Gut Microbiota Profile in Children with IgE-Mediated Cow’s Milk Allergy and Cow’s Milk Sensitization and Probiotic Intestinal Persistence Evaluation

Food allergy (FA) and, in particular, IgE-mediated cow’s milk allergy is associated with compositional and functional changes of gut microbiota. In this study, we compared the gut microbiota of cow’s milk allergic (CMA) infants with that of cow’s milk sensitized (CMS) infants and Healthy controls. The effect of the intake of a mixture of Bifidobacterium longum subsp. longum BB536, Bifidobacterium breve M-16V and Bifidobacterium longum subsp. infantis M-63 on gut microbiota modulation of CMA infants and probiotic persistence was also investigated. Gut microbiota of CMA infants resulted to be characterized by a dysbiotic status with a prevalence of some bacteria as Haemophilus, Klebsiella, Prevotella, Actinobacillus and Streptococcus. Among the three strains administered, B. longum subsp. infantis colonized the gastrointestinal tract and persisted in the gut microbiota of infants with CMA for 60 days. This colonization was associated with perturbations of the gut microbiota, specifically with the increase of Akkermansia and Ruminococcus. Multi-strain probiotic formulations can be studied for their persistence in the intestine by monitoring specific bacterial probes persistence and exploiting microbiota profiling modulation before the evaluation of their therapeutic effects.


Introduction
Food allergy (FA) is a common chronic condition characterized by hypersensitivity reactions to a certain food, such as cow's milk proteins [1].
The prevalence of IgE-mediated cow's milk allergy (CMA) in infancy and early childhood has been estimated at 2-3% worldwide [2]. In Europe, a large cohort study confirmed challenge-proven cow's milk allergy in <1% of children up to 2 years old [3]. Children with CMA in the first year of life carry an increased risk of being affected by subsequent other atopic diseases [4,5]. Therefore, understanding CMA etiology is needed for the management and prevention of the disease and its later impact on patients.

Demographic and Clinical Profiling of Subjects
During the study period of 3-years, 124 patients aged 10-15 months were evaluated for suspected CMA at our hospital units. Ninety-seven patients were excluded: 14 were treated with proton-pump inhibitors (PPI), 22 were still partially breastfed, 38 had taken a course of antibiotic therapy in the previous 6 months, 52 had probiotics and 55 had functional foods in the last 2 months; 67 infants had more than one reason for exclusion.
The main reasons for exclusion were the use of probiotics and feeding an extensively hydrolysed cow's milk formula with added functional foods (probiotics or prebiotics), prescribed by the paediatrician to replace breast milk (52 cases).
The study was proposed to 27 children. As one of them was diagnosed with celiac disease soon after the double-blind placebo-controlled food challenge (DBPCFC) he was excluded, hence 26 children were enrolled (median age 12.8 months).
The CMA group included 14 infants, the CMS group 12 DBPCFC-negative infants, and the Healthy group 14 age-and sex-matched healthy infants (Table 1). There was no significant difference between groups in anthropometric characteristics including sex, age, mode of delivery, duration of breastfeeding, and duration of cow's milk-free diet ( Table 1). Three out of 14 (21.4%) in the CMA group, 3/12 (25%) in the CMS group and 6/14 (43%) in the Healthy group were delivered by caesarean section (p = 0.6). Thirty-eight infants had been breastfed for more than two months, exclusively or complementarily (Table 1).
Skin prick test with fresh cow's milk returned positive in 11/14, 6/12, 0/14 in CMA, CMS and Healthy groups, respectively. Specific IgE for cow's milk was positive in 13/14 and 9/12, in CMA and CMS groups respectively.
After DBPCFC, all CMA infants followed a cow's milk protein's free diet and regularly took the probiotic mixture twice daily for 30 days without any relevant adverse events.

RT-PCR Analysis
At baseline, significant differences between B. breve and B. infantis and between B. longum and B. infantis were found for CMA and Healthy groups, respectively (Figure 1, Supplementary B. breve and B. longum were present in the gut microbiota of the three groups in the order of 10 3 -10 5 molecules/ul (Figure 1), without significant differences (Supplementary  Table S3). No trace of B. longum subsp. infantis was detected in the three groups.
During the probiotic intervention, no significant variations of B. breve and B. longum subsp. longum amounts were detected in the CMA group, while the amount of B. longum subsp. infantis, significantly increased from T1 to T3 and decreased since point T4 (Figure 2,  Supplementary Table S4).  B. breve and B. longum were present in the gut microbiota of the three groups in the order of 10 3 -10 5 molecules/ul (Figure 1), without significant differences (Supplementary  Table S3). No trace of B. longum subsp. infantis was detected in the three groups.

Gut Microbiota Profiling
During the probiotic intervention, no significant variations of B. breve and B. longum subsp. longum amounts were detected in the CMA group, while the amount of B. longum subsp. infantis, significantly increased from T 1 to T 3 and decreased since point T 4 (Figure 2, Supplementary Table S4). B. breve and B. longum were present in the gut microbiota of the three groups in the order of 10 3 -10 5 molecules/ul (Figure 1), without significant differences (Supplementary  Table S3). No trace of B. longum subsp. infantis was detected in the three groups.
During the probiotic intervention, no significant variations of B. breve and B. longum subsp. longum amounts were detected in the CMA group, while the amount of B. longum subsp. infantis, significantly increased from T1 to T3 and decreased since point T4 (

Gut Microbiota Profiling
Alpha diversity was calculated using Observed, Chao1 and Shannon indices in order to evaluate Operational Taxonomic Units (OTUs) evenness, rarity and richness for CMA, CMS, and Healthy groups, at baseline. No statistically significant differences were found in the three indices comparisons. However, CMS showed the highest number of total and rare OTUs (i.e., Observed and Chao I index, respectively), while CMA was characterized by the lowest OTUs richness (i.e., Shannon index) ( Figure 3).
Alpha diversity was calculated using Observed, Chao1 and Shannon indices in order to evaluate Operational Taxonomic Units (OTUs) evenness, rarity and richness for CMA, CMS, and Healthy groups, at baseline. No statistically significant differences were found in the three indices comparisons. However, CMS showed the highest number of total and rare OTUs (i.e., Observed and Chao I index, respectively), while CMA was characterized by the lowest OTUs richness (i.e., Shannon index) ( Figure 3). Beta diversity was computed to estimate the phylogenetic relatedness (i.e., unweighted UniFrac algorithm) of CMA, CMS and Healthy groups. CMA group was clearly separated from the other groups, as verified by the PERMANOVA test (p = 0.019) (Supplementary Figure S1).
After probiotics intake, no statistically significant changes of alpha diversity were observed in the CMA group. At T5 a statistically significant decrease was observed for all alpha diversity indices ( Figure 4).  Beta diversity was computed to estimate the phylogenetic relatedness (i.e., unweighted UniFrac algorithm) of CMA, CMS and Healthy groups. CMA group was clearly separated from the other groups, as verified by the PERMANOVA test (p = 0.019) (Supplementary Figure S1).
After probiotics intake, no statistically significant changes of alpha diversity were observed in the CMA group. At T 5 a statistically significant decrease was observed for all alpha diversity indices ( Figure 4). Alpha diversity was calculated using Observed, Chao1 and Shannon indices in order to evaluate Operational Taxonomic Units (OTUs) evenness, rarity and richness for CMA, CMS, and Healthy groups, at baseline. No statistically significant differences were found in the three indices comparisons. However, CMS showed the highest number of total and rare OTUs (i.e., Observed and Chao I index, respectively), while CMA was characterized by the lowest OTUs richness (i.e., Shannon index) ( Figure 3). Beta diversity was computed to estimate the phylogenetic relatedness (i.e., unweighted UniFrac algorithm) of CMA, CMS and Healthy groups. CMA group was clearly separated from the other groups, as verified by the PERMANOVA test (p = 0.019) (Supplementary Figure S1).
After probiotics intake, no statistically significant changes of alpha diversity were observed in the CMA group. At T5 a statistically significant decrease was observed for all alpha diversity indices ( Figure 4).  To detect differences in OTU composition, we compared the three groups, at phylum and genus levels.
At the phylum level, Verrucomicrobia was highest in Healthy and gradually decreased from CMS to CMA group (p < 0.05) (Supplementary Table S5). On the contrary, Firmicutes resulted more abundant in the CMS group and lowest in the healthy subjects (Supplementary Figure S3).
At the genus level, Haemophilus, Actinobacillus, Prevotella, Klebsiella and Streptococcus were associated with the CMA group (p < 0.05, Supplementary Table S6) while Parabacteroides and Granulicatella were more abundant in the Healthy group ( Figure 5).
No clearly defined patient clusters along the entire time-course was detected (PER-MANOVA = 0.829), (Supplementary Figure S2).
To detect differences in OTU composition, we compared the three groups, at phylum and genus levels.
At the phylum level, Verrucomicrobia was highest in Healthy and gradually decreased from CMS to CMA group (p < 0.05) ( Supplementary Table S5). On the contrary, Firmicutes resulted more abundant in the CMS group and lowest in the healthy subjects (Supplementary Figure S3).
At the genus level, Haemophilus, Actinobacillus, Prevotella, Klebsiella and Streptococcus were associated with the CMA group (p < 0.05, Supplementary Table S6) while Parabacteroides and Granulicatella were more abundant in the Healthy group ( Figure 5). To evaluate the influence of probiotic intake on gut microbiota modulation, we tested the OTU distribution during the T0-T5 time-course in the CMA group. At the phylum level, there was an increase of Verrucomicrobia during probiotic intake with a peak of abundance at T3 and a rapid decrease at T5. Proteobacteria showed a gradual increase during probiotic intake and this increment was maintained also at T5. On the contrary, Actinobacteria decreased during the time-course, even if an increase was registered from point T1 to point T3 ( Figure 6, Supplementary Table S7).
At the genus level, probiotic intake determined an increase of Akkermansia, Prevotella and Ruminococcus that was maintained also at T5. Blautia increased during the entire period of probiotic intake, until T3 point, while at T5 it started to decrease showing a level of abundance lower than that observed at T0. On the opposite way, Actinomyces, Enterococcus, Streptococcus and Sutterella resulted diminished after probiotic intervention (Figure 7, Supplementary Table S8). To evaluate the influence of probiotic intake on gut microbiota modulation, we tested the OTU distribution during the T 0 -T 5 time-course in the CMA group. At the phylum level, there was an increase of Verrucomicrobia during probiotic intake with a peak of abundance at T 3 and a rapid decrease at T 5 . Proteobacteria showed a gradual increase during probiotic intake and this increment was maintained also at T 5 . On the contrary, Actinobacteria decreased during the time-course, even if an increase was registered from point T 1 to point T 3 ( Figure 6, Supplementary Table S7).
At the genus level, probiotic intake determined an increase of Akkermansia, Prevotella and Ruminococcus that was maintained also at T 5 . Blautia increased during the entire period of probiotic intake, until T 3 point, while at T 5 it started to decrease showing a level of abundance lower than that observed at T 0 . On the opposite way, Actinomyces, Enterococcus, Streptococcus and Sutterella resulted diminished after probiotic intervention (Figure 7, Supplementary Table S8).

Discussion
Several comparative studies have been conducted on the intestinal microbiota composition of healthy versus allergic children, but their results are controversial [7,8,[23][24][25].
To the best of our knowledge only one study evaluated gut microbiota differences in food-allergic versus food sensitized and healthy controls [18].
We observed a lower microbial diversity in CMA infants, both in terms of richness and evenness indicating that bacterial OTUs are not equally abundant in the gut ecosystem, hence generating a disequilibrium or compositional dysbiosis. Indeed, previous studies found low intestinal diversity in CMA patients [26], supporting the theory of "microbial deprivation syndromes of affluence" [27]. According to this theory, the reduced microbial stimulation could lead to abnormal immune development in early life, a possible contributory cause of allergic disease [28]. This is in line with other previous results obtained for atopy and food allergy [29][30][31].
In our study harmful bacteria as Haemophilus and Klebsiella, Prevotella were associated with CMA.
Prior studies investigating the association of gut microbiota and allergic disease during infancy demonstrated inverse associations of Haemophilus with eczema [32] and food sensitization [18]. In vitro studies demonstrated that the pro-inflammatory activity of this bacteria leads to activation of inflammatory Th 2 pathways suggesting a possible role in allergy manifestation [33].
Additionally, Klebsiella spp. have been associated with pro-inflammatory responses [34,35]. Moreover, a higher abundance of this genus has been correlated with atopy in infants [23]. So, our results confirm its possible association to later development of pediatric allergy [36,37].
Interestingly, Haemophilus, Prevotella, Actinobacillus and Streptococcus gradually decreased in the gut microbiota from CMA to CMS to Healthy infants, leading to suppose a correlation of these bacteria to allergic status.
Probiotics have been proposed as a strategy for the management of CMA allergy. However, not all the studies based on probiotic intake demonstrate the actual ability of probiotics to colonize the GT. This makes it difficult to decide which strain and at what doses to administer it.
For this reason, we investigated the persistence of TRIBIF®probiotic strains in the gut microbiota at defined time-points. The mixture we studied is reported to get a significant improvement of symptoms and quality of life in allergic rhinitis and intermittent asthma [38], but no data have yet been provided about the real presence of probiotics in patients' microbiota.
From our study, B. breve and B. longum subsp. longum are present in the gut microbiota without significant differences in CMA, CMS and Healthy groups. B. breve and B. longum subsp. longum, commonly present in the gut of breastfed infants, are able to metabolize certain human milk oligosaccharides (HMOs) and carbohydrates released by other bifidobacteria. Moreover, B. longum subsp. longum persist in the gut microbiota also after weaning, due to its ability to utilize such diet-derived carbohydrates [39]. Our data confirm that they persist after weaning, and the probiotic supplementation has not produced any evaluable effect on the concentration of these species, which were already abundant in the infant gut microbiota. By contrast, B. longum subsp. infantis was totally absent at baseline in all groups. During probiotic intake, its amount increased significantly during the first 30 days, while it decreased after probiotic discontinuation. This indicates the survival of the bacteria during their transit in the gastrointestinal tract and their ability to colonize and to persist at least for some time in the gut microbiota of children affected by IgE-mediated CMA.
B. longum subsp. infantis is unique among gut microbes for its capacity to transport into its cytoplasm and to consume the full range of HMOs [40]. Then, it is presumable that B. infantis is a very early colonizer of the infant gut during breastfeeding, whereas after weaning this species is replaced by others more adaptable to the new environment. This may be the cause of the absence of B. infantis in all the groups at baseline.
It is known that the prolonged colonization of this species is a rare phenomenon [41], while both probiotic inoculation and continued provision of HMOs are likely to foster long-term colonization [42]. Our results are in agreement with these studies, indicating that between 10 and 15 months the colonization of B. infantis persists only if there is a continuous intake.
A beneficial impact of probiotic intake on gut microbiota modulation could be associated with the significant increase of Akkermansia that was also maintained after probiotic suspension. A. muciniphila is involved in the maintenance of intestinal barrier functions and in the containment of intestinal inflammation, playing a pivotal role in the host's overall health status [43].
Additionally, the persisting reduction of Streptococcus after the probiotic supplementation [44] can be considered a positive effect of the intervention, as Streptococcus is a potentially pathogenic microorganism that could be directly involved in atopy [45] especially considering that it resulted in a biomarker of CMA in the present study.
Blautia distribution increased during all the period of probiotic intake, until T 3 point, while at T 5 it started to decrease, suggesting a possible synergistic mechanism with the probiotic species for gut microbiota colonization.
Additionally, Ruminoccocus increased during the probiotic intake and after probiotic suspension. Interestingly, low relative abundance of Ruminococcaceae were registered at one week of age in infants that developed IgE-associated eczema, and low abundance of the genus Ruminococcus at that age was associated with exaggerated Toll-Like Receptor (TLR)-2-induced interleukin (IL)-6 and Tumor necrosis factor (TNF)-α responses at 6 months [46]. The increase of Ruminococcus during and after probiotic intake may be beneficial also considering its ability to produce ruminococcins, such as ruminococcin A, a bacteriocin that can inhibit the development of Clostridium species [47].
Our results indicate that B. infantis could have the biological plausibility for therapeutic actions for CMA infants, while the two other strains administered probably do not. As multi-strain probiotic formulations are largely marketed, we suggest that, according to the 5th Bradford Hill criterion [48], an evaluation of the colonization and persistence of their specific probiotics in the intestine should precede clinical studies.
This study has few limitations. It is not an efficacy study: we chose don't include any clinical evaluation of the trend of milk allergy symptoms, because we had no indication of the kinetics of the probiotic. Moreover, this probiotic formulation has not been described in efficacy studies in allergic children.
Another limitation is the small number of patients involved. However, we performed this study on a sample of children with homogeneous feeding, in a strict range of age which guarantees the uniformity stability of the gut microbiota composition.

Patients Enrolment
This was a prospective multicentric study. Infants aged 10-15 months, evaluated for suspected IgE-mediated CMA, were consecutively enrolled at the Allergy Division of Bambino Gesù Children's Hospital in Rome between January 2016 and December 2018. Infants who were still breastfed, those with gastrointestinal disease in the last 30 days, metabolic or congenital disorders were not included. We excluded infants who had been administered with antibiotics in the last 6 months, or probiotics, proton-pump inhibitors [49], fermented milk and other functional foods in the 4 weeks prior to the onset of the study.
For each patient, informed consent was obtained from their parents, and anamnestic data were recorded. The sensitivity to cow's milk was documented by a positive skin prick test (cut-off 3 mm wheel diameter) and/or allergen-specific IgE (ImmunoCAP ® Thermo Fisher Scientific, Uppsala, Sweden) with 0.35 kU/L as a lower limit of eligibility, as described elsewhere [50].
In positive patients, DBPCFC was then performed.

Food Challenges
Cow's milk was administered in seven incremental doses of 0.1, 0.3, 1, 3, 10, 30, and 100 mL at 30 min intervals for three and a half hours. Neocate ® was used as a placebo and the milk without lactose was blinded according to the standardized AAAAI/Europrevall protocol [51,52]. Oral food challenge was discontinued at the first onset of objective symptoms [53]. Patients were observed up to 6 h after starting the test.

Study Design and Faecal Sample Collection
The included patients were divided into two groups: CMA: patients sensitized to cow's milk proteins with DBPCFC-confirmed IgE-mediated cow's milk allergy; CMS: sensitized but tolerant to cow's milk proteins patients (negative DBPCFC). A group of age-and sex-matched children without suspect of CMA, regularly formulafed and subjected to the same exclusion criteria, was recruited at our immunization clinic (Healthy group).
For each subject (CMA, CMS, Healthy groups), a stool sample was collected at enrolment and stored at −80 • C at the Human Microbiome Unit of Bambino Gesù Children's Hospital in Rome until further processing.
Patients belonging to the CMA group were treated with a probiotic mixture (Bifidobacterium breve M-16V, Bifidobacterium longum subsp. longum BB536 and Bifidobacterium longum subsp. infantis M-63; TRIBIF ® , Valeas SpA, Milan, Italy). A dose of 3.5 × 10 9 colony forming units (CFU)/dose of each species was administered twice per day for 30 consecutive days. Fecal samples were collected during clinical visits at time T 0 (no probiotic intake), T 1 (7 days of probiotic intake), T 2 (15 days of probiotic intake), T 3 (30 days of probiotic intake), T 4 (after 60 days from probiotic discontinuation), T 5 (after 90 days after probiotic discontinuation). All samples were stored at −80 • C at the Human Microbiome Unit of Bambino Gesù Children's Hospital in Rome until further processing.
After the oral challenge, all the positive children were prescribed a milk-free diet with an extensively hydrolysed whey formula not containing prebiotics or probiotics (Aptamil SP, Danone-Nutricia, Utrecht NL, The Netherlands). Healthy controls continued to take their regular formula, while CMS children were allowed a free diet with the substitution of a formula not enriched with functional foods (Mio 2, Nestlé Vevey, Switzerland).
This study, designed according to the Declaration of Helsinki, was registered on Clinicaltrials.gov (ClinicalTrials.gov Identifier: NCT03639337; 21 August 2018), and was approved by Bambino Gesù Children's Hospital IRCCS Ethics Committee (protocol number 787_OPBG_2014).

Cloning and RT-PCR Assay Design
The pGEM ® -T Easy Vector System kit (Promega, Milan, Italy) was used to clone the purified PCR fragment into the PGEM vector with the use of Escherichia coli competent cells. The extraction of recombinant plasmids pGEM-BB (pGEM+B. breve), pGEM-BL (pGEM+B. longum subsp. longum) and pGEM-BI (pGEM+B. longum subsp. infantis) was then performed (Plasmid Miniprep Kit, Promega, Italy). Obtained DNA was quantified using the NanoDrop ND-1000 spectrophotometer and serial dilutions of plasmids were prepared in the range from 10 6 to 10 1 vector copy numbers and used to create a standard curve for the quantitative RT-PCR (qRT-PCR) assays. ITS rDNA cloned fragments were amplified and sequenced (Genetic Analyser 3500, Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions. FASTA sequences were aligned with CLUSTAL-W software (http://www.ebi.ac.uk/clustalw/) (accessed on 1 February 2021) and species-specific primers and TaqMan probes (Roche Diagnostics) design was carried out (Supplementary Table S1).

Stool Samples DNA Extraction
DNA extraction was executed on 200 mg of stool sample by the use of QIAmp Fast DNA Stool mini kit (Qiagen). The obtained DNA was quantified by NanoDrop ND-1000 and roughly diluted to 16 ng/µL for a total amount of 80 ng per assay.

qRT-PCR Assay
The amounts of B. breve, B. longum subsp. longum and B. longum subsp. infantis in faecal samples were measured by qRT-PCR in the Light Cycler 480 instrument (Roche Diagnostics, Mannheim, Germany) and following the procedure already described in detail [54].

16S rRNA Targeted-Metagenomics
The 16S rRNA targeted-metagenomics was performed on T 0 samples of CMA, CMS, and Healthy groups and on the follow-up samples of the CMA group (T 1 , T 3 and T 5 time-points).
The variable region V3-V4 of the 16S rRNA gene (∼460 bp) was amplified by using the primer pairs 16S_F 5 -(TCG TCG GCA GCG TCA GAT GTG TAT AAG AGA CAG CCT ACG GGN GGC WGC AG)-3 and 16S_R 5 -(GTC TCG TGG GCT CGG AGA TGT GTA TAA GAG ACA GGA CTA CHV GGG TAT CTA ATC C)-3 reported in the 16S Metagenomic Sequencing Library Preparation protocol (Illumina, San Diego, CA, USA). The first PCR reaction was set up using Fast Start Hifi Taq (Roche Diagnostics) with the following conditions: initial denaturation at 95 • C for 3 min, 32 cycles of denaturation at 95 • C for 30 s, annealing at 55 • C for 30 s, and extension at 72 • C for 30 s, and a final extension step at 72 • C for 5 min.
Twenty ul of KAPA Pure Beads (Roche Diagnostics) were employed to purify DNA amplicons. To obtain a unique combination of bar-code sequences, a second amplification step was prepared using Nextera technology (Illumina). The final library was cleanedup using 50 µL of KAPA Pure Beads, quantified using Quant-iT™ PicoGreen ® dsDNA Assay Kit (Thermo Fisher Scientific) and diluted in equimolar concentrations (4 nM).
The following steps consisted of samples' pooling, denaturation and dilution to 7 pM. Samples were then sequenced on an Illumina MiSeqTM platform (Illumina) following the manufacturer's specifications to generate paired-end reads of 250 base-length.

Biocomputational and Statistical Analyses
Illumina Miseq reads were first analyzed for their quality, length and chimera presence using the Qiime v1.8 pipeline. Then, sequences were organized into Operational Taxonomic Units (OTUs) with a 97% of clustering threshold of pairwise identity. PyNAST v.0.1 program was used to carry out a multiple sequence alignment against Greengenes 13_08 database with a 97% similarity for bacterial sequences [55]. The OTU multiple sequence alignment was used as well to build a phylogenetic tree [56]. In order to filter rarely distributed OTUs, and avoid noise (i.e., artifact taxa), taxa with a threshold of prevalence greater than 0.05% within the whole dataset were selected. Alpha and beta diversity were performed by phyloseq of R package [57] and PERMANOVA test was applied on beta diversity metrics with 9999 permutations to compare samples at different time points. With the aim to identify microbial biomarkers, differences in the relative abundances of taxa between groups and during the time-course of the CMA group was assessed by DESeq2 test at phylum and genera taxonomic level. In particular, for the genus level, only taxa with a relative abundance higher than 0.01 were considered for the computational analysis [58]. SPSS statistical software (version 21) was used for graphical analysis of probiotic species persistence during the time-course and to perform Wilcoxon signed rank test for paired comparison between time-points (Supplementary Tables S2 and S3). Student's t-test was applied for comparison of clinical variables of the population.

Conclusions
In conclusion, our study revealed that the gut microbiota of children with CMA is characterized by the prevalence of harmful bacteria such as Haemophilus, Klebsiella and Prevotella which can be considered as possible biomarkers associated with the disease.
Moreover, we can confirm the ability of B. longum subsp. infantis to pass the gastrointestinal tract and to persist in the gut microbiota of infants with CMA.
Potentially beneficial modulation of the gut microbiota during the probiotic intake was also observed and was maintained for some genera as Akkermansia and Ruminococcus.
Considering the protective and anti-inflammatory properties of B. infantis [40], it would be interesting to investigate, in a clinical trial, its clinical effect on food allergy symptoms.
Moreover, it would be interesting in a future study to evaluate the gut microbiota profile in children with a history of allergy with a negative oral challenge and sensitized at the enrollment, and compare that data with our results on sensitized children in order to investigate common fingerprints in terms of gut microbiota composition.