Failure of Hyperglycemia and Hyperinsulinemia to Compensate for Impaired Metabolic Response to an Oral Glucose Load

Abstract

Objective

To evaluate whether the augmented insulin and glucose response to a glucose challenge is sufficient to compensate for defects in glucose utilization in obesity and type 2 diabetes, using a breath test measurement of integrated glucose metabolism.

Methods

Non-obese, obese normoglycemic and obese Type 2 diabetic subjects were studied on 2 consecutive days. A 75g oral glucose load spiked with ¹³C-glucose was administered, measuring exhaled breath ¹³CO₂ as an integrated measure of glucose metabolism and oxidation. A hyperinsulinemic euglycemic clamp was performed, measuring whole body glucose disposal rate. Body composition was measured by DEXA. Multivariable analyses were performed to evaluate the determinants of the breath ¹³CO₂.

Results

Breath ¹³CO₂ was reduced in obese and type 2 diabetic subjects despite hyperglycemia and hyperinsulinemia. The primary determinants of breath response were lean mass, fat mass, fasting FFA concentrations, and OGTT glucose excursion. Multiple approaches to analysis showed that hyperglycemia and hyperinsulinemia were not sufficient to compensate for the defect in glucose metabolism in obesity and diabetes.

Conclusions

Augmented insulin and glucose responses during an OGTT are not sufficient to overcome the underlying defects in glucose metabolism in obesity and diabetes.

Introduction

Impaired carbohydrate oxidation is a well-recognized feature of obesity and diabetes. Investigations to date have implicated defects in multiple stages of glucose handling and metabolism, including the pancreatic response to ingestion (1), glucose absorption (2), glucose and insulin delivery to peripheral tissues (3, 4), tissue insulin receptor signaling responses (5-7), and transmembrane transport and intracellular glucose trapping by phosphorylation (7, 8) following insulin stimulation (insulin-mediated glucose uptake) or driven by mass action (glucose-mediated glucose uptake) (9-11). Most recently impaired mitochondrial function (12-14) has been added to this list.

The relative importance of these phenomena in impaired net carbohydrate oxidation in obesity and type 2 diabetes remains unclear. One argument has been that net tissue glucose uptake and oxidation are normalized by compensatory hyperinsulinemia and/or hyperglycemia, an idea that has found some experimental support (15, 16). However, other studies suggested that neither postprandial hyperinsulinemia nor postprandial hyperglycemia were sufficient to overcome the net metabolic defects in obesity and type 2 DM (17-19). Our group demonstrated using limb balance methods in humans that the net glucose delivery to skeletal muscle under steady-state conditions did not differ across insulin resistance groupings, suggesting that hyperinsulinemia was sufficient to compensate for this aspect of glucose handling (20). In a recent paper, Galgani and Ravussin (21) studied integrated whole-body glucose metabolism, measuring total carbohydrate oxidation by indirect calorimetry and production of labeled water following ingestion of labeled glucose as part of a traditional 75 gram glucose load. Nondiabetic subjects were studied, divided by clamp-derived insulin sensitivity into insulin resistant and insulin sensitive groups. There was no difference in rates of net glucose oxidation following glucose ingestion between these groups. There was however a lower glucose oxidation rate in insulin resistant subjects when adjusted for their greater glucose exposure following glucose ingestion. These results suggest that the modest hyperglycemic response was compensatory in this nondiabetic group.

We have recently utilized exhaled labeled CO₂ following ingestion of labeled glucose as an index of integrated carbohydrate handling (22-24). This method has been found to distinguish between lean, obese and obese/type 2 diabetic subjects, showing stepped decrements in areas under the curve for net glucose utilization (22, 24). Analogous to the Galgani method, by measuring a post-metabolism product of ingested glucose the breath method integrates all of the components of tissue and cellular glucose handling. This method is clearly dependent on the glucose excursion in response to ingestion, and in fact can serve as a surrogate measure of this phenomenon (24); we have also recently shown that it is inversely correlated to fasting indices of insulin resistance and to clamp-derived measures of insulin resistance (23). In contrast to effects observed using the deuterium labeling method of Galgani, our method demonstrates early and continuing divergence of tracer appearance between subject groups (arguing that this is not simply an effect of delayed glucose metabolism) (22). For current purposes the breath method offers a new opportunity to evaluate the question whether hyperinsulinemia and/or hyperglycemia following an oral challenge are sufficient to normalize tissue glucose metabolism.

Subjects, Materials and Methods

We performed secondary analyses of data from a recently published study (23). Full details of the methodology and participant characteristics are previously published (23); for convenience we provide a brief summary here. Participants were men and women, 18–65 years of age. Participants were categorized as non-obese (BMI <30 kg/m²), or obese (BMI ≥30 kg/m²). Diabetes mellitus was determined by prior physician diagnosis or defined at screening using a 2-hour 75-g oral glucose tolerance test (OGTT) under fasting conditions, applying the American Diabetes Association criteria of fasting glucose level >126 mg/dL or 2-h glucose level >200 mg/dL). Participants with diabetes could be treated with any combination of diet, exercise, insulin, or antidiabetes medications except thiazolidinediones or metformin. Volunteers were excluded if they had used thiazolidinediones within 6 months or metformin within 4 weeks. Stable antihypertensive or antihyperlipidemic medications were allowed. Volunteers were also excluded if not weight-stable for at least 6 months, if they had type 1 diabetes or rare variant forms of diabetes, were pregnant, had a concurrent acute or chronic medical illness likely to affect systemic fuel metabolism, used psychotropic medication including antidepressants, or had pulmonary disease or a history of current or past smoking.

Volunteers who qualified after this screening were scheduled for clinical research center admission on 2 consecutive days to undergo study measurements. Volunteers provided written informed consent for screening and main study participation. This project was overseen and approved by the Indiana University Institutional Review Board.

Measurements

The methodology has been previously published (23); a brief summary is provided. Body composition was measured by dual energy x-ray absorptiometry at the time of screening evaluation. OGTT and clamp studies were done in the morning following an overnight fast; participants consumed their usual diet except where provided by the study while admitted to the clinical research center. Participants with diabetes continued to take their diabetes medications, except morning treatments were withheld on each of the 2 measurement days until after the completion of the studies.

Anthropomorphic and blood pressure measures were performed the morning of the first testing procedure. Beginning at 7:30 a.m., baseline blood samples were drawn through an indwelling intravenous catheter, followed by a standard 75-g OGTT containing an added 150mg of [¹³C₆] glucose (catalog number CLM-1396; Cambridge Isotope Laboratories, Andover, MA). Breath samples were obtained at half-hourly intervals during the following 3 h. Blood samples were obtained concurrent with these breath samples. Upon completion of the OGTT, subjects were fed lunch and dinner ad libitum. Fasting from 8 p.m. was again implemented in anticipation of the second measurement day.

A 4-hour hyperinsulinemic euglycemic clamp procedure was performed on the second day using established procedures in our laboratory (20), based on the original method of DeFronzo (25). An insulin infusion rate of 120 mU/m²/min was used to ensure full suppression of endogenous glucose production in all subject populations (26, 27).

Analytical procedures

Breath [¹³C] glucose enrichment was measured using isotope ratio mass spectrometry (Metabolic Solutions, Nashua, NH). Screening laboratory measurements were performed by the Indiana University Health clinical laboratory using standard methodologies. Glucose measurements were performed at the bedside using a glucose oxidase method (YSI 2500 STAT glucose analyzer; Yellow Springs Instruments, Yellow Springs OH). Serum free fatty acids were measured using a colorimetric assay (Roche Diagnostics, Indianapolis, IN). Serum insulin levels were measured by radioimmunoassay (Millipore-Linco, St. Charles MO).

End points

Fasting and 2-hour glucose values and the area under the curve for glucose excursion above baseline to 180 min (glucose AUC180, calculated using the trapezoidal rule) were derived from OGTT glucose data. Similarly, breath measures of enrichment of exhaled breath for ¹³CO₂ were used to calculate breath AUC. In prior work we determined that the breath AUC180 value is the preferred single endpoint measure for this breath testing method (22-24).This value was used as the dependent variable for the current set of analyses.

The clamp glucose infusion rate-based calculation of glucose disposal rate (GDR) was adjusted for shifts in and out of the glucose space over each 20-min interval, with the average of values from the fourth hour of the clamp procedure taken as the GDR for each individual (23). This measure was then expressed per Kg fat-free mass (GDRffm) for the current set of analyses. The oral disposition index (oDI) was calculated using the OGTT-derived insulinogenic index (insulin30-insulin0/glucose30-glucose0) multiplied by the clamp-derived GDRffm. Free Fatty Acids (FFA) were measured in blood samples from the OGTT and the glucose clamp procedure. From these measures we calculated the Adipose Insulin Resistance Index (Adipose-IRi= FFA₀ min OGTT × Fasting plasma insulin) (28), and estimated adipose insulin sensitivity as the ability of insulin to suppress lipolysis (Δ FFA₁₂₀) = Free Fatty Acid _0min − Free Fatty acid 120min using clamp data.

Statistical analysis

Due to technical or scheduling problems we were not able to obtain DEXA measures on all participants. For the purposes of the current analyses involving body composition measures, we were therefore limited to 45 of the original 69 participants. The characteristics of this subset are described in Table 1. Where specific analyses could be done with the full dataset, these were performed in parallel; there were no systematic differences in the results obtained from the full dataset versus the subset with body composition measures available (not shown).

Table 1.

Population Descriptions. P values are presented for the one-way analysis of variance (ANOVA) comparing all three groups: ^alean versus obese, ^blean versus obese/type 2 diabetes, and ^cobese versus obese/type 2 diabetes. Adipose – IRi, adipose insulin resistance index; GDRffm, glucose disposal rate per kg fat free mass; FFA, Free Fatty Acid; OGTT, oral glucose tolerance test; Δ FFA120=FFA at time 0 of OGTT-FFA at 120 min of OGTT: DEXA, Dual- Energy X-ray Absorptiometry measurements of, whole –body tissue mass, whole-body fat mass, and whole-body lean mass

	Non-obese (33)		Obese (22)		Type 2 DM (13)
	Mean	SD	Mean	SD	Mean	SD	P(ANOVA)
Age (years)	39.1	12.3	40.2	10.0	53.2	4.0	0.011 (b, c)
Weight (kg)	74.2	10.1	105.0	18.9	95.3	18.2	<0.001
BMI (kg/m2)	25.1	2.5	35.6	5.0	30.3	4.9	<0.0001 (a, b)
Percent Fat (%)	30.8	11.0	43.9	8.9	30.3	11.3	0.001 (a,b)
Total fat mass (kg)	21.1	10.5	44.7	13.4	31.1	12.0	<0.0001 (a, b)
Total lean mass (kg)	47.0	15.2	56.4	11.0	59.8	11.6	0.04
Fasting Insulin (mU/mL)	11.4	5.2	24.0	35.8	15.6	5.6	0.23
Fasting glucose (mg/dL)	95.2	13.1	102.0	12.4	145.6	70.2	0.001
120 minute glucose (mg/dL)	107.7	35.0	116.3	26.4	275.5	102.9	<0.0001 (b, c)
OGTT Glucose AUC180 (mg/dL*min)	21391	3562	22227	4307	42896	16255	<0.0001 (b, c)
OGTT Insulin AUC180 (pmol/L*min)	9263	4547	14791	9916	10296	7876	0.08

Descriptive statistics were compared across groups using one-way analysis of variance, with subsequent pairwise testing using the Bonferroni method for variables with homogenous variance across group, or the Games-Howell method for variables with nonhomogenous variance. Age differed across the phenotypic groupings (Table 1, p=0.011), and age and sex were significant univariate determinants of the breath response (r=0.34, p=0.005 and r=-0.34, p=0.003 respectively). Therefore the analyses were adjusted for these demographic parameters. Age- and sex-adjusted univariate correlations were evaluated between the breath index of glucose oxidation and the metabolic variables of interested listed above.

Multivariable linear modeling analyses were performed to evaluate the primary determinants of the breath response, first in a hierarchical stepwise approach that allowed forced inclusion of variables of interest and then in standard forward stepwise fashion to extract the primary independent determinants. In the hierarchical modeling, age and sex were included as potential confounders. Insulin and glucose areas under the curve were evaluated initially to address the question of a compensatory effect; these parameters were also forced into the hierarchical models. To construct parsimonious models and minimize collinearity among parameters, 120 minute glucose and FFA concentrations were not included in the hierarchical models. The forward stepwise model did not include any forced parameters, and included age, sex, and variables that exhibited univariate associations with p<0.10. Adjusted breath responses were calculated from this model, for the further evaluation of relationships of adjusted breath response with insulin resistance, and to compare adjusted values across phenotypic groups. To allow implicit comparisons of the effects of incremental excursions versus total excursions of glucose and insulin, the modeling approaches were replicated using only incremental AUC parameters.

Results

The characteristics of the study participants evaluated are presented in Table 1. The expected group differences were seen in measures of obesity, glucose and insulin excursions after glucose ingestion, and fatty acid concentrations. Total fat mass differed between non-obese versus obese, and non-obese versus diabetic (P<0.0001 for ANOVA comparing all groups; p<0.05 for pairwise comparisons). There was a more modest but statistically significant difference in lean mass across groups (p=0.04), greater in the obese and diabetic subjects but not achieving significance in pairwise comparisons. Breath parameters also differed across groups, with graded reductions in tracer enrichment as obesity/dysglycemia increased (P<0.0001 across groups, all pairwise comparisons p<0.05), indicating progressively impaired net glucose uptake and oxidation despite the concurrent magnified glucose and/or insulin excursions.

Age- and sex-adjusted correlations of the breath data and various anthropomorphic and metabolic variables of interest are presented in Table 2. BMI (inversely) and insulin sensitivity (directly) were significantly related to breath ¹³CO₂ production. Total lean mass and total fat mass were also strongly inversely related to breath ¹³CO₂ production. Fasting and OGTT glucose values were inversely related to breath production of ¹³CO₂ following ingestion of labeled glucose. OGTT insulin area under the curve was not significantly related to breath ¹³CO₂ production, but the oral disposition index (a measure of early insulin production with the OGTT) was directly associated (i.e. greater early insulin production was related to greater breath ¹³CO₂). Free fatty acid concentrations were not significantly associated, but the fasting adipose insulin sensitivity index was related to breath ¹³CO₂ production.

Table 2.

Partial correlations of breath ¹³CO₂AUC180 with metabolic parameters. Pearson correlations adjusting for age and sex, presented as r value (P value). Adipose-IRi, Adipose-Insulin Resistance Index; GDRffm, glucose disposal rate per kg fat free mass; incAUC180 is the incremental AUC above baseline, whereas AUC180 is the total AUC over the 180 minute OGTT.

Correlations	∂13CO2 AUC180 min (‰*min)
BMI (kg/m2)	−0.50 (0.001)
Percent Fat (%)	−0.16 (0.30)
Total fat mass (kg)	−0.46 (0.002)
Total lean mass (kg)	−0.59 (<0.0001)
Fasting Plasma Insulin (mU/mL)	−0.25 (0.11)
Fasting glucose (mg/dL)	−0.35 (0.02)
120 minute glucose (mg/dL)	−0.43 (0.002)
OGTT Glucose AUC180 (mg/dL*min)	−0.40 (0.028)
OGTT Glucose incAUC180 (mg/dL*min)	−0.38 (0.001)
OGTT Insulin AUC180 (pmol/L*min)	−0.23 (0.14)
OGTT Insulin incAUC180 (pmol/L*min)	+0.04 (0.72)
Oral Disposition Index	+0.30 (0.049)
GDRffm (mg/kg/min)	+0.39 (0.01)
Fasting FFA (mmol/L)	−0.32 (0.055)
120 min FFA (mmol/L)	−0.31 (0.065)
Δ FFA120 (mmol/L)	−0.10 (0.56)
Adipose-IRi	−0.35 (0.034)

The results of multivariable modeling to identify the main determinants of breath ¹³CO₂ production are presented in Table 3 (Models 1-4, incorporating total glucose and insulin areas under the curve) and Table 4 (Models 5-8, incorporating incremental glucose and insulin areas under the curve). A hierarchical approach was taken to isolate effects of glucose absorption from those related to insulin sensitivity and body composition.

Table 3.

Determinants of ¹³CO₂ production following ingestion of ¹³C-labeled glucose, using total AUC for glucose and insulin. Age- and sex-adjusted models. Hierarchical linear regression modeling was performed, incorporating modules with parameters of glucose and insulin exposure (Model 1), metabolic variables (Model 2) and finally body composition measures (Model 3). Model 4 is the result of a multivariable regression incorporating using a forward stepwise approach. Values in the cells are direct beta terms (p values) for each variable as determinants of 13CO2 AUC180 within each model.

	Model 1	Model 2	Model 3	Model 4
Total Model R2adj (p value)	0.123 (0.076)	0.401 (0.004)	0.601 (9.5×10−5)	0.699 (2.6×10−8)
Change in F (p value of the change) from prior model	3.65 (0.037)	2.71 (0.033)	8.17 (0.002)	N/A
OGTT Glucose AUC180 (mg/dl*min*1000)	−1.13 (0.034)	−2.45 (0.014)	−1.35 (0.14)	+2.18 (0.044)
OGTT Insulin AUC180 (mU/dL*min*1000)	−0.81 (0.17)	+0.76 (0.42)	0.40 (0.64)
oDI		+9.10 (0.072)	+7.36 (0.08)	+7.71 (0.018)
BMI (kg/m2)		−2.76 (0.011)	−0.32 (0.82)
Fasting glucose (mg/dL)		+0.47 (0.12)	+0.27 (0.30)
120 min glucose				−0.42 (0.006)
GDRffm (mg/kg/min)		−0.51 (0.89)	+2.04 (0.60)
Fasting FFA (mmol/L)		−7.21 (0.74)	−30.07 (0.16)	−49.12 (0.001)
Adipose-IRi		−0.50 (0.19)	−0.38 (0.26)
DEXA total fat mass			+0.25 (0.76)	−0.37 (0.034)
DEXA total lean mass			−2.18 (0.002)	−1.63 (3.7×10−7)

Table 4.

Determinants of ¹³CO₂ production following ingestion of ¹³C-labeled glucose, using incremental AUC parameters for insulin and glucose. Models are constructed in parallel with those in Table 3. Age- and sex-adjusted models. Hierarchical linear regression modeling was performed, incorporating modules with parameters of glucose and insulin exposure (Model 5), metabolic variables (Model 6) and finally body composition measures (Model 7). Model 8 is the result of a multivariable regression incorporating using a forward stepwise approach. Values in the cells are direct beta terms (p values) for each variable as determinants of ¹³CO₂ AUC180 within each model.

	Model 5	Model 6	Model 7	Model 8
Total Model R2adj (p value)	0.047 (0.23)	0.376 (0.006)	0.500 (1.5×10−4)	0.668 (4.3×10−8)
Change in F (p value of the change) from prior model	1.47 (0.23)	3.99 (0.005)	8.06 (0.002)	N/A
OGTT Glucose incAUC180 (mg/dl*min*1000)	−1.77 (0.069)	−2.17 (0.027)	−1.06 (0.26)
OGTT Insulin incAUC180 (mU/dL*min*1000)	−0.33 (0.69)	+0.52 (0.52)	+0.22 (0.81)
oDI		+8.71 (0.09)	+7.10 (0.10)	+6.54 (0.048)
BMI (kg/m2)		−2.72(0.014)	−0.16 (0.91)
Fasting glucose (mg/dL)		−0.009 (0.96)	−0.005 (0.98)
120 min glucose (mg/dL)				−0.13 (0.002)
GDRffm (mg/kg/min)		−1.01 (0.79)	+1.75 (0.67)
Fasting FFA (mmol/L)		−13.91 (0.50)	−34.93 (0.09)	−44.83 (0.002)
Adipose-IRi		−0.29 (0.36)	−0.26 (0.31)
DEXA total fat mass			+0.21 (0.81)	−0.40 (0.029)
DEXA total lean mass			−2.22 (0.003)	−1.59 (1.0×10−6)

Age and sex, while related to parameters of interest in univariate analyses, were not important determinants of ¹³CO₂ production in the hierarchical models. Model 1 evaluated the contributions of total glucose and insulin exposures; this model explained approximately 15% of the variance in the ¹³CO₂ production, due to a contribution of glucose excursion but not insulin. Model 2 added metabolic parameters related to obesity and insulin resistance (which were individually strongly related to the breath parameter), early insulin response to the glucose load, and fasting glucose and fatty acid parameters. 120 minute glucose and free fatty acid values were not included due to concerns of collinearity with other included variables. Insulin sensitivity expressed per kg fat free mass was not significantly related to breath ¹³CO₂ production in this or the subsequent multiply adjusted model. Model 3 added DEXA-derived measures of body composition. Doing so removed the contribution of body mass index revealing a significant effect of lean mass but not fat mass as a determinant of ¹³CO₂ production.

With traditional forward stepwise modeling (Model 4), lean mass was again the most potent determinant of the breath response, with an inverse relationship (lower breath response with greater lean mass). Total fat mass was also modestly associated with the breath response in this model. Fasting free fatty acids, 120 minute OGTT glucose, and oDI (early insulin response to glucose) were also significant contributors to the breath response. A direct relationship with glucose excursion emerged from the selection process, along with an inverse relationship with 120 minute glucose. As with the hierarchical modeling, in this approach to modeling insulin resistance and insulin excursion were not independent determinants of the breath response.

A parallel set of analyses were performed using incremental AUC for glucose and insulin from the OGTT rather than total AUC (Table 4). The overall pattern that emerged from the hierarchical modeling was similar to those of Models 1-3, including a contribution from the incremental OGTT glucose excursion and BMI that were no longer significant after adding body composition to the model, and a final model with only lean body mass contributing significantly. The results of the forward stepwise modeling (Model 8) differed from analyses using total glucose excursion only in that the incremental glucose excursion parameter did not survive the selection process. .

Next we turned to the question of whether post challenge hyperinsulinemia and/or hyperglycemia are sufficient to overcome the effects of insulin resistance on net tissue glucose metabolism. Simple inspection of the raw data (Figure 1, left panels) suggests that the groupwise differences in insulin and glucose excursions were insufficient to compensate, as the raw breath data differed. Adjusting the breath response for the glucose or insulin exposures across the post-OGTT interval individually (Figure 1 right middle and lower panels) did not remove the group differences in breath response.

Figure 1.

Oral glucose tolerance results in the evaluated population. The panels on the right present ¹³CO₂ AUC₁₈₀ across groups, unadjusted in the right upper panel, then following simple adjustments for glucose excursion (right middle panel) or insulin excursion (right bottom panel).

It is evident from Table 1 and the modeling approaches that the groups differed in other parameters that could be contributing to the group differences in breath response. We therefore used the regression modeling results to adjust for the main determinants of the breath response (i.e. Models 4 and 8), and evaluate whether these adjusted breath values account for group differences. The results from this exercise using Model 4 are presented; essentially identical results were seen when adjustments using Model 8 were used (not shown). Figure 2 upper panel presents the adjusted breath response plotted against clamp-derived insulin sensitivity. The adjusted breath response remains significantly related to GDRffm, suggesting that this set of adjustments is not sufficient to explain the relationship of breath ¹³CO₂ production with insulin resistance. Further, the adjusted breath response remained significantly different across the subject groups (Figure 2 lower panel), suggesting that the primary factors underlying the breath response, including lean body mass, early insulin response, 120 minute OGTT glucose levels, and total glucose excursion are insufficient to account for the net metabolic defect in postprandial glucose utilization in obesity and type 2 diabetes. The lack of effect from adjusting for glucose and insulin excursion, combined with the observation that these excursions were not individually strong contributors to the breath response, argues that the hyperglycemia and hyperinsulinemia of obesity and type 2 diabetes are not sufficient to overcome the metabolic defect that is evident with an oral glucose tolerance test.

Figure 2.

Analyses of adjusted breath responses. Upper panel, adjusted breath response in relation to clamp-derived insulin resistance. Lower panel, breath responses further adjusted also for insulin resistance show persisting differences across subject groups. *Non-obese statistically different from the other groups.

Discussion

Exhaled ¹³CO₂ following ingestion of ¹³C-labeled glucose provides a measure of integrated glucose metabolism, from absorption through cellular oxidation and waste CO₂ disposal. We observed significant stepped reductions in this response in obese and obese/Type 2 diabetic individuals, but these groups also differ in a number of relevant phenomena including the insulin and glucose excursions following glucose ingestion, and tissue responsiveness to insulin. The current analyses were aimed at adjusting the breath response for these differences in order to rigorously evaluate the question of whether the augmented insulin and glucose responses were sufficient to overcome the intrinsic metabolic defects that underlie obesity and type 2 diabetes. The primary determinants of the breath response in this dataset included total lean mass, total fat mass, fasting fatty acid concentrations and measures of glucose and insulin response. In analyses adjusting for these effects, breath responses remained different across groups, arguing that overall the exaggerated insulin and glucose responses are not fully compensatory for the underlying metabolic defects.

We and others have previously found differences by obesity or insulin resistance status in the breath response following ¹³C-labeled glucose (22, 24, 29, 30). These observations, along with those currently presented, are in contrast to a recent study using an analogous method, ingesting ²H-labeled glucose and measuring appearance of labeled ²H₂O in total body water (21). In that study of 28 healthy men and women, divided into insulin resistant versus insulin sensitive subgroups by the median clamp-derived measure of insulin-stimulated glucose disposal, there was no significant difference in the rate of appearance of tracer in body water by insulin resistance status. However, there was a lower rate of tracer appearance in the insulin resistant subgroup after adjustment for the overall post-challenge glucose concentrations, suggesting that increased glucose excursions were compensating for an underlying metabolic defect in the more insulin resistant subset of their subjects. The differences between results from our studies and these data likely reflect important differences in the populations studied, with our studies evaluating a broader range of metabolic dysfunction. This interpretation is bolstered by prior observations using a water labeling method that showed significantly lower net glucose oxidation in type 2 diabetes (31), and a continuous relationship between the rate of appearance of tracer in body water and clamp-measured insulin resistance (32). Methodologic differences may also contribute to the contrasting observations; for example, ¹³C-labeled intermediates can be redirected to metabolic fates other than immediate mitochondrial uptake and oxidation. Also it is possible that there are differences in timing of the detectable appearance of the label in the detection medium.

Determinants of Integrated Glucose Handling

The particular set of determinants of the breath response that emerged from multivariable modeling is perhaps surprising. Most immediately, although 2hr OGTT glucose was inversely related to the breath response, the total glucose and insulin excursions, and the measures of insulin resistance, did not emerge as important determinants of the breath response when other factors were included. Further, while it is predictable on first principles that the total muscle mass is an important determinant of total tissue metabolism, the final relationship of this term was unexpectedly inverse in our study – the greater the muscle mass, the lower the breath response. This is presumably due to the fact that the obese and diabetic individuals, who demonstrate the lowest breath responses, have the greatest lean body mass as well as the greatest fat mass. One parameter that exhibited an expected relationship was fasting fatty acid concentration (reduced breath response with higher FFA), which emerged as a strong determinant in the multivariable modeling. This effect may reflect the balance of fuel availability and competition for mitochondrial oxidation; however, we did not design any aspect of this study to ask detailed questions about this interplay of fuel sources. Specifically designed studies will be needed to explore these possibilities.

Is the hyperinsulinemic/hyperglycemic response compensatory?

The principal result of these analyses is that the breath response remains reduced in obese normoglycemic and obese/Type 2 diabetic subjects compared to lean subjects after adjustment for relevant covariates, specifically including the glucose and insulin excursions. This indicates that the hyperglycemic and hyperinsulinemic responses following an oral glucose load are not sufficient to overcome the underlying metabolic defect in obese and type 2 diabetic subjects. This result is concordant with prior reports using dynamic insulin clamps (17, 18, 33) to mimic the OGTT response, with the obvious advantage that our results reflect native in vivo responses. Our analyses suggest that neither insulin resistance nor impairments in glucose-mediated glucose disposal are overcome.

Impaired tissue responses to insulin-mediated glucose disposal are well described features of obesity and type 2 diabetes. Our findings highlight the parallel impairment in glucose disposal, including glucose-mediated glucose disposal, which has been previously described as an important contributor to dysglycemia in Type 2 diabetes (9-11). It is also possible that other aspects of non-insulin mediated glucose handling contribute to the residual defect. The latter would include differences in intermediary metabolism, and mitochondrial defects in glucose oxidation (12, 19). Another potentially relevant factor is the meal-related change in metabolic rate (and, more generally, dynamic changes in the metabolic response following meal ingestion) (34-36) which is known to differ by insulin resistance status and could also contribute to residual group differences independent of insulin resistance. To determine whether such factors explain the quantitative differences in glucose handling we observe will require specifically designed studies that assess these phenomena.

Limitations

The principal limitations in these analyses are the reduced dataset due to unavailable body composition measures. This dataset was originally collected for the purpose of evaluating the relationship of the breath response with insulin resistance (23) and included a precise measure of insulin sensitivity, a precise measure of the breath response, and the current collection of potentially relevant covariates to allow for detailed analyses such as those presented. Given the high precision of the measures of insulin resistance the relationship with breath response in the adjusted models was unequivocally negative with the current sample size. The current analyses highlight measures that will be needed in order to further explore the questions raised here, such as calorimetry, measures of fatty acid dynamics, or measures of mitochondrial number or function.

Conclusions

The breath measurement of total glucose metabolism and oxidation is reduced in obesity and type 2 diabetes, reflecting impaired whole-body glucose metabolism following an oral glucose load. Multiple approaches to analyzing the contributions of exaggerated glucose and insulin exposure demonstrated that augmented insulin and glucose responses during an OGTT are not sufficient to overcome the underlying defects in glucose metabolism in obesity and diabetes.