am j clin nutr 2014 lecoultre 268 75

Coffee consumption attenuates short-term fructose-induced liverinsulin resistance in healthy men1–4

Virgile Lecoultre, Guillaume Carrel, Leonie Egli, Christophe Binnert, Andreas Boss, Erin L MacMillan, Roland Kreis,Chris Boesch, Christian Darimont, and Luc Tappy

ABSTRACTBackground: Epidemiologic and experimental data have suggestedthat chlorogenic acid, which is a polyphenol contained in greencoffee beans, prevents diet-induced hepatic steatosis and insulinresistance.Objective: We assessed whether the consumption of chlorogenicacid–rich coffee attenuates the effects of short-term fructose over-feeding, dietary conditions known to increase intrahepatocellularlipids (IHCLs), and blood triglyceride concentrations and to decreasehepatic insulin sensitivity in healthy humans.Design: Effects of 3 different coffees were assessed in 10 healthyvolunteers in a randomized, controlled, crossover trial. IHCLs, hepaticglucose production (HGP) (by 6,6-d2 glucose dilution), and fastinglipid oxidation were measured after 14 d of consumption of caffein-ated coffee high in chlorogenic acid (C-HCA), decaffeinated coffeehigh in chlorogenic acid, or decaffeinated coffee with regular amountsof chlorogenic acid (D-RCA); during the last 6 d of the study, theweight-maintenance diet of subjects was supplemented with 4 g fruc-tose $ kg21 $ d21 (total energy intake 6 SD: 143 6 1% of weight-maintenance requirements). All participants were also studied withoutcoffee supplementation, either with 4 g fructose $ kg21 $ d21 (highfructose only) or without high fructose (control).Results: Compared with the control diet, the high-fructose dietsignificantly increased IHCLs by 102 6 36% and HGP by 16 63% and decreased fasting lipid oxidation by 100 6 29% (all P ,0.05). All 3 coffees significantly decreased HGP. Fasting lipid ox-idation increased with C-HCA and D-RCA (P , 0.05). None of the3 coffees significantly altered IHCLs.Conclusions: Coffee consumption attenuates hepatic insulin resistancebut not the increase of IHCLs induced by fructose overfeeding. Thiseffect does not appear to be mediated by differences in the caffeineor chlorogenic acid content. This trial was registered at clinicaltrials.gov as NCT00827450. Am J Clin Nutr 2014;99:268–75.

INTRODUCTION

Obesity is associated with ectopic lipid deposition in hepaticand skeletal muscle cells. These ectopic lipid stores may result, atleast in part, from excessive intakes of dietary fats and sugars (1)and are thought to play a role in the pathogenesis of insulinresistance by generating intracellular lipid metabolites that in-terfere with insulin’s actions (2, 3). Epidemiologic surveys haveshown that high coffee consumption is associated with lowerincidences of obesity (4) and type 2 diabetes (5–10). Thesebeneficial effects of coffee are also observed with decaffeinated

coffee (7, 9, 11, 12), indicating that components in coffee, otherthan caffeine, play a role in these effects. Of the known activecompounds in coffee, polyphenols have been shown to protectagainst oxidative stress (13) and reduce risks for cardiovasculardisorders and diabetes (14–16). Coffee consumption is also as-sociated with lower incidences of nonalcoholic fatty liver dis-ease and the progression of nonalcoholic fatty liver disease tononalcoholic steatohepatitis or hepatocellular carcinoma (17–19).Furthermore, it has been observed that coffee polyphenols coulddownregulate lipogenic pathways and reduce liver fat accumu-lation in high-fat fed mice (13, 20, 21). These findings havesuggested that coffee polyphenols may specifically prevent tissuediet-induced ectopic lipid deposition and insulin resistance.

Chlorogenic acid, which is a polyphenol present in largeramounts in green coffee beans than in roasted coffee (22), hasbeen reported to improve glucose homeostasis (23–25), preventectopic fat accumulation in high fat-fed rodents (20, 26), andmay be responsible for some of the metabolic effects of coffee.In the current study. we used a short term hypercaloric fructoseoverfeeding protocol, which has been shown to increase intra-hepatocellular lipids (IHCLs)5 and reduce insulin sensitivity inhealthy subjects (27, 28), to assess whether the consumption of

1 From the Department of Physiology, University of Lausanne, Lausanne,

Switzerland (VL, GC, LE, C Binnert, and LT); the Service of Internal Med-

icine (GC) and the Service of Endocrinology, Diabetes and Metabolism (LT);

Lausanne University Hospital, Lausanne, Switzerland; the Department of

Clinical Research and Institute of Diagnostic, Interventional, and Pediatric

Radiology, University Bern, Bern, Switzerland (AB, ELM, RK, and C

Boesch); and Nutrition & Health Research, Nestle Research Center, Nestec

SA, Lausanne, Switzerland (C Binnert and CD).2 VL,GC, and LE contributed equally to this work.3 Supported by research funding from Nestec SA (to LT). The Swiss National

Science Foundation supported AB and ELM as well as magnetic resonance

investigations and evaluations (SNF 31003A_132935 and SNF 320030_

135743).4 Address reprint requests and correspondence to L Tappy, Department of

Physiology, University of Lausanne, 7 Bugnon Street, 1005 Lausanne, Switzerland.

E-mail: [email protected] Abbreviationsused: C, control diet; C-HCA, caffeinated coffee high in chloro-

genic acid; CRC, clinical research center; D-HCA, decaffeinated coffee high in

chlorogenic acid; D-RCA, decaffeinated coffee with regular amounts of chloro-

genic acid; HFr, high-fructose diet; HGP, hepatic glucose production; IHCL, intra-

hepatocellular lipid; IMCL, intramyocellular lipid; MR, magnetic resonance;

MRS, magnetic resonance spectroscopy; NEFA, nonesterified fatty acid.

Received July 2, 2013. Accepted for publication October 31, 2013.

First published online November 20, 2013; doi: 10.3945/ajcn.113.069526.

268 Am J Clin Nutr 2014;99:268–75. Printed in USA. � 2014 American Society for Nutrition

by guest on April 1, 2015

ajcn.nutrition.orgD

ownloaded from

http://ajcn.nutrition.org/

chlorogenic-rich coffees, independent of their caffeine content,would prevent IHCL accumulation and hepatic insulin re-sistance. We further postulated that chlorogenic-rich coffeesmay exert such beneficial effects by enhancing whole-bodyenergy expenditure and lipid oxidation.

SUBJECTS AND METHODS

Participants

Thirteen healthy men participated in the study; 10 subjectscompleted the study protocol. Subjects dropped out for personalreasons or minor adverse events related to fructose intake.Baseline variables of the 10 participants who completed the studyare shown in Table 1. Participants were light to moderate coffeeconsumers (#4 cups coffee/d) and had low to moderate physicalactivity levels (w30 min walking/d and #3 structured physicalactivities/wk). All subjects were in good physical health, werenonsmokers, and were not taking any drugs or medication at thetime of the study. The experimental protocol (version dated 20June 2008) was approved by the Ethical Committee of LausanneUniversity School of Medicine (protocol 144/08; accepted 4 Au-gust 2008). All participants provided informed, written consent.This trial was registered at clinicaltrials.gov as NCT00827450.

Study design

Volunteers were recruited by means of advertisements placedin the campus and university hospital. Volunteers were studied on

5 different occasions, each time with 1 of 5 different dietaryconditions over a 28-d period (days1–28), with a washout period$4 wk between each condition. The study design and conditionsare summarized in Figure 1. At the beginning of each sequence,participants received oral and written instructions to consumea balanced (55% carbohydrate, 30% fat, and 15% protein)weight-maintenance diet and avoid consuming foods rich inchlorogenic acid (eg, apples and artichokes) and caffeine. Par-ticipants consumed the prescribed diet on the basis of receivedinstructions from days 1 to 25. From days 26 to 28, subjectsreceived a controlled, balanced weight-maintenance diet (cal-culated to provide a total energy intake equal to basal energy

TABLE 1

Participant characteristics at inclusion1

Values

Age (y) 23 6 2

Weight (kg) 73.2 6 2.2

Height (cm) 180 6 2

BMI (kg/m2) 22.6 6 0.6

Fat mass (%) 17.6 6 1.2

Fat mass (kg) 12.9 6 0.9

Fat-free mass (kg) 60.3 6 1.9

Systolic blood pressure (mm Hg) 140 6 3

Diastolic blood pressure (mm Hg) 78 6 3

Resting heart rate (beats/min) 75.4 6 3.5

1All values are means 6 SEMs. n = 10 subjects.

FIGURE 1. Experimental design. In each of the 5 experimental conditions, volunteers received instructions (dietary instructions) to consume a balanceddiet and to avoid foods rich in chlorogenic acid for the first 25 d of the study period (28 d). After 14 d, body weight was assessed (thin vertical arrows).Thereafter, in the C condition, participants followed the same dietary instructions for 11 d, received an isocaloric, balanced diet for 3 d (controlled diet), afterwhich intrahepatocellular lipids and intramyocellular lipid concentrations were measured by using proton MRS (bold vertical arrows), body weight wasmeasured (thin vertical arrows), and FM (energy expenditure, substrate oxidation, hepatic glucose production, and lipolysis) was assessed (filled boxes). In theHFr condition, participants ingested 4 g fructose $ kg21 $ d21 added to their weight-maintenance diet (fructose overfeeding) for 6 d before the assessment ofintrahepatocellular lipid concentrations, body weight, and fasting glucose and lipid metabolism. Participants received coffees with varying amounts ofchlorogenic acid and caffeine for 14 d and 4 g fructose $ kg21 $ d21 for 6 d before metabolic assessments in the C-HCA, D-HCA, and D-RCA conditions.C, control diet; C-HCA, caffeinated coffee high in chlorogenic acid; D-HCA, decaffeinated coffee high in chlorogenic acid; D-RCA, decaffeinated coffee withregular amounts of chlorogenic acid; FM, fasting metabolism; HFr, high-fructose diet; MRS, magnetic resonance spectroscopy.

COFFEE AND FASTING GLUCOSE TURNOVER 269


ajcn.nutrition.orgD

ownloaded from


requirements predicted from the Harris Benedict equation (29)multiplied by 1.4 to account for a low-to-moderate level ofphysical activity). During this period, subjects received theirwhole diets as prepacked food items, with specific instructionsregarding the time at which each item was to be consumed andinstructions for preparation when needed. Participants werecarefully instructed not to consume any other foods or beveragesin excess of the distributed items, with the exception of water.On day 14, participants reported to the clinical research center(CRC) at Lausanne University Hospital to have their bodyweights measured. On day 28, subjects went to UniversityBern, where their IHCL and intramyocellular lipid (IMCL)concentrations were measured by using proton magnetic res-onance spectroscopy (MRS) as previously described (30). Onday 29, participants reported to the CRC where their bodyweights, blood metabolic profiles, hepatic glucose production(HGP), and net substrate oxidation rates were measured afteran overnight fast.

Each subject was studied on 5 different occasions as follows:

1) High-fructose diet (HFr): participants consumed, in addi-tion to their weight-maintenance diet, 4 g fructose $ kg bodyweight21 $ d21 as lemon-flavored drinks taken during mealsfrom days 23 to 28.

2) Subjects consumed 4 cups coffee/d prepared each with 3.4 gsoluble, partially roasted caffeinated coffee high in chloro-genic acid (C-HCA) (containing 3.0% caffeine and 9%chlorogenic acid; NESCAFE GreenBlend; Nestle) from days15 to 28 as well as 4 g fructose $ kg body weight21 $ d21

from days 23 to 28.

3) Subjects consumed the same diet with 4 cups decaffeinatedcoffee prepared with partially roasted coffee with a highchlorogenic acid content [decaffeinated coffee high in chloro-genic acid (D-HCA)] (containing 0.3% caffeine and 9%chlorogenic acid; Decaffeinated NESCAFE GreenBlend;Nestle).

4) Subjects consumed the same diet with 4 cups of coffeeprepared with roasted decaffeinated coffee with regularamounts of chlorogenic acid (D-RCA) (containing 0.3%caffeine and 3% chlorogenic acid; Decaffeinated NESCAFEGold; Nestle).

5) Subjects consumed, during 14 d, an energy-balanced dietwithout fructose supplementation (control condition).

In all conditions, except the control condition, fructose wasadded to the weight-maintenance diet during the last 6 d, whichresulted in a total energy intake that corresponded to 143 6 1%of energy requirements.

Participants were randomly allocated 1 of 5 study sequenceswith a balanced order of treatment [1: control diet (C), HFr,C-HCA, D-HCA, and D-RCA; 2: HFr, C-HCA, D-HCA, D-RCA,andC; 3: C-HCA,D-HCA,D-RCA, C, andHFr; 4: D-HCA,D-RCA,C, HFr, and C-HCA; and 5: D-RCA, C, HFr, C-HCA, and D-HCA].Sequences were generated at the sponsor site, and a list of thesequences was kept in a sealed envelope. The treatment order wasbalanced. Participants and investigators were blinded to the typeof coffee consumed, but not to the C or HFr because there was nosuitable placebo to coffee or fructose. Compliance was checked

by asking volunteers to return empty coffee bags to the laboratoryat each visit.

Assessment of ectopic lipid deposition

Measurements of IHCL and IMCL concentrations were ob-tained by proton MRS. Measurements were performed on aclinical 3T magnetic resonance (MR) system (TIM Trio; SiemensMedical) as previously described (30). In short, to quantifyIHCLs, a volume of interest of 19 mL was localized in the liverand evading large vessels and the proximity to extrahepatic fat byusing a double-echo localization sequence combined with Sie-mens’ prospective acquisition correction scheme (MRS echotime: 30 ms; repetition time: between 2.5 and 6 s depending onthe breathing rate). MRS was preceded by fast-spin echo MRimagining in 3 planes by using the same prospective acquisition-correction triggering to visualize the liver and to reliably re-produce the placement of the region of interest in follow-upexaminations. Automatic fitting of the MR spectra was per-formed with the home-written software Fitting Tool for Arraysof Interrelated Datasets (31) by using previous knowledge re-straints. The lipid spectrum was modeled with 9 Voigt lines. MRspectra were recorded with water presaturation to determinelipid and metabolite spectra (32 scans, with the center frequencyat 3 ppm) and without water suppression to acquire the watersignal as an internal standard (16 scans with a center frequencyat 4.7 ppm). IMCLs were measured in a volume of 2.4 cm3 inthe tibialis anterior of the right leg according to Boesch et al (32)but at 3 T and by using an echo time of 30 ms and repetition timeof 3s.

Measurement of fasting metabolism

On the morning of day 29, participants reported at 0700 to theCRC, Lausanne University Hospital, in the fasting state. Onarrival, participants were requested to void and were weighed.Thereafter, they were transferred to a bed where they remained ina semirecumbent position for the next 2.5 h. Two venous catheterswere inserted, one catheter into a vein of the left forearm, throughwhich infusions of 6,6-2H2 glucose (prime dose: 2 mg/kg over 10min followed by a continuous infusion of 20 mg $ kg21 $ min21)and 1,1,2,3,3-2H5 glycerol (prime dose: 90 mg/kg over 5 minfollowed by a continuous infusion of 9 mg $ kg21 $ min21) wereinfused for 2.5 h. The second catheter was inserted into a wristvein on the right side to allow for the periodic withdrawal ofblood samples. This hand was placed in a heated hand pad toensure partial arterialization of venous blood. Blood sampleswere obtained before tracer infusion (baseline) and at 120, 135,and 150 min after the start of tracer infusions for the measure-ment of tracer, hormones, and substrate concentrations. Re-spiratory gas exchange was measured throughout the 150-minperiod with the use of an open circuit indirect calorimeter,Deltatrac II (Datex Instruments).

Body composition

At baseline, body composition was assessed by using dualx-ray absorptiometry (iDXA; GE Healthcare), and body weightwas measured. On days 15 and 29, body weight was recorded tothe nearest 0.1 kg (Seca 708; Seca).

270 LECOULTRE ET AL


ajcn.nutrition.orgD

ownloaded from


Blood analysis

Plasma was immediately separated from blood by centrifu-gation for 10 min at 48C and 3600 3 g. Samples were storedat 2208C until analysis. Plasma glucose (Glucose Analyzer II;Beckman Instruments), triglycerides (Randox Laboratories),nonesterified fatty acids (NEFAs; Wako Chemicals), and urinarynitrogen (Randox Laboratories) were measured by using enzy-matic methods; insulin was assessed by using a radioimmuno-assay (Millipore).

After plasma deproteinization and partial purification over anion-and cation-exchange resins, plasma 2H2-glucose and 2H5-glycerolwere acetylated in the presence of acetic anhydride and pyridine,and glycerol concentrations and 2H2-glucose and 2H5-glycerolenrichments were measured by using gas chromatography–massspectrometry (Agilent Technologies).

Calculations

Basal metabolic and net substrate oxidation rates were cal-culated from respiratory gas exchange and the urinary nitrogenexcretion rate by using the equations of Tappy and Schneiter (33).HGP and glycerol rates of appearancewere calculated fromplasma2H2-glucose and 2H5-glycerol enrichment by using Steele’s equa-tions for steady state conditions.

Statistical analysis

A statistical analysis was performed on data obtained from the10 participants who completed all 5 experimental conditions.Two distinct hypotheses were tested. The first hypothesis was thatthe HFr would significantly alter the variables measured com-pared with the effect of the C (HFr compared with C; hypothesis1); the second hypothesis was that all 3 tested coffees modulatedthe effects of the HFr (C-HCA, D-HCA, and D-RCA comparedwith HFr; hypothesis 2). Both hypotheses were tested by usingmixed-model analysis and post hoc comparisons. Mixed modelswere fitted with the treatment as a fixed effect and the subject asa random effect. BMI at inclusion was added as a fixed covariatein initial models and removed if it did not significantly contributeto the goodness of fit. Compound symmetry was used as a cor-relation structure. A single mixed model was used to test hy-potheses 1 and 2 by treating the control condition as the referencegroup for testing the HFr and by using one-sided, Hommel-adjusted contrasts to compare each coffee with the HFr. Thisstatistical analysis was chosen on the basis of the hypothesis thatcoffee would lower IHCL concentrations, fasting HGP, andplasma triglyceride concentrations through mechanisms relatedto an increased energy expenditure and enhanced lipid oxidation.This hypothesis has been well supported by several reports thatshowed that coffee or the potentially active compounds of coffeeand tea increase lipolysis and lipid oxidation in humans (34, 35)or increase hepatic insulin sensitivity (21, 36). Therefore, weassumed that any alteration of these variables in a direction otherthan expected would reflect the effects of unaccounted con-founding factors, such as noncompliance to the prescribed diet orchanges in physical activity. Before the statistical analysis, datadistributions and equality of variance were checked by usingShapiro’s and Bartlett’s tests. IHCL and plasma triglycerideconcentrations were nonnormally distributed and were log trans-formed before the statistical analysis.

The sample size was calculated on the basis of previous datafrom our laboratory (37) that reported that IHCL accumulationwas blunted by a mean (6SD) of 4.36 3.5 mmol/kg after 4 d high-fat, high-protein consumption compared with that with a high-fatdiet [ie, an effect size (as defined by the ratio between the meanand SD of the difference) of 1.228]. The hypothesis was that,in $1 of the 3 experimental conditions (a = 0.016), the con-sumption of coffee for 14 d would decrease intrahepatic lipidaccumulation (primary outcome) to the same extent as woulda high-protein diet (37). With the assumption of a power of 80%,dropout rate of 20%, and Pitman asymptotic relative efficiencyof 0.864, the sample size was calculated to be 13 participants(PASS 6.0; NCSS). The statistical analysis was performed withR software (version 2.15.3) (38).

RESULTS

Body weight

Overall, body weight changes 6 SDs between days 14 and29 were 20.96 6 0.47, 0.28 6 0.23, 20.67 6 0.32, 0.34 6 0.4,and20.16 0.38 kg in control, HFr, C-HCA, D-HCA, and D-RCAconditions, respectively (P = NS).

Intrahepatic, intramuscular, and blood lipid concentrations

Compared with the C, the HFr increased IHCL concentrations(P , 0.001; Figure 2A) and total plasma triglycerides (P =0.018) significantly, whereas it decreased plasma NEFA (P ,0.001) and glycerol (P = 0.033) concentrations (Table 2). En-dogenous glycerol production tended to decrease (P = 0.051;Table 2). None of the 3 tested coffees significantly altered thesevariables (Table 2). Intramyocellular lipid concentrations wereunchanged after the HFr and were not altered by the differentcoffees (Table 2).

Basal metabolic rate and net substrate oxidation

The basal metabolic rate was 0.966 0.04, 0.996 0.03, 1.0360.05, 0.94 6 0.03, and 1.04 6 0.03 kcal/min with the C, HFr,C-HCA, D-HCA, and D-RCA respectively (P = NS). Comparedwith the C, the HFr decreased the net lipid oxidation (Figure 2B)and increased carbohydrate oxidation (Figure 2C) significantly(P = 0.011 and P = 0.0012, respectively). C-HCA and D-RCAincreased lipid oxidation compared with the HFr significantly(P , 0.001 and P = 0.025, respectively; Figure 2B). Comparedwith the HFr, both C-HCA and D-RCA significantly decreasedthe net carbohydrate oxidation (P = 0.0014 and P = 0.006, re-spectively), whereas this decrease showed a trend with D-HCA(P = 0.074) (Figure 2C).

HGP

The HFr significantly increased HGP (P = 0.0016, Figure 2D),but basal fasting plasma glucose and insulin concentrations didnot change significantly (Table 2). Compared with the HFr,D-HCA and D-RCA significantly decreased HGP (P = 0.031and P = 0.016, respectively, Figure 2D), whereas there was onlya trend with C-HCA (P = 0.057; Figure 2D). Fasting insulin andglucose concentrations remained unchanged compared withthose with the HFr (Table 2).



ajcn.nutrition.orgD

ownloaded from


DISCUSSION

A number of epidemiologic studies indicated that coffee con-sumption is associated with lower risk of diabetes and metabolicdisorders independently of caffeine intake (7, 9, 11, 12). Fur-thermore, human and animal studies have also suggested thatsome, as yet unidentified, coffee polyphenols protect againstoxidative stress, hepatic steatosis, and insulin resistance inducedby hypercaloric diets (13, 25, 39, 40). The purpose of this studywas to evaluate whether chlorogenic acid, which is a coffeepolyphenol present in a high amount in green coffee bean (41),may prevent the development of hepatic steatosis and hepaticinsulin resistance induced by short-term fructose overfeeding inhumans. To test this hypothesis, we observed the effects of 3

different coffees, 2 coffees of which had a high chlorogenic acidcontent with or without caffeine, and one coffee that had witha regular chlorogenic acid content, in healthy, young men overfedfor 6 d with fructose. Such fructose overfeeding is known to causea substantial increase in intrahepatic lipid concentrations withina few days, together with an increase in fasting plasma tri-glycerides and HGP (27, 28). However, the relative roles ofexcess energy intake and dietary fructose remain unknown. It wasrecently observed that the consumption of hypercaloric high-glucose or high-fructose diets increased IHCLs and impairedhepatic insulin sensitivity (42, 43), whereas isocaloric high-glucose or high-fructose diets did not change IHCLs (43), andan hypercaloric high-fat diet did not impair hepatic insulin

TABLE 2

Effect of short-term fructose overfeeding and coffee intake on fasting concentrations of metabolites related to glucose and

lipid metabolism and on skeletal muscle lipid accumulation1

Control HFr C-HCA D-HCA D-RCA

Triglycerides (mmol/L) 0.94 6 0.15 1.45 6 0.2y 1.31 6 0.29 1.6 6 0.23 1.62 6 0.42

NEFA (mmol/L) 0.53 6 0.06 0.31 6 0.06z 0.37 6 0.04 0.33 6 0.05 0.37 6 0.05

Glycerol (mmol/L) 51 6 7.2 37.1 6 4.9y 41.2 6 2.9 35.6 6 2.8 42.5 6 5.3

Glycerol Ra (mmol $ kg21 $ min21) 2.02 6 0.3 1.76 6 0.28 1.87 6 0.35 1.77 6 0.22 1.84 6 0.25

IMCLs (mmol/kg ww) 2.84 6 0.46 2.86 6 0.3 2.66 6 0.51 2.9 6 0.52 2.7 6 0.38

Glucose (mmol/L) 5.2 6 0.1 5.2 6 0.1 5.3 6 0.2 5.4 6 0.3 5.4 6 0.3

Insulin (mU/mL) 10.7 6 0.9 11.6 6 1 12.5 6 1.4 12.3 6 1.7 12.9 6 1.5

1Values are means6 SEMs. n = 10 participants. y,zCompared with the control condition, yP, 0.05 and zP, 0.01. P

values were obtained by using a single mixed model to test hypotheses 1 and 2 by treating the control condition as the

reference group for testing the HFr and by using 1-sided, Hommel-adjusted contrasts to compare each coffee with the HFr

(C-HCA, D-HCA, and D-RCA, respectively, compared with the HFr). C-HCA, caffeinated coffee high in chlorogenic acid;

D-HCA, decaffeinated coffee high in chlorogenic acid; D-RCA, decaffeinated coffee with regular amounts of chlorogenic

acid; HFr, high-fructose diet; IMCL, intramyocellular lipid; NEFA, nonesterified fatty acid; Ra, rate of appearance; ww, wet

weight.

FIGURE 2. Mean (6SEM) effects of short-term fructose overfeeding (HFr) and coffees with varying amounts of chlorogenic acid and caffeine on IHCL(A), fasting lipid oxidation rates (B), fasting glucose oxidation rates (C), and fasting glucose Ra (D). n = 10 participants. **,***Compared with theC condition, **P , 0.01 and ***P , 0.001; x,#,##,###Compared with the HFr for C-HCA, D-HCA, and D-RCA conditions, xP , 0.1, #P , 0.05, ##P ,0.01, and ###P , 0.001. P values were obtained by using a single mixed model to test hypotheses 1 and 2 by treating the C condition as the reference group fortesting the HFr and by using 1-sided, Hommel-adjusted contrasts to compare each coffee with the HFr. C, control diet; C-HCA, caffeinated coffee high inchlorogenic acid; D-HCA, decaffeinated coffee high in chlorogenic acid; D-RCA, decaffeinated coffee with regular amounts of chlorogenic acid; HFr,high-fructose diet; IHCL, intrahepatocellular lipids; Ra, rate of appearance.

272 LECOULTRE ET AL


ajcn.nutrition.orgD

ownloaded from


sensitivity (42). Therefore, it is likely that these effects are morebroadly related to carbohydrate overfeeding rather that to specificeffects of fructose.

Contrary to our expectation, none of the 3 coffees significantlyprevented the increase in IHCL concentration or development ofhypertriglyceridemia. The coffees also failed to increase en-dogenous glycerol production or glycerol and NEFA concen-trations, which indicated that the coffees did not overcome thefructose-induced inhibition of lipolysis. This result was incontrast with the observation that coffee consumption signifi-cantly reduced hepatic steatosis and improved glucose tolerancein high-fructose, high-fat fed rats independently of changes inbody composition (40). The very high doses of coffee extract andsimultaneous increase in dietary fat and fructose used in thisstudymay have accounted for the difference. The failure of coffeeto prevent the increase in IHCLs induced by short-term fructoseoverfeeding was also in contrast with epidemiologic studies thatshowed an inverse correlation between coffee intake and theprevalence of hepatic steatosis in humans (17, 44). Epidemiologicstudies provide information on long-term effects of coffee andmay reflect, in part, effects of coffee on body composition. Incontrast, our 6-d overfeeding period was too short to induceclinically relevant changes in body fat mass and essentiallyreflected metabolic changes directly induced by excess fructoseand energy consumption.

Although the coffees did not prevent an increase in IHCLs, all3 coffees significantly attenuated the increase in fasting HGPinduced by short-term fructose overfeeding. In addition, lipidoxidation was significantly increased with the coffee that con-tained caffeine and normal chlorogenic acid content and with thedecaffeinated, high–chlorogenic acid coffee (Figure 2B). We canonly speculate on the mechanisms by which coffee improvedhepatic insulin sensitivity at this stage. First, 2 of the 3 testedcoffees induced the stimulation of lipid oxidation and, hence,may have decreased intrahepatic concentrations of toxic lipidmetabolites, even in the absence of a decrease in IHCLs; thiseffect would bear some similitude with exercise increasingmuscle insulin sensitivity independently of changes in in-tramuscular lipid concentrations, which is a phenomenonreferred to as “the athlete paradox” (45). Second, we did notobserve significant changes in fasting energy expenditure on theday after exposure to the different coffees but hypothesized thatthe 24-h energy expenditure and substrate utilization may haveincreased during the initial days of coffee exposure (35); thiseffect may possibly have attenuated hepatic glycogen storageinduced by fructose overfeeding. Consistent with this hypothe-sis, an association between glycogen storage and an increase infasting HGP has been reported in healthy subjects after short-term carbohydrate overfeeding (46). Third, a decrease in fastingHGP may possibly be attributable to chlorogenic acid, which hasbeen reported to inhibit glucose-6-phosphatase (47, 48). Al-though the statistical design of this exploratory study did notinclude direct comparisons of the 3 coffees tested, our results didnot suggest that chlorogenic acid was primarily involved be-cause this effect was observed with the partially roasted coffees(C-HCA and D-HCA) and fully roasted coffee (D-RCA) despitea 3:1 difference in the concentration of chlorogenic acid. Coffeeextracts (39, 40, 49), single constituents of coffee such as caf-feine (49), coffee polyphenols (13, 20, 21, 25, 50), and coffeemelanoidins (13) have all been reported to prevent the impact of

dietary interventions on liver function, although sometimesat doses considerably higher than those in the current study.Therefore, it is possible that the reduction of endogenous glu-cose production was the result of the combination of severalcoffee bioactives. This issue may be even more complex becausethere may be genetically determined interindividual variationsregarding the health effects of coffee (51). However, additionalhuman studies, including measurements of energy expenditureand substrate oxidation throughout the period of coffee expo-sure, as well as animal studies that allow intrahepatic concen-trations of substrate and lipid intermediates to be assessed, willbe required to identify the mechanisms and bioactive com-pounds actually involved in these effects of coffee (34, 52, 53).

This study has several limitations that have to be kept in mindwhen evaluating the results. First, we studied the effects ofcaffeinated and decaffeinated, partially roasted coffees and de-caffeinated, roasted coffee but not those of roasted, caffeinatedcoffee. Furthermore, we limited our statistical analysis to thecomparison of each coffee with the HFr condition. Therefore, ourstatistical analysis did not include the direct comparison of thedifferent coffees and respective roles of chlorogenic acid andcaffeine present in the 3 coffees tested. These roles were onlyindirectly evaluated by assessing which coffee had, or had not,a significant effect. With this exploratory approach, we were ableto document a robust, significant effect of the coffees on hepaticinsulin resistance and lipid oxidation. However, additional,specifically designed studies will be needed to directly evaluatethe effects of coffee compounds of interest. Second, we evaluatedeffects of coffees with an experimental, short-term, hypercaloric,high-fructose diet. This protocol was convenient because it isknown to induce highly reproducible metabolic alterations thatmimic some features of the metabolic syndrome. However, themetabolic alterations induced by an HFr differ markedly fromthose observed in subjects with the metabolic syndrome in whomhigh fatty acid fluxes secondary to increased fat mass may playa prominent role in ectopic lipid deposition (54). Finally, we wereunable to provide study participants with any satisfactory placeboin the C and HFr conditions, and therefore, participants wereblinded to the type of coffee they consumed but not to the fact thatthey consumed coffee.

In conclusion, our current results indicate that the consumptionof several coffees with varying chlorogenic acid contents did notprevent intrahepatic lipid accumulation but significantly atten-uated the hepatic insulin resistance induced by short-term fructoseoverfeeding. These effects could not be specifically attributed tochlorogenic acid or caffeine and may possibly be caused by other,as yet unidentified, active compounds.

We thank the staff of the Department of Physiology and the CRC of the

University Hospital for their outstanding assistance; Julie Moulin and

Harrison Machaira (Nestle Research Center) for support with statistical anal-

ysis; Lionel Philippe (Nestle Research Center) for clinical trial monitoring;

Karin Zwygart for help with MR examinations; and all volunteers for their

participation in the study. Parts of this study were presented at the American

Diabetes Association 72nd Scientific Sessions, Philadelphia, PA, in 2012.

The authors’ responsibilities were as follows—LT, RK, C Boesch, and

CD: designed the study; C Binnert, GC, LE, and PS: recruited participants;

C Binnert, GC, and LE: performed tests and the laboratory analysis; AB and

RK: performed MR measurements; C Boesch, RK, and ELM: performed the

MRS analysis; VL and LE: performed the statistical analysis; LT and VL:

drafted the manuscript; and all authors: revised the manuscript, read and



ajcn.nutrition.orgD

ownloaded from


approved the final manuscript, and had access to the full data set. CD is an

employee of Nestec SA but was not involved in the study implementation

and analysis. C Binnert was an employee of Nestec SA at the time that this

study was performed. VL and LT had ultimate authority over the interpre-

tation of data, writing of the report, or submission of the manuscript. VL,

GC, LE, AB, ELM, RK, C Boesch, and LT had no conflicts of interest.

REFERENCES1. Zivkovic AM, German JB, Sanyal AJ. Comparative review of diets for

the metabolic syndrome: implications for nonalcoholic fatty liver dis-ease. Am J Clin Nutr 2007;86:285–300.

2. Kelley DE. Influence of weight loss and physical activity interventionsupon muscle lipid content in relation to insulin resistance. Curr DiabRep 2004;4:165–8.

3. Szendroedi J, Roden M. Ectopic lipids and organ function. Curr OpinLipidol 2009;20:50–6.

4. Lopez-Garcia E, van Dam RM, Rajpathak S, Willett WC, Manson JE,Hu FB. Changes in caffeine intake and long-term weight change inmen and women. Am J Clin Nutr 2006;83:674–80.

5. Bhupathiraju SN, Pan A, Malik VS, Manson JE, Willett WC, van DamRM, Hu FB. Caffeinated and caffeine-free beverages and risk of type 2diabetes. Am J Clin Nutr 2013;97:155–66.

6. Pereira MA, Parker ED, Folsom AR. Coffee consumption and risk oftype 2 diabetes mellitus: an 11-year prospective study of 28 812postmenopausal women. Arch Intern Med 2006;166:1311–6.

7. Sartorelli DS, Fagherazzi G, Balkau B, Touillaud MS, Boutron-RuaultMC, de Lauzon-Guillain B, Clavel-Chapelon F. Differential effects ofcoffee on the risk of type 2 diabetes according to meal consumption ina French cohort of women: the E3N/EPIC cohort study. Am J Clin Nutr2010;91:1002–12.

8. van Dam RM, Hu FB. Coffee consumption and risk of type 2 diabetes:a systematic review. JAMA 2005;294:97–104.

9. Salazar-Martinez E, Willett WC, Ascherio A, Manson JE, LeitzmannMF, Stampfer MJ, Hu FB. Coffee consumption and risk for type 2diabetes mellitus. Ann Intern Med 2004;140:1–8.

10. van Dam RM, Feskens EJ. Coffee consumption and risk of type 2diabetes mellitus. Lancet 2002;360:1477–8.

11. Loopstra-Masters RC, Liese AD, Haffner SM, Wagenknecht LE,Hanley AJ. Associations between the intake of caffeinated and de-caffeinated coffee and measures of insulin sensitivity and beta cellfunction. Diabetologia 2011;54:320–8.

12. Huxley R, Lee CM, Barzi F, Timmermeister L, Czernichow S, PerkovicV, Grobbee DE, Batty D, Woodward M. Coffee, decaffeinated coffee,and tea consumption in relation to incident type 2 diabetes mellitus:a systematic review with meta-analysis. Arch Intern Med 2009;169:2053–63.

13. Vitaglione P, Morisco F, Mazzone G, Amoruso DC, Ribecco MT,Romano A, Fogliano V, Caporaso N, D’Argenio G. Coffee reducesliver damage in a rat model of steatohepatitis: the underlying mecha-nisms and the role of polyphenols and melanoidins. Hepatology 2010;52:1652–61.

14. Goya L, Delgado-Andrade C, Rufian-Henares JA, Bravo L, Morales FJ.Effect of coffee melanoidin on human hepatoma HepG2 cells. Pro-tection against oxidative stress induced by tert-butylhydroperoxide.Mol Nutr Food Res 2007;51:536–45.

15. Hanhineva K, Torronen R, Bondia-Pons I, Pekkinen J, Kolehmainen M,Mykkanen H, Poutanen K. Impact of dietary polyphenols on carbo-hydrate metabolism. Int J Mol Sci 2010;11:1365–402.

16. Kishimoto Y, Tani M, Kondo K. Pleiotropic preventive effects of di-etary polyphenols in cardiovascular diseases. Eur J Clin Nutr 2013;67:532–5.

17. Molloy JW, Calcagno CJ, Williams CD, Jones FJ, Torres DM, HarrisonSA. Association of coffee and caffeine consumption with fatty liverdisease, nonalcoholic steatohepatitis, and degree of hepatic fibrosis.Hepatology 2012;55:429–36.

18. Montella M, Crispo A, Giudice A. HCC, diet and metabolic factors:diet and HCC. Hepat Mon 2011;11:159–62.

19. Schutte K, Bornschein J, Malfertheiner P. Hepatocellular carcinoma–epidemiological trends and risk factors. Dig Dis 2009;27:80–92.

20. Murase T, Misawa K, Minegishi Y, Aoki M, Ominami H, Suzuki Y,Shibuya Y, Hase T. Coffee polyphenols suppress diet-induced body fataccumulation by downregulating SREBP-1c and related molecules inC57BL/6J mice. Am J Physiol Endocrinol Metab 2011;300:E122–33.

21. Murase T, Yokoi Y, Misawa K, Ominami H, Suzuki Y, Shibuya Y, HaseT. Coffee polyphenols modulate whole-body substrate oxidation andsuppress postprandial hyperglycaemia, hyperinsulinaemia and hyper-lipidaemia. Br J Nutr 2012;107:1757–65.

22. Moon JK, Yoo HS, Shibamoto T. Role of roasting conditions in thelevel of chlorogenic acid content in coffee beans: correlation withcoffee acidity. J Agric Food Chem 2009;57:5365–9.

23. Greenberg JA, Boozer CN, Geliebter A. Coffee, diabetes, and weightcontrol. Am J Clin Nutr 2006;84:682–93.

24. McCarty MF. A chlorogenic acid-induced increase in GLP-1 pro-duction may mediate the impact of heavy coffee consumption on di-abetes risk. Med Hypotheses 2005;64:848–53.

25. Ho L, Varghese M, Wang J, Zhao W, Chen F, Knable LA, Ferruzzi M,Pasinetti GM. Dietary supplementation with decaffeinated green coffeeimproves diet-induced insulin resistance and brain energy metabolismin mice. Nutr Neurosci 2012;15:37–45.

26. Cho AS, Jeon SM, Kim MJ, Yeo J, Seo KI, Choi MS, Lee MK.Chlorogenic acid exhibits anti-obesity property and improves lipidmetabolism in high-fat diet-induced-obese mice. Food Chem Toxicol2010;48:937–43.

27. Faeh D, Minehira K, Schwarz JM, Periasamy R, Park S, Tappy L.Effect of fructose overfeeding and fish oil administration on hepatic denovo lipogenesis and insulin sensitivity in healthy men. Diabetes 2005;54:1907–13. (Published erratum appears in Diabetes 2006;55:563.)

28. Le KA, Ith M, Kreis R, Faeh D, Bortolotti M, Tran C, Boesch C, TappyL. Fructose overconsumption causes dyslipidemia and ectopic lipiddeposition in healthy subjects with and without a family history of type2 diabetes. Am J Clin Nutr 2009;89:1760–5.

29. Harris JA, Benedict FG. A biometric study of human basal metabolism.Proc Natl Acad Sci USA 1918;4:370–3.

30. Le KA, Faeh D, Stettler R, Ith M, Kreis R, Vermathen P, Boesch C,Ravussin E, Tappy L. A 4-wk high-fructose diet alters lipid metabolismwithout affecting insulin sensitivity or ectopic lipids in healthy hu-mans. Am J Clin Nutr 2006;84:1374–9.

31. Chong DG, Kreis R, Bolliger CS, Boesch C, Slotboom J. Two-dimensional linear-combination model fitting of magnetic resonancespectra to define the macromolecule baseline using FiTAID, a FittingTool for Arrays of Interrelated Datasets. MAGMA 2011;24:147–64.

32. Boesch C, Decombaz J, Slotboom J, Kreis R. Observation of intra-myocellular lipids by means of 1H magnetic resonance spectroscopy.Proc Nutr Soc 1999;58:841–50.

33. Tappy L, Schneiter P. Measurement of substrate oxidation in man.Diabetes Metab 1997;23:435–42.

34. Acheson KJ, Zahorska-Markiewicz B, Pittet P, Anantharaman K,Jequier E. Caffeine and coffee: their influence on metabolic rate andsubstrate utilization in normal weight and obese individuals. Am J ClinNutr 1980;33:989–97.

35. Hursel R, Viechtbauer W, Dulloo AG, Tremblay A, Tappy L, RumplerW, Westerterp-Plantenga MS. The effects of catechin rich teas andcaffeine on energy expenditure and fat oxidation: a meta-analysis.Obes Rev 2011;12:e573–81.

36. Beaudoin MS, Graham TE. Methylxanthines and human health: epi-demiological and experimental evidence. Handb Exp Pharmacol 2011(200):509-48.

37. Bortolotti M, Kreis R, Debard C, Cariou B, Faeh D, Chetiveaux M,Ith M, Vermathen P, Stefanoni N, Le KA, et al. High protein intakereduces intrahepatocellular lipid deposition in humans. Am J Clin Nutr2009;90:1002–10.

38. R Core Team. R: a language and environment for statistical computing.2.15.3 ed. Vienna, Austria: R Foundation for Statistical Computing,2013.

39. Matsuda Y, Kobayashi M, Yamauchi R, Ojika M, Hiramitsu M, InoueT, Katagiri T, Murai A, Horio F. Coffee and caffeine improve insulinsensitivity and glucose tolerance in C57BL/6J mice fed a high-fat diet.Biosci Biotechnol Biochem 2011;75:2309–15.

40. Panchal SK, Poudyal H, Waanders J, Brown L. Coffee extract atten-uates changes in cardiovascular and hepatic structure and functionwithout decreasing obesity in high-carbohydrate, high-fat diet-fed malerats. J Nutr 2012;142:690–7.

41. Sulaiman SF, Moon JK, Shibamoto T. Investigation of optimumroasting conditions to obtain possible health benefit supplement, anti-oxidants from coffee beans. J Diet Suppl 2011;8:293–310.

42. Lecoultre V, Egli L, Carrel G, Theytaz F, Kreis R, Schneiter P, Boss A,Zwygart K, Le KA, Bortolotti M, et al. Effects of fructose and glucose

274 LECOULTRE ET AL


ajcn.nutrition.orgD

ownloaded from


overfeeding on hepatic insulin sensitivity and intrahepatic lipids inhealthy humans. Obesity (Silver Spring) 2013;21:782–5.

43. Johnston RD, Stephenson MC, Crossland H, Cordon SM, Palcidi E,Cox EF, Taylor MA, Aithal GP, Macdonald IA. No difference betweenhigh-fructose and high-glucose diets on liver triacylglycerol or bio-chemistry in healthy overweight men. Gastroenterology 2013;145:1016–25.e2.

44. Gutierrez-Grobe Y, Chavez-Tapia N, Sanchez-Valle V, Gavilanes-Espinar JG, Ponciano-Rodriguez G, Uribe M, Mendez-Sanchez N. Highcoffee intake is associated with lower grade nonalcoholic fatty liverdisease: the role of peripheral antioxidant activity. Ann Hepatol 2012;11:350–5.

45. Goodpaster BH, He J, Watkins S, Kelley DE. Skeletal muscle lipidcontent and insulin resistance: evidence for a paradox in endurance-trained athletes. J Clin Endocrinol Metab 2001;86:5755–61.

46. Schwarz J-M, Neese RA, Turner S, Dare D, Hellerstein MK. Short-term alterations in carbohydrate energy intake in humans. Strikingeffects on hepatic glucose production, de novo lipogenesis, lipolysis,and whole-body fuel selection. J Clin Invest 1995;96:2735–43.

47. Henry-Vitrac C, Ibarra A, Roller M, Merillon JM, Vitrac X. Contri-bution of chlorogenic acids to the inhibition of human hepatic glucose-6-phosphatase activity in vitro by Svetol, a standardized decaffeinatedgreen coffee extract. J Agric Food Chem 2010;58:4141–4.

48. Ong KW, Hsu A, Tan BK. Anti-diabetic and anti-lipidemic effects ofchlorogenic acid are mediated by ampk activation. Biochem Pharmacol2013;85:1341–51.

49. Yamauchi R, Kobayashi M, Matsuda Y, Ojika M, Shigeoka S,Yamamoto Y, Tou Y, Inoue T, Katagiri T, Murai A, et al. Coffee andcaffeine ameliorate hyperglycemia, fatty liver, and inflammatory adi-pocytokine expression in spontaneously diabetic KK-Ay mice. J AgricFood Chem 2010;58:5597–603.

50. Shearer J, Farah A, de Paulis T, Bracy DP, Pencek RR, Graham TE,Wasserman DH. Quinides of roasted coffee enhance insulin action inconscious rats. J Nutr 2003;133:3529–32.

51. Cornelis MC, El-Sohemy A, Kabagambe EK, Campos H. Coffee,CYP1A2 genotype, and risk of myocardial infarction. JAMA 2006;295:1135–41.

52. Acheson KJ. Caffeine and insulin sensitivity. Metab Syndr Relat Dis-ord 2005;3:19–25.

53. Mure K, Maeda S, Mukoubayashi C, Mugitani K, Iwane M, KinoshitaF, Mohara O, Takeshita T. Habitual coffee consumption inversely as-sociated with metabolic syndrome-related biomarkers involving adi-ponectin. Nutrition 2013;29:982–7.

54. Yki-Jarvinen H. Nutritional modulation of nonalcoholic fatty liverdisease and insulin resistance: human data. Curr Opin Clin Nutr MetabCare 2010;13:709–14.



ajcn.nutrition.orgD

ownloaded from


am j clin nutr 2014 lecoultre 268 75

Documents