Ankylosing spondylitis is a chronic inflammatory rheumatic disease with a strong genetic component determining both susceptibility to, and severity of, the disease.^1,2 Genes of the human major histocompatibility complex (MHC), in particular HLA-B27, are major genetic factors influencing the familial clustering of ankylosing spondylitis, as evidenced by linkage and association studies.³ The strong association between ankylosing spondylitis and HLA-B27 has been known since 1973^4,5; however, the exact genetic mechanism of the pathogenic link remains enigmatic. Whether homozygosity for HLA-B27 contributes any additional effect on the disease is unclear. Contradictory reports on the effect of homozygosity for HLA-B27 have been published,^6,7 and there is no consensus as to whether HLA-B27 homozygosity increases the susceptibility to, or severity of, ankylosing spondylitis.

Several haplotypic and association studies suggest that more than one gene within the human MHC may influence the susceptibility to ankylosing spondylitis. An association between HLA-DRB1*01 and spondyloarthropathy has been reported in British and Mexican populations,^8,9 while a recent Sardinian study reported an association between HLA-DRB1*15-B27 haplotype and ankylosing spondylitis.^10,11

The objective of this study was to clarify the HLA associations in ankylosing spondylitis in a Finnish population. The Finnish population has a restricted gene pool,¹² a property which can be useful in dissecting the relevant disease causing polymorphisms.¹³ We have investigated the influence of HLA-DRB1 alleles and HLA-B27 homozygosity/heterozygosity in susceptibility to, and clinical manifestations of, ankylosing spondylitis in a Finnish population.

METHODS

RESULTS

In this dataset, the male to female ratio was 2.1:1. One hundred and twenty six subjects (47%) also had AAU, 18 (7%) had inflammatory bowel disease, six (2%) had psoriasis, and 210 (78%) had peripheral arthritis. The mean (SD) age at symptom onset was 27 (9) years; age at diagnosis, 35 (10) years; diagnostic delay, 8 (8) years; disease duration, 24 (11) years; age at study, 51 (11) years; BASDAI, 5.2 (1.9); and BASFI, 3.9 (2.3).

DISCUSSION

Previous studies on the influence of HLA-B27 homozygosity in the development of ankylosing spondylitis have been contradictory. A significant excess of HLA homozygotes among cases of ankylosing spondylitis has been reported,⁶ but this has not been confirmed by others.⁷ The contribution of HLA-B27 homozygosity is likely to be modest and large samples sizes are required to investigate this issue. In other studies, homozygosity for HLA has been reported to be associated with susceptibility to autoimmune diseases,²⁶ common variable immunodeficiency,²⁷ and an increased difficulty in clearing infections.²⁸

In our family based cohort, HLA-B27 homozygosity was significantly increased from that expected under HWE. There are genetic theories that could explain our findings. The increased susceptibility associated with HLA-B27 homozygosity could be explained by the threshold model of polygenic disease, where the presence of increased number of susceptibility alleles increases the likelihood of developing the disease. Another explanation could be the possibility that some non-HLA-B27 alleles are relatively protective. There are also theoretical molecular mechanisms that could explain the findings. First, HLA-B27 homozygotes may be more likely to carry abnormal HLA-B27 molecules such as homodimers or misfolded proteins. Second, HLA-B27 homozygotes may express an increased level of HLA-B27 molecules. The latter is supported by the observation of greater expression of HLA-B27 molecules in patients with ankylosing spondylitis than in healthy controls.²⁹ However, it is also possible that the genotyping method employed failed to detect some HLA-B alleles, which would subsequently increase the number of apparent homozygotes. This could be overcome by performing full HLA-B locus genotyping for all the homozygous cases. It is estimated that the frequency of the alleles that are not detected by this method is very low. This is supported by the fact that no Mendelian inconsistencies were detected in these families.

A previous report has suggested that HLA-B27 homozygosity may influence disease severity.⁷ A greater frequency of involvement of peripheral joints has been reported among the HLA-B27 homozygotes.⁶ Surprisingly, this study noted a significant, albeit marginal, decrease in BASDAI among the HLA-B27 homozygotes, suggesting that HLA-B27 homozygote patients may have milder disease than HLA-B27 heterozygotes. It is of note that BASDAI aims to measure the actual disease activity, which may have been decreased by more aggressive disease modifying drug treatment in patients with most severe disease. None of the other disease severity indices showed significant differences. These findings are consistent with the whole genome disease severity screen, in which no linkage between the disease severity indices and the MHC region was observed.³⁰

It has been proposed that HLA-B27 positive and HLA-B27 negative ankylosing spondylitis represent a heterogeneous group of phenotypically similar diseases that may have different aetiopathogenic mechanisms. HLA-B27 negative ankylosing spondylitis is rarely familial and is associated with a later age of disease onset.^31,32 The current study showed a significant association between HLA-B27 and an earlier age of symptom onset and diagnosis in ankylosing spondylitis patients, confirming the previously reported associations. However, HLA-B27 positivity may be used as a diagnostic tool and therefore cases will be detected earlier. A recent report suggested that HLA-B27 is also associated with an earlier onset of psoriatic arthritis.³³HLA-B27 appears to alter the threshold for developing the disease at an earlier age, but the pathogenic mechanism of this process remains unclear. The large difference between the average ages of onset suggests that the underlying disease process between the HLA-B27 positive and negative patients may be different. In sub-Saharan Africa the epidemic of HIV infection has been associated with a dramatic increase in the prevalence of HLA-B27 negative spondyloarthritis,³⁴ suggesting that environmental factors may play a crucial role in the development of HLA-B27 negative spondyloathropathy. HLA-B27 negative patients could merely represent phenocopies of classical HLA-B27 associated disease. Certainly, it should not be assumed that HLA-B27 positive and negative ankylosing spondylitis is homogeneous with regard to their aetiology or pathogenesis.

Previous studies have found several distinct associations between HLA-DRB1 alleles. Our case–control analysis showed that HLA-DRB1*07-B27 haplotype frequency was marginally increased compared with the controls, but this may merely be a reflection of small sample size and random statistical fluctuation. The logistic regression analysis on HLA-DRB1 alleles conditioned on HLA-B27 noted no statistically significant difference between distinct HLA-DRB1 alleles and ankylosing spondylitis susceptibility. Because of the significant LD within the MHC, HLA-DRB1 alleles were conditioned on HLA-B27. Owing to the extreme polymorphic nature of the HLA-DRB1 locus and the expected modest effects of HLA class II genes on susceptibility to ankylosing spondylitis, large scale studies are required to dissect these effects. Because of limited sample size this study was underpowered to detect weak effects.

An association between HLA-DRB1*08 and juvenile ankylosing spondylitis has been reported in Norwegian and Mexican cases.^35,36 A weak association between HLA-DRB1*08 and early age at symptom onset is reported here, supporting these previous findings. Such findings strongly suggest that both HLA-B27 and HLA-DRB1*08, either independently or as a haplotype, contribute to the genetic susceptibility of early onset disease. Significant positive LD between HLA-B27 and HLA-DRB1*08 was observed, raising the possibility that the observed association is caused by linkage disequilibrium with HLA-B27. In this dataset neither the HLA-DRB1*08-B27 haplotype or HLA-DRB1*08-nonB27 haplotypes were associated with an earlier age of symptom onset, and thus it is unlikely that HLA-B27 alone is responsible for the observed association.

Early onset of symptoms did not correlate with BASDAI or BASFI, suggesting that even if disease occurs earlier in the presence of predisposing genes, it is not more severe. This is consistent with previous studies, indicating that although cases with juvenile onset of ankylosing spondylitis are more likely to develop hip arthritis, their disease is otherwise similar in character to cases with adult onset disease ankylosing spondylitis.³⁷HLA-DRB1*03 allele and HLA-DRB1*03-nonB27 haplotype were associated with a later age of symptom onset, suggesting HLA-DRB1*03—or another gene on the HLA-B27 negative chromosomal strand—may be involved in determining this. Overall, these findings suggest that HLA-DRB1 alleles may influence the age of symptom onset of ankylosing spondylitis. However, the reported p values are not corrected for multiple comparisons made, and the associations reported here may merely be attributable to stochastic statistical fluctuations.

AAU is more common in HLA-B27 positive than HLA-B27 negative cases with ankylosing spondylitis,^38,39 confirmed by this study. There is no consensus among previous studies concerning the association between HLA-DRB1*08 and AAU. An association has been reported in a Japanese population,⁴⁰ and negative findings have been reported in Norwegian and Mexican studies.^35,36 A possible explanation is that the juvenile ankylosing spondylitis may increase AAU. However, a recent study noted that the prevalence of iritis correlates positively with disease duration, but not with age of symptom onset.⁴¹ The present study noted no significant association between HLA-DRB1*08 and AAU, and thus does not support the Japanese finding. The discrepant associations may be due to different ethnic background of the populations.

The disease severity was assessed using a structured questionnaire, which imposes limitations on the method. Peripheral arthritis and other extra-articular manifestations of the disease were defined merely using the questionnaire. Systematic radiographic disease severity indices such as BASRI were not available. However, BASDAI and BASRI are closely correlated (Calin A, personal communication). Also, BASDAI and BASFI are highly heritable,² suggesting that the random fluctuation does not significantly reduce their accuracy. Future studies using radiographic disease severity indices are warranted for a better assessment of the effect of HLA genes on disease severity.

The complexity of the LD and the density of genes make the detection of the causative variants a challenging task in the HLA region. In the healthy Finnish population HLA-B27 has previously been reported to be in LD with HLA-DRB1*01, *04, and *08.⁴² Positive LD between HLA-B27 and HLA-DRB1*01 and *08, and negative LD between HLA-B27 and HLA-DRB1*02, *03, *11, and *13 was observed. The British and Finnish populations have a similar overall LD pattern.⁸ The finding of this study of a significant overrepresentation of HLA-B27-homozygotes—and of previous studies of overrepresentation of HLA-B60 in HLA-B27 positive and negative cases^1,43—have major implications for studies of MHC genes other than HLA-B in ankylosing spondylitis. The past practice of matching cases and controls for HLA-B27 antigen status is clearly not adequate, and future studies must match at least for HLA-B27 at the allelic level.

APPENDIX

The sequence based method involved eight group specific PCR amplifications with the primers HLA-DRB1*01 (5′CAGTGTCTTCTCAGGTGGCT), HLA-DRB1*15/16 (5′GGCCGCCTTGTGACCGGATG), HLA-DRB1*03/08/11/13/14 (5′GCCTCAGGAAGACAGAGGAG), HLA-DRB1*04 (5′CTTGGGATCAGAGGTAGATTTT), HLA-DRB1*07 (5′CGGCGTCGCTGTCAGTGTT), HLA-DRB1*09 (5′CAGTTAAGGTTCCAGTGCCA), HLA-DRB1*10 (5′CCCACAGCGTTCTTGGAGG), and HLA-DRB1*12 (5′AGTGTCTTCTCAGGACGCCA) prepared in a 50:50 mix with a generic reverse primer with an M13-21 sequencing tag (5′TGTAAAACGACGGCCAGTGCCGCTGCACTGTGAAGCTCTC). Amplification was carried out in a 10 μl reaction mixture containing 50 ng DNA, 10 mM Tris-HCl pH 8.3, 50 mM KCl, 1.5 mM MgCl₂ (for HLA-DRB1*01 and HLA-DRB1*03/08/11/13/14) or 2.5 mM MgCl₂, 25 μM of each dNTP, 0.25 U AmpliTaq Gold polymerase (Applied Biosystems, Warrington, UK), and 0.4 μM primers. The cycling conditions were as follows: 94°C for 14 minutes; 35 cycles of 94°C for 30 seconds, 55°C (for HLA-DRB1*15/16, HLA-DRB1*04 and HLA-DRB1*10) or 60°C for 30 seconds, and 72°C for 30 seconds. PCR products were separated on a 3% agarose gel stained with ethidium bromide and visualised under ultraviolet light. PCR products were subsequently sequenced using M13-21 Big Dye dye primer sequencing kit (Applied Biosystems) using following conditions: 15 cycles of 96°C for 10 seconds, 55°C for 5 seconds, 70°C for one minute, and 15 cycles of 96°C for 10 seconds and 70°C for one minute. The products were separated on 4.75% polyacrylamide gels using ABI 373 semiautomated sequencer (Applied Biosystems) and analysed using Sequencing Analysis version 3.0 and Factura version 2.0.1 (PE Applied Biosystems).