The effect of L-rhamnose on intestinal transit time, short chain fatty acids and appetite regulation: a pilot human study using combined 13CO2/H2 breath tests

Background. The appetite-regulating effects of non-digestible carbohydrates (NDC) have in part previously been attributed to their effects on intestinal transit rates as well as microbial production of short chain fatty acids (SCFA). Increased colonic production of the SCFA propionate has been shown to reduce energy intake and stimulate gut hormone secretion acutely in humans. Objective. We investigated the effect of the propiogenic NDC, L-rhamnose, on gastrointestinal transit times using a combined 13CO2/H2 breath test. We hypothesised that L-rhamnose would increase plasma propionate leading to a reduction in appetite, independent of changes in gastrointestinal transit times. Design. We used a dual 13C-octanoic acid/lactose 13C-ureide breath test combined with breath H2 to measure intestinal transit times following the consumption of 25 g d−1 L-rhamnose, compared with inulin and cellulose, in 10 healthy humans in a randomised cross-over design pilot study. Gastric emptying (GE) and oro-caecal transit times (OCTTs) were derived from the breath 13C data and compared with breath H2. Plasma SCFA and peptide YY (PYY) were also measured alongside subjective measures of appetite. Results. L-rhamnose significantly slowed GE rates (by 19.5 min) but there was no difference in OCTT between treatments. However, breath H2 indicated fermentation of L-rhamnose before it reached the caecum. OCTT was highly correlated with breath H2 for inulin but not for L-rhamnose or cellulose. L-rhamnose consumption significantly increased plasma propionate and PYY but did not significantly reduce subjective appetite measures. Conclusions. The NDCs tested had a minimal effect on intestinal transit time. Our data suggest that L-rhamnose is partially fermented in the small intestine and that breath H2 reflects the site of gastrointestinal fermentation and is only a reliable marker of OCTT for certain NDCs (e.g. inulin). Future studies should focus on investigating the appetite-suppressing potential of L-rhamnose and verifying the findings in a larger cohort.


Introduction
Epidemiological evidence suggests that non-digestible carbohydrates (NDC) play a key role in body weight management [1][2][3]. A number of underlying mechanisms have been suggested for this effect including the promotion of satiety, increased intraluminal viscosity, delayed absorption of nutrients and divergent effects on small and large intestinal transit times following NDC consumption [4]. In addition, it has been suggested that the production of short chain fatty acids (SCFA) via the microbial fermentation of NDC may play a role in mediating their appetite-suppressing effects. Increasing the colonic production of the SCFA propionate has been reported to significantly increase postprandial concentrations of the anorexigenic gut hormones peptide YY (PYY) and glucagon-like peptide-1 (GLP-1), and to reduce ad libitum energy intake acutely in humans [5]. The identification of naturally propiogenic NDCs (i.e. NDC that lead to increased relative propionate production following their consumption) that successfully suppress appetite is therefore an attractive research avenue.
L-rhamnose, a non-absorbable monosaccharide, has previously been reported to be propiogenic [6,7]. It is not actively absorbed in the small intestine and reaches the colon intact where it is fermented by a limited microbial community [8]. Inulin, a fructo-oligosaccharide, is an NDC that resists enzymatic digestion in the small intestine and is delivered to the large intestine intact [9]. Dietary supplementation with inulin has been shown to significantly increase caecal production of SCFA compared with cellulose in rodents [10] and been associated with a reduction in energy intake in humans, albeit at very high doses (>35 g d −1 ) [11]. It is however currently unknown how L-rhamnose compares to inulin in terms of its propionate-producing and appetite-suppressing capacity. In contrast, cellulose is often used as a reference fibre due to its low and incomplete fermentability by the human gut microbiota [12]. We were interested to dissect the role of these NDCs on intestinal transit and SCFA production in relation to appetite regulation.
In the present study, we examined whether alterations in intestinal transit time or SCFA production profiles were associated with measures of appetite in healthy humans. We used a dual 13 C-octanoic acid/ lactose ureide breath test [13,14], both of which have been validated against scintigraphy [13,15], combined with breath H 2 , which is routinely used as a measure of OCTT for some NDCs [14], to measure GE and OCTT rates following the consumption of L-rhamnose. We hypothesised that L-rhamnose would significantly increase circulating propionate concentrations and would have a similar intestinal transit rate compared with inulin (positive control) and cellulose (negative control).

Subjects and methods
This study was carried out in accordance with the Declaration of Helsinki and was approved by the London-Harrow Research Ethics Committee (REC) (REC Reference Number 14/LO/0704). All subjects provided informed, written consent.
This pilot study was a randomised, double-blind, cross-over trial involving three study visits. The primary outcome measure was intestinal transit time and secondary outcome measures were plasma SCFA and PYY profiles and finally appetite scores. This was a pilot study because prior data on the effects of L-rhamnose on intestinal transit times was not available to undertake a sample-size calculation.

Participant recruitment
Participants were recruited via existing volunteer databases and poster advertisements. Healthy men and women aged between 18 and 65 years with BMI (kg m −2 ) 20-35 were eligible for inclusion. Exclusion criteria included the following: weight gain or loss 5% body weight in the preceding two months, current smokers, substance abuse, excess alcohol intake, any chronic illness or gastrointestinal disorder, pregnancy, or use of antibiotics and medications likely to interfere with energy metabolism. Eligible participants were randomised into the study by the sealed envelope randomisation service (www. sealedenvelope.com).
Dietary supplements L-rhamnose (Carbosynth Limited, Compton, UK) was the interventional supplement used in this study. A dose of 25 g d −1 was chosen as L-rhamnose has previously been shown to increase serum propionate at this dosage [6,7]. Inulin (Beneo HP, DKSH, London, UK) was used as a positive control (25 g d −1 ). Inulin contains a range of fructose polymers with an average chain length of 23. Cellulose (ACI Group Ltd, Slough, UK) was used as a negative control (25 g d −1 ). Lactose-[ 13 C] ureide and unlabelled lactose ureide were synthesised at the University of Glasgow as previously described [16].

Study day protocol
Participants were asked to avoid caffeine, alcohol and strenuous exercise, and to record their food intake for the 24 h before each study visit. Participants consumed the same evening meal before each visit in order to promote consistency between visits. Participants were also asked to consume 1 g unlabelled lactose ureide 12-24 h before their visit. This induced microbial activity to rapidly hydrolyse lactose-[ 13 C] ureide, leading to a more definitive rise in breath 13 CO 2 from fermentation [14]. Finally, participants were asked to arrive having fasted for 12 h prior to their study visit.
Study visits were conducted between January and May 2015 in the National Institute for Health Research (NIHR)/Wellcome Trust Imperial Clinical Research Facility (CRF), Hammersmith Hospital, London, United Kingdom. All participants attended three separate study visits 7 days apart. An overview of the study visit protocol is given in figure 1.
Prior to the consumption of the test meal (0 min), body composition was assessed by bioelectrical impedance analysis (Tanita BC-418 analyser; Tanita Corporation, Japan). Participants were asked to empty their bladders and change into a set of hospital scrubs prior to the collection of body composition data, in order to promote consistency between visits. In addition, a number of baseline samples were collected in duplicate (>10 min apart) prior to breakfast. Baseline blood samples were collected via a peripheral cannula in order to assess circulating SCFA concentrations. Baseline breath H 2 measurements were collected in real-time using a handheld breath H 2 metre (EC60 Gastrolyser Breath Hydrogen Monitor; Bedfont Scientific). Baseline breath samples were collected in order to measure 13 CO 2 natural abundance and involved blowing through a straw into an Exetainer (Labco Co., High Wycombe, UK) until vapour condensed at the bottom of the tube. Baseline subjective feelings of appetite ('how hungry/full do you feel?') and mood ('how nauseous do you feel?') were also collected using a series of 100 mm VAS. The left extremity of the visual analogue scale (VAS) was labelled with 'not at all' and the right-hand extremity was labelled with 'extremely'. Participants were asked to draw a vertical line on the VAS depending on how intensely they were experiencing each feeling.
Participants received a standardised breakfast (∼470 kcal; supplementary table 1 is available online at stacks.iop.org/JBR/12/046006/mmedia) at 0 min consisting of 100 mg (1-13 C)-Octanoic acid (Sercon Ltd, Crew, UK) and 500 mg lactose-[ 13 C] ureide and 25 g L-rhamnose, inulin or cellulose, according to the participant's randomisation. Participants were provided with a standardised snack (∼650 kcal) at 180 min and a standardised meal (∼550 kcal) at 360 min (supplementary table 1). Further venous blood and breath samples were collected and VAS were completed throughout the study visit (figure 1). After 480 min, the cannula was removed and participants were discharged from the CRF.
Finally, participants were asked to collect breath samples hourly for 4 h following discharge and to collect a breath sample as soon as they woke up on the morning after their study visit in order to keep tracking breath 13 CO 2 enrichment. In addition to this, participants were asked to consume the same evening meal after each visit to promote consistency between visits.
Blood sample preparation 5 ml venous blood samples were collected in plain 5 ml glass tubes and kept on ice until centrifugation. Samples were centrifuged at 2500 relative centrifugal force at 4°C for 10 min. Resulting plasma was aliquoted into cryovials and immediately frozen at −80°C until assaying for SCFA and PYY. Breath sample analysis Breath samples were analysed by isotope ratio mass spectrometry [17], a method used for measuring the relative abundance of isotopes in any given sample, as previously described [14].

Curve-fitting of breath 13 CO 2 data
The fraction of tracer recovered in breath per hour (PDR h −1 ) was calculated using Microsoft Excel software using the following equation: where VCO 2 is the CO 2 production rate (estimated to be 300 mmol m −2 body surface h −1 ; [18], with body surface estimated according to Haycock et al [19]), breath 13 C enrichment is the 13 C enrichment of breath over baseline (ppm excess), mg dose is the amount of substrate administered (mg), MW is the molecular weight of substrate, dose enrichment is the 13 C enrichment of substrate (ppm 13 C), and n is the number of isotopically labelled atoms in the substrate molecule. Breath data were analysed using a modified version of the curve-fitting techniques developed by Ghoos et al [13] and Maes et al [20], as previously pub- The variables a i , b i , c i and d i each describe a facet of each peak (i 1 , i 2 ) in the breath test data; a is the scaling factor related to the amplitude of the peak, b relates to the ascending slope, c relates to the downslope of the peak and d was introduced to facilitate the fitting of multiple peaks, as previously described [14]. The function was fitted to the data using Microsoft Excel software. The nonlinear regression Solver function was used to minimise the difference (root mean square) between the fitted curve and the raw data by adjusting the variables a, b, c and d for both curves simultaneously [21]. Two parameters defined from the fitted curve were used to describe the gastric emptying (GE) rate; t 1/2 and t max, Both have previously been shown to significantly correlate with scintigraphy data, supporting their use as measures of GE [20]. t 1/2 is defined as the timepoint at which 50% of exhaled 13 CO 2 is recovered under the GE curve. t max is defined as the time from substrate ingestion to the time of maximal breath enrichment. Both t 1/2 and t max were calculated as previously described [20] with the minor modification of adding the value of d 1 to each value to account for any delay in GE. The 'd' term for the OCTT curve, d 2 , was calculated as a measure of OCTT. It was defined as the time at which breath 13 CO 2 excretion, from the modelled data (peak 2) was non-zero and was arbitrarily set at 0.01-the time when breath 13 CO 2 excretion reached 0.01 PDR h −1 . This was undertaken using the GOALSEEK function in Excel to return the time (t) when the equation describing peak 2 (using the best-fit parameters) was equal to 0.01.

Breath H 2 excretion times
The first sustained rise in breath H 2 was determined for each volunteer on each visit and was considered a marker of OCTT.

Metabolic and hormone analysis
Thawed samples were vortexed and aliquoted into fresh cryotubes containing heparin sodium (2 μl/ 100 μl sample; 1000 IU ml −1 , Wockhardt UK Ltd). Plasma PYY was measured using a previously established in-house specific and sensitive radioimmunoassay [22,23]. Plasma SCFA were measured using an Agilent 7000C Triple Quadrupole GC/MS System according to a previously published method [24]. Data processing was performed using Agilent MassHunter Quantitative Analysis software and SCFA concentrations were integrated from a freshly prepared calibration curve.
Composite appetite score A composite appetite score (CAS) was calculated from the VAS as previously described using the following formula [25]: Statistics Data were checked for normality (defined as P0.01) using the Shapiro-Wilk test. Results are presented as means±standard error of the mean (SEM) or medians (interquartile range) if non-parametric. Missing VAS, breath H 2 and PYY data points were replaced with the mean group value at that timepoint. The average of two separate baseline measurements for breath H 2 , CAS, plasma SCFA and plasma PYY data, which were collected before breakfast, were used for the 0 min measurement. Baseline values were compared using a one-way analysis of variance (ANOVA).
Hunger 100 Fullness Desire to Eat Appetite for Meal 4 .
If baseline values were significantly different between groups, results are presented as percentage change from baseline. Area under the curve (AUC) and area over the curve (AOC) were calculated using the trapezoidal rule. AUC/AOC and breath data were compared using a one-way repeated measures ANOVA with post-hoc Fischer LSD test. If nonparametric, data were compared using a Friedman's test with post-hoc Wilcoxon signed rank tests. Postprandial SCFA and PYY concentrations were compared with baseline concentrations using paired t-tests or Wilcoxon's signed rank tests. R 2 was calculated using linear regression. Significance was considered P<0.05. Statistical analysis was performed using IBM SPSS Statistics v23-v24. Graphs were prepared using GraphPad Prism v5.0-7.0.

Participants
Participant characteristics are given in table 1. Ten healthy men and women participated in this study.

Breath 13 C enrichment profiles and intestinal transit times
The profiles of breath 13 CO 2 recovery and breath H 2 concentrations following the dual 13 C breath test and the consumption of cellulose, inulin and L-rhamnose are shown in figure 2.
A two-peak model was fitted to the data and used to determine GE and OCTT; individual data and group means are given in figure 3 and table 2, respectively.
There was a small but significant difference for GE rate (t 1/2 ) between groups (P=0.020; table 2). Post-hoc tests revealed that the t 1/2 was longer for L-rhamnose compared with cellulose (effect size: mean±SEM; 0.35±0.13 h, P=0.023) and inulin (0.30±0.11 h, P=0.023), which suggests that L-rhamnose delayed GE by 19.5 min on average. Similarly, there was a significant difference in the t max parameter for GE rate between groups (P=0.017). Post-hoc analysis revealed that the t max for L-rhamnose was longer compared with cellulose (0.45±0.16 h, P=0.018) and inulin (0.34±0.11 h, P=0.012), which also supports the observation that L-rhamnose delayed GE versus cellulose and inulin.
There was no significant difference between groups for the 'd' term parameter calculated to describe OCTT (d 2 , P=0.21). However, there was a significant difference in H 2 excretion times between groups (P<0.001). Post-hoc tests revealed there was a shorter time to sustained H 2 excretion time for L-rhamnose (P=0.004) and inulin (P=0.005) compared with cellulose, suggesting a shorter OCTT for L-rhamnose and inulin as determined by H 2 excretion. L-rhamnose also had a shorter time to sustained H 2 excretion time compared with inulin suggesting a shorter OCTT (P=0.005).
There was no significant difference in t 1/2 , t max , OCTT (d 2 ) or breath H 2 excretion times between visits independent of the NDC consumed (P=0.34-P=0.90).
Correlation between OCTT as assessed by 13 CO 2 and H 2 excretion The correlations between 13 CO 2 and H 2 excretion times following consumption of cellulose, inulin or L-rhamnose are shown in figure 4. As previously described by Schneider et al [26], 13 CO 2 from lactose-[ 13 C] ureide fermentation significantly correlated with the H 2 excretion times following the consumption of inulin (r 2 =0.89, P<0.001; figure 4). However, for both cellulose and L-rhamnose 13 CO 2 data there was no significant correlation with the H 2 excretion times (P=0.63 and 0.77, respectively).

Plasma PYY
While the measurement of plasma PYY concentrations was not an a priori outcome measure, the significant increase in circulating propionate concentrations observed following the consumption of L-rhamnose prompted us to measure plasma PYY. There was no difference in the plasma PYY concentrations between groups at baseline (P=0.47; figure 6). There was a significant difference in the AUC 0-420 for PYY between treatments (P=0.001; figure 6(B)). Post-hoc analysis revealed that the AUC 0-420 for PYY was significantly larger for L-rhamnose than cellulose (effect size: 9.  Composite appetite score There was a significant difference in CAS between groups at baseline (P=0.006). This was due to participants reporting significantly lower CAS at baseline of the L-rhamnose visit compared with cellulose (effect size: 5.6±2.3 mm, P=0.033) and inulin (effect size: 9.7±3.2 mm, P=0.015). Therefore data are expressed as percentage change from baseline ( figure 7). There was no difference between treatments for CAS area over the curve (AOC) 0-480 (P=0.37).

Discussion
A dual 13 C breath test combined with breath H 2 excretion was used in this pilot study in order to evaluate effects on gut transit of three different NDC; cellulose, inulin and L-rhamnose. The aim of conducting this study was two-fold: (1) to ascertain if variations in transit time could explain an observed effect on appetite; (2) to examine the variability in the optimum time-window for assessment of appetite where SCFA production from NDC is maximal and most likely to have an impact on satiety for future studies. L-rhamnose significantly delayed GE when compared with cellulose and inulin (t 1/2 : by 21 and 18 min, respectively). While the caloric load and nutritional composition of a meal are understood to affect GE rates [27], these factors are unlikely to have caused the differences observed in this study as the test meals were provided at exactly the same time in each group, were isoenergetic and had the same nutritional composition. Alternatively, the rate of GE has previously been shown to be strongly influenced by the osmolality of the contents of the duodenum, with increased osmolality slowing GE [28,29]. Therefore, L-rhamnose may have resulted in a delay in GE versus inulin and cellulose due to its greater osmotic load. However, it is currently unknown whether the modest reduction in GE rate observed Each data point represents one participant. t 1/2 , t max and d 2 were determined from the modelled 13 C data in order to describe the GE rate and OCTT; t 1/2 was defined as the timepoint at which 50% of exhaled 13 CO 2 is recovered, t max was defined as the time from substrate ingestion to the time of maximal breath enrichment and d 2 was defined as the time at which breath 13 CO 2 excretion was non-zero. The H 2 excretion time was defined as the time at which there was the first sustained rise in breath H 2 and was measured in real-time using a breath H 2 monitor. Abbreviations: GE, gastric emptying; IRMS, isotope ratio mass spectrometry; OCTT, orocaecal transit time.
following the consumption of L-rhamnose would have a significant impact on eating behaviour.
Despite detecting differences in GE rates between the NDC there were no differences in OCTT as measured from breath 13 C excretion. However, in contrast, breath H 2 excretion times suggest a significantly shorter transit time for L-rhamnose and inulin versus cellulose, and a significantly shorter transit time for L-rhamnose versus inulin. This difference in gut transit times as assessed using breath H 2 and 13 CO 2 was an unexpected finding and not in line with the original hypothesis. The H 2 -based inulin breath test has previously been used in multiple research studies as a reliable measure of OCTT and has been shown by Schneider et al to correlate well with the lactose-[ 13 C] ureide breath test data following the consumption of inulin [26]. Similarly, in the present study, transit time determined from the breath H 2 excretion times significantly correlated with the 13 C data following the consumption of inulin thus validating our data modelling approach and confirming fermentation of inulin in the caecum. However, for both cellulose and L-rhamnose the 13 C data does not correlate with the H 2 excretion data. It is possible that differences in the sites of fermentation of these substrates may explain the findings. Despite the perception that L-rhamnose reaches the colon intact [7], the early rise in breath H 2 in the present study occurs before the stomach has fully emptied and before L-rhamnose has reached the caecum according to the 13 C data suggesting that L-rhamnose is being partially fermented in the small intestine. In support of this hypothesis, Boojink et al reported the presence of species related to Ruminococcus obeum in the ileal effluent of humans [30], which has previously been shown to ferment L-rhamnose via the propanediol pathway leading to the production of propionate [8]. In contrast to L-rhamnose and inulin, cellulose is thought to have low or incomplete fermentability [12]. However, in the present study there was a small but sustained rise in breath H 2 from 6 h following the consumption of cellulose, consistent with observations from previous studies [31], and suggests that cellulose is fermented more distally in the colon compared with inulin [32]. There was also a similar later rise in breath H 2 following consumption L-rhamnose, which corresponds with the 13 C data and appears to represent fermentation of L-rhamnose in the colon. Together, the presented data suggest there was no difference in OCTT for the three NDCs tested but a difference in the fermentability and the site of fermentation within the intestine of the substrates. Furthermore, these data suggest that breath H 2 is only a suitable measure of OCTT following the consumption of certain NDCs (e.g. inulin) and its use as a marker of OCTT is unreliable following consumption of some NDCs (e.g. L-rhamnose and cellulose). Therefore, it is proposed that breath H 2 should only be used as a reliable marker of fermentability and not OCTT unless independent verification of the site of fermentation can be established. OCTT. a Parametric data were compared using a one-way repeated measures ANOVA with post-hoc Fischer LSD test. Non-parametric data were compared using a Friedman's test with post-hoc Wilcoxon Signed Ranks Test. b t 1/2 was defined as the timepoint at which 50% of exhaled 13 CO 2 is recovered. c T max was defined as the time from substrate ingestion to the time of maximal breath enrichment and was calculated using the following equation , a measure of OCTT, was defined as the time at which breath 13 CO 2 excretion, from the modelled data (peak 2) was non-zero and was arbitrarily set at 0.01-the time when breath 13 CO 2 excretion reached 0.01 PDR h −1 . This was undertaken using the GOALSEEK function in Excel to return the time (t) when the equation describing peak 2 (and using the best-fit parameters) was equal to 0.01. e H 2 excretion time was defined as the time at which there was the first sustained rise in breath H 2 concentrations. f Indicate significant differences in post-hoc tests.
Plasma propionate concentrations were significantly higher following consumption of L-rhamnose versus inulin and cellulose, which supports the hypothesis that L-rhamnose is propiogenic [6,7]. Plasma propionate concentrations were elevated above baseline from 120 to 360 min (P=0.023). This early increase in plasma propionate is observed following breath H 2 excretion and is detected in the peripheral circulation before L-rhamnose has reached the caecum according to the 13 C data. Together, these data further support the proposal that L-rhamnose does not reach the colon intact, as previously suggested, and is partially fermented in the small intestine. Interestingly, plasma acetate concentrations were higher following L-rhamnose compared with inulin, which was unexpected. One possible explanation for this increase in acetate could be due to differences in the rate of fermentation of the NDC owing to differences in chain length. For example, other short chain inulin molecules like fructo-oligosaccharides are more rapidly fermented versus long-chain inulin [33], which undergoes less rapid fermentation [9].
The co-expression of the SCFA receptors, free fatty acid receptor (FFAR) 2 and FFAR3, in PYY and GLP-1 secreting enteroendocrine cells has prompted suggestions that SCFA production following the consumption of NDC may stimulate the secretion of these anorexigenic gut hormones [5]. In acute studies, increased colonic propionate has previously been shown to increase plasma PYY and GLP-1 concentrations in humans [5]. Therefore, it was originally hypothesised that the consumption of L-rhamnose would reduce appetite via an increase in colonic propionate production, which would stimulate the secretion of anorexigenic gut hormones. Although our data suggests that the increase in plasma propionate concentrations following the consumption of L-rhamnose may be due to an increased production of propionate in the small intestine as well as in the colon, the propionate data prompted us to measure plasma PYY concentrations [34]. As hypothesised, the increase in peripheral propionate levels observed following consumption of L-rhamnose was associated with a . Correlation between orocaecal transit times as assessed by 13 CO 2 and H 2 excretion. Correlation between OCTT (h) as measured using the 13 CO 2 and H 2 excretion times for (A) cellulose, (B) inulin and (C) L-rhamnose (n=10). OCTT was defined as the time at which breath 13 CO 2 excretion, from the modelled data (peak 2) was non-zero or as the time at which there was the first sustained rise in breath H 2 concentrations, respectively. R 2 was calculated using linear regression. Abbreviations: OCTT 13CO2, orocaecal transit time assessed with lactose-[ 13 C] ureide breath test; OCTT H2 , orocaecal transit time assessed with H 2 breath test. significant increase in plasma PYY concentrations versus inulin and cellulose and is the first time to our knowledge that L-rhamnose has been shown to increase plasma PYY concentrations. Interestingly, postprandial PYY was most elevated after L-rhamnose appeared to arrive in the colon. The increase in postprandial PYY concentrations noted in the current study suggests that L-rhamnose could influence satiety. However, as appetite was not the primary outcome of this pilot study, the only appetite measure collected were VAS. While there was no effect of treatment on the CAS, as this study was not adequately powered to measure differences in VAS these data should be treated with caution. Interestingly, since the completion of the current study, Darzi et al reported that 25.5 g L-rhamnose does not reduce appetite [35]. However, ad libitum food intake was measured at 420 min, which is significantly later than the peak propionate and PYY concentrations noted in our study (240-300 min), suggesting effects on appetite may have been missed and should be addressed in future studies. Finally, a notable finding in the current study is the large inter-and intra-individual variability in GI transit times as outlined in figure 3, which others have previously reported [36,37]. While the data from the current study are in line with the literature, the variations in gut transit times further emphasise the challenges when designing such studies aiming to detect differences in appetite.
The use of a dual 13 C breath test, which has previously been validated against scintigraphy [13,15], and a human cross-over study design are major strengths of this trial. However, the small sample-size in this pilot study is a limitation. Future studies are required in order to verify our findings in a larger cohort.
In summary, the presented data suggest that the NDC tested have minimal effects on gut transit time. Despite previously being considered to be fermented exclusively in the colon, the presented 13 C breath data, breath H 2 and plasma SCFA data suggest that L-rhamnose also undergoes microbial fermentation in the small intestine. Furthermore, our data suggests that breath H 2 should only be used as a reliable marker of fermentability and not OCTT unless corroborated by an independent measure of OCTT. Finally, despite elevated propionate and PYY concentrations following consumption of L-rhamnose in addition to L-rhamnose slowing GE rate, L-rhamnose did not suppress subjective feelings of appetite. Further work is needed to assess whether these effects would alter eating behaviour.