Assessment of a new logistic model in the preoperative evaluation of adnexal masses
Related Experiment Videos
J L Alcázar1, T Errasti, C Laparte
1Department of Obstetrics and Gynecology, Clinica Universitaria de Navarra, School of Medicine, University of Navarra, Pamplona, Spain.
Objective:
To assess a new logistic regression model developed to predict malignancy in adnexal masses.
Methods:
In the first part of this study, we developed a logistic model by applying logistic regression analysis in a series of 268 adnexal masses (203 benign and 65 malignant lesions) in 248 patients (mean age, 43.6 years; SD, 14.2 years) evaluated and treated at our institution. Eleven parameters were entered in the logistic regression analysis in a forward stepwise way. In the second part of the study, we evaluated the model's diagnostic performance in a further set of 135 adnexal masses (103 benign and 32 malignant tumors) in 129 patients (mean age, 44.4 years; SD, 14.6 years). This diagnostic performance was compared with that of age, tumor volume, Sassone's and Ferrazzi's B-mode ultrasonographic morphologic scoring systems, serum cancer antigen 125 level, and the tumor's lowest resistive index. Comparison was done by calculating the area under the receiver operating characteristic curve.
Results:
In logistic analysis, only menopausal status, the presence of papillary projections, the logarithm of the cancer antigen 125 value, tumor blood flow location, and the lowest resistive index were retained in the model. The model had the best area under the curve (0.97), significantly higher than patient age (area under the curve, 0.78; P = .001), tumor volume (area under the curve, 0.68; P < .0001), cancer antigen 125 (area under the curve, 0.88; P = .008), lowest resistive index (area under the curve, 0.85; P = .011), Ferrazzi's scoring system (area under the curve, 0.89; P = .01), and maximal peak systolic velocity (area under the curve, 0.71; P< .0001). Comparison with Sassone's scoring system (area under the curve, 0.91) did not reach statistical significance, but a clear trend was found (P = .116).
Conclusions:
The model had the best diagnostic performance for discriminating between benign and malignant adnexal masses. A clinical prospective evaluation is needed to confirm its actual value.