Anifrolumab for Skin-Only Lupus: A Drug Built for a Different Patient
Her disease has never once shown up anywhere but her skin, and it's winning. The disagreement is whether a drug built and tested for systemic lupus is a real option for her, or a mismatch dressed up as a shortcut past two established alternatives.
T.O., a 42-year-old woman who manages a bakery's early-morning shift and has spent the last eighteen months layering makeup over discoid plaques on her scalp and cheeks before opening, has subacute and discoid cutaneous lupus erythematosus confirmed by two separate skin biopsies, with a positive ANA and anti-Ro/SSA antibody but a persistently normal anti-dsDNA, normal complement, and no arthritis, no serositis, and no organ involvement in eighteen months of follow-up. She has failed hydroxychloroquine at maximum tolerated dose, a six-month trial of quinacrine added on top of it, and potent topical and intralesional corticosteroids; two of her discoid lesions have started to scar, one at the hairline visibly enough that she has begun wearing a scarf to work, something she says customers have started to ask her about directly, which she finds harder some mornings than the disease itself. She has no hypertension, no diabetes, and nothing else that would slow wound healing, so the lesions still advancing rather than settling into old scars are advancing on lupus alone.
Her SLEDAI-2K today is 2 — the two points come entirely from the malar/discoid rash item, with every systemic domain scoring zero, which is the actual shape of her problem: real, disfiguring, scarring disease that by the numbers looks nothing like the population TULIP-1 and TULIP-2 enrolled, both of which required a baseline SLEDAI of at least 6 with evidence of active disease beyond skin alone. That gap matters directly to what's being proposed today. What the 2025 ACR guideline actually names for her phenotype is narrower than it is often quoted as being: dapsone appears there as a conditional first addition for bullous lupus, which she does not have, while the drug it names for cutaneous disease refractory to antimalarials and immunosuppressants is lenalidomide. Dapsone is used for discoid and subacute disease on long practice and case-series evidence rather than on that recommendation, which does not make it wrong, only weaker-warranted than the phrase “guideline-recognized” suggests. Lenalidomide, meanwhile, is a thalidomide analog carrying the same REMS pregnancy program she has already refused thalidomide over, with substantially less peripheral neuropathy — so the option the guideline actually points to is the one her stated objection half applies to.
Joint clinic, eighteen months into refractory disease
She's told us directly she won't take thalidomide — the contraception requirements and the neuropathy risk aren't hypothetical concerns to her, they're a dealbreaker. Cutaneous lupus runs on the same interferon pathway anifrolumab blocks, and TULIP-1 and TULIP-2 both reported CLASI improvement as a secondary endpoint, with case series since then in refractory skin-predominant disease reporting the same thing outside the trials. I'd start it now rather than push her toward a drug she's already refused.
I want to name the actual gap here plainly: TULIP-1 and TULIP-2 required a baseline SLEDAI of 6 with systemic activity to enroll. Hers is 2, entirely from skin. The approval doesn't cover the population she's actually in — it covers a sicker, more systemic patient. And I want to be exact about dapsone, because we keep calling it guideline-recognized for her: the guideline names it for bullous lupus. For her phenotype it names lenalidomide. Dapsone is still the cheaper, better-monitored, more reversible thing to try, and she hasn't tried it — but let's argue for it on what it actually is, a well-worn practice option, rather than borrowing authority from a recommendation written about a different presentation.
I'm not dismissing the real-world signal — I'm saying it's real-world, not the trial population's own data, and that distinction should matter more than it's being given credit for right now.
You're right that the population mismatch is real — I'm not arguing otherwise. But dapsone typically takes four to six weeks to show real benefit, sometimes longer, and one of her lesions is already scarring at the hairline. That's not a number we get to redo later if it turns out too slow.
I'd try dapsone and anifrolumab in parallel rather than sequentially given the stakes on that specific lesion — not because the population argument is wrong, but because the cost of being slow here is permanent in a way the cost of overlapping two mechanisms isn't.
Agreed: start dapsone after G6PD screening, and start anifrolumab in parallel rather than waiting through dapsone's full onset, given the active scarring at the hairline. Hydroxychloroquine and quinacrine continue unchanged.
Not agreed: whether parallel initiation was the right call versus a stricter sequential trial of dapsone alone — the pharmacologist's concern about prescribing outside the population anifrolumab was actually studied in wasn't resolved, only outweighed for this specific lesion's timeline. If the scarring stabilizes on dapsone alone within six weeks, stopping anifrolumab rather than continuing both indefinitely is the explicitly named next decision point.