The public health dimension of chronic kidney disease: what we have learnt over the past decade Free

Abstract

INTRODUCTION

This year marks the 10th anniversary of the full inauguration of World Kidney Day. In 2007, the International Society of Nephrology and the International Federation of Kidney Foundations decried that for too long, chronic kidney disease (CKD) had been considered ‘uncommon, without consequences, and untreatable until the stage of kidney failure’ [1], calling for a wider recognition of CKD as a global public health problem. At the time, there was considerable controversy about the definition and classification of CKD [2]. There was disagreement over how ‘common’ CKD was and what treatment strategies were evidence based. In this review, we examine the major advances made in the last decade in clarifying these issues under the paradigm of CKD as a ‘common, harmful and treatable’ disease.

THE DEFINITION OF CKD WAS REFINED

Novel equations for estimating glomerular filtration rate were developed and validated

Estimation of glomerular filtration rate (GFR) is crucial for determining renal function in CKD. The Modification of Diet in Renal Disease (MDRD) equation had been the prevailing equation for estimating GFR based on creatinine clearance since 1999 [3]. It was simplified from six to four variables in 2000 and re-expressed to use standardized creatinine in 2006. However, the MDRD equation received criticism for its systematic underestimation at higher GFR.

Over time, progressively larger studies have been conducted to improve GFR estimates, including the Lund-Malmö equations [4]. In 2009, the CKD Epidemiology Collaboration (CKD-EPI) published a novel equation for estimated GFR (eGFR) based on creatinine in 12 150 individuals, CKD-EPI eGFR_Cr [5]. CKD-EPI eGFR_Cr made use of the same four variables as MDRD—serum creatinine, age, sex and race—but introduced a spline in log serum creatinine, with a knot at 0.7 mg/dL for women and 0.9 mg/dL for men. The coefficient for serum creatinine was a larger negative value when above the knot, and a smaller negative value when below the knot. The introduction of this spline enabled CKD-EPI to achieve a lower bias at high GFR compared with MDRD [6]. It modestly increased accuracy, improved generalizability and strengthened risk prediction [6].

Apart from being inversely related to the GFR, the ideal filtration marker should be minimally influenced by non-GFR determinants. Cystatin C has been suggested as a filtration marker since its non-GFR influences were initially claimed to be negligible. Cystatin C appears to be a stronger risk factor of outcomes than serum creatinine [7]. In 2012, the CKD-EPI published an estimating equation based on cystatin C alone and in combination with creatinine (CKD-EPI eGFR_Cys and CKD-EPI eGFR_Cr-Cys) based on 5352 individuals. This equation had the benefits of using standardized creatinine and cystatin C, and being validated across the full range of GFR [8]. Precision was increased and bias was decreased when both creatinine and cystatin were used to estimate GFR, rather than just creatinine or just cystatin. The CKD-EPI equations have been adopted by Kidney Disease: Improving Global Outcomes (KDIGO) as the standard for reporting eGFR in Caucasian and African-American populations [9].

Stratification of stage 3 and incorporation of albuminuria into the definition of CKD improved its clinical utility

The definition and classification of CKD issued by the Kidney Disease Outcomes Quality Initiative (KDOQI) in 2002 [2] and endorsed by the KDIGO in 2004 unified the language of CKD and paved the way for an explosion in CKD research. However, in the subsequent years, the 2002 KDOQI definition and classification came under increasing criticism that the stages did not fully predict prognosis, the criteria neglected to include etiology or the presence of albuminuria, and the threshold of 60 mL/min/1.73 m² over-diagnosed early stages of disease.

To address these concerns, KDIGO conducted an unprecedented meta-analysis of 45 cohorts totaling 1.5 million people worldwide and convened a Controversies Conference to make evidence-based recommendations [10]. This meta-analysis arrived at several important findings. First, the increased relative risk for mortality and kidney outcomes was statistically significant at an eGFR of 60 mL/min/1.73 m², supporting this cutoff for defining CKD. Second, this pattern was true at younger and older age groups, suggesting that it was not necessary to create different definitions by age. Third, in the range of eGFR between 30 and 59 mL/min/1.73 m², there was a steep rise in risk with lower eGFR, favoring suggestions to subdivide stage 3 into two substages, 3A and 3B, split at 45 mL/min/1.73 m². Fourth, the independent predictive ability of albuminuria at all levels of eGFR supported the suggestion to add albuminuria as a component of the staging system.

These recommendations were incorporated in the revised KDIGO guidelines in 2012 [9]. As summarized in Figure 1, the new classification and staging of CKD takes note of cause (C), GFR (G) and albuminuria (A). The guidelines opted for three categories of albuminuria (with cutoffs at albuminuria of 30 mg/g and 300 mg/g) for simplicity, but added that it is worth noting when A3 individuals have nephrotic range proteinuria (albuminuria >2200 mg/g or proteinuria > 3000 mg/g), as nephrotic range proteinuria confers additional risks and is usually indicative of specific renal pathology such as glomerulonephritis.

FIGURE 1

CKD is classified based on cause (C; not depicted in matrix), GFR category (G) and albuminuria category (A). Reproduced with permission from [9].

Open in new tabDownload slide

Other national societies have since expressed agreement with the KDIGO 2012 guidelines, with some differences. The British guidelines do not include cause of disease in CKD staging [11]. The Canadian guidelines suggest adding an albuminuria category for nephrotic range proteinuria [12].

CKD IS COMMON

CKD is prevalent in the USA and worldwide

In the USA, prevalence is estimated from the National Health and Nutrition Examination Survey (NHANES). Between NHANES 1988–94 and NHANES 1999–2002, the prevalence of reduced (<60 mL/min/1.73 m², or stages 3–5) eGFR based on standardized creatinine rose from 4.7% to 6.5% [13]. Since 2003, the prevalence has stabilized at 6.9 in 2003–04, 6.7 in 2005–06, 6.5 in 2007–08, 6.4 in 2009–10 and 6.9 in 2011–12 [14]. Unfortunately, sensitivity analyses show that even a 0.04 mg/dL higher mean serum creatinine can contribute to a 23% higher CKD prevalence [15]. Therefore, without careful laboratory standardization, inferences about CKD prevalence differences across time and location should be made with caution.

Globally, a new meta-analysis of 100 prevalence studies estimates that worldwide prevalence at CKD stages 1 through 5 is 3.5%, 3.9%, 7.6%, 0.4% and 0.1%, respectively [16]. The majority of global CKD lies in stage 3. This follows on the heels of a meta-analysis of 33 prevalence studies which estimated that age-standardized worldwide prevalence of CKD stages 3–5 is 4.7% in men and 5.8% in women [17]. However, both studies acknowledged limitations from the paucity of standardized creatinine assays worldwide and variable adoption of the KDIGO 2012 classification, including lack of distinction between CKD stages 3A and 3B.

Global CKD prevalence varies with geography, age and sex. The sex balance is comparable at younger age groups, but women overtake men in CKD prevalence in middle age, and even more so in older age, especially with CKD stages 3–5 [17]. As expected, CKD prevalence rises with age [17].

With the exception of Iran, CKD prevalence is highest in the high-income regions such as Europe, USA, Canada and Australia, compared with low- and middle-income regions in Sub-Saharan Africa and India (Table 1) [16]. However, low- and middle-income countries tend to have younger population structures than high-income countries. Under 70 years of age, age-specific CKD prevalence is higher in low- and middle-income countries than in high-income countries, whereas above 70 years of age, the reverse is true [17]. This is consistent with the reality that patient access to pharmacotherapy and renal replacement therapy (RRT) is much stronger in high-income countries. Tellingly, Taiwan, Japan and the USA have the highest prevalence of treated end-stage renal disease (ESRD) [18]. These countries are characterized by robust registries, universal RRT access and high survival.

Prevalence estimates highlight areas of need. A meta-analysis of 123 country-level prevalence studies estimated that 4.902 million patients worldwide were in need of RRT in 2010 [19]. However, only 2.618 million patients were receiving RRT, suggesting that approximately 2.284 million patients were at risk of premature mortality related to RRT accessibility. RRT use is projected to more than double by 2030 to 5.439 million, mostly due to economic development leading to greater access to RRT [19]. The greatest growth is predicted to occur in Asia.

The etiology of CKD varies by country

In all developed countries and some developing countries, diabetes and hypertension are the main causes of CKD. In developing countries in Asia and Sub-Saharan Africa, where the transition from infectious disease to chronic disease is ongoing, CKD is often caused by glomerulonephritis and infectious diseases, due to poor sanitation, contaminated water and disease vectors [20]. In some developing countries, the balance of CKD etiologies is shifting. For example, in China, glomerulonephritis was surpassed by diabetes in 2011 as the leading cause of CKD [21]. The rising global prevalence of diabetes and hypertension will likely drive a rise in global CKD in the decades ahead.

Populations in China and Belgium are at risk for aristolochic acid nephropathy, a progressive renal interstitial fibrosis [22]. Rural populations in parts of Africa and Asia are at risk for nephrotoxicity from herbal remedies. Populations in Sri Lanka and India are at risk for CKD of unknown origin, a disease characterized by extensive interstitial fibrosis, tubular atrophy and rapid progression without proteinuria, possibly due to heavy metal contamination [23].

The advent of genome-wide association studies (GWAS) has linked several dozen loci to eGFR or albuminuria [24], including the UMOD gene, which encodes uromodulin, the most abundant protein in urine [25]. Ironically, these GWAS discoveries have turned the tables from the previous paradigm of ‘bench to bedside’ to ‘bedside to bench’: population studies point to causal genetic loci whose functional variants and mechanisms or action then need to be characterized.

African-Americans suffer disproportionately from non-diabetic causes of CKD, including focal segmental glomerulosclerosis (FSGS) and hypertension-attributed ESRD. The S342G and I384M variants of APOL1, a gene encoding apolipoprotein L1, are associated with FSGS and ESRD with an odds ratio of 10.5 and 7.3, respectively [26]. However, prospective studies found weaker associations (approximately a 2-fold risk) [27, 28], leaving risk prediction uncertain and leading researchers to search for a ‘second hit’ that accelerates CKD progression in APOL1-susceptible individuals. APOL1 susceptibility variants are present in 30% of African-American chromosomes and nearly absent in European-Americans [26]. These APOL1 variants may have been sustained as they confer protection against Trypanosoma brucei rhodesiense in African sleeping sickness.

GFR AND ALBUMINURIA STAGES ARE STRONG PREDICTORS OF A WIDE RANGE OF OUTCOMES

GFR and albuminuria stages predict cardiovascular mortality, acute kidney injury and ESRD

During the past decade there has been an expansion in the evidence base establishing that eGFR and albuminuria are independently associated with a wide range of renal and cardiovascular outcomes worldwide. The KDIGO Controversies Conference catalyzed the birth of the CKD Prognosis Consortium, which, in its inaugural year, produced three large-scale meta-analyses across general population cohorts, high-risk cohorts and CKD cohorts [29–31].

In the general population, a meta-analysis of nine cohorts consisting of 845 125 subjects found that the risk for ESRD was unrelated to eGFR, when eGFR remained above 75 mL/min/1.73 m², but increased exponentially at lower levels, after adjustment for albuminuria [32]. Log albuminuria was linearly associated with log ESRD risk without thresholds. This association of eGFR (with threshold) and albuminuria (without threshold) was also found to be true for progressive CKD and acute kidney injury (AKI) [32]. In addition, a meta-analysis of 21 cohorts consisting of 1 234 182 subjects from the general population found that reduced eGFR and increased albuminuria were independently associated with all-cause and cardiovascular mortality [29].

Figure 2 visualizes the relative risks of all-cause mortality, cardiovascular mortality, ESRD, AKI and progressive CKD in general population cohorts. For all five outcomes, there is a graded increase in risk of outcomes at higher levels of albuminuria categories independent of eGFR. Risk of outcomes rises by eGFR of 60 mL/min/1.73 m² and rises steeply in the 30–59 mL/min/1.73 m² range. This pattern holds for all outcomes, suggesting that groups with characteristics that make them at higher risk for one outcome are also at higher risk for all outcomes. The relative risks are higher for kidney outcomes than for all-cause and cardiovascular mortality, reflecting a greater specificity of association of eGFR and albuminuria with renal outcomes, as expected.

FIGURE 2

Adjusted hazards ratios (HR) or odds ratios (OR) for all-cause mortality, cardiovascular mortality, ESRD, AKI and progressive CKD in general population cohorts, with respect to eGFR on the horizontal axis and urine albumin to creatinine ratio (ACR) in color. The three lines represent ACR <30 mg/g or dipstick negative and trace in blue; ACR 30–299 mg/g or dipstick 1⁺ in green; and ACR ≥300 mg/g or dipstick  2⁺ in red. Solid circles indicate statistical significance compared with the reference point (P < 0.05), whereas solid triangles indicate non-significance. Red arrows indicate eGFR of 60 mL/min/1.73 m², the current threshold value for CKD. Diamonds indicate the reference points for the hazards ratios: at eGFR of 95 mL/min/1.73 m² and ACR of 30 mg/g or dipstick negative. Reproduced with permission from [10].

Open in new tabDownload slide

In individuals at high risk (defined as having a history of hypertension, diabetes or cardiovascular disease) for CKD, eGFR and albuminuria also independently predict a wide range of outcomes. A meta-analysis of 10 cohorts consisting of 266 975 subjects at risk for CKD found that reduced eGFR and log albuminuria were both associated with increased risk for all-cause mortality. Reduced eGFR and log albuminuria also associated independently with cardiovascular mortality, even after adjustment for traditional cardiovascular risk factors [31].

This pattern also holds in the CKD population. A meta-analysis of 13 cohorts consisting of 21 688 subjects with CKD found that lower eGFR and higher albuminuria both independently associate with mortality and ESRD [30]. A meta-analysis of five cohorts consisting of 79 519 subjects with CKD found that lower eGFR and higher albuminuria are associated with AKI [33]. A meta-analysis of three cohorts consisting of 3804 subjects with CKD found that eGFR and albuminuria in combination outperformed any single modifiable traditional cardiovascular risk factor in predicting cardiovascular mortality and stroke, but not coronary heart disease [34].

The presence of CKD worsens cardiovascular disease prognosis, as the pathophysiology of cardiovascular disease differs in the presence of CKD. At mild to moderate stages of the CKD (3A and 3B), the incidence of cardiovascular mortality exceeds the incidence of kidney failure. Only at stage 4 CKD does the risk of kidney failure begin to overtake that of cardiovascular events [35]. Reduced eGFR and albuminuria are also associated with increased risk of hospitalization with acute myocardial infarction, congestive heart failure, peripheral vascular disease and coronary artery bypass grafting or percutaneous coronary intervention [36]. A recent meta-analysis showed that eGFR and albuminuria improve cardiovascular risk prediction in all cohort types (general population, high risk for CKD and CKD cohorts) and suggested that differences in the literature may be explained by a stronger association with cardiovascular disease (CVD) mortality and heart failure than with coronary heart disease [34].

Improved risk prediction tools are emerging

While the staging of CKD by eGFR provides broad categories of prognosis, models have been created to predict the onset of kidney failure or mortality to provide individualized prognosis and improve planning for dialysis, transplantation and advance directives. Tangri et al. [37] introduced a model for predicting progression to kidney failure based on age, sex, eGFR and albuminuria, optionally adding serum calcium, serum phosphate, serum bicarbonate and serum albumin. Originally validated in a cohort of 4942 subjects with CKD stages 3–5, with a C-statistic of 0.917, the Tangri model was subsequently validated in 31 cohorts including 721 357 and 23 829 ESRD cases worldwide, demonstrating good discrimination, with a C-statistic of 0.90 [38]. However, risk for ESRD varied across cohorts: non-North American cohorts had 32.9% lower risk for ESRD at 2 years and 16.5% lower risk for ESRD at 5 years. All-cause mortality is another important endpoint for risk prediction in CKD. For example, Johnson et al. [39] introduced a model for predicting death based on age, sex, eGFR, diabetes, hypertension and anemia. This was validated in a cohort of 6541 subjects with CKD with a C-statistic of 0.70.

CKD risk prediction tools have not been as widely integrated into mainstream care, unlike in cardiovascular disease, but this may change as models are validated and appreciation for their use increases. Risk prediction of cardiovascular events within CKD and using CKD markers is a promising area for future advances.

KIDNEY FAILURE CAN BE PREVENTED

Evidence-based treatment can delay progression to kidney failure

Angiotensin-converting enzyme inhibitor (ACEI) and angiotensin II receptor blocker (ARB) monotherapy has long been a mainstay of CKD therapy, starting at >300 mg/24 h albuminuria for non-diabetic CKD and >30 mg/24 h albuminuria for diabetic CKD [9]. The ONTARGET trial, conducted in 25 620 subjects with atherosclerotic disease or diabetes with end-organ damage, showed that ACEI-ARB combination therapy with telmisartan and ramipril had a negligible additional benefit in blood pressure (BP) reduction, and increased the risk of hypotension, syncope, hyperkalemia, doubling of serum creatinine and dialysis [40]. Therefore, KDIGO does not recommend ACEI-ARB combination therapy [9].

Intensive BP control in CKD may not lead to added benefits. A systematic review of three trials totaling 2272 subjects with CKD showed that aggressive BP reduction to 130/80 mmHg versus 140/90 mmHg did not improve cardiovascular and renal outcomes [41], except in patients with proteinuria. Additionally, aggressive BP reduction caused more adverse events. However, a meta-analysis of 11 trials totaling 9287 subjects with CKD found that intensive BP reduction (defined variably) decreased the risk of a composite renal outcome [42]. The recent SPRINT trial consisting of 9361 subjects enriched with CKD patients showed that an intensive systolic BP goal of 120 mmHg versus 140 mmHg reduced risk of a composite cardiovascular outcome [43]. This suggests that in some patients, strict BP goals may have benefits. At this time, KDIGO recommends a target BP of  ≤140/90 mmHg, with exception for patients with albuminuria ≥30 mg/24 h, who can be targeted to ≤130/80 mmHg [9].

Statins reduce cardiovascular risk in CKD. The SHARP trial, comprising 9270 subjects with CKD, found that simvastatin and ezetimibe reduced major atherosclerotic events compared with placebo [44]. However, the results among the dialysis subpopulation alone were not statistically significant. There was no effect on development of ESRD or doubling of serum creatinine. A meta-analysis of 80 trials totaling 51 099 subjects with CKD found that statins reduced all-cause mortality, cardiovascular mortality and cardiovascular events in non-dialysis CKD, with a more modest effect in dialysis-dependent CKD [45]. Accordingly, KDIGO recommends initiating statin therapy in ≥50-year-old adults with CKD and younger patients with select risk factors. In dialysis-dependent CKD, statin therapy should not be initiated but may be continued if already started [46].

Most recently, the Empagliflozin, Cardiovascular Outcomes, and Mortality in Type 2 Diabetes trial (EMPA-REG) trial comprising 6185 subjects with diabetic CKD found that empagliflozin reduced creatinine doubling and need for RRT compared with placebo [47]. Empagliflozin is a sodium-glucose co-transporter 2 inhibitor, which also has thiazide-like effects in lowering BP and reducing sympathetic tone. The Canagliflozin and Renal Events in Diabetes with Established Nephropathy Clinical Evaluation Trial (CREDENCE) trial, which is ongoing until 2019, will determine if canagliflozin carries similar benefits for diabetic CKD.

Adjunctive therapies have been studied in a number of meta-analyses and systematic reviews. In CKD with acidosis, sodium bicarbonate administration lowers serum creatinine and need for dialysis [48]. At this time, KDIGO supports bicarbonate supplementation in CKD patients with serum bicarbonate <22 mmol/L [9]. Reduced protein intake reduces renal mortality [49]. Accordingly, KDIGO suggests lowering dietary protein intake to 0.8 g/kg/day when GFR falls <30 mL/min/1.73 m² and avoiding high dietary protein intake of  >1.3 g/kg/day in CKD patients at risk of progression [9]. Reduced salt intake is associated with a reduction in BP and albuminuria in both diabetic CKD [50] and non-diabetic CKD [51] patients. Accordingly, KDIGO recommends lowering salt intake to <2g/day of sodium unless contraindicated [9].

There is currently insufficient evidence regarding uric acid-lowering therapy, bisphosphonate therapy and vitamin D supplementation in CKD. A systematic review of 110 trials found that there is not yet enough evidence to support widespread implementation of CKD screening of asymptomatic adults [52], a consensus reflected in the KDIGO guidelines [9]. However, screening for CKD is cost-effective in high-risk groups, including patients with diabetes, hypertension or old age [53].

Comprehensive strategies are needed to address the growing CKD burden in lower and middle income countries

Of the 2.618 million patients receiving RRT noted in the ‘Prevalence’ section above, 92.8% reside in high-income and high- to middle-income countries, compared with 7.2% residing in low-income and low- to middle-income countries—a 70-fold prevalence gradient [19]. The largest treatment gaps exist in Asia and Africa. In China, India, Indonesia, Pakistan and Nigeria, which collectively comprise half of the world’s population, only ∼25% of patients requiring RRT are able to receive RRT. Not surprisingly, gross national income is associated with the availability of RRT [54].

Because CKD is a result of social, economic and physical health determinants, CKD management should be a nationally funded, preventive, interdisciplinary effort with attention to region-specific needs and etiologies [55]. Since 2003, Taiwan, which has the highest incidence and prevalence of ESRD worldwide, has banned aristolochic acid-containing herbs, launched public awareness campaigns, increased CKD research funding and introduced an interdisciplinary care program for patients before they progress to ESRD [56]. This comprehensive effort has paralleled a reduction in the annual incidence of ESRD in Taiwan from 432 per million population in 2005 to 361 per million population in 2010.

Furthermore, not all CKD patients need to be referred to nephrologists. As CKD becomes more commonplace, awareness should be raised of the new CKD staging and management guidelines [9] among general health professionals. KDIGO clarifies the criteria for referral to specialist kidney care to be CKD with AKI; GFR <30 mL/min/1.73 m²; albumin to creatinine ratio ≥300 mg/g; progressive; red blood cells >20/high-power field; refractory to four or more antihypertensive agents; persistent potassium abnormalities; recurrent or extensive nephrolithiasis; or hereditary kidney disease [9]. KDOQI adds that referral should also be made for CKD with albuminuria of uncertain etiology or albuminuria refractory to ACEIs and ARBs [57]. Recognizing the political and economic hurdles that must be surmounted to support RRT programs that are robust and accessible to all, it is crucial that we strengthen primary and secondary prevention measures.

CONCLUSION

Great strides have been made in the past decade under the paradigm of CKD as a common, harmful and treatable condition. The CKD-EPI equations for estimating GFR and standardized creatinine and cystatin assays paved the way for global estimates of CKD prevalence. The incorporation of albuminuria in the KDIGO classification of CKD and the associated meta-analyses paved the way for new research linking reduced GFR and increased albuminuria to a wide range of adverse renal and cardiovascular outcomes. Polymorphisms in uromodulin and apolipoprotein L1 and other genes were linked to CKD. Risk prediction tools are emerging to refine prognosis for individual patients. Referral criteria have been clarified, and standard treatment now avoids dual blockade therapy while BP control targets are less certain.

Looking forward, global efforts such as those outlined in the International Society of Nephrology framework, Global Kidney Health 2017 and Beyond [58] are needed. We must promote efforts to achieve uniform lab measurement of creatinine and albuminuria and adoption of the KDIGO 2012 classification to enable valid comparisons of CKD burden across time and region. We must increase the number, size and quality of clinical trials relating to CKD, especially in low- and middle-income countries, to facilitate evidence-based management. We hope the next decade of CKD epidemiology will be characterized by greater research that is standardized, randomized and global.

CONFLICT OF INTEREST STATEMENT

J.C. reports grants from National Kidney Foundation and is a member of Global Hyperkalemia Council (sponsored by Relypsa, no personal compensation). In addition, J.C. has an A110Q13 patent on GFR estimation: PCT/US2015/044567 provisional patent (Coresh, Inker and Levey) filed 15 August 2014, entitled ‘Precise estimation of glomerular filtration rate from multiple biomarkers’. The technology is not licensed in whole or in part to any company. Tufts Medical Center, Johns Hopkins University and Metabolon Inc. have a collaboration agreement to develop a product to estimate GFR from a panel of markers (25 June 2016).

REFERENCES