Should proton pump inhibitors be used for glucocorticoid-related upper gastrointestinal bleeding prevention in patients with autoimmune liver disease? A real-world cohort study.

BACKGROUND AND AIMS: Glucocorticoids (GCs) as a key therapeutic option for autoimmune liver diseases (AILDs) are associated with an increased risk of upper gastrointestinal bleeding (UGIB). Proton pump inhibitors (PPIs) are widely used for the prevention and treatment of drug-related UGIB. However, it is unclear whether PPI can be used for the prevention of GC-related UGIB in patients with AILD. The aim of the study was to identify risk factors for UGIB and evaluate the prophylactic efficacy of PPI in patients with AILD undergoing GC. METHODS: We retrospectively compared the occurrence of UGIB and PPI use in patients with AILD receiving GC between January 2005 and May 2025. Univariate and multivariate Cox regression were analyzed to identify independent influencing factors. Propensity score matching (PSM) was utilized to evaluate the prophylactic effect of PPI administration at different time points. RESULTS: A total of 364 patients with AILD treated with GC were included with a median follow-up duration of 29 months. Thirty-two patients (8.8%) experienced UGIB. Compared to patients without UGIB, the UGIB group had a higher proportion of PBC+AIH (90.6% vs. 64.8%; p = 0.001) and moderate to severe esophagogastric varices (56.3% vs. 19.9%; p < 0.001), as well as higher alkaline phosphatase (ALP) (269 vs. 153; p < 0.001) and immunoglobulin M (IgM) (3.82 vs. 2.28; p = 0.007) and lower platelet count (PLT) (100 vs. 121; p = 0.015). After multivariate Cox adjustment, concurrent with PLT (HR = 0.993, 95% CI: 0.986-1.000, p = 0.048), moderate to severe esophagogastric varices (HR = 3.876, 95% CI: 1.798-8.355, p = 0.001) were independently associated with UGIB. In this cohort, PPI users had a higher risk of UGIB; however, this association may be driven by confounding factors (log-rank p = 0.008). The duration of PPI use did not significantly affect the cumulative risk of UGIB (log-rank p = 0.426). Furthermore, neither baseline (log-rank p = 0.343), within 1 year after baseline (log-rank p = 0.132), nor PPI use for more than 1 year after baseline (log-rank p = 0.195) significantly reduced the cumulative risk of UGIB. CONCLUSIONS: In our real-world cohort, PPI use does not prevent GC-related UGIB among patients with AILD and may even be associated with a higher bleeding risk; however, this observed association is likely driven by confounding factors and should not be interpreted as a causal relationship.