Trickier Patients, Fewer Adverse Events in High-risk PCI: Impella PROTECT III Registry

Greater familiarity with the device and better patient selection appear to be driving adverse events down among high-risk patients whose PCI is supported using the Impella (Abiomed) left ventricular assist device, suggest the latest data from the single-arm PROTECT III postapproval study. And while the field is still clamoring for a larger, contemporary randomized trial, these registry data suggest that despite an increasingly complex patient mix, the risk of adverse outcomes is shifting downward.

The results provide a contemporary update to the PROTECT II randomized controlled trial, published 8 years ago in Circulation, William O’Neill, MD (Henry Ford Hospital, Detroit, MI), said when he presented the data in the “Best of Abstracts” session at the virtual TCT Connect 2020.

PROTECT II trial found no difference in 30-day major adverse events, its primary endpoint, between patients randomized to Impella or intra-aortic balloon pump (IABP). By 90 days, however, a per-protocol comparison found fewer events with Impella (40.0% vs 51.0% with IABP; P = 0.023). Based on PROTECT II—despite the miss on its primary endpoint—and its predecessor PROTECT I, Impella received US Food and Drug Administration premarket approval in 2015 and has seen its use expand in the ensuing years. Those numbers have been tracked in the PROTECT III registry.

An up-to-date look at how and when the Impella is being used, and with what results, helps fill in some of the gaps, O’Neill suggested. “We felt that a more-robust analysis of real-world data with modern angiographic techniques and the results of best practices would be required.”

Sunil Rao, MD (Duke University Medical Center, Durham, NC), who commented on the analysis for TCTMD, agreed that “it seems to be a reasonable reflection of how Impella is being used in clinical practice, with the caveat that these are all patients who got the Impella. So we don’t know the denominators—patients getting complex PCI without Impella support, site-level variation, etc.”

And while an observational analysis is inherently vulnerable to confounding, Rao said in an email, these data lay the foundation for the PROTECT IV trial, set to launch soon.

Further knowledge about mechanical circulatory support (MCS) in nonemergent high-risk PCI is a key priority in PCI research right now, added Rao. “These devices are likely to be incredibly helpful, but have significant risks and are very, very expensive. We need to understand which patients benefit and which ones don’t. The only way to do that is with an appropriately powered randomized trial.”

PROTECT II-Like Patients

This interim analysis of PROTECT III, an ongoing US FDA postapproval study, involved 1,143 patients undergoing elective nonemergent PCI with Impella 2.5 and Impella CP at 45 sites between March 2017 and September 2019.

Among the PROTECT III patients, 494 were considered to be “PROTECT II-like,” with an LVEF ≤ 35% with unprotected left main disease or an LVEF ≤ 30% with three-vessel disease, and none of the following complications: STEMI within 24 hours of Impella placement, cardiogenic shock, renal failure (creatinine < 4 mg/dL), or low platelet count (< 75,000/mm³).

PROTECT II-like patients in PROTECT III tended to be older and were more likely to be female while less likely to be white than Impella-treated patients in PROTECT II. They also were more likely to have prior stroke and renal insufficiency but less likely to have prior MI, though not all of the above differences reached statistical significance. Their mean LVEF levels were similar at 22.64% and 23.4%, respectively.

Additionally, patients in this registry analysis had a higher number of vessels treated as well as greater prevalence of left main and triple-vessel disease than did patients in the randomized trial. O’Neill drew attention to the more-frequent atherectomy use in PROTECT III (37.04% vs 14.80% in PROTECT II), which might be one reason for the longer mean support time (mean 5.32 vs 1.90 hours (P < 0.001 for both). Contrast volumes were lower, and more patients were discharged from the cath lab on device support in recent clinical experience.

“All this goes to show that patients were getting sicker, older, more complicated, with equally impaired ventricular function,” O’Neill said, while “lesions were longer, more heavily calcified, and much more complex in PROTECT III.”

Angiographic results showed slightly lower SYNTAX scores before PCI and greater reduction in SYNTAX score post-PCI for PROTECT II-like patients versus the original PROTECT II group, O’Neill continued, “meaning that there was more revascularization.” Ischemia jeopardy scores were similar across the two studies at baseline, but the PROTECT III patients showed greater gains in this metric, he noted. “This really allows us to understand that much more extensive and complete revascularization was performed in PROTECT III versus PROTECT II.”

Moving to in-hospital outcomes, O’Neill showed that clinical events across a range of endpoints suggest that Impella use in recent years is safe, with patients experiencing lower rates of major events across a range of ischemic, arrhythmic, and bleeding outcomes. Today, “because of the improvement in the procedure and the periprocedural management (the groin management), there’s a better safety profile,” he told attendees.

In-Hospital Outcomes

Stroke and vascular complications were low and similar for both cohorts. Highlighting the bleeding reduction, O’Neill pointed out that in PROTECT III, “we are now able to use the Perclose technique, [image]-guided access, and better management of the groin site.”

PROTECT II had originally defined periprocedural MI as a post-PCI rise in CK-MB of 3 times the upper limit of normal (ULN) or greater within 72 hours post-of the procedure. To dovetail with PROTECT III, O’Neill and colleagues used a more-recent definition, outlined in a 2014 paper, that set the bar at development of new Q-waves or a CK-MB elevation of eight times the ULN or greater within the same time frame; when CK-MB was not available, troponin values were used.

At 90 days, the rate of MACCE—defined as death, MI, stroke/TIA, or revascularization—were significantly lower in PROTECT II-like patients than in the Impella-treated PROTECT II group (15.0% vs 21.9%; P = 0.035) and those who received IABP in the randomized trial (P < 0.0001).

The researchers matched patients according to age, gender, LVEF, diabetes, renal insufficiency, prior MI, and ACS, arriving at 325 pairs. To account for the higher periprocedural MI rate in PROTECT II, they conducted a landmark analysis at 3 days; even beyond this, MACCE were significantly reduced through 90 days.

“We believe that this data suggests the results are really getting much better for these patients,” justifying the upcoming PROTECT IV randomized trial that incorporates best practices, O’Neill concluded. In that study, Impella will be compared against traditional therapy consisting of either no treatment or other forms of hemodynamic support.

Hope for PROTECT IV

Panelist Alaide Chieffo, MD (San Raffaele Scientific Institute, Milan, Italy), said after O’Neill’s presentation that, while awaiting large randomized trials, “at least we have some fresh data that can support everyday practice.” She asked O’Neill whether they have been able to identify what factors might be leading to superior outcomes in PROTECT III, such as centers’ expertise or specific protocols. “Because we have seen actually it’s a more-complex population,” she observed. “So why are they doing better?”

The shift speaks to improved operator experience, O’Neill suggested. “Even in PROTECT II, year-by-year experience grew and outcome improved. The technique now is 10 years old. It’s achieved a degree of maturity. I think people know better which kinds of patients to recruit.”

Asked by TCTMD whether the degree of differences between PROTECT II and III were convincing, Rao said: “These patients were compared against a historical control of different patients, different PCI eras, etc. It’s one of the most confounded ways of making comparisons.” Before and after propensity-matching, “they are pretty much the same curves. There is too much confounding here to make any firm conclusions,” he commented.

At the 2019 American Heart Association Scientific Sessions, a database study sparked headlines by showing worse outcomes with Impella versus no MCS among nearly 50,000 patients. Rao, senior author of that paper, which was published in Circulation, pointed out that both their analysis and the “current presentation are observational and suffer from measured and unmeasured confounders.

“PROTECT III,” he continued, “doesn’t have a contemporary control group of patients undergoing PCI without support or with other forms of MCS. This is the equipoise that sets up the PROTECT IV randomized trial. I am hopeful that PROTECT IV will go to completion and we will have good data to guide practice.”

Randomized trials of already-approved devices are notoriously difficult, because operators already convinced of their benefits are often reluctant to enroll patients who have a 50:50 chance of not getting the therapy. With Impella, however, a growing chorus of voices—even those of self-described converts—have called for randomized data both in high-risk PCI as well as in the setting of acute MI complicated by cardiogenic shock.

“PROTECT IV is a very important study,” Rao stressed. “Kudos to Abiomed for supporting this trial. It is now incumbent on the clinical community to realize we have equipoise and enroll patients.”