Histologic Features Do Not Reliably Predict Mismatch Repair Protein Deficiency in Colorectal Carcinoma: The Results of a 5-Year Prospective Evaluation

    loading  Checking for direct PDF access through Ovid

Abstract

Most major professional medical organizations advocate universal screening for Lynch syndrome in colorectal carcinoma; however, some allow for a selective screening approach based on clinicopathologic factors including assessment of histologic features of mismatch repair protein deficiency (MMRD). We performed a prospective evaluation for histopathologic features of MMRD in colorectal carcinomas that underwent universal screening for Lynch syndrome to evaluate the ability of histology to predict MMRD. In total, 947 resected colorectal carcinomas over a 5-year period were prospectively analyzed for histologic features of MMRD and for DNA mismatch repair protein abnormalities. Histologic features of MMRD were reported as present in 281 of 947 (30%) tumors with only 109 (39%) cases demonstrating MMRD by immunohistochemistry. Histologic features of MMRD had a sensitivity of 74% [95% confidence interval (CI), 66%-80%], specificity of 78% (95% CI, 75%-81%), positive predictive value of 39% (95% CI, 32%-44%), and negative predictive value of 94% (95% CI, 92%-96%). Histologic features of MMRD in left colon/rectal tumors had a significantly lower sensitivity of 56% (95% CI, 41%-77%) compared with right colon tumors (P=0.02). Histologic rereview identified that tumor-infiltrating lymphocytes (TILs) were most likely to be incorrectly reported as absent, and 72% of cases incorrectly assessed as lacking TILs demonstrated MMRD by immunohistochemistry. We demonstrate that histologic features of MMRD do not reliably predict the presence of MMRD by immunohistochemistry. Interpretative errors in the assessment of histologic features of MMRD occur, particularly for TILs and in tumors of the left colon/rectum.

Related Topics

    loading  Loading Related Articles