Objective: This study aimed to examine the pattern of recurrence in patients with endometrioid endometrial cancer and to identify clinically important prognostic factors in the recurrent population.

Methods: With institutional review board approval, a retrospective review identified 1061 patients who underwent primary surgery and treatment of endometrioid endometrial cancer at our institution from 1994 to 2007. Of this cohort, 77 (7.2%) patients developed a recurrence. Clinical factors were recorded, and Spearman correlation coefficients and χ² test were used to determine associations between groups. Kaplan-Meier survival estimates and Cox proportional-hazards model were used to determine how prognostic variables affected survival after recurrence (RS) and overall survival (OS).

Results: Of 77 patients, site of recurrence was not available in 5 patients. The distribution of recurrence in the remaining 72 patients was as follows: isolated vaginal 18% (13/72), nonvaginal pelvic 12% (9/72), distant 31% (22/72), abdominal 24% (17/72), and nodal 15% (11/72). There was an overrepresentation of advanced stage (P < 0.001) and high-grade (P < 0.003) at presentation in the recurrent group. Median OS was 3.4 years and median RS was 1.3 years. Low-grade tumors, early stage, and those with less than 50% myometrial invasion were associated with a significant OS and RS advantage. Age-adjusted isolated vaginal recurrence presented with a 1.2-year RS survival advantage (P < 0.03). An age-adjusted Cox proportional hazard ratio model incorporating significant prognostic variables demonstrated that the only independent variable associated with worse OS and RS was increased histologic grade with a hazard ratio of 2.31 (95% confidence interval, 1.25-3.97) for RS and 2.44 (95% confidence interval, 1.41-4.62) for OS.

Conclusions: Those patients with high-grade histology at the time of initial diagnosis manifest a decreased OS and RS, suggesting that the intrinsic biology of the tumor has the greatest prognostic importance.