RT Journal Article SR Electronic T1 Estimation of probability of malignancy using a logistic model combining color Doppler ultrasonography, serum CA125 level in women with a pelvic mass JF International Journal of Gynecologic Cancer JO Int J Gynecol Cancer FD BMJ Publishing Group Ltd SP 92 OP 98 DO 10.1136/ijgc-00009577-200602001-00015 VO 16 IS Suppl 1 A1 A. S. Mousavi A1 S. Borna A1 S. Moeinoddini YR 2006 UL http://ijgc.bmj.com/content/16/Suppl_1/92.abstract AB The goal of this study was to develop a scoring system using combination of Doppler characterization of pelvic/ovarian lesions and serum CA125 level. Our purpose was to maximize the preoperative discrimination between benign and malignant entities. In a prospective study, a total of 101 patients were evaluated preoperatively using a standard transvaginal ultrasound and color Doppler imaging with pulse spectral analysis and serum CA125 level within a week prior to surgery. The variables that were analyzed by the multivariate logistic regression method are as follows: tumor structure, ascites, presence of septum, the peak systolic velocity (PSV), the resistance index (RI), and serum CA125 level. Of the 101 patients qualified for the study, 48 patients were diagnosed with benign (47.5%) and 53 (52.5%) with malignant tumors. Each criterion used alone provides statistically significant discrimination between benign and malignant tumors. Four criteria could be combined in a malignancy score which is calculated using the product of the serum CA125 level (1 if CA125 ≥40 U/mL and 0 if CA125 <40 U/mL), the result of sonography for presence of septum in tumor (1 if there was septum ≥3 mm, 0 if there was no septum or <3 mm), result of Doppler flow imaging as RI (1 if RI ≤0.5 and 0 if RI >0.5) and the PSV (1 if PSV ≥40 cm/s and 0 if PSV <40 cm/s). This scoring system devised was statistically more effective discriminator between cancer and benign lesions than formal methods. Using malignancy score cutoff level of two, the sensitivity was 98% (CI 88.62–99.9.), the specificity was 85% (CI 71.62–93.45), the positive predictive value was 87.5%, and the negative predictive value was 97.6%. Area under curve of receiver operative characteristic curves was 0.987 (CI 0.971–1.004). These values were statistically more significant than those obtained from the independent use of RI, PSV, or serum CA125 level at their optimum decision values (P < 0.05). There is a need for a prospective evaluation of this score using a larger sample of patients.