Impact of a probiotic product on bowel habits and microbial profile in participants with functional constipation: A randomized controlled trial

Objective To investigate the clinical efficacy of a multi‐strain probiotic product on bowel habits and microbial profile in participants with functional constipation. Methods This was a randomized, double‐blind, placebo‐controlled and parallel‐arm study. Altogether 94 otherwise healthy adults aged 18 to 65 years with symptoms of functional constipation were randomized as part of the intention‐to‐treat population. The participants received a placebo or the probiotic product (1.5 × 1010 CFU/day), consisting of Lactobacillus acidophilus DDS‐1, Bifidobacterium animalis subsp. lactis UABla‐12, Bifidobacterium longum UABl‐14 and Bifidobacterium bifidum UABb‐10 over 4 weeks. Outcomes included the patient assessment of constipation‐symptom (PAC‐SYM) questionnaire, stool frequency and consistency, and microbial profile. Results There were no significant between‐group differences in the PAC‐SYM score, despite significant within‐group differences (P < 0.001) over the study period. The probiotic group showed a faster normalization of stool frequency and consistency, with most participants achieving a normalized profile after 1 week. Fecal samples of the probiotic group exhibited higher relative abundance of Ruminococcaceae (P = 0.0047), including the Ruminococcus genus, and lower relative abundance of Erysipelotrichaceae (P = 0.0172) at end‐point compared with baseline. Placebo group samples showed similar abundance profiles over the study, with the exception of Clostridiaceae, which was lower at the study end‐point (P = 0.0033). Among treated participants, all four probiotic strains were significantly more abundant after the intervention. Conclusions No significant differences were observed in symptomology, with both groups showing a more than 20% improvement. However, the probiotic helped modulate bowel function earlier than the placebo, with a corresponding shift to a more fibrolytic microbiota.


| INTRODUCTION
Functional gastrointestinal disorders, including functional constipation, are among the most frequently observed conditions in clinical practice. 1 Functional constipation is symptom-based, non-organic in origin and commonly diagnosed by Rome IV diagnostic criteria. 2 The disorder has no known structural abnormalities, infectious or metabolic causes. 3 It has an estimated prevalence of 14% (95% CI 12-17%) in adults, 4 representing a significant health care burden. Constipation and digestive symptoms associated with discomfort result in absenteeism from work and lost productivity as well as reduced quality of life and increased medical costs. 5 Furthermore, patients' perception of their symptoms may be amplified by physiological, intrapsychic, and sociocultural factors that influence their daily life activities. 1,5 The etiology and pathophysiology of functional constipation are most likely multifactorial. Previous studies have found chronic constipation to be related to dysbiosis [6][7][8][9] and such alterations have been suggested as a possible pathophysiologic mechanism. 10 Prolonged gastrointestinal transit time may also lead to dysbiosis, which can in turn affect motility as well as immune and barrier function. 11,12 Nevertheless, research into the relationship between constipation and a dysbiotic microbiome is in its infancy.
The clinical management of functional constipation remains challenging. Current management options include diet and lifestyle changes, as well as the use of bulking agents, stool softeners, osmotic and stimulant laxatives, and prescription drugs. 3,13 However, many such products have limitations due to a lack of efficacy, inconsistent symptom response or safety concerns. 14 Lately, probiotics have been adopted as an adjunct approach to normalize intestinal transit time and alleviate symptoms. 15 Probiotics are defined as "live microorganisms that, when administered in adequate amounts, confer a health benefit to the host". 16 The safety profile of probiotics and fermented foods has been well documented in the general population; however, recent safety determination guidance for strain assessments 17 and commentaries for probiotic use in at-risk populations 18 have been published. A recent systematic review and meta-analysis concluded that probiotics might improve whole gut transit time and stool frequency profile in adults with functional constipation, albeit with high heterogeneity across studies. 15 In a separate meta-analysis, probiotic supplementation was found to decrease intestinal transit time, with statistically superior effects observed in adults with constipation or of advanced age. 19 However, the ability of probiotics to improve intestinal transit times is generally believed to be strain-specific 20 and improvements observed in transit time have not necessarily been associated with improved symptomology. As such, more studies are warranted examining the utility of probiotics in relieving functional constipation. 20 Also, despite evidence of an association of functional constipation with dysbiosis, few randomized controlled trials with probiotics have looked at microbial profiling and symptom outcomes simultaneously.
A probiotic blend, consisting of Lactobacillus acidophilus DDS-1, Bifidobacterium animalis subsp. lactis UABla-12, B. longum UABl-14 and B. bifidum UABb-10, was previously shown, in a pilot clinical trial, to improve symptomology in participants with irritable bowel syndrome. 21 The current randomized controlled trial aimed to evaluate the clinical efficacy of this multi-strain probiotic on bowel habits, symptomology, microbial profiling and strain recovery in otherwise healthy participants with functional constipation.

| Study design
This was a prospective, randomized, placebo-controlled, double-blind and parallel-arm study. Investigational visits took place at screening and at baseline (d 0), mid-point (d 15 ± 3) and end-point (d 29 ± 3) of the 4-week intervention period. At screening, the participants' complete medical history, concomitant therapies and inclusion criteria were reviewed, their characteristics and vital sign measures were recorded and fasted blood was collected for hematological and biochemical assessments.
Participants meeting the screening criteria entered a 2-week runin period, which involved the completion of a daily bowel habit diary, 3-day food records and a physical activity questionnaire. Eligibility was confirmed at baseline, and included a review of their inclusion criteria, a physical examination, vital sign assessments, weight and body mass index (BMI) measurements, completion of patient assessment of constipation -symptom (PAC-SYM) and patient assessment of constipation -quality of life (PAC-QoL) questionnaires, and, if applicable, a urine pregnancy test. The study diaries, including the records of daily bowel habits, food and physical activity questionnaires were collected. Eligible participants were randomized to receive placebo or probiotic capsules (1:1). All study personnel were blinded to the product. The randomization list was computer-generated with a block size of 4. The block size was not disclosed to the investigators and allocation was blinded to the participants as well as the site staff. Enrolled participants were instructed to take one capsule per day, before or during a meal, beginning at the day after randomization (d 1).
Throughout the intervention period, patients recorded their daily bowel habits, 3-day food records, physical activity and compliance.
Participants returned for investigational visits after weeks 2 and 4 of the intervention period, which included the completion of PAC-SYM and PAC-QoL questionnaires and the collection of study diary information. Fecal sample collection was performed prior to baseline and the end of study visits for microbial profiling analysis.

| Outcome assessments
The primary outcome was the change in the PAC-SYM score after Food records were assessed for their total caloric and macronutrient intake. Compliance was recorded by examining unused study product and confirmed via participants' diary.

| Microbial profiling
Participants collected fecal samples, as close as possible but prior to week 0 and 4 visits, inside a specimen collection hat, which ensured no contact with toilet water or urine. Samples were maintained at The paired-end reads were merged using USEARCH, 24 and the resulting sequences were compared to an in-house strain database and Greengenes (version 13.5). 25 All sequences with an identity ≥99% to a unique strain were assigned a strain operational taxonomic unit (OTU). Sequences were mapped via USEARCH against the OTU representative sequences to calculate strain abundances. The remaining (non-strain) unique sequences were quality filtered and clustered at 97% by UPARSE, and representative consensus sequences per de novo OTU were determined. Each representative OTU sequence was assigned a taxonomic classification using mothur's Bayesian classifier, which was trained against the Greengenes reference database of 16S rRNA gene sequences clustered at 99%.
Alpha diversity (ie, within-sample diversity) and beta diversity (ie, sample-to-sample dissimilarity) metrics were calculated. For the latter, Bray-Curtis dissimilarity was used to calculate the abundanceweighted sample pair-wise differences. 26 Significant differences among discrete continuous or categorical variables was assessed using permutational analysis of variance. 27 Metagenomic inference of 16S rRNA sequenced samples was applied as described previously. 28

| Strain recovery and identification
Each quantitative PCR (qPCR) target region was amplified individually from genomic DNA with primer sequences specific to the L. acidophilus, B. animalis subsp. lactis, B. longum, and B. bifidum strains (Table S1). The PCR cycling conditions were 1 cycle of 2 minutes at 95 C, 25 cycles of 1 minute at 95 C, 1 minute at 53 C, 15 seconds at 72 C, followed by 10 minutes at 72 C. PCR products for each strain were purified by the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) and cloned into a plasmid to generate copy number standards that ranged from 20 to 2 000 000 copies for absolute quantifi-

| Statistical analysis
The primary efficacy outcome on which the sample size calculation was based was the change in the PAC-SYM score between groups from week 0 to week 4 of the intervention period. Assuming alpha equaled to 5% and a power of 80%, the sample size was estimated with a difference in PAC-SYM of 4.08 units between groups, and a standard deviation (SD) of 6.36 units. The above assumptions were based on prior studies with probiotic 29 and plant extract 30 interventions that used the PAC-SYM score to assess symptoms. Accounting for premature withdrawal, 100 participants were forecasted for randomization, with 50 participants in each group.
The primary and secondary objectives were assessed on the intention-to-treat (ITT) population. All the statistical analyses were performed using the R statistical package version 3.2.3 (R Core Team, 2015). Significance tests were two-sided, and a P value of < 0.05 was regarded as significant. Descriptive statistics are presented as mean ± SD, or median and interquartile range (IQR) for continuous variables or as numbers and percentages for qualitative variables. Differences in baseline characteristics were assessed using an independent Student's t-test or a non-parametric Mann-Whitney-Wilcoxon test for continuous variables, or Fisher's exact test for categorical variables, respectively. To test the differences between groups over the intervention period, an analysis of covariance (ANCOVA) was used, taking into account covariates that were identified by multiple linear regression. The dependent variable was the change in the value of the variable (or its logarithmic transform) at the later visit; the intervention was the factor of interest, and the value of the variable (or its logarithmic transform) at the baseline visit was a covariate. The following variables at baseline were prespecified as possible confounders: sex, BMI, age, baseline physical activity, alcohol consumption, and smoking habits. These variables were included in the model as additional covariates, and a reduced set of confounders were identified by stepwise regression. The final reduced model was used for the formal ANCOVA test. For intractably non-normal variables, the non-parametric Mann-Whitney U test was used to compare changes from baseline between the placebo and probiotic groups.
Microbial profiling and strain recovery assessments utilized a paired Wilcoxon signed-rank test to identify significant differences between week 0 and week 4 of the intervention period. The Mann-Whitney U test was employed on unpaired genomic delta data.

| Baseline characteristics of the participants
The baseline characteristics of the ITT population are presented in Table 1, and were shown to be homogeneous in terms of age, sex, body weight, BMI, heart rate, alcohol consumption and smoking status (all P > 0.05). Diastolic blood pressure was lower in the probiotic group (9.2 kPa vs 9.7 kPa, P = 0.039); however, the absolute difference was similar to that in systolic blood pressure (14.8 kPa vs 15.2 kPa, P = 0.255) and was not significant when accounting for multiple comparisons. The ITT population comprised more women (75.5%) than men (24.5%). The participants were deemed to have functional constipation per Rome III criteria. There were no significant differences in parameters of functional constipation, including PAC-SYM, PAC-QoL, BSS and CSBM at baseline (all P > 0.05).

| Caloric intake and exercise questionnaire
A dietary assessment of macronutrients was performed throughout the study period. The probiotic group showed a higher carbohydrate caloric intake at baseline (P = 0.014), but no within-group changes were observed over the study period (Table 2). Further, no significant differences were observed in intake of total energy, proteins, lipids or water over the intervention period (all P > 0.05). Using the IPAQ scoring protocol, weekly physical activity was estimated and categorized as sedentary/light, moderate, or vigorous intensity.
Between the two groups, there were no significant differences in the total exercise scores, including vigorous, moderate, and sedentary domains (P > 0.05).

| Questionnaire scores
There was no significant difference in the PAC-SYM total score at week 4 (P > 0.05) when comparing probiotic group with the placebo group (Table 3) However, similar within-group reductions (P < 0.001) were observed in the placebo group. The assessment of the PAC-QoL total score did not show significant differences between groups at week 4 (P > 0.05), with both groups showing significant within-group reductions (P < 0.001). Similarly PAC-QoL physical discomfort, psychological discomfort, worries and concerns, and satisfaction subscales (Table S2) were significantly reduced at week 4 in both the probiotic and placebo groups (P < 0.01). Regarding beta diversity, samples did not cluster or separate according to time point in either group. However, it was determined that sex (P = 0.012) and BSS score (P = 0.001), contributed to overall beta diversity. Dietary data from the food records was assessed and was found to not be correlated to the beta diversity of the samples.

| Metagenomic inference
There were no significant differences in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway abundances between week 0 and week 4 in the probiotic or placebo groups. Among KEGG orthologs, in participants receiving probiotic, 10 features had an unadjusted P value of < 0.05 and an absolute fold-change of greater than 2, with nine of week 0 (Table S4). However, no features passed multiple testing correction. In participants receiving placebo no features had an absolute fold-change of greater than 2.

| Safety profile
There were no significant differences in hematology (Table S5) or biochemical safety parameters (Table S6) between placebo or probiotic groups over the study period. Mean values were within clinical reference ranges pre and post-intervention. Additionally, blood pressure, heart rate, weight and BMI remained similar throughout the study (Table S7). There were a total of 25 adverse events, with 14 reported in the placebo group and 11 reported in the probiotic group (Table S8). A total of nine adverse events, primarily gastrointestinal, were found to be possibly related to the study product, with five reported in the placebo group and four reported in the probiotic group.

| DISCUSSION
The present study was a randomized, double-blind, placebo-controlled trial to assess a probiotic product consisting of L. acidophilus DDS-1, P value (between groups) Score per week Mean ± SD P value (within group) Mean ± SD P value (within group)
The current study enrolled on average middle-aged, predominately female participants, with normal to overweight BMI, representing a typical cross-section of the functional constipation population. 34  week, compared with a placebo. 15 Interestingly, the reported effect in the pooled analysis coincided with a reduction in whole gut transit time by 12.4 h, 15 suggesting that the probiotic group in the current study may have experienced improved intestinal transit. However, this remains to be determined.
The gastrointestinal microbiota has been reported to play a role in gut motility. Early studies in germ-free rats demonstrated delayed gastric emptying and colonic transit in comparison with their conventional counterparts. 9 Further, colonizing germ-free rats with L. acidophilus and B. bifidum helped normalize intestinal transit. 38 The mechanisms for this are likely to be varied and may involve endproducts of bacterial fermentation as well as modulation of the neuroendocrine function or the immune response. 39  The current study demonstrated a significantly higher relative abundance of Ruminococcaceae within the probiotic group over the study period. In contrast, the placebo group showed a significant decrease in Clostridiaceae. Ruminococcaceae, the second most abundant Firmicutes family in gut environment, 43 persist in communities of fibrolytic organisms and are well adapted to utilize or degrade complex and otherwise indigestible plant material. 44 As a result, they help generate short-chain fatty acids (SCFA) such as butyrate, acetate, and propionate, which can be used for energy by the host. 45 The Ruminococcaceae family is commonly associated with Clostridium clusters IV/XIVa, essential bacteria that produce SCFA. 46 Further, Ruminococcaceae abundance has been shown to be positively correlated with BSS scores and faster intestinal transit. 47  While probiotic products are ultimately supported by randomized controlled trials demonstrating health benefits, it is also of importance to assess the presence of the live organisms post-transit. 20 In the current study, participants receiving the probiotic had a significant increase in number of genome equivalents for all four strains evaluated, while those who received the placebo had no detectable shift in the strains evaluated. It should be noted that the probiotic product did not significantly disrupt the microbiota composition, in line with a recent systematic review of probiotics which reported no effects on alpha diversity, richness, or evenness. 52 In the current study, richness was maintained in the probiotic group, while simultaneously decreasing in the placebo group. However, no changes were observed when considering richness and evenness. No significant changes in beta diversity were found either, also in line with the systematic review, 52 although beta diversity did appear to be positively correlated with stool consistency.
Placebo response can be high in participants with bowel disorders, potentially leading to altered outcomes reported by the patient, and the power calculation for the current study may have underestimated this response. The study did not include a placebo run-in period. The inclusion of such a period may have helped control the placebo response of participants entering the trial by excluding high responders. 53,54 Longer run-in phases have also been associated with a less pronounced placebo response. 55 Second, the PAC-SYM score was not included as part of the primary inclusion criteria. Aligning the study inclusion with the study outcomes has been recommended, due to significant variation in functional constipation outcomes. 56 These limitations in the current study design may have ultimately affected its ability to differentiate between groups. As it relates to microbial profiling, the study did not assess colonic mucosal microbiota, which may present differences from fecal microbiota in functional constipation. 7 Also, while the study employed Rome III criteria, Rome IV criteria have since been released, which takes the view that functional bowel disorders exist on a continuum instead of as discrete disorders. 57 In summary, the multi-strain probiotic product was shown to be well tolerated but did not significantly affect symptomology, in part due to a significant placebo response. Nevertheless, the probiotic exhibited potential in modulating bowel habits and microbial profile in participants with functional constipation.