Breastfeeding may have a long-term effect on oral microbiota: results from the Fin-HIT cohort

The Developmental Origins of Health and Disease (DOHaD) theory has been a target of many studies to explain the global epidemic of non-communicable diseases. DOHaD theory proposes that exposures during critical development periods, such as the first 1000 days of life, could unleash metabolic programming that is able to modify structure and function of organs and systems, impacting health status later in life [1‐3].

Recently, the role of microbiota in the context of DOHaD has been evaluated [4]. Gut microbiota is known to be established during the first two years of life, reaching stability by the third year [5, 6]. Perturbations of microbiota during early life have been associated with later inflammatory or immune-mediated diseases, such as allergy, asthma, and obesity [7‐10], indicating that microbiota also has a critical period of development.

Evidence of possible protection of breastfeeding against obesity and related outcomes in adulthood supports the DOHaD theory [11, 12]. An explanation for the protective effect is its role during the gut microbiota establishment period [7, 13]. Relative to formula-fed infants, breastfed infants seem to have a lower abundance of Clostridium and a predominance of Bifidobacteria and Lactobacilli in the gut [6, 8, 14‐18].

Studies have suggested that also the infant oral microbiota is influenced by the type of feeding [19‐21]. Proteobacteria and Actinobacteria phyla were more abundant in the oral cheek of breastfed neonates, while formula-fed infants had a higher predominance of Bacteriodetes [20]. In examining saliva culture data of three-month-old infants, vital Lactobacillus species were found in breastfed but not formula-fed ones [19]. A longer term effect, assessed at four and 12 months after birth, reinforced differences in oral microbiota composition between breast and formula-fed infants [21].

Although differences exist between gut and oral microbiota compositions, some similarities have been described [22‐24]. Community types present in these sites seem to be associated, i.e. types of bacteria detected in one site are able to predict those found in the other [22, 23]. It has been suggested that oral microbiota seeds gut microbiota in infancy. Observations that both mother’s milk and infant’s faeces are colonized by some identical bacteria support the hypothesis that human milk is an important source of bacteria, contributing to define gut microbiota composition [22].

Considering that oral microbiota is relatively stable over time in healthy individuals, that perturbations could favour non-oral diseases such as type 2 diabetes [24, 25], and that saliva is easily collected, the use of saliva samples represents a unique opportunity to investigate factors associated with microbial composition. As far as we know, no study has evaluated whether the association between breastfeeding and oral microbiota composition is maintained until adolescence, which is a critical period in the life course for the prevention of chronic diseases. We aimed to evaluate whether the type of feeding during the first six months of life is associated with oral microbiota diversity and composition in Finnish adolescents. Our hypothesis is that those who received infant formula at early life (combined or not with breastmilk) had a lower exposure to breastmilk (in quantitative terms) compared to those who were fed only with breastmilk, which could influence the microbiota development.

Of the 423 adolescents at baseline of Fin-HIT, 52% were female; their mean age and BMI were 11.7 years and 18.0 kg/m² (73% were normal weight), respectively. Most participants were born by vaginal delivery (81.1%) and received infant formula (solely or combined with breastmilk) (58.6%) during their first six months of life. The majority of participating parents were female (87.4%), Finnish speakers (90.3%), and had a university degree (59.6%) (Table 1).

Mean (standard deviation) values of alpha diversity indices were 2.9 (0.3) and 10.1 (3.1) for Shannon and Inverse Simpson, respectively, and these did not differ between the No formula and Infant formula groups (Table 1). Beta diversity was also similar between groups (P - value = 0.881) (Fig. 2). In the No formula group, vaginal delivery was more frequent than in the Formula group (88.3% vs. 76.1%, P - value = 0.002) (Table 1). Alpha diversity was not associated with adolescents’ or parents’ characteristics (Table 2).

In 423 samples of adolescents’ saliva, 22,924,455 raw sequences reads were obtained and 1049 OTUs were evaluated. In total, 11 phyla, 16 classes, 24 orders, 44 families, and 76 genera were identified. As depicted in Fig. 3, a great predominance of phylum Firmicutes was found, followed by Bacteroidetes, Proteobacteria, and Actinobacteria (panel A), while the predominant genera were Veillonella, Prevotella, and Streptococcus (panel B).

Comparing groups of feeding type according to abundance of bacteria, no difference was observed at phylum level. Three OTUs (#019, #232, and #158) were more abundant in No formula than in Infant formula group (log2fold changes/ adjusted p-values: − 0.920/ < 0.001, − 0.328/ 0.001, and − 0.577/ 0.004, respectively) (Fig. 4; and see table at Additional file 1 for detailed information). These three OTUs belonged to phylum Firmicutes, class Clostridia, and order Clostridiales; OTU #019 belonged to genus Eubacteria and #232 and #158 to genus Veillonella. Considering the entire genera, genus Eubacteria (log2fold change − 0.742, adjusted p - value < 0.0003), but not genus Veillonella, was more abundant in the No formula group.

We investigated the long-term association of early life feeding type with oral microbiota composition in adolescence. A few differences emerged in abundances of some specific genera according to early feeding type, but our data do not suggest a significant impact on the diversity of microbiota. Abundances of commensal OTUs belonging to genera Veillonella and Eubacteria were higher in adolescents who had not been fed with formula than in others who had received infant formula, combined or not with breastmilk.

To the best of our knowledge this is the first study to explore associations of early feeding type with oral microbiota diversity and composition at this stage of life. Such relationships have been investigated mainly during the first year of life [19‐21]. Despite higher abundance of certain OTUs, the overall composition (beta diversity) or the alpha diversity of oral microbiota did not differ between adolescents with different early life feeding type. In general, high phylogenetic diversity in gut microbiota has been associated with several health outcomes [31], but long-term health implications of this picture are unclear yet. We speculate that our negative results regarding overall microbiota diversity and composition could result from a stronger impact of other exposures during childhood such as weaning age, time of introduction of solid foods, Westernized eating pattern and use of antimicrobials [18, 31‐33].

As far as the gut is concerned, it was reported that microbiota composition in breastfed babies tended to be more stable and has less diverse bacterial community compared with formula-fed babies [34, 35]. From three years of age, their composition converge to resemble that of an adult gut [5]. A multicentre cohort of 903 infants (including Finns) followed from three to 46 months of age showed that breastfeeding explained the greatest part of microbiome establishment during the first year of life. Also, breastfed infants had lower diversity than those weaned at an earlier stage. As age increased and breastmilk exposure decreased, diversity became more similar between groups, supporting the hypothesis that weaning is an important event affecting maturation of the gut microbiome [6]. Some studies have evaluated the long-term effect of breastfeeding on gut microbiota [6, 18]. A prospective birth cohort study of 281 Dutch children reported that breastfeeding duration was associated with gut microbiota composition in children aged 6–9 years [18].

Little is known how these changes occur in saliva, driven by diet, physical activity and antibiotics administration, which could alter the host health status. Although gut and oral microbiota had distinct compositions, community types of these sites seem to be associated [22, 24]. A study conducted with neonates 20 days old, found a positive correlation of subdominant family Lactobacillaceae in saliva and faeces samples of the same infant, and showed that Staphylococcus spp. and Streptococcus spp. were shared between the two ecosystems and also with human milk [22]. Comparable results have been reported in studies assessing gut and oral microbiota; i.e. early feeding seems to impact on microbiota profile [6, 14, 19, 20].

Comparisons with our findings are limited because we were the first to evaluate the long-term effect of early feeding on oral microbiota. Furthermore, exposures occurring during infancy, particularly related to antibiotics use, may affect microbiota composition mainly at the gut level [36, 37]. The protective role of breastfeeding against increased body adiposity and antibiotic use in childhood, especially by promoting beneficial microbiota, was eliminated by the use of antibiotics in early life [38]. Additionally, differences in microbiota composition from one population to another are recognized [39‐41].

According to unpublished data of the Fin-HIT, most of the participants used antimicrobial drugs during their first years of life, which could contribute to a lasting impact on oral microbiota diversity and composition at adolescence. Regarding diet, it is suggested that especially the high and frequent intake of sucrose and other fermentable carbohydrates result in the accumulation of acidogenic and aciduric microorganisms, driving a pathogenic biofilm community formation. Particularly, high frequency of sugar intake could disrupt the homeostasis between commensals and pathogens resulting in dysbiosis, which could increase chances of caries, inflammation and periodontitis in more susceptible individuals [42]. Thus, we suppose that sugar intake during childhood could be an important factor affecting diversity and composition of oral microbiota. A previous study, conducted in public child health service in Finland, showed that sugar introduction occurs early in life. Almost half of six month old children were receiving sugar-sweetened beverages and more than 90% of those older than 16 months were receiving sweets [43]. Further, another Finnish study showed that 95% of one year-old babies consume mass-produced baby foods, of which some are sweetened with juice-concentrates (70–80% of sugar) [44].

Our finding of a major predominance of phylum Firmicutes in the human oral cavity, followed by Bacteroidetes, Proteobacteria, and Actinobacteria is consistent with previous studies in one to two month-old [20] and 12 month-old children [21]. A study conducted in 38 healthy, full-term, vaginally delivered neonates from Australia found similar alpha diversity and proportion of Firmicutes between breastfed and formula-fed groups (96.3 vs. 95.3%) [20], in accord with our findings in an older age group. A higher relative abundance of Bacteroidetes and a lower abundance of Actinobacteria and Proteobacteria were also observed in formula-fed infants than in breastfed infant.

Holgerson et al. [19] showed that three month old exclusively breastfed infants clustered separately from formula-fed infants, according to their oral microbiota. In total, 14 probes differed significantly between these two feeding types. Among these probes, cultivable Lactobacilli and Eubacterium yurii were more abundant in breastfed than exclusively formula-fed infants, in whom the mean number of species per child was higher. Lactobacillus, but not Eubacterium, has been consistently reported in breastmilk, and thus, the presence of the former in infants’ oral microflora was expected. We do not know whether the higher abundance of Eubacteria observed in our breastfed adolescents could be detected earlier, during their infancy. Another study reported higher species richness in 4-month-old formula-fed infants and marked differences in saliva microbiota composition between feeding type subgroups [21]. However, at 12 months of age these differences were no longer significant.

We found that Eubacteria and Veillonella, both genera of the normal oral microbiota, were more abundant in adolescents of No formula group compared with those who received infant formula in early life. Eubacteria is an atypical genus in samples of breastmilk, but Veillonella, Gemella, Rothia, Lactobacillus, Streptococcus, and Staphylococcus have been previously described as the most abundant genera in breastmilk [45]. The association of No formula with higher Veillonella abundance seems favourable, considering that this genus has a beneficial effect on dental plaques by metabolizing lactate to weaker acids such as acetate and propionate [46]. Since lactate is a cariogenic factor produced by Streptococcus mutans, increased abundance of Veillonella could be interpreted as a compensatory effect to protect against dental injuries. Higher abundance of the species Veillonella parvula was previously described in subgingival biofilm samples from healthy subjects compared with those with chronic periodontitis, and its presence was inversely correlated with inflammatory biomarkers from gingival crevicular fluid [47]. Periodontal disease is considered a risk factor for cardiovascular disease, and the proposed underlying mechanism was based on increased circulating pro-inflammatory cytokines [48].

Strict anaerobic bacteria from the genus Eubacteria are chemoheterotrophs, i.e. are unable to synthesize their own organic molecules, requiring mixed organic acids from the host. Thus, Eubacteria energy is obtained from carbohydrates or protein metabolism, resulting in end-products such as butyrate and acetate [49]. These short-chain fatty acids have been associated with beneficial effects, such as cardiovascular and colonic disease prevention [50]. The presence of Eubacteria in our sample of healthy adolescents was expected since this a member of normal saliva microflora and plaques. Our methods were unable to identify species of Eubacteria that would be desirable since periodontal pathogenic Eubacteria species were previously described [47, 51]. A study of patients undergoing haemodialysis identified higher taxa of Eubacterium nucleatum in sulcular fluid from the periodontal pocket of non-diabetics than diabetics. Despite being a pathogen, the prevalence of periodontitis was similar between the groups [51].

Although evidence has suggested beneficial roles of Veillonella and Eubacteria, it was not possible to extrapolate these in our sample since the study design does not allow comparing health outcomes according to abundances of specific bacteria. Prospective studies are needed to investigate whether individuals with distinct abundances of specific bacteria have different chances of developing cardiovascular disease or other outcomes.

A strength of this study was the large sample size in which next-generation sequencing technology was used to analyse microbiota. This is a sub-study of a well-designed prospective cohort, including reliable information obtained from standardized questionnaires and national health registers [26]. Statistical analyses were duly adjusted, minimizing confounding bias. The main limitations of our study comprised a design impeding establishment of causality and retrospective data collection (feeding type answered by parents), which could generate recall bias. Considering that we evaluated associations of an event occurred long time ago (type of feeding) and another occurring in adolescence, there is a possibility of recall error, especially among parents who had multiple children. However, it has been reported that information about breastfeeding recalled by mothers after 20 years still shows reasonable accuracy [52]. We could not consider “exclusive breastfeeding” since no information regarding water, tea, and food intake in early life was available, or “exclusive formula-fed” (not combined with breastmilk) since its prevalence in our study was relatively low (2.4%). Previously, these categories of “exclusive” feeding type were clearly separated by microbiota composition, but partially breastfed infants were interspersed between these groups [19]. Despite this limitation that could have attenuated differences, we still detected some differences according to feeding type. We speculate that differences identified are explained by a higher exposure to breastmilk at early life by participants of the No formula group compared to those of the Infant formula group.

Finally, the lack of oral health status could be regarded as a limitation. It is known that diet, especially in combination with poor oral health, salivary dysfunction, scarce fluoride exposure and poor oral hygiene, can modulate oral microbiota resulting in dysbiosis [42]. Since Finland public oral health care services are freely available for individuals younger than 18 years, most children and adolescents have relatively good oral health, although differences exist depending on maternal educational level and region. Not having information on oral health could contribute to confounding bias and thus is a limitation in this study.

The type of feeding in early life is not associated with overall changes in composition or diversity but with some alterations in adolescents’ oral microbiota (higher abundance of Veillonella and Eubacteria). The clinical relevance of these findings and their health implications need to explored in future studies. Other exposures during childhood (e.g. e.g. antimicrobial use, sugar consumption and oral health) could play a stronger role in adolescents’ oral microbiota diversity and composition than the early feeding. Further prospective studies, considering the type of feeding in early life, age of weaning, and other factors during childhood, are needed to confirm our findings.