Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Ankylosing spondylitis susceptibility loci defined by genome-search meta-analysis


In genome scans of ankylosing spondylitis (AS), with the exception of the HLA loci, linkage has not been easy to replicate across studies. We applied the genome-search meta-analysis (GSMA) method to genome scans of AS and spondyloarthropathy (SpA) to assess evidence for linkage across studies. Three AS genome scans and one SpA scan including 430 families with 1,048 affected individuals were used. All four original genome scans mainly analyzed Caucasian families. Seven bins had both Psumrnk and Pord<0.05, suggesting these bins most likely contain AS-linked loci; bin 6.2, 6.1, 6.3, 16.3, 19.2, 17.1, and 16.4. The GSMA produced significant genome-wide evidence for linkage on chromosome 6p22.3–6p21.1 (Psumrnk=0.000003), including the HLA locus. In addition to the HLA-B27 locus, strong linkage evidence was found on chromosome 6p25.3–6p22.3 (Psumrnk=0.0013) and 6p21.1–6p15 (Psumrnk=0.043). In the GSMA of four genome scans including one SpA study, the bin 9.4 (9q21.32–9q33.1) was newly found for linkage (Psumrnk=0.043, Pord=0.013). This GSMA added the evidence of the HLA loci as the greatest susceptibility factor to AS and showed evidences of chromosome 6, 16q, 19, 17p, and 9q as non-HLA susceptibility loci.


Ankylosing spondylitis (AS) is a chronic inflammatory disorder characterized by inflammation in the spine and sacroiliac joints causing initial bone and joint erosion and subsequent ankylosis (Brown et al. 2002). The peripheral joints and the entheses are frequently involved, and inflammation may involve extra-articular sites such as the uvea, aorta, heart, lungs, and kidneys. AS is the second-most common cause of inflammatory arthritis worldwide, with a prevalence of 0.2–0.9% in white populations (Braun et al. 1998).

Strong genetic factors are implicated in the etiology of the disease. The sibling recurrence risk ratio has been known to be 82 (Brown et al. 2000), and twin studies estimate that disease heritability exceeds 90% (Brown et al. 1997). The human leukocyte antigen (HLA) B27 is the first genetic factor identified in AS and confers the greatest susceptibility to AS (Brewerton et al. 1973). However, there is increasing evidence that non-HLA genes also contribute to AS susceptibility (Breban et al. 2003). Although the prevalence of AS is correlated with the prevalence of the HLA-B27, only 1–5% of B27-positive individuals develop AS (Calin et al. 1983), and HLA-B27 explains less than 50% of the total genetic risk for AS (Brown et al. 1998). The remaining gene loci have yet to be identified. Previous AS genome scans have shown the HLA loci, including HLA-B27, as the strongest linkage region and non-HLA loci as the additional susceptibility regions (Brown et al. 2003; Laval et al. 2001; Miceli-Richard et al. 2004; Zhang et al. 2004). However, with the exception of the HLA loci, linkage has not been easy to replicate across the studies. This is unsurprising because for rare genes, or genes with a weak effect, nonparametric linkage studies are likely to require several hundreds, possibly thousands, of affected sibling pairs to provide sufficient power to detect linkage (Risch and Merikangas 1996). There have been three genome scans of AS and one genome-wide study of spondyloarthropathy (SpA). SpA includes a spectrum of related disorders comprising the prototype AS, psoriatic arthritis, reactive arthritis, arthritis associated with inflammatory bowel disease, and undifferentiated SpA. Until now, no systematic statistical assessment of previous results has yet been carried out in AS and SpA.

Meta-analysis combines the linkage results from several studies, providing greater statistical power. A genome scan meta-analysis is one step toward defining genomic regions that harbor disease susceptibility loci and can identify regions that may contain disease genes in the pooled studies and identify regions where the genetic effect is too small to be detected in an individual study. Recently, the genome-search meta-analysis (GSMA) method has been proposed as a robust genome scan meta-analysis technique (Levinson et al. 2003). We applied the GSMA method to four genome scans of AS and SpA to assess evidence for linkage across the studies using published data. In addition, in order to confirm the significance provided by the GSMA, we applied Fisher’s method to the bins that are most likely to contain linked loci as an alternative method (Guerra et al. 1999).

Materials and methods

Selection of genome scans

Four genome scans were identified though MEDLINE search (Brown et al. 2003; Laval et al. 2001; Miceli-Richard et al. 2004; Zhang et al. 2004). Three genome scans were performed in AS (Brown et al. 2003; Laval et al. 2001; Zhang et al. 2004) and one was studied in SpA (Miceli-Richard et al. 2004). They consisted of three Caucasian and one mixed ethnicity population including 430 families (579 sibling pairs) with 1,048 affected individuals (Table 1). All four original genome scans mainly analyzed Caucasian families. The GSMA method assumes a uniform map in each scan, so we did not consider the second stages of genome scans where candidate regions were more densely mapped or samples were modified. Linkage data such as LOD or NPL scores were obtained from the result tables or linkage graphs in each study for the rank-ordering procedure. LOD scores were used in two genome scans (Brown et al. 2003; Laval et al. 2001), and NPL scores were selected in two genome scans (Miceli-Richard et al. 2004; Zhang et al. 2004). We performed GSMA in three AS genome scans and four genome scans including three AS and one SpA genome scans, respectively.

Table 1 Linkage loci in individual genome scans

Genome scan meta-analysis

The GSMA was performed as described (Wise et al. 1999). In brief, the autosomes were divided into 120 30-cM bins defined by Genethon markers (CEPH-Genethon Integrated Map Web site). On the Marchfield map, the average bin width was 29.1 cM. Each marker was placed within one of these bins on the basis of its location on the Genethon or Marshifield map (available at For each study, each bin was assigned a within-study rank (Rstudy) based on the maximum linkage score within the bin. Bins were ranked in descending order (120 = most significant result). The summed rank across studies was computed for each bin (Rsumrnk). A weighted GSMA was carried out to allow results to reflect the relative contribution of each study. For the weighted analysis, each Rstudy value was multiplied by its study’s weight \(\left( {\sqrt {N[{\text{affected cases}}]} } \right),\) divided by the mean of this value over all studies. Two point-wise P values were determined, Psumrnk and Pord, as described and determined by 10,000 permutations of the weighted data set. Psumrnk is the probability of observing a bin’s summed rank by chance, and Pord is the probability of observing the jth place bin’s summed rank in jth place bins in randomly permuted data. The empirical criteria for bins most likely to contain linked loci were both Psumrnk and Pord <0.05 and the criterion for genome-wide significance was Psumrnk<0.000417 (0.05 corrected for 120 bins). Therefore, we considered the GSMA results to most likely contain linked loci if both Psumrnk and Pord were <0.05 and to have a genome-wide evidence of linkage if the GSMA Psumrnk was <0.000417.

Fisher’s method for combining P values

As an alternative, we also utilized Fisher’s method for combining P values (Guerra et al. 1999). Fisher’s method is a procedure of combining P values for the evaluation of several independent tests of the same null hypothesis. Under the null hypothesis of no-linkage P values, p1, p2, ...., pn from n, independent studies are uniformly distributed on the interval (0, 1) and the combination of P values, −2 Σ i=1, n ln pi, is distributed as a χ2 random variable with 2n° of freedom (df). P values are gained by the formula \( - \Phi {\left[ {{\text{sign(LOD)}}{\sqrt {21{\text{n(10) $|$ LOD $|$ }}} }} \right]}\) (Φ is the distribution function for standard normal distribution) or chi-square distribution of NPL score with 1 df. Unfortunately, using the original Fisher’s method to some of the genome scan data can introduce a bias in the distribution. This is because most of the model-free linkage methods produce one-tailed LOD scores that truncate all at LOD=0. To overcome this, we used a P value of 0.72 for LOD=0 (Province et al. 2001).


Each genome scans

Individual genome scans have shown significant linkages in HLA and non-HLA regions. Table 1 shows significant loci in each genome scan. All studies have shown HLA loci as the strongest linkage region. The highest linkage loci outside chromosome 6 were different across four studies. Chromosome 9q34.2, 19q13.31, 11q24.1, and 13q14.2 were the most significant loci outside of chromosome 6 in each study, respectively (Table 2).

Table 2 Characteristics of genome scans included in the meta-analysis. AS ankylosing spondylitis, SpA spondyloarthropathy

Genome-search meta-analyses in AS

In order to increase homogeneity for genetic linkage studies, first of all, we performed the GSMA in AS genome scan studies. Figure 1 shows the summed ranks for each bin giving Psumrnk. The summed ranks (vertical axis) are plotted against the bin location by a single point plotted for the summed rank for each bin with chromosome numbers (horizontal axis). A total nine bins lie above 95% confidence level (P=0.05), and five bins are above 99% confidence level (P=0.01). Table 3 summarizes the highest 10% of bins when ordered by summed rank. Out of nine bins (Psumrnk<0.05), seven had both Psumrnk and Pord<0.05, suggesting these bins most likely contain AS-linked loci: bins 6.2, 6.1, 6.3, 16.3, 19.2, 17.1, and 16.4.

Fig. 1

Genome-search meta-analysis (GSMA) results of genome scans in ankylosing spondylitis (AS) (a), and AS and spondyloarthropathy (SpA) (b). Individual chromosomes were subdivided into 30 cM bins (represented by a dot), and bins were ranked by the significance after summing weighted data across the studies. Significance levels corresponding to 99% (Psumrnk<0.01) and 95% (Psumrnk<0.05) are shown by the horizontal lines

Table 3 Psumrnk and Pord for bins with the highest-summed ranks (P<0.1)

The strongest evidence for linkage based on Psumrnk of the AS GSMA occurred on chromosome 6p22.3–6p21.1 (bin 6.2) containing the HLA-B27 region. The AS GSMA produced significant genome-wide evidence for linkage on this chromosome 6p22.3–6p21.1 (P=0.000003). The nominally significant Psumrnk of the adjacent bins, 6.1 and 6.3, provided additional evidence for linkage. The significant linkage evidence was found on 6p25.3–6p22.3 (Psumrnk=0.0013) and 6p21.1–6p15 (Psumrnk=0.043) in the adjacent bins. It is likely that the apparent linkage to bins 6.1 and 6.3 merely indicates the linkage of AS to HLA, and the evidence does not indicate a possibility of the existence of other loci near HLA. There may be also other loci on chromosome 6, but it is not supported by the evidence obtained.

The significant linkages on non-HLA loci were found on chromosome 16q12.2–16q23.1 (Psumrnk=0.0049), 19p13.2–19q13.43 (Psumrnk=0.0068), 17p13.3–17p12 (Psumrnk=0.035), and 16q23.1–16q24.3 (Psumrnk=0.036). All had both Psumrnk and Pord<0.05.

Genome search meta-analyses in four studies, including three AS and one SpA

We performed GSMA in all four studies, including one SpA genome scan. Ten bins lie above 95% confidence level (P=0.05), and three bins are above 99% confidence level (P=0.01) (Fig. 1). Table 3 summarizes the highest 10% of bins when ordered by summed rank. Out of the ten bins (Psumrnk<0.05), seven had both Psumrnk and Pord<0.05, suggesting these bins most likely contain linked loci: bin 6.2, 6.1, 6.3, 19.2, 16.3, 17.1, and 9.4. Bin 9.4 (9q21.32–9q33.1) was newly found for linkage in the GSMA of four genome scans, including one SpA study (Psumrnk=0.043, Pord=0.013). The strongest evidence for linkage based on Psumrnk of the GSMA occurs on chromosome 6p22.3–6p21.1 (bin 6.2) containing the HLA-B27 region. This region met a genome-wide significance for linkage (P=0.0000008). The second-highest summed ranks were assigned to bins 6p25.3–6p22.3 (Psumrnk=0.0001) and 6p21.1–6q15 (Psumrnk=0.0016). The next significant linkages were found on chromosomes 19p13.2–19q13.43 (Psumrnk=0.01), 16q12.2–16q23.1 (Psumrnk=0.0368), 17p13.3–17p12 (Psumrnk=0.038), 19p13.2 –19p13.43 (Psumrnk=0.01), and 9q21.32–9q33.1 (Psumrnk=0.043). All had both Psumrnk and Pord<0.05.

Fisher’s method for combining P values

Fisher’s method of the combination of probabilities from test of significance is a used tool for the synthesis of linkage evidence across studies. We applied Fisher’s combined P value method to all of the bins that met an empirical criteria of both Psumrnk and Pord<0.05: bins 6.2, 6.1, 6.3, 16.3, 19.2, 17.1, 16.4, and 9.4. Fisher’s combined P values support the significant GSMA P values of the bins for linkage (Table 3).


Family and twin studies of AS have demonstrated that predisposition to the disease was not exclusively related to HLA-B27, suggesting that additional susceptibility genes are expected in AS (Jarvinen. 1995). Previous AS and SpA genome-wide studies have shown linkages in HLA and non-HLA regions (Brown et al. 2003; Laval et al. 2001; Miceli-Richard et al. 2004; Zhang et al. 2004). Our meta-analysis provides more evidence confirming that HLA-B27 is a major genetic factor to susceptibility to AS. The highest evidence for linkage was observed in the HLA loci, and the linkage obtained a genome-wide significance (P=0.000003). The nominally significant Psumrnk of the adjacent bins, 6.1 and 6.3, provided additional evidence for linkage here. This might be shown because in simulated data, bins adjacent to those containing disease loci often also achieve nominal significance (Cordell et al. 2001), or this finding may suggest the possibility that genes on non-HLA chromosome 6 regions also may play an important role in susceptibility to AS. Beside confirming linkage of the HLA region, the GSMA found evidence for linkage at non-HLA loci such as chromosomes 16q, 19, and 17p. A recent simulation study based on the schizophrenia GSMA indicated that bins with Psumrnk and Pord<0.05 had the highest probability of containing a gene (Levinson et al. 2003). Additionally, we also performed the bin-specific Fisher’s combining P values when both Psumrnk and Pord are <0.05 for further assuring the significance. The significance of the bins with both Psumrnk and Pord <0.05 by AS GSMA was all supported by Fisher’s method.

Previous genome scans have indicated non-HLA susceptibility loci outside chromosome 6, including 9q34.2, 19q13.31, 11q24.1, and 13q14.2. In the GSMA, the linkage evidence on non-HLA loci was shown on 16q12.2–16q23.1, 19p13.2–19q13.43, 17p13.3–17p12, and 16q23.1–16q24.3 across the studies (both Psumrnk and Pord<0.05). Linkage studies may have minimal power to detect genes of weak effect, but the combined evidence for linkage at these loci may be increased across studies. In contrast, false-positive results observed in individual studies should decrease in significance. Among non-HLA loci with strong linkage evidence observed in previous genome scans, chromosome 19q was confirmed in the GSMA. Additionally, the GSMA shows that chromosomes 16q, 19, and 17p may harbor non-HLA genes conferring a risk to AS. These regions may be interesting loci for further positional candidate gene studies. However, these data do not exclude the possibility that other chromosomal regions harbor AS susceptibility loci that could not be detected by the present methods or that substantially influence risk in only one or a few populations. GSMA does not also consider X- or Y-chromosome data, so no conclusions can be reached on possible linkage on the chromosomes.

The GSMA of all four genome scans, including one SpA scan, showed similar results with the GSMA data of three AS genome scans except for the linkage on chromosome 9q21.32–9q33.1. The 9q21.32–9q33.1 loci was found only in the GSMA, including the SpA genome scan. This finding suggests that the chromosome 9q region may confer susceptibility to SpA as well as AS.

The linkage loci shown by this meta-analysis may provide a basis for the location of AS susceptibility genes. It is interesting to examine these regions for candidate genes. For example, a possible candidate gene on chromosome 19q13.2 is transforming growth factor beta1 (TGF-β1) gene, which plays a crucial role in inflammatory processes, extracellular matrix synthesis, bone remodeling, and fibrosis and may be important in the biological pathways related to the expression of AS. One study has shown that the TGF-β1 + 1632 polymorphism was associated with a younger age of symptom onset, suggesting that TGF-β1 polymorphisms play a role in AS (Jaakkola et al. 2004).

A review of genome-wide scans of complex diseases has shown that true linkage remains hard to find (Altmuller et al. 2001). A complex disease might have many low- or modest-risk genes involved, with a low probability that the same results would be found in different genome-wide scans. Gene discovery in complex human diseases has been complicated by genetic heterogeneity, genes with small effect, and requirement for large samples (Goring et al. 2001). One potential solution to this issue is to combine data from multiple studies. Meta-analysis is an emerging method in the linkage analysis of complex diseases. The combination of evidence from multiple studies may prove to be critical to the successful localization of genes of small or modest effect in complex diseases. Therefore, the linkage results presented in the GSMA represent an important advance in searching non-HLA loci.

In conclusion, the GSMA added evidence of the HLA loci as the greatest susceptibility factor to AS and shows evidence of chromosomes 6, 16q, 19, 17p, and 9q as non-HLA susceptibility loci. These results provided a basis for future positional candidate gene studies in non-HLA loci.


  1. Altmuller J, Palmer LJ, Fischer G, Scherb H, Wjst M (2001) Genomewide scans of complex human diseases: true linkage is hard to find. Am J Hum Genet 69:936–950

    CAS  Article  Google Scholar 

  2. Braun J, Bollow M, Remlinger G, Eggens U, Rudwaleit M, Distler A, Sieper J (1998) Prevalence of spondylarthropathies in HLA-B27 positive and negative blood donors. Arthritis Rheum 41:58–67

    CAS  Article  Google Scholar 

  3. Breban M, Said-Nahal R, Hugot JP, Miceli-Richard C (2003) Familial and genetic aspects of spondyloarthropathy. Rheum Dis Clin North Am 29:575–594

    Article  Google Scholar 

  4. Brewerton DA, Hart FD, Nicholls A, Caffrey M, James DC, Sturrock RD (1973) Ankylosing spondylitis and HL-A 27. Lancet 1:904–907

    CAS  Article  Google Scholar 

  5. Brown MA, Kennedy LG, MacGregor AJ, Darke C, Duncan E, Shatford JL, Taylor A, Calin A, Wordsworth P (1997) Susceptibility to ankylosing spondylitis in twins: the role of genes, HLA, and the environment. Arthritis Rheum 40:1823–1828

    CAS  Article  Google Scholar 

  6. Brown MA, Kennedy LG, Darke C, Gibson K, Pile KD, Shatford JL, Taylor A, Calin A, Wordsworth BP (1998) The effect of HLA-DR genes on susceptibility to and severity of ankylosing spondylitis. Arthritis Rheum 41:460–465

    CAS  Article  Google Scholar 

  7. Brown MA, Laval SH, Brophy S, Calin A (2000) Recurrence risk modelling of the genetic susceptibility to ankylosing spondylitis. Ann Rheum Dis 59:883–886

    CAS  Article  Google Scholar 

  8. Brown MA, Wordsworth BP, Reveille JD (2002) Genetics of ankylosing spondylitis. Clin Exp Rheumatol 20:S43–S49

    CAS  PubMed  Google Scholar 

  9. Brown MA, Brophy S, Bradbury L, Hamersma J, Timms A, Laval S, Cardon L, Calin A, Wordsworth BP (2003) Identification of major loci controlling clinical manifestations of ankylosing spondylitis. Arthritis Rheum 48:2234–2239

    Article  Google Scholar 

  10. Calin A, Marder A, Becks E, Burns T (1983) Genetic differences between B27 positive patients with ankylosing spondylitis and B27 positive healthy controls. Arthritis Rheum 26:1460–1464

    CAS  Article  Google Scholar 

  11. Cordell HJ (2001) Sample size requirements to control for stochastic variation in magnitude and location of allele-sharing linkage statistics in affected sibling pairs. Ann Hum Genet 65:491–502

    CAS  Article  Google Scholar 

  12. Goring HH, Terwilliger JD, Blangero J (2001) Large upward bias in estimation of locus-specific effects from genomewide scans. Am J Hum Genet 69:1357–1369

    CAS  Article  Google Scholar 

  13. Guerra R, Etzel CJ, Goldstein DR, Sain SR (1999) Meta-analysis by combining p-values: simulated linkage studies. Genet Epidemiol (Suppl 1):S605–S609

    Article  Google Scholar 

  14. Jaakkola E, Crane AM, Laiho K, Herzberg I, Sims AM, Bradbury L, Calin A, Brophy S, Kauppi M, Kaarela K, Wordsworth BP, Tuomilehto J, Brown MA (2004) The effect of transforming growth factor beta1 gene polymorphisms in ankylosing spondylitis. Rheumatology 43:32–38

    CAS  Article  Google Scholar 

  15. Jarvinen P (1995) Occurrence of ankylosing spondylitis in a nationwide series of twins. Arthritis Rheum 38:381–383

    CAS  Article  Google Scholar 

  16. Laval SH, Timms A, Edwards S, Bradbury L, Brophy S, Milicic A, Rubin L, Siminovitch KA, Weeks DE, Calin A, Wordsworth BP, Brown MA (2001) Whole-genome screening in ankylosing spondylitis: evidence of non-MHC genetic-susceptibility loci. Am J Hum Genet 68:918–926

    CAS  Article  Google Scholar 

  17. Levinson DF, Levinson MD, Segurado R, Lewis CM (2003) Genome scan meta-analysis of schizophrenia and bipolar disorder, part I: methods and power analysis. Am J Hum Genet 73:17–33

    CAS  Article  Google Scholar 

  18. Miceli-Richard C, Zouali H, Said-Nahal R, Lesage S, Merlin F, De Toma C, Blanche H, Sahbatou M, Dougados M, Thomas G, Breban M, Hugot JP (2004) Significant linkage to spondyloarthropathy on 9q31–34. Hum Mol Genet 13:1641–1648

    CAS  Article  Google Scholar 

  19. Province MA (2001) The significance of not finding a gene. Am J Hum Genet 69:660–663

    CAS  Article  Google Scholar 

  20. Risch N, Merikangas K (1996) The future of genetic studies of complex human diseases. Science 273:1516–1517

    CAS  Article  Google Scholar 

  21. Wise LH, Lanchbury JS, Lewis CM (1999) Meta-analysis of genome searches. Ann Hum Genet 63:263–272

    CAS  Article  Google Scholar 

  22. Zhang G, Luo J, Bruckel J, Weisman MA, Schumacher HR, Khan MA, Inman RD, Mahowald M, Maksymowych WP, Martin TM, Yu DT, Stone M, Rosenbaum JT, Newman P, Lee J, McClain JA, West OC, Jin L, Reveille JD (2004) Genetic studies in familial ankylosing spondylitis susceptibility. Arthritis Rheum 50:2246–2254

    CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Young Ho Lee.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Lee, Y.H., Rho, Y.H., Choi, S.J. et al. Ankylosing spondylitis susceptibility loci defined by genome-search meta-analysis. J Hum Genet 50, 453–459 (2005).

Download citation


  • Ankylosing spondylitis
  • Spondyloarthropathy
  • Genome scan
  • Linkage
  • Meta-analysis

Further reading


Quick links