Controversies in Chronic Kidney Disease Staging

Abstract

In 2002, a new chronic kidney disease staging system was developed by the US National Kidney Foundation. The classification system represented a new conceptual framework for the diagnosis of chronic kidney disease (moving to a schema based on disease severity defined by the glomerular filtration rate). While the introduction of the staging system stimulated significant clinical and research interest in kidney disease, there has been vigorous debate on its merits. This mini-review aims to summarise the recent controversies that have been raised since the introduction of the new classification.

Introduction

In 2002, the US National Kidney Foundation (NKF) released a new classification of chronic kidney disease (CKD).¹ This classification system represented a new conceptual framework for the diagnosis of CKD, moving away from an aetiology-based system to a schema based on disease severity defined by glomerular filtration rate (GFR). The introduction of the CKD staging system stimulated significant interest in kidney disease, increasing CKD research, placing CKD on the public health agenda, and increasing awareness of CKD at both the population and individual level. The new staging system also helped to provide the impetus to standardise serum creatinine assays across laboratories and countries, and to develop increasingly accurate GFR estimating equations. The majority of nephrologists would acknowledge the important advances seen since the classification system was introduced in 2002.

However, the introduction and use of the staging system has also prompted quite vigorous debate within nephrology, led by Glassock and Winearls^2–4 among others.^5,6 The purpose of this short review is to examine the controversies surrounding the current CKD staging system and to indicate possible future directions and modifications.

The Current CKD Staging System

Before discussing the controversies with the current CKD staging system, it is important to provide detail on the current CKD Classification.¹ The system contains three components (Tables 1 and 2): an anatomical or structural component (kidney damage), a functional component (GFR) and a temporal component (the abnormalities must be present for more than three months). In the majority of cases, the functional component based on GFR is estimated using an eGFR equation, most commonly the modified Modification of Diet in Renal Disease (MDRD) study equation.⁷ According to the classification system, a subject is diagnosed with CKD if they have evidence of kidney damage or an estimated GFR (eGFR) <60 mL/min/1.73m². Kidney damage is defined as the presence of pathological abnormalities or markers of damage including abnormalities in blood and urine tests and in imaging studies (Table 1). Once diagnosed with CKD, the severity is classified in five stages based on the level of kidney function defined by the GFR (Table 2).

Table 1.

Definition of Chronic Kidney Disease. Adapted from K/DOQI Clinical Practice Guidelines on Chronic Kidney Disease.²³

Criteria Kidney damage for 3 months, as defined by structural or functional abnormalities of the kidney, with or without decreased GFR, manifest by either: Pathological abnormalities; or Markers of kidney damage, including abnormalities in the composition of the blood or urine, or abnormalities in imaging tests. GFR <60 mL/min/1.73m2 for ≥3 months, with or without kidney damage.

Table 2.

CKD Staging System. Adapted from K/DOQI Clinical Practice Guidelines on Chronic Kidney Disease²³ and White et al.¹⁷

			Estimated Prevalence using CKD-Epi Study Equation†
Stage	Description	GFR (mL/min/1.73m2)	% (95% CI)	No.
1	Kidney Damage* with Normal GFR	>90	2.29 (1.72, 3.06)	275,453
2	Kidney Damage* with Mild ↓ GFR	60–89	3.43 (2.64, 4.44)	411,383
3a	Moderate ↓ GFR§	45–59	4.75 (3.60, 6.26)	570,905
3b	Moderate ↓ GFR§	30–44	0.78 (0.53, 1.14)	93,923
4	Severe ↓ GFR§	15–29	0.27 (0.14, 0.53)	32,530
5	Kidney Failure§	<15 (or on dialysis)	NA	16,751

^*Kidney damage represents abnormalities in blood and urine tests including haematuria and albuminuria/proteinuria or in imaging studies.

^†Prevalence estimates in the Australian general population aged ≥25 years (about 12 million individuals). Population totals for stage 5 CKD are the number of Australians receiving renal replacement therapy as at 31 December 2007 (ANZDATA Registry Report 2008. Eds McDonald S, Excell L, Livingston B. Australia and New Zealand Dialysis and Transplant Registry, Adelaide, South Australia).

^§Subjects may or may not have evidence of kidney damage.

Controversies in the CKD Staging System

The concerns regarding the current CKD Staging System can be grouped into four broad areas: issues with methodology (in relation to the use of GFR estimating equations and of albuminuria), overuse or misdiagnosis of ‘CKD’, appropriateness of the thresholds or cut-offs for CKD, and finally diagnosing CKD without any consideration of aetiology.

1. Methodology

A first major criticism with the CKD staging system has surrounded the accuracy of the eGFR estimates using the MDRD study equation especially at higher levels of true GFR (>60 mL/min/1.73m²). The MDRD eGFR equation was developed in a population with CKD, and there have been legitimate concerns regarding the imprecision and bias of the equation to estimate true GFR. While estimates using the MDRD equation at GFR below 60 mL/min/1.73m² demonstrate low levels of bias and high levels of precision, it is clear that as true GFR increases above 60 mL/min/1.73m², the equation underestimates the GFR, leading to misclassification of subjects with CKD.⁸ For example, the CKD Epidemiology (CKD-Epi) Collaboration assessed the performance of the MDRD study equation in a large (N=5,504) diverse population, all of whom had both a measured GFR (mGFR) and a standardised serum creatinine measurement.⁹ The median bias (difference) between the mGFR and the eGFR at eGFR levels of 30 to 59 mL/min/1.73m² was only 1.7 mL/min/1.73m² whereas it was 9.5 in the 60 to 89 mL/min/1.73m² group.

The concerns regarding the inaccuracy of the MDRD study equation at higher levels of eGFR prompted the CKD Epidemiology Collaboration to pursue the development of a new eGFR equation which retained the accuracy of the MDRD study equation at lower levels of eGFR but increased the accuracy at levels eGFR levels >60 mL/min/1.73m². The development and validation of the CKD-Epi equation was published in 2009¹⁰ demonstrating improved bias at eGFR levels above 60 mL/min/1.73m² compared to the MDRD equation; the mean difference between mGFR and eGFR being 3.5 versus 10.6 mL/min/1.73m² for CKD-Epi and MDRD respectively. Similar to the MDRD equation, the CKD-Epi equation has coefficients for serum creatinine, age, gender and race, making the potential implementation of the equation into automatic laboratory eGFR reporting very straightforward.

Finally, as with serum creatinine of five to ten years ago, there is an increasing recognition of the need to standardise urinary albumin assays and to provide consensus on how to measure and report proteinuria (either as albumin or total protein) in subjects with CKD.¹¹ It is certainly clear that an assessment of proteinuria by determination of albumin or total protein concentration (either in the laboratory or by dipstick) contributes very important prognostic information. Standardising methods of measurement as well as reporting of proteinuria will assist practitioners in the assessment and monitoring of patients with CKD.

2. Overuse or Misdiagnosis of ‘CKD’

The second major criticism of the CKD Staging System was the definition of chronic kidney disease as an eGFR<60 mL/min/1.73m² without the need for any other measures of kidney damage such as proteinuria or haematuria. Major general population epidemiological studies, such as those derived from the NHANES Survey in the US,¹² described a high prevalence of CKD (of just under 8%). Of particular concern were the high prevalence estimates of between 30 and 40% seen in the elderly. These estimates were largely driven by subjects (especially females) with an eGFR of 50 to 59 mL/min/1.73m² without proteinuria. Many nephrologists consider that a large number of these elderly subjects are in fact normal, with the eGFR values representing the ‘normal’ age related decline in renal function, and they have questioned the automatic labelling of this group as having CKD.³

In concert with these concerns, the implementation of eGFR automatic reporting led to a marked increase in ‘unnecessary’ referrals to nephrology clinics.¹³ In two separate analyses of the Alberta Kidney Disease Network Database,^14,15 significant increases in referrals to nephrology clinics were evident after the implementation of eGFR automatic reporting. This increase in referral was seen largely in the group with eGFR<60 mL/min/1.73m², and in those subjects aged >60 years. Similarly, data from Australia¹⁶ demonstrated a rise in the number of referrals to public hospital CKD clinics coupled with a fall in the proportion of those referrals which were deemed to be appropriate by the treating nephrologists. There was however a greater absolute number of CKD patients that were referred appropriately.

The development of the CKD-Epi equation and its implementation will likely help alleviate a number of these concerns. Data from the AusDiab Study¹⁷ demonstrates a significant reduction in the population prevalence of CKD from 13.4% to 11.5% with the application of the CKD-Epi equation, equivalent to ‘curing’ 200,000 subjects of their CKD. This reduction in prevalence was predominantly driven by the reclassification of subjects with stage 3a (eGFR 45 to 59 mL/min/1.73m²) to no CKD. In addition, a number of subjects with stage 2 CKD were reclassified to stage 1 CKD. Importantly, the reclassified subjects overall cardiovascular risk profile was similar to those subjects classified with no CKD by both equations, demonstrating that the new equation is selecting out those subjects at low cardiovascular and renal risk. Similar data has also been reported in a cohort study from the USA.¹⁸

3. Appropriateness of Thresholds or Cut-Offs for CKD Stages

The current CKD Staging System divides subjects with kidney disease into pre-defined stages according to eGFR. The separation into stages based on severity, while inherently arbitrary, was meant to aid identification of individuals at increased cardiovascular and renal risk, and to help facilitate the management of patients with CKD. Glassock and Winearls^2,3 have argued that the separation of stages 1 and 2 CKD is problematic, especially when one considers the difficulties with accuracy of the MDRD formula as outlined above, and that separation cannot be justified on the basis of mortality risk. However, recent data for the CKD Prognosis Consortium has refuted this argument.¹⁹ It is now clear that both, all cause and cardiovascular mortality, are independently increased for subjects with stages 1 and 2 CKD, and even for subjects from the general population with ‘high normal’ levels of albuminuria (a urinary albumin/creatinine ratio of 1.1 versus 0.6 mg/mmol) and eGFR >60mL/min/1.73m².¹⁹

There have also been concerns regarding whether all of stage 3 CKD should be termed a disease. In their seminal paper,²⁰ Go and colleagues divided stage 3 CKD into the two parts: stage 3a (45–59 mL/min/1.73m²) and stage 3b (30–44 mL/min/1.73m²). They demonstrated an increased risk of all cause and cardiovascular mortality with increasing severity of CKD stage but with a quite marked increase in the risk seen at stage 3b or greater. The issue of an independent adverse risk associated with stage 3a CKD remained subject to debate. Again, the CKD Prognosis Consortium data has helped inform the debate by demonstrating significant increased risk of all cause and cardiovascular mortality in subjects with stage 3a CKD.¹⁹

The presence of microalbuminuria is currently sufficient to diagnose CKD stages 1 and 2. Whether microalbuminuria truly represents kidney damage or just reflects inherent vascular disease has also been debated.⁶ However, it is clear that albuminuria, whether micro or macro, is not just a potent cardiovascular risk factor¹⁹ but also independently predicts the development of end-stage renal disease (ESRD).²¹ In fact, the combination of eGFR and albuminuria alone, predicted the development of ESRD as efficiently as a more complex model that also included other clinical risk factors such as hypertension and diabetes mellitus.²¹

4. Aetiology of CKD

According to the current staging system, a subject has CKD if they have evidence of kidney damage, with or without a reduced eGFR, or an eGFR of <60 mL/min/1.73m² with or without kidney damage. There is no consideration given to aetiology in the current system. Many nephrologists have been uneasy about the lack of consideration of aetiology in the ‘diagnosis’ of CKD. However, classification systems based solely on aetiology are not without precedent, with the NYHA classification of heart failure being an example. In response to this criticism, it has been suggested that the impending NKF ‘Kidney Disease: Improving Global Outcomes’ (KDIGO) revision of the CKD classification will incorporate a statement regarding the underlying aetiology of CKD in the revised classification system.²²

Potential Changes to CKD Staging

KDIGO has recently formed an international work group to update and revise the current CKD staging system.²² The deadline for publication of the updated guidelines is sometime in 2012. Potential changes to the current system include the addition of a measure of proteinuria in all stages of CKD, including a consideration of the cause of CKD, splitting stage 3 into 3a and 3b, and recommending the use of the CKD-Epi eGFR equation over the MDRD study equation.²²

Conclusions

The introduction of the CKD staging system in 2002 has proved an important advance in the conceptual framework for managing and assessing risk in patients with CKD. There have been many challenges to the current system and there is a certainly a need for revision to incorporate new research findings and recent advances in the diagnosis and management of CKD. The exact nature of the revised guideline, however, remains to be seen.