Effects of Preceding Ethanol Intake on Glucose Response to Low-Dose Glucagon in Individuals With Type 1 Diabetes: A Randomized, Placebo-Controlled, Crossover Study

Approximately 20% of hospital admissions for diabetes-related severe hypoglycemia are associated with ethanol consumption (1). Individuals with type 1 diabetes are therefore recommended to limit their ethanol consumption, to consume ethanol with meals and snacks, or to reduce their insulin dose to avoid hypoglycemia during and after drinking (2). Whether the ethanol-associated hypoglycemia is caused by missed meals or the glycemic effects of ethanol is not fully understood (3). Ethanol typically induces hypoglycemia 8–12 h after consumption (4), which may be related to inhibition of hepatic gluconeogenesis (5) and aggravated by blunted symptoms of hypoglycemia (6) and impairment of cognitive function (7).

In nonmedical environments, a 1.0-mg glucagon dose is the first-line of treatment for severe hypoglycemia. Glucagon has, however, been proposed to be inefficient in the treatment of ethanol-associated severe hypoglycemia (8,9). Glucagon primarily stimulates hepatic glycogenolysis, whereas ethanol suppresses hepatic gluconeogenesis (10). Thus, hepatic glucose production during ethanol intoxication is driven by glycogenolysis and might eventually lead to hepatic glycogen depletion (11). Consequently, glucagon may be ineffective in restoring euglycemia after ethanol intoxication due to empty glycogen depots.

Low-dose glucagon as an adjunct to intensified insulin therapy has shown promising results in treatment and prevention of mild hypoglycemia (12,13), especially in closed-loop studies with automatic delivery of insulin and glucagon (14). However, due to ethanol’s potential impairment of glucagon efficacy, participants in some of those studies were only allowed a daily ethanol consumption of a maximum of two to three drinks (15).

We hypothesized that the ability of low-dose glucagon to increase plasma glucose would not be impaired in individuals with type 1 diabetes after ethanol consumption compared with no ethanol consumption. Thus, we compared the glucose-restoring effect of subcutaneous (s.c.) low-dose glucagon during insulin-induced mild hypoglycemia in late hours after a meal with ethanol (0.8 g/kg) or without ethanol consumption.

We performed a double-blinded, randomized (1:1), placebo-controlled, crossover study. The study included a screening day, a run-in period, and two overnight study visits: one visit with ethanol consumption and one visit without ethanol consumption in random order with an interval of at least 7 days (Supplementary Fig. 1). The study was conducted in accordance with the Declaration of Helsinki, registered at ClinicalTrials.gov (NCT02881060), and approved by the Regional Committee on Health Research Ethics, Hillerød, Denmark (H-16027080) and the Danish Data Protection Agency, Copenhagen, Denmark.

Participants were recruited from the outpatient diabetes clinic at Hvidovre Hospital at University of Copenhagen and gave written informed consent to participation. Eligibility was determined after the screening day, where participants who had fasted came to the research facility for a physical examination and for blood sampling; samples were analyzed for HbA_1c, kidney biomarkers, and lipid profiles (Supplementary Table 1). Questionnaires to identify hypoglycemia unawareness (16–18), alcohol dependency (19), and alcohol dehydrogenase insufficiency (20) were completed. Further, participants were tested for orthostatic hypotension (21), beat-to-beat variations (21), and for pancreatic β-cell capacity by an intravenous glucose-arginine stimulatory test (22). Participants were included if they were 18–70 years of age, had type 1 diabetes ≥3 years, had BMI of 20–28 kg/m², used an insulin pump ≥1 year, used carbohydrate counting and bolus calculators in their insulin pumps for all meals and corrections, consumed <14 alcoholic drinks per week, and had at least once consumed ≥4 alcoholic drinks of 12 g ethanol on a single occasion within the last year. Key exclusion criteria were intolerance or allergies to glucagon, lactose, and ethanol, history of gastroparesis, pheochromocytoma, or sleep disorders, symptoms of alcohol dehydrogenase deficiency, liver failure, and history of alcoholism or drug abuse.

After the screening, eligible participants were randomly assigned (1:1) in blocks (2 or 4) to start with the ethanol visit or the placebo visit. The person responsible for the randomization list, generated by www.sealedenvelope.com, prepared the drinks for each visit but had no further involvement in the study. The drinks were isovolumetric, isotonic, and served in sealed transparent jugs to minimize the chances of spotting differences between the drinks. Participants ingested the drinks using straws.

Before the first study visit, participants’ insulin pump settings (basal insulin rates, insulin action time, insulin correction factors, and carbohydrate-to-insulin ratios) were optimized during 2–3 weeks (23). Furthermore, participants were instructed to avoid ethanol consumption and strenuous exercise as well as hypoglycemia (continuous glucose monitor value <3.5 mmol/L or capillary meter glucose <3.9 mmol/L) 24 h before arrival. If these requirements were not met, the study visit was postponed by ≥2 days. On each overnight study visit, participants were admitted to the research facility from 5:30 p.m. to the next morning. They were instructed to arrive having fasted for 5–6 h, aiming for plasma glucose (PG) of 5.0–10.0 mmol/L. A cubital intravenous catheter was placed to draw blood samples and was covered by a heating pad.

After initial blood samples were collected at 6:00 p.m., a standardized dinner (basmati rice, roasted chicken breast, and Uncle Ben’s Tikka Masala sauce: 1 g carbohydrate/kg, 50% carbohydrate, 30% protein, and 20% fat) was served with diet lemonade plus 0.8 g ethanol/kg Skyy vodka (Campari Group, San Francisco, CA) or diet lemonade plus water of equal volume and tonicity. The dinner had to be eaten within 20 min, and the drink had to be consumed within 60 min. A meal insulin bolus (NovoRapid; Novo Nordisk, Bagsværd, Denmark) was administered s.c. using the bolus calculator in the pump aiming for a PG of 6.0 mmol/L.

At 10:00 p.m., we reduced the basal insulin infusion rate to 90% on both study visits to reduce the risk of nocturnal hypoglycemia. Participants were then instructed to sleep until blood ethanol was estimated to be fully metabolized, regardless of the study visit (24) (see the modified Widmark formula in the Supplementary Data). If PG <4.5 mmol/L during sleep, a 50% glucose bolus was given intravenously to raise the PG to 6.0 mmol/L (25). An Alco-Sensor FST breathalyzer (Intoximeter) was used to confirm that the ethanol concentration at the estimated time was undetectable. We calculated that blood ethanol should be undetectable 480–540 min after the consumption on both visits (Supplementary Data).

Once the breathalyzer confirmed undetectable ethanol concentrations, the basal insulin infusion rate was reset to 100%, and an insulin bolus was given s.c. to induce a mild hypoglycemia. The insulin bolus size was calculated to lower the actual PG level to 3.0 mmol/L based on the individual’s insulin correction factor. In addition, the amount of insulin not given during the period of basal insulin rate reduction, equal to 10% of basal rate, was given. We accounted for the reduced insulin rate because insulin depots most likely would be reduced when the insulin bolus was given (26). When PG reached 3.9 mmol/L (t = 0), 100 µg glucagon (GlucaGen; Novo Nordisk) was given s.c., followed by another 100 µg glucagon 2 h later (t = 120) (Supplementary Fig. 2). Two glucagon boluses were given to determine whether the responses to the first and second glucagon bolus would differ between visits.

Throughout the study visits, blood samples, blood pressures, and heart rates were taken at prespecified times. Severity of hypoglycemic symptoms (Edinburgh Hypoglycemia Scale), side effects to glucagon (visual analog scales), and cognitive performance (Digit Substitution Span test, Stroop test, and the Trail Making B test) were assessed. Blood samples were analyzed for PG, plasma ethanol (Cobas 6000/8000; Roche Diagnostics), plasma glucagon, serum insulin aspart, serum free fatty acids (FFAs), serum triglycerides (TGs), plasma ketones, serum growth hormone (GH), and serum cortisol by assays previously described (21,27).

Primary outcome was the incremental peak PG caused by the first 100 µg glucagon bolus. Secondary outcomes were the overall glucose response to first and second glucagon bolus, changes in plasma glucagon, serum insulin, GH, cortisol, FFAs, TGs, and plasma ketones as well as intensity of hypoglycemia symptoms and cognitive performance. Safety issues regarding glucagon treatment (nausea, headache, tiredness) were assessed with visual analog scales.

The study required 11 participants be included to be 90% sure that the lower limit of a two-sided 95% CI was above the noninferiority limit of −1.0 mmol/L, based on the assumptions that the SD for incremental plasma glucose (between and within subjects) was 0.8 mmol/L (21,27). Twelve participants were included to account for study visit sequence. We did not account for drop-outs in the power calculations but continued recruitment. The difference of the time course of PG, counterregulatory hormones, and metabolites was tested with a linear mixed model with random effects. Repeated-measures ANOVA with a compound symmetry covariance structure was used to test the differences between the two study visits, adjusting for the study order. If data had a skewed distribution, logarithmic transformations were used in the ANOVA. If the transformations could not normalize the distribution, data were regarded as nonparametric and analyzed untransformed with Wilcoxon signed rank tests. Spearman correlations were used to determine relationships. A P value of <0.05 was considered statistically significant. Nonpredefined outcomes were Bonferroni adjusted for multiple comparisons within each parameter, meaning that the P values were multiplied with the number of tests for each parameter. The adjusted and unadjusted P values are presented in the tables. Data are presented as median (interquartile range) or as mean ± SEM if not otherwise stated.

A total of 13 participants were screened. One participant withdrew due to needle phobia during the first study visit. Twelve participants (six women) completed both overnight study visits (six started with the ethanol visit); they were 37 (31–51) years of age, had type 1 diabetes for 16 (13–23) years, had an HbA_1c of 57 (51–59) mmol/mol or 7.3% (6.8–7.5), had a BMI of 23.9 (21.9–24.6) kg/m², used insulin pumps for 6 (6–9) years, and had a total daily insulin dose of 37 (31–44) IU. Their usual alcohol consumption was 9 (6–10) drinks per week and had at least in one setting tolerated 6 (5–9) drinks within the last year. No participants reported hypoglycemia unawareness, but partial hypoglycemia unawareness was identified in two participants by the Pedersen-Bjerregaard method, three participants by the Gold methods, and four participants by the Clarke method (28). No participants had orthostatic hypotension, but one participant had impaired beat-to-beat RR variation. Residual β-cell function was negligible in all participants (stimulated C-peptide level <60 pmol/L). Biochemical and hematological markers were within normal ranges at screening (Supplementary Table 1).

Participants had similar preprandial plasma glucose levels at the placebo visit compared with the ethanol visit (5.6 ± 0.48 vs. 6.7 ± 0.76 mmol/L, P = 0.3). The postprandial glucose response, measured as peak PG, area under the curve (AUC), and PG level 3 h after the meal were similar between the two visits. During sleep from 180 to 480 min, the PG levels did not differ between the visits. Glucose was intravenously infused in five participants during the ethanol visit compared with six participants during the placebo visit to avoid nocturnal hypoglycemia. The total amount of glucose infused tended to be slightly lower on the ethanol visit than on the placebo visit (0.1 ± 0.03 vs. 0.27 ± 0.08 g glucose/kg body weight, P = 0.08) (Fig. 1 and Table 1).

Insulin was administered at a slightly lower PG level at the ethanol visit compared with the placebo visit (7.1 ± 0.4 vs. 8.1 ± 0.8 mmol/L, P = 0.15), and the time to achieve hypoglycemia was almost 60 min shorter at the ethanol visit compared with the placebo visit (127 ± 17 vs. 186 ± 15 min, P = 0.002). Glucagon was administered at similar PG levels (3.9 ± 0.1 vs. 3.7 ± 0.1 mmol/L, P = 0.07). The first glucagon bolus tended to produce a lower incremental peak PG (primary end point: 2.0 ± 0.5 vs. 2.9 ± 0.3 mmol/L, P = 0.06), a lower incremental AUC (87 ± 40 vs. 191 ± 37 mmol/L × min, P = 0.08), and a lower 2-h PG level (3.6 ± 1.0 vs. 4.9 ± 0.4 mmol/L, P = 0.05) on the ethanol visit compared with the placebo visit. The second glucagon bolus gave similar incremental glucose responses, with no differences between visits (Table 2). However, the PG level from 0 to 120 min after the second bolus remained, in absolute values, 1.8 ± 0.7 mmol/L lower on the ethanol visit compared with the placebo visit.

On the ethanol visit, participants drank from 47.2 to 72.8 g ethanol (∼4–6 drinks) within 1 h, resulting in a peak ethanol concentration of 1.0 ± 0.1 g/L at 81 ± 23 min after the study start (Fig. 1). All participants could distinguish the ethanol from the nonethanol drink, whereas the blinded investigator (A.R.) was only 50% correct.

The insulin bolus was administered once breath ethanol concentrations were undetectable after the scheduled time of full ethanol metabolism at 8 h. In five participants, the concentrations remained detectable after 8 h and became undetectable after a median of 8.5 (range 8–9) h on the ethanol visit. Plasma ethanol concentrations were zero from time of the insulin bolus to the end of both study visits.

The meal insulin bolus size did not differ between the ethanol visit and the placebo visit (7.9 ± 0.7 vs. 8.2 ± 1.0 IU, P = 0.93). The overall profiles for exogenous insulin and endogenous glucagon concentrations did not differ between the visits (Fig. 1). However, the AUC for glucagon increased during sleep on ethanol visit compared with placebo visit (Table 1). Otherwise, no differences were observed regarding the AUC for insulin and glucagon between the study visits (Tables 1 and 2).

The insulin doses required to induce hypoglycemia were similar on the ethanol visit and the placebo visit (2.5 ± 0.4 vs. 2.7 ± 0.4 IU, P = 0.7). Concentrations of insulin were similar at the time of hypoglycemia at both visits. The AUC for insulin concentrations after the first and the second glucagon injection did not differ between visits (Table 2). Plasma glucagon concentrations were similar before the first glucagon administration (3.3 ± 0.7 vs. 2.3 ± 0.6 pmol/L, P = 0.7). The administered 100 µg glucagon was equivalent to 1.4 ± 0.1 µg glucagon/kg. The plasma glucagon incremental peak and AUC after the first and the second glucagon bolus did not differ between study visits (Table 2).

Preprandial levels of cortisol and GH were similar between study visits. At the ethanol visit, the postprandial GH response tended to be lower compared with placebo (AUC 16 ± 26 vs. 325 ± 185 ng/mL, P = 0.05). However, the postprandial cortisol peak was significantly higher on the ethanol visit than on the placebo visit (216 ± 17 vs. 167 ± 32 nmol/L, P = 0.04). No significant differences in cortisol or GH profiles were otherwise observed between the study visits (Table 1). At 3 h after meal and ethanol consumption, GH levels were significantly reduced compared with the placebo visit (0.67 ± 0.5 vs. 6.2 ± 2.1 ng/mL, P = 0.007). The GH level remained lower during sleeping phase on the ethanol visit compared with the placebo visit (AUC 445 ± 142 vs. 2,393 ± 502 nmol/L × min, P = 0.0003) (Fig. 2 and Table 1).

At the time of the insulin bolus, cortisol levels were significantly higher at the ethanol visit compared with the placebo visit (277 ± 23 vs. 142 ± 17 nmol/L, P = 0.0001). However, cortisol levels were similar at the time of hypoglycemia and remained similar for the rest of the study visits (Fig. 2). In contrast, GH levels were similar immediately before the insulin bolus but showed a significantly increased response after the first glucagon dose compared with the placebo visit (AUC_0–240 1,002 ± 215 vs. 611 ± 125 ng/mL × min, P = 0.048) (Table 2).

FFA, TG, and ketones did not differ between study visits with regard to fasting level and peak level from 0 to 3 h after meal intake (Table 1). However, the postprandial TG and ketone levels were higher at the ethanol visit compared with the placebo visit. Furthermore, the 3-h FFA levels were significantly lower after ethanol intake than after placebo intake, whereas the 3-h TG and ketone levels were significantly higher after ethanol intake. The differences in FFA and TG levels remained throughout sleeping period. The ketone levels increased less during sleeping phase on the ethanol visit compared with the placebo visit (Fig. 2 and Table 1).

Before insulin was given, the FFA and ketone levels were significantly lower at the ethanol visit compared with the placebo visit. In contrast, TG levels were significantly higher before the insulin bolus at the ethanol visit compared with the placebo visit. The differences diminished after the insulin bolus, resulting in similar levels of FFA, TG, and ketones at the time of hypoglycemia (Table 2). However, after the first glucagon administration, FFA and ketone levels were significantly reduced at the ethanol visit compared with placebo visit. The TG levels were similar after the glucagon administrations (Table 2).

The subjective intensity of hypoglycemia and the cognitive performance did not differ at the measured times between study visits. Noteworthy, when ethanol was fully metabolized, participants scored equally on the cognitive performance test between study visits (Supplementary Fig. 3). No differences were observed between study visits regarding cognitive performance before, during, and after insulin-induced hypoglycemia (Supplementary Fig. 3).

No differences were observed between the study visits regarding visual analog scale registrations for side effects before, during, and after the insulin-induced hypoglycemia or after each of the glucagon injections. Furthermore, blood pressure and pulse did not differ between study visits (Supplementary Fig. 4).

We found that in individuals with type 1 diabetes, a s.c. 100 µg glucagon bolus after ethanol intake tended to cause a lower incremental glucose peak and a lower overall glucose response compared with placebo, with P values ranging from 0.05 to 0.07. The glucose response to a second 100 µg glucagon bolus, 2 h after the first bolus, did not differ between visits or from the glucose response to the first glucagon bolus (Table 2).

Notably, even though similar insulin doses were given, the time to achieve hypoglycemia was significantly shorter after ethanol compared with the placebo intake. The insulin levels at hypoglycemia were similar between study visits, even though plasma glucose level was 1.0 mmol/L lower on the ethanol visit compared with the placebo visit. However, no correlation with the glucose response to glucagon was found for the time to achieve hypoglycemia (r = 0.16, P = 0.44) or the insulin level at hypoglycemia (r = −0.32, P = 0.1). In this study, we have no direct measurement of insulin sensitivity; therefore, whether postethanol intoxication may enhance insulin sensitivity and thereby indirectly attenuate the glucose response to glucagon remains uncertain. To our knowledge, no studies have investigated the late effects of ethanol on insulin sensitivity in individuals with type 1 diabetes.

Participants in this study achieved a plasma ethanol peak of 1.0 g/L within 75 min after consuming four to six drinks. We chose this relatively high ethanol intake to achieve the full effect of ethanol and yet avoid the symptoms of severe ethanol intoxication (i.e., vomiting, confusion, seizures, and unconsciousness), which could have biased the glucose response (29). Whether the glucagon effects are influenced by ethanol in a dose-related manner has not been investigated.

The study was designed to determine the glucose-restoring effects of glucagon after evening and overnight ethanol intoxication. Participants were admitted to the clinic overnight to ensure that the study visits were comparable in avoidance of hypoglycemia, activity level, sleep quality, and amount of meal and drink intake. The study was therefore not designed to investigate the effects of ethanol on glucose metabolism before the glucagon administrations (30–33). We chose to administer two 100 µg glucagon boluses to determine whether the first glucagon bolus would impair the glucose response to the second glucagon bolus on the ethanol visit compared with the placebo visit. Castle et al. (34) showed that the glucagon response to an eighth glucagon bolus did not differ from the first bolus. Likewise, the current study showed that the glucose response to a second glucagon bolus were similar to the first bolus, despite the prior ethanol consumption.

Ethanol intake did not affect postprandial plasma glucose excursions from 0 to 180 min. During the “sleeping phase,” a minor nonsignificant reduction in plasma glucose was observed before induction of hypoglycemia. However, the intravenous glucose infusions that were given to avoid hypoglycemia, despite reduction in basal insulin infusion rate to 90% on both study nights, disturbed the interpretation of ethanol effects on nocturnal glucose metabolism. At least, the amount of glucose infusions overnight did not differ between study visits.

Glucagon and cortisol levels were not affected postprandially but increased at the end of sleep after ethanol intake compared with placebo intake (4). On one hand, this may be explained by a counterregulatory response to the lower plasma glucose level on the ethanol visit (33). On the other hand, GH levels were significantly suppressed during the sleep phase compared with the placebo visit, which in previous studies has been related to a suppression of the dawn phenomena and the risk of hypoglycemia (4,35,36).

The nocturnal levels of FFA and ketone bodies were lower at the ethanol visits. This may be related to ethanol-induced suppression of fat oxidation (37). Once insulin was administered, the elevated levels of FFA and ketones decreased to the same level as during the placebo visit. Interestingly, the increase in FFA after glucagon administration was suppressed at the ethanol visit. We can speculate that this reduced elevation of FFAs partly explains the attenuated glucose response to glucagon via reduction in gluconeogenesis (38).

The serum TG profile was opposite that of FFA, with slightly higher nocturnal TG level on the ethanol visit than on the placebo visit. The metabolism of ethanol is known to increase acetate levels that may have increased TG synthesis (39). However, no differences were seen on TG levels after glucagon administration at the two visits.

To our knowledge, this is the first study that investigates the effects of low-dose glucagon after preceding ethanol intake in individuals with type 1 diabetes. A recent unpublished study in fasting individuals with type 1 diabetes demonstrated that the hyperglycemic effect of 50 µg glucagon was not impaired despite a breath ethanol content corresponding to a blood concentration of 1.0 g/L (as determined by breath tests) (40). We chose to focus on the late effects because the ethanol-associated hypoglycemia typically occurs 8–12 h after ethanol intake (3). Ethanol and glucagon affect two different pathways of hepatic glucose production, namely gluconeogenesis and glycogenolysis, respectively (10). Thus, initially, when liver glycogen is available, the glucose-restoring effect of glucagon may not be influenced by ethanol consumption. However, the glucose response to glucagon may be attenuated in the late phase after ethanol-induced glycogen depletion.

The use of low-dose glucagon as an add-on to intensified insulin therapy has shown promising results and been suggested as a new treatment approach for type 1 diabetes (12). This is especially promising in studies with dual-hormone closed-loop systems that automatically deliver insulin and glucagon in response to a measured glucose value (41). One of the main critical points for the dual-hormone systems is the potential safety and efficacy issues regarding low-dose glucagon use. First, no long-term studies with glucagon have confirmed the safety of continuous use of glucagon. Second, glucagon has proven not to be efficient in treating mild hypoglycemia during high serum insulin concentrations (42). Finally, 1 week of a low-carbohydrate diet may also impair the effect of glucagon to restore plasma glucose (27). Our study provides additional clinical information on glucagon efficacy by demonstrating that glucagon is capable of treating insulin-induced mild hypoglycemia after preceding ethanol intake, although the response may be slightly attenuated.

This study has some limitations. First, the ethanol drink could not be blinded for the participants, all of whom correctly distinguished the ethanol visits from the placebo visits. The blinding failure may have affected the subjective measurements of cognitive function, hypoglycemia awareness, and side effects. Nevertheless, we did not find any differences in these parameters between the study visits during hypoglycemia and 1 h after glucagon administration (Supplementary Fig. 3). Second, it may be argued that the study was underpowered. Although we had calculated that 11 participants would be needed to provide sufficient power, the SD for and the incremental plasma glucose peak (primary outcome) was larger than expected (21,27). Furthermore, we found a difference of 0.9 mmol/L, which was lower than our prespecified clinically relevant difference between study visits. Thus, the study was not powered to see a difference of 0.9 mmol/L, which after all may be of clinical relevance. In addition, we made multiple comparisons that primarily should be assessed as explorative and hypothesis-generating data. Third, we allowed glucose and insulin concentrations to be variable during the preglucagon period. We chose this study design to mimic real-world settings and to minimize prior glucose infusion that may bias the late effect of ethanol on glucagon efficacy. Finally, we only included participants who were insulin pump users, in good glycemic control, and with regular ethanol intake but without known ethanol abuse. The strengths of the study are that the amounts of glucose and insulin administered and that the glucose level before glucagon administration were similar between the study visits.

In conclusion, the glucose response to glucagon tended to be reduced 8–9 h after ethanol intake compared with placebo. However, after ethanol intake, 100 µg glucagon could still increase plasma glucose by 2.0 mmol/L from mild hypoglycemia. Low-dose glucagon may therefore be a reasonably effective treatment option for mild hypoglycemia after preceding ethanol intake in both closed-loop and open-loop systems.

Acknowledgments. The authors thank the study participants and acknowledge Alis Sloth Andersen (Department of Endocrinology, Hvidovre Hospital, University of Copenhagen, Hvidovre, Denmark) for assistance with randomization. Furthermore, the authors thank Lene Brus Albæk (Department of Biomedical Sciences, University of Copenhagen, Copenhagen, Denmark) for glucagon analysis, Reingard Raml (Joanneum Research, Graz, Austria) for the insulin aspart analysis, Jette Nymann (Department of Biochemistry, Hvidovre Hospital, University of Copenhagen) for the analysis of cortisol, GH, ethanol, TGs, and FFAs, and Gitte Kølander Hansen (Department of Obesity Biology, Novo Nordisk A/S, Måløv, Denmark) for the ketone body analysis.

Funding. This work was funded by a research grant from the Danish Diabetes Academy supported by the Novo Nordisk Foundation, the Danish Diabetes Association, and the Poul and Erna Sehested Hansen Foundation.

Duality of Interest. K.N. serves as an adviser to Medtronic, Abbott, and Novo Nordisk; owns shares in Novo Nordisk; has received research grants from Novo Nordisk, Zealand Pharma, and Roche; and has received fees for speaking from Medtronic, Roche, Rubin Medical, Sanofi, Zealand Pharma, Novo Nordisk, and Bayer. I.I.K.S. has received a research grant from Zealand Pharma and has received fees for speaking from Rubin Medical and Roche Diabetes Care. T.R.C. works for Novo Nordisk A/S and owns shares in Novo Nordisk A/S and Zealand Pharma A/S. J.J.H. has consulted for Merck Sharp & Dohme, Novo Nordisk, and Roche. S.M. has served as a consultant or adviser to Amgen, AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, Intarcia Therapeutics, Johnson & Johnson, Merck Sharp & Dohme, Novo Nordisk, Novartis Pharma, and Sanofi; has received a research grant from Novo Nordisk; and has received fees for speaking from AstraZeneca, Bristol-Myers Squibb, Eli Lilly, Merck Sharp & Dohme, Novo Nordisk, Novartis Pharma, and Sanofi. S.S. serves on the continuous glucose monitoring advisory board for Roche Diabetes Care and served as a consultant for Unomedical. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. A.R. analyzed data and wrote and edited the manuscript. A.R., K.N., and S.S. conceived the idea and designed the study. A.R. and R.T. performed the study. K.N., R.T., I.I.K.S., T.R.C., J.J.H., and S.M. provided data analysis and reviewed and approved the final manuscript. S.S. interpreted the data and reviewed, edited, and approved the final manuscript. A.R. and S.S. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of the study were presented as a poster at the 77th Scientific Sessions of the American Diabetes Association, San Diego, CA, 9–13 June 2017.