Endometrial stromal sarcoma: analysis of treatment failures and survival.

The objective of this retrospective multicenter study was to assess the rates, times, and sites of recurrences of 26 patients with low-grade endometrial stromal sarcomas (ESS) (<10 mitoses per 10 high-power fields (HPF) and 40 patients with high-grade ESS (>10 mitoses/10 HPF). Surgery was the initial therapy for all patients. Postoperative treatment was given without well-defined protocols. The median follow-up of survivors was 92 months (range, 4-167). Low-grade ESS: Of the 20 patients with disease confined to the uterus, 5 (25.0%) developed pelvic recurrence after a median of 36 months (range, 4-108). Of the 6 patients with disease outside the uterus, only one recurred in the pelvis after 93 months. High-grade ESS: Of the 20 patients with disease confined to the uterus, 11 (55.0%) developed recurrent disease after a median of 5 months (range, 2-76). The relapse occurred in the pelvis in 3 patients, in upper abdomen in 3, in upper abdomen and extraabdominal sites in 1, and in the pelvis and upper abdomen or extraabdominal sites in 4. Of the 12 patients with extrauterine disease confined to the pelvis, 9 (75.0%) developed recurrent disease after a median of 12 months (range, 1-49). The relapse occurred in the pelvis in 3 patients, in extraabdominal sites in 3, and in the pelvis and upper abdomen or extraabdominal sites in 3. Of the 8 patients with extrauterine disease outside the pelvis, 7 died of disease and 1 is currently alive with progressive disease after 24 months. The disease-free survival was significantly better for low-grade than that for high-grade ESS (P = 0.0001). By log-rank test the disease-free survival of high-grade ESS patients was related to stage (P = 0.0466) and mitotic count (P = 0.0014), but not to age. Cox model showed that mitotic count was the only independent prognostic variable for high-grade ESS (P = 0.006). In conclusion, low-grade and high-grade ESS have a completely different biological aggressiveness and clinical behavior.