Reviews/Psoriasis Psoriatic Arthritis/Tofacitinib major adverse cardiovascular events versus TNF inhibitors in rheumatoid arthritis
Psa Jsaf · PSA-JSAF-01 · v2
Borderline: within 4 points of the Established/Likely boundary, so a re-grade could land either side.
In the trial built for exactly these patients, older rheumatoid arthritis patients with a heart-risk factor, more heart attacks and strokes happened on tofacitinib than on the older TNF-blocker drugs [1]. Every follow-up look at that trial's data pointed the same way, and the excess grew in the highest-risk patients such as older people, smokers, and those with existing heart disease [2][3][6][8]. Two things kept the panel cautious. First, the trial's own headline heart-risk number, while pointing to more events, was not statistically firm on its own [1]. Second, all of the evidence comes from that single trial, and the one outside study in the corpus reports no results yet [9]. So the direction of harm is well supported, but it rests on one trial rather than several independent ones.
The source trial [1] found higher combined-dose event rates with tofacitinib than with a TNF inhibitor (3.4% vs 2.5%) and did not show noninferiority in this risk-enriched population. Every post hoc analysis pointed the same direction, with hazard ratios above 1 and steeper risk in higher-risk strata: elevated risk in smokers and those aged 65 or older [2], a high-risk 'age >=65 or ever smoker' group [3], gradients by baseline cardiovascular risk [6], and higher event rates in both regional cohorts [8]. The excess intensified where baseline risk was greatest, reaching HR 4.07 (95% CI 1.20-13.82) in patients with prior atherosclerotic disease and no statin use [5]. The consistency of direction is the corpus's central demonstration. The contested point: The division is whether 'increased risk' holds across the whole population or only in its high-risk pockets: [3] finds no detectable increase in patients under 65 who never smoked, and [5] finds the excess near-null with statin use (HR 1.17). It also turns on endpoint composition, since [4] reports the broad ischaemic plus heart-failure composite did not appear different, with the positive signal driven by myocardial infarction and venous thromboembolism.
All 9 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Psoriasis Psoriatic Arthritis, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
The primary combined-dose MACE HR is 1.33 with 95% CI 0.91-1.94 [1] - the interval includes the null. The trial's independently significant excess was cancer (HR 1.48, 1.04-2.09), not MACE. 'Increased risk' for MACE rests on incidence difference and failed noninferiority, not a significant primary MACE estimate.
Single-trial dependence: all eight included papers are ORAL Surveillance. Eight citations simulate a literature where there is one trial. The only potentially independent source [9] reports no results, so no replication survives in this corpus.
Where it remains uncertain. The primary combined-dose MACE hazard ratio of 1.33 has a 95% CI of 0.91 to 1.94 that includes the null [1], and the signal localises to high-risk sub-strata rather than the whole population [3][5].
Aleatory
The ASCVD/no-statin subgroup estimate is very imprecise, HR 4.07 with a 95% CI of 1.20 to 13.82 [5], reflecting few events in a small stratum.
Epistemic
Every included paper [1]-[8] derives from one trial; the only potentially independent source [9] reports no results, so no replication exists in the corpus.
Inferential
The primary combined-dose MACE HR 1.33 has a 95% CI of 0.91 to 1.94 [1], so the association rests on incidence difference and failed noninferiority, not a significant primary estimate.
The load-bearing papers from a corpus of 9, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Primary RCT evidence: elevated MACE incidence and HR in the exact population; noninferiority failed. CI crosses 1.
“the incidences of MACE and cancer were higher with the combined tofacitinib doses (3.4% and 4.2%) than with a TNF inhibitor (2.5% and 2.9%). The hazard ratios were 1.33 (95% CI, 0.91 to 1.94) for MACE and 1.48 (95% CI, 1.04 to 2.09) for cancers; the noninferiority of tofacitinib was not shown in this cardiovascular risk-enriched population.”
Panel auditInternal validity: high · scope of inference: RA patients enriched for cardiovascular risk (the population the claim targets), comparing tofacitinib against TNF inhibitors for MACE.Key limitations: MACE CI crosses 1 (non-significant) · combined doses pooled · no absolute N in abstract
Post hoc dose-level MACE HRs confirming direction in aged >=50.
“Total MACE and MI incidence rates and risk were higher with tofacitinib vs TNFi (HRs >1). MACE HR was 1.24 (5 mg) and 1.43 (10 mg) vs TNFi.”
Panel auditInternal validity: medium · scope of inference: RA patients aged >=50 in ORAL Surveillance, examining MACE risk factors and tofacitinib-vs-TNFi risk.Key limitations: post hoc · no CI or p reported · conference abstract
Localises the excess risk to high-risk sub-strata; null in lower-risk patients - bounds the claim's generality.
“'Age >=65 years or ever smoker' defined a high-risk group with increased risk of MACE, myocardial infarction, VTE and all-cause death with tofacitinib versus TNFi (HRs 1.41-5.19). In patients aged <65 years and never smokers, there was no detectable risk increase with tofacitinib versus TNFi”
Panel auditInternal validity: medium · scope of inference: Stratified RA (and cross-programme) patients, distinguishing a high-risk age/smoking group from a group with no detectable increase.Key limitations: risk not uniform across the claim's population · HR range pools multiple endpoints · no detectable increase in <65 never-smokers
The composite of all ischaemic CV events plus heart failure did not appear different, even though MI alone and total CV risk (VTE-driven) were higher.
Panel auditInternal validity: medium · scope of inference: RA patients in ORAL Surveillance under extended/composite CV endpoint definitions, with subgroup signals in ASCVD and age >=65.Key limitations: composite differs from claim's MACE · MI higher but ischaemic composite not · no CI reported
4 of 9 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
A directional MACE excess in this population is consistently supported across the source trial and its post hoc analyses, but single-trial dependence and a primary MACE CI that crosses 1 preclude an Established verdict.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
No independent replication. All eight scored papers are one trial [1]-[8]; the sole potentially independent source [9] carries no results, so the association has never been reproduced in a separate cohort. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A large independent observational or randomised comparison of tofacitinib versus TNF inhibitors reporting MACE with effect estimates in the same population.
Open question: Does the cardiovascular excess with tofacitinib versus TNF inhibitors reproduce outside ORAL Surveillance? Why it is still open: Every scored paper in the corpus is that one trial, and the only outside source [9] reports no results.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 18 Sep 2026 · protocol v0.19