The present study describes the initial results of tumor response characteristics using irRECIST1.1 among advanced NSCLC patients treated with commercially-prescribed nivolumab in the clinical setting, and provides a direct comparison of the assessment results between irRECIST1.1 and RECIST1.1. The response rate in this cohort was 14% both by irRECIST1.1 and by RECIST1.1. No patient experienced pseudoprogression during the study period. Lesion-based assessment showed significant differences of responses across organs, with adrenal lesions and lymph nodes being more responsive and liver lesions being less responsive to therapy. To our knowledge, this is the first study that demonstrated the organ-specific tumor responses in patients treated with immune checkpoint inhibitors.
The overall response rate of 14% in the present study is similar to prior clinical trials of nivolumab or pembrolizumab in NSCLC, where response rates of 14.5–19.4% were reported [4, 8, 9, 12]. All responders in the present cohort had PR and no patient achieved CR; this is also similar to clinical trial results, where CR was noted in 0 of 117 nivolumab-treated squamous NSCLC patients [4], in 0.8% among 495 pembrolizumab-treated NSCLC patients [9], and 1% among 292 nivolumab-treated non- squamous NSCLC patients [12]. Tumor burden change at BOR per patient had a wide range (−66.8 to +278.1%) with a median of 3.9% in the present study, indicating a wide range of responses and an apparent heterogeneity of sensitivity to PD-1 inhibitors among advanced NCCLC population as noted in the trials [4, 8, 12]. Demographics and clinical characteristics including histology and smoking history showed no association with response to nivolumab. In the recent two trials of PD-1 inhibitor, the response rate was similar between squamous and nonsquamous NSCLC [8, 9], which is consistent with our results. Smoking history was associated with a higher rate of response to nivolumab in a phase 1 study [8]; however, such observation was not noted in our cohort, likely due to our small sample size. Other molecular markers such as PD-L1 positivity of tumor tissue were not available in most patients in this clinical retrospective cohort.
Head-to-head comparisons between immune-related response evaluations and the conventional RECIST assessments are lacking in most of the trials of immune-checkpoint inhibitor therapy, as recently described by Chiou et al. [27]. Therefore, providing such comparisons was one of the major goals of the present study. Assessment of PR was fully concordant between irRECIST1.1 and RECIST1.1, in this cohort with no cases of pseudoprogression. The discordance of BOR assessment between two criteria was only noted between SD and PD, either because of the requirement of confirmation for PD or the inclusion of new lesions in the total tumor burden by irRECIST1.1, which contributed equally to the discordance of BOR. These two features of irRECIST1.1 also contributed to longer TTP by irRECIST1.1 (irTTP) compared to TTP by RECIST1.1, which is an expected consequence of immune-related response evaluations as reported in a prior study of melanoma patients treated with ipilimumab [25]. Median TTP by RECIST1.1 was 1.9 months in the present cohort; in spite of the relatively short follow-up period of the present study, the result is overall similar to the previous studies in the trial cohorts reporting median PFS of 1.9-3.7 months based on RECIST [4, 8, 9, 12]. The present study focused on TTP, rather than PFS, as its major purpose was the differences derived from two tumor response criteria, which TTP reflects most accurately and is not affected by death as an event.
Pseudoprogression, or initial progression with tumor burden increase followed by subsequent response, is a challenging phenomenon during immune checkpoint inhibitor therapy. None of the patients in the present study experienced pseudoprogression during the treatment period of the study, which can be due to the small number of patients studied in a relatively short period of time during this initial clinical experience. In a phase 1 study of nivolumab in advanced NSCLC, 5% (6/129) of the patients experienced pseudoprogression, which were within 20 weeks since initiation of therapy [8]. On the other hand, a recent study et al. in 44 NSCLC patients treated with trials of PD-1 and PD-L1 inhibitors, 4 patients (9%), all of whom received PD-L1 inhibitors, had experienced pseudoprogression at 3 months of therapy [28]. Further investigations are needed in larger cohorts to determine the exact incidence of pseudoprogression in advanced NSCLC patients subclassified according to the types of immune-checkpoint inhibitors used in the treatment regimen.
The appearance of new lesions presents another management dilemma during immune-checkpoint inhibitor therapy because it may indicate true progression or pseudoprogression. In the present study, one-third of patients developed new lesions, most commonly in the liver. Though no patients showed partial response based on irRECIST measurements at the time of appearance of new lesions, 44% of the patients (8/18) had measureable tumor burden changes within the range of SD. Notably, 2 of these patients had new target lesions and had SD even though the measurements of these new lesions were added to the total tumor burden. These observations are indicative of the merit of the use of irRECIST for detection of stable disease in the setting of appearance of new lesions.
Most patients with follow-up scan after the appearance of new lesions showed progression of new lesions, and none of the new lesions subsequently responded during therapy. Though very few reports describe the detailed locations and behaviors of new lesions during immune-checkpoint inhibitor therapy, in a recent study by Caramella et al., 3 of the 4 patients with pseudoprogression had new lesions in the chest (2 in the lung and one in the mediastinum) during PD-L1 inhibitor therapy [28]. Precise descriptions of new lesion location and its behavior during therapy are needed in future studies to further understand the phenomenon of pseudoprogression in the setting of new lesions, and to differentiate it from true progression for better treatment guidance.
Lesion-based response assessment resulted in notable differences in the response rate and the degree of tumor size decrease across anatomic organs of the lesions. Adrenal lesions and lymph nodes are more responsive with a greater tumor size decrease, while liver lesions were less responsive with less size decrease. Although the exact mechanisms of this observation are uncertain, it may indicate the impact of different tumor microenvironments in different organs on immune-related tumor response. The liver has an intrinsic immune suppressive microenvironment [29, 30], which may help tumors to escape from anti-tumor immune attacks during therapy, resulting in less tumor shrinkage. Lymph nodes as source organs of immune cells may be at least partly contribute to the observed greater response of lymph nodes as target lesions. Adrenal gland is one of the most common site of extrathoracic metastasis from NSCLC, and is also a major effector organ of hypothalamic–pituitary–adrenal (HPA) axis and is responsible for synthesis and action of cytokines [31, 32]. Therefore, the organ is known to have immune-modulating properties via activation of the HPA axis as well as via cell-cell mediated immune-adrenal interactions [31, 32]. The unique properties of the organ may be related to a greater degree of response in adrenal lesions.
To our knowledge, lesion-based immune-related responses across different organs have not been previously described in detail in NSCLC patients. Given the significant difference of responses across different organs noted in this study, lesion-based analysis may provide clinically significant information because the anatomic distribution of metastasis and selection of target lesions prior to therapy may affect the response outcome. Further investigations are needed to study the consistency of the observations in larger cohorts of NSCLC and to understand the underlying mechanisms. Notably, the results of a recent investigation by Ribas et al. in advanced or metastatic melanoma treated with pembrolizumab in phase 1b trials indicate that the response rate may be higher in patients with lung metastases (M1b disease) than in those with other visceral metastases (M1c disease) [33]. Another study in 337 melanoma patients treated with pembrolizumab or nivolumab demonstrated that patients with liver metastases were less likely to respond to treatment, while patients with lung metastases were more likely to respond [34]. These results and the observations in the present NSCLC study are similar in that patients with liver metastases are less likely experience tumor response, which is consistent with the immunosuppressive environment in the liver. The presence of lung lesions was not indicative of response in our cohort as opposed to the studies in melanoma, which is likely due to the presence of primary lung tumor in many of the advanced NSCLC patients. It is necessary to further investigate the similarities and differences of lesion-based responses across different tumor types, to understand if the differences are due to tumor microenvironment alone or are also related to the interactions between tumor and its microenvironment.
A limitation of the study includes the small number of patients treated at a single institution during the initial clinical experience studied with a retrospective design. However, such study design allowed us to report the evaluation of tumor response characteristics in NSCLC patients treated with commercially-prescribed nivolumab in a “real-world” setting, which has not been studied in detail despite an abundance of published data from prospective trials [4, 8, 9, 12]. Lack of molecular markers such as PD-L1 tissue staining in most patients in this clinical cohort limits the assessment of predictive values of such markers. Short follow-up time in this initial clinical experience also limits the evaluation of late responses and durable stability of the disease. These issues as well as the prognostic values of irRECIST1.1 assessments remain to be investigated with more mature follow-up data in a larger number of patients.