The adverse effects of diabetes on osteoarthritis: update on clinical evidence and molecular mechanisms

SUMMARY

Projected increases in the prevalence of both diabetes mellitus (DM) and osteoarthritis (OA) ensure their common co-existence. In an era of increasing attention to personalized medicine, understanding the influence of common comorbidities such as DM should result in improved care of patients with OA. In this narrative review, we summarize the literature addressing the interactions between DM and OA spanning the years from 1962 to 2014. We separated studies depending on whether they investigated clinical populations, animal models, or cells and tissues. The clinical literature addressing the influence of DM on OA and its therapeutic outcomes suggests that DM may augment the development and severity of OA and that DM increases risks associated with joint replacement surgery. The few high quality studies using animal models also support an adverse effect of DM on OA. We review strengths and weaknesses of the common rodent models of DM. The heterogeneous literature derived from studies of articular cells and tissues also supports the existence of biochemical and biomechanical changes in articular tissues in DM, and begins to characterize molecular mechanisms activated in diabetic-like environs which may contribute to OA. Increasing evidence from the clinic and the laboratory supports an adverse effect of DM on the development, severity, and therapeutic outcomes for OA. To understand the mechanisms through which DM contributes to OA, further studies are clearly necessary. Future studies of DM-influenced mechanisms may shed light on general mechanisms of OA pathogenesis and result in more specific and effective therapies for all OA patients.

Introduction

Diabetes mellitus (DM) and osteoarthritis (OA) are common diseases that are predicted to increase in prevalence in the US and worldwide^1,2. Because of the resultant rise in the coexistence of these two diseases and a burgeoning body of evidence suggesting that DM may adversely affect articular tissues and exacerbate OA, it seems an opportune time to review literature in this area and consider ways in which these diseases interact. The ultimate goal of this field of study is to understand OA in the diabetic patient in order to individualize therapies and prognosis for patients having both of these common diseases.

DM is defined by metabolic abnormalities resulting from dysfunction in the glucose-handling machinery of the body. It occurs in two major forms. Type 1 DM (T1DM) is caused by reduction in quantities of insulin which regulates glucose metabolism. It is the most common form of DM in children and may have an autoimmune or post-infectious etiology. Although type 2 DM (T2DM) may occur in children, it comprises the vast majority of adult-onset DM. T2DM typically involves insulin resistance at the cellular level. The metabolic abnormalities of DM are shared by all forms of the disease; effective reduction of insulin activity causes prolonged hyperglycemia which leads to osmotic and oxidative stress and results in damage to the kidneys, eyes, nerves, and other tissues. By the year 2035, the number of adults with DM worldwide is estimated to be 592 million¹.

Like DM, OA is also rising in prevalence and affects millions². It is typically defined as a degenerative articular process characterized by eroded articular cartilage, altered subchondral and peri-cartilagenous bone, mild to moderate synovial inflammation, and pain. While cartilage damage is the final common endpoint in OA, primary abnormalities of other tissues such as tendon, bone, or muscle may contribute to or even cause OA. DM, especially T2DM, and OA share many epidemiologic features. They are both complex diseases with considerable clinical heterogeneity and multifactorial etiologies involving interactions of genetic and environmental factors. They also share common risk factors. Aging is a risk factor for both T2DM and OA. The prevalence in the US of DM is 3.3/1000 among individuals aged 18–44, and rises to 15.4/1000 for ages 65–79³. Similarly, OA dramatically increases with age, affecting 13.5% of adults 25 years and older and 33.6% of those over the age of 65⁴. Another important risk factor for both diseases is obesity. The association of OA with obesity is well-supported⁵, and obesity occurs in the majority of people with T2DM⁶.

OA and DM frequently co-exist simply by chance due to their high prevalence and shared risk factors. Nearly half (47.3%) of patients with DM have some form of arthritis⁷. The presence of co-morbid conditions typically increases the care needs of individual patients, decreases the effectiveness of care, and escalates health–related costs. In addition, therapeutic strategies that emphasize personalized medicine and take into account co-morbid conditions may result in improved outcomes for patients with OA⁸. The development of OA may also complicate DM. While not the focus of this review, there is increasing evidence that OA adds to the burden of cardiovascular disease, which is higher than average in DM patients⁹.

For this narrative review, we have searched the literature using combinations of the following key words: “osteoarthritis”, “diabetes”, “hyperglycemia”, “articular cartilage”, “bone”, “tendon” and “ligament”. We critically assess the literature exploring the effects of DM on both OA and its management strategies. The few clinical studies addressing this issue strongly support an association between DM and OA. While animal studies are also quite sparse and suffer from some methodological weaknesses, they also suggest that DM may accelerate OA progression. Further, current evidence derived from basic science studies suggests that DM may adversely affect the homeostasis and repair of articular tissues. These research studies give credibility to the hypothesis that DM has clinically relevant effects on OA and warrant further study.

Clinical studies

Population-based studies of DM and OA

There are only a handful of studies that directly address the effects of DM on OA. Two cross-sectional population-based studies using questionnaires found that DM was significantly more common in an OA cohort than a non–OA cohort^10,11. In contrast, another study found no increase in DM among OA patients; however, the study subjects’ healthier lifestyle may have influenced the results¹². The observation that specific features of OA may be more common with DM – such as increased enthesophytes or more severe cartilage degeneration – supports an interaction between these diseases^13,14.

Multiple studies have used rates of arthroplasty as a surrogate for OA severity in examining the role of DM in OA. In a cohort of patients undergoing arthroplasty for OA, no relationship was found between T2DM and bilateral OA or generalized OA after adjustment for age, sex, BMI, and other confounders for OA¹⁵. However, the study design based on cohort analysis of arthroplasty patients limits the conclusions. This is due to two issues; 1) a bias effect since OA severe enough to require arthroplasty eliminates potential study subjects with early stage OA, and 2) a survivor effect since uncontrolled diabetes may have been a cause to deter study subjects from arthroplasty.

To avoid these effects, two more recent studies examine a cohort of DM patients with non-DM patients as control^16,17. The first study used the Veterans Affairs Health Administration (VHA) national administrative database, and DM was defined as having two DM-related in- or out-patient visits or being prescribed anti-diabetic medication¹⁶. The cohorts included 450,000 with DM and over 3 million without DM. Rates of total joint arthroplasty were increased in patients with DM. The statistically significant odds ratios ranged from 1.2 (primary total knee) to 3.4 (revision total hip). When these data were stratified by age, the highest risk ratios for each joint were found in the youngest age group studied (46–55 years of age). The major advantage of this study was its size, while the major disadvantage was that data on body mass index (BMI) were not available. Overall, if one uses arthroplasty as a surrogate for severe OA, this study suggests that DM has a negative effect on OA by either increasing its prevalence or by increasing the rate of progression¹⁶.

A similar but smaller study¹⁷, analyzed retrospective data from a prospective study with data collected over 20 years from a regional hospital in Northern Italy. Of the 927 subjects that began the prospective study, 69 met criteria for T2DM, defined as having a fasting blood glucose level over 126 mg/dL or a clinical diagnosis with use of anti-DM medication. The authors found a significant and positive association of DM with arthroplasty; relative risk was about two and remained significant in models that controlled for age, sex, and BMI. Using a subset of the same population (347 total, 43 with T2DM), the authors found worse functional measures in the T2DM group. In particular, the pain variable was significantly higher in both the KOOS (Knee Injury and Osteoarthritis Outcome Score) and the WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index) questionnaires. In addition, knee synovitis was significantly more common in people with DM. The major advantages of this study were that known risk factors for OA, particularly BMI, were controlled in the statistical models and that the characteristics of OA (pain, function, and synovitis) were further probed. The disadvantage of this study was its small population size.

Despite the differences in approach, these two population-based studies both found that DM doubles the risk for arthroplasty^16,17. The VHA database was large enough to determine that persons with DM have arthroplasty at younger ages¹⁶; while the smaller single hospital study identified that the hazard risk associated with DM remained after adjustment for other predictors for OA¹⁷. Obviously, arthroplasty is not a perfect outcome measure in OA, and additional clinical studies comparing individual outcome measures such as pain, function, or structural damage will be necessary in the future.

Other clinical studies of DM and OA

Several recent studies began to approach the role of DM in contributing to important outcome measures in OA, including OA progression and pain. The effect of DM on OA progression was recently studied in a cohort of 559 subjects with well-characterized knee OA¹⁸. At baseline, 6.6% of the participants had T2DM. In men with T2DM, annualized joint space narrowing was statistically significantly higher than in non-diabetic men matched for BMI, age, hypertension and dyslipidemia. Of all the components of the metabolic syndrome, only DM was identified as an independent risk factor for knee OA progression. In a study of 530 subjects with radiographic hand OA, primary analysis showed no clear association between pain and DM¹⁹. Subset analysis, however, showed higher pain scores in patients with DM who had erosive OA defined as radiographic evidence of more than one IP joint with erosions. The impact of these conclusions is uncertain since the field does not have standard classification criteria for erosive OA. Certainly, further validation of these findings in OA is warranted.

Effect of DM on OA clinical care

The impact of DM on clinical care has been studied frequently, with particular attention paid to the effects of DM on arthroplasty outcomes. Arthroplasty patients with DM sampled from the Nationwide Inpatient Sample database have increased postsurgical death^20,21. In contrast, older patients in the Medicare database had no increased risk of postsurgical death with DM^22,23. Several studies have identified decreased functional outcomes for arthroplasty performed on patients with DM^24,25. The most commonly reported DM-associated complication is the increased rate of infection^{21,22,25–29}. DM may also increase the need for revision arthroplasty^25,30,31. Some arthroplasty outcomes may pertain to metabolic effects on joint tissues such as defective bone healing, while others reflect general risks of DM associated with major surgical procedures. All of these effects can add considerable cost to clinical care^20,21,32.

Effect of DM on human articular tissues

Studies of tissues from patients demonstrate that changes capable of promoting OA occur in articular tissues in patients with DM. For example, ankle cartilage from DM patients was softer with lower stiffness indices and larger permeability parameters than cartilage from non-diabetic patients³³. Despite the limitations of this study – including the use of frozen tissues, the focus on a joint infrequently involved in OA, and the significant age difference of control subjects – the observed changes in biomechanical properties of articular cartilage might contribute to the development of OA. Advanced glycation endproducts (AGE) may accumulate prematurely in diabetic tissues and affect biomechanics^34–37. A single, small study using human tissue showed increased levels of the AGE pentosidine in bone from patients with DM at the time of joint replacement, but no significant changes in pentosidine levels in cartilage³⁸. Tendon abnormalities may also contribute to OA. DM patients demonstrate biochemical and biomechanical changes in tendons that include decreased rupture threshold and disordered collagen fibrils³⁹. There is also mounting evidence that bone healing is defective in DM patients^40–46. Increased rates of non-healing microfractures in patients with DM may alter bone mechanics and promote OA, and they may also contribute to poor arthroplasty outcomes. Multiple clinical studies demonstrate increased fracture risk in post-menopausal women with T2DM which is not simply linked to low bone mineral density on clinical densitometry^41,42. This paradox may be explained by changes in bone microarchitecture, including increased cortical porosity⁴³, excess mineralization, and reduced subchondral bone heterogeneity⁴⁴ with DM. Patients with DM have reduced serum biomarkers of bone turnover as well as elevated sclerostin levels^45,46. Such changes could lead to increased bone stiffness which could accelerate OA.

Animal studies

The presence of multiple comorbidities and other inherent challenges in designing clinical studies to delineate mechanisms responsible for increased OA in DM patients have logically led to the use of animal models of these diseases. While primate and rabbit models of DM have been described^47–49, rodents are most commonly used to replicate human DM^50,51. Very few studies directly address the severity or progression of OA in diabetic animals.

Animal models of T1DM

In T1DM, insulin production is absent or significantly impaired⁵². Genetic models of T1DM such as the Akita mouse (C57BL/6-Ins2^Akita) are available on several background strains^53,54. A non-genetic and very common method of modeling T1DM is to chemically damage the pancreatic islets with single or multiple dose regimens of streptozotocin (STZ)^55–57. There are a few rat models of T1DM including the LEW.1AR1/-iddm⁵⁰ and the BBDP rat⁵⁸ both of which involve spontaneous autoimmune pancreatic damage.

Animal models of T2DM

In order to model human T2DM, the animal should have hyperglycemia non-fasting blood glucose >250 mg/dL^50,51,59,60, insulin insensitivity, and preferably additional characteristics such as polyuria and polydipsia^49,50. Further information on protocols for characterizing DM in rodents can be found in the literature^61,62. However, as with models of T1DM, no mouse model perfectly mirrors human T2DM disease. The relative contributions of obesity to acceleration of OA in DM are particularly important in T2DM and its models. While isolating the biomechanical effects of obesity from the effects of DM is itself a challenge, in addition, obesity has well-known metabolic consequences that add further complexity to the situation. These include increased production of inflammatory cytokines such as IL-1, TNFα, and adipokines (e.g., adiponectin, leptin, and resistin)⁶³ that promote inflammation and may accelerate OA. Because obesity and T2DM almost always co-exist in human T2DM, animal studies may be particularly useful in separating the pathological contributions derived from these different mechanisms.

A popular model of T2DM is the diet-induced obesity model (DIO) which attempts to simulate human obesity-induced T2DM but often results in only modestly increased glucose levels^51,64–66. The db/db mouse, ob/ob mouse, and fa/fa rat have monogenetic defects that disrupt leptin signaling resulting in hyperphagia and morbid obesity⁶⁷. This leptin signaling defect is rare in human T2DM^60,64,67, and T2DM models created using polygenic mutations are better models of human T2DM. The KK.Cg-A^y/J mouse was developed by crossing a spontaneously diabetic strain (KK) with the yellow obese strain (A^y) resulting in a mature-onset T2DM phenotype⁶⁸. Both DM and control normal glycemic siblings are obese⁶⁹. The NONcNZO10/LtJ mouse (NcZ10) is a polygenic model of T2DM with onset of DM in adulthood. The NcZ10 model requires a chow content of 10–11% fat for higher penetrance (90–100%)⁵⁹. The TALLYHO/JngJ mouse (TH) has polygenic and maturity-onset diabetes and has high penetrance in males^59,70. A short list of commonly available rodent models of T1DM and T2DM is presented in Table I. Readers are encouraged to review additional details in the references provided.

Table I.

Commonly used rodent models for type 1 and type 2 DM

Model name* (short name)	DM type	Preferred control	Advantages	Disadvantages	References
C57BL/6-Ins2Akita (Akita)	T1DM	Same strain background	Available on other backgrounds	Genetic background modulates DM phenotype	50,52–54
Streptozotocin (STZ)	T1DM	Same strain with vehicle treatment	Economical Can be used in most rodent strains and other species	Drug toxicity	50,52,55
Alloxan monohydrate (Alloxan)	T1DM	Same strain with vehicle treatment	Economical Can be used in most rodent strains and other species	Less efficacious than STZ method	49,50,52
Biobreeding diabetes-prone rat (BBDP)	T1DM	Biobreeding diabetes-resistant rat (BBDR)	Larger size (rat) Polygenic 80–95% incidence No sex bias	Will not develop DM in a pathogen-free environment	58,71
Diet-induced obesity (DIO)	T2DM	Same strain with vehicle treatment (low fat chow)	Economical Can model pre-diabetes state Works well with C56BL/6 and NON strains	Mild or no hyperglycemia	51,64,66
C57BLKS-Leprdb (db/db)	T2DM	Heterozygous (+/db) littermates	One of the most studied models No sex bias	Monogenic Juvenile-onset Early death	51,59,67
C57BLKS-Lepob (ob/ob)	T2DM	Heterozygous (+/ob) littermates	A well-studied model No sex bias	Monogenic Mild, short-term hyperglycemia	511,67
Zucker diabetic fatty rats (ZDF)	T2DM	ZDF lean	Larger size (rat)	Monogenic No or mild hyperglycemia DM in males only	51,67
KK.Cg-Ay/J (KKAy)	T2DM	Homozygous, non-diabetic siblings	Polygenic T2DM Maturity onset	Females have lower blood glucose levels	68
NONcNZO10/LtJ (NcZ10)	T2DM	NON/LtJ, SWR or FVB (Swiss-derived strains) on low fat diet	Polygenic T2DM Maturity onset	DM in males only Requires high fat diet for high penetrance	59,60
TALLYHO/JngJ (TH)	T2DM	NON/LtJ, SWR, or FVB (Swiss-derived strains). C57Bl/6 strain has also been used.	Polygenic T2DM Maturity onset High penetrance	DM in males only	59,70,72

^*Where relevant, the most commonly used background strain is indicated although other backgrounds may be available. DM.

Animal studies investigating a link between DM and OA

There is an unfortunate paucity of quality research studies in animal models of DM relevant to OA. A recent study using STZ-induced T1DM in mice suggested the presence of cartilage damage after 8 weeks of hyperglycemia, and showed elevated levels of circulating AGEs⁷³. Both abnormalities were ameliorated by the use of the DM drug pioglitazone. The authors concluded that this drug response implicated PPARγ in this effect, but it was unclear whether this was related to improvement of hyperglycemia or PPARγ inhibition. One valuable study applied the DIO model to the C57Bl/6 strain and made careful measures of physiological data and histological OA outcomes⁷⁴. This study combined a high fat diet (60% by calorie) with the meniscal ligament injury model to induce OA. Higher OA scores (increased joint degeneration) were found in the mouse group receiving both high-fat diet and ligament injury. However, since hyperglycemia was not established until the last month of the experiment, it is unclear whether the hyperadiposity⁷⁵ or the rising hyperglycemia was the driving factor for greater OA progression. None the less, this is a valuable study demonstrating acceleration of joint degeneration in association with metabolic changes typically seen in DM patients.

Effects of DM and hyperglycemia on articular cartilage

Biomechanical properties of cartilage are highly influenced by the composition of the extracellular matrix (ECM), and there is some evidence that metabolic abnormalities associated with DM alter cartilage ECM. Early studies in animal models of DM have shown that decreased collagen production⁷⁶ and increased proteoglycan catabolism⁷⁷ occurs in DM cartilage. Increased proteoglycan catabolism in DM animals has also been observed in non-articular connective tissues⁷⁸.

Effects of DM and hyperglycemia on bone

Studies showing delayed fracture healing in animals with DM support the ample clinical data associating DM with bone abnormalities. The STZ mouse model demonstrates that poor diabetic fracture healing is related to premature resorption of cartilage at the fracture callus⁷⁹, and that this was due to high levels of the inflammatory cytokine TNFα in mice with DM⁸⁰.

Effects of DM and hyperglycemia on tendons and ligaments

Animal models of DM fairly consistently show histologic and biochemical abnormalities in tendons and ligaments, as well as less well-characterized biomechanical changes, such as lower Young’s modulus and increased intra-substance failure^81–83. Further, several studies identify delayed tendon healing after injury in the presence of DM^84–87. These types of tendon and ligament abnormalities are known to promote OA⁸⁸.

Effects of DM and hyperglycemia on synovium

The synovium of T1DM rats is abnormal and contains fibrotic tissue with higher amounts of type I collagen and lower quantities of types III and V collagen⁸⁹. Synovial pathology is a significant contributing factor to OA; early inflammatory changes in the synovium may cause damage that then creates long-term production of catabolic mediators⁹⁰.

Potential mechanisms

Basic science studies have also identified some potential mechanisms linked to DM-influenced end-organ joint damage. Mediators such as hyperglycemia, AGEs, sorbitol, adipokines, and cytokines act through oxidative, osmotic, and inflammatory mechanisms to produce tissue damage (Table II). Further complexity is added by the participation of similar molecular participants in multiple catabolic pathways and the contributions of similar metabolic derangements in obesity.

Table II.

Mechanisms present in diabetes that can damage joint tissues

Potential DM mechanisms	Contributing pathways	Examples of major mediators	Examples of molecular participants	References
Inflammation	AGE/RAGE, innate immune pathways	Cytokines, Adipokines, Reactive oxygen species	MMPs, TLR	91,92
Oxidative stress	Glucose transporters AGE/RAGE	Reactive oxygen species	GLUT, ATP, ADP	93–95
Osmotic stress	Polyol pathway	Sorbitol	p38 MAPK	96,97

AGE = advanced glycation end products, RAGE = receptor for AGE, MMP = matrix metalloproteinase, TLR = toll-like receptors, GLUT = glucose transporter family, ATP = adenosine triphosphate, ADP = adenosine diphosphate, MAPK = mitogen-activated protein kinase.

An increasing recognition of a key role for inflammation in both OA⁹⁸ and DM provides an important mechanistic link between these two conditions. Significant synovitis occurs in OA and may be exacerbated by the increased levels of inflammatory cytokines, adipokines, and prostaglandins seen in DM tissues⁶³. Signaling through pathways of innate immunity, such as toll-like receptors, also may produce inflammation in both DM⁹² and OA⁹⁹.

In a hyperglycemic environment, there is increased production of reactive oxygen species contributing to tissue damage. Cellular transport of glucose becomes critical and may contribute to excess oxidative stress. One group found that human OA chondrocytes from donors aged ≥66 years cultured in high glucose are unable to decrease GLUT1 protein or reduce glucose transport activity in comparison with normal chondrocytes from young donors (age 28–35 years)^94,100. They also found that high glucose conditions favored production of oxidants and promoted matrix catabolism which would accelerate OA. However, age and media osmolarity were not controlled in those experiments¹⁰¹. The effects of high glucose may be associated with impaired function of ATP-sensitive K+ channels which couple GLUT to intracellular ATP/ADP levels and membrane potential^102,103.

The AGE/RAGE system also plays a role in DM end-organ damage through induction of inflammation and/or increased oxidative stress. Collagen has an extraordinary low turnover in many connective tissues and as such is prone to modification by AGEs. The formation of AGEs is accelerated by high tissue levels of glucose³⁸. AGEs signal through RAGE (receptor for AGEs) and other receptors to produce various deleterious effects on chondrocytes including inflammation and cytokine-mediated catabolism^104–106 and have been postulated to play a role in end-organ damage in DM³⁸. Further, AGE mediated cross-linking of collagen can alter a tissue’s biomechanical properties as shown for cartilage and tendon^35–37. Cross-linking by AGEs may also inhibit ECM turnover by restricting access to proteolytic sites¹⁰⁶. On the other hand, a recent study in dogs suggests that an artificial increase in AGE levels alone using repeated ribose injections did not accelerate OA in a mild injury model¹⁰⁷, but we know little about the effect of AGEs in the context of the diabetic milieu. Thus, whether or how AGE’s play a significant role in OA remains unclear.

In the polyol pathway, aldose reductase converts glucose to sorbitol and galactose to galactitol. This mechanism is activated in DM with excess polyols leading to cellular osmotic stress¹⁰⁸. Although not yet directly linked to OA, there is some evidence that this pathway is activated in DM in intervertebral disc cartilage and enhances matrix catabolism via p38 MAPK activation¹⁰⁹.

Although not covered in detail in this review, additional DM-relevant pathways have been proposed. For example, there is considerable evidence that adipokines may induce inflammation and have adverse effects on cartilage^75,110 and tissue healing¹¹¹. Because altered adipokine levels are seen in obesity in both the absence and presence of diabetes¹¹², the contribution of adipokines to OA in obese patients with DM will require further study. Alteration in angiogenesis, autophagy, and apoptosis are also associated with end-organ damage in OA^58,113–116. Insulin receptors are present on chondrocytes¹⁰⁰, and thus, excess insulin as seen in T2DM patients may also damage cartilage. In one study, PPARγ downregulation was shown to occur in articular chondrocytes exposed to high glucose media, but methodologic challenges warrant confirmation of this finding⁷³. Whether one or more pathways are involved, which pathway is most relevant, and how molecular mediators intersect multiple pathways will require additional study.

Conclusions

In summary, increasing evidence from the clinic and the laboratory supports an adverse effect of DM on the development, severity, and therapeutic outcomes for OA. The etiology and clinical manifestations of OA are complex, and currently we know little about how the multiple mechanisms altered in DM may affect OA that originates from different causes. Further, the clinical impact of DM on OA may be underestimated by the high prevalence of undiagnosed diabetes³. A deeper understanding of OA in the setting of DM could result in significant improvements in clinical care. For example, reducing OA-related joint pain may allow DM patients to perform the exercise necessary for cardiovascular health. A full appreciation of these disease interactions may also reduce the increased medical costs associated with arthroplasty and other surgeries in patients with comorbid DM^20,21,32. To understand the mechanisms through which DM contributes to OA, further work is clearly necessary. Future studies of DM-influenced mechanisms may shed light on the general mechanisms of OA pathogenesis and result in more specific and effective therapies for all OA patients.

As there is no perfect animal model that entirely recapitulates either T1DM or T2DM, compromises are likely to be required with any one model. Ideally the selected model(s) should be well characterized in terms of the following:

1. Level and consistency of hyperglycemia. Chemically induced models of T1DM have been known to lose their hyperglycemic state over time due to β-cell regeneration. Hyperglycemia drops among some monogenic T2DM models.

2. Age at onset of diabetes. T2DM and OA are more common amongst older humans, therefore the use of older animals in these research studies should be considered and the lifespan of the model should be sufficient to allow development of OA.

3. Sex differences. There is a gender bias in diabetes severity and age of onset In many rodent models of diabetes.

4. Appropriateness of control group. Different background strains may have greatly different susceptibility to obesity and change in blood glucose.

5. Comorbid conditions. It may be desirable to have present comorbid conditions to answer the specific research questions. Scientists should be aware that not all DM models have been fully characterized for comorbid conditions.