Introduction
The numbers of patients with osteoporosis or type 2 diabetes mellitus (T2DM) are increasing in aging and westernized societies. Both disorders predispose the elderly to disabling conditions and eventually raise mortality. Osteoporosis leads to fractures and T2DM leads to vascular complications. Although osteoporosis and T2DM have been viewed as separate diseases, accumulating evidence indicates that similar pathophysiological mechanisms underlie both of them.
Interaction between bone metabolism and glucose and fat metabolism through osteocalcin and Wnt signaling
Osteocalcin (OC), a protein secreted by osteoblasts into the circulation, has several hormone-like features. Recent animal studies have shown that OC action is involved in the control of bone metabolism, as well as glucose metabolism and fat mass. In studies of OC-deficient mice, Lee et al. showed that OC functions as a hormone that improves glucose metabolism and reduces fat mass. Administration of recombinant uncarboxylated OC to wild-type mice fed a high-fat-diet-regulated gene expression in pancreatic β-cells and adipocytes (including adiponectin expression) and prevented development of metabolic diseases, obesity and hyperglycemia. The results of two recent experiments suggest the existence of a bone–pancreas endocrine loop in which bone resorption-dependent activation of OC occurs by way of the insulin receptor: insulin signaling in osteoblasts decreases osteoprotegerin expression in a Forkhead box O1-dependent manner. This promotes bone resorption and acidification of the bone extracellular matrix, which in turn promotes OC decarboxylation and, as a result, β-cell proliferation, insulin secretion and insulin sensitivity. Several clinical studies have also confirmed the relationship between OC, and glucose and fat metabolism in humans. We have shown previously that, in T2DM patients, the serum concentration of OC was negatively correlated with that of plasma glucose and with atherosclerosis parameters. We have also found that, in a population of patients evaluated for treatment of DM and osteoporosis, the serum concentration of undercarboxylated OC is correlated negatively with plasma glucose concentration and fat mass, and positively with serum adiponectin concentration. These experimental and clinical findings suggest that bone metabolism and glucose or fat metabolism are etiologically related through the action of OC (Figure 1). However, high glucose concentrations impair OC expression and secretion from the osteoblastic cells, and treatment of hyperglycemia in T2DM patients has increased serum OC concentrations in patients. Thus, in cross-sectional studies in humans, the negative correlation between the concentration of serum OC and plasma glucose may be a result of hyperglycemia and not a cause for the suppression of insulin secretion and sensitivity.
Wnt signaling is also thought to be a common pathogenic pathway, important in osteoporosis and DM. A missense mutation in low-density lipoprotein receptor-related protein 6, the co-receptor for the Wnt protein, has been genetically linked to osteoporosis, as well as DM, hyperlipidemia and coronary artery disease. In addition, several studies have documented that T-cell-specific transcription factor-4, the partner of β-catenin in the canonical Wnt-signaling pathway, is the strongest T2DM susceptibility gene. Manolagas and Almeida, and other researchers have suggested that antagonism of Wnt signaling by oxidative stress diverts β-catenin from T-cell-specific transcription factor-mediated transcription to Forkhead box O-mediated transcription and contributes to the development of osteoporosis, as well as to insulin resistance and hyperlipidemia. In osteoblasts and their progenitors, inactivation of Forkhead box O1 by insulin signaling leads to decreased osteoblastogenesis by increasing reactive oxygen species abundance and oxidative stress, a mechanism implicated in the DM-related bone fragility, as discussed in a later section.
Involvement of fat tissue in bone metabolism through adiponectin
The positive correlation between bone mineral density (BMD) and fat mass, including the mass of visceral adipose tissue, suggests that there is a relationship between body fat and bone mass. However, the correlation between BMD and fat mass becomes negative when the mechanical loading effect of body weight on bone mass is considered, suggesting that body weight itself, but not fat mass, is related to increased BMD. Adiponectin is specifically and highly expressed in visceral, subcutaneous and bone marrow fat depots. It is also abundant in plasma and may have important roles in the regulation of energy homeostasis and insulin sensitivity. We and others have shown that osteoblasts have an adiponectin receptor and that the proliferation, differentiation and mineralization of osteoblastic cells is enhanced by adiponectin. We have also found that serum adiponectin concentration is associated with BMD, bone turnover and the presence of vertebral fractures (VFs) in T2DM patients. In addition, we have found that, in poorly controlled T2DM patients, the serum adiponectin concentration before treatment predicts the subsequent increase in serum OC upon glycemic control. These findings suggest that adiponectin and OC, secreted from fat tissue and osteoblasts, respectively, are involved in the interplay between the metabolisms of glucose or fat and bone (Figure 2).
BMD and fracture risk in T2DM
As a large proportion of the elderly population is affected by T2DM and osteoporosis, many clinical studies have investigated the association between the two. Although BMD is considered a gold standard for evaluating fracture risk in non-DM osteoporosis, accumulating evidence shows that T2DM patients have high fracture rates, despite the absence of decreased BMD. A recent meta-analysis showed that T2DM patients had higher hip BMD than non-DM controls (z-score=0.27), despite an increased risk of hip fracture (1.4-fold). These data suggest that BMD values may not reflect bone fragility in T2DM. Another meta-analysis showed that hip fracture risk is increased 1.7-fold in T2DM patients. In contrast, recent data have indicated that the femoral neck BMD T- and fracture risk assessment-scores are associated with the risk of hip and non-spinal fractures among older adults with T2DM. These data suggest that BMD measurements are meaningful in T2DM. However, the same study has also shown that fracture risk was higher in T2DM subjects than in non-DM controls with a given T-score and age or with a given fracture risk assessment score. These findings suggest that femoral neck BMD and fracture risk assessment scores may be partially effective in assessing hip and non-spinal fracture risk in T2DM patients, although they may not be as sensitive as in non-DM subjects.
In contrast, little is known about the risk of VF and its association with BMD. To explore this association, we examined Japanese T2DM patients and non-DM controls. We found that, after adjustment for age, body mass index and lumbar BMD by logistic regression analysis, the presence of T2DM was an independent risk factor for prevalent radiographic VFs in women (odds ratio (OR)=1.9), as well as men (OR=4.7). Another logistic regression analysis showed that BMD at any site was not significantly associated with the presence of VFs in T2DM patients, in contrast to the significant association in non-DM controls. We examined the distribution of lumbar BMD as a function of age in non-DM and T2DM women (Figure 3). Non-DM subjects with VFs were clustered in the region of higher age and lower lumbar BMD. In contrast, there was no association with age or lumbar BMD for T2DM subjects with VFs. In addition, although quantitative ultrasound is considered useful in evaluating bone quality, we found that calcaneal quantitative ultrasound did not discriminate between T2DM patients with prevalent VFs and those without VFs. Thus, T2DM patients may have an increased risk of VFs that is independent of BMD or quantitative ultrasound values. These findings suggest that bone quality rather than bone mass may define bone fragility that causes hip fractures and VFs in T2DM.
Surrogate markers for assessing fracture risk in T2DM
As BMD is not sensitive enough to assess the risk of osteoporotic fractures in T2DM, it is important to explore the etiology of DM-related bone fragility to identify diagnostic markers that supplement the partial effectiveness of BMD.
Pentosidine.
Advanced glycation end products (AGEs) are formed by the Maillard reaction, a collection of sequential nonenzymatic chemical glycoxidation reactions of protein amino groups. With normal aging, AGEs accumulate in various tissues, including the kidney and brain, and in coronary artery atherosclerotic plaques. However, hyperglycemia results in accelerated AGE formation, suggesting that AGEs have a pivotal role in the development of complications in DM patients. In addition, previous studies have revealed that AGEs accumulate in bone tissue, and the receptor for AGE (RAGE) is expressed in human bone-derived cells, suggesting that AGEs are associated with DM-related bone fragility.
Several studies have shown that AGEs have a negative impact on bone. AGEs inhibit the synthesis of type 1 collagen and OC, as well as the formation of mature bone nodules in osteoblasts. We previously demonstrated that a combination of high glucose and AGE levels in vitro additively or synergistically inhibits the mineralization of osteoblastic cells through a glucose-induced increase in expression of RAGE. These findings suggest that accumulation of AGEs in bone may cause osteoblastic dysfunction. AGEs are also known to increase osteoclast activity. Previous in vitro and in vivo experiments have indicated that the number of resorption pits is increased when osteoclasts are cultured on AGE-modified dentin slices; moreover, AGE-modified bone particles implanted subcutaneously in rats are resorbed to a much greater extent than non-AGE-modified bone particles. In addition, RAGE knockout mice have fewer osteoclasts and a significantly higher bone mass than wild-type mice. Taken together, these findings indicate that AGE accumulation inhibits the differentiation and mineralization of osteoblasts and enhances the activity of osteoclasts, possibly leading to uncoupled bone turnover and resultant bone fragility.
Accumulation of AGE in bone is also negatively associated with the material properties of bone. Collagen crosslinks have critical roles in the determination of bone strength. AGE-type crosslinks, which are formed spontaneously by non-enzymatic glycation and oxidation reactions, are thought to be associated with the brittleness of collagen fibers, whereas physiological crosslinks (enzymatic crosslinks) strengthen collagen fibers and thus increase bone strength. For example, spontaneously diabetic WBN/Kob rats have decreased abundance of enzymatic crosslinks and increased abundance of AGE-type crosslinks; as a result, they show a deterioration in bone strength, despite the lack of BMD reduction.
Pentosidine, one of the few AGEs characterized to date, is well known and chemically well defined. As the formation of pentosidine requires both glycation and oxidation, serum pentosidine concentration is considered a useful marker for glycoxidation. Several studies have revealed that pentosidine content in cortical or trabecular bone from vertebrae or femur is negatively associated with mechanical properties, and that the pentosidine content of cortical and trabecular bone derived from patients with femoral neck fracture is higher than that of age-matched controls. However, because invasive procedures such as bone biopsy are necessary for preparing specimens, assessment of bone pentosidine content is not easily done clinically. A recent study revealed that the plasma content of pentosidine is significantly and linearly correlated with that in the cortical bone. These data suggest that serum pentosidine concentration could be used as a surrogate marker for pentosidine content in bone and could indicate bone strength. We previously showed that serum pentosidine concentration was associated with the prevalence of VFs in postmenopausal T2DM women (OR=2.50 per s.d. increase). This association was independent of BMD, suggesting that it reflects bone quality rather than bone mass. In addition, an observational cohort study has shown that urine pentosidine concentration is associated with increased clinical fracture incidence in subjects with DM (relative hazard=1.42 per 1 s.d. increase in log pentosidine). Urinary pentosidine concentration at baseline also predicts future VFs in non-DM patients treated with bisphosphonate. Therefore, serum and urine pentosidine concentrations might be useful markers for assessing fracture risk in both T2DM and non-DM patients.
Endogenous secretory RAGE.
RAGE, a member of the immunoglobulin superfamily of cell surface receptors, interacts with multiple ligands, including AGEs. Crossbreeding of transgenic mice overexpressing human RAGE in their vascular cells with transgenic mice that develop insulin-dependent DM shortly after birth produced offspring that developed more progressive histological changes associated with DM nephropathy than did controls. This result confirmed that RAGE expression stimulates the development of DM complications. Endogenous secretory RAGE (esRAGE), a splice variant of one of the naturally occurring secretory forms of RAGE, contains the extracellular domains of the RAGE receptor, but lacks the transmembrane and cytoplasmic domains. Secreted esRAGE is thought to act as a decoy receptor that binds and neutralizes AGEs, reducing the activity of intercellular signal pathways that use RAGE. Indeed, administration of a genetically engineered soluble form of murine RAGE suppresses the development of diabetic atherosclerosis in a dose-dependent manner in streptozotocin-treated apoE-null DM mice. These findings suggest that enhanced RAGE activity is also linked to reduced bone strength in DM patients. Given the neutralizing activity of esRAGE, it is possible that the ratio of the serum concentrations of esRAGE and AGEs is more strongly linked to clinical bone problems, such as fractures, than is either parameter alone.
We found previously that the esRAGE/pentosidine ratio was significantly lower in T2DM patients with VFs than in patients without VFs. Multivariate logistic regression analysis, adjusted for age, height, weight, HbA1c, serum creatinine, DM duration, therapeutic agents, DM complications, osteoporotic risk factors and lumbar BMD, identified serum esRAGE concentration and esRAGE/pentosidine ratio as factors associated with the presence of VFs. The association was independent of BMD in men (OR=0.46 and OR=0.34 per s.d. increase, respectively) and in women (OR=0.32 and OR=0.14 per s.d. increase, respectively). These results show that serum esRAGE concentration and the esRAGE/pentosidine ratio are more useful than BMD for assessing the risk of VFs in T2DM patients.
Assessment and prevention of fracture risk of T2DM patients in clinical practice
In the face of the partial effectiveness of BMD in assessing fracture risk in T2DM, the major clinical problems in preventing fractures in daily practice are how to assess the risk and when to start therapies. Although the markers related to AGEs are potential candidates for fracture-risk assessment, it is unclear whether they can be used to predict the occurrence of new fractures in T2DM patients.
A large study of the incidence of VFs in postmenopausal osteoporosis showed that, during a several-year study period, patients with VFs were more likely than those without, to suffer from new VFs and hip fractures. This finding was independent of BMD. Thus, the presence of prevalent VFs might be useful for assessing bone quality in individual patients. Patient histories of existing non-VFs are also established risk factors for additional fractures. We found that 38 and 31% of men and women, respectively, with T2DM had prevalent VFs, as determined by X-ray. Sixteen percent of each group had histories of previous non-VFs. Thus, if patients with T2DM undergo spinal X-ray examination and are questioned about their fracture histories, it is likely that bone fragility will be identified in about half of the patients; those patients will need treatment for osteoporosis to prevent fractures. These procedures are simple, and we recommend that they be used by all physicians engaged in treating T2DM.
A recent study revealed that hip fracture-risk was higher, for a given T-score, in T2DM patients than in non-T2DM subjects. Therefore, when prevalent fractures are absent, osteopenic patients with T2DM and with BMD T-scores between –1.0 and –2.5 should undergo treatment to prevent fractures.
Large-scale studies of Caucasian and Chinese populations have shown that hip and other osteoporotic fracture risks are increased in patients with long-term or established T2DM. Fracture risks were not increased in those with impaired glucose tolerance or newly diagnosed T2DM. Thus, the severity and duration of T2DM should be considered when patients are started on drugs for osteoporosis.
It is unclear whether drugs for osteoporosis, such as bisphosphonates or selective estrogen-receptor modulators, are effective in preventing fractures in T2DM. However, recent studies have shown that the fracture-preventive potential of these drugs is similar in subjects with or without T2DM, suggesting that T2DM does not affect drug efficacy. Therefore, patients both with and without T2DM should receive the same treatments for osteoporosis.
Summary
T2DM populations are expanding in every country. It is therefore problematic that BMD testing is not sensitive enough to assess fracture risk—especially prevalent radiographic VF risk—in T2DM patients. If doctors use BMD values alone to diagnose osteoporosis, T2DM patients who are at risk of fracture may not be included in fracture-prevention programs. Practitioners should be aware of the importance of evaluating prior VFs and fracture histories by spinal X-ray and interview, as well as by measuring BMD. Combining all of these procedures would broaden the treatment of osteoporosis to include the T2DM population.
In addition, further studies are needed to clarify whether surrogate biochemical markers such as serum concentrations of pentosidine and esRAGE are useful for predicting the occurrence of new fractures in T2DM patients, and whether the sensitivity and specificity of these markers are similar to those of BMD measurement in osteoporotic non-DM patients.