2.1. Study Design and Patients
This was a double-blind, randomized, placebo-controlled clinical trial (study registered at ClinicalTrials.gov
: NCT01869426), performed to assess the efficacy and safety of a lyophilized high-concentration multi-strain probiotic mixture in reducing colic symptoms in exclusively breastfed infants.
The study was carried out between April 2015 and April 2016 in the Pediatric Outpatient Service of the Neonatology and Neonatal Intensive Care Unit (NICU) section of the Department of Biomedical Science and Human Oncology, of “Aldo Moro” University of Bari, Italy.
To be eligible for the study protocol, infants had to be healthy and well-fed, born at term, aged between 30 and 90 days, exclusively breastfed before and throughout all the study period, with a diagnosis of infantile colic (defined as crying episodes lasting three or more hours per day and occurring at least three days per week [16
], within seven days prior to enrolment).
Exclusion criteria were: (a) major acute or chronic diseases; (b) gastrointestinal diseases and gastro-esophageal reflux; (c) use of probiotics/antibiotics the week before or during randomization; (d) gastrointestinal malformations; and (e) concurrent enrolment in other clinical trials.
Written informed consent was obtained from a parent/guardian, in accordance with the local ethics committee, which reviewed and approved the study protocol.
All infants enroled were randomized to receive, either oral probiotics or the placebo. Randomization was performed using a computer-generated two-treatment allocation sequence (nQuery Advisor v.7.0 software, Statistical Solutions Ltd., Cork, Ireland). To avoid disproportionate numbers of patients in each group, a randomization scheme was performed in blocks of four participants. Randomization codes were secured until all data were analyzed.
All participants, as well as scientific and medical personnel dedicated to the study, were blind to the group assignment. An independent person prepared the randomization schedule and took care of the packaging and labelling of the products.
Parents were requested to administer to their infants 10 drops of the probiotics, once a day, for 21 days directly in the mouth, preferably in the morning before feeding. Instructions for keeping and maintaining the product were also provided according to manufacturer indications. Parental compliance was assessed by a diary. The parents were also asked to return used bottles.
Even though maternal diet can influence colic frequency, no specific dietary restrictions during lactation were recommended, except to avoid any commercial products containing probiotics [17
]. Parents were also instructed to avoid any other methods of managing infant colic.
The active study product was a medium chain triglycerides oil suspension containing a lyophilized high-concentration multi-strain probiotic mixture of four different strains of lactobacilli (L. paracasei DSM 24733, L. plantarum DSM 24730, L. acidophilus DSM 24735, and L. delbrueckii subsp. bulgaricus DSM 24734), three strains of bifidobacteria (B. longum DSM 24736, B. breve DSM 24732, and B. infantis DSM 24737), and one strain of Streptococcus thermophilus DSM 24731, produced at Danisco-Dupont, WI, USA and currently sold in Continental Europe, USA and Korea under the brand Vivomixx®, Visbiome® and DeSimone Formulation respectively.
Ten drops of the formulation delivered a dose of 5 billion colony-forming units (CFU) of L. paracasei DSM 24733, L. plantarum DSM 24730, L. acidophilus DSM 24735, L. delbrueckii subsp. bulgaricus DSM 24734, B. longum DSM 24736, B. breve DSM 24732, B. infantis DSM 24737, and finally of Streptococcus thermophilus DSM 24731. Maltodextrine was used as an excipient. The medium chain triglycerides oil suspension complied with baby food requirements as per regulation 1881/2006 and directives 2006/141 and 2006/125.
The placebo was characterized by an identical formulation, but without any probiotic. The active and placebo products were identical and supplied in the same bottle, in order to maintain a double-blind status. All study products were kept refrigerated until use. The randomization codes were secured during each phase of the trial and opened only after final data analysis.
Both the active and placebo study products were donated by Prof De Simone/CD Investments who had no role in any phase of the study.
According to the study protocol, when an infant was referred with a diagnosis of infantile colic and fulfilled all criteria for enrolment, parents were asked to record colic symptoms for one week. If symptoms fulfilled the “infant colic” definition during that period, infants were enrolled in the study protocol.
2.2. Analysis of Infant Colic Symptoms
A structured diary to record gastrointestinal events, such as a feeding schedule, daily episodes of fussing/crying, the number of minutes of inconsolable crying per day, the number of bowel movements per day, and stool consistency (according to the Bristol Stool Form Scale for children) [18
] was given to parents. The diaries used in this study were designed to be easy to use and to be suitable for a diverse sample of the population. “Fussing” was referred to elements of movement, facial expression, and voice that are usually interpreted as expressing a negative emotion.
Parents also recorded the times of administration of the study products, family quality of life, both at enrolment and at the end of the study, as well as any adverse events, such as constipation, vomiting, and skin reactions.
Quality of life was assessed by a 10-cm visual analogue scale with a numerical rating scale from 0 (worst possible well-being) to 10 (perfect well-being), as already described previously [19
]. Parents were encouraged to contact the study personnel whenever needed.
On enrolment (day 0), a pediatric medical examination was performed and the following information was collected: (1) gestational age; (2) type of delivery; (3) birth weight; (4) anthropometric data at entry; (5) family history of gastrointestinal disease; and (6) family history of atopy.
Follow-up visits were scheduled at 7, 14, and 21 days after the beginning of the administration of study products, performed by the same referring pediatrician.
At the end of the study all the diaries were collected. Diary analysis and data entry were performed independently by two experienced investigators, both blinded from treatment allocation.
2.3. Analysis of Infant Stool Samples
To evaluate if probiotic supplementation showed a real impact on intestinal microbiota composition, two stool samples of infants were collected, once at enrolment and then 21 days after the beginning of the administration of study products. All samples were collected from diapers in sterile plastic tubes and stored at −80 °C until analysis. qPCR was used to quantify bifidobacteria and lactobacilli using genus-specific primers and conditions, as already described previously [20
Each sample of feces was also prepared for proton nuclear magnetic resonance (1H-NMR) by mixing 80 mg of stool for 5 min with 1 mL of bi-distilled water and centrifuged at 4 °C and 18,630 g for 15 min. The supernatant (700 μL) was added to 100 μL of a D2O solution of 3-(trimethylsilyl)-propionic-2,2,3,3-d4 acid sodium salt (TSP) 10 mM, used as a reference for NMR chemical-shift, buffered at pH 7.00 by means of 1M phosphate buffer.
H-NMR spectrum was recorded at 298 K with an AVANCE III spectrometer (Bruker, Milan, Italy) operating at a frequency of 600.13 MHz. According to Ventrella et al. [21
], a CPMG-filter composed of 400 180° pulses of 24 μs each, separated by 400 μs intervals, was employed to suppress the signals from broad resonances. Presaturation was applied to suppress the HOD residual signal. To acquire each spectrum, 256 transients were summed up, measuring 32 K data points spanning 7184 Hz of spectral window, with a 2.28 s acquisition time.
To obtain quantitative values by NMR, 5 s of recycle were chosen, in agreement with the relaxation time of the protons under investigation [22
H-NMR spectra were baseline-adjusted by means of peak detection according to the “rolling ball” principle implemented in the baseline R package [23
Each spectrum was then linearly corrected, so to make the baseline points randomly spread around zero. Signals were not manually aligned, which was different from previous investigations [25
Probabilistic quotient normalization [26
] was applied to the entire array of spectra to remove the effects of differences in water and fibers content among samples.
The signals were assigned by comparing their chemical shift and multiplicity with the Human Metabolome Database [27
] and Chenomx software library (Chenomx Inc., Edmonton, AB, Canada, ver 8.1). Molecules with unknown chemical structures were also detected and quantified. Following Ventrella et al. [21
], these molecules will be referred to throughout the text with an x followed by the chemical shift in ppm of their signal used for quantification (i.e., x–4.20).
2.4. Statistical Analysis
The primary outcome measures was the rate of responders and non-responders in improvement of colic symptoms. Success rate was defined by a reduction of the daily average crying time ≥50%, expressed in minutes.
Secondary outcome measures were: (1) average crying time per day; (2) parental quality of life; (3) any other gastrointestinal events; (4) anthropometrical evaluations; (5) side effects; (6) amount of lactobacilli and bifidobacteria in the stool samples of infants; (7) metabolomics evaluation of feces by means of 1H-NMR.
A reduction of the daily average crying time ≥50% was used to calculate the sample size that was 26 patients per group. For α = 0.05, β = 0.10, and an estimated SD within groups of 55%, 26 patients were needed per group [19
]. Considering a potential dropout rate of 20%, thirty-three subjects per group were enroled.
Data from the first visit forms, diaries, and the results of the analysis of stool samples were reported in a database created by Google Drive software (Google ILC, Mountain View, CA, USA).
Statistical analyses were performed using Stata12MP (StataCorp LLC, College Station, TX, USA) and a per-protocol approach.
Quantitative variables with normal distribution were compared using the Student’s t-test. The Mann-Whitney U Test was used for non-normally distributed variables. Proportions were compared using the χ2 test or Fisher Exact Test, as appropriate.
Linear regression and logistic regression were performed to determine the effects of probiotic supplementation, vaginal delivery, family history of atopy and family history of gastrointestinal disease (determinants) on the duration of crying at 21 days, the total average crying minutes and the treatment response at day 14 and at day 21 (outcomes). In the logistic regression, the adjusted odds ratio, with 95% CI, were calculated and a z-score test was performed. Coefficiency with 95% CI and t values were calculated for linear regression. For all tests, a p value of <0.05 was considered as significant.
In the fecal metabolomics evaluation, molecules whose concentration varied in relation to the investigated treatments were looked for by calculating the T21–T0 differences and then they were compared by means of Mann-Whitney U test.
To highlight the underlying trends characterizing the samples, a robust Principal Component Analysis (rPCA) was built on the molecules concentration, centered and scaled to unity variance, according to Hubert [28
For each PCA model, we calculated the scoreplot and the projection of the samples in the PC space, tailored to highlight the underlying structure of the data. We calculated the correlation plot, relating the concentration of each variable to the components of the PCA model, therefore, tailoring the samples to highlight the most important molecules in determining the trends emphasized by the scoreplot. The same subjects were followed along the entire experiment, giving rise to a repeated measurement data structure. Molecule concentrations at timepoint T0
were subtracted from every timepoint, as suggested by Ndagijimana [29