1. Top of page
  2. Abstract
  6. Acknowledgements


Type IX collagen proteoglycan is an important protein in collagen networks and has been implicated in hip osteoarthritis (OA). We studied 2 COL9A1 markers (509-8B2 and 509-12B1) in relation to radiographic OA, within the framework of the Rotterdam Study, a population-based study of 7,983 subjects ages 55 years and older.


We used 2 different designs, as follows: 1) a linkage study of 83 probands with multiple joints affected with radiographic OA and their 221 siblings, yielding 445 sibpairs who participated in the study, and 2) an association study in a series of 71 patients with radiographic hip OA and 269 controls without radiographic OA. All subjects were characterized for the 2 COL9A1 markers, 509-8B2 and 509-12B1. The mean test was used to assess the proportion of alleles shared in concordantly affected and unaffected sibpairs. The chi-square test was used to compare the allele distributions in patients and controls.


Affected sibpairs with radiographic hip OA shared alleles identical by descent at markers 8B2 and 12B1 significantly more often than expected (mean ± SD 0.66 ± 0.07 and 0.65 ± 0.08, respectively; P < 0.05). No excess sharing for radiographic OA was observed at other joint sites. When comparing the frequency of marker 8B2 and 12B1 alleles in subjects with radiographic OA and controls, the frequency of 8B2 alleles in subjects with radiographic OA differed significantly(P = 0.01) from that in controls.


Our data suggest that susceptibility for hip OA is conferred within or close to the COL9A1 gene in linkage disequilibrium with the COL9A1 509-8B2 marker.

Osteoarthritis (OA) is a complex disorder that is common worldwide (1) and is the leading cause of disability and pain among the elderly (2). Results of family-based and candidate gene studies demonstrated a clear genetic component, particularly for early-onset primary OA (3, 4). One of the main pathologic characteristics of OA is degradation of hyaline cartilage. The collagen fibril network is one of the main components of hyaline cartilage, and this network maintains the integrity of hyaline cartilage and prevents its degradation (5). Type IX collagen links type II collagen–containing fibrils to the rest of the cartilage matrix and thus plays a role in cartilage integrity (6). Type IX collagen is composed of 3 genetically distinct α polypeptide chains, i.e., α1(IX), α2(IX), and α3(IX), which are encoded by COL9A1, COL9A2, and COL9A3, respectively (6).

A deficiency of polypeptide chain α1(IX) has been shown to lead to functional abnormalities in type IX collagen fibrils, which lead to instability of hyaline cartilage (7). This observation suggests that mutations in the COL9A1 gene that lead to a nonfunctional α1(IX) polypeptide may be implicated in OA. Some evidence supports this view. OA develops in transgenic mice that express a nonfunctional protein as well as in knockout mice (8), which suggests that COL9A1 may be a candidate gene for OA in humans. Some evidence suggests that COL9A1 is involved in hip OA (9, 10). Studies of affected sibpairs demonstrated linkage of COL9A1 to severe forms of hip OA in women (9, 10). However, association studies failed to show any relationship. Another question that remains to be answered is whether COL9A1 is involved in joints other than the hip.

In the present study, we investigated 2 polymorphisms in the COL9A1 gene (509-8B2 and 12B1) in relation to radiographic OA at different joint sites, in 2 independent studies: a sibpair study including 445 sibpairs with radiographic hip, knee, or hand OA or spinal disk degeneration, and an association study of 71 patients with radiographic hip OA and 269 controls without radiographic OA.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Study population.

The present study was conducted within the framework of the Rotterdam Study, a population-based cohort study of chronic diseases in 7,983 subjects ages 55 years and older (11). The medical ethics committee of the Erasmus Medical Center approved the study, and written informed consent was obtained from all participants. Baseline examinations took place between 1990 and 1993 by means of a structured interview in which a standardized questionnaire was used. Figure 1 presents a flow chart of participation of the study population. From the total cohort of subjects ages 55 to 65 years (n = 2,593), a random cohort including 944 noninstitutionalized persons was drawn, and subjects were scored for radiographic OA of the hip, knee, or hand and for disk degeneration of the spine.

thumbnail image

Figure 1. Flow chart showing participation of probands and siblings in the linkage study and cases and controls in the association study. ROA = radiographic osteoarthritis.

Download figure to PowerPoint

Radiographic examination.

Radiographs of the hip, knee, and hand joints of participants of the Rotterdam Study and their siblings were scored for the presence of radiographic OA. Radiographs of the spine were evaluated for the presence of disk degeneration using the standard Kellgren/Lawrence (K/L) grading system (12). A diagnosis of radiographic OA was considered for any joint with a K/L score of ≥2. Two independent readers scored all radiographs. After assessment of each set of ∼150 radiographs, the scores of the 2 readers were evaluated. Whenever the scores differed by ≥2 points, or when the score given by one reader was 1 and the score given by the other reader was 2, a consensus was agreed upon. All radiographs were scored before genotyping, with the readers blinded to clinical data.

Linkage study.

For the linkage analysis, probands were derived from the random cohort. Persons who had radiographic OA at ≥2 joint sites of the 4 joint groups (i.e., hips, knees, hands, or spine) were selected as probands. Individuals with both radiographic hand OA and disk degeneration of the spine, which was the most common combination observed, also had to have Heberden's nodes in order to be included as probands. This criterion was applied to maximize the probability of a subject having a genetic form of radiographic OA. A total of 221 probands (response rate 88%) were willing to contribute to the study, yielding 708 live-borne siblings (Figure 1). Four hundred fifty siblings of 101 probands were not eligible for the study due to sibling death, refusal, emigration, disease, or nonresponse. In total, 258 siblings and 120 probands derived from 120 pedigrees were included in the study (Figure 1). The siblings were examined at the research center, using the same protocol and methods that were used to examine participants in the random cohort.

Association study.

Within the random cohort, 72 persons with radiographic hip OA were genotyped. The 269 persons who did not have radiographic OA of the hip, knee, or hand joints were selected as controls (Figure 1).

Genotyping for COL9A1 markers 509-8B2 and 12B1.

Participants were genotyped for COL9A1 509-8B2 and 12B1 short tandem repeat polymorphisms (STRPs) according to the protocol described by Warman and colleagues (13). In the sibpair study, genotyping was successful for 85 probands and 241 siblings. In the association study, genotyping was successful for 71 cases and all controls, except for 8B2 in 1 control subject (Figure 1).

Statistical analysis.

Linkage study.

Familial relationships between siblings were confirmed using genealogic data. Six half-sibs were excluded from the analysis (Figure 1). Mendelian inconsistency in pedigrees was checked using the MARKERINFO module. Given the sibling genotypes in nuclear families, this module reconstructs sibling genotype sets and thereafter the parental genotypes. Pedigrees with Mendelian inconsistency are identified whenever 1 or 2 alleles of the studied markers in any sibling do not match with the family genotype sets. Two probands and 14 full siblings who belonged to 4 pedigrees and had Mendelian inconsistencies in 1 or both of the 2 markers were excluded from the analysis (Figure 1). The remaining 83 probands and 221 siblings (belonging to 100 pedigrees) yielded a total of 445 sibpairs.

Sibpairs were classified as follows: concordantly affected (both siblings had radiographic OA), concordantly unaffected (neither sibling had radiographic OA), and discordant (1 sibling was affected, and the other sibling was unaffected at the studied joint site). We used the mean test, which is a powerful test for additive inheritance, to compare the average proportion of alleles shared identical by descent (IBD) with the expected value of 0.5 (14). On average, sibpairs share half of alleles of a given locus IBD. Concordantly affected sibpairs should share >50% of alleles IBD at COL9A1 if this locus is linked to radiographic OA. The analysis was adjusted for age and sex, which are the 2 major determinants of OA. Sibpair data were analyzed using the program Statistical Analysis for Genetic Epidemiology, version 4.4 (New Orleans: Louisiana State University Medical Center).

Association study.

Allele and genotype frequencies for the 8B2 and 12B1 markers were estimated by counting alleles and estimating the sample proportion. Allele and genotype proportions were tested for Hardy-Weinberg equilibrium. The chi-square test was used to compare allele frequencies between subjects with radiographic OA and controls.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Tables 1 and 2 show the characteristics of the study population. Compared with probands, the mean age of siblings was significantly higher (P < 0.001), and the mean bone mineral density was significantly lower (P < 0.05). In the final analysis, each pedigree contributed, on average, 4.5 (range 1–36) sibpairs to the linkage study (Table 2). Among the probands, 33% had radiographic hip OA, 78% had radiographic knee OA, 78% had radiographic OA of the hand joints, and 64% had spinal disk degeneration. Among the siblings, 7% had radiographic hip OA, 19% had radiographic knee OA, 75% had radiographic OA of the hand joints, and 79% had spinal disk degeneration. In the association study, no significant differences in sex, body mass index, or bone mineral density between patients with radiographic hip OA and controls were observed. Patients were slightly (1 year) older than controls (P = 0.05). The allele and genotype proportions were in Hardy-Weinberg equilibrium.

Table 1. Characteristics of the study population*
CharacteristicLinkage studyAssociation study
  • *

    BMI = body mass index; BMD = bone mineral density.

  • P < 0.05 versus siblings.

  • P ≤ 0.05 versus controls.

Age, mean ± SD years60.90 ± 2.7165.80 ± 8.0260.76 ± 2.4359.71 ± 2.84
Women, %69.2250.2541.6649.07
BMI, mean ± SD kg/m227.36 ± 4.2326.71 ± 4.0126.45 ± 3.4625.65 ± 3.26
BMD, mean ± SD cg/cm20.91 ± 0.130.86 ± 0.140.90 ± 0.130.87 ± 0.13
Table 2. Frequency of families by the number of sibpairs
No. of sibpairsNo. of familiesNo. of sibpairs contributed

Table 3 shows the results of the linkage analysis in affected and unaffected sibpairs. Affected sibpairs with radiographic hip OA (n = 11) had significantly increased allele sharing (P < 0.05) in the COL9A1 8B2 (mean ± SD 0.66 ± 0.07) and 12B1 (0.65 ± 0.08) markers shared IBD. The 11 sibpairs with radiographic hip OA belonged to 9 families consisting of a total of 19 siblings (1 family contributed 3 affected siblings). Among the sibpairs with radiographic OA at the hip joints, 3 pairs were homozygous for the COL9A1 8B2 allele 5/allele 6 genotype (i.e., both siblings had the 5/6 genotype), 2 pairs were homozygous for 5/5, and 1 pair was homozygous for 4/6. The remaining sibling pairs were heterozygous for 8B2 (i.e., 2 pairs had a 5/5 and 5/6 genotype set, 1 pair had 5/2 and 5/6, 1 pair had 5/4 and 9/4, and 1 pair had 5/6 and 9/6). When considering the 12B1 marker, 2 sibpairs were homozygous for the 4/6 genotype, 1 pair for the 4/8 genotype, and 1 pair for the 4/4 genotype. The rest of the sibpairs were heterozygous for 12B1 (i.e., 2 sibpairs had 4/4 and 4/8 genotype sets, 2 pairs had 4/6 and 5/6, 1 pair had 4/8 and 8/8, 1 pair had 4/4 and 4/6, and 1 pair had the 3/6 and 3/5 genotype set). No significant differences for the other joints were observed. The numbers of alleles shared in affected and unaffected sibpairs were extremely similar, suggesting that there is no evidence for a role of COL9A1 in radiographic OA of other joints.

Table 3. Frequency of COL9A1 509-8B2 and 509-12B1 alleles shared IBD according to site of radiographic OA*
COL9A1 marker/sibpair phenotypeJoint site with radiographic OA
  • *

    Values are the mean ± SD (n). Data for discordant sibpairs are not presented.

  • P < 0.05, indicating a significant increase in the proportion of alleles shared identical by descent (IBD), based on the expected value of 0.5.

 Concordantly affected0.66 ± 0.07 (11)0.49 ± 0.05 (41)0.51 ± 0.02 (251)0.52 ± 0.02 (243)
 Concordantly unaffected0.50 ± 0.02 (327)0.51 ± 0.02 (205)0.52 ± 0.06 (26)0.54 ± 0.06 (39)
 Concordantly affected0.65 ± 0.08 (11)0.50 ± 0.05 (30)0.50 ± 0.02 (251)0.50 ± 0.02 (243)
 Concordantly unaffected0.49 ± 0.02 (327)0.49 ± 0.02 (212)0.52 ± 0.07 (26)0.50 ± 0.06 (39)

The frequency of the 8B2 or 12B1 alleles was not significantly different between controls without radiographic OA and the total population. Table 4 shows the frequency of the 8B2 and 12B1 alleles according to the presence of radiographic hip OA. The frequency of 8B2 alleles differed significantly (P = 0.01) between subjects with radiographic hip OA and controls without radiographic OA. The frequency of 12B1 alleles was not significantly different between subjects with radiographic hip OA and controls without radiographic OA.

Table 4. Frequency of COL9A1 509-8B2 and 12B1 alleles according to the presence of radiographic hip OA*
 Radiographic hip OAP
  • *

    Values are the number (frequency). OA = osteoarthritis.

  • Alleles with a frequency of <0.05.

COL9A1 marker 509-8B2  0.01
 Allele 550 (0.35)231 (0.43) 
 Allele 627 (0.19)140 (0.26) 
 Allele 714 (0.10)29 (0.05) 
 Allele 810 (0.07)22 (0.04) 
 Allele 920 (0.14)62 (0.12) 
 Other21 (0.15)52 (0.10) 
COL9A1 marker 509-12B1  0.10
 Allele 214 (0.10)96 (0.18) 
 Allele 450 (0.35)172 (0.32) 
 Allele 527 (0.19)67 (0.12) 
 Allele 615 (0.11)44 (0.08) 
 Allele 75 (0.03)33 (0.06) 
 Allele 823 (0.16)92 (0.17) 
 Other8 (0.06)34 (0.06) 


  1. Top of page
  2. Abstract
  6. Acknowledgements

In this population-based study, we observed that affected sibpairs with radiographic hip OA shared a significantly higher number of alleles IBD at 2 markers at COL9A1 (the 8B2 and 12B1 STRPs). Furthermore, in the association study, we observed that the 8B2 marker was significantly associated with radiographic hip OA.

The positive linkage of the COL9A1 locus in our sibpairs confirmed earlier findings of linkage in female sibpairs with hip OA (9, 10), although we could not stratify for sex because the numbers were too low for a meaningful statistical analysis. Despite the fact that in our study the number of sibpairs was small, the excess of sharing was statistically significant. Also, in our association study, we observed a significant relationship between the COL9A1 8B2 marker and radiographic hip OA. The relevance of our finding is not completely clear, because the significance was marginal, various alleles together contribute to the association, and no association with the nearby 12B1 marker was observed. One previous association study on the relationship between the COL9A1 markers 8B2 and 12B1 and radiographic OA has been reported. In that study, no association of 8B2 or 12B1 with severe hip OA was observed in 146 women selected from families with OA (9).

Two important points should be considered when interpreting the difference between our findings and those of Loughlin and colleagues (9). First, in contrast to a linkage study, the relationship in an association study can be easily missed because the marker used in the 2 studies is not very powerful for association analysis due to a large number of rare alleles. The genetic information content of a marker depends on the heterozygosity index, a function of marker allele frequencies, as well as the location of the marker on genome map and the functional effect of the marker variants. In the present study, the polymorphic nature of the studied markers resulted in multiple strata of cases and controls, thus demolishing the power of the association study. Second, although we hypothesize that the COL9A1 locus contributes to OA susceptibility, it is likely that the 8B2 marker is not causally related to radiographic OA. Marker 8B2 is located in COL9A1 intron 4, which resides 17.7 kb downstream of the start of a 65-kg haplotype block within COL9A1. This haplotype block is encompassed between intron 1 (−501) and intron 34 (+32) (9). Thus, marker 8B2 may be in strong linkage disequilibrium with other COL9A1 mutations. Marker 12B1 is located 14.3 kb upstream of exon 1 and resides outside the COL9A1 haplotype block. Furthermore, COL9A1 mapped to a region where other fibril-associated collagen with interrupted triple helix (FACIT)–like collagens (e.g., COL19A1) (15) have been also mapped. Although the association of marker 8B2 with hip OA might be explained by linkage disequilibrium with adjacent loci, which suggests that an OA susceptibility locus may map near to the COL9A1 locus, several experimental studies support the role of the COL9A1 locus in OA (7, 8). Those studies (7, 8) showed that early-onset OA developed in COL9A1–knockout mice.

Taken together with earlier findings, our results suggest that OA susceptibility may map within or near the COL9A1 gene, with 509-8B2 being simply a marker for this. Our sibpair data showed no evidence for a role of COL9A1 in other forms of OA. Further studies are necessary to identify the underlying mutation in COL9A1 or within a nearby OA susceptibility locus.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We greatly appreciate the contributions to the Rotterdam Study from the general practitioners and pharmacists of the Ommoord district.


  1. Top of page
  2. Abstract
  6. Acknowledgements
  • 1
    Lawrence RC, Helmick CG, Arnett FC, Deyo RA, Felson DT, Giannini EH, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum 1998; 41: 77899.
  • 2
    Odding E, Valkenburg HA, Algra D, Vandenouweland FA, Grobbee DE, Hofman A. Associations of radiological osteoarthritis of the hip and knee with locomotor disability in the Rotterdam Study. Ann Rheum Dis 1998; 57: 2038.
  • 3
    MacGregor AJ, Antoniades L, Matson M, Andrew T, Spector TD. The genetic contribution to radiographic hip osteoarthritis in women: results of a classic twin study. Arthritis Rheum 2000; 43: 24106.
  • 4
    Felson DT, Couropmitree NN, Chaisson CE, Hannan MT, Zhang Y, McAlindon TE, et al. Evidence for a Mendelian gene in a segregation analysis of generalized radiographic osteoarthritis: the Framingham Study. Arthritis Rheum 1998; 41: 106471.
  • 5
    Mendler M, Eich-Bender SG, Vaughan L, Winterhalter KH, Bruckner P. Cartilage contains mixed fibrils of collagen types II, IX, and XI. J Cell Biol 1989; 108: 1917.
  • 6
    Diab M, Wu JJ, Eyre DR. Collagen type IX from human cartilage: a structural profile of intermolecular cross-linking sites. Biochem J 1996; 314: 32732.
  • 7
    Hagg R, Hedbom E, Mollers U, Aszodi A, Fassler R, Bruckner P. Absence of the α1(IX) chain leads to a functional knock-out of the entire collagen IX protein in mice. J Biol Chem 1997; 272: 206504.
  • 8
    Fassler R, Schnegelsberg PN, Dausman J, Shinya T, Muragaki Y, McCarthy MT, et al. Mice lacking α1 (IX) collagen develop noninflammatory degenerative joint disease. Proc Natl Acad Sci U S A 1994; 91: 50704.
  • 9
    Loughlin J, Mustafa Z, Dowling B, Southam L, Marcelline L, Raina SS, et al. Finer linkage mapping of a primary hip osteoarthritis susceptibility locus on chromosome 6. Eur J Hum Genet 2002; 10: 5628.
  • 10
    Mustafa Z, Chapman K, Irven C, Carr AJ, Clipsham K, Chitnavis J, et al. Linkage analysis of candidate genes as susceptibility loci for osteoarthritis-suggestive linkage of COL9A1 to female hip osteoarthritis. Rheumatology (Oxford) 2000; 39: 299306.
  • 11
    Hofman A, Grobbee DE, de Jong PT, van den Ouweland FA. Determinants of disease and disability in the elderly: the Rotterdam Elderly Study. Eur J Epidemiol 1991; 7: 40322.
  • 12
    KellgrenJH, LawrenceJS, editors. The epidemiology of chronic rheumatism, atlas of standard radiographs. Oxford: Blackwell Scientific; 1963.
  • 13
    Warman ML, Tiller GE, Polumbo PA, Seldin MF, Rochelle JM, Knoll JH, et al. Physical and linkage mapping of the human and murine genes for the α1 chain of type IX collagen (COL9A1). Genomics 1993; 17: 6948.
  • 14
    Guo X, Elston RC. Two-stage global search designs for linkage analysis I: use of the mean statistic for affected sib pairs. Genet Epidemiol 2000; 18: 97110.
  • 15
    Yoshioka H, Zhang H, Ramirez F, Mattei MG, Moradi-Ameli M, van der Rest M, et al. Synteny between the loci for a novel FACIT-like collagen locus (D6S228E) and α1 (IX) collagen (COL9A1) on 6q12-q14 in humans. Genomics 1992; 13: 8846.