Predicting response to ³²P patch brachytherapy in pathological scars: A model integrating multimodal ultrasound and...

To develop and validate a predictive model integrating ultrasound features and clinical characteristics for assessing the efficacy of ³²P patch brachytherapy in pathological scars.

A total of 48 patients with pathological scars (yielding 169 total lesions) who received ³²P patch brachytherapy at the Department of Nuclear Medicine, the Second Affiliated Hospital of Anhui Medical University (April 1, 2023 - December 31, 2025) were retrospectively analyzed. The patients were randomly divided into a training set and an internal validation set at an approximate 7:3 ratio, yielding 119 and 50 lesions, respectively. Lesions in each set were further divided into an effective group and an ineffective group based on the therapeutic response. Univariate mixed-effects logistic regression analysis was employed to compare ultrasound and clinical characteristics between the two groups. Variables with statistical significance were entered into a multivariate mixed-effects logistic regression analysis to identify independent predictors of treatment efficacy and to construct a predictive model. The discriminative performance of the model was internally validated using the validation set.

Scar thickness, vascular index (VI), disease duration, and mean shear wave velocity (SWV) were identified as independent predictors of the efficacy of ³²P patch brachytherapy in pathological scars (all P < 0.05). The predictive model demonstrated an area under the ROC curve (AUC) of 0.969 (95% CI: 0.944-0.994) in the training set. The calibration curve indicated good agreement between predicted and observed outcomes (χ² = 3.598, P = 0.891). Internal validation in the validation set yielded an AUC of 0.790(95% CI: 0.664-0.916).

A multivariable regression model incorporating scar thickness, VI, disease duration, and mean SWV provides a reliable tool for predicting the efficacy of ³²P patch brachytherapy for pathological scars. This model may assist in optimizing clinical treatment strategies and reducing potential short- and long-term risks associated with radiotherapy.