Cancer risk with tumor necrosis factor alpha (TNF) inhibitors: meta‐analysis of randomized controlled trials of adalimumab, etanercept, and infliximab using patient level data

Abstract

Uncertain short- and long-term cancer risks with anti-TNF therapies is a concern, and led to a recent black box warning. This meta-analysis, requested by the European Medicines Agency, aimed at better assessing short-term risks by using meta-analytic techniques based on individual patient data from all corporate-sponsored randomized controlled trials (RCTs) of adalimumab, etanercept, and infliximab. All 74 RCTs of TNF inhibitors of at least 4 weeks duration were provided to independent investigators, including case narratives for events occurring between trial start until 30 days after planned end of treatment and indicating a possible cancer. Relative risks were estimated using Bayesian piecewise exponential models. One hundred thirty (0.84%) of 15,418 individuals randomized to anti-TNF therapy were diagnosed with cancer, compared to 48 (0.64%) of 7486 individuals randomized to comparators. The relative risks associated with all anti-TNF were 0.99 (95%CI 0.61-1.68) for cancers excluding non-melanoma skin cancer (NMSC), and 2.02 (95%CI 1.11-3.95) for NMSC. There were indications of differences in the relative risks for the three anti-TNF drugs, but also of differences across the cancer rates in the three comparator arms for adalimumab, etanercept, and infliximab. Despite a reassuring overall short-term risk, we could neither refute nor verify that individual anti-TNF therapies affect the short-term clinical emergence of cancer. Despite representing the best available evidence, statistical precision, and differences in baseline cancer risk and reporting detail between trials of adilumumab, etanercept, and infliximab hampered distinction of drug-specific from trial effects, illustrating the challenges in safety-assessments using RCT meta-analyses. Long-term risk assessment requires observational studies.