Study population
Between January 2015 and April 2016, healthy women who were pregnant or who had recently delivered healthy term infants and lived within Yolo and Sacramento counties in California were recruited and subsequently provided written informed consent to enroll in the study. Enrollment criteria for study participation were based on limiting the number of confounding variables that could influence the infant gut microbiome. Inclusion and exclusion criteria for mothers were as follows: women 21 to 45 years of age, in their third trimester of pregnancy or had delivered an infant within the past 4 days, planned to exclusively breastfeed their infants for at least the first 3 months postnatal, lived in a developed nation for the past 10 years, did not plan to administer probiotic supplements to their infants during the study duration unless they were allocated to the B. infantis group, who had not been diagnosed with any chronic metabolic disease or obesity, and who were non-smoking. Medications during labor, including antibiotics, were recorded but not used as an exclusion criterion. Inclusion and exclusion criteria for infants born to qualified mothers included gestational age at birth ≥37 weeks, birth without medical complications (such as respiratory distress syndrome, birth defects, and infection), no exposure to any oral or intravenous antibiotics 72 h postnatal, and no consumption of infant formula 24 h prior to the Day 7 postnatal at-home lactation consultation visit.
Study design
This IMPRINT study was a parallel, partially-randomized, controlled 2-month trial. The University of California Davis Institutional Review Board approved all aspects of the study (IRB #: 631,099). This trial was registered on
ClinicalTrials.gov Identifier: NCT02457338. Prior to the initiation of the study, three separate randomization schemes were generated using a random number generator in Excel. Participants were stratified to one of the three randomization schemes based on mode of delivery—vaginal delivery, cesarean section (time of membrane rupture before delivery ≤6 h), or cesarean section (time of membrane rupture before delivery >6 h). Stratified randomization was utilized because mode of delivery, as well as the time when membranes rupture before cesarean section delivery, have been shown to influence early infant intestinal microbial colonization [
30]. Randomization to lactation support alone (LS) or lactation support plus
B. infantis supplementation (BiLS) was in a 1:1 ratio for all three randomization schemes in blocks of ten. After receiving informed consent for the infant, the clinical coordinator assigned enrollment identification numbers according to the schedule. The first fifteen participants enrolled in the study were not included in the stratified randomization schedule due to unavailability of the
B. infantis product during the trial period. The first eight infants enrolled in the study were assigned to the LS group, but three had withdrawn or were screen-failed and thus only five infants received the intervention. The subsequent seven infants were assigned to the BiLS group, but three had withdrawn or were screen-failed and thus only four infants received the intervention. Parity and mode of delivery were not different between the two assigned groups for these fifteen participants.
After meeting major postpartum study criteria at enrollment (Day 3 or 4), infants were randomized into the BiLS or LS group. On Day 7, infants were screened for the consumption of infant formula within the past 24 h. On Days 3 or 4, 7, 15, 22, 33, and 61, study personnel visited mothers’ homes to conduct study procedures. On all six visits, mothers filled out questionnaires about their and their infants’ health, GI symptoms, occurrence of fever, illness, and number and reasons for sick doctor visits. Mothers collected infant stool samples from their infants’ diapers before Day 6 (baseline) and on Days 10, 14, 21, 25, 29, 32, 40, 50, and 60 and stored them in their kitchen freezers. Infant weight was measured by study personnel with a digital infant scale (Tanita) on Days 33 and 61. Participants received breastfeeding support at their homes by the study’s internationally board certified lactation consultant (IBCLC) prenatally and on Days 3 or 4, 7, and 15. On Days 22, 33, and 61 postnatal, study personnel transported samples from participants’ homes to the UC Davis campus on dry ice and stored at −80 °C.
Infants randomized into the BiLS group received one daily serving of B. infantis in their homes for 21 consecutive days starting on Day 7 and continuing through Day 27. During the Day 7 lactation consultation visit, mothers were trained by their lactation consultant to mix each B. infantis serving with 5 mL of their breast milk in a plastic medicine cup, and to syringe or finger-feed the mixture to their infants. Each daily serving of B. infantis EVC001 (ATCC accession Number SD-7035) consisted of one 625-mg sachet, delivering a minimum 156 mg of live bacteria (minimum 1.8 × 1010 CFU) plus 469 mg of lactose as the excipient. The 18 billion CFU per sachet was the minimum guaranteed CFU count as determined by the product specification. Because this is a live microorganism there is potential loss over time. As such, the sachets were produced with a 50% overage to account for potential losses during packaging and long term storage. This means the range of the dose delivered was 18–28 billion CFU per dose. The product was stored in the freezer at −20 °C and suffered no loss from the first infant enrolled to the last infant enrolled. Mothers received 21 sachets, plus four extra sachets that were to be used in the event of damage or misplacement. All sachets were kept frozen in mothers’ kitchen freezers until time of use, and mothers were instructed to keep all used and unused sachets provided. Compliance was assessed on Days 22 and 33 by counting and recording the number of empty B. infantis sachets.
Infant stool samples without labeled group assignments were provided to Evolve BioSystems, Inc. (Davis, CA USA) for the analysis of total infant fecal
Bifidobacterium. Group assignments were unblinded to Evolve BioSystems, Inc. post microbial analysis. For DNA extraction, approximately 100 mg of the frozen collected stool samples were extracted using the Zymo Research Fecal DNA kit, according to the manufacturer’s instructions. Polymerase chain reaction (PCR) amplification was conducted using methods as previously described, with minor modifications [
31]. Briefly, 5 μL of extracted DNA were used as template for a 20-μL reaction, using primers Bif F (5′-GCGTGCTTAACACATGCAAGTC-3′), Bif R (5′- CACCCGTTTCCAGGAGCTATT-3′), and Bif P (5′-TCACGCATTACTCACCCGTTCGCC-3′). Reactions were carried out with Taqman Universal MasterMix II with Uracil-N glycosylase (Life Technologies) [
31] using a Life Technologies QuantStudio 3 Real-Time PCR machine. Samples were assayed in duplicate. A standard curve was prepared from
Bifidobacterium longum subsp.
infantis EVC001 using the same extraction protocol as used for the stool samples for quantification of fecal
Bifidobacterium.
Infant gastrointestinal health and tolerability
Infant GI tolerability during
B. infantis EVC001 supplementation was assessed by mothers on a daily basis starting with Day 1 (or retrospectively if mothers were enrolled on Day 4 postnatal) until Day 61. On each day, mothers recorded the following information about their infants in daily logs: consumption of breast milk defined as suckling at the breast for at least five minutes or consuming any volume in a bottle; intake of other liquids or solids; amount of infant formula consumed; intake of probiotics that were not used in the study; intake of any oral antibiotics or administration of intravenous antibiotics; intake of any over-the-counter or prescribed medications; intake of vitamins, supplements, and herbs; number of spit-ups—less than five, five to ten, or more than ten; number of stools; consistency of stools using a modified Amsterdam infant stool scale—watery, soft, formed, hard [
32] (Additional file
1); blood in the stool; body temperature above 100.3 °F; ratings of GI-related symptoms using a continuous scale of 0 (“not noticeable”) to 10 (“most severe”), including general irritability (“how irritable was your baby?”), upset (“if your baby vomited or spit up, how upset was he/she after?”), and discomfort (“rate your baby’s discomfort in passing stool or gas”). Mothers also rated the frequency of their infant’s flatulence as “never, sometimes, often, very often” on a daily basis. Mothers filled out questionnaires to report any adverse events experienced by their infants during each at-home visit (Days 7, 15, 22, 33 and 61). Mothers recorded the following about their infants on weekly questionnaires: episodes of colic—defined as crying for more than 3 h per day for at least 3 days per week [
33], eczema diagnosis by a primary-care provider, number of sick doctor visits, illnesses, and medications used (Additional file
2).
Statistics
Data from the daily logs and retrospective questionnaires were binned into three time periods: baseline (Days 1–6), intervention (Days 7–27), and post-intervention (Days 28–61). For retrospective questionnaires, Day 7 data were binned as baseline. Means and proportions were calculated for continuous variables and categorical variables across all three time periods. Proportions for binary categorical variables were calculated as number of days reported/total number of days in each study period, and number of infants/total number of infants in each intervention group. The calculated values were multiplied by 100 to generate percentages.
For this Phase I study, the sample size was based on differences in infant fecal
B. infantis, which was calculated using the means and standard deviations from a previous study on breastfed infants [
22]. To detect a standardized inter-group difference of 1.3 z-scores in infant fecal
B. infantis with 90% power and α = 0.05, assuming equal standard deviations with a 20% attrition rate, 30 infants were needed in each group. Intent-to-treat analysis was performed of mother-infant dyads who initiated the study by Day 7 when final screening criteria were met. Statistical analyses were performed in IBM SPSS Statistics version 24 and figures were generated in PRISM v.7. Statistical significance was considered as
p < 0.05. Continuous data were checked visually for normality with histograms and quintile-quintile plots as well as numerically with the Shapiro-Wilk test and equality of variances using Levene’s statistic. Non-normal data were Log
10 transformed and confirmed again for normality prior to conducting parametric analyses.
To determine differences in total infant fecal Bifidobacterium between groups, Mann-Whitney U test was conducted using GraphPad Prism v7. Baseline demographics, maternal health, pregnancy history, and infant feeding and GI symptoms were compared between the LS and BiLS groups using the Pearson Chi-square Test for Independence (categorical variables), Mann-Whitney U Test, or one-way ANOVA (continuous variables). For normally-distributed continuous data, repeated measures ANOVA was performed with group and time as fixed factors, parity as the covariate, and group by time as the interaction term. If time was significant, multiple comparison post-hoc analysis with Bonferroni correction was carried out to compare baseline, intervention, and post-intervention data. Group differences in stool consistency, flatulence, and spitting-up were analyzed by logistic regression.