Guselkumab Reduces Disease- and Mechanism-Related Biomarkers More Than Adalimumab in Patients with Psoriasis: A VOYAGE 1 Substudy

Background Psoriasis is an immune-mediated inflammatory disease characterized by activation of IL-23–driven IL-17–producing T cell and other IL-23 receptor–positive IL-17–producing cell responses. Selective blockade of IL-23p19 with guselkumab was superior to blockade of TNF-α with adalimumab (ADA) in treating moderate-to-severe psoriasis. Objective: Pharmacodynamic responses of guselkumab versus ADA were compared in patients with psoriasis in VOYAGE 1. Design Inflammatory cytokine serum levels were assessed (n = 118), and lesional and nonlesional skin biopsies were collected (n = 38) in patient subsets at baseline and 4, 24, and 48 weeks after treatment to evaluate pharmacodynamic responses of guselkumab versus those of ADA. Results Guselkumab provided rapid reductions in serum IL-17A, IL-17F, and IL-22 levels by week 4 versus at baseline, which were maintained through weeks 24 and 48 (P < .001). The magnitude of reduction of IL-17A and IL-22 at week 48 and IL-17F at weeks 4, 24, and 48 were greater with guselkumab than with ADA (all P < .05). In the skin, guselkumab reduced the expression of IL-23/IL-17 pathway–associated and psoriasis-associated genes. Conclusion These data provide extensive characterization of pharmacodynamic anti-inflammatory responses to IL-23p19 and TNF-α inhibition in human blood and tissue over time with FDA-approved doses of guselkumab and ADA. Trial registration:ClinicalTrials.govClinicalTrials.gov (NCT02207231).

IL-23, a key regulatory cytokine produced by activated antigen-presenting cells, such as inflammatory monocytes and dendritic cells in psoriatic skin (Mehta et al, 2021), affects the expansion and maintenance of T17 cells (eg, CD4 þ T helper 17 and IL-17eproducing CD8 þ T cells) through transcriptional control regulated by RORgt (Gaffen et al, 2014;Kim et al, 2022;Teng et al, 2015).T17 cells, which secrete the proinflammatory effector molecules IL-17A and IL-17F, are a significant source of IL-17A in psoriatic skin and play a pivotal role in psoriasis pathogenesis (Elloso et al, 2012;Hawkes et al, 2018;Kim et al, 2022).IL-17A directly regulates keratinocyte-expressed genes involved in innate immune defense, including antimicrobial defensins, S100 family proteins, and lipocalin, as well as a range of C-X-C motif chemokine ligands that regulate neutrophil trafficking (Krueger et al, 2012).IL-17F shares a gene locus with, is coexpressed with, and has a similar protein structure and biologic activity as IL-17A; IL-17F is elevated in the sera and lesional skin (LS) of patients with psoriasis (Blauvelt and Chiricozzi, 2018;Elloso et al, 2012;Johansen et al, 2009;Reich et al, 2021;Soderstrom et al, 2017).IL-23 also affects IL-22 production from both T17 and T helper 22 cells, which have been identified as a source of IL-22 in normal and psoriatic blood and skin (Basu et al, 2012;Benham et al, 2013;Hijnen et al, 2013).IL-22 induces keratinocyte hyperplasia, increases synthesis of S100 proteins, accelerates loss of surface keratinocytes, and helps to eliminate pathogens (Fujita, 2013;Nograles et al, 2008).IL-22 can also have effects divergent from those of IL-17 associated with disruption of normal keratinocyte differentiation in psoriasis (Lowes et al, 2014;Nograles et al, 2008).Thus, both T17 and T helper 22 cells are increased in psoriatic plaques, and the cytokines they produce are critical for inflammatory processes in psoriatic skin.
Guselkumab is a fully human, IgG gamma mAb that inhibits IL-23 by targeting the p19 subunit (Sofen et al, 2014).Guselkumab binds selectively to human IL-23 with high specificity and affinity and blocks extracellular IL-23 binding to the cell surface IL-23 receptor, disrupting IL-23emediated signaling, activation, and cytokine cascades (Zhuang et al, 2016).In the phase 3 VOYAGE 1 trial, guselkumab demonstrated superior efficacy compared with placebo and the TNF-a antagonist adalimumab (ADA) in patients with moderate-to-severe plaque psoriasis (Blauvelt et al, 2017).In this study, peripheral blood and skin samples were collected from patients in VOYAGE 1, and pharmacodynamic (PD) responses to guselkumab (vs ADA) treatment were assessed over 48 weeks.
Serum samples were collected from all patients, and a random subset of 118 demographically matched patients was identified for serum biomarkers analysis.In addition, LS and non-LS (NL) samples were collected from 38 patients who provided consent to participate in an optional skin biopsy substudy to evaluate PD effects on psoriasis transcriptomics.Baseline demographic and disease characteristics were generally comparable across the randomized treatment groups, although minor imbalances were noted in the skin biopsy population (younger mean age in the ADA group, higher body mass index for the guselkumab group, lower body surface area with psoriasis for the placebo group), which may reflect the overall limited number of participants in this substudy (Table 1).
Outcomes PD effect on serum cytokines associated with the IL-23/IL-17 pathway.
Serum analytes included in this study were selected on the basis of an analysis of previous guselkumab phases 1 and 2 psoriasis studies (Gordon et al, 2015;Sofen et al, 2014) and a literature review (Blauvelt et al, 2015b;Sua ´rez-Farin ˜as et al, 2012;Teng et al, 2015).Baseline substudy serum samples from patients with psoriasis (n ¼ 118) contained significantly (all P < .01)higher concentrations of IL-17A, IL-17F, IL-22, IL-8, CCL22/macrophage-derived chemokine (MDC), and CCL4/MIP-1b than an independent, healthy control serum cohort (n ¼ 25) (Figure 1 and Table 2).Baseline concentrations of these key cytokines were comparable across treatment arms (Figure 1 and Table 2).In contrast, baseline serum IL-23 levels were not significantly different between participants with psoriasis and healthy controls (P ¼ .10).
Both guselkumab and ADA significantly reduced serum CCL22/MDC levels versus baseline at weeks 4, 24, and 48 (all P < .001)(Figure 2e); however, there were no significant differences between them.ADA but not guselkumab significantly reduced CCL4/MIP-1b serum levels at weeks 4, 24, and 48 (all P < .001)(Figure 2f), which is consistent with targeting TNF-a (Roach et al, 2002).Both treatment groups demonstrated nonsignificant trends for reduction in serum CXCL8/IL-8 from baseline through week 48 (data not shown).

Psoriatic skin transcriptomic profiling at baseline.
A 1832gene, core psoriatic skin transcriptomic signature (PSTR) previously identified through a meta-analysis (meta-analysis derived-3 transcriptome) (Tian et al, 2012) was used to assess differential gene expression between LS and NL biopsies at baseline in VOYAGE 1 through microarray profiling.Among these, 1824 (99.6%) genes exhibited the same direction of differential expression in VOYAGE 1 skin samples as shown in the meta-analysis, highlighting the consistency in the PSTR across studies (r ¼ 0.881) (Figure 3a); however, the overall magnitude of differential gene expression was compressed in this study (slope ¼ 0.524) (Figure 3a).This may be due to the use of an Affymetrix chip that utilized only perfect match probes (Affymetrix GeneChip HC HG-U133 Plus perfect match 96-array), compared with previous studies where data A Blauvelt et al.
were generated using chips with both perfect match and mismatch probes.To accommodate this difference, we chose a fold change (FC) >1.5 (instead of 2) and associated false discovery rate (FDR) <0.05 as selection criteria to distinguish differentially expressed genes (DEG).A total of 3323 individual probe sets, representing 2383 genes, were identified as the PSTR between LS and NL at baseline by paired t-test among 37 patients.
PD effect on PSTR in response to treatments.
Percentage of improvement or normalization in expression of individual probe sets (genes) was used to quantify the PD treatment   3b).Similarly, a larger number of PSTR genes achieved >75 and >90% improvement in expression with guselkumab than with ADA at all 3 time points, with larger differences at weeks 4 (32 vs 8% for >75% improvement and 14 vs 2% for >90% improvement) and 48 (87 vs 60% for >75% improvement and 74 vs 30% for >90% improvement) (Figure 3c and d).The majority of PSTR genes that achieved >75% improvement with ADA (63, 94, and 96% at weeks 4, 24, 48, respectively) also achieved >75% improvement with guselkumab.In contrast, 27, 20, and 30% of PSTR genes that achieved >75% improvement were unique to guselkumab at weeks 4, 24, and 48, respectively, whereas only 3, 5, and 3% of PSTR genes that achieved >75% improvement were unique to ADA at weeks 4, 24, and 48,  respectively (Figure 3eeg).Improvement in the PSTR in LS biopsies (n ¼ 102) after treatments was positively correlated with corresponding improvement in PASI (r ¼ 0.68, P < .001)(Figure 3h).The residual disease expression profile or molecular scar for a given treatment was defined as a PSTR gene subset with <75% improvement and remaining FC >1.5 in healed LS (HLS) collected from patients when they achieved a PASI 100 response.As shown in Figure 3i, 3.5% of PSTR genes continued to be differentially expressed in ADA-treated HLS (n ¼ 6) versus only 0.5% in guselkumab-treated HLS (n ¼ 25), with 0.4% of PSTR genes common to both (3.2 vs 0.1% were unique to ADA and guselkumab, respectively).The differences in transcriptional changes reported in Figure 3i might be more directly related to the differences in the treatment mechanisms of action between guselkumab and ADA.  in the PSTR were observed in the majority of these pathways with guselkumab versus ADA, at all 3 time points assessed (Figure 4 and Table 3).The largest early treatment differences (guselkumab minus ADA percentage improvement at week 4) were observed for pathways involved in cell cycle control, hyperplasia, and altered differentiation that characterize psoriasis (ie, mitotic roles of polo-like kinase  3).
Improvement in disease-associated gene expression in immune/ skin inflammation gene signatures.
We further evaluated immune and skin inflammation gene signatures reported to be associated with psoriasis (Brodmerkel et al, 2019).Overall differential expression between LS and NL at baseline was evaluated in 124 curated gene sets by gene set variation analysis (GSVA).A total of 108 gene sets exhibited significant differences (FDR < 0.05) in GSVA scores between LS and NL biopsies (Figure 5a), whereas differences in GSVA scores between LS and NL biopsies at baseline were more efficiently normalized by guselkumab than by ADA at each follow-up time point (Figure 5b).
Among these, 71 gene sets were further selected as enriched with PSTR genes (on the basis of FDR < 0.05 from a hypergeometric enrichment test and >20% in psoriasis DEG profile and >5 in the PSTR gene set).Consistent with findings in Ingenuity Pathway Analysis pathways and GSVA results, better improvements in the PSTR were observed with guselkumab than with ADA in almost all of these gene sets, at all 3 time points, with larger differences at weeks 4 and 48 (Figure 6 and Table 4).
The largest early treatment differences at week 4 were observed among gene signatures modulated by IFN-a, IL-1, and TNF-a in keratinocytes (Figure 6a and Table 3), whereas the largest treatment differences at week 48 were observed among gene signatures modulated by IL-17 and IL-22 (Figure 6c and Table 3).Better improvement in genes enriched for regulatory T cells were detected with guselkumab than with ADA as early as week 4 and were sustained through week 48 (Pfoertner et al, 2006).
We used qRT-PCR to confirm and extend observations from the skin gene expression profiles determined by microarray analyses.A total of 121 genes from 38 skin biopsy samples (guselkumab, n ¼ 17; ADA, n ¼ 13; placebo to guselkumab, n ¼ 8) were profiled using a Fluidigm Real-Time PCR panel (complete results are provided in Table 5).The baseline LS versus NL differences and weeks 4, 24, and 48 LS FC versus baseline within the 3 treatment groups are summarized in Table 6.Six genes from the Fluidigm RT-PCR panel failed quality control and were reprofiled using a standard qRT-PCR method.Results from 3 of the rerun genes (IL17A, IL17F, and IL22) plus 6 other genes (IL23A, IL12A, TNF, IFNg, DEFB4A, and CXCL10) profiled using the Fluidigm RT-PCR panel are summarized in Figure 7aei.Findings show that guselkumab treatment led to reductions in both IL-17A and IL-22 expression as early as week 4, whereas ADA showed almost no effect on IL-22 expression and minimal (nonsignificant) reductions of IL-17A at the same early time point.

DISCUSSION
We characterized blood and tissue PD responses in individuals with psoriasis treated with IL-23p19 blockade using guselkumab versus TNF-a blockade using ADA over the course of 48 weeks.Serum biomarkers associated with psoriasis and the IL-23/IL-17 and IL-22 immunologic pathways were elevated at baseline and were reduced with guselkumab to a greater extent than with ADA.Aberrant psoriasisassociated gene expression was documented in LS versus NL at baseline, with gene expression normalization occurring as early as week 4 (and progressively sustained over 48 weeks) with guselkumab; the breadth and depth of gene expression normalization were more extensive with guselkumab than with ADA at all time points assessed.These data provide a comprehensive characterization of PD antiinflammatory responses in human blood and tissue with IL-23p19 and TNF-a inhibition over time using FDA-approved doses of guselkumab and ADA, respectively.
Baseline serum IL-17A, IL-17F, and IL-22 levels were shown to be higher in patients with moderate-to-severe psoriasis than in healthy controls.Serum CCL22/MDC and CCL4/MIP-1b levels also were elevated at baseline in the psoriasis study population versus healthy controls (Table 2 and Figure 1).These data are consistent with previous observations that markers associated with the IL-23/IL-17 pathway (ie, IL-17A, IL-17F, IL-22, and CCL22/MDC) are elevated in serum from patients with psoriasis (Fotiadou et al, 2015;Sua ´rez-Farin ˜as et al, 2012).IL-23 levels were consistently low in the serum at baseline, and no significant changes in serum IL-23 levels were observed with guselkumab treatment.The Single Molecule Counting Erenna Assay used to measure serum IL-23 only detects free IL-23 and not guselkumab-bound IL-23 (data on file).Therefore, the lack of effect of guselkumab on serum IL-23 is likely not due to interference of guselkumab in the detection of free IL-23.In addition, IL-23 is a local mediator of inflammation by activating IL-23 receptorepositive cells in the tissue microenvironment (Pawlak et al, 2022;Sherlock and Cua, 2021), and serum IL-23 may not be reflective of local IL-23 levels in LS.The enhanced impact on the transcriptomic profile with guselkumab supports a mechanistic basis for differences beyond simply better clinical response.Furthermore, guselkumab treatment led to significantly greater reductions in IL-17A, IL-17F, and IL-22 serum levels than both placebo and ADA treatments.Guselkumab achieved rapid reduction in serum levels of IL-17A and IL-17F, which were significantly lower than at baseline by week 4 and were maintained at weeks 24 and 48.In addition, guselkumab achieved significantly greater reductions in IL-17A and IL-17F levels than placebo at all time points assessed.Similarly, guselkumab achieved significantly greater reduction of IL-17F serum levels at weeks 4, 24, and 48 than ADA, however, at only week 48 for IL-17A.Guselkumab also reduced serum IL-22 levels to a greater degree than ADA at week 48.
In addition to expected reductions in serum levels of effector cytokines associated with the IL-23/IL-17 and IL-22 pathways, other markers associated with psoriasis were also reduced to a greater extent with guselkumab than with placebo, including CCL22/MDC and CXCL8/IL-8, albeit to a moderate extent.However, ADA but not guselkumab decreased serum CCL4/MIP-1b levels, a marker associated with psoriasis and downstream of TNF-a (Pedrosa et al, 2011;Roach et al, 2002).This observation is aligned with the predominant role of the IL-23/IL-17 axis in driving psoriasis pathogenesis, which is selectively antagonized by guselkumab over ADA.
Taken together, these observations suggest that by blocking IL-23, a key regulatory cytokine that is required for the expansion of T17, T helper 22, and group 3 innate lymphoid  A Blauvelt et al.
Guselkumab Reduces IL-23/IL-17 Biomarkers cells, guselkumab reduces the ability of these cells to produce key effector cytokines.IL-23 also augments the production of IL-22 from group 3 innate lymphoid cells (Teng et al, 2015;Ward and Umetsu, 2014).Group 3 innate lymphoid cells lack rearranged antigen-specific receptors, rely on the transcription factor RORgt, and respond to IL-1b and IL-23; a subset of group 3 innate lymphoid cells is associated with production of IL-17A and IL-22 in psoriasis lesions (Villanova et al, 2014).In addition, IL-23 maintains some regulatory capacity for Tc17 cells (Ciric et al, 2009;Teunissen et al, 2014).Increased expression of genes induced by IL-17A in keratinocytes and TCR genes, including genes previously determined to be a part of the PSTR or present as a psoriasis molecular scar (Tian et al, 2012), was more efficiently normalized by guselkumab than by ADA at all time points assessed (Figure 4).Better improvement in expression of genes enriched for regulatory T cells by guselkumab than by ADA was detected as early as week 4 and sustained through week 48.This finding is consistent with a recent substudy of the phase 3 ECLIPSE trial that showed better maintenance of efficacy over time with guselkumab than with secukinumab, an IL-17A blocker (Reich et al, 2019).Among CD11c þ HLA-DR þ mononuclear phagocytes, CD64 bright CD163 À CD14 bright CD1c À CD1a À inflammatory monocyte-like cells were identified as the predominant IL-23eproducing cells in LS, and guselkumab but not secukinumab reduced the relative numbers of tissueresident memory T cells and maintained the levels of antiinflammatory regulatory T cells, which may contribute to prolonged maintenance of clinical response observed with guselkumab treatment (Mehta et al, 2021).
A limitation of this study is the challenge in separating normalization of the skin transcriptional profile from clinical improvements.Some of the differences in transcriptional changes at later time points (especially at week 48) might reflect differences in PASI responses.However, at week 4, there were no significant differences in PASI improvements between the guselkumab-and ADA-treated subgroups with skin biopsy samples for biomarker analyses (also shown previously in Blauvelt et al [2017] for all patients); therefore, early transcriptional changes may be independent of and precede clinical responses seen at later time points and may better reflect differences derived from blocking their respective therapeutic targets: IL-23p19 and TNFa (Gordon et al, 2019).In addition, IL-23 is a local mediator of inflammation that functions by activating IL-23 receptorepositive cells in the tissue microenvironment (Pawlak et al, 2022;Sherlock and Cua, 2021), and serum IL-23 may not be reflective of local IL-23 levels in LS.The enhanced impact on transcriptomic profile with guselkumab supports a mechanistic basis for differences beyond simply better clinical response.
In summary, IL-23p19 blockade by guselkumab in patients with psoriasis (vs TNF-a blockade by ADA) more dramatically reduced serum concentrations of cytokines IL-17A, IL-17F, and IL-22, which are associated with the IL-23/IL-17 pathway, thereby limiting the downstream activity of these effector inflammatory cytokines.Similarly, normalization of psoriasisassociated inflammatory genes was greater in LS samples from guselkumab-treated patients than in those from ADA-treated patients.Differences in protein and gene expression signatures with treatment should be interpreted in the context of differences in distribution of clinical responses between compared patient The differential expression in 124 curated gene sets related to immune cell and skin inflammation between LS and NL at baseline were evaluated by GSVA.71 gene sets were selected based on having significant differences in GSVA scores (FDR <0.05) and with !6% and >20% of genes in PSTR (2383 genes with FC >1.5).populations and, hence, significant differences to which associated signaling pathways have been suppressed.The molecular PD findings reported in this analysis are consistent with the differential clinical responses observed between guselkumab and ADA treatment in the VOYAGE 1 study and correlate with the depth of therapeutic responses seen with guselkumab versus ADA treatment in patients with moderate-to-severe psoriasis.

MATERIALS AND METHODS
Trial design and participants VOYAGE 1 (NCT02207231) was a phase 3, randomized, double-blind, multicenter, placebo-controlled, and active-comparatorecontrolled study of guselkumab in patients (male 72.6%, White 81.7%) with moderate-to-severe plaque-type psoriasis, which has been previously described (Blauvelt et al, 2017).Study participants were adults with a diagnosis of plaque-type psoriasis (with or without psoriatic arthritis) for !6 months before the first administration of study drug.Moderate-to-severe plaque-type psoriasis was defined by an Investigators Global Assessment !3, PASI !12, and an involved body surface area !10%.Eligible participants must have been candidates for and may have previously received either systemic therapy or phototherapy for psoriasis.

Intervention
At week 0, 837 eligible participants were randomized (2:1:2) to 1 of 3 treatment arms: group I (n ¼ 329, 100 mg guselkumab at weeks 0, 4, and 12 and every 8 weeks thereafter through week 44); group II (n ¼ 174, placebo at weeks 0, 4, and 12 and then 100 mg guselkumab at weeks 16 and 20 and every 8 weeks thereafter through week 44); and group III (n ¼ 334, 80 mg ADA at week 0, followed by 40 mg at week 1 and every 2 weeks thereafter through week 47).The open-label guselkumab treatment period began at week 48 and extended through week 264.Participants in all 3 treatment arms received 100 mg guselkumab at week 52 and every 8 weeks thereafter through week 252 (Figure 8).Sterling Institutional Review Board was a central United States institutional review board for the VOYAGE 1 studies (4809C).Additional details can be found in Supplementary Table S1.Participants provided written informed consent before study initiation.Biomarker data were collected through week 48 of the study.Exploratory analyses were not based on a prespecified analysis plan.

Biomarker assessments
Blood sample collection and analysis.
A Blauvelt et al. 1 psoriasis population and unaffected individuals, an independent, demographically matched healthy control cohort (n ¼ 25 serum samples collected independent of the VOYAGE 1 study) (Table 7) was obtained commercially from Bioreclamation.Blood samples were analyzed at Janssen Research & Development, LLC.
Skin sample collection and analysis.
LS biopsies were isolated from a representative psoriatic target lesion (!3 cm) located on the trunk or extremities with approximately !0.5 mm of induration and approximately 5 mm from the edge of the plaque.Four 4-mm skin biopsies were obtained from the identified target lesion at week 0 prior to injection of study agent and at weeks 4, 24, and 48 after treatment (1 biopsy per visit).One additional punch biopsy was obtained at week 0 from uninvolved (ie, macroscopically normal or NL) skin at a body site similar to that of the biopsied lesion.All skin samples were collected, snap frozen in liquid nitrogen, and stored at À80 C until RNA extraction.RNA was extracted using QIAsymphony RNA Kit (Qiagen) and later hybridized to GeneChip HT HG-U133þ perfect match 96-Array Plate (Affymetrix).

qRT-PCR.
Data from qRT-PCR were generated by BioProcessing Solutions (RUCDR Infinite Biologics) with a Fluidigm BioMARK system using a panel for 121 preselected genes.Raw data were processed using Fluidigm RT-PCR software.b-Actin and 18s assays were run as controls on each array and for each gene/target.Delta Ct was calculated using b-actin as normalizer (reference control).Data quality control was performed to check the reproducibility among 3 replicates and the number of missing (undetermined) values.Six genes (IL23A, IL12p35, TNF, IFGg, DEFB4, and CXCL10) profiled using Fluidigm RT-PCR panel plus 3 rerun genes (IL17A, IL17F, and IL22) were included in the Figure 7 plot because of their direct relevance to IL-17/IL-23 pathway The other rerun genes-S100A7A, IL23R, and EBI3-were not included in the plots owing to space limitations; however, their FCs are included in Table 8.

Serum biomarker analysis.
Welch's t-test was applied for each individual biomarker, in log 2 -transformed concentrations, between 118 patients with psoriasis at baseline (week 0) and 25 healthy controls randomly selected according to the age, sex, race, and body mass index distribution of the psoriasis population (Table 7) (samples collected from Bioreclamation) to assess for differences in baseline biomarker profiles between the study psoriasis population and healthy controls; an ANOVA was applied to evaluate for differences in baseline biomarker profiles among participants with psoriasis assigned to the 3 treatment arms (guselkumab, ADA, and placebo).PD effects in response to individual treatment were determined at weeks 4, 24, and 48 against baseline (week 0), using analysis of covariance linear mixed-effects models (lme) in R (https://cran.rproject.org/web/packages/emmeans/index.html) for each log 2 concentration by including treatment group, time, time by group interaction, and baseline log2 concentration as fixed effects and patient as a random effect.
Adjustment for multiple pairwise comparisons was implemented using BenjaminieHochberg FDR to account to account for multiple comparisons to baselines at different time points.The summary table of the P-values and the multiple comparisonecorrected FDR are attached in Table 8.

Skin transcriptomics analysis.
Gene expression measures in individual skin biopsies were determined at the probe set level (n ¼ 54,715) using the Robust Multi-array Average methodology.The quality of the entire dataset (183 skin samples) was assessed using a principal component analysis of whole-genome expression profiling, and no apparent outlier samples were detected.
Differential gene expression between psoriatic LS and NL at baseline (week 0) was evaluated using a paired t-test among 37 patients.P-values from paired t-tests were adjusted for multiple  A Blauvelt et al.
Guselkumab Reduces IL-23/IL-17 Biomarkers hypotheses using the BenjaminieHochberg FDR procedure.A FC >1.5 and FDR <0.05 were employed as criteria to determine the PSTR at baseline.For each probe set (or gene) in the PSTR, percentage of improvement in expression in LS in response to individual treatment at a given time point was calculated as the negative ratio of the average of the log 2 ratio of treatment (LS after treatment vs LS at baseline) over the average of the log 2 ratio of disease profiling (LS vs NL at baseline) among all patients in the given treatment arm.The molecular scar was defined for each treatment as the subset of PSTR genes with <75% improvement and remaining FC >1.5 in healed LS collected from patients when they achieved PASI100 response.Canonical pathways in the QIAGEN Ingenuity Pathway Analysis software (QIAGEN, https://digitalinsights.qiagen.com/IPA)enriched with PSTR genes were identified by a hypergeometric test (Kra ¨mer et al, 2014).A total of 56 pathways were selected with P < .01 and having >20% and !5 pathway genes as PSTR genes.For each of the 56 pathways, the percentage of improvement in expression in LS in response to individual treatments at a given time point was calculated as the average of the percentage of improvement among all PSTR genes in the pathway, with associated standard error.
Separately, the overall differential expression in 124 curated gene sets related to immune cell and skin inflammation between LS and NL at baseline was evaluated by GSVA.A total of 108 gene sets exhibited significant differences in GSVA scores between LS and NL at baseline (FDR < 0.05).Overall, 71 of these 108 gene sets were enriched with PSTR genes (FDR < 0.05, having >20% and !5 genes in each gene set as PSTR genes).For each of the 71 gene sets, percentage of improvement in expression in LS in response to individual treatment at a given time point was calculated as the average of percentage of improvement among all PSTR genes in the pathway, with associated standard error.

Clinical trial information
The primary results of this study have been published with the ClinicalTrials.govidentifier NCT02207231 and can be found in the abstract and Materials and Methods: ClinicalTrials.govidentifier NCT02207231 for the abstract and NCT02207231 for the Materials and Methods.Further identifying details have been added to the Material and Methods.Blood sample collection and analysis were performed at Janssen Research & Development, LLC.

A
Blauvelt et al.

Figure 5 .
Figure 5. Treatment effect on immune/skin inflammation gene sets by GSVA.(a) Volcano plot comparing GSVA scores of 124 gene sets between LS and NL at baseline.The 79 gene sets enriched with psoriasis DEG 1 between LS and NL at baseline are highlighted in red.Dashed horizontal line indicates an FDR ¼ 0.05.Gray dots represent the remaining gene sets not enriched by psoriasis DEG.(b) Heatmap of difference in GSVA scores in 108 gene sets (rows), corresponding to comparisons (columns). 1Gene set with !6% and >20% of genes differentially expressed (FC > 1.5) between LS and NL at baseline.ADA, adalimumab; DEG, differentially expressed gene; FC, fold change; FDR, false discovery rate; GSVA, gene set variation analysis; GUS, guselkumab; LS, lesional skin; NL, nonlesional skin; WK, week.

Figure 6 .
Figure 6.Normalization of disease-associated gene expression by GUS versus ADA in immune cell and skin inflammation gene sets.GSVA was used to evaluate differential expression of 124 gene sets between baseline LS and NL.Percentage improvement (mean þ SE) in the PSTR among the top 20 sets ranked by the difference in improvement between GUS and ADA at weeks (a) 4, (b) 24, and (c) 48 is shown.AH gene sets are from (Asifa)(Haider et al, 2008).AD, atopic dermatitis; ADA, adalimumab; Atheros, atherosclerosis; GSVA, gene set variation analysis; GUS, guselkumab; IPA, Ingenuity Pathway Analysis; KC, keratinocyte; LPS, lipopolysaccharide; LS, lesional skin; Mph, macrophage; NL, nonlesional skin; PSTR, psoriatic skin transcriptomic signature; RHE, epithelial cell; SE, standard error; Th, T helper; T reg , regulatory T cell.
1 qPCR: baseline LS versus NL differences and FCs by different treatment groups from all 121 qPCR genes assessed using Fluidigm RT-PCR panel are shown.2 Large FCs are capped at >1000 or < À1000 fold, similar to previous reports (reference: Sofen et al [2014]) 3 45_CALM1 and 75_CALM1 used two different qPCR assays for the same gene.A Blauvelt et al.Guselkumab Reduces IL-23/IL-17 Biomarkers

Table 1 .
Summary of Baseline Demographics and Clinical Disease Characteristics; Participants Randomized at Week 0 in the Serum Biomarker and Skin Biopsy Subpopulations

Table 4 .
Normalization of Disease-Associated Gene Expression by GUS Versus ADA in Immune Cell and Skin Inflammation Gene Sets (n [ 71) 1

Table 5 .
Full qPCR Results: Baseline LS Versus NL Differences and FC (vs Baseline) by Different Treatment Groups 1,2 A Blauvelt et al.

Table 5 .
Continued A Blauvelt et al.

Table 8 .
P-Values and Multiple Comparison Corrected FDR