FormalPara Key Points

Atopic dermatitis (AD) is a chronic skin disease associated with intense itch and reduced quality of life.

This analysis demonstrated the association among itch severity, poor sleep, and reduced work productivity in patients with moderate-to-severe AD. Treatment with abrocitinib was shown to reduce itch, which in turn improved sleep outcomes, resulting in less work impairment.

1 Introduction

Atopic dermatitis (AD) is a chronic, relapsing inflammatory skin disease that affects up to 20% of children and 5–10% of adults worldwide [1,2,3,4,5,6]. AD is typically characterized by intense pruritus (itch), eczematous lesions, and dry skin [5], of which itch is the most bothersome symptom [4, 7, 8]. Uncontrolled chronic itch is often worse at night and can lead to sleep disturbance/loss and decreased quality of life (QoL) [2, 9,10,11,12,13], which can in turn adversely impact function and productivity. Increasing disease severity augments the negative impact of AD on work productivity [3, 14,15,16].

The pathophysiology of itch in AD is influenced by extrinsic factors (i.e., allergens and irritants), genetic predisposition, and a type-2 immune response mediated via the signaling of pruritic and inflammatory cytokines (i.e., interleukin [IL]-31, IL-4, and IL-13) [17,18,19,20]. Itch likely contributes to sleep disturbance in patients with AD and other inflammatory diseases that involve upregulation of IL-4 and IL-13, given the significant increase in sleep impairment seen in patients with psoriasis and itch compared with those without pruritus [21,22,23]. As such, it is possible that systemic therapies targeting key pruritic cytokines in the itch pathway may also act to improve sleep quality and subsequently other aspects of daily life in patients with AD.

Abrocitinib is an oral, once-daily, selective Janus kinase 1 (JAK1) inhibitor approved for the treatment of moderate-to-severe AD in adults and adolescents [24,25,26,27]. Abrocitinib was shown to be a well-tolerated and effective treatment for moderate-to-severe AD in multiple clinical studies (phase IIb, JADE MONO-1, JADE MONO-2, JADE COMPARE, JADE TEEN, JADE REGIMEN, and JADE DARE) [28,29,30,31,32,33,34]. Significantly greater proportions of patients treated with abrocitinib (200 mg or 100 mg) achieved an Investigator Global Assessment (IGA) response of clear (0) or almost clear (1) with ≥2-grade improvement, a ≥75% improvement in Eczema Area and Severity Index score (EASI-75) response, and/or ≥4-point improvement from baseline in Peak Pruritus Numerical Rating Scale (PP-NRS4) response, than those treated with placebo [30]. In the phase III JADE MONO-1 and JADE MONO-2 trials, abrocitinib resulted in greater decreases from baseline in PP-NRS score compared with placebo [22, 23]. Additionally, in JADE MONO-2, greater decreases were observed with abrocitinib in Work Productivity and Activity Impairment Questionnaire-Atopic Dermatitis (WPAI-AD), Patient-Oriented Eczema Measure sleep item (POEM sleep item), and SCORing AD Sleep Visual Analog Scale (SCORAD Sleep Loss VAS) scores compared with placebo [30, 35].

The direct and indirect effects of abrocitinib on improving itch, sleep, and work productivity have not been evaluated [13]. This supplemental (post hoc) analysis used data from JADE MONO-2 to examine the relationships between itch and sleep disturbance and between itch and overall work impairment. In addition, a mediation model was developed to simultaneously evaluate interrelationships among treatment, itch, sleep, and work productivity. We thereby sought to quantify the direct and indirect effects of abrocitinib on work productivity through analysis of itch and sleep.

2 Methods

2.1 Study Design

JADE MONO-2 (NCT03575871) was a multicenter, international, phase III, randomized, placebo-controlled, parallel-group trial. Patients with moderate-to-severe AD were randomly assigned (2:2:1) to receive abrocitinib 200 mg, 100 mg, or placebo for 12 weeks. Full study details, including the primary efficacy and safety results, have been published previously [30]. The study was conducted in compliance with the ethical principles originating in or derived from the Declaration of Helsinki and in compliance with all International Council for Harmonization Good Clinical Practice Guidelines. All local regulatory requirements were followed. This research was approved by institutional review boards or ethics committees at each study site. An internal review committee monitored the safety of patients throughout the study. All patients provided written informed consent.

2.2 Patients

Eligible patients were aged ≥12 years with clinical diagnosis of moderate-to-severe AD (IGA ≥3, EASI ≥16, percentage of body surface area involvement [%BSA] ≥10, PP-NRS [defined as the worst itch experienced in the last 24 h; used with permission of Regeneron Pharmaceuticals, Inc. and Sanofi [36]] ≥4 on the day of the baseline visit) and a recent (within 6 months) history of inadequate response to topical corticosteroids or calcineurin inhibitors given for ≥4 weeks or an inability to receive topical treatment because it was medically inadvisable. Previous dupilumab use was permitted if it was discontinued >6 weeks before study initiation. Concomitant use of topical (corticosteroids, calcineurin inhibitors, tars, antibiotic creams, or antihistamines) or systemic therapies for AD was prohibited [30]. Rescue medication (including topical corticosteroids) was prohibited during the study. Full inclusion and exclusion criteria are published elsewhere [30].

2.3 Post Hoc Analysis

A repeated-measures longitudinal model [37, 38] was employed to estimate relationships between itch severity (PP-NRS or Night Time Itch Scale [NTIS], scored 0–10, higher scores indicating worse itch) as a predictor and sleep disturbance (POEM sleep item or SCORAD Sleep Loss VAS) as an outcome. The relationship between itch severity (PP-NRS or NTIS) as a predictor and work productivity (WPAI-AD) as an outcome was also investigated. In addition, the relationship between POEM sleep item and SCORAD Sleep Loss VAS was examined.

PP-NRS and NTIS were assessed daily during the 28-day screening period and the first 2 weeks after randomization, followed by single measurements at week 4, week 8, and week 12. As a result, in the analysis NTIS and PP-NRS values for baseline (week 0), week 1 and week 2 are represented by the mean of all available weekly measurements (if at least 4 observations were available). The POEM sleep item and SCORAD Sleep Loss VAS were collected at baseline and weeks 2, 4, 8, and 12.

WPAI-AD was assessed in adult patients at baseline and week 12 to measure the effect of AD on work productivity and regular activities during the previous 7 days. The activity impairment component of this questionnaire was completed by all adult patients, while the work impairment component was completed by only those adult patients who were employed during the study period and assessed the level of impairment in productivity while working. The WPAI-AD is designed in such a way that questions that are not applicable (i.e., impact of AD on work productivity) can be skipped by patients who are not employed and as such, for these patients, absenteeism, presenteeism, and overall work productivity were not calculated. As this omission is by design, these domains should not be considered as ‘missing’.

To study the appropriateness of the linear approximation of the relationship between the predictor and the outcome, two models were investigated. The main model, with the predictor as a continuous variable, imposed a linear relationship between the predictor and the outcome. A model with the predictor as a categorical variable (sensitivity analysis) was also studied whereby a functional relationship between the outcome and the predictor is not imposed. The models assume that measurements from the same patient are not independent, which means that the error terms covary over time within a patient, and the error at one time point is allowed to covary with the error at another time point for any patient [38]. The covariance of the error terms for the same patient over time was modeled with a spatial power covariance structure that allows for time intervals that are not evenly spaced [38]. The models, which were fit in SAS with Prox Mixed and the Repeated statement, use all available data and assume that data were missing at random.

A mediation model, which can help elucidate the mechanism of action of a drug, was used to investigate the direct and indirect effects of the treatment (abrocitinib [200 mg and 100 mg] vs placebo) on overall work impairment (WPAI-AD), and how these effects were influenced by itch (PP-NRS and NTIS), and sleep (POEM sleep item and SCORAD Sleep Loss VAS), denoted as mediators [39]. The initial model included indirect causal paths from abrocitinib to overall work impairment through itch severity (latent variable F1) and sleep (latent variable F2) as well as direct effects of abrocitinib to overall work impairment (representing all other possible paths; Supplementary Fig. S1, see electronic supplementary material [ESM]). Variables e_ppnrs, e_ntis, e_poem, e_scrd, and e_wpai represent error terms, which reflect unexplained variation due to random measurement error (score unreliability). Variables e_itch and e_sleep represent disturbance terms, which reflect unexplained variation due to omitted (unmeasured) causes and score unreliability in the corresponding latent variables. Based on the results of the initial model, the mediation model can be re-specified to resolve any irregularities and contradictions.

3 Results

3.1 Patients

This post hoc analysis used data from the JADE MONO-2 study (n = 391), which included patients randomized to treatment with abrocitinib 200 mg, abrocitinib 100 mg, or placebo. The baseline disease and demographic characteristics of these patients have been previously reported and were generally similar across treatment groups (Table 1) [30]. Patients from the JADE MONO-2 study were mostly male (59%) with a mean age of 35.1 years and an average PP-NRS score of 7 [30]. For this post hoc analysis, only those patients with at least one observation for each of the parameters assessed in the respective associations were included (Table 2).

Table 1 Demographic and baseline characteristics
Full size table
Table 2 Summary of patient numbers and individual patient-reported outcome observations included in analysis
Full size table

3.2 Linearity Between Outcomes Through a Longitudinal Model

All available data from baseline to week 12 were included in a repeated-measures longitudinal model, with predictors as continuous or categorical variables. Similar results were observed in models in which the predictor was used as a continuous variable versus a categorical variable, supporting the linearity assumption for (a) the relationships between itch (PP-NRS and NTIS) and sleep (SCORAD sleep loss VAS and POEM sleep items) (Fig. 1); (b) the relationships between PP-NRS as a predictor and either POEM sleep item or SCORAD Sleep Loss VAS as the outcome, or WPAI as the outcome for work productivity (Fig. 2); and (c) the relationship between NTIS as a predictor and either POEM sleep item or SCORAD Sleep Loss VAS as the outcome, or WPAI as the outcome for work productivity (Fig. 3).

Fig. 1
figure 1

Relationships between a PP-NRS and NTIS and b SCORAD and POEM sleep items. NTIS Night Time Itch Scale, POEM Patient-Oriented Eczema Measure, PP-NRS Peak Pruritus Numerical Rating Scale, SCORAD Sleep Loss VAS SCORing of Atopic Dermatitis Sleep Loss Visual Analog Scale, WPAI-AD Work Productivity and Activity Impairment-Atopic Dermatitis

Full size image
Fig. 2
figure 2

Relationships between outcomes and PP-NRS as the anchor (continuous predictor or categorical predictor): a PP-NRS and POEM Sleep item, b PP-NRS and SCORAD Sleep Loss VAS, and c PP-NRS and WPAI overall work impairment. NTIS Night Time Itch Scale, POEM Patient-Oriented Eczema Measure, PP-NRS Peak Pruritus Numerical Rating Scale, SCORAD Sleep Loss VAS SCORing of Atopic Dermatitis Sleep Loss Visual Analog Scale, WPAI Work Productivity and Activity Impairment questionnaire

Full size image
Fig. 3
figure 3

Relationships between outcomes and NTIS as the anchor (continuous predictor or categorical predictor): a NTIS and POEM sleep item, b NTIS and SCORAD Sleep Loss VAS, and c NTIS and WPAI overall work impairment. NTIS Night Time Itch Scale, POEM Patient-Oriented Eczema Measure, PP-NRS Peak Pruritus Numerical Rating Scale, SCORAD Sleep Loss VAS SCORing of Atopic Dermatitis Sleep Loss Visual Analog Scale, WPAI Work Productivity and Activity Impairment questionnaire

Full size image

The relationship between itch and the POEM sleep item showed that PP-NRS (N = 390) scores of 0, 4–6, and 10 and NTIS (N = 389) scores of 0, 4–6, and 9–10 are associated with no days, 3–4 days per week, and every day of disturbed sleep, respectively (Fig. 4a): every 1-point improvement in itch corresponded to an approximate 0.4-point improvement in the POEM sleep item. The relationship between itch and WPAI-AD showed that PP-NRS (N = 220) scores of 0–1, 4–5, and 10 and NTIS (N = 219) scores of 0–1, 4–5, and 10 are associated with 0–10%, 20–30%, and >50% overall work impairment, respectively (Fig. 4b): every 1-point improvement in itch corresponded to an approximate 5% improvement in WPAI-AD overall work impairment. With the interpretations of the relationships between outcomes found in the longitudinal model, further investigation of specific paths between outcomes was evaluated through mediation modeling.

Fig. 4
figure 4

Interpretation of the association between itch and a sleep and b overall work impairment. NTIS Night Time Itch Scale, POEM Patient-Oriented Eczema Measure, PP-NRS Peak Pruritus Numerical Rating Scale, SCORAD Sleep Loss VAS SCORing of Atopic Dermatitis Sleep Loss Visual Analog Scale, WPAI-AD Work Productivity and Activity Impairment-Atopic Dermatitis questionnaire version 2.0

Full size image

3.3 Mediation Modeling

The mediation model investigated the effect of abrocitinib on work productivity with itch and sleep as mediators. Data from 148 patients at week 12 were available for analysis. The direct effect of abrocitinib on overall work impairment was found to be 2.2% and not statistically significant (p = 0.95); the overall indirect effect was 97.8% and statistically significant (p = 0.0027), indicating that abrocitinib affects overall work impairment predominantly indirectly (Fig. 5a).

Fig. 5
figure 5

Mediation modeling: a direct and overall indirect effects, b path via itch, c path via sleep, and d path via itch and sleep. WPAI-AD Work Productivity and Activity Impairment—Atopic Dermatitis questionnaire version 2.0. Solid, single-headed arrows indicate direct paths; Dashed, single-headed arrows, indicate indirect paths

Full size image

Furthermore, the path from abrocitinib to overall work impairment through itch was 0.8% (p = 0.97) and through sleep was 22.2% (p = 0.27) (Fig. 5b & c, respectively).

Finally, the effect of treatment on overall work impairment via itch and sleep (i.e., treatment improves itch, improved itch improves sleep, and improved sleep improves overall work impairment) was 74.9% and statistically significant (p = 0.02) (Fig. 5d).

Based on the above results, the mediation model was re-specified: (a) the direct path from treatment to WPAI was excluded, (b) the path from itch (Factor 1) to WPAI was excluded, and (c) the path from treatment to sleep (Factor 2) was also excluded from the model (Supplementary Fig. S2, see ESM). In this re-specified model, treatment affects WPAI overall work impairment only indirectly, first via improving itch, followed by improved sleep outcomes, which in turn reduces work impairment.

4 Discussion

Itch is the most common symptom for patients with AD and affects several aspects of everyday life [40]. Itch, particularly at night, interferes with sleep and the normal cycle of physical and mental relaxation and thus can adversely affect daytime physical and mental function and alertness [13, 41,42,43,44]. This analysis systematically establishes relationships among the patient-reported assessments of itch, sleep, and work productivity and identifies the path that best models the effects of treatment on work productivity through itch and sleep outcomes, evaluating their relative contributions. The relationships between itch and sleep and between itch and work productivity were approximately linear and indicate that greater itch severity is strongly associated with greater sleep disturbance/loss and has a notable impact on work impairment. Severe PP-NRS or NTIS (≥7) is associated with ≥5 days/week of sleep disturbance and ≥30% overall work impairment, with each 1-point reduction in itch severity reducing overall work impairment by 5%.

Intense itch, sleep disturbances, and poor work productivity are all bothersome impacts of AD that have an adverse effect on patients’ QoL [13, 44]. Improvements in these symptoms can provide substantial changes in a patient’s daily life. The mediation model quantitatively shows that, in patients with moderate-to-severe AD, abrocitinib improves overall work productivity in a sequential process beginning with treatment effects on itch, which translate into improvement in sleep that eventually manifests as lesser impairment in work productivity.

This analysis was limited by its post hoc nature and the relatively small and varied number of observations for employed patients in the mediation model (53.7%–98.5% of total sample size). In addition, the contribution of factors (other than itch) to sleep disturbance, such as increased monoamine and cytokine dysregulation, and decreased melatonin and skin temperature [44], were not considered, although one may argue that at least some of the effects mediated by these additional variables may be accounted for within itch. Furthermore, this model does not explore dose relationships.

5 Conclusions

In JADE MONO-2, the relationships between itch and sleep outcomes and between itch and work productivity were approximately linear, demonstrating that reduction in itch severity was associated with improved sleep and work productivity [45]. Empirical evidence for the mechanism of action of abrocitinib suggests that the effect of abrocitinib on work productivity is predominantly mediated by an indirect path wherein treatment with abrocitinib improves itch severity, which reduces sleep loss/sleep disturbance that, in turn, improves work productivity. Future research using data from clinical studies with longer duration of treatment may expand these observations.