A Ruxolitinib Trial Won on Lesion Counts. Did It Ever Have the Power to Say Anything About Pain?

A letter published in the Journal of the American Academy of Dermatology this fall revisits a single number buried in a trial result from last November: in a 16 week phase 2 study of ruxolitinib cream for mild to moderate hidradenitis suppurativa, pain and itch scores improved by roughly the same amount whether patients got the active drug or plain vehicle cream. The letter, "Beyond abscess and inflammatory nodule counts: Reflecting on placebo and pain in the ruxolitinib trial," by Wan Yang and Menglong Li, argues that this detail deserves more scrutiny than a footnote.

The trial itself, published in JAAD in November 2025, tested 1.5% ruxolitinib cream (marketed as Opzelura) against vehicle in 69 adults with Hurley stage I or II disease. On its primary endpoint, lesion counts, the drug won: an LS mean change of -3.61 versus -2.42 for vehicle (P<0.05), and a HiSCR50 response rate of 79.2% versus 50.0% on vehicle alone. The trial authors attributed the flat pain and itch results to low baseline pain scores, meaning there was little room for either arm to improve.

Yang and Li's letter drew a direct reply, published in the same journal on June 2, 2026, under the title "Response to Yang and Li... Reflecting on placebo and pain in the ruxolitinib trial." Its authors are Martina Porter and Melinda Gooderham, both dermatology investigators tied to the original trial, and M. Celeste Ferreira-Cornwell, who is affiliated with Incyte, the company that makes Opzelura. That the rebuttal carries an industry co-author does not make it wrong, but it does mean readers should weigh the defense of the trial's design as coming partly from the sponsor whose drug is seeking a new indication, not solely from independent academics.

What the exchange actually tests is how much a 69-person, 16-week phase 2 study can tell anyone about pain relief. That sample size was built to detect a lesion-count signal, and it did. It was not built with the statistical power to separate two pain trajectories that may differ by a smaller margin than AN counts do. A trial can be genuine evidence for its primary endpoint and simultaneously too small to say anything reliable about a secondary one, which is the gap this letter and its rebuttal are arguing over without ever quite naming it that way.

The pattern is not unique to this drug. A cross-trial analysis covering 11 phase 2 and 3 hidradenitis suppurativa studies found a median placebo HiSCR50 rate of 31%, published in the British Journal of Dermatology. A related post hoc analysis found that patients classified as placebo responders showed real declines in CRP, a blood marker of inflammation, suggesting the high vehicle response in these trials is not simply noisy measurement but may reflect genuine, still unexplained disease fluctuation. Separately, pooled data from the PIONEER trials found that only about 47% of patients who hit the HiSCR lesion-count threshold also reached a clinically meaningful improvement on a pain scale, and only about 60% did on quality of life scores, meaning lesion counts and patient-felt improvement routinely diverge in this disease.

A comparable split showed up recently in a competing JAK inhibitor program: a phase 2 trial of upadacitinib in moderate to severe hidradenitis suppurativa, published in JAAD, found HiSCR50 rates of 38% on drug versus 25% on historical placebo, with pain response rates of 36% versus 22.5%, a similarly narrow margin. Read next to the ruxolitinib letter, it suggests the pain-placebo gap in this class of drugs is a recurring feature of small early-phase hidradenitis suppurativa trials generally, not a flaw isolated to one study.

Two phase 3 trials, TRuE-HS1 and TRuE-HS2, are already registered on ClinicalTrials.gov and will enroll far more patients over a longer window. Whether their protocols change how pain is measured, and whether they arrive with enough statistical power to say something definitive about it, is the question this letter leaves open rather than answers.