N-Acetylcysteine Plus IV Saline Versus IV Saline With or Without Placebo

Our main meta-analyses indicated that compared with IV saline alone, low-dose N-acetylcysteine (1200 mg/daily or less) had a borderline clinically important decrease in CIN in patients receiving either intra-arterial or IV contrast media (risk ratio 0.75; 95 % CI: 0.63 to 0.89) or when either low (1200 mg daily or less) or high-dose (> 1200 mg daily) N-acetylcysteine was used in patients receiving LOCM (risk ratio 0.69; 95 % CI: 0.58 to 0.84). The strength of evidence was low for the first comparison (low-dose N-acetylcysteine) and moderate for the second comparison (in patients receiving LOCM), primarily due to limitations in the quality of studies and inconsistency in results. In comparison, a highly cited meta-analysis published by the Annals of Internal Medicine in 2008 reported a relative risk of 0.62 (95% CI 0.44 to 0.88) for preventing CIN when studies were combined irrespective of the dose of N-acetylcysteine.²¹⁵ An older meta-analysis, published in Lancet in 2003, reported a relative risk of 0.44 (95% CI 0.22 to 0.88) for preventing CIN with N-acetylcysteine.²¹⁶ In a recent meta-analysis published in PLoS One in 2013, the risk ratio for CIN with N-acetylcysteine was 0.68 (95% CI 0.46 to 1.02).¹¹ One study has questioned whether N-acetylcysteine is effective at preventing CIN or if it simply reduces serum creatinine.²¹⁶ This is an important finding; however, the reduction in serum creatinine reported as significant was measured at 4 hours, and it was insignificant at 48 hours, which was the timeframe for the assessment of CIN in this report.

Our review included many more studies than any of those reviews, and showed a much smaller effect for both high-dose and low-dose N-acetylcysteine. Our sensitivity analysis showed a clinically important benefit (greater than 25% relative risk reduction) with N-acetylcysteine plus IV saline compared with IV saline alone in reducing the incidence of CIN when LOCM was used, but not when IOCM was used. Although this difference could be due to methodological differences between the two sets of studies, the results were relatively consistent among the studies involving use of LOCM, while the 95% confidence interval of the aggregate risk ratio from studies involving use of IOCM ruled out a clinically important benefit. These findings raise the possibility that the effectiveness of N-acetylcysteine could vary by type of contrast media.

The risk of CIN generally is considered to be higher with intra-arterial than with IV administration of contrast media, raising the possibility that N-acetylcysteine could have greater benefit in patients receiving intra-arterial contrast media. When we stratified the analysis by route of administration of contrast media, the pooled risk ratios suggested the possibility of a difference in the effectiveness of N-acetylcysteine in the direction of having a greater effect with IV than intra-arterial contrast media: high-dose N-acetylcysteine (pooled risk ratio 0.78 versus 0.55, respectively for intra-arterial versus IV administration); low-dose N-acetylcysteine (pooled risk ratio 0.77 versus 0.62, respectively for intra-arterial versus IV administration). However, fewer studies have involved IV contrast media than intra-arterial contrast media, with resulting CIs that were much wider for studies involving IV contrast media than for studies involving intra-arterial contrast media. Thus, the evidence is insufficient to determine whether the effectiveness of N-acetylcysteine in preventing CIN differs according to whether IV versus intra-arterial administration was used. In contrast to a previous meta-analysis which reported a pooled relative risk of 0.20 (95% CI: 0.07 to 0.57) for preventing CIN in patients receiving IV contrast for a CT scan,²²⁵ our analyses did not demonstrate a clear benefit of N-acetylcysteine for patients receiving IV contrast media. The previous meta-analysis included studies in which CIN was defined not only by change in serum creatinine but also by changes in cystatin C. In addition, in some of the studies included in this meta-analysis, the time frame for the definition of CIN was longer than 72 hours. These differences may explain why the previous analysis came to a different conclusions. More studies could help to determine whether there is a clinically important benefit of administering N-acetylcysteine to patients receiving an imaging test when the contrast media is administered IV.

Pre-test serum creatinine level may be an important covariate associated with CIN. Wu et al., 2013²²⁵ found that the risk of CIN was reduced with N-acetylcysteine in patients with a baseline serum creatinine greater than 1.2 mg/d. They did not find a statistically significant benefit of N-acetylcysteine in patients with a baseline serum creatinine less than 1.2 mg/d. When we performed a sensitivity analysis similar to what Wu et al performed, we found that the mean baseline serum creatinine for each study was not associated with a difference in the effect of N-acetylcysteine on the incidence of CIN. This difference in results can be explained by somewhat different criteria for inclusion in the review, and our inclusion of studies that showed no benefit with N-acetylcysteine. Since it is plausible that pre-test serum creatinine level may be associated with an increased risk of CIN, further studies could help to elucidate whether N-acetylcysteine would be beneficial in patients with a high preexisting serum creatinine level.

Because of the great variability in study protocols as well as the conflicting results of the available clinical trials, the recommendations for N-acetylcysteine administration vary by organization. For example, the joint American College of Cardiology/American Heart Association 2012 guidelines do not recommend the use of N-acetylcysteine for patients receiving intra-arterial contrast in cardiac procedures.²²⁶ In comparison, the 2012 Kidney Disease: Improving Global Outcomes (KDIGO) Clinical Practice Guideline for Acute Kidney Injury suggests using oral N-acetylcysteine with IV fluids in patients at increased risk for CIN, while acknowledging that the quality of evidence is very low.²²⁶ The KDIGO recommendation is based on the argument that although the overall benefit for N-acetylcysteine is not consistent or overwhelming, it is inexpensive, appears to be safe, and has been shown in many studies to have an effect in reducing the risk of CIN.²¹ Our analysis reveals a clinically important effect of low-dose N-acetylcysteine and is consistent with the KDIGO guidelines. Although N-acetylcysteine is inexpensive, and appears to be safe, the evidence may not be strong enough to support routine use, especially without stronger evidence on clinical outcomes other than the incidence of CIN.

Sodium Bicarbonate Versus IV Saline

Our meta-analysis demonstrated with low strength of evidence that IV sodium bicarbonate did not differ from IV saline in the incidence of CIN, although the confidence interval for the aggregate effect estimate was not precise enough to rule out the possibility of a clinically important benefit with sodium bicarbonate. The strength of evidence also was low that IV sodium bicarbonate did not produce a clinically important reduction in mortality or the need for renal replacement therapy when compared with IV saline. However, we found evidence for possible benefit of using sodium bicarbonate to prevent CIN in patients receiving LOCM although the observed difference was not statistically significant. Our main result is contrary to the conclusion of a recent meta-analysis of 19 clinical trials ¹⁰⁷ investigating the effect of IV sodium bicarbonate. Our analysis included 19 RCTs which compared only IV sodium bicarbonate versus IV saline. In comparison, 5 of the 19 trials in the other meta-analysis were of combination regimens of IV sodium bicarbonate and N-acetylcysteine which may have biased the results in favor of sodium bicarbonate. This difference in the included studies may help to explain why we did not find a clinically important effect favoring IV sodium bicarbonate administration. Only two studies used IV contrast media administration, and hence it is difficult to draw a conclusion about the effect of bicarbonate administration on the prevention of CIN in patients receiving IV contrast media.^109,114

N-Acetylcysteine Plus IV Saline Versus IV Sodium Bicarbonate

We found seven RCTs^{36,46,56,58,70,74,132} and two observational studies^97,133 addressing the effects of N-acetylcysteine with concurrent administration of IV saline compared with IV sodium bicarbonate. However, the evidence was insufficient to support a conclusion about the comparative effectiveness of these two interventions in their ability to prevent CIN. We found no other meta-analyses on this head-to-head comparison. Limitations of the head-to-head comparison of N-acetylcysteine with concurrent administration of IV saline compared with IV sodium bicarbonate included the small number of studies, the varying regimens of fluid administration and N-acetylcysteine dosing, the variations in follow-up time, and variation in inclusion criteria which predispose to CIN, as we described in the results section. If additional studies are done to assess the comparative effectiveness of these two interventions, it would be important to focus on comparing IV sodium bicarbonate to N-acetylcysteine with IV saline especially in the setting of administration of LOCM, as both of these interventions demonstrated a clinically important benefit in this subgroup of patients. Again, it would be important to investigate this in patients with a high baseline serum creatinine in whom the risk of developing CIN is likely higher.

Statins

We found a clinically important protective effect against CIN when statins were administered in combination with IV fluids compared with IV fluids alone (8 RCTs), or in combination with N-acetylcysteine compared to N-acetylcysteine alone (5 RCTs), but the effect was only statistically significant in the latter comparison. We saw this treatment effect for both of the above comparisons in populations with chronic kidney disease,^{137,141,142,144,145,153,154,156-158} diabetes mellitus^145,153,158 cardiac disease,^146,152,156 and in general populations.^141,155

These results are consistent with five^227-231 out of six recent meta-analyses on the comparison of statins versus IV saline. The one recent meta-analysis that does not agree with the presence of a clinically important benefit included four studies and had a CI wide enough to not rule out a clinically important effect.²³² One of the meta-analyses showing significant decreases in CIN in the statin group did not show a decrease in CIN in patients with chronic kidney disease greater than stage 3.²²⁸

Currently, protocols for prevention of CIN in the United States do not include the use of statins. It may be time to reassess the role of statins in preventing CIN, especially since statins are readily available, easy to administer, and relatively inexpensive. Although our findings have moderate strength of evidence, there are also reasons to move forward cautiously. First, it is important to note that all studies evaluating the effect of statins to reduce the incidence of CIN were done using intra-arterial administration of contrast media. Hence, its protective effect against CIN for IV contrast media administration is not known. Second, it is possible that the findings reported in the studies of statins could be partly explained by a direct effect of statins on glomerular filtration rate that is independent of a protective effect on kidney function, as has been reported in one study.²³³

Adenosine Antagonists Plus IV Saline Versus IV Saline

Our analyses showed insufficient evidence to demonstrate an overall effect of theophylline or aminophylline plus IV saline when compared with IV saline alone for the prevention of CIN. There were wide variations in the effect estimates for individual studies, ranging from a ten-fold decrease in the risk of developing CIN with theophylline¹⁶⁸ to an almost 6-fold increase in the risk of developing CIN with theophylline.¹⁶⁷ Although our test of heterogeneity demonstrated that almost half of the uncertainty in the latter estimate could be explained by differences between studies, the p-value around this estimate was not statistically significant. Clinically, the variation could be explained by the heterogeneity of the populations in the studies, which ranged from patients with stable coronary artery disease⁶⁶ to those with moderate to severe chronic kidney disease.³¹ A previous meta-analysis showed that the administration of theophylline or aminophylline was associated with less of a decline in kidney function than if it was not given.²³⁴ However, IV saline was not administered in all the studies. In addition, the authors were unable to comment on the incidence of CIN based on the information provided in the articles. The authors of a meta-analysis looking at the effects of theophylline reported a trend toward a reduction in the incidence of CIN with theophylline use, but noted that the findings were inconsistent across studies.²³⁵

Overall, the evidence on the effects of adenosine antagonists on CIN was limited by medium study limitations based on the five criteria described in the methods for assessing risk of bias for individual studies, and considerable inconsistency and imprecision in the effect estimates. Only one of the relevant studies looked at IV contrast media administration; this may be relevant because the effect of prophylactic agents on CIN may differ depending on the route of contrast media administration, as mentioned previously.^6,236 The evidence also suffered from a lack of reporting on secondary outcomes such as need for dialysis, prolonged hospitalization, in-hospital mortality, and adverse drug effects. In this situation, the evidence seems insufficient to support much investment in further studies of the use of adenosine antagonists in preventing CIN.

Renal Replacement Therapy Versus IV Fluids

Hemodialysis and hemofiltration are invasive and expensive procedures that carry risks, but can remove some of the administered contrast. Our analyses did not demonstrate a decreased incidence of CIN in individuals receiving hemodialysis. However, limitations of the studies we found include small sample size, lack of rigorous controls, and uncertainties about the magnitude of delays between contrast administration and initiation of hemodialysis.

The studies comparing hemofiltration to IV saline reported that patients with severe chronic kidney disease have a lower risk for CIN with hemofiltration, especially when hemofiltration is started before the contrast media administration. These conclusions are limited by the fact that we only found two studies reporting this, and both were from the same authors and same institution. Another limitation is that the control groups received IV saline, while the patients undergoing hemofiltration received IV sodium bicarbonate as part of the procedure. Hemofiltration is expensive and requires patients to be admitted to and monitored in an intensive care unit. Furthermore, based on the design flaws in the reported trials and the paucity of studies examining this, further research is needed before proposing to expose patients to this invasive procedure as a prophylactic measure. It is important to note that the benefit of hemofiltration was only seen when it was initiated before the contrast media was given. Therefore, any added benefit may not be from removal of the contrast media, and it is proposed that the benefit may be secondary to the ability to provide more vigorous hydration. Clinical trials comparing hemofiltration with IV fluid protocols, and stronger trials that include investigation of the pharmacodynamics of the contrast media elimination during hemofiltration, may help better understand this procedure and its potential benefits.

Several additional limitations should be noted. Renal injury after contrast media administration occurs rapidly, and in these studies, hemodialysis may have been started too late to provide a significant benefit. Furthermore, the removal of creatinine by hemodialysis or hemofiltration limits the assessment of CIN as an outcome. While a false decrease in serum creatinine due to hemodialysis or hemofiltration is expected to bias the results toward a protective effect on the incidence of CIN, the results for hemodialysis actually suggested possible harm. The lack of a clinical benefit of renal replacement therapy may also be secondary to adverse events directly caused by the procedure (e.g., hypotension that may worsen kidney injury). Based on these results and the limitations and risks of the procedures, evidence is insufficient to support a clinically important benefit of renal replacement therapy.

Our findings coincide with the previously published systematic review by Cruz,²³⁷ which concluded that renal replacement therapy does not provide any protection against CIN. That systematic review included additional studies that did not meet our inclusion criteria (a total of nine RCTs and two non-randomized RCTs).

Ascorbic Acid Versus IV Fluids

We found eight RCTs evaluating the use of ascorbic acid to prevent CIN. Our results showed a clinically important and statistically insignificant effect on CIN when administered in combination with IV fluids compared with IV fluids alone, and an unimportant effect when administered in combination with IV fluids and compared with N-acetylcysteine. We saw these results in populations with chronic kidney disease undergoing intra-arterial contrast media administration for coronary procedures. Overall, the strength of evidence was low for the finding that ascorbic acid given with IV fluids did not have a clinically important effect on preventing CIN when compared with IV fluids alone.

These results are consistent with but not as strong as those shown by a recent meta-analysis on the same comparison by Sadat el al.^{181,227-232,238} Sadat el al. included data from nine RCTs comparing ascorbic acid with other treatments, and showed that patients receiving ascorbic acid had 33 percent less risk of CIN than those receiving other interventions. Our analysis included all of the five studies covered by Sadat et al. with the addition of one recent trial by Dvorsak et al.¹⁸³ Sadat et al.'s results may differ in that they included in their review the results of three abstracts with positive results and another study that compared ascorbic acid versus N-acetylcysteine.¹⁸⁸

Based on our review, the dose, timing and duration of ascorbic acid administration for prophylaxis against CIN did not affect the results. We also found that ascorbic acid did not have a clinically important benefit when compared with N-acetylcysteine.

Miscellaneous Comparisons

Many studies identified in our search did not fall into any of the main comparison groups listed above. For all of the miscellaneous comparisons, we were unable to support conclusions on the effectiveness of one intervention versus the other in preventing CIN.

Surprisingly little evidence exists on the comparative effectiveness of different regimens for giving fluids to patients receiving intra-vascular contrast media, despite the fact that current clinical practice often involves use of oral hydration alone. Oral hydration is a simple and potentially cost-effective strategy for preventing CIN, if proven to be as effective as IV saline. Unfortunately, few studies investigated oral hydration versus IV saline. Hence, more studies are needed to investigate the effectiveness of oral hydration versus IV saline, especially for intra-arterial contrast procedures such as coronary angiography.

Overall Limitations

One of the biggest limitations of our systematic review is the marked heterogeneity of the study protocols, populations, definitions of CIN, and follow-up times in the studies. The heterogeneity limited our ability to assess all of the comparisons of interest. Because studies varied in their use and definition of kidney insufficiency as an inclusion criterion, and often did not report results stratified by baseline kidney function, it was very difficult to assess how the effectiveness of interventions might vary according to baseline kidney function. The studies generally did not report results in a manner that would permit assessment of how the effects of interventions might differ by other characteristics of patients. Also, some of the studies we found were excluded because their definition of CIN did not match our pre-specified definition; this is one of the reasons why our findings sometimes differed from those of other meta-analyses. We also found that studies examining the risk of CIN with different types of contrast media generally provided little detail about clinical indications for the diagnostic or therapeutic procedures, whether imaging was done on an urgent or elective basis or other details such as the severity of renal impairment.

A major limitation is that it is very difficult to apply the existing evidence to patients receiving IV contrast media because the vast majority of studies focused on patients receiving intra-arterial contrast media. It is possible that the risk of CIN is very low with the LOCM and IOCM protocols now used routinely with IV imaging. However, studies generally did not report results in a way that allows for determination of how the effects of interventions might differ by differences in the type, route, or volume of contrast media used.

Another limitation is that studies were very inconsistent in reporting on longer-term clinical outcomes that would be more important to patients than whether their serum creatinine level increased or their glomerular filtration rate decreased. In general, the evidence was insufficient to support conclusions about the comparative effects of interventions on long-term clinical outcomes.

The results of the review are susceptible to bias in the available evidence. Many of the included studies had important study limitations, including problems with selection bias (from inadequate methods for allocating patients to treatment assignments), detection bias (from limited blinding of outcome assessments), attrition bias (from incomplete outcome assessments), and reporting bias (from selective reporting of outcomes). In addition, publication bias is a concern in this body of literature, as reported by Vaitkus et al., 2007²³⁹ who showed that the estimated effectiveness of N-acetylcysteine was greater in published articles than in unpublished abstracts. Despite our extensive search, we may have missed studies that have not been presented in a publicly available forum. Although we did not find evidence of asymmetry of results by study precision, statistical techniques have limited ability to detect publication bias. In general, we would expect the overall results of existing biases in this body of evidence to lead to an overestimate of the effectiveness of interventions.

Although we included a broad search, our meta-analysis may overestimate the effect of prevention strategies to reduce CIN if studies with negative results were not reported in the sources we searched. The studies span over two decades and over time there may have been changes in the practice of CIN prevention such as increased screening, variation in definition of acute kidney injury, and variation in hydration. Such changes could contribute to observed differences in outcomes.

It is beyond the scope of this report to make a recommendation about screening for CIN. However, we acknowledge that CIN might be under-reported because patients often are discharged immediately after the imaging procedures are done.

Finally, this comprehensive review highlights the generally low strength of evidence on interventions for preventing CIN, while indicating that the greatest reduction in risk of CIN has been achieved with low-dose N-acetylcysteine in patients receiving LOCM, or with statins plus N-acetylcysteine.

Future Research