Development and internal validation of a clinical response-estimation model incorporating treatment-course exposure...

Response to hematoporphyrin monomethyl ether photodynamic therapy (HMME-PDT) varies among patients with port-wine stain (PWS), and treatment response is associated with both patient clinical characteristics and lesion properties. This study sought to establish a prediction model to assess the likelihood of essential cure in PWS patients after HMME-PDT.

This study retrospectively included PWS patients who received HMME-PDT in our department. All patients completed final efficacy assessments with complete data for the formal prediction model, which focused on "essential cure" (defined as ≥90% lesion clearance). The primary multivariable logistic regression and prediction model included age at initial treatment, log-transformed lesion area, anatomical site, clinical lesion type, and number of treatment sessions, and was internally validated via fixed stratified five-fold cross-validation.

Among the 855 PWS patients included in this study, essential cure was observed in 253 (29.6%). Younger age, smaller lesion area, Face/neck location, pink clinical lesion type, and receiving 3 or ≥4 treatment sessions were associated with a higher probability of status of essential cure. The final five-factor model showed good discrimination (AUC, 0.850), acceptable calibration (Brier score, 0.136; calibration slope, 0.933; Hosmer-Lemeshow P = 0.802), and interpretable internal validation performance.

A five-factor clinical response-estimation model incorporating age at initial treatment, lesion area, anatomic site, clinical lesion type, and number of treatment sessions showed good internal performance for estimating status of essential cure after HMME-PDT in patients with PWS. The model may aid treatment assessment and patient communication, but external validation is needed before clinical use.