This study showed a 1.3-fold increased risk of fracture in current TZD users versus nondiabetic patients. Overall, current TZD users had no significantly increased risk of fracture compared with users of other AD drugs. The risk of fracture with current TZD use was, however, increased in women, among prolonged users of TZDs, and risks of foot/ankle and tibia/fibula fractures were significantly increased.
Our finding of an increased risk of fracture in women using TZDs versus other AD drug users is in line with previous studies. In a meta-analysis from randomized controlled trials [10
], the pooled data from the five trials that reported fracture risk by sex showed a significantly increased risk of fractures among women (OR = 2.23, 95 % CI 1.65–3.01) but not among men (OR = 1.00, 95 % CI 0.73–1.39). Another recent trial on rosiglitazone use presented a similar result: there was an increased fracture risk in women (RR = 1.82, 95 % CI 1.37–2.41) but not in men (RR = 1.23, 95 % CI 0.85–1.77) [27
]. In observational studies, the use of TZDs has also been linked with an increase in fracture risk [12
]. Habib et al. [12
] found in their cohort study that TZD use was associated with a 1.6-fold increased risk of fracture in women but not in men. Conversely, Douglas et al. [13
] reported an increased risk of fracture in both women and men who were exposed to TZDs. In a nested case–control study [11
] an increased risk of fracture with TZD use was only apparent in the group with cumulative exposure of 15 TZD prescriptions or more (HR = 2.86, 95 % CI 1.57–5.22). This result is in line with our finding of a significantly increased risk of osteoporotic fracture for current TZD users in the stratum with highest cumulative exposure.
The association between TZD use and fracture risk may be explained by the drug itself but also by the underlying disease because both T2DM and the use of TZDs seem to be independently associated with an increased risk of adverse skeletal effects. There is substantial evidence from in vitro studies that activation of PPARγ, which is stimulated by TZDs, causes a predominant formation of adipocytes from mesenchymal stem cells, while the development of osteoblasts is suppressed [4
]. Moreover, BMD was significantly reduced at the lumbar spine and hip in women exposed to TZDs in two randomized controlled trials [8
In addition, epidemiological studies suggest that T2DM is an independent risk factor for adverse effects on the skeleton [15
]. Even though patients with T2DM have on average a higher BMI [28
] as well as a higher BMD [16
] compared with the general population, which would in theory protect them against fractures, previous studies have found an increased risk of fractures in T2DM [15
]. This finding may be explained by different mechanisms. High blood glucose might have a direct toxic effect on bone cells [17
] and may cause a negative calcium balance by increased urinary calcium excretion [18
] and reduced intestinal calcium absorption [30
]. There is evidence that vitamin D metabolism is altered in diabetic patients [19
], which may lead to bone loss. Further, functional hypoparathyroidism may result in low bone turnover [32
]. Complications of diabetes, such as neuropathy and angiopathy, may also contribute to an increased risk of fracture as well as disease severity in general [21
To evaluate to what extent the association between TZD use and fracture risk is related to the drug or to the underlying disease, we created a proxy indicator of disease severity. We defined four different stages based on a prescription scenario that is often used in the treatment of T2DM. Patients who were currently using insulin (stage 4) had a significantly greater risk of osteoporotic fracture than patients in stages 1 and 2, which supports the hypothesis that the underlying disease is involved. If the drug itself were strongly related to fracture risk, we would expect a higher risk in patients who were currently exposed to TZDs (stage 3). However, patients in stage 3 had a similar risk of osteoporotic fracture as patients in stage 4. Therefore, we would argue that the association between TZD use and fracture is at least partially confounded by the underlying disease.
Strengths of our study include a reasonable sample size and duration of follow-up. Our source population included the total Danish population, and controls were therefore population-based. There was detailed longitudinal information on drug prescribing.
There are several limitations. Data on smoking and body mass index (BMI) were not available, and therefore, unmeasured confounding may have biased our findings. Patients with T2DM have on average a higher BMI than the general population [28
], which is protective against fracture [33
]. In our study, this may have led to underestimation of a true association between T2DM/TZD use and fracture. However, we did adjust our analyzes for the use of statins, which is probably correlated with high BMI. Using a proxy indicator of disease severity, we may have underestimated the role of the underlying T2DM as a confounder in the association between TZD use and fracture. We used current treatment with insulin as a proxy indicator for the highest level of disease severity (stage 4). The risk of fracture for patients in this stage may be biased by potential direct effects of insulin on bone. Although the role of insulin on fracture risk is still unclear, it has been suggested that higher insulin levels in T2DM may preserve and increase bone density and bone strength [30
]. If insulin were indeed protective against fractures, this may have resulted in underestimation of a true association between disease severity and fracture risk.
In conclusion, we found that patients who were currently using TZDs had a similar risk of osteoporotic fracture as patients who were currently using insulin. When observational studies assess the risk of fracture in patients using TZDs, the underlying T2DM should be taken into account.