The ketone bodies (KBs) D-3-hydroxybutyrate (D-3HB) and acetoacetate (AcAc) play a role in starvation and have been associated with insulin resistance. The dose–response relationship between insulin and KBs was demonstrated to be shifted to the right in type 2 diabetes patients. However, KB levels have also been reported to be decreased in obesity. We investigated the metabolic adaptation to fasting with respect to glucose and KB metabolism in lean and obese men without type 2 diabetes using stable glucose and D-3HB isotopes in a two-step pancreatic clamp after 38 h of fasting. We found that D-3HB fluxes in the basal state were higher in lean compared to obese men: 15.2 (10.7–27.1) vs. 7.0 (3.5–15.1) µmol/kg lean body mass (LBM)·min, respectively, P < 0.01. No differences were found in KB fluxes between lean and obese volunteers during the pancreatic clamp (step 1: 6.9 (1.8–12.0) vs. 7.4 (4.2–17.8) µmol/kg LBM·min, respectively; and step 2: 2.9 (0–7.2) vs. 3.4 (0.85–18.7) µmol/kg LBM·min, respectively), despite similar plasma insulin levels. Meanwhile, peripheral glucose uptake was higher in lean compared to obese men (step 1: 15.2 (12.3–25.6) vs. 14.7 (11.9–22.7) µmol/kg LBM·min, respectively, P ≤ 0.05; and step 2: 12.5 (7.0–17.3) vs. 10.8 (5.2–15.0) µmol/kg LBM·min, respectively, P ≤ 0.01). These data show that obese subjects who display insulin resistance on insulin-mediated peripheral glucose uptake have the same sensitivity for the insulin-mediated suppression of ketogenesis. This implies differential insulin sensitivity of intermediary metabolism in obesity.
Starvation initiates an integrated metabolic response to prevent hypoglycemia and energy depletion (1). The ketone bodies (KBs) D-3-hydroxybutyrate (D-3HB) and acetoacetate (AcAc) play an important role in this adaptation: starvation induces an increase in plasma KB concentration and turnover serving as an important fuel for the central nervous system (2).
Ketogenesis occurs primarily in the liver. During fasting, lipolysis rates from white adipose tissue are increased, resulting in release of nonesterified fatty acids (NEFAs) into plasma (3). These fatty acids are degraded through β-oxidation within liver mitochondria, resulting in the production of acetyl-CoA. Acetyl-CoA is then either incorporated into the tricarboxylic acid cycle or channeled into the ketogenesis pathway (4). The transport of newly synthesized KBs from the hepatocytes to the circulation and the subsequent uptake in extrahepatic tissues occurs via diffusion and the monocarboxylate transporters 1, 2, and 4 (ref. 5,6). Ketogenesis is strongly suppressed by insulin and is stimulated in states of insulin deficiency and glucagon excess (4,6).
In addition to its physiological role in fasting, ketogenesis has been associated with insulin resistance: the dose–response relationship between circulating insulin and total KB (TKB) concentration was demonstrated to be shifted to the right in type 2 diabetes patients. This was interpreted as evidence of insulin resistance with respect to ketogenesis (7). However, it has been shown that obesity is associated with lower plasma D-3HB levels and a lower D-3HB oxidation rate compared to lean individuals (8,9,10). These observations question the relevance of the reported association of KBs with insulin resistance (7).
In this study, we investigated the relation between TKB turnover and glucose metabolism in 12 lean healthy and 12 obese nondiabetic men after short-term fasting (38 h), using stable isotopes and the pancreatic clamp technique. We hypothesized TKB turnover rates (TKB Ra) not to be different between lean and obese subjects when studied during similar plasma insulin levels.
Methods and Procedures
Twelve healthy lean and 12 healthy obese male subjects were recruited via advertisements in local magazines. Criteria for inclusion were (i) absence of a family history of diabetes in the lean group but not in the obese group; (ii) age 18–60 years; (iii) BMI 20–25 kg/m2 in the lean group, BMI >30 kg/m2 in the obese group; (iv) normal oral glucose-tolerance test according to American Diabetes Association criteria (11) in the lean group whereas in the obese group, glucose intolerance (but not diabetes) was accepted; (v) no excessive sport activities, i.e., less than three times per week; and (vi) no medication. Subjects were in self-reported good health, confirmed by medical history, routine laboratory and physical examination. Written informed consent was obtained from all subjects after explanation of purposes, nature, and potential risks of the study. The study was approved by the Medical Ethical Committee of the Academic Medical Center of the University of Amsterdam.
For 3 days before the fasting period, all volunteers consumed a weight-maintaining diet containing at least 250 g of carbohydrates per day. Then, the subjects started fasting at 2000 h 2 days before the study day until the end of the study.
Volunteers were admitted to the metabolic unit of the Academic Medical Center of the University of Amsterdam at 0730 h. Subjects were studied in the supine position and were allowed to drink water only. A catheter was inserted into an antecubital vein for infusion of stable isotope tracers, insulin, somatostatin, glucagon, and glucose. Another catheter was inserted retrogradely into a contralateral hand vein and kept in a thermoregulated (60 °C) Plexiglas box for sampling of arterialized venous blood.
In all the studies, saline was infused as NaCl 0.9% at a rate of 50 ml/h to keep the catheters patent. [6,6-2H2]glucose (>99% enriched; Cambridge Isotope Laboratories, Andover, MA) was used to study glucose kinetics. D[2,4-13C2]-3HB (>99% enriched; Cambridge Isotope Laboratories) was used to study KB turnover.
At T = 0 h (0800 h), blood samples were drawn for determination of background enrichments and a primed continuous infusion of the isotopes was started: [6,6-2H2]glucose (prime, 8.8 µmol/kg; continuous, 0.11 µmol/kg·min) and D[2,4-13C2]-3HB (prime, 1.0 µmol/kg; continuous, 0.10 µmol/kg·min) and continued until the end of the study. After an equilibration period of 2 h (38 h of fasting), three blood samples were drawn for glucose and KB enrichments and concentrations and one for glucoregulatory hormones and NEFA. Thereafter (T = 2.5 h) a two-step pancreatic clamp was started with low-dose insulin infusions to suppress ketogenesis: step 1 included an infusion of insulin at a rate of 4.5 mU/m2·min (Actrapid 100 IU/ml; Novo Nordisk Farma, Alphen aan den Rijn, the Netherlands). Glucose 5% was started when necessary to maintain a plasma glucose level of 5 mmol/l. [6,6-2H2]glucose was added to the 5% glucose solution to achieve glucose enrichments of 1% to approximate the values for enrichment reached in plasma and thereby minimizing changes in isotopic enrichment due to changes in the infusion rate of exogenous glucose. At the same time (T = 2.5 h) infusion of somatostatin (250 µg/h; Somatostatine-ucb; UCB Pharma, Breda, the Netherlands) was started to suppress endogenous insulin and glucagon secretion, as well as a glucagon infusion (1 ng/kg·min) (Glucagen; Novo Nordisk Farma) to replace endogenous glucagon concentrations. Somatostatin and glucagon infusions ran throughout step 1 and 2 of the clamp.
Plasma glucose levels were measured every 5 min at the bedside. After 2 h (T = 4.5 h), five blood samples were drawn at 5-min intervals for determination of glucose and D-3HB enrichments and concentrations. Another blood sample was drawn for determination of glucoregulatory hormones and NEFA. Hereafter, insulin infusion was increased to a rate of 7.5 mU/m2·min (step 2). Plasma glucose levels were continued to be measured every 5 min, and the final five blood samples were drawn at 5-min intervals for determination of glucose and D-3HB enrichments and concentrations (T = 6.5 h).
Body composition and indirect calorimetry
Body composition was measured with bioelectrical impedance analysis (Maltron BF-906; Maltron International, Essex, UK). Respiratory energy expenditure was measured continuously during the final 20 min of both the basal state and step 2 of the clamp by indirect calorimetry using a ventilated hood system (Sensormedics model 2900; Sensormedics, Anaheim, CA).
Glucose, KB, and NEFA measurements
Plasma glucose concentrations were measured with the glucose oxidase method using a Beckman Glucose Analyzer 2 (Beckman; Palo Alto, CA, intra-assay variation 2–3%). KB samples were drawn in chilled sodium fluoride tubes and directly deproteinized with ice cold perchloric acid 6% (1:1). AcAc and D-3HB were measured spectrophotometrically using D-3-hydroxybutyrate dehydrogenase (COBAS-FARA centrifugal analyzer; Roche Diagnostics, Almere, the Netherlands). Plasma NEFA concentrations were determined with an enzymatic colorimetric method (NEFA-C test kit; Wako Chemicals, Neuss, Germany): intra-assay variation: 1%; interassay variation: 4–15%; detection limit: 0.02 mmol/l.
[6,6-2H2]glucose enrichment was measured as described earlier (12). After centrifugation of the acidified KB samples, the supernatants were neutralized with KOH and HClO4. Following centrifugation, the supernatants were acidified with HCl and extracted twice with ethyl acetate. The combined extracts were dried under nitrogen at room temperature. D-3HB and AcAc were converted to their t-butyldimethylsilyl derivatives with MTBSTFA and pyridine. The sample was injected into an Agilent 6890/5973 MSD gas chromatograph/mass spectrometer system (Agilent Technologies, Palo Alto, CA). Separation was achieved on a Varian (Middelburg, the Netherlands) CP-SIL 19CB column (30 m × 0.25 mm × 0.15 µm). Selected ion monitoring (electron impact), data acquisition and quantitative calculations were performed using the Agilent ChemStation software. Ions were monitored at m/z 275 for unlabeled D-3HB, m/z 277 for D[2,4-13C2]-3HB, m/z 273 for unlabeled AcAc, and m/z 275 for [2,4-13C2]-AcAc. The enrichments in D-3HB and AcAc were determined from standard curves of known enrichments injected in the same run.
Insulin was measured with a chemiluminescent immunometric assay (Immulite 2000 system; Diagnostic Products, Los Angeles, CA), intra-assay variation: 3–6%, interassay variation: 4–6%, detection limit: 15 pmol/l. Basal state plasma insulin levels were measured with an ultra sensitive human insulin RIA kit (HI-11K; Linco, St Charles, MO; detection limit 1.5 pmol/l) to measure plasma insulin levels below the detection limit of the chemiluminescent immunometric assay. Glucagon was determined with the Linco 125I radioimmunoassay (St Charles). Intra-assay variation at 71 ng/l 10%, at 147 ng/l 9%; interassay variation at 84 ng/l 5%, at 192 ng/l 7%; detection limit 15 ng/l. Cortisol, epinephrine, and norepinephrine were determined as described previously (12).
Calculations and statistics
Resting energy expenditure and glucose and fat oxidation rates were calculated from O2 consumption and CO2 production as reported previously (13). Endogenous glucose production (EGP) and peripheral glucose uptake (rate of disappearance (Rd)) were calculated using the modified forms of the Steele Equations as described previously (14,15). EGP and Rd were expressed as µmol/kg lean body mass (LBM)·min. TKB production (rate of appearance/Ra) was calculated by using formulas for steady state as previously described (16,17):
where F is the infusion rate (µmol/kg·min), [A] and [B] are the plasma concentrations of AcAc and D-3HB, respectively and TTR A and TTR B are the tracer–tracee ratios of AcAc and D-3HB, respectively. TKB Ra was expressed as µmol/kg LBM·min and as µmol/kcal as proposed earlier for lipolysis (18,19). The KB ratio was calculated as (D-3HB)/(AcAc) (6). TKB metabolic clearance rate (MCR) was calculated as TKB Ra/(TKB).
A two-way ANOVA with repeated measures with factor 1 (time), factor 2 (lean vs. obese), and interaction was performed for substrate concentrations and kinetics as well as hormone levels between groups when these were assessed at all three occasions (basal, step 1 and 2 of the clamp). If significant effects were detected for factor 2 and interaction (or interaction alone), a post hoc Bonferroni test was performed, taking into account multiple comparisons, to assess differences between lean and obese volunteers at the different time points.
Other comparisons between groups were performed using the Mann–Whitney U-test. Comparisons within groups (T = 2, 4.5, and 6.5 h) were performed with the Wilcoxon signed-rank test. Correlations were expressed as Spearman's rank correlation coefficient (ρ). The SPSS statistical software program version 14.0 (SPSS, Chicago, IL) was used for statistical analysis. Data are presented as median (minimum–maximum).
Anthropometric data are presented in Table 1. Lean men were younger compared to obese men. As expected from the inclusion criteria, there were significant differences in weight, BMI, body fat mass, and LBM. Fasting glucose levels were not different between both groups, but glucose concentrations after the oral glucose-tolerance test were significantly higher in the obese subjects. Data on routine laboratory examinations prior to inclusion are presented in Table 1. Two obese subjects were glucose intolerant, but no diabetes was found.
Table 1. Subject characteristics
Resting energy expenditure, glucose, KB kinetics, and NEFA
Total resting energy expenditure was higher in obese subjects compared to lean subjects in the basal state: 2,086 (1,623–2,200) kcal/day vs. 1,797 (1,428–2,067) kcal/day, respectively, P < 0.01; as well as at the end of step 2 of the clamp: 1,952 (1,226–2,107) vs. 1,682 (1,293–1,958), respectively, P < 0.01.
Plasma glucose levels were significantly higher in the obese subjects compared to the lean subjects in the basal state, step 1 and step 2 of the clamp (Table 2). EGP did not differ between lean and obese subjects in the basal state after 38 h of fasting but was lower in lean subjects during step 1 and step 2 of the clamp (Figure 1). No change in EGP was observed between step 1 and the basal state in obese subjects (P = 0.2), but a significant decrease was observed for step 2 compared to step 1 (P < 0.05). EGP decreased in the lean subjects during the clamp; step 1 vs. basal state and step 2 vs. step 1: P < 0.01 and P < 0.01, respectively. Rd of glucose was lower during step 1 and 2 of the clamp in obese men compared to lean controls (Table 2).
Table 2. Glucose, ketone body, and lipid metabolism measurements
Data on KB are presented in Table 2 and Figure 1. Plasma D-3HB was lower in obese men in the basal state, but no differences were found during the clamp; plasma AcAc was not different between subjects in the basal state, step 1 and step 2 of the clamp. TKB Ra (in either µmol/kg LBM·min or µmol/kcal) was significantly lower in obese subjects in the basal state, but during the clamp, no differences were observed (Figure 1). TKB Ra decreased within lean subjects during the clamp; step 1 vs. basal state and step 2 vs. step 1: P < 0.01 and P < 0.01, respectively. No significant change in TKB Ra was observed between step 1 and the basal state in obese subjects, but a significant decrease was observed for step 2 compared to step 1 (P < 0.01).
The MCRTKB showed no differences in the basal state and step 1 of the clamp (Table 2). During step 2, obese subjects had lower MCRTKB compared to lean subjects. The KB ratio (two-way ANOVA repeated measure: ns, P < 0.01 and P < 0.01 for factor 1 (time), factor 2 (lean vs. obese), and interaction, respectively) was lower in the basal state in obese subjects compared to lean subjects: 2.1 (1.3–3.8) vs. 3.8 (1.9–7.0), respectively, P < 0.01. The KB ratios during the clamp were not statistically different between obese and lean subjects (step 1: 2.1 (1.1–3.8) vs. 2.2 (1.0–4.3), respectively; step 2: 1.7 (0.6–4.0) vs. 1.5 (0–3.5), respectively).
Obese men had lower plasma NEFA levels in the basal state, but higher plasma NEFA levels during step 1 of the clamp compared to lean men (Table 2). During step 2 of the clamp, no significant differences were observed. In the basal state, plasma NEFA correlated with TKB Ra in lean men; a trend toward a correlation was found in obese men (Figure 2).
Insulin levels were higher in obese subjects compared to lean subjects in the basal state (Table 3). However, there were no differences during step 1 and 2 of the clamp (Table 3). In the basal state, there was a significant negative correlation between plasma insulin levels and TKB Ra in obese but not in lean men (Figure 2).
Table 3. Glucoregulatory hormones
Plasma glucagon levels were not different in the basal state but were significantly higher in obese subjects during the clamp. Plasma cortisol and epinephrine levels were not significantly different between obese and lean subjects at all time points (Table 3). Norepinephrine levels were higher in obese subjects in the basal state and step 1 of the clamp, but no difference was found during step 2 of the clamp.
In patients with type 2 diabetes, a right-shifted dose–response curve for insulin vs. ketogenesis has been reported, suggesting resistance on insulin-mediated suppression of ketogenesis (7). The present study was designed to investigate KB metabolism (i.e., TKB Ra) in obese insulin-resistant subjects after 38 h of fasting in relation to glucose metabolism. We found that, after short-term fasting, rates of ketogenesis are higher in lean subjects, but this difference is no longer present after reaching similar plasma insulin levels. This implies that there is no insulin resistance on ketogenesis in obese subjects during short-term fasting.
We did not confirm earlier data on lower EGP in obese subjects in the basal state after 38 h of fasting (9). However, we did express EGP as µmol/kg LBM min because the central nervous system is part of the LBM and accounts for the majority of glucose uptake during fasting (20).
Plasma D-3HB and TKB levels as well as TKB Ra were higher in lean subjects compared to obese subjects in the basal state as shown earlier (8,9,10). There are multiple ways to explain this difference. First, the higher insulin levels in our obese subjects can inhibit the activity of the key enzyme in ketogenesis, 3-hydroxy-3-methylglutaryl-CoA synthase (mHS), in the liver (4,6,21). Indeed, plasma insulin levels correlated negatively with TKB Ra in the obese subjects in the basal state, whereas the lack of this correlation in lean subjects may be explained by the suppressed plasma insulin levels after 38 h of fasting. Second, it has been demonstrated that plasma glucose levels inhibit ketogenesis independently from insulin (22). It is unlikely, however, that the higher plasma glucose levels in the obese volunteers induced lower plasma KB levels because the suppressive effect of glucose on KBs was observed during profound hyperglycemia (i.e., 12 mmol/l) (22). Finally, and more importantly, the lower TKB Ra in obese subjects in the basal state may be explained by lower NEFA levels for these are reported to increase KB production independently of insulin (17). Besides decreasing hepatic ketogenesis, insulin also inhibits lipolysis by inhibiting hormone-sensitive lipase in adipose tissue (4,6). Indeed, we found a convincing correlation of NEFA and TKB Ra in the basal state in lean and obese subjects.
Initially, MCRTKB did not differ between lean and obese men, which emphasizes the notion that the rate of utilization of KBs is proportional to their circulating levels (4). The lower MCRTKB in obese subjects during step 2 of the clamp is not explained by differences in TKB Ra or plasma TKB levels.
Our study does not support the data by Singh et al. (7) who demonstrated a significant right shift of the dose–response relationship between circulating insulin and TKB concentrations in patients with type 2 diabetes. This can be explained by differences in study design because we studied obese insulin-resistant subjects with approximately fourfold higher insulin levels in the basal state, whereas the plasma insulin levels in the patients of Singh were similar to lean controls. Furthermore, our volunteers had been fasting for 38 h as opposed to the overnight fast in their study (7). In contrast, it may be arguable that the fractional decline in ketogenesis in lean volunteers was greater than in the obese subjects implicating a reduced insulin sensitivity of ketogenesis in the latter. Unfortunately, we have not measured TKB Ra in the obese volunteers during somatostatin infusion prior to exogenous insulin infusion to compare low plasma insulin levels between lean and obese men. However, at similar insulin levels, TKB Ra did not differ between lean and obese subjects suggesting that insulin infusion resulted in similar plasma insulin levels and TKB Ra in both groups.
In this study, we used a single isotope method to analyze TKB Ra (16). Earlier, the use of a single isotope model in measuring TKB Ra was debated because of the rapid interconversion of AcAc and D-3HB (23). Later, it was discussed by Beylot et al. that the single isotope method provided a reasonable estimate of TKB Ra even in the postabsorptive state when interconversion between AcAc and D-3HB is relatively high (16,17). Our volunteers had plasma KB levels well above postabsorptive levels indicating less interconversion of AcAc and D-3HB thereby permitting a single isotope analysis (16). Moreover, during the clamp, KB ratios were not different. On theoretical grounds, our results in the basal state could be overestimated in the lean group by 15%. However, the enormous difference in TKB Ra between the lean and obese subjects in the basal state cannot be due to such overestimation.
The studied groups were not matched for age. It has been shown that aging in healthy men is associated with increased KB levels during prolonged fasting but not in the postabsorptive state (24). Therefore, our findings are not likely to be attributed to age differences in our groups.
The obese volunteers had higher plasma free fatty acid levels during step 1 and 2 of the clamp that did not result in higher TKB Ra levels. This may be explained by the overruling inhibitory effect of insulin on ketogenesis, i.e., increased insulin and glucose availability will increase malonyl-CoA thereby inhibiting CPT1 and β-oxidation and thus ketogenesis. Moreover, mitochondrial HMG-CoA synthase is also inhibited by insulin (4). Plasma glucagon levels were not different in the basal state, but this does not preclude glucagon from being a regulator of ketosis in lean subjects. Although the exact role of glucagon in ketogenesis is unclear to date (17,25), Miles et al. showed that glucagon increases KB production when NEFA concentrations are increased in the setting of isolated insulin deficiency (25). This may explain why the higher plasma glucagon levels during the clamp did not result in higher TKB Ra in obese men compared to lean men.
In conclusion, we show in this study that insulin-mediated suppression of ketogenesis is similar in obese and lean subjects. Meanwhile, obese subjects display a lower insulin-mediated peripheral glucose uptake. This implies differential insulin sensitivity of intermediary metabolism in obesity.