Introduction

Performing an exercise session can increase reactive species production beyond the body’s capacity for antioxidant defense, resulting in the oxidation of cellular components1,2,3,4,5. On the other hand, reactive oxygen species (ROS) and reactive nitrogen species (RNS) that are transiently generated during exercise participate in signaling for cellular adaptations and muscle strength production. However, high levels of ROS/RNS can lead to contractile dysfunction, resulting in reduced performance1,2,4. Furthermore, oxidative stress (OS) chronic manifestation is known to be involved with a large number of diseases, including cardiovascular disease, cancer and diabetes. And thus, preventing them can help maintain performance and health6.

The magnitude and duration of OS in physically active individuals are influenced by the type, intensity and duration of exercise, training status antioxidant intake, and by the interaction of these factors3,5,7. Physical training associated with low antioxidant intake may represent a period of increased vulnerability to OS7. On the other hand, in most cases, a well-balanced and varied diet seems to meet physically active individuals’ antioxidant requirements4,5and thus help in injury prevention and the maintenance of performance and health5. In these scenarios, Dietary antioxidant Capacity (DaC) may be a valuable tool in physically active populations.

DaC is based on the cumulative and synergistic activities of all antioxidants present in the consumed foods8,9,10and has a relationship with diet quality markers10. Although it has been primarily studied in the context of chronic diseases8,11. DaC has been validated as a useful methodological tool for predicting plasma and dietary antioxidant concentrations in a young healthy population9. Furthermore, DaC has been found to be inversely associated with central adiposity (waist circumference) as well as with metabolic (glycemia, total cholesterol: HDL-c ratio, triglycerides) and OS biomarker (oxidized-LDL) in healthy young adults8. Nonetheless, to our knowledge, there are no studies investigating DaC in physically active individuals using the range of OS markers. Considering the potential for elevated OS in physically active individuals, it is imperative to thoroughly examine the correlation between DaC and OS biomarkers, alongside with markers of body composition. This knowledge in physically active individuals is imperative given the importance of OS modulation in the context of health and physical training, especially in periods of high physiological demand, and of the potential of antioxidant-rich diets to minimize the deleterious actions of reactive species without apparently impairing training adaptations1,2. Moreover, this study will contribute to improving nutritional interventions to enhance body composition in physically active individuals. Soccer referees were chosen as active individuals for their constant involvement with physical demands of match play.

We hypothesized that higher DaC values ​​are positively associated to an improvement in OS biomarkers and body composition variables. Therefore, this study aimed to analyze DaC associations with OS biomarkers, and body composition measures in healthy and physically active individuals – elite national soccer referees.

Methodology

Sample

This was an observational, cross-sectional study. The sample consisted of 20 elite male soccer referees who were members of the Brazilian Soccer Confederation (CBF). Sampling was intentional and non-probabilistic. One individual was excluded due to having metal prostheses. The study was approved by the Human Research Ethics Committee of the Federal University of Santa Catarina (UFSC), CAAE 82584318.0.0000.0121, and file no. 2.572.301. All participants signed an informed consent form. All methods were performed in accordance with the relevant guidelines and regulations.

Data collection

The study occurred in November 2018. All data collection was performed in the morning, ordered at the Department of Nutrition at UFSC, Brazil, by trained professionals. For blood and urine collection, and body composition assessment, participants were instructed to fast for 10–12 h, wear appropriate clothing, abstain from caffeine and alcohol 24 h prior, and not exercise on the day before the collection. Moreover, at the body composition assessment participants were barefoot and the urine was collected 30 min before the body composition record. Participants alongside with health profile analysis (laboratory measurements of serum biochemical metabolites and OS biomarkers in whole blood, plasma, serum, or urine), also completed questionnaires regarding socioeconomic data, physical activity level, and food intake.

Determining the physical activity level

The short version of the International Physical Activity Questionnaire (IPAQ) was used to establish the participants’ level of physical activity. This instrument was validated in a sample of the Brazilian population12. The classification followed the Guidelines for Data Processing and Analysis of IPAQ – Short Version. The questionnaire evaluates the weekly energy expenditure according to walking, and moderate and vigorous activities. The value is obtained by multiplying the time and the number of days per week spent on each activity by the metabolic equivalent of the task (MET) pre-established by IPAQ itself for a given activity.

Anthropometry and body composition

Body mass was recorded on an electronic scale (model RIW200 Welmy®, São Paulo, SP, Brazil), with 100-g accuracy. Height was measured on a stadiometer (Alturaexata®, Belo Horizonte, BH, Brazil), with 1-mm accuracy. Waist circumference was measured using an inelastic tape measure. The percentage of body fat (%BF) and lean mass were evaluated by dual-energy X-ray absorptiometry (DXA) (Lunar ProdigyAdvance, General Electric-GE®, Madison, WI, USA). The technical error of measurement for body composition assessment using DXA ranges between 1 and 2% for body fat (BF) and lean mass, indicating high precision for this assessment technique13,14,15. In our study, we minimized the error of measurement through regular equipment calibration, rigorous operator training and standardized procedures during all measurements. Furthermore, all measurements were conducted by the same researcher.

Food intake and dietary antioxidant capacity

To evaluate dietary intake, 24-hour recall were used. Participants were instructed to complete three recalls on non-consecutive days, one on the weekend and two on weekdays. The first two were collected by telephone and the last one was collected in person. To reduce possible biases, the Multiple-Pass Method16was used. The same interviewer collected all the recalls for each participant. The data obtained were converted into grams or milliliters, using the home measurement Table17 or weighed on an analytical scale (model YP-B20002, Bioscale®, Paraná, PR, Brazil). Energy, carbohydrate, protein, and lipid intakes were assessed using the NDSR program (Nutrition Data System for Research®Software, Graduate Pack 2017 version; NCC Food and Nutrient Database, University of Minnesota, Minneapolis, MN, USA). Inter and intrapersonal variability of macronutrient intake was adjusted. For DaC assessment, a table with over 3,100 registered foods, divided into 24 food groups, was used to calculate the antioxidant capacity of each food consumed by the participants10. The DaC score was calculated by adding the individual ferric reducing-antioxidant power assay values of each food, and it was then expressed as DaC in mmol/day. The DaC values were adjusted for total energy value according to the residual method18.

Preparation of biological samples

Blood was collected by puncture of the median antebrachial vein using a vacuum system (Vacuntainer-BD, São Paulo, Brazil) in dry tubes or with Ethylene diamine tetra-acetic acid by a trained and qualified professional. Plasma and serum samples were obtained by centrifugation (1,000 g for 10 min, at 4 °C). For urine collection, the volunteers were instructed to collect the first urine in the morning in a previously provided container containing Butylated Hydroxytoluene (BHT) as a preservative.

The following OS markers were evaluated: F2-isoprostane, total antioxidant state (TAS), total oxidative status (TOS), reduced (GSH) and oxidized glutathione (GSSG) and the antioxidant enzymes superoxide dismutase (SOD), catalase and glutathione peroxidase (GPx).

For GSH and GSSG evaluation, whole blood was transferred to micro tubes containing 100 µL of 310 mM N-ethylmaleimide (NEM) per milliliter of blood. A hemolysed blood sample from 100 µL of erythrocytes with 1 mL of hemolysing solution (4 nM MgSO4 and 1 nM acetic acid) was used for determination of GPx, SOD, and catalase enzymes and hemoglobin (Hb) concentration. Serum samples were stored at −80 °C for the subsequent TAS and TOS dosage. In order to determine F2-isoprostane, a 900 µL aliquot of urine was transferred to a micro tube and mixed with 10 µL of 2 mM BHT, a fat-soluble antioxidant, and then frozen at −80 °C.

Biochemical analysis

GSH concentration was calculated in whole blood by HPLC, and the results were expressed as µmol/g Hb. GSSG was determined in whole blood by spectrophotometry (UV-1800 – Shimadzu®Tokyo, Japan), and the results expressed as nmol/g Hb. Both according to previously described procedures19.

GPx activity was measured by monitoring the oxidation of the reduced tetrasodium salt of β-nicotinamide adenine dinucleotide 2’-phosphate (NADPH) in the presence of hydrogen peroxide20, and the results were expressed as mU/mg Hb. SOD activity was measured by the dismutation of the superoxide radical into hydrogen peroxide and water (Sigma-Aldrich®Saint Louis, Missouri, USA), and the results were expressed as USOD/mg Hb. Catalase activity was measured by its peroxidative function21, and the results expressed as U/g Hb. Hb, which was used to standardize the results of GSH, GSSG, SOD, catalase and GPx, was evaluated by a routine colorimetric technique using a spectrophotometer (UV-1800 – Shimadzu® Tokyo, Japan) with a Labtest® kit (Lagoa Santa, Minas Gerais, Brazil).

TAS was evaluated according to the previously described method22, with the results expressed as Trolox equivalent mmol/liter. Serum TOS was evaluated according to the previously described method23 with the results expressed as H2O2 equivalent µmol/liter.

The free form of F2-isoprostanes in urine was quantified using the 8-Isoprostane ELISA kit (Cayman Chemical, Ann Arbor, Michigan USA), according to the manufacturer’s protocol. Urine samples were thawed and diluted 10-fold prior to analysis. F2-isoprostane concentration was normalized by 1 mg creatinine and expressed as pg/mg urine creatinine24.

For serum biochemical analyses of uric acid and triglyceride, determination was performed by the standard method on the Dimension RxL Max® automated analyzer (Siemens Healthcare Diagnostics Inc., Marburg, Germany).

Statistical analysis

Statistical analyses were performed using STATA, version 13.1 (StataCorp LP, USA). Data normality was evaluated using the Shapiro-Wilk test, kurtosis and asymmetry analysis. Multiple linear regression analyses were performed to examine the association between DaC with OS biomarkers, and between DaC with body composition variables. The Variance Inflation Factor (VIF) was evaluated to detect multicollinearity. Residuals analysis was performed to investigate the fit of the regression models based on the residuals. To this end, a normality test, Shapiro-Wilk, was performed. The significance level adopted was 5% (p < 0.05). For each model analyzed, the regression coefficient (β), the fitted R2, and its 95% confidence interval (95% CI) were estimated. To measure the effect size, Cohen’s f2 (f2) was estimated using the formula f2 = R2/1 – R222,23. Effect size was interpreted as small (between 0.02 and 0.14), medium (between 0.15 and 0.34), or large (≥0.35).

Results

The variables for body composition, food intake, and OS biomarkers are shown in Table 1.

Table 1 Sample characterization, dietary intake variables and oxidative stress biomarkers (n = 20).
Full size table

Of the 24 food groups that compose DaC (Supplementary Table I), six groups showed more than 5% contribution alone to the DaC. Among of these three groups contributed more than 50% of the total DaC: beverages, fruits, and cereals. The beverage group, represented mainly by coffee, contributed with 27,87% of DaC.

The association between DaC and OS biomarkers is shown in Table 2. DaC was inversely associated with the lipoperoxidation marker F2-isoprostane (p = 0.044) after fitting for age and triglycerides (Adjusted Model), and the effect size was large (f2 = 0.66). Furthermore, DaC was inversely associated with the GPx enzyme activity (p = 0.048) after fitting for protein intake and total MET (Adjusted Model), and the effect size was small (f2 = 0.02). There were no significant associations between DaC and the other OS biomarkers.

Table 2 Association between Dietary antioxidant capacity (DaC) and oxidative stress blood biomarkers (n = 20).
Full size table

Table 3 shows the linear regression values between DaC and body composition parameters. DaC was inversely associated to %BF (p = 0.025) after fitting for height, family income and lean mass (Adjusted Model), and the effect size was large (f2 = 1.25). The associations between DaC and body mass, lean mass, and waist circumference showed large effect sizes (f2 = 2.50, f2 = 2.77, and f2 = 1.00, respectively) (Adjusted Model), although they were not significant, (p = 0.159, p = 0.116, p = 0.189, respectively).

Table 3 Association between Dietary antioxidant capacity (DaC) and body composition parameters (n = 20).
Full size table

Discussion

Our hypothesis that higher DaC values ​​are positively associated to improvement in OS biomarkers, and body composition variables in healthy individuals and physically active was confirmed, in part, by the association with F2 isoprostane, and %BF, by this study. To the best of our knowledge, these are the first reports of associations between DaC with OS biomarker, and with body composition in healthy physically active individuals.

In this study, the major contributor to DaC (27,87%) was the beverage group, which was mostly represented by coffee. The mechanisms underlying the antioxidant effects of coffee are still poorly understood in vivo25. However, evidence in an animal model suggested that coffee’s antioxidant capacity was greater than its individual bioactive components (i.e., chlorogenic acid, caffeic acid, and caffeine)26. Although, it is important to consider that the interactions of the bioactive constituents of a specific food item, and especially of the diet per se, are remarkably intricate and can result in additive, synergistic, inhibitory, neutralizing, or even opposing effects on a specific biological event27.

In this study, DaC was inversely associated with the lipoperoxidation marker F2-isoprostane. For every 1 mmol increase in DaC, there was a decrease of 3.06 pg F2-isoprostane/mg urine creatinine. Despite the lack of information relating DaC to OS indicator responses in an exercise situation, our result was consistent with the independent and inverse association between DaC and plasma LDL-oxidized levels observed in healthy young adults8. Furthermore, the association between DaC and a lipoperoxidation indicator in elite soccer referees can be supported, at least in part, by intervention studies using antioxidant-rich foods in physically active healthy adults and in athletes28,29,30,31,32.

In physically active men, consumption of dark chocolate (40 g) for two weeks prior to performing a bicycle exercise protocol (1.5 h at 60% maximal oxygen uptake (VO2max) with increments up to 90% VO2max every 30 s, followed by cycling at 90% VO2max to exhaustion) resulted, compared to control, in lower basal plasma levels of F2-isoprostane and oxidized LDL, and smaller increases in such markers immediately and one h post-exercise29. In another study, physically active men received chocolate drinks (100 mL) with high flavonoid content (HFCD) or low flavonoid content (LFCD) and performed an incremental bicycle test or not two h after the interventions. Plasma levels of F2-isoprostane were lower with HFCD vs. LFCD four hour after dietary interventions only, and two and four hours after intakes in the exercise condition30. In addition to possible benefits in acute exercise, antioxidant-rich foods appear to have the potential to improve adaptive responses to training. In soccer players, sesame consumption (40 g) during 28 days of training promoted, at the end of such a period, a greater reduction in baseline serum malondialdehyde (MDA) levels compared to placebo33.

TAS analysis in plasma samples allow the detection of non-enzymatic antioxidants, including dietary ones, as well as synergistic interactions between antioxidants, although it does not provide information on the nature of the compounds34. The large effect size (f2= 0.51) on the strength of the direct association between DaC and plasma TAS in soccer referees was consistent with the proposal that DaC may be a good predictor of plasma antioxidant status in healthy populations of both sexes9,35. However, TAS may change in response to acute exercise and/or training31, which could, therefore, influence the associations between DaC and TAS. TAS increased 24 and 48 h after a marathon in trained runners. However, TAS before and after the race was higher when athletes had consumed cherry juice (~ 240 mL for 7 days) as compared to placebo31.

DaC showed an inverse and significant association with blood GPx activity. Considering GPx participation in the elimination of lipid peroxides, in addition to other reactive species36,37, this finding seems to be partly in line with the associations of DaC with urinary F2-isoprostane and plasma TAS levels. Although DaC was not significantly associated with GSH and GSSG concentrations, together our results could partly support the proposal that dietary antioxidants could exert a GSH sparing effect38. SOD and catalase activities showed no significant associations with DaC in the present study. Our results agreed with the ineffectiveness of dietary intervention on adaptive antioxidant enzyme responses to training31,32. Supplementation with lemon verbena (Lippia citriodora) extract for 21 days did not influence the increase in the activities of enzymes catalase, GPx and glutathione reductase in blood after 21 days of training (90 min of treadmill running) in that period39.

It is worthy of note here that changes in the activity of antioxidant enzymes induced by acute exercise and/or training depend on the exercise protocol employed1. Moreover, the mechanism of antioxidant action partly involves the activation of transcription factors, such as the transcription factor Nrf-2 (nuclear factor erythroid 2-related factor 2) and NFκB (nuclear factor kappa B), whose activities can be modulated both by exercise and by dietary bioactive constituents1,37. Therefore, the complex modulatory potential of the diet added to the adjustments inherent to exercise may hinder a more integrative interpretation of the associations between DaC and OS markers. Nevertheless, overall, the associations found in the present study were consistent with the effects of antioxidant-rich foods on OS markers observed in studies with dietary intervention associated with exercise.

With regard to body composition, our findings indicated, for the first time, an inverse association between DaC and %BF, where for each 1 mmol increase in DaC, there was a 0.8% reduction in BF. This result was consistent with the strong inverse association of DaC with waist circumference (f2= 1.00). Studies have shown inverse associations of DaC with central adiposity - based on waist circumference measurement - and BMI in healthy young adults and elderly individuals8,40. In a study on obese participants, DaC was the major contributor to reduction in body mass and obesity markers (BMI, waist circumference, and %BF) compared to several dietary components, including omega-3 fatty acid intake, glycemic load, and the Healthy Eating Index41. Therefore, our study and those by others8,40seem to indicate that, as in obesity41, DaC is inversely associated with indicators of worsening body composition in healthy individuals, including those who are physically active.

We observed that DaC showed strong direct association (f2= 2.50) with body mass. Apparently, this result disagreed with previous findings, which indicated inverse associations of DaC with BMI and dietary energy density in healthy individuals40,41. However, such an association between DaC and body mass may have reflected the direct and strong association (f2= 2.77) occurring between DaC and lean mass, particularly due to the fact that lean mass is a component of body mass and susceptible to training- and diet-induced gains1,2 1,2. Therefore, the associations between DaC and lean mass need to be further investigated in physically active populations and athletes, which could help to understand, for example, the influence of DaC on muscle responses to acute exercise and training in the future.

While increasing antioxidant intake from whole foods may improve body composition and reduce OS, which are the focus of this study, isolated antioxidant supplements, may not be advisable as they could interfere with the body’s natural adaptation to physical exercise5. Taken together, our findings showed that DaC is inversely associated with %BF and the OS marker F2-isoprostane, suggesting that consuming foods rich in antioxidants may improve wealth indicators, such as body composition and OS biomarkers.

A limitation of the present study is the cross-sectional nature of the analyses, which precludes making inferences regarding causality. Another important limitation is the sample size, which may affect the interpretation and generalizability of results. Thus, the effect size estimation was used as a complement to the test for statistical significance. Our study did not provide information on associations between DaC and GSH: GSSG and TAS: TOS ratios, which represent important redox-balance indicators. Unfortunately, in multiple regression analyses including such dependent variables, VIF measurements indicated multicollinearity and residual analyses suggested that none of the various models tested were suitable.

We highlight the following aspects as strengths in the study: the measurement of F2-isoprostanes, which is a valid and reliable marker to assess OS in vivo; the investigation of robust range of OS biomarkers; the pioneering investigation on the relationship between DaC and %BF; the use of a reference methodology, DXA, to assess body composition; and the use of a validated methodology to collect food recalls, the Multiple-Pass Method. In addition, the process was carefully standardized from data collection to the input of dietary information in the software used. The table utilized to calculate DaC contains foods types from all over the world, which may over or underestimate the values of foods consumed by the population in this study. However, possible errors inherent to food consumption data were circumvented by using standardization from data collection to tabulation, food choice, and calculation.

Conclusion

Our results suggest that, in physically active individuals, increased DaC may bring benefits related to markers of OS, shown in these findings by decreasing urinary F2 isoprostane and increasing GPx activity, partly explained by the effect of GPx on participation in the elimination of lipid peroxides. The increase in DaC also showed an improvement in body composition, demonstrated by the lower %BF in the study subjects. Another finding, at first contradictory, was the strong direct association between CaD and body mass, which can be explained by the strong direct association of CaD with lean mass.