In-Depth Topic Review

American

Journal of

Nephrology

Published online: October 15, 2014

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Erythropoiesis-Stimulating Agents (ESA) for

Preventing the Progression of Chronic Kidney
Disease: A Meta-Analysis of 19 Studies
AdrianCovic a IonutNistor a Mihaela-DoraDonciu a RalucaDumea a
DavideBolignano b DavidGoldsmith c

Nephrology Department, Faculty of Medicine, University of Medicine and Pharmacy Gr. T. Popa, Iasi, Romania;
CNR-Institute of Clinical Physiology, Reggio Calabria, Italy; c Renal Unit, Guys and St. Thomas NHS Foundation
Hospital, Kings Health Partners, London, UK

Key Words
Chronic kidney disease Anemia
Erythropoiesis-stimulating agents ESA
CKD-progression Meta-analysis Systematic review

Abstract
Background: The effect of anemia correction on kidney
function in chronic kidney disease (CKD) patients remains
unclear. As 1940% of patients with CKD receive an erythropoiesis-stimulating agent (ESA), this is a potentially important consideration. Summary: We conducted a systematic
review and meta-analysis of randomized trials to January 1,
2014 in adult patients with CKD stages 1 to 4. Selection criteria for studies: randomized controlled trials of at least
2 months duration. Patients were allocated to ESA versus
placebo, no treatment, or different ESA doses with the purpose of achieving a higher versus a lower hemoglobin target. The analyzed outcomes were the need for renal replacement therapy, doubling of serum creatinine, change in GFR
(ml/min), mortality and withdrawal of treatment due to adverse events. A total of 19 trials (n = 8,129 participants with
CKD stage 14) were reviewed. There was no difference in
the risk of end-stage kidney disease (RR, 0.97 [CI 0.831.20],
17 trials, 8,104 participants), change in GFR (Mean Difference
[MD] 0.45 [2.21, 1.31], 9 trials, 1,848 participants) or with-

2014 S. Karger AG, Basel

02508095/14/04030263$39.50/0
E-Mail [email protected]
www.karger.com/ajn

drawal of treatment due to adverse events (RR, 1.18 [CI 0.77

1.81], 10 trials, n = 1,958 participants) for patients at higher
hemoglobin (Hb) targets. Furthermore, no statistically significant differences in mortality (Risk Ratio [RR] 1.10 [CI 0.90
1.35], 16 trials, n = 8,082 participants) were observed. Key
Messages: There is no evidence that ESA treatment affects
renal function in patients with CKD. Use of these agents
should not therefore be influenced by considerations about
2014 S. Karger AG, Basel
influencing CKD progression.

Introduction

Chronic kidney disease (CKD) is a worldwide health

burden, affecting about 1015% of the Western adult
population [1]. It is associated with premature death, cardiovascular disease, infection and cancer; it also consumes many valuable healthcare resources [25]. Prevention of CKD progression therefore is pivotal. Control of
blood pressure (BP) by blockade of the renin-angiotensin-aldosterone system (RAAS) is the cornerstone of current CKD preventive treatment, delaying both cardiovascular and renal disease progression [6, 7]. Alternative
nephroprotective therapies have yet uncertain values
(statins, pentoxyphillin, bicarbonate) [8]. Despite the
Dr. Ionut Nistor
University of Medicine and Pharmacy Gr. T. Popa
Nephrology Department, Faculty of Medicine
University Street no 16, RO700115 Iasi (Romania)
E-Mail [email protected]

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Methods
The systematic review and meta-analysis was performed according to a previously published protocol (http://www.crd.york.
ac.uk/PROSPERO/display_record.asp?ID = CRD42014007162).
We searched MEDLINE (January 1966 to 1st of January 2014)
and the Cochrane Controlled Clinical Trials Register Database
(through Issue 1 of 12, January 2014); we also hand searched reference lists of textbooks, articles, and scientific proceedings for relevant articles (online suppl. table1; for all online suppl. material,
see www.karger.com/doi/10.1159/000366025).
We included all RCTs and quasi-RCTs with a study period of
at least 2 months duration and assessed the effects of different hemoglobin (Hb) targets in pre-dialysis patients with anemia of CKD
on the prevention of the progression of CKD. The first period of
randomized crossover studies was also considered for inclusion.
There was no language restriction. The higher Hb target could be
achieved by EPO ( or ) or darbepoetin; the lower target could be
achieved by lower doses of the same drugs or by a placebo or no
treatment or blood transfusion.
Studies enrolling any patient with CKD stages 14 (as defined
by the Kidney-Disease Outcomes and Quality Initiative [K-DOQI]
guidelines: stage 1 = GFR 90 ml/min/1.73 m2; stage 2 = GFR 60
89 ml/min/1.73 m2; stage 3 = GFR 3059 ml/min/1.73 m2; stage 4=
GFR 1529 ml/min/1.73 m2; stage 5 = GFR <15 ml/min/1.73 m2
not requiring dialysis) were included. We excluded studies enrolling patients with CKD stage 5 (GFR <15 ml/min/1.73 m2) and who
were receiving hemodialysis or other forms of dialysis including
peritoneal dialysis. We also included studies on kidney transplanted patients if participants had residual GFR >15 ml/min/1.73 m2.

264

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Data on the effects of ESAs on the following outcome measures

were analyzed:
(1) Primary outcome:
(a) Need for renal replacement therapy (RRT);
(2) Secondary outcomes:
(b) Doubling of serum creatinine;
(c) A composite outcome death or need for RRT (whichever
came first) since numerous authors [16] argue that the main outcome for CKD is in fact death [17, 18], and only CKD survivors
reach dialysis in order to verify the previous result.
(d) Renal function measured as end-of-treatment GFR (ml/
min or ml/min/1.73 m2) and serum creatinine value at the end of
the study (mg/dl or mol/l); change in GFR (ml/min);
(e) End-of-treatment BP systolic BP and diastolic BP (mm
Hg);
(f) Adverse events (cardiovascular events, myocardial infarction and stroke, or any other serious adverse events as defined by
the authors of the included studies);
(g) End-of-treatment urinary albumin/protein excretion (24 h
proteinuria or 24 h albuminuria in g/dl);
(h) All-cause or cardiovascular-mortality (online suppl. fig.5).
Data extraction was carried out independently by the reviewers, using standard data extraction forms. It was planned that studies reported in non-English language journals would be translated
before assessment. It was also planned that where more than one
publication of one trial existed, reports would be grouped and the
most recent or complete dataset would be used. Disagreements
between reviewers were resolved in consultation with AC.
The quality of included studies was assessed independently by
(IN) and (DG) without blinding to authorship or journal according to recommendations from the Cochrane Collaboration [19].
The quality items assessed were selection bias (random sequence
generation, allocation concealment), performance bias (blinding
of patients and investigators), detection bias (blinding of outcome
assessors), attrition bias (incomplete outcome data), reporting bias
(selective reporting) and other forms of bias (significant different
group comparisons, funding sources, early termination of a trial).
We considered the RCT to be at a high risk of selective outcome
reporting when investigators did not report data for all-cause mortality, need for renal replacement therapy, doubling of serum creatinine and adverse events or patient symptoms.
For dichotomous outcomes (all-cause mortality, number of patients starting RRT), the results were expressed as relative risks
(RR) with 95% confidence intervals (CI). Data were pooled using a
random-effects model, but the fixed-effects model was also analyzed to ensure robustness of the model chosen and susceptibility
to outliers. Where continuous scales of measurement were used to
assess the effects of treatment (serum creatinine value at the end of
the study, GFR, BP), the mean differences and its 95% CI were used.
Heterogeneity was analyzed with a chi-squared test on N-1 degrees of freedom, with an alpha of 0.05 used for statistical significance and with the I2 test [20, 21]. An I2 value of >50% indicates
statistically large heterogeneity among the included studies. If substantial statistical heterogeneity were noted (I2 >50%), we planned
to explore individual study characteristics and those of subgroups
of the main body of evidence. Plausible reasons for variations in
treatment effect (heterogeneity) were explored using subgroup
analysis specifically for the length of follow-up, and type of ESA
used and number of participants. All analyses were performed using Revman 5.1 (2011, The Cochrane Collaboration, UK).

Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

currently available therapy, CKD patients remain at a

highly increased risk for progression toward end-stage
renal disease and concomitant cardiovascular complications [9].
The effect of anemia correction on kidney function
and progressive GFR loss is an under-researched topic. In
this regard, the available literature is contradictory, with
some researchers suggesting that correction of anemia
with ESAs is beneficial for preventing further renal functional decline, while others suggest potential detrimental
effects [10, 11].
Recently, comprehensive meta-analysis focusing on
the impact of anemia treatment on mortality and cardiovascular impacts of anemia treatment was published [12].
They did not, however, focus on or report conclusive information on the impact of anemia treatment on renal
function over time. Since some recent studies [1315]
have suggested that anemia treatment has a beneficial impact on renal function, we decided to perform a systematic analysis of the existing literature in order to establish
if different degrees of anemia correction (low or high Hb
targets) by ESAs may have different impact on renal function trajectories in CKD patients.

5,256 of records identified

through database searching:

1 of additional records
identified through other
sources

4,188 results: Medline

1966 to January 2014
1,068 results:
Cochrane database of
systematic reviews to
issue 1 of 12, January 2014

4,127 of records after

duplicates removed
3,808 of records excluded:
Not RCTs
Other intervention
4,127 of records
screened

Other population
Surrogate outcomes
Follow-up less than 2 months

300 of full-text articles

excluded with reasons:
108 full-text articles
excluded non-RCT;
319 of full-text articles
assessed for eligibility

19 studies included in
qualitative synthesis

citations retrieved by individual searches

and number of trials included in the systematic review.

Results

Search Results
The combined search of MEDLINE, CENTRAL databases, and hand search by experts identified 5,257 citations, of which 3,808 were excluded (non-RCTs or studies
evaluating other interventions not relevant to this review).
Full-text assessment of 319 potentially relevant articles resulted in the identification of 19 eligible trials enrolling a
total of 8,129 patients [10, 11, 1315, 2235] (fig.1).
ESA for Prevention of CKD-Progression

94 full-text articles excluded

not appropriate intervention;
58 full-text articles excluded
for being duplicates.

19 studies included in
quantitative synthesis
(meta-analysis)

Characteristics of the Included Studies

Baseline studies characteristics are presented in table1. The population sample sizes varied between 88 and
4,038 patients [10, 35]. Study duration ranged from twelve
weeks [2224], to 3.3 years [30]. The mean of the Glomerular filtration rate at the trial outsets spanned from 16
to 51 ml/min 1.73 m2. Baseline albuminuria/proteinuria
ranged from 0.5 to 3.1 g/day. Only two studies did not
include patients with diabetes mellitus [10, 24]. One study
included only renal transplant recipients [15].
Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

265

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Fig. 1. Flowchart showing the number of

40 full-text articles excluded

non-relevant outcome;

266

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Watson
1990 [23]

Clyne
1992 [24]

Revicki
1995 [25]

Kleinman
1989 [22]

Study

No

Diabetic
nephropathy:
not reported

Diabetes
mellitus:
exclusion
criterion

Diabetic
nephropathy:
9%

Diabetic
nephropathy:
57%

Baseline
kidney
disease

Quality of life, Allcause mortality,

cardiovascular
mortality, Adverse
events
All patients received
oral iron supplements
not to exceed 200 mg
elemental iron per day
if their serum iron
decreased below 50 /
dl or the iron/total
iron-binding capacity
decrease <20%

12 months
Arm 1: no ESA
treatment
Arm 2: ESA 50 U/kg/
week 3 to achieve and
maintain HCT >35%

HCT (%)
Arm 1:
26.83.6
Arm 2:
26.84.5

Not
Arm 1:
reported n = 40
Arm 2:
n = 43

Creatinine
clearance:
Arm 1
(no ESA):
10.04.1
Arm 2
(ESA):
10.24.1

Number of patients
starting RRT during
study period, change in
quality of life,
Glomerular filtration
rate at end of study;
Serum creatinine level
at end of study; blood
pressure at end of study;
Number of patients with
new antihypertensive
treatment

3 months
Iron therapy was
available for all
patients is the ferritin
levels were within the
normal range

Arm 1: No ESA
injection, standard care;
Arm 2: Aiming HCT
level of >30% with EPO
intravenously injection
once a week

Not
Arm 1:
reported n = 12
Arm 2:
n = 10

Arm 1:
938
Arm 2:
868

Insulin
creatinine
clearance:
Arm 1
(no ESA):
8.03.0
Arm 2
(ESA):
10.04.0

Renal function
outcomes, serum
potassium levels,
adverse events and
blood pressure control

Renal function
outcomes, adverse
effects, quality of life

Study end points

3 months

Not
Arm 1
reported (no ESA):
n=6
Arm 2
(ESA):
n=5

Not
reported

Cr-DTPA
clearance:
Arm 1:
12.43.5
Arm 2:
13.024.4

3 months

Study
duration,
months

All patients required

Arm 1: Aiming Ht
iron supplementation
level of 38% with
placebo injection
Arm 2: Aiming Ht level
of 38% with rHuEPO
(100 units/kg body
weight, subcutaneous
injection three times per
week for 12 weeks)

Not
Arm 1:
reported n = 7
Arm 2:
n=7

Iron therapy was

Arm 1: placebo
available for all
subcutaneously, thrice
patients
weekly at no less than
48-h intervals, for 12
weeks or until a target
hematocrit of 3840%
was attained
Arm 2: receive 100 U/kg
of r-HuEPO, thrice
weekly at no less than
48-h intervals, for 12
weeks or until a target
hematocrit of 3840%
was attained

Co-intervention in
treatment arm (n),
type of ESA, notes

Intervention

Use of
No. of
ACE
patients
inhibitor
or ARB

HCT (%)
Arm 1
(placebo):
28.2
Arm 2
(ESA): 28.1

Baseline Hb
level, g/l
(mean SD)

Creatinine
values from
3 to 11 mg/dl

Baseline renal
function, ml/
min/1.73 m2
(mean SD)

Table 1. Characteristics of the populations and interventions in the included trials

ESA for Prevention of CKD-Progression

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

267

Furuland
2003 [27]

Gouva
2004 [10]

Kuriyama
1997 [26]

Study

No

Table 1. (continued)

Diabetes
mellitus was
an exclusion
criterion

Diabetic
nephropathy
(type 2):
2415%

Diabetes
mellitus:
Arm 1
(no ESA):
58%
Arm 2
(ESA):
55%

Baseline
kidney
disease

Cockcroft
Gault equation:
Arm 1:
25.79.1
Arm 2:
22.36.0

Arm 1
(means SD):
101.05
Arm 2
(means SD):
101.06

Not
Arm 1:
reported n = 45
Arm 2:
n = 43

The primary end point

was a composite of
doubling of creatinine,
renal
replacement, or death;
hospitalizations,
hypertension, severe
anemia, allergic local or
systemic reactions
Subcutaneous EPO
alpha (Eprex,
Janssen-Cilag)
Arm 1: Aiming Hb
level range of 130 g/l
Arm 2: start
erythropoietin only
when Hb level range
<90 g/l

Protocol:
until all
patients had
completed
2years of
follow-up.
Because
of EPO alpha
withdraw,
patients
were followed
for a median
of 22.5
months
(interquartile
range 16 to
24months)

Quality of life, all-cause

mortality,
cardiovascular/non
cardiovascular
mortality, renal
function,
hospitalization,
serious adverse
events,
thromboembolic
events, blood
pressure control
12 months
weeks to
19 months

IV iron to keep
transferrin saturation
>20% and serum
ferritin between
400800 mg/dl;
Epoetin alfa
administered
subcutaneously; In
the low Hb group,
the target level was
obtained with or
without ESA
treatment

Arm 1: Aiming Hb
level 135150 g/l in
females and 145160
g/l in males
Arm 2: Aiming Hb
level 90120 g/l

Not
Arm 1:
reported n = 36
Arm 2:
n = 36

Arm 1:
10610
Arm 2:
1097

Doubling of serum
creatinine, mortality
rates, adverse events,
need for dialysis,
proteinuria level,
diabetes control

9 months

Iron therapy was

under the
responsibility of each
investigator

Arm 1: Aiming HCT

level of 38% with
placebo injection
Arm 2: Aiming HCT
level of 38% with EPO
intravenously injection
once a week for 36
weeks, when HCT
level reached 38%
doses were adjusted to
maintain it at a level of
3335%

24-h urine
collection,
iohexol
clearance or
Cr-EDTA
clearance:
Arm 1: 169
Arm 2: 176

Arm 1:
n = 31
Arm 2:
n = 42

Arm 1:
68.0%
Arm 2:
59.0%

HCT (%)
Arm 1:
27.91.8
Arm 2:
272.3

Creatinine
clearance:
Arm 1:
17.17.2
Arm 2:
19.17.2

Study end points

Study
duration,
months

Co-intervention in
treatment arm (n),
type of ESA, notes

Intervention

Use of
No. of
ACE
patients
inhibitor
or ARB

Baseline Hb
level, g/l
(mean SD)

Baseline renal
function, ml/
min/1.73 m2
(mean SD)

268

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Levin
2005 [29]

CHOIR
2006 [31]

10

ECAP
2006 [30]

Roger
2004 [28]

11

Study

No

Table 1. (continued)

Diabetic
nephropathy:
2627%

Diabetic
nephropathy:
46.850.8%

Diabetes
mellitus:
Arm 1: 41%
Arm 2: 35.1%

Diabetes
mellitus:
Arm 1: 24%
Arm 2: 33%

Baseline
kidney
disease

MDRD Arm 1:
30.310.5
MDRD Arm 2:
28.38.9

MDRD Arm 1:
36.717.0
MDRD Arm 2:
37.117.9

Arm 1:
115.010.0
Arm 2:
116.09.0

Arm 1:
101.09.0
Arm 2:
101.09.0

Arm 1:
91.0%
Arm 2:
87.0%

Arm 1:
73.7%
Arm 2:
74.2%

Arm 1:
n = 192
Arm 2:
n = 193

Arm 1:
n = 715
Arm 2:
n = 717

Rate of GFR decline (by

measuring plasma
iohexol clearance),
need for renal
replacement therapy,
hospitalizations,
cardiovascular and
thrombotic adverse
events, deaths, quality
of life, nutritional
status, blood pressure
control
7.4 months
for group 1
and 8.3
months for
group 2
Epoetin (Eprex)
therapy was
administered
subcutaneously once
per week. Because of
labeling changes for
Eprex, this study was
terminated
prematurely
Arm 1: Aiming Hb level
range (130150 g/l)
Arm 2: Aiming Hb level
range (110120 g/l)

Time to the composite

of death, myocardial
infarction,
hospitalization for
congestive heart failure,
or stroke; time to RRT,
hospitalization by any
cause, quality of life

16 months
Both groups of
patients initially
received epoetin alfa
subcutaneously
weekly; Iron
Arm 1: % = 31.2
Arm 2: % = 29.9

Arm 1: Aiming Hb level

135 g/l
Arm 2: Aiming Hb level
113 g/l

Changes in LVMI,
renal function,
functional cardiac
status, QoL from
baseline to 24 months

24 months

Subcutaneous epoetin
alpha; iron
supplementation as
necessary

Arm 1: Aiming Hb level

range between 120 and
140 g/l
Arm 2: start
erythropoietin only
when Hb level range
<90 g/l

Arm 1:
n = 78
Arm 2:
n = 74

Arm 1:
37.2%
Arm 2:
29.7%

Arm 1:
117.67.6
Arm 2:
117.38.0

Cockcroft
Gault equation:
Arm 1:
37.413.6
Arm 2:
35.113.0
MDRD Arm 1:
29.711.1
MDRD Arm 2:
27.89.3

Changes in LVMi in
2years, renal function,
cardiac performance,
and quality of life,
renal deterioration
was assessed on the
basis of the time to the
onset of renal
replacement therapy,
the calculated
creatinine clearance,
and nuclear estimations
of GFR

24 months

Subcutaneous epoetin
alpha

Arm 1: Aiming Hb level

range between 120 and
130 g/l
Arm 2: start
erythropoietin only
when Hb level range
<90 g/l, aiming for a Hb
level between 90 and g/l

Arm 1:
n = 75
Arm 2:
n = 80

Arm 1:
74.0%
Arm 2:
74.0%

Arm 1
(means SD):
1129
Arm 2
(means SD):
1128

Cockcroft
Gault equation:
Arm 1:
2611
Arm 2:
2511

Study end points

Study
duration,
months

Co-intervention in
treatment arm (n),
type of ESA, notes

Intervention

Use of
No. of
ACE
patients
inhibitor
or ARB

Baseline Hb
level, g/l
(mean SD)

Baseline renal
function, ml/
min/1.73 m2
(mean SD)

ESA for Prevention of CKD-Progression

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

269

Study

CREATE
2006 [11]

Macdougall
2007 [32]

ACORD
2007 [33]

No

12

13

14

Table 1. (continued)

Diabetes
mellitus
type 1 or 2:
100%

Diabetic
nephropathy:
22.023.4%

Diabetes
mellitus:
Arm 1: 27%
Arm 2: 25%

Baseline
kidney
disease

Arm 1: Aiming Hb level

11.01.0 g/dl
Arm 2: Aiming Hb level
11.01.0 g/dl;
Patients in group 1
started epoetin at an
early stage and those
in group 2 were
monitored every
2months until their Hb
concentration had fallen
to a trigger level of
9.0g/dl

Not
Arm 1:
reported n = 65
Arm 2:
n = 132

Arm 1:
108.96.0
Arm 2:
107.66.6

Not
Arm 1
reported
(median
(IQR)): 11.9
(11.312.2)
Arm 2 (median
(IQR)): 11.7
(11.312.0)

Cockcroft
Gault equation:
Arm 1:
25.7512.23
Arm 2:
23.2610.15

Cockcroft
Gault equation
(median
(IQR)):
Arm1:
51 (3967)
Arm 2:
46 (3555)

Arm 1: Aiming Hb level

range of 130150 g/l
Arm 2: Aiming Hb level
range 105115 g/l

Arm 1: Aiming Hb level

range of 130150 g/l
Arm 2: Aiming Hb level
range 105115 g/l

Arm 1:
n = 301
Arm 2:
n = 302

Arm 1:
70.0%
Arm 2:
69.0%

Arm 1
(means SD):
116.06.0
Arm 2
(means SD):
1166.0

Cockcroft
Gault equation:
Arm 1:
24.96.3
Arm 2:
24.26.0

Arm 1:
n = 88
Arm 2:
n = 82

Intervention

Use of
No. of
ACE
patients
inhibitor
or ARB

Baseline Hb
level, g/l
(mean SD)

Baseline renal
function, ml/
min/1.73 m2
(mean SD)

subcutaneous
erythropoietinepoetin beta;

Subcutaneous
erythropoietinepoetin beta; Iron
supplementation
(intravenous or oral)
was recommended at
the discretion of the
investigators

Co-intervention in
treatment arm (n),
type of ESA, notes

left ventricular mass

(LVM), progression of
renal failure, exercise
tolerance, epoetin-a
doses, number of
patients who withdrew
because of starting
dialysis or death
24.110.8
months
for group 1
and 21.110.8
months
for group 2

Primary end point was

change in left
ventricular mass
index (LVMI).
Secondary end points
included
echocardiographic
variables, renal
function, quality of
life, and safety

Time to the first

cardiovascular event,
death from any cause;
congestive heart
failure; hospitalization
for any cause; changes
in left ventricular mass
index, left ventricular
volume, and left
ventricular fractional
shortening; time to
initiation of renal
replacement therapy;
changes in nutritional
status, quality of life,
need for dialysis,
decrease in the
estimated GFR
Mean
follow-up
duration of
36 months
(1,044 days
for arm 1 and
1,092 days
for arm 2)

15 months of
follow-up

Study end points

Study
duration,
months

270

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Akizawa 2011
[37]

NEPHRODIAB2 Diabetic
2011 [14]
nephropathy
(type 2):
5062.8%

17

18

TREAT
2009 [35]

16

Diabetes
mellitus:
Arm 1: 21.1%
Arm 2: 22.5%

Type 2
diabetes
100%

Diabetes
mellitus:
Arm 1: 23.9%
Arm 2: 12.2%

Cianciaruso
2008 [34]

15

Baseline
kidney
disease

Study

No

Table 1. (continued)

Arm 1:
82.6%
Arm 2:
83.8%

Arm 1:
97.7%
Arm 2:
89.1%

Arm 1:
114.78.10
Arm 2:
114.210.6

Cockcroft and
Gault Arm 1:
37.09.1
Cockcroft and
Gault
Arm 2:
36.69.4
MDRD Arm 1:
29.612.8
MDRD Arm 2:
31.814.3
Arm 1:
91.57.9
Arm 2:
91.88.6

Arm 1:
n = 2,012
Arm 2:
n = 2,026

Arm 1:
79.7%
Arm 2:
80.0%

Arm 1:
median 105.0
Arm 2:
median 104.0

MDRD Arm 1:
median 34
MDRD Arm 2:
median 33

MDRD Arm 1:
19.046.96
MDRD Arm2:
18.576.32

Arm 1:
n = 46
Arm 2:
n = 49

Arm 1:
78.2%
Arm 2:
73.4%

Arm 1:
116.07.0
Arm 2:
117.08.0

Cockcroft
Gault equation:
Arm 1:
28.515.6
Arm 2:
29.414.0
MDRD Arm 1:
26.217.4
MDRD Arm 2:
26.714.0

Arm 1:
n = 161
Arm 2:
n = 160

Arm 1:
n = 43
Arm 2:
n = 46

Use of
No. of
ACE
patients
inhibitor
or ARB

Baseline Hb
level, g/l
(mean SD)

Baseline renal
function, ml/
min/1.73 m2
(mean SD)

The number of
patients with and the
occurrence rate of
adverse events, changes
in LVMI, functional
cardiac status, QoL
from baseline to
week 12
The study
lasted 12
months for
both groups
High Hb group
(11.013.0 g/dl)
received darbepoetin
alfa (DA) and the low
Hb group (9.011.0
g/dl) received
recombinant human
erythropoietin
(rHuEPO)
Arm 1: Aiming Hb level
range between 110 and
130 g/l
Arm 2: Aiming Hb level
range 90110 g/l

Change from baseline

in eGFR at 24 months;
mean daily iron use
and mean weekly ESA
use, QOL, critical
events (e.g., death from
all causes, myocardial
infarction, stroke)

24 months
Iron and ESA therapy
modality was under
the responsibility of
each investigator (type
of iron and ESA was
free). ESA therapy
wasnot started unless
serum ferritin was
<200 g/l

Death from any cause,

cardiovascular events,
time to the composite
outcome of death or
end-stage renal disease,
quality of life
Median
follow-up
duration of
29.1 months

Patients in the
placebo group were
assigned
To receive
darbepoetin alfa as a
rescue agent if the Hb
level fell below 9.0 g
per deciliter, with a
return to placebo once
the hemoglobin level
was 9.0 g per deciliter
or higher

Arm 1: Aiming Hb level

range of ~130 g/l
Arm 2: start
erythropoietin only
when Hb level range
<90 g/l

Arm 1: Aiming Hb level

range (110129 g/l)
Arm 2: Aiming Hb level
range (130149 g/l)

Changes in LVMI,
renal function,
functional cardiac
status, QoL from
baseline to 24 months

24 monthsintended;
12-month
analysis
carried out

subcutaneous epoetin
alpha

Arm 1: Aiming Hb level

range between 120 and
140 g/l
Arm 2: start
erythropoietin only
when Hb level range
<90 g/l

Study end points

Study
duration,
months

Co-intervention in
treatment arm (n),
type of ESA, notes

Intervention

Trial Quality
Trial quality was sub-optimal on overall. Allocation
concealment was adequate in three of 19 trials (16%) [29,
32, 34] and unclear in the remaining 10 out of 19 trials
(85%). A consistent high risk of bias concerning the blinding of participants and personnel was found for the majority of included studies, with exception of one, in which
the investigators stated the double-blinded design of the
trial, but without providing any additional information
[23] and two studies with low risk of bias [22, 35]. Only
three of the 19 trials (16%) were analyzed on an intentionto-treat basis [11, 35, 37]. The dropout rate ranged from
0 to 15% and did not differ between the treatment and
control groups (fig.2).

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

271

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

eGFR at the end

of study, the
variation in eCrcl,
doubling of
serum creatinine,
proteinuria,
number of
patients who
progressed to
ESRD, time to
return to dialysis,
graft survival,
occurrence
of acute graft
rejection, patient
survival, QoL
24 months
Arm 1: Aiming Hb level
range of 130150 g/l
Arm 2: Aiming Hb level
range 105115 g/l
CAPRIT 2012
[15]
19

On patients
with posttransplant
anemia only;
nodata about
the incidence
ofdiabetes
mellitus

Arm 1:
n = 63
Arm 2:
n = 62
Arm 1:
69.8%
Arm 2:
71.0%
Arm 1
(means SD):
104.09
Arm 2
(means SD):
106.07
Cockcroft
Gault equation:
Arm 1:
38.910.6
Arm 2:
37.811.0
MDRD Arm 1:
34.410.0
MDRD Arm 2:
33.710.8

Subcutaneous
erythropoietinepoetin beta

Study end points

Study
duration,
months
Co-intervention in
treatment arm (n),
type of ESA, notes
Intervention
Use of
No. of
ACE
patients
inhibitor
or ARB
Baseline Hb
level, g/l
(mean SD)
Baseline renal
function, ml/
min/1.73 m2
(mean SD)
Baseline
kidney
disease
Study
No

Table 1. (continued)

ESA for Prevention of CKD-Progression

Four studies reported the decline in renal function or

progression of renal disease as the primary endpoint [10,
14, 15, 26]. Only one study reported early termination
due to safety issues [30].
Baseline Hb levels varied from 9.0 [10, 27] to 12.9 g/
dl [14]. In trials performed earlier (between 1990 and
2005), there was a trend toward a higher Hb target level
(in the intervention group the upper Hb level limit was
15.0 g/dl [11, 27, 30, 33]). After 2006, the anemia guideline committee of the KDOQI raised the suggested Hb
cut-offs to 11.0 g/dl for the lower limit (evidence-based
recommendation) and 13.0 g/dl for the upper limit
(clinical practice recommendation) [36]. Following
these guideline modifications, the more recent trials referred to a lower limit for the high Hb target (13 g/dl
[35], 13.5 g/dl [13]), with two notable exceptions: 14.9
g/dl [14] and 15 g/dl [15].
The treatment comparisons in the identified trials
were: (a) ESA versus placebo or no treatment or, (b) different doses of ESA therapy as needed to achieve two
(higher vs. lower) hemoglobin target levels. Doses of ESA
used in the trials were not available in one third of the
study reports (table2).
As for the type of ESA, Erythropoietin alpha was used
in eight trials [10, 2732, 34], erythropoietin beta in three
[11, 15, 33] and darbepoetin alpha in two studies [35, 37].
In old studies, the type of ESA was not specified [14, 22
24, 26].
In the control arm, different treatment strategies
were used. Treatment comparisons included different
types of ESA versus placebo or no treatment unless Hb
levels decreased below a specified value [10, 11, 2224,
26, 2830, 3235], or different doses of ESA [14, 15, 31,
37].

Table 2. Baseline, target and achieved hemoglobin concentrations for the included studies
Baseline Hgb
concentration, g/dl

Target Hgb concentration, g/dl

Achieved Hgb concentration, g/dl

high/active group

low/no treatment

high/active group

low/no treatment

Kleinman et al., 1989 [22]

HCT of 28.2%

HCT of 3840%

HCT of 3840%

HCT of 35.8%

HCT of 28.3%

Watson et al., 1990 [23]

HCT of 282%

HCT of 38%

No target

HCT of 352%

HCT of 262%

Clyne et al., 1992 [24]

8.60.8

HCT of >30%

No target

11.71.1

9. 41.0

Revicki et al., 1995 [25]

HCT of 26.83.6%

HCT of 35%

No target

Not reported

8.91.2

Kuriyama et al., 1997 [26]

HCT of 27.91.8%

HCT of 3335%

HCT <30%

11.831.33

8.430.63

Furuland et al., 2003 [27]

10.61

13.015.0

9.012.0

14.31.1

11.31.3

Gouva et al., 2004 [10]

9.011.6

13.0

Unless Hb <9.0

12.5

12.5

Roger et al., 2004 [28]

11.013.0 to 10.012.0 12.013.0

Unless Hb <9.0 at 2
consecutive clinic visits
or Hb <8.0 at any visit

12.20.7

11.01.0

Levin et al., 2005 [29]

11.013.5

12.014.00.5 but not >14 Unless Hb <9

Not reported

11.41.2

CHOIR 2006 [31]

<11.0

13.5

11.3

Not reported

Not reported

ECAP 2006 [30]

Men: <13.0
Women: <12.5

13.015.0

11.012.0

16.2

16.1

CREATE 2006 [11]

11.012.5

13.015.0

10.511.5

Not reported

Not reported

Macdougall et al., 2007 [32] 11.00.5

11.01

Unless Hb <9.0 for

maintaining Hb level
at 11.01

Not reported

Not reported

ACORD 2007 [33]

13.015.0

10.511.5

13.015.0

>10.511.5

Cianciaruso et al., 2008 [34] 11.60.7

12.014.00.5

Unless Hb <9.0

Not reported at the

end of the study

Not reported at the

end of the study

TREAT 2009 [36]

<11.0

>13.0

Unless Hb 9.0

12.5 (IR, 12.012.8)

10.6 (IR, 9.911.3)

NEPHRODIAB2 2011 [14]

10.012.9

13.014.9

11.012.9

<12.0

>13.0

Akizawa et al., 2011 [37]

<10.0

11.013.0

9.011.0

Not reported (achieved Not reported

target range)

CAPRIT 2012 [15]

10.50.8

13.015.0

10.511.5

13.11.7

Not reported

11.41.0

Color version available online

Study

Random sequence generation (selection bias)

Allocation concealment (selection bias)
Blinding of participants and personnel (performance bias)
Blinding of outcome assessment (detection bias)
Incomplete outcome data (attrition bias)
Selective reporting (reporting bias)
Other bias
0%

Fig. 2. Risk of bias graph available accord-

Low risk of bias

25%

Unclear risk of bias

50%

75%

100%

High risk of bias

272

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

ing to recommendations from the cochrane collaboration.

Reduction in GFR at the End of the Study. Doubling

of Serum Creatinine
Nine studies reported changes in GFR, and did not reveal a significant difference between the two groups (9
trials, 1,828 participants, Mean Difference (MD) 0.45;
95% CI 2.211.31 ml/min), although for this outcome
measure, the heterogeneity was important (2 = 302.4;
I2= 97%) (fig.4, table3).
Doubling of serum creatinine was reported only by
three small studies (n = 286 participants) [10, 15, 26].
There was a statistically significant difference indicating
a 47% reduction in the risk of doubling of the serum
creatinine when aiming at higher Hb values (fig.3, table3).
Serum Creatinine or GFR Value at the End of the
Study
There was no difference between the high and the low
Hb target in serum creatinine values or GFR (ml/min)
atthe end of the study. This analysis included four studies
for the creatinine outcome and 8 studies for the GFR outcome (fig.4); an important heterogeneity was found for
both outcomes (2 = 15.29; I2 = 80% and 2 = 29.90; I2 =
77%).
Proteinuria at the End of the Study and Changes in
Proteinuria from Baseline
Proteinuria was reported only by three studies [14, 30,
33]. One study [33] reported that there was no change in
urine protein, a more recent study [14] reported no statistically significant difference between baseline proteinuria and end-of-treatment proteinuria and the oldest one
[30] reported the worsening of proteinuria in a dichotomous format with more events in the low Hb group (10
events vs. 4 events, n = 195 participants per arm).
ESA for Prevention of CKD-Progression

Blood Pressure at the End of the Study. Changes in

BP
Targeting for a higher Hb level was not associated
with higher systolic or diastolic BP levels (5 studies; 934
participants; MD = 1.83; 95% CI 0.794.45 mm Hg).
Heterogeneity across these 5 studies was moderate (2 =
9.33; I2 = 57%). However, the CREATE study [11] mostly contributes (32%) to the overall weight of this analysis
(fig.4).
Adverse Effects
The number of participants who discontinued the
treatment due to adverse events was reported by ten studies as 1,958 participants, and these studies did not reveal
any statistically significant difference between the two
comparison groups (fig.3, table3). Also, there was no difference for the number of patients experiencing at least
one adverse event, including one serious event during the
intervention period (RR = 1.01, 95% CI 1.001.03, 8 studies, 7,108 participants). Heterogeneity was low for both
the analyses (I2 = 0%).
However, there was an 18% higher risk of having at
least one serious adverse event during the intervention
period for higher as compared to lower Hb targets (6
studies, 6,268 participants, RR 1.18, 95% CI 1.011.07).
Finally, the risk of having at least one hospitalization
during the intervention period was 11% higher in the
higher-Hb target (3 studies, 1,910 participants, RR = 1.11,
95% CI 1.001.23; I2 = 0%) (fig.3).
Investigation for Sources of Heterogeneity
Heterogeneity in the effects of lower or higher Hb targets on renal function at the end of treatment (end of
study GFR and change in GFR from baseline) was explored through sub-group analyses based on the type of
ESA used (erythropoietin alpha, erythropoietin beta, darbopoetin alpha or undefined ESA), number of participants (less or more than 100) and study duration (less or
more than 12 weeks) (online suppl. fig.1). Causes of heterogeneity were not explored for the serum creatinine
outcome since few studies looked at this endpoint. The
analysis showed that the type of treatment, number of
participants and length of follow-up did not change the
effects of anemia corrections at different Hb targets on
renal function. In addition, a subgroup sensitivity analysis, with or without data available from the TREAT 2009
trial [35] (fig.3) and with or without data available from
the CAPRIT trial [15] (transplanted patients), was performed for all dichotomous and continuous outcomes
that included this study in the analysis (online suppl. fig.2
Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

273

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Quantitative Analysis
Number of Patients Starting RRT during the Study
Period
With regard to the number of patients that required
the initiation of RRT during the study, there were no data
that would favor either of the two groups (low versus high
Hb target) (17 trials, 8,104 participants, RR 0.97; 95% CI
0.831.15). The heterogeneity for this outcome was moderate: (2 = 28.59, I2 = 44%) (fig.3, table3).
For the composite outcome all-cause mortality or
RRT, there was no statistically significant difference between the high Hb and the low Hb arm (17 trials, 8,104
participants, RR = 0.95; 95% CI 0.811.12); heterogeneity
was again moderate (2 = 38.23, I2 = 58%) (fig.3).

Favours high-Hb group

Events
Total

Low-hemoglobin group
Events
Total

1.2.1 Number of patients starting RRT during study period

Clyne 1992
1
2
12
Revicki 1995
16
13
43
Kuriyama 1997
14
20
42
Furuland 2003
2
0
36
Roger 2004
24
15
75
Gouva 2004
10
18
45
Levin 2005
11
8
78
CHOIR 2006
155
134
715
ACORD 2007
2
3
88
CREATE 2006
127
111
301
ECAP 2006
3
1
195
Macdougall 2007
30
63
64
Cianciaruso 2008
4
4
46
TREAT 2009
338
330
2,012
NEPHRODIAB2 2011
2
5
46
Akizawa 2011
21
26
161
CAPRIT 2012
3
13
63
Subtotal (95% CI)
4,022
763
766
Total events
Heterogeneity: Tau2 = 0.03; Chi2 = 28.59; df = 16 (p = 0.03); I2 = 44%
Test for overall effect: Z = 0.32 (p = 0.75)

1.2.4 Doubling of serum creatinine

22
42
Kuriyama 1997
Gouva 2004
2
63
CAPRIT 2012
6
45
Subtotal (95% CI)
150
30
Total events
Heterogeneity: Tau2 = 0.08; Chi2 = 2.91; df = 2 (p = 0.23); I2 = 31%
Test for overall effect: Z = 2.41 (p = 0.02)
1.2.5 Number discontinued due to adverse events
Kleinman 1989
1
7
Watson 1990
2
5
Revicki 1995
1
43
Furuland 2003
5
36
Levin 2005
1
85
CREATE 2006
17
301
Macdougall 2007
2
65
ECAP 2006
6
195
Cianciaruso 2008
5
46
Akizawa 2011
7
161
Subtotal (95% CI)
944
Total events
47
Heterogeneity: Tau2 = 0.00; Chi2 = 8.27; df = 9 (p = 0.51); I2 = 0%
Test for overall effect: Z = 0.76 (p = 0.44)

26
10
11
47

0
0
3
1
3
10
8
6
5
7
43

Risk ratio
MH, Random, 95% CI

Year

0.5%
5.5%
7.0%
0.3%
5.9%
4.8%
3.1%
15.6%
0.8%
15.9%
0.5%
11.6%
1.4%
18.1%
1.0%
6.4%
1.7%
100.0%

0.42 [0.04, 3.95]

1.14 [0.63, 2.07]
0.52 [0.31, 0.85]
5.00 [0.25, 100.63]
1.71 [0.97, 3.00]
0.53 [0.28, 1.02]
1.30 [0.56, 3.06]
1.16 [0.94, 1.43]
0.62 [0.11, 3.62]
1.15 [0.94, 1.40]
3.00 [0.31, 28.59]
0.98 [0.72, 1.35]
1.07 [0.28, 4.01]
1.03 [0.90, 1.18]
0.37 [0.08, 1.83]
0.80 [0.47, 1.37]
0.23 [0.07, 0.76]
0.97 [0.83, 1.15]

1992
1994
1997
2003
2004
2004
2005
2006
2006
2006
2007
2007
2008
2009
2011
2011
2012

0.8%
7.5%
9.1%
0.5%
8.0%
6.7%
4.5%
16.3%
16.5%
1.3%
0.8%
13.4%
2.2%
1.6%
8.5%
2.6%
100.0%

0.42 [0.04, 3.95]

1.14 [0.63, 2.07]
0.52 [0.31, 0.85]
5.00 [0.25, 100.63]
1.71 [0.97, 3.00]
0.53 [0.28, 1.02]
1.30 [0.56, 3.06]
1.16 [0.94, 1.43]
1.15 [0.94, 1.40]
0.62 [0.11, 3.62]
3.00 [0.31, 28.59]
0.98 [0.72, 1.35]
1.07 [0.28, 4.01]
0.37 [0.08, 1.83]
0.80 [0.47, 1.37]
0.23 [0.07, 0.76]
0.94 [0.77, 1.16]

1992
1994
1997
2003
2004
2004
2005
2006
2006
2006
2007
2007
2008
2011
2011
2012

10
40
31
36
43
80
74
717
302
82
195
132
49
2,026
43
160
62
4,082

0.5%
5.4%
7.1%
0.6%
5.8%
5.5%
3.6%
14.6%
14.8%
0.8%
1.6%
10.8%
1.5%
16.9%
2.3%
6.1%
2.1%
100.0%

0.42 [0.04, 3.95]

1.06 [0.60, 1.89]
0.50 [0.32, 0.80]
6.00 [0.76, 47.36]
0.56 [0.33, 0.97]
1.71 [0.97, 3.00]
1.30 [0.49, 2.20]
1.22 [1.03, 1.45]
1.20 [1.02, 1.42]
0.62 [0.11, 3.62]
0.57 [0.17, 1.92]
0.94 [0.70, 1.27]
1.33 [0.38, 4.66]
1.04 [0.96, 1.13]
0.56 [0.22, 1.41]
0.84 [0.50, 1.42]
0.25 [0.09, 0.69]
0.95 [0.81, 1.12]

1992
1994
1997
2003
2004
2004
2005
2006
2006
2006
2007
2007
2008
2009
2011
2011
2012

31
62
43
136

65.1%
10.8%
24.1%
100.0%

0.62 [0.45, 0.87]

0.20 [0.04, 0.86]
0.52 [0.21, 1.29]
0.53 [0.31, 0.89]

1997
2004
2012

7
6
40
36
87
302
132
195
49
160
1,014

2.0%
2.3%
3.7%
4.1%
3.6%
31.2%
7.9%
14.7%
13.3%
17.4%
100.0%

3.00 [0.14, 63.15]

5.83 [0.34, 99.23]
0.31 [0.03, 2.86]
5.00 [0.61, 40.70]
0.34 [0.04, 3.22]
1.71 [0.79, 3.66]
0.51 [0.11, 2.32]
1.00 [0.33, 3.05]
1.07 [0.33, 3.44]
0.99 [0.36, 2.77]
1.18 [0.77, 1.81]

1989
1989
1994
2003
2005
2006
2007
2007
2008
2011

10
40
31
36
80
43
74
717
82
302
195
132
49
2,026
43
160
62
4,082

1.2.2 Number of patients starting RRT during study period_Treat trial excluded
Clyne 1992
1
2
10
12
Revicki 1995
16
13
40
43
Kuriyama 1997
14
20
31
42
Furuland 2003
2
0
36
36
Roger 2004
24
15
80
75
Gouva 2004
10
18
43
45
Levin 2005
11
8
74
78
CHOIR 2006
155
134
717
715
CREATE 2006
127
111
302
301
ACORD 2007
2
3
82
88
ECAP 2006
3
1
195
195
Macdougall 2007
30
63
132
64
Cianciaruso 2008
4
4
49
46
NEPHRODIAB2 2011
2
5
43
46
Akizawa 2011
21
26
160
161
CAPRIT 2012
3
13
62
63
Subtotal (95% CI)
2,056
2,010
425
436
Total events
Heterogeneity: Tau2 = 0.06; Chi2 = 28.59; df = 15 (p = 0.02); I2 = 48%
Test for overall effect: Z = 0.57 (p = 0.57)
1.2.3 Starting RRT and all cause mortality: composite outcome
Clyne 1992
1
2
12
Revicki 1995
16
14
43
Kuriyama 1997
15
22
42
Furuland 2003
6
1
36
Gouva 2004
13
22
45
Roger 2004
24
15
75
Levin 2005
12
11
78
CHOIR 2006
207
170
715
CREATE 2006
158
132
301
ACORD 2007
2
3
88
ECAP 2006
4
7
195
Macdougall 2007
31
68
64
Cianciaruso 2008
5
4
46
TREAT 2009
750
725
2,012
NEPHRODIAB2 2011
6
10
46
Akizawa 2011
22
26
161
CAPRIT 2012
4
16
63
Subtotal (95% CI)
4,022
1,276
1,248
Total events
Heterogeneity: Tau2 = 0.04; Chi2 = 38.23; df = 16 (p = 0.001); I2 = 58%
Test for overall effect: Z = 0.59 (p = 0.56)

Weight

Risk ratio
MH, Random, 95% CI

Color version available online

Study or Subgroup

0.5 0.7
1
1.5 2.0
Favours high-Hb group Favours low-Hb group

274

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Fig. 3. Effect of high Hb target versus low Hb target on dichotomous outcomes: 1.2.1 number of patients starting RRT during study period. 1.2.2 number of patients starting RRT during study period_TREAT trial excluded. 1.2.3 number of patients starting RRT and allcause mortality: composite outcome. 1.2.4 doubling of serum creatinine. 1.2.5 number discontinued due to adverse events.

Favours high-Hb group

Events
Total

Low-hemoglobin group
Events
Total

Weight

Risk ratio
MH, Random, 95% CI

1.2.6 Number of patients with at least one adverse event during the intervention period
CHOIR 2006
607
589
688
10.8%
686
CREATE 2006
279
273
302
7.8%
300
ECAP 2006
144
140
195
1.2%
195
Macdougall 2007
62
126
132
4.3%
65
TREAT 2009
1,880
1,878
2,019
67.2%
2,004
Akizawa 2011
154
153
160
8.3%
161
NEPHRODIAB2 2011
22
27
43
0.1%
46
CAPRIT 2012
41
33
49
0.3%
63
Subtotal (95% CI)
3,588 100.0%
3,520
3,189
3,219
Total events
Heterogeneity: Tau2 = 0.00; Chi2 = 4.38; df = 7 (p = 0.74); I2 = 0%
Test for overall effect: Z = 1.67 (p= 1.10)
1.2.7 Number of patients with at least one hospitalisation during the intervention period
369
715
334
717
94.6%
Gouva 2004
CHOIR 2006
4
45
6
43
0.7%
ECAP 2006
32
195
26
195
4.6%
Subtotal (95% CI)
955 100.0%
955
405
366
Total events
Heterogeneity: Tau2 = 0.00; Chi2 = 1.01; df = 2 (p = 0.60); I2 = 0%
Test for overall effect: Z = 1.97 (p = 0.05)

Risk ratio
MH, Random, 95% CI

Year

1.03 [0.99, 1.08]

1.03 [0.98, 1.08]
1.03 [0.91, 1.16]
1.00 [0.94, 1.07]
1.01 [0.99, 1.03]
1.00 [0.95, 1.05]
0.76 [0.52, 1.11]
0.97 [0.74, 1.26]
1.01 [1.00, 1.03]

2006
2006
2007
2007
2009
2011
2011
2012

1.11 [1.00, 1.23]

0.64 [0.19, 2.10]
1.23 [0.76, 1.98]
1.11 [1.00, 1.23]

2004
2006
2007

Color version available online

Study or Subgroup

0.5 0.7
1
1.5 2.0
Favours high-Hb group Favours low-Hb group

Fig. 3. Effect of high Hb target versus low Hb target on dichotomous outcomes: 1.2.6 number of patients with at least one adverse event
during the intervention period. 1.2.7 number of patients with at least one hospitalization during the intervention period.

Table 3. Summary of findings

Outcome

Trials reporting
more than one
event/total number
of trials included in
the outcome

Number of
patients
included in
the outcome

Relative
effect

95% CI

Median
treatment
duration,
months

Quality of
evidence*

Number of patients starting RRT during study period

Doubling of serum creatinine
Number discontinued due to adverse events
Reduction in GFR at the end of the study, ml/min

17/17
3/3
9/10
9/9

8,104
286
1,958
1,828

0.97
0.53
1.18
0.45

(0.83, 1.15)
(0.31, 0.89)
(0.77, 1.81)
(2.21, 1.31)

18.41
18.5
15.25
21.67

Moderate
Very low
Low
Very low

*Definition of quality of evidence: High quality = additional research is very unlikely to change our confidence in the estimates of
effect; Moderate quality = additional research is likely to have an important impact on our confidence in the estimate of effect; Low quality = additional research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the
estimate; Very low quality = any estimate of the effect is very uncertain.

and 3) showing roughly the same effect. For the TREAT

[35] subgroup sensitivity analysis, the analysis performed
without the inclusion of this trial did not reveal significant effects estimates differences (with TREAT, effect estimates: 0.97 (0.83, 1.15) and without TREAT: 0.94 (0.77,
1.16)) (fig.3). As new data regarding a higher prevalence
of resistance to ESA treatment in US-CKD patients compared to non-US patients has been recently hypothesized
[38], a geographical-subgroup sensitivity analysis was
conducted for the adverse events outcome including a
comparison of relative effect and heterogeneity with studies comprising US participants versus studies with only

non-US participants. There were no significant effect estimates or heterogeneity differences found (online suppl.
fig.4).

ESA for Prevention of CKD-Progression

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

Discussion

275

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Key Findings
Our analysis included 19 studies, enrolling 8,129 participants with CKD stages 14. There was no difference
in risks for end-stage kidney disease, reduction in GFR,
serum creatinine or GFR (ml/min) values, or in protein-

High-hemoglobin group
Mean
SD
Total

Low-hemoglobin group
Mean
SD
Total

Mean difference
Weight IV, Random, 95% CI

Mean difference
IV, Random, 95% CI

1.3.1 Reduction in GFR at the end of the study (ml/min)

75 10.0%
160 13.2%
62 13.4%
41 13.4%
293 12.9%
74
9.1%
132 10.8%
43
7.5%
61
9.8%
941 100.0%

2.10 [0.75, 4.95]

0.10 [0.78, 0.58]
3.60 [3.99, 3.21]
0.70 [0.37, 1.03]
0.50 [0.48, 1.48]
1.70 [5.06, 1.66]
0.86 [3.27, 1.55]
3.60 [7.89, 0.69]
2.00 [0.96, 4.96]
0.45 [2.21, 1.31]

13.8
3
19
7
6.3
12.2
10
7.1

62 12.2%
8 13.2%
293 15.5%
21 11.2%
43 14.4%
43 12.9%
61 14.4%
6
6.2%
537 100.0%

8.50 [13.24, 3.76]

2.00 [6.11, 2.11]
1.10 [1.45, 3.65]
3.00 [2.40, 8.40]
5.80 [9.13, 2.47]
5.80 [10.09, 1.51]
2.10 [1.22, 5.42]
2.60 [12.28, 7.08]
2.24 [5.23, 0.76]

12
9
12
11
11

62 20.6%
8
9.4%
302 31.8%
31 18.1%
61 20.1%
464 100.0%

0.00 [3.88, 3.88]

5.00 [12.39, 2.39]
2.00 [0.16, 3.84]
7.00 [2.55, 11.45]
2.00 [2.00, 6.00]
1.83 [0.79, 4.45]

ACORD 2007
5.5
8.29
85
3.4
9.92
Akizawa 2011
2.3
3.18
160
2.4
3.06
CAPRIT 2012
2.4
1.1
63
6
1.1
Cianciaruso 2008
2
0.9
37
1.3
0.5
CREATE 2006
3.6
6.7
287
3.1
5.3
Levin 2005
7.4
11.1
78
9.1
10
Macdougall 2007
6.96
8.28
65
7.82
7.5
NEPHRODIAB2 2011
5.1
7.8
46
8.7
12.2
Roger 2004
8
9
66
6
8
Subtotal (95% CI)
887
2
2
2
Heterogeneity: Tau = 5.99; Chi = 302.41; df = 8 (p < 0.00001); I = 97%
Test for overall effect: Z = 0.50 (p = 0.62)

Color version available online

Study or Subgroup

1.3.2 GFR at end of study (ml/min)

CAPRIT 2012
36.5
13.2
63
28
Clyne 1992
10
6
11
8
CREATE 2006
18.1
11.5
289
19.2
Furuland 2003
13
10
19
16
Gouva 2004
21.9
9.4
45
16.1
NEPHRODIAB2 2011
26.7
7.8
46
20.9
Roger 2004
19.9
9
66
22
Watson 1990
12.9
8.94
5
10.3
Subtotal (95% CI)
544
Heterogeneity: Tau2 = 13.38; Chi2 = 29.90; df = 7 (p < 0.0001); I2 = 77%
Test for overall effect: Z = 1.46 (p = 0.14)
1.3.3 Diastolic blood pressure at end of study (mm Hg)
CAPRIT 2012
79
10
63
79
Clyne 1992
85
7
12
90
CREATE 2006
79
11
300
77
Furuland 2003
90
6
29
83
Roger 2004
79
12
66
77
Subtotal (95% CI)
470
Heterogeneity: Tau2 = 4.72; Chi2 = 9.33; df = 4 (p = 0.05); I2 = 57%
Test for overall effect: Z = 1.37 (p = 0.17)

20
10
0
10
20
Favours high-hemoglobin
Favours low-hemoglobin

Fig. 4. Effect of high Hb target versus low Hb target on continuous outcomes: 1.3.1 reduction in GFR at the end of the study (ml/min).

1.3.2 GFR at end of study (ml/min). 1.3.3 diastolic blood pressure at end of the study (mm Hg).

276

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

overall quality of the investigational and trial design approaches used to address this key end-point was suboptimal. Finally, data on proteinuria were reported only by
three studies.
Two rather different studies included in our metaanalysis seemed to show protective outcomes for the deployment of ESAs in CKD. These were the studies by
Gouva et al. (2004) [10] and the more recent CAPRIT
study from 2012 [15].
The Gouva study was one of the smallest ones (n = 88)
of the 19 included in the meta-analysis, and was focused
on non-diabetic patients with a wide range of serum creatinine concentrations (26 mg/dl) and mild-to-moderate anemia (Hb 911 g/dl), from 14 clinical centers in
Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

uria, at the end of the study periods when aiming at higher hemoglobin (Hb) targets. Furthermore, no statistically
significant differences in a composite outcome need for
RRT plus mortality (Risk Ratio [RR] 0.95 [CI 0.811.12],
17 trials, n = 8,104 patients) were observed. Withdrawal
of treatment due to adverse events was similar, as well as
the general control of BP, between high and low Hb targets cohorts of patients.
Some important shortcomings in the manner of reporting outcomes deserve mentioning. Only four studies
out of 19 had renal endpoints as primary outcomes; the
remaining studies only described these events as secondary outcomes. The studies did not generally feature or
recruit from rapidly-progressing CKD cohorts. The

Comparison to Other Studies

It was our intention to assess the evidence culled from
publications until 2014 date range. The first comparison
should be done with two early systematic reviews [40, 41],
ESA for Prevention of CKD-Progression

published in 2004, although the evidence was quite limited, 4 trials including 77 patients in one meta-analysis
and 5 trials with 207 patients in the second. Nevertheless,
both meta-analyses showed no difference between Hb
groups regarding the renal progression outcome.
The most meaningful comparison should probably
be made with a 2010 meta-analysis by Palmer et al. [12]
that evaluated the effects of ESA in CKD patients. Our
work has in addition three subsequent published studies
included. Most importantly, our main aim was to focus
on assessing different renal outcomes, which were not
the main output focus of the Palmer paper, whose principal objective was a cardiovascular event/mortality
analysis.
Strengths and Limitations
Our review has a number of strengths and limitations.
Strengths include a systematic search of medical databases, data extraction and analysis, and trial quality assessment by two independent reviewers. We also carefully
analyzed the most important potential renal end-points
including loss of renal function, and dialysis initiation.
This has not been done before. Another strength of our
review was a detailed examination of trials quality and
heterogeneity. The major limitation is the lack of longterm studies analyzing the efficacy of ESA on mortality,
progression of CKD, and entry onto dialysis. Another
limitation is the populations included in the 19 studies,
and how many of them were truly demonstrating progressive loss of GFR prior to enrollment. The majority of
the included studies enrolled a relatively small number of
patients; they were powered to observe differences in surrogate end-points rather than patient-focused outcomes,
and used nonoptimal measures of renal function change
over time. Also, in our aim to study ESA role in CKD progression, our search found limited data for all proposed
outcomes. The absence of data for some relevant outcomes (e.g., doubling of serum creatinine) might represent an open pathway for future research. In addition, the
preponderance of high risk of bias regarding the lack of
blinding could be considered a possible limitation.
Implications for Practice
In CKD patients with anemia, the current KDIGO
[42] and ERBP guidelines [43] suggest that a careful exploration of the risk-benefit relationship between Hb,
ESAs, and outcomes, should be undertaken for all patients. In some cases, it is expected that clinically relevant benefits of using ESAs to raise Hb will be seen, but
not in all cases, while there are definite risks associated
Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

277

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

Greece. It was non-blinded, the intervention being early

versus later use of ESA. To be eligible for inclusion, patients also had to have stopped all ACEI and/or ARBs for
2 months, and to have controlled BP on non-ACE/ARB
therapy. These stringent, unusual, selection criteria make
it very challenging to extrapolate the findings of this
study, since a large proportion of current CKD patients
are on ACE/ARB-based drug therapy for BP control and
proteinuria reduction. During follow-up, 13 (3 deaths)
versus 23 (4 deaths) patients reached the end point of
doubling of creatinine, renal replacement, or death, in the
early versus deferred treatment arms, respectively. Thus,
the difference between the two arms was largely caused
by the initiation of renal replacement (a notoriously subjective, symptom-led, judgment).
The CAPRIT study [15] was a two-year, open-label trial involving 125 post-renal transplant patients. Beside the
well-known multifactorial pathogenesis of post-renal
transplant anemia (including the presence of immunosuppressive therapy and the altered inflammatory environment), the most notable anemia risk factor still remains and that is the loss of renal allograft function [39].
This is consistent with the loss of endogenous EPO production in CKD patients. The findings of the CAPRIT
trial suggested improved allograft survival among patients treated to a higher Hb level.
Indeed, in the CAPRIT, the patient admixture and comorbidity were highly different from those seen in other
anemia studies: <5% of subjects having primary diabetes,
and only around 15% subjects with prior cardiovascular
history. Adverse event rates were very low compared to
CREATE [11], CHOIR [31] and TREAT [35] studies. Additionally, there was little or no information on iron status, or on its therapeutic usage. The ESA dose used,
4,0006,000 IU Epoetin weekly, was about half to a third
seen in the other trials.
Most importantly, there was neither presumption nor
demonstration of declining renal function at study entry;
no renal biopsy information was obtained to differentiate
the immunologically driven loss of GFR from many other potential causes, and no direct measurements of GFR
were undertaken. Patient entry back onto dialysis was inevitably partly subjective, and symptom-based, and so
with higher Hb values was intrinsically less likely to happen.

with active therapy. However, in the quest for nephroprotection to prevent progressive loss of renal function
and the eventual need for dialysis, adjusting the patients Hb concentration using ESAs does not seem,
from our evidence, to have a compelling logic to support it.

Disclosure Statements
Adrian Covic received speaker honoraria from Amgen, Roche,
Fresenius Medical Care and Abbott, and is a member of the
European Renal Best Practices Board.
David Goldsmith received speaker honoraria from Roche,
Amgen, Vifor, Takeda, Sandoz.

Acknowledgements

Role of Funding Source

None to declare.

No funding source was available for this study.

References

278

13 Tsubakihara Y, et al: High target hemoglobin

with erythropoiesis-stimulating agents has
advantages in the renal function of non-dialysis chronic kidney disease patients. Ther
Apher Dial 2012;16:529540.
14 Villar E, et al: Anemia normalization in patients with type 2 diabetes and chronic kidney
disease: results of the NEPHRODIAB2 randomized trial. J Diabetes Complications 2011;
25:237243.
15 Choukroun G, et al: Correction of postkidney
transplant anemia reduces progression of allograft nephropathy. J Am Soc Nephrol 2012;
23:360368.
16 Centers for Disease Control and Prevention:
National Chronic Kidney Disease Fact Sheet:
General Information and National Estimates
on Chronic Kidney Disease in the United
States, GA: U.S. Department of Health and
Human Services, CDC, 2010, National
Chronic Kidney Disease Fact Sheet 2010.
17 Dalrymple LS, et al: Chronic kidney disease
and the risk of end-stage renal disease versus
death. J Gen Intern Med 2011;26:379385.
18 Tonelli M, et al: Chronic kidney disease and
mortality risk: a systematic review. J Am Soc
Nephrol 2006;17:20342047.
19 Higgins JPT, Greeen S (eds): Cochrane Handbook for Systematic Reviews of Interventions
Version 5.1.0. The Cochrane Collaboration,
2011. Available from http://www.cochranehandbook.org.
20 Higgins JP, et al: Measuring inconsistency in
meta-analyses. BMJ 2003;327:557560.
21 Thompson SG, Sharp SJ: Explaining heterogeneity in meta-analysis: a comparison of
methods. Stat Med 1999;18:26932708.
22 Kleinman KS, et al: The use of recombinant
human erythropoietin in the correction of
anemia in predialysis patients and its effect on
renal function: a double-blind, placebo-controlled trial. Am J Kidney Dis 1989; 14: 486
495.
23 Watson AJ, et al: Treatment of the anemia of
chronic renal failure with subcutaneous recombinant human erythropoietin. Am J Med
1990;89:432435.

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

24 Clyne N, Jogestrand T: Effect of erythropoietin treatment on physical exercise capacity

and on renal function in predialytic uremic
patients. Nephron 1992;60:390396.
25 Revicki DA, et al: Health-related quality of life
associated with recombinant human erythropoietin therapy for predialysis chronic renal
disease patients. Am J Kidney Dis 1995; 25:
548554.
26 Kuriyama S, et al: Reversal of anemia by
erythropoietin therapy retards the progression of chronic renal failure, especially in
nondiabetic patients. Nephron 1997; 77: 176
185.
27 Furuland H, et al: A randomized controlled
trial of haemoglobin normalization with epoetin alfa in pre-dialysis and dialysis patients. Nephrol Dial Transplant 2003; 18:
353361.
28 Roger SD: Effects of early and late intervention with epoetin on left ventricular mass
among patients with chronic kidney disease
(stage 3 or 4): results of a randomized clinical
trial. J Am Soc Nephrol 2004;15:148156.
29 Levin A, et al: Canadian randomized trial of
hemoglobin maintenance to prevent or delay
left ventricular mass growth in patients with
CKD. Am J Kidney Dis 2005;46:799811.
30 Rossert J, et al: Effect of early correction of
anemia on the progression of CKD. Am J Kidney Dis 2006;47:738750.
31 Singh AK, et al: Correction of anemia with
epoetin alfa in chronic kidney disease [see
comment]. N Engl J Med 2006; 355: 2085
2098.
32 Macdougall IC, Temple RM, Kwan JT: Is early treatment of anaemia with epoetin-alpha
beneficial to pre-dialysis chronic kidney disease patients? Results of a multicentre, openlabel, prospective, randomized, comparative
group trial. Nephrol Dial Transplant 2007;22:
784793.
33 Ritz E, et al: Target level for hemoglobin correction in patients with diabetes and CKD:
primary results of the Anemia Correction in
Diabetes (ACORD) Study. Am J Kidney Dis
2007;49:194207.

Covic/Nistor/Donciu/Dumea/Bolignano/
Goldsmith

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

1 Collins AJ, et al: United States Renal Data

System 2011 Annual Data Report: atlas of
chronic kidney disease and end-stage renal
disease in the United States. Am J Kidney Dis
2012;59(1 suppl 1):A7, e1e420.
2 Go AS, et al: Chronic kidney disease and the
risks of death, cardiovascular events, and hospitalization. N Engl J Med 2004; 351: 1296
1305.
3 Kerr M, et al: Estimating the financial cost of
chronic kidney disease to the NHS in England.
Nephrol Dial Transplant 2012; 27(suppl 3):
iii73iii80.
4 Fried LF, et al: Kidney function as a predictor
of noncardiovascular mortality. J Am Soc
Nephrol 2005;16:37283735.
5 Iff S, et al: Reduced estimated GFR and cancer
mortality. Am J Kidney Dis 2014;63:2330.
6 Casas JP, et al: Effect of inhibitors of the reninangiotensin system and other antihypertensive drugs on renal outcomes: systematic review and meta-analysis. Lancet 2005; 366:
20262033.
7 Strippoli GF, et al: Angiotensin converting
enzyme inhibitors and angiotensin II receptor
antagonists for preventing the progression of
diabetic kidney disease. Cochrane Database
Syst Rev 2006;4:Cd006257.
8 Gansevoort RT, et al: Chronic kidney disease
and cardiovascular risk: epidemiology, mechanisms, and prevention. Lancet 2013; 382:
339352.
9 Jha V, et al: Chronic kidney disease: global dimension and perspectives. Lancet 2013; 382:
260272.
10 Gouva C, et al: Treating anemia early in renal
failure patients slows the decline of renal
function: a randomized controlled trial. Kidney Int 2004;66:753760.
11 Drueke TB, et al: Normalization of hemoglobin level in patients with chronic kidney disease and anemia. N Engl J Med 2006; 355:
20712084.
12 Palmer SC, et al: Meta-analysis: erythropoiesis-stimulating agents in patients with chronic kidney disease. Ann Intern Med 2010; 153:
2333.

ESA for Prevention of CKD-Progression

38 De Nicola L, et al: Responsiveness to erythropoiesis-stimulating agents in chronic kidney

disease: does geography matter? Drugs 2014;
74:159168.
39 Molnar MZ, et al: Prevalence and management of anaemia in renal transplant recipients: data from ten European centres. Nephron Clin Pract 2011;117:c127c134.
40 Strippoli GF, et al: Hemoglobin targets for the
anemia of chronic kidney disease: a metaanalysis of randomized, controlled trials. J
Am Soc Nephrol 2004;15:31543165.

41 Cody June D, et al: Recombinant human

erythropoietin for chronic renal failure anaemia in pre-dialysis patients. Cochrane Database Syst Rev 2005;3. DOI: 10.1002/14651858.
CD003266.
42 Drueke TB, Parfrey PS: Summary of the
KDIGO guideline on anemia and comment:
reading between the (guide)line(s). Kidney
Int 2012;82:952960.
43 Locatelli F, et al: Kidney Disease: ImprovingGlobal Outcomes guidelines on anaemia
management in chronic kidney disease: a
European Renal Best Practice position statement. Nephrol Dial Transplant 2013;28:1346
1359.

Am J Nephrol 2014;40:263279
DOI: 10.1159/000366025

279

Downloaded by:
198.143.33.33 - 4/6/2016 6:15:54 PM

34 Cianciaruso B, et al: Italian randomized trial

of hemoglobin maintenance to prevent or delay left ventricular hypertrophy in chronic
kidney disease. J Nephrol 2008;21:861870.
35 Pfeffer MA, et al: A trial of darbepoetin alfa in
type 2 diabetes and chronic kidney disease. N
Engl J Med 2009;361:20192032.
36 KDOQI, KDOQI Clinical Practice Guideline
and Clinical Practice Recommendations for
anemia in chronic kidney disease: 2007 update of hemoglobin target. Am J Kidney Dis
2007;50:471530.
37 Akizawa T, et al: Positive outcomes of high
hemoglobin target in patients with chronic
kidney disease not on dialysis: a randomized
controlled study. Ther Apher Dial 2011; 15:
431440.