Abstract B017: Single-cell proteogenomic profiling reveals immune cell networks in renal cell carcinoma

Introduction: Single-cell proteogenomic profiling has enabled the interrogation of complex milieus of cell types and their corresponding states that interact to regulate the antitumor immune response. While several landmark studies employing single-cell RNA sequencing (scRNAseq) have provided a detailed taxonomy of lymphoid and myeloid phenotypic states within the renal cell carcinoma (RCC) microenvironment, these studies have only been performed on small patient cohorts. As such, the ability to infer coordinated interactions between cellular phenotypes remains limited. Here, we leverage single-cell proteogenomic approaches to map immune cell phenotype co-occurrences across a heterogeneous patient cohort. Experimental procedures: We performed 5’ scRNAseq on 75 dissociated tumor samples from 59 patients undergoing partial or radical nephrectomy using the 10X genomics platform. Paired single-cell TCR/BCR sequencing (scTCR/BCRseq) and cytometry by time-of-flight (CyTOF) were additionally performed on 64 and 48 samples, respectively. Results: scRNAseq captured a total of ~350,000 high quality cells of RCC, stromal and immune (myeloid and lymphoid) origin, across a range of histologies (clear cell 69%, papillary 11%, chromophobe 13%, other 8%) and stages (I 53% II 7%, III 31%, IV 9%), with proportions of major cell types being consistent between scRNAseq and CyTOF. Interestingly, paired scTCRseq revealed a subset of patients (~40%) that displayed intratumoral clonally hyperexpanded TCRs (> 100 copies/clone), with CD8+ cells expressing an exhausted phenotype being most dominant. These samples were also co-enriched for CXCL13+ CD8+ T cells as well as proinflammatory CXCL9/10+ macrophages, a relationship we confirmed upon calculating pairwise correlations between all immune cell phenotype proportions across all samples. Notably, RCC cells from these samples expressed an antigen presentation meta-program identified by non-negative matrix factorization and pathway analysis. A “pro-inflammatory” cell network gene signature was derived and employed across the clear cell RCC (KIRC) cohort of TCGA and IMMotion 151 data sets, to reveal pro-inflammatory enriched RCCs displayed worse survival yet improved response to immunotherapy-based regimens, respectively. Conclusions: Our study provides the largest single-cell proteogenomic characterization of RCC to date, detailing co-occurring immune cell phenotypes across heterogenous patient tumor microenvironments. We leveraged t