Search

2561 Background: Lymphocyte Activation Gene 3 (LAG3) or CD223 is an immune checkpoint that can be found on various T cells: CD4+, CD8+, regulatory T cells (Tregs), natural killer T cells, natural killer cells, and plasmacytoid dendritic cells. The expression of LAG3 molecule acts to increase T-cell exhaustion, leading to decreased tumor killing as well as an increase in immune suppressive cytokine release. Many clinical trials of LAG3 inhibitors have had modest effects, but recent data suggests that the LAG3 antibody relatlimab together with nivolumab (anti-PD1) provided greater benefit than nivolumab alone in patients with melanoma. Methods: The RNA expression levels of 397 genes in various types of solid tumors from 514 patients seen at the UCSD Moores Cancer Center were analyzed at a CLIA-licensed laboratory, OmniSeq (https://www.omniseq.com/). Following removal of germline variants, synonymous variants, indels and SNVs with < 5% VAF, TMB is reported as mutations/megabase. Transcript abundance was normalized to internal housekeeping gene profiles and ranked (0-100 percentile) in a standardized manner to a reference population of 735 tumors spanning 35 histologies. Odds ratio for high LAG3 expression was calculated and Bonferroni corrected for multiple genes and cancer histologies with > 40 samples. Results: A total of 116 (22.6%) tumors had high LAG3 (≥75) across 32 different histologies. Cancers with the highest proportion of LAG3 were neuroendocrine (47%), uterine (43%), sarcoma (33%), breast (31%), ovarian (30%), pancreatic (24%), lung (20%), stomach (16%), and colorectal (15%). There was significant association for high LAG3 with high PD-L1 (adj P < 0.001), high PD-1 (adj P < 0.0014), high PD-L2 (adj P < 0.0014), high CTLA-4 (adj P < 0.0014), TMB ≥10 mt/mb (adj P = 0.0504). There was no significant association between histologies colorectal (adj P = 0.1834), breast (adj P = NS), ovarian (adj P = NS), pancreatic (adj P = NS), or gender (adj P = 0.272). Conclusions: High LAG3 was found in almost a quarter of tumor samples and significantly associated with other immune checkpoints with FDA-approved drugs. Ongoing studies combining LAG3 inhibitors and specific immune checkpoint inhibitors may yield more clinical benefit if individualized immunomic transcript interrogation is undertaken, rather than population-based approaches without employment of rationally combined agents matched to each patient’s cancer.

2623 Background: Immune checkpoint inhibitors (ICIs) have emerged as effective treatments in non-small cell lung cancer (NSCLC). While the clinical utility of single agent ICI or in combination with chemotherapy has been well established, there remains an unmet need for the development of biomarkers that can better predict response. To address this need, we developed and applied a combination genomic and immune biomarker strategy to ICI-treated NSCLC patients which identified distinct patient subgroups with differential benefit among single agent or combination ICI treatment strategies. Methods: A discovery cohort (DC) of 5450 tumors across 37 histologies were evaluated by comprehensive genomic and immune profiling of the tumor immune microenvironment. Individual and combination biomarker assessment included PD-L1 IHC, TMB, tumor inflammation (TIGS), cell proliferation (CP) and cancer testis antigen burden (CTAB). From this cohort, combinations of molecular and immune biomarkers were identified and applied to a retrospective cohort (RC) of 225 metastatic NSCLC patients treated with pembrolizumab + chemo or pembrolizumab alone to correlate with response. Comparison of objective response rates (ORR) was performed using Chi-square test. Kaplan-Meir analysis was performed to test for differences in overall survival (OS) and 1-year OS. Results: Unsupervised analysis of the DC revealed four distinct biomarker combination groups that describe underlying tumor immunobiology: tumor dominant (CTAB, TMB, CP High), proliferative (CP High), inflamed (TIGS High), and checkpoint (PDL1, TIGS and TMB High). Application of these biomarker groups to the RC demonstrated significant differences in response to ICI regimens between groups (p = 0.04). Patients in the proliferative group (35.1%, 79/225; median PD-L1 = 20% TPS) treated with single agent pembrolizumab showed a significantly higher ORR (59%; 16/27) compared to pembrolizumab + chemo (27%; 14/52; p = 0.005), significantly improved 1-yr OS (p = 0.03), and trend towards better OS (p = 0.14). Importantly, patients in the inflamed group (16%, 36/225; median PD-L1 = 1% TPS), suggested that pembrolizumab + chemo (ORR 26.1%; 6/23) was not associated with ORR compared to pembrolizumab (ORR 31%; 4/13, p = 0.76), or OS (p = 0.37) and 1-yr OS (p = 0.57). Conclusions: Comprehensive genomic and immune profiling may identify PD-L1 low NSCLC patients who benefit from single agent pembrolizumab. PD-L1 low NSCLC patients with a proliferative phenotype may benefit from single agent pembrolizumab, whereas PD-L1 low cases with an inflamed phenotype may benefit from both single agent and combination pembrolizumab. Although further clinical validation of these predictive biomarker combinations is required, this data-driven approach demonstrates the potential to provide treatment decision support when selecting an ICI therapeutic strategy in lung cancer.

e21167 Background: CGIP analyzes FFPE tumor tissue by DNA/RNA sequencing for SNVs, indels, copy gain/loss, fusions, splice variants, MSI and TMB, along with PD-L1 IHC. A purported advantage of CGIP in NSCLC is the ability to identify targeted and immunotherapy biomarkers to inform clinical management. However, the extent to which CGIP supports treatment decisions and benefits NSCLC patients in various treatment settings is limited. Methods: A retrospective analysis of OmniSeq CGIP results (June 2017-March 2019) and real-world clinical data (through March 2020) for NSCLC patients (n = 300) was performed to evaluate treatment strategies at Roswell Park Comprehensive Cancer Center. Patient targeted and immunotherapies following CGIP were classified as “matched” to biomarker results (established or potentially clinically significant) at the indication level (single or multi-marker results, histology, treatment line) based on AMP/ASCO/CAP guidance for strength of biomarker clinical evidence. We estimated overall survival (OS) from CGIP report date for patients who first received either matched therapy or chemotherapy (and no subsequent matched therapy), and assessed the predictive value of matched therapy for OS in the first or subsequent line setting, adjusting for clinicopathologic covariates. Results: Most CGIP tested patients were female (55%), stage IIIB/IV (89%), ECOG < 2 (83%), non-squamous (86%), treatment naïve (62%), ever smokers (88%). 74% (228) of patients were treated post-CGIP, with 71% receiving at least one matched therapy. Matched therapies received in the frontline setting were supported by the highest (Tier 1A) category of evidence more often than subsequent line therapies (97% vs. 68%). 90% of patients with oncogenic driver mutations received targeted agents (17% of total) and 57% received matched immunotherapy. In the frontline setting, compared to chemotherapy, OS was highest for patients who first received matched targeted therapy (median = 23.4 mo; HR 0.26; p = .004; 95% CI 0.13-0.68) vs matched immunotherapy (median = 17.9 mo; HR 0.38; p = .001; 95% CI 0.21-0.69). Subsequent line, OS was also highest for patients who first received matched targeted therapy (median not est., mean = 27.5 mo; HR 0.20; p = .063; 95% CI 0.04-1.09) vs matched immunotherapy (median = 17.4 mo; HR 0.20; 95% CI 0.04-1.09), however, these differences were non-significant. Conclusions: CGIP supports evidence-based clinical decision making for NSCLC in the first and subsequent line settings and leads to improved survival for patients who receive matched targeted or immunotherapy compared to chemotherapy. Better predictive markers are needed to identify NSCLC patients who are more likely to respond to immunotherapies. Heterogeneity of patient biomarker profiles and treatment strategies over time in real world practice are a challenge to assessing CGIP efficacy.

2555 Background: Although immune checkpoint blockade (ICB) has revolutionized cancer treatment, not all patients with cancer benefit from ICB. One possible explanation for poor responders/resistance is the variable expression level of the target molecules (e.g., PD-1 and PD-L1) in the tumor microenvironment. There are recent or ongoing trials targeting variable pathways for immune evasion (e.g., LAG3 or IDO1). It is therefore of interest to know the expression levels related to variable immune checkpoints so that clinical trials can focus on the patients who can benefit from the cognate treatment. Methods: Overall, 514 patients with various solid tumors seen at the University of San Diego, Moores Center for Personalized Cancer Therapy were analyzed. The expression levels of checkpoint markers (ADORA2A, BTLA, CD276, CTLA4, IDO1, IDO2, LAG3, NOS2, PD-1, PD-L1, PD-L2, PVR, TIGIT, TIM3, VISTA, and VTCN) in the tumor samples were measured through RNA sequencing and normalized to internal housekeeping gene profiles, and ranked from 0 to 100 percentile based on a reference population. The expressions of each checkpoint marker were correlated with cancer types, microsatellite instability (MSI), tumor mutational burden (TMB), and programmed death-ligand 1 (PD-L1) status on immunohistochemistry. Results: In this cohort, 60% were female, median age of 60, and included 30 different tumor types, with colorectal cancer being the most common (27%). The rank values of all checkpoint markers were distributed broadly from 0 to 99 or 100. CD276 and NOS2 had the highest (68th percentile) and lowest (13.5 percentile) median rank values, respectively. When rank values were categorized to “Low” (0-24), “Intermediate” (25-74), and “High” (75-100), 41.6% of patients showed high expression of CD276 while only 13% showed high expression of PD-L1. Each patient had a distinctive protfolio of the categorical expression levels of 16 checkpoint markers. Several checkpoint markers, especially NOS2, showed a significant correlation with cancer type. (median rank values in colorectal, stomach, pancreatic, and breast cancer were 79, 76, 5 and 0 respectively, p < 0.001) Five markers (IDO1, LAG3, PD-1, PD-L1, and TIGIT) showed significant correlation with MSI, while seven markers (CTLA4, IDO1, LAG3, PD-1, PD-L1, PD-L2, and TIGIT) were significantly associated with positive PD-L1 status. However, no significant association was seen based on TMB or tissue-specific grouping of patients. Conclusions: The expression of immune checkpoint markers varies from patient to patient, though transcript expression of several markers correlates with cancer type, MSI, and PD-L1 status. Clinical trials with patient selection based on the expression level of checkpoint markers matched to the corresponding ICB drug are warranted.

8 Background: We performed microsatellite instability testing and expression immune profiling of endometrioid and serous carcinomas to identify immunotherapeutic targets and histologic attribute correlates. Methods: Microsatellite instability and RNA-seq of 395 immune-related genes were performed in 37 endometrioid and 53 serous carcinomas. Each gene was ranked against a reference population and subsequently compared to a separate clinical database reference of diverse neoplasms to perform Wilcoxon ranked sum test comparisons. Benjamin-Hochberg corrected p-values are reported. Results: VTCN1, IDO1, LAG3, and components of the mTOR/AKT1 pathway were overexpressed in both endometrioid and serous carcinomas compared to the clinical database population. CD276 and NT5E were uniquely overexpressed in the endometrioid set (Table). In the endometrioid set ADGRE5 was overexpressed in endometrioid primaries when compared to their metastatic counterparts. Serous metastases relatively overexpressed HLA-DPA1 compared to their primary counterparts. Ten of 26 endometrioid cancers with available microsatellite instability status were microsatellite unstable. All serous carcinomas were microsatellite stable. There was no correlation between microsatellite instability status and expression of any gene in either endometrioid or serous cancers. Conclusions: We identified high RNA-expression of VTCN1, IDO1, CD276, and LAG3 in endometrioid and serous carcinomas while NT5E (ADORA2A ligand) was found to be uniquely upregulated in endometrioid carcinomas. These markers may play a role in immune suppression in gynecological cancers and are potential immunotherapeutic targets. The upregulation of ADGRE5 in primary endometrioid cancers and upregulation of HLA-DPA1 in metastatic serous cancers suggests immunologic differences which may be relevant to the formation of metastatic disease.

e15024Background: Immunotherapy for PD-1 axis driven tumors, defined as tumors with high expression of PD-1, PD-L1, or PD-L2, are generally considered more responsive to checkpoint inhibitors than ...