Radiomics adds little to clinical factors for predicting PD-L1 status after chemoradiation in esophageal cancer

A combined clinical-radiomics model predicted post-treatment PD-L1 expression in esophageal squamous cell carcinoma with a weighted AUC of 0.8214, but improvement over a clinical-only model (AUC 0.8171) was minimal. Findings are hypothesis-generating and require external validat…

• The study analyzed pre- and post-neoadjuvant chemoradiotherapy CT scans from 101 patients with locally advanced ESCC, extracting 1873 radiomic features; six were ultimately selected via logistic regression.
• Predictive models integrated radiomics, clinical variables (e.g., radiation technique, total dose), and deep learning outputs (ResNet, TabNet).
• The authors caution that the model predicts PD-L1 expression, not immunotherapy response, and requires external validation before clinical application.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.