Role of Hepatic Glycogen on Nocturnal Gluconeogenesis in Type 2 Diabetes Mellitus

Abstract

Context

Higher gluconeogenesis (GNG) contributes to higher nocturnal endogenous glucose production (EGP) in type 2 diabetes (T2D). Studies using ¹³C magnetic resonance spectroscopy (MRS) have confirmed lower hepatic glycogen content in subjects with T2D than in subjects with no diabetes (ND).

Objective

We determined the role of glycogen loading (GL) vs nonglycogen loading (NGL) on the contribution of GNG to nocturnal EGP in T2D.

Methods

In total, 14 subjects with T2D and 15 matched subjects with ND were studied on 2 occasions, with GL (60% carbohydrate) vs NGL (40% carbohydrate) isocaloric meals for 3 days, in random order in the overnight state. [6,6-²H₂] glucose was infused to measure EGP, deuterium labelled water was used to measure GNG, and ¹³C MRS scans were performed in fed and fasted states to measure hepatic glycogen content.

Results

Hepatic glycogen content and nocturnal EGP were higher (P < .05) in GL vs NGL in both cohorts. The % GNG to EGP averaged ∼50% in subjects with ND throughout the night after both meals. In contrast, % GNG to nocturnal EGP in T2D was lower with GL vs NGL and matched the pattern observed in subjects with ND with GL lowering overnight rates of GNG in subjects with T2D.

Conclusion

Selective targeting of GNG at night with appropriate medications could reduce nocturnal and early morning fasting hyperglycemia and hepatic insulin resistance in people with T2D.

The liver has a fundamental role in regulating postabsorptive and postprandial carbohydrate metabolism. Glucose homeostasis is maintained by the liver in the postabsorptive state by controlling endogenous glucose production (EGP) (gluconeogenesis [GNG] and glycogenolysis [GGL]) and during postprandial state by facilitating hepatic glucose uptake leading to hepatic glycogen synthesis and storage. Hepatic glycogen synthesis, GGL, and GNG are regulated by balance of circulating glucose, insulin, and glucagon concentrations.

Studies using ¹³C magnetic resonance spectroscopy (MRS) have confirmed lower hepatic glycogen content in subjects with T2D than in those with ND before and after dinner (1). We have previously established that hepatic glycogen synthesis through the UDP glucose pathway is reduced in T2D entirely due to reduced flux through the direct pathway, implying a functional defect in the rate-limiting hepatic glucokinase activity (2, 3). We have also shown that the insulin dose response for hepatic glucose uptake is shifted to the right (4) in T2D. Taken together, lower hepatic glucose uptake due to a functional glucokinase defect contributes, at least in part, to lower hepatic glycogen content in individuals with T2D. Krssak et al (1) demonstrated that despite lower hepatic glycogen content, predinner EGP was higher due to higher GNG in subjects with T2D than in those ND. We have recently published that higher GNG during the latter part of the night contributes to higher fasting EGP in T2D (5). Therefore, lower hepatic glycogen content could contribute to dysregulation of nocturnal GNG and GGL. We reasoned that the higher rate of GGL in those with T2D in the early part of the night (5) would deplete an already lower hepatic glycogen store sooner, leading to falling rates of GGL with reciprocally rising rates of GNG as the night progresses (5). Therefore, we hypothesized that increasing postdinner hepatic glycogen content (via glycogen loading [GL] meals at dinner) would restore regulation of nocturnal GGL and GNG in T2D. To do so, we planned an optimal approach to maximize the differences in hepatic glycogen content in T2D and ND with minimal perturbation of nocturnal and early morning fasting glucose while estimating the effects of this difference on nocturnal glucose turnover and rates of GNG and GGL.

Research Design and Methods

The University of Virginia (UVA) Institutional Review Board for health sciences research (IRB-HSR) approved the study. Eligible subjects reported in the morning after an overnight fast, at the UVA Clinical Research Unit (CRU). After signing informed consent, subjects were screened, involving history, physical examination, and blood and urine testing to ensure subjects met enrollment criteria. A urine pregnancy test was performed whenever appropriate and a negative result confirmed before proceeding. Dietary history was taken to ensure adherence to a weight maintaining diet and physical activity assessed using the Paffenbarger activity questionnaire. Body composition was measured using multifrequency bioelectrical impedance (Inbody 770, Inbody Co. Ltd, Korea) (6). The enrollment criteria included age of 30-80 years and body mass index (BMI) range of 21 to 35 kg/m². Subjects with ND were frequency matched for age and BMI. Subjects with T2D with HbA1C ≤8.5% with lifestyle management or monotherapy (metformin or sulfonylureas) or HbA1c ≤7.5% on 2 oral hypoglycemic agents were included. However, use of insulin, SGLT2 inhibitors, thiazolidinediones, DPP4 inhibitors or GLP-1 receptor agonists, presence of microvascular or macrovascular complications, apart from stable background retinopathy were exclusionary. Pregnant and nursing women and persons with upper gastrointestinal disorder, debilitating illnesses, anemia, cardiac, and hepatic or renal disease were excluded. Individuals on medication affecting gastrointestinal motility and glucose tolerance, except those on stable thyroid hormones, were excluded. All subjects with ND with a history of diabetes in first-degree family members, gestational diabetes, or prediabetes were excluded.

Subjects with T2D were withdrawn from oral hypoglycemic agents 7 to 10 days prior to study visits and were asked to monitor fingerstick fasting blood glucose. If fasting glucose values were >300 mg/dL on 2 consecutive days they were asked to resume their antidiabetic medications and withdrawn from the study. Menstruating female subjects were studied in the follicular phase of their menstrual cycle. In order to control body weight changes to <2% between screening and study visits, subjects were advised to refrain from unusual and unaccustomed physical activity and adhere to a weight maintaining diet during the study.

All subjects were studied on 2 occasions, following 3 days of GL or nonglycogen loading (NGL) isocaloric meals, in random order. Both study visits were separated by at least 3 weeks to allow complete elimination of deuterium-labeled water. Both study visits were identical except for the preceding GL or NGL meal.

The isocaloric solid meals were 60:20:20 (% carbohydrate: % protein: % fat) for GL and 40:20:40 (% carbohydrate: % protein: % fat) for NGL. The subject's total estimated daily calorie intake (33 kcal/kg/day) was based on Harris Benedict + 20% calorie requirements, and the total calories were distributed evenly over 3 meals (33% per meal). Meals were prepared by the research dietician of the UVA CRU for 3 days prior to the overnight study. During this period, subjects were asked to continue habitual exercise/activity which was self-monitored with an activity tracker (Fitbit LLC, Google, Mountain View, CA).

Subjects reported to the CRU at 16:00 hours, were fed an evening meal. and ingested 3 timed doses of 1.67 g/total body weight each of deuterium-labeled water (²H₂O) for estimation of GNG with the first dose accompanying the evening meal. Four hours after the meal at ∼20:00 hours all subjects underwent ¹³C MRS imaging scan for estimation of liver glycogen in the fed state (fed scan) (7). At ∼21:30 hours an intravenous cannula was placed in the forearm for tracer infusions and a second was inserted on the contralateral hand to obtain arterialized venous blood samples with the heated-hand technique (8). A primed continuous infusion of [6,6-²H₂] glucose (11.8 mg/kg prime; 0.118 mg/kg/min) was started at 22:30 hours and continued overnight until 07:00 hours to estimate nocturnal EGP. All water consumed ad libitum during the study visit was enriched with ∼2% deuterium-labeled water to maintain steady state. Subjects were asked to sleep and lights were turned off at 22:00 hours. Blood samples were obtained throughout the night at timed intervals. ¹³C MRS scan was repeated on the second day at 07:00 hours to estimate liver glycogen in the fasted state (fasted scan) (Fig. 1A). Thereafter, infusions were discontinued, subjects were provided breakfast, and dismissed with instructions on care of the venipuncture sites.

Figure 1.

(A) Schematic representation of the experimental design. (B) CONSORT flow diagram. After meeting the inclusion criteria, study participants (no diabetes [ND] and type 2 diabetes [T2D]) received 3 days of glycogen loading (GL) and nonglycogen loading (NGL) meals in a crossover randomized fashion.

Analytical Techniques

Samples, were stored at −80 °C until analysis. Plasma glucose was measured by the glucose oxidase method (YSI, Xylem Inc, Yellow Spring, OH). Plasma enrichment of [6,6-²H₂] glucose, [5-²H] glucose, and [2-²H] glucose were measured by gas chromatography-mass spectrometry (5975 GC-MSD, Agilent, Santa Clara, CA) as described (3, 9, 10). Insulin and glucagon were measured by radioimmunoassay (Millipore Cat# PI-12 K, RRID:AB_2801580 and Millipore Cat# GL-32 K, RRID:AB_2757819) (Millipore Sigma, Danvers, MA) (2).

Liver glycogen was measured by ¹³C MRS using a 3-Tesla whole-body MR imager/spectrometer (Prisma-fit-fit, Siemens Healthineers, Malvern, PA) and a 12-cm ¹³C radio frequency (RF) coil (PulseTeq Ltd, Surrey, UK), which included a ¹H RF coil that was used for acquisition of proton scout images. The RF coil also included a small fiducial marker mounted in the center of the coil, which was filled with an aqueous solution of ¹³C-enriched urea (Sigma-Aldrich, Inc., St. Louis, MO) thus producing a signal in both proton images and ¹³C spectra. The ¹³C RF coil was placed laterally over the right lobe of the liver, and held in place with an elastic strap. Proton scout images included a 3-plane localizer, a 3-dimensional (3D) RF-spoiled gradient-echo acquisition, and a free-breathing cine acquisition based on a balanced steady-state free precession pulse sequence. These proton acquisitions were used to confirm proper coil placement, ensure consistent coil positioning between paired acquisitions in each subject, and allow measurement of the distance between the RF coil and the surface of the liver. Following a ¹³C transmitter calibration acquisition, glycogen ¹³C spectra were acquired during free breathing using a pulse-acquired free induction decay sequence (TR 500 ms, data acquisition 0.25 ms after the center of a 0.30-ms rectangular 90° excitation RF pulse, 256 complex data points for each free induction decay over 32 ms, and 1800 averages during 15 minutes). Following spectroscopy, the ¹³C RF coil was removed and the subject was positioned in the standard ¹H spine and body array RF coils. A 3D T1-weighted RF-spoiled gradient-echo image set was acquired during a breath-hold, and used to determine liver volume. The scanner was operated within FDA approved limits of radiofrequency heating specific absorption rate (SAR) and peripheral nerve stimulation due to field gradient switching.

Glycogen ¹³C spectra were processed using iNMR commercial software (version 6.3.3, Mestrelab Research, Santiago, Spain) and the data were decoupled. Following initial processing (0 filling to 2048 points, 40-Hz line broadening, Fourier transform, phase correction, and baseline correction), glycogen signal integrals were determined by line fitting in iNMR. The absolute glycogen concentration was calculated from the in vivo glycogen integrals, referenced to signal integrals from a 100 mM glycogen phantom, with corrections for RF-coil to liver-surface distance, coil loading, and liver volume in the sensitive region of the ¹³C RF coil relative to glycogen phantom volume (11, 12).

Calculations

Rates of EGP were estimated throughout the night but emphasis was on 01:00, 04:00, and 07:00 hours (5, 13). Since the tracer to tracee ratio was constant, steady-state equations of Steele were applied. Rates of GNG were calculated by multiplying the plasma C5 glucose to C2 glucose ratio by EGP while rates of GGL were calculated by subtracting the rate of GNG from the rate of EGP (5, 14).

Statistical Analyses

The experimental design was configured with parallel cohorts (ND and T2D) with primary hypotheses of within cohort comparisons (ie, GL vs NGL meals). For sample size planning purposes, a paired t-test was utilized. At the time of study design, no preliminary data existed on the effect of GL, so the effect size observed was utilizing a similar overnight study estimating GNG (5). For the studied participants with T2D, the mean (SD) GNG (µmol/kg/min) rate at 07:00 hours was 12.9 (2.8) and 10.6 (3.0) under test and control conditions. This resulted in a change of 2.05 (1.68) between the 2 conditions. Assuming a similar magnitude of change of 2 units is clinically relevant and a SD of the difference of 1.75 (to be conservative), a sample size of 9 was required to achieve 80% power at the α = .05 level of significance. The protocol was written to study up to 15 participants per cohort to provide additional precision in the estimates and address the uncertainty in the GL effect.

A linear mixed model with compound symmetry variance structure was fit to the data that had repeated measurements at 01:00, 04:00, and 07:00 hours. This model specified parameters for study group (T2D vs ND), time of night (01:00, 04:00, and 07:00 hours) and the GL treatment (GL vs NGL). This model estimated 12 (2 × 3 × 2) individual means for each outcome measure, with GNG planned as the primary endpoint at time of study planning. In order to summarize the effects of the model, model-based means were constructed to provide pooled estimates of the differences in means of interest. In some cases, several of the model-based means were averaged together (eg, a “diabetes” effect was obtained by comparing the mean of the 01:00, 04:00, and 07:00 hours measures under both GL conditions and unloaded conditions, viz, 6 values averaged for each patient group). All P values are unadjusted and reported in conjunction with estimates and 95% CI. Statistical analysis was performed using R version 4.1.2.

Results

Subject Characteristics

Thirty-four subjects (17 each with T2D and ND) were screened for the study. One subject with T2D did not meet enrollment criteria and 2 subjects in each group withdrew after successful screening due to non-study–related reasons (scheduling difficulty, time off from work). Twenty-nine subjects (14 with T2D and 15 ND) completed all study procedures (Fig. 1B).

Sex, weight, BMI, lean body weight and % body fat were not different between groups (Table 1). However, the ND group were significantly younger than the T2D group (P = .05) and fasting plasma glucose before stopping antidiabetes medications and HbA1c levels were higher in T2D (P < .001).

Table 1.

Subject characteristics

	No diabetes (n = 15)	Type 2 diabetes (n = 14)	P value
Gender (M/F)	6/9	8/6	.356
Age (years)	54.5 (14.0)	63.2 (7.7)	.049
Body weight (kg)	82.5 (12.5)	82.4 (11.0)	.978
Body mass index (kg/m2)	28.5 (2.9)	28.0 (2.8)	.653
Lean body mass (kg)	53.8 (10.4)	55.0 (11.0)	.763
Body fat (%)	34.8 (7.6)	33.3 (7.5)	.616
Fasting plasma glucose (mmol/L)	4.7 (0.4)	8.0 (3.5)	<.001
HbA1C (%)	5.3 (0.4)	7.1 (1.4)	<.001
HbA1C (mmol/mol)	34.7 (4.3)	53.9 (14.9)	<.001
Duration of diabetes (years)	NA	6.8 (3.7)

Mean (SD) of subjects with ND and subjects with T2D.

Abbreviations: T2D, type 2 diabetes; ND, no diabetes.

Plasma Glucose, Insulin and Glucagon concentrations

The experimental design was implemented to study physiological differences in the overnight periods. As expected, there was a pronounced difference in overnight glucose between T2D and ND (P < .0001). The concentrations under the GL condition were 0.82 (95% CI 0.35 to 1.29; P = .0008) mmol/L (∼14 mg/dL) higher than in the NGL condition in subjects with T2D in contrast to the ND group who did not experience a change in overnight glucose concentrations (change NG vs NGL: −0.02, 95% CI −0.50 to 0.47; P = .95) (Fig. 2). Insulin concentrations demonstrated a pronounced cohort effect (P = .0033) and differences with respect to GL vs NGL conditions in the subjects with T2D but not in those with ND. Specifically, the estimated differences in insulin concentrations attributable to GL were 18.0 (95% CI 7.7 to 28.2; P = .0007) pmol/L and 7.8 (95% CI −2.8 to 18.5; P = .146) pmol/L for the subjects with T2D and ND, respectively.

Figure 2.

Plasma glucose (A), insulin (B), and glucagon (C) concentration measured during primed continuous infusion of [6,6-²H₂] glucose among subjects with no diabetes (ND) and subjects with type 2 diabetes (T2D) after a nonglycogen loading (NGL) and glycogen loading (GL) meal.

There were changes in glucagon concentrations overnight as a result of the GL, albeit the pattern of effect varied by subject cohort. For the subjects with T2D, there was a numeric reduction in glucagon concentrations (pg/mL) (−4.3, 95% CI −9.5 to 0.8; P = .10). For the ND group, there was a numeric increase (3.1, 95% CI −2.3 to 8.5; P = .25). The results of the changes were that the subjects with T2D and those with ND had similar glucagon concentrations during the GL meal (difference −0.3, 95% CI −15.9 to 15.3; P = .97) and glucagon concentrations within about 10% of one another during the NGL meal (difference 7.1, 95% CI −8.4 to 22.7; P = .36).

Tracer/Tracee Ratio of [6,6-²H₂] Glucose, Plasma [5-²H] Glucose, Plasma [2-²H] Glucose, and Ratio of C5/C2 Glucose

The tracer/tracee ratio of [6,6-²H₂] glucose remained flat throughout the overnight period following NGL and GL meals in either cohort (Fig. 3), thereby permitting steady-state calculations of glucose turnover. Enrichment of plasma [5-²H] glucose (C5 glucose), plasma [2-²H] glucose (C2 glucose), and C5/C2 ratios are presented in Table 2. C5 glucose, C2 glucose, and C5/C2 ratios were not significantly different in ND group than in subjects with T2D overnight (P = .66, P = .57, and P = .49, respectively) over both GL conditions (NGL and GL) and the 3 measurements taken at 01:00, 04:00, and 07:00 hours. Enrichment of C5 glucose and the resulting C5/C2 ratios were significantly higher with NGL than GL meals in both ND and T2D whereas C2 glucose was not (Table 2).

Figure 3.

Tracer/tracee ratio of [6,6-²H₂] glucose measured throughout the overnight period in subjects with no diabetes (ND) and subjects with type 2 diabetes (T2D) after a nonglycogen loading (NGL) and glycogen loading (GL) meal.

Table 2.

MPE of C5 glucose, C2 glucose, and ratio of C5/C2 glucose

Time of day	01:00 Mean (SD)	04:00 Mean (SD)	07:00 Mean (SD)
[5-2H] glucose (C5) (%MPE)
NGL meal
ND	0.32 (0.09)*	0.34 (0.11)*	0.35 (0.11)*
T2D	0.27 (0.10)*	0.32 (0.09)*	0.36 (0.10)*
GL meal
ND	0.25 (0.11)	0.28 (0.12)	0.31 (0.13)
T2D	0.20 (0.08)	0.26 (0.09)	0.29 (0.09)
[2-2H] Glucose (C2) (%MPE)
NGL meal
ND	0.52 (0.10)	0.51 (0.10)	0.51 (0.10)
T2D	0.53 (0.10)	0.52 (0.10)	0.52 (0.10)
GL meal
ND	0.51 (0.11)	0.51 (0.11)	0.51 (0.11)
T2D	0.52 (0.09)	0.52 (0.09)	0.52 (0.09)
C5/C2 ratio
NGL meal
ND	0.61 (0.13)*	0.66 (0.15)*	0.68 (0.16)*
T2D	0.51 (0.14)*	0.61 (0.10)*	0.68 (0.12)*
GL meal
ND	0.47 (0.15)	0.54 (0.13)	0.58 (0.15)
T2D	0.38 (0.14)	0.50 (0.17)	0.56 (0.16)

Mean (SD) of subjects with ND and subjects with T2D after NGL and GL meal.

Abbreviations: GL, glycogen loading; MPE, molar percentage enrichment; NGL, nonglycogen loading; ND, no diabetes; T2D, type 2 diabetes.

^* P < .007, NGL vs GL.

Rates of EGP, GNG, GGL, and Liver Glycogen

The changes in EGP showed consistent profiles between the GL and NGL conditions overnight, with EGP being higher during GL vs NGL throughout the night in both cohorts. During the NGL challenge, the participants with T2D had elevated GNG whereas EGP and GGL were suppressed (P < .001 for each in the overall pooled analysis) (Fig. 4 and Table 3). These were observed for all the time points during the night except for the estimation of GNG at 04:00 hours; this difference did not reach statistical significance (−0.8, 95% CI −1.8 to 0.2; P = .10). Liver glycogen concentrations varied markedly between the GL and NGL conditions (P < .001) in both the fed and fasted states (Fig. 4 and Table 3).

Figure 4.

Endogenous glucose production (EGP) (A), gluconeogenesis (GNG) (B), and glycogenolysis (GGL) (C) at 01:00, 04:00, and 07:00 hours, and liver glycogen (D) at fed and fasted state after nonglycogen loading (NGL) and glycogen loading (GL) meals in no diabetes (ND) and type 2 diabetes (T2D). ND NGL and ND GL, T2D NGL and T2D GL compared after NGL and GL meal. *P < .05 vs GL. §P < .05 vs ND based on linear mixed model analysis.

Table 3.

Difference estimate between NGL and GL among subjects with ND and subjects with T2D

Outcome	Time of day (h)	ND NGL (n = 13) Mean (SD)	ND GL (n = 12) Mean (SD)	Difference estimate (95% CI)	P value	T2D NGL (n = 13) Mean (SD)	T2D GL (n = 14) Mean (SD)	Difference estimate (95% CI)	P value
GNG, µmol/kg FFM/min	01:00	8.1 (1.7)	7.3 (2.1)	−1.0 (−2.0, 0.02)	.056	8.8 (2.7)	7.1 (2.8)	−1.6 (−2.6, −0.7)	.001
	04:00	7.9 (1.8)	7.2 (1.7)	−0.8 (−1.8, 0.2)	.129	8.6 (2.0)	7.8 (2.8)	−0.8 (−1.8, 0.2)	.101
	07:00	9.5 (2.1)	9.0 (1.9)	−0.6 (−1.6, 0.4)	.245	10.8 (3.0)	9.7 (3.1)	−1.1 (−2.0, −0.1)	.03
	Pooled overnight	8.5 (7.3, 9.7)	7.7 (6.5, 8.9)	−0.8 (−1.4, −0.2)	.011	9.4 (8.2, 10.5)	8.2 (7.0, 9.4)	−1.2 (−1.7, −0.6)	<.001
EGP, µmol/kg FFM/min	01:00	13.4 (2.1)	15.8 (2.2)	2.8 (1.6, 4.0)	<.001	17.4 (3.4)	19.1 (4.6)	1.8 (0.6, 3.0)	.003
	04:00	12.1 (1.9)	13.6 (1.9)	1.9 (0.7, 3.2)	.002	14.2 (2.5)	15.7 (2.7)	1.6 (0.4, 2.8)	.008
	07:00	14.2 (2.8)	15.8 (3.2)	2.4 (1.2, 3.7)	<.001	15.7 (2.7)	17.3 (2.4)	1.7 (0.5, 2.9)	.005
	Pooled overnight	13.3 (12.1, 14.8)	15.5 (14.5, 17.2)	2.4 (1.7, 3.1)	<.001	15.6 (14.2, 17.0)	17.4 (16.0, 18.7)	1.7 (1.0, 2.4)	<.001
GGL, µmol/kg FFM/min	01:00	5.3 (2.5)	8.5 (3.0)	3.6 (2.3, 5.0)	<.001	8.6 (3.1)	12.0 (5.1)	3.5 (2.1, 4.8)	<.001
	04:00	4.2 (2.5)	6.4 (2.3)	2.7 (1.3, 4.0)	<.001	5.6 (1.8)	7.9 (3.7)	2.4 (1.1, 3.8)	<.001
	0700	4.7 (3.3)	6.8 (3.2)	2.6 (1.2, 4.0)	<.001	4.9 (1.9)	7.6 (3.3)	2.8 (1.5, 4.1)	<.001
	Pooled overnight	4.8 (3.2, 6.4)	7.8 (6.2, 9.4)	3.0 (2.1, 3.8)	<.001	6.3 (4.7, 7.9)	9.2 (7.6, 10.8)	2.9 (2.1, 3.7)	<.001
Liver glycogen, mmol	Fed state	255.7 (93.0)	369.5 (114.1)	113.8 (84.1, 143.5)	<.001	198.5 (84.6)	265.2 (102.5)	64.1 (33.5, 94.6)	<.001
	Fasted state	192.9 (82.4)	290.7 (91.4)	97.8 (68.1, 127.4)	<.001	142.8 (63.8)	204.7 (74.6)	59.2 (28.7, 89.8)	<.001
	Pooled effect	224.0 (176.0, 273.0)	330.0 (282.0, 379.0)	105.8 (84.8, 126.8)	<.001	173.0 (124.0, 222.0)	235.0 (186.0, 283.0)	61.6 (39.8, 83.5)	<.001

Mean (SD) of EGP, GNG, and GGL at 01:00, 04:00, 07:00 hours, liver glycogen at fed and fasted state, and pooled effect (overnight) in subjects with ND and subjects with T2D after (NGL) and (GL) meal. Difference estimate (95% CI) between NGL and GL meal at 01:00, 04:00, 07:00 hours, for EGP, GNG, and GGL, fed and fasted state for liver glycogen, and pooled effect (overnight) in ND and T2D.

Abbreviations: EGP, endogenous glucose production; FFM, fat free mass; GGL, glycogenolysis; GL, glycogen loading; GNG, gluconeogenesis; ND no diabetes; NGL, nonglycogen loading; T2D, type 2 diabetes.

Discussion

We have previously shown that EGP remains elevated through the night in patients with T2D compared with patients without diabetes, without a clear dawn phenomenon in patients with T2D (5). We have also shown that while GNG and GGL rates change through the course of the night in patients with T2D (higher GNG and lower GGL), they are largely constant through the night in patients without diabetes (5). In the current work, we went further by exploring the impact of different carbohydrate meal compositions in the interplay between GNG and GGL to normalize the contribution to that observed in people without T2D.

Prior studies have shown that liver glycogen by ¹³C MRS was lower in patients with T2D than in controls in the fed state (15). However, a comprehensive evaluation of whether restoration of hepatic glycogen by nontherapeutic means can improve nighttime glucose control has not been studied. The pattern of contribution of nocturnal GNG and GGL in T2D differs from that in healthy humans (5). Adequate glycogen stores overnight are important for maintaining overnight normoglycemia. Having lower glycogen stores due to hepatic glucokinase defect in T2D may cause subsequent depletion of glycogen in the early part of the night and increased GNG in the latter part of the night. We observed that by restoring hepatic glycogen content overnight using a slightly higher, but physiologic and germane to daily life, carbohydrate content in the meal resulted in reduction of GNG at 01:00 hours and 07:00 hours in T2D. However, this was at a cost of slightly higher rates of EGP due to higher GGL throughout the night in T2D following a GL diet. In contrast, in the ND cohort, a higher carbohydrate meal did not change rates of GNG but the rates of GGL were higher throughout the night than during the lower carbohydrate meal. Taken together, GL by increasing carbohydrate (vs fat) proportions in mixed meals increased rates of EGP in both T2D and ND cohorts by increasing rates of GGL throughout the night. In contrast, GL reduced rates of GNG in subjects with T2D but did not alter the contribution of GNG to EGP in those with ND. It is important to mention that these results were obtained in the absence of antidiabetes medications. However, the results lead the way for refining strategies to mitigate nocturnal hyperglycemia.

In individuals with T2D, the higher rates of nocturnal EGP with a GL diet resulted in higher circulating glucose ∼0.8 mM or 14 mg/dL vs NGL meal and in a small rise in insulin concentrations, implying slight worsening of insulin resistance overnight. However, fasting glucose concentrations in the morning at 0:700 hours during GL were higher by only ∼0.52 mmol/L (9 mg/dL) in subjects with T2D implying a small clinical effect size. In the ND group, while GL increased EGP through the night, glucose concentrations remained unchanged compared with NGL. Interestingly though, insulin concentrations also tended to be higher throughout the night during GL than NGL in the ND group, implying that further increasing postdinner hepatic glycogen content by providing slightly more carbohydrates (60% of calorie content) in subjects with ND could perhaps cause some mild insulin resistance in these individuals as well. However, in a prior study conducted by our group in healthy humans in which hepatic glycogen content was measured using radioisotopes, the results indicated that while higher glycogen content in the liver did affect EGP and percent contribution of GNG, it did not worsen hepatic insulin resistance (16).

As far as hepatic glycogen content was concerned, the strategy we applied with a GL meal program for 3 days prior to study was effective in achieving significant increases (by 20-40%) in net glycogen content in both groups in the fed state that continued through the overnight fasted period. Interestingly, GL in subjects with T2D resulted in hepatic glycogen content in the fed state to be similar to those with ND in the fed state after NGL meals. Likewise, the hepatic glycogen content in GL in subjects with T2D in the fasted state approximated that in the ND group in the NGL fasted state. Taken together, this implies that GL in subjects with T2D could replenish hepatic glycogen stores postdinner and in the overnight fasted state to approach that observed in the ND group without GL, thus, at least in part, overcoming the functional defect in hepatic glucokinase activity. However, though GL meals restored hepatic glycogen content in T2D to that observed in ND normally and reduced rates of GNG to those observed in ND, fasting (07:00 hours) glucose and insulin concentrations were higher by ∼ 9 mg/dL and ∼13 pmol/L.

We believe that this slight increase in overnight glucose and insulin concentrations may not be clinically relevant in the light of the potential benefit of restoring overnight glycogen stores to that observed in the ND group. However, the long-term effects of GL will need to be carefully evaluated in T2D in future studies. Restoring postdinner hepatic glycogen content in T2D increased the quantity of glycogen available for GGL in these individuals. To capture the maximal meal effects on net hepatic glycogen content, we opted to measure hepatic glycogen content approximately 4 hours after completion of the GL vs NGL evening meals (17, 18). Additionally, we independently conducted pilot studies approved by the IRB-HSR in both subjects with ND and subjects with T2D with both fasted and fed scans to optimize the glycogen MRS protocol and confirmed that ∼4 hours after the meal provided the highest glycogen content in the liver in both subjects with ND and subjects with T2D.

GL meals resulted in similar increases in liver glycogen in subjects with vs without T2D, and similar increases in GGL rates through the night in both group of cohorts. In contrast, the reduction in GNG rates was more pronounced in T2D than in ND. The mechanism for greater suppression of GNG in T2D is perhaps due to autoregulation of GNG by glycogen stores in the liver coupled with the slightly higher insulin concentrations overnight observed with GL meal resulting in lesser suppression of GGL but greater suppression of GNG as we have previously shown during hyperinsulinemic pancreatic hyperglycemic clamp conditions (8). Secondly, we have previously shown that higher amount of insulin is required to suppress GNG than GGL in humans (19), and the insulin dose response curves for EGP are shifted to the right and coupled with defects in insulin action and insulin secretion in T2D, higher insulin concentrations would be required to normalize nocturnal EGP in this cohort (4). Given that hepatic insulin resistance preexists in T2D, much higher insulin concentrations would be necessary to suppress GNG and hence EGP in T2D in the setting of lower hepatic glycogen stores, thus implying that there could be possible hepatic hypercompensation.

Of note glucagon concentrations were similar between the 2 cohorts studied with the different meals thereby explaining the comparable GGL in both cohorts. Interestingly, estimated fractional rates of GNG in T2D receiving GL meals were comparable to the ND group with NGL meals. Moreover, by early morning (ie, 07:00 hours) subjects with T2D who were fed GL meals maintained a relative contribution of GNG and GGL that was similar to what was observed in the ND group at the beginning of the night (ie, 01:00 hours). Hence the relative contributions of GNG and GGL are to some extent restored by the GL meal fed at dinnertime in T2D.

Bisschop et al, have previously conducted studies in Europe with relatively small cohort of healthy (n = 6) subjects, and T2D (5F/2 M) subjects with very high-carb (89%) and very high-fat (89%) diets (20, 21). The studies and the resultant conclusions cannot be compared due to the differences in experimental design. In contrast to Bisschop's studies, we utilized physiologically relevant amounts of carbohydrates which are typically consumed by people living in the United States. We quantified glycogen stores in the nighttime fed state and early morning fasted state in order to estimate nighttime GNG and EGP. To our knowledge, a comprehensive evaluation of contribution of hepatic glycogen with NMR spectroscopy coupled with estimations of nighttime partitioning of GNG and GGL to nocturnal glycemia has not been previously undertaken by any group prior to this study.

Our study has some limitations. Interventional studies of this nature require timed collection of blood and hence impractical and difficult to conduct with multiple permutations and combinations of dietary macronutrients keeping in mind blood volume restrictions and inconvenience to research subjects participating in nighttime studies. Additionally, these studies were conducted during the COVID pandemic. Hence, we were restricted in terms of the studies that could be conducted in the radiology suite and CRU. We do not believe this impacted the outcome.

The amount of carbohydrate used permitted us to study these subjects without large changes in nocturnal glucose concentrations once they were off antidiabetes medications, which otherwise would have precluded them from participation. It also remains to be seen whether different timing of the evening meal might have a larger impact on nocturnal EGP. In order to minimize the potential for confounding, those with ND were frequency matched for BMI to subjects with T2D but the ND group were younger (P = .05). However, the analysis plan, which was focused largely on estimating changes within an individual over time and between GL experimental conditions was not likely to be affected. In particular, the repeated measures design of the statistical analysis considered each individual as their own control comparing the data obtained over time and within time between GL and NGL conditions. The effect of age, and other measured and unmeasured variables, was estimated into the random intercept in the model and differences in the means between conditions were not likely influenced by age.

Finally, while there has been some concern in the field about potential tracer recycling; we have not observed such a process in prior studies (5) or the current work (data not shown).

Dietary and therapeutic strategies focused on differential inhibition of these pathways, or time-centered changes (ie, decreasing GGL early in the night or GNG later in the night) may hold the key to appropriate suppression of EGP and improvement of nocturnal and early morning hyperglycemia in people with T2D. We believe that focusing on different strategies to restore liver glycogen stores, such as glucokinase activators, metformin which suppresses GNG and insulin which inhibits GGL may provide rationale strategies for restoring nocturnal EGP to that observed in healthy humans without diabetes. Though there was a small change in EGP it is relevant to point out that the plasma glucose concentration after the GL meal in T2D averaged ∼14 mg/dL higher than what was observed in the NGL meal. We do not believe this slight increase is clinically meaningful or detrimental to health of these subjects, as these subjects outside of the research study would be taking their customary doses of GL agents. We also recognize that perhaps drug targets would be more beneficial than manipulating diet, but given that the average human being with diabetes changes the amount of carbs consumed day to day the results of dietary carbohydrate manipulation are nevertheless very meaningful and insightful. The studies were conducted in the absence of antidiabetes medications (ie, during a wash out period). While the amount of carbohydrate provided as part of the research during the high GL meals is not atypical of what an average person with T2D in the United States consumes in their daily diet, the absence of GL medications resulted in the higher fasting glucose concentrations observed (∼14 mg/dL). Hence in the real world if/when individuals with T2D on precise nighttime GL agents were to consume a slightly higher carbohydrate meal at supper (∼60% vs 50%) they would near normalize glycogen stores to what is observed in healthy people with ND, and the GL medications they are on would also mitigate the slight increase in GGL, EGP, and nocturnal glucose concentrations. Hence, it is possible that careful control of carbohydrate consumption in the evening meal in conjunction with appropriate nighttime medication can improve glycemia overnight in T2D.

In summary, our study shows that by restoring liver glycogen stores in people with T2D with a higher carbohydrate meal at dinner, we are able to near normalize the relative contributions of GNG in the overnight state in people with T2D to that observed in healthy people without diabetes. This is an important finding, and advancement as we previously have reported that patients with T2D experience a dysregulation of both nocturnal GNG and GGL, with progressive increase in GNG and a proportional decrease in GGL through the night (5). It is possible that by selectively targeting GNG with appropriate medications, we could reduce nocturnal and early morning fasting hyperglycemia and hepatic insulin resistance in people with T2D. Approaches looking at different strategies to achieve this (different carbohydrate compositions or different timing of the meals), as well as therapeutic strategies specifically targeting GNG (eg, metformin) may hold the key for the normalization of nocturnal EGP in T2D.

Abbreviations

3D

3-dimensional

BMI

body mass index

CRU

Clinical Research Unit

EGP

endogenous glucose production

GGL

glycogenolysis

GL

glycogen loading

GNG

gluconeogenesis

IRB-HSR

Institutional Review Board for health sciences research

MRS

magnetic resonance spectroscopy

ND

no diabetes

NGL

nonglycogen loading

RF

radio frequency

SAR

specific absorption rate

T2D

type 2 diabetes

Funding

The work received support from the National Institute of Health grants DK 029953 (R.B.), DK 085516 (A.B.), and Core facilities of the Vanderbilt University Medical Center DK 059637 (MMPC) and DK 020593 (DRTC).

Author Contributions

R.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data analysis. A.B. and R.B. designed and conducted the studies, analyzed data, and wrote the manuscript and R.C. analyzed the data and edited the manuscript. J.M. conducted ¹³C NMR liver scans and analyzed data, edited manuscript. U.U. analyzed data, edited manuscript, F.B. edited manuscript.

Disclosure

The authors have no conflicts of interest to disclose.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Clinical Trial Information

Clinical Trials.gov ID: NCT04416204 (registered June 4, 2020).

Prior Presentation

This study was in part previously published in abstract form at the American Diabetes Association Scientific Sessions meeting, Diabetes. 2023 June 20; 72(Supplement_1):1580-P.