Glucagon‐like peptide‐1(GLP‐1) receptor agonists: potential to reduce fracture risk in diabetic patients?

Abstract

This review summarizes current knowledge about glucagon‐like peptide 1 receptor agonists (GLP‐1 RA) and their effects on bone metabolism and fracture risk. Recent in vivo and in vitro experiments indicated that GLP‐1 RA could improve bone metabolism. GLP‐1 could affect the fat‐bone axis by promoting osteogenic differentiation and inhibiting adipogenic differentiation of bone mesenchymal precursor cells (BMSCs), which express the GLP‐1 receptor. GLP‐1 RA may also influence the balance between osteoclasts and osteoblasts, thus leading to more bone formation and less bone resorption. Wnt/β‐catenin signalling is involved in this process. Mature osteocytes, which also express the GLP‐1 receptor, produce sclerostin which inhibits Wnt/β‐catenin signalling by binding to low density lipoprotein receptor‐related protein (LRP) 5 and preventing the binding of Wnt. GLP‐1 RA also decreases the expression of sclerostin (SOST) and circulating levels of SOST. In addition, GLP‐1 receptors are expressed in thyroid C cells, where GLP‐1 induces calcitonin release and thus indirectly inhibits bone resorption. Furthermore, GLP‐1 RA influences the osteoprotegerin(OPG)/receptor activator of nuclear factor‐κB ligand (RANKL)/receptor activator of nuclear factor‐κB (RANK) system by increasing OPG gene expression, and thus reverses the decreased bone mass in rats models. However, a recent meta‐analysis and a cohort study did not show a significant relationship between GLP‐RA use and fracture risk. Future clinical trials will be necessary to investigate thoroughly the relationship between GLP‐1 RA use and fracture risk in diabetic patients.

Introduction

The fracture risk of type 2 diabetic patients is increased compared with non‐diabetic individuals 1, 2. Even if the bone mineral density (BMD) in type 2 diabetic patients is not low, or even higher, the fracture risk of ankle, hip and upper extremity is still increasing 3, 4, 5. The rising fracture risk cannot be simply explained by BMD. Bone strength reduced as a result of diabetes may not be completely reflected in the measurement of BMD. Poor bone quality may be another important factor that increases fracture risk in diabetic patients.

Previous studies have shown beneficial effect of glucagon‐like peptide 2 (GLP‐2) and glucose‐dependent insulinotropic polypeptide (GIP) on bone formation or resorption 6, 7, 8, 9, but the effects of GLP‐1 on bone metabolism remains unclear. Recent in vivo and in vitro experiments showed that the glucagon‐like peptide 1 receptor agonist (GLP‐1 RA) can improve bone metabolism. The primary objective of this review was to focus on the current knowledge about GLP‐1 RA and their effects on bone physiology and fracture risk and explore the therapeutic potential of GLP‐1 RA in the treatment of osteoporosis in diabetic patients.

Overview of GLP‐1 and GLP‐RA

The discovery of the incretins (including GLP‐1 and GIP) opens up a novel therapy in the treatment of diabetes. GLP‐1, a 30/31‐amino acid hormone, was found to be the second incretin hormone following the discovery of GIP. GLP‐1 is a tissue‐specific post‐translational proteolytic product of the proglucagon gene which is secreted by intestinal L‐cells in response to nutrient ingestion and stimulates insulin secretion from pancreatic β cells 10, 11, 12. Various forms of GLP‐1 are produced from their full length precursors by the action of prohormone convertase 1/3 13 and appear to show similar insulinotropic effects in humans. GLP‐1(7–37) and GLP‐1(7–36) NH2 are thought to be biologically active and exert an insulinotropic effect on pancreatic β cells 14. In humans, the majority of GLP‐1 in the circulation is GLP‐1(7–36) NH2 15. The half‐life of bioactive GLP‐1 in the circulation is approximately 2 min due to rapid inactivation by the widespread proteolytic enzyme dipeptidyl peptidase‐4 (DPP‐4) 16, 17. DPP‐4 is a serine protease that specifically cleaves dipeptides from the amino terminus of proteins or oligopeptides that contains an alanine or proline residue in position two, hence inhibiting their activity 18. GLP‐1, which contains a penultimate alanine residue, is metabolized rapidly to GLP‐1(9–37) or GLP‐1 (9–36) NH2, which have lost their insulinotropic effects 18, 19. More remarkably, DPP‐4 is also found on the position directly adjacent to the sites of GLP‐1 secretion. Consequently, over half of GLP‐1 has already been inactivated by DPP‐4 in the portal circulation 10. Ultimately, less than 5% of intact GLP‐1 reaches the systemic circulation 20. Glomerular filtration and tubular uptake and catabolism are the major route of GLP‐1 elimination 21.

GLP‐1 plays its biologic role primarily by binding to specific receptors, the GLP‐1 receptor (GLP‐1R) 22, 23, 24, which belongs to the G‐protein coupled receptor family. It can activate adenylate cyclase and increase levels of intracellular cyclic adenosine monophosphate (cAMP) in pancreatic β cells which then stimulates glucose‐dependent insulin secretion 20. What is more, GLP‐1 supresses appetite and lowers the rate of gastric emptying 25. In addition to its well‐known insulinotropic effects, GLP‐1 exerts its critical effect on a variety of biological processes in diverse tissues and organs that express GLP‐1R, including the bone tissue 20.

Different kinds of GLP‐1 RA are being developed as anti‐diabetic drugs, as GLP‐1 increases post‐prandial insulin secretion and suppresses post‐prandial glucagon secretion 26. In recent years, GLP‐1 RA (e.g. exenatide and liraglutide) have been widely and successfully used in the clinic. Exenatide, a 39 amino acid peptide that was originally isolated from the venom of the Heloderma suspectum lizard, exhibits approximately 53% amino acid identity sequence with native GLP‐1 and is a highly potent agonist for the mammalian GLP‐1 receptor 27. In contrast to native GLP‐1, exenatide is not a substrate for DPP‐4, as it contains a glycine at position two, and thus exhibits a longer circulating half‐life. Liraglutide is a human GLP‐1 analogue with 97% amino acid homology to native human GLP‐1. Two amino acid modifications enable its non‐covalent binding to albumin, thereby extending the half‐life to 13 h. Liraglutide exhibits all the actions of native GLP‐128.

Diabetes and osteoporosis

Although osteoporosis and T2DM are traditionally regarded as unrelated disease entities, emerging evidence indicates that they share many features including genetic predispositions, molecular mechanisms and pathophysiological mechanisms. ITGA1, a new candidate locus affecting both blood glucose and BMD, may partly explain the genetic contribution to the epidemiological observations linking T2DM and osteoporosis 29. Besides, regulatory factors, including insulin, peroxisome proliferator activated receptor gamma (PPARγ), gut hormones such as GLP‐1, and bone derived hormone osteocalcin are involved in the coordinated regulation of bone and energy homeostasis 30. Wingless‐related integration site (Wnt)/β‐catenin signalling, which plays an important role in bone metabolism, is also thought to be a common element in the pathogenesis of osteoporosis and diabetes. The anomalies of this signalling pathway may lead to the occurrence of type 2 diabetes and osteoporosis. A single missense mutation in low density lipoprotein receptor‐related protein 6 (LRP6), the co‐receptor of the Wnt signalling pathway, was genetically associated with osteoporosis and diabetes 31. Mice lacking osteocalcin show glucose intolerance 32, 33. Contrary to predictions from mouse models, Schwartz et al. found that bisphopshonate‐treated patients with low osteocalcin levels do not have an increase in the incidence of diabetes 34.

BMD is used as a strong predictor of fracture risk in clinical practice, but it does not encompass every aspect of bone strength. The fracture risk of T1DM increases as a consequence of decreased BMD that result from absolute deficiency of insulin, insulin‐like growth factor‐1 (IGF‐1) and lower values of peak bone mass 35. However in T2DM, increased load on bone because of obesity and hyperinsulinaemia as a result of insulin resistance may lead to increased bone formation. Several studies on osteoporosis and T2DM have demonstrated conflicting results, with BMD variously reported to be decreased 36, 37, increased 38, 39, 40 or unchanged 41, 42. Despite the general increased BMD of individuals with T2DM 2, type 2 diabetic patients have a higher risk of fracture than non‐diabetic patients with the same BMD measurement 43. T2DM influences bone quality rather than BMD, which explains only 70 ~ 75% of the variance in bone strength, while other important factors, including the impaired bone micro‐architecture, the accumulation of micro‐fractures, the bone remodelling imbalance 44, or increased tendency to fall, may be ignored.

Besides, the increased bone fracture risk in diabetic patients could, in part, be due to the influence of anti‐diabetic drugs. Previous research has shown that long term treatment with thiazolidinediones (TZDs) is associated with an increased risk of fracture in women with T2DM compared with other glucose‐lowering agents 45. The effect of TZD on bone fractures could be due to a specific inhibition of osteoblast differentiation and activity 46. In pharmacology, TZDs act as PPARγ activators 47. In fact, PPARγ is expressed in bone and adipose tissue, and is able to switch the pluripotent mesenchymal stem cells (MSC) toward adipocyte at the expense of osteoblast formation 48. As a result, adipogenesis exceeds osteogenesis.

GLP‐1 and osteoporosis

GLP‐1 and Wnt/β‐catenin signalling pathway

GLP‐1 exerts its influence by binding to its receptor. Its effect on bone metabolism can be speculated in some GLP‐1R deletion animal experiments. Yamada et al. found that GLP‐1R‐deficient mice showed cortical osteopenia, bone fragility, increased osteoclastic numbers and bone resorption activity 49 (Table 1). In another similar study, Mabilleau et al. also demonstrated that GLP‐1R knockout mice presented with a significant reduction in ultimate load, yield load, stiffness, total absorbed, post‐yield energies, cortical thickness and bone outer diameter, but the mineral quantity and quality were not significantly influenced 50. What is more, the improvement of bone metabolism and anti‐osteoporosis effects of GLP‐1 are demonstrated in rodent experiments, mainly in the osteoblast lineage. Its significant role in the improvement of bone metabolism has been reported in aged ovariectomized rats models 51, streptozotocin‐induced type 2 diabetic and fructose‐induced insulin‐resistant (IR) rat models 52 and hyperlipidic rat models 53. This intrigues clinical doctors for its potential application in osteoporosis.

Table 1.

Published molecular and pre‐clinical studies on the association between GLP‐1 and bone metabolism

Author et al.	Study subject	Study method	Main result
Yamada 49	GLP‐1R knockout mice, bone marrow cells and osteoblasts	Exendin‐4 and calcitonin treatment	GLP‐1R knockout mice have cortical osteopenia and bone fragility as well as increased osteoclastic numbers and bone resorption activity and reduced levels of calcitonin mRNA transcripts in the thyroid. Exendin‐4 increased calcitonin gene expression in the thyroid of mice
Mabilleau 50	Male GLP‐1R knockout mice	Analyze the presentation of GLP‐1R knockout mice	GLP‐1R knockout mice presented with a significant reduction in ultimate load, yield load, stiffness, cortical thickness, bone outer diameter and the maturity of the collagen matrix. But the mineral quantity and quality did not change significantly
Ma 51	Old ovariectomy rats	Exendin‐4 administration lasted for 16 weeks	Exendin‐4 not only inhibited bone resorption by increasing the OPG : RANKL ratio, but also promoted bone formation by increasing the expression of OC and Runx2 in old ovariectomy rats
Nuche‐Berenguer 52	Streptozotocin‐induced type 2 diabetic rats, fructose‐induced insulin‐resistant rats	Continuous infusion of GLP‐1 for 3 days	GLP‐1 increased OC and OPG in type 2 diabetic, insulin‐resistant rats and RANKL in type 2 diabetic rats. GLP‐1 induced an insulin‐ and PTH‐ independent bone anabolic action in insulin‐resistant and type 2 diabetic rats
Nuche‐Berenguer 53	Hyperlipidic rats	Continuous infusion of GLP‐1 and exendin‐4 for 3 days	GLP‐1 and exendin‐4 similarly reversed the decreased femoral and vertebral bone mass by increasing OC gene expression and the OPG : RANKL ratio in hyperlipidic rats
Sanz 57	hMSCs	Intervention with GLP‐1 in cell proliferation and cell differentiation	GLP‐1 significantly reduced the expression of PPARγ, C/EBPα, and LPL and prevented cell differentiation into adipocytes in hMSCs
Nuche‐Berenguer 66	Osteoblastic MC3T3‐E1 cells	Analysis of GLP‐1 binding and cross‐linking studies	GLP‐1 can directly and functionally interact with osteoblastic cells independent of the cAMP‐linked GLP‐1 receptor, possibly through a GPI/IPG‐coupled receptor
Kim 78	Type 2 diabetic OLETF rats, osteocyte‐like MLO‐Y4 cells and osteocytes of rat femurs	Investigated the presence of GLP‐1 receptors and the effect of exendin‐4 treatment through RT‐PCR, Western blot and confocal microscopy	GLP‐1 receptor was present on MLO‐Y4 cells and osteocytes of rat femurs. Exendin‐4 reduced the levels of SOST/sclerost in MLO‐Y4 cells. Besides, exendin‐4 reduced serum levels of SOST, increased serum levels of osteocalcin and femoral BMD in type 2 diabetic OLETF rats
Gier 88	Thyroid tissue samples with medullary thyroid carcinoma, C cell hyperplasia, papillary thyroid carcinoma, and normal human thyroid	Immunofluorescence for expression of calcitonin and GLP‐1 receptors	The neoplastic and hyperplastic lesions of thyroid C cells express the GLP‐1 receptor and GLP‐1 receptor expression is also detected in 18% of papillary thyroid carcinomas and in C cells in 33% of control thyroid lobes in humans
Hegedus 89	T2DM or non‐diabetic obese patients receiving liraglutide treatment	CT concentrations were measured at 3‐month intervals for no more than 2 years	No significant change.
Bjerre 90	The thyroid of mice, rats, cynomolgus monkeys and humans	The activation of the thyroid GLP‐1 receptor with GLP‐1 RA	GLP‐1 RA stimulated calcitonin release, up‐regulation of calcitonin gene expression and subsequently C‐cell hyperplasia in rats. In contrast, humans and/or cynomolgus monkeys had low GLP‐1 receptor expression in thyroid C‐cells and GLP‐1 RA did not activate adenylate cyclase or generate calcitonin release in primates

BMD, bone mineral density; cAMP, cyclic adenosine monophosphate; C/EBPα, CCAAT/enhancer‐binding protein α; GLP‐1RA, glucagon‐like peptide‐1 receptor agonists; hMSCs, human mesenchymal stem cells; LPL, lipoprotein lipase; OC, osteocalcin; OP, osteoporosis; OPG, osteoprotegerin; PPARγ, peroxisome proliferator‐activated receptorγ; PTH, parathyroid hormone; RANKL, receptor activator of nuclear factor‐kappaB ligand; Runx2, runt‐related transcription factor 2; SOST, sclerostin.

Bone formation is a series of events that include the differentiation of mesenchymal precursor cells into osteoblast precursors, the maturation of osteoblasts and the formation and the mineralization of matrix 54. The fat‐bone axis is involved in this important course. Osteoblasts and adipocytes differentiate from a common mesenchymal precursor cell, BMSCs 55, which can differentiate into osteoblasts, adipocytes as well as other cell types under the influence of local growth factors, hormonal regulators and transcriptional factors 56. GLP‐1 in humans can promote proliferation of BMSCs and inhibit adipocyte differentiation 57. Consistent with these findings, Ma et al. also found that exenatide can promote the osteogenic differentiation and inhibit the adipogenic direction of differentiation of BMSCs 51. Previous research has confirmed that the GLP‐1 receptor is expressed in BMSCs and immature osteoblasts but not in mature osteoblasts 57, 58, 59. Evidence from in vitro experiments suggests no direct effect of GLP‐1 in either osteoblasts or osteoclasts 49. This evidence suggests that BMSCs could be the key targets of GLP‐1 51 (Figure 1). The balance between osteoblast/adipocyte differentiation of BMSCs plays an important role in bone homeostasis. If too many BMSCs differentiate towards adipocytes, the number of osteoblasts will be reduced and bone absorption exceeds bone formation, resulting in the reduction of BMD and osteoporosis. Vermas et al. found that the number of adipocyte of bone marrow in patients with osteoporosis significantly increased 60. This indicated that reduction of osteoblasts is often accompanied by an increase in adipocyte ratio. Similarly, in the ageing mouse model, when fat cells in bone marrow increased, bone formation decreased significantly 61. Therefore, the direction of differentiation of BMSCs plays a crucial role in the pathogenesis of osteoporosis 62, 63, 64. Examples of drugs that increase adiposity (especially visceral fat) at the expense of osteoblast differentiation are glucocorticoids 65. Although it is widely believed that GLP‐1 does not directly affect osteoblasts or osteoclasts, Nuche‐Berenguer et al. found that GLP‐1 can directly and functionally interact with osteoblastic cells, possibly through a glycosylphosphatidylinositols(GPI)/inositolphosphoglycans (IPG)‐coupled receptor that is different from the cloned GLP‐1 receptor in pancreas 66.

Figure 1.

Model for the influence of GLP‐1 on osteogenesis. Bone homeostasis is regulated by the balance between osteoblastic bone formation and osteoclastic bone resorption. An imbalance between these two factors will lead to osteoporosis. GLP‐1 not only inhibits adipocyte differentiation from BMSCs, but also inhibits osteoblast differentiation, and thus improves bone metabolism. GLP‐1 glucagon‐like peptide‐1; BMSCs bone mesenchymal precursor cells

The canonical Wnt signalling pathway regulates the differentiation and maturation of the osteoblast. The signal pathway includes low density lipoprotein receptor related protein (LRP) 5/6, β‐catenin, GSK‐3β and other related regulative factors, such as Dkk1 and sclerostin 67. LRP5 gene mutations can lead to different abnormal bone phenotypic abnormalities, including osteoporosis pseudoglioma syndrome (OPPG) and increased bone mass 68. In animal experiments, LRP5 knockout mice are characterized by decreased bone mass, resulting from inactivating canonical Wnt signalling pathways 69. Wnts are secreted glycoproteins, consisting of 350 ~ 400 amino acids and 23 ~ 24 residues of conserved cysteine 70, many of which are likely to participate in intramolecular disulfide bonds. In BMSCs, canonical Wnt signalling regulates a reciprocal relationship between adipogenesis and osteogenesis. In previous research, Wnt signalling is regarded as a key pathway for β cell growth and differentiation. Recent findings link GLP‐1 as a direct activator of this pathway. GLP‐1 could also stabilize β‐catenin by binding to its receptor and increasing the level of intracellular cAMP 71. In the inactive state of the Wnt/β‐catenin pathway, Dkk1 forms a ternary complex with transmembrane proteins kremen and LRP5/6 and then induces rapid endocytosis and removal of the Wnt receptor LRP5/6 from the plasma membrane. Activation of the Wnt signalling pathway promotes bone formation while inactivation of the pathway leads to reduced bone mass 72, 73, 74. The activation of the GLP‐1 receptor and the Wnt signalling pathway showed similar effect on adipocyte differentiation and bone metabolism. Specific gene expression of adipocyte differentiation including PPARγ, C/EBPα and LPL decreased 57 but bone formation‐related gene including Runx2/cbfa1 69, 74, OPG 51, 52, 53, and OC 51, 52, 53 increased (Figure 2).

Figure 2.

GLP‐1 and Wnt signalling in the regulation of bone formation (A) In the inactive state of the Wnt/β‐catenin pathway, Wnt is inhibited by a decoy receptor sFRP, where as its co‐receptor LRP5/6 binds to inhibitory protein sclerostin or Dkk. Activated GSK‐3β can result in proteosomal degradation of β‐catenin, which then inhibits osteoblast differentiation. Specific gene expression of adipocyte differentiation including PPARγ, C/EBPα, and LPL and bone formation negative regulatory genes SOST increased but bone formation positive regulatory genes including Runx2/cbfa1, OPG, and OC decreased. (B) The Wnt/β‐catenin pathway is activated by the binding of the Wnt to a co‐receptor complex LRP5/6 and frizzled family member. Disruption of the GSK‐3β inhibitory complex stabilizes the β‐catenin, which is then translocated into the nucleus and activates transcription. GLP‐1 could also stabilize β‐catenin by binding to its receptor and increasing the level of intracellular cAMP. The changes in the expression of related genes are the opposite of Figure 2A. sFRP secreted frizzled related protein; Dsh dishevelled; LRP5/6 lipoprotein receptor related proteins 5 and 6; Dkk1 dickkopf1; APC adenomatous polyposis coli; GSK‐3β glycogen synthesis kinase‐3β; Frat1 frequently rearranged in advanced T‐cell lymphomas‐1; cAMP cyclic adenosine monophosphate; TCF/LEF T‐cell factor/lymphoid enhancing factor; PPARγ peroxisome proliferator‐activated receptorγ; C/EBPα CCAAT/enhancer‐binding protein α; LPL lipoprotein lipase; SOST sclerostin; Runx2/cbfa1 runt‐related transcription factor 2/core binding factor alpha1; OPG osteoprotegerin; OC osteocalcin

Regulation of the Wnt/β‐catenin signalling pathway is carried out mainly by proteins that either act as competitive binders of Wnts. Sclerostin (SOST), which might contribute to the pathogenesis of bone loss in T2DM, is the major one. SOST is released by mature osteocytes and inhibits Wnt/β‐catenin signalling by binding to LRP5 and preventing the binding of Wnt protein to LRP5/6 75. Serum SOST levels of the diabetic patients were significantly higher than non‐diabetic patients and were increased with age 76. The high SOST level might indicate increased osteocyte activity in post‐menopausal patients with T2DM 77. Osteocytes could be another target of GLP‐1 RA and GLP‐1 receptor is expressed in osteocytes as showed in osteocyte‐like MLO‐Y4 cells and osteocytes of rat long bone 78. In in vitro experiments, GLP‐1 RA decreases the expression of SOST in MLO‐Y4 cells whereas it decreases the circulating levels of SOST in T2DM rat models in in vivo experiments. These findings demonstrate that GLP‐RA might increase the femoral BMD through decreasing the expression of SOST/sclerostin in osteocytes in T2DM rat models 78.

GLP‐1, body weight and osteoporosis

DPP‐4 inhibitors (DPP4‐I) are used to prolong the action of GLP‐1, so their effect on bone may be similar to GLP‐1. To explore this possibility, a meta‐analysis was performed to compare DPP4‐I with placebo or other anti‐diabetic drugs in all randomized clinical trials with a duration of at least 24 weeks. This meta‐analysis showed a 40% reduction of fracture risk for DPP4‐I users compared with other patients 79 (Table 2). However, a retrospective population based cohort study demonstrated no different risk of fracture comparing DPP4‐I users with controls 80. In another meta‐analysis, Mabilleau et al. found that GLP‐1 RA did not reduce bone fracture risk in T2DM compared with the use of other anti‐diabetic medications 81. Consistent with this study, Driessen et al. also found that GLP1‐RA use (the median duration of actual GLP1‐RA use was 1.2 years) was not associated with a decreased risk of bone fracture as compared with users of other anti‐hyperglycaemic drugs in a population‐based cohort study 82. These clinical results seem to demonstrate that GLP‐RA use is not related to bone fracture risk. However we should realize that the average duration of GLP1‐RA and DPP‐I use is relatively short which may make it fail to exert an effect on bone fracture risk. Besides, GLP‐1 users often have a high BMI which may influence the final result. Careful assessment of the incidence of fractures in clinical trials will be important.

Table 2.

Clinical studies of the relationship between GLP‐1 RA and DPP‐4 inhibitors use and bone fracture

Author et al.	Study subject	Study method	Main result
Monami 79	Type 2 diabetic patients treated with DPP‐4 inhibitors with a duration of at least 24 weeks	Meta‐analysis of bone fracture risk	DPP‐4 inhibitors were associated with a reduced risk of fractures even after the exclusion of comparisons with thiazolidinediones or sulfonylureas
Driessen 80	Type 2 diabetic patients treated with DPP‐4 inhibitors from the CPRD database	retrospective population based cohort study of bone fracture risk	DPP4‐I use was not associated with fracture risk compared with controls and with other non‐insulin anti‐diabetic drug users
Mabilleau 81	Type 2 diabetic patients treated with GLP‐1 RA in randomized clinical trials	Meta‐analysis of bone fracture risk	The use of GLP‐1RA does not reduce the risk of bone fracture in T2DM compared with the use of other antidiabetic medications
Driessen 82	Type 2 diabetes patients treated with GLP‐1RA from the CPRD database.	Population‐based cohort analysis of bone fracture risk	There was no decreased risk of fracture with current use of GLP‐1RA compared to never‐GLP‐1RA use. Osteoporotic fracture risk was also not decreased by current GLP‐1RA use

CPRD, Clinical Practice Research Datalink; GLP‐1RA, glucagon‐like peptide‐1 receptor agonists.

A primary difference between GLP‐1 RA and DPP‐4I lies in weight loss, which often accompanies treatment with GLP‐1 RA. The weight loss effect of GLP‐1 RA is beneficial to many overweight or obese type 2 diabetic patients who are advised to lose weight. The positive association between body weight and BMD has previously been established in some studies 83, 84. Bone mobilization and loss could result from weight loss. According to Canadian osteoporosis guidelines, obvious weight loss is a risk factor for osteoporosis 85, 86. A positive effect of GLP‐1 RA on bone homeostasis could be weakened by weight loss‐induced mechanical loading and bone mass decrease, thus causing neutrality of GLP‐1 RA treatment on bone fracture risk in clinical trials. However from the perspective of the fat‐bone axis, this positive effect could be independent of weight loss. Apart from the direct effect of soft tissue mass on bone through skeletal loading, GLP‐1 can also act as a mediator of the fat and bone cells 87, but the exact mechanism remains unclear.

GLP‐1 and calcitonin

Calcitonin may contribute to the improvement of bone metabolism with GLP‐1 treatment. GLP‐1 receptors are expressed in thyroid C cells, where GLP‐1 induces post‐prandial calcitonin release and thus bone resorption is indirectly inhibited 49. However, these previous findings reported in rodents may not apply to humans. Recent studies have found that only 33% of the normal thyroid C cells express GLP‐1 receptors 88. Clinical experimental data do not support an effect of GLP‐1 receptor activation on serum calcitonin levels in humans in a clinical trial with over 5000 subjects receiving 2 years liraglutide treatment 89. As a result, considering the differences between rodents and humans, it is plausible that the observed bone protective effects of GLP‐1 in rodent models may be the result of direct effects on bone cells and indirect effects mediated via calcitonin. In contrast, non‐human primate and human thyroid glands and the human TT cell line, which is derived from human medullary thyroid carcinoma, express very low levels of GLP‐1 receptors. The human TT cell line, unlike rat thyroid C‐cell lines, does not respond to GLP‐1 RA with an acute release of calcitonin 90. Collectively, it seems that the influence of GLP‐1 RA on the human thyroid is limited.

GLP‐1 and OPG/RANKL/RANK

The discovery of the osteoprotegerin (OPG)/receptor activator of nuclear factor‐κB ligand (RANKL)/receptor activator of nuclear factor‐κB (RANK) system in the past decade provides us with a better understanding of bone metabolism, especially bone resorption. OPG, RANK and its ligand RANKL, which all belong to the family of tumour necrosis factor receptors (TNFR), mediate osteoclastogenesis 91. In the skeleton, RANKL stimulates the differentiation, activation and survival of osteoclasts and thus enhances bone resorption 5, 8. OPG acts as a decoy receptor by binding to RANKL therefore preventing the interaction with its receptor RANK that is necessary for the activation of RANKL 92. Consequently, OPG acts as a negative regulator of osteoclast differentiation, activation and survival and therefore inhibits bone resorption 93. High RANKL and low OPG levels are associated with high bone turnover and bone loss, whereas low serum RANKL may relate to accumulation of microdamage and reduced bone quality 94. An imbalance of OPG/RANKL/RANK expression associated with diabetes may contribute to the delay of fracture repair during the course of diabetes 95. Elevated OPG in patients with T1DM may be the body's response to increased bone resorption 96. New drugs targeting the OPG/RANKL/RANK system proved to be efficient in reducing bone resorption and preventing bone loss in post‐menopausal osteoporosis 97. Recently, Nuche‐Berenguer et al. observed that exenatide administration exerts osteogenic effects in streptozocin‐induced type 2 diabetes and fructose‐induced insulin‐resistant rats. They found that GLP‐1 RA administration increased OPG gene expression and the OPG/RANKL ratio and reversed the decreased femoral and vertebral bone mass in these rats 52. Their subsequent studies demonstrated that GLP‐1 and exendin‐4 are similarly efficient in reversing the bone alterations in the hyperlipidic‐related rat model 53.

Conclusion

Reduced bone strength in type 2 diabetic patients involves many complex factors. Some oral anti‐diabetic drugs even aggravate the impaired bone turnover and increase bone fracture risk. The recently reported bone‐related effects of GLP‐1, together with their known glucose‐lowering action, make them candidates for reducing fracture risk in diabetic conditions. Results from in vivo and in vitro experiments are quite encouraging. However, the clinical data about the relationship between GLP‐1RA and fracture risk are disappointing. A meta‐analysis and cohort study did not find any relationship between GLP‐RA use and fracture risk. Also, the influence of GLP‐1 RA on human thyroid is limited. However, we should not jump to a conclusion prematurely. Weight loss may weaken the protective effect of GLP‐1 RA on bone. The short duration of GLP1‐RA use and high BMI of GLP‐RA users are two main shortcomings of the clinical trials. Careful assessment of the fracture risk in future clinical trials with GLP‐1RA will be necessary.

Competing Interests

All the authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare no support from any organization for the submitted work, no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years and no other relationships or activities that could appear to have influenced the submitted work.

Luo, G. , Liu, H. , and Lu, H. (2016) Glucagon‐like peptide‐1(GLP‐1) receptor agonists: potential to reduce fracture risk in diabetic patients?. Br J Clin Pharmacol, 81: 78–88. doi: 10.1111/bcp.12777.