Associations Between Uraemic Toxins and Gut Microbiota in Adults Initiating Peritoneal Dialysis

Abstract

Declining kidney function contributes to the accumulation of uraemic toxins produced by gut microbiota, leading to the uraemic syndrome. This study aimed to identify associations between uraemic toxins, diet quality, symptoms and the gut microbiota in individuals initiating peritoneal dialysis. A cross-sectional analysis of baseline data from participants in a longitudinal study was conducted. Symptom scores using the Integrated Palliative Care Outcomes Scale-Renal were recorded. Plasma p-Cresyl sulfate, indoxyl sulfate and trimethylamine N-oxide were measured using liquid chromatography-mass spectrometry. Gut microbiota was determined using 16S rRNA sequencing. Multivariate linear models examined associations across the cohort. Data from 43 participants (mean age 61 ± 13 years; 70% male; median eGFR 7 mL/min/1.73 m²) were analysed. Diabetes was the primary cause of kidney disease (51.2%). Patients were classified into 'high' (n = 18) and 'low' (n = 26) uraemic toxin groups using K-means clustering. The 'high' group had a lower eGFR (p < 0.05) but no differences in diet quality or symptom scores. Significant differences in alpha and beta diversity were observed between the groups (p = 0.01). The 'high' group had increased Catenibacterium, Prevotella, Clostridia, and decreased Ruminococcus gnavus abundances. Multivariate models identified 32 genera associated with uraemic toxins, including positive associations of Oscillospiraceae UCG-002 and UCG-005 with p-cresyl sulfate, and negative associations with Actinomyces and Enterococcus. Patients with kidney failure initiating peritoneal dialysis have distinct uraemic toxin profiles, associated with differences in microbial diversity. This phenotype was also associated with differences in residual kidney function but not with diet or symptom severity. Longitudinal studies are required to determine causality and guide therapeutic interventions.

Keywords: chronic kidney disease; diet therapy; gut microbiome; kidney failure; peritoneal dialysis; uraemic toxin.

Figure 1

Prevalence of patient-reported symptoms. Physical symptoms associated with the uraemic syndrome were prevalent in this study population (n = 43). Most commonly reported symptoms were weakness or lack of energy, followed by constipation.

Figure 2

Uraemic toxin clusters. Clustering participants (n = 43) based on their baseline plasma uraemic toxin levels revealed two distinct groups. One cluster, termed the “high uraemic toxin group” (red; n = 18), exhibited elevated levels of uraemic toxins across all measured serum/plasma concentrations. In contrast, the other cluster, classified as the “low uraemic toxin group” (green, n = 26), displayed either negligible or lower concentrations of one or more uraemic toxins. In the low uraemic toxin group, mean serum free PCS = 2.1 (SD 1.3), total PCS = 69.3 (SD 46.1), free IS = 2.8 (SD 1.5), total IS = 40.6 (SD 20.6) and TMAO = 47.7 (SD 40.9). The high uraemic toxin group had mean serum free PCS = 9.0 (SD 6.8), total PCS = 176.3 (SD 104.7), free IS = 9.2 (SD 5.1), total IS = 84.5 (SD 42.4) and TMAO = 81.4 (SD 59.3).

Figure 3

Differences in selected outcomes between uraemic toxin cluster groups. Clinical outcomes, including biochemical markers such as (A) eGFR, (B) creatinine, (C) urea, and (D) total participant-reported symptom scores. Dietary outcomes, including the (E) overall plant-based diet quality index [PDI], (F) healthy PDI, (G) unhealthy PDI, (H) protein and (I) fibre intake, and their (J) ratio. Baseline comparisons between these groups revealed participants in the high group had significantly worse kidney function compared to those in the low group (mean eGFR = 5.89 (SD 2.22) vs. 8.04 (SD 2.44) mL/min/1.73², p < 0.01. Statistical differences were determined using either the Wilcoxon test or t-test, depending on normality and assumptions. ns = not statistically significant; ** p-value < 0.01.

Figure 4

Differences in microbial diversity outcomes between uraemic toxin cluster groups. (A) Differences in alpha-diversity metrics, assessed using the Wilcoxon test (Mann–Whitney U test) (* p < 0.05). (B) PCA plot illustrating differences in microbiota profiles between cluster groups. Analysis is based on relative abundance data at the genus level that was arcsine square-root-transformed. Dissimilarities between samples using the Bray–Curtis dissimilarity index. Relative abundance filtering required taxa to be present in at least 20% of the total samples. p-value was derived from a Permutational Multivariate Analysis of Variance (PERMANOVA), adjusting for age, gender, and diabetes, using 999 permutations. Analysis was completed on available data from N = 34 participants.

Figure 5

Associations between microbial taxa (genus level) and plasma uraemic toxin concentrations and total iPOS-renal symptom score. * Adjusted p-value < 0.20, ** adjusted p-value < 0.05. p-values were derived from linear regression models, where uraemic toxin concentrations and iPOS-renal symptom scores were log-transformed prior to analysis. Models were adjusted for age, gender, diabetes status, and blood albumin levels. Taxa relative abundances were arcsine square-root transformed and filtered to include only taxa present in at least 20% of samples prior to analysis. Analysis was conducted based on completed data from N = 34 participants data. PCS, p-Cresyl sulfate; IS, Indoxyl sulfate; iPOS, Integrated Palliative Care Outcomes Scale.