DNA methylation-based subclassification of psoriasis in the Chinese Han population

Abstract

Psoriasis (Ps) is an inflammatory skin disease caused by genetic and environmental factors. Previous studies on DNA methylation (DNAm) found genetic markers that are closely associated with Ps, and evidence has shown that DNAm mediates genetic risk in Ps. In this study, Consensus Clustering was used to analyze DNAm data, and 114 Ps patients were divided into three subclassifications. Investigation of the clinical characteristics and copy number variations (CNVs) of DEFB4, IL22, and LCE3C in the three subclassifications revealed no significant differences in gender ratio and in Ps area and severity index (PASI) score. The proportion of late-onset (≥40 years) Ps patients was significantly higher in type I than in types II and III (P = 0.035). Type III contained the smallest proportion of smokers and the largest proportion of non-smoking Ps patients (P = 0.086). The CNVs of DEFB4 and LCE3C showed no significant differences but the CNV of IL22 significantly differed among the three subclassifications (P = 0.044). This study is the first to profile Ps subclassifications based on DNAm data in the Chinese Han population. These results are useful in the treatment and management of Ps from the molecular and genetic perspectives.

Keywords

psoriasis / DNA methylation / subclassification

Introduction

Psoriasis (Ps) is a common inflammatory skin disorder caused by genetic and epigenetic factors with various environmental triggers in predisposed individuals [¹,²]. Plaques of Ps involve any part of the body in a localized or generalized way, especially favoring extensor surfaces. The characteristic histological changes in patients with Ps include hyperkeratosis, parakeratosis, and orthokeratosis. Ps is clinically classified into five subtypes, namely, plaque, guttate, inverse, pustular, and erythrodermic. The latest global report by the World Health Organization indicated that plaque Ps is the most common type, account for 58% to 97% of all Ps patients [³].

DNA methylation (DNAm) is associated with multiple diseases, including Ps, because of its effect on the transcriptional regulation and the control of promoter/exon usage and alternative splicing [⁴–⁸]. Recent research on molecular classification has revealed individual molecular signatures and developed reliable classification methods [⁹], especially in cancer [¹⁰,¹¹], complementing the deficiency of clinical or histological classification. Ps subclassifications usually depend on various clinical phenotypes, onset age, disease severity, and disease morphology [¹²], and substantial differences behind the similar appearance cannot be distinguished.

Molecular subclassification reveals detailed information about the clinical management of diseases [¹³,¹⁴]. Researchers used gene expression data to profile classification models of Ps on the basis of the novel incremental feature of a selection algorithm; the final Ps classification model shows a highly stable prediction accuracy and utilizes only three features from two unique genes, IGFL1 and C10orf99 [¹⁵]. A comprehensive analysis of gene expression in paired lesional and non-lesional psoriatic tissue samples revealed distinct molecular subclassifications of plaque Ps within the clinical phenotype [¹⁶]. However, molecular subclassification based on DNAm has not been performed in Ps, and the relationship between molecular subclassifications and clinical features remains to be established. In the present study, we describe a subclassification of Ps based on DNAm data from 114 plaque Ps biopsy samples and compare the clinical characteristics of different subclassifications. Basing from the DNA copy number variations (CNVs) of three Ps susceptibility genes (DEFB4, IL22, and LCE3C) that are associated with Ps [¹⁷–²⁰], we compare the CNV frequency difference of these genes in different Ps subclassifications.

Materials and methods

Psoriasis DNA methylation data

Previously, we conducted an epigenome-wide association analysis of plaque Ps by using Illumina Infinium Human-Methylation450k microarray, in which a genome-wide DNAm profile of 264 Ps-specific DNAm loci was discovered among psoriatic skin tissues, uninvolved psoriatic skin tissues, and unaffected skin tissues [⁷,²¹]. In this study, we focused on the 264 Ps-specific loci and performed methylation-based subclassification in the Chinese Han population. Sample collection, methylation experiment, locus selection, and statistical analysis were performed as described in previous study [⁷].

Consensus clustering analysis

Consensus Clustering (CC) was performed to detect clusters on the basis of the DNAm data of 114 Chinese Han patients with Ps. CC provides quantitative evidence for determining the number and association of possible clusters within a data set [²²,²³]. CC involves subsampling from a set of items and determines the clustering of specified cluster counts (k). Then, pairwise consensus values, the proportion of two items occupying the same cluster and the number of times they occurred in the same subsample, are calculated and stored in a symmetric consensus matrix (CM) for each k. CM has both row and column tables, in which consensus values range from 0 (never clustered together) to 1 (always clustered together) marked by white to dark blue. CM is ordered by CC and depicted as a dendrogram at the top of a heatmap. Cluster memberships are marked by colored rectangles between the dendrogram and the heatmap. Consensus cumulative distribution function (CDF) plot shows the cumulative distribution functions of the CM for each k, which allows a user to determine at what number of clusters (k) the CDF reaches an approximate maximum; thus, consensus and cluster confidence is at a maximum. Delta area plot shows the relative change in the area under the CDF curve comparing k and k−1, which allows a user to determine the relative increase in consensus and the k value at which no appreciable increase exists. The CM, CDF plot, and Delta area plot enable a user to decide upon a reasonable cluster number. ConsensusClusterPlus implements CC in R language and adds new functionality and visualizations [²³]. Partitioning around medoids (PAM) clustering method was applied in the CC analysis by using 1-Pearson correlation metrics as the distance between clusters. In CC analysis, 80% of items and all feature resamplings were selected.

Silhouette width analysis

Silhouette plot is a graphical aid to the interpretation and validation of cluster analysis [²⁴]. Each cluster is represented by a silhouette, which is based on the comparison of tightness and separation. The silhouette shows which objects lie well within their clusters and which ones are merely somewhere in between clusters. The entire clustering is displayed by combining the silhouettes into a single plot, allowing an appreciation of the relative quality of the clusters and an overview of the data configuration. The average silhouette width (ASW) provides an evaluation of clustering validity and is used to select an appropriate number of clusters. The summary measure of ASW is as follows: ASW≤0.25 shows that no substantial structure has been found, 0.25<ASW≤ 0.50 shows the structure is weak and could be artificial, 0.50<ASW≤0.70 shows that a reasonable structure has been found, and 0.70<ASW≤1.0 shows that a strong structure has been found [²⁵].

Copy number detection and analysis of IL22, DEFB4, and LCE3C

IL22 gene exon 1 (Hs00146600_CN), LCE3C gene exon 1 (Hs02550639_CN), custom-designed DEFB4 assay (DEFB4_CCKAK1P), and RNase P reference assay (assay ID: 4403328, Applied Biosystems, Foster City, CA, USA) were used to detect the copy numbers of these three genes. ABI PRISM 7900HT Real-time PCR instrument and the SDS 2.2 software package (Applied Biosystems, Foster City, CA, USA) were used to call copy numbers for each sample. The thermal-cycling conditions were as follows: 95 °C for 10 min followed by 45 cycles at 95 °C for 15 s and 60 °C for 60 s. The copy number of each target sequence was determined by the relative quantitation using the comparative CT method based on the assumption that the RNaseP gene has two copies of the DNA segment in the calibrator sample. In addition, Pearson Chi-square test was used to assess the differences in copy numbers and clinical parameters among different subclassifications. Statistical analysis was performed using the R program (http://www.r-project.org). In consideration of the relatively small research sample size in the present study, statistical significance was considered at P<0.1.

Results

Consensus clustering identifies three psoriasis subclassifications

CC was used to analyze the DNAm data, and the cluster number k was defined as 2−10 (Supplementary Fig. 1). When k = 2 (Fig. 1 A), the membership number of type II was nearly twice that of type I, and the CDF value was relatively small if the consensus index was low (Fig. 1 D). When k = 4 (Fig. 1 C), a new small cluster appeared in the clusters, and the Delta area plot (Fig. 1 E) shows that the relative change in the area under the CDF curve noticeably decreased. The CM shows each cluster equally, and consensus index is relatively high at k = 3 (Fig. 1 B). Patients with Ps were primarily divided into three subclassifications (type I with 29 Ps, type II with 39 Ps, and type III with 46 Ps). Silhouette width analysis was employed to evaluate the relative quality and data configuration of the three clusters. The silhouette width values of the three clusters (Fig. 1 F) are 0.26, 0.26, and 0.14, and the ASW value is 0.21, indicating a relatively weak structure. The trace plot (Supplementary Fig. 2), Cluster-Consensus plot (Supplementary Fig. 3), and Item-Consensus plot (Supplementary Fig. 4) are auxiliary for the determination of Ps subclassification.

Characteristics and distribution of psoriasissubclassifications

The characteristics of the three subclassifications are shown in Table 1, and details of the characteristic distributions are shown in Fig. 2. Clinical characteristics, including gender, onset age, smoking status, and Ps area and severity index (PASI) score, were evaluated, and a comparative analysis of the clinical characteristics of the three subclassifications was performed (Fig. 3). The proportions of females and males in the three subclassifications were different, with type II having the lowest proportion of females, but no significant difference was observed among the three subclassifications (P = 0.62). Analysis of patient onset age revealed that the proportions of early-onset patients (<40 years old) and late-onset patients (≥40 years old) were close in types II and III, whereas type I had a higher proportion of late-onset patients and lower proportion of early-onset patients compared with the two other types. The difference among the three subclassifications was statistically significant (P = 0.035). Analysis of Ps patients who smoked revealed that type III had the smallest proportion of smokers and the largest proportion of non-smokers, and a statistical difference existed in the three subclassifications (P = 0.086). A comparative analysis of the PASI scores of Ps patients showed that the highest frequency of 3≤PASI<9 was in type II, and the difference among the three subclassifications was not statistically significant (P = 0.25).

Copy numbers of DEFB4, IL22, and LCE3C in the three psoriasis subclassifications

The DEFB4 gene encodes an antibiotic peptide defensin β4 and affects the function of the innate immune system [²⁶]. IL22 encodes the inflammatory factor interleukin 22, which encodes protein that is involved in the inflammatory regulatory pathway and may have a role in the pathogenesis of Ps [²⁷]. LCE3C is a gene encoding late keratinocyte cornea, which is involved in the repair of skin tissue [²⁸]. A gap between LCE3C and LCE3B often forms LCE3BC-del and is closely related to the pathogenesis of Ps [²⁸]. We investigated the copy number of DEFB4, IL22, and LCE3C in the three subclassifications (Fig. 4). The copy number deletion of IL22 and LCE3C were found in types I and III, respectively. For all three genes with different frequencies of CNV in the three subclassifications, the highest frequency of copy number≥3 was found in LCE3C. The CNV frequency difference of IL22 in the three subclassifications was significant (P = 0.044). However, the CNV frequency difference of DEFB4 and LCE3C was not significant (P = 0.36 and P = 0.47, respectively) in the three Ps subclassifications.

Discussion

Ps is a persistent, chronic, and common skin disorder that changes the life cycle of skin cells and causes cells to build up rapidly on the skin surface [²⁹]. Patients with Ps are divided into different subtypes according to clinical characteristics, onset age, and PASI score. Phenotypic heterogeneity of Ps refers to clinical presentation, comorbid disorders, disease progression, and treatment response. The natural history of Ps was evaluated in a large American survey with different affected sites, onset ages, disease severity levels, familial histories, and remission factors [³⁰]. As for the evaluation of treatment, variable responses to the same therapy options, including topical therapies, phototherapy, conventional, and biological systematic therapies, were observed in patients with Ps [³¹]. These heterogeneities might be attributed to the subtle distinctions in the genetics of different individuals, suggesting the presence of intrinsic molecular subtypes.

DNAm is associated with human development and various diseases, and multiple epigenetic studies have shown that DNAm is associated with Ps and regulates the expression of Ps susceptibility genes [³²]. A previous epigenome-wide association study in the Chinese Han population found nine methylation differential sites associated with Ps, and CYP2S1, ECE1, EIF2C2, MAN1C1, and DLGAP4 are negatively correlated with DNAm [⁷]. The relationship between different differentiation sites and gene expression was compared, and three CpG sites were found to regulate the expression of C1orf106, DMBX1, and SIK3 [²¹]. Interestingly, genome methylation data were used to identify the molecular subtypes of some diseases [¹⁴]. For Ps, Roberson et al. clustered the 50 most differentially methylated sites that separate psoriatic from normal skin samples with uninvolved skin exhibiting intermediate methylation [³³]. However, this study did not classify Ps patients into different subtypes. Hence, we described a molecular subclassification of Ps based on DNAm data in the Chinese Han population, evaluated the clinical characteristics, and investigated the CNVs of three susceptible genes (LCE3C, IL22, and DEFB4) in different subclassifications.

In this study, we divided patients with Ps into 2–10 parts based on the DNAm data. When k is larger, a more stable consistency value can be obtained (Supplementary Fig. 1). However, given the limited number of samples and the clinical division of more subclassifications that easily lead to complications, we divided the patients into less than five subclassifications. When the patients were divided into two subclassifications, the membership number of type II was considerably higher than that of type I, and two subtypes cannot be more specific to reflect the characteristics of subtypes. Hence, we did not divide Ps patients into two subclassifications in this study. The main comparison was the division of three and four subclassifications when the cluster stability and consistency within the cluster project were achieved. A coherence matrix of k = 4 shows that the membership number of type III was very small; thus, the difference between the various subclassifications was expanded. When Ps patients were divided into three subclassifications (k = 3), each cluster sample was evenly distributed, and the consensus index was relatively high, indicating a more stable clustering. Therefore, we divided Ps patients into three subclassifications in this study.

We evaluated the clinical characteristics (gender, onset age, smoking status, and PASI score) of Ps in the different subclassifications. Previous studies have shown no difference in the gender ratio of patients with Ps [³⁴]. In the present study, the proportion of men and women were close in types I and III, whereas the proportion of females was significantly lower in type II than in the two other types. However, no statistical significant difference was observed in the three subclassifications. Among the three subtypes, type II contains the highest frequency of Ps patients with 3≤PASI<9. However, the difference in the frequency of PASI score was not statistically significant among the three subclassifications, indicating that the severity of Ps has no direct relationship in each subclassification, which is different from the classification depending on the PASI score.

Clinically, the onset age of Ps is an important basis for the classification of Ps subtypes [³⁵], and a positive association between the prevalence of smoking and Ps as well as an association between smoking and severity of Ps have been observed [³⁶]. Studies have shown that smoking affects the risk of Ps and plays an important role in the pathogenesis of Ps [³⁷]. In the analysis of subtype onset age, the ratio of patients with onset age≥40 years old in types II and III was significantly higher than that of patients with onset age<40 years old. The difference between the three subclassifications was statistically significant. Smoking statuses were compared and analyzed in the three subclassifications. The proportion of smokers in type III was the smallest, and the proportion of non-smokers was the largest with significant difference. Therefore, in the molecular subclassification of Ps on the basis of DNAm data, onset age and smoking status may play important roles in Ps.

In addition to the clinical factors and characteristics of Ps, we also attempted to explore the intrinsic factors of the disease, such as genetic variation differences that may exist between different classifications. Studies have found that the CNV changes of DEFB4, IL22, and LCE3C are closely associated with the pathogenesis of Ps or interact with other genetic factors to increase Ps risk [¹⁸–²⁰,³⁸–⁴⁰]. Therefore, the CNV of the three susceptible genes in Ps were compared between the different subclassifications. Results showed that the CNV frequency difference of DEFB4 and LCE3C in the different subclassifications was not significant, whereas that of IL22 was significant. These results show that the CNV of the IL22 gene may be one of the factors that are involved in this classification.

This study is the first to utilize DNAm data for profiling Ps molecular subclassifications, which will enrich the classification of Ps and provide an in-depth understanding of the differences among different molecular subtypes. Thus, the present study can serve as a reference to optimize the management of Ps and provide a theoretical basis for the study of Ps target drugs. Nevertheless, the present study has some shortcomings that should be improved and overcome. The collected sample size is relatively small, which may not be able to reflect completely the differences between the various subclassifications of Ps and may lead to a low ASW. For future studies, more Ps patients need to be collected to validate the identified subclassifications in our study, and some more susceptibility genes should be analyzed. New statistical methods could also be used to clarify the difference among these various subtypes in different tissue/cell types.

Conclusions

Three molecular Ps subclassifications were profiled through DNAm data analysis in the Chinese Han population. Comparative analysis showed statistically significant differences in age onset, smoking status, and CNV of IL22. The differences in gender ratio, PASI score, CNVs of DEFB4 and LCE3C were not significant. The subclassification analysis of Ps is helpful in the management and treatment of Ps from the molecular and genetic perspectives, providing new molecular insights and theoretical basis for individual treatment of patients with Ps.

References

[1] Nestle FO, Kaplan DH, Barker J. Psoriasis. N Engl J Med 2009; 361(5): 496–509

[2] Gudjonsson JE, Krueger G. A role for epigenetics in psoriasis: methylated cytosine-guanine sites differentiate lesional from nonlesional skin and from normal skin. J Invest Dermatol 2012; 132(3 Pt 1): 506–508

[3] WHO. Global Report on Psoriasis. United States: World Health Organization, 2016

[4] Eckhardt F, Lewin J, Cortese R, Rakyan VK, Attwood J, Burger M, Burton J, Cox TV, Davies R, Down TA, Haefliger C, Horton R, Howe K, Jackson DK, Kunde J, Koenig C, Liddle J, Niblett D, Otto T, Pettett R, Seemann S, Thompson C, West T, Rogers J, Olek A, Berlin K, Beck S. DNA methylation profiling of human chromosomes 6, 20 and 22. Nat Genet 2006; 38(12): 1378–1385

[5] Maunakea AK, Chepelev I, Cui K, Zhao K. Intragenic DNA methylation modulates alternative splicing by recruiting MeCP2 to promote exon recognition. Cell Res 2013; 23(11): 1256–1269

[6] Gervin K, Vigeland MD, Mattingsdal M, Hammerø M, Nygård H, Olsen AO, Brandt I, Harris JR, Undlien DE, Lyle R. DNA methylation and gene expression changes in monozygotic twins discordant for psoriasis: identification of epigenetically dysregulated genes. PLoS Genet 2012; 8(1): e1002454

[7] Zhou F, Wang W, Shen C, Li H, Zuo X, Zheng X, Yue M, Zhang C, Yu L, Chen M, Zhu C, Yin X, Tang M, Li Y, Chen G, Wang Z, Liu S, Zhou Y, Zhang F, Zhang W, Li C, Yang S, Sun L, Zhang X. Epigenome-wide association analysis identified nine skin DNA methylation loci for psoriasis. J Invest Dermatol 2016; 136(4): 779–787

[8] Elliott G, Hong C, Xing X, Zhou X, Li D, Coarfa C, Bell RJ, Maire CL, Ligon KL, Sigaroudinia M, Gascard P, Tlsty TD, Harris RA, Schalkwyk LC. Intermediate DNA methylation is a conserved signature of genome regulation. 2015; 6: 6363

[9] Brunet JP, Tamayo P, Golub TR, Mesirov JP. Metagenes and molecular pattern discovery using matrix factorization. Proc Natl Acad Sci USA 2004; 101(12): 4164–4169

[10] Cho YJ, Tsherniak A, Tamayo P, Santagata S, Ligon A, Greulich H, Berhoukim R, Amani V, Goumnerova L, Eberhart CG, Lau CC, Olson JM, Gilbertson RJ, Gajjar A, Delattre O, Kool M, Ligon K, Meyerson M, Mesirov JP, Pomeroy SL. Integrative genomic analysis of medulloblastoma identifies a molecular subgroup that drives poor clinical outcome. J Clin Oncol 2011; 29(11): 1424–1430

[11] Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, Miller CR, Ding L, Golub T, Mesirov JP, Alexe G, Lawrence M, O’Kelly M, Tamayo P, Weir BA, Gabriel S, Winckler W, Gupta S, Jakkula L, Feiler HS, Hodgson JG, James CD, Sarkaria JN, Brennan C, Kahn A, Spellman PT, Wilson RK, Speed TP, Gray JW, Meyerson M, Getz G, Perou CM, Hayes DN; Cancer Genome Atlas Research Network. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 2010; 17(1): 98–110

[12] Raychaudhuri SK, Maverakis E, Raychaudhuri SP. Diagnosis and classification of psoriasis. Autoimmun Rev 2014; 13(4-5): 490–495

[13] Noushmehr H, Weisenberger DJ, Diefes K, Phillips HS, Pujara K, Berman BP, Pan F, Pelloski CE, Sulman EP, Bhat KP, Verhaak RG, Hoadley KA, Hayes DN, Perou CM, Schmidt HK, Ding L, Wilson RK, Van Den Berg D, Shen H, Bengtsson H, Neuvial P, Cope LM, Buckley J, Herman JG, Baylin SB, Laird PW, Aldape K; Cancer Genome Atlas Research Network. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell 2010; 17(5): 510–522

[14] Bell D BA, Birrer M, Chien J, Cramer D, Dao F, Dhir R, DiSaia P, Gabra H, Glenn P, Godwin A, Gross J, Hartmann L, Huang M, Huntsman D, Iacocca M, Imielinski M, Kalloger S, Karlan B, Levine D, Mills G, Morrison C, Mutch D, Olvera N, Orsulic S, Park K, Petrelli N, Rabeno B, Rader J, Sikic B, Smith-McCune K, Sood A, Bowtell D, Penny R, Testa J, Chang K, Dinh H, Drummond J, Fowler G, Gunaratne P, Hawes A, Kovar C, Lewis L, Morgan M, Newsham I, Santibanez J, Reid J, Trevino L, Wu Y-, Wang M, Muzny D, Wheeler D, Gibbs R, Getz G, Lawrence M, Cibulskis K, Sivachenko A, Sougnez C, Voet D, Wilkinson J, Bloom T, Ardlie K, Fennell T, Baldwin J, Gabriel S, Lander E, Ding LL, Fulton R, Koboldt D, McLellan M, Wylie T, Walker J, O'Laughlin M, Dooling D, Fulton L, Abbott R, Dees N, Zhang Q, Kandoth C, Wendl M, Schierding W, Shen D, Harris C, Schmidt H, Kalicki J, Delehaunty K, Fronick C, Demeter R, Cook L, Wallis J, Lin L, Magrini V, Hodges J, Eldred J, Smith S, Pohl C, Vandin F, Raphael B, Weinstock G, Mardis E, Wilson R, Meyerson M, Winckler W, Getz G, Verhaak R, Carter S, Mermel C, Saksena G, Nguyen H, Onofrio R, Lawrence M, Hubbard D, Gupta S, Crenshaw A, Ramos A, Ardlie K, Chin L, Protopopov A, Zhang J, Kim T, Perna I, Xiao Y, Zhang H, Ren G, Sathiamoorthy N, Park R, Lee E, Park P, Kucherlapati R, Absher M, Waite L, Sherlock G, Brooks J, Li J, Xu J, Myers R, Laird PW, Cope L, Herman J, Shen H, Weisenberger D, Noushmehr H, Pan F, Triche T Jr, Berman B, Van Den Berg D, Buckley J, Baylin S, Spellman P, Purdom E, Neuvial P, Bengtsson H, Jakkula L, Durinck S, Han J, Dorton S, Marr H, Choi Y, Wang V, Wang N, Ngai J, Conboy J, Parvin B, Feiler H, Speed T, Gray J, Levine A, Socci N, Liang Y, Taylor B, Schultz N, Borsu L, Lash A, Brennan C, Viale A, Sander C, Ladanyi M, Hoadley K, Meng S, Du Y, Shi Y, Li L, Turman Y, Zang D, Helms E, Balu S, Zhou X, Wu J, Topal M, Hayes D, Perou C, Getz G, Voet D, Saksena G, Zhang J, Zhang H, Wu C, Shukla S, Cibulskis K, Lawrence M, Sivachenko A, Jing R, Park R, Liu Y, Park P, Noble M, Chin L, Carter H, Kim D, Karchin R, Spellman P, Purdom E, Neuvial P, Bengtsson H, Durinck S, Han J, Korkola J, Heiser L, Cho R, Hu Z, Parvin B, Speed T, Gray J, Schultz N, Cerami E, Taylor B, Olshen A, Reva B, Antipin Y, Shen R, Mankoo P, Sheridan R, Ciriello G, Chang W, Bernanke J, Borsu L, Levine D, Ladanyi M, Sander C, Haussler D, Benz C, Stuart J, Benz S, Sanborn J, Vaske C, Zhu J, Szeto C, Scott G, Yau C, Hoadley K, Du Y, Balu S, Hayes D, Perou C, Wilkerson M, Zhang N, Akbani R, Baggerly K, Yung W, Mills G, Weinstein J, Penny R, Shelton T, Grimm D, Hatfield M, Morris S, Yena P, Rhodes P, Sherman M, Paulauskis J, Millis S, Kahn A, Greene J, Sfeir R, Jensen M, Chen J, Whitmore J, Alonso S, Jordan J, Chu A, Zhang J, Barker A, Compton C, Eley G, Ferguson M, Fielding P, Gerhard D, Myles R, Schaefer C, Mills Shaw K, Vaught J, Vockley J, Good P, Guyer M, Ozenberger B, Peterson J, Thomson E. Integrated genomic analyses of ovarian carcinoma. Nature 2011; 474(7353): 609–615

[15] Guo P, Luo Y, Mai G, Zhang M, Wang G, Zhao M, Gao L, Li F, Zhou F. Gene expression profile based classification models of psoriasis. Genomics 2014; 103(1): 48–55

[16] Ainali C, Valeyev N, Perera G, Williams A, Gudjonsson JE, Ouzounis CA, Nestle FO, Tsoka S. Transcriptome classification reveals molecular subtypes in psoriasis. BMC Genomics 2012; 13(1): 472

[17] Jaradat SW, Cubillos S, Krieg N, Lehmann K, Issa B, Piehler S, Wehner-Diab S, Hipler UC, Norgauer J. Low DEFB4 copy number and high systemic hBD-2 and IL-22 levels are associated with dermatophytosis. J Invest Dermatol 2015; 135(3): 750–758

[18] Hollox EJ, Huffmeier U, Zeeuwen PL, Palla R, Lascorz J, Rodijk-Olthuis D, van de Kerkhof PC, Traupe H, de Jongh G, den Heijer M, Reis A, Armour JA, Schalkwijk J. Psoriasis is associated with increased β-defensin genomic copy number. Nat Genet 2008; 40(1): 23–25

[19] Prans E, Kingo K, Traks T, Silm H, Vasar E, Kõks S. Copy number variations in IL22 gene are associated with psoriasis vulgaris. Hum Immunol 2013; 74(6): 792–795

[20] Li M, Wu Y, Chen G, Yang Y, Zhou D, Zhang Z, Zhang D, Chen Y, Lu Z, He L, Zheng J, Liu Y. Deletion of the late cornified envelope genes LCE3C and LCE3B is associated with psoriasis in a Chinese population. J Invest Dermatol 2011; 131(8): 1639–1643

[21] Zhou F, Shen C, Xu J, Gao J, Zheng X, Ko R, Dou J, Cheng Y, Zhu C, Xu S, Tang X, Zuo X, Yin X, Cui Y, Sun L, Tsoi LC, Hsu YH, Yang S, Zhang X. Epigenome-wide association data implicates DNA methylation-mediated genetic risk in psoriasis. Clin Epigenetics 2016; 8(1): 131

[22] Monti STP, Mesirov J, Golub T. Consensus clustering: a resampling-based method for class discovery and visualization of gene expression microarray data. Mach Learn 2003; 52(1/2): 91–118

[23] Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics 2010; 26(12): 1572–1573

[24] Rousseeuw PJ. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J Comput Appl Math 1987; 20: 53–65

[25] Kaufman LRP. Finding Groups in Data: an Introduction to Cluster Analysis . New York: John Wiley & Sons, Inc., 1990

[26] Hellgren O, Sheldon BC. Locus-specific protocol for nine different innate immune genes (antimicrobial peptides: β-defensins) across passerine bird species reveals within-species coding variation and a case of trans-species polymorphisms. Mol Ecol Resour 2011; 11(4): 686–692

[27] Sabat R, Ouyang W, Wolk K. Therapeutic opportunities of the IL-22-IL-22R1 system. Nat Rev Drug Discov 2014; 13(1): 21–38

[28] Pajic P, Lin YL, Xu D, Gokcumen O. The psoriasis-associated deletion of late cornified envelope genes LCE3B and LCE3C has been maintained under balancing selection since Human Denisovan divergence. BMC Evol Biol 2016; 16(1): 265

[29] Lowes MA, Bowcock AM, Krueger JG. Pathogenesis and therapy of psoriasis. Nature 2007; 445(7130): 866–873

[30] Farber EM, Nall ML. The natural history of psoriasis in 5,600 patients. Dermatologica 1974; 148(1): 1–18

[31] Boehncke WH, Schön MP. Psoriasis. Lancet 2015; 386(9997): 983–994

[32] Pollock RA, Abji F, Gladman DD. Epigenetics of psoriatic disease: a systematic review and critical appraisal. J Autoimmun 2017; 78: 29–38

[33] Roberson ED, Liu Y, Ryan C, Joyce CE, Duan S, Cao L, Martin A, Liao W, Menter A, Bowcock AM. A subset of methylated CpG sites differentiate psoriatic from normal skin. J Invest Dermatol 2012; 132(3 Pt 1): 583–592

[34] Fan X, Yang S, Sun LD, Liang YH, Gao M, Zhang KY, Huang W, Zhang X. Comparison of clinical features of HLA-Cw*0602-positive and -negative psoriasis patients in a Han Chinese population. Acta Derm Venereol 2007; 87(4): 335–340

[35] Queiro R, Tejón P, Alonso S, Coto P. Age at disease onset: a key factor for understanding psoriatic disease. Rheumatology (Oxford) 2014; 53(7): 1178–1185

[36] Richer V, Roubille C, Fleming P, Starnino T, McCourt C, McFarlane A, Siu S, Kraft J, Lynde C, Pope JE, Keeling S, Dutz J, Bessette L, Gulliver WP, Haraoui B, Bissonnette R. Psoriasis and smoking: a systematic literature review and meta-analysis with qualitative analysis of effect of smoking on psoriasis severity. J Cutan Med Surg 2016; 20(3): 221–227

[37] Armstrong AW, Armstrong EJ, Fuller EN, Sockolov ME, Voyles SV. Smoking and pathogenesis of psoriasis: a review of oxidative, inflammatory and genetic mechanisms. Br J Dermatol 2011; 165(6): 1162–1168

[38] Stuart PE, Hüffmeier U, Nair RP, Palla R, Tejasvi T, Schalkwijk J, Elder JT, Reis A, Armour JAL. Association of β-defensin copy number and psoriasis in three cohorts of European origin. J Invest Dermatol 2012; 132(10): 2407–2413

[39] Hüffmeier U, Bergboer JG, Becker T, Armour JA, Traupe H, Estivill X, Riveira-Munoz E, Mössner R, Reich K, Kurrat W, Wienker TF, Schalkwijk J, Zeeuwen PL, Reis A. Replication of LCE3C-LCE3B CNV as a risk factor for psoriasis and analysis of interaction with other genetic risk factors. J Invest Dermatol 2010; 130(4): 979–984

[40] Coin LJ, Cao D, Ren J, Zuo X, Sun L, Yang S, Zhang X, Cui Y, Li Y, Jin X, Wang J. An exome sequencing pipeline for identifying and genotyping common CNVs associated with disease with application to psoriasis. Bioinformatics 2012; 28(18): i370–i374

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag GmbH Germany, part of Springer Nature