Impella Protected Percutaneous Coronary Intervention Outcomes Compared With Intra-aortic Balloon Pump: A Contemporary View

Highlights

  • •

    In the current study the win ratio of Impella versus IABP-supported high-risk PCI favors Impella.

  • •

    Favorability increases when the first Impella patients of each site are excluded.

  • •

    Favorability is observed in patients with ULM, atherectomy or CTO.

  • •

    Win ratio analyzes results align with prior analyzes with conventional statistics.

Traditional cardiovascular trials combine adverse events into composites, ignoring the clinical importance and weight of endpoints. The Win Ratio (WR) is a contemporary statistical technique overcoming these limitations. We aimed to evaluate outcomes of high-risk percutaneous coronary intervention supported with Impella versus intra-aortic balloon pump, by pooling data from the PROTECT-II and PROTECT-III studies, using the WR. All patients from PROTECT-II RCT (P-II) and patients from PROTECT-III (P-III) who met P-II inclusion/exclusion criteria were pooled. The WR was based on independently adjudicated major adverse cardiac and cerebrovascular events at 90 days with following hierarchy: (1) mortality; (2) stroke; (3) spontaneous myocardial infarction; (4) rehospitalization; and (5) peri-procedural myocardial infarction. All major adverse cardiac and cerebrovascular events were analyzed as time-to-event outcomes, except peri-procedural myocardial infarction (binary endpoint). Sub-analyzes included: (1) complex cases: patients with atherectomy or unprotected left main or chronic total occlusion, (2) all patients excluding firsts from P-II (learning cases); and (3) Impella P-II and P-III cohorts separately. Win statistics (WR, net benefit, and win odds) were calculated. The primary analysis (719 Impella and 211 intra-aortic balloon pump-supported PCI) yielded a WR of 1.691 in favor of Impella (1.314 to 2.176, p < 0.001), with net benefit of 0.166 (0.084 to 0.247, p < 0.001) and win odds of 1.398 (1.187 to 1.645, p < 0.001). The WR, net benefit and win odds for complex cases remained statistically significant in favor of Impella. Excluding first patients resulted in increased win statistics compared to primary analysis. In conclusion, pooled WR analyzes from P-II and P-III studies demonstrated improved high risk PCI outcomes up to 90 days with Impella compared to intra-aortic balloon pump.

The evidential landscape on the need for revascularization in patients with reduced left ventricular ejection fraction is very complex. The long-term results of the STICHES trial suggested improved survival in patients treated with coronary artery bypass surgery, whereas the REVIVED trial did not show a benefit of PCI versus medical therapy in a more stable population with a low percentage of angina prevalence. The choice of PCI versus coronary artery bypass surgery is largely based on coronary anatomy, hemodynamics and patient factors, such as surgical eligibility; as a result, PCI is often considered for high-risk comorbid patients with advanced age, complex anatomic lesions, who are ineligible for, or refuse, surgical revascularization. ,

In the high-risk PCI (HRPCI) space, few randomized studies stand out to date. In the Balloon Pump-Assisted Coronary Intervention Study randomized controlled trial no clinical differences were observed between intra-aortic balloon pump (IABP) and controls in the immediate postprocedural period, but at 51-month follow-up, patients in the IABP arm had a 34% reduced mortality. , The subsequent PROTECT II (P-II) support with Impella 2.5 did not result in statistically significant superiority in the 30-day primary end-point compared to IABP ; however, a post hoc evaluation of the PROTECT-II that re-examined the study with more practical, contemporary, and consensual periprocedural myocardial infarction (pMI) endpoint definitions concluded that the use of Impella results in significantly lower major adverse cardiovascular events at 90 days, compared to IABP. PROTECT III (P-III) was a prospective, multicenter, single-arm, observational, postapproval study. Within the P-III population, patients considered “P-II like” were identified, and their outcomes was analyzed and compared to P-II. This study reported further improvements in outcomes in P-III compared to the P-II, highlighting advances in HRPCI standard of care.

The Win Ratio (WR) is hierarchical composite endpoint analysis tool that has the following advantages compared to traditional composite endpoints: it prioritizes the most clinically meaningful events (for example, death), and ensures all events are included in the analysis, regardless of time to event, thus avoiding time to first event bias. The WR has been adopted in several large contemporary cardiovascular trials and represents a new paradigm in cardiovascular clinical research. ,,, It has not yet been applied to P-II or P-III. P-II outcomes were analyzed using a 10-item composite major adverse cardiac and cerebrovascular events (MACCE) rate at 90 days. The P-III study used a 4-item composite MACCE rate at 90 days, defined as death, myocardial infarction (MI), stroke or transient ischemic attack (TIA), and repeat revascularization. , The objective of this study was therefore to re-evaluate the outcomes of the combined P-II and P-III data using WR as described by Ryan et al. for the controlled trial of high-risk coronary intervention with percutaneous left ventricular unloading (CHIP-BCIS3) trial ( Figure 1 ).

Figure 1

Win ratio approach in patient cohorts from PROTECT II and PROTECT III studies.

Methods

An IRB was not required for this analysis as it utilized prespecified outcomes from P-II and P-III clinical trials. Patient consent had been previously obtained for all patients.

Patients

All patients from the P-II RCT and “P-II like” patients from the P-III study were included. Inclusion and exclusion criteria for P-II have been previously described. The “P-II-like” patient cohort from P-III is a subset of all P-III patients and includes patients that met the inclusion and exclusion criteria of P-II. The inclusion criteria were: nonemergent (i.e., elective or nonurgent) PCI on an unprotected left main or last patent coronary vessel with a left ventricular ejection fraction (LVEF) ≤35%, or patients with 3-vessel disease and LVEF ≤30%. Major exclusion criteria included recent MI with persistent elevation of cardiac enzymes, left ventricular thrombus, platelet count of ≤75,000/mm 3 and creatinine ≥4 mg/dl. ,

The randomization methodology for P-II, and HRPCI procedural details for both P-II and P-III have been described elsewhere. Briefly, in P-II: crossover from 1 study arm to the other was not allowed. Physicians were instructed to aim for the most complete revascularization possible, within a single procedure. For P-III, patients were treated with HRPCI in accordance to each participating hospital’s standard of care and operator discretion. ,

Device description

P-II patients received either an IABP, or the Impella 2.5 (Abiomed, Danvers, MA), a 12F axial flow, rotary blood pump mounted on a 9F catheter deployed in a retrograde fashion across the aortic valve. P-III patients received either the Impella 2.5, or the next generation device the Impella CP, which became commercially available in 2012. The Impella CP is an intravascular 14F microaxial flow pump with a 9F catheter shaft.

Endpoints

Outcomes included death, myocardial infarction (MI, peri‑procedural of spontaneous), stroke/TIA, and any repeat revascularization and hospitalization. For both P-II and P-III, an independent clinical event committee adjudicated all outcomes for protocol definitions and relatedness. For P-II, as this was a randomized controlled trial, the clinical event committee was blinded to treatment group assignment. Myocardial infarction was defined as described by Dangas et al. Briefly, MI was associated with the development of new Q-waves or CK-MB elevation >8× ULN within 72 hours post-PCI for peri‑procedural MI (pMI), or as >2× ULN beyond 72 hours of the PCI for spontaneous MI (sMI). Mortality, stroke, sMI and rehospitalization were captured as time to event outcomes, whereas pMI was simply defined as a binary endpoint. , The study duration was 90 days post-HRPCI.

Statistical Analyzes

A WR analysis was performed to compare treatment groups, using an unmatched approach, where each participant in the IABP cohort was compared to each individual participant in the Impella group. Whereas the follow-up period was 90 days, comparisons of time-to event endpoints were limited to the shortest duration of follow-up available for each patient. As shown in Figure 1 , the following endpoints were included in the WR in the following order: (1) mortality, (2) stroke, (3) sMI, (4) re-hospitalization, and (5) pMI. For the primary analysis, all patients from P-II and P-III were included, and win statistics for the hierarchical sequence, as well as for each individual endpoint were calculated.

The following subgroup analyzes were conducted: (1) as prespecified in P-II, patients with atherectomy or unprotected left main (ULM) or chronic occlusion were analyzed separately, to evaluate the impact of particularly challenging PCI procedures, on outcomes. (2) Learning curve was evaluated by excluding the first patients from the P-II cohort. This subgroup analysis was also prespecified in P-II. The rationale for excluding these patients, is that at the time of P-II, many sites were still inexperienced in the use of the device, and their first case may reflect a true learning case. We did not exclude the first IABP case as IABP has been used since ∼1968 and most sites were therefore experienced in its use at the time of P-II. We also did not exclude the first Impella case from P-III because by the time of P-III, participating sites were experienced in the use of Impella. (3) Outcomes over time were analyzed as follows, a cohort including all IABP and Impella patients from P-II alone was compared to the cohort used for the primary analysis (all IABP and all Impella) and to a cohort of patients including all IABP and Impella from P-III alone.

The win statistics included the WR, win odds (WO) and net benefits (NB). As described previously, these 3 statistics complement each other to show the strength of the treatment effect. With Pt = probability/proportion of win in the Impella group, and Pc = probability/proportion of win in the IABP group, the WR equals Pt/Pc, the NB equals Pt– Pc and the WO equals (Pt + 0.5 [1– Pt– Pc])/(Pc + 0.5 [1– Pt– Pc]). All analyzes were performed using the R WINS Package, version 1.5.1, published 2025-07-02. This package has been previously described and used in similar analyzes. The Tau value (τ) was set at 0.1 (difference of 0.1 unit is the minimum difference needed to consider outcomes better in 1 group vs the other), the significance level was set at α = 0.05, with 2-sided p values.

Results

Patient cohorts

Baseline characteristics of all patients are shown in Table 1 . A total of 211 IABP versus 719 Impella patients (503 from P-III and 216 from P-II) were included. Impella patients were slightly older than IABP patients (68.8 [standard deviation (SD): 11.1] vs 66.9 [SD: 10.8], p = 0.034). No significant differences were observed in rates of index comorbidities between cohorts. Average LVEF values were low and not different between groups, (23.1 [SD: 7.3] vs 24.1 [SD: 6.3], p = 0.077).

Table 1

Baseline characteristics of all patients (N (%) unless otherwise specified)

IABP Impella p Value
N 211 719
Age (mean (SD)) 66.9 (10.8) 68.8 (11.1) 0.034
Male 173 (82.0) 557 (77.5) 0.190
Caucasian 158 (74.9) 531 (73.9) 0.833
Hypertension 179 (84.8) 628 (87.3) 0.406
Diabetes 104 (49.3) 392 (54.5) 0.207
Prior Stroke 36 (17.1) 102 (14.2) 0.356
Renal Insufficiency 63 (29.9) 178 (24.8) 0.162
COPD 61 (28.9) 163 (22.7) 0.076
LVEF (mean (SD)) 24.1 (6.3) 23.1 (7.3) 0.120
NYHA of III or IV 96 (64.0) 286 (72.2) 0.077

The rate of individual endpoints are shown in Table 2 . These rates are based on a nonhierarchical analysis as all endpoints are used in the win ratio analysis. Mortality, sMI and stroke did not show differences between the groups. Statistical differences were observed for rehospitalization (39% vs 25%, p < 0.001) and pMI (7% vs 2%, p = 0.005).

Tags:
Aug 8, 2026 | Posted by in CARDIOLOGY | Comments Off on Impella Protected Percutaneous Coronary Intervention Outcomes Compared With Intra-aortic Balloon Pump: A Contemporary View

Full access? Get Clinical Tree
Get Clinical Tree app for offline access