A Proposal for a Classification for Recurrent Endometrial Cancer: Analysis of a French Multicenter Database From the FRANCOGYN Study Group.

OBJECTIVE: Endometrial cancer (EC) recurrences are relatively common with no standardized way of describing them. We propose a new classification for them called locoregional, nodal, metastasis, carcinomatosis recurrences (rLMNC). PATIENTS AND METHODS: The data of 1230 women with EC who were initially treated by primary surgery were included in this French multicenter retrospective study. Recurrences were classified based on dissemination pathways: (1) locoregional recurrence (rL); (2) nodal recurrence (rN) for lymphatic pathway; (3) distant organ recurrence (rM) for hematogenous pathway; and (4) carcinomatosis recurrence (rC) for peritoneal pathway. These pathways were further divided into subgroups. We compared recurrence free survival and overall survival (OS) between the 4 groups (rL/rN/rM/rC). RESULTS: The median follow-up was 35.6 months (range, 1.70-167.60). One hundred ninety-eight women (18.2%) experienced a recurrence: 150 (75.8%) experienced a single-pathway recurrence and 48 (24.2%) a multiple-pathway recurrence. The 5-year OS was 34.1% (95% confidence interval [CI], 27.02%-43.1%), and the median time to first recurrence was 18.9 months (range, 0-152 months). The median survival after recurrence was 14.8 months (95% CI, 11.7-18.8). Among women with single pathway of recurrence, a difference in 5-year OS (P < 0.001) and survival after recurrence (P < 0.01) was found between the 4 rLNMC groups. The carcinomatosis group had the worst prognosis compared with other single recurrence pathways. Women with multiple recurrences had poorer 5-year OS (P < 0.001) and survival after recurrence (P < 0.01) than those with single metastasis recurrence, other than women with peritoneal carcinomatosis. CONCLUSIONS: This easy-to-use and intuitive classification may be helpful to define EC recurrence risk groups and develop guidelines for the management of recurrence. Its prognosis value could also be a tool to select homogenous populations for further trials.