Characterization of the immune tumor microenvironment (TME) to inform personalized medicine with immuno-oncology (IO) combinations

Abstract

Background: Enhanced understanding of the TME can enable precision medicine- driven patient (pt) selection to identify pts more likely to benefit from IO therapy (Hellman MD, et al. N Engl J Med 2018; Overman MJ, et al. Lancet Oncol 2017). Methods: The Cancer Genome Atlas (TCGA) RNA data (melanoma, NSCLC, RCC, UC, SCCHN, GEJ) were normalized and grouped as inflamed (INF), intermediate (INT), or non-INF. Binary associations of PD1 with IO targets (LAG3, IDO1, FOXP3, GITR, CSF1R, KIRDL1, CTLA4) were studied. Unsupervised clustering and pt-level gene expression profiling (GEP) were performed. A separate set of tumors was analyzed by IHC (N=228; LAG-3, IDO-1, FOXP3, GITR, CSF-1R, NKp46, PD-L1, MHCI, MHCII, CD8, CD68, CD163) and matched mRNA analysis (EdgeSeq; n=128). IHC results were integrated into an algorithm for IO combination selection. Results: TCGA analysis showed associations of IO targets and PD1 (mean Pearson r6 SEM =0.62±0.03; P<0.0001). Unsupervised clustering revealed discrete groups of INT tumors with high T-cell anergy, regulatory T cell, or myeloid signatures. Pt-level GEP showed INT/low-PDL1 tumors as most likely to have outlier IO targets suggestive of functional relevance. IHC showed clustering of IO targets by INF level, with outliers in INT/low-INF tumors and variability by tumor type. Observations were verified by selected IHC markers showing significant association of expression level and INF score: IDO-1, LAG-3 (non-INF vs INT P=0.17-1.9E-04; INT vs INF P=0.001-0.03); FOXP3, GITR, NKp46 (INT vs INF P=0.001-0.049). CSF-1R did not show significant associations. These data aided in the design of the ADaptiVe biomarker trial that InformS Evolution of therapy (ADVISE; NCT03335540), where prospective treatment selection (nivolumab+ second IO agent) is driven by analysis of pretreatment biopsies. Initial clinical implementation will also be presented. Conclusions: Translational data reveal potentially actionable biomarkers, which are being assessed in the ongoing ADVISE trial as an initial clinical foray into personalized IO therapy.