A proteomic biomarker discovery platform for predicting clinical benefit of immunotherapy in advanced melanoma.

Abstract

10037Background: Immune checkpoint inhibitor-based immunotherapies that target CTLA-4 and the PD-1/PD-L1 axis have revolutionized the treatment of advanced melanoma due to their remarkable clinical benefit. However, only a limited number of patients respond to treatment. Therefore, biomarkers to identify appropriate candidates who will benefit from such therapy are needed. Our previous studies have identified therapy-induced, host-mediated mechanisms that drive resistance to a variety of cancer treatment modalities. Here, we explored whether assessing the systemic host-mediated response to immunotherapy can serve as a basis for predicting clinical outcome in melanoma patients. Methods: The cohort consisted of 34 advanced melanoma patients receiving anti-PD-1 monotherapy or anti-PD-1 and anti-CTLA-4 combination therapy. Clinical benefit was assessed. Plasma samples were obtained from patients at baseline and 2-4 weeks after a single treatment. Proteomic profiling of plasma samples was performed using ELISA-based protein arrays. A generalized linear model (GLM) was applied to a subset of the cohort (n = 13) to identify a proteomic signature that can predict clinical response to treatment. The predictive signature was then tested on the entire cohort (n = 33), excluding one patient with stable disease. Results: We identified a 10-protein signature that accurately distinguishes between responders and non-responders with an area under the curve (AUC) of 0.84 (confidence interval: 0.69-0.99, p-value 5.56E-04), and sensitivity and specificity of 0.94 and 0.79, respectively. These results are currently being validated in a larger cohort in an ongoing prospective study (PROPHETIC trial, NCT04056247). To explore the biological basis of resistance to immunotherapy, we performed a pathway enrichment analysis. Multiple mechanisms of resistance were identified in the non-responder group, including signaling pathways associated with immunosuppression and inflammation. Comparison between the two treatment modalities revealed pathways unique to each treatment, implying important differences between the two regimens. Conclusions: Our study provides insights into mechanisms of resistance to immunotherapy and paves the way towards the discovery of novel predictive biomarkers for patient stratification in melanoma.