Two Boxed Warnings, One New Diagnosis: Steroid-Sparing Choice in Cranial GCA
A single patient, newly diagnosed with cranial giant cell arteritis. The disagreement isn't over whether she needs a steroid-sparing agent — it's over which of two real, documented risks in her own history should carry more weight against the other.
Rosalind T., 74, spent thirty years as a hospice chaplain and still leads a weekly grief-support group out of her church basement — work she has no intention of giving up now that a headache has put her in a rheumatologist's office instead of leading it. The headache started three weeks ago along her right temple, joined a week later by an ache in her jaw that stopped her chewing a piece of bread, and four days ago by ninety seconds of gray fog across her right eye that cleared before she'd finished being frightened by it. A temporal artery biopsy came back showing the granulomatous panarteritis of giant cell arteritis; her ESR and CRP, 78 and 62, have started falling on the prednisone 60mg a day started two weeks ago, and the amaurosis hasn't recurred. The urgency of that first visit has passed. What's left is the steroid-sparing choice the team now has to make, and her chart carries two real facts pulling in opposite directions. Three years ago she was hospitalized for diverticulitis — IV antibiotics, five days on a ward, no perforation, no surgery — but a real, documented episode of exactly the disease process tocilizumab's own labeling flags for gastrointestinal perforation risk. She has also been in atrial fibrillation for six years, anticoagulated on apixaban, which puts her baseline thrombotic risk already under active management before anyone adds a JAK inhibitor's own boxed warning for the same category of harm.
Tocilizumab, weekly subcutaneous injection, is the drug GiACTA actually built GCA's evidence base around: more than half of that trial's weekly-dosed patients reached sustained glucocorticoid-free remission at week 52, against roughly one in seven on placebo. Upadacitinib is newer, oral, FDA-approved for GCA only since last year on the strength of SELECT-GCA — a smaller but real placebo-controlled margin, just under half reaching sustained remission against about three in ten on placebo. Neither trial excluded a patient with her diverticulitis history outright, and the two were never randomized against each other, so choosing between them means reading two different populations against her own chart rather than reading one trial against its comparator. The bowel-perforation caution attached to IL-6 blockade was drawn overwhelmingly from patients with active or recent diverticular disease, not a remote, resolved episode three years cold — while ORAL Surveillance, the trial underwriting the JAK class's own vascular warning, was built specifically to stress-test a cardiovascular-risk-enriched rheumatoid arthritis population, not an anticoagulated one. Which of her two three-years-apart facts actually predicts what happens to her is the whole question in the room.
Choosing the steroid-sparing agent, two weeks in
Her diverticulitis is the fact I'd weight first, because it already happened to her specifically — it isn't a population-level statistic, it's a bowel that has already shown it can perforate-adjacent inflammation under stress. Tocilizumab's labeling names exactly that population for perforation risk. I'd start upadacitinib.
She's also told us plainly she doesn't want another weekly injection added to a regimen that already includes apixaban and a bisphosphonate — that's a real preference, not the deciding factor, but it points the same direction.
I'd read the JAK warning more literally than that. ORAL Surveillance is the trial the entire class warning rests on, and it was deliberately enriched for cardiovascular risk factors to find exactly this signal — it's doing its job. She's not a hypothetical cardiovascular risk; she has documented atrial fibrillation and is already anticoagulated for it. Stacking a JAK inhibitor's own thrombotic signal on top of a person whose baseline risk already required lifelong anticoagulation isn't a theoretical concern, it's the same category of harm doubled.
The perforation caution you're citing concentrates in patients with active or recent diverticular disease — hers was three years ago, treated medically, fully resolved. That's a colder risk than yours.
Both of you are extrapolating, and neither trial was built for her specific overlap. But I'd separate what's structural from what's dynamic: her diverticulitis reflects a colonic wall vulnerability that doesn't go away just because it's quiet, while her atrial fibrillation is a monitored, actively managed variable — she's already anticoagulated and followed by cardiology, and any new thrombotic signal would show up in a relationship that's already being watched. I'd start upadacitinib, coordinate a check-in with her cardiologist before the first dose, and treat the diverticulitis history as the less recoverable of her two risks if either one materializes.
Agreed: start upadacitinib 15mg daily alongside the planned 26-week prednisone taper, with a cardiology check-in scheduled before the first dose given her anticoagulated atrial fibrillation.
Not fully settled: how tightly to monitor afterward. The rheumatologist wants a lower threshold for stopping upadacitinib at the first new cardiovascular symptom, given the class warning; the geriatrician thinks that threshold, applied to every anticoagulated patient on a JAK inhibitor, would functionally rule the drug out for a population it may still serve well — and would rather judge symptoms as they arise than pre-commit to a stopping rule neither trial actually tested.