Review of the NOBLE Trial

PCI versus CABG in treatment of unprotected left main stenosis: a prospective, randomized, open-label, non-inferiority trial

Lancet 2016;388:2743-52

Background: PCI was commonly used for the management of patients with left main coronary artery disease and favorable pathology (i.e., absence of complex lesions). This was recommended by the guidelines at the time and based mainly on the prespecified subgroup of 705 patients with left main disease in the SYNTAX trial.

The NOBLE trial sought to more rigorously test the hypothesis that PCI with drug eluting stents was non-inferior to CABG in patients with unprotected left main coronary artery stenosis.

Patients: Inclusion criteria for study enrollment were stable angina pectoris, unstable angina, or acute coronary syndrome with a left main lesion with visually assessed stenosis diameter >/= 50% or fractional flow reserve </= 0.8 in the ostium, mid-shaft, or bifurcation with no more than 3 additional non-complex lesions.

Complex lesions were defined as chronic total occlusions, bifurcation lesions requiring 2 stent techniques or lesions with calcified or tortuous vessel morphology.

Exclusion criteria were STEMI within 24 hr, being considered too high risk for CABG or PCI (no further specification), or expected survival of less than 1 year.

Baseline characteristics: The trial randomized 1,201 patients - 598 to PCI and 603 to CABG. The average age of patients was 66 years and 78% were men.

The average BMI was 28 kg/m2. Approximately 65% had hypertension, 80% were taking a stain, 33% had a previous PCI, 15% had diabetes, 20% were active smokers, and approximately 50% of patients had an NYHA class between II and IV. Notably, NYHA class was not evenly distributed between the PCI and CABG groups with the CABG group having a higher percentage of patients with NYHA scores >I.

The number of target lesions was 2 in both groups. The average SYNTAX score was 22.5 (for reference, it was 28.7 in the SYNTAX trial). The percentage of patients with stable angina was 82% and acute coronary syndrome was 18%.

Patients were enrolled from 36 different centers but 5 centers contributed 42% of them. A screening log from these 5 centers was presented in the main paper. For these centers, the enrolled/screened ratio was 52%. The most common reasons for exclusion were unsuitable coronary anatomy for PCI.

Procedures: Patients were assigned 1:1 to undergo PCI with drug eluting stents or CABG. After treatment of 73 patients with PCI, the Biolimus-eluting stent became the recommended study stent. Randomization was stratified by sex, the presence of a distal left main bifurcation lesion, and diabetes.

The intent was to achieve complete revascularization of all vessels with significant lesions. The techniques for CABG and PCI as well as post-treatment medicines were chosen based on local practice but included 75-150 mg of aspirin lifelong. In both groups, patients with acute coronary syndrome received 75 mg of clopidogrel daily for 12 months. All patients in the PCI group also received 75 mg clopidogrel daily for 12 months and prasugrel or ticagrelor could be substituted at the discretion of the PCI operator.

Endpoints: The primary endpoint was a composite of death from any cause, non-procedural myocardial infarction, repeat revascularization, or stroke.

Analysis was performed based on the intention-to-treat principle. The primary analysis was a non-inferiority analysis. The main hypothesis tested was non-inferiority of PCI to CABG, assessed as the upper limit of the 95% CI of the hazard ratio (HR) of PCI to CABG, not exceeding 1.35 at median 3 years of follow-up.

It should be noted that the sample size calculation for NOBLE was based on 1 year event estimates from the SYNTAX trial that were extrapolated to 2 years. The sample size of 1,200 patients came from the original hypothesis test of non-inferiority corresponding to a total of 275 events at median 2 years of follow-up. However, due to slow accrual of events, assessment of the primary endpoint could not be reached within the limit of the study and so the primary endpoint assessment was pushed back to 3 years of follow-up. Despite this, the prespecified event total (275 events) was not reached.

Results: 1,201 patients were enrolled, which included 598 in the PCI group and 603 in the CABG group; however, only 592 patients in each group were entered into the final intention-to-treat analysis. Compared to CABG, PCI increased the primary composite endpoint (28% vs 18%; HR 1.51; 95% CI 1.13-2.00). Since the upper bound of the 95% CI was >1.35, PCI did not pass the test for non-inferiority. The primary outcome was driven mainly by differences in non-procedural MI (6% vs 2%; HR 2.87; 1.40-5.89) and repeat revascularization (15% vs 10%; HR 1.50; 1.04-2.17), which favored the CABG group. There was no difference in all-cause mortality (11% vs 9%; HR 1.08; 0.67-1.74) or its sub-component of cardiovascular mortality (3% vs 3%; HR 0.92; 0.44-1.90). The outcome of stroke was numerically higher with PCI but the difference was not statistically significant (5% vs 2%; HR 2.20; 0.91-5.36). When examining the Kaplan-Meier curves for the primary outcome, the curves start to diverge at 1 year.

PCI was associated with significantly lower 30 day morbidity compared to CABG exemplified by 3 outcomes: duration of index treatment admission (2 vs 9 days; p<0.0001), reoperation for bleeding (4% vs <1%; p<0.0001), and blood transfusion (2% vs 28%; p<0.0001).

The only subgroup information provided in the main paper is based on SYNTAX score. No interactions were identified based on the score but only a small percentage of patients in the trial (9%) had high scores, recall the average SYNTAX score of trial participants was 22.5.

Conclusions: In patients with left main coronary artery stenosis, who have suitable anatomy for PCI, PCI was not found to be non-inferior to CABG for the 4 component primary endpoint used in this trial. This difference was driven by higher rates on non-procedural MI and repeat revascularization in the PCI group. However, despite noted improvements in these nonfatal outcomes, definite tradeoffs in terms of other endpoints associated with morbidity were found. CABG was associated with a significantly longer length of stay, reoperations and blood transfusions. Data on all-cause hospitalizations following the index treatment is not provided.

One limitation of the NOBLE trial was its low power due to the slow accrual of events necessitating changes to the hypothesis testing protocol, which was originally supposed to occur after less time had elapsed and 275 events occurred. The above results were presented after only 201 events.

Updated 5-year outcomes were presented in a later publication in 2020 that included a total of 275 events, as originally specified. There were no significant differences with these results compared to the original publication overall or for any of the individual components of the primary endpoint: MACCE (28% vs 19%), death (9% vs 9%), non-procedural MI (8% vs 3%), repeat revascularization (17% vs 10%).

In summary, the NOBLE trial showed PCI was not non-inferior to CABG in respect to certain non-fatal outcomes with a NNT of approximately 20-25 patients to prevent 1 non-procedural MI or repeat revascularization procedure. These differences must be balanced against the increased morbidity of CABG.

Unlike most trials involving CABG, the NOBLE trial did allow for inclusion of patients with ACS but they made up a small percentage of patients (18%) and no additional information is provided on these patients in the main manuscript.