Association of human milk oligosaccharides and nutritional status of young infants among Bangladeshi mother–infant dyads

Human milk oligosaccharides (HMOs) support the development of a healthy gut microbiome and the growth of infants. We aimed to determine the association of different HMOs with severe acute malnutrition (SAM) among Bangladeshi young infants. This study was nested within a single-blind, randomized, pilot clinical trial (NCT0366657). A total of 45 breastmilk samples from mothers of < 6 months old infants who had SAM (n = 26) or were non-malnourished (n = 19) and were analyzed for constituent HMOs. Of the infants with SAM, 14 (53.85%) had secretor mothers, and 11 (57.89%) of the non-malnourished infants had secretor mothers. A one-unit increase in the relative abundance of sialylated HMOs was associated with higher odds of SAM in age and sex adjusted model (aOR = 2.00, 90% CI 1.30, 3.06), in age, sex, and secretor status adjusted model (aOR = 1.96, 90% CI 1.29, 2.98), and also in age and sex adjusted model among non-secretor mothers (aOR = 2.86, 90% CI 1.07, 7.62). In adjusted models, there was no evidence of a statistically significant association between SAM and fucosylated or undecorated HMOs. Our study demonstrates that a higher relative abundance of sialylated HMOs in mothers’ breastmilk may have a negative impact on young infants’ nutritional status.


Scientific Reports
| (2022) 12:9456 | https://doi.org/10.1038/s41598-022-13296-w www.nature.com/scientificreports/ Human milk contains a number of bioactive components, among which human milk oligosaccharides (HMOs) have been reported to exhibit pivotal roles in the overall health and nutritional status of infants 11 . Human milk contains human milk oligosaccharide (HMO) more than 20 g/L in colostrum and mature milk contains approximately 12-13 g/L 12 . Also the composition of HMOs vary with maternal nutritional status and dietary intake 13 . The building blocks of these milk oligosaccharides are the five monosaccharides, namely: D-glucose (Glc3), D-galactose (Gal), N-acetylglucosamine (GlcNAc), L-Fucose (Fuc), and sialic acid (Sia; N-acetyl neuraminic acid [Neu5Ac]) 11 . Usually, HMOs range from 3 to 8 sugars in size, mostly containing fucose as a constituent although highly complex oligosaccharides containing up to 32 sugars have been reported 14 .
HMOs are not digestible by infants and arrive intact to the large intestine, where they exert prebiotic roles by supporting the development of selected beneficial groups of the gut microbiota, including Bifidobacterium sp and some Lactobacillus sp. by providing metabolic products for their existence, growth, and ultimate colonization in the gut of infant 15,16 . In addition to helping in healthy gut microbiome, HMOs deliver a number of assistances to infants, including brain development 17 , acting as decoys for harmful organism 18 , and averting disease and infection 19 . Certain HMOs have also been found to reduce frequencies of moderate to severe diarrhea among infants, in particular diarrhea caused by Campylobacter and calicivirus 20 .
Human milk oligosaccharides differ between secretor and non-secretor mothers 21 , whereby fucosylation in the human milk oligosaccharides are a result of gene products that regulate Lewis and secretor blood group types 21 . Concurrent shifts in gut microbiota depending on secretor status have been previously reported [22][23][24] . Subsequently, a Malawian study reported that mothers with undernourished infants had an overall decreased content of milk oligosaccharides 25 . It was also observed that particular human milk oligosaccharides support age-appropriate growth of infants 26 .
A recent study reported that milk oligosaccharides influences the nutritional status of the offspring 27 . However, there has been no report on the variation of HMO content and its effect on the nutritional outcome of young infants, particularly among those of under six months of age. In light of this knowledge gap, we conducted an observational study at the Dhaka Hospital of icddr,b, where we aimed to determine possible relationships between HMO content of breast milk with the nutritional status, anthropometric outcomes, and abundance of Bifidobacterium infantis in the gut of these young infants.

Materials and methods
Study site and participants. This study was conducted at the Dhaka Hospital of icddr,b. We recruited 49 mothers of infants who were already enrolled in a randomized controlled pilot trial (RCT) entitled "Pilot of a prebiotic and probiotic trial in young infants with severe acute malnutrition (NCT0366657)". These infants were between 2 and 6 months of age, of either sex, admitted to the Dhaka Hospital of icddr,b due to gastrointestinal or respiratory infection and were categorized into two groups based on their nutritional status: 1) SAM: infants with weight for length Z score of less than − 3 (WLZ < − 3) or with the presence of bilateral pedal edema, irrespective of anthropometric measurements and non-malnourished: infants with weight for length Z score of greater than − 1 (WLZ > − 1). The final analysis included 45 infants; 4 samples (3 SAM and 1 non-malnourished infant) were excluded because the secretor status of the women could not be determined.
Collection of breast milk samples. Breast milk samples from mothers were collected under aseptic conditions. Mothers were asked to stop feeding the infant 30 min before sample collection. Iodine-soaked gauze was used to clean the breast. Samples were collected in a sterile bottle using a breast pump. The samples were aliquoted in pre-labeled cryovials and flash-frozen in liquid nitrogen immediately after collection. The samples were transferred to − 80 °C freezers within 24 h of collection and were stored until further analyses.
HMO extraction and quantitation. The technique of extraction of milk oligosaccharide and quantification has been described in detail in the cited literature 28,29 . In brief, 25 µL aliquots of each breastmilk sample were diluted and defatted via centrifugation. Proteins were precipitated with ethanol and the resulting glycans were reduced with 1M NaBH 4 for 90 min in a dry incubator set at 65 °C and then purified on solid-phase extraction graphitized carbon cartridges (GCC). The resulting eluents were combined, followed by reconstitution by solvent evaporation and dilution for subsequent analyses. HPLC QqTOF Mass Spectrometry was used for quantitation of the individual HMO constituents, using previously optimized protocols 29 . The specific structures were identified and assigned by matching retention time and exact mass which was matched with the data available in previously developed HMO libraries 30,31 and the glycans were categorized in the manner as followed in previous publications 28-32 . Evaluation of secretor status from breast milk. Secretor status was determined by phenotyping method, as described elsewhere following earlier published methods 33 . Briefly, oligosaccharides with known α(1,2)-Fuc linkages were identified by matching exact mass and retention times with a previously developed inhouse library 30,31 . The signal responses of the most abundant α(1,2)-Fuc containing structures, namely 2'-fucosyllactose (2'FL), lactodifucotetraose (LDFT), difucosyllacto-N-hexaose a(DFLNHa) and trifucosyllacto-N-hexaose (TFLNH), were summed and normalized to the total HMO abundance present in a given sample. Secretor status was assigned based on a previously established and validated threshold 33 . The sum of the relative abundances was obtained and when it exceeded this threshold the mother was deemed to be a secretor; conversely, if it fell below the threshold they were deemed to be non-secretors.
Identifying human milk oligosaccharides. Fucosylated and sialylated HMOs. To identify the HMOs, we followed reports from previously published literature 34  www.nature.com/scientificreports/ N-acetylhexoseamine (HexNAc), fucose, N-acetylneuraminic acid (sialic acid). The compound name of HMOs was expressed with a four-digit number. For example, the compound identifier 4110 indicated that the compound has 4 hexoses, 1 HexNAc, 1 fucose, and 0 sialic acid molecule. Thus, compounds with a nonzero number in the last position are sialylated HMO; those with a nonzero number in the third position are fucosylated HMO.
Undecorated HMOs. Other than fucosylated or sialylated HMOs are called undecorated HMOs.

Evaluation of the B. infantis levels in the fecal samples of the infants at enrolment. Stool
samples were collected at baseline, prior to the start of supplementation. Stool samples collected were aliquoted in pre-labeled cryovials and flash-frozen in liquid nitrogen within 20 min after defecation, as practiced in other studies 35 . Aliquots were preserved in − 80 °C freezers until analysis. The laboratory assays were carried out at the Laboratory of Evolve bioSystems, Inc California. The levels of B.infantis, the predominant member of the gut flora among young infants was measured from the collected stool samples by qPCR, using protocols described elsewhere 36 .

Statistical analysis.
We presented the characteristics of mothers and infants using percentage and median (interquartile range (IQR)), as appropriate. To assess the association of the relative abundance of different HMOs with severe acute malnutrition, simple and multivariable binary logistic regression models were used separately for fucosylated, sialylated, and undecorated HMOs. As the sample size is relatively small, we used firth logistic regression instead of conventional logistic regression. Firth logistic regression provides more reliable estimates in instances of small sample size, complete separation, and rare events using penalized maximum likelihood estimation 37 . We expressed the strength of association as odds ratio (OR) and adjusted odds ratio (aOR) with a 90% confidence interval (90% CI) for each unit increase in the relative abundance of a certain HMO.
Since this is an exploratory study, to better understand the relationship between the relative abundance of the HMOs and SAM, we fitted infant age-and infant sex-adjusted models, and infant age-, infant sex-, and mother's secretor status-adjusted models. We also fitted infant age-and infant sex-adjusted models separately for secretor and non-secretor mothers. To examine the discriminative performance of the logit models for the detection of SAM, the area under the receiver operating characteristic (AUROC) curves was estimated 38 .
All statistical tests were two-sided. Since the study is pilot in nature, the statistical significance was evaluated at p < 0.10, and 90% CIs was reported 39 . Data analysis was done in Stata v15.1 (Stata Corp, College Station, TX, USA). Z scores for nutritional status were calculated using WHO Anthro software.
Ethical approval. The randomized controlled trial was registered at ClinicalTrials.gov (NCT03666572) on 12/09/2018 (https:// clini caltr ials. gov/ ct2/ show/ NCT03 666572) and the research protocol was approved by the Institutional Review Board (IRB) of International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b) on 16 February, 2018 (approval number: PR-17112). All clinical aspects of the study were supervised by the investigators and study physicians. All activities were conducted in accordance with the Helsinki Declaration of 1975, revised in 1983. Names of the participants were de-identified prior to analysis.
Patient consent for publication. Details that disclose the identity of the participants or their information were kept confidential. All procedures were approved by the appropriate ethics committee and have therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Informed consent was taken prior to sample collection and intervention.
In comparison to the mothers of non-malnourished infants, the mothers of the SAM infants were younger in age (23.42 years vs. 25.89 years), had lower educational qualification (38.46% vs. 21.05%, education level less than 5 years), had babies delivered with lower birth weight (2.76 kg vs. 3.04 kg) and lower proportion of term infant (65.38% vs. 89.74%). Higher proportion of cesarean section (LUCS) was predominant among the SAM infants in comparison to the non-malnourished infants (50% vs. 21.05%). Moreover, SAM infants presented with a short duration of diarrheal illness, when compared to the non-malnourished infants (4.46 days vs. 7.26 days). Antibiotic use before hospitalization was significantly higher in non-malnourished infants (34.78% vs. 94.44%, p < 0.01). Fecal B. infantis levels baseline were lower in non-malnourished infants compared to the SAM infants (6.00 vs. 4.88 Log10 CFU/µg of DNA), but this difference was not statistically significant (Table 1). Figure 1 shows the distribution of individual oligosaccharides in breastmilk of mothers of SAM and nonmalnourished infants. In the analysis, we identified 73 HMOs out of which 26 HMOs had a chemical name and others only had a compound identifier.
In Table 2 shows the association of malnutrition status of the young infants with different types of HMOs. Sialylated HMOs were associated with higher odds of severe acute malnutrition status in age and sex adjusted model (AOR = 2.00, 90% CI 1. 30 www.nature.com/scientificreports/ severe acute malnutrition but there was no significant association between these. Undecorated HMOs showed association with higher odds severe acute malnutrition among the young infants in all the models but associations were not statistically significant. In Supplementary Table, we included the conventional logistic regression. Different adjusted models showed modest to strong ability to distinguish association with different HMOS and severe acute malnutrition (AUROC range: 0.74-0.86) (Supplementary Figs. 1,2,3).

Discussion
The study was conducted among 45 infant-mother dyads. The data presented here were based on the HMO analysis of the breast milk of mothers who had either non-malnourished or SAM infants. In the study, we observed that the association of sialylated HMOs with severe acute malnutrition among young infants. Conversely, fucosylated HMOs and undecorated HMOs had no significant association with severely malnourished infants. To our knowledge, ours is the first study to assess the relation of different HMOs with malnutrition status among Bangladeshi infant-mother dyads without any supplementation of HMOs.
Breast milk contains complex oligosaccharides that escape digestion by the intestinal enzymes and transit through the gastrointestinal tract of the infant 40 . The variety of HMOs appears to be endless; to date over 200 different structures of HMOs have been identified in breast milk. HMO profile is affected by host secretor status, encoded by fucosyltransferase 2 (FUT2) gene, based on the secretor/Lewis blood group 21,41 . Secretor status phenotyping depends upon the expression of the histo blood group antigens (HBGAs): fucosyltransferase FUT2 (secretor gene), and FUT3 (Lewis gene) in secretions such as saliva and breast milk 42 . These antigens may act as innate host factors that differentially influence the susceptibility of individuals and their offspring 43 . The analytical procedure that was used in this study for the secretor status phenotyping of the mothers involves the measurement of a certain level of HBGAs in the breast milk samples, using a set cut-off value 33 . FUT2 and FUT3 genes that are rendered non-functional due to distinct polymorphisms are thus unable to express the corresponding HBGAs in the secretions 44 . Henceforth, the secretor status of such individuals with non-functional FUT2 and FUT3 genes cannot be determined by analytical methods, as used in our study. This is a probable explanation of their "Undetermined" secretor status.
The oligosaccharide abundance in breast milk is also affected by maternal genetics, particularly by the FUT2 gene, which encodes an enzyme accountable for the addition of a fucose residue at the α1-2 position on a backbone of abundant glycans containing galactose 45 . The absence or trace amount of α1-2 fucosylated oligosaccharides is responsible for the decreased amount of HMOs in non-secretor mothers' breast milk 33 .
In our study, we have observed an association of higher odds of severe acute malnutrition among the young infants with sialylated HMOs and this association was statistically significant. On the contrary, fucosylated HMOs were less likely to be associated with severely malnourished infants. Recent studies demonstrated that www.nature.com/scientificreports/ human milk oligosaccharides (HMOs) are linked to growth in early infancy. In a study conducted among 37 mother-infant dyads, it was observed that an increase in lacto-N-fucopentaose (LNFP) I was linked with lower infant weight at 1 and at 6 months and with lower lean and fat mass at 6 months. Further, At 6 months, LNFP II and disialyl-lacto-N-tetraose (DSLNT) were related to higher fat mass 46 . The amount of fucosylation in breast milk changes over the course of lactation 47 , which may affect the protection deliberated to an infant. HMOs containing α1,2-fucosyl linkages have been revealed to help the growth of members of the Bifidobacterium genus due to prebiotic action 48 . Different HMOs have been shown to exert an impact on the weight-for-age z score (WAZ) of infants, particularly 3'-SL, as observed in the Gambian study 49 . In this study, the relative sialylation of HMOs was not a predictor for WAZ of the infants 49 . However, in our study, we observed that severely malnourished infants consumed breast milk which had significantly higher sialylated HMOs than non-malnourished infants. Sialylated HMOs was also found to have impact on linear growth in animal model study 50 . The report from a Malawian cohort showed that  Table 2. Association of sialylated, fucosylated, and undecorated HMOs with severe acute malnutrition among young infants using firth logistic regression. For model 1: adjusted odds ratio (aOR) (90% CI) was adjusted for age and sex. For model 2: adjusted odds ratio (aOR) (90% CI) was adjusted for age and sex and secretor status. For model 3: it was for secretor mothers only and adjusted odds ratio (aOR) (90% CI) was adjusted for age and sex. For model 4: it was for non-secretor mothers only and adjusted odds ratio (aOR) (90% CI) was adjusted for age and sex. Abbreviations: aOR, adjusted odds ratio; CI, confidence interval, HMO, human milk oligosaccharide. www.nature.com/scientificreports/ among only the secretor mothers, the total abundance of HMOs was positively associated with increments in length-for-age z-score (LAZ) from 6 to 12 months of age 51 . Analysis of HMOs of two Malawian birth cohort also reported that severely stunted infants had significantly low sialylated HMO 25 . Observation of this study was also used in animal model study which indicated relationship between sialylated HMO and growth promotion 25 . On enrolment our study infants were with severe acute malnutrition, so we could not follow them to get association of linear growth with sialylated HMOS. In our model, there was significant association with sialylated HMOs with severe acute malnutrition both in age, sex-adjusted model and age, sex, secretor status adjusted model. Comparatively lower sample size in this study may have concealed the significant effect of fucosylated HMOs on the nutritional status of the infants enrolled in our study. Recent review article reported that the results of different studies on HMO were also contradictory due to mixed interpretations 52 . With our exploratory study we have found association of sialylated HMOs with severe acute malnutrition but we could not find causal association from this cross-sectional design. The infant's intestinal microbiome predominantly consists of two subspecies of Bifidobacterium longum: subsp. infantis (B. infantis) and subsp. longum (B. longum). Competitive growth of B. infantis in the neonatal intestine has been connected to the use of human milk oligosaccharides (HMO) 53 . Previous studies have shown that different fucosylated HMOs have an impact on the incidence of diarrhea 49 . Our study infants exhibited diarrhea on admission, so we were unable to establish a specific role of any of the constituent HMOs on diarrheal incidences among these young infants with different nutritional statuses. We have observed that malnourished infants presented with shorter duration of diarrhea compared to non-malnourished infants. Malnourished infants present with increased disease severity due to their impaired immune system which could have an impact on early presentation of such infants 54 .

Sialylated HMOs
Fucosylated HMOs and N-glycans on milk proteins are beneficial for the growth of healthy gut microbiota, acting as prebiotics 25,48 . Changes in gut microbiota depending on secretor status have been reported earlier [22][23][24] . The impact of maternal secretor status in microbiota is well documented and affects different bacterial groups during lactation 55 . Different studies reported that HMOs in breast milk are key factors which promote the development of members of the Bifidobacterium genus in the gut 56,57 . Bifidobacterium longum subsp. infantis (B. infantis) is unique among other gut microbes as this members of this particular subspecies of B.longum areequipped with the appropriate machinery to consume the full range of human milk oligosaccharides (HMOs) 58 . A large proportion of breast milk comprises fucosylated HMOs which can also be utilized by select B. breve strains and B. bifidum through extracellular fucosidase activity 59,60 . The gut microbiota is often exposed to a variety of antimicrobial agents, like antibiotics, which are used to combat bacterial infection in humans 61 . This antibiotic therapy may cause depletion of the beneficial members of the gut microbiota, a phenomenon known as "dysbiosis' 62 . Results from our study suggest that different types of HMOs consumed by infants have an impact on the nutritional status of the infants. However, the use of antibiotic may have masked the effect of HMOs on gut microbiota especially of the Bifidobacterium species which showed that even with exclusive breastfeeding, the levels of Bifidobacterium infantis was comparatively lower in non-malnourished infants. This association explains context-specific correlation, which also has causational support. Different studies also reported that infants without exposure to antibiotics presented with a higher percentage of Bifidobacteriaceae 63,64 . A number of metagenomics studies have illustrated associations between the altered gut microbiota and child malnutrition. Early depletion of Bifidobacterium longum appears to be the first step in gut microbiota alteration in severe acute malnutrition 65,66 since any change in the relative abundance of Bifidobacterium in early infancy marks a change in the infant gut microbiome, potentially indicating dysbiosis.
The smaller sample size of this study makes it underpowered for certain study findings. Nevertheless, differences were detected in the relative abundance of particular HMO constituents in the breast milk of mothers of infants of different nutritional statuses, and as a result, the concentration of each constituent HMO was not measured. Moreover, the SAM infants were only partially breastfed and this variable might have had influenced the overall study findings and we could not assess the total amount of HMOs. A prospective large data set is required to have clear understanding on relationship of HMOs with ponderal as well as linear growth of infants.
This study provides primary observational data that the nutritional status of infants is affected by breast milk HMO composition of mothers and that additional supplementation of certain deficient HMO along with B. longum subsp infantis could be an intervention approach for the management of malnutrition in early infancy.

Data availability
This data set contains some personal information of the study patients (such as name, admission date, month, area of residence). Our IRB has required that the personal information of the participants is not disclosed. Thus, the policy of our centre (icddr,b) is that we should not make the availability of whole data set in the manuscript, the supplemental files, or a public repository. However, data related to this manuscript are available upon request and for researchers who meet the criteria for access to confidential data may contact with Armana Ahmed (armana@icddrb.org) to the Research Administration of icddr,b (http:// www. icddrb. org/).