Immunotherapy clinical trials suggest predictive immune-related biomarkers are tissue specific.

Abstract

e14206 Background: RNA-based immune landscape signatures (RILS) portray the breadth of immune and stromal cell activity in the tumor microenvironment (TME). Their predictive and prognostic capability for patients with solid tumors and treated with various therapies has been reported. Methods: Single and multi-variable survival analysis, multi-variable linear models, publicly available data on clinical trials NCT00796445, NCT01621490, NCT02108652. Results: Using publicly available data, RILS were compared to each other and to tumor mutational burden (TMB) to determine their relative ability to identify responders/non-responders and estimate patient survival in a series of immunotherapy clinical trials. These immunotherapy trials include patients treated with a PD1 inhibitor, PD-L1 inhibitor or MAGE-A3 tumor antigen combined with immunostimulant (MAGE-A3). Although no single RILS/TMB biomarker emerged as a clear indicator of immunotherapy response and survival for all cases, there were certain immune-related signatures that recurred as significant (p < .05) or had near-significance (p < .10) related to a specific cancer or therapy. Immune signatures for T cell trafficking, ratios of PD1 to PD-L1 expression, cytotoxic lymphocytes and B cells were consistently predictive and prognostic in melanoma patients treated with the PD1 inhibitor, nivolumab or MAGE-A3. However, while these same signatures were not predictive/prognostic in bladder and kidney/ureter cancer patients treated with the PD-L1 inhibitor, atezolizumab, biomarkers significantly associated with outcome included TMB, TGFB1, and the ratio of T cell trafficking to M2 tumor associated macrophage activity. Since TMB and various immune signatures generally worked better together than individually in predicting outcome, there is potential for utilization of a combination of biomarkers and immune signatures in the design of effective immunotherapy clinical trials. Conclusions: In summary, our results suggest effective immunotherapy trials may require tissue-specific biomarker development. Further, a combination of immune signatures and biomarkers such as TMB shows markedly increased utility compared to single factor such as TMB, PD-1 or PD-L1 expression alone.