RSS Feeds
Meta-analysis: Effectiveness of Drugs for Preventing Contrast-Induced Nephropathy
- Aine M. Kelly, MD, MS;
- Ben Dwamena, MD;
- Paul Cronin, MD, MS;
- Steven J. Bernstein, MD, MPH; and
- Ruth C. Carlos, MD, MS
- From the University of Michigan and Veteran Affairs Ann Arbor Healthcare System, Ann Arbor, Michigan.
Abstract
Background: N-Acetylcysteine, theophylline, and other agents have shown inconsistent results in reducing contrast-induced nephropathy.
Purpose: To determine the effect of these agents on preventing nephropathy.
Data Sources: Relevant randomized, controlled trials were identified by computerized searches in MEDLINE (from 1966 through 3 November 2006), EMBASE (1980 through November 2006), PubMed, Web of Knowledge (Current Contents Connect, Web of Science, BIOSIS Previews, and ISI Proceedings for the latest 5 years), and the Cochrane Library databases (up to November 2006). Databases were searched for studies in English, Spanish, French, Italian, and German.
Study Selection: Randomized, controlled trials that administered N-acetylcysteine, theophylline, fenoldopam, dopamine, iloprost, statin, furosemide, or mannitol to a treatment group; used intravenous iodinated contrast; defined contrast-induced nephropathy explicitly; and reported sufficient data to construct a 2 × 2 table of the primary effect measure.
Data Extraction: Abstracted information included patient characteristics, type of contrast media and dose, periprocedural hydration, definition of contrast-induced nephropathy, and prophylactic agent dose and route.
Data Synthesis: In the 41 studies included, N-acetylcysteine (relative risk, 0.62 [95% CI, 0.44 to 0.88]) and theophylline (relative risk, 0.49 [CI, 0.23 to 1.06]) reduced the risk for contrast-induced nephropathy more than saline alone, whereas furosemide increased it (relative risk, 3.27 [CI, 1.48 to 7.26]). The remaining agents did not significantly affect risk. Significant subgroup heterogeneity was present only for N-acetylcysteine. No publication bias was discerned.
Limitations: All trials evaluated the surrogate end point of contrast-induced nephropathy as the primary outcome. The lack of a statistically significant renoprotective effect of theophylline may result from insufficient data or study heterogeneity. True study quality remains uncertain.
Conclusion: N-Acetylcysteine is more renoprotective than hydration alone. Theophylline may also reduce risk for contrast-induced nephropathy, although the detected association was not significant. Our data support the administration of N-acetylcysteine prophylaxis, particularly in high-risk patients, given its low cost, availability, and few side effects.
Editors' Notes
Context
-
Contrast-induced nephropathy is a common cause of acute renal failure in hospitalized patients. Clinicians use a variety of contrast agents to reduce the risk for contrast-induced nephropathy, including N-acetylcysteine, theophylline, fenoldopam, dopamine, furosemide, mannitol, and bicarbonate.
Contribution
-
Although all of the agents included in this analysis reduced the risk for contrast-induced nephropathy, this meta-analysis of 33 trials involving 3622 patients found the strongest evidence for the effectiveness of N-acetylcysteine, mannitol, and theophylline when compared with periprocedural hydration alone.
Caution
-
Available studies examined laboratory end points (such as an increase in serum creatinine levels) rather than clinical end points (such as dialysis or death).
—The Editors
Contrast-induced nephropathy, defined as an increase in serum creatinine greater than 25% or 44.2 µmol/L (>0.5 mg/dL) within 3 days of intravascular contrast administration in the absence of an alternative cause, is the third most common cause of new acute renal failure in hospitalized patients (1, 2). Contrast-induced nephropathy develops in 0% to 10% of patients with normal renal function (3). However, the incidence may be as high as 25% in patients with preexisting renal impairment or certain risk factors, such as diabetes, congestive heart failure, advanced age, and concurrent administration of nephrotoxic drugs (3). Large doses of intravenous contrast and use of high-osmolar contrast agents in patients with renal impairment also increase the risk for contrast-induced nephropathy (4–6). High-osmolar contrast agents are more rarely used now. The risk difference between iso-osmolar agents, such as iodixanol, and low-osmolar agents, such as iopamidol, ioxaglate, or iohexol, is less clear (7–9). Most episodes of contrast-induced nephropathy are not detected clinically because patients are asymptomatic. However, contrast-induced nephropathy may increase the risk for renal failure and is associated with dialysis, prolonged hospital stay, increased health care costs, potentially irreversible reduction in renal function, and death (10).
Use of preprocedural fluids and low-osmolar or iso-osmolar contrast agents has been shown to decrease the risk for contrast-induced nephropathy (11–13). These measures suffice for many patients; however, the risk is reduced but not eliminated in some patients—even when iso-osmolar contrast is used (14, 15). Other studies have evaluated the use of N-acetylcysteine, theophylline, fenoldopam, and other agents as preventive strategies in contrast-induced nephropathy; the results have been heterogeneous and are difficult to compare across the different treatment strategies. Given the widespread use of iodinated intravascular contrast agents, an improved understanding of the potential value of these agents has important patient safety and cost implications.
We conducted a meta-analysis of the literature to quantify the effects of individual strategies on the prevention of contrast-induced nephropathy and to facilitate comparison of preventive effects across strategies.
Methods
Study Search Strategy
We performed a computerized search by using standard meta-analytic techniques (16) to identify relevant articles in MEDLINE (from 1966 through 3 November 2006), EMBASE (1980 through November 2006), PubMed, Web of Knowledge (Current Contents Connect, Web of Science, BIOSIS Previews, and ISI Proceedings for the latest 5 years), and the Cochrane Library databases. For the MEDLINE search, we used the following combination of keywords: [renal failure or kidney failure to include all subheadings] and [contrast media or iopamidol or iodine or ioxaglic acid or iodine compounds or iohexol or urography or drug hyper sensitivity or tomography, X ray computed or diatrizoate] and [hydration or fluid therapy or water or dehydration or skin or nutritional support or body water] and [clinical trial or randomized controlled trial] and [prospective trial or prospective studies or clinical trials] and [adult or middle aged or aged] and [N-Acetylcysteine or acetylcysteine] or [theophylline] or [mannitol] or [dopamine] or [fenoldopam] or [bicarbonate]. For the PubMed, Cochrane Library Database, and Web of Knowledge searches, we used the search words renal failure, contrast medium, hydration, randomized controlled trial, N acetyl cysteine, Theophylline, Mannitol, Fenoldopam, Dopamine and Bicarbonate. We included English-, French-, German-, Spanish- and Italian-language studies and clinical trials and excluded review articles and nonhuman studies. We combined this strategy with a manual search of reference lists from identified articles.
Study Selection
We included a study if 1 of the treatment groups received N-acetylcysteine, theophylline, fenoldopam, iloprost, statin, dopamine, trimetazidine, bicarbonate, ascorbic acid, furosemide, or mannitol. Criteria for inclusion were randomized, controlled trials that compared treatment with control; used intravenous iodinated contrast; explicitly defined contrast-induced nephropathy; and sufficiently reported data to construct a 2 × 2 table and calculate the primary effect measure (relative risk reduction). Where data were missing, we contacted the original authors for the relevant information.
Data Extraction
One reviewer examined the abstracts to determine whether the study met the inclusion and exclusion criteria. Two reviewers separately abstracted complete articles according to a standardized form for studies meeting criteria. Abstracted information included patient characteristics (mean age, proportion of men and patients with diabetes mellitus or hypertension, and mean baseline creatinine level), type of radiologic or cardiologic imaging, inclusion and exclusion criteria, type of contrast media and dose used, periprocedural hydration, specific definition of contrast-induced nephropathy, prophylactic agent dose and route, and serum creatinine level at baseline and at 48 hours after contrast injection.
Analysis of Renoprotective Agents
The primary outcome was the development of contrast-induced nephropathy, defined as an absolute increase in baseline serum creatinine greater than 44.2 µmol/L (>0.5 mg/dL) or a relative increase greater than 25% at 48 hours after contrast injection. For trials missing this datum, we contacted the original authors to get the number of patients with this outcome. We calculated individual study relative risks and 95% CIs before aggregation. Subsequently, we obtained overall and subgroup summary risk ratios by random-effects modeling of the binary data from the multiple 2 × 2 tables. We used the method of DerSimonian and Laird (17), with the estimate of heterogeneity taken from the inverse variance fixed-effect model. We used the metan module in Stata, version 9.0 (Stata, College Station, Texas), to perform data synthesis.
We performed subgroup evaluation of each therapeutic regimen. In studies comparing 2 dosage regimens of the same intervention with a single control group (18–20), we considered the same-study dosage groups as representing a single intervention to avoid double-counting of shared control observations. When we identified only 1 study that examined a given therapy, we assigned that study to a group termed “other” and pooled data from all such studies together. This group included 1 study each on the use of iloprost; trimetazidine; mannitol; bicarbonate; ascorbic acid; and combinations of furosemide, dopamine, and mannitol and furosemide and dopamine. We used relative risk ratios to estimate the treatment effects.
Assessment of Methodological Quality
Criteria for quality assessment included concealment of allocation, similarity of both groups at baseline regarding prognostic indicators, eligibility criteria, blinding of patient, blinding of care provider, blinding of outcome assessor, point estimates and measures of variability for the primary outcome measure, and inclusion of an intention-to-treat analysis (21). Any disagreements in abstracted data between the reviewers were adjudicated by a third reviewer. We explored potential heterogeneity in estimates of treatment efficacy attributable to each quality criterion by using meta-regression.
Assessment of Heterogeneity
We used Forest plots to visualize the extent of heterogeneity among studies. We also examined I2, a standard test for heterogeneity that measures the degree of inconsistency across studies. I2 values, which range from 0% to 100%, describe the proportion of variation in treatment effect estimates that is due to genuine variation rather than sampling error (22). A value of 0% indicates no observed heterogeneity. Higgins and colleagues (22) suggest describing I2 values of 25%, 50%, and 75% as low, moderate, and high, respectively. We obtained the group-specific and overall I2 as standard output of the metan program.
We performed an Egger precision-weighted linear regression test as a statistical test of funnel plot asymmetry and publication bias (23).
All statistical analyses were performed with Stata.
Results
Study Identification
Our initial search yielded 619 citations and references. We excluded 531 studies on the basis of our criteria, including nonclinical trials; trials not conducted on humans; trials not reported in English, French, German, Spanish, or Italian; trials reporting only nonnephropathy outcomes; and trials using nonclinical outcome measures, leaving 88 studies that met the inclusion criteria (Figure 1). We reviewed abstracts from the 88 articles and excluded an additional 23 trials, including nonrandomized clinical trials; trials not conducted on humans; trials not reported in English, French, German, Spanish, or Italian; trials reporting only nonnephropathy outcomes; and trials that used nonclinical outcome measures, leaving 65 studies for full publication review. The full articles were then reviewed, and a further 24 studies were excluded for reasons similar to those just mentioned. After the final screening, 41 randomized clinical trials met our inclusion criteria (18–20, 24–59), involving 6379 patients who had elective radiographic procedures involving contrast agents.
Study Characteristics
The trials were published between 1994 and 2006, and the Table shows their characteristics. Fifteen trials were performed in the United States (9, 20, 24, 25, 30–32, 34, 44, 48, 49, 51, 52, 58, 59), and 26 trials were performed elsewhere (18, 19, 26–29, 33, 35–43, 45–47, 50, 53–57, 59). Thirty-four trials evaluated patients with impaired renal function (9, 19, 20, 24–29, 32–46, 48–52, 56–59), defined as serum creatinine levels greater than 106.1 to 132.6 µmol/L (>1.2 to 1.5 mg/dL). We had insufficient data to separately evaluate patients with normal renal function. Only 3 trials evaluated patients with normal and impaired renal function, and 2 trials evaluated only patients with normal renal function (30, 31, 47, 53, 54). One trial evaluated patients having computed tomography (58); the rest evaluated patients having cardiac catheterization. The average age of the study patients was greater than 65 years in all but 8 studies (9, 18, 29, 37, 40, 41, 45, 48), and all studies included patients with diabetes. Dosing regimens for each trial are detailed in the Table. The outcome measure of contrast-induced nephropathy was reported in all studies. Changes in serum creatinine levels were reported at 48 hours in most trials (9, 18–20, 24–47, 49–59), although we used outcomes reported at 72 hours for 1 trial (48).
Analysis of Renoprotective Agents
Of the evaluated agents, N-acetylcysteine significantly decreased the risk for contrast-induced nephropathy compared with saline alone (relative risk, 0.62 [95% CI, 0.44 to 0.88]) (Figure 2). Although seemingly renoprotective, the effects of theophylline on nephropathy prevention were not significant (relative risk, 0.49 [CI, 0.23 to 1.06]). In the heterogeneous group of treatments for which only a single study was identified (labeled “other”), only ascorbic acid (relative risk, 0.46 [CI, 0.23 to 0.90]) and bicarbonate (relative risk, 0.12 [CI, 0.02 to 0.95]) significantly reduced contrast-induced nephropathy. Furosemide (relative risk, 3.27 [CI, 1.48 to 7.26]) increased the risk for contrast-induced nephropathy.
The intervention and control columns show the number of events among the total number of participants randomly assigned to the group for each study. We estimated heterogeneity within subgroups by using the I2 statistic.
Assessment of Methodological Quality
The Appendix Table presents the quality characteristics of each study. Most studies included patients with similar baseline characteristics (94%) or specific inclusion characteristics (90%). Most also presented variance estimates of treatment effects (59%) or blinding of patients to treatment (51%). Fewer than half of the studies reported concealment of allocation (47%) or blinding of care providers to treatment (43%). Few studies noted outcome evaluation by individuals blinded to treatment assignment (6%) or an intention-to-treat design (8%). In exploratory analysis, only the quality characteristic of explicitly stating specific inclusion criteria (P = 0.007) independently contributed to heterogeneity across study efficacies.
Assessment of Within-Group Heterogeneity and Publication Bias
Treatment effect estimates within the N-acetylcysteine group showed moderate heterogeneity (I2 = 55%; P < 0.001). As expected, we found a moderate to high level of heterogeneity among the pooled studies that each examined a different therapy (labeled “other”) (I2 = 61%; P = 0.024). No other groups demonstrated significant within-group heterogeneity (Figure 2). No significant publication bias was discerned (bias coefficient, −0.55; P = 0.20).
Discussion
In our meta-analysis of 41 randomized trials, we found that preprocedural treatment with N-acetylcysteine effectively reduced the risk for contrast-induced nephropathy. Theophylline also produced larger risk reductions than previously mentioned; however, the effects of this agent were not significant. Not all agents analyzed had beneficial effects—fenoldopam; furosemide; mannitol; and the combination of furosemide, dopamine, and mannitol had odds ratios greater than 1. Our findings for N-acetylcysteine support previous studies (60–63). To date, no meta-analyses have studied preprocedural dopamine or statins for the prevention of contrast-induced nephropathy. Our findings for theophylline support previous studies that showed a risk reduction (64). However, the effects of theophylline were not statistically significant in our study. In contrast, Ix and colleagues (64) found borderline statistical significance when they limited their analysis to studies using concomitant intravenous fluids or contrast volumes greater than 100 mL and no statistical significance when their analysis was limited to studies of only coronary angiography patients or where the theophylline was given within 1 hour of the procedure. Bagshaw and Ghali (65), however, did not find a statistically significant effect, similar to our findings.
N-Acetylcysteine is extremely inexpensive at 23 cents for a 500-mg tablet (price as of 17 January 2007 at http://www.shopping.com), is readily available, and is easily administered. Side effects and drug interactions are very rare with continued use and are highly unlikely to result from the limited use for renal protection. Therefore, although no formal cost-effective analysis has been performed to date, these findings support the use of N-acetylcysteine in selected at-risk patients.
Hydration and iso-osmolar or low-osmolar contrast agents, such as iodixanol, are all associated with a decreased incidence of contrast-induced nephropathy in patients with renal impairment (creatinine clearance <1 mL/s [<60 mL/min]) (44). In a recent meta-analysis, McCullough and colleagues (15) found that although low-osmolar contrast agents reduced the risk for contrast-induced nephropathy by two thirds, they did not totally eliminate the risk. Thus, protective agents must still be considered for patients with severe renal impairment who are to receive large volumes of contrast agents.
Our meta-analysis has several limitations. All included trials evaluated the surrogate end point of contrast-induced nephropathy as the primary outcome. Contrast-induced nephropathy was defined as an increase in serum creatinine of more than 44.2 μmol/L (>0.5 mg/dL) or 25% from baseline values, which represents a minor deterioration in renal function in patients with chronic renal failure. Even in high-risk patients, contrast-induced nephropathy is almost always transient and only rarely requires dialysis. Only the trial by Kay and colleagues (46) examined length of hospital stay as an end point and found a significant reduction in length of stay among patients given N-acetylcysteine. Despite the reported association of contrast-induced nephropathy with impaired outcomes, no trial has examined clinical end points, such as dialysis dependency or in-hospital morbidity and mortality. The clinical relevance of the renoprotective effects of N-acetylcysteine, dopamine, and other agents is therefore debatable, whereas periprocedural hydration is of proven benefit (66). In addition, it is possible that we did not detect a significant effect for theophylline because of study heterogeneity or insufficient data.
Because we primarily identified and used published studies, our results are weighted on the findings of published trials. The exclusion of unpublished data is generally associated with an overestimate of the true effect in meta-analysis (67). The single most common reason for inability to publish a trial is the lack of statistical significance, although some have suggested that the quality of unpublished data is not comparable to that accepted by peer-reviewed journals (68). In addition, many of the included studies did not have high quality scores, and many did not specify that they met the quality criteria, with the true quality remaining uncertain.
Strengths of our study include the comprehensive search strategy and the careful statistical methods used. We identified 41 trials with a total of 6379 patients and evaluated multiple therapeutic agents within 1 analysis framework, allowing side-by-side comparison of the efficacies across agents.
Our meta-analysis shows that N-acetylcysteine is the most effective agent for preventing contrast-induced nephropathy in patients with chronic renal insufficiency. Whether this risk reduction translates into a benefit in clinical outcomes remains to be proven. The reported association of contrast-induced nephropathy with increased morbidity, mortality, and hospital stay might justify the use of N-acetylcysteine as a routine intervention for prophylaxis of contrast-induced nephropathy, given that N-acetylcysteine is readily available and inexpensive and has a favorable side effect profile.
The results of this meta-analysis should be evaluated in head-to-head empirical studies of active agents to identify the most efficacious regimen for preventing contrast-induced nephropathy. However, our findings indicate that the use of such oral agents as N-acetylcysteine is reasonable in high-risk patients who are to receive large or repeated volumes of contrast agents. We believe that the lack of significant side effects and the low cost justifies use of these agents while empirical data on clinical outcomes mature.
Article and Author Information
-
Grant Support: Funded in part by the National Institutes of Health and the National Cancer Institute (grant 1 K07 CA108664 01A1) and the General Electric–Association of University Radiologists Radiology Research Academic Fellowship.
-
Potential Financial Conflicts of Interest: None disclosed.
-
Requests for Single Reprints: Aine M. Kelly, MD, MS, Department of Radiology, Division of Cardiothoracic Radiology, University of Michigan Hospitals, B1 132K Taubman Center, 1500 East Medical Center Drive, Ann Arbor, MI 48109; e-mail, ainekell{at}med.umich.edu.
-
Current Author Addresses: Drs. Kelly and Cronin: Department of Radiology, Division of Cardiothoracic Imaging, University of Michigan, 1500 East Medical Center, Ann Arbor, MI 48109.
-
Dr. Dwamena: Division of Nuclear Medicine, University of Michigan, 1500 East Medical Center, Ann Arbor, MI 48109.
-
Dr. Bernstein: Department of Internal Medicine, University of Michigan, 1500 East Medical Center, Ann Arbor, MI 48109.
-
Dr. Carlos: Department of Radiology, Division of Magnetic Resonance Imaging, University of Michigan, 1500 East Medical Center, Ann Arbor, MI 48109.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
Related Clinical Content
