Transit

A total of 65 patients successfully completed whole gut transit time (WGTT) measurements at baseline, 28 on ondansetron and 37 on placebo. Their results are shown in Table 26. There were no significant differences between treatments at baseline; however when differences in WGTT between baseline and week 12 were compared, ondansetron tended to show a bigger increase (Figure 21), mean (SD) being 3.8 (9.1) on ondansetron versus a decrease –2.2 (10.4) on placebo, p = 0.011. Comparing the changes in regional transit times (right, left and rectosigmoid colon) between ondansetron and placebo (Table 27), only the increase in rectosigmoid transit was statistically significant.

TABLE 26

Whole gut transit time in hours at baseline and change at week 12 on ondansetron and placebo

FIGURE 21

Change in WGTT in hours. This shows the change in WGTT in hours, which increased significantly more on ondansetron compared to placebo, p = 0.011.

TABLE 27

Regional gut transit time in hours at baseline and change at week 12 on ondansetron and placebo median (interquartile range)

Barostat

Data are available on just eight patients allocated to ondansetron and 10 on placebo of whom only seven and six, respectively, also underwent a second study on treatment, meaning we were seriously underpowered to assess any effect. See Tables 28 and 29 below for further details at baseline.

TABLE 28

Demographics and barostat values at baseline

TABLE 29

Change in thresholds for pain and urgency from baseline to week 12 on ondansetron vs. placebo

Thus, despite showing a substantial change in the mean volume to reach urgency threshold (Figure 22), which on ondansetron rose by 84 (SD 61) ml while on placebo rose from 38 to 76, mean difference 38 (SD 48) ml, there was wide variability in this effect, which was non-significant, p = 0.26.

FIGURE 22

Change in urgency threshold volume by treatment. The change from baseline on treatment was assessed by barostat and showed a non-significant increase in volume on ondansetron.

Thus, using these figures, we calculate that to detect the difference in volume threshold urgency with 90% power would require 16 subjects per group showing as suspected, we were significantly underpowered. However, using a crossover design and paired test in the same individual we would need to study only 10 individuals, which could easily be done separate from any long-term trial.

The difference in the change in pain threshold pressure (Figure 23) was –3.500 ± 7.0 mmHg, which would however have required 85 per group to achieve 90% power to detect such a difference emphasising the small effect size on pain. Unlike alosetron there was an even smaller effect size for compliance, which increased just –1.5 ± 3.1, to detect such a difference would require 97 in each group suggesting that ondansetron does not affect compliance. Given the very minimal effect on pressure pain threshold, it seems likely that it acts on the local reflexes to reduce urgency rather than a purely sensory effect.

FIGURE 23

Pain threshold in mmHg. This was assessed from the barostat using the ascending method of limits at baseline and after 12 weeks treatment showing no significant effect of either time, treatment or interaction, p = 0.89, 0.4 and 0.4, respectively.

High-resolution colonic manometry

Regrettably very few subjects completed the two manometry visits before and after treatment. Nine completed baseline assessment (5 allocated to ondansetron, 4 to placebo) and there was 1 dropout on ondansetron at week 12 assessment, leaving 4 and 4 to compare. Another complication was the wide variability in the depth of insertion of the manometry catheter. Some were only recording rectal activity [Figure 24, Panel (a)], while others reached the ascending colon [see Figure 24, Panel (b)].

FIGURE 24

Plain X-ray at end of recording showing a very variable positioning achieved. Panel (a) shows recording ports mostly in the rectosigmoid region, while panel (b) shows recording ports in transverse colon missing the rectosigmoid altogether.

Obviously, these data can only be used to develop hypotheses to be tested by further studies but nevertheless there were some interesting results.

We did look at the frequency of HAPCs in the 4 hours of recording, which at baseline (n = 8) were (median, range) 1 (0–3) and on ondansetron (n = 4) were 0, 1, 11, 6 and on placebo, (n = 4) 0, 0, 2, 0. With such small numbers there were no significant differences between treatments, p = 0.25, Mann–Whitney.

Subject 78 showed marked clinical improvement on ondansetron with large falls in urgency (80.7 falling to 10.0), stool frequency (4.9 falling to 2.1) and BSFS (6.5 falling to 3.0). The results are displayed in Figures 25 and 26. There is a striking change in patterns of motility with the baseline showing many rapid moving, almost simultaneous contractions which would be predicted to cause urgent defaecation, which fits perfectly with the patient’s symptoms that were extreme.

FIGURE 25

Manometry: example of response to a meal. Left-hand image shows position of manometry catheter, and the upper tracing shows pressure waves before ondansetron during a meal when contractions appear to rapidly traverse the sigmoid colon and rectum. In contrast, (more...)

FIGURE 26

Manometry: example of effect of ondansetron. Raw data shown in the upper panel demonstrate increased activity in sigmoid colon on ondansetron. The bottom left panel shows an increase in the power in the retrograde contraction (negative phase difference), (more...)

Subject 58 also had good traces at both baseline and on ondansetron. Figure 27 shows an increase in postprandial contraction of ondansetron and a shift in the phase spectrum indicating an increase in retrograde propagated contractions. These would push material from the rectum into the sigmoid, which would reduce urgency induced by liquid reaching the sensitive anorectal margin.

FIGURE 27

Change in WGTT in hours.

Thus, manometry is inconclusive. However, it does provide pilot data and a plausible hypothesis, namely that ondansetron stimulates retrograde activity in the rectosigmoid region, and this could be tested in future studies.

Stool % water

Combining all participants’ stools water content at baseline was mean (SD) 72.4 (6.5) %, n = 48. There were no significant differences between ondansetron or placebo groups, either at baseline or week 12 (Table 30). There was no significant correlation with transit time, r = 0.02, p = –0.88, n = 48.

TABLE 30

Stool % water by visit and allocation

Loose stool responders to ondansetron (defined by < 1 day reduction in days per week with loose stools) tended to have less watery stool % water = 71.3 (6.3), n = 17 versus non-responders, stool % water = 75.5 (3.3), n = 4, but the difference was not significant, p = 0.16.

At baseline, there was no correlation between stool % water and urgency score, Pearson’s r = 0.11, n = 52, p = 0.42, nor between stool % water and days with loose stool, stool consistency nor number of stools per day, p = 0.8, 0.8 and 0.2, respectively.

Faecal proteases

These were assessed at baseline and week 12 and the results are shown in Table 31. There was wide variability but no significant change by visit or treatment, p = 0.8, two-way ANOVA.

TABLE 31

Faecal proteases in trypsin units/mg protein

Contrary to our previous findings,¹² we found no correlation between baseline protease and WGTT, Spearman’s r = –0.0037, p = 0.98, n = 39. However, when looking at the change from baseline in FP and WGTT at 12 weeks, Spearman’s r = mean (95% CI) 0.41 (0.03 to 0.69), there is a statistically significant correlation p = 0.03 (Figure 27).

Change in WGTT in hours [delta WGTT plotted against associated change in FP (delta protease) in trypsin units/mg protein for all subjects]. Spearman’s r = mean (95% CI) 0.32 (0.02 to 0.60), n = 34, p = 0.03.

Faecal bile salts

Faecal bile acids were assessed at baseline as shown in Table 32. Stools with more watery stool had lower total bile acid concentration, correlation with stool % water, Spearman’s r = 0.56, N = 48.

TABLE 32

Faecal bile acids concentration at baseline mean (SD)

As Table 33 shows, at baseline there was wide variability. After 11–12 weeks of treatment, there was a numerical fall in primary bile acids on both ondansetron and placebo with a non-significant rise in the secondary/primary ratio.

TABLE 33

Effect of ondansetron vs. placebo on primary and secondary bile acids in μM per litre mean (SD)

Using the total sample of 51 at baseline, there was no correlation between the secondary/primary bile acids ratio and WGTT, Spearman’s r = –0.08 (0.39–0.75), p = 0.9.