48Immunotherapy response correlates in metastatic clear cell renal cell carcinoma from circulating cell-free epigenomes

Abstract Background Immune checkpoint inhibitors (ICIs) are a mainstay of treatment for metastatic clear cell renal cell carcinoma (mccRCC). With a growing number of treatment options for mccRCC, biomarkers are needed to guide treatment selection by predicting benefit from ICIs. Cell-free DNA (cfDNA)-based assays that measure tumor DNA in plasma are promising biomarkers for treatment response in several cancers, but their role in RCC remains unclear. Here, we tested whether epigenomic profiling of cell-free DNA can identify or predict radiographic responses to immunotherapy in mccRCC. Methods Plasma samples were obtained from 180 patients (400 total samples collected, up to 4 samples per patient) with mccRCC enrolled in the DOD-funded Kidney Cancer Research Consortium ctDNA study (NCT04883827). Patients received an immune checkpoint inhibitor and were categorized as non-responders or responders based on CT scan results obtained at 3- or 6-months following treatment initiation. Cell-free chromatin immunoprecipitation sequencing (cfChIP-seq) was performed on patient plasma to profile H3K4me3, a histone modification associated with active gene promoters. DNA methylation was profiled with cell-free methylated DNA immunoprecipitation sequencing (cfMeDIP-seq). H3K4me3 signal at the promoters of ccRCC-associated genes variation between pre-treatment and on-treatment timepoints was assessed for non-responders and responders. We calculated changes in H3K4me3 signal for on- and pre-treatment plasma draws. We then compared these changes between non-responders and responders using the Wilcoxon sum-rank test. The immune cell types contributing to cfDNA were inferred from cfMeDIP-seq data. Results Comparing pre- and on-treatment timepoints drawn between 1-5 months after starting therapy, cfDNA promoter H3K4me3 at RCC-associated genes was concordant with radiographic response in 10 of 11 patients (91%), with signal increasing from pre- to on-treatment samples in 6 of 7 non-responders (86%) and decreasing in 4 of 4 responders (100%). The median change in H3K4me3 differed significantly for non-responders and responders (P = .02). Analysis of cfDNA methylation profiles identified a higher proportion of memory CD8 T cell-derived cfDNA in non-responders compared to responders in pre-treatment samples (P = .003). Conclusions This pilot study suggests that epigenomic profiling of cfDNA may identify pre-treatment and on-treatment biomarkers of ICI response. Profiling of larger cohorts is underway to test the generalizability of these findings.