Abstract 948: Epigenomic correlates of checkpoint blockade immunotherapy resistance

Abstract

Abstract The development of immune checkpoint inhibitors (ICI) has revolutionized treatment for numerous cancer types. However, a significant fraction (~45%) of patients treated with ICI fail to exhibit a sustainable clinical response to the drug. A number of previous studies have conducted genomic and transcriptomic profiling of patients treated with ICI, yielding a number of promising insights into the mechanisms underlying ICI resistance. However, to our knowledge, there has been no comprehensive epigenomic profiling study of ICI resistance in a clinical cohort. We hypothesize that epigenomic longitudinal profiling - conducted in parallel with transcriptomic and genomic profiling - will provide novel evidence to confirm existing hypotheses regarding ICI resistance mechanisms and potentially uncover novel regulatory pathways involved in ICI resistance. We chose to profile H3K27ac histone mark since it is a marker of active enhancers and promoters. We first optimized a high-throughput “chip-in-a-tip” H3K27ac ChIP-seq experimental pipeline for processing snap frozen melanoma samples. We applied this method along with RNA-seq and Whole Exome Sequencing (WES) to N=30 samples (N=13 pre-treatment, N=17 on/post treatment) from N=11 metastatic melanoma patients (N=5 responders/partial-responders, N=6 progressive disease) treated with anti-PD1. Differential H3K27ac binding & expression analysis between responders and non-responders significanlty implicates MHC class I peptide loading and processing, BAT3 complex, BCL3, JAK/STAT signaling, and NFKB pathways as correlates of ICI resistance. Motif analysis provides significant evidence for STAT1 and several novel binding motifs for mediating changes in H3K27ac binding between responders and non-responders. Joint epigenomic and expression differential analysis between pre-treatment and on/post-treatment samples suggests that a large number of pathways, including TCR signaling, Notch signaling, VEGF signaling, MAPK signaling, NFKB signaling, and IL-3 & IL-6 pathways are significantly changed as a result of ICI treatment. Finally, to address differential tumor purity and immune infiltration in our samples, we develop a Bayesian probabilistic model for ChIP-seq deconvolution model aimed at assigning putative cell-type identity to differential peaks. By leveraging immune ChIP-seq profiles from the Roadmap Epigenomics project and melanoma cell-line ChIP-seq from literature, along with estimates of tumor purity from matched RNA-seq/WES data, we are able to assign a putative cell-type-of-origin for peak clusters. We show in simulations that our model assigns reasonable estimates of cell-type identity. We then apply this method to our data to illustrate potential uses of this model in assisting us with interpreting our ChIP-seq results. Citation Format: Alvin H. Shi, Li-Lun Ho, Stuart Levine, Vinod Yadav, Jamie Cheah, Christian Soule, Dennie T. Frederick, David Liu, Genevieve Boland, Manolis Kellis. Epigenomic correlates of checkpoint blockade immunotherapy resistance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 948.