ESTIMATION OF PROBABILITY OF MALIGNANCY USING A LOGISTIC MODEL COMBINING PHYSICAL-EXAMINATION, ULTRASOUND, SERUM CA-125, AND SERUM CA-72-4 IN POSTMENOPAUSAL WOMEN WITH A PELVIC MASS - AN INTERNATIONAL MULTICENTER STUDY

Background. To assess the differential diagnostic potential of physica l examination, ultrasound, the serum CA 125 assay, and serum CA 72-4 a ssay, and the contribution of each parameter to a logistic model predi cting the probability of malignancy in postmenopausal patients present ing with a pelvic mass. Patients and methods. In a multicenter, prospe ctive study a total of 155 patients were evaluated preoperatively usin g a standard protocol for pelvic examination, transvaginal (occasional ly additional abdominal) ultrasound, and serum CA 72-4 (cutoff level 3 U/ml) and CA 125 (cutoff level 35 U/ml). Results. Fifty-nine malignan t (39%) and 92 benign (61%) pelvic tumors were found in addition to 4 borderline tumors (3%), Forty-three patients appeared to have ovarian carcinoma, FIGO Stage III or IV in 28 cases. Borderline tumors were ex cluded from the statistical calculations. The diagnostic accuracy of e ach single parameter, i.e,, pelvic examination, ultrasound, and serum CA 125 and CA 72-4 in discriminating between benign and malignant pelv ic masses gave highly similar results (81, 76, 78, and 81% respectivel y). Best sensitivity was found in pelvic examination (92%); best speci ficity was found in CA 72-4 (93%). Using logistic regression analysis the power of pelvic examination appeared to be the most relevant (adju sted odds ratio 12.1), followed by ultrasound (odds ratio 9.7), serum CA 125 (odds ratio 5.0), and serum CA 72-4 (odds ratio 4.9). Age appea red to be nonpredictive. The logistic model gives a correct prediction in 87% of all cases. Conclusions. The addition of serum CA 72-4 to th e combination of pelvic examination, ultrasound, and serum CA 125 lead s to an improved discrimination between malignant and benign pelvic ma sses. (C) 1998 Academic Press.