Separate and Combined Glucometabolic Effects of Endogenous Glucose-Dependent Insulinotropic Polypeptide and Glucagon-like Peptide 1 in Healthy Individuals

Food ingestion leads to secretion of several gut hormones, including the incretin hormones glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide 1 (GLP-1) (1,2). GIP is secreted from enteroendocrine K cells primarily located in the proximal part of the intestines (duodenum and jejunum), whereas L cells secreting GLP-1 are more prevalent in the distal gut (ileum and colon) (3). GIP and GLP-1 contribute to the so-called incretin effect (i.e., the potentiation of postprandial insulin secretion by gastrointestinal factors) (4). The separate contributions of GIP and GLP-1 to the incretin effect have not been quantified, but, based on exogenous hormone infusions, GIP has been suggested as the predominant incretin hormone in healthy subjects in some (4,5) but not in all studies (2,6). In patients with type 2 diabetes, the incretin effect is diminished, contributing importantly to the postprandial hyperglycemia observed in these patients (7,8).

The physiological effects of GLP-1 have been revealed in studies using the GLP-1 receptor (GLP-1R) antagonist exendin(9-39)NH₂ (9,10). Several exendin(9-39)NH₂-based studies have confirmed that endogenous GLP-1 delays gastric emptying, stimulates insulin secretion, decreases glucagon secretion, and lowers appetite (10–12). It has not been possible to study the physiological effects of endogenous GIP in a similar way due to lack of a suitable receptor antagonist. Intravenous administration of GIP has shown that GIP effects include glucose-dependent insulin secretion (during high plasma glucose) (13) and glucagon secretion (during low plasma glucose) (14), decreased bone resorption (15), increased deposition of triacylglycerides in the adipose tissue (16), and increased blood flow to the intestines and adipose tissue (16,17).

Recently, we observed that infusions of the naturally occurring GIP fragment GIP(3-30)NH₂ act as a GIP receptor (GIPR) antagonist (18–21) able to block effects of exogenous GIP on insulin secretion (18) and on adipose tissue (22) in humans. This antagonist will therefore help to elucidate the effects of endogenous GIP. To investigate the physiological effects of endogenous GIP alone and in combination with endogenous GLP-1 on glucose metabolism and glucose-regulating hormones, we studied healthy men during four separate oral glucose tolerance tests (OGTTs) with concomitant infusions of GIP(3-30)NH₂, exendin(9-39)NH₂, the combination of the two, and placebo, respectively. We hypothesized that the two incretin hormones equally contribute to postprandial glucose tolerance and that antagonism of the incretin hormone receptors will eliminate the incretin effect and, thus, exert additive effects on postprandial glucose excursions.

The protocol was approved by the Scientific-Ethical Committee of the Capital Region of Denmark (identification number H-16033104) and the Danish Data Protection Agency (local number HGH-2018–036; I-Suite number 6508). The study is registered at ClinicalTrials.gov (registration number NCT03133741).

Eighteen men (Table 1) were included. Inclusion criteria were age 20–70 years, BMI >19.0 kg/m², hemoglobin A1C <6.5% (48 mmol/mol), and fasting plasma glucose (FPG) <7 mmol/L. Exclusion criteria were use of medication that could not be paused for 12 h, diabetes, first-degree relatives with diabetes, abnormal blood biochemistry (blood hemoglobin and plasma liver enzymes [alanine aminotransferase and ASTs]), plasma creatinine, and urine albumin-to-creatinine ratio. All participants gave written informed consent before inclusion.

Synthetic human GIP(3-30)NH₂ (custom synthesized by Caslo, Lyngby, Denmark) and exendin(9-39)NH₂ (catalog number H-8740; Bachem, Bubendorf, Switzerland) were demonstrated to be >95% and >97% pure, respectively, and identical to the natural peptides by high-performance liquid chromatography, mass, and sequence analysis. Exendin(9-39)NH₂ was dissolved in sodium chloride (9 mg/mL) with 0.2% human albumin (CSL Behring, Marburg, Germany), and GIP(3-30)NH₂ was dissolved in sodium hydrogen carbonate with 0.2% human albumin. After sterile filtration and test for sterility and pyrogens by the Capital Region Pharmacy (Herlev, Denmark), vials were stored at −20°C pending use. On study days, vials were thawed and prepared for infusion under sterile conditions by dilution to a total volume of 250 mL in sodium chloride (9 mg/mL; Fresenius Kabi, Uppsala, Sweden) with 0.2% human albumin. Placebo infusions were 250 mL sodium chloride with 0.2% human albumin.

Each participant served as his own control and attended four 75-g OGTTs on four separate study days in a randomized order with infusion of GIP(3-30)NH₂ (260 min at a rate of 800 pmol/kg/min), exendin(9-39)NH₂ (20 min at 1,000 pmol/kg/min followed by 240 min at 450 pmol/kg/min), GIP(3-30)NH₂ and exendin(9-39)NH₂, or placebo. An uninvolved laboratory technician dissolved the peptides and/or mixed the placebo infusions; thus, the content was unknown for the participants and investigators.

On each study day, the participant arrived after a 48-h period without alcohol consumption or strenuous physical activity and an overnight fast (∼10 h) including liquids. A cannula was placed in a cubital vein in each arm, one for blood sampling and the other for peptide infusions. The hand and forearm for blood sampling were wrapped in a heating pad (45°C) to arterialize the venous blood. At time −20 min, the peptide infusions were started, and at time 0–5 min, 75 g of glucose (the OGTT) plus 1.5 g of acetaminophen dissolved in 300 mL water was ingested. At time 240–270 min, an ad libitum meal of pasta bolognese was served (energy content per 100 g: 147 kcal, 5.9 g fat, 17 g carbohydrates, and 5.6 g protein) and consumed during continuous infusions.

Blood samples were drawn 30, 15, and 0 min before and 15, 30, 45, 60, 90, 120, 180, and 240 min after initiation of the OGTT. For bedside analysis of plasma glucose, blood was collected in sodium fluoride-coated tubes and immediately centrifuged for 30 s (∼7,500g, room temperature). For analysis of GIP, GLP-1, glucagon, pancreatic polypeptide (PP), GIP(3-30)NH₂, and exendin(9-39)NH₂, blood was collected into chilled EDTA tubes to which were added a specific dipeptidyl peptidase 4 inhibitor (valine pyrrolidide, 0.01 mmol/L) (gift from Novo Nordisk, Måløv, Denmark). For analysis of insulin and C-peptide, blood was sampled in dry tubes with serum separator gel and clot activator (silica particles) and left at room temperature for 20 min for coagulation. For analysis of acetaminophen (paracetamol), blood was sampled in tubes with lithium and heparin. All tubes were centrifuged for 15 min (2,900g, 4°C). Plasma and serum samples were stored at −20°C until analysis. Questionnaires about hunger, satiety, fullness, potential meal consumption, thirst, comfort, tiredness, and nausea (10-mm visual analog scales [VASs] on paper) were answered every 30 min. The replies (millimeters) were measured with a ruler. The amount of food consumed during the ad libitum meal was determined by weighing meals and leftovers, and the appearance, smell, taste, off-notes, and overall impression of the meal were assessed by VAS.

Plasma glucose was measured by the glucose oxidase method (model 2300 STAT Plus analyzer; YSI Incorporated, Yellow Springs, OH) bedside. Serum insulin and C-peptide were measured with two-sided electrochemiluminescence assays (Roche/Hitachi Modular Analytics; Roche Diagnostics, Mannheim, Germany). Total GLP-1 (23), total GIP (24), PP (25), GIP(3-30)NH₂ (18), and exendin(9-39)NH₂ (26) were measured in plasma by radioimmunoassays as previously described. Plasma glucagon was measured by ELISA (10–1271–01; Mercodia, Uppsala, Sweden).

Results in the text are reported as mean ± SD and in figures mean ± SEM unless otherwise stated. All calculations of the area under curve (AUC) were based on the trapezoidal rule, and AUC values are reported for the period 0–240 min. To assess the postprandial period ended by plasma glucose levels’ return to baseline, baseline-subtracted AUC (bsAUC) is reported for the period 0–180 min. Baselines are calculated as mean of the −30, −15, and 0 min values or, when only available, −30 and 0 min.

Statistical analyses were performed with GraphPad Prism 7.02 (GraphPad Software, San Diego, CA). A two-sided P value <0.05 was used as significance level. One-way repeated-measures ANOVA (rmANOVA) with Greenhouse-Geisser correction and Tukey multiple comparison were used to test for changes and differences among bsAUC, baseline, peak, and time to peak values. Post hoc, the effects of exendin(9-39)NH₂ infusions on insulin levels were assessed: bsAUC values for insulin were analyzed in SAS Enterprise Guide 7.15 HF3 (SAS Institute Inc., Cary, NC) by a mixed-model rmANOVA with bsAUC values for insulin as fixed effect and study participant as random effect. Each of the following was subsequently included separately in the model as fixed effect: age, BMI, fat mass, glucagon, GIP, GLP-1, the integrated index of β-cell function, or insulinogenic index.

Insulin secretion rate (ISR) values were based on deconvolution of C-peptide concentrations using age, height, weight, and population-based variables for C-peptide kinetics as previously described (27,28). Insulin/glucose, C-peptide/glucose, and ISR/glucose ratios were calculated for each data point and summarized as a time curve. AUCs for these time curves were calculated as described above. Insulinogenic index was calculated as the change in insulin value from baseline to 30 min divided by the change in glucose for the same period (Δinsulin_{0–30 min}/Δglucose_{0–30 min}) (29). An integrated index of β-cell function throughout the postprandial period was calculated as AUC of ISR divided by AUC of glucose (30). β-Cell glucose sensitivity (30) was calculated as the slope of the linear regression of ISR from 0 min to time of peak value (y values) plotted against the glucose concentrations for the same period (x values).

Eighteen healthy men were included in and completed the study (Table 1). One participant had mild vasovagal reactions to the cannulations on all study days, but it resolved quickly and before infusions were initiated. We observed no reactions to the infusions, and the participants did not report any discomfort.

Plasma levels of GIP(3-30)NH₂ reached steady state after 20 min of infusion (time 0 min) with mean concentrations of 57 ± 16 nmol/L during the study day with coinfusion of exendin(9-39)NH₂ and 59 ± 16 nmol/L during coinfusion with saline (Fig. 1A). Exendin(9-39)NH₂ in plasma reached steady state after 50 min (time 30 min) with a mean concentration (time 0–240 min) of 212 ± 117 nmol/L during the coinfusion with GIP(3-30)NH₂ and 194 ± 98 nmol/L during the coinfusion with saline (Fig. 1B).

At baseline, infusions with exendin(9-39)NH₂ alone as well as combined with GIP(3-30)NH₂ caused higher FPG levels than the placebo infusion (Fig. 2 and Table 2). During placebo infusion, the OGTT increased plasma glucose concentrations from 5.3 ± 0.37 mmol/L at baseline to a maximum of 9.3 ± 2.0 mmol/L after 43.3 ± 16 min (Fig. 2). GIP(3-30)NH₂ and exendin(9-39)NH₂ each had higher peak glucose levels of 10.3 ±2.1 mmol/L (P = 0.0101) and 10.6 ± 2.2 mmol/L (P = 0.0006), respectively, during the separate infusions. An additive effect was observed with the combination of GIP(3-30)NH₂ and exendin(9-39)NH₂ (11.8 ± 1.9 mmol/L), which was significantly higher than the three other interventions (Fig. 2 and Table 2). The time to peak glucose levels was significantly prolonged by ∼14 min during both infusion with GIP(3-30)NH₂ + exendin(9-39)NH₂ and infusion with GIP(3-30)NH₂ alone, but was not affected by exendin(9-39)NH₂ alone (Table 2). bsAUC during exendin(9-39)NH₂ was similar to placebo, whereas GIP(3-30)NH₂ caused significantly higher glucose excursions (bsAUC) compared with placebo (P = 0.0012). Infusion of GIP(3-30)NH₂ + exendin(9-39)NH₂ resulted in the highest bsAUC (P < 0.0001 compared with placebo) (Fig. 2B and Table 2) and significantly higher bsAUC than each of the antagonists. Based on fold changes in bsAUC from each participant, GLP-1R antagonism, GIPR antagonism, and the combination increased plasma glucose excursions by 1.9 ± 2.0-, 2.6 ± 3.1-, and 3.6 ± 5.0-fold compared with placebo.

No significant differences in baseline values of insulin, C-peptide, glucagon, or PP were observed. During all four interventions, the OGTT stimulated insulin secretion (Fig. 3A–C), but the response was significantly reduced by infusion with GIP(3-30)NH₂ (Table 2). In 6 of the 18 participants, infusion of exendin(9-39)NH₂ caused significantly higher insulin and C-peptide levels compared with the three other interventions. This phenomenon has been reported in similar studies previously (31–34) and does not seem to be related to age, BMI, fat mass, glucagon, GIP, GLP-1, or the integrated index of β-cell function. However, a significant association was evident for bsAUC of insulin between the insulinogenic index (Fig. 4A and Table 2) calculated from the placebo infusion and the interventions (insulinogenic index × intervention, P = 0.0088) when evaluated by mixed-model ANOVA.

The profiles of insulin (Fig. 3A), C-peptide (Fig. 3B and C), insulin/glucose ratio (Fig. 3D), and ISR/glucose ratio (Fig. 3G) showed similar changes during the four infusions, but the magnitude of responses was significantly reduced during infusion of GIP(3-30)NH₂ alone, and there was no further reduction upon combination with exendin(9-39)NH₂ (Fig. 3A–D and Table 2). However, C-peptide/glucose ratios were significantly reduced compared with placebo during exendin(9-39)NH₂ and GIP(3-30)NH₂, as well as during the two antagonists combined (Fig. 3E and F).

Based on the values of bsAUC for C-peptide/glucose, insulin secretion was reduced by 8.6 ± 16% during exendin(9-39)NH₂ infusion (P = 0.0910), by 30 ± 17% during GIP(3-30)NH₂ infusion (P = 0.001), and by 37 ± 16% during infusion of exendin(9-39)NH₂ + GIP(3-30)NH₂ (P = 0.001) compared with placebo.

During placebo infusion, oral glucose induced a decrease in glucagon that was significantly diminished by the infusions of exendin(9-39)NH₂ and unaffected by GIP(3-30)NH₂ (Fig. 3H and Table 2). PP responses to oral glucose were similar during all four interventions (Fig. 3I and Table 2). As hypoglycemia is a powerful stimulant of PP secretion, the surge in PP after time 180 min was ostensibly due to the decline in plasma glucose, sometimes below fasting levels (35).

β-Cell function, assessed by the insulinogenic index (Δinsulin_{0–30 min}/Δglucose_{0–30 min}), appeared to be mainly affected by GIP(3-30)NH₂, which caused a significant reduction (Fig. 4A and Table 2). Likewise, the β-cell glucose sensitivity (slope of ISR vs. glucose from 0 min to glucose peak) was significantly reduced by GIP(3-30)NH₂ and the combined exendin(9-39)NH₂ and GIP(3-30)NH₂ infusion compared with placebo (Fig. 4B and Table 2). Infusion with GIP(3-30)NH₂ resulted in impaired integrated index of β-cell function during the postprandial period (AUC_ISR/AUC_glucose), which was even more pronounced during coinfusion of GIP(3-30)NH₂ + exendin(9-39)NH₂ (Fig. 4C and Table 2), indicating individual and additive effects of the two incretin hormones.

Plasma concentrations of total GIP were similar during all four experiments (Table 2). The postprandial bsAUC for total GIP was slightly increased during infusion with GIP(3-30)NH₂ compared with both exendin(9-39)NH₂ and placebo infusions (Fig. 5B and Table 2). Infusion with exendin(9-39)NH₂ was the only intervention that resulted in a significantly increased peak concentration of total GIP compared with placebo (P = 0.0108) (Fig. 5A). Baseline values of GLP-1 were similar (Table 2), but during infusions with exendin(9-39)NH₂ bsAUC as well as peak concentrations of total GLP-1 were clearly and significantly increased (Table 2) as previously reported (31–34).

Acetaminophen was undetectable in all baseline samples. Infusions with exendin(9-39)NH₂ accelerated gastric emptying, resulting in ∼11 min earlier and significantly higher peak values (Fig. 6A and Table 2).

The ad libitum meal was evaluated by VAS (0–100 mm) with a mean score for appearance of 73 ± 21 mm, smell 76 ± 15 mm, taste 78 ± 2.0 mm, off-taste 10 ± 16 mm, and overall impression 76 ± 16 mm (n = 17). One study participant refused to consume the ad libitum meal due to the choice of dish. Infusion with exendin(9-39)NH₂ increased the bsAUC for hunger significantly (Fig. 6C), but the amount of food consumed or VAS for satiety, fullness, prospective food consumption, comfort, nausea, or thirst was similar during the four interventions (Fig. 6B and D–I).

In the current study, the separate and combined effects of endogenous GIP and GLP-1 during an OGTT were evaluated by infusions of specific antagonists of the GIPR and GLP-1R, respectively, and we find that the two incretin hormones have additive insulinotropic effects. Furthermore, the data indicate that endogenous GIP may be responsible for a greater proportion of the insulin response to oral glucose than endogenous GLP-1 in healthy men.

The incretin effect is usually estimated by comparing the insulin responses of an OGTT and an isoglycemic i.v. glucose infusion (36). By infusing GIPR and GLP-1R antagonists, we can now confirm that endogenous GIP and GLP-1 increase the insulin secretion in healthy individuals additively and contribute by at least 40% to the total insulin response to a 75-g OGTT. Furthermore, assessment of the effect of endogenous GIP and GLP-1 during OGTT on plasma glucose (bsAUC of glucose) was estimated to be at least 57 ± 15% (bsAUC_{both antagonists}/bsAUC_placebo as percent of placebo), reflecting that 57 ± 15% of the glucose disposal during an OGTT seems to be a result of the incretin hormones GIP and GLP-1.

The contributions of the two hormones to the incretin effect have previously been estimated in studies applying i.v. administration of GIP and GLP-1. Depending on the protocol, data have suggested that the two hormones contribute about equally or that either GIP or GLP-1 explains the majority of the incretin effect in healthy subjects (2,5,6). Based on the current data, it seems that endogenous GIP lowers postprandial glucose levels to a greater extent than endogenous GLP-1 and that, while GIPR antagonism affects glycemic levels during the whole postprandial period, GLP-1R antagonism primarily affects the glucose excursions in the early postprandial phase. Thus, during the exendin(9-39)NH₂ infusions, plasma glucose responses showed increased and significantly delayed peak glucose concentrations (by >10 min and 1.3 mmol/L), but the overall glucose excursion (as assessed by bsAUC) equaled that seen with placebo. In contrast, during the infusion of GIP(3-30)NH₂, the bsAUC for glucose was significantly higher than during placebo [and exendin(9-39)NH₂] infusion, but there was no delay in time to peak of glucose, and the increase in peak concentration of glucose was only 1.0 mmol/L (P = 0.0101 compared with placebo). Combined with the accelerated absorption of acetaminophen during exendin(9-39)NH₂ infusions, these results would support that GLP-1 antagonism primarily affected gastric emptying and secondarily the insulin secretion (4,37).

Because the effects of the receptor antagonists are highly dependent on the degree of receptor blockage, we chose doses that previously resulted in high plasma concentrations (18,31,34). GIP(3-30)NH₂ is pharmacologically well characterized (18–21) and has a Kd value of 3.4 nmol/L (19); the plasma concentrations reached in this study were >15-fold higher. Exendin(9-39)NH₂ has been extensively used as a tool for studying the effects of endogenous GLP-1. Its affinity for the GLP-1R is high and almost similar to that of GLP-1 (9). The plasma concentrations obtained in this study exceed the plasma concentrations of endogenous GLP-1 ∼1,000-fold. Thus, the GLP-1Rs are theoretically fully blocked, whereas the GIPRs are at least 80% blocked (18). However, the mechanism of action of endogenous GLP-1 is complex and possibly involves activation of afferent sensory neurons in the gut and liver, which locally may be exposed to considerably higher GLP-1 concentrations. Furthermore, GLP-1 acts in a paracrine manner within the pancreatic islets (38), and this means that completeness of GLP-1R blockade cannot be estimated from infusion experiments. For GIP, a similar neural signaling mechanism is not known to exist.

The incretin effect is severely reduced in patients with type 2 diabetes (39), most likely due to diminished insulinotropic effect of GIP in these patients (40). GIP(3-30)NH₂ may represent an important tool to delineate this pathophysiological phenomenon and evaluate its contribution to the hyperglycemic state of type 2 diabetes.

GLP-1 reduces energy intake via peripheral and central neuronal pathways (41), and in obese individuals, exendin(9-39)NH₂ has been shown to increase food intake (42). In the current study, we did not observe any difference in amount of food ingested (ad libitum meal) during the different infusions. Nevertheless, in contrast to the infusions with GIP(3-30)NH₂ and exendin(9-39)NH₂ + GIP(3-30)NH₂, respectively, the infusion with exendin(9-39)NH₂ alone did increase hunger (Fig. 6C). This effect might be disturbed by GIPR antagonism or depleted by the higher plasma glucose levels during the infusion of exendin(9-39)NH₂ + GIP(3-30)NH₂. A lack of effect of exendin(9-39)NH₂ on food consumption has been reported in previous studies (42,43) and could be due to simultaneously increased levels of peptide YY (10,30,34,37), stimulating satiety (44). A role for endogenous GIP in the regulation of appetite sensations and food intake has not been identified previously and has not been suspected because of absent effects of exogenous GIP (45–48). Our results show that blockage of the GIPR using GIP(3-30)NH₂ for a period of ∼4.5 h had no effect on appetite sensations (as assessed by VAS) or the amount of food consumed during an ad libitum meal, in agreement with data from previous GIP infusion studies (46,48).

Administration of exendin(9-39)NH₂ to humans may have several effects including changes in gastric emptying, glucose levels, and glucagon levels (10,49,50). We found a paradoxically increased insulin secretion (serum insulin and C-peptide levels) not accompanied by correspondingly decreased glucose levels. Interestingly, we found a significant association with the insulinogenic index of each study participant, indicating that paradoxically high insulin levels are related to low insulin sensitivity. This phenomenon has been reported previously, primarily in healthy subjects undergoing OGTT or mixed-meal test (31–34), and is not present when glucose is administered intraduodenally (51). When evaluating insulin secretion from C-peptide/glucose or insulin/glucose ratios (describing insulin secretion in relation to prevailing glucose concentrations), the stimulation largely disappeared (Fig. 3D–F). This indicates that the increased insulin secretion could be a response to the hyperglycemia arising from exendin(9-39)NH_2–induced stimulation of glucagon secretion and acceleration of gastric emptying. The increased GLP-1 release from the enteroendocrine L cells occurring during exendin(9-39)NH₂ administration is most likely a result of a disturbed feedback loop involving somatostatin-secreting cells (52) and may also involve exendin(9-39)NH_2–induced acceleration of gastric emptying, but is unlikely to play a role in the increased insulin secretion observed during exendin(9-39)NH₂ infusion as GLP-1Rs on the β-cells are blocked. Nevertheless, other changes occurring during GLP-1R antagonism, including peptide YY secretion from L cells and GIP secretion from K cells (a weak response was actually observed in the present experiments [Fig. 5A and Table 2]), may contribute to explain the lack of effect of exendin(9-39)NH₂ on typical GLP-1 effects such as food intake and insulin secretion as previously described (31,43,48).

In healthy men, endogenous GIP and GLP-1 additively contribute to OGTT-induced insulin secretion and glucose tolerance, with GIP apparently having the greatest effect.

Acknowledgments. The authors thank the study participants for loyalty and commitment and Sisse Marie Schmidt and Inass Al Nachar (both from Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen, Gentofte Hospital, Hellerup, Denmark), and Lene Albæk (Department of Biomedical Sciences, University of Copenhagen, Copenhagen, Denmark) for laboratory assistance.

Funding. The clinical studies were conducted at Clinical Metabolic Physiology, Steno Diabetes Center Copenhagen (Gentofte Hospital, Denmark) and supported by the European Foundation for the Study of Diabetes, the Novo Nordisk Foundation, and the Hørslev Foundation.

Duality of Interest. GIP(3-30)NH₂ as a therapeutic agent is protected by intellectual property rights owned by University of Copenhagen (PCT/DK2015/050266). L.S.G., M.B.N.G., and M.B.C. are minority shareholders of Antag Therapeutics. B.H. is a minority shareholder in Bainan Biotech. M.H.J. is an industrial PhD student employed by Antag Therapeutics ApS. A.H.S.-U. is a shareholder in and Chief Executive Officer employed by Antag Therapeutics ApS. N.C.B. is an industrial PhD student employed by Zealand Pharma A/S. J.J.H. is a minority shareholder and board member of Antag Therapeutics and has been a consultant for, served on scientific advisory panels of, and been given speaker honoraria for Novo Nordisk and Merck Sharp & Dohme/Merck. M.M.R. is a minority shareholder in and consultant for Antag Therapeutics, minority shareholder and chair of the board of Bainan Biotech, and consultant for Synklino. F.K.K. has served on scientific advisory panels, been part of speakers’ bureaus for, served as a consultant to, and/or received research support from Amgen, AstraZeneca, Boehringer Ingelheim, Carmot Therapeutics, Eli Lilly and Company, Gubra, MedImmune, Merck Sharp & Dohme/Merck, Norgine, Novo Nordisk, Sanofi, SNIPR Biome, and Zealand Pharma and is a minority shareholder in Antag Therapeutics. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.S.G., M.M.H., S.S., and A.R.L. performed the study. L.S.G., B.H., M.H.J., M.B.N.G., and S.V. performed the radioimmunoassay and ELISA measurements. L.S.G., B.H., A.H.S.-U., M.B.C., J.J.H., M.M.R., and F.K.K. performed the data analysis. L.S.G., A.H.S.-U., N.C.B., M.B.C., T.V., J.J.H., M.M.R., and F.K.K. designed the study and wrote the study protocol. L.S.G. and F.K.K. wrote the manuscript. L.S.G., M.M.H., B.H., M.H.J., M.B.N.G., A.H.S.-U., S.V., S.S., A.R.L., N.C.B., M.B.C., T.V., J.J.H., M.M.R., and F.K.K. critically edited the manuscript and approved the final version. L.S.G. and F.K.K. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.