Management of Left Main Coronary Artery Disease.

Introduction

Left main coronary artery disease (LMCAD) portends higher prognostic risk as a result of the large myocardial territory at risk, ranging from 75% to 100%, depending on the dominance of the left coronary circulation. Diagnosis and management of significant LMCAD continues to be a source of clinical apprehension and uncertainty. LMCAD is not uncommonly found in stable patients undergoing coronary angiography and is often associated with concomitant coronary artery disease (CAD). Current clinical practice guidelines from both the American College of Cardiology/American Heart Association and the European Society of Cardiology recommend revascularization for all patients with ≥50% stenosis of the left main coronary artery (LM), regardless of symptomatic status or associated ischemic burden.1 The anatomic extent and complexity of CAD are major factors in deciding on the best management approach of LMCAD. For example, isolated LMCA lesions involving the ostium or shaft do well with either percutaneous coronary intervention (PCI) or coronary artery bypass graft surgery (CABG). However, distal LM bifurcation lesions or those associated with complex multivessel disease may do better with surgical revascularization. Other key elements of consideration include surgical operative risk, left ventricular function, acuity of clinical presentation, likelihood of achieving complete revascularization, and patients’ informed preference.

Medical Management of LMCAD

The use of guideline‐directed secondary prevention and lifestyle interventions should be implemented and encouraged for all patients with LMCAD just as they are indicated for patients with non‐LM CAD. In the COURAGE (Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation) trial,2 similar outcomes were observed for an initial strategy of optimal medical therapy compared with initial revascularization in patients with stable non‐LM CAD. The safety of deferred revascularization in patients with stable LM disease is less well known, but current clinical practice guidelines strongly recommend revascularization in all patients with ≥50% stenosis of the LM.1 The basis for these class IA recommendations is predicated on post hoc data derived from subgroup analyses of 185 patients with LM disease from 2 randomized clinical trials (RCTs) of patients with chronic stable angina conducted in the late 1970s and early 1980s3, 4 demonstrating the superiority of surgical revascularization over medical therapy on 5‐ to 10‐year survival.5, 6, 7, 8, 9 These early RCTs were conducted in an era when medical therapy was, by contemporary standards, limited. For example, only 66% of “medically managed” patients with LM in those early trials received β‐blockers and only 19% were taking aspirin. These trials antedated the current widespread use of disease‐modifying pharmacologic interventions (such as statins, inhibitors of the renin‐angiotensin‐aldosterone system, and more effective antiplatelet agents such as P2Y12 inhibitors), which reduce adverse cardiovascular events in patients with CAD.2, 10, 11, 12, 13, 14, 15

Patients with LM disease are not a homogenous group. In fact, even antedating our current era of evidence‐based optimal medical therapy, those with 50% to 70% LM stenosis or with preserved left ventricular (LV) function were found to have more favorable survival while receiving medical management alone (66% 3‐year survival) compared with more severe LM disease >70% (41% 3‐year survival) or with reduced LV function.5, 7 Even these older studies from 30 to 40 years ago were able to identify patients with LM disease who were at a relatively lower risk with medical therapy compared with patients with high‐risk features such as >70% LM stenosis, poor LV function, elevated LV end‐diastolic pressure, or prior myocardial infarction (MI).5, 7 Moreover, Conley et al5 reported that the 1‐year survival rate of patients with nonrevascularized left main stenosis ≥70% ranged between 50% and 62% in those with a history of congestive heart failure, resting chest pain, resting ST‐T–wave changes, LV dysfunction, or LV end‐diastolic pressure >15 mm Hg, as compared with a survival rate ranging from 81% to 95% in the absence of these clinical variables.5 However, the more favorable outcomes in these “lower‐risk” groups reflected the abysmal outcomes of more severe LM disease or comorbidities (>50% death rate over 3–4 years).5, 7 The absolute risks in even the lower‐risk categories (20–30% death rate over 3–4 years) makes even the most courageous physician nervous about deferring revascularization based on these clinical and angiographic characteristics alone regardless of significant advances in modern guideline‐based medical therapy. Most contemporary large RCTs assessing clinical outcomes from medical versus revascularization therapy have excluded patients with LMCAD by design. Thus, it remains unproven whether optimal medical therapy may be a safe and appropriate therapeutic alternative to revascularization in certain, select subsets of lower‐risk patients with stable LM CAD. Part of the obstinacy of our guidelines for LM disease is likely related to our poor ability to discriminate significant from nonsignificant LM disease from angiographic and clinical characteristics. Increasingly, newer invasive techniques are now being employed to help guide decision‐making regarding revascularization therapies for LM disease.

Contemporary Evaluation of LMCAD

Most patients with significant LMCAD are symptomatic given the large myocardial territory it subtends. Nonetheless, angiographically significant LMCAD is incidentally found not so infrequently in stable patients undergoing coronary angiography. In the absence of critical LM stenosis or corroborating clinical presentation, the hemodynamic significance of incidental or intermediate LM lesions warrants further consideration. The reliance of our current clinical practice guidelines on angiographic lesion severity as the sole determinant of risk and the de facto threshold for CABG surgery decision‐making is outdated in the current era of clinically proven noninvasive and invasive modalities in determining the functional and prognostic significance of angiographically challenging coronary lesions. In particular, visual estimates of intermediate LM stenoses (30–70%) from coronary angiography have significant interobserver variability.16 Conversely, the reproducibility and accuracy of the angiographic assessment of LM stenosis ≥70% is excellent.16, 17 Thus, revascularization strategies based solely on the angiographic appearance of LMCAD stenosis of intermediate severity (between 50% and 70%) may be inappropriate. Aside from the unwarranted surgical risk, premature CABG for potentially noncritical lesions may ultimately be harmful to patients because of low graft patency rates in this setting and up to a 6‐fold higher rate of accelerated obstruction of bypassed native coronary vessels, which renders subsequent PCI of native vessels technically challenging, if needed, for symptom relief.18

Noninvasively, certain high‐risk features are suggestive of significant LM or equivalent disease, including: (1) Duke treadmill score ≤11, (2) stress‐induced sustained ventricular tachyarrhythmia or nonsustained ventricular tachyarrhythmia >30 seconds or ST‐segment elevation, (3) exercise LV ejection fraction ≤35%, (4) large reversible anterior perfusion defect (≥10% LV involvement on nuclear perfusion or ≥12.5% LV involvement on cardiac MRI) or multiple reversible perfusion defects of moderate size, (5) stress‐induced LV dilation or increased lung uptake in the setting of moderate perfusion defect or large fixed perfusion defect, or (6) echocardiographic wall motion abnormality involving >2 segments developing at a low‐dose dobutamine (≤10 mg/kg per minute) or at a low heart rate (<120 beats per minute).19

Invasively, more detailed evaluation of the anatomic severity and hemodynamic significance of clinically ambiguous LM lesions can be obtained by intracoronary imaging with intravascular ultrasound (IVUS) imaging or physiologically with pressure wire assessment of fractional flow reserve (FFR).

Use of Intracoronary Imaging in the Evaluation of Myocardial Ischemia in LM Lesions

The use of IVUS is particularly helpful in the determination of plaque extent and characteristics within the LM, as well as in determining ostial involvement of daughter branches. IVUS can provide an estimate of the ischemic burden of the LM lesion, and its use following LM PCI improves clinical outcomes.20, 21 In the multicenter prospective LITRO study22 of intermediate LM stenosis between 25% and 60%, deferring revascularization of LM lesions with minimal luminal area (MLA) of ≥6 mm2 (53% of lesions) was safe and associated with favorable outcomes at 2 years of follow‐up (cardiac death‐free survival of 97.7%). The disagreement between angiography and this IVUS criterion for a significant stenosis was substantial. One third of patients with an insignificant angiographic stenosis of <30% had an MLA of <6 mm2, whereas 43% of patients with angiographic LM stenosis ≥50% had a prognostically favorable MLA of ≥6 mm.22 Moreover, a cutoff value of 6.0 mm2 also agrees with a theoretical value derived from fractal geometry. A study confirmed that the linear law was more exact than the Murray's law that largely underestimated the calculated mother‐vessel diameter. Using the currently established 3.0 mm2 as the best cutoff MLA for the LM branches, the calculated LM‐MLA cutoff by linear law was 5.8 mm.23

Although IVUS is good at describing the anatomical extent of disease, the LM‐MLA cutoff for determining a functionally significant lesion differs between populations and thus these values need to be interpreted with caution. In a Korean study, the MLA cutoff for FFR <0.80 was 4.5 mm2 with a sensitivity of 77% and a negative predictive value of 75%,24 whereas another US study yielded a cutoff value of 5.9 mm2 with a sensitivity and specificity of 93% and 94%, respectively, for an FFR <0.75.25 The most plausible explanation for this discrepancy appears to relate to differences in the reference size of the coronary arteries. For example, the average LM‐MLA in patients in the Korean study was 4.8 mm2 versus 7.6 mm2 in the US study. Another study comparing LM lesions between 99 white North American and 99 Asian patients showed that Asian patients had a significantly smaller LM‐MLA (5.2±1.8 mm2 versus 6.2±1.4 mm2, respectively; P<0.0001).26

The other value of IVUS is to ensure stent optimization of LM PCI. IVUS can ensure adequate expansion and apposition of stents after LM PCI, which improves clinical outcomes following LM PCI, particularly in patients with distal LM lesions and those treated with a 2‐stent strategy.20

Optical coherence tomography is another intracoronary imaging modality that is often employed for lesion characterization and PCI guidance in non‐LMCAD, particularly given its higher resolution imaging. As compared with IVUS, optical coherence tomography is a light‐based technology that requires a contrast flush to clear the blood column, which makes imaging of aortic ostial lesions in the LM challenging. Another drawback of optical coherence tomography use in LM imaging is its limited penetration depth (2–3 mm) as compared with IVUS (4–8 mm), given that the average LMCA diameter is 3.5 to 4.5 mm. Finally, literature regarding clinical outcomes or correlation with physiology for optical coherence tomography in LMCAD is lacking.

Use of FFR in the Evaluation of Myocardial Ischemia in LM Lesions

Whereas IVUS is a better tool for determining the anatomical extent of disease, FFR is a better tool for assessing the hemodynamically significance of an LM stenosis. A poor correlation was noted between quantitative coronary angiography and FFR, which further highlights the shortcomings of reliance on angiography alone in evaluating LM lesions.17 Similar to non‐LM lesions, angiographically intermediate LM lesions with an FFR of ≥0.80 can have revascularization deferred with favorable long‐term outcomes.17 The use of instantaneous wave‐free ratio (iFR) is well established in non‐LM lesions based on the DEFINE‐FLAIR (Functional Lesion Assessment of Intermediate Stenosis to Guide Revascularisation)27 and iFR‐SWEDEHEART (Instantaneous Wave‐free Ratio versus Fractional Flow Reserve in Patients With Stable Angina Pectoris or Acute Coronary Syndrome)28 trials, which both showed similar favorable outcomes by deferring revascularization of lesions with an iFR of >0.89. However, specific outcome studies evaluating iFR in LMCAD are warranted before iFR is liberally adopted as the sole determinant of revascularization in patients with intermediate LM lesions.

Although IVUS and FFR correlate well,29 there are limitations to both techniques that may favor one modality over the other. For example, IVUS area measurements may be limited by distortion caused by minor differences in the rotational speed of the imaging element, a noncoaxial orientation of the catheter, or extensive arterial calcification. On the other hand, FFR may be limited by the frequent presence of significant downstream stenoses, which may underestimate or overestimate the hemodynamic significance of the LM lesion. However, this is of more concern when there is severe disease present in both the left anterior descending and circumflex arteries. Fearon et al30 showed that when only one major branch of the left main has severe disease, downstream disease does not have a clinically relevant impact on the evaluation of intermediate LM stenosis with the pressure wire placed in the nondiseased branch. Based on their findings, if the FFR of the LM is either ≤0.80 or >0.85, then it can be assumed that the LM lesion is hemodynamically significant or insignificant, respectively. However, if the FFR is between 0.81 and 0.85, then the hemodynamic significance of the LM lesion cannot be accurately determined if the combined FFR of the LM and the downstream disease is ≤0.45.30 In such situations, IVUS guidance is preferred.20, 21 Accordingly, in the modern era of contemporary CAD decision‐making, an invasive assessment of stenosis severity is certainly complementary to a purely angiographic assessment and may be superior.31

Revascularization of LMCAD

Existing clinical practice guidelines continue to advocate CABG surgery as the singular class I indication for myocardial revascularization. However, more recent RCTs and registry studies in LM CAD support PCI as a reasonable alternative in select patients with less complex LM anatomy. Currently, in the US guidelines, PCI has a class IIa recommendation (“is reasonable”) in select patients with isolated LM stenosis involving the ostium or shaft and without coexisting multivessel disease and the risk of surgical bypass is increased.1 PCI has a class IIb recommendation (“may be reasonable”) in patients with LM stenosis involving the distal bifurcation or with less complex coexisting multivessel disease as defined by a low or intermediate SYNTAX score (≤33) and who have an elevated surgical risk. The current US guidelines recommend against PCI in patients who are good candidates for surgical bypass with coexisting complex multivessel disease as defined by highest tertile of the SYNTAX score (≥33).1

The guidelines are based primarily on the hypothesis‐generating findings of the prespecified and powered subgroup of patients with LMCAD in the SYNTAX (Synergy Between Percutaneous Coronary Intervention With TAXUS and Cardiac Surgery) trial,32, 33, 34 with consideration of other smaller randomized trials that were underpowered to provide a conclusive answer on the optimal revascularization strategy including the LE MANS (Left Main Coronary Artery Stenting) trial (n=100; bare‐metal stents [BMS])35, 36, 37 and PRECOMBAT (Bypass Surgery Versus Angioplasty Using Sirolimus‐Eluting Stent in Patients With Left Main Coronary Artery Disease) trial (n=600; sirolimus‐eluting stents).38, 39 The absence of conclusive evidence motivated the conduct of 2 large RCTs comparing CABG and PCI for LM revascularization (EXCEL [Evaluation of XIENCE versus Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization] trial40 and NOBLE [Nordic‐Baltic‐British Left Main Revascularization Study]40, 41), both of which have been recently published but not reflected in the guidelines as of yet. The Table summarizes key elements of the main RCTs comparing CABG and PCI for revascularization of LMCAD. Most have demonstrated similar intermediate and long‐term outcomes of mortality or MI, albeit with consistently higher rates of repeat revascularization with PCI. Some demonstrated improved long‐term survival with CABG in a select group of higher‐risk patients attributable to complex coronary anatomy33, 34 or patients with diabetes mellitus and an LM equivalent of 3‐vessel CAD.42 In fact, no dedicated RCT has compared outcomes of PCI versus CABG for LMCAD specifically in patients with diabetes mellitus. In subgroup analyses of the LM RCTs, patients with diabetes mellitus had similar outcomes with PCI compared with CABG, although none of the studies were powered to show a difference. The prespecified subgroup analysis of patients with diabetes mellitus in the SYNTAX trial (including LM and 3VD) showed significantly higher rates of repeat revascularization with PCI but no difference in the composite of all‐cause death, stroke, or MI.43 Patients with LV dysfunction are another high‐risk group for which data are lacking with respect to the optimal mode of revascularization, as they have been excluded from most large randomized trials of LM or equivalent CAD. Nonetheless, evidence from older trials demonstrated a clear survival advantage of revascularization over medical management in patients with LM with LV dysfunction.5

Table 1

Left Main Randomized Clinical Trials of PCI vs CABG in Perspective

	LE MANS35, 37	Boudriot44	PRECOMBAT38, 39	SYNTAX32, 33, 34	EXCEL40	NOBLE41
	Poland	Germany	North Korea	Europe/United States	United States/Europe	Northern Europe
Diabetes mellitus prevalence, %
PCI	19	40	34	26	30	15
CABG	17	33	30	24	28	15
LVEF, %
PCI	54 (SD=117	65 (55–70)	62 (SD=8)	N/R	57 (SD=10)	60 (55–65)
CABG	54 (SD=7)	65 (55–68)	61 (SD=9)		57 (SD=9)	60 (52–64)
Sample size, No.	105	201	600	705	1905	1201
LM disease severity	≥50% Angiography only	≥50% Angiography only	≥50% Angiography only	≥50% Angiography only	≥70% Angiographically, or 50% to 69% and hemodynamically significant	≥50% angiographically, or FFR ≤0.80
Distal LM stenosis, %
PCI	56	74	67	58	82	81
CABG	60	69	62	64	79	81
LM disease+3VD, %
PCI	60	11	40.7	35.1	17.2	N/A
CABG	75	17	41.0	38.1	19.4	N/A
Average SYNTAX
PCI	25.2+8.7	24.0 (19–29)	24.4	30.2+12.7	26.9+8.8	22.5+7.5
CABG	24.7+6.8	23.0 (14.8–28)	25.8	29.6+13.5	26.0+9.8	22.4+8.0
Composite end point	Death/MI/CVA/TVR	Death/MI/revascularization	Death/MI/CVA/TVR	Death/MI/CVA/TVR	Death/MI/CVA	Death/MI/CVA/TVR
Follow‐up (longest)	10 y	1 y	5 y	5 y	3 y	3 y
Composite outcome	Noninferior (52.2% vs 62.5%)	Inferior (19% vs 13.9)	Noninferior (17.5% vs 14.3%)	Noninferior (36.9% vs 31%)	Noninferior (15.4% vs 14.7%)	Inferior (28% vs 18%)
Death	Noninferior (21.6% vs 30.2%)	Noninferior (2% vs 5%)	Noninferior (5.7% vs 7.9%)	Noninferior (12.8% vs 14.6%)	Noninferior (8.2% vs 5.9%)	Noninferior (11% vs 9%)
Stroke	Noninferior (4.3% vs 6.3%)	N/R	Noninferior (0.7% vs 0.7%)	Superior (1.5% vs 4.3%)	Noninferior (2.9% vs 2.3%)	Noninferior (5% vs 2%)
MI	Noninferior (8.7% vs 10.4%)	Noninferior (3% vs 3%)	Noninferior (2% vs 1.7%)	Noninferior (8.2% vs 4.8%)	Noninferior (8.3% vs 8.0%)	Inferior (6% vs 2%)
Revascularization	Noninferior (26.1% vs 31.3%)	Inferior (14% vs 5.9%)	Inferior (13% vs 7.3%)	Inferior (26.7% vs 15.5%)	Inferior (12.9% vs 7.6%)	Inferior (15% vs 10%)
IVUS	Recommended	Infrequent	At discretion 91%	Infrequent	Recommended 77%	Recommended 74%
FFR guidance	N/R	N/R	N/R	Infrequent	Recommended 9.0%	Recommended
Stent for PCI group	BMS DES only if ≤3.8	SES	SES	PES	EES	BES First‐generation DES (7.7%)
LIMA use in CABG group, %	72	99	94	N/R	99	96
Off pump, %	1.9	46	64	N/R	29	16

3VD indicates 3‐vessel coronary artery disease; BES, biolimus‐eluting stent; BMS, bare‐metal stent; CABG, coronary artery bypass graft surgery; CVA, cerebral vascular accident; DES, drug‐eluting stent; EES, everolimus‐eluting stent; FFR, fractional flow reserve; IVUS, intravascular ultrasound; LIMA, left internal mammary artery; LM, left main coronary artery; LVEF, left ventricular ejection fraction; MI, myocardial infarction; N/A, not available; N/R, not required; PCI, percutaneous coronary intervention; PES, paclitaxel‐eluting stent; SD, standard deviation; SES, sirolimus‐eluting stent; TVR, target vessel revascularization. Please refer to text for complete trial names.

Hybrid bypass is another revascularization approach that combines coronary bypass using a minimally invasive direct coronary artery bypass approach of grafting the LIMA to LAD artery and PCI to the remaining vessels in an attempt to achieve the most desired aspects of each revascularization strategy.45 While evidence in support of this hybrid approach is limited, ongoing large randomized trials such as the National Institutes of Health–funded HYBRID (Hybrid Coronary Revascularization) trial (ClinicalTrials.gov Identifier: NCT03089398) should shed further light on the theoretical benefit of this revascularization strategy.

The Heart Team Approach

Regardless of which method of revascularization is used, current guidelines stress the importance of a “heart team” approach to management of complex coronary disease including left main disease. The heart team weighs the risks and benefits of PCI, surgery, or medical treatment alone, taking into account the patient's informed preference. Such teams typically include an interventional cardiologist, a cardiac surgeon, and a noninvasive physician to determine the appropriate management plan. The complexity and extent of coexisting CAD with the intention of achieving complete revascularization should be adequately considered in any discussions by the heart team and the patient. Major adverse cardiovascular events including mortality are higher in patients with incomplete revascularization than those with complete revascularization regardless of the revascularization strategy.46 This may explain the findings of SYNTAX demonstrating improved outcomes with CABG in patients with more complex coronary anatomy, since complete revascularization was achieved more frequently with CABG than with PCI in SYNTAX and other studies.34

Several risk models are used in clinical decision‐making to predict the relative impact of specific risk factors on outcomes with cardiac surgery. While none of these risk models are inclusive of all potential risk characteristics, they do offer general estimates of associated operative morbidity and mortality that help the heart team in selecting the best management option for a specific patient and to counsel patients accordingly. The most widely used surgical risk score in the United States is the Society of Thoracic Surgeons score.47 It classifies operative risk based on predicted risk of mortality into low (<4%), intermediate (4% to <8%), high (8% to <12%), or extreme (≥12%). The heart team is particularly imperative in weighing the risks and benefits of surgery in the high‐ and extreme‐risk population. Additional clinical factors that are not included in most risk models also need to be considered by the heart team in making management recommendations including frailty metrics, cognitive status, surgical recovery and social support, quality of life, life expectancy, patient's preference and religious beliefs, and any potential concerns regarding tolerance or adherence with long‐term dual antiplatelet therapy. An algorithm for managing patients with LM disease by the heart team is suggested in Figure 1.

Figure 1

Proposed algorithm for heart team management of left main coronary artery (LM) disease. CABG indicates coronary artery bypass graft surgery; CTO, chronic total occlusion; DAPT, dual antiplatelet therapy; FFR, fractional flow reserve; IVUS, intravascular ultrasound; LAD, left anterior descending coronary artery; OMT, optimal medical therapy; PCI, percutaneous coronary intervention; STS, Society of Thoracic Surgeons.

PCI Technical Considerations

The heart team may also discuss procedural aspects of PCI and its likelihood of achieving complete revascularization if this is recommended. Even though not supported by conclusive data, the use of hemodynamic support may offer a safety net that allows for safe and complete revascularization in patients at highest risk. Most experts agree that hemodynamic support should be strongly considered in patients with reduced ejection fraction, decompensated hemodynamics (elevated end‐diastolic LV pressure >20 mm Hg, systolic blood pressure <100 mm Hg, or mixed venous oxygen saturation <55%), or expected prolonged ischemic time caused by complexity of disease and/or need for atherectomy.48

Drug‐eluting stents are preferred for LM revascularization as they offer improved survival and fewer adverse cardiovascular events compared with BMS, particularly when guided by IVUS imaging.21, 36 Although the majority of LMCAD involves the distal bifurcation, randomized trials specifically addressing the optimal PCI revascularization technique in these lesions are scarce.49 Generally, provisional stenting has been advocated as the preferred approach in bifurcation lesions, as it is technically simpler with at least similar outcomes to a systematic 2‐stent strategy.50, 51, 52 However, this was challenged recently by the randomized DKCRUSH‐II (Randomized Study on Double Kissing Crush Technique Versus Provisional Stenting Technique for Coronary Artery Bifurcation Lesions),53 which included ≈17% LM lesions, showing that a 2‐stent strategy using the double kissing (DK) crush technique (Figure 2) is superior to provisional stenting particularly in more complex lesions. More recently, a dedicated randomized trial comparing DK crush technique with provisional stenting for left main distal bifurcation lesions (DKCRUSH‐V) demonstrated lower rates of target lesion failure with DK crush technique at 1 year (5% versus 10.7%, P=0.02) and stent thrombosis (0.4% versus 3.3%, P=0.02).54 In DKCRUSH‐III, the same investigators showed the superiority of the DK crush technique over culotte at 3 years with significantly lower rates of major adverse cardiac events (8.2% versus 23.7%, P<0.001) and stent thrombosis (0% versus 3.7%, P=0.007).55 Notably, in the recent NOBLE trial, 87.7% of LM PCI involved the distal bifurcation, of which 63.3% underwent a provisional strategy, 23.9% had culotte stenting, and only 4% underwent LM stenting using the crush technique. Whether a dedicated RCT of PCI versus CABG using the DK crush technique preferentially for distal LM bifurcation stenting would demonstrate superior outcomes with PCI is thought‐provoking.

Figure 2

Step‐by‐step illustration of the double kissing crush 2‐stent bifurcation technique.

Regardless of the revascularization strategy, emphasis should be placed on employing techniques shown to improve outcomes with either strategy, along with adherence to optimal medical management and lifestyle habits. Noteworthy, neither off‐pump CABG56 nor the use of bilateral internal mammary arteries57 has been shown to improve CABG outcomes in RCTs.

RCTs of LM Revascularization

The Table summarizes the main RCTs of CABG versus PCI for LM revascularization (detailed description in Data S1). Patients had a low prevalence of diabetes mellitus compared with other PCI studies (15–30% of patients), were defined by ≥50% LM stenoses, had generally preserved left ventricular ejection fraction, and 60% to 80% had distal left main lesions. Important differences include the use of IVUS to optimize stent deployment in the PCI group in the more recent EXCEL and NOBLE trials. Overall, they all demonstrated noninferiority of PCI as compared with CABG with respect to survival, particularly in patients with low to intermediate anatomic complexity. However, increased need for repeat revascularization after PCI continues to be the main advantage of CABG over PCI. Their findings are mostly consistent despite their heterogeneity including sample size, stent types, and utilization of adjunct imaging and physiologic guidance. Although there are 6 RCTs of PCI versus CABG, only 4 used drug‐eluting stents and had more than 1 year of follow‐up. In the LE MANS study,35, 36, 37 65% of the PCI group were treated with BMS, while in the study by Boudriot et al,44 data were only presented for up to a year after randomization. Both studies were underpowered for most outcomes, with 105 patients in the LE MANS study and 201 patients in the Boudroit study. The 4 larger trials using drug‐eluting stents in the PCI arm and predominantly LIMA grafts in the CABG arm were the SYNTAX, PRECOMBAT, EXCEL, and NOBLE trials. The report from the SYNTAX study is a subgroup analysis of the SYNTAX trial in patients with LM disease.33, 34

Figure 3 shows pooled odds ratios of the 4 main RCTs employing drug‐eluting stents for the end points of major adverse cardiac or cardiovascular risk, all‐cause mortality, MI, stroke, and total repeat revascularization. Overall, the studies show an increased risk of major adverse cardiac or cardiovascular end points, driven not by all‐cause mortality but by higher rates of MI and revascularization with PCI. However, the pooled estimate for MI had significant heterogeneity, particularly between EXCEL and NOBLE.

Figure 3

Pooled analyses of randomized controlled trials comparing percutaneous coronary intervention using drug‐eluting stents with coronary artery bypass graft surgery in left main coronary artery revascularization. A, Pooled estimates of major adverse cardiac and cerebrovascular events. B, Pooled estimates of death. C, Pooled estimates of myocardial infarction. D, Pooled estimates of stroke. E, Pooled estimates of all repeat revascularization. CI, confidence interval; RR, relative risk. Please refer to text for complete trial names.

With the recent findings of EXCEL and NOBLE, debates have spurred reconciling the apparently disparate findings of these 2 well‐conducted trials. However, an in‐depth look at their design and findings reveal more consistencies than initially appreciated. Both trials enrolled similar patients but had notable differences that account for the disparate findings. Firstly, the primary end points differed between the 2 studies with repeat revascularization being excluded in EXCEL and periprocedural MI being excluded in NOBLE. Both NOBLE and EXCEL had higher rates of repeat revascularization (Figure 3E) and nonprocedural MI (Figure 3C) in the PCI groups (EXCEL: 4.3% versus 2.7%, P=0.07; NOBLE: 6% versus 2%, P=0.004) and this drove the higher major adverse cardiac or cardiovascular end point risk from PCI. In EXCEL the primary end point was death, stroke, or MI and was similar between the PCI and CABG groups. The superiority of CABG in NOBLE with regards to its composite primary end point was primarily driven by higher rates of total revascularization (15% versus 10%, P=0.03) and nonprocedural MI (6% versus 2%, P=0.004) with PCI. Similarly, EXCEL demonstrated higher rates of repeat revascularization with PCI as compared with CABG (12.6% versus 7.5%, P<0.001) as a secondary end point. The addition of stroke and revascularization for the major adverse cardiac or cardiovascular end points in EXCEL, as shown in Figure 3A, is consistent with NOBLE and higher in the PCI group.

While the rate of MI in EXCEL was similar between PCI and CABG at 3 years (8.0% versus 8.3%, P=0.64), this was primarily driven by higher rates of periprocedural MI with CABG (5.9% versus 3.6%, P=0.02). In fact, nonprocedural MI in EXCEL, the end point used in NOBLE, tended to be higher with PCI at 3 years (4.3% versus 2.7%, P=0.07) in line with the NOBLE findings (Figure 3C). In addition, the longer length of follow‐up with NOBLE might have allowed for outcomes in favor of CABG to emerge. The highly anticipated 5‐year follow‐up data from EXCEL should shed further light on the long‐term performance of PCI as compared with CABG, particularly given concerns about a signal of divergence of event curves at 3 years in favor of CABG. Consistent with other studies, mortality did not differ between revascularization strategies in both studies (Figure 3B).

Conclusions

LMCAD is one of the most challenging conditions encountered in clinical practice. Current practice guidelines in support of indiscriminate revascularization of all LM lesions ≥50% are based on older trials in an era when medical therapy was limited and before the use of invasive physiological assessment of stenosis severity. In fact, the same evidence suggests that medical management of patients at lower risk might be associated with favorable outcomes. Although smaller studies support the use of FFR and IVUS to define lower‐risk groups with LM disease who could be treated by optimal medical therapy alone, larger trials assessing clinical outcomes over longer follow‐up are needed to fully assess this strategy.

Current clinical guidelines strongly recommend surgical revascularization for LMCAD (class IA) with PCI considered a reasonable alternative (class II) in select patients with less complex anatomy (SYNTAX score of <33) and clinical characteristics that predict an increased risk of adverse surgical outcomes.8 Advances in CABG and PCI techniques assessed in recent randomized trials show that PCI for LMCAD is a safe option with similar long‐term survival rates to CABG surgery, particularly in those with low and intermediate anatomic risk.34, 40 However, patients with PCI need close clinical follow‐up, as they may have a higher need for repeat revascularization in the future. It is expected that the results of EXCEL and NOBLE will determine the next guidelines for the foreseeable future, as forthcoming trials of this magnitude are unlikely to be pursued from economic and priority viewpoints unless marked advances in revascularization technologies emerge. Importantly, a heart team approach for shared decision‐making should be the standard of care for all cases of LMCAD.

Disclosures

None.

Data S1. Summary of the Six Randomized Clinical Trials Comparing PCI with CABG in Left Main Coronary Artery Disease Revascularization.

Click here for additional data file.