Pediatric Nephrology

, Volume 23, Issue 11, pp 1933–1940

Diabetes and chronic kidney disease: lessons from the Pima Indians


    • Division of Nephrology, MS#40Childrens Hospital Los Angeles
    • Division of Nephrology, Department of PediatricsUniversity of Southern California Keck School of Medicine

DOI: 10.1007/s00467-008-0763-8

Cite this article as:
Lemley, K.V. Pediatr Nephrol (2008) 23: 1933. doi:10.1007/s00467-008-0763-8


Although diabetic nephropathy is a very rare cause of kidney failure during childhood, the underlying events leading to progressive kidney injury begin during childhood in many patients with type 1 diabetes mellitus (T1DM) and in increasing numbers of children with type 2 diabetes mellitus (T2DM). The Pima Indians of Arizona represent an exceptionally thoroughly studied population suffering from very high rates of T2DM and diabetic nephropathy (T2DN). This population well illustrates the often inexorable progression from glomerular hyperfiltration to microalbuminuria to overt proteinuria and loss of glomerular filtration rate (GFR), paralleled by the accumulation of mesangial matrix and basement membrane, glomerular hypertrophy, loss of podocytes and eventual glomerular sclerosis and interstitial fibrosis. Structural changes quantitatively account for the loss of GFR in T2DN. The mechanism of albuminuria (and its relationship to GFR loss) is much less clear. There is strong functional and structural evidence for defects in glomerular size-selectivity (shunts) due to podocyte pathology, but only beginning at relatively high levels of proteinuria (albumin/creatinine ratios > 3000 mg/g). Podocyte loss accompanies, and may underlie, the loss of glomeruli to sclerosis. At this point, most evidence in humans suggests detachment of intact podocytes from the glomerular basement membrane, rather than apoptosis, as the predominant mechanism of podocyte loss.




Diabetes is the single largest cause of end-stage kidney failure worldwide in adults. In contrast, due to the smaller incidence of childhood-onset diabetes and the years required for the development of significant renal injury, diabetic nephropathy is a rare cause of kidney failure in children. In the 2007 NAPRTCS report, diabetes accounted for 11 out of 9506 transplants []. So why should a review of diabetic nephropathy as a cause of chronic kidney disease (CKD) be of interest to pediatric nephrologists? One reason is that type 1 diabetes mellitus (T1DM) typically begins in childhood. Although overtly impaired renal function is rare during childhood, the pathophysiologic events that will eventuate in kidney damage are present early on in many patients (as manifested by albuminuria). In addition, the incidence of type 2 diabetes mellitus (T2DM) is growing rapidly in children, paralleling an alarming rise in childhood obesity. Youth-onset T2DM may be associated with higher risks of adult morbidity and mortality than adult-onset disease [1], likely due to the longer duration of exposure to the diabetic milieu. If early interventions are expected to be effective in slowing or arresting the development of diabetic nephropathy in adult patients, they may need to be initiated in childhood.

Why is the study of diabetic nephropathy in Pima Indians of particular interest? First of all, this population has a greatly disproportionate risk of developing T2DM and diabetic nephropathy (T2DN). Historically, about 50–60% of the tribal population of the Gila River Indian Community has developed diabetes by late adulthood; of these, a large proportion eventually develop T2DN. Although the rate of end-stage renal disease (ESRD) development has historically been quite high in this population, there is evidence that it has declined since about 1990, possibly due to the slowing of disease progression by increased use of angiotensin-converting enzyme (ACE) inhibitors [2]. In addition, diabetic nephropathy in this population represents a relatively “pure” model of the disease. Compared to European-origin populations, Pima Indians with T2DM are remarkable for their lack of significant hypertension or hyperlipidemia early in the course of chronic kidney disease (CKD). The Pima Indians also manifest a relatively accelerated course of T2DN when compared to European-origin populations: because of the early onset of T2DM, the former typically experience kidney failure in their 30s and 40s versus ages 50–60 years in the latter [3]. Equally important is the exceptional willingness of the Pima Indians of the Gila River reservation near Phoenix, Arizona, to participate in an intensive longitudinal study of diabetes by the NIH1, which has been pursued since Dr. Peter Bennett started his seminal studies of this population in the 1960s. Routine biennial screening for impaired glucose tolerance is offered to the entire population from childhood (>5 years of age). Thus, the entire natural history of T2DN has been open to investigation for several decades.

The clinical course of diabetic nephropathy: Type 1 and Type 2

Since this review will concentrate on the development of nephropathy (CKD) only in one group of individuals with T2DM, it is reasonable to ask how applicable findings in this group are to diabetic nephropathy in general. That is, does T2DM differ from T1DM with respect to various aspects of nephropathy? In European populations, the risk of developing overt proteinuria and of impaired kidney function with T1DM and T2DM have been reported to be similar [4]. In the past, about 20–40% of T1DM patients eventually developed overt proteinuria, with 5–15% progressing to renal failure. The incidence of kidney failure seems to have declined (at least in some populations) since the initiation of more intensive therapies after the Diabetes Control and Complications Trial (DCCT), although it is possible that this is simply because the onset of nephropathy has been delayed [5]. The risk of nephropathy with T2DM, in groups other than the Pima Indians, is more difficult to estimate than in the case of T1DM since our current knowledge of the onset and incidence of asymptomatic T2DM itself is much less certain. Many affected individuals first present with overt diabetic complications later in life, after an unknown duration of T2DM. About 20% of those who develop overt DN progress to kidney failure. This rate appears to be lower than the rate in Pima Indians, possibly because the older European populations are censored by cardiovascular mortality before they develop kidney failure.

In most cases, the pathology of T2DN has been thought to be essentially indistinguishable from that of T1DN. This was confounded somewhat by early descriptions of a substantial minority (23%) of patients with T2DM and proteinuria showing non-diabetic renal lesions on biopsy [6]. More recently, atypical diabetic pathological changes have been described in a substantial minority of patients with T2DM and proteinuria [7]. Despite this, the detailed structure-function relationships in T1DN and T2DN appear to be quite similar [8].

The clinical course of diabetic nephropathy in Pima Indians

Physiologic aspects of diabetic nephropathy

Glomerular hyperfiltration [a glomerular filtration rate (GFR) greater than that in non-diabetic Pima Indian controls] occurs with the development of impaired glucose tolerance in Pima Indians [9], as it does in other populations. Whether increased renal size accompanies hyperfiltration in early diabetes, as in other populations, has not been reported. The development of albuminuria occurs some years later in the evolution of diabetic renal injury. The level of albuminuria is generally subdivided into microalbuminuria [spot urine albumin/creatinine ratio (ACR) between 30 and 299 mg/g or a timed urinary albumin excretion (AER) of 20–199 mcg/min] and macroalbuminuria or overt proteinuria (ACR ≥ 300 mg/g; AER ≥ 200 mcg/min). As with other groups (for example, adolescents with T1DM [10]), many individuals with established microalbuminuria will spontaneously return to normoalbuminuria independent of treatment [11].

There is a fairly rapid acceleration of the level of proteinuria, hypertension and loss of GFR (leading to end-stage kidney failure) once macroalbuminuria is established. Half of patients will reach end-stage within 10 years of developing stable overt proteinuria. The rate of GFR loss seems to be significantly greater in Pima Indians than in European populations (vide infra) [12].

Structural aspects of diabetic nephropathy

There is a progressive accumulation of extracellular matrix material, both in the mesangium and the glomerular basement membrane (GBM). It is not clear whether this is an independent, time-dependent process in diabetes (a diabetic “clock”), or if it is causally related to the development of nephropathy. Pagtalunan et al. [13] reported approximately equal mesangial volume fractions and GBM widths in diabetic Pima Indians with microalbuminuria and in subjects with long-term [14 ± 1 (SE) years] diabetes and normal urinary albumin excretion (UAE), suggesting that such matrix accumulation is independent of the development of nephropathy. Glomerular volume has been reported not to be greater in Pima Indians with diabetes than in those without diabetes, although this report was based on autopsy specimens from subjects having a number of co-morbidities [14]. In contrast, a cross-sectional biopsy study of Pima Indians demonstrated an increase in glomerular volume in subjects with “early diabetes” (≤6 years and normal UAE) to long-term normoalbuminuria and further to overt nephropathy [13]. Increased glomerular volume with time has also been described in persistently normoalbuminuric young subjects with T1DM [15].

More significant with respect to the eventual loss of renal function than the above-mentioned structural changes is the development of glomerular sclerosis and interstitial fibrosis. Glomerular sclerosis in diabetes may have a different character than the segmental lobar obliteration seen in other renal diseases. It is often characterized by marked acellular mesangial expansion, most classically represented by Kimmelstiel–Wilson nodules. Significant sclerosis may exist despite a GFR still within the “normal” non-diabetic range [13], likely due to compensatory hyperfunction of patent glomeruli. A more recently appreciated structural feature that may be relevant to loss of renal function is a progressive loss of glomerular podocytes. This has been demonstrated in both cross-sectional and longitudinal studies [13, 16] and becomes significant about at the point of transition from micro- to macroalbuminuria. Baseline podocyte number has also been shown to predict the future development of albuminuria [17]. Widespread podocyte (foot process) detachment and endothelial cell pathology [presumably due to perturbations in “signaling” to endothelial cells from podocyte-derived vascular endothelial growth factor A (VEGF-A] has recently been reported in T1DN [18] and is also observed in Pima Indians (personal communication J. Weil (2008)), (Fig. 1).
Fig. 1

a Electron micrograph of peripheral capillary walls of a 32-year-old microalbuminuric Pima Indian with type 2 diabetes mellitus (T2DM). Notice several areas denuded of podocyte foot processes (arrows). Courtesy of Dr. J. Weil. b Electron micrograph of peripheral capillary wall from the same 32-year-old microalbuminuric Pima Indian as in a. Note the moderately thickened basement membrane and subendothelial deposits and interposition (arrows). Courtesy of Dr. J. Weil

Mechanisms of GFR loss in Pima Indians with diabetic CKD

What is the mechanism by which renal function is lost during the development of advanced diabetic nephropathy in Pima Indians with T2DM? Given that the driving (Starling) forces for glomerular filtration probably do not decrease with the development of overt nephropathy [19], we need to look to structural changes to account for the loss of GFR. Two basic types of structural changes may be distinguished in this regard: loss of functional glomeruli (either due to glomerular sclerosis or to a disconnection of glomeruli from the tubules—leading to so-called atubular glomeruli) and loss of the intrinsic filtration capacity of the individual glomeruli. The former has been shown to be relevant in diabetic Pima Indians [13], with the incidence of global glomerular sclerosis increasing from 5 to 19% between the stages of microalbuminuria and overt nephropathy concomitant with a decline in GFR from 156 ± 10 (SE) to 103 ± 12 mL/min per 1.73 m2. Atubular glomeruli have not been found in a small number of diabetic Pima Indians investigated [13], although this form of injury may be relatively common in T1DM subjects with advancing renal injury [20]. Loss of the intrinsic filtration capacity of individual glomeruli may be due to either loss of filtration surface area or a decrease in the hydraulic permeability of the glomerular capillary wall. Divergent findings have been reported in different patient populations on the loss of filtration surface area associated with mesangial expansion [13, 21]. Although the filtration surface density has been reported to decline in conjunction with mesangial expansion in T1DN [21], in diabetic Pima Indians increases in total glomerular volume with advancing nephropathy tend to mitigate the effects of increasing mesangial volume on the absolute filtration surface area [13], so that the latter remains roughly equal to that in non-nephropathic Pima Indians with early diabetes (although decreased by roughly 14% compared to microalbuminuric individuals). The effects of thickening of the GBM and foot process broadening on local hydraulic permeability are slightly more pronounced: the calculated hydraulic permeability of the glomerular capillary wall is 18% less in subjects with overt nephropathy than in microalbuminuric individuals. Together with the increased incidence of glomerular sclerosis, these structural changes quantitatively account for the above-mentioned decrease in GFR that occurs between the stages of microalbuminuria and overt nephropathy.

As noted above, the rate of loss of GFR in macroalbuminuric T2DN seems to be greater in Pima Indians than in other well-studied populations. Nosadini et al. [22] reported a decrease of 3 mL/min per 1.73 m2 per year in measured GFR over 4 years in 34 European adults with macroalbuminuric T2DN, while Christensen and colleagues [23] reported a loss of 5.3 mL/min per 1.73 m2 per year over a median of 4 years of follow-up in 34 diabetic subjects. In contrast, Lemley [12] calculated a rate of GFR loss of 13.8 mL/min per 1.73 m2 per year [using an estimated body surface area (BSA) of 2.0 m2] in 34 Pima Indians with macroalbuminuric T2DN. This greater rate of loss of GFR among the Pima Indians may be largely an artifact of their large glomerular sizes. If the rate of loss of glomeruli to sclerosis were the same in all groups, assuming a rough equivalency of the driving forces for filtration, the greater glomerular size (and filtration surface area) of the Pima Indians should lead to about a two-fold increase in the amount of GFR lost per glomerulus lost. In support of this interpretation [12], the rate of loss of GFR in nine of these macroalbuminuric Pima Indians who had baseline renal biopsies was proportional to the glomerular volume (P = 0.04). Since GFR loss was also inversely proportional either to the baseline podocyte number per glomerulus (P = 0.04) or the podocyte density in the tuft (P = 0.04), it is possible that a low podocyte density (due to glomerulomegaly) may in addition accelerate the rate of sclerosis of glomeruli [24].

Other aspects of diabetic renal injury

More than two decades ago, Mogensen and Viberti and their colleagues reported that the risk of eventual development of nephropathy was substantially increased in T1DM subjects with microalbuminuria [25, 26]. In Pima Indians with T2DM, the rate of loss of GFR over 4 years has been shown to be highly significantly (log-linearly) correlated with the baseline urinary albumin/creatinine ratio [19]. What does the development of proteinuria have to do with a loss of GFR? The simplest answer is that proteinuria is a marker for glomerular injury, and sustained glomerular injury eventually leads to irreversible structural changes, including glomerular sclerosis. An alternate (although not mutually exclusive) view is that filtered protein stimulates an injury response by the proximal tubule that leads to interstitial inflammation and eventual tubular atrophy and interstitial fibrosis [27]. There is a continuing debate about whether the development of albuminuria in diabetes represents the effects of defects in glomerular size/charge selectivity or is due to changes in the handling of large amounts of albumin (presumed to be filtered by the glomerulus even under normal circumstances) by the tubule [28]. There does appear to be a fairly abrupt change of regime with respect to glomerular size selectivity in Pima Indians with T2DM at a level of proteinuria well above the traditional cut-off point for abnormal albuminuria [11]. The glomerular shunt parameter ω0 (a model-based parameter describing the magnitude of the non-selective macromolecule defect or “shunt” through the glomerular capillary wall [29]) first rises above the normal range for ACR values above 3000 mg/g (Fig. 2). This strongly suggests that lower (but still abnormal) levels of albuminuria are due to factors other than defects in size selectivity, although the data do not distinguish between a loss of glomerular charge restriction and defective albumin reabsorption by the proximal tubule. The magnitude of the shunt does correlate with podocyte foot process width [11], suggesting that podocyte injury is responsible for this higher level of proteinuria (Fig. 3). Significant correlations between quantitative proteinuria and both podocyte foot process width and podocyte number/density have been reported in other populations with T2DM as well [30] and between albuminuria and foot process width in adolescents with T1DM [31], including subjects with normoalbuminuria [32].
Fig. 2

Graph of the shunt parameter (ω0) against the urinary albumin/creatinine ratio for 42 microalbuminuric (open circle) and 31 macroalbuminuric (closed triangle) diabetic Pima Indians. The horizontal dashed lines represent the 25th and 75th percentiles for normoalbuminuric control subjects. Notice that ω0 values outside the normal range first occur for ACR > 3000 mg/g. Reprinted from [11] with permission of the American Society of Nephrology
Fig. 3

The shunt parameter ω0 (a marker of defective glomerular size selectivity) increases with increasing podocyte foot process widths in ten Pima Indians with macroalbuminuria (r = 0.69, P = 0.03). Reprinted from [11] with permission of the American Society of Nephrology

At this time, both with respect to mechanisms of proteinuria and of nephropathy progression, the nephrology world seems to be divided into glomerulocentric (podocentric) and tubulocentric camps, with little prospect for doctrinal reconciliation in the immediate future.

As mentioned above, a loss of podocytes with progression of CKD has been demonstrated in Pima Indians with T2DM [13, 16]. In a longitudinal study, the number of podocytes declined significantly [from 547 ± 150 (SD) to 356 ± 69 per glomerulus] in 12 initially microalbuminuric individuals who underwent serial biopsies about 4 years apart [16]. Half of these individuals had progressed to overt proteinuria by the time of the second biopsy. Since podocytes appear to be necessary in order to “maintain” intact glomerular structure [33], podocytopenia may contribute directly to the development of glomerular sclerosis, as it seems to in other glomerular diseases [34]. In addition, glomerulomegaly found in Pima Indians may act synergistically with podocyte loss to cause an even more dramatic decrease in podocyte density, possibly accelerating the development of glomerular sclerosis [24].

How are podocytes “lost” from the glomerular tuft in diabetic nephropathy? Three general mechanisms may be considered: necrosis, apoptosis and the detachment of intact podocytes from the GBM. There is essentially no evidence of necrosis and no direct evidence supporting apoptosis as a major mechanism of podocyte loss in human diabetes. Apoptosis of podocytes has been repeatedly demonstrated in cell culture; however, it must be remembered that cultured podocytes also undergo active cell proliferation, which is not characteristic of podocytes in situ, raising the question of whether apoptosis is also an artifact of these particular experimental conditions. One careful study using TUNEL staining and electron microscopy reported no glomerular apoptosis (but pronounced tubulointerstitial apoptosis) in five patients found to have diabetic nephropathy on renal biopsy [35]. The only other published study on this topic showed increased glomerular apoptosis, without however specifically quantifying the degree of podocyte apoptosis [36]. Thus, the relevance of podocyte apoptosis to the development of CKD in diabetic nephropathy remains largely unresolved. Our own experience (unpublished observations) does not suggest that apoptosis is a common finding in podocytes from biopsies of Pima Indians with T2DN.

The detachment of intact podocytes from the GBM has been reported in non-Pima Indian diabetic humans [37]. The percentage of urinary podocytes showing markers of apoptosis has not been reported in published studies on diabetes. The presence of apoptotic podocytes in the urine does not establish that they detached from the GBM due to apoptosis. Given the loss of anchoring to the GBM matrix and the “hostile” chemical conditions in the tubule and bladder (acidic, hypoxemic), podocytes may enter an apoptotic pathway after being shed from the tuft. The rate of podocyte shedding has been shown to be affected by various treatments [38].

Genetic and environmental influences on disease progression

A study using segregation analysis has suggested that a pattern of familial aggregation of nephropathy in diabetic Pima Indians may derive from the effects of a single major gene [39]. A single gene responsible for familial aggregation however has not been described. Numerous studies have shown that diabetic subjects with the angiotensin converting enzyme (ACE) deletion (D) polymorphism (compared to the insertion polymorphism) have an increased risk of developing albuminuria and nephropathy. The frequency of the ACE D allele in Pima Indians is significantly lower than in Caucasians (0.29 vs 0.52) and the attributable variance in ACE serum activity is consequently less (6.5% vs 18%) [40]. Thus, it is not surprising that no published studies have shown an effect of ACE genotype on nephropathy risk in Pima Indians. A recent study of polymorphisms in the meprin β metalloprotease gene suggested a possible association with nephropathy risk [41].

There is presumably a genetic basis to the glomerulomegaly, and related glomerulopenia, seen in the Pima Indians and some other populations at increased risk for nephropathy progression. It is not clear whether or not the glomerulomegaly seen in this population is related solely to obesity. At this point, no specific genetic locus has been identified to account for glomerulomegaly in humans, although podocyte-specific genetic alteration in the von Hippel-Lindau/Hypoxia-inducible factor 1 axis has recently been shown to lead to glomerulomegaly in mice [42].

Generational effects such as intrauterine growth retardation or macrosomia (both as complications of maternal diabetes during pregnancy) may lead to a nephron deficit or “programmed” abnormalities of glucose metabolism on the basis of gene imprinting [43] increasing the penetrance of diabetic nephropathy and perhaps accelerating its course.

Clinical implications and future directions

Probably the main “lesson” that can be drawn from studies on Pima Indians with T2DM and T2DN is that significant structural lesions occur before overt GFR depression is found and can account for the irreversible and progressive loss of GFR that characterizes advancing diabetic nephropathy. The presence of a normal GFR despite widespread sclerosis [13] suggests that therapeutic interventions must be undertaken early in the course of disease, before podocyte loss or overt proteinuria occurs (and perhaps before even microalbuminuria is established). Similarly, structural changes (foot process broadening, glomerular “occlusion”, interstitial expansion, glomerulomegaly) have been described even in young, persistently normalbuminuric subjects with normal (or elevated) GFR [15, 44], suggesting that effective therapies may need to be instituted in childhood in many cases. The IRMA-2, RENAAL and IDNT studies (all 2001) were all aimed at investigating whether or not angiotensin receptor blockade (ARB) prevents disease progression in adults with T2DM and various levels of proteinuria. These studies showed that worsening of proteinuria and even the risk of doubling of serum creatinine (RENAAL and IDNT studies) can be reduced with ARB. A double-blind, placebo-controlled multi-year study on whether or not early use of angiotensin blockade in normoalbuminuric and microalbuminuric Pima Indians with T2DM can prevent or retard changes in function (albuminuria, GFR), structure (podocyte loss) or renal gene expression is currently being analyzed. This study (which includes over 150 biopsies) also presents an opportunity to examine more closely the incidence of atubular glomeruli in diabetic Pima Indians.

To more effectively target and apply these interventions, it is vital to elucidate the mechanism of renoprotection provided by angiotensin blockade. In the past, such agents have been presumed to work via their effects on glomerular hemodynamics, but now it seems likely that they may be directly protective of podocytes by antagonizing the effects of ambient angiotensin II on the podocyte [45]. The (somewhat disputed) protective effects of beta blockers on the progression of diabetic nephropathy may similarly be due to a direct protective action on podocytes [46]. Animal studies using conditional, podocyte-specific knockouts of angiotensin or catecholamine receptors may help to clarify whether this protection is direct or indirect.

A number of novel renoprotective therapies have been suggested, including statins, antagonists to connective tissue growth factor (CTGF) and therapies directed against advanced glycosylation end-products (AGE) and their receptors (RAGE), such as pyridoxamine. Much more investigation of the effects of these therapies in diabetic humans is required.

Finally, albuminuria is not a perfect biomarker of progression risk. Despite its clear predictive associations with nephropathy risk [25, 26], microalbuminuria can resolve spontaneously in up to one-third of patients [10, 11]. Given our mechanistic understanding of disease progression, (cumulative) podocyturia may well be a better biomarker of progression risk than albuminuria [47]. Unfortunately, determination of quantitative urinary podocyte excretion is not readily available as a practical laboratory technique. It is possible that less labor-intensive techniques [such as Enzyme-Linked ImmunoSorbent Assay (ELISA) for urinary nephrin excretion or urinary mRNA of various podocyte-specific proteins] will prove as predictive and more practical.


The NIH study has included Pima Indians and Tohono O’odham (Papago) Indians from the Gila River Indian Community.



Many of the studies reported here represent the collaborative work of groups at Stanford University and the Phoenix Epidemiology and Clinical Research Branch of the NIDDK. Drs. Bryan D. Myers, Timothy W. Meyer, Peter H. Bennett and Robert G. Nelson and Ms. Kristina L. Blouch, Linda Anderson and Lois I. Jones have been principal colleagues in these investigations. Support for some of this work was provided by NIH DK 54600, the Juvenile Diabetes Foundation (#195080) and a Faculty Scholar Award from Satellite Dialysis Corporation, Inc.

Copyright information

© IPNA 2008