Abstract
Chronic kidney disease (CKD) is strongly associated with increased risks of progression to end-stage kidney disease (ESKD) and mortality. Clinical trials evaluating CKD progression commonly use a composite end point of death, ESKD or serum creatinine doubling. However, due to low event rates, such trials require large sample sizes and long-term follow-up for adequate statistical power. As a result, very few interventions targeting CKD progression have been tested in randomized controlled trials. To overcome this problem, the National Kidney Foundation and Food and Drug Administration conducted a series of analyses to determine whether an end point of 30 or 40% decline in estimated glomerular filtration rate (eGFR) over 2–3 years can substitute for serum creatinine doubling in the composite end point. These analyses demonstrated that these alternate kidney end points were significantly associated with subsequent risks of ESKD and death. However, the association between, and consistency of treatment effects on eGFR decline and clinical end points were influenced by baseline eGFR, follow-up duration and acute hemodynamic effects. The investigators concluded that a 40% eGFR decline is broadly acceptable as a kidney end point across a wide baseline eGFR range and that a 30% eGFR decline may be acceptable in some situations. Although these alternate kidney end points could potentially allow investigators to conduct shorter duration clinical trials with smaller sample sizes thereby generating evidence to guide clinical decision-making in a timely manner, it is uncertain whether these end points will improve trial efficiency and feasibility. This review critically appraises the evidence, strengths and limitations pertaining to eGFR end points.
INTRODUCTION
Chronic kidney disease (CKD) is a major public health problem affecting 8–12% of adults in developed countries [1, 2]. Even in its early stages, CKD is associated with significantly increased risks of end-stage kidney disease (ESKD) and all-cause and cardiovascular mortality, independent of other cardiovascular risk factors [3, 4]. Although prevention of CKD progression and cardiovascular complications has been identified as a major research priority from the perspectives of patients, healthcare providers and policy-makers [5], current available therapies are limited to blood pressure lowering agents, such as angiotensin-converting enzyme inhibitors and angiotensin-receptor blockers [6, 7] and statins [8]. These agents are only modestly effective and most people with CKD continue to suffer further declines in kidney function [9] and unacceptably high rates of cardiovascular morbidity and mortality [10, 11]. Therefore, there is an urgent need to conduct randomized controlled trials to evaluate more treatment options in this patient population.
Conducting trials in kidney disease progression is challenging because it frequently takes many years for CKD to progress to ESKD and because dropout rates from trials are often high due to the high number of cardiovascular events. As a result, these trials need a large sample size and several years of follow-up. To overcome this problem, investigators have commonly used a composite end point of death, ESKD (typically defined as needing dialysis or kidney transplantation) and serum creatinine doubling. Serum creatinine doubling approximately corresponds to a halving of glomerular filtration rate (GFR) and generally accounts for the bulk of the events comprising the composite kidney end point. However, it is a poorly validated surrogate end point.
Recently, the National Kidney Foundation (NKF) and Food and Drug Administration (FDA) conducted a scientific workshop to determine whether another surrogate end point of a <50% reduction in eGFR (henceforth referred to as lesser eGFR declines) can effectively substitute for serum creatinine doubling in the composite kidney end point. The NKF-FDA Scientific Workshop Committee subsequently published a series of meta-analyses and a simulation study addressing this point. The committee concluded that while a 30% eGFR decline ‘may’ be an acceptable surrogate end point in ‘some’ circumstances, a 40% decline is more broadly acceptable across a wider range of baseline eGFRs and warrants ‘careful’ consideration in CKD progression trials. For both 30 and 40% eGFR declines, the committee recommended a minimum follow-up of at least 2–3 years and repeat serum creatinine measurements at baseline and after reaching the end point to confirm the magnitude of the decline. For 30% eGFR decline, the workshop recommended examination of the pattern of treatment effects on eGFR, specifically including early acute effects.
From the perspectives of patients, clinicians, health policy-makers and trialists, the NKF-FDA Scientific Workshop recommendations offer some encouraging news. There is currently a dearth of high-level evidence guiding shared clinical decision-making in nephrology, which has the lowest rate of published randomized controlled trials among all medical specialties [12, 13]. This is partly because significant barriers to mounting adequately powered randomized controlled trials include slow recruitment and limited financial resources [14]. The use of valid surrogate end points would allow investigators to design clinical trials with relatively smaller sample sizes, shorter follow-up durations and lower costs. Importantly, a relatively shorter duration of follow-up may allow earlier implementation of trial results into clinical practice. This article critically appraises the recently published meta-analyses [15–18] and discusses the challenges, strengths and weaknesses of various clinical and surrogate ‘kidney function’ end points in CKD progression trials.
ESTABLISHED KIDNEY DISEASE PROGRESSION END POINTS
Table 1 describes the advantages and disadvantages of the existing and proposed end points in CKD progression trials. Table 2 describes the results of a few selected randomized trials using various kidney end points. These trials were hand-searched and selected to describe how the use of various kidney end points has evolved in the last 25 years.
Advantages and disadvantages of end points in CKD progression trials
| End point | Advantages | Disadvantages |
|---|---|---|
| Clinical end point | ||
| Death | Well defined Important to everyone due to the irreversible nature Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Occurs late in the disease process Relatively fewer event rates Trials require longer follow-up, larger sample size, more funding and other resources Not feasible to conduct trials in the earlier stages of the disease process |
| End-stage kidney disease (dialysis or transplantation for at least 1–3 months, or reaching eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2) | Clinically important as it indicates irreversible loss of an organ, needs disproportionate healthcare resources, associated with increased mortality and reduced quality of life Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Not well defined Timing of initiation of dialysis or transplantation is highly subjective Small chance of recovery of kidney function |
| Surrogate end point | ||
| Doubling of serum creatinine (corresponds to 57% reduction in eGFR if using the CKD-EPI equation) | It is strongly associated with subsequent risks of ESKD and death Treatment effect on doubling of serum creatinine is same as that on ESKD Significantly higher number of events than death and ESKD in a relatively shorter period, thus substantially decreasing the required sample size and the follow-up period | Serum creatinine is an imprecise marker of kidney function, affected by several non-GFR factors Not a patient-reported outcome Not always an irreversible event Its clinical importance is less than that of death and ESKD |
| 40% decline in eGFR over 2–3 years | Number of event rates even higher than doubling of serum creatinine, which will reduce the required sample size by 20–35% It will allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It will allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Affected by non-GFR determinants of serum creatinine Treatment effect on 40% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
| 30% decline in eGFR over 2–3 years | Number of event rates even higher than 40% eGFR decline, which may reduce the required sample size even further It may allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It may allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Cannot be used if an acute effect on GFR is present Cannot be used if the treatment effect is proportional (larger effect in fast progressors than non-progressors) Cannot be used in the advanced (later) stages of CKD Affected by non-GFR determinants of serum creatinine Treatment effect on 30% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is substantially less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
| End point | Advantages | Disadvantages |
|---|---|---|
| Clinical end point | ||
| Death | Well defined Important to everyone due to the irreversible nature Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Occurs late in the disease process Relatively fewer event rates Trials require longer follow-up, larger sample size, more funding and other resources Not feasible to conduct trials in the earlier stages of the disease process |
| End-stage kidney disease (dialysis or transplantation for at least 1–3 months, or reaching eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2) | Clinically important as it indicates irreversible loss of an organ, needs disproportionate healthcare resources, associated with increased mortality and reduced quality of life Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Not well defined Timing of initiation of dialysis or transplantation is highly subjective Small chance of recovery of kidney function |
| Surrogate end point | ||
| Doubling of serum creatinine (corresponds to 57% reduction in eGFR if using the CKD-EPI equation) | It is strongly associated with subsequent risks of ESKD and death Treatment effect on doubling of serum creatinine is same as that on ESKD Significantly higher number of events than death and ESKD in a relatively shorter period, thus substantially decreasing the required sample size and the follow-up period | Serum creatinine is an imprecise marker of kidney function, affected by several non-GFR factors Not a patient-reported outcome Not always an irreversible event Its clinical importance is less than that of death and ESKD |
| 40% decline in eGFR over 2–3 years | Number of event rates even higher than doubling of serum creatinine, which will reduce the required sample size by 20–35% It will allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It will allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Affected by non-GFR determinants of serum creatinine Treatment effect on 40% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
| 30% decline in eGFR over 2–3 years | Number of event rates even higher than 40% eGFR decline, which may reduce the required sample size even further It may allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It may allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Cannot be used if an acute effect on GFR is present Cannot be used if the treatment effect is proportional (larger effect in fast progressors than non-progressors) Cannot be used in the advanced (later) stages of CKD Affected by non-GFR determinants of serum creatinine Treatment effect on 30% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is substantially less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; GFR, glomerular filtration rate.
Advantages and disadvantages of end points in CKD progression trials
| End point | Advantages | Disadvantages |
|---|---|---|
| Clinical end point | ||
| Death | Well defined Important to everyone due to the irreversible nature Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Occurs late in the disease process Relatively fewer event rates Trials require longer follow-up, larger sample size, more funding and other resources Not feasible to conduct trials in the earlier stages of the disease process |
| End-stage kidney disease (dialysis or transplantation for at least 1–3 months, or reaching eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2) | Clinically important as it indicates irreversible loss of an organ, needs disproportionate healthcare resources, associated with increased mortality and reduced quality of life Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Not well defined Timing of initiation of dialysis or transplantation is highly subjective Small chance of recovery of kidney function |
| Surrogate end point | ||
| Doubling of serum creatinine (corresponds to 57% reduction in eGFR if using the CKD-EPI equation) | It is strongly associated with subsequent risks of ESKD and death Treatment effect on doubling of serum creatinine is same as that on ESKD Significantly higher number of events than death and ESKD in a relatively shorter period, thus substantially decreasing the required sample size and the follow-up period | Serum creatinine is an imprecise marker of kidney function, affected by several non-GFR factors Not a patient-reported outcome Not always an irreversible event Its clinical importance is less than that of death and ESKD |
| 40% decline in eGFR over 2–3 years | Number of event rates even higher than doubling of serum creatinine, which will reduce the required sample size by 20–35% It will allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It will allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Affected by non-GFR determinants of serum creatinine Treatment effect on 40% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
| 30% decline in eGFR over 2–3 years | Number of event rates even higher than 40% eGFR decline, which may reduce the required sample size even further It may allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It may allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Cannot be used if an acute effect on GFR is present Cannot be used if the treatment effect is proportional (larger effect in fast progressors than non-progressors) Cannot be used in the advanced (later) stages of CKD Affected by non-GFR determinants of serum creatinine Treatment effect on 30% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is substantially less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
| End point | Advantages | Disadvantages |
|---|---|---|
| Clinical end point | ||
| Death | Well defined Important to everyone due to the irreversible nature Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Occurs late in the disease process Relatively fewer event rates Trials require longer follow-up, larger sample size, more funding and other resources Not feasible to conduct trials in the earlier stages of the disease process |
| End-stage kidney disease (dialysis or transplantation for at least 1–3 months, or reaching eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2) | Clinically important as it indicates irreversible loss of an organ, needs disproportionate healthcare resources, associated with increased mortality and reduced quality of life Easy to interpret Useful for cost-effective analysis Widely accepted by drug approving authorities | Not well defined Timing of initiation of dialysis or transplantation is highly subjective Small chance of recovery of kidney function |
| Surrogate end point | ||
| Doubling of serum creatinine (corresponds to 57% reduction in eGFR if using the CKD-EPI equation) | It is strongly associated with subsequent risks of ESKD and death Treatment effect on doubling of serum creatinine is same as that on ESKD Significantly higher number of events than death and ESKD in a relatively shorter period, thus substantially decreasing the required sample size and the follow-up period | Serum creatinine is an imprecise marker of kidney function, affected by several non-GFR factors Not a patient-reported outcome Not always an irreversible event Its clinical importance is less than that of death and ESKD |
| 40% decline in eGFR over 2–3 years | Number of event rates even higher than doubling of serum creatinine, which will reduce the required sample size by 20–35% It will allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It will allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Affected by non-GFR determinants of serum creatinine Treatment effect on 40% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
| 30% decline in eGFR over 2–3 years | Number of event rates even higher than 40% eGFR decline, which may reduce the required sample size even further It may allow trials to be conducted in a shorter period of follow-up, such as 2 or 3 years It may allow trials to be conducted in patient populations with near normal or mildly decreased eGFR Easy to interpret due to widespread use of automatic reporting of eGFR | Cannot be used if an acute effect on GFR is present Cannot be used if the treatment effect is proportional (larger effect in fast progressors than non-progressors) Cannot be used in the advanced (later) stages of CKD Affected by non-GFR determinants of serum creatinine Treatment effect on 30% eGFR decline may be substantially less (weaker) than that of doubling of serum creatinine, thus will not reduce the sample size as expected Not always an irreversible event, needs to be confirmed by another measurement after 1 month Not a patient-reported outcome Its clinical importance is substantially less than that of death and ESKD It is not a substitute for death or ESKD, but for doubling of serum creatinine |
CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; GFR, glomerular filtration rate.
Description of trial end points in selected chronic kidney disease progression trials
| Trial | Study population, interventions and follow-up | Kidney outcome measures |
|---|---|---|
| Locatelli 1991 [32] | Population: (n = 456) Creatinine clearance <60 mL/min, serum creatinine 133–619 μmol/L; excluded: diabetes and nephritic syndrome Interventions: Low-protein diet (n = 226) versus normal protein diet (n = 230) Baseline kidney function: Data according to the randomization groups or collective data were not reported. Follow-up: 2 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total 69 events/15%) Low-protein diet (27/12%) versus normal protein diet (42/18%); P = 0.059 Secondary: Data on individual events was not reported. Death: 2 (1%) versus 3 (1%) |
| Lewis 1993 [33] | Population: (n = 409) Type 1 diabetes mellitus, proteinuria ≥0.5 g/day, serum creatinine ≤2.5 g/dL (221 μmol/L) Interventions: Captopril (n = 207) versus placebo (n = 202) Baseline kidney function: 24-h creatinine clearance 84 ± 46 mL/min versus 79 ± 35 mL/min Follow-up: Median 3 years (range 1.8–4.8 years) | Primary: Doubling of serum creatinine (total 68/17%) Captopril (25/12%) versus placebo (43/21%); P = 0.007 Secondary: Mean rate of decline in creatinine clearance: 11 ± 21%/year versus 17 ± 20%/year; P = 0.03 Death or ESKD: 23 (11%) versus 42 (21%); P = 0.006 Death: 8 (4%) versus 14 (7%) ESKD (dialysis or transplantation): 20 (10%) versus 31 (15%) |
| Klahr 1994 [41] MDRD Study 1 | Population: (n = 585) GFR 25–55 mL/min/1.73 m2 Intervention 1: Usual protein diet versus low-protein diet Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 38.6 ± 8.9 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Usual protein diet 12.1 (10.5–13.8) mL/min/3 years versus low-protein diet 10.9 (9.2–12.5) mL/min/3 years; P = 0.3 Intervention 2: Usual BP 12.3 (10.6–14) mL/min/3 years versus low BP 10.7 (9.1–12.4) mL/min/3 years; P = 0.18 Secondary: (Data according to the randomization groups was not reported): Death: 15 (3%) ESKD: 12 (2%) Rapid decline in GFR (>50 reduction in GFR if baseline GFR ≤ 40 mL/min/1.73 m2 or ≥20 mL/min/1.73 m2 reduction in GFR if baseline GFR > 40 mL/min/1.73 m2): 60 (10%) |
| Klahr 1994 [41] MDRD Study 2 | Population: (n = 255) GFR 13–24 mL/min/1.73 m2 Intervention 1: Low-protein diet versus very low-protein diet with keto acid-amino acid supplement Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 18.5 ± 3.4 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Low-protein diet 4.4 (3.7–5.1) mL/min/year versus very low-protein diet 3.6 (2.9–4.2) mL/min/year; P = 0.07 Intervention 2: Usual BP 4.2 (3.6–4.9) mL/min/year versus low BP 3.7 (3.1–4.3) mL/min/year; P = 0.28 Secondary: (Data according to the randomization groups was not reported): Death: 15 (6%) ESKD: 94 (37%) |
| Toto 1995 [34] | Population: (n = 77) Long-standing hypertension, GFR ≤ 70 mL/min/1.73 m2), normal urine sediment, excluded: diabetes Interventions: Strict BP control (n = 42) versus conventional BP control (n = 35) Baseline kidney function: GFR 34.6 ± 2.3 versus 41.9 ± 3.15 mL/min/1.73 m2 Follow-up: Mean 40.5 ± 1.8 months versus 42.2 ± 2.1 months | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope −0.31 ± 0.45 versus −0.05 ± 0.5 mL/min/1.73 m2/year; P > 0.25 Secondary: 50% reduction in GFR (or doubling of serum creatinine) or ESRD or death: 12 (29%) versus 7(20%); P > 0.25 50% reduction in GFR (or doubling of serum creatinine): 4 (10%) versus 5(14%) ESKD: 7 (17%) versus 2 (6%) Death: 1 (2%) versus 0 (0%) |
| Maschio 1996 [35] | Population: (n = 583) serum creatinine 1.5–4 mg/dL (133–354 μmol/L), 24-h creatinine clearance 30–60 mL/min Interventions: Benazepril (n = 300) versus placebo (n = 283) Baseline kidney function: Creatinine clearance 42.9 ± 11.6 mL/min versus 42.3 ± 10.6 mL/min Follow-up: 3 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total events 88/15%) Benazepril 31 (10%) versus placebo 57 (20%); P < 0.001 Secondary: Doubling of serum creatinine: total events 86 (15%) Data according to the randomization groups was not reported. Dialysis: total events 2 (<1%) Data according to the randomization groups was not reported. Death: 8 (3%) versus 1 (<1%) |
| The GISEN Group 1997 [50] REIN Study Stratum 2 | Population: (n = 166) Non-diabetic nephropathy, creatinine clearance 20–70 mL/min/1.73 m2, proteinuria ≥1 g/day (≥3 g/day for stratum 2); excluded: Type 1 diabetes mellitus Interventions: Ramipril (n = 78) versus placebo (n = 88) Baseline kidney function: GFR 40.2 ± 19 mL/min/1.73 m2 versus 37.4 ± 17.5 mL/min/1.73 m2 Follow-up: Mean 16 months | Primary: Change in GFR (plasma clearance of non-radioactive iohexol), results expressed as mean (SE): 0.53 (0.08) versus 0.88 (0.13) mL/min/month; P = 0.03 Secondary: Doubling of serum creatinine or ESKD: 18 (23%) versus 40 (45%); P = 0.02 ESKD: 17 (22%) versus 29 (33%); P = 0.2 Death: 2 (3%) versus 1 (1%) |
| Kuriyama 1997 [36] | Population: (n = 73) serum creatinine 2–4 mg/dL, hematocrit <30% Interventions: No EPO (n = 31) versus EPO (n = 42) Baseline kidney function: Creatinine clearance 17.1 ± 7.2 mL/min versus 19.1 ± 7.2 mL/min Follow-up: 3 years | Primary: Doubling of serum creatinine (total 48 events/66%) No EPO group 26 (84%) versus EPO group 22 (52%); P = 0.0003 Secondary: Dialysis: 20 (65%) versus 14 (33%) Death: 2 (%) versus 1 (%) |
| Brenner 2001 [10] (RENAAL) | Population: (n = 1513) Type 2 diabetes mellitus, albuminuria >300 mg/g (or >0.5 g/day), serum creatinine 1.3–3 mg/dL (115–265 μmol/L) or 1.5–3 mg/dL (133–265 μmol/L) if weight >60 kg; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Losartan (n = 751) versus placebo (n = 762) Baseline kidney function: serum creatinine 1.9 ± 0.5 mg/dL versus 1.9 ± 0.5 mg/dL Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESRD or death (total 686 events/45%) Losartan 327 (44%) versus placebo 359 (47%); P = 0.02 Secondary: Doubling of serum creatinine: 162 (22%) versus 198 (26%); P = 0.006 ESKD: 147 (20%) versus 194(26%); P = 0.002 Death: 158 (21%) versus 155 (20%); P = 0.88 ESKD or death: 255 (12%) versus 300 (39%); P = 0.01 Doubling of serum creatinine or ESRD: 226 (30%) versus 263 (35%); P = 0.01 Reciprocal of serum creatinine (median slope): −0.056 dL/mg/year versus −0.069 dL/mg/year; P = 0.01 Median change in GFR: −4.4 versus −5.2 mL/min/1.73 m2/year; P = 0.01 |
| Lewis 2001 [20] (IDNT) | Population: (n = 1715) Type 2 diabetes mellitus, proteinuria >0.9 g/day), serum creatinine 1.0–3 mg/dL (88–265 μmol/L) in women or 1.2–3 mg/dL (106–265 μmol/L) in men; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Irbesartan (n = 579) versus placebo (n = 569) versus amlodipine (n = 567) Baseline kidney function: Serum creatinine irbesartan 1.67 ± 0.53 mg/dL versus placebo 1.69 ± 0.57 mg/dL versus amlodipine 1.65 ± 0.61 mg/dL Follow-up: Mean 2.6 years | Primary: Composite of doubling of serum creatinine or ESKD (need for dialysis or transplantation or serum creatinine ≥6 mg/dL [≥560 μmol/L] or death (total 644 events/38%) Irbesartan 189 (33%) versus placebo 222 (39%); P = 0.02 Irbesartan 189 (33%) versus amlodipine 233 (41%); P = 0.006 Secondary: Doubling of serum creatinine: Irbesartan 98 (17%) versus placebo 135 (24%); P = 0.003 and versus amlodipine 144 (25%); P < 0.001 ESKD: Irbesartan 82 (14%) versus placebo 101 (18%); P = 0.07 and versus amlodipine 104 (18%); P = 0.07 Death: Irbesartan 87 (15%) versus placebo 93 (16%); P = 0.57 and versus amlodipine 83 (15%); P = 0.8 Mean change in creatinine clearance: Irbesartan −5.5 ± 0.36 versus placebo −6.5 ± 0.37 versus amlodipine −6.8 ± 0.37 mL/min/1.73 m2/year |
| Wright 2002 [42] (AASK) | Population: (n = 1094) Hypertensive renal disease, GFR 20–65 mL/min/1.73 m2; excluded: diabetes mellitus, proteinuria > Intervention 1: Low BP (n = 540) versus usual BP (n = 554) Intervention 2: Ramipril (n = 436) versus amlodipine (n = 217) versus metoprolol (n = 441) Baseline kidney function: GFR mean(SE) Low BP 46(12.9) mL/min/1.73 m2 versus usual BP 45.3(13.2) mL/min/1.73 m2 Ramipril 45.4(12.8) mL/min/1.73 m2 versus amlodipine 45.8(12.9) mL/min/1.73 m2 versus metoprolol 45.8(13.4) mL/min/1.73 m2 Follow-up: median 3.8 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope, results expressed as mean (SE) Intervention 1: Low BP −2.21 (0.17) versus usual BP −1.95 (0.17) mL/min/1.73 m2/year; P = 0.24 Intervention 2: Ramipril −1.81 (0.17) versus metoprolol −2.42 (0.17) mL/min/1.73 m2/year; P = 0.07 Amlodipine −1.60 (0.34) versus metoprolol −2.68 (0.20) mL/min/1.73 m2/year; P = 0.0.04 Secondary: (overall results described): Composite of ≥50% reduction in GFR or ≥25 mL/min/1.73 m2 from baseline or ESRD (dialysis or transplantation) or death: total 340 (31%) events (170 GFR events, 84 ESKD, 77 deaths). GFR event or ESKD: 263 (24%) events (179 GFR events, 84 ESKD). ESKD or death: 251 (23%) events (171 ESKD, 80 deaths) ESKD alone: 171 (16%) (death censored) |
| Barnett 2004 [51] | Population: (n = 250) Type 2 diabetes mellitus, urinary albumin excretion rate 11–999 μg/min, serum creatinine <1.6 mg/dL (<141 μmol/L), GFR > 70 mL/min/1.73 m2 Interventions: Telmisartan (n = 120) versus enalapril (n = 130) Baseline kidney function: GFR 91.4 ± 21.5 mL/min/1.73 m2 versus 94.3 ± 22.1 mL/min/1.73 m2 Follow-up: 5 years | Primary: Change (between the baseline value and the last available value during the 5-year treatment period) in GFR (plasma clearance of iohexol) Telmisartan −17.9 versus enalapril −14.9 mL/min/1.73 m2 over 5 years Secondary: Annual change in GFR (overall results): −7.6, −5.6 and −3.6 mL/min/1.73 m2 at 1, 2 and 3 years, respectively. Data according to the randomization groups was not reported. Death: 6 (5%) versus 6 (5%) ESKD (pre-specified end point): Data not reported |
| Ruggenenti 2005 [52] (REIN-2) | Population: (n = 338) Non-diabetic nephropathy, proteinuria ≥1 g/day, creatinine clearance <45 mL/min/1.73 m2 if proteinuria 1–3 g/day or creatinine clearance <70 mL/min/1.73 m2 if proteinuria ≥3 g/day; excluded: type 1 diabetes mellitus Interventions: Intensified BP control (n = 169) versus conventional BP control (n = 169) Baseline kidney function: GFR (mean ± SD) Intensified BP control (35.9 ± 18.6 mL/min/1.73 m2) versus conventional BP control (34.1 ± 18.1 mL/min/1.73 m2) Follow-up: median 19 months | Primary: Time to ESKD ESKD: intensified BP control 38 (23%) versus conventional BP control 34 (20%); P = 0.99 Secondary: GFR slope (plasma clearance of non-radioactive iohexol), expressed as median (IQR) rate of GFR change: −0.22 (0.06 to 0.55) versus −0.24 (0.0001 to 0.56) mL/min/1.73 m2/month; P = 0.62 Creatinine clearance slope, expressed as median (IQR) rate of GFR change: −0.26 (0.03 to 0.53) versus −0.25 (0.0001 to 0.75) mL/min/1.73 m2/month; P = .59 Death: 3 (2%) versus 2 (1%) |
| Hou 2006 [19] | Population: (n = 328) Non-diabetic renal disease, serum creatinine 1.5–5 mg/dL (133–442 μmol/L), creatinine clearance 20–70 mL/min/1.73 m2, proteinuria >0.3 g/day Interventions: Group 1 (serum creatinine 1.5–3 mg/dL): Benazepril in all participants (n = 104); Group 2 (serum creatinine 3.1–5 mg/dL): Benazepril (n = 112) versus placebo (n = 112) Baseline kidney function: Calculated GFR: Group 1: 37.1 ± 6.3 mL/min/1.73 m2, Group 2: Benazepril 26.3 ± 5.3 mL/min/1.73 m2 versus 25.8 ± 5.3 mL/min/1.73 m2 Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESKD or death Group 1: 22 (22%) Group 2 (total events 109/51%): Benazepril 44 (41%) versus 65 (60%); P = 0.004 Secondary: (Group 2 results only): Reciprocal of serum creatinine (median slope): −0.09 dL/mg/year versus −0.11 dL/mg/year; P = 0.02 Median change in MDRD eGFR: −6.8 versus −8.8 mL/min/ 1.73 m2/year; 0.006 Data on individual events of doubling of serum creatinine and ESKD were not reported. Death: 1 (1%) versus 0 (0%) |
| Pfeffer 2009 [53] (TREAT) | Population: (n = 4038) type 2 diabetes mellitus, MDRD eGFR 20–60 mL/min/1.73 m2, Hb ≤ 11 g/dL, transferrin saturation ≥15% Intervention: Darbepoetin alfa (n = 2012) versus placebo (n = 2026) Baseline kidney function: median (IQR) Darbepoetin alfa 34(27–43) mL/min/1.73 m2 versus placebo 33(26–42) mL/min/1.73 m2 Follow-up: median 29.1 months | Primary: Death or ESKD (dialysis >30 days or transplantation or death within 30 days of starting dialysis or refusing renal replacement therapy despite of physician recommendation) total events (1276/32%) Darbepoetin alfa 652 (32%) versus placebo 618 (31%); P = 0.29 Secondary: ESKD: 338 (17%) versus 330 (16%); P = 0.83 Death: 412 (21%) versus 395 (20%); P = 0.48 Fatal or non-fatal stroke: 101 (5%) versus 53 (2.6%); P < 0.001 |
| Fried 2013 [31] (VA NEPHRON-D) | Population: (n = 1448) type 2 diabetes mellitus, albuminuria ≥300 mg/g, MDRD eGFR 30–89.9 mL/min/1.73 m2; excluded: non-diabetic renal disease Intervention: Combination therapy ACE inhibitor and ARB (n = 724) versus ARB monotherapy (n = 724) Baseline kidney function: mean ± SD eGFR Combination therapy 53.6 ± 15.5 mL/min/1.73 m2 versus ARB monotherapy 53.7 ± 16.2 mL/min/1.73 m2 Follow-up: Median 2.2 year | Primary: Composite of absolute reduction in eGFR ≥ 30 mL/min/1.73 m2 if baseline eGFR ≥ 60 mL/min/1.73 m2 or ≥50% reduction in eGFR if baseline eGFR < 60 mL/min/1.73 m2 or ESRD (dialysis or eGFR < 15 mL/min/1.73 m2) or death- total events (284/20%) Combination therapy 132 (18%) [59 GFR events, 18 ESKD, 55 deaths] versus ARB monotherapy 152 (21%) [78 GFR events, 23 ESKD, 51 deaths]; P = 0.3 Secondary: GFR event or ESKD: 77 (11%) versus 101 (14%); P = 0.1 ESKD: 27 (4%) versus 43 (6%); P = 0.07 Death: 63 (9%) versus 60 (8%); P = 0.75 eGFR slope: −2.7 versus −2.9 mL/min/1.73 m2/year; P = 0.17 |
| de Zeeuw 2013 [30] (BEACON) | Population: (n = 2185) type 2 diabetes mellitus, eGFR 15–29 mL/min/1.73 m2 Intervention: Bardoxolone methyl (n = 1088) versus placebo (n = 1097) Baseline kidney function: mean ± SD eGFR Bardoxolone methyl 22.2 ± 4.3 mL/min/1.73 m2 versus placebo 22.5 ± 4.6 mL/min/1.73 m2 Follow-up: 9 months | Primary: ESKD (dialysis >12 weeks or transplantation) or death from cardiovascular causes- total events (138/6%) Bardoxolone methyl 69 (6%) versus placebo 69 (6%); P = 0.92 Secondary: ESKD: 43 (4%) versus 51 (5%); P = 0.35 Death from cardiovascular causes: 27 (2%) versus 19 (2%); P = 0.23 Death from any cause: 44 (4%) versus 31 (3%); P = 0.1 Hospitalization for heart failure or death from heart failure: 96 (9%) versus 55 (5%); P < 0.001 Composite outcome of non-fatal myocardial infarction, non-fatal stroke, hospitalization for heart failure or death from cardiovascular causes: 139 (13%) versus 86 (8%); P < 0.001 Absolute change in eGFR, expressed as mean (95% CI): +5.5 (5.2 to 5.9) mL/min/1.73 m2 versus −0.9 (−1.2 to −0.5) mL/min/1.73 m2; P < 0.001 |
| Haynes 2014 [54] (SHARP) | Population: (n = 6245) Serum creatinine ≥1.7 mg/dL (≥150 μmol/L) in men or ≥1.5 mg/dL (≥130 μmol/L) in women Intervention: Simvastatin plus ezetimibe (n = 3116) versus placebo (n = 3129) Baseline kidney function: Mean ± SD MDRD eGFR Simvastatin plus ezetimibe 26.6 ± 12.9 mL/min/1.73 m2 versus placebo 26.6 ± 13.1 mL/min/1.73 m2 Follow-up: 4.8 years | Pre-specified subsidiary renal outcome: ESKD (dialysis or transplantation)- total events (2141/34%) Simvastatin plus ezetimibe 1057 (34%) versus placebo 1084 (35%); P = 0.41 Tertiary: ESKD or death: 1477 (47%) versus 1513 (48%); P = 0.34 ESKD or doubling of serum creatinine: 1189 (38%) versus 1257 (40%); P = 0.09 Death: 635 (20%) versus 635 (20%); P = 0.93 Doubling of serum creatinine: 370 (12%) versus 415 (13%); P = 0.11 eGFR slope, expressed as mean (SE): −1.66 (0.07) versus −1.83 (0.07) mL/min/1.73 m2/year; P = 0.1 |
| Torres (2014) [55] (HALT-PKD Study B) | Population: (n = 486) Autosomal dominant polycystic kidney disease, age 18–64 years, eGFR 25 to 60 mL/min/1.73 m2 Intervention: Lisinopril plus telmisartan (n = 244) versus lisinopril plus placebo (n = 242) Baseline kidney function: Mean ± SD CKD-EPI eGFR Lisinopril plus telmisartan 48.5 ± 11.5 mL/min/1.73 m2 versus lisinopril plus placebo 47.9 ± 12.2 mL/min/1.73 m2 Follow-up: 5.2 years | Primary: Composite of death or ESKD or a 50% decline in baseline eGFR (total events 231/48%) Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) Secondary: 50% decline in baseline eGFR: total events 182 (37%), Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) ESKD: total events 112 (23%), Lisinopril plus telmisartan 46 (19%) versus lisinopril plus placebo 66 (27%) Death: total events 9 (2%), Lisinopril plus telmisartan 4 (2%) versus lisinopril plus placebo 5 (2%) Rate of change in eGFR, expressed as mean (95% CI): 3.91 (−3.65 to −4.17) mL/min/1.73 m2/year versus −3.87 (−3.61 to −4.14) mL/min/1.73 m2/year |
| Trial | Study population, interventions and follow-up | Kidney outcome measures |
|---|---|---|
| Locatelli 1991 [32] | Population: (n = 456) Creatinine clearance <60 mL/min, serum creatinine 133–619 μmol/L; excluded: diabetes and nephritic syndrome Interventions: Low-protein diet (n = 226) versus normal protein diet (n = 230) Baseline kidney function: Data according to the randomization groups or collective data were not reported. Follow-up: 2 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total 69 events/15%) Low-protein diet (27/12%) versus normal protein diet (42/18%); P = 0.059 Secondary: Data on individual events was not reported. Death: 2 (1%) versus 3 (1%) |
| Lewis 1993 [33] | Population: (n = 409) Type 1 diabetes mellitus, proteinuria ≥0.5 g/day, serum creatinine ≤2.5 g/dL (221 μmol/L) Interventions: Captopril (n = 207) versus placebo (n = 202) Baseline kidney function: 24-h creatinine clearance 84 ± 46 mL/min versus 79 ± 35 mL/min Follow-up: Median 3 years (range 1.8–4.8 years) | Primary: Doubling of serum creatinine (total 68/17%) Captopril (25/12%) versus placebo (43/21%); P = 0.007 Secondary: Mean rate of decline in creatinine clearance: 11 ± 21%/year versus 17 ± 20%/year; P = 0.03 Death or ESKD: 23 (11%) versus 42 (21%); P = 0.006 Death: 8 (4%) versus 14 (7%) ESKD (dialysis or transplantation): 20 (10%) versus 31 (15%) |
| Klahr 1994 [41] MDRD Study 1 | Population: (n = 585) GFR 25–55 mL/min/1.73 m2 Intervention 1: Usual protein diet versus low-protein diet Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 38.6 ± 8.9 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Usual protein diet 12.1 (10.5–13.8) mL/min/3 years versus low-protein diet 10.9 (9.2–12.5) mL/min/3 years; P = 0.3 Intervention 2: Usual BP 12.3 (10.6–14) mL/min/3 years versus low BP 10.7 (9.1–12.4) mL/min/3 years; P = 0.18 Secondary: (Data according to the randomization groups was not reported): Death: 15 (3%) ESKD: 12 (2%) Rapid decline in GFR (>50 reduction in GFR if baseline GFR ≤ 40 mL/min/1.73 m2 or ≥20 mL/min/1.73 m2 reduction in GFR if baseline GFR > 40 mL/min/1.73 m2): 60 (10%) |
| Klahr 1994 [41] MDRD Study 2 | Population: (n = 255) GFR 13–24 mL/min/1.73 m2 Intervention 1: Low-protein diet versus very low-protein diet with keto acid-amino acid supplement Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 18.5 ± 3.4 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Low-protein diet 4.4 (3.7–5.1) mL/min/year versus very low-protein diet 3.6 (2.9–4.2) mL/min/year; P = 0.07 Intervention 2: Usual BP 4.2 (3.6–4.9) mL/min/year versus low BP 3.7 (3.1–4.3) mL/min/year; P = 0.28 Secondary: (Data according to the randomization groups was not reported): Death: 15 (6%) ESKD: 94 (37%) |
| Toto 1995 [34] | Population: (n = 77) Long-standing hypertension, GFR ≤ 70 mL/min/1.73 m2), normal urine sediment, excluded: diabetes Interventions: Strict BP control (n = 42) versus conventional BP control (n = 35) Baseline kidney function: GFR 34.6 ± 2.3 versus 41.9 ± 3.15 mL/min/1.73 m2 Follow-up: Mean 40.5 ± 1.8 months versus 42.2 ± 2.1 months | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope −0.31 ± 0.45 versus −0.05 ± 0.5 mL/min/1.73 m2/year; P > 0.25 Secondary: 50% reduction in GFR (or doubling of serum creatinine) or ESRD or death: 12 (29%) versus 7(20%); P > 0.25 50% reduction in GFR (or doubling of serum creatinine): 4 (10%) versus 5(14%) ESKD: 7 (17%) versus 2 (6%) Death: 1 (2%) versus 0 (0%) |
| Maschio 1996 [35] | Population: (n = 583) serum creatinine 1.5–4 mg/dL (133–354 μmol/L), 24-h creatinine clearance 30–60 mL/min Interventions: Benazepril (n = 300) versus placebo (n = 283) Baseline kidney function: Creatinine clearance 42.9 ± 11.6 mL/min versus 42.3 ± 10.6 mL/min Follow-up: 3 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total events 88/15%) Benazepril 31 (10%) versus placebo 57 (20%); P < 0.001 Secondary: Doubling of serum creatinine: total events 86 (15%) Data according to the randomization groups was not reported. Dialysis: total events 2 (<1%) Data according to the randomization groups was not reported. Death: 8 (3%) versus 1 (<1%) |
| The GISEN Group 1997 [50] REIN Study Stratum 2 | Population: (n = 166) Non-diabetic nephropathy, creatinine clearance 20–70 mL/min/1.73 m2, proteinuria ≥1 g/day (≥3 g/day for stratum 2); excluded: Type 1 diabetes mellitus Interventions: Ramipril (n = 78) versus placebo (n = 88) Baseline kidney function: GFR 40.2 ± 19 mL/min/1.73 m2 versus 37.4 ± 17.5 mL/min/1.73 m2 Follow-up: Mean 16 months | Primary: Change in GFR (plasma clearance of non-radioactive iohexol), results expressed as mean (SE): 0.53 (0.08) versus 0.88 (0.13) mL/min/month; P = 0.03 Secondary: Doubling of serum creatinine or ESKD: 18 (23%) versus 40 (45%); P = 0.02 ESKD: 17 (22%) versus 29 (33%); P = 0.2 Death: 2 (3%) versus 1 (1%) |
| Kuriyama 1997 [36] | Population: (n = 73) serum creatinine 2–4 mg/dL, hematocrit <30% Interventions: No EPO (n = 31) versus EPO (n = 42) Baseline kidney function: Creatinine clearance 17.1 ± 7.2 mL/min versus 19.1 ± 7.2 mL/min Follow-up: 3 years | Primary: Doubling of serum creatinine (total 48 events/66%) No EPO group 26 (84%) versus EPO group 22 (52%); P = 0.0003 Secondary: Dialysis: 20 (65%) versus 14 (33%) Death: 2 (%) versus 1 (%) |
| Brenner 2001 [10] (RENAAL) | Population: (n = 1513) Type 2 diabetes mellitus, albuminuria >300 mg/g (or >0.5 g/day), serum creatinine 1.3–3 mg/dL (115–265 μmol/L) or 1.5–3 mg/dL (133–265 μmol/L) if weight >60 kg; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Losartan (n = 751) versus placebo (n = 762) Baseline kidney function: serum creatinine 1.9 ± 0.5 mg/dL versus 1.9 ± 0.5 mg/dL Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESRD or death (total 686 events/45%) Losartan 327 (44%) versus placebo 359 (47%); P = 0.02 Secondary: Doubling of serum creatinine: 162 (22%) versus 198 (26%); P = 0.006 ESKD: 147 (20%) versus 194(26%); P = 0.002 Death: 158 (21%) versus 155 (20%); P = 0.88 ESKD or death: 255 (12%) versus 300 (39%); P = 0.01 Doubling of serum creatinine or ESRD: 226 (30%) versus 263 (35%); P = 0.01 Reciprocal of serum creatinine (median slope): −0.056 dL/mg/year versus −0.069 dL/mg/year; P = 0.01 Median change in GFR: −4.4 versus −5.2 mL/min/1.73 m2/year; P = 0.01 |
| Lewis 2001 [20] (IDNT) | Population: (n = 1715) Type 2 diabetes mellitus, proteinuria >0.9 g/day), serum creatinine 1.0–3 mg/dL (88–265 μmol/L) in women or 1.2–3 mg/dL (106–265 μmol/L) in men; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Irbesartan (n = 579) versus placebo (n = 569) versus amlodipine (n = 567) Baseline kidney function: Serum creatinine irbesartan 1.67 ± 0.53 mg/dL versus placebo 1.69 ± 0.57 mg/dL versus amlodipine 1.65 ± 0.61 mg/dL Follow-up: Mean 2.6 years | Primary: Composite of doubling of serum creatinine or ESKD (need for dialysis or transplantation or serum creatinine ≥6 mg/dL [≥560 μmol/L] or death (total 644 events/38%) Irbesartan 189 (33%) versus placebo 222 (39%); P = 0.02 Irbesartan 189 (33%) versus amlodipine 233 (41%); P = 0.006 Secondary: Doubling of serum creatinine: Irbesartan 98 (17%) versus placebo 135 (24%); P = 0.003 and versus amlodipine 144 (25%); P < 0.001 ESKD: Irbesartan 82 (14%) versus placebo 101 (18%); P = 0.07 and versus amlodipine 104 (18%); P = 0.07 Death: Irbesartan 87 (15%) versus placebo 93 (16%); P = 0.57 and versus amlodipine 83 (15%); P = 0.8 Mean change in creatinine clearance: Irbesartan −5.5 ± 0.36 versus placebo −6.5 ± 0.37 versus amlodipine −6.8 ± 0.37 mL/min/1.73 m2/year |
| Wright 2002 [42] (AASK) | Population: (n = 1094) Hypertensive renal disease, GFR 20–65 mL/min/1.73 m2; excluded: diabetes mellitus, proteinuria > Intervention 1: Low BP (n = 540) versus usual BP (n = 554) Intervention 2: Ramipril (n = 436) versus amlodipine (n = 217) versus metoprolol (n = 441) Baseline kidney function: GFR mean(SE) Low BP 46(12.9) mL/min/1.73 m2 versus usual BP 45.3(13.2) mL/min/1.73 m2 Ramipril 45.4(12.8) mL/min/1.73 m2 versus amlodipine 45.8(12.9) mL/min/1.73 m2 versus metoprolol 45.8(13.4) mL/min/1.73 m2 Follow-up: median 3.8 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope, results expressed as mean (SE) Intervention 1: Low BP −2.21 (0.17) versus usual BP −1.95 (0.17) mL/min/1.73 m2/year; P = 0.24 Intervention 2: Ramipril −1.81 (0.17) versus metoprolol −2.42 (0.17) mL/min/1.73 m2/year; P = 0.07 Amlodipine −1.60 (0.34) versus metoprolol −2.68 (0.20) mL/min/1.73 m2/year; P = 0.0.04 Secondary: (overall results described): Composite of ≥50% reduction in GFR or ≥25 mL/min/1.73 m2 from baseline or ESRD (dialysis or transplantation) or death: total 340 (31%) events (170 GFR events, 84 ESKD, 77 deaths). GFR event or ESKD: 263 (24%) events (179 GFR events, 84 ESKD). ESKD or death: 251 (23%) events (171 ESKD, 80 deaths) ESKD alone: 171 (16%) (death censored) |
| Barnett 2004 [51] | Population: (n = 250) Type 2 diabetes mellitus, urinary albumin excretion rate 11–999 μg/min, serum creatinine <1.6 mg/dL (<141 μmol/L), GFR > 70 mL/min/1.73 m2 Interventions: Telmisartan (n = 120) versus enalapril (n = 130) Baseline kidney function: GFR 91.4 ± 21.5 mL/min/1.73 m2 versus 94.3 ± 22.1 mL/min/1.73 m2 Follow-up: 5 years | Primary: Change (between the baseline value and the last available value during the 5-year treatment period) in GFR (plasma clearance of iohexol) Telmisartan −17.9 versus enalapril −14.9 mL/min/1.73 m2 over 5 years Secondary: Annual change in GFR (overall results): −7.6, −5.6 and −3.6 mL/min/1.73 m2 at 1, 2 and 3 years, respectively. Data according to the randomization groups was not reported. Death: 6 (5%) versus 6 (5%) ESKD (pre-specified end point): Data not reported |
| Ruggenenti 2005 [52] (REIN-2) | Population: (n = 338) Non-diabetic nephropathy, proteinuria ≥1 g/day, creatinine clearance <45 mL/min/1.73 m2 if proteinuria 1–3 g/day or creatinine clearance <70 mL/min/1.73 m2 if proteinuria ≥3 g/day; excluded: type 1 diabetes mellitus Interventions: Intensified BP control (n = 169) versus conventional BP control (n = 169) Baseline kidney function: GFR (mean ± SD) Intensified BP control (35.9 ± 18.6 mL/min/1.73 m2) versus conventional BP control (34.1 ± 18.1 mL/min/1.73 m2) Follow-up: median 19 months | Primary: Time to ESKD ESKD: intensified BP control 38 (23%) versus conventional BP control 34 (20%); P = 0.99 Secondary: GFR slope (plasma clearance of non-radioactive iohexol), expressed as median (IQR) rate of GFR change: −0.22 (0.06 to 0.55) versus −0.24 (0.0001 to 0.56) mL/min/1.73 m2/month; P = 0.62 Creatinine clearance slope, expressed as median (IQR) rate of GFR change: −0.26 (0.03 to 0.53) versus −0.25 (0.0001 to 0.75) mL/min/1.73 m2/month; P = .59 Death: 3 (2%) versus 2 (1%) |
| Hou 2006 [19] | Population: (n = 328) Non-diabetic renal disease, serum creatinine 1.5–5 mg/dL (133–442 μmol/L), creatinine clearance 20–70 mL/min/1.73 m2, proteinuria >0.3 g/day Interventions: Group 1 (serum creatinine 1.5–3 mg/dL): Benazepril in all participants (n = 104); Group 2 (serum creatinine 3.1–5 mg/dL): Benazepril (n = 112) versus placebo (n = 112) Baseline kidney function: Calculated GFR: Group 1: 37.1 ± 6.3 mL/min/1.73 m2, Group 2: Benazepril 26.3 ± 5.3 mL/min/1.73 m2 versus 25.8 ± 5.3 mL/min/1.73 m2 Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESKD or death Group 1: 22 (22%) Group 2 (total events 109/51%): Benazepril 44 (41%) versus 65 (60%); P = 0.004 Secondary: (Group 2 results only): Reciprocal of serum creatinine (median slope): −0.09 dL/mg/year versus −0.11 dL/mg/year; P = 0.02 Median change in MDRD eGFR: −6.8 versus −8.8 mL/min/ 1.73 m2/year; 0.006 Data on individual events of doubling of serum creatinine and ESKD were not reported. Death: 1 (1%) versus 0 (0%) |
| Pfeffer 2009 [53] (TREAT) | Population: (n = 4038) type 2 diabetes mellitus, MDRD eGFR 20–60 mL/min/1.73 m2, Hb ≤ 11 g/dL, transferrin saturation ≥15% Intervention: Darbepoetin alfa (n = 2012) versus placebo (n = 2026) Baseline kidney function: median (IQR) Darbepoetin alfa 34(27–43) mL/min/1.73 m2 versus placebo 33(26–42) mL/min/1.73 m2 Follow-up: median 29.1 months | Primary: Death or ESKD (dialysis >30 days or transplantation or death within 30 days of starting dialysis or refusing renal replacement therapy despite of physician recommendation) total events (1276/32%) Darbepoetin alfa 652 (32%) versus placebo 618 (31%); P = 0.29 Secondary: ESKD: 338 (17%) versus 330 (16%); P = 0.83 Death: 412 (21%) versus 395 (20%); P = 0.48 Fatal or non-fatal stroke: 101 (5%) versus 53 (2.6%); P < 0.001 |
| Fried 2013 [31] (VA NEPHRON-D) | Population: (n = 1448) type 2 diabetes mellitus, albuminuria ≥300 mg/g, MDRD eGFR 30–89.9 mL/min/1.73 m2; excluded: non-diabetic renal disease Intervention: Combination therapy ACE inhibitor and ARB (n = 724) versus ARB monotherapy (n = 724) Baseline kidney function: mean ± SD eGFR Combination therapy 53.6 ± 15.5 mL/min/1.73 m2 versus ARB monotherapy 53.7 ± 16.2 mL/min/1.73 m2 Follow-up: Median 2.2 year | Primary: Composite of absolute reduction in eGFR ≥ 30 mL/min/1.73 m2 if baseline eGFR ≥ 60 mL/min/1.73 m2 or ≥50% reduction in eGFR if baseline eGFR < 60 mL/min/1.73 m2 or ESRD (dialysis or eGFR < 15 mL/min/1.73 m2) or death- total events (284/20%) Combination therapy 132 (18%) [59 GFR events, 18 ESKD, 55 deaths] versus ARB monotherapy 152 (21%) [78 GFR events, 23 ESKD, 51 deaths]; P = 0.3 Secondary: GFR event or ESKD: 77 (11%) versus 101 (14%); P = 0.1 ESKD: 27 (4%) versus 43 (6%); P = 0.07 Death: 63 (9%) versus 60 (8%); P = 0.75 eGFR slope: −2.7 versus −2.9 mL/min/1.73 m2/year; P = 0.17 |
| de Zeeuw 2013 [30] (BEACON) | Population: (n = 2185) type 2 diabetes mellitus, eGFR 15–29 mL/min/1.73 m2 Intervention: Bardoxolone methyl (n = 1088) versus placebo (n = 1097) Baseline kidney function: mean ± SD eGFR Bardoxolone methyl 22.2 ± 4.3 mL/min/1.73 m2 versus placebo 22.5 ± 4.6 mL/min/1.73 m2 Follow-up: 9 months | Primary: ESKD (dialysis >12 weeks or transplantation) or death from cardiovascular causes- total events (138/6%) Bardoxolone methyl 69 (6%) versus placebo 69 (6%); P = 0.92 Secondary: ESKD: 43 (4%) versus 51 (5%); P = 0.35 Death from cardiovascular causes: 27 (2%) versus 19 (2%); P = 0.23 Death from any cause: 44 (4%) versus 31 (3%); P = 0.1 Hospitalization for heart failure or death from heart failure: 96 (9%) versus 55 (5%); P < 0.001 Composite outcome of non-fatal myocardial infarction, non-fatal stroke, hospitalization for heart failure or death from cardiovascular causes: 139 (13%) versus 86 (8%); P < 0.001 Absolute change in eGFR, expressed as mean (95% CI): +5.5 (5.2 to 5.9) mL/min/1.73 m2 versus −0.9 (−1.2 to −0.5) mL/min/1.73 m2; P < 0.001 |
| Haynes 2014 [54] (SHARP) | Population: (n = 6245) Serum creatinine ≥1.7 mg/dL (≥150 μmol/L) in men or ≥1.5 mg/dL (≥130 μmol/L) in women Intervention: Simvastatin plus ezetimibe (n = 3116) versus placebo (n = 3129) Baseline kidney function: Mean ± SD MDRD eGFR Simvastatin plus ezetimibe 26.6 ± 12.9 mL/min/1.73 m2 versus placebo 26.6 ± 13.1 mL/min/1.73 m2 Follow-up: 4.8 years | Pre-specified subsidiary renal outcome: ESKD (dialysis or transplantation)- total events (2141/34%) Simvastatin plus ezetimibe 1057 (34%) versus placebo 1084 (35%); P = 0.41 Tertiary: ESKD or death: 1477 (47%) versus 1513 (48%); P = 0.34 ESKD or doubling of serum creatinine: 1189 (38%) versus 1257 (40%); P = 0.09 Death: 635 (20%) versus 635 (20%); P = 0.93 Doubling of serum creatinine: 370 (12%) versus 415 (13%); P = 0.11 eGFR slope, expressed as mean (SE): −1.66 (0.07) versus −1.83 (0.07) mL/min/1.73 m2/year; P = 0.1 |
| Torres (2014) [55] (HALT-PKD Study B) | Population: (n = 486) Autosomal dominant polycystic kidney disease, age 18–64 years, eGFR 25 to 60 mL/min/1.73 m2 Intervention: Lisinopril plus telmisartan (n = 244) versus lisinopril plus placebo (n = 242) Baseline kidney function: Mean ± SD CKD-EPI eGFR Lisinopril plus telmisartan 48.5 ± 11.5 mL/min/1.73 m2 versus lisinopril plus placebo 47.9 ± 12.2 mL/min/1.73 m2 Follow-up: 5.2 years | Primary: Composite of death or ESKD or a 50% decline in baseline eGFR (total events 231/48%) Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) Secondary: 50% decline in baseline eGFR: total events 182 (37%), Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) ESKD: total events 112 (23%), Lisinopril plus telmisartan 46 (19%) versus lisinopril plus placebo 66 (27%) Death: total events 9 (2%), Lisinopril plus telmisartan 4 (2%) versus lisinopril plus placebo 5 (2%) Rate of change in eGFR, expressed as mean (95% CI): 3.91 (−3.65 to −4.17) mL/min/1.73 m2/year versus −3.87 (−3.61 to −4.14) mL/min/1.73 m2/year |
AASK, (the African American Study of Kidney Disease and Hypertension); AASK, African American Study of Kidney Disease and Hypertension; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; BEACON, Bardoxolone Methyl Evaluation in Patients with Chronic Kidney Disease and Type 2 Diabetes Mellitus, the Occurrence of Renal Events Study; BP, blood pressure; CI, confidence intervals; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration equation; eGFR, estimated glomerular filtration rate; EPO, erythropoietin; ESKD, end-stage kidney disease; GFR, glomerular filtration rate; HALT-PKD, The Halt Progression of Polycystic Kidney Disease trial; IDNT, Irbesartan Type II Diabetic Nephropathy Trial; IQR, inter-quartile range; MDRD, Modification of Diet in Renal Disease Study; RENAAL, Reduction of End points in NIDDM with the Angiotensin II Antagonist Losartan Study; REIN-2, Ramipril Efficacy in Nephropathy 2 Study; SD, standard deviation; SE, standard error; SHARP, Study of Heart and Renal Protection; TREAT, A Trial of Darbepoetin Alfa in Type 2 Diabetes and Chronic Kidney Disease; VA NEPHRON-D, The Veterans Affairs Nephropathy in Diabetes Study.
Description of trial end points in selected chronic kidney disease progression trials
| Trial | Study population, interventions and follow-up | Kidney outcome measures |
|---|---|---|
| Locatelli 1991 [32] | Population: (n = 456) Creatinine clearance <60 mL/min, serum creatinine 133–619 μmol/L; excluded: diabetes and nephritic syndrome Interventions: Low-protein diet (n = 226) versus normal protein diet (n = 230) Baseline kidney function: Data according to the randomization groups or collective data were not reported. Follow-up: 2 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total 69 events/15%) Low-protein diet (27/12%) versus normal protein diet (42/18%); P = 0.059 Secondary: Data on individual events was not reported. Death: 2 (1%) versus 3 (1%) |
| Lewis 1993 [33] | Population: (n = 409) Type 1 diabetes mellitus, proteinuria ≥0.5 g/day, serum creatinine ≤2.5 g/dL (221 μmol/L) Interventions: Captopril (n = 207) versus placebo (n = 202) Baseline kidney function: 24-h creatinine clearance 84 ± 46 mL/min versus 79 ± 35 mL/min Follow-up: Median 3 years (range 1.8–4.8 years) | Primary: Doubling of serum creatinine (total 68/17%) Captopril (25/12%) versus placebo (43/21%); P = 0.007 Secondary: Mean rate of decline in creatinine clearance: 11 ± 21%/year versus 17 ± 20%/year; P = 0.03 Death or ESKD: 23 (11%) versus 42 (21%); P = 0.006 Death: 8 (4%) versus 14 (7%) ESKD (dialysis or transplantation): 20 (10%) versus 31 (15%) |
| Klahr 1994 [41] MDRD Study 1 | Population: (n = 585) GFR 25–55 mL/min/1.73 m2 Intervention 1: Usual protein diet versus low-protein diet Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 38.6 ± 8.9 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Usual protein diet 12.1 (10.5–13.8) mL/min/3 years versus low-protein diet 10.9 (9.2–12.5) mL/min/3 years; P = 0.3 Intervention 2: Usual BP 12.3 (10.6–14) mL/min/3 years versus low BP 10.7 (9.1–12.4) mL/min/3 years; P = 0.18 Secondary: (Data according to the randomization groups was not reported): Death: 15 (3%) ESKD: 12 (2%) Rapid decline in GFR (>50 reduction in GFR if baseline GFR ≤ 40 mL/min/1.73 m2 or ≥20 mL/min/1.73 m2 reduction in GFR if baseline GFR > 40 mL/min/1.73 m2): 60 (10%) |
| Klahr 1994 [41] MDRD Study 2 | Population: (n = 255) GFR 13–24 mL/min/1.73 m2 Intervention 1: Low-protein diet versus very low-protein diet with keto acid-amino acid supplement Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 18.5 ± 3.4 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Low-protein diet 4.4 (3.7–5.1) mL/min/year versus very low-protein diet 3.6 (2.9–4.2) mL/min/year; P = 0.07 Intervention 2: Usual BP 4.2 (3.6–4.9) mL/min/year versus low BP 3.7 (3.1–4.3) mL/min/year; P = 0.28 Secondary: (Data according to the randomization groups was not reported): Death: 15 (6%) ESKD: 94 (37%) |
| Toto 1995 [34] | Population: (n = 77) Long-standing hypertension, GFR ≤ 70 mL/min/1.73 m2), normal urine sediment, excluded: diabetes Interventions: Strict BP control (n = 42) versus conventional BP control (n = 35) Baseline kidney function: GFR 34.6 ± 2.3 versus 41.9 ± 3.15 mL/min/1.73 m2 Follow-up: Mean 40.5 ± 1.8 months versus 42.2 ± 2.1 months | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope −0.31 ± 0.45 versus −0.05 ± 0.5 mL/min/1.73 m2/year; P > 0.25 Secondary: 50% reduction in GFR (or doubling of serum creatinine) or ESRD or death: 12 (29%) versus 7(20%); P > 0.25 50% reduction in GFR (or doubling of serum creatinine): 4 (10%) versus 5(14%) ESKD: 7 (17%) versus 2 (6%) Death: 1 (2%) versus 0 (0%) |
| Maschio 1996 [35] | Population: (n = 583) serum creatinine 1.5–4 mg/dL (133–354 μmol/L), 24-h creatinine clearance 30–60 mL/min Interventions: Benazepril (n = 300) versus placebo (n = 283) Baseline kidney function: Creatinine clearance 42.9 ± 11.6 mL/min versus 42.3 ± 10.6 mL/min Follow-up: 3 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total events 88/15%) Benazepril 31 (10%) versus placebo 57 (20%); P < 0.001 Secondary: Doubling of serum creatinine: total events 86 (15%) Data according to the randomization groups was not reported. Dialysis: total events 2 (<1%) Data according to the randomization groups was not reported. Death: 8 (3%) versus 1 (<1%) |
| The GISEN Group 1997 [50] REIN Study Stratum 2 | Population: (n = 166) Non-diabetic nephropathy, creatinine clearance 20–70 mL/min/1.73 m2, proteinuria ≥1 g/day (≥3 g/day for stratum 2); excluded: Type 1 diabetes mellitus Interventions: Ramipril (n = 78) versus placebo (n = 88) Baseline kidney function: GFR 40.2 ± 19 mL/min/1.73 m2 versus 37.4 ± 17.5 mL/min/1.73 m2 Follow-up: Mean 16 months | Primary: Change in GFR (plasma clearance of non-radioactive iohexol), results expressed as mean (SE): 0.53 (0.08) versus 0.88 (0.13) mL/min/month; P = 0.03 Secondary: Doubling of serum creatinine or ESKD: 18 (23%) versus 40 (45%); P = 0.02 ESKD: 17 (22%) versus 29 (33%); P = 0.2 Death: 2 (3%) versus 1 (1%) |
| Kuriyama 1997 [36] | Population: (n = 73) serum creatinine 2–4 mg/dL, hematocrit <30% Interventions: No EPO (n = 31) versus EPO (n = 42) Baseline kidney function: Creatinine clearance 17.1 ± 7.2 mL/min versus 19.1 ± 7.2 mL/min Follow-up: 3 years | Primary: Doubling of serum creatinine (total 48 events/66%) No EPO group 26 (84%) versus EPO group 22 (52%); P = 0.0003 Secondary: Dialysis: 20 (65%) versus 14 (33%) Death: 2 (%) versus 1 (%) |
| Brenner 2001 [10] (RENAAL) | Population: (n = 1513) Type 2 diabetes mellitus, albuminuria >300 mg/g (or >0.5 g/day), serum creatinine 1.3–3 mg/dL (115–265 μmol/L) or 1.5–3 mg/dL (133–265 μmol/L) if weight >60 kg; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Losartan (n = 751) versus placebo (n = 762) Baseline kidney function: serum creatinine 1.9 ± 0.5 mg/dL versus 1.9 ± 0.5 mg/dL Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESRD or death (total 686 events/45%) Losartan 327 (44%) versus placebo 359 (47%); P = 0.02 Secondary: Doubling of serum creatinine: 162 (22%) versus 198 (26%); P = 0.006 ESKD: 147 (20%) versus 194(26%); P = 0.002 Death: 158 (21%) versus 155 (20%); P = 0.88 ESKD or death: 255 (12%) versus 300 (39%); P = 0.01 Doubling of serum creatinine or ESRD: 226 (30%) versus 263 (35%); P = 0.01 Reciprocal of serum creatinine (median slope): −0.056 dL/mg/year versus −0.069 dL/mg/year; P = 0.01 Median change in GFR: −4.4 versus −5.2 mL/min/1.73 m2/year; P = 0.01 |
| Lewis 2001 [20] (IDNT) | Population: (n = 1715) Type 2 diabetes mellitus, proteinuria >0.9 g/day), serum creatinine 1.0–3 mg/dL (88–265 μmol/L) in women or 1.2–3 mg/dL (106–265 μmol/L) in men; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Irbesartan (n = 579) versus placebo (n = 569) versus amlodipine (n = 567) Baseline kidney function: Serum creatinine irbesartan 1.67 ± 0.53 mg/dL versus placebo 1.69 ± 0.57 mg/dL versus amlodipine 1.65 ± 0.61 mg/dL Follow-up: Mean 2.6 years | Primary: Composite of doubling of serum creatinine or ESKD (need for dialysis or transplantation or serum creatinine ≥6 mg/dL [≥560 μmol/L] or death (total 644 events/38%) Irbesartan 189 (33%) versus placebo 222 (39%); P = 0.02 Irbesartan 189 (33%) versus amlodipine 233 (41%); P = 0.006 Secondary: Doubling of serum creatinine: Irbesartan 98 (17%) versus placebo 135 (24%); P = 0.003 and versus amlodipine 144 (25%); P < 0.001 ESKD: Irbesartan 82 (14%) versus placebo 101 (18%); P = 0.07 and versus amlodipine 104 (18%); P = 0.07 Death: Irbesartan 87 (15%) versus placebo 93 (16%); P = 0.57 and versus amlodipine 83 (15%); P = 0.8 Mean change in creatinine clearance: Irbesartan −5.5 ± 0.36 versus placebo −6.5 ± 0.37 versus amlodipine −6.8 ± 0.37 mL/min/1.73 m2/year |
| Wright 2002 [42] (AASK) | Population: (n = 1094) Hypertensive renal disease, GFR 20–65 mL/min/1.73 m2; excluded: diabetes mellitus, proteinuria > Intervention 1: Low BP (n = 540) versus usual BP (n = 554) Intervention 2: Ramipril (n = 436) versus amlodipine (n = 217) versus metoprolol (n = 441) Baseline kidney function: GFR mean(SE) Low BP 46(12.9) mL/min/1.73 m2 versus usual BP 45.3(13.2) mL/min/1.73 m2 Ramipril 45.4(12.8) mL/min/1.73 m2 versus amlodipine 45.8(12.9) mL/min/1.73 m2 versus metoprolol 45.8(13.4) mL/min/1.73 m2 Follow-up: median 3.8 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope, results expressed as mean (SE) Intervention 1: Low BP −2.21 (0.17) versus usual BP −1.95 (0.17) mL/min/1.73 m2/year; P = 0.24 Intervention 2: Ramipril −1.81 (0.17) versus metoprolol −2.42 (0.17) mL/min/1.73 m2/year; P = 0.07 Amlodipine −1.60 (0.34) versus metoprolol −2.68 (0.20) mL/min/1.73 m2/year; P = 0.0.04 Secondary: (overall results described): Composite of ≥50% reduction in GFR or ≥25 mL/min/1.73 m2 from baseline or ESRD (dialysis or transplantation) or death: total 340 (31%) events (170 GFR events, 84 ESKD, 77 deaths). GFR event or ESKD: 263 (24%) events (179 GFR events, 84 ESKD). ESKD or death: 251 (23%) events (171 ESKD, 80 deaths) ESKD alone: 171 (16%) (death censored) |
| Barnett 2004 [51] | Population: (n = 250) Type 2 diabetes mellitus, urinary albumin excretion rate 11–999 μg/min, serum creatinine <1.6 mg/dL (<141 μmol/L), GFR > 70 mL/min/1.73 m2 Interventions: Telmisartan (n = 120) versus enalapril (n = 130) Baseline kidney function: GFR 91.4 ± 21.5 mL/min/1.73 m2 versus 94.3 ± 22.1 mL/min/1.73 m2 Follow-up: 5 years | Primary: Change (between the baseline value and the last available value during the 5-year treatment period) in GFR (plasma clearance of iohexol) Telmisartan −17.9 versus enalapril −14.9 mL/min/1.73 m2 over 5 years Secondary: Annual change in GFR (overall results): −7.6, −5.6 and −3.6 mL/min/1.73 m2 at 1, 2 and 3 years, respectively. Data according to the randomization groups was not reported. Death: 6 (5%) versus 6 (5%) ESKD (pre-specified end point): Data not reported |
| Ruggenenti 2005 [52] (REIN-2) | Population: (n = 338) Non-diabetic nephropathy, proteinuria ≥1 g/day, creatinine clearance <45 mL/min/1.73 m2 if proteinuria 1–3 g/day or creatinine clearance <70 mL/min/1.73 m2 if proteinuria ≥3 g/day; excluded: type 1 diabetes mellitus Interventions: Intensified BP control (n = 169) versus conventional BP control (n = 169) Baseline kidney function: GFR (mean ± SD) Intensified BP control (35.9 ± 18.6 mL/min/1.73 m2) versus conventional BP control (34.1 ± 18.1 mL/min/1.73 m2) Follow-up: median 19 months | Primary: Time to ESKD ESKD: intensified BP control 38 (23%) versus conventional BP control 34 (20%); P = 0.99 Secondary: GFR slope (plasma clearance of non-radioactive iohexol), expressed as median (IQR) rate of GFR change: −0.22 (0.06 to 0.55) versus −0.24 (0.0001 to 0.56) mL/min/1.73 m2/month; P = 0.62 Creatinine clearance slope, expressed as median (IQR) rate of GFR change: −0.26 (0.03 to 0.53) versus −0.25 (0.0001 to 0.75) mL/min/1.73 m2/month; P = .59 Death: 3 (2%) versus 2 (1%) |
| Hou 2006 [19] | Population: (n = 328) Non-diabetic renal disease, serum creatinine 1.5–5 mg/dL (133–442 μmol/L), creatinine clearance 20–70 mL/min/1.73 m2, proteinuria >0.3 g/day Interventions: Group 1 (serum creatinine 1.5–3 mg/dL): Benazepril in all participants (n = 104); Group 2 (serum creatinine 3.1–5 mg/dL): Benazepril (n = 112) versus placebo (n = 112) Baseline kidney function: Calculated GFR: Group 1: 37.1 ± 6.3 mL/min/1.73 m2, Group 2: Benazepril 26.3 ± 5.3 mL/min/1.73 m2 versus 25.8 ± 5.3 mL/min/1.73 m2 Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESKD or death Group 1: 22 (22%) Group 2 (total events 109/51%): Benazepril 44 (41%) versus 65 (60%); P = 0.004 Secondary: (Group 2 results only): Reciprocal of serum creatinine (median slope): −0.09 dL/mg/year versus −0.11 dL/mg/year; P = 0.02 Median change in MDRD eGFR: −6.8 versus −8.8 mL/min/ 1.73 m2/year; 0.006 Data on individual events of doubling of serum creatinine and ESKD were not reported. Death: 1 (1%) versus 0 (0%) |
| Pfeffer 2009 [53] (TREAT) | Population: (n = 4038) type 2 diabetes mellitus, MDRD eGFR 20–60 mL/min/1.73 m2, Hb ≤ 11 g/dL, transferrin saturation ≥15% Intervention: Darbepoetin alfa (n = 2012) versus placebo (n = 2026) Baseline kidney function: median (IQR) Darbepoetin alfa 34(27–43) mL/min/1.73 m2 versus placebo 33(26–42) mL/min/1.73 m2 Follow-up: median 29.1 months | Primary: Death or ESKD (dialysis >30 days or transplantation or death within 30 days of starting dialysis or refusing renal replacement therapy despite of physician recommendation) total events (1276/32%) Darbepoetin alfa 652 (32%) versus placebo 618 (31%); P = 0.29 Secondary: ESKD: 338 (17%) versus 330 (16%); P = 0.83 Death: 412 (21%) versus 395 (20%); P = 0.48 Fatal or non-fatal stroke: 101 (5%) versus 53 (2.6%); P < 0.001 |
| Fried 2013 [31] (VA NEPHRON-D) | Population: (n = 1448) type 2 diabetes mellitus, albuminuria ≥300 mg/g, MDRD eGFR 30–89.9 mL/min/1.73 m2; excluded: non-diabetic renal disease Intervention: Combination therapy ACE inhibitor and ARB (n = 724) versus ARB monotherapy (n = 724) Baseline kidney function: mean ± SD eGFR Combination therapy 53.6 ± 15.5 mL/min/1.73 m2 versus ARB monotherapy 53.7 ± 16.2 mL/min/1.73 m2 Follow-up: Median 2.2 year | Primary: Composite of absolute reduction in eGFR ≥ 30 mL/min/1.73 m2 if baseline eGFR ≥ 60 mL/min/1.73 m2 or ≥50% reduction in eGFR if baseline eGFR < 60 mL/min/1.73 m2 or ESRD (dialysis or eGFR < 15 mL/min/1.73 m2) or death- total events (284/20%) Combination therapy 132 (18%) [59 GFR events, 18 ESKD, 55 deaths] versus ARB monotherapy 152 (21%) [78 GFR events, 23 ESKD, 51 deaths]; P = 0.3 Secondary: GFR event or ESKD: 77 (11%) versus 101 (14%); P = 0.1 ESKD: 27 (4%) versus 43 (6%); P = 0.07 Death: 63 (9%) versus 60 (8%); P = 0.75 eGFR slope: −2.7 versus −2.9 mL/min/1.73 m2/year; P = 0.17 |
| de Zeeuw 2013 [30] (BEACON) | Population: (n = 2185) type 2 diabetes mellitus, eGFR 15–29 mL/min/1.73 m2 Intervention: Bardoxolone methyl (n = 1088) versus placebo (n = 1097) Baseline kidney function: mean ± SD eGFR Bardoxolone methyl 22.2 ± 4.3 mL/min/1.73 m2 versus placebo 22.5 ± 4.6 mL/min/1.73 m2 Follow-up: 9 months | Primary: ESKD (dialysis >12 weeks or transplantation) or death from cardiovascular causes- total events (138/6%) Bardoxolone methyl 69 (6%) versus placebo 69 (6%); P = 0.92 Secondary: ESKD: 43 (4%) versus 51 (5%); P = 0.35 Death from cardiovascular causes: 27 (2%) versus 19 (2%); P = 0.23 Death from any cause: 44 (4%) versus 31 (3%); P = 0.1 Hospitalization for heart failure or death from heart failure: 96 (9%) versus 55 (5%); P < 0.001 Composite outcome of non-fatal myocardial infarction, non-fatal stroke, hospitalization for heart failure or death from cardiovascular causes: 139 (13%) versus 86 (8%); P < 0.001 Absolute change in eGFR, expressed as mean (95% CI): +5.5 (5.2 to 5.9) mL/min/1.73 m2 versus −0.9 (−1.2 to −0.5) mL/min/1.73 m2; P < 0.001 |
| Haynes 2014 [54] (SHARP) | Population: (n = 6245) Serum creatinine ≥1.7 mg/dL (≥150 μmol/L) in men or ≥1.5 mg/dL (≥130 μmol/L) in women Intervention: Simvastatin plus ezetimibe (n = 3116) versus placebo (n = 3129) Baseline kidney function: Mean ± SD MDRD eGFR Simvastatin plus ezetimibe 26.6 ± 12.9 mL/min/1.73 m2 versus placebo 26.6 ± 13.1 mL/min/1.73 m2 Follow-up: 4.8 years | Pre-specified subsidiary renal outcome: ESKD (dialysis or transplantation)- total events (2141/34%) Simvastatin plus ezetimibe 1057 (34%) versus placebo 1084 (35%); P = 0.41 Tertiary: ESKD or death: 1477 (47%) versus 1513 (48%); P = 0.34 ESKD or doubling of serum creatinine: 1189 (38%) versus 1257 (40%); P = 0.09 Death: 635 (20%) versus 635 (20%); P = 0.93 Doubling of serum creatinine: 370 (12%) versus 415 (13%); P = 0.11 eGFR slope, expressed as mean (SE): −1.66 (0.07) versus −1.83 (0.07) mL/min/1.73 m2/year; P = 0.1 |
| Torres (2014) [55] (HALT-PKD Study B) | Population: (n = 486) Autosomal dominant polycystic kidney disease, age 18–64 years, eGFR 25 to 60 mL/min/1.73 m2 Intervention: Lisinopril plus telmisartan (n = 244) versus lisinopril plus placebo (n = 242) Baseline kidney function: Mean ± SD CKD-EPI eGFR Lisinopril plus telmisartan 48.5 ± 11.5 mL/min/1.73 m2 versus lisinopril plus placebo 47.9 ± 12.2 mL/min/1.73 m2 Follow-up: 5.2 years | Primary: Composite of death or ESKD or a 50% decline in baseline eGFR (total events 231/48%) Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) Secondary: 50% decline in baseline eGFR: total events 182 (37%), Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) ESKD: total events 112 (23%), Lisinopril plus telmisartan 46 (19%) versus lisinopril plus placebo 66 (27%) Death: total events 9 (2%), Lisinopril plus telmisartan 4 (2%) versus lisinopril plus placebo 5 (2%) Rate of change in eGFR, expressed as mean (95% CI): 3.91 (−3.65 to −4.17) mL/min/1.73 m2/year versus −3.87 (−3.61 to −4.14) mL/min/1.73 m2/year |
| Trial | Study population, interventions and follow-up | Kidney outcome measures |
|---|---|---|
| Locatelli 1991 [32] | Population: (n = 456) Creatinine clearance <60 mL/min, serum creatinine 133–619 μmol/L; excluded: diabetes and nephritic syndrome Interventions: Low-protein diet (n = 226) versus normal protein diet (n = 230) Baseline kidney function: Data according to the randomization groups or collective data were not reported. Follow-up: 2 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total 69 events/15%) Low-protein diet (27/12%) versus normal protein diet (42/18%); P = 0.059 Secondary: Data on individual events was not reported. Death: 2 (1%) versus 3 (1%) |
| Lewis 1993 [33] | Population: (n = 409) Type 1 diabetes mellitus, proteinuria ≥0.5 g/day, serum creatinine ≤2.5 g/dL (221 μmol/L) Interventions: Captopril (n = 207) versus placebo (n = 202) Baseline kidney function: 24-h creatinine clearance 84 ± 46 mL/min versus 79 ± 35 mL/min Follow-up: Median 3 years (range 1.8–4.8 years) | Primary: Doubling of serum creatinine (total 68/17%) Captopril (25/12%) versus placebo (43/21%); P = 0.007 Secondary: Mean rate of decline in creatinine clearance: 11 ± 21%/year versus 17 ± 20%/year; P = 0.03 Death or ESKD: 23 (11%) versus 42 (21%); P = 0.006 Death: 8 (4%) versus 14 (7%) ESKD (dialysis or transplantation): 20 (10%) versus 31 (15%) |
| Klahr 1994 [41] MDRD Study 1 | Population: (n = 585) GFR 25–55 mL/min/1.73 m2 Intervention 1: Usual protein diet versus low-protein diet Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 38.6 ± 8.9 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Usual protein diet 12.1 (10.5–13.8) mL/min/3 years versus low-protein diet 10.9 (9.2–12.5) mL/min/3 years; P = 0.3 Intervention 2: Usual BP 12.3 (10.6–14) mL/min/3 years versus low BP 10.7 (9.1–12.4) mL/min/3 years; P = 0.18 Secondary: (Data according to the randomization groups was not reported): Death: 15 (3%) ESKD: 12 (2%) Rapid decline in GFR (>50 reduction in GFR if baseline GFR ≤ 40 mL/min/1.73 m2 or ≥20 mL/min/1.73 m2 reduction in GFR if baseline GFR > 40 mL/min/1.73 m2): 60 (10%) |
| Klahr 1994 [41] MDRD Study 2 | Population: (n = 255) GFR 13–24 mL/min/1.73 m2 Intervention 1: Low-protein diet versus very low-protein diet with keto acid-amino acid supplement Intervention 2: Usual BP versus low BP Baseline kidney function: Overall GFR 18.5 ± 3.4 mL/min/1.73 m2 Follow-up: Mean 2.2 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope (without adjustment for the body-surface area), results expressed as mean (95% CI) Intervention 1: Low-protein diet 4.4 (3.7–5.1) mL/min/year versus very low-protein diet 3.6 (2.9–4.2) mL/min/year; P = 0.07 Intervention 2: Usual BP 4.2 (3.6–4.9) mL/min/year versus low BP 3.7 (3.1–4.3) mL/min/year; P = 0.28 Secondary: (Data according to the randomization groups was not reported): Death: 15 (6%) ESKD: 94 (37%) |
| Toto 1995 [34] | Population: (n = 77) Long-standing hypertension, GFR ≤ 70 mL/min/1.73 m2), normal urine sediment, excluded: diabetes Interventions: Strict BP control (n = 42) versus conventional BP control (n = 35) Baseline kidney function: GFR 34.6 ± 2.3 versus 41.9 ± 3.15 mL/min/1.73 m2 Follow-up: Mean 40.5 ± 1.8 months versus 42.2 ± 2.1 months | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope −0.31 ± 0.45 versus −0.05 ± 0.5 mL/min/1.73 m2/year; P > 0.25 Secondary: 50% reduction in GFR (or doubling of serum creatinine) or ESRD or death: 12 (29%) versus 7(20%); P > 0.25 50% reduction in GFR (or doubling of serum creatinine): 4 (10%) versus 5(14%) ESKD: 7 (17%) versus 2 (6%) Death: 1 (2%) versus 0 (0%) |
| Maschio 1996 [35] | Population: (n = 583) serum creatinine 1.5–4 mg/dL (133–354 μmol/L), 24-h creatinine clearance 30–60 mL/min Interventions: Benazepril (n = 300) versus placebo (n = 283) Baseline kidney function: Creatinine clearance 42.9 ± 11.6 mL/min versus 42.3 ± 10.6 mL/min Follow-up: 3 years | Primary: Composite of need for dialysis or doubling of serum creatinine (total events 88/15%) Benazepril 31 (10%) versus placebo 57 (20%); P < 0.001 Secondary: Doubling of serum creatinine: total events 86 (15%) Data according to the randomization groups was not reported. Dialysis: total events 2 (<1%) Data according to the randomization groups was not reported. Death: 8 (3%) versus 1 (<1%) |
| The GISEN Group 1997 [50] REIN Study Stratum 2 | Population: (n = 166) Non-diabetic nephropathy, creatinine clearance 20–70 mL/min/1.73 m2, proteinuria ≥1 g/day (≥3 g/day for stratum 2); excluded: Type 1 diabetes mellitus Interventions: Ramipril (n = 78) versus placebo (n = 88) Baseline kidney function: GFR 40.2 ± 19 mL/min/1.73 m2 versus 37.4 ± 17.5 mL/min/1.73 m2 Follow-up: Mean 16 months | Primary: Change in GFR (plasma clearance of non-radioactive iohexol), results expressed as mean (SE): 0.53 (0.08) versus 0.88 (0.13) mL/min/month; P = 0.03 Secondary: Doubling of serum creatinine or ESKD: 18 (23%) versus 40 (45%); P = 0.02 ESKD: 17 (22%) versus 29 (33%); P = 0.2 Death: 2 (3%) versus 1 (1%) |
| Kuriyama 1997 [36] | Population: (n = 73) serum creatinine 2–4 mg/dL, hematocrit <30% Interventions: No EPO (n = 31) versus EPO (n = 42) Baseline kidney function: Creatinine clearance 17.1 ± 7.2 mL/min versus 19.1 ± 7.2 mL/min Follow-up: 3 years | Primary: Doubling of serum creatinine (total 48 events/66%) No EPO group 26 (84%) versus EPO group 22 (52%); P = 0.0003 Secondary: Dialysis: 20 (65%) versus 14 (33%) Death: 2 (%) versus 1 (%) |
| Brenner 2001 [10] (RENAAL) | Population: (n = 1513) Type 2 diabetes mellitus, albuminuria >300 mg/g (or >0.5 g/day), serum creatinine 1.3–3 mg/dL (115–265 μmol/L) or 1.5–3 mg/dL (133–265 μmol/L) if weight >60 kg; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Losartan (n = 751) versus placebo (n = 762) Baseline kidney function: serum creatinine 1.9 ± 0.5 mg/dL versus 1.9 ± 0.5 mg/dL Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESRD or death (total 686 events/45%) Losartan 327 (44%) versus placebo 359 (47%); P = 0.02 Secondary: Doubling of serum creatinine: 162 (22%) versus 198 (26%); P = 0.006 ESKD: 147 (20%) versus 194(26%); P = 0.002 Death: 158 (21%) versus 155 (20%); P = 0.88 ESKD or death: 255 (12%) versus 300 (39%); P = 0.01 Doubling of serum creatinine or ESRD: 226 (30%) versus 263 (35%); P = 0.01 Reciprocal of serum creatinine (median slope): −0.056 dL/mg/year versus −0.069 dL/mg/year; P = 0.01 Median change in GFR: −4.4 versus −5.2 mL/min/1.73 m2/year; P = 0.01 |
| Lewis 2001 [20] (IDNT) | Population: (n = 1715) Type 2 diabetes mellitus, proteinuria >0.9 g/day), serum creatinine 1.0–3 mg/dL (88–265 μmol/L) in women or 1.2–3 mg/dL (106–265 μmol/L) in men; excluded: type 1 diabetes mellitus and non-diabetic renal disease Interventions: Irbesartan (n = 579) versus placebo (n = 569) versus amlodipine (n = 567) Baseline kidney function: Serum creatinine irbesartan 1.67 ± 0.53 mg/dL versus placebo 1.69 ± 0.57 mg/dL versus amlodipine 1.65 ± 0.61 mg/dL Follow-up: Mean 2.6 years | Primary: Composite of doubling of serum creatinine or ESKD (need for dialysis or transplantation or serum creatinine ≥6 mg/dL [≥560 μmol/L] or death (total 644 events/38%) Irbesartan 189 (33%) versus placebo 222 (39%); P = 0.02 Irbesartan 189 (33%) versus amlodipine 233 (41%); P = 0.006 Secondary: Doubling of serum creatinine: Irbesartan 98 (17%) versus placebo 135 (24%); P = 0.003 and versus amlodipine 144 (25%); P < 0.001 ESKD: Irbesartan 82 (14%) versus placebo 101 (18%); P = 0.07 and versus amlodipine 104 (18%); P = 0.07 Death: Irbesartan 87 (15%) versus placebo 93 (16%); P = 0.57 and versus amlodipine 83 (15%); P = 0.8 Mean change in creatinine clearance: Irbesartan −5.5 ± 0.36 versus placebo −6.5 ± 0.37 versus amlodipine −6.8 ± 0.37 mL/min/1.73 m2/year |
| Wright 2002 [42] (AASK) | Population: (n = 1094) Hypertensive renal disease, GFR 20–65 mL/min/1.73 m2; excluded: diabetes mellitus, proteinuria > Intervention 1: Low BP (n = 540) versus usual BP (n = 554) Intervention 2: Ramipril (n = 436) versus amlodipine (n = 217) versus metoprolol (n = 441) Baseline kidney function: GFR mean(SE) Low BP 46(12.9) mL/min/1.73 m2 versus usual BP 45.3(13.2) mL/min/1.73 m2 Ramipril 45.4(12.8) mL/min/1.73 m2 versus amlodipine 45.8(12.9) mL/min/1.73 m2 versus metoprolol 45.8(13.4) mL/min/1.73 m2 Follow-up: median 3.8 years | Primary: Rate of decline in measured GFR (renal clearance of 125I-iothalamate) or GFR slope, results expressed as mean (SE) Intervention 1: Low BP −2.21 (0.17) versus usual BP −1.95 (0.17) mL/min/1.73 m2/year; P = 0.24 Intervention 2: Ramipril −1.81 (0.17) versus metoprolol −2.42 (0.17) mL/min/1.73 m2/year; P = 0.07 Amlodipine −1.60 (0.34) versus metoprolol −2.68 (0.20) mL/min/1.73 m2/year; P = 0.0.04 Secondary: (overall results described): Composite of ≥50% reduction in GFR or ≥25 mL/min/1.73 m2 from baseline or ESRD (dialysis or transplantation) or death: total 340 (31%) events (170 GFR events, 84 ESKD, 77 deaths). GFR event or ESKD: 263 (24%) events (179 GFR events, 84 ESKD). ESKD or death: 251 (23%) events (171 ESKD, 80 deaths) ESKD alone: 171 (16%) (death censored) |
| Barnett 2004 [51] | Population: (n = 250) Type 2 diabetes mellitus, urinary albumin excretion rate 11–999 μg/min, serum creatinine <1.6 mg/dL (<141 μmol/L), GFR > 70 mL/min/1.73 m2 Interventions: Telmisartan (n = 120) versus enalapril (n = 130) Baseline kidney function: GFR 91.4 ± 21.5 mL/min/1.73 m2 versus 94.3 ± 22.1 mL/min/1.73 m2 Follow-up: 5 years | Primary: Change (between the baseline value and the last available value during the 5-year treatment period) in GFR (plasma clearance of iohexol) Telmisartan −17.9 versus enalapril −14.9 mL/min/1.73 m2 over 5 years Secondary: Annual change in GFR (overall results): −7.6, −5.6 and −3.6 mL/min/1.73 m2 at 1, 2 and 3 years, respectively. Data according to the randomization groups was not reported. Death: 6 (5%) versus 6 (5%) ESKD (pre-specified end point): Data not reported |
| Ruggenenti 2005 [52] (REIN-2) | Population: (n = 338) Non-diabetic nephropathy, proteinuria ≥1 g/day, creatinine clearance <45 mL/min/1.73 m2 if proteinuria 1–3 g/day or creatinine clearance <70 mL/min/1.73 m2 if proteinuria ≥3 g/day; excluded: type 1 diabetes mellitus Interventions: Intensified BP control (n = 169) versus conventional BP control (n = 169) Baseline kidney function: GFR (mean ± SD) Intensified BP control (35.9 ± 18.6 mL/min/1.73 m2) versus conventional BP control (34.1 ± 18.1 mL/min/1.73 m2) Follow-up: median 19 months | Primary: Time to ESKD ESKD: intensified BP control 38 (23%) versus conventional BP control 34 (20%); P = 0.99 Secondary: GFR slope (plasma clearance of non-radioactive iohexol), expressed as median (IQR) rate of GFR change: −0.22 (0.06 to 0.55) versus −0.24 (0.0001 to 0.56) mL/min/1.73 m2/month; P = 0.62 Creatinine clearance slope, expressed as median (IQR) rate of GFR change: −0.26 (0.03 to 0.53) versus −0.25 (0.0001 to 0.75) mL/min/1.73 m2/month; P = .59 Death: 3 (2%) versus 2 (1%) |
| Hou 2006 [19] | Population: (n = 328) Non-diabetic renal disease, serum creatinine 1.5–5 mg/dL (133–442 μmol/L), creatinine clearance 20–70 mL/min/1.73 m2, proteinuria >0.3 g/day Interventions: Group 1 (serum creatinine 1.5–3 mg/dL): Benazepril in all participants (n = 104); Group 2 (serum creatinine 3.1–5 mg/dL): Benazepril (n = 112) versus placebo (n = 112) Baseline kidney function: Calculated GFR: Group 1: 37.1 ± 6.3 mL/min/1.73 m2, Group 2: Benazepril 26.3 ± 5.3 mL/min/1.73 m2 versus 25.8 ± 5.3 mL/min/1.73 m2 Follow-up: Mean 3.4 years | Primary: Composite of doubling of serum creatinine or ESKD or death Group 1: 22 (22%) Group 2 (total events 109/51%): Benazepril 44 (41%) versus 65 (60%); P = 0.004 Secondary: (Group 2 results only): Reciprocal of serum creatinine (median slope): −0.09 dL/mg/year versus −0.11 dL/mg/year; P = 0.02 Median change in MDRD eGFR: −6.8 versus −8.8 mL/min/ 1.73 m2/year; 0.006 Data on individual events of doubling of serum creatinine and ESKD were not reported. Death: 1 (1%) versus 0 (0%) |
| Pfeffer 2009 [53] (TREAT) | Population: (n = 4038) type 2 diabetes mellitus, MDRD eGFR 20–60 mL/min/1.73 m2, Hb ≤ 11 g/dL, transferrin saturation ≥15% Intervention: Darbepoetin alfa (n = 2012) versus placebo (n = 2026) Baseline kidney function: median (IQR) Darbepoetin alfa 34(27–43) mL/min/1.73 m2 versus placebo 33(26–42) mL/min/1.73 m2 Follow-up: median 29.1 months | Primary: Death or ESKD (dialysis >30 days or transplantation or death within 30 days of starting dialysis or refusing renal replacement therapy despite of physician recommendation) total events (1276/32%) Darbepoetin alfa 652 (32%) versus placebo 618 (31%); P = 0.29 Secondary: ESKD: 338 (17%) versus 330 (16%); P = 0.83 Death: 412 (21%) versus 395 (20%); P = 0.48 Fatal or non-fatal stroke: 101 (5%) versus 53 (2.6%); P < 0.001 |
| Fried 2013 [31] (VA NEPHRON-D) | Population: (n = 1448) type 2 diabetes mellitus, albuminuria ≥300 mg/g, MDRD eGFR 30–89.9 mL/min/1.73 m2; excluded: non-diabetic renal disease Intervention: Combination therapy ACE inhibitor and ARB (n = 724) versus ARB monotherapy (n = 724) Baseline kidney function: mean ± SD eGFR Combination therapy 53.6 ± 15.5 mL/min/1.73 m2 versus ARB monotherapy 53.7 ± 16.2 mL/min/1.73 m2 Follow-up: Median 2.2 year | Primary: Composite of absolute reduction in eGFR ≥ 30 mL/min/1.73 m2 if baseline eGFR ≥ 60 mL/min/1.73 m2 or ≥50% reduction in eGFR if baseline eGFR < 60 mL/min/1.73 m2 or ESRD (dialysis or eGFR < 15 mL/min/1.73 m2) or death- total events (284/20%) Combination therapy 132 (18%) [59 GFR events, 18 ESKD, 55 deaths] versus ARB monotherapy 152 (21%) [78 GFR events, 23 ESKD, 51 deaths]; P = 0.3 Secondary: GFR event or ESKD: 77 (11%) versus 101 (14%); P = 0.1 ESKD: 27 (4%) versus 43 (6%); P = 0.07 Death: 63 (9%) versus 60 (8%); P = 0.75 eGFR slope: −2.7 versus −2.9 mL/min/1.73 m2/year; P = 0.17 |
| de Zeeuw 2013 [30] (BEACON) | Population: (n = 2185) type 2 diabetes mellitus, eGFR 15–29 mL/min/1.73 m2 Intervention: Bardoxolone methyl (n = 1088) versus placebo (n = 1097) Baseline kidney function: mean ± SD eGFR Bardoxolone methyl 22.2 ± 4.3 mL/min/1.73 m2 versus placebo 22.5 ± 4.6 mL/min/1.73 m2 Follow-up: 9 months | Primary: ESKD (dialysis >12 weeks or transplantation) or death from cardiovascular causes- total events (138/6%) Bardoxolone methyl 69 (6%) versus placebo 69 (6%); P = 0.92 Secondary: ESKD: 43 (4%) versus 51 (5%); P = 0.35 Death from cardiovascular causes: 27 (2%) versus 19 (2%); P = 0.23 Death from any cause: 44 (4%) versus 31 (3%); P = 0.1 Hospitalization for heart failure or death from heart failure: 96 (9%) versus 55 (5%); P < 0.001 Composite outcome of non-fatal myocardial infarction, non-fatal stroke, hospitalization for heart failure or death from cardiovascular causes: 139 (13%) versus 86 (8%); P < 0.001 Absolute change in eGFR, expressed as mean (95% CI): +5.5 (5.2 to 5.9) mL/min/1.73 m2 versus −0.9 (−1.2 to −0.5) mL/min/1.73 m2; P < 0.001 |
| Haynes 2014 [54] (SHARP) | Population: (n = 6245) Serum creatinine ≥1.7 mg/dL (≥150 μmol/L) in men or ≥1.5 mg/dL (≥130 μmol/L) in women Intervention: Simvastatin plus ezetimibe (n = 3116) versus placebo (n = 3129) Baseline kidney function: Mean ± SD MDRD eGFR Simvastatin plus ezetimibe 26.6 ± 12.9 mL/min/1.73 m2 versus placebo 26.6 ± 13.1 mL/min/1.73 m2 Follow-up: 4.8 years | Pre-specified subsidiary renal outcome: ESKD (dialysis or transplantation)- total events (2141/34%) Simvastatin plus ezetimibe 1057 (34%) versus placebo 1084 (35%); P = 0.41 Tertiary: ESKD or death: 1477 (47%) versus 1513 (48%); P = 0.34 ESKD or doubling of serum creatinine: 1189 (38%) versus 1257 (40%); P = 0.09 Death: 635 (20%) versus 635 (20%); P = 0.93 Doubling of serum creatinine: 370 (12%) versus 415 (13%); P = 0.11 eGFR slope, expressed as mean (SE): −1.66 (0.07) versus −1.83 (0.07) mL/min/1.73 m2/year; P = 0.1 |
| Torres (2014) [55] (HALT-PKD Study B) | Population: (n = 486) Autosomal dominant polycystic kidney disease, age 18–64 years, eGFR 25 to 60 mL/min/1.73 m2 Intervention: Lisinopril plus telmisartan (n = 244) versus lisinopril plus placebo (n = 242) Baseline kidney function: Mean ± SD CKD-EPI eGFR Lisinopril plus telmisartan 48.5 ± 11.5 mL/min/1.73 m2 versus lisinopril plus placebo 47.9 ± 12.2 mL/min/1.73 m2 Follow-up: 5.2 years | Primary: Composite of death or ESKD or a 50% decline in baseline eGFR (total events 231/48%) Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) Secondary: 50% decline in baseline eGFR: total events 182 (37%), Lisinopril plus telmisartan 115 (47%) versus lisinopril plus placebo 116 (48%) ESKD: total events 112 (23%), Lisinopril plus telmisartan 46 (19%) versus lisinopril plus placebo 66 (27%) Death: total events 9 (2%), Lisinopril plus telmisartan 4 (2%) versus lisinopril plus placebo 5 (2%) Rate of change in eGFR, expressed as mean (95% CI): 3.91 (−3.65 to −4.17) mL/min/1.73 m2/year versus −3.87 (−3.61 to −4.14) mL/min/1.73 m2/year |
AASK, (the African American Study of Kidney Disease and Hypertension); AASK, African American Study of Kidney Disease and Hypertension; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; BEACON, Bardoxolone Methyl Evaluation in Patients with Chronic Kidney Disease and Type 2 Diabetes Mellitus, the Occurrence of Renal Events Study; BP, blood pressure; CI, confidence intervals; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration equation; eGFR, estimated glomerular filtration rate; EPO, erythropoietin; ESKD, end-stage kidney disease; GFR, glomerular filtration rate; HALT-PKD, The Halt Progression of Polycystic Kidney Disease trial; IDNT, Irbesartan Type II Diabetic Nephropathy Trial; IQR, inter-quartile range; MDRD, Modification of Diet in Renal Disease Study; RENAAL, Reduction of End points in NIDDM with the Angiotensin II Antagonist Losartan Study; REIN-2, Ramipril Efficacy in Nephropathy 2 Study; SD, standard deviation; SE, standard error; SHARP, Study of Heart and Renal Protection; TREAT, A Trial of Darbepoetin Alfa in Type 2 Diabetes and Chronic Kidney Disease; VA NEPHRON-D, The Veterans Affairs Nephropathy in Diabetes Study.
Composite end point
A composite end point of (i) death, (ii) development of ESKD and (iii) serum creatinine doubling has been widely used and accepted in CKD progression trials [10, 19, 20]. The principal advantage of combining these clinically meaningful end points is that they increase the total number of expected events thereby leading to smaller, shorter duration and less costly trials.
Ideally, individual components of a composite end point should (i) be similar in importance to patients, (ii) occur with similar frequency and (iii) be affected by the intervention to a comparable extent and direction as the other components [21, 22]. However, serum creatinine doubling is not a ‘patient-reported’ outcome, defined as a measure of how a patient feels or functions in relation to a health condition and its therapy without interpretation by healthcare professionals or anyone else [23]. Therefore, the patient-level importance of serum creatinine doubling is substantially less than that of death and ESKD. These components are also not similar in terms of frequency of event rates and treatment effect sizes. Therefore, the current kidney end point does not strictly meet the recommendations for an ideal composite end point. It is important to note that while the substitution of one surrogate end point by another is likely to reduce the required sample size, it may come at the cost of effect imprecision and uncertainty [24]. Consequently, the individual components of a composite end point should be separately analyzed and reported [25].
Death
Death is indisputably the most important clinical end point, because it is both clearly defined and important to patients, clinicians and policy-makers. However, death generally occurs significantly less frequently than ESKD, which in turn occurs less frequently than serum creatinine doubling. Thus, trials using death alone as an end point will require a significantly larger sample size than those using the composite kidney end point (Table 1).
End-stage kidney disease
ESKD is typically defined as the initiation of renal replacement therapy (either dialysis for >30 days or kidney transplantation). It indicates a permanent organ loss and is characterized by heightened mortality, morbidity, poor quality of life and appreciable use of healthcare resources. Therefore, ESKD is an important and core outcome. However, there are several challenges in using ESKD as a clinical end point. First, there is substantial variation and subjectivity in the timing of renal replacement therapy initiation in clinical practice. In the pre-Initiating Dialysis Early and Late (IDEAL) Study era, there was a trend toward initiation of maintenance dialysis at higher eGFR levels over time [26–28]. Second, ∼1% of patients who commence chronic dialysis subsequently stop due to kidney function recovery [29]. Third, the minimum duration of dialysis required to adjudicate the ESKD event has varied from 30 days to 12 weeks in different trials [30]. Fourth, a small number of study participants will erroneously not be adjudicated for the ESKD end point if they opt for conservative care or refuse dialysis and do not start maintenance dialysis. The VA NEPHRON-D (The Veterans Affairs Nephropathy in Diabetes) Study attempted to circumvent this issue by defining ESKD as reaching an eGFR < 15 mL/min/1.73 m2 (along with initiation of maintenance dialysis) [31]. This expanded definition is more likely to capture ESKD events, even when renal replacement therapy is not commenced. However, this definition should ideally have an additional requirement of eGFR < 15 mL/min/1.73 m2 for >3 months and be restricted to participants with a baseline eGFR > 25 mL/min/1.73 m2 to avoid including short-term, temporary eGFR reductions as CKD progression events. In the Irbesartan in Diabetic Nephropathy Trial (IDNT), the investigators included serum creatinine ≥6 mg/dL (530 μmol/L) in the definition of ESKD [20]. These points highlight the importance of using a uniform definition of ESKD in future trials.
Serum creatinine doubling
Serum creatinine doubling has been increasingly used as a surrogate end point in CKD progression trials [32–36]. It offers several advantages. First, being an intermediate event occurring prior to ESKD, it could theoretically reduce the duration of trial follow-up. Second, its inclusion in the composite end point substantially enriches the end point by accounting for 50–60% of all events (Table 1) [10, 20]. Third, it is a dichotomous outcome and can be easily used in time-dependent or survival analyses. Finally, since it roughly corresponds to halving of GFR, serum creatinine doubling was thought to be of similar importance to ESKD. In a recent systematic review and meta-analysis, Jun et al. [37] assessed the correlation between doubling of serum creatinine and occurrence of ESKD using data from 20 randomized trials. The treatment effect ratio (calculated as the treatment effect on ESKD/the treatment effect on doubling of serum creatinine) was close to 1 (0.98, 95% CI 0.85–1.14), suggesting that the treatment effect on serum creatinine doubling was nearly the same as that on ESKD.
However, there are several limitations to using end points that are based on the serum creatinine measurement. First, serum creatinine is an imprecise biomarker of GFR [38, 39]. Second, kidney function decline is generally not linear over time [40]. Third, as seen in the Modification of Diet in Renal Disease (MDRD) Study and the African American Study of Kidney Disease (AASK) trials, GFR reduction during the first 3–4 months of follow-up may reflect reversible hemodynamic changes attributable to the study interventions rather than true progression of CKD [41, 42]. Fourth, numerous factors other than GFR can influence serum creatinine concentration. Fifth, informative censoring bias may occur due to the competing events of death and renal replacement therapy initiation when serum creatinine doubling is analyzed as an individual end point. Finally, there is no evidence to confirm the concordance of treatment effects on serum creatinine doubling and mortality.
LESSER-ESTIMATED GLOMERULAR FILTRATION RATE DECLINE
The advent of widespread automated laboratory reporting of eGFR, most commonly using either the MDRD or Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, has fueled a push among researchers to consider eGFR decline as an alternative surrogate end point to serum creatinine doubling in clinical trials [43, 44]. A 50% decline in eGFR approximately equates to a serum creatinine doubling, although the corresponding values for the MDRD and CKD-EPI equations are 55 and 57%, respectively. This alternative CKD progression end point suffers from the same limitations as serum creatinine doubling that were detailed earlier.
More recently, researchers have advocated for a lesser decline in eGFR, such as 30 or 40%, as an alternative CKD progression end point, as such events will occur earlier and more frequently than a halving of eGFR. For example, Lambers Heerspink et al. reported that in the Reduction of Endpoints in NIDDM with the Angiotensin II Antagonist Losartan (RENAAL) study, replacement of serum creatinine doubling by 40 and 30% eGFR declines increased the event rates from 33.6 to 48.8% and 61.4%, respectively [45]. These respective event rates in the IDNT trial increased from 28.1 to 39.1% and 51.5%.
The advantage of using a lesser eGFR decline end point particularly applies to trial populations with normal or only mildly decreased baseline eGFR values, in which 3- to 5-year event rates for established kidney end points are uniformly low. For example, in the Tolvaptan Efficacy and Safety in Management of Autosomal Dominant Polycystic Kidney Disease and Its Outcomes (TEMPO 3:4) trial, the baseline eGFR was 81 mL/min/1.73 m2 and no deaths were observed during the 3-year follow-up period [46]. Although the investigators did not report the event rates of ESKD and serum creatinine doubling, they were likely to be low. In such scenarios, a lesser GFR decline end point would allow the investigators to conduct the trial within a reasonable follow-up period.
Validity of lesser GFR decline as a surrogate end point
The NKF-FDA Scientific Workshop reported a series of four studies to determine whether a lesser eGFR decline (using the CKD-EPI equation) can be used in kidney trials [47]. The results of these studies are described in Table 3 and discussed subsequently (Figure 1).
Description of the FDA-NKF Scientific Workshop studies
| Study | Description of study | Main results |
|---|---|---|
| Coresh [15] CKD Prognosis Consortium | Design: Patient-level meta-analysis Participants: ESKD analysis: 22 cohort studies, 1.5 million individuals Mortality analysis: 35 cohort studies, 1.7 million individuals Average eGFR (CKD-EPI Equation): Lower eGFR stratum 48 mL/min/1.73 m2 Higher eGFR stratum 92 mL/min/1.73 m2 Predictors: Lesser declines in eGFR (ranging from 0 to 40%) over a baseline period ranging from 1 to 3 years Primary outcome: ESKD (initiation of renal replacement therapy or death due to renal disease) during the subsequent follow-up after the initial baseline period Secondary outcome: Death | Lesser eGFR decline events occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 2-year baseline period and subsequent risks of ESKD and death. 30% eGFR decline over 2 years: HR for ESKD 5.4 and 6.7 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 1.8 and 1.6 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 2 years: HR for ESKD 10.2 and 15.3 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 3.5 and 2.4 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) |
| Lambers Heerspink [18] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Mean ± SD baseline eGFR (CKD-EPI Equation): 49.2 ± 24.9 mL/min/1.73 m2 Predictors: 30 and 40% declines in eGFR from baseline to 12 months Primary outcome: Composite of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine during the subsequent follow-up Secondary outcome: Above composite end point plus death Median follow-up after the 12-month baseline period: 2.0 (IQR 1.2–3.1) | The events of 30 and 40% eGFR decline occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 1-year baseline period and subsequent risks of ESRD and death, after adjusting for age, sex, treatment assignment, baseline eGFR and proteinuria. 30% eGFR decline over 1 years: HR for ESKD 9.6 (reference 0% decline ∼ stable kidney function) HR for death 7.3 (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 1 years: HR for ESKD 20.3 (reference 0% decline ∼ stable kidney function) HR for death 14.2 (reference 0% decline ∼ stable kidney function) |
| Inker [17] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Studies categorized into 5 types of interventions: RAS blockade versus control; RAS blockade versus calcium channel blocker; intensive blood pressure control; low-protein diet; and immunosuppressive therapy Index test: Alternative end points of lesser declines in eGFR (ranging from 30 to 40%) over a baseline period ranging from 12 to 24 months and throughout study duration Reference test: Established composite end point of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine Median follow-up: 3.62 years Analysis: Agreement between the established and alternative end points was calculated by Bayesian mixed models. | Treatment effects attenuated for lesser declines, particularly for 20 and 30% when compared with the established end points (suggestive of a weaker treatment effect). For the intervention of RAS blockade versus control, treatment effect increased for 30 and 40% declines at shorter interval (suggestive of a stronger treatment effect). For the intervention of low-protein diet, treatment effect increased for all alternative end points (suggestive of a stronger treatment effect). This result further amplified at shorter intervals. |
| Greene [16] | Design: Simulation study Index test: Alternative composite end point based on ESKD and either 30 or 40% eGFR decline Reference test: Established composite end point of ESKD or 57% eGFR decline Analysis: Comparison of the risk of type 1 errors for established and alternative end points. | Compared with a 57% eGFR decline, a 40% eGFR decline resulted in >20% reduction in the required sample size for a 2-year study. Use of 30% eGFR declined reduced the required sample size only in the absence of an acute treatment effect on eGFR. |
| Study | Description of study | Main results |
|---|---|---|
| Coresh [15] CKD Prognosis Consortium | Design: Patient-level meta-analysis Participants: ESKD analysis: 22 cohort studies, 1.5 million individuals Mortality analysis: 35 cohort studies, 1.7 million individuals Average eGFR (CKD-EPI Equation): Lower eGFR stratum 48 mL/min/1.73 m2 Higher eGFR stratum 92 mL/min/1.73 m2 Predictors: Lesser declines in eGFR (ranging from 0 to 40%) over a baseline period ranging from 1 to 3 years Primary outcome: ESKD (initiation of renal replacement therapy or death due to renal disease) during the subsequent follow-up after the initial baseline period Secondary outcome: Death | Lesser eGFR decline events occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 2-year baseline period and subsequent risks of ESKD and death. 30% eGFR decline over 2 years: HR for ESKD 5.4 and 6.7 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 1.8 and 1.6 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 2 years: HR for ESKD 10.2 and 15.3 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 3.5 and 2.4 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) |
| Lambers Heerspink [18] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Mean ± SD baseline eGFR (CKD-EPI Equation): 49.2 ± 24.9 mL/min/1.73 m2 Predictors: 30 and 40% declines in eGFR from baseline to 12 months Primary outcome: Composite of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine during the subsequent follow-up Secondary outcome: Above composite end point plus death Median follow-up after the 12-month baseline period: 2.0 (IQR 1.2–3.1) | The events of 30 and 40% eGFR decline occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 1-year baseline period and subsequent risks of ESRD and death, after adjusting for age, sex, treatment assignment, baseline eGFR and proteinuria. 30% eGFR decline over 1 years: HR for ESKD 9.6 (reference 0% decline ∼ stable kidney function) HR for death 7.3 (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 1 years: HR for ESKD 20.3 (reference 0% decline ∼ stable kidney function) HR for death 14.2 (reference 0% decline ∼ stable kidney function) |
| Inker [17] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Studies categorized into 5 types of interventions: RAS blockade versus control; RAS blockade versus calcium channel blocker; intensive blood pressure control; low-protein diet; and immunosuppressive therapy Index test: Alternative end points of lesser declines in eGFR (ranging from 30 to 40%) over a baseline period ranging from 12 to 24 months and throughout study duration Reference test: Established composite end point of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine Median follow-up: 3.62 years Analysis: Agreement between the established and alternative end points was calculated by Bayesian mixed models. | Treatment effects attenuated for lesser declines, particularly for 20 and 30% when compared with the established end points (suggestive of a weaker treatment effect). For the intervention of RAS blockade versus control, treatment effect increased for 30 and 40% declines at shorter interval (suggestive of a stronger treatment effect). For the intervention of low-protein diet, treatment effect increased for all alternative end points (suggestive of a stronger treatment effect). This result further amplified at shorter intervals. |
| Greene [16] | Design: Simulation study Index test: Alternative composite end point based on ESKD and either 30 or 40% eGFR decline Reference test: Established composite end point of ESKD or 57% eGFR decline Analysis: Comparison of the risk of type 1 errors for established and alternative end points. | Compared with a 57% eGFR decline, a 40% eGFR decline resulted in >20% reduction in the required sample size for a 2-year study. Use of 30% eGFR declined reduced the required sample size only in the absence of an acute treatment effect on eGFR. |
CKD, chronic kidney disease; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration equation; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; HR, hazard ratio; IQR, inter-quartile range; RAS, renin–angiotensin system.
Description of the FDA-NKF Scientific Workshop studies
| Study | Description of study | Main results |
|---|---|---|
| Coresh [15] CKD Prognosis Consortium | Design: Patient-level meta-analysis Participants: ESKD analysis: 22 cohort studies, 1.5 million individuals Mortality analysis: 35 cohort studies, 1.7 million individuals Average eGFR (CKD-EPI Equation): Lower eGFR stratum 48 mL/min/1.73 m2 Higher eGFR stratum 92 mL/min/1.73 m2 Predictors: Lesser declines in eGFR (ranging from 0 to 40%) over a baseline period ranging from 1 to 3 years Primary outcome: ESKD (initiation of renal replacement therapy or death due to renal disease) during the subsequent follow-up after the initial baseline period Secondary outcome: Death | Lesser eGFR decline events occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 2-year baseline period and subsequent risks of ESKD and death. 30% eGFR decline over 2 years: HR for ESKD 5.4 and 6.7 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 1.8 and 1.6 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 2 years: HR for ESKD 10.2 and 15.3 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 3.5 and 2.4 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) |
| Lambers Heerspink [18] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Mean ± SD baseline eGFR (CKD-EPI Equation): 49.2 ± 24.9 mL/min/1.73 m2 Predictors: 30 and 40% declines in eGFR from baseline to 12 months Primary outcome: Composite of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine during the subsequent follow-up Secondary outcome: Above composite end point plus death Median follow-up after the 12-month baseline period: 2.0 (IQR 1.2–3.1) | The events of 30 and 40% eGFR decline occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 1-year baseline period and subsequent risks of ESRD and death, after adjusting for age, sex, treatment assignment, baseline eGFR and proteinuria. 30% eGFR decline over 1 years: HR for ESKD 9.6 (reference 0% decline ∼ stable kidney function) HR for death 7.3 (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 1 years: HR for ESKD 20.3 (reference 0% decline ∼ stable kidney function) HR for death 14.2 (reference 0% decline ∼ stable kidney function) |
| Inker [17] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Studies categorized into 5 types of interventions: RAS blockade versus control; RAS blockade versus calcium channel blocker; intensive blood pressure control; low-protein diet; and immunosuppressive therapy Index test: Alternative end points of lesser declines in eGFR (ranging from 30 to 40%) over a baseline period ranging from 12 to 24 months and throughout study duration Reference test: Established composite end point of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine Median follow-up: 3.62 years Analysis: Agreement between the established and alternative end points was calculated by Bayesian mixed models. | Treatment effects attenuated for lesser declines, particularly for 20 and 30% when compared with the established end points (suggestive of a weaker treatment effect). For the intervention of RAS blockade versus control, treatment effect increased for 30 and 40% declines at shorter interval (suggestive of a stronger treatment effect). For the intervention of low-protein diet, treatment effect increased for all alternative end points (suggestive of a stronger treatment effect). This result further amplified at shorter intervals. |
| Greene [16] | Design: Simulation study Index test: Alternative composite end point based on ESKD and either 30 or 40% eGFR decline Reference test: Established composite end point of ESKD or 57% eGFR decline Analysis: Comparison of the risk of type 1 errors for established and alternative end points. | Compared with a 57% eGFR decline, a 40% eGFR decline resulted in >20% reduction in the required sample size for a 2-year study. Use of 30% eGFR declined reduced the required sample size only in the absence of an acute treatment effect on eGFR. |
| Study | Description of study | Main results |
|---|---|---|
| Coresh [15] CKD Prognosis Consortium | Design: Patient-level meta-analysis Participants: ESKD analysis: 22 cohort studies, 1.5 million individuals Mortality analysis: 35 cohort studies, 1.7 million individuals Average eGFR (CKD-EPI Equation): Lower eGFR stratum 48 mL/min/1.73 m2 Higher eGFR stratum 92 mL/min/1.73 m2 Predictors: Lesser declines in eGFR (ranging from 0 to 40%) over a baseline period ranging from 1 to 3 years Primary outcome: ESKD (initiation of renal replacement therapy or death due to renal disease) during the subsequent follow-up after the initial baseline period Secondary outcome: Death | Lesser eGFR decline events occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 2-year baseline period and subsequent risks of ESKD and death. 30% eGFR decline over 2 years: HR for ESKD 5.4 and 6.7 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 1.8 and 1.6 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 2 years: HR for ESKD 10.2 and 15.3 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) HR for death 3.5 and 2.4 for lower and higher eGFR strata, respectively (reference 0% decline ∼ stable kidney function) |
| Lambers Heerspink [18] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Mean ± SD baseline eGFR (CKD-EPI Equation): 49.2 ± 24.9 mL/min/1.73 m2 Predictors: 30 and 40% declines in eGFR from baseline to 12 months Primary outcome: Composite of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine during the subsequent follow-up Secondary outcome: Above composite end point plus death Median follow-up after the 12-month baseline period: 2.0 (IQR 1.2–3.1) | The events of 30 and 40% eGFR decline occurred more frequently than 57% eGFR decline during the baseline period. Strong association between 30 and 40% eGFR decline over a 1-year baseline period and subsequent risks of ESRD and death, after adjusting for age, sex, treatment assignment, baseline eGFR and proteinuria. 30% eGFR decline over 1 years: HR for ESKD 9.6 (reference 0% decline ∼ stable kidney function) HR for death 7.3 (reference 0% decline ∼ stable kidney function) 40% eGFR decline over 1 years: HR for ESKD 20.3 (reference 0% decline ∼ stable kidney function) HR for death 14.2 (reference 0% decline ∼ stable kidney function) |
| Inker [17] | Design: Patient-level meta-analysis Participants: 9488 participants from 37 randomized controlled trials Studies categorized into 5 types of interventions: RAS blockade versus control; RAS blockade versus calcium channel blocker; intensive blood pressure control; low-protein diet; and immunosuppressive therapy Index test: Alternative end points of lesser declines in eGFR (ranging from 30 to 40%) over a baseline period ranging from 12 to 24 months and throughout study duration Reference test: Established composite end point of ESKD, eGFR < 15 mL/min/1.73 m2 if baseline eGFR > 25 mL/min/1.73 m2, or doubling of serum creatinine Median follow-up: 3.62 years Analysis: Agreement between the established and alternative end points was calculated by Bayesian mixed models. | Treatment effects attenuated for lesser declines, particularly for 20 and 30% when compared with the established end points (suggestive of a weaker treatment effect). For the intervention of RAS blockade versus control, treatment effect increased for 30 and 40% declines at shorter interval (suggestive of a stronger treatment effect). For the intervention of low-protein diet, treatment effect increased for all alternative end points (suggestive of a stronger treatment effect). This result further amplified at shorter intervals. |
| Greene [16] | Design: Simulation study Index test: Alternative composite end point based on ESKD and either 30 or 40% eGFR decline Reference test: Established composite end point of ESKD or 57% eGFR decline Analysis: Comparison of the risk of type 1 errors for established and alternative end points. | Compared with a 57% eGFR decline, a 40% eGFR decline resulted in >20% reduction in the required sample size for a 2-year study. Use of 30% eGFR declined reduced the required sample size only in the absence of an acute treatment effect on eGFR. |
CKD, chronic kidney disease; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration equation; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; HR, hazard ratio; IQR, inter-quartile range; RAS, renin–angiotensin system.
![Characteristics of an ideal surrogate end point. The intervention has mechanisms of action through the causal pathway of the disease process (modified from reference [48] with permission).](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/ndt/31/9/10.1093_ndt_gfv269/3/m_gfv26901.jpeg?Expires=1750202023&Signature=td-BkqEGW9mPflH-3X85Og4yw-ASG0aSJMN9ifC8MGHBYdDZ14rFMml5HKryZ7RtiTDiShKySiESu~OfMapUOvhYdDpXIXkEXN43fU6sfUDg497PFrizkhx10PAa0VGC8vjnQcEPFSZ5XwF~558vIA9sUTTPg3NXrWeKkrjA8G3dmpgaV00I5gKyDFaL8cavNF5lIYjVjDgALpXNNaziqyENi2ud~WExR9ddZcS-q6UjXvH7eVdXXjnicYKArwdbFPUxSz33LH8oHIqP6qqBh9ZlnjyViVt4eDt1qDV81jd~bExhpkdrbq5-NCJJx18GHOeJfAzy8FTJDxq4Ec9SDQ__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Characteristics of an ideal surrogate end point. The intervention has mechanisms of action through the causal pathway of the disease process (modified from reference [48] with permission).
Association between lesser GFR decline and clinical end points
Coresh et al. [15] reported correlations of various magnitudes of eGFR decline (ranging from 57% decline to stable eGFR to eGFR increase) over the baseline period of 1–3 years with the risks of ESKD and death during subsequent years using patient-level data on 1.7 million individuals from 35 observational cohort studies and randomized controlled trials participating in the CKD Prognosis Consortium. As the magnitude of eGFR decline decreased from 57 to 40% and 30%, the hazard ratio estimates for 2-year baseline period for ESKD progressively decreased (32.1, 10.2 and 5.4, respectively). However, the cumulative event prevalences increased (0.79, 3.2 and 6.9%, respectively), leading to the percentage population attributable ESKD risk peaking around 40 and 30% eGFR declines at 31 and 44%, respectively. It should be noted that although there were substantial differences in event rates, cumulative prevalences and percentage population attributable ESKD risks between the lower (average 48 mL/min/1.73 m2) and higher (average 92 mL/min/1.73 m2) baseline eGFR strata, associations between the surrogate and clinical end points were ‘qualitatively’ similar. Associations between various lesser eGFR declines and death were also similar to those observed with the clinical end point of ESKD.
In another meta-analysis involving 9488 participants from 37 randomized trials, Lambers Heerspink et al. [18] reported associations between eGFR decline at 12 months (57, 40 or 30%) and time to first occurrence of ESKD (treatment with dialysis or transplantation) or the composite end point of ESKD, untreated kidney failure (defined as eGFR < 15 mL/min/1.73 m2 in those with a baseline eGFR > 25 mL/min/1.73 m2) and serum creatinine doubling. Similar to the meta-analysis by Coresh et al., as the eGFR decline criterion was progressively reduced from 57 to 40% to 30%, cumulative event prevalences increased (0.5, 7.8 and 16.1%, respectively), while the hazard ratios for the composite end point (73, 20.3 and 9.6, respectively) decreased, although they remained high enough to preserve their statistical and clinical significance. However, unlike the Coresh meta-analysis, there was no effect modification by baseline eGFR on meta-regression. Furthermore, Lambers Heerspink et al. did not report subgroup analyses according to baseline eGFR (cut-off 60 mL/min/1.73 m2). A secondary analysis that included death in the composite end point showed a similar association. The results of sensitivity analyses adjusting for baseline eGFR, proteinuria and treatment assignment were not different suggesting that the prognostic significance of a lesser eGFR decline is similar across different levels of proteinuria.
These two studies showed that the surrogate end points of 30 and 40% eGFR declines have statistically and clinically significant associations with the clinical end points of ESKD and death. The studies also demonstrated that these associations differed ‘quantitatively’ according to the clinical setting, such as baseline eGFR and follow-up duration. Therefore, the validity of the lesser GFR decline end point in one clinical setting may not be generalizable to another clinical setting.
Concordance of treatment effects on the surrogate and clinical end points
Inker et al. examined the concordance of treatment effects between various eGFR declines (57, 40, 30 and 20%) and ESKD or a composite end point of ESKD, untreated kidney failure and serum creatinine doubling (using the same definitions as Lambers Heerspink above) over variable follow-up durations (12, 18 or 24 months and complete study duration) in a patient-level meta-analysis of 43 trials involving 12 821 participants and five types of interventions [renin–angiotensin system (RAS) blockade versus control, RAS blockade versus calcium channel blocker, intensive blood pressure control, low-protein diet and immunosuppressive therapy] [17]. After a median follow-up of 3.62 years, event rates for 30 and 40% eGFR declines were 41 and 32%, respectively, whereas those of ESKD, eGFR < 15 mL/min/1.73 m2 and serum creatinine doubling were 16, 9 and 16%, respectively. For each of the five interventions, hazard ratios for 57% eGFR decline, ESKD and the composite end point were similar, while those for lesser eGFR declines were progressively attenuated, indicating weaker treatment effects. However, for the low-protein diet intervention, the treatment effect became stronger. Thus, the treatment effect estimates for the lesser eGFR declines varied substantially for different types of interventions and follow-up durations (particularly shorter durations). A similar pattern was reported when death was included in the composite end point. This study importantly highlighted that the validity of the surrogate end point of eGFR decline was ‘intervention-specific’. Therefore, the lesser eGFR decline end point should not be automatically extrapolated to another intervention in that clinical setting if the two interventions differed in (i) their effects on the causal pathways or (ii) their off-target effects or (iii) the magnitude and duration of effects on the causal pathways [48, 49].
Does the use of a lesser eGFR decline reduce the required trial sample size?
Most trials enroll participants with a wide range of kidney disease diagnoses, baseline eGFR values and proteinuria levels. Since a heterogeneous study population will include progressors as well as non-progressors of disease, the assumptions of event rates for sample size calculations are not straightforward. Further complexities are added by the non-linearity of eGFR decline, the acute hemodynamic effect on eGFR, the long-term treatment effect size, the accrual period and the duration of follow-up. Importantly, the evidence emerging from the NKF/FDA scientific workshop meta-analyses suggests that the validity of a lesser eGFR decline end point is potentially modified by all of these factors.
To address these issues, Greene et al. [16] conducted an extensive but elaborate simulation study, which evaluated 3060 scenarios representing the 13 largest CKD progression trials. The index (alternative) test was a composite end point of ESKD or eGFR decline of 30 and 40%. The reference (established) test was a composite end point of ESKD or 57% eGFR decline. The main analysis was the comparison of the risk of type 1 errors for established and alternative end points. The results showed the established end point of ESKD or 57% eGFR decline preserved the low risk of type 1 error. The investigators pre-specified that a type 1 error ≥10% was not acceptable as it could lead to erroneous conclusions of benefit or harm. Substitution of 30% eGFR decline for 57% eGFR decline reduced the required sample while preserving the low risk of type 1 error only in the absence of an acute treatment effect on eGFR (<±1.25 mL/min/1.73 m2). In contrast, the observed increase in the type 1 error for 40% eGFR decline was much smaller than that for 30% eGFR decline. These findings were applicable to a study population with a baseline eGFR of 42.5 mL/min/1.73 m2 for a 2-year study and a baseline eGFR of 67.5 mL/min/1.73 m2 for a 2- or 3-year study. In most scenarios, where the long-term treatment effect was uniform among the fast progressors and non-progressors, the use of a lesser eGFR decline was acceptable. However, if the treatment effect was proportional to the rate of eGFR decline, a larger effect was observed on the slopes of fast progressors. In this scenario, the hazard ratio estimates attenuated since the end points occurred predominantly among the fast progressors at larger eGFR declines than lesser declines. Greene et al. acknowledged that despite an extensive evaluation, uncertainty usually remained.
Limitations of a lesser eGFR decline end point
While a lesser eGFR decline-based end point may reduce the required sample size and the duration of trial follow-up, investigators and clinicians should carefully examine the following points.
Non-GFR determinants of serum creatinine
Any eGFR-based kidney end point will be problematic for interventions that are likely to have undesired effects on non-GFR determinants of serum creatinine. For example, the Bardoxolone Methyl Evaluation in Patients with Chronic Kidney Disease and Type 2 Diabetes Mellitus: the Occurrence of Renal Events (BEACON) Study, showed a statistically and clinically significant difference in eGFR of 6.4 (95% CI 5.9–6.9) mL/min/1.73 m2 between the bardoxolone methyl and placebo groups (P < 0.001) [30]. However, this improvement in eGFR did not translate into a reduction in the risk of ESKD, and may have resulted from a loss of muscle mass.
Clinical setting in which the trial is conducted
The association between a lesser eGFR decline and subsequent risk of ESKD depends on clinical factors, including baseline eGFR, duration of follow-up, acute effect on eGFR, episodes of acute kidney injury and proportional or uniform treatment effect on progressors and non-progressors. Therefore, the applicability of a lesser eGFR decline end point is not uniform across all clinical settings and interventions. It should be considered on a case-by-case basis.
eGFR decline is not a clinical end point
It should be always kept in mind that a lesser eGFR decline is a surrogate end point, not a clinical end point. In an ideal scenario, the surrogate end point lies in the causal pathway of the disease process that is impacted by the intervention and all the treatment effects on the clinical end point are mediated via the surrogate end point (Figure 1). If the intervention has any unfavorable ‘off-target’ effects, it significantly reduces the validity of the surrogate end point. For example, in the BEACON Study, despite a significant improvement in eGFR, significantly more patients in the bardoxolone methyl group were hospitalized for heart failure or died from heart failure (HR 1.83, 95% CI 1.32–2.55, P < 0.001), indicating an unfavorable ‘off-target’ effect [30].
eGFR is not a patient-reported outcome
The end point of lesser eGFR decline is based on a biomarker. It does not provide any information on important aspects of patients' health status, such as how they feel or function. It does not come directly from patients and needs to be interpreted by clinicians. Therefore, its patient-level importance is limited.
Use in the composite end point
The patient-level importance, frequency of events and treatment effect of a lesser eGFR decline are not similar to those of the clinical end points of death and ESKD. Treatment effect size on the surrogate end point may attenuate or accentuate depending on the type of intervention and duration of follow-up. Therefore, at this stage, it is uncertain whether a composite end point of death, ESKD and a lesser eGFR decline will improve trial efficiency.
Long-term data on patient safety
The use of a lesser eGFR decline will generally permit shorter trial durations (typically 2–3 years). Although a major short-term safety signal is likely to be detected (as in the BEACON Study), such trials will not collect important data on long-term patient safety.
SUMMARY
The field of nephrology is significantly limited by the lack of a validated surrogate end point, which would permit smaller, shorter duration, cheaper and more feasible trials of interventions targeting CKD progression with more reliable and rapid translation into clinical practice compared with the current established end points of death, ESKD and serum creatinine doubling. Recent comprehensive analytic work by the NKF-FDA Scientific Workshop has concluded that the end point of a 40% eGFR decline over 2–3 years was broadly acceptable as a substitute for doubling of serum creatinine in CKD progression trials, while a 30% eGFR decline may be an acceptable end point only in the absence of an acute effect on eGFR. However, it must be noted that a lesser eGFR decline is not a ‘one size fits all’ type of surrogate end point and investigators would need to carefully examine whether the intervention has any acute effect on eGFR and whether the treatment effect is proportional to the rate of GFR decline. The take home message is that the decision to use a 30 or 40% eGFR decline as a surrogate end point should be considered on a case-by-case basis. Investigators and clinicians should be aware of the strengths and limitations of clinical evidence based on eGFR decline as an outcome.
CONFLICT OF INTEREST STATEMENT
S.V.B., S.C.P. and D.W.J. have received a project grant from the National Health and Medical Research Council of Australia to conduct a randomized control trial to study the effect of allopurinol on the progression of chronic kidney disease (trial registration: Australia and New Zealand Clinical Trials Registry 12611000791932).
Comments