Detecting deficient DNA mismatch repair in stage II and III colon cancers.

Abstract

419 Background: Deficient DNA mismatch repair (MMR) results in microsatellite instability (MSI) that is detected in ∼15% of sporadic colon cancers. MMR status has been shown to provide prognostic and predictive information. We developed a model to predict MMR deficiency using clinically available data, and thereby facilitate the selection of patient tumors for MMR testing. Methods: Data were utilized from stage II and III colon carcinoma patients (n = 2016) who participated in 5-fluorouracil-based adjuvant studies (NCCTG, FFCD, NCIC, GIVIO, NSABP) and an Italian cohort. MMR status in tumors had been determined by MSI testing or by immunohistochemistry for hMLH1 and hMSH2 proteins. Logistic regression and a recursive partitioning and amalgamation analysis was used to identify factors (histologic grade, gender, tumor site, stage, age, lymph node status, T-stage) predictive of MMR status. Results: Of the cancers, 357 (17.7%) showed deficient MMR. Tumor site was the most important predictor of MMR status followed by histologic grade, then stage (II vs. III) and then gender. Distal tumors had a low likelihood of deficient MMR (5% rate overall), whereas proximal tumors had a greater likelihood of deficient MMR (30%). For patients with proximal tumors, the addition of histologic grade and stage increased the prediction of deficient MMR (Table). Using tumor site, histologic grade, and stage, the logistic regression model showed excellent discrimination (c-statistic = 0.80). Conclusions: Routine clinicopathological data can facilitate the identification of MMR deficient cases. Tumor site and histologic grade were the strongest predictors of MMR deficiency. Within proximal, poorly differentiated tumors, stage was highly predictive. These findings suggest that our model can assist in selecting sporadic colon cancers for MMR testing for use in clinical decision-making, especially for stage II patients. [Table: see text] [Table: see text]