Atopic Dermatitis Linked to Higher Incident Psoriasis Risk

Key Takeaways

• In a US electronic health record cohort, atopic dermatitis was associated with significantly higher incident psoriasis risk than in matched controls without atopic dermatitis.
• Among systemically treated patients, biologics were associated with lower psoriasis incidence than conventional systemic immunosuppressants.
• The atopic dermatitis association remained positive but weakened when acne was the comparator.
• The lower incidence associated with biologics generally persisted across evaluable sensitivity analyses, although the treatment-comparison lag analysis was only nominally significant.

Atopic dermatitis (AD) and psoriasis are traditionally considered to be immunologically distinct, yet their clinical presentations can overlap. Psoriasis emerging during systemic AD treatment raises a difficult question: does it reflect underlying disease-associated risk, treatment exposure, or both? Patients receiving systemic therapy provide an opportunity to examine the disease association separately from the treatment comparison.

Investigators conducted a preregistered retrospective cohort study of incident psoriasis risk in atopic dermatitis using the United States (US)-based TriNetX US Collaborative Network federated electronic health records (EHRs) platform after scoping and a design freeze. They matched patients in pairs on 14 covariates. The disease comparison included 351,069 matched pairs with a first-recorded AD diagnosis or a general-health-examination control encounter without AD. The treatment comparison included 5,439 matched pairs with AD, classified by first eligible systemic treatment. An AD diagnosis had to be recorded near treatment initiation; prior exposure to the opposing treatment class or Janus kinase inhibitors excluded eligibility. The biologic cohort included dupilumab, tralokinumab, and lebrikizumab, but use of the latter two was negligible; the pooled comparison was predominantly informed by dupilumab. Conventional systemic immunosuppressants (cvIS) included methotrexate, azathioprine, mycophenolate mofetil, and ciclosporin. Incident psoriasis vulgaris was the endpoint. Follow-up began 1 day after index and was capped at 2,595 days for AD versus controls and 601 days for biologics versus cvIS; investigators also conducted sensitivity checks.

Psoriasis was documented in 1,088 of 351,069 patients with AD (0.31%) versus 290 of 351,069 matched controls (0.08%). The primary hazard ratio (HR) was 3.81 (95% confidence interval [CI], 3.35–4.34). The association remained in the same direction across prespecified sensitivity checks. With acne as the comparator, the association attenuated to HR 1.73 in the Kaplan–Meier analysis and HR 1.39 in a post-hoc multivariable Cox analysis, compared with the primary HR of 3.81 against general-health-examination controls.

Psoriasis was documented in 14 of 5,424 biologic recipients (0.26%) versus 79 of 5,352 cvIS recipients (1.48%). The primary HR was 0.20 (95% CI, 0.11–0.35). Dupilumab-only, treatment-anchored, calendar-era, methotrexate-only, and multivariable Cox analyses generally retained the lower direction with biologics. A treatment comparison incorporating a lag pointed in the same direction but was only nominally significant.

Unmeasured AD severity, treatment selection, and residual confounding complicate both associations, according to the authors. Early psoriasis may have been treated before formal diagnosis, while differences in healthcare use and residual calendar-era effects could influence recorded outcomes. Psoriasis diagnoses were code-based and not clinically adjudicated, leaving room for misclassification despite matching and robustness checks. The lower comparative incidence with biologics runs counter to concern that type 2-targeted treatments might trigger psoriasis, but it does not establish a preventive effect. The finding also contrasts with a prior TriNetX cohort report of 58% higher psoriasis risk with dupilumab versus other systemic agents; the authors noted design limitations in that analysis and cautioned against causal conclusions from either study.

The authors interpreted AD as associated with higher subsequent psoriasis incidence and biologic treatment in AD as associated with lower incidence than matched cvIS treatment, without assigning causality to either finding. Whether psoriasis emerging during treatment reflects disease-associated risk or a treatment effect remains unresolved.

Clinician Questions

Was incident psoriasis associated with atopic dermatitis in patients younger than 18 years?

Investigators observed the same direction of association between atopic dermatitis and incident psoriasis in patients younger than 18 years and those aged 18 years or older. Too few psoriasis outcomes occurred among younger patients to evaluate the biologic-versus-cvIS treatment comparison in that age group.

How was incident psoriasis vulgaris identified in the atopic dermatitis cohorts?

Incident psoriasis vulgaris in the atopic dermatitis comparisons was identified by an ICD-10-CM L40.0 diagnosis code recorded after cohort entry, without clinical adjudication. Intermediate or overlapping psoriasiform phenotypes could have been coded elsewhere.

Did the atopic dermatitis biologic comparison follow patients after treatment changes?

For patients with atopic dermatitis classified by their first eligible biologic or conventional systemic immunosuppressant treatment record, follow-up continued regardless of later therapy changes in an intention-to-treat-like framework. Discontinuations, dose adjustments, and sequential therapy were therefore not fully represented.

Recommended Reading

• For more on psoriasis treatment in psoriatic disease: Assessing the Evidence for GLP-1 Receptor Agonists in Psoriatic Disease
• For more on the lived burden of psoriasis: The Full Picture of Psoriasis: Recognizing the Invisible Burden