Physical activity, sedentary behavior and microbiome : A systematic review and meta-analysis

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC-ND 4.0 https://creativecommons.org/licenses/by-nc-nd/4.0/ Physical activity, sedentary behavior and microbiome : A systematic review and metaanalysis © 2024 The Authors. Published by Elsevier Ltd on behalf of Sports Medicine Australia Published version Pérez-Prieto, Inmaculada; Plaza-Florido, Abel; Ubago-Guisado, Esther; Ortega, Francisco, B.; Altmäe, Signe Pérez-Prieto, I., Plaza-Florido, A., Ubago-Guisado, E., Ortega, F., & Altmäe, S. (2024). Physical activity, sedentary behavior and microbiome : A systematic review and meta-analysis. Journal of Science and Medicine in Sport, 37(11), 793-804. https://doi.org/10.1016/j.jsams.2024.07.003 2024 Journal of Science and Medicine in Sport 27 (2024) 793–804 Contents lists available at ScienceDirect Journal of Science and Medicine in Sport journal homepage: www.elsevier.com/locate/jsams Review Physical activity, sedentary behavior and microbiome: A systematic review and meta-analysis Inmaculada Pérez-Prieto a a Department of Biochemistry and Molecular Biology I, Faculty of Sciences, University of Granada, Spain ,b b Instituto de Investigación Biosanitaria ibs.GRANADA, Spain , ⁎ ⁎ Corresponding authors. E-mail addresses: [email protected] (I. Pérez-Prieto), [email protected] (A. Plaza-Florido), orteg [email protected] (F.B. Ortega), [email protected] (S. Altmäe). Social media: @InmaPerezPrieto (I. Pérez-Prieto). , AbelPlaza-Florido c c Department of Physical Education and Sports, Faculty of Sport Sciences, Sport and Health University Research Institute (iMUDS), University of Granada, Spain ,d d Pediatric Exercise and Genomics Research Center, UC Irvine School of Medicine, United States , Esther Ubago-Guisado b,c , Franc isco B. Ortega c,e e CIBER de Fisiopatología de la Obesidad y Nutrición (CIBEROBN), Instituto de Salud Carlos III, Granada, Spain ,f f Faculty of Sport and Health Sciences, University of Jyväskylä, Finland , ⁎ , Signe Altmäe a,b,g g Division of Obstetrics and Gynaecology, Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Huddinge, Stockholm, Sweden ,h h Department of Gynaecology and Reproductive Medicine, Karolinska University Hospital, Huddinge, Stockholm, Sweden ar ti cl e i n f o Article history: Received 16 May 2023 Received in revised form 18 May 2024 Accepted 2 July 2024 Available online 9 July 2024 Keywords: Exercise Microbiota Omics 16S rRNA gene sequencing Metagenomics ab st ra c t Background: The effects of physical activity and sedentary behavior on human health are well known, however, the molecular mechanisms are poorly understood. Growing evidence points to physical activity as an important modulator of the composition and function of microbial communities, while evidence of sedentary behavior is scarce. We aimed to synthesize and meta-analyze the current evidence about the effects of physical activity and sedentary behavior on microbiome across different body sites and in different populations. Methods: A systematic search in PubMed, Web of Science, Scopus and Cochrane databases was conducted until September 2022. Random-effects meta-analyses including cross-sectional studies (active vs. inactive/athletes vs. no n-athletes) or trials reporting the chronic effect of physical activity interventions on gut microbiome alpha-diversity in healthy individuals were performed. Results: Ninety-one studies were included in this systematic review. Our meta-analyses of 2632 participants indicated no consistent effect of physical activity on microbial alpha-diversity, although there seems to be a trend toward a higher microbial richness in athletes compared to non-athletes. Most of studies reported an increase in short-chain fatty acid-producing bacteria such as Akkermansia, Faecalibacterium, Veillonella or Roseburia in active individuals and after physical activity interventions. Conclusions: Physical activity levels were positively associated with the relative abundance of short-chain fatty acid-producing bacteria. Athletes seem to have a richer microbiome compared to non-athletes. However, high heterogeneity between studies avoids obtaining conclusive information on the role of physical activity in microbial composition. Future multi-omics studies would enhance our understanding of the molecular effects of physical activity and sedentary behavior on the microbiome. © 2024 The Authors. Published by Elsevier Ltd on behalf of Sports Medicine Australia. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). https://doi.org/10.1016/j.jsams.2024.07.003 1440-2440/© 2024 The Authors. Published by Elsevier Ltd on behalf of Sports Medicine Australia. This is an open access article under the CC BY-NC-ND license (http://creativecommons. org/licenses/by-nc-nd/4.0/). Key points • Most studies indicate that ph ysical activity alters the microbiome composition, mainly affecting the relative abundance of short-chain fatty acid-producing bacteria with health benefits, although its influence on microbiome diversity is unclear. Therefore, the existing evidence needs to be quantified using meta-analytic methods in the physical activity-microbiome field. • Our search identified high h eterog eneity in study popu lations and characteristics of physical activity in terventions, most of evidence based on cross-sectional studies using self-reported questionnaires to assess physical activity, lack of reference pipelines for microbiome analysis and relevant covariates missing in statistical analyses, especially diet. • More research on the effect of sedentary behavior on microbiome composition is needed. • Integrated multi-omics studies on bigger sample size are warranted to clarify the molecular effects of physical activity and sedentary behavior on human microbial communities. I. Pérez-Prieto, A. Plaza-Florido, E. Ubago-Guisado et al. Journal of Science and Medicine in Sport 27 (2024) 793–804 • Our systematic review and meta-analysis suggests that (1) physical activity levels were positively associated with the relative abundance of short-ch ain fatty acid-producing bacteria; and that (2) athletes seem to have a richer microbiome compared to non-athletes. 1. Introduction It is well known that physical activity (PA) (i.e., any movement produced by skeletal muscles which demands a higher energy expenditure than in resting conditions) can improve different health-related outcomes such as insulin resistance, adiposity, and fitness, among othe rs. 1,2 A related yet different construct is sedentary behavior (SB) (i.e., a behavior characterized by an energy expenditure of 1.5 or fewer metabolic equivalents [METs]), and is associated with a higher risk of different diseases. 3,4 Thus, increasing PA and reducing SB have been considered to prevent and treat multiple chronic diseases. 5 However, the molecular mechanisms underlying the health benefits of PA (acute or chronic effects) and the adverse effects of SB on health are poorly understood. 6 In the last decades, a new insight of the human being as a set of microbial and human cells has emerged. 7 The collectio n of microorganisms including bacteria, vi ruses, fungi and archaea that inhabits our body is defined as the microbiota and is a t least as abundant as the number of human cells. 8 The genomes of th e abovementioned microorganism s (i.e., microbiota) are called the microbiome, which is considered “our second genome” and “our last organ” due to its important role in human physiology. 9,10 Microbiome composition is profiled through metagenomics approaches such as marker gene and shotgun metagenomic sequencing. While marker gene sequencing targets a specific sequence of a gene (e.g., 16S rRNA gene) to provide a microbial classification that lacks accuracy at the species level, shotgun metagenomics consists of the sequencing of all microbial genomes within a sample, allowing a deeper taxonomic composition at species level and dete ct ing vira l and eukaryo tic DNA. Metagenomics studies (e.g., marker gene sequencing and shotgun metagenomic sequencing) led to the characterization of microbial composition using three common analyses: (1) alpha-diversity, that characterizes the microbial diversity within a sample considering richness and evenness (i.e., the number and the relative abundance of microbes); (2) beta-diversity, which measures the diversity between samples assigning a numerical value for every pair of samples to determine microbial community-level dissimilarities; and (3) differential abundance analysis, that identifies those microo rganisms that differ in abundance when compared different samples. There is evidence indicating that environmental and lifestyle factors such as pollutants, drugs, diet, lack of PA and increased SB, among others, may have a negative impact on microbiome compos ition and function leading to the disruption of the microbial hom eostasis. 11–14 In fact, microbial imbalances have been associated with the development of multiple diseases such as obesity, 15,16 type 2 diabetes, 17 and cancer, 18,19 among others. Thus, there is a growing interest to determine the composition of the “healthy core” microbiome and the factors that could shape the microbial communities, such as PA and SB, in order to design new therapeutic interventions. 20,21 Particularly, PA has been proposed as one of the modulators of the host-associated microbiome, while little is known about the effect of SB on microbial communities due to the limited number of studies. 22–24 Recent advances in meta-omics-based studies (i.e., marker gene sequencing, metagenomics, meta-transcriptomics, meta-proteomics, and meta-metab olomics) allow the identification of the molecular pathways regulatedbyPA. 25 Thus, the effect of PA on the microbiome, especially on the gut m icrobial communities, is a rese arch topic of increasing interest. 26,27 In the last years, several systematic reviews reported the effects of PA on the gut microbiome of healthy adults, 23,28–30 older adults 31 and adults with obesity and type 2 diabetes. 32,33 In addition, a syst ematic review on the effect of aerobic athletic performance has been recently published. 34 However, the afo rementioned systematic reviews showed inconsistent findings from observational and intervention studies. 23,28–33 Therefore, there is a need to synthesize the whole body of knowledge about the effect of PA and SB on the microbiome including healthy (e.g., non-athletes and professional athletes), unhealthy populations (e.g., obesity, diabetes, cancer), different stages of life (i.e., children, young and older adults), and different body niches (e.g., gut, saliva, vagina, etc.) through metagenomics approaches. In addition, there is still a lack of meta-analytic studies quantifying the effect of PA on the microbiome within non-athletic populations so far. Therefore, the current study aimed: (1) to summarize all the studies available about the relationship of PA and SB (observational and intervention studies) with the human-associated microbiome performing metagenomics and (2) meta-analyze the available data. 2. Material and methods This systematic revi ew and meta-analysis was condu cted following the Preferred Reporting I tems for Systematic Reviews and MetaAn alyses guidelines (PRISM A). 35 The review protocol was registered in the International Prospective R egister of S ystematic Revie ws (PROSPERO; http://www.crd.york.ac.uk/PROSPERO) with the reference number: CRD42022298526. 2.1. Search strategy A systematic search was conducted in PubMed , Web of Science, SCOPUS, and Cochrane electronic databases from inception to September 29, 2022. Search terms were included based on the sport science and microbiome terms of interest. Table 1 includes a list with the Table 1 Definition of main microbiome-related terms used in this systematic review. Term Definition Microbiota Collection of microorganisms, including bacteria, archaea, viruses and fungal communities, that collectively inhabit a particular environment (e.g., gut, blood, vagina, etc.) Microbiome Collection of genomes of the microorganisms inhabiting a particular environment Alpha-diversity Diversity within a sample taking into account both the number of microorganisms in a sample (richness) and their distribution (evenness) Shannon Diversity Index Alpha-diversity estimator of microbial richness and evenness within a sample or niche Chao1 Index Alpha-diversity abundance-based estimator of microbial richness within a sample or niche Beta-diversity Diversity between samples taking into account a distance matrix that reflects how compositionally different the samples are from one another (i.e., dissimilarity between samples) Meta-omics Refers to those techniques, including marker gene sequencing, metagenomics, meta-transcriptomics, meta-proteomics, and meta-metabolomics, which directly examine the phylogenetic markers, genes, transcripts, proteins, or metabolites from a microbial community Marker gene sequencing (e.g., 16S rRNA gene sequencing) DNA sequencing method to identify the microbes present in a given microbial community through the analysis of a sequence variation (i.e., hypervariable region) of a single ubiquitous gene (e.g., 16S ribosomal RNA gene) Metagenomics (or shotgun metagenomics) DNA sequencing method to assess the entire functional gene content of a given microbial community. It provides a much greater specific identification of the microbes compared to marker gene analysis (e.g., 16S rRNA gene), in which classification is normally limited to the genus level as multiple species may have the same sequence within the studied hyper variable region Meta-transcriptomics RNA sequencing method to assess the transcriptionally active microbes of a given microbial community, providing knowledge of the functional activity of these microorganisms Meta-proteomics Method to characterize the entire microbial protein complement of a sample Meta-metabolomics Method to identify the microbial metabolites present in a sample 794 main terms and their definitions related to microbiome field used in this syst ematic search. Electronic Supplementary Table S1 illustrates the search terms and strategy for each database. I. Pérez-Prieto, A. Plaza-Florido, E. Ubago-Guisado et al. Journal of Science and Medicine in Sport 27 (2024) 793–804 2.2. Study selection criteria The inclusion criteria were as follows: (1) all observational studies (longitudinal or cross-section al) that report the associa tion of PA and/or SB with microbiome; and (2) all original studies that included the effect of PA (acute and/or chronic effects) on microbiome. The exclusion criteria were: (1) studies addressing the effect of PA (acute or chronic effects) on microbiome containing diet modifications, probiotics and prebiotic supplements or caloric restriction, in which it was not possible to isolate the independent effect of PA, (2) non-eligible publication types, such as editorials, study protocols, letters to the editors, meeting abstracts, or review articles, and (3) studies written in any language other than English or Spanish. The selection process of the studies resulting from the literature search was performed using the software “Covidence” (https://www.covidence. org/), which detected duplicates. After removing the duplicates, the articles were first independently filtered by title/abstract screening by two researchers (I.P.P and A.P.F). Those articles that met the inclusion criteria were selected for the full-text review. Conflictive articles were solved through common consensus by the same researchers (I.P.P and A.P.F). Any article that did not meet the eligibility criteria was excluded. The quality assessment of the included studies was independently conducted by I.P.P and A.P.F (see Electronic Supplementary Material Appendix S1). 2.3. Data extraction For each study, one researcher (I.P.P) conducted the data extraction including the following information: ( 1) author's name and date of publ ication, (2) study design, (3) characteristics of the p opulation (number of participants, sex, age and ethnicity), (4) characteristics of the exposure (i.e., PA or SB), (5) sample origin, (6) dependent outcome (i.e., DNA extraction method, detection method of the microbiome and sequencing platform), (7) dietary record, (8) microbiome analyses, (9) main findings and (10) raw data availability. A second researcher (A.P.F) performed a double-check for data correction. 2.4. Data synthesis and meta-analysis We conducted three meta-analyses including cross-sectional studies (active vs. inactive/athletes vs. n on-athletes) or trials reporting the chronic effect of PA interventions on gut microbiome diversity (specifically alpha-diversity, expressed by the Shannon diversity and Chao1 indexes) in healthy individuals. This decision was made considering the limited availability of microbiome data and the heterogeneity observed acros s the identified studies (see Electronic Supplementary Material Appendix S1 for detailed explanation). Statistical analyses were performed using the Comprehensive MetaAnalysis software (version 3; Biostat Inc., 1385, NJ, USA). The effect size was calculated using Cohen's d and 95 % confidence intervals (CIs) for standardized mean difference (SMD). Pooled SMD was estimated using a random-effects model. Heterogeneity between studies was assessed using the I 2 statistics, which represents the percentage of total variation ac ross studies, considering I 2 values of 25 %, 50 %, and 75 % as low, moderate and high heterogeneity, respectively. 36 Ap value ofless than 0.05 was considered statistically significan t. 3. Results 3.1. General overview PRISMA checklist 2020 reflects the appropriateness of the methods performed in this systematic review and meta-analysis (Electronic Supplementary Tables S2, S3). Fig. 1 illustrates the PRISMA flow diagram of the search process. A total of 12,503 articles were detected across the four databases, and after removing the duplicates and noneligible articles, 91 studies were included in this systematic review: 50 observational studies (all cross-sectio nal), 37–86 9 studies reported the acute effects of PA (e.g ., followi ng a marathon, rowin g, etc.) on microbiome, 25,87–94 and 32 studies reported the chronic effects of PA on microbiome (1 7 non-RCT, 13 RCT and 2 randomized controlled cros s-over t rials ). 24,95–125 Of the 50 cross-sectional s tudies, 8 were eligible (based on availa bility of microbiome diversity data and healthy participants) for the first meta-analysis comparing groups of high and low PA levels in non-athletes, 41,42,44,46,53,73,76,84 and 11 were included in the second meta-analysis comparing athletes vs. non-athletes. 42,56,57,59,64,67,80,82,84,85,112 Of the 32 intervention studies, 7 were selected for the third meta-analysis, to evalua te the chronic effects of PA on microbiome alpha-diversity. 102,104,105,107,116,119,125 Seventeen studies reported significant associations between PA 40,44–47,50,52,53,70,72–75,78 or SB 39,41,71 and micro bial diversity (i. e., alphaand/or beta-diversity), and 19 studies found significant differences in the relative abundance of specificbacteriainactive vs. inactive participants 37,39–41,45–48,50,51,53–55,70,73,75–77,84 (Table S8). Sixteen studies fou nd significant differences in microbial diversity 56,57,59,60,62–66,68,69,80–82,85,86 and 13 in the abundance of specific microbial taxa 56,60,61,63–66,80–82,84,85,112 between athletes vs. non-athletes, professional vs. amateur or athletes from different sports. Three studies detected significant differences in alpha-diversity, 90,91,93 while 8 studies described significant changes in the relative abundance of certain bacteria after acute PA interventions. 25,87,88,90–94 Seventeen studies detected significant differences in alphaand/or betadiversity, 24,95,97,98,101,103,106,109,110,112,115,118,119,121,122,124,125 and 24 studies described significant changes in the relative abundance of certain bacteria after chronic PA interventions. 24,95,97,99–103,105–107,109,110,113–116,118–123,125 The sa mple sizes ranged from 1 91 to 2183. 72 Fifty-three studies involved both male and female participants, while 12 were exclusively conducted on women and 25 on men (see Table S8). One study did not report the gender of the participants. 118 Regarding age, 5 studies recruited children (i.e., 7–12 years) and/or adolescents (i.e., 13–17 years), 67 included young an d middle-aged adults (i.e., 18–64 years), 12 older adults (i.e., ≥65 years), 3 studies combined adolescents and adults and 4 adults of different ages (Table S8). Fiftynine studies were performed on healthy individuals, while 32 studies included participants with different diseases such as obesity or breast cancer, among others (Table S8). Regarding the exposure, 26 cross-sectional studies recorded PA using self-r eported questionnaires, 37–42,44–55,72–75,77–79,84 whereas 8 studies included PA data registered by accelerometry 39,43,45,46,51,70,73,76 (Table S8). Additionally, four studies reported SB data expressed as time per sedentary breaks/bouts or screen time. 39,41,71,79 Twenty-two cross-sectional stu dies recruited ath letes from different sports such as rugby, athletics or football, among others. 56–69,80–86,112 Six stu dies analyzed the effects of a marathon, footrace or rowing race on microbiomes, 25,88,90,91,93,94 three reported the effect of a single bout of PA (i.e., no sport competition) on microbiomes 87,89,92 and 32 conducted a long-term PA intervention ranged from two weeks 99 to thirty-four weeks, 122 mostly consistin g of aerobic training 24,95,98,99,105,106,111,113,114,116,118–120,122,125 or a combination of aerobic and resistance training. 96,97,100–104,107–110,112,115,117,121,123,124 Most of the studies analyzed the gut microbiome, with the exceptio n of te n whic h colle cted sa liva, oral, oropharyng eal, musc le, blood o r v ag inal samples. 37,52,60,71,86,87,89,110,120,123 Concerning the detection method, 78 studies conducted the 16S rRNA gene sequencing ap proach to charac terize the microbiome, 16 performed metagenomics analyses and two studies focused on metatranscriptomics (i.e., microbial RNA-sequencing) (see Table S8). Twenty-one studies did not report die tary data for all the participants (Table S8). One study performed a control of diet (each 795 participants received the same kind of food) during the PA intervention. 88 I. Pérez-Prieto, A. Plaza-Florido, E. Ubago-Guisado et al. Journal of Science and Medicine in Sport 27 (2024) 793–804 Records identified from: Databases (n = 12503) -PubMed -Web of Science -SCOPUS -Cochrane Library Records removed before screening: Duplicate records removed using automated function in the tool COVIDENCE (n = 6744) Records screened by title/abstract (n = 5759) Records excluded (n = 5630) Reports assessed for eligibility (n = 129) Reports excluded: Reason 1: Experimental studies with other lifestyle intervention (n = 6) Reason 2: Irrelevant independent variable (n = 12) Reason 3: Irrelevant outcome (n = 5) Reason 4: Congress abstract (n= 8) Reason 5: Book chapter (n=1) Reason 6: Duplicate (n = 5) Reason 7: No accessible (n = 1) Studies included in review (n = 91) Studies included in meta-analysis (n = 24) Identification of studies via databases and registers noitacifitnedI Screening Included Fig. 1. Search process according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 flow diagram. Fig. 2 shows a graphical summary of the main findings. Specificoutcomes of microbial composition identified in the articles are discussed and interpreted in the context of the current kn owledge in the Discussion section. For further details, see Electronic Supp lementary Appendix S2. 3.2. Quality assessment Among the 50 cross-sectional studies, 26 were categorized as high quality (quality score ≥ 75 %), whereas 24 as low quality (quality score < 75) (Electronic Supplementary Table S4). Regarding the 9 studies about the acute effects of PA, 8 studies were considered to have a high quality and 1 showed a low qua lity ( Electronic Supplementary Table S5). Concerning the 32 studies (15 RCTs and 17 non-RCTs) that reported the chronic effects of PA interventions, one RCT presented a high quality and 14 a low quality (Electronic Supplementary Table S6), while 12 non-RCTs were categorized as high quality and 5 as low quality studies (Electronic Supplementary Table S7). 3.3. Meta-analysis 3.3.1. First meta-analysis (cross-sectional studies): high vs. low PA levels This meta-analysis united 1814 participants from 8 studies, where 1157 belonged to the high PA and 657 participants to the low PA groups. No significant differences were reported betwe en the groups of high and low PA levels on alpha-diversity represented by the Shannon diversity index (SMD = −0.101, 95 % CI −0.386–0.184, p = 0.488, I 2 = 33.581) and Chao1 index (SMD = −0.127, 95 % CI −0.563–0.309, p = 0.568, I 2 = 13.774) (Fig. 3A). 3.3.2. Second meta-analysis (cr oss-sectional studies): athletes vs. non-athletes This meta-analysis comprised 651 participants from 11 studies, including 329 athletes and 322 non-athletes. No significant differences were repo rted between the groups of athletes and non-athle tes on alp ha-diversity using the Shannon diversity index (SMD = −0.113, 95 % CI −0.441–0.215, p = 0.501, I 2 = 0.000). However, athletes tended to present a higher alpha-diversity compared to non-athletes when Chao1 index was used as an indicator of microbial alpha-div ersity (SMD = 0.482, 95 % CI −0.026–0.991, p = 0.063, I 2 =0.000) (Fig. 3B). 796 I. Pérez-Prieto, A. Plaza-Florido, E. Ubago-Guisado et al. Journal of Science and Medicine in Sport 27 (2024) 793–804 Fig. 2. Summary of the main characteristics and findings of the studies included in this systematic review. A) Exposure: in cross-sectional studies, the effect of physical activity (PA), sedentary behavior (SB) or athletic performance on microbiome was analyzed. In intervention studies, the ac ute or chronic effects of PA on microbiomewereevaluated. B) Microbiome outcomes: samples from different body sites (gut, saliva, blood, muscle and vagina, among others) were analyzed by distinct detection methods (16S rRNA sequencing and shotgun metagenomic sequencing for DNA-based microbiome analysis; meta-transcriptomic [RNA sequencing] for RNA-based microbiome analysis). C) Main findings: relevant results concerning alphaand beta-diversity and differential abundance analysis are shown. D) Metabolic effects of PA-microbiome interaction. Growing evidence indicates that PA increases the abundance of members of the Firmicutes phylum, bacteria able to produce short-chain fatty acids (SCFAs). SCFAs produced by the gut microbiome by processing nutrients from diet may have positive effects in the intestine, improving barrier function and inflammation state. A crosstalk between the gut microbiome and skeletal muscle through lactate (generated during PA) and its conversion to SCFAs may improve athletic performance. SCFAs have been also linked to promoting neurogenesis (through brain-derived neurotrophic factor [BDNF]), improving hypothalamic–pituitary–adrenal (HPA) axis control, reducing inflammation and the risk of psychological diseases (e.g., depression, anxiety). A microbiome-dependent gut-brain connection mediated by microbial metabolites (i.e., fatty acid amides [FAAs], such as N-oleoylethanolamide [OEA]) has been discovered in mice, which enhances exercise performing and motivation by increasing dopamine signaling during PA. Recent studies suggest that disruption of microbial ecosystem may lead to the growth of proteolytic microbes able to produce trimethylamine-N-oxide (TMAO), an important metabolite that in elevated concentration has been linked to adverse cardiac events and chronic kidney diseases (CKD). This figure was created with BioRender.com. 3.3.3. Third meta-analysis (intervention studies): chronic effects of PA The third meta-analysis united 167 participants from 7 studies, where 118 were allocated to a PA group and 49 to a control group. No significant differe nces were found between the PA and control groups on alpha-diversity using the Shannon diversity index (PA group: SMD = 0.132, 95 % CI −0.124–0.388, p = 0.312, I 2 = 0.000; control group: SMD = 0.110, 95 % CI −0.288–0.508, p = 0.587; I 2 = 0.000) or Chao1 index (PA group: SMD = −0.080, 95 % CI −0.454–0.295, p = 0.677, I 2 = 0.000; co ntrol group: SMD = 0.001, 95 % CI −0.454–0.457, p = 0.995; I 2 =0.000)(Fig. 3C). 4. Discussion This is the first systematic review that summarizes the current evidence about the effects of PA and SB on the human-associated microbiome across differen t body site s and in different populations. The main findings of this systematic review and meta-analysis were: (1) there was no consistent effect of PA on modifying microbial alphadiversity, although most of studies support that PA (observational and intervention studies) induces changes in microbiome composition with the increase of short-chain fatty acid (SCFA)-producing bacteria such as Akkermansia, Roseburia or Veillonella, among others; (2) there is very limited evidence of the effect of SB on microbiome; (3) few stu dies assessed PA data by objective methods (i.e., accelerometry); (4) there are few studies about the acute effect of PA on microbiome; (5) available studies are h ardly comparabl e due to heterogeneity of the participants (i.e., age, sex, health status), wide use of different selfreported questionnaires to record PA, lack of standardized criter ia to stratify participants in active/sedentary groups in cross-sectional studies and different characteristics of PA interventions (e.g., type, intensity, duration); (6) most of studies did not include diet as a confounder in their statistical analyses; and (7) well-designed multi-omics studies (i. e., metagenomics, m eta-transcri ptomics, meta-prot eomics and meta-metabolomics) are warranted to clarify the effect of PA and SB on the human-associated microbiomes. 4.1. Cross-sectional studies: physical activity and s edentary behavior (no n-at hletes) Microbiome diversity is considered a direct measure of gut health in humans, and a loss of diversity has been linke d to a higher risk of obesity, type 2 diabetes, and cancer, among others. 126 In this systematic review, four studies found that the gut microbiome of children and adults with higher PA levels showed higher alpha-diversity, compared to those who rarely or never exercised. 40,46,72,73 Similarly, two studies reported a positive association between PA level and gut alpha797 diversity in participants with different diseases. 47,74 A positive correlation between average PA intensity and vaginal microbiome alpha-diversity was also found in healthy college-aged women. 52 However, other studies in individuals with different age and health condition s reported negative or no associations, 37,39,41–45,48,50,53,70,75–77,84 as is also detected in our meta-analysis of 1814 participants (Fig. 3A). Heterogeneity in study population (i.e., health status, sex, age), methodological aspects (i.e., use of diverse self-reported questionnaires, different pipelines to analyze the microbiome, etc.), varying criteria to stratify participants based on PA level and lack of control of relevant covariates (e.g., diet) in statistical analyses may contribute to the discrepant findingsacrossstudies. Infact, Langsetmo et al. demonstrated different results depending on the method for measuring PA, where self-reported PA was positively associated with beta-diversity, 45 while objectively measured PA recorded by accelerometry (expressed as step counts) showed no associations. 45 I. Pérez-Prieto, A. Plaza-Florido, E. Ubago-Guisado et al. Journal of Science and Medicine in Sport 27 (2024) 793–804 Fig. 3. Panel A) shows the meta-analysis of the high-PA vs. low-PA level's effects on Shannon diversity and Chao1 indexes (i.e., alpha-diversity metrics). Eight studies were finally included (Shan non diversity index 42,44,46,53,73,76,84 ; Chao1 in dex 41,42,46,53,84 ). Panel B) indicates the meta-analysis of the athletes vs. non-athletes' effects on both alpha-diversity me trics, i.e., Shannon diversity (8 studies 42,56,57,59,67,82,84,85 ) and Chao1 indexes (9 studies 42,56,59,64,80,82,84,85,112 ). Panel C) shows the meta-analysis of the PA intervention (up) vs. control's effects (down) on the Shannon diversity (7 studies 102,104,105,107,116,119,125 ) and Chao1 index (3 studies 105,107,119 ). Due to the lack of studies, we included both RCTs 104,105,107,116,125 and nonRCTs 102,119 in the same meta-analysis. The bottom meta-analyses reflect the effect of time in the absence of PA intervention since it only includes the control groups that were available from the RCTs. We did not use the control groups of Cronin et al. 104 and Bielik et al. 125 , as they consumed a protein or probiotic supplement. Regarding SB, Bressa et al. reported that less time in sedentary bout s was positively associated with alpha-diversity (Shannon and Chao1 indexes) in premenopausal women. 39 In contrast, there were no significant differences in alpha-diversity when compared the gut microbiom e of physically active women (those who perform at least 3 h of PA per week) and sedentary women (i.e., those who perform <3 h). 39 Whisner et al. did n ot find any significant differences in alpha-di versity parameters acr oss quartiles of SB in a c ohort of college students. 41 However, a later study detected a lower alphadiversity in the saliva of children who reported high sedentary screen (sitting) ti mes. 71 Interestingly, recent evidenc e indicates a positive association between SB and Streptococcus, detected in feces and saliva. 70,71 Streptococcus has been described as a key bacteria in disease such as old -onset colorectal cancer. 127 The existence of an oral-gut microbiome crosstalk has been proposed, highlighting a possible association between oral microbial alterations, oral–gut microbiome axis and the pathogenesis of different diseases such as gastrointestinal disease or colorectal cancer. 128 Thus, more future research is needed to unravel the role of SB as a potential modulator of microbial communities. There are more consistent findings abo ut the associations between PA and the gut microbiome, mostly at lower taxonomic categories. At phylum level, Firmicutes seems to be more abundant in th e gut of those individua ls with higher PA levels, 40,50 although several studies foun d the inverse association. 53–55 Si nce Firmicutes has been associated with fiber, 129 different dietary habits may be partially explaining variabili ty between the studies. Inte restingly , growing evidence s upports that PA increases the abundance of a Firmicutes-belonging group of c ommensal bacteria able to produce SCFAs from non-digestible carbohydrates ingested through diet, such as butyrate, propionate and acetate. 130 Most of the included studies reported higher abundance s of SCFA-producing bacteria from Lachnospiraceae and Erysipelotrichaceae families, 40,41,51,70,73 and Roseburia, Coprococcus, Lachnospira, Blautia and Faecalibacterium genera, among others, in more active individuals compared to those with lower PA levels. 40,41,45,46,51,77 Particularly, Bressa et al. quantified 798 the relative abundance of Akkermansia muciniphila, Faecalibacterium prausnitzii and Roseburia hominis by real-time PCR (qPCR) and detected higher abundances in physically active compared to inactive women. 39 SCFAs have been linked to good health due to their role on metabolic function, being substrates for energy metabolism as well as important signaling molecules implicated in the gut-microbiota axis and in the regulation of the immune respons e. 131,132 Since the availability of SCFAs is influenced by bot h, the ingestion of nutritional co mponents and th eir digestion directed by the gut microbes, 130 the previous results coul d indicate SCFAs as the key molecular link between PA, diet and microbiom e. I. Pérez-Prieto, A. Plaza-Florido, E. Ubago-Guisado et al. Journal of Science and Medicine in Sport 27 (2024) 793–804 4.2. Cross-sectional stu dies: athletes vs. non -athletes Available evidence generally agrees on a trend toward the increase of the gut microbial diversity in athletes of different sport disciplines compared to non-athletes (see our meta-analysis of 651 participants; Fig. 3B). Further, a recent meta-analysis evaluated microbial alphadiversity of shotgun metagenomics data of the gut microbiomes of 207 athletes of different sports and 107 non-athletes and found a significantly higher species richness in athletes compared to non-athletes. 133 However, it is also well known that specific dietary requirements are usually implemented based on the duration and intensity of PA training, 134 which makes it difficult to determine the isolated effect of athletic performance on the microbial communities. In 2014, Clarke et al. reported, for the first time, a positive association between athletic performance and alpha-diversity parameters, when compared the gut microbiome by 16S rRNA sequencing of a group of professional rugby play ers and sedentary participants with low and high BMI (i.e., BMI ≤25 or >28, respectively). 56 However, the athletes' enhanced diversity was also associated with high protein consumption in this group. Barton et al. 57 re-analyzed the participants from Clarke et al. to evaluate the microbiome diversity with the shotgun metagenomic sequencing, confirming the previous results. 56 More recently, Penney et al. analyzed the combined effects of diet and athletic performance in th e gut microbiome of those participants, and found a significant association with alpha-diversity when combined the effect of both athletic performance and dietary habits. 67 Later studies described an enriched microbial diversity in athletes with special diets, compared to sedentary participants, 68,69 and others did not find any significant differences between athletes and non-athletes with similar dietary patt erns. 60,61,65 In contrast, 2 studies reported a higher alpha-diversity in athletes compared to sedentary participants with similar dietary habits. 59,64 Large variety of sports disciplines included in the abovementioned stud ies (marathon runners, bodybuilders, cross-country skiers, rugby players, etc.) can be also contributing to inconsistency of the results. So far, the isolated effect of athletic performance, independently of diet, is still unclear. Since diet is one of the most important modulators of the microbiome, differences in nutritional habits may also affect the relative abu ndance of sp ecific microorganisms. 135 In fact, high-digestible carbohydrate diets have been related to the growth of SCFA-producing bacteria. Clarke et al. reported a higher abundance of Firmicutes phylum and a decreased abundance of Bacteroidetes in rugby players compared to sedentary individuals with high BMI. 56 Both groups presented a distinct nutritional profile, with an increased consumption of protein, fiber, carbohydrate and monounsaturated and polyunsaturated fat in the athletes group. A later study also described a higher abu ndance of Firmicutes and lower levels of Bacteroidetes in rugby players compared to non-athletes. 82 Accordingly to these findings, animal and human studies have positively associated Firmicutes to fiber intake but negatively to fa t consumption, while Bacteroidetes showed the opposite association. 129 Additionally, later metabolic pathway analyses revealed that rugby players had an enriched profile of SCFAs. 57 Other SCFA - producer, F. prausnitzii, was also found tobe moreabundant in senior athletes compared to older sedentary participants after adjusting for different covariates, including diet. 65 Morishima et al. found an increase of Faecalibacterium in female runners compared to non-athletes, and a higher abundance of succinate, a SCFA th at can be produced by Faecalibacterium. 80 Liang et al. reported that professional martial art athletes had an enriched m icrobiome c ompared to amateurs, and identified changes in the abundance of several bacteria after adjusting for different confounders including diet. 63 Furthermore, one study found higher diversity and Firmicutes/Bacteroidetes ratio in female elite compared to non-elite athletes. 66 However, metagenomics and meta-transcriptomics analyses conducted by Petersen et al. only detected differences at transcriptomic (R NA) level, highlighting the need for more microbiome studies at func - tional level. 58 Few studies have identified significant microbial shifts in relation with the type of sport. 61,62,64 Interestingly, O'Donovan et al. compared athletes from 16 different sports and found specific bacterial taxa such as Anaerostipes hadrus, F. prausnitzii and Bacteroides caccae, differently abundant between sports with a moderate-dynamic component (e.g., fencing), high-dynamic and low-static components (e.g., field hockey), and high-dynamic and static components (e.g., rowing). 62 4.3. Acute effects of PA Most of the studies aimed to analyze potential changes in the gut microbial composition following a marathon. 25,88,91 In this sense, two studies detected an incre ase in Firmicutes/Bacteroidetes ratio of the gut microbiome in long-distance runners post-race. 91,93 Significantly, Grosicki et al. also detected a higher abundance of Veillonella, accordingly to the results obtained by Sche iman et al. 25,91 The last stu dy proposed a microbiome-encoded enzymatic mechanism that cou ld partially explain how microbiome and its metabolites (i.e., SCFAs) may contribute to enhance athletic performance. 25 After detecting a higher abundance of Veillonella in runners after the race, they observed that administration of Veillonella atypica in a mouse mode l improved run time and demonstrated its capability of metabolically converting the exercise-induced lactate into propionate in the colon to subsequently re-enter the systemic circulation. In search of confirmi ng these findings, Moitinho-Silva et al. quantified the relative abundance of V. atypica by qPCR and sequencing in a subset of elite athletes (mainly cyclists and triathletes) and sedentary participants, but failed to find any significant differences between the groups. 112 These contrasting results cou ld be partially explained by several limitations of the last study such as the lack of dietary data for the athletes group. Other studies have detected an increase in several SCFA-producing bacteria, including Coprococcus_2, Dorea or Roseburia after a marathon or a transoceanic rowing race. 88,90 Although the number of human studies is still limited, these findings supp ort emerging evidence of the existence of a crosstalk between the gut microbiota and skeletal muscle through lactate (generated during exercise) and its conversion to SCFAs by the gut microbes which, consequently, could improve athletic performance. 27 In fact, SCFAs have been recently defined as “biotics” (substances able to modulate the microbiome by increasing the abundance of beneficial microbes) that could be used as an exogenous microbiome modulation approach for improving health and athletic performance. 136 Interestingly, a recent study discovered a gut–brain connection in mice that enhances athletic performance by increasing dopamine signaling during PA . 137 These results indicate that motivation for PA is influenced by the gut micro bes derived-metabolites, s uggesting a microbiomedependent mechanism for explaining inter-individual variability in PA motivation and performance. On the other hand, the acute effect of a bout of PA on the microbiome continues to be a scarcely investigated topic. Tabone et al. followed this approach analyzing fecal samples from athl etes who underwent a moderate-intensity treadmill session until volitional exhaustion and detected change s in six bacteria (Romboutsia, Escherichia coli TOP498, Ruminococcaceae UCG-005, Blautia, Ruminiclostridium 9and Clostridium phoceensis). 92 Overall, acute interventions collect serum samples where 799 potential change s can be detected earlier compar ed to fecal ones . In this context, one study c ollected blood and fecal samples of myalgic encephalomyelitis/chronic fatigue syndrome participants and detected changes at major bacterial phyla such as Actinobacteria, Bacteroidetes, Firmicutes and Proteobacteria, in both samples after a cycle ergometer maximal exercise test. 87 A later study exclusively analyzed viral reads (i.e., virome) from blood samples by RNA-seq and did not detect any difference s after acute PA. 89 Thus, more research directed to analyze blood microbiome is needed to accurately assess the short-term effect of PA, specially, meta-transcriptomics and metametabolomics could be a novel and useful approach to study the active microbiome in the context of acute eff ects of PA. I. Pérez-Prieto, A. Plaza-Florido, E. Ubago-Guisado et al. Journal of Science and Medicine in Sport 27 (2024) 793–804 4.4. Chronic effects of PA To further deepen the overall knowledge of the chronic effects of PA on human microbiome and, generally, the host health, several clinical trials have been published in the last years. 24,95–125 Our meta-analysis of 167 participants is the first analysis that quantifies those trials in healthy participants, indicating that controversial results for alphadiversity are consistently found (Fig. 3C). Two studies performed in healthy adults that underwent a 12-week aerobic PA intervention (3 sessions of 30 min per week) or 7-week high-intensity interval training (consisting of swimming lengths) reported an increase in microbial alpha-diversity. 118,125 Conversely, Moitinho-Silva et al. detected a slight decrease in alpha-diversity after an aerobic PA intervention (6 weeks; 3 s essions of 30 min per week) in healthy adults, 112 although no differences were observe d in another group subje cted to a stre ngth training. 112 Another study recruited hea lthy adults to undergo a PA intervention (a erobic and resistanc e training; 8 weeks; 3 sessions of 90 min per week), but no significant changes in alpha-diversity were detected after the intervention. 104 Most of the studies in unhealthy individuals did not report any significant changes in alpha-diversity after a PA intervention. 24,95,96,99,100,102,103,110,114,115,121,122,124 However, the chronic effect of PA on microbiome composition becomes clearer in the beta-diversity analysis, where more studies agree on a significant dissimilarity in the microbial communities of the individuals after longterm PA. 24,95,98,101,103,106,109,110,115,118,119,121,122,124,125 Interestingl y, Allen et al. observed how differences in the beta-div ersity detected at baseline between the participants with normal-weight and obesity disappeared after an aerobic PA intervention (6 weeks; 3 of 30–60 min sessions per week). 24 Different study designs (17 non-RCT, 13 RCT and 2 rand omized controlled cross-over trial s), health sta tus of participants (15 studies with healthy and 17 with unhealthy populations), characteristics of PA interventions (type, duration, and intensity), and methodological differences in microbiome analysis, diet, among other factors, might partially influence the varying results obtained. In accordance with observational studies, 39,40,46,47,70,73,76,77,80 an incr ease in SCFA-producing bacteria such as Lachnospiraceae, Verrucomicrobiaceae, Lachnospira, Akkermansia, Veillonella, Faecalibacterium, Bifidobacterium and Roseburia was also reported in participants with different age and health conditions (including obesity, prediab etes and insu lin resistance, among ot hers) a fter PA interventions ranging from 2 to 34 weeks. 24,95,99–102,109,110,114,115,119,121,122 More specifically, Liu et al. described an increase in A. muciniphila and an improvement in insulin sensitivity after a 12 weeks-concurrent PA intervention in men with prediabetes that were classified as responders compared to non-responders. 100 Later studies have also reported an increase in A. muciniphila in participants with overweight/obesity or type 2 diabetes after long-term PA. 109,121 A. muciniphila has been related to prevention of multiple metabolic diseases such as obesity, metabolic syndrome and type 2 diabetes. 138 In a recent publication, a multi-omics approach (transcriptomics, proteomics, metabolomics and lipidomics) investigated the underlying molecular mechanism of A. muciniphila in obesity. It concluded that A. muciniphila re duced lipid accumul ation and downregulated the expression of genes related to adipogenesis and lipogenesis in adypocites. 139 These studies point to A. muciniphila as a promising microbial target (potentially m odulated by PA) with therapeutic effects in obesity and other metabolic diseases. At genus level, Akkermansia has also been widely found to be positively associated with PA in cross-sectional studies 39,56 and increased after PA interventions. 95,100 4.5. Future directions Our comprehension about the effect of PA (little research is conducted on SB) on microbial communities is still in it s infancy, which is partially attributed to significant methodological variability between the studies. This heterogeneity highlights the need to perform well-designed studies focusing on specific detailed populat ions and establishing reference pipelines to ensure the accuracy and comparability of the results . To ensure th e reproducibilit y and co mparability of the future studies in the field, we recommend the researchers to follow the recent good practice guidelines 140,141 when microbiome analyses are performed. Most of cross-sectional studies in this systematic review recorded PA measures by self-reported questionnaires. Accelerometry has been widely demonstrated to be a more valid and comparable method for objectively collecting participants' PA and SB levels. 142 Therefore, more accelerometry-based studies will allow researchers to apply standardized criteria to classify participants based on the use of cut-points for PA and SB which will reduce the inconsistency between study findings and reveal the accurate association of PA and SB with microbiome. In intervention studies that assess the chronic effects of PA on microbiome, we detected a low quality in the RCTs. These results could be partially explained by the use of a checklist 143 with a stricter scale for the quality assessment. Since most of the studies analyzed D NA sequence s regardless of microbial variability or functionality (only two studies performed a meta-transcriptomic analysis), we are not close to d etermine t he functional microbes susceptible to PA. Moreover, future multiomics analyses (i.e., combining metagenomics, metatranscriptomics, meta-proteomics and meta-metabolomics) would further unravel the complex host-microbial molecular pathways implicated in the molecular response to PA. In this regard, the Molecular Transducers of Physical Activity Consortium (MoTrPAC) 6 will provide a powerful source of information to advance our understanding of PA's effects on the microbiome in humans and animal models performing multi-omics an alyses. 4.6. Limitations and strengths Due to the lack of available information, an important limitation of our meta-analyses was the use and transformation of directly reported data from the articles instead of re-analyzing raw data to reduce potential bias introduced by applying different methodologies and pipelines across studies. Besides, limited information prevented us from additionally analyzing other microbiome outcomes of interest, such as the differential abundance of key bacteria. Future studies should make publicly available raw sequences generated from sequen cing platforms to allow future meta-analyses to cover these gaps in the literature. To our knowledge, we present the first meta-analysis conducted withi n non-athletes population, including more than 2600 participants from 24 studies. We also performed sub-group analyses according to study design to ensure homogeneity. Moreover, we follow ed a rigorous and reliable methodology previously validated 144,145 to obtain numerical data when they were unavailable. Additional strengths of our systematic review are the elaboration according to PRISMA guidelines, use of four different search databases (PubMed, Web of Science, SCOPUS and Cochrane), and performance of qualit y assessment with validated tools specific for each study design, which ensure the scientific rigor. 800