Ultrasound radiomics integrated with an immune-related gene signature for predicting response and resistance to immune checkpoint inhibitors in cutaneous melanoma: a retrospective radiogenomic study with biological validation.

Ultrasound radiomics combined with an immune-related gene signature predicts melanoma response to immune checkpoint inhibitors with 78-88% accuracy.

• Why it matters: Accurate biomarkers for ICI response in melanoma are lacking, and current methods like PD-L1 and tumor mutational burden are imprecise, limiting personalized treatment.
• What they did: The study analyzed 218 melanoma patients using ultrasound radiomic features to create a US-score and integrated transcriptomic data to derive an immune-related gene signature (IRGS), validated across multiple datasets and in vitro.
• The result: The combined model achieved high predictive accuracy (AUC up to 0.88), effectively stratified patients by response and survival, and revealed ultrasound features reflect tumor-immune microenvironment, enabling non-invasive treatment guidance.