Several epidemiological and experimental data support the hypothesis that diabetes could be an independent risk factor for osteoarthritis (OA), at least in some patients, leading to the concept of a diabetes-induced OA phenotype. If confirmed, this new paradigm will have a dramatic impact on prevention of OA initiation and progression.
Statistics from Altmetric.com
Osteoarthritis (OA) is the leading cause of musculoskeletal handicap in the world. Ageing and obesity are the two main risk factors for OA. Since prevalence of these conditions are going to exponentially expand, an epidemic of the disease in the next decade is expected, leading to a dramatic increase in the number of total joint replacements and so entails significant costs to society. In order to attenuate the individual and societal consequences, and because no disease-modifying drugs have proven their efficacy yet, any preventive policies should have a dramatic impact on the quality of life and on countries economy. A better delineation of the different risk factors of the disease should lead to a better personalisation of the preventive messages delivered by doctors and stakeholders. To date, the main OA phenotypes described in the literature are ageing, post-traumatic, hormonal, genetic and metabolic OA.1
Metabolic OA has been recently individualised based on recent data showing an increased incidence of OA in patients who are overweight/obese even in non-weight bearing joints. In that case, mechanical overload being not able to explain this increased incidence, a new paradigm based on the role of systemic mediators called adipokines and defined as cytokines produced by fat adipose tissue, has been proposed.2 Moreover, recent epidemiological studies have strengthened this hypothesis by showing an increased incidence of OA in patients with metabolic syndrome (MetS). MetS, also known as syndrome X, is defined as a condition mixing several independent risk factors for cardiovascular (CV) events, including insulin resistance (identified by type 2 diabetes, impaired fasting glucose or impaired glucose tolerance) plus any two of the following: hypertension, elevated plasma triglycerides, decreased high-density lipoprotein cholesterol, obesity, proteinuria. Indeed, estimation of the prevalence of diabetes reaches over 10% of the population in industrialised countries, coexisting with obesity in specific geographic patterns because of a convergence of prevailing social norms, community and environmental factors, socioeconomic status and genetic risk factors among ethnically similar groups.3 The demonstration of an association between MetS and OA is challenging because obesity, a component of the MetS, is also a strong risk factor for knee OA.4 However, a large body of evidence indicates that OA is part of a generalised metabolic disorder in which various interrelated metabolic factors contribute to the OA process.5 A recent logistic regression analysis assessing the association between MetS and population-weighted variables in a representative sample of the general US population showed appealing results.6 Interestingly, MetS was prevalent in 59% of the OA population and in 23% of the population without OA. Each of the five CV risk factors that comprise MetS was more prevalent in the OA population. This association remained strong when obesity was controlled for. It is noteworthy that in this work, prevalence of diabetes was 30% in the OA population versus 13% in the control population. The role of diabetes independently of obesity as a risk factor for OA remains unclear.
The first paper describing an association between OA and diabetes was published in 1961 (table 1).7 The authors looked at the occurrence of definite radiological OA in 6 anatomical areas of 30 patients with diabetes and 30 matched controls. The correlation was statistically significant for feet and knees. OA patterns were also studied in 809 patients with knee or hip joint replacement due to OA according to the presence of non-insulin-dependent diabetes.8 Patients with non-insulin-dependent diabetes more frequently had bilateral OA (adjusted OR 2.2; 95% CI 0.8 to 6.4). An association with generalised OA was not seen in this study maybe due to statistical considerations. The clinical, pathological and epidemiological relationships between fasting plasma glucose concentrations (FPG) and the sites of lesion in OA were evaluated in 1026 patients.9 The mean FPG (99±22.2 mg/dl) was significantly higher in OA (p<0.01) than in the normal controls (88±19.9 mg/dl). Interestingly, erythrocyte sedimentation rate (p<0.01) and pain at rest (p<0.02) were higher in patients with non-insulin-dependent diabetes. In all, 1003 women aged 45–64 from the Chingford population study completed risk factor questionnaires for CV events.10 For knee OA in either knee, the variables significantly associated were raised blood glucose OR 1.95 (95% CI 1.08 to 3.59) and moderately raised serum cholesterol OR 2.06 (95% CI 1.06 to 3.98). In a recent study testing the hypothesis that vascular cell adhesion molecule 1 could be a predictor of severe knee or hip OA, the authors compared patients who did (n=60) or who did not (n=852) undergo joint replacement surgery in 1990–2005 due to severe OA.11 Although not discussed in this paper, an increased prevalence of non-insulin-dependent diabetes was seen in the group of patients who had joint replacement (18.3% vs 6.3%) but this difference was not statistically significant (p=0.06). In contrast, one negative case-control study found no association between an impaired oral glucose tolerance test or non-insulin-dependent diabetes and another found a tendency but not a statistical difference.12 13
Based on these case-control studies, we cannot definitively conclude that diabetes/hyperglycaemia is an independent risk factor for OA since many potential confounders may interfere with the results, such as age, weight, level of activity etc. However, taken together, there may be a positive signal for an independent correlation between OA and diabetes.
In vitro data
Chondrocytes are glycolytic cells able to sense the concentration of glucose present in the cartilage matrix, the synovial fluid and in a less extent the subchondral bone being the sources for glucose.14 By sensing glucose, chondrocytes will then respond appropriately by adjusting their cellular metabolism. Chondrocytes express multiple isoforms of the glucose transporter (GLUT)/SLC2A family of glucose/polyol transporters which represent the first rate-limiting step in glucose use.15 Among them, GLUT-1 is especially important as it is regulated by anabolic and catabolic stimuli.16 However, there are very few studies assessing the role of high extracellular glucose concentrations in articular chondrocyte functions. Hyperglycaemia decreases dehydroascorbate transport into chondrocytes, which can compromise the synthesis of type II collagen, and hyperglycaemia increases reactive oxygen species (ROS) production, known to be major deleterious mediators for cartilage destruction.17 18 Interestingly, although normal chondrocytes can adjust their intracellular concentration to local glucose concentration, OA chondrocytes exposed to high glucose are unable to down regulate GLUT-1, accumulating more glucose and producing more ROS.19
Advanced glycation end products (AGEs), the products of non-enzymatic glycation and oxidation of proteins and lipids, accumulate in several diabetic tissues such as the vasculature due to hyperglycaemia.20 21 Similarly, AGEs accumulate in ageing and in OA cartilage leading to matrix stiffness, becoming more sensitive to mechanical stress.22 Along with their deleterious effects on cartilage matrix, AGEs can also bind to membrane receptors called RAGE (for ‘receptor for age glycation end products’) present on many cell types including chondrocytes.20 23 Once bound to RAGE, AGEs trigger the activation of different signalling pathways leading to the over expression of proinflammatory and prodegradative mediators and to some alterations in the chondrocyte differentiation phenotype.24,–,26 To the best of my knowledge, to date no experimental studies have assessed the level of AGEs formation in diabetic cartilage compared to normal cartilage, even in human or in animal cartilage. However, it seems realistic to consider that an increased concentration of glucose in the diabetic cartilage matrix environment would lead to the same deleterious result.
In vivo data
In the well known streptozotocin-induced diabetes rat model which mimics type I diabetes, cartilage becomes resistant to the anabolic action of insulin-like growth factor 1 (IGF-1), a condition that is correctable by hypophysectomy, suggesting a metabolic impairment at the tissue level.27 In this model, it is noteworthy that after 70 days, there are intense remodelling and collagen deposition in the synovium.28
Human experimental data
Motor and sensory dysfunction of muscle may be important factors in the pathogenesis of articular damage.29 It is well accepted that patients with OA have muscle weaknesses and a vibratory sense loss in the regional OA joint.30 Interestingly, this neurological dysfunction is present locally and throughout the body, at least in patients with hip OA, suggesting a generalised alteration of the peripheral nervous system.31 Two main neurological syndromes have been described in diabetes. The first is the Charcot neuroarthropathy, a rare but devastating complication leading to joint deformity and eventually amputation or secondary OA. The second is a symmetric, mainly sensory polyneuropathy often accompanied by autonomic neuropathy.32 This latter diabetic neuropathy could be one of the suggested alteration of the peripheral nervous system seen in patients with OA leading to muscle weaknesses and joint laxity. I speculate that such peripheral nerve impairment induced by diabetes could be an added risk factor for OA in patients with diabetes.
Diabetes was first considered as a non-inflammatory disease. However, we now know that hyperglycaemia can trigger a low-grade systemic inflammation which would explain the increased risk of CV events seen in patients with diabetes.33,–,36 Low-grade systemic inflammation is associated with cartilage loss.37 Although MetS is a well known association with mild systemic inflammation, I propose that an independent hyperglycaemia-induced systemic inflammation may also have an impact on the progression of OA.
Based on these strong signals for a correlation between diabetes and OA (see figure 1), it is time to encourage the scientific community to perform prospective studies devoted to confirm or invalidate this hypothesis. In vitro and in vivo experimental studies should now be designed in order to have a better understanding of the mechanisms underlying the potential interactions between both diseases. These clinical and experimental studies would also help to decipher the respective roles of type 1 and type 2 diabetes in this suspicion of increased risk for OA. Finally, a better knowledge of the specificities of a diabetes-induced OA phenotype compared to the others should lead to a personalised approach of preventive and curative treatments for OA.
▶ What is the prevalence of OA in patients with non-insulin-dependent and insulin-dependent diabetes?
▶ Is insulin a potential protector against cartilage degradation?
▶ What are the characteristics of a diabetes-induced OA phenotype?
▶ Are antidiabetic drugs active for counteracting the OA process? (Such as peroxisome proliferator-activated receptor γ agonists, etc.)
I would like to thank Professor C Boitard for many helpful discussions.
Competing interests None.
Patient consent Obtained.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.