Impact of mother's own milk vs. donor human milk on gut microbiota colonization in preterm infants: a systematic review

INTRODUCTION

Infants born very preterm (i.e., < 32 weeks of gestation) constitute a vulnerable population, characterized by a significant risk of mortality and short- and long-term morbidities^[1]. Enteral feeding strategies are suggested to reduce those risks^[2]. Enteral feeding with mother’s own milk (MOM) improves short- and long-term health outcomes compared to preterm formula (PF). This includes lower rates of necrotizing enterocolitis (NEC) and sepsis, and better neurodevelopmental outcomes^[3-7].

In the very preterm population, achieving an exclusive MOM diet is not always feasible^[8]. Historically, when sufficient MOM was unavailable, very preterm infants received PF. In recent years, with the establishment of more donor human milk (DHM) banks, DHM is recommended as the second-best option by organizations such as the World Health Organization and the European Society for Paediatric Gastroenterology, Hepatology and Nutrition^[5,9-13]. Albeit probably not to the same extent as MOM, DHM also leads to fewer adverse neonatal outcomes, such as NEC, as compared to PF^[7,14]. Research comparing neonatal health outcomes between predominantly MOM or DHM diets is scarce and the available results are inconsistent^[7,15-17]. Some studies indicate that the incidence of NEC and other morbidities may be dependent on the quantity of MOM provided to the infants, indicating that higher MOM exposition confers greater benefits to the very preterm infant compared to DHM^[18].

A potential mechanism through which human milk exerts its beneficial effects is via the composition and function of the gut microbiome^[19,20]. Bioactive factors in MOM, such as lactoferrin, human milk oligosaccharides, secretory immunoglobulins, and the milk microbiota, influence the development of the preterm gut microbiota^[21]. However, it is essential to recognize that some of these bioactive factors are affected by processing methods, including Holder pasteurization and multiple freeze-thaw cycles, to provide preterm recipients with a safe product of DHM. In addition, biological factors such as the lactation stage and characteristics of the donating mother may also influence the quality of the donated milk^[22]. Subsequently, differences in composition between MOM and DHM may result in a difference in gut microbial colonization of preterm infants.

Currently, only a limited number of studies have compared the effect of DHM vs. MOM on the gut microbiota composition in preterm infants, hampering drawing a firm conclusion. This is important however, as clear associations between early-life gut microbiota composition and short- and long-term adverse outcomes, including NEC, sepsis, or neurodevelopmental impairment, have previously been described^[23-25].

The aim of this systematic review is to assess and provide a qualitative synthesis of available literature that reports on the effects of DHM vs. MOM on the gut microbiota composition in preterm infants, with specific attention to microbial diversity and taxonomic composition.

METHODS

RESULTS

Included studies

The search process is shown in Figure 1. A total of 1,028 articles were identified through the database search. After removing duplications, 717 articles were initially screened by title and abstract, of which 671 articles could be excluded. In total, 46 articles underwent full-text review, of which 38 articles were excluded. Thus, a total of 8 original studies could be included in our qualitative synthesis^[19,31-37]. Backward snowballing did not yield any additional suitable articles.

Figure 1. PRISMA 2020 flow diagram of the search and selection process. PRISMA: Preferred reporting items for systematic review and meta-analysis protocols.

Description of the studies

Characteristics of the eight included studies are summarized in Table 1 and depicted in Figure 2. Seven out of the eight studies were prospective observational cohorts. The remaining study by Kumbhare et al. was a randomized, controlled, open-label trial comparing bovine milk-based vs. human milk-based fortifier added to MOM or, if insufficiently available, to DHM^[34]. However, the authors also described fecal microbiota results for two post-hoc cohorts where infants were stratified whether they had predominantly received MOM or DHM, irrespective of the fortifier allocation.

Figure 2. Overview of studies included in this review. On the X-axis, the sampling period is shown, expressed in weeks of PMA. (A) Subgroup A consists of studies where infants received 100% human milk (MOM/DHM); (B) Subgroup B consists of studies where infants received a combination of human milk (MOM/DHM) and PF. Note: To express the maturity stage of infants at the time of sample collection, we calculated PMA during sampling collection time points. For the start of the PMA range, we added the earliest time point of sample collection to the lowest mean or median gestational age from both groups (MOM and DHM). For the end of the PMA range, we added the latest time point of sample collection to the highest mean or median gestational age from both groups. For studies that stated sampling at the day of achieving full enteral feeding, but not explicitly mentioning at which day of life this was, we arbitrarily chose ten days of age as the mean age of achieving full enteral feeding in order to calculate the PMA during sample collection^[38]. N means number of stool samples collected at one or multiple time points. ^*N for all time points combined. Created with Biorender.com. PMA: Postmenstrual age; MOM: mother’s own milk; DHM: donor human milk; PF: preterm formula; GA: gestational age; DoL: day of life; WoL: week of life; FEFA: full enteral feeding achieved.

Table 1

Characteristics of studies included in this review

Author, year, country [ref]	Design	Primary aim	Study population	Number of infants* (MOM/DHM)	Baseline characteristics MOM group	Baseline characteristics DHM group	Subgroup and definition of feeding groups
Subgroup A
Arboleya et al., 2020, Spain[31]	Mono-center cohort study	Effect of DHM on Bifidobacteria	GA < 34 weeks	42 (in total)	GA: 30.9 (2.44) BW: 1,351 (353) MoD: 46% vaginal	GA: 30.3 (2.7) BW: 1,246 (302) MoD: 42% vaginal	A MOM: 100% DHM: 100%
Ford et al., 2019, USA[33]	Mono-center cohort study	Effect of human milk on microbiota development	BW < 1,500 grams	90 (58/32)	GA: 28.7 (2.0) BW: 1,107 (242) MoD: 18% vaginal	GA: 28.4 (2.5) BW: 1,066 (294) MoD: 12% vaginal	A MOM: > 50% DHM: > 50%
Kumbhare et al., 2022, Canada[34]	Stratified cohorts from a mono-center randomized controlled open-label trial	Gut microbiome, oxidative stress and inflammation in milk fortifiers	GA 26-30 weeks	30 (20/10)	GA: 27.9 (1.0) BW: 1,013 (159) MoD: 20% vaginal	GA: 28.4 (0.86) BW: 1,053 (159) MoD: 20% vaginal	A MOM: > 50% DHM: > 50%
Piñeiro-Ramos et al., 2020, Spain[32]	Mono-center cohort study	Composition of microbiota and urinary metabolites	GA < 32 weeks and/or BW < 1,500 grams	36 (18/18)	GA: 28 (27-29) BW: 1,300 (300) MoD: 45% vaginal	GA: 29 (28-30) BW: 1,400 (200) MoD: 45% vaginal	A MOM: 80% DHM: 80%
Subgroup B
Cong et al., 2017, USA[36]	Bi-center cohort study	Effect of feeding type on gut microbial patterns	GA 28 to 32 + 6 weeks	18 (15/3)	GA: 31.1 (1.8) BW: 1,444.2 (442.7) MoD: 40% vaginal**		B MOM: > 70% DHM: > 70%
Gregory et al., 2016, USA[19]	Mono-center cohort study	The influence of nutrition on the microbiome	GA < 32 weeks	20 (10/10)	GA: 28.4 (1.5) BW: 1,044 (258) MoD: 10% vaginal	GA: 28.4 (2.1) BW: 1,070 (422) MoD: 10% vaginal	B MOM: > 50% DHM: > 50%
Morais et al., 2021, Portugal[37]	Mono-center cohort study	Impact of different feeding types on fecal microbiota and ALP activity	GA < 32 weeks	95 (75/20)	GA: 28.3 (2.1) BW: 1,123 (345) MoD: 45% vaginal	GA: 28.6 (1.8) BW: 1,173 (284) MoD: 35% vaginal	B MOM: > 50% DHM: > 50%
Parra-Llorca et al., 2018, Spain[35]	Mono-center cohort study	Effect of nutrition on gut microbiota composition	GA < 32 weeks and/or BW < 1,500 grams	62 (34/28)	GA: 28.9 (1.9) BW: 1,228 (301) MoD: 59% vaginal	GA: 29.8 (2.4) BW: 1,304 (262) MoD: 40% vaginal	B MOM: > 80% DHM: > 80%

^*In groups of interest; ^**for all included infants, not stated separately for DHM or MOM group. MOM: Mother’s own milk; DHM: donor human milk; GA: gestational age (in weeks), expressed as mean (SD) or median (IQR); BW: birthweight (in gram), expressed as mean (SD) or median (IQR); MoD: mode of delivery, expressed as percentages; ALP: alkaline phosphates.

The criteria that were used to stratify infants in one of the two feeding groups varied among the studies. Studies used varying thresholds from a minimum of 50%^[33,34,37] to 70%^[36], 80%^[32,35], or by exclusively receiving 100% MOM or 100% DHM^[31] of the total milk volume intake. In the study by Gregory et al., infants were also stratified based on whether they had received exclusively MOM, while those in the DHM group switched from exclusively DHM (duration of 6 to 25 days) to PF once achieving full enteral feeding^[19]. Eventually, four studies were grouped into Subgroup A^[31-34], and Subgroup B consisted of the other four^[19,35-37]. Six studies excluded children with NEC and/or sepsis^{[19,31,34-37]}. Ford et al. reported an overall NEC rate of 2.4% among the study infants^[33]. Piñeiro-Ramos et al. reported only one NEC case (5%) in the DHM group^[32].

A total of 389 infants were included, with analyses conducted on 1,832 unique fecal samples collected from the second day of life up to the 90th day of life. The mean/median gestational age in the included cohorts ranged from 27.9 to 31.1 weeks, and the mean/median birth weights ranged from 1,013 to 1,444 grams. Cesarean section was the predominant mode of delivery in all but one study^[35]. Overall, there were no major clinically meaningful differences in reported baseline characteristics between infants in the MOM and DHM cohorts in each of the included studies.

Two studies used a cross-sectional approach, with one sample per individual^[32,35] collected at the time of full enteral feeding achieved, while the remaining six studies collected multiple samples at different time points during admission.

The applied gut microbiota analysis techniques in each study are listed in Supplementary Table 1. Next-generation sequencing was employed in all eight studies to measure specific 16S rRNA gene regions. In the study by Arboleya et al., the profiling is specific to the species composition of the genus Bifidobacterium only, whereas others studied overall gut microbiota composition^[31].

Taxonomy

The primary taxonomy differences between groups from birth till PMA 36 weeks are summarized in Table 3 and graphically depicted in Figure 3. Below are recurring differences mentioned across multiple studies, categorized by phylum, to illustrate commonalities in this heterogeneous section. The MOM group serves as a reference group in text, tables, and figures. No consistent differences at the taxonomic level were apparent between subgroup A and subgroup B. Therefore, no distinction between the subgroups was made in the text.

Figure 3. Graphical depiction of taxonomic differences at various PMA intervals. (A) subgroup A consists of studies in which infants received 100% human milk (MOM/DHM). Arboleya et al. were excluded from this figure because they only reported differences within the Bifidobacterium genus^[31]; (B) subgroup B consists of studies in which infants received a combination of human milk (MOM/DHM) and PF. Created with Biorender.com. This figure is a visual complement to Table 3, where taxonomic differences resulting from DHM consumption in comparison to MOM at each time point are described. PMA: Postmenstrual age; MOM: mother’s own milk; DHM: donor human milk; PF: preterm formula.

Firmicutes

Seven studies reported differences in the Firmicutes phylum at different taxonomic levels^[19,32-37].

Ford et al., Morais et al., and Parra-Llorca et al. reported a higher abundance of Firmicutes at the phylum level^[33,35,37] in DHM-fed infants. Within the class Bacilli, two studies reported differences at the order level^[19,36], two studies reported differences at the family level^[32,35], and four studies reported differences at the genus level^[33-35,37]. There are conflicting results at the order level, with Cong et al. stating a lower abundance of Lactobacillales and Bacillales at all three time points^[36], whereas Gregory et al. stated the abundance of the two orders to be higher in DHM-fed infants^[19]. Within the reported outcomes in the class Bacilli, there are commonalities in the reported differences within the Staphylococcaceae family. Piñeiro-Ramos et al. reported higher abundances of the Staphylococcaceae family^[32], and in the same direction, Parra-Llorca et al., Morais et al., and Ford et al. showed an increased abundance of the Staphylococcus genus^[33,35,37].

Four studies stated differences at different taxonomic levels within the class Clostridia. Parra-Llorca et al. reported a higher abundance of the Clostridiaceae family^[35] in DHM-fed infants. Khumbare et al. and Parra-Llorca et al. stated a higher abundance of the genus Clostridium, whereas only Morais et al. stated a lower abundance in the DHM cohorts^[34-37].

Actinobacteria

Seven studies reported consistent differences in the Actinobacteria phylum at different taxonomic levels^{[19,31,33-37]}. Ford et al., Morais et al., and Parra-Llorca et al. stated lower abundances of Actinobacteria at the phylum level in the DHM groups^[33,35,37]. Cong et al. stated Actinomycetales at the order level have lower abundances^[36]. There are quite some commonalities in the Bifidobacteriales order in that Cong et al. stated lower abundance of the Bifidobacteriales order, while Ford et al., Morais et al., and Parra-Llorca et al. stated lower abundance of the Bifidobacterium genus and Khumbare et al. stated lower prevalence of the Bifidobacterium genus in the samples from mostly DHM-fed infants^[33-35,37]. Arboleya et al. focused solely on the Bifidobacterial population at the species level^[31]. Although in both feeding groups, B. breve and B. longum ssp. longum were the species with the highest abundances during the study period, there were some differences in the abundances of other species. Notably, B. animalis ssp lactis, B. bifidum, and B. dentium were higher in the DHM group in terms of relative abundance. Only Gregory et al. stated no differences in the succession of Bifidobacteriales^[19].

Proteobacteria

Ford et al. reported an increasing relative abundance of Proteobacteria in all study subjects over time^[33]. Five studies reported differences in the Proteobacteria phylum at different taxonomic levels, with conflicting results^[32,34-37]. Morais et al. stated lower abundances of the Proteobacteria phylum^[37]. In the Enterobacterales order, Cong et al. stated higher abundances of Enterobacterales^[36]. Khumbare et al. and Parra-Llorca et al. stated lower abundances of an unclassified genus of the Enterobacteriaceae family^[34,35]. In addition, Morais et al. stated a lower abundance of the E. coli species and Serratia genus, whereas Parra-Llorca et al. stated a higher abundance of Serratia^[35,37].

In the Pasteurellales order, a similar pattern of inconsistency is visible. Cong et al. stated a lower abundance of the Pasteurellales order^[36]. Piñeiro-Ramos et al. stated a higher abundance of the Pasteurellaceae family^[32]. At the genus level, Parra-Llorca et al. stated higher abundances of the Aggregatibacter genus and lower abundances of the Haemophilus genus^[35].

Bacteroidetes

Three studies reported consistent differences in the Bacteroidetes phylum at different taxonomic levels^[33,35,37]. At the phylum level, Morais et al. stated a lower relative abundance of Bacteroidetes^[37]. At the genus level, both Ford et al. and Parra-Llorca et al. stated a lower abundance of Bacteroides^[33,35].

DISCUSSION

In this systematic review, we observed differences in the fecal microbiome of preterm infants fed predominantly with DHM, compared to those predominantly fed with MOM. These differences manifest in both diversity and composition across various taxonomic levels and are reported across several time points, from as early as the second day of life to approximately 36 weeks PMA. In the following sections, we will discuss these findings in more detail.

Our review provides evidence indicating that preterm infants predominantly fed with DHM exhibit reduced alpha diversity in their gut microbiome compared to those fed predominantly with MOM. However, this evidence is solely documented within subgroup B, which encompasses studies incorporating formula feeding alongside MOM or DHM. Therefore, the inferred effect could also be related to the quantity of formula milk received, rather than the difference between MOM and DHM feeding. These findings align with existing evidence on the microbiome composition of infants fed PF vs. MOM, reporting higher alpha diversity in the latter cohorts^[39]. The only study that described an increase in alpha diversity in the DHM group focused only on Bifidobacterium species, rather than assessing overall diversity^[31].

The included studies employed different metrics for alpha diversity evaluation, including the Shannon- or Chao1-indices and OTU richness. Several studies reported multiple measures for alpha diversity with conflicting outcomes. For example, Morais et al. showed a difference in the Chao1 index, but not in Shannon index findings, underscoring that the results on alpha diversity are dependent not only on the analyzed cohort, but also on the selected metric^[37]. This discrepancy limits the comparability of published data.

Our review yields mostly consistent evidence for differences in beta diversity, with distinct clustering of the gut microbiome based on feeding type from the tenth day of life onwards. These differences are apparent in both subgroups A and B. The two studies assessing the effects of multiple demographic factors on the Bray-Curtis dissimilarity index, showed that feeding type contributed more to the observed difference than other factors tested, such as gestational age, birth weight, postnatal age, or antibiotic administration^[19,36]. Contradicting results were present in the study by Morais et al., which found no differences in beta diversity between the two feeding groups. This apparent incongruity might be due to the relatively small sample size (61 samples, of which 14 samples were from the DHM group) included for 16S rRNA sequencing compared to the other included studies^[37].

At the taxonomic level, there was large heterogeneity in reporting methods. While some studies reported outcomes at the phylum level, others reported at the order, family, genus and/or species levels. Despite this methodological heterogeneity, consistent evidence emerges regarding several compositional differences. DHM feeding appeared to be associated with an increased abundance of the Staphylococcaceae family, possibly attributable to elevated levels of the Staphylococcus genus, as well as the Clostridiaceae family, likely attributable to the increased abundance of the Clostridium genus.

Conversely, a decreased abundance was observed in the Bifidobacterium genus and the Bacteroidetes phylum in DHM-fed preterm infants compared to those fed predominantly MOM. In addition to a diminished overall abundance of Bifidobacterium, Arboleya et al. showed several differences in the relative abundances of various Bifidobacterial species present between both feeding groups, with differences starting to occur as early as the second day of life^[31].

The observed differences in the gut microbiota between both subgroups might be associated with different health outcomes through various (patho)physiological processes. While reduced alpha diversity levels have been linked to adverse health outcomes in adults, it is unknown if it pertains to preterm infants as well^[40]. At the compositional level, the observed summarized differences in this systematic review have been linked to adverse health outcomes. An increased abundance of the Staphylococcaceae family or Staphylococcus genus has been linked to late-onset sepsis in preterm infants^[41,42]. Similarly, increased abundances of different genera and species within the Clostridia order have been linked to NEC in preterm infants^[43-46]. Moreover, reduced Bacteroidetes and Bifidobacterium abundance have also been associated with NEC^[47-51]. Notably, these compositional differences are not only confined to short-term neonatal outcomes; the abundance of Bifidobacterium, Staphylococcus, and Bacteroidetes during the neonatal period has been linked to adverse neurodevelopmental outcomes at two years of corrected age^[24,52-54]. In addition, term-born infants generally have higher abundances of Bacteroidetes and Bifidobacterium and lower abundances of Staphylococcus compared to preterm infants^[55]. Therefore, the results of this systematic review may indicate greater deviance from a healthy full-term microbiome when infants are provided with DHM as compared to MOM.

However, recent studies show that not only microbiome composition, but also its metabolic activity and effects on the host’s immune system influence different health outcomes^[56,57]. This is underscored by some studies included in this systematic review. One study highlights different concentrations of fecal SCFAs, metabolites linked to intestinal immune and endocrine responses, among different types of feeding^[31]. Another study showed higher calprotectin levels, indicative of more intestinal inflammation in the DHM group^[34].

To the best of our knowledge, this is the first systematic review focusing solely on the difference in gut microbiome composition between infants predominantly fed MOM or DHM. One other systematic review on health outcomes between MOM- and DHM-exposed infants provides information on gut microbiome diversity only when it is mentioned in the included articles^[18]. However, that review was designed to find health outcomes rather than microbiome outcomes, so key publications reporting only gut microbiota are not included. In addition, two other recent systematic reviews on the associations of feeding practices and the gut microbiome in preterm infants did not focus on DHM, resulting in the omission of relevant studies on this topic^[58,59]. However, the findings in these reviews are coherent with our results.

Our review entails multiple strengths, including the categorization of studies involving infants exclusively human milk-fed and those with mixed feeding regimens. This stratification enabled us to mitigate possible bias by the potential effects caused by any consumption of PF. Furthermore, since both gestational age and chronological age may influence microbiota composition, we reported the PMA in Figures 2 and 3 for easier comparison when analyzing stool samples. However, this approach oversimplifies the potential impact of individual maturity on gut microbiome composition. Limitations of this review include the inability to provide a quantitative synthesis due to the considerable heterogeneity in reported outcome measures in the original studies. This includes variations in the use of different diversity indices, the use of different gut microbiome analysis techniques, and the preferred reporting styles for taxonomic outcomes. Additionally, most studies did not account for the mode of delivery which may have a confounding effect on microbiota composition. Several studies excluded infants with adverse neonatal outcomes such as mortality, NEC, and sepsis, possibly omitting infants with the most aberrant gut microbiome development, thus potentially biasing the results of this review. However, this can be seen as an advantage as well, where relatively well-doing preterm infants form a more homogeneous group with less external influences, so that the influence of milk type is better recognized.

The absence of randomized controlled trials (RCTs) comparing MOM to DHM is a significant limitation of reviews on this topic. The reliance on observational studies necessitates cautious interpretation of findings due to potential confounding variables. Future research should link health outcomes to gut microbiota changes attributable to feeding with DHM vs. MOM. In addition, the mechanisms through which MOM exerts its benefits to the infant should be investigated and used to improve existing DHM processing techniques to increase benefits. For instance, Holder pasteurization results in significant loss of bioactive compounds like lactoferrin, cytokines, and growth factors, which are vital for a healthy gut microbiome in infants^[60]. Alternative pasteurization techniques such as high-temperature short-time pasteurization, high-pressure pasteurization, and Ultraviolet C may better preserve these compounds and thereby affect the microbiome in infants, although studies on this matter are sparse^[61]. A novel concept of inoculation of DHM with bits of MOM shows promising in-vitro results in restoring the human milk microbiome^[62,63]. However, in-vivo studies are needed to fully assess the effects of inoculated DHM on clinically important outcomes.

In conclusion, our review reveals several differences in gut microbiota development of preterm infants fed predominantly DHM vs. those fed predominantly MOM. The gut microbial composition shows distinct clustering based on feeding type, with the DHM group potentially displaying lower alpha diversity levels. At the compositional level, a DHM diet is associated with an increased abundance of the Staphylococcaceae and Clostridiaceae family and a lower abundance of Bacteroidetes and the Bifidobacterium genus. The differences persist throughout the follow-up period examined in the included studies. Importantly, these observed differences have been associated with adverse health outcomes both in the short and long term, such as an increased incidence of NEC, sepsis, and neurodevelopmental delay beyond infancy.