Correlation of pathogenic POLE mutations with clinical benefit to immune checkpoint inhibitor therapy.
Benjamin Garmezy, Jinesh Gheeya, Kyaw Zin Thein, Patrick G. Pilié, Wanlin Wang, Jordi Rodon Ahnert, Kenna Shaw, Funda Meric‐Bernstam, Timothy A. Yap
Abstract
3008 Background: Mutations in DNA polymerase epsilon ( POLE) may induce DNA replication errors, increasing neoantigen load and potentially enhancing clinical benefit to immune checkpoint inhibitors (ICI). We present a clinicopathologic analysis of patients (pts) with advanced cancers harboring POLE mutations and their response to ICI therapy at MD Anderson Cancer Center. Methods: We used targeted exome sequencing via CLIA-certified next generation sequencing assays to identify pts with POLE-aberrant tumors and their co-occurring mutations. The pathogenicity of each POLE mutation was annotated utilizing InterVar and ClinVar databases. Chi-square analysis was performed. Results: Tumors from 12,947 pts were analyzed and 448 (3.5%) pts had a mutation or copy number variation in POLE (3.5%), comparable to the TCGA PanCancer Atlas (4.0%). Clinical data were available for 293 pts; the most common cancers were colorectal (14.7%), non-small cell lung (13.7%), cholangiocarcinoma (13.3%) and melanoma (10.2%). There were 267 unique co-mutations, including KRAS (23.0%), ARID1A (21.5%), BRCA2 (18.7%), ATM (18.4%), CDKN2A (17.5%), BRAF (15.3%), EGFR (15.3%), ATRX (12.6%), CREBBP (11.7%), APC (11.3%), ATR (11.0%), BRCA1 (11.0%) and CDK12 (10.4%). POLE variants were annotated in all pts: pathogenic/likely pathogenic (n = 34, 11.6%), benign/likely benign (61, 20.8%), and variant of unknown significance (198, 67.6%). 104 (35.8%) of 293 pts with POLE mutations received PD-1/L1 inhibitors as monotherapy or in combination. 93 (88.4%) of 104 pts were evaluable for response: Radiological CR 4.3% (n = 4), PR 26.9% (n = 25), SD 22.6% (n = 21), PD 46.2% (n = 43), for a clinical benefit rate (CR + PR + SD) of 53.8%. Pathogenic status of POLE mutation was associated with clinical benefit to PD-1/L1 inhibitors (p = 0.04). TMB (p = 0.44), PD-L1 (p = 0.11), and MSI (p = 0.66) status were not associated with pathogenic status. MSI-H status was not over-represented in pts with ICI clinical benefit (p = 0.36). Conclusions: Pathogenic POLE mutations were associated with clinical benefit to ICI therapy. Further studies are warranted to validate POLE mutations as a predictive biomarker. Multiple co-occurring DNA damage response mutations were found, which may contribute to ICI clinical benefit.