Deficient Mismatch Repair Subtypes in Vietnamese Colorectal Cancer: Clinicopathologic Associations, Predictive Modeling, and IHC-PCR Concordance.

Microsatellite instability (MSI) caused by deficient mismatch repair (dMMR) influences prognosis and immunotherapy response in colorectal cancer. The dMMR phenotype includes MutL-deficient (dMutL) and MutS-deficient (dMutS) subtypes, which may differ in their clinicopathologic profiles, yet data from Southeast Asian populations remain limited. Our study aimed to investigate the associations between clinicopathological features with MSI/dMMR status and concordance in dMMR identification between immunohistochemistry (IHC) and polymerase-chain reaction (PCR) in Vietnamese patients. Two hundred and twelve colorectal cancer patients were included. Tumor tissues underwent hematoxylin and eosin staining, IHC staining (for four MMR proteins: MLH1, PMS2, MSH2, and MSH6), and PCR assay. Clinicopathological characteristics were collected and compared between dMMR tumors and their subtypes (dMutL and dMutS) with proficient MMR (pMMR). Firth's logistic regression was conducted to construct the model predicting each subtype of dMMR based on clinicopathological characteristics. A PCR assay was performed on selected cases, and the concordance in MMR results between IHC and PCR was examined. dMMR tumors accounted for 25.5% of total tumors. Compared to pMMR tumors, dMutL tumors showed differences in tumor location, vascular invasion, lymphovascular invasion, histopathologic grade, and pTNM stage, whereas dMutS tumors showed differences in tumor size and pTNM stage. MMR results between IHC and PCR showed high concordance (Kappa = 0.88, 95% CI: 0.78-0.97), with IHC demonstrating 100% sensitivity, 87.5% specificity, and 88.9% precision relative to PCR. The dMutL predictive model exhibited good discrimination and calibration, but the dMutS model exhibited poor performance. dMMR status, especially dMutL, might be associated with clinicopathologic characteristics, which might support decision-making in clinical practice. These findings require validation in larger cohorts but may inform clinical screening practices in resource-limited settings.