Introduction
The development of acute kidney injury (AKI) after cardiac surgery, known as cardiac surgery-associated acute kidney injury (CSA-AKI), is frequent. The incidence is estimated at 20–40% of patients undergoing cardiac surgery, with renal replacement therapy (RRT) required in 1.6–5.8% of cases [
1‐
3]. The occurrence of AKI is a major complication worsening the prognosis after cardiac surgery, increasing peri-operative mortality by a factor of 3–8 [
4‐
6]. In the long term, CSA-AKI is independently associated with increased risk of end-stage renal disease and death [
7,
8].
Many mechanisms are involved in the development of CSA-AKI. Cardiac surgery is frequently associated with low systemic output secondary to cardiopulmonary bypass (CPB) [
9,
10] leading to renal ischemia [
11]. In addition, ischemia–reperfusion injury after CPB may worsen renal injury because of oxidative stress and inflammatory response. Finally, hemolysis during CPB leads to nitric oxide scavenging, thus resulting in vasoconstriction and reduced renal perfusion [
12]
Many pharmacological and non-pharmacological interventions have been tested to reducing the incidence of CSA-AKI. However, randomized controlled trials (RCTs) have presented discordant results [
13]. Several meta-analyses were carried out, but none clearly demonstrated the superiority of a particular intervention to reduce the incidence of CSA-AKI [
14‐
16]. Recently, 2 network meta-analyses found that natriuretic peptide may be the best pharmacological intervention to prevent CSA-AKI [
17,
18]. However, these studies did not consider non-pharmacological interventions, which could be of interest in preventing CSA-AKI.
In this systematic review and meta-analysis, we assessed the effectiveness of non-pharmacological interventions to reduce the incidence of CSA-AKI.
Discussion
Our meta-analysis identified 10 non-pharmacological interventions to reduce CSA-AKI. We found the use of GDP, RIPc and pulsatile blood flow during CPB associated with significantly reduced incidence of CSA-AKI. However, no intervention had high quality of evidence. Conversely, 2 interventions (restrictive transfusion strategy and tight glycemic control) had high quality of evidence for a lack of effect on reducing CSA-AKI incidence. Although we restricted our search to recent trials, the definition of CSA-AKI was heterogeneous across trials with nearly half of the trials using a non-consensus definition of AKI, despite a trend to increased use of a consensus definition of AKI over the years.
This is the first meta-analysis synthesizing all non-pharmacological interventions during the peri-operative period to reduce the risk of CSA-AKI. Our search strategy was extensive, without language restriction, thus limiting the risk of missing important trials, with a robust and standardized methodology based on the Cochrane Handbook including the GRADE approach for evaluating the certainty of evidence. The protocol was prospectively registered in PROSPERO. Our meta-analysis brings new insights into the prevention of CSA-AKI, with important implications for clinical practice and future research. We provide a comprehensive summary of non-pharmacological interventions for prevention of CSA-AKI including an evaluation of their level of evidence. While recent meta-analyses focused on specific interventions such as RIPc, pulsatile flow during CPB or GDP, no previous meta-analysis has concentrated on CSA-AKI prevention for the following interventions: red blood cell transfusion strategies, tight glycemic control, minimally invasive extracorporeal circulation, epidural analgesia, KDIGO care, targeting high-arterial pressure or hyperoxia during CPB.
In our meta-analysis, GDP was significantly associated with reduced incidence of CSA-AKI. GDP is a recent intervention [
27], derived from the goal-directed therapy used in the care of critically ill patients [
28]. It is based on directly monitoring the oxygen delivery (DO2) during CPB, thus allowing the perfusionist to target above a critical threshold of DO2, usually 280 ml/min/m2. In practice, the DO2 target can be achieved by an increase in pump flow or RBC transfusion. In one of the trials included in our meta-analysis [
29], DO2 was achieved only by pump-flow adjustment. Moreover, in the other trial [
30], RBC transfusion rate did not differ between the 2 groups, which suggests that optimization of DO2 is mainly achieved by adjusting pump flow. A recent meta-analysis on GDP including both RCTs and an observational study, also found a beneficial effect on prevention of CSA-AKI, especially for AKI stage 1 but not for stages 2 or 3 [
31]. While focusing on RCTs, our results support this with a moderate quality of evidence because only two RCTs have been performed.
RIPc was the most evaluated intervention in our meta-analysis, with more than 30 trials, and was associated with decreased incidence in CSA-AKI. RIPc is a simple, non-invasive and inexpensive strategy. During the last 2 decades, several trials and meta-analyses assessing RIPc in cardiac surgery were published, with discordant results. In 2015, 2 large multicenter RCTs (ERRICCA [
32] and RIPHeart [
33]) failed to show a clinical benefit of RIPc in reducing the incidence of CSA-AKI. Several hypotheses could explain the discrepancies. First, they could be related to a small study effect, the tendency for small trials to show larger treatment effects than larger trials within a meta-analysis [
34] and reflecting dissemination bias. In our meta-analysis, we found a clear asymmetry in the funnel plot, thus suggesting a small study effect. We used random-effects models, which give more weight to small studies and may overestimate the treatment effect. In a sensitivity post-hoc analysis using a fixed-effects model, we found consistent results, with significant results for RIPc, but we still cannot exclude a possible impact of the small study effect on our result. Another possible cause of discrepancies may be the population included in the 2 large RCTs, with high-risk patients in the ERICCA trial or conversely low-risk patients in the RIPHeart trial. In our meta-analysis, we included all patients undergoing cardiac surgery, with no restriction on risk. Third, general anesthesia mainly involved propofol in both trials; however, some data suggest that propofol may inhibit the cardioprotective effect of RIPc [
35] and reduce the impact of RIPc on CSA-AKI. Finally, the different modalities of RIPc could explain the discrepancies. In both ERICCA and RIPHeart trials, RIPc was applied to the arm, whereas a clinical study showed a larger benefit on endothelial ischemia–reperfusion injury when RIPc was applied to the leg than the arm [
36].Two previous meta-analyses assessing RIPc [
37,
38] showed results consistent with our finding. However, neither of these studies evaluated the quality of evidence. RIPc is not recommended in current practice guideline due to insufficient strength of evidence [
39]. In our meta-analysis we found a benefice of RIPc on reduction of CSA-AKI with a moderate level of evidence related to the small study effect, highlighting the need of further RCTs to confirm this benefit.
Laminar flow during CPB affects microcirculatory perfusion via endothelial damage [
40,
41], which could alter renal perfusion. In this context, pulsatile flow seems more natural than laminar flow in CPB, and physiological studies found that it can improve microcirculatory parameters [
42] and oxygen extraction and decrease systemic vascular resistance [
43]. Pulsatile flow during CPB can be performed by two 2 means: directly by the CPB machine via the blood pump or with an IABP. Pulsatile blood flow delivered by the CPB, essentially used for low-risk patients, had no significant benefit for CSA-AKI, but the use of IABP in high-risk patients significantly reduced the incidence of CSA-AKI. IABP is a cardiac assistance with potential harmful effects [
44] and therefore cannot be proposed for all patients. During the last decade, two meta-analyses assessing pulsatile flow during CPB [
45,
46], showed a potential beneficial effect for CSA-AKI prevention. However, they did not focus on RCTs only, were not registered on PROSPERO and did not assess risk of bias. In our meta-analysis, we found that pulsatile blood flow could reduce CSA-AKI but with a very low quality of evidence related to high heterogeneity and risk of bias in trials. This very low quality of evidence highlights the need of conducting further RCTs for this intervention.
The other interventions did not have a significant effect on reducing the incidence of CSA-AKI. Some interventions exhibited very low or low level of evidence, such as the use of epidural analgesia or KDIGO care bundle; therefore, more trials are needed to assess their usefulness to reduce the incidence of CSA-AKI. In contrast, our meta-analysis is the first to show a high quality of evidence for the lack of effect of restrictive transfusion strategy and tight glycemic control for preventing CSA-AKI. Consistent with our results, 2 recent meta-analyses showed a lack of benefit of a restrictive transfusion strategy in cardiac surgery [
47,
48]. Given these findings, further trials assessing these interventions are unnecessary.
No intervention was associated with a reduction in need for RRT after cardiac surgery, which may be due to the low proportion of patients requiring RRT and the lack of reporting of this outcome in some trials.
Our results highlight several issues with important implications regarding studies of CSA-AKI. First, a large number of trials did not report CSA-AKI in their outcomes and were therefore excluded. Hence, our results, based on secondary outcomes, may not be exhaustive because some trials may have assessed some of our secondary outcomes without including CSA-AKI among their outcomes. The number of RCTs of cardiac surgery that did not consider AKI in their outcomes was still high, with 40% of trials excluded for this reason. Although AKI remains one of the most common complications [
49] with a high impact on short- and long-term survival after cardiac surgery [
50], it is not part of the core outcome set for adult cardiac surgery trials [
51]. Because AKI is systematically assessed after cardiac surgery, it seems important and simple to consider that outcome for future trials. Second, despite having restricted our selection to a recent period with a consensus, definition of AKI remained heterogeneous across trials including the most recent ones. The first consensus definition of AKI was proposed in 2004 (RIFLE), with an update in 2007 (AKIN) and 2012 (KDIGO). However, we included trials published from 2004 that were planned before this date. The proportion of consensus definitions of AKI increased in recent trials, which is encouraging but remains insufficient. Only 65% of the trials published in the last 5 years reported a consensus definition for AKI. Therefore, we highlight the importance of using a consensus definition of AKI in trials [
52].
From a comprehensive perspective, considering the growing utilization of multimodal.
protective strategies in cardiac surgery [
53,
54], our meta-analysis may help implementing an evidence-based care bundle that includes interventions showing a reduction in the risk of AKI, supported by at least a moderate level of evidence, such as GDP and RIPc.
In contrast, interventions with a high quality of evidence for the lack of benefit on reducing the incidence of CSA-AKI such as tight glycemic control and transfusion strategy should not be incorporated into a future bundle of care for preventing CSA-AKI.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.