Previous Article in Journal
Piperine as a Multifunctional Epigenetic Modulator: Integrative Molecular Insights into Cancer and Chronic Disease Therapy
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
IJMS
Volume 27
Issue 11
10.3390/ijms27115147
ijms-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Integrative Transcriptomic, Network, and Genomic Analysis of Peripheral Blood Mononuclear Cells Identifies Candidate Genes Associated with Dupilumab Clinical Response in Atopic Dermatitis Patients

by
Martina Krušič
1,
Mario Gorenjak
1,
Uroš Potočnik
1,2,3,*,† and
Maruška Marovt
4,5,*,†
1
Center for Human Molecular Genetics and Pharmacogenomics, Faculty of Medicine, University of Maribor, Taborska ulica 8, 2000 Maribor, Slovenia
2
Faculty of Chemistry and Chemical Engineering, University of Maribor, Smetanova ulica 17, 2000 Maribor, Slovenia
3
Department for Science and Research, University Clinical Centre Maribor, Ljubljanska ulica 5, 2000 Maribor, Slovenia
4
Faculty of Medicine, University of Maribor, Taborska ulica 8, 2000 Maribor, Slovenia
5
Department of Dermatovenerology, University Clinical Centre Maribor, Ljubljanska ulica 5, 2000 Maribor, Slovenia
*
Authors to whom correspondence should be addressed.
†
These authors contributed equally to this work.
Int. J. Mol. Sci. 2026, 27(11), 5147; https://doi.org/10.3390/ijms27115147 (registering DOI)
Submission received: 30 March 2026 / Revised: 1 June 2026 / Accepted: 4 June 2026 / Published: 5 June 2026
(This article belongs to the Section Molecular Pathology, Diagnostics, and Therapeutics)
Browse Figure
Versions Notes

Abstract

Atopic dermatitis (AD) is among the most common chronic inflammatory diseases. Due to the heterogeneous presentation of AD, patient response to treatment may differ considerably. Therefore, there is a pressing need for biomarkers associated with response to biological therapies. Thus, we aimed to identify blood-based candidate biomarkers associated with response in patients treated with dupilumab. The present study applied a multi-stage integrative analytical framework combining transcriptomic profiling, functional enrichment, co-expression network analysis, and genomic variant analysis to identify potential biomarkers. Eighteen dupilumab-naïve patients were enrolled in the transcriptomic analysis, with blood samples collected at baseline and after 16–18 weeks of therapy; five patients were identified as non-responders. Additionally, genotyping was performed in 34 patients. We identified a set of candidate genes (RPL18A, RPS28, FAU, MASTL, AURKA, TAF2, BUB1B, and RNF135) and genomic variants that may reflect underlying biological mechanisms influencing therapeutic response. However, given the limited sample size, these findings should be considered exploratory and hypothesis-generating. Finally, our study identified exploratory candidate genes potentially associated with variability in dupilumab treatment response. Moreover, our study represents an incremental contribution to existing knowledge, opening avenues for research that may ultimately lead to personalized medicine.

1. Introduction

Atopic dermatitis (AD) is the most common chronic inflammatory skin disease [1,2,3], affecting approximately 20% of children and 10% of adults worldwide [4]. Research findings indicate an increasing incidence and prevalence of the disease [5], with environmental factors also contributing significantly [6].
The disease is heterogeneous, which is reflected in a wide spectrum of clinical characteristics. Among the most common symptoms are eczematous lesions, severe itching, and dry skin [2,7]. Pruritus, one of the most characteristic symptoms, drives patients to scratch, which consequently leads to sleep disturbances and further exacerbation of the disease, thereby significantly affecting patients’ quality of life [8,9].
Due to the heterogeneous nature of AD, it is expected that patients will respond differently to treatment. A set of evaluated genetic biomarkers represents a pivotal tool for the development of personalized medicine [10,11]. In recent years, the number of biological therapies has increased substantially, leading to a growing need to identify reliable biological markers of treatment response.
Despite effective therapies, more than 55% of adult patients with moderate to severe AD report inadequate disease control [12,13]. Although many studies are not yet applicable in clinical practice, each contributes new insights and a deeper understanding of the pathogenesis of AD and opens opportunities for the identification of new therapeutic targets.
Through a comprehensive literature review, we observed a notable lack of studies investigating the blood transcriptome in patients with AD treated with dupilumab. In particular, there is a lack of studies that examine differences in blood transcriptome profiles between responders and non-responders to dupilumab treatment. In line with this, our study aimed to identify potential biomarkers of treatment response in the blood of patients with moderate-to-severe AD. By focusing on peripheral blood, an easily accessible and minimally invasive source, we sought to characterize gene expression differences between responders and non-responders and thereby contribute to the development of response-associated molecular tools for personalized treatment strategies. Because of the exploratory design and limited cohort size, the present study was intended primarily for candidate biomarker discovery and biological hypothesis generation. Importantly, all patients included in the transcriptomic analysis were naïve with respect to dupilumab therapy, allowing the identification of baseline molecular signatures that are not confounded by prior biologic treatment effects.

2. Results

2.1. RNA Sequencing Analysis

To detect gene expression signals potentially masked by treatment and ongoing inflammation, we conducted a three-stage RNA-seq analysis (Supplementary Figure S1).
First, we performed stage one differential gene expression analysis. In this stage, we compared responders’ follow-up samples with healthy controls to identify differentially expressed genes influenced by dupilumab therapy. This analysis revealed 4,526 differentially expressed genes with an adjusted p-value < 0.05. ENTREZ IDs of genes identified in stage one are presented in Supplementary Table S1.
We removed significant therapy-related genes identified in stage one before proceeding to stage two. Stage two included two substages aimed at identifying differentially expressed genes associated with inflammation. In the first substage, we compared non-responders with responders’ follow-up samples, whereas in the second substage, we compared non-responders’ follow-up samples with healthy controls. We identified 383 differentially expressed genes in substage one and 572 genes in substage two, all with p-value < 0.05. We combined the unique differentially expressed genes identified in both substages to define inflammation-related genes and removed them from the analysis before proceeding to stage three. This set included 929 unique genes (ENTREZ IDs are presented in Supplementary Table S2).
Prior to stage three, we removed previously identified inflammation-related genes. Thus, stage three was performed on the full gene set, excluding only the inflammation-associated genes identified in stage two, as the samples were obtained prior to therapy (naïve to dupilumab treatment). In stage three, we compared non-responders’ and responders’ baseline samples and revealed 817 significantly differentially expressed genes (p-value < 0.05). The identified genes represent a first step toward identifying candidate molecular signals associated with dupilumab response. Stage three differentially expressed genes are presented in Supplementary Table S3.

2.2. Gene Ontology Analysis

In addition, we carried out gene ontology analysis (GO) on statistically significant differentially expressed response-associated genes. Of the 817 genes, we excluded from the analysis those genes that did not have a symbol. In total, we included 791 statistically significant genes in the GO analyses, 338 upregulated and 453 downregulated. Furthermore, we performed GO analysis for molecular function and biological processes separately for up- and downregulated genes. In the enrichment analysis, the upregulated genes were associated with one GO term related to molecular function, which comprised 12 unique genes. No GO terms related to biological processes were identified for the upregulated genes. In contrast, the downregulated genes were associated with 56 GO terms related to molecular function, comprising 108 unique genes. Additionally, 56 GO terms related to biological processes were identified for the downregulated genes, including 133 unique genes. In total, we obtained 181 unique genes listed at the enriched GO terms. All identified GO terms are presented in the Supplementary Table S4.
Genes listed in the GO analysis were excluded from the set of differentially statistically significant genes to obtain the subset of genes not represented in any enriched GO term (non-GO-enriched genes). This resulted in 326 upregulated and 272 downregulated genes (Supplementary Tables S5 and S6).

2.3. Gene Scores and Association Testing

After GO analysis, we calculated gene score variables for each individual for both GO terms and non-GO-enriched gene terms to enable association testing with the outcome. Thereafter, we used bias-reduced logistic regression, where we fitted a model with dupilumab response as the dependent outcome variable and gene score adjusted for sex and age as independent variables. We performed regression models for each gene score variable separately.
For GO terms with upregulated genes, we identified one significant GO term (GO:0003735, p-value = 0.046) associated with the outcome. Moreover, gene scores derived from upregulated non-GO-enriched genes were also significantly associated with the outcome (p-value = 0.040). On the other hand, for GO terms with downregulated genes, we identified seven significant GO terms associated with the outcome (GO:0044839, p-value = 0.047; GO:0000086, p-value = 0.044; GO:1902749, p-value = 0.045; GO:0010389, p-value = 0.048; GO:1901990, p-value = 0.048; GO:0007093, p-value = 0.038; GO:1901991, p-value = 0.049), while no significant association was observed for non-GO-enriched genes.
Additionally, genes were extracted from GO terms whose gene scores were significantly associated with the outcome (p-value < 0.05). Finally, we obtained 36 unique genes from the enriched terms associated with the outcome (RPL23, MRPL11, RPL18A, RPL22L1, RPS12, RPS28, RPL27, FAU, MRPL28, MRPS12, RPS23, RPL21, BLM, FBXO5, MASTL, AURKA, CCNA2, AXIP1, MRE11, CDC25A, NDC80, TAF2, CDK1, LATS1, KIF14, BARD1, TAOK1, CDC14A, RBL1, ZW10, ECD, ANKRD17, BUB1B, CDC73, GIGYF2, and ATF2).

2.4. Weighted Gene Co-Expression Network Analysis

For non-GO-enriched genes, we performed weighted gene co-expression network analysis (WGCNA) on 326 significantly upregulated genes derived from the non-enriched gene set, showing a significant gene score association with the outcome. We applied the WGCNA algorithm to construct gene co-expression networks and selected a soft-thresholding power corresponding to a scale-free topology fit of R2 = 0.8. Next, we identified gene modules through average hierarchical clustering combined with dynamic tree cutting. We set the minimum module size to 20. In total, we obtained five distinct modules, comprising 14 (gray), 109 (turquoise), 104 (blue), 53 (brown), and 46 (yellow) genes (Supplementary Figure S2). The gray module 0 contained genes that were not assigned to any co-expression module.
Furthermore, we screened the modules for the promising hub genes. Therefore, we calculated gene expression profiles for each given module. For each module, we performed bias-reduced logistic regression to assess the association with the outcome, using the module gene score as the predictor and adjusting for sex and age. Regression models were fitted separately for each module, with the blue module 2 showing the strongest tendency of association with the outcome (p-value = 0.0519). All observed modules, the number of genes in each module, and p-values for outcome association of each module are listed in Table 1.
For module 2, module membership (kME) was calculated for each gene to assess its correlation with the module eigengene. Genes with kME > 0.9 were considered highly connected hub genes and were selected for further analyses. Based on this, we identified four hub genes, SLC2A6, HMOX1, RNF135, and ADAP1.

2.5. Genomic Profiling and Association Analysis

A total of 40 genes were identified and subsequently used as targets for genomic integration analysis. We performed genomic profiling on eighteen naïve patients and an additional sixteen long-term responders. First, a genome-wide association analysis was conducted in response to dupilumab therapy. Integration was performed by extracting variants located within ±100 kb of the 40 target genes. Statistically significant (p-value < 0.05) targeted genomic signals were found in the gene regions of RPL18A, RPS28, FAU, MASTL, AURKA, TAF2, BUB1B, ATF2, and RNF135. All identified variants were pruned at an R2 threshold of 0.5 to retain independent signals, resulting in a final set of 10 variants across 9 genes (Table 2). For all SNPs, regional Manhattan plots were created and are presented in the Supplementary Figures S3–S11.
Additionally, we retrieved eQTL information from publicly available data sources, the GTEx Portal and the eQTL Catalogue, to confirm gene–SNP interactions. One identified variant, rs590142, showed a significant eQTL association with gene expression in whole blood. For two SNPs in the ATF2 gene (rs2788522 and rs12613111), no eQTL data or proxy SNPs expressed in blood or blood cells were identified. Therefore, these variants were excluded from downstream analyses. For an additional seven SNPs, proxy variants (D’ > 0.7) showing eQTL associations with gene expression in blood or blood cells were identified. eQTL data obtained for selected variants is present in Table 3.
Based on the retained eQTL-supported variants, Mendelian randomization (MR) analysis using the inverse-variance weighted method was subsequently performed to further explore the association of genes and the outcome through variants as instrumental variables.
MR analysis suggested potential associations between genetically predicted gene expression and dupilumab treatment response for all analyzed genes (RPL18A, RPS28, FAU, MASTL, AURKA, TAF2, BUB1B, and RNF135), with all associations reaching statistical significance (p-value < 0.05). As only one blood-related eQTL instrument was available for each gene, MR was performed using single-SNP instruments and therefore provided only limited support for causality.

2.6. Exploratory External Dataset Comparison

To further explore the consistency of the identified transcriptional signals, we analyzed an independent dataset from the Gene Expression Omnibus (GEO), GSE208405, comprising 46 samples (six non-responders and 40 responders). Several investigated genes showed nominal expression differences between responder groups, with RPS28 demonstrating the strongest signal among the analyzed candidate genes (Table 4). Because clinically defined dupilumab response groups were not available, responder status was inferred indirectly from eosinophil-associated transcriptomic patterns. Therefore, this analysis should be interpreted as an exploratory external comparison rather than a formal validation.

2.7. RT-qPCR Assessment of Candidate Genes

To further assess expression patterns of the prioritized candidate genes, reverse transcription quantitative polymerase chain reaction (RT-qPCR) was performed. Relative gene expression (2−ΔΔCt) was calculated between the non-responders relative to the responders and is presented in Table 5 (Figure 1). Logistic regression was performed. None of the investigated genes reached statistical significance after logistic regression (Table 5). However, BUB1B showed suggestive significance (p-value = 0.099) in the logistic regression model. Importantly, BUB1B, AURKA, and RPS28 showed concordant expression direction compared with the RNA-seq analysis.

3. Discussion

To the best of our knowledge, few studies have applied a multi-stage integrative approach combining RNA-seq, GO analysis, co-expression network modeling, and genomic profiling to investigate potential biomarkers of dupilumab response in AD. By systematically removing treatment- and inflammation-related genes, we aimed to uncover underlying gene expression patterns potentially associated with therapeutic response. For this reason, we used blood samples, as they are more accessible and less invasive compared to tissue biopsies.
The strength of the present study is the three-stage analytical approach, which was designed to reduce therapy- and inflammation-related signals that could obscure baseline response-associated molecular patterns. To sum up, first, we identified genes influenced by therapy by comparing responders’ follow-up samples with healthy controls. Subsequently, we removed therapy-related genes to unmask inflammation-associated genes, enabling a more defined analysis of baseline sample differences between responders and non-responders. Thus, an important strength of our study is that all patients included in the transcriptomic analysis were naïve with respect to dupilumab treatment, ensuring that the identified gene expression patterns reflect intrinsic baseline biology associated with treatment response rather than transcriptional changes induced by prior therapy or inflammation. Together, we identified 817 significant differentially expressed genes representing candidate transcriptional signals potentially associated with therapeutic response.
Furthermore, to enrich identified genes, we performed a detailed analysis of the molecular functions and biological processes associated with these genes. GO analysis of the upregulated genes identified a single significantly enriched GO term, “structural constituent of ribosome.” In contrast, GO analysis of molecular functions for downregulated genes revealed 56 terms predominantly associated with protein kinase activity, chromatin organization and epigenetic regulation, and DNA-related ATP-dependent processes, with additional enrichment in ubiquitin-mediated protein modification and cytoskeletal functions. Moreover, GO analysis of downregulated genes revealed 56 terms associated with biological processes, mainly involving cell cycle, DNA replication, and chromatin regulation.
Moreover, we calculated gene scores for each GO term and the set of non-GO-enriched genes. Thus, allowing us to test the association of each enriched GO term or non-GO-enriched set with dupilumab response as the outcome. After fitting the regression models, we obtained 36 genes listed at GO terms and non-GO-enriched genes associated with the outcome. For non-GO-enriched genes, we performed WGCNA to further identify the association drivers. We obtained five modules, of which one module showed the tendency for significance threshold (module 2 blue, p-value = 0.0519). This module consists of 104 genes. Finally, we identified four hub genes based on module kME (SLC2A6, HMOX1, RNF135, and ADAP1).
In addition, to integrate the aforementioned findings with genomic regions, we extracted variants located within ±100 kb of the total 40 identified genes. Downstream genomic integration yielded a final set of eight prioritized candidate genes: RPL18A, RPS28, FAU, MASTL, AURKA, TAF2, BUB1B, and RNF135.
To the best of our knowledge, none of these SNPs has been previously associated with AD. Among identified genes, RPL18A, RPS28, and FAU are listed in the enriched GO term “structural constituent of ribosome”. Those genes predominantly encode ribosomal proteins involved in protein synthesis and cellular metabolism. The identification of SNPs in these genes in our AD cohort points to a potential involvement of translational machinery in disease pathogenesis. This is consistent with emerging evidence that dysregulation of ribosomal function may contribute to altered immune responses and inflammatory processes [14]. RPL18A (Ribosomal protein L18a) was previously identified as a gene capable of promoting B-cell transformation [15]. Furthermore, B cells have been implicated in the pathogenesis of atopic dermatitis, particularly through regulatory B cells (Bregs), which can modulate inflammatory responses via IL-10–mediated suppression of eosinophil activation and skin inflammation [16,17]. In addition, it was shown that dupilumab treatment increases transitional B cells in severe asthma [18]. Thus, the identification of RPL18A in our study may indicate that B-cell-related pathways contribute to variability in dupilumab treatment response.
RPS28 (Ribosomal protein S28) has also been recently identified as a hub gene in allergic rhinitis, an IgE-mediated inflammatory disease that shares immunological mechanisms with atopic dermatitis, suggesting that RPS28 may contribute to pathways involved in allergic inflammation [19].
FAU (FAU ubiquitin-like and ribosomal protein S30 fusion) encodes a fusion protein composed of the ubiquitin-like protein FUBI and ribosomal protein S30, a component of the 40S ribosomal subunit involved in protein synthesis (NCBI Gene: https://www.ncbi.nlm.nih.gov/gene/2197, accessed on 24 March 2026). To our knowledge, beyond this mechanistic role in ribosomal function, FAU has not previously been implicated in immune-mediated diseases, and our findings represent the first evidence linking this gene to dupilumab treatment response in AD.
TAF2 (TATA-box binding protein associated factor 2) encodes a subunit of the TFIID transcription initiation complex involved in RNA polymerase II–dependent gene transcription. Beyond this fundamental role in transcriptional regulation, its involvement in immune-mediated diseases has only been reported in the context of prognosis and treatment response in acute myeloid leukemia [20].
Most notably, the genes AURKA (Aurora kinase A) and BUB1B (BUB1 mitotic checkpoint serine/threonine kinase B), which were previously associated with AD [21], were also identified in the present study. AURKA belongs to the serine/threonine kinase family and is widely recognized in cancer research [22,23,24]. Moreover, research has shown that expression levels of AURKA are increased in psoriatic lesions compared to healthy controls [25]. Upregulation of AURKA in inflammatory skin conditions has been associated with enhanced activation of the AIM2 inflammasome and the AKT/mTOR pathway, leading to increased keratinocyte-mediated inflammation [25]. A recent study by Liu and colleagues [21], aimed to uncover molecular mechanisms and potential therapeutic targets in AD using bioinformatics and experimental validation. Their protein–protein interaction network analysis identified five hub genes (AURKA, BUB1B, CCNB1, MELK, and TTK) associated with AD progression. The analysis of immune cell infiltration revealed strong associations between these genes and several immune cell types, including activated CD4 T cells, Th2 cells, regulatory T cells (Tregs), and central memory CD8 T cells. They also showed that mRNA expression levels of AURKA and BUB1B are upregulated in AD lesions. Moreover, protein BUB1B promotes homologous recombination-mediated DNA damage repair in breast cancer cells [26]. It shows that BUB1B enhances DNA repair capacity by activating the PI3K/AKT signaling pathway, which supports cancer cell survival. The PI3K/AKT pathway plays a very important role in several inflammatory diseases, including AD [27,28]. Its role is well established in psoriasis [29]. Interestingly, it has been shown that treatment with baricitinib inhibits the activation of PI3K/AKT and mTOR signaling pathways [30]. Several studies have shown that targeting the PI3K/AKT/mTOR signaling pathway alleviates AD symptoms [31,32].
In addition, MASTL (Microtubule-associated serine/threonine kinase-like) has been shown to regulate the PI3K/AKT/mTOR signaling axis through the MASTL–PP2A/B55 kinase–phosphatase module, thereby modulating AKT activity and downstream metabolic and signaling processes [33]. Importantly, this kinase–phosphatase module links cell-cycle regulation with PI3K/AKT/mTOR signaling and contributes to feedback mechanisms that restrain AKT activity during sustained mTOR signaling [33]. Given the central role of the PI3K/AKT/mTOR pathway in inflammatory diseases, these observations further support the potential involvement of MASTL in pathways relevant to AD pathogenesis. Several genes identified in our study have been associated with the PI3K/AKT/mTOR signaling pathway, which has previously been proposed as a potential therapeutic target in AD [34]. Taken together, these observations support the involvement of AURKA, BUB1B, and MASTL in AD-related signaling pathways and highlight the relevance of our findings, identifying candidate genes potentially associated with dupilumab treatment response.
RNF135 was the only gene identified within the non-GO-enriched gene set. This is particularly interesting because it was not associated with any enriched GO terms, indicating that its potential involvement in AD and dupilumab response may represent a previously unrecognized biological mechanism.
RNF135 (Ring finger protein 135) plays an important role in innate immune signaling by promoting RIG-I–mediated interferon responses during viral infection [35,36,37,38]. This immune regulatory function suggests that RNF135 may influence inflammatory pathways, making its identification in our analysis particularly intriguing despite its absence from enriched GO terms. Given the central role of interferon and innate immune signaling in inflammatory diseases, our findings may point to a previously unrecognized mechanism contributing to variability in dupilumab treatment response.
Finally, exploratory external dataset comparison demonstrated partial overlap with previously identified differentially expressed genes, although directionality of expression was not fully concordant between datasets. Therefore, these findings should be interpreted as evidence that related immune-associated transcriptional programs involving overlapping genes may be captured across datasets, rather than as direct replication of individual gene-level effects. Importantly, RT-qPCR showed concordant expression direction for BUB1B, AURKA, and RPS28 compared with the RNA-seq analysis, providing additional supportive evidence for these candidate genes despite the limited statistical power of the cohort. Although statistically significant expression differences were not observed after logistic regression adjustment, BUB1B demonstrated a suggestive association with treatment response and remained consistently downregulated in non-responders across analytical approaches.
Nevertheless, our study has several limitations. First, the relatively small sample size may reduce statistical power and limit the generalizability of the findings. In particular, the limited number of non-responders increases the possibility of false-positive findings and reduces statistical stability across transcriptomic and genomic analyses. The limited sample size also affects the stability of WGCNA-derived modules; therefore, the WGCNA results should be interpreted as exploratory co-expression patterns rather than definitive network structures. Second, Mendelian randomization analyses should be interpreted primarily as exploratory prioritization approaches, preventing formal sensitivity analyses such as MR-Egger, weighted median estimation, heterogeneity testing, or pleiotropy assessment. Third, blood-based biomarkers may not fully capture local cutaneous pathophysiological processes in AD, potentially limiting biological interpretation. Finally, RT-qPCR validation was also performed on a limited cohort, reducing the statistical power and robustness of validation analyses. Consequently, all identified candidate biomarkers should be interpreted cautiously until validated in larger independent cohorts.
In conclusion, this study applied a multi-stage integrative analytical framework combining transcriptomic profiling, functional enrichment, co-expression network analysis, and genomic variant analysis to identify potential biomarkers associated with dupilumab treatment response in patients with AD. Through systematic filtering of therapy- and inflammation-related signals, we identified a set of candidate genes and genomic variants that may reflect underlying biological mechanisms influencing therapeutic response. Several of the identified genes are functionally linked to pathways involved in immune regulation, cellular proliferation, and PI3K/AKT/mTOR signaling, supporting their potential relevance in AD pathophysiology. Although the present findings are exploratory and limited by sample size, convergent signals across multiple analytical approaches consistently highlighted overlapping candidate genes. Because nominal p-value thresholds were used in several downstream analyses, the possibility of false-positive findings cannot be excluded. Therefore, the identified genes and pathways should primarily be considered hypothesis-generating candidate signals requiring validation in larger independent cohorts rather than clinically validated predictive biomarkers. Overall, our findings provide additional insight into molecular mechanisms potentially underlying variability in dupilumab response and may support future development of personalized therapeutic strategies in AD.

4. Materials and Methods

4.1. Ethical Approval

For the purposes of the study, ethical approval was obtained from the Ethics Committee of the Republic of Slovenia (ethical approval number: KME 0120-433/2023/3). The study was conducted in accordance with the principles of the Declaration of Helsinki. All participants received information about the course of the study prior to voluntarily signing the informed consent form to participate in the research.

4.2. Enrolled Subjects

All enrolled patients were of Caucasian Central European ethnicity. The study included patients who were treated at the Department of Dermatovenerology at the University Medical Center Maribor and had been diagnosed with AD for at least 12 months. To be classified as having moderate-to-severe AD, dupilumab-naïve patients had to meet at least one of the following criteria: EASI ≥ 7.1, vIGA-AD ≥ 3, BSA ≥ 10%, or a pruritus score of ≥4 on the numeric VAS scale. Patients with immunodeficiency and those who were concomitantly treated with other systemic therapies for AD were excluded from the study. Additionally, among dupilumab-naïve patients, those who had received systemic corticosteroids, methotrexate, mycophenolate, or other immunosuppressive agents within the previous 4 weeks were excluded.
In addition, nine healthy controls were also included in the present study. Healthy controls had been previously recruited at the Center for Human Molecular Genetics and Pharmacogenomics, Faculty of Medicine, University of Maribor.
For an RNA-seq experiment, we enrolled eighteen patients with moderate-to-severe AD, who were naïve to dupilumab treatment. Peripheral venous blood samples were collected at baseline and 16–18 weeks after initiation of dupilumab therapy. Out of the eighteen patients, five were classified as non-responders, defined as not achieving EASI75. Transcriptomic profiles of peripheral blood from nine healthy controls were used to identify differentially expressed genes associated with inflammation and treatment.
The genotyping analysis included 34 patients with AD, comprising 18 patients from the initial cohort (13 responders and 5 non-responders) and 16 additional patients classified as long-term good responders to dupilumab, defined as maintaining a good response for 1–3 years. Demographic data and clinical parameters for patients are presented in Table 6 and Table 7, respectively.

4.3. DNA and RNA Isolation

DNA and RNA were extracted from peripheral blood mononuclear cells using the Monarch® Spin gDNA Extraction Kit (New England Biolabs, Ipswich, MA, USA) and the innuPREP Micro RNA Kit (iST Innuscreen GmbH, Berlin, Germany), respectively. DNA and RNA with sufficient concentration and quality were further included in downstream analyses.

4.4. RNA Sequencing and Differential Gene Expression Analysis

RNA samples were sequenced at BGI facilities using BGI Optimal Dual-mode mRNA Library Prep Kit PE150 on the DNBSEQ G400 (BGI, Shenzhen, China). Raw data containing adapter sequences or low-quality reads were removed using the SOAPnuke tool by BGI [39].
A three-stage approach was employed for differential expression analysis, similar to that used in a previously published article by our group [40]. In stage one, differentially expressed genes influenced by dupilumab therapy were identified. These genes were excluded prior to the second stage. Stage two consisted of two substages aimed at identifying differentially expressed genes associated with inflammation. The union of unique differentially expressed genes identified in both substages was considered to represent inflammation-related genes and was subsequently removed from the analysis prior to the third stage. In the final stage, differentially expressed genes potentially associated with response to dupilumab were determined.
Blood transcriptome analysis for each stage was performed using a custom pipeline developed by the Center for Human Molecular Genetics and Pharmacogenomics in the environment R 4.5.2 (R Core Team, Vienna, Austria), employing various packages as described in previously published studies [41]. In stage one, genes with an adjusted p-value < 0.05 were considered differentially expressed. In all subsequent stages, a p-value < 0.05 was used as the significance threshold. These downstream analyses were intended as exploratory prioritization steps within the integrative framework rather than definitive transcriptome-wide inference. Therefore, nominal significance thresholds were used to avoid excessive signal loss in the context of a limited sample size.

4.5. Gene Ontology Analysis

Gene ontology (GO) analysis was performed in R 4.5.2 using the clusterProfiler 4.16.0 package [42,43]. Importantly, GO was conducted separately for significant up- and downregulated differentially expressed genes. Genes with log2FC > 0 represented upregulated genes, while those with log2FC < 0 indicated downregulated ones. Enrichment analysis was carried out for molecular function and biological processes, with significance thresholds set at p-value < 0.05 and q-value < 0.1. Genes listed at enriched GO terms were extracted and subsequently excluded from the set of significant differentially expressed genes to identify non-GO-enriched genes.

4.6. Gene Scores, Logistic Regression, and Weighted Gene Co-Expression Network Analysis

After GO analysis, gene scores for each GO term and for non-GO-enriched genes were calculated for each patient. To avoid collinearity in logistic regression, bias-reduced logistic regression was performed in R 4.5.2 using the brglm2 1.0.1 package [44]. The logistic regression model was adjusted for age and sex. Significant gene scores for GO terms and non-GO-enriched genes were obtained from the logistic regression results. Genes in significant GO terms were further used for targeted genomic profiling. In addition, significant gene scores (p-value < 0.05) for non-GO-enriched genes were further analyzed using weighted gene co-expression network analysis (WGCNA) in R 4.5.2 using the WGCNA 1.74 package [45]. Identified hub genes were defined with module membership (kME) greater than 0.9.

4.7. Genomic Profiling and Association Analysis

For 34 DNA samples, genotyping was performed. Genotyping was conducted with Illumina Infinium Global Screening Array V4.0 (Illumina Inc., San Diego, CA, USA) by the Institute of Clinical Molecular Biology in Kiel, Germany. Extensive quality control was carried out as previously described [46]. Genotype imputation was performed using a TOPMed imputation server and a TOPMed r3 panel with the Minimac4 imputation algorithm, an Rsq filter of >0.3, and Eagle 2.4 phasing [47,48,49]. Association analysis was performed between non-responders and responders using Wald logistic regression implemented in PLINK 2.0 software, and using genotypic dosages [50]. An association analysis was conducted using a regression model adjusted for age and sex. Targeted genomic profiling and integration with genomic data were performed by extracting variants located within ±100 kb of previously identified and enriched differentially expressed genes from RNA-seq and GO analyses, respectively. Variants were considered statistically significant at p-value < 0.05, prior to identifying independent variants through LD pruning using the SNPclip tool from LDlink software 6.0.0, with a threshold R2 of 0.5 [51]. Variants residing in selected genomic regions are presented in regional Manhattan plots constructed using LocusZoom [52]. Tissue expression quantitative trait loci (eQTLs) were estimated using the GTEx Portal Database (https://www.gtexportal.org/home/, accessed on 5 March 2026) or the eQTL Catalogue (https://www.ebi.ac.uk/eqtl/, accessed on 5 March 2026). When the gene or SNP of interest was not available in the GTEx portal or the eQTL Catalogue, proxy SNPs (D’ > 0.7) were identified using the LDproxy tool from the LDlink software 6.0.0 and examined for expression in blood or blood cells [51].
Additionally, two-sample Mendelian randomization (MR) analysis was conducted using the inverse-variance weighted (IVW) random-effects model implemented in the MendelianRandomization 0.10.0 R package [53]. MR analyses were performed for genes for which eQTL data in blood or blood cells were available. The corresponding SNPs, or their highest linkage disequilibrium (LD) proxies when eQTLs were reported, were used as instrumental variables. eQTL effect sizes with their corresponding standard errors were used as variant–exposure associations, while beta coefficients and their respective standard errors from association analyses were used as variant–outcome associations. The causal effect of gene expression on dupilumab treatment response was considered statistically significant at p-value < 0.05. Because only one blood-related eQTL instrument was available for each gene, causal inference from MR analyses remains limited and should be interpreted cautiously.

4.8. Exploratory External Dataset Comparison

Finally, an independent dataset was used for exploratory external comparison of the prioritized transcriptional signals [54]. The dataset was downloaded from the GEO database, specifically GSE208405 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE208405, accessed on 4 May 2026). Before the differential gene expression analysis, the samples were filtered, and only those from patients with AD whose blood samples had been taken prior to dupilumab therapy were selected. Together, 46 samples were included in the downstream analysis.
Differential gene expression analysis was conducted as described in the aforementioned paragraph. Absolute immune deconvolution estimates derived from publicly available RNA-seq data were used to identify eosinophil-enriched (“eosinophil-hot”) samples. Samples with detectable eosinophil infiltration were operationally classified as non-responders (N = 6) based on observations reported in the original study associated with the dataset [54].

4.9. RT-qPCR Assessment of Candidate Genes

Nucleotide sequences of the exons of the target genes were obtained from the NCBI Gene database (https://www.ncbi.nlm.nih.gov/gene/, accessed on 4 May 2026). Isoform non-specific primers were hand-picked and designed using Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/, accessed on 4 May 2026) [55] and Primer3 [56]. Primers for reference genes ACTB and B2M were obtained from a previous study [57]. Primer sequences are summarized in Table 8. All primers were synthesized by Sigma-Aldrich (Merck KGaA, Darmstadt, Germany). mRNA was obtained from 18 (5 non-responders and 13 responders) enrolled individuals as described in the aforementioned paragraph. A concentration of 50 ng/µL of mRNA was transcribed into cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied BiosystemsTM, Thermo Fisher Scientific, Waltham, MA, USA). Next, 2 µL of 10-fold diluted cDNA was used as a template. RT-qPCR gene expression assay was carried out using Lightcycler 480® SYBR Green I Master Mix and Lightcycler 480® real-time thermocycler (Roche, Basel, Switzerland) according to the manufacturer’s instructions. Melting curves for each sample were analyzed after each run in order to confirm the specificity of amplification. Raw Ct values were obtained from two run-independent technical replicates for each sample. Normalization of raw data was performed using geometric averaging of reference genes, and relative expression was calculated using the 2−ΔΔCt method [58]. Stability of reference genes was statistically assessed using the relative 2−ΔCt’calculation. In addition, logistic regression was conducted using IBM SPSS Statistics (Version 31, IBM Corp., Armonk, NY, USA). Analysis was adjusted for sex and age. Statistical significance was defined as a p-value < 0.05.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ijms27115147/s1.

Author Contributions

Conceptualization, M.M.; methodology, M.K., M.M. and M.G.; software, M.K., M.M. and M.G.; validation, M.M., M.K. and M.G.; formal analysis, M.M., M.K. and M.G.; investigation, M.K., M.G. and M.M.; resources, M.M.; data curation, M.K., M.G. and M.M.; writing—original draft preparation, M.K., M.M. and M.G.; writing—review and editing, M.G., M.M., M.K. and U.P.; visualization, M.K., M.M. and M.G.; supervision, M.M. and U.P.; project administration, M.M. and U.P.; funding acquisition, M.M. and U.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Slovenian Research and Innovation Agency research core funding P3-0427 and No. I0-0029, by University Clinical Center Maribor, internal grant no. IRP-2021/02-18 and by the Republic of Slovenia, the Ministry of Higher Education, Science and Innovation, and the European Union from the European Regional Development Fund, grant RIUM.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Republic of Slovenia (protocol code KME 0120-433/2023/3). The present study was evaluated and approved on 14 November 2023.

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The original data presented in the study are openly available in the Gene Expression Omnibus (GEO) repository at GSE326320.

Acknowledgments

The authors would like to thank the patients for participating.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
ADAtopic dermatitis
AUCArea under the curve
BSABody surface area
EASIEczema Area and Severity Index
eQTLTissue expression quantitative trait loci
GOGene ontology
MRMendelian randomization
ROCReceiver operating characteristic
VASVisual analog scale
vIGA-ADValidated Investigator Global Assessment for Atopic Dermatitis
WGCNAWeighted Gene Co-Expression Network Analysis

References

  1. Weidinger, S.; Novak, N. Atopic dermatitis. Lancet 2016, 387, 1109–1122. [Google Scholar] [CrossRef]
  2. Afshari, M.; Kolackova, M.; Rosecka, M.; Čelakovská, J.; Krejsek, J. Unraveling the skin; a comprehensive review of atopic dermatitis, current understanding, and approaches. Front. Immunol. 2024, 15, 1361005. [Google Scholar] [CrossRef]
  3. Meledathu, S.; Naidu, M.P.; Brunner, P.M. Update on atopic dermatitis. J. Allergy Clin. Immunol. 2025, 155, 1124–1132. [Google Scholar] [CrossRef]
  4. Carrascosa-Carrillo, J.M.; Aterido, A.; Li, T.; Guillén, Y.; Martinez, S.; Marsal, S.; Julià, A. Toward Precision Medicine in Atopic Dermatitis Using Molecular-Based Approaches. Actas Dermosifiliogr. 2024, 115, 66–75. [Google Scholar] [CrossRef]
  5. Asher, M.I.; Montefort, S.; Björkstén, B.; Lai, C.K.; Strachan, D.P.; Weiland, S.K.; Williams, H. Worldwide time trends in the prevalence of symptoms of asthma, allergic rhinoconjunctivitis, and eczema in childhood: ISAAC Phases One and Three repeat multicountry cross-sectional surveys. Lancet 2006, 368, 733–743. [Google Scholar] [CrossRef]
  6. Bonamonte, D.; Filoni, A.; Vestita, M.; Romita, P.; Foti, C.; Angelini, G. The Role of the Environmental Risk Factors in the Pathogenesis and Clinical Outcome of Atopic Dermatitis. BioMed Res. Int. 2019, 2019, 2450605. [Google Scholar] [CrossRef]
  7. Langan, S.M.; Irvine, A.D.; Weidinger, S. Atopic dermatitis. Lancet 2020, 396, 345–360. [Google Scholar] [CrossRef]
  8. Chrostowska-Plak, D.; Reich, A.; Szepietowski, J.C. Relationship between itch and psychological status of patients with atopic dermatitis. J. Eur. Acad. Dermatol. Venereol. 2013, 27, e239–e242. [Google Scholar] [CrossRef]
  9. Silverberg, J.I.; Gelfand, J.M.; Margolis, D.J.; Boguniewicz, M.; Fonacier, L.; Grayson, M.H.; Simpson, E.L.; Ong, P.Y.; Chiesa Fuxench, Z.C. Patient burden and quality of life in atopic dermatitis in US adults: A population-based cross-sectional study. Ann. Allergy Asthma Immunol. 2018, 121, 340–347. [Google Scholar] [CrossRef]
  10. Mastraftsi, S.; Vrioni, G.; Bakakis, M.; Nicolaidou, E.; Rigopoulos, D.; Stratigos, A.J.; Gregoriou, S. Atopic Dermatitis: Striving for Reliable Biomarkers. J. Clin. Med. 2022, 11, 4639. [Google Scholar] [CrossRef]
  11. Bakker, D.; de Bruin-Weller, M.; Drylewicz, J.; van Wijk, F.; Thijs, J. Biomarkers in atopic dermatitis. J. Allergy Clin. Immunol. 2023, 151, 1163–1168. [Google Scholar] [CrossRef]
  12. Simpson, E.L.; Guttman-Yassky, E.; Margolis, D.J.; Feldman, S.R.; Qureshi, A.; Hata, T.; Mastey, V.; Wei, W.; Eckert, L.; Chao, J.; et al. Association of Inadequately Controlled Disease and Disease Severity With Patient-Reported Disease Burden in Adults With Atopic Dermatitis. JAMA Dermatol. 2018, 154, 903–912. [Google Scholar] [CrossRef]
  13. Simpson, E.L.; Bieber, T.; Guttman-Yassky, E.; Beck, L.A.; Blauvelt, A.; Cork, M.J.; Silverberg, J.I.; Deleuran, M.; Kataoka, Y.; Lacour, J.P.; et al. Two Phase 3 Trials of Dupilumab versus Placebo in Atopic Dermatitis. N. Engl. J. Med. 2016, 375, 2335–2348. [Google Scholar] [CrossRef]
  14. Das, A.S.; Basu, A.; Mukhopadhyay, R. Ribosomal proteins: The missing piece in the inflammation puzzle? Mol. Cell Biochem. 2025, 480, 785–797. [Google Scholar] [CrossRef]
  15. Wolf, I.; Bouquet, C.; Melchers, F. cDNA-library testing identifies transforming genes cooperating with c-myc in mouse pre-B cells. Eur. J. Immunol. 2016, 46, 2555–2565. [Google Scholar] [CrossRef]
  16. Kader, H.A.; Sabih Ur Rehman, S.; Saraswathiamma, D.; Shiek, S.S.; Bencomo-Hernández, A.A.; Moorakkan, U.R.; Haider, M.T.; Munawar, N.; Iratni, R.; Serfling, E.; et al. NFATc1 deficiency in B cells ameliorates atopic dermatitis. Sci. Rep. 2025, 15, 25170. [Google Scholar] [CrossRef]
  17. Lee, D.; Jo, M.G.; Min, K.Y.; Choi, M.Y.; Kim, Y.M.; Kim, H.S.; Choi, W.S. IL-10(+) regulatory B cells mitigate atopic dermatitis by suppressing eosinophil activation. Sci. Rep. 2024, 14, 18164. [Google Scholar] [CrossRef]
  18. Lommatzsch, M.; Dost, M.; Jaishankar, N.; Weise, M.; Stoll, P.; Virchow, J.C.; Bratke, K. Dupilumab treatment increases transitional B cells in severe asthma. Allergy 2023, 78, 2055–2057. [Google Scholar] [CrossRef]
  19. Li, L.; Zhang, Y.; Wang, J.; Zhang, L.; Gao, Y.; Ji, X.; Wang, T.; Zhao, F. GZMA silencing inhibits JAK2/STAT1 pathway and improves allergic rhinitis. Gen. Physiol. Biophys. 2025, 44, 93–106. [Google Scholar] [CrossRef]
  20. Shi, H.Z.; Wang, M.W.; Huang, Y.S.; Liu, Z.; Li, L.; Wan, L.P. A telomere-related gene risk model for predicting prognosis and treatment response in acute myeloid leukemia. Heliyon 2024, 10, e31705. [Google Scholar] [CrossRef]
  21. Liu, X.; Lu, X.; Wang, L.; Du, Q.; Li, X.; Li, Y.; Ding, X.; Lai, Y.; Chen, X. Bioinformatics analysis of immune infiltration and cell cycle hub genes in atopic dermatitis. Biochem. Biophys. Rep. 2025, 44, 102253. [Google Scholar] [CrossRef]
  22. Shi, S.; Qiu, Y.; Jin, Z.; Zhou, J.; Yu, W.; Zhang, H. AURKA Identified as Potential Lung Cancer Marker through Comprehensive Bioinformatic Analysis and Experimental Verification. Crit. Rev. Eukaryot. Gene Expr. 2023, 33, 39–59. [Google Scholar] [CrossRef]
  23. Farid, A.A.; Afify, N.A.; Alsharnoby, A.A.; Abdelsameea, E.; Bedair, H.M. Predictive Role of AURKA rs 1047972 Gene Polymorphism and the Risk of Development of Hepatocellular Carcinoma. Immunol. Invest. 2022, 51, 1211–1221. [Google Scholar] [CrossRef]
  24. Taylor, N.J.; Bensen, J.T.; Poole, C.; Troester, M.A.; Gammon, M.D.; Luo, J.; Millikan, R.C.; Olshan, A.F. Genetic variation in cell cycle regulatory gene AURKA and association with intrinsic breast cancer subtype. Mol. Carcinog. 2015, 54, 1668–1677. [Google Scholar] [CrossRef]
  25. Tang, H.; Tang, X.; Guo, Z.; Cheng, H.; Zheng, X.; Chen, G.; Huang, H.; Wang, W.; Gao, J.; Sheng, Y.; et al. AURKA facilitates the psoriasis-related inflammation by impeding autophagy-mediated AIM2 inflammasome suppression. Immunol. Lett. 2021, 240, 98–105. [Google Scholar] [CrossRef]
  26. Luo, X.; Wei, Y.; Lin, H.; Xiao, N.; Zhao, W. BUB1B promotes homologous recombination-mediated DNA damage repair in breast cancer cells through the PI3K/AKT signaling pathway. Oncol. Rep. 2025, 54, 147. [Google Scholar] [CrossRef]
  27. Kumar, S.; Sharma, R.; Komal, K.; Kumar, D.; Ghosh, R.; Kumar, M. Unlocking the molecular pathway of atopic dermatitis: Journey so far and roads ahead. Inflammopharmacology 2025, 33, 5203–5234. [Google Scholar] [CrossRef]
  28. Huang, M.; Dong, L.; He, Y.; Cai, X.; Lu, Y.; Xu, J.; Xiao, F. YTHDF1 regulates the proliferation and differentiation of keratinocytes via the PI3K/AKT signaling pathway in atopic dermatitis. Allergol. Int. 2025, 74, 591–604. [Google Scholar] [CrossRef]
  29. Huang, T.; Lin, X.; Meng, X.; Lin, M. Phosphoinositide-3 kinase/protein kinase-B/mammalian target of rapamycin pathway in psoriasis pathogenesis. A potential therapeutic target? Acta Derm. Venereol. 2014, 94, 371–379. [Google Scholar] [CrossRef]
  30. Chen, S.; Li, C.; Tu, Z.; Cai, T.; Zhang, X.; Wang, L.; Tian, R.; Huang, J.; Gong, Y.; Yang, X.; et al. Off-label use of Baricitinib improves moderate and severe atopic dermatitis in China through inhibiting MAPK and PI3K/Akt/mTOR pathway via targeting JAK-STAT signaling of CD4(+) cells. Front. Pharmacol. 2024, 15, 1324892. [Google Scholar] [CrossRef]
  31. Yang, F.; Tanaka, M.; Wataya-Kaneda, M.; Yang, L.; Nakamura, A.; Matsumoto, S.; Attia, M.; Murota, H.; Katayama, I. Topical application of rapamycin ointment ameliorates Dermatophagoides farina body extract-induced atopic dermatitis in NC/Nga mice. Exp. Dermatol. 2014, 23, 568–572. [Google Scholar] [CrossRef]
  32. Bae, M.J.; Lim, S.; Lee, D.S.; Ko, K.R.; Lee, W.; Kim, S. Water soluble extracts from Actinidia arguta, PG102, attenuates house dust mite-induced murine atopic dermatitis by inhibiting the mTOR pathway with Treg generation. J. Ethnopharmacol. 2016, 193, 96–106. [Google Scholar] [CrossRef]
  33. Sanz-Castillo, B.; Hurtado, B.; Vara-Ciruelos, D.; El Bakkali, A.; Hermida, D.; Salvador-Barbero, B.; Martínez-Alonso, D.; González-Martínez, J.; Santiveri, C.; Campos-Olivas, R.; et al. The MASTL/PP2A cell cycle kinase-phosphatase module restrains PI3K-Akt activity in an mTORC1-dependent manner. Embo J. 2023, 42, e110833. [Google Scholar] [CrossRef]
  34. Roy, T.; Boateng, S.T.; Uddin, M.B.; Banang-Mbeumi, S.; Yadav, R.K.; Bock, C.R.; Folahan, J.T.; Siwe-Noundou, X.; Walker, A.L.; King, J.A.; et al. The PI3K-Akt-mTOR and Associated Signaling Pathways as Molecular Drivers of Immune-Mediated Inflammatory Skin Diseases: Update on Therapeutic Strategy Using Natural and Synthetic Compounds. Cells 2023, 12, 1671. [Google Scholar] [CrossRef]
  35. Sid, H.; Schusser, B. Advancing immunity and disease resistance in chickens through genome editing. J. Anim. Sci. Biotechnol. 2026, 17, 33. [Google Scholar] [CrossRef]
  36. Oshiumi, H.; Matsumoto, M.; Hatakeyama, S.; Seya, T. Riplet/RNF135, a RING finger protein, ubiquitinates RIG-I to promote interferon-beta induction during the early phase of viral infection. J. Biol. Chem. 2009, 284, 807–817. [Google Scholar] [CrossRef]
  37. Wallach, D.; Kovalenko, A. Phosphorylation and dephosphorylation of the RIG-I-like receptors: A safety latch on a fateful pathway. Immunity 2013, 38, 402–403. [Google Scholar] [CrossRef]
  38. Oshiumi, H.; Miyashita, M.; Inoue, N.; Okabe, M.; Matsumoto, M.; Seya, T. The ubiquitin ligase Riplet is essential for RIG-I-dependent innate immune responses to RNA virus infection. Cell Host Microbe 2010, 8, 496–509. [Google Scholar] [CrossRef]
  39. Chen, Y.; Chen, Y.; Shi, C.; Huang, Z.; Zhang, Y.; Li, S.; Li, Y.; Ye, J.; Yu, C.; Li, Z.; et al. SOAPnuke: A MapReduce acceleration-supported software for integrated quality control and preprocessing of high-throughput sequencing data. Gigascience 2018, 7, gix120. [Google Scholar] [CrossRef]
  40. Gorenjak, M.; Gole, B.; Goričan, L.; Jezernik, G.; Prosenc Zmrzljak, U.; Pernat, C.; Skok, P.; Potočnik, U. Single-Cell Transcriptomic and Targeted Genomic Profiling Adjusted for Inflammation and Therapy Bias Reveal CRTAM and PLCB1 as Novel Hub Genes for Anti-Tumor Necrosis Factor Alpha Therapy Response in Crohn’s Disease. Pharmaceutics 2024, 16, 835. [Google Scholar] [CrossRef]
  41. Gorenjak, M.; Zupin, M.; Jezernik, G.; Skok, P.; Potočnik, U. Omics data integration identifies ELOVL7 and MMD gene regions as novel loci for adalimumab response in patients with Crohn’s disease. Sci. Rep. 2021, 11, 5449. [Google Scholar] [CrossRef]
  42. Yu, G.; Wang, L.G.; Han, Y.; He, Q.Y. clusterProfiler: An R package for comparing biological themes among gene clusters. Omics 2012, 16, 284–287. [Google Scholar] [CrossRef] [PubMed]
  43. Wu, T.; Hu, E.; Xu, S.; Chen, M.; Guo, P.; Dai, Z.; Feng, T.; Zhou, L.; Tang, W.; Zhan, L.; et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation 2021, 2, 100141. [Google Scholar] [CrossRef]
  44. Kosmidis, I. R Package, Version 1.1.0; Brglm2: Bias Reduction in Generalized Linear Models; CRAN: Vienna, Austria, 2023.
  45. Langfelder, P.; Horvath, S. WGCNA: An R package for weighted correlation network analysis. BMC Bioinform. 2008, 9, 559. [Google Scholar] [CrossRef] [PubMed]
  46. Anderson, C.A.; Pettersson, F.H.; Clarke, G.M.; Cardon, L.R.; Morris, A.P.; Zondervan, K.T. Data quality control in genetic case-control association studies. Nat. Protoc. 2010, 5, 1564–1573. [Google Scholar] [CrossRef]
  47. Taliun, D.; Harris, D.N.; Kessler, M.D.; Carlson, J.; Szpiech, Z.A.; Torres, R.; Taliun, S.A.G.; Corvelo, A.; Gogarten, S.M.; Kang, H.M.; et al. Sequencing of 53,831 diverse genomes from the NHLBI TOPMed Program. Nature 2021, 590, 290–299. [Google Scholar] [CrossRef]
  48. Das, S.; Forer, L.; Schönherr, S.; Sidore, C.; Locke, A.E.; Kwong, A.; Vrieze, S.I.; Chew, E.Y.; Levy, S.; McGue, M.; et al. Next-generation genotype imputation service and methods. Nat. Genet. 2016, 48, 1284–1287. [Google Scholar] [CrossRef]
  49. Fuchsberger, C.; Abecasis, G.R.; Hinds, D.A. minimac2: Faster genotype imputation. Bioinformatics 2015, 31, 782–784. [Google Scholar] [CrossRef]
  50. Chang, C.C.; Chow, C.C.; Tellier, L.C.; Vattikuti, S.; Purcell, S.M.; Lee, J.J. Second-generation PLINK: Rising to the challenge of larger and richer datasets. Gigascience 2015, 4, 7. [Google Scholar] [CrossRef]
  51. Machiela, M.J.; Chanock, S.J. LDlink: A web-based application for exploring population-specific haplotype structure and linking correlated alleles of possible functional variants. Bioinformatics 2015, 31, 3555–3557. [Google Scholar] [CrossRef]
  52. Pruim, R.J.; Welch, R.P.; Sanna, S.; Teslovich, T.M.; Chines, P.S.; Gliedt, T.P.; Boehnke, M.; Abecasis, G.R.; Willer, C.J. LocusZoom: Regional visualization of genome-wide association scan results. Bioinformatics 2010, 26, 2336–2337. [Google Scholar] [CrossRef]
  53. Yavorska, O.O.; Burgess, S. MendelianRandomization: An R package for performing Mendelian randomization analyses using summarized data. Int. J. Epidemiol. 2017, 46, 1734–1739. [Google Scholar] [CrossRef] [PubMed]
  54. Möbus, L.; Rodriguez, E.; Harder, I.; Boraczynski, N.; Szymczak, S.; Hübenthal, M.; Stölzl, D.; Gerdes, S.; Kleinheinz, A.; Abraham, S.; et al. Blood transcriptome profiling identifies 2 candidate endotypes of atopic dermatitis. J. Allergy Clin. Immunol. 2022, 150, 385–395. [Google Scholar] [CrossRef] [PubMed]
  55. Ye, J.; Coulouris, G.; Zaretskaya, I.; Cutcutache, I.; Rozen, S.; Madden, T.L. Primer-BLAST: A tool to design target-specific primers for polymerase chain reaction. BMC Bioinform. 2012, 13, 134. [Google Scholar] [CrossRef] [PubMed]
  56. Untergasser, A.; Cutcutache, I.; Koressaar, T.; Ye, J.; Faircloth, B.C.; Remm, M.; Rozen, S.G. Primer3—New capabilities and interfaces. Nucleic Acids Res. 2012, 40, e115. [Google Scholar] [CrossRef]
  57. Gorenjak, M.; Repnik, K.; Jezernik, G.; Jurgec, S.; Skok, P.; Potočnik, U. Genetic prediction profile for adalimumab response in Slovenian Crohn’s disease patients. Z. Gastroenterol. 2019, 57, 1218–1225. [Google Scholar] [CrossRef]
  58. Livak, K.J.; Schmittgen, T.D. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 2001, 25, 402–408. [Google Scholar] [CrossRef]
Ijms 27 05147 g001
Figure 1. Violin plots of 2−ΔCt expression for selected genes.
Figure 1. Violin plots of 2−ΔCt expression for selected genes.
Ijms 27 05147 g001
Table 1. WGCNA results.
Table 1. WGCNA results.
ModuleModule ColorNumber of Genes in the Modulep-Value for the Module
0gray140.0584
1turquoise1090.0734
2blue1040.0519
3brown530.0657
4yellow460.0551
Table 2. Identified variants.
Table 2. Identified variants.
ChrBase Pair (GRCh38)dbSNP IDA1Genep-Value
1917,831,128rs3212777GRPL18A0.0339
198,315,265rs8110376CRPS280.0454
1165,157,627rs75018393TFAU0.0432
1027,250,013rs590142CMASTL0.0409
2056,295,713rs6069696AAURKA0.0322
8119,695,360rs9297602GTAF20.0265
1540,157,612rs55940969TBUB1B0.0454
2175,160,299rs2788522TATF20.0219
2175,037,495rs12613111CATF20.0416
1730,925,135rs11868576TRNF1350.0421
Chr: chromosome; A1: effect allele.
Table 3. eQTL of selected variants.
Table 3. eQTL of selected variants.
dbSNP ID (Gene)Proxy SNPD’Tissuep-Value
rs3212777 (RPL18A)rs1119228430.7083blood †6.60 × 10−13
rs8110376 (RPS28)rs29139441whole blood ‖1.70 × 10−8
rs75018393 (FAU)rs29578791whole blood ‖6.00 × 10−5
rs590142 (MASTL)//whole blood ‖5.20 × 10−14
rs6069696 (AURKA)rs22980160.9201macrophage: IFNg 18 h †6.88 × 10−10
rs9297602 (TAF2)rs168929381whole blood ‖6.40 × 10−6
rs55940969 (BUB1B)rs620200251whole blood ‖5.00 × 10−5
rs11868576 (RNF135)rs65052190.9047whole blood ‖8.30 × 10−5
‖ Data obtained from the GTEx portal; † Data obtained from the eQTL Catalogue.
Table 4. Results of the independent dataset.
Table 4. Results of the independent dataset.
Genelog2FCCI ‖p-Value
RPL18A−0.4809−1.0318 to 0.07000.0855
RPS28−0.8069−1.5704 to −0.04330.0388
FAU−0.4038−0.8955 to 0.08790.1051
MASTL0.1165−0.1939 to 0.42700.4538
AURKA0.2827−0.1012 to 0.66660.1451
TAF20.3597−0.0347 to 0.75420.0729
BUB1B0.4093−0.1405 to 0.95910.1408
RNF135−0.0904−0.3090 to 0.12810.4091
‖ 95% confidence interval.
Table 5. Relative gene expression between non-responders and responders.
Table 5. Relative gene expression between non-responders and responders.
Gene2−ΔΔCtMinimal ErrorMaximal Errorp-Value ‖
RPL18A1.1490.9631.3720.139
RPS281.5260.7782.9950.299
FAU1.0500.8851.2450.562
MASTL0.9670.6901.3560.406
AURKA0.6550.5290.8120.386
TAF21.1280.6302.0200.803
BUB1B0.7240.5031.0420.099
RNF1351.1790.9501.4630.326
‖ p-value from logistic regression analysis.
Table 6. Demographic data for patients included in the study.
Table 6. Demographic data for patients included in the study.
Naïve Patients
(N = 18)		Long-Term Good Responders
(N = 16)		Healthy Controls
(N = 9)
Mean age at time of sampling (SD)27.44 (9.20)31.88 (13.92)43.11 (6.58)
Sex, N (%)
Female6 (33.3%)7 (43.8%)4 (44.4%)
Male12 (66.7%)9 (56.3%)5 (55.6%)
Treatment outcome, N (%)
Response13 (72.2%)
Non-response5 (27.8%)
SD: Standard deviation.
Table 7. Clinical parameters for dupilumab-naïve patients.
Table 7. Clinical parameters for dupilumab-naïve patients.
Naïve Patients (N = 18)
Baseline ParametersFollow-Up Parametersp-Value *
BSA [%]
Mean (SD)30.39 (25.20)3.28 (4.06)<0.001
Median201.50
Minimum50
Maximum9012
vIGA-AD
Mean (SD)2.81 (0.54)0.69 (0.70)<0.001
Median31
Minimum20
Maximum42
VAS pruritus
Mean (SD)7.44 (1.88)2.56 (2.20)<0.001
Median7.502.50
Minimum30
Maximum107
VAS sleep disturbance
Mean (SD)5.67 (3.42)0.72 (1.67)0.002
Median80
Minimum00
Maximum106
EASI
Mean (SD)16.96 (9.09)1.60 (2.13)<0.001
Median15.850.55
Minimum5.100
Maximum39.505.70
* Non-parametric Wilcoxon signed-rank test; BSA: body mass index; vIGA-AD: validated investigator global assessment scale for atopic dermatitis; VAS: visual analog scale; EASI: eczema area and severity index.
Table 8. Primer sequences of target and reference genes.
Table 8. Primer sequences of target and reference genes.
GeneForward Primer (5′–3′)Reverse Primer (5′–3′)
TAF2ACTTCTTCCGAGTTACAGGCATGGTGGGTTCTTAGTCAACATG
MASTLTGCATCCAATAACTCAGAACCAAATTCGCCCATCATCAACGG
FAUCATTGCCCCGGAAGATCAAGTAGCCCGACCTGTCTTCTTC
BUB1BGATCATGTCCACGCTTCAGGACTCTCCTTCCCACCTTGAG
RNF135CAACGAACTGAGCATCCTGGATGGGCATGAGGAGGAAGAC
RPS28CGTGTGCAGCCTATCAAGCCTCGCTCTGACTCCAAAAGG
RPL18AGGAGATTGTCTACTGTGGGCGCGATCTCCTCCACCTTCAT
AURKAAATTCTTCCCAGCGCATTCCAAGTCTTCCAAAGCCCACTG
B2M †TTCTGGCCTGGAGGCTATCTCAGGAAATTTGACTTTCCATTC
ACTB †CATCGAGCACGGCATCGTCATAGCACAGCCTGGATAGCAAC
† Reference genes.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Krušič, M.; Gorenjak, M.; Potočnik, U.; Marovt, M. Integrative Transcriptomic, Network, and Genomic Analysis of Peripheral Blood Mononuclear Cells Identifies Candidate Genes Associated with Dupilumab Clinical Response in Atopic Dermatitis Patients. Int. J. Mol. Sci. 2026, 27, 5147. https://doi.org/10.3390/ijms27115147

AMA Style

Krušič M, Gorenjak M, Potočnik U, Marovt M. Integrative Transcriptomic, Network, and Genomic Analysis of Peripheral Blood Mononuclear Cells Identifies Candidate Genes Associated with Dupilumab Clinical Response in Atopic Dermatitis Patients. International Journal of Molecular Sciences. 2026; 27(11):5147. https://doi.org/10.3390/ijms27115147

Chicago/Turabian Style

Krušič, Martina, Mario Gorenjak, Uroš Potočnik, and Maruška Marovt. 2026. "Integrative Transcriptomic, Network, and Genomic Analysis of Peripheral Blood Mononuclear Cells Identifies Candidate Genes Associated with Dupilumab Clinical Response in Atopic Dermatitis Patients" International Journal of Molecular Sciences 27, no. 11: 5147. https://doi.org/10.3390/ijms27115147

APA Style

Krušič, M., Gorenjak, M., Potočnik, U., & Marovt, M. (2026). Integrative Transcriptomic, Network, and Genomic Analysis of Peripheral Blood Mononuclear Cells Identifies Candidate Genes Associated with Dupilumab Clinical Response in Atopic Dermatitis Patients. International Journal of Molecular Sciences, 27(11), 5147. https://doi.org/10.3390/ijms27115147

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop