A clinicopathologic multivariate analysis affecting recurrence of borderline ovarian tumors.

OBJECTIVE To evaluate the risk factors associated with recurrence of borderline ovarian tumors that may be used as evidence of the efficacy of select preventive procedures. METHODS Various clinicopathologic factors of 234 patients with borderline ovarian tumors admitted to our hospital between January 2001 and June 2007 were reviewed. Univariate and multivariate logistic regression models were constructed to evaluate the risk factors for odds ratio (OR) and statistical significance. The survival was assessed by the Kaplan-Meier method and proportional hazards model. RESULTS Recurrence of borderline ovarian tumors was observed in 26 cases and the median time to recurrence was 29.4 months. Of these cases, 5 occurred involving the ipsilateral ovary, 9 involved the contralateral ovary, and 12 spread to the pelvic peritoneum, including 3 patients who had progressed to invasive carcinoma. No tumor-related deaths were reported. The results of the multivariate logistic regression analysis showed that conservative surgical procedures (OR=2.304; p=0.024), cyst rupture (OR=2.213; p=0.038), advanced FIGO stage (OR=4.114; p=0.000), microinvasion (OR=2.291; p=0.046), and peritoneal implants (OR=2.101; p=0.016) may be independent predictive factors of recurrence. The proportional hazards model identified surgical procedure (relative risk, RR=3.752, p=0.007), cyst rupture (RR=1.985, p=0.006), FIGO stage (RR=3.746, p=0.001), microinvasion (RR=1.153, p=0.009) and peritoneal implants (RR=2.742, p=0.010), as independently related to disease-free survival. CONCLUSIONS Although patients with borderline ovarian tumors have an excellent prognosis, the risk of recurrence remains. Identification of patients with high-risk factors is essential for offering more selective treatments to prevent recurrence.