Enhanced Fat Oxidation Through Physical Activity Is Associated With Improvements in Insulin Sensitivity in Obesity

The primary study group consisted of 25 obese (BMI >30 kg/m²) volunteers (9 men and 16 women), aged 39.0 ± 4 years, who participated in a combined program of exercise and caloric restriction-induced weight loss. An additional seven normal weight volunteers (four women and three men) completed an identical physical activity program but without weight loss. Improvements in insulin sensitivity in these two groups were also compared with a subset of obese men (n = 6) and women (n = 10) in a separate intervention (18) who had lost a similar amount of weight and total body fat through caloric restriction but without exercise. Baseline characteristics, including BMI, insulin sensitivity, and age, were similar in these two obese groups.

Volunteers were weight stable (±2 kg body wt) for at least 6 months before the study. None of the volunteers had type 2 diabetes, nor were they participating in any regular exercise before the study. Individuals with coronary heart disease, peripheral vascular disease, or clinically significant hyperlipidemia (plasma triglycerides >3.95 mmol/l or total cholesterol levels >7.76 mmol/l) were excluded. Individuals with treated or untreated hypertension were excluded. The protocol was approved by the University of Pittsburgh Institutional Review Board, and all volunteers gave written informed consent.

Whole-body fat mass (FM) and fat-free mass (FFM) were assessed by dual-energy X-ray absorptiometry (Lunar model DPX-L; Lunar, Madison, WI) using software version 1.3Z.

Cross-sectional areas and location of adipose tissue within the abdomen and thigh were determined using computed tomography (CT) imaging (9800 CT scanner; General Electric, Milwaukee, WI) and commercially available software (Sice-O-Matic; Tomovision, Montreal, Canada). Abdominal subcutaneous and visceral adipose tissue were measured in one image acquired at the L4-L5 vertebral disc space using an established method (25). Mid-thigh muscle, adipose tissue, and muscle attenuation characteristics were also determined with CT, defined by regions of interest with attenuation values for adipose tissue (−30 to −190 Hounsfield units; [HU]) and skeletal muscle (0–100 HU). Muscle area was further characterized by its mean attenuation value within that range, representing a marker of muscle lipid content such that lower attenuation values reflect higher lipid content (26). Thigh adipose tissue was further distinguished by manual tracings as intermuscular thigh adipose tissue, subfascial adipose tissue, and subcutaneous adipose tissue, as described previously (27).

Maximal aerobic capacity (Vo_2max) was measured using an incremental protocol on an electronically braked cycle ergometer (Sensormedics, Yorba Linda, CA). Heart rate, blood pressure, and electrocardiogram were recorded before, during, and immediately following this test. Oxygen consumption (Vo₂) was calculated via direct calorimetry (Sensormedics 2900). The heart rate-Vo₂ relationship was plotted for each person in order to provide individualized exercise intensity prescriptions and also to estimate energy expenditure during their exercise sessions. The exercise test was terminated when the subject reached volitional exhaustion or had a respiratory exchange ratio (RER) ≥1.15.

At ∼8:00 a.m. at the General Clinical Research Center, following an overnight fast, subjects ingested 40 g · body surface area (m⁻²) of a glucose solution (NERL Diagnostics, East Providence, RI). Plasma glucose and insulin were measured before glucose ingestion and at 30, 60, 90, and 120 min following glucose ingestion. Total area under the oral glucose tolerance test (OGTT) curve for glucose was computed using a trapezoid approximation procedure, using zero as the baseline.

Insulin sensitivity was determined as the rate of glucose disposal R_d during the last 30 min of a 4-h hyperinsulinemic (40 mU · m⁻² · min⁻¹)-euglycemic clamp. Subjects were instructed to consume a weight-maintaining diet containing at least 200 g carbohydrate for at least 3 days before measurements of insulin sensitivity and to avoid strenuous activity for 36–48 h preceding these studies. Postintervention metabolic assessments were performed 36–48 h following the last exercise session. On the evening before measurement of insulin sensitivity, subjects received a standard dinner (42 kJ/kg; 50% carbohydrate, 30% fat, 20% protein) and then fasted overnight in the General Clinical Research Center until completion of the glucose and insulin infusions. A catheter was placed in a forearm vein for the insulin infusion (Humulin; Eli Lilly, Indianapolis, IN), and an additional catheter was inserted into a radial artery for blood sampling. Euglycemia was maintained using an adjustable infusion of 20% dextrose. No tracer was administered to determine glucose disposal since hepatic glucose production was expected to be completely suppressed at this insulin infusion rate in these nondiabetic volunteers. Plasma glucose was determined at 5-min intervals during the clamp.

Whole-body indirect calorimetry was performed in the postabsorptive state and during the last 30 min of insulin infusion, using an open-circuit spirometry metabolic monitor system (DeltaTrac, Anaheim, CA), in order to calculate fat and glucose oxidation from respiratory gas exchange (28). An overnight, timed urine collection (∼12 h duration) was obtained to estimate whole-body protein oxidation.

A 16-week program of exercise training was conducted after completion of baseline metabolic and body composition assessments. Subjects were asked to participate in a minimum of four and a maximum of six exercise sessions weekly. At least one exercise session per week was supervised for each participant to assure that the target exercise intensity and duration was achieved. Subjects were instructed on the proper use of a wireless heart rate monitors (Polar, Kempele, Finland) to record exercise duration and mean heart rate for estimation of weekly caloric expenditure. Logs of exercise sessions were kept, including exercise duration and average heart rate. For the first 4 weeks, subjects were instructed to exercise for at least 30 min per session at an intensity of 60–70% of their maximal heart rate. At week 8, volunteers performed a submaximal Vo₂ exercise test on a cycle ergometer to reestablish the heart rate-energy expenditure relationship. During weeks 5–8, exercise sessions were increased to 40 min at the same intensity. During weeks 9–16, exercise sessions were continued at 40 min and the intensity increased to 75% of maximal heart rate.

To achieve a goal of 10% weight loss, a reduction of 500–1,000 kcal/day (based on recent food records/history) and low-fat (<30% of calories from fat) diet was implemented. The weight loss program was administered individually to subjects, so that a nutritionist/behaviorist met with each subject weekly and body weight was recorded. All subjects kept detailed 7-day food records for the entire 16-week intervention. The 16-week caloric restriction-induced weight loss program conducted in the subset of obese individuals has been described previously (29).

Plasma glucose during the OGTT and glucose clamp were measured using an automated glucose oxidase reaction (2300 Glucose Analyzer; YSI, Yellow Springs, OH). Serum insulin was determined using commercially available radioimmunoassay kits (Pharmacia, Uppsala, Sweden).

Data are presented as mean ± SE, unless otherwise indicated. Changes in whole-body fatty acid oxidation, insulin sensitivity (R_d), physical fitness (Vo_2max), and body composition were compared using paired t tests. Differences in insulin sensitivity among the intervention groups were examined using a two-way ANOVA (group × time). Bivariate and multivariate linear regression analysis was used to determine whether the changes in physical fitness, fatty acid oxidation, or body composition were associated with improvements in insulin sensitivity. All statistics were performed using JMP version 3.1.6 for the Macintosh (SAS, Cary, NC).

At completion of the intervention, volunteers had lost an average of 8% (7.7 ± 1.1 kg) of initial body weight. Most of this was comprised of the loss of fat mass (5.8 ± 0.8 kg), which represented a mean loss of 19% of fat mass, as shown in Table 1, along with other changes in body composition. There was a modest decrease in FFM.

All volunteers underwent CT imaging for regional fat distribution, but unfortunately CT data for eight subjects were lost due to irreparable damage to a storage disk. Of those subjects for whom CT data are available, abdominal visceral adipose tissue was reduced by 24% and, similarly, subcutaneous abdominal adipose tissue was decreased by 21% (Table 1). Within the thigh, the largest absolute loss was of subcutaneous thigh adipose tissue, but as a proportion of baseline, loss of intermuscular thigh adipose tissue was greatest at 32% (P < 0.05), as compared with the loss of subcutaneous (21%) or subfascial (17%) adipose tissue. Mean muscle attenuation values did not change. Of the changes in regional fat depots, only the change in visceral fat was different in men and women; men lost significantly more visceral fat than women (35 vs. 15% loss, P < 0.05).

Physical fitness (Vo_2max) increased on average by 19.5 ± 2.8% (Table 1). Average weekly exercise energy expenditure for the group was 1,114 ± 124 kcal/week. The average intensity per exercise session was 7.4 ± 0.5 kcal/min, and the average energy expenditure per exercise session was 252 ± 20 kcal. Although the exercise prescription was based on uniform guidelines, there was considerable variation among participants in their average duration and intensity of exercise.

Exercise combined with caloric restriction improved (P < 0.01) insulin sensitivity by 49 ± 10% (Fig. 1) and was not different in men and women. Of the 25 volunteers who completed the intervention, only 2 subjects failed to exhibit improvements in insulin sensitivity. Improvement in insulin sensitivity was due entirely to an improvement in nonoxidative glucose disposal, which increased by 159 ± 35%. There was no change in rates of insulin-stimulated glucose oxidation (Fig. 1 and Table 2). Systemic energy expenditure during insulin-stimulated conditions was not altered by the intervention (Table 2).

Oral glucose tolerance did not change as a result of the intervention (area under the curve: pre-OGTT 846 ± 34 mmol · l⁻¹ · min⁻¹; post-OGTT 849 ± 33 mmol · l⁻¹ · min⁻¹), and insulin secretion during the OGTT, as reflected by the 2-h insulin value, was only modestly reduced (P < 0.05) from 49.9 ± 7.6 to 38.3 ± 5.6 μU/ml.

The rate of systemic energy expenditure expressed per kilogram FFM did not change following intervention (Table 2), though the absolute rate of energy expenditure was reduced, reflecting a reduction in FFM (Table 1). Systemic postabsorptive respiratory quotient (RQ) was significantly lower following intervention, reflecting a 25% increase (P < 0.01) in resting fat oxidation. Correspondingly, postabsorptive glucose oxidation was significantly reduced (Table 2). The proportion of energy derived from fat oxidation during fasting conditions increased from 38 ± 3 to 52 ± 3%. The contribution of protein oxidation during postabsorptive conditions was minor (7.0 ± 0.5 mg/min; <4%) and did not change following intervention. The intensity of exercise and the amount of weekly exercise performed, but not changes in fat mass, were correlated with the change in fasting RQ (Table 3).

We examined, initially using bivariate analyses, potential correlates of the improvement in insulin-stimulated glucose metabolism induced by the weight loss and physical activity intervention. We also examined whether there were baseline (preintervention) characteristics that were predictive of the intervention-induced change in insulin sensitivity (Table 4). Those who were more insulin resistant before the intervention had greater improvements in insulin sensitivity, as did those who had low fasting rates of fat oxidation at baseline. This may imply that those who need the intervention the most, i.e., those who are more insulin resistant, are indeed the ones who benefit the most (have greater improvements). It is possible, however, that this was influenced by regression to the mean, resulting in an overestimation of the association between baseline and change values. Initial patterns of adipose tissue distribution or level of physical fitness did not predict improved insulin sensitivity.

Next, we examined whether changes induced in body composition, fitness, and fasting metabolism predicted the improvement in insulin sensitivity (Table 4). The loss of total body fat, subcutaneous abdominal fat, and subfascial thigh fat, but not the selective loss of visceral fat or subcutaneous thigh fat, was associated with the improved insulin sensitivity. Changes in Vo_2max and exercise energy expenditure tended to be associated with the improved insulin sensitivity. However, the strongest simple correlate with improved insulin sensitivity was increased fasting rates of fat oxidation and, accordingly, the reduced fasting RQ, as shown in Table 3 and Fig. 2. Similar associations were observed with respect to the nonoxidative component of insulin sensitivity.

Stepwise multivariate regression analysis was then used to examine the interaction and interdependence of these changes in physiologic and body composition parameters. Increased fasting rates of fat oxidation emerged as the strongest predictor of improved insulin sensitivity, accounting for 52% of the variance. After accounting for increases in fasting rates of fat oxidation, the selective loss of subfascial thigh fat emerged next, followed by the loss in total body fat as independent predictors of this improvement. These three variables combined to account for 84% of the variance in the improvement in insulin sensitivity. Enhanced postabsorptive fat oxidation remained a significant correlate of improved insulin sensitivity after adjusting for the loss of body fat or improved physical fitness in the model.

To further examine whether the addition of physical activity provided more metabolic benefit than weight loss alone, we compared the findings from the current research volunteers with that attained in a separate group of obese subjects (BMI 33.3 ± 2.7 kg/m²) who had completed a weight loss intervention without physical activity and with a nonobese group who completed a physical activity intervention without concomitant weight loss. Figure 3 illustrates the improvements in insulin sensitivity in a subset of men and women (9 women and 7 men) who completed the exercise and caloric restriction intervention, as compared with those who either completed a program of caloric restriction only (29) (10 women and 6 men) or exercise without weight loss (4 women and 3 men). The subset of individuals from the two groups were matched for fat mass lost: 7.7 ± 0.8 kg in the group who exercised and 7.8 ± 0.7 kg for the group who lost weight without exercise. Subjects in these groups were of similar age (36 ± 5, 38 ± 6, and 39 ± 5 years for exercise-only, diet-only, and exercise plus diet groups, respectively). Identical protocols, including timing of the glucose clamps following intervention, were followed in each of these three groups, although it is possible that the separate interventions were confounded by time effects. Improvements in insulin sensitivity were similar in those who exercised without weight loss (16%) and in those who lost weight without exercise (22%). The improvement in the group who combined dieting with exercise, however, was 61%, indicating a threefold higher (P < 0.01) improvement in insulin sensitivity compared with either the exercise-only group or the weight loss-only group.

Physical activity and moderate sustained weight loss are advocated for the treatment of obesity, the insulin resistance syndrome, and the prevention of type 2 diabetes. It is not clear how physical activity and weight loss interact in achieving these effects. Even a single session of moderate to strenuous intensity exercise can induce a transient improvement in insulin sensitivity (13), effects that appear to wane within a few days (15). Exercise also increases oxidative enzyme activity in skeletal muscle and induces related biochemical and morphologic changes that would seem to confer a metabolic basis for improved insulin sensitivity. However, in clinical interventions carefully controlled with respect to matching energy balance, exercise in the absence of weight loss has been found to have modest (17,30) or no (16) effect on insulin sensitivity. These findings have suggested that exercise might be of lesser value than weight loss as an intervention for the insulin resistance syndrome. We sought to address this apparent conundrum in two ways. First, we employed exercise interventions in conjunction with weight loss, an intervention model that more appropriately matches clinical recommendations than either intervention in isolation. Second, we sought to employ a more comprehensive characterization of “insulin resistance,” including aspects of both insulin-stimulated and fasting metabolism.

The improvement in insulin sensitivity in these obese subjects was greater than those observed for other studies examining exercise or weight loss alone (16,17). Similarly, Bogardus et al. (30) reported a 27% improvement in insulin sensitivity when exercise was superimposed upon diet, as compared with no improvement with diet only. Ross et al. (17) found that weight loss induced by either caloric restriction or exercise improved insulin sensitivity by 60%, as compared with a 30% increase with exercise without weight loss. Therefore, these results suggest a synergistic effect of weight loss and exercise to improve insulin resistance. The timing of the insulin sensitivity measure after the last exercise bout could also affect the magnitude of the improvements in insulin sensitivity across studies. It is important to account for the influence of acute exercise on insulin sensitivity. On the other hand, delaying the insulin sensitivity measure too long after the last exercise may result in a “de-training” effect and thereby mask the effects of chronic exercise. We chose 36–48 h after the last exercise bout to perform the glucose clamp to avoid both the acute exercise effects and potential de-training effects. This likely contributed to greater improvements with exercise training compared with other studies that performed the insulin sensitivity measure 4–6 days following the last exercise bout (16,17,30).

Improved insulin sensitivity through either weight loss or exercise has been associated with the loss of abdominal visceral fat (17,29). Obese subjects in the present study lost a significant amount of fat from several region-specific depots, including abdominal fat and subfascial thigh fat. In multivariate analysis, only the loss of subfascial thigh fat was independently correlated with the improvement in insulin sensitivity after adjusting for the change in systemic fatty acid oxidation. The attenuation of muscle on CT as a marker of muscle lipid (26) did not change, although muscle attenuation can increase with diet-induced weight loss (29). However, lipid within muscle can be higher in endurance-trained athletes (23), raising the possibility that weight loss and exercise have counterbalancing effects on muscle lipid, perhaps explaining the lack of change in muscle attenuation.

Our findings indicate that the combination of exercise and weight loss enhances not only the insulin-stimulated capacity for glucose utilization, but also enhances the capacity for fat oxidation during fasting conditions. The improvement in fat oxidation during fasting conditions was associated with characteristics specific for the exercise intervention, namely intensity and duration of physical activity, but was not associated significantly with overall or regional loss of adiposity. In turn, the magnitude of improvement in insulin-stimulated glucose metabolism was strongly related to the concomitant increase in fat oxidation during fasting conditions.

Impaired insulin-stimulated glucose metabolism is the most well-established manifestation of skeletal muscle insulin resistance, but it is also recognized that another very important facet of skeletal muscle insulin resistance is altered patterns of fat oxidation. Normally, insulin effectively suppresses fat oxidation (31), but this suppression of fat oxidation is impaired in insulin resistance (18). Additionally, it has long been noted that in those with normal insulin sensitivity, skeletal muscle has a high reliance upon fat oxidation during fasting conditions (32). In contrast, among those with obesity and insulin resistance, and in those with type 2 diabetes, fasting rates of fat oxidation are reduced in skeletal muscle (19). The decrease in fasting rates of fat oxidation and the lack of further suppression of fat oxidation during insulin are facets of insulin resistance that we have previously characterized as “metabolic inflexibility” (33). In prior cross-sectional studies, we observed that impaired fat oxidation during fasting predicts severity of insulin-resistant glucose metabolism (18). We had also noted that following diet-induced weight loss, but without changes in patterns of physical activity, there was improvement in insulin-stimulated glucose metabolism but there was not a significant change in fasting rates of fat oxidation (18). Thus, in weight loss without exercise, there appeared to be a separation in the effects on those aspects of insulin resistance related to insulin-stimulated glucose metabolism from those aspects regulating rates of fat oxidation. The current study sought to probe whether the addition of exercise might address insulin resistance in a manner different from weight loss alone.

There are several conceptual reasons to postulate that physical activity might modify the aspects of insulin resistance related to the capacity for fat oxidation. First, exercise training at moderate intensity is typically associated with induction of a higher capacity for, and reliance upon, fat oxidation during the exercise session (34). Second, chronic exercise effects, namely increased activity of oxidative enzymes and increased capillary density, might facilitate fatty acid utilization both at rest and during physical activity. Another consideration is perhaps more conceptual and is related to the “negative energy balance,” whereby during negative energy balance created by calorie reduction, muscle has a relatively passive role. In contrast, during physical activity, muscle generates a negative energy balance and is comprised of consumption of muscle glycogen and muscle triglyceride (24,35).

In contrast to the decrease in energy expenditure induced by weight loss through reductions in caloric intake (18), the lack of change in resting energy expenditure despite weight loss was likely due to the exercise component. Therefore, the current findings that an exercise intervention increases the reliance on fat oxidation while maintaining resting energy expenditure are in accord with these prior observations. The novel finding of the current study that helps to extend the clinical implications of these prior data is that the augmentation of resting rates of fat oxidation are a specific metabolic correlate of the amplitude of improvement in insulin sensitivity.

In summary, the improvement in insulin sensitivity resulting from a program combining exercise and diet is associated with an increased reliance on fat oxidation during fasted conditions. This enhanced fat oxidation is likely due to exercise and not caloric restriction. Therefore, the greater improvements in insulin sensitivity in obese subjects who perform regular exercise with weight loss, as compared with those who lose weight without exercise, are likely mediated by changes in skeletal muscle fatty acid metabolism.

This work was funded by R01DK-49200-04 (to D.E.K), K24 DK02782 (to D.E.K), and K01-AG-00851 (to B.H.G) and by the Obesity Nutrition Research (1P30DK46204) and General Clinical Research (5 M01RR00056) Centers.

We thank the volunteers for their participation in this study. We appreciate the time, dedication, and effort of the intervention staff (Pat Harper, Juliet Mancino, Anne Mathews, Donna Wolf, Kellie McCormack, and Sara Fleet) and nurses (Carol Kelley and Janet Krulia).