Moderate-to-Severe Atopic Dermatitis: A JAK Inhibitor Against a Clotting History
Her eczema has failed two potent topical regimens and a course of a topical calcineurin inhibitor. The fastest, most effective oral option carries a boxed warning that lands directly on the one thing in her history nobody can talk around.
Renata S., a 29-year-old pastry chef, has had eczema since she was small enough that her mother still keeps a photograph of her in oven mitts she wasn't supposed to be wearing to bed. It has never fully gone away, but the last eight months have been the worst stretch in years — flares across both forearms and the backs of her knees that no longer respond to the potent topical steroid her dermatologist started, or to six weeks of tacrolimus ointment layered on top of it. She is up three or four times a night scratching, and has started trading away the 5 a.m. prep shift she used to volunteer for because she can't reliably wake up functional.
Her exam today puts roughly a third of her body surface involved, an EASI score solidly in the severe range, and excoriations deep enough on both forearms to worry about secondary infection if nothing changes soon. The obvious next step is a systemic agent, and on pure speed and depth of effect the oral JAK1 inhibitor upadacitinib is the strongest option on the table — the Measure Up 1 and Measure Up 2 trials showed itch improvement measured in days rather than weeks against placebo, and Heads Up, the one trial that put the two drugs directly against each other, gave upadacitinib the higher EASI-75 rate at week 16. The complication sitting in her chart is a single, unprovoked deep vein thrombosis at 26, thoroughly worked up afterward with no thrombophilia found, six months of anticoagulation, and nothing since. That single line is exactly the risk category named in the boxed warning now carried by every drug in the JAK class, regardless of how different the AD population looks from the rheumatoid arthritis cohort where the signal was first measured.
She has otherwise been a healthy 29-year-old apart from the DVT and the eczema itself — no other chronic conditions, no other medications, a family history that includes her mother's own milder eczema but nothing suggesting an inherited clotting disorder beyond what her own post-DVT workup already ruled out. That baseline matters here specifically because it removes the easiest reason to discount the boxed warning as irrelevant noise: there is no competing explanation for her prior clot, no transient provoking factor beyond a long flight, which is exactly the profile that makes a clinician take an unprovoked-VTE history seriously rather than filing it as a closed, unrelated chapter. She has said plainly that she wants to feel like herself again before her sister's wedding in four months, which gives the group a real, if soft, timeline to weigh against whichever path they choose.
Weighing the fastest option against her own chart
If we're being honest about her actual quality of life, upadacitinib gets her further, faster, than anything else we could start today. She's not failed one topical regimen, she's failed two, and she's now missing income over this.
I want to be clear I'm not dismissing the warning — I'm asking whether a boxed warning built from a rheumatoid arthritis trial in an older, comorbidity-heavy population is the same warning in a 29-year-old with one clean unprovoked event and no thrombophilia on formal workup.
I'd push back on treating the population difference as doing more work than it actually does. ORAL Surveillance is the trial that produced the class-wide warning, and yes, it enrolled an older RA population — but the FDA's own extension of that warning across every JAK inhibitor, in every approved indication including AD, was a deliberate choice to treat the mechanism, not the population, as the source of risk.
And unprovoked VTE is specifically named in that rationale as one of the risk factors the warning is meant to flag for. That's not a generic caution being read too literally — it's a warning built for exactly the patient in front of us.
Dupilumab doesn't close this gap by magic — it closes it by not touching a pathway anyone has ever connected to clotting. It's slower. Most patients don't see real separation from placebo on itch until somewhere in the second or third week, and full skin clearance takes longer than that. But slower control and a reopened clotting risk are not costs of the same size.
I'd rather give her six honest weeks of a drug with no plausible mechanism for the one thing she's already lived through, than the fastest drug in the room with a black box that happens to name her specific history.
Agreed: start dupilumab, continue topical tacrolimus on flexural sites in the interim, and set a formal re-check at week 12 rather than judging response earlier than the drug's own trial timeline would predict.
Not agreed, and left explicit rather than smoothed over: whether upadacitinib should be reconsidered if dupilumab underperforms. The pharmacologist's reading of the warning doesn't soften with time; the dermatologist's view is that a clean twelve weeks on anticoagulation-adjacent monitoring, if it ever came to that, could change the calculus. Nothing was decided about that contingency — only that reaching it would require its own conversation, not a default fallback.