Introduction
Endoscopic ultrasound (EUS)-guided tissue acquisition is the standard of care for diagnosing solid pancreatic masses. However, the optimal needle type and the impact of adjunctive techniques, including rapid on-site evaluation (ROSE), fanning, suction, contrast, and stylet use, remain uncertain.
Aims & Methods
This study aimed to compare the diagnostic performance of different EUS-guided needle biopsy techniques and to evaluate the impact of key procedural covariates on outcomes. A Bayesian random-effects NMA was conducted using randomized controlled trials evaluating EUS-guided needle biopsy for solid pancreatic lesions. Outcomes included diagnostic accuracy (contrast yield), sample adequacy, suction technique performance, ROSE, fanning, stylet use (all dichotomous), number of passes (continuous), and adverse events. Effect estimates were reported as risk ratios (RR) or mean differences (MD) with 95% credible intervals (CrIs). Treatment ranking was assessed using Surface Under the Cumulative Ranking Curve (SUCRA). Meta-regression evaluated the modifying effects of ROSE, fanning, and stylet use.
Results
The NMA included 14 to 18 interventions per outcome, depending on the network. For diagnostic accuracy, Menghini_22G_FNA was the top-ranked intervention (SUCRA = 0.80), followed by FNB_22G_Contrast (SUCRA = 0.67). For number of passes, Menghini-side-ported_22G_FNA required the fewest passes (SUCRA = 0.84), with Menghini_22G_FNA (SUCRA = 0.79) and Menghini_25G_FNA (SUCRA = 0.78) also performing well. For sample adequacy, Franseen_22G_FNB was superior (SUCRA = 0.94), significantly outperforming Antegrade-Bevel_20G_FNB (RR 121.24, 95% CrI 1.42 to 1,741,347.85) and Lancet_22G_FNA (RR 0.56, 95% CrI 0.25 to 1.11), though with substantial uncertainty. For ROSE-guided procedures, Fork-Tip_22G_FNB was the best (SUCRA = 0.84), followed by Fork-Tip_25G_FNB (SUCRA = 0.79). For fanning technique, Fork-Tip_22G_FNB ranked first (SUCRA = 0.85). For stylet use, Fork-Tip_22G_FNB again ranked highest (SUCRA = 0.84), though the overall effect of stylet was negligible. For suction technique, Franseen_22G_Dry-Suction demonstrated the highest diagnostic yield (SUCRA = 0.89), followed by Franseen_25G _Dry-Suction (SUCRA = 0.75), while no-suction approaches (e.g., Lancet_22G_No-Suction, Lancet_25G_No-Suction) and M-series needles ranked lower, with several comparisons showing wide credible intervals indicating heterogeneity. For adverse events, Franseen_25G_FNB demonstrated the safest profile (SUCRA = 0.94). Meta-regression revealed no statistically significant effect for ROSE (coefficient 0.219, 95% CrI -1.530 to 6.406), fanning (0.079, 95% CrI -0.323 to 0.518), or stylet use (-0.018, 95% CrI -2.805 to 4.455), indicating that these covariates did not independently modify treatment effects across networks.
Conclusion
This network meta-analysis identifies Menghini_22G_FNA as the optimal needle for diagnostic accuracy, Menghini-side-ported_22G_FNA for minimizing the number of passes, and Franseen_22G_FNB for sample adequacy. Adjunctive techniques including ROSE, fanning, and stylet use did not demonstrate statistically significant independent effects on diagnostic performance, though certain needle types showed improved rankings when used with these techniques. These findings provide evidence-based guidance for selecting EUS needles for solid pancreatic mass sampling and suggest that technique selection should be tailored to the specific clinical priority (accuracy, efficiency, or safety).