Role of Presurgical Gut Microbial Diversity in Roux-en-Y Gastric Bypass Weight-Loss Response: A Cohort Study

The rising prevalence of obesity and associated diseases carries a significant health and economic burden. Bariatric surgery is currently the most effective strategy to achieve sustained weight loss in morbidly obese patients, where the most common surgical approaches are Roux-en-Y gastric bypass (RYGB) and sleeve gastrectomy (SG) [1, 2]. It has been described that these types of surgeries induce significant changes in the gut microbiota, likely reflecting both anatomical modifications of the gastrointestinal tract and its consequent changes in dietary habitats, digestion time, intestinal absorptive surface, and pH [3, 4]. However, it is clear now that not all microbes are commonly affected. For instance, the increase in Streptococcaceae abundance is commonly observed in both surgeries, while the increase in bacterial diversity and in Veillonella abundance seems to be more related to RYGB [5]. Gut dysbiosis is thought to play a relevant role in the development of obesity and associated complications. The mechanisms include an increased calorie harvest from the diet, a change in the functioning of the intestinal barrier, and production of metabolites that affect satiety signals among others [6]. Thus, it has been hypothesized that gut microbial alterations induced by bariatric surgery are related to the weight loss and metabolic improvements seen after the intervention [7].

Currently, the selection of the surgical approach in bariatric surgery requires a consensus between the patient and surgeon based on BMI, dietary habits, and comorbidities. However, it has been shown that either bacterial composition or the associated changes after bariatric surgery could play a significant role in both weight loss and diabetes remission [8]. Thus, the identification of potential gut microbial markers for the development of a precision medicine approach that incorporates the role of the gut microbiota could be of great value.

Several studies have followed gut microbial variations after bariatric surgery [3]; however, most of these studies have been performed in populations from developed countries, while data from developing countries, where overweight and obesity have become an important public health problem, are still scarce. Notably, numerous environmental characteristics such as diet, geography, and medication use play a significant role in gut microbial composition [9, 10]. Therefore, it is important to address this question in cohorts from developing countries [11, 12]. The aim of this study was to characterize the changes in gut microbiota composition in an adult Mexican cohort before and after RYGB and to explore whether presurgery gut microbiota or surgery-related changes in the microbial community were related to the extent of weight loss.

A prospective study was conducted with participants recruited from the Bariatric Surgery Program at Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán in Mexico City from 2015 to 2016. This program performs systematic multidisciplinary evaluations and interventions for surgery preparation, performance, and follow-up of patients with complicated obesity. For the present study, we included men and women, 18 years and older, with a BMI ≥40 kg/m² or BMI ≥35 kg/m² in the presence of chronic diseases associated with obesity and who were programmed to undergo RYGB. Participants with psychopathology limiting the capacity for adequate protocol participation, acute diseases, systemic inflammatory conditions, chronic bowel disease, uncontrolled medical and/or mental health diseases, and noncompliance to the clinical indications were excluded. This study protocol was reviewed and approved by the Ethics Committee of Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, approval number DIA-14040-14, and conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonization Guidelines for Good Clinical Practice. All the patients provided written informed consent before any protocol procedure. STROBE guidelines were followed during the preparation of this report.

This observational cohort prospective study included three visits for all participants: at the baseline (before surgery), 6, and 12 months after surgery. The following assessments took place in these visits: anthropometric and physiological measurements (including weight, height, body composition, and blood pressure), biochemical parameters, general and gastrointestinal symptoms, comorbidities, and medication use. Percentage weight loss from baseline was calculated as (initial weight − final weight)/initial weight × 100, while excess weight loss (%EWL) was calculated as the difference between the initial BMI divided by the difference between the initial BMI and a BMI of 25 kg/m² [13]. Following previous recommendations for presurgical weight loss [14], the program implemented routine prescription of casein or whey-based 800–1,000 calorie diets 2 weeks before surgery with the goal to reduce liver size and facilitate the procedure based on published protocols [15]. After bariatric surgery, the diet is gradually modified until the patient can tolerate solid foods. Over the course of a year, the diet transitions to a more normal eating pattern.

Biochemical measurements were carried out by the hospital on fasting serum samples and included serum glucose, lipids, and HbA1c, while C-reactive protein, serum vitamin D, iron, and bile acids were evaluated before and 12 months after surgery. Type 2 diabetes was defined as at least two measurements with HbA1c ≥ 6.5%, fasting plasma glucose ≥126 mg/dL, 2-h plasma glucose during a 75 g oral glucose tolerance test ≥200 mg/dL, and/or random plasma glucose ≥200 mg/dL [16], hypertension as an average (in at least two occasions) systolic blood pressure (BP) ≥140 mm Hg, diastolic BP ≥90 mm Hg, or self-reported use of antihypertensive medication [17] and dyslipidemia as total cholesterol (TC) > 200 mg/dL, HDL-c <40 mg/dL in men and <50 mg/dL in women, or triglycerides >150 mg/dL [18]. Diabetes remission was defined as a fasting glucose ≤100 mg/dL or HbA1c <6% in the absence of antidiabetic medications [19]. Hypertension remission was defined as systolic/diastolic BP <140/90 mm Hg without antihypertensive medication during follow-up [20]. Remission of dyslipidemia was considered as LDL-c <160 mg/dL, total cholesterol <240 mg/dL, triglycerides <200 mg/dL, and HDL-c >40 in men and >50 in women, without lipid-lowering medication at follow-up visits [21].

RYGB was performed as follows: a gastric pouch of approximately 50 mL was constructed. Then, the small bowel was divided 40–60 cm from the ligament of Treitz. An end-to-side gastro-jejunostomy was handsewn, and a jejuno-jejunostomy was constructed using a linear stapler. Alimentary limb length was approximately 150 cm. The mesenteric defect and the Petersen space were systematically closed using nonabsorbable sutures.

Baseline fecal samples were provided by each participant after the surgery was scheduled and the presurgery very-low-calorie diet was indicated, while postsurgery samples were obtained after 12 ± 4 months. Briefly, a fresh stool sample was collected in a sterile plastic cup by participants and then delivered to the clinic, where it was stored at −70°C until further analysis. Total DNA was extracted from the stool sample using the commercial kit (QIAamp DNA Stool Mini kit-Qiagen) according to manufacturer instructions. DNA concentration and purity were determined with a Nanodrop 2000c spectrophotometer (Thermo Scientific^®).

The composition of the gut microbiota in fecal samples was determined by 16S rRNA gene sequencing. The V3-V4 region of the 16S rRNA gene was amplified using the methodological approach developed by Illumina. Amplicons were sequenced at the Core Sequencing Unit of Instituto Nacional de Medicina Genomica on an Illumina MiSeq platform (2 × 250 bp).

The raw demultiplexed sequences were processed using QIIME 1.9. Briefly, quality filters were used to remove sequences containing ambiguous bases or low-quality reads (Phred quality score <30). Operational taxonomic unit (OTU) read counts were calculated using the QIIME pipeline (version 1.9.1; default parameters) with closed reference OTU picking at 97% identity against the Greengenes database (version 13_08). Potential chimeras were detected with USearch61 and excluded from further analysis. Alpha diversity metrics, Shannon, Chao, and observed OTUs, were calculated at a standardized sequencing depth. The differences in the overall microbial community structures were explored using weighted and unweighted UniFrac distance matrices. Taxonomical classification was performed to generate phylum-to-genus-level composition matrices, and the analysis was confined to OTUs that constituted ≥0.1% of the total reads.

The primary endpoint was the composition of the microbiota in stool samples before and after the surgery. The secondary endpoint was the changes in gut microbiota composition in relation to EWL.

Statistical analysis was conducted using the open-source software R, version 3.5.1, and figures were generated using ggplot2. For comparison of clinical variables, before and after surgery, either mean ± SD or median (min–max) are displayed, and paired T test or paired Wilcoxon were performed according to variable distribution. Nonparametric tests (paired Wilcoxon) were performed when comparing microbiota data before and after the surgery, and Benjamini-Hochberg multiple testing adjustment was applied (P_FDR < 0.1 indicates significance). Linear discriminant analysis effect size was used to identify differentially abundant taxa between subjects with and without T2D at baseline, where a p value <0.05 and a log linear discriminant analysis LDA score ≥2 were used for identification of potential biomarkers. PERMANOVA on Unifrac distance matrices was performed to quantify the variation explained by the surgery based on 2,000 permutations. To reveal associations between changes in gut microbial abundance and weight loss, Spearman’s rank-order correlation coefficients were calculated between delta abundance (T12 -T0) of significantly differentiated bacteria and %EWL.

Twenty patients who underwent RYGB met the study criteria. 40% were men and the mean age was 38.75 ± 8.56 years. 30% of patients had T2D and 50% had hypertension or dyslipidemia. None of the patients reported antibiotic intake 2 months before sample collection. Anthropometric, clinical, and metabolic parameters at baseline, 6 and 12 months are shown in Table 1. There was an important progressive reduction of weight, BMI, and fat mass throughout this period (p < 0.05). Percentages of weight loss and EWL were 27.0% and 28.0% at 6 months and 53.6%, and 67.0% at 12 months (p < 0.001). Fasting glucose, HbA1c, triglycerides, and LDL-c were significantly decreased, while HDL-c increased throughout the follow-up period (p < 0.05). This was accompanied by an increase in total serum bile acids and a decrease in C-reactive protein. All patients with diabetes experienced remission in all periods of assessment, compared to patients with hypertension (20% at 6 months and 15% at 12 months) and dyslipidemia (10% at 6 months, mostly attributed to remission of hypoalphalipoproteinemia).

Patients’ gut microbiota before surgery was dominated by Bacteroidota, followed by Bacillota (formerly Firmicutes) and Pseudomonadota (formerly Proteobacteria), which accounted for around 99% of the counts (online suppl. Fig. S1; for all online suppl. material, see https://doi.org/10.1159/000535397). Specifically, among Bacteroidota, the most abundant genera were Prevotella (median abundance [MA] = 21.6%) and Bacteroides (MA = 19.5%), while for Bacillota, an unclassified genus from Ruminococcaceae family (MA = 9.4%) and other from Lachnospiraceae family (MA = 4.1%), where the genera with the highest MA. At baseline, Lachnospira relative abundance was positively correlated with BMI (rho = 0.707, p = 0.001) and body fat percentage (rho = 0.665, p = 0.004), while an unclassified genus from Christensenellaceae family showed a negative correlation with both anthropometric variables (rho = −0.478, p = 0.044 and rho = −0.510, p = 0.036, respectively); however, after FDR correction, only the association of Lachnospira abundance with BMI remained significant (P_FDR < 0.05). Despite the fact there were no significant differences in alpha diversity indices between subjects with and without T2D, LefSe analysis showed that those with T2D had higher abundance of Enterobacteriaceae family and genus Mitsuokella, while a lower abundance of Streptococcus (online suppl. Fig. S2).

To detect changes in the gut microbiota after RYGB surgery, we analyzed diversity and composition before and 12 months after surgery. Multivariate analysis based on unweighted Unifrac distance demonstrated a moderate change in β-diversity after RYGB surgery that explained around 4% of gut microbial variation, but this was only as a trend (p = 0.071), while no significant effect was observed on weighted Unifrac distance.

We next assessed the impact of bariatric surgery on microbial alpha diversity indices. As observed in Figure 1, the three estimated indices showed a significant increase when compared to samples prior to surgery (p < 0.05), and these results were similar in subjects with and without T2D (online suppl. Fig. S3).

Four out of 37 identified genus-level phylotypes were significantly enriched in fecal samples after RYGB (Fig. 2). Participants showed a significant increase in the relative abundance of Veillonella, Streptococcus, Oscillospira (Bacillota), and an unclassified genus from the Enterobacteriaceae family (Pseudomonadota) (P_FDR < 0.1). In contrast, no genera were significantly depleted after surgery (online suppl. Table 1). Since the cohort included 6 patients with T2D, we analyzed whether the observed results were similar in these patients. Despite the small sample size, the increase in the relative abundance of Veillonella, Oscillospira, and Streptococcus was significant or close to significant, while the increase in the unclassified genus of Enterobacteriaceae was no longer observed (online suppl. Fig. S4).

To understand whether changes in gut microbiota after surgery were related to weight loss magnitude, we calculated Spearman’s rho coefficients between the delta values of alpha diversity indices, as well as between the delta abundance of the four phylotypes that significantly changed after RYGB and % EWL. Despite that there were no significant associations between changes in alpha diversity and % EWL (p > 0.5), the increase in Veillonella abundance was negatively correlated with % EWL (p = 0.060, Fig. 3a). In patients with T2D before the surgery, the change in Veillonella abundance was still negatively correlated with % EWL (although not significant, rho = −0.772, p = 0.220). Interestingly, after surgery, Veillonella abundance was positively correlated with BMI (rho = 0.732, p = 5.4 × 10⁻⁴) and as a trend with body fat percentage (rho = 0.459, p = 0.063).

Given that a higher increase in Veillonella abundance was related to lower weight loss, we explored if basal characteristics of the gut microbiota could predispose to a greater increase in the abundance of this genus. As observed in Figure 3b–d, a lower bacterial richness and diversity before surgery was significantly associated with a greater increase in Veillonella abundance after surgery (p < 0.05). Interestingly, when stratified by presurgery diabetes status, this association was only observed in individuals without T2D (online suppl. Table 2).

In this study of patients who underwent RYGB, sustained weight loss (between 26 and 29%) and an improvement or remission of chronic diseases was observed 12 months after surgery. These improvements were accompanied by changes in alpha diversity and gut microbial composition, and the latter was associated with the magnitude of weight loss. Furthermore, baseline gut microbial diversity seems to be related to the taxonomic changes observed after RYGB.

Regardless of the overwhelming disruption to the gut microbiota caused by bariatric surgery, it has been suggested that some of the changes are only partially sustained 12 months after the surgery [22]. In our study, 1 year after RYGB, a significant increase in bacterial richness and diversity was observed. This is consistent with many reports [5, 22, 23], suggesting that systemic and anatomical changes induced by RYGB can restore the putative loss of microbial diversity induced by obesity and related comorbidities. Regardless of the latter, and in line with previous studies, the observed changes in alpha diversity indices were not related to excess weight loss [24, 25].

Although overall differences in β-diversity were not detected, differences in various individual taxa after surgery were found. For instance, relative abundance of oral and small intestine-derived bacteria such as Streptococcus and Veillonella was significantly increased, which is consistent with previous studies [26, 27]. The rearrangement of the gastrointestinal tract involved by RYGB results in changes in bile acid production and luminal pH [28]. Particularly, it is suggested that the bile loop produced by RYGB, but not SG, promotes a greater reabsorption of bile acids in the small intestine, subsequently increasing serum concentrations, as observed in our study [29]. This increased reabsorption is thought to then limit the delivery of bile acids to the colon, which can influence the growth of bile acid-sensitive bacteria. In fact, it has been suggested that the observed increased abundance of Veillonella and Streptococcus could result from this effect [30‒32]. In addition, previous studies have shown that the absence of gastric acid also promotes the growth of Enterobacteriaceae family members, as well as the genus Veillonella [33]. Notably, the increase in the unclassified genus from Enterobacteriaceae family was observed only in subjects without prior T2D. The discrepant results for changes in this Pseudomonadota could arise from a different status at baseline as stratification by the presence of complications is not usually performed. Thus, our result is likely related to higher Enterobacteriaceae abundance observed at baseline in subjects with T2D, which is consistent with previous reports in Mexican and other populations [34, 35]. Overall, it is clear that certain bacteria are favored by the new conditions generated after surgery; however, their role in metabolic improvements, if any, remains to be clarified [4].

In our study, despite that most patients achieved a successful %EWL, a greater increase in Veillonella abundance after surgery was associated with lower %EWL and greater BMI post-surgery. Veillonella is a commensal known as lactate utilizer [36]; however, its association with metabolic improvements after RYGB, when observed, has been inconsistent [37, 38]. Veillonella uses lactate for propionate production, which can increase the energy harvest from the diet [39]. Thus, we can speculate that, in the context of surgery-induced malabsorption, an increased energy harvest from the diet could limit weight loss. Additionally, in a large Mexican cohort Veillonella was identified as a “bad metabolic health marker” [34]. However, whether Veillonella is implicated in weight loss or it is simply a microbial signature remains to be clarified. In fact, the gut microbiota constitutes a complex ecosystem where microbes perform cohesive interactions as a community. Therefore, it would be interesting to perform an integrative analysis of the microbiota by coabundance as it may provide a better understanding of the functional role of Veillonella after surgery and its relationship with metabolic traits [40].

It is worth mentioning that some of the alterations in gut microbes after bariatric surgery are affected in a procedure-dependent manner [41]. Recent studies have shown that species of Veillonella are more enriched by RYGB than by SG [26, 38]. Interestingly, in our study, a lower richness and diversity before the surgery, in non-T2D patients, was related to the higher increase in Veillonella abundance. Gut microbial diversity has been suggested to be a marker of greater stability and robustness, providing functional redundancy and colonization resistance [42]. Thus, we could hypothesize that under the modified anatomical conditions of RYGB, a lower diversity of the gut microbiota could allow the colonization of oral-derived bacteria and thus predispose to a higher increase in Veillonella abundance and possibly to a lower %EWL. The absence of this correlation in T2D patients could be related to a different baseline gut microbial configuration; however, due to the small sample size of T2D patients (n = 6), a further analysis may not have sufficient power. Finally, although more studies are needed, it will be interesting to assess whether, in patients with a low microbial richness, other types of bariatric surgery, such as SG, could provide a greater metabolic benefit.

We acknowledge several limitations. First, the sample size may impact the power of the study, although it was still comparable with similar studies published so far [43]. Second, the use of fecal samples does not necessarily reflect intestinal metabolism; thus, for future studies, the use of intestinal biopsies for this type of study will greatly enhance the knowledge. Third, by using 16S gene sequencing, whether specific Veillonella species are related to weight loss could not be identified nor the bacterial pathways or genes related to this effect. Fourth, our results are based on bacterial relative abundance. Despite this approach being similar to other studies [44‒47], it has been shown that changes in microbial biomass can influence the results. Fifth, information on samples consistency, gastrointestinal movements, or other factor that can significantly influence gut microbial or composition or the outcome such as diet and exercise were not collected. Thus, our findings may be interpreted with caution. Lastly, the follow-up was only a year, thus whether this increase also reflects long-term weight regain is still warranted. Notwithstanding these limitations, strengths of this study are that the fecal samples were collected before the presurgical diet, which is a common confounder in these types of studies [48], and that RYGB was performed by the same surgical team, which decreases some degrees of the variability of the results. Finally, up to our knowledge, this is the first report on Mexican patients which is an understudied population with a great prevalence of obesity. Thus, the present findings could improve the understanding of factors that influence weight loss after surgical treatment and identify strategies to improve its efficacy. In summary, our observation that the baseline gut microbial signature influenced taxonomic changes associated with weight loss, in the setting of RYGB and its metabolic benefits, contributes to the characterization of gut microbial profiles that could be clinically relevant to develop precision therapies considering the microbiome to improve weight loss and health outcomes [49].

We thank Dr. José Adolfo Pérez-Ramírez for his collaboration to this work.

This study protocol was reviewed and approved by the Ethics Committee of Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, approval number DIA-14040-14, and conducted in accordance with the Declaration of Helsinki and the International Conference on Harmonization Guidelines for Good Clinical Practice. All the patients provided written informed consent before any protocol procedure. STROBE guidelines were followed during the preparation of this report.

The authors have no conflicts of interest to declare.

This work was supported by a CONACYT research grant SALUD-2014-1-23425 to M-RF.

B.A.-P., M.S.-S., M.H., and M.R.-F. designed and conducted the study; S.M.-R. performed microbiota analysis; S.M.-R., M.R.-F., and B.A.-P. performed data analysis and interpretation; R.S.-C., A.T.-Q., V.S.-F., and E.G.-O. contributed to data collection and database generation; E.G.-O. and D-R.A. processed biological samples; S.M.-R., B.A.-P., and M.R.-F. wrote the manuscript; and C.A.-S. contributed to the discussion. All authors read and approved the manuscript.

Bárbara Antuna-Puente and Marcela Rodríguez-Flores contributed equally to this work.

All data generated or analyzed during this study are included in this article and its online supplementary material. Further inquiries can be directed to the corresponding author.