Effects of fructose and glucose overfeeding on hepatic insulin sensitivity and intrahepatic lipids in healthy humans


  • Disclosure: LT has received research support from Nestec SA for one of the studies included in this report. Kim-Anne L is presently an employee of Nestec SA. The other authors do not have any conflict of interest with regard to this manuscript.

  • Funding agencies: This study was supported by grants Nr 138428/1 and 132935 from the Swiss national Foundation for Science to LT, CB, and RK.



To assess how intrahepatic fat and insulin resistance relate to daily fructose and energy intake during short-term overfeeding in healthy subjects.

Design and methods:

The analysis of the data collected in several studies in which fasting hepatic glucose production (HGP), hepatic insulin sensitivity index (HISI), and intrahepatocellular lipids (IHCL) had been measured after both 6-7 days on a weight-maintenance diet (control, C; n = 55) and 6-7 days of overfeeding with 1.5 (F1.5, n = 7), 3 (F3, n = 17), or 4 g fructose/kg/day (F4, n = 10), with 3 g glucose/kg/day (G3, n = 11), or with 30% excess energy as saturated fat (fat30%, n = 10).


F3, F4, G3, and fat30% all significantly increased IHCL, respectively by 113 ± 86, 102 ± 115, 59 ± 92, and 90 ± 74% as compared to C (all P < 0.05). F4 and G3 increased HGP by 16 ± 10 and 8 ± 11% (both P < 0.05), and F3 and F4 significantly decreased HISI by 20 ± 22 and 19 ± 14% (both P < 0.01). In contrast, there was no significant effect of fat30% on HGP or HISI.


Short-term overfeeding with fructose or glucose decreases hepatic insulin sensitivity and increases hepatic fat content. This indicates short-term regulation of hepatic glucose metabolism by simple carbohydrates.


There is a growing concern regarding a role of added sugars in the pathogenesis of obesity, insulin resistance, and non-alcoholic fatty liver disease (NAFLD) (1). In high fructose- or sucrose-fed rodents, it has been observed that liver fat deposition and hepatic insulin resistance occurred early after exposure to these sugars, while muscle fat deposition and muscle insulin resistance were delayed by several weeks (2). In healthy male subjects, 6 days of fructose overfeeding increased intrahepatocellular lipids (IHCL) (3) and impaired hepatic insulin sensitivity (4). Furthermore, 4-10 weeks of fructose overfeeding impaired the glycemic response to oral glucose in normal weight (5) or overweight (6) subjects. According to data reported on overweight patients with impaired glucose tolerance, a blunted suppression of hepatic glucose production (HGP) may account for such an alteration in the glycemic curve (7). In contrast, euglycemic hyperinsulinemic clamps failed to document an effect of fructose overfeeding on muscle insulin sensitivity in healthy subjects (3, 8) or in type 2 diabetic patients (9).

These observations suggest that fructose may contribute to the development of insulin resistance and NAFLD. Whether these effects are observed at usual dietary intakes, or are specifically due to fructose, remain however unknown. We, therefore, analyzed relevant data collected in several published and unpublished studies (NCT 00523562 and 00827450) in which healthy young males were subjected to an overfeeding diet with various amounts of fructose, glucose, or saturated fat for 6-7 days.

Methods and Procedures

Fifty-five healthy young males, (mean age: 22.5 ± 1.6, weight: 71.8 ± 6.4 (SD), height: 1.79 ± 0.06, mean body mass index; 22.4 ± 1.6 kg/m2), with no family history of diabetes were included. Each subject was studied on two occasions, once after 6-7 days on a balanced, weight maintenance diet (control diet; C), and another time after overfeeding for 6-7 days with the same weight maintenance diet supplemented with one of the following: 1.5 g (F1.5; n = 7), 3.0 g (F3; n = 17), or 4.0 g fructose/kg/day (F4; n = 10), 3.0 g/kg/day glucose (G3; n = 11), or 30% total energy as saturated fat (fat30%; n = 10). Fructose and glucose were provided as 20% drinks to be consumed with main meals. No major gastro-intestinal tract-related side effects were reported with regard to fructose intake. Some volunteers however reported mild abdominal discomfort and bloating during the first day, but these symptoms cleared spontaneously during the following days. The control diet and overfeeding protocol were administered in a randomized order, and were separated by at least 4 weeks, except for F1.5, in which C and F1.5 were administered sequentially.

At the end of each period, subjects had their IHCL concentrations measured with 1H magnetic resonance spectroscopy, either on a 1.5 T (F1.5, G3, fat30% and part of F3, n = 8) or on a 3 T (F3 (n = 9) and F4) MR scanner (3, 8, 10, 11). Fasting blood samples were obtained for the measurement of plasma glucose and insulin concentrations. Fasting HGP was measured by isotope dilution analysis using primed-continuous infusions of 6,6 2H2 glucose, with minor between studies differences (3, 8, 10, 11). Hepatic insulin sensitivity index (HISI) was calculated as 100/(HGP × fasting insulin concentration) (12). All protocols were approved by the Ethical Board for Human Research of the Canton de Vaud and were registered at clinicaltrial.gov (NCT 00523562 and 00827450).

Normality of data distribution was checked prior to analysis by a Shapiro–Wilk test for all variables. When required, data were normalized using the Box-Cox transformation. Comparisons were performed by mixed-model analysis and post hoc Tukey tests on both raw data, and on post-overfeeding data expressed as % control. The latter transformation was done to normalize for between studies methods variations. To evaluate the effects of the dose of fructose, data obtained in C, F1.5, F3, and F4 were compared to each other; to evaluate whether the observed metabolic changes were due to specific effects of fructose or to excess energy, data obtained in F3 (which corresponded to an average 32 ± 1% excess energy intake) were compared with G3 (37 ± 1% excess energy intake) and fat30%. Statistical significance level was set at P < 0.05.

All calculations were performed with “R” version 2.14.1 (www.cran.R-project.org).


Average body weight, fasting glucose, insulin, HGP, IHCL, and HISI obtained both after C and overfeeding are shown in Table 1. Overall, body weight was not changed significantly after each overfeeding condition.

Table 1. Body weight, blood glucose and insulin concentrations, endogenous glucose production, hepatic insulin sensitivity index, and intrahepatocellular lipids in response to overfeeding
  Fructose overfeeding Glucose overfee dingFat overfeeding
F1.5 (n = 7)F3.0 (n = 17)F4.0 (n = 10)G3.0 (n = 11)Fat30% (n = 10)
  • Data are expressed as mean ± 1 SD.

  • Control: data obtained after 6 or 7 days weight maintenance, balanced diet data observed after overfeeding are expressed both in absolute values (OF) and as % control values (OF%).

  • a

    P < 0.05 vs. C.

  • b

    P< 0.01 vs. C.

  • c

    P < 0.05 vs Fru1.5.

  • *, d

    P < 0.05 vs Fru3.0.

  • e

    %C is expressed as the average of the individual changes, and is not strictly equals the ratio of the averages.

Body weight (kg)69.3 ± 7.569.2 ± 7.3 70.7 ± 6.471.4 ± 6.3 73.2 ± 6.873.5 ± 6.8 71.9 ± 5.772.9 ± 5.7 73.8±6.873.9 ± 7.7 
Energy intake (kcal/day)2467 ± 1582882 ± 202117 ± 12654 ± 1613511 ± 234132 ± 12726 ± 1653,902 ± 273143 ± 22847 ± 1523,721 ± 216137 ± 12,749 ± 1903,573 ± 247130 ± 0
Glucose (mg/dL)88.9 ± 6.689.1 ± 2.3100.7 ± 7.593.9 ± 4.395.7 ± 5.9102.0 ± 6.592.7 ± 5.593.6 ± 7.6101.0 ± 6.590.0 ± 6.289.6 ± 7.899.5 ± 5.593.3 ± 3.895.0 ± 6.2101.9 ± 4.9
Insulin (mU/L)8.4 ± 2.09.6 ± 3.0115.7 ± 30.79.4 ± 4.212.5 ± 7.0b133.4 ± 45.2b10.7 ± 2.911.6 ± 3.2110.0 ± 19.79.0 ± 2.39.7 ± 2.6115.6 ± 43.1d8.5 ± 2.58.9 ± 2.4d106.2 ± 22.1d
EGP (mg/kg/min)2.43 ± 0.212.43 ± 0.3599.1 ± 10.82.04 ± 0.212.07 ± 0.23101.1 ± 7.81.81 ± 0.222.11 ± 0.37b,d116.3 ± 9.9b,c,d2.16 ± 0.212.30 ± 0.16b108.0 ± 11.1b,d2.09 ± 0.352.17 ± 0.33104.5 ± 10.0
HISI5.2 ± 1.64.5 ± 1.791.0 ± 32.26.1 ± 2.34.8 ± 2.2b80.0 ± 22.1b5.6 ± 1.64.4 ± 1.1a80.6 ± 14.1b5.5 ± 1.44.7 ± 1.096.0 ± 37.06.0 ± 1.45.6 ± 1.594.9 ± 24.2
IHCL (mmol/kg)6.0 ± 3.05.7 ± 2.5114.3 ± 64.59.0 ± 8.018.5 ± 15.8b,c213.3 ± 86.4b,c13.1 ± 7.923.7 ± 15.2b,c202.3 ± 114.7b,c12.9 ± 15.016.1 ± 15.1158.7 ± 92.2a,d11.6 ± 8.021.9 ± 17.2a190.4 ± 74.3b

IHCL concentrations remained unchanged after F1.5, but were nearly doubled with F3 (P < 0.01) and F4 (P < 0.001). F4 did not further increase IHCL however. IHCL increased ∼90% with fat overfeeding (P < 0.001), and were increased by about 60% with G3 (P < 0.05) (Table 1).

Fructose overfeeding did not significantly alter plasma glucose concentration at any of the doses tested. Insulin levels were unchanged in F1.5 and F4 but were increased by 33% after F3 (P < 0.01). HGP did not change significantly with F1.5 and F3, but increased significantly with F4 (P < 0.001). HISI decreased significantly with F3 and F4 (P < 0.001 and P < 0.01, respectively) (Table 1).

G3 significantly increased HGP (P < 0.01), but did neither decrease HISI, nor change fasting plasma glucose and insulin. Fat30% did not significantly alter any of these parameters.


The present data confirm the hypothesis that fructose overfeeding decreases hepatic insulin sensitivity. This occurred within one week of fructose exposure, before important changes in body composition took place. Large amounts of fructose, well above the average US consumption (13), were however required to detect significant changes in fasting HGP or HISI. In addition, fructose overfeeding nearly doubled IHCL.

The effects of short-term glucose overfeeding were, however, very similar to those of fructose. Glucose caused a significant increase in HGP. Although HISI was not decreased, this nonetheless indicates that glucose, as fructose, impaired hepatic insulin's effects. Hepatic insulin resistance may indeed result in an increased fasting insulin concentration, an increased HGP, and/or a decreased HISI (12). In our analysis, fasting insulin concentration and HISI were significantly altered with F3, while HGP was increased and HISI decreased with F4, and HGP was increased with G3. Altogether, this indicates that hepatic insulin resistance was present in all three conditions. The number of subjects included, and hence the statistical power to detect an effect of treatment, was higher with F3 than with other conditions, however.

In addition, glucose, as fructose, caused a significant increase in IHCL. This is at odds with a recent study which did not observe any increase in IHCL in healthy subjects after a 4-week overfeeding with either fructose or glucose (5). This discrepancy may possibly be explained by the higher daily amount of sugar administered in the present studies. At similar excess energy intake, the effects of fat overfeeding contrasted with those of both fructose and glucose. Although fat overfeeding led to an increase in IHCL of comparable magnitude, it did alter neither HGP nor HISI. This indicates that simple sugars, but not fat, rapidly decrease hepatic insulin sensitivity, and that this effect is not directly linked to intrahepatic fat deposition.

The present results document early effects of excess nutrients intake, occurring before clinically significant changes in body composition, but do not allow to draw conclusions regarding their long-term effects. Nonetheless, they point to a significant modulation of hepatic glucose metabolism occurring within days after exposure to excess intake of simple sugars. Consistent with this early effect of dietary sugars, it has been reported, in overweight patients, that an energy-restricted low-carbohydrate, but not a low-fat diet, improved hepatic insulin sensitivity within 2 days. Both diets led to similar weight losses and improvement of muscle insulin sensitivity after several weeks (14).

In summary, our observations indicate that hepatic insulin sensitivity is regulated by short-term changes in the intake of simple sugars, but not of fat. This was observed at very high, hypercaloric intakes for both fructose and glucose. Whether these early changes play a role in the development of metabolic diseases in the long-term, or reflect merely adaptive changes to nutrients' intake, remain to be further evaluated.


This study was performed at the Clinical Research Center of Lausanne University Hospital, and was supported by grants 320030-138428 and 310030-121995 from the Swiss National Foundation for Science to LT, CB and RK.