OBJECTIVETo assess the value of computed tomography( CT) in the staging and predicting respectability of primary advanced ovarian carcinoma.

METHODSThe data of preoperative abdomen and pelvis CT scan in 64 women with Stage II or IV ovarian carcinoma were collected from tumor registry database. All CT scans were analyzed retrospectively without knowledge of the operative findings, and the stage as based on CT was compared with the surgical and pathological findings. Residual lesion of < or = 2 cm in maximal diameter was considered as an optimal surgical result. Twenty-senven of these 64 patients (42.2%) underwent optimal cytoreduction surgery for residual disease C2 cm in diameter. Based on the ability of each parameter in predicting cytoreductive surgery outcome, 11 radiographic features were selected for the final model. Each predictive parameter was assigned a numeric value (1 to 7). Sensitivity, specificity, positive predictive value( PPV) , negative predictive value( NPV),and accuracy were calculated for each predictive parameter. Receiver operating characteristic( ROC) curve was used to assess the ability of the model to predict surgical outcome. The correlation between CT stage and surgical-pathologic stage was analyzed by Chi-square test and Spearman's rho analysis.

RESULTSThe overall accuracy of CT staging for advanced ovarian carcinoma was 87. 5% ; 86. 5% and 91.7% for stage III and IV patients respectively. The correlation between CT stage and surgicopathologic stage was found to be comformable. In the final predictive index model, when a predictive index scoreed > or = 2, the overall accuracy, sensitivity and specificity was 70. 3% , 67.6% and 74. 1% for identifying patients for suboptimal surgery. The PPV and the NPV was 78. 1% and 62. 5% , respectively. The ROC curve was generated with an area under the curve = 0. 792+/-0. 055 using the predictive index scores.

CONCLUSIONCT has a high accuracy in staging and a moderate ability to predict resectability for advanced ovarian carcinoma. Therefore, the predictive index model may be useful in the management of ovarian carcinoma patients.