Higher dietary fructose is associated with impaired hepatic adenosine triphosphate homeostasis in obese individuals with type 2 diabetes§


  • Potential conflict of interest: R.J.J. published a lay book (“The Sugar Fix”) that discusses the potential role of fructose in obesity and fatty liver and has a patent application on lowering uric acid to reduce fatty liver disease.

  • The study was supported by the National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Disorders (NIH/NIDDK; grant nos.: RO1-DK060427 and UO1-DK57149) and the John Hopkins University School of Medicine General Clinical Research Center (M01-RR00052). R.J.J. is supported by grant HL-68607. M.F.A. is supported by a NIH/NIDDK K23 Career Development Award (K23-DK062116).

  • §

    This study was presented at the American Association for the Study of Liver Diseases 60th Annual Meeting, Boston, MA, October 30-November 3, 2009.


Fructose consumption predicts increased hepatic fibrosis in those with nonalcoholic fatty liver disease (NAFLD). Because of its ability to lower hepatic adenosine triphosphate (ATP) levels, habitual fructose consumption could result in more hepatic ATP depletion and impaired ATP recovery. The degree of ATP depletion after an intravenous (IV) fructose challenge test in low- versus high-fructose consumers was assessed. We evaluated diabetic adults enrolled in the Action for Health in Diabetes Fatty Liver Ancillary Study (n = 244) for whom dietary fructose consumption estimated by a 130-item food frequency questionnaire and hepatic ATP measured by phosphorus magnetic resonance spectroscopy and uric acid (UA) levels were performed (n = 105). In a subset of participants (n = 25), an IV fructose challenge was utilized to assess change in hepatic ATP content. The relationships between dietary fructose, UA, and hepatic ATP depletion at baseline and after IV fructose challenge were evaluated in low- (<15 g/day) versus high-fructose (≥15 g/day) consumers. High dietary fructose consumers had slightly lower baseline hepatic ATP levels and a greater absolute change in hepatic α-ATP/ inorganic phosphate (Pi) ratio (0.08 versus 0.03; P = 0.05) and γ-ATP /Pi ratio after an IV fructose challenge (0.03 versus 0.06; P = 0.06). Patients with high UA (≥5.5 mg/dL) showed a lower minimum liver ATP/Pi ratio postfructose challenge (4.5 versus 7.0; P = 0.04). Conclusions: High-fructose consumption depletes hepatic ATP and impairs recovery from ATP depletion after an IV fructose challenge. Subjects with high UA show a greater nadir in hepatic ATP in response to fructose. Both high dietary fructose intake and elevated UA level may predict more severe hepatic ATP depletion in response to fructose and hence may be risk factors for the development and progression of NAFLD. (HEPATOLOGY 2012;56:952–960)

The increasing prevalence of nonalcoholic fatty liver disease (NAFLD) parallels the rise in obesity and type 2 diabetes mellitus (T2DM). Patients with obesity and T2DM have not only a higher prevalence, but also more severe forms of NAFLD (i.e., steatohepatitis, hepatic fibrosis, or cirrhosis).1 The rapid rise in NAFLD supports the role for environmental factors, which, in tandem with predisposing genetic factors, likely contribute to the pathogenesis and epidemic of NAFLD. In recent decades, there has not only been an increase in total energy consumption, but also a shift in the types of nutrients consumed.

Fructose is a simple monosaccharide found in plant sources. Most commercially available fructose is in the form of a disaccharide (i.e., mixture of glucose and fructose) in the form of high-fructose corn syrup (HFCS). In the United States, fructose consumption has more than doubled in the past 30 years and has paralleled the rise in obesity and T2DM.2 Before 1900, Americans consumed approximately 15 g of fructose per day (4% of total calories), mainly through fruits and vegetables. However, by 1994, Americans consumed approximately 55 g of fructose per day (10% of total calories),3 which is primarily accounted for by the marked increase in soft-drink consumption.4, 5 Despite conservative estimates, patients with NAFLD consume 2- to 3-fold more fructose-containing beverages than matched controls.6 The parallel trend in rise of obesity, T2DM, NAFLD, and fructose consumption makes fructose an attractive target for investigation.

Fructose induces both metabolic syndrome and NAFLD independent of energy intake.7-10 In fact, free fructose and glucose combinations (such as seen with HFCS) induce fatty liver more than sucrose, despite the same fructose content.9 In overweight or obese adults, consumption of fructose-sweetened, but not glucose-sweetened, beverages increases de novo lipogenesis, promotes dyslipidemia, impairs insulin sensitivity, and increases visceral adiposity.11 Furthermore, a longitudinal study of women (n = 91,249) followed up over 8 years showed that those who consumed ≥1 serving of soft drinks per day were at twice the risk of developing T2DM as those who consumed <1 serving per month.12 Thus, it is plausible that habitual and/or excessive fructose consumption may not only increase the risk for NAFLD,6, 13 but also exacerbate liver injury and promote fibrosis progression in NAFLD.14

Unlike glucose metabolism, there is no negative feedback mechanism regulating the phosphorylation of fructose to prevent hepatic adenosine triphosphate (ATP) depletion.15 Upon entering the hepatocyte, fructose is rapidly phosphorylated by fructokinase to generate fructose-1-phosphate. Fructose-induced hepatic ATP depletion has been demonstrated with low concentrations of fructose (1 mM) in a variety of cell types16, 17 and in humans by both phosphorus magnetic resonance spectroscopy (31P MRS)18, 19 and by liver biopsy.20 Cellular ATP depletion can cause an arrest in protein synthesis and induce inflammatory and pro-oxidative changes.16, 17, 20 Consistent with these findings, HFCS increases fatty acid synthesis,21 increases endoplasmic reticulum stress, promotes activation of the stress-related kinase, Jun N-terminal kinase, induces mitochondrial dysfunction, and increases apoptotic activity22, 23 in liver cells. Habitual fructose consumption may therefore lead to an unfavorable energy balance in the liver, thus enhancing the susceptibility of hepatocytes to injury.24

Fructose metabolism also causes rapid intracellular generation of uric acid (UA). When fructose is rapidly phosphorylated, intracellular phosphate levels fall, resulting in the stimulation of adenosine monophosphate (AMP) deaminase. Consequently, the increased stimulation of AMP deaminase shunts AMP toward the production of UA, as opposed to the regeneration of ATP by AMP kinase (AMPK).25 After fructose ingestion, serum UA can increase by 1-4 mg/dL within 30 minutes.26 Furthermore, in subjects who chronically consume a high-fructose diet, fructose administration results in an enhanced rise in serum UA.26 Thus, increased UA may serve as a biomarker for increased fructose consumption and, potentially, as a marker of hepatic ATP depletion. Recent studies also suggest that UA may itself have proinflammatory and -oxidative effects16, 17 that could be involved in the development and progression of NAFLD.27-29 Finally, both cell-culture and experimental studies suggest that the continuous exposure to fructose results in the up-regulation of both transporters (e.g., Glut5) and enzymes (e.g., fructokinase) involved in fructose metabolism.30 Consistent with these data, subjects with NAFLD had higher hepatic fructokinase messenger RNA (mRNA) levels, compared to subjects with other forms of chronic liver disease (CLD).6

We proposed the following hypotheses. First, subjects with higher habitual intake of fructose may be susceptible to lower ATP levels. Second, subjects with higher UA levels (either as a consequence of increased fructose consumption or as a surrogate marker of impaired hepatic energy homeostasis) may be at increased risk for hepatic ATP depletion from increased dietary consumption of fructose. To test these hypotheses, we evaluated the relationship of dietary fructose consumption and baseline UA levels to the baseline, nadir, and duration of hepatic ATP depletion by 31P MRS in a subset (n = 25) of subjects enrolled in the Action for Health in Diabetes (Look AHEAD) Fatty Liver Ancillary Study.


ADP, adenosine diphosphate; ALT, alanine aminotransferase; AMP, adenosine monophosphate; AMPK, adenosine monophosphate kinase; ATP, adenosine triphosphate; BMI, body mass index; CLD, chronic liver disease; FFQ, food frequency questionnaire; Hb1Ac, glycosylated hemoglobin; HFCS, high-fructose corn syrup; IR, insulin resistance; IV, intravenous; Look AHEAD, Action for Health in Diabetes Study; MR, magnetic resonance; MRS, magnetic resonance spectroscopy; NAFLD, nonalcoholic fatty liver disease;NCI, National Cancer Institute; PDEs, phosphodiesters; Pi, inorganic phosphate; PMEs, phosphomonoesters; 31P MRS, phosphorous magnetic resonance spectroscopy; T2DM, type 2 diabetes mellitus; UA, uric acid.

Patients and Methods

Human Subjects.

The design and methods of the Look AHEAD study have been previously described.31 All participants were recruited by means of public advertisement and underwent complete medical history, examination, and laboratory tests to exclude viral hepatitis and other major diseases. Participants were eligible if they were between the ages of 45 and 76 years, had T2DM, a body mass index (BMI) ≥25 kg/m2, and were able to complete a maximal exercise test. Exclusion criteria included known CLD, cirrhosis, inflammatory bowel disease requiring treatment in the past year, consumption of >14 alcoholic drinks per week, previous bariatric surgery or use of weight-loss medications, uncontrolled medical conditions (e.g., glycosylated hemoglobin [HbA1c] >11% or blood pressure ≥160/100 mm/Hg), use of systemic corticosteroids, known conditions that would limit their lifespan (e.g., cancer), or their adherence to the study protocol (e.g., inability to engage in moderate exercise). Participants who weighed over 350 pounds or who had any contraindication to magnetic resonance (MR) imaging were excluded from the MR portion of the study. Between January 2002 and April 2004, 244 Look AHEAD study subjects enrolled at Johns Hopkins University (Baltimore, MD) also participated in the Look AHEAD Fatty Liver Ancillary Study. Informed consent was obtained from each participant included in the study, which was approved by the institutional review board.

Experimental Protocol.

As a part of both the parent Look AHEAD trial, participants underwent extensive data collection at baseline and screening. Age, sex, race/ethnicity, and medication use were obtained by questionnaire. Lifetime alcohol use was estimated using the validated Skinner Lifetime Drinking History questionnaire.32 Weight, height, and waist circumference were directly measured using standardized techniques. Blood samples were obtained in all patients after an overnight fast and included UA, serum aminotransferases, Hb1Ac, creatinine, and lipid levels. Serum UA was quantified by an autoanalyzer.

Usual food and nutrient intake in the preceding 6 months were obtained using a food frequency questionnaire (FFQ) modified slightly from the Diabetes Prevention Program FFQ. Estimates of food and nutrient intake were conducted by the Look AHEAD Diet Assessment Center using the National Cancer Institute (NCI) Health Habits and History Questionnaire/DietSys program (version 3.0, 1993; NCI, Rockville, MD), and the dataset was provided to the Look AHEAD Data Coordinating Center. The nutrient database was modified from the Diabetes Prevention Program database to incorporate new foods added for the Look AHEAD FFQ.33 Nutrient values for the added foods were obtained primarily from the Nutrition Data System for Research (version 4.01_30, 1999; Nutrition Coordination Center, Minneapolis, MN).

1H and 31P MRS were carried out on a 1.5-T whole body scanner (Philips Gyroscan ACS-NT; Philips Medical Systems, Best, The Netherlands), and hepatic fat and ATP were measured as previously reported.18, 34 1H MR spectra were processed in the frequency domain using an in-house software program, “CSX” (http://mri.kennedykrieger.org/). Areas under the water and fat signals were determined by integration after zero filling to 2,048 data points and exponential broadening of 3 Hz. Percentage of hepatic fat was determined according to fat * 100/(water + fat). 31P spectra data were processed using a circle-fitting (CFIT) program, as previously described.35 31P MRS allows for reproducible quantification and of phosphorus-containing metabolites.36 Hepatic 31P MRS detected in six resonances (i.e., phosphomonoesters [PMEs], inorganic phosphate [Pi], phosphodiesters [PDEs], and the nucleotide triphosphates, including γ, α, and β signals) resolved sequentially. These latter three peaks are commonly referred to as ATP signals, although the γ and α signals may include adenosine diphosphate (ADP), and uridine, guanosine, inosine, and cytosine triphosphates contribute to these signals as well.37 “Hepatic ATP” was expressed as β-ATP/total phosphorus.

A representative smaller sample (n = 25) of participants underwent a fructose challenge test,34 performed in the morning (between 6:30 a.m. and 9:30 a.m.) after an overnight fast. After intravenous (IV) catheter placement, a slow infusion of isotonic saline solution was started. After two baseline 31P MR spectra were obtained, fructose (250 mg/kg of body weight), dissolved in 100 mL of isotonic saline solution, was rapidly infused over 30-60 seconds; further spectra were then collected every 5 minutes for 1 hour. The slow saline infusion was continued until the end of the study. Of the 25 subjects who underwent a fructose challenge test, 16 subjects had FFQ and an assessment of UA.

Statistical Analysis.

This pilot study was conducted as exploratory hypothesis-generating research to assess whether hepatic ATP depletion and/or UA levels may be associated with increased dietary fructose consumption and response to IV fructose challenge. As a result of convenience sampling, this study could not be sufficiently powered to judge significance. We restricted our analyses to the 16 individuals with data on fructose intake, UA, and dynamic 31P MRS. We defined “high” fructose consumption as fructose consumption >15 g/day, a threshold in keeping with dietary fructose intake from vegetables and grains alone.3 For all the analyses, we also used tertiles of fructose and fructose as continuous variables. Hyperuricemia was defined as UA >5.5 mg/dL. Differences in baseline hepatic ATP (ß-ATP/total phosphorus), nadir value of ATP, and recovery by fructose consumption (high versus low) and UA level (hyperuricemia versus normal) were assessed using nonparametric tests because of the non-normal distribution of 31P MRS data and the small sample size. Differences in other 31P MRS parameters (e.g., α-ATP/Pi, β-ATP/Pi, γ-ATP/Pi, PME/Pi, and PDE/Pi) were also evaluated and compared among the groups. All statistical analyses were conducted using Stata 9.2 software (StataCorp LP, College Station, TX) and SAS 9.1 software (SAS Institute, Cary, NC) and were not performed at the Look AHEAD Data Coordinating Center. Differences were considered statistically significant when P values were less than 0.05. A “borderline” P value of ≤0.06 was considered noteworthy of consideration as a trend toward significance.


Clinical Characteristics of the Study Population.

Of those enrolled in the Look AHEAD Fatty Liver Ancillary Study (n = 244), 25 subjects had a successful magnetic resonance spectroscopy (MRS) and completed an IV fructose challenge test. With the exception of lower BMI and total caloric intake, our study cohort was comparable in age, gender, serum UA, total fructose intake, percent liver fat, liver biochemistries, or use of insulin-sensitizing agents, compared to those Look AHEAD subjects who did not undergo an IV fructose challenge test (Table 1).

Table 1. Characteristics of Participants Who Completed a Fructose Challenge Test
CharacteristicsIV Fructose Challenge TestP Value
Not Completed (N = 219)Completed (N = 25)
  1. Abbreviation: AST, aspartamine aminotransferase.

Age, years61.1 (6.1)60.3 (7.1)0.580
Gender, % female52580.450
Race, % white/other68880.200
BMI, kg/m236.3 (6.0)32.9 (3.0)0.001
Serum uric acid, mg/dL5.2 (1.3)4.9 (1.2)0.760
Total calorie intake, cal/day1,950 (1,391, 2,591)1,502 (1,182, 1,771)0.010
Fructose g/day16.8 (11.4, 24.3)17.1 (12.7-24.0)0.640
Liver outcomes   
 Steatosis, % liver fat5.2 (2.4, 11.6)5.8 (3.9, 16.1)0.140
 ALT21 (16, 28)23 (18, 31)0.310
 AST18 (15, 23)22 (18, 25)0.080
Use of metformin, %48460.980
Use of thiazolidinediones, %31500.070
Use of insulin, %1600.020

NAFLD (defined as >5% hepatic fat by MRS) was noted in 16 of 25 (64%) subjects. Among those with NAFLD, fat content ranged from 5% to 29%. Average fructose consumption in the high- versus low-fructose group was 22.3 ± 1.95 versus 11.13 ± 1.33 g/day (P < 0.001). Total energy intake in the high- versus low-fructose group was 1,716 ± 242 versus 1,497 ± 160 calories per day (P = 0.046). In the study cohort, serum UA in subjects with high versus low UA was 6.39 ± 0.25 versus 4.35 ± 0.18 mg/dL (P < 0.001).

Summary of the Results of the ATP Fructose Challenge Test.

Of the 25 patients who completed the IV fructose challenge test, patients with high dietary fructose consumption had lower mean hepatic ATP and ATP/total phosphate ratio at baseline, as compared to those who consumed lower amounts of fructose (Fig. 1). Patients who consumed higher amounts of fructose also had lower β-ATP/Pi and γ-ATP/Pi at nadir and 50 minutes after the IV fructose challenge. Similarly, at baseline, patients with hyperuricemia had lower mean baseline hepatic ATP levels as well as lower ATP levels at the nadir and 50 minutes after IV fructose challenge. The mean hepatic ATP/total phosphate ratio in patients with hyperuricemia dropped further than in patients without hyperuricemia (P = 0.04), suggesting less hepatic “reserve” (Fig. 2), although levels at baseline and at recovery were comparable.

Figure 1.

Response to IV fructose challenge, by fructose intake.

Figure 2.

Response to IV fructose challenge, by UA level.

Changes in 31P MRS metabolites at baseline and 50 minutes post-IV fructose challenge, as a function of the level of fructose consumption, are depicted in Table 2. When compared to baseline ATP levels, patients with high fructose consumption had significant declines in α-ATP and a trend toward a decline in β-ATP post-IV fructose challenge (P = 0.002 and 0.06, respectively). In contrast, α-ATP and β-ATP did not decline significantly from baseline in low-fructose consumers after the acute fructose challenge (P = 0.56 and 0.1, respectively). There was a significant difference between high- and low-fructose consumers after the acute fructose challenge occurred in α-ATP (P = 0.05).

Table 2. 31P MRS Metabolites at Baseline and 50 Minutes Post-IV Fructose, by Fructose Intake
MetabolitesHigh Fructose (≥15 g/day) (N = 9)P Value Δ Baseline-50 Minutes in High-Fructose ConsumersLow Fructose (<15 g/day) (N = 7)P Value Δ Baseline-50 Minutes in Low-Fructose ConsumersP Value Δ Baseline-50 Minutes in High- Versus Low-Fructose Consumers
α-ATP/total Pi     
 Baseline0.30 (0.02) 0.24 (0.02)  
 Nadir0.17 (0.03) 0.17 (0.02)  
50 minutes0.22 (0.01) 0.21 (0.02)  
 Absolute change0.08 (0.02)0.0020.03 (0.02)0.10.05
 Percent change−24.52 (5.3) −11.87 (5.8) 0.13
β-ATP/total Pi     
 Baseline12.5 (0.7) 13.2 (0.7)  
 Nadir6.4 (3.5) 7.6 (2.9)  
 50 minutes10.2 (1.0) 12.2 (1.3)  
 Absolute change2.3 (1.1)0.061.0 (1.6)0.560.48
 Percent change−17.3 (8.5) −5.3 (13.1) 0.44
γ-ATP/total Pi     
 Baseline0.13 (0.01) 0.15 (0.004)  
 Nadir0.05 (0.02) 0.06 (0.2)  
 50 minutes0.09 (0.01) 0.09 (0.01)  
 Absolute change0.03 (0.01)0.050.06 (0.007)<0.0010.06
 Percent change−22.61 (10.24) −40.03 (3.8) 0.14
PME/total Pi     
 Baseline0.11 ( 0.01) 0.12 (0.01)  
 Nadir0.11 (0.01) 0.09 (0.04)  
 50 minutes0.18 (0.01) 0.16 (0.02)  
 Absolute change−0.06 (0.02)0.01−0.04 (0.01)0.030.37
 Percent change80.86 (29.32) 40.70 (19.39) 0.30
PDE/total Pi     
 Baseline0.18 (0.01) 0.15 (0.01)  
 Nadir0.12 (0.03) 0.11 (0.04)  
 50 minutes0.14 (0.02) 0.13 (0.02)  
 Absolute change0.04 (0.02)0.080.02 (0.02)0.320.57
 Percent change−20.13 (8.81) −11.75 (11.01) 0.56

The relationships among changes in 31P MRS metabolites at baseline and 50 minutes after IV fructose challenge and the presence or absence of hyperuricemia are depicted in Table 3. Patients with hyperuricemia had a trend toward a decline in α-ATP, compared to patients without baseline hyperuricemia (P = 0.06). The median decline in β-ATP/Pi ratio was lower at nearly all time points (5-50 minutes) after IV fructose challenge in patients with, compared to those without, baseline hyperuricemia (Fig. 3). A statistically significant greater nadir was noted in patients with NAFLD versus without NAFLD (7.15 versus 4.58; P = 0.03). No association between hepatic ATP levels and alanine aminotransferase (ALT), BMI, or alcohol consumption was observed.

Table 3. 31P MRS Metabolites at Baseline and 50 Minutes Post-IV Fructose, by UA Levels
MetabolitesIncreased UA (≥5.5 mg/dL) (N = 8)P Value Δ Baseline- 50 Minutes Increased UANormal UA (<5.5 mg/dL) (N = 17)P Value Δ Baseline-50 Minutes Normal UAP Value Mean Δ Baseline-50 Minutes High Versus Normal
α-ATP/total Pi     
 Baseline0.3 (0.02) 0.3 (0.01)  
 Nadir0.15 (0.03) 0.17 (0.03)  
 50 minutes0.2 (0.03) 0.2 (0.01)  
 Absolute change0.1 (0.02)0.0040.04 (0.01)0.0030.06
 Percent change−28.9 (6.2) −14.3 (4.2) 0.06
β-ATP/total Pi     
 Baseline13.9 (1.5) 13.3 (0.5)  
 Nadir4.5 (2.3) 7.1 (2.8)  
 50 minutes10.92 (1.4) 10.9 (0.7)  
 Absolute change−2.9 (1.4)0.07−2.4 (0.9)0.020.76
 Percent change−18.5 (9.4) −16.2 (7.2) 0.85
γ-ATP/total Pi     
 Baseline0.14 (0.01) 0.14 (0.01)  
 Nadir0.04 (0.02) 0.06 (0.02)  
 50 minutes0.08 (0.01) 0.09 (0.01)  
 Absolute change0.06 (0.01)<0.0010.05 (0.01)<0.0010.47
 Percent change−43.4 (5.8) −32.9 (6.4) 0.31
PME/total Pi     
 Baseline0.15 (0.01) 0.11 (0.01)  
 Nadir0.09 (0.03) 0.11 (0.02)  
 50 minutes0.15 (0.01) 0.19 (0.01)  
 Absolute change−0.003 (0.02)0.86−0.07 (0.01)<0.0010.005
 Percent change9.91 (15.8) 81.8 (20.3) 0.03
PDE/total Pi     
 Baseline0.15 (0.01) 0.18 (0.01)  
 Nadir0.13 (0.03) 0.11 (0.04)  
 50 minutes0.15 (0.01) 0.14 (0.01)  
 Absolute change0.0003 (0.01)0.970.04 (0.01)0.00940.09
 Percent change3.7 (9.1) −19.5 ( 6.1) 0.04
Figure 3.

Median (interquartile range) changes after IV fructose, by UA status.


Our study shows that fructose intake triggers transient declines in hepatic Pi, α-ATP, β-ATP, and γ-ATP, findings consistent with hepatic ATP utilization during the initial phases of fructose metabolism. Interestingly, individuals with obesity and T2DM who habitually consumed increased dietary fructose were more susceptible to hepatic α-ATP and γ-ATP depletion after an acute IV fructose bolus than similar patients who consumed less dietary fructose. Thus, fructose provides a metabolic perturbation to the liver that can be utilized to characterize interindividual differences in hepatic energy homeostasis as well as variability in disease severity and progression among patients with NAFLD.

We speculate that the association between habitual consumption of high-fructose diets and susceptibility to hepatic ATP depletion after an acute fructose challenge may reflect, at least in part, a compensatory up-regulation of fructose-metabolizing enzymes in high-fructose consumers. Hydrolysis of ATP during fructose metabolism generates ADP and AMP. The latter is either rephosphorylated by AMPK to regenerate ATP or further degraded to adenosine and, ultimately, UA. UA tends to accumulate when the rate of ATP hydrolysis outstrips its regeneration. Thus, it is particularly interesting that hyperuricemic subjects had lower baseline α-ATP/Pi as well as a greater absolute and percent change from baseline in α-ATP, β-ATP, and γ-ATP after IV fructose challenge than those with normal serum UA levels. Together, these data suggest that habitual consumption of high-fructose–containing diets provides a metabolic challenge that may impair hepatic energy homeostasis in patients with underlying insulin resistance (IR). Furthermore, increased serum UA may serve as a surrogate serologic marker identifying individuals who are unable to replenish liver ATP stores effectively.

The decrease in absolute levels of hepatic ATP in viral and alcoholic hepatitis38, 39 and obesity18, 19 has been interpreted as “energy deficit” or impaired “ATP homeostasis.” Humans with IR and hepatic steatosis also have decreased hepatocellular ATP.40 Even in metabolically well-controlled T2DM, hepatic energy metabolism could be impaired when compared to age- and BMI-matched and young lean controls.40 Individuals with T2DM had 26% and 23% lower γ-ATP (1.68 ± 0.11, 2.26 ± 0.20, and 2.20 ± 0.09 mmol/L; P < 0.05) than age- and BMI-matched controls and young healthy individuals, respectively. Furthermore, they had 28% and 31% lower Pi than did individuals from the matched control and young healthy control groups (0.96 ± 0.06, 1.33 ± 0.13, and 1.41 ± 0.07 mmol/L; P < 0.05). Even after adjustment for hepatic lipid volume fraction, hepatic ATP and Pi related negatively to hepatic insulin sensitivity (r = −0.665, P = 0.010; r = −0.680, P = 0.007), but not to whole body insulin sensitivity. These data suggest that impaired hepatic energy metabolism and IR could precede the development of steatosis in individuals with T2DM.40

Likewise, it is conceivable that mitochondrial ATP synthesis might also be decreased in prediabetic patients with NAFLD. In support of this contention, patients with nonalcoholic steatohepatitis exhibit alternations and/or abnormalities of their mitochrondria.41 Impaired energy homeostasis could also result from hepatocellular γ-ATP depletion resulting from increased ATP utilization by energy-demanding processes, such as Na+/K+ adenosine triphosphatases, lipogenesis, or gluconeogenesis. Although loss of functional hepatocytes resulting from necrosis and replacement with fat and collagen may serve as yet another explanation for hepatic γ-ATP depletion, our study group of subjects with obesity and T2DM lacked overt clinical or laboratory evidence of liver damage. In such subjects, a dietary history of increased fructose consumption correlated with reduced hepatic content of ATP, suggesting that the metabolism of fructose may provide a previously unsuspected threat to hepatic energy homeostasis.

An alternative explanation may reflect the impaired ATP generation in response to fructose ingestion that is unique to fructose metabolism. As discussed above, fructose is known to induce transient ATP depletion as a result of its rapid phosphorylation.42 The scavenger enzyme, AMP deaminase 2, reclaims additional phosphates from ADP and, in the process, generates the waste product, UA. Of note, AMPK is the master regulator of cellular energy flux in the liver. Under normal physiologic conditions, increased cellular content of AMP activates AMPK and results in the prompt regeneration of ATP. However, under conditions where AMP kinase activity is low (as may occur in the setting of IR), AMP is deaminated and increased production of UA (as opposed to ATP) is favored. Fructose is also known to up-regulate both its main transporter (i.e., Glut5) and its major enzyme (i.e., fructokinase).30 Both fructokinase protein and activity in murine hepatocytes increase after incubation with fructose.6 Furthermore, laboratory rats fed diets high in fructose show an increase in Glut5 in the intestinal epithelium and an increase in fructokinase in their liver, compared to controls.30 Likewise, subjects with NAFLD and higher intake of fructose have higher levels of fructokinase mRNA in their liver biopsies, compared to control subjects with liver disease.6 Humans given a high-fructose diet show a more marked increase in UA in response to fructose.26 These studies suggest that the effects of fructose to up-regulate its enzymes could lead to a greater ATP depletion and hyperuricemia in response to fructose. In turn, a more severe ATP depletion could be a mechanism for potentiating cell injury in subjects with NAFLD.

Elevated UA predicted both the baseline and nadir of ATP depletion. This finding, which may be explained by a tight link between the generation of UA and ATP depletion in response to fructose administration, suggests that UA may be a biomarker of fructokinase activity levels. The higher UA increase induced by fructose in patients with cirrhosis therefore appears to be a good marker of the diseased liver's inability to efficiently resynthesize ATP from its breakdown products.43 Alternatively, the habitual consumption of fructose may not allow for the efficient resynthesis of ATP. Regardless, these data could provide an explanation for why an elevated UA level may be a predictor for NAFLD. However, we did not find UA as a predictor for more advanced liver disease in our recent study, whereas the amount of fructose intake did correlate with hepatic fibrosis.14 Clearly, further studies are needed to better understand the role of UA in NAFLD and the progression of liver disease.

Our study has some limitations. First, our study was an observational, cross-sectional study without a true control population and no randomized intervention designed to affect the endpoints. Second, histology was unavailable for analysis, becqause liver biopsies are not considered ethical in subjects without any evidence of liver disease. Third, this study was not powered to assess clinically significant differences between groups. Despite this limitation, interesting insights regarding the potential mechanism(s) that may underlie fructose-related liver injury were gained. Fourth, our study population consisted of patients with known T2DM who had already received nutritional counseling. Thus, total fructose and caloric intake were lower in our study population than might have been observed in a general population. Despite this, the striking finding was that we were still able to show a difference in hepatic ATP and baseline and after IV fructose challenge in subjects who consumed more fructose. Further, despite this low threshold for defining fructose consumption, differences in UA levels also correlated with the severity of ATP depletion observed in response to fructose.

In conclusion, patients with obesity and T2DM with increased habitual dietary fructose consumption show reduced hepatic ATP concentrations, compared to those with minimal dietary fructose intake. These data support our hypothesis that increased dietary fructose consumption may impair hepatocellular energy homeostasis and thus could be a risk factor for progressive liver injury. Furthermore, hyperuricemia may serve as a surrogate marker of hepatic ATP depletion after exposure to fructose in patients with IR and, potentially, NAFLD. The presence of hyperuricemia in patients with IR may help clinicians to identify patients “at risk” for cellular injury from fructose and hence may highlight subjects at risk for progression of NAFLD. Impaired hepatic energy homeostasis attributable to increased dietary fructose consumption underscores the urgent, dire need for increased public awareness of the risks associated with high-fructose consumption.


The Fatty Liver Subgroup of the Look AHEAD Research Group includes Jeanne M. Clark, M.D., M.P.H. (PI), Charalett Diggs, R.N. (PC), Anna Mae Diehl, M.D. (former PI; now at Duke University, Durham, NC), Frederick L. Brancati, M.D., M.H.S., Stephen Crawford, Ph.D., Susanne Bonekamp, Ph.D., D.V.M., Alena Horska, Ph.D., Mariana Lazo, M.D., Ph.D., Sc.M., and Steven Solga, M.D., from Johns Hopkins University (Baltimore, MD).