Osteoporosis, characterized by reduced bone strength and an increased risk for low-trauma fractures, increases dramatically with age.1 About 40% of women in developed countries will experience an osteoporosis-related fracture in the course of their lifetime, with men experiencing approximately one-third to one-half the risk of women.2, 3
Diabetes is also more frequent in the elderly, and therefore frequently coexists with osteoporosis.4, 5 Furthermore, there has also been a global increase in the prevalence of obesity, with obesity-related diabetes currently affecting over 366 million adults worldwide and projections that this will reach 552 million by 2030.6 Type 1 diabetes (T1D), and more recently type 2 diabetes (T2D), have been associated with increased fracture risk,7–9 despite the observation that areal bone mineral density (BMD) from DXA is higher in individuals with T2D compared with nondiabetic individuals.10–13
Given the growing body of research showing that T2D is an independent risk factor for fracture,14, 15 it is timely to review the proposed underlying pathophysiologic mechanisms and ask how this might be incorporated into risk assessment and management algorithms. This perspective provides a brief summary of the available evidence on how T2D impacts bone metabolism and fracture risk.
Biologic Mechanisms of T2D on Bone
The evidence that fracture risk is increased in T2D, paradoxically, despite normal areal BMD,13, 16 has led to the hypothesis that there are diabetes-associated alterations in skeletal properties. These alterations, which potentially include abnormalities in material, dynamic, and microarchitectural skeletal properties, may contribute to the increased fracture risk in T2D.
High glucose levels in T2D lead to accumulation of advanced glycosylation end-products (AGEs) in the organic bone matrix by a process known as nonenzymatic glycation (the Maillard reaction).17–20 Hemoglobin A1c is a common example of an early-stage glycation product. In contrast to normal enzymatic cross-linking in collagen (with pyridinoline, for example) which gives bone its toughness and scaffolding properties, AGE cross-links lead to biomechanically more brittle bone that has lost its toughness and is less able to deform before fracturing.21 Pentosidine, the best studied AGE,17, 22 when measured in urine, was associated with a 42% increase in clinical fracture incidence in T2D23 and when measured in serum was increased in diabetics who had vertebral fractures.24 Bone pentosidine in nondiabetic patients was greater in hip fracture patients than in controls25, 26 and predicted the whole bone fracture properties of human vertebrae, independently of BMD.27 These data, although limited, suggest that AGE-distorted collagen likely may render the bone more fragile in T2D regardless of BMD.
Increased AGEs may weaken bone by decreasing bone formation. There is evidence suggesting that AGEs interfere with normal osteoblast development,28 function29 and attachment to the collagen matrix.30 Moreover, low bone formation also works in the opposite direction to further increase AGEs, as, for example, with high bisphosphonate dosages.21 Numerous data suggest that skeletal dynamics are reduced in T2D, with a disproportionate reduction in bone formation.31, 32 Biochemical markers of bone formation have generally been shown to be reduced,31–33 although it still has not been established definitively that T2D is characterized by low bone formation.34–36 Few histomorphometric data exist; one study reported lower bone formation in T2D on biopsy, but the numbers were very small.37 Recent data suggest that sclerostin, the osteocyte product that inhibits the anabolic Wnt pathway, is increased in T2D,38 although a clear relationship between sclerostin levels and biochemical markers of bone formation was not confirmed. Interestingly, the known inverse relationship between PTH and sclerostin was not observed,38 perhaps suggesting that there may be an absence in T2D of the usual inhibitory effect of PTH on sclerostin production.
Recent studies suggest that in T2D, trabecular bone mass and structure are intact and perhaps even enhanced, whereas the cortical compartment is preferentially compromised.39, 40 This is noteworthy because: (a) the cortex makes up 80% of the skeleton, (b) cortical bone is present primarily at nonvertebral sites, and (c) in T2D, most of the fractures occur at sites that are rich in cortical bone.9, 10, 12, 16, 41, 42 Increased cortical porosity in particular has been reported at the radius in female diabetics who have fractured, as measured by intracortical pore volume fraction via high-resolution peripheral quantitative computed tomography (HR-pQCT) (Fig. 1).40 Although endosteal cortical remnants can be mistakenly interpreted as trabeculae,43 true increases in cortical porosity could be an important cause of increased fracture risk in T2D, because it reduces bone strength yet is undetectable by DXA.44
The mechanisms for increased fracture risk in T2D, therefore, appear to include both material and structural abnormalities. These abnormalities could well be intertwined, because abnormal collagen is also likely to be abnormally mineralized. An overly glycated collagen matrix, confounded by a low turnover state, in the setting of subtle cortical abnormalities, may lead to compromised biomechanical competence. Our current clinical mainstay, DXA, cannot ascertain these properties of bone.45 Future studies will be helpful to gain insights into the pathogenesis of fractures in T2D at a mechanistic level.
Epidemiology of T2D, BMD, and Fracture
There is substantial evidence that older adults with T2D have a higher risk of hip fractures.9, 16 A recent meta-analysis of 12 studies reported a relative risk of 1.7 (95% CI: 1.3 to 2.2) for T2D and hip fracture.9 Risk was elevated in men and women, and there was no evidence of regional differences between Europe and the United States. Less information is available for other regions of the world. A study in Taiwan reported increased hip fracture risk with T2D.46 The risk of all clinical fractures also appears to be increased with T2D. The most recent meta-analysis reported a summary RR of 1.2 (95% CI: 1.0 to 1.5),9 and subsequent studies have reported similar results.10, 47, 48
Surprisingly, given this increased fracture risk, BMD is generally higher in those with T2D.16, 49 In a meta-analysis, Vestergaard reported an increased Z-score of 0.41 at the spine and 0.27 at the hip associated with T2D.16 Higher body mass index (BMI) was identified as an important contributing factor. These findings were primarily based on results from Europe and North America. Results from studies in East Asia have been less consistent.42, 50–52 In this region, T2D is not as strongly associated with overweight and obesity. In a study of older women in China that stratified on BMI, diabetes was associated with higher BMD in the overweight group but with lower BMD in the normal weight group.53
BMD, central to fracture risk prediction in older adults, is also predictive of fractures in those with diabetes. However, those with diabetes tend to fracture at a lower BMD.14 This may be partly because of more frequent trauma, because diabetes is associated with an increased frequency of falls.54, 55 Factors that may contribute to poorer balance in those with diabetes include peripheral neuropathy, reduced vision, nephropathy, and lower muscle strength.56–58 Studies in older adults suggest that diabetes reduces muscle quality, assessed as strength for a given muscle mass.59 However, in studies of diabetes and fracture that controlled for fall frequency, diabetes remained independently associated with higher fracture risk, indicating that increased falls do not fully account for this higher risk10, 41, 60
The epidemiological evidence thus points to a deficit in diabetic bone that is not fully captured by cross-sectional DXA measurements. One contributing factor may be more rapid bone loss with T2D, reported in older white women61–63 and in older men.64 Greater weight loss in those with T2D accounts for some, but not all, of this increased rate of bone loss.63 Bone geometry has also been investigated as a possible reason for reduced bone strength in diabetes. Strength-to-load ratios (QCT) at the spine and femoral neck were not improved in older adults with diabetes although areal BMD (DXA) was higher.13 In a study of older men, volumetric BMD (pQCT) was higher, but bone area was smaller at the distal radius and tibia.39 Smaller cross-sectional area suggests that stimulation of periosteal apposition, normally observed with greater loading,65 may be reduced in T2D.
Established risk factors for fracture identified in broader populations also contribute to fracture risk in those with T2D.7, 48, 66 Considering diabetes-specific risk factors, thiazolidinedione (TZD) use is associated with bone loss and increased fracture risk, particularly in women.67 However, TZD use probably does not account for the increased risk of fracture observed with diabetes, because most studies included substantial follow-up time before wider use of these medications. Insulin use is associated with increased fracture risk,48, 60, 68, 69 but this is most likely a marker of more severe diabetes rather than the result of a direct effect on bone strength. Longer duration of diabetes appears to increase fracture risk.11, 47 Macrovascular and microvascular complications related to diabetes, particularly the presence of multiple complications, may be associated with fracture risk but current studies are limited.7, 48, 70 The effects of glycemic control on fracture risk and bone density are not clearly understood. In the ACCORD randomized trial, there was no difference in the rate of fractures between the intensive and standard glycemic control groups, but TZD use was higher in the intensive control group.71 Observational studies have provided limited and inconsistent results.7, 16, 49
T2D and Fracture Risk Assessment: The Case for Inclusion in FRAX
The WHO fracture risk assessment tool (FRAX) is a computer-based algorithm (http://www.shef.ac.uk/FRAX) primarily intended for use in primary care.72, 73 FRAX calculates fracture probability from easily obtained clinical risk factors: age, sex, body mass index, prolonged use of glucocorticoids, current smoking, alcohol intake of three or more units per day, a parental history of hip fracture, secondary osteoporosis, rheumatoid arthritis, prior fragility fracture, and (optionally) femoral neck BMD or T-score. The output, estimated probabilities for major osteoporotic fracture (hip, clinical spine, humerus, or forearm) and hip fracture over 10 years, has been shown to improve fracture prediction over T-score alone.74
Diabetes is not a primary entry variable in the current FRAX construction. T1D is indirectly considered in FRAX as one of the secondary causes of osteoporosis, increasing the calculated fracture probability when BMD is not known but not when BMD is included in the risk calculation. At the time that FRAX was released in 2008, the effect of secondary causes on fracture risk was conservatively assumed to be mediated by their effect to decrease bone density. Rheumatoid arthritis was the only secondary cause of osteoporosis considered independent of BMD in the FRAX algorithm, with uncertainty as to whether this was true for other disorders.75
Two recent reports have shown that for a given FRAX probability or T-score and age, the risk of fracture among individuals with diabetes is higher than the risk in nondiabetics. 14, 15 Schwartz et al.14 analyzed data from three major prospective observational studies totaling 9,449 women (770 with T2D) and 7,436 men (1,199 with T2D). Femoral neck T-score and FRAX probability predicted hip and nonspine fractures in women and men with diabetes. However, for a given T-score and age or for a given FRAX probability, those with diabetes had a higher fracture risk than those without diabetes. Concordant findings were reported in a subsequent study of incident fractures in 3,518 men and women with diagnosed diabetes (predominantly T2D) and 36,085 without diabetes from a large clinical cohort in Manitoba, Canada,15 Diabetes was a significant predictor of subsequent major osteoporotic fracture (adjusted hazard ratio [aHR] = 1.61, 95% confidence interval [CI] 1.42 to 1.83) after controlling for age, sex, medication use, and FRAX risk factors, including BMD. Diabetes was also associated with significantly higher risk for hip fractures (aHR 6.27, 95% CI 3.62 to 10.87 in those aged <65 years, aHR 2.22, 95% CI 1.71 to 2.90 in those >>65 years). FRAX underestimated observed major osteoporotic and hip fracture risk in diabetics even when adjusted for competing mortality (Fig. 2). Together, these studies provide compelling evidence that FRAX underestimates the risk of osteoporotic fractures in individuals with T2D.
A complementary question is whether diabetes identifies a risk amenable to therapeutic intervention—“reversibility of risk.”72, 73, 76 Randomized clinical trials powered to show a reduction in fracture risk in individuals with diabetes and elevated fracture risk have not been performed and might well be unethical. Post hoc analyses of pivotal clinical trials would have low power to show a beneficial effect in the small subgroups of individuals with diabetes. However, in a nationwide cohort study from Denmark, users of antiresorptive drugs (n = 103,562) were compared with age-matched and gender-matched controls from the general population (n = 310,683).77 Patients on bisphosphonates and raloxifene had a higher risk of hip, spine, and forearm fractures, reflecting the underlying reason for their prescription. No difference was observed in the effects of treatment between patients with diabetes and nondiabetic controls, or between patients with T1D and T2D. The authors concluded that diabetes did not affect the fracture-preventive potential of bisphosphonates or raloxifene.
T2D and Fracture Risk Assessment: The Case Against Inclusion in FRAX
T2D meets many but not all criteria for inclusion in FRAX as a primary entry variable. As noted above, there is compelling evidence that T2D is a risk factor for fracture and that the risk is independent of BMD (and FRAX). Moreover, there is a case for “reversibility” of risk with pharmacological intervention. Lacking, however, is the international dimension, and studies to date have been confined to North America.14, 15 In the case of fracture risk, the available information suggests a heterogeneous association with fracture risk.16 It is important to determine the international validity of any candidate risk factor.78 On the assumption that this hurdle can be readily overcome, what are the remaining impediments to the incorporation of diabetes into FRAX?
The output of FRAX is the probability of fracture. This metric differs from other risk engines79, 80 in that probability integrates the risk of fracture and the risk of death.78 A FRAX variable strongly affected in this way is smoking, which carries a moderate risk of fracture81 but has a minor effect on fracture probability because smoking increases the risk of death as well as the risk of fracture. T2D is also associated with approximately a twofold increase in mortality15 and this will compete with the fracture hazard. The consideration of fracture and death hazards is, however, more complex.
One of the strengths of FRAX is that it is derived from the primary data in population-based cohorts from around the world. The use of primary data for the model construct permits the determination of the predictive importance in a multivariable context of each of the risk factors, as well as interactions between risk factors, and in this way, optimizes the accuracy with which fracture probability can be computed.74 This FRAX matrix of covariates is used both for the fracture hazard and for the death hazard. If a candidate risk factor were totally independent of the other risk factors, then incorporation into a FRAX model would be relatively straightforward. We already know this not to be the case for T2D in that there is a significant interaction with age such that the hazard ratio for fracture appears to be higher in younger individuals.15
For all these reasons, the inclusion of T2D as an input variable to FRAX is premature. More research is required not only on the risk of fracture associated with T2D but also its dependence on the other risk variables in FRAX and their independent effect on the death hazard. This demands the collection of new population cohorts that include such information, as well as the other FRAX variables in sufficient numbers and with wide geographical representation.
Recommendations and Conclusions
Skeletal health in T2D continues to pose an unresolved paradox of increased fracture risk with higher bone density. Continuing investigation into the underlying mechanisms promises to advance our understanding of osteoporosis and diabetes, potentially providing new markers for identification of those at higher risk, as well as new approaches to prevention. In current clinical practice, because BMD is central to fracture prediction, a consequence of this paradox is a lack of suitable methods, including FRAX, to predict fracture risk in older adults with T2D. An improved ability to identify diabetic patients at higher risk of fracture promises to translate into improved fracture prevention, because patients with diabetes appear to benefit from antiresorptive treatments for osteoporosis in the same way as nondiabetic patients.
The available research information can inform the clinician how to temper clinical judgment on the existing output of the FRAX models. This can be exercised in several ways, but should be undertaken in the context of FRAX-based intervention thresholds (eg, 20% probability of a major fracture in North America). Thus, in a patient with diabetes below but close to a FRAX-based intervention threshold (eg, 18% probability of a major fracture), the physician may recommend treatment. The upward revision of a patient with a fracture probability of >20% has little clinical utility in making a decision to treat. A second option is to use the rheumatoid arthritis channel in patients with diabetes. A third measure might be to model the potential impact of diabetes on FRAX in much the same way that has been done when considering the effects of glucocorticoid dose or lumbar spine BMD on fracture probability.82, 83
There is growing awareness of the inadequacy of current methods to predict fracture risk in T2D patients, who constitute a substantial portion of the older population. This perspective aims to advance discussion regarding how to most effectively improve our ability to predict fractures in this population. The option of adding diabetes to the FRAX algorithm is appealing but requires additional data from large population-based cohorts, and consideration of how to accommodate differences between T1D (reduced BMD with greater relative risk) and T2D (higher BMD with greater prevalence).9, 16, 68 In any case, the need for an improved method for identification of fracture in older adults with T2D remains an important priority for osteoporosis research.
WDL: Speaker fees from Amgen. Research support from Novartis, Amgen, Genzyme. Advisory boards for Novartis and Amgen. MRR: Research support from NPS Pharmaceuticals. AVS: Research support from GlaxoSmithKline. JAK: Nothing to declare for FRAX and the context of this paper, but ad hoc consultancies for: Industry: Abiogen, Italy; Amgen, USA, Switzerland, and Belgium; Bayer, Germany; Besins-Iscovesco, France; Biosintetica, Brazil; Boehringer Ingelheim, UK; Celtrix, USA; D3A, France; Gador, Argentina; General Electric, USA; GSK, UK, USA; Hologic, Belgium and USA; Kissei, Japan; Leiras, Finland; Leo Pharma, Denmark; Lilly, USA, Canada, Japan, Australia, and UK; Merck Research Labs, USA; Merlin Ventures, UK; MRL, China; Novartis, Switzerland and USA; Novo Nordisk, Denmark; Nycomed, Norway; Ono, UK, and Japan; Organon, Holland; Parke-Davis, USA; Pfizer USA; Pharmexa, Denmark; Procter and Gamble, UK, USA; ProStrakan, UK; Roche, Germany, Australia, Switzerland, and USA; Rotta Research, Italy; Sanofi-Aventis, USA; Schering, Germany and Finland; Servier, France and UK; Shire, UK; Solvay, France, and Germany; Strathmann, Germany; Tethys, USA; Teijin, Japan; Teva, Israel; UBS, Belgium; Unigene, USA; Warburg-Pincus, UK; Warner-Lambert, USA; Wyeth, USA. Governmental and NGOs: National Institute for Health and Clinical Excellence (NICE), UK; International Osteoporosis Foundation; National Osteoporosis Guideline Group (NOGG), UK; INSERM, France; Ministry of Public Health, China; Ministry of Health, Australia; National Osteoporosis Society, UK; WHO.
Authors' roles: All listed authors contributed to the conception and interpretation of data; participated in drafting the manuscript; and approved the final version of the submitted manuscript.