Open- source computed tomography (CT)-based sarcopenia assessment in women with suspicious adnexal masses: a feasibility and reproducibility study

Aim

Ovarian cancer is the most lethal gynecologic malignancy, with prognosis linked to stage at diagnosis. Ultrasonography is the first-line tool for adnexal mass evaluation, while computed tomography (CT) supports staging and surgical planning. Sarcopenia—low skeletal muscle mass—is a prognostic factor in oncology and measurable on routine CT. Its role in women with indeterminate or malignant adnexal masses remains unclear. This study aimed to assess the feasibility, reproducibility, and clinical utility of CT-based sarcopenia analysis using open-source software in this population.

Materials and Methods

In this prospective pilot study, 38 women with suspicious adnexal masses (classified as indeterminate or malignant by International Ovarian Tumor Analysis Simple Rules, Ovarian-Adnexal Reporting and Data System, or subjective assessment) underwent CT-based sarcopenia analysis. Axial CT images at the L3 level were segmented using 3D Slicer. Skeletal muscle index (SMI, cm²/m²) was calculated, with sarcopenia defined as SMI <38.5 cm²/m². Interobserver agreement was evaluated with Bland–Altman plots and Cohen’s kappa. Associations were tested using chi-square, and diagnostic performance for malignancy prediction was assessed by sensitivity, specificity, predictive values, likelihood ratios, and area under the receiver operating characteristic curve.

Results

Sarcopenia analysis was feasible and reproducible (κ = 0.839). Significant associations were observed with advanced tumor stage (P = .028) and peritoneal carcinomatosis (P = .033), but not with histopathological malignancy, lymphadenopathy, adjacent invasion, or suspicion of alternative primary tumor. Diagnostic performance for malignancy prediction was limited.

Conclusion

CT-based sarcopenia analysis using open-source software is feasible and reproducible. Although associated with advanced disease, sarcopenia did not enhance malignancy prediction and should not guide preoperative risk stratification. Its potential value may lie in identifying patients who could benefit from prehabilitation or targeted nutritional support before treatment.