Worse inflammatory profile in omnivores than in vegetarians associates with the gut microbiota composition

The role of dietary habits for risk and for protection against cardiometabolic diseases is largely recognized [1]. While a high-fat low-fiber western diet has been associated with type 2 diabetes mellitus, dyslipidemia and cardiovascular disease, a vegetarian diet with a beneficial cardiometabolic profile and lower rates of cardiovascular events [2, 3]. Moreover, the ability of diet to decrease all-cause and cardiovascular mortality was demonstrated prospectively in vegetarians from the adventist health study-2 [4]. It is known that the body fat accumulation is an underlying mechanism of cardiometabolic disease by triggering a low-grade inflammation and insulin resistance [5, 6]. More recently, there is a growing interest in the participation of gut bacteria mediating the diet-induced cardiometabolic risk [7‐9]. Animal studies have contributed to understanding how bacterial communities influence energy extraction, fat deposition, inflammatory status and insulin sensitivity [10‐13].

In humans, numerous factors like delivery type, breastfeeding, antibiotic use and dietary habits have been implicated in the interindividual gut microbiota variability and diseases risk [14‐18]. In contrast to the initial studies in lean and obese animals [11], metagenomic analyses have indicated that the identification of relative abundance of the two major phyla, Firmicutes and Bacteroidetes, may not be sufficient to predict body adiposity and obesity-related diseases [19].

Inconsistent results regarding the initially described increased Firmicutes/Bacteroidetes ratio (F/B ratio) in human obesity have motivated deepening the knowledge of the phylogenetic structure of microbial communities in population exposed to different environments [20‐22]. Studies in some populations have detected associations of dietary patterns with the microbiota composition and certain chronic diseases [21, 23‐25]. Underlying pathways of these associations are not completely clarified but it is recognized the ability of a high-fat diet to increase intestinal permeability allowing lipopolysaccharides access in bloodstream [26‐28]. A gut microbiome-induced endotoxemia has been associated with increased risk of type 2 diabetes mellitus [7, 13, 29].

Scarce data are available in South America countries characterized by a large variety of foods and dietary habits. The study ADVENTO—Analysis of diet and lifestyle for cardiovascular prevention in 7th-day Adventists (http://​www.​estudoadvento.​org), conducted in a sample of the Brazilian population, has represented a unique opportunity to investigate how vegetarianism and exposure to animal food are associated with the gut microbiota and cardiometabolic risk profile. We hypothesized that dietary-dependent microbial composition influences inflammatory status, insulin resistance, and cardiovascular risk in individuals undergoing distinct dietary habits.

The aims of this study were: to describe the abundance of major phyla and genera in the gut microbiota of non-diabetic Brazilians classified according to dietary types (strict vegetarian, lacto-ovo-vegetarian, and omnivore); and to compare their inflammatory status, insulin resistance index, and cardiovascular risk profile.

Among 268 participants, 54.2% were women and 41.4% had weight excess. Stratifying according to dietary type, 66 were strict vegetarians, 102 lacto-ovo-vegetarians, and 100 omnivores, and groups did not differ according to male-to-female ratios and age (49.6 ± 8.5, 49.6 ± 8.6, 49.1 ± 8.2 years, p = 0.878, respectively). The frequencies of weight excess [26% (95% CI 15–37) versus 38% (95% CI 29–47) versus 55% (95% CI 45–65), pre-diabetes [21% (95% CI 11–31) versus 29% (95% CI 20–38) and 36% (95% CI 27–45)], and hypertension [18% (95% CI 9–27) versus 26% (95% CI 17–34) and 33% (95% CI 24–43)] were higher in the omnivores than in lacto-ovo-vegetarians and strict vegetarians, respectively.

Thirteen phyla were identified and five were present in all the fecal samples analyzed. Large relative abundances of Firmicutes (40.7 ± 15.9%) and Bacteroidetes (39.5 ± 19.9%) were found (Fig. 1), followed by Proteobacteria. Stratifying by dietary type, the proportion of Firmicutes (Fig. 1, panel a) was lower and of Bacteroidetes was higher in the strict vegetarian group (Fig. 1, panel b) when compared to lacto-ovo-vegetarians and omnivores. No differences were found in the comparisons of the F/B ratio between subsets of participants with normal and weight excess, considering the whole sample as well as within each dietary type.

The abundance of OTUs according to dietary types was also compared at genus level. Strict vegetarians had higher rates of Prevotella genus that belongs to the Bacteroidetes phylum (Fig. 2, panel a), and a higher Prevotella/Bacteroides ratio (8.4 ± 0.3) than the other groups. On the other hand, strict vegetarians had a lower proportion of Faecalibacterium genus from the Firmicutes phylum than lacto-ovo-vegetarians, and both vegetarian groups had higher proportions than in omnivores (Fig. 2, panel b).

Differences of selected OTUs between omnivores and non-omnivores (strict plus lacto-ovo-vegetarians), members of phyla Firmicutes, Proteobacteria, and Actinobacteria, assessed by the DESeq2, were shown in Fig. 3. The omnivore group showed much higher abundances of genera Succinivibrio and Halomonas, which belong to Proteobacteria, while non-omnivores had more OTUs from Actinobacteria and Roseburia genus from Firmicutes, compared to the counterparts.

The omnivorous group showed higher mean values of anthropometric data and worse lipid profile than the others (Table 1). Non-significant increases in fasting plasma glucose was verified across the groups, but omnivores exhibited higher insulin and HOMA-IR values (p < 0.001). Medians of CRP, LPS, and TNF-α/IL-10 ratio increased gradually from the vegetarian toward the omnivorous group.

The hypothesis that diet should influence systemic inflammatory status, insulin resistance and cardiovascular risk profile, via gut microbiota composition, is supported in the present study. In a developing country population of South America, the association between dietary habits and abundance of certain bacterial genera was demonstrated. Non-diabetic Brazilians undergoing distinct dietary types showed that vegetarians had a more favorable gut microbiota composition, characterized by less Firmicutes and more Bacteroidetes than omnivores. Furthermore, among the Firmicutes there was a predominance of genera associated with beneficial phenotypes, while among the Bacteroidetes an expected higher proportion of Prevotella. These findings suggest that exposure to animal foods could favor a pro-inflammatory intestinal environment, favoring endotoxemia, systemic inflammation and insulin resistance that are involved in the deterioration of the cardiometabolic risk profile.

In agreement with other studies [1‐4, 30], the Brazilian strict vegetarian Adventists exhibited a low-risk cardiometabolic profile, particularly when compared to the omnivores. Lower frequencies of obesity, hypertension, and pre-diabetes were observed in the strict vegetarians, similar to the results of the adventist health study-2 [3, 4] and also coherent with reports of fewer cardiovascular events in strict vegetarians [30, 39, 40]. Their food preferences—vegetables, fruits, and whole grains—are rich in fibers and micronutrients, which contribute to reduce oxidative stress, an underlying mechanism of these diseases [41]. Fiber-derived short-chain fatty acids (SCFA), mainly butyrate, acetate, and propionate, are facilitated by the presence of certain commensal bacteria that belong to phylum Firmicutes [42‐44].

Our findings of lower proportion of Firmicutes and higher proportion of Bacteroidetes in strict vegetarians compared to omnivores may not be attributed to the differences in body adiposity of the participants. In fact, a meta-analysis did not confirm previous assumptions, based on from animal and human studies, that obesity was associated with an increased F/B ratio [19, 45].

Analyzing particularly the Firmicutes subpopulations, we found that strict vegetarians had an increased abundance of the two most recognized butyrate-producing bacteria—Roseburia and Faecalibacterium [43, 46] in comparison with omnivores. Butyrate is a major energy source for colonocytes and promotes the expression of tight junction proteins, enhancing the intestinal barrier function and consequently, it protects against the LPS translocation [43, 47]. Our finding of lower LPS concentration in the strict vegetarian group is coherent with these effects. Also, the anti-inflammatory action of butyrate due to the nuclear factor κB inhibition in colonic cells [43, 46] is supported by our results since lower values of inflammatory markers, CRP and TNF-α/IL-10 ratio, were detected in the same group. However, our study design precluded establishing cause-effect relationship. Therefore, despite belonging to the Firmicutes phylum, Roseburia and Faecalibacterium genera were shown to be associated with a beneficial metabolic profile in our vegetarians, characterized by lower body adiposity and better lipid profile and insulin resistance index, which are along the same line of previous reports [16, 48].

Bacteroidetes were relatively more frequent in the microbiota of strict vegetarians than in omnivores, and their subpopulations were mainly composed of the genera Prevotella and Bacteroides. A higher Prevotella/Bacteroides ratio was seen in the strict vegetarians. These findings are consistent with others that reported higher Prevotella abundance in individuals with a plant-based diet and predominance of Bacteroides in non-vegetarians [15, 16, 49, 50]. Also, investigations on the interaction between long-term dietary patterns and microbiota using genera clusters found an association of Prevotella enterotype with fiber-enriched diets, as well as Bacteroides enterotype with protein and animal fat [15]. One study is in disagreement, since no significant difference between the gut microbiota composition of vegetarians and omnivores was observed [47].

Our observations of higher LPS, CRP, TNF-α/IL-10 ratio, and HOMA-IR values in the omnivorous group reinforce previous hypothesis that a saturated fat-enriched diet could induce inflammation and insulin resistance [7, 26, 28]. We speculate that exposure to animal foods could have contributed to alter the gut microbiota composition favoring an increase in LPS and generating endotoxemia. LPS are present in outer membrane of gram-negative bacteria and its ability to reduce the expression of tight junction proteins and increase intestinal permeability were demonstrated [29]. Also, there is evidence that microbiota-derived LPS in circulation, by binding to TLR4, trigger inflammation, deteriorate insulin signaling, and cause metabolic disturbances [7, 51, 52]. TLR4-deficient mice were recognized as resistant to the inflammatory activation induced by obesity or free fatty acids and protected from insulin resistance [52]. We suppose that, in the omnivorous participants, animal food consumption could have favored an enrichment of gram-negative bacteria, increased gut permeability and activation of immune response. Their higher levels of some inflammatory markers corroborate for this pathophysiological mechanism, anticipating the alterations in traditional cardiovascular risk factors. Actually, we detected an overrepresentation in some OTUs (increased Succinivibrio and Halomonas abundances) from the Proteobacteria phylum, which is known as a major group of gram-negative bacteria that prefer proteins as main energy source [49]. A similar result has already been reported in another study in which Proteobacteria were more abundant in European children fed with western diet compared to Africans fed predominantly with vegetarian diet [16].

Also, overrepresentation of class Bacilli, belonged to Firmicutes phylum, was verified among omnivores, which could be somehow unexpected since these bacteria are not gram-negative. However, our result is concordant with a report of an association of high abundance of Bacilli with the western diet [53] and diabetes [54], both conditions associated with low-grade inflammation and insulin resistance. Inflammatory markers and HOMA-IR were, in fact, higher in our omnivorous participants compared to vegetarians.

In lacto-ovo-vegetarians, we observed a higher proportion of Firmicutes and among these, of the Faecalibacterium genus. This finding was previously reported [50] and it was suggested that dairy products and eggs might be substrates for these bacteria [55]. Since these participants were exposed to this kind of animal foods and an intermediate cardiometabolic risk level, between the strict vegetarians and omnivores, was expected. Despite clinical parameters within the normal ranges, they already exhibited signs of a pro-inflammatory and reduced insulin sensitivity condition.

The main limitation is related to the cross-sectional design impeding the establishment of temporal or causal relationships. Also, the lack of detailed nutrient information inquiry made our study even more speculative. On the other hand, our study has the strength of reporting data on the gut microbiota at a deep level in a considerable number of individuals consuming with distinct dietary patterns. As far as we know, this approach is unique in South America countries. These data were associated with traditional risk factors and emergent cardiometabolic markers, suggesting possible mechanisms by which diet-mediated bacteria could participate in the genesis of prevalent diseases linked by the insulin resistance.

We called attention to the importance of gut microbiota assessment for understanding how diet participates in the pathogenesis of cardiometabolic diseases. Whether deeper taxonomic classification could provide clues in the investigation of the pathophysiological mechanisms of these complex diseases requires further investigation.

In conclusion, our data support that there are differences in gut microbiota composition of individuals consuming distinct types of diet, who differ according to their inflammatory and metabolic profiles. Based on the findings relative to bacteria abundances and on their recognized actions in the metabolism, we suggest that exposure to animal foods may favor an intestinal environment which could trigger systemic inflammation and insulin resistance-dependent metabolic disorders.