Analysis of risk factors associated with recurrence of granulomatous lobular mastitis and establishment of a prediction model

Abstract

Objectives: To explore the risk factors that are associated to the recurrence of granulomatous lobular mastitis (GLM) after treatment and to develop a nomogram model that can predict the risk of recurrence. Methods: Medical records and follow-up data from GLM patients treated at our facility were retrospectively analyzed. In order to identify independent risk factors for GLM recurrence, the patients were split into groups with and without recurrence using univariate and multivariate logistic regression analyses. A predictive nomogram model was built using these variables, and its performance was evaluated by plotting the calibration curve, receiver operating characteristic (ROC), area under the curve (AUC), and decision curve analysis (DCA) curves. Results: We evaluated the data of 287 GLM patients, of which 36 patients (12.5%) experienced recurrence, while the remaining 251 patients (87.5%) were included in the non-recurrence group. The affected side’s nipple discharge was detected using multivariate logistic regression, nipple inversion on the affected side, surgical treatment, and younger age as the independent risk factors for recurrence (P < 0.05). The model had an AUC of 0.805 (95% CI: 0.737–0.872), 81.7% sensitivity, and 66.7% specificity. The calibration curve was consistent with the ideal curve, and the DCA indicated good clinical utility of the model. Conclusion: Independent risk factors for GLM recurrence include younger age, surgical treatment, nipple discharge on the affected side, and nipple inversion on the affected side. The nomogram based on these factors demonstrated good predictive value for GLM recurrence.