Sequential laxative-probiotic usage for treatment of irritable bowel syndrome: a novel method inspired by mathematical modelling of the microbiome

Abstract

The gut microbiome plays an important role in human health. However, its response to external intervention is complex. A previous study showed that the response to Clostridium butyricum (CB) treatment of irritable bowel syndrome (IBS) is heterogeneous. We proposed that mathematical model simulation of the microbiota may help to optimize the management of IBS-associated microbiota. In this study, a novel mathematical non-extinction and defecation normalized (NEDN) model was generated for stable simulation of the dynamic nature of gut microbiota. In silico simulation revealed that a laxative may create a favourable opportunity for Clostridium cluster XIVa to shift the microbiota. An explorative clinical trial was conducted to compare three CB regimens in an IBS cohort: laxative, interval of 2 weeks and CB administration for 2 weeks (L2P); laxative immediately followed by CB administration (LP) for 2 weeks; and CB administration for 2 weeks (P). The LP regimen optimally relieved the IBS symptoms and shifted the microbiota closer to those of the healthy subjects during 2 weeks of CB intake. These results indicate that integration of biological/mathematical approaches and clinical scenarios is a promising method for management of microbiota. Additionally, the optimal effect of sequential laxative-CB usage for IBS treatment warrants further validation.Clinical trial registration numbers: NCT02254629.Date of registration: October 2, 2014.

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1

The non-extinction and defecation-normalized (NEDN) model of the faecal microbiome and its dynamic properties. (A) The schematic diagram of the NEDN model. (B) The inherent growth rate (vector α) and the inter-genus interaction (matrix β) for the genera in the NEDN model. The microbiome data used to fit this model are publicly available data from Caporaso et al.. (C) Iterative simulation of the NEDN model of the faecal microbiome showing a self-stabilization trend. (D) The NMDS (upper left) and Bray distance-based cluster (lower left) analysis of 1,000 random simulated microbiomes ( +) and 332 experimental microbiomes (dots) in the previously published data. Both types of microbiome were simulated by the NEDN model, and their stable composition was analysed in a similar manner (NMDS, upper right; cluster, lower right). The starting microbiome and the corresponding simulation-stabilized microbiome are shown in the same colour in the NMDS analysis. (E) The compositions of the five stable microbiomes. These five stable statuses correspond to the numbers in the NMDS plot of the stable microbiome (D, upper right), and the genus colour is identical to that in (C).

Figure 2

In silico intervention changes the microbiome based on the NEDN model. The simulation starts with the stable status 5 as the initial microbiome (except for the bottom left that starts with stable status 4). Two types of in silico intervention are included: (1) probiotic-like that increases the abundance of a given genus and (2) laxative-like that reduces the abundance of all genera to 1% of their original value. The probiotic-like intervention is administered either once at day 50 or daily from day 51 to day 64. The laxative-like intervention is administered at day 50 (if applicable). The time frame is extended until a microbiome reaches a stable status.

Figure 3

The flow chart and completeness of the exploratory trial.

Figure 4

The symptoms, QOL, and self-reported relief during 2 weeks of C. butyricum intervention. A-F, the individual Likert-7 score changes during 2 weeks of C. butyricum intervention. The values correspond to the score at the end of week 2 minus the baseline score for the LP and P groups and to the score at the end of week 4 minus the score at the end of week 2 for the L2P group. Lower score corresponds to better condition of the patients. G and H, the changes in the summed symptom score and the impact on QOL during C. butyricum intervention for 2 weeks. Lower values correspond to the better score. F, the self-reported relief score at the end of C. butyricum intervention (week 2 for the LP and P groups and week 4 for the L2P group). Higher values correspond to better condition of the patients. The Kruskal–Wallis p value for overall comparison and the Dunn’s post hoc comparison p value are indicated in each figure. Dunn’s p < 0.025 indicates a significant difference between the two groups.

Figure 5

The microbiome response to the C. butyricum intervention. (A) The comparison of the alpha diversity composition. The p values of the Wilcoxon test or Kruskal–Wallis test with Dunn’s post hoc comparison are indicated in each plot. (B) The NMDS analysis is based on the Euclidean distance. (C) The comparison of the weighted UniFrac distances within the groups. The p values of the Wilcoxon test or Kruskal–Wallis test with Dunn’s post hoc comparison are indicated in each plot. Colours are consistent across the figure. Black colour and grey dots represent the healthy reference group. The L2P group is plotted in orange, the LP group is plotted in red, and the P group is plotted in blue. Green represents the mixed samples from the IBS recipients. The plot patch ‘b’ represents the baseline sample; ‘2’ corresponds to the end of week 2, ‘4’ corresponds to the end of week 4, ‘w’ represents the watery fecal samples after laxative, and ‘h’ corresponds to the healthy reference group.