Janus kinase (JAK) inhibitors have become important treatment options for immune-mediated inflammatory diseases such as rheumatoid arthritis, particularly when conventional synthetic or biologic disease-modifying antirheumatic drugs are insufficient or not tolerated 1‒4. At the same time, concerns about cardiovascular and thrombotic safety have intensified following randomized and regulatory evidence, leading to changes in warnings and prescribing recommendations 1‒4.In this issue, Yoon et al. 5 provide an important pharmacovigilance-based perspective on the cardiovascular safety profile of JAK inhibitors compared with tumor necrosis factor inhibitors (TNF inhibitors) in a Korean national adverse event reporting system 5. Using the Korea Adverse Event Reporting System (KAERS) from 2015 to 2020, the authors conducted a disproportionality analysis and identified a higher reporting odds ratio (ROR) for total cardiovascular events (CVEs) with JAK inhibitors versus TNF inhibitors, with the strongest signal observed for thrombosis 5. The main reported estimates were ROR 4.90 (95% CI 2.80–8.59) for total CVEs and ROR 12.70 (95% CI 5.10–31.66) for thrombosis, while the signal for major adverse cardiovascular events (MACE) was not statistically significant (ROR 2.03, 95% CI 0.69–6.02) 5.This study is particularly valuable for several reasons. First, it contributes evidence from an Asian population, which remains underrepresented in post-marketing JAK inhibitor cardiovascular safety research. The authors explicitly position their study in this gap, noting that prior pharmacovigilance evaluations have largely relied on Western reporting systems such as FAERS and VigiBase 5. Second, the use of an active comparator (TNF inhibitors) is clinically meaningful. In disproportionality analyses, comparator choice has a major influence on signal interpretation, and comparison against another treatment class used for similar immune-mediated conditions is more clinically informative than comparison against all drugs in the database 5‒7. The authors also appropriately acknowledge that TNF inhibitors may have cardioprotective effects, which could influence relative reporting patterns 5.The findings should be interpreted within the limitations of spontaneous reporting data. The manuscript appropriately frames the analysis as signal detection rather than causal inference 5‒7. KAERS, like other pharmacovigilance systems, is subject to under-reporting, over-reporting, heterogeneity in report quality, and incomplete information on clinically important variables, including comorbidities, concomitant medications, baseline cardiovascular risk, and treatment exposure details 5‒7. Importantly, the absence of denominator data prevents estimation of incidence or absolute risk and limits any direct inference about comparative patient-level safety 6. In addition, heightened regulatory and clinical attention to JAK inhibitor safety may influence reporting behavior itself, including stimulated reporting after safety communications 8.One of the most useful aspects of the paper is that the authors distinguish between total CVEs, thrombosis, MACE, and other CVEs rather than relying on a single composite cardiovascular narrative 5. This distinction is important because the thrombosis signal was strong and statistically significant, whereas the MACE signal was not. The article does not overstate this difference; instead, the authors place their findings in the context of prior evidence, including randomized and observational studies, and emphasize that cardiovascular risk patterns may differ by outcome type, population characteristics, and study design 1, 5, 9. This measured interpretation is a major strength, especially in an area where heterogeneous outcomes are often grouped together broadly.The subgroup analyses by age and sex are also clinically relevant, although they require careful interpretation. The authors report stronger disproportionality signals in patients aged ≥50 years and in women, particularly for thrombosis and other CVEs among women 5. However, the manuscript also notes that age information was missing for a large proportion of reports in both treatment groups, and the database structure does not allow adequate adjustment for important confounders, including prior thrombosis history and reproductive risk factors. The authors appropriately acknowledge these limitations and avoid definitive biological claims regarding sex-specific mechanisms 5.Another strength is the discussion of individual JAK inhibitors. Although thrombosis signals were observed for both tofacitinib and baricitinib relative to TNF inhibitors, the authors do not overinterpret these findings as evidence of class-uniform effects. Instead, they contextualize their results using prior pharmacovigilance and observational work suggesting possible heterogeneity across JAK inhibitors and note that differences in data source, comparator strategy, and population characteristics complicate direct comparisons across studies 5, 9. This is a methodologically sound way to present drug-specific observations within a spontaneous reporting framework.From a clinical and policy perspective, the most appropriate takeaway is a cautious, risk-stratified approach rather than a blanket conclusion about the JAK inhibitor class. This is consistent with current regulatory and clinical practice trends, in which JAK inhibitors remain important options but require closer consideration of cardiovascular, thrombotic, and malignancy risk in treatment selection 2‒4. In that context, pharmacovigilance studies such as this one are most useful when they sharpen surveillance priorities and identify outcomes or subgroups that warrant follow-up using designs that can better address confounding and estimate event rates 6, 7.Overall, Yoon et al. 5 provide a meaningful addition to the post-marketing cardiovascular safety literature on JAK inhibitors by expanding the geographic evidence base and using an active comparator relevant to clinical practice. Their findings reinforce the importance of distinguishing thrombosis from broader cardiovascular composites and of interpreting disproportionality signals within the methodological limits of spontaneous reporting systems. The next step for the field will be methodologically robust comparative studies in Asian populations that can incorporate baseline cardiovascular risk, disease severity, treatment duration, and dose to determine whether the signal patterns observed in KAERS correspond to differences in absolute or relative risk in routine care.The authors thank the editorial team for the invitation to contribute this commentary.The authors declare no conflicts of interest.No funding was received for this work.B.M. conceptualized the commentary and drafted the manuscript. A.L.H. critically revised the manuscript for intellectual content. Both authors approved the final version.
Mikaeili et al. (Thu,) studied this question.