A ketone monoester drink reduces the glycemic response to an oral glucose challenge in individuals with obesity: a randomized trial

ABSTRACT

Background

Exogenous ketones make it possible to reach a state of ketosis that may improve metabolic control in humans.

Objectives

The main objective of this study was to determine whether the ingestion of a ketone monoester (KE) drink before a 2-h oral-glucose-tolerance test (OGTT) would lower blood glucose concentrations. Secondary objectives were to determine the impact of KE on nonesterified fatty acid (NEFA) concentration and glucoregulatory hormones.

Methods

We conducted a randomized controlled crossover experiment in 15 individuals with obesity (mean ± SD age: 47 ± 10 y; BMI: 34 ± 5 kg/m²). After an overnight fast, participants consumed a KE drink [(R)-3-hydroxybutyl (R)-3-hydroxybutyrate; 0.45 mL/kg body weight] or taste-matched control drink 30 min before completing a 75-g OGTT. Participants and study personnel performing laboratory analyses were blinded to each condition.

Results

The KE increased d-β-hydroxybutyrate to a maximum of ∼3.4 mM (P < 0.001) during the OGTT. Compared with the control drink, KE reduced glucose (−11%, P = 0.002), NEFA (−21%, P = 0.009), and glucagon-like peptide 1 (−31%, P = 0.001) areas under the curve (AUCs), whereas glucagon AUC increased (+11%, P = 0.030). No differences in triglyceride, C-peptide, and insulin AUCs were observed after the KE drink. Mean arterial blood pressure decreased and heart rate increased after the KE drink (both P < 0.01).

Conclusions

A KE drink consumed before an OGTT lowered glucose and NEFA AUCs with no increase in circulating insulin. Our results suggest that a single drink of KE may acutely improve metabolic control in individuals with obesity. Future research is warranted to examine whether KE could be used safely to have longer-term effects on metabolic control. This trial was registered at clinicaltrials.gov as NCT03461068.

Introduction

Research on potential therapeutic effects of the ketone body β-hydroxybutyrate (βHB) has gained substantial attention in recent years (1-3). Hepatic ketogenesis can be achieved via dietary changes, such as fasting or carbohydrate-restriction (e.g., ketogenic diet), resulting in an increase in circulating βHB to concentrations of ∼0.5–5.0 mM, primarily to provide an alternate fuel source for the brain (4-6). Through binding to G protein-coupled receptors and eliciting epigenetic modifications, βHB also possesses several signaling functions that regulate lipolysis (7), inflammation (8), and oxidative stress (9, 10). In addition, classic studies have observed that infusing βHB can lower circulating glucose and nonesterified fatty acids (NEFAs) (11, 12). Thus, raising circulating βHB could have potential therapeutic use in metabolic disease.

The recent development of exogenous ketones, such as the ketone monoester (KE) (R)-3-hydroxybutyl (R)-3-hydroxybutyrate, now makes it possible to reach a state of ketosis without the need for severe dietary alteration or use of invasive methods (13). Using this KE, we recently demonstrated that raising circulating βHB through oral supplementation was effective at reducing glucose and NEFA concentrations during an oral-glucose-tolerance test (OGTT) in young, healthy individuals (14). Compared with the control drink, the glycemic response was reduced by ∼17% and NEFA AUC was reduced by ∼44% (14). These observations, and other studies (15, 16), suggest that individuals with obesity or insulin resistance might benefit from strategies that elevate circulating βHB. However, βHB metabolism is altered with aging, obesity, and insulin resistance, which could limit the ability of ketone supplements to exert these potentially beneficial metabolic effects (17-20). At present, to our knowledge all human studies involving exogenous ketones have examined young, healthy volunteers and, therefore, results may not be applicable to older individuals with obesity or related metabolic impairments.

The main objective of this study (NCT03461068) was to determine whether the ingestion of KE before a 2-h OGTT could lower glucose concentrations in individuals with obesity. Additional objectives were to determine how acute KE ingestion altered NEFA concentrations, glucoregulatory hormones, and estimates of insulin sensitivity. Based on our study using the same oral ketone supplement and protocol (19), we hypothesized that raising βHB through oral KE ingestion before an OGTT would reduce glucose and NEFA AUCs as well as improve estimates of insulin sensitivity. Because infusion of βHB causes a robust (39%) increase in cerebral blood flow (CBF) (21), we also capitalized on this study design to explore heart rate, blood pressure, and common carotid artery (CCA) hemodynamics, which as far as we know have never been measured in humans after KE ingestion.

Methods

Ethical approval

The study conformed to the standards set by the Declaration of Helsinki and all participants provided written informed consent during the initial screening visit. The study was approved by the University of British Columbia Clinical Research Ethics Board (ID H16-01846). In March 2018, the clinical trials registry was modified. The optional substudy that was included in the consent form was eliminated owing to lack of interest from participants. The objective of this substudy was to assess glucose concentrations using a continuous glucose monitor after a single dose of KE consumed before a standardized lunch and dinner.

Participants

Fifteen participants (5 men and 10 women) volunteered for the study. Inclusion criteria required participants to be between the ages of 30 and 65 y and have a BMI (in kg/m²) ≥28 with a waist circumference >102 cm for males and >88 cm for females. Exclusion criteria included 1) taking medication affecting glucose or lipid metabolism; 2) following a low-carbohydrate diet or intermittent fasting protocol; 3) consuming ketone supplements; 4) engaged in intensive exercise training; 5) diagnosed with diabetes or heart disease; or 6) pregnant during the study. Baseline characteristics and the recruitment flowchart are presented in Table 1 and Figure 1, respectively.

TABLE 1.

Baseline characteristics of participants¹

	Overall	Men	Women
n	15	5	10
Age, y	46.6 ± 10.0	42.8 ± 13.9	48.5 ± 7.6
Body weight, kg	96.1 ± 20.9	111.2 ± 13.0	88.6 ± 20.4
Height, m	1.69 ± 0.09	1.80 ± 0.01	1.64 ± 0.06
BMI, kg/m2	33.7 ± 5.1	34.2 ± 3.6	33.5 ± 5.8
Waist circumference, cm	103.4 ± 12.6	108.8 ± 7.8	100.5 ± 14.1
Systolic blood pressure, mm Hg	128 ± 11	134 ± 6	126 ± 11
Diastolic blood pressure, mm Hg	85 ± 8	84 ± 8	86 ± 8
Glycated hemoglobin, mmol/mol	37 ± 6.6	34 ± 4.4	38 ± 7.7
Glycated hemoglobin, %	5.5 ± 0.6	5.3 ± 0.4	5.6 ± 0.7
Fasting glucose, mM	5.6 ± 0.9	5.3 ± 0.5	5.7 ± 1.1
Fasting insulin, pmol/L	153 ± 76	180 ± 95	140 ± 64
HOMA-IR, AU	2.9 ± 1.4	3.3 ± 1.8	2.6 ± 1.2
Fasting nonesterified fatty acids, mM	0.46 ± 0.22	0.36 ± 0.15	0.51 ± 0.24

¹Values are mean ± SD. AU, arbitrary units.

FIGURE 1.

CONSORT flow diagram. CVD, cardiovascular disease; OGTT, oral-glucose-tolerance test.

Baseline testing (visit 1)

Body weight (kg), height (cm), blood pressure (mm Hg), and waist circumference (cm) were measured using standardized methods. Participants were given a 24-h food log to complete on the day before their first experimental condition (visit 2). Participants refrained from exercise and alcohol consumption for 24 h before the experimental trials (visits 2 and 3).

Experimental trials (visits 2 and 3)

Data were collected at the University of British Columbia, Okanagan Campus (Kelowna, Canada) in the Exercise, Metabolism, and Inflammation Laboratory from April 2018 to December 2018. Participants reported to the laboratory after an overnight fast (≥10 h). After 10 min supine rest, assessment of right CCA blood flow shear patterns was performed using a 10-MHz multifrequency linear array duplex ultrasound (Terason t3200, Teratech). Blood pressure and heart rate were taken in the supine position pre– and post–CCA blood flow scans (Omron Healthcare Co. Ltd, Model BP786CANN). An indwelling intravenous catheter (BD Nexiva, Becton Dickinson Infusion Therapy Systems Inc.) was then inserted into an antecubital vein for repeated blood sampling. Blood was drawn into 1 × 4 mL serum tubes and 1 × 2 mL P800 K₂EDTA tubes (BD Vacutainer, Becton Dickinson Infusion Therapy Systems Inc.) for isolation of serum and plasma. Seven intravenous blood draws were performed for each condition. The first collection (−30 min) was immediately after the fasting blood flow measurement. Participants then consumed a KE supplement (ΔG^® from TΔS Ltd; 0.45 mL/kg body weight or 482 mg/kg body weight) in the form of clear viscous liquid mixed with water and fruit-flavored calorie-free artificial sweetener (Mio, Kraft Foods) in a total volume of 150 mL (222 ± 48 kcal, mean ± SD). Immediately afterward, participants were given 20 mL of a calorie-free sports drink (Gatorade G2) in an attempt to remove any remaining flavor. In the control drink condition, participants consumed 140 mL water combined with 10 mL bitter flavor (Symrise, 648352) with the same sweetener followed by the calorie-free sports drink. Participants wore a nose clip while consuming the supplement in both conditions to further mask the drinks. Fifteen minutes later, a second blood flow measurement was taken, followed by the collection of another blood sample (0 min) and the immediate consumption of a 75-g OGTT drink (Thermo Scientific, Fisher Scientific Company). Another 5 blood draws were taken at 15, 30, 60, 90, and 120 min after ingestion of the glucose drink. At each time point, βHB was measured in whole blood using β-ketone strips (Precision Neo; Abbott Laboratories). Two additional blood flow measurements were performed immediately before the 60- and 120-min blood sample collections. After each trial, participants were asked to complete a gastrointestinal distress questionnaire. After the first experimental condition (visit 2), a copy of the 24-h food log completed the day before was provided to them and they were asked to repeat the exact diet over the 24 h before their next experimental condition (visit 3).

In a randomized crossover design (randomization was generated by the research coordinator using the online research randomizer: https://www.randomizer.org/), participants returned to the laboratory (after a ≥10-h fast) ≥48 h later to complete the alternate condition. Adherence to the 24-h diet, exercise, and alcohol guidelines was confirmed. If no significant deviations were observed, the protocol for the third visit was the same as for the previous one, except that participants received the alternate drink. Five female participants were premenopausal and completed both experimental conditions in the follicular phase (between 3 and 9 d after the beginning of their menstrual cycle). Drinks were prepared by the research assistant following the randomization sequence, the order of which was not revealed until after data analyses. Participants and study personnel performing laboratory blood sample analyses were blinded to experimental conditions using coding (A for the first and B for the second visit). Only the researcher preparing the drinks and collecting blood samples was not blinded.

KE drink

The KE drink (ΔG^®) was provided in a bottle by TΔS Ltd as a clear, viscous liquid. The full processes for the production of (R)-3-hydroxybutyl (R)-3-hydroxybutyrate have been described (22). Once ingested, (R)-3-hydroxybutyl (R)-3-hydroxybutyrate is hydrolyzed by nonspecific carboxylesterases and esterases located throughout the gastrointestinal tract, blood, liver, and other tissues, generating d-βHB and (R)-1,3-butanediol. Both metabolites enter the portal circulation, with (R)-1,3-butanediol being further metabolized in the liver by alcohol and aldehyde dehydrogenase to form d-βHB (13, 23).

Blood samples

At each time point, βHB was measured in whole blood (serum tubes) using a Precision Neo ketone monitor and test strips (Abbott Laboratories). Serum tubes then sat at room temperature for 30 min before being centrifuged at 1500 × g for 15 min at 4°C, whereas the P800 K₂EDTA tubes were centrifuged immediately after collection. Serum and plasma samples were then stored at −80°C before blinded batch analyses. Blood metabolites were analyzed using commercially available kits—serum NEFA (Wako Diagnostics HR Series), glucose (Glucose hexokinase, Pointe Scientific Inc.), lactate (Lactate, Pointe Scientific Inc.), and triglycerides (Triglyceride, Pointe Scientific Inc.)—on a Chemwell 2910 automated analyzer (Awareness Technologies). Plasma C-peptide, insulin, glucagon-like peptide 1 (GLP-1), glucagon, and leptin were assessed using a Milliplex MAP Human Diabetes Magnetic Bead Panel (MilliporeSigma) on a Bio-Plex MAGPIX multiplex reader (Bio-Rad Laboratories Inc.). All assays were run in duplicate, except lactate and triglycerides, which were run singly.

CCA measurements

Right CCA blood velocity and vessel diameter were measured using a 10-MHz multifrequency linear array duplex ultrasound (Terason t3200, Teratech). Arterial diameter was measured using B-mode imaging, whereas pulse-wave mode was used to concurrently measure peak blood velocity. The insonation angle (always 60°) was unchanged throughout each test and all images were optimized in accordance with recent standardized guidelines (24). Mean blood flow was determined as half of the time-averaged maximal velocity multiplied by the cross-sectional luminal area for a minimum of 12 cardiac cycles (24). Mean shear rates were calculated using our edge-detection software as 4 times the peak blood velocity divided by vessel diameter (25). Our between-day and within-day CVs for the assessment of CCA diameter were 1.4% and 1.3%, respectively.

There were difficulties in gaining reliable images and blood flow velocity recordings for some participants, due to very deep vessel structures resulting from excess adiposity and/or carotid stenosis/plaque with turbulent blood flow; therefore, 10 participants were included for CCA vascular outcomes.

Visual analog scales and blinding

At the 120-min time point of each experimental condition, participants were asked to complete visual analog scales (VASs) (26) assessing the following 5 symptoms over the previous 2.5 h: nausea, urge to vomit, bloating, belching, and cramps—and to guess whether they had consumed the control or ketone supplement by answering the question, “What condition do you think you were in today?”

Calculations

The oral glucose sensitivity (OGIS) index was computed using the model-based method of Mari et al. (27). Two-hour AUCs were calculated using GraphPad Prism version 8.0.0 (GraphPad Software Inc.) and included time points 0–120 min, except for triglycerides, which included time points 0, 60, and 120 min. In order to account for any changes in basal glucose or hormones that occurred from −30 to 0 min after consumption of the KE or control drink, incremental AUCs were also calculated.

Statistical analysis

Data were analyzed using SPSS version 23 (SPSS Inc.). Normality was assessed using Q–Q plots and Shapiro–Wilk tests within each experimental condition. Square root transformations were used on GLP-1 AUCs to achieve a normal distribution. AUCs were compared between experimental conditions using paired Student's t tests. VAS differences between conditions were assessed using Wilcoxon's Signed Rank test. A linear mixed-effects model including all time points (−30 to 120 min; condition and time as fixed factors and subject as a random factor) was used to determine the treatment effects, except for triglycerides for which only time points −30, 0, 60, and 120 min were used. Significant interactions were followed up with preplanned contrasts comparing the control drink with KE within each time point using Bonferroni corrections for multiple comparisons. Cohen's d effect size was calculated for all of the significant preplanned comparisons. Significance was set at P < 0.05. Data in Table 1 and Table 2 and the figures are presented as mean ± SD, whereas nonparametric data in Table 3 are presented as median with range.

TABLE 2.

MAP and heart rate responses throughout an oral-glucose-tolerance test for both ketone and Ctrl drink conditions¹

	−30 mins		0 min		+60 mins		+120 mins		P values
Variable	KE	Ctrl	KE	Ctrl	KE	Ctrl	KE	Ctrl	Time	Condition	Time by condition
MAP, mm Hg	98 ± 9	99 ± 8	98 ± 12	101 ± 9	97 ± 13	99 ± 10	97 ± 11	98 ± 10	0.25	0.019	0.80
Heart rate, bpm	69 ± 7	68 ± 10	67 ± 9	64 ± 10	73 ± 8*	65 ± 9	73 ± 8*	63 ± 8	0.046	0.001	0.001

¹ n = 15. Values are mean ± SD. A linear mixed-effects model (condition and time as fixed factors and subject as a random factor) was used to determine the treatment effects. Significant interactions were followed up with preplanned contrasts comparing Ctrl drink with KE drink within each time point using Bonferroni corrections for multiple comparisons. *P ≤ 0.001 compared with Ctrl drink within time point, Bonferroni adjusted post hoc. P values < 0.05 are statistically significant. bpm, beats per minute; Ctrl, control; KE, ketone monoester; MAP, mean arterial pressure.

TABLE 3.

Gastrointestinal symptoms assessed at the end of the KE supplement and Ctrl drink experimental trials¹

	Nausea		Urge to vomit		Bloating		Belching		Cramps
	Range	Median	Range	Median	Range	Median	Range	Median	Range	Median
Ctrl	0, 9	0	0, 11	0	0, 22	0	0, 25	0	0, 9	0
KE	0, 5	0	0, 16	0	0, 18	0	0, 5	0	0, 3	0
P values	0.60		0.71		0.34		0.32		0.13

¹ n = 15. Values are median and range in millimeters (0–100). Differences between conditions were assessed using Wilcoxon's Signed Rank test. Ctrl, control; KE, ketone monoester.

In order to calculate sample size, we used the effect size of d = 1.1 for the difference in glucose AUC (primary outcome) based on our previous KE supplement OGTT study (14). With a 2-tailed α level of 0.05 and 90% power, a sample size of 11 was calculated using G*Power (version 3.1.9.3; Heinrich Heine University, Düsseldorf). In order to preserve power and account for dropouts or missing blood samples, we aimed to recruit 15 participants.

Results

Baseline characteristics of the participants are presented in Table 1. The βHB and glucose responses are presented in Figure 2. A significant condition-by-time interaction was found for βHB (P < 0.001), with all time points except −30 min (P = 1.00, d = 0.0) being higher after KE supplementation than after the control drink (all P < 0.001, d = 4.5; Figure 2A). The KE supplement significantly increased βHB AUC compared with the control drink (+1565%, P < 0.001, d = 5.4; Figure 2B). There was a main effect of time (P < 0.001) and condition (P < 0.001) observed for glucose, with glucose being lower in the KE condition (Figure 2C). The primary outcome of glucose AUC was lower in the KE condition than in the control (−11%, P = 0.002, d = 1.2; Figure 2D).

FIGURE 2.

Changes over time and 2-h AUC after a single drink of KE supplement or Ctrl drink. Drinks were consumed in the fasted state followed 30 min later by a 75-g oral-glucose-tolerance test. Linear mixed-effects models (condition and time as fixed factors and subject as a random factor) were used to determine the treatment effects. Significant interactions were followed up with preplanned contrasts comparing Ctrl drink with KE within each time point using Bonferroni corrections for multiple comparisons. AUCs were compared between experimental conditions using paired Student's t tests. (A) βHB. Condition-by-time interaction P < 0.001. *P < 0.001 compared with Ctrl drink within time point, Bonferroni adjusted post hoc (n = 15). (B) βHB AUC. ^‡P < 0.001 compared with Ctrl drink (n = 15). (C) Glucose. Condition-by-time interaction P = 0.354, main effects of time P < 0.001 and condition **P < 0.001 (n = 15). (D) Glucose AUC. ^†P = 0.002 compared with Ctrl drink (n = 15). Dotted lines represent the Ctrl drink and solid lines the KE drink. Data are presented as mean ± SD in changes over time figures and as individual data and mean in AUC figures. Ctrl, control; KE, ketone monoester; βHB, β-hydroxybutyrate.

Insulin and C-peptide responses are presented in Figure 3. Significant main effects of time were found for insulin and C-peptide (both P < 0.001) but there were no significant effects of condition or condition × time interactions (all P > 0.401). There were no significant differences observed between KE and the control drink for insulin AUC (+5%, P = 0.207, d = 0.4; Figure 3B) or C-peptide AUC (−3%, P = 0.499, d = 0.2; Figure 3D). The OGIS index improved by 11% in the KE condition (P = 0.01, d = 0.9; Supplemental Figure 1). Incremental AUC (iAUC) for glucose and insulin revealed similar results to respective total AUC, whereas C-peptide iAUC was reduced (−17%, P = 0.039, d = 0.7; Supplemental Figure 1).

FIGURE 3.

Changes over time and 2-h AUC after a single drink of KE supplement or Ctrl drink. Drinks were consumed in the fasted state followed 30 min later by a 75-g oral-glucose-tolerance test. Linear mixed-effects models (condition and time as fixed factors and subject as a random factor) were used to determine the treatment effects. Significant interactions were followed up with preplanned contrasts comparing Ctrl drink with KE within each time point using Bonferroni corrections for multiple comparisons. AUCs were compared between experimental conditions using paired Student's t tests. (A) Insulin. Condition-by-time interaction P = 0.989, main effects of time P < 0.001 and condition P < 0.401 (n = 14). (B) Insulin AUC (n = 14). (C) C-peptide, condition-by-time interaction P = 0.702, main effects of time P < 0.001 and condition P = 0.990 (n = 14). (D) C-peptide AUC (n = 14). Dotted lines represent the Ctrl drink and solid lines the KE drink. Data are presented as mean ± SD in changes over time figures and as individual data and mean in AUC figures. Ctrl, control; KE, ketone monoester.

NEFA responses are presented in Figure 4. A condition-by-time interaction was observed for serum NEFAs (P < 0.001; Figure 4A). Preplanned contrasts comparing the 2 conditions within each time point revealed significant differences at times 15 min (P < 0.001, d = 1.1) and 30 min (P = 0.003, d = 0.8) but no significant differences at any other time point. NEFA AUC was also decreased by 21% after KE compared with the control drink (P = 0.009, d = 0.8; Figure 4B).

FIGURE 4.

Changes over time and 2-h AUC after a single drink of KE supplement or Ctrl drink. Drinks were consumed in the fasted state followed 30 min later by a 75-g oral-glucose-tolerance test. Linear mixed-effects models (condition and time as fixed factors and subject as a random factor) were used to determine the treatment effects. Significant interactions were followed up with preplanned contrasts comparing Ctrl drink with KE within each time point using Bonferroni corrections for multiple comparisons. AUCs were compared between experimental conditions using paired Student's t tests. (A) NEFAs. Condition-by-time interaction P < 0.001. *P < 0.005 compared with Ctrl drink within time point, Bonferroni adjusted post hoc (n = 15). (B) NEFA AUC. **P < 0.01 compared with Ctrl drink (n = 15). (C) GLP-1. Condition-by-time interaction P = 0.144, main effects of time P < 0.001 and condition ^†P < 0.001 (n = 14). (D) GLP-1 AUC. ^‡P < 0.001 compared with Ctrl drink (n = 14). Dotted lines represent the Ctrl drink and solid lines the KE drink. Data are presented as mean ± SD in changes over time figures and as individual data and mean in AUC figures. Ctrl, control; GLP-1, glucagon-like peptide 1; KE, ketone monoester; NEFA, nonesterified fatty acid.

A significant main effect of condition was observed for GLP-1 (P < 0.001; Figure 4C), lactate (P = 0.038), and glucagon (P = 0.049) (Figure 5), with no significant condition-by-time interactions (all P > 0.14). The KE supplement significantly decreased GLP-1 AUC (−31%, P = 0.001, d = 1.2; Figure 4D) and iAUC (−39%, P = 0.043, d = 0.6, data not shown) compared with the control drink. Compared with the control drink, the KE supplement significantly increased glucagon AUC (11%, P = 0.030, d = 0.7; Figure 5D), whereas no differences were observed for lactate AUC and iAUC (−5%, P = 0.213, d = 0.3; 4%, P = 0.725, d = 0.1, respectively). No differences in triglycerides (P ≥ 0.210; Figure 5B) and triglyceride AUC (+2%, P = 0.741, d = 0.1) between conditions were observed.

FIGURE 5.

Changes over time and 2-h AUC after a single drink of KE supplement or Ctrl drink. Drinks were consumed in the fasted state followed 30 min later by a 75-g oral-glucose-tolerance test. Linear mixed-effects models (condition and time as fixed factors and subject as a random factor) were used to determine the treatment effects. Significant interactions were followed up with preplanned contrasts comparing Ctrl drink with KE within each time point using Bonferroni corrections for multiple comparisons. AUCs were compared between experimental conditions using paired Student's t tests. (A) Lactate. Condition-by-time interaction P = 0.462, main effects of time P < 0.001 and condition ^†P = 0.038 (n = 15). (B) Triglycerides, condition-by-time interaction P = 0.992, main effect of time P = 0.210 and condition P = 0.572 (n = 15). (C) Glucagon, condition-by-time interaction P = 0.826, main effect of time P < 0.001 and condition ^†P = 0.049 (n = 13). (D) Glucagon AUC. *P < 0.05 compared with Ctrl drink (n = 13). Dotted lines represent the Ctrl drink and solid lines the KE drink. Data are presented as mean ± SD in changes over time figures and as individual data and mean in AUC figures. Ctrl, control; KE, ketone monoester.

Blood pressure and heart rate measurements are presented in Table 2. Mean arterial pressure (MAP) was lower and heart rate higher during the KE trial than for the control drink (both P < 0.05). No significant differences in CCA cerebrovascular conductance (P = 0.160) or any other hemodynamic variables were observed (Supplemental Table 1).

No significant differences between conditions for symptoms of nausea, urge to vomit, bloating, belching, and cramps were found (Table 3). Nine participants were able to properly guess the condition they had (control or ketone), whereas 6 participants were incorrect.

Discussion

The objectives of this study were to evaluate the acute impact of an oral KE supplement on OGTT responses in individuals with obesity. We found that a single drink of KE 30 min before an OGTT significantly reduced our primary outcome of glucose AUC with a large effect size. Several secondary outcomes were also reduced with large effect sizes, including NEFA AUC. Glucose AUC was lower with no change in the insulin or C-peptide AUCs, supporting a potential mechanism involving improved insulin sensitivity. Similarly to our work in young, healthy, lean adults (14), GLP-1 AUC was also lower in the ketone condition than for the control drink. Our results support the notion that the improvement in glucose tolerance after the ingestion of KE is not driven by increased insulin secretion but may be related to improved insulin sensitivity, which was estimated in the current study using the OGIS index.

Ketone ester improves glucose tolerance

To the best of our knowledge, this is the first study to directly assess the metabolic effects of exogenous oral ketones in adults with obesity. We observed that glucose AUC decreased by 11% during a 2-h OGTT after consumption of the KE. We have demonstrated that this same KE supplement and protocol improved the glycemic response by ∼17% in young, healthy, lean adults (14). Stubbs et al. (15) also reported that the KE drink raised βHB concentrations to between 1.3 and 4.7 mM and reduced circulating glucose by ∼15% in both the fasted and fed states in young, healthy participants. Binkiewicz et al. (28) reported that βHB infusion reduced glucose, with greater effects in lean than in obese children. A decrease of ∼10% in steady-state glucose was also shown in nonobese and obese adults during a βHB infusion (29). Both studies were conducted in the fasted state and, importantly, the study conducted by Binkiewicz et al. (28) showed an accompanying increase in insulin with ketone infusion, perhaps due to the extremely high concentrations of βHB achieved (i.e., ∼15 mM). In our study, we did not find any significant differences in insulin concentrations between conditions; however, it did appear that insulin (155 ± 74 compared with 207 ± 95 pmol/L) and C-peptide (585 ± 225 compared with 713 ± 291 pmol/L) increased after KE consumption from time −30 to 0 min (both P < 0.01) (Figure 3A, B). The potential for βHB to stimulate insulin secretion in the fasting state has been observed in several studies (14, 15, 30) and it has been suggested by Mikkelsen et al. (31) to occur when βHB is increased rapidly to concentrations >2 mmol/L. The lack of difference in insulin between conditions during the OGTT is likely due to the high insulin secretion induced by the oral glucose load, which overrides the small differences between conditions seen before the OGTT. Nevertheless, insulin secretion does not appear to be the main driver for the observed fall in glucose and NEFAs because 1) no differences in insulin concentrations between conditions were recorded throughout the 2-h OGTT in either of our KE studies (14); and 2) the glucose- and NEFA-lowering effects of βHB are present in individuals with type 1 diabetes and in healthy humans who showed no concomitant changes in (12, 28, 29, 32), or even decreased (31), insulin concentrations with ketone infusion.

The mechanisms underlying the glucose-lowering effects of KE are not entirely clear. The small, yet immediate, increase in insulin secretion after KE ingestion may be sufficient to inhibit hepatic glucose production, which is supported by the fall in fasting glucose concentration seen in this study and others (31). Direct effects of βHB on hepatic glucose output have also been described (11, 31). A reduction in gluconeogenic precursors such as alanine, glycerol, or lactate might also be involved (29). Based on the immediate reduction in NEFA concentrations in the KE condition, it is likely that lipolysis was decreased, reducing the availability of glycerol, as found with βHB infusion (31, 33, 34). Lactate was also significantly lower in the KE condition (main effect of condition P = 0.038), which could influence gluconeogenesis. Insulin and C-peptide were not augmented throughout the OGTT yet GLP-1, which has potent effects on postprandial insulin secretion, was reduced. There is no indication that the KE affects digestion and absorption (13, 15) and studies showing that KE can lower fasting glucose (11, 31) would appear to discount altered gastric emptying as a mechanism, but we cannot rule out this possibility. Finally, it is conceivable that some conversion of βHB to acetoacetate occurs in the liver, which could increase the intracellular NAD(H):NAD ratio, contributing to reduced endogenous glucose production (35).

Ketone ester affects fat metabolism

When elevated, βHB binds to GPR109A, which inhibits adipose tissue lipolysis (7). In the present study, a 21% decrease in NEFA AUC was seen during the OGTT after KE ingestion. Acutely reducing NEFA concentrations with acipimox has been shown to reduce hepatic glucose output (36), whereas lipid infusion stimulates glucose release from the liver (37). Thus, it is likely that the lower concentrations of NEFAs in the KE condition may have contributed to the associated lower glycemic response, although other mechanisms could be involved (36). Contrary to Stubbs et al. (15), who observed a decrease in triglycerides after KE ingestion, no such effects were observed in our study. This could be related to the differences in the macronutrient content of the drinks provided in the 2 studies.

Role of obesity

Based on the HOMA-IR score (2.9 ± 1.4), the participants in this study displayed evidence of insulin resistance. When compared with the same relative amount of KE in healthy, lean adults (19), βHB concentrations in circulation reached a similar peak (3.4 mM compared with 3.2 mM) but stayed elevated longer throughout the OGTT in participants with obesity. Although this could be related to the higher absolute intake, some evidence suggests that βHB utilization is affected by age, obesity, and/or insulin resistance (17-20). Although the participants with obesity experienced a more sustained elevation of βHB in response to the KE supplement, the relative percentage changes in glucose and NEFAs appeared attenuated compared with the young, lean individuals. Potential differences in the maximum change in glucose (23.1 compared with 12.3 mg/100 mL) and NEFAs (875 compared with 625 mmol/L) after ketone infusion between lean and obese children have been observed (28). In participants with type 2 diabetes, sodium acetoacetate infusion altered neither plasma glucose nor hepatic glucose production, but did lower NEFA concentrations compared with saline infusion (38). Future studies are needed to determine the impact of KE drinks in individuals with different levels of insulin resistance.

Impact of ketone ester on hemodynamics

Acute and chronic hyperglycemia reduce CBF (39), but only chronic hyperglycemia decreases cerebral glucose utilization (40). Notably, the reduction in cerebral glucose utilization occurs alongside increased cerebral ketone utilization (41), thereby maintaining overall cerebral oxidative metabolism. Decreased resting CBF, coupled with adverse vascular remodeling (42), likely contributes to the incidence of cerebrovascular disease in patients with obesity and type 2 diabetes; therefore, improving blood flow patterns via KE could prove useful. Indeed, systemic ketone concentrations dictate cerebral ketone uptake and oxidation (31), and acute ketone infusion increased CBF linearly by 30–40% (21, 43). As a proxy index of CBF, assuming no changes in extracranial blood flow (24), we did not observe changes in flow or shear patterns in the CCA. Although CCA flow was unchanged, the KE rapidly increased plasma ketone bodies (≤3 mM), indicating that delivery of βHB to the brain would be greatly increased. However, acute hyperglycemia after the OGTT likely counteracted any increases in CBF and related shear patterns induced by the KE (39).

Another noteworthy hemodynamic observation was the 3- to 4-mm Hg reduction in MAP throughout the OGTT in the KE condition. This was coupled with a baroreflex-mediated compensatory increase in heart rate. Although we cannot find evidence in humans, rodent work indicates that βHB reduces sympathetic tone (44) via a free fatty acid receptor 3 pathway at the level of the sympathetic neurons (45). Such changes in vasomotor sympathetic nervous activity could explain the observed changes in MAP and heart rate after the KE drink and warrant further exploration.

Strengths and limitations

This study was conducted in adults with obesity who were taking neither glucose- nor lipid-lowering medications, so it is unknown if our results will apply to individuals with type 2 diabetes who are taking glucose-lowering agents. We were not powered to assess sex differences, but based on exploratory analysis in healthy, lean adults (14) and the current study, the effect of KE does not appear to differ between males and females. Finally, because no tracers, assessments of tissue-specific redox state [NAD(H):NAD], or measures of gastric emptying were used in our study, we were unable to assess the direct mechanisms responsible for the observed metabolic effects of KE (i.e., reduced hepatic glucose output, lower lipolysis, altered digestion/absorption, etc.). Similarly to past studies examining the impact of whey protein preloading or ketone infusion on glucose control (31, 46-49), we did not use an energy-matched placebo in our design, raising the possibility that calorie intake and not necessarily the increase in βHB concentration contributed to the effects. Our study findings are also limited to the 2-h period after a single drink of KE. Future studies looking at the glucose-lowering effect of repeated KE drinks over the longer term and the potential mechanism of actions are clearly needed in order to optimize and better understand the physiological effects of this supplement.

Conclusions

Consuming an exogenous KE supplement lowered both the glycemic and NEFA responses to an OGTT in individuals with obesity. Our results support the notion that the improvement in glucose tolerance after the ingestion of KE was not driven by increased insulin secretion and was accompanied by a direct effect of βHB on adipose tissue lipolysis. It is possible that the observed metabolic effects of βHB are feedback mechanisms, where βHB partly replaces glucose as fuel while limiting its own synthesis by inhibiting lipolysis. The practical implications of using KE supplementation to lower glucose and mediate favorable changes in hemodynamic function (i.e., reduced blood pressure) are currently unclear. The bitter taste and relatively high cost of the KE drink would need to be weighed against the magnitude of improvement and the currently unknown side effects of chronic supplementation such as change in blood pH, which could have a variety of metabolic effects. Additional, longer-term studies are warranted to test the clinical translation of our findings.

Notes

Supported by Canadian Institutes of Health Research (CIHR) New Investigator Salary Award MSH-141980, 2015–2020 (to JPL), Michael Smith Foundation for Health Research (MSFHR) Scholar Award 16890, 2017–2022 (to JPL), and Heart and Stroke Foundation of Canada Grant-in-Aid G-17-0018639 (to JPL).

TΔS Ltd. provided the ketone ester, ΔG®. The intellectual property covering the uses of ketones and ketone esters is owned by BTG Ltd., the University of Oxford, the NIH, and TΔS Ltd. Should royalties ever accrue from these patents, KC, as co-inventor, will receive a share of the royalties under the terms proscribed by Oxford University.

Supplemental Figure 1 and Supplemental Table 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.

Abbreviations used: CBF, cerebral blood flow; CCA, common carotid artery; GLP-1, glucagon-like peptide 1; iAUC, incremental AUC; KE, ketone monoester; MAP, mean arterial pressure; NEFA, nonesterified fatty acid; OGIS, oral glucose insulin sensitivity index; OGTT, oral-glucose-tolerance test; VAS, visual analog scale; βHB, β-hydroxybutyrate.