Pooling Functional Genetic Screens to Identify Genetic Correlates of Tumor Immunotherapy Response

Abstract

Allison E. B. Chang and Eliezer Van Allen Purpose: Recent advances in immunotherapy have resulted in unprecedented breakthroughs in cancer treatment, with previously devastating diseases like metastatic melanoma seeing substantial increases in expected survival. However, only some patients respond to immunotherapy, while others are either completely unresponsive or later develop resistance. Ongoing work has been directed at determining which patients are the best candidates for immunotherapy and, moreover, whether treatments can be developed to “rescue” immunotherapy response in patients who are resistant. So far, this work has uncovered several genetic correlates of immunotherapy response, which have provided some clues as to the fundamental mechanisms underlying the tumor immune response—especially the interferon-γ (IFN-γ), tumor necrosis factor (TNF), NF-kB, and antigen presentation pathways. But questions remain as to what might be actionable genetic biomarkers that can help stratify patients, guide treatment decisions, and develop new immunotherapy drugs. Part of the challenge lies in the fact that it can be difficult to discriminate signal from noise among genetic data, and many studies have used different experimental approaches to genetic screens, making it unclear how to compare their results. We carried out the first pooled analysis of in vitro and in vivo CRISPR screens in an effort to identify and prioritize genetic correlates of tumor immunotherapy response. Methods: We pooled extended data from four in vitro and in vivo CRISPR screens to identify genes that were significantly altered following exposure to immune pressure across multiple studies. Results: We identified 10 genes that were significantly enriched (i.e., candidate genes conferring sensitivity to immune pressure) following exposure to immune therapy in 3 or more CRISPR data sets, and we found 16 genes that were significantly depleted (i.e., candidate genes conferring resistance to immune pressure) following exposure to immune therapy in 3 or more CRISPR data sets. Several of these genes belong to pathways already implicated in immunotherapy response (e.g., NF-kB, IFN- γ, and antigen presentation pathways), but many others belong to pathways with less well known functions—including two candidate genes conferring sensitivity to immune pressure (Nf1 and Tjap1) and ten candidate genes conferring resistance to immune pressure (Calr, Ube2n, Ccs, Pigu, Gpaa1, Maea, Atg5, Larp4, Vps4b, Fitm2). Conclusions: Our results confirm and extend previous findings, emphasizing the importance of known immune pathways and also identifying several promising novel candidate genes that could represent new patient biomarkers or drug targets.