A Mutation in Cartilage Oligomeric Matrix Protein (COMP) Causes Early-Onset Osteoarthritis in a Large Kindred Study


Professor Yuan-Tsong Chen, Institute of Biomedical Sciences, Academia Sinica, 115 Taipei, Taiwan.Tel: 886–2-27899007; Fax: 886–2-27825573; E-mail: chen0010@ibms.sinica.edu.tw


We performed a genome-wide linkage analysis to identify susceptibility loci in a large six-generation extended family previously reported with early-onset osteoarthritis (OA) DNA sequencing was performed to investigate involvement of the COMP (Cartilage oligomeric matrix protein) gene in this family. The region covering D19S884, D19S226, and D19S414 on chromosome 19p following genome-wide scan from 70 individuals of this kindred showed significant linkage, with a maximum point LOD (logarithm of the odds ratio) score of 2.51 at D19S226. Direct sequencing of the COMP gene, the most plausible candidate gene in the region, identified a c.2152C>T substitution in exon 18 which resulted in a substitution of tryptophan for arginine at position 718 located in the C terminal globular domain of the gene product. A total of 26 individuals were identified with this mutation of which 21 affected individuals had the mutation, and the other five younger individuals (18.6 ± 11.3 years of age) carried the mutation without symptoms. The results indicate that COMP is the disease susceptibility gene and the c.2152C>T mutation in exon 18 could cause early-onset OA phenotypes in this kindred, which is compatible with a previous report that this mutation also causes a mild form of multiple epiphyseal dysplasia (MED).


Osteoarthritis (OA) is a degenerative disease of the joint characterized by degradation of the articular cartilage, remodeling of subchondral bone, osteophyte formation and sclerosis (Poole, 1999). Diverse pathogenetic mechanisms have been implicated in the severity and progression of OA, and there is a strong genetic association with early-onset familial OA (Spector et al., 1996; MacGreger et al., 2000). Early-onset OA refers to the condition developing before the age of 40, though symptoms usually occur gradually over time. In rare cases an individual sees symptoms as early as 25 years of age. It usually takes 10 to 15 years from the time of the first symptoms for the disease to become serious and debilitating. A number of candidate genes, predominantly those that encode structural proteins of the extracellular matrix, have been identified as potential susceptibility loci for idiopathic OA. Research on genetic loci involved in familial OA has particularly emphasized the COL2A1 gene, which encodes type II collagen, the major collagen of articular cartilage (Williams et al., 1995). An analysis of degenerative arthritic disorders has shown that an Arg75-Cys mutation in the COL2A1 gene is associated with childhood-onset OA (Lopponen et al., 2004).

In addition to COL2A1, a number of genes linked to cartilage protein or cartilage-related metabolic pathways, including COL1A1, COL9A1, COL9A2, COL9A3, COL11A1, COL11A2, VDR (vitamin D receptor), ER (estrogen receptor), IGF-1 (insulin-like growth factor 1), aggrecan, and TGFβ1, have been identified as susceptibility candidate genes in association analyses (Meulenbelt et al., 1998; Ushiyama et al., 1998; Meulenbelt et al., 1999; Uitterlinden et al., 2000; Yamada et al., 2000, Jackson et al., 2010). Given the pathogenesis of OA, we considered that structural proteins of the matrix, including cartilage oligomeric matrix protein (COMP), and proline/arginine-rich and leucine-rich repeat protein (PRELP) could also play a role in the disease (Posey & Hecht, 2008). There is an inverse relationship between hypermobility and hand and knee OA, and hypermorbility is associated with lower serum COMP levels. Genetic variations of the COMP gene may account for some subgroups of benign joint hypermobility (Chen et al., 2008a).

We previously reported one large extended six-generation kindred with early-onset OA and autosomal dominant inheritance. The clinical phenotypes of this family resembled the primary generalized early-onset OA caused by COL2A1 mutation; however, COL2A1 was excluded as the causative gene after linkage analysis and direct sequencing of COL2A1 (Mu et al., 2009). In this study, we performed a genome-wide linkage analysis followed by candidate gene sequencing to identify the susceptibility gene in this large extended family with a familial form of OA.

Materials and Methods


A six-generation family with autosomal dominant inheritance of early-onset OA (Fig. 1) was identified and recruited at the National Taiwan University Hospital as previously reported (Mu et al., 2009). The kindred consisted of 274 members from two pedigrees that shared the same ancestor through the proband's great grandmother. Seventy cases from this family were recruited for phenotypic evaluation and genotyping.

Figure 1.

A six-generation kindred with early-onset OA. Those numbered were enrolled subjects. Arrow indicates the proband (IV 1). OA phenotype without COMP mutation (c.2152C>T): IV 15, IV 58, V 41, and V 16; COMP mutation (c.2152C>T) without symptoms: VI 50, VI 26, VI 40, VI 51, and VI 68

Phenotype Assessment

All recruited cases were evaluated by the previously reported physician (S.C. Mu) to eliminate interexaminer variability (Mu et al., 2009). Height, weight, body-mass index, arm span, blood pressure, blood uric acid, and rheumatoid factor levels were recorded for these cases. All but two of the 25 affected patients (IV 18 and IV 44) consented to the radiological evaluation. Anterior–posterior and lateral views of radiographs of the hands, spine, knees, and hips were obtained from affected and clinically suspected cases. Radiographic hip OA was defined as a joint space ≥2.5 mm (Aspelund et al., 1996; Chaisson et al., 1997). Knee, hand, and spine OA were defined as the presence of Kellgren/Lawrence changes of grade ≥ 2. All radiographs were read by one radiologist (L.K. Chen) for grading the severity of OA. The final diagnosis of OA was based on history, physical examination, and/or radiographic findings.

The study was approved by the institutional review board, and informed consent was obtained from all of the participants.


Genomic DNA was extracted using the Puregene DNA Isolation Kit (Gentra Systems, Minneapolis, MN, USA) according to the manufacturer's instructions. Genotyping was performed for 400 highly polymorphic microsatellite markers using the ABI PRISM Linkage Mapping Set v 2.5 HD5 (Applied Biosystems, Foster City, CA, USA). The average heterozygosity of markers was 0.72% with an estimated spacing of 10 cM. Allele sizing was calculated using the GeneMapper software program (Applied Biosystems). Allele calling and binning were performed using the SAS program.

DNA Sequencing

COMP DNA sequence variants of all 19 exons and exon–intron junctions were determined by direct sequencing using an Applied Biosystems 3730 DNA analyzer. Primers for polymerase chain reaction (PCR) were designed using the Primer3 program (http://frodo.wi.mit.edu/primer3/). PCR reactions were performed in a final volume of 50 μl containing 0.2 μM each primer, 5 mM Tris-HCl (pH 8.3), 5 mM KCl, 1.5 mM MgCl2, 0.01 mM EDTA, 0.05 mM DTT, 200 μM dNTPs and 2.5 unit Taq™ (Violet, Taipei, Taiwan). Amplification conditions consisted of an initial denaturation of 3 min at 94 °C, followed by 20 cycles of touchdown PCR for 30 s at 94 °C, 30 s at 65 °C (decreased 0.5 °C per cycle), 40 s at 72 °C, and a final 20 cycles of 30 s at 94 °C, 30 s at 55 °C, and 40 s at 72 °C.

To ensure that any sequence variant identified was not a polymorphism, we included 96 normal subjects as controls. These 96 subjects were randomly selected from the Han-Chinese Cell and Genome Bank in Taiwan described previously (Pan et al., 2006), in which more than 3300 controls were collected and randomly selected through registry.

Statistical analyses

Mendelian inheritance of the 400 microsatellite markers of this pedigree was verified using the PedCheck program (http://watson.hgen.pitt.edu/register/docs/pedcheck.html). Two-point analyses using the MLINK program were carried out for the most significant results obtained in the region covering D19S884, D19S226, and D19S414.


Clinical assessment

A total of 70 individuals were recruited for the study, 25 of whom were diagnosed with OA based on history, physical examination, and/or radiographic findings (Fig. 1). Because of age-dependent variation in this early onset OA family, all subjects in generation VI who had no symptoms were assigned as unknown status.

The onset age and severity of OA varied among family members. The average onset age was 29.7 ± 16.1 years stratified with 37.3 ± 19.2 years in generation IV, 29.8 ± 13.7 years in V and 12.0 ± 7.2 years in VI (Table 1). The height and arm span of the affected male members were shorter than those of the unaffected males (155.9 ± 11.4 cm vs. 164.5 ± 16.0 cm for height, P= 0.010; 158.4 ± 12.5 cm vs. 165.3 ± 16.7cm for arm span, P= 0.027) (Table 2). Most patients recalled that one of the first signs of illness was difficult in performing squatting without a waddling gait. There was no evidence for mental retardation, central or peripheral nervous system disease, myopathy, muscle weakness, ocular disease, internal organ abnormality, spontaneous fracture, joint contracture, or hypermobility in the kindred. The proband (IV 1; age 45 years) had sclerosis that was present over C1 and C2 (KL grade 4); C1 to C7 showed formation of osteophytes (KL grade 3). The first radiographic signs in a male patient (V: 4) were cysts in the femoral head and narrow joint space at age 36 years (Fig. 2A). Decreased joint space and osteophyte formation were evident in both knee joints of a female patient in the study family (V 62) at age 40 years (Fig. 2B). Radiological assessments of the kindred detected no signs of delayed ossification or hypoplasia of the epiphyses of the long tubular bones. There were three clinically affected individuals in generation VI with onset ages of 10, 20, and 6 years, respectively (VI55, VI61, and VI25). Either knee or hip joints of radiological evaluations in VI 55 and VI 61 displayed no significant abnormalities at the onset ages (Fig 2C, 2D). Patient VI 25 presented with slow running and a strange running posture, and showed abnormalities in epiphyseal size and contours of both knees (Fig 2E, 2F) without delay of ossification in radiographic study of the hip and knee at 9 years of age. The radiographic study of three female patients (IV 58, V 16, and IV 15) with onset ages of 62, 24, and 25 years showed narrow joint space and osteophyte formation of both knee and both hip joints (Fig 3A–C). A 19-year-old asymptomatic female (VI 68) showed no abnormalities in both knee joints (Fig 3D); however, she was later found to have COMP mutation (see below). The other OA cases in this family showed classical degenerative changes of joint space upon radiological evaluation.

Table 1.  Onset age of OA, and numbers of affected, unknown and COMP mutation-carrying cases in the kindred
GenerationA/U/G casesOnset age (yrs)
  1. A: affected.

  2. U: unknown.

  3. G: COMP gene mutation.

 IV6/0/437.3 ± 19.2
 V16/0/1429.8 ± 13.7
 VI3/15/812.0 ± 7.2
All individuals25/15/2629.7 ± 16.1
Table 2.  Physical parameters in the kindred family
 AffectedUnaffectedP value
Male (n = 14)Female(n = 11)Male(n = 28)Female(n = 17)MaleFemale
  1. *p value > 0.05 significant level.

  2. BMI: Body Mass Index.

Body weight (kg) 62.6±14.1 60.4 ± 9.9 67.9 ± 14.1 59.3 ± 10.80.2470.608
Body height (cm)155.9±11.4154.3 ± 5.1164.5 ± 16.0155.9 ± 5.30.010*0.569
BMI (kg/m2) 25.4±4.2 25.3 ± 4.0 24.0 ± 4.0 24.4 ± 4.70.4270.331
Arm span (cm)158.4±12.5154.0 ± 7.5165.3 ± 16.7154.8 ± 6.10.027*0.738
Figure 2.

(A) Radiographs of both hip joints of a male patient (V 4). At age 36 years, the first radiographic signs were cysts in the femoral head and a narrow joint space. (B) Radiographs of both knee joints of a female patient in the study family (V 62). At age 40 years, decreased joint space and osteophyte formation were present. (C) Radiographs of both knee joints of a male patient (VI 55) showed no abnormalities at the age of 24 years. (D) Radiographs of both hip joints of a male patient (VI 61) showed no abnormalities at the age of 22 years. (E) and (F) Radiographs of both hip and knee joints of 9- year-old boy (VI 25) and abnormalities in epiphyseal size and contours of both knees were found.

Figure 3.

(A) Radiographs of both knee joints of a female patient (IV 58). Her onset age was 62 years old. The radiographic study showed narrow joint space and osteophyte formation of both knee joints. (B) Radiographs of both hip joints of a female patient in the study family (V 16). Her onset age was 24 years old. At age 36 years, decreased joint space and osteophyte formation were present. (C) Radiographs of both knee joints of a female patient (IV 15) showed markedly decreased joint space. Her onset age was 25 years old. (D) Radiographs of both knee joints of a female patient (VI 68) showed no abnormalities at the age of 20 years. The patients (IV 58, V 16, and IV 15) were presumably affected subjects without COMP mutation. The latter patient (VI 68) carried the COMP mutation but without symptoms and signs of OA

Genetic assessment

The clinical presentations in the familial early onset OA patients resembled those previously reported in patients with a COL2A1 gene mutation (Knowlton et al., 1990; Rivtaniemi et al., 1995). However, we excluded involvement of the COL2A1 gene by linkage analysis of short tandem-repeat polymorphism (STRP) markers and direct sequencing of the COL2A1 gene (Mu et al., 2009). A genome-wide scan with polymorphic microsatellite markers was then performed for 70 DNA samples (numbered in Fig. 1) from 25 individuals with OA manifestations, 15 of unknown disease status, and 30 unaffected family members. One region, covering D19S884, D19S226, and D19S414 on chromosome 19p, showed a highly significant linkage with the disease, with a maximum point LOD score of 2.51 at D19S226 (Fig. 4). A search for potential candidate genes in the flanking region showed that the COMP gene was the most likely candidate gene. The COMP gene was then directly sequenced in eight members of the index family (IV 1, 2; V 3–7; VI 33). A c.2152C>T substitution in exon 18 was identified after sequencing all 19 exons and exon–intron junctions of the COMP gene in all the affected members. This mutation, located in the C terminal globular domain of the gene product, resulted in the substitution of tryptophan for arginine at position 718. Further analysis of this mutation in all of the 70 enrolled subjects in this family revealed that a total of 26 individuals carried this mutation, of which 21 were affected with OA and 5 had no symptoms. All of the latter 5 individuals were in generation VI (VI 26, 40, 50, 51, and 68 and mean age 18.6 ± 11.3 years) and were assigned “unknown status” for OA. None of the 30 unaffected members had the c.2152C>T substitution. In addition, there were four clinical and radiological OA subjects (IV 58, V16 and V 41, onset ages, 25, 62, 24, and 39 years, respectively) without the COMP mutation. Individual VI 25 had the earliest age of onset (6 years old). To ensure there were no other gene mutations that may account for this early onset, we sequenced the COL19A, MATN3, and SLC26A2 genes and identified several SNPs, including two non-synonymous SNPs, one in COL9A1 (rs1135056, Gln621Arg) and one in SLC26A2 (rs30832, Ile574Thr); however, no mutations were found. No other sequence variations were found in sequencing of the COMP gene. The c.2152C>T substitution was not detected in 96 normal Taiwanese subjects.

Figure 4.

(A) Two-point analysis using the MLINK program. One region covering D19S884, D19S226, and D19S414 on chromosome 19p showed highly significant linkage with the disease, with a maximum point LOD score of 2.51 at D19S226. (B) After sequencing all 19 exons and exon–intron junctions of the COMP gene, we identified the c.2152C>T substitution in exon 18. The mutation resulted in the substitution of tryptophan for arginine at position 718 (R718W).


Early degeneration of cartilage in the joints is triggered by factors that are incompletely understood and results in premature OA. In the present report, we extended our studies on a previously identified family with early-onset joint disease. The clinical symptoms and radiographic images of the involved joints were similar to those observed in patients with idiopathic OA. Physical and laboratory findings ruled out most secondary causes of OA, including diabetes, and metabolic and endocrine disorders.

In total, 26 individuals in this family carried the c.2152C>T mutation in the COMP gene, 21 of which were affected and showed classical OA phenotypes. The other five individuals who carried the mutation but showed no symptoms (VI 26, VI 40, VI 50, VI 51, and VI 68) had an average age of 18 years (12, 38, 10, 14, and 19 years of age, respectively), and thus may have been too young for the appearance of symptoms. Of the 25 affected OA individuals, four did not have the mutation (generation IV 15, 58 and V 16, 41), but had common features and radiological studies of OA. The average age of onset was 30 years (25, 62, 24, and 39 years; Table 3). These four subjects are likely classical OA with multifactorial inheritance. Consistent with this notion is that these four individuals were taller than the affected patients with COMP mutation. Individuals IV 15, 58 and V 16 were female, with heights of 159, 159, and 162 cm, respectively (Table 3), taller than the affected female patients with COMP mutation (160 ± 1.7 vs. 152.1 ± 4.1 cm, P= 0.041). The other individual was a male of height 159 cm who was also taller than the average height of male patients with COMP mutation (155.6 ± 11.8 cm). Individual VI 25 who had the COMP mutation, appeared to have a more severe disease with onset by 6 years of age; however, sequencing of the COL19A, MATN3, and SLC26A2 genes revealed no mutations.

Table 3.  Summary of the main clinical findings in four clinically affected patients without COMP mutation and five individuals carried the mutation but without clinical symptoms
GroupsAge/onset age(yrs)Body height (cm)Body weight (kg)Symptoms*Radiography
  1. Clinical symptoms.

  2. *Radiography: Involved joints by radiography.

  3. A group: OA without COMP mutation.

  4. B group: Asymptomatic COMP mutation patients.

A group
 IV 1567/2515965Knee arthralgiaKnee
 IV 5865/6215972Knee arthralgiaKnee
 V 1644/2416258Hip/knee arthralgia; low back painHip/Knee/Spine
 V 4156/3915989Knee arthralgiaKnee
B group
 VI 5010/-14155.5NoNormal
 VI 2612/-15048.2NoNormal
 VI 4038/-15371NoNormal
 VI 5114/-15446.5NoNormal
 VI 6819/-15956.1NoNormal

The identified c.2152C>T substitution in exon 18 of the COMP gene located on chromosome 19p12 resulted in the nonconservative substitution of a hydrophobic tryptophan for the positively charged arginine at position 718. This mutation is located in the C terminal globular domain of the COMP gene. Although we did not perform functional analysis of the tryptophan substitution, the 718 position of arginine is conserved across the species, including bonobo, bovid, horse, mouse, and rat. Furthermore this substitution was not detected in any of 96 normal Taiwanese subjects. Therefore, the mutation identified in this study is not a polymorphism but is likely a disease causing variant.

The COMP gene contains 19 exons and encodes a 552 kDa pentameric adhesive glycoprotein expressed primarily in cartilage, ligaments, and tendons. A vast majority of previously identified COMP mutations (around 95%) are clustered within exons 8–14. The remaining 5% of mutations are in exons 16 and 18. Previous studies have shown that mutations in the COMP gene cause severe forms of dominant multiple epiphyseal dysplasia (MED) and the more severe disorder of pseudoachondroplasia (PSACH) (Briggs & Chapman, 2002). PSACH is one of the more common skeletal dysplasias and is characterized by disproportionate short-limbed dwarfism and ligamentous laxity. Growth retardation becomes apparent between the ages of one and three years. The hands and feet are short and broad. Platyspondylisis is evident on radiographs (Briggs & Chapman, 2002; Mabuchi et al., 2003). PSACH and MED mutations variably affect the cellular trafficking behavior of COMP and the extent of defective trafficking correlates with clinical phenotype (Chen et al., 2008b).

The second type of skeletal dysplasia caused by mutation of the COMP gene is MED which presents early in childhood and is typically characterized by bilaterally symmetrical short limb dwarfism and waddling gait, with coxa vara, genu valgum/varum, and abnormalities of the tibiotalar joint. The X-ray findings showed delayed ossification of the epiphyses of the long tubular bones (Briggs & Chapman, 2002; Jakkula et al., 2003; Mabuchi et al., 2003; Kennedy et al., 2005a; Kennedy et al., 2005b). The mutation (c.2152C>T, R718W) identified in this report has been described previously (Jakkula et al. 2003; Mabuchi et al., 2003; Kennedy et al., 2005a) These patents either presented as a severe Fairbank form of MED phenotype (Mabuchi et al., 2003) or a childhood-onset MED with myopathy with or without elevation of creatine kinase (Jakkula et al., 2003); all have radiological evidence of MED and delayed ossification of epiphyses in multiple joints. In Jakkula's study, the diagnosis of MED in affected children led to the identification of an affected father and grandmother who had not received a specific diagnosis but who were assumed to have the early-onset OA. According to the phenotypes of this large six-generation kindred, all were diagnosed as early-onset OA. The presence of the same mutation (c.2152C>T, R718W) and the radiographic findings in a 9-year-old boy (VI 25), suggest that the family might carry a known MED mutation and probably had MED. The congenital disorders of MED have a specific expression in childhood but only a nonspecific expression of early-onset OA in adults. The diagnosis of mild form MED is underevaluated and is still not widely appreciated among orthopedists and rheumatologists. We may make the diagnosis of mild form MED depending on genetic analysis and radiologic findings in children, rather than based on obscure phenotypes. In summary, our results suggest that COMP is the disease susceptibility gene and the c.2152C>T mutation in exon 18 can cause mild form MED mimicking as early-onset OA phenotype.


We would like to thank the participating patients and their families. We also thank Dr. Chang Tsung-Hsun in Department of Orthopedics of Lin Kou Chang Gung Medical Center for professional consultation. No conflict of interest has been declared by the authors. This work was supported by the National Science and Technology Program for Genomic Medicine from the National Science Council, Taiwan (NGC:NSC98–3112-B-001–022 and NCC:NSC98–3112-B-001–021), and the Academia Sinica Genomic Medicine Multicenter Study Program (40–05).