Endoscopic ultrasound-guided versus percutaneous transhepatic gallbladder drainage in surgically high-risk patients with acute cholecystitis: a systematic review and meta-analysis

Percutaneous transhepatic gallbladder drainage (PTGBD) has been the conventional non-operative option for acute cholecystitis in patients at high surgical risk, but endoscopic ultrasound-guided gallbladder drainage (EUS-GBD) with lumen-apposing metal stents (LAMS) is increasingly preferred because it removes the external catheter. We compared the two techniques, with device- or drainage-related unplanned readmission-the outcome specified in our public registration-as the primary endpoint.

Four bibliographic databases and two trial registries were searched from inception to 31 January 2026, with the search updated to 30 July 2026. 6 comparative studies were included (one randomized trial and five observational cohorts; 712 patients, EUS-GBD 322 and PTGBD 390). Odds ratios (OR) were pooled with random-effects restricted maximum likelihood (REML) models using the Hartung-Knapp-Sidik-Jonkman adjustment, with leave-one-out and method-sensitivity analyses and GRADE assessment. The protocol was registered on the Open Science Framework before data extraction.

The primary outcome was reported at the patient level in three studies. Device- or drainage-related unplanned readmission occurred in 14/200 (7.0%) of EUS-GBD patients versus 91/245 (37.1%) of PTGBD patients. Every contributing study favoured EUS-GBD and each study-level interval excluded unity, but the pooled Hartung-Knapp interval did not (OR 0.098, 95% CI 0.008-1.22; I² = 68%), and the estimate was not robust to leave-one-out omission. Technical success was lower with EUS-GBD in all six studies (technical failure 13/322 (4.0%) vs. 4/390 (1.0%); OR 0.31, 95% CI 0.087-1.11), although this interval also included unity. Clinical success (OR 0.99, 95% CI 0.33-2.99), 30-day adverse events (OR 0.55, 95% CI 0.024-12.57) and 30-day all-cause mortality (OR 0.97, 95% CI 0.076-12.51) did not differ significantly, with intervals too wide to exclude clinically important differences in either direction. Certainty was very low for the primary outcome and for adverse events, and low for the remainder.

In surgically high-risk patients with acute cholecystitis, EUS-GBD achieved clinical success and mortality comparable to PTGBD, with a small and consistent numerical reduction in technical success, and with a device-related readmission burden that was lower in every contributing study but not robustly lower on pooled analysis. The present evidence does not establish EUS-GBD as a universally superior first-line replacement for PTGBD. Adequately powered randomized trials using standardized, patient-level endpoints and longer follow-up are required.