Prediction of disease persistence after conization for microinvasive cervical carcinoma and cervical intraepithelial neoplasia grade 3.

We attempted to determine the significant variables and to predict the probability of disease persistence after conization for microinvasive cervical carcinoma and cervical intraepithelial neoplasia grade 3 (CIN3). We analyzed 133 patients from 2001 to 2002 who had a subsequent hysterectomy after conization. The histological findings of the cone specimens, together with the clinical parameters, were correlated with the presence of residual dysplasia in the hysterectomy specimen. The probability of having residual dysplasia was calculated based on the function of the significant variables obtained by logistic regression analysis. Of the 133 patients, 42 (31.6%) had residual disease in their hysterectomy specimens. Using multivariate analysis only for the postmenopausal state, positive endocervical curettage, positive margin, and microinvasive carcinoma were predictive of residual dysplasia. The probabilities of having residual dysplasia were about 0.99, 0.84, 0.4, 0.07, and 0.01 in patients with a presence of all four, any three, any two, any one, and no risk factors, respectively. The best cutoff probability determined by the receiver operating characteristic curve was 0.32, yielding a sensitivity of 81% and a specificity of 88%. Based on these results, patients with the presence of any two or more of the risk factors mentioned above should be considered as a high-risk group for having disease persistence after conization for the treatment of CIN3 and microinvasive carcinoma.