The genetic epidemiology of osteoarthritis - Semantic Scholar

The genetic epidemiology of osteoarthritis Ana M. Valdes and Tim D. Spector Department of Twin Research and Genetic Epidemiology, St Thomas’ Hospital Campus, Kings College London School of Medicine, London, UK Correspondence to Dr Ana M. Valdes, PhD, Department of Twin Research and Genetic Epidemiology, St Thomas Hospital, London SE1 7EH, UK Tel: +44 207 188 6765; fax: +44 207 188 6718; e-mail: [email protected] Current Opinion in Rheumatology 2010, 22:139–143

Purpose of review Osteoarthritis is the most common form of arthritis in the elderly and is influenced by both genetic and environmental risk factors. The scope of the present article is to offer an overview of recent developments in the genetic epidemiology of knee and hip osteoarthritis, with particular emphasis on published genomewide association studies (GWAS). Recent findings Candidate gene studies and genomewide linkage studies have identified genes in the bone morphogenetic pathway (e.g. GDF5), the thyroid regulation pathway (DIO2) and apoptotic pathways as involved in genetic risk of large joint osteoarthritis. GWAS have reported structural genes (COL6A4), inflammation-related genes (PTGS2/PLA2G4A) and a locus on chr 7q22 (GPR22 and four other genes in the same linkage disequilibrium block) associated with osteoarthritis. Summary Genetic studies have identified polymorphisms associated with osteoarthritis and related end-points. These include genes in signaling cascades involved in joint and bone biology, as well as genes in inflammatory pathways and a cluster of five genes in perfect linkage disequilibrium in the 7q22 region. Keywords genomewide association study, meta-analysis, osteoarthritis Curr Opin Rheumatol 22:139–143 ß 2010 Wolters Kluwer Health | Lippincott Williams & Wilkins 1040-8711

Introduction Genetic epidemiology is the study of genetic factors that determine the distributions and dynamics of diseases in populations using a variety of analysis methods [1]. With the advent of high-throughput genotyping it has become possible to genotype over one million single nucleotide polymorphisms (SNPs) per individual and to genotype 10s of 1000s of individuals in a single study. This, and the fact that current SNP maps have over 85% coverage of the genome in whites [2], means that genomewide association approaches have the potential to uncover the genetic contribution to complex human diseases such as osteoarthritis. The etiologic insights from such studies should also lead to improved diagnostic and prognostic tools and inform the development of more specific therapies for large joint osteoarthritis. The scope of the present article is to offer an overview of recent developments in the genetic epidemiology of knee and hip osteoarthritis, with particular emphasis to published genomewide association studies (GWAS). Osteoarthritis can affect different joints, including the hand and the facet joints in the spine. The focus of the current review is on knee and hip osteoarthritis that results in over 500 000 total joint replacement surgeries in the United States alone and is the cause of a huge personal and economic burden [3]. 1040-8711 ß 2010 Wolters Kluwer Health | Lippincott Williams & Wilkins

Osteoarthritis is a multifactorial disease with genetic and environmental determinants. All cases are probably affected by both genetics and environment, with a continuous distribution between the extremes of predominantly genetic or predominantly environmental [4]. For example, the risk of posttraumatic osteoarthritis after a meniscal injury of the knee is strongly affected by a family history of osteoarthritis, by the presence of nodal osteoarthritis of the hand, by obesity and by sex [5].

Genetic contribution to osteoarthritis Evidence for a genetic predisposition to osteoarthritis was reported as early as the 1940s [6]. Familial aggregation is the sine qua non for continuing the genetic analysis of a disorder and several studies have reported the risk ratio for a relative of an affected individual compared with the population prevalence ranging between 2 and 8 (see [7] for review). In addition, several classical twin studies have been carried out on osteoarthritis. Comparing the resemblance of identical twins for a trait or disease with the resemblance of nonidentical twins offers the first estimate of the extent to which genetic variation determines variation of that trait or ‘heritability’. The heritability of DOI:10.1097/BOR.0b013e3283367a6e

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

140 Epidemiology and health-related services

osteoarthritis has been calculated in twin sets after adjustment of the data for other known risk factors such as age, sex and BMI. Such findings show that the influence of genetic factors in radiographic osteoarthritis of the hip is 60% and 39% for knee osteoarthritis in women independent of known environmental or demographic confounding factors. Classical twin studies and familial aggregation studies have also investigated the genetic contribution to cartilage volume and progression of disease [8,9]. Below we discuss some of the genes deriving from linkage and association studies. Some of the genes discussed below have been found to be associated with osteoarthritis in at least two independent populations with statistical significance after adjusting for multiple testing; however at the time of this writing no studies have reported an association in whites with any single osteoarthritis phenotype achieving P less than 5 108 (which is the threshold considered for genomewide significance) and the only gene to achieve this level of significance in Asians (DVWA) is not associated with osteoarthritis in whites.

Linkage analysis Genetic linkage occurs when a locus involved in the trait of interest and alleles at nearby markers are inherited jointly. Familial co-segregation of a trait of interest with one or several genetic markers forms thus the basis of linkage analysis, which seeks to place a disease locus on the genome [1]. Several genomewide linkage scans have been published to date based on small families or twins of affected relatives collected in the UK, Finland, Iceland, the United States and the Netherlands (see [10] for references). In the case of large-joint osteoarthritis in the past couple of years the most significant development along these lines has been the identification of the region in chromosome 14q32.11 by Meulenbelt et al. [11] by carrying linkage analysis on 179 affected siblings and four trios with generalized osteoarthritis. Association tests in other cohort parts of the same study found a common deiodinase, iodothyronine, type II gene (DIO2) haplotype, exclusively containing the minor allele of rs225014 and common allele of rs12885300. This haplotype had an odds ratio (OR) 1.79 (P < 2.02 105) with hip osteoarthritis in women. The DIO2 gene encodes an intracellular enzyme in the thyroid pathway, responsible for the local bioavailability of thyroid hormone in specific tissues, including the growth plate. It converts inactive thyroid prohormone (T4) to active T3 hormone, which in turn binds to thyroid receptors in the nucleus and activates thyroid hormone-responsive genes. The role of thyroid hormone regulation in genetic risk of osteoarthritis may be an attractive therapeutic target, as drugs to intervene in this pathway have been in use for other therapeutic indications for a long time and are well

characterized in terms of their safety and pharmacokinetics [12].

Candidate gene studies Once the approximate location of a disease gene has been found, association studies can be conducted in that region alone to pinpoint the gene and allelic variants involved. Genetic association analysis of candidate gene regions without any preceding linkage analysis has a long history of discovering single marker allele associations. A number of candidate gene studies for knee and hip osteoarthritis have been carried out in the past with varying levels of success and certainty, some of which are discussed below. Cartilage degrading enzymes

The primary cause of cartilage degradation is elevated proteolytic activity that degrades aggrecan and type II collagen. Loss of aggrecan is an early and crucial event in the progression of cartilage destruction and is attributed to proteolytic cleavage of the aggrecan core protein. For many years, it was believed that matrix metalloproteases (MMP-3) was an important proteinase responsible for this breakdown of aggrecan. MMP-3, as well as many other MMPs cleave aggrecan at the Asn341/Phe342 bond, generating specific aggrecan fragments. These MMPgenerated aggrecan fragments, however, are not detected in in-vitro models of cartilage degradation or in synovial fluid of osteoarthritis patients [13]. Rather, mouse models of osteoarthritis have clearly demonstrated that the enzymatic activity of ADAMTS-5 mediates aggrecan-induced turnover of articular cartilage after physical or cytokinemediated injury [14]. Nevertheless, a large genetic association analysis of ADAMTS-5 SNPs with hip, hand and knee osteoarthritis failed to find convincing evidence as to role that genetic variation at this gene may have for selecting high-risk versus low-risk patients [15]. Studies on the role that genetic variation at other aggrecanases and metalloproteinases has on the risk of osteoarthritis have also failed to demonstrate any significant genetic associations [16]. Apoptosis and mitochondrial DNA

The important role of apoptosis in osteoarthritis has been demonstrated in both in-vitro and in-vivo models. A significantly higher proportion of apoptotic chondrocytes is found in samples from osteoarthritis patients than in those from age matched controls (see [17] for review). Two main pathways of programmed cell death have now been elucidated: a death-receptor-mediated pathway and a mitochondrial-mediated pathway. The death receptors are a subset of the tumor necrosis factor (TNF)-receptor family of cell surface molecules such as Fas. The mechanisms through which stimulation of Fas by the Fas


The genetic epidemiology of osteoarthritis Valdes and Spector 141

ligand initiates apoptosis have been extensively investigated, and it has been shown that cell death in primary osteoarthritis chondrocytes is mediated by the Fas pathway [18]. Mitochondrial DNA (mtDNA) variation has been associated with risk of osteoarthritis of the knee [19] and of the hip [20]. Given the role of mtDNA damage on mitochondria-driven apoptosis, it is possible to speculate that the involvement of mtDNA haplotypes on osteoarthritis risk may relate to apoptotic pathways. More recently we reported that genetic variation at the ANP32A gene, which encodes a tumor suppressor molecule that plays a regulatory role in apoptosis and interferes with canonical Wnt signaling in vitro is associated with hip osteoarthritis in women [21] (rs7164503 OR ¼ 0.67, P < 3.8 104 after meta-analysis of four independent cohorts). Such results suggest that additional research into the role that genetic variation in apoptotic signaling molecules plays in susceptibility to osteoarthritis might help further our understanding of the molecular pathways involved in osteoarthritis.

Meta-analyses Individual studies may be hampered by sample size limitations that result in lack of statistical power [22] and false positive results due to publication bias and chance. Meta-analyses based on consortium efforts may help overcome some of these limitations. A recent effort along those lines was recently published reporting the results of a collaborative meta-analysis of individuallevel data from 14 research groups on GDF5 and FRZB polymorphisms and osteoarthritis phenotypes. The authors found consistent evidence for an involvement of the promoter polymorphism at the GDF5 gene and knee osteoarthritis weaker evidence for hip osteoarthritis having tested more than 5000 cases of each hip and knee and approximately 8000 controls [23]. Analyses for FRZB polymorphisms and haplotypes did not reveal any statistically significant signals except for a borderline association of rs288326 with hip osteoarthritis (P ¼ 0.019; 4346 hip osteoarthritis cases and 8034 controls) [23]. Other smaller meta-analyses of various genes that were identified in Asian samples and were not reproducible in European-descent samples have been published [24].

Genomewide association studies With the recent advances of high-throughput SNP genotyping technology, GWAS have become possible. These studies take advantage of linkage disequilibrium, that is, the fact that in any given choromosomal region in the genome alleles at physically nearby loci are associated in the population. The markers analyzed need not be functional, but may simply be in linkage disequilibrium with

the functional variant [1]. However, the large-scale nature of these studies introduces a multiple testing issue, making the need for replication of any positive associations an absolute requirement. GWAS are nevertheless a powerful approach for unlocking the genetic basis of complex diseases such as osteoarthritis. The method has been used successfully in several common diseases [25]. Notable advantages include its comprehensiveness and the potential for finding susceptibility genes at previously unknown location or with no known relationship to the disease. The first wave of GWAS in osteoarthritis has now been published. Although these studies clearly show that there is no definitive and common highly penetrant allele that causes osteoarthritis, some interesting genes have emerged from these studies. The findings help to better delineate the types of genes and genetic variants that are involved in osteoarthritis and provide substantial material for future research. The first large-scale (500 000 markers) GWAS to be published was a pooled scan [26] on knee osteoarthritis. The variant identified by this study did not achieve genomewide significance but was replicated in three independent cohorts that fell in the 50 region of the gene encoding the cyclooxygenase 2 (COX-2) and the cytosolic phospholipase enzymes, both involved in prostaglandin synthesis [rs4140564 OR ¼ 1.55, 95% confidence interval (CI) 1.30–1.85, P < 6.9 107]. The data did not allow us to determine whether the association observed was due to the gene encoding COX-2 or cytosolic phospholipase-a (cPLA2a) or both. A second GWAS on knee osteoarthritis, but testing fewer markers was published by a Japanese group [27]. After screening only 94 knee osteoarthritis cases and 658 controls for fewer than 80 000 SNPs, a subset of 2153 SNPs were tested in a second set of 646 knee osteoarthritis cases and 631 controls. Two variants in what was up to then considered to be a hypothetical gene were identified as the most strongly associated. The locus was named double von Willebrand factor A (DWVA), and two coding variants, rs11718863 and rs7639618, were confirmed to be strongly associated with risk of knee osteoarthritis in additional Chinese and Japanese samples (OR ¼ 1.43, P < 7 1011). The two SNPs were also found to influence the binding of the DVWA protein to b-tubulin, and the authors hypothesized that tubulins and microtubules might be protective factors in the pathogenesis of osteoarthritis [27]. Subsequently, however, the two SNPs were found to have no effect on knee or hip osteoarthritis in whites samples. In a first study, Valdes et al. [28] found a slight trend for an association in the opposite direction in 999 UK knee osteoarthritis cases and 1166 controls and no association with hip osteoarthritis. A second study of


142 Epidemiology and health-related services

European samples also failed to find any associations between the variants in the DVWA locus and osteoarthritis in European patients [29], suggesting that the tubulin binding effect apparently by this molecule is not involved in whites’ osteoarthritis. The newly identified DVWA was shown to actually represent the human gene coding for the collagen VI a4 chain (COL6A4) [30]. These results highlight the large heterogeneity between Asian and white samples with regards to genetic risk of osteoarthritis. More recently, Kerkhof et al. [31] performed a GWAS testing more than 500 000 SNPs in 1341 osteoarthritis cases and 3496 white controls from the Netherlands. SNPs associated with at least two osteoarthritis phenotypes were subsequently analyzed in a further 14 938 osteoarthritis cases and 39 000 controls. After replication and metaanalysis the minor allele of rs3815148 on chromosome 7q22 (upstream of the GPR22 gene) was associated with a 1.14-fold increased risk (95% CI 1.09–1.19) for knee and/or hand osteoarthritis (P ¼ 8 108), and also with increased risk for knee osteoarthritis progression. The association with knee osteoarthritis alone was 1.16 (1.09–1.24) P less than 2 106 and with hand osteoarthritis alone [OR ¼ 1.14 (1.06–1.23), P < 3 104], no association was seen with hip osteoarthritis. This SNP mapped to a large haplotype block of over 500 kilobases that included at least five genes. The authors [31] identified another SNP strongly associated with GPR22 expression levels in lymphocytes in almost complete linkage disequilibrium with rs3815148. GPR22 encodes a G-protein coupled receptor with an unknown ligand. Immunohistochemistry experiments showed absence of GPR22 in normal mouse articular cartilage or synovium but GPR22 positive chondrocytes were found in the upper layers of the articular cartilage of mouse knee joints that were challenged by in-vivo papain treatment or in the presence of interleukin-1b suggesting that GPR22 may be the gene involved. However, other genes in the haplotype block such as COG5, DUSL4, BCAP29 and HBP1 could not be excluded as being involved in disease risk. No overlap exists between genes and regions suggested by linkage scans and those found by GWAS. On the contrary, GDF5, which is a strong biological candidate uncovered by a relatively small association scan, is found as one of the top 12 regions associated with knee osteoarthritis in the Dutch study [31].

Conclusion In spite of the fact that osteoarthritis has a very strong genetic contribution, as of today no genomewide significant associations have been identified in European samples and the only gene to show that level of association in Asians (DVWA) is not associated in whites. Never-

theless, several genes have been associated in two or more independent cohorts after adjusting for multiple testing (PTGS2, ANP32A, GDF5, DIO2 and a gene cluster in the 7q22 region). The pathways of these genes [apopotosis, bone morphogenetic protein (BMP) signaling, inflammation, thyroid signaling] are helping to increase our understanding of the molecular pathogenesis of osteoarthritis. Radiographic hip osteoarthritis has a strong familial aggregation (ls ¼ 4.27–5.07) and a strong heritability (60%) whereas radiographic knee osteoarthritis has weaker familial aggregation (ls ¼ 1.66–2.13) and a weaker heritability (39%; see [7] for references). Nevertheless, the strongest osteoarthritis genetic associations reported to date (GDF5, the 7q22 cluster, DVWA) are all with knee osteoarthritis and the reported associations with hip osteoarthritis [11,24] are much more modest. This suggests that there may be a considerable amount of heterogeneity in the hip osteoarthritis phenotype that is not captured by simple definitions such as total hip replacement or radiographic grade. During the coming years, as additional genetic and functional studies offer a more complete picture of the genetic contribution to osteoarthritis and the underlying molecular pathways, additional targets for novel therapies and improved diagnostic and prognostic tests should become available. Identifying individuals at greatest genetic and molecular risk for progression will have important implications for clinical trial design and should substantially improve our ability to diagnose and treat this disease that represents a major healthcare burden in industrialized societies.

Acknowledgement The present study was supported by EC framework 7 programme grant 200800 TREAT-OA.

References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest of outstanding interest Additional references related to this topic can also be found in the Current World Literature section in this issue (pp. 230–231). 1

Stein CM, Elston RC. Finding genes underlying human disease. Clin Genet 2009; 75:101–106.

2

The International HapMap Consortium. A second generation human haplotype map of over 3.1 million SNPs. Nature 2007; 449:851–861.

3

Losina E, Walensky RP, Kessler CL, et al. Cost-effectiveness of total knee arthroplasty in the United States: patient risk and hospital volume. Arch Intern Med 2009; 169:1113–1121.

4

Dieppe PA, Lohmander LS. Pathogenesis and management of pain in osteoarthritis. Lancet 2005; 365:965–973.

5

Englund M, Lohmander LS. Risk factors for symptomatic knee osteoarthritis fifteen to twenty-two years after meniscectomy. Arthritis Rheum 2004; 50:2811–2819.

6

Stecher RM. Heberden’s nodes: heredity in hypertrophic arthritis of the finger joints. Am J Med Sci 1941; 210:801–809.

7

Valdes AM, Spector TD. The contribution of genes to osteoarthritis. Med Clin North Am 2009; 93:45–66.

8

Zhai G, Ding C, Stankovich J, et al. The genetic contribution to longitudinal changes in knee structure and muscle strength: a sibpair study. Arthritis Rheum 2005; 52:2830–2834.


The genetic epidemiology of osteoarthritis Valdes and Spector 143 9

Zhai G, Hart DJ, Kato BS, et al. Genetic influence on the progression of radiographic knee osteoarthritis: a longitudinal twin study. Osteoarthritis Cartilage 2007; 15:222–225.

10 Lee YH, Rho YH, Choi SJ, et al. Osteoarthritis susceptibility loci defined by genome scan meta-analysis. Rheumatol Int 2006; 26:996–1000. 11 Meulenbelt I, Min JL, Bos S, et al. Identification of DIO2 as a new susceptibility locus for symptomatic osteoarthritis. Hum Mol Genet 2008; 17:1867–1875. Genetic variation in the DIO2 gene, an enzyme in the thyroid pathway amenable to therapeutic intervention, was found to be associated with risk of osteoarthritis. 12 Eisenberg M, Distefano JJ. TSH-based protocol, tablet instability, and absorption effects on L-T4 bioequivalence. Thyroid 2009; 19:103–110. 13 Tortorella MD, Malfait AM. The usual suspects: verdict not guilty? Arthritis Rheum 2003; 48:3304–3307. 14 Stanton H, Rogerson FM, East CJ, et al. ADAMTS5 is the major aggrecanase in mouse cartilage in vivo and in vitro. Nature 2005; 434:648–652. 15 Rodriguez-Lopez J, Mustafa Z, Pombo-Suarez M, et al. Genetic variation including nonsynonymous polymorphisms of a major aggrecanase, ADAMTS5, in susceptibility to osteoarthritis. Arthritis Rheum 2008; 58:435–441. Although animal models have shown the importance of ADAMTS-5 in the pathogenesis of osteoarthritis, this well powered association study could not detect an association between variation at the ADAMTS5 gene and osteoarthritis of the hand, the hip or the knee. 16 Rodriguez-Lopez J, Pombo-Suarez M, Loughlin J, et al. Association of a nsSNP in ADAMTS14 to some osteoarthritis phenotypes. Osteoarthritis Cartilage 2009; 17:321–327. 17 Kim HA, Blanco FJ. Cell death and apoptosis in osteoarthritic cartilage. Curr Drug Targets 2007; 8:333–345. 18 Wei L, Sun XJ, Wang Z, Chen Q. CD95-induced osteoarthritic chondrocyte apoptosis and necrosis: dependency on p38 mitogen-activated protein kinase. Arthritis Res Ther 2006; 8:R37. 19 Rego-Pe´rez I, Fernańdez-Moreno M, Fernańdez-Lo´pez C, et al. Mitochondrial DNA haplogroups: role in the prevalence and severity of knee osteoarthritis. Arthritis Rheum 2008; 58:2387–2396. 20 Rego I, Fernańdez-Moreno M, Fernańdez-Lo´pez C, et al. The role of European mtDNA haplogroups in the prevalence of hip osteoarthritis in Galicia (northern Spain). Ann Rheum Dis 2010; 69:210–213. Variation in mtDNA is implicated in risk of large joint osteoarthritis. 21 Valdes AM, Lories RJ, van Meurs JB, et al. Variation at the ANP32A gene is associated with risk of hip osteoarthritis in women. Arthritis Rheum 2009; 60:2046–2054.

22 Ioannidis JP, Trikalinos TA, Khoury MJ. Implications of small effect sizes of individual genetic variants on the design and interpretation of genetic association studies of complex diseases. Am J Epidemiol 2006; 164:609– 614. 23 Evangelou E, Chapman K, Meulenbelt I, et al. Large-scale analysis of asso ciation between GDF5 (rs143383) and FRZB (rs7775 and rs288326) variants and hip, knee and hand osteoarthritis. Arthritis Rheum 2009; 60:1710–1721. The largest meta-analysis of osteoarthritis candidate genes to date. 24 Dieguez-Gonzalez R, Calaza M, Shi D, et al. Testing the druggable endothelial differentiation gene 2 knee osteoarthritis genetic factor for replication in a wide range of sample collections. Ann Rheum Dis 2009; 68:1017– 1021. One of several examples in which osteoarthritis susceptibility loci discovered in Asians are not found to be associated in whites. 25 Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 2007; 447:661–678. 26 Valdes AM, Loughlin J, van Meurs JB, et al. Genome-wide association scan identifies a prostaglandin-endoperoxide synthase 2 variant involved in risk of knee osteoarthritis. Am J Hum Genet 2008; 82:1231–1240. 27 Miyamoto Y, Shi D, Nakajima M, et al. Common variants in DVWA on chromosome 3p24.3 are associated with susceptibility to knee osteoarthritis. Nat Genet 2008; 40:994–998. The first genomewide association scan in Asian samples identified the DVWA locus as strongly associated with risk of knee osteoarthritis. 28 Valdes AM, Spector TD, Doherty S, et al. Association of the DVWA and GDF5 polymorphisms with osteoarthritis in UK populations. Ann Rheum Dis 2009; 68:1916–1920. 29 Meulenbelt I, Chapman K, Dieguez-Gonzalez R, et al. Large replication study and meta-analyses of DVWA as an osteoarthritis susceptibility locus in European and Asian populations. Hum Mol Genet 2009; 18:1518–1523. 30 Wagener R, Gara SK, Kobbe B, et al. The knee osteoarthritis susceptibility locus DVWA on chromosome 3p24.3 is the 50 part of the split COL6A4 gene. Matrix Biol 2009; 28:307–310. 31 Kerkhof JM, Lories RJ, Meulenbelt I, et al. A genome-wide association study identifies a locus on chromosome 7q22 to influence susceptibility for osteoarthritis. Arthritis Rheum 2010; 62: (in press). This is the first large-scale (>100,000 markers) nonpooled GWAS for osteoarthritis, including data for knee, hand and hip osteoarthritis and extensive replication from a large number of cohorts.
