Racial Differences In Peripheral Insulin Sensitivity and Mitochondrial Capacity in the Absence of Obesity

Abstract

Context:

African-American women (AAW) have an increased risk of developing type 2 diabetes compared with Caucasian women (CW). Lower insulin sensitivity has been reported in AAW, but the reasons for this racial difference and the contributions of liver versus skeletal muscle are incompletely understood.

Objective:

We tested the hypothesis that young, nonobese AAW manifest lower insulin sensitivity specific to skeletal muscle, not liver, and is accompanied by lower skeletal muscle mitochondrial oxidative capacity.

Participants and Main Outcome Measures:

Twenty-two nonobese (body mass index 22.7 ± 3.1 kg/m²) AAW and 22 matched CW (body mass index 22.7 ± 3.1 kg/m²) underwent characterization of body composition, objectively assessed habitual physical activity, and insulin sensitivity with euglycemic clamps and stable-isotope tracers. Skeletal muscle biopsies were performed for lipid content, fiber typing, and mitochondrial measurements.

Results:

Peripheral insulin sensitivity was 26% lower in AAW (P < .01), but hepatic insulin sensitivity was similar between groups. Physical activity levels were similar between groups. Lower insulin sensitivity in AAW was not explained by total or central adiposity. Skeletal muscle triglyceride content was similar, but mitochondrial content was lower in AAW. Mitochondrial respiration was 24% lower in AAW and correlated with skeletal muscle insulin sensitivity (r = 0.33, P < .05).

Conclusion:

When compared with CW, AAW have similar hepatic insulin sensitivity but a muscle phenotype characterized by both lower insulin sensitivity and lower mitochondrial oxidative capacity. These observations occur in the absence of obesity and are not explained by physical activity. The only factor associated with lower insulin sensitivity in AAW was mitochondrial oxidative capacity. Because exercise training improves both mitochondrial capacity and insulin sensitivity, we suggest that it may be of particular benefit as a strategy for diabetes prevention in AAW.

African-American women (AAW) have an increased risk of developing type 2 diabetes compared with Caucasian women (CW) (1). Data from the National Health and Nutrition Examination Survey from 1999 to 2002 revealed that the prevalence of diagnosed diabetes in AAW >20 years old in the U.S. population was 12.2% compared with 4.5% in CW (1). Reasons behind this racial disparity are not entirely understood, but lower insulin sensitivity has been observed (2). Yet, the nature and etiology of lower insulin sensitivity in AAW remain poorly understood. For instance, whereas obesity is associated with insulin resistance in both skeletal muscle and liver, it is unknown whether an identical pattern contributes to the disparities in insulin sensitivity seen between AAW and CW. Furthermore, insulin resistance is associated with adiposity, particularly visceral and hepatic fat accumulation in obesity and diabetes (3, 4), but paradoxically, AAW women have lower visceral and hepatic fat (2, 5). These observations suggest that insulin sensitivity may be phenotypically distinct in AAW and not caused by traditional risk factors. Finally, physical activity plays an important role in insulin sensitivity (6), but objectively measured physical activity is rarely included in studies examining racial differences in insulin sensitivity.

Skeletal muscle characteristics such as fiber composition, lipid accumulation, and mitochondrial oxidative capacity have been shown to be associated with insulin resistance in this tissue (7, 8). Limited data suggest that AAW may have a lower percentage of type I oxidative muscle fibers (9) and lower oxidative capacity in the calf muscle (10). These observations suggest that phenotypical differences in skeletal muscle may underlie the differences in insulin sensitivity between AAW and CW.

In this study, we tested the hypothesis that the lower insulin sensitivity of lean AAW is predominantly attributable to skeletal muscle, not liver, and examined the extent to which it could be explained by differences in physical activity. In addition, we examined whether lean AAW manifest differences in skeletal muscle lipid accumulation and mitochondrial oxidative capacity that might be linked to such a metabolic phenotype. To address these issues, young nonobese AAW were compared with matched CW using euglycemic clamps and muscle biopsies to characterize insulin sensitivity, fiber-type composition, lipid accumulation, and mitochondrial oxidative capacity.

Subjects and Methods

Study participants

Twenty-two AAW and 22 CW matched for body mass index (BMI), weight, and age were enrolled. Inclusion criteria included age >18 years, stable body weight over the past 3 months, nonpregnant or lactating, and sedentary (<20 minutes of activity, 3×/wk). Exclusion criteria included significant disease or unstable medical condition, diabetes mellitus, and any drug treatment that alters glucose metabolism. All research participants gave written informed consent. The protocol was approved by the University of Pittsburgh institutional review board.

Participant characteristics

Body composition was assessed by dual-energy x-ray absorptiometry (Lunar iDXA; GE Healthcare). Fat distribution was assessed by examining fat within the trunk region and in the arms and legs (appendicular fat). The proportion of body fat within the trunk was calculated by dividing trunk fat by appendicular plus trunk fat. A graded exercise test was conducted with an electronically braked cycle ergometer (maximal oxygen uptake [VO₂max] test). Free-living physical activity was assessed using multisensor activity monitors (SenseWear MF Armband; BodyMedia) (11) for approximately 1 week (6.1 ± 1.4 days). After completion of these tests, an inpatient overnight stay was scheduled (at least 3 days after exercise test). A schematic detailing study procedures is presented in Figure 1.

Figure 1.

Schematic of timing and details of study procedures.

Insulin sensitivity and metabolic flexibility

Participants reported to the Clinical and Translational Research Center at ∼5:00 pm for an overnight stay where they received a standardized meal (10 kcal/kg; 50% carbohydrate, 15% protein, 35% fat) and then fasted until completion of study procedures. The following morning, baseline blood was collected, and a primed (210 mg/m²), continuous (2 mg/min/m²) infusion of [6,6-²H₂]glucose was initiated for assessment of endogenous glucose production. A 2-step euglycemic clamp was started with a 2-hour infusion of insulin (Humulin-R) at 15 mU/m²/min, followed by 2 hours at 40 mU /m²/min (12) (see Figure 1). Euglycemia (89 ± 4 mg/dL) was maintained with a variable 20% dextrose infusion enriched with [6,6-²H₂]glucose. Rates of glucose disposal (M) and endogenous glucose production (EGP) were calculated by non–steady-state equations based on plasma [6,6-²H₂]glucose enrichment determined by gas chromatography mass spectrometry (12, 13). The primary measure of peripheral insulin sensitivity (primarily skeletal muscle) (14) was the M/I parameter, which is calculated as the rate of glucose disposal (M) divided by plasma insulin (I) concentration (15). The hepatic insulin resistance index, a metric of hepatic insulin sensitivity, was calculated as the product of basal EGP × fasting insulin concentration (16). Insulin levels were measured by ELISA (Millipore Human Insulin Elisa Kit EZHI-14K). Plasma free fatty acid levels were measured using gas chromatography with flame ionization detection (17).

Carbohydrate and fat oxidation were determined using indirect calorimetry (Parvo Medics' TrueOne 2400; Metabolic Measurement Systems). This procedure took place before the clamp was initiated and repeated during the last 30 minutes of each step of insulin infusion to assess metabolic flexibility (calculated as steady-state respiratory quotient [RQ] minus baseline RQ). Nonoxidative glucose disposal was calculated by subtracting glucose oxidation from total glucose disposal.

Skeletal muscle biopsy and tissue analysis

After baseline calorimetry and before the start of the insulin infusion, a percutaneous muscle biopsy (vastus lateralis) was performed (18). The biopsy sample was cleared of blood and adipocytes before being partitioned for analyses. From the total biopsy sample, ∼20 mg was placed into a standardized buffer for high-resolution respirometry (19), ∼30 mg was placed in mounting medium and placed in isopentane cooled with liquid nitrogen for 2 to 3 minutes and then placed into liquid nitrogen for histochemical analysis (18), and ∼10 mg was placed in 2.5% glutaraldehyde for transmission electron microscopy (EM).

High-resolution respirometry

Muscle samples were permeabilized with saponin (2 mL of 50 μg/mL saponin in buffer solution) and treated with blebbistatin (25μM) to inhibit contraction before placement into the Oxygraph system (Oroboros, Oxygraphy-2k; Oroboros Instruments) held at 37°C. Basal (state 4) respiration was determined through the addition of glutamate (5mM), malate (2mM), succinate (10mM), and palmitoylcarnitine (25μM) (19). Maximal coupled (state 3) respiration was then assessed using an ADP titration from 37.5μM to 4000μM calculated as Vmax using standard Michaelis-Menten kinetics. Maximal uncoupled (state U) respiration was then determined after the addition of p-trifluoromethoxy carbonyl cyanide phenyl hydrazone (2μM, 3 additions). Cytochrome c (10μM) was added to assess the integrity of the outer mitochondrial membrane. There was no significant effect of cytochrome c on respiration, nor was there a racial difference in response (data not shown).

Histochemical analysis

Muscle fiber type was assessed using antimyosin monoclonal antibodies (12). Neutral intramyocellular lipid content (primarily triglycerides) was measured using semiquantitative Oil Red O staining (18). Succinate dehydrogenase activity, a marker of oxidative capacity, was measured using histochemical methods (18).

Mitochondrial content

Mitochondrial content was determined by the parameter of mitochondrial volume density measured by standard stereological methods as in previous studies (12, 13). This parameter expresses the percentage of cell volume occupied by mitochondria and reflects mitochondrial content. For each biopsy specimen (n = 12 AAW, n = 14 CW), at least 3 random longitudinal tissue sections were obtained and imaged at ×30 000 magnification. For each specimen, 18 to 25 random micrographs were acquired and downloaded into digital image analytical software (Metamorph version 6.3; Molecular Devices Corp). A144-point quadratic grid for point counting was overlapped for quantification of mitochondrial fractional density in a blinded fashion.

Statistics

Participant characteristics and skeletal muscle characteristics were compared using general linear models ANOVA (Proc GLM, SAS version 9.3 for Windows; SAS Institute Inc). Parameters assessed during the 2-step hyperinsulinemic-euglycemic clamp were compared using 2-way ANOVA (primary analysis is state by race but also as time by race to assess early non–steady-state time points). The primary measure of peripheral insulin sensitivity (glucose disposal rate divided by insulin level at steady state 2) was not normally distributed. Therefore, statistical comparisons of peripheral insulin sensitivity were conducted using log-transformed data. Adjustments for some parameters were made using ANOVA with appropriate covariates (eg, fat-free mass [FFM], proportion of type 1 fibers, VO₂max, mitochondrial density, and plasma insulin). Data are presented as means ± SD or least squares means ± SEM when covariates are used. Post hoc tests for differences in group means were accomplished using Fisher's least significant difference test. Pearson correlation coefficients were used to examine associations between measured parameters. Stepwise regression was used to model peripheral insulin sensitivity.

Results

Characteristics of participants (Table 1)

Table 1.

Participant Characteristics^a

	AAW (n = 22)	CW (n = 22)	P
Age, y	22.8 ± 4.0	24.3 ± 5.5	.29
Body composition
BMI, kg/m2	22.7 ± 3.1	22.7 ± 3.1	1.00
Weight, kg	62.0 ± 9.2	63.0 ± 9.0	.73
FFM, kg	44.8 ± 5.8	42.4 ± 5.1	.16
Fat, kg	17.2 ± 5.8	20.5 ± 6.4	.08
% body fat	27.1 ± 6.7	32.1 ± 6.8	<.02
Waist, cm	74.9 ± 8.8	80.5 ± 6.7	<.03
Trunk fat, kg	6.9 ± 2.9	9.1 ± 3.8	<.03
Appendicular fat, kg	9.5 ± 3.1	10.6 ± 3.0	.25
Fat distribution (trunk/total), %	41 ± 5	45 ± 7	<.04
Physical activity and fitness
Moderate activity, min/d	100 ± 44	85 ± 57	.34
Vigorous activity, min/d	23 ± 19	27 ± 43	.72
Steps/d	9958 ± 4472	8721 ± 3452	.31
VO2max, ml/min	2020 ± 410	2320 ± 491	.04
VO2max, ml/kg FFM/min	45.1 ± 6.9	54.4 ± 7.8	<.01
Plasma metabolic measures
Fasting glucose, mmol/L	4.7 ± 0.4	4.6 ± 0.3	.32
Fasting insulin, pmol/L	24 ± 16	22 ± 12	.66
Fasting free fatty acids, μmol/L	405 ± 168	420 ± 182	.64

^aValues are mean ± SD.

The AAW and CW were similar in age, BMI, and weight. Despite similar BMI, body fat (percentage), waist circumference, and fat within the trunk area were lower in AAW. Objectively measured habitual physical activity levels were similar between groups, with no differences in time spent engaged in moderate (metabolic equivalent level >3 and <6) or vigorous (metabolic equivalent level >6) physical activity or in ambulation (steps per day). Maximal aerobic capacity (VO₂max) was lower in AAW, suggesting differences in oxidative capacity. Fasting insulin, glucose, and free fatty acids were similar between groups.

Insulin sensitivity

Similar steady-state glucose levels were achieved in both groups (Table 2). As shown in Figure 2A, peripheral insulin sensitivity (M/I) was 26% lower in AAW, a difference that was noticeable even before the first steady state was reached (30 and 60 minutes). This difference was significant with (P < .001) or without (P < .001) adjustment for VO₂max as a covariate.

Table 2.

Two-Step Hyperinsulinemic-Euglycemic Clamp

	Steady State 1a		Steady State 2a		P
	AAW	CW	AAW	CW	Race	State
Glucose metabolism
Plasma glucose, mmol/L	5.0 ± 0.2	4.9 ± 0.4	4.9 ± 0.2	5.0 ± 0.2	.90	.16
Plasma insulin, pmol/L	98 ± 22	86 ± 23	265 ± 118	208 ± 71	<.03	<.0001
Glucose disposal rate, mg/min/kg FFM	4.11 ± 0.83	4.95 ± 1.13	9.46 ± 2.49	10.47 ± 3.03	<.04	<.0001
Peripheral insulin sensitivity (M/I), mg/min/kgFFM/μU insulin · mL	0.26 ± 0.08	0.37 ± 0.13	0.25 ± 0.13	0.34 ± 0.15	<.001	.48
Substrate oxidation
Metabolic flexibility (ΔRQ)	0.06 ± 0.03	0.06 ± 0.05	0.14 ± 0.04	0.12 ± 0.05	.46	<.0001
Fat oxidation rate, mg/minb	33 ± 5	44 ± 4	6 ± 5	27 ± 4	<.001	<.0001
Glucose oxidation rate, mg/minb	107 ± 11	103 ± 10	172 ± 10	157 ± 10	.38	<.0001
Nonoxidative glucose disposal, mg/minb	74 ± 20	112 ± 19	248 ± 20	296 ± 19	<.05	<.0001

^aSteady states 1 and 2 correspond to the last 20 minutes of insulin infusion at 15 or 40 mU/m²/min. Values are mean ± SD unless indicated otherwise.

^bLeast-squares means ± SEM, with FFM as covariate.

Figure 2.

Tissue-specific insulin sensitivity. A, Skeletal muscle insulin sensitivity; at every time point during the clamp, insulin sensitivity (M/I) was significantly lower in AAW. *, P < .01. B, EGP during the clamp was lower in AAW and CW (P < .01). C, Hepatic insulin sensitivity; EGP rate normalized for plasma insulin was similar between AAW and CW at every time point tested during the clamp, indicating no differences in hepatic insulin sensitivity between groups. D, Percent suppression of EGP by insulin was nearly identical between AAW and CW. Bars indicate mean ± SE. Black bars represent AAW; white bars represent CW.

Hepatic insulin sensitivity was similar between groups when examined by several complementary approaches. At baseline, the fasting hepatic insulin resistance index was nearly identical in AAW and CW (12.3 ± 8.9 vs 12.9 ± 9.0 mg/min/kg FFM/μU · ml, P = .82). In the hyperinsulinemic conditions of the clamp, unadjusted EGP rates were lower in AAW (P < .05) (Figure 2B). However, rates of EGP adjusted for plasma insulin revealed similar hepatic insulin sensitivity between groups (Figure 2C). The percent suppression of EGP by insulin was similar between groups (Figure 2D). Finally, when percent EGP suppression was adjusted for plasma insulin as a covariate, no racial difference was observed (steady state 1, 65.6% ± 3% vs 66.1% ± 3%; steady state 2, 99.6% ± 3% vs 94.1% ± 3% in AAW and CW, P = .31).

During the clamp, circulating free fatty acid levels were similar in AAW and CW at 30 minutes (151 ± 90 vs 137 ± 109 μmol/L, P = .66), at 60 minutes (79 ± 57 vs 86 ± 67 μmol/L, P = .81), and at 120 minutes (steady state) (60 ± 63 vs 66 ± 47 μmol/L, P = .86).

Whole-body substrate oxidation

Fasting baseline RQ was similar in AAW and CW (0.79 ± 0.04 vs 0.78 ± 0.05; P = .29). Both groups responded to insulin with comparable increases in RQ, indicating similar metabolic flexibility (Table 2). Glucose oxidation increased in response to insulin in both AAW and CW, and conversely, fat oxidation was suppressed in both groups.

Skeletal muscle tissue characteristics

Given that AAW displayed similar levels of habitual physical activity but lower aerobic capacity (VO₂max) and lower insulin sensitivity specific to skeletal muscle, we examined skeletal muscle tissue for differences in oxidative capacity. In AAW, there was a lower percentage of type I oxidative (41.1% ± 10.0% vs 48.1% ± 11.0%; P < .05) and reciprocally higher percentage of type II fibers (58.9% ± 10.0% vs 51.9% ± 11.0%; P < .05). There was also lower succinate dehydrogenase content (32.1 ± 8.5 vs 37.7 ± 6.3 arbitrary units [AU]; P < .03) and lower mitochondrial content by EM (3.18% ± 1.42% vs 4.21% ± 0.75%; P < .04). Representative electron micrographs from an AAW and a CW are presented in Figure 3, A and B. Both groups showed similar intramyocellular lipid content in type I (7194 ± 2029 vs 7009 ± 2293 AU; P = .79) and type II (3763 ± 1580 vs 4133 ± 1721 AU; P = .40) muscle fibers.

Figure 3.

Mitochondrial content and respiration. A and B, Representative electron micrographs (white arrow indicates a mitochondrion) of muscle from a CW (A) and an AAW (B). C, Mitochondrial respiration. Bars represent mean ± SE. Black bars represent AAW; white bars represent CW. *, Statistical significance. Respiration was significantly lower in AAW during state 4 respiration (basal, P < .03), at every time point during state 3 ADP titration (maximal coupled respiration, P < .001), and During State U respiration (maximal uncoupled respiration, P < .0001).

High-resolution respirometry of permeabilized muscle fibers showed lower oxidative capacity in AAW (Figure 3). This was observed for respiratory state 4 (basal, P < .03), state 3 (maximal coupled, P < .001), and state U (uncoupled, P < .0001). ADP titration revealed no significant difference in the ADP concentration at half maximal oxygen flux in AAW compared with CW (256 ± 80 vs 277 ± 79 μM; P = .40).

After adjusting for type I fiber content as a covariate, AAW still showed lower state 3 (336 ± 16 vs 396 ± 16 pmol/sec/mg dry weight, P < .02) and state U respiration (366 ± 17 vs 474 ± 20 pmol/sec/mg dry weight, P < .001). After adjusting for mitochondrial content as a covariate, state 3 (334 ± 23 vs 420 ± 21 pmol/sec/mg dry weight, P < .02) and state U (370 ± 19 vs 483 ± 19 pmol/sec/mg dry weight; P < .001) respiration were still lower in AAW.

Correlates of insulin sensitivity

Because AAW had lower adiposity and similar physical activity levels, these could not explain the lower peripheral insulin sensitivity. We examined which physiological variables were associated with peripheral insulin sensitivity (Table 3). Adiposity, VO₂max, physical activity, and basal RQ were not correlated with peripheral insulin sensitivity. There was also no association between type I or type II muscle fiber composition or skeletal muscle lipid or mitochondrial content and peripheral insulin sensitivity. However, there was an association between maximal state 3 respiration and state U respiration and peripheral insulin sensitivity.

Table 3.

Relationships Between Metabolic Characteristics and Peripheral Insulin Sensitivity

Parameter	ra	P
BMI	0.01	.97
% body fat	0.19	.22
VO2max, ml/min/kg FFM	0.12	.43
Moderate physical activity	−0.04	.78
Vigorous physical activity	−0.06	.72
Basal RQ	−0.03	.89
% type I fibers	−0.01	.96
% type II fibers	−0.16	.33
Oil Red O, type I fibers	−0.23	.15
Mitochondrial content	0.06	.78
State 3 respiration	0.33	<.04
State U respiration	0.33	<.05

^aPearson correlations between peripheral insulin sensitivity (M/I) at steady state 2 and metabolic parameters.

Discussion

AAW are at increased risk for development of type 2 diabetes (1). Although it is likely that more than one risk factor explains this elevated risk, lower insulin sensitivity is considered to be an important factor. Yet, this phenomenon is poorly understood in AAW. A number of studies employing iv glucose tolerance tests or euglycemic clamps (without isotope tracer) have reported that overweight or obese AAW have lower insulin sensitivity than CW (2, 20–22). However, because participants in those studies were overweight or obese, it is not fully understood whether AAW have a greater vulnerability to the detrimental effect of obesity upon glucose metabolism or an intrinsic difference that is present even in the absence of elevated adiposity. This distinction is important to better understand the natural history of insulin resistance and diabetes in AAW. Another unanswered question is whether liver and skeletal muscle contribute equally to the racial differences in insulin sensitivity. We are aware of no studies employing euglycemic clamps that have also employed isotope tracer methodology to differentiate between hepatic and peripheral insulin sensitivity in lean AAW. To address these issues rigorously, we employed stable isotope tracer methodology coupled with 2-step euglycemic clamps in young lean AAW and CW. Finally, given the importance of habitual physical activity and body composition on the outcomes, these were objectively measured. With this approach, novel findings were generated and revealed new insights into differences in glucose metabolism between AAW and CW.

First, we demonstrated that young lean AAW have no defects in hepatic insulin sensitivity. None of the complementary approaches revealed diminished suppression of EGP in response to insulin. In contrast, AAW had lower peripheral insulin sensitivity. In hyperinsulinemic conditions, this is largely accounted for by lower skeletal muscle tissue insulin sensitivity (14). Therefore, the phenotypic expression of lower whole-body insulin sensitivity in lean young AAW appears to be specific to skeletal muscle, and not liver. This pattern deviates from the pattern typically observed in obesity and type 2 diabetes, which affects both tissues. We do note, however, that once obesity develops in AAW, it is likely that both hepatic and skeletal muscle glucose metabolism are eventually affected. In support of this concept, we are aware of one study in AAW with a mean BMI in the overweight range. By employing iv glucose tolerance tests with stable isotope tracer and mathematical modeling, overweight AAW were shown to have reduced glucose disposal and lower suppression of EGP (23). However, as pointed out in that report, lack of statistical power prevented a subgroup analysis of lean vs overweight/obese participants, and the mathematical model has yet to be validated against model-independent measures. In contrast, our data show that in the absence of obesity, there are early manifestations of racial differences in glucose metabolism that are specific to skeletal muscle.

Increased adiposity is associated with the development of insulin resistance. However, the lower insulin sensitivity in AAW in our study was not explained by increased adiposity because total and central adiposity were actually lower in AAW. This is in line with published reports demonstrating lower central fat accumulation and lower insulin sensitivity in AAW compared with CW (2). The racial difference in insulin sensitivity was also not explained by circulating free fatty acid concentrations or intramyocellular lipid accumulation, because both were similar in AAW and CW. These observations reinforce the concept that the lower insulin sensitivity in skeletal muscle in AAW is due to mechanisms intrinsic to skeletal muscle and apparently distinct from that of excess lipid accumulation seen in obesity and type 2 diabetes.

Physical activity plays an important role in insulin sensitivity (6). For this reason, the study was designed a priori to objectively measure physical activity levels with multisensor activity monitors (11). Physical activity levels were similar in AAW and CW. Therefore, our findings, including differences in mitochondrial function and insulin sensitivity, could not be explained by racial differences in physical activity. Interestingly, despite similar physical activity levels, cycle ergometer tests revealed that AAW had a lower maximum aerobic capacity (VO₂max). Our finding that the lower VO₂max did not explain the lower insulin sensitivity in AAW has also been observed in AA girls (24) and in AAW (25). Interestingly, AA girls had a lower VO₂max, but reported higher levels of physical activity than Caucasian girls (24). However, our finding of lower aerobic capacity despite the same level of objectively assessed habitual physical activity led us to postulate that skeletal muscle tissue in AAW is phenotypically distinct not only in terms of glucose metabolism, but also in mitochondrial oxidative capacity.

To address these issues, we performed detailed studies of skeletal muscle biopsies. One factor that is associated with skeletal muscle oxidative capacity and has been shown to be different in insulin-resistant vs insulin-sensitive individuals is muscle fiber type composition (7, 8, 26). A previous study demonstrated a lower proportion of type I fibers in AAW, but this difference was observed only in obese women (9). In our study, there were fiber-composition differences, but they were subtle and did not explain the racial difference in insulin sensitivity. We did observe lower succinate dehydrogenase staining, lower mitochondrial content by EM, and lower mitochondrial respiration in AAW. The lower mitochondrial respiration was still evident after adjusting for either fiber type or mitochondrial content. Both state 3 and state U respiration were approximately 24% lower in AAW. These findings are similar to observations in individuals with type 2 diabetes and can be ascribed to an involvement of the electron transport chain and oxidative phosphorylation system (27). Overall, our findings point to a coordinated decrease in oxidative capacity at the level of both mitochondrial content and function.

Several reports have linked lower mitochondrial oxidative capacity in muscle to insulin resistance in this tissue. This has been reported in type 2 diabetes and aging, leading to a hypothesis that mitochondrial capacity is a contributory factor in the development of insulin resistance (28, 29). The association between mitochondria and insulin sensitivity is also observed dynamically, because aerobic exercise training in obese individuals resulted in increased skeletal muscle mitochondrial content, which appeared to be closely associated with the degree of improvement in muscle insulin sensitivity (13). Despite these associations, a definitive role for mitochondria in the etiology of insulin resistance has not been settled (4), and potential mechanisms remain unclear. It has been postulated that reduced mitochondrial oxidative capacity might predispose to subtle impairments in lipid oxidation and thus exaggerate intramyocellular lipid accumulation and/or generation of insulin resistance-inducing lipid metabolites (29). Other reports suggest that altered mitochondrial respiration per se may not be as important, and instead, increased mitochondrial production of reactive oxygen species may be the significant factor responsible for inducing insulin resistance (30). Alternatively, we view as plausible that the abnormalities in mitochondrial oxidative capacity in muscle may not be the key determinant of insulin resistance alone. Rather, it may represent an early biomarker of underlying abnormalities that play a role in the pathogenesis of insulin resistance.

It is interesting that in our study, the only factor correlated with the lower insulin sensitivity in AAW was mitochondrial capacity, not traditional factors. Although this does not definitively establish causation for the racial difference in insulin sensitivity, it suggests that the development of lower insulin sensitivity in AAW may be linked to mitochondrial metabolism in some way. Our findings agree with reports of lower in vivo mitochondrial metabolic flux in African-Americans assessed by ³¹P-magnetic resonance spectroscopy (10, 25) and lower mitochondrial function correlated to lower insulin sensitivity in AAW (25).

The lower insulin sensitivity in AAW was not associated with metabolic inflexibility. This finding was unexpected, because metabolic inflexibility is generally observed in insulin-resistant individuals (31) and has been linked to lower mitochondrial content in skeletal muscle (32). On the other hand, a recent report suggested higher metabolic flexibility in AAW (33). The decoupling between insulin resistance and metabolic flexibility in AAW in our study indicates that the phenotype of lower insulin sensitivity in lean AAW is, in this aspect, different from that seen in obesity and type 2 diabetes.

Although the phenotype of lower insulin sensitivity in lean AAW seems to be different from that typically reported in obesity and type 2 diabetes, we noticed that it resembles the phenotype observed in lean young offspring of parents with type 2 diabetes. The lower insulin sensitivity observed in these individuals has been reported to be due predominantly to lower skeletal muscle insulin sensitivity (34, 35), and the lower glucose disposal is due entirely (34, 36) or almost entirely (35, 37, 38) to lower nonoxidative glucose disposal. These observations resemble those we now report in lean young AAW compared with CW. The similarities also appear to exist at the mitochondrial level; lower mitochondrial content (38), oxidative capacity (36, 39), and decreased tricarboxylic acid cycle flux (40) have been observed in the offspring of parents with type 2 diabetes. These similarities suggest that the lower insulin sensitivity in AAW might involve mechanisms akin to those in individuals at genetic risk for type 2 diabetes.

Our study has certain limitations that deserve discussion. First, the correlation between mitochondrial respiration and insulin sensitivity does not conclusively establish causality. Therefore, it is unclear whether the lower oxidative capacity in AAW is part of the panoply of abnormalities that accompany insulin resistance, a marker of future type 2 diabetes, or play a contributory role in its pathogenesis. Second, we did not study men, and therefore, it is unclear whether our findings are gender-specific. Although the prevalence of diabetes is higher in AA vs Caucasian men (9.8% vs 6.0%) the racial disparity is not as great as that seen in women (12.2% vs 4.5%), and this is why we have focused our research in women (1).

In summary, we conclude that in the absence of obesity, AAW have a distinct form of lower insulin sensitivity that predominantly targets skeletal muscle, not liver, and shows intact whole-body metabolic flexibility. In addition, skeletal muscle of AAW not only displays lower glucose disposal but also lower mitochondrial content and respiration. Because this phenotype is characterized by both lower insulin sensitivity and mitochondrial oxidative capacity, exercise training is likely to be ideally suited to improve both parameters concomitantly and reduce diabetes risk in AAW.