Baseline high-frequency ultrasound thickness and doppler vascularity predict early imaging response to intralesional corticosteroid therapy in keloids: A compound betamethasone cohort study

Abstract: Background: High-frequency ultrasound (HFUS) parameters such as scar thickness and intralesional vascularity may reflect keloid biological activity and predict treatment response, yet evidence supporting their baseline predictive value remains limited. Objectives: To evaluate whether baseline HFUS-derived thickness and Doppler vascularity can predict early imaging response to intralesional corticosteroid therapy in keloids and to explore their association with fibrosis-related molecular signatures. Methods: This single-center prospective cohort study enrolled 150 keloid patients receiving standardized compound betamethasone injections. Baseline HFUS measurements included scar thickness, Adler vascularity grade and shear wave elastography (SWE)-derived elastic modulus, along with Vancouver Scar Scale (VSS) assessment. We defined early imaging response at week 4 as percentage thickness reduction (ΔTh%4W). Multivariable logistic regression and receiver operating characteristic (ROC) analysis identified predictors of poor early response (ΔTh%4W < 30%). We analyzed an exploratory molecular subgroup (n=40) for miR-205-5p and RUNX2 expression. Results: Baseline thickness (OR=2.41) and Doppler vascularity independently predicted poor early response. The combined HFUS model showed good discriminative performance (AUC=0.90). In the exploratory molecular subgroup, Doppler-positive lesions showed lower miR-205-5p and higher RUNX2 expression; RUNX2 correlated positively with thickness (r=0.92). At the week-24 follow-up (n=136), the overall recurrence rate was 20.6%, with higher rates in high-activity lesions (52.6%) than in low-activity lesions (7.1%). Conclusion: Baseline HFUS thickness and Doppler vascularity may be potential non-invasive predictors of early response to intralesional corticosteroid therapy in keloids. The HFUS-based activity classification shows clinical potential in identifying patients likely to respond poorly to injection monotherapy, but requires external validation. The observed molecular associations are exploratory and need further functional validation.