Pan-cancer landscape of oncogenic POLE and POLD1 alterations.

Abstract

2650 Background: POLE and POLD1 encode key enzymes for DNA proofreading. Oncogenic alterations in POLE and POLD1 associated with defective proofreading can lead to tumor hypermutation and increased sensitivity to immune checkpoint inhibition. Prior studies have identified enrichment of POLE and POLD1 alterations in colorectal and endometrial cancers. However, characterization of POLE and POLD1 alterations in the pan-cancer setting has been limited. Utilizing the American Association for Cancer Research (AACR) Project Genomics Evidence Neoplasia Information Exchange (GENIE) v19.0, we performed a comprehensive analysis of oncogenic POLE and POLD1 alterations across tumor types. Methods: The AACR GENIE v19.0 database was used to select tumor samples that were profiled for molecular alterations (somatic mutations, structural variants, copy number alterations) in POLE or POLD1. POLE and POLD1 alterations were annotated by OncoKB to determine if they were oncogenic. Tumor samples were analyzed for co-occurring molecular alterations. Categorical variables were compared using chi-square or Fisher exact tests, as appropriate. P-values were adjusted with Benjamini-Hochberg correction to control false discovery rate (significance for q<0.05). Results: Across 188,863 tumor samples that were profiled for POLE or POLD1, there were known oncogenic alterations in 786 samples (0.4%). Of note, there were 18,170 molecular alterations in POLE and POLD1 that were classified as variants of unknown significance at time of analysis. The most common tumor types with oncogenic alterations were endometrial (n=374 of 7573 samples profiled; 4.9%), colorectal (143/17821; 0.8%), glioma (49/12795; 0.4%), NSCLC (35/30542; 0.1%), ovarian (26/7728; 0.3%), breast (25/16813; 0.2%), bladder (25/6433; 0.4%), cancer of unknown primary (19/6544; 0.3%), melanoma (13/7965; 0.2%), and pancreatic (10/10369; 0.1%), and esophagogastric (9/6503, 0.1%). The most common mutations in POLE were V411L, E18K, P286R, S297F, and A456P. The most common mutations in POLD1 were R1016H, R1016C, R689W, L606M, and D402N. Oncogenic alterations in POLE and POLD1 were enriched in endometrial cancer (OR 22.8; q<0.001) and colorectal cancer (OR 2.14; q<0.001), with no other tumor types showing enrichment. Within the oncogenic POLE/POLD1 cohort, the most common co-occurring molecular alterations were in PIK3CA (n=561, 71.4% in altered vs 12.9% in unaltered, q<0.001), APC (n=526, 68.7% vs 11.8%, q<0.001), ATM (n=523, 67.8% vs 7.3%, q<0.001), and PTEN (n=523, 66.5% vs 8.7%, q<0.001). Conclusions: To our knowledge, this study represents the largest molecular analysis of oncogenic POLE and POLD1 alterations in a pan-cancer setting. These findings suggest that oncogenic POLE and POLD1 alterations may be rarely identified in a broad range of tumor types. Further studies are needed to classify the functional implication of many variants of unknown significance in POLE and POLD1.