Prognostic Impact of Histologic Subtypes in Mismatch Repair‐Deficient/Microsatellite Instability‐High Colorectal Cancer: A Single‐Center Retrospective Study of 1127 Stage 0– <scp>IV</scp> Patients

ABSTRACT Background Deficient mismatch repair/microsatellite instability‐High (dMMR/MSI‐H) colorectal cancer (CRC) generally confers a favorable survival yet is paradoxically enriched for poorly differentiated (Por) histology, which is linked to aggressive behavior. We evaluated how MMR/MSI status and histological subtype relate to stage distribution and prognosis. Methods We conducted a single‐center retrospective study of 1127 Stage 0–IV CRCs. MMR/MSI status was assessed by immunohistochemistry (MMR‐IHC) and/or MSI testing. Clinicopathological features and relapse‐free survival (RFS) were analyzed, with survival analyses focused on 900 curatively resected Stages I–III cases. Results MMR‐IHC and MSI testing showed 98.4% concordance; 104 patients (9.2%) had dMMR/MSI‐H tumors. In proficient MMR/microsatellite stable (pMMR/MSS) CRC, well/moderately differentiated (Wel/Mod) histology predominated (~90%) across all stages. By contrast, dMMR/MSI‐H tumors exhibited a stage‐dependent histological shift, with Por histology increasing from 11.5% in Stage I to 60.0% in Stage IV. Among Stage I–III patients, dMMR/MSI‐H was associated with superior RFS compared with pMMR/MSS CRCs, with statistical significance confined to Stage II. In Stage III, Por histology was strongly associated with early relapse and remained an independent adverse factor in multivariable models. Notably, within Stage III dMMR/MSI‐H CRC, Por histology was associated with distinctly worse RFS, whereas no recurrences were observed among Wel/Mod cases. Conclusion While confirming established prognostic trends, this study highlights that dMMR/MSI‐H CRC is histologically heterogeneous. A Por subset, prevalent in advanced disease, appeared to attenuate the survival advantage and was associated with early recurrence in Stage III, although validation is required due to limited sample sizes in subgroups.