Back

Time-Varying Cardiovascular Risk of Febuxostat versus Allopurinol in Gout: A One-stage Meta-analysis

Zhang, S.

2025-12-30 rheumatology
10.64898/2025.12.29.25343180 medRxiv
Show abstract

ObjectivesWhether febuxostat is associated with an increased cardiovascular risk compared to allopurinol in patients with gout remains controversial, with major randomized trials reporting conflicting results. This study aims to perform a comprehensive meta-analysis using reconstructed individual participant data (IPD) to evaluate the time-varying cardiovascular and mortality risk of febuxostat versus allopurinol in gout. MethodsWe conducted a one-stage individual participant data meta-analysis. PubMed, Embase, and Cochrane Library were searched up to November 2025 for randomized controlled trials and propensity score-matched observational studies reporting Kaplan-Meier curves for cardiovascular risk or all-cause mortality. Individual patient data were reconstructed from published curves. Time-varying hazard ratios (HRs) were estimated using a mixed-effects Cox model with treatment-by-time interaction across pre-specified intervals (0-12, 12-24, 24-48, >48 months). ResultsFour studies (n=19,090) were included. Analysis revealed significant time-varying effects. For cardiovascular risk, HRs were 0.89 (95% CI 0.79 to 0.99) at 0-12 months, 0.58 (0.50 to 0.68) at 12-24 months, 0.86 (0.74 to 0.99) at 24-48 months, and 1.09 (0.81 to 1.46) after >48 months. For all-cause mortality, HRs were 0.70 (0.62 to 0.79), 0.39 (0.33 to 0.45), 0.75 (0.65 to 0.86), and 1.02 (0.77 to 1.34) across the same intervals, respectively. ConclusionThe cardiovascular and mortality risk of febuxostat relative to allopurinol is time-dependent, showing a significant early reduction that attenuates over time. These findings advocate for a time-aware approach to clinical management and monitoring.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.