a conservative strategy straight after moderate/high-risk stress testing, without any coronary imaging to rule out left main disease, yet they had a similar death/MI outcome vs invasive strategy. Invasive strategy increased stroke and requirement for dialysis in ISCHEMIA-CKD trial and appeared particularly harmful in stage 4 CKD, wherein it also increased death.76 Despite their very high risk, ISCHEMIA-CKD patients did not benefit from revascularization (cardiac mortality ~27% at 2.2 years).
All the above 4 trials (COURAGE, FAME 2, BARI 2D and ISCHEMIA) included patients with stable symptoms that are reasonably controlled with medical therapy, not ACS, not daily angina (~excluded in ISCHEMIA), and not angina on short walks. In those patients, revascularization does not improve death or MI even if ischemia is severe or CAD is multivessel. Severe, refractory symptoms and ACS remain appropriate indications for coronary angiography and revascularization. PCI is superior for angina relief and is particularly helpful in patients with severe functional limitation, and in patients with angina and a combination of low heart rate/low blood pressure, whose myocardial demands are inherently low (low rate-pressure product) and whose angina is mainly driven by the severe stenosis rather than hemodynamic variables.73,77
Furthermore, it is unclear if PCI improves survival in patients with left main or extensive multivessel disease who would derive a survival benefit from CABG, but who are not candidates for CABG or whose SYNTAX score is ≤22. By extrapolating the results of CABG vs. PCI trials, PCI is presumed to improve survival in those patients (PCI and CABG are associated with equivalent survival in select patients with multivessel or left main CAD); this has not been directly proven.
IX. PCI vs. CABG in multivessel and left main disease
CABG outcome is less dependent on the complexity and diffuseness of lesions than PCI, and when a graft is placed, the whole proximal 6–8 cm of the coronary artery is protected from MI induced by plaque rupture (Table 3.3). Keep in mind that plaque ruptures and acute coronary occlusions occur overwhelmingly in the proximal third of the coronary arteries, a segment protected by CABG (80-90% of acute LAD and LCX plaque ruptures are in the proximal 5 cm).78 Hence, CABG is consistently associated with a lower risk of MI and angina recurrence than PCI. Even when MI occurs after CABG, it is less likely to be fatal and more likely a small MI, as compared with patients receiving medical therapy or PCI (CASS registry, BARI trial, and SYNTAX trial).79
Table 3.3 Reasons for superiority of CABG vs. PCI.
| CABG success is much less affected by the anatomical complexity (e.g., CTO) and the diffuseness of CADWhile PCI treats focal disease, a graft improves flow to the whole coronary territory, including segments with moderate diffuse disease, and protects from MI resulting from occlusion of the proximal 6–8 cm coronary segment. Plaque ruptures and acute coronary occlusions occur overwhelmingly in the proximal 5 cm of coronary arteriesVery long longevity of LIMA graftMore complete revascularization with CABG (vessels with CTO are sometimes left untreated in a multivessel PCI strategy) |
BARI and ARTS trials – In the balloon angioplasty era, the BARI trial randomized very select patients with focal multivessel CAD to CABG vs. PCI. In comparison with PCI, CABG dramatically reduced mortality in diabetic patients by an absolute 16% at 5 years, and dramatically reduced repeat revascularizations in all patients.80 The benefit on repeat revascularizations was shown in the BMS era as well (ARTS trial).81 The superiority of CABG was seen despite the very careful selection of patients with non-extensive CAD amenable to PCI (<10% of screened patients were randomized to CABG vs. PCI).
Isolated proximal LAD disease – A meta-analysis of randomized trials of CABG vs. PCI for isolated proximal LAD disease suggests the lack of mortality difference, although repeat revascularizations were much lower with CABG (pre-DES era).82
SYNTAX trial – In the DES era, the SYNTAX trial randomized patients with three-vessel and/or left main disease to CABG vs. PCI with DES. This trial included patients with extensive, complex CAD, and graded the angiographic severity of CAD using the SYNTAX score. In the overall trial, at 5 years of follow-up, CABG was associated with a significant reduction in MI (~10% vs. 4%), a marked reduction in the need for repeat revascularizations (26% vs. 14%), but no mortality difference (13.9% vs. 11.4%). CABG significantly reduced mortality by an absolute 8% in the high SYNTAX scores (>32).83,84 Conversely, patients with a low SYNTAX score (≤22) had comparable death, MI and even repeat revascularization rates with CABG and PCI, whether they had three-vessel or left main disease. Regarding 5- and 10-year mortality, CABG and PCI were equipoise for the left main subset (=left main +1, 2 or 3-vessel CAD), but CABG reduced mortality in 3-vessel CAD without left main. Interestingly, CABG outcomes were not affected by SYNTAX score, i.e., CABG success and post-CABG survival were not affected by angiographic complexity, in contradistinction with PCI. The only pitfall of CABG was the higher early risk of stroke (2.2% vs. 0.6% at 1 year).
The SYNTAX score assigns a number for the location of each stenosis (e.g., left main 5, proximal LAD 3.5, proximal LCx 1.5, OM 1, RCA 1), and multiplies this number by 2 in case of a 50–99% stenosis, and 5 in case of a CTO. Additional points are added at every lesion for tri- or bifurcation, long disease, calcium, and CTO complexity. Overall, the score emphasizes proximal stenoses (especially LAD) and angiographic complexity, especially CTO.
FREEDOM trial – Diabetic patients with two- or three-vessel CAD involving the LAD were randomized to CABG vs. PCI with DES.85 At 5 years of follow-up, CABG significantly reduced mortality vs. PCI, almost as much as in the high SYNTAX group of SYNTAX trial (16.3% vs. 10.9%). It reduced MI (~14% vs. 6%) at the price of an increase in postoperative stroke and a higher early postoperative mortality. The benefit in these diabetic patients was consistent across all SYNTAX score groups, including the low SYNTAX group.
EXCEL and NOBLE trials of left main disease- Both EXCEL and NOBLE trials specifically randomized patients with left main disease and mainly low or intermediate SYNTAX score (≤32) to CABG vs. PCI; 81% of patients had distal left main disease, and 15-29% had diabetes. In both trials, CABG was slightly superior to PCI at 5 years of follow-up.86,87 In EXCEL, the 5-year composite death/MI/stroke was not different, but CABG slightly yet significantly reduced mortality (13 vs. 9.9%), nonprocedural MI (6.8 vs. 3.5%), and repeat revascularizations (target and non-target, 16.9 vs 10%). In the CABG group, the composite outcome was increased in the first 30 days but reduced beyond 30 days (survival curves crossed at 1 year). In NOBLE, CABG significantly reduced the primary outcome, which included repeat revascularizations (28.9% vs 19%), and reduced nonprocedural MI, but did not reduce mortality.
For three-vessel CAD- Both SYNTAX and FREEDOM trials support CABG. PCI is an option in patients with a low SYNTAX score ≤22 and no diabetes.
For left main disease- Both EXCEL and NOBLE trials support PCI as an alternative to CABG, including in complex distal left main. In fact, PCI seems a more viable option in left main disease than complex 3-vessel CAD, as per SYNTAX left main analysis. PCI is associated with lower early stroke and periprocedural MI, but higher late MI and revascularizations.
Note some of the features of the SYNTAX, FREEDOM, NOBLE and EXCEL trials:
LVEF was normal or >40% in almost all patients. Very few patients had HF (4% in SYNTAX trial).
Patients with acute MI, including large acute NSTEMI, were excluded. The superiority of CABG is extrapolated to those patients whose event rates are higher than stable CAD.
While the CABG benefit on repeat revascularization emerges early within the first year, the survival and MI benefit only starts to emerge beyond 2 years of follow-up, after an early perioperative hazard (in all trials). This benefit appeared sooner in the high SYNTAX