Paper Details

ABSTRACT

Introduction
Four Bruton tyrosine kinase (BTK) inhibitors are approved for B-cell malignancies. Ibrutinib (2013) is limited by cardiovascular toxicity, particularly atrial fibrillation (AF) and hypertension. Acalabrutinib (2017) and zanubrutinib (2019) were designed for greater selectivity, and pirtobrutinib (2023) for reversible binding, all hypothesized to cause less off-target toxicity. Whether these design differences translate to safer real-world profiles has not been characterized in a single analysis with all four drugs on the market. We used the FDA Adverse Event Reporting System (FAERS, renamed the Adverse Event Monitoring System in March 2026) to compare their post-marketing profiles across the shared calendar window.

Methods
We analyzed FAERS data from 2023 Q1, the first quarter during which all four BTK inhibitors were on the US market, through 2026 Q1. After deduplication, reports listing each drug as primary suspect were identified using word-boundary regular expressions. Disproportionate reporting was assessed against the full database and in all six pairwise head-to-head comparisons. A signal required an ROR lower confidence bound above 1, PRR of 2 or more with chi-square of 4 or more, IC025 above 0, and at least three cases. Predefined event sets covered cardiovascular, hemorrhagic, infectious, and cytopenic outcomes. Pre-registered sensitivity analyses varied suspect role, reporter type, and co-reported BTK inhibitors; a post-hoc analysis restricted to CLL-indication reports. The protocol was registered on the Open Science Framework (osf.io/hnduw) before any data were extracted.

Results
We identified 15,613 ibrutinib, 7,170 acalabrutinib, 2,239 zanubrutinib, and 459 pirtobrutinib primary-suspect reports. Ibrutinib reproduced its established cardiovascular profile against the full database (AF ROR 11.19, 95% CI 10.35-12.09; hypertension 2.77, 2.51-3.05), and cytopenias appeared across all four agents, confirming method validity.

In head-to-head comparisons against ibrutinib (all values ROR with 95% CI), AF was reported significantly less often with acalabrutinib (0.39, 0.32-0.47), zanubrutinib (0.47, 0.35-0.63), and pirtobrutinib (0.19, 0.07-0.51). Acalabrutinib was also lower across five composite entities: atrial arrhythmia (0.39), hypertension (0.42), neutropenia (0.50), thrombocytopenia (0.68), and bleeding (0.74). These findings persisted when restricted to CLL-indication reports (8,036 ibrutinib, 3,612 acalabrutinib, 733 zanubrutinib), with atrial arrhythmia reductions unchanged for acalabrutinib (0.38) and zanubrutinib (0.46). Pirtobrutinib showed elevated cardiac failure of similar magnitude against all three comparators (2.07-2.20).

Three apparent zanubrutinib signals did not withstand scrutiny. Hepatitis B reactivation (19.26 versus acalabrutinib) rested on 6 cases, and 309 zanubrutinib reports originated from China. Ecchymosis (4.42) and purpura (5.89) rested on 11 cases each. Haematologic differences versus ibrutinib attenuated or reversed under healthcare-professional restriction (neutropenia 2.02 to 0.91; thrombocytopenia 1.69 to 0.65); professionals filed 54.5% of zanubrutinib versus 32.9% of ibrutinib reports.

Conclusion
To our knowledge, this is the first FAERS analysis of all four approved BTK inhibitors reporting all six pairwise comparisons. Reduced AF reporting versus ibrutinib was observed for all three newer agents, held across every sensitivity analysis and under CLL-only restriction. Acalabrutinib showed generally lower reporting across the composite entities examined, robust to reporter type. The elevated cardiac failure signal for pirtobrutinib may relate to its use after covalent BTK inhibitor failure, in patients with longer disease duration and greater cumulative cardiac exposure, though disproportionality cannot distinguish this from a drug effect. Several apparent zanubrutinib signals were not robust: the hepatitis B signal may reflect higher seroprevalence in its reporting population, and the haematologic differences may reflect greater laboratory monitoring by professional reporters. These are spontaneous reports subject to notoriety and channeling bias and do not reflect incidence. Where differences emerge, drug selection may be informed by patient-level comorbidity risk. Within this class, the reporting profile appears to vary by agent, and comparative claims from spontaneous data warrant explicit testing for reporting artifacts.