Mismatch repair-deficiency specifically predicts recurrence of atypical endometrial hyperplasia and early endometrial carcinoma after conservative treatment: a multi-center study

Objective: Deficient expression of mismatch repair proteins (MMR) has been suggested to be a predictor of resistance of atypical endometrial hyperplasia (AEH) and early endometrial carcinoma (EEC) to conservative treatment. Aims: To assess the predictive value of MMR immunohistochemistry in patients conservatively treated for AEH and EEC, and to calculate its predictive accuracy. Materials and methods: All patients with AEH or EEC conservatively treated with hysteroscopic resection plus progestins in two referral centers from January 2004 to July 2019 were retrospectively assessed. Immunohistochemistry for MMR was ad hoc performed. Study outcomes were: (i) the association of a deficient immunohistochemical expression of MMR with resistance and recurrence of AEH and EEC after conservative treatment, and (ii) the accuracy of MMR immunohistochemistry in predicting the outcome of conservative treatment. Relative risk (RR) for the associations, and sensitivity, specificity and area under the curve (AUC) on receiver operating characteristic curve for the predictive accuracy were calculated. Results: Sixty-nine women, (47 AEH and 22 EEC) were included; deficient MMR expression was observed in 8.7% of cases. Resistance to conservative treatment was more common in MMR-deficient than MMR-proficient cases (33.3% vs 15.9%; RR = 2.1), but with no statistical significance (p = 0.2508). On the other hand, recurrence was significantly more common in MMR-deficient than MMR-proficient cases (100% vs 26.4%; RR = 3.8; p < 0.0001). In predicting recurrence, a deficient immunohistochemical expression of MMR showed sensitivity = 22.2%, specificity = 100%, and AUC = 0.61. Conclusion: Deficient MMR immunohistochemical expression does not imply resistance of AEH/EEC to conservative treatment. On the other hand, MMR-deficiency appears as a highly specific predictor of recurrence of AEH/EEC after initial regression.