Position statement: the epidemiology, pathogenesis and risk factors of osteoarthritis of the knee
Osteoarthritis (OA) represents a large burden on healthcare resources worldwide with continually increasing prevalence. This has led to renewed interest in the causes and pathogenesis of the condition. In recent years, there has been a move away from a simple ‘wear and tear’ model of cartilage to one of a complex inflammatory process involving cellular and extracellular derangements that allow a catabolic state to dominate. Ultimately, OA is now seen as pan-joint disease involving synovium, menisci, ligaments and muscle, in addition to cartilage. There are several classification systems including radiographic, MRI-based, clinical and combined classification systems. As radiographs only detect OA in latter stages, there has been a focus on early diagnosis using MRI and serum biomarkers. New physiological MRI sequences can now measure the proteoglycan content in cartilage and new semiquantitative analyses have been developed to score total knee joint involvement in the disease process. Serum biomarkers can be divided into those that are collagen breakdown products and those that are inflammatory cytokines; these can be used in early detection of OA before radiographic appearances arise. The risk factors for OA include ageing, knee injury, obesity, altered limb alignment, impaired muscle strength, female gender, heavy physical work and genetic susceptibility. Research continues to identify the mechanisms involved that lead to OA development, with possibly unique processes underpinning each risk factor. Our understanding of the pathophysiology of OA will continue to improve in the next few years which should lead to new intervention strategies that target different processes. More informative MRI sequences will continue to be developed and the optimum combination of biomarkers to detect early OA will need to be identified. Genetic studies will continue to identify new susceptibility loci that could be targeted in future therapies.
The Osteoarthritis Research Society International (OARSI) defines osteoarthritis (OA) as ‘a disorder involving movable joints characterized by cell stress and extracellular matrix degradation initiated by micro- and macro-injury that activates maladaptive repair responses including pro-inflammatory pathways of innate immunity’.1 This in turn manifests initially as abnormal joint tissue metabolism, and subsequently by anatomic and physiological derangements. Clinically, this can manifest as cartilage degradation, bone remodelling, osteophyte formation, with patients presenting with joint inflammation, pain and loss of normal joint function (box 1).
Clinical and radiographic diagnostic criteria for osteoarthritis according to the American College of Rheumatology*2
Knee pain for most days of previous month
Osteophytes at joint margins on radiographs
Synovial fluid typical of osteoarthritis (laboratory)
Crepitus on active joint motion
Morning stiffness≤30 min duration
*Knee osteoarthritis (clinical and radiographic) if 1, 2 or 1, 3, 5, 6 or 1, 4, 5, 6 are present.
Prevalence and societal impact
OA is one of the most common causes of disability in adults. The prevalence increases with age, with a remarkable 13.9% of the population over 25 years old affected, and 33.6% of the population over 65 years old affected.3 It has a significant impact on society, compromising quality of life and productivity, and weighs heavily on national healthcare resources.4 ,5
Costs of OA care have risen over recent decades accounting for up to 1–2.5% of the gross national product for countries such as the USA, Canada, the UK, France and Australia.6
The WHO has developed a tool called the disability-adjusted life year to help gauge the burden of musculoskeletal disease on the world’s population.7Figure 1 represents international data collected from the WHO regarding the specific burden of OA worldwide. There is a large and growing burden in developed and developing nations, which will only increase with growing elderly populations.8 ,9 Most of the prevalence statistics originate from Western nations, but the prevalence of symptomatic knee OA in the West appears to be comparable to rural and urban areas within Asia.10
The Framingham Osteoarthritis Study12 reported prevalence rates of knee OA in a population in Massachusetts, USA, and a similar population-based study using identical methodology and definitions was performed in the Beijing Osteoarthritis Study.13 A similar radiographic and symptomatic prevalence of OA was found among Chinese men and their age-matched American counterparts (prevalence ratio 0.9 and 1.02, respectively) but elderly women in Beijing had higher radiographic and symptomatic prevalence rates compared with their age-matched American counterparts (prevalence ratio of 1.45 and 1.43, respectively); possible reasons given were genetic differences and higher physical activity in the Chinese population.13 Of those with knee OA, lateral compartment OA was much more common in the Chinese population (28.5% in Beijing females vs 11% in Framingham females; 32.3% Chinese males vs 8.8% Framingham males) which has been postulated to arise from more valgus distal femoral alignment.14 ,15 In sharp contrast, hip OA was 80–90% less prevalent in the Beijing population compared with the Framingham cohort, which may be due to morphological advantages in terms of better hip sphericity and less femora-acetabular impingement.16 ,17
Data from the US’ National Health and Nutrition Examination Survey (NHANES) as well as a population-based study in Johnston County, North Carolina, showed that knee OA is up to twice as common among African-Americans than Caucasians.18 ,19 The pattern of OA is also different with African-Americans tending to have more tri-compartmental disease and more severe radiographic changes in the tibiofemoral joint, and lateral joint space narrowing.20 Suggested reasons include differences in genetics, obesity, bone mineral density, occupational physical demands, diet and other lifestyle factors.20
The prevalence of risk factors for OA also differs between developed and developing nations. In developing nations, for example, obesity is often less prevalent (though now increasing), while higher proportions of the population have occupations requiring heavy physical labour, squatting, kneeling and climbing; the latter may in part explain why knee OA is more common in rural communities of China than in urban ones.10 Many Asian countries also currently have rapidly ageing populations which will increase the OA burden in the near future.7
Historic approach to OA of the knee
OA has historically been seen as an age-related degeneration of articular cartilage, in contrast to the inflammatory destruction of rheumatoid arthritis, and with little regard for other predisposing factors.21 The result being detected at a late stage with reduced joint space on traditional plain radiographs and ultimately joint replacement seen as the only and final solution.
There has been a shift from the belief that it is a uniform process mainly affecting articular cartilage to a heterogeneous condition with a plethora of predisposing risk factors working through different pathogenic mechanisms, including inflammation. All tissues within the joint become affected but the loss of articular cartilage and subchondral changes remain the most striking features.22 ,23 There is now a focus on early detection with MRI and biochemical markers, thereby allowing earlier, and possibly even preventative, intervention.24 ,25
Reviews and current concept articles
There has been renewed interest in the biological processes underpinning OA, which alongside better imaging and biochemical analysis has furthered our knowledge greatly. The role of inflammation, synovium, genetics and a variety of biochemical pathways open up the possibilities for more personalised therapy
Current state of the art
Classification of OA
Plain film radiographs and symptomology have historically been the mainstay of classifying OA. Recently, MRI assessment of the joint has entered clinical practice, while the use of serum and urinary biochemical markers is a rapidly developing research area for early detection. A common factor in radiographic and MRI assessments is the changes that occur in the subchondral bone.43 ,44 An emerging concept is that imaging in OA should not only assess cartilage and bone but also the menisci, synovium, fat and muscle, appreciating OA as a ‘whole-organ’ disease
Methods to classify osteoarthritis
1. Radiographic imaging45
2. MRI; including whole-organ scoring46
3. Clinical symptoms, including stiffness, swelling, knee range of motion and knee crepitus2
4. Combination of symptoms and imaging2
Classification strategies for radiographic imaging focus on the secondary changes of joint space narrowing, subchondral sclerosis and osteophyte formation
Osteoarthritis Research Society International Grading System (OARSI) for medial and lateral tibiofemoral joint space narrowing47
Although plain radiographic imaging classification is the most commonly quoted, and has endured for many years, the most limiting aspect of this classification is that it often does not detect joint degeneration until a more advanced stage.
A more complete picture of intra-articular disease is revealed by MRI. Advances in MRI technology have seen better resolution and new sequences developed that improve its ability to detect bone marrow lesions, cartilage abnormalities, ligamentous and meniscal damage, joint fluid changes and osteophyte formation, as well as macromolecular changes, which often precede morphological changes.29
Two broad types of MRIs can now be performed: ‘morphological’ which is the conventional technique visualising intra-articular structures, and ‘physiological’ which comprise state-of-the-art MRI sequences that are able to detect and measure proteoglycan content within cartilage, the concentration of which is altered early in the disease process . Physiological sequences currently remain research tools
Advanced MRI sequences for imaging knee OA
In MRI, the most common features that indicate OA are cartilage thinning and subchondral bone oedema48; however, as knee OA is increasingly being seen as a pan-joint disease, this had led to semiquantitative MRI-based scoring systems that grade the involvement of several other intra-articular structures in addition . These use conventional MRI sequences with a knee coil, but are time consuming and largely used in research.
MRI can not only be used to grade the severity of OA but may be able to help predict its onset. Non-contrast-enhanced T2-weighted MRI has been used to detect synovitis in non-OA patients who then go on to develop the disease indicating it may be a useful predictor in the very early stages.25 ,51This supports the notion that synovial inflammation has a pivotal role to play in the pathogenesis of OA, and highlights the importance of MRI assessment of extracartilaginous structures.
The final stages of OA can take decades to develop, but there are changes in the biochemical environment within the affected joint and subsequently blood that can be used to detect pathology early on in the disease process. In addition to advanced imaging, serum and urinary markers have been used to detect OA in its early stages, looking for bone, cartilage and/or synovial degradation products.52 ,53
Research has focused on two main groups of biomarkers, with studies in animal models and humans. The first group includes the degradation products of bone and cartilage such as C-terminal telopeptide of type II collagen, cartilage oligomeric matrix protein, collagen type II-specific neoepitope, an aggrecan neoepitope, a number of matrix metalloproteinases and procollagen type I amino-terminal properties.54–59The second group includes proinflammatory and anti-inflammatory cytokines that can be further divided according to those that aid diagnosis such as interleukin (IL)-6 and IL-15,24 ,60–62 those that reflect the burden of disease such as IL-1β, tumour necrosis factor (TNF)-α and vascular endothelial growth factor,63–67 and those that may be prognostic such as IL-6 and IL-10.68 ,69 There is yet to be any large-scale study to determine their validity but it is likely that a combination of markers from both groups will be needed to improve the sensitivity and specificity of such a tool in OA (figure 3).
Pathogenesis of OA
There is a clear association between OA and the ageing process12 ,70 ,71and also mechanical injury.72 ,73 Initially, there is a loss of glycosaminoglycans from cartilage which lowers the osmotic pressure, leaving it softer and with diminished resistance to compressive forces.74 A repair response then leads to increased production of proteoglycans and collagen type II, and chondrocyte proliferation and clustering.75
However, increased expression of inflammatory cytokines and proteases means catabolic activity dominates, and cartilage is degraded; initially fibrillation occurs in the superficial zone, followed by deep fissures and full chondral loss. Implicated cytokines include IL-6, IL-1β, TNF-α, IL-10, IL-13 and IL-4.76–81 They act through proinflammatory and anti-inflammatory pathways, and are also involved in angiogenesis and chemotaxis.82–84
A recent laboratory study showed OA-affected chondrocytes exhibit higher levels of the zinc importer ZIP8 leading to an increased level of intracellular zinc that triggers a catabolic cascade through increased expression of matrix metalloproteinases (MMP3, MMP9, MMP12, MMP13).85 Interestingly, in a mouse model, this also led to subchondral sclerosis in addition to cartilage damage, but no synovitis or osteophytes.85
State-of-the-art microarray analysis of OA chondrocytes suggests that there is faster messenger RNA (mRNA) decay with short-lived transcripts of genes involved in the regulation of transcription and programmed cell death.86Affected chondrocytes also exhibited more short-lived transcripts of genes involved in extracellular matrix turnover, contributing to phenotypical instability in these chondrocytes.86
The role of synovium
The role of synovial inflammation in the pathophysiology of cartilage degradation has been a controversial issue, but recently has won favour with studies showing lymphocytic infiltration and perivascular lymphoid aggregates with release of inflammatory mediators that directly promote cartilage degradation.87–89 Recent research shows that the procatabolic mediators alarmins S100A8 and S100A9 are increased in OA joints and appear to be closely associated with synovial inflammation, allowing a potential target for treatment.90 ,91 Similarly proteinase-activated receptor-2 is elevated in human OA cartilage and synovium, and in vitro studies have shown that its ablation can modulate synovial macrophage activation and thereby confer chondroprotection.92
There is evidence that the pain associated with OA originates from inflamed synovium, with increased levels of the proalgesic nerve growth factor (NGF) found in synovial fibroblasts and macrophages.93 Chondrocytes also appear to have the capacity to produce NGF,94 ,95 and this has successfully been exploited in a study using anti-NGF antibodies to modulate pain in a rat model.96
OA is a heterogeneous disease with differing phenotypes which could be due to different risk factors working through distinct processes.97 Some of the main risk factors are discussed below, with specific studies to illustrate the evidence for each
Salient risk factors for osteoarthritis
1. Advancing age
2. Body mass index>30 kg/m2
3. Female gender
4. Limb malalignment
5. Poor muscle strength and control
6. Previous knee injury
7. Heavy physical labour including kneeling and squatting
8. Genetic predisposition
9. High-level athletic activity
A 22-year prospective study of Finnish participants98 researched the association of new cases of OA over time, diagnosed by physicians using information on disease histories, symptoms and standardised clinical examinations. The risk of developing knee OA was strongly associated with the heaviest category of physical stress at work (compared with the lightest category), and past knee injury.
A recent meta-analysis pooling 24 observational studies (20 997 participants) concluded that knee injury is a major risk factor for the development of knee OA irrespective of study design and definition of knee injury.99 It identified knee injury as one of the few modifiable risk factors for OA, advocating its prevention in future public health programmes.99
In a separate study, follow-up of 44 surgically treated knee dislocations showed that at a mean of 10 years follow-up, the incidence of OA was 23%,100 again indicating the significant association between major knee trauma and the subsequent development of OA.
Anterior cruciate ligament (ACL) rupture and meniscal tear are major independent risk factors for OA.72 A study of 205 soccer players with ACL injuries showed that 41% developed advanced degenerative changes at 14 years follow-up, compared with 4% of uninjured knees.101 In a similar study of female soccer players, 51% had radiographic OA 12 years after ACL injury.102 There is also a high predisposition to OA with meniscal injury; a multicentre study showed that at 30-month follow-up, untreated meniscal injuries had an OR of 5.7 greater risk of OA.103 A fourfold risk was seen after partial meniscectomy at 16-year follow-up.104
Although it is conventionally believed that OA arises in such injured joints due to the abnormal biomechanical forces acting in the destabilised knee, studies suggest an alternative or additional mechanism. Studies in ACL-deficient knees,105 ,106 and recently those researching meniscal tears,89 ,107have shown proteases and inflammatory cytokines are released into the joint following injury that may directly degrade cartilage. Consequently, intra-articular inhibition of IL-1 in a mouse model significantly reduced cartilage degradation and synovitis following tibial plateau fracture.108
Impairments in muscle function
Muscle weakness, altered muscle activation patterns and proprioceptive deficits are commonly found in association with knee OA.109 Improvement of muscle strength (particularly quadriceps) is a key component of conservative management of knee OA and has not only been found to be effective in improving pain, physical function and quality of life110 but also to reduce the risk of developing symptomatic OA in the first instance.111 This effect does not appear to result from any detectable structural changes on knee MRI.112 The underlying mechanism is not fully understood, but one theory is that the knee extensors shock absorb and stabilise the knee during loading, so that their deficiency can lead to excessive mechanical stress on articular cartilage.109 ,113
Knee OA has a strong female sex preponderance.114 In a prospective 12-year study of 315 495 Norwegians, the rate of knee replacement was double for women (0.55%) than men (0.28%).115 Obesity is a stronger risk factor for knee OA in women than in men. Women also appear to be affected more than men if categorised in the moderate OA group with weaker performance scores and greater impairment to activities of daily living.116
Some of the reasons for this are speculated to be: women lose knee articular cartilage at a faster rate than men, female human articular chondrocytes may function better when oestrogen is available, male human articular chondrocytes are more responsive to vitamin D metabolites than female cells, vitamin D receptors and mRNA for inflammatory cytokines are differentially expressed in degenerated cartilage in a sex-specific fashion, and subchondral bone osteoblasts exhibit sex-specific responses to oestrogen.117 One case–control study suggests that low intake of vitamins D C are possible risk factors for OA, especially in females.118
Heavy physical work
The relationship of heavy physical work to OA has been the subject of considerable interest and investigation.119 Of particular note, vibration, repetitive movement, long hours of kneeling, squatting and standing have been shown to be associated with an increased risk of development of OA.120 ,121 Frequent knee bending while loading is another activity that has been related to cartilage degeneration.122
A case–control study in Germany involving 1310 patients with and without symptomatic knee OA suggested a dose–response relationship between kneeling/squatting and symptomatic knee OA.123 Occupational risks such as jumping or climbing stairs/ladders, however, did not correlate with symptomatic knee OA.123 There are particularly deleterious interactions of high body mass index (BMI) with kneeling/squatting and heavy lifting.124
A cross-sectional study of 2439 participants with OA showed that exposure to non-elite running at any time in life was not associated with higher odds of prevalent knee pain, symptomatic OA or radiographic OA.125 Indeed, sporting activities may even have a protective effect if not associated with traumatic injury.125 ,126 A separate literature review concluded that low-distance and moderate-distance running is not associated with OA, with long-distance running, barefoot and minimalist shoes being inconclusive.127
The literature is less optimistic with regard to higher level athletic activity with some low-level evidence that it may be associated with later development of hip and knee OA.128
In a prospective 12-year study of 315 495 Norwegians, increased BMI was directly correlated with increased risk of requiring knee replacement.115 Men with a BMI>27 had a sixfold increased incidence of knee replacement compared with men with a BMI<23. Women with a BMI>26 had an 11-fold increase compared with women with a BMI<21. Combining heavy labour with high BMI was particularly hazardous, with the risk increasing to 12-fold in men and 16-fold in women.
Increased awareness of the relationship between obesity and metabolic and inflammatory activities has made researchers rethink the role of obesity and OA. The relationship of obesity to the development of knee OA is not necessarily proportional to the severity of obesity. Development of OA appears to be strongly related to the coexistence of disordered glucose and lipid metabolism.129 ,130 Cytokines associated with adipose tissue, including leptin, adiponectin and resistin, may influence OA through direct joint degradation or control of local inflammatory processes.129 ,131
Metabolic risk factors including obesity, hypertension, dyslipidaemia and impaired glucose tolerance (collectively known as the ‘metabolic syndrome’) raise not only the risk of occurrence of OA, but also its progression.132 This risk rises with the increasing number of metabolic risk factors present, illustrating the significant relationship between these risk factors and OA.133Obesity is a particular issue in Western countries and is one of the few modifiable risk factors that can be a powerful tool in the reduction of OA prevalence in these countries.134
A meta-analysis of four genome-wide association studies with a cumulative sample size of 6709 cases and 44 439 controls in Caucasian populations identified a significant susceptibility locus for knee OA on chromosome 7q22.135 Other studies have uncovered a likely role in OA for the genes encoding structural extracellular matrix components (such as DVWA) and molecules involved in prostaglandin metabolism (such as DQB1 and BTNL2).136 A powerful association study showed that the genetic polymorphisms predisposing to knee OA differed from those for hip OA.137Recently, the genes encoding NCOA3, SULF2, ALDH1A2 have all been associated with hip and hand OA, but not necessarily knee OA .138
With an ageing population globally, OA is a growing problem with an increasing burden on the health of populations and the economies of nations worldwide.8 ,9 Future health strategies will thus need to focus on earlier diagnosis and preventative intervention in order to cope with such demand. This will be aided by advances in our understanding of OA pathophysiology that will continue to be made through laboratory-based studies, which have recently flourished in recent years with goal-orientated and clinically relevant research. Despite OA being a disease that has affected generations from antiquity, we are only starting to comprehend the complex processes involved.21 As we acknowledge the role of synovial inflammation and other intra-articular structures in OA, new pathways will continue to be discovered and with every discovery, there is the potential for targeted therapy.26 ,90 ,91 The next 3–5 years will reveal if indeed different risk factors work towards OA through ‘signature pathways’ that can then be targeted in personalised therapies.92 ,108
Technological advances will continue to be made in MRI, further enhancing our understanding of the pathophysiological processes and also allowing earlier diagnosis.25 ,51 The role of physiological MRI in the knee will assume a bigger role in the years to come and the complex and time-consuming analysis currently involved in semiquantitative measurements will become streamlined.26
In the next 5 years, biomarkers have the potential to make a great difference in the way we diagnose and treat patients. From detecting the disease process at such an early stage, treatments can be developed to tackle the pathophysiology in its most early stages, when it may be possible to abort or delay progression.24 Research in the next few years should focus on elucidating the best combination of biomarkers to most sensitively and specifically detect early OA, for different risk factor profiles. Indeed, the OARSI has made recommendations that future OA clinical trials should collect biospecimens to allow measurement of biomarkers for a number of reasons, including the early detection of disease, the assessment of treatment efficacy and to identify patient subgroups that may respond better to certain types of intervention.141
Valuable recommendations have also been made to ensure future trials use standardised nomenclature that will aid definition of different OA phenotypes in an attempt to unify disease concepts in what is a heterogeneous disease.142 In the next few years, we should see the development of tools that predict risk of development to OA, analogous to the fracture risk scores in osteoporosis. Risk stratification can be divided into risks of OA development and risks of OA progression.142 Earlier diagnosis of OA will be the first major step in better treatment strategies and there are sensible calls for consensus around more sensitive and specific diagnostic criteria for OA, including biomarkers, to allow formulation of tools for identifying disease in its early preradiographic and/or molecular stages—Reliable Early Disease Identification (REDI).142
The various risk factors for OA will continue to be researched with specific attention to the pathological mechanisms that may be unique to each risk factor.129 ,130 Public health education should focus on the two main modifiable risk factors of obesity and knee injury to try and stem the dramatic rise in OA rates expected globally.8 ,9 In the field of genetics, we will see further mapping and identification of risk loci, additional functional studies, and further integration with other genome-wide approaches, with the newfound realisation that different polymorphisms may exist for different sites of OA in the skeleton .138
1. Continued research into the pathogenesis of osteoarthritis (OA) including the role of synovium and other intra-articular structures, with targeted therapy for implicated pathways (such as those in inflammation).
2. Further research into the risk factors of OA and the unique pathophysiological processes that may underpin each one.
3. Risk factor scoring for predicting total OA risk.
4. Public health campaigns to address modifiable risk factors such as obesity and knee injury in an attempt to reduce the impact of dramatic increases in OA prevalence globally.
5. Further advanced MRI sequences to detect cartilage degeneration early by measuring proteoglycan content, with faster analysis of results.
6. Use of serum biomarkers to detect OA in early stages, with identification of the best combinations for optimum sensitivity and specificity.
7. Using consensus nomenclature in future trials to unify disease concepts to allow identification of specific OA phenotypes.
8. Further identification of genetic susceptibility loci with possible targeted therapy.
Contributors SE-T prepared the manuscript and EA and DP revised and co-authored the final manuscript.
Competing interests None declared.
Provenance and peer review Commissioned; externally peer reviewed.