Human milk oligosaccharide supplementation in irritable bowel syndrome patients: A parallel, randomized, double‐blind, placebo‐controlled study

Human milk oligosaccharides safely and beneficially impact bifidobacteria abundance in healthy adults, while their effects in patients with irritable bowel syndrome (IBS) are unknown. Hence, we aimed to determine the dose of 4:1 mix of 2’‐O‐fucosyllactose and Lacto‐N‐neotetraose (2’FL/LNnT) that increases fecal bifidobacteria abundance without aggravating overall gastrointestinal symptoms in IBS patients in a randomized, double‐blind, controlled study. Additionally, the impact of 2’FL/LNnT on the fecal bacterial profile was assessed.


| INTRODUC TI ON
The pathophysiology of irritable bowel syndrome (IBS), a functional bowel disorder characterized by altered bowel habits and abdominal pain, remains ambiguous. 1 However, subgroups of IBS patients show an unbalanced intestinal microbiota profile 2 associated with severity of symptoms. 3,4 Moreover, a recent systematic review of gut microbiota in IBS described a decrease in the genus Bifidobacterium in some patients. 5 Potentially, the gut microbiota composition may be modulated by diet, prebiotics, probiotics, synbiotics, and even non-absorbable antibiotics. 6 To date, only a few intervention studies have explored the effect of prebiotics, non-digestible compounds that stimulate specific bacterial growth, in IBS patients, with diverse study outcomes.
Trans-galactooligosaccharides have been shown to promote bifidobacteria growth and alleviate gastrointestinal (GI) symptoms 7 and short-chain fructo-oligosaccharides improved digestive comfort. 8 In contrast, another study demonstrated that short-chain fructo-oligosaccharides had no therapeutic value for IBS patients, 9 and a recent meta-analysis of 11 randomized controlled studies concluded that prebiotics did not improve GI symptoms but increased bifidobacteria. 10 Human milk oligosaccharides (HMO) are unique, complex glycans found in high concentrations (5 -25 g/L HMO) in human breast milk. 11 They are selective substrates for specific intestinal bacteria, protect against infection by blocking pathogen attachment to epithelial cells, promote immunomodulatory activity, and improve gut barrier function. 12 The most abundant structural classes are fucosylated HMO (35%-50%) and neutral core HMO (42%-55%).
2'-O-Fucosyllactose (2'FL) and Lacto-N-neotetraose (LNnT), respectively, are major component of each class. 13,14 HMO reach the colon intact 15,16 and undergo fermentation by specific bacteria, such as bifidobacteria. 17,18 A 2:1 mix of 2'FL and LNnT has been demonstrated to influence the fecal microbiota composition, including an increase in bifidobacteria in infants. 19 Furthermore, supplementation with either 2'FL or LNnT alone or in a mix was safe and promoted fecal bifidobacteria in healthy adults. 20 However, there are no published studies of the effects of HMO in IBS patients.
We hypothesized that a 4:1 mix of 2'FL/LNnT will be well tolerated and able to beneficially modulate gut microbiota in IBS patients, in particular inducing an increase in Bifidobacterium spp. abundance.
Hence, we aimed to determine the dose of a 4:1 mix of 2'FL/LNnT which increases fecal bifidobacteria abundance, without negatively influencing GI symptoms. Additionally, we assessed the effects on other IBS-related symptoms and on the general fecal microbiota profile.

| Study design and population
This was a phase II, parallel, double-blind, randomized, placebo-controlled study in adult male and female IBS patients (n = 61 at randomization). The patients were randomized (1:1:1) into three groups, consuming either active product (two groups) or a placebo product (one group) for four weeks with a four-week follow-up period after the end of the intervention. The design of the study is graphically presented in Figure 1.
The study population comprised patients (18-75 years of age) with IBS symptoms of at least moderate severity (IBS Symptom Severity Scale (IBS-SSS) score ≥175) 21

Key Points
• Supplementation with a human milk oligosaccharides (HMO) mix of 2'-O-fucosyllactose and lacto-N-neotetraose (2'FL/LNnT) promotes fecal Bifidobacterium spp. and is well tolerated in healthy adults. However, the effects in patients with irritable bowel syndrome (IBS) have not been assessed.
• The 10g dose of 2'FL/LNnT influences overall fecal microbiota composition, specifically induces growth of bifidobacteria, without aggravating gastrointestinal symptoms in IBS patients after a 4-week intervention.
• HMO might be favorable in order to restore altered gut microbiota in IBS.

| Study intervention
Patients were randomized to receive one of the three different interventions for 4 weeks; placebo, 5 g or 10 g doses of 2'FL/LNnT  Patients taking the full dose for ≥24 days during the 28-day supplementation period were considered fully compliant. A physical examination was done by a study physician. Study subjects, investigators, and sponsor were all blinded to treatment allocation until the end of the study.
The patients received equipment for fecal sampling including cooling kits. Samples were collected and kept at −20°C freezers maximum 4 days prior to a visit (visits 2, 3, and 4). Fecal samples were stored at −80°C after each visit.

| Bifidobacteria and microbiota profiling
The fecal samples were analyzed using a commercially available genome-based microbiota test (GA-map™ Dysbiosis Test, Genetic Analysis AS). Briefly, fecal samples were homogenized prior to mechanical disruption of bacterial cells, total bacterial genomic DNA isolated with magnetic beads, and hypervariable regions V3-V9 amplified by 16S rRNA polymerase chain reaction (PCR), using fiftyfour probes, validated to be GI disorder-specific, targeting more than 300 bacterial strains. 26 Probe intensity signal correspondent to fecal bacterial abundance was detected and measured by BioCode 1000A 128-Plex Analyzer (Applied BioCode).

| Data analyses; primary, secondary, and exploratory endpoints
The primary endpoint was to determine the dose of 2'FL/LNnT that increases Bifidobacterium spp. abundance without aggravating GI symptoms, measured by GSRS-IBS. As secondary efficacy endpoints, IBS severity, measured by IBS-SSS, bowel habits (stool consistency), and anxiety and depression were assessed. Moreover, the effect on fecal microbiota and the proportion of responders, defined as a patient with a bifidobacteria abundance increase ≥50% at the end of the intervention period, relative to baseline were explored. All eligible and randomized patients were included in the analysis of the comparisons of clinical questionnaires (intention-to-treat analysis, ITT; n = 60).
Statistical analyses were chosen based on normality of distribution determined by boxplots. For the effect on the fecal microbiota profile, patients who completed the intervention were included (per-protocol analysis, PP; n = 58). All analyses were performed after unblinding.

| Statistical analyses
This study was explorative, and no formal sample size calculation was performed. Instead, it was based on the sample size of a previous study in healthy adults which showed microbiota modulation after supplementation with HMO. 20 For this study, the sample size was slightly increased to address the inherent background noise of microbiota results and the statistical analysis significance. Natural logarithmic transformation was used for the microbiota data to obtain a more homogeneous variance. All tests used an alpha of .05 as cutoff for significance and two-sided confidence intervals.

F I G U R E 1
Overall study design. Additional information about statistical methodology can be found in Appendix S3.
Demographic information at baseline was analyzed based on normality of distribution and compared between the groups.
Categorical variables were analyzed with chi-square test, while continuous variables were analyzed by using one-way ANOVA with Bonferroni's correction or Kruskal-Wallis test.

| Demographics and clinical characteristics
In total, 73 IBS patients were screened for eligibility between January 2017 and April 2018, and of these, 61 patients (41 women and 20 men) were randomized into one of the three intervention groups: showing the number of screened, randomized, and completed subjects placebo (n = 21), 5 g (n = 20), and 10 g (n = 20) 2'FL/LNnT. Fifty-nine out of 61 patients completed the study to the follow-up visit (week 8). There were no major changes in diet. Fifty-eight patients reported consumption of the study product for ≥24 of the 28 days (Table S1).
One subject did not fulfill the compliance criteria but took full dose for 23 days and was compliant to all other study procedures, and was therefore included in all the analysis. Two patients, one from the placebo group and one from the 10 g 2'FL/LNnT group, discontinued prematurely after 2 weeks of intervention due to increased IBS symptoms. Additionally, after completion of the intervention, one patient in the placebo group was diagnosed with Crohn's disease. This patient was excluded from the ITT and PP analyses (  Table 1).
The exploratory analysis (PP analysis) excluded the patients who discontinued the intervention (n = 58). Sex, age, body mass index, and IBS subtype classification did not differ between the three groups at baseline. However, patients in the 5 g 2'FL/LNnT group demonstrated a lower GSRS-IBS total score at baseline compared to the placebo and 10 g 2'FL/LNnT groups (P = .03) ( Table 1).

| Dose effect of 2'FL/LNnT on Bifidobacterium spp
The 10 g 2'FL/LNnT group presented with a higher abundance of fecal

| Dose effect of 2'FL/LNnT on IBSrelated symptoms
The overall mean of GSRS-IBS total score at baseline was 49.29 (9.42), implying that on average patients reported "mild to moderately severe discomfort" on the five main domains of GI symptoms. No differences in overall GI symptom severity (GSRS-IBS total score) between groups were identified at week 4 ( Figure 4A) or week 8 ( Figure 4B).  (Table 2).

Within
Furthermore, there were no differences between groups or within groups at week 4 or week 8 regarding IBS symptom severity (IBS-SSS).
However, the placebo and 5 g groups demonstrated a tendency toward milder symptoms (See Table S2, Figure S1). Regarding bowel habits, pla-   baseline (See Table S3). Lastly, anxiety and depression subscales did not show any differences between groups at baseline, week 4, or week 8 (See Table S4).

| Effect of 2'FL/LNnT on microbiota profile
According to multivariate factor discriminant analysis by OPLS-DA, the microbiota profile of patients in the placebo ( Figure 5A) and Error bars correspond to 95% confidence range. Asterisks identify statistically significant bacteria taxa *P < .05; **P < .01; ***P < .001 at week 4 as compared to baseline ( Figure 5D). Some bacterial taxa showed a significant change during the intervention within the placebo or 5 g group (See Table S5), although these changes were not reflected in modulation of the overall microbiota profile.

| Microbiota profile in 2'FL/LNnT responders and non-responders
In the full study cohort, 24 patients (41%) met the criteria for response, that is, ≥50% increase of bifidobacteria abundance, ). However, the 10 g 2'FL/LNnT group demonstrated good fitness and predictability, that is, good discrimination between responders and non-responders ( Figure 6C). The bacterial taxa most important for differentiating responders from non-responders in the 10 g 2'FL/ LNnT group were high abundance of Actinobacteria, Eubacterium halli, Eubacterium biforme, Lactobacillus spp. A, and Coprobacillus cateniformis ( Figure 6D). However, the microbiota profile before the start of the intervention could not predict the treatment response in any of the groups (See Table S6).

| D ISCUSS I ON
In this study, daily intake of 10 g 2'FL/LNnT increased abundance of fecal Bifidobacterium spp. without negatively influencing GI symptoms in patients with IBS. Further, the product was well tolerated as no worsening of IBS symptoms, bowel habits, anxiety, or depression were detected. Moreover, the dose of 10 g 2'FL/LNnT influenced overall fecal microbiota composition, and responders, defined by bifidobacteria increase ≥50%, could be discriminated from non-responders based on the microbiota composition.
Several factors are considered to influence the pathophysiology of IBS, and growing evidence shows that the intestinal microbiota plays a key role. A well-balanced intestinal microbiota, comprising beneficial members such as bifidobacteria, is crucial to ensure the production of important compounds that improve intestinal health. HMO have been shown to increase bifidobacteria abundance in infants and healthy adults, but their ability to alleviate symptoms in IBS patients is unknown. The study was designed to determine the dose of a 4:1 mix of 2'FL/ LNnT that could increase fecal bifidobacteria abundance, without aggravating GI symptoms. The ratio used aimed to reflect the proportion of these two oligosaccharides in human breast milk. 14 The study was not designed to detect improvement and included only lower doses than 20 g, 20 as IBS patients are considered more sensitive than healthy subjects. Similar to healthy adults, 20 four-week intake of 5 and 10 g 2'FL/LNnT in adult IBS patients was well tolerated and did not deteriorate GI symptoms, as assessed by GSRS-IBS total score and individual items of GSRS-IBS. Surprisingly, the placebo group showed a modest tendency to improve GI symptoms. Importantly, the intervention did not increase gas production, a potential risk previously raised due to the non-absorbable properties of other prebiotics. 27,28 Only two patients, one from the placebo group and one from the 10 g 2'FL/LNnT group, discontinued the study due to worsening of GI symptoms. Thus, the low dropout rate together with lack of GI symptom deterioration in IBS patients completing the intervention suggests that the product was well tolerated, as the study was expected to demonstrate.

2'FL and LNnT have been identified as substrates for
Bifidobacterium spp., in both formula-fed infants and healthy adults. 19,20 Higher levels of bifidobacteria in elderly seem to correlate with healthy status and longevity, 29 whereas reduction in this genus has been associated with several conditions, including IBS, inflammatory bowel diseases, and metabolic disorders. 5,29,30 This study showed, similar to a previous study in a healthy cohort, 20 that a daily intake for four weeks of 10 g, but not 5 g, 2'FL/LNnT increased fecal bifidobacteria abundance in IBS patients. The lack of persistent effects during the washout period may indicate the necessity of continuously supplementation for maintaining increased bifidobacteria abundance.
Consumption of prebiotics induces growth of specific bacteria, but has also been shown to modulate the overall gut microbiota composition. 7,19,[31][32][33] In line with this, the results from the current trial suggest that consumption of 10 g 2'FL/LNnT not only increased Bifidobacterium spp. abundance but also modulated the overall gut microbiota profile. In the 10 g 2'FL/LNnT group, increased abundance of Actinobacteria (class that includes Bifidobacterium spp.), Alistipes, Bacteroides spp., and Prevotella spp. was seen after four weeks. Interestingly, these bacterial taxa have previously been demonstrated to be in lower abundance in IBS patients relative to healthy individuals, 2 so restoring the abundance of these bacteria may be associated with improved health. However, the study design did not allow for investigating whether these specific bacteria were directly metabolizing 2'FL/LNnT or influenced by cross-feeding. In parallel, despite significant modulation of some bacterial taxa within the placebo and the 5 g 2'FL/LNnT group was observed, these populations might not have been relevant enough to impact the overall gut microbiota profile.
As part of the secondary endpoints, microbiota modulation in responders and non-responders was compared. According to previous experience from determining fecal microbiota composition over time in ulcerative colitis patients in remission (manuscript in preparation), the normal individual monthly variation of Bifidobacterium spp. abundance is <30%. We therefore chose to define response to the intervention as ≥50% increase (after the intervention period) of relative abundance of Bifidobacterium spp. The overall microbiota composition was differently modulated by the HMO mix in responders and non-responders, and the difference was most evident in the 10 g 2'FL/LNnT group. The modulation of microbiota profiles might reflect higher abundance of HMO utilizing bifidobacteria in responders compared to non-responders at the start of the interven- as well as Eubacterium spp. and Lactobacillus spp., as seen among responders in this study, is often associated with saccharolytic processes and improved gut health. [38][39][40] Hence, HMO could potentially be used for IBS patients with unbalanced gut microbiota, to re-establish a healthy microbiota profile.
To our knowledge, this is the first study aiming to determine the daily dose of 2'FL/LNnT that increases abundance of Bifidobacterium spp. without negatively influencing GI symptoms in IBS patients.
Even with its strengths and promising results, there are limitations with the study. Being a truly exploratory study, the size of the study population was small and did not allow for subgroup analyses regarding IBS subtypes, which likely also influenced the fit and predictive ability of microbiota profile analysis models. Moreover, the study did not intend to assess a clinical effect of the intervention but to evaluate the tolerability of the study product. The inclusion of patients of all IBS subtypes, potentially with different disease driving mechanisms, may also have influenced the study outcome. However, since this is the first study exploring effects of the 2'FL/LNnT on IBS patients, we did not want to exclude any IBS subtype. Further, the relatively short intervention period did not allow to study the long-term effect, and this should be considered for future studies, since longer treatment periods may lead to an effect on both bifidobacteria and clinically relevant endpoints. Moreover, the use of GA-map ™ technology might have excluded some relevant taxa that could be detected by other techniques, as 16S rRNA gene sequencing. However, the commercially available test used in the study simplified the establishment of gut bacterial profiles and may also be applied in a clinical setting.
In conclusion, four-week daily intake of 10 g 2'FL/LNnT increased the abundance of Bifidobacterium spp. without aggravating GI symptoms in IBS patients. This dose was well tolerated and did not induce worsening of IBS symptoms, bowel habits, anxiety, or depression, and most patients completed the four weeks of intervention without significant side effects. Moreover, the 10 g 2'FL/ LNnT modulated overall fecal microbiota composition, and responders, defined by bifidobacteria increase ≥50%, could be discriminated from non-responders based on fecal microbiota modulation. Thus, this intervention might be favorable in order to restore gut microbiota of IBS patients toward a healthier profile.