Richmond, VA - A new meta-analysis of trials comparing PCI with medical therapy in patients randomized later than 12 hours after MI has shown a significant benefit in cardiac function and mortality in the PCI group [1].

Lead author Dr Antonio Abbate (Virginia Commonwealth University, Richmond) commented to heartwire: "Our results suggest that in the post-MI period there does appear to be a benefit of PCI in preventing late mortality, and this appears to be brought about by a reduction in heart failure."

The results could appear to contrast with the OAT trial, the largest study in this field, which showed no benefit of PCI, with a trend toward more repeat MIs with PCI than with medical treatment. But OAT lead investigator, Dr Judith Hochman (New York University School of Medicine, New York), told heartwire that this meta-analysis has several caveats and should not be used as an argument against the OAT results. Most important, she said, the meta-analysis included many different types of patients—it mixed open and closed arteries and included some studies that selected patients with ischemia, whereas OAT studied only patients with a totally occluded artery without significant symptoms or severe ischemia.

"The pathophysiology and evidence bases for those with open vs closed arteries and those selected for ischemia are different. If post-MI the artery is totally occluded and there is no severe ischemia, triple-vessel disease, clinical instability, or symptoms that require PCI, then PCI is not recommended. If there is an open artery in the late phase, then it would depend on ischemia, clinical course, and anatomic determinants. Grouping all these things together does not help to inform medical practice. The ACC/AHA guidelines clearly recognize this and have different recommendations based on these multiple patient subsets," Hochman said.

Abbate agrees that the benefit of PCI is not the same for all patients. "Our results suggest that most benefit from PCI appears to be gained in patients who have longer life expectancy, those who have viable myocardium with ischemia, and those without a totally occluded infarct-related artery," he said. "But we still saw an overall benefit, and we believe the OAT results on their own have been overinterpreted," he added.

He explained that prior to OAT, eight small studies had been published on late PCI in MI patients, and all but one had suggested a survival benefit with revascularization. While the OAT study failed to show any benefit of PCI, Abbate pointed out that the trial had difficulty enrolling patients, as many cardiologists believed PCI was beneficial for this group, leaving potentially less ideal candidates available for randomization. Hochman disputes this assertion. "The lack of PCI benefit in OAT was seen at high-enrolling centers that enrolled almost all eligible patients as well as low-enrolling centers," she told heartwire.

After the subsequent SWISSI II study, which showed a survival benefit for patients randomized to late revascularization, Abbate et al conducted the current meta-analysis, which he says "is not meant to be an alternative to the OAT study but rather an integration of information available to medical practice with a diverse assessment of the question."

The meta-analysis, to be published in the March 4, 2008 issue of the Journal of the American College of Cardiology, included 10 studies involving a total of 3560 patients one to 26 days post-MI who were randomized to PCI or medical therapy. Results showed improved survival in the PCI patients.

Ejection fraction was also improved in the PCI group (+4.4% change; 95% CI 1.1-7.6; p=0.009), which Abbate said suggested that the survival benefit of PCI is probably mediated by a reduction in adverse remodeling and heart failure. "Reinfarction was similar in both arms, and if anything there was a trend toward fewer MIs in the PCI arm," he said.

He noted that the median time of PCI in the meta-analysis was 12 days, but they included patients 12 hours or more after MI, whereas OAT enrolled patients three to 28 days post-MI. "There are many patients who come in 12 to 72 hours after an MI. These people were not studied in OAT." But he pointed out that the BRAVE-2 study looked specifically at patients in the 12-72-hour window post-MI and showed a benefit of PCI.

"I hope that our results will keep the question of late PCI open. We could do with another study in this area, as there are thousands of patients in this group in the US each year, but less than 3600 have been enrolled in clinical trials. Of these, more than 50% came from OAT. Our data tell us that there is more we need to learn in this area. I'm not saying all patients should undergo PCI based on this meta-analysis, but I think we should keep the discussion open."

Hochman was critical of the new meta-analysis, pointing out several weaknesses. The first of these was that it included both open and closed arteries. "These are apples and oranges—two very different post-MI populations," Hochman commented to heartwire. "Testing the late open-artery hypothesis requires angiographic confirmation of a total occlusion, which was the case in only six of their 10 studies. In fact, approximately 40% of the non-OAT patients in this meta-analysis did not have total occlusions, and the finding of benefit from late PCI is driven by four studies that included patients who did not have occluded arteries," she noted. She added that a recently published meta-analysis by Ioannidis and Katritsis included only studies with total occlusions and, like OAT, found no reduction in clinical events with PCI [2].

In response, Abbate argues that 84% of patients in the meta-analysis had total occlusion, making it unlikely that the mortality benefit was driven exclusively by the 16% of patients with subtotal occlusion. He adds: "If patients are apples and oranges with many individual characteristics, a study on only apples may not be useful when dealing with oranges, whereas a study on fruit in general may be equally applicable to apples as well as oranges. Similarly, we felt that it was better to include all studies and all patients and then perform a meta-regression to identify potential cofactors. If we had not included the four studies, we would have missed an opportunity to analyze outcome on 910 patients, the majority of whom did have total infarct-related artery occlusion. Furthermore, I am not sure that total and subtotal occlusions are two different disease processes; we believe that they are two presentations of the same process and that significant functional and clinical overlap exists between the two."

Hochman also raised other questions about some of the studies included in the meta-analysis: "They included SWISSI II, which enrolled patients over 10 years ago. Optimal medical therapy, which has been shown to reduce events, was quite different then. They also included BRAVE-2, which tested an invasive strategy (PCI, CABG, or no revascularization based on anatomy) vs conservative strategy. Once they have opened that door and included a broad range of patients, there are thousands of patients available from numerous other studies that they did not include. Conversely, a study that showed an adverse effect of PCI on the LV remodeling (TOAT) was not included."

Abbate noted that in BRAVE-2 only eight patients received bypass surgery rather than PCI, which is unlikely to have changed the 42-day outcome, and he says they decided not to include the TOAT data in the cardiac-function analysis because true baseline values were not available even after they contacted the investigators. He adds that the Ioannidis meta-analysis did include the TOAT study but still showed a 2% greater improvement in left ventricular ejection fraction with late revascularization, "proving in our opinion that a benefit indeed exists."

Hochman said another weakness of the meta-analysis was that it used mean values for the ejection fractions and volumes at baseline and follow-up, rather than paired values. "This means they are not comparing the same patients, which is not valid statistically. SWISSI II, the study with the largest apparent ejection-fraction benefit from PCI, had almost two-thirds missing follow-up data in the medical group, vs one-quarter missing data in the PCI group. No imputation or statistical methods can correct for this." Abbate concedes that the lack of paired data is a limitation, but he argues that statistical methods have been designed and proven reliable when paired data are lacking.

Commenting on the new meta-analysis for heartwire, Dr Brahmajee Nallamothu (University of Michigan, Ann Arbor), who was not involved in either study, said he also had concerns about the appropriateness of combining results from many of these trials. He noted that in the ALKK study, for example, most patients received fibrinolytic therapy and had a patent infarct-related artery at the time of late PCI, so the potential benefits of late PCI are more likely related to stabilization of the infarct-related artery and not the "open-artery" hypothesis, which is a quite different clinical scenario from the one studied in OAT.

But Nallamothu added that not all late-MI patients fit neatly into the specific criteria evaluated by OAT and that there is certainly a role for late PCI in many of them, a decision that may be driven by the presence of symptoms or ischemia and the appearance of the infarct-related artery. "What Abbate and colleagues found is not unexpected when you closely examine the trials that were included. And their conclusions in no way nullify the findings of OAT or its implications for the types of patients it studied. In fact, the most important message of this meta-analysis may be that stable patients after AMI are a diverse group, and decisions about late PCI in them should be individualized (and directed by the guidelines when possible)," he concluded.