Validation of a targeted next-generation sequencing approach to detect mismatch repair deficiency in colorectal adenocarcinoma

Abstract

Mismatch repair protein deficiency is a hallmark of cancers associated with Lynch syndrome and is a biomarker for response to immunotherapy. With the increasing adoption of cancer next-generation sequencing, there has been a movement to develop screening approaches that take advantage of the unique mutational signatures of mismatch repair-deficient tumors. Here, we develop a sequencing-based metric that distinguishes mismatch repair-deficient from mismatch repair-proficient colorectal adenocarcinomas with comparison to immunohistochemical staining. We find that a single criterion of three or more single base pair insertion or deletion mutations per megabase sequenced, occurring in mononucleotide repeat regions of four or more nucleotides, is sufficient to detect mismatch repair deficiency with 96% sensitivity and 100% specificity in a training set of 241 cancers and 96% sensitivity and 99% specificity in a validation set of 436 additional cancers. Using data from the same cohort, we also find that sequencing information from only three genes-ARID1A, KMT2D, and SOX9-is sufficient to detect mismatch repair-deficient colorectal adenocarcinomas with 76% sensitivity and 98% specificity in the validation set. These findings support the notion that targeted next-generation sequencing already being performed for clinical or research purposes can also be used to accurately detect mismatch repair deficiency in colorectal adenocarcinomas.