How Gut Bacteria Relate to Muscle Strength and Movement in Older Adults

We performed a search of NIH/PubMed, MEDLINE (Ovid), Embase and Scopus using the following search string: (“microbiome” OR “microbiota” OR “gut microbiome” OR “gastrointestinal microbiome”) AND (“elderly” OR “aged” OR “aging” OR “ageing” OR “older persons”) AND (“sarcopenia” OR “skeletal muscle” OR “muscle mass” OR muscle strength” OR “muscle atrophy” OR “grip strength” OR “gait speed” OR “physical performance” OR “muscle quality” OR “lean mass” OR “frailty”) AND (“prebiotic” OR “probiotic” OR “synbiotic” OR “symbiotic”). Exploded MeSH or thesaurus terms were used where possible. Studies were imported into a reference manager (Endnote X9™) for deduplication.

The above keyword search terms were used on two separate occasions as below:

Search 1 (20 September 2021): for the period between January 2004 and September 2021. This time interval reflects next-generation sequencing techniques for gut microbiota.

Search 2 (3 July 2023): for the period between September 2021 and July 2023.

Therefore, all articles published between January 2004 and July 2023 were considered. Grey literature sources were not searched. Reference lists of selected studies were also searched manually.

Duplicate articles returned from the search were removed (DB). Titles and abstracts of all returned citations were then independently screened for eligibility by two reviewers (DB and MC). The same reviewers then screened the relevant full-text articles for inclusion based on our inclusion/exclusion criteria. Disagreements were resolved by discussion with a third reviewer (SW). A list of excluded papers and explanations for exclusion can be found in. Supplementary Table S1

Observational, cross-sectional, longitudinal and interventional study designs published in English were considered. Articles were deemed eligible if: (1) study populations were older adults (defined as mean age >50 years); (2) assessed outcomes that included at least one of muscle strength (e.g., grip strength), muscle mass (e.g., appendicular lean mass) or physical performance (e.g., gait speed), in alignment with a clinical diagnosis of sarcopenia and (3) clinical trials included relevant baseline measurements. Editorials, commentary articles, letters to the editor, theses, dissertations and book chapters were excluded. Trials that did not distinguish treatment and control groups or interventions using combined exercise components were excluded. Figure 1 depicts the PRISMA flow diagram for this systematic review.

Two reviewers (DB and MC) extracted data independently using the Cochrane Collaboration template [49]. Data extracted from all studies included author(s), publication year and country, study design, sample characteristics (number, mean age and gender when available) and key findings. Methods for assessment of dietary habits and physical activity, specific outcome measures (how muscle mass, strength and physical performance were assessed), and gut microbiota analysis (e.g., DNA extraction and sequencing) were included for synthesis and comparison of findings. In addition to the above, data extracted from interventional studies included characteristics of intervention (species/strains, dose, duration) and comparator (e.g., placebo) and results were mapped to an intervention. No assumptions were made on missing or unclear data. In the event of missing and/or unclear data, the authors of the article were contacted in an attempt to provide this information. In cases where this contact was unsuccessful, the article was excluded. Discrepancies were resolved by a third reviewer (SW).

Endnote X9™ was used for synthesis and collation of data. Eligible papers were grouped into (i) observational (cross-section and case–control) and (ii) interventional (parallel and cross-over) studies. Due to the heterogeneity of included studies, a meta-analysis of outcomes was not possible. Instead, a qualitative review was performed.

Two independent reviewers conducted a quality appraisal of the included studies using the revised Cochrane Risk of Bias 2 (RoB 2) [50] and critical appraisal tools from the Joanna Briggs Institute (JBI) [51]. Bias in observational studies was assessed using JBI critical appraisal tools for analytical cross-sectional and case–control studies (DB and MC). The methodological quality of interventional trials was analysed according to the RoB 2 framework for both parallel and cross-over design studies (DB and SW).

Eight studies were conducted in Asia [54,61,62,65,67,68,69,70], six in Europe [36,52,57,59,63,64], four in the USA [55,56,58,60], one in Canada [66] and one in Brazil [53]. Two studies were conducted on women only [55,68] and one study recruited men only [58]. Fifteen studies recruited both women and men [36,52,53,54,56,60,61,62,63,64,65,66,67,69,70] while two studies did not report the biological sex of participants [57,59]. In total, 4071 participants were included, of which approximately 51.6% were female. The mean age of participants across all 20 studies was 69.9 ± 8.3 with sample size ranging from 17 to 1417 participants. General characteristics of the included studies are outlined in Table 1.

Muscle strength was evaluated in 18 studies. Hand grip strength (HGS) was utilised in sixteen studies [36,52,53,55,57,58,60,61,62,64,65,66,67,68,69,71], six studies used repeat chair stands (RCS) [36,59,60,61,67,71], and both were measured in five studies [36,60,61,67,71]. Table 4 provides a summary of the methods used for evaluating muscle strength, mass and physical performance.

Seventeen studies evaluated muscle mass. Bioelectric impedance analysis (BIA) [36,53,55,61,65,67,68,69,70] and dual-energy X-ray absorptiometry (DXA) [57,58,59,60,63,64,71,72] were the most common methods utilised in nine and eight studies, respectively, while magnetic resonance imaging (MRI) was used in one study [56]. Muscle mass (quantity) was reported in a variety of ways, including absolute values such as fat mass (FM), fat-free mass (FFM), lean body mass (LBM), skeletal muscle mass (SMM), appendicular skeletal muscle mass (ASMM) or appendicular lean mass (ALM); indices such as skeletal muscle index (SMI) or appendicular skeletal muscle index (ASMI); or by incorporating adjustments to account for body size (height, weight, BMI) such as ALM/ht², LBM/ht², ASM/body weight, ALM/BMI.

Among the 13 studies that evaluated physical performance, the short physical performance battery (SPPB) was performed in seven studies [36,57,58,60,63,67,69]. Gait speed (GS) was assessed in six studies, reporting walking time across 4- [36,52,65], 6- [67], and 10 m distances [71]. GS was evaluated in the study by Wang et al. [68]; however, the distance was not specified. Six-minute walking distance (6MWD) was measured in two studies [56,64], and the timed up and go (TUG) test was evaluated in one [71]. Dillon et al. [60] evaluated the 400 m walking time and stair climb test.

Taxonomic composition profiling of faecal samples was undertaken in 14 studies. Faecal metagenomic sequencing was utilised to quantify the microbiota in two studies [36,67], while 12 studies performed 16S rRNA sequencing, amplifying hypervariable regions V4 [55,58,59] and V3–V4 [60,61,62,63,64,65,66,69,70]. Various alpha-diversity indices were used to evaluate differences in the richness and evenness of samples in 11 studies [58,59,60,61,62,63,64,65,66,67,69] and variability in the community composition across samples (beta-diversity) was reported in 10 studies [36,58,59,61,62,64,65,66,67,68]. Gut microbial functions were evaluated by predictive metagenome profiling (PICRUSt) in two studies [61,62]. A summary of methods for evaluating gut microbiota and microbiome is provided in Table 5.

Quality assessment using JBI critical appraisal tools for analytical cross-sectional and case–control studies are displayed in Figure 2 and Figure 3, respectively. The risk of bias for parallel and cross-over design interventional studies using RoB 2 is displayed in Figure 4 and Figure 5. Of the observational studies, the majority addressed the criteria and thus were rated quite high. However, only 3/6 studies addressed the criteria of identifying and controlling for confounding factors. This was similar for the case–control studies, with 4/8 studies not identifying and controlling for confounding factors. Moreover, measurement of exposure between case and control was not identified in 4/8 studies. For the parallel study designs, the majority met all criteria, and the risk of outcome assessment was mostly low. One study [52] raised some concerns regarding the reporting of selected data. In addition, 3/5 studies were funded by the supplement/product company and listed co-authors that worked or have been funded before by said supplement/product company [53,56,57]. Of the six interventional studies, only three registered their trial with their respective clinical trials registry [52,53,54]. The one cross-over study [55] was rated overall high risk, specifically regarding carry-over effects, selection of reported data and study funded by the supplement/product company and listed co-authors working or have been funded before by said supplement/product company.

In one of the larger population studies, secondary analysis revealed significant differences in beta diversity (Bray–Curtis distance) between individuals with and without sarcopenia [67]. The authors reported that several bacterial species, including D. piger, C. symbiosum, H. effluvii, B. fluxus, C. innocuum, C. and C. citroniae, were significantly associated with the presence of sarcopenia and its severity [67]. The same authors conducted a smaller case–control study but only in women and found nine distinct bacterial species were significantly enriched in sarcopenic versus 13 species in the non-sarcopenic individuals, including associations with muscle mass [68]. From these studies, the relative abundance of D. piger, P. copri and B. longum, were considered potential taxonomic biomarkers of sarcopenia. However, with low to moderate AUC values for P. copri (AUC: 0.372) and B. longum (AUC: 0.647), only D. Piger was identified as most important bacteria for discriminating between sarcopenia and non-sarcopenic (AUC: 0.852). The ratio Firmicutes/Bacteroidetes (F/B) could also be considered a biomarker of sarcopenia, with Wang and colleagues reporting a higher F/B ratio in sarcopenic individuals compared to controls [68]. The F/B ratio is widely considered to be important for maintaining normal intestinal homeostasis, and a decrease in the ratio is often observed in inflammatory diseases [73]. Similarly, Wu and colleagues found the ratio of Prevotella/Bacteroidetes (P/B) to be higher (1.5 times) in sarcopenic compared to non-sarcopenic individuals [69]. In addition, the genus Coprococcus was positively associated with sarcopenia, while the family Lachnospiraceae demonstrated a negative relationship. The P/B ratio is considered an important indicator of diet and lifestyle, and like the F/B ratio, it may also be a possible biomarker for a distinction between control and sarcopenia individuals [74]. In another cross-sectional study, higher levels (5 times) of F. prausnitzii were evident in non-sarcopenic individuals compared to individuals with primary sarcopenia and these levels were significantly associated with higher hand grip strength [66]. Relative abundances of F. prausnitzii and B. longum are likely beneficial species for healthy aging and muscle, given their role in short-chain fatty acid (SCFA) production. In addition, F. prausnitzii is associated with improved insulin sensitivity, anabolic balance and alleviation of local intestinal and systemic inflammation [26,75,76]; all beneficial to healthy muscle mass and function.

The addition of physical frailty to the microbiome–sarcopenia relationship was recently explored by Picca and colleagues [77]. Biomarkers of physical frailty appeared to be identified along with sarcopenia using machine classifier techniques. Among the gut microbes contributing to the model, Oscillospira and Ruminococcus were more abundant in such individuals, while Barnesiellaceae and Christensenellaceae were higher in the non-physically frail/sarcopenic. While very little is known about a possible functional link between Oscillospira and physical frailty and sarcopenia, the butyrate-producing bacteria Ruminococcus is in line with previously reported associations between Ruminococcus abundance and frailty [33]. Indeed, a secondary analysis by Tavella and colleagues [64] found muscle mass (SMI) to be negatively correlated with Ruminococcus 2 along with Blautia, Fusicatenibacter, and Subdoligranulum, whereas SMI was positively correlated with the Christensenellaceae R7 group. In another study, SMI was significantly positively correlated with Blautia and Bifidobacterium in men, while in women, higher SMI was associated with higher read counts of Eisenbergiella, Flavonifractor, Eggerthella, and Erysipelotrichaceae incertae sedis [70]. Whether and how Ruminococcus abundance impacts muscle metabolism and function in the context of sarcopenia and/or frailty warrants further investigation. Finally, Lim et al. [62] reported a significant association between interindividual variation in microbiota composition (Bray–Curtis distance) and frailty-related metadata, including grip strength and the Korean Frailty Index (FI) score. Grip strength was not significantly associated with various alpha diversity indices, although a weak association was observed with Bacteroides and Alistipes.

Based on the observations in older sarcopenic and/or frail individuals, there appears to be an overall apparent shift of gut microbiota composition towards a lower abundance of butyrate-producing bacteria with physical frailty and sarcopenia, suggesting a positive role for these microbes in muscle function and skeletal mass potentially via direct action on the muscle or secondary action via reductions in inflammation. Indeed, Kang and colleagues showed a decrease in the abundance of butyrate-producing genera (Lachnospira, Fusicantenibacter, Roseburia, Eubacterium and Lachnoclostridium) in sarcopenic and “probable” sarcopenic subjects [61]. The apparent protective effect of butyrate-producing bacteria on outcomes related to sarcopenia suggests a potential benefit of enhancing the colonisation and/or growth of these bacteria through diet or probiotic supplementation.

With this in mind, Barger et al. found that FFQ-reported diets high in fibre positively impacted butyrate-producing genera (including Ruminococcus, Lachnospira and Clostridia), and gene counts for butyrate production (KEGG IDs: K01034↗, K01035↗), with authors suggesting a likely mechanism of action that may link dietary fibre intake with higher levels of whole-body lean mass (%) and physical function in older adult men [58]. Cox and colleagues underwent a secondary analysis of older female individuals to explore differences in microbiota and muscle mass/strength in those with poor appetite (Simplified Nutritional Appetite Questionnaire [SNAQ] score < 14) compared with normal appetite [59]. Differences in bacteria species involved in producing butyrate were evident between those with good and poor appetite, with reductions in Lachnospira as well as significantly higher mean chair stand time (indicating lower muscle strength) observed in those with poor appetite [59]. However, this was based on within-group observations only and no direct correlation between relative bacterial abundance and performance measures was undertaken. Notwithstanding, both poor and good appetite groups were matched on dietary similarity via a healthy eating index, which may suggest an influence of appetite on bacteria abundance beyond dietary intake alone [78]. Conversely, it may also reflect a limitation of using self-report mechanisms for capturing dietary data [79].

Finally, in two separate studies that examined faecal butyrate levels with relatively healthy 60–70 year olds, faecal butyrate levels were found to be positively correlated with sex-adjusted SMI [65] and greater grip strength [60]. These observations add further support to the mechanisms of the microbiota-gut-muscle axis via the production of SCFA butyrate; however, it should be noted that only 5% of all microbially produced SCFAs can be found in feces and thus other measures (i.e., microbiome and blood) may be required to confirm such a relationship [80].

Few studies have explored the effects of dietary (gut-modulating) interventions on muscle mass and function in healthy and/or sarcopenic individuals. The studies included in the review used single- and multi-strain probiotics with or without prebiotics or combined with other ingredients. Two studies investigated the effects of multi-strain synbiotic formulations on muscle strength and/or lean muscle mass. Treatment for 3 months with a synbiotic (6 g FOS + L. paracasei, L. rhamnosus, L. acidophilus and B. lactis, all at 10⁸ to 10⁹ CFU) resulted in no difference in HGS or FFM compared with placebo in older male and females [53]. Conversely, an 18-week crossover trial in older women given a high protein maintenance diet combined with either a prebiotic (inulin), a multi-strain (1.54 × 10⁹ CFUs B. bifidum and 4.62 × 10⁹ CFUs each of B. breve, B. longum, L. acidophilus, and L. plantarum) probiotic, or synbiotic (prebiotic + probiotic) resulted in a small, but significant increase in FFM [55]. This study also analysed the microbiota, and while bacterial diversity was not significantly impacted, Lactococcus and Streptococcus were enhanced, while butyrate-producing genera Roseburia and Anaerostipes were suppressed [55]. Several limitations existed in this study, including a short intervention period (2 weeks), carry-over effects (confirmed by faecal qPCR recovery) due to cross-over design and other dietary factors (increasing protein intake during the study)—all potentially influencing the results and thus should be interpreted with caution.

In another study, patients with sarcopenia consuming an experimental formula (L. paracasei, leucine and omega-3 fatty acids) for 60 days demonstrated significant improvements in HGS and SPPB scores [57], however, because the intervention was a multi-ingredient product, it is difficult to ascertain which ingredient or ingredients led to the improvements [57]. Performance improvements were also observed after consuming a single-strain probiotic formulation containing L. plantarum for 18 weeks [54]. Significantly lower TUG times, but not 10 m walking test, were noted in the intervention compared to the placebo group [54]. Again, limitations existed within the study, such as limited recording of the actual amounts of food intake and physical activity during the study, which may impact the outcomes. While positive effects were observed, given no microbiome analysis, it is difficult to ascertain possible mechanisms of action. Finally, Liu et al. reported a non-significant improvement in 6MWD following daily oral supplementation of a postbiotic (1000 mg urolithin A) compared with placebo [56]. Urolithin is a gut microbiome-derived metabolite that has been shown to stimulate mitophagy, improve muscle function in older animals and induce mitochondrial gene expression in older humans.

Several limitations exist within the studies included: Firstly, there is substantial variation regarding the setting (e.g., sample size, participant health status or co-morbidity, and geographic location) and the methodology of the studies included in this review. Several critical methodologies differ from one study to another. For example, faecal sample collection and processing, the platform used for sequencing, and sequencing depth can all impact results [81,82]. Interestingly, the time of day for stool sample collection appears to be associated with the gut microbiome. A small (n = 6) cross-sectional study of healthy female volunteers who defecated more than once daily reported the second stool of the day had lower bacterial richness and diversity and significantly higher total SCFA concentrations compared with the first stool [82]. Habitual diet and physical activity (or their potential changes during the intervention period for experimental trials) did not appear to be adequately assessed or controlled in multiple studies. Therefore, the observed gut microbiota outcomes may have been impacted by these lifestyle factors, and it is not possible to draw firm conclusions about these interactions. Many different methods were used for evaluating the strength, muscle mass, and/or performance of the study populations, which limits the comparability of the outcomes. Moreover, in the case of using BIA and DXA, the accuracy and reliability of results may differ between methods and formulas used by the respective techniques. Although not critical to our review, the specific criteria that were followed for detecting or confirming sarcopenia were reported in some (9/20), but not all studies. Perhaps most importantly, in all described clinical trials, the dose, duration and intervention used differed from each other. Therefore, definitive conclusions about their effectiveness cannot be extrapolated. Additionally, a lack of longitudinal studies, particularly with post-intervention follow-ups, restricts understanding of both effect and effect duration. Finally, over half (4/6) of the interventional studies did not evaluate any changes in microbial composition or diversity following treatment, so attempting to determine mechanisms of action or causal relationships is limited.

This review identified a lack of longitudinal studies investigating the effects of targeted interventions on gut microbiota and/or its microbiome in older populations. Further research is needed in the form of controlled clinical trials, including different types of gut-modulating therapeutics, larger sample sizes of more homogenous populations, and extended follow-up periods. Ideally, these would be of adequate dose and duration and be designed as randomised, double-blinded, and placebo-controlled to avoid complications such as inadequate wash-out that may be experienced with cross-over study designs [83]. Future studies should ensure they include measures for capturing detailed information on diet composition, physical activity, and other factors that influence gut microbiota, such as sleep patterns and stress levels [84,85]. Insofar as possible, we recommend utilising direct methods (e.g., accelerometry) in addition to self-reported questionnaires to measure physical activity [86]. Likewise, methods that provide for both quantitative and qualitative analysis (such as 3–5-day food records or multiple 24 h dietary recalls) should be used for dietary assessment, leveraging digital platforms when practical, to minimise burden, maintain compliance and optimise data quality [87,88]. It would also be relevant to document medication use and consumption history of prebiotic- or probiotic-containing supplements. Identification and sub-analysis of baseline gut microbiota may help explain interindividual differences following treatment (“responders” and “non-responders”) and give valuable insight into underlying mechanisms of action. Adopting universal and standard methods for evaluating skeletal muscle strength, mass, and physical performance would better allow the combining of data for statistical modelling.