Many researchers are becoming increasingly aware of the importance of the bone not only in the skeleton structure, but also in energy metabolism.[1-3] In particular, osteocalcin (OC), a protein secreted by osteoblasts and released from the bone matrix during osteoclastic bone resorption, has been proven to be involved in glucose metabolism. This process is induced by increasing insulin secretion and cell proliferation in the beta cells of the pancreas and by regulating insulin sensitivity through increased secretion of adiponectin.[3,4] Many cross sectional studies have shown that serum OC concentration are inversely proportional with plasma glucose levels, and fasting insulin, but directly proportional to lower insulin resistance (IR), insulin secretion, and serum adiponectin concentration.[5-7]

Bisphosphonates can reduce fracture risk through the reduction of osteoclastic activity and significantly decrease serum OC concentration.[8] We can speculate that significantly reduced levels of OC by long-term bisphosphate treatment may provoke IR, with subsequent decrease in insulin secretion eventually leading to changes in blood glucose.

We therefore conducted a prospective study to investigate the association between changes in glucose or insulin secretion and changes in OC for users of bisphosphonate.

We enrolled 84 participants in this study and classified the subjects according to their fasting glucose levels into the normal group (n=60) and the impaired fasting glucose (IFG, n=24) group (Table 1). After 16 weeks of treatment in total subjects, fasting glucose levels increased significantly from 5.3±0.56 to 5.5±0.75 mmol/L (Table 2). In the subgroup analysis, fasting glucose in the normal group also increased from 5.0±0.35 to 5.4±0.62 mmol/L (P<0.0001) after treatment (Table 3), while no significant changes were seen in the IFG group (Table 4). There were no significant statistical changes in fasting insulin levels, HOMA-IR, and HOMA-B% in the total, and IFG group after 16 weeks, but HOMA-IR significantly increased in normal group only (Table 3). Significant decreases were also seen in Total OC, Glu-OC, and Gla-OC levels in each group. There were no statistical correlations between changes in osteoalcin and glucose parameters after the 16 week-treatment in the total subjects, and normal group (Tables 5, 6).

Knockout mice lacking insulin receptors in their osteoblast have low levels of Glu-OC in their plasma and decreased osteoblast differentiation. These effects, in turn, can impair insulin secretion.[4] It has been found that an endocrinological feedback loop between the bone and the pancreas. Both Glu-OC and total OC have been inversely associated with fasting glucose, fasting insulin, haemoglobin A1c (HbA1C), and HOMA-IR levels in men and women. [10,11] In their study of middle-aged Korean men, Hwang et al.[12] noted that Glu-OC was found to enhance B while Gla-OC was associated with improved insulin sensitivity. Iki et al.[13] found that the level of Glu-OC, but not intact OC, showed a negative correlation with the glycemic index and IR. Shea et al.[14] insisted that circulating OC in its uncarboxylated form was not associated with IR, but total OC and the protein in its Gla form were inversely associated with IR. From the clinical perspective, many cross-sectional studies have found an association between OC and glucose metabolism. However, the few longitudinal studies that have been conducted do not provide sufficient evidence. Furthermore, no study has examined the effects of bisphosphonate on glucose metabolism and insulin secretion or resistance. Therefore, our study was designed to examine the dynamic changes in glucose after bisphosphonate therapy and to show that changes in OC are associated with changes in glucose and insulin homeostasis. We investigated the correlation between the glycemic index and changes in OC after 16 weeks, as bisphosphonates sufficiently suppress bone turnover at 3 to 6 months.

The present study demonstrated that bisphosphonate decreased the levels of two forms of OC, but these changes were not correlated with insulin secretion or resistance. Glucose levels significantly increased after 16 weeks of treatment, and HOMA-IR also increased in normal subjects, but these changes were not observed in whole subjects or IFG group. Furthermore, these changes were within normal range and not associated with changes in OC levels. It is not clear why blood glucose levels and HOMR-IR increased in patients treated with bisphosphonate, but it may be coincidental finding because of small subjects. Although bisphosphonate may affect glucose homeostasis, its effect may be weak or attenuated by other factors. Many clinical trials estimating the efficacy for osteoporosis treatment have not reported that bisphosphonates cause DM as an adverse event.[15] In fact, among users of alendronate, the risk of developing type 2 (T2) DM was reduced in a nationwide cohort study in Denmark.[16] As an antiresorptive treatment, hormone replacement therapy (HRT) can not only decrease OC, but also decrease fasting glucose and insulin levels and lower the incidence of diabetes in randomized trials.[17-19] Furthermore, Hwang et al.[20] demonstrated in a retrospective cohort study that circulating OC levels were not associated with the risk for T2 diabetes in middle-aged male subjects. Even though OC may affect glucose metabolism as in animal system, we can speculate that OC may have weak effects on glucose metabolism for humans, and its impact may be affected by another controlling mechanism.

Our study has some limitations. The sample size is small, and we did not include enough subjects in the IFG group. We also excluded a diabetes group. Furthermore, we did not perform oral glucose tolerance tests to verify glucose status. A 16 week-trial may not be long enough to observe a causal relationship, and a longer-term study including a larger number of subjects is needed for more accurate data.

Despite some limitations, this is a prospectively designed trial. Our findings suggest that bisphosphonate treatment for osteoporosis reduces OC, but that this change is not correlated with glucose homeostasis.