Pre-emptive TIPS for gastric variceal bleeding in patients with cirrhosis (GAVAPROSEC): an open-label randomised clinical trial

Abstract

Background: In patients with cirrhosis, variceal glue obliteration is recommended for the treatment of acute fundal variceal bleeding, and for the prevention of rebleeding in combination with non-selective β blockers (NSBB). We evaluated the efficacy of pre-emptive transjugular intrahepatic portosystemic shunt (p-TIPS) in this setting.

Methods: This open-label randomised trial was conducted at 17 tertiary centres in France. Patients with cirrhosis and acute fundal variceal bleeding (excluding type 1 gastro-oesophageal varices), who achieved initial haemostasis with endoscopic glue injection and vasoactive therapy, and remained stable for at least 12 h, were randomly assigned to receive either a covered p-TIPS within 72 h or to continue with on-demand glue obliteration sessions combined with NSBB. Randomisation was centralised, stratified by centre, and performed in blocks of four (1:1 ratio). The primary composite endpoint was all-cause mortality or clinically significant rebleeding within 1 year. Analyses were conducted in the modified intention-to-treat (mITT) population. This completed study was registered on ClinicalTrials.gov (NCT03705078).

Findings: Between Jan 3, 2019, and Feb 25, 2023, 292 patients were screened, of whom 105 were enrolled and randomly assigned. After excluding two patients who were randomly assigned by error and two who withdrew consent, 101 patients were included in the mITT population (mean age 58·2 years [SD 9·7], 81 [80%] male, 91 [90%] alcohol-related cirrhosis, mean MELD score 14·3 [SD 5·0]). Of these 101 patients, 47 were allocated to p-TIPS and 54 to glue obliteration and NSBB. The 1-year probability of being free from death or rebleeding was 77% (95% CI 62-87) in the p-TIPS group versus 37% (24-50) in the glue obliteration and NSBB group (hazard ratio 0·25 [95% CI 0·12-0·51]; p<0·0001). Rescue TIPS was required in 20 (37%) patients in the control group. Glue migration was reported in eight patients (three [6%] in the p-TIPS group and five [9%] in the control group). One case of cardiac decompensation occurred in the p-TIPS group. The 1-year cumulative incidence of hepatic encephalopathy was similar between groups (35% [95% CI 21-49] vs 32% [19-45]).

Interpretation: In patients with cirrhosis and acute bleeding from fundal varices, p-TIPS significantly reduced the risk of rebleeding or death at 1 year and should be considered as first-line therapy.

Funding: French Ministry of Health.