Card (IRC), a comprehensive molecular and immunological assay that uses five testing modes to detect several known markers of the host anticancer immune response.

Methods

167 formalin-fixed, paraffin-embedded (FFPE) cancer samples of diverse histologies were evaluated by IRC to measure transcript levels of genes related to T cell receptor signaling and tumor infiltrating lymphocytes (RNA-seq) and mutational burden (DNA-seq). Resultant data was QC filtered, normalized and ranked based on an assorted reference population of various tumor types. Gene signatures and mutational burden were determined using these ranked values with a rank value > 85th percentile considered high. Tumors are also defined as inflamed or non-inflamed based upon RNA-seq analysis of CD8. Tumors in upper 50th percentile of rank for CD8 are considered inflamed, while those in the lower 50th percentile are considered non-inflamed. RNA-seq analysis of CD8 had been previously calibrated against quantitative image analysis using the Aperio platform.

Results

The immune activated phenotype represented 20% (n=35) of the tested samples, of which 32 were inflamed (91%). 31% (n=11) of these samples had co-expression of the primary immune biomarker PD-L1 or demonstrated a high mutational burden. However, 66% (n=23) of the immune activated tumors were primary biomarker negative but instead expressed multiple secondary immune biomarkers that are potential checkpoint inhibitor therapy targets.

Conclusions

Immune Report Card profiles the tumor immune microenvironment to outline the immune biology of tumor samples. IRC is not only able to identify samples with highly expressed primary immune biomarkers such as PD-L1, MSI, or mutational burden, but using RNA-seq can also identify secondary biomarkers in many samples. These secondary biomarkers shed light on the underlying biological immune-state of the tumor microenvironment, potentially identifying additional mono or combination immunotherapies in PD-L1 negative inflamed tumors. With the ever-increasing numbers of FDA-approved therapies and clinical trials, IRC offers a robust tool to identify patients that might benefit from these options.

P17 The immune-excluded phenotype beyond colorectal cancer

Jeffrey Conroy1, Carl Morrison1, Sean Glenn1, Wiam Bshara2, Sarabjot Pabla1, Blake Burgher1, Vincent Giamo1, Jonathan Andreas1, Mary Nesline1, Mark Gardner1, Antonios Papanicolau-Sengos1, Igor Puzanov2, Marc Ernstoff2
1OmniSeq, Inc., Buffalo, NY, USA; 2Roswell Park Cancer Institute, Buffalo, NY, USA
Correspondence: Jeffrey Conroy (Jeffrey.conroy@omniseq.com)

Background

Absence of tumor infiltrating lymphocytes in the tumor proper, referred to as the immune-excluded phenotype, has been directly described in the peer-reviewed literature for colorectal cancer, but indirectly for melanoma and other solid tumors in the field of immunotherapy discussions. As part of our clinical immune cell analysis using a New York State CLEP approved assay, Immune Report Card (IRC), we routinely perform CD8 and CD3 immunohistochemistry (IHC) to determine the infiltrating versus the non-infiltrating (excluded) phenotype for all tumor types.

Methods

For all tumor types tested including carcinoma, sarcoma, and melanoma the immune-excluded phenotype is defined as a restriction of more than 95% of all CD8+ T cells present in a tumor tissue section to the periphery or interstitial stromal areas and not actively invading nest or groups of neoplastic cells. Tumors are also defined as inflamed or non-inflamed based upon RNA-seq analysis of CD8 from the same tissue section and comparison to a reference population of several hundred prior analyzed samples. Tumors in upper 50th percentile of rank for CD8 gene expression are considered inflamed, while those in the lower 50th percentile are considered non-inflamed. RNA-seq analysis of CD8 had been previously calibrated against quantitative image analysis using the Aperio platform. A total of 100 consecutive samples tested, excluding colorectal cancer, were evaluated for the excluded infiltrate phenotype.

Results

The immune-excluded phenotype was identified in all tumor types with the most frequent association seen in lung cancer and the least frequent being melanoma. The immune-excluded phenotype was more common in non-inflamed tumors with a direct correlation to CD8 rank, but was also seen in some moderate to highly inflamed tumors. The most common histological pattern of the immune-excluded phenotype was restriction of CD8+ T cells to the interstitial stromal areas.

Conclusions

The immune-excluded phenotype does occur in tumor types other than colorectal cancer. Providing a precisely defined nomenclature and method of testing for the immune-excluded phenotype in inflamed and non-inflamed tumors is important for both clinical and research purposes.

P18

Withdrawn

P19 Genomic determinants of response to pembrolizumab in gastric cancer

Razvan Cristescu, Diane Levitan, Julie Kobie, Andrew Albright, Ping Qiu, Robin Mogg, Jared Lunceford, Xiao Qiao Liu, Xinwei Sher, Michael Nebozhyn, Gladys Arreaza, Wendy Blumenschein, Mark Ayers, Terrill McClanahan, Andrey Loboda, David Kaufman, Rita Dalal, Minori Koshiji
Merck & Co., Inc., Kenilworth, NJ, USA
Correspondence: Razvan Cristescu (razvan_cristescu@merck.com)

Background

The mutation load of somatic non-synonymous variants (ML) is associated with response to anti CTLA-4 and PD-1/PD-L1 immunotherapies in select tumors, likely because neoepitope formation is not subject to central immune tolerance. An IFNγ gene expression profile (GEP) characteristic of a tumor T cell inflamed environment is also related to response to anti PD-1/-L1 therapy. ML and GEP have been reported as predictive in a pan-tumor setting and in HNSCC. This study evaluated relationships between ML and response, independent predictive value of ML, GEP and PD-L1 IHC in patients with gastric cancer treated with pembrolizumab monotherapy, and the association of response with TCGA molecular subtypes (MSI, EBV, CIN, and GS).

Methods

Whole exome sequencing (WES) and GEP were assessed in FFPE tumor specimens from previously treated patients with gastric cancer from KEYNOTE 059 cohort 1. ML, neoantigen load (NL), EBV status, and clonality were assessed by state-of-the-art WES analytics. GEP score is a weighted sum of normalized expression values of 18 genes. Molecular subtypes were determined using DNA genomic features. Statistical testing of ML and response, ML and GEP or PD-L1 relationship in EBV-negative patients, and association with molecular subtypes were pre-specified.

Results

85/259 patients (33%) had WES results, 76 of which had GEP results and 84 of which had PD-L1 results. There were 7 responders (CR or PR) and 78 non-responders. Five patients were identified as EBV-positive (EBV+), 3 as MSI-H, 45 as CIN, and 32 as GS. All MSI-H patients and no EBV+ patients were responders. For non-MSI-H responders (n=4), response was not significantly associated with the remaining TCGA subtypes EBV, CIN and GS. In EBV-negative (EBV-) patients, association between ML and objective response (OR) was statistically significant (p= 7E-4; AUROC 0.83). ML was not significantly correlated with either GEP or PD-L1. In a joint model, ML was significantly associated with response after adjusting for GEP (p=4E-3) or PD-L1 (p=2E-3) in EBV- patients. NL and clonality-weighted ML were also significantly associated with response in EBV- patients (p=2E-3 and 8E-4, respectively).

Conclusions

As reported in other indications, ML and GEP or PD-L1 were independently predictive of response to pembrolizumab monotherapy in gastric cancer. This suggests that ML and GEP or PD-L1 may help characterize responses to anti PD-1 therapies and novel treatment regimens in gastric cancer.

Trial Registration

ClinicalTrials.gov Identifier NCT02335411

P20 Distinct tumor immune profiles in non-small cell lung cancer (NSCLC) revealed by automated quantitative digital image analyzes of immune biomarker densities and spatial relationships

Marie Cumberbatch1, Elaine Foster2, James Clay2, Sunita Mistry2, Andrew Lawrence2, Christopher Womack2, David Fairley2
1Immune Insight Limited, Macclesfield, UK; 2HistologiX Limited, Nottingham, UK
Correspondence: Marie Cumberbatch (marie.cumberbatch@immuneinsight.com)

Background

The tumor microenvironment comprises complex interactions between infiltrating immune cells and tumor cells. Understanding these interactions may provide valuable insights into mechanisms of response and resistance, and aid clinical decisions. In the present study, immune cell densities and PD-L1 expression in NSCLC samples were quantified to determine immune status of each sample prior to conducting sophisticated spatial analyzes to establish the proximity and ratio of cytotoxic T cells and regulatory T cells to tumor expression of PD-L1.

Methods

Serial sections of formalin-fixed paraffin-embedded NSCLC samples (n=9; Tissue Solutions, Glasgow, UK) were stained immunohistochemically using well validated assays for CD3, CD8, Foxp3 and PD-L1. Whole slide digital images were generated and viable tumor regions annotated by a clinical pathologist. The immune profile of each sample was established by applying digital image analysis tools available in the Indica HaloTM platform to determine immune cell densities and PD-L1 expression. Selected images were subsequently registered and spatial analyzes between immune components performed.

Results

Tumor membrane PD-L1 scored by a pathologist ranged from 0% to 90% positivity with variable intensity of staining. Digital image analysis of PD-L1 using a membrane algorithm coupled with a classifier tuned to analyze tumor only delivered comparable PD-L1 positivity. Immune cell densities ranged from minimal infiltration up to 2315, 1512 and 340 cells/mm2 for CD3, CD8 and Foxp3, respectively. Variable CD8:Foxp3 ratios were observed ranging from 1.3 to 16.5, and no direct relationship between immune cell infiltration and PD-L1 expression was evident. Spatial analyzes of the proximity of CD8 and Foxp3 to PD-L1 revealed samples (n=2) exhibiting 15-20% of cytotoxic T cells within 5μm of PD-L1 with few adjacent Foxp3+ cells (CD8:Foxp3 ratio >10.0). In contrast, we identified samples (n=2) where CD8+ T cells were within close proximity (5μm) of the immunosuppressive influences of both PD-L1 and Foxp3+ regulatory T cells (CD8:Foxp3 ratio <2.6).

Conclusions

We have demonstrated using next generation digital image analysis tools that the potential immunosuppressive influences within close proximity of cytotoxic T cells within a tumor can be mapped. This approach has potential to identify tumors where cytotoxic T cells are in close proximity to PD-L1 and those where cytotoxic T cells may be in the vicinity of additional immunosuppressive influences, thus identifying patients most likely to benefit from different immunotherapies.

P21 Correlation of inflammatory biomarkers and patient-reported outcomes in patients with urothelial carcinoma treated with durvalumab

Arnold Degboe1, Srikala Sridhar2, Jingsong Zhang3, Pralay Mukhopadhyay1, Ashok Gupta1, Lorin Roskos1, Xiang Guo1, Terence Friedlander4
1AstraZeneca, Gaithersburg, MD, USA; 2Princess Margaret Cancer Centre, Toronto, ON, Canada; 3Moffitt Cancer Center, 12902 USF Magnolia Drive, FL, USA; 4UCSF Medical Center, San Francisco, CA, USA
Correspondence: Arnold Degboe (andrea.hammons@PAREXEL.com)

Background

A correlation between changes in inflammatory/cachexia biomarkers, tumor shrinkage, and overall survival (OS) has been demonstrated in patients with urothelial carcinoma (UC) receiving durvalumab [1]. The current analysis assessed the relationship between inflammatory/cachexia biomarkers and patient-reported outcomes (PROs), and differences in PROs between responders versus non-responders in patients with UC treated with durvalumab.

Methods

In a phase 1/2, dose-escalation study of durvalumab, Functional Assessment of Cancer Therapy—Bladder Cancer (FACT-BL) and the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) were administered before other procedures at screening, day 1 of doses 1 (d1; baseline assessment), 3 (d29), 4 (d43), 5 (d57), 7 (d85), 9 (d113), and every 8 weeks thereafter. For 126 patients, the association between PRO scores and maximum percent changes in tumor size, albumin level, neutrophil/lymphocyte ratio (NLR), and durvalumab clearance from baseline was assessed by Spearman’s correlation analysis.

Results

Pretreatment FACT-BL scores were similar for responders and non-responders, with patients reporting relatively high health-related quality of life (QoL) at baseline. Compared with non-responders, clinical tumor responders reported statistically significant improvement in FACT-BL total scores (mean change of 17.1 vs -4.13, P=0.0005), FACT-BL bladder cancer–specific symptoms (mean change of 5.8 vs -0.56, P=0.0047), and FACT-BL Trial Outcome Index (mean change of 14.3 vs -1.9, P=0.0005) at Day 113. Sustained improvement in PROs was observed in 43% of the responders compared with 12%–18% of non-responders. Fifty percent of patients with a clinical tumor response also reported a clinically meaningful change in UC symptoms at Day 113 using a threshold based on minimum important difference (MID) defined as ½ baseline standard deviation, which was sustained over time (≥2 consecutive visits with MID improvement vs baseline). Clinical tumor responders also showed higher mean scores and improvement from baseline in EORTC QLQ-C30 functional and global health status/QoL scales.

FACT-BL scores and EORTC QLQ-C30 functioning improvement were correlated significantly with decreased tumor size, increased albumin, decreased NLR, and decreased durvalumab clearance (Table).

Conclusions

PROs were correlated significantly with inflammatory/cachexia biomarkers and durvalumab clearance in advanced UC; additionally, there was a strong correlation between clinical tumor response and FACT-BL and EORTC QLQ-C30 scores.

Trial Registration

ClinicalTrials.gov Identifier NCT01693562

Reference

1. Powles T, Jin C, Zheng Y, et al. J Clin Oncol. 2017; 35 (suppl; abstr 3035).

P22 Up-regulation of a T and NK cell gene signature in peripheral blood is associated with mRNA-based immunotherapy in lung and prostate cancer

Henoch S Hong1, Madeleine M Hipp1, Fatma Doener1, James M Billingsley2, Linus Backert3, Oliver Kohlbacher3, Andreas Schröder1, Ute Klinkhardt1, Sabine Brutlach1, Uwe Peukert1, Ulrike Gnad-Vogt1, Mariola Fotin-Mleczek1, Sven D Koch1
1CureVac AG, 72076 Tübingen, Germany; 2Emory University, Atlanta, GA, USA; 3Applied Bioinformatics, University of Tübingen, 72076 Tübingen, Germany
Correspondence: Fatma Doener (fatma.doener@curevac.com)

Background

CV9201 and CV9104 represent novel mRNA-based cancer immunotherapeutics targeting multiple tumor antigens associated with either non-small cell lung cancer (NSCLC) or prostate cancer (PC), respectively. We sought to comprehensively profile transcriptional changes in peripheral blood of cancer patients after repeated mRNA treatment.

Methods

Whole-genome transcriptome profiling was performed in blood samples from 46 patients with localized, intermediate or high risk PC and samples from 22 stage IV NSCLC patients before and after repeated intradermal immunizations with protamine-formulated mRNA. The 46 study participants with PC included a cohort of 16 untreated control subjects. Expression data was analyzed by gene set enrichment analysis and modular approaches using blood transcriptional modules. Phenotypic leukocyte analyzes were performed by flow cytometry.

Results

Peripheral blood samples post mRNA treatment were characterized by an enrichment of various modules consistent with a T and/or NK cell profile in NSCLC as well as PC patients. Notably, this up-regulation of effector lymphocyte genes was completely absent in the PC control subjects. We observed an enrichment of a myeloid cell signature at baseline compared to samples derived after mRNA immunotherapy. The T and NK cell gene signature and the myeloid gene signature were inversely correlated and non-overlapping. Furthermore, correlation analyzes between gene expression data with findings from the phenotypic analyzes, suggest changes in the cellular composition of peripheral blood cells after treatment.

Conclusions

The consistent transcriptional up-regulation of T and NK cell modules in post treatment samples in different cancers and the absence of it in the untreated control cohort suggest an association with intradermal mRNA immunotherapy including its potential immunostimulatory effects.

P23 Deep molecular and immune-infiltrate stratification of cancer identifies mechanistic subtypes and predicts response to checkpoint inhibition

Janusz Dutkowski, Roy Ronen
Data4Cure, Inc., La Jolla, CA, USA
Correspondence: Janusz Dutkowski (janusz@data4cure.com)

Background

Checkpoint inhibitors are revolutionizing cancer therapy but identifying patients which are likely to achieve clinical benefit remains a major challenge. While relatively simple predictors such as PD-L1 expression or tumor mutation load have shown promise in predicting responders in specific cancer types, they fail to explain or predict response for many patients, suggesting that a more comprehensive understating of biology underlying tumor and immune system interactions and immunotherapy response is needed.

Methods

We developed an integrated machine-learning system that stratifies cancers using multimodal omics data and associated profiles of immune infiltration and immune pathway activity. This framework holistically determines the underlying pathway-level architecture of genomic and epigenetic alterations in each cancer using a new Model-based Network-informed Stratification approach (MNS, [1] for early version). It further uses a multiscale pathway mapping approach [2, 3] to predict subtype-specific immune-infiltrates, activation of key immune pathways, and, ultimately, response to checkpoint inhibition.

Results

The new framework is used to interrogate the tumor-immune system in colorectal and lung adenocarcinomas. Stratification of 533 TCGA lung adenocarcinomas into four subtypes (LUAD_MNS4) allows for reanalysis of pembrolizumab trial data [4] by matching the trial patients to the LUAD_MNS4 to infer immune infiltration and pathway activation signatures for each patient and predict patient response to pembrolizumab (Figure 1). In colorectal cancer, a total of six subtypes (COADREAD_MNS6) are identified along with subtype-specific profiles of pathway mutations and copy-number alterations. Two subtypes are hypermutated and one (COADREAD_MNS6_S4) is associated with high predicted CD8 T cell infiltration. The system also uncovers subtypes S1 and S6 which link WNT/β-catenin pathway mutations with immunologically cold tumors, suggesting the possibility of a similar mechanism for immune evasion as recently discovered in specific subtypes of melanoma [5].

Conclusions

Comprehensive systems-level interrogation of tumor genomic architecture and microenvironment may improve our understanding of tumor-immune biology and allow for more accurate patient stratification for immunotherapy.

References

1. Hofree, Shen, Carter, Gross, Ideker. Network-based stratification of tumor mutations. Nat Methods. 2013;10(11):1108–1115.

2. Dutkowski, Kramer, Surma, Balakrishnan, Cherry, Krogan, Ideker. A gene ontology inferred from molecular networks. Nat Biotechnol. 2013; 31(1):10.1038/nbt.2463.

3. Yu, Kramer, Dutkowski, Sharan R, Ideker T. Translation of Genotype to Phenotype by a Hierarchy of Cell Subsystems. Cell Syst. 2016;2(2):77–88.

4. Rizvi et al. Mutational landscape determines sensitivity to PD-1 blockade in non–small cell lung cancer. Science. 2015; 348(6230):124–128.

5. Spranger, Bao, Gajewski. Melanoma-intrinsic β-catenin signalling prevents anti-tumour immunity. Nature. 2015;523(7559):231-5.

P24 Systematic literature review of PD-L1 assays, their scoring algorithms and validation metrics

Margarita Udall1, Maria Rizzo2, Juliet Kenny2, Jim Doherty1, Eric Faulkner3
1Pfizer Inc., New York, NY, USA; 2Evidera, London, UK; 3Evidera, Morrisville, NC, USA
Correspondence: Margarita Udall (megancibulas@gmail.com)

Background

Programmed cell death protein 1 (PD-1) is a cell-surface receptor expressed on T cells. The interaction between PD-1 and its ligand, programmed death ligand 1 (PD-L1), down-regulates immune responses. Many cancer types express PD-L1 and, via the PD-1/PD-L1 interaction, evade immune destruction. Antibodies targeting this interaction are approved treatments for several cancers with a variety of PD-L1 tests available to measure PD-L1 expression. A systematic literature review was conducted to evaluate these commercially available PD-L1 assays, their scoring algorithms and validation data in bladder/urothelial, lung, gastric, or ovarian cancer.

Methods

A search was conducted in Medline and EMBASE and supplemented by abstracts from recent oncology conferences (January 2013 to November 2016). Studies meeting pre-defined criteria (such as cancer types of interest, with validation test data) were extracted and key trends summarized.

Results

26 primary studies were identified from a total of 950 records. All the included studies reported data relating to PD-L1 assays in lung cancer using immunohistochemistry (IHC) testing; one study also reported on bladder/urothelial cancer. Significant heterogeneity was reported among the available tests for PD-L1; for example, cut-offs used by test scoring algorithms to define PD-L1 positivity ranged from ≥1% to ≥50%. Studies also differed as to whether they evaluated tumor cells only and/or tumor-infiltrating immune cells. However, all assays developed within clinical-trial settings met recognized acceptance criteria of ≥90% inter-reader concordance. In head-to-head comparisons, there was poor concordance between PD-L1 assays developed outside of clinical trials compared with companion diagnostics developed within clinical trial-settings.

Conclusions

Published data on PD-L1 testing relates predominantly to IHC assays in lung cancer. Review of this evidence suggests the need to develop standardized testing methods, or, minimally, some common approach for test scoring processes. The current variability of such testing could be a source of confusion among treating physicians and might complicate the uptake of effective treatments.

P25 A systematic literature review (SLR) of the predictive value of programmed death ligand 1 (PD-L1) tests on clinical outcomes

Margarita Udall1, Jim Doherty1, Amber Martin2, Juliet Kenney3, Eric Faulkner4, Pratibha Chander1, Paul Robbins1, SueAnn Dahm1
1Pfizer Inc., New York, NY, USA; 2Evidera, Waltham, MA, USA; 3Evidera, London, USA; 4Evidera, Morrisville, NC, USA
Correspondence: Margarita Udall (megancibulas@gmail.com)

Background

Promising trial results have led to FDA approval of several drugs that target programmed cell death protein 1 (PD-1) and its ligand (PD-L1). However, the predictive value of PD-L1 expression on clinical outcomes is unclear. Therefore, an SLR was conducted to evaluate evidence on the predictive value of PD-L1 expression in bladder, gastric, lung, and ovarian cancers.

Methods

The SLR included a search of Medline and EMBASE databases from 1/1/2010 to 15/9/2016 using keywords for PD-L1 paired with terms related to predictive value and relevant abstracts from recent oncology conferences. Studies were selected using pre-defined criteria and then qualitatively summarized.

Results

The SLR identified 29 primary studies. There was limited evidence on bladder (n=4), gastric (n=1), and ovarian (n=1) cancer. Bladder cancer studies showed a trend towards an association between response outcomes and PD-L1 expression, while results were more variable for survival outcomes. Most studies (n=23) were related to lung cancer and were inconsistent. Data for pembrolizumab was most supportive, showing PD-L1 expression is associated with increased probability of response across treatment-naïve, previously-treated, and advanced or metastatic patient subgroups. For the other therapies there was variation across sub-groups. Evidence on the predictive value of combinations of biomarkers was limited to two durvalumab studies, which found improved outcomes when PD-L1 was detected along with additional biomarkers.

Conclusions

Evidence on the value of PD-L1 as a predictive biomarker is currently heterogeneous and relates mostly to lung cancer. While some studies suggest a correlation between PD-L1 expression and clinical outcomes, the SLR indicated that PD-L1 expression alone may not identify those likely to respond to treatment. Factors such as level of expression, cell phenotype, and variability in testing methods are possible reasons for these differences. Ambiguity in the validity of PD-L1 tests adds complexity regarding its use as a standalone biomarker. Further research is needed to explore the relationship between biomarker expression, cancer phenotypes, and treatment outcomes.

P26 Certain KIR/KIR ligand genotypes influence patient response to immunotherapy in neuroblastoma patient

Paul M. Sondel1, Amy K. Erbe1, Wei Wang1, Lakeesha Carmichael1, Anna Hoefges1, KyungMann Kim1, Wendy B. London2, Jacquelyn A. Hank1, Mitchell B. Diccianni3, Arlene Naranjo4, Michael D. Hogarty5, Julie R. Park6, Alice L. Yu7
1University of Wisconsin-Madison, Madison, WI, USA; 2Boston Children’s Hospital/Dana-Farber Cancer Institute, Boston, MA, USA; 3University of California San Diego, San Diego, CA, USA; 4University of Florida, Gainesville, FL, USA; 5Children's Hospital of Philadelphia, Philadelphia, PA, USA; 6Seattle Children's Hospital, Seattle, WA, USA; 7University of California San Diego, San Diego, CA, USA
Correspondence: Paul M. Sondel (aerbe@wisc.edu)

Background

In 2010, a phase III randomized trial (ANBL0032; NCT00026312) conducted by the Children’s Oncology Group (COG), found that patients treated with an immunotherapy regimen of dinutuximab, GM-CSF, IL-2 and isotretinoin had improved event-free survival (EFS) and overall survival (OS) compared to treatment with isotretinoin alone. Dinutuximab (monoclonal/chimeric anti-GD2 antibody) acts in part via engaging NK cells in antibody-dependent cellular cytotoxicity (ADCC). Killer Immunoglobulin-like Receptors (KIR) are a family of receptors expressed by NK cells that can influence their function. The genotypic profiles of inhibitory KIR/KIR-ligands have previously been shown to influence the response of neuroblastoma patients to immunotherapy. We investigated whether EFS and OS were associated with KIR/KIR-ligand genotypes and this immunotherapy regimen.

Methods

Of the 226 pts randomized in ANBL0032, 174 pts had DNA available to genotype for correlations with outcome (isotretinoin alone: n=86; Immunotherapy: n=88; >5yr follow-up if no event). KIR gene status was determined via SYBR green melt curve analysis; KIR-ligands were determined by PCR-SSP reactions using the KIR HLA Ligand SSP typing kit (Olerup). Algorithms were used to assess associations of inhibitory KIRs with their respective KIR-ligands and clinical outcome. Log-rank tests and Cox proportional hazards regression models were used to compare EFS/OS by genotype group; adjustment was made for non-proportional hazards as needed using time-dependent covariates.

Results

In this randomized trial of neuroblastoma patients, we found that certain KIR/KIR-ligand genotypes were associated with significantly improved outcome for patients receiving the COG immunotherapy regimen vs. those that received isotretinoin alone; whereas the patients with the complementary KIR/KIR-ligand genotypes had no significant difference in EFS and OS regardless of whether they received immunotherapy vs. isotretinoin alone.

Conclusions

For this study, certain KIR/KIR-ligand genotypes, but not others, are associated with improved outcome for patients treated with immunotherapy compared to isotretinoin alone. As validation of these KIR/KIR-ligand findings has not yet been performed in other studies comparing patients receiving immunotherapy to those not receiving immunotherapy, further investigation is required. However, if validated, KIR/KIR-ligand genotyping could potentially be used as a predictive biomarker, allowing for administration of this COG immunotherapy regimen to those that might best benefit. Enhancements to anti-GD2 mAb-based therapy, based on preclinical and early clinical data, are being evaluated in efforts to improve its efficacy. Further studies of KIR/KIR-ligand associations with outcome in subsequent trials of immunotherapeutic regimens for children with neuroblastoma will be needed to determine the potential clinical utility of these KIR/KIR-ligand findings.

P27 Optimized workflow improves the characterization of tumor-infiltrating T cells

Cesar Evaristo, Ramona Siemer, David Agorku, Janina Brauner, Olaf Hardt, Christian Dose, Anne Richter
Miltenyi Biotec GmbH, Bergisch Gladbach, Germany
Correspondence: Cesar Evaristo (cesare@miltenyibiotec.de)

Background

Immunotherapy has proven clinical efficacy and tremendous potential in multiple cancers. Syngeneic mouse tumor models represent the gold standard to analyze effects of immunotherapy, as they possess a fully competent immune repertoire. However, the amount and composition of tumor-infiltrating leukocytes (TILs) is highly variable, complicating the analysis of individual subpopulations. In particular, small subpopulations might escape analysis as they could get lost in the background noise. When working with large cohort sizes, even immunophenotyping of TILs by flow cytometry is time consuming and data processing highly work intensive. Therefore, pre-enrichment of TILs is highly desirable to increase the sensitivity of analysis and save time and effort during flow cytometry.

Methods

We have established workflows combining tissue dissociation, T cell isolation and phenotyping. Tumor dissociation was automated using the gentleMACS™ Octo Dissociator and optimized for epitope preservation. Moreover, isolation of tumor infiltrating T cells was improved by developing new CD4-, CD8- and Pan-T cell specific enrichment reagent for magnetic cell sorting, based on MACS® Technology, direct from dissociated tumor tissue. Finally, we used optimized panels of recombinant REAfinity fluorescently-labelled antibodies to phenotypically characterize tumor infiltrating T cells.

Results

Our workflows were validated in B16.F10, B16-OVA, CT26 and 4T1 mouse tumor models. Optimized tumor dissociation was essential for isolation of specific sub-populations of tumor-specific T cells. Tumor-infiltrating T cells were isolated to purities above 80% and yields ranging from 60-95%. Time of downstream analysis was reduced up to 50-fold while enhancing the detection and phenotypic characterization of T cell subpopulations within the tumor.

Conclusions

Standardized processing of tumor samples and magnetic isolation of tumor infiltrating T cells greatly reduces time and cost of downstream analysis while significantly increasing reproducibility and the quality of data obtained from TIL analysis.

P28 Profiling of T cell responses to tumor-associated antigens in lung cancer patients treated with checkpoint inhibitors

Wendy Broom, Kyle Ferber, Johanna K. Kaufmann, Judy Jacques, Christine McCoy, Michael O'Keeffe, Crystal Cabral, Yana Ostrovsky, Ning Wu, Jason Dobston, Theresa Zhang, Pamela Carroll, Jessica Flechtner
Genocea Biosciences, Inc., Cambridge, MA, USA
Correspondence: Wendy Broom (jessica.flechtner@genocea.com)

Background

While the recent approval of checkpoint inhibitor therapies (CPIs) has transformed the therapeutic landscape of advanced lung cancer, the early identification of patients who will benefit from CPIs requires improvement. In addition to neoantigen-specific responses, the success of CPI therapies can in part be attributed to reinvigoration of T cells specific for tumor-associated antigens (TAAs), providing an opportunity to explore using TAA-specific T cell response profiles for patient stratification. Here we utilized ATLAS™, a high-throughput antigen discovery platform that can profile autologous CD4+ and CD8+ T cell recall responses to putative antigens, to profile TAA-specific T cell responses in NSCLC.

Methods

Based on literature review, 77 common antigens were selected to be cloned as full-length proteins in E. coli, either alone or co-expressed with listerolysin O to facilitate MHC II and I presentation, respectively. PBMCs from patients with stage III or IV lung cancer, primarily NSCLC, post-checkpoint inhibitor treatment were analyzed. CD4+ and CD8+ T cells were isolated and non-specifically expanded. Autologous CD14+ monocytes were differentiated into monocyte-derived dendritic cells and co-incubated in an ordered array with the bacterial libraries for antigen processing and presentation. T cells were added and IFN-γ and TNF-α secretion was quantified in the supernatant after 21h, using a custom Meso Scale Discovery kit. Significant antigen-specific recall responses were defined as greater than 3 median absolute deviations from the median response to negative control clones.

Results

ATLAS identified broad TAA-specific T cell responses across the cohort of lung cancer patients, with a greater breadth of CD8+ than CD4+ response, and little overlap between subsets. In addition to T cell activation, antigen-specific inhibition of T cell activity was also observed. T cell response profiles were consistent over at least three weeks in patients for which longitudinal samples were available.

Conclusions

ATLAS is a valuable tool to profile tumor-associated T cell responses in advanced lung cancer non-invasively, providing the opportunity for cohort analysis to identify blood-based signatures that correlate with disease subtype, stage, or responder status to CPI. The consistency of observations over time supports the reproducibility of the platform, validating the utility of ATLAS in anti-tumor response profiling. Future prospective studies will compare T cell response profiles prior to and after CPI therapy to define signatures predictive of clinical response.

P29 Incorporating premixed staining cocktail for improving the workflow for whole blood immunophenotyping

Iliana Gonzalez, William Redmond, Yoshinobu Koguchi, William Miller, Tanisha Meeuwsen, Valerie K. Conrad, Ana Howells-Ferreira, Tomasz M. Poplonski
Earle A. Chiles Research Institute, Portland, OR, USA
Correspondence: Iliana Gonzalez (ilianaleti@gmail.com); William Redmond

Background

Whole blood immunophenotyping is indispensable for monitoring immune responses to cancer immunotherapy. We have successfully introduced an automated liquid handler equipped with a two-dimensional barcode reader for the cocktail preparation that enabled us to prepare antibody cocktails for multicolor flow cytometry for more than 800 fresh whole blood patient samples over last two years. Although it is reliable, our current workflow takes up 2-3 hours to prepare cocktails and is not easily scalable to meet an increasing demand. As a solution, we evaluated the compatibility of custom-made dried cocktail reagents.

Methods

As brilliant violet dyes are not compatible with the dried reagent format, we have re-established the staining panels for T cells, B cells, NK cells, monocytes and dendritic cells with other fluorochromes. Stained cells are analyzed by CytoFlex (Beckman Coulter).

We will evaluate custom-made dried antibody cocktails based on our revised immunophenotyping panels, focusing on the comparative analysis between liquid and dried cocktails for ease of use and the quality and reproducibility of staining.

Results

We have successfully reduced from the current 13-color panels to 10-color ones without compromising the ability to detect immune subsets and their activation status by omitting redundant staining. The comparative study between liquid and dried cocktails is underway.

Conclusions

The dried format allows multicolor antibodies to be dried within the analytical tubes, facilitating their ease of use. Unlike liquid cocktails, dried cocktails require no refrigeration, titration, or manipulation before using. Therefore, it not only saves time but also reduces potential errors and obviates the need to manage individual antibodies. With a shelf life of at least 12 months, the reagents also offer potential savings in reagent costs by reducing waste due to expiration or tandem breakdown in standard liquid formulation.

P30 Mass spectrometry-based test predicts outcome on anti-PD-1 therapy for patients with advanced non-small cell lung cancer, including those with brain metastases

Sarah Goldberg1, Lucia Jilaveanu1, Harriet Kluger1, Veronica Chiang1, Amit Mahajan1, Bing Xia1, Matthew Ribeiro1, Heinrich Roder2, Joanna Roder2, Carlos Oliveira2, Julia Grigorieva2, Mirte Muller3, Anna-Larissa Niemeijer4, Robert Schouten3, Adrianus de Langen3,4, Egbert Smit3
1Yale School of Medicine, New Haven, CT, USA; 2Biodesix, Steamboat Springs, CO, USA; 3Netherlands Cancer Institute, Amsterdam, Netherlands; 4VU University Medical Center, Amsterdam, Netherlands
Correspondence: Julia Grigorieva (julia.grigorieva@biodesix.com)

Background

Checkpoint inhibition has become a standard of care therapy for advanced non-small cell lung cancer (NSCLC), but only a minority of patients achieves a durable response. PD-L1 expression is often used to identify patients likely to have good outcomes. However, patients with low PD-L1 expression can have long-term treatment benefit and this biomarker is dynamic and requires tumor tissue. A less invasive approach that could provide predictive information regarding treatment outcomes could be useful.

Methods

Pre-treatment blood samples were collected from patients with advanced NSCLC. Cohort A (n=98) received nivolumab, 3mg/kg every 2 weeks in a compassionate use program, 59% (32%) in 2nd (3rd) line. Patients in cohort B (n=32) had at least one new or progressing brain metastasis and were treated on a clinical trial with pembrolizumab 10mg/kg every 2 weeks. Spectra were generated from all samples using a sensitive method of MALDI mass spectrometry and processed to render them comparable. Spectra from cohort A were combined with clinical data using modern machine learning methods to create a test able to identify patients likely to have better or worse overall survival (OS), as assessed by reliable test set evaluation methods. The locked test was applied to cohort B, blinded to all clinical data.

Results

Using mass spectral features associated with acute response and wound healing, we produced a test able to identify patients likely to have better or worse OS (HR=0.41, log-rank p=0.005) and progression-free survival (PFS) (HR=0.56, log-rank p=0.022) within cohort A (Figure 1). Test classification was an independent predictor of OS and PFS in multivariate analysis, with 1 year survival of 66% in the good outcome group and 36% in the poor outcome group. Thirty-four percent of patients were classified to the good outcome group. In cohort B, 25% classified to the good outcome group, which had 1 year survival of 88%, compared to 24% in the poor outcome group (Figure 2). The HR for OS between groups was 0.36 (log-rank p=0.062).

Conclusions

We developed a blood-based test able to identify patients with advanced NSCLC likely to have better or worse survival with anti-PD-1 therapy. It showed promise in a validation cohort of patients with NSCLC with progressing brain metastases. Further validation of the test in additional, larger cohorts is planned and necessary to assess clinical utility.

Trial Registration

Clinical Trials.gov Identifier NCT02085070

P31 Proteomic biomarker analysis of metastatic melanoma patients treated with anti-PD-1 checkpoint blockade

Paolo Antonio Ascierto1, Mariaelena Capone1, Antonio Maria Grimaldi1, Domenico Mallardo1, Ester Simeone1, Heinrich Roder2, Krista Meyer2, Senait Asmellash3, Carlos Oliveira2, Joanna Roder2
1Istituto Nazionale Tumori IRCCS Fondazione “G. Pascale”, Naples, Italy; 2Biodesix, Steamboat Springs, CO, USA; 3Biodesix, Boulder, CO, USA
Correspondence: Heinrich Roder (julia.grigorieva@biodesix.com)

Background

Biomarkers identifying who will derive benefit from anti-PD-1 therapy from pre-treatment parameters would further clinical understanding of PD-1 blockade and help development of alternative treatments for patients unlikely to benefit. We evaluate the performance of a serum proteomic test, BDX008, in metastatic melanoma patients treated with anti-PD-1 agents and investigate the role of BRAF mutation status.

BDX008, a pre-treatment test associated with acute phase reactants, wound healing and complement activation, stratifies patients into two groups, BDX008+ and BDX008-, with better and worse outcomes on immunotherapy. Following development and preliminary validation [1], it has been applied to independent cohorts in melanoma [2] and lung cancer [3].

Methods

Pre-treatment serum samples were available from 71 patients (under IRB); 70 pre-treated with ipilimumab. BRAF mutations were identified in 25 patients; 39 were wild type (WT). BDX008 results were generated by Biodesix blinded to clinical data.

Results

Overall, BDX008+ patients had significantly better OS (HR=0.40 (95%CI:0.21-0.76), p=0.005) and PFS (HR=0.54 (95%CI:0.32-0.93), p=0.027) than BDX008- patients. BDX008 was a significant predictor of disease control (p=0.002) and trended to significance for response (p=0.056). Outcomes were numerically inferior for patients with BRAF mutations (median OS/PFS: 160/85 days for BRAF mutations; 479/181 days for BRAF WT respectively). BDX008+ patients had superior OS (HR=0.20 (95%CI:0.07-0.59), p=0.004) and PFS (HR=0.45 (95%CI:0.21-0.99), p=0.046) in the BRAF WT subgroup, but not in the BRAF mutation subgroup (OS: HR=0.62 (95%CI:0.24-1.59), p=0.318; PFS: HR=0.57 (95%CI:0.24-1.40), p=0.223), possibly due to smaller sample size.

Conclusions

BDX008 stratified patients into groups with better and worse PFS and OS. The difference between BDX008+ and BDX008- groups was consistent across BRAF status, but did not reach statistical significance in smaller subgroups. Patients with BRAF mutations may have worse outcomes than those with BRAF WT with anti-PD-1 agents, possibly due to prior targeted therapies, and more previous lines of therapy overall. BDX008 as a possible predictive biomarker warrants further prospective evaluation

References

1. Grossi F, Rijavec E, Biello F, et al: Evaluation of a Pretreatment Serum test for Nivolumab Benefit in Patients with NWeber J, Martinez A, Roder H, et al: Pre-treatment patient selection for nivolumab benefit based on serum mass spectra. J Immunother Cancer. 2015;3(Suppl 2):P103.

2. Weber J, Roder H, Asmellash S, et al. A mass spectrometry-based serum test to predict outcome of treatment with nivolumab: Analysis of samples taken during therapy. Cancer Research. 2016;76:4891.3.

3. Grossi F, Rijavec E, Biello F, et al. Evaluation of a Pretreatment Serum test for Nivolumab Benefit in Patients with Non-Small Cell Lung Cancer. J Thorac Oncology. 2017;12:S1322.

P32 Differential association of myeloid cell and IFN-γ associated proteins with clinical response to durvalumab treatment in urothelial bladder cancer

Xiang Guo1, Wendy White1, Denise Jin2, Yanan Zheng2, Rajesh Narwal1, Yu Gu1, Ashok Gupta1, Pralay Mukhopadhyay3, Brandon Higgs1, Tony W. Ho3, Lorin Roskos1
1MedImmune, Gaithersburg, MD, USA; 2MedImmune, Mountain View, CA, USA; 3AstraZeneca, Gaithersburg, MD, USA
Correspondence: Xiang Guo (andrea.hammons@PAREXEL.com)

Background

Durvalumab, a human monoclonal antibody that binds PD‑L1, was recently granted accelerated approval for the treatment of patients with locally advanced or metastatic urothelial carcinoma (UC) who have disease progression during or following platinum-containing chemotherapy. While PD-L1 expression in tumor or infiltrating immune cells has been valuable in the prediction of clinical responses, new biomarkers and prognostic factors should be investigated to support precision medicine approaches.

Methods

In a phase 1/2 trial, durvalumab was administered intravenously at 10 mg/kg to 182 patients with UC Q2W for up to 12 months. PD-L1 positivity was defined as ≥25% expression in tumor cells or tumor-infiltrating immune cells. Ninety-three serum proteins with inflammatory, growth factor, or metabolic functions were measured by multiplex immunoassay in samples from 158 patients prior to durvalumab treatment.

Results

Univariate analysis identified IL8, CRP, and IL6 to be the most significant prognostic factors among all measured proteins for overall survival (OS) of UC patients receiving durvalumab (log-rank P<0.00001). Their baseline concentrations were significantly higher in progressive disease (PD) compared with stable disease (SD), complete or partial responders (Kruskal rank sum P<0.01). Fast progressors (n=36) who received only 1-2 doses of durvalumab demonstrated higher baseline levels of the 3 proteins versus progressors who stayed on treatment for >2 doses of durvalumab (n=60; Mann-Whitney P<0.001). Similarly, higher macrophage colony-stimulating factor (M-CSF) levels were associated with shorter OS and PD. In contrast, higher IFNγ-inducible proteins, CXCL9 and CXCL10, were associated with longer OS, along with TNF-related weak inducer of apoptosis (TWEAK). Using Cox stepwise proportional hazards modeling, we evaluated the impact of protein markers, PD-L1 status, and various clinicopathological factors on OS for 120 patients with complete data for all variables measured. IL8 levels above median showed the highest hazard ratio (HR=5, P<0.001), while CRP and IL6 were no longer associated with OS when the effects of other factors were taken into account. CXCL9 and TWEAK levels below median had HRs of 3.4 (P<0.001) and 2.9 (P=0.002), respectively, in the multivariate model, which also included low albumin levels (HR=4, P<0.001), liver metastasis (HR=3, P=0.001), and low PD-L1 (HR=2, P=0.03).

Conclusions

Our results demonstrate the association of high serum concentrations of myeloid cell-associated proteins and low concentrations of IFNγ-inducible proteins with shorter OS in UC patients receiving durvalumab. Myeloid-associated proteins may identify patients with myeloid cell-regulated resistance mechanisms to immune checkpoint blockers, which may be further explored as predictive biomarkers for combination therapy with macrophage-targeting agents.

Trial Registration

ClinicalTrials.gov Identifier NCT01693562

P33 The presence of effector immune cells in human CD200 positive tumor samples supports the CD200 immune checkpoint as a novel therapeutic target

Mark Hamilton1, Rui-Ru Ji2, Tony Arulanandam1, Sharon Barr1
1Alexion Pharmaceuticals, Inc., New Haven, CT, USA; 2Alexion Pharmaceuticals, Inc., Lexington, MA, USA
Correspondence: Mark Hamilton (mark.hamilton@alexion.com)

Background

CD200 (OX-2) is an immune checkpoint protein expressed by a number of immune cells (e.g. B-cells, T cells, macrophages), and non-immune cells, (e.g. endothelial cells, neurons). CD200 binds to its receptor (CD200R) expressed on antigen-presenting cells and T cells, and is believed to play an important role in normal immune homeostasis. Overexpression of CD200 by tumor cells implicates CD200 in tumor-mediated immunosuppression and regulation of anti-tumor activity. Samalizumab is a fully humanized therapeutic monoclonal antibody against CD200 and its binding to CD200 disrupts the immunosuppressive CD200-CD200R interaction. Samalizumab is currently under investigation in two Phase 1 trials in patients with acute myelogenous leukemia and solid tumors (NCT02987504 and NCT03013998).

Recent evidence suggests that the presence of immune effector cells within tumors is critical for clinical response to immune checkpoint therapy. Similarly, the expression of the immune modulatory target within the tumor may correlate with response to therapy. The purpose of this study was to identify tumor types which may be most sensitive to samalizumab, based on CD200 expression and the presence of immune effector cell populations within the tumor.

Methods

Publically available tumor gene expression data* was mined to evaluate CD200 expression in multiple tumor types and to develop an immune cell gene signature that may correlate with response to samalizumab. To confirm the gene expression data, we analyzed a series of adult and pediatric tumor sections by immunohistochemistry for expression of CD200 and the infiltration of immune cells.

Results

Gene expression values for CD200 were described as the percentage of samples with expression higher than the median for the set of all tumors. A “samalizumab competent” gene signature was developed from genes which co-correlate with CD200R expression across tumor types, and includes markers of T cells and macrophages; this signature was used to identify tumor types which harbor CD200R-expressing immune infiltrates that may mediate sensitivity to samalizumab. CD200 gene expression and gene signature scores were calculated for all tumor samples and expressed as the percentage of tumor samples which express the signature at a higher level than the overall median. Immunohistochemistry and gene expression data were highly concordant.

Conclusions

We explored both CD200 expression and the presence of CD200R+ immune infiltrates in multiple tumor types. These data point to a role for CD200 signaling in certain tumors, and the potential for anti-tumor activity of samalizumab.

P34 Pre-existing anti-therapeutic antibodies against Fc-region determinants shared by rituximab and hu14.18K322A are associated with outcome in a phase I trial of hu14.18K322A

Jacquelyn Hank1, Jacob Goldberg1, Fariba Navid2, Amy Erbe1, Victor Santana3, Jacek Gan1, Fenna de Bie1, Amal Javaid1, Anna Hoefges1, Lakeesha Carmichael1, Michael Bishop3, Michael Meagher4, Stephen Gillies5, Janardan Pandey6, Paul Sondel1
1University of WI-Madison, Madison, WI, USA; 2Children’s Hospital Los Angeles, Los Angeles, CA, USA; 3St. Jude Children’s Research Hospital, Memphis, TN, USA; 43St. Jude Children’s Research Hospital, Memphis, TN, USA; 5Provenance Biopharmaceuticles, Carlisle, MA, USA; 6Medical University of South Carolina, Charleston, SC, USA
Correspondence: Jacquelyn Hank (hank@humonc.wisc.edu)

Background

Tumor-reactive monoclonal antibody (mAb) therapy has been successful in a growing number of clinical settings, including the use of anti-GD2 mAb as a treatment component for high-risk neuroblastoma. Following treatment with a therapeutic mAb, patients may generate an endogenous antibody response against that mAb. Depending upon the strength and specificity of this antibody against the mAb, this anti-therapeutic antibody response can potentially augment the clearance, or neutralize the molecular functions, of the mAb. For chimeric or humanized therapeutic mAbs, these neutralizing antibodies are often directed against the hypervariable, antigen-binding region (idiotype), of the mAb’s Fab components. Patient derived antibody responses directed against non-variable sites on chimeric, humanized, or fully-human mAbs have been less well studied. We describe the detection of Pre-treatment Anti-Therapeutic Antibodies (PATA) in the sera of patients with neuroblastoma who were treated in a Phase I study of the anti-GD2 mAb hu14.18K322A. We sought to serologically characterize these PATA and evaluate for any clinical correlations.

Methods

Enzyme-linked immunosorbent assays (ELISA) were performed on pre- and post-treatment patient sera to detect the presence of anti-therapeutic antibodies. Sera from PATA+ patients were further characterized in ELISA for reactivity against other known antibodies. The association between the PATA data and clinical outcome data were assessed.

Results

Among samples from 38 patients obtained prior to mAb therapy, 9 (24%) patients demonstrated serological recognition of hu14.18K322A. Unlike the sera from patients who generate post-treatment neutralizing antibody responses, PATA sera did not diminish mAb detection in circulation or inhibit its antigen binding ability in vitro. In 5 of 5 PATA+ patients, IgM is a component of PATA reactivity. Sera from 8 of 9 PATA+ patients demonstrated preferential recognition of the Fc-portions of 14.18K322A and rituximab (both IgG1) as compared with the Fab portions of these mAbs. PATA recognized 14G2a and mouse IgG2a isotype mAbs, but did not recognize mouse IgG1 isotype or the fully human panitumumab (IgG2) mAb. Of the 38 treated patients, only 4 (all PATA+) demonstrated no disease progression for >2.5 years without receiving further therapy (p =0.002; Fisher’s exact test).

Conclusions

This study demonstrates an association between the presence of pre-existing antibodies against a tumor-reactive mAb and a favorable outcome. This association may suggest that PATA are playing a role in augmenting mAb-based anti-tumor effects. Further screening for the presence of non-neutralizing antibodies, analyzes of their clinical correlates, identification of their immunologic targets, and potential anti-tumor mechanism(s) are warranted.

P35

Withdrawn

P36 Computational tissue analysis–based quantification of tumor-infiltrating leukocytes using morphometrics in immunohistochemistry stained NSCLC samples

Elliott Ergon, Allison Harney, Will Paces, Kristin Wilson, Karen Ryall, Daniel Rudmann, Brooke Hirsch, Joseph Krueger
Flagship Biosciences, Westminster, CO, USA
Correspondence: Joseph Krueger (aharney@flagshipbio.com)

Background

Quantification of tumor-infiltrating lymphocytes (TILs) is predictive of patient response to immunotherapy regimens. Because of the role of anti–PD-L1 and anti–PD-1 treatments in promoting tumor-specific cytotoxic T cell responses, the identification and quantification of TILs and PD-L1–positive tumor cells by immunohistochemistry (IHC) may provide important information for disease prognosis. Using Computational Tissue Analysis (cTA™)–based morphometric features instead of an IHC-based lymphocyte assay enables a simpler workflow for TIL assessment and can be combined with additional IHC assays, enabling the identification and classification of immune cell subsets within the complex tumor microenvironment.

Methods

Non–small cell lung cancer (NSCLC) samples were stained with IHC assays for CD45, CD3, CD8, and PD-L1. Serial sections were stained with appropriate isotype-negative controls. cTA™ tools were used to determine the morphometric parameters that identified hematoxylin-stained leukocytes on isotype-stained slides and, once developed, to quantify the total immune cell population in tumor nests and surrounding stroma. The same morphometric ruleset was used to quantify total leukocytes in CD45-, CD3-, CD8-, and PD-L1–stained tissues. The results of this analysis were then compared with the results of the biomarker-specific assay to evaluate the frequency of biomarker-positive TILs in the context of total infiltrating leukocytes quantified by the cTA™ ruleset.

Results

Morphometric parameters were established that determined the number of hematoxylin-stained leukocytes correlated with the number of leukocytes identified by CD45 in serial sections. The relative populations of CD3-positive and CD8-positive TILs were consistent with available literature findings. The cTA™ morphometric ruleset separated PD-L1–positive leukocytes from PD-L1–positive tumor staining in the tumor nests and surrounding stromal tissue.

Conclusions

Methods were developed that use morphometric features to identify the total number of infiltrating immune cells in tumor tissues stained with hematoxylin and eosin or by IHC biomarker assays. These methods provide an additional dimension of data without requiring additional biomarker staining. In some IHC assays, like anti–PD-L1 assays, both the immune infiltrate and tumor cells may stain positive, and the accuracy and precision of quantifying one population by visual pathology is hindered by the staining of the second population. The cTA™-based methods can be used to aid pathologists in the interpretation of these types of biomarker assays in which both tumor cells and immune infiltrate may stain positive

P37 Combining in situ vaccination with checkpoint blockade enhances an endogenous anti-tumor B-cell response resulting in tumor-specific humoral memory

Zachary Morris1, Clinton Heinze1, Claire Baniel1, Peter Carlson1, Jacquelyn Hank1, Raghava Sriramaneni1, Emily Guy1, Jasdeep Kler1, Sara Busche1, Stephen Gillies2, Hans Loibner3, Alan Korman4, Alexander Rakhmilevich1, Paul Sondel1
1Wisconsin Institutes of Medical Research, Madison, WI, USA; 2Provenance Biopharmaceuticals, Carlisle, MA, USA; 3APEIRON Biologics, Vienna, Austria; 4Bristol-Myers Squibb, Sunnyvale, CA, USA
Correspondence: Zachary Morris (cheinze@wisc.edu)

Background

In a murine model of 5-week established (~200mm3) melanoma, we previously reported that combined treatment with radiation and intratumor (IT) injection of anti-GD2 hu14.18-IL2 immunocytokine (IC) results in an in situ vaccine effect, rendering most mice disease-free and eliciting tumor-specific T cell-dependent immunologic memory. In the treatment of larger 7-week (500 mm3) melanomas, combining this in situ vaccine with anti-CTLA-4 resulted in greater complete tumor regression (73%) and enhanced survival compared to dual combinations of radiation, IT-IC, and anti-CTLA-4. All mice (17/17) rendered disease-free by this triple combination therapy exhibited immunologic memory. Here, we evaluate whether this combination may trigger an endogenous anti-tumor antibody response.

Methods

GD2+ B78 melanoma cells were injected subcutaneously on the flank of C57BL/6 mice. After 5 weeks, B78 tumors were treated with 12Gy, IT injection of hu14.18-IL2 on days 6-10 after radiation, and IP injection of anti-CTLA-4 on days 3, 6, and 9 after radiation. Blood was drawn from mice via facial vein bleeds prior to treatment, at 10 day intervals thereafter to day 50, and from disease-free animals >90 days after radiation. Serum from these samples was tested by flow cytometry against the closely related GD2- B16 melanoma (parental to B78) and the unrelated GD2+ Panc02-GD2 pancreatic tumor lines for the presence of tumor-specific IgM and IgG antibodies. Functional capacity of tumor-specific serum was tested using Complement-Dependent Cytotoxicity (CDC) assays.

Results

A tumor-specific endogenous IgG response was observed against GD2- B16 melanoma in untreated tumor-bearing, untreated mice, and levels of this tumor-specific IgG declined over the first ~20 days following combined treatment with 12Gy + IT-IC + anti-CTLA-4. Beginning 20-30 days after this treatment, a tumor-specific IgM response against GD2- B16 melanoma was identified in most mice. T and this briefly peaked and then declined by day 50. A renewed tumor-specific IgG response was observed in serum from mice rendered disease-free with combined in situ vaccine and anti-CTLA-4, and the level of this tumor-specific IgG increased modestly following subcutaneous re-challenge of these mice with B78 cells.

Conclusions

In this preclinical melanoma model, combined treatment with 12Gy + IT-IC + anti-CTLA-4 augments an endogenous antibody response to the B78 tumor, resulting in a memory humoral response in animals rendered disease-free. This may suggest an opportunity to improve this treatment regimen by using the endogenous anti-tumor B cell response as a biomarker of adaptive anti-tumor immunity.

P38 A strategy to assess contributions of individual agents in combination immunotherapy trials

Traci Hilton1, Christopher Paustian1, Yoshinobu Koguchi2, Adi Mehta3, Fridtjof Lund-Johansen4, Brian Boulmay, Rui Li, Kyle Happel, Tarsem Moudgil, Sachin Puri, Christopher Dubay, Brenda Fisher, Rogan Rattray, Mary Campbell
1Ubivac, Portland, OR, USA; 2Robert W. Franz Cancer Center, Earle A. Chiles Research Institute, Providence Cancer Center, Portland, OR, USA; 3Oslo, Norway; 4Oslo University Hospital Rikshospitalet, Oslo, Norway
Correspondence: Traci Hilton (traci.hilton@ubivac.com)

Background

A recent FDA workshop sought discussion on how to demonstrate the contribution of each individual agent in combination immunotherapy trials. Here we outline a strategy of coordinated assessment of B and T cell response kinetics to a complex cancer vaccine as a possible model for such an evaluation. This strategy was applied to assess patients receiving DPV-001 DRibbles®, a dendritic cell-targeted microvesicle (proteasome blocked autophagosome) vaccine derived from adenocarcinoma and mixed histology cancer cell lines. It contains multiple TLR agonists and > 200 potential NSCLC antigens, many as prospective altered-peptide ligands or neoantigens. Based on preclinical studies where anti-OX40 significantly (P<0.05) improved survival and apparent cures, a clinical trial of DPV-001 plus anti-OX40 is planned and will use the proposed strategy to monitor the impact of this combination for cancer immunotherapy.

Methods

Patients received induction cyclophosphamide followed by 7 vaccines at 3-week intervals. The first vaccine was given intranodally; subsequent vaccines intradermally. Patients were randomized to receive DRibble alone (A), or with imiquimod (B) or GM-CSF (C). Thirteen pts were enrolled (Arm A: 5; B: 4; C: 4). PBMCs and serum were collected at baseline and at each vaccination to assess changes in antibodies (Ab) (Protoarray, microsphere affinity proteomics (MAP)), cytokines (Quanterix), PBMCs (flow cytometry) and TCR repertoires (Adaptive immunoSEQ).

Results

IgG levels to 5000 proteins was assessed thrice prior to vaccination and at 3-week intervals. For some antigens, IgG responses peaked and then returned to baseline with new Ab responses developing or being augmented at each time point. In others, Ab responses were maintained at multiple time points. Ab responses were detected against proteins whose genes were commonly upregulated in NSCLC, in some cases this upregulation was associated with significantly reduced survival (TCGA). Evaluation of CD4 and CD8 T cell clones by TCRSeq identified significantly (p=0.002) increased clonal expansion compared to normal controls (n=3). Similar to Ab responses, T cell clonal expansion exhibited expansion followed by apparent contraction.

Conclusions

This monitoring strategy identified that continued vaccination was associated with induction of new IgG Ab responses, inferring that new CD4 T cell responses were developing with repeated vaccination. Expansion and apparent contraction of CD4 and CD8 T cell clones, consistent with basic immunological principles, provides insights into combination immunotherapy strategies that might be used to augment response as well as a method to monitor for that effect.

Trial Registration

ClinicalTrials.gov Identifier NCT01909752

P39 Use of ex vivo histoculture to identify potential predictive biomarkers for the ICOS agonist antibody, JTX-2011

Heather Hirsch, Jason Reeves, Tong Zi, Amit Desphande, Guang Yang, Alexander Needham, Jenny Shu, Christopher Harvey, Sriram Sathyanaryanan, Jennifer Michaelson, Emma Lees, Elizabeth Trehu, Debbie Law
Jounce Therapeutics, Inc., Cambridge, MA, USA
Correspondence: Debbie Law (hhirsch@jouncetx.com)

Background

ICOS (Inducible T cell CO-Stimulator) is a co-stimulatory molecule expressed primarily on T lymphocytes. Clinical and preclinical data suggest that ICOS plays an important role in the immune response to cancer. Therefore, we generated JTX-2011, an ICOS agonist antibody currently in clinical development in advanced solid tumors in the ICONIC trial. In preclinical studies, single agent efficacy correlates with the percentage of ICOS-expressing T cells within the tumor. Thus, ICOS expression is being used as a biomarker to enrich for patients in Phase 2 of the ICONIC trial. Building on our biomarker-driven strategy, we have explored additional potential predictive biomarkers using ex vivo tumor histoculture which allows for in vivo-like analysis of therapies using patient intact tumor tissue. Herein, we report on the results of such analysis, including assessment of the induction of an IFN-gamma gene signature.

Methods

Tumor processing: Fresh human tumor samples were obtained post-surgery through the Cooperative Human Tissue Network. A section of each tumor was cut and fixed for IHC. 300 mM slices of remaining tumor were placed in a 6-well plate. Treatments were added into the medium and plates were incubated at 37°C. Tumor slices were stored in RNAlater after incubation.

RNA extraction and QC: Tumor slices were lysed using Qiagen’s TissueLyser processor and FFPE samples were deparaffinized. RNA was extracted from FFPE and fresh tumor samples, quantified using Quibit, and QC’d using AATI’s Fragment Analyzer.

Gene expression: Gene expression for interferon signature and other genes of interest were performed using Taqman qPCR probes or NanoString nCounter using the Human Immunology V2 panel.

IHC: ICOS (Spring SP98) and PD-L1 (CST E1L3N) levels were assessed by IHC.

Results

Treatment of NSCLC and HNSCC histoculture samples with either JTX-2011 or nivolumab, alone and in combination, led to induction of an IFN-gamma gene signature in a subset of the treated samples. This induction correlated with baseline ICOS expression for JTX-2011-treated samples, and with PD-L1 expression for nivolumab-treated samples. Additionally, expression of an ICOS RNA gene signature correlated with IFN-gamma induction in JTX-2011-treated samples.

Conclusions

Ex vivo histoculture is a robust tool that can be used to assess candidate predictive biomarkers, identify novel potential biomarkers, and interrogate post-dose responses to novel therapeutics. As part of the Jounce Translational Platform, we have used it to build on our JTX-2011 biomarker-driven strategy as we continue to generate new hypotheses to be tested in the context of the ICONIC trial.

P40 Deep proteomic and transcriptomic analysis of sorted T cells with a simple, integrated workflow

Kit Fuhrman1, Douglas Hinerfeld1, Katherine Drake2, Gary Geiss1
1Nanostring Technologies, Inc., Seattle, WA, USA; 2Cytobank, Inc., Santa Clara, CA, USA
Correspondence: Kit Fuhrman (jhorsman@nanostring.com)

Background

In addition to the long-standing efforts to understand and manipulate the immune system in the treatment of autoimmune diseases, the immune system is increasingly becoming a direct target for cancer therapeutics. These cancer immunotherapy efforts require deep biological characterization to better understand and characterize multiple rare cell populations, necessitating next-generation methods to expand or augment current flow cytometry methods. Recent advances in flow and mass cytometry have greatly expanded the number of cell parameters that can be interrogated resulting in an improved understanding of Immune system heterogeneity. These technologies, however, remain limited in the number and types of analytes that can be examined in a single sample.

Methods

We utilized a novel workflow integrating flow cytometry cell sorting with the nCounter® Vantage 3D™ RNA:Protein Immune Cell Profiling Assay to deeply characterize multiple flow-sorted immune cell populations. The NanoString nCounter platform enables the highly multiplexed digital analysis of both RNA and protein from a single biological specimen for multiple research applications. This workflow enabled us to interrogate 30 cell surface proteins and 770 immune-related RNA starting with cells in suspension. We used the unsupervised machine learning algorithms viSNE and SPADE on the Cytobank informatics platform to characterize the expression signatures across all measured proteins and RNA in each of the sorted cell populations.

Results

PBMC were co-stained with both fluorescently-labeled and NanoString DNA barcoded antibodies followed by the isolation of different T cell populations by standard flow cytometry. Using our novel workflow, we subsequently analyzed 30 proteins and 770 RNA from each sorted population with no additional staining. Demonstrating the value of this workflow in analyzing potentially rare cell populations, cells were titrated to determine the sensitivity of the workflow, producing concordant protein data from 5000 cells down to only 500 cells.

Conclusions

High-plex RNA data was obtained without the requirement for additional molecular biology methods, such as RNA purification or sequencing library construction, reducing potential for technical biases. Streamlined data analysis and visualization was accomplished by combining Cytobank with nSolver™ software packages to rapidly analyze this high-plex data. This experiment illustrates ‘3D Flow™’ methodology, which is ideally suited for incorporation into cell sorting workflows, simultaneously producing high-plex, multiomic data from multiple rare and potentially clinically-relevant cell populations.

P41 IgG antibodies correspond with T cell responses to tumor neoantigens

Tyler Hulett1, Shawn Jensen1, Larry David2, Ashok Reddy2, Phillip Wilmarth2, Carmen Ballesteros-Merino1, Christopher Dubay1, Michael Afentoulis1, Bernard Fox1
1Earle A. Chiles Research Institute, Portland, OR, USA; 2Oregon Health & Science University, Portland, OR, USA
Correspondence: Tyler Hulett (tylerhulett@gmail.com)

Background

Herein we report a survey of humoral and cellular immunity to tumor neoantigens which demonstrates interrelationships between IgG antibody and T cell responses specific to individual tumor peptides. Although a high background of preexisting autoantibody signals to thousands of normal proteins are frequently observed in IgG biomarker surveys, little work has been done to determine whether such antibodies might represent a broader immunologic history that enhances CD8+ T cell responses or adapts in concert with them. Improved understanding of any antigen-specific relationships between IgG and T cells could lead to improved immune monitoring for cancer patients and a deeper understanding of what features define clinically-relevant tumor antigens.

Methods

We sought to screen for IgG antibodies to peptides centered at single nucleotide variant and wild-type versions of known mutation sites in our 4T1 vaccine model, and hypothesized that patterns in these antibody profiles would relate to vaccine-induced T cell responses to those same mutation sites. We vaccinated female BALB/c mice with an autophagosome vaccine derived from 4T1 mammary carcinoma cells, then screened serum for IgG binding to a custom array of single nucleotide variant and paired wild-type peptides from those 4T1 cells.

Results

Vaccine-induced IgG signal changes correlated to antigens with stronger-predicted MHCI binding domains, indicating an overlap between the short peptide antigens recognized by antibody and CD8+ T cells. Interestingly, these IgG signal changes occurred on top of a background landscape of preexisting IgG signals. To determine whether these observations related to T cell immunity, we stimulated CD8+ T cells from naïve and vaccinated animals with a diverse selection of the top-predicted MHCI binding minimal 8-11mer peptides matching the mutation sites profiled on the antibody array. Similar to the antibody data, vaccine-induced CD8+ recognition of 4T1 cells was most improved by tumor peptides with stronger-predicted MHCI affinity. In addition, minimal peptides from Wdr33:H13Y, the mutation site garnering the strongest preexisting and post-vaccination IgG signals, also generated the largest post-vaccination IFNγ responses in multiple independent T cell experiments.

Conclusions

These results demonstrate that preexisting and post-vaccine serum antibody signals share important features with antitumor T cell responses. This could have important implications toward understanding why natural surveillance and clinical therapies either succeed or fail in creating immune responses to individual tumor antigens.

P42 The checkpoint inhibitor TTI-621 (SIRPαFc) stimulates innate and adaptive immune responses in patients with hematologic and solid tumor malignancies

Lisa Johnson1, Stephen Ansell2, Robert Chen3, Ian Flinn4, Michael Maris5, Owen O'Connor6, Alexander Lesokhin7, John Thompson8, Oleg Akilov9, Christiane Querfeld3, Matthew Taylor10, Meghan Irwin1, Tina Catalano1, Penka Petrova1, Eric Sievers1, Robert Uger1
1Trillium Therapeutics Inc., Mississauga, ON, Canada; 2Mayo Clinic, Rochester, MN, USA; 3City of Hope, Duarte, CA, USA; 4Sarah Cannon Research Institute/Tennessee Oncology, Nashville, TN, USA; 5Sarah Cannon Research Institute/Colorado Blood Cancer Institute, Denver, CO, USA; 6Columbia University Medical Center, New York Presbyterian Hospital, New York, NY, USA; 7Memorial Sloan Kettering Cancer Center, New York, NY, USA; 8University of Washington/Seattle Cancer Care Alliance, Seattle, WA, USA; 9University of Pittsburgh, Pittsburgh, PA, USA; 10Oregon Health and Science University, Portland, OR, Canada
Correspondence: Lisa Johnson (lisa@trilliumtherapeutics.com); Robert Uger

Background

CD47 is an immune checkpoint that binds to SIRPα and delivers an anti-phagocytic (“do not eat”) signal, suppressing macrophage phagocytosis. Tumor cells frequently over-express CD47 to evade macrophage-mediated destruction. TTI-621 (SIRPαFc) is an immune checkpoint inhibitor that promotes phagocytosis of tumor cells by blocking CD47 and engaging activating FcRs on macrophages. TTI-621 may stimulate T cell responses through enhanced antigen presentation and alteration of the tumor microenvironment. Two Phase 1 studies are underway evaluating the safety and tolerability of TTI-621 following intravenous delivery in subjects with hematologic malignancies (Study TTI-621-01; NCT02663518) and intratumoral injection in subjects with solid tumors and mycosis fungoides (Study TTI-621-02; NCT02890368). We report ongoing pharmacodynamic assessments from these trials.

Methods

In Study TTI-621-01, subjects with relapsed/refractory hematologic malignancies received weekly IV infusions of TTI-621 at 0.2 mg/kg as monotherapy or 0.1 mg/kg in combination with Rituximab. At the investigator’s discretion, the protocol allows dose intensification in 0.1 mg/kg increments to a maximum of 0.5 mg/kg. Serial blood draws were obtained to assess receptor occupancy, cytokine/chemokine levels, T cell repertoire and immune phenotype. In Study TTI-621-02, TTI-621 was administered intratumorally starting at 1 mg/injection up to a maximum of 10 mg/injection to subjects with percutaneously accessible solid tumors and mycosis fungoides. Peripheral cytokines/chemokines, immune phenotype and gene expression were assessed in serial blood draws. Biopsies were collected to assess the impact of TTI-621 on the tumor microenvironment using gene expression, T cell repertoire, and multi-spectral imaging analyzes. Reported data are current as of July 26, 2017.

Results

Subjects receiving IV TTI-621 had increases following infusion in cytokines including MIP-1a, MIP-1b, TNF-a, IL-6, and IL-8 with median peak concentrations of 917.7, 1644.2, 392.2, 262.2, and 400.4 pg/ml at Week 1 and 168.9, 176.4, 38.0, 17.9 pg/mL at Week 6, respectively. NanoString analysis of peripheral blood and tumor tissue pre- and post- intratumoral injection of TTI-621 indicated activation of the innate and complement systems, including modulation of IFN-stimulated genes, such as IFI27. Increases in CD8+ and NK cells were observed in the peripheral blood of subjects. An induction of peripheral T cell clonality was observed in the majority of subjects exhibiting objective responses following IV TTI-621.

Conclusions

Pharmacodynamic analysis demonstrated that TTI-621 stimulates the innate and adaptive immune responses in relapsed/refractory cancer patients. Importantly, changes in the TCR repertoire correlated with objective responses, suggesting that innate checkpoint inhibition with TTI-621 is capable of driving an anti-tumor T cell response.

P43 Assessment of pharmacodynamic effects of immuno-oncology agents in cynomolgus monkeys using high-content gene expression profiling

Veena Kandaswamy1, Amelie Forest1, Thompson Doman2, Manisha Brahmachary1, Krishna Chodavarapu1, Jason Manro2, John Rhoden2, Bing Han2, Nate Elliott3, Sarah Warren3, Tim Riordan3, Christina Bailey3, Ruslan Novosiadly1, Michael Kalos1, Gerald Hall1
1Eli Lilly & Company, New York, NY, USA; 2Eli Lilly & Company, Indianapolis, IN, USA; 3NanoString Technologies, Seattle, WA, USA
Correspondence: Gerald Hall (kandaswamy_veena@lilly.com)

Background

Non-human primate studies serve as valuable tools for preclinical safety and pharmacokinetic (PK) evaluation; however, their utility in Pharmacodynamic (PD) assessment of immunomodulatory agents in immuno-oncology is limited. Here, we assessed PD effects of immunomodulatory agents in Cynomolgus monkeys (Macaca fascicularis) using a high-content gene expression platform (nCounter Non-Human Primate (NHP) Immunology Panel, NanoString) that covers 20 immunologically relevant pathways.

Methods

NHP nCounter codeset used in the study, covers 770 genes and spans around 20 immune-related signaling pathways including but not limited to toll-like receptor, interferon signaling, innate and adaptive immune response. Blood was collected from naïve Cynomolgus monkeys, before and at various time points after intravenous administration of experimental immunotherapies. Based on previously generated Quantigene Plex gene expression data from four different studies, nCounter analysis was focused on select time points/samples along with respective controls. Two different sample types, purified RNA from peripheral blood cells and whole blood lysates, were also evaluated.

Results

nSolver software (NanoString) was used for quality control and normalization of the data. The positive control scaling factor was within the recommended range of 0.3-3, indicating that there were no platform-associated sources of variation. Due to the variability in the negative controls background subtraction was excluded from analysis. Expression levels of housekeeping genes (HKGs) were similar across samples, and all 16 HKGs were included in the analysis. Approximately 300-350 genes were below lower limit of detection under conditions tested and were removed from analysis. The nCounter assay demonstrated comparable performance on purified total RNA and whole blood lysates. Depending on the immunomodulatory agent studied, the number of differentially expressed genes (pre- vs post-dose) ranged from 3 to 76. Principal Component Analysis revealed that gene expression levels post-dose segregated from those of pre-dose. Most of the gene expression changes observed were indicative of peripheral innate immune response exemplified by upregulation of genes attributable to monocytes/macrophages and/or granulocytes.

Conclusions

Taken together, the data indicate that the nCounter gene expression platform was capable of detecting immune-related PD effects of immunomodulatory agents in Cynomolgus monkeys using both RNA and whole blood lysates. High-content molecular platforms such as nCounter, can significantly enhance PD assessment and broaden understanding of immune-related changes in NHP studies which may facilitate informative decision making and PK/PD modeling in clinical trials.

P44 Patient selection strategies and pharmacodynamic assays for CCR4 antagonists

Abood Okal, William Ho, Brian Wong, Paul Kassner, Gene Cutler
FLX Bio Inc., South San Francisco, CA, USA
Correspondence: Paul Kassner (pkassner@flxbio.com)

Background

Regulatory T cell (Treg)-mediated suppression of effector T cells in the tumor microenvironment (TME) can diminish anti-tumor immune responses. The CCR4 receptor can mediate recruitment and accumulation of Treg in the TME making it an ideal target for improving anti-tumor immune responses. We have previously reported on the development of potent and selective CCR4 antagonists. As these move towards the clinic, it is important to have a strategy for selecting patients most likely to respond to this therapy and to measure CCR4 engagement with our inhibitors in these patients once clinical trials begin.

Methods

Data from The Cancer Genome Atlas (TCGA) was mined for expression of relevant genes and signatures. Tumors were classified based on their inferred levels of immune infiltrate, and signatures for Treg and CD8+ T cells. RNAScope® (RNA in situ hybridization) was used on more than 400 tumor biopsy cores to confirm expression levels and localization of key genes. For receptor occupancy, whole blood from healthy donors or cancer patients was incubated with CCR4 antagonist prior to addition of fluorescently-labeled hCCR4 ligand, hCCL22 (A647-hCCL22). At 37°C and in the absence of a CCR4 antagonist, A647-hCCL22 stimulates receptor internalization and accumulation of fluorescence signal. Cells were then antibody-stained and interrogated by flow cytometry to measure CCR4-inhibition in Treg.

Results

Many tumor types show elevated expression of the CCR4 ligands, CCL17 and CCL22, and increases in Treg and CD8 signatures, with head and neck (HN), stomach, lung, and breast cancer types showing strong enrichment. Subtype analysis shows further immune signature enrichment in subtypes such as Triple Negative Breast and EBV+ stomach cancers. RNAScope analysis supports these observations, with lung and HN, in particular, showing strong clustering of Treg and CD8 cells. In the receptor occupancy assay, our CCR4 inhibitors demonstrate dose-dependent reductions in A647-hCCL22-induced internalization. This assay is highly reproducible within donors, on different days, and between operators. This signal stability makes it a robust PD assay for our First-in-Human trials.

Conclusions

Based on RNA expression and in situ hybridization analysis, we hypothesize that patients with a variety of tumors including types of breast, lung and HN cancers, would be good candidates for treatment with CCR4 antagonists. Once these patients are enrolled in clinical trials, robust assays to monitor receptor occupancy and establish pharmacokinetic-pharmacodynamic relationships are ready for deployment.

P45 The inhibitory checkpoint molecule NKG2A is upregulated on tumor infiltrating NK cells and CD8 T cells in human head and neck tumors

Michael Korrer, Young Kim
Vanderbilt University Medical Center, Nashville, TN, USA
Correspondence: Young Kim (michael.j.korrer@vanderbilt.edu)

Background

Immunotherapy has revolutionized cancer therapy by targeting checkpoint molecules found on endogenous immune cells. Presently, the most commonly targeted checkpoint molecules are CTLA-4 and PD-1, which results in response rates of approximately 25% in head and neck squamous cell carcinomas (HNSCC). To improve treatment, additional immune checkpoint molecules expressed in the tumor microenvironment must be identified. Natural Killer Group 2 A (NKG2A) is an inhibitory receptor found on NK cells and CD8 T cells, the receptor for which is HLA-E, a non-classical MHC molecule often overexpressed in solid tumors. This study aimed to identify if NKG2A is expressed on tumor infiltrating NK cells and CD8 T cells from human HNSCC tumors as it is a potential therapeutic target.

Methods

Fresh human HNSCC tumors were digested with miltenyi human tumor dissacotiation kit and Gentle Macs machine following manufactures instructions. Cells were then stained for surface phenotyping or frozen for functional assays.

Results

We analyzed and compared tumor infiltrating NK cells and CD8 T cells from HNSCC patients with matched PBMC by flow cytometry for the expression of activating and inhibitory receptors. We found a unique population of effector memory PD-1+ NKG2A+ CD8 T cells which was absent from the blood. NKG2A+ PD-1+ CD8 T cells expressed higher levels of CTLA-4 and LAG3 as well as produced lower IFNg than NKG2A- PD-1+ CD8 T cells. Interestingly, NKG2A+ PD-1+ CD8 T cells expressed higher levels of both Perforin and Granzyme B, suggesting that these cells are cytotoxically potent. We found a similar upregulation of NKG2A in the NK cell population. NK cells from the primary tumor, but not the blood of HNSCC patients, significantly increased expression of inhibitory KIR2DL4, PD-1 and NKG2A. In addition, we determined that the ligand for NKG2A, HLA-E, was abundantly expressed on CD45+ monocytes and T cells, but not on CD45- cells in the tumor.

Conclusions

We believe that this study provides the first characterization of human tumor-infiltrating NKG2A+ PD-1+ CD8 T cells. We have shown that NKG2A+ CD8 T cells express the highest levels of cytotoxic markers, suggesting an enhanced capacity to kill tumor cells, yet also the highest levels of checkpoint molecule expression. As we have also shown that NKG2A is upregulated in tumor infiltrating NK cells, we believe this makes NKG2A an ideal therapeutic target to improve anti-tumor cytotoxic responses. Ongoing studies are underway to determine the effect of inhibiting NKG2A on tumor immune responses.

P46 The impact of anti-PD-1 treatment on the immune cells and their correlation with the circulating tumor cells in patients with non-small cell lung cancer

Eleni-Kyriaki Vetsika1, Despoina Aggouraki1, Galaktia Kallergi1, Zaharoula Lyristi1, Aristeidis Koukos1, Despoina Kourougkiaouri1, Konstantinos Rounis2, Vassilis Georgoulias1, Athanasios Kotsakis2
1University of Crete, Faculty of Medicine, Heraklion, Greece; 2University General Hospital of Heraklion, Crete, Greece, Heraklion, Greece
Correspondence: Athanasios Kotsakis (thankotsakis@hotmail.com)

Background

The major immune-check point, programmed cell death-1 (PD-1), plays a pivotal role in tumor immune escape. The interaction of PD-1 with its ligand (PD-L1) results in T cells exhaustion, and the blockade of this interaction can partially restore T cell function. Antibodies targeting PD-1 and PD-L1 have been approved for treatment of advanced Non-Small Cell Lung Cancer (NSCLC). In this pilot study, we aimed to investigate the effect of anti-PD-1 treatment on the frequencies of circulating PD-1+ and PD-L1+ immune cells (ICs), as well as their correlation with circulating tumor cells (CTCs) in NSCLC patients.

Methods

Peripheral blood samples were collected from 24 advanced NSCLC patients before and after 1 and 3 cycles of anti-PD-1 treatment. Flow cytometry and immunocytochemistry were used to quantify the changes of the frequencies of PD-1- and PD-L1-expressing ICs and CTCs, respectively.

Results

At baseline, CTCs and PD-L1+CTCs were negatively correlated with PD-1- CD8+ (p=0.03 and p<0.0001, respectively). Moreover, PD-1- CD4+ T had a reverse correlation with PD-L1+CTCs (p=0.04), whereas PD-1+ CD4+ T cells were positively correlated with PD-L1+CTCs (p<0.0001). A significant decrease in PD-1+CD4+ and PD-1+CD8+ T cells was observed after 3 administrations of anti-PD-1 antibody (p=0.01 and p=0.05, respectively). In contrast, PD-1-CD4+ (p=0.05) and PD-1-CD8+ (p=0.05) levels were increased in response to anti-PD-1 therapy. After one dose of anti-PD-1 therapy, these increased levels of PD-1- CD8+ and PD-1- CD4+ T cells were associated with reduced levels of CTCs (p<0.0001). Most importantly, PD-L1+CTCs were reversely correlated with PD-1-CD8+ (p<0.0001) and PD-1-CD4+ T cells (p<0.0001) following 3 doses of treatment. Patients, with non-progressive disease after 3 doses of treatment, had reduced percentages of PD-1+CD4+ (p=0.05) and PD-1+CD8+ T cells (p=0.002), whereas the percentages of PD-1- CD4+ (p=0.04) and PD-1- CD8+ (p=0.004) T cells were increased compared to baseline. Finally, the levels of CD4+PD-L1+ Tregs and PD-L1+G-MDSCs (granulocytic-MDSC) were decreased after 1 dose of anti-PD-1 therapy (p=0.01 and p=0.02, respectively). No correlation was observed between CTCs and immunosuppressive cells at any time point.

Conclusions

These data indicate that anti-PD-1 therapy seems to exert an effect on circulating immune cells and provides evidence for their possible interaction with CTCs via the PD-1/PD-L1 axis, in NSCLC patients. These findings may pave the way for additional studies in a larger cohort in order to document its impact in NSCLC patients.

P47 Gene expression analysis of IL-13Rα2 in human adrenocortical carcinoma correlates with poor survival

Abhinav Kumar, Ian Bellayr, Raj Puri
Food and Drug Administration, Silver Spring, MD, USA
Correspondence: Raj Puri (abhinav.kumar@fda.hhs.gov)

Background

Adrenocortical carcinoma (ACC) is a rare disease where cancer cells develop in the outer layer of the adrenal gland. Because of unreliable early detection, the five year survival rate of subjects with ACC is only 65%. In a variety of cancers including glioblastoma, ovarian, and pancreatic cancers, expression of the interleukin-13 receptor α2 (IL-13Rα2) has been shown to be a prognostic biomarker and potential target for immunotherapy. Using RNA sequencing data from the Genomic Data Commons (GDC), we analyzed the association between IL-13Rα2 gene expression levels with duration of survival and disease outcome in the ACC patients.

Methods

ACC datasets were accessed from the GDC [https://portal.gdc.cancer.gov/legacy-archive] in January 2017. Seventy-nine samples, 48 female and 31 male, were categorized and merged using JMP Genomics. Samples were then evenly distributed into three groups based on low, medium or high levels of IL-13Rα2 expression. Statistical analyzes were performed to compare survival, hormone levels, and tumor reoccurrence with IL-13Ra2 expression.

Results

Subjects with high IL-13Rα2 expressing tumors (n = 27) showed a statistically significant lower survival rate (56%) compared to patients with low IL-13Rα2 expressing tumors (88%) (n = 26) (p= .0135). Subjects with high serum hormone levels (cortisol, estrogen, androgen) exhibited a worse outcome; high IL-13Rα2 expression was associated with a greater incidence of high serum hormone level and poorer survival compared to low IL-13Rα2 expressing subjects. Subjects with new tumor events (n = 35) such as reoccurrence or metastasis exhibited a worse survival rate and high IL-13Rα2 expression showed a higher rate of new tumor events when compared to low expression levels (p= .0042). Individuals with new tumor events exhibited a lower rate of survival, whereas individuals with high IL-13Rα2 expression had a low rate of survival regardless of whether a new tumor event occurred or not. Subjects with metastatic tumors (n = 17) had a lower survival rate compared to those without metastatic tumors, but no correlation with survival for subjects with medium or high expression levels of IL-13Rα2 was observed.

Conclusions

IL-13Rα2 expression levels were a major determinant of outcome in ACC subjects; high IL-13Rα2 levels associated with a lower rate of survival, higher incidence rate and worse outcome with excess hormone production, and more new tumor events. Additionally, subjects with higher expression levels of IL-13Rα2 exhibited a low survival rate regardless of new tumor events or metastases. These results indicate that IL-13Rα2 may be a prognostic biomarker in subjects with ACC.

P48 Different impact of immune cell infiltration and HLA class I expression in lymph node vs. cutaneous/subcutaneous metastases as predictive markers in melanoma patients treated with ipilimumab

Tímea Balatoni1, Anita Mohos2, Eszter Papp1, Tímea Sebestyén3, Gabriella Liszkay1, Judit Oláh4, Anita Varga4, Zsuzsanna Lengyel5, Gabriella Emri6, Soldano Ferrone7, Andrea Ladányi1
1National Institute of Oncology, Budapest, Hungary; 2Semmelweis University, Budapest, Hungary; 3St. John's Hospital, Budapest, Hungary; 4Albert Szent-Györgyi Medical Center, University of Szeged, Szeged, Hungary; 5University of Pécs, Pécs, Hungary; 6Faculty of Medicine, University of Debrecen, Debrecen, Hungary; 7Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA
Correspondence: Andrea Ladányi (ladanyi@oncol.hu)

Background

Immunomodulatory antibodies targeting immune checkpoints are gaining ground in cancer treatment, with simultaneous efforts to search for clinically usable predictive markers which could help making treatment decisions. The aim of our study was to explore tumor-infiltrating immune cells and HLA class I expression as potential biomarkers of response to ipilimumab and survival in patients with metastatic melanoma.

Methods

In pretreatment surgical tissue samples (52 lymph node and 34 cutaneous or subcutaneous metastases from 30 patients), intratumoral immune cell infiltration was determined using antibodies against a panel of 11 markers (CD4, CD8, CD45RO, CD20, CD134, CD137, FOXP3, PD-1, NKp46, CD16, CD68). Tumor infiltration by each cell type was evaluated with regard to response to treatment and patients’ survival. Expression level of HLA class I antigens and its correlation with immune cell densities as well as with response to ipilimumab and disease outcome was also examined.

Results

Immune cell infiltration in lymph node metastases was markedly higher compared to skin or subcutaneous ones in the case of the majority of markers studied (9 of 11). Evaluated separately in the two locations, in the group of lymphoid metastases significantly larger amount of CD4+, CD8+, FOXP3+, CD134+ lymphocytes, CD20+ B cells and NKp46+ NK cells was found in responders compared to nonresponders. The most pronounced differences were seen in the case of FOXP3 and CD8, which also showed association with patients’ survival. Infiltration level of CD45RO+, PD-1+, CD16+ and CD68+ cells correlated with survival but not with treatment response. On the other hand, analyzing subcutaneous/cutaneous metastases revealed significant associations with clinical response or survival only in the case of CD16+ and CD68+ cells. Furthermore, HLA class I antigen expression level of lymphoid metastases, but not that of skin or subcutaneous ones, showed correlation with T cell density and with patients’ survival.

Conclusions

Our results indicate different predictive impact of immune cell infiltration and HLA class I expression in lymphoid vs. non-lymphoid metastases of melanoma patients treated with ipilimumab. This finding points to the potential importance of analyzing metastases of different locations separately in order to reveal all existing predictive associations, part of which could be obscured in the case of joint evaluation because of high variation in immune cell prevalence.

P49 Characterization of immune checkpoint marker expression and infiltrating lymphoid and myeloid immune cells in the tumor microenvironment by RNA in situ hybridization

Annelies Laeremans, Na Li, Jeffrey Kim, Xiao-Jun Ma, Emily Park
Advanced Cell Diagnostics, Newark, CA, USA
Correspondence: Annelies Laeremans (alaeremans@acdbio.com)

Background

Interactions between tumor cells and surrounding immune cells in the tumor microenvironment (TME) play a key role in tumor progression and treatment response with accumulating evidence indicating a crucial role for tumor infiltrating immune cells. Although infiltrating T cells have been correlated with improved clinical outcome, tumor infiltrated lymphocytes are ineffective in eradicating tumors due to their inhibition by immune checkpoint molecules. Cancer immunotherapy blocking the PD-1/PD-L1 immune checkpoint pathway is an established treatment with long-lasting clinical benefits. However, the majority of patients are resistant or relapse after initial response, showing the need for understanding resistance mechanisms and stratifying patients using predictive biomarkers. Also, immunosuppressive factors in the local TME contribute to tumor evasion by recruiting and modulating regulatory T cells and myeloid-derived cells.

Methods

In this study, we used RNAscope® ISH to evaluate in situ expression profiles of therapeutic checkpoint targets in the TME of 30 NSCLC and 30 ovarian cancer archived FFPE tissue samples. In addition, immune infiltration of lymphoid and myeloid cells in the TME was investigated.

Results

Specific checkpoint target molecules were visualized in a highly specific and sensitive manner in individual cells within tissue morphological context. Multiple checkpoint molecules, including PD1, PD-L1, TIM3, and CTLA-4, were detected in the same immune environment, especially in highly inflamed tumors. Unexpectedly, we observed tumor cell-intrinsic expression of TIM3, LAG3, PD-L2, and GITR in a subset of samples. Furthermore, PD1 was often co-expressed with other therapeutic checkpoint targets, including LAG3, TIM3, and TIGIT, in the same infiltrating immune cells. Expression patterns of key functional markers FOXP3, IFNg, CXCL10, and CCL22 combined with cell lineage markers CD4, CD8, CD68, and CD163 demonstrated infiltration of lymphoid and myeloid cells across samples. Recruitment of CXCL10-expressing immune cells in local tumor regions was visualized. The presence of regulatory T cells (FOXP3+CD4+) was evaluated in relation to CCL22 expression in immunosuppressive local environment. IFNg positive CD8 cells were rarely detected.

Conclusions

These findings highlight the utility of RNAscope® ISH in understanding how cancer cells evade the host immune surveillance and ultimately develop resistance against checkpoint blockades. This approach visualizes states of lymphoid and myeloid cells present in the TME by detecting key functional molecules, such as cytokines and chemokines, in addition to immune co-inhibitory and lineage molecules. Considering the highly plastic immune environment with a spectrum of maturation and polarization, characterizing the immunomodulatory phenotype by RNAscope® ISH could offer fundamental insights in contributing factors in immunosuppressive environment.

P50 Application of the immunoscore as prognostic biomarkers in patients with epithelial ovarian cancer

Shin-Wha Lee1, Hee Jung Jung1, Young-Jae Lee1, Ha-Young Lee2, Yong-Man Kim1, Young-Tak Kim1
1Ulsan University, ASAN Medical Center, Seoul, Republic of Korea; 2ASAN Institute for Life Science, Seoul, Republic of Korea
Correspondence: Shin-Wha Lee (swhlee@amc.seoul.kr)

Background

The analysis of single parameters alone may not provide sufficient insights about complex immune system–tumor interactions. This study is to validate the immune contexture as prognostic biomarkers in high-grade serous ovarian cancer (HGS-OC) and to find new era of immunoscore in HGS-OC.

Methods

We collected FFPE samples from 187 patients with HSOC and produced TMA samples. We accomplished the OPAL multiplex IHC assay for the quantitative analysis of immune markers, including CD4, CD8, CD20, FoxP3, PD-L1, and CK. Multiplex Biomarker Imaging and inForm® Image Analysis Software was used.

Results

FIGO stage III and IV patients were 84.5% (158/187). The optimal debulking surgery was done in 66.8% (125/187). The 3-year disease-free survival and 5-year overall survival were 35.1% and 50.0%, respectively. Any single marker was not related to the survival including CD8, FoxP3, and PD-L1. However, high CD8:FoxP3 and CD8:PD-L1 ratios were correlated with the good survival. In cox regression model, the risk factors for HGS-OC survival were FIGO stage (HR 1.784, 95% CI: 1.295-2.457, p<0.001) and platinum resistance (HR 4.257, 95% CI: 2.753-6.582, p<0.001). Additionally, CD8:PD-L1 ratio was a favorable prognostic factor (HR 0.621, 95% CI: 0.042-0.917, p=0.017).

Conclusions

These findings indicate that, although any single immune marker is not related to the survival, CD8:FoxP3 and CD8:PD-L1 ratios provide the positive correlation with the prognosis in HGS-OC. Especially, CD8:PD-L1 ratio is prognostic biomarker which is comparable to clinical biomarkers. The next study for immunoscore is necessary to define immunoscore in ovarian cancer.

P51 FGFR2b expression and baseline immune signature to guide FPA144 development in urothelial cancer

Yeonju Lee, Nerissa Mendoza, Janine Powers, Lee Clark, James Hnatyszyn, Carmen Ladner, Helen Collins, Kevin P. Baker, Majid Ghoddusi
Five Prime Therapeutics, Inc., South San Francisco, CA, USA
Correspondence: Majid Ghoddusi (yeonju.lee@fiveprime.com)

Background

Five Prime Therapeutics, Inc. has developed an FGFR2b-specific humanized monoclonal antibody, FPA144, to treat cancer patients with overexpression of the FGFR2b receptor. FPA144 blocks ligand binding and has been glycoengineered for enhanced antibody-dependent cell-mediated cytotoxicity. Our preclinical models showed FPA144 reprograms the tumor microenvironment by recruiting natural killer cells to the tumor, upregulating PD-L1 expression, and enhancing T cell infiltration [1]. FPA144 is currently in clinical development in gastric and urothelial cancers (UC). FPA144 completed dose escalation in patients with solid tumors and was well tolerated with no dose-limiting toxicity. During dose escalation, a patient with metastatic UC treated with FPA144 at 3 mg/kg exhibited a complete response [2]. Currently, UC patients have enrolled in the dose expansion portion of the trial. We characterized baseline immune cells in the tumor microenvironment and their relationship with FGFR2b in UC patients to guide potential development of FPA144 in combination with other therapies.

Methods

We selected a cohort of archival primary UC formalin-fixed paraffin-embedded whole tumor sections (n=34) with known FGFR2b expression and performed fluorescent IHC on these sections using the proprietary FPR2-D mouse monoclonal antibody, which is specific for FGFR2b. We also performed multi-panel multiplex immunofluorescent assays to quantify immune cell compositions in sequential sections of UC samples.

Results

FGFR2b H scores positively correlate with tumor-infiltrating CD163, PD-L1/CD8 and PD-L1/CD68 cells. From baseline immune compositions and FGFR2b expression level, we clustered cases into three distinct groups: 1) Immune desert (low CD8 T cells, tumor-associated macrophages (TAM), and PD-L1 expression in the whole tumor) with low or negative FGFR2b; 2) Immune excluded (high CD8 T cells, TAM, and PD-L1 expression in tumor-associated stroma) with low to moderate FGFR2b; 3) Inflamed (high CD8 T cells, TAM, and PD-L1 expression in the whole tumor) with moderate to high FGFR2b.

Conclusions

The UC cases with high FGFR2b expression have higher tumor-infiltrating T cells, TAM and PD-L1 expression compared to FGFR2b low or negative tumors. Profiling FGFR2b expression and baseline tumor associated immune cells could help FPA144 development strategy in UC patients. We are currently exploring the utility of combining FPA144 with PD-1/PD-L1 blockade in clinic.

Trial Registration

ClinicalTrial.gov Identifier NCT02318329

Reference

1. Powers J, et al. FPA144, a therapeutic monoclonal antibody targeting the FGFR2b receptor, promotes antibody dependent cell-mediated cytotoxicity and stimulates sensitivity to PD-1 blockade in the 4T1 syngeneic tumor model in mice. AACR. 2016.2. Ghoddusi M, et al. FGFR2b represents a novel target for treatment of urothelial cancer. EORTC-NCI-AACR Molecular Targets. 2016.

P52 Partially exhausted T lymphocyte directed neoadjuvant immunotherapy in unresectable Stage III melanoma

Lauren Levine, Clinton Wu, Kelly Mahuron, Katy Tsai, Alain Algazi, Michael Rosenblum, Michael Alvarado, Adil Daud
Helen Diller Comprehensive Cancer Center, University of California, San Francisco, San Francisco, CA, USA
Correspondence: Lauren Levine (lauren.levine@ucsf.edu); Adil Daud

Background

Programmed death 1 (PD-1) inhibition activates partially exhausted cytotoxic T lymphocytes (peCTL) and induces tumor regression. We have previously demonstrated that the peCTL frequency predicts response to anti–PD-1 monotherapy and combination CTLA-4/PD-1 blockade in metastatic melanoma. However, the utility of this assay in the neoadjuvant setting has not been established. In the current study, 15 patients with unresectable Stage III Melanoma were assessed for response to neoadjuvant anti-PD-1 or anti-PD-1/CTLA-4 inhibition as assigned by peCTL frequency.

Methods

Pretreatment tumor samples from 15 patients with locally advanced melanoma underwent multiparameter flow cytometric analysis. Patients received neoadjuvant anti–PD-1 monotherapy or combination therapy and best overall response to therapy (BOR) as evaluated by Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1) and irAE incidence and severity by NCI Common Terminology Criteria for Adverse Events version 3.0 (CTCAE v3.0) were tabulated by clinicians blinded to immunophenotypic analysis. Patients with low peCTL burden (<20%) received anti-PD-1/CTLA-4 inhibitors whereas those with high peCTL (>20%) were assigned to anti-PD-1 monotherapy. Patients with documented follow up history and evaluable immunophenotype were included in efficacy and safety analysis.

Results

Of 82 patients screened, 15 met criteria for further analysis. Patients were 60% male with mean age of 58 years. Twelve patients had Stage III disease with LDH<ULN, and 13 out of 15 patients were BRAF WT. Prior therapy included CTLA-4 monotherapy in 4 of 12 patients, with only 1 patient with prior PD-1 inhibitor exposure or intralesional therapy. Eleven patients received anti-PD-1 monotherapy and 4 received CTLA-4/PD-1 blockade and were evaluable for BOR at time of analysis. Eleven out of 15 patients achieved complete response, with two partial responses, one patient exhibiting stable disease and one patient with progressive disease per RECIST v1.1 criteria. Of the 9 patients that underwent surgery, 6 attained complete pathologic response. Relapse free survival was 26 months and all patients remained alive at time of analysis. Four cases of disease relapse occurred, without correlation with peCTL frequency. Treatment was well tolerated, with grade 1/2 diarrhea, pruritus, arthralgia, and hypothyroidism most frequently reported. Grade 3/4 adverse effects included two isolated cases of gastritis and anaphylaxis.

Conclusions

Our results suggest the efficacy of peCTL directed neoadjuvant treatment in unresectable Stage III melanoma with the large proportion of patients achieving complete treatment response with acceptable toxicity profile. These promising data from this exploratory analysis merit further investigation with a larger cohort validation study.

Trial Registration

UCSF IRB Protocol 138510

P53 The crosstalk between PD-1, CD137 and OX40 in head and neck cancer

Robert Ferris (jil217@pitt.edu)
UPCI, Pittsburgh, PA, USA

Background

Head and neck squamous cell carcinoma (HNSCC) is one of the malignant diseases with highly immunosuppressive tumor microenvironment, which limits the efficacy of cancer therapies. Tumor infiltrating lymphocytes (TIL) express immune checkpoint receptors in the microenvironment, which can be divided into two groups, inhibitory receptors, such as Programmed Death 1 (PD-1), and costimulatory receptors, such as CD137 and OX40. However, the T cell in TIL exhibit dysfunctional status and the inhibitory receptors has more dominant effect than the costimulatory receptor. The aim of this study was to investigate whether there is crosstalk between PD-1, CD137 and OX40 pathways in TIL expressing multiple immune checkpoint receptors. [1, 2, 3, 4, 5]

Methods

We used PD-L1 (the ligand of PD-1) beads alone or in combination with CD137 or OX40 agonists to treat TIL T cell for 48 hour. Also we used anti-CD3/CD28 beads treated PBL to mimic the TIL environment.

Results

TIL T cells treated with anti-CD3/CD28/PD-L1 beads expressed less CD137 and OX40, and produced less IFN-g than control-treated TIL T cells. Activated PBL T cells treated with anti-CD3/CD28/PD-L1 beads showed similar results as TIL. CD137 or OX40 stimulation could partially reverse this effect.

Conclusions

The result suggest the potential for intracellular signaling cross-talk. We are investigating the pathways is involved in the PD-1, CD137 and OX40 crosstalk.

References

1. Pardoll, D. M. The blockade of immune checkpoints in cancer immunotherapy. Nature Reviews Cancer. 2012; 12(4), 252-264.

2. Ferris, R. L. Immunology and immunotherapy of head and neck cancer. Journal of clinical oncology. 2015; 33(29), 3293-3304.

3. Croft, M. Regulation of PI-3-kinase and Akt signaling in T lymphocytes and other cells by TNFR family molecules. 2013.

4. Chen, S., Lee, L. F., Fisher, T. S., Jessen, B., Elliott, M., Evering, W. & Wang, H. Combination of 4-1BB agonist and PD-1 antagonist promotes antitumor effector/memory CD8 T cells in a poorly immunogenic tumor model. Cancer immunology research. 2015; 3(2), 149-160.

5. Li, J., Jie, H. B., Lei, Y., Gildener-Leapman, N., Trivedi, S., Green, T. & Ferris, R. L. PD-1/SHP-2 inhibits Tc1/Th1 phenotypic responses and the activation of T cells in the tumor microenvironment. Cancer research. 2015; 75(3), 508-518.

P54 Multiplex IHC immuno-oncology panel for standardized profiling of the immune status based on spatial and functional characterization of the tumor microenvironment

Svenja Lippok1, Florian Leiss1, Katrin Schneider1, Tobias Wiestler1, René Korn1, Dasa Medrikova1, Moritz Widmaier1, Martin Hager1, Ivan Kanchev1, Anthony Masci2, Gela Sia2, Chris Kerfoot2, Lisa Dauffenbach2, Ralf Huss1
1Definiens AG, 80636 Munich, Germany; 2Mosaic Laboratories, Lake Forest, CA, USA
Correspondence: Svenja Lippok (slippok@definiens.com)

Background

Characterization of the immune status of patients will become increasingly important to personalize cancer treatments [1]. As the number of treatment options is expected to dramatically increase in the next years, standardized information that guides treatment decision is highly needed. Therefore, we developed a multiplex Immuno-Oncology (IO) panel to profile the immune status of patient tumors into biologically meaningful and clinically actionable categories which support therapy suggestions.

Methods

Three chromogenic IHC assays were developed and optimized for image analysis (PD-L1/CD68/CD3, PD1/FoxP3/CD8, and Granzyme B) using brown (DAB), red and green chromogens for several indications. Definiens’ Cognition Network Technology® was used for spatially resolved image analysis of biomarker expression levels combining rule-based object hierarchy detection with machine-learning strategies. Verifications strategies have been developed in preparation for the application in prospective stratification studies. Marker-positive cells were detected and classified as PD-L1+ tumor cells, (exhausted) CD8+ T cells, CD3+ T cells, regulatory T cells, and macrophages. Global and local densities were computed within regions annotated as Tumor Center and Invasive Margin as well as within automatically detected tumor epithelium (TE) and non-epithelial regions (NE).

Results

Based on the measured densities, NSCLC patients were grouped into one of six possible immune status categories and biomarker profiles have been computed. Additionally, heterogeneity between patients within one category is discussed against the backdrop of therapy suggestions and applicability in further indications. The intra-tumoral biomarker heterogeneity is visualized with cell co-occurrence heat maps and regional context maps, providing insights on inhibition status, effector status, and checkpoint blockade (Fig. 1).

Conclusions

Our IO Panel serves as a ready-to-use approach for standardized immune profiling that quantifies the set of seven and subsequently more biomarkers reproducibly on a continuous scale. Definiens’ Tissue Phenomics® technology combined with Mosaic’s multiplex IHC assays provides next generation spatial and functional resolution of the tumor microenvironment. In the future, we envision applications that facilitate the identification of novel prognostic and predictive signatures, with the goal to support stratification and guide eligibility of patients for a broad range of therapeutics.

Selected maps are shown for two immune status categories belonging to either hot tumors (top) or cold tumors (bottom). A: Light blue: PD-L1 negative T cells, Darker blue: PD-L1 negative T cells (NE), Light Green: Macrophages (TE), Dark Green: Macrophages (NE), Orange: PD-L1 positive T cells, Red: PD-L1 positive tumor cells. B: Red/Light Blue: PD-L1 positive/negative tumor cells, Orange: Macrophages, Green: T cells.

Reference

1. Yuan, et al. Journal for ImmunoTherapy of Cancer. 2016; 4:3
Fig. 1 (abstact P54).

Regional context maps displaying potential checkpoint blockade (A) and PD-L1 landscape (B)

P55 Non-fucosylated anti-CTLA-4 antibody enhances vaccine-induced T cell responses in a non-human primate pharmacodynamic vaccine model

John Loffredo, Raja Vuyyuru, Vanessa Spires, Sophie Beyer, Maxine Fox, Jon Ehrmann, Katrina Taylor, John Engelhardt, Alan Korman, Robert Graziano
Bristol-Myers Squibb, Princeton, NJ, USA
Correspondence: John Loffredo (jtloffredo@uwalumni.com)

Background

Ipilimumab, which blocks the CTLA-4 pathway, is an effective immunotherapy treatment for melanoma and in combination with nivolumab for other malignancies. To improve the potency of this immune checkpoint inhibitor, the Fc region of ipilimumab was altered to enhance antibody-dependent cellular cytotoxicity (ADCC) activity and thereby the potential for increased T-regulatory cell depletion. Here, we assessed the ability of a non-fucosylated form of ipilimumab (anti-CTLA-4-NF) to modulate T cell activity in a non-human primate pharmacodynamics (PD) vaccine model.

Methods

Mauritian cynomolgus macaques (MCMs) were preselected for the high frequency allele Mafa-A1*063 and vaccinated intramuscularly with 2 non-replicating adenovirus serotype 5 (Ad5) viral constructs, encoding simian immunodeficiency virus (SIV) Gag or Nef proteins. Following vaccination, the animals received a single dose of either a vehicle control, ipilimumab, or anti-CTLA-4-NF. Ipilimumab and anti-CTLA-4-NF were administered intravenously at either 1 or 10 mg/kg. Peripheral blood was sampled pre-treatment and then longitudinally over 7 weeks. Vaccine-elicited T cell responses and major T cell subsets were measured by flow cytometry, including the use of peptide-loaded MHC class I tetramers, and IFN-gamma ELISPOT assays.

Results

A single dose of ipilimumab or anti-CTLA-4-NF resulted in augmentation of antigen-specific CD8+ T cell responses at early (2-3 weeks post-vaccination) and later time points (>6 weeks). The resulting peripheral immune responses included T cells directed against the SIV protein inserts as well as against the Ad5 viral vector itself. Both treatments also enhanced circulating CD8+ and CD4+ T cell proliferation as measured by intracellular Ki-67 expression. Anti-CTLA-4-NF demonstrated enhancement of the vaccine-induced T cell responses over that observed in ipilimumab-treated MCMs. This included increased CD8+ CD3+ tetramer+ frequencies of 2 immunodominant SIV-specific T cell responses by flow cytometry (at 3 weeks post-vaccination: Nef RM9 median, 4.16% with anti-CTLA-4-NF vs 1.95% with ipilimumab treatment; Gag GW9 median, 3.08% with anti-CTLA-4-NF vs 1.17% with ipilimumab treatment).

Conclusions

Using this non-human primate PD vaccine model, we observed that immune checkpoint blockade with anti-CTLA-4 augments T cell activation and proliferation in circulating blood. These results suggest that, in addition to the use of non-human primates in toxicology models, the evaluation of antigen-specific T cell responses in MCMs provides meaningful preclinical pharmacological assessment of human drug candidates that target immune modulation. The improved T cell activity observed in these studies also lend support for the use of anti-CTLA-4 antibodies for vaccine enhancement, broadening the potential utility of next-generation anti-CTLA-4 antibodies beyond cancer therapy.

P56 Widespread human T cell receptor variable gene polymorphism revealed by long amplicon TCRB repertoire sequencing: Implications for the prediction and interpretation of immunotherapy outcome

Tim Looney1, Alex Glavin1, Sarabjot Pabla2, Sean Glenn2, Lauren Miller3, Denise Topacio-Hall3, Elizabeth Linch3, Alice Zheng1, Jeffrey Conroy2, Blake Burgher2, Carl Morrison2, Geoffrey Lowman3, Mark Andersen3, Fiona Hyland1
1Thermo Fisher Scientific, South San Francisco, CA, USA; 2OmniSeq, Buffalo, NY, USA; 3Thermo Fisher Scientific, Carlsbad, CA, USA
Correspondence: Geoffrey Lowman (geoffrey.lowman@thermofisher.com)

Background

Human T cell antigen receptors play a critical role in protective adaptive immune responses to infectious disease and cancer, but are also implicated in autoimmune disease and the emergence of severe immune-mediated adverse events during immunotherapy. The antigen specificity of the T cell receptor is determined in part by the sequence of the CDR and Framework regions encoded by the TCRB variable gene. Previous studies of population sequencing data indicate that current antigen receptor allele databases, such as IMGT, fail to capture a significant portion of human variation, though interpretation of this data is challenging. Here we sought to use long amplicon multiplex sequencing of rearranged TCRB receptors to validate putative novel human variable gene alleles previously recovered from 1000 genomes data.

Methods

TCRB rearrangements were amplified from cDNA template from 85 Caucasians undergoing treatment for melanoma using Ion AmpliSeqTM-based multiplex Framework 1 and Constant gene primers to produce ~330bp amplicons. Samples were sequenced in multiplex using the Ion 530TM chip to produce ~1.5M raw reads per sample. Raw data was uploaded to Ion ReporterTM for clonotyping and identification of rearrangements containing variable gene sequences absent from the IMGT database. Putatively novel sequences were compared with those reported in the Lym1k database of alleles recovered from 1000 genomes sequence data.

Results

We identified 15 variants of variable gene alleles, absent from the IMGT database, which result in amino acid changes to the CDR or Framework regions of the T cell receptor. Typically, a single individual was found to be heterozygous for a novel variant, though we note two instances of novel alleles that are found in multiple individuals within this cohort. We also find evidence for novel variable gene alleles that are absent from the Lym1k database, potentially due to challenges in inferring receptor alleles from short-read population sequencing studies.

Conclusions

We find evidence for significant human diversity in TCRB variable gene alleles beyond what is currently represented in the IMGT database. TCRB sequencing using multiplex Framework 1 and Constant gene targeting primers is ideally suited for studying the role of T cell antigen receptor diversity in autoimmune disease and the emergence of immune-mediated adverse events during immunotherapy.

P57 Insights into the tumor microenvironment and therapeutic T cell manufacture revealed by long amplicon immune repertoire sequencing

Geoffrey Lowman1, Tim Looney2, Elizabeth Linch1, Lauren Miller1, Denise Topacio-Hall1, Alex Pankov2, Alice Zheng2, Reidun Hartberg3, Hilde Almåsbak3, Tor Espen Stav-Noraas3, Anette Kullmann3, Fiona Hyland2, Mark Andersen1
1Thermo Fisher Scientific, Carlsbad, CA, USA; 2Thermo Fisher Scientific, South San Francisco, CA, USA; 3Thermo Fisher Scientific, Oslo, Norway
Correspondence: Geoffrey Lowman (geoffrey.lowman@thermofisher.com)

Background

TCRB immune repertoire analysis by next-generation sequencing is emerging as a valuable tool for research studies of the tumor microenvironment and potential immune responses to cancer immunotherapy. Here we describe a multiplex PCR-based TCRB sequencing assay that leverages Ion AmpliSeqTM library construction chemistry and the long read capability of the Ion S5TM 530 chip to provide coverage of all three CDR domains of the human TCR beta chain. We demonstrate use of the assay to evaluate tumor-infiltrating T cell repertoire features and monitor manufacture of therapeutic T cells.

Methods

To evaluate assay accuracy we sequenced libraries derived from 30 well-studied T cell lymphoma rearrangements, then compared our results with those reported by another commercially available immune repertoire sequencing technology. We next used the assay to profile tumor infiltrating T cell repertoires for a cohort of 19 individuals with non-small cell lung cancer. We correlate repertoire features with gene expression profiling data. We then harnessed the long read capability of the assay to profile T cells at various stages of the therapeutic T cell manufacturing process.

Results

Long read TCRB sequencing of 30 reference rearrangements yielded strong linearity in detection of clonal frequencies in reference spike-in experiments. We demonstrate the quantitative nature of the assay by studying populations of counted T cells. Profiling of T cell repertoires in non-small cell lung cancer samples revealed a positive correlation between the number of clones detected and CCR7, CXCR5 and CD3 expression. T cell Evenness (i.e. normalized Shannon Entropy) correlated most strongly with myeloid-specific genes and markers for T cell exhaustion and was anti-correlated with IFNG expression. Sequencing of therapeutic T cells during manufacture revealed a steady increase in the evenness of clone sizes during in vitro expansion and allowed for quantification of clonal expansion.

Conclusions

These results demonstrate: 1) The accuracy and versatility of immune repertoire sequencing using Ion AmpliSeqTM library construction 2) The benefit of combining targeted gene expression and repertoire profiling for studies of the tumor microenvironment 3) The utility of repertoire sequencing covering all CDR regions in monitoring the manufacture of therapeutic T cells.

P58 Public NY-ESO-1 specific TCRs as novel biomarkers for immune monitoring of NY-ESO-1 positive cancer patients

Hailing Lu1, Seth Pollack2, Neeta Somaiah3, Sant Chawla4, Patrick Hwu3, Marissa Vignali5, Julie Rytlewski5, Sacha Gnjatic6, Jan ter Meulen1
1Immune Design, Seattle, WA, USA; 2Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 3MD Anderson Cancer Center, Houston, TX, USA; 4Santa Monica Oncology Center, Santa Monica, CA, USA; 5Adaptive Biotechnologies, Seattle, WA, USA; 6Icahn School of Medicine at Mount Sinai, New York, NY, USA
Correspondence: Hailing Lu (hailing.lu@immunedesign.com)

Background

T cell clonotypes with shared T cell receptors (public TCRs [pTCR]) are involved in the immune response to chronic viral infections, however, their role in immune responses to cancer is largely unknown. We evaluated the association of NY-ESO-1 specific pTCR sequences and survival in solid tumor patients treated with LV305 or CMB305, which are active immunotherapies based on the dendritic cell targeting lentiviral vector platform ZVex®, expressing the cancer-testis antigen NY-ESO-1.

Methods

Peripheral blood mononuclear cells (PBMC) were collected before and after patients with NY-ESO-1 positive solid tumors, including soft tissue sarcomas, received therapy with LV305 or CMB305 (n=64). PBMC were subjected to deep sequencing to study the repertoire of the TCRVB-CDR3 region.

Results

The TCR-VB CDR3 amino acid sequences of three NY-ESO-1 specific pTCR clones obtained through in vitro culture from a LV305 patient with a near complete response were fully conserved in 41/56 (73.2%) of LV305/CMB305 patients and 54% of 539 healthy blood donors. Induction of NY-ESO-1 pTCR on LV305 or CMB305 therapy (baseline negative to positive, or doubling of frequency) was observed in 31% of patients and was associated with a trend towards better overall survival (data as of ASCO 2017). Querying TCR databases from multiple published clinical trials revealed NY-ESO-1 pTCR sequences in blood of patients with melanoma (6/13), renal cancer (1/3), and glioblastoma (6/13), and with a lower incidence in tumor biopsies.

Conclusions

We have identified NY-ESO-1 specific public TCRs in the PBMC of cancer patients undergoing active immunotherapy, as well as in healthy blood donors. In patients, the induction of pTCR appeared to be associated with better survival, whereas their presence in healthy blood donors may indicate frequent low-level baseline T cell immunity against this cancer testis antigen. pTCR should be investigated as a prognostic or predictive biomarker of cancer immunotherapies targeting NY-ESO-1, and possibly other cancer-testis antigens.

P59 Multi-color flow cytometry comparative analysis with DA-Cell(TM) centrifugation-less washing

Xavier Le Guezennec (sheila@curiox.com)
Curiox Biosystems, San Carlos, CA, USA

Background

Sample preparation for multi-color flow cytometry analysis requires staining of suspension cells with a mix of fluorescently labelled antibodies. This preparation requires the need to centrifuge and pack cells multiple times in order to wash away the excess of fluorescent antibody unbound to cells. Here, we explore a novel way to stain cell suspension by using wall less DropArray and Laminar flow wash properties of DA-Cell.

Using PBMC samples stained with various multi-color panels of 6, 11 and 21 antibodies, standard centrifugation or DA-Cell wash method was performed before acquisition with flow cytometry. We then subjected multi-parametric acquired flow data to unbiased clustering analysis with Cytofkit Bioconductor package.

Clustering Analysis and flow cytometry raw acquisition revealed improved segregation of surface markers intensity and cell population when DA-Cell wash method was used as compared to standard centrifugation method. Human annotation of multiple immune cell population type was also facilitated with DA-Cell wash method as compared to standard centrifugation wash method based on stochastic non-linear embedding (T-SNE) clustering map results. DA-Cell method led in unbiased analysis to identification of CD4 CD57 cells and switch memory B cells absent from standard centrifugation wash method.

The superior separation and identification of cell population with DA-Cell offers a new perspective to study various rare immune cell populations.

Methods

Using PBMC samples stained with various multi-color panels of 6, 11 and 21 antibodies, standard centrifugation or DA-Cell wash method was performed before acquisition with flow cytometry. We then subjected multi-parametric acquired flow data to unbiased clustering analysis with Cytofkit Bioconductor package.

Results

Clustering Analysis and flow cytometry raw acquisition revealed improved segregation of surface markers intensity and cell population when DA-Cell wash method was used as compared to standard centrifugation method. Human annotation of multiple immune cell population type was also facilitated with DA-Cell wash method as compared to standard centrifugation wash method based on stochastic non-linear embedding (T-SNE) clustering map results. DA-Cell method led in unbiased analysis to identification of CD4 CD57 cells and switch memory B cells absent from standard centrifugation wash method.

The superior separation and identification of cell population with DA-Cell offers a new perspective to study various rare immune cell populations.

Conclusions

The superior separation and identification of cell population with DA-Cell offers a new perspective to study various rare immune cell populations.

Consent to publish

Consent granted.

P60 Comparative study of flow cytometry analysis between conventional centrifuge method and centrifuge-less DA-Cell™ method from the perspective of cell retention

Xavier Le Guezennec (sheila@curiox.com)
Curiox Biosystems, San Carlos, CA, USA

Background

The preparation of cell suspension samples for flow cytometry commonly involves staining of cells with a specific fluorescent marker for detection of a cell population or a cell phenotype. For over half a century this standard preparation has required the need to centrifuge and pack cells at 500g for multiple times in order to wash away the excess of fluorescent marker unbound to cells. This “worldwide standard” remains however, inherently loses a significant portion of the cell population, particularly when cells are fixed and permeabilized. The biggest challenge imposed by the loss of cells is a well-known belief and intuition that the loss of cells varies depending on cell types and processing. The biggest challenge imposed by the loss of cells is a well-known belief and intuition that the loss of cells varies depending on cell types and processing.

Methods

Here, we present a new convenient methodology to wash suspension cells based on unique laminar flow properties of DropArray plate technology DA-Cell. This technology offers after multiple wash more than 95 % retention of millions of suspension cells and uses natural gravity without cells packing. The cell retention of 95 % can further increase when incubation time of cells and reagents increases by 10 – 20 min as the slight increase of the incubation time helps precipitating cells onto a plate surface by gravity. As an automated method DA-Cell wash is performed conveniently and consistently in 2-4 minutes as compared to a common 20-40 minutes requirement with a centrifugation wash based method.

Results

Application of this new technology on various sample preparation for flow cytometry analysis showed improvements in cell viability, segregation of cell population and detection of rare immune cell population when compared to conventional sample preparation. Various examples of multi-color flow cytometry analysis will be presented to demonstrate the performance of this new technology.

Conclusions

With minimal hands on time, DA-Cell technology enables researchers to conduct complex multi-parametric flow cytometry cell assays without the cellular stress or long time associated with centrifugation based procedures.

Consent to publish

Consent granted.

P61 Cellular and genomic disease signature of peripheral blood mononuclear cells in patients with malignant pleural mesothelioma

Zachary R. Mallon1, Christophe Poulet1, Amanda Enstrom1, Somayeh Honarmand1, Aimee L. Murphy1, Raffit Hassan2, Dirk G. Brockstedt1, Chan C. Whiting1, Nitya Nair1
1Aduro Biotech, Inc., Berkeley, CA, USA; 2National Cancer Institute, Bethesda, MD, USA
Correspondence: Nitya Nair (zmallon@aduro.com)

Background

Recent data on the incidence malignant pleural mesothelioma (MPM) and the continued large-scale use of asbestos throughout the developing world portends an epidemic of asbestos-related disease. MPM is an aggressive and fatal cancer with few treatment options. Recent advances in large scale genomic and high throughput cellular analyzes now provide the tools to more easily attain markers of disease status and potential responsiveness to immunotherapeutics.

Methods

Here we present pre-treatment cellular and genomic biomarker data on a cohort of chemotherapy-naïve MPM patients, and demographically matched healthy donors (HD). MPM patients were enrolled in a Phase 1b study utilizing CRS-207, a live, attenuated Listeria monocytogenes strain engineered to express the tumor-associated antigen, mesothelin. Four different multi-color flow cytometry panels were used to provide resolution on major immune cell populations of T cells, γδ T cells, B cells, dendritic cells, monocytes, and natural killer cells. Together, these panels provided deeper resolution on 39 distinct subpopulations of major immune cell subsets. RNA from these cells was used to perform multiplex gene expression analysis on 770 genes using the Nanostring nCounter PanCancer Pathway Panel.

Results

FACS analysis yielded numerous subpopulations with statistically significant differences between MPM patients and healthy controls. Differences in immune populations were analyzed by median and significant findings included populations of CD4+ T cells, CD8+ T cells, B cells, classical monocytes, and monocytic myeloid derived suppressor cells. Class comparison and hierarchical clustering of gene expression data revealed genomic markers that were significantly expressed in MPM compared to healthy controls. Immune subset deconvolution of gene expression data provided similar findings as FACS analysis and corroborated this disease signature across experimental platforms.

Conclusions

Understanding a patient’s biological disease signature can aide in diagnosis, as well as in making informed choices about therapies amidst the complex and broadening immunotherapeutic landscape. Until recently, existing biomarker data in MPM has been limited to a small number of serological markers and limited immune analysis. Here, we present the first comprehensive report of a MPM disease signature from the cellular and genomic perspectives. Correlation of patient baseline disease signatures with treatment outcome may yield biomarkers predictive of treatment efficacy. Predictive signatures are being investigated in the on-going Phase 1b study of CRS-207 and chemotherapy, as well as in the Phase 2 study of CRS-207 with pembrolizumab in MPM patients who failed prior treatment.

Trial Registration

ClinicalTrials.gov Identifier NCT01675765; NCT03175172

P62 Circulating T cell subpopulations correlate with immune inflammatory signatures at the tumor site in melanoma and non-squamous non-small cell lung cancer

Nataly Manjarrez-Orduño1, Selena Kansal1, Laurence Menard1, Sergey Lezhnin2, Can Jiang1, Paul Kayne1, Robin Edwards1, Suzanne Suchard1, Paul Fischer1, Bijal Kakrecha1, Chiang Yu1, Julie Carman1, Hongyue Dai2, Steven Bernstein1, Steven Nadler1
1Bristol-Myers Squibb, Princeton, NJ, USA; 2M2GEN, Tampa, FL, USA
Correspondence: Nataly Manjarrez-Orduño (nataly.manjarrez@gmail.com)

Background

Checkpoint inhibitors have transformed cancer therapy, although outcomes can be variable. Response to anti-PD1 therapies depends in part on tumor type and characteristics, particularly its mutation burden, as well as pre-existing antitumor immunity. These parameters may constitute the basis for patient stratification strategies for several immuno-oncology therapies. This analysis aimed to characterize the relationship between the local immune environment at the tumor site and peripheral immunity in melanoma and non-squamous non-small cell lung cancer.

Methods

Melanoma (n=42) and non-squamous non-small cell lung (n=40) FFPE tumor biopsies and peripheral blood mononuclear cells (PBMC) were provided by the Moffitt Cancer Center. All patients were appropriately consented through Total Cancer Care (TCC) protocol. None of these patients had received treatment with checkpoint inhibitors. For local immune responses, we used gene expression data in FFPE tissues to calculate inflammatory signature scores in the tumors [1]. For systemic responses, we performed flow cytometry for T cell subpopulations in PBMC. Control PBMC (n=27) were obtained from the BMS employee volunteer blood donation program.

Results

We observed that there are distinct circulating T cell maturation patterns that distinguish between control samples and those belonging to patients with the studied tumors. We detected particular circulating T cell profiles in patients with melanoma and non-squamous lung cancer who have an inflammatory milieu in the tumor as determined by inflammatory signature score.

Conclusions

These findings represent progress in the characterization of peripheral immunity, as it relates to the inflammatory immune status in the tumor. The results also provide potential approaches to measure the immune response which are more easily accessible compared to tumor biopsies.

Reference

1. Spranger, Bao, Gajewski. Melanoma-intrinsic beta-catenin signaling prevents anti-tumor immunity. Nature. 2015;523: 231-235.

P63 Measurement of the immune-suppressor nitric oxide within immune cell subsets in patients receiving ipilimumab with a peptide vaccine

Isaac P. Foote1, Zhihua Chen1, Y. Ann Chen1, Adam W. Mailloux1, Alisha Agrawal1, Braydon J. Schaible1, Amod Sarnaik1, Jeffrey S. Weber2, Anders E. Berglund1, James J. Mulé1, Joseph Markowitz1
1Moffitt Cancer Center, Tampa, FL, USA; 2NYU, New York, NY, USA
Correspondence: Joseph Markowitz (joseph.markowitz@moffitt.org)

Background

Biomarkers are unavailable to accurately predict clinical response to checkpoint blockade therapy. We employed the SPADE (spanning-tree progression analysis of density-normalized events) algorithm to precisely cluster immune cell phenotypes. Because phenotypic identification is laborious (Qiu et al., 2011, Nature Biotechnology, 29(10):887-891), we developed a novel algorithm to address this challenge. We explored whether nitric oxide levels are differentially expressed in distinct immune cell subsets before and after checkpoint blockade therapy, which may serve as a reliable biomarker.

Methods

Patients with resected stage III/IV melanoma were treated with ipilimumab plus a peptide vaccine. Pre and post treatment peripheral blood mononuclear cells (PBMCs) at week 13 were available for analyzes 44 patients pre-treatment samples, 35 patients had both pre/post treatment samples; Sarnaik et al., 2011, 17(4):896-906). Two flow cytometry panels (100,000 live cell events; LSR II flow cytometer) were constructed to compare PBMCs collected pre and post ipilimumab/vaccine treatment (Myeloid panel: nitric oxide stain (DAF-FM), HLADR, CD33, CD11b, CD14, CD15, and CD11c; Lymphoid panel: DAF-FM, CD3, CD4, CD8, CD25, CD127, CD56, CD19, and CD11c). Controls included: flow cytometric compensation beads to establish robust compensation matrices, fluorescence minus one controls to set negative and positive gates, isotype controls to control for patient variations, and a live/dead marker. SPADE was utilized to cluster and differentiate cell populations based on marker expression, and then a novel algorithm, MPAT-R (multi-parameter phenotyping analysis tool in R) was developed to quickly determine cell phenotypes in the SPADE tree.

Results

By the MPAT-R algorithm, 58/200 lymphoid and 116/200 myeloid phenotypes were differentially expressed after ipilimumab/vaccine treatment (Wilcoxon signed-rank test, False Discovery Rate (FDR) <0.05). Ten myeloid and 7 lymphoid phenotypes post-treatment trended with relapse-free survival (RFS) at one year (p <0.05, FDR >0.05, Wilcoxon rank-sum test). Two lymphoid and myeloid populations pre-treatment trended with RFS at one year (Cox regression, Wilcoxon rank-sum test, p <0.05, FDR >0.05). Two interesting populations associated with increased RFS were: a) Monocytic myeloid cells without NO (HLADR+CD33+/lowCD11b+CD14+/lowDAF-FMneg) and b) naïve/central memory T cells containing intermediate levels of NO (CD8+CD127+CD25-DAF-FM+/low).

Conclusions

To our knowledge, this is the first application of SPADE to pre/post clinical immunotherapy PBMC samples. The MPAT-R algorithm efficiently detected immune cell phenotypes associated with increased RFS as potential biomarkers for immune checkpoint blockade therapy with ipilimumab. Future studies will focus on the potential immunosuppressive role of NO identified in precise immune cell subsets both in the circulation and in the tumor microenvironment.

P64 Analytical validation of digital spatial profiling - a novel approach for multiplexed characterization of protein distribution and abundance in FFPE tissue sections

Kristi Barker, Heather Metz, Chris Merritt, Lucas Dennis, Philippa Webster, Joseph Beechem
NanoString Technologies, Inc., Seattle, WA, USA
Correspondence: Chris Merritt (cmerritt@nanostring.com)

Background

Characterization of the spatial distribution and abundance of key proteins within tissues enables a better understanding of biological systems in many research areas, including immunology and oncology. However, it has proven difficult to perform such studies in a highly-multiplexed manner on FFPE tissue sections. To address this unmet need, we have developed a novel imaging platform, Digital Spatial Profiling (DSP), designed to simultaneously analyze 10’s to 100’s of proteins from discrete regions by detecting oligos conjugated to antibodies that can be released via a UV-cleavable linker. Here we describe the validation of the oligo-conjugated antibodies used with this new approach to highly multiplexed protein analysis.

Methods

Various antibodies targeting important immuno-oncology proteins (CD3, CD4, CD8, CD45, PD1, PD-L1, etc.) were tested for specificity and sensitivity. Immunohistochemisty was performed on FFPE human tissues including tonsil and tumor samples, as well as human cell line pellets to evaluate binding specificity in both unconjugated and oligo-conjugated antibodies. The sensitivity and dynamic range of oligo-conjugated antibodies were tested using FFPE cell pellets with target-specific positive and negative cells at different ratios (0:100, 5:95, 20:80, 50:50, 100:0) and specific limits-of-detection (LODs) were determined. An interaction screen was performed to evaluate potential deleterious effects of multiplexing antibodies. Finally, a human tissue microarray (TMA) containing normal and cancer tissues was employed to assess assay robustness.

Results

Immunohistochemical analysis of antibodies integrated into the cocktail displayed indistinguishable staining patterns on control tissues and cell lines for both unconjugated and oligo-conjugated antibodies. Mixed-proportion cell pellet assays revealed strong correlations between observed counts above background and positive cell numbers in a region of interest, allowing us to accurately calculate LODs. For example, CD3 displayed a LOD of 4% when assayed using a cell pellet mixture containing increasing numbers of CD3+ CCRF-CEM cells with CD3- HEK293T cells. Antibody interaction studies showed similar count values for antibodies alone or in combination (R2 > 0.8). Finally, TMA hierarchical clustering analysis demonstrated expected patterns for immune cells and tumor cells across normal and disease tissue types.

Conclusions

These results demonstrate that indexing oligo conjugation does not interfere with antibody specificity and that these conjugated antibodies are robust reagents for quantification of protein abundance. Continued work on the DSP platform will expand the library of antibodies accessible for profiling.

P65 Spatially-resolved, multiplexed digital characterization of protein abundance in FFPE tissue sections: application in preclinical mouse models

Yan Liang, Alison VanSchoiack, JingJing Gong, Chris Merritt1, Dwayne Dunaway, Jaemyeong Jung, Isaac Sprague, Sarah Warren, Joseph Beechem
NanoString Technologies Inc., Seattle, WA, USA
Correspondence: Yan Liang (cmerritt@nanostring.com)

Background

Characterization of the abundance, distribution, and colocalization of key immunoregulatory proteins within the tumor microenvironment is necessary for a thorough understanding of tumor immune responses. Historically, immunohistochemistry and immunofluorescence have been used to assess spatial heterogeneity of proteins in tissue slices, but these techniques are of limited utility due to the challenge of measuring multiple targets in parallel. We recently developed a platform to enable spatially-resolved protein detection with the potential to simultaneously quantify up to 800 targets from a single formalin-fixed paraffin-embedded (FFPE) sample slide, termed Digital Spatial Profiling (DSP). To demonstrate preclinical applications of DSP, we have developed an assay to detect and quantify ~20 key immuno-oncology (IO) targets in mouse FFPE tissue sections.

Methods

DSP uses DNA oligo tags covalently linked to detection reagents (primary antibodies) via a UV photocleavable linker to identify targets in situ and enable quantitation via the standard nCounter® technology. A slide-mounted FFPE tissue section is incubated with a cocktail of oligo-labeled primary antibodies, and a serial section is stained with low-plex visible/fluorescent probes (e.g. nuclear staining probes, or select antibody pairs such as anti-CD3) to generate an image of the FFPE tissue slice morphology. Regions of interest (ROI) in the tissue/tumor are then identified and sequentially illuminated with UV light to release the DNA-oligos. Following UV illumination, the photocleaved oligos are released into the aqueous layer above the tissue slice, collected via microcapillary aspiration, and stored in an individual well of a microtiter plate. Oligos are then hybridized to nCounter optical barcodes to permit ex-situ digital counting of as many as 800 different analytes localized within a single ROI, which can be referenced using image capture software.

Results

We demonstrate preclinical applications of this technology by characterization of a panel of immune proteins on mouse FFPE tumor and normal tissue sections. We demonstrate that this approach enables protein detection at single cell resolution, and enables simultaneous multiplexed detection of AKT, B7-H3, Beta-2M, Beta-Catenin, CD3, CD4, CD8, CD45, CD68, PD-1, PD-L1, GZMB, Ki67, pan-cytokeratin, Stat3 and additional key IO targets.

Conclusions

The ability to measure DNA, RNA, and protein at up-to 800-plex from single slices of FFPE tissue may improve the early evaluation of drug targets with high-resolution spatial information, enable the discovery of key immune biomarkers in mouse tissue (tumors and inflammation et al), and accelerate the preclinical development of immunotherapies.

P66 Digital spatial profiling platform allows both spatially-resolved, multiplexed measurement of solid tumor and immune-associated protein distribution and abundance using a single FFPE tissue section

Youngmi Kim, Chris Merritt, Giang Ong, Kristi Barker, Jaemyeong Jung, Isaac Sprague, Yan Liang, Sarah Warren, Philippa Webster, Dwayne Dunaway, Joseph Beechem
NanoString Technologies Inc., Seattle, WA, USA
Correspondence: Youngmi Kim (cmerritt@nanostring.com)

Background

Spatial characterization of immune activations and tumor proteins within tumor tissues enables a better understanding of immunology and oncology. However, it has proven difficult to perform such studies in a highly multiplexed manner using limited samples. To address this unmet need, we have developed an imaging and tissue-sampling platform designed to simultaneously analyze hundreds of tumor and immune proteins in a single FFPE tissue section with spatial resolution. Using this approach, we demonstrate that a single FFPE tissue section is enough to obtain multi-dimension data sets, such as solid tumor protein expression, immune activity, and cancer pathways with an emphasis on markers important for cancer immunotherapy.

Methods

We developed a novel optical-barcode based microscope that can spatially resolve up to 800 different proteins or RNAs on a fixed tissue (Digital Spatial Profiling, DSP). DSP probes are not multiplex-limited by spectral resolution. Instead, colors are determined using barcode indexing oligos that are conjugated to antibodies or nucleic-acid probes via UV-cleavable linkers. The UV-cleavage is precisely controlled by a programmable digital micromirror device that can illuminate discrete regions ranging from entire tissue microenvironments to single-cell and subcellular regions. Following spatially-defined UV light exposure, released indexing oligos are siphoned off the tissue surface via a microcapillary tip. For quantification of signal, the UV-cleaved oligos from spatially-resolved regions of interest (ROIs) are hybridized to NanoString barcodes, providing digital counts of the protein or RNA targets in each ROI using standard NanoString nCounter® instruments.

Results

Using this novel approach, we demonstrate multiplexed detection from discrete micro regions within a tumor enables systematic interrogation of tumor protein expression and immune activity inner or outer regions of tumors. In addition, the whole tissue analysis allows us to characterize and link global distribution of tumor protein expression, immune activity and cancer pathways.

Conclusions

The simplicity of the DSP platform allows high-resolution, high-multiplexed, spatially-resolved tumor and immune protein characterization in any laboratory capable of performing immunohistochemistry procedures. DSP platform provides a potential method that can bridge the gap between translational research discovery and clinical applications.

P67 Standardization of tumor infiltrating lymphocytes isolation for flow cytometric analysis

William Miller, Yoshinobu Koguchi, William Redmond
Earle A. Chiles Research Institute, Portland, OR, USA
Correspondence: William Redmond (wmgaj.miller@gmail.com)

Background

Cancer immunotherapy has become a mainstay of cancer treatment. These agents can target the tumor/lymphocyte interaction in the tumor microenvironment. Therefore, analysis of tumor infiltrating lymphocytes (TIL) needs to be part of comprehensive immunomonitoring in patients receiving immunotherapy. There are two major approaches to isolate TIL, enzymatic digestion and mechanical dissociation. While enzymatic digestion may give better yields, it potentially destroys epitopes. In contrast, mechanical digestion may preserve epitopes, but can suffer from poor yields and reduced viability of TIL. As standardization of TIL isolation for flow cytometry is key to accurate and precise phenotyping of TIL populations, we compared these different approaches for cell yield, epitope preservation and viability.

Methods

Three different isolation techniques were tested: 1) Miltenyi tumor dissociation kit, 2) enzymatic digest, and 3) mechanical isolation.

Results

The Milenyi protocol and lab prepared enzymatic digests delivered higher yield of TIL from tumor, but proteases in the enzyme mix cleaved proteins and chemokine receptors on the cell surface of T cells. Mechanical dissociation preserved the integrity of the surface antigens and chemokine receptors, but negatively impacted the ability to breakdown the the tumor matrix and release TIL populations into a homogenous single cell suspension. A combination of mechanical dissociation through sieves and 15 minutes enzymatic digest buffered with HSA at room temperature yielded high recovery of Tregs, retention of surface markers and chemokine receptors and good viability.

Conclusions

The combination of enzymatic digestion and mechanical dissociation provided the best result. Due to small tumor size, frequently <1g, and high heterogeneity from tumor to tumor, it is challenging to validate TIL isolation methods for reproducibility and precision. Therefore, we are evaluating a surrogate system using mouse tumors, which allows us to directly compare methods and test proficiency of technologists.

P68 Immunogenomic evolution and atypical response to atezolizumab in a patient with metastatic triple-negative breast cancer (mTNBC)

Luciana Molinero1, Yijin Li1, Ching-Wei Chang1, Sophia Maund1, Maureen Berg2, Jeanne Harrison2, Marcella Fassò1, Carol O’Hear1, Priti S. Hegde1, Leisha A. Emens3
1Genentech, Inc., South San Francisco, CA, USA; 2Johns Hopkins School of Medicine, Baltimore, MD, USA; 3Johns Hopkins University School of Medicine, Baltimore, MD, USA
Correspondence: Leisha A. Emens (molinero.luciana@gene.com)

Background

mTNBC has poor prognosis and few treatment options. Atezolizumab (anti–PD-L1) demonstrated clinical activity in mTNBC, which has been linked to high levels of tumor-infiltrating lymphocytes (TILs) and PD-L1 expression. We describe the immunologic and genomic evolution of TNBC across sequential therapies in a patient with a 31-year history of TNBC and complete response (CR) to atezolizumab.

Methods

A 48-year-old woman enrolled in the Phase Ia atezolizumab study (PCD4989g) on March 5, 2013. She presented in 1986 with early TNBC and had surgery and radiotherapy (XRT). Two locoregional recurrences were managed with surgery and adjuvant chemotherapy. In 2009, she was diagnosed with mTNBC and was sequentially treated with capecitabine, gemcitabine-carboplatin-iniparib (GCI), XRT and an experimental vaccine. She then received atezolizumab for 1 year with partial response (PR), pseudoprogression and reversion to PR. After 1 year off-treatment, she progressed and reinitiated atezolizumab, achieving a CR. Ten tumor tissues obtained between 2008-2015 were assessed by immunohistochemistry, RNA-seq (Illumina) and DNA-seq (FoundationOne).

Results

Tumor-immune microenvironment (TiME) biomarkers, including CD8, TILs and PD-L1, increased after capecitabine exposure, remaining high after GCI and XRT and through pseudoprogression on atezolizumab. At relapse post-atezolizumab, expression of TiME biomarkers decreased, particularly PD-L1. Immune-related RNA signatures (T and B cells, cytotoxicity and antigen presentation) confirmed these findings. Angiogenesis signature varied over time and was highest at early time points, lower post-capecitabine, higher before GCI, lowest at pseudoprogression and increased at relapse post-atezolizumab. Proliferation signatures were highest in early tumors, decreasing over time. TNBC subtyping confirmed evolution of TiME from luminal androgen receptor (LAR) to basal-like immune suppressed (BLIS), then to basal-like immune active (BLIA) post-capecitabine and through pseudoprogression, with reversion to BLIS at relapse post-atezolizumab. Genomic profiling suggested possible driver mutations in RB1 and TP53. MYC amplification was lost over time, possibly associated with decay of the proliferation signature. The incidence of subclonal somatic mutations peaked post-XRT and were undetectable post-atezolizumab. Likewise, tumor mutational burden was highest post-XRT and lowest at pseudoprogression.

Conclusions

This case report describes the evolution of tumor immunity and molecular subtypes with sequential therapies over time in a patient with TNBC and CR to atezolizumab. Immune, stromal and genomic biomarkers of TiME demonstrated temporal plasticity across multiple therapies, including atezolizumab. These findings suggest that TiME is pliable and may be strategically manipulated to maximize response to immunotherapy.

Trial Registration

ClinicalTrials.gov Identifier NCT01375842

Consent to publish

Written informed consent was obtained from the patient for this publication. A copy of written consent is available upon request.

P69 A pan-cancer diagnostic, prognostic and targetable biomarker pipeline defining the failed immune-response

Anne Monette1,2, Antigoni Morou1,2, Louise Rousseau2, Jean-Baptiste Lattouf1, Daniel Kaufmann1, Igor Jurisica3, Rejean Lapointe1,2
1University of Montreal, Montreal, QC, Canada; 2CRCHUM, Montreal, QC, Canada; 3University of Toronto/7Princess Margaret Cancer Centre and Techna, Toronto, ON, Canada
Correspondence: Anne Monette (anne.monette@mail.mcgill.ca)

Background

Tumor infiltrating lymphocytes (TILs) represent positive prognostic and predictive indicators of patient outcomes. Their signature genes are, however, still under tight control by the overwhelming tumor microenvironment. Therefore, despite many of these genes reflecting effector T cell subsets, their totalities actually belong to a failed immune response (FIR) against cancer. This was best first exemplified by association of renal cell carcinoma (RCC) TILs with poor prognosis, yet was later also validated to be a reversible TIL condition by use of immune-checkpoint modulators. Since circulating T cells can share TIL tumor-reactivity and neoantigen-specificity, we used profiles of circulating RCC CD8+ and CD19+ lymphocytes and their paired TILs and normal tissue infiltrating immune cell (TIIC) counterparts as a model of a FIR signature. Commonalties of this signature were observed in several other solid cancers and represents a diagnostic, prognostic, and targetable biomarker model pipeline defining novel and feasible immunotherapeutic targets.

Methods

A rapid extraction protocol isolated RNA from CD8+ effector and CD19+ antigen presenting cells (APC) isolated from paired TILs, TIICs and PBMCs, and matched normal donor PBMCs. Microarray analysis interrogating the expression of all genes and RNA isoforms, was followed by bioinformatics probing an additional 500 RCC validation cohort. This provided prognostic linkage to biomarker expression, revealing novel antagonistic and agonistic targets, which were refined using clinical datasets from 11,000 additional patients with lung, breast, gastric, and ovarian cancers. Feasible novel immune-promoting targets exclusively expressed by immune cells, and in the context of cancer, included receptors and plasma membrane-associated proteins, or those having pre-existing targeting small molecules. These were further investigated for protein-protein interaction (PPI) and associated pathways, coexpression, and comprehensive literature reviews establishing relevant immune-cell function.

Results

Validation of original hits from the discovery cohort provided strong evidence of which cancers were most related at the TIL level, and PPI revealed that pan-cancer targets were functionally associated in novel pathways (p = 1.85e-10). Coexpression dynamics reflected key proteins responsible for APC:effector TIL communication, and results also showed that lung and breast cancer TILs and circulating PBMCs were most similar to other cancers, and that a FIR is skewed towards a diminished expression of genes otherwise expressed by the healthy population.

Conclusions

We present a strict, comprehensive pipeline defining a pan-cancer targetable immune-signature having many pre-existing and tested targeting small-molecules, in an effort to develop more feasible and broad reaching diagnostics and precision medicine to counter cancer, foiled by the surge of bystander TIL-signatures.

P70 Halioseek™, a dual CD8 and PD-L1 IVD assay to improve NSCLC patients stratification

Florence Monville1, Emmanuel Prestat1, Laurent Vanhille1, Julien Adam2, Sergiu Coslet3, Caroline Laugé1, Caroline Davin1, Merle Sophie Losch1, Luciana Batista1, Jacques Fieschi1
1HalioDx, Marseille, France; 2Gustave Roussy Cancer Campus, Villejuif, France; 3MImAbs, Marseille, France
Correspondence: Florence Monville (fabienne.hermitte@haliodx.com)

Background

Immune checkpoint inhibitors (ICI) improved non-small cell lung cancer (NSCLC) survival in 20% of the patients. Their administration is decided based on the expression level of PD-L1 on tumor cells despite limited predictive value of PD-L1 alone. We hypothesized that ICI action depends on both PD-L1 expression and on the presence of tumor infiltrating lymphocytes (TILs) in the tumor microenvironment.

We developed Halioseek™ according to IVD regulatory processes, a standardized PD-L1/CD8 dual-staining assay. In addition to PD-L1 detection, critical information is provided on TILs through the detection of CD8+ cells on the same tissue section. Halioseek™ includes a Digital Pathology (DP) analysis module to determine CD8+ cell density and a proximity index between CD8+ and PD-L1+ cells. Here we show the main analytical performance of Halioseek™ and concordance with two PD-L1 IVD assays.

Methods

Samples are double-stained with anti-PD-L1 (clone HDX3) and anti-CD8 (clone HDX1) antibodies, respectively revealed with DAB and fast-red substrates. The assay is fully automated on the Benchmark XT platform (Ventana). Digital images of stained slides are obtained using a whole slide scanner (Hamamatsu). Percentage of PD-L1 positive tumor cells (PD-L1+ TC) is estimated by a pathologist; DP analysis is performed using Halioseek™ DP module.

Staining accuracy of PD-L1 by HDX3 antibody was evaluated by comparison with SP263 IVD assay (Ventana) and 22C3 IVD assay (Dako) on 100+ NSCLC including resections and biopsies. Overall, Positive percent and negative percent agreements were assessed at 1%, 5%, 10% and 50% of PD-L1+ TC corresponding to published clinical cut-offs.

The precision of PD-L1 assay was evaluated in terms of repeatability, inter-primary antibody batches (3 lots), inter-revelation kit batches (3 lots) and inter-instruments (2 stainers) using 332 slides from 11 independent NSCLC samples covering the range of the assay.

Results
  • PD-L1 staining accuracy: overall agreements with SP263 were above 90% and with 22C3 above 85%.

  • PD-L1 precision: none of the tested parameters had a significant impact on results showing the robustness of the assay.

  • CD8+ cell density: distribution spanned between 30 and 1200 cells/mm2 ; global CV was below 13%.

Conclusions

Halioseek™ is a new robust IVD assay leveraging the advantages of DP to combine TILs and PD-L1 quantification within the tumor microenvironment. Halioseek™ could have a higher predictive performance than existing IVD tests and could fill a major gap in the management of ICI administration. As next step we intend to investigate predictive performance of the assay on ICI treated patient’s samples.

P71 Peptide-induced whole blood gene expression analysis for peptide-specific T cell response

Taku Murakami1, Keiji Dohi2, Kenji Kuwabara2, Mieko Ogura1, Masato Mitsuhashi1
1Hitachi Chemical Co. America, Ltd., Irvine, CA, USA; 2Shionogi & Co., Ltd., Osaka, Japan
Correspondence: Taku Murakami (tmurakami@hitachi-chemical.com)

Background

Peptide-based cancer vaccine is a promising approach in cancer immunotherapy. To monitor the efficacy of therapy, a simple tool to monitor specific T cell response to peptide vaccine is in need. Although MHC multimer and Elispot have been used frequently in various vaccine studies, the former requires specialized reagents and instrument and the latter requires long incubation and imaging instrument.

Methods

Whole blood ex vivo peptide stimulation assay was developed to investigate peptide-specific T cell responses. A 60 μL heparin blood was incubated in triplicate with class I peptide for 4 hours at 37°C to form a complex among major histocompatibility complex (MHC) – peptide – T cell receptor (TCR). Following the incubation, mRNA was isolated and quantified by real-time RT-qPCR. Fold increase of mRNA was calculated against solvent control in triplicate.

Results

We identified that IFNG, CXCL10 and GMCSF mRNA were significantly induced (fold increase >2, p value <0.05) by CMV peptide in 4, 4 and 2 out of the 8 healthy donors, respectively. IFN-gamma Elispot assay confirmed at least 7 of the subjects were positive following 2-week in vitro sensitization. On the other hand, these genes were not induced by HIV peptide, which was corroborated by Elispot with in vitro sensitization. To compare the assay performance with Elispot, 0 to 300 cells of CMV-peptide specific cytotoxic T-lymphocyte (CTL) were spiked in 1x105 cells of PBMC and assayed by both methods. The limit of detection was less than 3 to 10 cells for both assays and a linear correlation between mRNA quantity and Elispot count was obtained (Pearson correlation; IFNG, -0.942 (p = 4.9x10-6), CXCL10, -0.690 (p = 1.3x10-2), and GMCSF -0.888 (p = 1.1x10-4)). In a study with 12 samples of whole blood blindly spiked with peptide-specific CTL at 2.5 to 12.5x103 cells/mL, the assay sensitivity and specificity of each mRNA were 17% and 83% for IFNG, 100% and 100% for CXCL10, and 100% and 33% for GMCSF. These data suggest that these markers especially CXCL10 could be a useful marker to monitor peptide-specific T cell response.

Conclusions

The whole blood mRNA assay developed in this study is useful to measure peptide-specific T cell responses, which could be used to monitor the efficacy of peptide vaccine therapy.

P72 Analysis of biomarkers from a cohort of advanced melanoma patients previously exposed to immune checkpoint inhibition treated with entinostat (ENT) and pembrolizumab (PEMBRO)

Melissa L. Johnson1, Dmitry Gabrilovich2, Fang Wang2, Rene Gonzalez3, Mateusz Opyrchal4, David Tamang5, Emmett Schmidt6, Michael L. Meyers7, Peter Ordentlich5, Sanjiv S. Agarwala8
1Sarah Cannon Research Institute, Nashville, TN, USA; 2The Wistar Institute, Philadelphia, PA, USA; 3University of Colorado Comprehensive Cancer Center, Aurora, CO, USA; 4Roswell Park Cancer Institute, Buffalo, NY, USA; 5Syndax Pharmaceuticals, Inc., Waltham, MA, USA; 6Merck & Co., Inc., Kenilworth, NJ, USA; 7Syndax Pharmaceuticals, Inc., New York, NY, USA; 8St. Luke’s Cancer Center and University Health Network, Easton, PA, USA

Background

Treatment options following immune checkpoint inhibition remain an area of active investigation across multiple cancer indications, including advanced melanoma. Results from the ongoing ENCORE 601 study demonstrated that ENT, a class I selective HDAC inhibitor, combined with PEMBRO elicited a 31% response rate in a cohort of 13 patients with melanoma progressing on or after a PD-1 blocking antibody. Assessment of blood and tissue-based biomarkers may be useful to understand how patients who respond to ENT plus PEMBRO differ from those who do not.

Methods

ENCORE 601 employs a Simon 2-stage design to assess activity of ENT 5mg PO weekly combined with PEMBRO 200 mg IV every 3 weeks across 4 cohorts: 1) advanced anti-PD-(L)1–naïve NSCLC; 2) advanced NSCLC progressed on anti-PD-(L)1; 3) advanced melanoma progressed on anti-PD-(L)1; 4) advanced anti-PD-(L)1–naïve microsatellite stable colorectal cancer. Pre-treatment and Cycle 2 Day 15 (C2D15) biomarker analysis included gene expression by Nanostring Pan-Cancer Immune Profiling Panel, PD-L1 expression by immunohistochemistry, and levels of CD8+ T cells and LOX1+CD15+ MDSCs by immunofluorescence staining on paraffin embedded human tissue sections. Phenotypic evaluation of immune cell subsets was conducted in peripheral blood samples collected pre-treatment, C2D1 and C2D15.

Results

All patients enrolled in Stage 1 of the melanoma cohort (n=13) received a prior anti-PD-1 therapy with eight also receiving prior ipilimumab, and two receiving prior BRAF inhibitors. Four patients (31%) had a partial response (PR; 3 confirmed, 1 unconfirmed), and four additional patients (31%) had stable disease. Pre-and post-treatment tissue samples were evaluated for PD-L1 expression (n=10 pre; 7 post), gene expression (n=7 pre; 7 post), CD8+ T cells (n=9 pre; 9 post) and MDSCs (n=9 pre; 9 post). In 3 of the 7 patients whose tissue samples were analyzed for gene expression (2 with confirmed PRs, one with progressive disease), non-inflamed tumors at pre-treatment were converted to inflamed. Increases in both T cells and MDSCs were observed only in those tissue samples converted from non-inflamed to inflamed. Overall, regardless of clinical outcome, we noted a decrease in MDSCs (-35.7%; n=9) and an increase in CD8+ T cells (47.4%; n=9) between pre-and post-treatment biopsies.

Conclusions

ENT plus PEMBRO shows promising activity with an overall response rate of 31% observed in Stage 1. Enrollment to Stage 2 continues. Preliminary biomarker analysis supports our hypothesis that the addition of ENT restores inflammation in the tumor microenvironment necessary for successful re-treatment with an anti-PD-(L)1 blocker.

P73 First-in-human neoadjuvant study of the immunogenomic impact of the oral IDO inhibitor epacadostat (INCB024360) on the tumor microenvironment of advanced ovarian cancer

Junko Matsuzaki1, Amit Lugade1, Wiam Bshara1, Katherine Collins1, Sarah Warren2, Patrick Danaher2, Lucas Dennis2, Antonios Papanicolau-Sengos1, Leonard D’Amico3, Nirasha Ramchurren3, Angela Omilian1, Kevin Eng1, Sebastiano Battaglia1, Robert Newton4, Peggy Scherle4, Lance Leopold4, Steven Fling3, Richard Shine3, Martin Cheever3, Melissa Geller5, Kunle Odunsi1
1Roswell Park Cancer Institute, Buffalo, NY, USA; 2Nanostring, Seattle, WA, USA; 3CITN, Seattle, WA, USA; 4Incyte Corporation, Wilmington, DE, USA; 5University of Minnesota, Minneapolis, MN, USA
Correspondence: Kunle Odunsi (kunle.odunsi@roswellpark.org)

Background

The tryptophan catabolizing enzyme indole-amine 2,3 dioxygenase 1 (IDO1) has been identified as a potent immunosuppressive mechanism that fundamentally alters the ovarian tumor microenvironment (TME) and effector T cell function in ovarian cancer. Epacadostat is a novel oral IDO1 inhibitor that suppresses systemic tryptophan catabolism and is currently being evaluated in ongoing clinical trials. Although the pharmacodynamics (PD) effects of epacadostat in reducing kynurenine concentration in plasma has been demonstrated in pre-clinical and clinical studies, it is currently unknown whether the PD effects will occur at the TME and alter it to become more immunogenic.

Methods

Seventeen patients with newly diagnosed stage III or IV ovarian cancer underwent pre-treatment tumor biopsy and paracentesis. Subjects were scheduled to receive 600mg epacadostat twice daily for 2 weeks, followed by standard tumor debulking surgery. Peripheral blood and tumor specimens were utilized to assess the phenotype and functional status of multiple immune populations in the TME, define changes in the tumor genomic profile and gene expression by whole-exome sequencing, Nanostring, and RNAseq, TCR clonal evolution, and impact on humoral immune responses to cancer-testis antigens.

Results

Expression profiling identified an upregulated IFNγ induced gene signature in a subset of post-treatment tumor specimens, indicative of reactivation of IFNγ signaling in the TME. Likewise, in increase in CD8+ T cells was detected in some, but not all subjects. Of note, IFNγ signaling and increases in CD8+ frequency was not consistently observed in all subjects. Evidence of reduced IDO1 enzyme activity, determined by Kyn:Trp ratio, was observed not only in plasma but also in ascites fluid for a subset of patients (5 of 7 evaluable ascites fluid specimens). Although IDO1 activity was blocked, IDO1 expression within the TME was not affected by epacadostat. Mitochondrial spare respiratory capacity of T cells in peripheral blood increased in 8 of 12 patients after treatment. Additional immunogenomic analyzes are underway to identify molecular and metabolic pathways impacted by epacadostat along with determining the clonal evolution of T cells with reactivity to mutational antigens.

Conclusions

This is the first demonstration of the effect on tumor tissue of inhibition of IDO1 enzyme activity. IDO1 inhibition by epacadostat resulted in an increase in CD8+ T cell and IFNg gene signatures in the TME. Additional multi-dimensional immunogenomic data from on-going analyzes will be presented.

Trial Registration

ClinicalTrials.gov Identifier NCT02042430

P74 Increasing the levels of anti-beta glucan antibodies by administration of intravenous immunoglobulin (IVIG) induces immunopharmacodynamic (IPD) responses of a novel immunotherapeutic Imprime PGG

Nadine Ottoson1, Keith Gorden1, Blaine Rathmann1, Xiaohong Qiu1, Ben Harrison1, Michele Gargano1, Nandita Bose1, Gautam Jha2, Arkadiusz Dudek3
1Biothera Pharmaceuticals, Eagan, MN, USA; 2The University of Minnesota, Minneapolis, MN, USA; 3HealthPartners Institute, St. Paul, MN, USA
Correspondence: Nadine Ottoson (nottoson@biothera.com)

Background

There is a critical need for rational combination immunotherapies that have mechanism-driven predictive biomarkers. Imprime PGG (Imprime), is a novel, intravenously (i.v.) administered innate immunomodulator currently in clinical development as a combination therapy with checkpoint inhibitors in biomarker-selected patients. Imprime is a soluble β-glucan PAMP that requires immune complex formation with serum anti-beta glucan antibodies (ABA) for its functionality. Ex vivo human studies, a healthy volunteer phase I trial and retrospective analyzes of clinical studies have demonstrated that IPD changes and clinical responses mediated by Imprime correlated with serum ABA levels. Ex vivo studies have also shown that innate immune functionality in subjects with lower ABA values can be restored by supplementation with purified ABA or ABA-containing IVIG. Herein we present a case study of a cancer patient with low ABA levels demonstrating enhanced Imprime-induced PD responses post IVIG administration.

Methods

A 54-year-old female with metastatic colorectal adeno-carcinoma was offered immunotherapy with bevacizumab, cetuximab and Imprime as part of a compassionate use program after she could not tolerate first line therapy with FOLFOX and bevacizumab. The patient was dosed in 4 week cycles for 12 cycles. Imprime and cetuximab were administered i.v. weekly. Bevacizumab was administered every 4 weeks from cycle 2 through 8. Evaluation of serum ABA levels from cycles 1-6 confirmed low values in this patient. To boost ABA levels, IVIG was added to dosing beginning at cycle 6. ABA, complement, and cytokine levels in serum (ELISA and Luminex), Imprime binding and immune cell phenotyping (flow cytometry) were measured prior to and within 30 minutes after Imprime administration.

Results

Compared to the weeks prior to adding IVIG to the dosing regimen, IVIG infusion resulted in increased serum ABA levels at the End of Infusion (EOI), which then dropped to the baseline levels in the subsequent weeks. Concomitant with the ABA increase, serum C5a levels peaked at the EOI. Furthermore, serum chemokines such as IL-8 also increased at EOI with the most pronounced change observed during cycles 7 and 8. Increased ABA levels also correlated with significant Imprime binding on neutrophils and monocytes. Importantly, minimal PD changes were observed with Imprime dosing alone in the cycles prior to IVIG administration. Disease remained stable for 10 months.

Conclusions

These human data provide the first evidence of rescue of Imprime-driven PD responses in a cancer patient by supplementation of ABA, a crucial pre-requisite for the therapeutic activity of Imprime.

P75 iSEND is an algorithmic model specific for advanced non-small cell lung cancer patients treated with PD-1/PD-L1 inhibitors

Wungki Park1, Vaia Florou1, Sandra Algaze1, Diana Saravia1, Deukwoo Kwon1,2, Gilberto Lopes1
1University of Miami Miller School of Medicine, Miami, FL, USA; 2Sylvester Comprehensive Cancer Center, Miami, FL, USA
Correspondence: Gilberto Lopes (wungkipark@gmail.com)

Background

We have shown that the iSEND model may be predictive of clinical outcomes for advanced NSCLC (aNSCLC) patients treated with nivolumab but not with chemotherapy or TKI.[1] However, little is known about its potential performance for other advanced solid cancer patients treated with PD-1/PD-L1 inhibitors (PD-1/PD-L1i).

Methods

We evaluated the clinical outcomes of 370 solid cancer patients who received PD-1/PD-L1i and compared the performance of iSEND in aNSCLC (n=203) and other advanced solid cancers (n=167). Patients were either treated with nivolumab (n=237), pembrolizumab (n=109), or atezolizumb (n=24). As described in our previous reports, the iSEND model (immunotherapy, Sex, ECOG [Performance status], NLR [Neutrophil-to-Lymphocyte Ratio] & DNLR [Delta NLR = NLR after treatment - pretreatment NLR]) was developed. We stratified each treatment group by iSEND and compared progression free survivals (PFS) and clinical benefit rates (CBR) at 12+/-2 weeks in the iSEND Good, Intermediate, and Poor groups.

Results

Median follow-up was 9.1 (95% CI: 8.0-10.2). In the aNSCLC, The 3-, 6-, 9-, and 12-months PFS rates were 83%, 70%, 62%, and 58% in the iSEND Good group, 71%, 40%, 29% and 20% in Intermediate group, and 60%, 18%, 18%, and 18% in Poor group, respectively (p<0.0001). Median PFS was unreached in iSEND Good, 4.2 months; 95% CI [2.8-5.6] in iSEND Intermediate, and 2.9 months; 95% CI [2.7-3.1] in iSEND Poor groups. (Fig. 1). In contrast in other advanced solid cancers, median PFSs were 5.4, 8.2, and 3.7 months in iSEND Good, Intermediate, and Poor groups, respectively (p=0.415). The area under the curves (AUC) of the iSEND score for CBR at 12+/-2 weeks for aNSCLC patients treated with PD-1/PD-L1i was 0.720, (95% CI: 0.648-0.793, p<0.0001). The AUCs of iSEND for CBR in other advanced solid cancers was not significant (Fig. 1).

Conclusions

In our single-institution retrospective cohort, the iSEND model showed a predictive potential specifically for advanced NSCLC patients treated with PD-1/PD-L1i but not for other advanced solid cancers treated with PD-1/PD-L1i.

Reference

1. Saravia D, Laderian B, Park W, Desai A, Vargas F, Elias R, Warsch S, Mudad R, Ikpeazu C, Ishkanian A, Balfe L, Jahanzeb M. Journal of Thoracic Oncology. 2016; 12(1):S1327.
Fig. 1 (abstract P75).

Kaplan-Meier curves for PFS in aNSCLC vs. other advanced solid cancers treated with PD-1/PD-L1i

P76 Usefulness of automated multidimensional flow cytometry analyzes for monitoring the status of T and APC compartments in breast cancer patients under neoadjuvant chemotherapy

David Bernal-Estévez1, Carlos Parra-López2
1Fundación Salud de los Andes, Bogota, Colombia; 2Universidad Nacional de Colombia, Bogota, Colombia
Correspondence: Carlos Parra-López (caparral@unal.edu.co)

Background

Immune monitoring systems based on flow cytometry (FC) are usually analyzed by manual strategies of gating based on two-dimensional plots. Recently the forthcoming of new technologies such as CyTOF and advances of multi-parametric FC analyzes demand new approaches to analyze massive immune related data. New algorithms like SPADE, FLOCK, tSNE, and more recently CITRUS, have provided new tools for automated multidimensional FC analysis. These tools have been used mainly for the cellular hierarchy analysis of bone marrow derived cells and in leukemia patients but to a lesser extend for immune-monitoring of patients under anti-tumor therapy. In this work we explore the usefulness of these automated systems for the immune-monitoring of breast cancer patients under neo-adjuvant chemotherapy.

Methods

We used data from an in vitro system to monitor by FC various immunological readouts in T and APCs in PBMCs from breast cancer patients (BCPs) before, and after three cycles of neo-adjuvant anti-tumor therapy with Doxorubicin and Cyclophosphamide (A/C). We applied different automated algorithms including SPADE, FLOCK and CITRUS to process FC data on the status of T and APC compartments in these patients.

Results

Using CITRUS, we detected ex vivo alteration in the frequency of Dendritic Cells (DCs) with a plasmacytoid phenotype in BCPs patients before treatment that was recovered to the levels observed in healthy donors after three cycles of A/C chemotherapy. Regarding T cells, we confirmed that before therapy T cells in BCPs exhibited marked unresponsiveness to a polyclonal stimulus evidenced by a deficient TCR internalization and low expression of CD25, CD69, and CD154 that was also recovered in BC patients after chemotherapy.

Conclusions

Our results let us to argue that automated systems are useful tools for the immune-monitoring by FC of BCPs under chemotherapy and for the analysis of cell populations with a complex immune-phenotype not analyzable through manual gating.

P77 The novel IL-2 cytokine immune agonist NKTR-214 harnesses the adaptive and innate immune system for the treatment of solid cancers

Salah Eddine Bentebibel1, Chantale Bernatchez1, Cara Haymaker1, Michael Hurwitz2, Patrick Hwu1, Mario Sznol2, Nizar Tannir1, Sandra Aung3, Michael Imperiale3, Mary Tagliaferri3, Christie Fanton3, Ernesto Iacucci3, Jonathan Zalevsky3, Ute Hoch3, Adi Diab1
1The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2Yale School of Medicine, New Haven, CT, USA; 3Nektar Therapeutics, San Francisco, CA, USA
Correspondence: Salah Eddine Bentebibel (mvphillips@phillipsgilmore.com)

Background

The novel, CD-122 biased PEGylated IL-2 cytokine immune agonist NKTR-214 harnesses the potent immune stimulatory benefits of the IL-2 pathway to maximize anti-tumor responses, and minimize unwanted biological effects. NKTR-214 has been safely administered in the outpatient setting, with low-grade hypotension as the most clinically significant adverse event. NKTR-214 robustly increases expression of activation and proliferation markers on immune cells in blood and tumor. Evidence of clinical activity has been observed in IO naïve and relapsed/refractory patients. Here we describe the phenotype of proliferating T cells in the blood and present an in-depth characterization of tumor-infiltrating immune cells using gene expression analysis and TCR profiling.

Methods

Pre- and on-treatment blood and tumor biopsies from patients receiving 0.003 to 0.009 mg/kg NKTR-214 monotherapy were analyzed. Blood samples (n=22 pts) were analyzed with 16-color multi-parameter flow-cytometry. Tumor biopsies (n=10 pts) were analyzed for gene expression using the Nanostring PanCancer immune panel. Differentially expressed genes in pre- and on-treatment samples were identified and mapped to public databases for pathway enrichment analysis. Tumor T cell receptor (TCR) sequencing was conducted at Adaptive Biotechnologies, and analyzed using the R package LymphoSeq.

Results

In blood samples, we observed increased expression of activation markers (ICOS and OX40) as well as co-inhibitory receptors (PD-1, CTLA-4, TIM-3 and Lag-3) in both CD4+ and CD8+ T cells proliferating in response to NKTR-214. Differentially expressed genes (n=54) mapped to several innate and adaptive immune system pathways, including NK cell mediated cytotoxicity, IL-2, IL-12, and TCR signaling pathways. Pairwise comparison of matched baseline and on-treatment tumor biopsies confirmed increased expression of genes associated with T cell signaling (ie. CD3G/D/E, CD247, CD8A and CD8B) and function (ie. PRF1, GNLY, GZM, and IFNG), as well as upregulation of several immune checkpoint and costimulatory genes (ie. PD1, PDL1, LAG3, IDO, and CTLA4). Gene expression changes were observed at all dose levels and across tumor types studied. Tumor TCR sequencing indicates changes in clonality with a trend towards increased frequency of specific clones on-treatment.

Conclusions

A single dose of NKTR-214 modulates both the innate and adaptive immune system. Changes in clonality and increased frequency of specific clones suggest efficient remodeling of the T cell repertoire. Upregulation of checkpoint and costimulatory genes provides insights for potential combination therapies. Based on the favorable safety profile and strong correlative biomarker data, a phase 1/2 trial combining NKTR-214 and nivolumab is currently enrolling.

P78 Immune, molecular and T cell repertoire landscape of 235 resected non-small cell lung cancers, paired normal lung and peripheral blood mononuclear cells

Alexandre Reuben1, Rachel Gittelman2, Jiexin Zhang1, Runzhe Chen1, Kelly Quek1, Luis Vence1, Irina Fernandez-Cubelo1, Carmen Behrens1, Jianjun Gao1, Erik Yusko3, Ryan Emerson3, Sharon Benzeno3, Marissa Vignali3, Christopher Tipton4, Ali Jalali1, Won-Chul Lee1, Jun Li1, Xifeng Wu1, Yuanqing Ye1, Agda Eterovic1, Latasha Little1, Curtis Gumbs1, Vancheswaran Goplakarishnan1, Chantale Bernatchez1, Cara Haymaker1, Marie-Andrée Forget1, Lorenzo Federico1, Tina Cascone1, Harlan Robins5, Emily Roarty1, Jaime Rodriguez1, Edwin Parra1, Jennifer Wargo1, James Allison1, Padmanee Sharma1, Jianhua Zhang1, Jack Lee1, Boris Sepesi1, Stephen Swisher1, Don Gibbons1, John Heymach1, Andy Futreal1, Ignacio Wistuba6, Jianjun Zhang6
1MD Anderson Cancer Center, Houston, TX, USA; 2Adaptive Biotechnologies, Seattle, WA, USA; 3Adaptive Technologies, Seattle, WA, USA; 4Emory University School of Medicine, Atlanta, GA, USA; 5MD Anderson Cancer Center, Seattle, WA, USA; 6MD Anderson, Houston, TX, USA

Background

Non-small cell lung cancer (NSCLC) is characterized by a high mutational load. Accordingly, it is also among the tumor types that respond to immune checkpoint blockade, presumably mediated by the anti-tumor T cell response. However, the lung is continuously exposed to the outside environment, which may result in a continuous state of inflammation against outside pathogens rather than tumor cells. Therefore, further investigation into the T cell repertoire and T cell phenotypes across normal lung and tumor is warranted.

Methods

We performed T cell receptor (TCR) sequencing and whole exome sequencing (WES) on PBMC, normal lung, and tumor from 235 NSCLC patients. Of these patients, 96 were also subjected to WES and TCR sequencing. We further performed Cytometry by Time-of-Flight (CyTOF) on 10 NSCLC tumors and paired normal lung tissues to phenotype immune and T cell subsets.

Results

Comparison of the T cell repertoire showed 9% (from 4% to 15%) of T cell clones were shared between normal lung and paired tumor. Furthermore, among the top 100 clones identified in the tumor, on average 57 (from 0 to 95) were shared with paired normal lung tissue. Interestingly, T cell clonality was higher in the normal lung in 89% of patients suggesting potential differences in the immune response and immunogenicity. A substantial number of somatic mutations were also identified not only in NSCLC tumors (average 566; from 147 to 2819), but also in all morphologically normal lung tissues (average 156; from 50 to 2481). CyTOF demonstrated striking differences in the immune infiltrate between normal lung and tumor, namely a unique GITR+ T cell subset (0.96%) which was entirely restricted to the normal lung. Conversely, increases in regulatory T cell frequency (CD4+FoxP3+) were observed in the tumor (10.4% vs 1.7% in normal lung), further highlighting the differences in T cell phenotype and response across normal lung and tumor.

Conclusions

These results suggest that a substantial proportion of infiltrating T cells in NSCLC tumors may be residential T cells associated with response to environmental factors. However, normal lung and NSCLC tumors carry T cells of distinct phenotypes including increases in immunosuppressive T cells within the tumor which may further highlight the differences in the anti-tumor immune response.

P79 T cell inflamed gene expression profile (GEP) analysis of pembrolizumab- and ipilimumab-treated patients with advanced melanoma in the multicenter, randomized, open-label phase 3 KEYNOTE-006 study

Antoni Ribas1, Caroline Robert2, Jacob Schachter3, Georgina Long4, Ana Arance5, Jean-Jacques Grob6, Matteo Carlino7, James Larkin8, Andrea Webber9, Robin Mogg9, Mark Ayers9, Jared Lunceford9, Fang Liu9, Michael Nebozhyn9, Andrew Albright9, Wendy Blumenschein9, Clemens Krepler9, David Kaufman9, Nageatte Ibrahim9, Adil Daud10
1University of California, Los Angeles, Los Angeles, CA, USA; 2Gustave Roussy and Paris-Sud University, Villejuif, France; 3Ella Lemelbaum Institute for Melanoma, Sheba Medical Center at Tel Hashomer, Ramat Gan, Israel; 4Melanoma Institute Australia, University of Sydney, Mater Hospital and Royal North Shore Hospital, Sydney, Australia; 5Hospital Clinic de Barcelona, Barcelona, Spain; 6Aix Marseille University, Hôpital de la Timone, Marseille, France; 7Westmead and Blacktown Hospitals, Melanoma Institute Australia, and University of Sydney, Blacktown, Australia; 8Royal Marsden Hospital, London, UK; 9Merck & Co., Inc., Kenilworth, NJ, USA; 10University of California, San Francisco, San Francisco, CA, USA
Correspondence: Antoni Ribas (aribas@mednet.ucla.edu)

Background

The independently validated T cell–inflamed GEP has been consistently associated with response to the anti–programmed death 1 (PD-1) antibody pembrolizumab. We analyzed gene expression in baseline tumor samples from the KEYNOTE-006 study of pembrolizumab versus the anti-cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4) antibody ipilimumab to determine the relative predictive effects of the T cell–inflamed GEP on clinical outcomes in advanced melanoma.

Methods

Patients were randomized to pembrolizumab 10 mg/kg every 2 or 3 weeks or 4 doses of ipilimumab 3 mg/kg every 3 weeks. Total RNA from baseline tumor samples was isolated from 4-μm formalin-fixed, paraffin-embedded tissue sections and analyzed on the NanoString nCounter gene expression platform. Overall survival (OS) was the prespecified primary end point at final analysis; progression-free survival (PFS) and best objective response (BOR) were secondary end points. For efficacy comparison of pembrolizumab versus ipilimumab across T cell–inflamed GEP levels, the scores were categorized into tertiles (low: GEP < -0.28; middle: -0.28 < GEP < 0.07; high: GEP > 0.07), as prespecified in the statistical analysis plan, and Cox proportional hazards modeling was used to compare the treatment effect of the pooled pembrolizumab arms versus ipilimumab within each subgroup. Data cutoff for clinical analyzes was December 3, 2015.

Results

Of 834 enrolled patients, gene expression data were available for 427 (309 pembrolizumab; 118 ipilimumab) with OS/PFS data and for 375 (269; 106) with BOR data. Baseline characteristics were similar between the GEP and overall patient populations. Association of the T cell–inflamed GEP with OS was confirmed within each treatment group (P<0.0001 pembrolizumab; P=0.0127 ipilimumab). Evaluation of the between treatment group effect on OS was significant in all tertile subgroups (hazard ratio [HR]=0.61, P=0.0282 [low]; HR=0.60, P=0.0224 [middle]; and HR=0.44, P=0.0042 [high]) but was most pronounced in the high group. Qualitatively similar results were observed for PFS. Although the T cell–inflamed GEP was not associated with BOR among ipilimumab-treated patients (area under the receiver operating characteristic [AUROC]=0.511; P=0.4723), it was associated with BOR among pembrolizumab-treated patients (AUROC=0.696; P<0.0001). Association between the T cell–inflamed GEP and tumor PD-L1 expression by immunohistochemistry and the relationship with clinical outcomes will be presented.

Conclusions

T cell–inflamed phenotype as measured by the GEP shows a differential association with OS for pembrolizumab versus ipilimumab, which may be explained by the ability of the GEP score to predict BOR in the tumor microenvironment with anti−PD-1, but not anti−CTLA-4, therapy.

Trial Registration

ClinicalTrials.gov Identifier NCT01866319

P80 Testing systems immune monitoring by mass cytometry for clinical correlative research

Caroline Roe, Allison Greenplate, Jonathan Irish
Vanderbilt University, Nashville, TN, USA
Correspondence: Jonathan Irish (caroline.e.roe@vanderbilt.edu)

Background

Systems immune monitoring during cancer treatment can track therapy response and reveal biomarkers [1]. In metastatic melanoma, this approach has implicated proliferating T cell subsets as a cellular effector mechanism for checkpoint inhibitors [2, 3]. Aims here were: 1) develop a robust cancer immune monitoring panel for multi-center clinical correlative research conducted by a core, and 2) generate pilot data to train and test computational tools employing machine learning algorithms.

Methods

Peripheral blood mononuclear cells (PBMC) were collected with informed consent from melanoma patients with local Institutional Review Board (IRB) approval and in accordance with the Declaration of Helsinki (N = 10 samples from 5 patients). For each patient, PBMC were collected before and approximately 3 weeks post-treatment with anti-PD1 immunotherapy. A commercially available antibody kit, Fluidigm's Complete Human T cell Immuno-Oncology panel kit, was chosen to focus on T cell subsets, memory, activation, and immuno-oncology proteins. Also tested was the addition of proliferation marker Ki67 and cell identity markers CD19 and CD14. Data were collected using a Helios mass cytometer, normalized using normalization beads, and analyzed with viSNE, Cytobank, and MEM [4] following established methods [5].

Results

Individual patient immunophenotypes were consistent pre- and post- anti-PD-1 therapy. Addition of markers to the commercial antibody sets was straightforward. CD19 and CD14 proved unnecessary for automatic identification of cell subsets by viSNE and MEM. Ki67 enabled detection of T cell proliferation. Because Ki67 is less abundant in cells, mass cytometry panel design should be used to detect Ki67 in a channel with high sensitivity.

Conclusions

Use of mass cytometry and commercially available metal conjugated antibodies provided a robust method for systems immune monitoring in cancer therapy compatible with correlative research in larger clinical studies.

References

1. Greenplate, et al. Systems immune monitoring in cancer therapy. Eur J Cancer. 2016; 61:77-84.

2. Huang, et al. T cell invigoration to tumor burden ratio associated with anti-PD-1 response. Nature. 2017.

3. Spitzer, et al. Systemic immunity is required for effective cancer immunotherapy. Cell. 2017;168(3): 487-502 e15.

4. Diggins, et al. Characterizing cell subsets using marker enrichment modeling. Nat Methods. 2017;14(3):275-278.

5. Diggins, Ferrell, Jr., Irish. Methods for discovery and characterization of cell subsets in high dimensional mass cytometry data. Methods. 2015; 82:55-63.

P81 The immunomodulatory effects of cancer therapy on IFN-gamma responses in the periphery

Laura Rosa Brunet1, James Crooks2, Andrew MacDonald2, Samuel LaBrie3, Dominic Eisinger3
1Immodulon Therapeutics Ltd, Uxbridge, UK; 2Manchester University, Manchester, UK; 3Myriad RBM, Austin, TX, USA
Correspondence: Laura Rosa Brunet (lrb@immodulon.com)

Background

The immune system is recognized as an important ally in the fight against cancer. Indeed patients’ survival outcome can be predicted by their immune response bias at the tumor site. With the ever expanding landscape of cancer therapy options, including immunotherapeutic agents, it has become critical to determine how patients’ immune responses are affected by treatment. In light of the challenges of repeatedly accessing tumor tissue, investigating immune responses in the periphery assumes increased importance.

Methods

IMAGE-1 is a randomised open-label, phase II, first-line, proof of concept study (NCT01303172), investigating the combination of IMM-101, an immunomodulatory treatment comprising of heat-killed whole cell Mycobacterium obuense (NCTC13365), with Gemcitabine for the treatment of advanced pancreatic cancer. Treatment with IMM-101+Gemcitabine increased median survival to 7.0 months in metastatic pancreatic cancer patients compared to 4.4 months following treatment with Gemcitabine alone (1). We measured levels of IFN-gamma, a critical cytokine for anti-tumor responses, in patients’ serum samples. To overcome technical difficulties associated with measuring serum IFN-gamma, we used an immunoassay (LLOQ=9.5fg/ml) developed by Myriad RBM for the Quanterix Simoa™ platform. This approach provides ultra-sensitive measurement of biomarkers, achieving orders-of-magnitude greater sensitivity than conventional immunoassay platforms.

Results

We report that peripheral IFN-gamma levels increased following initiation of treatment (Gemcitabine±IMM-101). In the initial 50 serum samples analyzed, there was a statistically significant increase in IFN-gamma levels from time 0 (screening/randomization; 0.12pg/ml, 95% CI 0.084, 0.18pg/ml) to week 13 (0.41pg/ml, 95% CI 0.26, 0.66pg/ml), when patients had been scheduled to receive 3 cycles of Gemcitabine (1000 mg/m2 for 3 consecutive weeks out of 4) ± IMM‑101 (1mg, intradermally on 6 occasions). Our previous data show that IMM-101 activates primary murine and human dendritic cells, and in murine models instigates IFN-gamma production by a variety of innate and adaptive immune cells. We investigated whether patients that received IMM-101 in addition to Gemcitabine had increased serum IFN-gamma levels. Preliminary data suggest that patients treated with Gemcitabine+IMM-101 (n=15) had higher IFN-gamma secretion at week 13 (0.57pg/ml, 95% CI 0.33, 0.98pg/ml) than patients (n=9) treated with Gemcitabine alone (0.24pg/ml, 95% CI 0.099, 0.58pg/ml).

Conclusions

We are in the process of confirm these findings in further serum samples from IMAGE-1 patients and addressing the specific cellular sources of IFN-gamma in murine experimental models.

Reference

1. Dalgleish, et al. Randomised, open-label, phase II study of gemcitabine with and without IMM-101 for advanced pancreatic cancer. Br J Cancer. 2016;115:989-96

P82 Potential biomarkers in mRCC patients treated with nivolumab

Maria Sayapina1, Dmitry Nosov2, Nikita Savelov3, Nina Lyubimova1, Yuriy Timofeev1
1N.N. Blokhin Russian Oncology Research Center, Moscow, Russian Federation; 2The Central Clinical Hospital for Presidential Affairs, Moscow, Russian Federation; 3The Moscow City Oncology Hospital No. 62, Istra, Russian Federation
Correspondence: Maria Sayapina (maria.sayapina@mail.ru)

Background

Checkpoint inhibitors produce meaningful survival benefit in mRCC patients (pts). Predictive biomarkers of response to anti-PD1 therapy are still under evaluation in clinical trials. Prospectively we explored biomarkers that might have potential predictive and prognostic value in mRCC pts on nivolumab therapy.

Methods

Patients with mRCC who had 1 or more prior lines of therapy received nivolumab at 3 mg/kg q2w (n=23) from 2015 to 2017. Blood samples were collected at baseline and 2 mo after treatment start. Serum levels of TGF-β1, IL-17A, and soluble PD-1 (sPD-1) were measured by ELISA. Pre-treatment specimens from primary tumors (n=17) were evaluated for PD-L1 and FOXP3 expression on tumor-infiltrating immune cells (IC) using the IHC assay (SP142; EP340).

Results

With a median follow-up of 10 mo, PFS was 4 mo (95% CI: 1,37-10,04) and 1-year PFS was 17%, median overall survival was not reached. The optimum diagnostic cut-off value (mean + 2SD) for TGF- β1 was 20 ng/mL (sensitivity 43,75%, specificity 100%), for IL-17A was 0,5pg/mL (sensitivity 40%, specificity 88,23%), for sPD-1 was 0,45 ng/mL (sensitivity 40%, specificity 88,8%). Higher TGF-β1 level at baseline was associated with shorter PFS (6 vs 1,5 mo; p=0,03). Higher baseline IL-17A (p=0,0174) and sPD-1(p=0,039) levels were significantly associated with higher ORR (21,7%). Using the ≥0,5% and ≥1% cut-off for PDL-1 expression, 11/17 (64%) and 8/17 (47%) pts were considered PD-L1 positive respectively. When a cut-off of ≥0,5% was used clinical effect (tumor shrinkage >10%) was higher in the PD-L1-positive group than in the PD-L1-negative group (54,5% vs 33,3%). Using the ≥10% cut-off for FOXP3, 11/17 pts were considered positive (64%). Clinical effect was 45% in the FOXP3-positive group and 50% in the FOXP3-negative group. Expression of PD-L1 and FOXP3 on IC didn’t not have a significant effect on PFS (p>0,05).

Conclusions

Our results demonstrated improved survival benefit in patients with lower levels TGF- β1 (<20ng/mL). Elevated baseline levels of IL-17A (>0,5pg/mL) and sPD-1 (>0,45ng/mL) were significantly associated with ORR. PD-L1 and FOXP3 IC expression have no predictive or prognostic value for pre-treated RCC pts receiving nivolumab therapy.

P83 Characterizing cancers by prevalence on the PD-L1/CD8 axis

Guenter Schmidt1, Keith Steele2, Armin Meier1, Thomas Herz1, Tze Heng Tan1, Brandon Higgs2, Sonja Althammer1, Ralf Huss1
1Definiens AG, Munich, Germany; 2MedImmune LLC, Gaithersburg, MD, USA
Correspondence: Guenter Schmidt (gschmidt@definiens.com)

Background

Checkpoint inhibitor-based immunotherapies are successful across various cancers, particularly, type I cancers, characterized by tumoral PD-L1 expression and by a repertoire of tumor-infiltrating lymphocytes, respond well to therapies targeting PD1/PD-L1 [1]. While it has been shown that combinatorial CD8(+) and PD-L1(+) cell densities provide predictive value for durvalumab in NSCLC [2], we present in this study profiling results for multiple indications. The findings are set into the context of tumor mutation burden (TMB) which has been proposed as additional predictor for response to immunotherapies [3].

Methods

Image analysis (Definiens, Munich, Germany) detected PD-L1(+) tumor and CD8(+) immune cells in pathologist annotated tumor center in serial, IHC-stained tissue sections (Ventana SP263 and SP239 resp.) from multiple cancer indications. Automatic image alignment enabled the computation of tile-based (size=64μm) cell coverage statistics: percentage of tiles with at least one PD-L1(+) (CD8(+) resp.) cell, and co-occurrence C-score [4] of tiles containing at least one PD-L1(+) and one CD8(+) cell. The indications were ranked per TMB as published in [3], where the authors used comprehensive genomic profiling using the FoundationOne assay (Cambridge MA, USA).

Results

In Fig. 1, PD-L1(+) and CD8(+) low and high patients were defined by median cut-points (1% PD-L1(+) tiles, 36% CD8(+) tiles). The population prevalences in the PD-L1(+)HighCD8(+)High group correlate with mutational burden (Spearman's r=0.96, p<0.01), with the notable exception of BLAD (similar prevalence reported in [5]). PANC prevalence dominates the PD-L1(+)LowCD8(+)High group (>2σ). The CD8(+)Low groups are heterogenous in prevalences.

Figure 2 shows the mean PD-L1(+) and CD8(+) C-score in all cancers of this study. Mutation-high NSCLC is characterized by high C-scores, whereas mutation-low PANC is characterized by low C-scores, suggesting dysfunctional CD8(+) cells or tumor specific mutations suppressing response to IFNγ.

Conclusions

This study presents an automated profiling method for cancers, which may provide clinical decision support for which type of tumor a certain type of immunotherapy works best. While our findings are consistent with population statistics on tumor mutational burden, they highlight the relevance of spatial analysis in tissue for a deepened understanding of the mechanisms of diseases.

References

1. Teng, et al. doi:10.1158/0008-5472.CAN-15-0255

2. Althammer, et al. doi:10.1186/s40425-016-0191-4

3. Chalmers, Connelly, et al. doi:10.1186/s13073-017-0424-2

4. Novak, et al. doi:10.1111/j.1365-2486.2011.02482.x

5. Faraj, et al. doi:10.1016/j.urology.2014.10.020
Fig. 1 (abstract P83).

Cancer prevalences in the PD-L1(+) and CD8(+) low and high groups. Prevalences of seven cancer indications (see inset, with number of patients), grouped by classification into PD-L1(+) and CD8(+) low and high tile percentage tumors. The indications were sorted with respect to decreasing TMB as defined in [3]

Fig. 2 (abstract P83).

Mean of C-scores. Mean of C-scores (x-axis) for seven cancer indications (y-axis), whereas the indications are sorted by tumor mutaional burden (TMB). The ranking of indications in C-scores correlates well with TMB (Spearman's r=0.96, p<0.01)

P84 Relevance of the microenvironment and MHC class I mediated immune escape mechanisms of tumors for anti-tumoral immune responses

Barbara Seliger (barbara.seliger@uk-halle.de)
Institute of Medical Immunology, Halle, Germany

Background

Abnormalities of classical and non-classical MHC class I antigens were frequently found in tumor cells of distinct origin resulting in evasion of tumor cells from T cell-mediated immune surveillance. The underlying molecular mechanisms of these alterations are complex and include transcriptional, epigenetic and posttranscriptional control of MHC class I antigens and components of the antigen processing machinery (APM). Furthermore, the escape of tumor cells from immune surveillance is also linked to changes in the cellular composition of the tumor microenvironment and peripheral blood mononuclear cells.

Methods

Tumor cells were analyzed for the expression of HLA class I APM components by qPCR, Western blot analysis and/or flow cytometry. Immunohistochemistry and multispectral imaging was employed to determine the frequency, localization and topology of suppressor and effector immune cells.

Results

While impaired expression of classical MHC class I and HLA-G is often due to transcriptional downregulation, alterations in the histone acetylation and methylation appear not to be so frequent. Defects in the IFN signal transduction, oxidative status of MHC class I antigens as well as altered expression of immune modulatory microRNAs and RNA-binding proteins affects the expression of MHC class I APM and/or HLA-G. In addition, the extracellular matrix protein biglycan has been shown to increase MHC class I surface expression, which was associated with enhanced immune cell responses and reduced expression of members of the TGF-β pathway. However, not only tumor-, but also therapy-induced effects are responsible for shaping MHC class I antigen expression. Furthermore, high levels of immune suppressive cells in the tumor were associated with a worse prognosis, but in particular the distance of immune cells to each other, and to tumor cells has prognostic value. This could be further extended by combining the immune cell infiltration with MHC class I APM component expression. In addition, downregulation of MHC class I APM component expression has been shown to occur could upon adoptive cell therapy.

Conclusions

Thus, these data suggest that the molecular make up of tumor cells might shape the immune response, which might serve as biomarkers for prediction of immune-based therapies. In conclusion, a better understanding of the immune evasion mechanisms is still required for improving T cell-based, individualized immunotherapies, and appears to be of importance in the development of resistances to these therapies, but opens also new venues for therapeutic intervention.

P85 HAAH and MMP9 are complementary cancer biomarkers and potentially enhanced predictors of metastasis

Mark Semenuk, Eleanor Ghanbari, Anokhi Cifuentes, Anjali Kalra, Michael Lebowitz, Hossein Ghanbari
Panacea Pharmaceuticals Inc, Gaithersburg, MD, USA
Correspondence: Mark Semenuk (msemenuk@panaceapharma.com)

Background

When used together, complementary biomarkers associated with mediators of cancer cell mobility and invasiveness should be useful early disease and patient outcome predictors. The present investigation is an initial quantitative assessment of HAAH (aspartyl (asparaginyl) beta hydroxylase) and MMP9 (matrix metalloproteinase 9) in serum/serum exosomes from cancer patients to evaluate their concerted role in metastasis.

Cancer-specific cell surface HAAH functions by enzymatically modifying a number of motif-restricted protein targets including Notch. It thereby indirectly triggers events leading to metastasis. MMP9 is a well-known predictor of metastasis due to its inherent effect on the process of proteolytically-assisted tumor cell escape, albeit not completely useful as a cancer biomarker on its own. We propose here that up-regulated HAAH is a prerequisite for metastasis and that in turn MMP9 is an enabler of this process.

Methods

We detect serum and exosomal HAAH by a simultaneous-homologous ELISA format using an in house manufactured reagent kit comprising pre-coated microplates and pre-formulated reagents. Serum and exosomal MMP9 was detected with a commercial reagent kit ELISA (Abcam). Exosomes were prepared using a 50% polyethylene glycol 6000/0.5 M NaCl solution added to serum, centrifugation, and reconstitution. CEA positive cancer and healthy serum samples were commercially obtained (Complex Antibodies) or through off site collaborators.

Results

Similar recovery of HAAH (58.3%) and MMP9 (54.0%) from exosomes compared to the source CEA positive cancer serum samples was observed, suggesting exosome co-localization. Further, both biomarkers were quantified in a larger investigation of samples obtained from high risk volunteers that potentially could have cancer. In this ongoing field study of 48 accumulated serum samples, 9 were positive for HAAH. Of these 9 samples 6 were positive for MMP9. Therefore at least three subjects had elevated HAAH that was not paralleled by increased MMP9 levels.

Conclusions

While HAAH and MMP9 are both expected to be closely associated with metastatic activity of cancer cells, both co-localize in cancer derived exosomes, and both appear to be regulated by the same transcription factor(s), their expression in serum samples are mostly coincident but sometimes may differ. This may explain differences in metastatic potential. These studies are ongoing and focused upon determining whether using both biomarkers could lead to a more accurate prediction of metastatic potential.

P86 Altered immune cell repertoire and activity after checkpoint blockade immunotherapy with nivolumab in renal cell carcinoma

Michael Staehler1, Chiara Massa2, Annabel Spek1, Daniel Bethmann2, Jurgen Bukur2, Anja Mueller2, Claudia Wickenhauser2, Bernard A. Fox3, Barbara Seliger2
1Ludwig Maximilian University, Munich, Germany; 2Martin Luther University, Halle, Germany; 3Robert W. Franz Cancer Research Center, Portland, OR, USA
Correspondence: Michael Staehler (mstaehler@me.com)

Background

Programmed cell death protein-1 (PD-1) and its ligand (PD-L1) provide tumor cells with mechanisms to escape from immune surveillance. Antibodies directed against these checkpoint molecules have been developed and successfully employed for the treatment of solid tumors including renal cell carcinoma (RCC). Despite promising clinical benefits, only 20 - 30% of tumor patients respond to these therapies. In order to search for predictive or prognostic biomarkers the immune cell repertoire in the peripheral blood of RCC patients was determined prior and during anti-PD-1 therapy with the PD-L1 antibody nivolumab.

Methods

We included 43 patients who had an indication for second line therapy for clear cell RCC with nivolumab. Blood specimen and tumor samples were collected at baseline and three months post-Nivolumab onset and at a clinical event such as progression or remission.Cell differentiations were performed as well as using multicolor flow cytometry and nanostring technology. In addition, tumor biopsies obtained before treatment were analyzed by multispectral imaging (MSI).

Results

Peripheral blood and tumor lesions of seven RCC patients have been analyzed in detail: 1/7 RCC patients had a stable disease (14 %), 3/7 (43 %) patients a partial or complete remission. One RCC patient had serious adverse effects upon treatment resulting in discontinuation of therapy, while two other RCC patients had a progressive disease and therefore were shifted to chemotherapy. Immunohistochemical evaluation by MSI indicated a higher frequency of Foxp3+ cells in the three progressive RCC patients in the stroma of the tumor, and particularly in the central core, whereas lower levels of CD163+ macrophage were present in the core of the tumor. Increased levels of CD4+ Foxp3+ regulatory T cells (Treg) were also present in the peripheral blood of progressive RCC patients. During the therapy, variation in the Treg frequency of peripheral blood did not correlate with clinical outcome of patients, whereas a reduction in HLA-DR+ antigen presenting cells did. Nanostring analyzes highlighted enhanced levels of interleukin (IL)-2 transcript in the peripheral blood of the two progressive RCC patients, in contrast to the diminished IL-2 mRNA levels in the three responders.

Conclusions

These data suggest that the evaluation of a larger number of RCC patients combining these three different techniques might allow identifying a set of reliable markers that predict responsiveness to nivolumab therapy in these patients.

P87 Results of epigenetic-based quantitative PCR assisted immune cell counting analysis in bavituximab SUNRISE trial subgroup

Nikoletta L. Kallinteris1, Thomas O. Kleen2, Min Tang1, Shen Yin1, Tobi Guennel3, Jennifer Lai1, Victor Nowakowski2, Steven Olek2, Steve King1, Joseph S. Shan1
1Peregrine Pharmaceuticals, Tustin, CA, USA; 2Epiontis GmbH, Berlin, Germany; 3Precision Medicine, Frederick, MD, USA
Correspondence: Nikoletta L. Kallinteris (lisa_s@msn.com)

Background

SUNRISE (NCT01999673), a global, double-bind, randomized Phase III trial of docetaxel plus bavituximab (D+B) or docetaxel plus placebo (D+P) in previously treated non-squamous non-small cell lung cancer (NSCLC), demonstrated similar overall survival (OS) in the intent-to-treat population (n = 597). In the subgroup of 93 patients who received subsequent immune checkpoint inhibitors (ICI), median overall survival (mOS) was not reached (95% CI, 15.2-NA) in the D+B group (n=46) and was 12.6 months (95% CI, 10.4-17.8) for patients in the D+P group (n=47) (HR for death, 0.46; P = 0.006). Epigenetic-based quantitative real-time PCR assisted cell counting (qPACC) of immune cells in blood was used to potentially identify predictive biomarkers.

Methods

DNA was isolated from peripheral blood mononuclear cells (PBMC) from randomized patients and treated with bisulfite leading to conversion of de-methylated cytosine (epigenetically active) residues into uracil, but left methylated ones unaffected. The de-methylation status of DNA regions, previously identified of being specific to respective immune cell phenotypes sorted by FACS: CD3+, CD4+, CD8+, TFH, TH17, PD1, Foxp3, naïve CD8, MDSC, CD14+, NK56+, B-cells, GNLY, and CCR6+, was quantitated by qPACC. The number of de-methylated gene copies per biomarker was quantitated and translated into % total of cells in the sample. Each biomarker was classified as high or low based on the median value across all patients and correlated to OS. Hazard ratios (HR) and confidence intervals (CI) were estimated using a Cox proportional-hazards model.

Results

Pre-treatment (pre-tx) samples were evaluable by qPACC for 62 (32 D+B, 30 D+P) out of the 93 patients who received ICI as next line therapy after the SUNRISE assigned treatment. High pre-tx (≥ median) levels correlated with statistically significant OS benefit favoring D+B for the following biomarkers: CD3+ (HR=0.37, p=0.023), CD4+ (HR=0.32, p=0.012), CD8+ (HR=0.42, p=0.036), PD-1 (HR=0.33, p=0.017), GNLY (HR=0.25, p=0.011), FoxP3 (HR=0.34, p=0.032), naïve CD8+ (HR=0.32, p=0.034), B-cells (HR=0.24, p=0.007), MDSC (HR=0.33, p=0.037) NK56+ (HR=0.29, p=0.026). Low (<) pre-tx levels of THF correlated with OS favoring D+B (HR=0.34, p=0.033). No OS difference was observed for pre-tx high or low levels for TH17, CCR6, or CD14+.

Conclusions

High pre-tx levels (≥median) of circulating immune cells including T cells, B-Cells, MDSCs, NK cells correlated with significant improvement of OS in patients who received D+B then ICI compared to D+P then ICI in the SUNRISE trial. These results support further investigation of these markers in future bavituximab clinical trials.

P88 CD96, a new candidate for checkpoint blockade in human hepatocellular carcinoma

Zhigang Tian, Cheng Sun
Institute of Immunology, The Key Laboratory of Innate Immunity and Chronic Disease (Chinese Academy of Medical Science), School of Life Sciences and Medical Center, University of Science & Technology of China, Hefei, China
Correspondence: Zhigang Tian (charless@ustc.edu.cn)

Background

It was reported that murine CD96 functions as a checkpoint in natural killer (NK) cell functional exhaustion. To date, blocking CD96 has not been demonstrated to be of substantial benefit in patients with cancer. Here, we demonstrated that human CD96 works as a checkpoint in hepatocellular carcinoma (HCC) patients.

Methods

In this study, we analyzed CD96 expression and the related dysfunction of NK cells located in intra- or peritumor regions of liver tissue samples from HCC patients, in addition to analyzing disease outcomes.

Results

Through the use of paired peritumoral tissues (PT) and IT from 236 HCC patients, we found that increased expression of CD96 on NK cells in intratumoral but not peritumoral regions, along with increased expression of its ligand CD155 and a poor prognosis. Then we investigated the expressions of CD226, TIGIT and CD96 (Immunoglobulin-like family members and share the common ligand CD155) in 54 fresh tumor tissues from HCC patients and 20 normal livers from health controls. We observed increased CD96 and decreased TIGIT expressions in NK cells from IT of HCC patients, and an increased number of CD96+ NK cells in HCC patients is associated with poor clinical characteristics and prognosis.

Human CD96+ NK cells from health controls exhibited functional exhaustion, showing decreased IFN-γ and TNF-a productions, impaired cytolysis in response to in vitro stimulation. CD96 blockade can enhance NK cytotoxicity on CD155 expressing target cells, and play a protective role against tumor through boosting NK cell immune response. Global transcriptomic analysis of sorted CD96+ and CD96- hepatic NK cells further demonstrated exhausted features of CD96+ NK cells comparing to CD96- NK cells.

Further, we demonstrated that high levels of TGF-β in HCC patients induced this increased expression of CD96 in NK cells. Blocking TGF-β specifically inhibited CD96 expression in NK cells. In addition, we compared other two receptors, CD226 and TIGIT, which share common ligand CD155 with CD96, and found CD96 play a more important role in NK exhaustion.

Conclusions

Overall, these data demonstrate that targeting CD96 is a potential approach for cancer immunotherapy.

P89 Plasma-derived exosomes carrying CTLA-4, PD-1 and PD-L1 in head and neck squamous cell carcinoma patients treated with immunotherapy is associated with disease outcome

Marie-Nicole Theodoraki1, William E. Gooding2, Robert L. Ferris1, Theresa L. Whiteside1
1University of Pittsburgh, Pittsburgh, PA, USA; 2Hillman Cancer Center, Pittsburgh, PA, USA
Correspondence: Marie-Nicole Theodoraki (marie.nicole.theodorakis@gmail.com)

Background

Exosomes, the smallest subset of extracellular vesicles, present in body fluids serve as a cellular communication network. In the tumor microenvironment, exosomes carry messages from the tumor to local and distant cells and are also potential biomarkers of tumor progression. Exosomes express immunomodulatory molecules and shape antitumor immunity.

Methods

The role of exosomes as circulating biomarkers of response to immunotherapy was evaluated by monitoring changes in the protein cargo of exosomes isolated from plasma of patients with head and neck squamous cell carcinoma (HNSCC) treated with a combination of Cetuximab, Ipilimumab and radiotherapy. We retrospectively evaluated changes in the exosome protein cargo at multiple time points during therapy to assess the correlation of these changes with clinical outcome. Patients (n=18) with previously untreated advanced HNSCC enrolled in the phase I clinical trial (NCT01935921) donated plasma specimens at baseline, during (week 5, week 14), after (6 months) and long-term after immunotherapy. Exosomes were isolated from plasma by size exclusion chromatography. The protein content of exosomes in fraction #4 (30-150nm vesicles) was measured. Immunocapture with biotinylated anti-CD3 mAb was used to separate T cell-derived from non-T cell-derived exosomes. CD3+ exosomes bound to streptavidin-labeled beads were harvested. The uncaptured exosomes, derived from CD3- cells, were bound to beads using biotinylated anti-CD63 mAb. Flow cytometry-based antigen detection on bead-bound exosomes was performed using labeled Abs specific for CTLA-4, PD-1 and PD-L1. Ratios of positive exosomes/total plasma exosomes were assessed for correlation with patients’ recurrence status at last follow-up (median of 14 months).

Results

Increases in the protein levels of total exosomes during therapy associated with disease recurrence. Decreases in frequency of CTLA-4+ exosomes in CD3+ and CD3- fractions or of PD-1 in the CD3+ fraction occurred as early as 5 weeks into therapy. Patients segregated into two groups based on increasing vs. decreasing PD-L1+ exosomes/total exosomes. 4 patients with increasing CTLA-4+ and PD-L1+ exosomes during therapy had recurrence and died of disease. The 5th patient with this “unfavorable” exosome profile remains disease free (NED). 13 patients who had “favorable” profiles (decreased CTLA-4+ and PDL1+ exosomes) have not recurred.

Conclusions

Changes in the ratio of exosomes carrying checkpoint proteins/total exosomes during immunotherapy associated with oncologic outcome. The checkpoint molecules content of T cell-derived vs non-T cell-derived exosomes defined “favorable” and “unfavorable” immune profiles that associated with the patients’ responses to therapy. Plasma exosomes in cancer patients emerge as early markers of response to immune therapies and outcome.

P90 RNA-based immune gene set signatures demonstrate immune modulation by RXDX-106, a novel TAM (TYRO3, AXL, MER) family small molecule kinase inhibitor

Elizabeth Tindall, Heather Ely, Yumi Yokoyama, Colin Walsh, Erin Lew, Amy Diliberto, Robin Nevarez, Jack Lee, Ian Silverman, Joanne Oh, Gary Li, Amanda Albert, Jason Christiansen, Robert Shoemaker
Ignyta, San Diego, CA, USA
Correspondence: Elizabeth Tindall (etindall@ignyta.com)

Background

The TAM family of receptor tyrosine kinases, including TYRO3, AXL, and MER, is pivotal in regulating the immune system. In the tumor microenvironment, TAMs promote anti-inflammatory/pro-tumorigenic activities. RXDX-106 is a potent and selective TAM inhibitor in preclinical development, shown to have immunomodulatory effects. We herein illustrate that (1) MC38, a syngeneic mouse model of colorectal cancer, is sensitive to RXDX-106, and (2) using RNA sequencing, we demonstrate RXDX-106 treatment results in the recruitment of various immune cell subtypes to the tumor microenvironment. Flow cytometry analysis supports key RNA sequencing findings.

Methods

MC38 tumor-bearing C57BL/6 mice were treated with and without RXDX-106 for 7 days. Tumor tissue was collected for cellular and molecular characterization to measure immune modulation, via flow cytometry and total RNA sequencing respectively. Single sample gene set enrichment analysis (ssGSEA) was applied to RNA sequencing data to calculate normalized enrichment scores (NES) for 15 adaptive and 13 innate immune cell types, based on previously published gene sets. RNA-sequencing significance levels are presented as false discovery rates (FDR) to correct for multiple hypothesis testing. Significance of flow cytometry data are indicated by probability (P) values calculated using the two-tailed T-test.

Results

MC38 tumors. A high proportion (66%, 33 of the top 50) of individual genes differentially expressed between treated and untreated mice have immunomodulatory roles. Furthermore, ssGSEA suggested a significant enrichment (FDR<0.01) of T cell subtypes, including T follicular helper (Tfh) cells (FDR=0.0005), Type 1 T helper (Th1) cells (FDR=0.0005), and central memory CD8+ T cells (FDR=0.0008) with RXDX-106 treatment. flow cytometry data revealed a significant increase in overall tumor infiltrating lymphocyte (TIL) recruitment (P<0.0001), as well as a non-significant trend towards increased CD8+ T cell infiltration in RXDX-106 treated mice. Innate immune cells, such as plasmacytoid dendritic cells (pDCs; FDR=0.0005) and macrophages (FDR=0.0005), also had significantly elevated NES values in treated mice. A negative correlation (r=-0.7) between macrophage NES and M2 secreted VEGFA expression was observed and may suggest infiltrating macrophages are polarized to the pro-inflammatory/anti-tumorigenic M1 phenotype upon treatment with RXDX-106. Flow cytometry data supported the observed increase in tumor associated macrophages (P=0.0008), and demonstrated M1 polarization by a highly significant (P<0.0001) increase in M1/M2 ratio with RXDX-106 treatment.

Conclusions

We have demonstrated the application of an RNA sequencing based approach to detect immunomodulatory effects and provided evidence supporting immunomodulatory activity of RXDX-106 in vivo.

P91 A harmonized evaluation system for expression of HLA class I antigens in formalin-fixed paraffin embedded cancer tissues

Toshihiko Torigoe1, Nobuyoshi Hiraoka2, Naoya Nakamura3, Eri Atsuyama4, Shingo Toji5, Yuji Heike4
1Sapporo Medical University School of Medicine, Sapporo, Japan; 2National Cancer Center Hospital, Tokyo, Japan; 3Tokai University School of Medicine, Isehara, Japan; 4St Luke’s International University, Tokyo, Japan; 5Medical Biological Laboratories Co., Ltd., Ina, Japan
Correspondence: Nobuyoshi Hiraoka (torigoe@sapmed.ac.jp); Yuji Heike

Background

Expression of human leukocyte antigen class I antigens (HLA-I) on tumor cells is crucial for target recognition of cytotoxic T cells (CTLs). Therefore, it should be an important biomarker for CTL-based cancer immunotherapy to evaluate the expression levels of HLA-I on tumor cells. A monoclonal anti-pan HLA-I antibody, EMR8-5 is the first reliable antibody that is suitable for immunohistochemistry (IHC) of formalin-fixed paraffin embedded (FFPE) tissues [1]The aim of this study is to establish the harmonized protocol to evaluate expression levels of HLA-I on cancer cells in FFPE tumor tissues.

Methods

First of all, we determined a standardized method of staining, and proposed tentative criteria to evaluate expression levels of HLA-I in tissue sections; (+++) as ≥90% of cancer cells are strongly positive, in which the staining density is equal or stronger than that in lymphocytes or endothelial cells, (++) as <90% and ≥50% of cancer cells are strongly positive, (+) as <50% and >10% of cancer cells are strongly positive or >10% of cancer cells are weakly positive, in which the staining density is weaker than that in lymphocytes or endothelial cells, and (-) as ≤10% of cancer cells are positive. Reference IHC photo atlas was made for each type of cancer. After IHC, the microscopic images were converted into digital photo files, and the immunolabeled cells were analyzed on the standardized calibrated monitors (ColorEdge CG2420, EIZO, Japan). Then we examined inter-observer reproducibility of the criteria by multi-institutional study using each 20 cases of 7 cancers (colorectal, breast, prostate, pancreatic, ovarian, lung, and kidney cancers), which had been surgically resected in National Cancer Center Hospital, Tokyo.

Results

IHC with anti-pan HLA-I antibody was performed using automated IHC slide staining system. Expression profiles of HLA-I were characteristic in each cancer type. We made reference photo atlas of IHC to harmonize the evaluation for each type of cancer. According to the reference atlas, four certified pathologists scored the expression levels of HLA-I independently. Inter-observer reproducibility for each cancer was within moderate to substantial agreement levels (Fleiss’s kappa values were within 0.5 to 0.7).

Conclusions

Expression of HLA class I antigens can be evaluated with an acceptable reproducibility in several cancers using FFPE tissues.

Reference

1. Pathol. Int. 2012;62;303-8

P92 Early biomarker correlates of severe neurologic events and cytokine release syndrome in ZUMA-1, a multicenter trial evaluating axicabtagene ciloleucel in refractory aggressive non-hodgkin lymphoma

Frederick Locke1, Marika Sherman2, John Rossi2, Allen Xue2, Michelle Mojadidi2, Sattva Neelapu3, Armin Ghobadi4, Lazaros Lekakis5, David Miklos6, Caron Jacobson7, Olalekan Oluwole8, Tanya Siddiqi9, Yi Lin10, Patrick Reagan11, William Go2, Jeff Wiezorek2, Adrian Bot2
1Moffitt Cancer Center, Tampa, FL, USA; 2Kite Pharma, Santa Monica, CA, USA; 3The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 4Washington University School of Medicine, Saint Louis, MO, USA; 5University of Miami Health System, Sylvester Comprehensive Care Center, Miami, FL, USA; 6Stanford University School of Medicine, Stanford, CA, USA; 7Dana-Farber Cancer Institute, Boston, MA, USA; 8Vanderbilt-Ingram Cancer Center, Nashville, TN, USA; 9City of Hope National Medical Center, Duarte, CA, USA; 10Mayo Clinic, Rochester, MN, USA; 11University of Rochester Medical Center, Rochester, MN, USA
Correspondence: Sattva Neelapu (kate.trueblood@nexusggmed.com)

Background

Axi-cel, an autologous anti-CD19 chimeric antigen receptor (CAR) T cell therapy, demonstrated an 82% objective response rate including 54% complete remissions in refractory aggressive NHL patients.1 NE and CRS are associated with CAR T cell therapy.1 Post-treatment levels of serum analytes, measured after the onset of clinical toxicities, were associated with severe NE and CRS.2 Lymphodepleting conditioning-related (day 0) and early post-CAR T cell treatment (day 1) biomarkers associated with NE and CRS may provide important mechanistic insight and inform toxicity management.

Methods

Serum samples were obtained pre and post axi-cel infusion, from 101 refractory aggressive NHL patients in the Phase 2 ZUMA-1 (NCT02348216) trial [1]. A panel of 44 analytes representing major categories of immune function were measured as described before. [2] Post hoc statistical analyzes evaluated associations of analyte levels on day 0 and day 1, in patients with/without grade ≥3 NE/CRS. Univariate Wilcoxon two-sample tests were applied and adjusted for multiplicity, to identify important co-variates. Multivariate machine learning (random forest analysis [3]) was conducted to rank the most influential co-variates.

Results

Grade ≥3 NE were observed in 28/101 (28%) patients. Those developing grade ≥3 NE had increased IL-15 and decreased perforin (P=0.0006 and P=0.0011, respectively) at Day 0, and increased IL-15, MCP-1 and IL-6 (P=0.0021, 0.0037 and 0.0085, respectively) and decreased perforin (P=0.0025) at Day 1. Multivariate analysis identified identical analytes as the univariate analysis and ranked the importance of analytes relative to each other. Grade ≥3 CRS was observed in 13/101 (13%) patients. Those developing grade ≥3 CRS had no statistically significant differences in analyte levels at Day 0, and increased IL-15 at Day 1 (adjusted P=0.026).

Conclusions

These analyzes indicate IL-15 in NE/CRS pathogenesis post- anti-CD19 CAR T cells treatment, consistent with previous reports [4, 5, 6]. Association between early serum markers (day 0 and 1) and grade ≥3 NE may support development of clinically useful prevention and treatment algorithms.

Trial Registration

ClinicalTrials.gov Identifier NCT02348216

References

1. Locke, et al. #CT019. AACR. 2017.

2. Locke, et al. #CT020. AACR. 2017.

3. Segal and Xiao. WIREs Data Mining Knowl Discov. 2011.

4. Turtle, et al. JClinOncol. 2017.

5. Rossi, et al. #46. SITC. 2016.

6. Kochenderfer et al. JCO. 2017.

P93 Small molecule binders of Ly6K inhibit tumorigenic growth and induce immunity via TGFb/Stat1 and PDL1 pathway

Megha Kumar1, Jinya Chen1, Idalia Cruz1, Purushottam Tiwari1, Aykut Uren1, Maksymilian Chruszcz2, Michael B. Atkins1, Louis M. Weiner1, Geeta Upadhyay1
1Lombardi Comprehensive Cancer Center, Georgetown University Medical center, Washington, DC, USA; 2Lombardi Comprehensive Cancer Center, Georgetown University Medical center, Columbia, SC, USA
Correspondence: Geeta Upadhyay (gu6@georgetown.edu)

Background

We have previously shown that members of the Ly6 gene family, namely Ly6E, Ly6D, Ly6H and Ly6K are expressed in multiple types of solid human cancers and that the increased expression of these genes is associated with poor outcome [1]. We also have found that Ly6K and Ly6E are required for in vivo tumor growth and anti-tumor immune response [2]. Ly6K is an attractive target due to its cancer cell-specific expression; normal organ expression of Ly6K is limited to testis.

Methods

We used surface plasmon resonance technology to identify two small molecule binders of Ly6K. These molecules showed strong binding to Ly6K (kd range 1-2mM) but not to Ly6E and Ly6D.

Results

Combined small molecule treatment led to reduced TGFb signaling and reduced PDL1 expression in MDA-MB-231 cells. Homology modeling showed that the two small molecules bind to the N-terminus and C-terminus of the mature Ly6K protein, respectively. Treatment of E0771 tumors in a syngeneic C57Bl/6 model with individual small molecules reduced tumor growth, while combination treatment with the two small molecules completely eliminated tumor growth. The treatment was halted after tumor disappearance. Three weeks later, the cured mice were re-challenged with E0771 cells, with complete, durable protection against tumor growth (Fig. 1).

Conclusions

These results suggest that Ly6K inhibitors have anti-tumor properties and induce host-protective anti-tumor immunity. We are currently investigating the molecular basis of these findings in vivo. Our data implicate the novel biomarker Ly6K as a therapeutic target and immune modulator in multiple tumor types with high expression of Ly6K.

References

1. Luo L, McGarvey P, Madhavan S, Kumar R, Gusev Y, Upadhyay G. Lymphocyte antigens 6 complex gene family members drive tumorigenesis. Oncotarget. 2016; 7(10):11165-93, DOI:10.18632. PMID: 26862846

2. AlHossiny, Luo L, Frazier W, Steiner, Gusev, Kallakury, Glasgow, Creswell, Madhavan, Kumar, Upadhyay. Lymphocyte antigens 6 complex K/E-transforming growth factor-beta axis promotes breast cancer tumorigenesis.Cancer Research.Published Online April 11, 2016; doi: 10.1158/0008-5472.CAN-15-2654. PMID: 27197181
Fig. 1 (abstract P93).

Treatment with Ly6K binding small molecule induce immunity

P94 The RareCyte® platform for identifying rare antigen-specific circulating CD4 and CD8 T cells in Merkel cell carcinoma

Lance U'Ren1, Natalie Vandeven2, Erika Hayes3, Jennifer Chow1, Jackie Stilwell1, Paul Nghiem2, Eric Kaldjian1
1RareCyte Inc., Seattle, WA, USA; 2University of Washington, Seattle, WA, USA; 3Benaroya Research Institute, Seattle, WA, USA
Correspondence: Lance U'Ren (luren@rarecyte.com)

Background

Merkel cell carcinoma (MCC) is an often-lethal skin cancer associated with the Merkel cell polyomavirus (MCPyV) in 80% of cases. Although checkpoint immunotherapies have proven efficacious for a cohort of MCC patients, there is still a need to identify predictive biomarkers. Tumor antigen-specific T cells in peripheral circulation can be reflective of their phenotype within the tumor microenvironment making them attractive candidates as cellular biomarkers. However, the frequency of MCC antigen-specific CD4 and CD8 T cells can often be below the detection limit of flow cytometry making them difficult to find. We thus employed the highly sensitive RareCyte platform to detect and phenotype rare antigen-specific CD4 and CD8 T cells.

Methods

MCC patient peripheral blood mononuclear cells (PBMC) was stained and plated onto chamber well slides and imaged on the CyteFinder® 6-channel fluorescent scanning microscope. CD4 and CD8 T cells were identified by MCPyV-specific tetramer staining, and characterized with various exhaustion markers, including PD1, LAG3, TIM3, and CTLA4. Single MCPyV-specific CD4 T cells were picked and placed individually into tubes using an integrated computer-controlled retrieval system (CytePicker®). RNA was extracted and single cell sequencing was performed on the alpha and beta chains of the T cell receptor (TCR).

Results

We consistently identified MCPyV -specific CD4 and CD8 T cells below the detection limit of flow cytometry without the need for enrichment or large blood volumes. Furthermore we have high confidence in the cells identified and retrieved due to the morphologic information obtained with imaging that is not possible by flow cytometry. We were able to identify patients that had exhausted MCPyV-specific T cells in circulation and characterize the expression of exhaustion markers. We were also able to sequence the TCR from single rare MCPyV-specific CD4 T cells retrieved using the platform.

Conclusions

Using this approach to identify rare but functionally important MCPyV-specific T cells will enhance our understanding of the immune response against this oncogenic virus and may help guide patient selection for checkpoint therapies.

P95 Development and application of a 6-channel immunofluorescence assay for investigating PDL1 and IFR1 expression on circulating tumor cells

Lance U'Ren, Daniel Campton, Jackie Stilwell, Eric Kaldjian
RareCyte Inc, Seattle, WA, USA
Correspondence: Lance U'Ren (luren@rarecyte.com)

Background

There is a need for non-invasive predictive biomarkers of response to anti-PD1/PDL1 therapies. Assessment of circulating tumor cells (CTCs) is a rational approach to non-invasive sampling of tumors to understand PDL1 phenotype. IFN-gamma signals through the JAK/STAT cascade to induce PDL1 via the Interferon Regulatory Factor-1 (IRF1) transcription factor and is a potent inducer of PDL1 expression in tumor cells. Previous studies comparing PDL1 and IRF1 tissue staining of melanoma showed that IRF1 had a higher predictive value than PDL1 of response to anti-PD1/PDL1 therapy [1]. Using CTC models and the RareCyte platform for CTC identification, we developed a multi-parameter assay that allows simultaneous PDL1 and IRF1 assessment after CTC identification. We applied this assay to clinical samples of breast and Merkel cell carcinoma (MCC).

Methods

Peripheral blood from normal donors or cancer patients was collected into RareCyte blood collection tubes. PDL1(+) and PDL1(-) CTC models were created by culturing A549 overnight with or without 10ng/ml INF-gamma. The A549 cells were then spiked into normal donor blood and buffy coats isolated from 7.5ml of blood by AccuCyte® separation and spread onto slides. Slides were stained with a 6-marker panel that included antibodies to pan-cytokeratin (CK), EpCAM, CD45, CD45, PDL1, IRF1, and a nuclear dye on the Leica Bond Rx auto-stainer. Slides were scanned with CyteFinder® and CTCs identified by CK and/or EPCAM positivity and negative CD45 staining. Confirmed CTCs were then assessed for expression of PDL1 and IRF1; cell compartment was also recorded for IRF1 staining.

Results

In unstimulated A549s, PDL1 staining was absent but cytoplasmic staining of IRF1 was often detectable. In contrast, IFN-gamma stimulated A549s demonstrated PDL1 staining and strong nuclear staining for IRF1. In patient samples, PDL1 was identified in a minority of CTCs in both breast cancer and MCC. PDL1(+) CTCs identified had concomitant strong nuclear IRF1 signal. IRF1 was seen in the majority of the PDL1(-) CTCs but staining could be either cytoplasmic and/or nuclear.

Conclusions

In this study we demonstrate the feasibility of combining PDL1 and IRF1 biomarkers with CTC identification for non-invasive assessment. The cellular distribution of IRF1 in combination with PDL1 expression may add predictive value to the assay.

Reference

1. Smithy, Moore, Pelekanou, Rehman, Gaule, Wong, Neumeister, Sznol, Kluger, Rimm. J Immunother Cancer. 2017; 5:25.

P96 Tumor mutation burden, microsatellite instability and chromosomal instability analysis using low pass whole genome sequencing of single circulating tumor cell

Angel Rodriguez, Jerry Lee, Ramsay Sutton, Rhett Jiles, Yipeng Wang, Mark Landers, Ryan Dittamore
Epic Sciences, San Diego, CA, USA
Correspondence: Yipeng Wang (yipeng.wang@epicsciences.com)

Background

Tumor Mutation Burden (TMB), Microsatellite Instability (MSI) and Chromosomal Instability (CIN) represent the majority of Genomic instability (GI) in metastatic patients. Recent studies show that TMB and MSI are emerging immune checkpoint inhibitor drug sensitivity biomarkers, and CIN is a sensitivity marker for PARP inhibitors. Assessment of GI in bulk tumor samples is well explored, but it is limited by sample availability and tumor heterogeneity. Analysis of ctDNA is feasible for TMB and MSI but not CIN, and it also suffers in sensitivity and specificity in patients who harbor subclonal GI, limiting its clinical utility. Epic Sciences’ Circulating Tumor Cell (CTC) platform employs a non-enrichment based approach and could provide insight into subclonal heterogeneity. Here we present downstream single cell GI assay(s) for the detection of TMB, MSI and CIN from individual CTCs.

Methods

Contrived samples were prepared by spiking prostate cancer cell lines, LNCaP, PC3 and VCaP, into healthy donor blood. Red blood cells were lysed, nucleated cells deposited onto slides, slides immunofluorescently stained (DAPI, CK, CD45, and Androgen Receptor), and identified cancer cells individually isolated from the slides. Each recovered cell was lysed, whole genome amplified (WGA), shotgun library prepared, and low pass whole genome sequenced using Illumina NextSeq 500. Data were analyzed for TMB scores and large scale transitions. MSI was measured using Qiagen Type-It microsatellite PCR kit for four sites. Samples from metastatic castration resistant prostate cancer (mCRPC) patients were included to evaluate clinical feasibility.

Results

TMB scores for LNCaP (average 652) were significantly higher than PC3 (558), VCaP (548), and WBC from healthy donor (540) with p<0.01. MSI assay confirmed that LNCaP is the only MSI-H cell line with insertions/deletions found in 3 of 4 microsatellite sites. No MMR deficiencies were found in PC3 and VCaP. LST analysis shows PC3 (average 33) and VCaP (33) have much higher LST scores than LNCaP (11). A wide range of TMB (375-861) and LST scores (0-70) are observed inter- and intra- mCRPC patient samples.

Conclusions

These data demonstrate the feasibility of detecting three types of genomic instabilities at the single cell level using the Epic Sciences CTC Platform. Inter- and intra-patient heterogeneity is observed in the small patient cohort. Additionally, the results confirm that MSI and HRD are likely mutually exclusive driver events, driving tumor selection in mCRPC. Further studies are on-going to investigate the correlation of GI markers with PARPi and I/O checkpoint inhibitor responses.

P97 Simultaneous characterization of rare immune cell subpopulations and PD-L1 expressing CTCs in peripheral blood of cancer patients

Adam Jendrisak, Angel Rodriguez, Jiyun Byun, Ryon Graf, Yipeng Wang, Mark Landers, Ryan Dittamore
Epic Sciences, San Diego, CA, USA
Correspondence: Yipeng Wang (yipeng.wang@epicsciences.com)

Background

Expression of PD-L1 on tumor and immune markers in tumor tissue are associated with improved response to PD-1 and PD-L1 checkpoint inhibitors. However, each alone has limited predictive utility. Multimodal characterization of both the tumor and host immune system is an unmet medical need for the improved prediction of response to immunotherapy. Metastatic lesions are likely to be under-sampled and require a liquid biopsy, given tumor heterogeneity and evolution and temporal changes in the host immune system. We sought to examine expression of PD-L1 on circulating tumor cells (CTCs) as well as characterize rare immune cell populations with a non-invasive liquid biopsy. Examining dynamic biomarker changes in longitudinal samples could enable the development of novel diagnostic tools for response prediction and pharmacodynamics studies related to immunotherapy.

Methods

Blood samples from lung cancer patients were collected and shipped to Epic Sciences. Contrived samples were also developed by spiking cancer cell lines into healthy donor blood. Red blood cells were lysed, and nucleated cells plated onto glass slides. Slides were stained with DAPI as well as multiple immune cocktails and imaged. Targets included pan-CK, CD45, PD-L1, CD4, CD8, Ki-67, LAG-3, and TIM-3. Approximately 3 million nucleated cells per slide were examined through advanced digital pathology pipelines to detect and quantify changes in T cell populations to infer immune activation, exhaustion, suppression, and circulating tumor burden.

Results

Epic Sciences’ rare cell detection platform has an analytically validated limit of detection of 1 cell/mL of blood. Three immuno-panels were developed to profile leukocyte subpopulations and CTCs simultaneously. PD-L1+ leukocytes detected in 30 out of 33 lung cancer patients, and the incidence of PD-L1+ leukocytes ranged from 0% to 0.138% (average 0.0213%, median 0.0118%). PD-L1+ CTCs were observed in the presence of both high and low counts of PD-L1+ leukocyte populations.

Conclusions

The low limit of detection of the Epic Sciences CTC platform coupled with ability to archive patient blood samples allowed for retrospective precise quantification of leukocyte subpopulations and PD-L1 expression on CTCs retrospectively. Development of a liquid biopsy based platform that can simultaneously measure immune biomarkers in CTCs as well as on leukocytes will allow for real time assessment and monitoring of response to immune checkpoints inhibitors.

P98 Immunological profiling of baseline and resected biopsies from locally/regionally advanced/recurrent melanoma treated with neoadjuvant combination ipilimumab (3mg/kg or 10mg/kg) and high dose IFN-α2B

Sarah Warren1, Priyanka Vallabhaneni2, Jack El-Sawada2, Andrew White1, Xing Ren1, Alessandra Cesano1, Joseph Beechem1, Ahmad Tarhini2
1NanoString Technologies, Inc., Seattle, WA, USA; 2Hillman Cancer Center, University of Pittsburgh Medical Center, Pittsburgh, PA, USA
Correspondence: Sarah Warren (swarren@nanostring.com)

Background

Melanomas are the most responsive tumor type to immune checkpoint blockade, but clinical response rates remain suboptimal. Gene expression profiling of the pretreatment tumor biopsies is an attractive approach to developing predictive biomarkers of response because it is able to directly measure the activity of the immune system within the tumor. Furthermore, characterization of the transcriptional changes induced by treatment allows insights into the mechanism of action of the therapy. UPCI 11-063 is a clinical trial to assess the safety of two doses of ipilimumab combined with high dose IFN-α2B in the neoadjuvant setting. Both treatment regimens induce clinical responses, with reduced toxicity observed in the low dose ipilimumab arm. This study is a follow on biomarker analysis to investigate gene expression in pretreatment and post treatment biopsies.

Methods

This was a Phase I trial of 30 patients with locally/regionally advanced/recurrent melanoma who received either high (10 mg/kg) or low (3 mg/kg) dose ipilimumab plus high dose IFN-α2B. Tumor biopsies collected prior to treatment and at the time of surgery (6-8 weeks after starting neo-adjuvant treatment) were profiled on the NanoString® platform.

Results

Transcriptional profiling of tumor biopsies collected prior to treatment or at the time of surgery was performed in order to find patterns of gene expression that correlated with pathological complete response (pCR), radiologic response (RR) or recurrence-free survival (RFS). In the baseline tumor biopsies, a number of genes were positively associated with pCR, including several members of the WNT signaling pathway. The T cell inflamed Gene Expression Profile (GEP), a signature of a peripherally suppressed immune response in the tumor, was associated with longer RFS. Differential gene expression of paired pre- and post-treatment biopsies identified a number of genes upregulated in response to ipilimumab/IFN treatment, including signatures of multiple immune cell populations, including T cells and macrophages, which were associated with pCR.

Conclusions

The combination of ipilimumab/IFN as a neoadjuvant for treatment of melanoma has the potential to improve response rates by activating quiescent immune responses via multiple pathways. Expression profiling of pre and post treatment samples suggested a mobilization of immune cells to the tumor, consistent with activation of local immunity. The GEP, which predicts response to pembrolizumab in a variety of solid tumors, may also be useful in the setting of ipilimumab/IFN. Further work in this or similar sample sets may extend these observations.

Trial Registration

ClinicalTrails.gov Identifier NCT01608594

P99 Biomarker analysis from the OpACIN trial (Neo-/adjuvant ipilimumab + nivolumab (IPI+NIVO) in palpable stage 3 melanoma)

Lisette (EA) Rozeman1, Lorenzo Franchi1, Thomas Kuilman1, Oscar Krijgsman2, Alexander (ACJ) van Akkooi1, Pia Kvistborg1, Hans van Thienen1, Bauke Stegenga3, Dana Cullen3, Brian Lamon3, John (JBAG) Haanen1, Alessandra Cesano4, Sarah Warren4, Annegien Broeks1, Ton Schumacher1, Christian Blank1
1Netherlands Cancer Institute, Amsterdam, Netherlands; 2o.krijgsman@nki.nl, Netherlands, Netherlands; 3Bristol-Myers Squibb, Lawrence Township, NJ, USA; 4NanoString Technologies, Seattle, WA, USA
Correspondence: Sarah Warren (swarren@nanostring.com)

Background

The combination of IPI+NIVO induces high response rates and improved overall survival in late stage melanoma. T cell checkpoint inhibition is of greatest value at the moment of TCR triggering and therefore dependent on the amount of antigen present, indicating that adjuvant immunotherapy will work most efficiently, when initiated prior to surgery.

Methods

Two-arm Phase 1b feasibility trial consisting of 20 high risk AJCC stage 3B/C melanoma patients with palpable nodal disease receiving the combination of IPI 3mg/kg and NIVO 1mg/kg, either adjuvant four courses after surgery, or split neo-adjuvant and adjuvant.

Results

In this update 20 patients are evaluable. Neo-adjuvant application of IPI+NIVO was feasible and no surgery-associated adverse events were attributed to (neo-) adjuvant therapy. 18/20 patients had to stop earlier due to grade 3/4 toxicities. Neo-adjuvant IPI+NIVO reduced tumor load in 8/10 patients (3 pCR, 4 near pCRs [minimal remaining micro metastases], 1 pPR [<50% vital tumor cells], 1 pSD and 1 pPD). So far, none of the responders in the neo-adjuvant arm has relapsed. Relapse was observed for the 2 non-responders within the neo-adjuvant arm and for 3 patients within the adjuvant arm.

We will present at the detailed biomarker analysis, comprising DNA and RNA sequencing, NanoString® RNA and protein profiling of PBMC, NanoString digital spatial profiling of tumors, multiplex immunohistochemistry of tumors and TCR sequencing. We found that mutational load did not correlate with response and that IFN signature, inflamed signature and BATF3 signature were mainly positive predictive. Using NanoString spatial profiling, we found that PD-L1, PD-1, β2M were associated with favorable outcome.

Conclusions

The combination of IPI+NIVO in the (neo-)adjuvant treatment setting for high risk stage 3 melanoma patients is promising and currently tested in an international phase 2 randomized trial comparing different combination schemes (OpACIN-neo trial, NCT02977052) with the aim of preserving efficacy, but reducing toxicity. Biomarkers identifying patients responding upon neo-adjuvant IPI+NIVO and remaining relapse-free for a long time, will help to select the patients that need and don’t need to be exposed to IPI+NIVO associated toxicity.

Trial Registration

ClinicalTrials.gov Identifier NCT02437279

P100 Pretreatment gene expression correlation with clinical response to pembrolizumab or nivolumab in metastatic melanoma

Sarah Warren1, Patrick Danaher1, Tobias Schatton2
1NanoString Technologies, Seattle, WA, USA; 2Brigham-Women's Hospital, Boston, MA, USA
Correspondence: Sarah Warren (swarren@nanostring.com)

Background

PD-1 checkpoint blockade with the therapeutic antibodies, pembrolizumab or nivolumab, improves survival in patients with metastatic melanoma. However, not all patients experience clinical benefit. The NanoString T cell Inflamed Gene Expression Profile (GEP) measures genes associated with CD8+ T cells, natural killer cells, IFN signaling, and antigen presentation that are reflective of a pre-existing, PD-1-suppressed adaptive immune response within the tumor, and associated with positive treatment outcomes to pembrolizumab in advanced-stage melanoma and other solid tumors. Here, we evaluate the GEP in pre-treatment tumor biospecimens obtained from patients receiving either nivolumab or pembrolizumab. Further, this study examines differential gene expression in PD-1-responsive vs. non-responsive melanomas to identify potential resistance mechanisms, which could be targets for therapeutic intervention.

Methods

Formalin-fixed, paraffin-embedded pre-treatment tumor biopsies from patients receiving either nivolumab or pembrolizumab in the community setting were profiled with NanoString’s nCounter gene expression codesets. A total of 110 patients were included from three independent cohorts. Clinical benefit to PD-1 therapy was defined as an immune-related complete or partial response. PD-L1 expression on tumor cells was evaluated by immunohistochemistry in 36 samples from the pembrolizumab-treated cohort using the 22C3 anti-PD-L1 antibody clone, and staining was scored by two independent pathologists.

Results

The GEP predicted clinical benefit from pembrolizumab (n=85), but was not predictive for nivolumab (n=25), although the small sample size may limit statistical power. We plan to expand the nivolumab cohort prior to presentation of these results. Additionally, the GEP was more predictive of response to pembrolizumab than PD-L1 IHC, as assessed by ROC curves. Furthermore, clinical response to either therapeutic antibody was associated with an enrichment of genes expressed by lymphoid cells, as determined by cell type-specific gene expression. Additional analyzes of the total cohort are ongoing and will be presented.

Conclusions

The TIS predicted clinical response to pembrolizumab, but appears not to predict response to nivolumab. Validation and further studies will be required to determine if this observation holds true in larger cohorts, and if it relates to differences between the two agents.

P101 Multiplex biomarker detection in FFPE tumor samples

Mike Millar, Louise Welch, Dawn Lyster, Mariana Beltran, Sevi Giakoumelou
Aquila BioMedical, Edinburgh, UK

Background

Immuno-oncology (IO) drug discovery research requires a multi-staged approach: pre-clinical in vitro assays, in vivo studies and histological analysis. We demonstrate the power of incorporating a specialist histological approach in formalin fixed paraffin embedded (FFPE) tissues early in the drug development process can provide high value data to complement functional in vitro assay readouts to inform on cancer immunotherapy clinical trial strategy.

Simultaneously imaging multiple IO markers on a section provides morphological, pathological and spatial information of the juxtaposition of different immune cell types (quantified by image analysis techniques).

Multispectral imaging approaches:
  1. 1.

    Single immunofluorescence (IF), immunohistochemistry (IHC) and RNAscope (mRNA)

     
  2. 2.

    Dual and triple IF, IHC and RNAscope

     
  3. 3.

    Multiplex IF, IHC and RNAscope

     
  4. 4.

    Mixed multiplex detection of protein, mRNA and/or protein-mRNA

     
  5. 5.

    Quantitative image analysis

     

Methods

4 μm FFPE sections were stained using validated protocols on automated staining platforms (Leica Bond III/RX). Imaging performed using colorimetric and fluorescent scanners.

Results

Automated single-plex IHC or IF demonstrates protein (Figure 1) and mRNA (Figure 2) expression. Multiple protein targets are visualised using colorimetric (Figure 3) and 4-plex protein IF staining (Figures 4-7). It is possible to stain 6 or 7 markers on one section (Figures 8 & 9), visualising the cellular location and spatial distribution of markers in relation to each other.

Conclusions

Specialised histological staining approaches utilised during drug discovery research provide valuable insights into the spatial expression of IO biomarkers. A pragmatic approach to antibody selection and validation powerfully combines with RNAscope to facilitate demonstration of novel and challenging IO biomarkers in situ, and can complement pre-clinical in vitro assay data.

P102 Mining the periphery for tumor-relevant B and T cells

Katherine L. Williams1, Lukas Pfannenstiel2, Eldar Giladi1, Yvonne Leung1, Xiaobin Tang1, Shuwei Jiang1, Dongkyoon Kim1, Kevin Williamson1, Sean M. Carroll1, Ngan Nguyen1, Gregg Espiritu Santo1, Sini Reponen1, Xiaomu Chen1, Norman M. Greenberg1, Daniel Emerling1, Wayne Volkmuth1, Brian Gastman2, Guy Cavet1
1Atreca, Inc., Redwood City, CA, USA; 2Cleveland Clinic Lerner Research Institute, Cleveland, OH, USA
Correspondence: Brian Gastman (kwilliams@atreca.com); Guy Cavet

Background

Identification of tumor-specific T and B cells is key to understanding the immune response to cancer, but tumor tissue is only available in rare circumstances. Here we identify biomarkers for more accurate identification of tumor-reactive B and T cells from peripheral blood.

Methods

Matched peripheral blood and tumor samples from four donors with either advanced head and neck cancer or melanoma were analyzed by flow cytometry using markers including CD3, CD4, CD8, CD19, CD20, CD28, CD38, HLA-DR, NKG2D and TIGIT. From the periphery, we index sorted CD3+CD8+PD1+ T cells and collected CD19+CD20-CD38+ plasmablasts (PBs), as these are phenotypes likely to harbor tumor-reactive cells, and generated paired α/β and heavy/light chains from single cells. For tumor samples, CD8+ T cells, CD19+CD20-CD38+ PBs and CD19+CD20+ B cells were sorted and RNA was extracted for deep sequencing of β and heavy chains. Tumor-derived cells were deep sequenced to identify the largest possible numbers of clones present in tumor. T cell lineages were defined as having identical β-CDR3 nucleotide sequences, while B cell lineages were defined as having identical VH genes, and 80% amino acid similarity in the H-CDR3.

Results

We generated an average of 390 (314-468) paired α/β sequences from peripheral blood CD3+CD8+PD1+ T cells, and 4,692, 4,968, 7,346 and 108,032 clones from deep sequencing of tumor CD3+CD8+ T cells. 63% of peripheral CD8+PD1+ T cell clones overlapped with tumor-resident CD8+ clones. Among peripheral CD8+PD1+ T cells, CD28- clones had significantly greater overlap with tumor-resident T cell clones than CD28+ clones (p=0.005, Fisher’s exact test). There was also a trend toward greater overlap with tumor clones for the CD38+ subset of CD8+PD1+ peripheral T cells. Additional analyzes are underway to evaluate other T cell markers.

In analysis of B cell responses in two donors, 53.2% and 52.4% of peripheral plasmablast lineages overlapped with tumor B cell lineages. Additionally, the peripheral PBs that overlapped with the tumor population had a higher number of nucleotide mutations than the peripheral PBs that did not overlap (32.49 vs 30.08, p=0.06, two-sided t-test).

Conclusions

In this study, we demonstrated significant overlap between peripheral plasmablasts, peripheral PD1+CD8+ T cells and the tumor lymphocyte populations of advanced cancer patients. We further demonstrated that markers such as CD28 may be used to enrich for tumor-resident clones. Subsequent studies will expand upon these analyzes with additional donors and markers.

References

1. Van der Loos, Houtkamp, de Boer, Teeling, van der Wal, Becker. Immunohistochemical detection of interferon-gamma: fake or fact? J Histochem Cytochem. 2001;49(6):699-710.

2. Toth, Mezey. Simultaneous visualization of multiple antigens with tyramide signal amplification using antibodies from the same species. J Histochem Cytochem. 2007;55(6):545-554.

3. Tsujikawa, Kumar, Borkar, Azimi, Thibault, Chang, Balter, Kawashima, Choe, Sauer, El Rassi, Clayburgh, Kulesz-Martin, Lutz, Zheng, Jaffee, Leyshock, Margolin, Mori, Gray, Flint, Coussens. Quantitative multiplex immunohistochemistry reveals myeloid-inflamed tumor-immune complexity associated with poor prognosis. Cell Rep. 2017;19(1):203-217.

P103 Tumor infiltrating lymphocyte (TIL) percentage as a prognostic biomarker for overall and relapse free survival in Hu14.18-IL2 treated resectable recurrent stage 3 or 4 melanoma patients

Richard Yang1, Mark Albertini2, Jacquelyn Hank2, Hans Loibner3, Stephen Gillies4, Paul Sondel2, Erik Ranheim2
1University of Wisconsin - Madison Hospitals and Clinics, Madison, WI, USA; 2University of Wisconsin - Madison School of Medicine and Public Health, Madison, WI, USA; 3Apeiron Biologics, Vienna, Austria; 4Provenance Biopharmaceuticals, Carlisle, MA, USA
Correspondence: Richard Yang (rkenny168@gmail.com); Paul Sondel

Background

Recurrent stage 3 or 4 melanoma patients (pts) have a poor prognosis. The presence of tumor infiltrating lymphocytes (TILs) have shown prognostic value in pt survival in melanoma, particularly for pts receiving checkpoint blockade immunotherapy. We investigated the prognostic value of TILs identified on H&E sections of tumors from pts who have already undergone one cycle of hu14.18-IL2 (an immunocytokine fusion protein of anti-disialoganglioside antibody and IL-2) immunotherapy compared to hu14.18-IL2 naïve tumors.

Methods

Between 2008-2012, twenty-three stage 3 or 4 recurrent, but resectable, melanoma pts were enrolled in CO-05601 (NCT00590824) and randomized to receive the first of 3 courses of hu14.18-IL2 either before or after surgical resection of all sites of disease. At the time of surgery, two pts within this trial were found to be not completely resectable and were taken off study. Following routine pathologic assessment, representative sections of each tumor were coded and given to our board certified hematopathologist (author EAR) for TIL assessment. Scatter plots were made for exploratory analyzes to interrogate the relationship between treated TILs and survival outcomes with stratification based on whether pts had hu14.18-IL2 treatment. Kaplan Meier survival curves interrogated the relationship between TILs and survival outcomes (relapse free survival [RFS] and overall survival [OS]) in relation to treatment status.

Results

There was a significant correlation between increased percentage of TILs and increased RFS (p=0.0166; n=13) and OS (p=0.0223, n=13) in pts whose tumors were obtained after hu14.18-IL2 treatment. These correlations were not seen in pts whose tumors were obtained prior to hu14.18-IL2 (not significant, n=8). In addition, when comparing pt tumors with high TILs (above the median of 5%, n=11) and low TILs (below 5%, n=10), there was an increase in RFS (p=0.0446, median survival: high TILs - not reached, low TILs - 4.65 months) and a trend towards increased OS (p=0.0540, median survival: high TILs - not reached, low TILs - 45 months, Fig. 1a-b). Additionally, hu14.18-IL2 treated pt tumors with high TILs (n=6, median RFS and OS not reached) showed significantly increased RFS (p=0.0159) and OS (p=0.0208) compared to hu14.18-IL2 treated pt tumors with low TILs (n=5, median RFS - 4.2 months; OS - 41.3 months (Fig. 1C-D), whereas hu14.18-IL2 naïve tumors did not show this relationship.

Conclusions

These results suggest that resection specimens and biopsies obtained post-hu14.18-IL2 immunotherapy should be further investigated to monitor for TILs, as they may be associated with a beneficial immunotherapeutic effect.

Trial Registration

ClinicalTrials.gov Identifier NCT00590824
Fig. 1 (abstract P103).

TIL Percentage Prognosticates Increased OS and RFS in Hu14.18-IL2 Treated Melanoma Patients

P104 A study of PD-L1 diagnostic assay concordance in urothelial carcinoma

Magdalena Zajac1, Marietta Scott1, Marianne Ratcliffe2, Alan Sharpe1, Paul W Scorer1, Craig Barker1, Hytham Al-Masri3, Marlon Rebelatto4, Jill Walker1
1AstraZeneca, Cambridge, UK; 2AstraZeneca, Macclesfield, UK; 3Hematogenix, Tinley Park, IL, USA; 4Medimmune, Gaithersburg, MD, USA
Correspondence: Magdalena Zajac (jubilee.stewart@parexel.com)

Background

In urothelial carcinoma (UC), multiple pharmaceutical companies are investigating programmed cell death ligand-1 (PD-L1) expression using different antibody clones, staining protocols, and scoring algorithms. A stronger understanding of the comparative technical performance of these assays will allow appropriate interpretation of clinical outcomes for patients with UC treated with different anti-PD-1/PD-L1 therapies.

Methods

325 tumor biopsy samples from patients with UC were assessed using 4 commercially available PD-L1 diagnostic assays: Ventana SP263, Ventana SP142, Dako 28-8 and Dako 22C3. Assays were performed in an CLIA accredited laboratory, per the device protocol and scored for tumor cell (TC) and immune cell (IC) PD-L1 staining. Analytical concordance was calculated pairwise between assays using the Spearman (ρ) rank correlation coefficient. Classification concordance, including agreement between clinically relevant scoring algorithms, was investigated using overall/positive/negative percentage (OPA/PPA/NPA) agreement at multiple cut-offs.

Results

Data indicated good association, with a Spearman correlation coefficient of ≥0.8 for each pairwise comparison of the 4 assays for IC, and ≥0.7 for TC. When applying the SP263 clinical algorithm to 22C3 there is good agreement on positive cases with SP263 (OPA, PPA, NPA >80%) and vice versa.

Conclusions

Analytically, SP263, 22C3, and 28-8 assays showed similar performance for TC and IC PD-L1 expression. However, while SP142 showed similar analytical performance to the other 3 assays for IC staining, it appeared to be the least sensitive with the fewest samples showing PD-L1 staining on TC. This confirms previous observations from studies on assay performance in NSCLC and HNSCC.

When comparing scoring algorithms, caution should be taken when applying any algorithms utilising TC measures of PD-L1 to SP142 due to differences in sensitivity of the assay. Concordance achieved with SP263 algorithm applied to the 22C3 assay and vice versa builds optimism that those 2 assays could be used interchangeably in informing patient treatment decisions for durvalumab.

P105 Infino: Bayesian inference to distinguish immune cell expression phenotypes, estimate immune infiltration into tumor microenvironment, and investigate exhausted T cell phenotype

Maxim Zaslavsky1, Jacqueline Buros Novik1, Eliza Chang1, Alexander Huang2, Josephine Giles2, Robert Orlowski2, E. John Wherry2, Jeffrey Hammerbacher1
1Icahn School of Medicine at Mount Sinai, New York, NY, USA; 2Perelman School of Medicine, Philadelphia, PA, USA
Correspondence: Maxim Zaslavsky (maxim@hammerlab.org)

Background

Robust quantification of immune cell infiltration into the tumor microenvironment may shed light on why only a small proportion of patients benefit from checkpoint therapy. Indeed, the immune contexture has recently been associated with prognosis and response to therapy [1], and the exhausted T cell phenotype is likely to be implicated in response to checkpoint blockade [2]. However, traditional measurement of immune cell content around a tumor using cell surface markers is low-throughput, requiring manual intervention by a pathologist. Computational alternatives instead estimate immune cell type abundances by deconvolving the gene expression mixture measured in the tumor microenvironment.

Methods

After constructing synthetic gene expression mixtures with known immune cell abundances, we evaluate the performance of four existing immune cell mixture deconvolution methods [3]. Our analysis discovers that they have high confidence but substantial deconvolution error for certain mixtures. We then propose a new Bayesian method, infino, that learns the relationships between immune cell types to estimate their abundances in RNA-seq data. Modeling these relationships is crucial for our deconvolution ability, enabling investigation of phenotypes of interest.

Results

Using Bayesian generative modeling, infino estimates a probability distribution for the abundance of each immune cell type in an RNA-seq mixture. We demonstrate comparable performance to state-of-the-art infiltrate quantification methods, as well as more rigorous quantification through infino’s probabilistic estimates at all levels of the immune cell type hierarchy. Additionally, we extend infino to estimate stromal and tumor cell proportions for a complete accounting of the tumor microenvironment. We present results from a patient cohort, as well as infino’s estimation of the exhausted T cell phenotype, to investigate variation in response to checkpoint blockade.

Conclusions

By modeling the relationships between immune cell types, our novel Bayesian approach to immune infiltrate quantification produces robust deconvolution of gene expression mixtures like those found in the tumor microenvironment. Our results help probe the differential effectiveness of immunotherapy.

References

1. Galon, Mlecnik, Bindea, Angell, Berger, Lagorce, Lugli, Zlobec, Hartmann, Bifulco, et al.Towards the introduction of the ‘Immunoscore’ in the classification of malignant tumors. J Pathol. 2014; 232(2):199-209.

2. Sen, Kaminski, Barnitz, Kurachi, Gerdemann, Yates, Tsao, Godec, LaFleur, Brown, et al. The epigenetic landscape of T cell exhaustion. Science. 2016; 354(6316):1165-1169.

3. Hackl, Charoentong, Finotello, Trajanoski. Computational genomics tools for dissecting tumor-immune cell interactions. Nat Rev Genet. 2016; 17(8):441-58.

P106 Bioinformatic analysis of the tumor immune landscape to prioritize combinations with antibody-drug conjugates (ADCs)

Xiaoling Zhang, Song Wu, Brandon W. Higgs, Koustubh Ranade
MedImmune, Gaithersburg, MD, USA
Correspondence: Xiaoling Zhang (zhangx@medimmune.com)

Background

Immunotherapy has transformed the treatment of multiple tumors, yet significant unmet need remains. Antibody-drug conjugates (ADCs) have had limited success, with some evidence to support that they can reshape the tumor immune-environment to increase sensitivity to immunotherapy. Thus, combining immunotherapy and ADCs may enhance the benefit of both; however it is unclear how to prioritize such combinations. In this study, 52 ADC targets under clinical investigation were analyzed by bioinformatics approaches to examine the intersection of expression of these ADC targets and the tumor immune landscape to identify targets and pathways that may increase the likelihood of success of ADC through immunotherapy combinations.

Methods

Thirty-seven ADC targets were analyzed across 16 solid tumor types and 15 ADC targets were analyzed in 2 hematological cancers using transcriptome data in TCGA. Co-expression of ADC targets with inflamed or non-inflamed tumor microenvironment was defined based on a previously reported IFNγ mRNA expression signature that showed strong association with ORR and survival in patients treated with durvalumab. Top ranked IFNγ-associated ADC targets were further analyzed to identify the common molecular signatures shared by these targets.

Results

Out of 592 combination counts of 37 ADC targets in solid tumors, 163 (28%) were associated with inflamed signature whereas 67 (11%) were associated with non-inflamed signature. Out of 30 combination counts of 15 ADC targets in hematological cancers (AML and DLBCL), 14 (47%) were associated with inflamed signature whereas 5 (17%) were associated with non-inflamed signature. Top ranked ADC targets were associated with inflamed signature in a majority of tumor types (>10 out of 16), including those known to be either highly responsive or less responsive to immune checkpoint inhibitors. In-depth profiling of the top ranked ADC targets identified epithelial-mesenchymal transition (EMT) and tumor invasion/metastasis to be the most commonly shared signatures among these ADC targets.

Conclusions

ADC target selection is usually based on differential expression between tumor and normal tissue but the tumor immune landscape is not considered. With the success of immunotherapy, our bioinformatics analysis suggests that selection of ADC targets co-expressed with immune inflamed, EMT and tumor invasion/metastasis signatures may be a strategy to further triage candidate ADC targets to increase the likelihood of success for effective combination of ADC and immunotherapies.

Cancer Vaccines

P107 CMV gB/pp65 eVLPs formulated with GM-CSF as a therapeutic vaccine against GBM

Catalina Soare, Tanvir Ahmed, Julie Joseph, Anne-Catherine Fluckiger, Barthelemy Ontsouka, Abebaw Diress, Jasminka Bozic, Marc Kirchmeier, Francisco Diaz-Mitoma, David E. Anderson
VBI Vaccines, Cambridge, MA, USA
Correspondence: David E. Anderson (danderson@vbivaccines.com)

Background

Failures of many past therapeutic cancer vaccines can be attributed to several common shortcomings including the inherently poor immunogenicity of the tumor antigens used, the failure to expand CD4+ T helper cell responses that could sustain potent CD8+ CTLs, and limited breadth of immunity directed against multiple antigens/epitopes that could avoid the rapid immunoselection of tumor escape variants.

Numerous independent laboratories using several techniques (ISH, immunohistochemistry, T cell recognition of autologous primary tumors) have demonstrated cytomegalovirus (CMV) antigens in over 90% of GBM tumors. Memory CD4+ and CD8+ T cells are most frequently directed against the gB and pp65 antigens at frequencies that can exceed 10%. Thus, CMV gB and pp65 represent attractive, highly immunogenic “foreign” antigen components of a vaccine against GBM that can exploit high pre-cursor frequencies of CD4+ and CD8+ T cells.

Methods

Enveloped virus-like particles (eVLPs) are produced after transfection of HEK 293 cells with a plasmid encoding murine leukemia virus Gag plasmid fused in-frame with CMV pp65 antigen, which gives rise to particles. Co-transfected CMV gB plasmid enables particles budding from the cell surface to incorporate the gB protein into the lipid bilayer.

Results

In ex vivo studies using PBMCs from GBM patients we have demonstrated that gB/pp65 eVLPs restimulate both CD4+ and CD8+ T cells at frequencies comparable to those observed for healthy CMV+ subjects, and that formulation with GM-CSF augments IFN-g and CCL3 secretion. Biodistribution studies have demonstrated that the gB/pp65 eVLPs remain at the intradermal injection site for at least two weeks, with eVLP particles appearing in draining lymph nodes within a few hours. Mechanistic studies demonstrate that monocyte uptake of eVLP particles induces proinflammatory cytokines, which are amplified by gB expression on the surface of the gB/pp65 particles. Immunization of CMV+ rhesus macaques with a macaque-specific version of the vaccine candidate confirmed the safety of the vaccine and the ability to boost pre-existing IFN-g-secreting T cell responses.

Conclusions

An IND for a phase I/IIa trial in recurrent GBM patients was filed with FDA in July, with subject enrollment expected in Q4 2017.

P108 Induction of folate receptor alpha-specific Th17 T cell immunity in ovarian cancer patients

Matthew Block1, Kimberly Kalli1, Jacob Allred1, Courtney Erskine1, Allan Dietz1, Michael Gustafson1, Yi Lin1, Keith Knutson2, Martin Cannon3
1Mayo Clinic, Rochester, MN, USA; 2Mayo Clinic, Jacksonville, FL, USA; 3University of Arkansas for Medical Sciences, Little Rock, AR, USA
Correspondence: Matthew Block (block.matthew@mayo.edu)

Background

The nature of the endogenous immune response to ovarian cancer (OC) is a strong predictor of clinical outcomes; regulatory T cells (Tregs) predict poor survival, whereas interleukin 17 (IL-17)-producing T cells (Th17s) are favorably prognostic. We have developed a method by which ex vivo-matured dendritic cells (DCs) can preferentially drive expansion of Th17s [1]. A vaccine utilizing OC-specific Th17-inducing DCs might diminish OC-related Tregs and reduce the rate of OC recurrence.

Methods

Patients with stage IIIC and IV OC were enrolled following completion of initial surgery and chemotherapy. DCs were generated from autologous peripheral blood mononuclear cells (PBMCs) that were cultured according to our Th17-inducing protocol [1] and pulsed with peptides from the OC antigen folate receptor alpha (FRα). Mature antigen-loaded DCs (FRαDCs) were delivered to patients via intradermal injection in a series of five immunizations and up to seven boosters. Patient PBMCs were collected before and after vaccine induction and were assessed via Enzyme-Linked ImmunoSpot (ELISpot) assay for FRα-specific T cells capable of secreting IL-17 or IFNγ. ELISpot responses were defined as at least a doubling of FRα peptide- or protein-specific cytokine-secreting cells.

Results

Nineteen patients were enrolled and immunized with vaccine, with one patient subsequently excluded due to ineligibility. No grade 3 or higher adverse effects were seen. Vaccination resulted in the development of both Th17 and Th1 immunity to vaccine constituent FRα peptides as well as whole FRα protein in the majority of patients. Antibody responses to FRα were also seen following vaccination in some patients.

Conclusions

Specific induction of Th17 immunity against self-antigens in cancer patients is feasible and safe. The results of this study pave the way to future trials aimed at determining if vaccine-induced Th17 immunity prevents disease progression and improves survival.

Trial Registration

ClinicalTrials.gov Identifier NCT02111941

Reference

1. Cannon, Goyne, Stone, Macdonald, James, Cobos, Chiriva-Internati.Modulation of p38 MAPK signaling enhances dendritic cell activation of human CD4+ Th17 responses to ovarian tumor antigen. Cancer Immunol Immunother. 2013;62:839-849.

P109 Anti-NY-ESO-1 immune response and survival benefit after LV305 therapy in patients with advanced sarcoma and other solid tumors

Neeta Somaiah1, Matthew Block2, Joseph W. Kim3, Geoffrey Shapiro4, Khanh Tu Do4, Patrick Hwu1, Joseph Paul Eder3, Robin Lewis Jones5, Sacha Gnjatic6, Hailing Lu7, Jan H. ter Meulen7, Michael Chen9, Chet Bohac8, Frank Hsu8, Seth Pollack10
1MD Anderson Cancer Center, Houston, TX, USA; 2Mayo Clinic, Rochester, MN, USA; 3Yale University, New Haven, CT, USA; 4Dana Farber Cancer Institute, Brookline, MA, USA; 5Royal Marsden NHS, London, UK; 6Tisch Cancer Institute, New York, CA, USA; 7Immune Design, Seattle, WA, USA; 8Immune Design, South San Francisco, CA, USA; 9Biostatistics, South San Francisco, CA, USA; 10Fred Hutchinson Cancer Research Center, Seattle, WA, USA
Correspondence: Seth Pollack (chet.bohac@immunedesign.com)

Background

LV305 is a dendritic cell targeting lentiviral vector encoding NY-ESO-1 designed to generate and expand anti-NY-ESO-1 T cells. This first-in-human study showed LV305 is safe, induced anti-NY-ESO-1 CD4/CD8 T cells, and provided durable disease control with a 1-yr overall survival (OS) of 81% in sarcoma patients (pts)(N. Somaiah, ASCO 2016). We now present immune response (IR) and long term OS data.

Methods

Previously treated, advanced NY-ESO-1+ solid tumor pts were eligible. Part 1 dose escalation, 4 cohorts received 3 or 4 intradermal injections every 3 weeks of 108, 109 or 1010 vector genomes (vg)/dose; 1010 vg/dose in part 2 expansion. Anti-NY-ESO-1 antibody (Ab) and T cell IR measured by ELISA, and ELISpot and T cell receptor sequencing, respectively.

Results

As of March 31, 2017, 39 pts (13 synovial sarcoma, 6 myxoid round cell liposarcoma, 4 other soft tissue sarcoma (STS), 1 Ewing’s sarcoma, 9 ovarian cancer, 5 melanoma, 1 NSCLC) were enrolled, median (m) duration of observation 17.2 months (mos). For the 24 sarcoma pts (mean age 48.1 yrs, 87.5% relapsed metastatic, 100% previously treated), mPFS 4.67 mos and mOS not reached (NR). Exploratory analysis of biomarker data showed: 1) Survival (mos (95% CI)) and presence of baseline anti-NY-ESO-1 Abs: For all pts combined, mOS was NR (17.9, NR) for NY-ESO-1 Ab+ (n=12) vs 27.8 (10.3, 27.8) for NY-ESO-1 Ab- (n=19), HR 0.28 p value 0.09. For STS patients (n=23), mOS was NR (NR, NR) (n=8) vs 18.8 (10.3, 27.8) (n=14), HR 0.0 p value 0.02, respectively. 2) Survival and LV305 induction of IR: Of 31 pts tested, 52% pts developed NY-ESO-1 specific T cells; 3% pts developed anti-NY-ESO-1 Abs after LV305 therapy. 3) Survival and IR: For all pts combined, mOS was 27.8 (18.8, 27.8) for IR+ (n=21) vs 10.3 (4.0, NR) for IR- (n=10), HR 0.32 p value 0.06. Pts who had anti-NY-ESO-1 IR both at baseline and induced had the best survival: mOS was 27.8 (17.9, 27.8) (n=14) vs 8.6 (6.0, NR) in pts who had no evidence of IR at baseline nor on therapy (n=4), HR 0.17 p value 0.018.

Conclusions

LV305 is a well-tolerated therapy which induces anti-NY-ESO-1 IR and may improve OS in sarcoma pts. Pts with baseline anti-NY-ESO-1 Abs or induced IR on LV305 therapy appear to have a better survival.

Trial Registration

ClinicalTrials.gov Identifier: NCT02122861P110

P110 MEK inhibition augments Talimogene laherparepvec (T-VEC)-mediated in vitro melanoma oncolysis and induces in vivo tumor regression

Howard Kaufman, Praveen Bommareddy, Anne Silk, Andrew Zloza, Sachin Jhawar
Rutgers University, New Brunswick, NJ, USA
Correspondence: Howard Kaufman (pkb38@gsbs.rutgers.edu)

Background

Herpes simplex virus, type 1 (HSV-1) encoding GM-CSF (Talimogene laherparepevec; T-VEC) is the first and only FDA approved oncolytic virus for the treatment of melanoma. We recently found that T-VEC induces apoptosis in melanoma cells in vitro, and this effect is enhanced with concurrent MEK inhibition. In this study, we sought to determine how MEK inhibition enhances tumor cell lysis and determine the therapeutic effects of combination treatment in a melanoma xenograft model

Methods

BRAF wild-type and mutated melanoma cell lines were plated in 96-well plates (104 cells per well) and treated with T-VEC (MOI 0.001-1.0). Cell viability was assessed by standard MTS assay. After establishing baseline viability results, cells were also treated with MEK inhibitors (Trametinib or PD0325901 at 1-100 nM) and cell viability determined by MTS assay. Viral replication was measured by plaque assay and the infection metric was analyzed using Lumacyte single-cell optical density. Apoptosis was determined by Annexin V flow cytometry and cleaved PARP immunoblotting. For in vivo experiments, NSG mice were challenged with SKMEL-28 (5x106) at day 0 and treated with intra-tumoral T-VEC (106 PFU) biweekly and/or trametinib (0.1 mg/kg) by oral gavage for 2 weeks. Tumor growth was measured by calipers. Statistical comparisons between treatment groups were determined using the student’s t test with P<0.05 being considered statistically significant.

Results

Combination T-VEC and trametinib significantly increased melanoma cell death (P<0.01). A similar effect was seen with a second MEK inhibitor (PD 0325901) (P<0.001). The combination caused a significant increase in viral replication and increased apoptosis. Combination treatment was associated with morphologic changes in tumor cells and an increased infection metric by Lumacyte. A PCA analysis distinguished virally-infected cells and those treated with both virus and MEK inhibitor from uninfected cells. A significant reduction in tumor volume was also observed in SKMEL-28 NSG xenografts during the combination treatment compared to monotherapy with T-VEC or MEK inhibition (Figs. 1 and 2).

Conclusions

Combination T-VEC and MEK inhibition enhances tumor cell killing and viral propagation in vitro and induces tumor regression in vivo. Combining MEK inhibitors with T-VEC represents an attractive therapeutic option and further studies are needed to understand the impact of this combination on induction of host anti-tumor immunity.

Reference

1. Somaiah, Neeta, et al. American Society of Clinical Oncology (ASCO) Annual Meeting. 2016; Abstract #3093.
Fig. 1 (abstract P110).

MEK Inhibition increases viral titers in T-VEC treated SKMEL28 Melanoma cell line

Fig. 2 (abstract P110).

Trametinib augments T-VEC mediated tumor regression in Xenograft melanoma model. NSG mice with SKMEL28 xenograft treated with Trametinib and/or T-VEC

P111 Initial safety and production efficiency of an autologous tumor lysate-loaded yeast cell wall particle vaccine

Tommy Brown1, Kaitlin Peace1, Myers John1, Timothy Vreeland2, Diane Hale1, Doreen Jackson1, Julia Green1, Jamie Berry3, Clifton Guy1, Herbert Garth1, Xianzhong Yu4, Thomas Wagner5, George Peoples6
1Brooke Army Medical Center, San Antonio, TX, USA; 2University of Texas MD Anderson Cancer Center, Houston, TX, USA; 3Womack Army Medical Center, Fort Bragg, NC, USA; 4Clemson University, Clemson, SC, USA; 5Orbis Health Solutions, Greenville, SC, USA; 6Metis Foundation, San Antonio, TX, USA
Correspondence: Tommy Brown (tommy.alan.brown.ii@gmail.com)

Background

Yeast cell wall particles (YCWP) have been used to create an autologous tumor lysate, particle-loaded, dendritic cell (TLPLDC) vaccine with encouraging initial clinical results. Direct inoculation of an autologous tumor lysate loaded, particle only (TLPO) vaccine eliminates the need for ex vivo DC loading, greatly reducing production requirements compared to TLPLDC. Our preclinical work showed that YCWP can be loaded with autologous tumor lysate (TL) and capped with silicate (Si) to seal in the TL content and attract a monocytic infiltrate. We have shown that silicate capping enhances TL retention, DC phagocytosis and cytoplasmic delivery of the YCWP contents. Here, we present the initial production efficiency and safety data from an exploratory phase I basket trial of the autologous TLPO vaccine.

Methods

YCWP are created from Saccharomyces cerevisiae by NaOH/HCl digestion of all non-cell wall components and washed with isopropanol/acetone resulting in β-glycan shells. Autologous TL is created by freeze/thaw cycling of ≥1mg of tumor and loaded into YCWP along with CpG oligonucleotides and a tetanus helper peptide by dissolution and freeze-drying. The TL-loaded YCWP are then capped with Si, aliquoted in single dose vials and cryopreserved. The TL protein content of production batches is determined by measuring the nitrogen content by combustion analysis, while Si cap thickness is calculated based on Si content by mass spectrometry. The vaccine is administered intradermally at 1.0x108 particles monthly x4 followed by boosters every 3 months until progression.

Results

Nine patients with stage I-IV malignancies(non-small cell lung, pseudomyxoma peritonei, prostate, ovarian, pancreatic, breast, desmoplastic small round cell tumor, esophageal, and renal cell carcinoma) aged 21-85(median 52) have been enrolled and received between 1 to 3 inoculations of TLPO to date. Enrollment is ongoing. Vaccine production has been successful in 100% of patients. In the production runs of the vaccine, the mean TL protein content was 16.56+/-0.56% of each inoculation by weight. The Si cap thickness was 0.0615+/-0.0009 microns (approximately 140 molecular layers of Si). In the 17 inoculations thus far, there has been no toxicity >grade 2. Patients are due for their first set of imaging shortly.

Conclusions

TLPO vaccine is reliably and efficiently produced from autologous tumor tissue with sufficient loading of TL and sufficient Si capping to maintain TL within the YCWP. The vaccine is well tolerated with minimal toxicity thus far. The favorable production characteristics and safety profile combined with improved production efficiency of the TLPO compared to the TLPLDC vaccine warrant continued development of this vaccine platform.

P112 Telomerase peptide vaccine treatment of patients with locally advanced or metastatic non-small cell lung cancer: Report from a phase I/IIA trial

Tormod Kyrre Guren1, Else Marit Inderberg2, Paal Fr. Brunsvig1, Odd Terje Brustugun3, Jon Amund Kyte1, Jens Bjørheim4, Wenche Rasch4, Claudius H. Reisse1, Gustav Gaudernack4, Steinar Aamdal1
1Oslo University Hospital, Oslo, Norway; 2University of Oslo, Oslo, Norway; 3Vestre Viken Hospital Trust, Drammen, Norway; 4Ultimovacs AS, Oslo, Norway
Correspondence: Paal Fr. Brunsvig (pfb@ous-hf.no)

Background

At the initiation of this study in 2013 second and later lines treatment of patients with stage III or IV NSCLC was chemotherapy or radiotherapy with limited median survival.

UV1 is a therapeutic cancer vaccine consisting of three long synthetic peptides of the enzyme telomerase (hTERT). The UV1 peptides comprise epitopes recognized by T cells from cancer patients experiencing long-term survival following vaccination with first-generation hTERT vaccines. The aim was to investigate the safety, induction of immune response (IR) towards the hTERT peptides and efficacy. Three different UV1-doses were investigated.

Methods

Patients with locally advanced or metastatic NSCLC received UV1 (100 μg, 300 μg or 700 μg) and 75 μg GM-CSF as adjuvant [EudraCT No. 2012-001852-20]. All patients had received prior lines of chemotherapy and/or radiation therapy, and had no evidence of disease progression at the time of inclusion. Safety was assessed according to CTCAE v. 4.0 and tumor responses according to RECIST v.1.1. Immune response (IR) against UV1 peptides was monitored by T cell proliferation assays and, if sufficient cell numbers, IFN-γ ELISPOT assays. Patients are to be followed up for five years after their first UV1 treatment. Results as of Q2-2017 are reported.

Results

18 patients (10 females), mean age 65.9 yrs. (48-76), with locally advanced or metastatic (stage III or IV) NSCLC were included. Ten patients had adenocarcinoma and eight had squamous cell carcinoma. Six patients were enrolled in each dose group. Patients received between 9 and 18 doses of UV1 each (mean 12.5 doses, total of 224). Treatment was generally well tolerated. Adverse events mainly included injection site reactions and fatigue. No treatment related serious adverse events were reported.

IR was induced against UV1 in 12 of 18 patients (67%). Fifteen patients were evaluated for tumor response after nine months, 12 patients had SD and three patients had PD. Median overall survival was 23.6 months. After two years, seven patients were alive, two and five in the 300 and 700 μg dose group, respectively.

Conclusions

UV1 is safe and well tolerated and induce an UV1 specific IR in the majority of the patients (67%). UV1 is currently under development in different solid tumors as combination therapy.

Trial Registration

EudraCT No. 2012-001852-20

P113 OncoPeptVAC: A robust TCR binding algorithm to prioritize neoepitope using tumor mutation (DNAseq) and gene expression (RNAseq) data

Papia Chakraborty, Amit Chaudhuri, Vasumathi Kode, Snigdha Majumdar, Priyanka Shah, Malini Manoharan, Ravi Gupta
MedGenome Inc, Foster City, CA, USA
Correspondence: Amit Chaudhuri (papia.c@medgenome.com)

Background

Neoepitopes are tumor-derived immunogenic peptides that arise from intracellular proteolytic processing of somatic mutation in protein coding genes. These peptides bind HLA Class I proteins and are presented on the surface by antigen presenting cells. Productive engagement of HLA Class I-bound peptide with T cell receptor (TCR) activates CD8+ T cells to generate cytotoxic T cells, which mediate lysis of the neoantigen-expressing tumor cells. Neoepitopes can be used as cancer vaccines to prime CD8+ T cells against tumor cells. As tumors accumulate hundreds of mutations during cancer development and only a small subset of these are immunogenic, identifying the neoepitopes requires accurate modeling of the steps involved in peptide production, presentation as well as TCR binding.

Methods

Most current pipelines prioritize neoepitopes based on the expression of mutant proteins and their HLA binding affinity. However, presentation of peptides on the surface is not sufficient to activate T cells. To improve the predictive power of our neoepitope prioritization pipeline and to circumvent selection biases that are inherent with using HLA binding as a proxy for neoepitope prediction, we developed a novel algorithm that prioritizes peptide interactions with the TCR. This algorithm uses features selected by analyzing crystal structures of TCR and HLA-peptide complex present in the Protein Data Bank. A neural network model was built to derive a composite score that includes, besides TCR binding, other features associated with the peptide, such as level of expression of the mutant allele, affinity of HLA binding, affinity of TAP binding and sensitivity to proteasomal processing.

Results

We validated our neural net prediction model on known immunogenic and non-immunogenic peptides and achieved superior accuracy, sensitivity and specificity of prediction compared to using the standard HLA binding affinity of ≤500 nM. By applying OncoPeptVAC neoepitope prioritization solution to 2.2 million unique somatic mutations, we identified ~700 immunogenic peptides derived from recurrent somatic mutations in all cancers. Several of these peptides were validated on a CD8+ T cell-activation assay.

Conclusions

Our in silico prioritization platform combined with the cell-based validation method is a powerful tool to identify therapeutic vaccines for personalized cancer immunotherapy applications.

P114 No toxicity and long-term survival up to 16-years from repeated oncolytic vaccine systemic immunotherapy in advanced melanoma: Teaching us lessons

Brendon Coventry1, Carrie Lilly1, Tony Michele2, Peter Hersey3
1University of Adelaide, Adelaide, Australia; 2Adelaide Oncology & Haematology, North Adelaide, South Australia, Australia; 3University of Sydney, Sydney, Australia
Correspondence: Brendon Coventry (brendon.coventry@adelaide.edu.au)

Background

We have previously reported very long-term complete responses with survival of ≥5 years to oncolytic melanoma vaccine therapy for patients with advanced melanoma. The Complete Response (CR) Rate was also higher than expected at 17% predominantly from therapy with vaccine alone [1]. Importantly, side-effects were negligible with no noticeable toxicity.

Methods

Extended analysis of patient survival was performed to investigate the presence of recurrence, late failure, morbidity, mortality and the longevity of responders.

Standard methods of outcome evaluation (CR, PR, SD, PD and overall survival) including Kaplan-Meier estimates, were performed.

Results

Clinical responses have continued to beyond 10-years with survival periods of now over 15-years in duration for some patients. These responders are being studied in more detail. CR's when obtained were durable in almost all cases.

Conclusions

Clinical responses and long-term survival using vaccine therapy alone is unusual, but continued to beyond 10-years with survival periods of now over 15-years is obtainable and this requires more investigation as a matter of priority. These responders are being studied in more detail. Some 84% of patients showed some clinically useful response, and no toxicity. These long-term survival [2] and high CR rates results were not from chance alone and were attributable to the vaccine administration & scheduling most likely creating successful in-vivo immunomodulation.

Trial Registration

Australian Clinical Trials Registry [ACTRN] 12605000425695

References

1. Responses and survival for advanced metastatic melanoma. J Immunother Cancer. 2014;2: 9-17.

2. Coventry, Baum, Lilly. Long-term survival in advanced melanoma patients using repeated therapies: successive immunomodulation improving the odds? Cancer Manag. Res. 2015;7:93-103.

P115 Inflammatory response after immunotherapy with a yeast-CEA therapeutic cancer vaccine in metastatic medullary thyroid cancer

Jaydira Del Rivero1, Ann Gramza2, Myrna Rausckhorts1, Marijo Bilusic1, Fatima Karzai1, Julius Strauss1, Lisa Cordes1, William L. Dahut1, Jeffrey Schlom1, James L. Gulley1, Ravi A. Madan1
1National Cancer Institute/National Institutes of Health, Bethesda, MD, USA; 2Medstar Georgetown Lombardi Comprehensive Cancer Center, Washington, DC, USA
Correspondence: Jaydira Del Rivero (jaydira.delrivero@nih.gov); Ravi A. Madan

Background

Medullary thyroid cancer (MTC) accounts for approximately 4% of thyroid carcinomas. Although vandetanib and cabozantinib are approved for metastatic MTC, toxicity limits their use. There are ongoing trials at the NCI exploring the role of immunotherapy in MTC patients.

Methods

In a series of two clinical trials (including a phase II MTC study; NCT01856920), 30 MTC patients have been treated with a yeast-based vaccine targeting CEA (Bilusic M et al., Immunol Immunother, 2014). The vaccine was scheduled every 2 weeks for 3 months; then monthly for 9 months and quarterly in the second year for patients without radiographic disease progression.

Results

3/30 (10%) of patients developed tumor site-specific inflammatory responses which are not typically seen in MTC. A 39-year-old woman with sporadic metastatic MTC and stable disease on restaging CT after 3 months of vaccine, developed dyspnea one week later. CT scan showed pleural and paricardial effusions at sites of known disease. After a work-up yielded no diagnosis, empiric steroids resolved symptoms in 48 hours and a follow-up CT-scan demonstrated resolution of the effusions. This patient from the phase I study was found to have robust antigen-specific T cell responses. Similarly, a 72-year-old woman with sporadic metastatic MTC presented with an enlarging neck mass at a known site of metastatic disease after 21 months of vaccine. Due to difficulty swallowing, she received high-dose steroids and empiric antibiotics at an outside hospital with resolution of her symptoms. A follow-up CT reported a decreased in size of the neck mass. Perhaps most interestingly, a 66-year-old woman with MEN2A and metastatic MTC returned to our clinic 17 months after completing the 2-year vaccine course with a rapidly enlarging lymph node at a known site of metastatic disease confirmed by CT. A biopsy of the lymph node was negative for tumor but demonstrated an abundance of CD3+ and CD20+ lymphocytes. The symptoms and lymph node size decreased within 1 week as did the patient’s calcitonin in the following weeks.

Conclusions

We present 3 patients with metastatic MTC who presented with an inflammatory event after treatment with a yeast-CEA therapeutic cancer vaccine. These unexpected inflammatory events may demonstrate the potential for immunotherapy in MTC. A follow-up up phase II is ongoing at the NCI treating patients (both with previous immunotherapy and immunotherapy-naïve) with pembrolizumab (NCT03072160) and analysis of the vaccine study is underway. Informed consent to publish had been obtained from all of the patients.

P116 Heat-inactivated modified vaccinia virus Ankara (MVA) as vaccine adjuvant

Ning Yang, Stewart Shuman, Taha Merghoub, Jedd Wolchok, Liang Deng
Memorial Sloan Kettering Cancer Center, New York, NY, USA
Correspondence: Liang Deng (dengl@mskcc.org)

Background

Recent discoveries of cancer neoantigens and the potential of the combination of cancer vaccination with immune checkpoint blockade to enhance vaccination effects have generated excitement and renewed interest. Developing novel and effective vaccine adjuvants that can enhance antitumor immune responses is critical for the success of cancer vaccine. We have recently shown that intratumoral injection of Heat-inactivated MVA (Heat-iMVA; by heating MVA at 55°C for 1 h) generates systemic antitumor effects via a STING-dependent mechanism and requires Batf3-dependent CD103+/CD8α+ dendritic cells (DCs). Heat-iMVA infection of conventional DCs (cDCs) induces higher levels of type I IFN and proinflammatory cytokines and chemokines than live MVA. It also induces DC maturation in a STING-dependent manner. The role of Heat-iMVA as a vaccine adjuvant has not been explored previously.

Methods

To test whether Heat-iMVA is an effective vaccine adjuvant, we first tested whether co-administration of chicken ovalbumin (OVA) with Heat-iMVA improves anti-OVA T cell and antibody responses. We then tested whether co-administration of irradiated OVA-expressing B16 cells with Heat-iMVA delays tumor growth in a therapeutic vaccination model. We also tested whether systemic delivery of anti-PD-L1 antibody would further improve vaccination efficacy. Mice were intradermally implanted with B16-OVA, they were vaccinated with either irradiated B16-OVA, or B16- OVA + Heat-iMVA three times on the contralateral flank in the presence of absence of anti-PD-L1 antibody. We measured tumor volumes and monitored survival of the mice.

Results

We found that addition of Heat-iMVA increased the percentage of OVA-specific CD8+ T cells and CD4+ T cells in the draining lymph nodes and spleens. The induction of OVA-specific CD8+ T cells was significantly reduced in Batf3-/- mice. It also boosted the OVA-specific IgG2c and IgG1 in the serum. In the therapeutic vaccination model, we found that vaccination with irradiated B16-OVA + Heat-iMVA extended the median survival from 16 days (with irradiated B16-OVA vaccination) to 23 days (**, P<0.01). With anti-PD-L1 antibody, vaccination with irradiated B16-OVA + Heat- iMVA extended the median survival from 20 days to 27 days (**, P<0.01). In vitro studies demonstrated that Heat-iMVA enhanced antigen cross-presentation by DCs.

Conclusions

Our results demonstrate that Heat-iMVA is a safe and potent vaccine adjuvant for peptide and irradiated whole cell vaccinations. Future studies will evaluate the role of Heat-iMVA as a vaccine adjuvant for tumor neoantigens.

P117 DNA-based cancer vaccines designed by SynCon® technology break tolerance in genetically diverse pre-clinical models

Bradley Garman1, Elizabeth K Duperret2, Bernadette Ferraro1, Jewell N Walters1, Kimberly A Kraynyak1, Emma Masteller1, Jian Yan1, Amir S Khan1, David B Weiner2, Laurent M Humeau1
1Inovio Pharmaceuticals, Inc., Plymouth Meeting, PA, USA; 2The Wistar Institute, Philadelphia, PA, USA
Correspondence: Laurent M Humeau (bradley.garman@inovio.com)

Background

Cancer vaccines targeting self-antigens have shown limited efficacy in the clinic due to self-tolerance mechanisms preventing or dampening an immune response against them. To overcome self-tolerance, we developed a novel, multi-phase DNA vaccine design strategy (SynCon® technology) that introduces subtle amino acid changes into native cancer antigens. These synthetic immunogens aim to generate cross-reactive T cells and/or promote epitope spreading. As a proof of concept, we first used SynCon® technology to design a novel DNA immunogen targeting the self-antigen FAP (SynCon® FAP) and investigated its ability to break tolerance in outbred CD-1 mice. Compared to commonly used inbred mouse models, CD-1 mice are more relevant for inferring immunogenicity in humans due to their genetic diversity. In addition, we designed a SynCon® immunogen targeting TERT (SynCon® TERT) and tested its ability to break tolerance in non-human primates (NHP), a highly relevant model for immunotherapeutic vaccine development.

Methods

CD-1 mice were immunized with either SynCon® FAP, a mouse native FAP vaccine, or vector control. Three immunizations were administered two weeks apart with electroporation following each immunization. For the NHP study, rhesus macaques were immunized with SynCon® TERT or native rhesus TERT vaccine as a control. Four immunizations were administered four weeks apart with electroporation following each immunization. Vaccine-induced immune responses were evaluated by IFN-γ ELISpot using species-specific native peptides.

Results

Compared to native mouse FAP, SynCon® FAP was more immunogenic in CD-1 mice (average 407 ELISpot-forming units (SFUs) per 106 splenocytes for SynCon® vs. 160 SFU/106 for native), suggesting that SynCon® FAP was capable of breaking tolerance. Importantly, 14/15 mice in the SynCon® FAP group generated an immune response above 100 SFU/106, compared to only 9/15 mice in the native FAP group. In rhesus macaques, while the native rhesus TERT was not able to induce any detectable immune responses following the fourth immunization, SynCon® TERT elicited an average of 353 SFU/106, indicating that the SynCon® TERT immunogen was capable of breaking tolerance in NHPs.

Conclusions

Taken together, we demonstrate that SynCon® technology can be used to design DNA immunogens that are capable of breaking tolerance in genetically diverse preclinical models. SynCon® DNA immunogens have the potential to break tolerance, induce immune responses, and provide clinical benefit for patients with diverse HLA haplotypes. Further investigation of SynCon® DNA immunogens in clinical trials is warranted.

P118 Gp96-Ig/Costimulator combination vaccine improves T cell priming, enhances immunity, memory, and tumor elimination

Louis Gonzalez, Louise Giffin, Jason Rose, George Fromm, Suresh De Silva, Taylor Schreiber, Jeff Hutchins
Heat Biologics, Durham, NC, USA
Correspondence: Jeff Hutchins (lgonzalez@heatbio.com)

Background

The recent excitement in the field of immuno-oncology has been driven largely by the clinical success of checkpoint inhibitors. This success is tempered by the fact that monotherapy succeeds in only 10-40% of patients. It is widely believed that to improve patient outcomes, new approaches that combine treatments with more than one functionality will be required.

Methods

We have developed a next generation cellular vaccine platform – referred to as ComPACT (COMbination Pan-Antigen Cytotoxic Therapy)that incorporates a tumor antigen chaperone (gp96-Ig) with T cell costimulation (Fc-OX40L), into a single anti-tumor cell line that secretes them both (Cancer Immunol Res. 2016 Sep 2; 4(9):766-78).

Results

ComPACT primes both antigen-specific CD4+ and CD8+ T cells, and stimulates activation of CD127+KLRG-1- memory precursor cells. Systemic administration of OX40 agonist antibodies led to toxic proliferation of non-specific CD4+ T cells, Tregs and systemic inflammatory cytokine production. Importantly, ComPACT led to high frequencies of IFNγ+, TNFα+, granzyme-b+ and IL-2+ antigen-specific CD8+ T cells at both priming and boosting, which enhanced rejection of established murine melanoma (B16.F10) and colon cancer (CT26) tumors and increased overall survival.

Conclusions

Here, we have assessed ComPACT in combination with a checkpoint inhibitor and an additional systemic administered T cell costimulator (TL1A) and show that they synergize effectively with antagonist antibody therapies, amplifying antigen-specific T cells, programming a memory response, and eliminating tumors. ComPACT/αPD1 or αPD-L1/TL1a combinations may therefore translate into an efficacious approach to treat human cancers.

P119 Optimization of antigenic composition of a dendritic-cell targeting MIP3α-antigen vaccine and the efficacy of additional IL-10 neutralization therapy in the B16F10 mouse melanoma model

James Gordy, Richard Markham
Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA
Correspondence: Richard Markham (jgordy2@jhu.edu)

Background

The chemokine MIP3α (CCL20) binds to CCR6 on immature dendritic cells. Vaccines fusing MIP3α to melanoma antigen gp100 have been shown to be moderately effective in therapeutically alleviating melanoma in mouse models. To optimize the therapy, our laboratory is exploring two avenues. First, we added agents to neutralize interleukin 10 (IL-10) at the tumor site (i.t.) or vaccination site to modulate immune responses. Second, we examined the effects of substitution and addition of the highly immunogenic tyrosinase-related protein 2 (TRP2) antigen to the MIP3α-gp100 vaccine construct.

Methods

The current studies utilize the B16F10 syngeneic, transplantable, mouse melanoma model system. The MIP3α-antigen DNA vaccine is administered intramuscularly (i.m.) into the tibialis muscle, followed immediately by i.m. electroporation. Constructs utilized include MIP3α fused to gp100, TRP2, or both. Vaccinations are given therapeutically, beginning at day 3 or 5 post challenge. Tumor sizes, growth, and survival were all assessed. Treatment responses were characterized by flow cytometric analysis of tumor infiltrate. The mechanism of i.t. αIL-10 efficacy was explored by RT-PCR and confirmed with a knockout mouse model.

Results

With this therapeutic protocol, we demonstrate for the first time that neutralizing IL-10 at either the vaccination or tumor site enhances the anti-tumor efficacy of a MIP3α-gp100 vaccine, leading to significantly smaller tumors, slower growing tumors, and increases in mouse survival. We discovered that IFNα-4 transcripts in the tumor were significantly upregulated in mice given vaccine and i.t. αIL-10 compared to vaccine alone. A mouse model with IFNαR1 knocked out eliminated the protection provided by i.t. αIL-10, demonstrating that the additional therapeutic value of i.t. αIL-10 is primarily mediated by type-I interferons. Also, we demonstrate that the MIP3α-gp100-TRP2 dual antigen and the MIP3α-TRP2 vaccine provided better protection compared to the MIP3α-gp100 vaccine but were not significantly different from each other.

Conclusions

Efficient targeting of antigen to immature dendritic cells with a chemokine fusion vaccine offers a potential alternative approach to the ex vivo dendritic cell antigen loading protocols currently undergoing clinical investigation. Neutralizing IL-10 at both the vaccination site and tumor site have enhanced vaccine efficacy, the latter shown to be due to type-I interferons. Additionally, MIP3α-TRP2 has proven to be a superior vaccine construct than MIP3α-gp100 in our system, and the MIP3α-gp100-TRP2 vaccine did not improve on the protection provided by MIP3α-TRP2. Further potential therapy optimization currently undergoing investigation offers promise for this line of investigation to become a novel melanoma therapy.

P120 Streamlining approaches for the selection of T cell neo-epitopes: from mutanome to therapeutic personalized cancer vaccine

Guilhem Richard1, Leonard Moise1,2, Matthew Ardito1, Frances Terry1, Gad Berdugo3, William Martin1, Anne De Groot1,2
1EpiVax Inc., Providence, RI, USA; 2University of Rhode Island, Providence, RI, USA; 3EpiVax Oncology LLC, Providence, RI, USA
Correspondence: Anne De Groot (jhealey@epivax.com)

Background

Next-generation sequencing has opened the door to precision cancer therapies targeting mutations expressed by tumor cells. However, most neo-epitopes selected by traditional T cell epitope prediction algorithms prove to be non-immunogenic. Poor predictive performance may partially be due to inclusion of mutated epitopes cross-conserved with self-epitopes recognized by the T cell receptor of regulatory (Treg), anergic or deleted T cells. Vaccination with self-epitopes can lead to weak effector responses, active immune suppression, and toxicity due to immune-mediated adverse effects.

Methods

We have developed Ancer, an advanced cancer T cell epitope identification and characterization tool that streamlines the selection of Class I and II neo-epitopes. Ancer is leveraging EpiMatrix and JanusMatrix, state-of-the-art tools that have been extensively validated in prospective vaccine studies for infectious diseases [1, 2]. Use of Ancer in the oncology field allows for the prioritization of neo-epitopes exhibiting reduced potential for inducing Tregs, whose activation continues to curtail efficacy of current cancer therapies.

Results

We validated Ancer’s predictive accuracy using datasets of HLA-bound peptides detected by mass spectrometry, which are independent of training sequence data used in model development. Analysis of sequences from Abelin et al. [3] shows a 96% agreement between Ancer predictions and peptides eluted from common Class I HLAs, while only 86% of these sequences are accurately recalled by NetMHC and NetMHCpan. Additional retrospective analyzes of a cancer immunogenicity study [4] demonstrate that Ancer selects immunogenic neo-epitopes with 72% accuracy, as compared to 21% accuracy when using public prediction tools.

Conclusions

These results demonstrate that Ancer may focus epitope candidate selection on higher value sequences than conventional algorithms. Neo-epitopes with low Treg activation potential may then be used to support the development of safer and more effective vaccines.

References

1. Moise, et al. iVAX: An integrated toolkit for the selection and optimization of antigens and the design of epitope-driven vaccines. Hum. Vaccines Immunother. 2015; 11, 2312–2321.

2. Wada, et al. A humanized mouse model identifies key amino acids for low immunogenicity of H7N9 vaccines. Sci. Rep. 2017; 7, 1283

3. Abelin, et al. Mass Spectrometry Profiling of HLA-Associated Peptidomes in Mono-allelic Cells Enables More Accurate Epitope Prediction Resource Mass Spectrometry Profiling of HLA-Associated Peptidomes in Mono-allelic Cells Enables More Accurate Epitope Prediction. Immunity 2017; 46, 315–326.

4. Strønen, et al. Targeting of cancer neoantigens with donor-derived T cell receptor repertoires. Science. 2016; 352, 1337–1341.

P121 Phase 1 dose escalation trial of CV301

Julius Strauss1, Jaydira Del Rivero1, Ravi A. Madan1, Jennifer Marte1, Alanvin Orpia2, Eva Wagner3, Cesar Pico Navarro4, Christopher Heery4, Jeffrey Schlom1, James Gulley1
1Center for Cancer Research, National Cancer Institute, Bethesda, MD, USA; 2Leidos Biomedical Research, Inc., Bethesda, MD, USA; 3Bavarian Nordic, Martinsried, Germany; 4Bavarian Nordic, Morrisville, NC, USA
Correspondence: James Gulley (chhe@bavarian-nordic.com)

Background

CV301 is a poxviral-based vaccine comprising a prime‑boost strategy with Modified Vaccinia Ankara (MVA, prime) and fowlpox (FPV, boost), which contain transgenes encoding two tumor associated antigens, CEA and MUC-1, as well as 3 costimulatory molecules (B7.1, ICAM-1 and LFA-3, called TRICOM). Preclinical evidence suggests CV-301 activates CEA and MUC-1 specific T cell responses, potentially increasing the clinical benefit of immune checkpoint inhibition. An open label phase 1/1b/2 trial was initiated to evaluate safety of CV301 alone (phase 1) and in combination with PD-1/L1 inhibition (phase 1b) to be followed by a randomized phase 2 trial evaluating the combination compared with PD-1/L1 inhibition alone. Here, we report the phase 1 safety evaluation of CV301.

Methods

This open-label, 3 +3 design dose-escalation phase 1 trial evaluated 3 dose levels of the priming dose of CV301 (MVA): 1, 2 or 4 s.c. injections of 4 x 108 Inf.U/0.5 mL on days 1 and 29. FPV booster doses were scheduled for all subjects subcutaneously at a 1 × 109 Inf.U/0.5 mL dose on weeks 9, 11, 13, 15, 19, 23, 27, 31, 35, 39, 43, 47, 51, 65, 78, 91 and 104.

Results

There were no dose limiting toxicities in the phase 1 evaluation of CV301 alone. The minimum required patient number was enrolled in each dose level (DL1: 3, DL2: 3, DL3: 6, total: 12). The most frequent treatment-related AEs were temporary and self-limiting, grade 1/2, and included injection site reactions (erythema, pruritus, pain, induration and/or swelling) and general symptoms (fever/chills, flu-like symptoms, headache, fatigue/weakness, nausea/vomiting, myalgia and arthralgia). Seven of the 12 patients have microsatellite stable colon cancer. Of these seven, two had preliminary evidence of benefit. One patient with stable disease at first restaging has marked reduction (60% fall) in tumor markers at 10 weeks. Another patient had unconfirmed partial response at first restaging (6 weeks). When 3 month restaging scans indicated progression, a PD-L1 inhibitor was initiated. Subsequently, tumor markers fell by 60%.

Conclusions

CV301 priming doses (MVA) can be safely administered to at least 4 injection sites concurrently at 4 x 108 Inf.U/0.5 mL per site for a total of 1.6 x 109 Inf.U, which is the recommended phase 1b and phase 2 dose. Booster doses (FPV) can safely be administered s.c. at 1 x 109 Inf.U/0.5mL. Subsequent analyzes will include immune response data. The trial will continue to enroll phase 1b and, subsequently, the randomized phase 2.

Trial Registration

ClinicalTrials.gov Identifier NCT02840994

P122 N-terminally extended proline residues influences the HLA class I antigen processing and decreases CTL responses

Ayumi Hongo, Takayuki Kanaseki, Serina Tokita, Vitaly Kochin, Toshihiko Torigoe
Sapporo Medical University, Sapporo, Japan
Correspondence: Takayuki Kanaseki (arkpon1717@gmail.com); Toshihiko Torigoe

Background

CD8+ T cells recognize peptide-MHC class I complexes presented on target cell surfaces; however, the precise mechanisms of antigen processing by which T cell epitopes are generated still remains unclear.

Methods

Here we extensively collected >800 naturally presented peptide sequences, and statistically analyzed an amino-acid profile of their N-terminal extensions. We combined natural HLA-A24 ligand sequences gained from HLA-ligandome analysis with their corresponding N-extension sequences gained from database search, and analyzed frequency of each amino acid at given positions of N-extensions (Position -15 to -1).

Results

Interestingly, we found that frequency of proline residue (Pro) was significantly reduced between P-3 and P-1, implying that peptide sequences following Pro might not be efficiently presented by HLA-A24. To further investigate this hypothesis, we prepared 293T cells expressing a panel of mini-gene constructs that carried N-extensions followed by a model epitope sequence, and ultimately found that presence of Pro at P-3 to P-1 significantly reduced IFN-g production from the CTLs specific to the epitope.

Conclusions

Thus, we consider that presence of Pro could be a new signature in HLA class I antigen processing that attenuates presentation and CTL recognition of the following epitope sequences.

P123 A novel oncolytic virus expressing multiple immune regulatory factors to change tumor immune microenvironment and to cause lasting complete tumor eradication in animal models

William Jia1, JUN Ding2, Erica Lee2, Dmitry Chouljenko2, Guoyu Liu2, Yanal Murad2, Will Liu2, Zahid Delwar