The Gut Microbiome in Autosomal Dominant Polycystic Kidney Disease: A Cross-Sectional Study

Abstract

Key Points:

1. Patients with autosomal polycystic kidney disease (ADPKD) display relevant alterations in gut microbiome signatures compared with a healthy control cohort.

2. Gut microbiome alterations in patients with ADPKD were associated with specific markers of ADPKD disease progression.

Background: Changes in gut microbiota signatures have been associated with CKD and nephrolithiasis and may thus be a factor explaining variability of outcome in autosomal polycystic kidney disease (ADPKD). We aimed to characterize the intestinal microbiome in a cross-sectional study of patients with ADPKD and to explore the potential effect of microbiome signatures on polycystic kidney disease progression.

Methods: This observational cross-sectional pilot study recruited 25 patients from the German ADPKD Tolvaptan Treatment Registry patient cohort and 12 healthy, age- and sex-matched control participants. The gut microbiome was analyzed by 16S ribosomal RNA gene profiling of stool samples. Bacteria-derived serum uremic toxins were measured using liquid chromatography coupled with tandem mass spectrometry. Microbiome data were correlated with age, kidney function, and markers of polycystic kidney disease progression like Mayo classification and arterial hypertension <35 years of age.

Results: Patients with ADPKD displayed a significantly decreased abundance of Actinobacteria including probiotic Bifidobacteriaceae and significantly increased abundance of Enterobacteriaceae. Those findings were independent of kidney function. Most notably, Streptococcaceae were significantly overrepresented in patients with Mayo classes 1D and 1E compared with 1A–1C. In addition, early onset of hypertension (<35 years of age) was associated with an increased abundance of Proteobacteria and a decreased abundance of Tannerelleaceae. Furthermore, patients with ADPKD revealed an increased abundance of Peptococcaceae with increasing age and declining kidney function. Finally, serum uremic toxin levels were significantly increased in patients with ADPKD, highly correlating with eGFR.

Conclusions: This pilot study suggests relevant changes in gut microbiota signatures of patients with ADPKD, which might be associated with rapid disease progression. These findings indicate that composition of the gut microbiota could potentially contribute to disease progression of ADPKD and the individual disease variability. Further investigation is warranted to assess the gut microbiota as a potential therapeutic target in ADPKD.

Trial registration: ClinicalTrials.gov NCT02497521.

Keywords: ADPKD; cystic kidney; polycystic kidney disease.

Graphical abstract

Figure 1

Microbiome characterization. (A) α-Diversity. Comparison of the total number of measured ASVs in the ADPKD and healthy cohorts using a Mann–Whitney U test. No significant difference could be found (P = 0.7702). (B) β-Diversity. β-Diversity was calculated as weighted UniFrac metric distances between samples, visualized with the PCoA and tested with the analysis of similarity test. No significant difference could be observed (PERMANOVA: P = 0.28; ANOSIM: P = 0.483; R=−0.001202). Axis 1 represents the principal coordinate that explains the largest data change, and axis 2 represents the principal coordinate that accounts for the largest proportion of the remaining data changes. 95% confidence levels assuming normal (--) distribution and 95% confidence ellipses (⎯). (C–E) Analysis of ASVs. Specific ASVs were analyzed and visualized using the LEfSe. An LDA score ≥±2 was defined as significant difference as previously described. Healthy control cohort was compared with the total ADPKD cohort (N=25; C), ADPKD subcohort with eGFR >60 ml/min per 1.73 m² (n=13; D), and ADPKD subcohort with eGFR >70 ml/min per 1.73 m² (n=7; E). All significantly different ASVs are displayed (white: healthy cohort; gray: ADPKD cohort). Negative LDA score defines reduction of abundance, positive LDA score increased abundance of ASVs, NA displays incomplete DNA fragments. Uncultured, displays complete DNA fragments of so-far unidentified bacteria. ADPKD, autosomal polycystic kidney disease; ANOSIM, analysis of similarities test; ASV, amplicon sequence variant; LDA, linear discriminant analysis; LEfSe, LDA effective size algorithm; NA, not applicable; PCoA, principal coordinate analysis; PERMANOVA, permutational multivariate analysis of variance; UniFrac, unique fraction.

Figure 2

Correlation of ASVs with Mayo classification. (A and B) Patients with ADPKD were divided on the basis of Mayo class as a marker of ADPKD disease progression: Mayo class 1A–1C (n=14) versus 1D+1E (n=9). The ASV abundance was analyzed using the LEfSe algorithm and LDA score. Significant differences in abundance were observed for the class of Bacilii with its order of Lactobacillales and its family Streptococacceae. (C) Analysis of the ASV abundance of the class Bacilii, order Lactobacillales and family Streptococacceae upon separate Mayo Classes using a Kruskal-Wallis test. *P values refer to the entire test-model. “Other classification” displays patients with ADPKD where the Mayo classification could not be determined.

Figure 3

Correlation of ASVs with early onset of art. hypertension. (A) Analysis of ASV abundance in patients with ADPKD with and without early onset of art. hypertension (<35 years of age) using the LEfSe algorithm and LDA score. (B) Significant differences in abundance were observed for phylum α-proteobacteria and β-proteobacteria and analyzed by Mann–Whitney U test. aHT, arterial hypertension.

Figure 4

Correlation of ASVs with age and eGFR and analysis of sUT. (A) Patients with ADPKD were divided into two subgroups by age (≤45 years of age, ≥50 years of age) and ASV abundance were analyzed using the LEfSe algorithm and LDA score. (B) Patients with ADPKD were divided into subgroups by age (≤45 years of age, ≥50 years of age) and kidney function (≥60 ml/min per 1.73 m², ≤45 ml/min per 1.73 m²) and ASV abundance were analyzed using the LEfSe algorithm and LDA score. (C) Key figures of the multiple linear regression analysis showing the relationship between the ASV count of the family Peptococcaceae in the gut microbiome of patients with ADPKD and the kidney function (eGFR) as well as the age at the time of stool sampling. The ANOVA states that the regression model as a whole is of significance (P value 0.013). The adjusted R² indicates that 26.6% of the total spread of ASV counts (Peptococcaceae) can be explained by the independent variables age and kidney function at the time of stool sampling. t Tests for the regression coefficients show that the constant and independent variable eGFR at stool sampling do not reach significance at the 95th confidence interval. (D) Partial regression plots depicting the linearity between the dependent variable ASV count of the family Peptococcaceae and the independent variables age at stool sampling and kidney function (eGFR) at stool sampling. The y axis provides the residues for the dependent variable ASV count of the family. The x axis provides the residues for the independent variables age at stool sampling or eGFR at stool sampling regressed to all other independent variables. (E) Analysis of sUT. Analyses of total and free sUT of healthy control and ADPKD cohort using mass spectrometry. pCS, IS and TMAO were measured as replicates and medians in micrometer were analyzed using a Mann–Whitney U test. IS, indoxylsulfate; pCS, p-cresyl-sulfate; sUT, serum uremic toxin; TMAO, trimethylamine N-oxide.