There has been extensive discussion and debate in terms of the outcome of the recent landmark coronary revascularisation trials and their implications for the future management of chronic stable angina (CSA) and prognostically significant coronary artery disease (CAD). While the Clinical Outcomes Utilizing Revascularisation and Aggressive Drug Evaluation (COURAGE) trial questioned the added benefit of percutaneous coronary intervention (PCI) over medical therapy in CSA, the PCI versus Coronary Artery Bypass Grafting for Severe CAD (SYNTAX) trial reported increased major adverse cardiac or cerebrovascular events (MACCEs) in patients treated with PCI versus coronary artery bypass grafting (CABG) for left main (LM) and three-vessel disease (3VD).1,2 In the drug-eluting stent (DES) era, we are now presented with the challenge of how to further improve PCI outcomes.
The revascularisation strategies in COURAGE and SYNTAX were guided by non-invasive assessments of ischaemia and/or angiography. Non-invasive assessment of ischaemia is unreliable, particularly in differentiating the severity of individual stenoses in multivessel disease (MVD).3 We also know from pressure-derived fractional flow reserve (FFR) studies that angiography is an unreliable method for assessing the functional severity of stenoses.4 These trials need to be re-assessed taking into account the results of the Fractional Flow Reserve versus Angiography for Multivessel Evaluation (FAME) study, which demonstrated that PCI guided by the functional assessment of lesions using FFR rather than angiography alone reduces major adverse cardiac events (MACE); this has led us to conclude that a functional approach to revascularisation may be the key to improving PCI outcomes.
COURAGE
The hypothesis of the COURAGE trial was that long-term clinical outcome in CSA would be superior in patients treated with PCI and optimal medical therapy (OMT) compared with those treated with OMT alone.1 This was a US/Canadian multicentre, randomised trial of 2,287 patients using bare-metal stents (BMS) with a median follow-up of 4.6 years. Inclusion required CSA with ≥80% stenosis in at least one major vessel or ≥70% stenosis with objective evidence of ischaemia (abnormal resting electrocardiogram [ECG] or positive non-invasive stress test). This was a low-risk population, excluding patients with unstable angina, markedly positive stress tests, LM disease, left ventricular ejection fraction (LVEF) <30% or PCI or CABG in the preceding six months.
There was no significant difference in the primary end-point of death and myocardial infarction (MI) (19 versus 18.5%) or in any of the secondary end-points (see Figure 1). Useful insights followed in a nuclear substudy that measured ischaemic burden pre- and post-therapy.5 This demonstrated that the greater the reduction in the magnitude of ischaemia, the greater the reduction in the risk of death or MI regardless of treatment strategy. A greater reduction in ischaemia was achieved by adding PCI to OMT than with OMT alone, suggesting that a potential benefit from PCI may have been masked by its non-targeted use in a low-risk population.
SYNTAX
The SYNTAX trial compared PCI with CABG for treating patients with LM or 3VD.2 This was a US/European multicentre, randomised study of 1,800 patients. A multidisciplinary team agreed that revascularisation with either PCI or CABG was appropriate prior to randomisation, and a SYNTAX score of disease complexity was calculated. PCI was performed with DES with the aim of complete revascularisation of all vessels ≥1.5mm with ≥50% stenosis.
The primary end-point of MACCE at one year was significantly higher in the PCI versus CABG group (17.8 versus 12.4%; p=0.002), largely driven by repeat revascularisation (13.5 versus 5.9%) (see Figure 2). The rate of death and MI was similar in the two groups, while stroke was less likely in the PCI group (2.2 versus 0.6%; p=0.003). While patients with low or intermediate SYNTAX scores had similar MACCE rates with PCI or CABG, those with high scores had a significantly higher event rate with PCI. In the PCI group, mean stent length was 86.1±47.9mm (range 8–324mm) and 33.2% of patients received >100mm of stent.
Strictly speaking from a statistical point of view, the failure to demonstrate non-inferiority of the primary end-point means that subgroup analysis can be hypothesis-generating only, but it nevertheless provides some important insights. In the subgroups of isolated LM and LM plus one-vessel disease (1VD), MACCE rates were lower with PCI compared with CABG. In the subgroups of LM plus two-vessel disease (2VD)/3VD and in 3VD, MACCE rates were higher with PCI compared with CABG.2 The advantage of CABG over PCI was greatest in those patients with SYNTAX scores in the highest tertile and/or those with diabetes.
SYNTAX clearly tells us that if we are to use PCI in patients with complex and extensive CAD we need to further reduce the need for repeat revascularisation. While most of the predictors of repeat revascularisation are non-modifiable, we can minimise the impact of procedure-related predictors by performing PCI only on lesions where it is clearly indicated and by attempting to minimise the length of treated segments.6
FAME
The FAME study hypothesised that FFR-guided PCI in MVD would be superior to angiography-guided PCI.7 This was a US/European randomised, multicentre study in 1,005 patients with >50% stenosis in two major vessels using DES. In the FFR-guided group, lesions were treated if the FFR was ≤0.80. Patients with LM disease, previous CABG and ST-segment elevation MI (STEMI) in the preceding five days were excluded. Importantly, the study population included those with previous MI, previous PCI, acute coronary syndromes, poor left ventricular function, diabetes, proximal left anterior descending (LAD) disease and chronic total occlusions.
There was a 5.3% reduction in MACE at one year in the FFR-guided group compared with the angiography-guided group (18.3 versus 13.2%; p=0.02) (see Figure 3). Interestingly, the reduction in MACE was not all driven by repeat revascularisation, with a significant reduction in the non-pre-specified composite end-point of death or MI. In 89.6% of patients, at least one stenosis had an FFR of ≤0.80, and 63% of lesions assessed had an FFR of ≤0.80. Even using quantitative coronary assessment (QCA), the unreliability of angiography in determining the functional severity of lesions was confirmed. As expected, in a randomised trial the mean number of lesions scheduled for treatment in each group was similar (2.7±0.9 versus 2.8±1.0; p=NS). However, FFR guidance resulted in fewer stents being required per patient compared with angiographic guidance only (1.9±1.3 versus 2.7±1.2; p<0.001), with a reduced mean stent length (37.9±27.8 versus 51.9±24.6; p<0.001). There was no difference in procedure time, with significantly less contrast and fewer costs in the FFR-guided group.
COURAGE, SYNTAX and FAME
So, how does FAME compare with COURAGE and SYNTAX, and how might the use of FFR in these two trials have influenced the PCI outcomes? To allow a comparison between the trials we will discuss only the 3VD subgroup from SYNTAX as LM disease was excluded from COURAGE and FAME (see Table 1). While the primary end-points in the three trials were different, further comparisons have been made by estimating an identical end-point definition of MACCE.8
An estimated 12-month MACCE rate of 21% in the PCI group in COURAGE is similar to 19.1% in the PCI arm of SYNTAX 3VD and 18.3% in the PCI–angiography arm of FAME. In COURAGE, the proportion of patients free from angina at one year was non-significantly higher in the PCI group compared with the medically treated group (58 versus 50%). In FAME, an additional 4% of patients in the FFR group were free of angina compared with the angiography group (82 versus 78%), suggesting that the use of functional assessment in COURAGE might have resulted in a larger reduction in angina in the PCI group compared with the medically treated group.
The MACCE rate in the PCI arm of the 3VD subgroup in SYNTAX was similar to the MACE rate in the angiography arm of FAME (19.1 versus 18.3%) (see Figure 4). If we compare the CABG arm of the 3VD subgroup in SYNTAX with the FFR arm of FAME, we see a marked reduction in the difference in MACCE rates (11.2 versus 13.2%) between the groups (see Figure 4). The gap between the groups essentially disappears if we apply an identical definition of MACCE, i.e. including stroke but excluding MIs with creatine kinase (CK)-MB three to five times the upper limit of normal (see Figure 5).
While comparing patient groups across different studies is not scientifically robust, it is useful in generating hypotheses. It is difficult to resist the extrapolation that if PCI in COURAGE and SYNTAX had been FFR-guided, it might have shown added benefit over OMT in CSA and achieved non-inferiority to CABG in the management of LM and 3VD.
Discussion
The next step in optimising the efficacy and safety of PCI is to move towards a more focused and functional approach, only treating physiologically significant stenoses. The goal should be ‘functionally complete revascularisation’ rather than angiographically complete revascularisation.7
The COURAGE nuclear substudy re-emphasised the lesson from previous myocardial perfusion studies that reducing myocardial ischaemia optimises prognosis.5 In addition, the PCI of Functionally Non-significant Stenosis (DEFER) study clearly demonstrated that stenting non-ischaemic stenoses (with a lesion event rate of approximately 1% per year) has no benefit over medical therapy, while stenting ischaemia-related stenoses (with a lesion event rate of approximately 5% per year) improves symptoms and outcome.4 With an estimated stent thrombosis risk of approximately 3% per year in SYNTAX, this clearly illustrates why stenting non-ischaemic stenoses provides no benefit and very likely causes harm. Therefore, the current widely used approach of performing PCI on all angiographically significant stenoses results in offsetting the benefits of appropriate stenting of ischaemic lesions with the unnecessary stent related risk of treating non-ischaemic lesions.
The higher MACCE rate with PCI in SYNTAX was largely driven by repeat revascularisation. Total stent length is a predictor of repeat revascularisation and one-third of patients in SYNTAX received more than 100mm of stent. This highlights the need for a more focused approach to stenting.6 In this context, an additional benefit of pressure wire guidance is that it offers the option of performing an FFR pullback in diffusely diseased vessels to identify local segments with significant pressure gradients, which can be treated by focal stent implantation, thus avoiding the need for extensive stenting. In some cases, a functional approach to revascularisation may lead to no PCI being performed or to a decrease in the number of lesions treated, while in other cases it will lead to performing PCI on angiographically under-estimated stenoses. As demonstrated in FAME, FFR-guided PCI led to a reduction in the mean number of stents and stent length per patient.7 Both of these parameters were predictors of repeat revascularisation in SYNTAX, so this more focused use of stents should lead to enhanced efficacy and safety after PCI.
FFR can be used to further optimise stent deployment. The FFR Post Stent Registry found that the FFR post-stenting was a strong independent predictor of six-month MACEs.9 With a post-stent FFR of >0.95, the event rate was 4.9%; with an FFR of 0.90–0.95, the event rate was 6.2%; with an FFR of <0.90, the event rate was 20.3%; and with an FFR <0.80, the event rate was 29.5%. The detection of a persistent pressure gradient across the stented segment may highlight under- or mal-deployment of the stent or the presence of persisting significant plaque without the stented segment, indicating the need for further optimisation.
Finally, we should consider whether functional revascularisation would improve outcomes after CABG. We know that bypass grafts on non-functionally significant lesions are more likely to be occluded at one year than bypass grafts on functionally significantly lesions (21.4 versus 8.9%).10 Despite this, there was no difference in angina class or repeat revascularisation between patients with patent or occluded grafts. This observation was also reported recently in the SYNTAX Le Man LM angiographic follow-up substudy. Sixteen per cent of bypass grafts for LM disease were occluded or >50% stenosed at 15 months, but there was no difference in the 9% MACCE rate between those with and without graft problems. This reflects the fact that placing a graft on a vessel with no functionally significant stenosis has minimal potential for an adverse outcome if that graft subsequently occludes. The native vessel almost always remains open and, of course, the patient did not need the graft in the first place. Placing a stent in a vessel with no functionally significant stenosis is an entirely different matter. If that stent occludes, the patient will almost certainly sustain an MI having had no potential for benefit from the index procedure anyway. This is perhaps one of the strongest arguments for FFR guidance during multivessel PCI.
Conclusion
We believe that aiming for functional as opposed to anatomical revascularisation will result in added benefit from PCI in CSA, and that the significant improvement in PCI outcomes seen with FFR guidance will result in FFR-guided PCI achieving non-inferiority to CABG.