JAK Inhibitor Candidacy After Two Failed TNF Inhibitors in a Patient with Cardiovascular Risk Factors
Two TNF inhibitors have already failed, and the next mechanism on the list carries a boxed warning built from a trial population this patient partly, but not entirely, resembles.
Harold T., a 63-year-old man, spends most weeknights now at his daughter's kitchen table helping his two grandsons with homework, a routine that started when her shift at the hospital moved to nights and hasn't stopped since. He retired three years ago from supervising a machine shop floor, work that left both his hands with the kind of wear his rheumatologist still asks about directly. His rheumatoid arthritis is nine years old and has now outlasted two TNF inhibitors: adalimumab worked well for three years before losing effect, and etanercept, tried next, never brought him below moderate activity in six months. His DAS28-CRP today is 5.1.
He also carries three findings that, taken together, are exactly what ORAL Surveillance screened for when it randomized patients fifty and older with at least one cardiovascular risk factor to tofacitinib or a TNF inhibitor: his LDL has needed atorvastatin for the past six years, his most recent HbA1c of 6.0% sits in the prediabetic range, and he quit smoking twelve years ago after two decades of a pack a day. That trial found more heart attacks, strokes, cancers, and blood clots with tofacitinib than with the TNF inhibitor comparator, a finding serious enough that the FDA extended its boxed warning to the entire JAK inhibitor class rather than to tofacitinib alone. What the label doesn't distinguish is how much cardiovascular risk is enough to count, and the trial's own post-hoc analyses have since begun to. The excess cardiovascular risk was concentrated in enrollees with established atherosclerotic disease; among those with no ASCVD history, MACE occurred at a similar rate on tofacitinib and on a TNF inhibitor, statin or no statin. He has no established ASCVD — his ten-year risk estimate lands at 9%, intermediate rather than high — which places him in the part of the trial where the two drug classes did not separate. The statin-mitigation finding usually quoted alongside this belongs to the ASCVD group and not to him, so it is the wrong thing to lean on even though he does take a statin. A second analysis is less reassuring: the absence of excess risk was clearest in enrollees under 65 who had never smoked, and at 63 with two decades of a pack a day behind him, he falls outside that group on one of its two criteria. Whether that reading changes anything for a man who has already used up two of the other available treatment mechanisms is the actual question in front of the room.
After two TNF inhibitors, choosing what's next
I'd move to tocilizumab rather than a JAK inhibitor. ORAL Surveillance enrolled exactly his profile — age fifty or older with at least one cardiovascular risk factor — and found more MACE, cancer, and VTE with tofacitinib than with a TNF inhibitor, which is why that warning now covers the whole class, not just tofacitinib. He hasn't tried an IL-6 inhibitor yet, and it doesn't carry that warning.
I manage his lipids and his glucose, and I want to push back gently on how the warning gets applied here. His ten-year ASCVD risk is 9% — intermediate, not high — and he has no established atherosclerotic disease, which is precisely where the trial's own post-hoc work located most of the excess risk. A label written for the trial's enrolled population isn't automatically written for its lowest-risk member.
Both points are real, and I don't think they actually conflict as much as they sound like they do. The label is appropriately broad because the trial wasn't powered to tell us exactly where the risk starts mattering within that population — so treating the whole enrolled group as covered by the warning is the conservative, correct reading of the data as it stands, not an overreach.
Where I'd push back on starting with the JAK inhibitor anyway is that real-world registries haven't reproduced the signal, and the trial's own post-hoc analyses put the excess where he isn't: in enrollees with established atherosclerotic disease. Among those without an ASCVD history, MACE rates were similar on tofacitinib and on a TNF inhibitor. I want to be careful with one thing, though — the statin subanalysis people reach for in this conversation is about the ASCVD group specifically. His atorvastatin is good medicine, but it isn't evidence that describes him, and I'd rather not borrow reassurance from a subgroup he isn't in.
The honest version is that he sits in the part of the trial that didn't separate, with one qualifier against him: the cleanest null was in enrollees under 65 who had never smoked, and he's a former pack-a-day smoker. So I'd concede the JAK inhibitor isn't off the table forever, but I'd try tocilizumab first given he still has an untried mechanism with no comparable warning, and keep the JAK inhibitor as the next option rather than a last resort if this doesn't work.
Agreed: tocilizumab started as the next mechanism. The rheumatologist and primary care physician agreed explicitly that a JAK inhibitor is not ruled out, only deferred, with his existing statin therapy noted in the chart as a specific factor to weigh if tocilizumab doesn't work.
Not agreed: how literally the boxed warning should be read onto a patient whose calculated risk sits at the borderline-moderate end of what the trial's entry criteria would allow. The primary care physician's view is that the label reasonably admits of individual risk stratification; the rheumatologist's view is that a class-wide warning is meant to be applied to the labeled population as written, not re-derived per patient. The choice made today didn't require settling which reading is correct.