SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Objective

To use a candidate gene approach to the identification of genetic markers that are significantly associated with ankylosing spondylitis (AS).

Methods

We genotyped 201 multiplex AS families with 1 exonic and 5 intronic single-nucleotide polymorphisms (SNPs) in TNAP, the gene that encodes tissue-nonspecific alkaline phosphatase, and performed family-based association analyses.

Results

In our cohort of 201 multiplex AS families, the TNAP haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) was significantly associated with AS (P = 0.032 by additive model). Haplotype-Based Association Testing (HBAT) analyses of AS families in which both men and women were affected showed that the same TNAP haplotype was significantly associated with AS (P = 0.002 by additive model). Using setafftrait code 1 0 0 in the HBAT program, testing specifically for affected men in AS families containing affected individuals of both sexes, this TNAP haplotype was also significantly associated with AS (P = 0.001 by additive model). The HBAT –p option (haplotype permutation test) was used to compute the “exact” P value via a Monte Carlo method for each haplotype (haplotype permutation test) and for the minimum observed P value among the haplotypes (whole marker permutation using the minimal P test), and both P values were statistically significant (2-sided P value for haplotype rs3767155 [G]/rs3738099 [G]/rs1780329 [T] = 0.00059, the smallest observed P value among all the individual haplotype scores = 0.003). Interestingly, this haplotype was not associated with AS in affected women from the same families.

Conclusion

Our results indicate that the TNAP haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) is a novel genetic marker in men that is significantly associated with AS in multiplex families containing affected individuals of both sexes.

Ankylosing spondylitis (AS) is a form of chronic arthritis that affects young adults in their most active and productive years. AS is usually diagnosed according to the modified New York criteria (1), as follows: limited motion of the lumbar spine, persistent lower back pain, limited chest expansion, and radiographic sacroiliitis. The course of the disease is variable. The sine qua non for AS is sacroiliitis, but the characteristic radiographic abnormalities often may not appear for 5–10 years after the onset of symptoms. In some patients, inflammatory disease is limited to the sacroiliac joints and spine; in others, the disease also affects the peripheral joints. In the most severe cases, patients experience progressive and unremitting spinal ankylosis, leading ultimately to “bamboo spine,” with marked pain, extreme stiffness, and severe disability.

AS has a strong genetic component. It is highly associated with HLA–B27, but analyses of recurrence risk among family members suggest that at least 3 other genetic loci in addition to HLA–B27 are required to confer full susceptibility to AS (2). Despite several large genome-wide linkage studies of AS families (3–6), the IL1 locus is the only significant non–major histocompatibility complex (non-MHC) susceptibility locus that has thus far emerged (7, 8). However, 2 family-based association studies, 1 of which examined 244 affected sibpairs, showed no evidence of linkage to genes in the IL1 cluster (9, 10). The difficulty in identifying non-MHC susceptibility genes in AS has suggested that AS patients comprise a heterogeneous population. It is possible that the similar pathology found in different subsets of AS patients might be due to polymorphisms in several different genes, the products of which might be involved in pathways critical for neo-ossification and joint inflammation. For this reason, it might be more fruitful to test specific candidate genes and to analyze the data separately in different types of multiplex families.

Several lines of evidence point to the potential importance of inorganic pyrophosphate (PPi) and inorganic phosphate (Pi) metabolism in chronic joint diseases. Several human and mouse diseases with excessive or insufficient bone formation have been related to defects in, respectively, the phosphodiesterases (such as nucleotide pyrophosphatase/phosphodiesterase 1), which produce PPi (11, 12), and the phosphatases (such as tissue-nonspecific alkaline phosphatase [TNAP]), which degrade PPi (13–15). The homozygous ank (progressive ankylosis) mouse has a loss-of-function nonsense mutation in the ank gene that codes for a regulator of PPi export, and these mice develop a condition characterized by pathologic calcium apatite crystal deposition in the synovial and subsynovial spaces, followed by chondro-osteophyte formation and eventual bony ankylosis of the affected joints (16). The phenotype of the ank/ank mouse suggests that the ank gene is involved in pathways that are critical for neo-ossification, which is the hallmark of AS.

It remains unclear why neo-ossification is restricted to sites of inflammation in humans with AS. Since ANKH is a growth factor (17) and an androgen-responsive gene (18), it is possible that cytokines generated by inflammation dysregulate the expression of ANKH, resulting in neo-ossification. We previously tested whether ANKH is one of the susceptibility genes for AS (19), and we later showed that ANKH is linked to, and associated with, AS in 201 Caucasian multiplex families (20). However, another study concluded that ANKH did not significantly contribute to susceptibility or specific disease expression in AS patients from the UK (21). The basis for the discrepancy between the results in the UK and the North American populations is not entirely clear. The differences could reflect clinical differences in the patient populations that were recruited, or they could be due to population-specific mechanisms of genetic susceptibility (20).

There were 2 types of families in our cohort of multiplex AS families: those with affected individuals of both sexes and those with affected individuals of only 1 sex. We showed that there was heterogeneity in the different types of families, as reflected by sex differences in the ANKH variants associated with AS (20). A recent study of French multiplex AS families also identified 2 groups of patients, one consisting predominantly of affected women and the other consisting predominantly of affected men (22). Patients from both groups showed similar axial symptoms, radiographic sacroiliitis, and uveitis. However, the group consisting predominantly of affected men had an earlier age at onset and a higher frequency of clinical enthesitis, peripheral arthritis, dactylitis, psoriasis, and inflammatory bowel disease.

Immunostaining studies have shown that ANKH, TNAP, and type X collagen colocalize in zones of hypertrophic and mineralizing growth plate cartilage (23). Enhanced ANKH staining was detected in human osteoarthritic cartilage as compared with normal cartilage (23, 24). TNAP staining was absent in normal human articular cartilage, but was present in chondrocytes from human osteoarthritic cartilage (23). Based on studies of embryonic chick hypertrophic chondrocytes, Wang et al (23) proposed the following model for how ANK expression might affect TNAP expression. Enhanced ANK PPi transport activity results in increased levels of extracellular PPi (ePPi) and hydrolysis of ePPi by TNAP. Pi resulting from PPi hydrolysis then enters the cell through Na+/Pi cotransporters (PiT-1 and PiT-2). Transport of Pi into the cell results in a further stimulation of TNAP expression. Regulation of the expression and activity of TNAP by ANK ensures that sufficient TNAP activity is available at the outer membrane surface of growth plate chondrocytes and matrix vesicles to remove PPi, which is a potent inhibitor of mineralization (24). It is unclear whether this model of regulation of TNAP expression by ANK in chicken chondrocytes is applicable to mammalian chondrocytes.

We hypothesized that the variable genetic determinants among AS patients may perturb different aspects of a common biochemical pathway to yield a common pathologic outcome characterized by aberrant ossification. Previously, our family-based association analyses showed that AS is significantly associated with a specific ANKH haplotype. In view of the interplay of ANKH and TNAP in pyrophosphate and phosphate metabolism and in ossification, we sought to determine whether specific TNAP variants are associated with AS.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

AS families.

The study group comprised 201 Caucasian AS multiplex families (a total of 226 nuclear families) (20). This group was recruited from the Toronto Western Spondylitis Clinic (23 families) and from other sites in the North American Spondylitis Consortium (178 families). All patients met the modified New York criteria for a diagnosis of AS (1), including radiographic evidence of sacroiliitis. Family members older than 40 years who had no history of signs or symptoms of spondylarthritis were recruited as unaffected individuals. Of the affected and unaffected individuals, 60% and 47%, respectively, were men. The ages of all study subjects ranged from 8 years to 75 years.

There were 90 AS nuclear families containing affected individuals of both sexes (76 pedigrees; 363 persons, with 95 affected men and 89 affected women) and 102 AS nuclear families containing affected individuals of only 1 sex (93 pedigrees [63 pedigrees with 138 affected men and 30 pedigrees with 64 affected women]; 418 persons). There were 34 AS nuclear families for which the sex information was incomplete or missing, and these families were therefore excluded from analyses relating to sex differences. AS patients in almost the entire cohort of AS multiplex families (195 of 201 families) were HLA–B27 positive.

The study was approved by the University Health Network Research Ethics Board and the Committee for the Protection of Human Subjects at the University of Texas Health Science Center at Houston.

Genotyping.

DNA from the affected and unaffected family members was prepared from peripheral blood lymphocytes using standard techniques. Genotyping was performed using 5 TNAP intronic single-nucleotide polymorphisms (SNPs; rs1472563 [G/A], rs4654957 [T/A], rs1256348 [G/A], rs3767155 [G/A], and rs1780329 [G/T]) and 1 exonic variant (rs3738099 [A/G]). Optimized allele discrimination assays for SNPs were purchased from Applied Biosystems (Foster City, CA). Selection of these SNPs was based on the frequency of the minor alleles (>5%), their location in the linkage disequilibrium (LD) blocks (25), and their availability from Applied Biosystems. Two of the 6 SNPs we genotyped are tag SNPs (rs1472563 and rs3767155). The plates were read using an ABI Prism 7900 sequence detection system (Applied Biosystems).

Statistical analysis.

The transmission disequilibrium test was used to test for transmission of specific alleles from heterozygous parents to affected offspring (26, 27). We computed the test statistics using the Family-Based Association Testing (FBAT) software, version 1.7.2 (online at http://www.biostat.harvard.edu/∼fbat/default.html) (27). This program uses data from nuclear families, sibships, pedigrees, or any combination thereof and provides unbiased tests with or without founder genotypes. Biallelic tests were performed using additive, dominant/recessive genetic models. Haplotype analyses were performed using the Haplotype-Based Association Testing (HBAT) routine in the FBAT program. The PBAT program (version 3.2) was used for a priori power computation of the FBAT statistics. The HBAT –p option (which uses the full conditional distribution of offspring haplotypes) was used to compute an “exact” P value via a Monte Carlo method for each haplotype separately and for the minimum observed P value among the haplotypes.

For analysis of affected men/women, the FBAT command “setafftrait” was used. Unaffected siblings and parents from the families were coded as 0 (unknown phenotype), affected men were coded as 2, and affected women were coded as 1. HBAT analyses using the setafftrait 1 0 0 command were used to test specifically for affected men, and analyses using the setafftrait 0 –1 0 command were used to test specifically for affected women. To test for differences between family-based association for affected men and women, the setafftrait 1 –1 0 command was used.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Association between specific TNAP variants and AS.

In a family-based association analysis, 201 multiplex AS families (with 226 nuclear families) were genotyped with 6 markers in the TNAP region: rs1472563 in the promoter region, rs4654957 and rs1256348 in intron 1, rs3767155 in intron 2, rs3738099 in exon 7, and rs1780329 in intron 10 (Figure 1). Using 3 different routines (numerical integration, approximation, and simulation), and with parameters as close as possible to our cohort of AS families, an estimated a priori power of ∼0.8 was achieved with the additive model. Where the etiologic variant is not typed, haplotype-based analysis is more powerful for association studies in which there is significant LD in the region of interest. According to HapMap Public Release no. 19 (October 24, 2005; online at http://www.hapmap.org) (25), there are 2 LD blocks in the human TNAP gene: rs1472563, rs4654957, and rs1256348 are located in LD block 1, while rs3767155, rs3738099, and rs1780329 are located in LD block 2 (Figure 1).

thumbnail image

Figure 1. Locations and spacings of TNAP single-nucleotide polymorphisms (SNPs) used for genotyping. The 2 linkage disequilibrium (LD) blocks (HapMap Public Release no. 19; October 24, 2005) are shown.

Download figure to PowerPoint

We performed haplotype analyses based on this information, and the results are summarized in Table 1. Haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) in LD block 2 was significantly associated with AS (P = 0.032 by additive model). Permutation test for this haplotype also showed a modest association with AS (P_2side = 0.028 by additive model). However, no significant P values were observed in the minimal P tests for each of the 2 data sets (Table 1).

Table 1. Haplotype-Based Association Testing for TNAP variants conducted in 226 ankylosing spondylitis nuclear families*
SNP, haplotypeHaplotype frequencyNo. of informative familiesZ scorePP_2side, by haplotype permutation testMinimal P
  • *

    A total of 201 pedigrees and 894 persons were evaluated, using an additive model and biallelic tests. The 2 linkage disequilibrium (LD) blocks in the human TNAP gene consist of rs1472563, rs4654957, and rs1256348 in LD block 1, and rs3767155, rs3738099, and rs1780329 in LD block 2. SNP = single-nucleotide polymorphism; P_2side = 2-sided P.

rs1472563, rs4654957, and rs1256348      
 G/T/G0.41800.690.4890.495 
 A/T/G0.3877−0.160.8710.816 
 G/A/G0.13450.110.9090.885 
 A/T/A0.0721−0.720.4690.521 
      0.327
rs3767155, rs3738099, and rs1780329      
 G/A/G0.4473−0.280.7820.792 
 A/A/G0.3774−0.930.3510.372 
 G/G/T0.12452.140.0320.028 
 G/A/T0.0622−0.250.8030.809 
      0.135

Assessment of sex differences in TNAP variants associated with AS.

It has been observed that long-term outcomes in AS, such as radiographic changes, are worse in men than in women (28–30). We have previously shown that there are sex differences in ANKH variants associated with AS (20). To assess whether similar sex differences are present in TNAP variants associated with AS, we reanalyzed our genotyping results along sex lines in 2 separate HBAT analyses, using the setafftrait command. HBAT analyses of the transmission of alleles to affected men showed that the haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) in LD block 2 was associated with AS (P = 0.008 by additive model) (Table 2). As expected, the minimal P tests for haplotypes in LD block 1 showed no significant P values. In contrast, modestly significant P values were achieved in the minimal P tests for haplotypes in LD block 2 (minimal P = 0.029 by additive model). In contrast to the results for affected men, no TNAP variants were significantly associated with AS in affected women (Table 2).

Table 2. Haplotype-Based Association Testing for TNAP variants specifically in affected men and affected women, conducted in 226 ankylosing spondylitis nuclear families*
SNP, haplotypeHaplotype frequencyNo. of informative familiesS scoreE(S) scoreZ scorePP_2side, by haplotype permutation testMinimal P
  • *

    A total of 201 pedigrees and 894 persons were evaluated, using an additive model and biallelic tests. In tests for affected men, setafftrait code 1 0 0 was used; in tests for affected women, setafftrait code 0 −1 0 was used. Values in boldface are highlighted for discussions. SNP = single-nucleotide polymorphism; E(S) score = expected S score; P_2side = 2-sided P.

  • Statistically significant.

Affected men        
 rs1472563, rs4654957, and rs1256348        
  G/T/G0.4163128.4121.31.410.1560.150 
  A/T/G0.3868111.0119.7−1.640.1010.102 
  G/A/G0.133750.248.60.410.6820.661 
  A/T/A0.071723.623.30.150.8830.868 
        0.289
 rs3767155, rs3738099, and rs1780329        
  G/A/G0.4466134.2134.5−0.070.9440.915 
  A/A/G0.3764119.4124.9−1.020.3030.300 
  G/G/T0.123151.643.22.640.0080.011 
  G/A/T0.061715.317.6−0.920.3530.418 
        0.029
Affected women        
 rs1472563, rs4654957, and rs1256348        
  G/T/G0.4156−71.7−74.60.630.5260.517 
  A/T/G0.3855−74.3−66.4−1.740.0820.073 
  G/A/G0.1331−32.8−33.90.320.7480.731 
  A/T/A0.0715−14.6−16.91.210.2260.228 
        0.418
 rs3767155, rs3738099, and rs1780329        
  G/A/G0.4453−89.3−90.80.370.7140.736 
  A/A/G0.3760−76.7−76.80.010.9910.982 
  G/G/T0.1232−32.0−30.9−0.320.7490.670 
  G/A/T0.0615−18.0−16.9−0.490.6240.701 
        0.983

Assessment of heterogeneity in multiplex AS families with regard to TNAP variants associated with AS.

There were 2 types of families in our cohort of multiplex AS families: those with affected individuals of both sexes, and those with affected individuals of only 1 sex. We have previously shown that there is significant heterogeneity in these different multiplex AS families (20). Thus, we assessed whether there was heterogeneity in both types of AS families with regard to TNAP variants associated with AS.

HBAT analyses were performed in 90 nuclear AS families containing both affected men and affected women. The haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) was significantly associated with AS (P = 0.002 by additive model) (Table 3). The permutation tests for this haplotype in LD block 2 also showed significant P values (P_2side = 0.001 by additive model). Significant minimal P values were also achieved with the data set in relation to haplotypes in LD block 2 (minimal P = 0.008 by additive model). Table 4 summarizes the results of the Hardy-Weinberg equilibrium tests performed, showing that none of the allele frequencies deviated from Hardy-Weinberg equilibrium, and thus, it is likely that there was no genotyping error. Similar analyses were performed in 102 nuclear AS families containing affected individuals of only 1 sex, and surprisingly, no TNAP variant or haplotype was associated with AS in these families (Table 3). Results of Hardy-Weinberg equilibrium tests for this analysis also suggested that there was no genotyping error (Table 4).

Table 3. Haplotype-Based Association Testing for TNAP variants conducted in 90 AS nuclear families containing affected individuals of both sexes and in 102 AS nuclear families containing affected individuals of only 1 sex*
SNP, haplotypeHaplotype frequencyNo. of informative familiesZ scorePP_2side, by haplotype permutation testMinimal P
  • *

    A total of 76 pedigrees and 363 persons were evaluated in the 90 ankylosing spondylitis (AS) nuclear families containing affected individuals of both sexes, and a total of 93 pedigrees and 418 persons were evaluated in the 102 AS nuclear families containing affected individuals of only 1 sex, using an additive model and biallelic tests. The 2 linkage disequilibrium (LD) blocks in the human TNAP gene consist of rs1472563, rs4654957, and rs1256348 in LD block 1, and rs3767155, rs3738099, and rs1780329 in LD block 2. SNP = single-nucleotide polymorphism; P_2side = 2-sided P; NC = not calculated (fewer than 10 informative families).

  • Statistically significant.

Families with affected individuals of both sexes      
 rs1472563, rs4654957, and rs1256348      
  G/T/G0.44320.630.5310.526 
  A/T/G0.33301.330.1830.188 
  G/A/G0.1419−1.110.2630.283 
  A/T/A0.0810−1.660.0960.098 
      0.369
 rs3767155, rs3738099, and rs1780329      
  G/A/G0.4329−0.460.6420.571 
  A/A/G0.3930−0.930.3500.349 
  G/G/T0.12142.990.0020.001 
  G/A/T0.058NCNCNC 
      0.008
Families with affected individuals of only 1 sex      
 rs1472563, rs4654957, and rs1256348      
  G/T/G0.40290.340.7350.660 
  A/T/G0.4029−0.660.5110.458 
  G/A/G0.13160.440.6610.766 
  A/T/A0.077NCNCNC 
      0.979
 rs3767155, rs3738099, and rs1780329      
  G/A/G0.4425−0.160.8690.927 
  A/A/G0.3625−0.700.4820.462 
  G/G/T0.12170.970.3310.263 
  G/A/T0.078NCNCNC 
      0.834
Table 4. Summary of Hardy-Weinberg equilibrium tests for genotyping error in the 90 AS nuclear families containing affected individuals of both sexes and in the 102 AS nuclear families containing affected individuals of only 1 sex*
SNPTotal no. of personsAllele frequencyχ2P
MajorMinor
  • *

    A total of 76 pedigrees and 363 persons were evaluated in the 90 ankylosing spondylitis (AS) nuclear families containing affected individuals of both sexes, and a total of 93 pedigrees and 418 persons were evaluated in the 102 AS nuclear families containing affected individuals of only 1 sex. SNP = single-nucleotide polymorphism.

Families with affected individuals of both sexes     
 rs14725632590.590.410.00970.922
 rs46549572140.840.160.66830.414
 rs12563482200.910.090.05140.821
 rs37671552560.640.370.72090.396
 rs37380992560.840.161.42210.233
 rs17803292590.790.210.32820.567
Families with affected individuals of only 1 sex     
 rs14725632910.540.463.70850.054
 rs46549572410.920.080.31190.577
 rs12563482410.830.171.67140.196
 rs37671552940.630.370.09220.761
 rs37380992860.860.140.71310.398
 rs17803292950.810.190.26880.604

When HBAT analyses using the setafftrait command were performed in AS multiplex families containing affected individuals of both sexes, significance was found in the haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) only in affected men (P = 0.001 by additive model), but not in affected women (Table 5). As expected, permutation tests for this haplotype in LD block 2 showed significant P values (P_2 side = 0.00059 by additive model). Minimal P tests for the data set relating to haplotypes in LD block 2 (but not LD block 1) were significant (P = 0.003 by additive model).

Table 5. Haplotype-Based Association Testing for TNAP variants specifically in affected men and affected women, conducted in 90 ankylosing spondylitis nuclear families containing affected individuals of both sexes*
SNP, haplotypeHaplotype frequencyNo. of informative familiesS scoreE(S) scoreZ scorePP_2side, by haplotype permutation testMinimal P
  • *

    A total of 76 pedigrees and 363 persons were evaluated, using an additive model and biallelic tests. The 2 linkage disequilibrium (LD) blocks in the human TNAP gene consist of rs1472563, rs4654957, and rs1256348 in LD block 1, and rs3767155, rs3738099, and rs1780329 in LD block 2. In tests for affected men, setafftrait code 1 0 0 was used; in tests for affected women, setafftrait code 0 −1 0 was used. Values in boldface are highlighted for discussions. SNP = single-nucleotide polymorphism; E(S) score = expected S score; P_2side = 2-sided P; NC = not calculated (fewer than 10 informative families).

  • Statistically significant.

Affected men        
 rs1472563, rs4654957, and rs1256348        
  G/T/G0.443647.044.50.710.4750.435 
  A/T/G0.333739.240.4−0.350.7240.743 
  G/A/G0.142221.821.20.220.8270.975 
  A/T/A0.08118.810.1−0.700.4840.469 
        0.819
 rs3767155, rs3738099, and rs1780329        
  G/A/G0.433749.749.60.040.9670.957 
  A/A/G0.393639.043.7−1.310.1880.207 
  G/G/T0.121620.713.93.240.0010.00059 
  G/A/T0.05107.38.6−0.720.4700.577 
        0.003
Affected women        
 rs1472563, rs4654957, and rs1256348        
  G/T/G0.4439−47.6−47.3−0.080.9310.003 
  A/T/G0.3340−44.3−37.1−1.870.0620.063 
  G/A/G0.1422−15.0−19.31.530.1250.148 
  A/T/A0.0812−7.6−9.91.360.1740.174 
        0.195
 rs3767155, rs3738099, and rs1780329        
  G/A/G0.4339−43.0−45.00.570.5650.597 
  A/A/G0.3943−48.0−45.0−0.150.8840.958 
  G/G/T0.1217−17.0−15.9−0.460.6490.527 
  G/A/T0.059NCNCNCNCNC 
        0.959

Table 6 summarizes the allele frequencies in 90 AS nuclear families containing affected individuals of both sexes. There were no intrinsic sex differences in the allele frequencies. However, we cannot conclude that there were sex differences in TNAP variants associated with AS, since we failed to show significant heterogeneity between affected men and affected women using the setafftrait command 1 –1 0 (data not shown).

Table 6. Summary of TNAP allele frequencies in men and women from 90 ankylosing spondylitis nuclear families containing affected individuals of both sexes*
SNP, alleleAllele frequency (%)χ2P
MenWomen
  • *

    Values are the number with the allele/total number of individuals (%). SNP = single-nucleotide polymorphism.

rs1472563    
 Major160/258 (62.02)147/262 (56.1)  
 Minor98/258 (37.98)115/262 (43.9)  
   1.8770.171
rs4654957    
 Major179/218 (82.11)188/218 (86.24)  
 Minor39/218 (17.89)30/218 (13.76)  
   1.3950.238
rs1256348    
 Major202/218 (92.66)201/224 (89.73)  
 Minor16/218 (7.34)23/224 (10.27)  
   1.1780.278
rs3767155    
 Major162/258 (62.79)165/260 (63.46)  
 Minor96/258 (37.21)95/260 (36.54)  
   0.0250.874
rs3738099    
 Major213/256 (83.20)218/258 (84.5)  
 Minor43/256 (16.80)40/258 (15.5)  
   0.1590.690
rs1780329    
 Major201/260 (77.31)208/258 (80.62)  
 Minor59/260 (22.69)50/258 (19.38)  
   0.8550.355

We did not find a significant association of any TNAP haplotype with affected men in the 102 AS nuclear families containing affected individuals of only 1 sex (Table 7). There were too few informative families (fewer than 10) to conduct HBAT analyses of affected women in these 102 AS families.

Table 7. Haplotype-Based Association Testing for TNAP variants conducted in 102 ankylosing spondylitis nuclear families containing affected individuals of only 1 sex*
SNP, haplotypeHaplotype frequencyNo. of informative familiesZ scorePP_2side, by haplotype permutation testMinimal P
  • *

    A total of 93 pedigrees and 418 persons were evaluated, using an additive model and biallelic tests. The 2 linkage disequilibrium (LD) blocks in the human TNAP gene consist of rs1472563, rs4654957, and rs1256348 in LD block 1, and rs3767155, rs3738099, and rs1780329 in LD block 2. In tests for affected men, setafftrait code 1 0 0 was used. SNP = single-nucleotide polymorphism; P_2side = 2-sided P; NC = not calculated (fewer than 10 informative families).

Affected men      
 rs1472563, rs4654957, and rs1256348      
  G/T/G0.40210.650.5130.515 
  A/T/G0.4021−1.120.2600.258 
  G/A/G0.13120.300.7640.741 
  A/T/A0.075NCNCNC 
      0.833
 rs3767155, rs3738099, and rs1780329      
  G/A/G0.4420−0.300.7660.671 
  A/A/G0.3619−0.350.7260.767 
  G/G/T0.12110.760.4440.588 
  G/A/T0.075NCNCNC 
      0.865
Affected women (NC)

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

In this study of the association between TNAP genetic markers and AS, analyzing 201 AS multiplex families, we found that a specific TNAP haplotype was significantly associated with AS, especially in affected men from AS families containing affected individuals of both sexes. What might be the biologic relevance of this result? A recent report on the regulation of sexually dimorphic genes in mice showed that while each gene functions similarly in both sexes, there is a direct correlation between sex and the amount of gene transcripts expressed (31). It was proposed that a large number of genes with small differences between the sexes could contribute to sex-biased susceptibility to common diseases (31). Serum alkaline phosphatase activity is higher in men than in women up to the age of 50 years (32). Our result therefore suggests that the level of alkaline phosphatase activity in young men from AS multiplex families containing affected individuals of both sexes might contribute to disease susceptibility.

The TNAP haplotype associated with AS contains an exonic variant (rs3738099 [G]) that has a histidine residue instead of a tyrosine residue at amino acid 263. This Tyr263His variant lies in the calcium-binding site vicinity of TNAP. However, it is not yet resolved whether this exonic variant has any functional significance. It is also unclear why this TNAP haplotype was not associated with AS in affected men from AS multiplex families containing affected individuals of only 1 sex. Heritability in AS is very complex. AS has a higher prevalence in the offspring of women with AS than the offspring of men with AS, and the sons of men with AS are 2.5 times more likely than the daughters to inherit the disease (33, 34).

Results of tests of heterogeneity between the sexes were not significant, but tests of heterogeneity are known to have lower power than tests of main effects. In fact, not only was no significant overtransmission of haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) detected among affected women, but there was also slightly less transmission than would be expected by chance (compare S versus E(S) scores for haplotype rs3767155 [G]/rs3738099 [G]/rs1780329 [T] in men versus women in Tables 2 and 5 [see numbers in boldface]). It will be of interest to assess whether there are any sex differences in alkaline phosphatase activities in AS multiplex families.

Our results highlight the heterogeneity that exists in multiplex AS families. In our cohort of the 2 types of AS families, there were differences in 2 features. First, in AS families containing affected individuals of both sexes, there were similar numbers of affected women and affected men (89 and 95, respectively, from 76 pedigrees). However, in the AS families containing affected individuals of only 1 sex, there were 138 affected men from 63 pedigrees and 64 affected women from 30 pedigrees. Thus, the ratio of AS families with only affected men versus only affected women is ∼2:1. Second, in the subset of families containing affected individuals of both sexes, men had a younger age at diagnosis (mean ± SD 27.8 ± 11 years; n = 94) as compared with women (32.6 ± 11 years; n = 101), whereas in the AS families containing affected individuals of only 1 sex, the men and the women had a similar age at diagnosis (28.6 ± 12 years [n = 130] versus 29.9 ± 12 years [n = 53]). However, FBAT analyses using the offset option (–o), which works for both quantitative and qualitative traits, did not show any significant association between age at diagnosis in the men or women and AS, even in the subsets of families, using the 6 TNAP markers (data not shown). This suggests that TNAP variants are responsible for disease susceptibility. We had too few informative families (fewer than 10; especially for haplotype rs3767155 [G]/rs3738099 [G]/rs1780329 [T]) to analyze whether the results are the same in the presence of associated comorbidities such as inflammatory bowel disease, psoriasis, and uveitis.

Statistical tests are available to control the false-positive rate at the nominal level over all polymorphisms considered. Since the Bonferroni correction is very conservative, especially when tests are correlated as in the present study, we used a more powerful test, permutation resampling, to analyze correlations between polymorphisms and test statistics (35). We used the HBAT –p option, which takes into account the full conditional distribution of offspring haplotypes, to compute an “exact” P value via the Monte Carlo method for each haplotype separately. In addition, it provided the minimum P test, which evaluates the statistical significance of the smallest observed P value among all the individual haplotype scores.

Based on this minimum P test, we concluded that the TNAP haplotype rs3767155 (G)/rs3738099 (G)/rs1780329 (T) is associated with AS in affected men from our cohort of AS multiplex families containing affected individuals of both sexes (biallelic tests; additive model). For our cohort of AS families with affected individuals of both sexes, there were similar numbers of affected women and affected men (89 and 95 respectively), and thus, the lack of a statistically significant association in the affected women in these families were not due to lack of power (Type II, or beta, error). Using the PBAT program with parameters close to our cohort of AS families, the a priori power estimation in our cohort of 201 AS multiplex families was ∼0.8 (additive model). This is probably an underestimate, since some of our families had more than 2 affected individuals (3–5) and PBAT power estimation takes into account only 2 affected individuals in a family. However, the subset of families containing only affected women (30 pedigrees) had a much lower a priori power estimation (∼0.2 by additive model). Thus, we cannot draw any conclusion about this subset of AS families (affected women only).

The chromosomal TNAP gene is located on chromosome 1p36.12. Genome-wide linkage studies for AS showed no evidence of linkage in this chromosome region. However, there are examples showing significant associations between gene variants and AS, even with genes that either are not located in regions that showed significant linkage in genome-wide linkage scans or are not linked to AS in family-based association studies. For example, we previously showed that ANKH is linked to and is associated with AS in our cohort of AS multiplex families, but the chromosomal 5p15.2 region in which ANKH is localized had no evidence of linkage in genome-wide linkage studies. The major locus encompassing the IL1 gene complex was associated with AS, and yet, a genome-wide linkage study and a family-based association study showed no evidence of linkage to genes in the IL1 cluster (9, 10). These discrepancies between results were likely due to insufficient power to consistently detect genes with “small effects.”

We previously showed that in AS multiplex families with affected individuals of both sexes, ANKH variants from an intronic region were significantly associated with AS only in affected women (20). In the present study, we showed that in the same families, a specific haplotype containing an exonic variant was significantly associated with AS only in affected men from families containing affected individuals of both sexes. Taken together, these results implicate genes involved in PPi metabolism as being important in the predisposition for AS. TNAP can generate PPi from nucleoside triphosphate (NTP) and can hydrolyze PPi and Pi from NTP, nucleoside diphosphate, nucleoside monophosphate, proteins, and lipids, while ANKH regulates the export of PPi from intracellular to extracellular compartments. Furthermore, our results illustrate the principle that polymorphisms that affect different genes can contribute to a common pathology when these gene products affect the same biochemical pathway.

We are aware that the numbers of informative families in our subsets of AS multiplex families are relatively small. This was due in part to a relatively low frequency (15%) of the minor allele of the TNAP exonic variant rs3738099 (G). Thus, our data would require confirmation at a later date with larger numbers of informative families. The use of multiplex AS families for association analyses has the advantage of enhancing the chances of identifying the genes involved. It remains unclear whether our findings from multiplex families might be directly applicable to sporadic cases of AS. This could be assessed using case–control studies, which involve recruitment of sporadic cases of AS and normal controls.

Since neo-ossification at sites of chronic inflammation is the hallmark of AS, this is proving to be an informative line of investigation into the pathogenesis of this disease.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Dr. F. W. L. Tsui had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Drs. Inman, Reveille, and F. W. L. Tsui.

Acquisition of data. Drs. H. W. Tsui and Reveille.

Analysis and interpretation of data. Drs. H. W. Tsui, Inman, and F. W. L. Tsui.

Manuscript preparation. Drs. H. W. Tsui, Inman, Reveille, and F. W. L. Tsui.

Statistical analysis. Dr. F. W. L. Tsui.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

We thank Dr. Cathy Barr for making the ABI sequence detection system available and Karen Wigg for reading the plates, and we thank Dr. Celia Greenwood for her expert advice with the statistical analyses and interpretation of the data. We also thank the Ontario Spondylitis Association and the Spondylitis Association of America for their assistance in recruiting the families included in this study.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES