Nomogram prediction model based on non-enhanced computed tomography for adnexal torsion: A retrospective multicenter study.

Abstract

Objective: To develop and validate an individualized nomogram for predicting adnexal torsion in women with abdominal pain and an adnexal mass based on preoperative non-contrast computed tomography (CT) findings.

Methods: This retrospective study included 200 women with surgically resected ovarian lesions who underwent preoperative non-contrast CT for abdominal pain from January 2017 to September 2023 in seven hospitals. The 200 patients were randomly divided into a development group (140 cases) and a validation group (60 cases). The independent risk factors for adnexal torsion from non-contrast CT signs were screened using univariate and multivariate logistic regression analyses. The corresponding nomogram prediction model was drawn according to the regression coefficients. The areas under the receiver operating characteristic curves (AUC) and the calibration plots were used to estimate the discrimination and calibration of the prediction model, respectively.

Results: The most common lesion causing adnexal torsion was ovarian cyst (30.1%), followed by mature cystic teratoma (25.7%), whereas malignant tumors and tubo-ovarian abscesses were more common in non-torsioned patients. Multivariable regression analysis showed that navel sign (odds ratio [OR] 14.78; P = 0.005), whirl sign (OR 38.05; P = 0.002), and peri-adnexal fat haziness (OR 0.20; P = 0.012) were independent significant parameters predicting adnexal torsion. The AUC of the development group and validation group were 0.903 and 0.861, respectively, and the calibration curve suggested good agreement between the predictive probability and the actual probability.

Conclusion: Navel sign, whirl sign, and peri-adnexal fat haziness, and an integrated nomogram derived from these non-contrast CT findings, can be useful for predicting adnexal torsion, especially when ultrasound and contrast-enhanced CT cannot be used.