Pathologic stage III endometrial carcinoma. Prognostic factors and patterns of recurrence.

Abstract

This review was done to assess the outcomes and patterns of recurrence in a group of patients with Stage III endometrial carcinoma that might help guide adjuvant therapy. A retrospective review was performed of 105 patients treated from 1970-1990 at three institutions. All patients underwent abdominal hysterectomy, with 60 having pathologic node assessment and 45 having cytologic examination of peritoneal washings. A single extrauterine site was involved in 75% of patients; 20% and 5% had two and three sites involved, respectively. All patients received postoperative external beam irradiation to the pelvis or pelvis and paraaortic regions for pathologically positive paraaortic nodes. Actuarial techniques were used to estimate the survival and recurrence rates. The 5-year disease-free survival rate for all patients was 64%. Univariate analysis revealed that the depth of myometrial penetration, the clear cell or papillary serous pathologic type, the histologic grade, and the number of extrauterine sites predicted disease-free survival. Cox regression revealed the grade and pathologic findings to be independent predictors of disease-free survival. The overall 5-year pelvic recurrence rate was 21%, with multivariate analysis revealing the grade to be the strongest prognostic factor. Pathologic findings and the number of involved extrauterine sites were the most important prognostic factors for abdominal recurrence and other sites of distant relapse. The subgroup of patients with low-grade endometrial tumors or superficial myometrial penetration has a low distant relapse rate. Local control remains the goal of therapy in these patients. Patients with high-grade tumors, deep myometrial penetration, clear cell or papillary serous histologic types, or two or more involved extrauterine sites are at high risk for distant recurrence that may include the abdomen. Investigative strategies delivering aggressive adjuvant therapy are appropriate.