Fractional flow reserve–guided PCI as compared with coronary bypass surgery (FAME III)– another brick in the wall!

Abstract

FAME III, a multi-centric randomized controlled trial compared fractional flow reserve (FFR) guided percutaneous coronary interventions (PCI) with coronary artery bypass grafting. The study confirmed that compared with FFR-guided-PCI, coronary artery bypass grafting was associated with a significantly lower incidence of the composite of mortality, stroke, MI, and repeat revascularizations at 1 year. Thus, FFR-guided-PCI failed to meet the “noninferiority” criteria.

Fractional Flow Reserve (FFR) guided Percutaneous Coronary Intervention (PCI) has been shown to be associated with improved outcomes in previous studies. The FAME (Fractional Flow Reserve Versus Angiography for Multivessel Evaluation) study concluded that FFR guided interventions led to better outcomes, compared to interventions based on angiography alone [1]. This was followed by the second study, where FFR-guided-PCI was compared with medical management, known as FAME II. This trial was terminated prematurely, as a significantly higher number of patients on medical management alone required urgent revascularization [2]. The confirmation of positive outcomes of FFR-guided PCI inevitably led to it being compared with coronary artery bypass grafting (CABG) in the FAME III study [3].

The FAME 3 trial was a randomized controlled trial, conducted across 48 centres in 16 countries. It was a non-inferiority trial, and the initial inferiority margin (Δ) of 1.45 was calculated based on the assumption that the primary event would occur in 12% of patients undergoing CABG. Subsequently newer data comparing CABG and PCI was published, where the primary event was found to occur only in 10% of the cases. This led to re-setting the inferiority margin at 1.65 by the trial steering committee.

The population studied included patients with triple vessel coronary artery disease (≥50% diameter stenosis in the three coronary arteries or a major side branch on visual estimation of angiographic images). Patients with left main stem stenosis were excluded from the study, as were those with recent STEMI (ST-elevation myocardial infarction), cardiogenic shock, and those with poor left ventricular ejection fraction (<30%). The intervention arm included patients undergoing PCI. It was mandatory to undertake FFR in this arm and according to the protocol, only those stenoses which had FFR ≤ 0.8 were to undergo PCI. The stents to be deployed included zotarolimus-eluting stents (Resolute Integrity or Resolute Onyx, Medtronic). The comparator arm was CABG and while the strategy of revascularization was left at the discretion of individual surgeons, multiple arterial grafting was strongly recommended. FFR in this arm was not mandatory. The outcome of interest was major adverse cardiac or cerebrovascular event (MACCE) rates at 1-year. MACCE was defined as all-cause mortality, myocardial infarction (MI) stroke or repeat revascularization. For the first time CABG was being compared with PCI guided by FFR, and thus was thought to be equipped to establish non-inferiority of PCI against CABG. The study was adequately powered. The initial sample size requirement was 1424 based on an inferiority margin of 1.45. The sample size requirement reduced to 1290 when the inferiority margin increased to 1.65. However the trial steering committee chose to recruit the initially planned sample of 1500 participants. All events were assessed by an independent clinical events committee who were blinded to the treatment allocation.

In the CABG arm, Left Internal Thoracic Artery (LITA) was used in 97% and multiple arterial grafts were used in 24.5% cases. Off pump coronary artery bypass grafting was the mode of myocardial revascularization in 24% of the patients and in 10% patients FFR was used prior to CABG. In contrast, FFR was measured in 82% of the lesions in the PCI arm with a mean FFR of 0.70. A FFR > 0.80 was present only in 24% of the cases in patients undergoing PCI. The distribution of SYNTAX scores was similar among PCI and CABG. Nearly one-third of the patients undergoing both PCI and CABG had low SYNTAX scores (0–22) and similar (PCI 18%, CABG 17.2%) number of patients had a high SYNTAX score (>32). A mean of 3.7 stents were used in the PCI arm and the mean number of distal anastomoses performed in the CABG group was 3.4. 48(3.2%) patients crossed over from one treatment arm to the other.

The study confirmed that compared with FFR-guided PCI, CABG was associated with a significantly lower incidence of the composite of mortality, stroke, MI and repeat revascularizations at 1 year. Even though the authors have reported a p value of 0.35 against the primary outcome, this p value was for the test of inferiority. The p value comparing the event rates, which is not mentioned, would be statistically significant as evidenced by the confidence interval (CI) of the Hazard Ratio {(HR) of 1.1–2.2. There was no difference between PCI and CABG with respect to the individual components of death (1.6% vs 0.9%), MI (5,2% vs. 3.5%), stroke (0.9% vs 1.1%) and need for re-vascularization (5.9% vs. 3.9%). However, the study was not powered to detect the differences in the individual components. The safety endpoints showed that CABG was associated with a significantly higher incidence of acute kidney injury (0.9% vs 0.1%), bleeding (3.8% vs. 1.6%), atrial fibrillation (14.1% vs. 2.4%) and re-hospitalizations within 30 days (10.2% vs. 5.5%). The study provides two very important findings. By comparing with the number of stents implanted in the Synergy between PCI with Taxus and Cardiac Surgery (SYNTAX) study, the study very elegantly showed that FFR-guided PCI leads to fewer implantation of stents in patients of similar complexity. The study also provided conclusive evidence that PCI did not meet the non-inferiority criteria compared with CABG.

The main issues of the study would be perhaps the definition of MI and the inherent weakness associated with a non-inferiority design. Procedural MI was defined along the Third and Fourth Universal definition of MI. One of the criteria included in both these definitions for patients undergoing PCI is a 5 times increase in Troponin-I along with presence of symptoms suggestive of myocardial ischemia or new ischaemic electrocardiographic (ECG) changes. However, the authors acknowledge that these were not recorded in all cases after PCI. They also concede that had it been recorded, the number of MI in the PCI arm would have been higher.

Interpretation of non-inferiorit

To understand the concept of non-inferiority we have to begin by looking at the delta value (Δ) (Fig. 1). The Δ value represents the inferiority margin. Thus, for PCI to be considered non-inferior, the CI of the HR should not have crossed the inferiority margin. As seen in Fig. 1, the CI (1.1–2.2) crosses the Δ value (inferiority margin) of 1.65. Based on this, the authors have concluded that “PCI was not found to be noninferior to CABG”. The term “not found to be non-inferior” needs further qualification. The natural language interpretation of “not non-inferior” would be to consider it to be “inferior”. However, statistically this is not the case (Fig. 1) and “not non-inferior” and “inferior” are statistically different concepts.

Fig. 1.

The point estimate of the study Hazard ratio is 1.5. The confidence interval ranges from 1.1 to 2.2. It crosses the delta point(Δ) of 1.65 and hence is “not non-inferior”. To be non-inferior the CI should not be crossing the Δ. To be inferior the CI should be completely on the higher side of the Δ

Even though the FAME III compared the safety of PCI and CABG along with the non-inferiority, there are several concerns with any inferiority trial. The first and foremost being how the inferiority margin is decided. The authors of the FAME-3 trial have explicitly and transparently described both the initial inferiority margin of 1.45 as well as the changed inferiority margin of 1.65. The second concern with all inferiority trials is the philosophy behind it. Non-inferiority trials have been considered to be unethical by some as they disregard the interest of patients [4], as in essence, non-inferiority trials test if the proposed treatment lies within the acceptable limit of inferiority. Thus, the ethical question in any non-inferiority trial is whether the patients are adequately informed, that they may be randomized to an intervention whose degree of inferiority is being tested. The research question during consenting the patient in the trial of FAME-3 could thus be rephrased as – the study was being done to check how much worse is FFR-guided-PCI compared to CABG? Lastly, non-inferiority trials should provide both intention-to-treat as well as per-protocol analyses, but FAME-3 surprisingly provides only intention-to-treat analyses [5].

A huge evidence base already exists [6–8] that confirms better outcomes of CABG in comparison with PCI, and FAME III, which was certainly a high-quality, well conducted study, once again echoes the same sentiment. The evidence base continues to build and the study, to quote Pink Floyd, is truly “another brick in the wall” of evidence supporting CABG.

Funding

None.

Declarations

Ethics approval

Not applicable being a review article.

Informed consent

Not applicable being a review article.

Conflict of interest

The author declares that there is no conflict of interest.