Candidate Gene Association Studies in Atopic Dermatitis in Participants of European and Asian Ancestry: A Systematic Review and Meta-Analysis

Atopic dermatitis (AD) has been extensively investigated for genetic associations utilizing both candidate gene approaches and genome-wide scans. Here, we comprehensively evaluated the available literature to determine the association of candidate genes in AD to gain additional insight into the etiopathogenesis of the disease. We systematically screened all studies that explored the association between polymorphisms and AD risks in cases of European and Asian ancestry and synthesized the available evidence through a random-effects meta-analysis. We identified 99 studies that met our inclusion/exclusion criteria that examined 17 candidate loci in Europeans and 14 candidate genes in Asians. We confirmed the significant associations between FLG variants in both European and Asian populations and AD risk, while synthesis of the available data revealed novel loci mapped to IL18 and TGFB1 genes in Europeans and IL12RB1 and MIF in Asians that have not yet been identified by genome-wide association studies. Our findings provide comprehensive evidence for AD risk loci in cases of both European and Asian ancestries, validating previous associations as well as revealing novel loci that could imply previously unexplored biological pathways.


Introduction
Atopic dermatitis (AD) is a common, chronic inflammatory cutaneous disease characterized by the development of recurrent eczematous lesions and intense itching [1]. AD affects about 20% of the worldwide infant population, displaying a lower prevalence during adulthood, incorporating both persistent as well as new clinical cases [1]. The increasing prevalence, nevertheless, of AD in the industrialized societies depicts the complex interactions of the stable genetic background with the constant perturbations of the environmental factors; AD is considered as a multifactorial disorder with a strong genetic background, accounting for approximately 75% of the total variability [2]. Numerous attempts have been conducted in order to decipher the polygenic etiology of AD, examining both the inter-individual variation of AD cases [3,4] as well as under the atopic march spectrum [5], including asthma, hay fever and AD. The majority of the associated loci lie in the central pathophysiological features of AD, characterized by the epidermal barrier dysfunction and the increased T helper (T H ) 2 cell-like inflammatory pattern in the skin [1]. Specifically, loss-of-function mutations in the filaggrin (FLG) gene, including the rs558269137 (2282del4) and rs61816761 (R501X) [6] Single-Nucleotide Polymorphisms (SNPs), which facilitates the formation of the cornified envelope, have been consistently associated with an increased risk for AD in various ethnic populations, due to their implication in the reduced granular layer [7]. Disease progression is additionally promoted by the participation of microbial alterations and physical damage [1], leading to the increased epithelial permeability and initiation of a type 2 immune response [1] mediated by the TSLP cytokine [8]. Genetic variants mapped in the type 2 cytokine cluster, such as the IL4 rs2243248 [9] and the IL13 rs1800925 [10] SNPs, regulate the secretion of several T H 2-related interleukins (ILs), with the major examples of the IL4 and IL13 that have been further targeted by modern therapeutic approaches [11].
Accumulating evidence from genome-wide association studies (GWASs) in participants of European ancestry (n = 116,863, n cases = 21,399) has identified 31 genetic variants that account for less than 20% of the general heritability [3], while extended rare variant approaches have extended the explained variation by 12.56% [4]. Contrastingly, the latest Asian GWAS reported four novel Asian-specific AD susceptibility loci from a total of 118,287 participants (n cases = 2639) [12]. However, discrepancies observed between the number of genome-wide scans conducted in both populations [13] limit the representation of diverse ethnic backgrounds in GWAS. In light of these high-throughput approaches, re-evaluation of the candidate gene studies and gene-by-interaction hypotheses is of paramount importance to complement the above results and further characterize the genetic architecture of complex disorders through genotyping of targeted biological pathways. Nevertheless, the relatively small sample size, lack of statistical power and potentially biased reportage of the derived results hinder the conducting of such approaches despite their established efficacy .
To address this issue, we assessed the inter-individual variability present in the genetic predisposition of AD, incorporating a conservative random-effects meta-analysis (REM) approach to synthesize the evidence derived from available studies and thus unveil putative risk loci associated with the disease onset in participants of European  and Asian  ancestry.

Search Strategy and Selection Criteria
We conducted a systematic literature search to identify studies that examined the inter-individual genetic variability in patients with AD. Searches in the Medline database (through PubMed) using keywords referring to the association ('association,' 'susceptibility') of genetic variants ('polymorphism,' 'variant,' 'SNP,' and related terms) with the AD trait ('atopic dermatitis,' 'eczema') from inception to 18 September 2022. Screening was performed by two independent authors in a standardized manner, following relevant titles and abstract reading before full-text reviewing. We considered all case-control candidategene studies that examined the genetic predisposition of atopic dermatitis through targeted genotyping, regardless of the age of onset and the allergic disease spectrum that includes hay fever, rhinitis and asthma. Studies included in our analysis were published in peerreviewed journals in the English language; variants incorporated in our meta-analysis should be investigated by at least two eligible studies. We excluded case reports, abstracts, animal studies, duplicate reported data and reports that did not provide the appropriate data for the calculation of effect size and confidence intervals.

Data Extraction
Two authors extracted concurrently and independently the eligible studies and extracted the appropriate data; discrepancies were resolved through re-examination of the respective publications until consensus was reached. Data extracted from the eligible studies referred to the first author, publication year, total sample size and the reported genotypes for further effect size computations. respective publications until consensus was reached. Data extracted from the eligible studies referred to the first author, publication year, total sample size and the reported genotypes for further effect size computations.
Odds Ratios (ORs), along with their 95% Confidence Intervals (95% Cis), were calculated for the allelic model of inheritance. To avoid the parallel assessment of non-independent SNPs in the same locus, we measured linkage disequilibrium (LD) using the LDpop tool provided by the National Institutes of Health (NIH; https://ldlink.nci.nih.gov/?tab=ldpop; accessed on 15 November 2022). R-squared values greater than 0.9 highlighted SNPs in high LD and were therefore considered as a single locus in the meta-analysis.

Statistical Analysis
We calculated OR and 95% CI for each respective heritance pattern by synthesizing the available study-specific evidence through the random-effects model. The presence of heterogeneity was assessed with the Cochran's Q test (considered significant at p-value < 0.1) and quantified via the I 2 metric, with 0% < I 2 < 25% reporting small, 25% < I 2 < 50% reporting moderate, 50% < I 2 < 75% reporting high and I 2 > 75% reporting very high heterogeneity. The predominant role of the FLG variants in the AD predisposition led us to further evaluate their combined effect due to their similar biological effects. To explore the inter-study variability, we employed the Harbord's modified test, a robust statistical analysis for the assessment of small-study effects in each meta-analysis. All statistical analyses were performed with the Stata 13.1 software (Stata Corp, College Station, TX, USA). The metan plugin was incorporated for our REMs, while the metabias plugin was utilized for the Harbord's modified test. Statistical significance threshold was set at p-value ≤ 0.05, while the p-value < 0.1 threshold was used for the Harbord's modified test.

FLG Gene
We confirmed all significant associations between the FLG loss-of-function (LOF) variants and increased AD risk from a total of 27 studies, incorporating 17,759 participants. Specifically, both The predominant role of the FLG locus in the disease onset, as well as the similar, trans-acting biological mechanism of the above variants [34], prompted us to investigate their effect under the spectrum of combined genotypes. Three combined genotypes (CGs) were identified from the included studies (CG1: rs558269137 and rs61816761; CG2: rs558269137, rs61816761 and rs138726443; CG3: rs558269137, rs61816761, rs138726443 and rs150597413); CG1 incorporated 5263 participants from 9 studies, CG2 included 2264 AD cases from 3 studies and CG3 was assessed in 5391 eczema patients from 4 studies. As expected, all CGs yielded significant associations (OR (95% CI)

SPINK5 Gene
Comparably to IL4 variants, two SPINK5 variants (rs2303063, rs2303067) were found in high LD (r 2 > 0.9) and were thus examined together from a total of three studies (n = 1280). However, we found that the above locus was significantly associated with AD risk (OR (95% CI): 1.26 (1.06-1.48); I 2 = 0.0%). Relevant, independent variants mapped to the SPINK5 gene yielded additional significant signals, with the examples of SPINK5  Table S2).

Discussion
Here, we performed a systematic review and meta-analysis of all available evidence considering the genetic predisposition to AD in participants of both European and Asian ancestry. Despite the expanding list of risk loci identified through genome-wide scans in AD, our REM shed light upon new risk loci mapped to the IL18 and TGFB1 genes in Europeans (Figure 2), while the contribution of IL12RB1 and MIF loci in AD predisposition in Asians was additionally characterized through our analysis, revealing results not previously reported by genome-wide scans (Figure 3). We thus provide novel insights into the genetic architecture of the disease, highlighting perturbed biological pathways that could be further implicated in the development of new therapeutic mechanisms.
Several of our findings have been previously discovered by GWASs in both participants of European and Asian ancestries. In particular, the present meta-analysis confirms the pivotal role of the FLG locus in disease susceptibility, as LOF mutations disrupt the production of the filaggrin protein in the stratum corneum. The impaired barrier function facilitates a loss of water and pH increase, as well as the entry of irritants and allergens, contributing to the onset and exacerbation of allergic reactions [113]. The consistency of these findings across European and Asian populations highlights the universal importance of FLG in maintaining skin homeostasis and its relevance in AD [114]. The presence of significant heterogeneity concerning FLG rs558269137 and rs61816761 in European population was also observed in a prior meta-analysis of FLG polymorphisms in AD [115], which could be attributed to several demographic factors and environmental exposures that mediate the increased atopic eczema risk [113]. We further validate the 11q13.5 rs7927894 SNP as a risk locus, a variant localized within a non-coding genomic region. Notably, the above variant has been additionally implicated in Crohn's disease and deregulated epithelial function, thus potentially contributing to the pathogenic mechanisms underlying AD [55]. The cytokine cluster located at 5q31, harboring both IL4 and IL13, represents a prominent risk locus implicated in the pathogenesis of the disease and has been extensively linked with AD through several GWAS-derived variants. Here, we identified the IL13 rs1800925 SNP, a regulatory variant that enhances the activity of the IL13 promoter in human Th2 lymphocytes, thereby augmenting the risk of allergic disorders [10]. Remarkably, this particular variant further exhibits a significant association with cutaneous T-cell lymphoma [116], a malignancy that has been observed to manifest in some cases of AD patients following the administration of dupilumab [117], an anti-IL4RA biological drug that inhibits both IL4 and IL13 signaling [11]. Another significantly associated variant that belongs to the 5q31 locus is the IL4 rs2243248 SNP, where our REM reported the protective effect of the minor rs2243248 G allele in AD risk, similar to asthma [9]. Relevant, previously identified risk variants refer to the SPINK5 nonsynonymous variants that disrupt the function of serine protease inhibitors in the integrity of the epidermal barrier, as well as the IL9 rs31563 variant that exhibits a regulatory role in IL9 expression, a cytokine known to contribute to B cell isotype switching from IgM to IgE synthesis [102].
Despite the prominence of GWAS outcomes as the gold standard in genetic association studies, there are still certain risk loci that have not been yet identified. We further report four novel loci, two belonging to Europeans ( Figure 2) and two Asian-specific ( Figure 3). In particular, we detected a significant association between the IL18 rs187238 polymorphism and AD risk in European cases ( Figure 2); this variant, residing within the IL18 promoter region, supports the modulation of the binding affinity of transcription factors and thus participates in the dysregulated Th2 response [118]. Our second novel addition in Europeans is the TGFB1 rs1800471 missense variant ( Figure 2) that significantly alters the inhibitory action of TGFB1 protein molecule during allergic reactions [119].
In addition, the promoter IL12RB1 rs393548 and rs436857 variants were significantly associated with AD risk in Asians ( Figure 3); the IL12RB1 receptor subunit is involved in the modulation of the IL12-dependent inhibition of IgE synthesis and Th2 cell function. Consequently, a potential decrease of the IL12RB1 expression perturbs the above inhibitory activity [101], thus leading to the dysregulation of downstream immune responses and potentially contributing to allergic reactions. Finally, our findings indicate a significant association between the MIF rs755622 C allele and increased AD susceptibility ( Figure 3); the rare C allele exhibits enhanced promoter activity relative to the common G allele, leading to differential MIF expression levels [103]. Studies conducted on mouse models of AD have demonstrated that MIF holds a significant role in the development of AD-related immune dysregulation through the induction of a type 2 immune response, as well as through fostering the recruitment of eosinophils in the cutaneous inflammation [120].
However, our study displays some constraints. Despite the assessment of small-study effect bias through the Harbord's modified test and the identification of six significant cases, the majority of studies included in our synthesis incorporated a relatively small sample size and are thus susceptible to biases [121]. In addition, discrepancies between the available data derived from the included studies, with the examples of disease activity, age of onset and relevant clinical information, restricted further stratification and subgroup examination in our analysis. The predominant role of the exposome in the AD predisposition [113] and the inclusion of such environmental risk factors in multivariate analyses is of paramount importance to gain a holistic understanding of AD etiopathology and thus unveil novel therapeutic targets.
In conclusion, we conducted a systematic review and meta-analysis of all available data regarding the genetic predisposition to AD in participants of both European and Asian ancestry. We report four novel risk loci that have not been previously shown by GWAS, partially contributing to the elucidation of the genetic architecture of the disease.
By exploring data from both ethnic groups, our study endeavors to provide a broader understanding and enhance the applicability of our findings across diverse populations. High-throughput investigation of putative risk loci in AD [122,123] could further unveil their functional role in the disease predisposition. In addition, the incorporation of clinical information for each included participant, as well as the vast number of environmental exposures associated with increased AD risk, shall facilitate the interpretation of diseaseassociated interactions and form the framework for precision medicine.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/genes14071456/s1, Table S1. Association between single nucleotide polymorphisms and atopic dermatitis susceptibility in patients of European descent. Table S2. Association between single nucleotide polymorphisms and atopic dermatitis susceptibility in patients of Asian descent.
Funding: This research received no external funding.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: All data are available from the corresponding author upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.