(CSPOT23) An Active Safety Surveillance Using Real-World Evidence (ASSURE) Approach to Pharmacovigilance Signal Evaluation: The case of infliximab and alternative autoimmune conditions

Background: Extensive literature exists on the topic of TNF-alpha inhibitors and the biological plausibility for risk of alternative autoimmune conditions or paradoxical reactions.

Objectives: To evaluate the association between infliximab (INF) and incident inflammatory bowel disease (IBD), rheumatoid arthritis (RA), plaque psoriasis (PsO), psoriatic arthritis (PsA), ankylosing spondylitis (AS) hidradenitis suppurativa (HS), dermatomyositis (DM), myositis, and autoimmune thyroiditis (AT) among patients with prior IBD, RA, PsO, PsA, and AS, utilizing our ASSURE framework.

Methods: We defined the four pillars of this analysis-Target: INF; Comparators: adalimumab (ADA), vedolizumab (VEDO), abatacept (ABT), ustekinumab (UST), secukinumab (SEC), azathioprine (AZA), and methotrexate (MTX); Indication: IBD, RA, PsO, PsA, and AS (separately); Outcomes: IBD, RA, PsO, PsA, AS, HS, DM, myositis, and AT (separately). A quantitative feasibility assessment identified 9 fit-for-purpose databases (7 U.S. and 2 non-U.S. databases). Comparative cohort (CC) and self-controlled case series (SCCS) designs were employed; the new-user CC design estimated adjusted hazard ratio (95% confidence interval) in an on-treatment analysis (all-time exposed) using a large-scale propensity score matching approach with 1:100 variable-ratio matching. The SCCS design estimated an incidence rate ratio during on-treatment compared to unexposed time. Patients were not considered at-risk if outcomes were observed before cohort entry. We assessed pre-defined analytic diagnostics (e.g. covariate balance, power) to determine whether to unblind results from each analysis and then pool estimates across databases using Bayesian random effects meta-analysis.

Results: All analyses of PsO and AS as an indication and those using MTX as a comparator (RA and PsA indication) failed pre-defined analytic diagnostics. PsA as an indication had no significant meta-analysis findings across all outcomes. IBD indication meta-analysis was significant for the outcomes of RA, comparator: VEDO 2.01 (1.12-3.62) and AZA 2.11 (1.24-3.59); and outcomes PsA, PsO, and AS with comparator AZA 2.13 (1.24-3.66), 1.83 (1.27-2.64), 1.99 (1.03-3.63), respectively. The RA indication analysis exhibited significant meta-analyses findings for IBD comparator: ADA 1.62 (1.04-2.53) and ABT 1.64 (1.09-2.46); and AS comparator ABT 2.15 (1.36-3.41). There were no significant meta-analysis findings produced by SCCS analyses.

Conclusions: Our analysis complements existing literature indicating an increased risk of alternative autoimmune conditions among infliximab treated patients. Several analyses using multiple databases, indicated populations, and comparators failed to pass pre-defined diagnostics.