Improved cognition after rifaximin treatment is associated with changes in intra- and inter-brain network functional connectivity

Abstract

Background: Rifaximin is a non-reabsorbable antibiotic which acts at gut level, and improves cognition and inflammatory parameters in minimal hepatic encephalopathy (MHE) patients, but not all patients show the same level of response. This study aims to assess brain activity, both within and between brain networks, following rifaximin treatment, considering the differences between response groups as well.

Methods: Twenty-two healthy controls and 53 patients with cirrhosis (22 without and 31 with MHE, diagnosed by Psychometric Hepatic Encephalopathy Score, PHES) performed psychometric, attention and coordination tests, and blood inflammatory parameters were measured. Resting-state functional magnetic resonance imaging (fMRI) acquisitions were performed on controls and MHE patients. Eighteen MHE patients underwent a rifaximin treatment for 6 months, after which all measures were repeated. fMRI images were analysed and changes after treatment were assessed.

Results: After rifaximin treatment, 13 patients improved their PHES score (Responder patients) while 5 did not (Non-responder patients). No significant decrease in blood ammonia was observed after rifaximin treatment, but there was a decrease in plasma inflammatory cytokines in responder patients. A global effect of rifaximin was detected on the sensorimotor and fronto-parietal networks. Responder patients showed a relative increase of thalamic network connectivity in comparison to non-responder patients. Before treatment, responder and non-responder patients showed connectivity differences in basal ganglia network. The connection of the sensorimotor and thalamic networks between them and with other networks suffered changes after treatment. These connections between networks mostly decreased after treatment. All changes and differences showed a significant level of correlation with the performance of psychometric tests and the blood levels of inflammatory biomarkers.

Conclusions: There was an improvement of the communication between executive, motor and attention-related brain areas, and their functional independence following rifaximin treatment. Patients who respond also show a less deteriorated connection involved in these functions before treatment. Results suggest that the improved inflammatory state of MHE patients, following rifaximin treatment would favour the observed changes in brain function and enhanced cognitive performance.

Keywords: Functional magnetic resonance imaging; Minimal hepatic encephalopathy; Psychometric tests; Rifaximin.

Fig. 1

Flow chart of the steps followed in the selection of the study sample and follow-up that led to the final study groups

Fig. 2

Spatial maps of the 13 resting-state networks identified and analysed in the study. Maps are thresholded at 3 < Z < 10. Images are shown following MNI convention

Fig. 3

Spatial maps of all the clusters in which significant results were observed during the intra-network functional connectivity analysis. Clusters are shown as follows: a MSMN-1, b LFPN-1, c LFPN-2, d LSMN-1, e LSMN-2, f THN-1, g BGN-1. All clusters (green) are pointed (yellow arrows). The resting-state network (red) associated with each cluster is thresholded at 3 < Z < 10. All results were cluster-corrected for multiple comparisons using family-wise error (FWE), in combination with a threshold of p < 0.001 at the uncorrected voxel level. Clusters were considered significant at p < 0.05 after FWE correction. Images are shown following MNI convention

Fig. 4

Boxplots showing z scores of inter-networks connections with significant changes after rifaximin treatment. A–D Considering all patients as one group, before and after treatment. E–H Stratifying patients by response group (Resp: responding patients; NoResp: non-responding patients). Significant pre- and post-treatment differences were analysed using a general linear model similar to a paired t-test and are indicated by (*): *p < 0.05, **p < 0.01. Significant interactions between treatment and response group were analysed using a general linear model similar to a 2-way mixed effect ANOVA and are indicated by (α): α p < 0.05. All p values were corrected for multiple comparisons using FDR, and differences were considered significant at p < 0.05 after correction