Diagnostic Performance of Magnifying Endoscopy With Crystal Violet Staining for Superficial Non‐ampullary Duodenal Epithelial Tumors: A Single‐center Prospective Study

Objectives We previously reported a potential diagnostic algorithm for superficial non‐ampullary duodenal epithelial tumors (SNADETs) using white‐light magnifying endoscopy with crystal violet staining (ME‐CV). This study aimed to determine the diagnostic performance of the scheme and compare it with the conventional white‐light endoscopy (WLE) scoring system in a prospectively accumulated cohort. Methods This was a single‐center prospective cohort study conducted over a 3‐year period. The primary endpoint was the diagnostic performance of ME‐CV in distinguishing Vienna classification (VCL) category 4/5 (C4/5) from category 3 (C3) SNADETs, including the positive predictive value (PPV) and negative predictive value (NPV), as evaluated by two expert endoscopists. We compared the diagnostic performance of the WLE and ME‐CV algorithms. Results Fifty patients with SNADETs were enrolled. The inter‐observer agreement for the WLE scoring system and the ME‐CV algorithm was good (kappa 0.66 and 0.63). The PPV and NPV of the ME‐CV algorithm, when applied by expert endoscopists, were 25.6% and 90.2%, respectively. The sensitivity, specificity, PPV, NPV, and accuracy of the WLE scoring system were 43.8%, 73.8%, 24.1%, 87.3%, and 69.0%, respectively. The sensitivity, specificity, and accuracy of the ME‐CV algorithm were 62.5%, 65.5%, and 65.0%, respectively. Comparison of the diagnostic performance between the two systems demonstrated the significantly higher sensitivity of the ME‐CV algorithm (WLE, 43.8%; ME‐CV, 62.5%; p = 0.029). Conclusions The ME‐CV algorithm had higher sensitivity than the WLE scoring system for distinguishing VCL C4/5 from VCL C3 in SNADETs, suggesting its potential to improve diagnosis and for selecting appropriate endoscopic resection.