Tumor PD-L1 expression and gene panel mutational profile as outcome predictors of PD-1-based checkpoint inhibition therapy in metastatic melanoma: A prospective multicenter DeCOG study.

Abstract

9568 Background: PD-1 checkpoint inhibition (CPI) has recently advanced to one of the most effective treatment strategies in melanoma. However, since a considerable proportion of patients shows upfront therapy resistance, baseline predictive biomarkers of therapy outcome are needed. Methods: This prospective multicenter study included metastatic melanoma patients whose formalin-fixed paraffin-embedded tumor tissue samples taken prior to the start of a systemic non-adjuvant therapy of any line were analyzed for PD-L1 expression on tumor cells by immunohistochemistry (clone 28-8, DAKO) and for COSMIC-annotated oncogenic mutations by 29-gene panel sequencing (MiSeq, Illumina). Clinical baseline and follow-up data were collected within the DeCOG multicenter skin cancer registry ADOREG. Results: From 09/2015 until 10/2020, 706 enrolled patients from 15 centers were evaluable for the endpoints best overall response (BOR), progression-free (PFS) and overall survival (OS). Thereof, 540 patients received PD-1-based CPI as first systemic treatment after tumor tissue analysis. 197/540 patients tested positive for PD-L1 (cut-off = 5%) in pre-treatment tumors, and revealed a favourable BOR (objective response 34.4% versus 19.1%; p < 0.0001), PFS (median 10.4 versus 4.2 months; p < 0.0001) and OS (median 45.1 versus 18.8 months; p = 0.001) compared to patients with PD-L1 negative tumors. 47/540 patients presented oncogenic mutations of three or more genes in pre-treatment tumors, and revealed a favourable BOR (objective response 46.8% versus 32.1%; p = 0.041), PFS (median 15.1 versus 6.1 months; p = 0.008) and OS (median not reached versus 25.2 months; p = 0.027) compared to patients whose tumors showed mutations in two or less genes. Multivariable Cox regression including sex, primary site, non-adjuvant systemic pre-treatment, serum LDH, and ECOG performance state demonstrated tumor PD-L1 expression and gene panel mutational profile as independent predictors of survival upon treatment with PD-1-based CPI. In contrast, in 106/706 patients treated with BRAF/MEK inhibitors as first systemic treatment after tumor tissue analysis, no association was found between tumor PD-L1 expression or gene panel mutational profile and therapy outcome. Conclusions: PD-L1 expression quantification and gene panel mutational profiling provide useful outcome predictors of PD-1-based CPI therapy in metastatic melanoma patients.