摘要Background Conization is being widely accepted for diagnosis and treatment of cervical intraepithelial neoplasia (CIN). There is controversy as to which factors are most predictive of a positive cone margin and the clinical significance of it. We conducted this study to identify the predictive factors and to evaluate the clinical significance of a positive cone margin in CIN Ⅲ patients.Methods A retrospective review was conducted of 207 patients who had undergone conization due to CIN Ⅲ from January 2003 to December 2005 at Peking Union Medical College Hospital. Of these, 67 had a subsequent hysterectomy. Univariate and multivariate analysis were utilized to define the predictive factors for a positive cone margin, and to compare the pathologic results of conization with subsequent hysterectomy.Results One hundred and fifty-one (72.9%) were margin free of CIN Ⅰ or worse, 37 (17.9%) had CIN lesions close to the margin and 19 (9.2%) had margin involvement. A total of 56 cases (27.1%) had positive cone margins (defined as the presence of CIN at or close to the edge of a cone specimen). Univariate analysis showed that the parity, cytological grade, multi-quadrants of CIN Ⅲ by punch biopsy, gland involvement, as well as the depth of conization were significant factors correlated with a positive cone margin (P<0.05). However the age, gravidity, grade of dysplasia in punch biopsy, as well as the cone methods were not significantly correlated (P >0.05). Multivariate analysis revealed that the cytological grade (OR=1.92), depth of conization (OR=2.03), parity (OR=3.02) and multi-quadrants of CIN Ⅲ (OR=4.60) were significant predictors with increased risk for positive margin. The frequency of residual CIN Ⅰ or worse in hysterectomy specimens was found to be 55.6% (20/36) in patients who were margin free, 71.4% (15/21) in patients with CIN occurring close to margin, and 80.0% (8/10) in patients with margin involvement. The frequency of residual CIN Ⅲ or worse was found to be 13.9% (5/36), 23.8% (5/21) and 50.0% (5/10) respectively in different groups.Conclusions Cytological grade, depth of conization, parity and multi-quadrants of CIN Ⅲ in punch biopsy were significant factors with increased risk in predicting a positive cone margin. Margin status of conization did not mean the presence or absence of CIN, but rather the varied frequency of residual CIN in specimens of subsequent hysterectomy. In view of this fact, it is suggested that the margin status of conization be a valuable surrogate marker for clinical management of CIN Ⅲ.