Incidence and impact of acute kidney injury on patients with implantable left ventricular assist devices: a Meta-analysis

Abstract Background We aimed to evaluate the acute kidney injury (AKI) incidence and its associated risk of mortality in patients with implantable left ventricular assist devices (LVAD). Methods A systematic literature search in Ovid MEDLINE, EMBASE, and Cochrane Databases was conducted through January 2020 to identify studies that provided data on the AKI incidence and AKI-associated mortality risk in adult patients with implantable LVADs. Pooled effect estimates were examined using random-effects, generic inverse variance method of DerSimonian-Laird. Results Fifty-six cohort studies with 63,663 LVAD patients were enrolled in this meta-analysis. The pooled incidence of reported AKI was 24.9% (95%CI: 20.1%–30.4%) but rose to 36.9% (95%CI: 31.1%–43.1%) when applying the standard definition of AKI per RIFLE, AKIN, and KDIGO criteria. The pooled incidence of severe AKI requiring renal replacement therapy (RRT) was 12.6% (95%CI: 10.5%–15.0%). AKI incidence did not differ significantly between types of LVAD (p = .35) or indication for LVAD use (p = .62). While meta-regression analysis did not demonstrate a significant association between study year and overall AKI incidence (p = .55), the study year was negatively correlated with the incidence of severe AKI requiring RRT (slope = −0.068, p < .001). The pooled odds ratios (ORs) of mortality at 30 days and one year in AKI patients were 3.66 (95% CI, 2.00–6.70) and 2.22 (95% CI, 1.62–3.04), respectively. The pooled ORs of mortality at 30 days and one year in severe AKI patients requiring RRT were 7.52 (95% CI, 4.58–12.33) and 5.41 (95% CI, 3.63–8.06), respectively. Conclusion We found that more than one-third of LVAD patients develop AKI based on standard definitions, and 13% develop severe AKI requiring RRT. There has been a potential improvement in the incidence of severe AKI requiring RRT for LVAD patients. AKI in LVAD patients was associated with increased 30-day and 1 year mortality.


Introduction
Implantable left ventricular assist devices (LVADs) are increasingly utilized as a bridge to heart transplantation or destination therapy for patients with end-stage heart failure [1][2][3][4][5][6][7]. The use of LVADs is shown to be associated with reduced mortality in patients on heart transplantation waiting lists, and they improve quality of life and functional status in advanced heart failure patients [8]. LVADs alleviate the cardiovascular load on a failing heart and have shown notable advantages in treating patients with advanced heart failure, providing prolonged survival and improvement in the quality of life [9,10]. Clinical outcomes after LVAD implantation have significantly improved over the past decade, with 1 year and 2 year survival of 83% and 73%, respectively [11,12]. In the United States, the number of LVAD implantations rose, from only 459 implants in 2008 to a total of 2,118 implants in 2017 [11].
This systematic review and meta-analysis were conducted to summarize the AKI incidence and mortality risk among adult patients with LVADs.

Methods
The protocol for this meta-analysis is registered with PROSPERO (no. CRD42020134592). The meta-analyses were conducted in adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [71].

Search strategy
Two investigators (CT and PL) independently searched for published clinical trials or observational studies indexed in MEDLINE, EMBASE and the Cochrane databases from inception to January 2020 using a search strategy (S1 in online Supplementary Data 1) that included the terms "left ventricular assist device", "LVAD", "ventricular assist device", "acute kidney failure", "acute kidney injury" and "renal replacement therapy". No language restrictions were applied in this systemic review and meta-analysis. A manual search for additional pertinent studies and review articles using references from the retrieved articles was also completed.

Study eligibility criteria
Two main criteria were used for study inclusion. First, the study had to report the incidence of AKI or severe AKI requiring renal replacement therapy (RRT), and AKI associated mortality risk in adult patients with LVADs aged at least 18 years. Second, the study had to include data assessing AKI incidence or mortality risk with 95% confidence intervals (CIs) (or sufficient raw data for the calculation). Patients were excluded if they only used a temporary, short-term, non-implantable LVAD during a hospitalization. Study eligibility was independently determined by two investigators (CT and PL). Differences were resolved by mutual consensus.
A standardized data collection form was used to obtain the following information from each study: title, name of the first author, year of study, year of publication, country of origin, number of participants, demographic data of participants, the method used to diagnose the outcomes of interest (AKI incidence and associated mortality), the average duration of followup, adjusted and unadjusted risk ratios and their corresponding 95% CI, and list of confounders that were adjusted for in the multivariate analyses. To ensure accuracy, both investigators independently performed this data extraction process. Any data discrepancy was resolved by referring back to the original articles. The Newcastle-Ottawa quality assessment scale was utilized to appraise the quality of observational studies [72].

Statistical analysis
The meta-analysis of combined data was performed using a random-effects, generic inverse variance method of DerSimonian and Laird [73]. We assessed the overall incidence of AKI, which was defined by the consensus definitions provided by the Risk, Injury, Failure, Loss of kidney function, and End-stage kidney disease (RIFLE) [74], Acute Kidney Injury Network (AKIN) [75], and Kidney Disease: Improving Global Outcomes) (KDIGO) [76] classifications. We did not impute missing values for any outcomes in our analyses. A randomeffect model was used to pool AKI incidence and AKIassociated mortality risk due to the possibility of between-study variance. Heterogeneity among included studies was statistically evaluated by the Cochrane's Q test and the I 2 statistic. Heterogeneity was considered insignificant when I 2 of 25%, low when I 2 of 26-50%, moderate when I 2 of 51-75%, and high when I 2 of !75% [77]. Per Cochrane, publication bias was assessed using a funnel plot. Funnel plot asymmetry was further confirmed with Egger's test if there were >10 available studies [78]. All analysis was performed using The Comprehensive Meta-Analysis software version 3.3.070 (Biostat Inc, New Jersey, USA). The data underlying the results presented in the study are available through the Open Science Framework (https://osf.io/8hk35/)

Results
Our search approach identified a total of 1,665 potentially eligible articles. We initially excluded 846 articles because they were case reports, correspondences, review articles, or studies involving in-vitro, animal, or pediatric patients. Six hundred fifty-eight duplicated articles were additionally excluded. After the review of 161 full-length articles, we subsequently excluded 67 articles because the data on AKI incidence and its associated mortality was not available, 28 articles because they were not observational studies or clinical trials, and 10 articles because they investigated AKI in shortterm LVAD use, not implantable LVAD [79][80][81][82][83][84][85][86][87][88]. Therefore, 56 cohort studies  with a total of 63,663 adult patients were included in this meta-analysis. Figure 1 demonstrates by flowchart the systematic review of the literature. Table 1 shows the characteristics of the included studies. The kappa for systematic searches, selection of studies and data extraction were 1.00, 0.91 and 0.98, respectively

Discussion
This meta-analysis supports that AKI is a common complication after LVAD implantation. The pooled incidence of post-LVAD AKI (using standard AKI definitions) and severe AKI requiring RRT was 37% and 13%, respectively. We found no significant difference in AKI incidence after adjusting for LVAD indication (bridging vs. destination therapy). Moreover, our analysis did not show any difference in AKI incidence between pulsatile and continuous flow LVAD devices. It was also noted that the incidence of AKI was higher (37% vs. 25%) when using standard AKI criteria, such as RIFLE, AKIN, and KDIGO. This may indicate that defining AKI using consensus criteria may improve the sensitivity of detecting AKI in LVAD patients. This meta-analysis further identified that AKI incidence remained constant over time, while the need for RRT due to AKI decreased significantly in more recent studies. The mechanisms of AKI among LVAD patients are complex and can be multifactorial [5,89,90]. Mechanical stress on red blood cells traveling through the LVAD leads to constant low-level hemolysis, potentially resulting in pigment nephropathy [5]. These patients also tend to have acquired von Willebrand disease, as the von Williebrand factor multimers suffer fragmentation when passing through the LVAD pump, leading to subsequent increased risk of AKI due to decreased effective blood volume secondary to bleeding from arteriovenous malformations or severe epistasis [5,8]. An additional concern is, the development of right heart failure following LVAD implantation, which is observed in approximately 20-50% of patients [14,[91][92][93][94]. This right heart failure could further potentiate renal venous congestion, compromised net effective renal perfusion pressure and decrease GFR [95,96]. Hemodynamic instability in the immediate post-operative period could exacerbate kidney ischemia and lead to acute tubular necrosis. Accelerated thrombogenicity secondary to the LVAD pump and blood stasis may trigger renal microemboli, as evidenced by the presence of kidney infarctions [97]. Yet an another proposed hypothesis for the development of worsening kidney function in LVAD patients is that the continuous flow of the LVAD might lead to a proliferation of afferent arteriolar smooth muscle cells and periarteriolitis, which causes an eventual decline in eGFR [98]. However, in our study, the incidence of AKI was similar between pulsatile-flow and continuous-flow LVADs, suggesting that the lack of pulsatility from continuous-flow LVADs might not be the cause of associated AKI. On the other hand, currently only a limited amount of pulsatility can be generated by LVADs using periodic speed steps, and it is considerably smaller in both flow increase and rate than what is found with natural pulsatile circulation [99]. Given the ongoing efforts to advance LVAD technology, future studies are needed to evaluate whether or not improvements in pulsatile-flow LVADs can reduce the incidence of post-LVAD implantation AKI.
The findings from our study demonstrated that LVAD patients who developed AKI had greater odds of 30-day and 1-year mortality. The pooled odds ratios were even higher in patients with severe AKI requiring RRT. It is emphasized that even an occurrence of AKI following LVAD implantation has long-lasting negative clinical impacts, especially if dialysis is required [40,100]. Post-implantation AKI is associated with right ventricular failure and arrhythmias, both of which are, in turn, associated with increased mortality [28]. Our study shows that LVAD patients with severe AKI requiring RRT are associated with 7.5-fold and 5.4-fold increased risks of 30-day and 1-year mortality, respectively. While the findings of our study suggested no significant changes in overall AKI incidence over the study years, the incidence of severe AKI requiring RRT appeared to decrease over study year significantly. This finding suggests potential improvements in the prevention, mitigation, and clinical management of severe AKI in LVAD patients. Interventions proposed to mitigate the incidence and severity of post-LVAD implantation AKI include maintenance of high mean arterial pressures (MAP) and coronary perfusion rates [96], inotropic support when needed, frequent monitoring of MAP via audible doppler ultrasound in combination with calibrated blood pressure measurement devices, and maintaining central venous pressures between 8 and 12 mm Hg via diuretics or extracorporeal ultrafiltration. Avoiding nephrotoxic medications postoperatively until hemodynamic stability is achieved has also been recommended. In patients with severe right heart failure, right ventricular assist devices may help decrease rightsided venous congestion and improve renal perfusion [96]. Future studies are required to assess whether these measures can significantly help to reduce AKI incidence or promote AKI recovery among LVAD patients, to improve patient survival rates ultimately.
Our systematic and meta-analysis is subject to certain limitations. First, all studies were observational in design, making them susceptible to potential selection bias. The potential sources of this heterogeneity included differences in variation in baseline characteristics (e.g., age, sex, ethnicity, and underlying chronic kidney disease), LVAD types, indications for LVAD, and outcome ascertainments. Second, the incidence of AKI is predisposed to several confounding factors. Our meta-analysis had a high degree of heterogeneities. However, we performed subgroup analyses after applying standardized AKI definitions and conducted metaregression analyses assessing the effects of the study year, LVAD types (pulsatile vs. continuous flow), and indications for LVAD implantation (bridge to transplant vs. destination therapy) on AKI incidence. The results from these additional analyses provided clinical insights that may highlight and stimulate the need for additional research to intervene on AKI in LVAD patients. Third, the data on the use of peritoneal dialysis as a modality of RRT in LVAD patients is limited, and all of the included studies defined RRT as either non-peritoneal continuous or intermittent renal replacement therapies. Lastly, AKI diagnoses in the included studies were solely based on the change in serum creatinine, which might underestimate the incidence of AKI [101][102][103][104][105]. Data on urine output or other AKI biomarkers data were limited [104,106] Furthermore, future studies using artificial intelligence to predict AKI among LVAD patients are needed [107].
In conclusion, AKI is a common complication among LVAD patients. There have been some potential improvements in the incidence rates of severe AKI requiring RRT in LVAD patients. AKI, while on LVAD, is associated with increased 30-day and 1-year mortality.

Disclosure statement
No potential conflict of interest was reported by the author(s).