The effectiveness and safety of the Impella ventricular assist device for high‐risk percutaneous coronary interventions: A systematic review

Small randomized controlled trials (RCTs) and observational studies have examined the effectiveness and safety of the Impella device, a percutaneous left ventricular assist device, in the setting of high‐risk percutaneous coronary intervention (PCI). However, data are sparse and results are conflicting. Our objective was to evaluate the effectiveness and safety of the Impella device in high‐risk patients undergoing PCI via a systematic review of the literature.

aortic pressure that cannot be improved to more than 60 mmHg by vasopressors [2]. Consequently, the TandemHeart was developed in the early 2000s. However, this device is associated with a bigger vascular profile, arterial and venous large cannula, and subsequent vascular complications [1,3]. Recently, the Impella device entered the market, with five models now available (Impella 2.5, Impella CP (or 3.5), Impella RD, Impella LD, and Impella 5.0), each delivering percutaneously an increasing cardiac output [4]. The most frequently used devices, Impella 2.5 and Impella 5.0, are capable of increasing cardiac output by up to 2.5 and 5.0 L/min, respectively [5][6][7][8]. Impella works by increasing the maximal blood flow by unloading blood from the left ventricle into the ascending aorta, resulting in improved coronary perfusion pressure and end-organ perfusion [5][6][7]. In addition to increasing cardiac output, it also decreases myocardial oxygen consumption and pulmonary capillary wedge pressure [6], potentially reducing the size of a myocardial infarction (MI) and accelerating its recovery. Circulatory support with Impella in high-risk PCI patients increases end-diastolic compliance and reduces end-diastolic wall stress by allowing a more effective unloading of the left ventricle (LV) compared to IABP. In addition to improving the cardiac output and coronary perfusion, this results in a decrease in the myocardial workload and metabolic need and reduces infarct size [9][10][11].

| M E TH ODS
Our systematic review followed a prespecified protocol and is reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [28].

| Search strategy
We systematically searched Medline (via Ovid), EMBASE (via Ovid), and the Cochrane Library databases, from inception to February 15th, 2016, for RCTs and observational studies that evaluated the Impella device. In all databases, we searched using the term "Impella*" with no restrictions. This search was supplemented by a hand-search of relevant studies, editorials, commentaries, and reviews on this topic.
Two independent reviewers (JA and NL) assessed all potentially relevant studies for inclusion. Any publication deemed potentially relevant by either reviewer during title/abstract screening was carried forward to full-text review. Disagreements during full-text review were resolved by consensus or, if necessary, by a third reviewer (KBF).

| Study selection
Studies were included if they (1) evaluated the use of the Impella 2.5 or 5.0 devices; (2) were conducted among high-risk patients undergoing PCI; (3) reported any of the following clinical outcomes: all-cause mortality, major adverse cardiac events (MACE), stroke, MI, repeat revascularization, or bleeding complications (including access site hematoma) at 28-30 days or at longer follow-ups; or any of the following angiographic or hemodynamic characteristics: duration of device support, number of lesions treated, cardiac output provided by the Impella device, angiographic success, and mean arterial pressure (MAP) (before support, on support, and after support); and (4) were published in English or French.
Controlled studies were restricted to those comparing the Impella device to IABP.
Studies were excluded if the number of participants receiving the Impella device was <10. We also excluded cross-sectional studies, reviews and meta-analyses, editorial, commentaries, and letters to the editor. Studies presented as abstracts or conference proceedings were not considered for inclusion as their results are often not final and they contain insufficient information to adequately assess study quality. Finally, basic science studies and those using animal models were also excluded.

| Data extraction
Two independent reviewers (JA and NL) extracted data using a standardized and pilot-tested data collection form, with discrepancies resolved by consensus or by a third reviewer (KBF). Data extracted included study characteristics (study design, sample size, patient population, country of study, presence of a control group, follow-up duration, number of crossovers, number of patients lost-to-follow-up), demographic characteristics (age and sex), clinical characteristics (prior coronary artery bypass grafting (CABG), prior PCI, prior MI, hypertension, diabetes, smoking, left ventricular ejection fraction (LVEF)), procedural and hemodynamic outcomes (duration of support, cardiac output, lesions treated, angiographic success, MAP (before, on, and after support)), and clinical outcomes (all-cause mortality, stroke, MI, MACE, and bleeding complications). For RCTs, data were extracted following an intention-to-treat principle, where patients were considered part of the treatment group to which they were randomized, regardless of treatment received. For observational studies, data were extracted based on treatment received.

| Quality assessment
Quality assessment was performed in duplicate using the Cochrane Risk of Bias Tool for RCTs [29] and using "A Cochrane Risk of Bias Assessment Tool: A Tool for Non-Randomized Studies of Interventions (ACROBAT-NRSI)" for observational studies. Quality assessment for RCTs was performed by characterizing the sequence generation; allocation concealment; blinding of participants, personnel, and outcome assessors; incomplete outcome data; selective outcome reporting; or other sources of bias in the specific trial. Each domain was assigned a "low," "high," or "uncertain" risk of bias. For observational studies, quality assessment was performed by assessing the bias due to confounding, bias in selection of participants into study, bias in measurement of interventions, bias due to departure from intended interventions, bias due to missing data, bias in measurement of outcomes, and bias in AIT ICHOU ET AL. selection of the reported results. Each was assigned a "low," "moderate," "serious," or "critical" risk of bias. All studies were included regardless of their quality.

| Search results
Our search identified 906 potentially relevant publications ( Figure 1).
Following the removal of duplicates and the screening of titles and abstracts, 25 publications were retrieved for full-text review. A total of 20 studies met our inclusion criteria and were included in our systematic review.
In both controlled and uncontrolled studies, the mean age of patients receiving the Impella device ranged from 57.9 to 79.8 years, and the percentage of male patients ranged from 59.1% to 100% (Tables 2 and 3). Mean LVEF was low, ranging from 23% to 37%, while the percentage of patients with previous MI was variable, ranging from 24% to 76%. Overall, patients had multiple co-morbidities and were at high procedural risk. Among comparative studies, patient characteristics were similar between the 2 groups (Table 2).

| Quality assessment
Three RCTs [5,6,27] were deemed to have a low risk of bias, based on the criteria in the Cochrane Risk of Bias tool. However, given the small number of patients randomized in these RCTs, these studies likely had insufficient statistical power to examine clinical outcomes. The IMPRESS in STEMI trial [26] was deemed to have high risk of bias given its early termination and the widening of inclusion criteria over time [26]. The 2 controlled observational studies [23,24] were found to both have serious risk of bias due to many methodological limitations, including confounding by indication. All other observational studies were inherently at serious or critical risk of bias due to their uncontrolled nature.

| Procedural and hemodynamic characteristics
The Impella device had favorable procedural characteristics overall (Table 4). The angiographic success rate was high, ranging from 90% to 100% across studies. The mean duration of support on the Impella device ranged from 38 min to 49 hr. The number of lesions treated with the Impella device was fairly consistent, ranging from 1.8 to 3.0 across studies [12,14,15,17,18,20,21].
The hemodynamic characteristics were also favorable ( Table 4).
The mean cardiac output provided by the Impella device was consistent across studies (2.1 L/min). Seven studies reported increases in MAP (before, on, and after support) with the Impella device; however, only 3 of these studies report both pre-and postprocedural MAP. Of these, the study by Dixon et al. [21] reported lower postprocedural  (Table 4).

| All-cause mortality
In controlled studies, the incidence of all-cause mortality among patients using the Impella device was low (Tables 5 and 6). All RCTs This study compared high-risk patients undergoing PCI using of Impella 2.5 with high-risk patients undergoing PCI using Impella 5.0.
| 1253 found similar rates of all-cause mortality between the Impella device and IABP (  [5], and the IMPRESS trial found 30-day mortality rates of 46% in the Impella group and 50% in the IABP group [27]. The much higher mortality rates in these trials relative to other studies are explained by differences in study population; these RCTs were conducted in patients with cardiogenic shock or acute MI undergoing emergent PCI.
In the controlled observational studies, the 30-day rates of allcause mortality were higher in the Impella arm than in the IABP arm in 2 of the 3 studies that reported this endpoint (

| MACE and individual cardiac outcomes
In controlled studies, the incidence of MACE was low (  Baseline demographic and clinical characteristics of high-risk patients undergoing PCI in controlled studies of the Impella device  Baseline demographic and clinical characteristics of high-risk patients using the Impella device while undergoing PCI in uncontrolled studies of the Impella device A total of 11 patients were upgraded from Impella 2.5 to Impella 5.0 within 30 days post-procedure.
c One patient received an IABP due to a malfunction of the Impella device during the procedure.
| 1255 TABLE 4 Procedural and hemodynamic characteristics for high-risk patients undergoing PCI and using the Impella device  Mean cardiac output reported in this study is the total cardiac output of the patient while on Impella, including the underlying cardiac output of the patient and the flow rate provided by Impella, whereas in the other studies, only the flow rate provided by Impella is reported.

TABLE 5
Thirty-day and long-term clinical outcomes in controlled studies of high-risk patients using either the Impella device or IABP while undergoing PCI AIT ICHOU ET AL.

| 1257
Finally, Ouweneel et al. [26] reported clinical outcomes for the Impella device and IABP at 90 days and 1 year; however, no significant differences were observed (Table 5).
These outcomes were also examined in one controlled observational study (Table 5). Schwartz et al. [24] reported lower 30-day rates of MACE and MI for Impella patients than in IABP patients. However, with a sample size of only 18 patients (13 patients and 5 IABP), no definitive conclusions can be drawn from these data. Rates of stroke reported by this study were similar across groups.

| D I SCUSSION
This study was designed to synthesize the available evidence regarding the effectiveness and safety of the Impella device in high-risk patients undergoing PCI. We found limited evidence available, with only 4 small RCTs and 16 observational studies examining its use. Included studies reported promising hemodynamic and procedural characteristics with the Impella device. In addition, the cardiac output provided by the Impella device was approximately 2.1 L/min, which is higher than that provided by IABP [30]. No differences in all-cause mortality between the Impella device and the IABP were observed; however, a notable Among uncontrolled studies, the rates of all-cause mortality and adverse events were heterogeneous due to differences in study populations and their underlying cardiovascular risk. Although bleeding parameters were not reported in controlled studies, uncontrolled studies suggest that Impella may result in some bleeding complications.
Importantly, most studies were small and inadequately powered to assess clinical events, and all observational studies had a high or critical risk of bias.
The Impella device offers many theoretical advantages over traditional cardiac assist devices due to its potential for increased cardiac output, continued function irrespective of timing or trigger, and stability, even in the presence of cardiac conditions such as arrhythmias. The effective unloading of the LV provided by the use of Impella clinically benefits patients with severely impaired LV function because it can effectively support the failing circulation and reduce the infarct size in case of MI [10]. A study by Cohen et al. [31] compared the procedural, hemodynamic, and clinical characteristics of patients undergoing highrisk PCI supported by Impella 2.5 in a multicentre registry (USpella)   Our study has some potential limitations. First, although we examined the effect of Impella on several outcomes, we did not assess its effect on hemolysis. Second, the individual studies included in our systematic review had small sample sizes and were thus underpowered to detect important differences in clinical events. Third, due to heterogeneity in study designs (e.g., presence of a comparator), study population, and reporting of results, we were unable to pool data across studies via meta-analysis. Fourth, many of the included studies were observational and thus subject to biases such as confounding by indication and by other variables. Such biases were considered in our quality assessment. Finally, as is true with all systematic reviews, there is the potential for publication bias.

| C ONC LUSI ON S
Our study was designed to synthesize the current data regarding the effectiveness and safety of the Impella device among high-risk patients undergoing PCI. We identified that there is limited evidence available concerning the effect of Impella on clinical events, particularly compared to IABP. Although procedural and hemodynamic results appear promising, there remains a need for large, multicenter RCTs to conclusively assess the effectiveness and safety of Impella.

C ONF LI C TS OF I N TER EST
The authors have no conflicts of interest to report.