JAK Inhibitor Therapy in Ulcerative Colitis: A Boxed Warning Borrowed From Another Population
A single patient with moderate-to-severe ulcerative colitis, weighing a JAK inhibitor against a class-wide boxed warning built almost entirely from an older rheumatoid arthritis population she does not resemble.
Dana K., a 29-year-old woman, teaches middle school earth science and is eleven weeks into training for her first half marathon, a plan now stalled by six to eight loose, bloody stools a day and a level of fatigue that has her sitting through her own planning period instead of grading. She was diagnosed with ulcerative colitis three years ago, initially controlled on mesalamine alone, then escalated to infliximab fourteen months ago when a flare didn't respond to budesonide. Infliximab worked well for the better part of a year before her symptoms crept back over the last two months despite dose intensification and a trough level that came back therapeutic — a true secondary loss of response, not underdosing. She has never smoked, takes no other medications, has never had a blood clot, and her only family cardiac history is a grandfather's heart attack at 81.
Her Mayo endoscopic subscore is 2, calprotectin 480, and the group's next move is a JAK inhibitor — tofacitinib or upadacitinib, both approved for moderate-to-severe UC after anti-TNF failure, which is exactly where she now sits. The complication is the boxed warning both drugs carry: increased risk of major cardiovascular events, malignancy, and venous thromboembolism, added class-wide across every JAK inhibitor in 2021 after ORAL Surveillance, a postmarketing safety trial run not in inflammatory bowel disease but in rheumatoid arthritis, and not in a general RA population but specifically in patients 50 and older with at least one additional cardiovascular risk factor. Tofacitinib controlled disease comparably to the TNF-inhibitor comparator in that trial and still failed it, because the endpoints being formally tested for noninferiority were the safety ones — and on both, MACE and malignancy, tofacitinib did not meet the noninferiority criterion. Dana is 29, has never smoked, and has none of the risk factors ORAL Surveillance actually enrolled for — the question in front of the group is how much of a warning built on that specific population is a warning about her.
Clinic, deciding the next line of therapy
Upadacitinib is where I'd go next, and I don't think the boxed warning changes that for her specifically. ORAL Surveillance, published in the New England Journal in 2022, enrolled rheumatoid arthritis patients age 50 and up who already carried at least one additional cardiovascular risk factor by design — that's not incidental to the result, it's the population the trial was built to stress-test. The excess MACE and malignancy tofacitinib showed against a TNF-inhibitor comparator came out of that specific group. Dana is 29 with a clean cardiovascular history. The FDA's own boxed-warning language, once you read past the headline, is written around patients 50 and older with at least one cardiovascular risk factor — she isn't that patient, and U-ACHIEVE and U-ACCOMPLISH, the actual ulcerative colitis trials for upadacitinib, enrolled a materially younger population without reproducing that signal.
I'd slow down before treating this as a population-mismatch problem that resolves itself. The FDA didn't write a boxed warning scoped to age 50-plus-risk-factor — it applied the warning class-wide, to every JAK inhibitor, across every approved indication, specifically because the agency judged the risk couldn't be confidently fenced off by age or comorbidity based on the data available. The malignancy signal in ORAL Surveillance included lymphoma and lung cancer, and while current smoking concentrated the lung-cancer risk, the lymphoma signal wasn't as cleanly age-restricted. I've watched colleagues reason their way past a class warning with an argument this clean before and regret it once more follow-up data came in.
I take the regulatory point — I'm not disputing what the label says. What I'm disputing is treating 'the FDA applied it class-wide' as evidence the risk actually generalizes, rather than evidence of how conservatively the agency chose to write the label given how the trial was designed and how limited long-term IBD-specific safety data still is.
Both of you are right about different parts of this. The mechanism supports the distinction she's drawing: upadacitinib is JAK1-selective rather than tofacitinib's broader JAK1/JAK3 inhibition, and U-ACHIEVE/U-ACCOMPLISH's own population looked more like her than ORAL Surveillance's did. But the label doesn't carve out an exception for her risk profile, and prescribing as if it did would be substituting our own risk stratification for the FDA's written one. The honest way to hold both of those is to proceed — her disease is active and she's already failed a TNF inhibitor — while documenting the counseling explicitly: no smoking, no personal or family VTE history, contraception review since JAK inhibitors carry teratogenicity concerns, and a plan to reassess if any cardiovascular risk factor emerges. That's not reinterpreting the warning; it's applying it to the specific patient it's actually meant to flag for.
Agreed: upadacitinib 45mg daily induction, with explicit boxed-warning counseling documented in the chart — no smoking, no personal or family VTE history, contraception reviewed given teratogenicity risk. The rheumatologist's caution wasn't overruled so much as folded into the monitoring plan: lipid panel and CBC at baseline and 12 weeks, and an explicit instruction to flag any new cardiovascular risk factor for reassessment rather than treating today's decision as fixed regardless of what develops.