The kidney's problem starts in the gut.
The kidney filters blood. That much is uncontroversial. What has shifted in the past decade is the recognition of where the things the kidney filters are coming from. A growing body of work, summarised in nephrology reviews and now extended to dogs and cats, places the gut microbiome upstream of the kidney as a meaningful source of the small molecules that accumulate in chronic kidney disease (Vaziri et al. 2013; Lim et al. 2021).
The framing is sometimes called the gut-kidney axis. The idea is straightforward. The microbial community in the colon ferments what reaches it from the small intestine. The products of that fermentation are absorbed into the portal circulation, processed by the liver, and excreted by the kidneys. When kidney function declines, those products accumulate. Some of them are toxic at the concentrations they reach. And the composition of the microbial community determines, in part, which products are produced and at what rate.
In a healthy gut, the dominant fermentation pattern is saccharolytic. Saccharolytic bacteria break down complex carbohydrates and dietary fibre, producing short-chain fatty acids such as butyrate, propionate, and acetate. These short-chain fatty acids are useful: they feed colonocytes, support intestinal barrier integrity, and have been shown to reduce kidney inflammation and fibrosis through several mechanisms including G-protein-coupled receptor signalling and histone acetylation (Kim et al. 2023).
In a dysbiotic gut, the balance tips. Proteolytic species rise, and the fermentation pattern shifts toward the breakdown of aromatic amino acids: tryptophan, tyrosine, and phenylalanine. The products of that shift are the precursors of indoxyl sulfate, p-cresyl sulfate, and a family of related compounds collectively known as protein-bound uraemic toxins. These are the molecules nephrologists worry about, because they are difficult to remove with conventional dialysis and are independently associated with the progression of kidney disease and cardiovascular events.
Vaziri and colleagues (2013) were among the first to characterise this shift in detail in human and rat chronic kidney disease, reporting that uraemia itself reshapes the gut community in a way that further amplifies the production of toxin precursors. The same pattern, with species-specific variations, has now been documented in dogs and cats.
The toxins: indoxyl sulfate and p-cresyl sulfate.
Two molecules carry most of the weight in the gut-kidney axis literature. Both are protein-bound. Both are produced by gut bacteria from dietary amino acids. Both are normally cleared by the kidney via tubular secretion. And both accumulate when kidney function declines.
Indoxyl sulfate is the sulfate conjugate of indole. Indole is produced from tryptophan by bacterial tryptophanase, an enzyme widely distributed across Escherichia coli, Proteus vulgaris, and several Bacteroides species (Summers et al. 2019). Once produced in the colon, indole is absorbed into the portal blood, conjugated in the liver to form indoxyl sulfate, and excreted in the urine. About 90% of the indoxyl sulfate in plasma is bound to albumin (Chen et al. 2018), which is why dialysis removes it inefficiently: the protein-bound fraction is not freely available for membrane transfer.
p-Cresyl sulfate is the sulfate conjugate of p-cresol, which is generated by gut bacteria from tyrosine and phenylalanine. The bacterial genera implicated in human studies include Bacteroides, Lactobacillus, Enterobacter, Bifidobacterium, and Clostridium, although the specific species mix varies by host. Like indoxyl sulfate, p-cresyl sulfate is produced in the gut, conjugated in the liver, and excreted in the urine.
What these molecules do at elevated concentrations is the reason the literature treats them as more than markers. Indoxyl sulfate has been shown in cell and animal studies to induce reactive oxygen species in renal tubular cells, inhibit tubular cell proliferation, increase expression of profibrotic cytokines, and promote glomerular sclerosis (Lim et al. 2021). It has also been shown to suppress erythropoietin production, which is part of the explanation for the anaemia that accompanies progressive kidney disease. p-Cresyl sulfate induces a comparable inflammatory transcriptional programme in renal tubular cells. The accumulation of these toxins is therefore not just a passive consequence of reduced kidney clearance; it is a contributor to further kidney injury, which is why the gut-kidney axis literature describes a self-perpetuating cycle.
The clinical question that follows is whether the cycle can be measured early, and whether intervening on the gut side of the axis can shift the kidney side. Both questions have been addressed in companion-animal studies.
What the cat data show.
The most direct feline evidence comes from a 2019 study by Summers and colleagues, published in the Journal of Veterinary Internal Medicine. Thirty client-owned cats with chronic kidney disease at IRIS stages 2 to 4 were compared with 11 healthy older control cats. Faecal samples were sequenced using 16S rRNA on the Illumina platform, and serum indoxyl sulfate and p-cresyl sulfate were measured by liquid chromatography tandem mass spectrometry.
The microbiome findings were directional rather than dramatic. Cats with chronic kidney disease had significantly lower Chao1 diversity and significantly fewer observed operational taxonomic units than healthy cats. The Shannon diversity index was lower in the kidney disease group but did not reach statistical significance. Beta diversity comparisons did not separate the groups cleanly. At the genus level, the kidney disease cats had significantly lower abundances of Holdemania, Adlercreutzia, Eubacterium, Slackia, and Mogibacterium. Prevotella was enriched in the kidney disease cats. The takeaway from the microbiome side: chronic kidney disease in cats is associated with a less diverse, compositionally shifted faecal community, but the shift is not large enough to be diagnostic on diversity metrics alone.
The toxin findings were sharper. Indoxyl sulfate concentrations were significantly higher in the kidney disease cats than the healthy controls. The unexpected and clinically important finding was the timing: cats at IRIS stage 2, the earliest stage at which most clinicians make a kidney disease diagnosis, already had indoxyl sulfate concentrations significantly elevated above controls, with no significant further increase from stage 2 to stages 3 and 4. The implication is that the uraemic toxin burden is established early in the disease course in cats, which makes early intervention biologically reasonable.
p-Cresyl sulfate concentrations did not differ significantly between groups in the Summers study, although the highest values in the kidney disease group were several-fold above the highest control value. The authors noted that the study was underpowered for p-cresyl sulfate; their post hoc calculation suggested 42 cats per group would be required to reach 80% power, against the 30 they had recruited.
Chen and colleagues (2018), working in Taiwan with a different feline cohort of 58 cats with chronic kidney disease at IRIS stages 2 to 4, extended the picture by linking indoxyl sulfate to disease progression. They defined progression as either a 0.5 mg/dL rise in serum creatinine within the same IRIS stage or an advance of one IRIS stage, both within a three-month window. Cats whose disease progressed had significantly higher baseline plasma indoxyl sulfate concentrations than cats whose disease remained stable, both at IRIS stage 2 (median 14.7 vs 10.6 mg/L) and at IRIS stage 3 (22.4 vs 11.07 mg/L). The receiver operating characteristic curves had area-under-curve values of 0.755 and 0.766 for stages 2 and 3 respectively, which is moderate discriminative performance. Baseline serum creatinine, by contrast, did not separate progressors from non-progressors at all.
In Chen's multivariable logistic regression, indoxyl sulfate remained an independent predictor of progression after adjustment for blood urea nitrogen, creatinine, phosphate, hemoglobin, hematocrit, and albumin. In other words, the gut-derived toxin was telling the clinician something the standard renal panel was not.
What the dog data show.
The canine evidence is more recent and somewhat less mature, but the directional findings are consistent with the feline literature. Kim and colleagues (2023), publishing in Frontiers in Veterinary Science, sequenced the faecal microbiomes of 19 dogs with chronic kidney disease at IRIS stages 1 to 4 alongside 10 healthy control dogs, using 16S rRNA V3-V4 region sequencing.
As in the feline studies, alpha and beta diversity metrics did not significantly differ between the kidney disease and control groups. The compositional shifts, however, were clear. The phylum Proteobacteria was significantly elevated in the chronic kidney disease group (median relative abundance 2.0% vs 0.3% in controls). At the family level, Enterobacteriaceae was significantly elevated. At the genus level, Enterococcus was elevated and Ruminococcus was reduced.
The most informative finding in the Kim study was the stage-progressive trend analysis using the Jonckheere-Terpstra test. As IRIS stage advanced, the relative abundance of two proteolytic genera, Klebsiella and Clostridium, climbed significantly. The relative abundance of the saccharolytic Ruminococcus fell significantly. At the species level, Collinsella intestinalis climbed with stage. Each of these shifts is consistent with the proteolytic-to-saccharolytic transition described in the human and rat literature: more bacteria producing the precursors of uraemic toxins, fewer bacteria producing short-chain fatty acids.
The species composition has additional functional meaning. Klebsiella is a well-described lipopolysaccharide producer. Lipopolysaccharide translocation across a compromised intestinal barrier is one of the proposed mechanisms by which gut dysbiosis amplifies systemic inflammation in chronic kidney disease (Kim et al. 2023). Collinsella, in mouse models, has been associated with downregulation of intestinal tight-junction proteins and increased proinflammatory cytokine production. The composition of a chronic kidney disease microbiome is therefore not a bystander; the species that thrive in dysbiosis are exactly the species whose products place additional load on a kidney that is already struggling.
Chen and colleagues (2018), in the same study that established indoxyl sulfate as a progression marker in cats, also enrolled 36 dogs with chronic kidney disease at IRIS stages 2 and 3. The pattern was the same. Dogs whose disease progressed had significantly higher baseline indoxyl sulfate concentrations than dogs whose disease remained stable. The receiver operating characteristic curves were sharper than in cats: area-under-curve 0.797 at IRIS stage 2 and 0.885 at IRIS stage 3. Multivariable logistic regression confirmed indoxyl sulfate as an independent predictor after adjustment for the standard renal markers.
Plasma indoxyl sulfate in dogs correlated significantly only with phosphate, not with the broader chemistry panel. The authors interpreted this as evidence that indoxyl sulfate is doing something distinct from the standard markers, rather than serving as a downstream summary of them. In their words, indoxyl sulfate served as a marker for predicting renal progression in both dogs and cats with stages 2 or 3 chronic kidney disease, and it was a better predictor than previously reported factors in dogs.
The oral connection.
The gut is the most-studied microbial entry point in the gut-kidney axis, but it is not the only one. The oral microbiome contributes too, and the most quantitative companion-animal evidence on this point comes from a retrospective cohort study by Trevejo and colleagues (2018), published in the Journal of the American Veterinary Medical Association.
The study used the Banfield Pet Hospital electronic record system to assemble a cohort of 169,242 cats seen at any of 829 primary-care hospitals between 2002 and 2013. Cats with a staged diagnosis of periodontal disease at study entry (n = 56,414) were frequency-matched on age and year of entry to cats with no record of periodontal disease (n = 112,828). The endpoint was a new diagnosis of chronic azotemic kidney disease during follow-up. Cox proportional hazards regression was used to estimate hazard ratios after adjustment for age, breed, sex, neuter status, body weight, history of general anaesthesia, history of cystitis, diabetes mellitus, and hepatic lipidosis.
The adjusted hazard ratio for chronic kidney disease was 1.33 in cats with stage 1 periodontal disease, 1.34 with stage 2, and 1.50 with stage 3 or 4, all relative to cats with no periodontal disease and all statistically significant. The risk was dose-related to severity: more advanced periodontal disease, higher kidney disease risk. Age was a strong independent predictor, as expected. Several breeds, including Siamese, Himalayan, and Abyssinian, also carried significantly elevated risk.
The mechanism the authors propose, drawing on the human literature, runs through systemic inflammation. Periodontal pathogens and their lipopolysaccharide products enter the systemic circulation through the inflamed gingival tissue. Chronic low-grade inflammation drives endothelial dysfunction and microvascular damage, including in the renal vasculature. Over years, this contributes to the structural and functional decline that becomes clinically detectable as chronic kidney disease.
For the gut-kidney axis, the implication is that the relevant microbial community is not just colonic. The oral microbiome is part of the same axis, and oral dysbiosis is associated with a measurably elevated long-term risk of kidney disease in cats. This is why a full microbiome view of an animal at risk includes the oral panel, not just the gut panel.
What this means for diet and intervention.
If the gut and oral microbiomes contribute to the uraemic toxin burden in chronic kidney disease, the question is whether modifying the microbiome modifies the toxin burden, and whether that translates into clinical benefit. Two recent companion-animal studies are directly relevant.
Ephraim and Jewell (2020), publishing in Metabolites, ran a randomised crossover study in 28 dogs with IRIS stage 1 chronic kidney disease (defined by serum SDMA above 14 µg/dL). Each dog was fed a control food, a low-soluble-fibre food with added betaine, and a high-soluble-fibre food with added betaine, each for ten weeks in a Williams Latin Square sequence. The added ingredients were 0.5% betaine and prebiotics: short-chain fructooligosaccharides at 0.27% or 0.41% and oat beta-glucan at 0.39% or 0.59%, depending on food.
The findings were directional and consistent. Plasma 4-methoxyphenol sulfate, eugenol sulfate, and 3-methoxycatechol sulfate, all uraemic solutes elevated in human haemodialysis cohorts, fell significantly on one or both test foods relative to control. Hydroxyproline, a marker of collagen degradation, fell. N-methylproline, a fibrosis biomarker, fell. Plasma omega-3 fatty acid concentrations rose. Serum creatinine fell significantly on both test foods relative to control, although all values remained within the normal range. The shifts in the gut microbiome composition itself were modest, which is consistent with a growing literature suggesting that the metabolic capacity of a microbiome can change without large compositional shifts in the underlying community.
Hall, Jewell, and Ephraim (2022), publishing in PLOS ONE, ran a parallel study in seven cats with chronic kidney disease at IRIS stages 1 and 2, using a betaine-and-prebiotic food formula similar to the canine study. Total body mass was significantly higher after eight weeks on the test food than after eight weeks on the control food, with no significant difference in food intake. All seven cats gained body mass on the test food. The urinary excretion of indoxyl sulfate and the related 5-, 6-, and 7-hydroxyindole sulfates fell on the test food, alongside reductions in plasma indole and p-cresol pathway markers. The plasma indoxyl sulfate burden, in this small study, was positively correlated with both serum SDMA and serum creatinine.
Two caveats are worth keeping front of mind. First, both studies involve a single product formulation from a single sponsor, and both report directional rather than disease-modifying outcomes. Whether the metabolite shifts translate into longer survival, slower IRIS-stage progression, or improved quality of life is not yet established in randomised clinical endpoint trials. Second, dietary management of chronic kidney disease in dogs and cats is the responsibility of the treating veterinarian, who weighs many factors beyond the gut-kidney axis, including phosphate restriction, protein quality, sodium content, and palatability for an animal that may already be inappetent. The microbiome data add to that conversation; they do not replace it.
Where BAARK fits.
The gut-kidney axis on a BAARK report.
BAARK's Complete Bundle measures the gut and oral microbiomes from a single mail-out collection. The functional metabolic profile in the report quantifies the indole-producing capacity, the p-cresol biosynthesis capacity, the trimethylamine and urease pathways, and the short-chain fatty acid production capacity, all derived from the gene content of the sequenced community. For an animal with vet-diagnosed kidney disease, or an animal in a breed at elevated kidney disease risk, this is the part of the report that maps most directly onto the literature reviewed above.
What the report does not do is make a kidney disease diagnosis. It does not detect, stage, or predict the progression of chronic kidney disease. The clinical diagnosis belongs with the veterinarian, who has access to the creatinine, SDMA, urinalysis, blood pressure, and imaging that an IRIS staging requires. What the BAARK report adds is the parallel microbiome view: a documented dysbiosis pattern, a quantified uraemic-toxin precursor pathway burden, and a baseline against which a retest after a renal-supportive diet trial can show whether the microbiome side has shifted.
This is the framing used in the Adelle demo report, where a cat at IRIS stage 2 with a vet-diagnosed early-stage kidney disease shows the gut-kidney axis pattern in the form the report delivers it. The report contains a full bacterial taxonomy, a functional metabolic profile with a uraemic-toxin precursor callout, an oral panel that flags periodontal-associated organisms, and a vet-collaboration band that points the conversation back to the treating clinician.
For owners and vets thinking about what the test is good for, the gut-kidney axis is one of the clearest use cases. The published evidence supports the existence of the axis in dogs and cats. The microbiome shifts described in Kim 2023 and Summers 2019 are the shifts the BAARK report quantifies. The uraemic toxin precursor pathways described in Chen 2018 are the pathways the BAARK functional profile reports. The dietary intervention literature suggests that change is possible. None of this replaces the kidney panel, but it adds a dimension the kidney panel does not measure.
If you are interested in the kit itself, the order page describes what is in the box, how the sample is collected, and how the report is delivered. The what we test page describes the targets of the assay in more detail.
The bottom line.
Chronic kidney disease in dogs and cats has a gut microbiome story. Cats at IRIS stage 2 already carry an elevated indoxyl sulfate burden. Dogs across IRIS stages 1 to 4 show a stage-progressive proteolytic-to-saccharolytic shift. In both species, plasma indoxyl sulfate predicts disease progression independently of the standard renal markers. And in cats, periodontal disease is associated with later kidney disease in a cohort large enough to make the association reliable.
This is the science layer behind the gut-kidney axis findings on a BAARK report. It is wellness profiling, not a diagnostic test. Decisions about diet, supplements, and clinical management belong with your veterinarian; the microbiome view is one input into that conversation, alongside the creatinine, SDMA, urinalysis, and blood pressure that the vet relies on for staging and monitoring.