Bachelor thesis
Exercise and the microbiome
Health effects of exercise on gut microbiome modulation in healthy, prediabetic, and diabetic cohorts
Author: Linnea Brengesjö Supervisor: Anna Blücher Examinator: Britt-Inger Marklund Date: May 28th, 2021
Sammanfattning
Diabetes har orsakat många dödsfall världen över men kan bekämpas åtminstone delvis med hjälp av diet och fysisk aktivitet. Tarmens mikrobiom har visats korrelera med både typ 1 och typ 2 diabetes, har reglerande effekter på immunsystemet och inverkar på hjärnfunktioner genom tarm-hjärna-axeln, delvis via metaboliter från mikroberna.
Det har länge varit känt att mat kan påverka mikrobiomet, men träning har också väckt intresse över det senaste årtiondet med studier som fokuserat mest på möss och atleter med någorlunda positiva resultat för dess förändring av mikrobiomet. Denna litteraturstudie syftar till att undersöka om träning kan ha effekt på mikrobiomet hos såväl friska människor som prediabetiker och diabetiker, och vad detta kan betyda för hälsan. En litteratursökning gjordes i databasen PubMed i januari 2021 som efter sortering enligt inkluderande kriterier gav 7 artiklar för granskning. Dessa använde olika metoder, undersökta grupper och träningsupplägg, vilket försvårar jämförelser men indikerar i linje med tidigare forskning att träning påverkar mikrobiomet, med en del skillnader i resultat beroende på individens status och träningsupplägg.
Keywords
Microbiome, microbiota, gastrointestinal, exercise, diabetes, prediabetes, metabolites, SCFA, inflammation.
Abstract
Diabetes has caused many deaths worldwide but can be combated at least partially by diet and physical activity. The gut microbiome shows correlation with both type 1 and type 2 diabetes, has modulatory effects on the immune system and implicates brain functions through the gut-brain axis, in part by microbial metabolites. Diet has long been known to impact the microbiome but exercise has gained interest within the last decade, with studies mostly done on rodents and athletes with somewhat positive results on its modulation of the microbiome. This literature study aims to evaluate whether exercise can influence the microbiome for healthy, prediabetic, and diabetic cohorts and what this might mean for host health. The database PubMed was searched for articles in January 2021 and inclusion criteria yielded 7 articles for review. These differed in methods, cohorts, and exercise interventions, and therefore cannot grant any strong evidence but indicate along with previous research that exercise affects the microbiome, with slight differences in responses depending on the individual’s current state and exercising methods.
Acknowledgements
To my classmates, who added to the growth, fun, and enjoyment in this journey.
To all the teachers of this program, for your time, patience, and knowledge.
To Anna Blücher, for sharing your wisdom, in addition to the encouragement and support in helping me along the way.
To Britt-Inger Marklund, for your immense help with the editing and finishing touches.
To my parents, for your continuous support and help in every season, especially in enabling me freedom to study – in dad’s favourite music abode.
To my dear Prince Chacko Johnson. I would have never been able to do this without you. Your enduring encouragement, love, and support is so valuable.
To my Lord and Saviour Jesus Christ, who is my everything – my life, joy, and peace – and without whom I would have never been where I am today.
Contents
List of abbreviations ___________________________________________________ v 1. Introduction ________________________________________________________ 1 2. Theoretical framework _______________________________________________ 1 2.1. General overview _________________________________________________ 1 2.1.1. The microbiome _______________________________________________ 1 2.1.2. Diabetes _____________________________________________________ 2 2.2. Microbiome functions ______________________________________________ 2 2.2.1. Immune responses _____________________________________________ 2 2.2.2. Microbial metabolites __________________________________________ 3 2.2.3. Gut-brain axis ________________________________________________ 4 2.3. The microbiota composition _________________________________________ 5 2.3.1. Microbiota assessment _________________________________________ 5 2.3.2. Beneficial microbes ____________________________________________ 5 2.3.3. Diseases and dysbiosis _________________________________________ 7 2.3.4. Minority commensal microbes ____________________________________ 8 2.4. Impacting the microbiota composition _________________________________ 9 2.4.1. Diet, probiotics, and prebiotics ___________________________________ 9 2.4.2. Exercise _____________________________________________________ 9 2.4.3. Exercise-induced changes ______________________________________ 10 3. Study purpose _____________________________________________________ 11 4. Methodology _______________________________________________________ 11 5. Results ____________________________________________________________ 11 5.1. Overview of measurements ________________________________________ 11 5.2. Study 1: Motiani et al. (2020) ______________________________________ 12 5.3. Study 2: Liu et al. (2020) __________________________________________ 13 5.4. Study 3: Taniguchi et al. (2018) _____________________________________ 15 5.5. Study 4: Karl et al. (2017) _________________________________________ 16 5.6. Study 5: Morita et al. (2019) _______________________________________ 17 5.7. Study 6: Kern et al. (2020) _________________________________________ 18 5.8. Study 7: Pasini et al. (2019) ________________________________________ 19 5.9. Summary of methods and results ____________________________________ 20 6. Discussion _________________________________________________________ 23 6.1. General considerations ____________________________________________ 23 6.1.1. Microbiota assessment ________________________________________ 23 6.1.2. Diet _______________________________________________________ 23 6.1.3. Adherence __________________________________________________ 24 6.1.4. Test timings _________________________________________________ 24 6.1.5. Environment _________________________________________________ 24
6.2. Study questions __________________________________________________ 25 6.2.1. Exercise and the microbiota composition __________________________ 25 6.2.2. Differences between healthy, prediabetics and diabetics ______________ 25 6.2.3. Correlations with other measurements ____________________________ 26 6.2.4. Intensity and mode of exercise ___________________________________ 29 6.2.5. Health aspects _______________________________________________ 30 7. Conclusion ________________________________________________________ 30 References ____________________________________________________________ I
List of abbreviations
6MWD – 6-minute walk distance ACTH – adrenocorticotropic hormone AWE – aerobic walking exercise (study 5) BCAAs – branched-chain amino acids BDNF – brain-derived neurotrophic factor BIKE – bike commuting group (study 6) BMI – body mass index
CFUs – colony forming units CNS – central nervous system CRP – c-reactive protein DBP – diastolic blood pressure DNA – deoxyribonucleic-acid FFA – free fatty acids
GABA – gamma-aminobutyric acid GBA – gut-brain axis
GI – gastrointestinal GU – glucose uptake
HbA1C – glycated haemoglobin A1C
HIITG – high intensity interval training group (study 1)
HOMA-IR – homeostatic model assessment for insulin resistance HPA – hypothalamic-pituitary-adrenal
IBD – inflammatory bowel disease IBS – irritable bowel syndrome IFN-γ – interferon gamma IL – interleukin
IP – intestinal permeability IR – insulin resistance
LBP – lipopolysaccharide binding protein LPS – lipopolysaccharide
MET – metabolic equivalents
MICT – moderate intensity continuous exercising group (study 1) MOD – moderate intensity exercising group (study 6)
NPY – neuropeptide Y
OGTT – oral glucose tolerance test OTUs – operational taxonomic units
p – statistical significance ROS – reactive oxygen species rRNA – ribosomal ribonucleic acid SBP – systolic blood pressure SCFAs – short chain fatty acids
sp. – unclassified species of bacteria (singular) spp. – unclassified species of bacteria (plural) T1D – type 1 diabetes
T2D – type 2 diabetes
TM – trunk muscle training group (study 5) TNF-α – tumour necrosis factor alpha
VIG – vigorous intensity exercising group (study 6) VO2max – maximum oxygen uptake
VO2peak – peak oxygen uptake
1. Introduction
Diabetes has globally caused many deaths and, according to the World Health Organization, was the direct cause of 1.5 million deaths in 2019 [1]. Evidence is accumulating that links the gut microbiome to both type 1 [2,3] and type 2 diabetes [4,5], as well as prediabetes [2], partially through microbe metabolites and inflammatory responses [5]. To combat diabetes, recommendations often concern dietary intake and regular physical activity, along with maintaining a healthy body weight [1]. Though there is a lot of evidence of how dietary components can influence the microbiome [6], physical activity has not received much attention until the last decade or so and has focused on trying to identify effects in rodents and understanding the athletic gut microbiome [7], though some studies have evaluated the effects of exercise on other human populations [7,8]. However, exercise does not produce the same results in all humans and individuals respond quite differently, depending on genetics and other factors [9,10]. Though an interesting question can be raised regarding if there are implications of the microbiome on such individual effects of exercise, this study first and foremost looks into whether exercise affects the microbiome in healthy, prediabetic, and diabetic cohorts.
2. Theoretical framework
For the sake of clarity, this study follows proposed definitions by Berg et. al. (2020) based on the first definition of the microbiome from 1988 [11]. Therefore, ‘microbiota’
refers to the microbes themselves; ergo, the bacteria, fungi, archaea, protozoa, and viruses [11,12]. Whereas ‘microbiome’ includes the microbiota plus its microbial internal and external structural elements, and microbial metabolites [11]. All further usage of these two terms refers to that which is within the gastrointestinal (GI) tract.
2.1. General overview
An introduction to the microbiome is presented below, followed by general considerations of diabetes, before looking into more details in later sections.
2.1.1. The microbiome
Within the microbiota, the bacterial component is the dominant one [13]. This, as well as the entire microbial community, varies greatly in diversity between people even in the absence of disease and over time within the same person [12,14] The microbiome changes along with diet, environment, medicines, and disease states [12]. Though the microbiota has some hereditary influences it has been estimated that this only accounts for 12% of the composition, while dietary changes can alter the composition within 24 hours and account for up to 57% of the induced changes [15]. So, commensal microbes are impacted by both habitual dietary intake and acute dietary changes [16]. Pro- and prebiotics are also able to influence the microbiome [12]. Beyond that, stress has a profound impact on changes of composition [6]. Thus, the entire microbiome is dynamic and subject to a number of changes in the life of the host.
The microbiome is essential to human normal function [12] and contributes to vital developmental processes and functioning of the immune system, gut epithelium, gut- associated lymphoid tissue, brain, and more [12]. As the microbiome matures it moves from a rather unstable state to a more long-term temporal stability of an equilibrium, dominated by Bacteroidetes and Firmicutes phyla followed by Actinobacteria and Proteobacteria [17]. Though more stable, this state is still subject to change [12].
Microbes of the microbiota reside on the luminal side of the epithelial cells, within and on the surface of the mucus layer. In a healthy gut, the bacteria help prevent pathogenic penetration of the epithelium. The microbiome also contributes to host metabolism, endocrine signalling, synthesis of vitamins and nutrients, and regulation of immune function. Many of these interactions are through what is called local crosstalk networks, within which there are specialised microbial communities that vary along the length and width of the GI tract and the layers of the mucus. [12]
2.1.2. Diabetes
Diabetes Mellitus is divided into two main classifications. The first is type 2 diabetes (T2D), characterised by pancreatic β-cell dysfunction in production of insulin, in combination with a reduced ability of peripheral tissues to respond to insulin facilitation of glucose uptake into the cells, causing insulin resistance (IR) [18]. IR increases as the disease progresses and is related to insulin sensitivity, which indicates the effectiveness of a given insulin concentration [9]. The causes for T2D can be multifactorial, though often coupled with inflammation and being overweight [18]. The second classification is type 1 diabetes (T1D), an autoimmune disease where most of the pancreatic β-cells are destroyed, causing a critical lowering of insulin production [18]. Diabetes patients therefore have hyperglycemia, which can be measured by tests such as fasting plasma glucose, 2h plasma glucose during an oral glucose tolerance test (OGTT), and glycated haemoglobin (HbA1c) [18]. If these reach above normal but not hyperglycemic levels, it is known as prediabetes; impaired glucose tolerance [18]. Metformin, a very common diabetes medication, has also been suggested to exert some of its antidiabetic effects through modulation of the microbiome [19].
2.2. Microbiome functions
This section includes a short overview of immune responses, the gut-brain axis, and the roles of some important microbial metabolites.
2.2.1. Immune responses
Immunomodulatory effects are a very integral part of the microbiome, as the microbiota helps in early development of the immune system by ‘teaching’ it what to react to, and this system contributes to keeping the microbiota composition in check and is important for developing a ‘healthy gut’ [20]. The healthy gut has a functioning integrity of the intestinal wall, as the tight junction proteins between the endothelial cells only allow entry of small molecules such as ions, water, and leukocytes [15]. This junction consists of transmembrane proteins [15,21] and its integrity can mainly be affected by heightened HPA-activation, oxidative stress, and ischemia [15,22]. Even zonulin protein, an analogue to zonula occludens toxin [23,24], can cause disassembly of the tight junctions [25].
The condition of a ‘leaky gut’ shows loosening of the tight junction protein structures, allowing lipopolysaccharides (LPS) from the surface of gram-negative bacteria to translocate and bind to toll-like receptors, triggering an inflammatory response [5]. This causes release of pro-inflammatory cytokines like tumour necrosis factor alpha (TNF- α), interleukin (IL) -1β and -6, that can further increase tight junction openings and result in heightened intestinal permeability (IP) and endotoxemia [15]. Increased levels of cytokines such as IL-6, IL-1, and TNF-α induce production of C-reactive protein (CRP), which is an acute phase inflammatory protein measured in serum to detect infections and inflammation [26]. Furthermore, LPS is also released into the blood in levels correlated with IP and endotoxemia [5,15]. An alternative measure for this is LPS binding
protein (LBP) [27]. Even zonulin can be detected in stool samples, along with calprotectin, and correlates with levels of IP [24]. Calprotectin is found in neutrophils, monocytes, and macrophages with an essential function in inflammatory processes, where its release can be triggered by bacterial components so the concentration in faeces increases and mirrors the level of intestinal inflammation [28].
Zonulin release can be triggered by different factors such as gliadin in gluten and certain bacteria (e.g. Escherichia coli) [23], and its heightened levels have been shown to precede T1D in rat models [25]. Zonulin levels also seem to associate with insulin sensitivity, through its impact on IL-6, and correlates with obesity and T2D [23].
Activation of the hypothalamic-pituitary-adrenal (HPA) axis can occur in response to prolonged stress which causes systemic glucocorticoid release and suppression of inflammation, but also hinders the ability to produce anti-inflammatory cytokines and uphold microbial defence [15]. Such conditions in mice have caused long-term change in HPA functioning, elevation of IL-6, increases in Clostridium bacteria, and reduction in relative abundance of Bacteroides [20,29].
Oxidative stress, on the other hand, is an excess of oxidants compared to antioxidants and results in disturbed oxidation processes which could cause molecular damage [22]. This is often seen with heightened amounts of reactive oxygen species (ROS), a type of free radical oxidants that are created by exogenous and endogenous processes [22], that can cause β-cell deterioration [30]. Local gut ischemia is often seen in prolonged and highly intense exercise and can cause a type of oxidative stress by increasing production of ROS and pro-inflammatory factors like TNF-α, IL-6, interferon-gamma (IFN-γ), and IL-1β [15]. The microbiome has been suggested to affect the level of ROS and antioxidants by production of metabolites, through fermentation and other microbial processes, creating a link between the microbiota composition and oxidative stress [22]. 2.2.2. Microbial metabolites
Evidence suggest that metabolites from microbial metabolic pathways are associated with nearly half of the blood metabolites [31]. Produced metabolites from the gut are most likely in the hundreds and are released into the blood and lymphoid circulation [20]. These impact the immune system and brain function and makes the microbiome resemble an endocrine organ [8]. Examples are enzymes that regulate immune system functioning, branch-chain amino acids (BCAAs), short chain fatty acids (SCFAs), and gamma-aminobutyric acid (GABA) [8,32].
SCFA is an umbrella term for molecules such as acetate, butyrate, and propionate, produced by different bacteria through fermentation of mainly nondigestible carbohydrates [33], so the dietary composition will greatly impact the microbiota’s ability to produce these molecules [15]. The different SCFAs have specific functions such as being used as nutrients by colonocytes and the central nervous system (CNS) microglia, thereby supporting their maturation and functioning [8,34]. They can also activate the sympathetic nervous system and stimulate release of serotonin from the mucosa [15]. SCFAs can even improve insulin sensitivity [34], regulate glucose homeostasis [33], and lower inflammatory responses [8]. A higher production of SCFAs has been associated with protection against inflammation [8], through different functions that are molecule specific and includes inhibiting insulin stimulation of adipocytes to store fat and thus reduce adipose infiltrate that could increase inflammation [33]. SCFAs
also impact the colonic pH and thereby affects intestinal motility, permeability, proliferation of epithelial cells, and the microbiota composition [15].
Butyrate is the main energy source for colonocytes [8] and can improve the intestinal barrier by increasing function of tight junctions [15], plus enhance mucin production which protects the gut epithelium [7]. Moreover, butyrate has anti-inflammatory effects, in part through regulation of function and migration of neutrophils [15]. Some studies have also indicated a suppression of TNF-α and IL-6 by butyrate, as well as some direct effects on sites of inflammation [33].
Propionate is another energy source for colonocytes, which can prevent mucosal degradation during prolonged stress responses [15] and hinder lipopolysaccharide (LPS) induction of cytokines such as IL-6 [33]. Adipocytes and intestinal cells have specific G- protein receptors that propionate and acetate act as ligands for, resulting in decreased inflammatory responses [20]. Propionate can even reduce fat storage in liver and visceral areas [33]. Acetate also seems able to impact adipocytes by inducing their lipolysis, and elevated levels of acetate are inversely correlated with plasma insulin levels [33].
Other microbial metabolites are GABA and BCAAs. GABA is partly produced by the gut microbes [32,35] but mainly by pancreatic β-cells [30]. This molecule, a modified form of glutamate and the major inhibitory neurotransmitter in the brain [36], could also aid in control of glucose homeostasis and suppress inflammation [30]. The microbiota can also produce BCAAs, and higher circulating levels of these are correlated with decreased insulin sensitivity [37].
Therefore, the metabolic activity in all sites of the GI tract is impacted by microbiota diversity. The microbial metabolites even affect the brain through the gut-brain axis. [22]
2.2.3. Gut-brain axis
The gut-brain axis (GBA) is a bidirectional communication between the autonomic and enteric nervous system [15]. There are two main ways through which these communicate:
Along the vagal nerve that runs from the brain stem through the digestive tract, and through gut hormones (e.g. GABA, NPY, dopamine) and microbe produced molecules (e.g. SCFAs, tryptophan/serotonin) [15]. SCFAs have a great impact on the GBA and are able to activate vagal afferent receptors [7]. Butyrate and propionate can also cross the blood brain barrier, influencing gene expression and neurotransmitter synthesis [20]. Interestingly, the vagal nerve has mainly afferent nerve fibres with up to 90% of them communicating from the gut to the central nervous system (CNS) [22], showing the importance of input to the CNS from the gut.
As an example of this communication, microbe-free rodents have an abnormal stress response even after a mild stress exposure with exaggerated levels of corticosterone, ACTH, and lower expression of brain-derived neurotrophic factor (BDNF) [8,20]. BDNF is essential for the growth and health of the CNS but also regulates the GI tight junction proteins and thus aids in the integrity of the gut [8]. Decreased hippocampal BDNF levels are associated with anxiety, depression, irritable bowel syndrome (IBS), and inflammatory bowel disease (IBD) [8]. These levels can be increased with probiotic supplementation of Bifidobacterium but aerobic exercise also seems able to induce gut levels [8]. Rodent studies also reveal that removal of the vagal nerve will eradicate effects of probiotic supplements on the HPA axis and depression [8]. Considering this, the GBA is an important link from the microbiome to the brain and affects the rest of
the body. Though the mechanisms are not quite understood, a healthy and diverse microbiome communicates with the CNS to exert tighter control on the HPA axis and increase BDNF regulation to induce neural and GI growth [8].
2.3. The microbiota composition
This section reviews how microbiota assessment can be done, followed by some microbes that are deemed beneficial and some that tend to associate with diseases.
2.3.1. Microbiota assessment
Studies on the microbiome and microbiota composition in humans generally rely on tests of faecal samples. Most methods for assessing the microbiota look at specific microbial DNA or RNA regions that are amplified, sequenced, and then analysed based on similarity with reference microbial genomic databases [12]. The most common methods are metataxonomic sequencing that focus on regions of marker genes, often from the ribosomal RNA (rRNA) [38] such as the 16S rRNA for bacteria or 18S rRNA for eukaryotes [38]. These are taxonomy specific and cannot capture anything else from the microbiota than the targeted gene [38]. To determine a wider range, other methods need to be used such as metagenomic sequencing of random microbial DNA [38].
When metataxonomics are analysed, the reads are statistically clustered into operational taxonomic units (OTUs) according to sequence similarity and classification [38], and compared with existing databases [39]. The richness of the microbiota is then determined by the number of individual OTUs, where the relative abundance and richness of the total OTUs make up the microbiota diversity [12].
Alpha (α) diversity indicates the diversity within one unique sample, while beta (β) diversity refers to that between different samples. Furthermore, dysbiosis concerns changes in the microbiome composition that lead to altered microbe-host homeostasis.
This is often seen in diseases with a reduction in microbial species, or a change in the systemic or intestinal inflammatory environment, or an altered metabolic relation between the host and microbes. [12]
2.3.2. Beneficial microbes
Bacteria are the most studied within the microbiota, where the two dominating phyla are Firmicutes and Bacteroidetes, covering about 90% of the bacteria [8,40]. The Firmicutes phylum contains over 250 genera including Lactobacillus, Lactococcus, Lachnospira, Lactiplantibacillus, Eubacterium, Streptococcus, Clostridium, and Blautia [8]. Bacteroidetes has around 20 genera where Bacteroides is the largest [8]. There are several other phyla also: Actinobacteria (e.g. Bifidobacterium spp.), Proteobacteria (e.g.
Escherichia spp.), Verrucomicrobia (e.g. Akkermansia spp.), and more [41].
On phylum level, Firmicutes are mainly butyrate producers while Bacteroidetes tend to generate mostly acetate and propionate [42]. More specifically looking at different genera, Blautia, Bifidobacterium, Eubacterium, Lachnospira, and Lactobacillus are among SCFA producers [8]. Bifidobacterium and Lactobacillus are also examples of GABA producers [32]. Dominating bacterial species for butyrate seem to be Faecalibacterium prausnitzii, Eubacterium rectale, Eubacterium hallii and Ruminococcus bromii [33]. For a simplified overview of bacteria relevant for this study, with their phylogenetic tree, see figure 1 below.
Figure 1: Phylogenetic tree of bacterial divisions, inspired by Barko et al. [12]. This is far from a full phylogenetic tree and only meant to give a quick overview of bacterial strains relevant for this study, so for the sake of simplicity not every branch is filled, nor any branch filled to completion. *Eubacterium halli is the basionym for Anaerobutyricum hallii. **C. difficile was previously counted to Clostridium genus. ***R.gnavus will probably be shifted to a different genera soon. All information collected from NCBI Taxonomy [41] .
Some bacteria (e.g. Blautia coccoides and Eubacterium rectale) even have the ability to convert lactic acid into butyrate [7]. Similarly, lactic acid can be used by the Veillonella genus to produce propionate and acetate [43]. Bacteroides also produce SCFAs, where B.
xylanisolvens mostly acts on cellulose (e.g. xylan) for its production [44]. Akkermansia muciniphila, on the other hand, can metabolise mucin into SCFAs which then induces production of mucin [45].
In summary, the specific bacteria all have slightly different functions and parts to play in the microbiome. So even though it is tempting to define a healthy microbial composition, this is not an easy task [11]. Instead, is has been suggested that it is not merely the presence of certain bacterial species but an interplay of the entire environment to keep a functional equilibrium of the microbiome, which thereby results in a ‘healthy’ state [14]. However, numerous studies have been conducted in order to find correlations between different microbes and disease states.
2.3.3. Diseases and dysbiosis
Larger fluctuations to the equilibrium cause dysbiosis in the microbiota composition and can result in an inflammatory state, which may be a factor in developing diseases such as T2D, IBS, cardiovascular diseases, autism spectrum disorder, and even allergies, mood disorders, and constipation [8,20,46]. Generally, a reduced diversity and richness of the commensal bacteria is associated with higher adiposity levels and insulin resistance [8], where a low α-diversity is associated with weight gain [47].
On phylum level, a lower abundance of Bacteroidetes with increased ratio of Firmicutes is correlated with obesity, IBS, T2D, and altered blood glucose levels [7,8], thereby increasing the likelihood of releasing LPS and inducing inflammation [48]. Proteobacteria are also associated with IBD and inflammation [49], and able to produce endotoxins [50]. Elevations in LPS levels have been found in T2D [12] and could, at least in part, explain why alterations in gut permeability have been seen in several diseases characterised by low grade inflammation [29]. Similar effects on this ratio of Firmicutes:Bacteroidetes also tend to surface in those on a high-fat diet [48]. But this shift in ratio is possible to counteract through mere weight reduction [20]. Interestingly though, such a relative increase in Firmicutes tends to be present in elderly individuals around the time the immune system starts declining, though the underlying mechanisms are unknown [51].
On genus and species levels, obesity and inflammation seem to correlate with reductions in Bacteroides and increases in Alistipes [52], where increases in the species A. shahii also correlate with T1D [2]. Patients with T2D and systemic inflammation also show reduced abundance of butyrate producing species within both Firmicutes and Bacteroidetes [33]. Considering insulin resistance, Prevotella copri is one bacteria with an ability to synthesise BCAAs, so its abundance may correlate with IR [53]. On the more positive side, Akkermansia muciniphila is negatively correlated with obesity, diabetes, and systemic inflammation [45], while Oscillospira abundance is positively correlated with leanness [54,55].
Evidence can appear contradicting however, such as for Streptococcus mitis which has been suggested to correlate with T1D in young children [56] but seem to protect against carcinogenic pathogens [57]. Also, even though Ruminococcus gnavus is normally present in a healthy gut with around <1% of the bacterial load [58] and can aid in germ- free mice growth impairments [59], its increased abundance correlates with inflammatory
diseases and induces release of TNF-α [58,60]. Increases in R. gnavus tend to also associate with a decrease in F. prausnitzii [60]. Even the genera Blautia can be elevated in patients with T1D [2,3], despite favourable aspects mentioned in the previous section.
Some bacteria are found to correlate with other diseases such as colorectal cancer, where an increased relative abundance of the Fusobacteria phylum is seen [61]. But Bacteroides xylanisolvens seem able to aid in anti-cancer treatment and responsiveness to this treatment can correlate with increases in Prevotella copri and Alistipes spp., along with higher Bacteroidetes relative abundance over the Firmicutes phyla [62]. One Firmicutes species, Clostridioides difficile (formerly known as Clostridium difficile
[63]), is known for its infections [63] and one risk factor for this infection is obesity [64]. There are several such opportunistic bacteria that can cause infections and stimulate immune responses [65], along with some fungi [66].
Fungi, such as Candida, are often identified in obese individuals and more abundant in both T1D and T2D, though more prevalent in type 1 [67]. T1D patients also tend to have increased levels of Escherichia coli, and Enterobacteriaceae but reduced abundance of Bifidobacterium [67].
2.3.4. Minority commensal microbes
Because the focus of microbiota studies has long been on bacteria and current methods are limited for studying the minority inhabitants such as fungi, viruses, and protozoa there is probably much we do not know about their roles yet. These are suggested to impact the entire microbiome and the intestinal environment through interkingdom interactions [13].
Fungi, mainly yeasts, have received some attention over the last few years and are suggested to be very important for the microbiome and may be far more sensitive to changes than the commensal bacteria [39]. This could be due to fungi feeding on carbohydrates, which would explain the strong link between Candida abundance and recent carbohydrate consumption [39]. Interestingly, Candida abundance seems inversely correlated with Bacteroides bacteria [39]. However, it appears as if lower diversity of the fungi community is associated with host health, rather than greater diversity [13]. Studies on rodents have also shown how fungi can impact inflammatory processes by either contributing to maturation of the immune system, or by promotion or inhibition of TNF- α and IL-6, or produce neurotransmitters (e.g. norepinephrine and histamine) and thereby impact the GBA [13,67]. One of the most studied fungi is Candida albicans, which in higher abundance is associated with diseases and could prevent microbiome recovery after disturbance of the equilibrium [13].
The viral gut inhabitants is another very recent and unexplored field, though their presence in the gut has been known for over a century [17] with estimated numbers that could be far more than bacteria [68]. Similarly to fungi, viral richness and diversity is associated with disease states but they are still important in controlled numbers [68]. Bacteriophages can impact bacterial genetics by functioning as vehicles for gene transfer, impacting bacterial evolution, diversity, and metabolism, and thus produce beneficial or disadvantageous changes in host risk of disease [68].
Considering archaea, the phyla Euryarchaeota could be richer in the microbiota than previously thought but their involvement in disease is rather uncertain [69]. So in
conclusion to this section without even looking into protozoa, there is definitely much uncertainty but also a cross-kingdom interaction that could be key to understanding the microbiome as a whole and its impact on host gut and body health.
2.4. Impacting the microbiota composition
What can be done to impact the microbiota if a profitable microbiome profile is our aim? This section very briefly answers how dietary components can be influential, and then turns to the main focus for this study: exercise. The concept and physiological effects of exercise are shortly reviewed, followed by its microbiotic effects.
2.4.1. Diet, probiotics, and prebiotics
Diet has a great impact on the microbiome. As mentioned earlier, the microbes of the gut can ferment components in the diet. For instance, a greater butyrate production is associated with low-fat, high-fibre diets [33], while high-fat diets can reduce relative abundance of Bifidobacterium, Lactobacillus and Bacteroides genus [48].
Fibre is one component majorly discussed on the topic of impacting the microbiome and many of them, though not all, can be classified as prebiotics [16]. Prebiotics is a non- digestible compound that, when metabolised by the microbiota, can have beneficial physiological effects on the host by altering microbiome composition or activity [16]. These tend to generally favour Bifidobacteria and Lactobacilli growth and may reduce adverse GI symptoms and levels of CRP [20].
Probiotics on the other hand, are “live microorganisms that, when administered in adequate amounts, confer a health benefit on the host” [70]. These can, dependent on the bacterial contents, reduce levels of TNF-α, IL-6, increase plasma concentration of tryptophan to possibly aid in serotonin synthesis, and help improve intestinal permeability [15,29]. Some strains may also lower incidence of respiratory tract illness in athletes [20] and are favourable for exercising individuals by influencing immune function, plus proliferation, function, and protection of intestinal epithelial cells [15]. For example, exercise-induced IP can be somewhat improved by supplementing probiotics, because of anti-inflammatory effects [71].
2.4.2. Exercise
With our attention turned to exercise, it is important to note that the nature of exercise is of either aerobic (typically endurance) or anaerobic type (typically strength/resistance training, or endurance at very high intensity), depending on the individual’s capacity for the exercise and thus determined, not necessarily by mode but, by which metabolic system is used within the active muscles. [10]
General physiological effects of exercise can be divided into acute and chronic effects, seen either during, immediately after, or weeks and months after regular exercise. Some acute effects of endurance training are redistribution of blood flow, heightened concentrations of by-products within the blood from muscles, and increases in cardiac output. The latter leads to a heightened maximal oxygen uptake (VO2max) and is induced alongside exercise intensity. This intensity along with mode, duration, and specificity strongly affects the chronic adaptations seen that are very individual and connected to heredity. However, these changes are reversible and must be maintained. [9]
Even though greater amounts of physical activity can give greater health benefits [9], adverse effects with injuries and negative health outcomes can occur if progression
causes excessive overload [10]. However, if there is no progression with increase in intensity or amount of exercise, no improvements will be seen [10]. Both endurance and resistance training can therefore yield health benefits if the individual starting point is respected and the intensity, duration, or amount is progressed over time [9,10].
2.4.3. Exercise-induced changes
Studies on rodents have yielded results that indicate exercise as able to exert protective effects on gut integrity, even in the presence of a high-fat diet [72]. The strains of gut bacteria that indicate the largest responses to exercise are within the Firmicutes phylum
[8]. Endurance exercise has resulted in shifts of the microbiota composition, with increased abundance of lactic acid and butyrate producing bacteria [73]. Moderate treadmill exercise for 30 minutes, 5 days per week for 4 weeks, was enough to change the microbiota composition with increases in bacteria such as Lactobacillus in non- obese, obese, and hypertensive rats [20]. Even short voluntary exercise with 5 days of wheel running increased Lactobacillus, Blautia, and Bifidobacterium in rodents [8]. Prolonged and strenuous endurance exercise in both humans and rodents has been shown to induce IP, which might be explained by reduced blood flow. A study on rats found that repairment of the GI epithelial barrier started after 3 hours of reperfusion, and was fully recovered after 24h, though the microbiota shifts did not recover even after 72h [20]. Similarly, adverse effects have been seen in correlation to endurance exercise in humans, where increased arterial resistance and sympathetic nervous system input reduced gut blood flow to such extents that prolonged exercise gave toxic effects and increased IP [7]. Related results were seen in another study of trained triathletes, where 93% of the participants had GI disturbances after completion of a race, most likely explained by gut ischemia and loosening of tight junctions [15]. Endurance athletes who exercise at very high intensities for prolonged periods of time tend to have GI disturbances coupled with strongly elevated concentrations of LPS in blood [15]. Exercise-induced stress can also induce levels of glucocorticoids and pro-inflammatory cytokines, and therefore result in increased IP and lowered production of serotonin [15]. On the other hand, Cook et al. found a more anti-inflammatory profile in endurance exercising mice, where exercise was associated with protection against influenza infection [73]. They also discovered reductions in inflammatory signalling molecules such as TNF-α, IL-1, and IFN-γ, plus elevations in anti-inflammatory cytokines IL-10 and -4 [73]. However, there are indications to suggest that voluntary exercise grants a more favourable response and shift of the microbiota with reduced GI symptoms, compared to forced exercise which seems to exasperate inflammation especially if coupled with food restriction [74,75]. Voluntary exercise in rodents has also resulted in higher faecal levels of SCFAs, than the sedentary control group that ate slightly more of the same diet than the exercising rats [76]. Lower or medium intensity exercise could also increase serotonin levels in the brain and lower depression and anxiety [15].
In a study from 2016, Estaki et al. [77] found a positive correlation between VO2max and faecal SCFA levels, indicating a higher production rate in fit humans. The same study also linked VO2max to 20% of the microbiota composition, granting a favourable diversity even when controlling for other factors, including diet. Furthermore, Allen et al. [34] revealed that progressive aerobic exercise for 3 days per week during 6 weeks increased SCFA production, coupled with greater microbiota diversity, some of which correlated with changes in body fat and VO2max but independent of diet. Most of these
adjustments were later reversed after detraining, indicating a similar effect as other chronic physiological changes; lost if not maintained.
In summary, the results so far are somewhat positive, even though mechanisms behind these changes are not quite understood.
3. Study purpose
The intention of this study was to evaluate whether physical exercise can influence the gut microbiome in healthy, prediabetic and diabetic individuals, and what this might mean for host health. This study also attempts to answer these specific questions:
(1) Does exercise influence the microbiota composition?
(2) Does the microbiome modulation differ for healthy, prediabetic, and diabetic cohorts?
(3) Is the microbiota modulation associated with either metabolites (SCFAs, GABA, or BCAAs), inflammation, glucose profile, cardiorespiratory fitness, or fat loss?
(4) Is there a particular form of exercise, in intensity or mode, that shows a more favourable response in the microbiome?
4. Methodology
For this literature study, the database PubMed was searched for articles in January 2021.
To limit the articles to those focusing on the subject, and therefore specifically using the word exercise along with either microbiota or microbiome in their title and/or abstract, these exact words were used for the search: “(Exercise[Title/Abstract]) AND ((Microbiota[Title/Abstract]) OR (Microbiome [Title/Abstract]))”. Filters used were:
“Full text” and “Clinical trial”. Some articles were not free on PubMed but instead accessed through other databases. The search yielded 34 articles, out of which 27 were excluded due to lack of relevance for this study as they did not examine the impact of exercise on the gut microbiome. Inclusion criteria were human participants that were deemed healthy or diagnosed prediabetic or diabetic, plus including some form of structured and planned physical activity, and assessing faecal samples for microbiota sequencing or culturing to evaluate the effects of the exercise on the microbiome. The final collection consisted of 7 articles. In a few of these, partial methods and results were referenced by the authors to a separate publication. These have been screened and their content included in the results here, to the extent of relevance for this study. In some studies there were more arms included after the original intervention (e.g. faecal transplantation to mice, or extra control cohort) that have been excluded for this study.
5. Results
Data from the seven different experimental studies is presented below. Some of these studies were difficult to decipher and results can therefore have been misinterpreted in some places, despite attempts to avoid this. To clarify methods, measurements, and results, tables are presented at the end. But first, a short explanation of some relevant measurements within the studies is provided.
5.1. Overview of measurements
Several of the studies presented values of the homeostasis model assessment of insulin resistance (HOMA-IR), which is calculated based on measurements of fasting plasma insulin concentration and fasting plasma glucose [78]. Therefore, whenever HOMA-IR is mentioned those tests are assumed.
One study reported Matsuda index, a value of whole-body insulin sensitivity calculated by results of an oral glucose tolerance test (OGTT), which increases when improved [79]. Some studies reported VO2peak as the measurement for cardiorespiratory fitness, which is often used if a maximal and true plateau is not met, and therefore VO2max is replaced with VO2peak [80]. One study used an alternative of estimating metabolic equivalents (MET), where intensity is generalised to different values for the given activity [9].
Body mass index (BMI), as reported from these studies and in table 1, was measured in kg/m2 but for simplicity this unit of measurement has been left out below.
Statistical calculations were done in all studies with confidence interval at 95% with significance level (p) ≤0.05, even though this was implied but not stated in all studies.
Only one study reported adjusted significance levels using the Benjamini-Hochberg procedure, with significance level reported as Q≤0.10. Another study used other means of reporting significance in comparing correlations with microbial species, and this index was compared in their results to the regular p<0.05 and therefore is reported below as ‘significance’ along with its value.
5.2. Study 1: Motiani et al. (2020)
Title: “Exercise training modulates gut microbiota profile and improves endotoxemia.”
[81]
Purpose: To examine the effects of short-term exercise on insulin-stimulated glucose uptake, fasting free fatty acid uptake, gut microbiota composition, and metabolic endotoxemia in prediabetic and T2D subjects.
Methods: Eighteen sedentary males and females with IR and diagnosed T2D or prediabetes were recruited in Finland. They had an average BMI of 29.9 and age between 40-55 years with no previous regular exercise training. Exclusion was done for any disease that might interfere with interpretation of results or endanger the subject’s health during the trial. Glucose and anthropometric profiles were assessed before and after intervention, including whole body fat, visceral and subcutaneous fat mass, BMI, blood pressure, glucose uptake (GU), insulin levels, whole body insulin sensitivity, and HbA1c. On screening day, OGTT and VO2peak tests were done, along with a physical examination by a medical doctor to rule out any individual health risk for participation in the trial. VO2peak was defined according to a 1-min mean value of oxygen consumption when cycling to exhaustion, and did not reach above 40. Faecal samples were taken the day after screening. Two days from screening day, intestinal GU was assessed. The opposite order of tests was repeated post-intervention.
Participants were randomised into two groups with different types of exercise programs:
High intensity interval training group (HIITG), or moderate-intensity continuous training (MICT). Both groups exercised 3 times per week for 2 weeks, in a supervised and progressive fashion. HIITG consisted of 30 seconds all out sprints on a stationary cycle with 4 minutes recovery between each. This group was familiarised with the training (two sprints) one week pre-trial. During the intervention, the number of sprints were increased from 4 to 6, with an added sprint every other session, and stayed at the top number of 6 sprints for the last few sessions. The MICT group did 40-60 minutes of cycling at 60% of VO2peak and increased in a similar fashion from initial 40 minutes to
60 by adding another 10 minutes every other session. All participants were advised to continue with their habitual diet and to abstain from caffeinated and alcoholic drinks but no control or assessment of dietary intake was mentioned.
Other measurements done were levels of interleukins, IFN-γ, TNF-α, LBP, CRP, calprotectin, and zonulin. Interleukin levels that did not exceed detection limits were excluded in the analysis. The authors conducted sequencing of 16S rRNA (v3-v4 region) with OTU cut-off value at 97% similarity, using the MiSeq-Illumina platform.
Medication levels were also considered in the statistical analyses.
Results: After the intervention both groups reduced 0.45kg in weight and 0.22 in BMI, with no significant difference between the groups (p=0.51 and p=0.50, respectively).
Whole body fat percentage and visceral fat mass reduction were significantly reduced after intervention (p=0.04 for both) but without any significant difference between groups (p=0.91 and p=0.18, respectively). The only glucose profile measurement that changed was HbA1c (p=0.003) but with no difference between groups (p=0.056).
VO2peak only improved after HIITG training (p=0.03). Inflammatory markers TNF-α and LBP were reduced in both training groups (p-values 0.03 and 0.02, respectively), with only a tendancy of reduction in CRP (p=0.08). No significant changes were observed in interleukins, cytokines, calprotectin, or zonulin.
Gut microbiota composition changed in both groups with increased relative abundance of Bacteroidetes (p=0.03) but no change in Firmicutes, which resulted in a decreased ratio of Firmicutes:Bacteroidetes (p=0.04). Both groups decreased abundance in Clostridium spp. (p=0.04) and non-significantly in Blautia spp. (p=0.051). HIITG increased in Lachnospira (p=0.025) while MICT increased in Veillonella (p=0.036). No significant difference was observed in richness or diversity in either group.
In the authors’ statistical analysis, a negative inverse association was seen at baseline between insulin-stimulated colonic GU and abundance of Firmicutes (p=0.03), Firmicutes:Bacteroidetes ratio (p=0.024), and Blautia (p=0.049). Similarly, GU had a positive correlation at baseline with abundance of Bacteroidetes (p=0.007). Whole-body insulin sensitivity was negatively associated with abundance of Blautia genus (p=0.04).
LBP also showed a positive correlation with reduction in HbA1c (p=0.02).
5.3. Study 2: Liu et al. (2020)
Title: ”Gut microbiome fermentation determines the efficacy of exercise for diabetes prevention.” [82]
Purpose: To explore the roles of differently shaped gut microbiota by exercise in glucose metabolism and insulin sensitivity.
Methods: Thirtynine overweight and obese Japanese men with prediabetes and/or impaired fasting glucose, were recruited in this study, with an average age of 43.8 years and BMI 29.2. All individuals were examined to rule out systemic, metabolic, or cardiovascular diseases, as well as infections, or use of antibiotics or probiotics. Before intervention, anthropometric measurements were done along with an OGTT, a 2h glucose test, faecal samples collection, assessment of VO2max, and HOMA-IR. Same tests were repeated 48-72 hours post-intervention.
For the intervention, all participants were randomised into either the exercise group (n=20) or a sedentary control group (n=19). The trial lasted 12 weeks, with a supervised exercise program at 80-95% of maximal heart rate. This included warmup, a treadmill station with intervals of 3-4 bouts at 85-95% of VO2max and active rest for 30-45 seconds in between sprints, followed by high intensity resistance training focused on large muscle groups, and then stationary bike intervals of 4-5 bouts lasting 45-60s while reaching 90-95% of VO2max with active recovery in between sprints for 60-75s, before finishing with cooldown and stretching. The exercises were progressively adjusted, to keep up individual intensity during the 12 weeks. For inclusion in results an adherence level of 85%, of the total number of exercise sessions, was required. All participants were instructed to not change habits regarding diet nor physical activity level, apart from the prescribed exercise, which were assessed once a month with questionnaires and one-on-one interviews, including food frequency questionnaires.
Levels of CRP were also measured, but no other markers of inflammation. Microbiota profiles were assessed by shotgun metagenome sequencing and compared to the MiDAS database with a threshold for 95% similarity. Metabolites were also measured in stool samples, including BCAAs, SCFAs, and GABA.
Results: Participants in the exercise group increased VO2max with nearly 22% (no p- value recorded), and reduced on average 1.28kg in body weight (p=0.003) and 0.14 in BMI (non-sign.). They also had a strongly significant change of fat mass (-2.21%, p<0.001) and lean mass (+2.16%, p<0.001). The sedentary group only showed negligible and non-significant anthropometric changes. Strong improvements were seen in insulin sensitivity in the whole exercise group, as measured by levels in fasting glucose (p<0.001), 2h glucose (p<0.001), and fasting insulin level (p=0.02), resulting in significant changes of HOMA-IR (p=0.031). Diet changes were negligible within both exercise and sedentary group, with only a slight and non-significant increase in fat intake in the exercising group. Both groups reduced intake from carbohydrates but non- significantly. Levels of CRP increased slightly but non-significantly in the sedentary group, yet were reduced by almost 30% in the exercising group (p=0.002).
The authors noticed a high individual variability within the exercising group, in changes of insulin sensitivity as measured by Matsuda index, which divided these individuals in groups of “responders” and “non-responders”. These two groups showed no significant differences in anthropometric changes or diet but in fasting insulin, Matsuda index, and resting heart rate change after intervention (p=0.03, p=0.017, and p=0.024, respectively;
measuring group difference in relative change). No intergroup difference was seen in diet or CRP levels of the responders and non-responders.
In the microbiota assessment, the significant changes seen in the exercise group were in relative abundance of six species within the Firmicutes, Bacteroidetes, and Proteobacteria phyla. However, there was a significant segregation between responders and non-responders, with the latter showing a much more similar profile to that of the sedentary group (p<0.001) after intervention. Only responders had an increase in Streptococcus mitis (p=0.036), reduction in Bacteroides xylanisolvens (p=0.049) and Alistipes putredinis (p<0.001), with the latter having been increased in non-responders (non-sign.) and sedentary controls (p-value not recorded). Also, responders had a reduction in Alistipes shahii (p=0.0079), which in contrast was increased amongst non- responders (p=0.031). Responders also had a reduction in Prevotella copri (p=0.024) and increased in abundance of Lachnospiraceae (p=0.036). Non-responders also had
decreases in Ruminococcus gnavus (p=0.031) and Escherichia coli (p=0.031), apart from the changes already mentioned. Akkermansia muciniphila also decreased after exercise in responders (p=0.0064), while E. coli was almost three times more abundant in non-responders at baseline (p-value not recorded).
Due to these differences, there was a significant association between microbiota profiles and the percentage of reduction in HOMA-IR (p<0.01). The greatest correlations with HOMA-IR were noted for A. shahii, S. mitis, and E. coli (significance < 0.05), while R.
gnavus and Eubacterium hallii had a greater correlation with Matsuda index (significance < 0.05) All these remained significant after controlling for body weight and adiposity levels.
The responders also showed a different metabolomic profile after intervention, with an increase in butyrate (p<0.001), propionate (p=0.006), and GABA (p<0.001) while non- responders reduced in butyrate (p=0.004), creating a significant relative change between the groups for butyrate (p=0.001). Responders decreased in amounts of BCAAs (p=0.005) which remained unaltered in non-responders, with a significant relative change between the responders and non-responders (p=0.035).
5.4. Study 3: Taniguchi et al. (2018)
Title: ”Effects of short-term endurance exercise on gut microbiota in elderly men.” [83]
Purpose: To evaluate whether endurance exercise modulates the gut microbiota in elderly subjects and if such changes associate with cardiometabolic phenotypes.
Methods: Thirty-one healthy Japanese men, between 62-76 years of age with a consistent lifestyle, diet, and body weight in the most recent decade, were recruited and finished this trial (two extra individuals dropped out). They had an average BMI of 22.9. Exclusion criteria were use of pre-/pro-/symbiotics, any present or previous GI- disorder, diabetes, cardiovascular disease, cancer, chronic kidney disease, autoimmune disorder, or previous participation in nutritional or exercise studies. Out of the original 33 participants, two had high levels of fasting glucose at baseline, six were taking antihypertensive drugs, and four had lipid-lowering medication. None of the medication administration statuses were changed during the intervention. All tests were done pre- trial, after 5 weeks, and post-intervention. These tests included assessment of VO2peak, blood pressure, anthropometric measurements, and HbA1c. Measurements post-trial were done 3 days after the last exercise session and without having taken medication that same morning.
The intervention was a 5 weeks, randomised cross-over trial. Participants were assigned to either endurance exercise or a sedentary control period at baseline, and switched after 5 weeks. For the exercise program, three supervised sessions per week on cycle ergometer were done with progressive intensity. The first week of cycling consisted of sessions with 30 minutes at 60% of VO2peak. For the second week the intensity was increased to 70% but same duration of sessions. The third week duration was increased to 45 minutes, yet with same intensity as the week prior. For the final two weeks, duration stayed at 45 minutes but intensity increased to 75% of VO2peak. Participants were instructed to maintain habitual diet and physical activity. To control for diet history and changes, food frequency questionnaires were taken prior to intervention, at 5 weeks, and post-trial. Physical activity was also assessed by questionnaires (apart from assigned exercise).
No inflammatory markers were measured in this study. Faecal samples were analysed for microbiota assessment by sequencing of the 16S rRNA gene (v3-v4), with OTUs at 97% cut-off value and compared to the Greengenes (2013) database.
Results: No change was seen in weight or BMI during either control or exercise periods. Slight changes in fat mass occurred, with a negligible increase during control period and similar reduction during the exercise program, but only intrahepatic fat percentage changed significantly (p=0.046). VO2peak was improved after exercise period (p<0.001) but reduced during sedentary period (non-significantly). Both systolic and diastolic blood pressure (SBP and DBP) reduced during all 10 weeks but with a larger reduction after exercise, though non-significantly. When comparing difference between the control and training period, both fasting glucose and HbA1c reduced significantly after exercise (p=0.004 and p<0.001, respectively). Fasting insulin did not show any statistically significant changes. Some dietary changes occurred such as greater intake of light-coloured vegetables after 10 weeks compared to baseline (p=0.015). Rice and seaweed intake also increased after 10 weeks and the mid-trial 5 weeks mark (p=0.01 and p=0.004, respectively).
The microbiota assessment uncovered neither a significant change in relative abundance at phylum, class or order levels, nor any changes in α- or β-diversity. But a significant decrease in relative abundance of C. difficile occurred after the exercise period, even when controlling for dietary changes (p=0.035). Also, increases were seen in Oscillospira after exercise (p=0.003), though this statistical significance disappeared after controlling for diet (p=0.236).
Differences in α-diversity correlated negatively with changes in SBP and DBP (p=0.02 and p=0.008, respectively). The changes in C.difficile also significantly and positively correlated with changes such as visceral fat (p=0.02), SBP (p=0.022), and HbA1c
(p=0.016) but negatively with VO2peak (p=0.007). Relative abundance of Oscillospira also correlated negatively with HbA1c (p=0.004) and body fat percentage (p=0.008).
5.5. Study 4: Karl et al. (2017)
Title: ”Changes in intestinal microbiota composition and metabolism coincide with increased intestinal permeability in young adults under prolonged physiological stress.” [84]
Purpose: To study the effects of physiological and metabolic stress on IP and intestinal microbiota composition, and identify associations between dietary intake, IP, inflammation, and the intestinal microbiota.
Methods: In this study, 73 Norwegian army soldiers (71 men, 3 women) over age 18 were recruited. Only 26 of these participants provided both pre- and post-trial faecal samples for microbiota assessment. No specific inclusion criteria were mentioned, apart from being stationed in the chosen military program and over age 18. Average age was 19.7 and BMI 23.7. Stool samples were taken during the 2 days prior to the start of exercise and within 36 hours of completion. Anthropometric measurements were done pre- and post-exercise, along with blood sampling.
The intervention consisted of a four-day arctic (skiing) military exercise, where all participants were block-randomised into groups depending on their weight, and
