Effect of Fructose Overfeeding and Fish Oil Administration on Hepatic De Novo Lipogenesis and Insulin Sensitivity in Healthy Men

Seven healthy male volunteers without family history of diabetes were recruited by advertisement. They were aged 22–31 years and had BMIs of 20.2–25.4 kg/m² (Table 1). All subjects were in apparent good health, were nonsmokers, and took no medications. The study was approved by the ethical committee of Lausanne University School of Medicine and a written consent was obtained from each subject after the nature of the study was explained.

Standing height was measured using a stadiometer. Body weight and hip and waist circumferences were measured before the last meal preceding each study (8). Body composition was estimated from subcutaneous skin fold thickness measurements at the biceps, triceps, subscapular, and suprailiac sites as described by Durnin and Womersley (9).

Each subject was studied on four occasions (Fig. 1). On one occasion, volunteers received 7.2 g of fish oil (1.2 g eicosapentaenoic acid and 0.8 g docasahexaenoic acid; Biorganic Omega-3, Gisand, Bern, Switzerland) per day for 28 days. It has been documented that such supplementation with fish oils leads to marked increases in n-3 fatty acids in serum phospholipids (10,11). On another occasion, subjects ingested 3 g of fructose (D-Fructose; Fluka Chemie Gmbh, Buchs, Switzerland) per kilogram of body weight per day (high-fructose diet) as a 20% fructose solution with the three main meals during the 6 days before the test. On the third occasion, fish oil supplementation was combined with high-fructose diet. Each subject also underwent a control test. Volunteers were instructed to avoid certain foods and to follow a balanced, isoenergetic diet, which was controlled by a food diary during an initial 3-day period. Thereafter, during the 3 days preceding each test, the subject followed a provided isoenergetic diet (15% proteins, 35% lipids, 40% starch, 10% mono- and disaccharides) partitioned into three meals at 0700, 1200, and 1900 and two snacks at 0900 and 1500. During the high-fructose diet, subjects received the same diet supplemented with 3 g · kg⁻¹ · day⁻¹ fructose, resulting in an hyperenergetic (800–1,000 kcal/day) diet containing 11% proteins, 26% lipids, 30% starch, 8% glucose and disaccharides, and 25% fructose. Subjects were told to avoid vigorous physical activity during the 6 days preceding the tests.

The order by which the four dietary conditions were applied was randomized, with an interval of 12 weeks after the two tests with fish oil administration to allow wash-out of fish oil between the experiments.

After each dietary condition, subjects underwent an overnight 13-h study (from 2200 to 1100 of the next morning), during which they stayed in bed and slept between 2200 and 0600. On the evening of the study, they took their last meal at 1830. At 2030, two indwelling catheters were inserted: one into a right wrist vein for blood sampling; the other into a vein of the controlateral forearm for infusions. From 2200 to 0730, 0.5 g/h of [1-¹³C]sodium acetate (10 mg/ml in NaCl 0.9%) was infused constantly. Whole-body glucose turnover was assessed with 6,6-²H₂ glucose infusion (bolus: 2 mg/kg; continuous: 20 μg · kg⁻¹ · min⁻¹) between 0500 and 1100. Basal blood samples were obtained at 0600, 0700, and 0730 for determination of hepatic DNL, endogenous glucose production (EGP), insulin, glucose, nonesterified fatty acid (NEFA), and triglyceride concentrations (referred as “fasting condition”).

To assess insulin sensitivity, insulin-mediated glucose disposal (6,6-²H₂ glucose, “hot infusion model”) (12), inhibition of EGP, and suppression of lipolysis (plasma NEFA concentrations) and lipid oxidation (indirect calorimetry), a two-step (0.2 mU · kg⁻¹ · min⁻¹ from t = 0 to t = 90, then 0.5 mU · kg⁻¹ · min⁻¹ from t = 90 to t = 180) hyperinsulinemic-euglycemic clamp (13) was performed between 0800 and 1100 (Fig. 1). Blood samples were collected at 30-min intervals to measure insulin, NEFA, and triglycerides concentrations. The glucose infusion rate (insulin resistance) during the clamp was used to evaluate insulin-mediated glucose disposal. Indirect calorimetry was performed using a ventilated canopy as described previously (14) from 0700 until 1100 (Fig. 1). Energy expenditure and substrate oxidation rates were measured using the equation of Livesey and Elia (15).

Blood samples were immediately centrifuged at 4°C to separate plasma, which was then frozen at −20°C until testing. Plasma glucose and lactate concentrations were measured with a glucose-lactate analyzer YSI 2300 STAT Plus (Yellow Springs, OH). Plasma NEFA and triglycerides concentrations were analyzed by a colorimetric method using commercial kits for NEFA (NEFA C; Wako Chemicals, Freiburg, Germany) and for triglycerides (Biomérieux Vitek, Switzerland). Commercial radioimmunoassay and enzyme-linked immunosorbent assay kits were used for determination of plasma insulin (Biochem Immunosystems, Freiburg, Germany), adiponectin (Linco Research, St. Charles, MO), and resistin (Human Resistin ELISA, BioVendor Laboratory Medicine, Czech Republic). During the clamp, plasma glucose concentrations were measured by the glucose oxidase method using a Beckman glucose analyzer II (Beckman Instruments, Fullerton, CA). Plasma 6,6- ²H₂ glucose isotopic enrichment was measured by gas chromatography–mass spectrometry (Model 5973; Hewlett-Packard, Palo Alto, CA) as described (16). Plasma VLDL-¹³C palmitate enrichment and mass isotopomers were measured as described by Hellerstein et al. (17). Gas chromatography–mass spectrometry was used for analysis of isotopic enrichments of plasma fatty acid-methyl esters from VLDL. For fatty acid-methyl esters analysis, a 25-m fused DB-1 silica column was used, with electron impact ionization ion at m/z 270–272 representing the parent M0 through the M2 isotopomers (17). Fractional DNL was calculated by the isotopomer distribution analysis technique. The ratio of excess double-labeled to excess of single-labeled species (EM2/EM1) in VLDL palmitate reveals the isotope enrichment of the true precursor pool for lipogenesis (hepatic cytosolic acetyl-CoA) by application of probability logic based on the multinomial expansion. The fractional contribution from DNL to VLDL fatty acid can then be calculated by the precursor-product relationship (17).

Data are expressed as means ± SE. Statistical analyses were performed by using STATA version 8.2 (StataCorp, College Station, TX) with P < 0.05 as level of significance. Global mean difference among the various conditions was tested using the Friedman’s test. When a significant difference was found, multiple comparisons between two conditions were performed using the paired Wilcoxon test. To test the existence of a trend, we performed the Page’s test (18).

Mean fasting parameters are summarized in Table 1. The subjects’ body weights and compositions were not affected significantly by high-fructose diet or fish oil supplementation, respectively. Insulin levels did not differ significantly between the four conditions. Fasting glycemia and lactatemia significantly increased after high-fructose diet (7 ± 2 and 58 ± 12%, respectively; P < 0.05) compared with control condition, whereas fish oil had no influence.

As shown in Fig. 2, fasting EGP was significantly higher (14 ± 3%, P < 0.05) after the high-fructose diet (13.5 ± 0.2 μmol · kg⁻¹ · min⁻¹) and the high-fructose diet plus fish oil (13.5 ± 1.4 μmol · kg⁻¹ · min⁻¹) compared with fish oil alone (11.8 ± 1.3 μmol · kg⁻¹ · min⁻¹) and control subjects (11.9 ± 0.5 μmol · kg⁻¹ · min⁻¹). Fish oil did not change EGP significantly. Fasting NEFAs (Table 1) were significantly lower (−39 ± 6%, P < 0.05) after high-fructose diet (243 ± 43 μmol/l) compared with control subjects (392 ± 43 μmol/l). Again, fish oil had no significant impact on fasting NEFA concentrations. Fasting adiponectin and resistin concentrations were similar under all four conditions (adiponectin: control subjects, 7.7 ± 1.3; fish oil alone, 7.3 ± 1; high-fructose diet, 7.9 ± 1.4; high-fructose diet plus fish oil, 7.6 ± 1.5 μg/ml; resistin: control subjects, 2.2 ± 0.7; fish oil alone, 2.1 ± 0.4; high-fructose diet, 2.0 ± 0.5; high-fructose diet plus fish oil, 2.4 ± 0.5 ng/ml). Fasting glucose oxidation was higher (13.7 ± 1.8 and 13.8 ± 1.4 vs. 9.2 ± 2.2 and 7.7 ± 0.9 μmol · kg⁻¹ · min⁻¹) and lipid oxidation lower (0.24 ± 0.12 and 0.23 ± 0.09 vs. 0.36 ± 0.12 and 0.61 ± 0.07 mg · kg⁻¹ · min⁻¹) after high-fructose diet and high-fructose diet plus fish oil than after fish oil alone or control subjects (P < 0.05 for both). High-fructose diet increased fasting triglycerides (1.13 ± 0.21 mmol/l) by 79 ± 22% (P < 0.05) compared with control subjects (0.63 ± 0.07 mmol/l), as shown in Fig. 3. Fasting triglycerides concentrations were significantly lower (37%, P < 0.05) in the high-fructose diet plus fish oil (0.83 ± 0.12 mmol/l) compared with the high-fructose diet alone, but both remained significantly higher than fish oil alone, and control subjects (P < 0.05). Figure 4 shows the percentage of fractional DNL. After fish oil alone and control subjects, DNL was very low (1.9 ± 0.44% and 1.6 ± 0.34%, respectively). In contrast, DNL significantly increased after the high-fructose diet to 9.4 ± 2.8% (P < 0.05). Mean DNL tended to be lower after the high-fructose diet plus fish oil (mean DNL: 7.5 ± 1.8%) compared with the high-fructose diet alone but the difference was not significant. The relationship between DNL and fasting plasma triglycerides in all four conditions is illustrated in Fig. 5.

Figure 6 shows plasma glucose and insulin concentrations as well as insulin resistance during hyperinsulinemic-euglycemic clamp. Glycemia was maintained at 4.9 ± 0.1 mmol/l and did not differ between the four conditions. Insulin concentrations and insulin resistance did not vary significantly between the four conditions. Similarly, glucose oxidation at t = 180 (control subjects, 16.0 ± 1.0; fish oil alone, 16.3 ± 1.3; high-fructose diet, 18.8 ± 0.6; high-fructose diet plus fish oil, 19.6 ± 0.4 μmol · kg⁻¹ · min⁻¹) and lipid oxidation at t = 180 (control subjects, −0.09 ± 0.06; fish oil alone, −0.22 ± 0.08; high-fructose diet, −0.14 ± 0.10; high-fructose diet plus fish oil, −0.22 ± 0.06 mg · kg⁻¹ · min⁻¹) did not differ significantly under all four conditions.

In Fig. 2, EGP is displayed in fasting condition and at t = 90 and t = 180 into the clamp. Although suppression of EGP in high-fructose diet (28% after t = 180) and high-fructose diet plus fish oil (24% after 180 min) were not significant, suppression in control subjects (44% at t = 180) and fish oil (36% at t = 180) were both significant (P < 0.05).

Figure 7 shows NEFA concentrations expressed as absolute values (Fig. 7A) and as percent of fasting values (t = 0) (Fig. 7B). During the clamp, NEFAs were higher after high-fructose diet compared with control subjects and fish oil alone. At times t = 90 and t = 150, differences between high-fructose diet and control subjects were significant (P < 0.05). Under high-fructose diet plus fish oil, NEFA concentrations tended to be lower compared with high-fructose diet alone throughout the clamp, but the difference did not reach statistical significance (Fig. 7A). In all four conditions, NEFA concentrations were significantly suppressed (P < 0.01) during the clamp. However, NEFA suppression was significantly stronger (P < 0.05) after control subjects (90% at t = 180) and after fish oil alone (91% at t = 180) compared with high-fructose diet (76% at t = 180) and high-fructose diet plus fish oil (77% at t = 180) (Fig. 7B).

Fructose overfeeding for 6 days led to the development of several features of the metabolic syndrome in these healthy, normal-weight volunteers. First, fructose overfeeding significantly increased plasma triglycerides concentration by 79%. This increase was associated with a sixfold stimulation of fractional hepatic DNL. Although the kinetics of VLDL particles were not assessed, it is likely that stimulation of hepatic DNL by fructose contributed to the observed hypertriglyceridemia (Fig. 5). Previous studies have indeed observed significant correlations between stimulation of fractional hepatic DNL and plasma triglycerides levels in humans fed high simple carbohydrate diets (4,19). Second, fructose increased fasting glycemia and EGP and blunted suppression of EGP during the clamp experiment, indicating hepatic insulin resistance (Fig. 2). Similar observations have been made in rodents after 2 weeks of high-sucrose diet (20). In rodents, as in our study, this increase in EGP was not associated with fasting hyperinsulinemia, suggesting that hepatic insulin resistance alone does not stimulate insulin secretion. Third, and quite unexpectedly, fructose impaired the suppression of plasma NEFA during low and moderate insulin infusion rate, indicating adipose tissue insulin resistance (Fig. 7).

The mechanisms by which fructose exerts these various effects cannot be determined from the present experiments and can only be speculated. Changes in adiponectin and resistin concentrations could have been suspected given the effects of these adipokines on whole body and hepatic insulin resistance (21,22). However, their concentrations were not altered by high-fructose diet or fish oil.

Fructose overfeeding exposes the liver to a huge carbohydrate load. The fructose carbons can be converted into glucose and glycogen, thus leading to increased hepatic glycogen stores (23,24). This process may contribute to enhanced EGP and to its impaired suppression by insulin (25). The suppression of glucose production during a hyperinsulinemic clamp may, however, not truly reflect what occurs during oral feeding when the liver will be exposed to higher concentrations of glucose due to portal glucose delivery. A portion of fructose carbons can be also converted into lactate, leading to hyperlactatemia and possible extrahepatic metabolic effects in muscle and adipose tissue (26,27). Finally, carbon atoms can be converted into fatty acids, thus enhancing hepatic DNL (28–30). However, these acute effects of fructose administration are unlikely to account for our observation of increased EGP, DNL, triglyceridemia, and lactatemia after an overnight fast (i.e., more than 10 h after ingestion of fructose). These effects are most likely explained by alterations of hepatic gene expression by high-fructose diet (28). In this regard, it is documented that high carbohydrates diets stimulate carbohydrate response element–binding protein (ChREBP), a hepatic transcription factor that upregulates the expression of lipogenic and glycolytic enzymes (31). ChREBP also regulates the expression of key enzymes of hepatic fructose metabolism. Although ChREBP is primarily regulated by glucose through the increase in intrahepatic glucose metabolites, it may possibly be stimulated as well during high-fructose diet. In addition, recent findings show that hepatic stress may mediate hepatic insulin resistance associated to high-fructose diet (32,33).

Regarding the effects of fructose overfeeding on adipose tissue, we have no mechanistic explanation to offer. Although it is generally accepted that fructose is nearly completely metabolized at first pass in the gut and the liver (34), it remains possible that some fructose reached the systemic circulation and exerted metabolic effects directly at the level of adipocytes. Alternatively, fructose metabolism in the liver may possibly exert effects on adipose tissue through the release of metabolites such as lactate (24). It could also be speculated that fructose overfeeding activated the sympathetic nervous system, as it has been demonstrated in rodents (35) and hence stimulated lipolysis.

Despite the observed hepatic and adipose tissue insulin resistance, whole-body insulin-stimulated glucose disposal was not decreased after six days of fructose overfeeding. Because skeletal muscle glucose metabolism is the major determinant of whole-body glucose disposal under conditions of hyperinsulinemia (36), this strongly suggests that muscle insulin sensitivity was not impaired. Similar observations were made in rodents after 2 weeks of a diet rich in sucrose (20). Rodents, however, were shown to subsequently develop whole-body insulin resistance after several weeks of high-sucrose feeding (20), and it is therefore likely that the duration of fructose overfeeding was too short to produce effects in skeletal muscle.

We observed that fish oil supplementation significantly blunted fructose-induced hypertriglyceridemia. The hypotriglyceridemic effect of fish oil in animals (37) and humans (38) is well known. The mechanisms by which this effect was attained can only be speculated on the basis of our data. Although hepatic DNL was not significantly reduced after fish oil supplementation, we cannot exclude that the number of subjects studied was too small to detect such an effect. Alternatively, fish oil may have reduced the secretion of triglycerides-rich particles and/or stimulated their extrahepatic clearance (39,40). Whatever the mechanism, the hypotriglyceridemic effect of fish oil was not associated with a reduction of EGP or with improved insulin-induced NEFA suppression (Fig. 2 and 7), indicating that it failed to impact the metabolic steps involved in hepatic and adipose tissue insulin resistance. In rats however, fish oil has been shown to prevent whole body insulin resistance (6) after several weeks of high sucrose feeding (41). This suggests that fish oil could have the potential to prevent development of whole-body insulin resistance induced by chronic fructose overfeeding in humans.

Our study design involved administration of an extra amount of fructose while leaving the other dietary intakes unchanged. It therefore resulted in both energy and fructose overfeeding. As such, it is representative of a condition where increased dietary fructose intake would not be compensated by a reduction of calories from other sources. It cannot, however, truly differentiate the effects of high-fructose intake per se and of energy total carbohydrate overfeeding. Only comparative studies involving subjects overfed with fructose versus starch or glucose will be able to address this issue.

In conclusion, we found a hepatic phenotype that was characterized by hepatic (and adipose tissue) insulin resistance, hypertriglyceridemia, and increased DNL after 6 days of fructose overfeeding. However, whole-body insulin-mediated glucose disposal remained unchanged, suggesting normal muscle insulin sensitivity. Experiments done in rodents, however, suggest that exposure to large amounts of fructose for several weeks may impair muscle insulin sensitivity as well. Despite its significant hypotriglyceridemic effect, fish oil supplementation did not improve the other metabolic alteration induced by fructose. Therefore, the overall consequences of fructose overfeeding over longer periods of time and the potential preventive properties of fish oil in the development of whole-body insulin resistance remain to be evaluated.

This study was funded by Grant 3200–067787 from the Swiss National Science Foundation (L.T.) and by the Nutricia Research Foundation.

Part of this work was presented as an abstract at the 1st International Congress on “Prediabetes” and the Metabolic Syndrome, Berlin, Germany, 13–16 April 2005.

The authors are indebted to Christiane Ruffieux and Guy van Melle for their precious help in statistical analyses.