Objective Although omentectomy is part of the staging and treatment of epithelial ovarian cancer (EOC), its performance in a patient with a grossly normal omentum—acknowledging its role in debulking gross tumor deposits—has never been definitively shown to improve survival.

Methods/Materials Using Surveillance, Epidemiology, and End Results data from 1998 to 2010, we identified patients with EOC and assessed their age, race, year of diagnosis, tumor grade, histologic subtype, International Federation of Gynecology and Obstetrics stage, lymph node dissection, nodal findings, and performance of omentectomy. We compared disease-specific survival (DSS) based on the presence or absence of omentectomy using log-rank univariate analysis, Cox multivariate analysis, and Kaplan-Meier survival curves.

Results A total of 20,975 patients with invasive EOC underwent surgical treatment. Initial univariate analysis indicated a lower mean DSS with performance of omentectomy. However, multivariate analysis demonstrated no significant association between DSS and performance of omentectomy (hazard ratio, 0.978; P = 0.506). The DSS was improved if lymphadenectomy was performed (hazard ratio, 0.60; P < 0.001). In recent years, there was a trend toward decreased performance of omentectomy.

To look specifically at patients without bulky omental disease, a subset analysis was done looking at patients with stage I-IIIA disease who had had lymphadenectomy performed. There were 5454 patients in the group who underwent an omentectomy and 2404 patients in the group who did not. No difference in DSS was seen between the groups based on performance of omentectomy (P = 0.89). However, the analysis was limited by the lack of Surveillance, Epidemiology, and End Results data on the extent of omentectomy, amount of residual disease, and adjuvant chemotherapy.

Conclusions In this analysis, performance of omentectomy in patients with EOC without bulky disease (≤stage IIIA) did not seem to confer improvement in survival. A randomized control trial would be needed to fully address this question.