Impella in High-risk Nonemergent PCI: Early Differences Disappear by 1 Year

Compared with no added support, use of the Impella percutaneous left ventricular assist device (Abiomed) in patients undergoing high-risk nonemergent PCI is associated with more periprocedural MI events, data from a single center show. Still, by 1 year, MACE and its components did not differ with or without mechanical circulatory support (MCS).

Speaking with TCTMD, lead author Lorenzo Azzalini, MD, PhD (Icahn School of Medicine at Mount Sinai, New York, NY), said the study adds to the slim and inconclusive evidence base on MCS in this setting.

PROTECT II, for instance, showed similar 30-day outcomes for high-risk PCI patients regardless of whether they received an intra-aortic balloon pump (IABP) or Impella. When that window was extended to 90 days and calculations were per protocol rather than based on intent to treat, Impella patients came out ahead with lower mortality than IABP patients (40% vs 51%; P = 0.023).

However, these positive signs from PROTECT II “must be considered hypothesis-generating since they’re derived from a post hoc analysis,” Azzalini said. “And this prompted us to conduct our study.” That work, with Samin K. Sharma, MD (Icahn School of Medicine at Mount Sinai), as senior author, was published online recently in Catheterization and Cardiovascular Interventions.

Another recent report, published earlier this year in the same journal, showed reduced in-hospital mortality with Impella versus IABP in a nationwide sample of 21,848 nonemergent PCI patients.

At his center, Azzalini noted, physicians began using Impella a decade ago, so have a lot of experience with the device. Here, the researchers chose to compare Impella to no support “because we think that IABP provides almost no support at all as far as cardiac output is concerned,” he explained, and this most closely resembles the “clinical dilemma faced by interventionalists nowadays: should I perform this procedure with Impella or can I get away with no support?”

Asked how to interpret their results showing no differences at 1 year, Azzalini was circumspect. “I think that it would be a very shallow and wrong interpretation to say that Impella provides no benefit compared to no support, since there is very ample literature showing superior hemodynamic support offered by Impella compared with intra-aortic balloon pump . . . both in cardiogenic shock and in high-risk PCI. So how do we reconcile this with our findings?”

One answer, he suggested, is that “it’s very hard to fully capture and adjust for the decision-making process that the operator goes through when facing such cases.”

Duane Pinto, MD (Beth Israel Deaconess Medical Center, Boston, MA), commenting on the results for TCTMD, also said it’s tough to draw conclusions from these data. “They made an attempt to have a control group and [did] propensity matching, which has an advantage over registries, but basically any retrospective look using mechanical support devices is fraught with some element of unmeasured confounding,” he observed. “Unless there’s a prospective look it’s hard to say that we have the definitive answer that Impella is worse, better, or the same.”

A Window Into Decision-making

Azzalini and colleagues reviewed data on 250 patients who underwent high-risk nonemergent PCI supported by Impella 2.5/CP between January 2009 and June 2018 at Mount Sinai Hospital, propensity matching them to 250 controls. Mean age was 69.5 years, and 76% of patients were men. Comorbidities were common; for example, 52% had diabetes, 45% had prior MI, and 41% had chronic kidney disease. Impella patients were significantly more likely than controls to have previously undergone CABG but less likely to have dyslipidemia. SYNTAX scores, though, were similar in the two groups.

In the matched analysis, there remained some procedural differences: patients in the Impella group were more likely to undergo left main PCI (26% vs 11%) and had a higher number of treated vessels (1.8 vs 1.3; P < 0.001 for both) than patients in the control group. The Impella patients also showed a trend toward more rotational atherectomy use (44% vs 37%; P = 0.10).

Azzalini attributed these findings to operators’ greater confidence when performing Impella-supported procedures.

Perhaps partially as a result of this more aggressive PCI, the co-authors say, periprocedural MI was more common in the Impella group (14.0% vs 6.4%; P = 0.005), as were major bleeding (6.8% vs 2.8%; P= 0.04) and need for blood transfusion (11.2% vs 4.8%; P = 0.008). But by 1 year, there were no differences in overall MACE, defined as all-cause death, MI, and TLR (31.2% vs 27.4%; P = 0.78), or any of its components.

On top of propensity matching, the researchers also performed doubly-robust Cox regression analysis. Here, too, Impella and no support were associated with similar 1-year MACE risk (HR 0.87; 95% CI 0.55-1.40).

MCS devices do a good job of providing superior hemodynamic support. We just need to be able to identify the right patients for the device. Lorenzo Azzalini

One thing the Mount Sinai researchers didn't report is residual SYNTAX score, a metric that would capture the completeness of revascularization. Looking ahead, the randomized PROTECT IV trial will hopefully provide some clarity on whether Impella enables more-complete revascularization and whether this would affect long-term outcomes, Azzalini noted.

Decision-making in nonemergent PCI is intricate and variable across centers, he pointed out. What’s known now, he concluded, is that “MCS devices do a good job of providing superior hemodynamic support. We just need to be able to identify the right patients for the device.”

In this data set, the patient population had a heavy comorbidity burden and tended to be undergoing complex procedures—not all high-risk cases are equal and decisions must be individualized, Azzalini urged. “We don’t want to send the wrong message, that the operator can go out there and start doing [rotational atherectomy] on the left main patient with an ejection fraction of 15 and elevated filling pressures.”

For Pinto, overall the paper restates a known concept: that more aggressive PCI can up the risk of periprocedural MI but, over time, lessens the risk of restenosis. “With an Impella in place, people are more aggressive with PCI,” he said, adding that higher-quality data are needed.

William Lombardi, MD (Heart Institute at UW Medical Center—Montlake, Seattle, WA), said that while the series is large, what’s missing is the “what we learned” takeaway. “It doesn’t give us any teaching points,” he commented.

Several key aspects are missing from the analysis, according to Lombardi. For one, there’s no look at how bleeding might have changed over time as operators became more comfortable with large-bore access. He, too, said that residual SYNTAX is crucial in understanding the quality of the PCI performed.

“The outcomes with Impella and the outcomes with PCI are far more related to the operator and their completeness of revasc rather than the pump or the stents. And as long as you continue to evaluate [results] based on the pump and the stents rather than the completeness of revasc, the data is murky,” Lombardi stressed.