Comparative diagnostic yield of different endoscopic techniques for tissue sampling of upper gastrointestinal subepithelial lesions: a network meta-analysis

Abstract

Background There is limited evidence on the comparative diagnostic performance of endoscopic tissue sampling techniques for subepithelial lesions. We performed a systematic review with network meta-analysis to compare these techniques. Methods A systematic literature review was conducted for randomized controlled trials (RCTs) comparing the sample adequacy and diagnostic accuracy of bite-on-bite biopsy, mucosal incision-assisted biopsy (MIAB), endoscopic ultrasound- guided fine-needle aspiration (EUS-FNA), and EUSguided fine-needle biopsy (FNB). Results were expressed as relative risk (RR) and 95%CI. Results Eight RCTs were identified. EUS-FNB was significantly superior to EUS-FNA in terms of sample adequacy (RR 1.20 [95%CI 1.05 1.45]), whereas none of the other techniques significantly outperformed EUS-FNA. Additionally, bite-on-bite biopsy was significantly inferior to EUSFNB (RR 0.55 [95%CI 0.33 0.98]). Overall, EUS-FNB appeared to be the best technique (surface under cumulative ranking [SUCRA] score 0.90) followed by MIAB (SUCRA 0.83), whereas bite-on-bite biopsy showed the poorest performance. When considering lesions <20mm, MIAB, but not EUS-FNB, showed significantly higher accuracy rates compared with EUS-FNA (RR 1.68 [95%CI 1.02 2.88]). Overall, MIAB ranked as the best intervention for lesions <20mm (SUCRA score 0.86 for adequacy and 0.91 for accuracy), with EUS-FNB only slightly superior to EUS-FNA. When rapid on-site cytological evaluation (ROSE) was available, no difference between EUS-FNB, EUS-FNA, and MIAB was observed. Conclusion EUS-FNB and MIAB appeared to provide better performance, whereas bite-on-bite sampling was significantly inferior to the other techniques. MIAB seemed to be the best option for smaller lesions, whereas EUS-FNA remained competitive when ROSE was available.