A Review of Experimental Models of Feline Kidney Disease

Abstract

Chronic kidney disease (CKD) is a leading cause of mortality in cats, yet no treatments currently exist to reverse or halt its progression. This lack of therapeutic options stems partly from a limited understanding of the disease pathogenesis and the complexities of its heterogeneous nature. Experimental models of kidney disease are crucial for advancing research and improving treatment outcomes. These models facilitate the identification of biomarkers, elucidate disease mechanisms, and enable the testing of potential therapies. Several feline-specific models, such as ischemia-reperfusion injury (IRI), remnant kidney (RK), and toxin-induced injury (TI), have been developed to study feline kidney disease. Each model has distinct advantages and limitations, making the careful selection of appropriate models critical for progressing research in feline nephrology. The IRI model mimics acute kidney injury that can progress to CKD, while the RK model induces CKD by partially removing kidney tissue, leading to glomerular hyperfiltration. The TI model, involving toxins like meloxicam, provides a simpler approach to studying kidney damage. Despite their utility, these models present challenges, including variability in outcomes, technical demands, and the need for refined methodologies. This review examines the strengths and weaknesses of these feline models and offers recommendations for researchers working to discover new biomarkers and develop effective treatments for CKD in cats.

Introduction

Kidney disease is one of the most common causes of mortality seen in feline practice,^1,2 and estimates of the prevalence of chronic kidney disease (CKD) in cats have ranged from 9.6% to 50%.^3,4 Despite the significant impact of this disease on veterinary medicine, advances in feline nephrology over the past few decades have been relatively modest. Currently, there are no treatments available to reverse or halt the progression of this devastating condition. The lack of therapeutic options is, in part, due to insufficient knowledge about the disease’s pathogenesis and slow progress in research within this area. A challenge in studying CKD is its relative heterogeneity; it appears to be triggered by different factors and presents with varying kidney changes.^5,6 In addition, CKD often occurs with other conditions, such as hypertension and hyperthyroidism, that can also affect kidney function.^5,6 Moreover, the biomarkers currently available for detecting and monitoring the progression of this disease are not very sensitive. One of the objectives in reducing the impact of CKD is to develop better tools for early detection as well as therapeutic options to prevent, halt, or reverse the disease’s progression. To achieve this, substantial progress in our understanding of the disease is essential.

The availability of suitable animal models can accelerate the discovery of biomarkers for disease staging, the development of therapies, and the design and testing of novel therapeutic strategies.⁷ Over the past few decades, pigs and rodents have been extensively used to study experimentally induced kidney disease.⁸ However, an apparent lack of exact disease mimicry of animal models may partially explain why no drugs have been approved for treating kidney disease in humans and other species.⁸ Moreover, it remains uncertain whether data generated from other species can be fully translated into feline nephrology. There are likely significant interspecies differences in the pathogenesis of kidney disease.⁸ Therefore, using feline models could expedite our understanding and advancement in this area, rather than relying solely on information obtained from other species.

The optimal experimental models should resemble, as much as possible, the naturally occurring events.⁸ However, this is a task difficult to accomplish in feline nephrology because the disease can be initiated by multiple causes (for example, amyloidosis, renal dysplasia, neoplasia, polycystic kidney disease, infection, immune-complex glomerulonephritis, urolithiasis)^5,6 that could result in pathogenesis events that are not necessarily similar. Furthermore, it has been speculated that intrarenal events are likely to be shaped by individual and environmental factors and/or repeat episodes of acute kidney injury (AKI).⁵ Experimental models of kidney disease hold numerous advantages over naturally occurring disease, including homogeneity of events and the ability to study the kidney response to controlled injury. While not perfect, the use of these models offers the opportunity to examine individual mechanisms in an accelerated time frame. Selection of a model that most closely approximates the specific clinical scenario intended for the use of the biomarker or therapeutic is crucial to maximizing the probability that findings translate into clinical studies.^7,8 Several feline models of kidney disease have been characterized and are used to study the disease’s pathogenesis and identify potential biomarkers to detect or monitor the progression of kidney disease. These models include ischemia-reperfusion injury (IRI), remnant kidney (RK), toxin-induced injury (TI), and postrenal obstruction (PO) models. Each model has advantages, limitations, and specific applications that are critical to consider when selecting an experimental model. In this review, we outline the strengths and weaknesses of these model systems and provide recommendations for researchers involved in discovering and validating new biomarkers and therapeutics for the early detection and treatment of kidney disease in cats. A summary of our findings is presented in Table 1.

Table 1.

A summary of described experimental models of feline kidney disease

	Experimental model
	Ischemia-reperfusion injury models9–13	Remnant kidney models14–21	Combined renal embolism and remnant kidney model22	Toxin-induced injury models23,24	Postrenal obstruction model25
Disease induction	Temporary surgical clamping of one or both renal arteries and veins	Partial ablation of one kidney in conjunction with simultaneous or delayed contralateral nephrectomy	Transcatheter administration of embolic microspheres with subsequent contralateral nephrectomy	Meloxicam 0.3 mg/kg SC every 24 h for 31 d	Urinary catheterization of male cats followed by catheter occlusion
Maximum duration of evaluation	120 d	736 d	178 d	47 d	3 d
Minimum reported time to abnormal serum creatinine	24 h	2 wk	Unclear	13 d	9 h
Survival	50% to 100%	78% to 100%	100%	100%	100%
Primary histologic lesions	Interstitial inflammation and fibrosis; tubular atrophy; periglomerular fibrosis; obsolescent glomeruli	Mesangial matrix expansion; tubular lesions; interstitial fibrosis; interstitial inflammation; glomerulosclerosis	Kidney fibrosis; tubular atrophy and loss; lymphoplasmacytic inflammation	Cortical tubular dilation; cortical and medullary tubular degeneration and necrosis; tubular inflammation	Unknown
Variability of the histologic lesions	Marked	Marked	Marked	Marked	Unknown
Modeling of AKI	Yes	No	Yes	Yes	Unknown
Modeling of CKD	Yes	Yes	Unknown	Unknown	Unknown
Size of a feasible experiment	Relatively small due to technical requirements	Relatively small due to technical requirements	Relatively small due to technical requirements	Small to large	Small to large
Main advantages	Reliable induction of kidney injury	Reliable induction of CKD with variable proteinuria and hypertension	Reliable induction of kidney injury and alteration of makers of kidney function	Small technical resource use and associated cost	Small technical resource use and associated cost
Main disadvantages	High technical resource use and associated cost	High technical resource use and associated cost	High technical resource use and associated cost	Variable induction of kidney injury	Unknown induction of kidney injury

Ischemia‑Reperfusion Injury Models

Kidney disease induced by IRI has been studied in cats. These models involve the temporary surgical clamping of one or both renal arteries and veins to produce ischemic AKI.^9–13 In rodents, the severity of AKI can be controlled by altering the duration and pressure of clamping,²⁶ as well as body temperature.²⁷ Only the duration of clamping has been investigated in cats. Bilateral clamping for 15 to 30 min resulted in no increase in serum creatinine after 21 d, although a reduction in glomerular filtration rate (GFR) was detected in one out of 6 cats.¹² Renal histopathologic changes were likewise mild, with rare streaks of tubular atrophy, mild inflammation, and fibrosis present in 50% of cats.¹² In addition to the unreliable induction of AKI, short clamp times induce insufficient injury to cause progression to CKD.⁸ In contrast, bilateral clamping for 60 min resulted in reduced GFR and increased serum creatinine and BUN in 4 out of 4 cats after 24 h.⁹ Mild brush border attenuation and relative tubular lumina expansion were present 60 min after clamping, and tubular necrosis, degeneration, and regeneration were observed in 2 cats that died 3 and 6 d postclamping. One cat survived to the end of the study period (14 d postclamping) and had a serum creatinine and BUN within the laboratory reference interval at this time.

To reduce mortality, subsequent studies have involved unilateral clamping only. Unilateral clamping for 60 min resulted in a significant reduction in mean GFR accompanied by a nonsignificant increase in mean serum creatinine one to 12 d postoperatively.¹⁰ In a different group of cats, the mean GFR was no longer significantly reduced from baseline, although mean serum creatinine was significantly elevated from baseline 70 d postoperatively. The explanation for these seemingly contradictory findings is not clear. The small number of cats and the variability common with GFR testing may be reasons for this discrepancy. Renal histopathology from cats euthanized between days 3 and 12 postclamping showed tubular necrosis, regeneration, and interstitial inflammation that was predominantly mononuclear. In cats euthanized between days 12 and 70, tubular atrophy, interstitial fibrosis, and interstitial mononuclear inflammation were present. Lesions were most severe within the corticomedullary junction, likely due to the high energy demands and limited oxygen supply of the distal portion of the proximal tubule (S3 segment) and thick ascending limb of the loop of Henle.^10,28 Unilateral clamping for 90 min increased mean serum creatinine on days one to 7 postclamping.¹¹ Mean serum creatinine then returned to baseline from days 30 to 60 postclamping, followed by a gradual increase with increased levels from days 90 to 180, with a lower mean GFR than baseline recorded on day 180. Renal histopathology at this time showed interstitial inflammation and fibrosis, tubular atrophy, periglomerular fibrosis, and obsolescent glomeruli. There was marked interindividual lesion severity in both studies with unilateral clamping,^11,12 which may reflect variable degrees of tubular regeneration following the ischemic insult.¹¹ Implementing a unilateral ischemic injury, Lourenço et al²⁹ documented greater genetic expression of molecular pathways associated with inflammation and fibrosis.

Since feline CKD is not typically unilateral, unilateral ischemic models will likely provide limited external validity. Also, compensatory hypertrophy of the nonclamped contralateral kidney precludes the accurate assessment of functional changes to the clamped kidney over time using clinically available serum and urinary biomarkers.^8,11

These problems can be overcome with nephrectomy of the contralateral kidney, which is typically delayed following clamping.³⁰ This allows a degree of recovery from the ischemic insult to occur that may otherwise be lethal if nephrectomy were performed simultaneously. It may allow the period of renal medullary ischemia to be prolonged postoperatively, as renal blood flow is redistributed away from the clamped kidney to the uninjured kidney.³⁰ Unilateral clamping for 60 min with contralateral nephrectomy 2 wk later increased serum creatinine immediately postnephrectomy.¹² After 3 mo, serum creatinine remained static or decreased, and mean GFR was not different between cats that underwent clamping and those that underwent unilateral nephrectomy only. Renal histopathology of the clamped kidneys at this time showed variable degrees of tubular atrophy and mononuclear inflammation.¹² In another study, unilateral clamping was performed for 90 min, with contralateral nephrectomy performed either 21 or 90 d later.¹³ This protocol was associated with a relatively high mortality rate, with 50% and 16.7% of cats with contralateral nephrectomy performed 21 and 90 d later, respectively, euthanized due to severe AKI within 7 d of nephrectomy. Renal histopathology of these cats showed tubular necrosis and interstitial lymphoplasmacytic inflammation. Cats that survived to the 120-d mark had a mean increase in serum creatinine and a reduction in GFR from baseline. Renal histopathology showed cortical fibrosis, tubular atrophy with variable inflammation, and glomerular obsolescence.¹³

IRI models require many technical resources, including a veterinary surgeon, an anesthesiologist, postoperative monitoring, and associated costs. A considerable limitation of this model is that the number of cats undergoing the surgical procedures in a day is very limited, at least under typical experimental circumstances. This may limit the sample size of experiments and slow the progression of research programs if available resources necessitate the staggering of experiments. Furthermore, novel AKI therapeutics identified using IRI models over the past decades have consistently failed phase 3 trials in people.⁷ It has been speculated that these models may not predict therapeutic responses in naturally occurring AKI, among other reasons.⁷ The requirement for general anesthesia also introduces numerous factors that may introduce interindividual variability in the magnitude of renal injury induced, such as anesthetic duration, refractory hypotension, and/or perioperative use of nonsteroidal anti-inflammatory drugs.³¹

Remnant Kidney Models

RK models involve the partial ablation of one kidney to create a “remnant” in conjunction with simultaneous or delayed contralateral nephrectomy.³² They are a reliable method of inducing CKD in cats,^14–21 resulting in glomerular hyperfiltration of remnant nephrons.¹⁷ Partial ablation has been most commonly performed by ligation of 5/6 of the renal artery branches,^18–21,32 although ligation of 50%¹⁷ and 4/6 of the branches^15,16 has also been reported. The procedure appears to be well tolerated even with simultaneous nephrectomy, with perioperative mortality reported in only one study.²¹ Cats in this study had contralateral nephrectomy performed 3 wk following partial ablation, with 22% euthanized within 29 d of nephrectomy due to a serum creatinine greater than 12 mg/dL. Most of these cats showed evidence of subtotal or total renal infarction, which the authors speculate may have resulted from thromboembolism of nonligated vessels or renal pedicle torsion.²¹ A reduction in GFR and an increase in serum creatinine and BUN have been reported between 2 and 8 wk postnephrectomy.^{14,17–19,21} Renal function most commonly remained static or showed mild improvement by 12 mo,^14–16 although one study²¹ reported an acute deterioration in renal function resulting in euthanasia in 40% of cats between 50 and 736 d postnephrectomy, with the severity of azotemia postnephrectomy predictive of median survival time. Improvement of renal function with time is likely referable to hypertrophy of the RK.¹⁷

There is variable induction of proteinuria, with some studies^14–17 reporting an increased mean urine protein-to-creatinine ratio (UPCR) 4 to 8 wk postnephrectomy, and others^18–20 reporting no difference in mean UPCR one to 11 wk postnephrectomy. In another study,²¹ 73.3% of cats had a UPCR less than 0.2, 20% had a UPCR of 0.2 to 0.4, and 6.7% had a UPCR greater than 0.4 from one to 45 mo postnephrectomy. An increase in mean systemic blood pressure has been reported from one week to 45 mo postnephrectomy.^14,18–20 Induced systemic hypertension is usually mild, with one study reporting that 53.3% of cats were normotensive, 26.7% prehypertensive, 6.7% hypertensive, and 13.3% severely hypertensive one to 45 mo postnephrectomy.²¹

Renal histopathologic changes in RKs are also variable. Mild mesangial matrix expansion, tubular lesions, interstitial fibrosis, interstitial inflammation, and glomerulosclerosis have been reported 6 to 12 mo postnephrectomy.^{14–16,18,20} These lesions did not occur in cats fed with low-protein and low-phosphorus diets.^14,15 In cats euthanized due between 50 and 736 d postnephrectomy due to a serum creatinine concentration greater than 12 mg/dL, anorexia for greater than 48 h, or other clinical signs unresponsive to medical therapy, renal histopathology findings included a focally extensive sterile abscess, infarction, necrosis, fibrosis, lymphoplasmacytic interstitial inflammation, and interstitial fibrosis.²¹

A variation of the RK model has been described,²⁰ which involves unilateral partial renal ablation by sectioning of the cranial and caudal poles (removing approximately 2/3 of renal mass), followed by wrapping the kidney in sterile silk and sterile cellophane. This was followed 2 wk later by partial ablation of the contralateral kidney by ligation of 5/6 of the branches of the renal artery. This model has been termed the “remnant-wrap technique.” The model induces a marked sustained systemic hypertension, decreased renal function, proteinuria, activation of the renin-angiotensin-aldosterone axis, and renal structural injury. The 24-h mean systolic blood pressure was significantly (P < 0.05) higher in cats with induced kidney disease using the remnant-wrap model, compared with the control group and cats in the RK group.²⁰

Similar to IRI models, RK models require a large amount of technical resources with the associated limitations on experimental size and the time frame required, representing a disadvantage of this model.

Combined Renal Embolism and Remnant Kidney Model

Recently, a minimally invasive partial kidney embolism combined with subsequent contralateral nephrectomy approximately 5 mo postembolism (day 150) in cats has been described.²² Delivery of embolic microparticles into a unilateral renal artery, followed by delayed contralateral nephrectomy, resulted in significant changes in biochemical markers of kidney function and histopathologic evidence of kidney fibrosis, tubular atrophy, and loss, as well as lymphoplasmacytic inflammation. All cats that completed the study had an International Renal Interest Society grade of ≥2 AKI.³³ Biomarkers of GFR postembolism appeared largely unchanged, which could hinder understanding of the effectiveness of the embolism procedures. In addition, the extent of chronic maladaptive changes was not evaluated; therefore, it remains to be determined if this model emulates naturally occurring CKD. Further research is needed to determine the optimal dose of embolic microparticles, the timing of the contralateral nephrectomy, and the time needed for stabilization following embolism and nephrectomy. Future studies could investigate the potential of modeling different stages of the disease by varying the interval between embolism and nephrectomy. Further characterization of this model is necessary; however, it is important to note that it introduces an added layer of complexity compared with the IRI and RK models, as it requires specialized expertise in interventional radiology.

Toxin-Induced Injury Models

Only one toxin-induced injury (TI) model has been described in cats, which employed an overdose of meloxicam (0.3 mg/kg SC every 24 h for 17 to 31 d) to induce AKI.^23,34 Although this protocol induced cortical and medullary damage in 4 out of 4 treated cats in one study,²³ AKI was only variably induced in another, with 2 out of 6 cats showing no increase in serum creatinine by day 31 and normal renal histopathology.^34,35 In this study, an increase in serum creatinine was first detected between days 13 and 26, following which it generally declined.³⁵ In 2 cats, serum creatinine was measured 16 d following the last meloxicam dose and was less than 1.6 mg/dL in both.³⁵ A return to baseline was seen in one cat, and an increase of 0.6 mg/dL from baseline was seen in the other. Renal histopathology showed multifocal cortical tubular dilation, cortical and medullary tubular degeneration and necrosis, and tubular (predominantly mononuclear) inflammation.³⁴ The mechanism of tubular injury induced by this model and factors influencing meloxicam tolerance are not known. Several renal cortex and medulla metabolomic and lipidomic changes caused by the repeated administration of meloxicam have been reported.²³ AKI in a rodent non‑steroidal anti‑inflammatory drug TI model occurred only in the context of water deprivation.²⁴ However, AKI was able to be induced without water deprivation in another rodent study,³⁶ and meloxicam has been associated with AKI in cats without apparent hemodynamic compromise.³¹ Future TI models using meloxicam could explore different dosing protocols to optimize the induction of AKI. Other TI models in rodents and other large animals have used cisplatin, rhabdomyolysis, gentamicin, and radiocontrast agents.⁸ However, the induction of kidney disease in cats with these drugs remains to be investigated.

TI models are inherently limited by doses and/or a combination of insults that do not occur under field conditions,⁸ and as such, the extent to which findings can be extrapolated to cats with naturally occurring kidney disease is not known. One of the most appealing aspects of TI-based models, compared with IRI and RK models, is that they require comparatively fewer technical resources. The extent of the renal injury may be controlled by adjusting the dose of toxin, and repeated dosages allow repeated insults to be modeled. These models also allow the induction of bilateral lesions, which is more representative of naturally occurring feline kidney disease. It is not known whether feline TI models induce hypertension, proteinuria, or CKD.

Postrenal Obstruction Model

PO of cats has been induced in one study²⁵ by urinary catheterization of male cats followed by catheter occlusion. Mean serum creatinine increased to greater than 2.6 mg/dL approximately 9 h postocclusion and increased to greater than 5.1 mg/dL after approximately 30 h. Mean serum creatinine subsequently returned to normal 72 h following relief of the occlusion. It is unclear if renal histopathologic lesions suggestive of intrinsic AKI occurred,²⁵ and further characterization of this model is needed before researchers can use it. Another group of researchers induced kidney disease by ligating one of the ureters for 28 d, combined with a contralateral nephrectomy on day 42 in 2 cats.³⁷ In one of the cats, serum creatinine, BUN, and symmetric dimethylarginine (SDMA) peaked on day 46 (serum creatine: 2.8 mg/dL; BUN 51.6 mg/dL; SDMA, 28 μg/dL) and then stabilized within their reference ranges. Urine specific gravity remained in the range of 1.015 to 1.066 until the end of the study. In the second cat, serum creatinine, BUN, and SDMA increased 2 d after the right nephrectomy and then stabilized until day 456, when they increased again and peaked on day. Mild interstitial fibrosis, interstitial inflammation, and tubular atrophy were observed.³⁷

The Problem of Existing Biomarkers of Feline Kidney Disease

A biomarker is a measurable indicator of a biologic state or condition.³⁸ In the context of experimental models of kidney disease, biomarkers are crucial for confirming model induction, monitoring disease progression, and evaluating the response to therapeutic interventions. Unfortunately, validated biomarkers for feline kidney disease are limited to biomarkers of GFR (for example, serum creatinine and SDMA), urine protein and glucose, and urine cast formation, all lacking high sensitivity and/or specificity.^39,40 While urinary biomarkers of tubular injury (for example, kidney injury molecule-1, neutrophil gelatinase-associated lipocalin, cystatin B) have been explored as potential earlier markers of kidney disease, their analytical validity and clinical utility require further investigation.³⁸

One significant limitation of biomarkers of GFR is that they fail to capture cellular events. In addition, there is a significant lag time between the initial renal insult and deterioration in GFR.^35,41 The practical challenge of this temporal disconnection between renal insult and the detection of altered GFR is the delayed confirmation of successful model induction. Existing biomarkers are also insensitive for identifying renal tubular changes, adaptive and maladaptive repair processes, and the extent of the lesions,^35,42 hindering the assessment of pharmacological interventions, such as antifibrotic drugs.

Undoubtedly, there is an urgent need for additional biomarkers to harness the potential of kidney disease models fully. Researchers aiming to induce kidney disease for purposes other than identifying novel biomarkers must thoroughly understand the limitations of currently validated biomarkers within the context of the specific model being used.

Future Directions and Conclusions

In addition to consuming large amounts of technical resources, the in vivo models described above also have substantial ethical and animal welfare ramifications. Recently, kidney organoids generated from human pluripotent stem cells have been used to model various human glomerular and tubular diseases in vitro.^8,43–45 However, current protocols produce organoids that mostly resemble fetal kidney tissue,⁴⁴ which may limit their utility in cats, where renal disease is most common in aged animals.^3,4 Kidney organoids include off-target cell populations and are often not appropriately organized, intrinsically forming a heterogeneous cell mass. Accordingly, the extent to which organoids exhibit normal kidney function (for example, tubular reabsorption/secretion and the production of hormones) remains unclear.^44,45 This problem is overcome with organs-on-chip technology, which utilizes constructs containing various compartments and multiple cell types to replicate tissue-tissue interfaces.⁴⁴ Functional glomerular and tubule chips have been developed,⁴⁵ although their adoption is limited by the substantial technical expertise required as well as access to the specific differentiated kidney cells required to build the desired chip. Feline kidney organoids or organ chips have yet to be described and deserve future research.

Modifications to existing in vivo models could be made to better replicate episodes of AKI thought to contribute to feline CKD.⁵ For example, an IRI model using serial, shorter clamp times, ideally performed employing minimally invasive surgical techniques, may be a more accurate model of naturally occurring kidney disease than single, prolonged clamp times. Prolonged, partial clamping could also be explored, given evidence for increased renal vascular resistance in various feline kidney diseases.⁴⁶ Measurement of ultrasonographic indices of renal vascular resistance⁴⁶ would determine if these in vivo models replicate this increased renal vascular resistance found in naturally occurring diseases. Finally, using minimally invasive renal biopsy techniques may allow serial renal biopsies to be taken following the initial injury, without substantially impairing renal function.⁴⁷

To summarize, feline IRI models with ≥60-min clamping of the renal vasculature produce reliable ischemic AKI. Bilateral clamping is not recommended due to a high risk of perioperative mortality. Unilateral clamping may be combined with delayed contralateral nephrectomy, which allows the function of the injured kidney to be more accurately assessed over time. RK models produce reliable CKD with variable induction of proteinuria and hypertension. A TI model using meloxicam produced variable AKI, and whether PO models induce intrinsic renal disease in cats is not currently known. The external validity of these models is unknown, given the range of possible etiologies of feline renal disease that is most commonly idiopathic. Despite this, the use of these models may facilitate the discovery of novel therapeutics⁴⁸ and early biomarkers of feline renal disease,^23,34 facilitating diagnosis of naturally occurring disease when initiating factors are still present.

None of these models perfectly replicates all the events associated with naturally occurring kidney disease. The choice of model should be made carefully, based on the specific lesion (for example, tubular damage compared with fibrosis) or event (for example, diminished GFR) to be mimicked, considering the strengths and limitations of each available option. Moreover, researchers must give special attention to the biomarkers selected, ensuring their reliability in confirming disease induction, evaluating the extent of induced changes, and tracking disease progression.

Conflict of Interest

The authors have no conflicts of interest to declare.

Funding

The Dr. Villarino kidney research program is supported by the Washington State University College of Veterinary Medicine Kay Yarborough Nelson Distinguished Professor in Feline Health Endowment, a generous gift from Mary Kay Fowler, Washington State University intramural grants program, and Washington Research Foundation.