The fecal microbiome and serum concentrations of indoxyl sulfate and p-cresol sulfate in cats with chronic kidney disease.

BACKGROUND: Intestinal dysbiosis has been documented in humans with chronic kidney disease (CKD) and is thought to contribute to production of the uremic toxins indoxyl sulfate (IS) and p-cresol sulfate (pCS). Characteristics of the fecal microbiome in cats with CKD and correlation to serum concentrations of uremic toxins are unknown.
OBJECTIVES: To characterize the fecal microbiome and measure serum IS and pCS concentrations of cats with CKD in comparison to healthy older cats. ANIMALS: Thirty client-owned cats with CKD (International Renal Interest Society stages 2-4) and 11 older (≥8 years) healthy control cats.
METHODS: Prospective, cross-sectional study. Fecal samples were analyzed by sequencing of 16S rRNA genes and Escherichia coli quantitative PCR (qPCR). Serum concentrations of IS and pCS measured using liquid chromatography tandem mass spectrometry.
RESULTS: Cats with CKD had significantly decreased fecal bacterial diversity and richness. Escherichia coli qPCR showed no significant difference in bacteria count between control and CKD cats. Cats with stage 2 (P = .01) and stages 3 and 4 (P = .0006) CKD had significantly higher serum IS concentrations compared to control cats. No significant difference found between stage 2 and stages 3 and 4 CKD. The pCS concentrations were not significantly different between CKD cats and control cats. CONCLUSIONS AND CLINICAL IMPORTANCE: Decreased fecal microbiome diversity and richness is associated with CKD in cats. Indoxyl sulfate concentration is significantly increased with CKD, and cats with stage 2 CKD may suffer from a similar uremic toxin burden as do cats with later stage disease.

analysis of similarity

blood urea nitrogen

chronic kidney disease

erythropoietin

International Renal Interest Society

indoxyl sulfate

linear discriminant analysis effect size

operational taxonomic units

principal coordinate analysis

p‐cresol sulfate

quality assurance/quality control

Quantitative Insights Into Microbial Ecology

quantitative PCR

symmetric dimethylarginine

urine specific gravity

INTRODUCTION

The wide array of microorganisms that reside in the intestinal tract (gut microbiome) play an important role in maintaining host health, and alteration of the gut microbiome has been associated with many illnesses in humans including chronic kidney disease (CKD). The uremia associated with CKD has been shown to negatively impact the gut microbiome in humans and rats causing intestinal dysbiosis.1, 2 In humans and rats with CKD, these changes shift the microbiota to a less diverse community that is dominated by certain bacterial families.1 Other instigators of intestinal dysbiosis in CKD patients include frequent use of antibiotics,3 dietary changes such as decreased fiber intake, and use of phosphate binders.2, 4, 5, 6

The intestinal dysbiosis associated with CKD has been shown to exacerbate the accumulation of colonic‐derived uremic toxins such as indoxyl sulfate (IS) and p‐cresol sulfate (pCS) because toxin‐producing bacterial families may be more prevalent in dysbiosis.5, 7, 8, 9 p‐Cresol is generated from the partial breakdown of tyrosine and phenylalanine by many intestinal obligate or facultative anaerobes including the genera Bacteroides, Lactobacillus, Enterobacter, Bifidobacterium, and Clostridium.5, 7, 10 Indoles are produced by the metabolism of dietary tryptophan by tryptophanase in intestinal bacteria such as Escherichia coli (E. coli), Proteus vulgaris, and Bacteroides spp.5, 7, 11 Indole and p‐cresol are absorbed into the blood, metabolized to IS and pCS in the liver, and excreted in the urine.5, 7 In people, the accumulation of IS and pCS in CKD has been associated with progression of the disease by inducing inflammation and damaging renal tubular cells,12, 13 promoting renal fibrosis,14, 15 and by stimulating progression of glomerular sclerosis.16 These uremic toxins also contribute to morbidity and mortality by impairing the nervous system,17 lowering erythropoietin (EPO) production,18 altering bone turnover,19, 20 and increasing the risk of cardiovascular disease.21, 22, 23

In veterinary medicine, IS concentrations have been correlated to the severity of CKD in cats and dogs,24 and were found to be an independent predictor for disease progression.25 However, the role of pCS has not been explored in veterinary medicine. Various studies have evaluated the gastrointestinal microbiome of cats in health and in disease,26, 27, 28, 29, 30, 31, 32, 33 but the fecal microbiome in cats with CKD and its association with IS and pCS have not been characterized. The primary aim of our study was to characterize the fecal microbiome of cats with CKD in comparison to healthy control cats. A secondary aim was to measure serum IS and pCS concentrations in the same population of cats. We hypothesized that cats with CKD would have an altered fecal microbiome and higher serum concentrations of IS and pCS when compared to healthy control cats.

MATERIALS AND METHODS

Study design and selection of animals

This prospective cross‐sectional study was performed at the Colorado State University Veterinary Teaching Hospital. All client‐owned cats with CKD that presented between August 2016 and August 2017 were eligible for enrollment into the study. Cats were considered to have CKD based on a serum creatinine concentration >1.6 mg/dL and urine specific gravity (USG) <1.035 or an increased serum creatinine concentration on at least 2 time points (over at least 3 months) in addition to an increased symmetric dimethylarginine (SDMA) concentration (>14 ug/dL).34 To be eligible for inclusion, CKD and control cats underwent a thorough evaluation that included a review of past medical record, complete physical examination (including body condition score), minimum database consisting of CBC, serum biochemistry panel, urinalysis, serum total thyroxine concentration, blood pressure, fecal flotation, SDMA (when USG > 1.035 and persistent serum creatinine concentration > 1.6 mg/dL), and urine protein:creatinine ratio (if urine dipstick testing detected at least 1+ protein). Physical examination and Doppler blood pressure were performed by a single trained individual for consistency. The CKD cats were staged based on the International Renal Interest Society (IRIS) guidelines.34 A control population of older (≥8 years) clinically healthy cats was recruited from employees, students, and staff of the veterinary teaching hospital. Cats were considered healthy based on unremarkable client history and past medical record, physical examination, and normal laboratory test results including serum creatinine concentration <1.6 mg/dL and USG >1.035.

Exclusion criteria included a history of antibiotic, probiotic, or antacid administration (eg, omeprazole, famotidine) <6 weeks before enrollment. Other medications or supplements including mirtazapine, maropitant, PO potassium supplementation, glucosamine/chondroitin joint supplements, and transdermal methimazole were allowed. In addition, cats were excluded if they had a history of uncontrolled hyperthyroidism and known or suspected gastrointestinal disease including intestinal parasitism and food‐responsive chronic enteropathy. A client questionnaire was provided to the owner and the following information was obtained: diet, current medications or supplements, medications or supplements administered in previous 3 months, appetite and fecal score at the time of enrollment, and frequency of vomiting. A table with descriptions of appetite, fecal, and vomiting scores is provided in Supporting Information (Table S1). The project was approved by the Institutional Animal Care and Use Committee at Colorado State University and all owners gave written informed consent before participation.

Fecal microbiome analysis

A fresh fecal sample was collected by the owner and placed on ice until frozen within 24 hours of collection. The samples were stored at −80°C until analysis of the fecal microbiome by sequencing of the 16S ribosomal RNA (16S rRNA) gene as previously described.35 Briefly, DNA was extracted following the manufacturer's instructions (Mo Bio Power soil DNA isolation kit, Mo Bio Laboratories, Carlsbad, California). The DNA then was amplified and sequenced (Illumina, Inc., San Diego, California) using primers 515 F (5‐GTGC CAGCMGCCGCGGTAA‐3) to 806 R (5‐GGACTACVSGGGTATCTAAT‐3) at MR DNA Laboratory (Shallowater, Texas). Sequences were processed and analyzed using Quantitative Insights Into Microbial Ecology (QIIME) v 1.9.36 Sequences were filtered for chimeras with USEARCH. The remaining sequences were clustered into operational taxonomic units (OTUs) by using an open reference approach in QIIME against the 97% clustered representative sequences from the Greengenes v 13.8 database.37 Before downstream analysis, sequences that were assigned to chloroplast, mitochondria, unassigned, and low abundance OTUs, containing <0.01% of the total reads in the dataset were removed. The samples were rarefied to an equal depth of 80 875 sequences per sample to account for unequal sequencing depth. The rarefaction depth was chosen based on the lowest sequence depth of samples. The sequences were deposited in the National Institutes of Health Sequence Read Archive under accession number SRP117611.

To evaluate bacterial species diversity within the fecal samples, alpha diversity was measured with the Chao1 and Shannon diversity indices, and observed OTUs metrics. Chao1 and Shannon indices take into account the number of observed OTUs and the evenness of bacterial species distribution within the fecal sample. Species richness can be defined as the number of unique OTUs and can be used as a proxy for bacterial species within the fecal sample.

To compare the bacterial communities among samples, beta diversity was evaluated with the phylogeny‐based weighted and unweighted UniFrac38 distance matrix and visualized using Principal Coordinate Analysis (PCoA) plots. Weighted UniFrac gives importance to the abundance of a particular OTU present in the fecal sample and unweighted does not. Linear discriminant analysis effect size (LEfSe) was used to elucidate bacterial taxa and genes that were associated with healthy control cats or CKD cats. The LEfSe was used in the Galaxy workflow framework (http://huttenhower.sph.harvard.edu/galaxy/) with the parameters set at α = 0.01, linear discriminant analysis score = 3.0. Quantitative PCR (qPCR) was used as described previously39 to evaluate E. coli bacterial counts between healthy control cats and CKD cats.

Assays for serum indoxyl sulfate and p‐cresol sulfate concentrations

Blood was collected in sterile non‐heparinized tubes and centrifuged at 5000 rpm for 5 minutes. Serum was harvested and samples were frozen and stored at −80°C until analysis by the Pharmacology Core at Colorado State University. Total IS and pCS serum concentrations were determined by liquid chromatography tandem mass spectrometry modified from a published method40 using a Waters Sunfire C8 5 μm, 4.6 × 50 mm column with a Phenomenex C18 Filter Frit Guard Cartridge on the Agilent 1200 Series Binary Pump SL high‐performance liquid chromatography system coupled to the 3200 Q‐TRAP triple quadrupole mass spectrometer (Applied Biosystems Inc., Foster City, California).

The calibration curves were constructed by spiking standard IS (Alpha Asear, Haverhill, Massachusetts) and pCS (Tokyo Chemical Industries, Tokyo, Japan) solutions into normal serum collected from a clinically healthy cat that had normal kidney function. Narageninine (1000 ng/mL) was used as the internal standard (Sigma‐Aldrich, St. Louis, Missouri). Assay performance for each batch was assessed utilizing at least 10% quality assurance. Quality assurance/quality control (QA/QC) samples were dispersed among unknown samples at concentrations representing low, mid, and high regions of the calibration curves, with acceptable batches showing <25% of QA/QC samples outside of the accepted level of 85% accuracy. The accuracy plus/minus coefficient of variation of QA/QC samples among the batches analyzed for this study were 93.7 ± 4.8% for IS and 92.5 ± 4.0% for pCS. Standard curves in spiked cat serum were linear over the range of 100‐50 000 ng/mL for both IS and pCS. The linearity of the curves was greater than r 2 = 0.99 using 1/x 2 weighting. As measured in cat serum, the lower limit of quantitation for this assay was based on the level of detection with >85% accuracy and a coefficient of variation <15%, and was determined to be 250 ng/mL for IS and 100 ng/mL for pCS.

Statistical analysis

For statistical analysis between the stages of CKD, stages 3 and 4 CKD cats were combined given the few stage 4 cats that were enrolled in the study. For all analyses, a value of P < .05 was considered significant. Normality was assessed by the Shapiro‐Wilk test.

Analysis of similarity (ANOSIM) test within PRIMER 6 software package (PRIMER‐E Ltd., Luton, UK) was used to analyze significant differences in microbial communities between healthy control cats and CKD cats. R values (range −1.0 to 1.0) were computed to evaluate the strength of the factors on the fecal samples. R values closer to 1.0 indicate a difference between the study groups. The remainder of the microbiome data was analyzed using commercial software (JMP Pro 11, SAS Software Inc., Cary, North Carolina). Because most datasets did not meet the assumptions of normal distribution, comparisons between healthy and disease groups were determined using nonparametric Kruskal‐Wallis tests (healthy cats and stage of CKD [stage 2, stages 3 and 4]) or a Mann‐Whitney U test (healthy control cats versus CKD [all stages combined]) in Prism software (Prism 7, Prism Graphpad Inc., La Jolla, California). The resulting P values of the Kruskal‐Wallis tests or Mann‐Whitney U test were adjusted for multiple comparison using Benjamini and Hochberg's false discovery rate41 at each taxonomic level. Post hoc Dunn's multiple comparison test was used to determine the bacterial taxa that were different between the groups. The OTU tables generated also were uploaded into Calypso,42 a web‐based application to evaluate the correlations between bacterial taxa and IS and pCS serum concentrations.

For IS and pCS concentrations, data were analyzed with Prism. Median IS and pCS concentrations between healthy control cats and CKD cats were compared using the Mann‐Whitney U test or Student t test depending on normality. Healthy control cats and the stage of CKD (stage 2, stages 3 and 4) were compared using Kruskal‐Wallis testing with Dunn's post hoc analysis. Spearman correlation coefficient (rho) was computed to evaluate the association between IS and pCS serum concentrations and the following variables: appetite score, fecal score, frequency of vomiting, body condition score, presence of proteinuria, exclusive feeding of prescription renal diet, number of unique OTUs, hematocrit, and serum creatinine concentration, blood urea nitrogen (BUN) concentration, and serum calcium, phosphorus, sodium, and potassium concentrations. Mann‐Whitney U test was used to compare IS and pCS concentrations between cats that ate ≥75% of their food and cats that ate <75% of their food.

RESULTS

Animals

Thirty cats with CKD (17 cats had IRIS stage 2 CKD, 11 cats had IRIS stage 3 CKD, and 2 cats had IRIS stage 4 CKD) and 11 healthy older control cats were enrolled in the study. Fecal microbiome analysis was performed on healthy control cats (n = 11) and CKD cats (n = 30). Serum IS and pCS concentrations were measured in healthy control cats (n = 10) and compared to CKD cats (n = 28). An inadequate amount of serum was available to perform IS and pCS concentrations on 3 cats.

Descriptions of age, body condition score, clinical scores, hematologic, and biochemical variables of cats in each group (healthy cats, stage 2 CKD cats, and stages 3 and 4 CKD cats) are presented in Table 1. The healthy control cats were significantly younger than the CKD cats (P < .001). As expected, compared with the healthy control group, CKD cats had a significant increase in serum concentrations of creatinine (P < .001) and BUN (P < .001). Additionally, the CKD cats had significantly lower hematocrit (P = .003) and appetite scores (P = .004), and higher vomiting scores (P = .009) and serum total calcium concentration (P = .04) compared with healthy control cats. Stage 2 (P = .008) and stages 3 and 4 (P = .007) CKD cats had lower appetite scores compared to healthy control cats. Stage 2 CKD cats had higher vomiting scores compared to healthy control cats (P = .005). No difference was found in fecal scores, presence of proteinuria, and serum phosphorus, sodium, and potassium concentrations between groups. Healthy control cats were being fed variable commercial maintenance diets. Nine of 30 CKD cats ate only a commercial renal diet, 3of 30 CKD cats ate a combination of commercial renal diet and maintenance diet, and 18 of 30 ate variable commercial maintenance diets. Current medications at the time of enrollment for the CKD cats included transdermal mirtazapine gel (4 of 30 cats), amlodipine (4 of 30 cats), PO potassium supplementation (3 of 30 cats), and transdermal methimazole, alendronate, levothyroxine, maropitant, psyllium powder, aluminum hydroxide, glucosamine/chondroitin joint supplement, transmucosal buprenorphine, and polyethylene glycol (1 cat each). None of the healthy control cats were on medications at the time of enrollment.

Table 1

Characteristics of study groups including healthy control cats, International Renal Interest Society (IRIS) stage 2 chronic kidney disease (CKD) cats and IRIS stages 3 and 4 CKD cats

	Healthy controls cats (n = 11)	CKD stage 2 (n = 17)	CKD stages 3 and 4 (n = 13)
Variable (reference interval)	Mean ± SD	Mean ± SD	Mean ± SD
Age (yr)	10.5 ± 1.3a	14.5 ± 3.2b	13.4 ± 3.9b
Hematocrit (32‐47%)	41 ± 4a	36.5 ± 4b	34.5 ± 5b
Sodium (149‐157 mEq/L)	153 ± 2.1	151 ± 1.2	154 ± 1.5
	Median (range)	Median (range)	Median (range)
BCS (1‐9)	5 (4‐8)	5 (2–8)	5 (3‐7)
Fecal score (1‐7)	2 (2)	2 (1‐3)	2 (2‐3)
Appetite score (0‐4)	0 (0‐1)a	1 (0‐3)b	1 (0‐3)b
Vomiting score (0‐3)	0 (0‐2)a	2 (0‐3)b	1.5 (0‐3)c
Creatinine (0.8‐2.4 mg/dL)	1.2 (0.7‐1.6)a	2 (1.6‐2.6)b	3.2 (2.9‐6.9)b
BUN (18‐35 mg/dL)	24 (20‐38)a	43 (20‐60)b	52 (33‐98b
Total calcium (9.2‐11.1 mg/dL)	9.8 (9.1‐11.3)a	10.1 (9.1‐10.7)b	10.6 (10‐14)b
Phosphorus (3.0‐6.0 mg/dL)	4.3 (2.9‐5.0)	3.7 (2.6‐5.6)	4.4 (3.3‐8.2)
Potassium (3.7‐5.4 mEq/L)	4.6 (4.2‐5.54)	4.5 (3.89‐5.31)	4.5 (4‐5.1)
Indoxyl sulfate (ng/mL)	1201 (202‐2860)a	3370 (746‐10 300)b	6385 (1020‐27 600)b
p‐Cresol sulfate (ng/mL)	2905 (901‐7220)	6940 (34‐30 600)	5300 (189‐35 300)
Observed OTUs	2575 (1274‐3208)a	2198 (1541‐3092)b	2032 (1427‐2511)b

Rows bearing a different superscript letter are significantly different from one another.

Abbreviations: BCS, body condition score; BUN, blood urea nitrogen; OUT, operational taxonomic unit.

The sequence analysis yielded 5 085 660 quality sequences for all analyzed samples (n = 41) (mean, 124 040; range, 80 890‐369 577). The samples were rarefied to a depth of 80 875 sequences per sample to account for unequal sequencing depth. When alpha diversity, as described by Chao1 and Shannon diversity indices, was compared between CKD cats and healthy control cats, the Chao1 diversity index was significantly lower in CKD cats (median, 4286; range, 2729‐6171) when compared to healthy control cats (median, 5099; range, 2258‐6633; P = .03). The Shannon diversity index was lower in CKD cats (median, 5.9; range, 4.1‐7.3) compared to healthy control cats (median, 6.4; range, 3.8‐7.4) but this finding was not statistically significant (P = .06). Species richness, as described by the number of unique OTUs, was significantly decreased in CKD cats (median, 2152; range, 1427‐3092) when compared to healthy control cats (median, 2575; range, 1274‐3208; P = .03; Figure 1). When the CKD cats were analyzed by IRIS stage and compared to healthy control cats, the Chao1 diversity index was significantly lower in stage 2 CKD cats and in stages 3 and 4 CKD cats compared to healthy control cats (P = .03). No significant differences were found in Shannon diversity index or species richness (number of unique OTUs) among stage 2 CKD cats, stages 3 and 4 CKD cats, and healthy control cats.

Figure 1

Scatter plot of the number of observed unique operational taxonomic units (OTUs) in chronic kidney disease (CKD) and healthy control cats. Species richness was significantly decreased in CKD cats when compared to healthy control cats (P = .03)

For evaluation of beta diversity, no significant difference was found in microbial communities between CKD cats and healthy control cats based on ANOSIM of weighted (R = −0.08) and unweighted (R = 0.07) UniFrac distances. Also, no significant differences in microbial communities were observed in CKD stages 2‐4 compared to healthy control cats (weighted R = −0.04; unweighted R = 0.06). The PCoA plots are provided in Supporting Information (Figure S1).

When individual bacterial groups were analyzed based on LEfSe, several bacterial taxa were identified as being significantly different among the groups. When comparing healthy control cats to CKD cats, CKD cats had significantly lower relative abundances of the genera Holdemania, Adlercreutzia, Eubacterium, Slackia, and Mogibacterium. When comparing healthy control cats to stage 2 CKD cats, the genera Eubacterium and Adlercreutzia were less abundant and the genus Prevotella was more abundant in stage 2 CKD cats. When healthy control cats were compared to stages 3 and 4 CKD cats, the genera Adlercreutzia and Eubacterium were less abundant and the genus Prevotella was enriched in stages 3 and 4 CKD cats. Based on univariate statistics, genus Prevotella within class Prevotellaceae was less abundant in cats with CKD, whereas Adlercreutzia and [Eubacterium] were less abundant in the stages 3 and 4 CKD cats. However, these changes were not significant after correcting for multiple comparisons. Quantitative PCR was performed for E. coli and no significant difference in bacterial count was found between healthy control cats and CKD cats.

Serum indoxyl sulfate and p‐cresol sulfate analysis

Serum concentrations of IS and pCS are shown in Table 1. The IS concentrations were found to be significantly higher in CKD cats compared to healthy control cats (P < .001). Significantly higher IS concentrations were seen in stage 2 CKD cats (P = .01) and stages 3 and 4 CKD cats (P < .001) in comparison to healthy control cats (Figure 2). No significant difference was found in IS concentrations between stage 2 CKD cats and stages 3 and 4 CKD cats. The pCS concentrations were not significantly different between CKD cats and healthy control cats.

Figure 2

Serum indoxyl sulfate (IS) concentrations in healthy control cats, chronic kidney disease (CKD) stage 2 cats, and CKD stages 3 and 4 cats. Significantly higher IS concentrations were seen in stage 2 CKD cats (P = .01) and stages 3 and 4 CKD cats (P = .0006) in comparison to healthy control cats. There was no statistical difference in IS concentrations between stage 2 CKD cats and stages 3 and 4 CKD cats

Serum concentrations of IS were found to be correlated to hematocrit (rho −0.68; P < .001), BUN (rho 0.71; P < .001), creatinine (rho 0.64; P < .001), and pCS (rho 0.41; P = .01). No correlation of either IS or pCS concentrations to appetite score, vomiting score, fecal score, body condition score, exclusive feeding of a prescription renal diet, number of unique OTUs, and presence of proteinuria, or serum phosphorus, calcium, sodium, and potassium concentrations were found. When IS and pCS concentrations were compared between cats that ate ≥75% of their food and cats that ate <75% of their food based on reported appetite scores at enrollment, serum concentrations of IS tended to be higher in cats that had a worse appetite (P = .07), but no significant difference in pCS concentrations was seen between the cats. The genera Bifidobacterium (rho −0.41; P = .01) and Eubacterium (rho −0.40; P = .01) were significantly correlated with pCS serum concentrations, whereas the genus Mogibacterium (rho −0.32; P = .05) was significantly correlated with serum IS concentrations.

DISCUSSION

Decreased fecal bacterial species diversity and richness in CKD cats compared to healthy control cats is consistent with previous assessment of the intestinal microbiome in humans with CKD and rat CKD models.1 The effect of uremia on the composition of the gut microbiome and its association with IS and pCS are highly variable among studies in rats and humans.1, 8, 43, 44, 45 in vitro studies in rats have shown that several bacterial families generate toxic compounds such as p‐cresol and indole.9, 46 However, our study did not show a correlation between the types of bacteria known to produce indole and p‐cresol in humans and serum IS and pCS concentrations in cats. Although E. coli is a documented tryptophanase‐producing bacteria in humans,10, 11 qPCR did not show quantifiable differences between CKD cats and healthy control cats, although the serum IS concentrations were significantly different between the 2 groups. In humans, bifidobacteria have been found to generate p‐cresol,10, 47 but in our study a negative correlation was found between the genus Bifidobacterium and serum pCS concentrations.

The reasons for differences between this study and studies in humans is likely multifactorial. First, species differences exist in the commonly isolated bacterial groups from the gastrointestinal tract between humans and cats, making it difficult to make a direct comparison.30 Second, although all dogs and cats harbor similar bacterial groups when analyzed at the level of the family or genus, much variability exists in the gut microbiome at the species level with generally only 5%‐20% overlap of species among individuals.48 This variability makes it difficult to incriminate a single bacterial species or genus as the culprit for increased serum IS and pCS serum concentrations in CKD patients. In addition, knowledge of the intestinal microbial composition alone does not lead to a complete understanding of its metabolic activity because of functional redundancy among different microbial species.49 Third, historical antibiotic use, variable diets, and prolonged intestinal transit times are well‐known factors that promote uremic toxin generation in people. Although these factors were highly variable among study cats, they were similarly variable for both the control and CKD cats. Further investigation is warranted before the composition of the responsible bacterial species contributing to the increased serum IS and pCS concentrations in CKD cats can be determined.

Serum IS concentrations were correlated with severity of CKD in cats in a previous study in which CKD was defined based on persistent azotemia or azotemia with smaller kidneys on imaging, without urinalysis.24 More specifically, this study demonstrated that all stages of CKD cats had significantly higher IS concentrations compared to non‐azotemic cats, and that stage 4 CKD cats had significantly higher IS concentrations compared to stages 2 and 3 cats. However, no significant difference in IS concentrations between CKD stage 2 and CKD stage 3 cats was seen, similar to our study. Serum IS concentrations were found to be significantly correlated with serum phosphorous concentrations in this previous study, and in both people and cats with CKD, IS and serum phosphorous concentration are reported to be important progression factors.24, 25, 50, 51, 52 These findings were not repeated in our study, which may have been because of lower numbers of cats with hyperphosphatemia or CKD IRIS stage 4 that were enrolled. Although there are similarities between the 2 studies, caution should be taken in directly comparing results because the enrolled populations are different. In our study, more stringent exclusion and inclusion criteria were used that centered on the primary aim of characterizing the fecal microbiome of cats with CKD. For example, stage 4 cats enrolled in the previous study might have recently received antibiotics, potentially altering their fecal microbiome and influencing uremic toxin concentrations.53

A negative correlation between hematocrit and IS concentration was observed in our study. This association may be coincidental because non‐regenerative anemia is common in CKD cats.24, 54 However, IS has been documented in humans to decrease EPO production by dysregulation of oxygen metabolism in tubular cells and impairing oxygen sensing in EPO‐producing cells in the kidney.18 Further investigation is required to determine if IS has a similar effect on EPO production in CKD cats.

In humans, serum pCS concentrations are higher in hemodialysis and CKD patients than in controls.43, 55 Although we did not show a significant difference in serum pCS concentrations between CKD cats and healthy control cats, the highest serum pCS concentrations in the healthy control group was 7220 ng/mL and 13 of 28 (46%) CKD cats had serum pCS concentrations higher than the healthy control maximum with the maximum serum pCS concentration in CKD cats (35 300 ng/mL) documented in a stage 4 CKD cat. Based on a post hoc sample size calculation, a sample size of 42 cats per group would be necessary to obtain 80% statistical power. Thus, it is possible that the inability to detect significance in pCS concentrations between the 2 groups was because of small sample size.

Indoxyl sulfate and pCS lead to progression of CKD,12, 13, 14, 15, 16 contribute to multi‐organ dysfunction,17, 18, 21, 56 and are associated with increased mortality in human CKD patients.17, 18, 21 In addition, because removal by hemodialysis in human patients is markedly lower for protein‐bound IS and pCS than for urea and creatinine,57 human medicine has focused on strategies to decrease production of IS and pCS including modulation of bacterial growth in the colon by dietary management,58 prebiotics, probiotics, and target adsorption of uremic toxins by the use of adsorbents.2, 5, 46 The generation of IS and pCS can be modulated by selectively increasing saccharolytic and decreasing proteolytic bacteria in the colon and by increasing intestinal transit time. Prebiotics and probiotics have been shown to influence the composition of the colonic microbiota and have been successfully used to decrease IS and pCS concentrations in human CKD patients.43, 59 In addition, increasing carbohydrate and fiber and decreasing protein intake have been shown to decrease IS and pCS concentrations.47, 58, 60 Adsorbents such as sevelamer hydrochloride61, 62 (Renalgel, Genzyme, Cambridge, Massachusetts) and AST‐12061, 62 (Kremezin, Kurecha Chemical Industry, Tokyo, Japan) also are used to limit intestinal absorption of IS and pCS. In veterinary medicine, there has been no published investigation into strategies to decrease gut‐derived uremic toxins in CKD patients and, based on a growing body of literature,24, 25 further exploration as a potential therapeutic target seems warranted.

Our study had several limitations. The control group could not be age‐matched to the CKD group. Because of the high prevalence of CKD in the geriatric cat population and the study exclusion criteria, an age‐matched healthy control population (ie, without CKD or concurrent illness) could not be identified in the referral hospital feline population. Another limitation of the study was the low number of enrolled CKD stage 4 cats. Because of the exclusion criteria (in particular, no antibiotics or antacid therapy for 6 weeks before enrollment), only 2 stage 4 CKD cats could be enrolled. Because of this, stage 3 and stage 4 CKD cats were combined into a single group. Also, 2 of 30 cats with CKD maintained a normal USG (>1.035) and were diagnosed based on persistently increased serum creatinine concentration >1.6 mg/dL over a 3‐month period and an increased SDMA serum concentration. According to IRIS guidelines, USG > 1.035 still can occur in some cats with azotemic CKD.34 A glomerular filtration rate study (iohexol or nuclear scintigraphy) to document decreased renal function would have been required to confirm renal dysfunction in these cats but was not feasible. Lastly, accurate determination of dietary protein intake was not possible because of the common practice of feeding a mixture or rotation of diets. Thus, analysis of a possible relationship between percent dietary protein and uremic toxin concentrations was not possible.

In summary, we demonstrated that CKD is associated with a decreased fecal bacterial species diversity and richness in cats. Indoxyl sulfate is significantly increased in cats with CKD and IRIS stage 2 CKD cats may suffer from a similar uremic toxin burden as do cats with later stage disease. Future studies are needed to further understand the interplay between the fecal microbiome and serum concentrations of uremic toxins.

CONFLICT OF INTEREST DECLARATION

Authors declare no conflict of interest.

OFF‐LABEL ANTIMICROBIAL DECLARATION

Authors declare no off‐label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

The project was approved by the IACUC at Colorado State University. All owners gave written informed consent before participation.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.

Figure S1: Principal coordinate analysis (PCoA) of unweighted and weighed UniFrac distances of 16S rRNA genes. Each dot represents an individual cat fecal sample. There was no significant difference in microbial communities between CKD cats and healthy control cats (A) and between the stages of CKD and healthy control cats (B).

Click here for additional data file.

Table S1: Clinical scoring system used to determine appetite, consistency of feces, and frequency of vomiting.

Click here for additional data file.