Uterine corpus involvement as well as histologic type is an independent predictor of ovarian metastasis in uterine cervical cancer.

OBJECTIVE This study aimed to investigate the independent risk factors for ovarian metastasis in cervical cancer. METHODS Among 1,040 consecutive patients who underwent operation for cervical cancer at our institution from January 1998 to July 2007, a total of 625 patients had both ovaries removed during primary operation were retrospectively selected by medical records. In order to determine clinicopathological risk factors for ovarian metastasis, we analyzed patients' demographics, FIGO stage, and other pathologic findings. The Chi-square or Fisher's extract tests were used to compare any association of clinicopathologic variables with ovarian metastasis. For multivariate analysis, the log regression models were used to determine independent predictors for ovarian metastasis. RESULTS Overall, ovarian metastasis was detected in fourteen (2.2%) patients: two of 473 patients with squamous cell carcinoma (0.4%) and twelve of 151 patients with non-squamous cell carcinoma (7.9%), respectively (p<0.0001). Univariate analysis represents age (45 years: p=0.347), histologic types (squamous vs. non-squamous, p<0.0001), FIGO stages (IA1-IIA 4 cm, p=0.054), stromal invasion (1/2, p=0.788), lymph node metastasis (positive vs. negative, p=0.007), parametrium (involved vs. uninvolved, p=0.145), upper vagina (involved vs. uninvolved, p=0.003), uterine corpus (involved vs. uninvolved, p<0.0001), and margin status (involved vs. uninvolved, p=0.017). By multivariate analysis, uterine corpus involvement was the only independent risk factor for ovarian metastasis (p=0.008), in addition to histologic types (p<0.0001). CONCLUSION Based on our study, uterine involvement of cervical cancer is an independent predictor for ovarian metastasis, except histologic types. Ovarian preservation in cervical cancer may be safely performed only when no involvement of uterine corpus is present.