Current insights into cow's milk allergy in children: Microbiome, metabolome, and immune response—A systematic review

The increasing prevalence of IgE‐mediated cow's milk allergy (CMA) in childhood is a worldwide health concern. There is a growing awareness that the gut microbiome (GM) might play an important role in CMA development. Therefore, treatment with probiotics and prebiotics has gained popularity. This systematic review provides an overview of the alterations of the GM, metabolome, and immune response in CMA children and animal models, including post‐treatment modifications. MEDLINE, PubMed, Scopus, and Web of Science were searched for studies on GM in CMA‐diagnosed children, published before 1 March 2023. A total of 21 articles (13 on children and 8 on animal models) were included. The studies suggest that the GM, characterized by an enrichment of the Clostridia class and reductions in the Lactobacillales order and Bifidobacterium genus, is associated with CMA in early life. Additionally, reduced levels of short‐chain fatty acids (SCFAs) and altered amino acid metabolism were reported in CMA children. Commonly used probiotic strains belong to the Bifidobacterium and Lactobacillus genera. However, only Bifidobacterium levels were consistently upregulated after the intervention, while alterations of other bacteria taxa remain inconclusive. These interventions appear to contribute to the restoration of SCFAs and amino acid metabolism balance. Mouse models indicate that these interventions tend to restore the Th2/Th1 balance, increase the Treg response, and/or silence the overall pro‐ and anti‐inflammatory cytokine response. Overall, this systematic review highlights the need for multi‐omics‐related research in CMA children to gain a mechanistic understanding of this disease and to develop effective treatments and preventive strategies.


| BACKG ROU N D
One of the most common food allergies in early childhood is cow's milk allergy (CMA). 1,2Allergic reactions can be IgE-mediated, non-IgE-mediated, or a mix of both. 35][6][7] At present, infants diagnosed with CMA are placed on an elimination diet consisting of an extensively hydrolyzed formula (EHF) or, if symptoms persist, an amino acid formula (AAF). 8cause of the increasing evidence linking food allergies with alterations in gut microbial composition, 9,10 modifying the gut microbiome (GM) with probiotics, prebiotics, or synbiotics has emerged as a promising way to prevent and treat allergies. 11However, there is still little mechanistic understanding of how the GM influences host immune health, leading to allergies, including CMA. 12 Recent technological innovations in the field of microbiome, proteomics, and metabolomics have opened new doors for research and provided opportunities to address the gap in understanding the role of GM in CMA.The objective of this systematic review is to further the understanding of the relationship between GM and CMA, by reviewing existing studies examining microbiome, metabolome, proteome, and immune response data on IgE-mediated CMA in children and animal models.

| Search strategy
A search in MEDLINE, PubMed, Scopus, and Web of Science was performed using the queries in Table S1.The search was limited to research articles published in English before March 1, 2023.

| Inclusion and exclusion criteria
Human case, case-control, and intervention studies were included only if they examined children with IgE-mediated CMA aged 0-12 years.The allergy had to be medically diagnosed by either a skin prick test (SPT) or an IgE-specific test combined with a cow's milk food challenge.In studies with fecal transplantation (FT), the IgE-mediated CMA status of the donor must be confirmed by the diagnosis criteria used for human studies.For studies reporting data on groups of subjects diagnosed with different types of CMA, only the group with IgE-mediated CMA was reviewed.For animal studies, only case-control, and intervention studies on models that included both sensitization and challenge steps were included.The studies were included only if they contained analytical data that examined the GM or metabolome and were excluded when they failed to meet the inclusion criteria, had unclear diagnosis, or involved antibiotic treatment.

| Study selection
Titles, abstracts, and methods were screened independently by two of the authors MVS, PZ, and DMH, and by a third author in case of disagreement.Subsequently, the full text of the studies marked as potentially eligible was retrieved and independently checked for eligibility by at least two of the authors MVS, PZ, DMH, and by a third author in case of disagreement or doubts.

| Data extraction
For human studies, the extracted data included general study details (author, year), participant information (age, sample size), CMA diagnosis, analytical data types, data acquisition techniques, measured analytical parameters, and significant results.For intervention studies, the intervention details were also extracted.If available, the age range for each group in the study was reported.When only the mean and standard deviation (SD) were available, the age was reported as mean ± SD.The results were split in two: increased and decreased variables between the compared groups.For animal intervention studies, the extracted data included general study details, model information, challenge information, intervention details, data acquisition techniques, measured analytical parameters, and significant results.

Key message
The gut microbiome (GM) may play an important role in the development of cow's milk allergy (CMA).Treatments targeting the GM, such as prebiotics, probiotics, and synbiotics, may therefore be key prevention and treatment strategies.This systematic review reports on 21 studies, including 13 human studies and 8 animal studies studying GM's relation to CMA.Our findings suggest that GM characterized by an enriched Clostridia class, reduced Lactobacillales order, and reduced Bifidobacterium genus is typical in CMA children.Our results highlight that mechanistic insights, which can be obtained by means of multiomics approaches, are required to study CMA and develop effective preventive and treatment strategies.

| Search strategy
Our search yielded 733, 479, 512, and 897 articles in, respectively, Scopus, PubMed, MEDLINE and Web of Science.Forty-nine studies were eligible for inclusion.Figure 1 shows the PRISMA 13 flow diagram.Of the 49 papers, 28 were excluded after careful consideration Table S2.

| Human studies
CMA diagnosis criteria and measured parameters in human studies are summarized in Table S3.
Two studies applied specific probes to target certain bacteria groups, 14,15 and six used universal probes or primers to target the V3 region, 19 V4 region 16,22 or both. 17,20,21x studies compared α-and β-diversity between CMA group and HC, three of them noted increased 16,19 or decreased 20 Shannon αdiversity difference in the CMA groups, and one reported β-diversity (unweighted UniFrac) difference between CMA group and HC. 21single study reported a higher total bacteria count in the CMA group.18 Firmicutes, Bacteroidetes, Actinobacteria, Proteobacteria, and Verrucomicrobia were the primary reported GM phyla.15][16][17][18][19]21 These included: total Firmicutes 17,21 ; the class Clostridia 17 ; the families Lachnospiraceae 16 and Ruminococcaceae 16,17 ; the genera Clostridium, 14,19 Faecalibacterium, 16 Lactobacillus, 18 Ruminococcus 16 and Subdoligranulum 19 and the species Clostridium coccoides 15 and Clostridium celerecrescens.19 Conversely, certain Firmicutes phylum, including the genus Granulicatella 21 and the families Streptococcaceae, 16 Enterococcaceae, 16 and Acidaminococcaceae, 20 decreased in the CMA groups.Additionally, enriched bacteria of the Firmicutes phylum, including the class Clostridia, were also observed in the infants who outgrew CMA.22 Bacteroidetes phylum members also showed varying changes in the CMA groups.14,17,[19][20][21] These included increased levels of the Flavobacteriaceae family, 17 the Bacteroides 14,19 and Prevotella 21 genera, along with reduced abundance of the Prevotellaceae family 20 and the Parabacteroides genus.21 Furthermore, several bacteria from the Proteobacteria phylum, including the Haemophilus, Actinobacillus, and Klebsiella genera, 21 and the Escherichia coli species, 19 increased in the CMA groups.In contrast, total Proteobacteria, 17 the Enterobacteriaceae family, 16,18 and the Escherichia genus 16 decreased.19 Additionally, the Verrucomicrobia phylum dropped in the CMA group. 21 Tw studies reported certain bacteria only in the CMA group or the HC. Th Clostridium celerecrescens species, 19 and the Burkholderiaceae, Nannocystaceae, Shewanellaceae, Thermomonosporaceae, and Flavobacteriaceae families were reported only in the CMA group.17 In contrast, the Bifidobacterium bifidum species 19 and the Methylophilaceae and Dietziaceae families were exclusively detected in the HC. 17 Metabolome modifications. Dereased total short-chain fatty acid (SCFAs), 14,17 along with increased butyrate and total branchedchain short fatty acids (BCSFAs), 15 were reported in CMA groups.
Besides, lower pyruvate, lactate, threonine, and proline, along with higher total esters, ketones, alcohol aldehydes, uridine, histidine, tyrosine, trimethylamine-N-oxide (TMAO), and arginine/ histidine, 14 and elevated organic acids were reported in CMA groups. 23tabolome-microbiome associations.Two studies examined the association between the GM and the metabolome. 15,17Positive correlations were found between the Clostridium genus and butyrate, the Clostridium coccoides species and BCSFAs, and the Bacteroides genus and propionate. 15Isocaproate and BCSFAs were negatively related to the Bifidobacterium genus. 15Additionally, lactate was found to be negatively correlated with Bacteroides genus 17 and Clostridium coccoides species, 15 but positively correlated with Bifidobacterium genus. 15

Age
a AEDS as basic disease for subjects in both case and control groups, and the age is calculated by the pooled mean and SD from the age groups provided in the article.

TA B L E 1 (Continued)
TA B L E 2 Characteristics of studies that compare CMA infants/children before and after intervention (intervention study).
The interventions also affected the Proteobacteria phylum 21 and its members.The Betaproteobacteria class, the Burkholderiales order, the Alcalligenaceae family, and the Sutterella genus increased in the treated group, 26 while some studies reported decreased levels of the Deltaproteobacteria class, 26 the Enterobacteriaceae family 18 and the Sutterella genus. 21In the Bacteroidetes phylum, studies reported the interventions increased levels of the Porphyromondaceae family 26 and the Prevotella genus, 21,26 and reduced levels of the Bacteroides and Prevotella genera. 14Additionally, the Actinobacteria phylum also underwent changes with interventions. 14,18,21,25,26The use of probiotic Bifidobacterium strains consistently elevated the Bifidobacterium genus. 21,25,26Increased Bifidobacterium were also noticed after lactose-supplemented EHF diet. 14In contrast, the Actinobacteria phylum 21 and its members, the genera Bifidobacterium, 18 Atopobium, 21 and Actinomyces, 21,26 were decreased by the treatments.The Verrucomicrobia phylum and its Akkermansia genus were found to increase in the treatment group. 21 addition to the taxonomy changes, enhanced α-diversity (chao1, observed species), 26 reduced total bacteria, 24 and a decreased ratio of the Eubacterium rectale/Clostridium coccoides species 25 were reported after probiotics, pectin-based thickened AAF and synbiotics treatments, respectively.

Metabolome modifications. After the
LGG-supplemented hydrolyzed whey formula (HWF) diet, CMA patients showed increased kynurenate and decreased 3-indoleacetate. 23ditionally, butyrate increased in LGG-supplemented extensively hydrolyzed casein (EHC) formula-treated CMA patients. 16Meanwhile, lactose-supplemented EHF-raised SCFAs, lactate, threonine, uridine, histidine, tyrosine, methionine, TMAO, phenylalanine, arginine/histidine and gamma-amino-butyrate/ lysine, and lowered the total esters, ketones, alcohols, aldehydes and valine/isoleucine in CMA patients. 14mune response.The single intervention study reporting findings on the immune response showed that Bifidobacterium bifidum reduced allergy symptoms, lowered serum IgE, and raised IgG 2 levels in CMA patients. 26The IgG 2 and IgE were respectively positively and negatively correlated with GM α-diversity (Chao1 index, observed species, community diversity index, and Shannon index).The intervention decreased the pro-inflammatory cytokines TNFα, IL-1β, and IL-6 and increased the anti-inflammatory cytokine IL-10 as well. 26A outcome.5][26] Two studies noted significant improvement in allergic symptoms after treatment, 24,26 and one reported 5 out of 12 infants in the treated group outgrew CMA after 6 months, compared to none in the control group. 16

| Animal studies
The animal studies include two studies on the GM, metabolome, and immune response, 27,28 four on the GM and immune response [29][30][31][32] and two on the metabolome and immune response 33,34 (Table 3).All animal models were on mice, details are provided in Tables S4 and S5.
In two studies comparing GM changes between CMA and sham mice, 27,30 one observed increased Simpson α-diversity in CMA-male-C57BL/6J mice but decreased Simpson and Shannon α-diversity in CMA-female-BALB/cJ mice. 30Regardless of the strain and gender, the β-diversity (Bray-Curtis) was significantly different between the two groups. 30Apart from the gender and strain-specific αdiversity difference, CMA mice showed enrichment in the phyla Bacteroidetes and Patescibacteria (female-C57BL/6J) but reduction in the phyla Verrucomicrobia, Proteobacteria (male-C57BL/6J) and Actinobacteria (female-C57BL/6J). 30Compared to mice colonized with feces from healthy children (healthy-colonized mice), an FT study reported that mice with feces from CMA children (CMAcolonized mice) had higher abundances of the Clostridiales order and the Clostridiaceae, Ruminococcaaceae, and Barnesiellaceae families, along with lower levels of the Lachnospiraceae, Erysipelotrichaceae, and Enterobacteriaceae families. 29At the genus level, the CMAmice exhibited higher Barnesiella and Clostridium_XIVa, 27 and CMA-colonized mice had enhanced Enterococcus, Ruminococcus, Coprobacillus, Blautia, and Parabacteroides. 29In contrast, the Lactobacillus, Parvibacter, 27 Streptococcus, and Salmonella 29 genera, as well as Anaerostipes caccae species 29 decreased in CMA and CMA-colonized mice.Additionally, the Bosea genus was absent in CMA mice. 27ecies and strains of the Lactobacillus and Bifidobacterium genera were used as probiotic in the CMA mouse models. 28,31ne study reported that five out of six probiotic strains reduced the total bacteria. 31Another found significant differences in GM β-diversity (Bray-Curtis, UniFrac) between control and treated groups but only the Lactobacillus rhamnosus species increased GM richness. 28At the family level, it was reported that Prevotellaceae and Marinifilaceae increased, whereas Helicobacteraceae, Lachnospiraceae, Deferribacteraceae, Clostridiaceae, Peptococcaceae, and Burkholderiaceae decreased after taking at least one probiotic. 28Interestingly, the Ruminococcaceae family increased with Lactobacillus rhamnosus treatment but decreased with Bifidobacterium longum subsp.infantis treatment. 28Furthermore, one study found that probiotic treatments with Lactobacillus rhamnosus and Bifidobacterium animalis subspecies lactis increased the Clostridium cluster IVa genus and the Clostridium leptum species. 31Conversely, more than three probiotic strains decreased Lactobacillus, Clostridium cluster I/II, Clostridium cluster XI, Enterococcus and Prevotella genera, as well as the Clostridium coccoides and Clostridium leptum species. 31ditionally, it was reported that prebiotic administration with partially hydrolyzed whey reduced the Lactobacillus genus and increased the Prevotella genus. 32

Metabolome modifications
Two studies examined fecal SCFAs in CMA mice with and without synbiotic intervention. 33,34They reported enhanced acetate, 33 butyrate, 33 and propionate 34 with a synbiotic diet.However, one study only observed reduced kynurenine and N-acetylkunurenine in probiotic-treated mice. 28Additionally, an FT study compared ileal transcription signatures between CMA and healthy-colonized mice. 29They found upregulated metabolism of monocarboxylic acid, arachidonic acid, linoleic acid, and pyruvate in CMA-colonized mice, while increased carbohydrate metabolic process in healthycolonized mice. 29

CMA outcome and immune response
Among all animal studies, only Feehley et al. 29 and Kostadinova et al. 34 correlated the immune response to the GM.Feehley et al. 29 reported that growth factor TGFβ receptor and ROR2 genes in CMA-colonized mice were positively correlated with the Lachnospiraceae family. 29Meanwhile, Kostadinova et al. 34 showed that propionate was positively correlated with FOXP3+ cell frequency in the colon. 342][33][34] Unlike postsensitization, 28 pre-sensitization 31 intake of Lactobacillus salivarius, Lactobacillus rhamnosus, and Bifidobacterium longum subspecies infantis successfully lowered the mast cells degranulation marker mucosal mast cell protease-1 (mMCP-1) 35 and BLG-specific IgE. 31 All strains lowered the IL-4 secretion and the BLG-specific sIgG 1to-sIgG 2a ratio 31 which indicates the overall T h 2-to-T h 1 response. 36e rest of the responses were strain-dependent.Lactobacillus rhamnosus and Bifidobacterium longum subspecies infantis increased T h 1 IFNγ and T reg IL-10 secretion in stimulated splenocytes, whereas Lactobacillus salivarius declined IFNγ secretion. 31Postchallenge administration of those probiotic strains predominantly induced regulatory response. 28All strains significantly increased TGFβ expression, while Lactobacillus rhamnosus and Lactobacillus salivarius interventions also increased FOXP3 and IL-10 expression.The post-sensitization intake resulted in overall cytokine suppression as well.The reduction in granulocyte-macrophage colony-stimulating factor (GM-CSF), IFNγ, IL-2, and IL-4 was common among the strains, while IL12p70, IL-10, IL-5, and IL-17A was strain-dependent. 28stadinova et al. 33,34 reported that synbiotic intake alone did not alleviate the acute allergic skin response but its combination with T cell-epitope-containing BLG peptides (PepMix) did. 33,34tably, the combined diet reestablished the lost T h 1/T h 2 balance as evidenced by the lymphocyte distribution in the small intestine lamina propria 33 as well as the increased transcription factor (Tbet/GATA3) and cytokine (IFNγ/IL-13) gene expression in the Peyer's Patches (PP). 34Right after the intervention the immune response was predominantly regulatory.It was characterized by an increase in the mRNA expression of FOXP3 over the GATA3 and RORγT in the PP, as well as higher FOXP3+ over GATA3+ and T reg over T h cell frequencies in mesenteric lymph node. 34Synbiotic addition had a site-dependent effect on IL-22 mRNA expression and also silenced the whey-stimulated splenocyte secretion of cytokines (IL-10, IL-5, IL-13, IL-17A, and IFNγ) which were induced by the PepMix intake. 33Kleinjans et al. 32 showed that the effect of prebiotics on allergic symptoms varied with the composition and treatment duration.

| D ISCUSS I ON AND CON CLUS I ON
In general, no clear conclusion can be drawn about the GM diversity modification in CMA children, because of limited data on βdiversity 21,30 and discordant results regarding α-diversity in both human 16,19,20 and animal 30 studies.
Taxonomic findings showed that the Bifidobacteriaceae family, including Bifidobacterium spp., were consistently reported lower in CMA children. 14,16,18,19This result aligns with the consensus on the protective function of Bifidobacterium spp. in early life. 37,38other noteworthy observation concerning GM in CMA children is the consistent increase of the Firmicutes phylum, [14][15][16][17][18][19]21 primarily associated with the Clostridia class. Conersely, decreased levels of bacteria of the Lactobacillales order were observed.16,21 The trends of Firmicutes alterations align with the findings of an animal study which reported higher Clostridium cluster XIVa and lower Lactobacillus genus in CMA mice.27 However, CMA and healthycolonized mice were both characterized with bacteria from the Clostridia class, with Anaerostipes caccae, a clostridial species, showing protective effects against CMA. 29Additionally, infants who resolved CMA were reported to have enriched Clostridia class at 3-6 months. 22Discordant results have also been reported regarding the protective or detrimental effect of the Clostridia class in food allergy. 39,40Therefore, despite the conflicting findings of the Clostridia class in this review, we lean towards suggesting that GM with enriched Clostridia class, reduced Lactobacillales order, and reduced Bifidobacterium genus is associated with CMA in early life.
Various intervention approaches, including probiotics, prebiotics, and synbiotics, were applied to restore the balance of GM and the metabolome in CMA children.Elevated Bifidobacterium genus was consistently observed post-treatment with Bifidobacterium strains as probiotics 21,25,26 or after lactose-supplemented EHF treatment. 14However, the impact on the Lactobacillales order in both CMA children and CMA mice was less clear.Increased levels of the Lactobacillaceae family were reported with Bifidobacterium-specific probiotics 26 and EHF in CMA children, 18 while decreased Enterococcus and Streptococcus genera were noted in Bifidobacterium-treated CMA children. 21Additionally, decreased levels of Lactobacillus genus were reported in CMA mice treated with Bifidobacterium and Lactobacillusspecific probiotics. 31,32Similarly, the effect on the Clostridia class varied.Higher levels of its members were reported in CMA children and mice treated with probiotics. 16,21,26,28,31Meanwhile, reduced Clostridia class members were also noted in CMA children treated with lactose-supplemented EHF or probiotics, 14,26 and in CMA mice treated with probiotics. 28,31Therefore, it is clear that the enhancement of Bifidobacterium after Bifidobacterium-specific treatment was commonly reported, however, the treatment effect on other bacteria remain inconclusive.Despite the uncertainty of most GM profile modifications, there are studies that reported improved allergic symptoms or a high-resolution rate in CMA children treated with probiotics or prebiotics. 16,24,26 addition to GM modifications, CMA children were reported to have decreased total SCFAs 14,16 and altered amino acids and nucleotide levels. 14,23These findings are consistent with a recent review on the metabolic changes in children with IgE-mediated food allergies, 41 and these metabolome changes appear to be restored with interventions.Increased SCFAs and balanced amino acids were reported after treatment with LGG or lactose-supplemented EHF. 14,23hanced levels of acetate, 33 butyrate, 33,34 and propionate 34 were also reported in synbiotic-treated CMA mice.
This systematic review provides an overview of the modifica- Thus, future work should examine broader range of metabolites known to be crucial in the crosstalk between the GM and host's immune system 41,42 and use untargeted metabolomics as hypothesisgenerating strategy.Only a single human study reported microbiome and immune response data and their relationship. 26Similarly, only a single animal study correlated transcriptomics and GM data, 29 including genes related to the immune response.Therefore, there is a need for both human and animal studies on the correlation of the GM to the immune response.Future animal studies can build on the general treatment outcome findings in the review, namely overall cytokine silencing, 28,33 restoration of the T h 2/T h 1 balance, 31,33,34 and induction of regulatory response. 28,31,34Moreover, future work can focus on parameters already connected to allergic tolerance acquisition in human, such as induction of T reg response, the production of TGFβ, IgG 4 , and IgA. 43No proteomics studies met our inclusion criteria, but a study on the fecal microbiome and metaproteome relationships in CMA children has been published after our inclusion date. 44Overall, discussions on multi-omics connections are rare in the reviewed studies, and none of the studies reported shotgun meta-genomics, meta-transcriptomics, or meta-proteomics for microbiome function information.Therefore, there is a clear need for more comprehensive multi-omics studies to gain a better mechanistic understanding of CMA in early life.These efforts would eventually lead to the development of better and more effective treatment and preventive strategies.

ACK N OWLED G M ENTS
We thank Ria Derkx (Wageningen University library) for her advice on the search strategy.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare that they have no known conflicts of interest.

PEER R E V I E W
The peer review history for this article is available at https:// www.webof scien ce.com/ api/ gatew ay/ wos/ peer-review/ 10. 1111/ pai.

R E FE R E N C E S
tions of the GM, metabolome, and immune response in IgE-mediated CMA children and CMA animal models.Comparing microbiome data between studies is challenging due to methodological variations, diverse intervention approaches, and the reporting of different taxonomic levels.Consequently, only general conclusions can be drawn based on family or higher taxonomic levels.Meanwhile, insights into metabolomics are restricted by limited scope of studied metabolites.
AUTH O R CO NTR I B UTI O N S Diana M. Hendrickx: Formal analysis; investigation; writing -review and editing; supervision.Mariyana V. Savova: Formal analysis; investigation; writing -original draft.Pingping Zhu: Investigation; writing -original draft; formal analysis.Amy C. Harms: Supervision; writing -review and editing.Renate G. van der Molen: Investigation; writing -review and editing.Clara Belzer: Conceptualization; funding acquisition; investigation; supervision; writing -review and editing.

FU N D I
N G I N FO R M ATI O N This study was part of the EARLYFIT project (Partnership programme NWO Domain AES-Danone Nutricia Research), funded by the Dutch Research Council (NWO) and Danone Nutricia Research (project number: 16490).