FormalPara Key Points

Regardless of intensity, volume, type of exercise and studied population, antioxidant parameters seem to increase and pro-oxidant indicators to decrease after physical training.

While data relating to redox balance after exercise training for specific populations are sparse, the available evidence suggests that the general population can undertake any kind of physical training to decrease oxidative stress and prevent and treat metabolic diseases.

1 Introduction

Worldwide, about 30 % of adults are insufficiently active [1], which is a greater risk factor than smoking, high cholesterol, and elevated blood pressure for cardiovascular disease and all-cause mortality [2, 3]. A sedentary lifestyle is associated with several diseases, such as obesity [4], type 2 diabetes mellitus (T2DM) [5], hypertension [6], coronary artery disease [7], and many others. Thus, physical fitness components have been frequently used in clinical and scientific practice as markers of health status [8, 9], highlighting the importance of regular physical exercise for health [10].

The conditions by which a sedentary lifestyle may lead to diseases are diverse, but usually are related to metabolic and immune disorders [11], which may be both preceded by [12] and/or result in degenerative processes such as oxidative stress (OS) and chronic inflammation [13] that in turn may accelerate the aging process. OS is the imbalance between reactive oxygen species (ROS), which are byproducts of aerobic metabolism, and the mechanisms of defense and repair, known widely as antioxidants [14]. Although ROS may trigger fundamental signaling pathways in skeletal muscle [15], its exacerbated production may lead to intra- and extra-cellular damage-induced malfunction of molecular mechanisms and chronic inflammatory conditions [16–19]. OS is associated with the development of various diseases, including atherosclerosis [20], cardiovascular diseases [21], T2DM [22], cancer [23, 24], and neurological diseases [25].

On the other hand, it is well-established that physical exercise may reduce inflammation in the long term [26–30]. Today, the scientific evidence is strong regarding how a physically active lifestyle attenuates OS. This attenuation may be one of the mechanisms responsible for the improvement in several clinical aspects, such as attenuated cellular aging [31], increased insulin sensitivity and lipid profile regulation [32], and reduced endothelial dysfunction [33] after exercise training.

Traditionally, OS is quantified by pro-oxidant parameters, which may indicate DNA damage, lipid peroxidation, or protein oxidation. On the other hand, these effects seem to be attenuated through the activity of antioxidants, which may be enzymatic or non-enzymatic, and reflect the total antioxidant capacity [34].

Regarding the OS response to physical exercise, it seems that exercise acutely increases ROS production followed by antioxidant compensation [35]. Longitudinally, studies have suggested that the effect on OS of physical exercise may vary according to the exercise mode, volume, intensity, and population, making it difficult to reach a consensus about the effects of exercise on oxidative balance. Vincent et al. [4] suggested that physical training would be the only method capable of reducing OS independently of body adiposity, when compared with other interventions such as caloric restriction, bariatric surgery, and pharmacotherapy or antioxidant supplementation. However, comparisons are quite difficult, as studies have been conducted using a wide variety of experimental designs, e.g., studied samples, interventions, and outcomes [36, 37]. Moreover, although several narrative reviews and different clinical trials have investigated the effects of exercise training on OS parameters, a pooled analysis may add some weight to the conclusions of individual studies. Therefore, to contribute to the literature, this review aimed to summarize the effects of regular physical exercise on pro- and antioxidant parameters through a systematic review and meta-analysis of controlled clinical trials. To the best of our knowledge, this is the first time such a review and analysis has been undertaken.

2 Methods

2.1 Search Strategy

We conducted a systematic review of the literature up to March 2016, including the following databases: PubMed, SCOPUS, and Web of Science. We also manually searched the reference lists of the articles. We chose the date of the first article published in 2004 [38] that related to exercise training effect on OS markers as the initial date of the search.

2.2 Inclusion/Exclusion Criteria

Studies proposed to be included in the review were checked for the following criteria: (1) the study was a full report published in a peer-reviewed journal; (2) the study assessed humans; (3) one or more exercise programs were performed; (4) the study was a randomized or non-randomized clinical/controlled trial; and (5) the keyword combination referred to exercise training. We included OS as part of a thorough and exhaustive search process.

Articles were excluded if (1) we could not obtain the full-text article; (2) exercise was not performed; (3) the study did not include a non-exercise control group; (4) articles were not written in English, Spanish, or Portuguese; and (5) the study used drug administration or nutritional supplementation before or after the exercise (Fig. 1).

Fig. 1
figure 1

Flow diagram for the strategy of searching for the studies

Full size image

2.3 Identification of Eligible Studies

Eligible studies were longitudinal experimental interventions executed in humans that analyzed the effects of exercise training on OS parameters and did not administer any drug or nutritional supplementation. At least two independent researchers read and evaluated the abstracts of all articles identified through the search and selected potentially eligible articles. They also carefully checked the reference list of every article to identify other potentially eligible studies.

2.4 Quality Assessment

The selected studies were subsequently analyzed using a methodological quality assessment (MQA) adapted from the Downs and Black Quality Index [39] and Quality Assessment for Longitudinal Studies [40]. This modified version consists of eight objective questions (Table 1). Each study was allocated a “1” for “yes” or a “0” for “no” for each question, as previously described [36], and responses were summed for a total of eight. A total score ≥6 indicated a high-quality study, a total score of 3–5 indicated moderate quality, and a score <3 indicated low quality.

Table 1 Methodological quality assessment questions
Full size table

2.5 Data Extraction

We extracted all relevant data from studies that met the eligibility criteria: characteristics of the sample, methodological design, exercise modality, training protocol (volume/intensity), method of measuring OS, and the OS outcomes after exercise training. For this analysis, the number of subjects (n) in each study was only considered for the intervention groups and their paired non-exercise group(s).

2.6 Statistical Analysis

We performed the meta-analysis by computing the standardized mean difference (SMD) (Hedges’ g) and using the random-effects model, and performed quantitative analyses on the confined data derived from the last measure of the intervention and control groups of each study. As previously suggested [41], we combined the outcomes for trials with more than one parameter (pro- or antioxidant). For studies with more than one intervention group (i.e., exercise training), we considered each comparison (treated vs. control) as an independent study. We assessed heterogeneity with the Cochran’s Q test and tau-squared (τ 2) and measured inconsistency (the percentage of total variation across studies due to heterogeneity) of effects across exercise interventions using the I 2 statistic proposed by Higgins et al. [42]. We visually and objectively assessed the risk of publication bias with Egger’s test, a funnel plot symmetry. The significance level was set at p < 0.05 for all analyses, except for Egger’s regression to assess funnel plot asymmetry, which was set at p < 0.1, as suggested [43]. We used the aforementioned procedures for two independent analyses, one for pro-oxidant and another for antioxidant parameters. To assess the statistical power of the included studies, we considered the F statistic for analysis of variance (ANOVA) within and between factors according to the number of groups (i.e., experimental 1 + experimental 2 + control) and moments (baseline vs. post-training) of each study. The statistical power was calculated considering an alpha of 5 % and effect size of 0.25 to reach at least 80 % (1−β = 0.8), to assume that the sample size was sufficient to avoid type 1 or 2 error [44, 45]. We used the software Comprehensive Meta-Analysis (CMA 3.0) and G*power (v. 3.0.10) to carry out all procedures.

3 Results

3.1 Search Results

The search of PubMed, SCOPUS, and Web of Science provided a total of 1573 citations and one identified through a reverse search. After adjusting for duplicates, 1086 articles remained. Of these, we discarded 1020 after abstract review because they did not meet the inclusion criteria. We examined the full text of the remaining 66 citations in more detail; 36 appeared to not meet the inclusion criteria as described. A total of 30 studies met all of the criteria and were therefore included in the systematic review. A further 11 studies were excluded from the meta-analysis because they did not provide enough data for quantitative analysis (Fig. 1).

3.2 Study Characteristics

All studies selected for this review were controlled trials published in English (n = 29) or Spanish (n = 1). The duration of the interventions ranged from 8 to 48 weeks, with two to seven exercise sessions per week. A total of 20 trials had supervised sessions, two were non-supervised, and eight did not report on this. The 30 studies included in the present review tested 38 interventions and involved 1346 participants. The interventions included aerobic training (n = 23), resistance training (n = 8), aerobic + resistance training combined (n = 3), and others (n = 5). A more explicit description of each exercise intervention is available in Table 2. The studies used 12 OS parameters, including seven pro-oxidants and five antioxidants (Table 3). All studies measured OS in the blood, except for one [46] that measured OS via muscle biopsy.

Table 2 Studies that evaluated the effects of exercise training on oxidative stress parameters
Full size table
Table 3 Oxidative stress parameters used in the included trials
Full size table

3.3 Methodological Quality Assessment

Table 4 shows the list of included studies with quality scores. The agreement between the two reviewers was 85.3 %. Disagreement was solved in a consensus meeting. We defined five studies as high quality (score ≥6), 24 as moderate quality (3 ≤ score ≤ 5), and one trial as low quality (score <3). Questions 5 and 7 had the lowest score ratio, with only 19.9 % and 26.7 % of the trials scoring ‘yes’ on these questions, respectively.

Table 4 Methodological quality assessment scores of the included studies
Full size table

3.4 Meta-Analysis

Individual and summary effects of the studies included in the meta-analysis (n = 19) are presented in the forest plot (Fig. 2). In the pooled analysis, exercise seems to reduce pro-oxidant parameters (SMD –1.08; 95 % confidence interval [CI] –1.57 to –0.58; p < 0.001) and increase antioxidants (SMD 1.45; 95 % CI 0.83–2.06; p < 0.001). Some evidence of heterogeneity and inconsistency was found for both pro-oxidant (Q = 23.5; τ 2 = 1.09 ± 0.49; I 2 = 18.96 %) and antioxidant (Q = 19.4; τ 2 = 1.83 ± 0.76; I 2 = 2.11 %) analyses [42].

Fig. 2
figure 2

Standardized mean difference (Hedges’ g) and 95 % confidence interval of each study and summary effect (dotted lines). The square size of each intervention expresses its respective relative weight. CI confidence interval

Full size image

3.5 Risk of Bias

The funnel plot (Fig. 3) shows evidence of possible asymmetry. The pro-oxidant analysis involved 15 trials, 19 interventions, and 747 individuals, and it achieved an objective asymmetry significance (p = 0.80). On the other hand, the antioxidant analysis involved 14 trials, 19 interventions, and 699 individuals, and it failed to achieve an objective asymmetry (p = 0.21).

Fig. 3
figure 3

Funnel plot of each intervention’s effect and the respective standard error. Egger’s regression intercept: pro-oxidant –2.7 (95 % confidence interval [CI] –6.6–1.2); p = 0.08; antioxidant 2.4 (95 % CI –3.5–8.3); p = 0.21

Full size image

4 Discussion

4.1 Summary of Evidence

In general, the presented body of evidence so far seems to be robust and shows a reduction in pro-oxidant parameters and an increase in antioxidant parameters as a result of physical exercise. The meta-analysis results also corroborate these outcomes. However, only 39.9 % of the trials reached adequate statistical power (1−β ≥ 0.8), making this a methodological problem, given that small sample sizes may increase the chance of type 2 errors, i.e., no differences could be detected even when they may exist. Moreover, a small sample size may increase dispersion measures, such as the standard error. These prevalent elevated dispersion measures may have been responsible for the risk of bias assessed in the funnel plot analysis (Fig. 3). Additionally, the trials varied among themselves in the studied sample, in the volume, intensity, and type of exercise, and in the OS markers used.

4.2 Population

4.2.1 Children

Three of the selected studies investigated the effect of exercise on the OS parameters in children [47–49]; one was classified as high quality [49] and two as moderate quality [47, 48]. Onur et al. [48] followed 15 children with asthma and 15 controls who took part in 8 weeks of aerobic training; they reported an improvement (p < 0.05) in the redox state (decrease in pro-oxidant parameters and increase in antioxidant parameters). Kelly et al. [47] and Dennis et al. [49] investigated overweight/obese children who undertook aerobic training and intermittent running games, respectively, and reported unchanged pro-oxidant parameters.

In summary, whether physical training can reduce OS in children is inconclusive. Additionally, exercise training may not be effective for reducing OS in the young and relatively healthy, whose physiological and antioxidant functions are well-preserved. However, it is noteworthy that this population could benefit from other aspects of physical exercise, such as improvement in physical fitness, oxygen consumption, body composition, lipid profile, and fasting blood glucose [49]. Further studies should better describe the possible adverse events or sample losses during the trials and adjust the results for confounders. Although only one study showed improvement in redox state after a period of physical training, other studies showed no worsening. Thus, as a practical application, physical exercise programs for children can reduce OS and, in turn, delay the onset of age-related diseases as well as treat diseases common in this age group (i.e., asthma) that are associated with an imbalance in oxidant/antioxidant status.

4.2.2 Age

Among the six studies that addressed this topic, five were classified as of moderate quality [50–54]. One study compared two groups that differed only by age, preventing a more accurate analysis of the subject. Garcia-Lopez et al. [50] followed 23 healthy middle-aged men (mean age 53.9 ± 2.3 years) who undertook 21 weeks of two different training modes (resistance and aerobic) and reported no improvement in antioxidant markers. Conversely, Karabolut et al. [53] investigated healthy men with a mean age of 40.3 ± 6.2 years who underwent 12 weeks of aerobic training; reported pro-oxidant and antioxidant improvements (p < 0.05) compared with baseline values.

Similarly, Fatouros et al. [51] and Beltran Valls et al. [54] studied elderly individuals (mean age 71.5 ± 6.5 and 72 ± 1 years, respectively) who took part in aerobic and resistance training and reported improvements (p < 0.05) in both antioxidant and pro-oxidant parameters, respectively. Furthermore, Azizbeigi et al. [52] tested healthy young men (mean age 22.3 ± 2.3 years) who underwent 8 weeks of resistance training and also reported improvement (p < 0.05) in pro-oxidant and antioxidant markers.

These results provide strong evidence that physical training may reduce OS and improve health status, which can be considered key to preventing several chronic diseases [4]. Moreover, exercise training seems to counteract the deleterious effects of aging, not only by combating the major triggers of OS but also by exerting additional antioxidant actions [55]. However, more studies are warranted to compare these effects in aged and young adults. It is noteworthy that older adults generally present a higher level of OS and other deleterious health markers such as inflammation [56, 57], which could make them not only more vulnerable to chronic diseases but also more able to benefit from the health-enhancing effects of physical training [58, 59]. On the other hand, although young adults are generally healthier and possibly exhibit lesser magnitude improvements [60], their healthier status also allows them to achieve higher internal and external outputs, which could lead to an elevated physiological adaptation to physical training. The main concerns for future trials are to adjust the results for confounders and statistical power.

In practical terms, adults engaged in physical exercise programs can reduce OS and, in turn, delay the onset of diseases related to aging. Similarly, elderly individuals who practice regular exercise may show improvement in their redox state and can thereby not only reduce the deleterious effects of chronic diseases already conferred but also prevent the onset of other age-related diseases.

4.2.3 Obesity, Type 2 Diabetes, and Metabolic Syndrome

Only one study [61], classified as moderate quality, investigated the effects of physical training on OS parameters in the context of body composition. These authors followed ten normal-weight men (mean body fat 25.9 ± 3.5 %) and 19 overweight/obese men (mean body fat 33.0 ± 1.4 %) who undertook 24 weeks of resistance training and reported a decrease (p < 0.05) in pro-oxidant indicators in both groups compared with the respective baseline values and their controls. The comparison between experimental groups revealed no significant differences. Moreover, the data were reported only graphically, precluding us from calculating the effect size and quantifying the response of both groups to the same stimulus. Thus, evidence is insufficient to verify the role of body composition on OS as related to physical training.

T2DM is a metabolic disease associated with elevated OS. The pathophysiological processes of T2DM may disrupt natural antioxidant defenses and increase ROS production [62, 63]. One high-quality [64] and three moderate-quality [65–67] studies investigated the effects of physical training on OS parameters in T2DM. Two moderate-quality studies [68, 69] described their samples as adults with metabolic syndrome. de Oliveira et al. [67] studied 31 individuals with T2DM who underwent 12 weeks of three different types of physical training (aerobic, resistance, and combined training); they reported unchanged antioxidant indicators. Furthermore, aerobic and combined training were effective in promoting an increase in sulfhydryl compared with baseline values (p < 0.05), showing a possible decrease of the pro-oxidant state. On the other hand, Mitranun et al. [64] followed 28 individuals with T2DM who undertook 12 weeks of two different types of aerobic training (continuous vs. intermittent) and reported an improvement in redox state for the intermittent group. Similarly, Kurban et al. [65] tested 12 weeks of power walking in 30 individuals with T2DM and reported an increase in total antioxidant status compared with baseline values. In subjects with metabolic syndrome, Gomes et al. [68] and Rosety-Rodríguez et al. [69] also reported an improvement in pro- and antioxidant indicators, respectively, after 12 weeks of aerobic training.

In summary, moderate evidence suggests that individuals with T2DM exhibit increased antioxidant defenses and lowered OS after physical training, which could be reflected in decreased ROS production and less activation of many detrimental pathways, including hexosamine pathways, formation of advanced glycation end products (AGEs), and protein kinase C β1/2 (PKCβ1/2) [70]. Furthermore, β cells have lower antioxidant enzyme levels (superoxide dismutase, catalase, and glutathione) and higher sensitivity to OS. β-cell exposure to OS may lead to increased p21 cyclin-dependent kinase inhibitor production, decreased insulin messenger RNA (mRNA), and adenosine triphosphate (ATP) and calcium flux reductions in mitochondria and cytosol-causing apoptosis [71, 72]. Therefore, in practical terms, an attenuation of OS in individuals with T2DM may play a fundamental role in the treatment and prevention of this disease.

4.2.4 Cardiovascular Diseases and Heart Transplant Recipients

Five studies addressed the effects of exercise on OS parameters in patients with cardiovascular diseases and in heart transplant recipients. Four were classified as moderate quality [38, 46, 73, 74] and one as high quality [75]. Beck et al. [74] investigated the effects of 8 weeks of resistance and aerobic training on OS parameters in pre-hypertensive young adults (mean age 20.9 ± 0.7 years) and reported an improvement in total antioxidant capacity for both groups (aerobic and resistance), a decrease in F2-isoprostanes (8-iso-F2: pro-oxidant indicator) for the resistance training group, and a curious increase in 8-iso-F2 in the aerobic group. Braith et al. [73] investigated 12 weeks of aerobic training in nine heart transplant recipients. The pro-oxidant marker (8-iso-F2) did not change, but the individuals who undertook aerobic training had an attenuated decrease in the diameter of the brachial artery and maintained flow-mediated dilation compared with the controls. On the other hand, Linke et al. [46] tested 24 weeks of aerobic training in 12 men with chronic heart failure and reported a decrease in pro-oxidant status (lipid peroxidation [LPx]) and an increase in antioxidant indicators (catalase [CAT] and glutathione peroxidase [GPx]) compared with baseline values.

In addition, Edwards et al. [38] followed nine men with coronary artery disease who underwent 12 weeks of aerobic training; they reported an improvement in redox state. However, in the high-quality study conducted by Luk et al. [75], 32 adults with this same pathology underwent 8 weeks of combined aerobic + resistance training, but the authors observed no improvements in pro-oxidant or antioxidant status.

In combination, these results suggest that exercise training improves redox status. However, some results seem to be contradictory, probably because cardiovascular diseases are often preceded by different metabolic dysfunctions, such as dyslipidemia and chronic hyperglycemia, which may increase extracellular ROS production [76, 77] and overwhelm the antioxidant defensive system, leading to greater metabolic dysfunction [78]. Because cardiovascular diseases are usually multifactorial, a population with this disease may be very heterogeneous [79], which in turn may be a confounding factor that does not allow solid conclusions regarding exercise and OS to be drawn in this group. Further, we suggest that future clinical trials report any adverse events, even if none occurred, especially in these at-risk populations. In practice, although results are conflicting, individuals with cardiovascular disease and those who have received a heart transplant who are engaged in physical exercise programs appear to have reduced OS, which may therefore mitigate the deleterious effects of these diseases.

4.2.5 Spinal Cord Injury and Rheumatoid Arthritis

People with chronic spinal cord injury seem to have higher levels of OS and may be more likely to have metabolic and cardiovascular complications than the general population [80], possibly because of their low levels of physical activity [81]. A moderate-quality study from Ordonez et al. [82] investigated the effects of physical training in nine men versus eight controls with spinal cord injury. The training consisted of 12 weeks of arm-cranking exercise, 30 min at 65 % of heart rate reserve three times per week, and resulted in a decrease in pro-oxidant markers (thiobarbituric acid reactive substances [TBARS] and protein carbonyls [pCarb]) and an improvement in antioxidant status (trolox equivalent antioxidant capacity [TEAC] and GPx). Therefore, physical training seems to reduce OS in men with spinal cord injury. Nevertheless, this particular topic is scarcely explored in this population, and more evidence is needed to provide reliable conclusions.

Wadley et al. [83] conducted a moderate-quality study with 19 middle-aged adults with rheumatoid arthritis; after 12 weeks of aerobic training, pro-oxidant parameters (LPx and 3-nitrotyrosine [3-NT]) reduced, but antioxidant status (CAT and TEAC) did not change. Stiffness, swelling, and progressive destruction of the joints are the primary characteristics of rheumatoid arthritis [84] and may be induced by chronic systemic inflammation and exacerbated ROS production [85]. Therefore, physical exercise, besides inducing an improvement in joint function [86], may also play an important role in the pathogenesis of this disease by balancing the redox state. Nevertheless, the evidence is insufficient to determine the effects of physical training on OS markers in this population.

4.2.6 Severe Depression and Parkinson’s Disease

The human brain is considered highly sensitive to oxidative damage because it possesses high levels of phospholipids and polyunsaturated fatty acids, both of which are very susceptible to oxidants and have high oxygen consumption and low levels of antioxidant enzymes [87, 88]. Therefore, some neurological disorders, such as severe depression and Parkinson’s disease, may be triggered by a constant exposure to OS [87, 89]. A thorough review by Camiletti-Moirón et al. [36] suggests that physical exercise may be a powerful tool to treat such diseases.

In this regard, Schuch et al. [90] conducted a high-quality study with severely depressed men: 15 patients and 11 controls underwent 12 weeks of aerobic training. The results showed no time effect in the within-group comparison, but a significant difference (p < 0.05) between groups (experimental vs. control) in pro-oxidant status (TBARS). This could indicate that, if not a reduction, at least an attenuation of OS may play a fundamental role in the pathogenesis of this disease [89]. However, the evidence is insufficient to determine the effects of physical training on OS markers in severely depressed men. Further, a moderate-quality study [91] investigated the effects of 8 weeks of resistance training on pro-oxidant and antioxidant indicators in 16 middle-aged adults with Parkinson’s disease. The authors reported no effect of exercise training (p > 0.05) for either the within-group or the between-group comparisons. Therefore, although the studies are of high quality, the data are too sparse to make any qualified conclusion about the effect of exercise training on OS in individuals with Parkinson’s disease. In practical terms, although literature are scarce, it does seem that physical training reduces OS in these health conditions, suggesting that regular physical exercise can prevent progression of these diseases.

4.3 Exercise Training

4.3.1 Aerobic Training

Aerobic exercise is the most frequently used type of training in research protocols. Studies included in this review investigated a variety of stimuli (e.g., continuous or intermittent), volume, intensity, and training equipment (e.g., treadmill, cycle-ergometer). A more detailed discussion on the particulars of each of these factors on the outcomes of aerobic training is beyond the scope of this review.

We found 19 studies (21 interventions) that used aerobic training, three of which were classed as high quality [64, 90, 92], and 16 were of moderate quality [38, 46–48, 50, 51, 53, 65–69, 73, 74, 82, 83]. The high-quality studies were consistent regarding the effects of aerobic training on OS parameters. One study [90] reported an unchanged (p > 0.05) redox status after intervention, and the other two [64, 92] reported an improvement (p < 0.05) in pro-oxidant and/or antioxidant parameters.

Taken together, strong evidence supports a positive effect of aerobic training on redox balance. The training-induced adaptations of OS increase the efficiency of the enzymatic and non-enzymatic antioxidant defense systems, leading to a greater mitochondrial capacity to scavenge free radicals [93] and a reduction in ROS production by the mitochondrial membrane potential at basal conditions [37, 94]. Therefore, aerobic training seems to be a suitable intervention to improve redox balance. Future trials should ensure statistical power is sufficient (1−β ≥ 0.8). In practice, the available literature, almost in its entirety, indicates that aerobic training can be a powerful strategy to reduce OS and therefore delay the onset of non-communicable chronic diseases.

4.3.2 Resistance Training

Resistance training is the exercise mode with the largest annual increase in prevalence among experimental studies on health and performance in exercise physiology-related publications in the last decade. Thus, resistance training has been postulated as an important tool to control metabolic and cardiovascular variables, and it has been tested and used for the treatment and prevention of several diseases, such as T2DM and hypertension [95]. Therefore, it is reasonable to infer that resistance training may promote beneficial exercise-induced adaptations in physiological systems that play a role in the pathogenesis of these diseases, such as chronic inflammation and OS [11, 12].

In that regard, six studies were classified as of moderate quality [50, 52, 61, 67, 74, 91] and showed the effects of resistance training on OS parameters. Three studies [50, 67, 91] indicated an improvement in the redox balance, and three [52, 61, 74] reported no differences after resistance training interventions. Unfortunately, the complexity of this training method and the heterogeneity of protocols may produce conflicting results. While researchers using aerobic training sessions usually take into account the total volume (e.g., running distance) and intensity (e.g., % maximal oxygen uptake [VO2], % heart rate reserve, or % of anaerobic threshold intensity), those using resistance training must consider the number of sets and repetitions until exhaustion, the total amount of exercise, and the recovery intervals. Therefore, although the results are not unanimous in showing improvements in redox balance, at least an attenuation was observed. More homogeneous protocols are needed to reach a consistent conclusion, as previously stated [37].

4.3.3 Combined Aerobic and Resistance Training

Combined aerobic and resistance training may be the most comprehensive approach, promoting broad adaptations, such as increased cardiopulmonary fitness and strength, respectively [96]. Nevertheless, few studies have investigated its effects on OS parameters. Only two included studies tested this method, one of moderate quality [67] and one of high quality [75]; both reported that redox status after training was no different from baseline values or that of the control group. Therefore, the findings are sparse, and more studies are needed to confer valid conclusions.

4.3.4 Tai Chi and Yoga

Yoga and tai chi have become increasingly popular in Western cultures as a method of exercise training for several populations [97, 98] and may be effective in health maintenance. Tai chi or yoga sessions are typically performed at around 40 % of VO2, which is below the recommended intensity to promote cardiovascular adaptations [99], but they may be effective for improving balance, lower-body flexibility, and muscular strength [97, 98]. Their effects on OS are poorly investigated and not adequately substantiated, since it has been suggested that chronic adaptations in redox balance are dependent on the intensity and type of exercise utilized. Therefore, exercise performed at very low intensities may not cause enough physiological stress to generate adaptation [37].

Despite this, moderate-quality studies [66, 100] and a low-quality study by Rosado-Pérez et al. [101] investigated the effects of tai chi [100, 101] and yoga [66] on OS parameters, and all three reported an improvement both in pro-oxidant and antioxidant indicators. These results may be partially explained by an additional adaptation besides the physiological effect by itself, since these exercise modes may help with adherence to training and better emotional control, thus helping practitioners to become more resilient and improve their overall health [102, 103], such as through a reduction in OS. Thus, yoga and tai chi are suggested as an alternative to reduce OS. Future studies should better describe the intervention and more rigorously control the intervention’s intensity to improve reproducibility and help experts infer the possible physiological mechanisms involved in the health-enhancing effects of tai chi and yoga. To the best of our knowledge, only three studies have investigated the effects of tai chi on OS indicators [66, 100, 101]. However, they all highlight that their practice seems to improve redox state. It is noteworthy that this modality is commonly practiced by older people who generally have a higher OS [57, 58] than young individuals. Thus, in practical terms, it is suggested that older people who wish to improve or preserve their redox state and therefore treat or delay the onset of chronic degenerative diseases use tai chi for both.

4.4 Perspectives and Limitations

The heterogeneity of training protocols, such as different volumes and intensities, precluded us from definitively identifying the most appropriate and effective exercise mode, duration, and intensity to reduce OS. Furthermore, the studied populations were not homogeneous. Regardless, a general analysis of the quantitative and qualitative results seems to point in the same direction, i.e., that redox state may be balanced by exercise training.

The wide range of pro-oxidant and antioxidant markers used in the studies may also influence the results. An attenuation of OS may indicate a reduction of lipid peroxidation, protein oxidation, and/or free radical production and an increase in antioxidant capacity, which may be caused by an increased amount and/or activity of several proteins, enzymatic or non-enzymatic, that operate in different steps of ROS neutralization [104]. Moreover, substances that have no direct antioxidant mechanism, such as repair enzymes, may also play an important role in OS after oxidative damage. Thus, any identified changes in the redox balance would be quite varied, and a detailed and specific discussion on this subject goes far beyond our scope, which may also be a limitation of this review. However, regardless of the OS markers under consideration, their response to exercise training seems to be similar. Future research should evaluate the minimum intervention required to promote benefits in redox state.

5 Conclusion

The effects of exercise training on OS parameters were analyzed in 30 trials that met our inclusion/exclusion criteria. The studies involved 38 different exercise interventions and 1346 participants. Most studies were of moderate quality (score ≥3 and ≤5) and not reach an adequate statistical power for the analysis (80 %). Given the wide heterogeneity of protocols, we are unable to indicate a specific intensity, volume, or exercise mode that is more effective in decreasing OS. Furthermore, the trials involved a wide variety of populations (e.g., children, obese and healthy adults, and individuals with T2DM, Parkinson’s disease, severe depression, and other diseases caused by immune or metabolic disorders), which in turn precluded us from making any specific conclusion. Conversely, the pooled analysis revealed that, regardless of intensity, volume, type of exercise, and studied population, the antioxidant indicators tended to increase and pro-oxidant indicators tended to decrease after training. In other words, exercise training seems to induce an antioxidant effect. Therefore, we recommend that people should perform some kind of exercise to balance the redox state, regardless of their health status, to improve health-related outcomes.