Abstract Immune checkpoint inhibitors are fast becoming a key therapy in the medical oncologist's toolbox across a diverse array of tumour types. The excitement surrounding the utility of these emerging agents is marred by the heterogeneous response observed among patients, highlighting the urgent need for clinically applicable predictive biomarkers. Significant progress has been made in identifying potentially relevant biomarkers including mutation burden and immune infiltration. However, it is still unclear which biomarkers are most predictive across various disease types, and how to integrate various markers of response and resistance for clinical interpretation. We hypothesized that comprehensive genomic profiling, combined with clinical data, would reveal differences in the immuno-oncologic phenotype that could be used as biomarkers for response to checkpoint inhibitor therapy. The Personalized OncoGenomics (POG) study at BC Cancer performs whole genome and transcriptome sequencing of metastatic disease across a diverse array of cancer types to comprehensively characterize cancers and inform clinical therapeutic decision-making. Here we performed in-depth genomic profiling of 64 POG patients who received immune checkpoint inhibitors encompassing multiple disease types including skin, lung, breast, pancreatic and colorectal cancers and sarcomas. Single nucleotide variants, structural variants, copy number alterations, and RNA expression derived from whole genomic and transcriptomic data were used to characterize genomic instability, neoantigen landscape, immune infiltration, and other potential biomarkers of clinical response to checkpoint inhibitors. This collection of putative immune-oncologic biomarkers were integrated into a multivariate model to stratify markers of the tumour immune response. Our analysis shows limited association between PD-L1 expression and checkpoint inhibitor response, consistent with the inadequate effectiveness of this as a universal marker, and shows stronger association between signatures of T cell infiltration based on RNA-Seq data. We also observed patients with low mutation burden that respond to checkpoint inhibitors and show high levels of immune infiltration, and cases with high mutation burden that harbour mechanisms of resistance including reduction in predicted neoantigen diversity. Additional mechanisms of therapeutic resistance observed in post treatment biopsies highlight disruption of antigen presentation and JAK1 mutations in resistant tumours. Our study helps to define which features distinguish patients most likely to respond to checkpoint inhibitors, and uses these features in selecting patients for immunotherapy treatment within the POG clinical trial. Citation Format: Hillary Pearson, Laura Williamson, Erin Pleasance, Scott Brown, Emma Titmuss, Martin Jones, Stuart Zong, Payal Sipahimalani, Yussanne Ma, Steve Jones, Robert Holt, Marco Marra, Janessa Laskin. Integrated whole genome profiling of the immune tumour interaction identifies predictive biomarkers of checkpoint inhibitor response in metastatic cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4346.
Read full abstract