|Home | About | Journals | Submit | Contact Us | Français|
Aging is both a natural and inevitable biological process. With advancing age, the kidneys undergo anatomical and physiological changes that are not only the consequences of normal organ senescence but also of specific diseases (such as atherosclerosis or diabetes) that occur with greater frequency in older individuals. Disentangling these two processes, one pathologic and the other physiologic, is difficult. In this review we concentrate on the glomerular structural and functional alterations that accompany natural aging. We also analyze how these changes affect the identification of individuals of advancing age as having chronic kidney disease (CKD) and how these changes can influence prognosis for adverse outcomes, including all-cause mortality, end-stage renal disease, cardiovascular events and mortality, and acute kidney injury. This review describes important shortcomings and deficiencies with our current approach and understanding of CKD in the older and elderly adult.
Normal aging of the kidneys must be discussed in the context of the various meanings of the term normal.1 Epidemiologists often use the term in a statistical sense to describe the usual range of variation, both in health and disease, as it appears in the unselected general population. The use of this definition and the currently used classification systems for chronic kidney disease (CKD) explains the very high reported prevalence of CKD in the elderly. CKD is diagnosed in healthy elderly subjects whose renal function is below a notional normal range. The more colloquial definition of normal implies good health for age. How does the anatomy of the kidney change with ‘natural and healthy’ aging in the absence of either a specific kidney disease or CKD risk factors? The term ‘senescence’ as opposed to ‘disease’ may also be used to describe these more predictable and physiological changes with aging. This latter approach to defining normal is commonly used for laboratory reference ranges.2 Further, with living kidney donation there is a need to clarify what is normal for the age of the kidneys, lest only young adults be permitted to be kidney donors. These living kidney donors also provide a unique opportunity to study normal aging by this latter definition, as kidney donors are systematically evaluated for CKD, and kidney tissue is often available for microscopic examination. Although not every marker of CKD has been discovered, and occult disease may remain undetected, existing evidence finds that even 70-year-olds with optimal health for their age have features of senescence in their kidneys not seen in healthy 20-year-olds. This controversy has been reviewed on many occasions,3–9 but this review will focus on a contemporary analysis of the changes of glomerular structure and function with aging and their clinical consequences.
The age-related findings on light microscopic evaluation of kidney biopsies can be divided into two groups: (1) nephrosclerosis and its features (glomerulosclerosis, tubular atrophy, interstitial fibrosis, and arteriosclerosis) and (2) morphometric analysis of microanatomy (particularly glomerular size). Numerous studies have shown an increased proportion of globally sclerotic glomeruli with aging.10–16 Even in living kidney donors there is an increased prevalence of glomerulosclerosis on the renal biopsy throughout the age spectrum: 19%, 38%, 47%, 65%, 76%, and 82% in the 18- to 29-, 30- to 39-, 40- to 49-, 50- to 59-, 60- to 69-, and 70- to 77-year-old age groups, respectively.16 The glomerulosclerosis that occurs with aging has an ischemic appearance with tuft collapse and intracapsular fibrosis, suggesting a primary vascular origin for the lesions. Some functional glomeruli also show ischemic capillary wrinkling of tufts, thickening of basement membranes, and mild intra-capsular fibrosis, all of which are precursors for glomerulosclerosis. Over time, shrinkage of the glomerular tufts toward the vascular pole with eventual sclerosis and collagen deposition filling Bowman’s space develops.17,18 Besides glomerulosclerosis, increased arteriosclerosis (fibrointimal hyperplasia), medial hypertrophy, and arteriolar hyalinosis occur with aging.12,16 With the sclerosis of juxtamedullary glomeruli, there is formation of a direct connection between afferent and efferent arterioles that bypasses the tuft.19 Tubular atrophy with surrounding areas of interstitial fibrosis also increases with aging.15,16,20
Glomerulosclerosis, arteriosclerosis, tubular atrophy, and interstitial fibrosis occur commonly together, and this constellation of findings constitutes nephrosclerosis. Indeed, it is hypothesized that fibro-intimal hyperplasia in small arteries with aging leads to glomerulosclerosis, followed by local tubular atrophy and interstitial fibrosis.21 A sclerosis score can be defined by the number of these different histological abnormalities present on an implantation biopsy of living kidney donors (Figure 1). Nephrosclerosis, defined by two or more of these abnormalities, increases progressively with age: 2.7% for 18- to 29-, 16% for 30- to 39-, 28% for 40- to 49-, 44% for 50- to 59-, 58% for 60- to 69-, and 73% for 70- to 77-year-olds. Nephrosclerosis may be universal in centenarians. An age-related decline in glomerular filtration rate (GFR) was not explained by these histological changes occurring with normal aging.16
Discrepant findings exist among studies of changes in the average glomerular size with aging. Some have reported the glomerular surface area or volume to decline with age,14,22 whereas others have found no change11 or an increase with age.23–25 It is not clear from these studies whether sclerotic glomeruli were excluded or included in the measurement of average glomerular size. The sclerotic glomeruli seen with aging are smaller than functional glomeruli and would contribute to any decline in average glomerular size with aging. However, compensatory hypertrophy of the remaining functional glomeruli also develops in response to sclerotic (presumed nonfunctional) glomeruli.24,26 Thus, both an increase in the proportion of small sclerosed glomeruli and an increase in the size of functional glomeruli may occur with age.
Besides glomerular size, glomerular density (number of glomeruli per area of cortex) is an inversely proportional surrogate for the average nephron size.27 Glomerular density inversely correlates with glomerular size and is a potent predictor of GFR decline in early IgA and membranous nephropathy.28,29 A decrease in glomerular density with aging is noted in biopsy samples where sclerotic glomeruli are <10% of the total glomerular population, consistent with an increased size of glomeruli and tubules with aging. Conversely, glomerular density increases with aging in biopsy samples where sclerotic glomeruli are >10% of the total glomerular population (Figure 2). This latter process becomes more dominant with age, as the proportion of small sclerotic glomeruli increases and there is more tubular atrophy.27 Autopsy microdissection studies have shown that tubular diverticula are also observed with increased frequency with aging.14
Autopsy studies allow thorough assessment of the macro-anatomy of the kidney, as well as concurrent sectioning and staining for the microanatomy of the cortical parenchyma. The main disadvantage of autopsy studies for studying normal aging is that the underlying renal pathology may have contributed to the cause of death. Substantially less renal pathology is observed among deceased kidney donors compared with autopsied patients at the same age, even after excluding known CKD and CKD risk factors.15 In autopsy series, a decrease in the kidney weight with age appears to begin in the fourth to fifth decade and results in a 10–30% decrease in kidney weight by the seventh to eighth decade.30,31 In a study limited to normal individuals with sudden death, there was no association between kidney weight and age after adjusting for body surface area.32
Assessment of the kidney size and other anatomical changes with age can be performed readily with ultrasound and computed tomography, in healthy individuals. As with autopsy studies, imaging studies in populations that exclude persons with comorbidity show less decline in the kidney volume with age than the studies that do not have this exclusion. In particular, studies in potential kidney donors have not found any evidence of a decline in kidney volume by computed tomography scan with age, although few persons over the age of 60 years were included.33–35 Factors other than age that seem to be more strongly correlated with kidney parenchymal volume include gender, body surface area (height and weight), and GFR. Imaging studies in populations with less exclusion of comorbidity, with larger sample sizes, and with older adults do find a decline in kidney parenchymal volume with age.36–38 The loss of kidney mass with age is accelerated with severity of atherosclerosis (as estimated from carotid intimamedia thickness).39 An increase in renal sinus fat with age may compensate for the decrease in the kidney parenchymal volume with age.36,37
Given the marked increase in glomerulosclerosis and tubular atrophy with age in kidney donors,16 why does kidney volume not decline in this population? The answer appears to be a compensatory increase of the volume of unaffected nephrons in response to loss of nephrons affected by glomerulosclerosis and tubular atrophy.11 There is both increased volume of functional glomeruli and decreased glomerular density with aging.23,24,26,27 There may be a threshold to this compensation around the age of 60 years, when the glomerular density no longer decreases because of enhancement of lesions of glomerulosclerosis and tubular atrophy (Figure 2).27 This hypothesis is consistent with autopsy and imaging studies that show an accelerated loss of parenchymal volume after middle age.30,31,40
In a series of 1957 potential kidney donors undergoing computed tomography angiograms and urograms, the prevalence of renal artery fibromuscular dysplasia and narrowing from atherosclerosis, focal scarring (focal cortical thinning), parenchymal calcification, and suspicious or indeterminate masses all increased with age.41 Simple renal cysts also become increasingly prevalent with age.42,43 Renal tubular diverticula that occur with aging are hypothesized to be the source of these simple cysts.14,44
Using stereological techniques, autopsied kidneys can be carefully sectioned to estimate the total number of glomeruli present. Given the increased proportion of sclerosed glomeruli, it is not surprising that the total number of functional glomeruli decreases with age.11 A less precise technique to determine the number of glomeruli in living kidneys is to multiply the cortical volume (imaging study) with the glomerular density in the cortex (renal biopsy). By this technique, the number of functional glomeruli also appears to decrease with the kidney donor’s age.25 There is an alternative approach for estimating the total number of functional glomeruli. The ultrafiltration coefficient (Kf) of the whole kidney is estimated from direct GFR and renal plasma flow (RPF) measurements. The average estimated Kf of individual glomeruli is calculated from light and electron microscopic measurements from the renal biopsy. The total number of functioning glomeruli is then estimated from the whole kidney Kf divided by the single nephron Kf.45–47 Using this technique, the numbers of functional glomeruli appear to be substantially decreased in older compared with younger kidney donors.46,47 What remains unclear is whether the total number of glomeruli (sclerosed and functioning) changes with age because of a process of glomerular resorption. Nephron endowment (the number of glomeruli a person is born with) is largely impacted by birth weight,48 and whether nephron endowment affects age-related changes in the kidney is unclear.
The alterations in renal function that accompany aging have been the focus of attention for over a half century. In the classic work of Homer Smith,49 a decline in standard urea clearance from 100% at the age of 30 years to 55% at the age of 89 years was described. The seminal contribution of Davies and Shock in 1950 (ref. 50) cemented the notion of an inexorable decline of GFR with aging. They studied 70 men (aged 25–89 years) free of clinical signs or a history of kidney or heart disease—some were normal volunteers, whereas others were hospitalized or were residents of a nursing facility with a variety of ‘chronic diseases’ including tuberculosis, syphilis, or generalized arteriosclerosis. Isolated systolic or systolic/diastolic hypertension was present in 11 of 70 subjects, 9 of whom were over 60 years of age. GFR measured by renal inulin clearance declined linearly beginning at the age of 30 years from an average of 123 ml/min per 1.73 m2 to 65 ml/min per 1.73 m2 at the age of 90 years (a 46% decline). The decline in RPF roughly paralleled the decline in GFR, although there was a slight increase in RPF at the age of 70 years and beyond. These early studies were all cross-sectional in nature.
This deficit was corrected in 1985 by the report of Lindeman, Tobin, and Shock51 on longitudinal studies of the rate of decline in renal function with age in the Baltimore Longitudinal Study of Aging. A total of 254 presumably ‘normal’ subjects (all men without renal disease or hypertension but some with non-proteinuric diabetes) were studied with serial urinary true endogenous creatinine clearance (Ccr) over a 5- to 14-year period. The average decline of Ccr was 0.75 ml/min/year. The calculated slopes of Ccr vs. time followed a normal Gaussian distribution. As many as 1/3 of the subjects showed a stable Ccr over time. A small number showed an actual increase in Ccr with aging. It was never reported whether the presence of diabetes and hyperglycemia (and concomitant ‘hyperfiltration’) influenced these increasing Ccr results. The statistical analysis did not adequately account for imprecision of Ccr measures, the limited number of observations, and multiple hypothesis testing; relatively few of the increasing slopes were likely a true increase in function. Nevertheless, this study has been widely cited to challenge the concept that the decline in renal function with age is inexorable. Rowe et al.52 had earlier studied many of the same subjects for changes in renal function also by means of Ccr. In 548 ‘normal’ subjects, the average Ccr was 140 ml/min per 1.73 m2 at the age of 30 years and declined to 97 ml/min per 1.73 m2 at the age of 80 years (a 31% difference). Longitudinal data on 293 ‘normal’ subjects suggested an acceleration of the decline with advancing age. They concluded that the decline in Ccr with aging is true ‘physiological’ renal aging (senescence), and not secondary to diseases that become more prevalent with advancing age.
Many years later, the controversy regarding the inevitability of the decline in renal function with aging was reawakened by Fliser et al.53,54 These authors suggested that the elderly population was heterogeneous—some have a decline in GFR explained by diseases that complicate aging (such as severe hypertension or congestive heart failure), whereas in the most healthy the decline in GFR is much more modest and not inevitable. Fliser et al. also proposed that the renal functional changes accompanying aging might be the consequence of an altered responsiveness to vasodilators (such as L-arginine, nitric oxide, acetylcholine and dopamine) and vasoconstrictors (angiotensin, nor epinephrine, endothelin).55 This thesis is based largely on observations that the filtration fraction increases with aging, due to a disproportionate fall in RPF relative to GFR.55–57 As noted above, the filtration fraction does not apparently begin to increase until after the age of 60 or 70 years, yet the decline in GFR begins at age 30–40 years, at least in cross-sectional studies. Nevertheless, it is clear that the aging kidney shows impaired endothelium-dependent vasodilatation (an nitric oxide–dependent response), particularly in the presence of hypertension (and albuminuria).58–60 Accumulation of asymmetric dimethyl arginine, an endogenous inhibitor of nitric oxide synthetase, may be partly responsible for the imbalance of vasodilatation and vasoconstriction in the aging kidney.61 In a classic study of renal vascular responses in normal aging men, Hollenberg et al.59 found that the vasodilatory response to acetylcholine or to an acute sodium load was impaired with aging, whereas the vaso-constrictive response to angiotensin was not altered with aging, consistent with a fixed (anatomic) lesion of blood vessels.59 Collectively, these studies strongly indicate that senescent changes in renal function are driven primarily by a vascular process. Lindeman et al.62 had previously shown that aging subjects without hypertension (mean arterial pressure <107 mm Hg) demonstrate much less decline of Ccr with aging, but a cause and effect relationship cannot be proven by such data.
Two additional observations are worth considering. One is the elegant studies of Hollenberg et al.63–65 in the Kuna Indians of Panama. This island-dwelling indigenous population (culturally adapted to consumption of large amounts of flavone throughout life) is remarkably free of the cardiovascular (CV) diseases commonly affecting the elderly in non-island Panamanians, and they do not show any progressive increase in blood pressure with age. Both GFR and RPF decline with age in the Kuna Indians—indeed, the slope of RPF and GFR relative to age was greater in the Kuna Indians compared with age-matched residents of Boston. This is one of the few studies that potentially disassociate the diseases of aging that are often attributed to a modern lifestyle from the renal function changes accompanying normal senescence. The second observation concerns changes in renal function in living donors of kidney allografts.66–68 Kidney donors are rigorously evaluated to ensure ‘health’ before donation, yet these cross-sectional studies consistently show a progressive decline in GFR with age even after excluding donors with CKD or CKD risk factors.
Because of the complexity of direct measurement of GFR (mGFR) by infusion of exogenous substances (e.g., iothalamate), formulas have been devised to estimate GFR (eGFR) from the serum concentration of endogenous markers that are filtered at the glomerulus.69–71 Creatinine is the endogenous marker that has been most widely used. The Cockcroft–Gault (C–G) formula,69 which estimates endogenous Ccr in ml/min not adjusted for body surface area, and the modification of diet in renal disease formula (eGFR-MDRD), which estimates GFR in ml/min adjusted to standard body surface area of 1.73 m2,70 are among the best studied serum creatinine (Scr)–based equations. A newer modification of the eGFR known as the CKD-EPI equation71 uses the same variables as the MDRD equation but was developed using a more diverse sample of patients. All creatinine-based equations contain age and gender as variables (only the C–G formula contains a weight variable). The inclusion of age (and gender) in the Scr-based equations is to provide ‘surrogacy’ for anticipated endogenous creatinine production rate, which inevitably declines with age most likely due to loss of lean body mass.5 Numerous cross-sectional studies have shown that eGFR (MDRD) and eCcr (C–G) decline with age in a manner very similar to that described by studies of mGFR, cited above, although the absolute values vary.66–68,72,73 Figure 3 shows a representative study of eGFR (MDRD) in normal ‘healthy’ kidney transplant donors.66 Note that the mean value for eGFR (MDRD) is about 105 ml/min per 1.73 m2 for 18- to 24-year-old male subjects (about 11% lower than average mGFR) and is about 102 ml/min per 1.73 m2 for 18- to 24-year-old female subjects (about 7% lower than average mGFR). The mean eGFR (MDRD) declines to about 80 ml/min per 1.73 m2 and 78 ml/min per 1.73 m2 in 65-year-old men and women, respectively. This equates to a 25 ml/min per 1.73 m2 or 24% decline in eGFR over 45 years of age for men and women, respectively.
These values for ‘normal’ eGFR (MDRD) in aging populations have important implications for the diagnosis of CKD in the elderly.74,75 Since 2002, with the publication of the NKF-KDOQI classification and staging schema for CKD,76 it has been possible to diagnose CKD solely on the basis of an isolated eGFR (MDRD) value of <60 ml/min per 1.73 m2. Inspection of Figure 3 demonstrates that a significant fraction of apparently normal aged individuals over the age of 60 years (more women than men) have eGFR values that are below 60 ml/min per 1.73 m2 but nearly always above 45 ml/min per 1.73 m2. In the eGFR 45–59 ml/min per 1.73 m2 range for healthy adults (kidney donors), the MDRD equation underestimates mGFR by 25% and the CKD-EPI equation underestimates mGFR by 16%.77 This likely occurs because these equations were developed using CKD patients with decreased muscle mass compared with healthy adults. This bias, combined with the relatively poor precision of the eGFR formulas in the range of >60 ml/min per 1.73 m2,70,78 can lead to misclassifying (overdiagnosing) ‘healthy’ older persons as having a ‘disease.’ This is the inevitable consequence of adopting a fixed threshold of eGFR for identifying CKD, without making some adjustment for the variation of eGFR seen with age and gender.9,74,75 With current criteria, the great majority of subjects diagnosed as having CKD Stage 3 are over 65 years of age.79
On the other hand, there would be justification for such identification of CKD if a modestly reduced creatinine-based eGFR (45–59 ml/min per 1.73 m2) was predictive of adverse outcomes among older adults. Such an adverse prognostic effect of a lower eGFR in older subjects has been the subject of intense investigation and some general principles have now been clarified, as a result of large observational studies and meta-analyses.80–91 These are summarized below:
Whether a creatinine-based eGFR at levels of 45–59 ml/min per 1.73 m2 at ages over 65 years can be considered an important independent risk factor for CV disease is doubtful.92 Adding eGFR (MDRD equation) to risk estimation among 27,620 subjects over the age of 55 years with documented CV disease contributed nothing of value to reassignment of risk compared with traditional risk scoring.92 It seems clear that the use of risk stratification to define a disease has its inherent limitations.
It should be remembered that all creatinine-based equations for estimation of eGFR contain an age variable. This can confound the observed relationship of eGFR with both all-cause mortality and CV morbidity and mortality, as age is itself a strong factor predisposing to both. Further, the use of age instead of actual muscle mass with creatinine-based equations for eGFR inflates both prevalence estimates of CKD and risk estimates of adverse outcomes in the elderly.93
Scr-based eGFR determinations also assume that sarcopenia is inevitable with aging, as age is used as a surrogate for muscle mass in these equations. To put this in perspective, consider a 20-year-old woman with an Scr level of 1.0 mg/dl. If there is no change in her Scr level with aging, she will still have developed CKD (Stage 3) by the time she is 50 years old (eGFR-MDRD =59 ml/min per 1.73 m2). It will be argued that she developed CKD that was not detected by a change in Scr level, because as she aged she lost muscle mass at the same rate as she lost GFR. But why is the age-related decline in GFR considered less inevitable and more of a disease than the age-related sarcopenia? If both age-related decline in GFR and sarcopenia were preventable, then this hypothetical patient could continue to have an Scr level of 1.0 mg/dl well into old age, but would still be labeled with CKD.
Cystatin C–based eGFR measurements appear to be superior for predicting morbidity and mortality risk,94,95 but this may be the result of Cystatin C levels covarying with underlying inflammation, which can itself be a factor in CV disease.96 Nevertheless, there are likely two different populations within a cohort of elderly persons with eGFR (MDRD or CKD-EPI) values of 45–59 ml/min per 1.73 m2—fit persons with well-preserved muscle mass and a senescence-related (benign) decline in GFR and those with actual CKD. In this setting, a confirmatory test such as an elevated cystatin C level might be helpful before the diagnosis of bona fide CKD can made. Indeed, the risk of adverse outcomes is only increased among those with a creatinine-based eGFR <60 ml/min per 1.73 m2 who also have a Cystatin C–based eGFR <60 ml/min per 1.73 m2.95
Normal aging is accompanied by progressive nephrosclerosis (glomerulosclerosis, tubular atrophy, arteriosclerosis, and interstitial fibrosis). Despite volume-losing lesions of glomerulosclerosis and tubular atrophy, overall kidney volume appears to be stable with aging, except in the very elderly. Compensatory hypertrophy of unaffected nephrons seems to preserve kidney volume. Indeed, glomerular size increases and glomerular density decreases with aging. GFR also declines with normal aging, but this is not clearly explained by changes in glomerular size, density, or glomerulosclerosis. Creatinine-based eGFR is imprecise and substantially underestimates mGFR. This contributes to the overdiagnosis of CKD in the elderly. Confirmatory testing with cystatin C and urine albumin excretion might help clarify which elderly truly have CKD and are at increased risk for disease-related adverse events. To the extent that a GFR decline is inevitable with aging, the use of a single GFR threshold for diagnosing CKD is not clearly justified.
This work was supported by the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK 090358 and K23DK078229). We also thank Dr Christopher Winearls for his review of an earlier version of this manuscript.
All the authors declared no competing interests.