2017-03-16
The authors reply [to: Everolimus-Eluting Stents or Bypass Surgery for Left Main Coronary Disease]
Publication
Publication
New England Journal of Medicine , Volume 376 - Issue 11 p. 1089
Belley-Côté et al. question
the noninferiority margin in the EXCEL trial. The
margin should be selected to support a reasonable
conclusion of therapeutic interchangeability
between the test and control treatments,1 and in
an active controlled trial, the margin may be
based on a putative placebo or expert consensus
of the minimal clinically important difference.2
Trial feasibility often necessitates composite end
points. The cardiology and surgical leadership in
the EXCEL trial unanimously agreed that death,
stroke, and large MI are clinically important end
points and that a margin of 4.2 percentage points
was acceptable. An examination of the absolute
95% confidence intervals of the 3-year difference
with respect to the primary end point in the
EXCEL trial (which accounts for the number of
patients enrolled) shows that PCI may range from
2.7 percentage points better to 4.0 percentage
points worse than CABG for treatment of left
main disease, with a Synergy between Percutaneous
Coronary Intervention with Taxus and Cardiac
Surgery (SYNTAX) score of 32 or lower (the
SYNTAX score reflects a comprehensive angiographic
assessment of the coronary vasculature,
with 0 as the lowest score and higher scores [no
upper limit] indicating more complex coronary
anatomy).
Both Belley-Côté et al. and Christiansen et al.
question the appropriateness of the EXCEL protocol
definition of MI. Our MI definition represents
extensive myonecrosis, which has been
strongly associated with subsequent death in
patients who underwent PCI and patients who
underwent CABG3 and avoids ascertainment
bias. Periprocedural MI, as defined in the EXCEL
trial, was significantly associated with death at
3 years in patients who underwent PCI (hazard
ratio, 3.04; 95% confidence interval [CI], 1.39 to
6.63) and in patients who underwent CABG
(hazard ratio, 2.44; 95% CI, 1.10 to 5.40), which
justifies its inclusion in the composite outcome.
We disagree with Christiansen et al. that repeat
revascularization (which favored CABG) deserves
equality with death, stroke, and large MI as a
primary end-point event. Repeat revascularization
is of no greater clinical consequence than
many other adverse outcomes that favored PCI,
such as major bleeding, renal failure, arrhythmia,
or serious infection. Comparisons between the
EXCEL trial and NOBLE4 are problematic because
a relatively thick-strut stainless steel stent
was used in NOBLE; the rate of thrombosis
among the patient who received this stent in
NOBLE was 4 times as high as that among the
patients who received the everolimus-eluting stent
used in the EXCEL trial. Moreover, in NOBLE,
periprocedural MI was not routinely assessed,
and a higher rate of stroke was observed in association
with PCI than with CABG, which could
not be explained.
Finally, Belley-Côté et al. implicate undue industry
influence in the design of the EXCEL trial.
Our hypotheses and end points were developed
by a large group of practicing interventional
cardiologists and cardiac surgeons (with guidance
from the Food and Drug Administration) and
were not prejudiced by the sponsor.
Additional Metadata | |
---|---|
doi.org/10.1056/NEJMc1701177, hdl.handle.net/1765/98907 | |
New England Journal of Medicine | |
Organisation | Erasmus MC: University Medical Center Rotterdam |
Stone, G. W., Sabik, J., Serruys, P., & Kappetein, A. P. (2017). The authors reply [to: Everolimus-Eluting Stents or Bypass Surgery for Left Main Coronary Disease]. New England Journal of Medicine, 376(11). doi:10.1056/NEJMc1701177 |