Author + information
- Somjot S. Brar, MD, MPH∗ ()
- Regional Department of Cardiac Catheterization, Kaiser Permanente Los Angeles Medical Center, Los Angeles, California
- ↵∗Address for correspondence:
Dr. Somjot S. Brar, Regional Department of Cardiac Catheterization, Kaiser Permanente Los Angeles Medical Center, 4867 Sunset Boulevard, 3rd Floor, Room 3753, Los Angeles, California 90027.
N-acetylcysteine and intravenous sodium bicarbonate infusion are 2 treatments that have been extensively studied for contrast-induced acute kidney injury (CI-AKI) prevention. Over time, the published research has evolved to confirm that these therapies are ineffective, a process that was slow and unnecessarily confusing. A careful appraisal of the clinical trial data on these treatments illustrates a cautionary tale of the effect of small trials and meta-analyses of these studies.
The purported benefits of sodium bicarbonate infusion and N-acetylcysteine were first described approximately 10 and 18 years ago, respectively. Since then, more than 80 clinical trials of these 2 therapies have been performed, including in combination. Summarizing a large number of studies in a traditional review is challenging; nevertheless, an appraisal of the studies is necessary and if appropriately performed is likely to be informative. A common contemporary approach is to perform a comprehensive search and review of all relevant studies, referred to as a systematic review, and, if appropriate, follow with a meta-analysis. In the simplest terms, a meta-analysis is a weighted average of results from multiple studies yielding a summary estimate of effect and corresponding confidence interval. In contrast to a typical review, in which the author is not obligated to identify all relevant studies and is permitted to selectively report certain trials and may emphasize those that are deemed to be of greater importance, the meta-analysis provides an objective framework that lessens the risk for bias.
The numerous clinical trials of these 2 therapies have been aggregated and analyzed in more than 30 meta-analyses, of which many conclude that these treatments are beneficial, often with a high degree of statistical significance (p < 0.01). A particular challenge in the interpretation of CI-AKI prevention research, including meta-analyses, is the abundance of small trials. Summary estimates in smaller clinical trials have a higher degree of uncertainty, and the treatment effect must be somewhat extreme to achieve statistical significance (p < 0.05). For example, 1 additional event in the experimental treatment group is sufficient to negate the purported benefit of these treatments in some trials. Small trials may also be of lower methodologic quality. Moreover, small trials with positive findings are more likely to be published than similarly sized trials with negative findings. This phenomenon is well recognized in medical research and can lead to an overestimation of the treatment effect (1,2). This small study bias can be somewhat attenuated by including small negative unpublished studies in meta-analyses; unfortunately, such studies can be exceedingly difficult to identify. Public registers of clinical trials were meant to address this, in part, but their impact remains to be seen.
Over time, larger trials found N-acetylcysteine and sodium bicarbonate to be ineffective for CI-AKI prevention. For example, 5 of the first 7 randomized trials evaluating sodium bicarbonate infusion versus sodium chloride showed statistically significant results in favor of sodium bicarbonate. In the subsequent 7 trials, which included larger studies, none observed a statistically significant benefit for sodium bicarbonate infusion. Interestingly, inclusion of these larger negative studies did not neutralize the positive findings of the initial group of smaller studies in meta-analyses. Summary estimates reported in many meta-analyses continued to suggest a statistically significant benefit for sodium bicarbonate. Remarkably, the treatment effects in the small positive trials are so great that the aggregate summary estimate remains significantly in favor of sodium bicarbonate infusion. This should not be construed as a failure of the meta-analysis concept itself; after all, this study design is not intended for generating summary estimates for studies with such extreme differences in outcomes. The meta-analytic framework does, however, allow identification of this phenomenon by quantifying heterogeneity between studies. A uniform finding in these meta-analyses is moderate to high levels of heterogeneity among studies of these therapies. In general, the likelihood of drawing correct inferences from a meta-analysis decreases with increasing heterogeneity.
Exploring causes of heterogeneity, clinical and statistical, among trials can provide much-needed insight into the trial data. For example, study sample size has been shown to be an important source of heterogeneity among these trials (3). When studies are stratified by sample size, the benefit of sodium bicarbonate is observed only in the group of studies with smaller sample sizes, and no benefit is seen in the larger studies. Recognizing the effect of sample size on outcomes can have an important impact on formulating a recommendation for a treatment. Unfortunately, not enough of an emphasis is placed on the heterogeneity assessment in some meta-analyses, with reports instead focusing on the summary estimate and confidence interval despite the known uncertainty of these estimates in this setting. Unlike the well-outlined process for calculating a summary estimate, evaluating sources of heterogeneity can be more complex and may require some creativity.
Over time, other randomized trials continued to suggest that neither of these 2 therapies were effective (4–7). This led to a marked decrease in the use of N-acetylcysteine and sodium bicarbonate, but there remained lingering doubt given the preponderance of meta-analyses perpetuating the myth of benefit. A large trial was deemed necessary, in part, to resolve the confusion propagated by these studies.
The PRESERVE (Prevention of Serious Adverse Events Following Angiography) trial provides further evidence that N-acetylcysteine and sodium bicarbonate are not effective for the prevention of CI-AKI. In this issue of JACC: Cardiovascular Interventions, Garcia et al. (8) report on the effect of sodium bicarbonate infusion and N-acetylcysteine for the prevention of CI-AKI in patients undergoing percutaneous coronary intervention. Of the 4,993 patients in the main trial, 1,161 underwent percutaneous coronary intervention. This population is of particular interest given that patients undergoing percutaneous coronary intervention are exposed to higher contrast volumes and thus are at higher risk for CI-AKI. For the primary composite endpoint of death, need for dialysis, or persistent kidney impairment at 90 days, the event rates were not significantly different for sodium bicarbonate versus sodium chloride infusion (2.6% vs. 4.0%) or N-acetylcysteine versus placebo (3.8% vs. 2.8%). The incidence of the secondary endpoint, a 25% or 0.5 mg/dl increase in serum creatinine at day 4, was 11% to 12% and not significantly different between treatments. The observed CI-AKI rate is similar to multiple other randomized CI-AKI prevention trials. It is encouraging that the incidence is now largely reproducible across several well-conducted trials. The data reaffirm that CI-AKI remains among the most common complications of cardiac catheterization, especially in patients with estimated glomerular filtration rates of 60 ml/min/1.73 m2 or less. These high rates of CI-AKI underscore the need for effective therapies for prevention; aggressive intravenous fluid administration remains the most promising treatment (9).
The PRESERVE trial is the largest CI-AKI prevention trial to date. The strengths of a large trial include the ability to focus on clinically meaningful adverse events as opposed to surrogates. The PRESERVE trial’s primary endpoint was a composite of all-cause mortality, need for renal replacement therapy, or persistent kidney impairment at 90 days. A missed opportunity may have been the omission of myocardial infarction in the composite endpoint. Prior studies have shown that patients who develop CI-AKI have an increased risk for adverse cardiovascular events, myocardial infarction in particular. Another benefit of a clinically oriented endpoint is a reduction in missing outcomes data. In most CI-AKI trials, the primary endpoint requires multiple serum creatinine assessments after cardiac catheterization, and about 10% of trial participants will have some missing laboratory data, even with the use of home phlebotomy services (10). The primary outcome was not available for 4% of the study participants in the trial, which is a general improvement over CI-AKI trials that recruit patients undergoing both in-hospital and ambulatory procedures.
The PRESERVE trial was born out of considerable confusion in published medical research regarding the effectiveness of sodium bicarbonate and N-acetylcysteine. Much of the purported benefit was reinforced by the publication of serial meta-analyses that included numerous small trials, emphasized statistically significant summary estimates, and often downplayed or disregarded the observed high levels of heterogeneity among trials. Careful assessment of heterogeneity should provide greater insight into the results, increase the scientific value of the report, and lessen the likelihood of erroneous conclusions. There is more to a meta-analysis than simply producing a single estimate of effect. The PRESERVE trial closes the lengthy chapter on N-acetylcysteine and sodium bicarbonate for CI-AKI prevention almost 2 decades after it began. The arc of the research universe can be long, but it bends toward the truth.
↵∗ Editorials published in JACC: Cardiovascular Interventions reflect the views of the authors and do not necessarily represent the views of JACC: Cardiovascular Interventions or the American College of Cardiology.
Dr. Brar has reported that he has no relationships relevant to the contents of this paper to disclose.
- 2018 American College of Cardiology Foundation
- Brar S.S.,
- Hiremath S.,
- Dangas G.,
- Mehran R.,
- Brar S.K.,
- Leon M.B.
- Maioli M.,
- Toso A.,
- Leoncini M.,
- et al.
- ACT Investigators
- Solomon R.,
- Gordon P.,
- Manoukian S.V.,
- et al.
- Garcia S.,
- Bhatt D.L.,
- Gallagher M.,
- et al.