Introduction
Atopic dermatitis (AD) is a common, chronic, inflammatory skin disease characterized by complex pathophysiology and heterogeneous clinical phenotypes [
1‐
3]. AD symptoms include intense itch, sleep disturbance, and skin pain, which significantly impact patient quality of life and work productivity [
4‐
6]. Emollients and topical corticosteroids (TCS) are the mainstay of AD therapy [
7‐
9]. In patients with moderate-to-severe AD for whom topical therapy does not adequately control the signs and symptoms of disease, addition of phototherapy and/or systemic treatment is recommended [
10,
11]. Systemic treatments can be associated with variable efficacy and unnecessary cycling through ineffective medications resulting in prolonged burden and delayed improvements [
10,
12]. Therefore, prediction of treatment efficacy is becoming increasingly important with the emergence of new therapies. An understanding of which patients are most likely to benefit from therapy can help tailor therapies to individual patient needs. Tailored therapeutic approaches can significantly refine the management of AD through improving patient experience with a treatment, increasing cost-effectiveness of a therapy, and ensuring that only patients who are likely to benefit from therapy are exposed to a given treatment.
Baricitinib, an oral selective Janus kinase (JAK) 1 and JAK2 inhibitor [
13], is indicated in the European Union and Japan and being evaluated in the USA and other countries for treatment of moderate-to-severe AD in adult patients who are candidates for systemic therapy. Dependent upon country-specific approvals, the recommended starting dose of baricitinib for adults with moderate-to-severe AD is 2 mg or 4 mg [
14‐
16]. In BREEZE-AD5, an ongoing, randomized, placebo-controlled, phase 3 trial of moderate-to-severe AD patients who had an inadequate response or intolerance to TCS, once-daily oral baricitinib 2-mg monotherapy improved several clinical signs and symptoms of AD at week 16 compared with placebo, with a safety profile consistent with previous studies of baricitinib 2 mg in AD [
17,
18]. The objective of this post-hoc analysis was to identify patients who are most likely to benefit from baricitinib 2 mg, using a proposed clinical tailoring approach based on baseline body surface area (BSA) affected at drug initiation and early clinical improvement, in the phase 3 monotherapy trial BREEZE-AD5.
Methods
Study Design and Patients
BREEZE-AD5 (NCT03435081) is an ongoing phase 3 randomized, double-blind, placebo-controlled trial conducted in the USA and Canada [
17]. Patients (
N = 440) were randomized 1:1:1 to receive once-daily placebo, baricitinib 1 mg, or baricitinib 2 mg (Fig. S1). Eligible patients were ≥ 18 years of age and had a diagnosis of AD, as defined by the American Academy of Dermatology [
19], ≥ 12 months prior to screening. Enrolled patients had moderate-to-severe disease, defined by baseline Eczema Area and Severity Index (EASI) score ≥ 16, validated Investigator Global Assessment for Atopic Dermatitis (vIGA-AD) score ≥ 3, and BSA involvement of ≥ 10%. Patients had a documented history of inadequate response or intolerance to topical therapies within 6 months before screening. Patients discontinued topical therapy 2 weeks and systemic therapy 4 weeks before randomization. This trial was conducted in accordance with the ethical principles of the Helsinki Declaration of 1964 and its later amendments and Good Clinical Practice guidelines and approved by the appropriate institutional review boards/ethics committees at each study site (Table S1), including the Quorum Review IRB (approval #33039). All patients provided written informed consent.
Outcomes
The primary endpoint of the trial was the proportion of patients who achieved 75% improvement in EASI score (EASI75) at week 16. Two secondary endpoints included the proportion of patients who achieved a vIGA-AD score of 0 (clear) or 1 (almost clear), with a ≥ 2-point improvement from baseline, and the proportion of patients who achieved a ≥ 4-point improvement from baseline in the Itch Numeric Rating Scale (NRS) [
17]. These endpoints were analyzed in this post-hoc analysis in the subgroup of patients most likely to benefit from baricitinib 2-mg treatment.
Statistical Analyses
In this post-hoc analysis, classification and regression tree algorithm (CART) was used to identify baseline predictors of response. A variety of baseline measures were assessed, including, but not limited to patient demographics, prior treatment history, baseline disease severity [such as vIGA-AD, EASI, BSA, Itch NRS, Atopic Dermatitis Sleep Scale, Dermatology Life Quality Index (DLQI), SCORing Atopic Dermatitis, Hospital Anxiety Depression Scale], and laboratory parameters. Response was defined as the proportion of patients achieving EASI75 at week 16 among those treated with baricitinib 2 mg in the intent-to-treat (ITT) population. Two-by-two contingency tables with associated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were generated between CART-identified baseline predictors and EASI75. As a sensitivity analysis, the same method was applied to identify baseline predictors for the proportion of patients who achieved vIGA-AD (0,1) at week 16. Sensitivity and NPV were examined closely. Sensitivity represents the proportion of patients with certain baseline characteristics, among those who were responders at week 16. A high sensitivity value indicates that most responders were identified by characteristics through the algorithm. NPV represents the proportion of patients who were not responders at week 16, among those without the same baseline characteristics. A high NPV indicates that patients in this group would likely not benefit from therapy. After baseline subgroups were identified using the CART method, response rate for EASI75, vIGA-AD (0,1), and ≥ 4-point improvement in Itch NRS by visits were analyzed with logistics regression models. Baseline disease severity (vIGA-AD), continuous version of endpoint at baseline, treatment group, baseline subgroup, and treatment-by-baseline subgroup interaction were included as model terms.
After identification of a baseline predictor of response, further analyses were conducted to determine if early response could be used to further refine the patient selection. The goal of this second analysis was to quickly identify any additional patients who are less likely to benefit from therapy and allow for quick clinical decision on whether to transition such patients to alternative therapies. Two-by-two contingency tables were used to evaluate the association between early response in skin inflammation (≥ 50% improvement in BSA) or itch (≥ 3-point improvement in Itch NRS) at week 4 or week 8, and response at week 16 for EASI75, vIGA-AD (0,1), and ≥ 4-point improvement in Itch NRS. Early measures of improvement (≥ 50% improvement in BSA and ≥ 3-point improvement in Itch NRS) were selected for being considered a meaningful improvement, easily noticeable by patients and physicians, and applicable in clinical practice [
20]. EASI75, vIGA-AD (0,1), and ≥ 4-point improvement in Itch NRS are accepted regulatory endpoints in AD, which allowed for comparison between the ITT population results and this subgroup analysis. Sensitivity, specificity, PPV, and NPV were used to evaluate the performance of the two-by-two classification table.
Data collected after any rescue or treatment discontinuation were considered as missing. All missing data were imputed as non-responder. To use the same ITT population for evaluating skin inflammation and itch simultaneously in association with early improvement and week 16 response, a conservative approach to define improver and responder was used: any patient with a baseline Itch NRS score < 3 was automatically classified as a non-improver for ≥ 3-point improvement in Itch NRS at week 4 or 8, and any patient with a baseline Itch NRS score < 4 was automatically classified as a non-responder for ≥ 4-point improvement in Itch NRS at week 16.
Discussion
Personalized medicine has long been a goal in clinical practice [
21,
22]. Providers seek to identify which patients are most likely to benefit from treatment in order to increase the probability of success with a given therapy and minimize risk of exposure to an ineffective treatment. In this study, we demonstrated that baricitinib 2 mg provides enhanced efficacy in patients with a baseline BSA ranging from 10% to 50%, with higher response rates for EASI75 and vIGA-AD (0,1), compared with patients with a baseline BSA > 50%.
Early assessment of clinical response was defined as an improvement of at least 50% in BSA or an improvement of at least 3 points in Itch NRS from baseline at week 4 or week 8. Early response in skin inflammation or itch resulted in an additional significant increase in patient response at week 16 among baricitinib 2-mg-treated patients with baseline BSA 10–50%, with as many as two-thirds of patients achieving an EASI75 at week 16 and over half of the patients achieving a vIGA-AD (0,1) at week 16. This is in contrast to 37.5% and 31.7% of responders observed for EASI75 and vIGA-AD (0,1), respectively, in the overall baseline BSA 10–50% group, and 29.5% and 24.0%, respectively, in the ITT population [
17]. These high efficacy results were observed while maintaining a very high NPV for the analyses, showing that an early decision for discontinuation of therapy is possible in patients with a BSA of 10–50% at baseline. Patients who respond to baricitinib 2 mg experience clinical benefit early in their course of treatment. Patients who do not respond to baricitinib 2-mg treatment within the first 4–8 weeks are unlikely to respond to treatment in the long term.
Although BSA and associated EASI scores were lower in patients with BSA 10–50% at baseline, overall burden of disease by symptoms was comparable between the groups. Patients’ baseline itch, sleep disturbance due to itch, skin pain, Patient Oriented Eczema Measures scores, and quality of life (DLQI) were comparable between the groups, confirming the significant burden of disease in patients with baseline BSA 10–50%. Although the proposed clinical tailoring approach uses baseline BSA, a small increase in itch response was also observed in patients with baseline BSA 10–50% compared with patients with baseline BSA > 50%. Importantly, 100% of the patients who responded to baricitinib 2 mg with a ≥ 4-point improvement in Itch NRS at week 16 had achieved a ≥ 3-point improvement by week 8 (Table
2, sensitivity = 1.0, NPV = 1.0).
A limitation to this study is that these are post-hoc subgroup analyses as opposed to prespecified subgroup analyses [
23]. The analyses focused on the 2-mg dose of baricitinib, and the lack of data from the 4-mg dose is a further limitation. In addition, patients discontinued topical and systemic treatments 2 and 4 weeks prior to randomization, respectively. The discontinuation of topical and systemic therapies weeks prior to the initiation of the study drug may lead to rescue earlier than if therapies were discontinued closer to the initiation of the clinical trial. In addition, this was a monotherapy trial. Although this study helped to better assess the impact of treatment in AD severity without the potential confounding effect of topical corticosteroids, topical calcineurin inhibitors, or topical phosphodiesterase-4 inhibitors, some of these treatments may be maintained in clinical practice during treatment. Other limitations are the lack of standardization in the assessment of BSA [
24], as well as the inconsistent performance of a full-body skin examination in clinical practice. The BSA affords a rapid assessment on physical examination that may help to guide clinical decision-making, particularly with the advent of new therapies in the treatment armamentarium for AD.
In conclusion, patients with moderate-to-severe AD affecting between 10% and 50% of their BSA accounted for the majority of responders to baricitinib 2 mg. The clinical assessment of patients after 4–8 weeks of initiation of baricitinib 2-mg treatment predicted which patients are likely to benefit from long-term therapy. This analysis may allow for a precision-medicine approach to therapy in moderate-to-severe AD.
Acknowledgements
The authors would like to thank the patients and study investigators who participated in the study.
Disclosures
Jonathan I. Silverberg served as a consultant and/or advisory board member for AbbVie, Arena, Asana, Bluefin, Boehringer-Ingelheim, Dermavant, Dermira, Eli Lilly and Company, Galderma, GlaxoSmithKline, Glenmark, Incyte, Kiniksa, Leo, Luna, Novartis, Pfizer, RAPT, Regeneron, Sanofi, receiving honoraria; served as a speaker for Regeneron-Sanofi; and received research grants from Galderma. Mark Boguniewicz served as an investigator for Incyte and Regeneron and a consultant and/or advisory board member for AbbVie, Eli Lilly and Company, Janssen, LEO Pharma, Pfizer, Regeneron, and Sanofi-Genzyme. Jill Waibel served as a consultant and/or investigator for, and/or received personal fees from AbbVie, Allergan, Almirall, AstraZeneca, Avita Medical, Biofrontera, Candela, Cytrellis, Dermira, Dominion Aesthetics, Eli Lilly and Company, Michaelson Diagnostics, Novartis, Pfizer, and Sciton. Jamie Weisman served as a speaker and/or investigator and/or has received grants and/or honoraria from AbbVie, Amgen, Biogen, Boehringer Ingelheim, Celgene, Eli Lilly and Company, Janssen, LEO Pharma, Merck, Novartis, Pfizer, Regeneron, Stiefel, and Valeant Pharmaceuticals. Lindsay Strowd has received support from Actelion, Eli Lilly and Company, Galderma, Pfizer, Regeneron, and Sanofi. Luna Sun, Yuxin Ding, and Meghan Feely are employees and shareholders of Eli Lilly and Company. Meghan Feely is also a clinical instructor at Mount Sinai. Fabio P. Nunes is a former employee of Eli Lilly and Company and is a current employee and shareholder of Janssen Pharmaceuticals Companies of Johnson and Johnson. Eric L. Simpson received grants and fees for participation as a consultant and principal investigator from Eli Lilly and Company, LEO Pharma, Pfizer, and Regeneron; grants for participation as a principal investigator from Galderma and Merck & Co.; and fees for consultant services from AbbVie, Boehringer Ingelheim, Dermavant Incyte, Forte Bio, Pierre Fabre Dermo, and Sanofi Genzyme.