Skip to content

Advertisement

  • Meeting abstracts
  • Open Access

33rd Annual Meeting & Pre-Conference Programs of the Society for Immunotherapy of Cancer (SITC 2018)

Washington, D.C., USA. 7-11 November 2018
Journal for ImmunoTherapy of Cancer20186 (Suppl 1) :114

https://doi.org/10.1186/s40425-018-0422-y

  • Published:

Poster Presentations

(ADCC/CDC)

P1 The identification of potent anti-tumor antibodies for ADC therapeutics from patients undergoing immunotherapy

Alexander Scholz, PhD1, Jerald Aurellano1, Michael Harbell, MS PhD1, Danhui Zhang, MD PhD1, Samantha O'Connor1, May Sumi, BS1, Beatriz Millare, BS1, Felix Chu, MS1, Sheila Fernandez1, Cathrin Czupalla1, Iraz Aydin, PhD1, Amy Manning-Bog, PhD1, Yvonne Leung, BS, PhD1, Kevin Williamson, BS PhD1, Chantia Carroll1, Dongkyoon Kim, BS PhD1, Xiaomu Chen, MS PhD1, Sean Carroll, BS, PhD1, Ish Dhawan, PhD1, Ngan Nguyen, BS PhD1, Shweta Thyagarajan1, Mark Whidden1, Gregg Espiritu Santo, BS PhD1, Nicole Haaser, MS1, Hibah Mahmood1, Guy Cavet, PhD1, Lawrence Steinman, MD2, Tito Serafini, PhD1, Wayne Volkmuth, BS PhD1, Jonathan Benjamin, MD, PhD1, William Robinson, MD2, Norman Greenberg, PhD1, Daniel Emerling, PhD1, Jell DeFalco1
1Atreca Inc, Redwood City, CA, USA; 2Stanford University School of Medicine, Stanford, CA, USA
Correspondence: Daniel Emerling (d.emerling@atreca.com)

Background

Anti-tumor therapy with antibody-drug conjugates (ADCs) is predicated on the identification of antibodies that demonstrate suitable selectivity for tumor cells that are also internalized upon binding their cognate target. Remarkably, only a select number of such antibodies with the propensity to internalize have been identified, limiting the range and breadth of ADC therapeutics in the clinic. Here we show that Atreca’s Immune Repertoire Capture (IRC™) technology can identify potent anti-tumor antibodies with internalization activity applicable for ADC therapeutics from patients undergoing immunotherapy.

Methods

We analyzed blood plasmablasts from patients with non-progressing metastatic cancer using IRC™ technology. Briefly, plasmablasts were collected from patients and paired heavy and light chain antibody sequences were then obtained from individual cells. Antibody sequences representing expanded clonal families were subsequently expressed and analyzed for their ability to (i) bind to human tumor and non-tumor tissues and (ii) internalize into cancer cells when labeled with a pH-sensitive dye. Those antibodies with a high internalization rate were directly conjugated with a cytotoxic agent (auristatin MMAE) and tested in an in vitro ADC assay.

Results

Patient-derived antibodies from several cancer types bound to human tumor tissue but not adjacent normal tissue and also internalized into A549 lung tumor cells. These internalizing antibodies were able to induce target cell death in vitro when conjugated directly or indirectly to a cytotoxic agent across several human tumor cell lines.

Conclusions

In this study we demonstrate that patient-derived antibodies which bind to public tumor-selective antigens and internalize into cancer cells can be identified by our IRC™ technology. Furthermore, we demonstrate that these antibodies can deliver a cytotoxic payload to target tumor cells to induce cell death.

Ethics Approval

The study was approved by Sutter Health Institutional Review Board, approval #2016.148-1

P2 Intratumoral application of hu14.18-IL2 for treatment of GD2+ pediatric malignancies: A novel immunotherapeutic approach aiming at in-situ vaccination

Romana Gugenberger, PhD1, Zachary Morris, MD, PhD2, Oliver Mutschlechner1, Paul Sondel, MD, PhD2, Hans Loibner, PhD1
1Apeiron Biologics AG, Vienna, Austria; 2University of Wisconsin, Madison, WI, USA
Correspondence: Hans Loibner (hans.loibner@apeiron-biologics.com)

Background

hu14.18-IL2 is an antibody-cytokine fusion protein that combines targeting and immune activation of a human IgG1 monoclonal antibody with the immune stimulatory function of IL2. The humanized antibody portion targets the GD2 ganglioside antigen expressed on a variety of tumors of neuroectodermal origin. Clinical efficacy of the immunocytokine by i.v. application has been shown already in several clinical trials in melanoma and neuroblastoma. Dose limiting toxicity relates to systemic IL2 toxicity. A novel approach was explored preclinically in murine tumor models to deliver hu14.18-IL2 locally by intratumoral (IT) injection aiming at induction of a systemic immune response (in-situ vaccination). We present here activity of the immunocytokine in vitro against various GD2 positive pediatric tumor cell lines. We also discuss a humanized mouse model based on patient-derived xenografts (PDX) by directly transplanting surgical material. Finally we will present the design of a clinical trial to explore safety and clinical activity of IT hu14.18-IL2 in patients with GD2+ pediatric malignancies.

Methods

Expression of the target antigen GD2 on human cell lines MG63 (osteosarcoma), TC-71 (Ewing’s sarcoma), RH41 (rhabdomyosarcoma) and Y79 (retinoblastoma) was analyzed by flow cytometry. Hu14.18-IL2 mediated ADCC and whole blood cytotoxicity (WBT) was determined by 51Cr release assays.

Results

We found expression of antigen GD2 on all cell lines derived from neuro-ectodermal pediatric malignancies. Hu14.18-IL2 was effective in mediating ADCC and WBT against all cell lines in vitro, and potency was found higher than that of the unconjugated chimeric anti-GD2 antibody ch14.18/CHO in osteosarcoma and retinoblastoma. The effects were antigen specific as addition of an anti-idiotypic antibody abrogated the cytolytic activity. A humanized mouse model (CD34+ cell engraftment and transplantation of patient derived GD2+ sarcoma tissue) with intra-tumoral application of the immunocytokine is presently set up.

Conclusions

Immunocytokine hu14.18-IL2 is effective in vitro against various GD2 positive pediatric malignancies by activation of both antibody and IL2 effector functions. Humanized mouse tumor models with GD2+ patient derived tumors may be useful to explore IT immunocytokine in vivo. A clinical phase I/II trial in several advanced pediatric GD2 positive tumors (mostly sarcomas; “basket study”) is in preparation with repeated IT administration of low doses of hu14.18-IL2 (in-situ vaccination).

P3 Evaluating antibody-mediated cellular cytotoxicity and potency of antibody-drug conjugates within three- dimensional tumor models

Chris Langsdorf, BS, Bhaskar Mandavilli, PhD, Yi-Zhen Hu, Aimei Chen, Bachelor of Science, Marcy Wickett
ThermoFisher Scientific, Eugene, OR, USA
Correspondence: Chris Langsdorf (chris.langsdorf@thermofisher.com)

Background

Three dimensional tumor spheroids provide biochemical conditions that closely resemble the tumor microenvironment in an intact organism. Noninvasive approaches such as fluorescence microscopy are highly advantageous as they allow for the study of these three dimensional systems. Here we investigate the penetration and potency of natural killer cells, cytotoxic T cells, and antibody-drug conjugates in three-dimensional models of breast and lung cancer.

Methods

Tumor spheroids were formed by incubating cancer cell lines overnight in Nunclon Sphera 96-well plates. Natural killer cells were isolated from human PBMCs using negative magnetic selection and expanded in culture for 16 days. Natural killer cells were added to SKBR3 breast cancer spheroids with or without trastuzumab. T cells were isolated from human PBMCs using negative magnetic selection and activated for 72 hours. Activated or resting T cells were added to lung cancer spheroids. Immune cell penetration and tumor cytotoxicity were evaluated using whole-spheroid imaging on a confocal high-content imaging system. Trastuzumab was site-specifically conjugated with monomethyl auristatin E (MMAE) and iFL pHrodo Red via SiteClick conjugation. Spheroids of HER2+ breast cancer cells were treated 48 hours with this antibody drug conjugate. ADC penetration and apoptosis were evaluated using confocal high-content imaging.

Results

Unstimulated T cells produced minimal cytotoxicity, similar to untreated spheroids. Activated T cells penetrated and produced significant cytotoxicity throughout cancer spheroids. SKBR3 breast cancer cells form a compact, viable spheroid. Addition of NK cells leads to moderate cytotoxicity, while addition of NK cells and trastuzumab results in substantial cytotoxicity and degradation of spheroid structure (Fig. 1). Trastuzumab labeled with iFL pHrodo Red becomes brightly fluorescent following specific endosomal internalization into breast cancer cells, but minimal toxicity is observed. Trastuzumab conjugated with both iFL pHrodo Red and MMAE internalizes into cells and results in cell killing (Fig. 2).

Conclusions

Fluorescence microscopy combined with novel cell and antibody labeling methods permits investigation of the penetration and potency of natural killer cells, cytotoxic T cells, and antibody-drug conjugates in three-dimensional solid tumor models.
Fig. 1 (abstract P3).
Fig. 1 (abstract P3).

See text for description.

Fig. 2 (abstract P3).
Fig. 2 (abstract P3).

See text for description.

P4 PBD-based anti-MICA/B antibody drug conjugate with a dual mechanism of action: direct tumor cell killing and restoration of NKG2D-mediated immunosurveillance

Florence LHOSPICE, Pharm D1, Laurent Pouyet, PhD2, Ester Morgado2, Romain Remark, PhD1, Delphine Bregeon1, Adeline Montbel1, Nadia Anceriz1, Mathieu Blery, PhD1, Ariane Morel, PhD1, Manel Kraiem1, Kenneth Crook1, Eric Vivier1, Yannis Morel, PhD1
1Innate Pharma, Marseille, France; 2MI-mAbs, Marseille, France
Eric Vivier (eric.vivier@innate-pharma.fr)

Background

MICA and MICB can be expressed at the surface of a wide variety of tumor cells upon stress, while having a very limited expression on healthy tissues. This makes MICA/B promising targets for the development of antibody drug conjugates (ADC). In addition, MICA and MICB serve as ligands for NKG2D, a potent activating receptor expressed on NK, CD8+T and γδ T cells. As a consequence, the expression of MICA and MICB promotes recognition and elimination of tumors by these lymphocytes through NKG2D engagement. However, in vitro and in vivo studies have reported that chronic engagement of NKG2D by its ligands induces NKG2D downregulation and lymphocyte dysfunction, leading to compromised immunity. We thus aimed to generate an ADC targeting MICA/B- expressing tumors with a dual function to achieve optimal therapeutic benefits: (i) killing of tumor cells and (ii) disrupting the interaction between MICA/B and NKG2D that induces impaired immunosurveillance [1] [2].

Methods

Antibodies were screened based on their binding affinity for the most frequent MICA/B alleles, as well as for their internalization, cytotoxicity and immunomodulatory properties. The MICA/MICB cross-reactive, pan-allele antibody with the highest affinity was conjugated to valine-alanine-pyrrolobenzodiazepine (PBD) dimers using bacterial transglutaminase-based site-specific conjugation to generate anti-MICA/B-PBD ADC

Results

Anti-MICA/B-PBD showed potent in vitro cytotoxicity against a variety of solid cancer cell lines as well as in vivo efficacy in both HCT116 human colon carcinoma and breast cancer patient-derived xenograft models. The immunomodulatory properties of anti-MICA/B-PBD were assessed in MICA-transgenic mice engrafted with MICA- expressing mouse melanoma B16-F10. Lastly, cell surface MICA/B expression and soluble form concentration were assessed in a large panel of samples from healthy donors and patients with various cancers in order to determine potential therapeutic indications for clinical development.

Conclusions

MICA/B molecules are attractive targets for an ADC approach based on their selective expression in a wide range of malignancies while showing restricted expression in healthy tissues along with manageable concentrations of their soluble form. The anti-MICA/B-PBD shows efficacy both in vitro and in vivo, paving the path for further evaluation towards clinical development.

References

1. Wiemann K, Mittrucker HW, Feger U, Welte SA, Yokoyama WM, Spies T, et al. Systemic NKG2D down-regulation impairs NK and CD8 T cell responses in vivo. J Immunol. 2005;175(2):720-9. PubMed PMID: 16002667.

2. Blery M, Vivier E. NKG2D-MICA Interaction: A Paradigm Shift in Innate Recognition. J Immunol. 2018;200(7):2229-30. Epub 2018/03/21. doi: 10.4049/jimmunol.1800176. PubMed PMID: 29555675.

P5 Anti-CD38 immunotherapy kills Treg (CD4+CD25+FoxP3+CD38hi) and Breg (CD19+CD24+CD38hi) cells and restores the anti-tumor T-cell repertoire in chronic lymphocytic leukemia (CLL)

Alak Manna, PhD, Sonikpreet Aulukh, MD, Laura Lewis-Tuffin, PhD, Taimur Sher, MD, Sikander Ailawadhi, MD, Rami Manuchakian, MD, Asher A. Chanan-Khan, Aneel Paulus, MD
Mayo Clinic, Jacksonville, FL, USA
Correspondence: Asher A. Chanan-Khan (Chanan-Khan.Asher@mayo.edu)

Background

CLL is the most common adult B-cell leukemia in western hemisphere. A subset of CLL cells immunophenotypically resembles B-regulatory cells (Bregs) and produce IL-10 and TGFβ that functionally imparts to them tumor-supportive properties. These cells are known to support and maintain T-regulatory (Treg) cells. Together, CLL Bregs and Tregs suppress CD8+ cytotoxic T-cell (cTLs) fostering an immunosuppressive and tumor promoting milieu that contribute to disease progression. We observed that a large proportion of CLL-Bregs and Tregs have a high CD38 receptor expression. This led us to hypothesize that eliminating them can potentially restore anti-tumor immune-effector response and is possible through anti-CD38 immunotherapy.

Methods

Blood peripheral mononuclear cells (PBMCs) were isolated from patients with a confirmed diagnosis of CLL (n=17) or healthy donors (n=6, control) under a protocol approved by the Mayo Clinic IRB. Characterization of CLL B-cells (CD19+CD5+), CLL-Bregs (CD19+CD24+CD38+IL10+), Tregs (CD4+CD25+CD127dimFoxP3+) was performed by flow-cytometry. Anti-CD38 immunotherapeutic, Daratumumab (Dara) was used in experiments. Intracellular IL-10 and FoxP3 were measured via fix/perm protocol followed by cytokine staining. Naïve T-cell to Treg transformation was determined using a trans-well co-culture assay. Extracellular IFNγ and IL-10 were measured via ELISA. Apoptosis was determined using annexin-V/PI staining. cTL proliferation was assessed via CFSE labeling of CD8+ sorted T-cells. For in vivo studies, a CLL patient-derived-xerograph (PDX) mouse model was established. (Figure 1)

Results

We noted that compared to healthy donors, CLL patients had a significantly higher % of Tregs (55.23±6.85%) and these Tregs had high CD38 expression (MFI=616.8±36.27). Consistent with our hypothesis, ex-vivo treatment of CLL patient PBMCs with Dara (1ug/mL) was highly lethal to CD38hi Bregs and Tregs. We also noted that CD38hi CLL-Bregs promoted transformation of naïve CD4+ T cells into Tregs in an IL-10/TGFβ dependent manner and neutralization of IL-10/TGFβ prevented this process. Notably, Dara treatment of naïve CD4+ T cells elicited same effect. Overall, Dara induced CLL cell death via ADCC, CDC, ADCP and mitochondrial/FcγR-mediated apoptosis. In Dara-treated CLL PBMC+T-cell co-cultures, ex vivo, we observed decreased IL-10 but increased IFN-y, Th17 and cTL counts. Similarly, in the PDX model, Dara-treated mice showed an increase in CD8+ and Th17 cells but a decrease in Bregs and Tregs.

Conclusions

Anti-CD38 immunotherapy is lethal to immunosuppressive CD38hi Breg/Treg cells and may improve anti-tumor T-cells function via modulating CLL immune-microenvironment. The results of these analyses have led to the approval of a phase-II clinical study that will be testing Dara in relapsed/refractory CLL patients.

Acknowledgements

Acknowledgement: Daniel Foundation of Alabama and Mayo Clinic Cancer Center (CA015083; A.A.C-K.), University of Iowa and Mayo Clinic Lymphoma SPORE Developmental Research Program (P50 CA097274; A.P.) and the Predolin Foundation (A.A.C-K).

Ethics Approval

Ethics Approval: The study was approved by Mayo Clinic Institutional Research Board (IRB# 14-009163)
Fig. 1 (abstract P5).
Fig. 1 (abstract P5).

Anti-CD38 immunotherapy kills Treg and Breg

P6 Anti-CD38 immunotherapy kills Treg (CD4+CD25+FoxP3+CD38hi) and Breg (CD19+CD24+CD38hi) cells and restores the anti-tumor T-cell repertoire in chronic lymphocytic leukemia (CLL)

Alak Manna, PhD, Sonikpreet Aulakh, MD, Laura Lewis-Tuffin, PhD, Taimur Sher, MD, Sikander Ailawadhi, MD, Rami Manuchakian, MD, Asher A. Chanan-Khan, Aneel Paulus, MD
Mayo Clinic, Jacksonville, FL, USA
Correspondence: Asher A. Chanan-Khan (Chanan-Khan.Asher@mayo.edu)

Background

CLL is the most common adult B-cell leukemia in western hemisphere. A subset of CLL cells immunophenotypically resembles B-regulatory cells (Bregs) and produce IL-10 and TGFβ that functionally imparts to them tumor-supportive properties. These cells are known to support and maintain T-regulatory (Treg) cells. Together, CLL Bregs and Tregs suppress CD8+ cytotoxic T-cell (cTLs) fostering an immunosuppressive and tumor promoting milieu that contribute to disease progression. We observed that a large proportion of CLL-Bregs and Tregs have a high CD38 receptor expression. This led us to hypothesize that eliminating them can potentially restore anti-tumor immune-effector response and is possible through anti-CD38 immunotherapy.

Methods

Blood peripheral mononuclear cells (PBMCs) were isolated from patients with a confirmed diagnosis of CLL (n=17) or healthy donors (n=6, control) under a protocol approved by the Mayo Clinic IRB. Characterization of CLL B-cells (CD19+CD5+), CLL-Bregs (CD19+CD24+CD38+IL10+), Tregs (CD4+CD25+CD127dimFoxP3+) was performed by flow-cytometry. Anti-CD38 immunotherapeutic, Daratumumab (Dara) was used in experiments. Intracellular IL-10 and FoxP3 were measured via fix/perm protocol followed by cytokine staining. Naïve T-cell to Treg transformation was determined using a trans-well co-culture assay. Extracellular IFNγ and IL-10 were measured via ELISA. Apoptosis was determined using annexin-V/PI staining. cTL proliferation was assessed via CFSE labeling of CD8+ sorted T-cells. For in vivo studies, a CLL patient-derived-xerograph (PDX) mouse model was established.

Results

We noted that compared to healthy donors, CLL patients had a significantly higher % of Tregs (55.23±6.85%) and these Tregs had high CD38 expression (MFI=616.8±36.27). Consistent with our hypothesis, ex-vivo treatment of CLL patient PBMCs with Dara (1ug/mL) was highly lethal to CD38hi Bregs and Tregs (Figure 1). We also noted that CD38hi CLL-Bregs promoted transformation of naïve CD4+ T cells into Tregs in an IL-10/TGFβ dependent manner and neutralization of IL-10/TGFβ prevented this process. Notably, Dara treatment of naïve CD4+ T cells elicited same effect. Overall, Dara induced CLL cell death via ADCC, CDC, ADCP and mitochondrial/FcγR-mediated apoptosis. In Dara-treated CLL PBMC+T-cell co-cultures, ex vivo, we observed decreased IL-10 but increased IFN-y, Th17 and cTL counts. Similarly, in the PDX model, Dara-treated mice showed an increase in CD8+ and Th17 cells but a decrease in Bregs and Tregs.

Conclusions

Anti-CD38 immunotherapy is lethal to immunosuppressive CD38hi Breg/Treg cells and may improve anti-tumor T-cells function via modulating CLL immune-microenvironment. The results of these analyses have led to the approval of a phase-II clinical study that will be testing Dara in relapsed/refractory CLL patients.

Acknowledgements

Acknowledgement: Daniel Foundation of Alabama and Mayo Clinic Cancer Center (CA015083; A.A.C-K.), University of Iowa and Mayo Clinic Lymphoma SPORE Developmental Research Program (P50 CA097274; A.P.) and the Predolin Foundation (A.A.C-K).

Ethics Approval

Ethics Approval: The study was approved by Mayo Clinic Institutional Research Board (IRB# 14-009163)
Fig. 1 (abstract P6).
Fig. 1 (abstract P6).

Anti-CD38 immunotherapy kills Treg and Breg

P7 Potent tumor-directed T cell activation and tumor inhibition induced by a 4-1BB x 5T4 ADAPTIR™ bispecific antibody

Michelle Nelson, PhD1, Gabriele Blahnik-Fagan1, Robert Bader1, Doreen Werchau, BS2, Anneli Nilsson2, Lill Ljung2, Jeannette Bannink, BS1, Danielle Mitchell1, Lynda Misher1, Catherine McMahan1, Maria Askmyr2, Anna Dahlman3, Peter Ellmark, PhD2, Gabriela Hernandez-Hoyos1, Sara Fritzell2,4
1Aptevo Therapeutics Inc., Seattle, WA, USA; 2Alligator Bioscience AB, Lund, Sweden; 3Allligator Bioscience AB, Lund, Sweden; 4Alligator Bioscience, Lund, Sweden
Correspondence: Gabriela Hernandez-Hoyos (ghoyos@apvo.com)

Background

4-1BB (CD137) is an important activation-induced co-stimulatory receptor that regulates immune responses of activated CD8+ T and NK cells, by enhancing proliferation, survival, cytolytic activity and IFN-γ production. The ability to induce potent anti-tumor activity by stimulating 4-1BB on tumor-specific cytotoxic T cells makes 4-1BB an attractive target for designing novel therapeutics for immuno-oncology. However, clinical development of a monospecific 4-1BB agonistic antibody has been hampered by dose-limiting hepatic toxicities. To minimize systemic immune toxicities, we have developed a novel 4-1BB x 5T4 bispecific antibody designed to direct tumor- specific T cell responses to the tumor microenvironment by stimulating 4-1BB function only when co-engaged with 5T4, a tumor-associated antigen.

Methods

ALG.APV-527 was built based the ADAPTIR™ platform with binding domains to 4-1BB and 5T4 generated using the ALLIGATOR-GOLD® human scFv library and subsequently optimized to increase binding affinity, function, stability and manufacturability. To assess its agonistic function, ALG.APV-527 was tested in NF-κB luciferase reporter systems and assays using primary cells in the presence or absence of cells expressing 5T4. To stimulate primary cells, enriched CD8+ T cells or unseparated PBMC were sub-optimally cultured with anti-CD3 antibody. Secretion of IFN-γ was measured at 72 hrs using ELISA or Luminex-based assays. To measure proliferation, PBMC were labelled with Cell TraceTM and CD8+ T cells were gated using multicolor flow cytometry. For tumor inhibition studies, the human colon carcinoma HCT116 xenograft model expressing endogenous levels of 5T4 was used. 5T4 expression was evaluated in normal human tissues and a range of different human tumors by immunohistochemistry (IHC).

Results

In vitro, ALG.APV-527 triggers luciferase reporter activity in the presence of 5T4-expressing cells. Using enriched T cells or whole unseparated PBMC, ALG.APV-527 induces a concentration-dependent increase of IFN-γ production when co-cultured with 5T4-expressing cells. ALG.APV-527 enhances primary CD8+ T cell proliferation preferentially over CD4+ T cells. Of significance, ALG.APV-527 is capable of inhibiting tumor growth in a human colon carcinoma xenograft model. IHC staining confirms that 5T4 is overexpressed in a range of solid tumors but not in normal tissues, indicating that ALG.APV-527 may primarily localize to the tumor improving potential for a achieving concentrations that demonstrate efficacy in a solid tumor setting.

Conclusions

ALG.APV-527 induces potent CD8+ T cell co-stimulation but only in the presence of 5T4 antigen. Based on preclinical data, ALG.APV-527 is a promising anti-cancer therapeutic for the treatment of a variety of 5T4- expressing solid tumors.

P8 Single-cell proteomic analysis of T cells stimulated by Bi-specific T-cell Engagers (BiTEs) shows robust and unique polyfunctional secretion profile

Sean Mackay, MBA1, Patrick Paczkowski1, Brianna Flynn, MS1, Kevin Morse1, Tiffany Coupet, BS2, Claire Godbersen, BS2, Charles Sentman2, Jing Zhou, MD, PhD1
1IsoPlexis, Branford, CT, USA; 2Geisel School of Medicine at Dartmouth, Lebanon, NH, USA
Correspondence: Jing Zhou (jing@isoplexis.com)

Background

T cell cytokines can drive anti-tumor activity and greater polyfunctionality (co-secretion of 2+ proteins per single cell) has been shown to be associated with improved clinical outcome in the study of CAR-T cell therapy and vaccine. We employed single-cell proteomics to fully evaluate the impact of BiTEs on polyfunctional T cells. BiTEs as cancer-targeting drugs link T cells with tumor by binding CD3 and a tumor antigen. Natural killer group 2, member D (NKG2D) ligand, such as MICA, expressed on more than 90% of human tumors but limited on normal tissues has emerged as appealing targets for BiTEs. This study has explored the polyfunctional profile of T cells by two BiTEs: B2-OKT3 (MICA x CD3) and hNKG2D-OKT3 (NKG2D ligands x CD3).

Methods

Blood T cells were negatively enriched from 3 healthy donors and incubated with K562 cells at a ratio of 1:2 in the presence of 250 ng/ml of B2-OKT3, hNKG2D-OKT3 or control Tz47-2C11. After 36 hours stimulation at 37°C, 5% CO2, CD4+ and CD8+ T cells were separated with anti-CD4 or anti-CD8 microbeads. Approximately 30,000 cells were loaded on the IsoCode single-cell chip (SCBC), pre-patterned with a 32-plex antibody ELISA array per cellular microchamber. Secreted proteins were captured from ~1500 single T cells after 16-hour-on-chip incubation at 37°C, 5% CO2. The T cell polyfunctional profile was evaluated across 5 functional groups: effector (Granzyme B, IFN-γ, MIP-1α, Perforin, TNF-α, TNF-β), stimulatory (GM-CSF, IL-2, IL-5, IL-7, IL-8, IL-9, IL-12, IL-15, IL-21), regulatory (IL-4, IL-10, IL-13, IL-22, TGF-β1, sCD40L, sCD137), inflammatory (IL-1β, IL-6, IL-17A, IL-17F, MCP-1, MCP-4), and chemoattractive (CCL-11, IP-10, MIP-1β, RANTES).

Results

Both B2-OKT3 and hNKG2D-OKT3 BiTEs enhanced single-cell polyfunctionality and polyfunctional strength index (PSI) of both CD4+ and CD8+ T cells when activated by K562 tumor cells compared to the Tz47-2C11 negative control (Figure 1). The polyfunctional response was mainly driven by effector cytokines, including Granzyme B, IFN-γ, and MIP-1α, chemoattractive MIP-1β, and regulatory sCD137. hNKG2D-OKT3 elicited more robust polyfunctional response of both CD4+ and CD8+ T cells to K562 cells stimulation than B2-OKT3. Detailed polyfunctional cell subsets with unique cytokine signatures induced by each BiTE are elucidated through high- dimensional single-cell visualizations of the data.

Conclusions

Single-cell proteomic analysis reveals a significantly upregulated polyfunctional profile of T cells induced by the BiTEs against tumor cells than the negative control, providing important insights into BiTE-triggered T cell activity as well as better evaluation and understanding of BiTE therapies.
Fig. 1 (abstract P8).
Fig. 1 (abstract P8).

T Cell Polyfunctional Strength Enhanced by BiTEs

P9 Identification and functional profiling of PD-L1 targeted engineered toxin bodies for antigen seeding technology and redirection of T cell response to tumors

Brigitte Brieschke, BS, Sara LeMar, Garrett Robinson, Aimee Iberg, PhD, Shaoyou Chu, PhD, Jack Higgins, PhD, Erin Willert, PhD, Hilario Ramos, PhD
Molecular Templates, Austin, TX, USA
Correspondence: Hilario Ramos (hilario.ramos@mtem.com)

Background

Molecular Templates’ Engineered Toxin Body (ETB) platform comprises recombinant immunotoxins leveraging Shiga-like Toxin A subunit (SLTA) properties of self-internalization, predictable retrograde transport, and lethal ribosomal inactivation with antibody binding domains to create targeted biologics capable of potent and specific direct killing of cancerous cells (MOA-1). Antigen Seeding Technology (AST) addition to the ETB scaffold provides a novel approach for redirection of preexisting memory cytotoxic T lymphocytes (CTLs) to cancerous cells (MOA-2). Both MOAs are designed to be functional in patients previously treated with standard of care agents. Here we describe the development of PD-L1 targeted ETBs with AST functionality capable of promoting cytolytic activity by CTLs recognizing a common Cytomegalovirus (CMV) viral antigen (HLA:A02 restricted CMV-pp65- NLVPVMATV, A2-pp65) on targeted tumor cells. We further describe the characteristics that distinguish the complementary mechanisms of action.

Methods

ETBs comprising SLTA fused to PD-L1 scFvs were engineered with or without A2-pp65 peptide. Human tumor cell lines expressing or lacking PD-L1 and HLA:A2 were used as target cells for cytotoxicity assays. Antigen restricted CTLs were expanded from human donors and used in co-culture models as effector cells.

Results

ETBs with potent direct cell kill activity have been identified to bind PD-L1 outside of or overlapping critical contact residues for PD-1. These ETBs have checkpoint inhibitor activity in a PD-1/PD-L1 blockade assay, though significantly less than their corresponding monoclonal antibody. A2-pp65 peptide fusion to ETBs resulted in cell- surface presentation of A2-pp65 peptide in complex with MHC-I and triggered efficient lysis by A2-pp65 specific CTLs in a target and HLA-restricted fashion. SLTA mutations which cause ER retention or do not inactivate ribosomes have no direct cell kill activity (remove MOA-1) but retain the antigen presentation and CTL directed lysis (maintain MOA-2). Additionally, genetic engineering identified modifications that could enhance MOA-1 activity without limiting MOA-2 activity, thus identifying molecules with optimal activity for clinical development. The predictable routing of ETBs support both MOAs and indicates a reduced threshold and routing requirement for MOA-2 as compared to MOA-1, allowing for broader cytotoxicity.

Conclusions

We have developed ETBs which bind distinct epitopes on PD-L1 and provide two unique and complementary mechanisms of action. Coupling both mechanisms of cytotoxicity into one molecule allows for potential to increase target penetrance, expand a prolonged immune response, and overcome resistance. In vivo syngeneic and xenograft studies are ongoing in preparation for clinical development of PD-L1 targeted ETBs with AST functionality in 2019.

P10 Local radiation with intratumoral anti-disialoganglioside (anti-GD2) and interleukin-2 (IL2) induces significant tumor responses with immunologic memory in a syngeneic murine NXS2 neuroblastoma model

Julie Voeller, MD1, Amy Erbe, PhD1, Kayla Rasmussen, MS1, Jacob Slowinski1, Sabrina VandenHeuvel1, Ravi Patel, MD, PhD1, Hans Loibner, PhD2, Stephen Gillies, PhD3, Jacquelyn Hank, PhD1, Alexander Rakhmilevich, MD, PhD1, Zachary Morris, MD, PhD1, Paul Sondel, MD, PhD1
1University of Wisconsin Madison, Madison, WI, USA; 2HL Bioscience Research GmbH, Vienna, Austria; 3Provenance Biopharmaceuticals, Carlisle, MA, USA
Correspondence: Julie Voeller (jvoeller@uwhealth.org)

Background

Neuroblastoma is the most common extracranial solid tumor in pediatrics. Standard therapy for patients with high-risk disease includes anti-disialoganglioside (anti-GD2) monoclonal antibody (mAb), GM-CSF (granulocyte- macrophage colony stimulating factor), and IL2 (interleukin-2)—an immunotherapeutic regimen that has significantly improved survival rates. Our lab has previously shown that local radiation therapy (RT) combined with intratumoral (IT) immunocytokine (IC; a fusion of hu14.18 anti-GD2 mAb and IL2) can cure mice with melanoma. We aimed to test and optimize this in situ vaccine approach in a murine neuroblastoma model.

Methods

Using the murine NXS2 neuroblastoma cell line, subcutaneous neuroblastoma tumors were established on the dorsal right flank of syngeneic A/J mice. Mice bearing 155mm3 tumors (engrafted about 2 weeks prior) received no RT or 12Gy RT to the tumor on treatment day 1, followed by daily intratumoral injections of either 50μg IC or PBS on treatment days 6-10. All mice with complete response were rechallenged on treatment day 90 by injecting NXS2 cells into the dorsal left flank.

Results

We observed improved tumor control (Figure A) and animal survival (Figure B; p<0.0001) when animals were treated with a combination of RT and IT-IC. Complete tumor regression was observed in 75% (9/12) of animals receiving RT and IT-IC, with 89% (8/9) of these rejecting rechallenge. Of all the other groups, only 9% (1/11) of animals receiving IT-IC alone and 33% (4/12) of animals receiving 12Gy and PBS had complete tumor regression.

Conclusions

Combined treatment with RT and intratumoral IC cures most mice bearing a single, 155mm3 NXS2 neuroblastoma tumor and induces immunologic memory. Our ongoing studies continue to investigate this immunotherapy regimen to test its effectiveness in more advanced disease as well as in other in vivo syngeneic murine neuroblastoma model systems.
Fig. 1 (abstract P10).
Fig. 1 (abstract P10).

See text for description.

Fig. 2 (abstract P10).
Fig. 2 (abstract P10).

See text for description.

P11 A CD25-targeted pyrrolobenzodiazepine dimer-based antibody-drug conjugate shows potent anti-tumor activity in pre-clinical models of solid tumors either alone or in combination with a PD-1 inhibitor

Francesca Zammarchi, PhD1, Karin Havenith, PhD1, Francois Bertelli2, Balakumar Vijayakrishnan2, Patrick van Berkel, PhD1
1ADC Therapeutics, London, UK; 2Spirogen/MedImmune, London, UK
Correspondence: Francesca Zammarchi (francesca.zammarchi@adctherapeutics.com)

Background

Regulatory T (Treg) cells infiltrate into various types of human cancers and contribute to the immunosuppressive tumor microenvironment [1]. The intratumoral balance between Tregs versus T effectors (Teffs) cells appears to impact the outcome of the immune system-mediated tumor eradication and numerous attempts are currently underway to reduce the CD25-expressing Tregs cells [2].

Methods

Sur301 is an antibody-drug conjugate (ADC) composed of PC61, a rat monoclonal antibody directed against mouse CD25, stochastically conjugated to a pyrrolobenzodiazepine (PBD) dimer via a protease-cleavable linker, with a drug-to-antibody ratio of 2.

Results

In vitro, sur301 demonstrated potent and specific cytotoxicity in a CD25-expressing mouse lymphoma cell line, while no specific cytotoxicity was observed in two CD25-negative murine colon cancer-derived cell lines, MC38 and CT26. All three cell lines were highly sensitive to SG3199, the PBD dimer toxin of sur301, irrespective of their CD25-status. In vivo, sur301 anti-tumor activity was investigated in the syngeneic MC38 and CT26 models, two immunogenic colon cancer models with tumor-infiltrating CD25-positive Treg cells [3]. Sur301 was administered either alone (0.1, 0.5 or 1 mg/kg, single dose) or in combination with an anti-PD1 antibody. A non-binding control ADC was used as negative control and tested as single dose at 1 mg/kg, either alone or in combination with an anti- PD1 antibody. Single doses of sur301 at 0.5 or 1 mg/kg induced strong and durable anti-tumor activity in both models. When tumor free survivor (TFS) animals were re-challenged with MC38 or CT26 cells, no mice developed new tumors, demonstrating tumor-specific protective immunity was generated by the single action of sur301. A single sub-optimal dose of sur301 at 0.1 mg/kg elicited limited anti-tumor activity when tested as single agent, but combination with an anti-PD1 antibody resulted in synergistic anti-tumor activity in both models. Combination of a single dose of sur301 at 0.5 or 1 mg/kg with the anti-PD1 antibody further increased the number of responders and none of the re-challenged animals developed new tumors. The non-binding control ADC showed no or significant reduced activity when tested at the same dose as sur301.

Conclusions

In conclusion, sur301 demonstrated potent in vivo activity against CD25-negative immunogenic solid tumors with infiltrating CD25-positive Treg cells. Sur301 in vivo activity was further enhanced by combination with anti-PD1 antibody. These data warrant further investigation of ADCT-301, a PBD-based ADC targeting human CD25, in patients with solid tumors, either alone or in combination with checkpoint inhibitors.

References

1. Sasidharan Nair V, Elkord E. Immune checkpoint inhibitors in cancer therapy: a focus on T-regulatory cells. Immunol Cell Biol, 2018. 96(1): p. 21-33.

2. Menetrier-Caux C, et al. Targeting regulatory T cells. Target Oncol, 2012. 7(1): p. 15-28.

3. Arce Vargas F, et al. Fc-optimized anti-CD25 depletes tumor-infiltrating regulatory T cells and synergizes with PD-1 blockade to eradicate established tumors. Immunity, 2017. 46(4): p. 577-586.

Ethics Approval

All in vivo study work was approved by the Institutional Animal Care and Use Committee at Charles River Laboratories, Morrisville, N.C..

Best Practices for Improving Cancer Immunotherapy

P12 Response and toxicity with immune checkpoint inhibition in older patients with non-small-cell lung cancer

Keval Yerigeri, BS1, Kristen Marrone, MD2, Jiajia Zhang, MD, MPH2, Julie Brahmer, MD2, Patrick Forde, MD2, Christine Hann, MD, PhD2, David Ettinger, MD2, Ronan Kelly, MD MBA2, Josephine Feliciano, MD2, Sarah Sagorsky, PA-C2, Michelle Turner, NP2, Valerie Rowe, NP2, Jarushka Naidoo, MD2
1Northeast Ohio Medical University, Copley, OH, USA; 2Johns Hopkins University School of Medicine, Sidney Kimmel Comprehensive Cancer Center, Baltimore, MD, USA
Correspondence: Kristen Marrone (kmarron1@jhmi.edu)

Background

Immune checkpoint inhibition (ICI) has rapidly become standard of care in advanced or metastatic non-small-cell lung cancer (NSCLC) treatment. Initial phase III clinical trials suggest ICI may have decreased efficacy in NSCLC patients ≥ 75 years old. The relationship between age-related immune system changes and ICI treatment is poorly understood.

Methods

The Johns Hopkins Upper Aerodigestive Diseases Immunotherapy Database was queried for all patients ≥ 75 years old treated with anti-PD-1/PD-L1 agents as part of a clinical trial or standard of care, from 2007 to 2018.

Results

Thirty-one patients ≥ 75 years old receiving anti-PD-1/PD-L1 agents for locally advanced or metastatic NSCLC were identified. Eleven patients were female, median age was 80.8 years (range: 75.1-90.6) with median ECOG PS=1 (range: 0-3). Twenty-seven patients received PD-1/PD-L1 monotherapy (nivolumab=16, pembrolizumab=10, atezolizumab=1) and 4 received combination (+chemotherapy=1, +ipilimumab=2, +additional ICI=1). Ten patients received ICI in the first-line setting (1L); 21 patients in the second-line or beyond (2L+). In 1L ICI monotherapy (n=8), median doses received was 5.5 (range: 2-19), median progression-free survival (mPFS) was 7.3m, and median overall survival (mOS) was 11.3m. In 2L+ patients, median dose administration was 4 (range: 1-24), mPFS was 7m and mOS was 7.6m. Across 1L and 2L+ ICI monotherapy patients, a rate of 81.5% all-grade toxicity was seen, of which 30% were high-grade (3+). All 1L and 2L+ ICI combination patients (n=4) experienced a toxicity, with 3 patients experiencing high-grade events. Across all patients, the most common low-grade toxicities were fatigue (n=8) and dyspnea (n=8). High-grade pneumonitis was seen in two 1L ICI monotherapy patients; 2L+ ICI monotherapy high-grade toxicities included dyspnea (n=4), hypoxia (n=1; Grade 5), pneumonitis (n=1), chest pain (n=1), delirium (n=1), aspiration (n=1), heart failure (n=1), lymphadenopathy (n=1) and pleural infection (n=1). Combination ICI high grade toxicities included pneumonitis (1L=1; 2L+=1) and rash (2L+=1). Across patients, reasons for treatment discontinuation included progressive disease (31%), with double the patients stopping for toxicity (62%) and treatment ongoing for 2 patients.

Conclusions

Our results indicate increased frequency and severity of toxicity in anti-PD1/PD-L1 treated older NSCLC patients, with decreased time to off treatment compared to landmark phase III studies. Survival data comparisons are limited in the setting of the current small sample size, but show interesting trends of decreased time on therapy and decreased overall survival. Further translational evaluation of senescent remodeling’s role in outcome and toxicity with ICI in older NSCLC patients is needed.

Acknowledgements

We would like to thank the patients and their families who agreed to participate in the Johns Hopkins Upper Aerodigestive Diseases Immunotherapy Database.

Ethics Approval

The study was approved by the Johns Hopkins School of Medicine’s Ethics Board, IRB00087582.

Biomarkers and Immune Monitoring

P13 Molecular profiling of anti-PD-1 treated melanoma patients reveals importance of assessing neoantigen burden and tumor escape mechanisms for clinical treatment

Charles Abbott, PhD1, Sean Boyle, PhD1, Eric Levy, PhD1, Rena McClory1, Sekwon Jang, MD2, Richard Chen1
1Personalis, Menlo Park, CA, USA; 2Inova, Fairfax, VA, USA
Correspondence: Charles Abbott (cabbo003@ucr.edu)

Background

Despite the remarkable response of some melanoma patients to checkpoint inhibitor therapy, significant numbers of patients do not achieve complete response. It is of great interest to identify biomarkers and mechanisms that influence immunotherapy effectiveness. Here we apply a comprehensive tumor immuno-genomics platform (ACE ImmunoID) to identify potential biomarkers of response to checkpoint blockade therapy.

Methods

We characterized the immuno-genomics of tumors from 31 stage III/IV melanoma patients who have received anti-PD-1 treatments to assess potential factors influencing response. Tumor responses to the therapy were evaluated using RECIST criteria with a median follow-up of 12 months. Immuno-genomic profiling was performed using Personalis’ ACE ImmunoID platform: an augmented exome/transcriptome platform and analysis pipeline. Analysis included assessment of tumor mutations, neoantigen characterization, HLA typing, gene expression quantification, and tumor micro-environment profiling. The molecular information of the tumors was then analyzed together with their corresponding clinical response.

Results

We observed a trend that higher neoantigen burden was associated with better progression free survival. Further investigation of patients with high neoantigen burden that failed to achieve complete response (3 PD, 1 PR) revealed potential resistant mechanisms to anti-PD-1 therapy. Specifically, we identified two of these patients with high expression of IDO1 or CTLA4, which may facilitate immune escape in a PD-1 independent manner. Additionally, we found two patients with mutations in their antigen presentation machinery (APM). The first patient had two independent HLA mutations in HLA-A and HLA-B (stop-gain mutation and splice site mutation, respectively), leading to the likely loss of surface expression of two classes of HLA-A and HLA-B proteins. In the second APM mutation patient we observed a frameshift deletion event detected in B2M at a very high frequency (80% AF) in their tumor. These APM mutations suggest reduced neoantigen presentation in these patients, likely underlying mechanisms for tumor escape.

Conclusions

While we observed the expected association between neoantigen burden and response to checkpoint blockade therapy, we also identified potential resistance mechanisms in patients that involve perturbations to antigen presenting machinery and high expression of non-targeted checkpoint genes. This highlights the potential importance of broad immuno-genomic profiling of patients that are candidates for receiving immunotherapy. We are continuing to increase our cohort size to identify additional mechanisms for immune evasion.

Ethics Approval

IRB file # 16-2427

P14 Durvalumab treatment-induced transcriptional changes in the tumor microenvironment associated with longer survival in patients with late stage Non-Small Cell Lung Cancer (NSCLC)

Ikbel Achour, PhD, Zachary Cooper, PhD, Sriram Sridhar, Maria Ascierto, PhD, Jixin Wang, Young Lee, Natasha Angra, Shaad Abdullah, MD, FACP, Rajiv Raja, PhD, Brandon Higgs, PhD, Maria Jure-Kunkel
Medimmune, Gaithersburg, MD, USA
Correspondence: Maria Jure-Kunkel (jure-kunkelm@medimmune.com)

Background

Baseline biomarkers including PD-L1, IFNg signature and tumor mutational burden (TMB) have demonstrated clinical utility in predicting overall survival in patients treated with anti-PD(L)1 therapies. However, changes in the tumor microenvironment on checkpoint therapy and their relationship to survival are poorly understood. Here, we systematically analyzed tumor microenvironment transcriptional profiles before and on durvalumab treatment and explored association with survival in patients with NSCLC.

Methods

CP1108/NCT01693562 is a nonrandomized phase 1/2 trial evaluating durvalumab (10 mg/kg, Q2W) in patients with solid tumors including advanced NSCLC (squamous and non-squamous). RNA sequencing was performed on 97 baseline tumors and 29 paired baseline and on-treatment (6 weeks) tumors. Gene and pathway level analyses were performed in relation to overall survival at baseline (prolonged OS >2yrs, n=23, compared to short OS <1yrs, n=61) and following durvalumab treatment (n=11 in both OS groups); clinical data cut off was 10/16/2017

Results

Among 763 genes differentially expressed at baseline between tumors from patients with prolonged compared to short survival (FC≥ 1.5 ; p≤0.05), gene signatures of CD8, Th1, T-agonist, T-effector, B, Natural Killer cells (NK), M1 macrophages, dendritic cells (DC), chemoattractant chemokines, and IFNg were expressed at higher levels (FC≥2 ; p≤0.04). Following durvalumab-treatment, in both OS groups, CD8 and T-effector gene signatures were significantly increased (FC≥3 ; p≤0.04), and IFNg,Th1, M1 and chemoattractant chemokine gene signatures were moderately induced (FC≥1.7 ; p≤0.1). Baseline levels of Th2, M2 macrophages and MDSC gene signatures did not significantly change on durvalumab. T-agonist, B, NK and CD1c+ DC cell gene signatures were further induced (FC≥2 ; p≤0.05) only in patients with OS>2 yrs. High expression levels of T-agonist, B cell, and CD1c+ DC signatures correlated with improved OS in early stage non-squamous NSCLC in TCGA. On-treatment reduction of at least 2-fold (p≤0.04) in genes involved in angiogenic, metabolic, cell-cell adhesion and cell cycle pathways including WNT7B, VEGFA, FASN, EVPL and CDKN2B were observed specifically in patients with prolonged OS compared to no change in patients with short OS.

Conclusions

Durvalumab treatment resulted in substantial changes in gene expression in the tumor microenvironment, with notable increases in B and dendritic cell signatures associated to prolonged survival. Our results provide new insights into the anti-tumor mechanism of PD-L1 blockade.

Trial Registration

NCT01693562

P15 Development of a robust, simplified method to measure receptor occupancy in peripheral blood from patients treated with a novel anti-PD-1 agent, AB122

Devika Ashok, PhD, Dana Piovesan, MSc, Sharon Zhao, Hema Singh, Steve Young, PhD, Matthew Walters, PhD, Lisa Seitz, MSc
Arcus Bio, Hayward, CA, USA
Correspondence: Devika Ashok (dashok@arcusbio.com)

Background

Exhausted T cells express high levels of immune checkpoint proteins, including programmed cell death-1 (PD-1) receptor. Preclinical and clinical data support the role of PD-1 and its ligand, programmed cell death ligand 1 (PD- L1), in promoting tumor evasion by curtailing immune responses. In a Phase 1 clinical trial of the anti-PD-1 monoclonal antibody AB122, we determined receptor occupancy (RO) in peripheral blood T cells using a directly conjugated competitive antibody method. We contrasted the data quality and derived RO values to previously established methodology described for nivolumab using biotinylated anti-human IgG4.

Methods

RO assays were developed using healthy donor peripheral blood mononuclear cells (PBMCs) spiked with AB122. We evaluated parameters including specimen stability, fresh vs. frozen samples, wash conditions, reagent concentrations and adapted the protocol for application to whole blood specimens to eliminate the need for PBMC isolation. Multi-color flow cytometry enabled determination of RO as well as proliferation status in individual T cell subsets using Ki67 as a functional readout of the effect of anti-PD-1 therapy. We developed the RO panel to work in conjunction with an intra-nuclear staining protocol for Ki67. This included identification of optimal clones for surface staining, blocking non-specific staining and selection of a clone for Ki67 identification. Finally, we deployed both RO determination protocols to evaluate an initial set of samples from cancer patients enrolled in the ongoing dose escalation Phase 1 study of AB122.

Results

Comparable RO data were obtained using both the AB122 competitive antibody and saturation methodologies using biotinylated anti-human IgG4 in PBMC samples from study subjects. Across all initial subjects tested, including different dose groups and time points, an average of ≥ 90% RO was observed using either method. In addition, a greater than 2-fold increase in Ki67+ T cell subsets was observed in approximately half of the patients.

Conclusions

Data from our Phase 1 dose-escalation cohorts demonstrates complete RO across a range of dosing regimens of AB122 and is consistent with data published for other anti-PD-1 antibodies. This optimized assay eliminates the need for multiple wash steps, decreases variability and enables testing with smaller numbers of cells. In addition, we have modified the direct competition method to enable direct assay of whole blood specimens allowing for a one- step staining process and preservation at a central lab to eliminate PBMC isolation.

P16 Better efficacy of PD-1 antibody predicted by immune-related adverse effects is impaired by high dose steroids

Xue Bai, MD1, Michelle Kim2, Gyulnara Kasumova2, Tatyana Sharova3, Justine Cohen, DO4, Donald Lawrence, MD4, Christine Freedman, RN4, Riley Fadden, NP4, Krista Rubin, MS, FNP-BC4, Ryan Sullivan, MD4, Keith Flaherty4, Genevieve M. Boland, MD, PhD2
1Massachusetts General Hospital Cancer Center, Harvard Medical School; Department of Renal Cancer and Melanoma, Peking University Cancer Hospital and Institute, Boston, USA; 2Department of Surgical Oncology, Massachusetts General Hospital, Harvard Medical School; Geisel School of Medicine at Dartmouth, Hanover, USA; 3Department of Surgical Oncology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; 4Massachusetts General Hospital Cancer Center, Harvard Medical School, Boston, MA
Correspondence: Genevieve M. Boland (gmboland@partners.org)

Background

PD-1 antibody has greatly improved the prognosis of unresectable or metastatic melanoma, and is now the standard first line therapy. It also brings a spectrum of immune-related adverse effects (irAEs). However, the correlation between the presence and timing of irAEs and the efficacy of PD-1 antibody remains elusive.

Methods

We retrospectively collected clinical data of pembrolizumab or nivolumab monotherapy-treated patients in Massachusetts General Hospital from 2009 to 2017. Correlations between irAEs and clinical outcomes were statistically analyzed.

Results

Of total 147 enrolled patients, 81 (55.1%) had irAE(s) (median 1/patient), 33 (22.4%) had severe irAE(s) (grade 3,4). The presence of irAE(s) was correlated with better therapeutic efficacy, which was impaired but not entirely offset by the application of high dose steroids, i.e. irAE+/high-dose steroids(-) subgroup had the best clinical outcome (median PFS 132.1 weeks, median OS not reached), followed by irAE+/high-dose steroids(+) (median PFS 43.0 weeks, median OS 182.6 weeks), which was better than irAE- subgroup (median PFS 11.4 weeks, median OS 74.7 weeks) (P<0.001 as for both PFS and OS). In total 158 irAEs were reported, among which 37 (23.4%) affected skin, 32 (20.3%) endocrine system, 31 (19.6%) muscle and joints, 21 (13.3%) gastrointestinal, 12 (7.6%) pulmonary, 9 (5.7%) hepatic, 4 (2.5%) renal, 4 (2.5%) neural, 2 (1.3%) pancreas, and 6 (3.8%) others. Median onset time of irAEs affecting each system was significantly different (P=0.036), early onset irAEs included neuropathy (6.9 weeks), hepatitis (7.3 weeks), and late onset irAEs included musculoskeletal (32.7 weeks), and cutaneous (27.6 weeks). Within the subpopulation of patients with irAE(s), those with musculoskeletal irAEs had better therapeutic response (ORR 84.6% vs. 48.1%) (P=0.005), longer median PFS (119.6 vs. 43.4 weeks) (P=0.010), and longer median OS (not reached vs. 189.6 weeks) (P=0.013). While cutaneous irAEs in total did not correlate with outcomes, the subset of vitiligo patients (n=8) had longer median PFS (not reached vs. 60.4 weeks) (P=0.032), and the tendency towards better therapeutic response (ORR 87.5% vs. 56.9%) (P=0.186) and longer median OS (not reached vs. 229.3 weeks) (P=0.191). Interestingly, rare irAEs* requiring high dose steroids (median onset of 7.7 weeks), were correlated with shorter median PFS (28.1 vs. 90.0 weeks) (P=0.006) and shorter median OS (120.0 weeks vs. not reached) (P=0.037).

Conclusions

The presence of irAE(s) serves as a prognostic biomarker during PD-1 antibody monotherapy. Application of high dose steroids impairs PD-1 antibody efficacy, and may impact subtype-specific predictive values of different irAEs.

P17 Combined MAGE-A1,3/6,4, and 10 expression levels quantified in solid tumors by (BaseScope™) RNA in situ hybridization (ISH) identify targets for immunotherapy

Anshika Bajaj, PhD1, Helly Xiao Yan Pimentel2, Bingqing Zhang, PhD2, Ruby Hsu, PhD2, Peter Berglund, PhD1, Jan Ter Meulen, MD, PhD1
1Immune Design, Seattle, WA, USA; 2Advanced Cell Diagnostics, Neward, CA, USA
Correspondence: Anshika Bajaj (anshika.bajaj@immunedesign.com)

Background

Ongoing clinical trials of cancer vaccines and adoptive cell therapies target members of the melanoma-associated antigen (MAGE-A) family, highly prevalent in tumors. However, no multiplexed diagnostic assay is available to quantify MAGE-A expression for ideal patient selection. The homologous nature with 50-80% sequence identity between the MAGE-A genes poses significant challenges to their specific detection in tumors. Due to antibody cross-reactivity there is limited capability of protein-specific assays to detect and distinguish between the various MAGE-A antigens. Here, an RNA ISH based assay was developed to assess and quantitate MAGEA1, MAGEA3/6, MAGEA4, MAGEA10 expression in normal and tumor tissue.

Methods

MAGEA (-A1, -A3, -A4, and -A10) specific probes were designed targeting sequences with minimal inter-gene identity. Tissue samples that passed quality control were evaluated for MAGEA expression. BaseScope™ LS Red ISH assays were performed on Leica Bond RX using the BaseScope LS kit on cell pellet arrays, tumor, and normal tissues.

Results

4 assays were developed, each designed to specifically detect RNA encoding MAGE-A1, -A3/6, -A4, or -A10. Experiments done in control cell lines (1 negative and 4 cells lines each expressing 1 of the 4 MAGEA genes) demonstrated that the assays were highly specific and sensitive for their respective target genes. MAGEA expression was assessed in 10 melanoma, head and neck, lung, and esophageal cancer biopsies. Samples were assigned dot scores based on semi-quantitative visual scoring of the number of dots (RNA molecules)/cell. Moderate-high expression of all 4 MAGEAs (score of 2-3) was observed in biopsies from 1/1 melanoma patients, 2/3 lung cancer patients, and 1/4 head and neck cancer patients. High MAGEA1 and 3 expression only (score of 3) was observed in 3/4 head and neck cancer patients. As expected, analysis of normal tissue samples except for testes revealed minimal signal.

Conclusions

Specific and sensitive BaseScope assays were developed for MAGEA1, MAGEA3/6, MAGEA4, and MAGEA10. The assays demonstrate inter-gene specificity, are amenable to multiplexing, and can potentially be used as a companion diagnostic in clinical trials targeting MAGEA antigens. Furthermore, these preliminary results demonstrate that these 4 MAGE-A antigens are highly prevalent in cancers such as head and neck, melanoma, and lung and support the development of an active immunotherapy based on Immune Design’s dendritic cell-targeting ZVex® vector platform.

P18 Preliminary evaluation of a novel whole slide multispectral assessment of seven markers: Potential to minimize bias in the characterization of the tumor immune environment

Carmen Ballesteros Merino, PhD1, Shawn Jensen, PhD2, Carla Coltharp, PhD3, Kristin Roman, MS3, Chichung Wang3, Nikhil Lonberg, HSDG2, Sebastian Marwitz2, Tarsem Moudgil, MS2, William Miller, BS2, William Redmond, PhD2, Yoshinobu Koguchi, MD, PhD2, Carlo Bifulco, MD2, Clifford Hoyt, MS3, Bernard A. Fox, PhD2
1Robert W Franz Cancer Center, Earle A Chiles Research Insititute, Portland, OR, USA; 2Robert W Franz Cancer Center, Earle A Chiles Research Institute, Portland, OR, USA; 3Perkin Elmer, Hopkinton, MA, USA
Correspondence: Bernard A. Fox (foxb@foxlab.org)

Background

PD-L1 expression and tumor-mutational burden enrich for patients that respond to checkpoint blockade, but these evaluations are only a component of the entire story. Recently, our lab reported that evaluation of specific cell-cell relationships provided a powerful biomarker for overall survival in patients with HPV- head and neck cancer (HNSCC). However, the areas selected for analysis were operator selected “hot spots”. This approach introduces the potential for unconscious bias in the selection process. To address this, we have sought to perform whole slide evaluations of sections to compare with hot spot analysis. This study is a preliminary report applying a novel set of fluorophores and filters that allow the visualization of seven colors on a whole slide.

Methods

Tissue samples included pellets of cultured lymphocytes and tumor specimens. A sample of the cultured lymphocytes that were fixed and embedded were analyzed by flow cytometry for immune markers. Formalin-fixed paraffin embedded (FFPE) sections were stained with antibodies for CD8, CD68, FoxP3, PD-1, PD-L1, cytokeratin and DAPI. PerkinElmer Opal reagents were used to identify markers and included standard and a new set of fluorophores that included Opal 480, 520, 570, 620, 690, and 780. Slides were imaged using a new scanning approach on a Vectra Polaris (PerkinElmer, Inc, Waltham, MA).

Results

Preliminary comparison of cells that were used to produce FFPE blocks by flow cytometry and multiplex IHC provided similar results for some markers. Determining optimal staining, exposure times and thresholds for analysis for this new method needs work, but the potential exists for effective evaluation of a whole slide with 7 different markers.

Conclusions

Our preliminary results provide reason to be optimistic that this approach can assess 7 colors in a whole slide. Whole sections labelled with 7 colors and spectrally unmixed supports deeper analysis of immune-biology on multiple scales, including re-analysis of spatial metrics based on emerging hypotheses about how cellular and expression distributions relate to disease progression and response to therapy.

P19 Molecular determinants of response to PD-L1 blockade across tumor types

Romain Banchereau1, Ning Leng1, Edward Kadel, BS1, Dorothee Nickles, PhD1, Steve Lianoglou, BS, MSc, PhD1, Oliver Zill1, Sushit Jhunjhunwala1, Luciana Molinero, PhD1, Mahrukh Huseni1, Marcin Kowanetz, PhD1, Richard Bourgon, BS, PhD1, Craig Cummings, PhD1, Sanjeev Mariathasan, PhD1, Priti Hegde, PhD1, Thomas Powles, MBBS, MD, MRCP2
1Genentech, South San Francisco, CA, USA; 2BART, London, UK
Correspondence: Romain Banchereau (banchereau.romain@gene.com)

Background

Immune checkpoint inhibitors targeting the PD-1/PD-L1 axis lead to durable clinical responses in subsets of cancer patients across multiple indications including non-small cell lung cancer (NSCLC), urothelial carcinoma (UC) and renal cell carcinoma (RCC). This work aims at determining whether unifying molecular profiles can predict response across these tumor types.

Methods

379 samples from three phase II trials were investigated. PD-L1 expression on tumor-infiltrating immune cells (IC), tumor mutation burden (TMB) and bulk transcriptome measurements were obtained before treatment with atezolizumab from 218 UC (IMvigor210), 83 NSCLC (POPLAR) and 78 RCC (IMmotion150) patients. Objective response was assessed by RECIST v1.1. Patients from a phase I atezolizumab monotherapy basket study (PCD4989g) were employed as an independent validation cohort. PD-L1 IC was assessed by immunohistochemistry (Ventana SP142: >1% of IC was defined as positive). TMB was assessed by whole exome sequencing. Bulk tumor transcriptomes were assessed by RNAseq.

Results

Initial analyses focused on responder prevalence in PD-L1 IC+ and/or TMBhigh individuals. They revealed variable results across tumor types with overall sensitivity/specificity of 76.4%/34.5% and 74.5%/55.4% for PD-L1 IC and TMB respectively. Importantly, no common TMB threshold predicted response across indications. Unsupervised analysis revealed that RCC tumors cluster away from UC and NSCLCs. Supervised analysis showed that PD-L1 IC expression correlated with myeloid and lymphoid signatures across tumor groups, while few immune genes associated with TMB. Modular transcriptional analysis failed to identify a unified tumor signature associated with response, although parallels were seen between NSCLC and UC, but not RCC. Using a linear model that accounted for genes associated with PD-L1 IC levels, the CDK4/6 inhibitor CDKN2A, which is frequently mutated in UC and NSCLC tumors, was identified as the most significant correlate of response to PD-L1 inhibition, highlighting the association of non-immune pathways to checkpoint blockade outcome. Tumor-related pathways including mismatch repair and senescence were enriched in responders with low tumor immune infiltrate. Finally, machine learning identified a 42-gene signature associated with outcome, which included both immune- and tumor-related components. This signature complemented TMB and PD-L1 IC to increase responder prevalence both in training and independent validation cohorts.

Conclusions

While no unifying gene signature correlated with response across tumor types, consistent overlaps were observed between UC and NSCLC, highlighting common mechanisms of response to PD-L1 inhibition between tumors from different origins. Machine learning can integrate high-dimensional datasets across indications to identify both immune- and tumor-related determinants of response to checkpoint blockade.

P20 A structured tumor-immune microenvironment in triple negative breast cancer revealed by multiplexed ion beam imaging

Leeat Keren1, Marc Bosse1, Robert West2, Sean Bendall, PhD1, Michael Angelo, MD, PhD1
1Stanford University, Stanford, CA, USA; 2Stanford, Stanford, CA, USA
Correspondence: Michael Angelo (mangelo0@stanford.edu)

Background

Cancer progression is a complex process that depends on the interplay between cells in the tumor, the microenvironment, and the immune system, which can act both to promote and suppress growth and invasion [1]. Triple-negative breast cancer (TNBC) is an aggressive form of invasive breast cancer lacking appreciable expression of therapeutic targets: estrogen receptor, progesterone receptor, and Her2 [2]. In terms of TNBC immunotherapy, no single biomarker has been sufficient for adequate patient stratification [3]. Consequently, there is still much interest in its tumor immune landscape: which immune cell types are present, which immunoregulatory proteins are expressed, and how these vary between patients.

Methods

We leveraged a next-generation tissue pathology imaging platform we have developed, Multiplexed Ion Beam Imaging [4] coupled to Time of Flight (MIBI-TOF) mass analysis, to perform a retrospective study on TNBC patients from the Stanford Pathology archive. With this we simultaneously quantified in-situ expression of 36 proteins covering identity, function and immune regulation at sub-cellular resolution in 41 TNBC patients. This data enabled us to develop a multi-step analysis pipeline for standardized processing of this multiplexed imaging cohort, including deep-learning-based segmentation, cell type identification, and spatial enrichment analysis of the tumor immune microenvironment.

Results

While the composition of tumor-immune populations varied widely between individuals, this heterogeneity could be reconciled by the overall amount of immune infiltration, where there was enriched co-occurrence and ordering of specific immune populations conserved across the cohort. Monocytes were at all levels of infiltrate while B and NK cells only in patients with the greatest immune cell density. At the same time, distinct immune populations expressed checkpoint proteins (i.e. PD1, PD-L1, IDO, and LAG3) in different patients, and patients that express one immunosuppressive pathway were more likely to express another. Data-driven analysis of spatial organization revealed either immune compartmentalized or immune mixed tumors. Most interestingly, this histological organization was significantly correlated with expression of checkpoint molecules (particularly PD1, PD-L1, and IDO) in a cell-type- and location-specific manner. Here, ordered immune structures along the tumor-immune border served as a hallmark of tumor compartmentalization and were linked to overall survival with standard chemotherapy.

Conclusions

Together, these data demonstrate an organization in the tumor-immune microenvironment that is structured in cellular composition, spatial arrangement, and expression of regulatory proteins. We elucidate these organizational features creating a resource for TNBC and provide a framework to apply highly multiplexed subcellular imaging to complex immune oncology.

Acknowledgements

S.C.B. is supported by a gift from Christy and Bill Neidig, the Damon Runyon Cancer Research Foundation (DRG- 2017-09), the NIH 1DP2OD022550-01, 5U19AI116484-02, and U19AIP97

*Corresponding author email: paola.nistico@ifo.gov.it9. M.A is supported by 1-DP5-OD019822. S.C.B and M.A. are jointly supported by 1R01AG056287–01, 1R01AG057915-01, and 1U24CA224309-01 from the NIH. L.K. is a Damon Runyon Fellow supported by the Damon Runyon Cancer Research Foundation (DRG-2292-17)

References

1. Chen DS, Mellman I. Elements of cancer immunity and the cancer–immune set point. Nature. 2017;541:321–330.

2. Denkert C, Liedtke C, Tutt A, von Minckwitz G. Molecular alterations in triple negative breast cancer—the road to new treatment strategies. Lancet. 2017;389:2430–2442.

3. Maleki Vareki S, Garrigós C, Duran I Biomarkers of response to PD-1/PD-L1 inhibition. Crit. Rev. Oncol. Hematol. 2017;116:116–124.

4. Angelo M, et al. Multiplexed ion beam imaging of human breast tumors. Nat. Med. 2014;20:436–42.

P21 Deep learning-based PD-L1 tumor cell (TC) scoring improves survival prediction compared to pathologists on durvalumab-treated NSCLC patients

Nicolas Brieu, PhD1, Ansh Kapil1, Aleksandra Zuraw, Dr1, Abraham Silva, MD1, Marlon Rebelatto, DVM, PhD, DACVP2, Keith Steele, DVM, PhD2, Guenter Schmidt, PhD1
1Definiens, Munich, Germany; 2MedImmune, Gaithersburg, MD, USA

Background

PD-L1 expression in non-small cell lung carcinoma (NSCLC) patients is commonly quantified by the tumor cell (TC) score estimated by pathologists. An accurate score is key to identify patients that could benefit from anti-PD- L1 check point inhibitor treatment, patients with high score being more likely to respond to such therapy [1]. Recent advances in deep learning algorithms for computer vision enable an accurate alternative to pathologist scoring via the identification of positive and negative tumor regions [2]. With statistical analysis of the clinical response, we show the predictive value of the automated scoring system and evaluate it against pathologist scoring.

Methods

The dataset consists of tissue sections of NSCLC patients from subsets of NCT01693562 [1] and NCT02000947 [3] clinical trials and stained with Ventana SP263 PD-L1 assay. Using two-fold cross validation, we train a deep semi- supervised convolutional neural network [2] for the automated segmentation of PD-L1 positive and PD-L1 negative tumor cell regions. Training is based on labeled patches generated from the manual annotation of positive and negative tumor cell regions by two pathologists on a subset of images (n=20) as well as on unlabeled patches generated from the remaining non-annotated images (n=305). The dataset for network application, TC score estimation and further statistical analysis consists of the non-annotated NSCLC samples of the durvalumab monotherapy clinical trial (NCT01693562) for which overall survival (OS), progression free survival (PFS) defined by RECIST criteria, and three pathologist scores are available (n=152). The automated score is estimated from the segmented regions as the relative area of the PD-L1 positive tumor cell region. Using leave-one-out cross-validation, we finally optimize the cutpoint between low and high scores regarding log-rank test associated with overall survival and progression free survival.

Results

The deep learning-based score is strongly correlated with the consolidated pathologist score obtained by majority voting (Pearson:0.80). It yields more significant OS and PFS stratifications in terms of Cox proportional hazards regression than the pathologist score for both the standard 25% cut-off [1] and the respective optimized cut-off (Figure.1 and Table.1).

Conclusions

Our results suggest that the proposed deep learning based system for PD-L1 TC scoring enables the retrospective stratification of durvalumab-treated NSCLC patients into predictive groups. Upon further improvement of the correlation to pathologists and confirmation of the presented results in a prospective trial, we envision that the proposed model could be used in a clinical routine setting to identify patients which may benefit from anti-PD-L1 therapy.

References

1. M. Rebelatto, J. Walker et al., Development of a programmed cell death ligand-1 immunohistochemical assay validated for analysis of non-small cell lung cancer and head and neck squamous cell carcinoma, Diagnostic Pathology (2016), 11:95, DOI 10.1186/s13000-016-0545-8

2. A. Kapil, N. Brieu et al., Deep Semi Supervised Generative Learning for Automated PD-L1 Tumor Cell Scoring on NSCLC Tissue Needle Biopsies, ArXiv (2018), https://arxiv.org/abs/1806.11036

3. S. Antonia, NA. Rizni et al., Safety and antitumour activity of durvalumab plus tremelimumab in non-small cell lung cancer: a multicentre, phase 1b study, Lancet Oncol (2016), 17(3):299-308, DOI 10.1016/S1470- 2045(15)00544-6

Ethics Approval

This works relies on data from NCT01693562 and NCT02000947 clinical trials (clinicaltrials.gov)
Fig. 1 (abstract P21).
Fig. 1 (abstract P21).

See text for description.

Table 1 (abstract P21).

See text for description.

P22 Impact of tumor inherent interferons on immune reactivity and personalized therapy in triple negative breast cancer

Natasha Brockwell, BBiomed (Hons)1, Jai Rautela3, Tim Molloy, PhD4, Sandra O'Toole, MD, PhD5, Vinod Ganju, MBBS, FRACP6, Belinda Parker1
1La Trobe Institute for Molecular Science, Bundoora, Australia; 2La Trobe Institute for Molecual Science, Bundoora, Australia; 3Walter and Eliza Hall Institute, Melbourne, Australia; 4St Vincents Centre for Medical Research, Sydney, Australia; 5Garvan Institute for Medical Research, Sydney, Australia; 6Hudson Institute for Medical Research, Frankston, Australia
Correspondence: Natasha Brockwell (n.brockwell@latrobe.edu.au)

Background

Triple negative breast cancer (TNBC) is known for its ability to rapidly metastasize within the first two years and its association with tumor infiltrating lymphocytes (TILs). As immune infiltrate has been associated with a good prognosis and therapeutic response in TNBC, immunotherapy is now being trialed. However, responses have been underwhelming to date and difficult to predict, leading to an inability to accurately weigh up the benefit-to-risk ratio for their implementation. Previous work done in our laboratory demonstrated that type I IFN signalling can increase the heat of the tumor, induce a tumor specific T cell responses and sensitize mice to checkpoint inhibitors [1]. This suggested that characterization of a tumors heat is imperative in deciding which patients are most likely to benefit from immunotherapy and the type of immunotherapy they should receive.

Methods

Multiplex immunohistochemistry using the OPAL method was utilized to characterize tumor heat through assessment of T cell subsets and effector status and novel IFN biomarkers. A TNBC cohort (n = 21) where sequential biopsies were taken pre, mid and post chemotherapy was used to assess the role of tumor heat in chemotherapeutic response and relapse. Two independent adjuvant TNBC cohorts (n = 398; n = 159) were used to validate findings. Murine TNBC cells were manipulated to have constitutive expression of the type I IFN pathway. Syngeneic mouse models were used to assess the role of inherent type I IFN signalling on chemotherapeutic response, survival and the immune landscape.

Results

We demonstrate the superior prognostic information that can be gathered from TIL characterization whereby T cell subsets and their effector function can be used to predict response to chemotherapy and relapse. Furthermore we identified a novel prognostic marker that indicates presence of an intact type I IFN signalling pathway. Patients with loss of this marker were up to eight times more likely to relapse with metastatic disease than those who retained the biomarker, this was prognostic in 3 independent TNBC cohorts. Overexpression of the type I IFN pathway in murine TNBC cells resulted in increased sensitivity to chemotherapy, decreased metastasis and promotion of a T cell inflamed tumor.

Conclusions

Our work suggests tumor inherent type I IFN signalling and TIL characterization predicts relapse. Immunotherapy aimed at increasing tumor heat may hold promise in those lacking immune activation or IFN signalling prior to chemotherapy or checkpoint inhibitors, supporting the notion of TME characterization pre-treatment to personalize therapy.

References

1. Brockwell N K, Owen K L, Zanker D, Spurling A, Rautela J, Duivenvoorden H M, Baschuk N, Caramia F, Loi S, Darcy P K, Lim E, and Parker B S, “Neoadjuvant Interferons: Critical for effective PD-1 based immunotherapy in TNBC,” Cancer Immunol. Res., Aug. 2017; Vol 5:871-884.

Ethics Approval

This study was approved by the latrobe animal ethics committee, approval number AEC15-62. This study was approved by the human resources ethics committee of the Royal Prince Alfred Hospital, approval number X15-0388 [SSA/16/RPAH/397].

P23 Centrifuge-less immunostaining of suspension cells for flow cytometry analysis by DA-Cell™ washer and plate for superior data and workflow

Namyong Kim, PhD, Melvin Lye, Namyong Kim, PhD
Curiox Biosystems
Correspondence: Melvin Lye (melvin@curiox.com)

Background

We describe the DA-Cell™ system, a novel wall-less plate and laminar-flow cell washer that enables the automated washing of suspension cells and retains more than 95% cells at a fraction of the time and with higher viability of cells.

Methods

The wall-less DropArray (DA) plate consists of an array of 96 hydrophilic spots separated and surrounded by a hydrophobic surface, which functions as a virtual wall. In a typical immunostaining assay, a 50 μl drop, containing cells and antibody mix, is dispensed on each spot of the DA plate. During incubation, cells settle on the surface of the spots. The plate then undergoes a laminar-flow washing process in DA-Cell washer by repeated cycles of aspiration and dispensing of buffer through two sets of nozzles. The controlled buffer flow minimizes turbulence and cell loss. The cell washing process only requires 3-4 minutes by eliminating the need of a centrifuge, which also reduces stress on the cells and possible cross-contamination of antibodies on cell membranes, leading to better segregation of cell populations in flow cytometry. Additional incubation time of 10-20 minutes improves cell retention to more than 99%. Most importantly, cell incubation and washing on a DA-Cell system minimizes operator variability as mixing and washing steps are mechanically controlled and mostly automated, significantly improving reproducibility and consistency of flow cytometry analysis.

Results

We show a series of immunostaining assays comparing the DA-Cell system with a conventional centrifugation. Based on the staining index, absolute cell counts and additional data, we demonstrate that the DA-Cell system produces superior data while simplifying and expediting cell preparation for flow cytometry analysis.

Conclusions

The DA-Cell system produces superior data while simplifying and expediting cell preparation for flow cytometry analysis.

Acknowledgements

Singapore Immunology Network

P24 Centrifuge-less red blood cell lysis and immunostaining of whole blood for flow cytometry using DA-Cell washer and plate

Namyong Kim, PhD, Melvin Lye, Namyong Kim, PhD
Curiox Biosystems, San Carlos, TX, USA
Correspondence: Melvin Lye (melvin@curiox.com)

Background

Blood cells are prime indicators of immuno-surveillance, and the ease of blood sampling makes blood analysis a key interest for clinical and research applications. While current flow cytometry methods are high-throughput and provide fine resolution in the segregation of white blood cell (WBC) populations, WBC enrichment involving red blood cell (RBC) lysis are laborious and typically performed manually, contributing to experimental variability especially as blood cells are sensitive to physical and chemical stress.

Methods

We describe RBC lysis and leukocyte immunostaining on a centrifuge-less platform DA-Cell™, using a novel wall-less plate and laminar flow washer. The DropArray (DA) plate consists of an array of 96 hydrophilic spots surrounded by hydrophobic surface, which functions as a virtual wall that separates each spot. The maximum volume of each spot can be increased to 300μL with an insert. During lysis, WBC settle to the surface of the spot, allowing the spent lysis buffer to be removed. After removal of the insert, the plate goes through a gentle 3-4min laminar-flow washing process in the DA-Cell washer, decreasing cell washing time by at least 50% while eliminating centrifugation that stresses cells and disrupts antibody binding.

Results

In studies comparing mouse whole blood lysis (1X RBC Lysis buffer, eBioscience) and antibody staining by conventional tube centrifuge and DA-Cell, DA-Cell achieved dramatically higher staining index and improved resolution of cell cluster by flow cytometry. CD45+ leukocyte recovery and viability was uncompromised compared to conventional tube centrifuge.

Conclusions

In summary, DA-Cell system provides gentle, fast and convenient blood lysis, while improving data quality with superior antibody staining.

P25 Consistent pharmacodynamics and immunological responses to the TLR9 agonist, SD-101, following intratumoral injection in multiple cancer types

Albert Candia, PhD1, Cristiana Guiducci, PhD1, Ezra Cohen, MD2, Ronald Levy, MD3, Mohammed Milhem, MBBS4, Antoni Ribas, MD, PhD5, Thomas Tuting, MD6, Erick Gamelin, MD PhD1, Robert Janssen, MD1, Robert Coffman, PhD1
1Dynavax Technologies, Berkeley, CA, USA; 2Moores Cancer Center, La Jolla, CA, USA; 3Stanford University Hospital and Clinics, Stanford, CA; 4University of Iowa Health Care, Iowa City, IA, CA; 5University of California at Los Angeles, Los Angeles, CA; 6University Hospital Magdeburg, Magdeburg, Germany
Correspondence: Albert Candia (acandia@dynavax.com)

Background

SD-101 is a synthetic class C CpG oligonucleotide agonist of Toll-like receptor 9. SD-101 stimulates dendritic cells to release interferon-alpha and mature into antigen presenting cells that effectively activate T cell responses. SD-101 is administered intratumorally (IT) and has been evaluated in combination with radiation therapy for lymphoma, and is currently being evaluated with pembrolizumab for melanoma and HNSCC. Pharmacodynamic and biomarker assessments across these three different tumor types offer mechanistic insights into the anti-tumor activity observed in the clinic.

Methods

Peripheral blood collected before and 24 hours after dosing was analyzed with Nanostring or qPCR by a panel of IFN responsive genes as an indirect measure of target engagement. Biopsies from injected lesions were collected prior to treatment and at specific post-dose time points, and gene expression was analyzed by Nanostring to evaluate the immunophenotype of the tumor environment. Tumor responses were assessed using Cheson criteria for lymphoma and RECIST v1.1 for melanoma and HNSCC.

Results

Type 1 IFN production was demonstrated by the activation of IFN responsive genes in peripheral blood. The range of induction on an individual basis (2 to 29 fold) and maximal induction on an averaged cohort basis (approximately 10-fold) were comparable across tumor types indicating similar mechanism of dendritic cell activation. Analysis of gene expression in tumor biopsies before and after treatment across the three tumor types shows a consistent increase in immune functions and cell types expected to contribute to anti-tumor activity. These changes are also consistent with the known mechanisms of action for SD-101 and include increases in CD8+, Th1, and NK cells. Responding patients (PR/CR) had average increases of at least 2-fold and as high as 9-fold in these cell types and functions. These increases in immune functions occurred in patients with or without prior checkpoint inhibitor therapy, and in the largest data set from melanoma patients, the changes in these cell types and functions correlated significantly with reductions in size of target lesions.

Conclusions

Biomarker assessments across three different tumor types following IT administration of SD-101 demonstrate consistent pharmacodynamic and biomarker activities consistent with its mechanism of action. The results suggest that activation of the innate immune system may be a core component of combination therapies in orchestrating an anti-tumor immune response in a wide range of cancer types.

Ethics Approval

The studies described were approved at the Institutional Review Boards of the respective clinical sites.

P26 High-dimensional flow cytometry of circulating immune cells predicts clinical responses to combination Immune Checkpoint Blockade (ICB) and Radiotherapy (RT) in Gastroesophageal Cancer (GEC)

Joseph Chao, MD, Wanqiu Hou, Yi-Jen Chen, MD, PhD, Helen Chen, Michael Tajon, Marwan Fakih, MD, Peter P. Lee, MD
City of Hope Comprehensive Cancer Center, Duarte, CA, USA
Correspondence: Peter P. Lee (plee@coh.org)

Background

While ICB has been promising, the majority of GEC patients do not respond to single agent anti-PD-1 therapy. Combination strategies are being explored to augment immune responses including combining ICB with RT. We are currently conducting a prospective trial testing palliative RT with pembrolizumab in patients with metastatic GEC. To study immune correlates of ICB, we employed high-dimensional flow cytometry single cell analyses to characterize blood immune profiling as predictive biomarkers of clinical response.

Methods

In this single institutional trial, patients received standard palliative RT 30 Gy over 10 fractions to a single site of disease. Pembrolizumab 200 mg was given concurrently with RT with first dose concordant with the first fraction. Cycles repeated every 3 weeks for up to 35 cycles in the absence of disease progression or unacceptable toxicity. Peripheral blood was collected at baseline prior to first fraction of RT (C1D1 pembrolizumab) and ~21 days after completion of RT (~C2D15 pembrolizumab). Blood comprehensive immune profiling was interrogated using three 15-color flow cytometry panels. Abscopal responses were assessed using RECIST1.1 of lesions out of the field of RT.

Results

In this current analysis, 5 patients were included. RECIST responses included 2 confirmed partial responses (PRs), and 3 patients with progressive disease. The 2 PRs have been durable lasting >12 months and ongoing at data cut- off. On C2D15, most patients demonstrated a decrease in CD56hi NK cells (p=0.04), CD1c+ dendritic cells (p=0.02), plasmacytoid dendritic cells (p=0.01), CD33hi myeloid-derived suppressor cells (p=0.02), and PD- 1+KLRG1+ exhausted CD8 T cells (p=0.005). Comprehensive immune profiling demonstrated a strong correlation of RECIST responses with low levels of circulating follicular helper T cells (r=0.99, p=0.001), and PD-1+BTLA+ exhausted CD4 T cells (r=0.91, p=0.03) at baseline C1D1. Analysis of immune changes over time also demonstrated a strong correlation of RECIST responses with an increase in circulating T cells (r=0.94, p=0.02) and nonclassical monocytes (r=0.92, p=0.03) as well as a decrease in Th2 cells (r=0.89, p=0.04) at C2D15 vs. C1D1.

Conclusions

Palliative RT plus pembrolizumab demonstrated encouraging activity in our dataset. Durable RECIST responses correlated with changes at C2D15 in circulating innate immune (T cells and nonclassical monocytes) as well as adaptive immune signatures (follicular helper T cells, Th2, and exhausted CD4 T cells). The ability to identify blood biomarkers early in ICB therapy that may predict durable clinical benefit is of significant clinical utility in GEC and warrants study in larger prospective cohorts.

Acknowledgements

Joseph Chao and Wanqiu Hou contributed equally to this work.

Trial Registration

NCT02830594

Ethics Approval

The study was approved by the Institutional Review Board of the City of Hope, approval number 16099.

P27 Tumor mutational burden assessment on FFPE samples using a targeted next-generation sequencing assay

Ruchi Chaudhary, PhD1, Dinesh Cyanam, PhD2, Vinay Mittal2, Charles Scafe2, Warren Tom, PhD2, Janice Au- Young2, Seth Sadis2, Fiona Hyland2
1Thermofisher Scientific, South San Francisco, CA; 2Thermo Fisher Scientific, South San Francisco, CA
Correspondence: Ruchi Chaudhary (ruchi.chaudhary@thermofisher.com)

Background

Recently, high Tumor Mutational Burden (TMB) was associated with significantly longer progression-free survival from immune checkpoint blockade combination therapy in NSCLC. Although TMB was originally determined by whole exome sequencing (WES) of matched tumor-normal samples, the high input requirement, complex bioinformatics, and long turn-around time makes this approach impractical for routine testing. Herein, we develop a targeted amplicon-based panel for computing TMB and detecting important variants from FFPE research samples.

Methods

A targeted panel was designed that included 409 key cancer genes covering 1.7 Mb of genomic region. Utilizing Ion AmpliSeq multiplex PCR chemistry, the workflow required only 20 ng of input DNA. The assay enabled a 2.5-day turn-around time from sample to report. The workflow enabled < 60 min of hands-on time for automated library preparation and templating on a batch of 4 samples. Sequencing was performed on Ion GeneStudio S5 System at sufficient coverage depth (~1200x) to support accurate variant detection with an analysis pipeline containing optimized variant calling parameters. A tumor only informatics workflow was developed that removed germline variants present in population databases. Two cell line samples and nine FFPE samples were analyzed by the tumor only workflow. Matched tumor-normal samples were analyzed by WES and the tumor samples were independently analyzed using the targeted TMB panel.

Results

An in-silico analysis using 10,000 exomes from the TCGA MC3 project demonstrated the panel could support high sensitivity (≥85%) and PPV (≥90%) necessary to stratify high and low mutation burden samples. TMB estimates on a normal diploid cell line (NA12878) was < 1 TMB for all 8 replicates. In a cancer cell line (HCC1143; expected TMB 8.33 mutations/Mb), the average TMB for 4 replicates was 6.11 (SD 0.43). TMB estimates obtained with the tumor only workflow using the targeted panel had high concordance (r2 = 0.87) with the TMB values obtained from the matched tumor/normal analysis using WES. For two samples with highest TMB by both assays, the assay detected loss of function mutations in MSH2 and TP53 genes. The informatics pipeline identified mutation signatures consistent with specific mechanisms such as UV and tobacco damage, and detected samples impacted by FFPE processing.

Conclusions

A simple workflow has been developed on the Ion Torrent sequencing platform to estimate TMB from FFPE and fresh frozen tumor research samples. This solution will advance research in immuno-oncology.

P28 Detection and validation of cancer immunotherapy biomarkers in blood and urine-based liquid biopsy

Simo Zhang2, Shidong Jia2, Amy Wang2, Chen Xie2
1Predicine, Inc., Hayward, CA, USA; 2Predicine, Hayward, CA, USA
Correspondence: Amy Wang (amy.wang@predicine.com)

Background

Immunotherapy response varies widely, making it difficult for physicians to know whether immunotherapy will be effective for a given patient. Indeed, ~80% or more patients with cancer fail to respond to checkpoint inhibitor immunotherapy. In addition to PD-L1IHC staining, recent studies reported that patients with deleterious mutations in mismatch repair (MMR) genes, high tumor mutation burden (TMB) or microsatellite instability (MSI) are also associated with better clinical response. As tissue biopsy represents a practical challenge due to its insufficient quantity or lack of access, noninvasive molecular has emerged as an efficient complementary test and attracted increasing attention in clinical development of cancerimmunotherapy. With Predicine’s gene RARDAR technology, we developed a blood-based PredicinePLUS NGS panel to capture genomic alterations in 180 cancer genes including tumor mutation burden (TMB) and microsatellite instability (MSI). Technical validation was performed to evaluate assay sensitivity, specificity and accuracy using reference samples with known genetic profiling. The panel has been further tested using tissue biopsy and plasma samples from cancer patients. The development of PredicinePLUS panel offer a comprehensive solution to stratify and monitor cancer patients who may benefit from cancer immunotherapy.

Methods

Nucleic acids were extracted from plasma samples and tested for DNA based SNV, CNV and gene rearrangement by proprietary pipeline.

Results

Mutation detection at DNA level can go down to 0.1% AF. At 0.25% expected AF, 94.4% SNVs were detected; at 0.1% expected AF, 78.6% SNVs were detected.High linearity was observed from detected and expected copy number from spiked-in cell lines; clinical validation of HER2 amplification in breast cancerHigh correlation between panel-TMB and TMB from WES on 14 cell lines. Consistency of panel-TMB at AF =0.5% or above in a series of dilution of reference materialsHigh correlation of panel-TMB with WES in a public study (Rizvi et al. Science 2015). D. high panel-TMB showed favorable PD-1 response in the public study.

Conclusions

A non-invasive PredicinePLUS NGS test was developed to support cancer immunotherapy clinical studies. A cfRNA-based PD-L1 mRNA assay was developed to monitor PD-L1 mRNA gene expression in circulation.

P29 Ensemble computational intelligence reveals novel molecular signatures of cancer biology and pan-cancer survival

Richard Williams, Thomas Chittenden, Nicholas Cilfone, x, Jeffrey Gulcher
Wuxi NextCODE, Cambridge, MA, USA
Correspondence: Thomas Chittenden (tchittenden@wuxinextcode,com)

Background

Next-generation sequencing has significantly advanced our understanding of cancer biology by providing a unique genomic perspective of the molecular states of human disease. However, combining multiple ‘omics’ measurements into biologically-relevant statistical computing frameworks to define causal molecular underpinnings of disease remains a significant challenge.

Methods

To address these issues, we developed novel feature learning approaches that enhance quantitative assessment of annotated tissues from The Cancer Genome Atlas. Our a priori biological-knowledge and data-driven network-based approaches improve performance and interpretability of both deep learning and probabilistic programming strategies.

Results

Herein, we demonstrate the utility of collapsing molecular signals, from five different -omics platforms, into integrated metagenes that are highly informative across roughly 8,200 tumors, encompassing 22 cancer types. We identified multiple immune related genes and pathways comparing cancer sub-types (e.g. CCR1, IFNA4, CD34, IL25 – kidney renal clear cell carcinoma vs. kidney renal papillary cell carcinoma), between 22 diverse cancer types (e.g. IL-20 and tumor necrosis factor production nested in negative regulation of cellular metabolic process), and associated with overall patient survival (e.g. FCGR2A, IFNE, TGFB1, IL23A, CD80 and type I interferon signaling pathway nested in cell proliferation).

Conclusions

Our results demonstrate the potential of deep learning methodologies to help revolutionize the analysis and interpretation of multi-omics data, how to identify more complex disease etiology than previous methods, and how to hypothesize putative ‘network driver genes’ of disease state and progression. Taken together, these aspects allow researchers to generate better novel hypotheses of therapeutic targets or diagnostic biomarkers.

P30 Tracking the cancer immune response using neural network deep learning of serial inflammatory marker data for forecasting timing of therapy

Brendon Coventry, MD PhD, Mohsen Dorraki, MSc, BS, Anahita Fouladzadeh, BSc, Andrew Alison, Derek Abbott, PhD
University of Adelaide, Adelaide, SA, Australia
Correspondence: Brendon Coventry (brendon.coventry@adelaide.edu.au)

Background

The immune response in advanced cancer patients is not static but fluctuates under homeostatic control around a mean level of inflammation indicative of the anti-cancer response occurring in the patient. Immunotherapy given during immune activation (as opposed to immune inhibition) might be expected to better induce stronger anti-cancer responses, therefore timing is likely to be important. In multiple studies, the key inflammatory marker C-reactive Protein (CRP) has been widely associated with cancer survival; predicting cancer risk; a bio-maker for tumour recurrence; as a marker in oncology for prognosis; and as a reliable tool for making critical treatment decisions for several cancer types. Since CRP is biomarker of immune system activity, and CRP concentrations exhibit low values in healthy subjects, the ability to forecast CRP trends might potentially guide clinical decisions in cancer therapies based on the inflammatory state existing in the patient at the precise time of treatment.

Methods

We investigated time-series analyses of our previous data sets from advanced melanoma and other ovarian cancer patient data using (i) Periodogram analysis and (ii) Recurrent Neural Networks (RNNs) using Long Short-Term Memory (LSTM)-based, approaches to predict the future state in a C-reactive protein (CRP) time-series in cancer patients. Deep learning provided CRP time-series forecasting.

Results

Using Periodogram methods, the time-series used in [1,2] did not contain enough data points in the measured time period to conclude whether the CRP data was periodic or not, particularly for the previously hypothesised period of seven days. Moreover, the study [3] provided a prescription for the minimum data sampling rate required for improved testing of a periodic CRP (or other biomarker) signal hypothesis. We abandoned this method in favour of investigating RNN approaches. The distribution of CRP was highly skewed, so a log(.) representation that is more symmetric and less skewed is recommended for CRP estimation. We challenged our data interpretation [1] and other data [4] for periodicity in either serial daily CRP measurements in melanoma patients, or in gynaecological cancer patients [2] with less frequent measurements.

Conclusions

Deep learning and other of machine learning-based approaches for biomedical signal analysis can be used to predict trends in C-reactive protein time-series, with greater accuracy than periodogram approaches. Deep LSTM RNN with 200 layers achieves the lowest prediction error. These approaches offer useful avenues for bio-marker monitoring. Forecasting CRP trends can provide potentially valuable information for guiding clinical decision-making for more accurately timing of therapies, including immunotherapies.

Acknowledgements

Australian Melanoma Research Foundation & Cancer Council SA/ SA Government/ SAHMRI for supporting the Trial Data collection and CRP monitoring,

References

1. Coventry BJ, Ashdown ML. Quinn MA, Markovic SN, Yatomi-Clarke SL, Robinson AP. CRP identifies homeostatic immune oscillations in cancer patients: a potential treatment targeting tool? J Transl Med. 2002; vol. 7, no. 102.

2. Madondo MT, Tuyaerts S, Turnbull B, Vanderstraeten A, Kohrt H, Narasimhan B, Amant F, Quinn M, and Plebanski M. Variability in CRP, regulatory T cells and effector T cells over time in gynaecological cancer patients: a study of potential oscillatory behaviour and correlation. Journal of Translational Medicine. 2014; 12:179.

3. Dorraki M, Fouladzadeh A, Salamon SJ, Allison A, Coventry BJ, and Abbott D. On detection of periodicity in C- reactive protein (CRP) levels, Sci. Rep., in press.

4. Leontovich AA, Dronca RS, Suman VJ, Ashdown ML, Nevala WK, Thompson MA, Robinson A, Kottschade LA, Kaur JS, McWilliams RR, Ivanov LV, Croghan GA, Markovic SN. Fluctuation of systemic immunity in melanoma and implications for timing of therapy. Front Biosci (Elite Ed). 2012 Jan 1;4:958-75.

P31 Prevalence of high microsatellite instability in cancer patients in the real world

Razvan Cristescu, PhD1, Kai-Li Liaw, PhD1, Scott Pruitt, MD, PhD1, Mark Ayers, PhD1, Jianda Yuan, MD, PhD1, Thao Vo, MD1, Senaka Peter1, Andrew Joe, MD PhD1, Darcy Hille1, Sun Young Rha2, Torben Steiniche3, Andrey Loboda, PhD1
1Merck & Co., Inc., Kenilworth, NJ, USA; 2Yonsei Cancer Center, Seoul, Korea, Republic of; 3Aarhus University Hospital, Aarhus, Denmark
Correspondence: Razvan Cristescu (razvan_cristescu@merck.com)

Background

High microsatellite instability (MSI-H) cancers are vulnerable to immunotherapies targeting the PD-1/PD-L1 pathway. PD-1–blocking mAb pembrolizumab was recently approved by the FDA for the treatment of MSI-H cancer, regardless of tumor histology or location. This analysis evaluated MSI-H prevalence by cancer type and stage using several real-world observational studies to examine the potential patient populations that might benefit from treatment.

Methods

Four data sources, including 1 database and 3 retrospective observational molecular epidemiology studies, were evaluated. Microarray analysis of loss of MLH1 gene expression was used as a surrogate for MSI-H status (<2.35 on log10 scale in quantile-normalized data) to assess prevalence of MSI-H in the Moffitt Cancer Center database of ~16,000 archived tumors. Results were compared with those of 3 epidemiologic studies that evaluated MSI-H status using a PCR-based assay on archival tissue of gastric, ovarian, endometrial, cervical, and 8 rare cancers, most (72%- 100%) from patients with stage III-IV disease.

Results

In the Moffitt database, for the following cancers, MSI-H prevalence by MLH1 expression loss was: endometrial (138/664, 20.8%), sarcoma (4/38, 10.5%), gastric (4/41, 9.8%), colorectal (194/2251, 8.6%), esophageal (2/49, 4.1%), kidney (2/60, 3.3%), cervical (2/72, 2.8%), melanoma (1/41, 2.4%), prostate (1/88, 1.1%), lung (22/2064, 1.1%.), ovarian (5/523, 1.0%). MSI-H prevalence was generally higher in stage I-II than in stage III-IV. In epidemiologic studies of advanced disease, MSI-H prevalence by PCR was assessed in Korean patients with gastric cancer (0/103, 0%) and in patients with ovarian (1/40, 2.5%), endometrial (7/49, 14.3%), cervical (1/44, 2.3%), and rare cancers (0/305, 0%) in Denmark.

Conclusions

Using MLH1 expression loss as a surrogate, a large database with comprehensive molecular data of cancer patients provided MSI-H prevalence estimates in many cancer types not usually tested for MSI-H. The results were generally comparable with those of epidemiologic studies that used PCR-based MSI testing. These real-world data found MSI-H tumors in multiple cancers and may help to identify patients who can potentially benefit from treatment with pembrolizumab.

Ethics Approval

The abstract is based upon 1 database and 3 retrospective observational molecular epidemiology studies; no ethics approval was required.

P32 Monitoring of M-MDSC vs. G-MDSC in clinical studies – which is more important? (monocytic versus granulocytic myeloid derived suppressor cells)

Henry Hepburne-Scott, PhD, Phoebe Bonner-Ferraby
Serametrix, Carlsbad, CA, USA
Correspondence: Phoebe Bonner-Ferraby (pbf@serametrix.com)

Background

It is now almost universal good practice to monitor MDSC during clinical trials for novel checkpoint inhibitors and combination therapies. This is because MDSC protect tumors from anti-PD-1 and other such drugs by infiltrating tumors and providing localized immunosuppression. MDSC monitoring is also increasingly important in combination studies to help assess the efficacy of putative anti-MDSC agents, such as HDAC inhibitors.However, MDSC is an elusive biomarker, a “catch-all” category of immature immune cells and immunologists rarely agree on a definitive phenotype. Even having navigated the minefield of markers there is still the question of whether monocytic or granulocytic MDSC are dominant in protecting tumors from checkpoint-mediated de-repression of cytotoxic T cells. Whilst a biomarker assay for M-MDSC is already well established and widely used, a reliable assay for G-MDSC has proved more challenging. Here we report the development and application of a novel assay for G-MDSC for use in I-O clinical studies.

Methods

A flow cytometry panel consisting of Lineage Cocktail, CD14, CD33, CD15 and HLA-DR was developed, validated and used to identify G-MDSC in peripheral whole blood samples drawn from cancer patients. To ensure operator independent gating of the continuous HLA-DR marker a computational algorithm was used to determine sample- specific thresholds for gating. Data for G-MDSC were compared with same-sample data for M-MDSC.

Results

The assay was successful in identifying G-MDSC in peripheral Whole Blood samples collected in 5mL Cyto-Chex BCT tubes. A single blood draw yielded sufficient material to measure both M-MDSC and G-MDSC from the same sample. However, during assay validation the stability of the G-MDSC was found to be significantly less than the stability of M-MDSC (logistical implications of this will be discussed). There was poor correlation between the two cell types: samples that were rich in M-MDSC did not always have high levels of G-MDSC and vice versa.

Conclusions

Whilst MDSC monitoring is a valuable part of some immune-oncology clinical development it may not be sufficient to rely on measuring the monocytic phenotype alone for all studies. The poor correlation between M-MDSC and G-MDSC observed in this study suggests that both types should be included if a biomarker program is to be truly effective. The use of a combo-MDSC assay will enable a fuller understanding of the mechanisms of resistance to checkpoint blockade.

P33 DNA damage detected by localized γH2AX is associated with elevated TILs and PD-L1 expression in human colorectal carcinomas

Shruti Desai, PhD, Parker Sulkowski, Aravind Kalathil, Ranjini Sundaram, Ila Datar, PHD, Charles Fuchs, MD, MPH, Patricia LoRusso, DO, Ranjit Bindra, Kurt A. Schalper, MD, PhD
Yale University School of Medicine, New Haven, CT, USA
Correspondence: Kurt A. Schalper (kurt.schalper@yale.edu)

Background

Cancer cells accumulate genomic aberrations due to the frequent combination of increased DNA damage and decreased DNA repair capacity. H2AX is a histone component of nucleosomes and its phosphorylation on Serine139 (γH2AX) is the first step in recruiting repair proteins upon DNA damage. DNA alterations can produce mutant neoantigens that are recognized as non-self and presented by the HLA system to trigger anti-tumor immune responses and mediate sensitivity to immune checkpoint blockers. To date, evidence for the association between active DNA damage in cancer cells and adaptive anti-tumor immune responses in intact tumors specimens remains elusive.

Methods

Using γirradiated cell line preparations and expression controls, we standardized a multiplexed quantitative immunofluorescence (mQIF) panel for simultaneous and localized measurement of DAPI (all cells), cytokeratin for tumor epithelial cells (AE1/AE3, DAKO), γH2AX to map active DNA damage (JBW301, Millipore), CD3 for T- lymphocytes (Rabbit polyclonal, DAKO) and PD-L1 (E1L3N, CST) in formalin-fixed paraffin-embedded (FFPE) tissue samples. We then used the assay to study 265 stage I-IV colorectal carcinomas (CRCs) from Yale represented in tissue microarray format. We analyzed the level of the targets, their association and correlation with major clinicopathologic variables and survival.

Results

The levels of γH2AX (but not total H2AX) were significantly higher in FFPE preparations of γirradiated HEK293 cells than in control/untreated cells or in morphologically normal human tissues. The radiation-induced γH2AX increase was abrogated in HAP1 cells with targeted deletion of the H2AX gene. Detectable γH2AX protein was found in 170 (64%) of CRCs with nuclear staining pattern and predominant expression in cytokeratin-positive tumor cells. Elevated tumor γH2AX was significantly associated with increased CD3+ tumor infiltrating lymphocytes (TILs) and PD-L1 protein expression (P<0.05). Elevated CD3, PD-L1 and γH2AX were associated with lower tumor stage and better overall survival in the cohort. No significant association was seen between the markers and age, gender or smoking status.

Conclusions

Tumor DNA damage as measured by nuclear γH2AX protein expression occurs in 64% of CRCs and is associated with increased anti-tumor immune responses and better prognosis. TILs and PD-L1 are prognostic in CRC. Our results support targeting DNA repair deficiency pathways in combination with immune stimulatory agents as therapeutic strategy in a proportion of CRCs.

Acknowledgements

Research supported by a Stand Up to Cancer Colorectal Cancer Dream Team Translational Research Grant (Grant Number SU2C-AACR-DT22-17)

P34 Clinical application of urinary cell free DNA as a marker for cancer

Jeffrey Ding, Debin Sun
Admera Medical Technology Corp., Suzhou, China
Correspondence: Jeffrey Ding (jeffrey.w.ding@gmail.com)

Background

The role of circulating cell free DNA holds great promise for individualized medicine for cancer. As it carries information on DNA from cells exfoliated in urine and from circulation, urinary cell free DNA (UcfDNA) is believed to have the potential of being a useful and ultra-noninvasive tool for cancer screening, diagnosis, prognosis, and monitoring of cancer progression and therapeutic effect [1]. However, compared with the widely studied cell- free DNA in blood, less is known about the role of UcfDNA. The presence of UcfDNA signals from tumor has remained controversial, possibly due to the lack of appropriate method and technology to robustly extract and detect the potentially highly degraded UcfDNA.

Methods

We developed and optimized UcfDNA extraction method from urine samples, and compared UcfDNA quantity and quality regarding the urine sampling time and the storage condition. As a proof of concept, we processed urine samples from pregnant women during their second trimester and used real-time quantitative PCR (qPCR) assays to successfully detect UcfDNA from fetal in maternal urine. We subsequently collected urine samples from lung cancer patients and applied the developed methods to extract UcfDNA. The qPCR assay with the blocker-based enrichment method was developed and used for EGFR mutation detection in UcfDNA (Image 1,2). We also explored the possible application of UcfDNA for the microsatellite instability(MSI) testing.

Results

We successfully detected EGFR mutations in UcfDNA which also found in the corresponding tumor tissue samples, demonstrating the UcfDNA extraction and genotyping was possible from urine samples of cancer patients (Image 3).

Conclusions

We have confirmed the existence of UcfDNA from tumor, and have shown that UcfDNA could be used for noninvasive cancer genetic test and cancer research.

References

1. Lu T, Li J. Clinical applications of urinary cell-free DNA in cancer: current insights and promising future. Am J Cancer Res. 2017 Nov 1;7(11):2318-2332.
Image 1 (abstract P34).
Image 1 (abstract P34).

See text for description.

Image 2 (abstract P34).
Image 2 (abstract P34).

See text for description.

Image 3 (abstract P34).
Image 3 (abstract P34).

See text for description.

P35 Development of biomarkers to assess adenosine generation & activity in support of clinical trials conducted with the adenosine receptor antagonist AB928

Daniel DiRenzo, PhD, Joanne Tan, PhD, Devika Ashok, PhD, Amy Anderson, PhD, Jenna Jeffrey, PhD, Lisa Seitz, MSc, Manmohan Leleti, PhD, Steve Young, PhD, Jay Powers, PhD, Matthew J. Walters, PhD
Arcus Biosciences, Hayward, CA, USA
Correspondence: Matthew J. Walters (mwalters@arcusbio.com)

Background

The high levels of adenosine (ADO) found in the tumor microenvironment have been shown to inhibit immune responses through activation of the A2aR and A2bR receptors on immune cells. The extracellular enzymes ecto-5’- nucleotidase (CD73) and tissue non-specific alkaline phosphatase (TNAP) catalyze the extracellular conversion of adenosine monophosphate (AMP) into ADO. We have previously shown that AB928, a dual A2aR/ A2bR antagonist, blocks the immunosuppressive effects of ADO in human cell culture systems and in mouse syngeneic tumor models. Herein, we describe the development of assays to measure the expression and activity of adenosine- generating enzymes in human tumor samples and peripheral blood. These assays are being implemented in ongoing clinical trials with AB928, to identify tumor types and patients most sensitive to adenosine receptor antagonism.

Methods

Gene expression data were extracted from The Cancer Genome Atlas (TCGA). To correlate protein and gene expression levels, immunohistochemistry (IHC) and NanoString analyses were performed on serial sections of formalin fixed paraffin embedded (FFPE) tumor tissue. Circulating levels of CD73 were quantified with an in-house developed CD73 ELISA and total AMP-ase enzymatic activity in plasma was determined using an AMP-Glo assay.

Results

TCGA gene expression analysis identified non-small cell lung, renal clear cell, triple-negative breast, ovarian, colorectal, and gastro-esophageal cancers as tumors that express high levels of adenosine processing enzymes. Specifically, combined CD73 and TNAP levels were highest in lung adenocarcinoma (CD73 = 4.024; TNAP = 4.753) whereas colorectal (CD73 = 4.101; TNAP = 0.7493) and ovarian (CD73 = 2.173; TNAP = 5.288) cancers were heavily biased towards CD73 or TNAP, respectively. IHC on human FFPE tumor microarrays demonstrated that NSCLC had the highest CD73 protein levels with a stark contrast between adenocarcinoma (70.5 +/- 17.3 μM2) and squamous cell carcinoma (7.3 +/- 1.9 μM2). In contrast, prostate cancer had low CD73 gene expression (median TPM = 1.82) and protein levels (0.99 +/- 0.07 μM2). Overall, there was strong agreement between TCGA data and IHC (R2 = 0.793) suggesting that transcript levels of CD73 broadly predict local protein levels. In addition, we are in the process of determining the relationship between tumoral CD73 levels and, using different methodologies, peripheral CD73 protein levels and enzymatic activity.

Conclusions

Collectively, these assays provide a detailed picture of the capacity of individual human tumors to generate adenosine and should enable the correlation of this information with peripheral activity/levels of adenosine- generating enzymes and potentially the clinical benefit of AB928.

P36 Defining the expression of Programmed Death-Ligand 2 in high grade glioma tumor microenvironment

Gifty Dominah1, Victoria Sanchez, BS1, John Lynes, MD1, Nicholas Adamstein2, Arnold Obungu, BS, BA3, Xiang Wang, MS1, Nancy Edwards, BA1, Edjah K. Nduom, MD1
1NINDS/NIH, Bethesda, MD, USA; 2Columbia University College of P&S, New York, NY, USA; 3Indiana University School of Medicine, Indianapolis, IN, USA
Correspondence: Edjah K. Nduom (edjah.nduom@nih.gov)

Background

Checkpoint blockade with anti-programmed death-1 (PD-1) therapy has been demonstrated as a promising treatment for many systemic cancers. Tumor programmed death-ligand 1 (PD-L1) expression has been shown to increase the response to anti-PD-1 immunotherapy in many tumor types. We have recently validated PD-L1 expression as a negative prognostic factor in high grade gliomas (HGG). However, PD-1 has two ligands, PD-L1 and PD-L2, and PD-L2 expression has not been characterized in HGG tissue. Accordingly, the potential prognostic and/or predictive value of PD-L2 has not yet been assessed. This study aims to establish reliable means to detect PD-L2 expression in HGG patient samples, validate this expression and define its clinical relevance.

Methods

The PD-L2 antibody clone 24F.10C12 was optimized for staining by immunofluorescence (IF) in PD-L2-plasmid transfected HEK293 cells. Further antibody validation was performed via immunohistochemistry (IHC), IF staining, and RNAscope in situ hybridization (ISH) using normal brain slides and human heart tissue as a positive control. After antibody validation, immunohistochemistry and RNAScope ISH was used to evaluate PD-L2 protein and mRNA expression in paraffin-embedded HGG slides.

Results

PD-L2 expression using clone 24F.10C12 was found in transfected HEK293 cells, but not in untransfected cells. By IHC, IF, and ISH, PD-L2 was expressed adjacent to blood vessels in normal brain and HGG slides. There was heterogeneity between HGG patient tumor samples as some tissues also had perinuclear expression of PD-L2. Moreover, HGG tissues that expressed low grades of PD-L1 showed high PD-L2 expression grades per IHC suggesting an inverse relationship between the two.

Conclusions

PD-L2 expression in HGG has a distinct cellular pattern and distribution, separate from the expression of PD-L1. PD-L2 expression should be further characterized in these tissues to determine the potential prognostic or predictive value of this marker for immune therapy of HGG patients.

References

1. Pratt D, Dominah G, Lobel G, Obungu A, Lynes J, Sanchez V, ... Maric D. Programmed death ligand 1 ss a negative prognostic marker in recurrent isocitrate dehydrogenase-wildtype glioblastoma. neurosurgery. 2018.

P37 Multiplexed biomarker quantification to assay and characterize T-cell activation

Shilan Dong, MS, Jason Cahoon, BS, Rachit Ohri, PhD
Enable Life Sciences LLC, Worcester, MA, USA
Correspondence: Rachit Ohri (rachit@enablelifesciences.com)

Background

Multiplexed biomarker based characterization has proven effective for disease diagnosis and therapeutics development [1]. We adapted a similar multiplexed biomarker strategy for characterizing T-cell activation, with the goal of developing a standardized cell-culture assay for immuno-oncology applications. The panel of chosen biomarkers expressed by T-cells included: [a] IL-2, important for downstream immune activation [b] TNF⍺, important in acute phase immune cell activation, differentiation and migration [c] IL-10, with the role of tumor-specific immune surveillance and mitigating pathologic inflammation, and [d] IFNγ, with the role of tumor- protection through immune cell activation including dendritic cells.

Methods

Activation of T-cells was induced by ionomycin and PMA (phorbol 12-myristate 13-acetate) [2] (Sigma-Aldrich, St. Louis). Jurkat cells [UMass, Worcester at 1X106 cells/ml] and primary CD3-positive pan T cells [iXCells Biotechnologies, San Diego at 8X105 cells/ml] were cultured in 24 well plates in RPMI 1640 media (10% FBS and 1% Penn-Strep). 24 h protein-level expression (in the cell supernatant) of IL-2, IFNγ, TNFα, and IL-10 was quantified in response to a wide concentration range for both activators (250 - 2000 ng/ml for ionomycin and 10 - 100 ng/ml for PMA). For the specific combination of 2000 ng/ml ionomycin and 50 ng/ml PMA, time-course expression levels were also determined over a 0h - 56h period. Biomarker quantification was pursued using Luminex methodology.

Results

Our multiplexed biomarker data indicates significant overall consistency of expression trends comparing one biomarker to another, though minor differences does occur (eg. the optimal time point for activation), even though comparative concentration levels varied. Additionally, consistency was also observed between the activation profiles of primary CD3+ pan T-cells and the Jurkat T-cell line for [a] the most synergistic combination of ionomycin and PMA for T-cell activation [i.e. 2000 ng/ml ionomycin and 50 ng/ml PMA] (IL-2 expression levels statistically higher i.e. p<0.05 or equal compared to any other combinations of PMA and ionomycin) (Figures 1A, 1B), as well as [b] the time-course of T-cell activation [i.e. biomarker plateauing or high expression most commonly in the 8h - 48h window] (Figures 2A-D). Despite overall consistency, nuanced differences were observed between primary T- cells and Jurkat cells (e.g. timing of the peaking or plateauing of individual biomarker expression), which may further vary between different sources of primary T-cells (to be evaluated).

Conclusions

In conclusion, our results provide the basis of a robust, standardized, sensitive and efficient assay for pan T-cell activation.

Acknowledgements

The authors acknowledge and thank MBI (Massachusetts Biomedical Initiatives), WPI (Worcester Polytechnic Institute), and University of Massachusetts Medical School for the use of their facilities.

References

1. Wagner JA. Strategic approach to fit-for-purpose biomarkers in drug development. Annu. Rev. Pharmacol. Toxicol. 2008; 48:631-651. 2. Manger B, Weiss A, Weyand C, Goronzy J, Stobo JD. T cell activation: differences in the signals required for IL 2 production by nonactivated and activated T cells. The Journal of Immunology. 1985; 135(6):3669-3673.
Fig. 1 (abstract P37).
Fig. 1 (abstract P37).

See text for description.

Fig. 2 (abstract P37).
Fig. 2 (abstract P37).

See text for description.

P38 Highly multiplex spatial immuno-profiling in FFPE tumor tissue with InSituPlex technology

Abdul Mohammed, PhD, Gourab Chatterjee, PhD, Kevin Hwang, PhD, Julie Xia, Amanda Bares, PhD, Michael Murphy, Eloise Wheeler, Armen Changelian, Katir Patel, PhD, Bonnie Phillips, PhD, Sean Downing, PhD, Mael Manesse, PhD
Ultivue, Cambridge, MA, USA
Correspondence: Sean Downing (sean.downing@ultivue.com)

Background

Innovative and future translational research tools are key to enabling the full impact of personalized medicine. Current pathology methods rely on chromogenic and H&E staining with low multiplexing capabilities, limiting the depth of information obtained from a single tissue sample. Fluorescence-based tissue staining and analysis enable quantification and higher multiplexing; however, current multiplex immunohistochemistry (mIHC) technologies not only compromise throughput and potentially damage the sample with each round of staining, but also require post- acquisition spectral unmixing. Using InSituPlex technology, a highly multiplexed assay (8-plex) can be carried out in a single work day, with maximum tissue preservation and no spectral unmixing.

Methods

InSituPlex technology was used to carry out high-multiplexed immuno-profiling on deidentified FFPE tissue sections. Samples included human tonsil as well as multiple tumor types in skin, lung, and colon. A panel of eight different markers was developed, including CD3, CD8, CD45RO, CD68, PD-1, PD-L1, FoxP3, as well as pan- cytokeratin and Sox10 as tumor markers. Staining of the eight markers was performed in a single run on individual slides, following a manual or automated protocol on the Leica BondRX. Images were acquired using commercially available fluorescence slide scanning platforms, including the Zeiss Axio Scan.Z1, without the need for linear unmixing. All images were analyzed using IndicaLabs HALO.

Results

Images from the eight-plex assay on tonsil tissue were compared to individually stained reference samples to confirm the specificity of the markers in the multiplex assay. In tumor samples, abundance of immune and tumor cells was characterized through cell counting. In addition, the different expression levels of PD-L1 in immune and tumor cell types were recorded over the entire section. Phenotyping and spatial distribution analysis was carried out to identify cytotoxic T-cells, memory T-cells, exhausted T-cells, regulatory T-cells, and macrophages on single sections. Reproducibility of the assay was also assessed using the data points from serial sections.

Conclusions

InSituPlex technology enables the staining of eight different immune and tumor markers on single FFPE sections, with a streamlined workflow and high reproducibility. Resulting images were then used to perform phenotyping of multiple subsets of T-cell population, macrophages, and tumor cells.

P39 PD-L1 expression on tumor versus antigen presenting cells investigated with multiplexed IHC using UltiMapper™ I/O assays

Amy Zhang, MS, Alexis Wong, PhD, Max Rubinstein, Laura Sciarra, PhD, Chakib Boussahmain, BS, Bonnie Phillips, PhD, Katir Patel, PhD, Sean Downing, PhD, Stephanie Hennek, PhD
Ultivue, Cambridge, MA, USA
Correspondence: Sean Downing (sean.downing@ultivue.com)

Background

In the field of immuno-oncology, there is great promise for improving outcomes by identifying meaningful biomarkers within the tumor microenvironment. The TME is highly heterogeneous, requiring characterization of what cell types are present and what cellular interactions are taking place. Multiplexed IHC is a promising approach to profile cell types and map interactions to identify useful biomarkers. Ultivue has developed UltiMapper assays that provide numerous advantages over other multiplexed IHC approaches including high multiplexing in situ, sample preservation, streamlined workflows, and versatile implementation. The checkpoint marker PD-L1 has become a common marker in immune-oncology, but the usefulness of this marker in insolation has been questioned. In this study, we investigate the expression of PD-L1 on both tumor cells and antigen presenting cells to better understand which cell phenotypes may be important biomarkers.

Methods

Multiplexed immunofluorescence was carried out using UltiMapper assays on multiple deidentified FFPE tissue samples, including lung, melanoma, colon, and breast. Each sample was stained and analyzed for the UltiMapper I/O PD-L1 and APC panel using serial sections. The UltiMapper I/O PD-L1 panel included the markers CD8, CD68, PD-L1, pan-cytokeratin, and Sox10. The UltiMapper I/O APC panel included markers CD11c, CD20, CD68, CD163, and MHCII. Staining was performed manually or using the Leica Bond Rx™ autostainer. Imaging was performed on various tissue scanners including the Zeiss Axio Scan.Z1, and image analysis was performed using HALO from Indica Labs.

Results

Cell phenotyping was carried out to measure the abundance of PD-L1 co-staining on macrophages, dendritic cells, B-cells, cytotoxic T-cells, and tumor cells. In tumor samples, PD-L1 expression was observed on both immune cells and tumor cells with a range of different expression levels and percent positive cells. Spatial analysis was employed to measure the distances between immune cells of differing phenotypes and tumor cells, which could be used to classify tumor samples hot or cold.

Conclusions

Multiplexed IHC is necessary to understand complex cancer biology and the mechanisms by which the immune system is activated or suppressed. The UltiMapper approach is efficient and easy to implement to achieve high quality multiplexed data on a range of tissue types. The UltiMapper I/O PD-L1 panel is complimented by the UltiMapper I/O APC panel to characterize PD-L1 expression on antigen presenting cells and tumor cells. The abundance and location of PD-L1 positive APCs may be a preferred biomarker over tumor cell PD-L1 expression.

P40 Intra-assay and inter-assay assessment of reproducibility and quantification of UltiMapperTM I/O PD-1 and PD-L1 immuno-oncology panels for tissue multiplexing

Bonnie Phillips, PhD, Katir Patel, PhD, Courtney Hebert, Jamie Buell, Sean Downing, PhD
Ultivue, Cambridge, MA, USA
Correspondence: Sean Downing (sean.downing@ultivue.com)

Background

The field of immuno-oncology has enthusiastically adopted multiplex IHC techniques to establish the spatial relationships between various immune cells in tumor biology in context. Multiplexing enables researchers to gain a deeper understanding and insight into the tumor microenvironment. Unfortunately, many of the multiplexing technologies currently utilized in the immuno-oncology field face a number of challenges, specifically in generating highly robust, reproducible, and easily quantifiable data sets. Ultivue’s UltiMapper I/O PD-1 and PD-L1 I/O kits that utilize InSituPlex (ISP) technology, a new method of multiplexed immunohistochemistry (IHC) that utilizes streamlined single antigen retrieval, staining, elongation, and detection steps allowing for the completion of the assay < 5hr. Here we assess these kits for intra-assay and inter-assay reproducibility and quantification.

Methods

Intra-assay reproducibility and quantification was accomplished by manually staining 5 serial sections from three different tissue types (tonsil, melanoma, NSCLC) with one set for each of the UltiMapper PD-1 (CD3, CD45RO, PD-1, CK/Sox10) and PD-L1 (CD8, CD68, PD-L1, CK/Sox10) I/O kits. Inter-assay assessment was determined by staining a single slide from a set of serial sections of each tissue type described above once a week for 5 consecutive weeks. Images were acquired using the Zeiss Axio Scan.Z1, without the need for linear unmixing allowing for direct whole slide imaging. Analysis was accomplished using IndicaLabs HALO software. Coefficient of variations (CV) were calculated based on resulting data.

Results

Analysis of intra-assay serial section images revealed that cell counts from section to section were within a CV of <10% across all markers, in all tissues, for both the UltiMapper PD-1 and PD-L1 kits. This included total cell counts, top 10% brightest cells, and all quartiles of cell counts based on fluorescence signal intensity. Similar results were seen for inter-assay comparisons over 5 weeks for both kits (<10% CV).

Conclusions

The results presented here indicate that InSituPlex technology is potentially much more reproducible than other tissue multiplexing techniques currently available, such as TSA. Histological standards for coefficient of variations in IHC based assays typically are <15%. Data presented here falls well within that standard indicating the potential for future translational applications. In conclusion, InSituPlex is a highly reproducible and quantifiable multiplexing staining technology across a variety of tissue types and markers, within a single run and over time.

P41 Characterization of circulating biomarkers in subjects with NSCLC using data independent acquisition mass spectrometry reveals host immune response mechanisms

Nicholas Dupuis, PhD, Jakob Vowinckel, PhD, Daniel Heinzmann, Claudia Escher
Biognosys, Schlieren, Switzerland
Correspondence: Nicholas Dupuis (nicholas.dupuis@biognosys.com)

Background

Identification of circulating biomarkers in cancer has proven utility in applications for early detection, differential diagnosis, predicting pre-treatment response to therapy, and treatment monitoring. More recently, circulating proteomic biomarkers have been evaluated as surrogate endpoints for early indication of benefit for immunotherapies. This last application is especially relevant during immunotherapy development where the optimal endpoint, overall survival (OS), can take longer to mature. Here, we present an unbiased survey of the circulating proteome of subjects with NSCLC to identify candidate biomarkers which may have utility in multiple stages of patient care.

Methods

Unbiased, data-independent acquisition (DIA) mass spectrometry was used to analyze plasma samples from subjects with Stage III-IV non-small cell lung cancer (NSCLC, n = 15) and age matched healthy donors (n = 15), enabling simultaneous sequencing and quantification of plasma proteins. Samples were prepared for mass spectrometry and spiked with a panel of standards covering 500 plasma proteins. All samples were analyzed using 1 hour gradients on a C18 column coupled to a Thermo Scientific Q Exactive HF mass spectrometer. Data was extracted using Spectronaut (Biognosys) with a sample specific spectral library and statistical analysis was conducted to identify disease associated biomarker candidates. Pathway analysis highlights dysregulated biological functions and predicts upstream regulatory pathways.

Results

A protein library was created containing 771 unique proteins. In DIA acquisition, 462 proteins were quantified across all samples. Univariate statistical testing identified 26 dysregulated proteins (20 up-regulated and 6 down- regulated; q-value > 0.05 and log2 fold change > 0.58). Multivariate (PLS-DA) analysis identified c-reactive protein (CRP) and serum amyloid a (SAA1/SAA2), complement C9, S100A8/S100A9, and leucine rich glycoprotein 1 (LRG1) as the most significantly changed proteins across sample groups. Significantly enriched pathways include acute phase response, complement system as well as IL-12 and IL-6 signaling. Similarly, upstream activated candidate pathways included STAT3, IL-6, and EZH2.

Conclusions

26 proteins were identified as candidate biomarkers and reflect the host immune response via acute phase response signaling, innate immune response (complement system), and other proinflammatory stimuli. Several of these markers have been linked to patient outcomes and poor prognosis. Accurate monitoring these proteins offers the possibility to define surrogate, molecular based, markers with multiple modes of utility.

P42 Expanding insights into the colorectal cancer tumor proteome; unbiased protein profiling reveals multiple proteomic-based tumor subtypes

Jan Muntel, Roland Bruderer, Nicholas Dupuis, PhD, Lukas Reiter
Biognosys, Zurich, Switzerland
Correspondence: Nicholas Dupuis (nicholas.dupuis@biognosys.com)

Background

Recent approvals of microsatellite instability (MSI) and PD-L1 testing, have expanded the tools available to identify tumor characteristics which predict patient responses to immunotherapies. However, even in MSI and PD-L1 positive subgroups, not all subjects achieve a durable response and work continues to identify tumor characteristics that will further predict the likelihood of patient response. To support and advance this area of research, new tools are being developed that provide deeper and unbiased views of the tumor proteome. Here, we characterize the protein expression profiles of 95 colorectal cancer tumors (CRC) using SWATH acquisition mass spectrometry (SWATH MS) to probe tumor phenotypic characteristics.

Methods

FFPE colon tissue samples (95 cancer, 10 healthy) from subjects with colorectal cancer across seven regions of the colon: cecum (16), ascending (17), right hepatic flexor (2), left splenic flexor (5), descending (12), sigmoid (21), nonspecific (22). Proteins were extracted from the tissue, processed to peptides, and injected on a Triart C18 column (YMC) coupled to a NanoLC 425 system (SCIEX). Eluted peptides were then analyzed with a TripleTOF® 6600 system (SCIEX) operated in SWATH mode. Data were analyzed in Spectronaut Pulsar X (Biognosys) with a project specific library.

Results

Across all samples, >4,500 protein groups were quantified (approximately 3,600 per sample). Data analysis revealed a large number of proteins (~1,000) were differentially expressed in the cancer cohort, including an elevation of proteins involved in translation which is consistent with increased tumor cell proliferation. Unsupervised clustering of the data separated the healthy and the cancer cohort and revealed three main proteomic subtypes within in the cancer cohort (A, B and C) which were largely distinguished by expression of cell adhesion proteins, including neuronal growth regulator 1 (NEGR1), a potential tumor suppressor. Interestingly, hepatocyte nuclear factor 4-alpha (HNF4A), a transcription factor which is known to be elevated in CRC, was most significantly overexpressed in subtype B, which correlates with protein signatures from MSI high samples from previous studies. Additional analysis of key protein networks related to CRC and MSI high status, as well as analysis of the mismatch repair proteins MSH2 and MSH6 expression, will be presented from this work.

Conclusions

High-throughput proteomic profiling of FFPE tissues using SWATH-MS enables the deepest phenotypic characterization of tumor tissue. Through global profiling, these analyses will help improve the functional understanding of the interplay between the expression of protein networks, tumor microenvironment, and response to immune-directed therapies.

P43 The presence of exhausted CD8+ T cells identifies a subset of immunogenic ER+ breast cancer patient tumors

Colt Egelston, PhD, Christian Avalos, Diana Simons, Min Hui Lim, Peter Lee, MD
Beckman Research Institute, City of Hope, Duarte, CA, USA
Correspondence: Colt Egelston (cegelston@coh.org)

Background

Estrogen receptor positive (ER+) breast cancers are generally thought to be less immunogenic and less immune infiltrated than triple negative breast cancer (TNBC). In TNBC the presence of tumor infiltrating lymphocytes (TILs) is predictive of response to chemotherapy and associates favorably with patient survival. However, in ER+ breast cancer the relationship between T cell infiltration and disease is less clear. Expression of both PD-1 and CD39 on exhausted CD8+ T cells has been described in murine models of chronic disease and recently human carcinomas. Here we profile human breast tumors for the frequency and phenotype of exhausted CD8+ T cells and their association with a unique immunogenic tumor microenvironment.

Methods

Fresh surgical tumor specimens were obtained from consented patients. Single cell suspensions were analyzed by flow cytometry for immune phenotyping and functional assessment of cytokine production by T cells. Single cell sorted T cells were subjected for whole transcriptome RNA sequencing and bulk sorted T cells were submitted for T cell receptor repertoire sequencing by Adaptive Biotechnologies. Formalin fixed tissues were then used for multispectral immunohistochemistry and RNA transcript analysis by Nanostring.

Results

Of 35 ER+ tumors assessed, 20% were found to have exhausted PD-1+ CD39+ CD8+ T cells at a significant frequency of total CD8+ TILs. On the contrary, 60% of TNBC tumors assessed contained exhausted PD-1+ CD39+ CD8+ T cells. These T cells showed a significant reduction of IFNγ, TNFα, and IL-2 production capacity. Further phenotyping of exhausted CD8+ T cells revealed a loss of expression of both the IL-7 receptor alpha (CD127) and KLRG1, suggesting terminal differentiation of these cells. Single cell and T cell receptor sequencing revealed distinct transcriptional and clonal signatures of PD-1+ CD39+ CD8+ T cells suggesting them to be a unique population in response to tumor antigen. Immunohistochemistry of tumor tissues showed that ER+ tumors with PD- 1+ CD39+ CD8+ T cells were significantly more infiltrated with T cells and showed characteristics of ‘hot’ tumors. Finally, Nanostring analysis of RNA transcripts from tumor tissues revealed that ER+ tumors containing PD-1+ CD39+ CD8+ T cells had a significantly higher expression of gene signatures of inflammation and IFNγ signaling.

Conclusions

This work suggests that a subset of ER+ breast cancer patients have inflamed, highly lymphocyte infiltrated tumors. These tumors are associated with the presence of exhausted CD8+ T cells, suggesting a tumor antigen driven response in these tumors and that these patients may benefit from immunotherapeutic interventions.

P44 Discovery and screening of protein biomarkers with the FirePlex Technology Platform

Timothy Erps, Amy Perea, PhD, Russell Neuner, PhD, Bianca Heinrich, PhD, Wayne Austin, Conor Rafferty, PhD, Matt Camilleri, Long To, James Murray, PhD, Daniel Pregibon, PhD, Elnaz Atabakhsh, PhD
Abcam, Inc, Cambridge, USA
Correspondence: Bianca Heinrich (bianca.heinrich@abcam.com)

Background

In patients and animal models, molecular biomarkers are used as indicators of normal and pathogenic processes. In drug discovery and screening pipelines, molecular biomarkers are used to assess the mechanism of action, efficacy, and toxicity of lead compounds. To address the need for rapid and sensitive quantitation of protein biomarkers, we have developed the FirePlex® Technology Platform. Utilizing patented hydrogel particles and a three-region encoding design, FirePlex Immunoassays allow for true, in-well multiplexing, providing flexible and customizable quantification of analytes.To facilitate biomarker discovery studies, we offer our standard-throughput FirePlex Immunoassays, which enable quantitation of up to 75 protein analytes per sample, from only 12.5 μl of input. These assays demonstrate 5 logs of dynamic range and sub-picogram/ml sensitivity, allowing for highly sensitive quantitation of analytes in serum, plasma, cell culture supernatant, urine, and saliva. Assays are run in 96well plate format, with readout on standard flow cytometers.

Methods

For drug discovery and screening studies, we offer our high-throughput FirePlex (FirePlex®-HT) Immunoassays for quantitation of up to 10 protein analytes per sample from only 6.25 μl biofluid input, in 384-well plate format. FirePlex-HT assays provide 3-4 logs dynamic range, demonstrate 1-100 pg/ml sensitivity, and have been validated in serum, plasma, and cell culture supernatant. The two-step workflow, no-wash assay format, and readout on high- content imagers limit hands-on time and are amenable to automation, thus making FirePlex-HT ideally suited for high-throughput screening studies.

Results

Here we present data from studies investigating cytokine profiling in human and rodent samples using the FirePlex immunoassays, and introduce the simplified workflow of the FirePlex-HT® immunoassays with data demonstrating the performance for quantifying key cytokines in multiplex, in biological samples.

Conclusions

Together, this novel combination of multiplexed, high-sensitivity assays and bioinformatics tools enables rapid quantitation of protein biomarker signatures in biofluid specimens.

P45 Tumor infiltrating T cells: complete workflows allow faster and improved analysis

Cesar Evaristo, PhD, Ramona Siemer, BTA, Philipp Steinbrueck, Zhongjie Yu, David Agorku, BS, Olaf Hardt, PhD, Christian Dose, PhD, Anne Richter, PhD
Miltenyi Biotec, Bergisch Gladbach, Germany
Correspondence: Cesar Evaristo (cesare@miltenyibiotec.de)

Background

Immunotherapies engaging T cells have proven clinical efficacy and tremendous potential. However, responses are often suboptimal. Further research is required to understand tumor-infiltrating leukocytes (TILs) biology and enhance outcomes. TIL analysis is technically challenging and labor intensive. Their number can be very low and small subpopulations might escape analysis as they get lost in the background noise. Importantly, tumor-infiltrating T cells are embedded in a highly immunomodulatory environment such that unbiased cell-intrinsic functional characterization is hindered. When working with large cohort sizes, even immunophenotyping TILs by flow cytometry is time consuming and data processing is laborious. Therefore, it is fundamental to use effective tools to streamline the workflow and to generate reliable data.

Methods

We established complete workflows combining tissue storage, dissociation, T cell isolation and phenotyping of mouse and human tumors. Tissues were processed immediately or stored in a solution that was shown to maintain cell viability and phenotype up to 48h after collection (Tissue Storage Solution™). Tumor dissociation was automated and optimized for epitope preservation using a tissue dissociator (gentleMACS™ Octo). We developed new T cell-specific enrichment reagents for magnetic cell sorting, incorporating novel technology enabling the removal of both superparamagnetic beads and antibody fragments (REAlease®). We labeled TILs using recombinant antibodies engineered to eliminate Fc receptor-mediated background (REAfinity™). Finally, flow cytometric analysis was performed using an automated analyzer (MACSQuant X™ and MACSQuant 16™).

Results

Optimal enzymatic dissociation was essential for analysis of critical tumor-specific sub-populations, such as PD1hiTim3+Lag3+CD39+CD8+ T cells present in tumors. Magnetic cell sorting resulted in enrichment of rare tumor infiltrating T cells by up to 500-fold, while maintaining activation status and phenotype. Isolation of T cells using REAlease technology allowed enrichment of subpopulations such as gamma delta T cells and Tregs with high purity. Use of REAfinity antibodies significantly diminished non-specific labeling of cells present in the tumor microenvironment. Use of MACSQuant analyzers decreased hands-on as well as total acquisition time by facilitating fast and fully automated sample processing, including sample mixing and absolute cell counting.

Conclusions

We optimized workflows that include standardized processing of tumor samples, newly developed tools for (semi-) automated magnetic isolation of tumor infiltrating T cells and automated flow cytometric analysis. These workflows greatly reduce experimental time and allow the performance of more complex experimental setups. We believe the use of these innovative tools and workflows can significantly increase the quality of the data obtained in immuno- oncology and immunotherapy research.

P46 Withdrawn

P47 Immune response in patients treated with autologous dendritic cells transduced with AdGMCA9 (DC- AdGMCAIX) in patients with mRCC from the phase I, open label, dose escalation and cohort expansion study

Izak Faiena, MD1, Nazy Zomorodian1, Beata Berent-Maoz1, Ankush Sachadeva1, Adrian Bot, MD, PhD2, Fairooz Kabinnavar1, Jonathan Said1, Gardenia Cheung-Lau1, Jia Pang1, Mignonette Macabali1, Tinle Chodon1, Xiaoyan Wang1, Paula Cabrera1, Paula Kaplan-Lezco1, Sandy Liu1, Begonya Comin-Anduix, PhD1, Allan Pantuck, MD1, Arie Belldegrun, MD, FACS1, Karim Chamie, MD1, Alexandra Drakaki, MD,PHD1
1UCLA, Los Angeles, CA, USA; 2Kite Pharma, Santa Monica, CA, USA
Correspondence: Izak Faiena (ifaiena@mednet.ucla.edu)

Background

Patients with metastatic RCC were treated in a phase I trial with autologous dendritic cells transduced by a replication deficient adenovirus comprised of GM-CSF+CAIX. Nine patients in three dose escalation cohorts (5, 15, and 50 X 106 cells/administration) were injected based on a 3+3 design.

Methods

An enzyme-linked immunospot (ELISpot) assay was used to determine the frequency of CAIX-specific IFN-γ producing T cells in blood. 15-mer overlapping peptides from CAIX protein, AdV5-pepton, and controls (+/-) were plated in Elispot plates pre-coated with anti-IFN-γ antibody. Subsequent to assay development, the number of T- cells responding to CAIX was calculated as above the lower limit of detection (LLD) (7 spots). After subtracting the backgrounds, fold change was calculated with respect baseline. The criterion for positive immunological response was defined as the mean fold change plus two. Further assessment included immunohistochemistry (IHC) staining of tissue from patients #4 (with PD) and #8 (with SD) for CAIX, CD4/8, Ki67, GrZ8, PD1/L1. The samples were scored based on percent positivity and staining intensity (Table 1). Tissue was obtained from the primary tumor prior to vaccination, and the target tumor at the end of the study period (18 months).

Results

ELISpot showed consistently positive responses against CAIX upon vaccination with DC vaccine, more prominently in patients in cohort 3 (high dose) as well as in those with longer time to progression (Figure 1). None of the treated patients showed an objective response. However, patient #8 who achieved stable disease (SD) lasting 18 months had more than 2-fold change in immune response over baseline on day 35 and 60 after the first vaccination cycle. All nine patients showed different degrees of immunological reaction to AdV5 at baseline and elevation at the end of the study. IHC showed that both patients had high CAIX expression in primary tumor and on the target lesion post vaccination. Immune infiltrates were seen at baseline in both subjects, with predominant CD4/8 T-cells in patient #8 with a high PD-1 expression in infiltrating lymphocytes without PD-L1 expression in the tumor environment.

Conclusions

DC-AdGMCAIX vaccination may elicit robust immunologic response against CAIX in patients with ccRCC. The findings of high PD-1 expression in the patient with SD in both the primary tumor and target lesion warrants future efforts to explore how combination therapies with biological response modifiers may further enhance clinical responses.

Acknowledgements

Supported by NCI RAID Initiative NSC 740833 and Kite Pharma

Trial Registration

NCT01826877

Ethics Approval

The study was approved by UCLA institutional review board IRB#12-000577
Fig. 1 (abstract P47).
Fig. 1 (abstract P47).

ELispot immune response in study subjects

Table 1 (abstract P47).

Immunohistochemistry staining

P48 A high baseline neutrophil-to-lymphocyte ratio in patients receiving anti-PD-1 therapy for head and neck cancer associates with poor prognosis, tumor hypoxia, and tumor M2 macrophage predominance

Corey Foster, MD1, Riyue Bao, PhD2, Sara Kochanny, BA1, Arun Khattri, PhD1, Rajesh Acharya, MS1, Allison Dekker, RN1, Yi-Hung Carol Tan, PhD1, Elaine Klema, BS1, Ryan Brisson, BS3, Vassiliki Saloura1, Alexander Pearson, MD, PhD1, Everett Vokes, MD1, Rom Leidner, MD4, Hisham Mehanna, PhD5, Tanguy Seiwert, MD1
1The University of Chicago Medicine, Chicago, IL, USA; 2The University of Chicago, Chicago, IL, USA; 3Oakland University, Rochester, MI, USA; 4Providence Cancer Center, Portland, OR, USA; 5University of Birmingham, Birmingham, UK
Correspondence: Tanguy Seiwert (tseiwert@medicine.bsd.uchicago.edu)

Background

The associations among a high neutrophil-to-lymphocyte ratio (NLR), prognosis, and the tumor microenvironment for patients with head and neck (H&N) cancer receiving anti-programmed death receptor 1 (PD-1) therapy are not established.

Methods

One-hundred-fourteen patients with metastatic H&N cancer received anti-PD-1 therapy with high baseline NLR defined as >8.77 (highest quartile). Logistic regression analyzed the association with overall response rate (ORR), and Kaplan-Meier methods and Cox proportional hazards regression models were used to test the association between NLR group and progression-free survival (PFS) and overall survival (OS). Tissue was available from 60 patients for gene expression profiling and pathway analysis using RNA sequencing. Expression abundance was quantified by Kallisto and normalized using the trimmed mean of M-values normalization method. Genes differentially expressed in the high vs. low NLR group were detected using limma voom with precision weights.

Pathway enrichment and upstream regulator prediction was performed through the use of IPA (QIAGEN Inc.).

Results

Median follow-up was 14.3 months. ORR was 22.3% with a trend towards lower ORR for the high NLR group (p=0.20). Median PFS was 1.7 months (95% confidence interval [CI]: 1.0-3.6) for high NLR patients and 3.7 months (95% CI: 3.2-6.2) for low NLR patients (p=0.01). Median OS was 9.6 months shorter in the high vs. low NLR group (3.8 months vs. 13.4 months, p<0.0001), and a decreased NLR 6 weeks into therapy associated with improved OS (p=0.0008). High NLR remained associated with OS (p=0.01) on multivariate analysis and trended toward independent association with PFS (p=0.07). High NLR associated with higher expression of CD163 (p=0.006) and HIF1-alpha (p=0.009), which was further supported by predicted activation of HIF1-alpha-target molecules based on experimental evidence from the Ingenuity Knowledge Base (activation z-score = 2.834, p=1.88E-04).

Conclusions

High baseline NLR associates with poor prognosis for patients with metastatic H&N cancer receiving anti-PD-1 therapy and correlates with increased expression of genes related to M2 macrophage predominance and hypoxia within the tumor microenvironment.

Ethics Approval

This study was approved by The University of Chicago Institutional Review Board.

P49 Comprehensive immune and molecular analysis of a cohort of non-small cell lung cancer (NSCLC) patients treated with a personal neoantigen vaccine, NEO-PV-01, in combination with anti-PD1

Joel Greshock1, Ramaswamy Govindan, MD2, Riley Curran1, Rana Besada1, Samantha Gates1, Victoria Kohler1, Meghan Bushway1, Julian Scherer, PhD1, Ying Sonia Ting1, Yuting Huang, Master of Science1, Yvonne Ware, Masters1, April Lamb1, Lisa D.Cleary1, Melissa Moles1, Richard Gaynor, MD1, Matthew Goldstein, MD, PhD1, Lakshmi Srinivasan1
1Neon Therapeutics, Inc; 2Washington University Medical School, Saint Louis, MO, USA
Joel Greshock (jgreshock@neontherapeutics.com)

Background

Neoantigens arise from DNA mutations and are important targets when presented on the surface of cancer cells for tumor-specific T cell responses. Vaccines targeting neoantigens have the potential to induce de novo and amplify pre-existing anti-tumor T cell responses. NEO-PV-01, a personal neoantigen vaccine, designed based on a patient’s tumor-specific mutations and are predicted to be presented by MHC molecules, is comprised of up to 20 long peptides and administered along with an immune adjuvant Poly-ICLC. Here we report comprehensive immune and molecular analysis observed in a cohort of metastatic non-small cell lung cancer (NSCLC) patients treated with NEO-PV-01 in combination with nivolumab (ClinicalTrials.gov: NCT02897765).

Methods

Tumor biopsies were scheduled for collection: i) prior to any treatment, ii) after 12 weeks of nivolumab monotherapy and iii) after completion of NEO-PV-01 vaccination. Tumor biopsies from each collection time point were analyzed for multiple immune and tumor markers by immunohistochemistry, gene expression and whole exome sequencing. Immune monitoring from peripheral blood samples was also evaluated at similar times for the presence of antigen-specific responses by IFNγ ELISPOT, intracellular cytokine staining, multi-parameter surface and functional phenotyping by FACS and the presence of cytolytic properties.

Results

IFNγ ELISPOT analysis with peripheral blood mononuclear cells (PBMCs) revealed neoantigen-specific CD4+ and CD8+ T cell responses that were only detected in the post-vaccination samples. Vaccine-induced immune responses were durable in one of the patients who reached the week 52 treatment timepoint. Neoantigen-specific T cells were of effector memory and central memory phenotype. Additionally, these cells were cytolytic and secreted IFNγ, TNFα and IL2. Assessment of serial tumor biopsies with repeat exome sequencing, gene expression, TCR repertoire analysis, immunohistochemistry and pathologic analysis will be presented.

Conclusions

NEO-PV-01 is immunogenic and leads to durable de novo neoantigen-specific immune responses in the peripheral blood of patients with metastatic NSCLC.

P50 Characterization of T cell receptor repertoire from FFPE extracted RNA from glioblastoma patients by anchored multiplex PCR and next-generation sequencing

Laura Johnson, PhD1, Josh Haimes1, Namitha Nair1, Angelo Porciuncula, PhD2, Kurt Schalper, MD, PhD2
1ArcherDX, Boulder, CO, USA; 2Yale School of Medicine, New Haven, CT, USA
Correspondence: Laura Johnson (ljohnson@archerdx.com)

Background

The adaptive immune system is involved in various disease conditions including cancer, chronic infection, autoimmune disease and transplant rejection. Adaptive immunity is mediated by B and T lymphocytes, which are activated upon antigen binding to antigen receptors expressed on their surface. Therefore, the spectrum of these antigen receptors, or immune repertoire (IR), provides a means to monitor adaptive immune responses to disease, vaccination and therapeutic interventions. Next-generation sequencing (NGS) of antigen receptor genes is a valuable tool in the study of disease states and responses to various interventions. The Archer® Immunoverse™ TCR assay quantitatively assesses the T-cell content and clonal diversity of patient derived RNA. Here, we describe the use of an Anchored Multiplex PCR (AMP)-based NGS assay to analyze RNA extracted from FFPE samples from glioblastoma patients during an immunotherapeutic clinical trial.

Methods

In this study RNA was extracted both pre and post treatment, and the effects of the two therapeutic interventions were compared. Testing was performed by two independent labs and with both NGS and NanoString™ based assays.

Results

The results of the two different assays showed concordance with the T cell content increasing significantly post immunotherapeutic intervention as detected by the NanoString assay (p=0.003) while the increase was not significant post standard of care treatment (p=0.438). The trend was the same with the Archer® Immunoverse™ assay where a larger increase in T cell clones was detected post immunotherapeutic treatment compared to post standard of care treatment (p=0.037 vs. 0.5275 respectively). The results we present also show that the number of detected TCR clones and clonotypes depended on sample input amount, sequencing depth and RNA quality.

Conclusions

The results of this study support the utility of the Archer® Immunoverse™ assay in the study of FFPE derived RNA from patient samples.

P51 Objective quantitative measurements of PD-L1 expression in tumor tissue by Phosphor Integrated Dots staining in patients with non-small cell lung cancer

Kazuyuki Hamada, MD, PhD1, Ryotaro Ohkuma, MD1, Takehiro Takahashi, MD, PhD1, Takeshi Setogawa2, Masaru Takahashi2, Hisatake Okada2, Hiroo Ishida1, Yutaro Kubota, MD, PhD1, Hirotsugu Ariizumi, MD PhD1, Etsuko Satoh, MD, PhD1, Yuya Hirasawa, MD1, Yasutsuna Sasaki, MD, PhD1, Kiyoshi Yoshimura, MD, PhD1, Takuya Tsunoda, MD1, Satoshi Wada, MD, PhD1
1Showa University, Tokyo, Japan; 2Konica Minolta, Inc., Tokyo, Japan
Correspondence: Satoshi Wada (st-wada@med.showa-u.ac.jp)

Background

The PD-1/PD-L1 signal suppresses activated T cells. Anti-PD-1 antibody treatment has an anti-tumor effect by blocking the PD-1/PD-L1 signaling, thereby reactivating exhausted T cells, especially tumor antigen-specific T cells. Several clinical studies were reported that the effect of anti-PD-1 antibody was associated with the number of CD8 T cells in tumor tissue. It was also reported that the effect of anti-PD-1 antibody was associated with the expression ratio (%) of PD-L1 in tumor tissue. However, the effect was also shown in cases with low PD-L1 expression ratio (%). Thus, the expression ratio (%) of PD-L1 in tumor tissue has not yet been a clearly predictive biomarker for anti-PD-1 antibody treatment. This may be a result of the pathologist individually measuring the percentage of PD-L1 positive cells in tumor tissue by immunohistochemistry (IHC).

Methods

Phosphor Integrated Dots (PID) staining is a highly accurate measurement method that detects the expression of specific molecules with high sensitivity and can be quantified as a particle number. In this study, we performed objective quantitative measurements of PD-L1 expression per unit area of tumor tissue by PID staining (PD-L1 PID score) in five patients with non-small cell lung cancer (three responders and two non-responders) treated with anti- PD-1 antibody. In addition, double staining with PD-L1 and CD8 T cells in tumor tissue was performed by PID and IHC, respectively.

Results

PD-L1 expression per unit area of tumor tissue was quantified by PID analysis. We also established double staining of IHC for CD8 T cells and PD-L1 with PID. It was proved that the PD-L1 PID score was significantly associated with the number of tumor-infiltrating CD8 T cells. Additionally, the effective cases of anti-PD-1 antibody tended to have a higher ratio of the PD-L1 PID score and the number of CD8 T cells.

Conclusions

PID staining is might be useful for PD-L1 measurement in tumor tissue. By conducting further case analysis and clarifying the relationship between the PD-L1 PID score and immune cells, characteristics of responders treated with anti-PD-1 antibody may be clarified.

Ethics Approval

The study was approved by Showa University School of Medicine‘s Ethics Board, approval number 2253.

P52 Emergence of an ICOShi CD4 T cell subset correlates with tumor reductions in subjects treated with the ICOS agonist antibody JTX-2011

Amanda Hanson1, Sean Lacey, MA Biostatistics1, Courtney Hart1, Ty McClure1, Ellen Hooper, MD1, Elizabeth Trehu, MD1, Deborah Law, DPhil1, Christopher Harvey, PhD1
1Jounce Therapeutics, Cambridge, MA, USA; 2Jounce Therapeutics, Inc., Cambridge, MA, USA
Correspondence: Christopher Harvey (charvey@jouncetx.com)

Background

Inducible T cell Co-stimulator (ICOS) is a costimulatory molecule expressed primarily on T lymphocytes that is upregulated upon cell activation. ICOS was identified as a potential target of interest based on clinical data from studies with anti-CTLA-4. Sustained ICOS upregulation was associated with clinical benefit, with preclinical data confirming a role for ICOS signaling in optimal anti-tumor activity. JTX-2011 is a first-in-class ICOS agonist antibody that has been demonstrated preclinically to have a tumor-centric dual mechanism of action through stimulation of CD4 T effector cells and depletion of intratumoral T regulatory cells. Clinical and biological activity of JTX-2011 is currently being evaluated in the advanced solid tumor setting in the ongoing Phase I/II ICONIC trial (NCT02904226).

Methods

Relapsed/refractory cancer patients received escalating doses of JTX-2011 as a monotherapy or in combination with nivolumab (240mg) administered q3w. Serial collection of peripheral blood mononuclear cells (PBMCs) was performed to enable longitudinal assessment of biological activity through flow cytometry-based assays, including target engagement (TE), and immunophenotyping (IP) including a limited assessment of T cell exhaustion.

Results

At the RP2D, peripheral TE demonstrated sustained (>70%) engagement over the entire dose cycle, and IP data demonstrated no consistently significant changes in T cell populations following JTX-2011 treatment. The potential of sustained agonism to drive phenotypic T cell exhaustion was assessed, and no evidence of phenotypic T cell exhaustion was observed in subjects treated with the combination of JTX-2011 and nivolumab. No conclusions could be reached in monotherapy setting due to limited sample availability. Further analysis of peripheral T cell phenotype demonstrated the emergence of an ICOShi subset of CD4 T cells in select subjects. Interestingly, the emergence of this cell population correlated with tumor reductions in both JTX-2011 monotherapy and combination subjects. Of the evaluable subjects assessed (N=37), emergence of the ICOShi subset was detected in 7/7 subjects with a reduction of their target lesion >30%, but not in any subject with best overall response of progressive disease.

Conclusions

Analysis of longitudinal blood samples from subjects treated with JTX-2011 suggests that the emergence of a distinct ICOShi population of peripheral CD4 T cells correlates with a radiographic response to JTX-2011 treatment. The emergence of this population may serve as a surrogate biomarker of response and may be useful in guiding future clinical development.

P53 X4P-001, an orally bioavailable CXCR4 antagonist, increases immune cell infiltration and tumor inflammatory status in the microenvironment of melanoma

Timothy Henion1, Robert Andtbacka, MD, CM, FACS, FRCSC2, Robert Pierce, MD3, Jean Campbell, PhD3, Melinda Yushak, MD, MPH4, Mohammed Milhem, MBBS5, Merrick Ross, MD6, Katie Niland7, Lu Gan, MD, PhD7, Sudha Parasuraman, MD7, Yan Wang, PhD7
1Acumen Medical Communications, Cambridge, MA, USA; 2Huntsman Cancer Institute, Salt Lake City, UT, USA; 3Fred Hutchison Cancer Research Center, San Diego, CA, USA; 4Emory University School of Medicine, Atlanta, GA, USA; 5University of Iowa, Iowa City, IA, USA; 6MD Anderson Cancer Center, Houston, TX, USA; 7X4 Pharmaceuticals, Cambridge, MA, USA
Correspondence: Robert Andtbacka (robert.andtbacka@gmail.com)

Background

The CXCR4/CXCL12 axis plays a central role in the trafficking of key immune cells in the tumor microenvironment (TME). Enhanced survival is reported in multiple syngeneic mouse models when a CXCR4 antagonist is combined with a check point inhibitor. X4P-001 is an oral, selective, allosteric inhibitor of CXCR4 that robustly inhibits the growth of murine B16-OVA melanoma. A biomarker-driven Phase 1b clinical study (NCT02823405) is on-going in melanoma patients to test the hypothesis that X4P-001 has the potential to modulate the TME in favor of an improved response to checkpoint inhibitors.

Methods

The primary objectives for this trial are to evaluate the safety and tolerability of X4P-001 alone and in combination with pembrolizumab in patients with metastatic melanoma and to characterize the effects of X4P-001 on tumor immune cell infiltrates. Serial biopsies of cutaneous or subcutaneous melanoma lesions, peripheral blood mononuclear cells, and serum samples were collected pre-dose, after three weeks of single-agent X4P-001, and after six weeks of combination treatment. Biopsies were assessed by immunohistochemistry and multiplex immunofluorescence for multiple protein markers, including CD8, FoxP3, and Granzyme B; and by NanoString analysis for changes in gene expression. Serum samples were assessed using the multi-analyte profile platform for key chemokines and cytokines.

Results

As of June 5th, 2018, 16 patients (median age 74.5 years, range 53-91) have been enrolled and all have completed treatment or are off study. X4P-001 alone increased infiltration of CD8+ T cells, granzyme B signal, antigen- processing and presentation machinery, such as HLA-DR, and both Tumor Inflammatory Signature (TIS) and IFN- gamma gene expression signature scores in the TME of select patients with paired evaluable biopsies. These biomarker responses were further enhanced when X4P-001 was combined with pembrolizumab. Adverse events related to either X4P-001 or pembrolizumab (greater than 10%) were diarrhea, fatigue, maculo-papular rash, dry eye, acute kidney injury, chills, decreased appetite, dry mouth, ocular hyperemia, oral candidiasis, and pruritus. These data, along with measurements of serum chemokines and cytokines, will be presented.

Conclusions

Treatment with single-agent X4P-001 consistently results in enhanced immune cell infiltration and activation in the TME leading to increases in TIS and IFN gamma gene expression signature scores. X4P-001 as a single-agent and in combination with pembrolizumab has an acceptable safety profile. These data support the use of X4P-001 to potentially improve outcomes for patients with tumors that are less responsive to checkpoint inhibitors.

Trial Registration

NCT02823405

Ethics Approval

Institutional Review Board approval was obtained from each participating center.

Consent

Each patient provided consent to participate in this clinical trial.

P54 Adenosine signature genes associate with tumor regression in renal cell carcinoma (RCC) patients treated with the adenosine A2A receptor (A2AR) antagonist, CPI-444

Andrew Hotson, PhD1, Stephen Willingham, PhD1, Lawrence Fong, MD2, John Powderly, MD, CPI3, Jason Luke, MD, FACP4, Mario Sznol, MD5, Saby George, MD, FACP6, Toni Choueiri7, Marios Giannakis, MD, PhD7, Brian Rini, MD8, Shivaani Kummar, MD9, Erik Evensen10, Ian McCaffery, PhD1, Chunyan Gu1, Long Kwei, PhD1, Ginna Laport1, Joseph Buggy1, Richard A. Miller, MD1
1Corvus Pharmaceuticals, Burlingame, CA, USA; 2University of California, San Fransisco, San Francisco, CA, USA; 3Carolina BioOncology Institute, Huntersville, NC, USA; 4University of Chicago, Chicago, IL, USA; 5Yale University Cancer Center, New Haven, CT, USA; 6Roswell Park Cancer Institute, Buffalo, NY, USA 7Dana Farber Cancer Institute, Boston, MA, USA 8Cleveland Clinic Foundation, Cleveland, OH, USA; 9Stanford University School of Medicine, Stanford, CA, USA; 10Basis Bioscience, Foster City, CA, USA
Correspondence: Richard A. Miller (rmiller@corvuspharma.com)

Background

Adenosine in the tumor microenvironment is immunosuppressive. CPI-444, an adenosine A2A receptor (A2AR) antagonist, restores immune function and is active in preclinical tumor models. CPI-444 is being investigated as monotherapy and in combination with atezolizumab (Tecentriq®) in an ongoing Phase 1/1b trial in patients (pts) with treatment-refractory renal cell carcinoma (RCC). Biomarker objectives included identifying tumor gene expression pathways associated with tumor response.

Methods

Tumor biopsies obtained at trial screening from pts with RCC (n=30) were analyzed for gene expression profiles with the Nanostring PanCancer Immune Panel that included 770 markers of immune activity and inflammation. The gene expression correlation (Spearman) matrix was hierarchically clustered (Ward’s method) to identify modules of genes that were co-expressed across tumors. Gene cluster expression intensity was compared between pts with evaluable best change in tumor size ≤0 (n=8) vs >0 (n=15). The composite adenosine gene signature score was calculated as the average of the Log2 of expression values for seven genes (CXCL1, CXCL2, CXCL3, CXCL5, IL1B, IL8, SERPINB2) shown to be induced in vitro in normal peripheral blood mononuclear cells by adenosine [1].

Results

Analysis of baseline RCC tumors revealed a cluster of correlated genes in T cell activation, IFNg-signaling, and antigen presentation pathways; this cluster was not significantly associated with tumor response to CPI-444 monotherapy or combination. However, the gene cluster containing the adenosine gene signature was associated with tumor regression (p=0.02). This cluster included 18 other genes such as chemokines (e.g. CCL20), complement genes, and serum amyloid A1. An additional cluster contained components of the adenosine signature, CXCL3 and CXCL5, and also associated with regression (p=0.04). In contrast, a cluster enriched for growth factor response genes associated with tumor progression (p=0.01) and negatively correlated with the adenosine signature cluster.

Likewise, a cluster containing CX3CL1 and complement inhibition also associated with tumor progression (p=0.04) and inversely with the adenosine signature. These data support the co-regulation of adenosine and other biological processes within the tumor microenvironment, such as a negative relationship between adenosine and both CX3CL1 and growth factor signaling.

Conclusions

CPI-444 anti-tumor activity in RCC was associated with baseline expression of adenosine responsive genes, and is consistent with the mechanism of action of CPI-444. Adenosine-mediated immunosuppression and growth factor pathway activation may represent alternative oncogenic processes that define RCC pt subsets and could provide pathway specific biomarkers for prognosis and pt selection.

Trial Registration

NCT02655822

References

1. Willingham S, et al. Identification of adenosine pathway genes associated with response to therapy with the adenosine receptor antagonist CPI-444. European Society for Medical Oncology poster presentation. Munich, Germany, 2018.

Ethics Approval

The protocol was approved by the institutional review board or ethics committee at each participating center.

P55 Within the secretome: Immunomodulatory role of extracellular vesicles in breast cancers

Sheeba Irshad, MD, PhD1, Atousa khiabany, Mres1, Fabian Flores-Borja, PhD1, Ines García Carcedo1, Felix Wong1, Fabienne Beuron, PhD2, Jose vicencio3, Andrew Tutt1, Tony Ng1
1Kings College London, London, UK; 2Institute of Cancer Research, London, UK; 3University College London, London, UK
Correspondence: Sheeba Irshad (sheeba.irshad@kcl.ac.uk)

Background

Extra-cellular vesicles (ECVs) are heterogeneous submicron-sized vesicles that vary in size, composition and surface biomarkers. Recently, evidence suggests ECVs can have dichotomic role in the regulation of the immune system, enhancing or suppressing an immune response depending on their cell of origin and functional state. We investigated the immunomodulatory functions of breast cancer cell line and patient-derived serum ECVs.

Methods

ECVs were isolated from cell line culture media and serum of patients diagnosed with early breast cancers (BC) using ultracentrifugation and gradient column methods, respectively. Isolated ECVs were characterized by nanoparticle flow, electron microscopy and dot blot analyses. ECVs were cultured with human PBMCs with or without ECVs for upto 5 days. Cell phenotypes, and their proliferation and activation states were evaluated by flow cytometry.

Results

Breast cancer cell-line ECVs significantly improved the survival of PBMCs in vitro (p=0.005). PBMCs cultured with ECVs had significantly higher CD4+T-cells (p=0.01) but reduced number of CD8+T-cells (p=0.001) compared to unstimulated PBMCs. No significant differences in the B-cell or NK-cell counts were observed. Assessment of the activation status demonstrated that BC cell line-derived ECV inhibited the activation of CD4+T-cells by anti-CD3/anti-CD28 activation beads (p=0.01) but no significant difference was observed in the activation status of CD8+T-cells. No differences in the proliferation rates of CD3+CD4+ or CD3+CD8+ T cells between ECV- stimulated versus no ECV stimulation was observed. Further analysis for the co-expression of CD45RO and CCR7 on CD4+ or CD8+ distinguished between naïve (CD45RO−CCR7+), central memory (CD45RO+CCR7+,Tcm), memory effector (CD45RO+CCR7−, TEM), and effector (CD45RO−CCR7−, Teff) T-cells. PBMC- ECV co- cultures significantly increased the percentage of CD4+ Tcm-cells (p=0.001) and CD8+ Tcm-cells (p=0.001), as compared to the decrease see in the frequency of the CD4+ Teff cells (p=0.001). Significant increase in T-regulatory cells populations in the PBMC-ECV co-culture conditions was observed (p=0.01). The effect of patient-derived serum ECVs on PBMCs in vitro are ongoing but have revealed differences in ECV secretion capacity within BC subtypes (TNBC, HER2+, ER+ ECVs). Functional validation of our preclinical results is awaited.

Conclusions

These results suggest that BC cells utilise ECVs to tame immune cells to promote an immunosuppressive microenvironment. The skewed maturation phenotype of T-cells following ECVs stimulation with increase in Tcm-cells suggest an accelerated maturation of naive T-cells. These reported differences in the immunomodulatory function of ECVs require further investigation.

P56 New Dextramer technology for T-cell epitope profiling, and single cell deep phenotyping using DNA barcodes, transcriptomic and genetic sequence analysis, on single cells

Kivin Jacobsen, PhD1, Liselotte Brix, phd1
1Immudex, Copenhagen, Denmark; 2Immudex Aps, Copenhagen, Denmark
Correspondence: Kivin Jacobsen (kj@immudex.com)

Background

Identification of disease-specific T-cell epitopes is key to the development of many novel vaccines and immunotherapies. Profiling disease-specific T cells, emerging during a cellular immune response e.g. in tumor development or destruction, is an important aspect of personalized immunotherapy. The epitope diversity of the human population is large, and the technologies for identifying disease-specific epitopes have been inadequate.We have developed a process in which DNA barcoded Dextramer reagents are used to simultaneously screen for hundreds, or potentially thousands, of T-cell epitopes in a few milliliters of blood. Single cell deep phenotyping is quickly emerging, analyzing both cellular proteins, transcriptome and genetics of single cells by sequence analysis.DNA barcoded Dextramer reagents extend these technologies, to include simultaneous analysis of antigen specific T cells, by single cell sequencing.

Methods

We present a process in which DNA barcoded Dextramer reagents (dCODE Dextramer, can be used to simultaneously screen for hundreds, or potentially thousands, of T-cell epitopes in one small patient sample. Similar to previously reported academic methods, CITE-seq and REAP-seq, it is now possible to combine DNA barcoded MHC Dextramer technology, with single cell analysis allowing direct correlation between the specificity of the antigen specific T cell and its cognate T cell receptor protein sequence, which has not been possible with existing technologies.

Results

Profiling of antigen-specific T cells, by simultaneous detection of a large numbers of T-cell specificities in the same small cell sample, was performed using DNA barcode MHC Dextramer analysis, followed by mass sequencing. Pools of 50 MHC Dextramer reagents was screened on PBMC samples from human donors, and the analysis could identify relevant antigen-specific T-cell populations. The results were confirmed by Dextramer analysis using conventional fluorochrome based flow cytometry.

Conclusions

DNA Barcoded Dextramer reagents in large libraries, allows for, high-throughput screening of antigen-specific T cells in limited sample material.DNA Barcoded Dextramer, together with single cell sequencing technology, can be used for deep T cell phenotyping by barcode sequencing and transcriptomic or genetic sequence analysis, on the single cell level.

P57 A rapid multi-color fluorescence imaging on frozen tissues

Dinesh Jaishankar, PhD1, Cormac Cosgrove, PhD2, Caroline Le Poole2
1Northwestern University, Chicago, IL, USA; 2Feinberg School of Medicine, Chicago, USA
Correspondence: Caroline Le Poole (caroline.lepoole@northwestern.edu)

Background

Immunofluorescence (IF) imaging is a commonly performed routine technique. To detect >four molecules that identify different cell types in a single image, multiplex imaging techniques now exist wherein the fluorophores are spectrally separated to avoid overlap [1]. Quantitative multispectral imaging on tissues can provide a wide range of information ranging from predicting response rates of immunotherapies to immune monitoring [2]. Current techniques to perform multispectral imaging are tedious, require expensive kits, and have been developed with paraffin-embedded sections in mind. Here, using readily available fluorophore-conjugated antibodies, and a rapid IF staining protocol, we imaged up to six different colors in a frozen mouse spleen section.

Methods

8μm sections of a naïve and frozen mouse spleen were cut, acetone fixed, and stored at -20oC until the staining process. An antibody cocktail comprising of primary-labeled antibodies to CD3, CD8, CD19, CD11b and CD206 was made. For the staining, sections were blocked using the Superblock, incubated with the antibody cocktail for 1hr at RT and coverslips were added with Prolong Diamond mounting medium. The DAPI stain was used to stain the nuclei. Single color and unstained slides were also processed the same way. Image acquisition, under 20x objectives, and the multispectral imaging analysis was performed on the Vectra 3.0 from Perkin Elmer.

Results

Using the spectral library generated from the single stained and unstained slides, the multispectral imaging analysis revealed the detection of six different colors on the frozen section.

Conclusions

For the first time, we show a rapid and easy method of detecting different markers on a single frozen tissue section.

This method has great potential for immune monitoring studies to measure the number and colocalization of immune and target cells.

Acknowledgements

We would like to thank Ryan Deaton, the imaging specialist at the University of Illinois at Chicago for helping with the multispectral imaging.

P58 The use of low-coverage sequencing of cell-free DNA for monitoring response to immune checkpoint inhibitors throughout treatment

Taylor Jensen, PhD1, Aaron Goodman2, Shumei Kato, MD2, Mina Nikanjam2, Christopher Ellison1, Gregory Daniels, MD, PhD2, Lisa Kim, MS2, Kimberly Kelly, MS1, Kerry Fitzgerald, PhD1, Erin McCarthy1, Prachi Nakashe, MS1, Amin Mazloom1, Graham McLennan, MS1, Daniel Grosu, MD, MBA1, Mathias Ehrich1, Razelle Kurzrock, MD2
1Sequenom, a LabCorp company, San Diego, CA, USA; 2University of California San Diego, San Diego, CA, USA
Correspondence: Taylor Jensen (taylor.jensen@integratedgenetics.com)

Background

Immune checkpoint inhibitors continue to revolutionize the cancer treatment paradigm. It has been observed that even some patients with advanced, refractory malignancies achieve durable responses; however, only a minority of patients benefit, demonstrating the importance of developing new biomarkers to predict and/or monitor patient outcome. While markers including PD-1/PD-L1 expression, microsatellite instability, and tumor mutational burden have been shown to have varying degrees of predictive power, they are not the ideal markers to monitor and differentiate response during treatment. Interrogating cell-free DNA (cfDNA) isolated from plasma (liquid biopsy) provides a promising noninvasive method for monitoring response.

Methods

Whole blood was collected in Streck BCT tubes and plasma subsequently separated using centrifugation. cfDNA was isolated from 4 mL of plasma using an automated process and used to prepare sequencing libraries. The total amount of cfDNA isolated from each plasma sample was quantified using droplet digital PCR. All sequencing libraries were then subjected to low-coverage, genome-wide sequencing using illumina HiSeq2500 instruments, generating a median of 35.4 million reads (~0.3X genomic coverage). A newly developed metric – the Genome Instability Number (GIN) – was utilized to measure and quantify the cumulative abundance of copy number alterations (CNAs) present in the cfDNA and additional algorithms were applied to identify the genomic loci containing CNAs.

Results

A series of 477 plasma aliquots prospectively collected at various time points throughout the treatment of 98 cancer patients receiving immunotherapy was measured in this study. These data built on previous results to further describe how the GIN can be used to discriminate clinical response from progression, differentiate progression from pseudoprogression, and identify hyperprogressive disease, as early as 4-6 weeks after treatment initiation. In addition, the cfDNA profiles of a small cohort of melanoma patients were evaluated at frequent intervals shortly after the initiation of checkpoint inhibitor therapy to better identify response kinetics. Finally, CNAs across the genome were analyzed to determine whether specific genes or genomic regions were associated with patient response to checkpoint inhibitors.

Conclusions

These data suggest that low coverage, genome-wide sequencing of cfDNA may have utility for monitoring response to immunotherapy in cancer patients.

Ethics Approval

This study was performed and consents were obtained in accordance with UCSD Institutional Review Board guidelines for specimen collection and data analysis (NCT02478931) and for any investigational treatments.

P59 Assessment of consistency of multiplex fluorescent immunohistochemistry data across multiple users utilizing different quantitative analysis strategies

Shawn Jensen, PhD, Carmen Ballesteros Merino, PhD, Sebastian Marwitz, Nikhil Lonberg, HSDG, Bernard A. Fox, PhD
Robert W Franz Cancer Center, Earle A Chiles Research Institute, Portland Providence Medical Center, Portland, OR, USA
Correspondence: Bernard A. Fox (foxb@foxlab.org)

Background

Tools that facilitate examination of the tumor microenvironment in cancer patients who either respond or do not respond to treatment are informative to the future design of immunotherapeutic strategies. Multiplex fluorescent immunohistochemistry (mIHC) is a technique enabling examination of the number and location of cells within the tumor microenvironment. Recently, we described the Cumulative Suppression Index (CSI), which examines the number of CD8+ T cells in the invasive margin of tumor combined with the number of FoxP3+ or PD-L1+ cells within 30 um of CD8+ T cells[1]. This CSI correlated with overall survival in a cohort of 119 patients with HPV- oral squamous cell cancer. Future application of the CSI will require reliable analysis methods with minimal variation across studies and users to enable comparative analysis. In this current study, we systematically compared the reliability of three different methods of enumerating cellular phenotypes of mIHC images across multiple users.

Methods

Primary tumors obtained from oral squamous cell cancer patients were sectioned and stained with antibodies to CD3, CD8, FoxP3, CD163, and PD-L1. Nine representative images were collected from one patient and analyzed using either commercial phenotyping software based on machine-learning to phenotype cells (C), commercial phenotyping software coupled with a Thresholding method (C+T), or the Thresholding method alone (T). Multiple independent users analyzed the same nine images determining the number of CD3+ PD-L1-, CD3+PD-L1+, CD8+PD- L1-, CD8+PD-L1+, FoxP3+, CD163+PD-L1-, CD163+PD-L1+, or PD-L1+ cells using each of the three methods.

Results

Analysis of the variation within each user across the three different analysis methods demonstrated tight correlation for the principle phenotypes of the CSI, namely CD8+PD-L1-, CD8+PD-L1+, FoxP3+, and PD-L1+ cells (Spearman Rank Correlation p<0.05). CD163+PD-L1+/- cells and CD3PD-L1+/- cells showed a correlation in only a fraction of the users which was partially influenced by low cell counts in a portion of those cellular phenotypes. More importantly, examining the reproducibility of data between users using the intraclass correlation coefficient demonstrated consistency across all phenotypes for all users using either the C+T or T methods (p<0.05).

Conclusions

Comparative analysis of mIHC data between multiple users requires confidence in a reproducible and consistent method for data analysis. These data demonstrate that C+T or T methods of analyzing data minimize inter-user variation when using the CSI mIHC panel tested in this study.

References

1. Feng Z, et al. Multiparametric immune profiling in HPV- oral squamous cell cancer. JCI Insight. 2017;2(14):e93652.

P60 Circulating tumor DNA assessment in plasma samples collected in Atezolizumab versus docetaxel in subjects with previously treated non-small-cell lung cancer (OAK) study

Yuqiu Jiang, PhD1, Namrata Patil, PhD2, Johnny Wu1, Wei Zou2, Stephanie Yaung1, Aarthi Balasubramanyam1, Susan Flynn2, Maureen Peterson2, Eric Peters2, Priti Hegde, PhD2, Simonetta Mocci2, Marcin Kowanetz, PhD2, John Palma1
1Roche Sequencing Solutions, Pleasanton, CA, USA; 2Genentech, South San Francisco, CA, USA
Correspondence: John Palma (john.palma@roche.com)

Background

Circulating tumor DNA (ctDNA) sequencing and analysis has the potential to transform clinical management of subjects with advanced NSCLC. It has been shown that mutation(s) or molecular tumor burden assessed in plasma using NGS could potentially serve as disease monitoring tool or therapy response predictions.

Methods

In this exploratory pilot study, longitudinal plasma samples have been analyzed for presence of ctDNA in NSCLC subjects from the OAK study. 108 subjects were selected based on their clinical response profiles of early and late responders, and early and late progressors. For this preliminary report, 102 baseline samples and subsequent plasma collected at C2D1, C3D1, and C4D1 were analyzed by NGS. The AVENIO ctDNA Surveillance kit** (Roche, Branchburg, NJ) were used for sequencing analysis. The Surveillance kit (200 kb size) contains 17 cancer driver genes and additional 180 frequently mutated genes in cancer. This kit is capable of detecting four mutation classes: SNVs, fusions, CNVs and InDels. Association of survivals with ctDNA level or change was interrogated with Cox regression model. Response to treatment was assessed using RECIST v1.1.

Results

102 of the 108 (94%) baseline plasma samples were successfully sequenced. All 102 (100%) samples had somatic variants detected. The most commonly mutated genes in tumors were TP53 (59/102 subjects), KRAS (21/102), APC (21/102), and NPAP1 (15/102). The median number of variants detected per subject was 7. Mutant molecules per milliliter (MMPM) was also assessed for each baseline samples. The median MMPM was 139, ranging from 1 to 1972 for these 102 samples. Survival and therapy response in relation to ctDNA level or change at different time points will be reported at the time of presentation.

Conclusions

ctDNA testing with molecular barcoded sequencing and digital background error suppression of a 197 gene panel offers high sensitivity for tumor variant detection. The study demonstrated that tumor variants can be detected in blood in pre-treatment samples using the AVENIO kit. Subsequent analysis will be performed with plasma samples collected at C2D1, C3D1, and C4D1, and reported at the time of submission.**For Research Use only; Not for diagnostic purposes.

Acknowledgements

We would like to acknowledge Dr. Dan Klass, Ph.D., Katrina Mayol, and Nasiema Wingate-Pearse for their insightful discussions.

P61 Development of an exosome / EV analysis pipeline for tumor and immune monitoring

Joshua Welsh, PhD1, Kevin Conlon, MD1, Milos Miljkovic, MD, MSc1, Julia Kepley, BS1, Jennifer Marte, BS MD1, Jason Savage, PhD1, Veronica Galli, PhD1, Katherine McKinnon, MS1, Katherine Calvo, MD, PhD2, Hong Zhang1, Cynthia Masison1, Fatima Karzai, MD1, Fatah Kashanchi3, Deborah Citrin1, Steve Jacobson4, James Gulley, MD, PhD, FACP1, Genoveffa Franchini, PhD1, Thomas Waldmann, MD1, Kevin Camphausen, MD, PhD1, Jennifer Jones, MD, PhD1
1National Cancer Institute, Bethesda, MD, USA; 2NIH Clinical Center, Bethesda, MD; 3George Mason University, Fairfax, VA, USA; 4NINDS, NIH, Bethesda, MD, USA
Correspondence: Jennifer Jones (jennifer.jones2@nih.gov)

Background

Exosomes and other Extracellular Vesicles (EVs) carry surface receptors that are characteristic of their cells of origin. Therefore, Extracellular Vesicles (EVs) have tremendous potential as non-invasive biomarkers for immunotherapy. We have developed a first-in-class pipeline to characterize EV heterogeneity and provide high- sensitivity quantification of informative EVs in biofluids before, during, and after treatment. This pipeline combines multiplex assays with high-resolution single EV flow cytometric methods together into a Mutiplex-to-Single EV Analysis (Mt-SEA) pipeline. With this pipeline, we are able to characterize a broad range of relevant EV subsets, while also accurately measuring the concentration of specific EV populations. With clinical cases, we demonstrate the performance of Mt-SEA method by confirming strong correlations of liquid biopsy EV repertoires with tumor burden and responses to treatment, including an abscopal immune response following radiation. Furthermore, EV analysis with Mt-SEA may identify previously unrecognized prognostic epitopes or EVs subsets.

Methods

To evaluate the use of this pipeline in an exploratory clinical cohort, we evaluated EVs from plasma samples of Adult T Cell Leukemia/Lymphoma patients receiving palliative radiation. Plasma was obtained before and after treatment. Multiplex EV capture beads were used with additional detection antibodies to identify more than 40 major EV subsets. General exosome and EV detection epitopes included CD63, CD9, and CD81. Tumor-specific epitopes for each patient included CD4, CD5, and CD25, based on available histo-/cyto-pathology results. Next, high-resolution single EV analyses were performed with nanoFACS sorting and a prototype nanoFCM analyzer.

Results

ATLL-derived EVs were detected in each pre-treatment sample, with reduced specific ATLL-derived EV subsets concentrations at the end of treatment. Furthermore, ATLL-specific EVs from patients with progressive systemic disease prior to treatment were found to carry CD44 and other stemness-associated epitopes, consistent with increasing tumor aggressiveness. Responses to treatment that were clinically evident mirrored changes in the Mt- SEA EV profiles, and Mt-SEA identified new candidate prognostic EV profiles associated with clinical outcomes that would not have been predicted.

Conclusions

The use of EVs as clinical biomarkers requires a combination of methods to broadly characterize EVs and rigorously enumerate specific selected EV subsets. Therefore, we developed the Mt-SEA pipeline. Mt-SEA provides unexpected insights into tumor biology, and detection of tumor-associated and immune-associated EVs and detection of EV repertoire changes during treatment paves the way to future evaluation of the Mt-SEA pipeline for personalized, bio-adaptive therapies in a wider range of tumor types.

Ethics Approval

The study was approved by the NCI IRB, with NIH intramural protocol number 02-C-0064.

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.

P62 A Pan-cancer view of the immune landscape in the tumor microenvironment via RNA and their potential for biomarkers in clinical trials

Wendell Jones (wendell.jones@q2labsolutions.com)
EA Genomics / Q2 Solutions

Background

Immune-based biomarkers are now commonly available when measuring the tumor microenvironment (TME), although many are based on immuno-histochemistry methods which sometimes have drawbacks: specificity, quantitation, variety. In addition, next-gen RNA sequencing and focused RNA assays are now much more accessible (for example, processing formalin-fixated material has a much higher degree of success) and are commonly employed in clinical trials to measure tumor activity and immune cell content in the TME.

Methods

Based on previous research [1-3] and pan-cancer information from TCGA, we have derived gene signatures for immune activity encompassing a dozen distinct subcomponents of the immune system and its response to tumor evolution. We have applied these immune signatures in multiple independent oncology datasets from the same indication and in multiple solid tumor indications (> 20) from TCGA and other sources to explain variation in disease-free and overall survival (DFS and OS). Using Cox multivariable analysis, we further have clarified the potential impact of these various immune components and tumor mutational burden (TMB) relative to therapy response and overall survival in the presence of multiple clinical covariates many of which are specific to each indication but which typically include patient age, tumor stage, and residual disease status. These signatures are available from a variety of RNA measurement platforms, including those that specialize in measuring difficult input material.

Results

The results include recognizing that the immune status, reflected in the expression-based signatures, play a large role (based on Hazard Ratio estimates) in explaining variation in survival in many indications, often having a larger potential impact than other biological factors or interventions. In addition, we show that there is often an interplay between cytotoxic and immune modulating cell activity in the TME (for example, cytotoxic lymphocytes vs. M2TAM cells) that explains much more variation than either factor by itself. We also show that these results are reproducible across multiple independent datasets for the same indication, implying they are robust. Further we show that while some indications have much commonality regarding the potential impact of specific immune components, other indications (such as pancreatic and kidney cancer) show more complex and sometimes counter- intuitive results regarding immune status which need to be explored further.

Conclusions

The results suggest relevant biomarkers for immune status that can be gleaned and developed from RNA measurements of the TME which can then be applied to clinical trials or converted into companion diagnostics as needed.

Acknowledgements

Some of the results presented are in part based upon data generated by the TCGA Research Network: http://cancergenome.nih.gov/

References

1. Newman AM, Liu CL, Green MR, Gentles AJ, Feng W, Xu Y, Hoang CD, Diehn M, Alizadeh AA. Robust enumeration of cell subsets from tissue expression profiles. Nature methods. 2015 May 1;12(5):453-7.

2. Rőszer T. Understanding the mysterious M2 macrophage through activation markers and effector mechanisms. Mediators of inflammation. 2015;2015.

3. Bindea G, Mlecnik B, Tosolini M, Kirilovsky A, Waldner M, Obenauf AC, Angell H, Fredriksen T, Lafontaine L, Berger A, Bruneval P. Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer. Immunity. 2013 Oct 17;39(4):782-95.

Ethics Approval

This study is a meta-analysis of data from other studies and thus did not require review by an institutional ethics board.

P63 Estimation of Microsatellite Instability (MSI) by next-generation sequencing using a novel MSI classification method

Asha Kamat, PhD1, Sameh El-Difrawy1, Annie Kraitcheva2, Alice Zheng1, Simon Crawley1
1Thermofisher Scientific, South San Francisco, CA, USA; 2ThermoFisher.com, Carlsbad, CA, USA
Correspondence: Asha Kamat (asha.kamat@thermofisher.com)

Background

Cancer-associated instabilities at microsatellite locations throughout the genome have been shown to be predictive of response to immunotherapy treatment.Here, we describe an NGS-based method to assess Microsatellite Instability (MSI) status in tumor-only and tumor-normal samples utilizing Ion AmpliSeq™ HD technology and an Ion GeneStudio™ S5 next-generation sequencer.

Methods

We have identified optimal chemistry and developed a novel algorithm to assess MSI status of samples using a large number of markers on an Ion GeneStudio S5™ Series sequencer. The diverse marker set includes monomers that vary in length between 10 BP and 40 BP in addition to di- and tri-nucleotide STR markers. The algorithm works with tumor-only or tumor-normal samples. Each sample is assigned an MSI score based on features that measure MSI response of markers in the assay.

Results

Combining a groundbreaking AmpliSeq™ HD workflow and a novel analysis method, we developed an assay that utilizes a diverse set of MSI markers. We evaluated performance of the assay and algorithm with a set of 50 samples including CRC, Endometrial and Gastric carcinomas tumor in both MSI-High and MSS status. The resulting scores were in concordance with results from capillary electrophoresis studies.

Conclusions

A next-generation sequencing based assay using multiple markers was developed to assign MSI status to tumor samples with great precision. The accuracy of the assay was validated using an orthogonal test. MSI status can be assigned using tumor-only or tumor-normal samples.

P64 Evaluation of PD-L1 and IRF-1 expression on circulating tumor cells as a predictive biomarker of checkpoint inhibitor response

Laura Kennedy, MD PhD1, Lance U'Ren, DVM, PhD2, Yao Sun2, Petros Grivas, MD, PhD1, Laura Chow, MD1, Vijayakrishna Gadi, MD, PhD1
1University of Washington, Seattle, WA, USA; 2Rarecyte, Inc., Seattle, WA, USA
Correspondence: Vijayakrishna Gadi (vkgadi@uw.edu)

Background

PD-1 CPI therapy can generate durable responses with fewer side effects compared to conventional cytotoxic chemotherapy. Unfortunately, CPI can induce an objective response in less than 15 - 20% of non-melanoma solid tumor patients [1-3]. Multiple biomarkers have been evaluated as potential factors predicting response, but none has shown reproducible clinical utility across tumor types. Higher PD-L1 expression in tumor tissue is associated with higher response rates, but a single tumor tissue sample may not reflect spatial and temporal variability in PD-L1 expression. Circulating tumor cells can be collected at multiple timepoints with minimal risk and may provide a more comprehensive and dynamic view of tumor heterogeneity. Higher IRF-1 expression in tumor tissue has been correlated with longer progression-free survival (PFS) in metastatic melanoma patients treated with CPIs [4]. We hypothesize that evaluating both PD-L1 and IRF-1 expression on CTCs may better predict patient response to PD-1 CPIs.

Methods

Patients with metastatic solid tumors receiving de novo CPI alone or in combination with other treatments are eligible. Patients undergo peripheral blood collection for CTC evaluation at 3 timepoints: prior to receiving the first dose of CPI, prior to the second dose of CPI, and 3-6 months after starting therapy or at the time of CPI discontinuation. PFS is defined as time between initiation of therapy and progression or death of any cause, with patients censored at the time of last follow up. To isolate CTCs from peripheral blood, we use the AccuCyte® kit to create slides and a Ventana Discovery Ultra autostainer to stain the slides with a pancytokeratin (CK)/EpCAM antibody cocktail, anti-CD45, and a nuclear dye. Using the CyteFinder® software, CTCs are identified by positive staining for CK/EpCAM and negative staining for CD45. PD-L1 and IRF-1 expression are determined using mean fluorescence intensity (MFI) in computationally estimated cell compartments.

Results

To optimize the assay, low and high PD-L1/IRF-1 were created by spiking A549 cells cultured with or without 10 ng/mL interferon-gamma into whole blood (Figure 1). We will present progression-free survival data and corresponding CTC data for the first set of patients enrolled. A number of these patients have tumor tissue samples with varying degrees of PD-L1 expression by IHC, which will allow correlation between tumor tissue and CTC PD- L1 expression.

Conclusions

CTC PD-L1 and IRF-1 expression may provide a more comprehensive predictive biomarker for patients starting PD-1/PD-L1 CPIs.

Acknowledgements

We would like to thank the patients and providers of the Head/Neck/Lung Oncology group, the Genitourinary Oncology group, the Gastrointestinal Oncology, and the Breast Oncology group at the Seattle Cancer Care Alliance for participating in this trial. We would like to thank Alisa Clein for assistance with clinical research coordination.

References

1. Borghaei H, et al. Nivolumab versus docetaxel in advanced nonsquamous non-small-cell lung cancer. New England Journal of Medicine. 2015;373:1627-1639.

2. Powles T, et al. Atezolizumab versus chemotherapy in patients with platinum-treated locally advanced or metastatic urothelial carcinoma (IMvigor211): A multicentre, open-label, phase 3 randomised controlled trial. Lancet. 2018;391:748-757.

3. El-Khouiery A, et al. Nivolumab in patients with advanced hepatocellular carcinoma (Checkmate 040): An open-label, non-comparative, phase 1/2 dose escalation and expansion trial. Lancet. 2017; 389:2492-2502.

4. Smithy J, et al. Nuclear IRF-1 expression as a mechanism to assess “capability” to express PD-L1 and response to PD-L1 therapy in metastatic melanoma. J Immunother Cancer. 2017;5:25.

Ethics Approval

This study was approved by the Fred Hutchinson Cancer Research Center Institutional Review Board, Committee D, Protocol #10031.
Fig. 1 (abstract P64).
Fig. 1 (abstract P64).

See text for description.

P65 IDO/HLA-DR expression and tumor mutational burden are complementary predictive biomarkers of anti- PD-1 immunotherapy in squamous cell carcinoma of the head and neck

Arun Khattri, PhD1, Ju Young Kim2, Riyue Bao, PhD3, Rajesh Acharya3, Yi-Hung Carol Tan, PhD3, Rom Leidner, MD4, Hisham Mehanna, PhD5, Nathan Roscoe6, Christine Vaupel, PhD6, Naveen Dakappagari6, Tanguy Seiwert, MD3, Sara Kochanny, BA7
1Department of Medicine, The University of Chicago, Chicago, IL, USA; 2Navigate Biopharma Services, Inc., Carlsbad, CA, USA; 3The University of Chicago, Chicago, IL, USA; 4Providence, Portland, OR, USA; 5Institute of Cancer and Genomic Sciences, Birmingham, UK; 6Navigate Biopharma, Carlsbad, CA, USA; 7University of Chicago, Chicago, IL, USA
Correspondence: Tanguy Seiwert (tseiwert@medicine.bsd.uchicago.edu)

Background

Background: PD-1 checkpoint blockade is active in squamous cell carcinoma of the head and neck (SCCHN) (e.g. Seiwert et al, Lancet Oncol, 2016). Biomarkers such as PD-L1 immuno-histochemistry (IHC) and markers of the T- cell inflammation (e.g. Interferon-γ gene signature, INF-G) identify tumors more likely to benefit from treatment. However, both protein- and genomic markers have not been concurrently examined to identify an ideal combination of predictive markers in SCCHN.

Methods

Methods: Pretreatment formalin-fixed paraffin-embedded biopsies of 82 anti-PD-1 treated recurrent/metastatic SCCHN patients were studied for key immune resistance markers, namely, PD-L1 (tumor, stroma, PD-1/PD-L1 interaction), Myeloid suppressor cells (CD11b/IDO1/HLA-DR/ARG1) and regulatory T cells (CD4/CD8/CD25/FOXP3/Ki67) by combining multiplexed immunofluorescence (IF) and novel quantitative spatial imaging algorithms via AQUA® (Automated Quantitation Analysis) technology. Concurrently, INF-G gene signature was assessed by RNA-seq and tumor mutational burden (TMB) via exome sequencing (or if not available 1000-gene panel-sequencing (OncoPlus)). Correlations with overall survival (OS) and progression free survival (PFS) were performed using optimized cut points.

Results

Results: Stromal PD-L1 level was more predictive of survival than tumor PD-L1 level or PD-1/PD-L1 interaction score (OS, p=0.039; PFS p=0.019). However, tumor- IDO1/HLA-DR expression was most predictive of survival (OS, p=0.0005, HR = 0.4; PFS, p=0.0015, HR = 0.42), outperforming PD-L1 and INF-G signature. The markers remain significant after adjusting for HPV status in the multivariate model. TMB (using a previously established cutpoint of 175mt/exome=10mt/MB) was also associated with PFS in the overall population (OS, p=0.13; PFS, p=0.016), but TMB performed much better in HPV(-) SCCHN (OS, p=0.0083; PFS, p=0.043). Interestingly, IDO1/HLADR expression and TMB were only weakly correlated (r=0.20, P=0.078). Four distinct subgroups of tumors were identified using this combination with the combined IDO1/HLA-DRhi + TMBhi group having 60% OS plateau. HPV (+) patients had significantly higher levels of IDO1/HLADR expression in their tumors compared with HPV(-) patients (p<0.0001). Unique biologic characteristics of each of the four subgroups were also identified.

Conclusions

Conclusions: Our results show that IDO1/HLADR expression by tumor cells is highly predictive of outcome in SCCHN patients treated with anti-PD-1 therapy. TMB provides distinct and complementary predictive information and performs particularly well in HPV(-) tumors. The combined IDO/HLA-DR – TMB analysis identifies unique subgroups of patients, which are relevant for combination drug development as well as potential clinical care after further prospective validation.

References

1. Seiwert TY, Burtness B, Mehra R, Weiss J, Berger R, Eder JP, Heath K, McClanahan T, Lunceford J, Gause C, Cheng JD, Chow LQ. Safety and clinical activity of pembrolizumab for treatment of recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-012): an open-label, multicentre, phase 1b trial. Lancet Oncol. 2016 Jul;17(7):956-965.

Ethics Approval

The study was approved by the IRB of the University of Chicago (IRB# 8980)

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal

P66 Evaluation of Immune – related Markers in circulating proteome and their association with atezolizumab efficacy in patients with 2L+ NSCLC

Marcin Kowanetz, PhD1, Ning Leng1, Joanna Roder, PhD2, Carlos Oliveira, PhD2, Senait Asmellash, PhD2, Krista Meyer, PhD2, Heinrich Roder, DPhil2, Marcus Ballinger, PhD1, David Shames, PhD1
1Genentech, South San Francisco, CA, USA; 2Biodesix, Steamboat Springs, CO, USA
Correspondence: Marcin Kowanetz (kowanetz.marcin@gene.com)

Background

Anti-PD-L1/PD-1 therapy has become a standard of care in NSCLC. However, understanding of the biological mechanisms of treatment efficacy and resistance is still incomplete. Here we examine the role of the circulating proteome in 2L+ NSCLC patients treated with atezolizumab (anti-PD-L1).

Methods

Using expression data for the circulating proteome from mass spectrometry of pre-treatment serum samples collected from 77 patients with NSCLC treated with atezolizumab (atezo) in a single arm Phase 1a study (NCT01375842) (development cohort) and machine learning methods, a test was developed to classify patients into “Good” and “Poor” prognosis groups. Protein set enrichment analysis (PSEA) was used to compare the underlying biology of classification of Good and Poor phenotypes. Blinded pre-treatment serum samples from 270 patients treated with atezo or docetaxel (doc) in the randomized Phase 2 study POPLAR (NCT01903993) were subsequently used for validation.

Results

The test stratified patients treated with atezo in the development cohort by overall survival (OS) (Good vs. Poor prognosis HR=0.23, p <0.001) and progression-free survival (PFS) (Good vs. Poor prognosis HR = 0.52, p = 0.012). PSEA identified trends in association of increased complement activation, acute inflammation and immune response type 2 in the Poor classification phenotype. In the blinded analysis of the validation cohort (POPLAR), 262 samples (97%) passed QC, and 134 (51%) were classified to the Good prognosis group. In the validation cohort, OS and PFS were associated with atezo efficacy compared to doc in the Good but not in the Poor prognosis group, with unadjusted classifier-treatment interaction p-value 0.005 for PFS and interaction p-value 0.001 for OS (Figure 1).

Conclusions

Patients characterized by complement activation, acute inflammation and immune response type 2 markers in their baseline serum appeared to derive less benefit from atezolizumab. The analysis of circulating-proteome-defined phenotypes may help to better understand the biological mechanisms beyond response and resistance to checkpoint inhibition in cancer patients.

Trial Registration

NCT01375842, NCT01903993
Fig. 1 (abstract P66).
Fig. 1 (abstract P66).

See text for description.

P67 Innovative combinatorial approach to characterize the immune landscape and analyze the tumor response after anti-PD-1 blockade in a 3D ex-vivo tumoroid system of non-small cell lung cancer

Melba Marie Page, PhD, Melanie Mediavilla-Varela, PhD, Jenny Kreahling, PhD, Soner Altiok, MD, PhD
Nilogen Oncosystems, Tampa, FL, USA
Correspondence: Melanie Mediavilla-Varela (melanie@nilogen.com)

Background

Cell lines and mouse models have provided valuable information in understanding the tumor microenvironment. The overall success in translating these results to the clinic is less than 10% and lacks the full complexity of the tumor microenvironment. Understanding these greater complexities is critical to the development of immuno-oncology therapies. Nilogen Oncosystems’ proprietary 3D-EXSM ex vivo drug screening platform analyzes fresh patient tumor tissue that remains embedded in its natural environment as a tumoroid. This approach includes a powerful combination of confocal image analysis, flow cytometry analysis, cytokine assays as well as gene expression analysis. With this model, we can visualize response, phenotype the tumor microenvironment and accurately determine the response to checkpoint inhibitors in non-small cell lung cancer (NSCLC) and correlates can be found among different platforms.

Methods

For the 3D-EX vivo platform, tumoroids were shaped from procured fresh tumor tissue from NSCLC cancer patients. They were then treated with Keytruda ex-vivo and treatment-mediated changes in TIL subpopulations were analyzed using confocal analysis, flow cytometric analysis, cytokine release by Bio-Rad’s 17-plex cytokine assay as well as gene expression by NanoString’s 770 gene Immune Panel.

Results

Ex vivo treatment of the 3D tumoroids with Keytruda, showed significant changes in T-cell activation and immune cell populations in 26% of NSCLC tumors. This was observed by the simultaneous increase in IFN-γ and TNF-α upon cytokine analysis. Furthermore, we found a differential expression of signature genes such as CD8, CXCL10, CXCL9, EOMES, Granzyme A/B, IFN-γ, related to T-cell subpopulations via Nanostring analysis, which was accompanied by an increase in Granzyme B via flow cytometry. Confocal imaging analysis allowed detection of immune cell mediated killing of tumor cells upon ex-vivo drug treatment.

Conclusions

The positive and negative associations between expression of immune function genes, TIL activation, and cytokine production by ex-vivo treatment shows that Nilogen Oncosystems’ 3D-EXSM platform can accurately recapitulate the tumor microenvironment. This comprehensive approach including the powerful visualization by confocal analysis provides profound technological advantages in analyzing the tumor immune microenvironment. With this combinatorial approach, the 3D-EXSM platform delivers a better understanding of the mechanism of action of immuno-oncology drugs that may aid in developing biomarkers that can be used for patient selection.

P68 Using high dimensional mass cytometry (CyTOF) and machine assisted analysis to detect biomarkers in the immunotherapy of cancer

Carsten Krieg, PhD1, Silvia Guglietta1, John Wrangle, MD1, Luis Cardenas, BS1, Mitchell Levesque2, Reinhard Dummer, MD2, Burkhard Becher3, Mark Rubinstein, PhD1, Mark Robinson4
1Medical University of South Carolina (MUSC), Charleston, SC, USA; 2University Hospital Zurich, Zurich, Switzerland; 3University Zurich, Zurich, SC, Switzerland; 4University Zurich (IMLS), Zurich, Switzerland
Correspondence: Carsten Krieg (KriegC@musc.edu)

Background

Immunotherapy has created a lot of enthusiasm in oncology but, as not all patients respond, it becomes evident that the strategy “one drug fits all” is not applicable to all patients and all cancers. Therefor, selection of patients benefiting from mono-immunotherapy, combination therapy or advice on which drug to use on therapy, ideally by biomarker, is in high demand.

Methods

Here we approached the problem by designing a customized workflow by using high-dimensional single cell mass cytometry combined with machine-learning bioinformatics for the in-depth characterization of single immune cells in predicting and monitoring immune responses. The analysis is data driven, can be adapted to high throughput approaches and can model arbitrary trial designs such as batch effects and paired designs. We tested our workflow on two studies:

Results

a) Predict response to anti-PD-1 immunotherapy in melanoma. In our discovery cohort peripheral blood mononuclear cells (PBMCs) from 20 patients with stage IV melanoma before and 12 weeks on anti-PD-1 therapy was analyzed. We observed a clear T cell response on therapy. The most evident difference in responders before therapy was an enhanced frequency of CD14+ CD16+HLA-DRhi classical monocytes. We validated our results using conventional flow cytometry in an independent exploratory cohort of 31 patients before therapy. Finally, we correlated enhanced monocyte frequencies before therapy initiation with clinical response and could show association with lower hazard, extended progression-free and overall survival.b) Monitoring the immune response in non small cell lung cancer patients refractory to anti-PD-1 immunotherapy treated with a novel combination immunotherapy of anti-PD-1 and an IL-15 superagonist. 21 patients with non small cell lung cancer who got refractory to anti-PD-1 treatment received a novel combination therapy of IL-15 superagonist plus anti-PD-1. A response in the CD8+ T cell compartment was observed characterized, among other factors, by expansion of TCR variety. Unexpected we observed a strong expansion of effector NK cells starting around day 4 of therapy.

Conclusions

Taken together, high throughput mass cytometry together with an unbiased artificial intelligence driven analysis workflow might support patient selection prior to therapy, select the right drug combination and identify new drugable cell populations.

Ethics Approval

These studies have been approved by the University of Zurich and Medical University of South Carolina Ethics Boards.

P69 Simultaneous quantification of activated immune cells and PD-L1 expressing circulating tumor cells (CTCs) in peripheral blood of cancer patients receiving checkpoint inhibitor therapy

Rachel Krupa, BS, Robin Richardson, Priscilla Ontiveros, Joseph Schonhoft, Yipeng Wang, PhD, MD, Jiyun Byun, MS, David Lu, Aaron Oh, Sean Nisperos, Mark Landers, MS, Ryan Dittamore, MS
Epic Sciences, San Diego, CA, USA
Correspondence: Rachel Krupa (rachel.krupa@epicsciences.com)

Background

Expression of PD-L1 on tumor and tumor infiltrating lymphocytes has been associated with improved response to PD-1 and PD-L1 checkpoint inhibitors, however clinical utility is limited. 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 by biopsy given tumor heterogeneity, clonal evolution, and temporal changes in the host immune system under therapeutic pressure. Therefore, we sought to expand the existing Epic Sciences’ non-invasive liquid biopsy platform to simultaneously examine expression of PD-L1 on circulating tumor cells (CTCs) as well as quantify changes in activated immune cell populations from a single sample in patients undergoing checkpoint inhibitor therapy. Examining dynamic biomarker changes both prior to and on therapy could yield novel diagnostic tools for response prediction and pharmacodynamics response for immunotherapy.

Methods

Blood samples from bladder, kidney, and prostate cancer patients undergoing checkpoint inhibitor therapy were collected at baseline and on-therapy (when available). Contrived samples were developed using cancer cell lines and healthy donor (HD) blood. Red blood cells were lysed and nucleated cells were plated onto glass slides and stained with DAPI and immune panels including pan-CK, CD45, PD-L1, CD4, CD8, and Ki-67. Approximately 3 million cells per slide were imaged through advanced digital pathology pipelines to detect and quantify changes in immune cell populations and to assess circulating tumor burden.

Results

An immuno-panel was developed to profile activated (CD8+Ki-67+ and CD4+Ki-67+) leukocyte subpopulations and PD-L1+ CTCs from a single blood sample. Feasibility was demonstrated in cell lines, HD, and patient blood. CTCs were detected in 76% (32/42) of samples tested. Of the 25 baseline samples tested, 12% (3/25) had PD-L1+ CTCs detected. No PD-L1+ CTCs were detected in the 13 on-therapy samples tested. Of 14 patients with matched samples, 50% (7/14) patients had an increase in activated CD4+ leukocytes and 14% (2/14) patients had an increase in activated CD8+ leukocytes in on-therapy samples compared to baseline. A decrease in activated CD4+ leukocytes was observed in 7% (1/14) patients and a decrease in activated CD8+ leukocytes was also observed in 7% (1/14) patients with therapy.

Conclusions

Development of a liquid biopsy based platform that can simultaneously measure immune biomarkers in CTCs and leukocytes will allow for real time assessment and monitoring of response to immune checkpoint inhibitors and may lead to novel diagnostic tools for response prediction.

P70 CD4+FOXP3+ regulatory T cells in the periphery of HNSCC patients demonstrate high phenotypic diversity depending on Treg subtype

Cornelius Kürten, MD1, Shanhong Lu2, Tullia Bruno, PhD3, Robert L. Ferris, MD, PhD3
1University of Essen, Essen, PA, Germany; 2Centre South University, Pittsburgh, PA, USA; 3University of Pittsburgh, Pittsburgh, PA, USA
Correspondence: Robert L. Ferris (ferrisrl@upmc.edu)

Background

Regulatory T cells (Treg) promote immune escape and are a putative biomarker for response to immunotherapy.

Indeed, previous data from our group demonstrated that Ki67+ Treg were elevated in head and neck cancer (HNSCC) responders to nivolumab (CheckMate-141). Importantly, phenotypic markers for different subsets of Treg have been described: naïve Treg (nTreg, CD45RA+FOXP3low), effector Treg (eTreg, CD45RA-FOXP3high) and non-suppressive Treg (nsTreg, CD45RA-FOXP3low). While these populations have been described in peripheral blood lymphocytes (PBL) and tumor-infiltrating lymphocytes (TIL) of cancer patients, the presence of suppressive eTreg in the blood is a biomarker for disease progression in HNSCC patients. Thus, a more detailed characterization of this cell population is warranted.

Methods

Multi-parameter flow cytometry (CD8, CD4, CD45RA, FOXP3, Neuropilin-1, CD39, PD-1, Tim-3, LAG-3, CTLA-4, TIGIT, CD69, pAKT, Ki67, Bcl2) was performed on PBL from HNSCC patients (n = 50). T cell function was assessed using intracellular staining for cytokine production (IFNy, TNFa) after a 4 hour stimulation with PMA/Ionomycin. Findings were correlated with patient demographics and clinical outcome.

Results

A higher percentage of eTreg express inhibitory receptors (IRs) (two-way ANOVA: PD-1: mean diff. 22%, p <0.0001; Tim-3: mean diff. 15 %, p < 0.01; CTLA-4: mean diff. 13 %, p < 0.05; TIGIT: mean diff. 26 %, p < 0.0001; CD39: mean diff. 40 %, p < 0.0001) compared to naïve/non-suppressive Treg population. IR expression correlated highly on eTreg (Spearman correlation e.g. for PD-1 with Tim-3 r2 = 0.70, PD1 with TIGIT r2 = 0.72, PD-1 with CD39 r2 = 0.82, all p < 0.0001), but not on the non-effector Treg. Interestingly, proliferative eTreg (measured by Ki67 expression) were strongly associated with better survival (Hazard Ratio: 0.29, p < 0.05).

Conclusions

eTreg are a distinct cell subtype in HNSCC, as shown by their differential expression of IRs and proliferation/activation markers. The higher IR expression suggests that eTreg are an important target of currently available immunotherapy drugs among the whole Treg population. This also underscores the putative value of this cell subset as a biomarker of response to monitor longitudinally during anti-PD-1 immunotherapy.

P71 Characterization of tumor mutational burden (TMB) and homologous recombination repair (HRR) mutations to assess correlation with immune checkpoint inhibitors (ICIs) response in renal cell carcinoma

Matthew Labriola, MD1, Jason Zhu, MD2, Rajan Gupta1, Shannon McCall, MD1, Jennifer Jackson, PhD3, James White, PhD3, Elizabeth Weingartner3, Eric Kong3, Peter Simone, PhD4, Eniko Papp, PhD3, Kelly Gerding, PhD3, Eun-Hae Kim, PhD3, John Simmons, PhD3, Daniel George, MD2, Tian Zhang, MD2
1Duke University Hospital, Durha, NC, USA; 2Duke Cancer Institute, Durham, NC, USA; 3Personal Genome Diagnostics, Baltimore, MD, USA; 4Pesronal Genome Diagnostics, Baltimore, MD, USA
Correspondence: Tian Zhang (tian.zhang2@duke.edu)

Background

The advent of immune checkpoint inhibitors (ICI) has revolutionized the treatment landscape for patients with metastatic renal cell carcinoma (mRCC) [1,2]. However, traditional biomarkers such as PD-L1 have not served as predictive markers of treatment response. Given the risk of toxicity and variable response rates, there is a need to develop more reliable predictive biomarkers to support precision immunotherapy. High tumor mutational burden (TMB) has been previously described as a robust biomarker for predicting ICI response in metastatic melanoma [3] and non-small cell lung cancer [4], but has not yet been fully explored in mRCC. Here, we describe the prediction of clinical outcomes for mRCC patients and ICI response using a solid tissue-based next-generation sequencing (NGS) assay to identify genetic correlates and tumor mutation burden.

Methods

34 patients with mRCC who had received ICI therapy at Duke Cancer Institute were identified. FFPE tumor samples from archival tissue banks were evaluated using Personal Genome Diagnostics elioTM tissue complete investigational NGS assay, screening for somatic variants across >500 genes, as well as TMB and microsatellite status. Clinical information was extracted from the medical record and tumor response was evaluated based on RECIST 1.1 criteria.

Results

16 of 34 patients displayed disease control (overall responses of: stable disease, partial response, or complete response) following ICI therapy. This patient cohort displayed a range of TMB scores from 0.4 to 12.2 mutations/Mb (predicted whole exome equivalent), with a mean TMB score of 2.8 mutations/Mb, exome equivalent. Overall, there was no significant difference in TMB scores between responders and non-responders, and no significant correlation between increased TMB score and response to ICI was found (Figure 1). Interestingly, genes related to DNA repair pathways, particularly homologous recombination (including FAM175A, RAD50, RECQL4, and SLX4), were more often found to be mutated in the ICI responder group compared to ICI non-responders. Of the 16 responders, 9 were found to harbor somatic mutations in at least one gene associated with DNA repair (Figure 2).

Conclusions

Overall, TMB did not appear to correlate to patient outcomes or ICI response in this mRCC patient cohort. However, NGS analysis showed an increase in somatic mutations in DNA repair genes in responders compared to non-responders. Recently, hereditary RCC syndromes have been mechanistically linked to defects in homologous recombination [5]. Our findings suggest that mutations in DNA repair pathway genes may correlate with ICI response and may have potential as a predictive biomarker for treatment success.

Acknowledgements

Study supported by research funds from Personal Genome Diagnostics (PGDx).

References

1. Motzer RJ, Escudier B, McDermott DF, et al. Nivolumab versus everolimus in advanced renal-cell carcinoma. New England Journal of Medicine. 2015; 373(19): 1803-13.

2. Motzer RJ, Tannir NM, McDermott DF, et al. Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma. New England Journal of Medicine. 2018; 378(14): 1277-90.

3. Goodman AM, Kato S, Bazhenova L, et al. Tumor mutational burden as an independent predictor of response to immunotherapy in diverse cancers. Mol Cancer Ther. 2017; 16(11): 2598-608.

4. Hellmann MD, Ciuleanu TE, Pluzanski A, et al. Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden. New England Journal of Medicine. 2018; 378(22): 2093-104.

5. Sulkowski PL, Sundaram RK, Oeck S, et al. Krebs-cycle-deficient hereditary cancer syndromes are defined by defects in homologous-recombination DNA repair. Nat Genet. 2018. Epub ahead of print.

Ethics Approval

This study was approved by Duke University’s Institutional Review Board, protocol number Pro00088779.
Fig. 1 (abstract P71).
Fig. 1 (abstract P71).

See text for description.

Fig. 2 (abstract P71).
Fig. 2 (abstract P71).

See text for description.

P72 Biomarker and preliminary pharmacodynamic evaluations of the PD-1 inhibitor ABBV-181 from an ongoing phase 1 clinical trial in patients with advanced solid tumors

Stacie Lambert, Gregory Vosganian, MD, Fiona Harding, PhD, James Sheridan, Stefan Englert, Dr, Sara Siggelkow, Apurvasena Parikh, PhD, Betty Wang, Kinjal Hew, Jaishree Bankoti, Marcia Stickler, Eileen Lee, Merriam Mcclellan, Daniel Afar, PhD, Anita Reddy, PhD, Stacie Lambert
AbbVie, Inc, Redwood City, CA, USA
Correspondence: Stacie Lambert (stacie.lambert@abbvie.com)

Background

ABBV-181 is a humanized, recombinant, modified IgG1 monoclonal antibody targeting programmed cell death 1 (PD-1). Here we present preliminary analyses of ABBV-181 pharmacodynamic (PD) data from an ongoing Phase 1 study in patients with solid tumors (NCT03000257).

Methods

Prior to initiation of clinical testing, ABBV-181 was characterized in vitro for PD-1 blocking ability and bioactivity and a PD-1 saturation assay was developed. In the ongoing Phase 1 study, patients with previously treated advanced solid tumors received ABBV-181 at 1, 3, or 10 mg/kg intravenous once every 2 weeks (Q2W) in dose escalation. Following dose finding, multi-histology, non-small cell lung cancer (NSCLC) and head and neck squamous cell cancer (HNSCC) cohorts were opened at 250 mg Q2W. PD-L1 expression was assessed on pre-treatment tumor samples (PD-L1 immunohistochemistry 28-8 pharmDx, 1% threshold). Whole blood samples were collected pre and post dosing to measure PD-1 receptor saturation and expression of Ki-67 and other biomarkers in circulating T cell populations by flow cytometry, while serum was collected pre and post dosing to measure cytokine biomarkers. Pharmacokinetic (PK) sample collections and PK-PD analyses were also conducted.

Results

In vitro characterization demonstrated that ABBV-181 blocked PD-1 binding to its ligands and increased IFNγ production in functional human and non-human primate peripheral blood mononuclear cell assays. T cells from non- human primates administered ABBV-181 showed PD-1 saturation. Preliminary clinical PD data were available for 72 patients treated with ABBV-181 (n=25 in dose escalation, n=47 in ongoing 250 mg Q2W dose expansion). PD- L1 was expressed on 32/60 (53%) available pretreatment tumor samples, with higher expression rates in the NSCLC and HNSCC cohorts compared to the multi-histology cohort. Rapid and sustained PD-1 saturation on circulating CD4 T central memory cells was observed at all ABBV-181 doses tested. A transient decrease in circulating T cell counts following dosing was also observed, consistent with prior reported clinical observations of PD-1 blocking agents. Increases in Ki-67+ CD8+ T cells within the first cycle were detected in approximately half of the tested patients. Significant increases in serum chemokines including IP-10 were also found post dosing with ABBV-181.

Conclusions

Preliminary results indicate that ABBV-181 demonstrates PD-1 receptor saturation and biological activity in peripheral blood at all clinical doses tested. PD data is consistent with the biological activities previously reported for other PD-1 blocking antibodies. Enrollment in these expansion cohorts continues at 250 mg Q2W and 500 mg once every 4 weeks as supported by ongoing PK-PD analyses.

Acknowledgements

AbbVie and the authors thank the patients participating in this clinical trial and all study investigators for their contributions.

Trial Registration

ClinicalTrials.gov, NCT03000257

Ethics Approval

This study (NCT03000257) was approved by each participating institution’s Ethics Board.

P73 Analysis of survival and mRNA expressivity in the tumor microenvironment of adenocarcinoma via K-means clustering algorithm

Sunyoung Lee, MD, PhD1, Andrew Baird, MD3, Jillian Dolan, BS4, Stuart Baird5, Fateeha Furqan, MD6, Shinyoung Park, MS1
1Roswell Park Cancer Institute, Williamsville, NY, USA; 2National Institute for Mathematical Sciences, Seoul, Korea, Republic of; 3University of Pittsburgh Medical Center, Pittsburgh, PA, USA; 4University at Buffalo School of Medicine, Buffalo, NY, USA; 5St. Lawrence University, Canton, NY, USA; 6Rochester General Hospital, Rochester, NY, USA
Correspondence: Sunyoung Lee (syandsy@gmail.com)

Background

Stromal elements in tumor microenvironment (TME) impact the response to cytotoxic chemotherapy and immunotherapy. Advances in mRNA-sequencing have improved our understanding of TME expressivity. However, few models exist to analyze immune crosstalk between TME elements and mRNA expressivity in terms of patient survival.

Methods

mRNA-seq of 3,758 adenocarcinoma tumors and 314 non-tumor tissues (lung, breast, esophageal, gastric, colorectal, pancreatic, ovarian, endocervical, and prostate adenocarcinoma) were obtained from the Cancer Genome Atlas (TCGA) and analyzed based on mRNA expression. The mRNA expressivity of 195 genes enriched in stromal components were arranged into 26 gene groups (tables 1 and 2). Using mRNA expressivity via K-means algorithm (50 cycles of machine learning), patients were clustered into two groups (high and low mRNA expression) for each gene group. Kaplan-Meier and correlation analyses were performed to assess the significance of each gene group in survival.

Results

Genes associated with immune activation correlate with better survival until 60 (lung), 110 (breast), 50 (esophageal), and 20 months (pancreatic). There is negative correlation with survival in gastric and colorectal, no correlation in ovarian, and positive correlation in endocervical. Angiogenesis is negatively associated with survival in colorectal (p<0.1), lung and gastric (p<0.05). Genes related to desmoplasia and immunosuppressive chemokines have a significant association with poor survival in gastric, colorectal, and endocervical (p<0.05), as well as ovarian (p<0.1). Pancreatic has extensive expression of genes associated with desmoplasia and immunosuppressive chemokines, but degree of expressivity does not correlate with survival. Neutrophils are negatively associated with survival in gastric and endocervical (p<0.05), as well as esophageal and ovarian (p<0.1), but positively correlate in breast (p<0.1). Genes associated with cancer stem cells negatively correlate with survival in pancreatic. In prostate, expressivity of immune-related genes was low, and no correlation was found except for genes related to type II IFN, which positively correlate with survival. Patients with enhanced expression of genes associated with type I IFN and antigen presentation in tumor tissue show increased expression of these genes in their respective non-tumor tissue samples.

Conclusions

Analysis of large data was assisted by K-means (machine learning) algorithm, showing that stromal genes have varied impact on survival in adenocarcinoma. Genes associated with immune activation have temporal correlation with survival, which seems to be a result of tumor immune escape. Expressivity of type I IFN and antigen presentation in non-tumor tissues is conserved in tumor tissues. Future prospective studies in response to chemotherapy and immunotherapy are warranted.
Table 1 (abstract P73).

Gene associated with each gene group

Table 2 (abstract P73).

Gene groups and association with survival in adenocarcinoma, analyzed by Kaplan-Meir analysis

P74 Validation the immune contexture as prognostic biomarkers in high-grade serous ovarian cancer

Shin Wha Lee, MD1, Ju-hyun Kim1, Soo-Jung Kim2, Yong-Man Kim1
1Ulsan University, ASAN Medical Center, Seoul, Korea; 2ASAN Institute for Life Science, Seoul, Korea, Republic of
Correspondence: Yong-Man Kim (ymkim@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.

Acknowledgements

This study was supported by the R&D Project, Asan Institute for Life Sciences, Seoul, Korea (2016-588).

P75 Development of a next-generation sequencing-based microsatellite instability assay (MSI-NGS) for solid tumor testing

Sean Glenn, PhD, Sarabjot Pabla, MSc, PhD, BS, Jonathan Andreas, MS, Blake Burgher, BS, RN, Jacob Hagen, Jeffrey Conroy, BS, Mary Nesline, MS, Antonios Papanicolau-Sengos, MD, Vincent Giamo, BS, MS, Felicia Lenzo, Maochun Qin, MD, MS, Yirong Wang, MS, Mark Gardner, Carl D. Morrison, MD, DVM
OmniSeq, Inc., Buffalo, NY, USA
Carl D. Morrison (carl.morrison@omniseq.com)

Background

Microsatellite instability (MSI) is as a screening test for Lynch syndrome (HNPCC), is FDA-approved as a companion diagnostic for checkpoint inhibition, and has demonstrated high positive predictive value for response to anti-PD1 therapy for MSI-high/dMMR patients. Typically, MSI analysis involves comparison of allelic profiles of microsatellite markers generated by amplification of DNA from matching normal and tumor samples using a fluorescent PCR-based assay. The requirement of needing matched normal DNA has limited the number of patients that can have testing performed due to the difficulty in obtaining adjacent benign tissue in the tumor specimen received. Utilizing advances in NGS technologies and bioinformatics, we have developed a targeted, multi-plexed NGS assay (MSI NGS) which robustly and accurately determines MSI status without the requirement of a matched normal DNA.

Methods

Utilizing previously published data sets, 29 highly significant loci within the genome were determined for interrogation and integration into the MSI NGS targeted assay. For each loci, the number of peaks and average indel lengths were utilized in a custom algorithm on a gold standard training set, where MSI status was previously determined by MSI PCR methods, to define cluster centroid metrics which determine MSI and MSS status. By this process cluster 1 and cluster 2 were assigned as “MSI” and cluster 3 was assigned as “MSS” with 100% PPV and 96% NPV. The Euclidean distance of each experimental sample is then calculated across all 29 loci, and the cluster centroid closest to the sample determines the MSI status.

Results

Utilizing a 100 gold-standard sample set the overall concordance of the MSI NGS assay to MSI PCR is extremely high with only two false negative calls (98% concordance). The two false negative cases can be attributed to the fact that these cases are in cluster 1 which is closer (Euclidean distance) to the centroid of cluster 3 (MSS) than the centroid of cluster 2 (MSI). To this end, the reported sensitivity and specificity of the accuracy study are 96% and 100% respectively with a PPV of 100% and an NPV of 96%.

Conclusions

The development of a MSI NGS assay, which has recently been NYS-CLEP approved for all solid tumors, allows for robust and accurate testing of MSI status without the need of matched normal tissue, a major hurdle with conventional testing.

Ethics Approval

OmniSeq’s analysis utilized deidentified data that qualified as non-human subject research under IRB protocol (BDR #073166) approved by Roswell Park Comprehensive Cancer Center (Buffalo, NY).

P76 Analysis of the complex immune-cell milieu of tumors from patients treated with immunotherapy to better understand clinical response

Shumei Kato, MD1, Ryosuke Okamura1, Mina Nikanjam1, Ramez Eskander, MD1, Paul Fanta, MD1, Suzanna Lee1, Sean Glenn, PhD2, Devin Dressman, PhD2, Sarabjot Pabla, MSc, PhD, BS2, Jeffrey Conroy, BS2, Mary Nesline, MS2, Antonios Papanicolau-Sengos, MD2, Felicia Lenzo2, Mark Gardner2, Carl Morrison, MD, DVM2, Razelle Kurzrock, MD1
1UC San Diego Moores Cancer Center, La Jolla, CA, USA; 2OmniSeq, Inc., Buffalo, NY, USA
Correspondence: Razelle Kurzrock (rkurzrock@ucsd.edu)

Background

Although immunotherapies, especially checkpoint inhibitors, have achieved salutary anti-cancer effect among patients with advanced cancers, most patients do not respond to immunotherapies. We comprehensively evaluated biomarkers associated with the “cancer-immunity cycle” among patients with diverse solid tumors to understand the immune landscape in metastatic cancers and the resistance mechanism for anti-PD1/PDL1 inhibitors.

Methods

Interrogation of key markers of the cancer-immunity cycle was carried out in patients (n=101) with diverse malignancies using a NYS Clinical Laboratory Evaluation Program (CLEP) approved targeted RNA sequencing assay (51 genes) developed in a Clinical Laboratory Improvement Amendments (CLIA) certified laboratory. Resultant gene expression data was QC filtered, normalized and ranked based on an assorted reference population of various tumor types. Gene signatures were determined using these ranked values with a rank value > 85th percentile considered high.

Results

The immune phenotypes of the patients’ tumor milieu demonstrated overexpression of multiple checkpoint blockade markers including PD-L1 (6.9%), PD-L2 (10.9%), CTLA4 (3%), LAG-3 (8.9%), TIM-3 (9.9%) and VISTA (15.8%). Overexpression of other cancer-immunity cycle markers were also observed including myeloid suppression markers (e.g. CCL2, CCR2 and CSF1R; 10-22%), metabolic immune escape markers (e.g. ADORA2A and IDO1; 9-16%) and T-cell primed markers (e.g. CD40, GITR, ICOS and OX40; 4-26%). Each patient had a unique cancer- immunity expression pattern that was distinctive from others in this cohort. Overexpression of TIM-3 and VISTA was associated with significantly shorter progression-free survival (PFS) from anti-PD1/PDL1 based therapies (P=0.007 and P=0.001 respectively).

Conclusions

Evaluation of the gene expression levels of biomarkers associated with the cancer-immunity cycle was feasible among diverse solid tumors using targeted RNA sequencing. Checkpoint blockade markers (TIM-3 and VISTA) were associated with shorter PFS with anti-PD1/PDL1 based therapies. All patients had unique immune related expression profiles suggesting more extensive molecular profiling beyond tumor mutation burden and PD-L1 status may be essential to personalize treatment options such as individualized combination immunotherapy.

Ethics Approval

All investigations followed the guidelines of the UCSD Institutional Review Board for data collection (Profile Related Evidence Determining Individualized Cancer Therapy, NCT02478931) and for any investigational therapies for which the patients consented. OmniSeq’s analysis utilized deidentified data that qualified as non-human subject research under IRB protocol (BDR #073166) approved by Roswell Park Comprehensive Cancer Center (Buffalo, NY).

P77 Identifying major immune-related subsets of GI tumors for clinical purposes

Amy Early, MD1, Sarabjot Pabla, MSc, PhD, BS2, Jeffrey Conroy, BS2, Mary Nesline, MS2, Sean Glenn, PhD2, Felicia Lenzo2, Antonios Papanicolau-Sengos, MD2, Blake Burgher, BS, RN2, Vincent Giamo, BS, MS2, Jonathan Andreas, MS2, Yirong Wang, MS2, Carl D. Morrison, MD, DVM2
1Roswell Park Comprehensive Cancer Center, Williamsville, NY, USA; 2OmniSeq, Inc., Buffalo, NY, USA
Correspondence: Carl D. Morrison (carl.morrison@omniseq.com)

Background

Gastrointestinal (GI) tumors, both colorectal and non-colorectal, have a low response rate to immune checkpoint inhibitors (ICIs) outside of the setting of microsatellite (MSI) unstable. Currently there is uncertainty how to evaluate microsatellite stable GI tumors for evidence of checkpoint blockade as PD-L1 IHC has minimal applications in this setting. Furthermore understanding other mechanisms of immunosuppression in GI tumors, such as myeloid or metabolic suppression and the degree of CD8+ T-cell infiltration, can have profound influence on the selection of immunotherapies in this patient population.

Methods

131 MSI stable GI tumors (82 colorectal, 41 pancreatic, 8 small bowel) from multiple institutions were evaluated for PD-L1 expression by IHC, TMB (DNA-seq), and expression of 54 immune-related genes (RNA-seq) that are the target of multiple immunomodulatory immunotherapeutics or evaluate tumor infiltrating lymphocytes (TILs) in a CLIA setting.

Results

PD-L1 IHC was positive (TPS>=1%) in 32 (24%) of cases, but with only 3 (2%) of cases strongly positive (TPS>=50%). In PD-L1 IHC negative cases (n=99; TPS<1%) the corresponding PD-L1 RNA-seq value was low in 88 (89%) and moderately high to high in 11 cases (11%). A high macrophage content was identified in 50 (31%) cases indicative of strong myeloid suppression. All of these cases over expressed one or more myeloid-related immunotherapeutic targets including CCR2, CSF1R, or TGFB1. The majority of these 50 cases (n=37; 74%) also over expressed VISTA or TIM3, or both, indicating that over expression of these two checkpoint blockade receptors are strong indicators of myeloid suppression. In a similar fashion PD-L2 was frequently over expressed in these 50 cases (n=23; 46%). While a low number of CD8+ T-cells were common in cases with strong myeloid suppression (6/50;12%) it was more common in the remaining 81 cases at 47% (38/81). This latter group of cases with few exceptions can be aptly described as immune deserts.

Conclusions

Using a more sophisticated approach to evaluating the tumor microenvironment in GI tumors 3 major groups including PD-L1 positive, myeloid suppression, and immune deserts can be identified that has major implications for the application of precision immunotherapy for this patient population.

Ethics Approval

OmniSeq’s analysis utilized deidentified data that qualified as non-human subject research under IRB protocol (BDR #080316) approved by Roswell Park Comprehensive Cancer Center (Buffalo, NY).

P78 More sensitive identification of T-cell receptor beta rearrangements with an augmented transcriptome method

Eric Levy, PhD, Sean Boyle, PhD, Gabor Bartha, Pamela Milani, Robin Li, Shujun Luo, Rena McClory, John West, MBA, Richard Chen
Personalis, Inc., Menlo Park, CA, USA
Correspondence: Eric Levy (eric.levy@personalis.com)

Background

With the growth of new immunotherapies, there is an increasing need for comprehensive immuno-genomic profiling of tumors to identify new potential biomarkers. This includes neoantigen identification, HLA, immuno-modulators, tumor microenvironment, and T-cell receptor (TCR) repertoire. However, limited sample amount, formalin-fixed paraffin embedded tissue (FFPE) degradation, and cost of multiple sequencing assays pose a significant barrier to comprehensive immuno-genomics biomarker characterization in clinical trials.To address these challenges, we developed an augmented, immuno-oncology optimized exome/transcriptome platform (ACE ImmunoID) that can also identify abundant TCR clones, from limited FFPE tumor biopsy samples.

Methods

We designed the next generation of our ACE ImmunoID platform to augment RNA profiling of TCR and BCR, including TCR beta. We next characterized the performance of our platform at profiling TCR beta. First, we analyze the impact of sequencing depth and input material amount on the observed repertoires of diverse PBMC samples. We test LOD by diluting well-characterized clonal T-cell line samples into PBMCs. Finally, we analyze patient- derived FFPE and FF tumors to understand the profiles of tumor-infiltrating immune repertoires and effects of FFPE damage.

Results

We observe that at our specified 400ng of fragmented input RNA and 100M cluster sequencing depth, we detect approximately 26,000 clones, enough to get a picture of the clonality of a sample. Our platform has higher sensitivity to TCR clones compared to non-augmented transcriptome methods at a comparable depth of sequencing. Furthermore, in comparison to a commercially-available deep TCR kit, we identify 93.9% of the top 1000 clones, showing that we confidently identify high-abundance clones. We also are able to reliably identify clones down 0.0004% RNA by mass in the mixture. We further tested on FFPE samples, showing our platform works well with degraded RNA.

Conclusions

Our ACE ImmunoID platform has been designed to enable sensitive detection of major TCR repertoire clones in addition to comprehensive biomarkers from exome/transcriptome results. Here we demonstrate that our platform achieves both a higher sensitivity for TCR clones compared to non-augmented transcriptome approaches and a high concordance with the top abundance clones derived from targeted TCR methods. We show our method is feasible with FFPE samples, making it practical for clinical trial use. In summary, by combining exome/transcriptome with TCR characterization into a single assay, our ACE ImmunoID platform enables comprehensive immuno-genomics characterization of a tumor sample while reducing overall sample requirements and cost.

P79 Change in neutrophil to lymphocyte ratio during treatment with immune checkpoint inhibitors predicts survival in patients with advanced cancer

Mingjia Li, MD, Dan Spakowicz, PhD, MS, Jarred Burkart, Sandip Patel, Marium Husain, MD, MPH, Kai He, MD, PhD, Carolyn Presley, Erin Bertino, Peter Shields, David Carbone, MD, PhD, Claire Verschraegen, MD, Greg Otterson, Kari Kendra, Mingjia Li, MD, Dwight Owen
The Ohio State University, Columbus, OH, USA
Correspondence: Dwight Owen (Dwight.Owen@osumc.edu)

Background

Baseline neutrophil to lymphocyte ratio (NLR) is known to be prognostic for patients with many cancer types treated with immune checkpoint inhibitors (ICI), including non-small cell lung cancer (NSCLC). We evaluated NLR at baseline and during treatment for patients who received ICI to evaluate the prognostic value of the change in NLR over time.

Methods

A retrospective review of patients with advanced cancer who received ICIs from 2011 to 2017 at the Ohio State University was performed with IRB approval. NLR was calculated as ratio of absolute neutrophil/lymphocyte counts, and considered elevated if ≥ 5. Overall Survival (OS) was calculated from the initiation of ICI to death of date or last follow-up. Significance of Cox Proportional-Hazards models were evaluated by log-rank test at Alpha = 0.05. All calculations were performed using the survival and survminer packages in R.

Results

677 patients were included in the analysis. NLR was collected at the initiation of ICI and at least one time after (median 21, IQR 8 days, Chart 1).Patients with baseline NLR ≤5 had median OS 592 days (95% Cl: 497-681) compared to median OS 224 days (95% Cl: 158-269) for patients with baseline NLR >5, P<0.001 (Figure 1). Patients with on-treatment NLR ≤5 had median OS 616 days (95% CI: 532-878) compared to median OS 177 days (95% CI: 143-242, P< 0.001 for patients with on-treatment NLR >5 (Figure 2). Subgroup analysis of 121 NSCLC patients at baseline and on-treatment NLR showed similar prognostic value (Figure 3 and 4).For patients with baseline NLR >5 but where on-treatment NLR normalized to ≤5, there was improved median OS of 433 days (95% Cl: 304-NA) compared to median 150 days (95% Cl: 120-214) for patients when NLR remained high, P< 0.001 (Figure 5). Similar results were seen in NSCLC patient with baseline NLR >5 with on-treatment normalization of NLR (P=0.0018, Figure 6).

Conclusions

We confirmed the prognostic value of baseline NLR in patients with advanced cancer treated with ICI, including metastatic NSCLC. We demonstrated that change in NLR over time may identify patients with poor prognosis at baseline who nevertheless benefit from ICI. To our knowledge, the association between dynamic changes in NLR during treatment with ICI and survival have not previously been reported in NSCLC. This biomarker is especially attractive because NLR can be easily obtained from routine labs.

Ethics Approval

The study was approved by the Ohio State University Institutional Review Board, approval number # 2016C0070
Fig. 1 (abstract P79).
Fig. 1 (abstract P79).

See text for description.

Fig. 2 (abstract P79).
Fig. 2 (abstract P79).

See text for description.

Fig. 3 (abstract P79).
Fig. 3 (abstract P79).

See text for description.

Fig. 4 (abstract P79).
Fig. 4 (abstract P79).

See text for description.

Fig. 5 (abstract P79).
Fig. 5 (abstract P79).

See text for description.

Fig. 6 (abstract P79).
Fig. 6 (abstract P79).

See text for description.

Chart 1 (abstract P79).
Chart 1 (abstract P79).

See text for description.

P80 Day-to-day profiling of T cell activation by measuring cellular markers and secreted cytokines in a rapid multiplexed cell/bead mixture assay

Zhaoping Liu, PhD (zhaopoing.liu@sartorius.com)
Sartorius, Albuquerque, NM, USA

Background

Optimized adoptive cell therapy protocols, and profiling drug candidates against immuno-oncology targets such as immune checkpoint proteins requires precise monitoring of ex vivo activation of human T lymphocytes. Here we describe the development of a large scale multiplexed assay using high throughput flow cytometry to daily profile T cell activation in human PBMCs treated with different activators (Figure 1).

Methods

On Day 0, PBMCs were stained with cell proliferation tracing dye before being plated into a 96-well plate. Cells were treated and cultured with this protocol: 3 different T cell modulators—CD3/CD28magnetic beads, phytohemagglutinin (PHA), and Staphylococcal enterotoxin B (SEB)—with 12 serial titrations and duplicate wells per dose. On culture Days 1, 3, and 6, cell and supernatant mixture samples were removed from the culture plate and evaluated without dilution using a multiplex cell/bead mixture assay that combined cell phenotype, T cell activation markers, cell proliferation, cell viability measurements and secreted cytokine analysis measured by bead-based ELISA. In each sample well of the assay plates, secreted levels of 3 cytokines (IL-4, IFNg and TNFa) were quantified by the standard curves generated from the standard wells in the same assay plate. In addition, over the course of a six-day activation, we generated multiple cellular readouts identifying the presence of different T cell subpopulations expressing early and late T cell action markers CD69, CD25 and HLA-DR. Cell proliferation, cell viability and cell number for different subpopulations were also evaluated. Each plate was read on the iQue Screener Plus system in approximately 20 minutes (Figure 2).

Results

In total, we measured 16 endpoints and generated 1152 data points per 96-well assay plate and 3456 data points from 3 assay plates ( days). We also acquired a total of 144 EC50s or IC50s (16-endpoint EC50/IC50s x 3 Days x 3 Activators), demonstrating the unique signature pattern of 3 different modulators on T cell activation (Figure 3).

Conclusions

These large-scale experiments with high throughput flow cytometry provide extensive T cell activation profiles with rapid assay turnaround time. The use of these assays can provide valuable information for optimizing immuno-oncology drug development such as checkpoint inhibitors and adoptive cell therapy development and manufacturing protocols.
Fig. 1 (abstract P80).
Fig. 1 (abstract P80).

Assay biochemistry

Fig. 2 (abstract P80).
Fig. 2 (abstract P80).

Gating strategy

Fig. 3 (abstract P80).
Fig. 3 (abstract P80).

High content readout

P81 Germline encoded TRBV polymorphism predicts adverse events during checkpoint blockade immunotherapy

Timothy Looney, PhD, Geoffrey Lowman, PhD, Asha Kamat, PhD, Fiona Hyland
Thermo Fisher Scientific, South San Francisco, CA, USA
Correspondence: Timothy Looney (timothy.looney@thermofisher.com)

Background

Identifying predictive biomarkers for immune related adverse events (IRAEs) during immunotherapy is a key objective of current immuno-oncology research. Polymorphism within the TCRB variable gene (TRBV) has been implicated in autoimmune disease and may be mechanistically linked to IRAEs. Efforts to evaluate TRBV polymorphism by traditional approaches such as whole genome sequencing (WGS) have been hampered by the repetitive nature of the TCRB locus and incomplete genome assembly. Here we employed a novel long-amplicon TCRB repertoire sequencing approach to evaluate the link between TRBV polymorphism and adverse events in 54 Caucasians receiving checkpoint blockade immunotherapy for cancer.

Methods

To circumvent the challenge in measuring TRBV polymorphism by WGS, we instead employed next-generation sequencing of rearranged TCRB chains using RNA extracted from peripheral blood. Our strategy utilized multiplex PCR via framework 1 and constant gene primers to create ~330 bp amplicons spanning the three beta chain CDR regions including the germline-encoded framework and CDR1 and 2 regions of the receptor. Resultant amplicons were sequenced via the Ion Torrent, annotated by comparison to the gold-standard IMGT database, then mined to construct TRBV allele profiles for each individual including, where detected, novel alleles not found in the IMGT database. Finally, we correlated TRBV allele profiles with adverse events annotations to detect alleles or sets of alleles associated with severe (grade 3 or higher) adverse events following immunotherapy.

Results

Sequencing of TCRB libraries yielded on average 29k clonotypes per individual with mean evenness (normalized Shannon entropy) of .85. Principal component analysis and k-means clustering of TRBV allele profiles revealed the presence of 4 major sets of coincident variable gene alleles which we term haplotype groups. The incidence of severe adverse events varied markedly across haplotype groups: one group, comprising approximately one fourth of the sample set, appeared completely protected against grade 3 or higher adverse events (0% incidence), while up to 57% of individuals in other haplotype groups had severe adverse events (p = 2.4E-3, Fisher’s exact test).

Conclusions

These data suggest that germline-encoded TRBV polymorphism may play a mechanistic role in autoimmune toxicity during checkpoint blockade immunotherapy. We find that a subset of Caucasians appear to be at low risk of IRAEs and thus may be particularly well suited for immunotherapy regimens having elevated incidence of toxicity. Current and future studies will further explore the utility of TRBV polymorphism as a predictive biomarker for IRAEs.

P82 Peripheral blood TCRB repertoire convergence and clonal expansion predict response to anti-CTLA-4 monotherapy for cancer

Li Zhang, MD PhD2, Timothy Looney, PhD1, David Oh, MD, PhD2, Denise Topacio-Hall, BS, MA1, Lawrence Fong, MD2
1Thermo Fisher Scientific, South San Francisco, CA, USA; 2University of California - San Francisco, San Francisco, CA, USA
Correspondence: Li Zhang (li.zhang@ucsf.edu)

Background

Tumor antigen-driven selection may expand T cells having T cell receptors (TCRs) of shared antigen specificity but different amino acid or nucleotide sequence in a process known as TCR convergence. Efforts to evaluate the biomarker utility of TCR convergence through TCRB repertoire sequencing have been hampered by the base substitution error rate of the Illumina platform, given that such errors may create artifacts resembling TCR convergence. Here we leverage the low base substitution error rate of the Ion Torrent platform to evaluate convergence as a predictive biomarker for response to anti-CTLA-4 monotherapy in a set of 22 individuals with cancer. For context, we compared convergence values obtained using this platform to those for the same samples interrogated with Illumina-based TCRB repertoire sequencing. Finally, we examined whether TCR convergence may be combined with measurements of clonal expansion to improve prediction of immunotherapy response.

Methods

Total RNA extracted from pretreatment peripheral blood leukocytes (PBL) from 22 recipients of Ipilimumab monotherapy. TCRB repertoire libraries were constructed by multiplex PCR via the Oncomine TCRB-LR assay, then sequenced using the Ion Torrent S5 to a target depth of 1.5M raw reads per library. To evaluate convergence within each repertoire we searched for instances where TCRB rearrangements were identical in amino acid space but had distinct nucleotide sequences within the CDR3. For a subset of samples, TCRB sequencing was performed in parallel via Illumina-based approaches.

Results

Sequencing of TCRB libraries yielded on average 31k clonotypes per individual with mean evenness (normalized Shannon entropy) of .82. TCR convergence was elevated in pretreatment PBL of responders compared to non- responders (mean frequency .022 vs .009; p=.03, Wilcoxon), and could discriminate responders from non- responders (AUROC = .77). Pretreatment evenness was reduced in responders vs non-responders and also predictive of response (AUROC = .74). A logistic regression model combining both features improved prediction of response (AUROC = .89).

Conclusions

These data suggest that PBL TCR convergence may serve as a predictive biomarker for response to anti-CTLA-4 monotherapy, potentially in combination with other immune repertoire features. Notably, measurements of TCR convergence appear to be sensitive to base substitution sequencing errors. These results highlight the impact of different sequencing approaches for assessing TCR repertoire.

P83 T cell receptor beta immune repertoire sequencing in several FFPE tissue types – Interrogation of the tumor microenvironment in archived tissue samples

Denise Topacio-Hall, BS, MA, Lauren Miller, BS, Elizabeth Linch, BS, Alice Zheng, PhD, Geoffrey M. Lowman, PhD, Timothy Looney, PhD, Mark Andersen, PhD
ThermoFisher Scientific, Carlsbad, CA, USA
Correspondence: Geoffrey M. Lowman (geoffrey.lowman@thermofisher.com)

Background

Immune repertoire sequencing is a valuable tool for studies of the tumor microenvironment and potential immune responses to cancer immunotherapy. Here we describe a T cell receptor beta (TCRβ) sequencing assay that leverages the low sample input requirements of AmpliSeq library preparation technology to extend the capability of targeted immune repertoire sequencing to include FFPE samples which can often be degraded and in short supply.

Methods

Evaluation of the highly diverse CDR3 region of TCRβ allows for T cell clone identification and frequency measurement. We demonstrate assay functionality with input of RNA or DNA samples, as well as flexibility in sequencing throughput and sample multiplexing capability. T cell repertoires were evaluated from as low as 10ng to 1ug of input material of varying repertoire diversity, such as sorted T cells, peripheral blood leukocytes, fresh-frozen tissue, and FFPE tissue from a variety of normal and cancerous tissues such as lung, colon, brain, spleen, lymph node, and thymus.

Results

Accuracy is demonstrated through the evaluation of samples comprised of known numbers of sorted T cells or spike-in experiments using well-studied lymphoma rearrangements. In order to test functionality of the assay with a range of degraded input material, RNA was controllably degraded with heat treatment at 90-95°C. In these systematically degraded samples we observe a strong correlation (r = 0.97) between the percentage of RNA molecules over 200bp in length and the amount of productive repertoire reads that the assay produces, while maintaining performance levels with samples with RIN values approaching 2. T cell richness and diversity in repertoires measured from FFPE tissue samples vary, as expected, depending on sample quality, disease state, and tissue of origin. To aid in sample input determination we present a complimentary qPCR assay, specific for T cell markers, which allows for sample T cell quantification and acts to guide optimal sample input ranges for library construction.

Conclusions

These data introduce a T cell immune repertoire sequencing solution for applications in a wide range of sample types including challenging FFPE preserved tissues. This assay is capable of profiling repertoire metrics from samples over a large range of input amounts from several tissue types. In addition, we demonstrate use of a qPCR assay for quantification of sample T cell content to guide sample input for TCRβ immune repertoire sequencing with samples with highly variable T cell content.

P84 Quantitative evaluation of tumor-infiltrating lymphocyte subsets and PD-L1 expression in lung cancer brain metastases

Benjamin Lu, MD, Richa Gupta, Hailey Wyatt, Matthew Ribeiro, Tyler Stewart, Veronica Chiang, MD, Joseph Contessa, Adebowale Adeniran, Harriet Kluger, MD, Lucia Jilaveanu, MD, Kurt Schalper, MD, PhD, Sarah B. Goldberg, MD, MPH
Yale Cancer Center, New Haven, CT, USA
Correspondence: Sarah B. Goldberg (sarah.goldberg@yale.edu)

Background

Lung cancer brain metastases (BrM) are associated with prominent morbidity and mortality. PD-1/PD-L1 inhibitors are safe and clinically active in patients with BrM in non-small cell lung cancer (NSCLC). While PD-L1 expression is associated with increased tumor infiltrating lymphocytes (TILs) and sensitivity to PD-1/PD-L1 inhibitors in extracranial tumors, the level and association between these markers in lung cancer BrM is unknown. Using spatially resolved/multiplexed tumor tissue analysis, we performed a comparative analysis of PD-L1 and major TIL subsets in primary lung cancers, BrM, and extracranial metastases (ECM).

Methods

We studied formalin-fixed paraffin-embedded tumor samples from a retrospective collection of 94 stage I-IV lung cancer patients from Yale between 2002-2013 represented in a tissue microarray. In total, 40 primary lung cancers, 63 BrM, and 15 ECM were included. Paired samples included primary-BrM from 11 patients and BrM-ECM from 12 patients. TIL density was determined by a semi-quantitative pathologist-based, scoring system using H&E preparations. Multiplexed quantitative immunofluorescence was used to evaluate PD-L1, CD4 for helper T-cells, CD8 for cytotoxic cells, and CD20 for B-lymphocytes. Signal for each marker was measured in marker-selected tissue compartments using the Automated Quantitative Analysis (AQUA) platform. We studied the association between markers and major clinicopathologic variables, including overall survival.

Results

Lung cancer histology included adenocarcinoma 62.5%, squamous cell carcinoma 11.5%, small cell 9.4%, and other NSCLC 16.7%. Only 8.5% of patients received immune checkpoint inhibitors. TIL density by pathologist read was significantly lower in BrM compared with primary lung tumors (p<0.0001). BrM had significantly lower levels of CD4+ T-cells (p=0.0416), CD8+ T-cells (p=0.0003), and CD20+ B-lymphocytes (p=0.0058) than primary lesions. Levels of tumor PD-L1 were comparable between BrM and primary lung tumors or ECMs (p>0.05). However, PD- L1:CD8 ratios were significantly higher in BrM compared with primary tumors (p=0.0024) or ECM (p=0.0322) without differences in PD-L1:CD4 ratios (p>0.05). Paired sample analyses demonstrated similar trends, though statistical significance was not achieved. There was no association observed between overall survival and TIL density, levels of TIL subsets, or PD-L1 expression.

Conclusions

Despite having lower levels of major TIL subsets, lung cancer BrM displayed similar PD-L1 expression compared with lung primary cancers and ECM. The latter indicates differences in the adaptive immune modulation of PD-L1 in BrM compared with extracranial tumors, suggesting alternative TIL-independent mechanisms sustaining PD-L1 expression in BrM. A better understanding of how the PD-1 axis differs in the brain microenvironment may help improve anti-PD-1/PD-L1 efficacy and reveal additional therapeutic targets

Ethics Approval

The study was approved by Yale University's Institutional Review Board, HIC# 1310012801.

P85 Correlation of clinical response and pathologic treatment effect after 4 weeks of preoperative PD- 1 blockade in primary head and neck squamous cell carcinoma (HNSCC)

Adam Luginbuhl, MD, Jennifer Johnson, MD, Madalina Tuluc, MD, Stacey Mardekian, MD, Chandala Chitguppi, MD, Larry Harshyne, PhD, Ralph Zinner, MD, Joseph Curry, MD, David Cognetti, MD, Ulrich Rodeck, MD PhD, Athanassios Argiris, MD, PhD, Adam Luginbuhl, MD
Thomas Jefferson University, Philadelphia, PA, USA
Correspondence: Jennifer Johnson (jennifer.m.johnson@jefferson.edu)

Background

Nivolumab, a PD-1 inhibitor, has been integrated into the clinical management of recurrent or metastatic HNSCC and is being evaluated in earlier stages of this disease. It is unclear whether imaging modalities, such CT and MRI, accurately reflect tumor response to immune checkpoint inhibition due to treatment-induced inflammatory changes at tumor sites. We sought to explore the relationship of imaging and pathology findings in the context of an ongoing neoadjuvant trial of preoperative nivolumab with or without tadalafil in resectable HNSCC.

Methods

Patients (n=17) with resectable primary HNSCC received nivolumab 240 mg IV Q 2 weeks for 2 doses and were randomized 1:1 to also receive tadalafil 10 mg PO for 28 days or not. Surgery was performed 4 weeks after the first nivolumab infusion. Tumor volumes were assessed pretreatment and on the day of surgery by CT scan. Resection specimens were graded histopathologically by two pathologists. Percent of treatment effect was determined by dividing the area of tumor showing changes consistent with treatment effect (fibrosis with chronic inflammation, foamy macrophage reaction and multinucleated giant cells) by total area containing treated and residual tumor. Radiographic effect was determined both by modified iRECIST and investigator assessment. Fischer exact test was performed to assess association between radiological and pathological tumor response. Percentage shrinkage of tumor was calculated in both radiological and pathological modalities and strength of association was calculated using Pearson correlation coefficient.

Results

Imaging results were used to stratify patients into response categories ranging from progression, stable disease and partial/complete response. Radiographic tumor shrinkage was observed in 11/17 (65%) patients with 16-100% volume reduction. A statistically significant relationship was noted between the radiological findings and treatment effects confirmed histopathologically (p = 0.009). A strong correlation was observed between these two groups (Pearson’s r = 0.7185; p = 0.001). All patients with stable disease or radiographic progression (6/17) at 4 weeks had no evidence of treatment effect in pathologic specimens.

Conclusions

In this treatment naive cohort, imaging approaches accurately captured treatment responses validated by histopathologic assessment of HNSCC surgical specimens obtained after two doses of nivolumab.

Acknowledgements

Sidney Kimmel Cancer Center at Thomas Jefferson UniversityBristol-Myers Squibb

Trial Registration

NCT03238365

Ethics Approval

The study was approved by Thomas Jefferson University Institutution‘s Ethics Board, approval number #17P.210

P86 T-cell receptor (TCR) repertoire features associated with disease-free survival following infusion with marrow-infiltrating lymphocytes (MILs)

Eric Lutz, PhD1, Alex Hopkins, PhD2, LAKSHMI RUDRARAJU, MS1, Elizabeth DeOlivera1, Ido Weiss, PhD1, Rachel Gittelman, PhD3, Erik Yusko, PhD3, Kathryn Boland, DVM, PhD3, Ivan Borrello, MD2, Kimberly A. Noonan, Ph.D1
1WindMIL Therapeutics Inc., Baltimore, MD, USA; 2Johns Hopkins University, Baltimore, MD, USA; 3Adaptive Biotechnologies, Seattle, WA, USA
Correspondence: Kimberly A. Noonan (noonan@windmiltherapeutics.com)

Background

MILs are an autologous T-cell product expanded from bone marrow (BM) being developed as a novel cell therapy for both hematological and solid malignancies. In a Phase I trial evaluating MILs in patients with advanced multiple myeloma, 6 (27.3%) of 22 patients achieved a complete remission (CR). Immune analyses demonstrated that the establishment of persistent tumor antigen-specific T cells in BM correlated with improved clinical responses [1].Herein, we sought to identify the repertoire of T cell clonotypes within MILs, including the subset that specifically recognize tumor antigens; to track and compare their frequencies in blood and BM before and after infusion; and to compare T cell repertoire characteristics, such as clonality, between clinical responders and non- responders.

Methods

The TCRb CDR3 was sequenced using Adaptive Biotechnologies’ immunoSEQ Assay and used to identify and track MILs T cell clonotypes. The immunoSEQ assay was used on 11 specimens (unsorted MILs, IFNg-capture- sorted tumor antigen-specific CD4+ and CD8+ T cells, and blood and BM collected pre-treatment and 60, 180 and 360 days post-infusion) from 6 patients (3 clinical responders who achieved a CR and 3 non-responders whose disease progressed) from the Phase I study.

Results

When cumulative frequencies of MILs T-cell clonotypes were tracked in BM and blood, there were significant differences between responders and non-responders. Responders had a lower frequency of clonotypes at baseline but showed larger and more persistent increases in the frequency of clonotypes in both BM and blood. At day 360, fold- change from baseline in the frequency of MILs in both compartments segregated responders from non-responders. In general, T-cell repertoires in MILs were highly polyclonal and no specific TCRb variable genes were enriched in tumor antigen-specific T cells suggesting that multiple antigens are targeted. In all 6 patients, MILs were more polyclonal than pre-expanded BM. However, starting repertoires were more polyclonal in responders, and responders had larger and more persistent post-infusion increases in clonality. At day 360, all 3 responders maintained an increase in clonality whereas clonality returned to baseline or lower in all 3 non-responders.

Conclusions

These data provide a 1st look at the repertoire of T cell clonotypes in MILs and how the repertoire evolves after treatment. The data also demonstrate the highly polyclonal nature of tumor antigen-specific T cells within MILs, which could provide an advantage against heterogeneous tumors.

References

1. Noonan KA, Huff CA, Davis J, Lemas M. V, Fiorino S, Bitzan J, Ferguson A, Emerling A, Borrello I. Adoptive transfer of activated marrow-infiltrating lymphocytes induces measurable antitumor immunity in the bone marrow in multiple myeloma. Sci. Transl. Med. 2015; 7: 288ra78.

Ethics Approval

The study was approved by the Johns Hopkins University IRB.

P87 Preliminary evidence of intratumoral activation and immunomodulatory effect of CX-072, a Probody therapeutic antibody prodrug targeting PD-L1, in a phase 1/2a trial

Susan K. Lyman1, Judi Gordon1, Amy DuPage1, Preeti Pramanik1, Bruce Howng1, Michael B. Winter1, Irina K. Popova1, Olga Vasiljeva1, James Jones1, Kenneth Wong, MA1, Victoria Singson1, Jennifer Richardson, PhD1, Beiyao Zheng, PhD1, Mark Stroh, PhD1, Lori Carman, RN1, Vanessa Huels1, Karen Autio, MS, MD2, Valentina Boni3, Daniel Cho, MD4, Javier Garcia-Corbacho5, Iván Victoria Ruiz5, Omid Hamid, MD6, Nataliya Uboha7, Elisabeth de Vries8, Anthony El-Khoueiry, MD9, Alexander Spira, MD, PhD, FACP10, Rachel Sanborn, MD11, Fiona Thistlethwaite, MD, PhD12, Hendrik-Tobias Arkenau13, Johanna Bendell14, Patrick Ott, MD, PhD15, Naiyer Rizvi, MD16, Matthias Will, MD1, W. Michael Kavanaugh, MD1, Aung Naing, MD, FACP17, Luc R. Desnoyers, Ph D1
1CytomX Therapeutics, Inc., South San Francisco, CA, USA; 2Memorial Sloan Kettering Cancer Center, New York, NY, USA; 3Hospital Universitario HM Sanchinarro, Madrid, Spain; 4Laura and Isaac Perlmutter Cancer Center, New York, NY, USA; 5Hospital Clinic de Barcelona, Barcelona, Spain; 6The Angeles Clinic and Research Institute, A Cedars-Sinai Affiliate, Los Angeles, CA, USA; 7University of Wisconsin, Madison, WI, USA; 8University Medical Center Groningen, the Netherlands, Groningen, Netherlands; 9USC Norris Comprehensive Cancer Center, Los Angeles, CA, USA; 10Virginia Cancer Specialists, Fairfax, VA, USA; 11Earle A. Chiles Research Institute, Providence Cancer Center, Portland, OR, USA; 12The Christie Hospital NHS Trust and University of Manchester, Manchester, UK; 13Sarah Cannon Research Institute-UK, London, UK; 14Tennessee Oncology, Nashville, TN, USA; 15Dana Farber Cancer Institute, Boston, MA, USA; 16Columbia University Medical Center, New York, NY, USA; 17The University of Texas, Houston, TX, USA
Correspondence: Luc R. Desnoyers (Luc@cytomx.com)

Background

CX-072 is a Probody™ therapeutic antibody prodrug directed against PD-L1. Probody therapeutics are masked antibodies designed to be selectively activated within the tumor microenvironment by tumor-associated proteases. CX-072 is designed to reduce systemic immune-related toxicities of anti-PD-L1 therapy, especially in combination with other drugs, while maintaining antitumor activity. PROCLAIM-CX-072 is a first-in-human, phase 1/2, open- label dose-finding trial investigating the safety and maximum tolerated dose of CX-072 as monotherapy and in combination with ipilimumab or vemurafenib. PROCLAIM-CX-072 patients have metastatic or recurrent solid tumors or lymphomas for which approved PD-1/-L1–based therapy is not available. Here we present the initial results of the tissue-based biomarker program intended to evaluate the mechanism of action of CX-072 in patients from PROCLAIM-CX-072.

Methods

Tumor biopsy and matched plasma samples were collected during the screening phase and after dosing of CX-072.

PD-L1 and CD8 expression were analyzed using immunohistochemistry. Relevant tumor-associated protease activity was measured by tissue zymography. Intratumoral CX-072 unmasking and activation were measured using capillary immunoelectrophoresis. Gene expression was profiled by NanoString.

Results

Results of the first 13 evaluable biopsies obtained are reported here. Nine of 12 (75%) predose biopsies had detectable levels of relevant protease activity. Two of 4 (50%) biopsies from patients treated with CX-072 at 3 mg/kg and 4 of 4 (100%) biopsies from patients treated at ≥10 mg/kg had detectable intratumoral activation of CX- 072, and the concentration of activated CX-072 in tumors increased with increasing dose. The preliminary calculated estimate of tumor receptor occupancy for patients receiving the 10 mg/kg dose was similar to that targeted for the PD-L1 inhibitor atezolizumab [1], and was consistent with quantitative systems pharmacology model predictions. Notably, the concentration of activated CX-072 measured in human tumor samples was similar to that associated with efficacy in a syngeneic preclinical tumor model. Consistent with the inhibition of the PD-L1 pathway by CX-072, we found an increase in CD8+ T cells and elevation of cytotoxic T-cell markers in the tumor of the one CX-072 monotherapy patient whose biopsy met evaluability criteria. These data support selection of the 10 mg/kg dose for clinical expansion cohorts.

Conclusions

These preliminary results show the presence of relevant protease activity, intratumoral Probody therapeutic activation, and biological effect of a Probody therapeutic in human subjects treated with CX-072. Taken together with previous data demonstrating stability of the masked Probody therapeutic in systemic circulation [2], these results support proof-of-mechanism for the Probody platform.

Acknowledgements

Editorial support was provided by ApotheCom (San Francisco, CA).

Trial Registration

ClinicalTrials.gov, NCT03013491

References

1. Stroh, M, Winter H, Marchand M, et al. Clinical pharmacokinetics and pharmacodynamics of atezolizumab in metastatic urothelial carcinoma. Clin Pharmacol Ther. 2017;102:305-312.

2. Autio K, Arkenau H-T, O’Neil B, et al. Preliminary results of the first-in-human, dose-finding PROCLAIM-CX- 072 trial of the PD-L1 Probody therapeutic CX-072 as monotherapy in patients (pts) with advanced solid tumors. J Clin Oncol. 2018;36(suppl):abstr 3071.

Ethics Approval

The animal study was performed in accordance with the Guide for the Care and Use of Laboratory Animals under a protocol (AP203) approved by the Institutional Animal Care and Use Committee at CytomX. The CX-072 clinical study was approved by all participating sites institutional review boards as well as the overseeing Ethics Board.

P88 Quantitative assessment and standardization of the programmed death 1 ligand 1 (PD-L1) immunohistochemistry companion diagnostic assays

Sandra Martinez-Morilla, PhD1, John McGuire1, Patricia Gaule, PhD1, Lauren Moore1, Balazs Acs1, Delphine Cougot2, David Rimm, MD, PhD1
1Yale University, New Haven, CT, USA; 2Horizon Discovery, Cambridge, UK
Correspondence: Sandra Martinez-Morilla (sandra.martinez-morilla@yale.edu)

Background

Programmed Death 1 Ligand 1 (PD-L1) Immunohistochemistry (IHC) is the only FDA approved predictive marker to identify responders to anti-PD1 axis drugs. Multiple PD-L1 IHC assays with various antibodies and cut-points have been used in clinical trials across tumor types. Comparative performance characteristics of these assays have been extensively studied qualitatively, but not quantitatively. Since PD-L1 is a continuous marker, we propose the use of a standardized PD-L1 Index TMA to objectively evaluate concordance between antibody assays for PD-L1 using quantitative image analysis.

Methods

A panel of 10 isogenic cell lines expressing various amounts of PD-L1 was developed by Horizon Dx and constructed as an Index Tissue Microarray (TMA). Identical but independent batches of isogenic cells lines were cultured to create 3 TMA batches at 3 separate timepoints to control for any batch effect that may occur. The TMAs were validated using a previously published quantitative immunofluorescence protocol (QIF-AQUA). Comparing antibodies E1L3N, SP142 and SP263, reproducibility was assessed between batches. We then compared US Food and Drug Administration (FDA)-approved 22C3, 28-8, SP142 and SP263 assays and E1L3N lab developed test (LDT). Digital image analysis was used to quantify chromogenic PD-L1 assays using the open-source QuPath platform.

Results

There was very high reproducibility between blocks (R2=0.875-0.995) with the IF assay. The IF concordance between antibodies was also extremely high (R2=0.986-0.987). The 4 FDA approved assays and the E1L3N LDT were compared on the Index TMAs using quantitative chromogenic assessment with QuPath. The assays for 22C3- FDA, 28-8-FDA, SP263-FDA and E1L3N-LDT were essentially identical. The SP-142-FDA assay failed to detect low expressing cell lines detected by the other 4 assays. Levey-Jennings analysis was done to show the value of using the index array over time as a standardization tool in the CLIA lab setting.

Conclusions

We have built a standardized Index TMA that spans the dynamic range of PD-L1 expression that shows reproducibility by QIF across independent blocks. Quantitative assessment of the 4 FDA assays and the E1L3N LDT shows the assays recognize the entire dynamic range in a reproducible manner, except for the SP-142 assay that fails to detect low PD-L1 expressers. We propose this commercial TMA as a useful standardization mechanism to compare results between institutions and to identify abnormalities while running routine clinical samples. A multi-institutional comparison study with this Index TMA with different assays and platforms is currently underway.

P89 Withdrawn

P90 Using artificial intelligence to predict response to immunotherapy

Anthony Milici, PhD, Navi Mehra, Jospeh S. Krueger, BS, PhD, Karen Ryall, PhD, BS, Jenifer Caldara, BS, Will Paces, BS, Kelsey Weigel, PhD
Flagship Biosciences, Branford, CT, USA
Correspondence: Jospeh S. Krueger (jkrueger@flagshipbio.com)

Background

PD-L1/PD-1 checkpoint blockade is the backbone for the myriads of combination therapies being developed; and thus effective PD-L1 IHC testing remains critical for predicting patient response to these therapies. Pathological interpretations applied to PD-L1 immunohistochemistry (IHC) as a response biomarker are becoming more complex, going beyond simple Tumor Proportion Score (TPS) and requiring more complex diagnostic algorithms which evaluate the role of PD-L1 expression in 1) The tumor cells; 2) Immune cells in the Tumor Microenvironment (TME); and 3) Tumor infiltrating lymphocytes (TILs); all whose spatial relationships are critical for understanding the immune contexture. This complex matrix of several different biological cell types and spatial relationships can quickly become impossible for a pathologist to record and report successfully.

Methods

Recent advances in computer computational ability, machine learning algorithms, and data science now allow the application of Artificial Intelligence (AI) methods to create data-rich profiles from Whole Slide Images (WSI) of tissue that capture this key tissue context information about PD-L1. In this study, tissue image analysis was applied to this setting to increase objectivity and reproducibility of scoring, and AI interpretation of the results used to increase accuracy and sensitivity of the diagnostic performance of existing PD-L1 IHC testing methods. In this manner, existing FDA approved PD-L1 IHC tests can be re-evaluated by AI approaches to create this value in the clinical setting, without change to the IHC assay or significant disruption in the normal procedures performed in pathology labs.

Results

In this study, we demonstrate how our WSI AI platform captures the different attributes of PD-L1 stained slides to create a summary, singular score or output for decision making. Flagship’s cTA® records the PD-L1 staining, morphological, organizational, and spatial aspects of a tissue section to create a more sophisticated scoring system and better diagnostic cutpoint in a cohort of patient tissues measured against clinical response.

Conclusions

By applying Flagship’s cTA® Artificial Intelligence to existing PD-L1 IHC CDx, clinical labs can now go beyond using tissue image analysis for improving objectivity and reproducibility, and can also create entirely new scoring approaches from existing PD-L1 IHC CDx for improved clinical performance.

P91 It’s pan-o’clock: Tumor and circulating lymphocytic pan-pathology targets of the failed immune response

Anne Monette, PhD1, Antigoni Morou, PhD2, Nadia Al-banna3, Louise Rousseau, RT4, Jean-Baptiste Lattouf, MD, FRCSC2, Sara Rahmati5, Tomas Tokar6, Daniel Kaufmann, MD2, Jean-pierre Routy7, Igor Jurisica, PhD8, Rejean Lapointe, PhD2
1University of Montreal / University of Montreal Hospital Research Centre / Lady Davis Institute for Medical Research / Jewish General Hospital, Montreal, Canada; 2University of Montreal / University of Montreal Hospital Research Centre, Montreal, QC, Canada; 3McGill University / University of Montreal Hospital Research Centre, Montreal, Canada; 4University of Montreal Hospital Research Centre, Montreal, QC, Canada; 5University of Toronto / Krembil Research Institute, Toronto, Canada; 6Krembil Research Institute, Toronto, Canada; 7McGill University Health Centre Chronic Viral Illnesses Service and Division of Hematology, Montreal, QUEBEC, Canada; 8University of Toronto / Krembil Research / Toronto Western Hospital / Techna Institute, Toronto, ON, Canada
Correspondence: Anne Monette (anne.monette@mail.mcgill.ca)

Background

Tumor infiltrating lymphocytes are widely associated with positive outcomes, yet carry key indicators of a systemic failed immune response against unresolved cancer. Cancer immunotherapies can reverse their tolerance phenotypes, while preserving tumor-reactivity and neoantigen-specificity shared with circulating immune cells.

Methods

We performed comprehensive transcriptomic analyses to identify gene signatures common to circulating and tumor infiltrating lymphocytes in the context of clear cell renal cell carcinoma. Modulated genes also associated with disease outcome were validated in several other cancer types. Using bioinformatics, we identified practical diagnostic markers and actionable targets of the failed immune response.

Results

On circulating lymphocytes, a minimal set of genes could efficiently stratify patients from healthy control donors.

From their associations with resistance to cancer immunotherapies and microbial infections, we have uncovered not only pan-cancer, but pan-pathology failed immune response profiles of effector and antigen presenting lymphocytes.

Conclusions

A prominent lymphocyte-specific cell migration pathway, is central to a panoply of diseases and tumor immunogenicity, correlates with multi-cancer recurrence in patients, and identifies a feasible, non-invasive approach to pan-pathology diagnoses.

Ethics Approval

The study was approved by the CHUM research ethics board Ethics Board, approval reference number SL07.053.

Consent

Written informed consent was obtained from the all study participants by the CHUM kidney biobank.

P92 Predictive plasma proteomic biomarkers of immunotherapy toxicity in patients (pts) with metastatic melanoma (MM)

Meghan Mooradian, MD1, Xuesong Gu2, Donald Lawrence, MD1, Justine Cohen, DO1, Tatyana Sharova1, Genevieve Boland, MD, PhD1, Towia Libermann2, Ryan J. Sullivan, MD1
1Massachusetts General Hospital, Boston, MA, USA; 2Harvard University, Boston, MA, USA
Correspondence: Ryan J. Sullivan (rsullivan7@mgh.harvard.edu)

Background

Mechanisms underlying immune checkpoint inhibition (ICI) efficacy and toxicity have yet to be fully elucidated and, to date, there are no reliable biomarkers predictive of the development of immune-related adverse events (irAEs) or efficacy.

Methods

Pts with MM who developed select irAEs (colitis, hepatitis and arthritis) while receiving ICI (anti-PD-1, anti-CTLA-4 or combination anti-PD-1/CTLA-4) were identified. Using SOMAscan (SomaLogic; Boulder, CO), a proteomics platform that enables measurement of 1,305 proteins, quantitative plasma protein profiles were generated from banked plasma samples to identify candidate protein biomarkers predictive of irAEs and of progression-free survival (PFS). Comparator samples from pts without toxicity served as matched controls. Baseline protein expression was assessed and in evaluable pts, change in protein expression pre- and post-treatment was performed. PFS data was captured.

Results

36 pts were tested; 28 with confirmed irAEs and 8 controls. Baseline expression of IL-17, CXCL10 and TGFβ1 was associated with irAE development. Baseline IL-17 and CXCL10 expression was especially prominent in cases of ICI-induced arthritis, whereas TGFβ1 was linked to both ICI-induced arthritis and colitis. Independent ELISA validation studies of the candidate biomarkers are underway and will be presented. The one-year PFS of the toxicity cohort was 54% and 65% in those treated with PD-1 inhibition +/- anti-CTLA-4, respectively. Elevated TGFβ1, as well as LGALS3 and DLL4, were associated with poorer outcomes.

Conclusions

This study demonstrates the feasibility of SOMAscan to identify potential biomarkers of toxicity and outcome.

Previous literature has linked IL-17 and CXCL10 in autoimmunity, and TGFβ1 with a more pleiotropic role in immune regulation and melanoma pathogenesis. Our findings illustrate their possible role in the development of irAEs and outcome. If confirmed to be mechanistically involved in irAEs, targeted inhibitors of these proteins may serve as effective methods to abrogate or even prevent ICI-induced toxicity while minimizing effect on efficacy.

Ethics Approval

The study was performed with the approval of the MGH IRB, under research protocol 11-181

P93 Pre-analytical variables affect myeloid-derived suppressor cell quantitation by flow cytometry

Chihiro Morishima, MD, Amy Wright, Angela Riggins, Minjun Apodaca
University of Washington, Seattle, WA, USA
Correspondence: Chihiro Morishima (chihiro@uw.edu)

Background

Background: Myeloid-derived suppressor cells (MDSC) have been found to play an important role in limiting immune responses in many disease states including cancer. Higher circulating MDSC levels have been associated with greater tumor burden, poorer response to immunotherapy, and poorer survival. Optimal measurement of MDSC levels could provide clinicians with a useful management and/or prognostic tool.

Methods

Methods: Whole blood was obtained from healthy and diseased subjects through a University of Washington Institutional Review Board-approved study, #51834. A nine color flow cytometric assay included fluorescently- labeled antibodies against CD45, CD3, CD19, CD20, CD56, CD16, HLA-DR, CD33, CD11b, CD14 and CD15, and BD Trucount beads for quantitation. Samples were analyzed using a BD LSRFortessa and FlowJo software v9.9.5. Total MDSC were defined as CD45+CD3-CD19-CD20-CD56-CD16-HLA-DR-CD33+CD11b+ cells, while the monocytic (M-MDSC) and granulocytic (G-MDSC) subsets were defined as CD14+ or CD15+, respectively.

Results

Results: To confirm that our MDSC whole blood (WB) assay identified the same cells reported by other groups using PBMC as source material, we performed standard Ficoll density centrifugation of WB. The MDSC population identified by our assay in WB (0.32% of CD45+ cells) was found among cells from the Ficoll interface (0.28% of CD45+), but not the Ficoll pellet (0.02% of CD45+). Surprisingly, the yield of total and M-MDSC was higher with EDTA compared to heparin tubes (median 68% and 83% greater, respectively) among 5 donors with simultaneous blood collection in the two tube types, tested within 4 hours of blood draw. In addition, the duration of time that WB was kept at room temperature prior to cell labeling affected the yield of MDSC identified. For blood collected in EDTA tubes, total MDSC numbers decreased slightly by medians of 9% (N=7) and 11% (N=5) at 8 and 24 hours, respectively. M-MDSC numbers decreased somewhat more by medians of 14% (N=7) and 53% (N=5) at 8 and 24 hours, respectively. Finally, bilirubin levels as low as 1.6 mg/dL could impair the accurate identification of MDSCs. After controlling for these pre-analytical factors, significant differences in MDSC levels were still found among patients with hepatocellular carcinoma (N=35) compared to healthy controls (N=35, p=0.001).

Conclusions

Conclusions: MDSC are a heterogenous group of cells, and their quantitation in WB can be affected by a number of pre-analytical variables. Consideration of these factors, and measurement using a material type that has not been manipulated, such as WB, is likely to yield the most accurate results.

P94 Measurement of adenosine and other immunomodulators in the tumor microenvironment by in vivo microdialysis

Nadege Morisot, PhD, Julien Roeser, PhD, Holden Janssens, PhD, Arash Rassoulpour, PhD
Charles River Laboratories, South San Francisco, CA, USA
Correspondence: Nadege Morisot (nadege.morisot@crl.com)

Background

Understanding the tumor microenvironment is essential to gaining a better understanding of cancer biology and pathophysiology, and to date has proven to be challenging. The ability to determine the levels of adenosine and/or other molecules in the extracellular environment of tumors could provide unique insight into molecular targets for novel therapies in the treatment of cancers.

Methods

We implemented in vivo microdialysis to measure oncomodulator levels in the tumor microenvironment of freely moving mice. Microdialysis probes were implanted within 4T1 or CT26 syngeneic tumors and in the contralateral rump of the same animals. Probes were perfused with a buffered solution and dialysate samples were continuously collected for several hours. Levels of 12 metabolites (adenosine, cAMP, cGMP, arginine, ornithine, putrescine, tryptophan, kynurenine, kynurenic acid, 3-hydroxy-kynurenine, anthranilic acid, and xanthurenic acid) and lactate in the dialysates were quantified by liquid chromatography–mass spectrometry (LC-MS). Additionally, some animals received erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA) through the microdialysis probe to examine the effects of adenosine deaminase inhibition on metabolite levels.

Results

We first confirmed that probe implantation did not affect the growth of 4T1 and CT26 allografts by comparing the volume of probe-implanted and -free tumor. Next, we found that adenosine levels were significantly elevated in both 4T1 and CT26 tumors, whilst the levels of cAMP and cGMP were reduced compared to non-tumor dialysis. Application of EHNA significantly increased the levels of Adenosine, cAMP, and cGMP in 4T1 tumors compared to basal levels. Additionally, we found that the levels of several kynurenine pathway metabolites were reduced within 4T1 tumors compared to the contralateral side. Interestingly, putrescine levels were elevated in CT26 but not 4T1 tumors compared to non-tumor dialysis. Finally, we observed high levels of lactate in both 4T1 and CT26 allografts.

Conclusions

Together, we have demonstrated the ability to detect levels of several signaling molecules that are believed to play a role in cancer biology. Additionally, we have been able to show that the reported levels are specific to the tumor microenvironment, and not a global circulating phenomenon within the same animal. In vivo microdialysis in murine tumor models may help in elucidating the mechanisms by which therapies such as chemotherapy and immune checkpoint inhibitors modulate the tumor microenvironment, helping identify the next-generation of therapies in oncology.

P95 Modulation of adenosine levels in the tumor microenvironment following treatment with anti-PD1 antibodies and oxaliplatin: an in vivo microdialysis study

Arash Rassoulpour, PhD, Nadege Morisot, PhD, Julien Roeser, PhD, Holden Janssens, PhD
Charles River Laboratories, South San Francisco, CA, USA
Correspondence: Nadege Morisot (nadege.morisot@crl.com)

Background

Elucidating the mechanisms by which cancer treatments modulate cellular signaling within the tumor microenvironment represents a major challenge in cancer research. In addition, the lack of reliable measurement of anti-cancer drug penetration in solid tumor may slow the development of effective systemic drug delivery strategies. We propose that microdialysis of the tumor microenvironment in syngeneic tumor models coupled to liquid chromatography–mass spectrometry may help filling in these gaps.

Methods

Mice bearing MC38 syngeneic tumors received control or anti-PD1 antibodies (5 mg/kg, i.p.) treatment on post-inoculation day 6, 9, 12 and 15. Next, animals were implanted with microdialysis probes within MC38 allografts. Dialysate samples from the tumor microenvironment were continuously collected in freely moving animal. In a separate cohort of MC38-bearing mice with microdialysis probes implanted in the tumor and contralateral subcutaneous (SC) space, baseline dialysate samples were collected, and mice received an acute systemic administration with oxaliplatin (10 mg/kg, i.p.). Dialysates were collected for two hours after treatment. Levels of adenosine, inosine and cAMP as well as the concentration of oxaliplatin in the tumoral and SC dialysates were quantified by liquid chromatography–mass spectrometry (LC-MS).

Results

We found that chronic treatment with an anti-PD1 antibody treatment reduced the levels of adenosine and inosine in MC38 allografts compared to control treated animals. Intriguingly, cAMP levels were higher in the tumor dialysates from anti PD-1 treated mice compared to controls. Moreover, oxaliplatin treatment produced a rapid and transient increase in adenosine and inosine levels in MC38 tumors. In contrast, changes in the concentration of these metabolites were much less pronounced in the SC space. From the same samples we were able to quantify the levels of oxaliplatin, and found that the levels in the core of the tumor were significantly less that in the SC of the same animal.

Conclusions

Here we bring proof-of-concept evidence that pharmacological and pharmacokinetic changes can be measured within tumor allografts in freely moving animals. Utilizing in vivo microdialysis we are able to simultaneously measure the levels of oncomodulators and anti-cancer drugs. Our experimental method may help the development of effective oncology therapeutic strategies.

P96 Assessment of RNA turbulence and PD-L1 expression on tumor-infiltrating lymphocytes in breast cancer

Apoorva Mylavarapu, BS1, Scott Morris2, Maulik Patel, PharmD, PhD3, Sandip Patel, MD4
1University of California at Los Angeles, David Geffen School of Medicine, San Diego, CA, USA; 2Paradigm Diagnostics, Phoenix, AZ, USA; 3Abbvie Inc., Redwood City, CA, USA; 4University of California San Diego, Moores Cancer Center, La Jolla, CA, USA
Correspondence: Maulik Patel (mpatel@abbvie.com)

Background

Triple negative breast cancer (TNBC) as a subtype is generally associated with poorer outcomes compared to other breast cancer subtypes. In this study, we explore the utility of a novel tissue-based biomarker, RNA turbulence score, derived from a panel of 56 gene targets. We describe the RNA turbulence landscape and examine potential association of RNA turbulence score with PD-L1 positive tumor infiltrating lymphocytes (TILs) in breast cancer tissue samples.

Methods

Biomarkers were examined in breast cancer tissue samples (N=516). Expression of PD-L1 on tumor/tumor infiltrating lymphocytes (TILs) was assessed by immunohistochemistry with two different anti-PD-L1 clones (22C3 and E1L3N). PD-L1 positivity for this analysis was defined as staining intensity ≥ 1+ with tumor/immune cell staining of >1%. Molecular testing was conducted at Paradigm Diagnostics utilizing a 56-gene panel for RNA turbulence assessment. RNA turbulence score was calculated by assessing the number of genes with mRNA over- expression by at least 5-fold or greater with a significance of p<0.01.

Results

Average RNA turbulence score was higher (12.45 vs. 10.18, Mann-Whitney p<0.001) in patients whose tumor samples were positive for PD-L1 expression on TILs (19%, N=55) than in negative samples (N=233). A total of 40% (18/45) of TNBC samples were positive for TILs expressing PD-L1. In contrast, 11% (8/70) of estrogen receptor/progesterone receptor (ER/PR) positive tumors had evidence of PD-L1 positive TILs. The greater percentage of samples expressing PD-L1 on TILs in TNBC samples than in ER/PR-positive samples was statistically significant (p=0.0005; Fisher’s exact test). Average RNA turbulence score was also higher in TNBC samples than in ER/PR-positive samples (11.96 vs. 9.79, Mann-Whitney p<0.01). ER/PR-positive samples had ESR1 exclusively over-expressed with AREG and ERBB2 relatively higher in expression in comparison to TNBC. In contrast, the top three genes exclusively over-expressed in TNBC were VEGFA, BAX, and LRP6.

Conclusions

RNA turbulence score is greater in breast cancer patients whose tumor samples were also positive for PD-L1 expressing TILs. Additionally, we observe greater PD-L1 positive TILs in TNBC samples in comparison to ER/PR-positive samples. These findings suggest that RNA turbulence score may in part identify an increased immune active tumor microenvironment as assessed by PD-L1 positive TILs. Studies correlating RNA turbulence with response to anti-PD(L)-1 directed therapies in TNBC are warranted.

P97 Different functionality of CD8 PD-1-positive CD28-negative T cells in the periphery and in the tumor of lung cancer patients

Belinda Palermo1, Ornella Franzese2, Mariangela Panetta1, Giulia Campo1, Fabiana Cecere1, Virginia Ferraresi1, Gabriele Alessandrini1, Francesco Facciolo1, Gennaro Ciliberto, MD1, Paola Nistico', MD1
1IRCCS - Regina Elena National Cancer Institute, Rome, Italy; 2School of Medicine, University of Tor Vergata, Rome, Italy
Correspondence: Paola Nistico' (paola.nistico@ifo.gov.it)

Background

The main objective of cancer immunotherapy is an efficacious control over tumor progression through the generation of a strong and persistent T-cell mediated immune response. CD8+ T cells are key players able to recognize and kill cancer cells, which experience phenotypic and functional changes due to the constant exposure to tumor-associated antigens, frequently in association with a dysfunctional state mediated by co-inhibitory receptors, including Programmed Death 1 (PD-1). Recently CD28 has been proven to be the main downstream target of PD-1- mediated signaling [1], and accordingly we have reported that a subset of Ag-specific CD8+ T-cell clones, characterized by PD-1 expression in the absence of CD28, show high proliferative capability and an AKT-dependent anti-tumor functionality sustained by ICOS [2-4]. Whereas, the co-expression of PD-1 and CD28 confers an exhausted phenotype and a defective anti-tumor functionality, reversible by PD-1 blockade. The role of these subsets in the tumor site is still not clarified, and there is a need to determine their presence and functionality to improve current therapies.

Methods

T cells were isolated from peripheral blood, tumoral (T) and adjacent non-tumoral (NT) tissue of lung cancer patients and analyzed by multiparametric flow cytometry for phenotypic characterization and polyfunctionality. Ag- specific CD8+ T cell clones (Melan-A and gp-100) were obtained as described [2-4].

Results

To elucidate the critical role of PD-1 in regulating CD8+ T-cell functionality, we have investigated the phenotypic and functional distribution of CD8+ T cells, with respect to PD-1 and CD28 expression, in the peripheral blood of patients with different solid tumors and at the tumor site of lung cancer patients. Preliminary results indicate that distinct PD-1+CD28- and PD-1+CD28+ CD8+ subsets could be found among T cells isolated from the periphery and both NT and T tissues. In the periphery we found that the differentiation and functional pattern of these T cells is similar to that identified in Ag-specific CD8+ T-cell clones [2-4], also in terms of GrzB, IFN-gamma and TNF- alpha production, ICOS and Ki-67 expression. Differently, when we compared periphery and tissue sites, we observed a heterogeneous phenotype and functionality of CD8+ T cells, also in terms of polyfunctionality and frequency of CD103+CD69+ T-resident as well as CD39+CD127- T-cell subsets, recently described as the major tumor-reactive T cells [5, 6].

Conclusions

These results highlight the critical role of PD-1 and CD28 molecules in regulating T-cell functionality and may help in the identification of biomarkers predicting the efficacy of anti-PD-1 therapy.

Acknowledgements

§ B.P. and O.F. contributed equally to this work. This work was supported by Associazione Italiana Ricerca sul Cancro (AIRC) IG n°15224

References

1. Hui E, Cheung J, Zhu J, et al. T cell costimulatory receptor CD28 is a primary target for PD-1-mediated inhibition. Science. 2017 Mar 31;355(6332):1428-1433.

2. Palermo B, Franzese O, et al. Antigen-specificity and DTIC before peptide-vaccination differently shape immune- checkpoint expression pattern, anti-tumor functionality and TCR repertoire in melanoma patients. OncoImmunology. In Press. https://doi.org/10.1080/2162402X.2018.1465163;

3. Franzese O, Palermo B, et al. Polyfunctional Melan-A-specific tumor-reactive CD8(+) T cells elicited by dacarbazine treatment before peptide-vaccination depends on AKT activation sustained by ICOS. OncoImmunology 2016. Feb 1;5(5):eP582

*Corresponding author email: rajesh.gottimukkala@thermofisher.com03;

4. Palermo B, et al. Dacarbazine treatment before peptide vaccination enlarges T-cell repertoire diversity of melan- a-specific, tumor-reactive CTL in melanoma patients. Cancer Res. 2010 Sep 15;70(18):7084-92.

5. Duhen T, Duhen R, Montler R et al. Co-expression of CD39 and CD103 identifies tumor-reactive CD8 T cells in human solid tumors. Nat Commun. 2018 Jul 13;9(1):2724.

6. Thommen DS, Koelzer VH, Herzig P et al. A transcriptionally and functionally distinct PD-1+ CD8+ T cell pool with predictive potential in non-small-cell lung cancer treated with PD-1 blockade. Nat Med. 2018 Jun 11.

Ethics Approval

The study was approved by Regina Elena National Cancer Institute's Ethics Board, approval number 1008/17

P98 High gene expression of estrogen and progesterone receptor is associated with increased T cell infiltration in patients with NSCLC

Michael Oh, MD, Jonathan Anker, MD, Young Kwang Chae, MD
Northwestern University Feinberg School of Medicine, Chicago, USA
Correspondence: Young Kwang Chae (young.chae@northwestern.edu)

Background

Recent reports have demonstrated that hormonal markers, including estrogen and progesterone receptor (ER and PR) status, may have prognostic and predictive relevance in non-small cell lung cancer (NSCLC). The precise impact of hormone signaling on clinical outcomes in NSCLC remains unclear, as initial studies have had diverging results. Here, we investigate the impact of hormone receptor status on tumor inflammation and survival in patients with NSCLC.

Methods

A dataset of NSCLC patients was obtained from The Cancer Genome Atlas (TCGA). RNA-Seq data was used to determined mRNA expression levels of ESR1 (ER-alpha), ESR2 (ER-beta), PGR (PR), and ARO (aromatase). Tumor infiltration by activated T cells was predicted using a previously-described method based on expression of immune metagenes [1]. Overall survival (OS) and progression-free survival (PFS) was assessed using Kaplan-Meier analyses with log-rank test.

Results

High levels of ESR1 was associated with significantly increased proportion of tumors infiltration by CD4+ (58% of tumors vs 19%, adjusted p < 0.001) and CD8+ activated T cells (55% vs 14%, adj. p < 0.001). Increased expression of PGR similar was associated with increased CD4+ (44% vs 18%, adj. p = 0.001) and CD8+ (48% vs 12%, adj p < 0.001) activated T cells. There were no significant differences based on ESR2 or ARO. These findings remained even after stratifying patients based on sex and histologic tumor type. In a multivariate logistic regression analysis, ESR1, PGR, and tumor mutation burden all were identified as independent factors predicting T cell infiltration.

However, greater expression of ESR1, PGR, or a combined measure of both genes did not confer greater overall or progression-free survival in this cohort.

Conclusions

Increased gene expression of estrogen receptor-α and progesterone receptor was associated with increased activated T cell infiltration in patients with NSCLC. The relevance of these findings will need be validated, potentially with clinical studies using immunotherapy based on hormone receptor status.

References

1. Angelova M, et al. Characterization of the immunophenotypes and antigenomes of colorectal cancers reveals distinct tumor escape mechanisms and novel targets for immunotherapy. Genome Biol. 2015;16:64.

P99 RRx-001 is a Phase 3-Ready small molecule dual inhibitor of CD47 and SIRPalpha

Pedro Cabrales, PhD1, Bryan Oronsky, MD, PhD2, Tony Reid, MD PhD2, Corey Carter, MD2
1University of California San Diego, La Jolla, CA, USA; 2EpicentRx Inc., La Jolla, CA, USA
Correspondence: Bryan Oronsky (boronsky@epicentrx.com)

Background

CD47 binds to SIRPα on the surface of macrophages delivering a “do not eat” signal to suppress phagocytosis. To evade macrophage-mediated destruction, tumor cells frequently overexpress CD47. One area of intense interest is the targeting of CD47 with monoclonal antibodies (mAbs), three of which, Hu5F9-G4, CC-90002, and TTI-621, have proceeded to clinical trials [1]. However, these mAbs have been associated with severe hemolytic anemia and thrombocytopenia. RRx-001 is a minimally toxic small molecule that dually downregulates CD47 on tumor cells and SIRPα on macrophages and triggers tumor associated macrophage phagocytosis of tumor cells in vitro and in vivo. RRx-001 is entering Phase 3 trials for the treatment of multiple cancer indications.

Methods

The effect of RRx-001 on the expression of CD47 and SIRPα on macrophages was evaluated with Western blotting and flow cytometry. An in vitro phagocytotic assay was used to determine whether RRx-001 promoted engulfment of A549 tumor cells by macrophages. Transcriptional mRNA profiling in murine tumor associated macrophages (TAMs) was performed to analyze the cytokine profile of TAMs in the presence or absence of RRx-001. Finally, nude mice bearing A549 tumors were treated with RRx-001 in the presence or absence or absence of clodronate to determine the effect of macrophage depletion on RRx-001 anticancer activity.

Results

RRx-001 was shown to downregulate CD47 and SIRPα expression on tumor cells and macrophages, respectively, and to promote the phagocytosis of high-expressing CD47 A549 tumor cells. RRx-001 also stimulated the production of pro-inflammatory cytokines in TAMs. In tumor bearing mice, depletion of macrophages by clodronate reduced the antitumor effects of RRx-001.

Conclusions

RRx-001 is a Phase 3-ready small molecule innate immune checkpoint inhibitor, which triggers tumor associated macrophage phagocytosis of high-expressing CD47 tumor cells. Dual downregulation of CD47 and SIRPα by RRx- 001 results in TAM repolarization and phagocytosis of tumor cells. Depletion of macrophages by clodronate in tumor-bearing mice reduced the antitumor effects of RRx-001 and further suggests that the target of RRx-001 is the macrophage.

References

Liu X, Kwon H, Li Z, Fu Y. Is CD47 an innate immune checkpoint for tumor evasion? J Hematol Oncol. 2017;10:12. DOI: 10.1186/s13045-016-0381-z

Ethics Approval

All applicable international, national, and/or institutional guidelines for the care and use of animals were followed.

P100 Exploring dissociated human tissues as an alternative to fresh tissue for multiple downstream applications

Rebecca Parker1, Shawn Fahl, PhD2
1Conversant Bio, Huntsville, AL, USA; 2Folio Conversant, Huntsville, AL, USA
Correspondence: Shawn Fahl (shawn.fahl@folioconversant.com)

Background

The foundation of the future of biomedical research requires access to highly-annotated primary human biospecimens. The logistical barriers of acquiring fresh tissue remains an impediment to advances in medicine, requiring the coordination of not only the tissue collection but also the downstream applications in the laboratory. Dissociation and cryopreservation of solid tissue provides a solution to this problem. These single cell suspensions remain viable following cryopreservation and ease the demands on large-scale experimental assays. Furthermore, these cells provide the ability to screen new biomarkers and therapeutic targets as they are uncovered without the need to source new fresh tissue.

Methods

Using this model, we have analyzed viably cryopreserved single cell suspensions generated from over 400 unique patients across 11 oncology indications by flow cytometry. Flow cytometry allows for the identification of cell surface marker expression on the single cell level and provides in-depth characterization of the cellular composition of the tumor microenvironment. This large-scale characterization revealed indication-specific trends to the tumor composition, which are vitally important considerations as the next-generation of therapeutic interventions are developed.

Results

N/A - Results pending late breaking submission requested.

Conclusions

Viably cryopreserved single cell suspensions from solid tissues provides numerous benefits to the logistical demands of sourcing fresh tissue. Using solid tumor indications as a model, we have demonstrated the utility of cryopreserved dissociated tumor cells to understand and screen the cellular composition of the tumor microenvironment. In particular, these results highlight the patient-specific heterogeneity of the tumor microenvironment, but also demonstrates indication-specific trends that are crucial when developing future immunotherapies.

P101 Evaluation of ctDNA mutations detected in plasma as potential correlate of immunotherapy efficacy in NSCLC patients

Namrata Patil, PhD2, Yuqiu Jiang, PhD3, Wei Zou2, Johnny Wu3, Stephanie Yaung3, Susan Flynn2, Maureen Peterson2, Eric Peters2, Priti Hegde, PhD2, John Palma3, Marcin Kowanetz, PhD2, David Shames, PhD2, Yuqiu Jiang, PhD3
1Genentech/ Roche, South San Francisco, CA, USA; 2Genentech, South San Francisco, CA, USA; 3Roche Sequencing Solutions, Pleasanton, CA, USA
Correspondence: Namrata Patil (patil.namrata@gene.com)

Background

Multiple recent studies have shown that tumor mutation burden (TMB) may be a surrogate for the overall neoantigens likely to presented for an effective immune response. High TMB, including TMB measured in blood (Gandara et al. 2017), has been associated with clinical benefit with checkpoint inhibitors in several malignancies. The goal of this analysis was to evaluate whether the number of mutations detected in ctDNA by the AVENIO Surveillance panel (200kb) correlates with immunotherapy efficacy in NSCLC patients.

Methods

A subset of 375 baseline plasma samples from 2L+ NSCLC subjects enrolled in study OAK (NCT02008227) were analyzed using the AVENIO ctDNA Surveillance kit**, which covers 200 kb (Roche, Branchburg, NJ), from 375 pts. Data from 108 patient samples have been analyzed so far. The Surveillance kit contains 17 cancer driver genes and an additional 180 frequently mutated genes in cancer. This kit is capable of detecting four mutation classes: SNVs, fusions, CNVs and Indels. Tumor tissue samples from these patients were also analyzed for tumor mutation burden (tTMB) using the FoundationOne assay.

Results

102 samples of the 108 (94%) baseline plasma samples that passed QC metrics were successfully sequenced. All 102 samples had somatic variants detected. The median number of variants detected per patient was 7. Mutant molecules per milliliter (MMPM) was also assessed for each of the baseline samples. The median MMPM was 139, ranging from 1 to 1,972 for these 102 samples. Preliminary analysis of the number of variants detected with a limited filters algorithm was positively correlated (r=0.56) with tissue Tumor Mutation Burden (tTMB).

Conclusions

ctDNA testing with molecular barcoded sequencing and digital background error suppression of a small 197 gene panel offers high sensitivity for tumor variant detection (all 102 samples with variants detected). This study demonstrated that tumor variants can be detected in blood in pre-treatment samples using the AVENIO Surveillance kit. Preliminary analysis suggests that the number of mutations detected correlate with immunotherapy efficacy, specifically PFS. Analysis of all 375 pts as well as association with clinical benefit to atezolizumab will be presented.

Trial Registration

NCT02008227

References

1. Gandara DR, Pawel JV, Sullivan RN, Helland A, Han J, Aix SP, Rittmeyer A, Barlesi F, Kubo T, Park K, Goldschmidt JH, Gandhi M, Yun C, Yu W, Matheny C, He P, Sandler A, Ballinger M, Fehrenbacher L. Journal of Clinical Oncology. 2017; 35:15_suppl, 9001-9001.

P102 Localized measurement and clinical significance of OX40 and OX40L expression in human non-small cell lung cancer (NSCLC)

Angelo Porciuncula, PhD1, Micaela Morgado1, Sima Zacharek, PhD2, Maria Toki, MD, MSc1, Kostas Syrigos3, Vamsidhar Velcheti, MD FACP4, Joshua Frederick, PhD2, Roy Herbst, MD, PhD1, Kurt A. Schalper, MD, PhD1
1Yale University, New Haven, CT, USA; 2Moderna Therapeutics, Cambridge, MA, USA; 3Athens School of Medicine, Athens, Greece; 4Cleveland Clinic, Pepper Pike, OH, USA
Correspondence: Kurt A. Schalper (kurt.schalper@yale.edu)

Background

Immunostimulatory therapies targeting co-inhibitory T-cell checkpoint pathways such as PD-1 and CTLA-4 produce lasting anti-tumor responses in a proportion of patients with diverse malignancies, including NSCLC. However, the majority of patients show primary resistance to treatment and a fraction of those initially responding have subsequent disease progression. Activation of co-stimulatory signals such as the OX40/OX40L pathway can favor anti-tumor immune responses and clinical trials modulating this pathway are currently ongoing. The expression, tissue distribution, biological associations and clinical significance of OX40/OX40L protein expression in human tumors remain largely unexplored.

Methods

Using formalin-fixed paraffin-embedded (FFPE) preparations from cell line transfectants and human tissue controls, we validated and standardized a multiplexed quantitative immunofluorescence (mQIF) assay for simultaneous and localized measurement of DAPI (all cells), cytokeratin for epithelial tumor cells (clone-AE1/AE2), OX40 (clone- SP195) and OX40L (clone-D6K7R). We used this panel to interrogate 619 stage I-IV NSCLCs from 3 retrospective cohorts represented in tissue microarray format (cohort#1 [Yale, n=280], cohort#2 [Greece, n=207]). In addition, we analyzed a collection of lung adenocarcinomas with molecular testing for major oncogenic drivers (cohort#3 [Yale, n=132]). The targets were measured in all cells of the preparation using fluorescence co-localization with DAPI and specifically in tumor (CK-positive) and stromal (CK-negative) cells. Associations between the markers and with major clinicopathological variables, driver mutations and survival were studied.

Results

Expression of OX40 protein was seen in 90% of NSCLCs in cohort#1 and 87% of cases in cohort#2. OX40 staining was predominantly located in stromal cells with membranous staining pattern. Detectable OX40L signal was identified in 9% of cases in the first cohort and 14% of the second collection with membranous/perinuclear staining pattern and predominant expression in CK-positive tumor cells. There was no clear association between OX40 and OX40L expression in the cohorts. The levels of OX40 and OX40L were not consistently associated with major clinicopathogical variables, level of T-cell infiltration or with the presence of oncogenic mutations in EGFR/KRAS in lung adenocarcinomas (cohort#3). Elevated levels of OX40L protein (but not OX40) were significantly associated with longer 5-year overall survival in both NSCLC cohorts (log-rank P=0.03 and P=0.04, respectively).

Conclusions

Our preliminary data show that OX40 protein is expressed in the majority of NSCLCs with predominant location in stromal/immune cells and does not segregate with a specific clinicopathological variant or molecular subtype. Elevated OX40L, observed in ~15% of NSCLCs, is unrelated with OX40 levels and has favorable prognostic value.

P103 Bulk and single-cell TCR and transcriptome profiling reveals significant phenotypic, spatial and temporal heterogeneity in the TIL repertoire of pancreatic cancer and melanoma patients

Chin Leng Tan1, Ignacio Heras-Murillo, MSc1, Katharina Lindner1, Aaron Rodriguez Ehrenfried1, Lena Appel1, Anna-Katharina Koenig2, Markus Buechler2, Ugur Sahin, MD3, Jessica Hassel2, Oliver Strobel2, Michael Flossdorf4, Rienk Offringa, PhD1, Isabel Poschke, PhD1
1German Cancer Research Center, Heidelberg, Germany; 2Heidelberg University Hospital, Heidelberg, Germany; 3Johannes Gutenberg University Mainz, Mainz, Germany; 4Technical University of Munich, Germany, Heidelberg, Germany
Correspondence: Isabel Poschke (i.poschke@dkfz.de)

Background

Presence of tumor-infiltrating lymphocytes (TIL) is associated with good survival in many cancers, and harnessing of the T-cell response through checkpoint inhibition or infusion of ex vivo expanded TIL can result in tumor regression. While pancreatic ductal adenocarcinoma (PDA) has been considered a poorly-immunogenic tumor, we detect TIL infiltrates - and aggregates - in the majority of patients with primary resectable disease. PDA TIL phenotype and expansion capacity resemble that of melanoma TIL - our benchmark of an immunogenic tumor, and 80% of in vitro expanded PDA TIL cultures exhibit reactivity against autologous tumors.

Methods

To gain a better understanding of TIL reactivity, we performed repertoire profiling of freshly isolated TIL by T-cell receptor (TCR) deep- and high-throughput single-cell sequencing and transcriptomics, as well as phenotypic and functional analysis of bulk and single-cell expansion cultures.

Results

Intra-tumoral and intra-patient heterogeneity is significant and highly individual: TIL repertoires from multiple regions of the same tumor show an overlap between 8.4-100%. Lesions within the same patient share between 0-70% of TCRs and tend to overlap less if biopsies are not acquired concurrently, indicating a continuous turn-over or reshaping of TIL composition. Notably, repertoire-sharing is most prominent among the largest TIL clones, possibly explained by efficient migration/re-circulation of some clones, or their maintenance by ubiquitously expressed (tumor-)antigens. Enrichment of highly frequent CDR3 sequences within the TIL repertoire suggests in situ proliferation in response to tumor-derived antigens. Droplet-based TCR- and transcriptome-analysis of >6000 PDA TIL cells revealed that dominant T-cell clones exhibit multiple distinct phenotypes, enriched for markers associated with activation, proliferation and exhaustion. Importantly, TIL in vitro expansion - with protocols established for therapeutic TIL application- induces dramatic shifts in repertoire composition, resulting in loss of dominant clones and enrichment of bystander clones with high proliferative potential.

Conclusions

Our findings call for careful sampling and optimized culture conditions for TIL infusion products and illustrate the need to probe T-cell reactivity directly ex vivo. The heterogeneous TIL response implies that therapeutic efficacy of TCR gene therapy using tumor-dominant TCRs could be more consistent than that of TIL therapy. Emerging high- throughput single cell technologies and resulting data will facilitate rapid identification of relevant TIL populations.

Ethics Approval

The study was approved by the local ethics committee at the medical faculty of Heidelberg university and conducted in accordance with the declaration of Helsinki.

P104 Radiomic features that can predict response to PD-1 inhibitors in late-stage Non-Small Cell Lung Cancer are also associated with tissue-derived measures of immune response

Prateek Prasanna, PhD1, Mohammadhadi Khorrami, PhD1, Pradnya Patil2, Kaustav Bera, MBBS1, Vamsidhar Velcheti, MD FACP2, Anant Madabhushi, PhD1
1Case Western Reserve University, Cleveland, OH, USA; 2Cleveland Clinic, Cleveland, OH, USA
Correspondence: Prateek Prasanna (pxp238@case.edu)

Background

Nivolumab is a FDA approved immune checkpoint inhibitor (ICI) for treatment of patients with chemotherapy refractory advanced NSCLC. The current standard for identifying candidates who would benefit from ICIs is sub- optimal. First, the role of computer-extracted textural descriptors (radiomic features) [1] on baseline CT in predicting response to ICIs, is investigated. Secondly, since the degree of immune response is reflective of a cancer’s ability to respond to ICIs, understanding the distribution of lymphocytic infiltration will help provide a morphological basis for the observed radiographic phenotypes. Towards that end, the predictive radiomic features are correlated with lymphocytic arrangement to understand their morphologic underpinning.

Methods

Non-contrast CT scans, before ICI treatment, were retrospectively acquired from 73 NSCLC patients. Patients with an objective response (complete/partial response) per RECIST after two cycles of nivolumab were defined as “responders” and patients with progressive disease were defined as “non-responders”. 454 intra-tumoral texture, 24 shape features and 7426 features from annular rings outside the expert-annotated nodules, capturing different measures of phenotypic heterogeneity, were extracted from the baseline scans. A linear discriminant analysis (LDA) classifier was trained using the most predictive features identified on the discovery set (n=29) and validated on the test set (n=44). Digitized H&E histology scans of baseline biopsies were available for 56 cases. The nuclei were first segmented [2] and then classified into either lymphocytes or non-lymphocytes using texture, shape, and color features. 76 features quantifying density or compactness of tumor infiltrating lymphocyte (TIL) clusters were subsequently extracted. A pairwise Spearman correlation analysis was performed between the TIL compactness measures and the top discriminating radiomic features.

Results

A combination of 2 intra-tumoral, 6 peri-tumoral and 1 shape delta radiomic feature yielded an area under the receiver operating characteristic curve (AUC) of 0.85 ± 0.05 within the discovery set and an AUC=0.81 within the validation set. TIL density was found to be statistically significantly (correlation coefficient=-0.5, p<0.05) correlated with a peritumoral Gabor [3] feature, as illustrated in Figure 1.

Conclusions

Radiomic features extracted from baseline CT scans were predictive of objective response to ICIs. The TIL compactness features model relationships between lymphocytes and their surrounding cells. Presence of an immune infiltration is more likely to manifest via unique textural patterns in the tumor environment. The radiomic features could therefore be capturing the degree of immune response, which in turn is known to be correlated with the likelihood that the tumor will have a favorable response.

Acknowledgements

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under award numbers 1U24CA199374-01, R01CA202752-01A1, R01CA208236-01A1, R01 CA216579-01A1, R01 CA220581-01A1, National Center for Research Resources under award number1 C06 RR12463-01, the DOD Prostate Cancer Idea Development Award; the DOD Lung Cancer Idea Development Award; the DOD Peer Reviewed Cancer Research Program W81XWH-16-1-0329, the Ohio Third Frontier Technology Validation Fund, the Wallace H. Coulter Foundation Program in the Department of Biomedical Engineering and the Clinical and Translational Science Award Program (CTSA) at Case Western Reserve University. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References

1. Prasanna P, Patel J, Partovi S, Madabhushi A, Tiwari P.. Radiomic features from the peritumoral brain parenchyma on treatment-naive multi-parametric MR imaging predict long versus short-term survival in glioblastoma multiforme: preliminary findings. European radiology. 2017; 27(10): 4188-4197.

2. Veta M, Van Diest PJ, Kornegoor R, Huisman A, Viergever MA, Pluim JP.. Automatic nuclei segmentation in H&E stained breast cancer histopathology images. PloS one. 2013 8(7): e70221.

3. Tuceryan M, Jain AK. Texture analysis. In Handbook of pattern recognition and computer vision (pp. 235-276).

Ethics Approval

The study protocol was approved under University Hospitals (UH) IRB 02-13-42C
Fig. 1 (abstract P104).
Fig. 1 (abstract P104).

Radiomic expression and TIL density patterns

P105 Response to immunotherapy in hepatocellular carcinoma, a single-institutional analysis

Petra Prins1, Bhavana Singh, MD2, AIwu He2
1Georgeotwn University, Washington, DC, USA; 2Medstar Georgetown University Hospital, Washington, DC, USA
Correspondence: Petra Prins (pap48@georgetown.edu)

Background

Despite advances in our understanding of the molecular pathways involved in hepatocellular carcinoma (HCC), therapeutic options remain limited and patient (pt) survival is dismal. Immunotherapy (IO) is one of the newer and more promising options for the treatment of HCC, with nivolumab being at the forefront. In the CheckMate 040 non-randomized, open-label, phase-1/2 study of nivolumab in pts with advanced liver cancer, a median overall survival (OS) of 15 months (m), an overall response rate of 15%, and a median duration of response of 16.6m was observed

Methods

In this single-institutional retrospective analysis, 30 pts with HCC received one of five different IO regimens. Thirteen pts had received sorafenib prior to initiation of IO. Patients received either atezolizumab plus bevacizumab, cemiplimab, pembrolizumab, nivolumab, or nivolumab plus ipilimumab until disease progression (PD) or unacceptable toxicity. Responses were assessed using RECIST v 1.1 criteria for stable disease (SD), partial response (PR) and PD, and blood biomarker levels, and were correlated with clinical outcomes like best response and progression free survival (PFS).

Results

Demographics for our cohort (n=30) were as follows: 73% were male and 27% female; 33% were African American, 30% Caucasian, 23% Asian, and 13% of other ethnicities. Of our cohort 33% had HBV and 30% had HCV as part of their disease etiology. A positive treatment response was observed in 90% of the pts with 67% having SD and 23% having a PR. Median OS was 20m (95%CI, 7.0-32.0) from the start of IO. A positive response to therapy significantly improved OS (20m SD versus 5m PD, p=0.005). A positive treatment response also showed clear PFS benefits: 20m (95%CI, 0-40.0) for PR, 7m (95%CI, 2.4-11.6) for SD and 3m (95%CI, 0-6.2) for PD (p=0.016). Data for circulating blood biomarkers and image biomarkers are currently being analyzed.

Conclusions

Overall, treatment with immunotherapy resulted in durable responses and favorable PFS and OS in pts with advanced HCC within our institution, making immunotherapy a valid option for the treatment of HCC. Additional correlation of IO treatment response with blood biomarkers might improve our understanding of treatment response.

P106 Multiplexed ion beam imaging (MIBI) for characterization of the tumor microenvironment (TME) across tumor types

Jason Ptacek, PhD, Rachel Finck, Murat Aksoy, Jay Tarolli, Jessica Finn
Ionpath, Inc, Menlo Park, CA, USA
Correspondence: Jessica Finn (jessica.finn@ionpath.com)

Background

Cancer arises from tumor cells taking advantage of complex relationships between stromal, vascular, and immune cell subsets. To date the ability to characterize the cellular composition and spatial organization of the TME has been limited by the techniques available to image the necessary number of biomarkers for broad phenotyping at a subcellular resolution. Here we show the applicability of MIBI for cell phenotype identification and their spatial relationships across multiple tumor types.

Methods

FFPE samples from tumor biopsies were imaged for their cellular composition and architecture using multiplexed ion beam imaging. Samples were stained with a panel of 15 antibodies, each labeled with a specific metal isotope. The panel was validated by comparing to single-plex IHC and included antibodies for tumor and immune cell subsets in addition to immunotherapy targets (PD-1, PD-L1). In MIBI the stained section is scanned via secondary ion mass spectrometry to image the tissue for expression of the antibody targets. Multi-step processing, including machine-learning-based segmentation, was used to produce images of the TME and determine both the frequency of cell subsets and the distance between immune cells and tumor cells.

Results

A total of 25 tumor specimens from 8 tumor types, plus control samples, were characterized for their immune profile, spatial organization of tumor and immune cells and their expression of PD-1 and PD-L1. Tumor-associated macrophages (TAMs) and tumor infiltrating lymphocytes (TILs) were observed in breast, gastric, lung, ovarian, and head and neck cancers. Nearest-neighbor immune:tumor distances revealed the level of mixing between tumor and immune cells (Figure 1). For example, ovarian serous carcinoma samples showed large numbers of infiltrating cytotoxic T cells and macrophages amongst tumor cells. However, the TMEs differed with one showing mixing of the populations and the second showing a compartmentalized organization. In contrast, an ovarian endometrioid carcinoma specimen showed much less robust immune cell infiltrate. Similar results of mixing or compartmentalization were observed for lung and gastric adenocarcinomas. Interestingly, PD-L1 expression was detectable in all gastric adenocarcinomas with the strongest expression observed in the tumor showing the most compartmentalization.

Conclusions

The function and phenotypes of cells can only be determined through the co-expression of multiple proteins.

Multiplexed imaging by MIBI reveals the complex tumor immune landscape by enabling the characterization of the spatial relationship of immune and tumor cells and the expression of immunoregulatory proteins. This work demonstrates the possibilities of MIBI for future patient stratification through characterization of the TME.
Fig. 1 (abstract P106).
Fig. 1 (abstract P106).

Spatial distribution of cells within tumor samples

P107 A comparative study of the PD-L1 IHC 22C3 and 28-8 assays in melanoma samples

Gabriel Krigsfeld, PhD1, Kim Zerba1, James Novotny Jr1, Shuntae Williams2, Michael Matthews, MS3, Hytham Al- Masri, PhD2, David Gold1, James White1
1Bristol-Myers Squibb, Princeton, NJ, USA; 2Hematogenix Laboratory Services, LLC, Tinley Park, IL, USA; 3Acupath Laboratories, Inc, Plainview, NY, USA
Correspondence: James White (James.White@bms.com)

Background

Nivolumab is a programmed death-1 (PD-1) receptor blocking antibody approved as monotherapy or in combination with ipilimumab for patients with unresectable or metastatic melanoma. The Dako programmed death ligand 1 (PD- L1) immunohistochemistry (IHC) 28-8 pharmDx is an FDA-approved PD-L1 complementary diagnostic for melanoma. While the Dako PD-L1 IHC 22C3 pharmDx is an FDA-approved companion diagnostic for several tumor types, it is not approved for melanoma. To date, studies have compared the 2 assays in non-small cell lung cancer and urothelial carcinoma. We report a comparison of the 28-8 and 22C3 assays on real-world melanoma samples, which provide new data to address the potential interchangeable use of these assays in clinical practice.

Methods

Formalin-fixed, paraffin-embedded melanoma samples were obtained from Acupath Laboratories, Inc (Plainview, NY) with basic demographic information. Staining and scoring of slides were performed at Hematogenix (Tinley Park, IL). The 28-8 and 22C3 PD-L1 assays were used per manufacturer’s protocols and evaluated using the 28-8 pharmDx tumor cell scoring algorithm [1]. Slides were randomized and pathologists were blinded to the assay at scoring, with paired 28-8 and 22C3 assay results read by the same pathologist. Samples that were heterogeneous (determined by bracket evaluation with H&E staining), had high melanin content, or failed to meet minimum criteria (≥100 tumor cells per 28-8 assay) were excluded from the final analysis.For the primary analysis, overall, positive, and negative percentage agreement (OPA, PPA, and NPA) at the ≥1% PD-L1 expression level were evaluated. Secondary analyses included agreement rates at the ≥5% expression level, Passing–Bablok regression, and Bland– Altman plots with summary statistics describing the differences between assays across the dynamic range.

Results

Of 265 samples analyzed, 202 were confirmed to contain melanoma tissue with quantifiable PD-L1 expression.

Average patient age was 66.4 years. Average sample age was <2 years (657 days [range 637–678]), and samples ranged from stage I to IV melanoma. High analytical concordance was observed between the 28-8 and 22C3 assays across paired melanoma samples. OPA, PPA, and NPA for all paired samples at the ≥1% PD-L1 expression level were 93.1%, 82.1%, and 97.3%, respectively. Identical PD-L1 expression values for the 28-8 and 22C3 assays were reported for 82.7% of samples, with a ≤10% difference in tumor cell membrane staining for 98.0% of samples.

Conclusions

These data support the potential interchangeability of the PD-L1 IHC 28-8 and 22C3 pharmDx assays for assessing tumor cell membrane PD-L1 expression on melanoma samples.

Acknowledgements

Medical writing assistance was provided by Bernard Kerr, PGDipSci, CMPP™, of Spark Medica Inc. (US), funded by Bristol-Myers Squibb. This study was funded by Bristol-Myers Squibb.

References

1. Dako. PD-L1 IHC 28-8 pharmDx SK005. Package Insert. 2017. [https://www.agilent.com/cs/library/packageinsert/public/128371004.PDF] Accessed July 30, 2018.

P108 HLA class I subtypes are associated with immune-related adverse events in patients with melanoma treated with ipilimumab

Anton Safonov, MD1, John Pluta1, Lu Qian2, Ravi Amaravadi1, Allison Applegate3, Elizabeth Buchbinder, MD4, Justine Cohen, DO4, Claire Friedman, MD5, Ruth Halaban, PhD6, F. Stephen Hodi, MD4, Christine Horak, PhD7, Douglas Johnson, MD, MSCI8, John Kirkwood, MD9, Tara Mitchell1, William Robinson, MD, PhD3, Lynn Schuchter, MD1, Jeffrey Sosman, MD10, Mario Sznol, MD6, Megan Wind-Rotolo, PhD7, Jedd Wolchok, MD, PhD5, Mingyao Li1, Peter Kanetsky2, Katherine L. Nathanson1
1Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; 2Moffitt Cancer Center, Tampa, FL, USA; 3University of Colorado, Denver, CO, USA; 4Dana Farber Cancer Institute, Boston, MA, USA; 5Memorial Sloan Kettering Cancer Center, New York, NY, USA; 6Yale University, New Haven, CT, USA; 7Bristol-Myers Squibb, Lawrenceville, NJ, USA; 8Vanderbilt Cancer Center, Nashville, TN, USA; 9University of Pittsburgh Medical Center, Pittsburgh, PA, USA; 10Feinburg School of Medicine, Northwestern University, Chicago, IL, USA
Correspondence: Katherine L. Nathanson (knathans@upenn.edu)

Background

Checkpoint immunotherapies have demonstrated substantial benefit in patients with melanoma. However, there are currently no established biomarkers to predict immune-related adverse events (irAEs). Because genetic variation at the human leukocyte antigen (HLA) locus is associated with increased risk of autoimmune disorders, we evaluated associations between irAEs and HLA subtypes.

Methods

We utilized clinical and genomic data from 269 chemotherapy and immunotherapy-naïve melanoma patients treated with ipilimumab monotherapy on the Bristol-Meyers-Squibb clinical trial CA184-169 (NCT01515189). DNA samples were genotyped on Affymetrix 6.0 and standard sample- and SNP-level quality control pipelines were conducted; only samples with high levels of European genetic ancestry (i.e. Caucasian) were further analyzed. Genotype imputation was completed using the Haplotype Reference Consortium Backbone. We used SNP2HLA to establish estimated allele dosages for each 4-digit HLA allele (n = 172), and subsequently collapsed these calls into one of nine established class I supertypes. We catalogued occurrences of immune-related adverse events by CTCAE grade and affected organ system. Logistic regression was performed to evaluate associations of class I HLA supertype allele dosage with irAE outcomes (grade < 2 vs grade ≥ 2) adjusting for ECOG score, dosage, and number of doses.

Results

HLA-A24 was significantly associated with hepatitis (Grade > 2), with meta-analysis odds ratio 3.69 [95% CI 1.75 to 7.51, p = 0.0005]. Odds ratio was 3.54 [95% CI 1.51 to 8.31, p = 0.0036] in the BMS dataset (remaining significant after correction for multiple comparisons), and OR 3.85 [95% CI 0.96 to 15.5, p = 0.056] from the consortium data set. The frequency of HLA-A24 in our population was 9.6%.

Conclusions

Our results demonstrate that a specific HLA subtype, HLA-A24, is associated with checkpoint-inhibitor related hepatitis. HLA-A24 has been explored in its dose-dependent response to hepatitis B and C vaccination. Our results potentially demonstrate unmasking of germline-mediated immune sensitivity by checkpoint inhibitors. This data warrants further study into HLA subtypes as predictive biomarkers for immune-related adverse effects. These findings should be confirmed in additional study populations, potentially with sequence-based HLA typing.

Trial Registration

NCT00135408, NCT00261365, NCT01515189

Ethics Approval

The protocol and all amendments were approved by the institutional review board or independent ethics committee for each study center

P109 Clinical efficacy of immune checkpoint inhibitors in patients with small cell lung cancer is associated with high tumor mutational burden and development of immune-related adverse events

Biagio Ricciuti, MD1, Suzanne Dahlberg1, Sasha Kravets1, Safiya Subegdjo1, Renato Umeton1, Adem Albayrak, MS1, Lynette Sholl2, Mark M. Awad1
1Dana Farber Cancer Institute, Boston, MA, USA; 2Brigham and Women's Hospital, Boston, MA, USA
Correspondence: Mark M. Awad (Mark_Awad@dfci.harvard.edu)

Background

Immune-checkpoint inhibitors (ICIs) have shown promising activity in only a fraction of patients with small cell lung cancer (SCLC), and factors associated with clinical benefit are not well characterized. High tumor mutational burden (TMB), quantified by whole exome sequencing, has been shown to predict response to ICIs in SCLC. However, whether targeted next generation sequencing (NGS) can be used to identify SCLC patients with high TMB who might benefit from treatment with immunotherapy is currently unknown. The relationship between the development of immune-related adverse events (irAEs) and immunotherapy response in SCLC is also unknown.

Methods

Patients with SCLC at the Dana-Farber Cancer Institute (DFCI) who received treatment with immunotherapy and/or had successful NGS were included in this study. TMB was determined using the DFCI NGS OncoPanel platform of >450 genes. The relationships between TMB, the development of irAEs, and clinical outcomes were determined among patients with SCLC treated with PD-1 inhibitors alone or in combination with a CTLA-4 inhibitor.

Results

Out of a total of 145 patients, 125 (86.2%) had successful NGS with TMB assessment and 64 (44.1%) were treated with ICIs. The median (range) TMB was 9.29 (1.21-33.89) mutations/megabase. Among 44 TMB-evaluable patients treated with ICIs, the median progression-free survival (mPFS) was significantly longer in the 21 patients with a TMB above median (“TMB-high”) compared to the 23 patients below median (“TMB-low”) (4.1 vs. 1.4 months, HR: 0.39 [95%CI: 0.19-0.77], P < 0.01) (Figure 1A). The median overall survival (mOS) was significantly prolonged in the TMB-high group compared to the TMB-low group (10.5 vs. 2.7 months, HR: 0.43 [95%CI: 0.20- 0.89], P = 0.02) (Figure 1B). Among 64 patients with SCLC treated with ICIs (44 TMB-evaluable), the 21 (32.8%) patients who developed at least one irAE had significantly longer mPFS (4.1 vs. 1.3 months, HR: 0.33 [95%CI: 0.19-0.56], P < 0.001) and mOS (12.5 vs. 3.2 months, HR: 0.35 [95%CI: 0.20-0.62], P < 0.01) compared to the 43 (67.2%) patients who did not develop irAEs (Figure 2 A,B). Patients with irAEs had a significantly higher mean TMB than those who did not experience any irAEs (13.09 vs. 9.41, P = 0.02).

Conclusions

Median PFS and OS on immunotherapy were significantly longer among patients with SCLC and high TMB and among those who developed at least one irAE. TMB determination by NGS may be a helpful biomarker for identifying patients who are likely to benefit from treatment with ICIs in SCLC.
Fig. 1 (abstract P109).
Fig. 1 (abstract P109).

See text for description.

Fig. 2 (abstract P109).
Fig. 2 (abstract P109).

See text for description.

P110 Modular repertoire analysis of blood transcriptomic immune response in metastatic renal cell carcinoma patients treated with Pazopanib

Darawan Rinchai, PhD1, Elena Verzoni2, Agata Cova2, Paola Squarcina2, Loris Cecco2, Paolo Grassi2, Raffaele Ratta2, Veronica Huber2, Matteo Dugo2, Monica Rodolfo2, Jessica Roelands, Master1, Damien Chaussabel1, Giuseppe Procopio2. Davide Bedognetti, MD, PhD1, Licia Rivoltini, MD2
1Sidra Medicine, Doha, Qatar; 2Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy
Correspondence: Davide Bedognetti (dbedognetti@sidra.org)

Background

Transcriptional modular repertoire analysis has been developed and successfully used as a basis for the selection of biomarkers and development of multivariate transcriptional indicator of disease progression in patients with systemic lupus erythematosus or infectious diseases. In the context of cancer immunotherapy, understanding the molecular mechanisms modulated by a given drug is critical to implement more efficient therapeutic approaches. Here, we employed a modular repertoire approach to investigate immune response in metastatic renal cell carcinoma patient treated with Pazopanib (Tyrosine kinase inhibitors).

Methods

Peripheral blood mononuclear cells collected from 8 metastatic renal cell carcinoma patients (mRCC) receiving first-line Pazopanib were prospectively analyzed at baseline, 3 and 6 months for whole-genome transcriptomic perturbations using Illumina HT12v4BeadChip. As for the best response, partial response was observed in 3 patients and stable disease in 5 patients. A set of 260 transcriptional modules was used for the analysis of this dataset using a pre-defined framework (1–3). A module is considered to be “responsive” to the treatment when significant changes in abundance are observed for a proportion of its constitutive transcripts that is greater that what could be expected by chance.

Results

We first assessed changes in transcript abundance at the modular level. The percentage of responsive transcripts constitutive of a given module was determined for each time point (group comparison) or individual comparison.

The group comparison analysis showed that module perturbations peaked at 3 months and decreased at 6 months. These perturbations include enrichment of modules M3.6 & M8.46 (Cytotoxic/NK), M4.11(Plasma cells), and M8.89 (Immune response) (Figure 1). In addition, individual-level analysis showed that Pazopanib administration was associated with the decreased of the immunosuppressive module M9.34 in 5 patients. Such module was up- regulated only in one patient who did not respond to treatment. Interestingly, a rapid increased of Interferon (IFN) modules (M1.2 & M3.4) was observed exclusively in responding patients (Figure 2).

Conclusions

These results suggest that Pazopanib has a strong immune modulatory effect and might reshape anti-tumor immunity by reducing immunosuppression and triggering cytotoxic mechanisms and IFN pathways. The peak of this immune modulation is observed 3 months after treatment. These data provide a strong rationale for exploring combinatorial immune-targeted therapy based on kinase inhibitors such as a Pazopanib and for implementing transcriptomic analysis of peripheral blood in the context of cancer immunotherapy.

References

1. Chaussabel D, Quinn C, Shen J, Patel P, Glaser C, Baldwin N, et al. A modular analysis framework for blood genomics studies: application to systemic lupus erythematosus. Immunity. 2008 Jul 18;29(1):150–64.

2. Obermoser G, Presnell S, Domico K, Xu H, Wang Y, Anguiano E, et al. Systems Scale Interactive Exploration Reveals Quantitative and Qualitative Differences in Response to Influenza and Pneumococcal Vaccines. Immunity. 2013 Apr 18;38(4):831–44.

3. Chaussabel D, Baldwin N. Democratizing Systems Immunology with Modular Transcriptional Repertoires Analyses. Nat Rev Immunol. 2014 Apr;14(4):271–80.
Fig. 1 (abstract P110).
Fig. 1 (abstract P110).

See text for description.

Fig. 2 (abstract P110).
Fig. 2 (abstract P110).

See text for description.

P111 Streamlined human immune monitoring with mass cytometry: 29 markers in a single tube with automated data analysis

Christina Loh, PhD1, Thiru Selvanantham, PhD1, Leslie Fung1, Michelle Poulin1, C. Bruce Bagwell2, Margaret Inokuma2, Clare Rogers, MS1, Greg Stelzer, PhD1, Steven Lott, PhD1
1Fluidigm, South San Francisco, CA, USA; 2Verity Software House, Topsham, ME, USA
Correspondence: Clare Rogers (clare.rogers@fluidigm.com)

Background

Immune monitoring is an essential method for quantifying changes in immune cell populations in health and disease.

The extreme heterogeneity of immune cells demands a high-parameter approach to more fully and efficiently quantify these changes. Mass cytometry is an ideal solution, enabling simultaneous detection of over 40 phenotypic and functional markers in a single tube of sample.

Methods

We developed a single-tube, 29-marker panel for mass cytometry based on the Human ImmunoPhenotyping Consortium (HIPC) consensus panel [1], expanded to allow identification of additional leukocyte subsets, particularly T cells. Automated data analysis with Verity Software House GemStone™ has been developed specifically for data collected with the panel, providing users with results in minutes, reducing time-to-answer and variability inherent in manual gating.

Results

Repeatability was tested with a single PBMC sample stained by a single technician in two technical replicates and acquired in triplicate on two Helios™ mass cytometers. SDs for percent of parent were 1% or less for 16 identified populations. Reproducibility was tested by determining the variability in measurements of five PBMC lots stained by five technicians and collected on two Helios instruments. CVs on mean percent of all measured populations were under 20%. R2 values for agreement of percent parent populations using the full 29 marker panel compared to smaller 10-to-12-marker panels for T, B and myeloid cell populations were 0.94 or higher. In addition, comparison of cell population frequencies determined using GemStone analysis with manual gating in WinList demonstrated a high degree of concordance.

Conclusions

We conclude that this panel kit can provide consistent immune population identification and enumeration for any given lot of PBMC, and that data generated by the kit is amenable to either manual or automated data analysis.

References

1. Maecker HT, McCoy JP, Nussenblatt R, Standardizing immunophenotyping for the Human Immunology Project. Nature Reviews Immunology 2012; 12: 191-200.

P112 Proteomic profiling of biomarkers for response to checkpoint immunotherapy in melanoma patients

Marijana Rucevic, PhD (m.rucevic@olink.com)
Olink Proteomics, Watertown, MA, USA

Background

Checkpoint immunotherapy has greatly improved clinical outcomes in patients with several malignancies including melanoma. However, only a subset of advanced melanoma patients generates durable responses to such immunotherapy. To date, there are no reliable methods to predict responders which urges a need to develop serological biomarkers of melanoma response to immune checkpoint therapy.

Methods

We applied an innovative and highly sensitive Proximity Extension Assay (PEA) for comprehensive profiling of ~1000 proteins in the plasma of 55 metastatic melanoma patients that received anti-PD-1 treatment whereas, 18 patients were previously treated with anti-CTLA-4 while 37 received anti-PD-1 monotherapy. Of the 55 patients, 42 were classified as having benefit from immunotherapy (‘responders’) and 13 had no benefit (‘non-responders’). Our study included analyses of samples collected at baseline, on-treatment (~67 days from baseline) and post-treatment from these patients.

Results

Proteome wide analysis identified changes in ~100 immune response proteins over the course of treatment with different characteristic behavior across the treatment period. The dynamic response proteins included current and new exploratory checkpoint targets (PD-1, PDL-1, LAG-3, Gal-9, CD27, TNFR2) as well as, many proteins associated with the events that are hallmarks of immune response to cancer, such as cytokines/chemokines, angiogenesis, vascular cell adhesion and oncogenic signaling proteins. Importantly, 7 proteins including HER3, SAA4, NID1, ST6GAL1, OPN, TIMP1 and CXCL13 were found distinctive between responders and non- responders at the first on-treatment time point and demonstrated potential to classify responders and non-responders Thus, they may represent a protein signature that can predict treatment response or serve as surrogate markers for efficacy to immune checkpoint therapy in melanoma.

Conclusions

Our study might provide better understanding of immune system interaction with checkpoint inhibitors and demonstrates potential to improve melanoma patient’s management that may have a significant impact to the field of immunotherapy.

P113 Digital spatial profiling of bone-marrow infiltrating immune cells in acute myeloid leukemia

Sergio Rutella, MD, PhD, FRCPath1, Jayakumar Vadakekolathu, PhD1, Sarah Church2, Heidi Altmann3, Elena Viboch, MS2, Jorn Meinel, MD3, Yan Liang, MD PhD2, Joseph Beechem, PhD2, Alessandra Cesano, MD, PhD2, Sarah Warren, PhD2, Gerhard Enhinger, MD4, Marc Schmitz, MD4, Martin Bornhauser, MD3
1Nottingham Trent University, Nottingham, UK; 2NanoString Technologies, Inc., Seattle, WA, USA; 3University Hospital Carl Gustav Carus, Dresden, Germany; 4Technische Universität Dresden, Dresden, Germany; 5The John van Geest Cancer Research Centre, Nottingham, UK
Correspondence: Sergio Rutella (sergio.rutella@ntu.ac.uk)

Background

The therapeutic approach in patients with acute myeloid leukemia (AML) has not changed substantially in >30 years. The discovery of new treatment strategies, including immunotherapy, remains a priority [1-3]. Herein, we employed Digital Spatial Profiling (NanoString Technologies, Seattle) to characterize the expression of 31 immuno- oncology (IO) proteins in 10 bone marrow (BM) samples from adult patients with newly diagnosed AML.

Methods

FFPE BM slides were incubated with fluorescent markers (CD3 to identify T cells, CD123 to identify AML blasts, and Syto83 to stain nuclei) to establish the overall tissue morphology, followed by a cocktail of antibodies conjugated with oligo tags via a photo-cleavable linker. We then identified regions of interest (ROIs: CD3-rich and CD3-poor) with visible light-based imaging and selected them for high-resolution multiplex profiling. Oligo tags from the selected ROIs released upon exposure to UV light, were collected via a micro-capillary tube, hybridized to 6-spot optical barcodes, and digitally counted using the NanoString nCounter® platform. In parallel, we measured the expression of 770 immune-related mRNAs (including 14 molecules covered by the DSP IO panel) using the Pan-Cancer Immune Profiling Panel™ (NanoString Technologies, Seattle).

Results

We selected 24 geometric ROIs per BM sample using fluorescent anti-CD3 and anti-CD123 antibodies. Overall, T cell abundance across the 10 BM biopsies was highly variable and median CD3 barcode counts were used to classify samples into CD3-rich and CD3-poor. In the context of the six CD3-rich biopsies (Fig. 1A), ROIs were assigned to either CD3low, CD3int and CD3high categories using the 25th and 75th percentile of the area-normalized CD3 barcode counts. The expression levels of PD1, B7-H3, CD45RO, FoxP3, CD4 and CD8 were significantly higher in CD3high compared with CD3int and CD3low ROIs. The density of CD8+ T cells with an activated/exhausted CD45RO+PD1+ phenotype correlated with PD-L1 expression, consistent with the establishment of adaptive immune resistance (Fig. 1B). A similarity matrix of IO proteins allowed us to identify co-expression patterns of monocyte/macrophage markers (CD14/CD68) and negative immune checkpoints (B7-H3 [CD276] and VISTA), as well as a correlation between PD-L1, Bcl-2 and PTEN within the selected ROIs (Fig. 1C). Finally, in situ expression of mRNA and barcode counts for CD19, CD14, CD68, CD56 and Bcl-2 were also significantly correlated.

Conclusions

This proof-of-concept study provides evidence for heterogeneous immune profiles and advances our understanding of the immuno-biology of AML. DSP could support the implementation of future immunotherapy clinical trials.

Acknowledgements

Grant support: Qatar National Research Fund (#NPRP8-2297-3-494), Roger Counter Foundation (Dorset, UK) and John and Lucille van Geest Foundation to S.R.

Ethics Approval

Bio-banked tissue biopsies were collected at Technische Universität Dresden, Germany, under approval by the SAL AML Biobank's Ethics Committee.
Fig. 1 (abstract P113).
Fig. 1 (abstract P113).

Digital spatial profiling of AML

P114 ImmunoINTEL, a flow cytometry based platform that identifies and quantifies the most critical cell subsets and related functional potential in dissociated solid tumors

Konstantin Salojin, PhD1, Christine Hauther1, Dai Liu1, Santosh Putta2, Norman Purvis, PhD1, Matt Westfall, PhD1
1Pierian Biosciences, Franklin, TN, USA; 2Qognit,Inc., Franklin, TN, USA
Correspondence: Konstantin Salojin (csalojin@pierianbio.com)

Background

Standard immunohistochemistry (IHC) approaches to tumor-infiltrating leukocyte (TIL) phenotyping yield limited information, as they utilize antibody (Ab) panels not broad enough to identify and functionally characterize the complexity of TIL subsets. A flow cytometry platform was developed that delivers quantitative, clinically relevant information to support the following: 1) Phenotypic and functional analysis of TILs to characterize the immune status within the tumor microenvironment (TME), and 2) Quantification of TIL and tumor cell (co-)expression profiles of targetable immunomodulatory receptors (IMR) and ligands (IMR-L).

Methods

Standardized flow cytometry set-up, QC and sample processing procedure were established. Automated instrumentation and sample processing procedures were implemented in a barcode-based workflow utilizing Hamilton STARlet automated liquid handlers fully integrated with an SMS/LIMS and multi-dimensional flow cytometry data analysis software. Six Ab panels, ranging from 6-12 colors were designed, optimized, and validated to profile PBMC and TIL subsets.

Results

Phenotypic composition of the following cell subsets in the TME was delineated and quantified using qualified antibody panels for: (1) T, B, and NK cells; (2) T regulatory cells; (3) myeloid cell subsets (classical/nonclassical monocytes, M1/M2-macrophages); (4) myeloid cells with suppressive phenotypes (M- and G-MDSC); (5) myeloid/plasmacytoid DCs; (6) epithelial and mesenchymal cells. The relative quantities of the TIL subsets were tabulated based on tumor type. The ImmunoINTEL panels were complemented with drop-in IMR and IMR-L markers (PD1/PDL1, TIGIT/CD112-CD155, TIM-3/Galectin-9, SIRPa/CD47, LAG-3/HLADR) to profile the functional status of TILs, and to study the cytotoxic potential of tumor TILs. Reliable separation of IMR/IMR-L positive TILs from negative TILs and IMR/IMR-L positive tumor cells from negative tumor cells was observed, with notable heterogeneity, across cell and tumor types. This included the elevated expression of several IMR-Ls on TILs and IMRs on epithelial and stromal cells, suggesting tumor- and TIL-intrinsic mechanisms modulating checkpoint interactions. Results demonstrate a highly reproducible data set with minimal variability in PBMC reference/control specimens (CVaverage < 5%; n = 30) and accurate quantitation of lymphoid, myeloid, epithelial, and mesenchymal cell subsets in tumor samples.

Conclusions

Stringent flow cytometry QC processes were developed and implemented to ensure quality and precision of phenotypic and functional analyses of dissociated TILs and tumor cells, capturing the most critical metrics of intra- tumor immune responses and providing quantitative characterization of IMR and/or IMR-L interactions for more refined selection of potential responders to immune checkpoint inhibitors.

Ethics Approval

This study was reviewed and approved by BioIVT’s and iSpecimen’s Ethics/Regulatory Committees, approval numbers: 0944, 0984, 0985, 01034, 01045, 01050, 01071, 01084, 01095, & 01096. All of the samples collected under the above listed approvals are in full compliance with applicable Good Clinical Practices as defined by U.S. Food and Drug Administration (FDA) and U.S. Department of Health and Human Services (HHS) regulations as well as the International Conference on Harmonization (ICH) guidelines.

P115 Predictive and pharmacodynamic biomarkers associated with treatment with the oral selective AXL Inhibitor bemcentinib in combination with pembrolizumab in patients with advanced NSCLC and Melanoma

Robert Holt, PhD1, David Micklem, PhD1, Anthony Brown1, Cornelia Schuster2, Oddbjørn Straume2, James Lorens, PhD1
1BerGenBio ASA, Bergen, Norway; 2Haukeland University Hospital, Bergen, Norway
Correspondence: James Lorens (james.lorens@uib.no)

Background

Bemcentinib (BGB324) is a first-in-class, oral, potent and highly selective inhibitor of the AXL tyrosine kinase currently in phase II clinical development across several cancer types. AXL over-expression has been observed in patients failing PD-1 therapy in several cancers whereas AXL inhibition via bemcentinib has shown synergistic effect with checkpoint blockade in pre-clinical models. Selective blockade of Axl by bemcentinib in combination with pembrolizumab in NSCLC and melanoma is currently being explored in two Phase 2 trials BGBC008 (NCT03184571) and BGBIL006 (NCT02872259). Here we report results of the biomarker research programmes designed to identify predictive and pharmacodynamic biomarkers associated with bemcentinib/pembrolizumab treatment.

Methods

Fresh pre-treatment tumour biopsies were mandatory for PD-L1 and AXL analysis by IHC. Plasma protein biomarker levels were measured using the DiscoveryMap v3.3 panel (Myriad RBM) at pre-dose and C2D1 to identify predictive and pharmacodynamic biomarkers associated with bemcentinib/pembrolizumab treatment. The co-localisation of AXL and PD-L1 in tumour infiltrating immune cells was determined using NeoGenomics MultiOmyx.

Results

One cycle of treatment with bemcentinib significantly altered soluble Axl protein levels in a subset of patients including those who had benefited from treatment. This observation was consistent across multiple disease indications and treatment regimes. Protein biomarkers predictive of patient benefit following bemcentinib treatment have been identified - correlations with AXL and PD-L1 IHC will also be presented. AXL is expressed in a subset of tumour infiltrating immune cells, primarily macrophages. Inaddition, AXL and PD-L1 were found to be co- expressed.

Conclusions

Predictive biomarker candidates were identified supporting potential companion diagnostics development for bemcentinib/pembrolizumab treatment. Pharmacodynamic biomarkers indicate that bemcentinib is selective and on target. AXL is expressed on macrophages and is co-expressed with PD-L1.

Trial Registration

NCT03184571 and NCT02872259

Ethics Approval

All relevant ethical and regulatory approvals were obtained.

P116 Quantitative measurement of CD8, CD68 and PD-L1 expression in a novel multiplex assay and associations of overall survival in non-small cell lung cancer (NSCLC) patients treated with anti-PD-1 Therapy

Fahad Shabbir, MD1, Jon Zugazagoitia, MD1, Yuting Liu, PhD candidate1, Katir Patel, PhD2, Brian Henick, MD3, Scott Gettinger, MD1, Roy Herbst, MD, PhD1, Kurt Schalper, MD, PhD1, David Rimm, MD, PhD1
1Yale School of Medicine, New Haven, CT, USA; 2Ultivue Inc, Cambridge, MA, USA; 3Columbia University Irving MedicalCenter, New York, NY, USA
Correspondence: Fahad Shabbir (fahad.ahmed@yale.edu)

Background

While PD-1 axis therapies have dramatically changed outcomes in some lung cancer patients, many patients don’t benefit from these immunotherapies. Quantitative immunofluorescence (QIF) may provide a method for selection of those that benefit. This approach has been limited by the fact that traditional fluorescent multiplexing using tyramide amplification and unique species antibodies result in complexity that would be challenging in the CLIA lab setting. Here we test a novel, single mix, multiplex approach to simultaneously assess CD8, CD68 and PD-L1 in immmuno- therapy treated NSCLC patients

Methods

A tissue microarray with 81 spots in two-fold redundancy was derived from Yale patients treated with immunotherapy between 2011and 2017. Both a traditional tyramide and species unique multiplex (DAPI, CD68, PDL1 and Cytokeratin) and a DNA-based kit (UltiMapper Kit 1: CD8, CD68, PD-L1 & Cytokeratin/SOX10 with a nuclear dye) were assessed using the PM2000 microscope and AQUA software. The UltiMapper antibody premix is a single step, antibody mix which decreases overall staining time in anticipation of CLIA lab usage. We validate this new assay by regression with conventional QIF and overall survival (OS).

Results

Reproducibility was measured using two slides from 2 different blocks on two different days and showed that they were significantly correlated for CD8 (R2 = 0.58), CD68 (R2 = 0.67) and PD-L1 (R2 = 0.71). Comparative regressions between the two different protocols for the expression of PD-L1 in both tumor and stromal compartments were excellent (R2 = 0.93, and 0.83). From a total of 81 patients, we excluded those whose samples were taken after therapy and those who received more than one immunotherapy resulting in 62 patients for outcome analysis using only the data from the Ultimapper assay. Assessment by OS showed significant relationships for high PD-L1 in tumor (21m vs 8m median OS, p=0.036) and high PD-L1 in CD68 positive macrophages (20m vs 10m median OS, p=0.012). Also, on the same slide, high CD8 was associated with better OS (22m vs. 11m median OS, p=0.011).

Conclusions

Expression levels of PD-L1 in tumor, but also in macrophages and the presence of CD8 in tumor show OS benefit in patients treated with immune checkpoint blockade in NSCLC. Future studies are required to evaluate this approach within a CLIA certified laboratory setting.

Acknowledgements

Sponsored research agreement between Yale and Ultivue (D. Rimm - PI)

Ethics Approval

Yale Human Investigation Committee protocol #9505008219.

P117 Host immune response in undifferentiated pleomorphic sarcoma – 10-year retrospective analysis

Joseph Sheridan, Andrew Horvai, Ross Okimoto, Rosanna Wustrack
University of California, San Francisco, San Francisco, CA, USA
Correspondence: Rosanna Wustrack (rosanna.wustrack@ucsf.edu)

Background

Undifferentiated pleomorphic sarcoma (UPS) is an aggressive soft-tissue sarcoma (STS) characterized by high rates of local and metastatic recurrence. Due to the paucity of therapeutic options, advanced disease remains lethal in a large majority of patients. An improved understanding of how the tumor microenvironment modulates UPS progression may enhance our ability to predict therapeutic responses and improve outcomes.

Methods

Thirty-six clinically annotated UPS patients collected over 10 years at a single institution with minimum five-year follow-up and available tumor specimens were included in this retrospective study. Using primary tumor specimens, we performed a targeted immunohistochemical analysis of the UPS microenvironment. We quantified expression of lymphocyte markers (CD8, CD20, CD68) and immune checkpoint protein (PD-L1) in all 36 UPS tumors using automated image analysis. The median percentage of positive cells for each subpopulation was used to define high expression vs. low expression. The Kaplan-Meier method was used to analyze OS and DFS; the association of specific TILs with OS and DFS was analyzed using the Log Rank Test.

Results

Factors that correlated with improved overall survival in our UPS cohort included localized disease (p=0.015), and use of intraoperative radiation therapy (IORT) or adjuvant radiation therapy (p=0.01). Our immunohistochemical analysis revealed the presence of TILs (CD8, CD20, CD68) and expression of immune checkpoint protein (PD-L1) in UPS tumors. Patients with a greater population of CD8+ TILs had a 5-year OS of 66% compared to those with lower levels of 28% (p=0.003, Figure 1). CD8+ T-cell expression in UPS tumors inversely correlated with local recurrence (p=0.04), suggesting CD8+ T-cell mediated immune surveillance. Interestingly, we also observed an increase in metastatic events in patients whose tumors harbored low CD8 expression compared to high CD8

expression (59% vs. 41%).

Conclusions

Through our quantitative immunohistochemical (IHC) analysis of immune cell subsets in UPS tumors, we identified improved survival in patients with increased infiltration of CD8+ T-Cells. Our study demonstrates that patients with low levels of CD8+ TILs are at increased risk of local (and potentially metastatic) recurrence. These findings underscore the importance of immune mediated tumor surveillance in UPS. Recent advancements in systemic immunotherapy further highlight the immunogenicity of UPS tumors and demonstrate the clinical impact of targeting the tumor microenvironment to improve outcomes for UPS patients.

Ethics Approval

Approval from the Institutional Review Board was obtained before beginning this study.
Fig. 1 (abstract P117).
Fig. 1 (abstract P117).

See text for description.

P118 Immunosuppression involving MDSC and IL-17 is associated with decreased levels of serum rapid turnover protein and shorter survival in patients with gastrointestinal cancer

Masahiko Shibata, MD, Kenji Gonda, MD, PhD, Takahiro Nakajima, Tatsuo Shimura, Koji Kono, Seiichi Takenoshita
Fukushima Medical University, Fukushima, Japan
Correspondence: Koji Kono (kojikono@fmu.ac.jp)

Background

Although a causal relationship of inflammation and immune function of cancer patients is more widely accepted today, the precise cell mechanisms mediating this relationship have not been elucidated. Accumulating evidence suggests that myeloid-derived suppressor cells (MDSC) may contribute to the negative regulation of immune responses during cancer and inflammation. IL-17 is a pro-inflammatory cytokine that is primarily secreted by T helper (Th)17 cells and has been reported to be associated with immunosuppressive conditions in patients with cancer. We have reported that systemic inflammation is closely associated with immunosuppression, malnutrition, and poor prognosis in several types of cancer. Cancer cachexia is a multifactorial condition characterized by hypoproteinemia, and systemic inflammation is a major cause of cachexia.

Methods

Rapid turnover proteins such as prealbumin (PA), retinol binding protein (RBP) and transferrin (TF), were measured and analyzed in correlation with prognosis in 288 patients with esophageal cancer (Study 1). In order to characterize the inflammation, the production of IL-17 was measured and MDSC (CD11b+CD14-CD33+) in peripheral blood were detected by flow cytometry in 106 patients including 43 with gastric and 63 with colorectal cancer. Blastogenic response of lymphocytes with PHA (SI: stimulation index) was used as a marker of cell-mediated immunity (Study 2).

Results

(Study 1) The patients were divided according to their average levels of PA, RBP and TF. The survival of the patients with higher PA, RBP or TF levels were significantly longer than of those with lower PA, RBP or TF levels. The levels of PA, RBP and TF were significantly inversely correlated with SI. (Study 2) The IL-17 production and MDSC levels were both increased along with disease advancement, and significantly correlated with each other. Further, they both were correlated with neutrophil/lymphocyte ratio (NLR), and CRP, inflammatory markers, and significantly inversely correlated with the levels of PA, RBP, and TF. The overall survival of the patients with higher IL-17 production or higher MDSC was shorter than of those with lower IL-17 production or lower MDSC.

Conclusions

IL-17-mediated inflammation may associate with immunosuppression involving MDSC, malnutrition and shorter survival and IL-17-targeting therapy may be effective for cancer treatment.

P119 Fab-selective proteolysis coupled with liquid chromatography-mass spectrometry for monitoring therapeutic antibodies in circulation

Takashi Shimada, PhD, Noriko Iwamoto, PhD
Shimadzu Scientific Instruments, Bothell, WA, USA
Correspondence: Takashi Shimada (tashimada@shimadzu.com)

Background

Individual cancer treatment according to drug efficacy indicators is very important matter. The genetic background, drug surrogate biomarker, and immune cell monitoring will be potentially incorporated into the future medicine as well as the expression level of target molecule or pharmacokinetic information. For the analysis of low-MW drugs, liquid chromatography-mass spectrometry (LC-MS) are often used. This is optimal approach for the structure identification and quantitation with sequential and comprehensive manner. However, for the high-MW biopharmaceuticals, pharmacokinetic parameters are often analyzed by ligand binding assays (LBA). The LBA may have some significant limitation. Therefore, analytical innovation should be indispensable independent of a variety of antibodies. For LC-MS, it should be developed the integrated and overall optimized approaches from complex biological samples because of the issue in separation and ionization suppression. We have focused on the two features: antibody structure-indicated analysis, and complementarity-determining region (CDR)-targeting quantitation. The ideal antibody analysis is possible using the selective quantitation of somatic mutated region CDRs.

Methods

IgGs were immobilized in resin pore (100 nm) via Fc, so that Fab was oriented to reaction solution. And proteolysis was performed by immobilized trypsin on the surface of FG nanoparticles (200 nm). Owing to these diameter difference, limited proteolysis on Fab was successful with maintaining the antibody specificity while minimizing the complexity or protease contamination. We named this method nano-surface and molecular-orientation limited (nSMOL) proteolysis.[1] And the generated signature peptides from each antibody were quantified using triple quadrupole LC-MS/multiple reaction monitoring (MRM) analysis.

Results

We have performed the assay validation development for Trastuzumab, Bevacizumab, Nivolumab, and more than 25 items according to the guideline on bioanalysis method validation.[2] The quantitation range in human plasma was obtained from about 0.2 to 300 μg/ml, which is enough to cover the clinical levels. And the verification of clinical samples has also been successfully with high-reproducibility.

Conclusions

The feature of nSMOL is the breakthrough solution for the accuracy, reproducibility, cost, general-purpose, QC and stability. We have some study design for the elucidation of overall mechanism of antibody drugs to discuss the relationship between drug level and efficacy or anti-drug antibody assay. And for practical use, we have another activity into the expanded clinical application, comparative assay of original/biosimilar, and therapeutic drug monitoring. Furthermore, the antibody profiles and distribution mechanism[3] in tumor tissues based on nSMOL may be expected to aid the acceleration of the biopharmaceuticals.

Acknowledgements

This study was partly collaboration results with Dr. Hamada A of National Cancer Center and Dr. Yonezawa A of Kyoto University.

References

1. Iwamoto N, Shimada T, Umino Y, Aoki C, Aoki Y, Sato TA, Hamada A, Nakagama H. Selective detection of complementarity-determining regions of monoclonal antibody by limiting protease access to the substrate: nano- surface and molecular-orientation limited proteolysis. Analyst. 2014;139(3):576-580.

2. Iwamoto N, Shimada T, Terakado H, Hamada A. Validated LC-MS/MS analysis of immune checkpoint inhibitor Nivolumab in human plasma using a Fab peptide-selective quantitation method: nano-surface and molecular- orientation limited (nSMOL) proteolysis. J Chromatogr B Analyt Technol Biomed Life Sci. 2016;1023-1024:9-16.

3. Terrell-Hall TB, Nounou MI, El-Amrawy F, Griffith JIG, Lockman PR. Trastuzumab distribution in an in-vivo and in-vitro model of brain metastases of breast cancer. Oncotarget. 2017 Jul 26;8(48):83734-83744.

P120 Validated next generation sequencing assay for the characterization of the T-cell repertoire from RNA

Jennifer Sims, Martin Buchkovich, PhD, Victor Weigman, Jason Powers, John Pufky, Jennifer Mason, PhD, Patrick Hurban, PhD
Q2 Solutions, Morrisville, NC, USA
Correspondence: Patrick Hurban (patrick.hurban@q2labsolutions.com)

Background

Activation of T-cells during cell-mediated immunity is initiated by the stimulation of the T- cell receptor (TCR) by major histocompatibility complex-antigen complexes. While the entire TCR chain is diverse, most of the diversity is concentrated in a hypervariable complementarity-determining region 3 (CDR3) loop, the center of the antigen- binding site for the TCR. The frequency of a specific CDR3 sequence within the T-cell repertoire is a surrogate for the abundance of its corresponding T-cell clone. Deep sequencing of the TCR CDR3 region assists with resolving T- cell diversity, and, in oncology, detects specific clones or changes in clonality associated with anti-tumor immune responses. Here, we have validated an RNA based TCRβ/γ next-generation sequencing assay for use with whole blood, peripheral blood mononuclear cells (PBMCs) and formalin-fixed paraffin-embedded (FFPE) tissues.

Methods

TCR analysis was performed using RNA or total nucleic acids derived from whole blood, PBMCs or tumor FFPE specimens. The TCRβ/γ sequencing assay entailed gene specific cDNA synthesis, preparation of sequencing libraries, TCR gene specific amplification, sample barcoding and sequencing. A 2x150 bp sequencing was performed and ≥2 million paired-end reads for each sample were obtained. Data analysis was performed using Archer Analysis software.

Results

Rarefaction analysis was used to characterize the analytical sensitivity of the assay at varying sequence depths and input amounts, using RNA from whole blood and PBMCs from healthy donors. The total number of unique TCRβ RNA fragments observed using 20-1,200 ng of RNA ranged from 2,890 to 94,969 for whole blood, and 8,433 to 174,350 for PBMCs. The assay detected 1 TCRβ RNA fragment in the background of up to 174,350 TCRβ molecules, yielding a sensitivity of 5.7 *10-6. 100% accuracy was observed for both TCRβ and TCRγ using pre- characterized T-cell derived cell lines. FFPE tumor specimens from patients with lung and breast adenocarcinoma were used to determine TCR repertoire and its diversity. Each tumor was classified as having either high (≥15%), medium (7-14%), or low (≤7%) TILs in FFPE sections using H&E analysis. Using 400ng RNA, the number of unique TCR fragments detected in FFPE samples ranged between 53-1,632. FFPE samples were either polyclonal or highly clonal, however TCR clonality did not always correlate with TIL count.

Conclusions

We have validated an RNA-based NGS assay for the analysis of TCRβ/γ in whole blood, PBMC and FFPE specimens. The assay can support both TCR repertoire analysis and minimal residual disease monitoring in T-cell malignancies.

P121 Interaction of immune checkpoints in tumor-stromal microenvironment of primary and chemoreduced retinoblastoma

Lata Singh, PhD1, Mithalesh Singh2, Seema Kashyap2, Seema Sen, MD2, Moshahid A. Rizvi, PhD1
1Jamia Millia Islamia, New Delhi, India; 2All India Institute of Medical Sciences, New Delhi, India
Correspondence: Moshahid A. Rizvi (mrizvi@jmi.ac.in)

Background

Interactions between malignant and non-malignant cells create the tumor microenvironment (TME). The non-malignant cells of the TME have a dynamic and tumor-promoting function at all stages of carcinogenesis. Cytotoxic T lymphocyte-associated antigen 4 (CTLA4), programmed death-1 (PD-1) and programmed death-ligand 1 (PD-L1) are key components of the immune checkpoint pathway. They play a crucial role in the regulation of T-cell activation and their expression in TME constitutes a predictive biomarker in cancers. It has recently been shown that chemotherapeutic agents could modify tumor microenvironment. Therefore, we investigated the expression of PD-1, PD-L1 and CTLA-4 in primary and chemoreduced retinoblastoma to define their significance in the tumor microenvironment with patient prognosis.

Methods

Expression of immune markers (PD-1, PD-L1 and CTLA-4 protein) was evaluated in 75 prospective cases of primary (Group I) and 25 cases of chemoreduced (Group II) enucleated retinoblastoma specimens by immunohistochemistry. mRNA expression of genes of interest were investigated by quantitative real time PCR (qPCR) and results were finally correlated with clinicopathological parameters and patient outcome by statistical analysis.

Results

Differential expression pattern of PD-1, PD-L1 and CTLA-4 proteins was found in both group I (primary retinoblastoma) and group II (chemoreduced retinoblastoma) cases. Immunohistochemistry showed cytoplasmic/membranous staining of these immune markers using their respective antibodies. Increased expression of PD-1, PD-L1 and CTLA-4 were found in stromal/immune cells of group II as compared to Group I. Expression of these immune markers showed significant correlation with poor tumor differentiation, tumor invasion and patient outcome (p<0.05).

Conclusions

This is the first of its kind study investigating the role of immune markers in primary retinoblastoma and their alteration in expression after chemotherapy. Tumor microenvironment of retinoblastoma showed expression of PD- L1 in primary patients and increased expression in PD-L1, CTLA-4 and PD-1 after chemotherapy. This paves the way for development of new strategies for treatment of chemoreduced retinoblastoma.

Acknowledgements

Dr. Lata Singh was supported by Department for Science and Technology (DST), Govt. of India, funding agency for providing National Post-Doctoral fellowship (N-PDF) and conducting this research.

Ethics Approval

This study was approved by Institute’s Ethical Committee, AIIMS (Ref. No. IEC-424/RP-6/2016)

Consent

Written consent was obtained from all the patient's guardian

P122 Validated next generation sequencing RNA-based assay of the IGHV mutation status and repertoire diversity

Jeran Stratford, Patrick Hurban, PhD, Victor Weigman
Q2 Solutions, Morrisville, NC, USA
Correspondence: Jeran Stratford (jeran.stratford@q2labsolutions.com)

Background

High-throughput sequencing of immune repertoire is increasingly used for clinical diagnosis, monitoring of residual disease, and development of cancer immunotherapies. The load of somatic hypermutations in the rearranged immunoglobulin heavy-chain variable region gene (IGHV) is a powerful prognostic biomarkers in CLL. CLL arising from unmutated IGHV cells (≥98% identity to closest germline rearrangement) are more aggressive and associate with poor prognosis, while CLL with mutated IGHV (<98% identity) have favorable outcomes and a higher rate of durable remissions after treatment with chemoimmunotherapy combinations. Historically, IGHV mutation status has been performed using Sanger sequencing looking at a single dominant clone. However, next- generation sequencing can reveal the sample's IGHV diversity. We have developed an NGS based IGHV assay with comparable results to Sanger sequencing. The assay is certified by the European Research Initiative on CLL (ERIC) [1].

Methods

Purified RNA is used as template to produce PCR amplicons of the VH1, VH3, VH4 and combined VH2/VH5/VH6/VH3-21 subfamilies. Productive PCR reactions are analyzed, sequencing libraries are prepared and sequenced using 2x300 paired-end Illumina MiSeq method. Paired reads within each sequenced amplicons are synthetically joined and aligned using all V-region sequences of the ImMuno GeneTics reference. Following filtering, the proportion of remaining reads corresponding to each IGHV gene is determined. The NCBI’s IgBlast software is used to assign a mutation status (<98% is mutated and ≥98% is unmutated) for each sequenced amplicons with ≥50% of reads aligned to a single gene. Sequenced amplicons with <50% of reads assigned to a single gene are classified “polyclonal”. A sample level mutation status is then determined by reconciling the mutation status for each sequenced amplicon. If all sequenced amplicons for a sample are polyclonal, the sample is classified as “polyclonal”. If the sample has any amplicons labeled as mutated and an absence of any unmutated labeled amplicons then the sample will be classified as mutated. If the sample has any amplicons with an unmutated status, the sample is classified as unmutated.

Results

For each sample we report the mutation status, percent identity to germline IGHV clonotypes, V/D/J assignment, and re-arrangement productivity. We also identify stereotype subset #2 and indicate its prognostic value.

Conclusions

Our assay has met ERIC guidelines [2] and was awarded IGHV assay certification. With its ability to characterize IGVH diversity, the utility of a certified NGS IGHV assay is critical in determining care for CLL patients now and providing a springboard for future biomarker development.

References

1. http://www.ericll.org/ighv-gene-mutational-analysis-certification/

2. Rosenquist R, Ghia P, Hadzidimitriou A, Sutton L-A, Agathangelidis A, Baliakas P, Darzentas N, Giudicelli V, Lefranc M-P, Langerak A, Belessi C, Davi F, Stamatopoulos K. Immunoglobulin gene sequence analysis in chronic lymphocytic leukemia: updated ERIC recommendations. Leukemia. 2017; 31:1477-1481.

P123 Peripheral blood lymphocyte responses in patients with renal cell carcinoma treated with high-dose interleukin-2

Rupal Bhatt, MD, PhD1, Lei Sun, Ph.D2, William Slichenmyer, MD3, Sean Rossi2, Juan Alvarez, PhD4, Wenxin Xu, MD1, Heather Losey, PhD2
1Beth Israel Deaconess Medical Center, Boston, MA; 2Alkermes, Inc., Waltham, MA, USA; 3Alacrita Consulting, Waltham, MA, USA; 4Merck & Co, Boston, MA, USA
Correspondence: Lei Sun (Lei.Sun@alkermes.com)

Background

High-dose interleukin-2 (HD IL-2) activates the expansion of immunosuppressive regulatory T cells (Tregs), cytotoxic CD8+ T cells and natural killer (NK) cells. Previous data show that immunosuppressive ICOS+ Tregs are significantly expanded after treatment with HD IL-2 [1], but no data are readily available that specifically quantify and compare the levels of expansion of cytotoxic effectors such as CD8+ T cells and NK cells relative to Tregs. This study was conducted with the primary goal to assess the pharmacodynamic effects of HD IL-2 on numbers of circulating CD8+ T cells, NK cells, and Tregs.

Methods

Whole blood samples were collected prior to the first dose and after the last dose of treatment cycles 1 and 2 from a cohort of renal cell carcinoma patients receiving treatment with HD IL-2. CD8+ T cells, NK cells, and Tregs were quantified by flow cytometry. Safety and antitumor activity were monitored throughout the study period. Response was assessed according to RECIST, and best response was recorded.

Results

Ten patients with renal cell carcinoma were enrolled: median age 55 (range 39-62), male/female 6/4, ECOG PS of 0=9/1=1, and median number of prior therapies 2 (range 1-3). All treatment emergent adverse events seen were consistent with the known adverse event profile of HD IL-2 [2]. Capillary leak syndrome was reported in 5 patients. Five of the 10 patients achieved best response of partial response, and 1 patient had mixed response. Administration of HD IL-2 resulted in robust expansion of circulating Tregs with a mean maximum expansion of ~4-fold as compared to ~2-fold expansion of circulating CD8+ T cells and NK cells.

Conclusions

The safety profile and clinical response observed in this small cohort of patients were similar to previous published data [2]. A more robust expansion of Tregs over CD8+ T cells and NK cells was observed in patients treated with HD IL-2, consistent with the known biological activities of IL-2.

Acknowledgements

This study was funded by Alkermes, Inc. The authors gratefully acknowledge the patients and their families who participated in this study.

References

1. Sim GC, et al, IL-2 therapy promotes suppressive ICOS+ Treg expansion in melanoma patients. J. Clin. Invest. 2014;124(1): 99–110.

2. Marabondo S and Kaufman HL. High-dose interleukin-2 (IL-2) for the treatment of melanoma: safety considerations and future direction. Expert Opin. Drug Saf. 2017;16(12):1347-1357

Ethics Approval

This study was approved by the Beth Israel Deaconess Medical Center IRB, protocol #06-105.

P124 Quantitative multichannel immunofluorescence imaging to assess the immune composition of the T cell- inflamed tumor microenvironment in bladder cancer

Randy Sweis, MD, Ken Hatogai, Danny Kim, Yuanyuan Zha, PhD, Alexander Pearson, MD, PhD, Gary Steinberg, MD, Thomas Gajewski, MD, PhD
University of Chicago, Chicago, IL, USA
Correspondence: Randy Sweis (rsweis@uchicago.edu)

Background

A T cell inflamed tumor microenvironment is linked to improved prognosis and response to immunotherapy in bladder cancer. This immune phenotype can be measured by the presence of tumor-infiltrating T-cells with various gene expression signatures reflective of the immune response. Tumor-infiltrating BATF3+ dendritic cells have been shown in murine models to be critical for both priming an immune response and for recruitment of effector CD8+ T cells to the tumor microenvironment. In human bladder cancer specimens, the presence and distribution of tumor- infiltrating BATF3-dendritic cells has not been previously evaluated.

Methods

We performed quantitative multiplex immunofluorescence imaging on 64 bladder cancer specimens from patients to investigate the population of immune-infiltrating cells present in the tumor such as BATF3 cells and CD8+ T cells (Figure 1). We used a previously described immune gene expression signature to determine the presence of absence of the T cell-inflamed tumor microenvironment based on RNA sequencing for a subset of these samples. Immune cells proportions were calculated relative to the tumor cell counts in 5 randomly selected regions of interest for each specimen.

Results

The proportion of BATF3+ cells per 1000 tumor cells ranged from 0 to 6.32 with an average of 0.68. There were no BATF3+ cells in 23 out of 64 specimens (36%). The proportion of CD8+ cells per 1000 tumor cells ranged from 0 to 135 with an average of 10.4. There were no CD8+ cells in 8 out of 64 specimens (13%). In a subset of 18 samples for which RNA sequencing data were available, we found a 39-fold higher proportion of BATF3+ cells in tumors that were T cell-inflamed by gene expression profiling compared with those that were non-T cell-inflamed (P=0.02). T-cell inflamed tumors also had a 2.4-fold higher CD8+ cell proportion compared with non-T-cell inflamed tumors, but this difference did not reach significance (P=0.30).

Conclusions

The presence of tumor-infiltrating BATF3+ dendritic cells correlates strongly with the presence of a T cell-inflamed tumor microenvironment by gene expression profiling. Further analyses are ongoing to assess the impact of spatial relationships of immune cells and their association with immunotherapy response and outcomes.

Ethics Approval

Specimens used in this study were obtained through a protocol approved by the University of Chicago Institutional Review Board (IRB 15550B)
Fig. 1 (abstract P124).
Fig. 1 (abstract P124).

Multichannel immunofluorescence for CD8 and Batf-3

P125 DNA methylation biomarkers for noninvassive detection of hepatocellular carcinoma

David Taggart, PhD1, Dhruvajyoti Roy, PhD1, Gen Li1, Dan Liu1, Lianghong Zheng1, Kang Zhang2
1The Laboratory for Advanced Medicine, Inc., West Lafyette, IN, USA; 2University of California San Diego, San Diego, CA, USA
Correspondence: David Taggart (david.taggart@lamoncogroup.com)

Background

The epigenetic inactivation of tumor suppressor genes by promoter hypermethylation is an important aspect of tumorigenesis. Indeed, aberrant methylation of CpG sites within genomic DNA isolated from cancer cells has been shown to correlate with clinically relevant information and has the potential to be used for cancer diagnosis and identification of the cancer tissue of origin. Malignant cells shed genomic DNA into circulation through both cell death and active release by viable cells. Therefore, investigating the methylation of cell-free DNA allows for the noninvasive detection and early diagnosis of cancers, such as hepatocellular carcinoma. Here, we identified and validated hepatocellular carcinoma-specific methylation markers for diagnosis of the disease with high sensitivity and specificity.

Methods

Banked samples were obtained for 130 subjects, including: 60 subjects diagnosed with hepatocellular carcinoma (Stage I to IV), 30 subjects without liver disease, 10 subjects diagnosed with benign liver disease and 30 subjects diagnosed with breast, colorectal or lung cancer. Samples were provided to the laboratory blinded for analysis. Cell- free DNA was then extracted from the samples, bisulfite converted, and DNA methylation was quantified by using the IvyGene® Platform. After data collection and analysis of all samples was complete, the samples were unblinded to calculate test performance.

Results

A total of 57 of the 60 samples drawn from subjects with hepatocellular carcinoma were correctly identified for an overall calculated sensitivity of 95%, with little difference between the sensitivity of detecting Stage I to Stage IV hepatocellular carcinoma (range 89% to 100%). Additionally, 29 of 30 samples drawn from subjects without liver disease and 9 of 10 samples drawn from subjects diagnosed with benign liver disease were correctly identified as non-cancer for a combined calculated specificity of 97.5%. Of the samples drawn from subjects with cancer other than liver cancer, 90% of breast cancer samples, 80% of colorectal cancer samples and 90% of lung cancer samples were correctly identified as not liver cancer, for a total calculated analytical specificity of 87%.

Conclusions

These data demonstrate the high diagnostic potential of cfDNA methylation markers in the blood for the detection of hepatocellular carcinoma. Indeed, quantification of cfDNA methylation may be a more sensitive and specific method for the detection of hepatocellular carcinoma than ultrasound, which is the current recommended imaging method for surveillance of high-risk populations.

Ethics Approval

This project was approved by the Institutional Review Boards (IRBs) of Sun Yat-sen University Cancer Center, Xijing Hospital, and West China Hospital.

P126 Peripheral immune monitoring identifies biomarkers of response and toxicity after neoadjuvant combination immunotherapy with ipilimumab (ipi) and high dose IFNα2b (HDI) in patients with melanoma

Ahmad Tarhini, MD, PhD1, Ghanashyam Sarikonda2, Arjun Khunger, MD3, Jack El-Sawada, MD4, Christian Laing2, Christine Vaupel, PhD2, Naveen Dakappagari2
1Cleveland Clinic and Case Comprehensive Cancer Center, Cleveland, OH, USA; 2Navigate BioPharma Services, Inc., A Novartis Subsidiary, Carlsbad, CA, USA; 3Cleveland Clinic, Cleveland, OH, USA; 4University of Pittsburgh Medical Center, Pittsburgh, PA, USA
Correspondence: Ahmad Tarhini (tarhina1@ccf.org)

Background

Neoadjuvant ipi-HDI given over 6 weeks for locally/regionally advanced melanoma showed preoperative radiologic response of 36% and pathologic complete response (pCR) of 32% [1]. Mechanistic studies may illuminate the underlying mechanisms of immune susceptibility and resistance and identify biomarkers of response and toxicity.

Methods

Peripheral blood mononuclear cells (PBMC) from treated patients (N=28) on this trial were tested at baseline (before initiating ipi-HDI), 6-weeks, 3-months and 12-months following neoadjuvant ipi-HDI). High complexity (14-color) flow cytometry designed to detect key immunological biomarkers such as MDSCs, regulatory T cells, PD-1 and TIM3 expression on T-cells, and differentiation of T-cells into Th1, Th2 or Th17 phenotype were used to evaluate the effects of immunological biomarkers on safety and efficacy. Statistical significance was determined using R-package employing Kruskal’s test.

Results

Higher levels of Th1 cells (defined as CD45RA- CCR6- CXCR3+ CCR4-) correlated with preoperative radiological response (p=0.0070) while higher Th2 cells (defined as CD45RA- CCR6- CXCR3- CCR4+) was associated with progressive disease (0.0092). In evaluating pathologic response, a higher multimarker score consisting of Th1 cells and CD8+ central memory T-cells was associated with pCR (p=0.0406), in contrast, higher TIM3 expression on T- cells correlated with gross viable tumor (p=0.0472). Higher levels of phenotypically naive and effector memory CD8+ T-cells (p=0.0140) or lower levels of Th2 cells were associated with lower toxicity as well (p=0.0243). Finally, a multimarker score consisting of higher CD19+ and CD8+ cells was associated with lower toxicity (p=0.0013) and vice versa

Conclusions

Peripheral immune monitoring may identify predictive biomarkers of response and toxicity following combination neoadjuvant immunotherapy. Validation and long term immune monitoring studies are ongoing and will be presented.

Trial Registration

https://clinicaltrials.gov/ct2/show/NCT01608594NCT01608594

References

1. Tarhini AA, Rahman Z, Lin Y, Vallabhaneni P, Tawbi HA-A, Gnan A, et al. Neoadjuvant combination immunotherapy with ipilimumab (3 mg/kg or 10mg/kg) and high dose IFN-a2b in locally/regionally advanced melanoma. Journal of Clinical Oncology. 2016;34(15_suppl):9585-.

Ethics Approval

The study was initiated after approval from the institutional review board (IRB) and was conducted in accordance with the Declaration of Helsinki

Consent

A University of Pittsburgh IRB approved written informed consent (IRB# PRO12020161) was obtained from all patients.

P127 Characterization of changes in tumor immune microenvironment after treatment with neoadjuvant combination immunotherapy with ipilimumab (ipi) and high dose IFNα2b (HDI) in patients with melanoma

Arjun Khunger, MD1, Jennifer Bordeaux, PhD2, Ju Young Kim2, Christine Vaupel, PhD2, Naveen Dakappagari2, Ahmad A. Tarhini, MD, PhD3
1Cleveland Clinic, Cleveland, OH, USA; 2Navigate BioPharma Services, Inc., A Nov, Carlsbad, CA, USA; 3Cleveland Clinic and Case Comprehensive Cancer Center, Cleveland, OH, USA
Correspondence: Ahmad A. Tarhini (tarhina1@ccf.org)

Background

Neoadjuvant ipi-HDI given over 6 weeks for locally/regionally advanced melanoma showed preoperative radiologic response of 36% and pathologic complete response (pCR) of 32% (Tarhini, et al. ASCO 2016). Mechanistic studies may reflect the underlying dynamics of tumor immune microenvironment (TME) in response to neoadjuvant treatment and offer important insights into immune mechanisms of response and resistance.

Methods

Tumor biopsy specimens of 28 patients with locally/regionally advanced melanoma, who were treated with neoadjuvant ipi-HDI were obtained at baseline and post-treatment assessment. Primary (archival) tumor samples collected at the time of diagnosis were also available. Multiplexed fluorescence immunohistochemistry combined with unique AQUA (Automated Quantitative Analysis) algorithms specifically designed to classify regulatory T cells (Tregs), Myeloid derived suppressor cells (MDSCs), CD3+ T cells, PD-1 expression, PD-L1 expression and IDO1/HLA-DR co-expression was used to assess tumor immune modulation in response to neoadjuvant immunotherapy. Statistical significance was determined using the paired t-test.

Results

Our analysis revealed that there was a significant decrease in the proportion of Tregs between pre-treatment primary and post-treatment biopsy samples (p=0.007). Also, there was a significant increase in the percentage of CD3+ T cells between pre-treatment primary and post-treatment biopsy samples (p= 0.005), supporting the hypothesis of increased generation of effector T-cells post-treatment. In correlation with pathologic response, there was a consistent trend towards a decrease in IDO1/HLA-DR co-expressing cells post-treatment as compared to pre- treatment primary (p= 0.067) and baseline (p=0.08) biopsy samples. No significant changes in expression of PD-1 or PD-L1 were seen between pre-treatment baseline and primary samples and post-treatment samples. In addition, no significant changes were detected in the immune profiles of pre-treatment baseline versus primary tumor biopsy.

Conclusions

A significant increase in CD3+ T cells and a significant decrease in Tregs was observed following treatment with neoadjuvant immunotherapy in our study. Consistent trends towards a decrease in IDO1/HLA-DR co-expression in responding patients support a role for IDO1 in immune resistance. Comparable TME profiles between preprimary and pre-treatment baseline tumor biopsies may be useful in guiding future studies.

Trial Registration

https://clinicaltrials.gov/ct2/show/NCT01608594NCT01608594

References

1. Tarhini AA, Rahman Z, Lin Y, Vallabhaneni P, Tawbi HA-A, Gnan A, et al. Neoadjuvant combination immunotherapy with ipilimumab (3 mg/kg or 10mg/kg) and high dose IFN-a2b in locally/regionally advanced melanoma. Journal of Clinical Oncology. 2016;34(15_suppl):9585-.

Ethics Approval

The study was initiated after approval from the institutional review board (IRB) and was conducted in accordance with the Declaration of Helsinki.

Consent

A University of Pittsburgh IRB approved written informed consent (IRB# PRO12020161) was obtained from all patients.

P128 Multiplexed immunofluorescent assay development for study of the PD-1/PD-L1 checkpoint in the tumor immune microenvironment (TIME)

Sneha Berry, MS, Sneha Berry, MS, Sneha Berry, MS, Nicolas Giraldo, MD PhD, Peter Nguyen, MS, Benjamin Green, BS, Haiying Xu, Aleksandra Ogurtsova, Abha Soni, DO, Farah Succaria, MD, Daphne Wang, MS, Charles Roberts, Julie Stein, MD, Elizabeth Engle, MSc, Drew Pardoll, MD, PhD, Robert Anders, MD, PhD, Tricia Cottrell, MD, PhD
Johns Hopkins University School of Medicine, Baltimore, MD, USA
Correspondence: Sneha Berry (jtaube1@jhmi.edu)

Background

Multispectral immunofluorescent (mIF) staining of formalin-fixed paraffin-embedded (FFPE) tissue allows spatially-resolved quantitative analysis of cell position and protein expression. The design and validation of mIF panels is a challenge. Our goal was to develop a 7-plex assay for characterizing PD-1 and PD-L1 expression, with high sensitivity for multiple markers and minimal bleed-through between fluorescent channels, while avoiding steric hindrance among markers occupying the same cellular compartment.

Methods

Single IF slides were stained for PD-1, PD-L1, CD8, FoxP3, CD163, and a tumor marker (e.g. Sox10/S100 for melanoma) using primary antibodies at manufacturer’s recommended concentrations and visualized with an Opal kit (PerkinElmer). Positive signal was compared to chromogenic IHC (n=3 tonsil specimens). In some instances, the kit’s HRP-polymer was substituted for one that provided greater amplification. Primary antibody titrations were performed, and the concentration with comparable signal to chromogenic IHC that showed the highest IF signal to noise ratio was selected. Using the selected primary antibody concentration, TSA dilution series were performed on n=5 tumor specimens to minimize bleed-through. Finally, the optimized single IF stains were combined into multiplex format, which was again validated to ensure no positivity loss. Images were scanned with the Vectra 3.0 and processed using inForm (Ver 2.3).

Results

The percent positive pixels for CD163, CD8, and tumor marker expression by IF were comparable to chromogenic IHC with manufacturer’s recommended protocols (p>0.05). However, PD-1, PD-L1, and FoxP3 showed ~50% loss of signal (p<0.05), which was recovered by replacing the Opal kit’s secondary HRP polymer with PowerVision (Leica). Unbalanced fluorescence intensities between 540 to 570 Opal dyes resulted in significant bleed-thorough and led to false positive pixels. This error was minimized >2 fold (2.5% to 1.1%) by concentrating the 570 dye and ensuring that this dye pair was used to study markers in different cellular compartments (nuclear FoxP3 vs. membrane CD8), so any residual bleed-through could be discounted during image analysis. Using the optimized panel, we are able to reliably identify cell types contributing PD-L1 and PD-1 to the TIME, and even resolve populations of PD-1high vs. PD-1low lymphocytes.

Conclusions

We demonstrate successful optimization of a 7-color multiplex panel characterizing the PD-1/PD-L1 axis to provide high quality data sets for whole slide or regional analysis of the TIME. With the use of multiparametric assays such as this, we hope to guide improved approaches to patient selection and potentially identify additional tumor types likely to respond to anti-PD-(L)1 immunotherapy.

Ethics Approval

The study was approved by Johns Hopkins University Institutional Review Board.

P129 Simultaneous single cell analysis of multiple analytes resolves T cell populations at high resolution

Sarah Taylor, PhD1, Katherine Pfeiffer1, Michael Stubbington, PhD1, Josephine Lee1, Jerald Sapida1, Liselotte Brix, phd2, Kivin Jacobsen, PhD2, Bertrand Yeung3, Xinfang Zhao3, Tarjei Mikkelsen1, Deanna Church, PhD1
110x Genomics, Pleasanton, CA, US; 2Immudex, Copenhagen, Denmark; 3BioLegend, San Diego, CA, USA
Correspondence: Sarah Taylor (sarah.taylor@10xgenomics.com)

Background

Characterization of lymphocyte types and understanding their antigen binding specificities are key to the development of effective therapeutics. Recent technological advancements have enabled the integration of simultaneous cell-surface protein, transcriptome, immune repertoire and antigen specificity measurements at single cell resolution, providing comprehensive, high-throughput characterization of immune cells.

Methods

Using the 10x Genomics Single Cell Immune Profiling Solution with Feature Barcoding technology in conjunction with TotalSeq™-C oligo-conjugated antibodies (BioLegend) and DNA barcoded MHC Dextramer® reagents (Immudex), we performed multi-omic characterization of PBMCs from cytomegalovirus (CMV) seropositive and seronegative donors. Next generation sequencing libraries were made following the 10x Genomics workflow, where gene expression and immune repertoire libraries are generated alongside libraries from DNA barcodes conjugated to antibodies or MHC Dextramer reagents, allowing quantification of cell surface proteins and identification of TCR specificities. Analysis was performed using the latest version of Cell Ranger (v3.0). The TCR-dist algorithm was used to identify clusters of related TCR sequences and enriched CDR3 motifs.

Results

Combining single cell TCR assembly with barcoding of MHC Dextramer reagents allowed the identification of full length, paired alpha and beta TCR sequences with specificity for known CMV antigens. Cells were also labelled with barcoded antibody reagents to allow cell type characterization based on surface protein expression to augment the transcriptomic information provided by the gene expression portion of the assay. The combination of these single cell assays allowed the identification of CD8+ cell populations with specificity for CMV/MHC. Multiple TCRs that bound CMV/MHC were observed and we identified enriched amino acid motifs within the TCR sequences. We analyzed the distribution of CMV-binding TCRs with similar sequences shared across individual donors and compared the observed CMV-binding TCR sequences with those previously reported in TCR-antigen databases.

Conclusions

The analytical approaches outlined here provide a systematic and scalable method for deciphering TCR-peptide MHC specificity, with clear implications in understanding the complexity of the tumor microenvironment. The multi-omic combination of gene expression, paired adaptive immune receptor repertoire, antibody-based detection of cell surface proteins and Dextramer-based analysis of antigen binding specificity for the same single lymphocytes allows the comprehensive characterization of immune cell populations at unprecedented resolution and throughput. Identifying discrete cellular phenotypes that underlie immune receptor specificity and antigen binding capabilities is critical for developing a better understanding of the adaptive immune response to cancer; leveraging this understanding will be key in the development of successful cellular and transgenic immunotherapies.

P130 A measurable immune response in the tumors and lymph nodes of patients with lymph node positive and lymph node negative breast cancer

Archana Thakur, PhD1, Dana Schalk1, Tayson Lin2, Johnson Ung1, Griffin Calme, BS2, Johnson Ung1, Lawrence G. Lum, MD, DSc1, Lydia Choi3
1University of Virginia Cancer Center, Charlottesville, VA, USA; 2Wayne State University, Detroit, MI, USA; 3Karmanos Cancer Institute, Detroit, MI, USA
Correspondence: Lawrence G. Lum (lgl4f@hscmail.mcc.virginia.edu)

Background

The functional significance of various immune cell subsets may provide important information for maximizing the prognostic value of immunoscoring. In early stage breast cancer patients with no detectable lymph node invasion relapse rate is about 20-30% while one-third of lymph node positive breast cancer patients remain free of distant metastasis. The prognostic gap in this group of patients remains and the vast majority of the patients in both groups had no prognostic marker to assist in driving clinical management based on risk factors. We hypothesize that immune cell population and its T cell activation status (Th1 type) in the resected breast tumors and sentinel lymph nodes may serve as immunological biomarkers of tumor aggressiveness and offer useful prognostic information to facilitate specific therapy related clinical decision as well as response to therapeutic options. Since the nodal immune environment is influenced in part by tumor-derived factors, we tested a small cohort for cytoplasmic expression of IFN-gamma and IL-10 to explore the association between Th1 and Th2 status of the TILs along with the triple staining for the CD3+ T cells, CD20+ B cells and CD68+ macrophages in tumors and lymph nodes biopsies with clinical outcomes of breast cancer patients.

Methods

Biopsies from four groups of patients were examined for T cell activation status, 1) no recurrence until present and node negative; 2) no recurrence until present and node positive; 3) recurrence in last 5 years and node negative; 4) recurrence in last 5 years and node positive. Tissue slides from sentinel node biopsies and tumor resections from thirty women who underwent sentinel node biopsy with invasive breast cancer resection were dual stained for CD3/IL-10, CD3/IFN-gamma or triple stained for CD3/CD20/CD68. Slides were imaged using a Digital Scanner, the density and intensity of each cell type and cytokine was recorded as the number of positive cells per unit tissue surface area.

Results

Our data show that patients who had no evidence of disease after 5 years had weak to moderate expression of IFN-gamma by T cells while patients who progressed rapidly had no expression of IFN-gamma by T cells in the tumors or in the lymph nodes.

Conclusions

Understanding of the immune cell environment within the tumors and lymph nodes may have implications for improving clinical outcome of cancer patients.

Ethics Approval

The study was approved by Wayne State University's IRB, approval number 634183

P131 High-plex predictive marker discovery for melanoma immunotherapy treated patients using NanoString® Digital Spatial Profiling

Maria Toki, MD, MSc1, Pok Fai Wong, MD, MPhil1, James Smithy, MD, MHS2, Harriet Kluger, MD1, Chris Merritt, PhD3, Giang Ong, MS3, Sarah Warren, PhD3, Joseph Beechem, PhD3, David Rimm, MD, PhD1
1Yale University School of Medicine, New Haven, CT, USA; 2Brigham and Women's Hospital Department, Boston, MA, USA; 3NanoString Technologies, Seattle, WA, USA
Correspondence: David Rimm (david.rimm@yale.edu)

Background

NanoString Digital Spatial Profiling (DSP)*, a technology previously validated to quantitative immunofluorescence, offers the capacity of highly multiplexed immune marker quantitative measurements with spatial resolution within specific regions of interest (ROI) on formalin-fixed, paraffin-embedded (FFPE) tissue. Here, we used NanoString DSP to explore the predictive value of a 44plex panel of immune markers measured in multiple compartments in a melanoma immunotherapy (ITx) treated cohort.

Methods

NanoString DSP technology uses a cocktail of primary antibodies conjugated to indexing DNA oligos. ROI on the tissue are selected with fluorescently labeled antibodies, and oligos from that region are UV cleaved, sipped up, and quantified on the nCounter® platform. Here, we used a tissue microarray (TMA) cohort which includes 60 Itx treated melanoma patients to identify potential predictive markers of survival and response using a 44plex panel of immune markers collected sequentially and quantified in three different compartments: in macrophage, leukocyte and the melanocyte ROI, defined by CD68+, CD45+ and S100+HMB45+ respectively. Each patient was represented by two cores and the counts for each marker were averaged.

Results

Biomarker counts were highly concordant across unique TMA cores from the same patient tumor. High concordance between DSP and quantitative fluorescence was also seen as a validation for the DSP method. Out of the 44plex panel measured in three different ROIs, 13 and 16 immune markers were found to be associated with prolonged progression free survival (PFS) and overall survival (OS). Perhaps the most striking finding was that, using a CD68 specific compartment marker measurement, we found that PD-L1 expression in macrophages and not in tumor was a predictive marker for PFS, OS and response. Other notable compartment-specific biomarkers found to be associated with outcome included beta 2-microglobulin, HLADR, CD8 and IDO1.

Conclusions

NanoString DSP capacity for highly multiplexed immune marker measurements on selected compartments allowed the discovery of multiple predictive markers in a single TMA section of a melanoma ITx cohort. This tool represents the highest multiplexed spatially informed method for discovery without the liabilities of cycling methods. These findings can be used for the construction of predictive signatures involving multiple compartments reflecting the complexity of the tumor microenvironment. Future studies on the DSP platform are underway with multiplexing capacity to nearly 1000 biomarkers on the same tissue section.*FOR RESEARCH USE ONLY. Not for use in diagnostic procedures.

Ethics Approval

The study was approved by the Yale Human Investigation Committee protocol #9505008219 and conducted in accordance with the Declaration of Helsinki.

P132 RNA sequencing of rare antigen-specific T cells and tissue micro-regions using the RareCyte platform

Lance U'Ren, DVM, PhD, Rebecca Podyminogin, Nolan Ericson, Eric Kaldjian, MD, Tad George, PhD
RareCyte Inc., Seattle, WA, USA
Correspondence: Lance U'Ren (luren@rarecyte.com)

Background

The immune system provides antigen-specific protection against pathogens as well as malignancies, both of which evolve strategies to evade immune surveillance and containment. Effective immune response often depends on activation of rare immune cell sub-types, whose function are influenced by the tissue micro-environment, the pathogen or cancer, and other factors. The RareCyte platform provides integrated multi-parameter imaging and retrieval capabilities that allow phenotypic identification and isolation of rare cells and microscopic regions of interest for sequence and transcript level analyses in order to study the complexity of host defense.

Methods

We utilized RareCyte’s CyteFinder platform to identify rare antigen-specific T cells by using tetramers against influenza-specific T cell receptors (TCR) in both unstimulated and influenza peptide stimulated samples. Additionally, we isolated 40 μm regions of interest (ROIs) from B and T cell regions of human tonsil tissue. Single cells and tissue ROIs were isolated with the integrated CytePicker module. SMART-seq2 whole transcriptome amplification was carried out on single retrieved cells and ROIs, followed by Illumina Nextera XT library preparation and sequencing on an Illumina MiSeq. Expression analysis was carried out with DESeq2 and TCR sequences were retrieved from the antigen-specific T cell RNAseq datasets utilizing the TraCeR software package. For 6-color imaging, tonsil and melanoma tissue sections were immunofluorescent stained with panels containing BV421, Alexa488, Alexa647, sytox orange, qDot-625, and qDot-800 fluorophores and scanned with the CyteFinder system.

Results

We validated T cell activation by gene expression analysis revealing upregulation of transcripts in pathways such as TCR signaling and inflammatory response/cytokine signaling and were able to match paired alpha/beta TCR sequences with previously published influenza antigen-specific T cell databases. Expression analysis referencing retrieved tonsil T cell zone ROIs against B cell zone ROIs resulted in an expected differential analysis, such as upregulation of CD8a, CCL19, and CCL21 and downregulation of CD38, CR2, and CXCL13.

Conclusions

We were able to confirm that anti-influenza T cells isolated with the RareCyte were specific for expected TCRs and that peptide stimulated cells had an activated phenotype through RNA sequencing. Additionally, we validate use of the instrument for the picking of tissue micro-regions with confirmation of the cell type and tissue micro- environment via RNA sequencing. We also demonstrate that the system can be used for 6-color tissue imaging of melanoma cells and tumor-associated immune cells.

Ethics Approval

Human samples in this study were procured from commercial vendors who collected them according to their established ethics policies

P133 Spatially resolved and multiplexed immunoprofiling of NSCLC using imaging mass cytometry reveals distinct functional profile of CD4 and CD8 TILs associated with response to immune checkpoint blockers

Franz Villarroel-Espindola, PhD1, Miguel Sanmamed, MD, PhD2, Jonathan Patsenker1, Ya-Wei Lin3, Brian Henick, MD4, Jovian Yu1, Mark Verburg5, Tayf Badri1, Jon Zugazagoitia, MD1, Daniel Carvajal-Hausdorf, MD6, Ruth Montgomery, PhD1, Roy Herbst, MD, PhD1, Lieping Chen, MD, PhD1, David Rimm, MD, PhD1, Tal Shnitzer, PhD3, Ronen Talmon, PhD3, Yuval Kluger, PhD1
1Yale School of Medicine, NEW HAVEN, CT, USA; 2Clínica Universidad de Navarra, New Haven, CT, USA; 3Israel Institute of Technology, Haifa, Israel; 4Columbia University, New York, NY, USA; 5Stony Brook University, New York, NY, USA; 6Clinica Alemana Universidad Desarrollo, New Haven, CT, USA
Correspondence: Franz Villarroel-Espindola (kurt.schalper@yale.edu)

Background

Reinvigoration of anti-tumor immunity by blocking co-inhibitory checkpoints has revolutionized the treatment of numerous human malignancies including non-small cell lung cancer (NSCLC). Understanding the mechanisms mediating anti-tumor effect and determinants of sensitivity/resistance to treatment remain major challenges. Efforts to comprehensively analyze intact tumor specimens have been limited by the number of targets that can be measured with preserved tissue architecture and in native cell conditions. We used a 29-marker IMC panel to study the immune composition of NSCLC and association with benefit to immune checkpoint blockers.

Methods

The IMC panel was established by validation of individual metal-conjugated primary antibodies in formalin-fixed/paraffin-embedded (FFPE) samples using cell-line controls, human tissue specimens and comparison with multiplexed immunofluorescence. Markers included structural tissue components: GAPDH/Histone- 3/LipoR/DNAintercalators; cell-phenotype markers: Vimentin/Cytokeratin/CD3/CD4/CD8/CD20/CD68/FOXP3/CD45RO; functional indicators: TBET/PD-1/LAG- 3/TIM-3/CD25/B2M/KI-67/GZB; and candidate immunomodulatory targets: PD-L1/PD-L2/B7-H3/B7-H4/PD- 1H/VISTA/IDO-1/CD47. We studied 12 NSCLC/normal-lung tissue pairs and 30 baseline tumor specimens from patients receiving PD-1 axis blockers represented in tissue microarrays. Treated cases included 12 with durable clinical benefit (DCB) and 18 without benefit (NDB). We analyzed the overall levels of the markers and profiles of individual T-cell subpopulations using MCDviewer software. We performed integrated and spatially-resolved analysis in marker-selected patches using mathematical integration of signal and training of a support vector machine classifier to distinguish patients with response/resistance to treatment.

Results

The IMC panel showed specificity for individual markers, reproducibility across experimental runs and concordance with immunofluorescence. NSCLCs showed increased CD4+/CD8+/CD20+ TILs with higher expression of functional markers than case-matched non-tumor lung tissue. In cases treated with immune checkpoint blockade we identified prominent differences in the T-cell profile between patients with benefit/DCB compared with those without response to treatment/NDB characterized by higher levels of effector memory CD8+/CD45RO+ TILs and lower levels of T-cell immune inhibitory receptors. Cases with primary resistance to treatment were associated with CD4+ or CD8+ TILs containing increased levels of both activation (CD25/TBET/GZB/Ki-67) and immune suppression/dysfunction markers (PD-1/LAG-3/FOXP3). The spatially resolved machine learning classifier stratified DCB/NDB cases with 90% of concordance.

Conclusions

We have standardized a 29-marker immuno-oncology multiplexed IMC panel for simultaneous, quantitative and spatially resolved analysis of multiple protein targets in FFPE specimens. Our results show higher adaptive immune response in NSCLC relative to patient-matched non-tumor lung and identify a distinct activated/dysfunctional profile of T-cells in patients with primary resistance to immune checkpoint blockers. Our work supports further exploration of machine learning-based classifiers to objectively integrate markers and use as biomarkers.

Acknowledgements

Thanks to Yale CyToF core facility and Fluidigm(R).Funding: Department of Defense Lung Cancer Research Program Career Development Award #LC150383; Yale SPORE in Lung Cancer P50CA196530; Stand Up To Cancer - American Cancer Society Lung Cancer Dream Team Translational Research Grant SU2C-AACR-DT17- 15; Lung Cancer Research Foundation grant.

Ethics Approval

The study was approved by Yale Human Investigation Committee protocol #1608018220

P134 Prognostic and predictive value of baseline biomarkers in advanced non-small cell lung cancer (NSCLC)

Huifen Wang, PhD, Robert McEwen, Ketan Patel, PhD, Binbing Yu, J Carl Barrett, Kris Sachsenmeier, PhD
AstraZeneca, Gaithersburg, MD, USA
Correspondence: Binbing Yu (yub@MedImmune.com)

Background

Expression levels of biomarkers measured prior to treatment may indicate high tumor burden and be useful to inform or optimize treatment options or be predictive of response.

Methods

We studied baseline status (post hoc) for several biomarkers that are commonly used in the “real world” clinical oncology setting, lactate dehydrogenase (LDH), neutrophil to lymphocyte ratio (NLR), carcinoembryonic antigen (CEA), and prostate-specific antigen (PSA). We used the FlatIron Health database to correlate baseline biomarker data with respect to overall survival (OS) to predict immune oncology (IO)/chemo/targeted therapy response in NSCLC. The study included patients diagnosed with stage IIIB/IV NSCLC, aged ≥18 years at diagnosis, who had a non-missing value of blood-based baseline biomarkers of interest, received ≥1 IO/chemo/targeted therapy, and who were not participating in clinical trials. Pre-specified values along with median values of each biomarker served as cutoffs to define patients with high and low level of each biomarker. Time from NSCLC diagnosis to first line (1 L) treatment start, and from 1 L treatment start to end of follow-up/death were used to define baseline and OS, respectively. Covariates of practice type, age at diagnosis, gender, histology, stage at diagnosis, smoking status, performance status, number of treatment lines, and 2 L treatment type (IO vs chemo/targeted therapy) were studied and included in the multivariate model to assess biomarkers’ prognostic value, and used for propensity score matching between the IO and chemo/targeted therapies.

Results

Of 10,678 patients included, most received 1 L (57.6%) and 2 L (26.8%) therapies. Of 716 patients on 1 L IO therapy, 18% remained on IO therapy; while 16.5% of patients on 1 L chemo/targeted therapy switched to IO therapy in 2 L. Baseline LDH, CEA, and NLR showed significant prognostic benefit for all treatments. A potential predictive value of LDH and CEA in response to IO therapy was observed: Patients on IO therapy with a low baseline biomarker level had improved OS vs patients on IO therapy with a high baseline biomarker level and chemo/targeted therapy-treated patients.

Conclusions

These data support the hypothesis that in NSCLC, patients with high tumor burden and/or specific leukocyte profiles are generally non-responsive to IO therapies.

P135 Integrative genomic and proteomic analysis identifies cancer subtypes and signaling networks associated with aberrant tumor expression of VISTA

Duncan Whitney, PhD1, Anna Ma, MS1, Ramachandra Katabathula, PhD2, Timothy Wyant1, Jefferson Parker1, Salendra Singh, MS2, David Tuck1, Vinay Varadan, PhD2
1Curis Inc, MA; 2Case Western Reserve University, Cleveland, OH, USA
Correspondence: Duncan Whitney (dwhitney@curis.com)

Background

VISTA is a negative regulator of T-cell and myeloid cell function and is gaining importance as a target for cancer immunotherapy [1-3]. VISTA is highly expressed in tumor-infiltrating leucocytes, particularly within the myeloid lineage. Recent evidence suggests that tumor cells themselves also express VISTA, exacerbating the immunosuppressive milieu within the tumor microenvironment [4]. Determining tumor subtypes that overexpress VISTA can inform the indication selection for VISTA-targeting agents and design of clinical trials in specific patient populations.

Methods

Genome-scale data was obtained from the TCGA Pan-Cancer Atlas dataset spanning 33 cancer types and >11,000 tumor samples [5]. DNA Methylation profiles were utilized to model leukocyte infiltration in individual tumor samples. The immune-cell component of VISTA expression was modeled using a nonlinear non-parametric regression model (LOESS). Non-immune compartment tumors with high VISTA expression, independent of leukocyte infiltration scores, were selected as likely over-expressing VISTA. VISTA-High tumor types and subtypes were determined and signaling network activities on a per-sample basis were derived using the InFlo systems framework. Pathways associated with aberrant VISTA expression were identified using a multivariate Gaussian model of gene expression and immune infiltration. Quantitative immunofluorescence (QIF) of VISTA protein expression was run on tissue microarrays (TMA): breast (n=602), lung (n=271), colorectal (n=2017), and ovarian (n=250) tumors. VISTA expression in stromal and tumor compartments was differentiated using labeled anti- cytokeratin and anti-VISTA antibodies.

Results

We identified 703 cancer samples (6.4%) exhibiting likely aberrant tumor cell expression of VISTA on tumor cells.

VISTA-high tumors were enriched (P<< 0.01) for Low Grade Glioma (LGG), Glioblastoma Multiforme (GBM), Kidney Renal Clear Cell Carcinoma (KIRC), Head and Neck Squamous Cell Carcinoma (HNSC), Sarcoma (SARC) and Mesothelioma (MESO). The CpG Island Methylator Phenotype (CIMP)-high of LGG and GBM, as well as the Atypical and Classical subtypes of HNSC were markedly enriched within the VISTA-High tumor population. We also identified subpopulations of lung (5.2%), breast (4.5%), colorectal (3.9%), and ovarian (4.1%) cancers to exhibit VISTA over-expression in the tumor compartment. Signaling networks identified as significantly associated with VISTA over-expression (P < 0.05) were identified. VISTA protein was expressed primarily in the stroma in nearly all (90-100%) of the TMA histospots, whereas between 5-10% showed VISTA expression in the tumor compartment, consistent with findings from the TCGA analysis.

Conclusions

Overexpression of VISTA on tumor cells was consistent between RNA analysis of TCGA and QIF- analyses of proteins in TMAs. This study represents the most comprehensive analysis of VISTA expression to date.

References

1. Lines JL, Sempere LF, Broughton T., Wang L, Noelle R. VISTA is a novel broad-spectrum negative checkpoint regulator for cancer immunotherapy. Cancer Immunol Res. 2014; (2): 510-7.

2. Nowak EC, Lines JL, Varn FS, Deng J, Sarde A, Mabaera R, Kuta A, Le Mercier I, Cheng C, Noelle RJ. Immunoregulatory functions of VISTA. Immunol Rev. 2017. 276: 66-79.

3. Gao J, Ward JF, Pettaway CA, Shi LZ, Subudhi SK, Vence LM, Zhao H, Chen J, Chen H, Efstathiou E, Troncoso P. VISTA is an inhibitory immune checkpoint that is increased after ipilimumab therapy in patients with prostate cancer. Nature medicine. 2017; 23(5): 551.

4. Boger C, Behrens HM, Kruger S, Rocken C. The novel negative checkpoint regulator VISTA is expressed in gastric carcinoma and associated with PD-L1/PD-1: A future perspective for a combined gastric cancer therapy? Oncoimmunology. 2017.

5. Thorsson V, Gibbs DL, Brown SD, Wolf D, Bortone DS, Ou Yang TH, Porta-Pardo E, Gao GF, Plaisier CL, Eddy JA, Ziv E, Culhane AC, Paull EO, Sivakumar IKA, Gentles AJ, Malhotra R, Farshidfar F, Colaprico A, Parker JS, Mose LE, Vo NS, Liu J, Liu Y, Rader J, Dhankani V, Reynolds SM, Bowlby R, Califano A, Cherniack AD, Anastassiou D, Bedognetti D, Rao A, Chen K, Krasnitz A, Hu H, Malta TM, Noushmehr H, Pedamallu CS, Bullman S, Ojesina AI, Lamb A, Zhou W, Shen H, Choueiri TK, Weinstein JN, Guinney J, Saltz J, R. A. Holt RA, Rabkin CE, Cancer Genome Atlas Research Network, Lazar AJ, Serody JS, Demicco EG, Disis ML, Vincent BG, Shmulevich L. The immune landscape of cancer. Immunity. 2018; 48: 812-30

P136 Genetic immunosignatures associate with progression-free survival in advanced soft tissue sarcoma patients treated on a Phase 2 trial of the VEGF receptor inhibitor axitinib plus pembrolizumab

Breelyn Wilky, MD1, SuFey Ong2, Sarah Warren, PhD2, Alessandra Cesano, MD, PhD2, Despina Kolonias1, Eric Wieder, PhD1, Deukwoo Kwon, PhD1, Andrew Rosenberg, MD1, Jonathan Trent, MD, PhD1, Krishna Komanduri, MD1
1University of Miami - SCCC, Miami, FL, USA; 2NanoString Technologies, Seattle, WA, USA
Correspondence: Breelyn Wilky (b.wilky@med.miami.edu)

Background

Vascular endothelial growth factor (VEGF) maintains the immunosuppressive tumor microenvironment by limiting T cell infiltration and promoting suppressive immune cell phenotypes. Accordingly, simultaneous blockade of VEGF with checkpoint inhibitors has led to improved immune cell infiltration and tumor responses in melanoma and renal cell carcinoma. In soft tissue sarcomas (STS), response rates to PD-1 monotherapy or dual CTLA4/PD-1 blockade are modest at 16-19% [1,2]. We hypothesized that addition of VEGF receptor inhibitor axitinib plus anti- PD1 checkpoint inhibitor pembrolizumab would improve responses in STS. In a Phase II study of axitinib/pembrolizumab in 30 patients with advanced STS (NCT02636725), we observed 4 month progression-free survival (PFS) of 47.7%. We obtained tumor biopsies from study patients and evaluated expression of immune- related gene signatures in respect to clinical outcomes.

Methods

Formalin-fixed, paraffin embedded core needle tumor biopsies were obtained from study patients at baseline and after 12 weeks on study treatment. RNA was extracted from unstained slides and hybridized with NanoString® IO360 beta gene expression panel (for research use only) prior to analysis on the nCounter® platform. After normalization with housekeeping genes and technical controls, expression of immune subset signatures was analyzed using research algorithms developed by NanoString.

Results

28 baseline and 14 on-treatment biopsies contained sufficient tumor for analysis after pathologist review. STS histologies included alveolar soft part sarcoma (ASPS, n=9), leiomyosarcomas (LMS, n=6), undifferentiated pleomorphic sarcoma (UPS, n=5), and other (n=9). Higher baseline expression of antigen presentation machinery, type 1 interferons, immunoproteosome, and interferon downstream signaling signatures was significantly associated with PFS > 4 months, whereas B cell, NK cell, and NK CD56dim signatures were negatively associated with PFS. Higher glycolysis gene expression in on-treatment samples relative to baseline was associated with PFS > 4 months. UPS significantly differed from other subtypes with higher baseline expression of proliferation, stroma, myeloid inflammation, PD-L2, B7-H3, and TGF-beta signatures. UPS immunoprofiles were significantly altered with treatment relative to other subtypes, with higher expression of myeloid inflammation, inflammatory chemokine, B cell, dendritic cell, and TGF-beta signatures. ASPS demonstrated higher baseline expression of endothelial cell, antigen presentation machinery, and apoptosis signatures, with lower proliferation relative to other subtypes.

Conclusions

This is the first analysis in STS patients treated with immunotherapy to correlate genetic expression signatures with clinical outcomes. Pathways identified by the NanoString IO360 beta panel will require prospective validation, but may ultimately serve as predictive biomarkers, and suggest alternative targets to further enhance efficacy of immunotherapy in STS.

Trial Registration

Clinicaltrials.gov: NCT02636725

References

1. Tawbi HA, Burgess M, Bolejack V, Van Tine BA, Schuetze SM, Hu J, D'Angelo S, Attia S, Riedel RF, Priebat DA, Movva S, Davis LE, Okuno SH, Reed DR, Crowley J, Butterfield LH, Salazar R, Rodriguez-Canales J, Lazar AJ, Wistuba II, Baker LH, Maki RG, Reinke D, Patel S. Pembrolizumab in advanced soft-tissue sarcoma and bone sarcoma (SARC028): a multicentre, two-cohort, single-arm, open-label, phase 2 trial. Lancet Oncol. 2017; 18(11):1493-1501.

2. D'Angelo SP, Mahoney MR, Van Tine BA, Atkins J, Milhem MM, Jahagirdar BN, Antonescu CR, Horvath E, Tap WD, Schwartz GK, Streicher H. Nivolumab with or without ipilimumab treatment for metastatic sarcoma (Alliance A091401): two open-label, non-comparative, randomised, phase 2 trials. Lancet Oncol. 2018; 19(3):416- 426.

Ethics Approval

The study was approved by the University of Miami Institutional Review Board, approval number 20150932.

P137 A systematic literature review (SLR) of tumor mutational burden (TMB) and efficacy with immunotherapy (IO) in lung cancer

Connor Willis, PharmD1, Michelle Fiander1, Dao Tran, PharmD2, Beata Korytowsky3, John-michael Thomas, PharmD3, Signe Fransen3, Florencio Calderon3, Teresa Zyczynski3, Lisa Siegartel3, Diana Brixner1, David Stenehjem2
1University of Utah, Salt Lake City, UT, USA; 2University of Minnesota, Duluth, MN, USA; 3Bristol-Myers Squibb, Plainsboro, NJ, USA
Correspondence: Diana Brixner (diana.brixner@utah.edu)

Background

TMB is an emerging biomarker that may predict response to IO. This SLR evaluates published evidence on TMB as a biomarker for efficacy of IO in lung cancers.

Methods

Cochrane SLR methodology was followed [1]. Searches were conducted through April 2018 using: Medline; EMBASE; EMCARE; and SCOPUS. Two clinical trial registries (clinicaltrials.gov; ICTRP) and published conference abstracts were also searched. Studies of any design assessing clinical efficacy (objective response rate [ORR], progression-free survival [PFS], overall survival [OS]) in lung cancer (NSCLC/SCLC) by TMB or the association of TMB with other biomarkers or enrichment factors were included.

Results

Database searches identified 3662 unique references, full text screening of 809 articles was conducted, and 81 studies met all inclusion criteria. TMB was reported primarily as total mutation count (n=32) or by mutations/megabase (mut/Mb) (n=40). TMB was categorized as low, intermediate, or high in 32 of 81 studies. Methods used to categorize TMB were predetermined thresholds (n=21), the 50th percentile (n=9), or other percentiles (n=2). All studies showed improved PFS (n=8) [2-9] and ORR (n=8) [2-4, 8-12] of IO in TMB high vs. low/intermediate. Significant efficacy results (p<0.05) for IO in TMB high were observed in 6 of 8 studies for PFS and 4 of 8 studies for ORR. Improved OS was observed in 7 of 9 studies with 2 of 9 studies showing significant results [2-4, 10, 12-16]. Results on OS are presented (Table 1) and highlight the small sample size of included studies. TP53 and EGFR mutations were positively and negatively associated with high TMB, respectively [17-19] [19-21]. The literature showed a mixed association of PD-L1 and TMB [2, 7, 22-29]. High TMB was associated with smoking history [7, 19, 30-40], squamous cell carcinoma [18, 33, 41, 42], and male gender [19, 39, 40].

Conclusions

This is the first SLR to assess the association of TMB and efficacy in lung cancer. Robust, adequately powered observational and prospective clinical studies should continue to assess TMB and other biomarkers to IO with clinical outcomes. This will validate ongoing data sets and further support precision treatment planning.

References

1. Cochrane handbook for systematic reviews of interventions, J.P.T. Higgins and S. Green, Editors. 2011.

2. Carbone DP, et al. First-line nivolumab in stage iv or recurrent non-small-cell lung cancer. N Engl J Med, 2017; 376(25): 2415-2426.

3. Goodman AM, et al. Tumor mutational burden as an independent predictor of response to immunotherapy in diverse cancers. Mol Cancer Ther, 2017; 16(11): 2598-2608.

4. Griesinger F, et al. Tumor mutation burden (TMB) is associated with improved efficacy of atezolizumab in 1l and 2l+ NSCLC patients. Oncology Research and Treatment, 2017; 40((Hellmann M.D.) Memorial Sloan Kettering Cancer Center, New York, USA): 220-221.

5. Hellmann MD, et al. Molecular determinants of response and resistance to anti-pd-(l)1 blockade in patients with NSCLC profiled with targeted next-generation sequencing (ngs). Journal of Clinical Oncology, 2017. 35(15).

6. Hellmann MD, et al., Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden. N Engl J Med, 2018. 378(22): 2093-2104.7.

7. Mahadevan, N., et al., Non-synonymous mutation burden in lung carcinoma is associated with durable clinical response to immune checkpoint blockade. Journal of Thoracic Oncology, 2017. 12(1): p. S428-S429.

8. Rizvi NA, et al., Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science, 2015. 348(6230): p. 124-8.

9. Roszik J, et al. Novel algorithmic approach predicts tumor mutation load and correlates with immunotherapy clinical outcomes using a defined gene mutation set. BMC Medicine, 2016; 14(1): 168.

10. Gettinger S, et al. Predictive value of measuring somatic mutations and tumor infiltrating lymphocytes for PD-1 axis therapy in non-small cell lung cancer (NSCLC). Journal of Thoracic Oncology, 2017; 12(1): S430-S431.

11. Haratani K, et al. Tumor immune microenvironment and nivolumab efficacy in EGFR mutation-positive non- small-cell lung cancer based on t790m status after disease progression during EGFR-tki treatment. Annals of Oncology. 2017; 28(7): 1532-1539.

12. Singal G, et al. Analyzing biomarkers of cancer immunotherapy (cit) response using a real-world clinico- genomic database. Annals of Oncology. 2017; 28((Abernethy A.) Medical Group, Flatiron Health, Inc., New York, USA): p. v404-v405.

13. Owada Y, et al. Correlation between mutation burden of tumor and immunological/clinical parameters in considering biomarkers of immune checkpoint inhibitors for non-small cell lung cancer (NSCLC). Journal of Clinical Oncology. 2017; 35(15).

14. Rothberg B G, et al. The prognostic impact of EGFR, KRAS and TP53 somatic mutations in curatively resected early-stage lung adenocarcinomas. Journal of Thoracic Oncology. 2017; 12(1): S623.

15. Schrock A B, et al. Pulmonary sarcomatoid carcinomas commonly harbor either potentially targetable genomic alterations or high tumor mutational burden as observed by comprehensive genomic profiling. J Thorac Oncol. 2017; 12(6): 932-942.

16. Chen M. et al. The potential clinical application of comprehensive genomic profiling in targeted therapy and immunotherapy of lung cancer. Journal of Thoracic Oncology. 2017;12(11):S2010.

17. Dong, Z.Y., et al., EGFR mutation correlates with uninflamed phenotype and weak immunogenicity, causing impaired response to PD-1 blockade in non-small cell lung cancer. Oncoimmunology, 2017. 6(11): p. e1356145.

18. Choi, M., et al., Mutation profiles in early-stage lung squamous cell carcinoma with clinical follow-up and correlation with markers of immune function. Annals of Oncology, 2017. 28(1): p. 83-89.

19. Goldberg, M.E., et al., The interaction of PD-L1, TMB, and genomic alterations in NSCLC. Cancer Research, 2017. 77(13).

20. Liu, S.Y., et al., Dual positive PD-L1 and cd8+ til represents a predominant subtype in NSCLC and correlates with augmented immunogenicity. Journal of Thoracic Oncology, 2017. 12(1): p. S237.

21. Liu, X., et al., Molecular biomarker study of programmed death receptor ligand 1 (PD-L1) in korean patients with lung adenocarcinoma [#4213], in AACR 2018 Proceedings: Abstracts 3028-5930. 2018, CTI Meeting Technology: Chicago.

22. Nakagomi, T., et al., New therapeutic targets for pulmonary sarcomatoid carcinomas based on their genomic and phylogenetic profiles. Oncotarget, 2018. 9(12): p. 10635-10649.

23. Schabath, M., et al., Molecular epidemiology of programmed cell death 1-ligand 1 (PD-L1) protein expression in non-small cell lung cancer. Journal of Thoracic Oncology, 2017. 12(1): p. S475-S476.

24. Senarathne, W., et al., Composition of the immune microenvironment differs between carcinomas metastatic to the lungs and primary lung carcinomas. Annals of Diagnostic Pathology, 2018. 33: p. 62-68.

25. Ross, J., et al., Immune checkpoint inhibitor (ICPI) efficacy and resistance detected by comprehensive genomic profiling (CGP) in non-small cell lung cancer (NSCLC) [#1138] Annals of Oncology, 2017. 28(Suppl 5).

26. He, L., et al., Evaluating genomic signatures predicting veliparib sensitivity in non-small cell lung cancer (NSCLC). Journal of Thoracic Oncology, 2017. 12(1): p. S267-S268.

27. Kadara, H., et al., Whole-exome sequencing and immune profiling of early-stage lung adenocarcinoma with fully annotated clinical follow-up. Annals of Oncology, 2017. 28(1): p. 75-82.

28. Kim, H.S. and J.Y. Han, Association of PD-L1 expression with tumor infiltrating immune cells and mutation burden in the high grade neuroendocrine carcinoma of the lung. Journal of Clinical Oncology, 2017. 35(15): p. Suppl 1.

29. Ono, A., et al., Clinical factors associated with mutation burden in non-small cell lung cancer. Annals of Oncology, 2017. 28((Yamaguchi K.) Hospital and Research Institute, Shizuoka Cancer Center, Shizuoka, Japan): p. v578-v579.

30. Quek, K., et al., Mutational landscape of non-small cell lung adjacent normal [#5363], in AACR 2018 Proceedings: Abstracts 3028-5930. 2018, CTI Meeting Technology: Chicago.

31. Reck, M., et al., Smoking history predicts sensitivity to parp inhibitor veliparib in patients with advanced non- small cell lung cancer. J Thorac Oncol, 2017. 12(7): p. 1098-1108.

32. Schrock, A.B., et al., Characterization of 298 patients with lung cancer harboring MET exon 14 skipping alterations. J Thorac Oncol, 2016. 11(9): p. 1493-502.

33. Shim, H.S., et al., Unique genetic and survival characteristics of invasive mucinous adenocarcinoma of the lung. J Thorac Oncol, 2015. 10(8): p. 1156-62.

34. Wang, C., et al., The correlation between mutation burden and disease free survival in patients with lung adenocarcinomas. Journal of Clinical Oncology, 2017. 35(15).

35. Xiao, D., et al., Analysis of ultra-deep targeted sequencing reveals mutation burden is associated with gender and clinical outcome in lung adenocarcinoma. Oncotarget, 2016. 7(16): p. 22857-64.

36. Xiao, D., et al., Integrative analysis of genomic sequencing data reveals higher prevalence of lrp1b mutations in lung adenocarcinoma patients with copd. Sci Rep, 2017. 7(1): p. 2121.

37. Isaka, M., et al., Integrated genomic analysis to assess the molecular signature of Japanese patients with non- small cell lung cancer. Journal of Thoracic Oncology, 2017. 12(11): p. S2292.

38. Kojima, H., et al., Genomic analysis to assess a molecular signature in Japanese patients with pulmonary high grade neuroendocrine carcinoma. Journal of Thoracic Oncology, 2017. 12(11): p. S2186.

39. Davis, A.A., et al., Association of tumor mutational burden (TMB) with DNA repair mutations and response to anti-PD-1/PD-L1 therapy in non-small cell lung cancer (NSCLC). Cancer Research, 2017. 77(13).

40. Park, W., et al., Correlating isend and tumor mutation burden (TMB) with clinical outcomes of advanced non- small cell lung cancer (ansclc) patients on nivolumab. Journal of Thoracic Oncology, 2017. 12(11): p. S2005-S2006.

41. Patel, K., et al., In silico analysis of non-synonymous snps (nsSNPs) and outcomes in non-small cell lung cancer (NSCLC) patients (pts) treated with immunotherapy (it). Journal of Clinical Oncology, 2017; 35(15).
Table 1 (abstract P137).

Overall survival with IO by TMB

P138 Quantitative assessment of cancer-associated fibroblasts and immunotherapy outcome in metastatic melanoma

Pok Fai Wong, MD, MPhil1, Wei Wei, MD, PhD2, Swati Gupta, PhD1, James Smithy, MD, MHS3, Harriet Kluger, MD1, David L. Rimm, MD, PhD1
1Yale School of Medicine, New Haven, CT, USA; 2Yale School of Public Health, New Haven, CT, USA; 3Brigham and Women's Hospital, Boston, MA, USA
Correspondence: David L. Rimm (david.rimm@yale.edu)

Background

Clinical benefit from programmed cell death 1 (PD-1) immune checkpoint blockade is limited to a subset of metastatic melanoma patients, so there is a need for predictive biomarkers. Because the cancer-associated fibroblast (CAF) population is the predominant stromal cell type within the tumor immune microenvironment, we hypothesized that pretreatment CAF profiles could be associated with immunotherapy outcome.

Methods

Pretreatment whole tissue sections from 117 melanoma patients treated with anti-PD-1 therapy (pembrolizumab, nivolumab, or ipilimumab plus nivolumab) from 2011–17 were collected from Yale Pathology archives. Multiplex immunofluorescence for CAF profiling was achieved by simultaneous detection of nuclei (DAPI), melanoma cells (S100, HMB45), and the CAF markers, Thy1 (7E1B11, Abcam), smooth muscle actin (SMA; 1A4, Dako), and fibroblast activation protein (FAP; EPR20021, Abcam). Cell phenotyping and counting were performed using inForm software (PerkinElmer) and protein expression was measured by the AQUA method of quantitative immunofluorescence (QIF). CAF parameters by both methodologies were correlated with best overall response as defined by Response Evaluation Criteria in Solid Tumors (RECIST) 1.1, progression-free survival (PFS), overall survival (OS), and immune markers previously measured in this cohort.

Results

Pretreatment CAF parameters, by cell counts or QIF, were not associated with RECIST tumor burden classifications for best overall response, objective response rate or disease control rate. In contrast, PFS (all P < 0.05) and OS (all P < 0.003) had significant positive associations with Thy1 and FAP cell counts, and negative associations with SMA cell count, which were specific to anti-PD-1 treated patients. Similar associations were not observed in a historical untreated melanoma cohort. In the absence of therapy, FAP was instead a negative prognostic biomarker (P = 0.01). The specific association of FAP with anti-PD-1 survival advantage suggests mechanistic involvement and warrants further study. Multivariable analyses also revealed statistically significant PFS and OS associations with the CAF parameters, particularly for FAP, independent of age, sex, mutation, stage, treatment, and prior immune checkpoint blockade. The QIF data showed similar trends. There was no correlation between the CAF parameters and CD8 or PD-L1 by either method of assessment.

Conclusions

Pretreatment CAF parameters, by cell counts or QIF, are associated with immunotherapy outcome in metastatic melanoma patients. Multiplex analysis of the tumor microenvironment has potential to be used as a companion diagnostic for precision immunotherapy and may be complementary to existing markers (CD8 and PD-L1).

Ethics Approval

The study was approved by the Yale Human Investigation Committee protocol #9505008219.

P139 Pharmacodynamic effects of CA170, a first-in-class small molecule oral immune checkpoint inhibitor (ICI) dually targeting V-domain Ig suppressor of T-cell activation (VISTA) and PD-L1

Timothy Wyant, PhD1, Funda Meric-Bernstam, MD2, David Tuck3, Yung-Jue Bang, MD PhD4, Anna Ma, MS3, Jeffrey Sosman, MD5, Adil Daud, MBBS MD6, John Powderly, MD, CPI7, Javier Garcia-Corbacho8, Manish Patel, MD9, James Lee, MD, PhD10, Kyu-Pyo Kim11, Joshua Brody, MD12, Sun Young Rha13, Erika Hamilton, MD14, Marta Gil Martin15, Santiago Ponce Aix, MD16, Radhakrishnan Ramchandren, MD17, Myung-Ju Ahn18, James Spicer, MD, PhD19, Simon Pacey20, Gerald Falchook, MD21, Funda Meric-Bernstam, MD2
1Curis Inc, Lexington, MA, USA; 2MDACC, Houston, TX, USA; 3Curis, Lexington, MA, USA; 4Seoul National University Hospital, Seoul, Korea; 5Northwstern, Chicago, IL, USA; 6UCSF, San Francisco, CA, USA; 7Carolina BioOncology Institute, Huntersville, NC, USA; 8Hospital Clinic Barcelona, Barcelona, Spain; 9FLorida Cancer Specialists SCRI, Sarasota, FL, USA; 10UPMC, Pittsburgh, PA, USA; 11AMC, Seoul, Korea, Republic of; 12Mt.Sinai, New York, NY, USA; 13Yonsei University Health, Seoul, Korea, Republic of; 14Tenn Oncology SCRI, Nashville, TN, USA; 15Catalan Institute of Oncology, Catalon, Spain; 16Hosptial Univesitario, Madrid, Spain; 17Karmanos, Detroit, MI, USA; 18Samsung Medical center, Seoul, Korea, Republic of; 19King's College Guy's Hospital, London, UK; 20University of Cambridge, Cambridge, UK; 21SCRI Healthone, Denver, CO, USA
Correspondence: Simon Pacey (hwwang.2004@gmail.com); Gerald Falchook

Background

VISTA was shown to independently suppress T cell responses and is expressed on both immune and tumor cells. VISTA expression is upregulated in tumors as a potential resistance mechanism after ICI therapy. As such, it has been considered a target for cancer immunotherapy. Pre-clinical studies have demonstrated dual blockade of PD-1 and VISTA can be synergistic. CA-170, novel oral dual inhibitor of VISTA and PD-L1/L2 is in phase 1 study with exploratory pharmacodynamic endpoints.

Methods

Longitudinal blood samples were collected for PBMC phenotyping. Archival tumor tissue was acquired on all patients and paired tumor biopsies (baseline and C2) were collected when feasible. IHC using select markers of interest and Nanostring immune transcriptome analyses were conducted on paired tumor samples.

Results

At the time of this analysis 41 patients had been treated across 6 dose levels (50 – 800 mg QD) with 39 having blood samples available for PBMC phenotyping, 19 patients had sufficient paired tumor biopsies for IHC, and 9 for Nanostring-based analysis. In the 39 patients with available blood samples, within 24 hours after the initial CA-170 dose, 28% showed an increase in the percentage of peripheral blood CD8 T cells that expressed CD69 (median +3.2 fold) and 21% showed an increase in CD8 T cells that expressed CD134+ (median + 2.8 fold). IHC staining analysis of paired tumor biopsies showed increases of CD8+ T cell population in 39% (1.5 fold to 13.8 fold) and CD11b+ myeloid population in 36% of samples (1.5 fold to 5 fold) at cycle 2. Additionally there tended to be increased expression of VISTA in 38% of evaluated samples (1.5 fold to 14.5 fold) post-treatment. Nanostring immune panel transcriptome analysis showed a trend for increased expression of signatures of T-helper (3/9 pairs), myeloid (9/9 pairs), and T-reg cells (3/9 pairs). Activation of IFN-γ gene signature and certain induced genes (CXCL9, CXCL10, CXCL11) showed an increase of at least one transcript ranging from 1.97 to 7.6 fold (7/9 pairs).

Conclusions

These data suggest that CA-170 treatment results in modulation of peripheral and intra-tumoral immune profile in treated patients. CA-170 treatment was associated with altered myeloid cell as well as T helper cell populations more frequently than CD8+ T cells. Increased expression of VISTA in tumor biopsies were noted post CA-170 treatment. Correlation between the PD effects and tumor response to CA-170 treatment is being explored.

Trial Registration

NCT02812875

Ethics Approval

The following institutions ethics boards have approved this study:MD Anderson Cancer Center, Seoul National University Hospital, Northwestern, UCSF, Carolina BioOncology Institute, Hospital Clinic de Barcelona, Florida Cancer Specialists/Sarah Cannon Research Institute, UPMC, Asian Medical Center, Mt. Sinai, Yonsei University Health System - Severance Hospital, Tennessee Oncology/Sarah Cannon Research Institute, Catalan Institute of Oncology, Hospital Universitario 12 de Octubre, Karmanos, Samsung Medical Center, King's College London, Guy's Hospital, University of Cambridge,Sarah Cannon Research Institute at HealthONE

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal

P140 Deep profiling of Asian NSCLC to identify the tumor antigen-specific T cells and the predictive potential of the patients treated with PD-1/PD-L1 blockade

Joe Yeong, MBBS, PhD1, Lisda Suteja2, Yannick Simoni3, Kah Weng Lau4, Sherlly Lim4, Jie Hua, Josh Loh4, Angela Takano4, Eng Huat Tan, MD2, Kiat Hon, Tony Lim4, S. W. Daniel Tan2, Evan W. Newell, PhD3
1Singapore Immunology Network/ Singapore General Hospital, Singapore, Singapore; 2National Cancer Center Singapore, Singapore, Singapore; 3Singapore Immunology Network, Singapore, Singapore; 4Singapore General Hospital, Singapore, Singapore
Correspondence: Tony Lim (lim.kiat.hon@singhealth.com.sg)

Background

Although PD-L1 tumor proportion score (TPS)(1), microsatellite instability status (MSI)(2) and interferon gamma (IFN-γ)(3) gene signature have been widely recognized as biomarkers to predict the responsiveness of PD-1/PD-L1 blockade treatment, more clinically robust and practical markers are needed to more precisely identify potential responders and also to warn of potential toxicity especially when combination immunotherapies are used. Asian Non-Small Cell Lung Cancer (NSCLC) are differ from their western counterparts in terms of etiology with low prevalence of smoking and high incidence of EGFR mutant positive adenocarcinoma(4). Following up on our recent study (Nature 2018)(5) that profiled and implicated CD39+CD8+ T cells as tumor antigen-specific T cell populations in both Asian NSCLC and colorectal cancer tumors, here we investigate the tissue localization and predictive potential of CD39+CD8+ T cells in the context of NSCLC. Because CD39 expression by CD8+ T cells in these tumors appears to be associated with tumor-reactivity, we hypothesize that patients with higher frequencies or densities of these cells should respond better to checkpoint blockade immunotherapy.

Methods

Quantifying these cells in-situ using formalin-fixed paraffin-embedded (FFPE) archival tissue is difficult due to the often-abundant expression of CD39 by tumor cells or other infiltrating immune cells, which confounds accurate quantification of densities/frequencies of CD39+CD8+ T cells by standard immunohistochemistry. To overcome this challenge, we developed a fully automated multiplex Immunohistochemistry (m-IHC) protocol with clinical autostainer to visualize and quantitate tumor infiltrating lymphocytes (TILs) on FFPE samples (Pathology 2018)(6). In addition to manual scoring by two pathologists, CD39+CD8+ T cells frequencies/densities, TILs and tumor cells expression were assessed using pathological image analysis software.

Results

Based on this approach, we found that the quantification of CD39+CD8+ T cells by CyTOF and m-IHC yield decent correlation which lays the foundation to translate this to clinical practice in near future. We also identified that the abundances of such subsets are associated with clinicopathological parameters such as EGFR mutation status. We then applied such quantification method on the pre-treatment FFPE biopsies samples of a pilot retrospective cohort of NSCLC treated with PD-1/PD-L1 blockade (N=20) and found that the abundance of these CD39+CD8+ T cells population are of predictive value. The prediction is independent to the abundance of CD8+ T cells as well as other TILs and CD39 expression on tumor cells.

Conclusions

Further study is ongoing to expand the cohort and explore the predictive potential of this biomarker compared to the TPS, MSI and FN-γ gene signature.

References

1. Reck M, Rodríguez-Abreu D, Robinson AG, Hui R, Csőszi T, Fülöp A, et al. Pembrolizumab versus Chemotherapy for PD-L1–Positive Non–Small-Cell Lung Cancer. New Engl J Med. 2016;375(19):1823-33.

2. Le DT, Uram JN, Wang H, Bartlett BR, Kemberling H, Eyring AD, et al. PD-1 Blockade in Tumors with Mismatch-Repair Deficiency. New Engl J Med. 2015;372(26):2509-20.

3. Ayers M, Lunceford J, Nebozhyn M, Murphy E, Loboda A, Kaufman DR, et al. IFN-γ–related mRNA profile predicts clinical response to PD-1 blockade. The Journal of Clinical Investigation. 2017;127(8):2930-40.

4. Cheng T-YD, Cramb SM, Baade PD, Youlden DR, Nwogu C, Reid ME. The International Epidemiology of Lung Cancer: Latest Trends, Disparities, and Tumor Characteristics. Journal of Thoracic Oncology. 2016;11(10):1653-71.

5. Simoni Y, Becht E, Fehlings M, Loh CY, Koo S-L, Teng KWW, et al. Bystander CD8+ T cells are abundant and phenotypically distinct in human tumour infiltrates. Nature. 2018;557(7706):575-9.

6. Lim JCT, Yeong JPS, Lim CJ, Ong CCH, Wong SC, Chew VSP, et al. An automated staining protocol for seven- colour immunofluorescence of human tissue sections for diagnostic and prognostic use. Pathology. 2018;50(3):333- 41.

Ethics Approval

The Centralized Institutional Review Board of SingHealth provided ethical approval for the use of patient materials in this study (CIRB ref: 2011/411/B)

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.

P141 Circulating tumor DNA ( ctDNA) a novel biomarker for immunotherapy response in advanced lung cancer

Meera Yogarajah, MD1, Ebenezer Appah, MD1, Katherine Neblett2, Clive Morris2, Greg Jones2, Vincent Plagnol2, Paul Walker, MD1
1East Carolina University, Greenville, NC, USA; 2Inivata, Research Triangle Park, NC, USA
Correspondence: Meera Yogarajah (meera.bhradeev@gmail.com)

Background

Circulating tumor DNA (ctDNA) can be predicative of outcomes in lung cancer. Moreover the ctDNA levels can correlate with changes in tumor burden in response to therapy. We assessed the utility of (ctDNA) levels as an early indicator of response to immune-therapy.

Methods

Twenty-nine patients with advanced non-small cell lung cancer initiated on immune therapy with anti-PD1/PDL-1 therapy either alone or in combination with platinum-based chemotherapy were enrolled in this prospective trial. Patients had baseline plasma samples collected prior to therapy and serially with the initial 4 cycles of immuno- or chemo-immunotherapy. ctDNA was assessed in plasma by InVisionFirst™ (Inivata) ctDNA NGS assay for detection and quantification of genomic alterations in 36 genes commonly mutated in NSCLC. The early trends of the ctDNA allele fractions were correlated with imaging responses post 4 cycles of therapy and subsequently with interval imaging.

Results

Patients included 31% male, 62 years median age, 21% squamous, 72% adeno, 7% others, median cycles 4. Of the 28 patients, 7 were not evaluable at the time of analysis as they only had baseline values or did not have response assessment imaging. Twenty two patients had evaluable imaging assessments which was done on completion of C4 and later at intervals determined by the treating oncologist. Clinical benefit was demonstrated in 16 (55%) patients (complete response (CR, n =4), partial response (PR, n = 5) or stable disease (SD, n = 7)); 6 patients had progressive disease (PD). In patients achieving CR/PR there was no detectable ctDNA at baseline (n=4) or there was complete clearance of ctDNA after completion of 3-4 cycles of immune therapy (n=4), except for 1 patient who demonstrated persistent ctDNA. Follow up imaging demonstrated continued beneficial responses, with most patients who had CR continuing to be in CR and patients who had PR persisting as PR or improving to CR. Patients with stable disease had varying levels of ctDNA with mildly increasing levels in some and stable low levels in others. All 6 patients with PD had detectable ctDNA at progression and some showed increasing levels.

Conclusions

Early decrease or clearance of ctDNA during immune therapy was correlated with positive clinical responses.

Absence of detectable ctDNA was indicative of overall a good response and prognosis. Increasing or newly detectable ctDNA was indicative of progressive disease or poor overall outcome. These results suggest a value for validation in an expanded patient cohort.

P142 The immunogenomic impact of indoximod on the tumor microenvironment of melanoma patients

Jiayi Yu, PhD1, Gabriela R. Rossi, PhD1, Ravindra Kohle, MD, PhD2, David Munn, MD2, Yousef Zakharia, MD3, Nicholas Vahanian, MD1, Eugene Kennedy, MD, FACS1, Charles Link, MD1
1NewLink Genetics, Ames, IA, USA; 2Medical College of Georgia, Augusta Univ, Augusta, GA, USA; 3University of Iowa, Iowa City, IA, USA
Correspondence: Gabriela R. Rossi (gabyrossi@icloud.com)

Background

The indoleamine 2,3-dioxygenase (IDO) pathway mediates immunosuppressive effects through the metabolism of tryptophan (Trp) to kynurenine (Kyn). This metabolic pathway triggers downstream signaling through the Trp sensors (GCN2 and mTOR) and the Kyn sensor (aryl hydrocarbon receptor, AhR) [1-4]. Indoximod is an orally administered, small-molecule IDO pathway inhibitor that reverses the immunosuppressive effects of low Trp and high Kyn that result from IDO activity. Preclinical studies demonstrate that indoximod has immunostimulatory effects involving three main cell types: CD8+ T cells, Tregs, and DCs. Indoximod increases proliferation of effector T cells, reprograms Tregs into helper T cells, and downregulates IDO expression in DCs. These effects are observed in both the presence and absence of IDO activity [5].

Methods

Patients with newly diagnosed unresectable locally advanced or metastatic melanoma in Phase 2 trial (NCT02073123) underwent pre-treatment tumor biopsy followed by a repeat biopsy after cycle 3 of pembrolizumab and indoximod. Fourteen pairs of tumor specimens (6 patients with objective response, and 8 non-responders) underwent RNA sequencing analysis and multiplex immunofluorescence staining to assess the phenotype and functional status of multiple immune populations in the tumor microenvironment (TME), define changes in the tumor genomic profile and gene expression. Baseline samples from the trial were used for predictive biomarker assessment (n= 38).

Results

Expression profiling identified up-regulation of multiple immune regulation pathways previously reported following pembrolizumab treatment. Importantly, comparison against published studies suggested immunologic and metabolic changes contributed exclusively by indoximod, including genes suggestive of increasing cytotoxicity and innate immune cell infiltration and activation (CD14, CD33, CD86, GZMM, CD11c, IRF8 among others). Melanoma- related genes were markedly decreased in responding patients compared to non-responding patients. Pro- inflammatory immunologic changes were observed only in the clinical-responder patients, while the non-responder patients showed minimal immunologic response. Consistent with the hypothesized mechanism of action for indoximod [5], IDO1 expression within the TME was downregulated upon treatment, especially in Ki67neg population. Considered as a predictive biomarker, patients with high IDO expression showed higher probability of response to treatment and longer progression-free survival (PFS). This result was independent of the expression levels of PD-L1.

Conclusions

The combination of indoximod and pembrolizumab induced multiple immunologic and metabolic changes in the TME. Comparison analysis indicates that some of these changes appear to be contributed exclusively by indoximod. High IDO1 expression at baseline shows correlation with clinical response to treatment.

Trial Registration

ClinicalTrials.gov Identifier NCT02073123

References

1. McGaha TL, et al. Immunol Rev. 2012;249(1):135-157.

2. Munn DH, et al. Immunity. 2005;22(5):633-642.

3. Metz R, et al. OncoImmunology. 2012;1(9):1460-1468. 4. Opitz CA, et al. Nature. 2011;478(7368):197-203.

5. Brincks EL, et al. Presented at: Annual Meeting of the American Association for Cancer Research (AACR); April 14-18, 2018; Chicago, IL. Abstract 3753.

P143 Rational combination of GITR agonism with PD-1 blockade in cancer patients

Roberta Zappasodi, PhD2, Cynthia Sirard, MD3, Yanyun Li, PhD MD2, Sadna Budhu, PhD2, Moshen Abu-Akeel2, Cailian Liu, MD2, Xia Yang2, Hong Zhong, BS2, Walter Newman, PhD3, Jingjing Qi2, Phillip Wong, PhD2, David Schaer2, Henry Koon, MD4, Vamsidhar Velcheti, MD FACP5, Michael Postow, MD2, Margaret K. Callahan, MD, PhD2, Jedd Wolchok, MD, PhD2, Taha Merghoub, PhD2
1Memorial Sloan Kettering Institute, New York, NY, USA; 2MSKCC, New York, NY, USA; 3Leap Therapeutics, Walpole, MA, USA; 4Case Western Reserve University, Cleveland, OH, USA; 5Cleveland Clinic, Pepper Pike, OH, USA
Correspondence: Taha Merghoub (merghout@mskcc.org)

Background

Despite the clinical successes of checkpoint blockade, many patients are refractory to these therapies, highlighting the need of more effective combination programs targeting alternative immune pathways. Engaging the T-cell co- stimulatory receptor glucocorticoid-induced TNFR-related protein (GITR) with agonist antibodies has shown promising activity in preclinical mouse models. Based on this rationale, we initiated the first in-human phase-I trial of GITR stimulation with the agonist antibody TRX518 in advanced solid cancer patients (NCT01239134). Treated patients showed frequent reductions in circulating regulatory T cells (Tregs) that correlated with intra-tumor Treg reductions. However, this was not sufficient to achieve a clinical benefit. Here, we investigate the mechanisms underlying resistance of advanced tumors to GITR agonism and provide the rationale to combine anti-GITR with checkpoint blockade in clinical trials.

Methods

Mice were implanted with B16F10 murine melanoma and treated with the anti-GITR DTA-1 alone or in combination with the anti-PD-1 RMP1-14. Composition, phenotype and function of tumor-infiltrating T cells were analyzed by flow cytometry, in vitro killing assays and molecular profiling.

Results

We modeled tumor sensitivity and refractoriness to anti-GITR therapy by treating B16F10-bearing mice on day 4 (responsive tumors) or day 7 (refractory tumors) after tumor implantation respectively. We found that intra-tumor Tregs were significantly reduced and effector-T-cell:Treg ratios increased in both responding and refractory tumors. However, time course analyses revealed complete lack of Treg accumulation in day-4-treated responding tumors, suggesting that Tregs may limit T-cell functionality during tumor development. Accordingly, in refractory compared to responding tumors, CD8+ T cells down-regulated activation/memory T-cell markers and up-regulated exhaustion markers. To overcome resistance to anti-GITR, we co-administered anti-PD-1 with the day-7 anti-GITR suboptimal treatment. This combination controlled tumor growth similar to the day-4 curative anti-GITR monotherapy and achieved 50% long-lasting complete responses. These effects were associated with intra-tumor infiltration of highly activated and cytolytic CD8+ T cells. Based on these results and considering the biologic activity and safety profile of TRX518, we have initiated the clinical investigation of TRX518 in combination with anti-PD-1 in patients with advanced solid tumors (NCT02628574). Noteworthy, 3 of the first patients enrolled in this study have manifested clinical responses.

Conclusions

These findings indicate that Treg elimination from advanced tumors is not sufficient to activate cytotoxic CD8+ T-cell responses unless the T-cell exhaustion process is concurrently blocked, underscoring the need to combine Treg- inhibiting/depleting immunotherapies with strategies that counteract exhaustion to regress advanced tumors. This study highlights the importance of developing rational, evidence-based combination immunotherapies.

Ethics Approval

All mouse procedures were performed in accordance with institutional protocol guidelines at MSKCC. All procedures involving human subjects, human material, or human data, or involving animals were in compliance with the ethical regulations.

P144 A fully-automated multiplex fluorescence IHC assay with whole slide multispectral imaging on mouse tissue: phenoptics™ quantitative pathology solutions translational workflow

Yi Zheng, PhD, Carla Coltharp, PhD, Ryan Dilworth, PhD, Rachel Schaefer, Linying Liu, Victoria Duckworth, William Kennedy, Darryn Unfricht, PhD, Peter Miller, MS, Milind Rajopadhye, PhD
PerkinElmer Inc., Hopkinton, MA, USA
Correspondence: Milind Rajopadhye (milind.rajopadhye@perkinelmer.com)

Background

Developing biomarker research strategies for clinically relevant therapies in immuno-oncology is predicated on the ability to execute fully translational research studies. Phenoptics™ quantitative pathology solutions (QPS) is a comprehensive, end-to-end solution consisting of multiplex fluorescence immunohistochemistry staining along with multispectral imaging and tissue analysis. It provides an effective and quantitative method to reveal multiple biomarkers or cell types and their interaction within tissue on a per-cell and per-tissue-context basis. It enables a deeper understanding of the biology of the tumor microenvironment. Recently, the applications for Phenoptics have been expanded from regional areas of interest in human tissue Immuno-oncology research into whole slide IO tissue research, including animal studies, creating a truly translational platform. Here we describe a robust, fully- automated 7-color Opal staining procedure on Formalin-fixed paraffin-embedded (FFPE) section of mouse breast cancer followed by a multispectral whole slide scan and image analysis.

Methods

FFPE samples from mouse breast cancer were immunostained using Opal Polaris™ 7-color Automation IHC Kit on the Leica BOND RX™ automated stainer. Multispectral fluorescence imagery was acquired on a Vectra Polaris® automated imaging system and analyzed with inForm® and MATLAB® software.

Results

We’ve applied the fully-automated Opal staining procedure to a mouse breast cancer sample by using the new Opal Polaris 7-Color Whole Slide reagents plus the newly developed Opal Polymer anti-Rabbit HRP Secondary Antibody. Opal multiplexed fluorescence IHC in conjunction with tissue and cell segmentation and phenotyping using inForm image analysis software allowed us to reliably interrogate subsets of T cells, macrophages, tumor cells and proliferating cells. With the ability to spectrally unmix the whole slide image, we are able to analyze and examine spatial relationships between specific phenotypes in a much larger scale, across tissue types for fully translational studies.

Conclusions

The fully-automated Opal multiplex fluorescence staining assay and whole slide multispectral imaging that we developed for FFPE tissue research is illustrated here with a mouse breast cancer model, expanding the Phenoptics workflow into animal studies. It allows for unique tumor microenvironment assessment in the whol