Randomised trials in left main disease: a NOBLE effort.
Michael J. Mack, M.D.
Mack, M. and D. R. Holmes (2016). “Randomised trials in left main disease: a NOBLE effort.” Lancet 388(10061): 2715-2716.
The NOBLE trial enrolled 1201 patients in 36 centres, randomly assigned to either PCI primarily with the Biolimus–eluting stent (Biomatrix Flex) or CABG. The primary endpoint was major adverse cardiac and cerebrovascular events including all-cause mortality, myocardial infarction, stroke, and repeat revascularisation at a median follow-up of 3 years. The trial was designed as a non-inferiority trial with a relatively wide confidence interval of 1·35. The primary endpoint occurred in 29% of patients in the PCI group and 19% of patients in the CABG group (HR 1·48 [95% CI 1·11–1·96]), exceeding the limit for non-inferiority and was in fact statistically significant for superiority of CABG over PCI (p=0·0079). Outcomes were similar in both intention-to-treat and per-protocol analyses. Regarding the individual components of the composite primary endpoint, 5 year Kaplan Meier estimates of all-cause mortality and stroke were the same but there were significantly fewer clinically apparent myocardial infarctions and fewer repeat revascularisation procedures with CABG. The conclusion of the trial is that CABG might be superior to PCI for treatment of left main stem coronary artery disease. There are a few specific findings of this trial that are noteworthy and somewhat surprising but there are also some concerns. The first is that the benefit of CABG was noted in all ranges of the SYNTAX score, which is contrary to the SYNTAX trial in which patients with less complex disease did as well with PCI as with CABG. This finding might partly be explained by the fact that 81% of the patients in NOBLE had bifurcation left main disease, which is more difficult to treat than ostial or trunk left main disease and might be associated with a worse outcome with PCI. The second finding is that there was a trend toward a higher incidence of stroke at 5 years with PCI, which is the opposite to what has been noted in most previous comparative trials. This result is perplexing in that with PCI at 30 days, there were no strokes. The stroke rate only gradually increased over time with PCI to an estimated 4·9% at 5 years. The reasons for the late stroke rate in PCI might be due to chance in the absence of any reasons for this late event. The third point is that the trial primary endpoint was changed from the original design. This trial was event driven, in which a specific number of events was needed; this requirement was based on events in the SYNTAX trial. When that number could not be reached in 5 years of follow-up, the primary endpoint was assessed at a median of 3 years. Might that have affected the results? Fourth, procedural purists on both sides will argue that optimal procedures were not done, with only 75% of the patients with PCI undergoing intravascular ultrasound assessment, only 93% of patients with CABG receiving a left internal mammary artery graft, and 86% of patients receiving at least one saphenous vein graft. However, we suspect that this is probably representative of real world practice. (Excerpt from text of this commentary, p. 2715-2716.)