Increased Brain Lactate Concentrations Without Increased Lactate Oxidation During Hypoglycemia in Type 1 Diabetic Individuals

  1. Kitt Falk Petersen3
  1. 1Department of Diagnostic Radiology, Yale University School of Medicine, New Haven, Connecticut
  2. 2Department of Psychiatry, Yale University School of Medicine, New Haven, Connecticut
  3. 3Department of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut
  4. 4Department of Cellular and Molecular Physiology, Yale University School of Medicine, New Haven, Connecticut
  5. 5Howard Hughes Medical Institute, Yale University School of Medicine, New Haven, Connecticut
  6. 6Department of Biomedical Engineering, Yale University School of Medicine, New Haven, Connecticut
  1. Corresponding author: Kitt Falk Petersen, kitt.petersen{at}


Previous studies have reported that brain metabolism of acetate is increased more than twofold during hypoglycemia in type 1 diabetic (T1D) subjects with hypoglycemia unawareness. These data support the hypothesis that upregulation of blood-brain barrier monocarboxylic acid (MCA) transport may contribute to the maintenance of brain energetics during hypoglycemia in subjects with hypoglycemia unawareness. Plasma lactate concentrations are ∼10-fold higher than acetate concentrations, making lactate the most likely alternative MCA as brain fuel. We therefore examined transport of [3-13C]lactate across the blood-brain barrier and its metabolism in the brains of T1D patients and nondiabetic control subjects during a hypoglycemic clamp using 13C magnetic resonance spectroscopy. Brain lactate concentrations were more than fivefold higher (P < 0.05) during hypoglycemia in the T1D subjects compared with the control subjects. Surprisingly, we observed no increase in the oxidation of blood-borne lactate in the T1D subjects, as reflected by similar 13C fractional enrichments in brain glutamate and glutamine. Taken together, these data suggest that in addition to increased MCA transport at the blood-brain barrier, there may be additional metabolic adaptations that contribute to hypoglycemia unawareness in patients with T1D.

Despite the increased availability of improved methods for managing glycemic control (i.e., continuous glucose monitoring), failing counterregulation and hypoglycemia unawareness still present a real burden in the daily life of type 1 diabetic (T1D) and advanced (insulin-deficient) type 2 diabetic patients (1,2). Recurrent episodes of hypoglycemia are considered to induce both the failure in counterregulatory hormone release and hypoglycemia unawareness, a concept known as hypoglycemia-associated autonomic failure (3,4).

Although the exact mechanisms of hypoglycemia unawareness are still unknown, studies have predominantly focused on adaptations related to nutrient transport into the brain and changes in brain energy metabolism. For example, changes in the transport of plasma glucose across the blood-brain barrier and consequently the brain glucose levels have been the topic of various studies (510). Other studies have focused on glycogen supercompensation, a hypothesis suggesting increased storage of glucose in astroglial glycogen after recurrent hypoglycemic events (1113). The increased astroglial glycogen would function as a glucose reserve during hypoglycemia. However, during a 50-h wash-in and wash-out study of [1-13C]glucose, control subjects showed higher levels of newly synthesized brain glycogen than hypoglycemia-unaware T1D subjects (11). Öz et al. (11) consequently concluded that glycogen supercompensation did not contribute to hypoglycemia unawareness in T1D patients.

Previously we have reported that brain transport and metabolism of acetate is increased more than twofold in intensively treated T1D subjects with hypoglycemia unawareness (14). These data support the hypothesis that upregulation of blood-brain barrier monocarboxylic acid (MCA) transport via MCA transporter 1 (15,16) may be a hallmark of hypoglycemia unawareness in T1D patients. In contrast to acetate, which circulates in plasma at relatively low concentrations (∼0.1 mmol/L), plasma lactate concentrations are ∼10-fold higher during hypoglycemia (17), making it a primary candidate for an alternative brain fuel (1821).

Lactate metabolism can play a central role in neuroenergetics, as suggested by the astrocyte-neuron lactate shuttle (22). The astrocyte-neuron lactate shuttle models the compartmentalized metabolism of glucose in astrocytes and neurons. It describes how glucose is metabolized through glycolysis in astrocytes, producing lactate. Lactate is then shuttled to neighboring neurons where it is oxidized. The astrocyte-neuron lactate shuttle is analogous to the intercellular lactate shuttle that was proposed earlier and describes skeletal muscle lactate metabolism (23).

We have shown in healthy subjects that there is sufficient lactate transport activity to supply ∼10% of the brain’s energy needs at physiological lactate concentrations (24). Increased blood-brain barrier transport capacity of MCAs, and thus lactate, could contribute to the maintenance of brain energetics during hypoglycemia, providing the brain with an increased influx of alternative substrates (14). However, to our knowledge, there is no direct evidence of increased brain transport and oxidation of plasma lactate in T1D patients. We therefore examined transport of lactate over the blood-brain barrier and its metabolic fate in healthy T1D patients and nondiabetic control subjects during a hypoglycemic clamp by measuring 13C label incorporation from intravenously administered [3-13C]lactate into brain lactate, glutamate (Glu), and glutamine (Gln) by 13C magnetic resonance spectroscopy (MRS).



Five healthy T1D patients (age 34 ± 5 years; BMI 23.0 ± 1.5 kg/m2) and six healthy control subjects matched for BMI (age 24 ± 1 years; BMI 23.5 ± 0.9 kg/m2) were recruited for this study. The T1D subjects all were in well to moderate glycemic control (HbA1c 7.6 ± 0.9%).

The T1D subjects were selected on the criteria of having experienced frequent hypoglycemic events based up on the Clarke questionnaire (25), with subject scores ranging from 5 to 3 with an average of 4.3 ± 0.9. The control subjects had normal fasting plasma glucose concentrations (4.9 ± 0.1 mmol/L), HbA1c 5.3 ± 0.1%, and were not taking any medications except for birth control pills. The purpose, nature, and potential complications of the studies were explained, and written consent was obtained from each subject. The protocol was approved by the Yale University Human Investigation Committee.

Hypoglycemic clamp studies.

All subjects presented at 7:00 a.m. the morning of the study in the Yale Magnetic Resonance Research Center after an overnight fast. Subjects with diabetes were instructed to take their usual evening dose of insulin and to abstain from their morning insulin dose. After intravenous catheters were inserted into each antecubital area for blood collection and for infusions, basal blood samples were collected for determination of plasma glucose, lactate, β-hydroxybutyrate, insulin, glucagon, and catecholamine concentrations.

At 8:00 a.m., the subjects were positioned in the supine position in the 4 Tesla MRS scanner. A primed-continuous infusion of insulin was initiated and kept constant at 40 mU/(m2 ⋅ min) while plasma glucose concentrations were measured every 5 min and allowed to decrease to 3.1 mmol/L and kept constant at this level with a variable infusion of 20% dextrose. The head of each subject was positioned over the 13C transmit/receiver coil, and the bed was slid into the magnetic resonance (MR) scanner. A primed-continuous infusion of [3-13C]l-lactate (Isotech, Miamisburg, OH) was started and continued for 90–120 min at a rate of 10 μmol/(kg ⋅ min) (Fig. 1). MR spectra were acquired continuously throughout the study, and blood samples were drawn at intervals of 5–10 min for the determination of plasma substrate and hormone concentrations and for determination of the enrichment of plasma [13C]lactate.

FIG. 1.

Schematic illustrating the time line of the hyperinsulinemic-hypoglycemic clamp, [3-13C]lactate infusion, and 13C MRS acquisition.

Measurement of metabolites and hormones.

Plasma glucose and lactate concentrations were measured every 5 min using a YSI 2700 STAT Analyzer (Yellow Springs Instruments, Yellow Springs, OH). Samples for hormones were taken every 15 min. Plasma concentrations of insulin and glucagon were measured with the use of double-antibody radioimmunoassay kits (Linco, St. Charles, MO). Plasma epinephrine and norepinephrine concentrations were measured with a three-step procedure that consisted of adsorption onto alumina (pH 8.6), elution with dilute acid, and analysis by high-pressure chromatography.

Fractional enrichments of plasma [13C]glucose and [13C]lactate were measured by gas chromatography–mass spectrometry (GCMS) using a Hewlett-Packard 5890 gas chromatograph (HP-1 capillary column, 12 × 0.2 × 0.33-mm film thickness; Hewlett Packard, Palo Alto, CA) interfaced to a Hewlett Packard 5971A mass selective detector operating in chemical ionization (CI) mode with isobutane as the reagent gas. Glucose was analyzed by GCMS as the glucose-pentaacetate. 13C isotopic enrichments of singly and multiply labeled isotopic isomers (m+1, m+2, m+3, m+4, and m+6) of glucose were determined using CI and monitoring ions 331–337. Singly labeled glucose was calculated from the ratio of the m+0 signal (m/z 331) and the m+1 signal (m/z 332). Lactate was analyzed by GCMS as the n-butyl ester-trifluoroacetate derivative. 13C isotopic enrichment (m+1) of lactate was determined using CI mode and monitoring ions 243 and 244.

MRS acquisition.

MR spectra were acquired using a 4 Tesla whole-body magnet equipped with a Bruker console (Bruker Instruments, Billerica, MA), as previously described (24). The RF-coil setup was a combination of a circular 13C coil (∅ 8.5 cm) for acquisition and two quadrature 1H surface coils (∅ 15 cm) for imaging, shimming, polarization transfer, and 1H decoupling. After scout imaging, shimming was performed using the FASTERMAP procedure (26), and decoupling power was calibrated. 13C MR spectra were acquired using a polarization transfer sequence optimized for detection of C4 of Glu and Gln (27) (repetition time [TR] = 2,500 ms, 128 averages), in combination with 3D ISIS localization and outer volume suppression. The volume of interest was a 90-mL voxel centered on the midline in the occipital-parietal lobe during the infusion of [3-13C]lactate.

Spectral processing and analysis.

Spectra were manually phase corrected, and Lorentzian (−2 Hz) and Gaussian (6 Hz) apodization and baseline correction up to second order was applied. Peak amplitudes were determined with an in-house software package written in MATLAB using an LCModel approach with each 13C resonance having independent amplitudes (28). Basis sets for peak fitting were acquired in phantom solutions using identical MRS acquisition conditions for Glu, Gln, N-acetyl aspartate (NAA), aspartate, creatine, and lactate. Glu and Gln C4 peaks were fitted with a spectrum averaged over the last 21 min of the time series. Lactate C3 (Lac C3) and NAA C3 and C6 peak amplitudes were fitted in a spectrum averaged over the complete time course. Concentrations of 13C Lac, Glu, and Gln were calculated using the averaged NAA C3 and C6 peak amplitudes and assuming a concentration for NAA of 11 μmol/g (29,30). Fractional 13C enrichment of Glu C4 and Gln C4 were determined assuming concentrations for Glu (9.8 μmol/g) and Gln (4.2 μmol/g) (31).

Measurement of brain lactate concentrations.

Brain lactate concentrations ([brain Lac]) were determined from the measured 13C concentration of 13C3 lactate ([brain LacC3]) by assuming that at steady state, the fractional 13C enrichment of C3 lactate (fe[brain LacC3]) was similar to that of Glu C4 (fe[GluC4]) (Eq. 1). This assumption is based on the lactate/pyruvate pool being the immediate precursor for acetyl-CoA, which in turn is the precursor for the Glu C4 and C5 carbons (32). A correction in the measured 13C concentration of brain Lac C3 was applied for the contribution of plasma [3-13C]lactate, assuming a plasma volume of 5% relative to total brain volume (33).


Metabolic modeling analysis.

Steady-state metabolism of lactate was modeled using a one-compartment model as depicted in Fig. 2. At steady state, the inflow of plasma lactate (Vin) relative to the outflow from the brain (Vout) and lactate oxidation in the tricarboxylic acid (TCA) cycle (VTCA) was derived (Eq. 2):Formulawhere CMRglc represents the glucose consumption rate and fe indicates fractional enrichment of the particular metabolite. Equation 2 was solved theoretically; no absolute values of CMRglc, VTCA, Vin, and Vout are derived from the present data. Previously we have described the linear relationship between plasma and brain lactate concentrations (24). To account for the possible effect of different plasma lactate concentrations between groups, we also normalized the results of Eq. 2. to the average plasma lactate concentration of the control group. We used the transport relationships described by Boumezbeur et al. (24) who showed that the relationship between plasma lactate concentration and lactate unidirectional transport is approximately linear at the concentrations of plasma lactate studied here.

FIG. 2.

One-compartment model describing incorporation of 13C label from [3-13C]lactate into the brain glutamate and glutamine pools. This figure illustrates the fluxes Vin (lactate influx), Vout (lactate efflux), CMRglc (glucose consumption), and VTCA (TCA cycle rate), which were considered to derive Eq. 2. BBB, blood-brain barrier; α-KG, α-ketoglutarate; MCT1, MCA transporter 1. Graphic, 13C-labeled carbon position. Lactate in the neuronal and glial compartments was treated as a single pool due to the rapid transfer of lactate between these cells (24).

Statistical analysis.

Group differences between control and T1D subjects were analyzed using two-tailed, unpaired Student t test, considering a P value <0.05 as statistically significant. All data are presented as mean ± SEM.


Basal plasma lactate concentrations were similar in control and T1D subjects (control, 0.80 ± 0.03 mmol/L; T1D, 1.06 ± 0.21 mmol/L; P = 0.21). Plasma β-hydroxybutyrate tended to be higher in T1D subjects (398 ± 184 μmol/L) than control subjects (83 ± 16 μmol/L, P = 0.09) at baseline. After insulin infusion, plasma β-hydroxybutyrate was lower and similar in both groups at the start of the [3-13C]lactate infusion (control, 44 ± 8 μmol/L; T1D, 60 ± 5 μmol/L; P = 0.11). After 40–60 min of insulin infusion, plasma glucose levels stabilized at 3.6 ± 0.1 mmol/L in the control and 3.2 ± 0.3 mmol/L in the T1D subjects (P = 0.10). M-values were not significantly different among the groups (control, 4.25 ± 0.69 mg/[kg ⋅ min]; T1D, 3.06 ± 0.59 mg/[kg ⋅ min]; P = 0.14). After the start of the [3-13C]lactate infusion, plasma lactate concentrations quickly increased to 1.8 ± 0.4 mmol/L in the control subjects and 1.3 ± 0.2 mmol/L in the T1D subjects (P = 0.05). The average 13C fractional enrichment of plasma lactate was 26.2 ± 4.0% in control and 31.6 ± 6.7% in T1D subjects (P = 0.13). Mean fractional 13C enrichments of plasma glucose ([1-13C]glucose) between 60 and 90 min were 1.3 ± 0.1% in the control subjects and 2.0 ± 0.1% in the T1D subjects (P = 0.0013).

The elevated plasma lactate levels, as anticipated from previous studies, led to a blunting of the counterregulatory response in control subjects with only a significant but small increase in epinephrine that was similar between groups (no difference between groups in glucagon, epinephrine, and norepinephrine concentrations). Figure 3 shows examples of 13C MR spectra averaged over the last 30 min of [3-13C]lactate infusion from a control and a T1D subject, respectively. Glu C4 13C fractional enrichment increased quickly after the infusion of [3-13C]lactate, reaching similar steady-state levels (corrected for 1.1% natural abundance 13C signal) of 2.8 ± 0.3% in the controls and 2.7 ± 0.2% in T1D subjects (P = 0.40). Gln C4 13C fractional enrichment was 1.9 ± 0.5% in the controls and 2.0 ± 0.4% in T1D subjects (P = 0.77). The ratio of 13C Gln C4/Glu C4 was 0.73 ± 0.11 in T1D subjects and 0.65 ± 0.11 in control subjects (P = 0.70) (Table 1).

FIG. 3.

13C MR spectra of a T1D subject (top) and control subject (bottom) averaged over the last 30 min of [3-13C]lactate infusion.


Steady-state fractional enrichments of brain Glu and Gln during intravenous infusion of [3-13C]lactate in control and T1D subjects

The calculated brain lactate concentrations were increased by more than fivefold in the T1D subjects (1.7 ± 0.6 μmol/g) compared with the control subjects (0.3 ± 0.2 μmol/g, P < 0.05) (Fig. 4). Furthermore brain lactate concentrations normalized to the average plasma lactate concentration were increased more than sixfold in the T1D subjects (2.2 ± 0.9 μmol/g) compared with the control subjects (0.3 ± 0.2 μmol/g, P < 0.05).

FIG. 4.

Total calculated lactate concentrations in brain.

Lactate influx into the brain, as a fraction of the brain TCA cycle rate and the flow of lactate out of the brain [Vin/(VTCA + Vout)], as estimated using Eq. 2, was similar in the control (0.11 ± 0.02) and T1D subjects (0.09 ± 0.01, P = 0.25). When normalized to the average plasma lactate level of the control group, Vin/(VTCA + Vout) was still similar between the control and the T1D subjects (0.11 ± 0.01 and 0.12 ± 0.01, respectively, P = 0.71).


In the current study, we examined whether lactate blood-brain barrier MCA transport and subsequently oxidation of blood-borne lactate in T1D subjects were increased as compared with nondiabetic individuals during mild hypoglycemia. In support of increased MCA transport capacity in the T1D subjects, we found elevated concentrations of lactate in the brain during the infusion of [3-13C]lactate. Surprisingly, despite the several-fold higher brain lactate levels in the diabetic subjects, the fractional entry of blood-borne lactate into the brain lactate pool [Vin/(VTCA + Vout)] did not appear any different from control subjects, given similar Glu C4 fractional enrichments.

The T1D group showed increased transport capacity for plasma lactate, as shown by brain lactate concentrations being comparable to the levels in plasma so that the rate of lactate influx would be similar to lactate efflux (Vout = Vin). A lack of net lactate influx is similar to what has been measured during euglycemia (24), indicating that at these mild levels of hypoglycemia, the T1D subjects have not downregulated glucose metabolism. In contrast, in the control subjects, brain lactate concentration was extremely low so that almost all lactate entering the brain was being oxidized (Vout ∼0), which indicates a reduction in glucose oxidation in the control subjects by a minimum of ∼11% (Eq. 2). Previous studies using positron emission tomography, MRS, and arterio-venous differences reported drops in glucose oxidation in control subjects between 25 and 45% (9,10,34). The actual drop in glucose oxidation in the control subjects may therefore have been considerably greater than 11%. The reduction in brain glucose metabolism in control subjects compared with the T1D subjects would also explain the similarity in the ratio Vin/(VTCA + Vout) despite higher unidirectional lactate transport (Vin) in the T1D subjects. At lower levels of hypoglycemia, it is possible that the net oxidation of lactate in the T1D subjects would increase to greater than that of the control subjects due to their increased lactate transport activity.

Brain lactate originates from both the plasma lactate and from glucose metabolism through glycolysis. In ideal experimental conditions, the glycolytic flux is unlabeled and would dilute the fractional enrichment of brain [13C]lactate and subsequently Glu. The increased calculated brain lactate concentration in T1D subjects could consequently be the result of increased (unlabeled) glycolytic flux relative to control subjects, as explained above. However, lactate is also an important precursor in gluconeogenesis. Glucose synthesized from [3-13C]lactate will be labeled in the C1 and C6 positions. In our experiments, gluconeogenesis is strongly inhibited by the infused insulin, but a small metabolic flux of 13C-labeled glucose needs to be considered. The levels of glucose fractional enrichment in the last 30 min of the study were similar across groups, small (∼1.5%) compared with the fractional 13C enrichment of lactate (∼30%) and therefore considered negligible. In addition, when analyzing fractional enrichment of Glu C4 between 20 and 40 min of [3-13C]lactate infusion (before any 13C-labeled glucose could have contributed to the Glu pool), results were similar to those from the steady-state analysis (control subjects, 2.4 ± 0.6%; T1D subjects, 2.1 ± 0.6%; P = 0.4). We would therefore argue that the small amounts of 13C-labeled plasma glucose did not weaken our interpretation that similar Glu C4 fractional enrichments despite higher brain lactate transport indicate preservation of glucose oxidation in the T1D subjects.

At present, the mechanism for the likely maintenance of brain glucose metabolism in hypoglycemic-unaware subjects during hypoglycemia is unknown. Although upregulated brain glucose transport has been reported in rodent models exposed to recurrent hypoglycemia (35,36), similar findings have not been reported in humans. Positron emission tomography studies using [11C]-3-O-methyl-d-glucose or [1-11C]glucose to assess glucose transport in humans have not found evidence of upregulation of glucose transport in unaware T1D or healthy subjects (5,37). Similarly, studies using 1H MRS have not found definitive evidence of a metabolically significant change in glucose transport in subjects with T1D and/or hypoglycemia unawareness under euglycemic conditions (8,38). In addition, a recent study by van de Ven et al. (7) did not show differences in brain glucose concentrations during both euglycemia and hypoglycemia in control and T1D subjects without hypoglycemia unawareness.

Recently, a new role for brain lactate was proposed, acting as a volume transmitter in addition to a metabolic substrate, with higher brain lactate concentrations stimulating neuronal activity and increased brain glucose metabolism (39). The mechanisms put forward include the NADH/NAD+ redox ratio and a cAMP pathway triggered by binding of lactate to the suggested G-protein–coupled receptor 81 (GPR81) (39). GPR81 is expressed in adipose tissue, and there seem to be indications of its presence in brain tissue as well (39). Future studies are required to confirm a role for lactate as volume transmitter in brain and its potential relevance in the sensing of the brain’s energy status and hypoglycemia. Activation of GPR81 by increased brain lactate concentrations in hypoglycemic-unaware T1D subjects could potentially be involved in regulating brain glucose metabolism in these individuals. Besides the potential role of GPR81, various other mechanisms have been described of lactate regulating redox-sensitive pathways (reviewed in 40).

The relative 13C labeling of Gln C4 and Glu C4 after [3-13C]lactate infusion closely resembled that from providing [1-13C]glucose as a substrate (Table 1), as was shown in our previous study (24). The resemblance of the 13C-labeling patterns of [1-13C]glucose and [3-13C]lactate as substrates indicates similar fractional enrichments for neuronal and glial lactate pools. This also implies that the transport rates for lactate between neurons and glia are at least similar or higher than the glucose oxidation rate, thus allowing the treatment of the lactate pool as one, shared by neurons and glia (Fig. 2).

In vivo 13C MRS can offer unique data of brain lactate transport and metabolism but is technically challenging and associated with high costs. 13C MRS studies are therefore often carried out using relatively small group sizes, which is a limitation of the studies. 13C MRS is also inherently limited in both spatial and temporal resolution. Our data were acquired from a relatively large volume of the brain. It is not inconceivable that smaller areas of the brain demonstrate different metabolic responses to hypoglycemia and the presence of blood-borne lactate. Similarly, the timing of the response to hypoglycemia can vary in different brain regions. Such spatial and temporal variations in metabolism cannot be detected with the method as used in the current study. Other limitations of the study are the assumptions required for the quantification and modeling of the MRS data and the slightly lower hypoglycemic glucose levels in T1D subjects compared with controls. However, the slightly lower plasma glucose levels in the T1D group makes the lack of increased lactate metabolism in those subjects even more remarkable. Because lower plasma glucose levels equal lower brain glucose levels, the T1D group experienced a somewhat more severe brain energy challenge. Nevertheless, despite higher brain lactate levels compared with controls, no increased lactate oxidation was detected in the T1D group.

In conclusion, our results showing increased brain lactate concentration in T1D subjects who experience regular hypoglycemic episodes further support our previous finding, using an acetate tracer, of increased MCA transport being a metabolic adaptation that may have a role in hypoglycemic unawareness. However, the present data highlight the differences between cerebral lactate and acetate metabolism, a result of brain metabolism being highly compartmentalized. The lack of increased lactate oxidation despite increased brain lactate levels in the T1D subjects is surprising and implies that other roles of lactate beyond being a metabolic fuel need to be explored. In addition, future studies using MRS should be able to answer whether under higher levels of brain activity (or lower glucose concentrations) the higher lactate transport activity becomes important for supporting metabolic demand, which we would anticipate, and to determine the degree to which glucose oxidation is relatively increased in the T1D subjects.


This work was supported by National Institutes of Health grants R01-NS-051854, R01-AG-034953-01A1 (D.L.R.), R01-AG-023686 (K.F.P.), R21-AA-018210 (G.F.M.), R01-DK-49230, R24-DK-085638, P30-DK-45735 (G.I.S.), R21-AA-019803 (G.F.M.), and UL1-RR-024139, a Distinguished Clinical Investigator Award from the American Diabetes Association (K.F.P.), and an award from the W.M. Keck Foundation. H.M.D.F. is supported by a fellowship of the American Institute for Cancer Research (10A087).

No potential conflicts of interest relevant to this article were reported.

H.M.D.F., D.L.R., and K.F.P. acquired and analyzed data and were involved in research and interpretation of the data and writing, reviewing, and editing of the manuscript. G.F.M. implemented MR acquisition and quantification methods. G.I.S. was involved in research and interpretation of the data and writing, reviewing, and editing of the manuscript. H.M.D.F. and D.L.R. 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.

Preliminary data from this study were presented at the 72nd Scientific Sessions of the American Diabetes Association, Philadelphia, Pennsylvania, 8–12 June 2012.

The authors thank Mikhail Smolgovsky (Yale-New Haven Hospital Research Unit), Irina Smolgovsky (Yale University School of Medicine), Yanna Kosover (Yale University School of Medicine), Donna D’Eugenio (Yale-New Haven Hospital Research Unit), Gina Solomon (Yale-New Haven Hospital Research Unit), and the Yale-New Haven Hospital Research Unit for expert technical assistance with the studies; and Terry Nixon, Peter Brown, and Scott McIntyre (Yale University Magnetic Resonance Research Center) for maintenance and upgrades to the MRS system. The authors thank the volunteers for their participation in these studies.


  • See accompanying commentary, p. 3024.

  • Received February 22, 2013.
  • Accepted May 18, 2013.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See for details.


| Table of Contents

This Article

  1. Diabetes vol. 62 no. 9 3075-3080
  1. All Versions of this Article:
    1. db13-0313v1
    2. 62/9/3075 most recent