Incidence and Impact of Acute Kidney Injury in Patients Receiving Extracorporeal Membrane Oxygenation: A Meta-Analysis

Background: Although acute kidney injury (AKI) is a frequent complication in patients receiving extracorporeal membrane oxygenation (ECMO), the incidence and impact of AKI on mortality among patients on ECMO remain unclear. We conducted this systematic review to summarize the incidence and impact of AKI on mortality risk among adult patients on ECMO. Methods: A literature search was performed using EMBASE, Ovid MEDLINE, and Cochrane Databases from inception until March 2019 to identify studies assessing the incidence of AKI (using a standard AKI definition), severe AKI requiring renal replacement therapy (RRT), and the impact of AKI among adult patients on ECMO. Effect estimates from the individual studies were obtained and combined utilizing random-effects, generic inverse variance method of DerSimonian-Laird. The protocol for this systematic review is registered with PROSPERO (no. CRD42018103527). Results: 41 cohort studies with a total of 10,282 adult patients receiving ECMO were enrolled. Overall, the pooled estimated incidence of AKI and severe AKI requiring RRT were 62.8% (95%CI: 52.1%–72.4%) and 44.9% (95%CI: 40.8%–49.0%), respectively. Meta-regression showed that the year of study did not significantly affect the incidence of AKI (p = 0.67) or AKI requiring RRT (p = 0.83). The pooled odds ratio (OR) of hospital mortality among patients receiving ECMO with AKI on RRT was 3.73 (95% CI, 2.87–4.85). When the analysis was limited to studies with confounder-adjusted analysis, increased hospital mortality remained significant among patients receiving ECMO with AKI requiring RRT with pooled OR of 3.32 (95% CI, 2.21–4.99). There was no publication bias as evaluated by the funnel plot and Egger’s regression asymmetry test with p = 0.62 and p = 0.17 for the incidence of AKI and severe AKI requiring RRT, respectively. Conclusion: Among patients receiving ECMO, the incidence rates of AKI and severe AKI requiring RRT are high, which has not changed over time. Patients who develop AKI requiring RRT while on ECMO carry 3.7-fold higher hospital mortality.

Despite these benefits, there have been a number of reports to highlight the concomitant occurrence of organ failures and complications including acute kidney injury (AKI), infections, thrombosis, bleeding and coagulopathy, and neurological events [17,18].The underlying mechanisms for AKI among patients requiring ECMO appear to be complex and include hemodynamic instabilities, inflammatory responses, coagulation-platelet abnormalities, and immune-mediated injury that arise from the primary underlying disease, premorbid conditions and the ECMO circuit [18][19][20][21][22][23][24][25][26][27][28].Due to previously non-uniform definitions of AKI, the reported incidences of AKI among patients requiring ECMO therapy ranged widely from 8% up to 85% [4,7,15,.In addition, the incidence and mortality associated with AKI in patients requiring ECMO and their trends remain unclear.
This systematic review was conducted with the aim to summarize the incidence (using standard AKI definitions) and the impact of AKI on mortality risk among adult patients on ECMO.

Information Sources and Search Strategy
The protocol for this systematic review and meta-analysis is registered with International Prospective Register of Systematic Reviews (PROSPERO no.CRD42018103527).A systematic literature review of EMBASE, Ovid MEDLINE, and the Cochrane Database of Systematic Reviews from database inception through March 2019 was conducted to summarize the incidence and impact of AKI on mortality risk among adult patients on ECMO.Two authors (C.T. and W.C.) independently performed a systematic literature search utilizing a search approach that consolidated the search terms "extracorporeal membrane oxygenation" OR "ECMO" AND "acute kidney injury" OR "acute renal failure."Further details regarding the search strategy utilized for each database are provided in Online Supplementary Data 1.No language restriction was implemented.A manual search for conceivably related articles utilizing references of the included studies was additionally performed.This systematic review was performed following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) statement [71].

Study Selection
Studies were included in this systematic review if they were clinical trials or observational studies that reported the incidence of AKI (using standard AKI definitions including RIFLE (Risk, Injury, Failure, Loss of kidney function, and End-stage kidney disease) [72], AKIN (Acute Kidney Injury Network) [73], and KDIGO (Kidney Disease: Improving Global Outcomes) classifications) [74], severe AKI requiring renal replacement therapy (RRT), and mortality risk of AKI among adult patients (age ≥ 18 years old) on ECMO.Eligible studies needed to provide the data to evaluate the incidence or mortality rate of AKI with 95% confidence intervals (CI).Retrieved articles were independently examined for eligibility by the two authors (C.T. and W.C.).Inconsistencies were discussed and resolved by shared agreement.The size of the study did not limit inclusion.

Data Collection Process
A structured data collecting form was adopted to gather the following data from individual study including title, name of authors, publication year, year of the study, country where the study was conveyed, type of ECMO, AKI definition, incidence of AKI, incidence of severe AKI requiring RRT, and mortality risk of AKI among patients on ECMO.

Statistical Analysis
We used the Comprehensive Meta-Analysis software version 3.3.070(Biostat Inc, Englewood, NJ, USA) to conduct the meta-analysis.Adjusted point estimates of included studies were consolidated by the generic inverse variance method of DerSimonian-Laird, which assigned the weight of individual study based on its variance [75].Due to the probability of between-study variance, we applied a random-effects model to pool outcomes of interest, including the incidence of AKI and mortality risk.Statistical heterogeneity of studies was assessed by the Cochran's Q test (p < 0.05 for a statistical significance) and the I 2 statistic (≤25%: insignificant heterogeneity, 26%-50%: low heterogeneity, 51%-75%: moderate heterogeneity and ≥75%: high heterogeneity) [76].The presence of publication bias was evaluated by both the funnel plot and the Egger test [77].

Results
A total of 1,632 potentially eligible articles were identified with our search approach.After excluding 644 articles that were either in-vitro studies, focused on pediatric patient population, animal studies, case reports, correspondences, or review articles, and 831 articles due to being duplicates, 157 articles remained for full-length article review.Seventy-three articles were subsequently excluded as they did not provide data on the incidence of AKI or mortality of AKI, while 33 articles were excluded because they were not clinical trials or observational studies.Ten studies [19][20][21][22][23][24][25][26][27][28] were additionally excluded because they did not use a standard AKI definition or did not report the incidence of severe AKI requiring RRT.Therefore, 41 cohort studies [7,15, with a total of 10,282 adult patients receiving ECMO were enrolled.The systematic review of the literature flowchart is demonstrated in Figure 1.The characteristics of the included studies are shown in Table 1.

AKI associated Mortality in Patients Requiring ECMO
Mortality rate and mortality risk associated with AKI in patients requiring ECMO are demonstrated in Tables 1 and 2, respectively.The pooled estimated hospital and/or 90-day mortality rates of patients with AKI and severe AKI requiring RRT while on ECMO were 62.0% (95%CI: 54.7%-68.8%,I 2 = 73%, Figure 4A) and 68.4% (95%CI: 62.6%-73.6%,I 2 = 87%, Figure 4B), respectively.The pooled OR of hospital mortality among patients receiving ECMO with AKI on RRT was 3.73 (95% CI, 2.87-4.85,I 2 = 62%, Figure 5A).When the analysis was limited to studies with confounder-adjusted analysis, the increased hospital mortality remained significant among patients receiving ECMO with AKI requiring RRT with pooled OR of 3.32 (95% CI, 2.21-4.99,I 2 = 82%, Figure 5B).Meta-regression showed that year of the study did not significantly affect hospital mortality among patients receiving ECMO with AKI requiring RRT (p = 0.86), as shown in Figure 6.
Meta-regression showed that year of the study did not significantly affect hospital mortality among patients receiving ECMO with AKI requiring RRT (p = 0.86), as shown in Figure 6.

Evaluation for Publication Bias
Funnel plots (Figure 7) and Egger's regression asymmetry tests were utilized to assess for publication bias in our meta-analyses evaluating the incidence of AKI and severe AKI requiring RRT while on ECMO.There was no publication bias as determined by the funnel plot and Egger's regression asymmetry test with p = 0.62 and p = 0.17 for the incidence of AKI and severe AKI requiring RRT, respectively.

Evaluation for Publication Bias
Funnel plots (Figure 7) and Egger's regression asymmetry tests were utilized to assess for publication bias in our meta-analyses evaluating the incidence of AKI and severe AKI requiring RRT while on ECMO.There was no publication bias as determined by the funnel plot and Egger's regression asymmetry test with p = 0.62 and p = 0.17 for the incidence of AKI and severe AKI requiring RRT, respectively.

Discussion
The findings of our meta-analysis demonstrate that patients who required ECMO had incidence rates of AKI (using standard AKI definitions) and severe AKI requiring RRT of 62.8% and 44.9%, respectively.Moreover, patients with AKI and severe AKI requiring RRT had high associated mortality rates of 62.0% and 68.4%, respectively.
Although the mechanisms underlying ECMO associated-AKI remains unclear, it is likely complex and multifactorial, including contributing factors such as primary disease progression, altered hemodynamics, low cardiac output syndrome, exposure to nephrotoxic agents (for management of underlying diseases), new-onset sepsis, high intrathoracic pressures, fluid overload, ischemia-reperfusion injury, release of proinflammatory mediators and oxidative stress, hemolysis and iron-mediated (hemoglobin-induced) renal injury, and hypercoagulable state resulting in renal microembolisms [4,8,68,78,79].Studies have demonstrated the activation of proinflammatory mediators such as tumor necrosis factor-alpha (TNF-α), interleukins (e.g., IL-1β, IL-6, IL-8) and other cytokine signaling cascades due to the continuous exposure of blood to non-biological and non-endothelialized ECMO interface [68,80,81].Activation of the inflammatory cascades can result in hyperdynamic vasodilated hypotensive states, leading to AKI [68,78].
Following the initiation of ECMO treatment, there are improvements in oxygenation and oxygen consumption as well as hemodynamics [3,[5][6][7][8][9].However, ischemia-reperfusion injury can also occur after the restoration of circulation to previously hypoxic cells and hypoperfused organs,

Discussion
The findings of our meta-analysis demonstrate that patients who required ECMO had incidence rates of AKI (using standard AKI definitions) and severe AKI requiring RRT of 62.8% and 44.9%, respectively.Moreover, patients with AKI and severe AKI requiring RRT had high associated mortality rates of 62.0% and 68.4%, respectively.
Although the mechanisms underlying ECMO associated-AKI remains unclear, it is likely complex and multifactorial, including contributing factors such as primary disease progression, altered hemodynamics, low cardiac output syndrome, exposure to nephrotoxic agents (for management of underlying diseases), new-onset sepsis, high intrathoracic pressures, fluid overload, ischemia-reperfusion injury, release of proinflammatory mediators and oxidative stress, hemolysis and iron-mediated (hemoglobin-induced) renal injury, and hypercoagulable state resulting in renal microembolisms [4,8,68,78,79].Studies have demonstrated the activation of proinflammatory mediators such as tumor necrosis factor-alpha (TNF-α), interleukins (e.g., IL-1β, IL-6, IL-8) and other cytokine signaling cascades due to the continuous exposure of blood to non-biological and non-endothelialized ECMO interface [68,80,81].Activation of the inflammatory cascades can result in hyperdynamic vasodilated hypotensive states, leading to AKI [68,78].
Following the initiation of ECMO treatment, there are improvements in oxygenation and oxygen consumption as well as hemodynamics [3,[5][6][7][8][9].However, ischemia-reperfusion injury can also occur after the restoration of circulation to previously hypoxic cells and hypoperfused organs, leading to the production of reactive oxygen species (ROS) and oxidative stress-mediated injury [68,78].In addition, ECMO-associated complications or adverse effects such as hemolysis, hemorrhage or thrombosis also can play important roles in the development of AKI [29,68,[82][83][84].Despite the advance of a new miniaturized ECMO system, hemolysis due to shear stress from the ECMO circuit has been reported among ECMO patients with incidences between 5% and 18% [17,[85][86][87].This can contribute to heme pigment-induced AKI [83,84].Although improvements in the ECMO technology have led to less thrombus development in its circuit with an improved capacity of the circuit to remove large emboli [68,82], smaller thrombi can still develop and result in renal microembolism [68,82], particularly with VA-ECMO [82].
The type of ECMO may also differently affect AKI risk.Our study demonstrated a higher incidence of AKI among patients requiring VA-ECMO (60.8%) than those requiring VV-ECMO (45.7%).While VV-ECMO is typically utilized for patients with isolated respiratory failure, VA-ECMO is used for combined severe cardiac and respiratory failure [4].In VA-ECMO, there is a mixture of pulsatile arterial flow from the native heart and non-pulsatile arterial flow from the ECMO pump.Conversely, VV-ECMO maintains pulsatile cardiac output, and alterations in renal perfusion may conceivably be smaller [4].Recent studies have shown that pulsatile flow may provide beneficial effects over non-pulsatile flow, especially protective effects on microcirculation and renal perfusion [88][89][90].The differences in patient population and pulsatility between the two types of ECMO are likely explanations underlying the higher AKI incidence among patients requiring VA-ECMO.
As there is no treatment available for AKI, management of AKI is limited to appropriate secondary preventive measures and supportive strategies [91][92][93][94][95][96].RRT in the form of continuous renal replacement therapy (CRRT) is often required among patients requiring ECMO with severe AKI [42,55,97].Our study demonstrated no significant correlation between the year of study and the incidence of AKI and/or severe AKI requiring RRT despite considerable changes in technology and practice of ECMO among adult patients.Furthermore, we showed a 3.7-fold increased risk of hospital mortality among ECMO patients with severe AKI requiring RRT.Thus, prevention and early identification of AKI among patients at-risk of ECMO-associated AKI could potentially play a crucial role in improved survival.Studies have shown several important AKI risk factors among patients requiring ECMO including older age, elevated lactate levels before ECMO initiation, high dose of inotropic drugs, severely reduced left ventricular ejection fraction, cirrhosis, postcardiotomy shock as an indication for ECMO, and finally ECMO pump speed and its duration [56,64,67].Lee et al. recently observed a lower AKI association with a higher ECMO pump speed [56].Although the underlying pathophysiology remains unclear, excessive ECMO pump speed has been shown to induce hemolysis and complement activation in vitro and animal model [98,99].In pediatric patients receiving ECMO, Lou et al. also demonstrated higher pump speeds are associated with hemolysis and a number of other adverse clinical outcomes [100].To prevent hemolysis-mediated kidney injury, it is suggested to limit pump revolutions/min (RPM) to safe levels (i.e., 3000 to 3500 RPM) in order to avoid excessive negative pressures generated within the pump [101].Future prospective studies are required to assess the effects of ECMO pump speed on AKI risk in ECMO patients.In addition, future studies creating risk prediction models for ECMO-associated AKI are needed to assist with the prevention of AKI in a timely manner, which could potentially lead to an improvement in patient survival.
Our study has several limitations.Firstly, there are statistical heterogeneities in our meta-analysis.Potential sources for heterogeneities were the variations in patient characteristics among the included studies.However, we performed subgroup analysis to assess the AKI incidence based on types of ECMO and a separate meta-analysis that only included studies with confounder-adjusted analysis for mortality risk.Another limitation was that AKI diagnosis was mainly based on serum creatinine [102][103][104] while the data on urine output and novel biomarkers for AKI [105][106][107][108] were limited.Lastly, this systematic

Figure 1 .
Figure 1.The flowchart for the systematic review.

Figure 1 .
Figure 1.The flowchart for the systematic review.

Figure 2 .
Figure 2. Forest plots of the included studies assessing (A) incidence rates of AKI while on ECMO and (B) incidence rate of severe AKI requiring RRT while on ECMO.A diamond data marker depicts the overall rate from each included study (square data marker) and 95%CI.

Figure 2 .
Figure 2. Forest plots of the included studies assessing (A) incidence rates of AKI while on ECMO and (B) incidence rate of severe AKI requiring RRT while on ECMO.A diamond data marker depicts the overall rate from each included study (square data marker) and 95%CI.

Figure 3 .
Figure 3. Meta-regression analyses showed that year of the study did not significantly affect (A) the incidence of AKI (p = 0.67) or (B) AKI requiring RRT (p = 0.83).The solid black line depicts the weighted regression line based on variance-weighted least squares.The inner and outer lines represent the 95%CI and prediction interval encompassing the regression line.The circles indicate log event rates in individual study.

Figure 3 .
Figure 3. Meta-regression analyses showed that year of the study did not significantly affect (A) the incidence of AKI (p = 0.67) or (B) AKI requiring RRT (p = 0.83).The solid black line depicts the weighted regression line based on variance-weighted least squares.The inner and outer lines represent the 95%CI and prediction interval encompassing the regression line.The circles indicate log event rates in individual study.

Figure 4 .
Figure 4. Forest plots of the included studies assessing (A) mortality rate of patients with AKI while on ECMO and (B) mortality rate of patients with severe AKI requiring RRT while on ECMO.A diamond data label serves as the overall rate from each study (square data marker) and 95%CI.

Figure 4 .
Figure 4. Forest plots of the included studies assessing (A) mortality rate of patients with AKI while on ECMO and (B) mortality rate of patients with severe AKI requiring RRT while on ECMO.A diamond data label serves as the overall rate from each study (square data marker) and 95%CI.

Figure 5 .
Figure 5. Forest plots of the included studies assessing (A) hospital mortality among patients receiving ECMO with AKI on RRT and (B) hospital mortality among patients receiving ECMO with AKI on RRT limited to studies with confounder-adjusted analysis.A diamond data label serves as the overall rate from each included study (square data marker) and 95%CI.

Figure 5 .
Figure 5. Forest plots of the included studies assessing (A) hospital mortality among patients receiving ECMO with AKI on RRT and (B) hospital mortality among patients receiving ECMO with AKI on RRT limited to studies with confounder-adjusted analysis.A diamond data label serves as the overall rate from each included study (square data marker) and 95%CI.

Figure 6 .
Figure 6.Meta-regression analyses showed that year of the study did not significantly affect hospital mortality among patients receiving ECMO with AKI requiring RRT (p = 0.86).The solid black line depicts the weighted regression line based on variance-weighted least squares.The inner and outer lines represent the 95%CI and prediction interval encompassing the regression line.The circles indicate log event rates in an individual study.

Figure 6 .
Figure 6.Meta-regression analyses showed that year of the study did not significantly affect hospital mortality among patients receiving ECMO with AKI requiring RRT (p = 0.86).The solid black line depicts the weighted regression line based on variance-weighted least squares.The inner and outer lines represent the 95%CI and prediction interval encompassing the regression line.The circles indicate log event rates in an individual study.

Figure 7 .
Figure 7. Funnel plot demonstrated no publication bias in analyses evaluating (A) incidence of AKI in patients requiring ECMO and (B) severe AKI requiring RRT.

Figure 7 .
Figure 7. Funnel plot demonstrated no publication bias in analyses evaluating (A) incidence of AKI in patients requiring ECMO and (B) severe AKI requiring RRT.

Table 2 .
Characteristics of studies included in this meta-analysis of AKI associated mortality risk among patients requiring ECMO.

Table 2 .
Characteristics of studies included in this meta-analysis of AKI associated mortality risk among patients requiring ECMO.