Periprocedural Adverse Events and Late Mortality: More Nuance From EXCEL

The risk of major adverse events during the periprocedural period is roughly fourfold higher with CABG surgery than with PCI for left main coronary artery disease, but regardless of the initial procedure, these major complications are associated with early and late mortality, according to a new analysis from EXCEL.

Investigators say that reducing the risk of major periprocedural events, both with surgery and PCI, could reduce the long-term risk of death in patients undergoing revascularization for left main disease.

“A major adverse event, as we defined them, was an independent predictor of 5-year mortality—both within the first 30 days and between 30 days and 5 years—to a fairly comparable degree after both PCI and CABG,” Gregg Stone, MD (Icahn School of Medicine at Mount Sinai, New York, NY), the study’s senior author, told TCTMD. “It suggests that if we can improve both procedures by reducing periprocedural complications, we can continue to improve prognosis most likely after both PCI and CABG.”

These prespecified nonfatal major adverse events included MI, stroke, TIMI major or minor bleeding, transfusion (≥ 2 U of blood), major arrhythmias, ventricular tachycardia or fibrillation requiring treatment, bradyarrhythmia requiring a temporary or permanent pacemaker, unplanned revascularization for ischemia, unplanned surgery or therapeutic radiologic procedures, renal failure, sternal wound dehiscence, severe infection or sepsis, prolonged intubation, and post-pericardiotomy syndrome.

Overall, the rate of any predefined nonfatal major adverse event during the 30-day periprocedural period was 11.9% in the PCI arm and 45.4% in the surgical arm, a rate that reflects the higher up-front risks of CABG surgery, say investigators.

“Even cardiac surgeons acknowledge, and patients and physicians all understand, that PCI is an easier procedure to go through with less periprocedural complications,” said Stone. “The benefits of surgery are really sustained late, with its greater durability. So, the purpose of coming up with this major adverse event endpoint was to document and [quantify] the actual difference in those periprocedural complications and how those complications contribute to the different prognosis after PCI and CABG surgery.”

If we can improve both procedures by reducing periprocedural complications, we can continue to improve prognosis most likely after both PCI and CABG. Gregg Stone

Cardiac surgeon Marc Ruel, MD (University of Ottawa, Canada), who wasn’t involved in the analysis, called the new paper an “interesting exercise,” but cautioned against ascribing cause-and-effect to the risk of death associated with periprocedural complications.

“It has been known for a long time that a strong correlation exists between short-term and long-term adverse events after surgery, even blood transfusions,” Ruel told TCTMD. “People who won’t live as long generally tend to be the same ones as those who are more likely to have complications after a procedure, but that doesn’t mean that there is a direct causal link between the two.”

Ruel acknowledged the higher up-front risk of complications with surgery compared with PCI, but said he believes they are justified considering the long-term durability of the procedure.

“You have to break some eggs to make an omelet,” he told TCTMD. “Obviously, the more invasive the procedure, there will be more complications you’ll have up front. CABG is more invasive, there’s no question—we have to work on the invasiveness of the surgery, and it’s been the theme of my career for 20 years—but it’s more definitive. If patients can withstand the invasiveness of CABG, there’s huge benefits at the end. It’s an investment.”

Interventional cardiologist Ankur Kalra, MD (Franciscan Health, Lafayette, IN), agreed, noting that while PCI is associated with less up-front morbidity, if patients can weather the early storm of CABG surgery, it is associated with fewer spontaneous MIs, less repeat revascularization, and better long-term survival.

He also said the higher rate of major adverse events after surgery would be expected because many of them, such as blood transfusions, renal failure, prolonged intubation, sternal wound dehiscence, post-pericardiotomy syndrome, infection requiring antibiotics or sepsis, and major arrhythmias, aren’t expected complications with PCI.

“The fact that there is a fourfold increase in major adverse events following surgery is not a surprise,” said Kalra. “The classic major adverse events we typically see in PCI trials are stroke, MI, death, and bleeding.”

Predictors of Mortality at 5 Years

The primary endpoint of the new EXCEL analysis, which was published in the February 13, 2023, issue of JACC: Cardiovascular Interventions, was all-cause mortality out to 5 years. During this time there were 117 and 87 deaths from any cause, and 61 and 48 cardiovascular deaths, in the PCI and surgical arms, respectively.

Overall, patients undergoing PCI had less periprocedural protocol-defined MI and TIMI major or minor bleeding. They were also less likely to require a transfusion, to have a major arrhythmia, to require unplanned surgery or a radiologic procedure, to develop renal failure and major infections/sepsis, or to require prolonged intubation. Stroke, unplanned repeat revascularization, and post-pericardiotomy syndrome were infrequent in both groups.

Any major adverse event during the 30 days after PCI was associated with a significantly higher risk of death at 5 years when compared to those who didn’t have a complication (OR 4.61; 95% CI 2.71-7.82). The same association held for those who had a periprocedural major adverse event in the CABG arm (OR 3.25; 95% CI 1.95-5.41). The risk of cardiovascular death was also significantly higher among the patients in both the PCI and surgery arms who had a periprocedural major adverse event (ORs 5.80 and 4.01, respectively).

You have to break some eggs to make an omelet. Marc Ruel

As Stone noted, any periprocedural complication was associated with all-cause and cardiovascular death within the 30-day period after the procedure as well.

In terms of individual events, periprocedural TIMI major or minor bleeding requiring a blood transfusion was an independent predictor of death at 5 years in both the PCI and surgery arms. Stroke, unplanned revascularization for ischemia, and renal failure were independent predictors of death only after CABG surgery. Unplanned revascularization after CABG surgery is a rare complication, Stone said, noting it suggests graft failure.

In a sensitivity analysis, the EXCEL researchers addressed the MI question given the controversy around their chosen definition, as well as the hullabaloo that arose when they initially failed to published MI event rates based on the Universal Definition, a secondary endpoint. Those have subsequently been published, including a detailed analysis of different MI definitions.  

Based on the Third Universal Definition, patients undergoing PCI were more likely to have a periprocedural MI than those treated with surgery. When they used the Third Universal Definition as part of their composite major adverse event endpoint, as opposed to the per-protocol MI definition, the occurrence of any major complication remained an independent predictor of death at 5 years in the PCI and CABG arms. In multivariable analysis, periprocedural MI based on the Third Universal Definition was an independent predictor of all-cause mortality after CABG surgery, but not after PCI.

Speaking with TCTMD, Ruel said the question of whether one definition of perioperative MI is better than another for predicting the risk of subsequent mortality has little relevance because the EXCEL trial captured mortality as part of the study’s primary composite endpoint of all-cause death, MI, or stroke at 3 years and in later 5-year follow-up.

“Much more important here is the fact that there were 38% increased odds of death with PCI over CABG at 5 years in EXCEL,” said Ruel. “One cannot argue with a dead person, especially in a randomized controlled trial of 1,905 patients which addresses one of the main killers of human beings, namely, left main coronary stenosis.”

Mortality Remains an Issue

The mortality question is one that has nagged at interpretations of EXCEL and led to significant dispute between those in the cardiac surgery community and interventional cardiologists. The 5-year results of EXCEL, which were presented in 2019, showed that deaths from all causes were 13.0% in the PCI arm compared with 9.9% in the CABG arm, a difference that led to a very public, high-profile fight between David Taggart, MD, PhD (University of Oxford, England), the chairman of the EXCEL surgical committee during the design and recruitment phase of the trial, and the rest of the investigators.

Eric Bates, MD (University of Michigan, Ann Arbor), who wrote an editorial accompanying this latest EXCEL analysis, believes too much is made of the difference in major adverse events between PCI and CABG.

“The fewer major adverse events after PCI compared with CABG is overemphasized in this report, with the conclusion implicitly suggesting that PCI should be a safer procedure because major adverse events is strongly associated with death,” writes Bates. “Underemphasized is that absolute all-cause death (12.5% vs 9.4%) and cardiovascular death (6.5% vs 5.2%) rates were higher after PCI compared with CABG.”

For Bates, it’s difficult to square the fourfold higher risk of periprocedural major adverse events with CABG compared with PCI in EXCEL with the fact that surgery has a lower overall mortality risk at 5 years.

As for the discrepancy between more periprocedural adverse events with CABG but less mortality, the EXCEL investigators point out that of the 117 deaths in the PCI arm at 5 years, 28.2% occurred in patients who had a major periprocedural adverse event. In other words, 72.8% of the deaths in the PCI arm were unrelated to periprocedural adverse events. The EXCEL investigators have previously explained that the difference in all-cause mortality with PCI was driven by more noncardiovascular deaths, including a higher rate of late malignancy and sepsis that remains unexplained. In the main EXCEL publication, there was no significant difference in the risk of cardiovascular death at 5 years between the two procedures.

Ruel, on the other hand, said the adjudication of clinical events in trials is “deeply imperfect.” For that reason, the most prominence should be given to all-cause mortality, an endpoint patients care most about and that favored surgery in EXCEL.

“Whether it’s cardiovascular, all-cause, or noncardiovascular death, however you want to name it, a death is a death,” added Kalra. “Death is something all of us strongly care about and at the end of the day, patients with left main disease who underwent CABG ended up surviving longer than patients who got PCI.”

While EXCEL mortality data continue to spark discussion, a subsequent meta-analysis that included EXCEL, published in the Lancet  in 2021, found no difference in the risk of all-cause mortality between PCI and CABG in patients with left main coronary disease. The Lancet analysis was intended to inform the European Society of Cardiology’s ongoing guideline review but there’s no word on when that will wrap up.

Kalra recommended physicians make decisions based on the totality of the evidence coupled with individual patient characteristics. For example, he noted, ostial left main disease in an otherwise healthy, active patient is favorable for PCI whereas distal left main disease may favor CABG. 

“Each patient is different, each anatomy is different,” he said. “Each patient has their own comorbidities. Trials are important, they should be done, and they inform practice, but it’s up to the practicing clinicians to ascertain which aspects of the trial are applicable to the patient in front of them.”