Abstract
The gut hormone glucose-dependent insulinotropic polypeptide (GIP) causes postprandial insulin release and inhibits bone resorption assessed by carboxy-terminal collagen crosslinks (CTX).
To study if GIP affects bone homeostasis biomarkers independently of insulin release and glycemic level.
Randomized, double-blinded, crossover study with 5 study days.
Ten male C-peptide-negative patients with type 1 diabetes.
On 3 matched days with “low glycemia” (plasma glucose in the interval 3 to 7 mmol/L for 120 minutes), we administered intravenous (IV) GIP (4 pmol × kg−1 × min−1), glucagon-like peptide 1 (1 pmol × kg−1 × min−1), or placebo (saline), and on 2 matched days with “high glycemia” (plasma glucose 12 mmol/L for 90 minutes), we administered either GIP or saline.
CTX, procollagen type 1 N-terminal propeptide (P1NP), and parathyroid hormone (PTH).
During low glycemia: GIP progressively suppressed CTX from baseline by up to 59 ± 18% compared with 24 ± 10% during saline infusion (P < 0.0001). Absolute values of P1NP and PTH did not differ between days. During high glycemia: GIP suppressed CTX from baseline by up to 59 ± 19% compared with 7 ± 9% during saline infusion (P < 0.0001). P1NP did not differ between days. GIP suppressed PTH after 60 minutes compared with saline (P < 0.01), but this difference disappeared after 90 minutes.
Short-term GIP infusions robustly reduce bone resorption independently of endogenous insulin secretion and during both elevated and low plasma glucose, but have no effect on P1NP or PTH after 90 minutes.
Food intake plays a pivotal role in the diurnal variation in biochemical markers of bone remodeling (1). Thus, following a meal, carboxy-terminal collagen crosslinks (CTX)—a biochemical marker reflecting the rate of osteoclastic bone resorption (2)—is reduced by ∼50% from baseline values (1, 3). Markers of osteoblastic bone formation are typically less affected, as evidenced by levels of procollagen type 1 N-terminal propeptide (P1NP), which are suppressed by ∼8% following glucose ingestion (3). The gut hormone, glucose-dependent insulinotropic polypeptide (GIP), which is released following intake of carbohydrates—fat as well as protein (4)—has been proposed as an important contributor to the meal-induced suppression of bone resorption, i.e., representing a link between food intake and bone homeostasis (5, 6). We have recently shown that CTX is suppressed following GIP infusion in healthy, young male subjects (7). In these studies, GIP infusion during fasting led to an ∼30% suppression of CTX. Intravenous (IV) glucose administration also caused slight CTX suppression (∼13%), but interestingly, the combination of GIP and hyperglycemia acted in synergy to suppress bone resorption by ∼50%. Importantly, from these experiments, it was not possible to separate the effects of GIP and glucose, respectively, and combined, from that of endogenous insulin secretion, which was also markedly stimulated by GIP and glucose administration. There is conflicting evidence concerning the role of insulin in bone metabolism and remodeling. The possibility of an antiresorptive and perhaps even direct anabolic effect in bone is supported by the demonstration of rodent insulin receptors exerting inhibitory effects on osteoclast-like cells (8) and mediating glucose uptake in an osteoblast cell line (9). Nonetheless, short term (2-hour) studies in humans have shown that hyperinsulinemia during euglycemia does not affect markers of bone resorption or formation, whereas hyperinsulinemia during hypoglycemia seems to inhibit bone resorption (1, 10). In contrast, a 4-hour hyperinsulinemic euglycemic clamp caused dose-dependent CTX reductions of 11% to 32% (11). Thus, endogenous insulin secretion and the concomitant glycemic levels may affect bone turnover and explain (some of) the effects observed with GIP. In the current study, we aimed to separate the bone-modulating effects of GIP alone from those of insulin and/or prevailing glucose concentrations on markers of bone turnover.
Materials and Methods
This is a randomized, double-blind, crossover study, consisting of 3 study days from a previously published study (12), combined with 2 additional study days (not previously published) performed in 10 subjects with type 1 diabetes. Thus, the same individuals were studied on 5 different test days performed in randomized order: 3 days with “low glycemia” and coinfusion of GIP, glucagon-like peptide 1 (GLP-1), or placebo (saline) and 2 days with “high glycemia” with either coinfusion of GIP or placebo.
Experimental procedures
To standardize study days, subjects were instructed to eat similarly and maintain a regular diet, as well as to avoid alcohol and strenuous physical activity for 3 days before each study day. Subjects were asked to wear a Guardian® REAL-Time continuous glucose monitor (Medtronic Danmark A/S, Copenhagen, Denmark) for 2 to 4 days before each study day to monitor glycemic variability and facilitate therapeutic decisions in relation to food intake, exercise, and insulin delivery, which would allow them to arrive at the hospital with a plasma glucose of 5 to 9 mmol/L. If it was not possible for a patient to control his morning plasma glucose at 5 to 9 mmol/L without eating or using rapid-acting insulin in the morning, then he was admitted the night before the study days and had his glucose levels monitored during a continuous IV insulin (Actrapid®; Novo Nordisk, Bagsværd, Denmark) infusion. During the study days, patients were placed resting in a bed for 2 hours and had cannulas inserted into both cubital veins for infusions and blood samples, respectively. The arm used for sampling blood was wrapped in a heating pad (∼50°C) for arterialization of the blood. At time 0 minutes, IV infusion of GIP in the dose 4 pmol × kg−1 × min−1, GLP-1 in the dose 1 pmol × kg−1 × min−1, or a matched volume of saline was initiated and continued for the remainder of the study day. Plasma glucose was measured bedside every 5 minutes, allowing the plasma glucose level to be clamped by an adjustable infusion of 20% glucose (weight-to-volume ratio). During high glycemia, plasma glucose was elevated using a bolus infusion of 50% glucose and then clamped at 12 mmol/L. During days with low glycemia, insulin (dose 1 mU × kg−1 × min−1 Actrapid®) was administered in the time interval 0 to 60 minutes, and exogenous glucose was administered (if necessary) to clamp plasma glucose above 2 mmol/L.
Participants
We included 10 male patients (mean ± standard deviation): 26 ± 4 years; body mass index: 24 ± 2 kg/m2; hemoglobin A1c: 7.3 ± 0.8% (57 ± 9 mmol/mol) with type 1 diabetes (positive glutamic acid decarboxylase 65 and/or islet cell antibodies), documented to be without measurable beta cell function, i.e., incremental C-peptide below detection limit (<0.16 nmol/L), following a 5-g arginine stimulation test, performed as previously described (12). The patients were treated with an insulin pump (n = 1) or basal and prandial insulin (n = 9). Eligibility criteria are described at https://www.clinicaltrials.gov/ct2/show/NCT03195257.
Ethics
Oral and written, informed consent was obtained from all participants before inclusion. The study complied with the Declaration of Helsinki (fifth revision, Edinburgh, 2000). The protocol was approved by the Scientific Ethical Committee of The Capital Region of Denmark (reg. no. H-D-2009-0078, with amendment no. 33269), and the study was registered with clinicaltrials.gov (clinical trial reg. no. NCT03195257).
Measurements
CTX was measured using a commercially available sandwich enzyme-linked immunosorbent assay (ELISA) kit, according to the manufacturer’s instruction (Serum CrossLaps® ELISA; Immunodiagnostic Systems Nordic A/S, Copenhagen, Denmark). The ELISA uses highly specific monoclonal antibodies directed against the β-isomerized 8-amino acid sequence EKAHD-β-GGR, derived from the C-terminal telopeptide region of collagen 1. Plasma P1NP was measured using the IDS-iSYS intact P1NP assay (Immunodiagnostic Systems Nordic A/S). Plasma parathyroid hormone (PTH) was measured using the IDS-iSYS Intact PTH assay. Both the P1NP and the PTH assay were carried out on a dedicated automated analyzer, iSYS (Immunodiagnostic Systems Nordic A/S), according to the manufacturer’s instructions. Both assays are chemiluminescence immunoassays. Plasma glucose, insulin, intact GIP, and total GLP-1 were measured, as previously described (12).
Statistical analysis
Results are reported as means ± standard deviation, unless stated otherwise. Integration was carried out using the trapezoidal rule. Plasma levels of hormones and biomarkers were also presented as percent of baseline, where baseline was the mean of two baseline samples taken at −10 and 0 minutes. Potential differences in plasma or serum concentrations of glucose, hormones, and biomarkers of bone turnover were explored with repeated-measures analysis of variance, reporting P values for differences over time or between interventions, i.e., GIP, GLP-1, or placebo; and for the interaction of intervention with time. If a significant interaction between intervention and time was documented (P < 0.05), then values at single time points were compared by the Holm-Sidak posttests, which correct for multiple comparisons. P < 0.05 were considered significant. Statistical evaluation and graphic presentation were made in Prism 7 (GraphPad Software, La Jolla, CA).
Results
Glucose and insulin
Plasma glucose and insulin are shown in Fig. 1. During the 3 matched days with low glycemia, plasma glucose levels were gradually lowered from mean levels of 7 ± 2 mmol/L by an insulin infusion and then raised again (plasma glucose between 3 ± 0.4 and 6 ± 1 mmol/L for 120 minutes; Fig. 1a). During 2 matched days with high glycemia, plasma glucose was clamped at 12 mmol/L for 90 minutes (Fig. 1b). Insulin levels increased similarly during insulin-induced hypoglycemia (i.e., days with low glycemia) and remained at basal substituted concentrations during high glycemia (Fig. 1c and 1d).
Glucose and insulin. Plasma/serum concentrations of (a and b) glucose and (c and d) insulin during insulin-induced (a and c) hypoglycemia or (b and d) hyperglycemia with IV infusion of GIP (♦), saline (□), or GLP-1 (shaded ○). Data are means ± standard error of the mean. Statistical analyses were done with repeated-measures analysis of variance and P-values denote differences over time (A), differences between the interventions (B), and differences owing to interaction between interventions and time (AB).
GIP and GLP-1
Baseline values of GIP were similar, with means of 13 ± 3 pmol/L for all days. During days with GIP infusion, similar mean steady-state concentrations were reached (overall means: 126 ± 26 pmol/L; data not shown). Baseline values of GLP-1 were similar with means of 21 ± 6 pmol/L (all days). During days with GLP-1 infusion, similar mean steady-state concentrations were reached (overall means: 205 ± 55 pmol/L; data not shown).
CTX
During low glycemia, CTX concentrations at baseline were similar (overall means: 439 ± 280 µg/L; Fig. 2a), and GIP increasingly suppressed CTX by up to 59 ± 18%, whereas CTX levels were reduced by 24 ± 10% maximally during placebo infusion (P < 0.0001). During GLP-1 infusions, CTX concentrations were suppressed similarly to the situation on the placebo days (Fig. 2c).
Bone resorption marker—CTX. Plasma concentrations of C-terminal telopeptide of type I collagen (CTX) during insulin-induced (a and c) hypoglycemia or (b and d) hyperglycemia with IV infusion of GIP (♦), saline (□), or GLP-1 (shaded ○). Data are means ± standard error of the mean. Statistical analyses were done with repeated-measures analysis of variance with Holm-Sidak multiple comparisons posttests. Significant differences are indicated as: *P < 0.05 and ***P = 0.0001 to 0.001. P-values denote differences over time (A), differences between the interventions (B), and differences owing to interaction between interventions and time (AB).
During high glycemia, baseline concentrations of CTX were similar (overall means: 375 ± 256 µg/L; Fig. 2b), and GIP suppressed CTX by up to 59 ± 19%, whereas a placebo infusion reduced levels by 7 ± 9% maximally (P < 0.0001 for the difference between GIP and placebo; Fig. 2d).
P1NP
Absolute plasma P1NP concentrations did not differ to a statistically significant degree among GIP, GLP-1, and placebo during “low or high glycemia” days (Fig. 3a–3d). During low glycemia and when expressed as percentage change from baseline, GIP increased P1NP from baseline by 12 ± 8% after 30 minutes compared with 2 ± 7% suppression during saline (P < 0.001), a difference still significant at time 60 minutes (6 ± 10% vs −3 ± 8%; P < 0.04), but with no difference between interventions after 90 minutes (Fig. 3c).
Bone formation marker—P1NP. Plasma concentrations of P1NP during insulin-induced (a and c) hypoglycemia or (b and d) hyperglycemia with IV infusion of GIP (♦), saline (□), or GLP-1 (shaded ○). Data are means ± standard error of the mean. Statistical analyses were done with repeated-measures analysis of variance with Holm-Sidak multiple comparisons posttests. Significant differences are indicated as: *P < 0.05 and ***P = 0.0001 to 0.001. P-values denote differences over time (A), differences between the interventions (B), and differences owing to interaction between interventions and time (AB).
PTH
Plasma PTH concentrations did not differ among GIP, GLP-1, and placebo during low glycemia (Fig. 4a and 4c). During high glycemia, GIP suppressed PTH significantly after 60 minutes, but not after 90 minutes (Fig. 4b and 4d).
PTH. Plasma concentrations of PTH during insulin-induced (a and c) hypoglycemia or (b and d) hyperglycemia with IV infusion of GIP (♦), saline (□), or GLP-1 (shaded ○). Data are means ± standard error of the mean. Statistical analyses were done with repeated-measures analysis of variance with Holm-Sidak multiple comparisons posttests. Significant differences are indicated as: **P = 0.001 to 0.01 and ***P = 0.0001 to 0.001. P-values denote differences over time (A), differences between the interventions (B), and differences owing to interaction between interventions and time (AB).
Discussion
We demonstrate that GIP exerts robust suppression of CTX independently of endogenous insulin secretion and that this occurs irrespective of high or low plasma glucose concentrations. The reductions in circulating CTX within 2 hours (without changes in insulin) suggest that insulin is of lesser importance for the bone-preserving properties of GIP in the immediate postprandial period, where insulin responses are normally boosted by the incretins. Our findings are therefore in line with the study by Clowes et al. (10), demonstrating that short-term hyperinsulinemia-euglycemia does not suppress bone resorption. Perhaps more importantly, our findings add to the increasing body of preclinical and clinical evidence demonstrating a role for GIP in bone turnover and suggesting that GIP is a central part of an entero-osseous axis in humans. This evidence includes the presence of GIP receptors on various osteoblasts (i.e., osteosarcoma) and osteoclast cell lines (13, 14) and the observation that GIP receptor knockout mice have smaller bone size, lower bone mass, and substantial deficiencies in bone microarchitecture and biomechanical properties (15, 16). Furthermore, activation of functional GIP receptors and secondary cyclic adenosine monophosphate accumulation leads to diminished resorptive activity of murine osteoclasts (17). Thus, by taking the present results into consideration, it is likely that the bone-preserving effects described in cell line studies and rodent models actually translate into a physiological function for GIP in humans. This notion is further supported by clinical data showing that functional mutations in the human GIP receptor (also causing a deficient insulin response) are associated with a lower bone mineral density and a 60% increased risk for nonvertebral fracture (18). Interestingly, there is also some evidence suggesting that GLP-1 may act upon immature bone cells, as well as more mature osteoblasts or -clasts to limit resorption and enhance bone mineral density [for a review, see Zhao et al. (19)]. Nonetheless, we did not find any effect of GLP-1, which we used as active control on the low glycemia days.
An effect of GIP and GLP-1 on the levels of P1NP, which is a peptide marker secreted from osteoblastic cells, reflecting the overall rate of osteoblastic bone formation (2), may also have been expected based on the literature. For example functional GIP receptor activation leads to reduced apoptosis and P1NP secretion in human osteoblastic SaOS2 cells (20). SaOS2 cells express GIP receptors but not GLP-1 receptors (14) and are interesting in a translational perspective, as they have been described to have an osteoblastic-labeling profile similar to mature human osteoblasts (21). Likewise, treatment with a GLP-1 receptor agonist for 4 months in aged, ovariectomized rats (22) and for one year in obese, middle-aged women (23) was associated with increased levels of P1NP. Yet, we only found a negligible (and likely clinically insignificant) effect of GIP at normoglycemia and no effect of GLP-1 on P1NP in our experiments. The reason for this is not clear but may relate to the short duration of exposure or the experimental conditions. In relation to this, GIP receptor expression on osteoblast-like cells is not a static phenomenon but rather seems to be downregulated by continued exposure to GIP and upregulated by removal of GIP exposure (5). In our study, circulating PTH concentrations were slightly suppressed by GIP during hyperglycemia after 60 minutes. This finding merits further studies and is hard to interpret without concurrent calcium measurements (which we did not perform), but as a major action of PTH is to cause calcium release from bone, a suppressive effect of GIP on PTH release may be consistent with a net anabolic effect on bone.
There are some limitations to the present data. We present short-term data on indirect biochemical measurements of markers of bone turnover, i.e., CTX and P1NP. Histomorphometric assessment of iliac crest bone biopsies is the gold standard for measuring bone turnover (2), but these markers may provide reliable information on short-term changes in bone turnover that correlate quite well with changes in bone mineral density and clinical outcomes, e.g., fractures on the longer term (24), particularly if intraindividual and diurnal variations are controlled for, as we did in our crossover design with investigations at the same time each experimental day. But, it is important to emphasize that preclinical evidence suggests that GIP needs to be administered as an intermittent injection to prevent bone loss (5), and the observed effect on CTX thus likely represents a temporary effect. Another important issue includes the applied GIP doses, which resulted in concentrations at steady state of ∼125 pmol/L. These are similar to the maximal concentrations that can be reached after a large, liquid mixed meal (25, 26) and at peaks throughout the day (27). Thus, the GIP stimulus should, in many postprandial situations, be considered in the high physiological range. Nonetheless, the physiological postprandial relevance of endogenous GIP and the potential for pharmacological GIP receptor activation in the treatment of osteopenia warrant further investigations.
Conclusion
In patients with type 1 diabetes without endogenous insulin secretion, the hormone GIP increasingly reduced bone resorption (estimated by CTX) during both elevated and low plasma glucose concentrations. GIP also seemed to induce a short-lived stimulatory effect on P1NP at low glycemia and a suppressive effect on PTH at hyperglycemia.
Abbreviations:
- CTX
carboxy-terminal collagen crosslinks
- ELISA
enzyme-linked immunosorbent assay
- GIP
glucose-dependent insulinotropic polypeptide
- GLP-1
glucagon-like peptide 1
- IV
intravenous
- P1NP
procollagen type 1 N-terminal propeptide
- PTH
parathyroid hormone.
Acknowledgments
The authors thank the participating patients and J. Purtoft, N. Kjeldsen, and S. M. Schmidt from the Center for Diabetes Research, Gentofte Hospital, University of Copenhagen (Hellerup, Denmark), and Nadia Quardon at the Department of Clinical Biochemistry, Rigshospitalet, Glostrup, University of Copenhagen (Copenhagen, Denmark) for laboratory assistance.
Clinical Trial Information: Clinicaltrials.gov no. NCT03195257 (registered 22 June 2017).
Author Contributions: M.B.C. contributed to study design, clinical experiments, and the data analysis, as well as drafted the manuscript. A.L. contributed to data analysis. S.C. contributed to the clinical experiments. N.R.J. and J.J.H. contributed with biochemical measurements. F.K.K. contributed to the study design and to data analysis. All authors (M.B.C., A.L., S.C., N.R.J., J.J.H., T.V., and F.K.K.) revised and approved the final manuscript.
Current Affiliation: S. Calanna’s current affiliation is Novo Nordisk A/S, Novo Allé, 2880 Bagsvaerd, Denmark.
Disclosure Summary: The authors have nothing to disclose.
