Article Text

Extended report
Human leucocyte antigen risk alleles for psoriatic arthritis among patients with psoriasis
  1. Lihi Eder1,
  2. Vinod Chandran1,
  3. Fawnda Pellet1,
  4. Sutha Shanmugarajah1,
  5. Cheryl F Rosen2,
  6. Shelley B Bull3,
  7. Dafna D Gladman1
  1. 1Centre for Prognosis Studies in the Rheumatic Diseases, Toronto Western Hospital, Toronto, Canada
  2. 2Division of Dermatology, Toronto Western Hospital, University of Toronto, Toronto, Canada
  3. 3Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Canada
  1. Correspondence to Dr Dafna D Gladman, University of Toronto Psoriatic Arthritis Clinic, Centre for Prognosis Studies in the Rheumatic Diseases, Toronto Western Hospital, 399 Bathurst Street, 1E-410B, Toronto M5T 2S8, Canada; dafna.gladman{at}


Aim Genes that differentiate patients with psoriatic arthritis (PsA) from those with cutaneous psoriasis (PsC) may serve as markers for the development of PsA in patients with psoriasis. The authors aimed to identify human leucocyte antigen (HLA) alleles that are associated with the development of PsA in patients with psoriasis.

Methods 712 adult patients with PsA, 335 adult patients with PsC and 713 healthy controls were genotyped for HLA-A, HLA-B, HLA-C, HLA-DR and HLA-DQ alleles. Differences in allelic distributions for each of the HLA loci were compared using a likelihood ratio test. Logistic regression analysis of multiple loci was performed to account for linkage disequilibrium. Haplotype information was inferred using the expectation–maximisation algorithm (given HLA-C and HLA-B genotypes) and analysed similarly.

Results The following HLA alleles were found to be significantly associated with patients with PsA compared to patients with PsC in multivariate regression analysis: B*08 (OR 1.61, p=0.009), B*27 (OR 5.17, p<0.0001), B*38 (OR 1.65, p=0.026) and C*06 (OR 0.58, p=0.0002). HLA-B*27, HLA-B*38 and HLA-C*06 frequencies were also significantly higher in patients with PsA than in healthy controls (B*27: OR 3.05, p<0.0001; B*38: OR 5.9, p<0.0001; HLA-C*06: OR 1.71, p<0.0001). The following haplotypes were independently associated with PsA compared to PsC: HLA-B*18-C*07 (OR 10.1, p=0.004), HLA-B*27-C*01 (OR 41.1, p<0.0001), HLA-B*27-C*02 (OR 19.9, p<0.0001), HLA-B*38-C*12 (OR 2.9, p=0.01), HLA-B*08-C*07 (OR 2.6, p=0.004) and HLA-B*57-C*06 (OR 0.5, p=0.03).

Conclusions Certain HLA-B and HLA-C alleles confer susceptibility to PsA among patients with psoriasis and may be used to identify patients with PsC who may develop PsA.

Statistics from

Psoriatic arthritis (PsA) can be viewed as a ‘disease within a disease’, as approximately 30% of patients with psoriasis develop PsA. One of the models that describe the relation between these diseases considers PsA as a more severe phenotype of cutaneous psoriasis (PsC) that occurs due to a greater number of susceptibility genes or environmental factors.

Studies have shown a strong association between PsC and the major histocompatibility complex (MHC) region,1,,5 particularly with a 300 kb segment in the MHC class I region on chromosome 6p21.3, known as PSORS1.6 7 Within this region, the largest and most consistently reported association is that with the human leucocyte antigen (HLA) C*068 allele, which is linked to earlier onset and more severe PsC.9 10

The MHC region, particularly HLA genes, has also been associated with PsA.11,,16 While HLA-C*06 is more frequent in patients with PsA than in the general population, this association is stronger with PsC than with PsA.17 In a case–control study that compared PsA to PsC, HLA-B*27and HLA-B*07 antigen frequencies were higher in PsA.11 HLA-DRB1*04 was reported to confer a risk for PsA,13 18 19 but several investigators subsequently found no such association.11 20 Evidence for the role of HLA in susceptibility to PsA has also been demonstrated by an increased sharing of HLA haplotypes among sibpairs concordant for PsA but not among those concordant for PsC.21

The great challenge in investigating the genetics of PsA is related to the difficulty in differentiating the genes that confer a risk for the joint disease from those that are associated with the cutaneous disease. Many previous genetic studies in PsA have used only healthy individuals as controls, preventing the determination of whether any significant association was related to the cutaneous disease or to joint manifestations of PsA. Additional limitations of previous studies include small sample sizes, potential misclassification of cases and controls and the lack of uniform classification criteria for the clinical definition of the PsA phenotype. Genes that differentiate patients with PsA from those with PsC can serve as markers for the development of PsA in patients with PsC. In this study, we aimed to identify HLA alleles that confer a risk for PsA in patients with psoriasis.


Study population

In this population-based case–control study, three groups of individuals from the same geographical region were compared. All participants were Caucasians (by self-report) from the Greater Toronto Area.

The PsA group included 712 adult patients with PsA who were recruited from the University of Toronto PsA cohort. All patients were carefully phenotyped by a rheumatologist and satisfied the Classification Criteria for Psoriatic Arthritis for the classification of PsA.22 These patients underwent assessment according to a standard protocol, which includes complete history taking, physical examination and laboratory evaluation at 6–12-month intervals and radiological assessment at 2-year intervals irrespective of symptoms. The latter includes radiographs of hands, feet, spine and sacroiliac joints.11 The PsC group included 335 patients with a diagnosis of psoriasis confirmed by a dermatologist. All of these patients were assessed by a rheumatologist to exclude those with inflammatory arthritis. Subjects were interviewed and examined according to a standardised protocol, as previously described.23 If there were clinical findings of inflammatory arthritis, enthesitis or spondylitis, the patient was excluded from the study. Where the diagnosis of PsA was not clear, imaging studies, including radiographs, ultrasound or MRI, were performed as indicated to investigate the nature of the abnormality. This process ensured that no participants had clinical inflammatory arthritis at enrolment. DNA from 713 healthy controls, as well as from ethnically matched unrelated healthy volunteers from the Toronto area, are available in the laboratory control biobank.

HLA typing

DNA was extracted from peripheral blood using a modified salting-out procedure (Gentra Puregene Blood Kit, Qiagen, Mississauga, Ontario, Canada). Extracted genomic DNA was amplified by PCR using locus-specific primers for each of the HLA-A, HLA-B, HLA-C, HLA-DR and HLA-DQ loci. PCR amplicons were identified by sequence-specific oligonucleotide probes using the reverse line blot technique (RELI SSO HLA typing kits, Invitrogen, Burlington, Ontario, Canada).24 Ambiguous results were resolved using sequence-specific primers (PCR-SSP).

Statistical analysis

Single-locus analysis

Since HLAs are expressed in a codominant manner, the frequency of individuals carrying at least one copy of each of the different HLA alleles was used for comparisons. Overall, the presence of each of 105 HLA alleles at 5 HLA loci was tested; of these, 11 rare alleles (frequency <1%) were excluded. A likelihood ratio test was used to assess the differences in allelic distribution between patients with PsA and patients with PsC, and between patients with PsA and healthy controls. The false discovery rate (FDR) approach was employed24 25 to account for multiple testing separately within each HLA locus. To account for multiple testing of five HLA loci, we applied a significance criterion of FDR-p<0.01 (p=0.05 divided by the number of tested loci).

Multi-locus analysis

After the initial comparisons of allele frequencies, multivariate logistic regression analysis was used to identify key genetic differences between PsA and PsC while accounting for LD(Linkage disequilibrium) between loci and dependence among allele frequencies within a locus. A targeted approach was employed to select variables for inclusion in the full regression model. Previous literature showed that alleles with the strongest PsA associations are within the HLA-B and HLA-C loci. Therefore, in a screening step, a separate regression model for each of these two HLA loci was constructed by backward selection. In the second step, a single full logistic regression model was constructed for both loci. This model included the following alleles: HLA-B*27, HLA-B*57, HLA-B*08, HLA-B*38, HLA-C*01, HLA-C*02, HLA-C*06, HLA-C*07, HLA-C*12 and HLA-DRB1*07. A stepwise logistic regression was used to identify HLA alleles occurring more or less frequently in patients with PsA compared to patients with PsC. Allelic associations that were considered statistically significant were retained in the multivariate regression if the p value from the two-sided Wald test was less than 0.05. Statistical analysis was performed using the statistical software SAS (version 9.2).

The association between HLA and subphenotypes of PsA

As in many complex diseases, there are large variations in the phenotype of PsA. It has been shown that refining the phenotype may strengthen genetic associations by decreasing the heterogeneity of the cases.25 Therefore, the PsA group was subdivided into mutually exclusive axial and peripheral PsA subgroups, and each group was then compared separately to patients with PsC. Axial PsA was defined by radiological evidence of either bilateral grade 2 (at least) sacroiliitis or unilateral grade 3 or grade 4 sacroiliitis at any point during the follow-up, with or without peripheral arthritis. Peripheral PsA was defined based on clinical and radiographical evidence of peripheral joint involvement and no evidence of axial involvement, as defined above. Patients with unilateral grade 2 sacroiliitis were excluded from the analysis. A statistical approach similar to that used for multi-locus analysis was employed to identify the association of HLA alleles with subtypes of PsA.

Since HLA-C*06 is strongly associated with age at onset of psoriasis,9 the PsA and PsC groups were subdivided into patients with early-onset psoriasis (age ≤40 years) and patients with late-onset psoriasis (age >40 years). The frequencies of HLA-C*06 were compared in these subgroups using the likelihood ratio test.

Haplotype analysis

Haplotype analysis was performed only for HLA-C and HLA-B alleles to confirm the single-locus findings because these loci have shown the strongest association with PsA in single-locus analysis. Haplotype information was inferred among patients with PsA and patients with PsC using the expectation–maximisation algorithm (SAS-Genetics),26 generating maximum likelihood estimates given a multi-locus sample of HLA-C and HLA-B genotypes.27 Posterior probabilities of all possible haplotypes for an individual—conditional on the observed genotypes—were estimated. The resulting posterior probabilities made it possible to test haplotype–disease association while accounting for uncertainty in haplotype estimation. Because of the imprecision involved in estimating the effects of low-frequency haplotypes, only those with an estimated frequency of more than 1% in the study population were included in the analysis. Stepwise logistic regression was used to identify haplotypes occurring at higher or lower frequencies in PsA compared to PsC. Haplotype associations were considered statistically significant if the p value from the two-sided Wald test was less than 0.05.


The demographic and clinical characteristics of the study population are presented in table 1.

Table 1

Demographic and clinical characteristics of the study population

The patients with psoriasis were slightly older than the patients with PsA. Unexpectedly, the patients with PsA had more severe psoriasis, as measured by the maximal Psoriasis Area Severity Index score in the first 3 years of follow-up. This finding may be partially explained by the fact that patients with psoriasis who were recruited from phototherapy centres were evaluated subsequent to phototherapy treatments that most likely improved their psoriasis. However, it is possible that patients with PsA have more severe psoriasis, as has been previously reported.28 There was no difference in the frequencies of a family history of psoriasis; however, more patients in the PsA group reported a family history of PsA, ankylosing spondylitis or inflammatory bowel disease.

Association between HLA alleles and PsA: single-locus analysis

Allele frequencies in the PsA group were compared to those in the PsC group to identify susceptibility alleles that confer a risk for PsA. Similar comparisons were made between the PsA group and the healthy controls. The complete results are presented in supplementary tables 1.1–1.5. Significance criteria of FDR-p<0.01 were applied.

The following HLA alleles were more frequent in PsA than in PsC: HLA-B*27 (OR 5.07, p=2×10−10, FDR-p<0.0001), HLA-C*01 (OR 2.55, p=0.0009, FDR-p=0.0036) and HLA-C*02 (OR 2.41, p=0.0005, FDR-p=0.003), while HLA-C*06 (OR 0.48, p=8×10−8, FDR-p<0.0001), HLA-B*57 (OR 0.58, p=0.001, FDR-p=0.01) and HLA-DRB1*07 (OR 0.62, p=0.0004, FDR-p=0.005) were less frequent in PsA than in PsC.

Association between HLA alleles and PsA: multivariate analysis

In a multivariate regression analysis, the following alleles were more frequent in PsA than in PsC: HLA-B*27 (OR 5.17, p<0.0001), HLA-B*08 (OR 1.61, p=0.009) and HLA-B*38 (OR 1.65, p=0.026) (table 2). Two of these alleles were also more frequent in patients with PsA than in healthy controls (single-locus analysis): HLA-B*27 (OR 3.05, p=2×10−11, FDR-p<0.0001) and HLA-B*38 (OR 5.9, p=5×10−14, FDR-p<0.0001). In contrast, HLA-C*06 was less frequent in the PsA group than in the PsC group (OR 0.58, p=0.0002) but was more frequent in the PsA group than in healthy controls (single-locus analysis) (OR 1.71, p=2×10−5, FDR-p<0.0001). The remaining HLA alleles were not found to be associated with PsA in the multivariate regression, suggesting that they were part of extended haplotypes with the more strongly associated alleles.

Table 2

ORs comparing HLA allele frequencies in PsA (n=710) versus PsC (n=334) using logistic regression analysis

Subgroup analysis based on age at onset of psoriasis

The association between PsA and HLA-C*06 remained significant in the early-onset-psoriasis group (table 3). These findings suggest that the lower frequency of HLA-C*06 in the PsA group compared to the PsC group is not related to differences between the two groups in the age at onset of psoriasis. In the late-onset-psoriasis group, the frequency of HLA-C*06 was slightly lower in the PsA group than in the PsC group, but not significantly lower, possibly due to the small number of individuals in each group.

Table 3

ORs comparing HLA-C*06 allele frequencies in PsA (n=687) versus PsC (n=334) by age at onset of psoriasis*

The association of HLA alleles with axial and peripheral PsA

A subgroup analysis was employed to identify new HLA alleles associated with subtypes of PsA (tables 4 and 5 and supplementary tables 4.1–5.1). One additional HLA allele was identified as being associated with a PsA subtype but not with the whole group of patients with PsA. After multivariate analysis, the frequency of HLA-B*39 was found to be increased in axial PsA compared to PsC (OR 2.51, 95% CI 1.25 to 5.01, p=0.009). This allele was not associated with peripheral PsA (OR 0.9, 95% CI 0.43 to 1.87, p=0.77). The remaining alleles associated with axial PsA were HLA-B*27, HLA-B*08 and HLA-B*38. Sixty-one patients with PsA had radiographical syndesmophytes that did not satisfy the definition of sacroiliitis, as described above (19 had grade 2 unilateral sacroiliitis, 42 had grade 1 or grade 0 sacroiliitis). The inclusion of these patients in the axial PsA group (316 patients overall) did not change the results.

Table 4

ORs comparing HLA allele frequencies in axial PsA (n=255) versus PsC (n=334) using logistic regression analysis

Table 5

ORs comparing HLA allele frequencies in peripheral PsA (n=323) versus PsC (n=334) using logistic regression analysis

Analysis of patients with peripheral PsA did not reveal new associations, as only HLA-C*06 and HLA-B*27 met significance criteria for association with this subgroup compared to PsC.

The association between HLA-B and HLA-C haplotypes and between PsA and PsC

A total of 107 possible haplotypes were estimated; however, due to low frequencies, only 23 common haplotypes were included in the analyses. The following haplotypes remained significantly associated with PsA compared to PsC after multivariate analysis (table 6 and supplementary table 6.1): HLA-B*18-C*07, HLA-B*27-C*01, HLA-B*27-C*02, HLA-B*38-C*12, HLA-B*08-C*07 and HLA-B*57-C*06. All of these haplotypes include the risk alleles that were identified in the single-locus analysis, with the only exception being HLA-B*18. For most of the haplotypes, the LD between loci was high and ranged from 0.6 to 0.98, illustrating the difficulty of investigating the independent effects of each of the HLA alleles.

Table 6

ORs comparing HLA-B/HLA-C haplotype frequencies in PsA (n=710) versus PsC (n=334) using logistic regression analysis


The MHC region on chromosome 6 and, in particular, the HLA class I region has been consistently associated with both PsC and PsA.11,,16 In this large case–control study, we aimed to identify PsA-specific genetic markers among the HLA alleles. We confirmed an association between HLA-B*27 and PsA that has been previously reported in a case–control comparison. Furthermore, HLA-C*06 was found to be associated with PsA compared to healthy controls and was significantly less frequent in patients with PsA than in those with PsC. Two additional HLA alleles, HLA-B*08 and HLA-B*38, were identified as potential genetic markers for PsA in patients with PsC. HLA-B*39 may be a potential marker for axial PsA.

We have previously assessed the association between HLA and PsA versus PsC using serological methods.11 In that study, HLA-B27 and HLA-B7 antigens conferred an increased risk for PsA, while HLA-B17 (split into HLA-B*57 and HLA-B*58), Cw6 and DR7 occurred with a lower frequency among patients with PsA than among those with PsC. HLA-B38 and HLA-B39 were associated with polyarthritis, while HLA-B27, Cw2 and DRw52 were associated with axial involvement. In the present study, we were able to replicate several of these associations using an independent sample, this time using molecular techniques to detect HLA alleles. It should be noted that only 50 of the patients with PsA had been included in the previous study, but none of the patients with psoriasis or the controls had been previously analysed.

In this report, we have shown that HLA-B*27 is a strong genetic marker for PsA compared to PsC. The absence of a detectable association between HLA-B*27 and the PsC group compared to healthy individuals suggests that HLA-B*27 is not a marker for skin disease. Previous studies in different ethnic groups have consistently shown that HLA-B*27 is an independent risk allele for PsA that is unrelated to the skin disease.11 29,,31 However, the prevalence rate of HLA-B*27 among patients with PsA in our study was 19.2% and ranged from 20% to 35% in previous studies—clearly much lower than its prevalence rate in ankylosing spondylitis (80–95% of the patients).32 Therefore, this allele can only account for a small proportion of the total genetic risk of PsA. Several studies have shown an association between HLA-B*27 and axial involvement in PsA.29 33 34 It has also been suggested that HLA-B*27 is associated with an earlier onset of joint manifestations in patients with PsC compared to non-carriers.34 Therefore, HLA-B*27 can be considered as the strongest HLA risk allele for PsA among patients with PsC and does differentiate the two conditions.

Interpretation of the association between HLA-C*06 and PsA is more challenging, since this allele confers a strong risk for PsC.8 35 36 In the present study, the frequency of HLA-C*06 was significantly higher in both patients with PsC and patients with PsA than in controls. However, HLA-C*06 frequency was significantly lower in patients with PsA than in patients with PsC. The association remained significant in the subgroup of patients with early-onset PsC, but not among those with late-onset disease. This appears to be inconsistent with the report of Ho et al,17 who analysed the association between HLA-C*06 and PsA. They concluded that HLA-C*06 was associated only with early-onset psoriasis and does not confer any additional susceptibility to PsA. However, no direct comparison between PsA and PsC was performed. Other studies have found a lower prevalence rate of HLA-C*06 among patients with PsA than among patients with PsC.31 37 In addition, psoriatic nail involvement, a clinical marker for PsA, is more common among patients with PsC who are negative for HLA-C*06.9 Although HLA-C*06 has not been associated with a particular clinical subtype of PsA, it has been found to increase the psoriasis–arthritis latency period.38 HLA-C*06 was also associated with a milder form of arthritis among patients with PsA. Ho et al39 reported that patients with PsA who carried HLA-C*06 had fewer damaged joints and lower active joint counts. The strong association of HLA-C*06 with PsC, despite its lower frequency in PsA than in PsC, is challenging to explain. HLA-C*06 may be involved in different mechanisms that lead to the joint disease and the skin disease—for example, antigen presentation and activation of the adaptive immune system for the skin and inhibition of the innate immune system through interaction with killer cell immunoglobulin-like receptors on natural killer cells in the joints. Another explanation may be related to the genetic heterogeneity of psoriasis. It is possible that different HLA alleles are associated with a subtype of psoriasis that is more likely to develop PsA. Therefore, although the phenotype of psoriasis is similar, one HLA allele is associated only with skin disease (HLA-C*06), while the other is associated with both skin and joint involvement. A potential candidate is HLA-C*12, which is associated with both PsA and PsC and had been reported in the past to be associated with PsC among Caucasians.5

Two additional alleles were significantly associated with PsA compared to PsC in multivariate analysis: HLA-B*08 and HLA-B*38. HLA-B*38 was strongly associated with PsA compared to both PsC (multivariate model) and controls (univariate model). HLA-B*38 has been reported to be more frequent among non-Ashkenazi Jews with PsA versus PsC and healthy controls in a small study performed in Israel.40 Another small study in Maryland also reported higher frequencies of HLA-B*38 in PsA than in PsC.37 Additionally, HLA-B*38 has been associated with more peripheral joint involvement.41 However, in the subgroup analysis of our study, an association of HLA-B*38 was detected in the group with axial PsA, but not in the group with arthritis restricted to the peripheral joints. HLA-B*38, along with HLA-C*12, is part of a common haplotype. In our study, this haplotype was associated with PsA compared to PsC. This haplotype has been reported to be present in up to 8% of the Jewish population and in 1.3–3.4% of non-Jewish Caucasians.42 In our study, the proportion of patients with Jewish ancestry was similar in the PsA and PsC groups (9.7% and 14%, respectively); therefore, it is unlikely that population stratification bias explains this association. We did not have information about Jewish ethnicity for most of the healthy controls. However, the proportion of HLA-B*38 among Jewish patients with PsA and with PsC were 39% and 34%, respectively. These figures are much higher than the reported prevalence rate of that allele in the Ashkenazi Jewish population (7–16%),42 suggesting that the association found between HLA-B*38 and PsA, compared to the control group, is valid and not explained by a low proportion of Jews in the latter group.

HLA-B*08 has not been previously reported to be associated with PsA. In our study, HLA-B*08 frequencies were higher among patients with PsA than among patients with PsC, but not different from healthy controls. In the literature, the extended haplotypes across HLA class I and class II genes—A*01-C*07-B*08-DRB1*03-DQA1*05-DQB1*02-DPA1*01-DPB1*04—have been associated with increased tumour necrosis factor (TNF) production following rubella vaccination.43 TNF is a major pro-inflammatory cytokine in PsA, and rubella vaccine was suggested to be a risk factor for the development of PsA among patients with psoriasis.44 HLA-B*08 is part of an ancestral haplotype (8.1 AH) that also includes HLA-A*01-C*07-DRB1*03 and TNF-308A.42 There have been conflicting results with regard to the association between PsA and TNF-308A polymorphism, which is part of 8.1 AH.45,,47 In summary, HLA-B*08 may be an independent genetic marker for PsA among patients with psoriasis. However, its strong LD with other HLA alleles and relevant genes in the MHC region precludes a more confident conclusion.

The utility of HLA risk alleles in identifying patients with psoriasis who have a higher risk of developing PsA can be estimated using the positive predictive value (PPV) for each of the identified HLA risk alleles. Assuming a 30% prevalence rate of PsA in patients with psoriasis, the PPV of the identified HLA risk allele was calculated as follows: HLA-B*08, PPV=0.42; HLA-B*27, PPV=0.64; HLA-B*38, PPV=0.43; HLA-C*06, PPV=0.35. Although the PPV is only moderate for several of the HLA alleles, in a population with a higher pretest probability of PsA, such as patients with psoriasis with non-specific musculoskeletal symptoms, the PPVs of these HLA alleles are likely higher. However, more information from prospective studies is required.

In summary, in this study, the independent association between HLA-B*27 and PsA was confirmed. Several additional alleles (C*06, B*38 and B*08) are potential genetic markers for PsA in patients with PsC and warrant further investigation.


Supplementary materials

  • Web Only Data

    Files in this Data Supplement:


  • Funding The University of Toronto PsA program was supported by a grant from the Krembil Foundation, as well as by The Arthritis Society SPARCC National Research Initiative. Dr Eder was supported by a fellowship grant from the Canadian Arthritis Network and by an Abbott PsA Fellowship. Dr Chandran was supported by a Canadian Institutes of Health Research—Clinical Research Initiative Fellowship and the Krembil Foundation.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the University Health Network.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.