FAME 3 fails to defame coronary artery bypass grafting: what went wrong in the percutaneous coronary intervention arm?

The primary results of the landmark Fractional Flow Reserve Versus Angiography for Multivessel Evaluation (FAME) 1 trial found in 2010 that case selection using physiological assessment of stenosis severity by measurement of fractional flow reserve (FFR) improves outcomes compared with angiography-guided percutaneous coronary intervention (PCI) [1]. Results were confirmed at 5 years follow-up [2, 3] influencing the current revascularization guidelines [4]. A subanalysis of the same study found a worrisome proportion of patients with angiographically 3 vessel coronary artery disease (3-VD) who actually, by functional significance, turned out to have either no stenoses or have been downgraded to one- or two-vessel coronary artery disease and received different treatment [5]. Because hypothetically, this whole group of angiographically 3-VD patients would have been candidates for coronary artery bypass grafting (CABG) but in retrospect would not have benefitted from it, the FAME 3 trial was conducted to address this question, if stenting of only the functionally significant lesions in patients with multivessel disease overcomes the ‘universal’ grafting, or at least matches it. Because current guideline recommendations for CABG were determined by older large randomized studies where PCI was done using bare-metal stents or first-generation drug-eluting stents [6–8], a contemporary assessment of the role of PCI in 3-VD has been eagerly awaited by the medical community. In a head-to-head comparison of 2 revascularization strategies for patients with 3-VD, PCI guided by FFR failed to match CABG surgery for the reduction of major adverse cardiovascular and cerebrovascular events (MACCE) [9]. Of those randomized to PCI, 10.6% met the 1-year primary end point of MACCE, compared with only 6.9% of patients assigned to CABG [hazard ratio (HR) 1.5 (confidence interval (CI) 1.1–2.2), P = 0.35 for noninferiority]. One thousand five hundred patients with angiographically 3-VD (at least 50% diameter stenosis by visual assessment in each of the 3 major epicardial or major side branches) not involving the left main were randomly assigned to FFR-guided PCI or CABG. All lesions were declared appropriate for either PCI or CABG by the local Heart Team. The average age was 65, the majority was male (>80%) and 29% had diabetes. Thirty-nine percentage of patients were diagnosed with acute coronary syndrome. The average SYNTAX I score was 26, with an average of 4.3 stenoses per patient. Chronic total occlusions were observed in 22% of the patients, and 68% had at least 1 bifurcation lesion. Recent ST-segment elevation myocardial infarction, cardiogenic shock and a left ventricular ejection fraction of <30% were all exclusion criteria. FFR evaluation of all stenoses was performed, excluding subtotal or complete occlusions. Twenty-four percentage of the studied lesions had an FFR of >0.80. PCI was conducted using only the latest generation of durable-polymer Zotarolimus-eluting stents in stenoses with an FFR of ≤0.80. Intracoronary imaging was performed at the discretion of the operator in around 12% of PCI procedures. FFR after PCI was conducted in 60% of patients, with a median value of 0.88. CABG was carried out in compliance with standard practice throughout all centres, and full arterial revascularization was strongly advised (24.5% had multiple arterial grafts, with the left internal mammary artery being utilized as a graft in 97%). If FFR was recorded during the diagnostic procedure (10% of the time), this information could be applied to guide CABG at the discretion of the surgeon. FFR invasively assesses the haemodynamic impact of coronary disease, thereby adding functional information to the two-dimensional anatomic assessment made by coronary angiography to determine the utility of revascularization. However, recent data have not uniformly shown FFR to provide a mortality benefit over angiography-guided revascularization [10–12] and further analyses may be warranted to define the role of FFR outside stable coronary artery disease (CAD) (as documented in the FAME and FAME 2 trials). Probably the crucial answer that FAME 3 needs to provide to move the field forward is whether the excess events in the PCI arm were due to target lesion revascularization (e.g. in-stent restenosis or stent thrombosis) or non-stented lesions deferred by FFR (e.g. unstable angina/non-ST-elevation myocardial infarction, ST-elevation myocardial infarction). Events leading to repeat revascularization were not defined according to target lesion. In other words, what was done wrong, FFR or PCI? Such a ‘lesion-centric’ question digresses the essence of whether CABG remains the treatment of choice for 3-VD with moderate-to-severe anatomic complexity as defined by the SYNTAX score. The difference between the 2 treatment strategies could be explained if the events were a consequence of de novo disease progression because the grafted vessel is protected from a possible atherothrombotic event upstream of the bypass, while the stented vessel would suffer an acute event caused by physiologically non-significant but unstable plaques. Indeed, in the dual antiplatelet therapy study, about 50% of recurrent events occurred in a non-target lesion [13]. This brings us back to the concept of vulnerable plaque and that other variables besides functional testing should be included in PCI decision-making. Plaque histopathology and prophylactic stenting (plaque sealing) were recently tested, showing a five-fold higher rate of major adverse cardiac events when thin-cap fibroatheromas were identified, despite the absence of ischaemia [14–16]. Second, there are valid criticisms of the PCI arm due to the low frequency of intracoronary imaging, which may have affected the outcome. About 4 stents per patient were implanted, with an average length of 80 mm, while intravascular imaging was only used in 11.7% of the cases. Although this may represent real-world practice, it is difficult to explain this small percentage in the context of complex multivessel disease and that all enrolled centres have this technology. Many meta-analyses have demonstrated a reduction in major adverse cardiac events, repeat revascularization and in-stent restenosis with the use of intracoronary imaging [17–21]. Furthermore, at least 1 artery was chronically occluded in 22% of the patients and therefore not available for FFR. It remains unknown how many of these chronic total occlusions were treated in the PCI arm. Completeness of revascularization is unknown in both arms, which is of importance given that full revascularization conferred a benefit in the SYNTAX trial [6]. Third, the periprocedural myocardial infarction definition adopted may have partially impacted the results. FAME 3 used the Third Universal Definition of MI, type 5 (CABG), with rates of 3.3% for PCI vs 2.3% for CABG. If the SCAI definition had been used, which is more prognostically relevant [22], the rates would have been 10.3% for PCI vs 14.5% for CABG and non-inferiority would have been met. Fourth, in the CABG group, periprocedural complications such as atrial fibrillation, acute kidney injury, major bleeding according to the Bleeding Academic Research Consortiu criteria and a 30-day rehospitalization rate were more prevalent. Repeat revascularization was included in the primary end point (5.9% PCI vs 3.9% CABG), but this event (usually an outpatient elective procedure) can be disputed as less important than major bleeding, acute kidney injury or unplanned rehospitalization, all of which favoured PCI. There did appear to be a benefit for PCI among patients with low SYNTAX scores (0–22), but investigators caution that this subgroup analysis is hypothesis-generating only. The difference between the curves increases with the SYNTAX score. Thus, the anatomical complexity should influence the decision-making, not the amount of ischaemia. Of course, an important limitation is the relatively short duration of follow-up (1 year) at this time. Coronary disease progression is accelerated after CABG [23] and around half of the venous grafts are found to be occluded at 5-years [24]. Surgery will be difficult to match, given that the 6.9%-incidence of events in the CABG group was almost half of the traditional 10–13% MACCE rate seen in previous PCI versus CABG trials for multivessel disease [6–8, 25, 26]. The surgical results were unexpectedly good, and it should be acknowledged that the surgical community has focused on quality improvement in recent years (more arterial grafting, off-pump surgery, minimally invasive direct coronary artery bypass, intraoperative Doppler graft assessment, etc.) and these marginal gains account for markedly improved results compared to trials and registries reporting 10 years ago. In the US guidelines for the treatment of stable ischaemic heart disease, CABG has a class I recommendation (level of evidence B) for the treatment of patients with 3-VD, with the guidelines stating ‘it is reasonable’ to choose CABG over PCI in patients with complex three-vessel CAD (SYNTAX score >22) who are good candidates for surgery. PCI, in this setting, has a IIb recommendation, meaning that it is “of uncertain benefit’ [4]. With the European guidelines for myocardial revascularization, PCI has a class IA recommendation for the treatment of patients with 3-VD and a low SYNTAX score (≤22). For those with an intermediate-to-high SYNTAX score, however, PCI is not recommended (class III). In contrast, CABG surgery has a class IA recommendation in patients regardless of the SYNTAX score [27]. FAME 3 reaffirms that not even the support of FFR guidance managed to diminish the benefits of CABG over multivessel PCI and reassures the anatomical stratification of these patients. The study is not a defeat of FFR in general but a defeat for physiology driven PCI in multivessel disease, given the high percentage of positive FFR. What we should take from these results is that many other factors such as intracoronary imaging, lesion preparation, calcium modification, proper bifurcation stenting and possibly in the near future identification of vulnerable plaques must be considered in extensive coronary disease, not just FFR. FFR must regain its recently lost ground with the FLOWER-MI, FUTURE and RIPCORD-2 trials [10–12]. Absolute differences are more important than relative differences and the magnitude of the differences in the 2 most important major events was small: death (0.7% favouring CABG) and stroke (0.2% favouring PCI). Patients need to be informed about all the risks and benefits of each procedure (absolute differences, not relative) to make an informed decision. There are many patients that look for trade-offs; some patients will accept a higher event rate to avoid a long recovery, and vice versa. There is still a signal that when decisions are made with appropriate guidance, either strategy probably can be employed for patients with single- and two-vessel CAD. Even for patients with 3-VD and lower SYNTAX scores, FFR-guided PCI produced decent results. As we saw in the SYNTAX trial [6], for the more complex patients with 3-VD, CABG stands out as the treatment of choice and a discussion with the surgeon, anaesthesist and patient remains imperative. The Heart Team has become an important paradigm in clinical practice, emphasizing the importance of team consensus on the optimal approach to revascularization. Overall, FAME 3 showed that PCI still needs to improve in the treatment of complex, multivessel coronary artery disease. One solution could be to reduce the amount of metal (malapposed and underexpanded) and switch to drug-coated balloons, which provide promising results [28–30]. Besides functional assessment, contemporary PCI practice should also include intracoronary imaging, extensive plaque modification and proper bifurcation techniques to obtain an irreproachable result with an improved long-term outcome. All of these criteria should be requisite for future research trials. What we do today will matter in 5–10 years.