Effect of Training Status on Regional Disposal of Circulating Free Fatty Acids in the Liver and Skeletal Muscle During Physiological Hyperinsulinemia

Nine male endurance athletes (e.g., triathletes, runners) and nine healthy sedentary men volunteered for the study. Endurance athletes competed at the national or international level and had engaged in regular endurance exercise for >10 years. Subjects included in the sedentary group did not exercise regularly. All study subjects were healthy, as judged by history, physical examination, and routine laboratory tests. None of the subjects were taking any medication, and all of them were instructed to suspend any physical activity and to consume a low-fat isocaloric diet, including at least 200 g of carbohydrate for 2 days before the study. Written, informed consent was obtained from all subjects after the study protocol had been accepted by the Joint Commission of Ethics of Turku University and Turku University Central Hospital.

[¹⁸F]FTHA was synthesized by previously described methods (8,13), leading to a radiochemical purity of >98%. Subjects were studied after a 10- to 12-h overnight fast. They were positioned in a 15-slice ECAT 931/08–12 tomograph (Siemens/CTI, Knoxville, TN) with a measured in-plane resolution of 6.7 mm and an axial resolution of 6.5 mm. Transmission scans were obtained, to correct subsequent emission scans for the attenuation of γ-photons. Two catheters were inserted, one into a left antecubital vein for the injection of [¹⁸F]FTHA and for the infusion of insulin and glucose, the other in a heated (70°C) contralateral radial vein for collection of arterialized venous blood. A primed-continuous infusion of insulin (48 mU · m⁻² · min⁻¹) was started, and euglycemia was maintained using a 20% glucose infusion, adjusted according to frequent plasma glucose measurements (14). After 115 min had elapsed from the beginning of insulin infusion, [¹⁸F]FTHA (178 ± 7 MBq) was injected, and a dynamic 32-min scan of the lower thoracic and liver regions (12 × 15, 4 × 30, 2 × 120, 1 × 180, and 4 × 300 s) was followed by a 15-min scan of the femoral region (6 × 150 s). Serum insulin, FFA, and plasma lactate concentrations were determined every 30 min, as detailed elsewhere (8,13). Blood samples for the measurement of arterialized plasma radioactivity and labeled metabolites were taken throughout the PET study.

All data were corrected for deadtime, decay, and photon attenuation and reconstructed in a 128 × 128 matrix. Final in-plane resolution in reconstructed and Hann-filtered (0.3 cycles/pixel) images was 9.5 mm FWHM (full width at half maximum). Serum metabolite analysis was performed as previously described (8).

Regions of interest (ROIs) were placed on two to four different planes within the right lobe of the liver. One to two small ROIs were placed in the left cardiac chamber to obtain blood time-activity curves, and care was taken to avoid spillover from the myocardial wall. For the femoral regions, ROIs were placed on the anterior, medial, and posterior muscle compartments over four adjacent planes. The nonmetabolized fraction of [¹⁸F]FTHA was used to correct the input function (8). Blood and muscle or liver time-activity curves were analyzed graphically (15), i.e., they were fitted to a linear equation in which the slope of the fitted line equals the influx rate constant of [¹⁸F]FTHA in the liver (LK_i) and in skeletal muscle (MK_i). FFA uptake indexes in the liver (LFU) and in skeletal muscle (MFU) were estimated by multiplying respective K_i values by mean serum FFA concentrations during PET scanning. For the liver, this represents a minimal estimate, given that the FFA concentration in the portal vein might be slightly higher than that in peripheral blood. To evaluate whether group differences in hepatic tracer retention might depend on tracer outflow from tissue, LK_i was fitted over subsequent time intervals, and LFU was calculated accordingly. In this analysis, a progressive decrease in the estimated parameter would indicate tracer loss (16). For skeletal muscle, previous data rule out this mechanism during insulin infusion in healthy subjects (17).

Maximal aerobic power (Vo_2max) was determined by direct analysis of respiratory gases during a continuous incremental maximal cycle ergometer test (Ergoline, Mijnhardt, the Netherlands). Body fat content was estimated from four skinfolds.

One-way ANOVA was used for group comparisons. Student’s paired t test was used for intraindividual comparisons, i.e., fast versus clamp metabolic data. Regression analyses were carried out according to standard techniques. Multivariate analysis was performed where appropriate to examine the impact of simultaneous changes in experimental parameters on regional fatty acid uptake across the study groups. All results are expressed as means ± SE. Level of significance was set at P ≤ 0.05.

Athletes had a lower percentage of fat mass (10 ± 1 vs. 17 ± 1%, P = 0.001) and higher Vo_2max (67 ± 2 vs. 48 ± 3 ml · min⁻¹ · kg⁻¹, P < 0.0001) than untrained subjects; the two groups were otherwise well matched for age (28 ± 1 vs. 27 ± 1 years, NS), BMI (21.6 ± 0.5 vs. 22.7 ± 0.6 kg/m², NS), and circulating substrate-hormone levels (Table 1). Insulin-mediated, whole-body glucose disposal rate was ∼25% higher in the athletes (Table 1).

The influx rate constant of FFA in skeletal muscle, MK_i, was significantly higher (by ∼25%) in athletes than in sedentary men (Fig. 1); the correspondent ∼25% increase in MFU (0.128 ± 0.014 vs. 0.168 ± 0.024) fell just short of statistical significance (P = 0.18), due to slightly higher serum FFA levels in sedentary individuals (Fig. 1). Normalizing for this factor [(MFU/individual FFA) × (group average FFA)] led to a significant difference (P = 0.03). In the whole study population, MK_i and MFU were positively correlated with plasma lactate levels and whole-body insulin-mediated glucose disposal (r = 0.63, P < 0.01); the latter relationship was mostly accounted for by the trained group (r = 0.65, P = 0.03), and MK_i tended to be positively correlated with Vo_2max (r = 0.45, P = 0.07).

Rapid tracer accumulation was observed in the liver, with a progressive increase of the tissue-to-blood radioactivity ratio. Linear fitting was good, with mean r values of 1.0 ± 0.0 in both groups (NS). Progressive fitting of liver data demonstrated time consistency in both groups, thus excluding the occurrence of tracer release from tissue throughout the scanning period (Fig. 2).

LK_i was decreased by ∼20% in athletes compared with sedentary men (Fig. 1). LFU was also decreased by 20%, but, similar to MFU, the differences achieved significance after being adjusted for circulating FFA levels (P = 0.04).

Multiple regression models were fitted to isolate the independent effect of endurance training on LFU. In these models, LFU was introduced as the dependent variable, and its relationships with the two categories of fitness (the independent variable) were examined while accounting for the influence of biochemical and anthropometric characteristics. An independent inverse association was found between the two training levels and LFU (P = 0.008) when adjusting for concurrent plasma lactate and insulin levels.

Collectively, the regional differences observed between the two study groups led to a 40% increase in the MFU-to-LFU ratio (per unit of tissue) in trained compared with untrained subjects (Fig. 1).

The present study verified the hypothesis that endurance training is independently associated with an increased uptake of FFAs by skeletal muscle and lower uptake by the liver during hyperinsulinemia. As a result, the MFU-to-LFU ratio was significantly shifted in favor of skeletal muscle in our athlete group. FFA uptake was quantified by PET and [¹⁸F]FTHA in the present study. Notably, circulating FFAs comprise a spectrum of different molecules; FTHA is a 17-carbon FFA analog, and the available comparison with 11C-palmitate supports its use as representative of long-chain fatty acids (18). [¹⁸F]FTHA enters mitochondria in proportion to the oxidative rate of FFA (13), but after formation of two acetyl-CoA moieties it cannot be oxidized, due to its sulfur atom, and remains trapped. The amount of tracer entering triglycerides is also mostly retained, considering the short experimental time versus the delay required for the release of newly formed triglycerides. Thus, the FFA uptake observed reflects the sum of both processes because the current technique does not allow their dissection. As compared with the leg balance technique, PET allows one to noninvasively and specifically target regional metabolism, and when limited to the present case, the lack of quantification of each intracellular metabolic pathway and rapid tracer breakdown are relative disadvantages. In the whole organism, with a total muscle mass of ∼40% of body weight (29 ± 1 kg in our subjects) and an average liver volume of 1.5 l, total MFU and LFU could be estimated from our results to ∼30% (muscle) and 25% (liver) of total serum FFA turnover, as previously measured in healthy individuals during insulin infusion studies (19). Notably, similar amounts of serum FFA were handled in the two organs despite a 15-fold difference in mass. Previous studies showing that postabsorptive lipid oxidation and skeletal muscle oxidative capacity are higher in trained compared with untrained subjects (6) rule against defective FFA breakdown as a mechanism for skeletal muscle triglyceride accumulation in athletes, and the evidence that muscle triglyceride content was unaffected during exercise led to the hypothesis that concurrent synthesis overrides hydrolysis in male subjects (20). The present study was conducted in the insulinized state, in which FFA oxidation is almost fully suppressed and triglyceride synthesis is enhanced and tightly dependent on FFA uptake. In this situation, we found a 25% increase in MFU and a 20% decrease in LFU in trained versus sedentary individuals. Such changes had been anticipated by previous evidence (12,21); to the best of our knowledge, the present study demonstrates them for the first time in humans. It should be noted that uptake values and group differences shown in the present work would most likely be amplified in a physiological postprandial situation, in which suppression of lipolysis is balanced by the introduction of lipids with the meal, and circulating FFA levels are higher than those during an insulin clamp. In this situation, an enhanced MK_i, as observed in our athletes, would support the notion of a causal link between higher MFU and increased FFA transporter/triglyceride content in endurance-trained humans (6,12,21). Further along this line of reasoning, the correlation observed between MFU or MK_i and known indicators of fitness, namely insulin-mediated glucose disposal and Vo_2max, may explain the observed link between muscle FFA transporters/triglyceride content and the degree of endurance training (6,21).

The current finding of a downregulation of LFU in athletes is compatible with the concept that higher insulin sensitivity and/or fitness may prevent excessive hepatic fat accumulation (10,11), particularly since the present experiments were conducted in a metabolic state in which hepatic triglyceride synthesis is stimulated and FFA catabolism is inhibited. Though the influence of visceral FFA release on the calculation of LFU could not be assessed, due to the lack of portal measurements, a higher suppression of visceral lipolysis in trained subjects, due to increased insulin sensitivity and decreased abdominal fat mass (22), would amplify the group differences reported here.

Different from findings in type 2 diabetic patients with defective insulin-regulated lipolysis and from (intralipid) studies revealing substrate competition at high serum FFA concentrations (4), our data showed no evidence of reciprocal inhibition between glucose and FFA uptake in skeletal muscle under normally suppressed serum FFA levels. On the contrary, skeletal muscle FFA and whole-body glucose uptake (mainly reflecting muscle uptake) were proportionally increased in all subjects, and they were augmented to the same extent, i.e., by ∼25%, in the athletes compared with the sedentary individuals. The current data underscore the composite nature of the insulin-sensitivity concept, of which the sole assessment of insulin-mediated glucose uptake represents one part only. Different substrates need to be available to support muscle performance at different intensity levels, as exercise capacity is optimized by predominant use of FFAs at low-moderate intensities and of glucose (from muscle glycogen and from blood) during high-intensity work (23). In keeping with this notion, the simultaneous increase in the uptake of both substrates, FFAs and glucose, in trained subjects might be not surprising when considering that hyperinsulinemia normally signals a situation in which provision of fuel is not rate limiting. In the whole body, glucose, but not lipid oxidation, was found to be enhanced in trained compared with sedentary subjects during euglycemic hyperinsulinemia (24). In the liver, glucose storage was shown to be increased, and net FFA uptake decreased, by prior exercise during hyperglycemia and mild hyperinsulinemia in dogs (25), which is in accord with the subsequent need for a prompt enhancement in fasting hepatic glucose production in athletes (26). In skeletal muscle, training was associated with increased protein content of GLUT4, insulin receptor, glycogen synthase, and glycogen synthase activity, enhancing glucose uptake during exercise and during insulin stimulation (27,28). Higher triglyceride stores were observed in skeletal muscle of trained men, together with elevated oxidative metabolism at rest (6) and triglyceride consumption during a bout of exercise (24); both glycogen and triglyceride repletion occurred, though with different time courses, in the postexercise period (29). Partially at variance with these findings, no influence of exercise or endurance training on muscle triglyceride stores could be found in men after a strict and prolonged dietary regimen (20). In our study, patients were instructed to consume a low-fat, high-carbohydrate isocaloric diet, reflecting the condition of the former rather the latter studies, and we cannot exclude that closer monitoring would have resulted in different findings. The cross-sectional nature of the present study does not allow us to dissect the outcome of endurance training from the effect of genetic or non–exercise-related environmental influences. Supporting the interaction between FFA and glucose metabolism, longitudinal studies (30) showed that enhanced fasting rates of fat oxidation were the strongest predictor of exercise-related improvement in insulin sensitivity. Interestingly, in animal studies (25), prior exercise augmented the capability of the liver to extract glucose under fasting and insulin-stimulated conditions.

Our findings are relevant to the therapeutic or preventive potential of exercise in patients with, or in those at risk of, type 2 diabetes, in whom numerous metabolic defects involve both glucose and lipid metabolism. In fact, drugs acting to reduce lipid content in the liver of type 2 diabetic subjects improved their glycemic control (5); those drugs were also shown to stimulate MFU in cell cultures (1), though the benefits of such an effect on whole-body metabolic homeostasis could not be explored due to the in vitro study design. In our data, the enhanced removal of circulating FFAs by skeletal muscle in the athletes may be one mechanism by which FFA uptake by other organs, e.g., the liver, was reduced, thus preventing excessive hepatic fat accumulation. The signal(s) for such organ cross-talk remain, however, putative.

We conclude that endurance training concertedly promoted insulin-mediated glucose and FFA disposal in skeletal muscle, while lowering hepatic FFA uptake. Consequently, serum FFAs were diverted from the liver to skeletal muscle. Such changes may result into a divergent pattern of fat accumulation in the two organs.