The Eltrekibart Trial's Headline Number Didn't Come From Its Own Placebo Arm

A letter responding to a Lilly-sponsored hidradenitis suppurativa trial appeared online October 6, 2026 in the Journal of the American Academy of Dermatology, exactly one year after the trial report it addresses. The underlying paper, by Forman and colleagues, published in JAAD on October 6, 2025 and later appeared in the February 2026 print issue (pages 530 to 538). The response, by Nanxiang Zhang and Meirong Xu, is not yet indexed in full text at its ScienceDirect listing, so this account relies on the original trial record and the public data trail around it, not on the letter's own wording.

The trial being scrutinized tested eltrekibart, an antibody that neutralizes the CXCR1/CXCR2 chemokine ligands thought to drive neutrophil activity in hidradenitis suppurativa. Sixty-seven patients were randomized 2 to 1 to drug or placebo across 17 sites in the United States and Australia, running from 2020 to 2022 under the registration NCT04493502. The trial's own prespecified primary endpoint, HiSCR50 response at week 16, came in at 48.9 percent for eltrekibart (47 patients) against 31.8 percent for placebo (20 patients), for a P value of .19. By the conventional threshold of .05, that result does not separate drug from placebo.

The number that circulated in coverage of the paper was different: 65.6 percent versus 32.3 percent, described as a 99.9 percent posterior probability of superiority. That figure did not come from comparing the trial's two randomized arms. It came from a Bayesian analysis that augmented the trial's small placebo group with historical placebo data drawn from separate phase 3 programs, a technique sometimes called borrowing. The method is established in some rare-disease and small-trial settings, but it answers a different question than the one the randomized comparison was built to answer, and a 67-patient trial with a P value of .19 on its primary endpoint is the base fact a reader has to hold alongside the borrowed-control figure.

Two of the trial's listed authors, Alice Nirula and Paul Klekotka, work for Eli Lilly, which developed eltrekibart under the designation LY3041658. That is standard disclosure for an industry-funded study, not evidence of misconduct on its own, but it sharpens the question of which analysis a sponsor chooses to foreground when the primary one falls short. The paper's conclusion reportedly used language describing eltrekibart's effect as disease-modifying, a claim that sits uneasily next to a missed primary endpoint in a 16-week, single-region trial with a placebo arm of only 20 patients.

Lilly did not wait for a definitive signal before moving forward. A second trial, NCT06046729, is already underway: more than 350 patients, subcutaneous dosing instead of the original trial's intravenous delivery, and a dose-finding design, with completion listed around mid-2026 on the company's own trial registry. That sequencing, a larger confirmatory study launched before the smaller one's contested result could be independently replicated, is itself a pattern worth watching rather than a verdict on the drug.

The field around eltrekibart has also shifted since the 2020 trial design. The FDA approved bimekizumab for moderate to severe hidradenitis suppurativa on November 20, 2024, giving the disease a second approved biologic alongside adalimumab and one with phase 3 data showing durable HiSCR50 and HiSCR75 responses through week 48. A Spherix Global Insights dermatologist survey from December 2025 found prescribing intent consolidating toward IL-17 inhibitors, with bimekizumab positioned as the primary beneficiary. Any eltrekibart approval would land in a market that already has an answer with a stronger evidence base than a P value of .19.

What the October 2026 letter specifically argues, whether it targets the borrowed-control statistics, the disease-modifying language, or something else entirely in the Forman paper, remains unconfirmed pending access to the full text at its ScienceDirect listing. The gap between the trial's primary result and the number that made it into headlines is already sitting in the public record, waiting for scrutiny, with or without the letter attached to it.