The Human Milk Oligosaccharide, 3′SL, in pre-weaning Milk influences Attention, Learning and Memory later in Life

Editor(s): Pisa E, Tomasi F, Macri S, Hauser J.

2′-Fucosyllactose Is Well Tolerated in a 100% Whey, Partially Hydrolyzed Infant Formula With Bifidobacterium lactis: A Randomized Controlled Trial

Human milk oligosaccharides are important components of breast milk. We evaluated feeding tolerance of the human milk oligosaccharide 2′-fucosyllactose (2′FL) in a 100% whey, partially hydrolyzed infant formula with the probiotic Bifidobacterium animalis ssp lactis strain Bb12 (β lactis, Test) as compared with the same formula without 2′FL (Control) in a randomized controlled trial of healthy infants enrolled at 2 weeks of age (±5 days). After 6 weeks of feeding the assigned formula, the primary outcome of tolerance was assessed using the Infant Gastrointestinal Symptom Questionnaire. Stooling, vomiting, spit-up, crying, and fussing were compared between groups. Seventy-nine infants were enrolled and 63 completed the study per protocol (30 Test, 33 Control). Infant Gastrointestinal Symptom Questionnaire scores were similar between groups (Test 20.9 ± 4.8, Control 20.7 ± 4.3, P = .82). Partially hydrolyzed infant formula with 2′FL and B lactis is tolerated well, as confirmed by a validated multi-symptom index.


Human milk oligosaccharides (HMOs) are unique and functional constituents of breast milk, being its third most abundant solid component, only lower in concentration than lactose and lipids.1-3 Approximately 200 different HMOs have been identified in breast milk with variable diversity and quantity of linear, branched, fucosylated, and sialylated HMOs occurring in all lactating women.4 2′-Fucosyllactose (2′FL) is a trisaccharide composed of glucose, galactose, and fucose, and it is one of the HMOs occurring in greatest predominance in human milk.3 Levels of 2′FL vary in human milk depending on a woman’s secretor blood group status, ethnicity, and stage of lactation.5 Based on an international assessment of HMO levels in human milk at approximately 2 months postpartum among healthy women living in 8 countries, levels of 2′FL ranged from a mean of 0.7 ± 0.1 g/L in milk collected in Ghana to a mean of 3.4 ± 0.4 g/L in milk collected in California.6 In contrast to human milk, bovine milk contains low levels of oligosaccharides with low prevalence of fucosylated oligosaccharides.7
Evidence suggests that HMOs are important components of the innate immunity provided by human milk that help provide direct protection as well as contribute to the development of the infant’s immune system.8 2′FL, in particular, serves as a selective prebiotic, an antiadhesive molecule against pathogens, and appears to modulate cellular immune responses.2,7-9 In vitro, 2′FL has been shown to block adhesion of Campylobacter jejuni, Rotavirus, Norovirus, and enterotoxigenic Escherichia coli, organisms known, in vivo, to cause infectious diarrhea in infants.10-13 Additionally, 2′FL has been shown to have direct cellular effects including ability to suppress inflammation, inhibit proliferation of peripheral blood mononuclear cells, and bind to lectins expressed on dendritic cells in vitro.14-16 Evidence from observational studies have further reflected a role of 2′FL in infant health. Infants breastfed by secretor mothers were reported to have enhanced levels of bifidobacteria and have earlier establishment of bifidobacteria-rich microbiota compared with infants breastfed by nonsecretor mothers.17 In an observational study of 93 breastfeeding mother-infant pairs, incidence of pathogen-specific diarrhea was significantly lower in infants who received 2′FL from their mothers’ milk compared with infants of nonsecretor mothers.18 In another study, Sprenger et al found significant associations between levels of 2′FL in mother’s breast milk and “any allergic disease,” immunoglobulin (Ig) E–associated disease, eczema, and IgE-associated eczema in Caesareansection-born infants, with higher 2′FL levels associated with lower risk.19
Randomized clinical trials evaluating the effects of 2′FL added to infant formulas have demonstrated safety as well as potential clinical benefits. Initial evidence came from a trial evaluating effects of 2′FL and lacto-n-neotetraose in an intact whey and casein-based infant formula wherein normal growth and feeding tolerance were demonstrated within the secondary outcomes, infants fed the formula with HMOs received fewer antibiotics and antipyretics than control-fed infants and had a gut microbiota closer to that of breastfed infants at the genus level.20,21 Another published trial similarly described appropriate growth in infants fed formulas containing 2′FL at levels of 0.2 and 1.0 g per liter of formula.22 Selective cytokine profiles of the infants fed formulas with 0.2 g/L or 1.0 g/L 2′FL were found to be similar to those of reference breastfed infants.23 Infant formulas with partially hydrolyzed whey protein (PHF-W) have beneficial effects on some manifestations of functional gastrointestinal (GI) disorders, such as regurgitation and stool consistency, and are often considered as nutrition options when formula-fed infants experience tolerance-related issues.24,25 Infant formula with a 100% whey, partially hydrolyzed protein base has also been shown to promote softer stools compared with stools of infants fed intact protein-based formula.25  The clinical trial described in this article was designed to assess the tolerance of this new PHF-W and probiotic Bifidobacterium animalis ssp lactis strain Bb12 (B lactis) with added 2′FL using a validated tool encompassing multiple parameters of feeding tolerance.26


Study Population

Healthy, full-term (≥37 weeks gestation; ≥2500 and ≤4500 g birth weight), singleton infants, ages 14 ± 5 days, who had been exclusively formula-fed for at least 3 days prior to enrollment were recruited for this trial. Infants already participating in a conflicting clinical trial and those being treated for reflux were excluded.

Ethical Approval and Informed Consent

The study protocol was approved by the Chesapeake Research Review, Inc, Institutional Review Board, Columbia, Maryland, with protocol reference 00022536. Informed consent was obtained from the parents or guardians of participating infants. Good clinical practice was followed by all sites throughout the study. The study was registered with ClinicalTrials. gov (NCT# NCT03307122).

Study Design

This was a randomized, controlled, double-blind multicenter study conducted from September 2017 to February 2018 at 7 sites throughout the United States. Subjects were enrolled at 14 ± 5 days of age after obtaining informed consent from their caregivers. Randomization to the Test or Control formula (Table 1) was performed by computer-generated assignment embedded in the centralized electronic data capture system. The randomization was performed in such a manner to assign subjects to formulas in a 1:1 ratio. Both formulas were made from 100% whey protein that was partially hydrolyzed, contained B lactis, and provided 0.67 kcal/mL and 2.2 g protein/L. The only difference between the 2 formulas was the addition of 0.25 g/L 2′FL to the Test formula. Subjects/caregivers, support staff, sponsor project manager, statisticians, and investigators were blinded as to the identity of the study formulas. The products were labeled with the same protocol number and color of label but were distinguished by the codes on the labels (4KN and MP9). At the enrollment visit (V0), the Infant Gastrointestinal Symptom Questionnaire (IGSQ) was administered and anthropometric measurements were taken by trained study staff. Caregivers began to feed the subjects with the randomly assigned formulas ad libitum after V0. After 42 days of feeding, subjects returned for a second visit (V1). For 2 days before V1, caregivers completed a diary of formula 

intake, stooling, spit-up, and vomit. At V1, the IGSQ and anthropometric measurements were repeated.

The Infant Gastrointestinal Symptom Burden Questionnaire

An infant’s ability to tolerate a formula or food can be assessed based on several parameters, including stool patterns, frequency of spitting up, degrees of flatulence, and general demeanor or mood. Caregiver reports on these parameters can be combined to yield a GI symptom burden score using the IGSQ. The IGSQ is a validated 13-item questionnaire that assesses an infant’s GI-related signs and symptoms as observed by caregivers/parents over the previous week in 5 domains: stooling, spitting up/vomiting, flatulence, crying, and fussing.25 Caregivers/parents provide one response after each question is read to them by a trained interviewer, and items are scored on a scale of 1 to 5, with higher values indicating greater GI distress. Total IGSQ scores are calculated by summing item responses. The possible range in scores is 13 to 65, where a score of 13 indicates no GI distress and a score of 65 represents extreme GI distress. The IGSQ has been utilized successfully in clinical trials assessing infant feeding tolerance.26-28

Two-Day Diary

Tolerance measures were documented by parents/caregivers over the 2 days prior to V1, which was the final visit, occurring after approximately 6 weeks of consumption of the Test and Control formulas. Caregivers were asked to record the amount of study formula consumed by the infant, the number of stools passed in a 24-hour period, consistency of each stool (watery, loose, soft, formed, hard), whether the infant had difficulty passing the stool (Yes or No), frequency of vomiting, spitting up (none, occasional [1-5 times/day], frequent [>5 times/day]), and durations of crying and fussing (<10, 10-30, 31-60, 61-120, 121-180, or more than 180 minutes). Spitting up was defined for caregivers/parents as “effortless return of small amounts of formula after feeding, usually about a mouthful” and vomiting as “a forceful return of larger amounts of formula.”

Adverse Events

Adverse events (AEs) were collected throughout the study and were assessed by the site investigator or designee for duration, intensity, frequency, and relationship to test product. AEs were classified into System, Organ, and Class categories. Within the infections and infestations System, Organ, and Class category, 5 AE clusters were identified: upper respiratory tract infection, viral upper respiratory tract infection, otitis media/ pharyngitis, thrush, and “other.”

Sample Size

The primary objective of this clinical trial was to compare IGSQ scores after infants were fed the formula (Test) containing 2′FL or the formula without 2′FL (Control). The sample size was estimated for the primary endpoint based on an assumption of noninferiority with a limit of 4 points higher on the IGSQ for the Test versus the Control group, assuming an α level of .05 and 80% power. Sample size calculations were based on data from previous trials with similar study designs and endpoints. Based on these parameters, the total sample size calculated was 28 subjects per formula group. Eighty infants were recruited in order to account for a dropout rate of approximately 30%.

Statistical Methods

Demographics and birth characteristics were analyzed using descriptive statistics and compared between formula groups using t tests for continuous measures and χ2 tests for categorical measures. IGSQ scores at V0 and V1 were compared between groups using t tests. Mean stool frequency per day over the 2-day diary period was compared between groups using a repeated measures model to account for multiple stools per day per baby. Stool consistency in each category was examined using percentages based on all stools passed and compared between groups first using a repeated measures ordinal model with 5 stool consistency categories. Additionally, stool consistency was grouped into categories of Soft + Loose versus Watery + Formed + Hard and examined in a repeated measures model. Difficulty in passing stool was examined at the subject level with a between-group comparison using Fisher’s exact test due to small cell counts. Crying, fussing, and vomiting frequency were compared using a repeated measures model to account for multiple measurements over the 2-day diary period for each baby. Spitting up was assessed first as a Yes/No variable for each baby on each diary day and examined with a repeated measures model. Next, frequency of spit-up was evaluated at the level of the baby as either Yes (spit-up on one or both diary days) versus No (no spit-up on either diary day) with comparisons made between groups using a t2 test. Weights and weightfor-age z scores and percentiles, as well as length and length-for-age z scores and percentiles, were calculated according to the World Health Organization growth charts ( resources/sas-who.htm) and length with strata of gender at each study visit. AEs were counted within formula groups at the event level as well as at the subject level.
All statistical analyses were performed using Stata (StataCorp, 2017, Stata Statistical Software: Release 15, StataCorp LLC, College Station, TX) or SAS/STAT software, Version 9.4 of the SAS System for Windows (SAS Institute Inc, Cary, NC). Statistical significance was assessed using an α level of 5% with 2-sided statistical tests unless otherwise specified.
The Intent-To-Treat (ITT) population was defined as all randomized subjects who took any amount of the study formula, while the Per Protocol (PP) population was defined by excluding subjects if they were unlikely to have had full exposure to the study formulas as defined by these criteria: had a break in study formula feeding longer than 3 days, were hospitalized for more than 3 days, had more than 4 teaspoons of complementary foods per day, and/or had more than 3 ounces of juice per day. The primary outcome of IGSQ scores at V1 was analyzed in the ITT and PP populations. Secondary outcomes including safety endpoints were analyzed only in the ITT population.
Customized, secure electronic case report forms and database (Medidata Solutions, Inc, New York, NY) were used to collect and store data. Study staff entered data into the eCRF database, and data were quality checked by a data manager before database lock.


Seventy-nine subjects were randomized. One subject withdrew consent before receiving study product. Seventy-eight subjects were, therefore, included in the ITT analyses (38 Test, 40 Control). In the Test group, 1 subject was lost to follow-up, 1 caregiver wished to withdraw, 3 withdrew due to AEs, and 3 were noncompliant with feeding only study formula. In the Control group, 1 subject was lost to follow-up, 1 caregiver wished to withdraw, 3 withdrew due to AEs, and 2 were noncompliant. Therefore, 30 subjects from the Test group and 33 subjects from the Control group were included in the PP analysis (Figure 1). 


Overall, subjects were at a mean age of 14 (±3.3) days at time of enrollment. The demographics of the ITT population are shown in Table 2. There were no differences in gender, ethnicity, delivery type, gestational age at birth, birth anthropometrics, whether the subject had siblings, maternal age, or maternal education between groups in the ITT population. The Test group had significantly more infants who were reported as having ever been breastfed compared with the Control group (20 [53%] vs 9 [22%], P = .006). The mean number of days of breastfeeding prior to enrollment, however, were not significantly different between groups (Test: 4.7 [± 4.0] days vs Control 3.6 [±4.7] days, P = .51).

Primary Outcome: Infant Gastrointestinal Symptom Questionnaire

The primary outcome was comparison of IGSQ scores for the PP population between groups at V1, occurring after 6 weeks of feeding. IGSQ scores for the Test and Control group were similar at baseline for both ITT and PP analyses (ITT: 21.0 ± 7.1 vs 21.7 ± 8.2 in Test vs Control, P = .68, and PP: 20.1 ± 6.6 vs 20.7 ± 7.5 in Test vs Control, P = .77). At V1, mean IGSQ scores were similar between groups (ITT: 22.5 ± 6.4 vs 20.8 ± 4.5 in Test vs Control, P = .19, and PP: 20.9 ± 4.8 vs 20.7 ± 4.3 in Test vs Control, P = .82) and within the hypothesized 4-point limit of noninferiority for both ITT and PP analyses (Table 3).

Secondary Outcomes

Stool Frequency, Consistency, and Ease of Passing. Based on data recorded by caregivers in the 2-day diaries, stool frequency did not differ between groups; infants in the Test group passed 2.2 (±1.8) stools per day and infants 

in the Control group passed 2.3 (±2.1) stools per day in the 2-day period prior to the final visit (P = .79). Stool consistency, based on an ordinal model utilizing 5 stool consistency categories, did not differ significantly between groups (P = .65) in the ITT population. Stools of infants fed the Test formula were either loose or soft 77% of the time (Figure 2). More stools were reported to be difficult to pass in the Control group (4 [3%] Test and 33 [21%] Control, P = .04); however, the number of infants with stools