Oral JAK inhibitors rapidly calmed severe drug eruptions in small hospital study

When a drug eruption keeps spreading despite the culprit medication being withdrawn and systemic corticosteroids already running at full dose, dermatologists have few well-evidenced options. A new retrospective observational case series from Shenzhen Nanshan People’s Hospital, published in Frontiers in Medicine, reports that adding an oral Janus kinase (JAK) inhibitor to ongoing treatment was associated with strikingly rapid disease control in nine hospitalized patients with extensive or difficult-to-control drug eruptions. In seven patients classified as a rescue treatment subgroup because their eruptions were still actively progressing after at least 48 hours of systemic corticosteroids, new lesions were first documented as absent within 24 hours in six of seven patients and within 48 hours in all seven. All seven began corticosteroid dose reduction within 96 hours, with a median interval of 48 hours after JAK inhibitor initiation.

The study reviewed all hospitalized patients with a clinically diagnosed drug eruption who received an oral JAK inhibitor at the department between March 2023 and January 2026. Nine patients met the eligibility criteria: five men and four women with a median age of 52 years. Their phenotypes ranged from extensive maculopapular eruptions, some with DRESS-like features, to two cases of definite drug reaction with eosinophilia and systemic symptoms (DRESS) scored by the RegiSCAR system, one probable acute generalized exanthematous pustulosis (AGEP), one erythema multiforme-like eruption, and one suspected maculopapular eruption superimposed on preexisting erythrodermic atopic dermatitis. Baseline body surface area (BSA) involvement ranged from 30% to 90%, although the 90% value in the atopic dermatitis case predominantly reflected the preexisting erythroderma rather than the drug eruption alone.

The authors deliberately separated treatment contexts rather than pooling all patients. Cases 1 through 7 formed the rescue subgroup: each had documented continued new-lesion formation during the 24 hours before JAK initiation, together with expanding BSA involvement, worsening erythema or edema, or uncontrolled severe pruritus, despite 48 to 96 hours of systemic methylprednisolone. Five of the seven had also received etanercept or intravenous immunoglobulin before or at the time of JAK initiation. Case 8, a patient with definite DRESS whose new lesions had already stopped forming, received upadacitinib instead to reduce residual inflammatory burden and enable steroid tapering. Case 9, an elderly woman with severe DRESS affecting liver and kidney, began baricitinib concurrently with corticosteroids and immunoglobulin, a sequence that made any independent contribution of the JAK inhibitor impossible to isolate.

Four different oral JAK inhibitors were used across the cohort: upadacitinib in three patients, tofacitinib in three, abrocitinib in two and baricitinib in one. Selection was individualized and off label, driven by phenotype, severity, comorbidities, affordability and clinical experience rather than any standardized protocol. The authors emphasize that the regimens reported, with doses and switching patterns documented patient by patient, should not be interpreted as dosing recommendations. Day 0 was defined as the first actual administration of the JAK inhibitor, and two dermatologists independently verified phenotypes, timelines, itch scores, lesion status, corticosteroid changes and safety outcomes from electronic records, with disagreements resolved by consensus.

The quantitative serial data were consistent with the timeline findings. Across all nine patients, the median Worst Itch Numerical Rating Scale score fell from 6 at baseline to 4 at 24 hours, 2 at 48 hours, 1 at 72 hours and 1 by day 7 or discharge. Every patient achieved a reduction of at least 3 points by 72 hours, a threshold the authors used as a descriptive marker of clinically important improvement. Median BSA involvement decreased from 55% at baseline to 30% at day 3 and 10% by day 7 or discharge. A sensitivity analysis excluding the patient whose baseline BSA was confounded by erythrodermic atopic dermatitis produced nearly identical results, with median involvement falling from 55% to 27.5% at day 3 and 10% by day 7 or discharge.

The statistical analyses were formal within-patient repeated-measures tests, and the authors are careful about what they show. Changes in WI-NRS across five time points yielded a Friedman test statistic of 35.04 with P less than 0.001 and a Kendall’s W of 0.973, indicating a highly consistent direction of improvement across individuals. Changes in BSA across three time points gave a Friedman statistic of 18.00, again with P less than 0.001 and Kendall’s W of 1.000. Post hoc comparisons of baseline with each follow-up point remained significant after Holm correction. However, because there was no untreated or conventional-treatment comparator group and most patients received concomitant systemic therapy, the authors state explicitly that these results demonstrate within-patient temporal change, not an independent JAK inhibitor treatment effect.

Safety findings deserve equal attention. Two herpes simplex-related events occurred during combined immunosuppression: one patient with preexisting HSV-2 infection experienced recurrent HSV-2 amid extensive concomitant immunosuppression, and another developed clinically diagnosed perioral herpes simplex while taking upadacitinib with corticosteroids. Two other patients developed marked hyperglycemia considered primarily related to systemic corticosteroid treatment. No new thrombosis, major cardiovascular event or clearly JAK-attributed hepatic or renal deterioration was documented, but the authors caution that laboratory monitoring and follow-up duration were incomplete in several patients, and the absence of documented events should not be read as evidence of safety. This caveat is particularly relevant given the known increased infection risk, including herpesvirus infection, associated with oral JAK inhibitors in dermatologic populations.

The biological rationale for the approach is substantial. Drug-responsive T cells in delayed drug hypersensitivity activate overlapping type 1, type 2, cytotoxic and pruritic cytokine programs, many of which depend on JAK-STAT signaling. Single-cell transcriptomic work in corticosteroid-refractory DRESS demonstrated pathway activation and guided successful tofacitinib treatment, while functional experiments showed that CCR8/CCL1-mediated type 2 memory responses and CXCR3/CXCL10-mediated cytotoxic T-cell activation in recalcitrant DRESS were attenuated by JAK inhibition. Spatial proteomic analysis of toxic epidermal necrolysis has identified interferon-STAT1 signaling as a major pathogenic program. Because JAK inhibition suppresses several downstream inflammatory circuits simultaneously rather than targeting one cytokine, the authors argue it may complement upstream treatments such as etanercept or immunoglobulin.

The study’s limitations are considerable and honestly stated. It was small, single-center, retrospective and uncontrolled, with heterogeneous phenotypes, individualized regimens and substantial concomitant treatment. New-lesion cessation was assessed at intermittent routine documentation points around 24, 48 and 72 hours, so the reported timings represent the first assessment documenting no further lesion formation rather than the exact biological moment lesions stopped. WI-NRS and BSA values were retrospectively reconstructed from routine records, diagnostic certainty and culprit drug attribution varied across cases, and follow-up was incomplete. Culprit drug withdrawal, corticosteroids, etanercept, immunoglobulin, supportive care, regression to the mean and the natural course of the eruptions may all have contributed to improvement. The retrospective rescue-treatment classification itself requires prospective validation.

Even with those constraints, the consistency of the temporal pattern is what makes the report notable. Every rescue-subpatient patient stopped forming new lesions within 48 hours of JAK inhibitor addition and began steroid tapering within 96 hours, a median interval far shorter than the 7 to 12 days reported before initial tapering in a published six-patient series of refractory DRESS treated with baricitinib or abrocitinib. The authors do not advocate routine JAK inhibitor use for uncomplicated maculopapular eruptions, which generally resolve with culprit drug withdrawal and symptomatic care. Instead, they position the findings as support for prospective controlled evaluation in selected patients with persistent progression, severe itch or heavy corticosteroid exposure, ideally incorporating immune biomarkers such as interleukin-5, CCL1, CXCL10 and phosphorylated STAT1 to identify who benefits most. Until such studies are done, these results stand as an encouraging but uncontrolled signal that rapid pathway-directed rescue therapy may be feasible in some of the hardest-to-treat drug eruptions.

Subject of Research: Adjunctive oral JAK inhibitor treatment for severe or difficult-to-control drug eruptions

Article Title: Efficacy of adjunctive oral Janus kinase inhibitors in hospitalized patients with extensive or difficult-to-control drug eruptions: a retrospective observational case series

Article References: Zhu, Y., Wang, X., Huang, Y., Wang, M., Lin, L., Li, J., Feng, P., Lu, Y., Liu, X., Chai, B., & Wang, P. (2026). Efficacy of adjunctive oral Janus kinase inhibitors in hospitalized patients with extensive or difficult-to-control drug eruptions: a retrospective observational case series. Frontiers in Medicine, 13, Article 1947886. https://doi.org/10.3389/fmed.2026.1947886

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1947886

Keywords: JAK inhibitors, drug eruptions, DRESS, AGEP, corticosteroids, pruritus, dermatology, case series, rescue treatment, herpes simplex, retrospective study, steroid-sparing