Frizzled-related protein variants are risk factors for hip osteoarthritis

Authors


Abstract

Objective

To examine the association of the Arg200Trp and Arg324Gly variants of FRZB with the risk and phenotype of radiographic osteoarthritis (OA) of the hip and serum levels of Frizzled-related protein (FRP) in a prospective cohort of elderly Caucasian women.

Methods

Radiographic hip OA status of patients was defined by the presence of severe joint space narrowing (JSN) (feature grade ≥3), a summary grade ≥3, or definite osteophytes (grade ≥2) and JSN (grade ≥2) in the same hip. Genotypes were obtained in 569 patients with radiographic OA of the hip and in 1,317 and 4,136 controls for the Arg200Trp and Arg324Gly variants, respectively. Serum FRP levels were measured by enzyme-linked immunosorbent assay. Multivariate logistic regression was performed.

Results

The minor allele frequency for the Arg200Trp polymorphism was 0.12 in the control group compared with 0.14 in the group with radiographic OA of the hip (P = 0.12), and the minor allele frequency for the Arg324Gly variant was 0.083 in the control group compared with 0.088 in the group with radiographic OA of the hip (P = 0.63). The multilocus genotypes available in 1,886 subjects suggested that inheritance of both minor alleles was a risk factor for developing OA characterized by JSN (P < 0.01). Patients with radiographic OA of the hip who were homozygous for the Arg200Trp minor allele had higher serum FRP levels than controls who were homozygous for the major allele.

Conclusion

Our data confirm findings of another study, that a rare haplotype with both Arg200Trp and Arg324Gly FRZB variants contributes to the genetic susceptibility to hip OA among Caucasian women, and that these polymorphisms may contribute to increased serum levels of proteins as biomarkers of OA.

The prevalence of symptomatic hip osteoarthritis (OA; OMIM 165720) is ∼3% among adults age ≥30 years in the US and accounts for a significant health care burden (1). Multiple risk factors, including age, obesity, joint injury, occupational factors, sex, and ethnicity, among others, have been well established for hip OA (1). Furthermore, genetic factors are believed to be associated with the development and progression of OA. Although several biomarkers for bone turnover and cartilage damage have been investigated, definitive markers for the diagnosis and progression of OA remain elusive. Biomarkers that are associated with a basic understanding of the genetic epidemiology and pathophysiology of OA could potentially lead to new diagnostic, prognostic, and therapeutic targets.

Investigators in several twin and family studies have reported a significantly increased risk for hip OA among siblings of affected individuals (2, 3). In addition, genome-wide scans, fine-scale mapping, and candidate gene association analyses have identified several loci that may be associated with hip OA (2, 3). One such locus was identified by 2 separate genome-wide scans for familial OA susceptibility (4, 5), and finer mapping suggested a peak linkage signal at D2S2284 (2q31.1) (6). Further single-nucleotide polymorphism (SNP) analysis of 8 candidate genes in this region demonstrated an association of hip OA with a functional SNP resulting in an Arg324Gly substitution at the penultimate residue of Frizzled-related protein (FRP) (7). In addition, a haplotype coding for substitutions of 2 highly conserved arginine residues (Arg200Trp and Arg324Gly) in FRZB was a strong risk factor for primary hip OA, with an odds ratio (OR) of 4.1 (7).

FRP is a member of a family of soluble Wingless (Wnt) antagonists (8, 9). Recent evidence has demonstrated that products of Wnt, Frizzled, secreted FRP (sFRP), and low-density lipoprotein receptor–related protein gene families play key roles in the development and maintenance of bones and joints (10–13). Consequently, polymorphisms of genes that alter protein expression or function in the Wnt signaling pathway are potential biomarker candidates for OA susceptibility and progression. We further investigated the association of the FRZB variants Arg200Trp and Arg324Gly in a large cohort of older postmenopausal Caucasian women with radiographic hip OA. Patients with radiographic hip OA were further subdivided into phenotypes characterized by bone formation with osteophytes or cartilage loss with joint space narrowing (JSN). The sera from patients and controls who were homozygous for Arg200Trp were then assayed for circulating levels of FRP.

PATIENTS AND METHODS

Study population.

All subjects were participants in the Study of Osteoporotic Fractures (SOF), a multicenter cohort study initiated in 1986 to determine risk factors for osteoporotic fractures in elderly women (14). Participants were all age ≥65 years at baseline and were recruited from population-based listings at 4 clinical centers in the US: Baltimore, MD; Minneapolis, MN; Monongahela Valley, PA (near Pittsburgh); and Portland, OR. Exclusion criteria for the parent study, the SOF, included bilateral hip replacement and an inability to walk unassisted; African American women were excluded because of their low risk of hip fracture. The study was approved by the institutional review boards at each of the institutions involved. All subjects provided written informed consent at enrollment and at each clinical examination.

A buffy coat specimen sufficient for genotyping was available from 6,975 participants who provided adequate consent for genetic studies (Figure 1). The present analysis was based on women with available genotype data who also had data on radiographic hip OA from baseline and visit-5 (mean 8.3 years after baseline) pelvis radiographs (15, 16).

Figure 1.

Flow chart of study subjects. SOF = Study of Osteoporotic Fractures; RHOA = radiographic osteoarthritis of the hip; JSN = joint space narrowing.

Genotyping.

Genotypes for FRZB Arg200Trp (rs288326) and Arg324Gly (rs7775) SNPs were obtained by 2 different protocols (Roche Molecular Systems, Alameda, CA). The Arg324Gly SNP was genotyped for the entire SOF cohort in conjunction with another study of osteoporosis, to conserve the patient material. The Arg200Trp SNP genotypes were separately determined for the patients with radiographic hip OA and ∼2.5 times as many randomly selected controls. Hence, genotypes of 570 patients with radiographic OA of the hip and 1,317 controls were obtained for the Arg200Trp SNP, and genotypes of 569 patients with radiographic OA of the hip and 4,136 controls were obtained for the Arg324Gly SNP.

The FRZB SNP rs288326 genotype was assayed by an allele-specific kinetic polymerase chain reaction (PCR) method previously described (17, 18). Two allele-specific primers for the noncoding strand and 1 common primer were used: allele G–specific primer, 5′-TTGTCTTTTATCCCAGTCATTC-3′; allele A–specific primer, 5′-TTGTCTTTTATCCCAGTCATTT-3′; and the common primer, 5′-CATCATGGCACTTAGTCTTTATCTC-3′. The DNA samples were amplified using the GeneAmp 5700 Sequence Detection System (Applied Biosystems, Foster City, CA) measuring fluorescence in real time, as follows: 2 minutes at 50°C for uracil N-glycosylase–mediated degradation of any carryover PCR product contamination, 12 minutes at 95°C for activation of CEA2 Gold polymerase (provided by David Birch, Roche Molecular Systems), then 45 cycles of 2-step amplification of 20 seconds at 95°C for denaturation and 20 seconds at 58°C for annealing/extension. Melting curves were obtained from 35 subsequent temperature increments, with +1°C stepwise increments for 1 minute beginning at 60°C. The genotypes were analyzed with software developed at Roche Molecular Systems. Each amplification plate used 3 positive controls with 3 known genotypes and 1 negative control. Two hundred seventy-six samples were genotyped in duplicate and with complete concordance.

The Arg324Gly polymorphism was genotyped in the context of a multiplex PCR amplification followed by allele-specific SNP detection with immobilized oligonucleotide probes in linear arrays, essentially as described previously (19). Primers were modified at the 5′ phosphate by conjugation with biotin. Purified human genomic DNA (10–50 ng) was amplified in a reaction volume of 50 μl with AmpliTaq Gold DNA polymerase using a GeneAmp PCR System 9600 thermal cycler (PE Biosystems, Foster City, CA) with the following cycling profile: an initial hold at 94°C for 7 minutes, then 33 2-step amplification cycles of 15 seconds at 95°C for denaturation and 60 seconds at 60°C for annealing/extension, and a final 5-minute product extension step at 68°C. Chromogenic detection of allelic variants following stringent hybridization of the biotinylated PCR products to the immobilized sequence-specific probes was performed on a Profiblot II T24 (Tecan, Research Triangle Park, NC). Roche Molecular Systems in-house software was used to scan the linear arrays on a Perfection 1670 scanner (Epson, Long Beach, CA) and to assign genotypes. Sixteen percent of samples were genotyped twice for this SNP, and all results were concordant.

Radiographic assessment.

Details of radiographic scoring methods for hip OA have been described previously (9, 16). Subjects were considered to have radiographic hip OA if they satisfied at least 1 of the following 3 criteria in an individual hip based on the reading from the visit-5 radiograph: severe JSN (Osteoarthritis Research Society International [OARSI] atlas [20] feature grade ≥3), definite osteophytes and JSN (both OARSI grades ≥2), or any 3 or more individual radiographic features (summary grade ≥3). For patients who underwent a total hip replacement between baseline and followup, radiographic hip OA status was assessed in the contralateral hip. Controls selected for Arg200Trp SNP genotyping had no radiographic evidence of hip OA and had not had a total hip replacement. The radiographic phenotype of patients was further characterized into those with JSN grade ≥3 (n = 326) in either hip and those who had at least 1 hip with a femoral osteophyte grade ≥2 and JSN grade ≤2 (n = 130) and who did not have severe JSN in either hip (Figure 1). Reproducibility of readings was good, with kappa values for interrater reliability of 0.66 for definite JSN (grade ≥2), 0.95 for severe JSN, 0.71 for definite osteophytes, and 0.65 for summary grades ≥2 (15, 21). For the present analysis, subjects with rheumatoid arthritis, Paget's disease of bone, or prior hip fracture were excluded.

Covariate assessment.

All study participants completed a self-administered questionnaire at the baseline and followup visits that included age, self-reported health status, current medication use (including estrogen, multivitamins, and other supplements), daily physical activity, and smoking history. Height was measured using a wall-mounted Harpenden stadiometer (Holtain, Dyfed, UK), and weight was measured with a balance beam scale; methods have been described previously (14, 15, 22). Bone mineral density (BMD) measurements of the hip were performed using dual x-ray absorptiometry (QDR 1000; Hologic, Waltham, MA), and BMD measurements of the calcaneus were performed using single x-ray absorptiometry (OsteoAnalyzers; Siemens-Osteon, Wahiawa, HI), with protocols and reproducibility that have been described previously (14). Data presented for subject characteristics are from visit 5.

Serum measurements of FRP.

We assayed levels of FRP in all patients with radiographic OA of the hip with available serum who were homozygous for the T allele at Arg200Trp, in an equal number of controls homozygous for the T allele, and in approximately twice as many randomly sampled patients and controls who were homozygous for the C allele. Ninety-six–well plates (Costar, Cambridge, MA) were coated with 1 μg/ml goat anti-human FRP (R&D Systems, Minneapolis, MN) in carbonate buffer (pH 9.5) overnight at 4°C. The plates were washed and blocked for 3 hours with filtered 2% bovine serum albumin (Sigma, St. Louis, MO). The sera were incubated overnight at 4°C diluted 1:10 and 1:30 in 50 ml diluent. The presence of bound FRP was detected with biotinylated goat anti-mouse/human FRP (R&D Systems) and streptavidin–horseradish peroxidase (Zymed, South San Francisco, CA). The reaction with a tetramethylbenzidine substrate (Kirkegaard & Perry, Gaithersburg, MD) was neutralized with 1M phosphoric acid, and the absorbance at 450 nm was read on a multiwell plate reader (Molecular Devices, Sunnyvale, CA). Commercially prepared FRP (R&D Systems) was used as a standard, and concentrations were determined using SoftMax Pro software (Molecular Devices). Each sample was assayed in duplicate in several repeated assays. The lower limit of detection was reliably 3 ng/ml, and we were unable to detect 100 ng/ml purified sFRP-1 in this assay as an assessment of cross- reactivity. The covariance was 5.9% intraassay and 8.2% interassay.

Statistical analysis.

To observe an effect comparable with that previously reported, in which haplotype frequency for the 2 minor alleles was 0.006 in the control group and 0.026 in the OA group, the present study would have to have >80% power with an alpha level of 0.05 for an OR of 3 with 569 patients and 1,317 controls and a population prevalence of radiographic hip OA of 10% (7). For the Arg200Trp polymorphism, based on an observed control frequency in women (0.11), our sample size of 569 patients and 2.5 controls per patient provided >90% power with an alpha level of 0.05 for an OR ≥3 in a recessive model using Quanto version 1.0 (online at http://hydra.usc.edu/gxe/) (23, 24). For the Arg324Gly polymorphism, based on an observed control allele frequency in women (0.07), our sample size of 570 patients and 7 controls per patient provided >80% power with an alpha level of 0.05 for an OR of 3.5 in a recessive model.

Differences in subject characteristics and BMD variables between patients with radiographic hip OA and participants without radiographic hip OA were assessed by Student's t-test for continuous variables and by chi-square test for dichotomous variables. Hardy-Weinberg equilibrium was assessed in both groups by chi-square test, and allele frequencies were calculated using the gene counting method. Multivariate logistic regression was used to estimate ORs and 95% confidence intervals (95% CIs) for each genotype and the risk of radiographic hip OA and its phenotypes. All analyses for the association of FRP Arg200Trp and Arg324Gly genotypes and radiographic hip OA definitions were performed both with and without adjustments for confounding variables including age, height, weight, estrogen use, and femoral neck BMD. Haplotype frequencies were determined using the estimated haplotype program HAP version 3 (25). Finally, the differences in serum FRP levels between all genotypes and radiographic hip OA groups levels were assessed by two-factor analysis of variance. Statistical analysis was performed using the statistical software program SAS version 8.2 (SAS Institute, Cary, NC).

RESULTS

Characteristics of the study subjects.

Among the 4,706 study subjects, 570 satisfied the criteria for radiographic hip OA, of whom 326 had JSN grade ≥3 and 130 had definite femoral osteophytes and no definite JSN (Figure 1). Five hundred sixty-nine of these 570 patients with radiographic hip OA had confirmed genotypes for both SNPs. The 569 patients with radiographic OA of the hip and 1,317 controls were genotyped for both FRZB SNPs to determine haplotype estimates. All 4,706 subjects were genotyped for the Arg324Gly polymorphism as part of a separate study of osteoporosis. Compared with the 1,317 control subjects who did not satisfy the radiographic hip OA criteria and were genotyped for both SNPs, the genotyped patients with radiographic OA of the hip were older and reported less daily physical activity, had poorer overall health status, and had higher femoral neck BMD (Table 1). All other subject characteristics and BMD variables were not significantly different between the 2 groups.

Table 1. Subject characteristics*
 Radiographic hip OA patients (n = 569)Controls (n = 1,317)P
  • *

    Except where indicated otherwise, values are the mean ± SD. Subjects who were genotyped for both minor alleles are included in these analyses. OA = osteoarthritis; BMI = body mass index; BMD = bone mineral density; DXA = dual x-ray absorptiometry.

  • By chi-square test for dichotomous variables and by Student's t-test for continuous variables.

Age, years79.6 ± 4.978.4 ± 4.8<0.01
Height, cm157.6 ± 6.3157.7 ± 6.30.73
Weight, kg66.3 ± 12.866.6 ± 14.00.68
BMI, kg/m226.7 ± 5.026.7 ± 5.20.98
Currently smoking, %4.45.20.48
Physical activity (walking), kcal/week462.0 ± 549.3533.3 ± 623.8<0.05
Age at menopause, years48.4 ± 5.448.1 ± 5.60.40
Current estrogen use, %17.618.50.65
Current vitamin D use, %48.349.90.54
Health status (good/excellent), %73.579.8<0.01
Area BMD by DXA, gm/cm2   
 Total hip0.74 ± 0.130.74 ± 0.160.55
 Femoral neck0.67 ± 0.140.63 ± 0.13<0.01
 Calcaneal0.37 ± 0.100.38 ± 0.100.04

Minor allele and genotype frequencies.

The frequency of the minor allele (T) for the Arg200Trp polymorphism in the entire cohort was 0.12, with frequencies of 0.14 in the patients with radiographic OA of the hip and 0.12 in the controls (P = 0.12) (Table 2). The frequency of the minor allele (G) of the Arg324Gly polymorphism was 0.088 in the patients with radiographic OA of the hip and 0.083 in the controls (P = 0.63). However, the frequency of the G allele encoding Arg324Gly was 0.11 in subjects with severe JSN (P = 0.04 versus controls). Genotype frequencies in the entire cohort did not deviate from Hardy-Weinberg equilibrium and were similar to those reported elsewhere (7, 26). In addition, the 2 SNPs were in linkage equilibrium (D′ = 0.01, r2 = 7.6 × 10−5).

Table 2. Frequency of alleles for the 2 coding single-nucleotide polymorphisms of FRZB*
 Minor allele frequency (no./total no.)P
Radiographic hip OA patientsControls
  • *

    OA = osteoarthritis; JSN = joint space narrowing.

  • By chi-square test for dichotomous variables.

Arg200Trp T allele   
 All0.14 (157/1,138)0.12 (307/2,632)0.12
 JSN0.14 (92/650)0.12 (307/2,632)0.14
 Osteophytosis0.13 (35/260)0.12 (307/2,632)0.51
Arg324Gly G allele   
 All0.088 (100/1,140)0.083 (684/8,272)0.63
 JSN0.11 (71/652)0.083 (684/8,272)0.04
 Osteophytosis0.054 (14/260)0.083 (684/8,272)0.15

The unadjusted OR for radiographic hip OA among those homozygous for the Arg200Trp T allele was 1.84 (95% CI 0.95–3.55) (P = 0.07) (Table 3). There was no association between homozygosity for the FRZB Arg200Trp polymorphism and radiographic hip OA characterized by JSN grade ≥3 (OR 1.43 [95% CI 0.60–3.41], P = 0.42). However, the odds of radiographic hip OA characterized by femoral osteophyte grade ≥2 and JSN grade ≤2 were significantly increased with the presence of the FRZB Arg200Trp polymorphism homozygous mutant (OR 3.37 [95% CI 1.40–8.12], P < 0.01) and remained highly significant after adjustment for potential confounders including age, weight, height, femoral neck BMD, and estrogen use (OR 3.18 [95% CI 1.27–7.96], P = 0.01) (Table 3).

Table 3. Association between the Arg200Trp and Arg324Gly polymorphisms and radiographic hip OA status*
GenotypeNo. of patients/total no. of subjects (%)Unadjusted OR (95% CI)PAge-adjusted OR (95% CI)PAdjusted OR (95% CI)P
  • *

    OA = osteoarthritis; OR = odds ratio; 95% CI = 95% confidence interval; JSN = joint space narrowing; NA = not applicable.

  • Adjusted for age, height, weight, estrogen use, and femoral neck bone mineral density.

Arg200Trp       
 All radiographic hip OA patients569/1,886 (30.1)
  CC428/1,459 (29.3)1.00 (referent)1.00 (referent)1.00 (referent)
  CT125/390 (32.0)1.14 (0.89–1.44)0.301.15 (0.90–1.46)0.271.20 (0.92–1.55)0.18
  TT16/37 (43.2)1.84 (0.95–3.55)0.071.89 (0.97–3.68)0.061.70 (0.85–3.41)0.13
 All patients with osteophytosis130/1,447 (8.7)
  CC102/1,133 (9.9)1.00 (referent)1.00 (referent)1.00 (referent)
  CT21/286 (7.3)0.80 (0.49–1.31)0.370.80 (0.49–1.31)0.380.83 (0.50–1.38)0.48
  TT7/28 (25.0)3.37 (1.40–8.12)<0.013.49 (1.45–8.43)0.0063.18 (1.27–7.96)0.01
 All patients with JSN325/1,642 (19.1)
  CC240/1,271 (18.9)1.00 (referent)1.00 (referent)1.00 (referent)
  CT78/343 (22.7)1.26 (0.95–1.69)0.111.28 (0.96–1.71)0.091.36 (0.99–1.86)0.06
  TT7/28 (25.0)1.43 (0.60–3.41)0.421.43 (0.60–3.43)0.421.21 (0.46–3.14)0.70
Arg324Gly       
 All radiographic hip OA patients570/4,706 (12.1)
  CC475/3,951 (12.0)1.00 (referent)1.00 (referent)1.00 (referent)
  CG90/726 (12.4)1.04 (0.81–1.32)0.781.06 (0.83–1.35)0.651.00 (0.78–1.30)0.99
  GG5/29 (17.2)1.53 (0.58–4.02)0.391.43 (0.54–3.79)0.481.00 (0.30–3.40)0.99
 All patients with osteophytosis130/4,266 (3.0)
  CC116/3,592 (3.2)1.00 (referent)1.00 (referent)1.00 (referent)
  CG14/650 (2.1)0.66 (0.38–1.17)0.160.67 (0.38–1.17)0.160.67 (0.38–1.18)0.17
  GG0/24 (0)NANANA
 All patients with JSN326/4,462 (7.3)
  CC259/3,735 (6.9)1.00 (referent)1.00 (referent)1.00 (referent)
  CG63/699 (9.0)1.33 (1.00–1.77)0.051.37 (1.02–1.82)0.0371.28 (0.94–1.74)0.12
  GG4/28 (14.3)2.24 (0.77–6.50)0.142.10 (0.71–6.10)0.181.26 (0.29–5.48)0.75

None of the patients with radiographic OA of the hip were homozygous for both minor alleles. Five patients with radiographic OA of the hip were homozygous for the Arg324Gly SNP. In the small number of subjects with at least 1 copy of this allele, there was an increased risk of developing JSN (unadjusted OR 1.33 [95% CI 1.00–1.77], P = 0.05).

Haplotype frequencies.

We estimated the frequency of the 4 haplotypes for the 2 exonic SNPs in patients with radiographic OA of the hip and in controls (Table 4). None of the common haplotypes demonstrated an association with increased risk of radiographic hip OA. The TG haplotype, which contains the rare alleles of both SNPs, showed the greatest proportionate increase in frequency and risk (0.02 in controls versus 0.04 in patients with radiographic OA of the hip; unadjusted OR 1.5 [95% CI 1.01–2.22], P < 0.05). This haplotype conferred a particular risk of developing radiographic hip OA with a phenotype characterized by severe JSN (unadjusted OR 1.90 [95% CI 1.22–2.96], P < 0.01). The proportions of subjects who had diplotypes with a T allele at the exon 4 SNP together with a G allele at the exon 6 SNP (Table 4) were 7% for all patients with radiographic OA of the hip (unadjusted OR 1.50 [95% CI 1.00–2.20], P < 0.05), 12% for those with severe JSN (unadjusted OR 2.1 [95% CI 1.42–3.23], P < 0.001), 5% for those with osteophytosis (unadjusted OR 1.07 [95% CI 0.49–2.4], P = 0.83), and 5% for controls. These analyses imply that inheriting copies of both rare SNPs is a particular risk factor for developing radiographic hip OA with a phenotype characterized by severe JSN.

Table 4. Frequency of the 9 possible compound genotypes for the Arg200Trp and Arg324Gly SNPs, and estimated frequency of the 4 haplotypes of the FRZB SNPs in radiographic hip OA patients and controls*
 Number (frequency)
Radiographic hip OA patientsPatients with JSNPatients with osteophytosisControls
  • *

    OA = osteoarthritis; JSN = joint space narrowing.

  • Includes T allele of exon 4 single-nucleotide polymorphism (SNP) and G allele of exon 6 SNP.

  • Numbers are estimated.

Arg200Trp/Arg324Gly diplotype    
 CC/CC375 (0.66)203 (0.62)95 (0.73)877 (0.67)
 CT/CC89 (0.16)51 (0.16)17 (0.13)207 (0.16)
 TT/CC10 (0.02)4 (0.01)4 (0.03)13 (0.01)
 CC/CG49 (0.08)26 (0.08)7 (0.05)148 (0.11)
 CT/CG35 (0.06)34 (0.10)4 (0.03)56 (0.04)
 TT/CG6 (0.01)3 (0.01)3 (0.02)8 (0.01)
 CC/GG4 (0.01)3 (0.01)0 (0.00)6 (0.004)
 CT/GG1 (0.002)1 (0.003)0 (0.00)2 (0.001)
 TT/GG0 (0.00)0 (0.00)0 (0.00)0 (0.00)
Haplotype    
 CC923 (0.81)517 (0.79)218 (0.84)2,165 (0.82)
 TC115 (0.10)62 (0.10)28 (0.11)241 (0.09)
 CG58 (0.05)41 (0.06)7 (0.03)162 (0.06)
 TG42 (0.04)30 (0.05)7 (0.03)66 (0.02)

Serum levels of FRP.

Sixteen patients were homozygous for the Arg200Trp variant, but only 5 patients were homozygous for the Arg324Gly variant. The sera of 11 patients with radiographic OA of the hip who were homozygous for the Arg200Trp variant and that of their controls were tested for FRP levels by enzyme-linked immunosorbent assay. There was a trend for the serum levels of FRP in patients with radiographic OA of the hip to be higher than those in controls (mean ± SD 21.9 ± 11.8 ng/ml versus 17.2 ± 9.9 ng/ml; P = 0.11 by Student's t-test). When the patients and controls were separated by genotype, the presence of radiographic hip OA had a greater influence than genotype on serum FRP levels (Figure 2).

Figure 2.

Serum levels of Frizzled-related protein (FRP) in patients with radiographic osteoarthritis of the hip (RHOA) and in controls, by Arg200Trp genotype. Stored serum samples obtained from patients with radiographic OA of the hip and from controls were randomly chosen within each homozygous genotype for the Arg200Trp polymorphisms and were assayed for FRP levels by sandwich enzyme-linked immunosorbent assay. Each sample was assayed in duplicate in several repeated assays. Shown are the averages of all the replicates for each individual tested. Bars show the group means. FRP levels were significantly higher in patients with radiographic OA of the hip (P = 0.04 by two-factor analysis of variance), but there was no statistical difference between genotypes (P = 0.06), and there was no significant interaction (P = 0.49).

DISCUSSION

Susceptibility to OA is heterogeneous and involves complex genetics. A previous study suggested that susceptibility to primary hip OA in women was strongly associated with a rare haplotype coding for substitutions of 2 highly conserved arginine residues (Arg200Trp and Arg324Gly) in FRZB. Twenty-eight patients had this haplotype compared with only 5 controls, giving an OR of 4.1 (P = 0.004) (7). Our results confirmed that the TG haplotype showed the greatest proportionate increase in frequency and risk (66 controls and 42 patients with radiographic OA of the hip had this haplotype, yielding frequencies of 0.02 and 0.04, respectively). Despite large numbers, the allele frequencies in the main radiographic hip OA population were not significantly different from those in the control population. However, our results suggest that Caucasian women homozygous for the minor allele of the Arg200Trp substitution in FRZB have a 3-fold higher risk of radiographic hip OA characterized by femoral osteophytes without significant JSN. Inheritance of a variant coding for the Arg324Gly substitution is a susceptibility factor for developing radiographic hip OA characterized by severe JSN, and this susceptibility increased severalfold when both minor allele variants were present.

Two previous studies identified an increased risk of OA associated with the Arg324Trp variant (7, 26). In one study, the patients were identified at the time of hip replacement surgery, and an independent association with the minor allele for the Arg200Trp variant was not found (7). However, differences between subject selection in that study and in the current study make direct comparison somewhat difficult. Individuals with bilateral hip replacements were excluded from the cohort in the original SOF study design. In addition, patients with radiographic OA of the hip in the current study were identified radiographically from a cohort of elderly Caucasian women and were characterized to evaluate the genetic association with the phenotype of hip OA, which may partly account for some of the allele frequency differences seen in the radiographic hip OA groups in the 2 studies.

In a Rotterdam population-based study, the Arg324Gly polymorphism has been associated with an increased risk of generalized OA, confirming a risk association and suggesting that these polymorphisms may have a more systemic than anatomic influence (26). The allele frequencies for both variants and their haplotypes were similar to the frequencies in controls in both the UK and Dutch studies (7, 26), indicating no difference in linkage disequilibrium or genetic background. In our study and in the Dutch study, however, carriers of the G allele of the Arg324Gly variant showed no increased risk for the occurrence of radiographic hip OA. In prospective populations, the Arg324Gly variant might accelerate progression of disease, as suggested by the association with a severe JSN phenotype reported here.

The soluble inhibitors of the Wnt signaling pathway have been implicated in the maintenance of adult bone density (11, 27, 28). The regulation of bone homeostasis by sFRP family members probably involves multiple mechanisms. The sFRP family has 2 main protein domains. There is the amino-terminal domain, which is cysteine rich and binds to Wnt as a soluble competitor for Frizzled receptor engagement (9). The midregion, which includes residue 200, has a netrin-like domain. In transfection studies, the Arg200Trp variant had no effect on the ability of FRP to diminish Wnt signaling via nuclear translocation of β-catenin (7). This substitution, however, may have other functional ramifications. Secreted FRP-1 has been shown to bind to RANKL (29), and it binds to a specific peptide motif that is not found in the Wnt family (30). An alternative ligand binding domain may explain a mechanism for the ability of FRP to depress proliferation in osteoblasts and increase osteoblast differentiation independent of signaling activity through β-catenin (31).

OA cartilage displays signs of Wnt/catenin activation and expresses FRP (7, 32, 33). Ectopic canonical Wnt signaling leads to enhanced ossification and suppression of chondrocyte formation (34). Activation of β-catenin in mature cartilage cells stimulated hypertrophy, matrix mineralization, and expression of terminal markers such as matrix metalloproteinase 13 (MMP-13) and vascular endothelial growth factor (35). In addition, β-catenin overexpression in chondrocytes strongly stimulated expression of matrix degradation enzymes such as MMP-2, MMP-3, MMP-7, MMP-9, membrane type 3 MMP, and ADAMTS-5. Thus, Wnt/catenin signals may activate cartilage matrix catabolism and may have roles in cartilage destruction under pathologic conditions.

The presence of FRP in the serum suggests that a homeostatic control mechanism may be augmenting the expression of FRP to counterbalance the activity of Wnt and β-catenin. Elevated FRP was detected in patients who did not carry either minor allele, suggesting that this is a physiologic response. The cellular source of the FRP remains to be determined; FRP may arise from cartilage itself or possibly from other mesenchymally derived cells. The further serum elevations of FRP in patients with radiographic hip OA who were homozygous for the T allele may have been due to a reduced ability of the protein to bind to extracellular matrix or cartilage components, such as the heparan sulfate proteoglycans (36–38).

The present study offers several strengths, including its large community-based cohort of elderly Caucasian women and a validated radiographic scoring system making phenotype-specific definitions to fully characterize the extent of radiographic hip OA. However, there are also several potential limitations. Individuals with bilateral hip replacements were excluded from the study, so we may have excluded subjects with the risk-conferring genotype. We divided our patients with radiographic hip OA into phenotypic subgroups and did not adjust our analyses for multiple comparisons, which increases the possibility that marginally significant findings are due to chance. Therefore, the JSN results should be interpreted with caution, and replication by others would be warranted. In addition, this study was performed in Caucasian women age ≥65 years, and our findings may not be generalizable to other populations.

In summary, the Arg200Trp TT genotype in the netrin domain of FRP was associated with increased risk of radiographic hip OA specifically characterized by femoral head osteophyte development. The TG haplotype was a risk factor for developing radiographic hip OA characterized by severe JSN. The presence of radiographic hip OA may influence the serum levels of proteins such as FRP. Biomarkers that reflect genetic susceptibility could be used in conjunction with other parameters of bone turnover and cartilage damage to ascertain the extent of disease and investigate the underlying defect.

Acknowledgements

We are grateful to K. Pekny for technical assistance and to N. Noon for secretarial support. We thank Dr. Dennis Carson for critical reading of the manuscript and Dr. John Loughlin for fruitful and insightful discussion. We appreciate the assistance of Dr. Eleazar Eskin and Hyun Min Kang with distance calculations.

Ancillary