Endoscopic ultrasound-guided cyanoacrylate injection for gastric varices: confirming the trajectory, navigating the uncertainty

10.1055/a-2879-0652 Bleeding from gastric varices remains one of the most demanding clinical scenarios in portal hypertension, with sustained 6-week mortality and substantial rebleeding rates even in well-resourced centers [ 1 ]. Direct endoscopic cyanoacrylate injection has historically been the cornerstone of hemostasis, but its limitations are well documented: pulmonary, cerebral, or splenic embolization, and post-injection ulcers [ 1 ] [ 2 ]. Over the past 5 years, endoscopic ultrasound (EUS)-guided strategies (EUS injection of cyanoacrylate and/or deployment of embolization coils) have emerged as anatomically rational alternatives, now endorsed by international guidelines, as preferred endoscopic options in expert hands [ 1 ] [ 2 ] [ 3 ]. In this issue of Endoscopy , Brigida et al. report a systematic review and network meta-analysis restricted to randomized controlled trials (RCTs), pooling 10 RCTs and 760 patients across 6 endoscopic interventions: direct endoscopic cyanoacrylate injection, endoscopic thrombin injection, EUS-guided cyanoacrylate injection, EUS-guided coil deployment, EUS-guided coil+cyanoacrylate injection, and large-volume band ligation [ 4 ]. Three findings stand out. First, EUS-guided cyanoacrylate (with or without coils) significantly reduced rebleeding compared with direct endoscopic cyanoacrylate injection (risk ratios 0.39 and 0.15, respectively). Second, no treatment outperformed direct cyanoacrylate injection in achieving complete variceal obliteration. Third, EUS-guided cyanoacrylate injection, large-volume band ligation, and thrombin injection were each associated with fewer composite adverse events than direct cyanoacrylate injection. “…the choice of endoscopic technique will continue to depend on the clinical scenario (active bleeding versus secondary prophylaxis) and on the expertise and equipment available, rather than on a universal best option.” By limiting inclusion to RCTs, the authors address the principal weakness of the previous network meta-analysis in this field, which pooled randomized and retrospective data [ 5 ]. These findings are consistent with the most recent pairwise meta-analysis suggesting that EUS-guided coil+glue reduces recurrent bleeding compared with direct endoscopic injection [ 6 ]. These encouraging results must, however, be viewed with caution. Certainty of evidence was uniformly low to very low. The network is sparse: several key edges (including those connecting EUS-coil and large-volume band ligation) are informed by a single trial each, so indirect estimates are inevitably fragile. More fundamentally, the distribution of effect modifiers across comparisons appears uneven. Child–Pugh class C ranged from 0% in some trials to over 70% in others. Moreover, the proportion of patients with active bleeding varied from 0 to 100%. The authors proposed sensitivity analysis to discriminate the active bleeders from secondary prophylaxis, but active bleeding is essentially driven by a single multicenter trial of large-volume band ligation. Such asymmetry challenges the transitivity assumption underpinning any network meta-analysis. Finally, 7/10 trials were conducted in Asia and 9/10 were single-center studies, raising legitimate questions about external generalizability and making clinical routine application difficult. There is also a gap between trial conditions and real-world emergency care. Most included trials were dominated by secondary prophylactic indications. In active variceal bleeding, EUS-guided therapy is rarely available out of hours: it requires an interventional endosonographer, a dedicated linear echoendoscope, coils, and a trained team, resources that few centers can mobilize at 3 a.m. Even in tertiary units, the pragmatic reality is that the bleeding patient receives direct endoscopic cyanoacrylate injection because it is what is on the trolley. This operational ceiling determines which comparator is clinically relevant for whom. A further interpretive difficulty lies in the composite adverse event outcome. The favorable safety signal for EUS-guided cyanoacrylate injection, large-volume band ligation, and thrombin injection is largely driven by mild events (fever, abdominal pain, mild ulcers), while severe events such as embolization and infarction were too rare for formal comparison. Yet it is precisely those rare embolic complications, including the fatal pulmonary event reported in one included RCT, that motivate the move away from direct cyanoacrylate injection in the first place. Until cohorts large enough to capture differences in severe adverse events become available, claims about comparative safety should be advanced cautiously. No included trial captured survival, quality-of-life, or health-economic outcomes, and definitions of complete obliteration differed across arms, introducing potential bias. Brigida et al. bring interesting data on thrombin injection. Endoscopic injection of human thrombin promotes localized fibrin clot formation without the systemic embolic risks of cyanoacrylate, causes markedly fewer post-injection ulcers, and requires no dedicated EUS infrastructure. The authors confirm a threefold reduction in composite adverse events versus direct cyanoacrylate injection. Although the evidence base remains thin (two RCTs), thrombin emerges as the most plausible technique to disseminate beyond expert centers, particularly where EUS is impractical and cyanoacrylate unavailable. Wider availability, harmonized dosing protocols, and adequately powered comparative trials, ideally benchmarked against EUS-guided combination therapy, are now needed. The findings must also be placed within a broader therapeutic landscape that the analysis cannot, by design, capture. Pre-emptive transjugular intrahepatic portosystemic shunt (TIPS) and balloon-occluded retrograde transvenous obliteration (BRTO) retain a central role for selected high-risk patients [ 1 ] [ 2 ] [ 3 ]. A recent RCT suggests that combining cyanoacrylate injection with BRTO or TIPS reduces rebleeding compared with endoscopy alone [ 7 ]. The recent French GAVAPROSEC trial provides timely evidence on this point: in cirrhotic patients with acute gastric variceal bleeding controlled after initial cyanoacrylate injection, TIPS reduced rebleeding compared with standard endoscopic plus pharmacologic therapy [ 8 ]. It suggests that the emergency algorithm should be conceived as a sequence – initial hemostasis by direct cyanoacrylate injection where EUS expertise is unavailable, followed by early multidisciplinary triage to TIPS or BRTO in high-risk patients – rather than as a competition between endoscopic techniques. Where, then, does this network meta-analysis leave the practicing endoscopist? It strengthens, with the most rigorous synthesis available to date, the rationale for preferring EUS-guided cyanoacrylate injection (augmented by coils where feasible) over direct endoscopic injection in centers with the requisite expertise, particularly for secondary prophylaxis. It vindicates the direction of travel of recent guidelines without resolving the residual uncertainty about which patients gain most. Furthermore, it makes a strong case for adequately powered, prospectively registered, multicenter, and ideally multinational, RCTs, stratified by Child–Pugh class and clinical indication, comparing EUS-guided combination therapy not only against conventional endoscopic injection but also against or in combination with pre-emptive radiological interventions. The Baveno VIII consensus workshop recently concluded in March 2026, and its updated statements are eagerly awaited. Until those recommendations or new data appear, the choice of endoscopic technique will continue to depend on the clinical scenario (active bleeding versus secondary prophylaxis) and on the expertise and equipment available, rather than on a universal best option. Publication History Article published online: 21 July 2026 © 2026. Thieme. All rights reserved. Georg Thieme Verlag KG Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany Refers to: Comparative efficacy and safety of endoscopic treatments for gastric varices: a systematic review and network meta-analysis Endoscopy eFirst DOI: 10.1055/a-2879-0652