Geospatial characterization of immune cell distributions and dynamics across the microenvironment in clear cell renal cell carcinoma

Abstract

IntroductionIn clear cell renal cell carcinoma (ccRCC), tumor-associated macrophage (TAM) induction of CD8+T cells into a terminally exhausted state has been implicated as a major mechanism of immunotherapy resistance, but a deeper biological understanding is necessary.MethodsPrimary ccRCC tumor samples were obtained from 97 patients between 2004 and 2018. Multiplex immunofluorescence using lymphoid and myeloid markers was performed in seven regions of interest per patient across three predefined zones, and geospatial analysis was performed using Ripley’s K analysis, a methodology adapted from ecology.ResultsClustering of CD163+M2 like TAMs into the stromal compartment at the tumor–stroma interface was associated with worse clinical stage (tumor/CD163+nK(75): stage I/II: 4.4 (IQR −0.5 to 5.1); stage III: 1.4 (IQR −0.3 to 3.5); stage IV: 0.6 (IQR −2.1 to 2.1); p=0.04 between stage I/II and stage IV), and worse overall survival (OS) and cancer-specific survival (CSS) (tumor/CD163+nK(75): median OS–hi=149 months, lo=86 months, false-discovery rate (FDR)-adj. Cox p<0.001; median CSS–hi=174 months, lo=85 months; FDR-adj. Cox p<0.001). An RNA-seq differential gene expression score was developed using this geospatial metric, and was externally validated in multiple independent cohorts of patients with ccRCC including: TCGA KIRC, and the IMmotion151, IMmotion150, and JAVELIN Renal 101 clinical trials. In addition, this CD163+ geospatial pattern was found to be associated with a higher TIM-3+ proportion of CD8+T cells, indicative of terminal exhaustion (tumor-core: 0.07 (IQR 0.04–0.14) vs 0.40 (IQR 0.15–0.66), p=0.05).ConclusionsGeospatial clustering of CD163+M2 like TAMs into the stromal compartment at the tumor–stromal interface was associated with poor clinical outcomes and CD8+T cell terminal exhaustion.