J Am Soc Nephrol 15: 761769, 2004

N-Acetylcysteine for the Prevention of Radiocontrast

Induced Nephropathy: A Meta-Analysis of Prospective
Controlled Trials
ABHIJIT V. KSHIRSAGAR,* CHARLES POOLE, AMY MOTTL,*
DAVID SHOHAM,* NORA FRANCESCHINI,* GAIL TUDOR,
MALAY AGRAWAL, CINDY DENU-CIOCCA,* E. MAGNUS OHMAN, and
WILLIAM F. FINN*
*Division of Nephrology and Hypertension, School of Medicine, University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina; Department of Epidemiology, School of Public Health, University of North
Carolina at Chapel Hill, Chapel Hill, North Carolina; Department of Biostatistics, School of Public Health,
University of North Carolina at Chapel Hill, Chapel Hill, North Carolina; and Division of Cardiology,
School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina.

Abstract. N-acetylcysteine has been recommended for patients

with renal insufficiency who are to receive radiocontrast media. However, trials of oral N-acetylcysteine for the prevention
of radiocontrast-induced nephropathy have yielded inconsistent results. A systematic review of patient and study characteristics was undertaken to discover possible explanations of
the inconsistencies. The databases MEDLINE, EMBASE, and
CENTRAL (1966 to March 2003) were searched in all languages, and conference proceedings from several professional
societies from the years 1999 to 2003 were also searched. Only
prospective controlled trials of oral N-acetylcysteine were included. Risk difference estimates and 95% confidence intervals
were calculated. The estimates were examined for evidence of
publication bias and heterogeneity. Stratified and meta-regression analyses were used to compare estimates by study and

patient characteristics. Identified were 16 studies, 15 published

and 1 unpublished. There was no evidence of publication bias,
but there was substantial evidence of heterogeneity, thus precluding reliance on a meaningful summary effect estimate.
Meta-regression identified several patient and study characteristics, with some evidence of association with study-specific
estimates. None of these characteristics, however, formed subsets of studies with results that were homogeneous enough to
aggregate. Research on N-acetylcysteine and the incidence of
radiocontrast nephropathy is too inconsistent at present to
warrant a conclusion on efficacy or a recommendation for its
routine use. Identified patient and study characteristics may be
responsible for some, but not all, of this inconsistency. A large,
randomized, placebo-controlled trial, a pooled analysis of patient-level data, or both may resolve this issue.

Over the last 3 yr, enthusiasm for the use of oral N-acetylcysteine to prevent radiocontrast-induced nephropathy has been
growing (1). A breakthrough study (2) first reported a large and
significant reduction in the risk of radiocontrast-induced nephropathy compared with placebo among patients with moderate renal dysfunction. Subsequent studies produced mixed
results (3 6). Recently, a large study (7), arguably with a
relatively high degree of power and sound methods, seemed to
confirm the beneficial effect of this agent. N-acetylcysteine is
now being routinely recommended as preventative therapy, as
an additive to low-osmolar contrast media and intravenous

hydration, for patients with reduced renal function who are to

undergo a planned exposure to radiocontrast media (8,9). Yet
the reasons for discrepant findings among the trials need investigation before the widespread use of N-acetylcysteine can
be recommended.
We therefore performed a systematic review and metaanalysis of available prospective controlled trials to quantify
and compare reported associations of oral N-acetylcysteine
with the incidence of nephropathy after exposure to radiocontrast media. For this review, we examined several key questions. Is there evidence of publication bias? Are there differences among the study results compatible with chance
variation? Are there characteristics of studies or participating
patients associated with meaningful changes in the association
of administration of N-acetylcysteine and the incidence of
radiocontrast-induced nephropathy? And most importantly,
does the existing literature currently support a conclusion regarding the effectiveness of administration of oral N-acetylcysteine to reduce the occurrence of radiocontrast-induced
nephropathy?

Received September 2, 2003. Accepted December 8, 2003.

Correspondence to Dr. Abhijit V. Kshirsagar, CB 7155, 348 MacNider Hall,
Chapel Hill, NC 27599-7155. Phone: 919-966-2561, ext. 259; Fax: 919-9664251; E-mail: sagar@med.unc.edu
1046-6673/1504-0761
Journal of the American Society of Nephrology
Copyright 2004 by the American Society of Nephrology
DOI: 10.1097/01.ASN.0000116241.47678.49

762

Journal of the American Society of Nephrology

Materials and Methods

Inclusion Criteria
The inclusion criteria were determined by the three study authors
(AVK, AM, NF) involved in planning the meta-analysis. Trials were
included in the systematic review only if they were prospective and
controlled. Trials had to include individuals with renal insufficiency,
defined as an average trial serum creatinine (SCr) concentration 1.2
mg/dl or more, or average creatinine clearance less than 70 ml/min per
1.73 m2. These individuals had to be assigned to oral N-acetylcysteine
or to placebo/no intervention before the administration of intravenous
radiocontrast. The comparison groups could not be assigned to other
interventions, such as fenoldopam or theophylline.

Identification of Relevant Trials

Identification of relevant trials was performed by three study authors (AVK, AM, NF). We performed a MEDLINE search of the
literature from 1966 to March 2003 using the terms N-acetylcysteine
and radiocontrast. Searches of EMBASE and the Cochrane Collaborations register, CENTRAL, were also performed between 1974 and
March 2003 using the terms N-acetylcysteine, contrast, and clinical
trial. In addition, we reviewed abstracts from the annual meetings of
the American Society of Nephrology, National Kidney Foundation,
American Heart Association, and Radiologic Society of North America from 1999 to 2002, and from the American College of Cardiology
from 1999 to 2003, to identify studies not yet published at the time of
our literature search. We also reviewed the bibliographies of original
and review articles that investigated N-acetylcysteine. We were also
aware of an ongoing trial at the University of North Carolina examining the effect of N-acetylcysteine on the prevention of radiocontrast
nephropathy directed by three of the contributing authors (MA, CDC,
EMO). The study was initially designed to enroll 100 patients but was
terminated after recruitment of 25 patients for reasons of study personnel. The study authors (MA, CDC, EMO) offered to share results
of their unpublished study after learning of our intention to perform
this systematic review.

Data Analyses
Specific data to be incorporated into the analysis were abstracted
independently from the articles by three researchers (AVK, AM, NF)
without masking. The three researchers extracted the information and
recorded it on preprinted forms. The three researchers then met to
confirm findings and to resolve any differences. We did not test
formally for agreement among the three researchers with a kappa
statistic. The extracted information included the following: definition
of radiocontrast-induced nephropathy, number of patients at initiation
of study, average patient age, prevalence of diabetes and congestive
heart failure, average baseline SCr concentration, hydration protocol,
total hydration volume, weight before and after administration to
assess hydration status, type of radiocontrast medium, average volume
of radiocontrast medium, dose of N-acetylcysteine, timing of N-acetylcysteine administration, change in SCr concentration after 48 h,
incidence of radiocontrast-induced nephropathy (as defined below),
and the proportion of individuals needing dialysis. Information that
was not clearly presented within the body of the article or abstract was
clarified by contacting the primary investigators (5,6,10,11).
The clinical end point investigated was the risk of radiocontrastinduced nephropathy measure at or after 48 h. Radiocontrast induced
nephropathy was defined as (1) a 0.5 mg/dl or more increase in SCr
from baseline after 48 h, or (2) a 25% or more increase in SCr from
baseline after 48 h. Our definition was based on the current literature
and was used by all studies incorporated in the meta-analysis. Phys-

J Am Soc Nephrol 15: 761769, 2004

iologically, the peak elevation of the SCr concentration after administration of radiocontrast media is known to occur after 48 h and
generally begins to drop after 96 h. Furthermore, studies have shown
that a creatinine change of 25% or more has the same mortality as
larger changes and is a sensitive predictor of the future need for
dialysis (12,13).

Statistical Analyses
Evidence of publication bias was investigated by means of funnel
plot, Begg and Mazumdars test (14), and the test of Egger et al. (15)
If there is no publication bias, the expected shape of this plot is a
symmetrical, funnel-shaped distribution of effect estimates above and
below the summary estimate, with the more precise estimates more
tightly clustered and the less precise estimates more widely dispersed
around the summary value. Begg and Mazumdars test is a rank
correlation test of the null hypothesis that the studies estimates are
not associated with their estimated standard errors. The test of Egger
et al. regresses the z-score on the standard error. The expected
intercept of this regression equation is zero when the estimates and
their standard errors are unassociated. Thus, both tests are tests of
funnel plot symmetry.
Heterogeneity of the trials was assessed by means of Cochrans Q
statistic, which has a 2 distribution with degrees of freedom equal to
one minus the number of estimates. When all trials produce unbiased
estimates of the same true population value and their results therefore
differ only by chance, the expected value of this statistic equals the
degrees of freedom (16). The test of homogeneity was conducted for
all studies and also for various subsets of studies, including the trials
that used a randomization scheme and trials that used both randomization and placebo. We also conducted tests of homogeneity for all
studies using relative risk or odds ratio as a means of sensitivity
analysis.
We performed meta-regression analyses to assess the association
between risk difference estimates from the trials and characteristics of
those trials and their participating patients. All study characteristics
were selected a priori as potentially influential. The small number of
trials precluded the use of multivariable meta-regression. Selected
characteristics were: (1) randomized trial (versus nonrandomized); (2)
N-acetylcysteine given the day before (versus day of); (3) placebo
used (versus no placebo); (4) abstract (versus peer-reviewed journal
article); (5) iso-osmotic contrast used (versus low-osmolality contrast
used); (6) average baseline SCr of all study subjects within each trial;
(7) difference in average baseline SCr between treatment and control
groups within each trial; (8) average contrast volume within each trial;
(9) difference in average volume between treatment and control
groups within each trial; (10) average age of subjects within each trial;
(11) difference in average age between treatment and control groups;
(12) average prevalence of diabetes within each trial; (13) difference
in diabetes prevalence between treatment and control groups within
each trial; and (14) the type of saline used, 0.9% or 0.45%. The
meta-regression was conducted both for all studies and, as a sensitivity analysis, for the studies in which randomization to treatment group
was a feature of the study design. We estimated the association
between treatment and outcome by using the risk difference with 95%
confidence intervals calculated for each study, and a summary or
aggregate risk difference by using the Metan program in Stata.

Results
Our search strategy yielded a total of 18 clinical trials of
N-acetylcysteine used to prevent radiocontrast-induced nephropathy (Figure 1). Sixteen studies met the inclusion criteria.

J Am Soc Nephrol 15: 761769, 2004

Figure 1. Flow diagram of search strategy and results.

The two studies were excluded because they used retrospective

controls (17,18). Investigation of publication bias, heterogeneity, and meta-regression was performed on these 16 studies.
We sought further clarification of presented data in four cases
(5,6,10,11).
Tables 1 and 2 list important study and subject characteristics of the 16 prospective controlled trials with a total of 1538
subjects. Twelve of the 16 studies assigned intervention by a
randomization process. Participants in ten studies received a
placebo, whereas participants in six did not receive a placebo.
Nine studies were published in peer-reviewed journals, six
were published as an abstract at the time of our search, and one
study was unpublished. All of the cited were published within
the last 3 yr, with the majority published in the year 2002.
Except for the study by Tepel et al. (2), the indication for
radiocontrast media exposure was for a coronary artery
angiography/intervention.
The median average SCr concentration was approximately
2.0 mg/dl. The average age of the study subjects showed them
to be advanced in age, over age 65 yr. Patients with diabetes
mellitus were included in all studies, with the prevalence
varying between 32% and 68%.
All 1538 individuals, whether in the N-acetylcysteine group
or in the placebo group, were provided intravenous saline
hydration before the administration of radiocontrast media;
0.45% saline, rather than 0.9% saline, was predominantly used.
Seven hundred twenty-one subjects were assigned to receive
oral N-acetylcysteine, and 817 individuals were assigned to
receive a control therapy. Individuals in the N-acetylcysteine
group were given the drug orally at a dose of 600 mg twice a
day 1 d before and on the day of exposure to radiocontrast

N-acetylcysteine and Radiocontrast Nephropathy

763

media in the majority of studies. In the study by Shyu et al.

(19), the dose of N-acetylcysteine was 33% less than the other
studies, whereas in the study by Agrawal (unpublished), the
initial dose of N-acetylcysteine was 33% higher than the other
studies. Finally, in the studies by Durham et al. (3) and DiazSandoval et al. (20), individuals were dosed at different times
than in the other studies.
Figure 2 displays the results of the funnel plots for all 16
trials. There was little evidence of publication bias in these
studies by visual inspection, and by the tests of Begg and
Mazumdar (14) and of Egger et al. (15). Figure 3 is a forest
plot displaying estimated risk difference from each of the 16
controlled trials. In parentheses after the year of publication is
the relative weight, which is one over the within-study variance
expressed as a percentage of the maximum of these values. In
other words, the shortest line is 100 and the longest line is 3.
The point estimates and 95% confidence intervals vary widely
among these studies, as confirmed by the homogeneity test (P
0.001). Roughly one-third of the point estimates suggest a
beneficial effect of N-acetylcysteine, one-third suggest no effect, and one-third suggest a beneficial effect of the control
intervention. Notably, the forest plot graphically demonstrates
the high degree of heterogeneity of study results in a great
degree of nonoverlap among the confidence intervals. We did
not present the aggregate risk difference estimate because of
the high degree of heterogeneity observed among the studies.
Of note, heterogeneity was also present and was highly significant when using relative risk or odds ratio as the primary
outcome measure.
Two sets of univariate linear meta-regressions were performed; the first set used all 16 trials, and is summarized in
Table 2. Most of the study characteristics were associated with
risk difference estimates from the individual studies to an
extent that, although imprecisely estimated, would be clinically
meaningful. For example, how the primary study authors defined radiocontrast nephropathyabsolute versus percentage
change in SCr changed the risk difference by 0.05. The
prevalence of diabetic patients was associated with a change in
the risk difference by 0.06; other notable examples include the
timing measurement of SCr and the use of iso-osmolar radiocontrast media. The level of baseline renal function, as assessed by SCr concentration, did not meaningfully affect the
summary estimate. None of these characteristics of the studies
and patients completely resolved the inconsistencies, however.
There was no clinically significant change in the identified
patient and study characteristics when meta-regression was
limited to only the 11 randomized clinical trials. There was no
difference in the incident need for hemodialysis between the
two groups.

Discussion
This systematic review suggests that currently, there is not a
wholly positive answer to the question: does the use of oral
N-acetylcysteine before radiocontrast media reduce the incidence of acute radiocontrast-induced nephropathy? The analysis of 16 prospective controlled clinical trials (with a total of
1538 subjects) suggests, first and foremost, that there is sub-

764

Journal of the American Society of Nephrology

J Am Soc Nephrol 15: 761769, 2004

Table 1. Summary of outcomes from the 16 prospective controlled trialsa

Study

Year

NAC Dosing

Tepel (2)

2000

600 mg bid 1 day before day of study

Adamian (47)

2002

600 mg bid

Allaqaband (10)

2002

600 mg bid 1 d before and 1 day after

Briguori (4)

2002

600 mg bid 1 d before and day of study

Diaz-Sandoval (20)

2002

600 mg bid 1 dose before and 3 d after

Durham (3)

2002

1200 mg 1 h before and 3 h after

Goldenberg (48)

2003

600 mg tid 1 d before and 1 d after

Kahlon (5)

2002

600 mg bid 1 d before and 1 d after

Oldemeyer (6)

2003

1500 mg bid 4 doses

Shyu (19)

2002

400 mg bid 1 d before and day of study

Vallero (42)

2002

600 mg bid 1 d before and day of study

Boccalandro (46)

2003

600 mg bid day of study and 1 d after

Kay (7)

2003

600 mg bid 1 d before and 1 d after

Loutrianakis (49)

2003

600 mg bid 1 d before and 1 d after

Nogareda (11)

2003

600 mg 1 before then bid 2 d

Agarwal (unpublished data)

a

800 mg 24 h before then 600 mg bid 1 d

Treatment
Arms

Patients
(n)

Incidence
of RCIN

NAC

41

1/41

Placebo
NAC
Control
NAC
Control
NAC
Control
NAC
Placebo
NAC
Placebo
NAC
Placebo
NAC
Placebo
NAC
Placebo
NAC
Placebo
NAC
Control
NAC
Control
NAC
Placebo
NAC
Placebo
NAC
Control
NAC
Placebo

42
35
22
45
40
92
91
25
29
38
41
41
39
23
22
49
47
60
61
12
8
73
106
102
98
24
23
47
137
11
14

9/42
1/35
8/22
8/45
6/40
6/92
10/92
2/25
13/29
10/38
9/41
4/41
3/39
5/23
4/22
4/49
3/47
2/60
15/61
2/12
0/8
9/73
13/106
4/102
12/98
8/24
3/23
0/47
8/137
2/11
2/14

NAC, N-acetylcysteine; bid, twice daily; tid, thrice daily.

stantial interstudy variability of estimated effects outcomes,

despite an outwardly similar design (prospective, controlled,
patients with baseline renal insufficiency). This interstudy variability was strongly confirmed by formal statistical testing for
heterogeneity and held even when only randomized trials were
analyzed.
We regarded the presence of such pronounced heterogeneity
to contraindicate reliance on any single summary estimate of
treatment effect. An aggregated summary estimate in such a
circumstance would be an oversimplification, as the presence
of heterogeneity clearly suggests the existence of study characteristics, patient characteristics, or both, that appreciably
influence the magnitude, and perhaps also the direction, of the
effect estimates that different trials produce. We did not subscribe to the view that incorporation of the among-study variance (random-effects model) into the computation of a sum-

mary estimate accommodates or accounts for inconsistent

results among the studies. Rather, we concurred with the observation of DerSimonian and Laird that in drawing inferences from heterogeneous but logically related studies. . .the
use of regression analysis to characterize differences in study
outcomes may be more appropriate (21) and have followed
that advice.
As such, univariate meta-regression analysis has suggested
that patient characteristics and study design may account for
discrepant results of the effect of N-acetylcysteine. On closer
examination, these characteristics are plausible explanations of
the observed heterogeneity and may serve to direct future
investigations of the use of N-acetylcysteine. Speculatively, the
identified characteristics may also guide clinicians to carefully
choose certain groups of patients for the intervention.
Take for example, the characteristic, the time of dosing of

2002

2002

2002

2002

2002

2003

2002

2003

2002

2002

2003

2003

2003

2003

Adamian (47)

Allaqaband (10)

Briguori (4)

Diaz-Sandoval (20)

Durham (3)

Goldenberg (48)

Kahlon (5)

Oldemeyer (6)

Shyu (19)

Vallero (45)

Boccalandro (46)

Kayd (7)

Loutrianakis (49)

Nogaredad (11)

RCT

CT

RPCT

BRPCT

CT

CT

RPCT

BRPCT

RPCT

BRPCT

RPCT

BRPCT

RCT

RCT

CT

RPCT

Designa

Unpublished

Abstract

Abstract

Peer reviewed

Peer reviewed

Peer reviewed

Peer reviewed

Abstract

Abstract

Abstract

Peer reviewed

Peer reviewed

Peer reviewed

Peer reviewed

Abstract

Peer reviewed

Publication Type

0.45% NS 12 h before
and 12 h after

0.45% NS 12 h before
and 12 h after

600 mg bid 1 d before

and 1 d after

600 mg bid 1 d before

and day of study

400 mg bid 1 d before

and day of study
600 mg bid 1 d before
and day of study

LHC
intervention
LHC
intervention

600 mg bid 1 d before

and 1 d after
600 mg immediately
before then bid 2 d
800 mg 24 h before then
600 mg bid 1 d

LHC

LHC RA
intervention
LHC
intervention

600 mg bid 1 d before

and 1 d after

1500 mg bid 4 doses

LHC

LHC
intervention

600 mg bid 1 d before

and 1 d after

LHC
intervention

600 mg bid day of study

and 1 d after

0.45% NS 12 h before
and 12 h after

600 mg tid 1 d before

and 1 d after

LHC
intervention

LHC

0.9% NS 70 ml/kg
bolus for 12 h before
and 12 h after

1200 mg 1 h before and

3 h after

LHC
intervention

0.45% NS 12 h before
and 12 h after

NA

NA

0.9% NS 12 h before
and 6 h after

0.45% NS 12 h before
and 12 h after

0.45% NS 12 h before
and 24 h after

0.45% NS 12 h before
and 12 h after

0.45% NS 12 h before
and 12 h after

0.45% NS up to 12 h
before and 12 h after

600 mg bid 1 dose before 0.45% NS 212 h

and 3 d after
before and 12 h after

LHC

Peripheral or
LHC
intervention

Peripheral or
LHC
intervention

NA

600 mg bid

LHC
intervention

0.45% NS 12 h before
& 12 h after

IV Hydration Regimen
(ml/kg/h)

600 mg bid 1 d before

day of study

Dosing of NAC

CT scan

Indication for
Contrast Mediab

Iohexol

Ioversol

NA

Iopamidol

Iodixanol

Iodixanol

Iopamidol

Iopamidol

Iohexol

Iopamidol

Iohexol

Ioxilan

Iopromide

Iodixanol

Iodixanol

Iopromide

Contrast
Medium
66 (11)
65 (15)
71 (10)
72 (10)
70 (10)

71 (10)
64 (9)

64 (9)
74 (2)
72 (2)
71 (12)
70 (10)
NA

66 (12)

71 (9)
77 (9)
75 (8)
70 (7)
70 (7)
NAe

65 (11)
66 (11)
69 (4882)
69 (5081)
NA

73 (6879)
75 (7080)
65 (15)
62 (12)

Placebo
NAC
Control
NAC

Control
NAC

Control
NAC
Placebo
NAC
Placebo
NAC
Placebo
NAC

Placebo
NAC
Placebo
NAC
Placebo
NAC
Control
NAC
Control
NAC
Placebo
NAC
Placebo
NAC
Control
NAC
Placebo

Average Age
(yr)

NAC

Treatment
Armsc

57

34
36

42

36
NA

58
39

67

64
NA

49
63

59
41

57

46
NA

38
45

32
40

43
44

68
53

33
57

32

Prevalence of
Diabetes (%)

158 (29)

122 (99 to 139)

155 (21)

147 (115 to 159)

111 (66 to 320)

NA

167 (53)
109 (68 to 263)

135 (39)
158 (53)

246 (70)
132 (44)

146 (57)
246 (70)

172 (41)
143 (71)

185 (60)

202 (44)
NA

137 (4)
194 (35)

136 (32)
146 (5)

179 (42)
134 (38)

173 (44)
195 (88)

211 (114)
166 (39)

220 (114)

Average Baseline Serum

Creatinine, mol/L (SD)

Average Contrast
Volume, ml (SD)

170 (88)

162 (125 to 189)

188 (76)

164 (128 to 200)

120 (70 to 380)

NA

191 (120)
130 (75 to 320)

192 (142)

115 (48)
NA

127 (73)
119 (3)

103 (68)
134 (71)

105 (65)

85 (42)
NA

189 (12)
77 (36)

200 (144)
179 (8)

133 (68)
194 (127)

303 (152)
115 (59)

75 (NA)
273 (113)

75 (NA)

BRPCT, Double blinded randomized placebo-controlled trial; RPCT, randomized placebo-controlled trial; RCT, randomized controlled trial; CT, controlled trial.
LHC, Left heart catheterization; RA, renal arteriogram.
c
N, NAC; P, placebo.
d
Median and interquartile range specified.
e
NA, not available.

Agrawal Unpublished
Data

2000

Year

Tepel (2)

Primary Author of
Study

Table 2. Summary of study and patient characteristics from the 16 prospective controlled trials

J Am Soc Nephrol 15: 761769, 2004

N-acetylcysteine and Radiocontrast Nephropathy
765

766

Journal of the American Society of Nephrology

Figure 2. Funnel plot with pseudo 95% confidence limits for all 16
trials. rd, risk difference; s.e., standard error.

Figure 3. Forest plot of weighted risk difference of all 16 controlled

clinical trials. Relative weight is reported in parentheses after year of
publication.

N-acetylcysteine. Studies whose participants were dosed with

N-acetylcysteine on the day before the procedure were more
likely do demonstrate a reduction of radiocontrast-induced
nephropathy than those studies dosing participants on the day
of the procedure. Individuals receiving their first dose of
N-acetylcysteine on the day before, rather than on the day of,
the radiocontrast media may have greater systemic levels of the
antioxidant agent. However, peak serum levels of N-acetylcysteine occur 1 h after oral administration, and the half-life of
reduced N-acetylcysteine is approximately 2 h (22). Reduction

J Am Soc Nephrol 15: 761769, 2004

in renal function, furthermore, should not alter the elimination

pathway for the agent.
Other important characteristics identified by the univariate
meta-regression include the age of subjects, prevalence of
diabetes mellitus, and the type and volume of radiocontrast
media used by the individual studies. Studies with a higher
proportion of elderly individuals or diabetic patients, or studies
that used a high volume of radiocontrast media or that used a
relatively high osmolality agent, were associated with a change
in the estimated risk difference in a direction favoring N-acetylcysteine. Most of these identified characteristics are known
to be risk factors for the development of radiocontrast-induced
nephropathy (23). Thus, it is conceivable that the effectiveness
of N-acetylcysteine may be greatest among a population of
individuals with a high baseline risk of developing radiocontrast-induced nephropathy.
Yet there were some inconsistent findings with respect to the
risk factors for contrast nephropathy. Notably, the level of
renal function at baseline was not associated with a change in
the estimated effect of N-acetylcysteine. It may be that SCr
concentration is too imprecise and insensitive as an estimate of
GFR (24). Furthermore, the estimated effect N-acetylcysteine
was greater among the group of individuals with intravenous
hydration greater than 12 h compared to the group receiving
less than 12 h of saline hydration. We have no pathophysiologic explanation for the finding, and speculatively, this may
reflect an intrinsic anomaly of the data set.
The findings of our systematic review should be interpreted
in the context of both intrinsic limitations of meta-analysis, and
in the context of our own study-specific (subject matter) limitations. In systematic reviews, the traditional unit of analysis
is each study, rather than patients. Thus, the power to detect a
difference in aggregate or to identify explanatory variables by
meta-regression is greatly diminished compared with large
primary trials with individual-level data (or to meta-analyses
with individual-level data). Furthermore, interpretation of any
results for study or patient characteristics that must be represented by study population average values or percentages are
prone to the ecologic fallacy (25).
Second, although we have strongly demonstrated heterogeneity, we were not able to identify any single characteristic of
studies or patients within levels or categories of which the
results are consistent. Nor, among the randomized trials, did
the results of the placebo-controlled trials differ materially
from those that did not use placebo controls. The small number
of studies made it difficult to determine whether multiple study
and patient characteristics in combination might render the
results more consistent.
Third, we predominately identified and used published studies. Our search strategy identified only one unpublished trial,
and thus our results are heavily weighted on the findings of
these published trials. We identified an unpublished study of
quality at our local institution. It is quite likely that other
unpublished studies exist at other institutions, and we hope that
the publication of this study will allow for the eventual analysis
of these other studies. The exclusion of unpublished data are
generally associated with an overestimate of the true effect in

J Am Soc Nephrol 15: 761769, 2004

N-acetylcysteine and Radiocontrast Nephropathy

767

Table 3. Meta-regression of selected patient and study characteristics

Stratum

RCIN definition (rise in SCrc)

0.5 mg/dl rise
8
25% rise
3
0.5 mg/dl or 25%/33%
5
rise
Study quality
BRPCT
4
RPCT or RCT
8
CT
4
Average of subjects
age 70 yr
6
age 70 yr
7
Baseline SCra
average 2.0 mg/dl
9
average 2.0 mg/dl
5
Average prevalence diabetes
mellitus
50%
9
50%
4
Missing baseline data
unknown
3
Radiocontrast type
low-osmolar
10
iso-osmolar
4
Contrast volume
average 150 ml
7
average 150 ml
6
NAC administration
day of procedure
4
day before to
11
procedure
Hydration regimen
12 h before
10
12 h before
4
Hydration type
0.45% saline
11
0.9% saline
2
48-h risk for RCIN
yes
13
no
3

Pooled Risk
Difference

Change in
Risk
Differencea

0.05
0.11
0.00

Refd
0.06
0.05

0.08
0.04
0.06

0.05
Ref
0.02

0.06
0.11

Studiesb

95% CI

0.19 to 0.07
0.12 to 0.21
0.21 to 0.12

2,3,5,10,11,19,46,48
4,7,47
6,20,45,49,

0.19 to 0.14

6,7,20,48
2,3,4,5,10,19,49,
11,45,46,47

Ref
0.05

0.18 to 0.08

2,4,5,7,46,
3,6,10,11,19,20,47

0.07
0.08

Ref
0.01

0.15 to 0.14

4,6,7,11,20,45,46,47,
2,3,5,10,19

0.07
0.13

Ref
0.06

0.20 to 0.09

2,3,4,6,7,10,11,20,
5,19,46,47

0.10

NA

45,48,49

0.08
0.04

Ref
0.04

0.12 to 0.20

2,3,4,5,6,7,19,20,48,
10,45,46,47

0.07
0.10

Ref
0.04

0.18 to 0.09

2,3,5,6,7,10,19
4,11,20,46,47,

0.01
0.03

Ref
0.05

0.09 to 0.18

3,11,20,46
2,4,5,6,7,10,19,45,48,49,

0.07
0.04

Ref
0.03

0.14 to 0.21

2,4,6,7,10,19,46,47,48,
3,5,20,45

0.05
0.07

Ref
0.01

0.18 to 0.19

2,3,4,6,10,19,20,45,46,48,
5,7

0.07
0.05

Ref
0.10

0.07 to 0.27

2,3,4,6,7,10,19,20,45,46,47,48,
5,11,49

a
RD difference estimated using metareg procedure in STATA, with tau-squared between-study variance estimated using restricted
maximum likelihood (REML).
b
Studies with reference number. refers to unpublished study by Agrawal.
c
SCr, serum creatinine.
d
Ref, reference group.

meta-analysis (26). By far the single most common reason for

the inability to publish a trial is the lack of statistical significance, although some have suggested that the quality of unpublished trials may not be comparable to those accepted in
peer-reviewed journals (27). The unpublished study included

in this analysis was of high quality, randomized, placebo

controlled, and masked. In addition, we did not use a masked
assessment of study quality, nor did we formally test agreement among the independent observers.
Acute renal failure remains a common clinical occurrence

768

Journal of the American Society of Nephrology

among hospitalized patients (28 30). The development of

acute renal failure increases morbidity (31,32), cost (33), and
mortality (34 37). Exposure to radiocontrast media is a common iatrogenic cause of acute renal failure among hospitalized
patients with chronic kidney disease (38). The use of nonionic,
low-osmolality radiocontrast media (39,40) and the administration of intravenous fluids (41,42) have attenuated the risk of
developing acute renal failure. Initial studies suggested a
highly beneficial effect of N-acetylcysteine, and clinicians
rapidly adopted the agent into their armamentarium.
This systematic review suggests that the role of oral N-acetylcysteine in the prevention of radiocontrast-induced nephropathy has yet to be defined. The literature as it currently exists
is profoundly heterogeneous, making any single summary estimate invalid. Thus, the analysis could not demonstrate an
added benefit of oral N-acetylcysteine among all individuals
with preexistent renal insufficiency. Meta-regression analysis
identified some important study and patient characteristics that
may partially explain the heterogeneity: the time of N-acetylcysteine administration, advanced age, presence of diabetes
mellitus, and the volume and type of radiocontrast media. It is
biologically plausible that these characteristics would affect the
relationship of N-acetylcysteine and therefore may serve to
guide future research in this field.
Our results may have some important immediate and longterm implications. First, rather than the indiscriminate use of
N-acetylcysteine, clinicians may better direct their efforts at
proven interventions (intravenous saline, low-osmolality/isoosmolality contrast media). Second, clinicians should be judicious in their decision to order radiographic tests. Third, given
the low side-effect profile and cost of N-acetylcysteine, a large,
randomized, placebo-controlled trial should be conducted.
Large randomized-controlled trials have been known to differ
with findings of meta-analyses (43,44). Alternatively, a pooled
analysis of individual patient data may be a more economical
way of answering the question and likely should be done as a
prelude to a mega-trial. In conclusion, the findings of this
systematic review do not support the routine use of N-acetylcysteine for the prevention of radiocontrast-induced
nephropathy.

J Am Soc Nephrol 15: 761769, 2004

4.

5.

6.

7.

8.
9.
10.

11.

12.

13.

14.

Acknowledgments
We thank Drs. Philip Klemmer, Susan L. Hogan, Seema Garg, and
Ronald J. Falk for their thoughtful comments and critiques of the
article in manuscript. Dr. Poole was supported by supported in part by
a grant from the National Institute of Environmental Health Sciences
(P30ES10126).

15.

16.

References
1. Safirstein R, Andrade L, Viera J: Acetylcysteine and nephrotoxic
effects of radiographic contrast agentsA new use for an old
drug. N Engl J Med 343: 210 212, 2000
2. Tepel M, Van Der Giet M, Schwarzfeld C, Laufer U, Liermann
D, Zidek W: Prevention of radiographic-contrast-agentinduced
reductions in renal function by acetylcysteine. N Engl J Med 343:
180 184, 2000
3. Durham JD, Caputo C, Dokko J, Zaharakis T, Pahlavan M, Keltz
J, Dutka P, Marzo K, Maesaka JK, Fishbane S: A randomized

17.

18.

controlled trial of N-acetylcysteine to prevent contrast nephropathy in cardiac angiography. Kidney Int 62: 22022297, 2002
Briguori C, Manganelli F, Scarpato P, Elia PP, Golia B, Riviezzo
G, Lepore S, Mariateresa L, Villari B, Colombo A: Acetylcysteine and contrast agentassociated nephrotoxicity. J Am Coll
Cardiol 40: 298 303, 2002
Kahlon JP, Moser L, Rosman H, Nordstrom C, Zafar A, Khairulah Q, Schreiberer T, LaLande T, Gardin J, Yamasaki H: Effectiveness of N-acetylcysteine for the prevention of radiocontrastinduced nephropathy: Is the jury still out? [Abstract]. Circulation
106: II-691, 2002
Oldemeyer J, Cichowski E, Wurdeman R, Packard K, Biddle W,
Mooss A: Prophylactic acetylcysteine is not effective in preventing contrast-induced nephropathy following coronary angiography [Abstract]. J Am Coll Cardiol 41: 328A, 2003
Kay J, Chow WH, Mao CT, Lo SK, Kwok OH, Yip A, Fan K,
Lee CH, Lam WF: Acetylcysteine for prevention of acute deterioration of renal function following elective coronary angiography and intervention: A randomized controlled trial. JAMA 289:
553558, 2003
Curhan G: Prevention of contrast nephropathy. JAMA 289: 606
608, 2003
Conti C: Contrast nephropathy or contrast creatinopathy. Clin
Cardiol 26: 5354, 2003
Allaqaband S, Tumuluri R, Malik AM, Gupta A, Volkert P,
Shaleve Y, Bajwa TK: Prospective randomized study of N-acetylcysteine, fenoldopam, and saline for prevention of radiocontrast-induced nephropathy. Catheter Cardiovasc Interv 57: 279
283, 2002
Nogareda G, Ricci D, IM P, Huckell V, Fox R, Chow J, Eichman
A, Buller C: Oral N-acetylcysteine administered immediately
prior to routine percutaneous coronary procedures ameliorates
contrast-induced nephrotoxicity [Abstract]. J Am Coll Cardiol
41, 2003
Gruberg L, Mintz GS, Mehran R, Gangas G, Lansky AJ, Kent
KM, Pichard AD, Satler LF, Leon MB: The prognostic implications of further renal function deterioration within 48 h of interventional coronary procedures in patients with pre-existent
chronic renal insufficiency. J Am Coll Cardiol 36: 15421548,
2000
McCullough P, Wolyn R, Rocher L, Levin R, ONeill W: Acute
renal failure after coronary intervention: Incidence, risk factors,
and relationship to mortality. Am J Med 103: 368 375, 1997
Begg C, Mazumdar M: Operating of rank correlation test for
publication bias. Biometrics 50: 1088 1101, 1994
Egger M, Davey S, Schneider M, Minder C: Bias in metaanalysis detected by a simple, graphical test. BMJ 315: 629 634,
1997
Deeks J, Altman D, Bradburn M: Statistical methods for examining heterogeneity and combining results from several studies in
meta-analysis. In: Systematic Reviews in Health Care, edited by
Egger M, Smith DG, Altman DG, 2nd Ed., London, BMJ Books,
2001, pp 285312
Mouhayer EN, Tadros G, Akinwanda AA, Campbell B, Blankenship JC, Iliadias EA: Prevention of contrast-induced renal
dysfunction with acetylcysteine in patients undergoing coronary
angiography [Abstract]. J Am Coll Cardiol 39: 1A, 2002
Gill N, Piccione E, Vido D, Clark B, Shannon R: The role of
acetylcysteine in preventing contrast-induced nephropathy following cardiac catheterization in patients with acute and chronic
renal failure [Abstract]. J Am Coll Cardiol 41: 23A 2003

J Am Soc Nephrol 15: 761769, 2004

19. Shyu K-G, Cheng J-J, Peiliang K: Acetylcysteine protects

against acute renal damage in patients with abnormal renal
function undergoing a coronary procedure. J Am Coll Cardiol 40:
13831388, 2002
20. Diaz-Sandoval LJ, Kosowsky BD, Losordo DW: Acetylcysteine
to prevent angiography-related renal tissue injury. The APART
Trial. Am J Cardiol 89: 356 358, 2002
21. DerSimonian, Laird N: Meta-analysis in clinical trials. Control
Clin Trials 7: 177188, 1986
22. Morgan LR, Holdiness MR, Gillen LE: N-acetylcysteine: Its
bioavailability and interaction with ifosfamide metabolites. Semin Oncol 10[Suppl 1]: S56 S61, 1983.
23. Murphy SW, Barrett BJ, Parfrey PS: Contrast nephropathy. J Am
Soc Nephrol 5: 125137, 2000
24. Hsu C, Chertow G, Curhan GC: Methodological issues in studying the epidemiology of mild to moderate chronic renal insufficiency. Kidney Int 61: 15671576, 2002
25. Greenland S: Quantitative methods in the review of epidemiologic literature. Epidemiol Rev 9: 130, 1987
26. McAuley L, Pham B, Tugwell P, Moher D: Does the inclusion of
grey literature influence estimates of intervention effectiveness
reported in meta-analyses. Lancet 356: 1228 1231, 2000
27. Cooke D, Guyatt G, Ryan G: Should unpublished data be included in meta-analyses: Current convictions and controversies.
JAMA 269: 2749 2753, 1999
28. Hou S, Bushinsky D, Wish J, Cohen J, Harrington J: Hospital
acquired renal insufficiency: A prospective study. Am J Med 74:
243248, 1983
29. Shusterman N, Strom B, Murray T, Morrison G, West S, Maislin
G: Risk factors and outcome of hospital-acquired acute renal
failure: Clinical epidemiologic study. Am J Med 83: 6571, 1987
30. Nash K, Hafeez A, Hou S: Hospital-acquired renal insufficiency.
Am J Kidney Dis 39: 930 936, 2002
31. Liano F, Garcia-Martin F, Gallego A: Easy and early prognosis
of acute tubular necrosis: A forward analysis of 228 cases.
Nephron 51: 307313, 1989
32. Pascual J, Orofino L, Liano F: Incidence and prognosis of acute
renal failure in older patients. J Am Geriatric Soc 38: 2530, 1990
33. Hamel MB, Phillips RS, Davis RB, Desbiens N, Connors AF Jr,
Teno JM, Wenger N, Lynn J, Wu AW, Fulkerson W, Tsevat J:
Outcomes and cost-effectiveness of initiating dialysis and continuing aggressive care in seriously ill hospitalized adults. Ann
Intern Med 127: 195202, 1997
34. Novis B, Rozien M, Aronson S, Thisted R: Association of
preoperative risk factors with postoperative acute renal failure.
Anesth Analg 78: 143149, 1994
35. Wardle E: Acute renal failure and multiorgan failure. Nephron
66: 380 385, 1994
36. Zanardo G, Michielon P, Paccagnella A: Acute renal failure in
the patient undergoing cardiac operation: Prevalence, mortality
rate and main risk factors. J Thorac Cardiovasc Surg 107:
1489 1495, 1994

N-acetylcysteine and Radiocontrast Nephropathy

769

37. Brivet G, Kleinknecht D, Landais P: Acute renal failure in

intensive care units: Causes, outcome, and prognostic features of
hospital mortalityA prospective, multicenter study. Crit Care
Med 24: 192198, 1996
38. Erley C: Acute renal failure associated with radiocontrast agents.
In: Acute Renal Failure, edited by Molitoris BA, Finn WF,
Philadelphia, W.B. Saunders, 2001, pp 376 382
39. Barrett B, Carlisle E: Meta-analysis of the relative nephrotoxicity
of high- and low-osmolality iodinated contrast media. Radiology
188: 171178, 1993
40. Rudnick M, Goldfarb S, Wexler L: Nephrotoxicity of ionic and
nonionic contrast media in 1196 patients: A randomized trial.
The Iohexol Cooperative Study. Kidney Int 47: 254 261, 1995
41. Solomon R, Werner C, Mann D, DElia, Silva P: Effects of
saline, mannitol and furosemide on acute decreases in renal
function induced by radiocontrast agents. N Engl J Med 331:
1416 1420, 1994
42. Mueller C, Buerkle G, Buettner HJ, Petersen J, Perruchond AD,
Eriksoon U, Marsch S, Roskamm H: Prevention of contrast
media-associated nephropathy: Randomized comparison of 2
hydration regimens in 1620 patients undergoing coronary angioplasty. Arch Intern Med 162: 329 336, 2002
43. LeLorier J, Gregoire G, Benhaddad A, Lapierre J, Derderian F:
Discrepancies between meta-analyses and subsequent large randomized, controlled trials. N Engl J Med 337: 536 542, 1997
44. DerSimonian R, Levine R: Resolving discrepancies between
meta-analysis and a subsequent large controlled trial. JAMA 282:
664 670, 1999
45. Vallero A, Cesano G, Garbo R, Minelli M, Quarello F, Formica
M: Nefropatia da contrasto in cardiologia interventistica: Assenza di vantaggi con impiego profilattico di N-acetilcisteina.
Giornale Ital Nefrol 19: 529 533, 2002
46. Boccalandro F, Amhad M, Smalling RW, Sdringola S: Oral
acetylcysteine does not protect renal function from moderate to
high doses of intravenous radiographic contrast. Catheter Cardiovasc Interv 58: 336 341, 2003
47. Adamian MG, Moussa I, Mehran R, Dangas G, Farkhou M,
Adamyan Y, Mohamed M, Negoita A, Reyes A, Losquadro M.
The role of mucomyst administration prior to percutaneous interventions on renal function in patients with chronic renal failure [Abstract]. J Am Coll Cardiol 39: 1A, 2002
48. Goldenberg I, Jonas M, Matetzki S, Schechter M, Agranat O,
Har-Zahav I, Elian D, Guetta V: Contrast-associated nephropathy and clinical outcome of patients with chronic renal insufficiency undergoing cardiac catheterization: Lack of additive benefit of acetylcysteine to saline hydration [Abstract]. J Am Coll
Cardiol 41: 537A, 2003
49. Loutrianakis E, Stella D, Hussain A, Lewis B, Steen L, Sochanski M, Leya F, Grassman E: Randomized comparison of fenoldopam and N-acetylcysteine to saline in the prevention of radiocontrast induced nephropathy [Abstract]. J Am Coll Cardiol 41:
327A, 2003

Access to UpToDate on-line is available for additional clinical information

at http://www.jasn.org/