Stopping a Janus kinase inhibitor can undo months of calm skin in a matter of days, and a new narrative review argues that the reason lies in the drugs’ own molecular design. Writing in Discover Immunity, a team led by Soroush Moradi of Tehran University of Medical Sciences synthesizes evidence on why inflammatory skin diseases such as atopic dermatitis, alopecia areata, vitiligo and psoriasis so often flare after treatment ends, and proposes a framework for safer deprescribing built around gradual dose reduction rather than abrupt cessation. The review, published open access on 2 November 2026, draws together randomized discontinuation trials, retrospective cohorts and mechanistic studies to explain what the authors call the relapse phenomenon.
The central mechanism concerns Type I JAK inhibitors, the class that includes ruxolitinib, tofacitinib, baricitinib and upadacitinib. These molecules compete with ATP for the kinase’s binding pocket and block signaling to STAT proteins, cutting off downstream inflammatory transcription. Crucially, however, they bind the kinase in its active, phosphorylated conformation. The review explains that this binding paradoxically stabilizes the enzyme in a primed, hyperphosphorylated state, restrained only by the presence of the drug. Most of the supporting evidence comes from non-dermatological models, particularly myeloproliferative neoplasms, and the authors are careful to note that direct validation in human inflammatory skin disease is still lacking, though they consider the extrapolated mechanism a highly plausible explanation for the sudden bursts of cytokine transcription seen clinically.
A second mechanism compounds the first. Under normal conditions, activated STAT proteins induce the expression of Suppressors of Cytokine Signaling, or SOCS proteins, which bind to JAK enzymes and degrade them, acting as a negative feedback brake on the pathway. Because JAK inhibitors suppress STAT signaling during treatment, they also suppress transcription of SOCS1 and SOCS3. When the drug is withdrawn, the review argues, a temporary molecular imbalance arises: an accumulation of hyperphosphorylated active JAKs meets depleted levels of the very proteins that would normally dampen the signal. The authors emphasize that the exact dynamics have yet to be definitively mapped in specific human dermatoses, but theorize that this lack of feedback permits unchecked signal amplification, facilitating the acute inflammatory flares observed after relapse.
The speed of the crash is dictated by pharmacokinetics. Unlike biologics, whose long elimination half-lives produce a self-tapering effect, oral JAK inhibitors have half-lives of roughly 2 to 12 hours. Within 24 hours of the last dose, serum concentrations fall below the IC50, a drop the authors describe as a pharmacokinetic cliff: receptor occupancy collapses and cytokine suppression is lost almost instantaneously. The review proposes that relapse potential reflects an interaction between drug clearance rate and the velocity of the underlying disease pathway, with rapid neural or cytotoxic pathologies reacting far faster than conditions driven by slower structural tissue remodeling.
Clinical data bear out the hierarchy. In atopic dermatitis, a phase 2b randomized trial of upadacitinib found that pruritus returned within less than a week of discontinuation, with EASI scores rising within a month and disease control regained promptly upon resumption. A comparative analysis reported that time to relapse was significantly shorter for selective JAK1 inhibitors than for dupilumab, roughly two months versus about one and a half years, with more than half of JAK1 inhibitor patients relapsing compared with dupilumab patients, a hazard ratio of 2.58. In a phase 2b study of abrocitinib monotherapy, fewer than one-third of patients sustained IGA 0/1 and EASI-75 responses during a four-week drug holiday, and inflammatory biomarkers returned to normal levels within four weeks of stopping. The review also cites a report of eight patients with inflammatory bowel disease who developed new eczematous lesions one week after JAK inhibitor discontinuation, two confirmed by biopsy as atopic dermatitis, all resolving when the drugs were restarted.
Alopecia areata shows some of the steepest discontinuation curves. In the randomized BRAVE-AA1 trial, 154 patients with severe alopecia areata who achieved disease control after 52 weeks of baricitinib were re-randomized to continue treatment or switch to placebo. By week 152, relapse, defined as worsening of more than 20 SALT points, occurred in approximately 80 percent of those who discontinued, against 7 percent of those who continued. No relapses appeared in the first four weeks, began as early as week 8 in a minority, and exceeded half by week 24. Relapses ranged from gradual thinning to rapid near-complete hair loss, but no rebound phenomena were reported, and retreatment restored response in a significant proportion of relapsed patients, though complete remission was uncommon. A randomized phase 2a cessation study of ritlecitinib and brepocitinib found median times to relapse of 16.1 and 24.1 weeks respectively, with only 18 to 39 percent maintaining regrowth during the drug-free period.
Vitiligo and psoriasis behave differently. Topical ruxolitinib, the only US FDA-approved JAK inhibitor for vitiligo, showed in an exploratory analysis that 75 percent of 16 responders maintained repigmentation after discontinuation while 25 percent experienced minor loss at lower doses, with baseline serum CXCL9, CXCL10 and interleukin-15 levels failing to distinguish relapsers from maintainers. The long-term TRuE-V extension study found that approximately 70 percent of patients who continued ruxolitinib cream maintained successful repigmentation at one year, compared with 40 percent of those who discontinued, with relapse in roughly one-third of the discontinuation group and nearly half of those relapses within the first four months. In psoriasis, the OPT Retreatment trial of tofacitinib showed most patients losing 50 percent of their PASI improvement within about 16 weeks, with gradual worsening and no rebound, while phase 3 deucravacitinib discontinuation data showed roughly one-third maintaining PASI75 responses by week 28 and a median time to relapse exceeding 28 weeks.
On this evidence the review builds what it calls a velocity-of-pathology framework. High-velocity diseases involving neural signaling, such as atopic dermatitis, or cytotoxic effector release, such as alopecia areata, can reactivate within hours, particularly when paired with short half-life agents like tofacitinib or abrocitinib, whose cessation allows rapid resensitization of pruritogenic sensory nerves or near-immediate disinhibition of cytotoxic CD8-positive NKG2D-positive T cells attacking the hair follicle. Low-velocity diseases involving structural remodeling, such as psoriasis and vitiligo repigmentation, possess biological inertia: even with short half-life drugs, the pathology takes weeks to re-establish, so the outcome is gradual relapse rather than rebound. The authors note that alopecia areata treated with pan-JAK inhibitors carries the highest reported risk of true rebound, defined as worsening beyond the pre-treatment baseline, while psoriasis treated with TYK2 inhibitors exhibits the lowest risk.
For management, the review advocates a slope-of-reduction principle: taper velocity should be inversely proportional to relapse potential. The primary objective is to identify a minimum effective dose that maintains adequate cytokine suppression while minimizing systemic exposure, and the authors argue that dose de-escalation, which preserves the dosing interval and steady-state concentrations, is preferable to frequency extension, which for drugs with 3 to 12 hour half-lives creates washout periods of 36 to 40 hours during which serum levels fall well below the IC50. A 50 percent rule, drawn from dermatologic guidelines, caps the initial dose reduction at half the current therapeutic dose. Supporting evidence includes the Flex Up trial, in which 68.5 percent of patients maintained an EASI-90 response after reducing upadacitinib from 30 to 15 milligrams, and 48.1 percent achieved EASI-90 after re-escalation from 15 to 30 milligrams, and a retrospective study showing that complete discontinuation in moderate-to-severe atopic dermatitis leads to relapse in over 70 percent of cases within 8 weeks. For alopecia areata, the authors recommend deferring tapering of baricitinib or tofacitinib until full regrowth, a SALT score below 10, has been maintained for six to twelve months, with immediate re-escalation if the score rises by more than 5 to 10 points.
The authors are candid about the limits of the evidence base. Because no established protocol exists for continuing JAK inhibitors after clinical improvement, most experts rely on clinical experience rather than evidence-based guidance, and clinical studies specifically designed to assess true disease modification through longitudinal molecular characterization and sustained off-treatment remission are lacking. Mechanistic claims rest heavily on preclinical models that inadequately represent human immunological memory, and a murine vitiligo study demonstrating persistence of skin-resident T cells despite JAK inhibition cannot be automatically generalized to humans. The review calls for future randomized trials explicitly evaluating tapering slopes and therapeutic bridging with topical agents or phototherapy, for Type II or allosteric inhibitors such as deucravacitinib that stabilize kinases in an inactive conformation, and for tissue-based rather than serum biomarkers capable of identifying patients whose disease carries latent volatility requiring ongoing suppression.
Subject of Research: Molecular mechanisms and tapering strategies for relapse after discontinuing Janus kinase inhibitors in inflammatory skin diseases
Article Title: Molecular mechanisms and clinical tapering strategies for Janus kinase inhibitor relapse in inflammatory dermatoses
Article References: Moradi, S., Rostami, A., Hasani, V., & Shariat Rad, P. (2026). Molecular mechanisms and clinical tapering strategies for Janus kinase inhibitor relapse in inflammatory dermatoses. Discover Immunity, 3(1), Article 21. https://doi.org/10.1007/s44368-026-00038-7
Image Credits: AI Generated
DOI: 10.1007/s44368-026-00038-7
Keywords: JAK inhibitors, atopic dermatitis, alopecia areata, vitiligo, psoriasis, relapse, rebound, pharmacokinetic cliff, SOCS proteins, dose tapering, Type I inhibitors, deucravacitinib
News Source: Louis Brooks. (November 5, 2026). JAK inhibitor relapse traced to primed enzymes and a pharmacokinetic cliff. Scienmag.
