Oral Janus kinase/SYK inhibition (ASN002) suppresses inflammation and improves epidermal barrier markers in patients with atopic dermatitis
Background: Moderate-to-severe atopic dermatitis (AD) has been associated with significant disease burden and systemic abnormalities and often requires systemic treatments.Currently, safe and effective oral systemic treatments for moderate-to-severe AD are not yet available. ASN002 is an oral inhibitor of the Janus kinase/spleen tyrosine kinase signaling pathways, targeting several cytokine axes (TH2/TH22/TH17/TH1) and epidermal differentiation.Objective: We sought to evaluate the effect of ASN002 on the cellular and molecular biomarker profile of patients with moderate-to-severe AD and to correlate changes in biomarkers to improvements in clinical severity measures and pruritus.Methods: Thirty-six patients with moderate-to-severe AD were randomized to groups with dose escalation of ASN002 (20, 40, and 80 mg) and a placebo group. Skin biopsy specimens were performed at baseline, day 15, and day 29. Gene expression.From athe Department of Dermatology, Laboratory of Inflammatory Skin Diseases, Icahn School of Medicine at Mount Sinai, New York; bthe Laboratory for Investigative Dermatology, Rockefeller University, New York; cAsana Biosciences, Princeton; and dInnovaderm Research, Montrealstudies were conducted by using microarray and quantitative RT-PCR, and cellular infiltrates and protein expression were studied by using immunohistochemistry.Results: ASN002 reversed the lesional skin transcriptome toward a nonlesional phenotype. It also rapidly and significantly suppressed key inflammatory pathways implicated in AD pathogenesis, including TH2 (IL4 receptor [IL4R], IL13, CCL13/ monocyte chemoattractant protein 4, CCL17/thymus and activation-regulated chemokine, CCL18/pulmonary and activation-regulated chemokine, CCL22/macrophage-derived chemokine, and CCL26/eotaxin-3), TH17/TH22 (lipocalins, PI3/ elafin, CCL20, S100A7/S100A8/S100A9, and IL36G/IL36RN),and TH1 (IFNG, CXCL9/CXCL11, and MX1) axes and barrier- related measures (filaggrin [FLG] and CLDN23). Significant improvements in AD gene signatures were observed predominantly in the 40- and 80-mg groups. Smaller and largely nonsignificant molecular changes were seen in the 20-mg and placebo groups.Conclusion: The Janus kinase/spleen tyrosine kinase inhibitor ASN002 significantly suppressed key AD inflammatory pathways, corresponding to clinical response. ASN002 might be an effective novel therapeutic agent for moderate-to-severe AD. (J Allergy Clin Immunol 2019;.)
Atopic dermatitis (AD) affects up to 10% of adults and 25% of children worldwide, with up to one third of patients having moderate-to-severe disease.1-5 AD is a heterogeneous disease pri- marily involving TH2 and TH22 activation, with variable TH1 and TH17 skewing depending on subtype.6-15 Although all subtypes show TH2/TH22 activation, some subtypes, including Asian, pedi- atric, and intrinsic AD, show increased TH17 activation.8,10-12 Despite its high prevalence and effect on quality of life, systemic treatment options for moderate-to-severe AD are limited, present- ing a large unmet need for safer and more efficacious treat- ments.16,17 Glucocorticoids, which harbor many side effects and are inappropriate for long-term use, are the only US Food and Drug Administration–approved oral therapeutic for AD.18 Other immune suppressants used off-label for AD, such as cyclosporine A, are inadequate for continuous use.19-22 Dupilumab, a novel TH2-specific mAb targeting IL-4 receptor (IL-4R) a, which was 80 mg of oral ASN002 and a placebo group for 28 days.49 Each patient was eval- uated at baseline (day 1), day 15, and day 29 for safety and efficacy, and skin biopsy specimens were taken for evaluating biomarkers (see the Methods sec- tion in this article’s Online Repository at www.jacionline.org for details). Addi- tional study design and clinical trial outcome data have been reported recently.In vitro biochemical assays ASN002, tamatinib, and tofacitinib were evaluated for inhibiting JAK1, JAK2, JAK3, and tyrosine kinase 2 (TYK2) activity by using an in vitro, time- resolved fluorescence resonance energy transfer assay. Inhibitory concentra- tions of 50% of ASN002, tamatinib, and tofacitinib were determined by using an in vitro radiometric assay. IL-17–mediated CCL20 release was evaluated by using neonatal human epidermal keratinocytes treated with ASN002 (1 mmol/ L), tamatinib (1 mmol/L), tofacitinib (1 mmol/L), an anti–IL-17 antibody, and IL-17 dimethyl sulfoxide. See the Methods section in this article’s Online Re- pository for details recently approved by the US Food and Drug Administration, has shown efficacy in patients with moderate-to-severe AD, with approximately 30% of patients approaching disease resolution (90% or greater reduction in Eczema Area and Severity Index [EASI] score).5,23-25
Oral Janus kinase (JAK) inhibitors targeting JAK1 or JAK1/ JAK2 signaling pathways have recently shown efficacy in early clinical trials for patients with moderate-to-severe AD.26-30 De- pending on their specificity, JAK inhibitors target cytokine axes involved in the pathogenesis of AD, specifically the TH2 (IL-4, IL-5, IL-13, and thymic stromal lymphopoietin), TH22 (IL-22), and TH1 (IFN-g, IL-12, and IL-23) pathways.26-28,31,32 Spleen tyrosine kinase (SYK), a nonreceptor tyrosine kinase, regulates B-cell and dendritic cell (DC) differentiation and TH17/IL-17 signaling and inhibits keratinocyte terminal differentiation.33-41 In mice SYK inhibition was shown to inhibit TH17 pathway signaling and improve terminal epidermal differentiation.41-43 Thus dual JAK-SYK inhibition might broaden the range of cyto- kines targeted to successfully resolve AD across the different dis- ease subtypes.9,10,12,21,44-48.ASN002 is the first oral dual JAK/SYK inhibitor tested in patients with AD. A phase 1B clinical trial has been performed recently in patients with moderate-to-severe AD over 4 weeks (ClinicalTrials.gov: NCT03139981),49 evaluating 3 daily doses (20, 40, and 80 mg) compared with placebo. Eight percent to 100% of patients receiving the higher doses of ASN002 achieved 50% or greater reduction in EASI score response by day 29.49 The present study evaluates the effects of the 3 ASN002 doses compared with placebo on the molecular and cellular profiles of lesional and nonlesional AD skin after 15 and 29 days of treatment. Overall, both higher doses of ASN002 significantly and rapidly reduced inflammatory cellular infiltrates, suppressed major inflammatory pathways, and improved the epidermal barrier abnormalities implicated in AD, and treatment response biomarkers were correlated with clinical responses.
This mechanistic study is part of a randomized, double-blind, placebo- controlled, phase 1B clinical trial (ClinicalTrials.gov: NCT03139981) conduct- ed at 10 centers across the United States and Canada. Thirty-six patients were enrolled and randomized 1:1:1:1 to dose escalation groups with 20, 40, and Immunohistochemistry was performed on frozen OCT-embedded cryostat tissues using purified mouse anti-human mAbs, as previously described.50 An- tibodies and dilutions can be found in Table E1 in this article’s Online Repos- itory at www.jacionline.org. Epidermal thickness and inflammatory cells were quantified per millimeter by using computer-assisted analysis software (Im- ageJ 1.42; National institutes of Health, Bethesda, Md), as previously described.10,51-54.Quantitative real-time PCR and gene array analysis RNA was profiled by using Affymetrix Human U133 Plus 2.0 gene arrays (Affymetrix, Santa Clara, Calif) available through the Gene Expression Omnibus (GSE133385) and real-time PCR with EZ-PCR Core reagents (Life Technologies, Grand Island, NY), as previously described and detailed in the Methods section in this article’s Online Repository.10,52,54-56 TaqMan Low Density Array/TLDA cards were used for quantitative RT-PCR (qRT-PCR) to analyze mRNA expression of inflammatory markers. Primers and probes used are listed in Table E2 in this article’s Online Repository at www. jacionline.org and were generated by using the Primer Express Algorithm (Applied Biosystems, Foster City, Calif).10 See the Methods section in this article’s Online Repository for details.All analyses were performed in the statistical language R (www.R-Project. org) and Bioconductor. Standard R packages, such as arrayQualityMetrics and GC Robust Multi-array Average, were used for quality control and normaliza- tion.11,57,58 Gene expression was modeled by using linear mixed-effect models with time, treatment, and tissue as a fixed interaction term and a random effect for each patient by using R’s Limma linear modeling framework. P values were adjusted for multiplicity by using the Benjamini-Hochberg procedure, which controls the false discovery rate.59 Spearman correlations between clin- ical scores and biomarkers were considered significant at a P value of less than.05. A nonparametric multivariate previously published m stat approach60 was used to integrate skin, blood, and clinical changes. See the Methods section in this article’s Online Repository for details.
RESULTS
Thirty-six patients were randomized to receive placebo or 20, 40, or 80 mg of ASN002 (n 5 9 per group) in an ascending-dose study design.49 There were no significant differences between study arms in baseline demographic data (age, sex, and clinical severity measures; see Table E3 in this article’s Online Repository at www.jacionline.org).49 Lesional and nonlesional biopsy specimens were collected at baseline, and lesional biopsy speci- mens were collected at days 15 and 29. The number of biopsy spec- imens available for analyses at each time point is listed in Table E4 in this article’s Online Repository at www.jacionline.org.ASN002 shows greater affinity for SYK (and TYK2) pathways and greater inhibition of IL-17–induced CCL20 compared with selected JAK inhibitors.To determine the affinity of ASN002 to JAK1, JAK2, JAK3, TYK2, and SYK compared with other JAK and/SYK inhibitors, we used an in vitro radiometric assay61 that determined the inhib- itory concentration of 50% of ASN002, tamatinib, and tofacitinib and also compared these data with reported upadicitinib62 and baricitinib63 data. ASN has much greater affinity for SYK and TYK2 compared with tofacitinib and greater affinity for SYK and TYK2 compared with tamatinib (see Table E5 in this article’s Online Repository at www.jacionline.org). ASN002 also shows decreasing affinity to JAKs (JAK2 > JAK3 > JAK1).Based on the SYK activity of ASN002, we hypothesized that IL-17 will be more suppressed by ASN002 compared with the pan-JAK tofacitinib. Thus we compared between inhibition of IL- 17–mediated CCL20 release in human keratinocytes by using similar concentrations (1 mmol/L) of ASN002, tamatinib, tofa- citinib, and an IL-17 neutralizing antibody. ASN002 showed the greatest inhibition of CCL20 release, which was similar to that of the IL-17 antagonist (see Fig E1 and the Methods and Results sections in this article’s Online Repository at www.jacionline.org).Epidermal hyperplasia was evaluated at baseline, day 15, and day 29 by using epidermal thickness, along with protein and gene expression of the proliferation marker keratin 16 (K16), as quantified by fold changes compared with baseline and placebo (Fig 1). ASN002 improved epidermal hyperplasia, as evident by decreased epidermal thickness on hematoxylin and eosin staining and a rapid reversal of K16 staining already evident at day 15 in 7 of 14 of the patients receiving the 40-mg (3/8) or 80-mg (4/6) doses, with 10 of 15 (4/7 receiving 40 mg and 6/8 receiving 80 mg) of these patients achieving more resolution of K16 stain- ing at day 29 (Fig 1, A and B, and see Fig E2, A, in this article’s Online Repository at www.jacionline.org).
Only minimal im- provements were seen in the 20-mg and placebo arms at day 29 (Fig 1, A and B), where only 4 of 15 (1/7 receiving 20 mg and 3/8 receiving placebo) patients had improvement in K16 staining by day 29. The 40- and 80-mg doses also induced significant re- ductions in K16 mRNA expression at days 15 and 29 compared with baseline (P < .05 for both doses) and improvements in epidermal thickness (P < .05 for 80 mg at day 29), whereas min- imal improvements and even exacerbations were seen in the 20- mg and placebo groups (Fig 1, D and E). We also evaluated protein expression of the epidermal differ- entiation marker filaggrin (FLG) by means of immunohistochem- istry. At baseline, lesional skin showed faint interrupted FLG staining compared with continuous and more intense staining in nonlesional skin (Fig 1, C). Both the 40- and 80-mg ASN002 doses induced more robust FLG staining already by day 15 and further at day 29 (Fig 1, C, and see Fig E2, B) compared with the discontinuous lesional staining in the placebo arm at both time points.We assessed cellular infiltrates by using immunohistochemical staining for T cells (CD31) and myeloid DCs (CD11c1; Fig 1, F and G, and see Fig E2, C and D). ASN002 significantly reduced CD31 T-cell infiltrates in lesional skin at day 29 compared with baseline, predominantly in the 80-mg group and approaching sig- nificance in the 40-mg group (P <.05 for day 29 [80 mg]; Fig 1, F), with no significant difference compared with placebo. Progres- sive reductions in numbers of CD11c1 DCs were seen in both the 40- and 80-mg arms at day 29 compared with baseline (P <.01 for both at day 29; Fig 1, G), with 80 mg approaching sig- nificance versus placebo (Fig 1, G, and see Fig E2, D). No significant decreases in T-cell or DC infiltrates were seen in the 20-mg or placebo arms.Affymetrix U133 Plus 2.0 gene arrays were used to define the AD skin phenotype or transcriptome (defined as differentially expressed genes [DEGs] between lesional and nonlesional skin) by using criteria of a fold change of 2 or greater and false discovery rate of less than 0.05 (see Table E6 in this article’s On- line Repository at www.jacionline.org), with DEGs depicted in 2 heat maps representing changes in the established robust meta- analysis–derived atopic dermatitis (MADAD) transcriptome (Fig 2)20,51,64 and in this study’s transcriptome (see Fig E3 in this article’s Online Repository at www.jacionline.org). Progres- sive improvements in genomic profiles were seen with both tran- scriptomes in the higher doses, with red-to-blue or blue-to-red transitions in skin lesions resembling nonlesional skin by day 29 compared with minimal changes in the 20-mg and placebo groups (Fig 2, A, and see Fig E3, A). Significant gene expression alterations in the MADAD transcriptome toward a nonlesional skin profile were evident at day 15 in the 40- and 80-mg groups and in the 40-mg group at day 29 (Fig 2 and see Table E7 in this article’s Online Repository at www.jacionline.org); smaller transcriptomic improvements were seen with the placebo and 20-mg groups, corresponding to modest clinical responses in these groups.49 The mean MADAD transcriptomic percentage improvements in the 40-mg group of 49.9% at day 15 and 79.29% at day 29 were significantly greater than the percentage improvements of 8.28% at day 15 and 21.81% at day 29 obtained in the placebo group (P < .001 at both time points; Fig 2, B). Sig- nificant genomic improvements compared with placebo of 45.77% (P < .001) and 22.91% (P < .01) were also seen at day 15 in the 80- and 20-mg groups, respectively. Similar genomic changes were seen in this study’s transcriptome, with significant mean transcriptomic improvements compared with placebo of 44.08% at day 15 and 48.26% at day 29 in the 80-mg group(P < .001) and 60.05% at day 15 and 69.25% at day 29 in the 40-mg group (P < .001; see Fig E3, B). FIG 1. Histology and barrier changes. A-C, Representative histologic images in the placebo and 40-mg treat- ment groups of hematoxylin and eosin (H&E; Fig 1, A), K16 (Fig 1, B), and FLG (Fig 1, C). The number in the lower right corner of the image for K16 staining (Fig 1, B) shows the number of patients with K16 staining after treatment. D, Fold change (FCH) in epidermal thickness at days 15 and 29 versus baseline in lesional skin. E, FCH of K16 mRNA expression in lesional and nonlesional skin at days 15 and 29 versus baseline. F, FCH of CD31 T cells in lesional and nonlesional skin at days 15 and 29 versus baseline. G, FCH of CD11c1 DCs in lesional and nonlesional skin at days 15 and 29 versus baseline. 1P < .1, *P < .05,**P < .01, and ***P < .001. Black stars, Significance of comparison between the placebo and ASN002 groups; red stars, significance of comparison versus baseline. FIG 2. MADAD transcriptomic changes. A, Heat map of genomic changes in the MADAD transcriptome in groups receiving 80, 40, and 20 mg of ASN002 compared with the placebo group. DEGs were defined by criteria of a fold change of greater than 2 and a false discovery rate of less than 0.05. Upregulated (red) and downregulated (blue) genes at baseline in lesional (LS) and nonlesional (NL) skin and after treatment with respective doses (80, 40, and 20 mg) in lesional skin at day 15 and day 29. B, Overall percentage im- provements in the MADAD transcriptome at days 15 and 29 with respective treatment doses of 80, 40, and 20 mg compared with placebo. **P < .01 and ***P < .001 compared with the placebo group on respec- tive treatment days. The light blue bar on the box plot represents mean 6 SEM values the groups treated with higher ASN002 doses compared with baseline. There was significant progressive suppression of TH2- related (IL4R, CCL13/monocyte chemoattractant protein 4, CCL17/thymus and activation-regulated chemokine, CCL18/pul- monary and activation-regulated chemokine, CCL22/macro- phage-derived chemokine, TNFSF4/OX40L, and FCER1A), TH17/TH22-related (IL36RN, IL36G, LCN2, and S100A7/S100A8/S100A9), and TH1-related (OASL, MX1, and IL12RB2) genes in the 40-mg group at both day 15 and day 29 compared with baseline (P < .05, Fig 3). Similarly, there were significant reduction in genes related to TH2 (CCL18, CCL22, CCL13, and FCER1A), TH17 (LCN2, IL36G, and PI3/elafin), and TH1 (OASL and MX1) markers at day 29 in the 80-mg group compared with baseline (P < .05, Fig 3). There was also significant FIG 3. Immune gene subset. Heat map of mean gene expression of a curated list of AD immune genes based on previous profiling studies. The table shows gene symbols with fold changes at days 15 and 29 versus day 1 for each treatment group of 80, 40, and 20 mg. 1P < .1, *P < .05, and **P < .01. LS, Lesional; NL, nonlesional downregulation of genes related to T cells/T-cell activation (IL15, CCR7, ITK, CD28, and CD5), DCs (CD1b and CD83), and gen- eral inflammation (MMP12 and TNFSF10/TRAIL), mostly in the 40-mg group (P <.05, Fig 3). Markers of general inflammation (MMP12) and DCs (CD1B, CD83, and CD86) were also significantly downregulated in the 80-mg group at day 29 (P < .05, Fig 3). Negative regulator genes (IL34 and IL37) were signifi- cantly upregulated in the 40-mg group at day 29, with only IL34 expression upregulated in the 80-mg group (P < .05, Fig 3). There were minimal gene modulations in the 20-mg and pla- cebo groups. ASN002 at 80 mg also significantly upregulates several barrier-related markers, including FLG2 and scielin (SCEL; P < .05), with FLG showing similar trends (P 5 .06). Additional barrier measures, including the tight junction marker claudin 23 (CLDN23) and CDH11, were significantly increased by 40 mg of ASN002 (P <.05, see Table E8 in this article’s Online Repository at www.jacionline.org).We confirmed modulation of selected key AD immune markers using qRT-PCR, as shown in individual graphs (Fig 4 and see Fig E4 in this article’s Online Repository at www.jacionline.org) and summary heat maps for all patients (Fig 5).20,55,57,66 Correspond- ing to microarray data, ASN002 decreased mRNA expression of markers related to general inflammation (MMP12), T-cell/natural killer (NK) cell activation (IL15, IL15RA, and IL2RA), innate im- munity (CXCL8/IL8 and IL17C), and the TH1 (IFNG, STAT1, CXCL9, CXCL10, IL12RB1, and CCL5), TH2 (IL13, IL31, IL10, IL4R, CCL13/monocyte chemoattractant protein 4, CCL17/ thymus and activation-regulated chemokine, CCL18/pulmonary and activation-regulated chemokine, CCL22/macrophage- derived chemokine, and CCL26/eotaxin-3), TH9 (IL9), and TH17/TH22 (IL22, IL32, S100A7, S100A8, S100A9, S100A12, IL19, PI3/elafin, CCL20, and IL23p19) axes (Figs 4 and 5 and see Fig E4). mRNA expression of markers related to T-cell/NK cell activation (IL15, IL15RA, IL2RA) and TH1 (IL12RB1, STAT1, CXCL9, and CXCL10) was significantly decreased compared with baseline and also compared with placebo at day 29, predominantly for the 40-mg group (P < .05, Fig 4 and see Fig E4). IL13 expression was significantly downregulated in both the 40- and 80-mg groups compared with baseline (P < .05 for both) and also compared with placebo for the 40-mg group FIG 4. Fold change (FCH) of inflammatory markers measured by using qRT-PCR. Fold changes of inflamma- tory markers in lesional skin at days 15 and 29 in the 80-mg, the 40-mg, and 20-mg treatment groups compared with the baseline and placebo groups are shown. Black stars, Significance of comparison be- tween the placebo and ASN002 groups; red stars, significance of comparison versus baseline. 1P < .1, *P < .05, **P < .01, and ***P < .001. FIG 5. Heat map of genes measured by using qRT-PCR. Heat map of mean gene expression measured by using qRT-PCR in lesional skin at days 1, 15, and 29 and in nonlesional skin at day 1. The table shows gene symbols with fold changes at days 15 and 29 versus day 1 for each treatment group of 80, 40, and 20 mg. 1P < .1, *P < .05, and **P < .01. LS, Lesional; NL, nonlesional (P <.05, Fig 4). Expression of TH2-induced chemokines (CCL13, CCL18, CCL22, and CCL26) was reduced in both the 40- and 80- mg groups, achieving greater significance compared with base- line and placebo for the 40-mg dose (Fig 4). Additional TH1 (IFNG), TH2 (IL4R), and TH17/TH22 (IL22, IL19, IL23A/ IL23p19, PI3/elafin, S100A7/S100A8/S100A9/S100A12, and IL32) axis genes were significantly downregulated in the 40-mg group compared with baseline at day 29 (P < .05, Figs 4 and 5 and see Fig E4). IL9/TH9 expression was downregulated at both higher doses at day 29, approaching significance in the 40-mg group (Fig 5 and see Fig E4). The T-regulatory marker FOXP3 showed a significant decrease in expression at day 29 in the 40- mg group (Fig 5). Epidermal cytokine and/or innate immune markers (CXCL8/IL8 and IL17C) showed small reductions with ASN002, approaching significance for IL17C in the group receiving 40 mg at day 29 and for CXCL8/IL8 in the group receiving 80 mg at day 29 (Fig 5 and Fig E4). A summary of RT-PCR and immunohistochemistry data by patient (ordered by EASI50 response) is shown as a heat map in Fig E5 in this article’s Online Repository at www.jacionline.org. To evaluate the relationship between clinical responses to ASN002 and changes in expression of molecular and cellular markers evaluated in lesional skin, percentage EASI score and percentage body surface area (BSA) improvements were correlated with changes in inflammatory and epidermal hyper- plasia (K16) markers at days 15 and 29 (Table I). Spearman cor- relation coefficients were used, with significance determined by a P value of less than .05. A heat map depicting correlations be- tween skin biomarkers with clinical scores and each other at days 15 and 29 are shown in Fig E6 in this article’s Online Re- pository at www.jacionline.org, with correlations between clin- ical severity and biomarkers shown in Table E9 (day 15) and Table E10 (day 29) in this article’s Online Repository at www. jacionline.org. Already at day 15, percentage EASI improvement significantly correlated with improvements in lesional expressions of markers of general inflammation (MMP12), TH1 (IFNG), TH2 (IL13, IL5, CCL13, CCL17, CCL22, and CCL26), and TH17/TH22 (IL22 and CCL20; P < .05; Table I). At day 29, percentage EASI score im- provements significantly correlated with additional markers, including those related to innate immunity (IL6), T-cell/NK cell activation (IL15), and TH1 (CXCL10) and TH17/TH22 (S100A7, S100A9, and S100A8) axes (P < .05, Table I). Levels of the epidermal hyperplasia marker (KRT16) also correlated with per- centage EASI score improvement at day 29 (P 5 .05, Table I).Improvement in BSA showed similar trends, showing strong significant correlations with changes in general inflammation (MMP12), TH2 (IL13, CCL22, and CCL17), and TH17/TH22 TABLE I. Biomarker correlations with clinical improvements levels), general inflammation (MMP12), T-cell/NK cell activation (IL2RA, IL15, and IL15RA), epidermal cytokine/innate immunity (IL6, IL17C, CXCL8/IL8, and IL1B), and TH1 (IL12RB1, jacionline.org). Although skin univariate analyses yielded signif- icant correlations, an integrated m score of lesional skin bio- markers showed a higher and more significant correlation (r 5 0.66, P 5.0006). The biomarkers included in the m score cor- relation included the innate cytokine IL6, the TH1/interferon- related chemokine CXCL10, the TH2-related chemokine CCL13, and the general inflammatory marker MMP12, reflecting various pathways modulated by JAK/SYK inhibition. On inte- grating skin and serum biomarkers, we obtained an increased cor- relation with clinical improvement (r 5 0.76, P 5 .00003), with an integrated biomarker profile that includes innate immunity(IL6), general inflammation (MMP12), T 2 (CCL13), and T 1/ (IL22 and CCL20) markers as early as day 15 (P < .05, Table I), with additional markers showing correlations at day 29 (P < .05, Table I).We also evaluated for correlations between reversal of the epidermal hyperplasia measured by using a histologic score (histoscore; obtained by averaging z scores for changes in epidermal thickness and K16 expression)52,54 and changes in bio- markers and clinical scores at days 15 and 29 (Table II). A summary of correlations with the histoscore is found for days 15 and 29 for Tables E11 and E12 in this article’s Online Repos- itory at www.jacionline.org. As expected, we found significant correlations between histoscore improvement and changes in epidermal hyperplasia (S100 levels, thickness, and KRT16. We evaluated the effect of ASN002 on previously defined AD- related gene signatures of upregulated or downregulated genes. These gene pathways include those related to T cells, DCs, epidermal differentiation complex, untreated keratinocytes and cytokine-treated keratinocytes, and T-cell cytokine re- sponses.20,55,57,64,69-72 Pathway improvements were quantified by percentage change from baseline in each treatment group compared with placebo. ASN002 induced strong and significant improvements compared with placebo in most evaluated path- ways at both higher doses by day 29 (Fig 6). Specifically, path- ways related to immune genes, genes upregulated in T cells, DCs, untreated keratinocytes, and keratinocyte treated by IL-1, IL-4, IFN-g, and TNF-a showed significant improvements from baseline in the 40-mg group and some in the 80-mg group compared with placebo by day 29 (P < .05, Fig 6). Gene sets related to IL-17–treated keratinocytes and the TH22/IL-22 pathway showed significant improvements of greater than 100% in the 40-mg group versus the placebo group (P <.05, Fig 6). The 20-mg and placebo groups showed only modest onset of clinical efficacy in patients with moderate-to-severe AD, with greater than 80% of patients achieving 50% or greater reduction in EASI scores at the higher (40 and 80 mg) doses and rapid significant reduction in Pruritus Numerical Rating Scale scores, with limited clinical efficacy in the 20-mg and placebo groups.49 The current study evaluated the effects of ASN002 on cellular infiltrates, inflammatory pathways, and skin barrier ab- normalities implicated in AD pathogenesis. At both higher doses, ASN002 suppressed CD31 T-cell and CD11c1 DC infiltrates and improved the 3 major components of barrier defects in patients with AD, including epidermal hyper- plasia, S100As, and epidermal differentiation measures, such as 10,20,65,72-77 in the epidermal barrier were paralleled by significant reductions in levels of TH2 (IL-13 and IL-31) and TH22/TH17 (IL-22 and S100) pathway cytokines,33,52,72 which are key to the pathogen- esis of AD and have been linked to downregulation of FLG74 and other epidermal barrier differentiation measures, such as tight junctions.33,52,72,74,76-80.The AD genomic signature (MADAD transcriptome)64 showed significant improvements in both the 40- and 80-mg groups as early as day 15, with further changes toward the nonlesional phenotype by day 29 and approximately 80% improvement from baseline in the 40-mg group. No significant changes were Number of Genes in Each Pathway FIG 6. Gene pathway analysis. Percentage improvements from baseline dysregulation of selected AD- related gene signature pathways in lesional skin at day 29 in respective treatment groups. Numbers at the bottom of the graph indicate numbers of genes in each gene set. 1P < .1, *P < .05, and ***P < .001. Black stars, Significance of comparison between placebo and ASN002 general inflammation (MMP12), TH2-related (IL4R, IL10, IL13, CCL13, CCL17, CCL18, CCL22, CCL26, and FCER1A), TH17/ TH22-related (LCN, PI3/elafin, CCL20, IL12RB2, S100As, IL36G, and), TH1-related (IFNG, CXCL9, CXCL11, MX1, and OASL), and barrier-related (FLG and CLDN23) products. In par- allel with cellular and molecular improvements, we also observed significant improvements in gene signatures of cytokine-treated keratinocyte and AD-associated pathways.11,20,52,66,67 These sig- nificant molecular improvements are likely due to the successful inhibition of JAK-STAT, as well as SYK, signaling pathways because JAK family kinases are involved in the TH2 (IL4, IL5, IL13, and thymic stromal lymphopoietin [TSLP]), TH22 (IL22), TH1 (IFNG and IL12),26-32 and TH17 (IL17)81 pathways and T- cell/NK cell/TH1 activation (IL15).82 JAK signaling has also been shown to have a direct effect on neuronal itch.81,82 Our data showed significant and robust early modulation of the itch cytokine IL-31, which signals through JAK1/JAK2, with 40 mg of ASN002. SYK signaling further regulates TH17, B-cell, and DC signaling34-39 and inhibits keratinocyte terminal differentiation.34-41,83.The rapid inhibition of various axes and amelioration of barrier pathology can be induced by the broadened spectrum of cytokines targeted with JAK/SYK inhibition.31,35-38,84,85 Previous studies have shown that concomitant JAK/SYK inhibition has a synergis- tic therapeutic effect on inflammatory diseases.33,85 Thus concomitant SYK inhibition can increase the clinical efficacy of JAK inhibition in patients with AD. Our study also correlated clinical responses to changes in molecular and histologic tissue responses. We found significant correlations between improvements in clinical scores (EASI and BSA) and reductions in levels of inflammatory markers, epidermal hyperplasia, and the composite histoscore combining epidermal thickness and K16 mRNA expression as early as day 15.54 Markers significantly correlated with clinical improvements are among those crucial to the AD pathogenesis and barrier abnor- malities, including TH2 (IL5 and IL13), TH17/TH22 (IL22 and CCL20), and TH1 cytokines and/or their associated chemokines (eg, CCL13/CCL17/CCL22, CCL26, CXCL10, S100As, and the epidermal proliferation/hyperplasia marker K16). Among skin biomarkers significantly associated with EASI and BSA score im- provements was also the proinflammatory cytokine IL-6, which has been associated with atherosclerosis and induction of C-reac- tive protein.86,87 Even greater correlations were obtained between changes in tissue biomarkers and the histoscores. Particularly high correlations were obtained between improvements in histo- scores at both day 15 and day 29 with IL-22/IL-17–induced S100As, as well as other IL-22/IL-17–associated markers, including PI3/elafin, IL23p19, and CXCL1. High correlations were also found between improvements in histoscores and sup- pression of TH2-associated measures, such as IL4R, CCL17, CCL22, and CCL18. Although individual serum biomarkers did not correlate with clinical disease improvements,49 on integrating changes in skin and serum biomarkers, the correlations with clin- ical improvement improved to 0.76 (compared with <_0.64 with individual skin biomarkers), with a profile that includes primarily skin biomarkers (IL6, TH2/CCL13, TH1/T-cell activation/IL15RA, and MMP12), which reflects the cytokine modulation by ASN002. Although skin biomarkers are generally more sensitive for assessing molecular improvement,49 combining skin and blood measures provides the most robust biomarker approach. These significant correlations might help define a set of tissue disease response biomarkers that can potentially serve to monitor treatment response in future studies with JAK and/or SYK inhibitors. Many of these biomarkers have been previously identified as key to AD pathogenesis and were shown to mirror treatment responses in patients with AD treated with broad (cyclosporine,20 narrow-band UVB,67 topical steroids,55 and cal- cineurin antagonists)73 and specific agents targeting the TH2(dupilumab54 and GBR83088), TH22 (fezakinumab),52 and TH17/ IL-23 (ustekinumab)89 axes. Although ASN002 potently in- hibited TH2-related (CCL17/CCL18/CCL22/CCL26), TH22/ TH17-related, and hyperplasia-related (S100As and K16) genes similar to dupilumab, it also suppressed TH1-related products (CXCL9/CXCL10, STAT1, and IL12RB1) that were not modulated by dupilumab.54,66 Of note, 40 mg of ASN002 showed more robust and significant inhibition of mRNA expression of the itch cytokine IL31 at both weeks 2 and 4 compared with dupilu- mab at either 4 or 16 weeks.54 Although overall TH2 inhibition with ASN002 at 4 weeks is similar to dupilumab at 4 weeks, greater TH17/TH22 inhibition is seen after 4 weeks of ASN002 treatment compared with dupilumab treatment for both 4 and 16 weeks, likely because of the indirect inhibition of the TH17/ TH22 pathways with dupilumab, whereas ASN002 exerts robust TH17/TH22 inhibition through SYK/TYK2 signaling. Because SYK and TYK2 are involved in IL-23/TH17/TH22 signaling, we believe that these data explain the particularly robust and early inhibition of the TH17/TH22 pathway in our data, as well as the ef- fects on hyperplasia. This is also supported by the pathway analyses we provide that show the greatest improvements in the TH17 and TH22 pathways (>100% improvement). Of note, inhibition of TH17/TH22 pathways with ASN002 in our AD cohort is much greater than that seen in patient with psoriasis, a TH17- centered disease, at weeks 1 and 2 with the pan-JAK inhibitor to- facitinib, which took much longer to improve psoriasis clinically and molecularly, likely because of the lack of direct modulation of the TH17 pathway.
A comparison of treatment response bio- markers across different therapeutics is listed in Table E13 in this article’s Online Repository at www.jacionline.org. As shown in the Table E13, ASN002 at 4 weeks shows overall larger responses in TH17/TH22 markers, as well as some TH2/TH1 markers, compared with other broad or specific systemic treatments, including narrow-band UVB,67 ustekinumab/anti-IL-12/ 23p40,92 fezakinumab/anti–IL-22,52 and GBR830/anti-OX40 at later weeks,88 as well as compared with a mild-to-moderate topical steroid at 4 weeks (see Table E13).55 Percentage improve- ment in the AD transcriptome with 40 mg of ASN002 at 4 weeks was greater than that seen with dupilumab at 4 weeks, and the magnitude of TH17/TH22 modulation at week 4 is even larger than with dupilumab at week 16.
Overall, ASN002 showed significant cellular and molecular suppression of key AD inflammatory pathway, such as including those related to the TH2, TH22/TH17, and TH1 pathways, and im- provements in epidermal barrier measures. Although AD is pri- marily TH2/TH22 driven, it is a heterogeneous disease with variable contributions of TH1 and TH17 immune pathways in distinct disease subtypes.6-8,13-15,66,93 For example, AD in both pediatric and Asian adult populations have strong TH2 and TH17 polarization, whereas AD in European American adults
10,12,94 of IL-17–induced release of CCL20 in keratinocytes by ASN002 compared with tofacitinib, a pan-JAK inhibitor, which indeed did not show robust TH17/TH22 modulation at week 2 in patients with psoriasis.90
Our study had several limitations, including a small sample size, short study duration of only 4 weeks, and nonsignificant differences in baseline EASI scores between the groups (the 20- and 80-mg groups were more severe with an EASI score of approximately 29, whereas the 40-mg and placebo groups had EASI scores of approximately 21), perhaps accounting for the greater clinical and biomarker effects seen in the 40-mg group versus the 80-mg group and the lack of efficacy in the 20-mg group. Future larger and longer studies in groups of diverse ethnicity and age are needed to further assess the clinical efficacy and associated molecular response to ASN002 in patients with moderate-to-severe disease across the AD spectrum.In summary, ASN002 couples clinical ASN-002 responses with robust tissue improvements in cellular infiltrates, inflammatory path- ways, and the associated barrier pathology of AD as early as day 15, with sustained progressive changes until day 29.