Original Investigation | Pediatrics

Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants
A Randomized Clinical Trial
Nicholas D. Embleton, MD; Thomas Sproat, MD; Sabita Uthaya, MD; Gregory R. Young, PhD; Shalabh Garg, MD; Vimal Vasu, MD; Andrea C. Masi, MRes;
Lauren Beck, MRes; Neena Modi, PhD; Christopher J. Stewart, PhD; Janet E. Berrington, MD

Abstract Key Points

Question What is the effect of an
IMPORTANCE The effect of using an exclusive human milk diet compared with one that uses bovine
exclusive human milk diet on gut
products in preterm infants is uncertain, but some studies demonstrate lower rates of key neonatal
microbiota and clinical outcomes in
morbidities. A potential mediating pathway is the gut microbiome.
preterm infants?

OBJECTIVE To determine the effect of an exclusive human milk diet on gut bacterial richness, diversity, Findings In this randomized clinical trial
and proportions of specific taxa in preterm infants from enrollment to 34 weeks’ postmenstrual age. of 126 preterm infants, there were no
significant differences in measures of
DESIGN, SETTING, AND PARTICIPANTS In this randomized clinical trial conducted at 4 neonatal gut microbial diversity in infants who
intensive care units in the United Kingdom from 2017 to 2020, microbiome analyses were blind to only received human milk products
group. Infants less than 30 weeks’ gestation who had only received own mother’s milk were recruited compared with those receiving bovine
before 72 hours of age. Statistical analysis was performed from July 2019 to September 2021. milk formula or fortifiers. There were no
differences in clinical outcomes.
INTERVENTIONS Exclusive human milk diet using pasteurized human milk for any shortfall in
Meaning These findings suggest that
mother’s own milk supply and human milk–derived fortifiers (intervention) compared with bovine
human milk products given to preterm
formula and bovine-derived fortifier (control) until 34 weeks’ postmenstrual age. Fortifier
infants to supplement a shortfall in
commenced less than 48 hours of tolerating 150 mL/kg per day.
mother’s own milk does not affect
clinical outcomes via microbial
MAIN OUTCOMES AND MEASURES Gut microbiome profile including alpha and beta diversity, and
mechanisms.
presence of specific bacterial taxa.

RESULTS Of 126 preterm infants enrolled in the study, 63 were randomized to control (median [IQR] + Supplemental content
gestation: 27.0 weeks [26.0-28.1 weeks]; median [IQR] birthweight: 910 g [704-1054 g]; 32 [51%] Author affiliations and article information are
male) and 63 were randomized to intervention (median [IQR] gestation: 27.1 weeks [25.7-28.1 weeks]; listed at the end of this article.

median [IQR] birthweight: 930 g [733-1095 g]; 38 [60%] male); 472 stool samples from 116 infants
were analyzed. There were no differences in bacterial richness or Shannon diversity over time, or at
34 weeks between trial groups. The exclusive human milk diet group had reduced relative abundance
of Lactobacillus after adjustment for confounders (coefficient estimate, 0.056; P = .03), but not after
false discovery rate adjustment. There were no differences in time to full feeds, necrotizing
enterocolitis, or other key neonatal morbidities.

CONCLUSIONS AND RELEVANCE In this randomized clinical trial in preterm infants using human
milk–derived formula and/or fortifier to enable an exclusive human milk diet, there were no effects on
overall measures of gut bacterial diversity but there were effects on specific bacterial taxa previously
associated with human milk receipt. These findings suggest that the clinical impact of human milk–
derived products is not modulated via microbiomic mechanisms.

TRIAL REGISTRATION ISRCTN trial registry identifier: ISRCTN16799022

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165

Open Access. This is an open access article distributed under the terms of the CC-BY-NC-ND License.

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 1/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

Introduction
Receipt of mother’s own breast milk (MOM) is associated with lower rates of neonatal morbidities in
preterm infants1-4 and improved long-term metabolic and neurocognitive outcomes.5-8 However,
many experience a shortfall in MOM supply necessitating the use of either bovine formula or
pasteurized human milk from donor milk banks or commercial suppliers.9,10
Only a small number of randomized clinical trials (RCTs) have explored the optimal strategy
when there is a shortfall of MOM.3 Although none of these trials were individually powered to
determine a reduction in NEC, a meta-analysis showed lower rates of NEC where donor human milk
(DHM) compared with formula milk was used as sole diet,3 but not significant in studies where DHM
or formula milk was used to make up any shortfall. In the 2 largest RCTs, one showed a lower rate of
NEC with DHM11 whereas another showed no difference.9 However, neither was powered to detect a
clinically important difference in surgical NEC.
Despite support from several professional organizations, practices toward the use of DHM in
neonatal intensive care units (NICU) vary. Data show DHM use in United Kingdom (UK) neonatal
networks ranges from less than 5% to greater than 30% of very preterm infants12,13 and many require
higher macronutrient intakes than are provided from unfortified human milk to enable growth.14
Clinical practice varies with some NICUs routinely using breast milk fortifiers whereas others may
never use fortifiers. Meta-analysis of the benefits of using fortifiers suggests faster growth but there
are no high-quality data on longer term functional outcomes and most used fortifiers of bovine
origin.15 An RCT showed significantly lower rates of NEC when an exclusive human milk diet
compared with a diet containing bovine-derived products was used,16 and a small RCT of commercial
human milk compared with formula milk as sole diet showed lower rates of NEC.17
The mechanisms of pasteurized human milk in reducing NEC may include reduced exposure to
bovine antigens, or the effects of functional components such as lactoferrin and human milk
oligosaccharides (HMOs)18-21 which may act via effects on gut microbiota. Studies have shown
differences in gut microbiota associated with dietary exposures, as well as differences prior to the
onset of NEC and sepsis, both major contributors to long-term outcomes.20,22-26 However, these
studies are observational and may be subject to residual confounding. We aimed to determine the
effect of an exclusive human milk diet to one containing bovine-derived products on gut microbiota
in very preterm infants in a randomized clinical trial. We hypothesized that gut bacterial diversity and
proportions of specific bacterial taxa would differ between trial groups as part of the mechanism by
which exclusive human milk diets benefits preterm infants. We also aimed to collect a range of
secondary clinical outcomes potentially affected by human milk compared with bovine milk.

Methods
In this randomized clinical trial we recruited preterm infants in the first 72 hours of life (born less than
30 weeks of gestation) who had not received any milk other than MOM from one of 4 NICUs in the
UK. We followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline and
provide the study protocol (Supplement 1) and CONSORT flow diagram (eFigure 1 in Supplement 2).
We excluded infants with major congenital or life-threatening abnormalities, or where study
continuation to 34 weeks’ postmenstrual age (PMA) was considered unlikely due to hospital transfer.
Based on previous work, we planned to recruit sufficient infants to have microbiome data from at
least 40 infants per group at 34 weeks’ PMA and anticipated recruitment of at least 100 infants to
allow for nonsurvivors, incomplete sample collection, or unanticipated hospital transfer. Where
infants were transferred to another hospital, we collected clinical outcomes but no further samples.
The study received approval from the Northeast, Tyne and Wear South, Research Ethics Committee
and was registered prospectively (ISRCTN16799022).
Parents had a discussion with a research team member, were given a Parent Information Sheet,
and provided written informed consent. We established a Study Steering Committee with an

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 2/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

independent clinical chair and additional members including a parent. The following amendments to
the original protocol were implemented following Study Steering and Ethics Committee approval:
(1) April 2018 after 25 recruits, approval to extend eligibility from less than 29 weeks’ gestation and
less than 48 hours age to less than 30 weeks’ gestation and less than 72 hours age; (2) August 2018,
addition of 2 further hospital sites; (3) April 2019, enrollment of more than 100 infants within original
study time window of 24 months to ensure sufficient infants with samples at 34 weeks’ PMA (see
Supplement 1).
Following parental consent, infants were randomized to standard (control) or exclusive human
milk diet (intervention) using a secure online web randomization program (Sealed Envelope [Sealed
Envelope Ltd]) with a 30% chance of simple random allocation and including 3 strata: (1) hospital
site; (2) less than 25 weeks’ gestation at birth; (3) singleton vs multiple birth status. Twins were
randomized independently. Feeding guidelines followed standard practice for the NICU with infants
commencing parenteral nutrition shortly after birth, use of buccal colostrum and early MOM,
multivitamin and iron supplements, and increases in milk feeds according to the attending clinical
team (Supplement 1).
The control group consisted of feeding with MOM and the use of preterm formula milk to make
up any shortfall MOM supply. The intervention group consisted of MOM with the use of a ready-to-
feed pasteurized human milk product (RTF 26, Prolacta Biosciences, Los Angeles, California) to make
up any shortfall in MOM. We used a breast-milk fortifier and aimed to start within 48 hours of
achieving a milk intake of 150 mL/kg per day. Infants in the control group received commercially
available, bovine-derived fortifier (Nutriprem [Nutricia Ltd] or SMA Fortifier [SMA Nutrition UK]) and
infants in the intervention group received a pasteurized human milk–derived fortifier (P+6, Prolacta
Biosciences). Two study sites used probiotics routinely and two did not. Infants remained on their
assigned diet until 34 weeks’ PMA but data on weight gain and morbidities were collected until
hospital discharge.
We recorded demographic and outcome data using the National Neonatal Research Database
and local medical records (Supplement 1) including gestation, birthweight, survival, cause of death,
intensive and high-dependency care days, age and weight at discharge, dietary exposures, days of
antibiotics, treated retinopathy of prematurity, chronic lung disease, NEC requiring surgery, and late
onset sepsis confirmed on blood culture. We planned to collect daily stool samples from the diaper,
labeled and immediately frozen on the NICU before frozen courier transfer after 2 to 4 months to a
−80 °C freezer for longer-term storage. We planned stool analysis at the following points: (A) baseline
(first sample after enrolment), (B) day of life (DOL) 10 (anticipated to be before ‘full’ feeds), (C) at full
feeds, (D) DOL 21-28, and (E) final sample collected before stopping dietary intervention at 34
weeks’ PMA.
There were 472 stool samples that underwent DNA extraction and sequencing27,28 using 100
mg of stool and the QIAGEN DNeasy PowerLyzer PowerSoil Kit. V4 16S rRNA gene sequencing was
performed by NU-OMICS using the 2x250 protocol on the Illumina MiSeq.29 Raw data processing
was performed as previously described where merging allowed zero mismatches and a minimum
overlap of 50 bases, and merged reads were trimmed at the first base with a q less than or equal to
5. Samples were rarefied at 2000 reads per sample.

Statistical Analysis
Statistical analysis was conducted from July 2019 to September 2021 in R version 3.6.2 (R Project for
Statistical Computing) and all visualizations were plotted using the ggplot2 package V.3.3.2.30
Shannon diversity and richness were calculated for each sample using the vegan package V.2.5-7. To
determine which covariates were associated with metagenome profiles at each time point, multiple
cross-sectional analyses using the adonis function from the vegan package were performed based
on Bray-Curtis dissimilarity. Each test was performed in a stepwise manner and subsequent P values
were adjusted for multiple comparisons using false discovery rate (FDR) adjustment (Benjamini-
Hochberg procedure).31 Linear mixed models (LMMs) were fit to the data using the lme4 package

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 3/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

with subject ID included as a random group intercept. The Multivariate Association with Linear
models 2 (MaAsLin2) package V.1.2.032 was used to determine significant genera associated with
covariates, while adjusting for potential confounders.33 All covariates used in the adonis analysis plus
DOL were included as fixed effects in the analysis, and subject ID was included as a random effect.
The arcsin square root transformation was performed on relative abundance data and default
MaAsLin2 parameters were used, with reporting based on P < .05 and q values provided in eTable 2
in Supplement 2. All other P values were adjusted for multiple comparisons using FDR adjustment
(Benjamini-Hochberg procedure), where adjusted P<.05 was considered significant.

Results
We enrolled 126 preterm infants, of which 63 were randomized to control (median [IQR] gestation:
27.0 weeks [26.0-28.1 weeks]; median [IQR] birthweight: 910 g [704-1054 g]; 32 [51%] male) and 63
were randomized to intervention (median [IQR] gestation: 27.1 weeks [25.7-28.1 weeks]; median
[IQR] birthweight: 930 g [733-1095 g]; 38 [60%] male) between September 2017 and September
2019 (eFigure 1 in Supplement 2). Demographic data are provided in Table 1. There were 86 infants
(68%) born at less than 28 weeks’ gestation and 76 (60%) had a birthweight less than 1000 g.
There were no significant group differences in prespecified clinical outcomes (Table 2). Twelve
infants died (4 in the control group and 8 in the intervention group) at a median (IQR) postnatal age
of 25 days (14-60 days) and 15 days (10-23 days), respectively (eTable 1 in Supplement 2). In the
infants who died, formula or ready-to-feed human milk represented less than 1% and 1% of all fluid
intake (including parenteral nutrition), respectively; and the median (range) proportion of all enteral
intake that formula and ready-to-feed represented in the control group was 1% (0%-100%) and 24%
(0%-99%) in the intervention group. None of the infants who died had NEC as the primary cause,
although surgical NEC affected 1 baby who died from each group. There was one serious adverse
event not prespecified in the protocol of a 23-week gestation female infant in the intervention group
diagnosed with late onset vitamin K deficiency bleeding on DOL 73. She received additional vitamin
K but did not develop serious sequelae and has normal development at 2 years.
We selected and analyzed 526 stool samples from 119 infants. Following raw data processing
and rarefaction at 2000 reads, a total of 472 stool samples from 116 preterm infants (n = 58
intervention; n = 58 control) remained in the analysis. The number of samples included at each time
point was as follows: 92 at time point A (median [IQR] DOL, 5 [4-7]), 89 at time point B (median [IQR]
DOL, 10 [10-11]), 96 at time point C (median [IQR] DOL, 15 [14-20]), 98 at time point D (median [IQR]
DOL, 25 [24-28]); 97 at time point E (median [IQR] DOL, 44 [36-52]) (eFigure 2 in Supplement 2).

Table 1. Baseline Characteristics of Recruited Infants by Study Group

Infants, No. (%)

Characteristic Control (n = 63) Intervention (n = 63)
Gestation, median (IQR), wk 27.0 (26.0 to 28.1) 27.14 (25.7 to 28.1)
Birthweight, median (IQR), g 910 (704 to 1054) 930 (733 to 1095)
SDS birthweight, median (IQR) −0.43 (−1.10 to 0.14) −0.42 (−0.85 to 0.17)
Multiple 15 (24) 15 (24)
Sex
Male 32 (51) 38 (60)
Female 31 (49) 25 (40)
Weight, g
<1000 40 (63) 36 (57)
>1250 5 (9) 8 (13)
Cesarean delivery 35 (56) 37 (59)
Antenatal steroids (any) 56 (88) 57 (90)
Abbreviation: SDS, standard deviation score.

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 4/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

We first sought to understand what demographic or outcome factors had the strongest
associations with the overall developing gut microbiome profiles through each time point. This
analysis found that NICU site, and the use of probiotics were the only covariates that had a
statistically significant association with the bacterial profiles at more than one time point, observed
from time points C to E, but not before (Figure 1). Sex was associated with bacterial profiles at time
point E only. Trial group was not found to be associated with bacterial profiles at any time point.
Because samples predominantly clustered by probiotic use in the NICU (eFigure 3 in Supplement 2),
we stratified analysis by probiotic use and focused on time point E (34 weeks’ PMA, study
completion). This stratified analysis based on unweighted UniFrac (ie, bacterial presence or absence)
showed a significant difference in trial group within NICUs using probiotics (R2 = 0.0791; P = .001),
but not nonprobiotic NICUs (R2 = 0.0573; P = .09) (Figure 2A). Weighted UniFrac (ie, considers
relative abundance) showed no association between trial group and bacterial profiles within
probiotic (R2 = 0.006; P = .76) and nonprobiotic (R2 = 0.056; P = .14) sites (Figure 2B).
We then used a generalized linear mixed model to explore longitudinal alpha diversity while
accounting for repeated measures, and included all samples, with adjustment for all covariates listed
in Figure 2A, showing that trial group (control vs intervention) was not associated with a difference
in the bacterial richness (coefficient estimate, −0.951; P = .23) or Shannon diversity (coefficient
estimate, –0.094; P = .06). Analysis of samples collected at time point E only showed both richness
(7 vs 6; P = .02) and Shannon diversity (1.2 vs 0.9; P = .04) were significantly associated with trial
group at study completion (eFigure 4A and 4B in Supplement 2). Further stratified analysis showed
the difference was primarily driven by NICUs using probiotics (eFigure 4C and 4D in Supplement 2).
The relative abundance of bacterial genera was explored using MaAsLin2 (Microbiome
Multivariable Association with Linear Models), which included all samples and time points, while
adjusting for covariates and controlling for repeated measures. In accordance with previous
work,18,27,34 Staphylococcus (coefficient estimate −0.173; P < .001) and Corynebacterium (coefficient
estimate, −0.006; P = .01) were negatively correlated with infant DOL, whereas Enterobacter
(coefficient estimate, 0.125; P < .001), Veillonella (coefficient estimate, 0.034; P < .001), Clostridium
(coefficient estimate, 0.0184; P < .001), Escherichia (coefficient estimate, 0.050; P < .001),
Bifidobacterium (coefficient estimate, 0.042; P < .001) were positively correlated with DOL (eTable 2
in Supplement 2). In addition to DOL, Bifidobacterium was higher in relative abundance in NICUs
using probiotics (coefficient estimate, 0.285; P < .001), as well as positively correlated with days of
MOM (P = .01) and negatively correlated with the number of days of antibiotics (coefficient estimate,
−0.044; P = .02) (eTable 2 in Supplement 2). The intervention trial group was significantly associated

Table 2. Clinical Outcomes for Recruited Infants by Study Groupa

Infants, No. (%)

Variable Control (n = 63) Intervention (n = 63) P value
Days of PN, median (IQR)b 13 (9 to 18) 11 (9 to 20) .67
Days of MOM intake, median (IQR)b 56 (33 to 76) 45 (21 to 74) .47
Any MOM intake at 34 wc 41 (68) 36 (65) .26
Days of antibiotics, median (IQR)b 11 (6.5 to 19) 12 (6 to 19.5) .92
Blood culture positive LOSb 8 (13) 6 (10) .78
Surgical NECb 2 (3) 1 (2) >.99 Abbreviations: LOS, late onset sepsis; MOM, mother’s
ROP requiring treatmentc 2 (3) 2 (3) >.99 own milk; NEC, necrotizing enterocolitis; PMA,
postmenstrual age; PN, parenteral nutrition; ROP,
Chronic lung diseasec 40 (67) 37 (67) >.99
retinopathy of prematurity; SDS standard
Diedb 4 (6) 8 (13) .23
deviation score.
PMA at hospital discharge, median (IQR), wkc 38.7 (37.3 to 40.9) 38.6 (36.9 to 40.1) .39 a
Data that are median (IQR) were analyzed using
Intensive care + high dependency days, 47 (26 to 81) 46 (20 to 78) .80 Mann-Whitney U test, mean (SD) were analyzed
median (IQR)b
using t tests, and No. (%) were analyzed using Fisher
Change in weight SDS, mean (SD)c
exact test.
Birth to 34 weeks −0.83 (0.59) −0.88 (0.60) .63 b
All 126 infants.
Birth to discharge −0.81 (0.95) −0.94 (0.97) .48 c
Excludes deaths prior to outcome.

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 5/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

Figure 1. Significance and Explained Variance (R2) of Covariates at Each Time Point

A Time point A

0.15

0.10
R2

0.05

0
Gestational Mode of Birth Sex Center Probiotic Days' MOM at Total days' NEC Trial Day
age delivery weight unit MOM 34 wk antibiotics or LOS group of life
Covariate
B Time point B
0.15

0.10
R2

0.05

0
Gestational Mode of Birth Sex Center Probiotic Days' MOM at Total days' NEC Trial Day
age delivery weight unit MOM 34 wk antibiotics or LOS group of life
Covariate
C Time point C
0.15

a
0.10
a
R2

0.05

0
Gestational Mode of Birth Sex Center Probiotic Days' MOM at Total days' NEC Trial Day
age delivery weight unit MOM 34 wk antibiotics or LOS group of life
Covariate
D Time point D

0.15

a
0.10
a
R2

0.05

0
Gestational Mode of Birth Sex Center Probiotic Days' MOM at Total days' NEC Trial Day
age delivery weight unit MOM 34 wk antibiotics or LOS group of life
Covariate
E Time point E
a
0.15

0.10 a
R2

0.05 a

0 Plot based on the results of adonis within each time

Gestational Mode of Birth Sex Center Probiotic Days' MOM at Total days' NEC Trial Day
age delivery weight unit MOM 34 wk antibiotics or LOS group of life point. MOM indicates mother’s own milk.
a
Covariate Denotes P < .05.

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 6/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

with a reduced relative abundance of Lactobacillus, (coefficient estimate, 0.056; P = .03) (eTable 2
in Supplement 2).

Discussion
We conducted an RCT in preterm infants exploring the effect of an exclusive human milk diet, with
prospective stool sampling and gut microbiota analyses. Exploration of the microbiome is complex,
and no single measure fully reflects the microbial community. We found that trial group had no
overall effect on gut microbiome richness or Shannon diversity, although infants randomized to
exclusive human milk had different unweighted microbiome profiles, but not in weighted analysis.
This suggests that the exclusive human milk diet has a greater effect on the low abundance taxa (ie,
those with low weighting in a weighted analysis). This is consistent with the finding that
Lactobacillus, a relatively low abundance genera was the only significant genera that differed
between trial groups, being lower in infants receiving the exclusive human milk diet. Lactobacillus
spp have previously been shown to be increased in breast-fed infants,35 although whether this is via
direct transfer of breast milk microbiome, or metabolism of breast milk–derived substrates is not
understood. Although the group difference in Lactobacillus was significant after adjustment for
confounders, this was no longer significant after FDR adjustment. Thus, the finding of lower
Lactobacillus relative abundance in the exclusive human milk diet differs to existing work and is
unlikely to be explained by human milk processing. Furthermore, our findings differ to a similar trial
that showed lower microbial diversity and higher relative abundances of Enterobacteriaceae and
lower abundances of Clostridium senus strictu in preterm infants also receiving an exclusive human
milk diet.36

Figure 2. Bacterial Profiles Between Trial Group at Time Point E (Study Completion) Only, Stratified by Probiotic Use on the NICU

Trial group
Control Intervention

A Unweighted UniFrac overall B Unweighted UniFrac, nonprobiotic unit C Unweighted UniFrac, probiotic unit
0.4 0.4 0.2
R2 = 0.0424; R2 = 0.0573;
P =.001 P =.09
0.2
0.2 0

0
PC2

PC2

PC2

0 –0.2
–0.2
R2 = 0.0791;
P =.001
–0.4 –0.2 –0.4

–0.25 0 0.25 0.50 –0.50 –0.25 0 0.25 0.50 –0.2 0 0.2 0.4
PC1 PC1 PC1

D Weighted UniFrac overall E Weighted UniFrac, nonprobiotic unit F Weighted UniFrac, probiotic unit
0.4 0.6 0.50
R2 = 0.0108; R2 = 0.0558; R2 = 0.00602;
P =.38 P =.14 P =.76
0.2 0.4 0.25

0 0.2
PC2

PC2

PC2

–0.2 0
–0.25

–0.4 –0.2
–0.50
–0.25 0 0.25 0.50 0.75 –0.2 0 0.2 0.4 –0.25 0 0.25 0.50 0.75
PC1 PC1 PC1

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 7/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

Bifidobacterium spp are enriched in the gut microbiome of term infants receiving breast milk
compared with formula, thought to be due to the provision of HMOs. In the current study,
Bifidobacterium relative abundance was not associated with an exclusive human milk diet. However,
all infants received substantial intakes of MOM which may have minimized differences between the
2 groups in terms of intakes of bioactive components.
Supplemental probiotics received by some infants in the study also exert major effects on the
gut microbiota,33 but analyses adjusting for probiotic use, as well as analyses focused on the study
completion time point stratified by probiotic site showed no differences.
There were no important group differences in key neonatal morbidities, weight gain, or length
of stay. Although the liquid fortifier used in the trial intervention group contributed 30% of intake
volume, there was no difference in days of MOM exposure or the number of infants still receiving
MOM at the end of the trial.

Strengths and Limitations

This study had some strengths and limitations. The strengths of this study are the RCT study design
with robust data collection and blinding of the analyses to group. Nonsurvival and unanticipated
hospital transfer prior to 34 weeks’ PMA was higher than anticipated so we made protocol
modifications to ensure we reached target recruitment within the 24-month window. We recruited
infants from 4 tertiary level NICUs representative of UK neonatal practice but acknowledge that
NICU-site has a large effect on gut microbiota and there may have been small differences in feeding
protocols. Microbiomic analysis used 16S rRNA gene sequencing but may not detect differences at
species level which require metagenomic approaches; metabolomic analysis may provide additional
insights. Changes in stool microbiota may not reflect microbiota in the small intestine where NEC
usually occurs. Additionally, although we had analyzed around half of our samples by February 2020,
our sequencing facilities became unavailable for the remainder of 2020 due to the COVID-19
pandemic during which all samples were stored at −80 °C.

Conclusions
The mechanisms of action of human milk in reducing NEC or other morbidities are poorly understood
but this study suggests that pasteurized human milk (or products derived from human milk) do not
exert a major impact on gut bacteria when used in addition to MOM, in contrast to observational data
in term infants where MOM exposure explains the largest differences in microbial patterns.35
However, in settings where MOM usage is low the impact may differ. It is possible that diet exerts
effects on health or disease without changes in gut microbiota, for example effects on the gut
epithelium. However, observational data from our group showed significant differences in the
exposure to specific HMOs in MOM between preterm infants who developed NEC associated with
changes in gut microbiota.18 Furthermore, we also demonstrated that gut microbiota in preterm
infants is rapidly changing, is NICU-specific and while differences exist before NEC, the sensitivity and
specificity of individual microbiomic parameters for subsequent disease is poor.24,26 Studies
exploring metabolomic correlates of the diet may provide further insights although adequately
powered RCTs are needed to determine the optimal clinical approach.23,37,38

ARTICLE INFORMATION
Accepted for Publication: January 14, 2023.
Published: March 1, 2023. doi:10.1001/jamanetworkopen.2023.1165
Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License. © 2023
Embleton ND et al. JAMA Network Open.

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 8/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

Corresponding Authors: Nicholas D. Embleton, MD, Ward 35 Neonatal Unit, Newcastle Hospitals NHS Trust, Royal
Victoria Infirmary, Richardson Road, Newcastle upon Tyne NE1 4LP, United Kingdom (nicholas.embleton@
ncl.ac.uk); Christopher J. Stewart, PhD, Translational and Clinical Research Institute, Newcastle University,
Newcastle upon Tyne, United Kingdom (christopher.stewart@ncl.ac.uk).
Author Affiliations: Newcastle Hospitals NHS Trust, Royal Victoria Infirmary, Newcastle upon Tyne, United
Kingdom (Embleton, Sproat, Berrington); Population Health Sciences Institute, Newcastle University, Newcastle
upon Tyne, United Kingdom (Embleton); Translational and Clinical Research Institute, Newcastle University,
Newcastle upon Tyne, United Kingdom (Sproat, Masi, Beck, Stewart, Berrington); Section of Neonatal Medicine,
School of Public Health, Imperial College London, London, United Kingdom (Uthaya, Modi); Chelsea and
Westminster Hospital, NHS Foundation Trust, London, United Kingdom (Uthaya, Modi); Northumbria University,
Newcastle upon Tyne, United Kingdom (Young); James Cook University Hospital, Middlesbrough, United Kingdom
(Garg); William Harvey Hospital, Ashford, Kent, United Kingdom (Vasu).
Author Contributions: Dr Embleton and Dr Stewart had full access to all of the data in the study and take
responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Embleton, Uthaya, Modi, Stewart, Berrington.
Acquisition, analysis, or interpretation of data: Embleton, Sproat, Uthaya, Young, Garg, Vasu, Masi, Beck, Stewart,
Berrington.
Drafting of the manuscript: Embleton, Beck, Stewart, Berrington.
Critical revision of the manuscript for important intellectual content: Embleton, Sproat, Uthaya, Young, Garg, Vasu,
Masi, Modi, Berrington.
Statistical analysis: Beck, Stewart, Berrington.
Obtained funding: Embleton, Uthaya, Modi, Stewart.
Administrative, technical, or material support: Embleton, Sproat, Young, Garg, Masi, Stewart, Berrington.
Supervision: Uthaya, Young, Stewart, Berrington.
Conflict of Interest Disclosures: Dr Embleton reported receiving grants from Danone Early Life Nutrition,
personal fees from Nestle Nutrition Institute Lecture (honoraria), personal fees from Astarte Lecture (honoraria),
and grants from NeoKare outside the submitted work; and Dr Embleton also provides nonremunerated advice to
NHS milk banks and is a previous member of ESPGHAN Committee of Nutrition. Dr Modi reported receiving grants
from Imperial College London during the conduct of the study; grants from Medical Research Council, grants from
National Institute for Health Research, grants from Chiesi Pharmaceuticals, grants from C-Path, grants from
European Health Data and Evidence Network, and grants from Health Data Research UK outside the submitted
work. Dr Stewart reported receiving grants from Wellcome Trust and grants from Lister Institute during the
conduct of the study; personal fees from Astarte Medical and personal fees from Nestle Nutrition Institute outside
the submitted work. Dr Berrington reported receiving grants from Danone during the conduct of the study;
personal fees from Nestle Nutritional Institute outside the submitted work. No other disclosures were reported.
Funding/Support: The study was sponsored by Newcastle Hospitals NHS Foundation Trust, Newcastle-upon-
Tyne, UK, and was funded by Prolacta Biosciences, California US who also provided human milk formula and
fortifier.
Role of the Funder/Sponsor: The sponsor and funder had no role in the design and conduct of the study;
collection, management, analysis, and interpretation of the data; preparation, review, or approval of the
manuscript; and decision to submit the manuscript for publication.
Data Sharing Statement: See Supplement 3.
Additional Contributions: We gratefully acknowledge the support of the following research nurses: Julie
Groombridge (Newcastle upon Tyne), Amanda Forster (JCUH, Middlesbrough), Shermi George, Claire Moloney
(William Harvey Hospital, East Kent), Suzan Jeffries and Izabela Andrewjewska (Chelsea and Westminster hospital,
London). They were not compensated. We are also grateful for the support provided by clinical teams, R&D
departments, and all the parents who have supported our research.

REFERENCES
1. Zhou J, Shukla VV, John D, Chen C. Human milk feeding as a protective factor for retinopathy of prematurity:
a meta-analysis. Pediatrics. 2015;136(6):e1576-e1586. doi:10.1542/peds.2015-2372
2. Villamor-Martínez E, Pierro M, Cavallaro G, Mosca F, Kramer BW, Villamor E. Donor human milk protects against
bronchopulmonary dysplasia: a systematic review and meta-analysis. Nutrients. 2018;10(2):9-15. doi:10.3390/
nu10020238
3. Quigley M, Embleton ND, McGuire W. Formula versus donor breast milk for feeding preterm or low birth weight
infants. Cochrane Database Syst Rev. 2018;6(6):CD002971. doi:10.1002/14651858.CD002971.pub4

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 9/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

4. Patel AL, Johnson TJ, Engstrom JL, et al. Impact of early human milk on sepsis and health-care costs in very low
birth weight infants. J Perinatol. 2013;33(7):514-519. doi:10.1038/jp.2013.2
5. Isaacs EB, Morley R, Lucas A. Early diet and general cognitive outcome at adolescence in children born at or
below 30 weeks gestation. J Pediatr. 2009;155(2):229-234. doi:10.1016/j.jpeds.2009.02.030
6. Isaacs EB, Gadian DG, Sabatini S, et al. The effect of early human diet on caudate volumes and IQ. Pediatr Res.
2008;63(3):308-314. doi:10.1203/PDR.0b013e318163a271
7. Singhal A, Cole TJ, Fewtrell M, Deanfield J, Lucas A. Is slower early growth beneficial for long-term
cardiovascular health? Circulation. 2004;109(9):1108-1113. doi:10.1161/01.CIR.0000118500.23649.DF
8. Fewtrell MS, Williams JE, Singhal A, Murgatroyd PR, Fuller N, Lucas A. Early diet and peak bone mass: 20 year
follow-up of a randomized trial of early diet in infants born preterm. Bone. 2009;45(1):142-149. doi:10.1016/j.
bone.2009.03.657
9. Corpeleijn WE, de Waard M, Christmann V, et al. Effect of donor milk on severe infections and mortality in very
low-birth-weight infants: The Early Nutrition Study randomized clinical trial. JAMA Pediatr. 2016;170(7):654-661.
doi:10.1001/jamapediatrics.2016.0183
10. Unger S, Gibbins S, Zupancic J, O’Connor DL. DoMINO: donor milk for improved neurodevelopmental
outcomes. BMC Pediatr. 2014;14(1):123. doi:10.1186/1471-2431-14-123
11. O’Connor DL, Gibbins S, Kiss A, et al; GTA DoMINO Feeding Group. Effect of supplemental donor human milk
compared with preterm formula on neurodevelopment of very low-birth-weight infants at 18 months:
a randomized clinical trial. JAMA. 2016;316(18):1897-1905. doi:10.1001/jama.2016.16144
12. Zipitis CS, Ward J, Bajaj R. Use of donor breast milk in neonatal units in the UK. Arch Dis Child Fetal Neonatal
Ed. 2014;100(3):F279-81. doi:10.1136/archdischild-2014-307606
13. Battersby C, Marciano Alves Mousinho R, Longford N, Modi N; UK Neonatal Collaborative Necrotising (UKNC-
NEC) Study Group. Use of pasteurised human donor milk across neonatal networks in England. Early Hum Dev.
2018;118(118):32-36. doi:10.1016/j.earlhumdev.2018.01.017
14. Agostoni C, Buonocore G, Carnielli VP, et al; ESPGHAN Committee on Nutrition. Enteral nutrient supply for
preterm infants: commentary from the European Society of Paediatric Gastroenterology, Hepatology and
Nutrition Committee on Nutrition. J Pediatr Gastroenterol Nutr. 2010;50(1):85-91. doi:10.1097/MPG.
0b013e3181adaee0
15. Brown JV, Embleton ND, Harding JE, McGuire W. Multi-nutrient fortification of human milk for preterm infants.
Cochrane Database Syst Rev. 2016;2016(5):CD000343. doi:10.1002/14651858.CD000343.pub3
16. Sullivan S, Schanler RJ, Kim JH, et al. An exclusively human milk-based diet is associated with a lower rate of
necrotizing enterocolitis than a diet of human milk and bovine milk-based products. J Pediatr. 2010;156(4):
562-7.e1. doi:10.1016/j.jpeds.2009.10.040
17. Cristofalo EA, Schanler RJ, Blanco CL, et al. Randomized trial of exclusive human milk versus preterm formula
diets in extremely premature infants. J Pediatr. 2013;163(6):1592-1595.e1. doi:10.1016/j.jpeds.2013.07.011
18. Masi AC, Embleton ND, Lamb CA, et al. Human milk oligosaccharide DSLNT and gut microbiome in preterm
infants predicts necrotising enterocolitis. Gut. 2021;70(12):2273-2282. doi:10.1136/gutjnl-2020-322771
19. Sproat T, Payne RP, Embleton ND, Berrington J, Hambleton S. T cells in preterm infants and the influence of
milk diet. Front Immunol. 2020;11:1035. doi:10.3389/fimmu.2020.01035
20. Granger CL, Embleton ND, Palmer JM, Lamb CA, Berrington JE, Stewart CJ. Maternal breastmilk, infant gut
microbiome and the impact on preterm infant health. Acta Paediatr. 2021;110(2):450-457. doi:10.1111/apa.15534
21. Demers-Mathieu V, Huston RK, Markell AM, et al. Differences in maternal immunoglobulins within mother’s
own breast milk and donor breast milk and across digestion in preterm infants. Nutrients. 2019;11(4):1-14. doi:10.
3390/nu11040920
22. Berrington JE, Stewart CJ, Embleton ND, Cummings SP. Gut microbiota in preterm infants: assessment and
relevance to health and disease. Arch Dis Child Fetal Neonatal Ed. 2013;98(4):F286-F290. doi:10.1136/
archdischild-2012-302134
23. Stewart CJ, Nelson A, Treumann A, et al. Metabolomic and proteomic analysis of serum from preterm infants
with necrotising entercolitis and late-onset sepsis. Pediatr Res. 2016;79(3):425-431. doi:10.1038/pr.2015.235
24. Stewart CJ, Marrs ECL, Nelson A, et al. Development of the preterm gut microbiome in twins at risk of
necrotising enterocolitis and sepsis. PLoS One. 2013;8(8):e73465. doi:10.1371/journal.pone.0073465
25. Abdulkadir B, Nelson A, Skeath T, et al. Stool bacterial load in preterm infants with necrotising enterocolitis.
Early Hum Dev. 2016;95:1-2. doi:10.1016/j.earlhumdev.2016.01.018

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 10/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023

 JAMA Network Open | Pediatrics Effect of an Exclusive Human Milk Diet on the Gut Microbiome in Preterm Infants

26. Stewart CJ, Marrs ECL, Magorrian S, et al. The preterm gut microbiota: changes associated with necrotizing
enterocolitis and infection. Acta Paediatr. 2012;101(11):1121-1127. doi:10.1111/j.1651-2227.2012.02801.x
27. Stewart CJ, Embleton ND, Marrs ECL, et al. Longitudinal development of the gut microbiome and metabolome
in preterm neonates with late onset sepsis and healthy controls. Microbiome. 2017;5(1):75. doi:10.1186/s40168-
017-0295-1
28. Stewart CJ, Embleton ND, Marrs EC, et al. Temporal bacterial and metabolic development of the preterm gut
reveals specific signatures in health and disease. Microbiome. 2016;4(1):67. doi:10.1186/s40168-016-0216-8
29. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a dual-index sequencing
strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform.
Appl Environ Microbiol. 2013;79(17):5112-5120. doi:10.1128/AEM.01043-13
30. Wickham H, Chang W, Henry L, et al. Elegant graphics for data analysis. Accessed January 25, 2023. https://
ggplot2.tidyverse.org
31. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple
testing. J R Stat Soc B. 1995;57:289-300. doi:10.1111/j.2517-6161.1995.tb02031.x
32. Mallick H, Rahnavard A, McIver LJ, et al. Multivariable association discovery in population-scale meta-omics
studies. PLoS Comput Biol. 2021;17(11):e1009442. doi:10.1371/journal.pcbi.1009442
33. Beck LC, Masi AC, Young GR, et al. Strain-specific impacts of probiotics are a significant driver of gut
microbiome development in very preterm infants. Nat Microbiol. 2022;7(10):1525-1535. doi:10.1038/s41564-022-
01213-w
34. Stewart CJ, Marrs ECL, Magorrian S, et al. The preterm gut microbiota: changes associated with necrotizing
enterocolitis and infection. Acta Paediatr. 2012;101(11):1121-1127. doi:10.1111/j.1651-2227.2012.02801.x
35. Stewart CJ, Ajami NJ, O’Brien JL, et al. Temporal development of the gut microbiome in early childhood from
the TEDDY study. Nature. 2018;562(7728):583-588. doi:10.1038/s41586-018-0617-x
36. Asbury MR, Shama S, Sa JY, et al; OptiMoM Feeding Group. Human milk nutrient fortifiers alter the developing
gastrointestinal microbiota of very-low-birth-weight infants. Cell Host Microbe. 2022;30(9):1328-1339.e5. doi:10.
1016/j.chom.2022.07.011
37. Wilcock A, Begley P, Stevens A, Whatmore A, Victor S. The metabolomics of necrotising enterocolitis in
preterm babies: an exploratory study. J Matern Fetal Neonatal Med. 2016;29(5):758-762. doi:10.3109/14767058.
2015.1017462
38. Morrow AL, Lagomarcino AJ, Schibler KR, et al. Early microbial and metabolomic signatures predict later onset
of necrotizing enterocolitis in preterm infants. Microbiome. 2013;1(1):13. doi:10.1186/2049-2618-1-13

SUPPLEMENT 1.
Trial Protocol

SUPPLEMENT 2.
eFigure 1. CONSORT Flow Diagram
eTable 1. Causes of Death in Recruited Infants
eFigure 2. Schematic Showing Samples Included From Each Infant Per Day of Life
eFigure 3. Importance of Probiotic Use in Shaping the Preterm Gut Microbiome
eFigure 4. Boxplots of Shannon Diversity at Time Point E (Study Completion) Stratified by Probiotic Use
eTable 2. MaAsLin2 Results for Significant Taxa at the Genus Level

SUPPLEMENT 3.
Data Sharing Statement

JAMA Network Open. 2023;6(3):e231165. doi:10.1001/jamanetworkopen.2023.1165 (Reprinted) March 1, 2023 11/11

Downloaded From: https://jamanetwork.com/ by b Wildberries on 03/02/2023