Gut-brain Axis Dysfunction Underlies FODMAP-induced Symptom Generation in Irritable Bowel Syndrome

Jie Wu; Imke Masuy; Jessica R Biesiekierski; Heather E Fitzke; Chinar Parikh; Laurel Schofield; Hafsa Shaikh; Anisha Bhagwanani; Qasim Aziz; Stuart A Taylor; Jan Tack; Lukas Van Oudenhove

Disclosures

Aliment Pharmacol Ther. 2022;55(6):670-682. 

In This Article

Materials and Methods

Study Subjects

We recruited female, right-handed or ambidexter healthy subjects and Rome IV IBS patients,[14] aged between 18 and 55 years old (Supplementary Material S1). Only females were recruited to avoid confounding sex differences and because females showed less variation in the volume of FODMAP solution consumed in our aforementioned study,[6] where FODMAP and control solutions were intragastrically administered until full satiation. The study was performed at the University Hospitals Leuven, Belgium from February 2018 to August 2019. Informed consent was obtained from all subjects before performing any study procedure.

Sample Size

In the absence of previous data on brain responses to FODMAP administration, sample size was calculated based on the significantly different increase in cramps after fructans compared to glucose in IBS patients vs HC in our previous study.[6] Assuming a similar effect size for this condition-by-group interaction and within-participant variability over time, a sample size of n = 13 per group yields 90% power at α = 0.05 as calculated for a linear mixed model with GLIMMPSE version 2.2.8 (Details at https://github.com/labgas/proj-fodmap-fmri/tree/main/sample_size_calculation).[15] It should be noted that we did not correct for the number of symptom outcomes/models in this power calculation, which is not commonly done, but should nevertheless be considered a limitation. However, we did apply stringent multiple testing correction within each model (see below). Power is virtually impossible to calculate for the brain imaging analysis due to the lack of previous data, but has arguably higher (within-subject) power compared to the behavioural data analysis given the higher number of repeated measures over time in each condition (49 for brain imaging data vs 7 for GI symptoms).

Study Design

This placebo-controlled, randomised, double-blind cross-over trial consisted of three study visits, with a washout period of at least 1 week. Figure 1 provides an overview of the procedures performed on each study visit. After an overnight fast, a feeding probe was positioned transnasally into the stomach (50–55 cm); its position was verified by measuring the pH of aspirated gastric fluid using pH strips. The probe was taped to the nose with adhesive tape to prevent displacement after which a 15-minute break was held to avoid any autonomic nervous system influences.[16]

Figure 1.

Overview of procedures performed on one study visit. GI, gastrointestinal; PANAS, positive and negative affect; POMS, profile of mood states; VAS, visual analogue scale

Then, the baseline abdominal MRI scan was acquired, followed by brain scanning using a pharmacological MRI design, starting with a 10-minute pre-infusion baseline. Then, 500 mL of solution was intragastrically infused at a constant speed of 60 mL/min. The solutions contained fructans (BioCare® Ltd, 40 g in 500 mL 0.9% saline, degree of polymerisation = 3–10; osmolality = 72 mOsmol/kg), glucose (Glucopur glucose powder, 40 g in 500 mL 0.9% saline), or 500 mL 0.9% saline only (Table S1). Functional brain imaging continued for 49 minutes after the start of the infusion; at 1 and 2 hours after the infusion start, abdominal MRI scanning was repeated.

GI symptoms (bloating, fullness, nausea, cramps, pain and flatulence) were scored on a 100 mm visual analogue scale (VAS) pre-, and 10, 20, 30, 40, 50, 60, 90, 120 and 150 minutes post-infusion. Emotional state was assessed using the positive and negative affect schedule (PANAS)[17] pre-, and 60, 90, 120 and 150 minutes post-infusion, and the profile of mood states (POMS)[18] (fear, anger, depression, vigour and fatigue) pre-, and 10, 20, 30, 40 and 50 minutes post-infusion.

The study was approved by the Ethics Committee of the University Hospitals Leuven, Belgium (number: S60607) and was performed in accordance with the declaration of Helsinki and BMJ guidelines. All authors had access to study data and reviewed and approved the final manuscript. The study was registered on www.clinicaltrials.gov as NCT04283487.

Randomisation and Blinding

The infusions were given in counterbalanced order with a washout period of at least 1 week. Allocation using https://www.randomization.com was done by a colleague not involved in the study, who also prepared the solutions; subjects and researchers were both blinded to the test solution. Data were entered prior to unblinding.

Abdominal MR Imaging Analysis

Abdominal MRI images were acquired using a 1.5 T whole-body scanner (Philips Medical Systems, Best, The Netherlands) in the supine position with no oral contrast. Motility was quantified from dynamic balanced turbo field echo (BTFE) images acquired during 20 second breath-hold using GIQuant®, a validated technique based on the dual registration of abdominal motion (DRAM) (Motilent Ltd, Ford, London).[19,20] Ascending colon gas and volume content were measured on T2-weighted anatomical images. A summary of the data processing pipeline is shown in Figure 2. Further details are provided in Supplementary Material S2, Tables S2 and S3, and Figure S1-S4.

Figure 2.

Overview of the abdominal MRI data processing pipeline

Brain MR Imaging Analysis

Functional brain MR images were acquired using a 3.0 T Philips Achieva DStream MR system with a 32-channel head coil (Philips Medical Systems). Data were analysed using Statistical Parametric Mapping (SPM12, Wellcome Trust Centre for Neuroimaging, UCL) implemented in MATLAB R2014b (MathWorks). Pre-processing, first (ie subject) level and second (ie group) level analysis were performed. Voxel-level threshold was set at P < 0·05 family-wise error (FWE) corrected for multiple testing. Voxel-based analysis was performed within a single mask of a priori pain-responsive regions of interest generated using "pain" as specific term of interest in the automated meta-analytical tool Neurosynth (https://neurosynth.org).[21] Further details are provided in Supplementary Material S3.

Statistical Analysis of Non-brain Imaging Data

Statistical analysis was performed in Statistical Analysis System (SAS) version 9.4 (SAS Institute). Data were considered statistically significant when P < 0.05. Demographical parameters were compared between IBS patients and HC using two-sample t tests.

Symptoms, abdominal parameters and the associations between both were analysed using marginal linear mixed models. Where the outcome variable or its logarithmic transformation was not normally distributed, the data were ordinalised into tertiles or quartiles and analysed using generalised linear mixed models with a cumulative logit link function. The variance-covariance structure providing the best fit was chosen based on the minimum Akaike's information criterion (AIC).

Comparison Between Conditions and Groups. Condition, time (both within-subject) and group (between-subject) were the categorical independent variables. A main effect of visit (categorical, within-subject) was added to control for putative visit differences. Effects of interest included the main effect of group, testing overall differences between IBS and HC, the group-by-time interaction effect, testing the differences in the time course between both groups, and the group-by-condition two-way interaction effect, comparing the differences between conditions between both groups. To clarify the latter effect, constituting our main hypothesis test, change from baseline values were compared between conditions (fructans vs glucose and fructans vs saline) within groups as well as between groups within conditions using planned contrast analyses, with stepdown Bonferroni multiple testing correction.

Covariation Between Abdominal Responses and Symptoms in IBS Patients. Differences in both GI symptoms and gut physiology parameters between conditions (fructans vs glucose and fructans vs saline) at each timepoint were calculated and, in case of colonic parameters, averaged over timepoints. Time and difference in abdominal parameters were added to the model as independent variables with differences in GI symptoms as dependent variable. The main effect of the abdominal parameters constitutes the effect of interest testing covariation between abdominal responses and symptom responses (over all timepoints).

processing....