ABSTRACT Esophageal squamous cell carcinoma (ESCC) remains a leading cause of cancer‐related mortality, with early detection being challenging. Although endoscopic screening can aid in diagnosis, its invasiveness and cost limit widespread compliance. To address this, we developed and validated a non‐invasive diagnostic panel based on urinary exosomal microRNAs (miRNAs) for early ESCC detection. Urine samples were prospectively collected from ESCC patients and healthy controls across five institutions, and exosomal miRNA profiles were obtained using small RNA sequencing. A diagnostic panel was constructed using machine learning with recursive feature elimination in a proof‐of‐concept (POC) cohort comprising 99 ESCC and 93 control samples, and its performance was evaluated by five‐fold cross‐validation. This panel was subsequently tested in a blinded validation cohort of 50 ESCC and 61 control samples. In the POC cohort, the multi‐miRNA panel achieved an area under the curve (AUC) of 0.90. In the validation cohort, the AUC was 0.85 (95% CI: 0.77–0.92), with 84% sensitivity and 66% specificity. The panel showed high sensitivity across disease stages—100% for Stage 0, 91% for Stage I, and 73% for Stages II–IV—and its performance was not significantly affected by sex, body mass index, alcohol use, or smoking. Furthermore, in early‐stage cases, the diagnostic score significantly declined after treatment ( p = 0.026). These findings suggest that the urinary exosomal miRNA panel may offer a practical screening approach for the early detection of ESCC.
Murano et al. (Mon,) studied this question.