Michael J Choi

Effectiveness of Prevention Strategies for Contrast-Induced Nephropathy: A Systematic Review and Meta-analysis

Iodine contrast medium is an essential component of many diagnostic and therapeutic procedures that involve medical imaging. One important side effect of iodine contrast is contrast-induced nephropathy (CIN), defined as an increase in serum creatinine levels of more than 25% or 44.2 mol/L (0.5 mg/dL) within 3 days of intravascular administration in the absence of an alternative cause (1). Because of increasing use of contrast media in radiologic and cardiologic procedures and the increasing prevalence of persons who are vulnerable to CIN (those with chronic kidney disease, diabetes mellitus, or hypertension, as well as elderly persons), kidney failure due to CIN is a substantial concern (2, 3). The reported incidence varies between 7% and 11% depending on the definition of CIN, study population, and setting (24). Some studies suggest that this incidence may be overestimated (4), especially when intravenous (IV) contrast media are used. An average additional cost of

Comparative Effect of Contrast Media Type on the Incidence of Contrast-Induced Nephropathy: A Systematic Review and Meta-analysis.

10345 is associated with a CIN-related hospital stay (5). Many strategies have been used to prevent CIN. They include oral hydration; volume expansion with sodium chloride or bicarbonate or both; administration of N-acetylcysteine; withdrawal of metformin, angiotensin-converting enzyme inhibitors, angiotensin IIreceptor blockers, or nonsteroidal anti-inflammatory drugs; hemofiltration or hemodialysis; statins; use of low-osmolar contrast media (LOCM), iso-osmolar contrast media (IOCM), or nonionic contrast media; and reducing the volume of contrast media administered. Despite these varied strategies, no clear consensus exists in clinical practice about the most effective intervention to prevent or reduce CIN. Many meta-analyses have been published, but almost all of them have focused on specific therapies or included subspecialtyspecific populations, which reduced the general applicability in clinical practice (611). The route of administration of contrast media may be a confounder because the baseline risk profile of patients having intra-arterial (IA) versus IV procedures may differ. Whether effectiveness of preventive interventions depends on the route of administration or the type of contrast media (IOCM or LOCM, the 2 types now in regular clinical use in the United States) is also unclear. We did a systematic review and meta-analysis to compare the preventive effect of strategies to reduce CIN, including subgroup analyses based on route of administration of contrast media or preventive strategies and the type of contrast media used. Methods We developed a protocol for this systematic review, which we posted online and registered in PROSPERO (CRD42013006217). The complete protocol is in the full report on which this article is based (12). Data Sources and Searches We searched MEDLINE, EMBASE, and the Cochrane Library through 30 June 2015 (Appendix Table). In addition, we searched the Scopus database for conference proceedings and other reports. We reviewed the reference lists of relevant articles and related systematic reviews to identify original articles that we might have missed. We also searched ClinicalTrials.gov and the U.S. Food and Drug Administration Web site. Appendix Table. Detailed Search Strategy Study Selection We included studies of patients of all ages. We identified observational and randomized, controlled trials (RCTs) that included administration of N-acetylcysteine, sodium bicarbonate, sodium chloride, statins, or ascorbic acid to prevent CIN. The study groups received IOCM or LOCM via IV or IA injection, CIN outcome was explicitly defined, and sufficient data were reported to calculate the primary effect measure (relative risk reduction of CIN). Secondary outcomes included the need for renal replacement therapy, cardiac events, and mortality. We included only RCTs for the meta-analyses. All data from other studies and other strategies to reduce CIN incidence (such as adenosine antagonists, renal replacement therapy, diuretics, antioxidants, and vasoactive agents) were analyzed and included in the full report (12). We excluded studies of high-osmolar contrast medium because it is no longer used in clinical practice in the United States. We did not contact the authors for original data. Data Extraction and Quality Assessment Two reviewers independently screened the titles and abstracts for eligibility and independently assessed each studys risk of bias by using 5 items from the Cochrane Risk of Bias Tool for RCTs (3). We solved disagreements by consensus or a third reviewer when consensus was not possible. At random intervals during screening, we did quality checks to ensure that eligibility criteria were applied consistently. The second reviewer checked the accuracy of the data extracted by the first reviewer. We graded the strength of evidence (SOE) on comparisons of interest for the key outcomes by using the grading scheme recommended in the Methods Guide of the Evidence-based Practice Center and considered the domains of study limitations, directness, consistency, precision, reporting bias, and magnitude of effect (13). Following the guidance of the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) Working Group (14), we rated evidence as precise if the total number of patients exceeded an optimum information size and the 95% CI excluded a risk ratio (RR) of 1.0. If the number of patients exceeded the optimum information size and the CI did not exclude the possibility of no difference (that is, RR of 1.0), we only rated the evidence as precise if the CI excluded the possibility of a clinically important benefit or harm (that is, RR <0.75 or >1.25). We classified the SOE pertaining to each comparison into 4 category grades (high, moderate, low, and insufficient) and assigned SOE grades by group consensus. The body of evidence was considered high-grade if study limitations were low and there were no problems in any other domain, and it was subsequently downgraded for each domain in which a problem was identified. If the magnitude of effect was very large, the SOE could be upgraded. Data Synthesis and Analysis The primary outcome was CIN, defined as an increase in serum creatinine levels of more than 25% or 44.2 mol/L (0.5 mg/dL) within 3 days of intravascular administration of contrast media. We calculated individual study RRs and CIs and then obtained overall and subgroup summary RRs by using a random-effects model. For large comparisons with 18 or more studies, we used the DerSimonianLaird random-effects estimator, with the estimate of heterogeneity taken from the inverse-variance, fixed-effect model (15). Although this method is often the standard estimator used by many meta-analysis software programs, it tends to underestimate CIs when fewer than 18 studies are compared (15). To compensate, we used the KnappHartung small-sample estimator approach for comparisons with fewer than 18 studies. This method allows for small sample adjustments to the variance estimates and calculates CIs on the basis of the t distribution with k 1 degrees of freedom (15). We used the Harbord modified test for small study effects to determine whether there was asymmetry in effect estimates. To assess the clinical importance of differences in CIN incidence, a binary outcome, we followed guidance for selecting a minimally important difference on the basis of the overall event rate in the studies (14). Our clinical experts decided that a relative risk reduction of 25% would be clinically important, which is consistent with the guidance that suggests a reduction of 20% to 30% in determining optimal information size. To account for factors that could be associated with a difference in CIN risk, we did a subgroup analysis on the basis of the route of administration (IA vs. IV) and type of contrast media (IOCM vs. LOCM), baseline serum creatinine level, sex, age, and prevalence of diabetes mellitus. A priori, we assumed that there would be considerable heterogeneity and therefore used a random-effects model. We also examined the I 2, which measures the degree of heterogeneity across studies (I 2 varies from 0% to 100%, with 0% indicating no heterogeneity). All statistical analyses were done in Stata, version 13 (StataCorp). Role of the Funding Source The Agency for Healthcare Research and Quality selected the topic and assigned it to the Johns Hopkins University Evidence-based Practice Center. The Agency assigned a task order officer who provided comments on the protocol and draft versions of the full evidence report. The Agency did not directly participate in the literature search, determination of study eligibility, data analysis or interpretation, or preparation of the manuscript for publication. Results The literature search revealed 86 RCTs on interventions for preventing CIN (Appendix Figure). These study results were published between 1998 and 2015. Six studies were funded by industry sources (1621), 16 were funded by academia or government agencies, 33 had no funding statement, and the remainder reported no conflicts of interest. All findings from these studies were analyzed and described in the full report (12). Appendix Figure. Summary of evidence search and selection. CIN = contrast-induced nephropathy; RCT = randomized, controlled trial. * 24647 gray literature results were also found. Total does not sum to 371 because the 2 reviewers were not required to agree on reasons for exclusion. N-acetylcysteine Plus IV Saline Versus IV Saline N-acetylcysteine is a direct scavenger of free radicals and improves blood flow through nitric oxidemediated pathways, which results in vasodilatation. As a result, both the antioxidant and vasodilatory properties of N-acetylcysteine are believed to protect against CIN. We included 54 RCTs on N-acetylcysteine plus IV saline versus IV saline with or without a placebo published since 2002 in the meta-analysis (1669). The studies varied widely in patient and intervention characteristic

Technical Expert Panel

Iodine contrast media are essential to many diagnostic and therapeutic procedures that involve imaging. An important potential side effect is contrast-induced nephropathy (CIN), most commonly defined in past studies as an increase in serum creatinine levels of more than 25% or 44.2 mol/L (0.5 mg/dL) within 3 days of intravascular contrast administration in the absence of an alternative cause (1). The precise mechanism of CIN is not entirely understood. The leading theories are that it results from hypoxic injury to the renal tubules induced by renal vasoconstriction or by direct cytotoxic effects of contrast media (2, 3); alternatively, some experts have arguedand recent evidence suggeststhat acute kidney injury occurring after intravascular contrast administration is caused by coexisting risk factors and is only coincidentally related to the contrast media, especially when administered intravenously (4, 5). Regardless of the cause, acute kidney injury after intravascular contrast administration remains a major concern for referring clinicians. Osmolality of contrast media is a key factor determining its tolerability (6). Since the 1990s, low-osmolar contrast media (LOCM) (2 to 3 times plasma osmolality) have been the standard of care for intravascular injection. A newer class of intravascular contrast, iso-osmolar contrast media (IOCM), is isotonic to plasma. Iodixanol is the only IOCM available for intravascular injection. The literature contains conflicting reports about whether iodixanol is associated with less risk for CIN than LOCM (7, 8). International guidelines from the Kidney Disease: Improving Global Outcomes Acute Kidney Injury Work Group mention IOCM and LOCM, but they do not make recommendations about selection between them (9). We did a systematic review of randomized, controlled trials (RCTs) to determine the comparative effects of different types of intravascular contrast media on CIN risk in patients having diagnostic imaging studies or image-guided procedures. We hypothesized that updating past reviews with more recent RCTs may help us understand conflicting reports about CIN risk. Some reports suggest that intra-arterial administration may be associated with greater risk than intravenous administration (4, 10, 11), so we also investigated whether the comparative effects vary according to the route of administration. Methods We developed and followed a review protocol, which is included in the full technical report on which this article is based (12). Data Sources and Searches We searched (without date or language restrictions) PubMed, EMBASE, and the Cochrane Library for RCTs published through 30 June 2015, as well as the Scopus database for conference proceedings and other reports (Appendix Table 1). We also reviewed the reference lists of relevant articles and related systematic reviews, searched ClinicalTrials.gov to identify ongoing studies, and asked an external expert panel to identify trials missing from our final list of eligible articles. Appendix Table 1. Search Strategy Study Selection We selected all RCTs that compared 1 or more contrast media types (LOCM or IOCM) with CIN incidence as the main outcome in patients having diagnostic imaging or image-based therapeutic procedures. Studies had to report the incidence of CIN based on serum creatinine levels or glomerular filtration rates at baseline and within 72 hours of contrast injection. Studies could involve patients of any age and preprocedure risk for CIN. There were no restrictions on how the contrast classes were compared, so studies comparing different types of LOCM and those comparing LOCM with IOCM were included. Two reviewers independently screened titles and abstracts to identify articles for inclusion. If necessary, the full text of articles was reviewed. Articles in a language other than English were excluded at the full-text level. Discrepancies between the 2 reviewers that remained after full-text review were resolved by consensus. At random intervals during screening, quality checks were done to ensure that eligibility criteria were applied consistently. Data Extraction and Quality Assessment For each eligible study, 1 investigator extracted pertinent data about study characteristics, patient population, imaging procedure type, comparisons, results, and statistical analysis. A second investigator reviewed the extracted data for accuracy. Discrepancies between the 2 investigators were resolved by consensus. Article and data management were done within the DistillerSR Web service (Evidence Partners). Two reviewers independently assessed each studys risk of bias using the following 5 items from the Cochrane Risk of Bias Tool for randomized studies: allocation sequence generation, allocation concealment, investigator blinding, incomplete outcomes, and selective outcome reporting (13). Discrepancies were resolved by consensus. Data Synthesis and Analysis When evaluating changes in CIN risk, we followed published guidelines for selecting a minimally important clinical difference based on the overall observed event rate in the studies (14). Taking into consideration the potential effect of CIN on a patients overall health and well-being, the clinical experts on our team decided that a 25% reduction in the relative risk for CIN would be clinically important, which is consistent with the published guidance suggesting a range of reduction in relative risk of 20% to 30% in determining optimal information size (14). For each comparison in our review, the study team assigned a grade (high, moderate, low, or insufficient) for the strength of evidence (SOE) associated with the entire group of studies that represented the particular comparison. Grades for SOE were assigned by consensus of the senior study team members (J.E., R.W., R.S., and E.B.). This grading scheme considered all of the following domains in the Agency for Healthcare Research and Quality guidelines for comparative effectiveness reviews: study limitations, precision, directness, consistency, reporting bias, and magnitude of effect (15). The study limitations domain was assessed by examining the risk-of-bias items for each study involved in the comparison. Study limitations were considered high if more than half of the studies in a group scored negatively in at least 1 of the risk-of-bias items, low if more than half of the studies in the group scored positively in all 5 risk-of-bias items, or medium if neither the high nor the low criteria were met. The precision domain was assessed by following guidance from the GRADE (Grading of Recommendations Assessment, Development and Evaluation) Working Group (14). We rated a group of studies as precise if the total number of patients exceeded the optimum information size (14) and the 95% CI excluded a pooled relative risk of 1.0. If the total number of patients exceeded the optimum information size but the CI did not exclude a relative risk of 1.0, we only rated the evidence as precise if the CI excluded the possibility of a 25% minimally important clinical difference as defined previously (relative risk <0.75 or >1.25). For the main outcome of interest, CIN, we calculated an optimum information size of 2000 patients based on an expected 0.1 probability of CIN and a minimally important relative risk of less than 0.75 or greater than 1.25. The SOE of a group of studies was graded high if the study limitations domain was considered low and all other SOE domains were scored positively. The SOE was downgraded for each domain that was scored negatively. If the magnitude of effect was very large, the SOE was upgraded. We did de novo meta-analyses of all studies on a given comparison if study heterogeneity was not important by clinical, qualitative, and statistical criteria (16). We calculated pooled risks by using a random-effects model and the DerSimonianLaird method (17). We used a funnel plot and the Harbord modified test for small study effects (18) to look for asymmetry in the reporting of results, which can be seen when publication bias exists. Analyses were done in Stata, version 13 (StataCorp). Role of the Funding Source The Agency for Healthcare Research and Quality selected the review topic and funded this research under a contract. A representative from the Agency provided technical assistance during creation of the full evidence report on which this article is based and provided comments on draft versions of that report (12). The Agency did not directly participate in the literature search; determination of study eligibility criteria; data collection, analysis, or interpretation; or preparation, review, or approval of the manuscript for publication. Results The literature search revealed 29 RCTs for summary and analysis (Figure 1). Five RCTs compared 2 or more types of LOCM in 826 patients (Appendix Table 2) (1923). Twenty-five RCTs compared the IOCM iodixanol with 1 or more types of LOCM in 5053 patients (Appendix Table 2) (19, 2447). One RCT reported data on both types of comparisons (19). In the 5 RCTs that compared LOCM, 4 studies scored negatively in 1 or more of the 5 risk-of-bias items (Appendix Table 3). In the 25 RCTs comparing iodixanol and LOCM, all studies scored negatively in 1 or more of the 5 risk-of-bias items (Appendix Table 4). Of the 29 RCTs included in our review, 14 (48%) studies (19, 20, 29, 3338, 4043, 45) received funding support from industry sources, all of which were contrast media manufacturers. Figure 1. Summary of evidence search and selection. * Sum of reasons for exclusion exceeds 443 because reviewers were not required to agree on the reason. Appendix Table 2. Study Characteristics Appendix Table 3. Risk of Bias for RCTs Comparing LOCMs* Appendix Table 4. Risk of Bias for RCTs Comparing Iodixanol and LOCMs* No study comparing 2 LOCM reported a statistically significant or clinically important difference between study groups in the incidence of CIN or a related measure of renal func