Genomic biomarkers in relation to PD-1 checkpoint blockade response.
25 Background: Somatic tumor mutational burden (TMB) and a T-cell inflamed gene expression profile (GEP) predict response to anti-PD-1/PD-L1 immunotherapies in multiple tumor types. We assessed the potential for GEP and TMB to jointly predict clinical response to pembrolizumab and to identify distinct, targetable patterns of biology that may modulate response/resistance. Methods: To assess the individual and joint clinical utility of TMB and GEP in a pan-tumor context, pembrolizumab-treated patients with advanced solid tumors and melanoma were stratified as 4 biomarker-defined clinical response groups (GEP low/TMB low, GEP low/TMB high, GEP high/TMB low, GEP high/TMB high; N > 300) based on cutoffs for TMB (ROC Youden Index associated) and GEP (selected via analysis of pan cancer data). TMB and GEP were used to guide transcriptome and exome analysis of tumors in 2 large databases (Moffitt, n = 2944; TCGA, n = 6978). Results: TMB and GEP had a low, but significant, correlation in these clinical datasets. ORR was highest in GEP high/TMB high (37-57%), modest in GEP high/TMB low (12-35%) and GEP low/TMB high (11-42%), and lowest in GEP low/TMB low (0-9%) groups. Within the Moffitt and TCGA databases, GEP and TMB again had a low correlation, demonstrating their potential joint utility for stratifying additional transcriptomic and genomic features of these datasets. Specific gene modules showed strong positive or negative and highly statistically significant associations with TMB, GEP or both in each dataset, and patterns were consistent between datasets. In particular, gene set enrichment analysis identified proliferative, stromal and vascular biology corresponding to specific TMB-defined subgroups within GEP high tumors. In TMB-high tumors, indication-dependent somatic DNA alterations in key cancer driver genes showed a strong negative association ( P< 1e-5) with GEP. Conclusions: This analysis shows that TMB and T-cell inflamed GEP score can stratify human cancers into groups with different response rates to pembrolizumab monotherapy, and identify patterns of underlying, targetable biology related to these groups. This approach may provide a precision medicine framework for evaluating anti-PD-1/L1-based combination therapy regimens. Clinical trial information: NCT01848834; NCT02054806; NCT01295827; NCT01866319.