Gut microbiota in patients with obesity and metabolic disorders — a systematic review

Background Previous observational studies have demonstrated inconsistent and inconclusive results of changes in the intestinal microbiota in patients with obesity and metabolic disorders. We performed a systematic review to explore evidence for this association across different geography and populations. Methods We performed a systematic search of MEDLINE (OvidSP) and Embase (OvidSP) of articles published from Sept 1, 2010, to July 10, 2021, for case–control studies comparing intestinal microbiome of individuals with obesity and metabolic disorders with the microbiome of non-obese, metabolically healthy individuals (controls). The primary outcome was bacterial taxonomic changes in patients with obesity and metabolic disorders as compared to controls. Taxa were defined as “lean-associated” if they were depleted in patients with obesity and metabolic disorders or negatively associated with abnormal metabolic parameters. Taxa were defined as “obesity-associated” if they were enriched in patients with obesity and metabolic disorders or positively associated with abnormal metabolic parameters. Results Among 2390 reports screened, we identified 110 full-text articles and 60 studies were included. Proteobacteria was the most consistently reported obesity-associated phylum. Thirteen, nine, and ten studies, respectively, reported Faecalibacterium, Akkermansia, and Alistipes as lean-associated genera. Prevotella and Ruminococcus were obesity-associated genera in studies from the West but lean-associated in the East. Roseburia and Bifidobacterium were lean-associated genera only in the East, whereas Lactobacillus was an obesity-associated genus in the West. Conclusions We identified specific bacteria associated with obesity and metabolic disorders in western and eastern populations. Mechanistic studies are required to determine whether these microbes are a cause or product of obesity and metabolic disorders. Supplementary Information The online version contains supplementary material available at 10.1186/s12263-021-00703-6.


Introduction
Obesity-related metabolic disorders, including type 2 diabetes (T2DM), cardiovascular diseases, and nonalcoholic fatty liver disease (NAFLD), affect 13% of the population and result in 2.8 million deaths each year [1,2], and are a significant socioeconomic burden to society. Pathophysiology of obesity and metabolic disorders is multi-factorial, and currently, therapies are limited. The role of intestinal microbiota in patients with obesity and metabolic disorders have been extensively studied in the past decade. Humanized mouse models showed that the microbiome in obese subjects appeared to be more efficient in harvesting energy from the diet and may thereby contribute to the pathogenesis of obesity [3,4]. However, observational studies reported inconsistent and inconclusive changes of intestinal microbiota in patients with obesity and metabolic disorders [5]. For instance, the Firmicutes and Bacteriodetes ratio (F/B ratio) is not a reproducible marker across human cohorts [6].
Microbial-based therapies such as probiotics aiming to reshape the gut microbial ecosystem have been increasingly explored in the treatment of obesity-related metabolic disorders [7,8]. Traditional probiotics, primarily consisting of Lactobacillus and Bifidobacterium have been shown to elicit weight loss in subjects with obesity yet the effect sizes were small with large variations of efficacy among different studies [9]. Emerging evidence showed that Akkermansia muciniphila was depleted in patients with obesity-related metabolic disorders. These results have led to mechanistic studies and clinical trials to test its efficacy in the management of obesity and metabolic disorders [10].
Age, geography, and dietary patterns largely affect the gut microbiome [11][12][13]. The gut microbiota of vegetarians was dominated by Clostridium species [14] whereas subjects who mainly consumed fish and meat had high level of F. prausnitzii [15]. In recent years, the prevalence of childhood obesity has increased sharply. However, only limited data has issued the function and structure of gut microbiota in children and adolescents with obesity [16].
We have therefore conducted a systematic review of case-control studies evaluating the microbiota in patients with obesity and metabolic disorders compared to lean, healthy controls to summarize the current evidence in the relationship between individual members of the microbiota and obesity. We aimed to identify novel candidates as live biotherapeutics to facilitate the treatment of obesity and metabolic disorders.

Search strategy
This systematic review was performed in accordance with the PRISMA 2009 guidelines [17]. We performed a systematic search of MEDLINE (OvidSP) and Embase (OvidSP) of articles published from Sept 1, 2010 to July 10, 2021 to identify case-control studies comparing gut microbiota in patients with obesity and metabolic disorder and non-obese, metabolically healthy controls. Search strategy is shown in the Appendix.

Study selection and patient population
Studies were included if they were (1) case-control studies comparing gut microbiota in patients with obesity and metabolic disorders and non-obese, metabolically healthy individuals (controls); (2) intestinal microbiota was assessed by next-generation sequencing (NGS; 16s rRNA amplicon or shotgun metagenomic sequencing); and (3) obesity was defined based on body mass index (BMI) ≥ 30kg/m 2 and metabolic disorders including type 2 diabetes mellitus, non-alcoholic fatty liver disease, cardiovascular disease, and metabolic syndrome were diagnosed according to respective guidelines (Table 1). Studies from all age groups were included. Studies were excluded if they were (1) case reports, reviews, metaanalyses, re-analysis of public datasets, or conference abstracts, (2) without data for individual bacterial groups, (3) not in English, and (4) not a case-control design. Studies of genetic-associated obesity such as Prader-Willi syndrome were also excluded.

Study outcomes
The primary outcome was the bacterial taxonomic changes in patients with obesity and metabolic disorders compared to non-obese, metabolically healthy controls. Secondary outcomes included the changes in bacteria diversity and F/B ratio, subgroup analysis of microbiota changes in adults and children with obesity and metabolic disorders, and in Eastern and Western populations. Data on microbiota community composition were extracted from each study. Taxa were defined as "lean-associated" if they were depleted in patients with obesity and metabolic disorders or negatively associated with abnormal metabolic parameters such as high body mass index (BMI), elevated fasting plasma glucose and elevated serum cholesterol. Taxa were defined as "obesity-associated" if they were enriched in patients with obesity and metabolic disorders or positively associated with abnormal metabolic parameters. Taxon at each level (phylum, class, order, family, genus) was only counted once for each study (i.e., if a genus was both depleted in obesity and negatively associate with fat mass in the same study, it was only counted once).

Eligibility assessment and data extraction
Two authors (JW, HW) independently reviewed studies and excluded based on titles, abstracts, or both to lessen the selection bias and then reviewed selected studies with full text for complete analysis. JW extracted data from studies and entered it into a designated spreadsheet. HW checked the accuracy of this process. The data were re-checked when there was a discrepancy. XZ arbitrated if the discrepancy cannot be resolved by consensus and discussion. The data collected included the following: participant characteristics, including age group, country, types of metabolic disorders, number of patients; types of specimens, microbiota assessment method, microbiome diversity, and Firmicutes/Bacteroides ratio.

Quality assessment
The Newcastle-Ottawa Scale was applied to assess the quality of included studies. The Newcastle-Ottawa Scale consists of 3 domains (maximum 9 stars); selection (is the case definition adequate, representativeness of the cases, selection of controls, definition of controls); comparability (comparability of baseline characteristics); and exposure (ascertainment of exposure, same method of ascertainment for cases and controls, attrition rate).

Microbiome assessment methods
Of the 58 studies assessing stool microbiome, 50 studies assessed the gut microbiota by using 16S ribosomal RNA (rRNA) gene sequencing, six used shotgun metagenomic sequencing and two studies applied both 16s rRNA and shotgun metagenomic sequencing. Both studies assessing biopsy microbiome applied 16S rRNA sequencing.

Difference of microbiota between adult and childhood obesity
The trend for most microbial changes in adult and childhood obesity were consistent. Studies reported Actinobacteria as lean-associated, while Proteobacteria and Firmicutes as obesity-associated in both adults and childhood obesity. However, discrepancies were observed for several genera. Three studies in adults consistently reported that Fusobacterium was obesityassociated, but controversial results were found in children [18,20,22,32,61,77]. Moreover, more studies reported that Dorea [39,46,49,77] and Ruminococcus [39,44,49,69] were obesity-associated in adults, while more studies reported them to be lean-associated in children [19,68]. Three studies consistently reported that Turicibacter was lean-associated in adults [44,66,69], but one study reported it to be obesity-associated in children  [20]. Notably, three studies in adults reported that the genus Bifidobacterium was lean-associated [22,57,58], while controversial results were found in children (3 lean-associated and 2 obesity-associated) [19-21, 38, 68]. These findings suggested that microbiota in childhood obesity and metabolic disorders were more heterogeneous compared with adults.

Difference of microbiota between the East and the West
Large discrepancies in gut microbiome in obesity and metabolic disorders were observed in studies from the East and the West. Four studies exclusively consisting of populations in the West reported that the Family Coriobacteriaceae was obesity-associated [27,38,53,71] whereas none in the East reported significant change of this bacterial family between obese subjects and controls.

Quality of the evidence
The Newcastle Ottawa Scale showed that all 60 studies provided an adequate explanation in the definition and selection method for patients with obesity and metabolic disorders (Table 4). Fifty-five (91.7%) of 60 studies did the same process for controls. Twenty (33.3%) and 27 (45%) studies demonstrated comparable data of sex and age in patients with obesity / metabolic disorders and controls.

Discussion
To our knowledge, this is the most comprehensive systematic review in microbiota and obesity and metabolic disorders, as we extracted the data of each available bacterial group using the lowest taxonomic level based on NGS of each included study. We believe that the findings reflect the best available current evidence demonstrating the relationship between individual bacterial taxa and obesity or metabolic disorders. Proteobacteria was the most consistently reported obesity-associated phylum. Several members of Proteobacteria, such as Proteus mirabilis and E. coli, were potential drivers of inflammation in the gastrointestinal tract [7,80,81]. Low-grade inflammation is a risk factor for developing metabolic diseases including atherosclerosis, insulin resistance, and diabetes mellitus [82]. Besides stool microbiota, obese subjects with T2DM also showed a high bacterial load with an increase in Enterobacteriaceae in plasma, liver, and omental adipose tissue microbiota [83].
Lactobacillus was reported to be an obesity-associated taxon and abundance was higher in the stool of patients with obesity and metabolic diseases. This food-derived probiotic genus showed relative low prevalence and abundance in the commensal gut microbiota [52]. Previous clinical trials of Lactobacillus, alone or in combination with Bifidobacterium, showed variable efficacy in weight loss in patients with obesity [9]. These inconsistent results indicated that the underlying mechanisms of Lactobacillus (at least some of its species) in the treatment of metabolic disorders warrant further investigation. Other commensal bacteria such as Bifidobacterium spp., Alistipes spp., and Akkermansia that constitute a large proportion of the gut microbiota were frequently observed to be higher in healthy individuals than obese, metabolically affected subjects. These species might therefore exert a more durable beneficial effect for the consideration in managing obesity compared with Lactobacillus.
Akkermansia muciniphila (Actinobacteria phylum), a species identified by NGS, was one of the most commonly reported lean-associated bacteria in obesity and metabolic diseases. A. muciniphila was reported to help modulate the gut lining which could promote gut barrier function and prevent inflammation caused by the "leaky" gut [84]. A clinical trial demonstrated that supplementation with A. muciniphila could reduce body weight and decrease the level of blood markers for liver dysfunction and inflammation in obese insulin-resistant volunteers [10]. Another proof-of-concept study showed that supplementation with five strains including A. muciniphila was safe and associated with improved postprandial glucose control [85]. These findings highlight the potential of specific live biotherapeutics in weight control in subjects with obesity and metabolic diseases.
Other genera that were consistently reported to be more abundant in lean healthy individuals than obese subjects were Alistipes (Bacteroidetes phylum) and Faecalibacterium (Firmicutes phylum). Alistipes could produce small amounts of short-chain fatty acids (SCFA, acetic, isobutyric, isovaleric, and propionic acid) [86] while Faecalibacterium is one of the major butyrate producers in the human gut [87,88]. SCFA have antiinflammatory properties [89] and may promote weight loss through the release of glucagon-like peptide 1 that promotes satiety and the activation of brown adipose tissue via the gut-brain neural circuit [90,91]. Butyrate could activate the GPR43-AKT-GSK3 signaling pathway to increase glucose metabolism by liver cells and improve glucose control in diabetes mice [92]. They could also inhibit the expression of PPARγ, increase fat oxidation in skeletal muscle mitochondria, and reduce lipogenesis in high-fat diet (HFD) mouse model [93].
We have identified several genera, including Bifidobacterium, Roseburia, Prevotella, and Ruminococcus, that were consistently reported to be lean-associated exclusively in subjects from the East. Bifidobacterium spp. are widely used probiotics proven to be safe and welltolerated and exhibited a significant effect in lowering serum total cholesterol both in mice and in humans [94]. Roseburia is another major butyrate-producing Table 4 Quality of each included study by the Newcastle Ottawa Scale NA not appliable genus of the human gut [95]. R. intestinalis could maintain the gut barrier function through upregulation of the tight junction protein [96]. Supplementation of R. intestinalis and R. hominis could ameliorate alcoholic fatty liver disease in mice [97]. Ruminococcus bromii is a keystone species for the degradation of resistant starch in the human colon [98]. Prevotella copri (Bacteroidetes phylum) was found to improve aberrant glucose tolerance syndromes and enhance hepatic glycogen storage in animals via the production of succinate [99]. However, a recent study also showed that the prevalence of P. copri exacerbated glucose tolerance and enhanced insulin resistance which occur before the development of ischemic cardiovascular disease and type 2 diabetes [100].
Only limited human studies in the current review reported an increased ratio of F/B in obesity. An increased ratio of F/B was shown in studies of the high-fat diet mouse model [6]. No taxon distinction was found to be specific for any type of metabolic disease. This was in line with a recent study that showed obesity, but not type 2 diabetes, was associated with notable alterations in microbiome composition [58].
The strength of this study is that we applied a robust method of grouping various types of disease-microbiome associations into "lean, metabolically healthy state" or "obese, metabolically diseased state." Despite various metabolic disorders may affect the gut microbiota in different manners, the inter-study variation often supersedes the intra-study variation between disease and control groups [101]. Overall, the most striking observation is the lack of consistency in results between studies. This probably relates to the limitations of the studies included in this review. Also, it relies on the striking stability and individuality of adult microbiota, changing over time. Heterogeneity between studies is often a problem in systematic reviews. Several different methods were used to assess the microbiota, which makes it difficult to compare results between studies and likely contributes to the differences in results. While the standardization of study protocol (sample storage, DNA extraction, sequencing, analysis methods, and stringent subject recruitment criteria) could potentially result in comparable data between studies, this remains a big challenge across different regions. Moreover, we excluded studies that used species-or group-specific primers for microbiota assessment because such methods could only capture certain bacterial groups. This limits the total number of studies included. For robust microbiota results that are comparable among studies, there need to be efforts for standardization of sample storage, DNA extraction, sequencing, and analysis methods among groups undertaking gut microbiota studies. Finally, longitudinal studies would allow for a more robust association of changes in the microbiota to changes in obesity and metabolic disorders.

Conclusions
This systematic review identified consistent evidence for several lean-associated genera that may have therapeutic potential for obesity and metabolic diseases. Besides A. muciniphila, species from genera Faecalibacterium, Alistipes, and Roseburia might also harbor therapeutic potentials against obesity and metabolic diseases. These results provided a guide for the future development of certain bacteria into live biotherapeutics that may be helpful for the management of obesity and metabolic disorders. Further in-vitro and in-vivo research are needed to elucidate their role in the management of obesity and metabolic diseases.