Introduction

Psoriasis is a complex skin disease. Its etiology is unknown, but genetics and the immune system play a key role in its development.1 Tumor necrosis factor (TNF), interleukin (IL)-12, IL-23 and the HLA-C*0602 allele are the main genetic risk factors for psoriasis.2 Genome-wide association studies and candidate gene studies have provided initial data on these risk factors. TNFα and the p40 subunit of IL-12 and IL-23 are the current therapeutic targets in moderate-to-severe psoriasis (anti-TNF drugs and ustekinumab, respectively).3 Therefore, genetic studies can reveal new therapeutic targets and increase our knowledge of the etiology and pathogenesis of psoriasis.

Moderate-to-severe psoriasis can be treated using several anti-TNF drugs, including etanercept, adalimumab and infliximab. Adalimumab and infliximab are antibodies, and etanercept is a dimeric fusion protein that acts against the proinflammatory cytokine, TNFα, that is involved in the development of psoriasis.3 Anti-TNF drugs are usually safe and well tolerated, but patients can develop adverse effects or not respond to treatment.3 Genetics can explain interindividual differences in the response to therapy. Few pharmacogenomic studies have been performed in psoriasis, and the results of the few studies performed have not been replicated.4, 5, 6, 7, 8, 9, 10, 11, 12 Therefore, we performed a candidate gene study to investigate single-nucleotide polymorphisms (SNPs) that can predict the response to anti-TNF drugs in Caucasian patients with moderate-to-severe plaque psoriasis.

Materials and methods

Study population

Our study included 144 patients recruited between 16 October 2007 and 17 December 2012 from four hospitals in Madrid, Spain (Hospital Universitario de la Princesa, Hospital Universitario Gregorio Marañón, Hospital Universitario Fundación Alcorcón and Hospital Universitario Infanta Leonor). All patients were Caucasians aged ⩾18 years with moderate-to-severe plaque psoriasis (defined according to the European consensus published by Mrowietz et al.13) requiring treatment with biological drugs. The patients were treated with anti-TNF agents according to the Summary of Product Characteristics. Only patients who were naive for anti-TNF treatment were included in the analysis. The parameter used to evaluate the effectiveness of treatment was the Psoriasis Area and Severity Index (PASI). Patients who achieved a 75% improvement over their baseline PASI (PASI75) were considered responders to treatment. We collected PASI75 data at 3, 6 and 12 months. All patients signed a written informed consent document to participate in this study. The protocol and informed consent document complied with Spanish legislation on biomedical research and were approved by the Ethics Committee for Clinical Investigation of Hospital Universitario de la Princesa.

Sample processing and genotyping

DNA was extracted from peripheral blood samples (EDTA 3 ml) using the MagNa Pure System (Roche Applied Science, Indianapolis, IN, USA) and quantified in a NanoDrop ND-1000 Spectrophotometer (Wilmington, DE, USA). Samples were stored at –80 °C in the Clinical Pharmacology Service (Madrid, Spain).

A total of 173 polymorphisms were evaluated using the IlluminaVeracode genotyping platform (Human Genotyping Unit-CeGen, Madrid, Spain). A description of the SNPs studied is shown in Supplementary Table S1 published by Prieto-Pérez et al.14

Statistical analysis

Allele and genotype frequencies, Hardy–Weinberg equilibrium and linkage disequilibrium were analyzed using SNPStats.15 The univariate analysis to evaluate the SNPs and the treatment response was performed using R 3.0.2. (SNPassoc package)16 and SNPStats. We tested different logistic regression models (codominant, dominant, recessive and additive). The optimal model was selected using the lower Akaike information criterion. SNPs with P⩽0.05 in the univariate analysis were included in the multivariate analysis. Moreover, when several SNPs were in linkage disequilibrium and were significant in the univariate analysis, we analyzed the association between haplotype and response using SNPstats. Statistical significance was set at P⩽0.05. The results were expressed as the odds ratio (OR), 95% confidence interval and P-value. Efficacy data were also analyzed using ITT-LOCF (intention-to-treat last observation carried forward) method.17

The clinical variables compared between responders and nonresponders included age at onset of psoriasis, type of psoriasis (I or II), weight, gender, presence or absence of psoriatic arthritis and age at prescription of the first anti-TNF drug.

Results

Population

The study population included 144 patients (84 men and 60 women) with moderate-to-severe plaque psoriasis treated with anti-TNF drugs. The phenotypic characteristics of the patients are shown in Table 1. The mean age at onset of psoriasis was 28.51±14.04 years and the mean age when the first anti-TNF agent was prescribed was 43.61±15.02 years.

Table 1 Phenotypic characteristics of psoriatic patients
Full size table

The first anti-TNF drug was etanercept in 74 patients (51.39%), adalimumab in 42 (29.17%) and infliximab in 28 (19.44%) patients. Throughout treatment, almost all patients (97%) were treated as monotherapy. A total of 106 patients achieved a PASI75 response at 3 months of treatment (73.61%). Data for the evaluation of the clinical response are missing in some patients owing to loss of follow-up, side effects, remission, intermittent therapy and/or patient’s wish (Table 1). Before 6 months, 6 patients decided not to continue with the treatment and we missed PASI evaluation at 6 months owing to lack of follow-up. In addition, 3 patients stopped anti-TNF treatment at 3–4 months because of lack of efficacy and 2 patients because of side effects. Before 1 year, other 8 patients did not wish to continue with the therapy. Moreover, the treatment was withdrawn because of lack of efficacy (N=8, 7–9 months), side effects (N=3, 6–8 months) or remission (N=1, 8 months). Therefore, 11 and 31 patients did not reach the visits at 6 and 12 months, respectively, and we included 133 and 113 patients to evaluate effectiveness at 6 months and 1 year of treatment. Of the 133 patients evaluated at 6 months, 102 achieved PASI75 (76.69%); of the 113 patients evaluated at 1 year, 92 achieved PASI75 (81.42%).

We did not obtain significant results for the following clinical variables evaluated for response to the treatment: age at onset of psoriasis (P=0.620), type I or II psoriasis (P=0.740), weight (P=0.830), gender (P=0.950), presence or absence of psoriatic arthritis (P=0.920), age at prescription of the first anti-TNF drug (P=0.380), presence of comorbidities (P=0.530 for dyslipemia, P=0.300 for hypertension, P=0.510 for diabetes, P=0.190 for obesity) and prior therapies (P=0.390 for methotrexate, P=0.200 for cyclosporine, P=0.980 for acitretin, P=0.055 for phototherapy, P=0.200 for efalizumab).

Effectiveness

All significant results in the univariate analysis at 3 months (PASI75) are shown in Supplementary Table S1 (22 significant SNPs). Only 6 SNPs were significant in the multivariate analysis (Table 2). Of these, 4 were associated with a lack of achievement of PASI75: AG/GG in rs2916205 (PGLYRP4-24; OR=3.62), CC in rs9304742 (ZNF816A; OR=7.66), AA in rs11126740 (CTNNA2; OR=20.56) and AG/GG in rs2546890 (IL12B; OR=3.22). However, patients with the CT/CC allele in rs96844 (MAP3K1; OR=0.17) and the CT/TT allele in rs12191877 (HLA-C; OR=0.30) responded better to anti-TNF drugs (Table 2).

Table 2 Summary of the results of univariate and multivariate logistic regression analyses for PASI75 at 3 months of treatment (N=144)
Full size table

All significant results in the univariate analysis at 6 months (PASI75) are shown in Supplementary Table S2 (17 significant SNPs). Only 5 SNPs were significant in the multivariate analysis (Table 3): carriers of the CT/CC genotype for rs1801274 (FCGR2A; OR=13.32), CT/TT for rs6311 (HTR2A; OR=5.60) and AG/GG for rs2546890 (IL12B; OR=4.14) were poor responders. Nevertheless, carriers of the CT/CC genotype of rs96844 (MAP3K1; OR=0.24) and carriers of CT/TT in rs6908425 (CDKAL1; OR=0.14) achieved a better therapeutic response (Table 3).

Table 3 Summary of results of univariate and multivariate logistic regression analyses for PASI75 at 6 months of treatment (N=133)
Full size table

We also performed a multivariate logistic regression analysis for PASI75 at 1 year of treatment (N=113; Table 4 and Supplementary Table S3). We only obtained significant results for rs2546890 in IL12B: patients with the AG/GG genotype were more likely to be nonresponders (Table 4). Moreover, rs2546890 is in linkage disequilibrium with rs6887695 in IL12B (that was also significant in the univariate analysis for PASI75 at 1 year of treatment, P=0.014; Supplementary Table S3). The haplotype results showed an association between the GC haplotype (rs2546890 and rs6887695, respectively) and nonresponse to anti-TNF drugs at 1 year of treatment (Supplementary Table S4).

Table 4 Summary of results of univariate and multivariate logistic regression analyses for PASI75 at 1 year of treatment (N=113)
Full size table

The haplotype analysis (Supplementary Table S4) showed an association between nonresponders (PASI75) and the GACCT haplotype (rs2916205, rs821421, rs3006448, rs3006452 and rs3006457, respectively) in PGLYRP gene (P=0.028), TT haplotype (rs12191877 and rs10484554, respectively) in HLA-C (P=0.018) and TA haplotype (rs2073048 and rs2022544, respectively) in C6orf10 (P=0.032) at 3 months of treatment. None of these haplotypes were significant at 6 months or 1 year of treatment. However, analysis of 2 SNPs in FOXP3 (rs2280883 and rs3761548) revealed significant results in men at 6 months (TC haplotype; P=0.025; Supplementary Table S4).

The results with ITT-LOCF method showed 6 significant SNPs in rs1801274 (FCGR2A; OR=6.53), rs6311 (HTR2A; OR=11.36), rs2546890 (IL12B; OR=3.27), rs6908425 (CDKAL1; OR=0.12), rs6028945 (MAFB; OR=0.12) and rs10945919 (QK1; OR=5.13) at 6 months of treatment (supplementary Table S5). Furthermore, we obtained 2 significant SNPs in rs96844 (MAP3K1; OR=0.20) and rs191190 (TNFRSF1A; OR=3.30) at 1 year of treatment (supplementary Table S6).

Discussion

Polymorphisms in PGLYRP4-24 (rs2916205), ZNF816 (rs9304742) and CTNNA2 (rs11126740) have been associated with psoriasis.18, 19, 20, 21 However, the association between these SNPs and response to treatment has not yet been studied. Therefore, our study is the first to show an association between several SNPs and response to anti-TNF drugs in patients with moderate-to-severe plaque psoriasis.

PGLYRP4 encodes a protein involved in immune responses that binds to peptidoglycan and bacterial lipopolysaccharide.22 This protein is found at the PSORS4 locus that is within the epidermal differentiation complex and is expressed in epithelial cells in the skin.18, 22, 23 Several polymorphisms in PGLYRP4 seem to increase the risk of developing Crohn’s disease22 that is associated with psoriasis.2 In addition, Sun et al.18 did not find an association between this gene and moderate-to-severe psoriasis in single SNP-based or haplotype-based tests in a case–control study.18 We confirmed these results in a recent publication.14 However, when we studied SNPs in the PGLYRP4 gene, we found that carriers of the G allele of rs2916205 were 3.62 times more likely not to respond to anti-TNF drugs (Table 2). Moreover, we found a significant GACCT haplotype in the PGLYRP gene (rs2916205, rs821421, rs3006448, rs3006452 and rs3006457, respectively) that was associated with no response to anti-TNF drugs at 3 months (PASI75).

In addition, we describe for the first time that C allele of rs9304742 in ZNF816A gene was 7.66 times more likely not to respond to treatment (Table 2). This gene encodes a zinc-finger protein that could play a role in the recognition of specific proteins and have several regulatory functions.19 Similarly, the SNP rs11126740 is found in CTNNA2 and considerably affected the response to treatment (PASI75) at 3 months in the study population. A allele carriers were 20.56 times more likely not to respond to treatment (Table 2). This gene encodes a catenin-α2 (cadherin-associated protein) involved in cellular adhesion. Overexpression of basal layer cadherins results in abnormal keratinocyte differentiation, a typical feature of psoriasis.24 El-Wahed Gaber et al.24 performed a study of gene expression of β-catenin (protein that interacts with α-catenin during cell adhesion) and demonstrated deregulation of β-catenin in psoriatic skin Therefore, these proteins may play a key role in the maintenance of normal tissue architecture.

Furthermore, the frequency of T allele carriers for rs12191877 was 70.8% for responders and 52.6% for nonresponders at 3 months of treatment (P=0.027; Table 2). This SNP has also been associated with psoriasis and is in linkage disequilibrium with HLA-C*0602.25, 26 HLA-C plays a role in the immune system by presenting peptides derived from the lumen of the endoplasmic reticulum. HLA-C*0602 and late-cornified envelope (LCE; I carriers) genotypes have been associated with better response to anti-TNF drugs at 24 weeks (PASI75; N=116).7 In this sense, we found that carriers of the T allele for rs12191877 (HLA-C) were 3.33 times more likely to respond to anti-TNFs (Table 2). We also found a significant TT haplotype for rs12191877 and rs10484554, respectively, in HLA-C that was associated with response to anti-TNF agents at 3 months (PASI75). In contrast, the HLA-C*0602 allele has been associated with poor response to anti-TNF drugs.6 Moreover, the HLA-C*0602 allele has been associated with an increased and faster response to ustekinumab.27 Therefore, the HLA-C gene may play a relevant role in the development of the disease, as well as in the response to biological drugs. Ryan et al.5 compared the frequencies of HLA-C genotypes between responders and nonresponders and found no significant results.

The results obtained for rs2546890 (IL12B) and rs96844 (MAP3K1) were replicated in our study at 3 and 6 months of anti-TNF treatment. Moreover, we replicated the association for rs2546890 at 1 year. Thus, these SNPs could be predictors of anti-TNF response in the short and long terms. IL-12 is a proinflammatory cytokine involved in the lymphocyte T helper 1 pathway. This cytokine has been widely associated with psoriasis.28, 29, 30 Moreover, rs2546890 in IL12 has been associated with psoriasis.26 We previously evaluated the influence of polymorphisms in IL12B (rs6887695 and rs3212227) on the response to anti-TNF drugs6 but found no association. However, our results did indicate that carriers of the G allele of rs2546890 (IL12B) were 3.22, 4.11 and 2.79 times more likely not to respond to treatment at 3 months, 6 months and 1 year, respectively, than patients carrying the AA genotype (Tables 2, 3, 4). In addition, we found an association between the GC haplotype (rs2546890 and rs6887695, respectively) and response at 1 year of treatment (Supplementary Table S4). Furthermore, we selected a series of SNPs in our previous study (see Table 2 by Prieto-Pérez et al.2). Among the SNPs,2 we found an association between carriers of the C allele of rs96844 (MAP3K1) and better response to anti-TNF drugs at 3 and 6 months of treatment (5.88-fold and 4.17-fold, respectively; Tables 2 and 3). Bowes et al.31 also found an association between the minor allele (G) of rs96844 (MAP3K1) and a good response to anti-TNF drugs in patients with rheumatoid arthritis (RA; N=428).31 However, the authors did not validate their results in an independent cohort.31

In addition, we obtained significant results for rs1801274 in FCGR2A at 6 months of treatment. Prieto-Pérez et al.2 showed this protein to be associated with the response to anti-TNF therapy in RA32, 33 (see Table 2 of Prieto-Pérez et al.2). FCGR2A encodes Fc-γ receptor on the cell surface of macrophages and neutrophils. In our study, carriers of the C allele of rs1801274 were 13.32 times more likely to be nonresponders to anti-TNF drugs. Our results were consistent with those of Ramírez et al.34 who showed an association between the T allele (rs1801274) and a better response to anti-TNF drugs in patients with psoriatic arthritis34 that is also associated with psoriasis.2 In contrast, Dávila-Fajardo et al.35 found an association between the C allele and a better response to adalimumab in patients with RA at 14 weeks. Julià et al.12 analyzed rs1801274 (FCGR2A) but obtained no significant results for PASI75 at 12 weeks of treatment with anti-TNF drugs in psoriatic patients.

Other SNPs associated with the response to anti-TNF drugs at 6 months of treatment in our population were rs6311 (HTR2A) and rs6908425 (CDKAL1). rs6311 in HTR2A (that encodes a serotonin receptor) was associated with late-onset psoriasis in a Thai population.36 Psoriasis is characterized by high production of cytokines that could be inhibited in the presence of 2,5-dimethoxy-4-iodoamphetamine (selective serotonin receptor agonist).37 This inhibition was more pronounced in carriers of the T and C alleles of rs6314 and rs1328674, respectively, in patients with RA.37 Moreover, we found an association between psoriatic carriers of the T allele of rs6311 (HTR2A) and a worse response to anti-TNF drugs (5.60-fold). In addition, a gene–gene interaction was described for HTR2A and the major genetic risk factor for RA, HLA-DRB1.38 These results suggest that HTR2A may be associated with the immune system and could be a risk factor for psoriasis. Besides, rs6908425 (CDKAL1) has been described as a risk factor for psoriasis.39 Our results showed an association between carriers of the T allele of rs6908425 (CDKAL1), which has an unknown function, and better response to anti-TNF drugs (7.14-fold) at 6 months. Nevertheless, the influence of this SNP on the response of anti-TNFs has not been studied yet.

Other haplotypes associated with response to anti-TNF drugs were TA for rs2073048 and rs2022544, respectively, in C6orf10 (PASI75 at 3 months) and TC for rs2280883 and rs3761548, respectively, in FOXP3 (PASI75 at 6 months in men). SNPs in C6orf10 and FOXP3 genes were previously associated with psoriasis, but not with response to anti-TNF drugs.25, 40, 41, 42 FOXP3 is expressed by regulatory T cells and has been associated with some autoimmune disorders.41 Although the function of the protein C6orf10 is not known, it has been described as a potential downstream effector of TNFα.25

Our sample size was limited by the number of patients treated in the dermatology department. However, the limited sample size was counterbalanced by an exhaustive follow-up of patients and analysis of their data. Otherwise, we did not have enough statistical power to analyze each drug independently. We lost patients during the study at 6 and 12 months, but it is not clear how the problem that represents missing data may be addressed.43 Our data showed patient nonresponders at 3 months who were responders at 6 months and vice versa. For this reason, the main methodological approach used to analyze long-term efficacy data was ‘as-treated analysis’. The problem with using ITT-LOCF for analyzing long-term efficacy results is that this analysis method assumes that efficacy will remain constant (consistent with the last known value).17 Hence, the results showed at 6 and 12 months of treatment should be considered as preliminary results because of missing data.

The SNPs in FCGR2A, HTR2A, IL12B and CDKAL1 were replicated with ITT-LOCF analysis. In addition, SNPs in IL12B (rs2546890) and MAK3K1 (rs96844) were significant at 6 months and 1 year of treatment. Hence, these SNP remain being long-term biomarkers. Moreover, ITT-LOCF analysis showed 2 new SNPs in MAFB (rs6028945) and QK1 (rs10945919) (Supplementary Table S5; PASI75 at 6 months) and another one (rs191190) in TNFRSF1A (Supplementary Table S6; PASI75 at 1 year) associated with response to anti-TNFs. The SNP in TNFRSF1A has been previously associated with psoriasis,14, 44 but not with response to the treatment. Furthermore, the SNPs in MAFB and QKI have been shown as possible predictors of response to anti-TNFs in patients with RA.45

In conclusion, ours is the first study to show an association between several SNPs—rs2916205 (PGLYR4), rs9304742 (ZNF816A), rs11126740 (CTNNA2), rs2546890 (IL12B), rs96844 (MAP3K1) and rs12191877 (HLA-C)—and response to anti-TNFs in psoriatic patients (PASI75 at 3 months of treatment). All of the SNPs have regulatory functions and are involved in immune responses or differentiation of keratinocytes in the skin. Therefore, they affect the development of psoriasis and the response to anti-TNF drugs during the initial stages of treatment. The only SNPs replicated at 6 months were rs2546890 (IL12B) and rs96844 (MAP3K1). We also obtained significant results for rs2546890 (IL12B) at 1 year of treatment (PASI75). Finally, rs1801274 (FCGR2A), rs6311 (HTR2A) and rs6908425 (CDKAL1) were associated with response to anti-TNFs at 6 months and could be biomarkers of longer-term response. The results at 6 and 12 months of treatment should be considered carefully because of missing data.

Few studies have evaluated the effect of polymorphisms on the response to anti-TNF drugs.4, 5, 6, 7, 8, 9, 10, 11, 12 Consequently, our findings add to current knowledge on the pharmacogenomics of moderate-to-severe plaque psoriasis. However, large studies continue to be necessary to validate biomarkers before they are routinely applied in clinical practice.