JAK inhibitor initiation was not associated with excess MACE (primary outcome), heart failure hospitalization, or VTE compared to TNF inhibitors in RA patients at elevated cardiovascular risk, though a modest excess in all-cause mortality was observed as a secondary outcome with no signal for any individual agent.
Key Findings
Results
JAK inhibitors were not associated with excess major adverse cardiovascular events (MACE) compared to TNF inhibitors in rheumatoid arthritis patients at elevated cardiovascular risk.
Primary outcome HR for MACE was 1.002 (95% CI 0.923–1.088)
Absolute risk difference was +2.11 per 1,000 over 5 years
Study included 18,757 matched pairs using 1:1 propensity score matching
MACE was defined as nonfatal myocardial infarction or nonfatal ischemic stroke
Follow-up began 90 days after treatment initiation
Results
All-cause mortality was modestly higher with JAK inhibitors compared to TNF inhibitors.
HR for all-cause mortality was 1.158 (95% CI 1.058–1.268; p = 0.0014)
Absolute risk difference was +10.4 per 1,000 over 5 years
E-value for this association was 1.586, indicating modest robustness to unmeasured confounding
The association was a secondary outcome and was not reproduced in any agent-specific cohort
Authors noted residual channeling bias cannot be excluded
Results
Heart failure hospitalization did not differ significantly between JAK inhibitors and TNF inhibitors.
HR for heart failure hospitalization was 1.019 (95% CI 0.955–1.088)
The confidence interval crossed 1.0, indicating no statistically significant difference
This was a prespecified secondary outcome
Results
Venous thromboembolism (VTE) risk did not differ significantly at the class level between JAK inhibitors and TNF inhibitors, though baricitinib showed a nominally significant VTE signal.
Overall class-level HR for VTE was 1.054 (95% CI 0.961–1.156), which was not statistically significant
Baricitinib specifically showed a nominally significant VTE HR of 1.364 (95% CI 1.028–1.809) based on 198 events
Baricitinib VTE finding was the only nominally significant agent-level finding in the study
Results
No individual JAK inhibitor agent showed a statistically significant excess in MACE or all-cause mortality in agent-specific analyses.
Mortality HRs by agent were: tofacitinib 1.052, baricitinib 1.213, and upadacitinib 1.090
None of these agent-level mortality findings reached statistical significance
Cohorts were separately matched by agent, sex, cardiovascular-risk stratum, and index era
Results
A prespecified negative control outcome (cataract) showed no excess risk with JAK inhibitors, supporting the validity of the study design.
HR for cataract (negative control) was 0.934 (95% CI 0.867–1.007)
The absence of a signal in the negative control suggests the propensity score matching adequately controlled for measured confounders
Cataract was prespecified as a negative control outcome
Results
No consistent effect modification was observed by sex, cardiovascular risk burden, or index year.
Subgroup analyses were conducted by sex, cardiovascular-risk stratum, and index era
No subgroup showed a consistent differential effect of JAK inhibitors versus TNF inhibitors
This lack of effect modification was consistent across all outcomes examined
Results
A three-component composite outcome including death was not statistically significant.
HR for the composite including death was 1.058 (95% CI 0.992–1.127)
The confidence interval marginally crossed 1.0, indicating no significant difference
This composite was designed to capture a broader safety signal beyond MACE alone
Methods
The study used target trial emulation with improved propensity score matching incorporating a broad range of covariates within the TriNetX database.
Eligible patients were adults with RA aged ≥50 years with ≥1 cardiovascular risk factor
5-year Kaplan-Meier cumulative incidences and absolute risk differences per 1,000 were reported
E-values were calculated to quantify sensitivity to unmeasured confounding
What This Means
This research compared the heart-related safety of two classes of drugs used to treat rheumatoid arthritis (RA): JAK inhibitors (JAKi) and TNF inhibitors (TNFi). Regulatory agencies have raised concerns that JAK inhibitors may increase the risk of heart attacks, strokes, blood clots, and death compared to TNF inhibitors. To investigate this, the researchers analyzed data from over 18,000 matched pairs of RA patients aged 50 or older who had at least one cardiovascular risk factor and had started one of these two drug classes. The study found that JAK inhibitors were not associated with a higher risk of major cardiovascular events (heart attacks or strokes) compared to TNF inhibitors — the primary finding of the study. There were also no significant differences in hospitalizations for heart failure or blood clot (venous thromboembolism) risk at the overall class level, though the drug baricitinib showed a nominally elevated VTE risk in a subgroup analysis.
However, the study did find a modest increase in all-cause (any-cause) deaths among patients taking JAK inhibitors compared to TNF inhibitors — about 10 additional deaths per 1,000 patients over 5 years. This finding was statistically significant but was not reproduced when researchers looked at individual JAK inhibitor drugs separately, and the signal was of modest strength when accounting for the possibility of unmeasured biases. No differences were seen when breaking down results by sex, level of cardiovascular risk, or time period of treatment.
This research suggests that the cardiovascular safety picture for JAK inhibitors compared to TNF inhibitors in RA patients at elevated cardiac risk is nuanced. The lack of excess heart attacks and strokes is notable, but the modest mortality signal adds uncertainty, and the authors caution that neither conclusion — that JAK inhibitors are clearly safe or clearly dangerous — can be drawn from these observational data alone. Residual biases, such as sicker patients being preferentially prescribed one drug over the other, cannot be fully ruled out, and the findings should inform but not replace shared clinical decision-making between patients and their doctors.
McClenahan R, Alamlih L. (2026). Cardiovascular and mortality outcomes with JAK versus TNF inhibitors in rheumatoid arthritis: a target trial emulation (CARDINAL-RA).. Frontiers in immunology. https://doi.org/10.3389/fimmu.2026.1908442