Background: Osteoradionecrosis (ORN) is a severe complication following radiotherapy (RT), particularly after mandibular reconstruction with a fibula free flap. Risk prediction remains challenging due to limited data. Methods: We conducted a multicenter retrospective study including 294 patients who underwent fibula free flap reconstruction and adjuvant RT. A predictive nomogram for ORN was developed using logistic regression with stepwise backward selection, validated via 10-fold cross-validation. Results: ORN occurred in 9.9% of patients. Eight variables were retained in the final model, which demonstrated excellent discrimination (AUC = 0.94) and good calibration (mean absolute error = 0.017). The optimal threshold for classification was 0.30, with the model achieving a sensitivity of 1.0, specificity of 0.80, and an overall accuracy of 0.82. Conclusion: This is the first nomogram developed to predict ORN following fibula free flap mandibular reconstruction. It provides a personalized risk estimate, supporting individualized clinical decision-making and improved pre- and postoperative planning. Prospective validation is warranted.
Nomogram for Predicting Osteoradionecrosis After Fibula Free Flap for Mandibular Reconstruction
Andrea Costantino;Filippo Marchi;Francesco Mazzola;Luca Calabrese;Giampiero Parrinello;Giorgio Peretti;
2025-01-01
Abstract
Background: Osteoradionecrosis (ORN) is a severe complication following radiotherapy (RT), particularly after mandibular reconstruction with a fibula free flap. Risk prediction remains challenging due to limited data. Methods: We conducted a multicenter retrospective study including 294 patients who underwent fibula free flap reconstruction and adjuvant RT. A predictive nomogram for ORN was developed using logistic regression with stepwise backward selection, validated via 10-fold cross-validation. Results: ORN occurred in 9.9% of patients. Eight variables were retained in the final model, which demonstrated excellent discrimination (AUC = 0.94) and good calibration (mean absolute error = 0.017). The optimal threshold for classification was 0.30, with the model achieving a sensitivity of 1.0, specificity of 0.80, and an overall accuracy of 0.82. Conclusion: This is the first nomogram developed to predict ORN following fibula free flap mandibular reconstruction. It provides a personalized risk estimate, supporting individualized clinical decision-making and improved pre- and postoperative planning. Prospective validation is warranted.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
