Detection of changes in regional colonic fermentation in response to supplementing a low FODMAP diet with dietary fibres by hydrogen concentrations, but not by luminal pH

Summary Background Carbohydrate fermentation plays a pivotal role in maintaining colonic health with excessive proximal and deficient distal fermentation being detrimental. Aims To utilise telemetric gas‐ and pH‐sensing capsule technologies for defining patterns of regional fermentation following dietary manipulations, alongside conventional techniques of measuring fermentation. Methods In a double‐blind crossover trial, 20 patients with irritable bowel syndrome were fed low FODMAP diets that included no extra fibre (total fibre content 24 g/day), or additional poorly fermented fibre, alone (33 g/day) or with fermentable fibre (45 g/day) for 2 weeks. Plasma and faecal biochemistry, luminal profiles defined by tandem gas‐ and pH‐sensing capsules, and faecal microbiota were assessed. Results Plasma short‐chain fatty acid (SCFA) concentrations (μmol/L) were median (IQR) 121 (100–222) with fibre combination compared with 66 (44–120) with poorly fermented fibre alone (p = 0.028) and 74 (55–125) control (p = 0.069), but no differences in faecal content were observed. Luminal hydrogen concentrations (%), but not pH, were higher in distal colon (mean 4.9 [95% CI: 2.2–7.5]) with fibre combination compared with 1.8 (0.8–2.8) with poorly fermented fibre alone (p = 0.003) and 1.9 (0.7–3.1) control (p = 0.003). Relative abundances of saccharolytic fermentative bacteria were generally higher in association with supplementation with the fibre combination. Conclusions A modest increase in fermentable plus poorly fermented fibres had minor effects on faecal measures of fermentation, despite increases in plasma SCFA and abundance of fermentative bacteria, but the gas‐sensing capsule, not pH‐sensing capsule, detected the anticipated propagation of fermentation distally in the colon. The gas‐sensing capsule technology provides unique insights into localisation of colonic fermentation. Trial registration: ACTRN12619000691145.


| BACKG ROU N D
Changes in dietary intake can readily and substantively modify the activity of colonic microbiota by altering the amount and type of substrates available for fermentation. For example, saccharolytic fermentation is associated with putative health benefits, generating gases (carbon dioxide, hydrogen and methane) and short-chain fatty acids (SCFA) while acidifying the lumen. 1 Short-chain fatty acids have a plethora of cellular effects via multiple pathways as shown in vitro and in vivo. 2 Butyrate, for example, plays essential nutritive roles for the colonic epithelium, exerts concentration-dependent anti-inflammatory, differentiative and anti-tumorigenic effects, but can be toxic at higher concentrations. 2 Importantly, butyrate mediates its colonic effects topically rather than via systemic delivery following its absorption. 2,3 Hence, local luminal concentrations and production are of crucial importance to its actions.
Dietary fibres are the major substrate for colonic SCFA production. Accumulated knowledge regarding the fermentation and interaction of specific dietary fibres in the colon has enabled dynamic models of regional fermentation to be developed based upon animal studies 4,5 and human observations. 6 These models are founded on three key concepts. First, slowly absorbed or non-digestible shortchain carbohydrates, fermentable oligosaccharides, disaccharides, monosaccharides and polyols (FODMAPs), when consumed in usual amounts, are rapidly fermented in the proximal colon leading to their depletion in the distal colon. Excessive fermentation proximally in the colon is potentially injurious, inducing barrier dysfunction and mucosal inflammation, 7 and may induce symptoms such as abdominal pain and bloating, especially in patients with irritable bowel syndrome (IBS). Reducing dietary intake of FODMAPs ameliorates such symptoms, 7 but such a strategy potentially reduces fibre intake, 8 which may reduce delivery of fermentative substrate to the distal colon. 3 Second, degradation of less rapidly fermented fibres, such as resistant starch (RS), is likely occur over a longer length of the colon but does not spread to the whole colon, unless very large doses are used. 9 This effect is likely to be due to the sheer amount of fibre available for fermentation. Such a strategy to spread fermentation to the distal colon is not favoured in clinical practice since it is associated with high levels of fermentation in the proximal colon and may induce gastrointestinal symptoms, particularly in patients with IBS.
The third concept involves spreading fermentation of fermentable substrates towards the distal colon, but without excessive fermentation proximally. This has been achieved by adding a nonor poorly fermentable fibre with fermentable fibres. Initial experiments were performed in rats in which luminal concentrations of fermentation products, specifically SCFA, were higher in the distal colon by combining wheat bran, a bulking fibre, with resistant starch than could be achieved by either alone. 5 The potential health value of such an approach was supported by suppression of tumorigenesis in rats using the fibre combination strategy, 5 likely to be related to increased delivery of butyrate to the distal epithelium. 10 Given the methodological difficulties in proving this directly in humans, a study was performed in pigs, whose colonic function better mimics that of humans. Indeed, a combination of RS and wheat bran spread fermentation evenly around the colon. 4 The increase in substrate and fermentation in the distal colon using this fibre combination was subsequently shown in healthy humans using large doses of RS and measuring SCFA concentrations in the faeces where residual starch was increased in the faces in association with wheat bran and RS supplementation. 9,11 In a recent study that utilised magnetic resonance imaging of the colon, the effect of psyllium (poorly fermentable) and inulin (fermentable) showed the spread of fermentation to the distal colon in humans. 12 Two potential mechanisms by which the effect of a poorly fermented fibre can lead to a distal spread of fermentation have been identified. The gel-structures formed by viscous fibres reduce the accessibility of fermentable fibres to the microbiota, which slows fermentation in vivo, but does reduce the amount of in vitro. 12 The other potential contributing factor is that the transit-hastening effects of bulking fibres propel the fermentable substrate along the colon. 4,6 Measurement of regional colonic fermentation is problematic.
Conventional techniques of assessing the delivery of SCFA to the colonic mucosa in humans comprise measurement of metabolite concentrations in faeces, which are more likely to be representative of fermentation in the distal colon and rectum and poorly represent their production, 13,14 and are subject to the artefacts of ongoing fermentation of residual carbohydrates in the faeces ex vivo. 15 Evaluation of breath hydrogen and methane, and plasma SCFA concentrations, can provide indirect insights into overall fermentation occurring in the intestine, but gives limited information on where the fermentation is occurring, and also has limited reproducibility. 2 Similarly, taxonomic and metabolic analyses of faecal microbiota offer limited insights into local processes.
Telemetric capsule technologies offer the opportunity to overcome such limitations via localised, real-time assessments of intraluminal metabolite concentrations in the ambulant person without physiological disruption. Luminal pH, measured by the wireless motility capsule (WMC), has been applied as marker of luminal fermentation due to the acidification related to SCFA, succinate and lactate formation, but pH is the sum of metabolic activities and hence not specific to fermentation. 16 In contrast, luminal hydrogen concentration offers a highly specific measure of local fermentation since hydrogen is only produced via saccharolytic fermentation and is rapidly consumed via microbial metabolic pathways or via absorption to systemic circulation. 1 The development of a telemetric capsule that samples volatile molecules, including simple gases such as hydrogen, through a semipermeable membrane offers a more specific option. 17 Thus, the current study aimed first, to evaluate the hypothesis that the WMC and gas-sensing capsule can detect changes in colonic fermentation and its distribution induced by manipulation of the types of dietary fibres consumed in humans. Second, the ability of luminal pH and hydrogen concentrations to detect changes in regional fermentation was compared. Third, the study aimed to compare the telemetric findings with those using conventional techniques of assessing fermentation. To do this, patients with IBS were studied under carefully controlled conditions by feeding them diets restricted in FODMAP content (to reduce proximal fermentation) and supplemented or not with sugarcane bagasse, a poorly fermented fibre comprised of fractions highly resistant to fermentation (~50% cellulose, 25%-35% hemicelluloses, 15%-25% lignin) 18,19 and possessing stool bulking properties, 20 alone or with a moderately fermentable RS (to change regional fermentative profiles). Regional changes in the colonic lumen were measured via tandem ingestion of telemetric capsules.

| Participants
The participants have been previously described in detail. 20

| Trial design and procedures
The trial design and most procedures have been previously described in detail. 20 Briefly, participants maintained typical dietary habits during a 7-day baseline before being randomised to receive one of three 14-day dietary interventions (detailed below). Both participants and investigators were blinded to the diets. The three dietary interventions were separated by a ≥21-day washout, where participants resumed typical dietary habits.
Trial procedures are illustrated in Figure S1. Assessments

| Interventional diets
The dietary interventions (designated 'Control', 'Sugarcane', 'Combination') were delivered via controlled feeding, where most food was provided to participants, as previously described. 20 Briefly, the Control diet comprised a base low FODMAP diet; the Sugarcane diet comprised the base low FODMAP diet supplemented with 10 g/ day fibre from sugarcane bagasse (Tamu Pty. Ltd.); the Combination diet comprised the base low FODMAP diet supplemented with 10 g/day fibre from sugarcane bagasse and 12 g/day RS from highamylose starch (Hi-Maize 1043, Ingredion). Other than fibre content, the diets were nutritionally identical (Table S1).

| Telemetric assessments
The telemetric capsules studied were the WMC (SmartPill™, Medtronic Australasia) that measures luminal pH, pressure and temperature and gas-sensing capsule (Atmo Gas Capsule, Atmo Biosciences) that detects a range of gas concentrations, predominantly hydrogen, carbon dioxide and oxygen, and temperature ( Figure S2). On day 9 during each dietary intervention, a sub-group of participants ingested the capsules in tandem in a random order after consuming breakfast comprising cereal with 250 mL lactosefree milk ( Figure S1), before fasting for 4 h and resuming the intervention diet as previously outlined in detail. 20 The 4-h fasting period was associated with delayed gastric emptying in some patients, 20 but this did not change the ability to assess colonic gas patterns.
Participants wore receivers corresponding to each capsule until their passing, confirmed by fall in temperature, signal loss following bowel movement and/or visual confirmation in collected faecal samples.
Colonic transit time was calculated from the time each capsule reached the ileocaecal junction to its excretion, as previously reported and validated. 22,23 Colonic pH was examined as an average across the entire colon, together with nadir and peak pH and their timing after the ileocaecal valve. Colonic hydrogen concentrations were expressed as a percentage of the gas detected within the lumen, together with peak concentration and its timing after the ileocaecal valve. The colon was segmented into quartiles based on relative transit time 24 to enable regional fermentation to be assessed. A coefficient of variation <15% was taken as a valid result. Faecal pH was measured with a calibrated pH probe (Five-Go pH meter, Mettler-Toledo) with the sample warmed to 25°C.

| Plasma metabolite assessments
Plasma samples were analysed in duplicate for SCFA content as previously described. 25 Briefly, 300 μL of plasma was spiked with 200 μM internal standard (1.68 mM heptanoic acid) and acidified using 10% sulfosalicylic acid, with 3 mL diethyl ether solvent added. The mixture was vortexed and then centrifuged (400 g, 2 min, 4°C) to clarify the organic layer, which was transferred into 50 μL 0.  Paired-end short-read sequence data generated on the Illumina MiSeq was processed using the VSEARCH package. 30 Demultiplexed paired-end sequences were passed through cutadapt for primer removal 31 and then merged prior to sequence quality filtering, followed by error correction, 32 chimera checking, 33 and clustering of sequences to Amplicon sequence variants (ASVs). 34 Taxonomic classification of bacterial ASVs was done using the IDTAXA algorithm implemented in the DECIPHER R package against the SILVA SSU r132 training set. 35

| Participants successfully adhered to dietary interventions
Twenty participants, 19 Figures S3 and S4. As previously reported, dietary adherence was excellent: participants consumed ≥89% of the fibre-containing meals across the diets, with minimal deviations from the diets reported. 20 The participants successfully reduced consumption of FODMAPs during all interventions and intake of fibre and/or RS was increased as planned during the Sugarcane and Combination diets. 20

| Effect of diets on plasma SCFA and faecal metabolite concentrations
Total plasma SCFA concentrations were median 64% and 83% higher in the Combination diet compared with those during the Control (p = 0.069) or Sugarcane diets (p = 0.028) respectively (Table 1; Figure 1). Across the diets, 92%-95% of plasma SCFA was acetate.
Propionate and butyrate plasma concentrations were near or below the lower limit of the assay and are not presented.
Total faecal concentrations of SCFA were similar across the diets ( Figure 1), as were the major SCFA ( Table 1)

| Colonic pH profiles from the wireless motility capsule
No differences in overall colonic pH or the level and timing of the pH nadir were observed across the diets (Table 2; Figure 2). Peak pH was higher during the Control compared with the Sugarcane (mean difference 0.4; p = 0.035) and Combination diets (0.5; p = 0.012), occurring numerically later in the Combination diet (at median 80% of colonic transit time) compared with that in the Control (67%, p = 0.423) and Sugarcane diets (57%, p = 0.092). Proximal-to-distal gradient of colonic pH increased similarly across the three diets with no statistical differences (Figure 3).

| Colonic hydrogen profiles from the gas-sensing capsule
Overall colonic hydrogen concentration tended to be higher during the Combination compared with Control (mean difference 1%; p = 0.076) and Sugarcane diets (1%; p = 0.052) ( Table 3; Figure 2).
Peak hydrogen concentration was similar across the diets but occurred later during the Combination compared with Control diet; when expressed in terms of absolute time from ileocaecal junction, the median difference was 7 h (p = 0.071) and, relative to colonic transit time, the median difference was 52% (p = 0.011). Regionally, hydrogen concentration tended to exhibit a proximal-distal fall during the Control and Sugarcane diets, and appeared to increase across proximal-to-distal quartiles in association with the Combination diet.
In Quartile 4, hydrogen concentration was more than two-fold higher during the Combination diet compared with the Control (p = 0.003) and Sugarcane diets (p = 0.003) (Figure 3).

| Faecal microbiota composition
The compositions of faecal microbiota were compared across the three diets (Figure 4). There were no differences in alpha-diversity or richness (data not shown). On unsupervised principal component analysis (PCA), there were no overall differences in composition.
However, when the most discriminative features in the data were analysed by sparse partial least-squares discriminant analysis, clear differentiation in microbial composition between the Combination and Control diets with less discrimination between the Sugarcane and the other two diets was observed, as shown in the PCA plots and heatmap. For individual ASVs, the major differences in general were a marked increase in relative abundance of Ruminococcus and

Control Sugarcane Combination
Faecal metabolite concentrations (μmol/g)  fibre supplementation had to be within the scope of normal clinical practice and supra-physiological doses avoided. 37 Patients with IBS were studied since FODMAP and fibre intake is of direct relevance to them, with doses used modest and well-tolerated. 38 Third, that fermentation patterns would be altered in its total amount and distribution along the colon was important so that shifts in regional fermentation might be detected. To do this, a fermentable substrate (RS) was used together with a poorly fermented fibre (sugarcane bagasse) since such a combination of fibres with these characteristics has been demonstrated by studies of healthy subjects and patients with IBS. 6,11,12 The background diet was low in FODMAP content as this would specifically reduce proximal fermentation, as shown via breath gas. 39 That the supplemented fibres differed markedly in fermentability had already been shown in vitro. 19 Hence, it was anticipated that distal fermentation would be enhanced by the sugarcane bagasse/RS combination. Finally, a cross-over design was essential to permit the evaluation of changes in indices that have considerable inter-subject variance, such as SCFA concentrations and faecal microbiota.
That overall colonic fermentation was increased by the combi- However, the location of fermentation within the colon is as important as its magnitude. As outlined above, clinical problems may arise from too much in the proximal colon (symptom genesis and mucosal injury) or too little in the distal colon (loss of protection from carcinogenesis or impairment of barrier function). 7 Conventional faecal and plasma measures provide few insights, but telemetric measurement of metabolites that are rapidly depleted by either metabolism or absorption at the site of production provides a unique opportunity to define variations along the colon. While localisation of key landmarks enabling assessment of regional gastrointestinal transit has been validated for the WMC and gas-sensing capsule, 23 the localisation within the large bowel itself is less precise and depends upon the net movement of the Control Sugarcane Combination pH Overall 6.9 (6.6-7.2) 6.7 (6.2-7.1) 6.7 (6.4-7.0) Quartile 1 6.5 (6.2-6.8) 6.5 (6.2-6.8) 6.3 (6.0-6.6) Quartile 2 6.9 (6.6-7.2) 6.7 (6.2-7.2) 6.9 (6.6-7.1) Note: Data shown as mean (95% CIs) for pH values, median (IQR) for time to pH nadir and peak, and analysed via linear mixed models. Significant differences (p ≤ 0.05) between the dietary interventions shown via shared superscripts.

TA B L E 2
Colonic pH profiles across the dietary intervention periods, including overall and regional colonic pH, as well as pH nadir and peak metrics.
capsules from proximal to distal colon. Hence, information from the colon was divided into quartiles with the first and last clearly being associated with proximal and distal colonic events, respectively, and the time to peak concentrations or nadir of the pH were applied in order to judge quantitative distribution of fermentation.
Luminal pH may be a useful marker for the degree of carbohydrate fermentation in the proximal colon, 3 but this was not investigated in the current study. The value of pH in the distal colon may be considerably reduced since more metabolites that contribute to the net pH, such as ammonia described earlier, occur there. Indeed, the pattern of pH across the quartiles was similar across all dietary arms and, therefore, provided little insight into changes in distal fermentation patterns due to its non-specificity.
In contrast, the specificity of hydrogen production to carbohydrate fermentation and the likely expectation that, within one individual, the hydrogen-disposal mechanisms will be similar along the colon, it might be anticipated that, within a single study, variations of hydrogen concentrations might reflect differences in production. Indeed,

AUTH O R S H I P
Guarantor of the article: Peter R. Gibson.