Physical Activity Offsets the Negative Effects of a High-fructose Diet

Amy J. Bidwell; Timothy J. Fairchild; Jessica Redmond; Long Wang; Stefan Keslacy; Jill A. Kanaley


Med Sci Sports Exerc. 2014;46(11):2091-2098. 

In This Article

Subjects and Methods

Twenty-two healthy male (n = 11) and female subjects (n = 11) between the ages of 18 and 25 yr were recruited from the Syracuse University community. Subjects had to be recreationally active, as determined by a PA questionnaire (3–4 d·wk−1 of moderate-to-vigorous activity for at least 20–60 min·d−1), with a body mass index (BMI) <27 kg·m−2 (mean, 22.5 ± 1.6 kg·m−2). All subjects completed an informed consent form approved by the Syracuse University institutional review board before participating in this study. Exclusion criteria included the use of lipid and/or glucose-lowering medications or other medications that may affect glucose and lipid metabolism (e.g., antidepressants, oral contraceptives, etc.), chronic nonsteroidal anti-inflammatory drug use (more than two times a week), daily antioxidant supplementation, orthopedic limitations to walking, type 2 diabetes or glucose intolerance, overt cardiovascular disease, hypertension, and/or an abnormal lipid profile. Subjects were excluded if they were currently ingesting more than one high-fructose drink per day (>20 g). Women started the intervention period within the first 7 d of their menstrual cycle to minimize the potential effects of estrogen on glucose/insulin concentration.[12]


Initially, subjects participated in a 1-wk control period to determine their normal PA (Table 1) and dietary habits. The two interventions were separated by a 2-wk washout period, and participants were randomly allocated to each intervention using a counterbalanced crossover experimental design. During each intervention period, the subjects' usual ad libitum diet was supplemented with an additional 75 g of fructose per day. The first intervention involved 2 wk of high PA levels (>12,500 steps) (FR+active). The second intervention involved low PA (<4500 steps) (FR+inactive). A study day with a fructose-rich meal was given at the beginning and at the end of each 14-d intervention, as described in the following section.

Prescreening Visit

All subjects completed a medical history, PA, and sedentary behavior questionnaire before the start of the intervention. After a review of the subject's activity level with the use of a validated questionnaire monitoring PA at work, during leisure time, and during sport,[10] the potential subjects were excluded from participation if their activity level included regular structured exercise more than five times per week or less than two times per week. Height and weight were measured, and body composition was assessed by air displacement plethysmography (BOD POD system; Life Measurement, Inc., Concorde, CA).[29] After anthropometric measurements were completed, subjects performed a graded exercise test on the treadmill using a protocol that has been previously published[13] to determine peak oxygen consumption (V̇O2peak).

After the initial visit, the subjects underwent a nutritional consultation with a registered dietitian to ensure compliance with the dietary intervention and to estimate normal fructose intake. Subjects were instructed to refrain from ingesting any added sugar such as sweetened beverages, fruit juices, pastries, and cookies aside from the study drinks during the intervention period. Furthermore, estimates of food intake during the control and intervention periods were collected by random 24-h recall (via telephone) two times per week using the US Department of Agriculture five-step multiple-pass method.[8] The same registered dietitian administered the recall to all subjects. Recalls were then analyzed using Diet Analysis Plus (version 7; Thomson Wadsworth, Thomson Corp., Independence, KY). After the initial visit, the subjects began a 1-wk control period, at which time, they were instructed to maintain normal activities of daily living and dietary habits. During this period, PA (steps) was monitored with the use of an accelerometer (ActiGraph® GT3X activity monitor; ActiGraph Corp., Pensacola, FL), which was later uploaded to the computer for further analysis. Subjects also were given pedometers (Accusplit®, Livermore, CA) to provide visual feedback regarding daily step count. Subjects were asked to refrain from any additional exercise throughout the study duration.

Visit 1

At the end of the 7-d control period, the subjects arrived at the Human Performance Laboratory at Syracuse University at 0700 h after a 12-h fast and no exercise 24 h before testing. Subjects had a catheter inserted into the antecubital vein by a registered nurse. Subjects then rested in a supine position for 30 min before obtaining two baseline blood samples (10 mL each). After baseline blood samples were obtained, a test meal was prepared for subjects. The test meal included the following: 139.5 g of Wegmans® large eggs, 65.55 g of Thomas Better Start Light® Multi-Grain English Muffins, 22.54 g of I Can't Believe Its Not Butter® Mediterranean Blend butter, and a high-fructose corn syrup drink consisting of 20.6 g of Swanson® fructose, 16.9 g of NOW® sports glucose, 1.1 g of Great Value® artificial sweetener, and 236.5 g of Vintage® sodium-free carbonated water (600 kcal, 45% CHO (25% fructose and 20% complex), 40% fat, 15% protein, and 5 g of fiber). After the test meal, blood samples were obtained at the following time points: 5, 10, 15, 20, 30, 40, 50, 60, 75, 90, 120, 150, 180, 210, 240, and 360 min. For the duration of the study day, subjects were instructed to sit on a reclining chair and abstain from any strenuous activity.


The subjects consumed a fructose-rich diet containing 74.9 g·d−1 of fructose (two 20-oz Lemon Lime WPOP® drinks; Rochester, NY) along with their ad libitum diet during both conditions. The ad libitum diet was chosen on the basis that sugar-sweetened beverages are usually consumed in conjunction with an ad libitum diet.[32] Previous research has indicated that within 7 d, metabolic abnormalities can occur with a high-fructose diet;[1] therefore, a 2-wk intervention was chosen to ensure that changes will occur. The subjects collected their beverages twice weekly from the Human Performance Laboratory. They were required to return their empty drink bottles to the laboratory once per week to assess drink compliance and to record step counts. Each subject met with a registered dietitian who assisted in maintaining eating habits and recording of weight during these weekly visits to the laboratory.

Visit 2

On the day after the 14-d intervention period, a postintervention test meal was provided using the same procedures as those in visit 1. After visit 2, subjects were instructed to maintain their normal activities of daily living and dietary habits for 2 wk. Previous research has indicated that a 2-wk washout period is adequate to normalize metabolic markers associated with hyperlipidemia.[20] Visits 3 and 4 were then completed before and a day after the alternate 2-wk intervention period, respectively, using the same procedures as those outlined for visits 1 and 2.

Metabolic Assays

A lipid profile (Cholestech LDX; Biosite International, San Diego, CA) was performed on the samples taken at -5, 0, 60, 120, 180, 240, and 360 min and measured TG, VLDL, total cholesterol, and glucose concentrations. The use of this equipment has previously been validated.[26] In addition, blood samples obtained at -5, 0, 5, 10, 15, 20, 30, 40, 50, 60, 75, 90, 120, 150, 180, 210, 240, and 360 min were transferred to BD Vacutainer® Plus Plastic EDTA tubes (Franklin Lakes, NJ), separated by centrifugation, divided into two sets of polypropylene tubes, and stored at -80°C for subsequent analysis. Insulin, TNF-α, and IL-6 were analyzed using Luminex xMap Technology (Linco Research, St. Charles, MO) on a Luminex 100/200 platform (Luminex Corp., Austin, TX). All procedures followed the manufacturer's instructions (Millipore, Billerica, MA), with quality controls within expected ranges for each assay (insulin: interassay coefficient of variation (CV) = 5.0%, intraassay CV = 4.0%; TNF-α: interassay CV = 9.9%, intraassay CV = 10.6%; IL-6: interassay CV = 10.3%, intraassay CV = 11.9%). C-reactive protein (CRP) assays were performed using the Quantikine assay kit (R&D Systems, Inc., Minneapolis, MN) (interassay CV = 6.5%, intraassay CV = 4.2%). Insulin sensitivity was calculated by the homeostatic model assessment method as previously described by Levy et al.[19] and the quantitative insulin sensitivity check index.[7]


All results were reported as mean ± SEM using SPSS 19.0 (Chicago, IL). Descriptive variables (n = 22) and dietary analysis were analyzed using a two-way repeated-measures ANOVA to depict differences in pre- and postintervention weight, BMI, percent body fat, and macronutrient consumption. Postprandial responses for all blood variables were determined by calculating the total area under the curve (tAUC) or incremental AUC (iAUC) (Excel; Microsoft Corp., Redmond, WA) and absolute change from peak to baseline concentrations (Δpeak) for all variables. A between-subject analysis was performed on all variables to depict differences in genders. A log transformation (log10) was used for data that were not normally distributed on the basis of visual appearance of skewed data. Transformations were applied so that the data more closely met the assumptions of normality on the basis of parametric analysis statistical procedures.[18] Lipid and inflammatory variables were analyzed using a three-way ANOVA with repeated measures to assess the changes in lipid measures and inflammatory markers over the 6-h test day: 2 (high vs low PA) × 2 (pre- vs postintervention) × 18 (time points). If a significant interaction was found, differences between time points were analyzed using a paired t-test with Bonferroni correction factor. Statistical significance for AUC and Δpeak concentrations was computed using a two-way repeated-measures ANOVA (intervention × pre–post). Pearson correlation coefficient was computed to determine any correlations between fasting TG and inflammatory variables. A priori significance was set at P < 0.05. Sample size was determined by a previous research with similar methodology.[3]