Author + information
- Received April 24, 2009
- Accepted May 3, 2009
- Published online July 1, 2009.
- Michael Reed, MD⁎,
- Pascal Meier, MD⁎,
- Umesh U. Tamhane, MD⁎,
- Kathy B. Welch, MS, MPH†,
- Mauro Moscucci, MD‡ and
- Hitinder S. Gurm, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Hitinder S. Gurm, Department of Medicine, Division of Cardiovascular Medicine, University of Michigan Cardiovascular Center, 2A394, 1500 East Medical Center Drive, Ann Arbor, Michigan 48109-5853
Objectives We sought to compare the nephrotoxicity of the iso-osmolar contrast medium, iodixanol, to low-osmolar contrast media (LOCM).
Background Contrast-induced acute kidney injury (CI-AKI) is a common cause of in-hospital renal failure. A prior meta-analysis suggested that iodixanol (Visipaque, GE Healthcare, Princeton, New Jersey) was associated with less CI-AKI than LOCM, but this study was limited by ascertainment bias and did not include the most recent randomized controlled trials.
Methods We searched Medline, Embase, ISI Web of Knowledge, Google Scholar, Current Contents, and International Pharmaceutical Abstracts databases, and the Cochrane Central Register of Controlled Trials from 1980 to November 30, 2008, for randomized controlled trials that compared the incidence of CI-AKI with either iodixanol or LOCM. Random-effects models were used to calculate summary risk ratios (RR) for CI-AKI, need for hemodialysis, and death.
Results A total of 16 trials including 2,763 subjects were pooled. There was no significant difference in the incidence of CI-AKI in the iodixanol group than in the LOCM group overall (summary RR: 0.79, 95% confidence interval [CI]: 0.56 to 1.12, p = 0.19). There was no significant difference in the rates of post-procedure hemodialysis or death. There was a reduction in CI-AKI when iodixanol was compared with ioxaglate (RR: 0.58, 95% CI: 0.37 to 0.92; p = 0.022) and iohexol (RR: 0.19, 95% CI: 0.07 to 0.56; p = 0.002), but no difference when compared with iopamidol (RR: 1.20, 95% CI: 0.66 to 2.18; p = 0.55), iopromide (RR: 0.93, 95% CI: 0.47 to 1.85; p = 0.84), or ioversol (RR: 0.92, 95% CI: 0.60 to 1.39; p = 0.68).
Conclusions This meta-analysis including 2,763 subjects suggests that iodixanol, when compared with LOCM overall, is not associated with less CI-AKI. The relative renal safety of LOCM compared with iodixanol may vary based on the specific type of LOCM.
- contrast media
- low-osmolar contrast media
- contrast-induced nephropathy
- contrast-induced acute kidney injury
Contrast-induced acute kidney injury (CI-AKI) is a common complication of intra-arterial and intravenous contrast administration. As the population ages, the incidence of CI-AKI likely will rise with the increased referrals for cardiac catheterization and radiology studies with intravenous contrast and with the increased prevalence of risk factors for CI-AKI, such as chronic kidney disease, diabetes, and anemia (1–3). CI-AKI typically occurs within the first 48 to 72 h after exposure to intravascular iodinated contrast medium. CI-AKI rarely directly results in immediate death or the need for immediate hemodialysis, but it is associated with increased cost, increased hospital stay, and increased in-hospital and long-term morbidity and mortality (2,4,5).
Sodium bicarbonate and acetylcysteine have been shown in multiple studies to prevent CI-AKI, although there is significant heterogeneity among the existing trials (6–9). Prevention of CI-AKI with prophylactic hydration and minimization of total contrast volume are well-established means of preventing CI-AKI (10,11).
The choice of the contrast media used may also influence the risk of CI-AKI. The use of high-osmolar contrast media results in more CI-AKI than the more contemporary low-osmolar contrast media (LOCM) or the iso-osmolar contrast medium, iodixanol (Visipaque, GE Healthcare, Princeton, New Jersey) (12,13). Early studies suggested that iodixanol is even less harmful to renal function than some types of LOCM are (14,15). A previous meta-analysis (16) suggested a dramatic reduction in CI-AKI with iodixanol when compared with LOCM. However, this analysis included trials that were not specifically designed to study the incidence of CI-AKI and in which renal function was not systematically determined in all patients. Thus, this study was limited by ascertainment bias (17).
These data provided the rationale for the American College of Cardiology/American Heart Association's revised 2007 guidelines for the management of patients with unstable angina/non–ST-elevation myocardial infarction, which now give a class IA recommendation for the use of iso-osmolar contrast media for coronary angiography in patients with chronic kidney disease (18). More recently, several larger randomized controlled trials have been published and have shown no difference in CI-AKI when iodixanol was compared with different types of LOCM (19–22). Also, a recent meta-analysis comparing iodixanol to a pool of nonionic LOCM showed no significant reduction in the rates of CI-AKI (23). There are significant pharmacological differences between various LOCM agents, and the disparate results of these trials may be in part related to the specific LOCM agent used. The objective of this systematic review and meta-analysis was to integrate the latest randomized clinical trials to provide a comprehensive comparison of the nephrotoxicity of the iso-osmolar agent iodixanol and the currently available LOCM.
We performed a computerized search to identify relevant articles from 1980 through November 30, 2008, using MEDLINE (National Library of Medicine, Bethesda, Maryland), Embase, ISI Web of Knowledge, Google Scholar, Current Contents, and International Pharmaceutical Abstracts databases, as well as the Cochrane Central Register of Controlled Trials. We used the following keywords: “iso-osmolar,” “iodixanol” (Visipaque, GE Healthcare, Princeton, New Jersey), “low-osmolar,” “iohexol” (Omnipaque, GE Healthcare, Princeton, New Jersey), “iopamidol” (Isovue, Bracco Diagnostics, Princeton, New Jersey), “iopromide” (Ultravist, Bayer HealthCare Pharmaceuticals, Montville, New Jersey), “ioversol” (Optiray, Tyco, Tyco Healthcare, Mallinckrodt Inc., Hazelwood, Missouri), “iomeprol” (Iomeron, Bracco Diagnostics, Princeton, New Jersey), “nephrotoxicity,” “contrast nephropathy,” “radiocontrast,” “renal failure,” and “randomized.” Abstract lists from the 2006 to 2008 scientific meetings of the American Heart Association, the American College of Cardiology, the European Society of Cardiology, and the Transcatheter Cardiovascular Therapeutics were viewed. Other Internet-based sources of information were also used to identify any other studies of interest. When needed, further information on the trials was obtained from the study investigators.
Study selection, data extraction, and quality assessment
A study was included if it randomized patients undergoing contrast media application to either LOCM or iodixanol, and if data on renal function were routinely ascertained in all patients. Data were independently abstracted by 3 reviewers (M.R., P.M., U.T.) and disagreements were resolved by consensus. Reviewers were not blinded to study authors or outcomes. Attempts were made to retrieve the data from the original source in unpublished studies. Baseline demographic, clinical, and procedural characteristics of each trial were recorded, including mean age, average contrast volume, prophylactic hydration volume, information about sex, diabetes, chronic kidney disease, type of procedure, type of LOCM used, the use of prophylactic therapies such as acetylcysteine, and the definition of CI-AKI used. We assessed trial quality by evaluating specific elements of study design (i.e., concealment of allocation during randomization, intention-to-treat analysis, and blinded assessment of outcome measures), but did not use a quality score given the limitations inherent to such an approach (24).
End points, data synthesis, and data analysis
The primary end point was the incidence of CI-AKI as defined by each individual study protocol. Secondary end points were the need for renal replacement therapy and mortality. Data from all the selected studies were combined to estimate the pooled risk ratios (RR) for iodixanol versus LOCM using a random-effects model. Significant between-study heterogeneity was expected regarding study populations and because different LOCM were used as comparators; therefore, a random-effects model was used to produce across-study summary RR with 95% confidence intervals (CI). All analyses were performed on an intention-to-treat basis. Continuity correction was used when an event did not occur in 1 group (25). We quantitatively evaluated the presence of heterogeneity across trials with the Q and I2 statistics, with an I2 value of 25% indicating low, 50% at least moderate, and 75% high between-study heterogeneity. To assess the effect of individual studies on the summary estimate of effect, we performed an influence analysis, in which the pooled estimates were recalculated by omitting 1 study at a time. Publication bias was tested by plotting a funnel plot, by formal testing using rank order correlation and Egger test of intercept, and by the Orwin fail-safe N (26–29). Duval-Tweedie trim and fill method was used to assess for presence of and impact of missing studies (30). An exploratory meta-regression using a mixed-effect model was performed to assess the impact of select variables (age, diabetes, baseline renal function, contrast volume) on the incidence of CI-AKI. Finally, an omnibus test was used to test the validity of each subgroup analysis (31). All analyses were performed using Comprehensive Meta-Analysis software, version 2.0 (Biostat, Englewood, New Jersey) and Statistical Analysis System software, release 9.2 for Windows 2002 to 2008 (SAS Institute, Cary, North Carolina).
A total of 120 articles were reviewed. Ninety articles were initially rejected because they did not directly compare iodixanol and LOCM, or because they did not document renal function as an outcome or safety end point. Of the 30 remaining studies, 14 were rejected because they were not randomized controlled trials. The remaining 16 studies satisfied the pre-determined inclusion criteria and were considered in the analysis (Fig. 1) (15,19–22,32–42). One study has not been published as a peer-reviewed report yet but was presented at the Transcatheter Therapeutics 2006 meeting (40).
Table 1 summarizes the characteristics of the 16 trials included in our analysis. Of the 2,763 subjects, a total of 1,383 were randomly assigned to iodixanol (Visipaque) and 1,380 to LOCM. Each of the 16 trials was designed as a superiority trial and used creatinine change or CI-AKI as a primary or secondary outcome. Nine of the trials involved coronary angiography alone, 1 trial involved coronary or peripheral angiography, 1 trial involved peripheral angiography, and 5 involved intravenous injection of contrast for computed tomography scans. Twelve of the trials (2,266 of 2,763 patients in the pooled analysis) only included patients with moderate or severe renal insufficiency, which was variably defined as serum creatinine ≥1.5 mg/dl and/or creatinine clearance <60 ml/min (14,20,33,37,39,40), estimated glomerular filtration rate <60 ml/min (19,22), creatinine clearance <60 ml/min (15), serum creatinine ≥1.7 mg/dl in men and ≥1.5 mg/dl in women (21), and serum creatinine >1.7 mg/dl (35). Thirteen trials included patients with diabetes mellitus. Ten trials used prophylactic hydration and 6 trials used prophylactic acetylcysteine. There were no relevant differences noted between the baseline characteristics of the iodixanol and LOCM groups of patients in each of the individual trials. The type of LOCM agent used in each study was ioxaglate (Hexabrix, Guerbet Group, Villepinte, France) (n = 4), iopromide (Ultravist, Bayer HealthCare Pharmaceuticals, Montville, New Jersey) (n = 4), iohexol (Omnipaque, GE Healthcare) (n = 3), iopamidol (Isovue, Bracco Diagnostics) (n = 3), ioversol (Optiray, Tyco Healthcare Group, Mallinckrodt Inc.) (n = 1), and iomeprol (Iomeron, Bracco Diagnostics) (n = 1). Ioxaglate is the only ionic dimer included; the other agents are nonionic monomers. The definition of CI-AKI varied among the trials. Most studies (n = 6) defined CI-AKI as an absolute increase of creatinine by at least 0.5 mg/dl (20–22,33,35,41) or as a relative increase by at least 25% (n = 4) (19,36,37). A few studies used a composite of the 2 definitions (n = 3) (15,34,38,40). One trial used the definition of a relative increase of at least 50% (39), and 2 trials based their definition on a decrease of glomerular filtration rate of at least 25% (32,42). For each study, we used the corresponding pre-defined primary end point.
Clinical end points
Contrast-induced acute kidney injury
The primary clinical end point of our meta-analysis was CI-AKI. There was an overall trend in favor of iodixanol, but no significant difference in the overall risk of CI-AKI between the iodixanol and the pooled LOCM groups. CI-AKI occurred in 128 of 1,379 patients receiving iodixanol and in 158 of the 1,375 patients assigned to LOCM treatment (RR: 0.79, 95% CI: 0.56 to 1.12; p = 0.189) (Fig. 2). There was moderate heterogeneity across studies regarding clinical or patient characteristics and protocols, and also according to formal tests (I2 = 44.7; Q = 23.5, p = 0.036). An influence analysis omitting 1 individual study at a time to determine if a single study dominated the results of the pooled analysis did not change the lack of difference between iodixanol and the pool of LOCM agents, suggesting that no single individual study overwhelmingly influenced the overall results of the pooled analysis.
In a subset analysis of the 8 studies involving 1,793 patients in the setting of coronary angiography and interventions, there was no difference in CI-AKI between patients randomized to iodixanol and LOCM (RR: 0.82, 95% CI: 0.55 to 1.21; p = 0.31) (Fig. 3). A subgroup analysis limited to data specific for diabetic patients (n = 914, 7 studies) demonstrated similar findings (RR: 0.64, 95% CI: 0.29 to 1.44; p = 0.285). Meta-regressions did not suggest an association between the relative effect of iodixanol and the average baseline creatinine level of each study (slope = –0.029, standard error [SE]: 0.028; p = 0.31) or any of the other variables studied (data not shown).
A stratified analysis by the specific LOCM explained part of the observed heterogeneity and demonstrated a relatively lower risk of CI-AKI with iodixanol than with iohexol (RR: 0.19, 95% CI: 0.07 to 0.56; p = 0.002) or ioxaglate (RR: 0.58, 95% CI: 0.37 to 0.92; p = 0.022) (Fig. 4). No relative difference in the risk of CI-AKI was observed between iodixanol and iopamidol (RR: 1.20, 95% CI: 0.66 to 2.18; p = 0.55), iopromide (RR: 0.93, 95% CI: 0.47 to 1.85; p = 0.84), or ioversol (RR: 0.92, 95% CI: 0.60 to 1.39; p = 0.68). Only 1 study used iomeprol and thus a pooled analysis was not performed, but this is the 1 study that actually shows a higher incidence of CI-AKI with iodixanol (20). An omnibus test was used to investigate the validity of the subgroup analysis and confirmed a significant difference in CI-AKI when comparing the type of LOCM (p = 0.004), but not when comparing the route (intravenous vs. intra-arterial, p = 0.07) of contrast administration.
There was no suspicion of publication bias based on funnel plot symmetry and according to formal testing (rank order correlation or Kendall taub: −0.022, 1-tailed p value: 0.48, Egger test intercept: −0.24, 95% CI: −1.83 to 1.35, 1-tailed p value: 0.37) (Fig. 5). The classic fail-safe N was not valid for our analysis because there was no significant difference between the 2 groups; thus, an Orwin fail-safe N was derived. Using an imputed RR of 0.3 favoring iodixanol over LOCM, 15 studies would be required to achieve a significant pooled RR of CI-AKI under 0.5. Using a less stringent RR of 0.39 as described in the prior meta-analysis (16), 30 such studies would be needed.
Need for hemodialysis
Need for hemodialysis was recorded in 12 trials. In 9 of these, no event was observed in either of the study arms. A total of 2 patients in the iodixanol arms (n = 1,031) required hemodialysis (range: 0% to 1.5%), and 9 patients in the LOCM arms (n = 1,036; range: 0% to 9.2%). This difference was not statistically significant (RR: 0.37, 95% CI: 0.08 to 1.68; p = 0.20).
Data for mortality were available from 9 studies (n = 1,491 patients). No death was recorded in either study arm in 6 of these trials. The overall mortality rate was not significantly different between the 2 treatment groups: 7 deaths in the 743 patients assigned to iodixanol (range: 0% to 4.9%) and 5 deaths in the 748 patients randomized to LOCM (range: 0% to 3.6%), with a summary RR of 1.38 (95% CI: 0.33 to 5.79, p = 0.663).
The main finding of this systematic review of the published studies and meta-analysis is that there is no significant difference in the incidence of CI-AKI when using iodixanol than when using a pool of ionic and nonionic LOCM. The vast majority of these trials (12 of 16) included subjects with elevated mean creatinine. Although the inclusion of both ionic and nonionic forms of LOCM as well as of intravenous and intra-arterial contrast injection helps make the results of this analysis more generalizable, it should also be noted that our subgroup analysis of only those trials involving coronary angiography also does not show a benefit of iodixanol over other LOCM.
A stratified analysis that indirectly compared the various types of LOCM demonstrated that some LOCM are relatively more likely to cause CI-AKI compared with iodixanol than others. In our stratified analysis, iodixanol is associated with a relatively lower rate of CI-AKI when compared with ioxaglate or iohexol, but there is no relative difference in CI-AKI when comparing iodixanol with iopromide, iopamidol, iomeprol, or ioversol. This suggests that the risk of CI-AKI cannot be explained by osmolarity alone (Table 2). Ionicity is a feature unique to ioxaglate among the LOCM, and our stratified analysis suggests this agent is relatively more likely to result in CI-AKI. Whether viscosity is a significant factor in the risk of developing CI-AKI is not known. In fact, the exact mechanism of CI-AKI is imperfectly established, but it is thought to involve renal medullary hypoxia due to renal artery vasoconstriction and hyperviscosity or direct renal cytotoxicity from contrast agents (43,44).
The lack of significant difference in mortality and need for hemodialysis between those treated with iodixanol and those treated with LOCM is not surprising, given the sheer infrequency of this outcome and given that the follow-up in many of these trials was very short. With that said, even modest elevations in serum creatinine after contrast exposure are associated with increased long-term morbidity and mortality. In a retrospective analysis of 9,067 post-PCI patients, the 1-year survival of patients with renal failure was 70.3% and without renal failure was 93.6% (4).
Although the pooled analysis trended in favor of iodixanol, this was not statistically significant—a finding that differs from the results of a previous meta-analysis published by McCullough et al. (16) that indicated that the use of iodixanol is associated with lower rates of CI-AKI than LOCM, particularly in patients with chronic kidney disease and diabetes (16). The subjects used in the analysis by McCullough et al. (16) were taken from a pool of trials that did not explicitly use CI-AKI as an end point, which may have resulted in ascertainment bias and lack of control of confounding variables. In fact, although creatinine typically peaks at 48 to 72 h after intravascular contrast exposure, only 40% of the subjects had creatinine checks at 48 h and less than 20% had creatinine checks at 72 h. In addition, 83% of the subjects in the LOCM arm received ioxaglate or iohexol, both of which were associated with relatively more CI-AKI in our stratified analysis. Finally, the study by McCullough et al. (16) only included trials through 2003, and there are several randomized controlled trials published since 2003 that we included in this current study. The data presented here provide an updated and more comprehensive look at the nephrotoxicity of iodixanol versus various LOCM. We only included trials designed to look at CI-AKI as an end point, thus ensuring adequate control of confounding variables in the 2 study arms and minimizing the likelihood of ascertainment bias. Our results corroborate those of a recent meta-analysis that showed no significant difference in the rates of CI-AKI when iodixanol was compared with a pool of nonionic LOCM (23). However, our study includes both ionic and nonionic forms of LOCM and provides a direct comparison of coronary angiography trials.
The 2007 ACC/AHA guidelines update gives a class IA recommendation for the use of iso-osmolar contrast media for coronary angiography in the setting of unstable angina or non–ST-segment elevation myocardial infarction and renal insufficiency (18). The findings of recent randomized trials and of this meta-analysis suggest a revision of this recommendation may be in order.
Finally, given the relative lack of difference in CI-AKI with use of iodixanol and some types of LOCM, cost emerges as an important factor in the choice of which agent to use. Pricing for LOCM and iodixanol varies among institutions depending on the size of the contract and the packaging of the materials. When considering the vast number of cardiac catheterizations, peripheral angiography procedures, and computed tomography scans with intravenous contrast performed each year in the U.S., choosing a less-expensive but equally efficacious agent could result in significant cost-reduction.
Our meta-analysis is subject to the limitations inherent to all meta-analysis. There is a risk of publication bias, although this tested nonsignificant in our study. There is possibility of limited power and a type II statistical error, but this is unlikely as demonstrated by the high Orwin fail-safe number. Also, not all trials reported standardizations on hydration or prophylactic medications used. Finally, although CI-AKI is associated with worse long-term prognosis, the included trials focused primarily on short-term elevations in creatinine rather than long-term outcomes of harder end points.
This systematic review and meta-analysis demonstrates that the use of iodixanol, when compared with a pool of LOCM, does not result in less nephrotoxicity. There appears to be no difference in the risk of nephrotoxicity when iodixanol is compared with iopamidol, iopromide, and ioversol, whereas iodixanol appears to cause less nephrotoxicity than ioxaglate and iohexol do. Further trials comparing various iso-osmolar and low-osmolar media are needed in order to determine the optimal contrast agent for use in patients with chronic renal insufficiency and the impact of individual contrast media on long-term outcomes.
The authors are especially grateful to Gupreet K. Rana, Clinical Education Librarian, Taubman Medical Library, University of Michigan, for her help during the search for published reports.
Dr. Moscucci participated in a continuing medical education program sponsored by Bayer, a manufacturer of contrast media.
- Abbreviations and Acronyms
- confidence interval
- contrast-induced acute kidney injury
- low-osmolar contrast medium/media
- risk ratio
- Received April 24, 2009.
- Accepted May 3, 2009.
- American College of Cardiology Foundation
- Rihal C.S.,
- Textor S.C.,
- Grill D.E.,
- et al.
- Briguori C.,
- Airoldi F.,
- D'Andrea D.,
- et al.
- Jo S.H.,
- Youn T.J.,
- Koo B.K.,
- et al.
- McCullough P.A.,
- Bertrand M.E.,
- Brinker J.A.,
- Stacul F.
- Detrenis S.,
- Meschi M.,
- Savazzi G.
- Anderson J.L.,
- Adams C.D.,
- Antman E.M.,
- et al.,
- American College of Cardiology, American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines for the Management of Patients With Unstable Angina/Non-ST-Elevation Myocardial Infarction), American College of Emergency Physicians, Society for Cardiovascular Angiography and Interventions, Society of Thoracic Surgeons, American Association of Cardiovascular and Pulmonary Rehabilitation, Society for Academic Emergency Medicine
- Thomsen H.S.,
- Morcos S.K.,
- Erley C.M.,
- et al.,
- Investigators in the Abdominal Computed Tomography: IOMERON 400 Versus VISIPAQUE 320 Enhancement (ACTIVE) Study
- Rudnick M.R.,
- Davidson C.,
- Laskey W.,
- Stafford J.L.,
- Sherwin P.F.,
- VALOR Trial Investigators
- Solomon R.J.,
- Natarajan M.K.,
- Doucet S.,
- et al.,
- CARE Study Investigators
- Egger M.,
- Davey Smith G.,
- Schneider M.,
- Minder C.
- Rosenberg M.S.
- West B.,
- Welch K.,
- Galecki A.
- Chalmers N.,
- Jackson R.W.
- Klow N.E.,
- Levorstad K.,
- Berg K.J.,
- et al.
- Mehran R.
- Tveit K.,
- Bolz K.D.,
- Bolstad B.,
- et al.