Diabetes and Fractures — An overshadowed association

Type 2 Diabetes

The data on fracture risk in type 2 diabetes are not entirely consistent. Several cohort studies [3–10] and a case-control study [11] have found an increased fracture risk in type 2 diabetes compared to the general population (see Table 1). Whereas, a few studies report a decreased [12, 13] or similar fracture [14, 15] risk in type 2 diabetes compared to the general population.

Most studies that report a positive association have limited their analysis to and have found that patients with type 2 diabetes have a higher risk of hip fractures. However, studies that evaluated all types of fractures [5, 8, 11] still found hip fractures to contribute the most to the fracture risk seen in type 2 diabetes. Wrist [9] and foot [5, 10] fractures also appear to be increased in type 2 diabetes but only in patients who are on treatment with either oral anti-diabetic agents (OAD)[9] or insulin[5, 10]. The risk for vertebral fractures is similar to non-diabetics [6, 10, 11]. The risk of hip fractures appears to be slightly higher in men compared to women with type 2 diabetes [2, 16] and slightly higher in black compared to white women with type 2 diabetes [10]. In a large prospective study of nurses aged 34–59 years [3], the incidence of hip fractures in females with type 2 diabetes is reported at 153 per 100,000 patients compared to 63 per 100,000. The incidence increases to 209/100,000 for those type 2 diabetic females treated with insulin.

A few studies have reported a similar or lower fracture risk in type 2 diabetes compared to the general population (see Table 1) [14] [13] [17] [12]. Of these, two [14, 15] showed no difference in fracture risk and both only adjusted for age. One was a small cross-sectional study [14] and the other [15], when further adjusting for calcaneal stiffness, actually found a significantly increased risk of hip fractures in type 2 diabetics compared to non-diabetic controls.

Of those studies that reported a lower fracture risk in type 2 diabetes compared to the general population two of them were unable to show this decrease to reach statistical significance, either on initial report [17] or when reanalyzed in a meta-analysis [2, 12]. The third study [13] included a large percentage (42%) of diet controlled diabetics. Interestingly, of the positive studies, those that showed only a minimal increase in fracture risk involved diet controlled [4] or early onset type 2 diabetics[8]. Perhaps, early on in the course of diabetes a higher bone mineral density (BMD) [13, 18] [5, 19] protects from fractures. However, as the diabetes progresses, certain factors such as hyperglycemia [9, 19] may impair bone quality, and this, then, overcomes the protective effects of a higher BMD and results in a higher risk of fractures. This hypothesis is supported by a study done by de Liefde et al [9] that showed a higher BMD in subjects with impaired glucose tolerance, who also had a lower fracture risk HR 0.8 (0.63–1.00), whereas treated type 2 diabetics had a higher fracture risk HR 1.69 (1.16–2.46) despite an equally high BMD.

Finally, a recent meta-analysis [2] of 12 positive and negative studies [4–7, 11, 12, 17, 20–23] found that the overall relative risk of hip fractures in type 2 diabetics significantly increased with a relative risk (RR) of 1.7 (1.3–2.2). The association became stronger when examining the 4 cohorts with more than 10 years of follow-up, RR 2.7(1.7–4.4). Foot fractures were the only other type of fracture found to be increased in type 2 diabetics in this meta-analysis [2].

Type 1 Diabetes

The risk of fractures is significantly greater in type 1 diabetes when compared to the general population as well as to patients with type 2 diabetes[3]. The evidence supporting a higher fracture risk in type 1 diabetes comes mostly from the same studies comparing fracture risk in type 2 diabetes to the general population, but the results for type 1 are more consistent (see Table 2) [11] [7, 24, 25]. Studies of type 1 diabetes have focused mostly on hip fractures and have found a higher risk of hip fractures in type 1 diabetes with a range of RR from 1.7–12.3 [2]. However, fracture risk was also found to be moderately increased at the spine and proximal humerus [11, 24].

The risk appears to be similar between men and women with type 1 diabetes although due to a smaller number of studies examining this relationship no definite conclusions can be drawn. A recent meta-analysis of 5 cohort studies showed an overall RR of 8.9 (CI 7.1–11.2) for hip fractures in type 1 diabetics [2]. The incidence of hip fractures in the Nurses Health Study was reported as 383 per 100,000 over a follow up period of 2.2 million person-years [3]. This is a 6-fold increase from the overall incidence of hip fractures in this population, reported at 63 per 100,000, and a 2.5-fold increase from the incidence of hip fractures in type 2 diabetics in this population.

Impaired Bone Quality

Longstanding or poorly controlled diabetes may offset the beneficial effects of higher BMD via glycation of bone proteins and cells. Advanced glycation end products (AGEs) may adversely affect collagen, osteocytes, and multipotent bone marrow stem cells [28, 29] [30, 31]. The ability of multipotent bone marrow stem cells to proliferate or differentiate into bone or adipose tissue is decreased in the presence of AGEs in in-vitro studies [31]. Osteoclast induced bone resorption is enhanced by AGEs in mouse models [32] and the addition of AGE to osteoblast-like cells significantly impairs their ability to produce collagen [30]. Glycated collagen in turn may be able to inhibit expression of osteoblasts [30] or become stiff and resistant to proteolytic degradation, necessary for bone turnover [33]. Rodent studies have shown impaired biomechanical properties and decreased bone formation in hyperglycemic rats. These findings were significantly attenuated when hyperglycemia was controlled [34]. Pentosidine, a cross-link structure between lysine and arginine residues, is a major AGE. A recent analysis of the Health ABC study participants[35] showed a significant association between urinary pentosidine levels and clinical fractures in elderly diabetics but not in those without diabetes. This finding supports the role of AGEs in impairing bone quality.

Few studies have looked at the association between HbA1c and fractures. Elevated HbA1c has been associated with decreased markers of bone turnover, serum osteocalcin and C-terminal telopeptide (β-CTX), in elderly nursing home patients with type 2 diabetes [15]. Another study showed that having both a higher BMI (>24) and HbA1c (>9%) was associated with a greater risk of multiple vertebral fractures on spine x-rays (OR 5.4 CI 1.2,26.1) [36]. Other studies have shown a lower incidence of fractures in diabetics with a higher HbA1c. In a study of type 2 diabetic women aged 50–74 years, 26% of those with a hip fracture had an HbA1c >9% as compared to 50% of those who did not have a hip fracture [7]. It needs to be noted, however, that the HbA1c in this population was determined at baseline, i.e. nine years prior to their hip fracture and thus may not be reflective of their glycemic control during the follow-up period. A longer duration of diabetes is associated with a greater fracture risk in most [3–5, 9, 16] but not all [8, 11] studies in type 2 diabetics. The greater risk of fractures reported in patients with advanced diabetes may be due to the prolonged exposure to hyperglycemia and to the detrimental effects of AGEs on bone. The duration after which studies find an escalation in fracture risk is approximately 12–14 years after the diagnosis of diabetes [2, 4].

In addition to the effect of hyperglycemia on bone quality, bone mineralization may be altered in diabetes due to impaired vitamin D and calcium metabolism [37, 38]. Diabetics are more likely to be vitamin D deficient compared to the general population [39, 40]. Vitamin D deficiency may be greater in type 2 diabetics compared to type 1 diabetics, even after adjusting for obesity which is known to contribute to a higher distribution volume of this fat-soluble vitamin [41]. Hyperglycemia has been shown to impair renal calcium absorption [42] and this finding is corrected when hyperglycemia is controlled [43]. Finally, the development of renal insufficiency with its associated secondary or tertiary hyperparathyroidism and impaired vitamin D metabolism may further contribute to impaired bone quality in diabetes.

Falls

Diabetes is associated with a higher risk of falls [44, 45]. Diabetic neuropathy and neuromuscular impairment have been shown to be a major risk factor for falls [45, 46]. Furthermore, diabetic neuropathy may lead to falls that are more severe and falls that are sideways, as opposed to forward or backward. Both of these fall characteristics have been associated with a greater risk of fractures [47, 48]. This is supported by findings linking diabetic neuropathy with calcaneal and metatarsal fractures [49] [50].

A higher risk of falls has also been seen in diabetes of longer duration, and as discussed above, longer duration of diabetes is also associated with a greater fracture risk [2]. Higher rates of diabetes-related complications such as neuropathy, impaired vision due to retinopathy or cataracts [22, 45], stroke [16, 51], and impaired renal function [45], more prevalent in advanced diabetes may underlie the higher risk of falls and fractures in this population.

Diabetics are also more likely to be on medications that increase fall risk. A large retrospective study of older diabetics (>66 years old) found that male and female subjects with diabetes not only had a higher risk of hip fractures, but also, when compared to age matched controls, they were more likely to be on medications that increased fall risk (59% vs. 47%, p<0.001). These medications include sedatives, opiates, anti-epileptics and anti-parkinsonian agents [16].

Severe hypoglycemia may be another contributor to fall risk. Intensive glycemic control (HbA1c<6%) has been associated with an increased risk of falls but only in those diabetics treated with insulin (OR 4.36, CI 1.3,14.5) not in those treated with oral agents [45]. Some [4, 16, 20] [5, 8] but not all [6, 7] studies have shown an increased risk of fracture with insulin treatment. The higher fracture risk associated with insulin use remains unchanged even when adjusted for duration of diabetes [5, 16]. A possible explanation for this finding may be an increased frequency of severe hypoglycemic events in insulin treated diabetics. This may be one reason for why type 1 diabetics have a greater risk of fractures compared to type 2 diabetics. Other potential contributors to the observed difference in fracture risk between type 1 and type 2 diabetes may be lower body mass index, longer duration of disease, higher rates of diabetic retinopathy resulting in impaired vision [52], and a higher incidence of associated autoimmune diseases such as celiac disease [53] potentially predisposing type 1 diabetics to calcium malabsorption.

Thiazolidinediones

Recent studies have suggested an association between thiazolidinedione treatment and an increased risk of bone fractures in patients with type 2 diabetes [54] [55, 56], as well as in subjects with pre-diabetes [57] and older women with diabetes [58]. Fractures associated with thiazolidinediones were first reported in 2006, as an adverse outcome in the ADOPT Trial, a large, double blind randomized controlled study comparing the glycemic durability in patients with type 2 diabetes receiving monotherapy with rosiglitazone, metformin or glyburide [59]. This study showed an increased number of fractures in women treated with rosiglitazone but not in men. There was an increased rate of fractures in the hand, humerus, and foot in these women but not in the hip [59]. The cumulative incidence of fractures was 15.1% after 5 years of rosiglitazone treatment as compared to 7.3% and 7.7% for metformin and glyburide treatment respectively. The average age of subjects was 56.3 ± 10.0, which was no different from the other treatment groups. However, the increase in fractures was also seen in younger, pre-menopausal subjects of this study [60]. Similar findings regarding fractures have also been reported with pioglitazone use in postmenopausal diabetic women [61]. A recent case control analysis of the UK General Practice Research Database, confirmed the association of thiazolidinedione use with incident fractures (OR 2.43) [62]. Fractures were mainly reported in the hips and wrist and, unlike prior studies, the association of fractures with TZDs was found to be independent of age or gender. The association of fractures and TZD use in men has been further confirmed by a cross-sectional study of 43 males with type 2 diabetes. This study revealed a higher rate of radiographically confirmed vertebral fractures in those on metformin plus rosiglitazone compared to those on metformin alone (66.7% vs. 27.3% p=0.01)[56]. These findings remained true even after adjustment for BMD and age. Based on in-vitro studies and animal models, TZDs seem to act on pluripotent mesenchymal bone marrow stromal cells to cause an increase in adipocyte precursors at the expense of osteoblast precursors. These changes shift the balance of osteoblast and adipocyte precursors leading to increased fat accumulation, bone loss and ultimately increased fracture risk [63, 64]. However, studies in humans have not been done and other potential mechanisms, including an indirect negative effect on osteoblasts via enhanced secretion of adipocyte factors, remains to be elucidated.