44 Liquid biopsy epigenomic profiling for the detection of sarcomatoid renal cell carcinoma

Abstract Background Sarcomatoid differentiation (SD) in renal cell carcinoma (RCC) is associated with poor survival and heightened response to immune checkpoint blockade. Detection of SD can be challenging due to spatial heterogeneity and sampling error. Herein, we introduce a novel tissue–informed epigenomic approach to noninvasively identify sarcomatoid differentiation in patients with RCC from cell-free DNA (cfDNA) using 1mL of plasma. Methods Chromatin immunoprecipitation and sequencing (ChIP-seq) for H3K27ac – a histone modification associated with active regulatory elements (REs) – was performed on pathologically reviewed clear cell RCC frozen tissue samples with and without SD (sarcomatoid-RCC and epithelioid-RCC, resp.) collected at the Dana-Farber Cancer Institute. Differentially marked REs between sarcomatoid and epithelioid subtypes were identified using DESeq2 (false discovery rate of q < 0.01). After establishing tissue signatures, ChIP-seq was then performed on cell-free chromatin (cfChIP-seq) in plasma from patients with sarc-RCC and epi-RCC. A Sarcomatoid Score was derived for each sample by aggregating the plasma H3K27ac signal at tissue-derived sarcomatoid-specific REs (sarc-REs), while normalizing to signal at epithelioid-specific REs (epi-REs). Scores were compared between the two groups using a Wilcoxon rank-sum test. A classifier was built to distinguish sarc-RCC from epi-RCC based on the Sarcomatoid Score and its performance was evaluated using the area under the receiver operating characteristic (AUROC) curve. Results We identified 25,919 differentially marked REs between 8 sarc-RCC and 8 epi-RCC tissue samples at a false discovery rate of q < 0.01. We selected 12,868 REs that are enriched in sarcomatoid vs. epithelioid. We generated cfChIP-seq profiles from plasma of 29 patients, 17 with sarc-RCC and 12 with epi-RCC. The Sarcomatoid Scores were significantly higher in sarc-RCC vs. epi-RCC plasma samples (p=1.6×10-4; Figure 1A). These scores achieved an AUROC curve of 0.9 for classifying patients with sarc-RCC from patients with epi-RCC (Figure 1B). Conclusions We present a proof-of-concept study in 1 cc of plasma for the detection of sarcomatoid differentiation in RCC based on the assessment of histone modification signals in cfDNA. This approach could help overcome the challenges of spatial heterogeneity and sampling error from tissue that make identification of sarc-RCC difficult. More generally, it establishes a paradigm for identifying histologic subtypes of cancer based on their epigenomic correlates from cfDNA, with possible therapeutic implications in real-time. Figure 1. (A) Sarcomatoid Score in plasma from patients with RCC at sarcomatoid-specific REs, comparing sarcomatoid (orange) and epithelioid RCC (blue). (B) ROC curves for distinguishing sarcomatoid from epithelioid RCC plasma samples using H3K27ac cell-free ChIP-seq signal at sarcomatoid-specific REs. ‘AUC’ indicates area under the ROC curve.