https://orcid.org/0000-0003-0683-3989
Figures
Abstract
Background
Physical activity (PA) is associated with various health benefits, especially in improving chronic health conditions. However, the metabolic changes in host metabolism in response to PA remain unclear, especially in racially/ethnically diverse populations.
Objective
This study is to assess the metabolic profiles associated with the frequency of PA in White and African American (AA) men.
Methods
Using the untargeted metabolomics data collected from 698 White and AA participants (mean age: 38.0±8.0, age range: 20–50) from the Louisiana Osteoporosis Study (LOS), we conducted linear regression models to examine metabolites that are associated with PA levels (assessed by self-reported regular exercise frequency levels: 0, 1–2, and ≥3 times per week) in White and AA men, respectively, as well as in the pooled sample. Covariates considered for statistical adjustments included race (only for the pooled sample), age, BMI, waist circumstance, smoking status, and alcohol drinking.
Results
Of the 1133 untargeted compounds, we identified 7 metabolites associated with PA levels in the pooled sample after covariate adjustment with a false discovery rate of 0.15. Specifically, compared to participants who did not exercise, those who exercised at a frequency ≥3 times/week showed higher abundances in uracil, orotate, 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1) (GPE), threonate, and glycerate, but lower abundances in salicyluric glucuronide and adenine in the pooled sample. However, in Whites, salicyluric glucuronide and orotate were not significant. Adenine, GPE, and threonate were not significant in AAs. In addition, the seven metabolites were not significantly different between participants who exercised ≥3 times/week and 1–2 times/week, nor significantly different between participants with 1–2 times/week and 0/week in the pooled sample and respective White and AA groups.
Conclusions
Metabolite responses to PA are dose sensitive and may differ between White and AA populations. The identified metabolites may help advance our knowledge of guiding precision PA interventions. Studies with rigorous study designs are warranted to elucidate the relationship between PA and metabolites.
Citation: Du Y, Li Y-Y, Choi BY, Fernadez R, Su K-J, Sharma K, et al. (2023) Metabolomic profiles associated with physical activity in White and African American adult men. PLoS ONE 18(11): e0289077. https://doi.org/10.1371/journal.pone.0289077
Editor: Trenton Honda, Northeastern University, UNITED STATES
Received: December 9, 2022; Accepted: July 11, 2023; Published: November 9, 2023
Copyright: © 2023 Du et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data can be accessed by contacting the data access committee of the study at aliu10@tulane.edu. There is NO ethnical or legal restrictions on sharing a de-identified data set as imposed by the Institutional Review Board.
Funding: This study was supported by funds from Edward G. Schlieder Endowment and startup funds of Tulane University, as well as grants from the National Institutes of Health (grant numbers R01AR069055, U19AG055373) awarded to HD. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: PA, physical activity; AA, African American; LOS, Louisiana Osteoporosis Study; BMI, body mass index; GPE, 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18; IRB, Institutional Review Board; QC, qualitative control; MS, mass spectroscopy; HESI-II, heated electrospray ionization; LIMS, laboratory information management system; AMP, adenosine monophosphate; ADP, adenosine diphosphate; ATP, adenosine triphosphate; DASH, dietary approaches to stop hypertension
Introduction
Many studies have documented the benefits of physical activity (PA) in preventing and managing various chronic health conditions and reducing mortality [1]. These conditions include but are not limited to cardiovascular diseases [2], type 2 diabetes [1], metabolomic syndrome (particularly obesity and insulin resistance) [3], cognitive function [4], musculoskeletal health [5], and mental health [6]. However, the underlying mechanisms of the health benefits of PA are largely unknown. Emerging studies have examined the pathophysiological pathways of PA on health outcomes through metabolomics approaches [7].
Metabolomics is a novel approach to systematically profile metabolites in biofluids, cells, and tissues at the molecular level [8]. Metabolites represent the downstream expression of the genome, transcriptome, and proteome and the final endpoints of the interaction of the genes, environment and lifestyles [9]. Studying metabolomic signaling is one of the most powerful approaches to revealing inherent omics variation closest to the disease risk/phenotype [8]; it is promising and useful to elucidate the mechanisms that link lifestyle behaviors and guide precision lifestyle intervention, including PA [10–12].
Most previous studies examining PA and metabolomics assessed the acute influence of exercise on metabolites. After reviewing 27 studies, Schranner and colleagues reported that the abundance of 196 metabolites changed significantly within 24 hours after endurance or resistance exercise in at least two experiments [13]. Similarly, another review study concluded that most studies focused on metabolite changes after the bout (s) of exercise, and cross-sectional studies are also needed to understand the human system’s response to chronic PA workloads [14]. In addition, available studies that examined the effect of chronic PA on metabolites used samples of adults from all age groups, such as those aged 40 and older with a majority being 65 and older [15], and those aged 25 and over with mean age of around 55 years old [16]. Various factors may complicate assessing the associations of PA and metabolites in a population using a wide age range, such as the health conditions cumulated throughout the aging process. Furthermore, increasing numbers of studies reported the possibility of molecular differences (e.g., metabolites) between women and men, and gender differences might need consideration when developing personalized and precision treatment and intervention [17–20]. Assessing metabolites responses to PA in different genders is key to more precisely elucidating the relationship between metabolites and PA. Last but not least, African Americans (AAs) are disproportionally affected by various chronic diseases compared to other race/ethnicity groups [21]. For instance, previous studies reported that metabolites associated with type 2 diabetes [22,23], and the risk of coronary heart disease [23] differ between different racial/ethnical groups. The metabolomic differences may reflect the underlying variations of genetic, environmental, and behavioral factors (e.g., physical activity) [22,24,25]. However, PA-related metabolomic profiles in AAs are rarely studied [26]. Therefore, this study aimed to explore metabolites associated with PA in White and AA men aged between 20–50 years old.
Materials and methods
Study population
The current study included 698 White and African American participants with available metabolomic data from the ongoing Louisiana Osteoporosis Study (LOS) (from 2011 to current). LOS is a cross-sectional study that aims to build a large sample pool and database to investigate the genetic and environmental factors for complex chronic health conditions (e.g., osteoporosis, obesity, and sarcopenia) in Southern Louisiana. The inclusion and exclusion criteria of LOS have been previously described [27,28]. Each participant signed a consent form before any data collection. All participants included in this study were men between 20 and 50 years old. The Tulane University Institutional Review Board approved this study, and the principles of the Declaration of Helsinki were followed [29].
Measurements
Physical activity.
We measured participation in leisure-time PA with the self-reported number of times the participants exercised weekly. Several scientific articles used these PA measures [30,31]. According to the Physical Activity Guidelines for Americans 2nd edition [32], participating in leisure-time PA at least 3 days per week is associated with significant health benefits, and participating in any amount of PA (e.g., 1 to 2 times a week) yields more health benefits compared to no PA [33,34]. Therefore, in the current study, we classified levels of PA as 0, 1–2, and ≥3 times per week. PA is any body movement that increases energy expenditure; exercise is a subcategory of PA referred to as planned, structured, and repetitive body movements that focus on improving or maintaining physical fitness [35]. In this current study, we only assessed PA by self-reported exercise frequency per week. To be noted, both PA and exercise will be used throughout as this study assessed PA levels by self-reported exercise.
Demographics and other covariates.
A questionnaire survey was used to collect age, gender, race/ethnicity, and other lifestyle behaviors. Self-reported race/ethnicity was identified by choosing from the following options: African American/Black, Asian, Caucasian/White, Hispanic/Latino, Native American/Pacific Islander, and others. Questions with yes/no responses assessed other behavioral factors regarding participant smoking status and alcohol consumption. Weight was measured in light indoor clothing using a calibrated balance beam scale, and height was measured using a calibrated stadiometer without shoes. We calculated body mass index (BMI) as weight (kg) divided by height squared (m2).
Metabolomic analyses.
We collected blood sample from each participant, centrifuged the blood sample at 1000G for 5 minutes, aliquoted the serum into 2-mLmicrotubes, and stored the serum at -80°C before further processing. Metabolon (Durham, NC) analyzed the serum samples of the current study. Samples were prepared using the automated MicroLab STAR® system from Hamilton Company. Specifically, to remove protein, dissociate small molecules bound to protein or trapped in the precipitated protein matrix, and recover chemically diverse metabolites, proteins were precipitated with methanol under vigorous shaking for 2 min (Glen Mills GenoGrinder 2000) followed by centrifugation. The resulting extract was divided into five fractions for different types of analyses and backups. Samples were placed briefly on a TurboVap® (Zymark) to remove the organic solvent. The sample extracts were stored overnight under nitrogen before preparation for analysis.
Several types of controls were analyzed in concert with the experimental samples, which included: a pooled matrix sample generated by taking a small volume of each experimental sample to serve as a technical replicate throughout the data set; extracted water samples to serve as process blanks; and a cocktail of qualitative control (QC) standards carefully chosen to not interfere with the measurement of endogenous compounds were spiked into every analyzed sample, and allowed for instrument performance monitoring and aided chromatographic alignment.
All methods utilized a Waters ACQUITY ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-Exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35,000 mass resolution. All the samples were analyzed in both positive and negative ion modes with heated electrospray ionization. The mass spectroscopy (MS) analysis alternated between MS and data-dependent MSn scans using dynamic exclusion. The scan range varied slightly between methods but covered 70–1000 m/z.
Bioinformatics.
The informatics system consisted of 4 major components, the Laboratory Information Management System (LIMS), the data extraction and peak-identification software, data processing tools for QC and compound identification, and a collection of information interpretation and visualization tools for use by data analysts. The informatics component hardware and software foundations were the LAN backbone, and a database server running Oracle 10.2.0.1 Enterprise Edition.
Metabolon’s hardware and software extracted, peak-identified, and QC processed the raw data. Three criteria formed the basis of the biochemical identifications: retention index within a narrow retention time/index window of the proposed identification, accurate mass match to the library +/- 10 ppm, and the MS/MS forward and reverse scores between the experimental data and authentic standards. A comparison of the ions presenting in the experimental spectrum to the ions presenting in the library spectrum determined the MS/MS scores. While there may be similarities between these molecules based on one of these factors, using all 3 data points can distinguish and differentiate biochemicals.
The design of the QC and curation processes ensured accurate and consistent identification of true chemical entities and removal of those representing system artifacts, misassignments, and background noise. Peaks were quantified using area-under-the-curve. Data normalization step was performed to correct variation resulting from instrument inter-day tuning differences. Essentially, each compound was corrected in run-day blocks by registering the medians to equal one (1.00) and normalizing each data point proportionately. We identified a total of 1,378 biochemicals, with 1,133 compounds of known identities (named biochemicals) and 245 compounds of unknown structural identities (unnamed biochemicals).
Statistical analysis
Kruskal-Wallis rank sum test was used to describe continuous sample characteristic variables; Pearson’s Chi-squared test was used to assess categorical sample characteristic variables. To identify metabolites significantly associated with PA, linear regression models were fitted using individual metabolites and PA levels as response and exposure variables, respectively,. We increased the number of covariates included in the regression models for statistical adjustments. Two dummy variables were used to model PA; each dummy variable compared PA performed 1–2 or ≥3 times a week with 0 times per week. Metabolite abundances were log-transformed for regression analysis to reduce the impact of extreme values. To assess the association between each metabolite and PA levels, we fitted three regression models for each metabolite: Model 1 included PA levels only; Model 2 adjusted for race and age, which were non-modifiable factors; and Model 3 additionally adjusted for modifiable factors, including BMI, waist circumference, smoking, and alcohol use. We tested whether the exercise dummy variables’ regression coefficients were equal to 0, which implies no association between PA levels and individual metabolites. Based on Benjamini and Hochberg’s method to correct for multiple testing [36], we converted raw p-values of Wald statistics were converted to false discovery rates (FDR or q-values), and the q-values/FDR less than 0.15 were considered significant. All model fittings had no multicollinearity issues based on the variance inflation factors. For the 7 metabolites selected by Model 3, we conducted stratified analysis by race (African American and White) to investigate whether the associations between PA and those metabolites could change by different race groups. More specifically, we fitted Model 3 (without the race/ethnicity variable) to African Americans and Whites separately to examine race-specific associations.
Pathway analysis
To identify pathways represented by the 7 significant metabolites, we conducted a pathway analysis using the web tool MetaboAnalyst 5.0 [37]. The pathway analysis integrated well-established (e.g., over-representation analysis) methods and novel algorithms and concepts (e.g., network topology analysis). The topology analysis accounts for the position of a metabolite in the pathway structure and provides an impact value ranging from 0 (minimum impact) to 1 (maximum impact) for each tested metabolite.
Results
Table 1 shows the characteristics of the study population. The mean age of the sample was 36.9±8.3, and over 44.5% (297) were African American. Approximately 61.4% of the sample reported exercising ≥3 times/week. Those who reported 0 regular exercise engagement were relatively older (mean age = 38.1±8.2) compared to those who exercised 1–2 times/week (mean age = 36.4 ±8.0) and ≥3 times/week (mean age = 36.0 ±8.2). A higher proportion of Whites (64%) reported exercising ≥3 times/week compared to African Americans (36%). The mean BMI (28.0 ±5.8) and waist circumference (86.0±14.0) were the lowest in those who exercised ≥3 times/week. Over 67% of those who exercised ≥3 times/week reported alcohol consumption compared to 54% of those who reported no regular exercise. We further assessed the sample characteristics by race/ethnicity groups and physical activity levels (Table 2). There was a trend that those who reported exercising ≥3 times/week were younger (mean age = 38.0 ±8.0), had the lowest BMI (27.0 ±5.2), and waist circumference (85.4±14.0), but the differences were not significant in AAs (p>0.05); On the other hand, it is all significant in Whites with p = 0.005 for age difference, p = 0.002 for BMI, and p<0.001 for waist circumference.
A total of 365 metabolites were related to PA levels (Table 3) before adjusting for any covariates. Of the 365 metabolites, we identified 105 metabolites with significant associations with PA levels (Table 3) after controlling for non-modifiable factors (race and age). After adjusting for both non-modifiable factors (race and age) and modifiable factors (BMI, waist circumference, smoking status, and alcohol drinking), 7 metabolites remained significant (salicyluric glucuronide, uracil, orotate, adenine, GPE, threonate, and glycerate) (Fig 1). The 7 metabolites associated with PA included 1 carbohydrate, 1 cofactor and vitamin, 3 nucleotides, 1 xenobiotic, and 1 lipid (Table 3).
From left to right, pairwise groups comparisons are displayed for 1–2 versus 0 (reference), ≥3 versus 0 (reference), and ≥3 versus 1–2 (reference). Point estimates and 95% confidence intervals are presented for the mean differences in individual metabolite concentrations between the two exercise groups. The value 0 indicates no mean difference between the groups.
Interestingly, the abundances of all 7 metabolites differed between those who reported no exercise and those who exercised ≥3 times/week in the pooled sample (Table 4 and Fig 1). Specifically, those who exercised ≥3 times/week had higher uracil, orotate, GPE, threonate, and glycerate compared to those who reported no exercise. Those who exercised ≥3 times/week had lower abundances of salicyluric glucuronide and adenine than those who reported no exercise. For the respective White and AA groups who reported exercising ≥3 times/week and those who reported no regular exercise, salicyluric glucuronide and orotate were significant only in AAs; adenine, 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1) (GPE) and threonate were significant in Whites; uracil and glycerate were significant in both AAs and Whites (Fig 2). Although the trends of metabolite alterations with PA levels were not significant in some comparison groups in Whites or AAs, the trends moved in the same direction among all examined comparison groups (Fig 2).
Up arrow indicates higher exercise frequency with higher metabolite concentration, and lower arrow indicates higher exercise frequency with lower metabolite concentration. GPE = 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (P-16:0/18:1).
We found that for either pooled outcomes or separated outcomes for Whites and AAs (Table 4 and Fig 1), there were no significant differences in the abundances of the 7 metabolites between those who reported no exercise and those who exercised 1–2 times/week, as well as no differences between those exercised 1–2 times/week and exercised ≥3 times/week. In the pooled sample, however, we found significant differences in the salicyluric glucuronide between those who reported no exercise and those exercised 1–2 times/week and in the orotate between those exercised 1–2 times/week and ≥3 times/week in the pooled sample.
In the pathway analyses, 4 of the 7 PA-related metabolites (orotate, uracil, glycerate, and adenine) were annotated in the 8 metabolic pathways, and only 2 metabolic pathways (pyrimidine and glycerolipid metabolism) were significant with p<0.05 (Table 5). Of the 8 pathways, 3 included uracil, and 4 included glycerate. The impact values of the metabolites on these pathways varied (from almost 0 to 0.12).
Discussion
In the pooled sample, compared to those who did not exercise regularly, those who exercised ≥3 times/week had significantly higher abundances of uracil, orotate, GPE, threonate, and glycerate and significantly lower abundances of salicyluric glucuronide and adenine; however, the differences were not significant between the other PA comparison groups (0 vs. 1–2 times/week and 1–2 vs. ≥3 times/week). In addition, uracil and glycerate differed in both White and AA groups, but the other 5 metabolites were significantly different in either Whites or AAs. The findings suggest that PA-related metabolite changes might be dose sensitive and vary among racial/ethnic groups. Longitudinal studies in diverse race/ethnicity populations are needed to confirm the findings. In addition, race/ethnicity is not only a biological trait, but also a sociological trait [38,39]; a big data approach to comprehensively, dynamically, and relatively precisely collect all relevant data could be helpful in further advancing our knowledge and understanding of the human metabolomic profile and its responses to modifiable factors, such as lifestyle behaviors.
In the pooled and the respective racial/ethnical samples, those who exercised ≥3 times/week had significantly higher uracil and glycerate than those who did not exercise regularly. Uracil is in the pyrimidine metabolism pathway and belongs to the class of organic compounds known as pyrimidines. Uracil participates in several enzymatic reactions and evidence indicates a relationship to various health conditions. For example, a narrative review reported that many studies demonstrated that pyrimidine metabolism dysfunction is closely associated with cancer progression. Precision cancer medicine arose from several cancer treatments targeting pyrimidine metabolism [40]. However, there are no previous reports about the relationship between uracil and PA. The development of precision lifestyle interventions requires studies that assess whether PA could alter uracil [40]. Glycerate is a hydroxy monocarboxylic acid anion derived from propionate and is the conjugate base of glyceric acid. A deficiency of human glycerate kinase leads to D-glycerate acidemia/D-glyceric aciduria. Our findings are consistent with a previous study that reported moderate-to-vigorous-intensity leisure-time PA is positively related to glycerate in both AAs and Whites in the respective and pooled samples [41]. Our findings suggest that PA may increase glycerate concentration, and the influence was dependent on regular PA frequency. In addition, glycerate positively correlates with healthy food intake [42] and may confound the relationship between PA and metabolomic abundance. Therefore, a rigorous study design is needed to confirm the effects of PA on metabolite abundances.
The pooled sample analysis revealed that AA but not White participants who exercised ≥3 times/week had a lower abundance of salicyluric glucuronide and a higher abundance of orotate than participants who reported no exercise. Salicyluric glucuronide is a metabolite of salicylic acid and aspirin and does not have a previously reported association with exercise. Orotate, or orotic acid, is classified as a pyrimidine monocarboxylic acid and is an uracil molecule that bears a carboxy substituent at position C-6. Previous studies reported that orotate might improve PA tolerance in trained athletes or patients with coronary artery disease [43,44].
In addition, compared to those who reported no exercise, we found higher abundances of threonate and GPE and a lower concentration of adenine in those who exercised ≥3 times/week in Whites but not in AAs. Adenine combines with the sugar ribose to form adenosine, which can bond with one to three phosphoric acid units, and yield adenosine monophosphate (AMP), adenosine diphosphate (ADP), and adenosine triphosphate (ATP). Our findings are consistent with animal studies. For example, an animal model study found that swimming exercises in rats with adenine-induced chronic kidney disease may result in favorable blood pressure compared to their control counterparts [45], suggesting that exercise may counteract adenine. In addition, our findings could be consistent with those reported in another human study. Zarebska and colleagues [46] tested the effect of exercise on adenine nucleotide levels in 20 endurance-trained adults. Blood plasma was collected at rest, during submaximal and maximal cycle ergometer exercise, and after early recovery. The study reported that plasma ATP and ADP concentrations increased up to 2.5-fold during maximal exercise. Even though not reported, an ATP and ADP increase could potentially result in an adenine decrease, one of the components forming ATP and ADP. Further study is needed to explore PA interventions on adenine concentration and how the changes may relate to health outcomes.
Regarding threonate, the current study findings are consistent with another study reporting that threonate was positively related to PA energy expenditure which was calculated by PA levels objectively measured by ActiGraph in an Asian population [47]. A positive relationship between threonate and a healthy diet may exist. For example, in AAs and Whites, the Healthy Eating Index and Dietary Approaches to Stop Hypertension (DASH) emphasize higher healthy food consumption (e.g., whole grains, fruits, and vegetables) and lower consumption of processed food reportedly have a positive relationship with threonate abundance [48]. Future studies may consider PA and dietary intake when assessing changes in threonate concentration, especially when examining in different populations. GPE is a lipid metabolite of the plasmalogen pathway. Plasmalogens act as an endogenous antioxidant produced by peroxisome, and plasmalogen production is a compensatory mechanism that protects the organism against higher oxidative stress [42]. It is reported that significantly higher GPE was found in athletes who performed high cardiovascular demand (CD; increased stroke volume and blood pressure) sports compared to low-moderate CD athletes, which may potentially place the high CD group at higher risk of cardiovascular disease [49]. The study concluded that the application of metabolites may help avoid potential disorders associated with excessive training. However, the study did not quantify the levels of GPE, and future studies could further assess the harmful threshold of GPE to avoid excessive training or physical activity. Diet studies also assessed GPE. For instance, one study found that the DASH diet is significantly negatively related to GPE compared to either a usual diet (macronutrient intake similar to average US consumption) or a fruit/vegetable diet in adults [50]; consistently, another study reported GPE was positively associated with dietary intake of red and processed meat and CKD progression in children with chronic kidney disease [51]. Further exploration of how multiple lifestyle behaviors, such as PA and diet, alters GPE concentration is warranted.
Notably, we only found significant alterations between those who reported no regular exercise and those who exercised ≥3 times/week. There was a trend of alterations in abundance, but not significant, between those without regular exercise and exercised 1–2 times/week and between those who exercised 1–2 times/week and exercised ≥3 times/week. The findings may suggest that metabolite responses to PA are dose sensitive. These findings are particularly important as they may help identify precision PA interventions when we further understand the pathophysiological mechanism of each metabolite, such as function, normal range, and interaction with other metabolites.
In addition, the abundance of some metabolites changed in African Americans but not in Whites, and vice versa. Some possible reasons might be the population differences in genes [52], and environmental and other lifestyle factors (e.g., diet). For instance, in the current study, we found that BMI was different across PA levels in Whites, but not in AAs; this could be partially attributed to potentially different dietary patterns, which was not assessed in this study. Particularly, racial/ethnic groups not only differ in biological traits, but also in social, economic and cultural backgrounds [38,39]. Historical and contemporary disparities in access to resources [53], living environment [54], discrimination [55], and stress [56] disproportionately affect AA communities in the US. These social economic and cultural factors may contribute to observed differences in metabolomic profiles of AAs and Whites [23]. These factors may also influence lifestyles such as the dietary patterns of the study participants [57,58]. Although not assessed in the current study, these factors (e.g., exposure to discrimination or chronic stress) could potentially confound or modify the association between physical activity and metabolomic profiles. Furthermore, the metabolomic profile reflects the interaction of genes, physical and social environments, and lifestyles [9]; it also reflects one’s health/disease status during the life course journey [59]. Therefore, a big data approach to comprehensively, dynamically, and relatively precisely collect all relevant data is warranted to advance our knowledge and understanding of the human metabolomic profile and its responses to modifiable factors, such as lifestyles, and physical and social environment. Such an approach may provide profound implications for developing precision interventions to decrease health disparities.
Limitations
The study is one of the first to assess the relationships between PA levels and metabolites in a racially/ethnically diverse population with an untargeted approach. Interpretation of the study findings requires consideration of the following limitations. First, PA was assessed by self-reported regular exercise frequency which might be subject to recall bias. Second, we only assessed regular weekly exercise frequency and did not consider the exercise type and intensity in the current study. However, evidence indicates that weekly exercise frequency is essential for health benefits [60,61]. Third, although we included various covariates in the models, other factors (e.g., dietary intake, sedentary behavior, diseases), which might be related to metabolite alternations were not included [7]. Fourth, given the study’s cross-sectional nature, causality cannot be inferred. Fifth, we only included White and AA men from one geographic area; the generalizability of the findings to women and other populations needs further investigation. Since there might be gender differences in metabolites between women and men, we will assess PA and metabolites in future studies should data be available. Sixth, looking at each racial/ethnical group might have reduced the power to detect some effects in a subgroup. However, the trends of metabolite alterations with exercise levels were in the same direction among all examined groups. Therefore, studies with larger sample sizes and rigorous study designs are needed to investigate whether metabolite responses to PA differ in other populations.
Conclusion
Compared to those who reported no regular exercise engagement, those who exercised ≥3 times/week had higher uracil, orotate, GPE, threonate, and glycerate metabolite abundances and a low abundance of salicyluric glucuronide and adenine. The findings are primarily consistent with previous reports in other populations. However, in either pooled outcomes or separated outcomes in respect of Whites and AAs, there were no significant differences in the abundances of the seven metabolites between those who reported no exercise and those who exercised 1–2 times/week, as well as between those who exercised 1–2 times/week and exercised ≥3 times/week. These findings suggest that metabolomic responses to exercise might be dose sensitive, which may advance our knowledge of guiding precision PA interventions. In addition, other factors such as dietary intake have been reported to significantly influence some metabolites (e.g., threonate, glycerate, GPE) associated with PA levels [48]; AAs and Whites shared similar alterations in some metabolites but not all. Therefore, studies with rigorous study designs are needed to elucidate the relationship between PA and metabolite responses for precision PA intervention development. Furthermore, a big data approach to comprehensively, dynamically, and relatively precisely collect all relevant data could further advance our knowledge and understanding of the human metabolomic profile and its responses to modifiable factors to reduce health disparities.
References
- 1. Warburton DER, Bredin SSD. (2017) Health benefits of physical activity: a systematic review of current systematic reviews. Current opinion in cardiology.32(5):541–56. pmid:28708630.
- 2. Pang Y, Kartsonaki C, Du H, Millwood IY, Guo Y, Chen Y, et al. (2019) Physical Activity, Sedentary Leisure Time, Circulating Metabolic Markers, and Risk of Major Vascular Diseases. Circulation Genomic and precision medicine.12(9):386–96. pmid:31461308.
- 3. Myers J, Kokkinos P, Nyelin E. (2019) Physical Activity, Cardiorespiratory Fitness, and the Metabolic Syndrome. Nutrients.11(7). pmid:31331009.
- 4. Erickson KI, Hillman C, Stillman CM, Ballard RM, Bloodgood B, Conroy DE, et al. (2019) Physical Activity, Cognition, and Brain Outcomes: A Review of the 2018 Physical Activity Guidelines. Medicine and science in sports and exercise.51(6):1242–51. pmid:31095081.
- 5. Chughtai M, Gwam CU, Mohamed N, Khlopas A, Sodhi N, Sultan AA, et al. (2017) Impact of Physical Activity and Body Mass Index in Cardiovascular and Musculoskeletal Health: A Review. Surgical technology international.31:213–20. pmid:29327778.
- 6. White RL, Babic MJ, Parker PD, Lubans DR, Astell-Burt T, Lonsdale C. (2017) Domain-Specific Physical Activity and Mental Health: A Meta-analysis. Am J Prev Med.52(5):653–66. pmid:28153647.
- 7. Kelly RS, Kelly MP, Kelly P. (2020) Metabolomics, physical activity, exercise and health: A review of the current evidence. Biochimica et biophysica acta Molecular basis of disease.1866(12):165936. pmid:32827647.
- 8. Clish CB. (2015) Metabolomics: an emerging but powerful tool for precision medicine. Cold Spring Harbor molecular case studies.1(1):a000588. pmid:27148576.
- 9. Srivastava S. (2019) Emerging Insights into the Metabolic Alterations in Aging Using Metabolomics. Metabolites.9(12). pmid:31847272.
- 10. Andraos S, Wake M, Saffery R, Burgner D, Kussmann M, O’Sullivan J. (2019) Perspective: Advancing Understanding of Population Nutrient-Health Relations via Metabolomics and Precision Phenotypes. Advances in nutrition (Bethesda, Md).10(6):944–52. pmid:31098626.
- 11. Després JP. (2020) Predicting longevity using metabolomics: a novel tool for precision lifestyle medicine? Nature reviews Cardiology.17(2):67–8. pmid:31754191.
- 12. Jacob M, Lopata AL, Dasouki M, Abdel Rahman AM. (2019) Metabolomics toward personalized medicine. Mass spectrometry reviews.38(3):221–38. pmid:29073341.
- 13. Schranner D, Kastenmüller G, Schönfelder M, Römisch-Margl W, Wackerhage H. (2020) Metabolite Concentration Changes in Humans After a Bout of Exercise: a Systematic Review of Exercise Metabolomics Studies. Sports medicine—open.6(1):11. pmid:32040782.
- 14. Sakaguchi CA, Nieman DC, Signini EF, Abreu RM, Catai AM. (2019) Metabolomics-Based Studies Assessing Exercise-Induced Alterations of the Human Metabolome: A Systematic Review. Metabolites.9(8). pmid:31405020.
- 15. Xiao Q, Moore SC, Keadle SK, Xiang YB, Zheng W, Peters TM, et al. (2016) Objectively measured physical activity and plasma metabolomics in the Shanghai Physical Activity Study. International journal of epidemiology.45(5):1433–44. pmid:27073263.
- 16. Ding M, Zeleznik OA, Guasch-Ferre M, Hu J, Lasky-Su J, Lee IM, et al. (2019) Metabolome-Wide Association Study of the Relationship Between Habitual Physical Activity and Plasma Metabolite Levels. American journal of epidemiology.188(11):1932–43. pmid:31364705.
- 17. Krumsiek J, Mittelstrass K, Do KT, Stückler F, Ried J, Adamski J, et al. (2015) Gender-specific pathway differences in the human serum metabolome. Metabolomics: Official journal of the Metabolomic Society.11(6):1815–33. pmid:26491425.
- 18. Lee J, Jung Y, Park JY, Lee SH, Ryu DH, Hwang GS. (2015) LC/MS-based polar metabolite profiling reveals gender differences in serum from patients with myocardial infarction. Journal of pharmaceutical and biomedical analysis.115:475–86. pmid:26299524.
- 19. Mittelstrass K, Ried JS, Yu Z, Krumsiek J, Gieger C, Prehn C, et al. (2011) Discovery of sexual dimorphisms in metabolic and genetic biomarkers. PLoS genetics.7(8):e1002215. pmid:21852955.
- 20. Patel MJ, Batch BC, Svetkey LP, Bain JR, Turer CB, Haynes C, et al. (2013) Race and sex differences in small-molecule metabolites and metabolic hormones in overweight and obese adults. Omics: a journal of integrative biology.17(12):627–35. pmid:24117402.
- 21. Babulal GM, Williams MM, Stout SH, Roe CM. (2018) Driving Outcomes among Older Adults: A Systematic Review on Racial and Ethnic Differences over 20 Years. Geriatrics (Basel, Switzerland).3(1). pmid:29657944.
- 22. Vasishta S, Ganesh K, Umakanth S, Joshi MB. (2022) Ethnic disparities attributed to the manifestation in and response to type 2 diabetes: insights from metabolomics. Metabolomics: Official journal of the Metabolomic Society.18(7):45. pmid:35763080.
- 23. Hu J, Yao J, Deng S, Balasubramanian R, Jiménez MC, Li J, et al. (2022) Differences in Metabolomic Profiles Between Black and White Women and Risk of Coronary Heart Disease: an Observational Study of Women From Four US Cohorts. Circulation research.131(7):601–15. pmid:36052690.
- 24. van Valkengoed IGM, Argmann C, Ghauharali-van der Vlugt K, Aerts J, Brewster LM, Peters RJG, et al. (2017) Ethnic differences in metabolite signatures and type 2 diabetes: a nested case-control analysis among people of South Asian, African and European origin. Nutrition & diabetes.7(12):300. pmid:29259157.
- 25. Kemppainen SM, Fernandes Silva L, Lankinen MA, Schwab U, Laakso M. (2022) Metabolite Signature of Physical Activity and the Risk of Type 2 Diabetes in 7271 Men. Metabolites.12(1). pmid:35050191.
- 26. Hu J, Balasubramanian R, Li J, Clish CB, Paynter NP, Albert CM, et al. (2019) Abstract MP55: Racial Differences in Metabolomic Profiles Between African Americans and Whites in the Women’s Health Initiative. Circulation.139(Suppl_1):AMP55–AMP.
- 27. Du Y, Zhao LJ, Xu Q, Wu KH, Deng HW. (2017) Socioeconomic status and bone mineral density in adults by race/ethnicity and gender: the Louisiana osteoporosis study. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA.28(5):1699–709. pmid:28236128.
- 28. Ning HT, Du Y, Zhao LJ, Tian Q, Feng H, Deng HW. (2021) Racial and gender differences in the relationship between sarcopenia and bone mineral density among older adults. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA.32(5):841–51. pmid:33231702.
- 29. (2014) World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. The Journal of the American College of Dentists.81(3):14–8. pmid:25951678.
- 30. Zhao Q, Shen H, Su KJ, Zhang JG, Tian Q, Zhao LJ, et al. (2018) Metabolomic profiles associated with bone mineral density in US Caucasian women. Nutrition & metabolism.15:57. pmid:30116286.
- 31. Zhou Y, Wu K, Shen H, Zhang J, Deng HW, Zhao LJ. (2019) Geographical differences in osteoporosis, obesity, and sarcopenia related traits in white American cohorts. Scientific reports.9(1):12311. pmid:31444391.
- 32. U.S. Department of Health and Human Services. Physical activity guidelines for Americans, 2nd Wasington, DC; 2018 [Available from: https://health.gov/sites/default/files/2019-09/Physical_Activity_Guidelines_2nd_edition.pdf.
- 33. O’Donovan G, Lee IM, Hamer M, Stamatakis E. (2017) Association of "Weekend Warrior" and Other Leisure Time Physical Activity Patterns With Risks for All-Cause, Cardiovascular Disease, and Cancer Mortality. JAMA internal medicine.177(3):335–42. pmid:28097313.
- 34. Stofan JR, DiPietro L, Davis D, Kohl HW 3rd, Blair SN. (1998) Physical activity patterns associated with cardiorespiratory fitness and reduced mortality: the Aerobics Center Longitudinal Study. Am J Public Health.88(12):1807–13. pmid:9842378.
- 35. Izquierdo M, Merchant RA, Morley JE, Anker SD, Aprahamian I, Arai H, et al. (2021) International Exercise Recommendations in Older Adults (ICFSR): Expert Consensus Guidelines. The journal of nutrition, health & aging.25(7):824–53. pmid:34409961.
- 36. Benjamini Y, Hochberg Y. (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal statistical society: series B (Methodological).57(1):289–300.
- 37. MetaboAnalyst. MetaboAnalyst5.0—user-friendly, streamlined metabolomics data analysis 2022 [cited 2022 08/09]. Available from: https://www.metaboanalyst.ca/MetaboAnalyst/ModuleView.xhtml.
- 38. Jackson FLC. (1992) Race and ethnicity as biological constructs. Ethnicity & disease:120–5. pmid:1467750
- 39. Van den Berghe PL. (1978) Race and ethnicity: a sociobiological perspective. Ethnic and racial studies.1(4):401–11.
- 40. Wang W, Cui J, Ma H, Lu W, Huang J. (2021) Targeting Pyrimidine Metabolism in the Era of Precision Cancer Medicine. Frontiers in oncology.11:684961. pmid:34123854.
- 41. Xu J, Liu G, Hegde SM, Palta P, Boerwinkle E, Gabriel KP, et al. (2021) Physical Activity-Related Metabolites Are Associated with Mortality: Findings from the Atherosclerosis Risk in Communities (ARIC) Study. Metabolites.11(1). pmid:33477977.
- 42. Palau-Rodriguez M, Garcia-Aloy M, Miñarro A, Bernal-Lopez MR, Brunius C, Gómez-Huelgas R, et al. (2020) Effects of a long-term lifestyle intervention on metabolically healthy women with obesity: Metabolite profiles according to weight loss response. Clinical Nutrition.39(1):215–24. pmid:30862367
- 43. Rosenfeldt FL. (1998) Metabolic supplementation with orotic acid and magnesium orotate. Cardiovascular drugs and therapy.12(2):147–52. pmid:9794088
- 44. Golf S, Happel O, Graef V, Seim K. (1984) Plasma aldosterone, cortisol and electrolyte concentrations in physical exercise after magnesium supplementation.
- 45. Ali BH, Karaca T, Al Suleimani Y, Al Za’abi M, Al Kalbani J, Ashique M, et al. (2017) The effect of swimming exercise on adenine-induced kidney disease in rats, and the influence of curcumin or lisinopril thereon. PloS one.12(4):e0176316. pmid:28445490.
- 46. Yegutkin GG, Samburski SS, Mortensen SP, Jalkanen S, González-Alonso J. (2007) Intravascular ADP and soluble nucleotidases contribute to acute prothrombotic state during vigorous exercise in humans. The Journal of physiology.579(Pt 2):553–64. pmid:17204504.
- 47. Xiao Q, Moore SC, Keadle SK, Xiang Y-B, Zheng W, Peters TM, et al. (2016) Objectively measured physical activity and plasma metabolomics in the Shanghai Physical Activity Study. International journal of epidemiology.45(5):1433–44. pmid:27073263
- 48. Kim H, Hu EA, E Wong K, Yu B, Steffen LM, Seidelmann SB, et al. (2021) Serum metabolites associated with healthy diets in African Americans and European Americans. The Journal of nutrition.151(1):40–9. pmid:33244610
- 49. Al-Khelaifi F, Donati F, Botrè F, Latiff A, Abraham D, Hingorani A, et al. (2019) Metabolic profiling of elite athletes with different cardiovascular demand. Scandinavian journal of medicine & science in sports.29(7):933–43. pmid:30929282.
- 50. Rebholz CM, Lichtenstein AH, Zheng Z, Appel LJ, Coresh J. (2018) Serum untargeted metabolomic profile of the Dietary Approaches to Stop Hypertension (DASH) dietary pattern. The American journal of clinical nutrition.108(2):243–55. pmid:29917038.
- 51. Ren X, Chen J, Abraham AG, Xu Y, Siewe A, Warady BA, et al. (2023) Abstract P212: Plasma Metabolomics of Dietary Intake of Protein-Rich Foods and Kidney Disease Progression in Children. Circulation.147(Suppl_1):AP212–AP.
- 52. Hsu S, Hoofnagle AN, Gupta DK, Gutierrez OM, Peralta CA, Shea S, et al. (2020) Race, Ancestry, and Vitamin D Metabolism: The Multi-Ethnic Study of Atherosclerosis. The Journal of clinical endocrinology and metabolism.105(12):e4337–50. pmid:32869845.
- 53. Feagin J, Bennefield Z. (2014) Systemic racism and U.S. health care. Social science & medicine (1982).103:7–14. pmid:24507906.
- 54. Gee GC, Payne-Sturges DC. (2004) Environmental health disparities: a framework integrating psychosocial and environmental concepts. Environmental health perspectives.112(17):1645–53. pmid:15579407.
- 55. Bleich SN, Findling MG, Casey LS, Blendon RJ, Benson JM, SteelFisher GK, et al. (2019) Discrimination in the United States: Experiences of black Americans. Health services research.54 Suppl 2(Suppl 2):1399–408. pmid:31663124.
- 56. Murkey JA, Watkins BX, Vieira D, Boden-Albala B. (2022) Disparities in allostatic load, telomere length and chronic stress burden among African American adults: A systematic review. Psychoneuroendocrinology.140:105730. pmid:35338946.
- 57. Lago S, Cantarero D, Rivera B, Pascual M, Blázquez-Fernández C, Casal B, et al. (2018) Socioeconomic status, health inequalities and non-communicable diseases: a systematic review. Zeitschrift fur Gesundheitswissenschaften = Journal of public health.26(1):1–14. pmid:29416959.
- 58. Müller-Riemenschneider F, Reinhold T, Berghöfer A, Willich SN. (2008) Health-economic burden of obesity in Europe. European journal of epidemiology.23(8):499–509. pmid:18509729.
- 59. Buergel T, Steinfeldt J, Ruyoga G, Pietzner M, Bizzarri D, Vojinovic D, et al. (2022) Metabolomic profiles predict individual multidisease outcomes. Nature medicine.28(11):2309–20. pmid:36138150.
- 60. Kemmler W, von Stengel S. (2014) Dose-response effect of exercise frequency on bone mineral density in post-menopausal, osteopenic women. Scandinavian journal of medicine & science in sports.24(3):526–34. pmid:23190199.
- 61. Powell KE, King AC, Buchner DM, Campbell WW, DiPietro L, Erickson KI, et al. (2018) The Scientific Foundation for the Physical Activity Guidelines for Americans, 2nd Edition. Journal of physical activity & health:1–11. pmid:30558473.