Prebiotic Administration to Chronic Kidney Disease Patients Modifies Their Fecal Microbiome and Host Metabolism

Abstract

Objective(s):

Prebiotics are believed to improve gut microbial dysbiosis and dysmetabolism in chronic kidney disease (CKD) patients. However, impact of prebiotics on gut microbial metagenome and dynamic changes in metabolome has not been clearly defined.

Methods:

We conducted a nonrandomized, open-label, three-phase pilot trial to investigate the effect of daily oral prebiotic, oligofructose-enriched inulin (p-inulin), on stool functional metagenome and changes in plasma, urine, and stool metabolites in 13 CKD patients. The study comprised a pretreatment phase (8 weeks), p-inulin treatment phase (12 weeks), and post-treatment phase (8 weeks).

Results:

During treatment phase, there was a significant increase in the abundance of Bifidobacterium adolescentis, Bifidobacterium longum, and Lachnospiraceae species. Microbial pathways related to carbohydrate degradation and amino acid biosynthesis were enriched during the treatment phase, but urea biosynthetic pathway was attenuated. In plasma, metabolic biosynthetic pathways for valine, leucine, and isoleucine were activated during the treatment phase. Microbial genes related to lipid metabolism were enriched during post-treatment. Abundance of several polar and nonpolar lipids were altered in plasma and stool samples during treatment and post-treatment phases. Pathway analysis for lipids indicated suppression of triglyceride biosynthesis in plasma and enhanced triglyceride degradation in stool during the treatment phase. Secondary bile acid levels in plasma, urine, and stool were significantly reduced during p-inulin consumption. Urine levels of indoxyl sulfate and p-cresol sulfate were reduced during treatment phase.

Conclusion(s):

P-inulin administration to CKD patients resulted in a distinct shift in toxin-generating proteolysis to amino acid biosynthesis and favorable changes in lipid metabolism.

Introduction

MICROBIAL COMMUNITIES RESIDING in human gut exist in complex, dynamic consortia with highly interconnected networks of metabolic interactions between themselves and with the host.¹ Metabolic synergy between humans and microbiota exist in metabolic, synthetic, and detoxification processes. This metabolic synergy is perturbed in patients with chronic kidney disease (CKD). Excess delivery of undigested protein to the distal colon and relative paucity of carbohydrate substrates drives enrichment of proteolytic bacteria, resulting in increased generation of proteolytic products such as ammonia, amines, thiols, phenols, and indoles in CKD patients.² A parallel reduction in saccharolytic bacteria results in decreased generation of saccharolytic products such as short-chain fatty acids (SCFA). Thus, delivering complex polysaccharide substrate to colon could restore metabolic synergy in CKD.

The International Scientific Association for Probiotics and Prebiotics defines prebiotic as a substrate that is selectively utilized by host microorganisms conferring a health benefit.³ In this study, we report the myriad microbiome and metabolic effect of a prebiotic, oligofructose-enriched inulin (p-inulin) in patients with CKD using state-of-the-art omics studies.

The primary objective of the Targeting Gut Microbiome with p-inulin in CKD (TarGut-CKD) trial was to test the feasibility of conducting a prebiotic-based clinical trial in patients with CKD. Detailed methods and recruitment was as described by Sohn et al.⁴ We performed shotgun metagenomic sequencing and expanded metabolomic profiling (primary metabolites, biogenic amines, bile acid metabolism, and lipidomics) of stool, urine, and plasma samples collected from TarGut CKD study participants. Metagenomics analysis permits more comprehensive analysis of the microbial community structure and assessment of the functional attributes of the microbiome than 16S sequencing. The aim of the present study is to determine the impact of p-inulin administration on metabolites in CKD patients and associate them with corresponding metabolic pathways in the gut microbial metagenome.

Materials and Methods

The study design, patients’ characteristics, and biosample collection strategies of the TarGut CKD study was as described before⁴ (Supplemental Figure S1). Briefly, in a nonrandomized, open-label, 3-phase pilot trial, we investigated the impact of p-inulin on the gut microbiome, as well as plasma, stool, and urine metabolites in patients with CKD. The study consisted of 3 sequential phases: a pretreatment phase (weeks 1–8), a p-inulin treatment phase (weeks 9–20), and a post-treatment phase (weeks 21–28). Thirteen individuals aged ≥18 years and estimated glomerular filtration rate of 15.0–50.0 mL/min/1.73 m² were included in the study. The primary exclusion criteria included: (1) use of prebiotics, probiotics, or antibiotics within the last 8 weeks; (2) consumption of probiotic yogurt in the past 2 weeks; (3) current infection, inflammatory bowel disease, chronic diarrhea, or C. difficile infection; and (4) a hemoglobin level <9.0 g/dL within the last 4 weeks. Blood, urine, and stool samples were collected weekly after fasting, aligned with the stool collection.

Blood, urine, and stool samples were collected weekly in a fasting state. Blood samples were preserved as serum and plasma aliquots at −80°C. Study participants were provided a commercial “toilet hat” specimen collection kit (Fisherbrand Commode Specimen Collection System; Thermo Fisher Scientific, Waltham, MA, USA) to collect stool samples at home. The stool samples were stored at 4°C, and transported to the clinical trial unit within 24 hours in Styrofoam coolers containing ice packs.

Inulin-type prebiotics are fructans—plant-derived oligo- and polysaccharides with predominantly fructosyl-fructose linkages. These are abundantly found in high-fiber vegetables such as sugar beets, leeks, onions, garlic, and asparagus. In this study, we used oligofructose-enriched inulin (p-inulin, Prebiotin^™) from Jackson GI Medical (Harrisburg, PA) which is is prepared by enzymatically shortening chicory-derived inulin into short-chain oligofructose (95% purity) and mixing it with long-chain inulin. This formulation allows fermentation to occur sequentially, with short-chain oligofructose fermented in the proximal colon and long-chain inulin in the distal colon. P-inulin was administered at 8 g twice daily dissolved in water without a specified volume. If gastrointestinal side effects occurred, the dose could be reduced to 4 g twice daily. Compliance was assessed by sachet count. The Block Food Frequency Questionnaire was completed at baseline and again at weeks 8, 20, and 28.⁵

In this study, we performed shotgun metagenomic sequencing, which provides functional insights at the species and strain level. We performed advanced metabolomic studies that included lipidomics. Integrative analysis of microbiome and metabolome thus offer deeper insights into microbial functions and interactions with the human host.

Shotgun Metagenomic Sequencing

Shotgun DNA sequencing was performed in stool samples collected in weeks 2, 6, 14, 20, 24, and 28 of the study. Briefly, samples were processed using Qiagen’s DNeasy PowerSoil extraction kit. Extracts were quantified using the Quant-iT PicoGreen dsDNA assay (Thermo Fisher Scientific, Waltham, MA). Libraries were prepared using the NexteraXT kit and a HiSeq 1 × 150-cycle v3 kit (Illumina, San Diego, CA) was used to sequence samples. Generation of taxa count tables from raw sequencing data used an approach previously described.⁶ The database used was generated by selecting the first 20 strains per species in RefSeq v87, prioritizing genomes annotated as “Complete Genome,” followed by “Chromosome,” then “Scaffold,” and finally “Contig.”

Untargeted Mass Spectrometry of Hydrophilic and Lipophilic Metabolites

Metabolomic studies were performed in plasma, stool, and urine samples at weeks 2, 4, 6 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, and 28 of the study. Untargeted mass spectrometry–based data acquisition was performed at the UC Davis West Coast Metabolomics Center using 3 different assays: (1) primary metabolism by gas chromatography time-of-flight mass spectrometry (GC-TOF MS) platform⁷; (2) biogenic amines by hydrophilic interaction chromatography–quadrupole time of flight mass spectrometry (HILIC-QTOF MS/MS); and (3) lipidomics by charged surface hybrid reversed phase chromatography (CSH)–QTOF MS/MS.

Primary metabolites were derivatized by methoximation and trimethylsilylation, and data were acquired using 0.5-μl injections on an Agilent 6890 GC with a 30-m-long, 0.25 - m i.d. Rtx-5Sil MS column (0.25-μm 95% dimethyl 5% diphenyl polysiloxane film) with additional 10-m integrated guard column is used (Restek, Bellefonte PA), and a temperature gradient from 50–330°C at a rate of 20°C/s with 1 mL/min helium carrier gas. MS was performed on a Leco Pegasus IV TOF MS at 70 eV at 250°C from 85–500 Da at 17 Hz. ChromaTOF version 4.50 was used for raw data processing, and Binbase was used for data alignment and compound identification.

Hydrophilic interaction chromatography QTOF MS/MS was performed using biphasic extraction of samples with cold methanol, methyl tert-butyl ether (MTBE), and water. Hydrophilic-layer samples were chromatographed using an Agilent 1290 Infinity LC system on a Waters Acquity UPLC BEH Amide column, 130 Å, 1.7 μm, 2.1 mm 3 100 mm at 45°C and a flow rate of 0.6 mL/min using a water/acetonitrile gradient buffered with 10 mM ammonium formate and formic acid mobile phases. One point 7 μl was injected for positive-mode electrospray ionization (ESI) analyses and 5 μl for negative mode ESI. MS was performed on a Sciex TripleTOF 6600 instrument with a 16.8-minute total run time from 50–1200 Da at 500 ms cycle times using gas temperature 300°C, ionization at 4.5 kV with a 10-V declustering potential, and data-dependent MS/MS analyses.

Lipidomics was performed in plasma and stool samples using the lipophilic-layer samples, chromatographed using an Agilent 1290 Infinity LC system on a Waters Acquity UPLC charged surface hybrid column (130 Å, 1.7 μm, 2.1 mm 3 100 mm) at 65°C and a flow rate of 0.6 mL/min using a water/acetonitrile/isopropanol ternary gradient buffered with 10 mM ammonium formate and formic acid. Isopropanol needle wash was essential to eliminate sample carryover. MS was performed on an Agilent 6530 QTOF for positive ESI and an Agilent 6550 QTOF for negative ESI-MS/MS analyses from 65–1700 Da using gas temperature 325°C, ionization at 3.5 kV with a 1 kV nozzle voltage, and data-dependent MS/MS analyses at 2 Hz. For both lipidomics and biogenic amines, data processing was performed using MS-DIAL software⁸ with NIST17 and LipidBLAST MS/MS libraries⁹ or compound identifications, in addition to in-house reference libraries available at MassBank.us. Datasets were normalized to the sum of internal standards. Indoxyl sulfate (IS) and p-cresol sulfate (PCS) were measured by targeted metabolomics.

Short-Chain Fatty Acid Profiling

Plasma SCFA measurement were done at the University of Michigan metabolomic core facility.¹⁰ Briefly, about 100 μL of plasma was transferred to a glass tube and 200 μL of a solution of 30 mM hydrochloric acid plus isotopically-labeled acetate (150 μM), butyrate (10 μM), and hexanoate (2 μM) in water was added to each sample, and then vortexed for 10 seconds. Three hundred μL of MTBE was added to each sample and vortexed.

The samples were centrifuged for 1 minute to separate the solvent layers, and the MTBE layer was then transferred to an autosampler vial for GC-MS analysis. Data were processed using MassHunter Quantitative analysis version B.07.00. SCFAs were normalized to the closest isotope-labeled internal standard and quantified using 2 replicate injections of 5 standards to generate a linear calibration curve, achieving an accuracy of over 80% for each standard.

Statistical Analysis

Descriptive statistics were applied to summarize the features of the study population. Means and standard deviations were used for continuous data, and frequencies and percentages for categorical data. The Shannon index was used for alpha diversity to assess the richness and evenness for each participant across the treatment phase using a mixed-effects model, treating the Shannon indices in each treatment phase as repeated measures. For beta diversity, the Bray-Curtis distance was used for microbial compositions and microbial pathways.

The results were visualized using principal coordinate analysis and analyzed with permutational multivariate ANOVA, with permutations constrained by time to evaluate changes in microbial compositions or pathways throughout the treatment phase. The difference between intra- and interparticipant variability was calculated and compared using an approximate randomization method. Changes in microbial composition or pathways across the treatment phase, relative to baseline (i.e., week 2), were assessed using linear mixed-effects models, treating the changes in each treatment phase as repeated measures. Changes in microbial composition or pathways across the treatment phase, compared to baseline (i.e., week 2), were evaluated using linear mixed-effects models, with changes in each treatment phase treated as repeated measures. Univariate analysis between each species and each pathway was performed using mixed-effects models. Similar analyses were used for metabolic data using the Euclidian distance. The Benjamini-Hochberg procedure was applied to adjust for multiple comparisons and control the false discovery rate. All analyses were conducted in R. Pathway analysis for metabolites was performed using MetaboAnalyst 6.0. Pathway analysis for lipids was performed using Bioinformatics Methodology for Pathway Analysis (BioPAN) comparing pretreatment to the treatment phase.¹¹

Results

Among 17 consenting participants, 13 completed the pretreatment (weeks 1–8), p-inulin treatment (weeks 9–20), and the post-treatment (weeks 21–28) phases of the study. Characteristics of the study participants are shown in Table 1. Four participants withdrew consent during the study. In our study, we observed 83.4% adherence to p-inulin. Analysis of the food frequency questionnaire showed a significant decrease in carbohydrate intake during the treatment phase.⁴ (Supplemental Table 2)

Table 1.

Characteristics of the Study Participants

Parameter	n (%) or Mean ± SD
Age (y)	68.1 ± 11.4
Male	11 (73.3%)
Race
Black or African American	12 (80.0%)
White	3 (20.0%)
BMI (kg/m2)	31.29 ± 7.27
Diabetes mellitus	7 (46.7%)
eGFR (mL/min/1.73 m2)	30.0 ± 9.87
UACR (mg/g)	860 ± 882
Hemoglobin (g/dL)	11.3 ± 1.33
BUN (mg/dL)	39.1 ± 11.9
Creatinine (mg/dL)	3.02 ± 1.35
Calcium (mg/dL)	9.53 ± 0.51
Phosphorus (mg/dL)	4.29 ± 0.84
Glucose (mg/dL)	125 ± 74.9
Medication use
Vitamin D	4 (26.7%)
Lipid lowering agents	8 (53.3%)
Insulin	1 (6.7%)

BMI, body mass index; BUN, blood urea nitrogen; eGFR, estimated glomerular filtration rate; SD, standard deviation; UACR, urine albumin-to-creatinine ratio.

Changes in Gut Microbiome

We performed shotgun metagenomic sequencing on stool samples collected during the pretreatment (weeks 2, 6), treatment (weeks 14, 20), and post-treatment (weeks 24 and 28) phases of the study. The fifty most abundant microbial species across the study phases are shown in Figure 1A. The abundance of Bifidobacterium longum (q = 0.002), Bifidobacterium adolescentis (q = 0.03), and Lachnospiraceae (q = 0.03) species increased during the treatment phase, while the abundance of Adlercreutzia equolifaciens (q = 0.02) and Clostridium bartlettii (q = 0.03) increased during the post-treatment phase compared to baseline (Fig. 1B). The change in alpha diversity assessed by Shannon index at patient level and over the phases of the study was not significantly different (Supplemental Fig. S3A and S3B). Beta diversity assessed by Bray-Curtis distance was also not significantly different over phases (Supplemental Fig. S3C). Between-sample variations were significantly larger than within sample variations (Supplemental Fig. S3D and S3E).

Figure 1.

Shotgun DNA sequencing was performed in stool samples collected in weeks 2, 6, 14, 20, 24, and 28 of the study. (A) Heatmap of the relative abundance of 50 most abundant microbial species in stool samples during the pretreatment (weeks 2 and 6), treatment (weeks 14, and 20), and post-treatment (weeks 24 and 28) phases. Color scale represents relative abundance. (B) Differentially abundant microbial species over phase. Bifidobacterium longum (q = 0.002), Bifidobacterium adolescentis (q = 0.03), and Lachnospiraceae (q = 0.03) species increased during the treatment phase, but Adlercreutzia equolifaciens (q = 0.02) and Clostridium bartlettii (q = 0.03) increased during the post-treatment phase. Y-axis values are shown as proportions.

We examined the metabolic pathways in the microbial genome. Heat map of the relative abundance of 50 most abundant microbial metabolic pathways during the pretreatment, treatment, and post-treatment phases are shown in Supplemental Figure 4.

Metabolomic Studies

Metabolomic studies were performed in the plasma, stool, and urine samples collected during pretreatment (weeks 2, 4, 6, and 8), treatment (weeks 10, 12, 14, 16, 18, and 20), and post-treatment (weeks 22, 24, 26, and 28) phases of the study. The 50 most abundant metabolites in plasma, stool, and urine are shown in Supplemental Figure 5A-C.

Carbohydrate Metabolism

We examined the enrichment of metabolic pathways in the microbial genome during the 3 study phases. Microbial pathways involved in glycolysis and degradation of sucrose, fucose, and rhamnose were enriched during the treatment phase (Fig. 2A). We also observed a decrease in the plasma level of sucrose and urine levels of glucose and fucose (a monosaccharide sugar) during the post-treatment phase (Fig. 2B). Despite the increase in carbohydrate fermentation, plasma levels of SCFA were not significantly altered over the study phases (Supplemental Fig. S6).

Figure 2.

P-inulin on significantly altered the carbohydrate metabolic pathways and associated changes in metabolites in stool, plasma, and urine in CKD patients. Pre = before treatment,during = during treatment, post = post-treatment. (A) Microbial pathways involved in sucrose, fructose, rhamnose, glucose, and fucose degradation; glycolysis; and hexitol fermentation were differentially expressed during the study phases. Sucrose degradation pathway II showed increased activation during treatment phase (q = 0.02). The superpathway of fructose and rhamnose degradation was activated during the post-treatment phase (q = 0.01). Glucose and glucose-1-phosphate degradation exhibited expression during the post-treatment phase (q = 0.007). Glycolysis pathway (q < 0.001) and hexitol fermentation to lactate, format, ethanol, and acetate pathway (q = 0.01) were activated during the treatment phase. Y-axis values are shown as proportions. (B) Urine levels of glucose (q = 0.04) and fucose (q = 0.03), and plasma levels of sucrose (q = 0.03), decreased during the post-treatment phase. Top panel = stool; middle panel = plasma; bottom panel = urine. P-inulin prebiotic, oligofructose-enriched inulin; CKD, chronic kidney disease.

Protein Metabolism

The urea cycle is the primary pathway for nitrogen disposal in humans. During the treatment phase the urea cycle pathway was reduced (Supplemental Fig. 7A), but metabolic pathways related to arginine, ornithine, lysine, threonine, methionine, isoleucine, alanine, serine, and glycine biosynthesis were enriched in the bacterial metagenome (Supplemental Fig. 7B). Pathway analysis for metabolites showed that metabolic pathways for valine, leucine, and isoleucine biosynthesis were enriched in the plasma samples during the treatment phase, with threonine, leucine, and valine as significant compounds within that pathway (Supplemental Fig. 7C).

Lipid Metabolism

Gut microbiota is a key regulator of lipid metabolism. In our study, 4 microbial pathways related to lipid metabolism were enriched during post-treatment phase (Fig. 3A). These include unsaturated fatty acids (FAs) biosynthesis, phospholipid remodeling, phosphatidylcholine (PC) acyl editing, glyoxylate cycle, and FA degradation.

Figure 3.

Effect of p-inulin on lipid metabolism. Pre = before treatment; during = during treatment; post = post-treatment. Y-axis values are shown as proportions. (A) Differentially abundant pathways over phase (FDR <0.1) All these pathways were enriched during the post-treatment phase compared to the pretreatment phase. (B) Intersubject variability for plasma lipids acquired under ESI (+) mode. Levene test of homogeneity: P = 0.002. Post hoc: Bonferroni correction. Intersubject variability is relatively stable at each time point. (C) Intersubject distance for plasma lipids acquired under ESI (−) mode. Levene test of homogeneity: P < 0.001. Post hoc: Bonferroni correction. Intersubject variability is relatively stable at each treatment phase. * indicates P < 0.05; ** indicates P < 0.01; *** indicates P < 0.001. Boxplots of log-transformed significant plasma and plasma and stool lipids acquired in ESI (+) mode (FDR <0.01) and ESI (−) mode (FDR <0.05). The unique ID of the metabolite is shown; first number indicates the number of carbon atoms in the chain, and subsequent after the colon signifies the presence of one double bond. Pre = before treatment; during = during treatment; post = post-treatment. (D) Differentially altered lipids that increased during the treatment phase and/or post-treatment phase in plasma and stool belong to the class of FAs, LPC, PC, PE, SM, MGDG, and β sitosterol. (E) In the stool, diacylglycerol and nonadecenoic acid levels increased but tetracosapentaenoic acid decreased during the treatment phase. P-inulin, prebiotic, oligofructose-enriched inulin; ESI, electrospray ionization; FA, fatty acid; FDR, false discovery rate; LPC, lysophosphatidylcholine; MGDG, monogalactosyldiacylglycerol; PC, phosphatidylcholine; PE, phosphatidylethanolamine; SM, sphingomyelins.

Intersubject Euclidean distance—a measure of variance in metabolite abundance—for plasma lipids acquired under ESI positive mode and negative mode are shown in Figure 3B and C. Significantly altered plasma lipids acquired under ESI positive and negative modes in plasma and stool are presented in Figure 3D and E. Differentially altered lipids during the study phases in plasma and stool belong to the class of FAs, lysophosphatidylcholine (LPC), phosphatidylcholine (PC), phosphatidylethanolamine (PE), sphingomyelins, and monogalactosyldiacylglycerol. They showed an increase during the treatment phase and/or in post-treatment phase.

Pathway analysis for lipids was performed using BioPAN comparing pretreatment to the treatment phase.¹¹ The most active lipid class reactions in plasma during the treatment phase were diacylglycerol (DG) > PE (Z = 2.33, predicted gene is CEPT1) and O-LPC->O-PC (Z = 1.725, predicted gene is LPCAT1) (Fig. 4A). Pathway analysis of metabolite levels in plasma revealed that biosynthesis of PE (DG > PE) was most activated pathway, meanwhile triglyceride (TG) biosynthesis (DG[34:1] > TG[56:6], DG[34:3] > TG[53:4], DG[34:2] > TG[53:3], DG[36:1] > TG[55:1]) and biosynthesis of PE (DG(38:6)->PE(38:6)) were the most suppressed pathways (Fig. 4B). The only significantly active lipid class reaction in stool was lysophosphatidylethanolamine > PE (Z = 1.667, predicted genes are MBOAT1, MBOAT2, LPCAT3, LPCAT4) (Fig. 4C). TG catabolism (TG[52:3]- > DG[36:3], TG[50:2] > DG[34:2], TG[52:2] > DG[36:2], TG[52:1] > DG[36:1]) was the most active pathway in the stool metabolite pathway analysis, meanwhile TG biosynthesis (DG[34:1] > TG[50:2], DG[34:2]->TG[54:7]) and biosynthesis of PC (PE[36:1]->PC[36:1]) were the most suppressed pathways (Fig. 4D). No lipid classes were significantly active in the urine samples.

Figure 4.

BioPAN was used to perform pathway analysis. This permits exploration of systematic changes in lipid pathways at different levels. Both pathways are highlighted, and changes in gene activity could also be predicted. Active pathways are green; suppressed pathways are purple. FA, fatty acid; LPC, lysophosphatidylcholine; MGDG, monogalactosyldiacylglycerol; PC, phosphatidylcholine; PE, phosphatidylethanolamine; SM, sphingomyelins; (A and B) Plasma lipid pathway; (C and D) stool lipid pathway. Circular node shape: glycerolipids and glycerophospholipids; triangular node: FAs; purple edge: negative Z-score; green edge: positive Z-score. (A) The most active lipid class reactions in plasma were DG > PE (Z = 2.33, predicted gene is CEPT1) and O-LPC->O-PC (Z = 1.725, predicted gene is LPCAT1). (B) Lipid class pathways in plasma samples: biosynthesis of PE (DG > PE) was the most activate pathway, meanwhile triglycerol (triglyceride [TG]) biosynthesis (DG(34:1) > TG(56:6), DG(34:3) > TG(53:4), DG(34:2)-D.>TG(53:3), DG(36:1) > TG(55:1)) and biosynthesis of PE (DG(38:6)->PE(38:6)) were the most suppressed pathways in plasma. (C) The only significantly active lipid class reaction in stool was lysophosphatidylethanolamine (LPE) > PE (Z = 1.667, predicted genes are MBOAT1, MBOAT2, LPCAT3, LPCAT4). (D) Lipid class pathways in stool samples. TG catabolism (TG[52:3]->DG[36:3], TG(50:2)->DG(34:2), TG(52:2) > DG(36:2), TG(52:1) > DG(36:1)) was the most active pathway, meanwhile TG biosynthesis (DG[34:1] > TG[50:2], DG(34:2) > TG(54:7)) and biosynthesis of PC (PE (36:1) > PC(36:1)) were the most suppressed pathways. BioPAN, Bioinformatics Methodology for Pathway Analysis; DG, diacylglycerol.

Bile acids play a vital role in lipid metabolism. We observed a significant reduction in the levels of secondary bile acid, deoxycholic acid in stool, plasma, and urine during the treatment phase (Fig. 5).

Figure 5.

Log-transformed intensity of deoxycholic acid samples (top panel = stool; middle panel = plasma; bottom panel = urine) before, during, and after treatment.

Changes in Purine Metabolism

Microbial pathways involved in the degradation of purine nucleotides were detected. PWY0–1297 superpathway of purine deoxyribonucleosides degradation and PWY-6353 purine nucleotides degradation II (aerobic) were increased during the post-treatment phase (Supplemental Fig. 8A). Relative abundance of adenine and its breakdown product, uric acid, decreased in the post-treatment phase (Supplemental Fig. 8B). Pathway analysis for metabolites showed that purine metabolic pathway in the stool was significantly enriched, with xanthosine, Inosine, adenine, and urate as significant compounds within the pathway (Supplemental Fig. 8C). A pyrimidine metabolite pathway in the urine was also significant, with L-glutamine, uridine, and orotate as significant compounds within the pathway (Supplemental Fig. 8D).

Microbial Metabolites

Urine levels of IS and PCS were reduced during the treatment phase without significant changes in plasma or stool (Supplemental Fig. 9A and B). The plasma levels of IS during treatment phase (0.06 ± 0.09 μmol/L) was not significantly different from pretreatment (0.17 ± 0.68 μmol/L) or post-treatment (0.06 ± 0.11 μmol/L) concentrations (P = .996). Similarly, PCS levels did not change significantly during treatment phase (29.73 (48.93)) compared to pretreatment (76.46 ± 217.70 μmol/L) and post-treatment (27.73 ± 37.54 μmol/L) values (P = .903). 5-hydroxy-3-indoleacetic acid, a byproduct of bacterial metabolism of serotonin was also reduced. In contrast, 3-(4-hydroxyphenyl) propionic acid, a metabolite produced by gut bacteria through the breakdown of dietary polyphenol, increased in the stool samples during the treatment phase (Supplemental Fig. 9C-E). 3-(3-hydroxyphenyl)-3-hydroxypropionic acid, a metabolite primarily produced by gut bacteria through the metabolism of aromatic amino acids, was higher in the urine during the treatment phase.

Heatmap of the correlations between microbial species and plasma metabolites that have at least one significant pairwise correlation is shown in Supplemental Figures S10-S12.

Discussion

We conducted a nonrandomized, open label, pilot trial to understand the effects of p-inulin administration on bacterial and host metabolism in patients with CKD. We combined shotgun metagenomic sequencing with metabolomic studies performed in plasma, stool, and urine samples during the pretreatment phase, treatment phase, and post-treatment phase of the study. Our results show that p-inulin consumption promotes abundance of specific bacteria that are known to be depleted in patients with CKD and also alters the composition and functional metagenome of gut microbiota, resulting in significant changes in metabolic functions. P-inulin administration was associated with enrichment of metabolic pathways related to carbohydrate fermentation and amino acid biosynthesis in microbiota and human host (Supplementary Fig. S13). Microbial pathways involved in lipid metabolism were also altered by p-inulin. Composite metabolites pathway analysis showed that biosynthesis of PEs was activated, but that of TG was suppressed in the CKD patients during p-inulin treatment. Thus, p-inulin consumption results in complex alterations in microbial metabolism as well as host cometabolism. We found that the impact of p-inulin consumption extends beyond carbohydrate metabolism to amino acid biosynthesis, lipid metabolism, and bile acid metabolism.

Inulin-type fructans are nondigestible carbohydrates that are fermented by selected bacteria, including bifidobacteria. In our study, although overall microbial composition and diversity was not altered by p-inulin consumption, the abundance of specific saccharolytic bacteria such as bifidobacteria was increased. The end products of inulin fermentation are SCFA, carbon dioxide, hydrogen, and/or methane. P-inulin promotes growth of bifidobacteria, reduces endotoxin generation, attenuates inflammation, decreases generation of uremic toxins, and improves metabolic function.^3,12

Many fiber-degrading bacteria are also SCFA-producing, such as Anaerostipes, which was increased in our study.^13–15 However, we did not find a significant change in plasma SCFA levels. This might be because only a small fraction of SCFA produced by the gut microbiota reaches the systemic circulation.¹⁶

Integral metabolism of amino acids by gut microbiota results in the production of a wide range of metabolites, including ammonia, amines, nitrogen compounds, phenols, and precursors to SCFA.¹⁷ We observed that the amino acid biosynthetic pathways were activated in the microbiome and in the host during the treatment phase. There was specific activation of valine, leucine, and isoleucine biosynthesis. Valine, leucine, and isoleucine are branched chain amino acids that are essential for protein synthesis. These amino acids are synthesized by bacteria through conserved pathways that are absent in mammals.¹⁸ The aspartate superpathway was altered by p-inulin supplementation. Aspartate is produced through a transamination reaction involving oxaloacetate, a key intermediate in the Krebs cycle. The biosynthesis of aspartate is at the interface between amino acid metabolism and carbohydrate metabolism. Our results indicate that delivery of carbohydrate as p-inulin shifts the microbial/host co-metabolism from proteolysis to carbohydrate fermentation and amino acid synthesis.

Lipids are small molecules that are essential for many biological functions, including cell membrane assembly, cell signaling, and energy metabolism. Patients with CKD have dyslipidemia, with elevated blood TG levels, increased low-density lipoprotein levels, and altered functionality of high-density lipoproteins.¹⁹ Disturbances in lipid metabolism are evident in early stages of CKD and parallels loss of kidney function.²⁰ In our study, we found significant changes in abundance of complex lipids during the treatment phase, and many of these changes were sustained during the post-treatment phase.^21,22 This effect is likely attributable to p-inulin–induced shifts in the gut metagenome, which continue to modulate host metabolic pathways. However, the reversal of certain lipid metabolic changes after supplementation ceased suggests that these effects are dependent on ongoing prebiotic intake; the decline in SCFA-producing bacteria and their metabolites diminishes their regulatory influence on lipid metabolism.

Qian et al reported that supplementation with B. adolescentis, as a probiotic, was associated with a decrease TG and low-density lipoprotein-C levels.²³ We also noted significant changes in lipid metabolic pathways in microbial metagenome during the post-treatment phase. The gut microbiota has been shown to affect lipid metabolism and lipid levels in blood and tissues in mouse models and in human subjects.^24,25 Gut microbiota–produced lipids may enter the circulation and affect phenotypic changes of the host.²⁶ We used BioPAN to identify metabolomic pathways and reactions, which combines existing knowledge of lipid metabolism with a statistical analysis functionality to identify activated or suppressed pathways.¹¹ BioPAN also permits prediction of the most likely lipid transforming genes modulated by the intervention. In our study, lipid metabolic pathway analysis suggested that TG biosynthesis was suppressed in plasma and its degradation was enhanced in the stool samples during the treatment phase compared to pretreatment phase.

Primary bile acids, including cholic acid and chenodeoxycholic acid, are synthesized in the liver from cholesterol, which are further conjugated to glycine or taurine and go through the process of enterohepatic circulation. Gut bacteria in the colon contribute to the modification of primary bile acids to generate secondary bile acids, including deconjugation and dehydroxylation, which increase their hydrophobicity and toxic and metabolic effects.²⁷ Among patients with moderate to severe CKD, elevated serum deoxycholic acid level is associated with higher coronary artery calcification volume score and lower baseline bone minder density.²⁸ During p-inulin treatment, there was a significant reduction in the level of deoxycholic acid in stool, plasma, and urine during the treatment phase.

In our study, we observed changes in several bacteria-related metabolites, including 5-hydroxy-3-indole acetic acid (IAA), a derivative of indole-3-acetic acid known as a bioactive tryptophan metabolite. Gut bacteria such as Bacteroides have been reported to convert tryptophan to indole-3-acetic acid. Indole-3-acetic acid exhibits a dual role, serving as a beneficial microbial metabolite under healthy conditions but acting as a uremic toxin in CKD. The observed shifts in 5-hydroxy-3-indoleacetic acid suggest alterations in microbial tryptophan metabolism, particularly via Bacteroides-mediated IAA production. As ligands for the aryl hydrocarbon receptor, these metabolites influence intestinal barrier integrity, immune regulation, and inflammation, while in CKD, their accumulation may exacerbate cardiovascular and renal complications.^29,30

3-(4-hydroxyphenyl) propionic acid, a major microbial phenolic metabolite produced by Clostridium sporogenes, which may function as a signaling molecule influencing host’s physiological regulation.³¹ It can also be derived from apigenin-7-glucoside, which is a flavonoid contained in red wine, artichokes and chamomile.³² 4-HPPA exhibits antioxidant, anti-inflammatory, and gut barrier protective properties, making it potentially beneficial in CKD by counterbalancing the harmful effects of uremic toxins.^29,33

3-(3-hydroxyphenyl)-3-hydroxypropionic acid is considered an abnormal phenylalanine metabolite generated through alternative microbial metabolism, particularly by Clostridia spp. Its presence may indicate dysbiosis or atypical microbial processing of phenylalanine.³⁴ Finally, we observed reduced urinary levels of IS and PCS during the treatment phase, without significant changes in plasma concentrations. These results suggest reduced generation of IS and PCS during the treatment phase.

Purines and pyrimidines are the nitrogenous bases of nucleotides. Some gut microbiota belonging to the Bacteroides, Escherichia coli, and Lactobacillus genera possess enzymes capable of degrading purines and pyrimidines. In a urate oxidase knockout mouse model of hyperuricemia, p-inulin administration reduced uric acid level.³⁵ In our study, we noted an enrichment of microbial purine degradation path-ways during the post-treatment phase and a reduction in relative abundance of adenine and uric acid in the stool samples of CKD patients. Metabolic pathway analysis showed that the purine metabolic pathway in the stool was significantly expressed. Alterations in purine metabolism during post-treatment phase could be due to lag time between change in microbiome and the metabolic impact on the host.

The strengths of the study include a clear study design, repeated sampling strategy within each phase of the study, and advanced metabolomics performed simultaneously in blood, urine, and stool samples. We show the association between microbial functional metagenomics and alteration of metabolite levels in CKD patients. Previous studies have largely focused on specific targeted metabolites and did not perform advanced metagenomic pathway analysis and integrated microbial and metabolomic analysis as in the present study. The study is not adequately powered to detect changes in some metabolites in plasma and stool. This is not a randomized controlled trial, but patient served as his own control and the postintervention phase allowed us to study the persistence of microbial and metabolomic signatures after stopping p-inulin. While the quantification of SCFA levels in stool samples could provide valuable insights into colonic microbial activity, we were not able to perform these measurements due to technical reasons. We acknowledge this as a potential limitation of the study. We observed that the mean energy intake among the study participants was low, which may reflect underreporting or limitations of the assessment tools used, representing another potential limitation of the study.

In conclusion, our study shows that p-inulin consumption by patients with CKD resulted in a plethora of metabolic effects that extends beyond carbohydrate and protein metabolism, affecting metabolism of lipids, bile acids, and nucleic acids. We also observed similar functional changes in microbial metabolic pathways, indicating that these are the results of host–microbial cometabolism. Several metabolites derived from protein fermentation by gut microbiota contribute to the uremic syndrome. Our results suggest that delivery of complex carbohydrates as a prebiotic to the distal colon shifts from toxin generating protein putrefaction process to amino acid biosynthesis. These exciting findings need to be tested in adequately powered large scale randomized controlled trial. Furthermore, studies are needed to evaluate the functional consequences of oral p-inulin to CKD patient health outcomes.

Practical Application

Excess protein breakdown by colonic microbiota results in generation of harmful uremic retention solutes. This study shows that p-inulin consumption by CKD patients results in structural and functional changes in gut microbiota and host cometabolism. Metabolomic studies show a shift in shift from toxin-generating proteolysis to amino acid biosynthesis in CKD patients during p-inulin treatment. Lipidomic studies revealed suppression of TG biosynthesis in plasma and enhanced TG degradation in stool. Plasma, urine, and stool levels of secondary bile acids were reduced. Thus, support that prebiotics could be a safe adjuvant therapy in the management of CKD.

Study Approval

The study abided by guidelines laid out by the Declaration of Helsinki and the Declaration of Istanbul. The trial was registered at Clinicaltrials.gov (NCT03348592). The study was approved by the Institutional Review Boards and the Data Coordinating Center. All participants provided written informed consent before initiating study procedures.

Supplementary Material

Supplementary Data

Supplementary data related to this article can be found at https://doi.org/10.1053/j.jrn.2025.10.015.

Consortium/study groups:

Pilot Studies in CKD Consortium.

Data Availability

Enquiries about data availability can be directed to the corresponding author.