The aging feline kidney: A model mortality antagonist?

Abstract

Traditional thinking views apparently non-programmed disruptions of aging, which medical science calls geriatric diseases, as separate from ‘less harmful’ morphological and physiological aging phenotypes that are more universally expected with passage of time (loss of skin elasticity, graying of hair coat, weight gain, increased sleep time, behavioral changes, etc). Late-life disease phenotypes, especially those involving chronic processes, frequently are complex and very energy-expensive. A non-programmed process of homeostatic disruption leading into a death trajectory seems inconsistent with energy intensive processes. That is, evolutionary mechanisms do not favor complex and prolonged energy investment in death. Taking a different view, the naturally occurring feline (Felis silvestris catus) renal model suggests that at least some diseases of late life represent only the point of failure in essentially survival-driven adaptive processes. In the feline renal model, individuals that succumbed to failure most frequently displayed progressive tubular deletion and peritubular interstitial fibrosis, but had longer mean life span than cats that died from other causes. Additionally, among cats that died from non-renal causes, those that had degrees of renal tubular deletion and peritubular interstitial fibrosis also had longer mean life span than those cats with no changes, even though causes of death differed minimally between these latter two groups. The data indicate that selective tubular deletion very frequently begins early in adult life, without a clear initiating phase or event. The observations support a hypothesis that this prolonged process may be intrinsic and protective prior to an ultimate point of failure. Moreover, given the genetic complexity and the interplay with associated risk factors, existing data also do not support the ideas that these changes are simple compensatory responses and that breed- or strain-based ‘default’ diseases are inevitable results of increasing individual longevity. Emerging molecular technology offers the future potential to further evaluate and refine these observations. At present, the existence of plastic and adaptive aging programming is suggested by these findings.

Diseases of advanced life traditionally have been viewed as disruptive, non-adaptive events that hold the potential to end life. These more ominous events have been segregated, at least conceptually, from more benign ‘expected’ aging phenotypes such as loss of skin elasticity, graying of the hair loss, weight gain, increasing sleep time, or behavioral changes. This segregated understanding is reflected in cataloging approaches to describing diseases of aging and recommending preventive or therapeutic actions (Hoskins 1995, Kraft 1998). In domestic cats (Felis silvestris catus), renal failure is a common example of potentially terminal late-life disease that seems to align with this traditional view.

In a different view, aging phenotypes often are associated with complex underlying biology at genomic, cellular, organ, and systems levels. Prolonged investment of metabolic energy at this level of complexity seems inconsistent with events that simply disrupt ‘natural aging’ to initiate a death trajectory (Heininger 2001, 2002). Rather, late-life diseases perhaps should be regarded as an integral part of the aging process, for which response programming is inherently plastic and highly adaptable (Heininger 2001, 2002). In this context, our data support an alternative hypothesis that the example phenotype of overt renal failure in domestic cats represents only the point of eventual failure of a lengthy but essentially survival-driven adaptive process.

Materials and methods

General methods

The relationship of renal changes to aging and mortality was evaluated from applicable components of a database consisting overall of post-mortem findings from 676 adult domestic cats. The data were collected over the years 1979–2001. All of the cats were maintained for life as residents of the same colony. They were housed indoors in cages or group pens, under the same environmental and health management system. Their environment was maintained at approximately 20–24°C; relative humidity varied around a 65% average. The colony was free of internal and external parasites and retroviruses, and routine vaccinations and health care were administered under established and supervised protocols. The population was reproductively self-supporting, with periodic additions of breeding cats to maintain management practices that minimized inbreeding. Known heritable diseases occurred infrequently because of careful breeding practices.

All cats were monitored daily and were evaluated by animal health technicians and by the attending veterinarian when quality of life was judged to be deteriorating in response to naturally occurring events. However, humane euthanasia was carried out only after necessity was established by thorough diagnostic assessment, sequential evaluation of clinical condition and quality of life, and consideration of prognosis, according to pre-established and uniformly applied colony practices.

Postmortem procedures

The study database included biographical data (signalment), gross necropsy, selected tissue histopathology, and related epidemiology. Gross necropsy was conducted as close to the time of death as possible, with observations recorded electronically during the procedure. Kidney and other tissues were harvested and preserved in 10% buffered formalin, processed and sectioned routinely, and stained with hematoxylin and eosin or other stains as required. Kidney tissue samples usually were obtained by bilateral full or partial mid-sagittal section, but lesions observed in other kidney areas also were included in sampling. Primary causes of death were confirmed by assessing the clinical course, results of gross post-mortem examination, and histopathological outcomes.

Analysis of renal morphology and age relationships

A morphological scoring system was established to facilitate analysis of renal histological observations. The components of this score were developed from standard histological descriptors for lesion (a) location (involved tissues within the organ), (b) region (dissemination of changes within the organ), (c) degree (severity), and (d) character (acute, chronic-active, chronic) (Table 1). This ordinal scoring system was evaluated within the study population for age-associated changes among its components, and for frequencies of specific observations.

Table 1.

Scoring system for renal histopathology

	Point assignment
Location
Glomerular	1
Tubular	1
Interstitial	1
Papillary	1
Pelvic	1
Vascular	1
Region
Focal	1
Multifocal	2
Focally disseminated	3
Coalescing	4
Diffuse	5
Degree
Mild	1
Mild–moderate	2
Moderate	3
Moderate–severe	4
Severe	5
Character
Acute	1
Chronic-active	2
Chronic	3
Total possible score	19

The data were first categorized by renal or non-renal cause of death for evaluation of age at death, both overall and within sequential age categories. Body condition of cats with renal cause of death was noted at necropsy; cats with non-renal cause of death were further segregated to compare age at death within body condition. These comparisons were done using Student's t-test. Tubular, interstitial, and glomerular locations then were evaluated specifically, using pair-wise comparison by age of cats that displayed those histological changes. Among cats that died from non-renal causes, causes of death also were compared between the two subclasses that either had or did not have histological renal changes at the time of death. These data were compared using χ² tests. SAS software was used for all data analysis except for heritability (SAS Institute 1999).

Recognizing the insensitivity of standard clinical chemistry to early and mid-term progressing changes, clinical reflection of ongoing renal changes was examined by measuring blood pressure in a cohort of 56 of these colony cats of various ages. The data were age-categorized and subjected to analysis of variance and regression analysis (Lawler et al 1995).

Heritability

Morphological observations of renal tissue from gross post-mortem examination and histopathological outcomes were defined as traits and were evaluated for heritability characteristics. The same procedure was applied to outcomes of serial Dual-Energy X-ray Absorptiometry (DEXA) evaluations of a cohort of 119 cats from this population, to further examine the relationship between body composition and aging. The major independent factors and the individual traits were evaluated for heritability using the ‘polygenic’ function of Sequential Oligogenic Linkage Analysis Routines (SOLAR) (Almasy and Blangero 1998). This function uses restricted error maximum likelihood to iteratively estimate the additive genetic variance (Lynch and Walsh 1998). Principal Component Analysis (PCA) was used to separate phenotypic variation into major independent factors.

Results

Renal morphology and age

The amount of change in the renal histopathology score and in the components location, region, degree, and character, was evaluated by age at death among 600 cats. Scores for these categories increased significantly (P<0.01) with older age at death (Table 2). The majority of histological changes involved the tubules and interstitium. Tubule loss with replacement fibrosis and mineralization, and peritubular interstitial fibrosis, were observed frequently. Medullary rays usually were affected prominently. Descriptively, the predominant changes were multifocal, moderate, and chronic-active. Focal, mild, and acute lesions were very infrequent (Table 3).

Table 2.

Analysis of variance for mean change in renal histology score components (cross-sectional, months of age at death, n=600)

	Change	P-value
Total score	0.034	<0.001
Location	0.002	<0.001
Region	0.008	<0.001
Degree	0.010	<0.01
Character	0.007	<0.001

Table 3.

Frequency of feline renal histological changes by location, region, degree, and character

Location
Pelvic	11
Papillary	30
Vascular	6
Tubular	432
Interstitial	412
Glomerular	75
Region
Focal	13
Multifocal	256
Focally disseminated	33
Coalescing	5
Diffuse	158
Degree
Mild	36
Mild–moderate	68
Moderate	188
Moderate–severe	62
Severe	111
Character
Acute	10
Chronic-active	281
Chronic	175

Cats that died or were euthanatased because of tubulointerstitial renal disease lived significantly (P<0.001) longer than those that died from other causes (Table 4). Among cats that had renal cause of death, body condition at the time of death was thin, normal, or obese in 48%, 34%, and 13%, respectively. Among cats that died from non-renal causes but had histological renal changes, mean life span also was longer than those cats without renal changes or renal cause of death, regardless of body condition at the time of death (Table 4).

Table 4.

Relationships of age to cause of death

Cause of death	N	Mean age at death (months)	P-value
Renal disease	187	140
Non-renal*	304	120	<0.001

*Non-renal death	Lesions	N	Mean age (months)	P-value
Thin body condition	Renal (+)	68	148
	Renal (−)	16	97	<0.001
Normal body condition	Renal (+)	50	116
	Renal (−)	18	74	<0.002
Obese body condition	Renal (+)	46	143
	Renal (−)	16	112	<0.001

When age-at-death categorized data were evaluated, cats in age categories older than about 108 months generally had greater prevalence of mortality due to renal disease (Table 5). Cats that died because of renal failure had higher but uniform mean histological scores across most age categories, compared to cats with other causes of death, but the latter group also frequently displayed renal changes, even in early adulthood (Table 5).

Table 5.

Age-based evaluation of renal changes (score) by cause of death

Age at death (month)	Renal cause death		Non-renal cause death		P-value
	N	Mean score *	N	Mean score
12–48	08	11.89	75	4.16	<0.001
49–72	13	12.85	44	7.11	<0.001
73–84	06	11.67	25	6.32	<0.05
85–96	09	12.22	25	7.20	<0.001
97–108	09	11.78	24	7.54	<0.05
109–120	18	13.39	26	7.00	<0.001
121–132	14	12.64	32	8.81	<0.05
133–144	10	10.80	16	11.06	>0.05
145–156	16	11.38	39	8.90	<0.05
157–168	17	11.24	29	8.97	<0.05
169–180	21	11.24	30	9.40	<0.05
181–192	19	11.58	17	10.18	>0.05
>192	22	12.22	15	8.13	<0.05

^*Mean histological score.

Pair-wise comparisons of tubular, interstitial, and glomerular sites showed that tubular change in particular occurred more frequently among older age groups (Table 6). Glomerular changes occurred much less frequently, in approximately 11% of the population. However, cats having glomerular changes died at a younger mean age than those without glomerular changes (119.5 vs 146.7 months, respectively, P<0.0008).

Table 6.

P-values for pair-wise comparisons among tubular (t), interstitial (i), and glomerular (g) renal histopathology (n=normal) for dependent variable=Age: PR>|t| for LSMean (i)=LSMean (j)

i	j *
	n	g	i	ig	t	tg	ti	tig
n		0.0487 †	0.2236	0.9583	<0.0001	0.0009	<0.0001	<0.0001
g	0.0487		0.2720	0.3633	0.8536	0.4456	0.3472	0.3543
i	0.2236	0.2720		0.7359	0.0443	0.0304	0.0002	0.0009
ig	0.9583	0.3633	0.7359		0.2681	0.1546	0.1253	0.1265
t	<0.0001	0.8536	0.0443	0.2681		0.4099	0.1117	0.1677
tg	0.0009	0.4456	0.0304	0.1546	0.4099		0.9400	0.9678
ti	<0.0001	0.3472	0.0002	0.1253	0.1117	0.9400		0.9374
tig	<0.0001	0.3543	0.0009	0.1265	0.1677	0.9678	0.9374

^*Comparisons represented are i (vertical column) to j (horizontal row).

^†Bold typescript denotes significant differences, P<0.05.

Among cats that died from non-renal causes, only the category that included multiple causes of death was different between the two subclasses (Table 7), demonstrating that different age-based mortality patterns were not responsible for the life span observations.

Table 7.

Comparison of renal tubulointerstitial changes by causes of death among cats dying of non-renal causes

Cause of death	Renal changes (+) *	Renal changes (−) *	P-value for difference †
Neoplasia	20 (10.3%)	9 (14.1%)	0.40
Heart disease	38 (19.5%)	9 (14.1%)	0.33
Pancreatitis	8 (4.1%)	3 (4.7%)	0.84
Diabetes mellitus	20 (10.3%)	10 (15.6%)	0.24
Liver disease	10 (5.1%)	3 (4.7%)	0.89
Inflammatory bavel disease	13 (6.7%)	5 (7.8%)	0.75
Oral cavity disease	6 (3.1%)	2 (3.1%)	0.98
Respiratory disease	25 (12.8%)	11 (17.2%)	0.38
Neurological disease	5 (2.5%)	3 (4.7%)	0.39
Osteoarthritis	5 (2.5%)	2 (3.1%)	0.81
Infection ‡	9 (4.6%)	3 (4.7%)	0.98
Dermatopathy	3 (1.5%)	1 (1.6%)	0.99
Accidental	2 (1.0%)	2 (3.1%)	0.24
Multiple	31 (15.9%)	1 (1.6%)	<0.01
n	195	64

^*Number (percent of population).

^†χ²

^‡Includes a variety of viral and bacterial diagnoses.

Analysis of cross-sectional blood pressure data using age categorization indicated that cats over age 3.8 years had higher blood pressures in general than younger cats, without subsequent age-related increases (Table 8). Regression analysis of individuals aged 6.2 years or less indicated that systolic arterial pressure (SAP), diastolic arterial pressure (DAP), and mean arterial pressure (MAP) increase at respective annual rates of 6.1, 8.5, and 7.5 mmHg over these years (Table 8).

Table 8.

Mean blood pressure ^* of adult cats by age group

Age group (months)	N	SAP (mmHg)	DAP (mmHg)	MAP (mm Hg)
2.2–3.8	16	137.3 A , y	102.5 A	114.5 A
4.0–6.2	12	151.3 AB , z	122.7 B	132.1 B
9.0–13.7	28	157.0 B	124.2 B	135.6 B

^AB=Means (columns) with different letters differ (P<0.05).

^yz=Means (columns) with different letters differ (P<0.10). SAP, DAP, MAP=systolic, diastolic, mean arterial pressure.

^*Least squares means.

Heritability

There was no significant heritability for the renal phenotypes, indicating that segregating (quantitative) genetic factors could account for no more than 20% of the observed morphological variability. Therefore, the renal phenotypes likely resulted from environmental (non-programmed) factors or fixed genetic traits, or from their interaction, but not directly from quantitative heritability alone.

Evaluation of serial DEXA data from 119 adult cats yielded between 1 and 12 scan replications during the lifetime of individuals. Heritability and principal component analyses of DEXA data suggested that multiple genes are involved in the expression of components of body composition (Table 9). Two heritable principal components, PC2 (h²=0.40, P<0.01) and PC6 (h²=0.74, P<0.01), explained 24.7% and 1.3% of the population variance, respectively (Table 9). The first principal component, PC1, which accounted for 55.0% of population variance (h²=0.33), was less strongly significant (P=0.038), possibly as a result of the number of variables tested. Lean, fat, and bone components of body composition tended to relate moderately to one another.

Table 9.

Heritability and principal component (PC) analyses of serial dual-energy X-ray absorptiometry (DEXA) data from 119 adult cats

	PC1	PC2	PC3	PC4	PC5	PC6	PC7
Percent of variance	55.0	24.7	12.2	4.1	2.2	1.3	0.5
Heritability	0.33*	0.40	0.00	0.21	0.11	0.74	0.00
Trait	Loadings
	PC1	PC2	PC3	PC4	PC5	PC6	PC7
Bone density	0.30	−0.07	0.85	−0.11	0.29	−0.29	−0.06
Tissue % fat	0.33	0.54	−0.08	−0.48	−0.21	0.06	−0.57
Body weight	0.37	−0.32	−0.49	−0.23	0.63	−0.25	−0.08
Body condition score	0.38	0.41	−0.01	0.64	0.36	0.39	−0.03
Lean grams	0.39	−0.43	−0.06	0.45	−0.49	−0.25	−0.40
Bone mineral content	0.41	−0.40	0.09	−0.30	−0.17	0.70	0.23
Fat grams	0.45	0.30	−0.13	−0.04	−0.29	−0.38	0.68

^*P=0.038 (suggests statistical significance given the number of traits tested). Bold typescript indicates significantly contributing loadings.

Discussion

Population maintenance and health surveillance practices in this large colony excluded or controlled infectious diseases, and environmental conditions were regulated and monitored. Inbreeding was minimal, known genetic diseases were diagnosed infrequently, and quantitative heritability appeared to exert little influence on the data, except for body composition. Therefore, it is not likely that our observations are consequent to the structure of the colony.

Older cats succumbed to overt renal disease with increased frequency, an observation that is compatible with reports that describe the general (community-based) cat population (Hamilton 1966, Lucke 1968, Lulich et al 1992). Histological changes in our population also are compatible with observations in previous reports of community-based populations (DiBartola et al 1987, Finco and Brown 1995). However, our data show clearly that cats with tubulointerstitial changes had longer life span regardless of cause of death, despite the fact that initial changes often can be demonstrated at the early adult life stage.

When cats that succumbed to non-renal causes of death were evaluated across categories of death-causing diseases, the only significant difference was a higher frequency among cats with renal changes in the category of multiple interacting causes of death, compared to cats with no renal changes. However, given that these also were generally older cats, this should not be surprising. In addition, multiple interacting causes of death in older cats typically have a renal component, indicating little likelihood of incorrect categorization. The numbers within a few cause-of-death categories were very small, and therefore these sparse data would not influence overall outcomes and interpretations in any meaningful way. These data demonstrate that other problems did not underlie the difference in mean age at death between the two groups without renal cause of death, and further underscore the observation that the outcomes of our study were not consequent to the structure of the colony.

Life-limiting glomerulopathy and severe inflammatory forms of interstitial renal disease may represent superimposed influences that occur less frequently. Recent research suggests candidate examples for these influences, such as incremental tubular damage caused by protein leakage through damaged glomeruli (Schieppati and Remuzzi 2003, Lees and Tryggvason 2004) or development of complications such as systemic hypertension (Brown et al 2004). Mean SAP values that were found in cats over age 3.8 years in our study could represent hypertension in some measurement systems but not others (Lawler 1997). Interestingly, cats in this colony rarely exhibited signs of systemic hypertension, and the observations do not suggest increasing progression into late middle age and senescence. Thus, the age distribution of the cats and the covert nature of the observed blood pressure differences are compatible with a hypothesis for secondary reflection of ongoing (adaptive) renal changes.

Cats that succumbed to renal failure frequently had morphological and pre-terminal metabolic changes indicating varying but substantial retained functional capacity, suggesting that other factors may direct actual transition to failure. Candidate causes for this latter phenomenon also can be recognized. For example, cellular stress induced by oxidative damage may play a ‘downstream’ role in ultimate organ or system failure (Busuttil et al 2003, Sastre et al 2003), but must necessarily reflect ‘upstream’ triggering mechanisms for changing patterns of gene expression.

Usual sequelae of primary ischemia (cell swelling, karyolysis, lysosomal rupture, and massive inflammation) are not commonly observed in the cat kidney or in a similar mouse model (Majno and Joris 1995, Schelling et al 1998, Sawashima et al 2000). In a feline model for renal tubular disease, the associated fibrosis was principally peritubular, suggesting that hypoxic sequelae may result mainly from local compromise of diffusion of oxygen and other nutrients (Sawashima et al 2000). In a murine renal model, upregulated FAS in tubular epithelial cells was shown to bind to FAS-ligand of adjacent tubular cells, suggesting that tubular loss is a fratricidal apoptosis that depends on specific signals (Schelling et al 1998). The low frequency of acute renal lesions among a large number of cats that died from acute or chronic non-renal causes in our study suggests that a discreet initializing event, such as ischemic episode, infection, or toxicity, may not be a prominent feature of this process. Thus, the histological nature of tubule loss in the feline and murine models both appear to be more compatible with an adaptive mechanism that operates at the cellular level to delete dysfunctional renal tubular cells and nephrons, and appear much less compatible with a prolonged energy-expensive investment in death.

Frequent observation of tubulointerstitial changes in younger adults documents early onset and also is compatible with the possibility that renal changes may represent a defensive adaptation, as symptomatic renal failure is less frequent during early adulthood and very common during late adult life. Underlying explanations for this phenomenon are not established at present. While domestic carnivores do not undergo a menopause, fertility clearly declines by about year 8 in queens, and slightly later in toms (Scott 1970, Lawler and Monti 1984, Lawler and Bebiak 1986). A renal adaptation as we describe could, for example, increase the likelihood of successful reproduction through greater metabolic stability. Thus, the onset of tubulointerstitial changes that appears well prior to reproductive senescence might have evolved in part as one homeostatic response to help preserve fertile reproductive life span through selective elimination of dysfunctional renal tubular cells and nephrons. However, this possibility also provides insufficient explanation for the long post-reproductive life span of domestic cats, suggesting at the least that additional genetic programming or epigenetic interaction may be involved. Indeed, the hypothesis that mechanistic programming for stress responses has evolved to modulate outcomes of intrinsic and extrinsic insults (albeit with species specificity) and thus delay organismal death, is within the present scope of the debate about aging theories (Bredesen 2004).

Deterioration during more advanced stages of chronic renal failure often is accompanied by progression to cachexia (Lulich et al 1992), although factors that influence body composition during late life may not reflect only secondary, pre-terminal degenerative processes. In studies of aging populations across species, death tends to be associated with more precipitous declines of body mass that are recognizable around the time that late-life population mortality increases (Lesser et al 1973, Lesser et al 1980, Yu et al 1982, Kealy et al 2002, Lawler et al 2005). In our study, the higher percentage of cats with kidney-related death and thin body condition suggests at least some prior transition from more obese body condition. This observation aligns well with typical clinical observations of early onset of very gradual change of body mass in (eventual) renal patients.

The observation that body composition has a quantitative genetic component was unexpected and indicates a need for re-evaluation of the underlying role of body composition during aging. Interestingly, lean, fat, and bone components of body composition related to one another only moderately in the PCA, suggesting further that expression of these individual body composition variables might result from multiple (and very possibly separate) underlying genetic and/or non-programmed processes. It is therefore possible that slowly progressive, pre-terminal loss of body mass reflects additional adaptive genetic programming, at least until late progressive stages are reached. In this context, death occurs as an apparent pathological outcome only at the point of multiple systemic adaptive failures that are associated with more advanced nephron deletion, but are driven by extra-nephron, extra-renal, or extrinsic metabolic factors.

Genetic evaluation did not reveal heritable components of renal tubulointerstitial phenotypes in this population, indicating that the phenotypes are either totally environmental in origin or they reflect fixed traits in the feline genome (or both). In this regard, it is well to first remember that aging, although reflecting response programming, also remains a highly plastic process. That is, aging is subject to non-programmed interactions with stress-resistance phenotypes that may be fixed (Clare and Luckinbill 1985, Finch 1990, 1997, Buck et al 1993, Scheiner 1993, Krebs and Loeschcke 1999, Martin 2005). Therefore, heritabilities of these likely would be quite modest in any event, and of course would not be expected in the case of fixed traits. Further, the possible influence of modulating interactions between fixed alleles and epigenetic influences is compatible with a working hypothesis that ultimately may re-characterize the role of overt disease in aging, centered about an emerging understanding of the role of genetic–epigenetic interactions (Jazwinski et al 1998, Robert and Labat-Robert 2000, Feinberg et al 2002, Issa 2002, Claus and Lubbert 2003, Kopelovich et al 2003).

The consequence of these observations ultimately may involve altering approaches to intervention–prevention paradigms. It must be recognized that at least some components of long-term intrinsic disease (aging) processes likely represent life-preserving adaptations. This, in turn, strongly indicates that current views of the role of ‘disease’ in aging are in need of significant reconsideration.