REVIEW URRENT C OPINION
Progress in risk prediction for people with chronic kidney disease Maarten W. Taal
Purpose of review Chronic kidney disease is common, but the associated risk of adverse outcomes is heterogeneous and methods for risk stratification are, therefore, required. We have reviewed recent progress in developing clinically applicable risk scores for people with chronic kidney disease. Recent findings Large epidemiological studies have confirmed that lower glomerular filtration rate and albuminuria are the most important risk factors for adverse outcomes in all populations studied. Ongoing research has identified several potentially important novel risk factors, including genetic factors, metabolomic factors, fibroblast growth factor 23, novel cardiovascular risk factors [CXC motif, ligand 12 (CXCL12) and ceruloplasmin], skin autofluorescence and inflammatory markers (serum-free light chains and circulating receptors for tumour necrosis factor). Several risk scores have been developed and one in particular (by Tangri et al.) performs well, uses variables commonly assessed in clinical practice and has been externally validated. Further studies are required to assess its performance in populations outside of secondary care and its usefulness in guiding decision making in clinical practice. Summary Although substantial progress has been made, the goal of achieving simple robust risk scores which are widely applicable and effective in guiding clinical decision making remains to be achieved. Keywords chronic kidney disease, outcomes, risk factor, risk prediction, risk score
INTRODUCTION Global focus on the epidemiology of chronic kidney disease (CKD) has revealed that it is common (affecting 10–14% of adult populations) and is associated with important adverse outcomes, including endstage kidney disease (ESKD), acute kidney injury, cardiovascular disease (CVD) and all-cause mortality. It is also clear, however, that these risks vary substantially between affected individuals; for example, in one large study, the risk of ESKD varied from 0.02 per 1000 person-years in the lowest risk category to as high as 195 per 1000 person-years [1]. This heterogeneity and the high prevalence of CKD make a strong case for the development of risk stratification tools to identify people at high risk of adverse outcomes for targeted intervention to ameliorate the risk and spare people at low risk the anxiety and burden of further investigation or treatment. Risk prediction tools have been developed for many other chronic diseases, most notably CVD, but these risk scores substantially underestimate risk in populations with CKD. In this paper,
we review recent progress towards the goal of clinically applicable risk prediction tools for people with CKD.
ESTIMATED GLOMERULAR FILTRATION RATE AND ALBUMINURIA AS RISK FACTORS The work of the CKD Prognosis Consortium has provided robust evidence that reduced estimated GFR (eGFR) and albuminuria are the principal factors independently associated with adverse outcomes in people with CKD. By combining cohorts Department of Renal Medicine, Royal Derby Hospital, Derby, Derbyshire, UK Correspondence to Maarten W. Taal, Department of Renal Medicine, Royal Derby Hospital, Uttoxeter Road, Derby, Derbyshire, DE22 3NE, UK. Tel: +441332 789344; fax: +441332 789352; e-mail: maarten. [email protected] Curr Opin Nephrol Hypertens 2014, 23:519–524 DOI:10.1097/MNH.0000000000000072
1062-4821 ß 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins
www.co-nephrolhypertens.com
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Diagnostics and techniques
KEY POINTS Large epidemiological studies have identified albuminuria and reduced GFR as the principal risk factors for adverse outcomes in CKD. Novel risk factors continue to be investigated to improve risk prediction and guide the development of new therapies. Progress has been made in developing risk scores for risk stratification in CKD. One risk score in particular performs well and has been externally validated, but the impact of its introduction to clinical practice has not yet been tested. The goal of achieving simple robust risk scores that are widely applicable and effective in guiding clinical decision making remains to be achieved.
from multiple studies, the group has been able to generate study populations of over a million people. Their analyses have shown that reduced eGFR and albuminuria are independent risk factors for ESKD [2], acute kidney injury [1], cardiovascular and allcause mortality [3,4]. Further analysis of subgroups has shown that these associations persist independent of the presence of diabetes [5] or hypertension [6] and are independent of age [7], sex [8 ] and ethnicity [9 ]. Recognition of the importance of these observations has prompted the Kidney Disease Improving Global Outcomes (KDIGO) and National Institute for Health and Care Excellence to recommend a revision of the classification system for CKD to reflect both reduced GFR and albuminuria [10 ]. This is an important development because for the first time, the CKD category (defined by GFR categories G1–5 corresponding to the previous CKD stages and albuminuria categories 1–3 corresponding to previous definitions of normoalbuminuria, microalbuminuria and macroalbuminuria) is explicitly linked to risk of adverse outcomes, and this already goes a long way to achieving simple risk stratification for large populations (Fig. 1). The primacy of GFR as a risk factor has focussed attention on methods for estimating it. Widespread adoption of the Modification of Diet in Renal Disease (MDRD) equation has revolutionized the detection of CKD and has made possible much of the epidemiological research discussed. Nevertheless, it was recognized from the outset that the MDRD equation was imperfect and particularly that it tended to underestimate true GFR at values above 60 ml/min/1.73 m2. This is particularly important because this is the threshold below which CKD may be diagnosed without other evidence of kidney damage. Recognition of these limitations prompted &
&
&
520
www.co-nephrolhypertens.com
the development of several other equations to estimate GFR from serum creatinine concentration culminating in a recommendation by KDIGO that the MDRD equation should be replaced by the CKD Epidemiology Collaboration (CKD-EPI) equation, which is more accurate than the MDRD equation and results in less bias, particularly at GFR values above 60. Further analysis by the CKD Prognosis Consortium found that eGFR estimated with the CKD-EPI equation results in a lower prevalence of CKD stage 3–5 (8.7 versus 6.3%) and affords better risk prediction than the MDRD equation. Amongst people classified as having eGFR 59–45 ml/min/ 1.73 m2 by the MDRD equation, 34.7% were reclassified as eGFR 89–60 ml/min/1.73 m2 by CKD-EPI, and those reclassified evidenced a lower incidence of adverse outcomes versus those not reclassified (e.g., incidence of all-cause mortality 9.9 versus 34.5 per 1000 person-years) [11]. The limitations of creatinine as a marker of GFR because of nonrenal factors that affect serum creatinine concentration, including muscle mass and diet, have been appreciated for decades, prompting a search for alternatives. Cystatin C, a peptide produced by all nucleated cells and, therefore, not affected by muscle mass, has emerged as the most promising alternative to creatinine. The production of reference material to standardize cystatin C assays has greatly improved the potential for clinical application, and equations have been developed to estimate GFR from serum cystatin C concentration or from creatinine and cystatin C together. The combined equation and CKD-EPI (creatinine) equation produced similar bias, but the combined equation gave better precision and accuracy [12]. Further work by the CKD Prognosis Consortium reported that when cystatin C was used to estimate GFR, reclassification to a higher GFR category (than assigned by eGFR creatinine) was associated with a lower risk of adverse outcomes. For example, amongst those with CKD-EPI (creatinine) eGFR 59–45 ml/min/1.73 m2, application of cystatin C resulted in reclassification of 42% to eGFR of at least 60 ml/min/1.73 m2 and reclassification was associated with a 34% reduction in adjusted risk of all-cause mortality [13 ]. It should be noted, however, that in this study, CKD was defined by a single eGFR value. It therefore remains to be established whether the use of cystatin C will improve risk prediction in people with CKD defined by two eGFR values at least 90 days apart, as required by the KDIGO definition. &
NOVEL RISK FACTORS Numerous studies over the past decade have identified risk factors in addition to GFR and albuminuria Volume 23 Number 6 November 2014
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Progress in risk prediction for people with CKD Taal
Persistent albuminuria categories description and range A1
A2
A3
Normal to mildly increased
Moderately increased
Severely increased
30 mg/mmol
GFR categories (ml/min/1.73 m2) description and range
Prognosis of CKD by GFR and albuminuria categories: KDIGO 2012
≥90
G1
Normal or high
G2
Mildly decreased
60–89
G3a
Mildly to moderately decreased
45–59
G3b
Moderately to severely decreased
30–44
G4
Severely decreased
15–29
G5
Kidney failure
Progress in risk prediction for people with chronic kidney disease Maarten W. Taal
Purpose of review Chronic kidney disease is common, but the associated risk of adverse outcomes is heterogeneous and methods for risk stratification are, therefore, required. We have reviewed recent progress in developing clinically applicable risk scores for people with chronic kidney disease. Recent findings Large epidemiological studies have confirmed that lower glomerular filtration rate and albuminuria are the most important risk factors for adverse outcomes in all populations studied. Ongoing research has identified several potentially important novel risk factors, including genetic factors, metabolomic factors, fibroblast growth factor 23, novel cardiovascular risk factors [CXC motif, ligand 12 (CXCL12) and ceruloplasmin], skin autofluorescence and inflammatory markers (serum-free light chains and circulating receptors for tumour necrosis factor). Several risk scores have been developed and one in particular (by Tangri et al.) performs well, uses variables commonly assessed in clinical practice and has been externally validated. Further studies are required to assess its performance in populations outside of secondary care and its usefulness in guiding decision making in clinical practice. Summary Although substantial progress has been made, the goal of achieving simple robust risk scores which are widely applicable and effective in guiding clinical decision making remains to be achieved. Keywords chronic kidney disease, outcomes, risk factor, risk prediction, risk score
INTRODUCTION Global focus on the epidemiology of chronic kidney disease (CKD) has revealed that it is common (affecting 10–14% of adult populations) and is associated with important adverse outcomes, including endstage kidney disease (ESKD), acute kidney injury, cardiovascular disease (CVD) and all-cause mortality. It is also clear, however, that these risks vary substantially between affected individuals; for example, in one large study, the risk of ESKD varied from 0.02 per 1000 person-years in the lowest risk category to as high as 195 per 1000 person-years [1]. This heterogeneity and the high prevalence of CKD make a strong case for the development of risk stratification tools to identify people at high risk of adverse outcomes for targeted intervention to ameliorate the risk and spare people at low risk the anxiety and burden of further investigation or treatment. Risk prediction tools have been developed for many other chronic diseases, most notably CVD, but these risk scores substantially underestimate risk in populations with CKD. In this paper,
we review recent progress towards the goal of clinically applicable risk prediction tools for people with CKD.
ESTIMATED GLOMERULAR FILTRATION RATE AND ALBUMINURIA AS RISK FACTORS The work of the CKD Prognosis Consortium has provided robust evidence that reduced estimated GFR (eGFR) and albuminuria are the principal factors independently associated with adverse outcomes in people with CKD. By combining cohorts Department of Renal Medicine, Royal Derby Hospital, Derby, Derbyshire, UK Correspondence to Maarten W. Taal, Department of Renal Medicine, Royal Derby Hospital, Uttoxeter Road, Derby, Derbyshire, DE22 3NE, UK. Tel: +441332 789344; fax: +441332 789352; e-mail: maarten. [email protected] Curr Opin Nephrol Hypertens 2014, 23:519–524 DOI:10.1097/MNH.0000000000000072
1062-4821 ß 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins
www.co-nephrolhypertens.com
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Diagnostics and techniques
KEY POINTS Large epidemiological studies have identified albuminuria and reduced GFR as the principal risk factors for adverse outcomes in CKD. Novel risk factors continue to be investigated to improve risk prediction and guide the development of new therapies. Progress has been made in developing risk scores for risk stratification in CKD. One risk score in particular performs well and has been externally validated, but the impact of its introduction to clinical practice has not yet been tested. The goal of achieving simple robust risk scores that are widely applicable and effective in guiding clinical decision making remains to be achieved.
from multiple studies, the group has been able to generate study populations of over a million people. Their analyses have shown that reduced eGFR and albuminuria are independent risk factors for ESKD [2], acute kidney injury [1], cardiovascular and allcause mortality [3,4]. Further analysis of subgroups has shown that these associations persist independent of the presence of diabetes [5] or hypertension [6] and are independent of age [7], sex [8 ] and ethnicity [9 ]. Recognition of the importance of these observations has prompted the Kidney Disease Improving Global Outcomes (KDIGO) and National Institute for Health and Care Excellence to recommend a revision of the classification system for CKD to reflect both reduced GFR and albuminuria [10 ]. This is an important development because for the first time, the CKD category (defined by GFR categories G1–5 corresponding to the previous CKD stages and albuminuria categories 1–3 corresponding to previous definitions of normoalbuminuria, microalbuminuria and macroalbuminuria) is explicitly linked to risk of adverse outcomes, and this already goes a long way to achieving simple risk stratification for large populations (Fig. 1). The primacy of GFR as a risk factor has focussed attention on methods for estimating it. Widespread adoption of the Modification of Diet in Renal Disease (MDRD) equation has revolutionized the detection of CKD and has made possible much of the epidemiological research discussed. Nevertheless, it was recognized from the outset that the MDRD equation was imperfect and particularly that it tended to underestimate true GFR at values above 60 ml/min/1.73 m2. This is particularly important because this is the threshold below which CKD may be diagnosed without other evidence of kidney damage. Recognition of these limitations prompted &
&
&
520
www.co-nephrolhypertens.com
the development of several other equations to estimate GFR from serum creatinine concentration culminating in a recommendation by KDIGO that the MDRD equation should be replaced by the CKD Epidemiology Collaboration (CKD-EPI) equation, which is more accurate than the MDRD equation and results in less bias, particularly at GFR values above 60. Further analysis by the CKD Prognosis Consortium found that eGFR estimated with the CKD-EPI equation results in a lower prevalence of CKD stage 3–5 (8.7 versus 6.3%) and affords better risk prediction than the MDRD equation. Amongst people classified as having eGFR 59–45 ml/min/ 1.73 m2 by the MDRD equation, 34.7% were reclassified as eGFR 89–60 ml/min/1.73 m2 by CKD-EPI, and those reclassified evidenced a lower incidence of adverse outcomes versus those not reclassified (e.g., incidence of all-cause mortality 9.9 versus 34.5 per 1000 person-years) [11]. The limitations of creatinine as a marker of GFR because of nonrenal factors that affect serum creatinine concentration, including muscle mass and diet, have been appreciated for decades, prompting a search for alternatives. Cystatin C, a peptide produced by all nucleated cells and, therefore, not affected by muscle mass, has emerged as the most promising alternative to creatinine. The production of reference material to standardize cystatin C assays has greatly improved the potential for clinical application, and equations have been developed to estimate GFR from serum cystatin C concentration or from creatinine and cystatin C together. The combined equation and CKD-EPI (creatinine) equation produced similar bias, but the combined equation gave better precision and accuracy [12]. Further work by the CKD Prognosis Consortium reported that when cystatin C was used to estimate GFR, reclassification to a higher GFR category (than assigned by eGFR creatinine) was associated with a lower risk of adverse outcomes. For example, amongst those with CKD-EPI (creatinine) eGFR 59–45 ml/min/1.73 m2, application of cystatin C resulted in reclassification of 42% to eGFR of at least 60 ml/min/1.73 m2 and reclassification was associated with a 34% reduction in adjusted risk of all-cause mortality [13 ]. It should be noted, however, that in this study, CKD was defined by a single eGFR value. It therefore remains to be established whether the use of cystatin C will improve risk prediction in people with CKD defined by two eGFR values at least 90 days apart, as required by the KDIGO definition. &
NOVEL RISK FACTORS Numerous studies over the past decade have identified risk factors in addition to GFR and albuminuria Volume 23 Number 6 November 2014
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Progress in risk prediction for people with CKD Taal
Persistent albuminuria categories description and range A1
A2
A3
Normal to mildly increased
Moderately increased
Severely increased
30 mg/mmol
GFR categories (ml/min/1.73 m2) description and range
Prognosis of CKD by GFR and albuminuria categories: KDIGO 2012
≥90
G1
Normal or high
G2
Mildly decreased
60–89
G3a
Mildly to moderately decreased
45–59
G3b
Moderately to severely decreased
30–44
G4
Severely decreased
15–29
G5
Kidney failure
