Statistics from Altmetric.com
To understand the evolution in the strategies for the treatment of left main stem (LMS) disease in light of the three principal trials relating to this topic: SYNTAX, Nordic-Baltic-British Left Main Revascularisation (NOBLE) and Evaluation of Xience Versus CABG for Effectiveness of Left Main Revascularization (EXCEL).
To understand the similarities and differences in the SYNTAX, NOBLE and EXCEL trials.
To understand the relevant role of the Heart Team in the decision process for the best treatment strategy for LMS disease.
Obstructive atheroma of the left main stem (LMS) is relatively unusual as it is present in only 4% of diagnostic coronary angiograms.1 Historically, disease of the LMS has been treated with particular respect as the LMS provides blood supply to at least two-thirds of the left ventricle,2 since prognosis after diagnosis of LMS obstruction is poor without treatment and because percutaneous coronary intervention (PCI) to the LMS usually involves a major bifurcation with higher calcific and fibrotic components than atherosclerosis elsewhere in the coronary tree.3 4 These historical data and technical challenges resulted in coronary bypass grafting (CABG) being the standard treatment of LMS narrowing especially because of the excellent results achieved using the left internal mammary artery graft to the left anterior descending artery.5
Encouraging outcomes when treating the LMS with PCI in patients unsuitable for surgery and technical advances in the predictability of PCI outcome have generated interest and reconsideration of the relative merits of PCI and CABG. Four small preliminary studies6–9 were performed, but provocative data from the randomised Synergy between PCI with Taxus and Cardiac Surgery (SYNTAX) trial substudy9 provided sufficient circumstantial evidence for the initiation of two international multicentre randomised comparative trials of surgery versus PCI for LMS disease: Evaluation of Xience Versus CABG for Effectiveness of Left Main Revascularization (EXCEL)10 and Nordic-Baltic-British Left Main Revascularisation (NOBLE).11 Both trials reported their primary endpoints in late 2016, and in this review we analyse current management of LMS disease and how this may change after the publication of these two landmark trials.
Management of significant left main stem disease before 2016
Identifying the optimal management for patients with LMS disease has many facets, but framing the clinical scenario requiring treatment is the first decision that the operator must categorise.
Currently the operator can ground this decision according to three main parameters:
elective (stable angina)
urgent (acute coronary syndrome (ACS))
emergency with LMS occlusion or impaired antegrade flow.
LMS anatomical complexity
Coronary anatomical complexity
isolated LMS disease
LMS disease+multivessel disease.
In patients presenting as an emergency, the extremely high mortality rate associated with acute LMS occlusion mandates immediate intervention usually with PCI.2 However, involvement of the anaesthetic and the cardiac surgical team in the management is ideal because of the risk of pulmonary oedema, cardiogenic shock and cardiac arrest, and to maintain a bail-out CABG option. In these clinical scenarios, the use of high-intensity circulatory support devices (eg, Impella or extracorporeal membrane oxygenation) may be increasingly considered.
When LMS disease is detected in a non-emergency setting, the anatomical complexity of the lesion together with the overall atheromatous burden in the coronary arteries and elsewhere in the peripheral vasculature is relevant, together with the assessment of left ventricular function and, importantly, the patient’s comorbidity. A discussion with colleagues within the Heart Team is in the best interest of both the patient and the individuals providing the care, and remains the cornerstone of best practice. The angiographic SYNTAX score is a parameter derived to express the overall anatomical complexity of coronary disease. According to the latest European guidelines, the SYNTAX score can guide the decision process in patients with stable coronary syndrome and LMS disease in the following manner12:
LMS disease with SYNTAX score ≤22: indication IA for both CABG and PCI
LMS disease with 22<SYNTAX score≤32: indication IB for CABG and IIaB for PCI
LMS disease with SYNTAX score >32: indication IB for CABG and IIIB for PCI.
These recommendations and use of the SYNTAX score were derived from the SYNTAX trial, which represented the only large randomised clinical trial available before 2016, addressing the comparison of PCI versus CABG in patients with three-vessel and LMS disease. The SYNTAX study was designed to assess the efficacy of PCI with a first-generation paclitaxel-eluting stent (Taxus, Boston Scientific) versus CABG in patients with three-vessel disease,13 with data and results for patients with LMS derived as hypothesis-generating from a powered and predefined subgroup of 705 patients.9 14
In the SYNTAX-LMS subset, patients treated with PCI and CABG appeared to have comparable rates of the combined endpoint of all-cause death, myocardial infarction (MI), cerebrovascular accident (CVA) and repeat revascularisation (HR 1.23 (0.95, 15.9), p=0.12) at 1 year.9 However, at 5-year follow-up there were significantly lower rates of CVA (HR 0.33 (0.12, 0.92), p=0.03) and a higher rate of repeated revascularisation (HR 1.82 (1.28–2.57), p<0.01) in the PCI group (table 1).
More importantly, when the cohort was stratified according to the anatomical complexity expressed by a low/intermediate (≤32) versus high (≥33) SYNTAX scores, there were possible differences in event rates. In the low/intermediate SYNTAX score group, there were no significant differences between PCI and CABG in the combined endpoint (31.3% vs 32.1%, p=0.74), MI (6.1% vs 3.8%, p=0.33), CVA (1.4% vs 3.9%, p=0.11) and repeat revascularisation (22.6% vs 18.6%, p=0.36), but interestingly there was a lower all-cause death rate observed in the PCI group (7.9% vs 15.1%, p=0.02).14 Conversely, in the high SYNTAX score group, outcomes were better in patients treated with CABG than those treated with PCI (combined endpoint 29.7% vs 46.5% p=0.003), which appears to have been principally driven by the higher repeat revascularisation rates (11.6% vs 34.1%, p<0.001).14
When considering these data from the SYNTAX trial, it is important to remember two important limitations:
The results pertaining to patients with LMS disease were from a subgroup analysis (constituting only 13% of patients who had isolated LMS disease) in a trial that showed the superiority of CABG over PCI.
PCI has evolved since completion of the SYNTAX trial. In the SYNTAX study a first-generation drug-eluting stent (DES) was used and coronary stenosis severity was assessed according to the angiogram, without resorting to fractional flow reserve (FFR) measurements and with minimal usage of intravascular imaging (13.5% of patients).15 Staged procedures were rarely performed resulting in high rates of incomplete revascularisation in patients treated with PCI.
NOBLE and EXCEL: impact on patient management?
Changes in the management of LMS disease may be expected after the publication of the NOBLE and EXCEL trials. However, it is important to understand the differences and similarities in these studies as well as the trial limitations.
The NOBLE was the first of the two trials comparing PCI versus CABG for LMS disease.11
The study had a non-inferiority design and randomised 1201 patients with LMS disease (enrolled in 36 centres) to percutaneous revascularisation with mainly biolimus-eluting stents (BES) versus CABG over a period of 6 years (from December 2008 to January 2015). LMS disease was defined according to an angiographic stenosis ≥50% and/or an FFR ≤0.80. Notably, however, a SYNTAX score ≤32 was not a prespecified inclusion criterion, but up to three additional non-complex coronary lesions were permitted in addition to an LMS lesion. Presentation with ACS was not an exclusion criterion, but patients with ST elevation myocardial infarction (STEMI) within 24 hours and patients with low life expectancy were excluded. The central role of the Heart Team was stressed in NOBLE and patients were enrolled in the study only if there was agreement between the interventional cardiologist and the cardiac surgeon. The primary endpoint was represented by major adverse cardiac and cerebrovascular events (MACCE) including death from any cause, non-procedural MI, repeat revascularisation and stroke at a median of 3 years of follow-up.
Notably no differences in baseline characteristics were observed between PCI and CABG groups, and isolated LMS disease occurred in 54% of cases, with bifurcation disease reported in 88% of cases requiring a double stent technique strategy in 35% of cases, according to the per-protocol analysis.16 The groups were matched for angiographic complexity with a mean SYNTAX score of 22.5 in the PCI arm versus 22.4 in the CABG arm (p=0.74). The primary endpoint findings are presented in table 1. Estimated MACCE rate at 5 years favoured CABG (HR 1.48 (1.11, 1.96), p=0.007 for superiority) mainly driven by a significantly higher rate of non-procedural MI (7% vs 2%, p=0.004) and higher repeat revascularisation (16% vs 10%, p=0.03) in the PCI group.11
Some of the issues highlighted within this trial are summarised below:
Is NOBLE a ‘true’ second-generation stent study?
It is important to note that the Biomatrix Flex (Biosensors, Morges, Switzerland) became the recommended study stent in 2010 after 73 patients had already been enrolled, with 44 patients (7.4% of the PCI cohort) who received first-generation sirolimus-eluting stents. This could have potentially affected the efficacy of PCI and may have contributed to the observed definite stent thrombosis rate of 2% in NOBLE, which is higher than contemporary second-generation and third-generation stent PCI trials.17–19
Does procedural MI matter?
In NOBLE the composite primary endpoint excluded procedural MI. It may be argued that this approach misses the capture of important myocardial events occurring early in the treatment plans of patients undergoing revascularisation.
Why was a prespecified median 3-year follow-up used versus a reported 5-year MACCE rate?
The primary endpoint of the NOBLE trial was changed while the study was being performed because of a lower than expected event rate. The initial follow-up was supposed to be 2 years, then extended to a 3-year median follow-up, but at the end of the fifth year of follow-up the observed event rate was still lower than anticipated. Results were extrapolated to a projected 5-year follow-up, and this may have preferentially represented the patients initially recruited within the trial and may have affected the final results.
Why did the SYNTAX score not predict outcome in NOBLE?
In apparent contrast with previous observations in the SYNTAX study, PCI failed to show any benefit compared with CABG in the low SYNTAX score subgroups of the NOBLE trial. It is possible that the high rate of observed bifurcation disease (88% in NOBLE vs 64.1% in SYNTAX study) may be relevant. It has been suggested that the extra technical challenge of LMS bifurcation treatment occurring in centres with limited familiarity with LMS treatment could be a possible explanation.
Why was there a high rate of stroke in the PCI arm?
This is a further peculiarity of the study, which reported a higher rate of stroke (5% vs 2%, p=0.07) in the PCI group, in contrast with previous studies. More interestingly, no strokes were reported in the PCI group in the first 30 days postprocedure. It appears likely that the ‘late stroke’ occurrence after PCI could be due to chance.
Published simultaneously with the NOBLE trial results, the EXCEL trial is the largest randomised clinical trial comparing PCI versus surgery specifically for the treatment of LMS disease.10
The EXCEL study was a non-inferiority study randomising 1905 patients (initially designed to enrol 2600) with LMS disease to CABG versus PCI with second-generation cobalt-chromium everolimus-eluting stent in 126 centres in 17 countries over a period of nearly 3 years (from September 2010 to March 2014). LMS disease was defined according to an angiographic stenosis ≥70% or an angiographic stenosis between 50% and 70% in the presence of an invasive (FFR) or non-invasive documentation of inducible ischaemia. It was a study requirement that enrolled patients who had a low-intermediate SYNTAX score (≤32), with SYNTAX score being calculated by the enrolling site. As in NOBLE, in the EXCEL trial the role of the Heart Team was stressed with inclusion of patients for whom agreement about equipoise of treatment was achieved between interventional cardiologist and cardiac surgeon. A detailed treatment schedule was agreed and optimal stent strategies were predefined by the steering group.
The primary endpoint of the study was represented by MACCE rate at 3-year follow-up. MACCE definition in EXCEL differed from that in NOBLE by including only all-cause mortality, MI and stroke. Secondary endpoints in EXCEL were (1) the rate of MACCE at 30 days and (2) the rate of all-cause mortality, MI, stroke and repeated revascularisation (a MACCE definition similar to that seen in the NOBLE trial) at 3 years.
No differences in baseline characteristics were observed between PCI and CABG groups and notably bifurcation disease was reported in 81% of patients randomised to PCI (a similar proportion to that seen in NOBLE).
The primary endpoint of the study demonstrated that PCI was not inferior to CABG (0.7% difference with upper 97.5% confidence limit of 4.0%, p=0.02 for non-inferiority) with no significant differences in the rates of MACCE at 3 years between the two treatment arms (HR 1.0 (0.79, 1.26), p=0.98). In addition, when ischaemia-driven revascularisation was added to the composite endpoint, no significant differences in events were found between those undergoing PCI versus CABG (23.1% vs 19.1% with HR 1.18 (0.97, 1.45), p=0.10 for superiority and p=0.01 for non-inferiority). Moreover, PCI was associated with a lower rate of 30 days of MACCE compared with CABG (4.9% vs 7.9%, p value for non-inferiority <0.001). Interestingly at 30 days, the main benefit of PCI over CABG appeared in terms of a lower rate of MI (3.9% vs 6.2%, p=0.02), mainly driven by a lower occurrence of periprocedural MI and STEMI in the PCI arm (p=0.02 and p=0.005, respectively).
Interestingly, no significant differences in the rates of the composite endpoint were found between the treatment groups when the cohorts were stratified by various prespecified subgroups including diabetes (n=505, 21.3% vs 19.4%, p=0.77) and high SYNTAX score (n=239, 16.9% vs 14.3%, p=0.49). Encouragingly, the definite stent thrombosis rate (0.7%) was considerably lower than the symptomatic vein graft occlusion rate (5.4%) (HR 0.12 (0.05, 0.28), p<0.001). This difference needs to be read also at the light of a significantly larger proportion of patients on P2Y12 inhibitor, on top aspirin, treatment in the PCI arm compared with CABG group both at discharge and after 30 days postdischarge (97.6% vs 32.6% at discharge, p<0.001; and 97.2% vs 31.8%, p<0.001).
The EXCEL study also requires a careful interpretation:
Is EXCEL a ‘true’ second-generation stent study?
Compared with NOBLE, patients in the EXCEL study received a second-generation DES in 98.4% of cases. This could be relevant with respect to the observed lower rate of definite stent thrombosis (0.7%) reported in the EXCEL trial compared with NOBLE (2.0%).
Does procedural MI matter?
The predefined primary composite endpoint did not include repeat revascularisation; however, it did include both non-procedural MI and large periprocedural MI (defined as creatine kinase MB (CK-MB) >10 times the upper reference limiting (URL), or CK-MB >5 times the URL plus electrocardiographic, angiographic or imaging confirmation of MI). This was considered important in order to capture clinically relevant events after the procedure and at the time of revascularisation.
Why a change in the sample size?
The EXCEL study was initially designed to have a 90% statistical power, thus requiring a recruitment of 2600 patients. While the study was ongoing the sample size was reduced to 1900 because of a slower than anticipated recruitment rate. However, the final sample size of 1900 patients ensured an 80% power for the primary endpoint and for the 30-day secondary point, and a 99% power for the 3-year secondary endpoint.
Does the SYNTAX score predict outcome in EXCEL?
Although a high SYNTAX score was a predefined exclusion criterion and locally assessed scores complied with this study requirement, post hoc core lab analysis of the angiographic data revealed that 24% of patients had a high SYNTAX score. This is a larger proportion than that seen in the NOBLE trial (9%). However, in common with the NOBLE trial, no clear benefit from PCI was observed in low and intermediate SYNTAX score subgroups, even though (converse to the NOBLE trial) a trend for a better outcome was observed for PCI in the low SYNTAX score subgroup and for CABG in high SYNTAX score subgroup.
Catch-up phenomenon of PCI curve at 3 years
Even though encouraging for PCI, it could be argued that the Kaplan-Meier curves appear to begin to cross at 3 years, meaning that the immediate benefit of PCI over CABG might be progressively lost in the long term.
LMS intervention from SYNTAX to NOBLE and EXCEL: is it time to recommend coronary stenting?
Despite contrasting results, the NOBLE and EXCEL trials have undoubtedly increased the available evidence regarding the treatment of LMS disease. When one considers the large amount of energy and resources expended to deliver these two large randomised clinical trials, it is unlikely that new studies of such size will be repeated. For this reason it is probable that the evidence derived from SYNTAX, NOBLE and EXCEL, and from their substudies, will dictate the rules of revascularisation strategy for LMS disease in the coming years.20
It is probably too early to think of a dramatic change in the current guidelines in light of the only short follow-up periods of these trials. Even though influenced by not trivial limitations, the NOBLE results introduce caution before considering stenting over surgery for all significant LMS disease in the low-intermediate SYNTAX score tertile.
Furthermore, the catch-up phenomenon observed in the PCI arm of the EXCEL study also invites caution before coming to overenthusiastic conclusions, and it warrants patience for longer follow-up data before changing the guidelines in favour of stenting.21
We believe that three main messages come from a careful analysis of the evidence collected so far. First, there does not appear to be any mortality benefit from CABG compared with PCI in patients with LMS disease. The two studies reflect a true and fair comparison between contemporary PCI in expert centres versus CABG in patients with LMS disease. In our opinion, it is highly improbable that higher rates of revascularisation with bilateral internal mammary artery and/or off-pump CABG would make any difference to the outcome. This statement is informed by the Arterial Revascularization Therapies Study (ARTS)22 and the CABG Off or On Pump Revascularization (CORONARY)23studies, and recognises that in most centres a single arterial graft with two vein grafts is the usual surgical approach.
Second, the Heart Team remains central to the decision-making process in the management of patients with LMS disease. Careful individual patient review will usually highlight particular features, either anatomical or clinical patient-related, that will favour a particular revascularisation approach. Maintaining an environment where teamwork is possible between interventional cardiologist and cardiac surgeon guarantees the best results for the individual patient.
Following presentation of these trial data, patients with less complex LMS disease, which spares the bifurcation, and patients at higher surgical risk are more likely to be treated with PCI. However, when PCI is to be considered, it should be done in expert hands with optimal stent technology. Intravascular ultrasound (IVUS) was used in the majority of cases in both trials and previous evidence has suggested its value during LMS therapy24. Notably, as reported at the Transcatheter Therapeutics 2016 meeting, IVUS was performed in 73% of cases in the EXCEL study, triggering further action (postdilation and/or further stenting) in 51.7% of cases and with a significant trend towards worse outcome in patients having a low minimum stent area (between 4.4 and 8.7 mm2). For this reason, careful lesion preparation, appropriate stent sizing and intravascular imaging-guided stent optimisation, together with emphasis on compliance to drug regimens, were all part of the EXCEL protocol and require emphasis if the trial outcomes are to be applicable to routine clinical practice.
Finally, both EXCEL and NOBLE were unable to show potential benefit of PCI in the low SYNTAX tertile as previously reported in the SYNTAX LMS subgroup. Furthermore, the NOBLE trial showed an unprecedented superiority of CABG over PCI in this low SYNTAX score tertile subgroup.
The possible explanations for these unexpected findings may be multifactorial. First, it should be noted that low-intermediate SYNTAX score was an inclusion criterion in EXCEL but not in NOBLE. Second, in the EXCEL study there was a marked discrepancy between site-assessed and core lab-assessed SYNTAX score, uncovering the difficulties in consistency and uniformity in the application of this stratification tool. Third, it should not be forgotten that the SYNTAX score was specifically developed for patients with three-vessel disease and/or LMS disease. The relative weighting of LMS disease within the score may have been inaccurate. These considerations, together with the wealth of new data, suggest the potential need for a new treatment stratification scoring system specific for LMS disease.
With the simultaneous publication of NOBLE and EXCEL studies in 2016, evidence regarding revascularisation in LMS disease has expanded dramatically. Both trials were large and well conducted, but because of their contrasting results and potential limitations, a uniform recommendation about the relative role of stenting compared with CABG in LMS disease is inappropriate. A possible algorithm to support and aid the decision-making process is shown in figure 1.
The role of careful angiographic and individual clinical risk assessment with recognition of local expertise within the Heart Team is the contemporary standard of care. In most patients and clinical scenarios, informed debate, with the patient’s best interests at the centre, will generate a consensual decision. This can be informed by the results of the NOBLE and EXCEL trials, and achieved without a return to confrontational acrimony evident in discussions about revascularisation strategy before the SYNTAX trial was performed.
What is known on this subject
SYNTAX study suggested potentially similar outcomes of percutaneous coronary intervention (PCI) and coronary bypass grafting (CABG) in patients with left main stem (LMS) disease with low-intermediate complexity of coronary anatomy. But SYNTAX was not an LMS-designed study and thus further evidence was necessary.
What might this study add?
Nordic-Baltic-British Left Main Revascularisation (NOBLE) and the Evaluation of Xience Versus CABG for Effectiveness of Left Main Revascularization (EXCEL) studies were the first two randomised studies specifically designed to address the comparison of PCI versus CABG in LMS disease.
The NOBLE study did not show the non-inferiority of PCI in LMS disease at 5 years.
The EXCEL study showed a non-inferiority of PCI in LMS compared with CABG at 30 days and 3 years of follow-up.
How might this impact clinical practice
There is no prognostic advantage to CABG compared with PCI performed in expert centres.
The Heart Team continues to have a crucial role in the decision process about the best revascularisation strategy for LMS disease in patients with low-intermediate SYNTAX scores.
CME credits for Education in Heart
Education in Heart articles are accredited for CME by various providers. To answer the accompanying multiple choice questions (MCQs) and obtain your credits, click on the ‘Take the Test’ link on the online version of the article. The MCQs are hosted on BMJ Learning. All users must complete a one-time registration on BMJ Learning and subsequently log in on every visit using their username and password to access modules and their CME record. Accreditation is only valid for 2 years from the date of publication. Printable CME certificates are available to users that achieve the minimum pass mark.
Contributors Manuscript drafting: GLDM and NP. Manuscript draft editing and approval: GLDM, NP and APB. Guarantor: APB.
Competing interests APB has received honorarium payments from Abbott Vascular, Boston Scientific and Medtronic. NP has received travel expenses for attendance to an educational meeting from Biosensors.
Provenance and peer review Commissioned; externally peer reviewed.
Author note The references marked with a * are considered key references for this paper.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.