Tugas Metaanalisis Kelompok 5

TUGAS PAPER
MATA KULIAH METODELOGI PENELITIAN DAN STATISTIKA
TOPIK:
METAANALISIS
Dosen Pengajar:
Dr. Budi Utomo, dr., M.Kes
Oleh:
Kelompok 5
Melia Bogari 012118246306
Karisma Septari Idamusaga 012118026318
Nurun Nabilla Junaedy 012118026311
Nuril Sudiyatma 012118026309
Isna Mahmudah 012118026314
Aldira Rasthy Krisdania 012118026316
Willy Gunawan 012118066305
Wisnu Wahyu Nugroho 012118066312
Nanda Daiva Putra 012118046302
Dinda Rozita Maharani 012118046303
MATA KULIAH DASAR UMUM

PROGRAM PENDIDIKAN DOKTER SPESIALIS-I
FK UNAIR-RSUD Dr. SOETOMO SURABAYA
2021
IDENTIFIKASI
Title Perbandingan angka kematian pasien penyakit jantung koroner

yang menjalani coronary artery bypass graft (CABG) dan
percutaneous coronary intervention (PCI)
Comparison of mortality rates of patients with coronary artery
disease undergoing coronary artery bypass graft (CABG) and
percutaneous coronary intervention (PCI)
Objective Mengidentifikasi, menggambarkan, mengevaluasi, dan
mensintesis semua bukti untuk tingkat kematian pasien dengan
penyakit jantung koroner yang menjalani CABG dan PCI.
Identifying, describing, evaluating, and synthesising all evidence
for the mortality rates of patients with coronary artery disease
undergoing CABG and PCI.
Participants/population Pasien dengan penyakit jantung koroner
Patients with coronary artery disease
Intervention(s), Pasien dengan penyakit jantung koroner yang menjalani prosedur
exposure(s) CABG
Patients with coronary artery disease undergoing a CABG
procedure
Comparator(s)/control Pasien yang menjalani prosedur PCI pada pasien penyakit arteri
koroner
Coronary artery disease patients who undergo the PCI procedure
Outcome Tingkat Kematian
Mortality rate
Keywords (coronary artery disease) AND (coronary artery bypass graft OR
CABG) AND (percutaneous coronary intervention OR PCI) AND
mortality rate OR mortality)
Protocol guideline PRISMA checklist and flowchart
STRATEGI PENCARIAN
Penelitian ini merupakan penelitian systematic review dan meta-analisis yang membahas
Perbandingan angka kematian pasien penyakit jantung koroner yang menjalani coronary artery
bypass graft (CABG) dan percutaneous coronary intervention (PCI). Penelitian systematic
review dapat memberikan informasi yang beragam antar beberapa studi yang dilakukan
sebelumnya. Penelitian juga dapat memberikan informasi baru dari beberapa artikel. Penelitian
juga dapat digunakan sebagai rekomendasi untuk pengembangan ilmu pengetahuan di dunia
medis terkait angka kematian pasien penyakit jantung koroner yang menjalani coronary artery
bypass graft (CABG) dan percutaneous coronary intervention (PCI),
Data yang dibutuhkan dalam penelitian ini merupakan data dari dalam beberapa artikel
ilmiah. Penelitian menggunakan literatur berupa artikel yang dipublikasikan selama lima tahun
terakhir yaitu tahun 2016 sampai 2021. Sumber data diawali dengan melakukan pencarian
jurnal akademik dalam basis data jurnal. Basis data jurnal yang digunakan adalah PubMed dan
Scopus. Dalam melakukan pencarian data jurnal yang sesuai dengan penelitian ini, maka kami
menggunakan kata kunci sebagai berikut:
a. coronary artery disease

b.CABG
c. PCI
d.Mortality rate
Dalam pencarian artikel, kata kunci digabungkan dengan Boolean yaitu AND and OR.
Penggunaan dari Boolean memiliki tujuan untuk memperluas pencarian artikel. Selanjutnya,
dapat memfokuskan pencarian pada artikel-artikel secara spesifik.
Tabel 1. Pencarian Literatur
Database Keyword
PubMed “coronary artery disease” AND “coronary

artery bypass graft OR CABG” AND
“percutaneous coronary intervention” OR
PCI” AND “mortality rate” OR mortality”
Scopus “coronary artery disease” AND “coronary
artery bypass graft OR CABG” AND
“percutaneous coronary intervention” OR
PCI” AND “mortality rate” OR mortality”
PubMed SCOPUS
Kriteria Inklusi:
1.Artikel terpublikasi 5
Hasil artikel yang ditemukan tahun terakhir.
2.Artikel dimuat dalam
jurnal internasional
3.Artikel dalam Bahasa
Inggris
Kriteria Eksklusi :
Screening
1.Artikel yang membahas tentang pasien dengan
coronary artery disease yang tidak berhubungan
tingkat kematian antara pemberian intervensi
CABG dan PCI
Artikel Akhir yang

dikaji
PRISMA FLOW DIAGRAM
Identification of studies via databases and registers
Records identified from: Records removed before

Databases (n = 2) screening:
Identification
Duplicate records removed (n

= 200)
PubMed (n=1439) & Scopus (n=956) Records marked as ineligible
by automation tools (n = 50)
Records removed for other
Registers (n = 2404) reasons (n = 673)
Records screened Records excluded

(n = 1481) studies irrelevant (n = 806)
Reports sought for retrieval Reports not retrieved

(n = 675) (n = 635)
Screening
Reports assessed for eligibility

(n = 40) Reports excluded:
25 Wrong outcomes
4 Wrong intervention
5 Wrong study design
Studies included in review

Included
(n = 6)
Reports of included studies
(n = 6)
DATA EKSTRAKSI
Authors Ramanathan K, Hideyuki Daniel J F M Gennaro Giustino, Park D, et al Ori Ben-Yehuda, et

et al Kawashima, MD, Thuijs, et al et al al
et al
Tahun 2017 2021 2019 2020 2020 2019
Judul Surgical Versus Mortality 10 Years Percutaneous Mortality After Ten-Year Impact of Large
Percutaneous After Percutaneous coronary Repeat Outcomes After Periprocedural
Coronary or Surgical intervention versus Revascularization Drug-Eluting Myocardial Infarction
Revascularization Revascularization coronary artery Following PCI or Stents Versus on Mortality after
in Patients with in Patients with bypass grafting in Coronary Artery Coronary Artery Percutaneous
Diabetes and Total Coronary patients with Bypass Grafting for Bypass Grafting Coronary Intervention
Acute Coronary Artery Occlusions three-vessel or Left Main Disease for Left Main and Coronary Artery
Syndromes left main Coronary Disease Bypass Grafting for
coronary artery Left Main Disease : an
disease: Analysis from the
10-year follow-up EXCEL Trial
of the
multicentre
randomised
controlled
SYNTAX trial
Metode population-based, Randomized Randomized Randomized Randomized Randomized
penelitian retrospective Controlled Trial Controlled Trial Controlled Trial Controlled Trial Controlled Trial
cohort study
British Columbia North American North American and North American and Korea USA
and European European countries European countries
countries
Population all patients older A total of 1,800 Patients aged 21 All patients were Patients were Patient with LM
than patients with years or older with required to have low eligible for diameter stenosis of
20 years of age de novo 3VD de-novo three- or intermediate participation in the >70% as estimated
with DM and and/or LM, who vessel disease and anatomic complexity trial if they had de visually or a stenosis
angiographically were deemed left main coronary of coronary artery novo stenosis of of 50%
confirmed MV- eligible artery disease disease, as defined the LMCA of more to <70% if determined
CAD (stenosis of for both PCI and by a site- determined than 50% (as by means of non-
>70% in 2 or CABG based on SYNTAX (Synergy estimated visually) invasive or invasive
more clinical judgment Between PCI With and had received a testing to be
major epicardial and Taxus and Cardiac diagnosis of stable haemodynamically
vessels, excluding the consensus of a Surgery) score of angina, unstable significant, a site-
the left main Heart Team <32 angina, silent assessed SYNTAX
coronary artery) ischemia, or non– score <32, and a
who underwent ST segment consensus among the
either PCI or elevation MI, in 13 members of the heart
isolated hospitals in Korea team regarding
CABG between from April 2004 to eligibility
October 1, 2007 August 2009. for revascularization
and January 31, Clinical and with either PCI or
2014 anatomic eligibility CABG
in British of all participants
Columbia had to be
considered by the
cardiologists and
surgeons at each
hospital to be
equivalently
suitable for both
PCI and CABG.
Jumlat total 4819 1800 1800 1905 600 1858
sample
Jumlah 1931 897 897 957 300 935
sample
CABG
Jumlah 2888 903 903 948 300 923
sample PCI
Variable The primary The primary The prespecified The primary The primary The primary endpoint
Outcome outcome was endpoint of this primary endpoint of endpoint of this outcome was a of the original study
yang diteliti the first study was all-cause the SYNTAXES study is to composite of major was the rate of a
occurrence of a mortality at 10 study was all-cause investigate the adverse cardiac and composite of death
major adverse years. The death at 10 years in incidence and cerebrovascular from any cause,
cardiac or secondary patients randomly impact on mortality events. Major stroke, or MI at 3
cerebrovascular endpoint was all- assigned to PCI of repeat secondary years, analysed by
event (MACCE), cause mortality at with drug-eluting revascularization outcomes included intention-totreat.
defined as a maximum stents versus after index the individual
composite available follow-up CABG. The percutaneous components of the
of all-cause secondary endpoint coronary primary composite
mortality, was all-cause death intervention (PCI) or outcome; a
nonfatal MI at maximum coronary artery composite of death,
(International available follow-up bypass grafting MI, or stroke, any
Classification of in patients randomly (CABG) for left revascularization
Disease-Tenth assigned to PCI main coronary artery and definite stent
Revision [ICD- with drug-eluting disease (LMCAD). thrombosis or
10] stents versus symptomatic graft
codes I21, I22) CABG. occlusion. All
and nonfatal participants in this
stroke (ICD-10 trial were invited to
codes I60 participate in 10-
to I64, H356, year follow-up
H341, The evaluations.
primary endpoint
of this
study was all-
cause mortality at
10 years. The
secondary
endpoint was all-
cause mortality at
maximum
available follow-
up H342, and
H348) after
revascularization.
Secondary
outcomes
included the
individual
components of
MACCE, repeat
revascularization
postdischarge
(RR), and a
composite of
MACCE and
repeat
revascularization
(MACCE[r])
Hasil At 30-days post- Of 1,800 From March, 2005, During 3-year At 10 years, a Periprocedural MI was
penelitian revascularization, randomized to April, 2007, 1800 follow-up, there primary outcome associated with
for ACS patients patients to the PCI patients were were 346 repeat event occurred in SYNTAX score,
the odds ratio for or CABG arm, 460 randomly assigned revascularization 29.8% of the PCI COPD, cross-clamp
MACCE favored patients had at least to the PCI (n=903) procedures among group and in 24.7% duration and total
CABG 0.49 (95% 1 lesion of TO. In or CABG (n=897) 185 patients. PCI of the CABG procedure duration,
confidence patients with group. Vital status was associated with group (hazard ratio and not using
interval [CI]: 0.34 TOs, the status of information at 10 higher rates of any [HR] with PCI vs antegrade
to 0.71), whereas TO recanalization years was complete repeat CABG, 1.25 [95% cardioplegia. By
among SIHD or revascularization for 841 (93%) revascularization CI, 0.93–1.69]). multivariable analysis,
patients MACCE was not associated patients in the PCI (12.9% vs. 7.6%; The 10-year PMI was associated
was not affected with 10-year all- group and 848 hazard ratio: 1.73; incidence of the with cardiovascular
by cause mortality, (95%) patients 95% confidence composite of death, death and all-cause
revascularization irrespective of in the CABG group. interval: 1.28 to myocardial death at 3 years
strategy (odds the assigned At 10 years, 244 2.33; p 1 4 0.0003). infarction, or [adjusted hazard ratio
ratio: 1.46; 95% treatment (PCI (27%) patients had Need for repeat stroke (18.2% vs (HR) 2.63, 95% CI
CI: 0.71 to 3.01; p arm: 29.9% vs. died after PCI and revascularization 17.5%; HR 1.00 1.19–5.81; P = 0.02
interaction 29.4%; adjusted 211 (24%) after was independently [95% CI, 0.70– and adjusted HR 2.28,
<0.01). With a hazard ratio [HR]: CABG (hazard associated with 1.44]) and all- 95% CI 1.22–4.29; P =
median follow-up 0.992; 95% ratio 1·17 [95% CI increased risk for 3- cause mortality 0.01, respectively].
of 3.3 years, the confidence interval 0·97–1·41], year all-cause (14.5% vs 13.8%; The effect of PMI was
late (31-day to 5- [CI]: 0.474 p=0·092). Among mortality (adjusted HR 1.13 [95% CI, consistent for PCI and
year) benefit of to 2.075; p = 0.982; patients with three- hazard ratio: 2.05; 0.75–1.70]) were CABG for
CABG over PCI and CABG arm: vessel disease, 151 95% confidence not significantly cardiovascular death
no longer varied 28.0% vs. 21.4%; (28%) of 546 had interval: 1.13 to different between (P interaction = 0.56)
by acuity of adjusted HR: died after 3.70; p 1 4 0.02) and the PCI and CABG and all-cause death (P
presentation, with 0.656; 95% CI: PCI versus 113 cardiovascular groups. Ischemia- interaction = 0.59).
a hazard ratio for 0.281 to 1.533; p = (21%) of 549 after mortality (adjusted driven target-vessel Peak post-procedure
MACCE in ACS 0.330). When TOs CABG (hazard ratio hazard ratio: 4.22; revascularization CK-MB >_10x URL
patients of 0.67 existed in left main 1·41 [95% CI 1·10– 95% confidence was more frequent strongly predicted
(95% CI: 0.55 to and/or left anterior 1·80]), and among interval: 2.10 to after PCI than after mortality, whereas
0.81) and the descending artery, patients with left 8.48; p < 0.0001) CABG (16.1% vs lesser degrees of
hazard ratio for the status of TO main consistently after 8.0%; HR 1.98 myonecrosis were not
SIHD patients of recanalization or coronary artery both PCI and CABG [95% CI, 1.21– associated with
0.55 (95% CI: revascularization disease, 93 (26%) of (pint 1 4 0.85 for 3.21). prognosis.
0.40 to 0.74; did not have 357 had died after both endpoints).
pinteraction = an impact on the PCI versus 98 Although target
0.28). mortality (34.5% (28%) of 348 after vessel
vs. 26.9%; adjusted CABG (0·90 [0·68– revascularization
HR: 0.896; 95% 1·20], and target lesion
CI: 0.314 to 2.555; p revascularization
p = 0.837). interaction=0·019). were both associated
There was no with an increased
treatment-by- risk for mortality,
subgroup interaction target vessel non–
with diabetes (p target lesion
interaction=0·66) revascularization
and no linear trend and non–target
across SYNTAX vessel
score tertiles (p revascularization
trend=0·30). were not.
Kesimpulan In diabetic "1. Among patients At 10 years, no In the EXCEL trial, In this 10-year In the EXCEL trial,
penelitian patients with MV- with TOs (n = significant repeat follow-up of the PMI was more
CAD, CABG was 460), there were no difference existed in revascularization PRECOMBAT common after CABG
associated with a significant all-cause death during follow-up trial that enrolled than PCI, and was
lower rate of differences in the between PCI using was performed less patients with strongly associated
long-term all-cause mortality first-generation frequently after LMCA disease, with increased
MACCE relative at paclitaxel-eluting CABG than PCI and there was no 3-year mortality after
to PCI for both 10 years between stents and CABG. was associated with significant controlling for
ACS and SIHD. patients with However, CABG increased mortality difference between potential confounders.
successfully provided a after both PCI and CABG in Only extensive
recanalized or significant survival procedures. the incidence of myonecrosis was
revascularized TOs benefit in patients Reducing the need major adverse prognostically
and those with three-vessel for repeat cardiac or important.
without, disease, but not in revascularization cerebrovascular
irrespective of the patients with left may further improve events, composite
assigned treatment main coronary long-term survival of death, MI, or
(PCI or CABG) artery disease. after percutaneous or stroke, and all-
and location of surgical treatment of cause mortality.
TOs (left main LMCAD. However, the study
coronary had insufficient
artery and/or LAD statistical power to
or other vessels). allow for a firm
2. When patients conclusion, hence
with TO were further research is
stratified according needed in this area.
to
the type of disease
(3VD or LM),
there was no
difference
between the PCI
and CABG arm in
terms of
the 10-year
mortality."
PRISMA CHECKLIST
Mengevaluasi setiap studi literatur yang digunakan dengan Prisma Checklist
Sebagai contoh pengisian adalah sebagai berikut;
PRISMA 2009 Checklist for systematic review
Section/topic # Checklist item Reported on page #
TITLE
Title 1 Identify the report as a systematic review. Describe in the title
ABSTRACT
Structured summary 2 Provide a structured summary including, as applicable: background; The eligibility criteria and limitations of the
objectives; data sources; study eligibility criteria, participants, and study are not well describe in abstract
interventions; study appraisal and synthesis methods; results; limitations; section.
conclusions and implications of key findings; systematic review registration Background, objective, data source, study
number. appraisal, and conclusion have described
well.
INTRODUCTION
Rationale 3 Describe the rationale for the review in the context of what is already known. Well describe,
However in introduction section the authors
wrote “Susie Linder-Pelz (1982, p. 580)
defines patient satisfaction as positive
evaluations of distinct dimensions of health
care” but in reference there is no article
authored by Susie Linder-Pelz.
Objectives 4 Provide an explicit statement of questions being addressed with reference to Not clearly
participants, interventions, comparisons, outcomes, and study design
(PICOS).
METHODS
Protocol and registration 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Well describe in method section
Web address), and, if available, provide registration information including
registration number.
Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report Well describe
characteristics (e.g., years considered, language, publication status) used as
criteria for eligibility, giving rationale.
Information sources 7 Describe all information sources (e.g., databases with dates of coverage, Not clearly describe in method section
contact with study authors to identify additional studies) in the search and
date last searched.
Search 8 Present full electronic search strategy for at least one database, including Well describe
any limits used, such that it could be repeated.
Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in Well describe
systematic review, and, if applicable, included in the meta-analysis).
Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, The data collection process was clearly
independently, in duplicate) and any processes for obtaining and confirming describe
data from investigators.
Data items 11 List and define all variables for which data were sought (e.g., PICOS, Not clearly
funding sources) and any assumptions and simplifications made.
Risk of bias in individual 12 Describe methods used for assessing risk of bias of individual studies NA
studies (including specification of whether this was done at the study or outcome
level), and how this information is to be used in any data synthesis.
Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). NA
Synthesis of results 14 Describe the methods of handling data and combining results of studies, if Method of handling data was done by 1)
done, including measures of consistency (e.g., I2) for each meta-analysis. Grouping the effective factors which
included, 2) Extracting main findings
Section/topic # Checklist item Reported on page #
Risk of bias across studies 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., NA
publication bias, selective reporting within studies).
Additional analyses 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta- Not clearly
regression), if done, indicating which were pre-specified.
RESULTS
Study selection 17 Give numbers of studies screened, assessed for eligibility, and included in the review, Flow diagram is well
with reasons for exclusions at each stage, ideally with a flow diagram. describe
Study characteristics 18 For each study, present characteristics for which data were extracted (e.g., study size, Describe systematically in
PICOS, follow-up period) and provide the citations. the table 1.
Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level Not clearly
assessment (see item 12).
Results of individual studies 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple Well describe in the section
summary data for each intervention group (b) effect estimates and confidence and table 2
intervals, ideally with a forest plot.
Synthesis of results 21 Present results of each meta-analysis done, including confidence intervals and The main finding from each
measures of consistency. study is well describe,
however the authors did not
describe about the
confidence interval.
Risk of bias across studies 22 Present results of any assessment of risk of bias across studies (see Item 15). NA
Additional analysis 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, Not clearly
meta-regression [see Item 16]).
DISCUSSION
Summary of evidence 24 Summarize the main findings including the strength of evidence for each main Well describe in discussion
outcome; consider their relevance to key groups (e.g., healthcare providers, users, section and summarize in
and policy makers). the table 2 – table 9.
In most of paragraph of
discussion section, the
authors wrote %x (e.g.,
%20, %45), but that all
have been repaired
Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level The outcome level review-
(e.g., incomplete retrieval of identified research, reporting bias). level, and study implication
are well describe, but the
limitation is not clearly
describe
Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and Well describe
implications for future research.
FUNDING
Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply Well describe
of data); role of funders for the systematic review.
From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(7):
e1000097. doi:10.1371/journal.pmed1000097
JACC: CARDIOVASCULAR INTERVENTIONS VOL. -, NO. -, 2020
ª 2020 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION
PUBLISHED BY ELSEVIER
Mortality After Repeat Revascularization

Following PCI or Coronary Artery Bypass
Grafting for Left Main Disease
The EXCEL Trial
Gennaro Giustino, MD,a Patrick W. Serruys, MD, PHD,b Joseph F. Sabik III, MD,c Roxana Mehran, MD,a,d
Akiko Maehara, MD,d,e John D. Puskas, MD,f Charles A. Simonton, MD,g Nicholas J. Lembo, MD,d,e
David E. Kandzari, MD,h Marie-Claude Morice, MD,i David P. Taggart, MD, PHD,j Anthony H. Gershlick, MD,k
Michael Ragosta III, MD,l Irving L. Kron, MD,l Yangbo Liu, MS,d Zixuan Zhang, MS,d Thomas McAndrew, PHD,d
Ovidiu Dressler, MD,d Philippe Généreux, MD,d,m,n Ori Ben-Yehuda, MD,d,e Stuart J. Pocock, PHD,o
Arie Pieter Kappetein, MD, PHD,p Gregg W. Stone, MDa,d
ABSTRACT
OBJECTIVES The aim of this study was to investigate the incidence and impact on mortality of repeat revascularization
after index percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) for left main coronary
artery disease (LMCAD).
BACKGROUND The impact on mortality of the need of repeat revascularization following PCI or CABG in patients with
unprotected LMCAD is unknown.
METHODS All patients with LMCAD and site-assessed low or intermediate SYNTAX (Synergy Between PCI With Taxus
and Cardiac Surgery) scores randomized to PCI (n ¼ 948) or CABG (n ¼ 957) in the EXCEL (Evaluation of XIENCE Versus
Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization) trial were included. Repeat revascu-
larization events were adjudicated by an independent clinical events committee. The effect of repeat revascularization on
mortality through 3-year follow-up was examined in time-varying Cox regression models.
RESULTS During 3-year follow-up, there were 346 repeat revascularization procedures among 185 patients. PCI was
associated with higher rates of any repeat revascularization (12.9% vs. 7.6%; hazard ratio: 1.73; 95% confidence interval:
1.28 to 2.33; p ¼ 0.0003). Need for repeat revascularization was independently associated with increased risk for 3-year
all-cause mortality (adjusted hazard ratio: 2.05; 95% confidence interval: 1.13 to 3.70; p ¼ 0.02) and cardiovascular
mortality (adjusted hazard ratio: 4.22; 95% confidence interval: 2.10 to 8.48; p < 0.0001) consistently after both PCI
and CABG (pint ¼ 0.85 for both endpoints). Although target vessel revascularization and target lesion revascularization
were both associated with an increased risk for mortality, target vessel non–target lesion revascularization and
non–target vessel revascularization were not.
CONCLUSIONS In the EXCEL trial, repeat revascularization during follow-up was performed less frequently after CABG than
PCI and was associated with increased mortality after both procedures. Reducing the need for repeat revascularization may
further improve long-term survival after percutaneous or surgical treatment of LMCAD. (EXCEL Clinical Trial; NCT01205776)
(J Am Coll Cardiol Intv 2020;-:-–-) © 2020 by the American College of Cardiology Foundation.
From aThe Zena and Michael A. Wiener Cardiovascular Institute, Icahn School of Medicine at Mount Sinai, New York, New York;
b
Imperial College of Science, Technology and Medicine, London, United Kingdom; cDepartment of Surgery, UH Cleveland Medical
Center, Cleveland, Ohio; dClinical Trials Center, Cardiovascular Research Foundation, New York, New York; eNewYork-Presby-
terian Hospital/Columbia University Medical Center, New York, New York; fMount Sinai Heart at Mount Sinai St Luke’s, New York,
New York; gAbbott Vascular, Santa Clara, California; hPiedmont Heart Institute, Atlanta, Georgia; iRamsay Générale de Santé,
ISSN 1936-8798/$36.00 https://doi.org/10.1016/j.jcin.2019.09.019

2 Giustino et al. JACC: CARDIOVASCULAR INTERVENTIONS VOL. -, NO. -, 2020
Repeat Revascularization and Mortality - 2020:-–-
I
ABBREVIATIONS terations in coronary stent technologies, Bypass Surgery for Effectiveness of Left Main Revas-
AND ACRONYMS technique, and pharmacotherapies have cularization) trial.
enhanced the efficacy and safety of
CABG = coronary artery bypass
grafting
percutaneous coronary intervention (PCI), METHODS
leading to lower rates of stent thrombosis,
CI = confidence interval
restenosis, and the need for repeat revascu- STUDY DESIGN. The EXCEL trial was an interna-
HR = hazard ratio
larization (1–6). Outcomes of coronary artery tional, open-label, multicenter, randomized trial that
IQR = interquartile range
bypass grafting (CABG) have also improved compared PCI using cobalt-chromium fluoropolymer-
LM = left main coronary artery
with the use of multiple arterial grafts, mini- based everolimus-eluting stents (XIENCE; Abbott
LMCAD = left main coronary
mally invasive techniques, and optimal med- Vascular, Santa Clara, California) versus CABG in pa-
artery disease
ical therapy (4,7–11). The need for repeat tients with LMCAD. The EXCEL trial design and
MI = myocardial infarction
revascularization is more common after PCI principal results have been previously reported (19).
PCI = percutaneous coronary
than CABG, although the differences be- In brief, inclusion criteria were LM diameter stenosis
intervention
tween the techniques are diminishing over of $70% as estimated visually or stenosis of 50%
TLR = target lesion
revascularization time (12–15). Although often considered a to <70% if hemodynamically significant by noninva-
TVR = target vessel
clinical endpoint of lesser importance sive or invasive testing. All patients were required to
revascularization compared with death, stroke, or myocardial have low or intermediate anatomic complexity of
infarction (MI), the need for repeat revascu- coronary artery disease, as defined by a site-
larization is associated with worse quality of life and determined SYNTAX (Synergy Between PCI With
exposes patients to new hospitalizations and proce- Taxus and Cardiac Surgery) score of #32. Consensus
dural risks (13,16–18). In addition, the need for a among the members of the heart team regarding the
repeat procedure after revascularization of the left eligibility for revascularization with either PCI or
main coronary artery (LM) may be associated with CABG was required. Clinical follow-up was performed
substantial morbidity and mortality given the large at 1 month, 6 months, and 1 year and then annually
amount of subtended myocardium at risk (19). We through 5 years. At the time of the present analysis,
therefore sought to characterize the incidence, pre- all patients had completed 3 years of follow-up. The
dictors, and consequences of the need for repeat primary endpoint of the EXCEL trial was the com-
revascularization after the PCI or CABG for LM coro- posite of death of any cause, stroke, or MI at a median
nary artery disease (LMCAD) using contemporary de- follow-up time of 3 years. Major powered secondary
vices and surgical techniques from the EXCEL endpoints included this composite endpoint at
(Evaluation of XIENCE Versus Coronary Artery 30 days and the composite of death, stroke, MI, or
Hopital Privé Jacques Cartier, Massy, France; jDepartment Cardiac Surgery, John Radcliffe Hospital, Oxford, United Kingdom;
k
University Hospitals of Leicester, Leicester, United Kingdom; lDivision of Cardiovascular Medicine, University of Virginia Health
m
System, Charlottesville, Virginia; Gagnon Cardiovascular Institute, Morristown Medical Center, Morristown, New Jersey; nHô-
pital du Sacré-Coeur de Montréal, Montréal, Québec, Canada; oDepartment of Medical Statistics, London School of Hygiene and
Tropical Medicine, London, United Kingdom; and the pErasmus University Medical Center, Rotterdam, the Netherlands. The
EXCEL trial was sponsored by Abbott Vascular. Dr. Giustino is a consultant for Bristol-Myers Squibb/Pfizer. Dr. Serruys is a
consultant for Abbott, Biosensors, Medtronic, Micell, Philips/Volcano, Xeltis, and HeartFlow. Dr. Sabik is a consultant for Med-
tronic, Technologies, SINOMED Edwards, and Sorin; and is an advisory board member for Medtronic Cardiac Surgery. Dr. Mehran
has received institutional research grant support from Eli Lilly/Daiichi Sankyo, Bristol-Myers Squibb, AstraZeneca, The Medicines
Company, OrbusNeich, Bayer, CSL Behring, Abbott Laboratories, Watermark Research Partners, Novartis Pharmaceuticals,
Medtronic, AUM Cardiovascular, and Beth Israel Deaconess Medical Center; is an executive committee member for Janssen
Pharmaceuticals and Osprey Medical; is a data and safety monitoring board member for Watermark Research Partners; is a
consultant for Medscape, The Medicines Company, Boston Scientific, Merck, Cardiovascular Systems, Sanofi, Shanghai BraccoSine
Pharmaceutical, and AstraZeneca; and holds equity in Claret Medical and Elixir Medical. Dr. Maehara has received institutional
grant support from Boston Scientific and Abbott; is a consultant for Boston Scientific and OCT Medical Imaging; and has received
speaking fees from Abbott. Dr. Simonton is an employee of Abbott Vascular. Dr. Lembo is a consultant and member of the
Speakers Bureau for Abbott Vascular, Boston Scientific, and Medtronic. Dr. Kandzari has received consulting honoraria from
Medtronic, Biotronik, and Boston Scientific; and has received research and grant support from Medtronic, Biotronik, and Boston
Scientific. Dr. Genereux has received speaking fees from Edwards Lifesciences, Medtronic, Tryton Medical, Cardinal Health, and
Cardiovascular Systems; has received consulting fees from Boston Scientific, Cardiovascular Systems, and Pi-Cardia; has received
institutional research grant from Boston Scientific; and holds equity in SIG.NUM, SoundBite Medical Solutions, Saranas, and Pi-
Cardia. Dr. Pocock is a consultant for Abbott Vascular. Dr. Kappetein is an employee of Medtronic. Dr. Stone has received speaking
honoraria from Terumo and Amaranth; and is a consultant to Reva. All other authors have reported that they have no relation-
ships relevant to the contents of this paper to disclose.
Manuscript received August 15, 2019; revised manuscript received September 4, 2019, accepted September 10, 2019.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. -, NO. -, 2020 Giustino et al. 3
- 2020:-–- Repeat Revascularization and Mortality
T A B L E 1 Rates of Time to First Repeat Revascularization With Percutaneous Coronary Intervention Versus Coronary Artery Bypass Grafting for Left Main Coronary
Artery Disease During 3 Years of Follow-Up
1 Year 3 Years
PCI CABG Hazard Ratio PCI CABG Hazard Ratio

(n ¼ 948) (n ¼ 957) (95% CI) p Value (n ¼ 948) (n ¼ 957) (95% CI) p Value
Any revascularization 6.9% (64) 4.6% (42) 1.51 (1.02–2.22) 0.04 12.9% (117) 7.6% (68) 1.73 (1.28–2.33) 0.0003
Revascularization with PCI 5.9% (54) 3.8% (35) 1.53 (1.00–2.33) 0.05 10.7% (97) 6.8% (61) 1.59 (1.16–2.19) 0.004
Revascularization with CABG 1.6% (15) 0.8% (7) 2.12 (0.86–5.20) 0.09 3.3% (30) 0.8% (7) 4.25 (1.87–9.68) 0.0002
Target vessel revascularization 6.3% (58) 4.2% (39) 1.47 (0.98–2.20) 0.06 11.0% (100) 7.1% (64) 1.56 (1.14–2.14) 0.005
Target lesion revascularization 5.4% (50) 4.0% (37) 1.33 (0.87–2.03) 0.19 9.4% (85) 6.8% (61) 1.38 (1.00–1.92) 0.052
Non–target lesion revascularization 1.8% (17) 0.2% (2) 8.45 (1.95–36.59 0.0006 3.3% (30) 0.7% (6) 5.00 (2.08–12.00) <0.0001
Non–target vessel revascularization 0.9% (8) 0.4% (4) 1.98 (0.60–6.57) 0.26 2.7% (24) 0.8% (7) 3.40 (1.47–7.89) 0.002
Ischemia-driven revascularization 6.8% (63) 4.4% (40) 1.56 (1.05–2.31) 0.03 12.7% (115) 7.5% (67) 1.73 (1.28–2.33) 0.0003
Revascularization with CABG 1.6% (15) 0.8% (7) 2.12 (0.86–5.20) 0.09 3.3% (30) 0.8% (7) 4.25 (1.87–9.68) 0.0002
Non–target lesion revascularization 1.8% (17) 0.1% (1) 16.9 (2.25–126.86) 0.0002 3.2% (29) 0.6% (5) 5.80 (2.24–14.97) <0.0001
Non–target vessel revascularization 0.8% (7) 0.4% (4) 1.73 (0.51–5.90) 0.38 2.5% (22) 0.8% (7) 3.11 (1.33–7.28) 0.006
Non ischemia-driven revascularization 0.5% (5) 0.2% (2) 2.47 (0.48–12.72) 0.26 1.0% (9) 0.3% (3) 2.96 (0.80–10.93) 0.09
Revascularization with CABG 0.1% (1) 0.0% (0) — 0.32 0.2% (2) 0.0% (0) — 0.16
Non-target lesion revascularization 0.0% (0) 0.1% (1) — 0.31 0.0% (0) 0.1% (1) — 0.31
Non-target vessel revascularization 0.1% (1) 0.0% (0) — 0.32 0.2% (2) 0.0% (0) — 0.16
Values are Kaplan-Meier estimated rate (number of events).

CABG ¼ coronary artery bypass grafting; CI ¼ confidence interval; PCI ¼ percutaneous coronary intervention.
ischemia-driven revascularization at 3 years. Defini- of each type of repeat revascularization event on

tions of the primary and major secondary endpoints all-cause, cardiovascular, and noncardiovascular
are reported elsewhere (19). Study monitors collected mortality at 3-year follow-up.
source documents of all primary and secondary end-
points for adjudication by an independent clinical STATISTICAL ANALYSIS. All analyses were per-
events committee. The extent and complexity of formed in the intention-to-treat population. Cate-
coronary artery disease and the SYNTAX score at gorical variables were compared using the chi-square
baseline were also assessed by an independent test or Fisher exact test. Continuous variables were
angiographic core laboratory. The investigation was compared using Student’s t-test or the Wilcoxon
approved by the ethics committee or Institutional rank sum test for skewed data. Event rates were on
Review Board at each center, and all patients pro- the basis of Kaplan-Meier estimates in time-to-first-
vided informed consent. event analyses and were compared using the log-
The present study is a secondary analysis from the rank test. Hazard ratios (HRs) with 95% confidence
EXCEL trial investigating the incidence, risk factors, intervals (CIs) for PCI versus CABG were generated
and prognostic impact of the performance of repeat using Cox regression models. Predictors of repeat
revascularization procedures following PCI and revascularization events were evaluated with multi-
CABG. The following type of adjudicated revascular- variate Cox regression models separately for patients
ization events were considered in this analysis: randomized to PCI or CABG, including clinical,
ischemia-driven revascularization, non-ischemia- angiographic, and procedural characteristics that
driven revascularization, target lesion revasculariza- were significantly associated with the outcome by
tion (TLR), target vessel revascularization (TVR), univariate analysis or were deemed to be clinically
target-vessel non-TLR, non-TVR, repeat revasculari- important for each type of index procedure (full list
zation with PCI, and repeat revascularization with of covariates for each model is included in the
CABG. A complete list of definitions for the different footnote of the respective table). The association of
types of repeat revascularization endpoints is re- repeat revascularization with the risk for mortality at
ported in Online Table 1. We evaluated the effect 3 years was evaluated using multivariate Cox
F I G U R E 1 Kaplan-Meier Curves for Repeat Revascularization Within 3 Years After Percutaneous Coronary Intervention or Coronary Artery Bypass Grafting
(A) Any repeat revascularization. (B) Target lesion revascularization. (C) Target vessel revascularization. (D) Target vessel non–target lesion revascularization.
(E) Non–target vessel revascularization. CABG ¼ coronary artery bypass grafting; CI ¼ confidence interval; PCI ¼ percutaneous coronary intervention.
regression models entering repeat revascularization, congestive heart failure, anemia, and ST-segment
any MI, and any stroke as time-varying covariates elevation MI or non–ST-segment elevation MI at
alongside other baseline covariates, including age, presentation. Two-sided p values #0.05 were
sex, SYNTAX score, diabetes, chronic kidney disease, considered to indicate statistical significance. All
analyses were performed using SAS version 9.4

T A B L E 2 Independent Predictors of Any Repeat Revascularization Within 3 Years After
(SAS Institute, Cary, North Carolina). Percutaneous Coronary Intervention or Coronary Artery Bypass Grafting for Left Main
Coronary Artery Disease
RESULTS
Adjusted Hazard Ratio 95% CI p Value
PCI group*
During a median follow-up time of 3 years (inter- Body mass index, per unit increase 1.04 1.00–1.07 0.04
quartile range [IQR]: 3 to 3 years), there were 346 Diabetes mellitus
repeat revascularization procedures among 185 pa- No diabetes mellitus 1.00 (reference) — —
tients (Online Table 2). Of these, 259 of 346 (74.9%) Without insulin treatment 1.19 0.76–1.86 0.45
were PCI procedures and 87 of 346 (25.1%) were CABG With insulin treatment 1.96 1.10–3.51 0.02
Hemodynamic support during the procedure 2.37 1.29–4.35 0.005
procedures. Overall, 102 patients (55.1%) underwent 1
Use of statin at discharge 0.30 0.16–0.58 0.0003
repeat revascularization procedure, 41 (22.2%) un-
CABG group†
derwent 2 procedures, and 42 (22.7%) underwent >2
Age, per 10-yr increase 0.71 0.55–0.92 0.01
events. The median time to the first repeat revascu- Female 1.64 0.94–2.86 0.08
larization procedure was 320 days (IQR: 141 to Peripheral vascular disease 2.14 1.05–4.35 0.04
616 days). Baseline clinical, angiographic, and proce-
Adjusted hazard ratios and 95% CIs were generated using multivariate Cox regression analysis. Only the cova-
dural characteristics in patients with versus without riates significantly associated with the outcome are displayed. *This model included the following covariates:
any repeat revascularization procedures after the in- age, sex, body mass index, diabetes mellitus, left main distal segment or bifurcation lesions, use of intravascular
ultrasound imaging, use of hemodynamic support during the procedure, core laboratory–assessed SYNTAX score,
dex PCI or CABG are reported in Online Tables 3 to 5. number of diseased non–left main vessels, and use of statin at discharge. †This model included the following
covariates: age, sex, body mass index, diabetes mellitus, hyperlipidemia, peripheral vascular disease, clinical
There were no significant differences in SYNTAX presentation with an acute coronary syndrome, core laboratory–assessed SYNTAX score, and number of arterial
score between patients with versus without repeat conduits used.
SYNTAX ¼ Synergy Between PCI With Taxus and Cardiac Surgery; other abbreviations as in Table 1.
revascularization at 3 years within both the PCI and
the CABG groups (Online Table 4). Medication use
over 3 years is reported in Online Table 6. Patients
who required repeat revascularization were more the 3-year follow-up was performed more frequently
likely to remain on dual-antiplatelet therapy through in patients randomized to initial PCI compared with
3 years within both the PCI and the CABG groups. CABG (3.3% vs. 0.8%; HR: 4.25; 95% CI: 1.87 to
Patients who required repeat revascularization had 9.68; p ¼ 0.0002).
higher rates of anginal symptoms at 3 years in both
the PCI and CABG arms (Online Table 7).
T A B L E 3 Predictors of All-Cause and Cardiovascular Mortality at 3 Years After
Percutaneous Coronary Intervention or Coronary Artery Bypass Grafting for Left Main
RISK FOR REVASCULARIZATION BY PCI AND CABG.
Coronary Artery Disease
Median time to the first repeat revascularization was
347 days (IQR: 182 to 570 days) after PCI and 257 days Adjusted 95% Confidence
Hazard Ratio Interval p Value
(IQR: 83 to 628 days) after CABG (p ¼ 0.13). Rates of
All-cause mortality (128 events)
time to first repeat revascularization over 3 years are
Any repeat revascularization* 2.05 1.13–3.70 0.02
reported in Table 1 and Figures 1A to 1E. Patients Any myocardial infarction* 4.03 2.43–6.67 <0.0001
assigned to PCI had higher rates of any repeat revas- Any stroke* 16.62 9.97–27.69 <0.0001
cularization at 3 years compared with those assigned Age, per 10-yr increase 1.39 1.10–1.77 0.006
to CABG (12.9% vs. 7.6%; HR: 1.73; 95% CI: 1.28 to Diabetes mellitus 1.69 1.17–2.44 0.005
2.33; p ¼ 0.0003). There were no significant differ- Anemia 2.15 1.45–3.18 0.0001
Cardiovascular mortality
ences between PCI and CABG in the rates of repeat
(74 events)
revascularization at 6 months (3.1% vs. 3.2%; HR:
Any repeat revascularization* 4.22 2.10–8.48 <0.0001
0.98; 95% CI: 0.59 to 1.63; p ¼ 0.93). Most of the dif- Any myocardial infarction* 5.30 2.86–9.83 <0.0001
ferences between the 2 strategies in the rates of Any stroke* 31.11 17.10–56.61 <0.0001
Age, per 10-yr increase 1.45 1.06–2.00 0.02
repeat revascularization emerged beyond 6 months
Congestive heart failure 2.04 1.04–4.00 0.002
(Online Figures 1 and 2) (4.4% vs. 9.9%; HR: 2.33; Anemia 2.27 1.35–3.81 0.04
95% CI: 1.59 to 3.41; p < 0.0001). The cause of repeat Diabetes mellitus 1.55 0.96–2.50 0.07
revascularization was stent thrombosis in 8 of 117

Adjusted hazard ratios and 95% confidence intervals were generated using multivariate Cox regression analysis.
patients (7.1%) after PCI and symptomatic graft oc- *Modeled as a time-varying covariate within the Cox regression model. The multivariate Cox regression model
included the following covariates: any repeat revascularization, any myocardial infarction, any stroke, age, sex,
clusion in 42 of 68 patients (62.7%) after CABG
diabetes, anemia, congestive heart failure, chronic kidney disease, ST-segment elevation myocardial infarction or
(p < 0.0001). Most repeat revascularizations were non–ST-segment elevation myocardial infarction as clinical presentation, core laboratory–assessed SYNTAX score,
and randomized assignment to PCI or CABG.
performed with PCI in both the PCI and CABG groups.
Abbreviations as in Tables 1 and 2.
Overall, repeat revascularization with CABG during
F I G U R E 2 Early and Late Risk for Mortality After Any Repeat Revascularization in the Overall Population
(A) All-cause mortality. (B) Cardiovascular mortality.
PREDICTORS OF REPEAT REVASCULARIZATION. cardiovascular mortality (adjusted HR: 4.22; 95% CI:
Predictors of any repeat revascularization at 3 years 2.10 to 8.48; p < 0.0001) but not noncardiovascular
after PCI or CABG are reported in Table 2. Higher body mortality (Online Tables 8 and 9). However, the
mass index, insulin-treated diabetes, and hemody- magnitude of the association between repeat revas-
namic support during the procedure were associated cularization and all-cause mortality was smaller
with a higher risk for repeat revascularization after compared with that of MI (adjusted HR: 4.03; 95% CI:
PCI, while statin use at discharge was protective 2.43 to 6.67; p < 0.0001) or stroke (adjusted HR: 16.62;
(adjusted HR: 0.30; 95% CI: 0.16 to 0.50; p ¼ 0.0003). 95% CI: 9.97 to 27.69; p < 0.0001). Of note, the risk for
Younger age, female sex, and peripheral vascular mortality after repeat revascularization peaked
disease were independent predictors of repeat within 30 days and then declined over time (Figure 2).
revascularization after CABG. The adjusted risk for 3-year all-cause and cardiovas-
cular mortality according to the subtypes of repeat
REPEAT REVASCULARIZATION AND MORTALITY. revascularization events is illustrated in Figure 3. TVR
At 3 years, there were 128 all-cause deaths, including and TLR were both associated with increased all-
74 cardiovascular deaths and 54 noncardiovascular cause and cardiovascular mortality. Conversely,
deaths. Independent predictors of all-cause and car- both target-vessel non-TLR and non-TVR were not
diovascular mortality at 3 years in the overall popu- associated with increased all-cause and cardiovascu-
lation are reported in Table 3. The need for repeat lar mortality. Of note, the need for repeat revascu-
revascularization was independently associated with larization using CABG was strongly associated with
increased risk for both all-cause mortality (adjusted increased all-cause mortality. The effect of repeat
HR: 2.05; 95% CI: 1.13 to 3.70; p ¼ 0.02) and revascularization on mortality according to the
F I G U R E 3 Association Between Type of Repeat Revascularization and Mortality Within 3 Years in the Overall Population
(A) All-cause mortality. (B) Cardiovascular mortality. CABG ¼ coronary artery bypass grafting; CI ¼ confidence interval; HR ¼ hazard ratio; PCI ¼ percutaneous coronary
intervention.
F I G U R E 4 Association Between Type of Repeat Revascularization and Mortality Within 3 Years After Percutaneous Coronary Intervention or
Coronary Artery Bypass Grafting
(A) All-cause mortality. (B) Cardiovascular mortality. CABG ¼ coronary artery bypass grafting; CI ¼ confidence interval; PCI ¼ percutaneous coronary intervention.
F I G U R E 5 Kaplan-Meier Curves for All-Cause and Cardiovascular Mortality After Any Repeat Revascularization
(A) All-cause mortality. (B) Cardiovascular mortality.
index revascularization strategy is illustrated in DISCUSSION

Figures 4 and 5. The effect of any repeat revasculari-
zation and its subtypes on both all-cause and car- The major findings of the present analysis from the
diovascular mortality was consistent in patients EXCEL trial in which we evaluated the incidence,
undergoing initial PCI and CABG, without evidence timing, and consequences of the need for repeat
of interaction. coronary revascularization following PCI or CABG for
C E NT R AL IL L U STR AT IO N Repeat Revascularization and Mortality After Percutaneous Coronary Intervention or

Coronary Artery Bypass Grafting for Left Main Coronary Artery Disease
7.6% underwent repeated Impact on Mortality of Repeat Revascularization

revascularization within 3 years
3-Year All-Cause Mortality
9.3%
CABG Any repeat revascularization p = 0.02
(N = 957)
90.7%
After index PCI
pint = 0.85
PCI Redo-CABG After index CABG
0.1 1 10 100
p = 0.0003 favoring CABG
3-Year Cardiovascular Mortality
Any repeat revascularization p < 0.0001

12.9% underwent repeated
revascularization within 3 years
After index PCI
21.1% pint = 0.85
PCI After index CABG
(N = 948)
79.6%
0.1 1 10 100
Redo-PCI CABG
Adjusted HR (95% CI)
Log Scale
Giustino, G. et al. J Am Coll Cardiol Intv. 2020;-(-):-–-.
After percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) for left main coronary artery disease and low or intermediate anatomic
complexity, the need for repeat revascularization was more common after PCI compared with CABG (left). Of note, the need for repeat revascularization was
independently associated with increased risk for both all-cause and cardiovascular mortality consistently after both PCI and CABG (right). The lower rate of repeat
revascularization after CABG compared with PCI may be one factor contributing to the long-term benefits of surgical revascularization. CI ¼ confidence interval;
HR ¼ hazard ratio.
LMCAD are as follows (Central Illustration): 1) repeat Although developments in technologies, tech-
revascularization procedures within 3 years were nique, and pharmacotherapies have enhanced both
performed more commonly after PCI than CABG, the efficacy and safety of revascularization, the need
mostly beyond the first 6 months after the index for repeat revascularization remains a frequent
procedure (of note, the need for revascularization adverse event after both PCI and CABG (1–5,20).
after PCI was infrequently due to stent thrombosis, Following PCI, stent-related complications or the
while after CABG repeat revascularization was most development of new obstructive native coronary le-
often prompted by symptomatic graft occlusion); 2) sions remote from the stented vascular segment
the performance of any repeat revascularization usually motivate most repeat revascularization pro-
procedure was an independent predictor of a subse- cedures (1–5). After CABG, the need for repeat revas-
quent increase in all-cause and cardiovascular mor- cularization is generally driven by progression of
tality within 3 years after both PCI and CABG, a risk native vessel disease distal to the site of anastomosis
that peaked within 30 days after the repeat procedure or by arterial or venous graft occlusion (1–5,20). Few
and then declined over time (repeat revascularization studies have previously examined in detail the timing
was not associated with increased risk for noncardiac and impact on mortality of the need for repeat
mortality); and 3) the magnitude and significance of revascularization after an index revascularization
the association between repeat revascularization and with PCI or CABG. In the era of first-generation drug-
mortality depended upon its subtype, with TVR and eluting stents, this subject was examined in a report
TLR being strongly associated with increased risk for from the SYNTAX trial in which 1,800 patients with
all-cause mortality, whereas target vessel non-TLR triple-vessel disease and/or LMCAD were randomized
and non-TVR were not associated with increased all- to CABG or PCI with paclitaxel-eluting stents (16). In
cause mortality. this study, PCI was associated with higher risk for
repeat revascularization at 5 years (13,17); repeat consistent with the larger amount of myocardium in
revascularization was an independent predictor of jeopardy after failed LM revascularization (16). The
the composite of death, stroke, or MI after initial PCI effect on mortality following repeat revasculariza-
but not after initial CABG, a finding driven mostly by tion in the present study was greater early after the
an increased risk of MI (16). event (within 30 days) and then attenuated over
In the present analysis from the EXCEL trial, we time, suggesting that the actual event of repeat
extend these prior observations to a larger LMCAD revascularization per se was associated with
cohort with low or intermediate anatomic complexity increased risk.
treated with contemporary PCI devices and CABG The association between repeat revascularization
techniques. Consistent with prior studies, the rates of and mortality is likely multifactorial and may be both
repeat revascularization were lower after CABG than causative and associative in nature. First, the need for
PCI, possibly related to the protective effect from repeat revascularization exposes patients to new
progressive atherosclerosis developing proximally to hospitalizations with its integral risks. Second, every
the surgically anastomosed segment. Conversely, revascularization procedure carries risk; in this re-
because PCI treats a only a focal target coronary gard, mortality was significantly greater after repeat
lesion, the rate of subsequent repeat revasculariza- revascularization by CABG but not after PCI, reflect-
tion will be influenced by both the complexity of the ing inherent differences in the risks of these 2 stra-
actual lesion affecting stent-related events and from tegies. This observation suggests that CABG should be
the development of new lesions upstream or down- reserved for repeat revascularization procedures that
stream from the stented vascular segment (non-stent- are not amenable to repeat PCI, irrespective of the
related events) (21–23). However, it has also been initial revascularization approach. Third, prolonged
shown that repeat revascularization procedures are dual-antiplatelet therapy after repeat revasculariza-
performed for less severe anginal symptoms and tion is associated with increased bleeding and, in
health status deterioration after PCI compared with some reports, mortality (28,29). Finally, the need for
CABG, which may reflect differences in the threshold repeat revascularization could represent a marker of
for or anatomic suitability of further revascularization more extensive coronary artery disease and comor-
after each procedure (13,17,24). This differential bidity burden; however, the baseline SYNTAX score
threshold may in part explain the more frequent use did not differ between patients who did and did not
of repeat revascularization after PCI compared with require repeat revascularization.
CABG (13,17,24). Of note, the absolute differences in Of note, both TVR and TLR were significantly
the rates of repeat revascularization between PCI and associated with an increased risk for all-cause and
CABG in the EXCEL trial were smaller than from the cardiovascular mortality, consistently after both PCI
SYNTAX trial, which may reflect the lower anatomic and CABG. Following LM PCI, repeat revasculariza-
complexity of the EXCEL population as well as the tion of a previously stented unprotected LM lesion
greater safety and efficacy of contemporary secondary to drug-eluting stent failure (e.g., in-stent
everolimus-eluting stents compared with paclitaxel- restenosis or stent thrombosis) is inherently associ-
eluting stents. Also, despite the slightly greater ated with poor prognosis because of the large area of
rates of revascularization in the PCI arm in EXCEL, subtended myocardium at risk. Similarly, after CABG,
the overall health status, quality of life, and freedom after graft failure either repeat revascularization
from angina at 3 years after PCI and CABG were not through PCI of a diseased graft or of the native
significantly different in this trial (25), in contrast to occluded coronary artery may be associated with
prior reports (26,27). adverse events (30,31). Finally, non-TVR (which in
By time-varying multivariate analysis, repeat this trial most commonly consisted of revasculariza-
revascularization was associated with an increased tion of the right coronary artery) and non–target
risk for both all-cause and cardiovascular mortality lesion TVR (in this trial including lesions distal to
through 3-year follow-up irrespective of the index the LM complex within the left anterior descending
revascularization strategy, although its impact was coronary artery and left circumflex coronary artery
smaller than that of a stroke or an MI. However, the territories) were not associated with an increased risk
adjusted hazard of mortality after TLR following for all-cause or cardiovascular mortality after either
initial LMCAD revascularization in EXCEL (HR: 2.47; PCI or CABG. Considering the lower periprocedural
95% CI: 1.34 to 4.55) was higher than that observed morbidity of PCI compared with CABG, the present
after PCI from a large drug-eluting stent database of analysis suggests that new approaches to reduce the
non-LM PCI (HR: 1.22; 95% CI: 1.03 to 1.45), need for stent-related repeat revascularization may
further improve the benefit/risk profile of PCI

compared with CABG. ADDRESS FOR CORRESPONDENCE: Dr. Gregg W.
Stone, The Zena and Michael A. Wiener Cardiovas-
STUDY LIMITATIONS. First, because this was a sec- cular Institute, Icahn School of Medicine at Mount
ondary analysis from a randomized controlled trial, Sinai, Cardiovascular Research Foundation, 1700
our findings should be considered hypothesis- Broadway, 9th Floor, New York, New York 10019.
generating. Second, detailed causes for repeat revas- E-mail: gregg.stone@mountsinai.org.
cularization were not prospectively collected; how-
ever, most repeat revascularization events were PERSPECTIVES
adjudicated as ischemia-driven. Third, at the time of
this analysis, only 3 years of follow-up were available; WHAT IS KNOWN? PCI is known to be associated
longer term follow-up (currently planned for 5 years) with higher risk for repeat revascularization compared
may demonstrate further differences between PCI with CABG. However, its prognostic significance after
and CABG in the rates of repeat revascularization and revascularization of unprotected LMCAD remains
their prognostic significance. Fourth, bias from re- unclear.
sidual confounding in our multivariate models
evaluating the association between repeat revascu- WHAT IS NEW? The need for repeat revasculariza-
larization and mortality cannot be excluded. tion after both PCI and CABG for unprotected LMCAD
was associated with increased mortality over 3 years,
although with an associated risk that was smaller than
CONCLUSIONS
that of MI or stroke. Of note, repeat revascularization
of the index target coronary vessel (e.g., LM or
Our findings in an unprotected LMCAD population
proximal left anterior descending coronary artery) was
suggest that the need for and performance of repeat
associated with increased mortality, whereas repeat
revascularization procedures have prognostic impli-
revascularization of the non-target vessel (e.g., right
cations, the magnitude of which depends on its
coronary artery) was not. Moreover, mortality was
indication and type of repeat revascularization. The
substantially greater after repeat revascularization by
lower rate of repeat revascularization after CABG
CABG than after PCI.
compared with PCI may be one factor contributing to
the favorable long-term prognosis of surgical revas-
WHAT IS NEXT? Measures to reduce the need for
cularization seen in some prior trials. It is also
repeat revascularization, including improved stent
plausible that measures to reduce repeat revascular-
platforms and implantation technique, use of pan-
ization, including improved drug-eluting stents and
arterial bypass grafting, and aggressive risk factor
implantation techniques, use of pan-arterial bypass
control with guideline-directed medical therapy, may
grafting, and aggressive risk factor control with
improve prognosis after both PCI and CABG.
optimal medical therapy, may improve prognosis af-
ter both PCI and CABG.
REFERENCES
1. Dangas GD, Claessen BE, Caixeta A, Sanidas EA, 4. Piccolo R, Giustino G, Mehran R, Windecker S. 9. Gaudino M, Antoniades C, Benedetto U, et al.
Mintz GS, Mehran R. In-stent restenosis in the Stable coronary artery disease: revascularisation Mechanisms, consequences, and prevention of
drug-eluting stent era. J Am Coll Cardiol 2010;56: and invasive strategies. Lancet 2015;386:702–13. coronary graft failure. Circulation 2017;136:
1897–907. 1749–64.
5. Palmerini T, Biondi-Zoccai G, Della Riva D, et al.
2. Giustino G, Baber U, Aquino M, et al. Safety and Stent thrombosis with drug-eluting and bare-metal 10. Modolo R, Chichareon P, Kogame N, et al.
efficacy of new-generation drug-eluting stents in stents: evidence from a comprehensive network Contemporary outcomes following coronary artery
women undergoing complex percutaneous coro- meta-analysis. Lancet 2012;379:1393–402. bypass graft surgery for left main disease. J Am
nary artery revascularization: from the WIN-DES Coll Cardiol 2019;73:1877–86.
collaborative patient-level pooled analysis. J Am 6. Giustino G, Chieffo A, Palmerini T, et al. Efficacy
11. Doenst T, Haverich A, Serruys P, et al. PCI and
Coll Cardiol Intv 2016;9:674–84. and safety of dual antiplatelet therapy after
CABG for treating stable coronary artery disease:
complex PCI. J Am Coll Cardiol 2016;68:1851–64.
JACC review topic of the week. J Am Coll Cardiol
3. Giustino G, Baber U, Salianski O, et al. Safety
7. Gaudino M, Taggart D, Suma H, Puskas JD, 2019;73:964–76.
and efficacy of new-generation drug-eluting
Crea F, Massetti M. The choice of conduits in
stents in women at high risk for athero- 12. Capodanno D, Gargiulo G, Buccheri S, et al.
coronary artery bypass surgery. J Am Coll Cardiol
thrombosis: from the Women in Innovation and Computing methods for composite clinical end-
2015;66:1729–37.
Drug-Eluting Stents collaborative patient-level points in unprotected left main coronary artery
pooled analysis. Circ Cardiovasc Interv 2016;9: 8. Alexander JH, Smith PK. Coronary-artery revascularization: a post hoc analysis of the DELTA
e002995. bypass grafting. N Engl J Med 2016;375:e22. registry. J Am Coll Cardiol Intv 2016;9:2280–8.
13. Kazi DS, Hlatky MA. Repeat revascularization is after drug-eluting stent implantation in women 27. Cohen DJ, Van Hout B, Serruys PW, et al.
a faulty end point for clinical trials. Circ Cardiovasc versus men: a sex-specific patient-level pooled- Quality of life after PCI with drug-eluting stents or
Qual Outcomes 2012;5:249–50. analysis of six randomized trials. Catheter Car- coronary-artery bypass surgery. N Engl J Med
diovasc Interv 2017;89:178–89. 2011;364:1016–26.
14. Kip KE, Hollabaugh K, Marroquin OC,
Williams DO. The problem with composite end 21. Fihn SD, Blankenship JC, Alexander KP, et al. 28. Palmerini T, Benedetto U, Bacchi-Reggiani L,
points in cardiovascular studies: the story of major 2014 ACC/AHA/AATS/PCNA/SCAI/STS focused et al. Mortality in patients treated with extended
adverse cardiac events and percutaneous coronary update of the guideline for the diagnosis and duration dual antiplatelet therapy after drug-
intervention. J Am Coll Cardiol 2008;51:701–7. management of patients with stable ischemic eluting stent implantation: a pairwise and
heart disease: a report of the American College of Bayesian network meta-analysis of randomised
15. Stolker JM, Spertus JA, Cohen DJ, et al.
Cardiology/American Heart Association Task Force trials. Lancet 2015;385:2371–82.
Rethinking composite end points in clinical trials:
on Practice Guidelines, and the American Associ-
insights from patients and trialists. Circulation
ation for Thoracic Surgery, Preventive Cardiovas- 29. Giustino G, Baber U, Sartori S, et al. Duration
2014;130:1254–61.
cular Nurses Association, Society for of dual antiplatelet therapy after drug-eluting
16. Parasca CA, Head SJ, Milojevic M, et al. Inci- Cardiovascular Angiography and Interventions, stent implantation: a systematic review and
dence, characteristics, predictors, and outcomes of and Society of Thoracic Surgeons. Circulation meta-analysis of randomized controlled trials.
repeat revascularization after percutaneous coro- 2014;130:1749–67. J Am Coll Cardiol 2015;65:1298–310.
nary intervention and coronary artery bypass
22. Neumann FJ, Sousa-Uva M, Ahlsson A, et al. 30. Brilakis ES, Rao SV, Banerjee S, et al. Percu-
grafting: the SYNTAX trial at 5 years. J Am Coll
2018 ESC/EACTS guidelines on myocardial revas- taneous coronary intervention in native arteries
Cardiol Intv 2016;9:2493–507.
cularization. Eur Heart J 2019;40:87–165. versus bypass grafts in prior coronary artery
17. Arnold SV, Magnuson EA, Wang K, et al. Do bypass grafting patients: a report from the Na-
23. Taggart DP. Percutaneous or surgical revas-
differences in repeat revascularization explain the tional Cardiovascular Data Registry. J Am Coll
cularization in multivessel coronary artery disease:
antianginal benefits of bypass surgery versus Cardiol Intv 2011;4:844–50.
synthesis from SYNTAX. Eur Heart J 2014;35:
percutaneous coronary intervention?: implications
2789–91. 31. Brilakis ES, O’Donnell CI, Penny W, et al.
for future treatment comparisons. Circ Cardiovasc
Qual Outcomes 2012;5:267–75. Percutaneous coronary intervention in native
24. Teirstein PS. Percutaneous revascularization is
coronary arteries versus bypass grafts in pa-
18. Palmerini T, Della Riva D, Biondi-Zoccai G, the preferred strategy for patients with significant
tients with prior coronary artery bypass graft
et al. Mortality following nonemergent, uncom- left main coronary stenosis. Circulation 2009;119:
surgery: insights from the Veterans Affairs
plicated target lesion revascularization after 1021–33.
Clinical Assessment, Reporting, and Tracking
percutaneous coronary intervention: an individual 25. Baron SJ, Chinnakondepalli K, Magnuson EA, Program. J Am Coll Cardiol Intv 2016;9:
patient data pooled analysis of 21 randomized et al. Quality-of-life after everolimus-eluting 884–93.
trials and 32,524 patients. J Am Coll Cardiol Intv stents or bypass surgery for left-main disease:
2018;11:892–902. results from the EXCEL trial. J Am Coll Cardiol
19. Stone GW, Sabik JF, Serruys PW, et al. Ever- 2017;70:3113–22.
olimus-eluting stents or bypass surgery for left KEY WORDS CABG, left main coronary
26. Abdallah MS, Wang K, Magnuson EA, et al.
main coronary artery disease. N Engl J Med 2016; artery, PCI, repeat revascularization
Quality of life after PCI vs CABG among patients
375:2223–35.
with diabetes and multivessel coronary artery
20. Sawaya FJ, Morice MC, Spaziano M, et al. disease: a randomized clinical trial. JAMA 2013; A PP END IX For supplemental tables and fig-
Short-versus long-term dual antiplatelet therapy 310:1581–90. ures, please see the online version of this paper.
Articles
Percutaneous coronary intervention versus coronary artery

bypass grafting in patients with three-vessel or left main
coronary artery disease: 10-year follow-up of the multicentre
randomised controlled SYNTAX trial
Daniel J F M Thuijs, A Pieter Kappetein, Patrick W Serruys, Friedrich-Wilhelm Mohr, Marie-Claude Morice, Michael J Mack, David R Holmes Jr,
Nick Curzen, Piroze Davierwala, Thilo Noack, Milan Milojevic, Keith D Dawkins, Bruno R da Costa, Peter Jüni, Stuart J Head, for the SYNTAX
Extended Survival Investigators*
Summary
Background The Synergy between PCI with Taxus and Cardiac Surgery (SYNTAX) trial was a non-inferiority trial that Published Online
compared percutaneous coronary intervention (PCI) using first-generation paclitaxel-eluting stents with coronary September 2, 2019
http://dx.doi.org/10.1016/
artery bypass grafting (CABG) in patients with de-novo three-vessel and left main coronary artery disease, and S0140-6736(19)31997-X
reported results up to 5 years. We now report 10-year all-cause death results.
See Online/Comment
http://dx.doi.org/10.1016/
Methods The SYNTAX Extended Survival (SYNTAXES) study is an investigator-driven extension of follow-up of a S0140-6736(19)32040-9
multicentre, randomised controlled trial done in 85 hospitals across 18 North American and European countries. *Investigators are listed in the
Patients with de-novo three-vessel and left main coronary artery disease were randomly assigned (1:1) to the PCI appendix
group or CABG group. Patients with a history of PCI or CABG, acute myocardial infarction, or an indication for Department of Cardiothoracic
Surgery, Erasmus University
concomitant cardiac surgery were excluded. The primary endpoint of the SYNTAXES study was 10-year all-cause
Medical Centre,
death, which was assessed according to the intention-to-treat principle. Prespecified subgroup analyses were Rotterdam, Netherlands
performed according to the presence or absence of left main coronary artery disease and diabetes, and according to (D J F M Thuijs MD,
coronary complexity defined by core laboratory SYNTAX score tertiles. This study is registered with ClinicalTrials.gov, Prof A P Kappetein PhD,
M Milojevic PhD, S J Head PhD);
NCT03417050.
Medtronic, Maastricht,
Netherlands
Findings From March, 2005, to April, 2007, 1800 patients were randomly assigned to the PCI (n=903) or CABG (n=897) (Prof A P Kappetein);
group. Vital status information at 10 years was complete for 841 (93%) patients in the PCI group and 848 (95%) patients Department of Cardiology,
Imperial College London,
in the CABG group. At 10 years, 244 (27%) patients had died after PCI and 211 (24%) after CABG (hazard
London, UK
ratio 1·17 [95% CI 0·97–1·41], p=0·092). Among patients with three-vessel disease, 151 (28%) of 546 had died after (Prof P W Serruys PhD);
PCI versus 113 (21%) of 549 after CABG (hazard ratio 1·41 [95% CI 1·10–1·80]), and among patients with left main University Department of
coronary artery disease, 93 (26%) of 357 had died after PCI versus 98 (28%) of 348 after CABG (0·90 [0·68–1·20], Cardiac Surgery, Heart Centre
Leipzig, Leipzig, Germany
pinteraction=0·019). There was no treatment-by-subgroup interaction with diabetes (pinteraction=0·66) and no linear trend
(Prof F-W Mohr PhD,
across SYNTAX score tertiles (ptrend=0·30). P Davierwala MD, T Noack MD);
Department of Cardiology,
Interpretation At 10 years, no significant difference existed in all-cause death between PCI using first-generation Cardiovascular Institute
Paris-Sud, Hopital Privé Jacques
paclitaxel-eluting stents and CABG. However, CABG provided a significant survival benefit in patients with Cartier, Ramsay Générale de
three-vessel disease, but not in patients with left main coronary artery disease. Santé, Massy, France
(M-C Morice PhD); Department
Funding German Foundation of Heart Research (SYNTAXES study, 5–10-year follow-up) and Boston Scientific of Cardiothoracic Surgery,
Baylor University Medical
Corporation (SYNTAX study, 0–5-year follow-up). Centre, Dallas, TX, USA
(M J Mack PhD); Department of
Copyright © 2019 Elsevier Ltd. All rights reserved. Cardiovascular Diseases and
Internal Medicine, Mayo Clinic,
Rochester, MN, USA
Introduction [95% CI 1·06–1·37], p=0·0038). However, the mean age (D R Holmes Jr MD); University
Several randomised trials1–8 have compared coronary of the patient population was 65 years, and thus the Hospital Southampton NHS
artery bypass grafting (CABG) versus percutaneous coro overall life expectancy of most patients exceeded this Foundation Trust and School of
nary intervention (PCI) with simple balloon angioplasty, follow-up time. Longer-term follow-up beyond 5 years is Medicine, University of
Southampton,
bare metal stents, or drug-eluting stents for the treatment required to determine the relative effectiveness of PCI Southampton, UK
of multivessel or left main coronary artery disease, but no versus CABG. The Synergy between PCI with Taxus and (Prof N Curzen PhD); Shockwave
significant differences in survival were demonstrated. Cardiac Surgery (SYNTAX) trial compared PCI with Medical Inc, Santa Clara, CA,
Results from a pooled analysis of individual patient data9 paclitaxel-eluting stents versus CABG in 1800 patients USA (K D Dawkins MD); Applied
Health Research Centre,
from 11 trials and 11 518 patients suggested that all-cause with de-novo three-vessel disease and left main coronary Li Ka Shing Knowledge
death was significantly lower after CABG versus PCI at artery disease, and reported similar survival among Institute of St Michael’s
5-year follow-up (9·2% vs 11·2%; hazard ratio [HR] 1·20 patients in the PCI and CABG groups after 5 years of Hospital (B R da Costa PhD,
www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X 1

Articles
Prof P Jüni MD), and

Department of Medicine, Research in context
Institute of Health Policy,
Management and Evaluation, Evidence before this study (first-generation) drug-eluting stents. The search did not
University of Toronto, We searched PubMed for articles published in English between identify studies reporting outcomes in patients with de-novo
Toronto, ON, Canada database inception and July 24, 2019, with the following three-vessel and left main coronary artery disease randomly
(Prof P Jüni); Institute of
Primary Health Care, University
search terms: “percutaneous coronary intervention”, “stents”, assigned to PCI with drug-eluting stents or CABG.
of Bern, Bern, Switzerland “coronary artery bypass grafting”, and “random*”.
Added value of this study
(B R da Costa); and Medtronic, Randomised controlled trials and meta-analyses comparing
Minneapolis, MN, USA The current study is the first randomised trial that reports
percutaneous coronary intervention (PCI) using stents versus
(S J Head) complete 10-year data on all-cause death in patients with
coronary artery bypass grafting (CABG) in patients with
Correspondence to: de-novo three-vessel and left main coronary artery disease after
three-vessel or left main coronary artery disease were
Dr Daniel J F M Thuijs, PCI with drug-eluting stents versus CABG. It provides important
Department of Cardiothoracic included. An individual patient data meta-analysis reported
insights into the relative effectiveness of PCI versus CABG
Surgery, Erasmus University that mortality outcomes favoured CABG over PCI at 5-year
Medical Centre, regarding the most robust and clinically relevant outcome—
follow-up in patients with multivessel disease, particularly
3015 GD Rotterdam, all-cause death. At 10 years, no significant difference was found
those with diabetes and more complex coronary artery
Netherlands in all-cause death between PCI using first-generation
d.thuijs@erasmusmc.nl disease, whereas no significant difference was identified in
paclitaxel-eluting stents and CABG. However, CABG provided a
See Online for appendix patients with left main coronary artery disease. It was
significant survival benefit in patients with three-vessel disease,
concluded that longer-term follow-up would be required to
but not in patients with left main coronary artery disease.
better define mortality differences between revascularisation
These findings can aid decision making for patients with
strategies. In our search, we only found two randomised
coronary artery disease who require PCI or CABG, accounting
controlled trials reporting survival outcomes at 10 years after
for differences in cardiovascular risk factors, coronary lesion
PCI versus CABG. In the MASS II trial, patients with multivessel
complexity (eg, SYNTAX score), and the presence of
disease had a 10-year all-cause death rate of 24·9% after PCI
three-vessel or left main coronary artery disease.
versus 25·1% after CABG (p=0·089). However, PCI was done
with bare metal stents and about 40% of patients had Implications of all the available evidence
two-vessel disease, and no patients with left main coronary Patients with complex, three-vessel coronary artery disease
artery disease were included. In the LE MANS trial, which who require revascularisation should undergo CABG as it
included only patients with left main coronary artery disease results in significantly lower all-cause death than PCI. In
(n=105), 10-year all-cause death was 21·6% after PCI versus selected patients with left main coronary artery disease, PCI is
30·2% after CABG (p=0·41). However, the sample size was a suitable alternative to CABG and provides similar 10-year
small and only 35% of PCI procedures were performed with survival.
follow-up (13·9% all-cause death in the PCI group vs aged 21 years or older with de-novo three-vessel disease
11·4% all-cause death in the CABG group, p=0·10).5,10,11 and left main coronary artery disease were enrolled with
This study, the SYNTAX Extended Survival (SYNTAXES) the following exclusion criteria: a history of PCI or CABG,
study, examined all-cause death after 10 years of follow-up acute myocardial infarction, or an indication for con
in patients randomly assigned to PCI or CABG in the comitant cardiac surgery (see appendix pp 56–58 for the
SYNTAX trial. complete list of inclusion and exclusion criteria).
The SYNTAXES study is registered at ClinicalTrials.gov
Methods as an investigator-driven extension of follow-up of
Study design and patients the SYNTAX trial. Medical Ethical Committee approval
The SYNTAX trial (NCT00114972) was a multicentre, for this study was granted at the institution of the
randomised controlled trial done in 85 hospitals across principal investigators (Erasmus University Medical
18 North American and European countries, with the Centre, Rotterdam, Netherlands, reference: MEC-2016-716).
aim of assessing non-inferiority of PCI with paclitaxel- Informed consent to obtain information on 10-year
eluting stents versus CABG in patients with de-novo vital status was waived, and follow-up was performed
three-vessel disease and left main coronary artery disease in accordance with local law and regulations of each
for the primary endpoint of major adverse cardiac or participating site and complied with the Declaration of
cerebrovascular events at 1 year. The rationale, design, Helsinki. Survival data were obtained by (electronic)
and 1-year primary endpoint results of the SYNTAX trial health-care record review and national death registry
have been published previously, as well as results at checks.
prolonged 3-year and 5-year follow-ups.1,5,12
The SYNTAX trial completed follow-up at 5 years Randomisation and masking
and was reinitiated as the SYNTAXES study to evaluate Randomisation and masking for the SYNTAXES study
survival up to 10 years (the protocol and CONSORT was the same as for the SYNTAX study. Briefly, patients
checklist are available in the appendix pp 26–58). Patients who were assessed as equally suitable for CABG or PCI
2 www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X

Articles
were randomly assigned (1:1) to one of the two treat and three-vessel disease subgroups and overall. After
ments, as described in detail in previous publications.1,5,12 allowing for an expected attrition rate of 3·5%, the
overall sample size of 1800 patients (900 per group)
Procedures resulted in 96% power to detect non-inferiority at a non-
Procedures were done according to local practice with the inferiority margin of 6·6% and a one-sided α level
intention to accomplish complete revascularisation of of 5%.
any vessel at least 1·5 mm in diameter with stenosis of All analyses were according to the intention-to-treat
50% or more, identified during pre-procedural heart team principle. Patients with missing follow-up data were
meetings.1 CABG could be performed with or without included in the analysis and censored at the time they
cardiopulmonary bypass and the use of arterial grafts was were lost to follow-up or at 5 years if their recruiting
strongly recommended yet not mandatory. PCI could be hospital did not participate in the 10-year follow-up. We
performed using a radial, femoral, or brachial approach.
Staged PCI procedures were allowed when performed
within 72 h of the initial treatment and during the same 4337 patients assessed for eligibility
hospital stay. Every patient was prescribed life-long
aspirin, and adherence to contemporaneous guideline-
2537 excluded
directed medical treatment was highly recommended.13 408 had a treatment preference
306 declined to participate after providing
informed consent, or the referring physician
Outcomes declined to accept the patient’s consent
The prespecified primary endpoint of the SYNTAXES 210 met exclusion criteria
study was all-cause death at 10 years in patients randomly 194 declined to participate before providing
informed consent
assigned to PCI with drug-eluting stents versus CABG. 79 had other reason
The secondary endpoint was all-cause death at maximum 51 underwent medical treatment
14 declined to undergo revascularisation
available follow-up in patients randomly assigned to PCI 1275 eligible only for enrolment in parallel registries
with drug-eluting stents versus CABG. 198 enrolled in PCI registry
The left main coronary artery disease subgroup 1077 enrolled in CABG registry
consisted of patients with any left main disease, either

isolated, or in combination with single-vessel, two-vessel, 1800 randomly assigned
or three-vessel coronary artery disease. The three-vessel
disease subgroup consisted of patients with coronary
artery disease involving all three vessels in the absence
of left main coronary artery disease.5,12 The anatomical 903 allocated to PCI 897 allocated to CABG
886 received allocated treatment 856 received allocated treatment
complexity of coronary artery disease was graded 17 did not receive allocated treatment 41 did not receive allocated treatment
according to the SYNTAX score during prerandom 11 underwent CABG 16 underwent PCI
6 did not have PCI or CABG 25 did not have PCI or CABG
isation heart team meetings, with higher SYNTAX
scores indicating more complex coronary artery disease.14
SYNTAX scores, according to core laboratory analyses, 3 from hospitals that did not 2 from hospitals that did not
were defined according to tertiles, with scores of 22 or participate in 10-year participate in 10-year
follow-up (with complete follow-up (with complete
lower defined as low, 23–32 as intermediate, and 33 or primary endpoint data at primary endpoint data at
higher as high.1,6 The European System for Cardiac 5 years, and followed up 5 years, and followed up
and alive) and alive)
Operative Risk Evaluation (EuroSCORE) was used to
assess operative risk. Diabetes was defined as patients
requiring treatment with oral hypoglycaemic agents or 900 from hospitals that participated in 895 from hospitals that participated in
insulin. Incomplete revascularisation was determined 10-year follow-up 10-year follow-up
841 with complete primary endpoint 848 with complete primary endpoint
post-procedurally by correlating the revascularised data at 10 years data at 10 years
lesions to those lesions identified during the preoperative 597 followed up and alive 637 followed up and alive
244 died 211 died
heart team meeting. 59 with incomplete primary endpoint 47 with incomplete primary endpoint
data at 10 years (lost to follow-up) data at 10 years
Statistical analysis (lost to follow-up)
The sample size of the SYNTAXES study was based on

the sample size considerations for the original trial, 903 included in intention-to-treat analysis 897 included in intention-to-treat analysis
which was powered for a non-inferiority comparison 59 censored at time of loss to follow-up 47 censored at time of loss to follow-up
3 censored at 5 years as hospitals did 2 censored at 5 years as hospitals did
of major adverse cardiac and cerebrovascular events not participate in 10-year follow-up not participate in 10-year follow-up
at 12 months between PCI and CABG (a complete
description of the sample size calculation is provided Figure 1: Trial profile
in the appendix p 6).1 Sample sizes were calculated Patient flow through the SYNTAX trial (0–5 years of follow-up) and the SYNTAX Extended Survival study (up to
for each of the left main coronary artery disease 10 years of follow-up). CABG=coronary artery bypass grafting. PCI=percutaneous coronary intervention.

Articles
the SYNTAX trial from the extended follow-up in the

PCI group (n=903) CABG group (n=897)
SYNTAXES study. Landmark analyses were accompanied
Age, years 65·2 (9·7) 65·0 (9·8) by a test for interaction between treatment effect and
Sex time (first 5 years versus subsequent period). We did a
Women 213 (24%) 189 (21%) sensitivity analysis of the primary endpoint using a
Men 690 (76%) 708 (79%) multivariable Cox model with stepwise forward selection
Body-mass index, kg/m² 28·1 (4·8) 27·9 (4·5) of covariates. Prespecified subgroup analyses were done
Diabetes in a hierarchical manner. For the primary subgroup
Requiring oral 231 (26%) 221 (25%) analysis according to the presence or absence of left main
medications or insulin
coronary artery disease, we used a prespecified Bonferroni
Requiring insulin 89 (10%) 93 (10%) correction, which allowed for the treatment-by-subgroup
Metabolic syndrome 339/737 (46%) 317/696 (46%) interaction to be tested at a two-sided α value of
Ever smoked 167 (18%) 196/890 (22%) 0·025 (0·05/2) in addition to comparing the primary
Previous myocardial 285/893 (32%) 300/887 (34%) endpoint between PCI and CABG in the overall popu
infarction
lation. For the secondary subgroup analyses according to
Previous stroke 35/899 (4%) 43/890 (5%)
presence or absence of diabetes and complexity of
Previous transient 39/901 (4%) 45/888 (5%)
ischaemic attack
coronary artery disease defined by ordered SYNTAX score
Hypertension 622 (69%) 574 (64%)
tertiles, we used an additional Bonferroni correction,
(≥130/85 mm Hg) which allowed for the interaction and the trend of log
Congestive heart failure 36/898 (4%) 47/880 (5%) HRs across ordered SYNTAX score tertiles to be tested at
Previous carotid artery 73 (8%) 75 (8%) a two-sided α value of 0·0125 (0·05/4). As this approach
disease allowed for the primary endpoint in the overall population
Hyperlipidaemia 705/896 (79%) 686/889 (77%) to be tested at a two-sided α value of 0·05, without
Angina requiring a significant test of the primary endpoint at the
Stable 514 (57%) 513 (57%) prespecified α level before proceeding with treatment-by-
Unstable 262 (29%) 251 (28%) subgroup interaction tests, it mitigated the inflation of
Ejection fraction <30% 12/891 (1%) 22/875 (3%) the type I error rate considerably when performing
EuroSCORE value 3·8 (2·6) 3·8 (2·7) multiple subgroup analyses, but did not fully control it.
Parsonnet score 8·5 (7·0) 8·4 (6·8) Prespecified subgroup analyses of the primary endpoint
SYNTAX score* 28·4 (11·5) 29·1 (11·4) by age, sex, and ordered SYNTAX score tertiles in left
Number of lesions 4·3 (1·8) 4·4 (1·8) main coronary artery disease and three-vessel disease
Total occlusion 217/897 (24%) 198/897 (22%) subgroups were considered exploratory. Additional post-
Bifurcation lesion 649/897 (72%) 651/890 (73%) hoc exploratory analyses in the left main coronary artery
Three-vessel disease 546 (60%) 549 (61%) disease subgroup were performed according to the
Left main coronary 357 (40%) 348 (39%)
presence or absence of additional vessel disease (ie,
artery disease, any isolated left main coronary artery disease, or left main
Isolated 42/357 (12%) 49/348 (14%) coronary artery disease in combination with one-vessel,
Plus one-vessel disease 67/357 (19%) 71/348 (20%) two-vessel, or three-vessel disease). All subgroup analyses
Plus two-vessel disease 112/357 (31%) 106/348 (30%) were done in unadjusted and adjusted manners
Plus three-vessel 136/357 (38%) 122/348 (35%) (the statistical analysis plan is available in the appendix
disease pp 59–66). The secondary endpoint of all-cause death at
Data are mean (SD), n (%), or n/N (%), unless otherwise noted. Percentages might maximum follow-up was analysed identically, including
not sum to 100% as a result of rounding. Data are reported according to the post-hoc subgroup analyses. Baseline characteristics of
intention-to-treat principle. CABG=coronary artery bypass grafting. patients with and without availability of 10-year follow-up
EuroSCORE=European System for Cardiac Operative Risk Evaluation.
PCI=percutaneous coronary intervention. *SYNTAX scores are reported according
were compared to address the potential for attrition bias.
to core laboratory analysed data. Analyses were performed using Stata, version 15, and
SPSS Statistics software, version 24.
Table: Baseline clinical and angiographic characteristics
This study is registered with ClinicalTrials.gov,
NCT03417050.
analysed the primary endpoint of 10-year all-cause death
using Kaplan-Meier curves, with a log-rank p value to test Role of the funding source
between-group differences at a two-sided α value of 0·05. The funders had no role in the SYNTAXES study design,
We used Cox proportional hazards models to estimate data collection, data analyses, interpretation of the data,
HRs with 95% CIs comparing PCI with CABG. We did or writing of the report. The corresponding author, and
landmark analyses in the overall population and in APK, MM, BRdC, PJ, and SJH, had full access to the data
prespecified subgroups, setting the landmark point at in the study and had final responsibility for the decision
5 years to distinguish the results of the 5-year analysis in to submit for publication.

Articles
Results A
From March, 2005, to April, 2007, 1800 patients were 100 PCI group
randomly assigned to undergo PCI with paclitaxel- CABG group
eluting stents (n=903) or CABG (n=897; figure 1). Clinical 90
and angiographic characteristics were well matched 80

between groups (table). Further details about the pro
70
cedural characteristics of the patients included in this
Probability of death (%)

study have been published previously1,5 and are included 60 HR 1·17 (95% CI 0·97–1·41);
in the appendix (p 8). p=0·092
50
Information on 10-year survival was collected between
40
March 1, 2017, and June 17, 2019. Two hospitals, which
included five patients, elected not to participate in the 30
SYNTAXES study. Information on vital status at 10-year
20
follow-up was complete in 841 (93%) patients in the
PCI group and 848 (95%) patients in the CABG group. 10
Baseline characteristics of patients with versus without 0
vital status at 10 years are provided in the appendix (p 9).
The median duration of follow-up was 11·2 years B
(IQR 7·7–12·1) overall and 11·9 years (11·2–12·3) in 100 0–5 years 5–10 years
HR 1·19 (95% CI 0·92–1·54) HR 1·15 (95% CI 0·89–1·50)
survivors. 90
The primary endpoint of all-cause death at 10 years
80
occurred in 244 (27%) of 903 patients after PCI and
211 (24%) of 897 patients after CABG (HR 1·17 [95% CI 70
0·97–1·41, p=0·092; figure 2A). Landmark analysis 60

between 5-year and 10-year follow-up identified that all-
cause death occurred in 119 (13%) patients after PCI and 50
in 106 (12%) patients after CABG (HR 1·15 [95% CI 40

0·89–1·50]; figure 2B). At maximum follow-up, PCI was
30
associated with higher all-cause death than was CABG
(303 [34%] vs 264 [29%], HR 1·18 [95% CI 1·00–1·39]; 20
appendix p 15). 10
There was a treatment-by-subgroup interaction accor
0
ding to presence or absence of left main coronary artery 0 1 2 3 4 5 6 7 8 9 10
disease (pinteraction=0·019). In the three-vessel disease Time since randomisation (years)
Number at risk
subgroup, all-cause death at 10 years occurred in PCI group 903 860 844 822 795 744 699 680 651 621 583
151 (28%) of 546 patients after PCI compared with CABG group 897 856 838 820 799 753 711 687 666 644 620
113 (21%) of 549 patients after CABG (HR 1·41 [95% CI
Figure 2: Kaplan-Meier curves for primary analysis of 10-year all-cause death (intention-to-treat population)
1·10–1·80]; figure 3A). In the left main coronary artery The probability of all-cause death in PCI versus CABG up to 10 years of follow-up (A) and landmark analysis
disease subgroup, all-cause death at 10 years occurred in according to a landmark point at 5 years (B). Because the widths of 95% CIs were not adjusted for multiple
93 (26%) of 357 patients after PCI versus 98 (28%) of comparisons in the landmark analysis, these intervals should not be used for inference about between-group
differences. CABG=coronary artery bypass grafting. HR=hazard ratio. PCI=percutaneous coronary intervention.
348 patients after CABG (HR 0·90 [95% CI 0·68–1·20];
figures 3B, 4). There was no treatment-by-subgroup
interaction according to diabetes status (pinteraction=0·66; At 10-year follow-up, the proportions of all-cause deaths
figures 3C, 3D, 4) and ordered SYNTAX score tertiles between PCI and CABG were similar. Prespecified
(ptrend=0·30; figures 4, 5). Results were similar in adjusted subgroup analyses identified that CABG resulted in
subgroup analyses at 10 years and in subgroup analyses significantly lower all-cause death than did PCI in
at maximum follow-up (appendix pp 16–24). Additional patients with three-vessel disease, whereas no significant
exploratory analyses and the prespecified sensitivity difference between PCI and CABG was identified in
analysis are provided in the appendix (pp 11–14, 25). patients with left main coronary artery disease. The
current study provides unique long-term insights into
Discussion survival after PCI versus CABG by extending follow-up to
The SYNTAX trial reported similar survival in patients 10 years, which could aid in decision making in
with de-novo three-vessel and left main coronary artery determining the optimal revascularisation strategy
disease randomly assigned to PCI with paclitaxel-eluting for patients with coronary artery disease. Moreover, the
stents versus CABG after 5 years of follow-up. The primary endpoint of all-cause death focuses on the most
SYNTAXES study is the first study to assess 10-year robust endpoint that is clinically relevant for both
survival after PCI with drug-eluting stents versus CABG. patients and physicians. In addition, follow-up at 10 years

Articles
A Three-vessel disease B Left main coronary artery disease

100 PCI group
90 CABG group
80
70
60
50
HR 1·41 (95% CI 1·10–1·80) HR 0·90 (95% CI 0·68–1·20)
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Number at risk
PCI group 546 517 506 490 477 449 417 407 389 372 346 357 343 338 332 318 295 282 273 262 249 237
CABG group 549 524 515 506 494 470 446 436 422 409 397 348 332 323 314 305 283 265 251 244 235 223
C Diabetes D No diabetes
100
90
80
70
60
50
HR 1·10 (95% CI 0·80–1·52) HR 1·20 (95% CI 0·96–1·51)
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Time since randomisation (years) Time since randomisation (years)
Number at risk
PCI group 231 210 206 198 190 178 164 160 151 146 128 672 650 638 624 605 566 535 520 500 475 455
CABG group 221 206 199 196 190 177 165 157 151 141 131 676 650 639 624 609 576 546 530 515 503 489
Figure 3: Kaplan-Meier curves for prespecified subgroup analysis of 10-year all-cause death (intention-to-treat population)
The probability of all-cause death in PCI versus CABG up to 10 years of follow-up in prespecified subgroups of patients with three-vessel disease (A), with left main
coronary artery disease (B), with diabetes (C), and without diabetes (D). p value for interaction for three-vessel disease versus left main coronary artery disease
was 0·019, and p value for interaction for diabetes versus no diabetes was 0·66. Because the widths of 95% CIs were not adjusted for multiple comparisons, these
intervals should not be used for inference about between-group differences. CABG=coronary artery bypass grafting. HR=hazard ratio. PCI=percutaneous coronary
intervention.
PCI group CABG group HR (95% CI) pinteraction
Type of coronary disease

Left main coronary artery disease 93/357 98/348 0·90 (0·68–1·20) 0·019
Three-vessel disease* 151/546 113/549 1·41 (1·10–1·80)
Medically treated diabetes
Yes 79/231 71/221 1·10 (0·80–1·52) 0·66
No 165/672 140/676 1·20 (0·96–1·51)
Coronary disease complexity
SYNTAX score ≤22 66/299 53/275 1·13 (0·79–1·62) 0·30†
SYNTAX score 23–32 78/310 71/300 1·06 (0·77–1·47)
SYNTAX score ≥33 98/290 82/315 1·41 (1·05–1·89)
0·5 0·8 1·0 1·25 2·0
Favours PCI Favours CABG
Figure 4: Forest plot of prespecified subgroup analyses of 10-year all-cause death (intention-to-treat population)
All-cause death after PCI versus CABG at 10-year follow-up in prespecified unadjusted subgroup analyses according to baseline characteristics. Because the widths of
95% CIs were not adjusted for multiple comparisons, these intervals should not be used for inference about between-group differences. CABG=coronary artery bypass
grafting. HR=hazard ratio. PCI=percutaneous coronary intervention. *Patients with coronary artery disease involving all three vessels in the absence of left main
coronary artery disease. †p value for trend of log HRs across SYNTAX score tertiles for subgroup analysis according to lesion complexity.

Articles
was complete for 94% of randomly assigned patients and

equally distributed between CABG and PCI. A SYNTAX score ≤22
100 PCI group
In the SYNTAX trial, PCI was performed with first- CABG group
90
generation drug-eluting stents that are no longer 80
available. Newer-generation drug-eluting stents have

70
been shown to be associated with significantly improved 60 HR 1·13 (95% CI 0·79–1·62)
mid-term (up to 3 years of follow-up) outcomes, in 50
cluding reduction of all-cause death.15 Moreover, the 40
larger adoption of fractional-flow reserve instead of 30
solely angiography-guided interventions, in combination 20
with the application of intravascular ultrasound, has 10
resulted in improved outcomes after PCI.16,17 Indeed, the 0
SYNTAX II study18 demonstrated that these develop 0 1 2 3 4 5 6 7 8 9 10
ments were associated with significant reductions in Number at risk
PCI group 299 293 289 283 275 261 245 240 232 223 207
adverse events during follow-up. Despite these improve CABG group 275 262 258 255 247 237 221 214 210 206 202
ments, most recent randomised trials have shown
that CABG remained consistently associated with lower B SYNTAX score 23–32
100
rates of repeat revascularisation at mid-term follow-
90
up compared with PCI, regardless of which type of
80
stent was used.2,4 Longer-term follow-up of trials com
70
paring contemporaneous PCI with CABG are therefore HR 1·06 (95% CI 0·77–1·47)
60
warranted to determine the relative effectiveness of PCI
50
versus CABG. 40
According to our prespecified subgroup analyses, 30
patients with more complex coronary disease (eg, 20
three-vessel disease and those with higher SYNTAX 10
scores) continued to derive a benefit of CABG over PCI 0
beyond the 5-year follow-up. These results underscore 0 1 2 3 4 5 6 7 8 9 10
the long-term impact of CABG over PCI that might Number at risk
PCI group 310 298 290 282 273 256 241 235 231 222 205
be attributable to two factors. First, coronary bypass CABG group 300 282 273 268 263 245 239 230 224 215 207
surgery offers the advantage of overcoming the overall
burden of complex and diffuse atherosclerotic disease C SYNTAX score ≥33
100
by constructing the anastomosis distal to diseased seg
90
ments, whereas PCI only treats significant flow-limiting
80
lesions without protecting the distally diseased vessels.
70
Second, CABG is associated with a higher rate of complete HR 1·41 (95% CI 1·05–1·89)
60
revascularisation than achieved with PCI.19–21 Particularly
50
in patients with diffuse and complex coronary disease,
40
PCI can be technically challenging and more frequently
30
results in incomplete revascularisation. More incomplete
20
revascularisation is associated with an increased risk of 10
death at 5-year follow-up, whereas minimal incomplete 0
revascularisation is not.22,23 In patients with low coronary 0 1 2 3 4 5 6 7 8 9 10
disease complexity for which complete revascularisation Number at risk
Time since randomisation (years)
with PCI is achievable, PCI is a suitable alternative to PCI group 290 265 261 253 243 225 210 202 185 173 169
CABG group 315 306 301 291 283 266 246 238 228 219 209
CABG.24 Finally, adherence to guideline-directed medical
therapy after revascularisation is important to adequately Figure 5: Kaplan-Meier curves for 10-year all-cause death in prespecified SYNTAX score tertile subgroups
treat any progression of coronary artery disease. (intention-to-treat population)
The FREEDOM Follow-On study25 found significantly The probability of all-cause death in PCI versus CABG up to 10 years of follow-up in prespecified subgroups of
patients with low SYNTAX scores (≤22; A), intermediate SYNTAX scores (23–32; B), and high SYNTAX
fewer deaths with CABG versus PCI at a median follow- scores (≥33; C). p value for trend was 0·30. SYNTAX scores were reported according to core laboratory analysed
up of 7·5 years in patients with multivessel disease. data. Because the widths of 95% CIs were not adjusted for multiple comparisons, these intervals should not be
Typically, patients with diabetes, who have more used for inference about between-group differences. CABG=coronary artery bypass grafting. HR=hazard ratio.
complex and progressive coronary disease, also benefit PCI=percutaneous coronary intervention.
from CABG compared with PCI.6 The current study,
however, found no survival difference between PCI and size (n=452) as compared with the FREEDOM trial
CABG in patients with diabetes at 10 years. This finding (n=1900). The length of follow-up could also have
could be due to chance related to the smaller sample affected the difference in results of the FREEDOM study

Articles
(median 7·5 years) and the current study (median indicates that patients with advanced coronary artery
11·2 years), because in our analysis the Kaplan-Meier disease, as reflected by increasing SYNTAX scores,
curves con verged further with follow-up prolonging have a benefit with CABG over PCI. Indeed, in the
after 7–8 years. Moreover, the inclusion of patients with subgroup of patients with three-vessel disease and a
left main coronary artery disease in the current study high SYNTAX score, PCI resulted in higher 10-year
could have had an effect on the relative benefit of CABG all-cause death than did CABG (HR 1·83 [95% CI
over PCI in the overall diabetic cohort. In the recent 1·20–2·81]; appendix p 21), indicating a significant
pooled analysis of PCI versus CABG randomised trials, survival benefit of CABG over PCI. This hypothesis is
diabetes was an effect modifier in patients with further corroborated by the reasons for exclusion from
multivessel disease but not in patients with left main randomisation in the SYNTAX trial; the majority of
coronary artery disease.9 patients were referred to CABG for having very
Results at the 5-year follow-up provided promising complex coronary artery disease (mean SYNTAX score
survival outcomes of PCI versus CABG in patients with was 37·8).30
left main coronary artery disease and was corroborated At maximum follow-up, CABG appeared to be associ
in the pooled analysis of trials.9 It is reassuring that PCI ated with a borderline survival benefit compared with
resulted in a similar number of deaths at 10 years PCI. It is important to note that the HR was similar at
compared with CABG, as shown in the current analysis. 10-year and maximum follow-up, but with additional
The LE MANS trial26 also reported similar survival deaths the statistical power was increased at maximum
outcomes at 10 years in patients randomly assigned follow-up. The differences in survival outcomes between
to CABG or PCI with bare metal stents or first- PCI and CABG at maximum follow-up were identified
generation drug-eluting stents, but in a smaller cohort only in patients with three-vessel disease but not left
(n=105). Similarly, the observational MAIN-COMPARE main coronary artery disease, similar to the 10-year
study27 (n=2240) found no survival difference between findings. Because of the limited number of patients at
PCI with bare metal stents or drug-eluting stents and risk at maximum follow-up, these results should be
CABG at 10-year follow-up. PCI for a focal left main interpreted as hypothesis-generating and could be used
lesion—ie, large in diameter with high flow—results for sample size calculation in randomised controlled
in better stent patency and is therefore a suitable trials comparing PCI with CABG.
alternative to CABG in selected patients with left main Additional limitations should be considered. First,
coronary artery disease. Nevertheless, 56% of patients the endpoint was all-cause death only. Although causes
with left main coronary artery disease who underwent of death could have provided additional insights into
PCI in the SYNTAX trial had a distal left main lesion.28 mechanisms of death that could potentially be related to
Moreover, in the EXCEL trial,4 80·5% of patients had a the revascularisation strategy, it was not feasible to
distal lesion that involved a bifurcation or trifurcation collect those data.31 Second, additional outcomes, such
lesion, and subgroup analyses according to the presence as myocardial infarction, stroke, stent thrombosis, and
or absence of a distal bifurcation or trifurcation lesion graft occlusion, were not assessed but are important
found no significant interaction. These data suggest to consider when choosing the most appropriate
that PCI can be an alternative to CABG not only in revascularisation strategy.
patients with relatively non-complex left main lesions, In conclusion, no significant differences in all-cause
but also in patients with more complex disease, as also death emerged between PCI with first-generation pacli
demonstrated in our analyses according to SYNTAX taxel-eluting stents and CABG at 10 years. Nonetheless,
scores. The NOBLE3 and EXCEL4 trials might provide in patients with three-vessel disease, CABG provided a
important additional insights in long-term outcomes significant survival benefit over PCI, whereas no treat
after PCI with second-generation stents versus CABG if ment differences were identified in patients with left
follow-up is prolonged to 10 years. main coronary artery disease. The decision to opt for
Despite the fact that the SYNTAX score was originally PCI or CABG in patients with three-vessel disease or
intended to predict major adverse cardiac and cerebro left main coronary artery disease should be put forward
vascular events at the 1-year follow-up,14 the recent by a multidisciplinary heart team that takes into
pooled analysis of randomised trials suggested an consideration the presence or absence of mortality
interaction between SYNTAX score tertiles and death, differences in patient subgroups. In addition, the
particularly in patients with multivessel disease and overall coronary lesion complexity (eg, SYNTAX score),
less so in patients with left main coronary artery and other cardiovascular risk factors of an individual
disease.9,29 In patients with left main coronary artery patient, such as diabetes and additional comorbidities,
disease, we confirmed the absence of an association together with a patient’s preference, should be included
between the SYNTAX score and 10-year all-cause death. in the discussion.
However, although in the current study the interaction Contributors
test was negative, the visual inter pretation of the DJFMT, APK, PWS, F-WM, M-CM, MJM, DRH, NC, PD, KDD, and
interaction in patients with three-vessel disease SJH designed the SYNTAX trial or SYNTAXES study, enrolled

Articles
patients, or collected the data. DJFMT, APK, PWS, MM, BRdC, PJ, 7 Fihn SD, Blankenship JC, Alexander KP, et al. 2014 ACC/AHA/
and SJH analysed and interpreted the data. BRdC and MM were the AATS/PCNA/SCAI/STS focused update of the guideline for the
study statisticians. The analyses were performed in twofold, with one diagnosis and management of patients with stable ischemic heart
team led by PJ and one team led by SJH and DJFMT, to ensure disease: a report of the American College of Cardiology/American
validity of analyses. DJFMT participated in the study design and Heart Association Task Force on Practice Guidelines, and the
American Association for Thoracic Surgery, Preventive
oversaw data collection and verification. DJFMT, APK, PJ, and SJH
Cardiovascular Nurses Association, Society for Cardiovascular
drafted the report, which was critically reviewed by all authors. Angiography and Interventions, and Society of Thoracic Surgeons.
All authors approved the final version of the manuscript for J Am Coll Cardiol 2014; 64: 1929–49.
submission. 8 Hueb W, Lopes N, Gersh BJ, et al. Ten-year follow-up survival of the
Declaration of interests Medicine, Angioplasty, or Surgery Study (MASS II): a randomized
APK is Chief Medical Officer, Vice President at Medtronic. PWS reports controlled clinical trial of 3 therapeutic strategies for multivessel
personal consultancy fees from Abbott Laboratories, Biosensors, coronary artery disease. Circulation 2010; 122: 949–57.
Cardialysis, Medtronic, Micell, Sino Medical Sciences Technology, 9 Head SJ, Milojevic M, Daemen J, et al. Mortality after coronary
Philips/Volcano, Xeltis, and Heartflow. MJM reports non-financial artery bypass grafting versus percutaneous coronary intervention
with stenting for coronary artery disease: a pooled analysis of
support from Edwards Lifesciences, Medtronic, and Abbott, outside the
individual patient data. Lancet 2018; 391: 939–48.
submitted work. NC reports grants from Boston Scientific Corporation,
10 Morice MC, Serruys PW, Kappetein AP, et al. Five-year outcomes in
Haemonetics, and HeartFlow; personal fees from Boston Scientific
patients with left main disease treated with either percutaneous
Corporation, Abbott, Haemonetics, and Heartflow; education coronary intervention or coronary artery bypass grafting in the
grant from Volcano Phillips; and non-financial support from synergy between percutaneous coronary intervention with taxus and
Haemonetics, Heartflow, Biosensors, and Edwards, outside the cardiac surgery trial. Circulation 2014; 129: 2388–94.
submitted work. KDD is the chief medical officer of Shockwave 11 Head SJ, Davierwala PM, Serruys PW, et al. Coronary artery bypass
Medical Inc and 4Tech Cardio Ireland, and is also on the Board of grafting vs. percutaneous coronary intervention for patients with
Directors of Avicena LLC, JenaValve Technology Inc, and InnovHeart three-vessel disease: final five-year follow-up of the SYNTAX trial.
srl, and is a senior adviser to Conformal Medical Inc. PJ reports grants Eur Heart J 2014; 35: 2821–30.
from Canadian Institutes of Health Research, AstraZeneca, Biotronik, 12 Ong AT, Serruys PW, Mohr FW, et al. The SYNergy between
Biosensors International, Eli Lilly, and The Medicines Company, percutaneous coronary intervention with TAXus and cardiac surgery
outside the submitted work; reports honoraria to the institution for (SYNTAX) study: design, rationale, and run-in phase. Am Heart J
participation in advisory boards from Amgen unrelated to the 2006; 151: 1194–204.
submitted work, but has not received personal payments by any 13 King SB 3rd, Smith SC Jr, Hirshfeld JW Jr, et al. 2007 Focused
pharmaceutical company or device manufacturer; serves as unpaid Update of the ACC/AHA/SCAI 2005 Guideline Update for
member of the steering group of trials funded by AstraZeneca, Percutaneous Coronary Intervention: a report of the American
Biotronik, Biosensors, St Jude Medical, and The Medicines Company; College of Cardiology/American Heart Association Task Force on
Practice Guidelines: 2007 Writing Group to Review New Evidence
and is a Tier 1 Canada Research Chair in Clinical Epidemiology of
and Update the ACC/AHA/SCAI 2005 Guideline Update for
Chronic Diseases funded by the Canadian Institutes of Health
Percutaneous Coronary Intervention, Writing on Behalf of the 2005
Research. SJH is Global Clinical Evidence Director at Medtronic. Writing Committee. Circulation 2008; 117: 261–95.
All other authors declare no competing interests. 14 Sianos G, Morel MA, Kappetein AP, et al. The SYNTAX Score:
Data sharing an angiographic tool grading the complexity of coronary artery
The SYNTAX Extended Survival study hereby declares that no data will disease. EuroIntervention 2005; 1: 219–27.
be made available to others. 15 Stefanini GG, Baber U, Windecker S, et al. Safety and efficacy of
drug-eluting stents in women: a patient-level pooled analysis of
Acknowledgments randomised trials. Lancet 2013; 382: 1879–88.
The SYNTAX trial was supported by Boston Scientific Corporation 16 Tonino PA, De Bruyne B, Pijls NH, et al. Fractional flow reserve
(Marlborough, MA, USA) during the first 5-years of follow-up. versus angiography for guiding percutaneous coronary
The SYNTAX Extended Survival study was funded by the German Heart intervention. N Engl J Med 2009; 360: 213–24.
Research Foundation (Frankfurt am Main, Germany) for 5–10 years of 17 Zhang YJ, Pang S, Chen XY, et al. Comparison of intravascular
follow-up. We thank all research coordinators, cardiothoracic surgeons, ultrasound guided versus angiography guided drug eluting stent
and cardiologists at participating hospitals who contributed to the implantation: a systematic review and meta-analysis.
SYNTAX Extended Survival study. BMC Cardiovasc Disord 2015; 15: 153.
18 Serruys PW, Kogame N, Katagiri Y, et al. Clinical outcomes of
References state-of-the-art percutaneous coronary revascularisation in patients
1 Serruys PW, Morice MC, Kappetein AP, et al. Percutaneous with three-vessel disease: two-year follow-up of the SYNTAX II study.
coronary intervention versus coronary-artery bypass grafting for EuroIntervention 2019; 15: e244–52.
severe coronary artery disease. N Engl J Med 2009; 360: 961–72.
19 Doenst T, Haverich A, Serruys P, et al. PCI and CABG for treating
2 Park SJ, Ahn JM, Kim YH, et al. Trial of everolimus-eluting stents stable coronary artery disease: JACC review topic of the
or bypass surgery for coronary disease. N Engl J Med 2015; week. J Am Coll Cardiol 2019; 73: 964–76.
372: 1204–12.
20 Gersh BJ, Frye RL. Methods of coronary revascularization—
3 Mäkikallio T, Holm NR, Lindsay M, et al. Percutaneous coronary things may not be as they seem. N Engl J Med 2005; 352: 2235–37.
angioplasty versus coronary artery bypass grafting in treatment of
unprotected left main stenosis (NOBLE): a prospective, 21 Head SJ, Mack MJ, Holmes DR Jr, et al. Incidence, predictors and
randomised, open-label, non-inferiority trial. Lancet 2016; outcomes of incomplete revascularization after percutaneous coronary
388: 2743–52. intervention and coronary artery bypass grafting: a subgroup analysis
of 3-year SYNTAX data. Eur J Cardiothorac Surg 2012; 41: 535–41.
4 Stone GW, Sabik JF, Serruys PW, et al. Everolimus-eluting stents or
bypass surgery for left main coronary artery disease. N Engl J Med 22 Garcia S, Sandoval Y, Roukoz H, et al. Outcomes after complete
2016; 375: 2223–35. versus incomplete revascularization of patients with multivessel
coronary artery disease: a meta-analysis of 89,883 patients enrolled
5 Mohr FW, Morice M-C, Kappetein AP, et al. Coronary artery in randomized clinical trials and observational studies.
bypass graft surgery versus percutaneous coronary intervention in J Am Coll Cardiol 2013; 62: 1421–31.
patients with three-vessel disease and left main coronary disease:
5-year follow-up of the randomised, clinical SYNTAX trial. Lancet 23 Farooq V, Serruys PW, Bourantas CV, et al. Quantification of
2013; 381: 629–38. incomplete revascularization and its association with five-year
mortality in the synergy between percutaneous coronary intervention
6 Sousa-Uva M, Neumann F-J, Ahlsson A, et al. 2018 ESC/EACTS with taxus and cardiac surgery (SYNTAX) trial validation of the
Guidelines on myocardial revascularization. Eur J Cardiothorac Surg residual SYNTAX score. Circulation 2013; 128: 141–51.
2019; 55: 4–90.

Articles
24 Bangalore S, Guo Y, Samadashvili Z, Blecker S, Xu J, Hannan EL. 29 Head SJ, Papageorgiou G, Milojevic M, Stone GW, Kappetein AP.
Everolimus-eluting stents or bypass surgery for multivessel Interpretation of results of pooled analysis of individual patient
coronary disease. N Engl J Med 2015; 372: 1213–22. data—authors’ reply. Lancet 2018; 392: 818.
25 Farkouh ME, Domanski M, Dangas GD, et al. Long-term survival 30 Head SJ, Holmes DR Jr, Mack MJ, et al. Risk profile and 3-year
following multivessel revascularization in patients with diabetes: outcomes from the SYNTAX percutaneous coronary intervention
the FREEDOM Follow-On study. J Am Coll Cardiol 2019; 73: 629–38. and coronary artery bypass grafting nested registries.
26 Buszman PE, Buszman PP, Banasiewicz-Szkrobka I, et al. JACC Cardiovasc Interv 2012; 5: 618–25.
Left main stenting in comparison with surgical revascularization: 31 Milojevic M, Head SJ, Parasca CA, et al. Causes of death following
10-year outcomes of the (left main coronary artery stenting) PCI versus CABG in complex CAD: 5-year follow-up of SYNTAX.
LE MANS trial. JACC Cardiovasc Interv 2016; 9: 318–27. J Am Coll Cardiol 2016; 67: 42–55.
27 Park DW, Ahn JM, Yun SC, et al. 10-year outcomes of stents versus
coronary artery bypass grafting for left main coronary artery disease.
J Am Coll Cardiol 2018; 72: 2813–22.
28 Morice MC, Serruys PW, Kappetein AP, et al. Outcomes in patients
with de novo left main disease treated with either percutaneous
coronary intervention using paclitaxel-eluting stents or coronary
artery bypass graft treatment in the Synergy Between Percutaneous
Coronary Intervention with TAXUS and Cardiac Surgery (SYNTAX)
trial. Circulation 2010; 121: 2645–53.

JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 77, NO. 5, 2021
ª 2021 THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION.
PUBLISHED BY ELSEVIER. ALL RIGHTS RESERVED.
Mortality 10 Years After Percutaneous or

Surgical Revascularization in Patients
With Total Coronary Artery Occlusions
Hideyuki Kawashima, MD,a,b Kuniaki Takahashi, MD,a Masafumi Ono, MD,a,b Hironori Hara, MD,a
Rutao Wang, MD,b,c Chao Gao, MD,b,c Faisal Sharif, MD, PHD,b Michael J. Mack, MD,d David R. Holmes, MD,e
Marie-Claude Morice, MD,f Stuart J. Head, MD, PHD,g Arie Pieter Kappetein, MD, PHD,g Daniel J.F.M. Thuijs, MD,g
Milan Milojevic, MD, PHD,g,h Thilo Noack, MD,i Friedrich-Wilhelm Mohr, MD, PHD,i Piroze M. Davierwala, MD,i
Patrick W. Serruys, MD, PHD,b,j Yoshinobu Onuma, MD, PHD,b on behalf of the SYNTAX Extended Survival
Investigators
ABSTRACT
BACKGROUND The long-term clinical benefit after percutaneous coronary intervention (PCI) or coronary artery bypass
grafting (CABG) in patients with total occlusions (TOs) and complex coronary artery disease has not yet been clarified.
OBJECTIVES The objective of this analysis was to assess 10-year all-cause mortality in patients with TOs undergoing
PCI or CABG.
METHODS This is a subanalysis of patients with at least 1 TO in the SYNTAXES (Synergy Between PCI With Taxus and
Cardiac Surgery Extended Survival) study, which investigated 10-year all-cause mortality in the SYNTAX (Synergy
Between PCI With Taxus and Cardiac Surgery) trial, beyond its original 5-year follow-up. Patients with TOs were further
stratified according to the status of TO recanalization or revascularization.
RESULTS Of 1,800 randomized patients to the PCI or CABG arm, 460 patients had at least 1 lesion of TO. In patients with
TOs, the status of TO recanalization or revascularization was not associated with 10-year all-cause mortality, irrespective of
the assigned treatment (PCI arm: 29.9% vs. 29.4%; adjusted hazard ratio [HR]: 0.992; 95% confidence interval [CI]: 0.474
to 2.075; p ¼ 0.982; and CABG arm: 28.0% vs. 21.4%; adjusted HR: 0.656; 95% CI: 0.281 to 1.533; p ¼ 0.330). When TOs
existed in left main and/or left anterior descending artery, the status of TO recanalization or revascularization did not have
an impact on the mortality (34.5% vs. 26.9%; adjusted HR: 0.896; 95% CI: 0.314 to 2.555; p ¼ 0.837).
CONCLUSIONS At 10-year follow-up, the status of TO recanalization or revascularization did not affect mortality,
irrespective of the assigned treatment and location of TOs. The present study might support contemporary practice
among high-volume chronic TO-PCI centers where recanalization is primarily offered to patients for the management of
angina refractory to medical therapy when myocardial viability is confirmed. (Synergy Between PCI With TAXUS and
Cardiac Surgery: SYNTAX Extended Survival [SYNTAXES]; NCT03417050; SYNTAX Study: TAXUS Drug-Eluting Stent
Versus Coronary Artery Bypass Surgery for the Treatment of Narrowed Arteries [SYNTAX]; NCT00114972)
(J Am Coll Cardiol 2021;77:529–40) © 2021 the American College of Cardiology Foundation. Published by Elsevier.
All rights reserved.
From the aDepartment of Cardiology, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands;
b
Department of Cardiology, National University of Ireland, Galway, Galway, Ireland; cDepartment of Cardiology, Radboud Uni-
versity, Nijmegen, the Netherlands; dDepartment of Cardiothoracic Surgery, Baylor Scott and White Health, Dallas, Texas, USA;
Listen to this manuscript’s e
Department of Cardiovascular Diseases and Internal Medicine, Mayo Clinic, Rochester, Minnesota, USA; fDépartement of Car-
audio summary by diologie, Hôpital Privé Jacques Cartier, Générale de Santé Massy, France; gDepartment of Cardiothoracic Surgery, Erasmus Uni-
Editor-in-Chief versity Medical Centre, Rotterdam, the Netherlands; hDepartment of Cardiovascular Research, Dedinje Cardiovascular Institute,
Dr. Valentin Fuster on Belgrade, Serbia; iUniversity Department of Cardiac Surgery, Heart Centre Leipzig, Leipzig, Germany; and the jNational Heart and
JACC.org. Lung Institute, Imperial College London, London, United Kingdom. Samir Kapadia, MD, served as Guest Associate Editor for this
paper. Javed Butler, MD, MPH, MBA, served as Guest Editor-in-Chief for this paper.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.
Manuscript received October 13, 2020; revised manuscript received November 20, 2020, accepted November 23, 2020.
ISSN 0735-1097/$36.00 https://doi.org/10.1016/j.jacc.2020.11.055

530 Kawashima et al. JACC VOL. 77, NO. 5, 2021
10-Year Mortality in Patients With Total Occlusion FEBRUARY 9, 2021:529–40
C
ABBREVIATIONS hronic total occlusions (CTOs) are a SYNTAX (Synergy between PCI with Taxus and Car-
AND ACRONYMS common observation in patients diac Surgery) trial (NCT00114972) beyond its origi-
with stable coronary artery disease nally planned final follow-up at 5 years (11–13). In
3VD = 3-vessel disease
(CAD) who are undergoing coronary angiog- brief, the SYNTAX trial was a multicenter, random-
AF = angina frequency
raphy with a point-prevalence between 15% ized controlled trial done in 85 hospitals across 18
CABG = coronary artery bypass
and 30% (1–3). The American College of Car- North American and European countries, which
grafting
diology/American Heart Association/Society adopted an “all-comer” design with minimum
CAD = coronary artery disease
for Cardiovascular Angiography and Inter- exclusion criteria (11). A total of 1,800 patients with
ventions guidelines (4) recommend that in de novo 3VD and/or LM, who were deemed eligible
CK = creatine kinase
patients with appropriate clinical indications for both PCI and CABG based on clinical judgment and
CK-MB = creatine kinase- and suitable anatomy percutaneous coronary the consensus of a Heart Team, were enrolled and
myocardial band
intervention (PCI) of a CTO is reasonable randomized in a 1:1 fashion either to receive PCI
CTO = chronic total occlusion
when performed by operators with appro- (n ¼ 903) with the default use of Taxus Express
HR = hazard ratio
priate expertise. The recommendation by paclitaxel drug-eluting stents (Boston Scientific,
LAD = left anterior descending the current European guidelines is that Marlborough, Massachusetts) or CABG (n ¼ 897).
artery
percutaneous revascularization of CTO The main result of the SYNTAXES study in terms of
LM = left main disease
should be considered in patients with angina vital status up to 10 years has been recently reported
PCI = percutaneous coronary
resistant to medical therapy or with a large (14). The median duration of follow-up was 11.2 years
intervention
area of documented ischemia in the territory (interquartile range: 7.7 to 12.1 years) overall and 11.9
SAQ = Seattle Angina
Questionnaire
of the occluded vessel (5). However, the re- years (interquartile range: 11.2 to 12.3 years) in sur-
sults from previous cohort studies have vivors (14). The SYNTAX and SYNTAXES trials were
TO = total occlusion
been inconsistent with regard to the poten- approved by the ethics committees at each investi-
tial survival benefit of successful versus failed CTO- gating center, and all patients provided their written
PCI (6–9). Coronary artery bypass grafting (CABG) informed consent prior to participation in the SYN-
for total occlusion (TO) showed that failed revascular- TAX trial. Follow-up was performed in accordance
ization of a non–left anterior descending artery (LAD) with local law and regulations of each participating
TO was not associated with increased risk of long- institution and complied with the Declaration of
term mortality, whereas in the study, all the TOs in Helsinki.
LAD were bypassed (10). Therefore, the clinical STUDY ENDPOINT. The primary endpoint of this
benefit of recanalization or revascularization of TO study was all-cause mortality at 10 years. The sec-
is still debated. In patients with TO and complex ondary endpoint was all-cause mortality at maximum
CAD undergoing PCI or CABG, the benefit of the 2 available follow-up. All analyses were performed ac-
respective revascularization approaches on the long- cording to the intention-to-treat principle. Vital sta-
term (10-year) outcome has not yet been clarified. tus was confirmed by (electronic) health care record
Furthermore, the impact of successful treatment of review and national death registries. Patients with
TO on 10-year all-cause mortality in PCI- or CABG- missing vital status were included in the analysis and
treated patients with 3-vessel disease (3VD) and/or censored at the last date of contact or observation.
left main disease (LM) also remains undefined. Two hospitals, which included 5 patients in total,
SEE PAGE 541 decided not to participate in the SYNTAXES study.
TOTAL OCCLUSION AND SYNTAX SCORE. In the
The aim of the present subanalysis of the SYN-
SYNTAX trial, the calculation of the anatomical SYN-
TAXES (Synergy between PCI with Taxus and Cardiac
TAX score was performed by the study sites and by an
Surgery Extended Survival) study is to investigate the
independent core laboratory blinded to the treatment
10-year mortality in patients with TOs and complex
assignment (15–18). The age and angiographic char-
CAD after PCI or CABG stratified according to the
acteristics of TO were specified as part of the
status of TO recanalization or revascularization and
anatomical SYNTAX score calculation; both the age of
3VD and/or LM.
the TO #3 months and >3 months were specified, but
METHODS both were considered as TO (15). The definition of the
TO required that there was absolutely no flow
STUDY DESIGN AND PATIENT POPULATION. The through the lesion (TIMI [Thrombolysis In Myocardial
present study is a post hoc subgroup analysis of the Infarction] flow grade 0). Antegrade flow beyond the
SYNTAXES study (NCT03417050), which was an TO maintained by bridging collaterals and/or ipsi-
investigator-driven extended 10-year follow-up of the collaterals did not invalidate the definition of TO.
JACC VOL. 77, NO. 5, 2021 Kawashima et al. 531
FEBRUARY 9, 2021:529–40 10-Year Mortality in Patients With Total Occlusion
C ENTR AL I LL U STRA T I O N All-Cause Mortality at 10 Years According to the Status of Total

Occlusion Recanalization or Revascularization and Randomized Treatment of Percutaneous
Coronary Intervention or Coronary Artery Bypass Grafting
40
29.9%
All-Cause Mortality (%)
30 29.4%
28.0%
21.4%
20
10
Log-rank p = 0.632
0
0 1 2 3 4 5 6 7 8 9 10
Years Since Randomization
No. at risk:
103 98 96 93 88 84 79 78 75 73 67
134 128 125 122 120 116 109 107 99 91 88
135 131 126 123 121 116 111 110 104 96 94
88 83 83 80 78 76 69 66 65 64 63
Total Occlusion Recanalization-PCI Total Occlusion Revascularization-CABG

Non-Total Occlusion Recanalization-PCI Non-Total Occlusion Revascularization-CABG
Kawashima, H. et al. J Am Coll Cardiol. 2021;77(5):529–40.
Kaplan-Meier curves of all-cause mortality at 10 years according to the status of total occlusion (TO) recanalization or revascularization
and randomized treatment of percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) (TO population in
SYNTAXES [Synergy Between PCI With Taxus and Cardiac Surgery Extended Survival] study, n ¼ 460). TO recanalization–PCI (blue line)
versus non-TO recanalization–PCI (red dotted line) versus TO revascularization–CABG (gray line) versus non-TO revascularization–CABG
(purple dotted line).
Pre-procedure, the presence of TO was evaluated whenever all TOs were recanalized or revascularized
by the study sites as a part of site-reported anatomical according to the sites.
SYNTAX score calculation. Post-procedural status In the PCI arm, the residual SYNTAX score was
of TO recanalization or revascularization was also quantified by an independent core laboratory un-
assessed by the sites according to the site-reported aware of, and blind to, patient’s revascularization
anatomical SYNTAX score on an intention-to-treat outcome (21). This score was calculated as the sum of
basis; the data were available in the electronic case the individual scores of coronary lesions with $50%
records of the SYNTAX population (19,20). To main- diameter stenosis in vessel $1.5 mm that were left
tain the consistency of TO assessment, the site- without PCI (19,21,22).
reported anatomical SYNTAX score was described MEASUREMENT OF ANGINA STATUS. In the SYNTAX
and tabulated in this paper. trial, health status was assessed in all patients at
In the present study, the total randomized patients baseline; at 1 and 6 months; and 1, 3, and 5 years after
were stratified according to the presence or absence randomization (23). In the present study, disease-
of a TO, and those with TO were further stratified specific health status was assessed using the Seattle
according to the post-procedural status of recanalized Angina Questionnaire (SAQ) (24). The SAQ is a 19-item
or revascularized TOs and 3VD and/or LM. Patients questionnaire that measures 5 domains of health
were stratified as recanalized or revascularized TO status related to CAD: angina frequency (AF); physical
F I G U R E 1 All-Cause Mortality at 10 Years According to Randomized Treatments of PCI or CABG in Patients With or Without TO
A Overall
40
Log-rank p = 0.583

30
27.6%
26.1%
20
10
0
0 1 2 3 4 5 6 7 8 9 10
Patient number at risk
TO 460 440 430 418 407 392 368 361 343 324 312
Non-TO 1,340 1,276 1,252 1,224 1,188 1,129 1,070 1,029 998 963 913
B PCI
40
Log-rank p = 0.710
30 29.6%
28.0%
20
10
0
0 1 2 3 4 5 6 7 8 9 10
TO 237 226 221 215 208 200 188 185 174 164 155
Non-TO 666 634 623 607 587 557 523 505 488 466 437
Kaplan-Meier curves of all-cause mortality at 10 years according to randomized treatments of percutaneous coronary intervention (PCI) or
coronary artery bypass grafting (CABG) in patients with or without total occlusion (TO) (total SYNTAXES [Synergy Between PCI With Taxus
and Cardiac Surgery Extended Survival] population, n ¼ 1,800): (A) overall; (B) PCI arm; and (C) CABG arm. TO (blue line) versus non-TO
(red line).
Continued on the next page
limitations; disease perception and/or quality of life; score of 100) or as having monthly angina (SAQ-AF
angina stability; and treatment satisfaction (24,25). score: 70 to 90), weekly angina (SAQ-AF score: 40 to
The SAQ-AF score ranges from 0 to 100, with higher 60), or daily angina (SAQ-AF score: #30) (23).
scores indicating fewer symptoms and better health STATISTICAL ANALYSIS. The mean SD for contin-
status. Consistent with previous studies, patients uous variables were compared using the Student’s
were categorized as being angina-free (i.e., SAQ-AF t-test. Binary variables were reported as counts
F I G U R E 1 Continued
C CABG
40
Log-rank p = 0.718
30
25.5%
24.2%
20
10
0
0 1 2 3 4 5 6 7 8 9 10
TO 223 214 209 203 199 192 180 176 169 160 157
Non-TO 674 642 629 617 601 572 547 524 510 497 476
and/or percentages and compared with the chi-square mortality between patients with and without
or Fisher exact test as appropriate. The cumulative TO (27.6% vs. 26.1%; unadjusted HR: 1.060; 95% CI:
incidence of the 10-year all-cause mortality up to 10 0.862 to 1.302; p ¼ 0.583) (Figure 1, Supplemental
years was assessed using the Kaplan-Meier method Table 2). The results remained consistent after sta-
and compared using the log-rank test. Hazard ratio tistical adjustment for the confounding factors
(HR) with 95% confidence interval (CI) was assessed (adjusted HR: 1.018; 95% CI: 0.810 to 1.280;
by a Cox proportional regression model. To adjust for p ¼ 0.877) (Supplemental Table 2).
potential confounding factors, the following variables
COMPARISON OF PATIENTS WITH OR WITHOUT
were entered into a multivariable Cox regression
RECANALIZED/REVASCULARIZED TOs. Baseline char-
model: age; sex; body mass index; medically treated
acteristics in patients with TO undergoing PCI and
diabetes; hypertension; dyslipidemia; current
CABG are shown in Table 1. As baseline characteris-
smokers; previous myocardial infarction; previous
tics, the PCI arm had a higher prevalence of history of
cerebrovascular disease; peripheral vascular disease;
hypertension than the CABG arm did. All TOs were
chronic obstructive pulmonary disease; chronic kid-
successfully recanalized in 43.5% of patients after PCI
ney disease (defined as creatinine clearance <60 ml/
and revascularized in 60.5% with CABG (p < 0.001).
min); left ventricular ejection fraction; clinical pre-
The all-cause mortality estimates up to 10 years
sentation (silent ischemia, stable angina, or unstable
according to the status of TO recanalization or
angina); disease type (3VD or LM); and anatomical
revascularization are shown in the Central Illustration
SYNTAX score. A p value <0.05 was considered to be
and Table 2. In the PCI arm, there was no significant
statistically significant. All data were processed using
difference in the 10-year all-cause mortality between
SPSS version 26.0 (IBM Inc., Armonk, New York).
patients with successfully recanalized TO and those
RESULTS without (TO recanalization 29.9% vs. non-TO recan-
alization 29.4%; unadjusted HR: 1.041; 95% CI: 0.645
COMPARISON OF PATIENTS WITH AND WITHOUT TO to 1.681; p ¼ 0.868). Similarly, in the CABG arm, the
UNDERGOING PCI OR CABG. Of 1,800 randomized mortality of patients with revascularized TO did not
patients (7,739 lesions), 460 patients (25.6%; 543 differ from those without revascularized TO (TO
lesions) had at least 1 lesion of TO and 1,340 pa- revascularization 28.0% vs. non-TO revascularization
tients (74.4%; 7,196 lesions) did not have any TO 21.4%; unadjusted HR: 1.311; 95% CI: 0.746 to 2.303;
lesion. Baseline characteristics of patients with or p ¼ 0.346).
without TO are shown in Supplemental Table 1. After multivariate analysis, the 10-year all-cause
There was no significant difference in the 10-year mortality in patients with successfully recanalized
T A B L E 1 Baseline Characteristics and Medical Therapy in Patients With TO Undergoing PCI or CABG
Overall (N ¼ 460) PCI (n ¼ 237) CABG (n ¼ 223) p Value
Age, yrs 64.6 10.4 64.7 10.3 64.5 10.5 0.822

Body mass index, kg/m2 28.3 4.6 28.4 4.7 28.2 4.5 0.667
Male 378 (82.2) 188 (79.3) 190 (85.2) 0.100
Medically treated diabetes 134 (29.1) 77 (32.5) 57 (25.6) 0.102
On insulin 49 (10.7) 27 (11.4) 22 (9.9) 0.596
Hypertension 302 (65.7) 167 (70.5) 135 (60.5) 0.025
Dyslipidemia 367 (80.5) 194 (82.9) 173 (77.9) 0.180
Current smokers 89 (19.4) 41 (17.3) 48 (21.6) 0.242
Previous myocardial infarction 181 (39.8) 83 (35.5) 98 (44.3) 0.053
Previous cerebrovascular disease 69 (15.1) 33 (14.0) 36 (16.2) 0.504
Previous stroke 22 (4.8) 814 (6.0) 8 (3.6) 0.236
Previous TIA 29 (6.3) 12 (5.1) 17 (7.7) 0.263
Previous carotid artery disease 41 (8.9) 19 (8.0) 22 (9.9) 0.487
Peripheral vascular disease 46 (10.0) 24 (10.1) 22 (9.9) 0.926
Chronic obstructive pulmonary disease 35 (7.6) 21 (8.9) 14 (6.3) 0.296
Chronic kidney disease 82 (19.3) 46 (20.7) 36 (17.7) 0.436
Creatinine clearance, ml/min 87.6 32.1 86.9 32.4 88.3 31.8 0.652
Left ventricular ejection fraction, % 55.7 13.2 55.6 13.3 55.7 13.1 0.984
Congestive heart failure 30 (6.6) 13 (5.5) 17 (7.7) 0.346
Clinical presentation 0.101
Silent ischemia 76 (16.5) 31 (13.1) 45 (20.2)
Stable angina 257 (55.9) 135 (57.0) 122 (54.7)
Unstable angina 127 (27.6) 71 (30.0) 56 (25.1)
EuroSCORE 3.7 2.7 3.8 2.7 3.7 2.8 0.670
Parsonnet score 8.7 6.9 8.8 6.8 8.5 7.0 0.605
Disease type 0.862
3VD 349 (75.9) 176 (74.3) 173 (77.6)
LM
LM only 0 0 0
LM þ 1VD 9 (2.0) 5 (2.1) 4 (1.8)
LM þ 2VD 39 (8.5) 23 (9.7) 16 (7.2)
LM þ 3VD 63 (13.7) 33 (13.9) 30 (13.5)
Number of lesions 4.6 1.5 4.6 1.5 4.7 1.6 0.429
Anatomical SYNTAX score 30.8 10.6 31.3 11.6 30.3 9.5 0.315
Any bifurcation 336 (73.0) 173 (73.0) 163 (73.1) 0.981
All TOs recanalization or revascularization 238 (51.7) 103 (43.5) 135 (60.5) <0.001
Number of stents — 5.0 2.2 — —
Total stent length, mm — 95.9 48.1 — —
Off-pump CABG — — 38 (17.0) —
LAD with LIMA — — 197 (88.3) —
LAD with arterial graft — — 214 (96.0) —
Total number of graft conduits — — 2.8 0.7 —
Number of artery graft conduits — — 1.4 0.7 —
Number of vein graft conduits — — 1.3 0.9 —
Values are mean SD or n (%). Dashes indicate that data were not available.
CABG ¼ coronary artery bypass grafting; EuroSCORE ¼ European System for Cardiac Operative Risk Evaluation; LAD ¼ left anterior descending artery; LIMA ¼ left internal
mammary artery; LM ¼ left main disease; PCI ¼ percutaneous coronary intervention; SYNTAX ¼ Synergy Between PCI With Taxus and Cardiac Surgery; TIA ¼ transient ischemic
attack; TO ¼ total occlusion; VD ¼ vessel disease.
or revascularized TOs was also similar to those When TOs existed in left main coronary artery
without, in the PCI arm (adjusted HR: 0.992; and/or LAD, the status of TO recanalization or
95% CI: 0.474 to 2.075; p ¼ 0.982) and in the revascularization did not have an impact on the
CABG arm (adjusted HR: 0.656; 95% CI: 0.281 to 10-year all-cause mortality (TO recanalization or
1.533; p ¼ 0.330), and these results were consistent revascularization 34.5% vs. non-TO recanalization or
in the mortality at maximum available follow-up revascularization 26.9%; adjusted HR: 0.896; 95% CI:
(Table 2). 0.314 to 2.555; p ¼ 0.837) (Table 2).
T A B L E 2 Association Between the Status of TO Recanalization or Revascularization and All-Cause Mortality at 10 Years and at Maximum Available Follow-up
TO Recanalization Non-TO Recanalization

or Revascularization or Revascularization Unadjusted Adjusted
(n ¼ 238) (n ¼ 222) HR (95% CI) p Value HR (95% CI) p Value
At 10 yrs
Overall (N ¼ 460) 67 (28.9) 56 (26.3) 1.112 (0.780–1.586) 0.557 0.900 (0.545–1.485) 0.680
PCI (n ¼ 237) 30 (29.9) 38 (29.4) 1.041 (0.645–1.681) 0.868 0.992 (0.474–2.075) 0.982
CABG (n ¼ 223) 37 (28.0) 18 (21.4) 1.311 (0.746–2.303) 0.346 0.656 (0.281–1.533) 0.330
Patients with at least 1 TO in LM and/or LAD (n ¼ 140) 23 (34.5) 19 (26.9) 1.404 (0.765–2.579) 0.274 0.896 (0.314–2.555) 0.837
Patients with at least 1 TO except in LM and/or LAD (n ¼ 320) 44 (26.6) 37 (25.8) 1.002 (0.647–1.552) 0.992 0.883 (0.454–1.715) 0.713
At maximum available follow-up
Overall (N ¼ 460) 75 (35.6) 71 (39.6) 0.951 (0.687–1.317) 0.762 0.894 (0.559–1.429) 0.640
PCI (n ¼ 237) 32 (34.0) 45 (39.8) 0.936 (0.595–1.473) 0.775 1.003 (0.488–2.061) 0.994
CABG (n ¼ 223) 43 (36.7) 26 (39.1) 0.986 (0.603–1.613) 0.955 0.621 (0.288–1.337) 0.223
Patients with at least 1 TO in LM and/or LAD (n ¼ 140) 23 (34.5) 26 (56.3) 1.013 (0.578–1.776) 0.964 0.842 (0.300–2.366) 0.744
Patients with at least 1 TO except in LM and/or LAD (n ¼ 320) 52 (39.5) 45 (33.5) 0.939 (0.62–-1.403) 0.759 0.940 (0.514–1.719) 0.841
Values are n (%) unless otherwise indicates. Number of deaths are the percentage based on Kaplan-Meier estimates. The HR show the risk of all-cause mortality in patients with versus without recanalized or
revascularized TOs.
CI ¼ confidence interval; HR ¼ hazard ratio; other abbreviations as in Table 1.
COMPARISON OF PATIENTS WITH TO STRATIFIED 10 years between patients with successfully

ACCORDING TO LM OR 3VD. When patients with TO recanalized or revascularized TOs and those
were stratified according to the disease type (LM or without, irrespective of the assigned treatment
3VD), there were no significant differences in the (PCI or CABG) and location of TOs (left main coro-
10-year all-cause mortality between PCI and CABG nary artery and/or LAD or other vessels).
(LM: PCI 30.5% vs. CABG 40.9%; HR: 1.539; 95% CI: 2. When patients with TO were stratified according to
0.814 to 2.911; p ¼ 0.185; and 3VD: PCI 29.3% vs. the type of disease (3VD or LM), there was no dif-
21.0%; HR: 0.673; 95% CI: 0.437 to 1.037; p ¼ 0.073) ference between the PCI and CABG arm in terms of
(Figure 2, Table 3). the 10-year mortality.
COMPARISON OF PATIENTS WITH OR WITHOUT TO
In the present study, among patients with TOs, the
WHO UNDERWENT PCI ACCORDING TO THE RESID-
success of revascularization approach was less
UAL SYNTAX SCORE. Table 4 shows the association
frequent with PCI than with CABG (43.5% vs. 60.5%;
between the residual SYNTAX score and all-cause p < 0.001). This relatively low TO success rate with
mortality at 10 years and at maximum available PCI may be due to the complexity of the TO lesion or
follow-up. Irrespective of the status of TOs, patients the unavailability of contemporary devices and
with the residual SYNTAX >8 (TO: 45.5% and non-TO: techniques during the SYNTAX trial enrollment (2005
54.7%) had worse survival up to 10 years compared to 2007). At that time, dedicated contemporary TO
with the residual SYNTAX score #8 (TO: 22.4% and classifications such as the J-CTO (Multicenter CTO
non-TO: 24.6%) (Figure 3, Table 4). These results were Registry in Japan) score (26), PROGRESS CTO (Pro-
consistent in the mortality at maximum available spective Global Registry for the Study of Chronic To-
follow-up (Table 4). tal Occlusion Intervention) score (27), and EuroCTO
ANGINA FREQUENCY IN PATIENTS WITH TO. The (CASTLE [previous CABG; Age >70 years; Stump
SAQ-AF score per each time point in patients with at anatomy, blunt or no; severe Tortuosity; Length of
least 1 TO is presented in Supplemental Figure 1. Both CTO >20 mm; Extent of calcification, >50% of the
CABG and PCI improved the ratios of patients who segment) score (28) did not exist, and therefore these
were angina-free at all follow-up points. scores were not used, although the anatomical SYN-
TAX score incorporated some components of above-
DISCUSSION
mentioned scores. The SYNTAX II trial enrolled
patients approximately 1 decade later between 2014
The main findings of this study are summarized as
and 2017 and investigated the impact of a contem-
follows:
porary PCI strategy on clinical outcomes in patients
1. Among patients with TOs (n ¼ 460), there were no with 3VD; the procedural success rate of TO recana-
significant differences in the all-cause mortality at lization was as high as 87% (29,30), presumably
F I G U R E 2 All-Cause Mortality at 10 Years According to Randomized Treatments of PCI or CABG in Patients With TO Stratified Into
LM or 3VD
50
40 40.9%
30.5%
30
29.3%
20 21.0%
10
Log-rank p = 0.021
0
0 1 2 3 4 5 6 7 8 9 10
61 58 57 56 54 53 51 50 46 42 40
50 45 44 42 40 38 34 33 30 28 28
176 168 164 159 154 147 137 135 128 122 115
173 169 165 161 159 154 146 143 139 132 129
LM-PCI
HR: 1.539; 95% CI: 0.814-2.911
LM-CABG
3VD-PCI
HR: 0.673; 95% CI: 0.437-1.037
3VD-CABG
Kaplan-Meier curves of all-cause mortality at 10 years according to randomized treatments of PCI or CABG in patients with TO stratified into
left main disease (LM) or 3-vessel disease (3VD) (TO population in SYNTAXES study, n ¼ 460). LM-PCI (blue line) versus LM-CABG (red line)
versus 3VD-PCI (gray dotted line) versus 3VD-CABG (purple dotted line). CI ¼ confidence interval; HR ¼ hazard ratio; other abbreviations as
in Figure 1.
because of improvements in technology and tech- compared to a conservative strategy, although major
nique of PCI for a TO such as the diversity of guide- adverse cardiac events were comparable between the
wires to cross a TO and the combination of antegrade 2 groups (33). In the DECISION-CTO (Drug-Eluting
and retrograde approach (31,32). Stent Implantation Versus Optimal Medical Treat-
To the best of our knowledge, this is the first ment in Patients With Chronic Total Occlusion) trial,
study evaluating the impact of TO recanalization which is the largest randomized trial of the CTO-PCI,
or revascularization on long-term ($10-year) mortal- during a median follow-up of 4.0 years, there was
ity in patients with complex CAD undergoing also no significant difference in the incidence of ma-
PCI or CABG. Regardless of the assigned treatment jor adverse cardiovascular events with the CTO-PCI
and location of TOs, the status of TO recanalization compared with incidence with the conservative
or revascularization did not influence the all- strategy (34). Our findings seem to corroborate
cause mortality at 10 years and at maximum avail- these results and suggest that long-term all-cause
able follow-up. The randomized EuroCTO trial mortality is similar between patients with successful
demonstrated a higher rate of freedom from angina in TO recanalization or revascularization and those
the CTO patients undergoing PCI at 1 year as without, and these results persist beyond 5 years.
The lack of association between recanalization or

T A B L E 3 Association Between the Randomized Treatments of PCI or CABG and All-Cause
revascularization status of TOs and long-term benefit Mortality at 10 Years and at Maximum Available Follow-Up According to the Disease Type
in the mortality might be related to the indiscriminate (3VD or LM) in Patients With TO
selection of TOs treated in the absence of myocar-
PCI CABG
dium viability assessment prior to treatment of TO (n ¼ 237) (n ¼ 223) HR (95% CI) p Value
lesions. In the SYNTAX trial, there was no mandated At 10 yrs

viability assessment, so that the revascularization Patients with 3VD (n ¼ 349) 50 (29.3) 35 (21.0) 0.673 (0.437–1.037) 0.073
strategy was based on anatomic findings without Patients with LM (n ¼ 111) 18 (30.5) 20 (40.9) 1.539 (0.814–2.911) 0.185
taking into consideration the extent of viable and
Patients with 3VD (n ¼ 349) 57 (37.4) 45 (32.4) 0.750 (0.507–1.108) 0.149
ischemic myocardium. To assess myocardium
Patients with LM (n ¼ 111) 20 (37.1) 24 (55.7) 1.641 (0.906–2.975) 0.102
viability subtended by a TO lesion, cardiac magnetic
resonance imaging with late gadolinium enhance- Values are n (%) unless otherwise indicated. Number of deaths are percentage based on Kaplan-Meier estimates.
The HR show the risk of all-cause death in patients undergoing PCI versus CABG.
ment can be used for assessing the presence of scar in
Abbreviations as in Tables 1 and 2.
the myocardium (35). The assessment of myocardium
viability by cardiac magnetic resonance imaging
might be helpful to decide whether recanalization or myocardial band (CK-MB) were used as primary car-
revascularization of TOs should be attempted. diac enzymes to rule out patients presented with
Finally, in our analysis, PCI improved the ratio of acute myocardial infarction. CK-MB was required
patients who were angina-free from 18.6% (baseline) only if CK was $2 the upper limit of normal before
to 63.8% (1 month), 63.6% (6 months), 67.6% (1 year), the procedure. Of 1,800 patients, 9 patients (6 pa-
68.2% (3 years), and 70.3% (5 years) (Supplemental tients with stable angina and 3 patients with unstable
Figure 1). The present study might support contem- angina) had pre-procedural elevated CK-MB values
porary practice among high-volume CTO-PCI centers (37). The rest of the patients (1,791 of 1,800; 99.5%)
where recanalization is primarily offered to patients did not have the elevated cardiac enzyme. In patients
for the management of angina refractory to medical with at least 1 TO (460 patients), only 2 patients
therapy when myocardial viability is confirmed. fulfilled the criteria of the pre-procedural elevated
STUDY LIMITATIONS. The present study is a post hoc CK-MB value. We could not completely exclude the
analysis and should be considered only as possibility where the TOs were “recent” occlusion
hypothesis-generating (36). By protocol design of the with or without the elevation of the cardiac enzyme.
SYNTAX trial, patients with acute myocardial infarc- Due to the unavailability of pre-procedural angiog-
tion were excluded. In 1,800 patients randomized in raphies 3 months prior to the procedure and specific
the SYNTAX trial, 260 presented with silent coronary anamnesis pertaining to a sudden episode of chest
ischemia, 1,027 presented with stable angina, and 513 pain possibly due to vessel occlusion, the age of TOs
presented with unstable angina. At the time of could not be clearly defined, with most cases
recruitment (March 2005 through April 2007), pre- being unknown. Furthermore, the low frequency of
procedure, creatine kinase (CK) and creatine kinase- TO recanalization or revascularization might dilute a
T A B L E 4 Association Between the Residual SYNTAX Score and All-Cause Mortality at 10 Years and at Maximum Available Follow-Up
Death/Patients Unadjusted HR (95% CI) p Value Adjusted HR (95% CI) p Value
At 10 yrs
Patients with TO and rSS #8 34/157 (22.4) Reference — Reference —
Patients with TO and rSS >8 34/76 (45.5) 2.539 (1.578–4.084) <0.001 1.683 (1.011–2.802) 0.045
Patients without TO and rSS #8 137/580 (24.6) 1.125 (0.773–1.683) 0.539 1.245 (0.837–1.850) 0.279
Patients without TO and rSS >8 41/77 (54.7) 3.798 (2.409–5.987) <0.001 4.100 (2.531–6.642) <0.001
Patients with TO and rSS #8 39/157 (27.1) Reference — Reference —
Patients with TO and rSS >8 38/76 (54.9) 2.600 (1.663–4.066) <0.001 1.774 (1.105–2.848) 0.018
Patients without TO and rSS #8 181/580 (38.9) 1.311 (0.927–1.853) 0.125 1.366 (0.948–1.967) 0.094
Patients without TO and rSS >8 43/77 (59.1) 3.624 (2.348–5.593) <0.001 4.035 (2.554–6.374) <0.001
Values are n/N (%) unless otherwise indicated. Number of deaths are percentage based on Kaplan-Meier estimates.
rSS ¼ residual SYNTAX score; other abbreviations as in Tables 1 and 2.
F I G U R E 3 All-Cause Mortality at 10 Years According to the Residual SYNTAX Score in Patients With or Without TO Who Underwent PCI
60
54.7%
50
45.5%
40
30
24.6%
22.4%
20
10
Log-rank p < 0.001

0
0 1 2 3 4 5 6 7 8 9 10
No. at risk:
TO, rSS ≤8 157 156 153 150 147 142 133 131 128 121 114
TO, rSS >8 76 66 64 61 57 55 53 52 44 41 39
Non-TO, rSS ≤8 580 570 561 551 538 509 478 461 446 425 398
Non-TO, rSS >8 77 57 55 49 42 41 38 37 36 35 33
Kaplan-Meier curves of all-cause mortality at 10 years according to the residual SYNTAX (Synergy Between PCI With Taxus and Cardiac Surgery) score (rSS)
in patients with or without TO who underwent PCI. TO, rSS #8 (blue line) versus TO, rSS >8 (red line) versus non-TO, rSS #8 (gray line) versus non-TO,
rSS >8 (purple line). Abbreviations as in Figure 1.
potential benefit on mortality. Moreover, beyond evidence for contemporary technology can be only
the initial 5 years of follow up, there are no data on derived from short-term follow-up studies. Finally,
the prevalence of myocardial infarction, rehospitali- the endpoint in the SYNTAXES study was all-cause
zation, and no information on quality of life mortality alone. However, the SYNTAXES study pro-
including angina status. In addition, the recanaliza- vides randomized data that was meticulously
tion or revascularization of TOs was assessed based collected and achieved a high follow-up rate of 93.8%
on the site decision. However, in the SYNTAX trial, for 10-year vital status (1,689 of 1,800 enrolled
the heart team was obliged to state before the patients) (14).
randomization process took place which vessel
needed to be revascularized. Procedural success CONCLUSIONS
rates of recanalization or revascularization of TOs
were available in the electronic case records of the At 10-year follow-up, the status of TO recanalization
SYNTAX population. Moreover, viability assessment or revascularization did not affect mortality, irre-
was not part of the protocol. In addition, the SYNTAX spective of the assigned treatment (PCI or CABG) and
trial was conducted between 2005 and 2007, with a location of TOs (left main coronary artery and/or LAD
predominant use of first-generation paclitaxel- or other vessels). The present study might support
eluting stents for treatment with PCI, which may contemporary practice among high-volume CTO-PCI
limit the generalizability of our findings to current centers where recanalization is primarily offered to
practices. However, it is unavoidable that the find- patients for the management of angina refractory
ings stemming from long-term follow-up data are to medical therapy when myocardial viability is
based on partially outdated technology, whereas the confirmed.
FUNDING SUPPORT AND AUTHOR DISCLOSURES

ADDRESS FOR CORRESPONDENCE: Dr. Patrick W.
The SYNTAXES study was supported by the German Foundation of Serruys, National University of Ireland, Galway,
Heart Research. The SYNTAX trial, during 0- to 5-year follow-up, Cardiology, University Road, Galway H91 TK33,
was funded by Boston Scientific Corporation. Both sponsors had
Ireland. E-mail: patrick.serruys@nuigalway.ie.
no role in the study design, data collection, data analyses and
interpretation of the study data, nor were involved in the decision Twitter: @HideyukiKawash2.
to publish the final manuscript. The principal investigators and
authors had complete scientific freedom. Dr. Hara has received a PERSPECTIVES
grant for studying overseas from the Japanese Circulation Society
and a grant from the Fukuda Foundation for Medical Technology,
outside the submitted work. Dr. Morice is a chief executive office COMPETENCY IN PATIENT CARE AND PROCEDURAL
and shareholder of the Cardiovascular European Research Center, a SKILLS: All-cause mortality is unrelated to coronary
contract research organization based in Paris having no role in this
artery patency 10 years following revascularization, irrespective
trial. Dr. Head is an employee of Medtronic, outside the submitted
work. Dr. Kappetein is an employee of Medtronic, outside the of initial revascularization strategy (percutaneous or surgical)
submitted work. Dr. Serruys has received personal fees from Bio- or anatomical location of total coronary artery occlusions.
sensors, Micell Technologies, Sino Medical Sciences Technology,
Philips/Volcano, Xeltis, and HeartFlow, outside the submitted TRANSLATIONAL OUTLOOK: Further studies are needed to
work. Dr. Onuma has received institutional research grants related
identify the factors that determine long-term clinical outcomes
to his work as the chairman of cardiovascular imaging core labs of
several clinical trials and a registry sponsored by industry, for in patients undergoing revascularization of completely occluded
which he receives no direct compensation. All other authors have coronary arteries.
reported that they have no relationships relevant to the contents of
this paper to disclose.
REFERENCES
1. Tsai TT, Stanislawski MA, Shunk KA, et al. total coronary occlusion angioplasty: analysis from 14. Thuijs D, Kappetein AP, Serruys PW, et al., for
Contemporary incidence, management, and the U.K. Central Cardiac Audit Database. J Am Coll the SYNTAX Extended Survival Investigators.
long-term outcomes of percutaneous coronary Cardiol 2014;64:235–43. Percutaneous coronary intervention versus
interventions for chronic coronary artery total coronary artery bypass grafting in patients with
8. Mehran R, Claessen BE, Godino C, et al., for the
occlusions: insights from the VA CART program. three-vessel or left main coronary artery disease:
Multinational Chronic Total Occlusion Registry.
J Am Coll Cardiol Intv 2017;10:866–75. 10-year follow-up of the multicentre randomised
Long-term outcome of percutaneous coronary
controlled SYNTAX trial. Lancet 2019;394:
2. Azzalini L, Jolicoeur EM, Pighi M, et al. Epide- intervention for chronic total occlusions. J Am Coll
1325–34.
miology, management strategies, and outcomes of Cardiol Intv 2011;4:952–61.
patients with chronic total coronary occlusion. Am 15. Serruys PW, Onuma Y, Garg S, et al. Assess-
J Cardiol 2016;118:1128–35. 9. Lee PH, Lee SW, Park HS, et al. Successful ment of the SYNTAX score in the Syntax study.
recanalization of native coronary chronic total EuroIntervention 2009;5:50–6.
3. Ramunddal T, Hoebers LP, Henriques JP, et al.
occlusion is not associated with improved long-
Chronic total occlusions in Sweden—a report from 16. Serruys PW, Chichareon P, Modolo R, et al. The
term survival. J Am Coll Cardiol Intv 2016;9:
the Swedish Coronary Angiography and Angio- SYNTAX score on its way out or . . . towards arti-
530–8.
plasty Registry (SCAAR). PLoS One 2014;9: ficial intelligence: part I. EuroIntervention 2020;
e103850. 10. Fefer P, Gannot S, Kochkina K, et al. Impact of 16:44–59.
coronary chronic total occlusions on long-term
4. Levine GN, Bates ER, Blankenship JC, et al. 2011 17. Serruys PW, Chichareon P, Modolo R, et al. The
mortality in patients undergoing coronary artery
ACCF/AHA/SCAI guideline for percutaneous coro- SYNTAX score on its way out or . . . towards arti-
bypass grafting. Interact Cardiovasc Thorac Surg
nary intervention: executive summary: a report of
2014;18:713–6. ficial intelligence: part II. EuroIntervention 2020;
the American College of Cardiology Foundation/
16:60–75.
American Heart Association Task Force on Practice 11. Ong AT, Serruys PW, Mohr FW, et al. The
Guidelines and the Society for Cardiovascular SYNergy Between Percutaneous Coronary Inter- 18. Sianos G, Morel MA, Kappetein AP, et al. The
Angiography and Interventions. Circulation 2011; vention With TAXus and Cardiac Surgery (SYNTAX) SYNTAX Score: an angiographic tool grading the
124:2574–609. study: design, rationale, and run-in phase. Am complexity of coronary artery disease. Euro-
Heart J 2006;151:1194–204. Intervention 2005;1:219–27.
5. Neumann FJ, Sousa-Uva M, Ahlsson A, et al., for
the ESC Scientific Document Group. 2018 ESC/ 12. Serruys PW, Morice MC, Kappetein AP, 19. Farooq V, Serruys PW, Garcia-Garcia HM,
EACTS Guidelines on myocardial revascularization. et al., for the SYNTAX Investigators. Percu- et al. The negative impact of incomplete
Eur Heart J 2019;40:87–165. taneous coronary intervention versus angiographic revascularization on clinical out-
coronary-artery bypass grafting for severe comes and its association with total occlusions:
6. Yamamoto E, Natsuaki M, Morimoto T, et al.,
coronary artery disease. N Engl J Med 2009; the SYNTAX (Synergy Between Percutaneous
for the CREDO-Kyoto PCI/CABG Registry Cohort-2
360:961–72. Coronary Intervention With Taxus and Cardiac
Investigators. Long-term outcomes after percu-
Surgery) trial. J Am Coll Cardiol 2013;61:
taneous coronary intervention for chronic total 13. Mohr FW, Morice MC, Kappetein AP, et al. 282–94.
occlusion (from the CREDO-Kyoto registry cohort- Coronary artery bypass graft surgery versus
2). Am J Cardiol 2013;112:767–74. percutaneous coronary intervention in patients 20. Zhang YJ, Iqbal J, Campos CM, et al. Prog-
7. George S, Cockburn J, Clayton TC, et al., for the with three-vessel disease and left main coro- nostic value of site SYNTAX score and rationale for
British Cardiovascular Intervention Society, Na- nary disease: 5-year follow-up of the rando- combining anatomic and clinical factors in decision
tional Institute for Cardiovascular Outcomes mised, clinical SYNTAX trial. Lancet 2013;381: making: insights from the SYNTAX trial. J Am Coll
Research. Long-term follow-up of elective chronic 629–38. Cardiol 2014;64:423–32.
21. Farooq V, Serruys PW, Bourantas CV, et al. 27. Christopoulos G, Kandzari DE, Yeh RW, et al. approach. Catheter Cardiovasc Interv 2012;79:
Quantification of incomplete revascularization and Development and validation of a novel scoring 3–19.
its association with five-year mortality in the system for predicting technical success of chronic
33. Werner GS, Martin-Yuste V, Hildick-Smith D,
Synergy Between Percutaneous Coronary Inter- total occlusion percutaneous coronary in-
et al., for the EUROCTO Trial Investigators.
vention With Taxus and Cardiac Surgery (SYNTAX) terventions: the PROGRESS CTO (Prospective
A randomized multicentre trial to compare revas-
trial validation of the residual SYNTAX score. Global Registry for the Study of Chronic Total
cularization with optimal medical therapy for the
Circulation 2013;128:141–51. Occlusion Intervention) score. J Am Coll Cardiol
treatment of chronic total coronary occlusions.
Intv 2016;9:1–9.
22. Genereux P, Palmerini T, Caixeta A, et al. Eur Heart J 2018;39:2484–93.
Quantification and impact of untreated coronary 28. Szijgyarto Z, Rampat R, Werner GS, et al.
34. Lee SW, Lee PH, Ahn JM, et al. Randomized
artery disease after percutaneous coronary inter- Derivation and validation of a chronic total coro-
trial evaluating percutaneous coronary interven-
vention: the residual SYNTAX (Synergy Between nary occlusion intervention procedural success
tion for the treatment of chronic total occlusion.
PCI With Taxus and Cardiac Surgery) score. J Am score from the 20,000-patient EuroCTO Registry:
Circulation 2019;139:1674–83.
Coll Cardiol 2012;59:2165–74. the EuroCTO (CASTLE) score. J Am Coll Cardiol
Intv 2019;12:335–42. 35. Kwong RY, Farzaneh-Far A. Measuring
23. Abdallah MS, Wang K, Magnuson EA, et al. for
myocardial scar by CMR. J Am Coll Cardiol Img
the SYNTAX Trial Investigators. Quality of life af- 29. Farooq V, van Klaveren D, Steyerberg EW,
2011;4:157–60.
ter surgery or DES in patients with 3-vessel or left et al. Anatomical and clinical characteristics to
main disease. J Am Coll Cardiol 2017;69:2039–50. guide decision making between coronary artery 36. Milojevic M, Nikolic A, Juni P, Head SJ.
bypass surgery and percutaneous coronary inter- A statistical primer on subgroup analyses. Interact
24. Spertus JA, Winder JA, Dewhurst TA, Deyo RA, Cardiovasc Thorac Surg 2020;30:839–45.
vention for individual patients: development and
Fihn SD. Monitoring the quality of life in patients
validation of SYNTAX score II. Lancet 2013;381: 37. Hara H, Serruys PW, Takahashi K, et al., for the
with coronary artery disease. Am J Cardiol 1994;
639–50. SYNTAX Extended Survival Investigators. Impact
74:1240–4.
30. Escaned J, Collet C, Ryan N, et al. Clinical of peri-procedural myocardial infarction on out-
25. Spertus JA, Winder JA, Dewhurst TA, et al. comes after revascularization. J Am Coll Cardiol
outcomes of state-of-the-art percutaneous coro-
Development and evaluation of the Seattle Angina 2020;76:1622–39.
nary revascularization in patients with de novo
Questionnaire: a new functional status measure
three vessel disease: 1-year results of the SYNTAX
for coronary artery disease. J Am Coll Cardiol
II study. Eur Heart J 2017;38:3124–34.
1995;25:333–41. KEY WORDS coronary artery bypass
31. Surmely JF, Tsuchikane E, Katoh O, et al. New grafting, long-term mortality, percutaneous
26. Morino Y, Abe M, Morimoto T, et al., for the J-
concept for CTO recanalization using controlled coronary intervention, SYNTAX, total
CTO Registry Investigators. Predicting successful
antegrade and retrograde subintimal tracking: the occlusion
guidewire crossing through chronic total occlusion
CART technique. J Invasive Cardiol 2006;18:334–8.
of native coronary lesions within 30 minutes: the
J-CTO (Multicenter CTO Registry in Japan) score 32. Brilakis ES, Grantham JA, Thompson CA,
A PPE NDI X For a supplemental figure and
as a difficulty grading and time assessment tool. et al. The retrograde approach to coronary ar-
tables, please see the online version of this
J Am Coll Cardiol Intv 2011;4:213–21. tery chronic total occlusions: a practical
paper.
Circulation
ORIGINAL RESEARCH ARTICLE
Ten-Year Outcomes After Drug-Eluting

Stents Versus Coronary Artery Bypass
Grafting for Left Main Coronary Disease
Extended Follow-Up of the PRECOMBAT Trial
Editorial, see p XXX Duk-Woo Park, MD*
Jung-Min Ahn, MD*
BACKGROUND: Long-term comparative outcomes after percutaneous Hanbit Park, MD
coronary intervention (PCI) with drug-eluting stents and coronary-artery Sung-Cheol Yun, PhD
bypass grafting (CABG) for left main coronary artery disease are highly Do-Yoon Kang, MD
Pil Hyung Lee, MD
debated.
Young-Hak Kim, MD
METHODS: In the PRECOMBAT trial (Premier of Randomized Comparison Do-Sun Lim, MD
of Bypass Surgery versus Angioplasty Using Sirolimus-Eluting Stent Seung-Woon Rha, MD
in Patients with Left Main Coronary Artery Disease), patients with Gyung-Min Park, MD
Hyeon-Cheol Gwon, MD
unprotected left main coronary artery disease were randomly assigned
Hyo-Soo Kim, MD
to undergo PCI with sirolimus-eluting stents (n=300) or CABG (n=300)
In-Ho Chae, MD
Downloaded from http://ahajournals.org by on April 9, 2020
in 13 hospitals in Korea from April 2004 to August 2009. The follow-up Yangsoo Jang, MD
was extended to at least 10 years for all patients (median, 11.3 years). Myung-Ho Jeong, MD
The primary outcome was the incidence of major adverse cardiac or Seung-Jea Tahk, MD
cerebrovascular events (composite of death from any cause, myocardial Ki Bae Seung, MD
infarction, stroke, or ischemia-driven target-vessel revascularization). Seung-Jung Park , MD
On behalf of the
RESULTS: At 10 years, a primary outcome event occurred in 29.8% of PRECOMBAT
the PCI group and in 24.7% of the CABG group (hazard ratio [HR] with Investigators
PCI vs CABG, 1.25 [95% CI, 0.93–1.69]). The 10-year incidence of the
composite of death, myocardial infarction, or stroke (18.2% vs 17.5%;
HR 1.00 [95% CI, 0.70–1.44]) and all-cause mortality (14.5% vs 13.8%;
HR 1.13 [95% CI, 0.75–1.70]) were not significantly different between
the PCI and CABG groups. Ischemia-driven target-vessel revascularization
was more frequent after PCI than after CABG (16.1% vs 8.0%; HR 1.98
[95% CI, 1.21–3.21).
CONCLUSIONS: Ten-year follow-up of the PRECOMBAT trial of patients
with left main coronary artery disease randomized to PCI or CABG did *Drs D.-W. Park and Ahn contributed
not demonstrate significant difference in the incidence of major adverse equally.
cardiac or cerebrovascular events. Because the study was underpowered, Key Words: coronary artery bypass
the results should be considered hypothesis-generating, highlighting the grafting ◼ coronary artery disease
◼ drug-eluting stents ◼ outcome
need for further research. assessment ◼ percutaneous coronary
intervention ◼ survival
REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifiers:
Sources of Funding, see page XXX
NCT03871127 and NCT00422968.
© 2020 American Heart Association, Inc.
https://www.ahajournals.org/journal/circ
Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039 xxx xxx, 2020 1

Park et al PCI Versus CABG for Left Main Coronary Disease
In the PRECOMBAT trial (Premier of Randomized

Clinical Perspective
ORIGINAL RESEARCH
Comparison of Bypass Surgery versus Angioplasty Us-

ing Sirolimus-Eluting Stent in Patients with Left Main
What Is New?
ARTICLE
Coronary Artery Disease), we randomly assigned pa-

• Long-term outcomes (beyond 5 years and up to
tients with LMCA disease to receive either PCI with
10 years) after percutaneous coronary interven- sirolimus-eluting stents or CABG.6 This trial showed no
tion with drug-eluting stents and coronary-artery significant difference between the 2 strategies in the
bypass grafting for left main coronary artery dis- incidence of major adverse cardiac or cerebrovascular
ease are highly debated and still limited. events and mortality at 2 and 5 years.6,15 To further
• In this 10-year follow-up of the PRECOMBAT trial characterize the long-term outcomes of PCI and CABG
(Premier of Randomized Comparison of Bypass in patients with LMCA disease, we now present the 10-
Surgery versus Angioplasty Using Sirolimus-Eluting year follow-up results of this trial.
Stent in Patients with Left Main Coronary Artery
Disease), there was no significant difference
between percutaneous coronary intervention and METHODS
coronary-artery bypass grafting in the incidence of The data, analytic methods, and study materials will not be
major adverse cardiac or cerebrovascular events, made available to other researchers for purposes of reproduc-
composite of death, myocardial infarction, or ing the results or replicating the procedure.
stroke, and all-cause mortality.
Study Design
What Are the Clinical Implications? The trial design and methods of the PRECOMBAT trial
• This extended follow-up of PRECOMBAT provides (NCT00422968) have been described previously.6,15 Briefly,
important insights on long-term outcomes, which the PRECOMBAT trial was a prospective, open-label, random-
could aid in decision-making for the optimal revas- ized trial that compared PCI with sirolimus-eluting stents with
cularization strategy in patients with left main cor- CABG in patients with LMCA disease. It was conducted at
onary artery disease. 13 hospitals in Korea between April 2004 and August 2009.
• However, our findings should be confirmed or Details of the trial organization and participating centers are
refuted through adequately powered, larger-sized provided in the Data Supplement.
trials with long-term follow-up. Although the PRECOMBAT trial was initially planned to
complete follow-up at 5 years in the original protocol,6 all
participating centers agreed to participate in the extended
A
lthough coronary-artery bypass grafting (CABG) 10-year follow-up study. This extended 10-year follow-up
surgery has traditionally been the mainstay of study was registered at https://www.clinicaltrials.gov as an
treatment for patients with left main coronary investigator-driven extension of follow-up of the PRECOMBAT
trial (NCT03871127) and was funded by the CardioVascular
artery (LMCA) disease, percutaneous coronary inter-
Research Foundation (Seoul, Korea). The sponsor had no role
vention (PCI) has undergone considerable evolution.1,2 in the study design nor in the collection, analyses, or inter-
Technical improvements in stent placement and the de- pretation of data. The institutional review board at each hos-
velopment of drug-eluting stents (DES) led to greater pital approved the protocol, and informed consent to obtain
use of PCI, and many studies have reported favorable information on 10-year outcomes was waived. Follow-up was
outcomes after PCI with DES for this complex disease.3,4 performed in accordance with the local law and regulations
Subsequently, multiple comparisons between the 2 of each participating site and complied with the Declaration
competing revascularization strategies (CABG vs PCI of Helsinki. The principal investigators had unrestricted access
with DES) have been conducted in randomized trials as to the data, prepared the article, and vouch for the complete-
well as registry studies,5–9 in most of which the 2 strate- ness and accuracy of the data and analyses and for the fidelity
of the trial to the protocol.
gies were associated with similar incidence of the com-
posite end point of death, myocardial infarction (MI),
stroke, or all-cause mortality. Patients, Randomization, and Procedures
However, data are still limited on long-term (beyond Patients were eligible for participation in the trial if they had
5 years) outcomes of PCI or CABG in patients with de novo stenosis of the LMCA of more than 50% (as esti-
LMCA disease. Available long-term studies showed mated visually) and had received a diagnosis of stable angina,
unstable angina, silent ischemia, or non–ST-segment eleva-
conflicting results,10–14 and some studies reported a
tion MI. Clinical and anatomic eligibility of all participants had
trend of late catch-up or crossover in the incidence
to be considered by the cardiologists and surgeons at each
of the primary composite outcome or all-cause death hospital to be equivalently suitable for both PCI and CABG. A
favoring CABG over PCI during extended follow- complete list of inclusion and exclusion criteria are provided in
up.11,13,14 Therefore, there remains uncertainty about Table I in the Data Supplement.
long-term outcomes warranting additional longer- Patients were randomly assigned, in a 1:1 ratio, to
term follow-up studies. undergo PCI with sirolimus-eluting stents or CABG. Central
2 xxx xxx, 2020 Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039

randomization was performed using an interactive web-based noninferiority comparison between PCI and CABG with
ORIGINAL RESEARCH
response system in permuted block sizes of 6 and 9, with strat- respect to the primary end point of major adverse cardiac
ification according to the participating center. Details of the or cerebrovascular events.6 All principal analyses were per-
PCI and CABG procedures have been described previously.6,15 formed according to the intention-to-treat principle, in other
ARTICLE
PCI was performed with standard interventional techniques words, treatment groups were defined according to the
according to local practice, and sirolimus-eluting stents were original randomization. A descriptive analysis was performed
used as the default device. Surgical revascularization was per- by presenting data as mean (SD) or number (proportion).
formed with standard bypass techniques, and the internal-tho- Continuous variables were compared with Student t test or
racic-artery graft was preferentially used for the left anterior the Wilcoxon rank-sum test, and categorical variables were
descending coronary artery. Dual antiplatelet therapy (aspirin compared with the χ2 test or Fisher exact test. Cumulative
and clopidogrel) was administered before PCI, and for at least event rates were calculated using the Kaplan-Meier esti-
1 year thereafter. Aspirin was indefinitely used after CABG, mates, with event or censoring times calculated from the
and concomitant use of clopidogrel was at the discretion of date of randomization. Risk differences and corresponding
the operators. During follow-up, guideline-directed medical 95% CIs with the Wald approach were reported. We also
therapy and management of risk factors for secondary preven- compared the primary and secondary outcomes between the
tion were highly recommended for all patients.6,15 2 groups using Cox regression models with robust standard
errors to account for the clustering effect of participating
site. For these models, all available follow-up data were used
Outcome and Follow-Up for long-term outcome analyses without censoring clini-
The primary outcome was a composite of major adverse car- cal events beyond 10 years. Patients lost to follow-up were
diac and cerebrovascular events (ie, death from any cause, included in the analyses for all outcomes by censoring at the
nonfatal MI, nonfatal stroke, or ischemia-driven target vessel date of last follow-up. The proportional-hazards assump-
revascularization). Major secondary outcomes included the tion was confirmed using the Schoenfeld residuals test and
individual components of the primary composite outcome; graphical log-minus-log method;18 no relevant violations of
a composite of death, MI, or stroke; any revascularization; the underlying assumption were found.
and definite stent thrombosis or symptomatic graft occlu- Sensitivity analyses were conducted with the use of the
sion. Outcome definitions are provided in Table II in the Data as-treated analyses (in which patients were compared based
Supplement. All primary and secondary outcome events were on the treatment they actually received) and the per-protocol
centrally adjudicated by an independent clinical-events com- analyses (which included only patients who actually received
mittee, with source documents at each hospital. The extent their randomly assigned treatment). We also assessed the
of disease and a SYNTAX score (Synergy between PCI with consistency of treatment effects in the prespecified subgroups
Taxus and Cardiac Surgery; which was developed while the using Cox regression models with tests for interaction: age
current trial was ongoing and thus measured as post hoc) (<65 vs ≥65 years), sex (male vs female), diabetes mellitus (yes
was independently assessed by an angiographic core labora- vs no), acute coronary syndrome (yes vs no), left main dis-
tory, in which members were blinded to randomization.6 The ease location (ostium or shaft vs distal bifurcation), extent of
SYNTAX score reflects a comprehensive angiographic assess- combined diseased vessels (isolated LMCA disease, or LMCA
ment of the coronary vasculature, with higher SYNTAX scores disease in combination with 1-vessel, 2-vessel, or 3-vessel dis-
indicating more complex coronary artery disease. ease), SYNTAX score category (scores of ≤22 defined as low,
According to the original protocol recommendation, clini- 23–32 as intermediate, and ≥33 as high),5,8,9 and complete
cal follow-up was performed at 1, 6, 9, and 12 months and revascularization (yes vs no). All statistical analyses were per-
then annually through 5 years.6 Ten years after the index formed using SAS software, version 9.4 (SAS Institute).
treatment, all participants in this trial were invited to partici-
pate in 10-year follow-up evaluations. During the extended
follow-up, if a patient was unwilling or unable to return to the
enrolling center, follow-up was maintained by the enrolling
RESULTS
investigators through telephone contact or medical records Patients and Treatment
obtained from other hospitals, as necessary. Information on
From April 2004 through August 2009, a total of 600
adverse clinical events and survival data (vital status, cause of
death, and date of death) was obtained through (electronic)
of patients with unprotected LMCA disease were ran-
healthcare record review and national death registry checks of domly assigned to PCI with sirolimus-eluting stents (300
the Korean National Health Insurance Service database, which patients) or to CABG (300 patients). The baseline clini-
was merged from the Statistics Korea database. The National cal and angiographic characteristics are summarized in
Health Insurance Service is a single-payer program of a univer- Table 1 and well balanced between the PCI and CABG
sal health coverage system in Korea and provides mandatory groups. The mean (±SD) age of the trial participants
health care for all Korean citizens, with an enrollment rate of was 62.3±9.7 years, 76.5% were men, and 32.0% had
more than 97%.16,17 medically treated diabetes mellitus. Distal left main bi-
furcation disease was present in 64.6% of the patients,
Statistical Analysis and the mean SYNTAX score was 24.8±10.3 (low in
This report provides descriptive information on all end point 42.4%, intermediate in 35.3%, and high in 22.3%).
events that occurred during 10-year follow-up. Therefore, Complete revascularization was achieved in 68.3%
we did not perform formal hypothesis testing for the in the PCI group and 70.3% in the CABG group.

Table 1. Baseline Characteristics of the Patients Procedural or operative data are provided in Table III in
ORIGINAL RESEARCH
PCI Group CABG Group the Data Supplement. In the PCI group, intravascular
Characteristic (N=300) (N=300)
ultrasound was used in 91.2% of the patients and a
ARTICLE
Age, y 61.8±10.0 62.7±9.5 mean of 2.7 stents were implanted per patient. In the
Male sex, No. (%) 228 (76.0) 231 (77.0) CABG group, 63.8% underwent off-pump surgery and
Body mass index* 24.6±2.7 24.5±3.0 93.6% underwent revascularization of the left anterior
Diabetes mellitus, No. (%) descending artery with an internal-thoracic-artery graft.
Any diabetes mellitus 102 (34.0) 90 (30.0)
Requiring insulin 10 (3.3) 9 (3.0)

Follow-Up and Outcomes
Hypertension, No. (%) 163 (54.3) 154 (51.3)
Hyperlipidemia, No. (%) 127 (42.3) 120 (40.0)

The median duration of follow-up in all patients was
11.3 years (interquartile range, 10.2 to 13.0; maximum
Current smoker, No. (%) 89 (29.7) 83 (27.7)
follow-up, 14.7 years). The flow of patients through the
Previous PCI, No. (%) 38 (12.7) 38 (12.7)
trial up to 10 years of follow-up are provided in Fig-
Previous myocardial infarction, No. (%) 13 (4.3) 20 (6.7)
ure I in the Data Supplement. Ten-year follow-up for all
Previous congestive heart failure, No. (%) 0 2 (0.7)
clinical end point events was achieved in 288 patients
Chronic renal failure, No. (%) 4 (1.3) 1 (0.3)
(96.0%) randomized to PCI and 288 patients (96.0%)
Peripheral arterial disease, No. (%) 15 (5.0) 7 (2.3) randomized to CABG, respectively. Vital status was veri-
Chronic obstructive pulmonary disease, No. (%) 6 (2.0) 10 (3.3) fied in all patients.
Family history of coronary artery disease, No. (%) 31 (10.3) 19 (6.3) The primary end point of major adverse cardiac or
Clinical presentation, No. (%) cerebrovascular events at 10 years occurred in 29.8% of
Stable angina or silent ischemia 160 (53.3) 137 (45.7) the patients in the PCI group and 24.7% of the patients
Unstable angina 128 (42.7) 144 (48.0) in the CABG group (HR with PCI vs CABG, 1.25 [95%
Recent myocardial infarction 12 (4.0) 19 (6.3) CI, 0.93–1.69]; Table 2 and Figure 1A). There was also
Left ventricular ejection fraction (%) 61.7±8.3 60.6±8.5
no significant between-group difference with respect
Electrocardiographic findings, No./total No. (%)
to the secondary composite outcome of death, MI, or
stroke (18.2 vs 17.5%; HR 1.00 [95% CI, 0.70–1.44])
Sinus rhythm 286/296 (96.6) 289/297 (97.3)
and death from any cause (14.5% vs 13.8%; HR 1.13
Atrial fibrillation 5/296 (1.7) 5/297 (1.7)
[95% CI, 0.75–1.70]) at 10 years (Table 2; Figure 1B

Other 5/296 (1.7) 3/297 (1.0)
and 1C, respectively). The incidence of MI and stroke
EuroSCORE† 2.6±1.8 2.8±1.9
at 10 years did not significantly differ between the 2
Left main disease location, No./total No. (%)
groups. However, the 10-year incidences of ischemia-
Ostium or shaft 99/299 (33.1) 111/294 (37.8) driven target-vessel revascularization and any revascu-
Distal bifurcation 200/299 (66.9) 183/294 (62.2) larization were higher after PCI than after CABG (Ta-
Extent of diseased vessel, No. (%) ble 2 and Figure 1D).
Left main only 27 (9.0) 34 (11.3)
Left main plus 1-vessel disease 50 (16.7) 53 (17.7)

Sensitivity and Subgroup Analyses

We performed an as-treated analysis comparing 327
SYNTAX score by core-laboratory assessment‡
patients who were actually treated with PCI and 272 pa-
tients who were actually treated with CABG (Figure I in
Mean 24.3±9.6 25.3±10.9
the Data Supplement); the HR for the primary outcome
Category, No./total No. (%)
with PCI was 1.51 (95% CI, 1.11–2.06; Table IV in the
Low (≤22) 131/291 (45.0) 109/275 (39.6)
Data Supplement). We also performed a per-protocol
Intermediate (23 to 32) 102/291 (35.1) 98/275 (35.6)
comparison of 276 patients randomly assigned to PCI
High (≥33) 58/291 (19.9) 68/275 (24.7)
who actually received PCI and 248 patients assigned
Complete revascularization 205 (68.3) 211 (70.3) to CABG who actually underwent CABG; the HR for
Plus–minus values are mean±SD. Percentages may not total 100 because of rounding. the primary outcome with PCI was 1.51 (95% CI, 1.09
CABG indicates coronary artery bypass grafting; and PCI, percutaneous coronary intervention.
*Body mass index is the weight in kilograms divided by the square of the height in meters.
to 2.10; Table V in the Data Supplement). In addition,
†The EuroSCORE (European System for Cardiac Operative Risk Evaluation) is a clinical model when we estimated the risks of cause-specific mortality
for calculating the risk of death after cardiac surgery based on patient, cardiac, and operative
factors. Possible scores range from 0 to 39, with higher scores indicating greater risk.
and nonmortality related outcomes in competing-risks
‡The SYNTAX score (Synergy between Percutaneous Coronary Intervention With Taxus framework, overall results were consistent (Table VI in
and Cardiac Surgery) reflects a comprehensive angiographic assessment of the coronary
vasculature, with a score of 22 or less indicating low anatomical complexity and scores of
the Data Supplement).
23 to 32 indicating intermediate anatomical complexity (0 is the lowest score and there is no The treatment effect for the primary outcome in pre-
upper limit).5 The SYNTAX score was measured by angiographic core laboratory assessment
and was available for 291 patients in the PCI group and 275 patients in the CABG group
specified subgroups is shown in Figure 2. The 10-year
who had available angiograms of sufficient image quality to make the assessment accurately. rate of primary outcome between PCI and CABG were

Table 2. Primary and Secondary Outcomes at 10 Years
ORIGINAL RESEARCH
PCI Group CABG Group Risk Difference
(N=300) (N=300) (95% CI)
Hazard Ratio
ARTICLE
Outcomes No. of Events (%) at 10 Years Percentage Points (95% CI)*
Primary outcome
Major adverse cardiac or cerebrovascular events† 87 (29.8) 72 (24.7) 5.2 (−2.1 to 12.4) 1.25 (0.93–1.69)
Secondary outcomes
Death, myocardial infarction, or stroke 53 (18.2) 51 (17.5) 0.7 (−5.6 to 6.9) 1.00 (0.70–1.44)
Death from any cause 42 (14.5) 40 (13.8) 0.7 (−5.0 to 6.4) 1.13 (0.75–1.70)
Cardiovascular cause 22 (7.8) 25 (8.7) −0.9 (−5.5 to 3.6) 0.96 (0.56–1.65)
Noncardiovascular cause 11 (3.9) 8 (2.9) 1.0 (−2.0 to 4.0) 1.55 (0.63–3.81)
Undetermined cause 9 (3.4) 7 (2.7) 0.8 (−2.2 to 3.7) 1.27 (0.50–3.22)
Myocardial infarction 9 (3.2) 8 (2.8) 0.4 (−2.4 to 3.2) 0.76 (0.32–1.82)
Q-wave 4 (1.4) 4 (1.4) −0.02 (−1.9 to 1.9) 0.82 (0.22–3.06)
Non-Q-wave 5 (1.8) 4 (1.4) 0.4 (−1. to 2.5) 0.71 (0.22–2.26)
Stroke 5 (1.9) 6 (2.2) −0.3 (−2.7 to 2.1) 0.71 (0.22–2.23)
Ischemia-driven target-vessel revascularization 45 (16.1) 22 (8.0) 8.1 (2.8 to 13.5) 1.98 (1.21–3.21)
Any revascularization 59 (21.3) 29 (10.6) 10.7 (4.6 to 16.7) 2.04 (1.33–3.11)
Stent thrombosis or symptomatic graft occlusion 4 (1.4) 10 (3.7) −2.3 (−4.9 to 0.3) 0.56 (0.20–1.55)
Event rates (%) shown are the incidences as estimated with the use of a Kaplan-Meier survival analysis of data from the intention-to-treat population.
CABG denotes coronary-artery bypass grafting; and PCI, percutaneous coronary intervention.
*Hazard ratios are for the PCI group as compared with the CABG group. For these models, all available follow-up data were used for long-term
outcome analyses without censoring clinical events beyond 10 years. The CIs that are reported in this table have not been adjusted for multiple testing
and therefore should not be used to infer definitive treatment effects.
†The primary end point of major adverse cardiac or cerebrovascular events was a composite of death from any cause, myocardial infarction, stroke,
or ischemia-driven target-vessel revascularization.
consistent across multiple subgroups, except for those Although cumulative evidence have suggested that
stratified by the extent of concomitant coronary artery PCI with DES is an acceptable alternative to CABG in
disease in which the event rate was higher after PCI than patients with LMCA disease,1–3 the relative benefit of
after CABG in patients with left main and 3-vessel disease. CABG and PCI has been substantially different over
Primary and key secondary outcomes, according time,9,11,13,14 but longer-term studies beyond 5 years
to the SYNTAX score tertiles, are shown in Figures II were still limited. Limited follow-up could penalize the
and III in the Data Supplement. There was no notable CABG group because the long-term benefits of CABG
trend across the ordered SYNTAX score tertiles in the might not be fully evident until 5 to 10 years after re-
incidence of the primary outcome; composite of death, vascularization.19,20 Therefore, the extended follow-up
MI, or stroke; and death from any cause. The rate of of PRECOMBAT provides important insights on long-
ischemia-driven target-vessel revascularization was sig- term outcomes, which could aid in decision-making for
nificantly higher after PCI than after CABG in the high the optimal revascularization strategy in patients with
SYNTAX score group. LMCA disease.
Recently, conflicting long-term findings from sev-
eral studies have been reported.11–14 The 10-year re-
port of the MAINCOMPARE registry (Revascularization
DISCUSSION
for Unprotected Left Main Coronary Artery Stenosis:
PRECOMBAT was a randomized trial specifically tar- Comparison of Percutaneous Coronary Angioplasty
geting patients with LMCA disease. In this longest Versus Surgical Revascularization) showed a benefit of
extended follow-up, we did not detect significant dif- CABG over PCI with DES on mortality and a composite
ference between PCI with sirolimus-eluting stents and of death, Q-wave MI, or stroke after 5 years.11 The
CABG in the primary composite endpoint of major ad- SYNTAX trial showed similar 10-year incidence of all-
verse cardiac or cerebrovascular events at 10 years. In cause death with PCI and CABG for LMCA disease.12
addition, the 10-year incidence of composite of death, The 5-year follow-up of the EXCEL trial (Evaluation of
MI, or stroke, and all-cause mortality were also similar XIENCE versus Coronary Artery Bypass Surgery for Ef-
between the 2 groups. The 10-year rate of ischemia- fectiveness of Left Main Revascularization) reported
driven target-vessel revascularization was 8 percentage no significant difference between PCI and CABG in
points higher with PCI than with CABG. the rate of the primary composite of death, stroke, or

ORIGINAL RESEARCH
ARTICLE
Figure 1. Time-to-event curves for the primary and key secondary outcomes through 10-year follow-up.
A, Results of the analysis of the primary composite outcome of death from any cause, myocardial infarction, stroke, or ischemic-driven target-vessel revasculariza-
tion at 10 years. The results of the analyses for key secondary outcomes are shown: (B) composite of death from any cause, myocardial infarction, or stroke; (C)
death from any cause; and (D) ischemia-driven target-vessel revascularization. Event rates were based on Kaplan-Meier estimates. The hazard ratios are for the
percutaneous coronary intervention (PCI) group as compared with the coronary artery bypass grafting (CABG) group. In each panel, the inset shows the same data
on an enlarged y axis.
MI.13 However, the 5-year incidence of all-cause death after PCI and CABG in several trials and meta-analy-
was significantly higher after PCI than after CABG. ses,9,12,21,22 the excess of all-cause mortality in EXCEL
By contrast, updated 5-year report of the NOBLE might be because of chance mainly driven by noncar-
trial (Nordic-Baltic-British Left Main Revascularisation diovascular deaths. Nonetheless, further studies are
Study) showed that PCI was associated with inferior required to resolve this conflicting issue, because all-
primary composite outcome compared with CABG, cause mortality is the most robust and unbiased index
which was mainly driven by higher rates of nonproc- for clinical assessment, and which is less likely influ-
edural MI and repeat revascularization, but all-cause enced by ascertainment bias.23
mortality was similar.14 In this extended report of the In the present trial, contrary to the intention-to-treat
PRECOMBAT trial, we did not detect a significant dif- analysis, the as-treated and the per-protocol analyses
ference between PCI and CABG in the rates of primary showed that PCI was associated with a higher 10-year
composite of major adverse cardiac or cerebrovascular incidence of primary endpoint compared with CABG,
events and all-cause mortality at 10 years. Recently, which was mainly driven by repeat revascularization.
the discrepancy in the long-term incidence of all-cause The recent study showed that need for repeat revascu-
mortality between trials has been highly debated. larization was independently associated with increased
Given that all-cause deaths were consistently similar risk for all-cause mortality and cardiovascular mortality

ORIGINAL RESEARCH
ARTICLE
Figure 2. Subgroup analyses of the primary composite outcome at 10 years.

Data are shown as the number of primary composite outcome (ie, composite of death from any cause, myocardial infarction, stroke, or ischemic-driven
target-vessel revascularization) events per total number of patients in that subgroup and the event rate. Event rates were based on Kaplan-Meier estimates;
thus, the rate is not the same as the ratio of the numerator and denominator. The hazard ratios are for the percutaneous coronary intervention (PCI) group
as compared with the coronary artery bypass grafting (CABG) group. The confidence intervals that are reported in this figure have not been adjusted for
multiple testing and therefore should not be used to infer definitive treatment effects. The P value for interaction represents the likelihood of interaction
between the subgroups and the treatment.
after LMCA revascularization.24 In this context, it war- the more appropriate revascularization strategy with
rants further studies to determine the potential clinical respect to primary and secondary outcomes. Similar
implications of a higher risk of repeat revascularization findings were also identified in other recent clinical tri-
after PCI than after CABG. However, per-treatment als.8,9,12–14,27 Although it is further determined whether
analyses showed that the imbalance in crossover rates the SYNTAX score should be central to the decision-
between groups modified the results of the primary making process for LMCA revascularization, compre-
intention-to-treat analysis. Particularly, a relatively high hensive approaches combining clinical and anatomic
rate of crossover from the PCI group to the CABG factors could be helpful for enhanced personalized
group could have biased our findings toward a neu- assessment of patient risk.28 Furthermore, a more inte-
tral effect on outcomes. Therefore, this interpretation grated PCI approach that incorporates coronary physi-
should be considered in a provisional and conserva- ology and imaging may substantially improve PCI out-
tive manner. Nevertheless, the per-treatment analyses comes in patients with multivessel or LMCA disease.29
might be informative as they closely mirror real-world The overall rates of adverse events and mortality in
clinical decision making. our trial were substantially lower than the event rates
Current guidelines have adopted the SYNTAX score in other trials.12–14 Although this disparity is not fully
to aid the choice of the appropriate revascularization elucidated, it may be partly explained by the differ-
strategy in patients with LMCA disease.25,26 However, ences in clinical or lesion characteristics, procedural
in our trial, SYNTAX score tertiles did not discriminate practice, or race or ethnicity. For instance, intravascular

ultrasound was performed in >90% of patients for effective but also safer than first-generation DES and
ORIGINAL RESEARCH
stent optimization and the proportion of off-pump bare-metal stents.32,33

surgery was high in our study. Also, the mean SYNTAX
ARTICLE
and EuroSCORE (European System for Cardiac Opera-

tive Risk Evaluation) were lower in our study than in Conclusions
the SYNTAX left main substudy.6,30 Nevertheless, the In this 10-year follow-up of the PRECOMBAT trial that
relative effect of PCI and CABG might be fairly tested enrolled patients with LMCA disease, there was no
in each trial setting. In addition, clinical practice stan- significant difference between PCI and CABG in the
dards at the institutions participating in this trial, as incidence of major adverse cardiac or cerebrovascular
well as the expertise of the interventional cardiologists events, composite of death, MI, or stroke, and all-cause
and cardiac surgeons who performed the procedures, mortality. However, the study had insufficient statisti-
may differ from those of other institutions and practical power to allow for a firm conclusion, hence further
tioners, potentially limiting the reproducibility of these research is needed in this area.
results in other settings.
The protocol definition of MI considerably varied
among trials5,8,9 and specific criteria of various defini- ARTICLE INFORMATION
Received January 28, 2020; accepted March 2, 2020.
tions can penalize a specific treatment group.31 We
The Data Supplement is available with this article at https://www.ahajournals.
used the original protocol definition of MI;6 among org/doi/suppl/10.1161/circulationaha.120.046039.
several criteria for MI, this definition may be the most
stringent (which included only new pathologic Q-wave Correspondence
after procedure MI during the index hospitalization and Seung-Jung Park, MD, Department of Cardiology, Asan Medical Center, Uni-
clinically-driven spontaneous MI during follow-up). It versity of Ulsan College of Medicine, 88, Olympic-ro 43-gil, Songpa-gu, Seoul,
05505, Korea. Email sjpark@amc.seoul.kr
might partly explain the lower rates of MI in our study.
To diminish uncertainty and to minimize ascertainment
Affiliations
bias for MI, further research and consensus are war-
Department of Cardiology (D.-W.P., J.-M.A., H.P, D.-Y.K., P.H.L., Y.-H.K., S.-J.P.),
ranted to implement a more applicable definition of
and Division of Biostatistics (S.-C.Y.), Asan Medical Center, Seoul, Korea. Korea
periprocedural MI not penalizing a specific revascular- University Anam Hospital, Seoul (D.-S.L.). Korea University Kuro Hospital, Seoul
ization group.31 (S.-W.R.). Ulsan University Hospital, Korea (G.-M.P.). Samsung Medical Center,
Seoul, Korea (H.-C.G., H.-S.K.). Seoul National University Hospital, Korea (H.-
Our study had several limitations. First, as PRECOM-
S.K.). Seoul National University Bundan Hospital, Seongnam, Korea (I.-H.C.). Yon-
BAT was an open-label trial, nonfatal outcomes could sei University Severance Hospital, Seoul, Korea (Y.J.). Chonnam National Univer-
have been influenced by the knowledge of the treat- sity Hospital, Gwangju, Korea (M.-H.J.). Ajou University Medical Center, Suwon,
ment received (ie, ascertainment bias). Second, the Korea (S.-J.T.). Catholic University of Korea, St. Mary’s Hospital, Seoul (K.B.S.).
limited reproducibility of trial findings in real-world

settings should be considered. Third, owing to the Acknowledgments
limited number of patients and low event rates, this The authors thank the staff of the PRECOMBAT trial, the other members of the
cardiac catheterization laboratories and the cardiovascular surgery departments
trial did not have sufficient statistical power to detect at the participating centers, and the study coordinators for their efforts in col-
a clinically significant difference in end points. Also, lecting clinical data and ensuring the accuracy and completeness of the data.
we cannot exclude the possibility that clinical events
were underreported because of the nonprespecified Sources of Funding
10-year follow-up and lack of yearly follow-up be- This study was partly supported by the CardioVascular Research Foundation
tween 5 and 10 years. Fourth, long-term medication (Seoul, Korea). The sponsors had no role in the design and conduct of the study;
collection, management, analysis, and interpretation of the data; preparation,
use after PCI and CABG varied, which reflects differ- review, or approval of the article; or decision to submit the article for publication.
ences in practice with respect to the 2 different treat-
ments. Unfortunately, we did not capture detailed Disclosures
information on concurrent cardiovascular medications Dr D.W. Park reports grants from Daiichi-Sankyo, ChongKunDang Pharm, and
during long-term follow-up. Although the extent to Daewoong Pharm, personal fees from Edwards, grants and personal fees from
which variability in medication use contributed to the Abott Vascular, and personal fees from Medtronic, all outside the submitted
work. Dr G.-M. Park reports grants from Yuhan Pharm outside the submitted
present results is uncertain, unmeasured confounding work. Dr S.-J. Park reports grants and personal fees from Abott Vascular, grants
owing to differences in subsequent medication care from Daiichi-Sankyo, grants from ChongKunDang Pharm and Daewoong
cannot be ruled out. Finally, because we evaluated the Pharm, grants and personal fees from Edwards, all outside the submitted work.
The other authors report no conflicts.
first-generation sirolimus-eluting stents in compari-
son with CABG, our findings should be confirmed or Supplemental Materials
refuted through 10-year (or longer) follow-up of the
Appendix (Trial Investigators, Participating Centers, and Organization)
recent EXCEL and NOBLE trials involving contempo- Data Supplement Tables I–VI
rary DES. Newer-generation DES were not only more Data Supplement Figures I–III

REFERENCES the PRECOMBAT study. J Am Coll Cardiol. 2015;65:2198–2206. doi:

10.1016/j.jacc.2015.03.033
ORIGINAL RESEARCH
1. Lee PH, Ahn JM, Chang M, Baek S, Yoon SH, Kang SJ, Lee SW, Kim YH, 16. Son JS, Choi S, Kim K, Kim SM, Choi D, Lee G, Jeong SM, Park SY, Kim YY,
Lee CW, Park SW, et al. Left main coronary artery disease: secular trends Yun JM, et al. Association of blood pressure classification in Korean young
in patient characteristics, treatments, and outcomes. J Am Coll Cardiol.
ARTICLE
adults according to the 2017 American College of Cardiology/American
2016;68:1233–1246. doi: 10.1016/j.jacc.2016.05.089 Heart Association guidelines with subsequent cardiovascular disease
2. Park DW, Ahn JM, Park SJ, Taggart DP. Percutaneous coronary interven- events. JAMA. 2018;320:1783-1792. doi: 10.1001/jama.2018.16501.
tion in left main disease: SYNTAX, PRECOMBAT, EXCEL and NOBLE-com- 17. Yun JE, Kim YJ, Park JJ, Kim S, Park K, Cho MS, Nam GB, Park DW.
bined cardiology and cardiac surgery perspective. Ann Cardiothorac Surg. Safety and effectiveness of contemporary P2Y12 inhibitors in an east
2018;7:521–526. doi: 10.21037/acs.2018.04.04
Asian population with acute coronary syndrome: a nationwide pop-
3. Park DW, Park SJ. Percutaneous coronary intervention of left main dis-
ulation-based cohort study. J Am Heart Assoc. 2019;8:e012078. doi:
ease: pre- and post-EXCEL (Evaluation of XIENCE Everolimus Eluting
10.1161/JAHA.119.012078
Stent Versus Coronary Artery Bypass Surgery for Effectiveness of Left
18. Cain KC, Lange NT. Approximate case influence for the proportional haz-
Main Revascularization) and NOBLE (Nordic-Baltic-British Left Main Re-
ards regression model with censored data. Biometrics. 1984;40:493–499.
vascularization Study) era. Circ Cardiovasc Interv. 2017;10:e004792. doi:
19. Alexander JH, Smith PK. Coronary-artery bypass grafting. N Engl J Med.
10.1161/circinterventions.117.004792.
2016;375:e22. doi: 10.1056/NEJMc1608042
4. Lee PH, Kwon O, Ahn JM, Lee CH, Kang DY, Lee JB, Kang SJ, Lee SW,
20. Velazquez EJ, Lee KL, Jones RH, Al-Khalidi HR, Hill JA, Panza JA, Michler RE,
Kim YH, Lee CW, et al. Safety and effectiveness of second-generation
Bonow RO, Doenst T, Petrie MC, et al; STICHES Investigators. Coronary-
drug-eluting stents in patients with left main coronary artery disease. J Am
artery bypass surgery in patients with ischemic cardiomyopathy. N Engl J
Coll Cardiol. 2018;71:832-841. doi: 10.1016/j.jacc.2017.12.032.
Med. 2016;374:1511–1520. doi: 10.1056/NEJMoa1602001
5. Serruys PW, Morice MC, Kappetein AP, Colombo A, Holmes DR, Mack MJ,
21. Head SJ, Milojevic M, Daemen J, Ahn JM, Boersma E, Christiansen EH,
Ståhle E, Feldman TE, van den Brand M, Bass EJ, et al; SYNTAX Investi-
Domanski MJ, Farkouh ME, Flather M, Fuster V, et al. Mortality after coro-
gators. Percutaneous coronary intervention versus coronary-artery bypass
nary artery bypass grafting versus percutaneous coronary intervention with
grafting for severe coronary artery disease. N Engl J Med. 2009;360:961–
stenting for coronary artery disease: a pooled analysis of individual patient
972. doi: 10.1056/NEJMoa0804626
data. Lancet. 2018;391:939–948. doi: 10.1016/S0140-6736(18)30423-9
6. Park SJ, Kim YH, Park DW, Yun SC, Ahn JM, Song HG, Lee JY, Kim WJ,
22. Palmerini T, Serruys P, Kappetein AP, Genereux P, Riva DD, Reggiani LB,
Kang SJ, Lee SW, et al. Randomized trial of stents versus bypass surgery
Christiansen EH, Holm NR, Thuesen L, Makikallio T, et al. Clinical out-
for left main coronary artery disease. N Engl J Med. 2011;364:1718–1727.
comes with percutaneous coronary revascularization vs coronary artery
doi: 10.1056/NEJMoa1100452
bypass grafting surgery in patients with unprotected left main coronary
7. Seung KB, Park DW, Kim YH, Lee SW, Lee CW, Hong MK, Park SW, Yun SC,
artery disease: a meta-analysis of 6 randomized trials and 4,686 patients.
Gwon HC, Jeong MH, et al. Stents versus coronary-artery bypass grafting
Am Heart J. 2017;190:54–63. doi: 10.1016/j.ahj.2017.05.005
for left main coronary artery disease. N Engl J Med. 2008;358:1781–1792.
23. Lauer MS, Blackstone EH, Young JB, Topol EJ. Cause of death in clinical
doi: 10.1056/NEJMoa0801441
research: time for a reassessment? J Am Coll Cardiol. 1999;34:618–620.
8. Stone GW, Sabik JF, Serruys PW, Simonton CA, Généreux P, Puskas J,
Kandzari DE, Morice MC, Lembo N, Brown WM 3rd, et al; EXCEL Trial doi: 10.1016/s0735-1097(99)00250-8
Investigators. Everolimus-eluting stents or bypass surgery for left main 24. Giustino G, Serruys PW, Sabik JF 3rd, Mehran R, Maehara A, Puskas
coronary artery disease. N Engl J Med. 2016;375:2223–2235. doi: JD, Simonton CA, Lembo NJ, Kandzari DE, Morice MC, et al. Mortality
10.1056/NEJMoa1610227 after repeat revascularization following PCI or CABG for left main dis-
ease: the EXCEL trial. JACC Cardiovasc Interv. 2020;13:375–387. doi:
9. Mäkikallio T, Holm NR, Lindsay M, Spence MS, Erglis A, Menown IB,

Trovik T, Eskola M, Romppanen H, Kellerth T, et al; NOBLE study inves- 10.1016/j.jcin.2019.09.019
tigators. Percutaneous coronary angioplasty versus coronary artery by- 25. Fihn SD, Blankenship JC, Alexander KP, Bittl JA, Byrne JG, Fletcher BJ,
pass grafting in treatment of unprotected left main stenosis (NOBLE): Fonarow GC, Lange RA, Levine GN, Maddox TM, et al. 2014 ACC/AHA/
a prospective, randomised, open-label, non-inferiority trial. Lancet. AATS/PCNA/SCAI/STS focused update of the guideline for the diag-
2016;388:2743–2752. doi: 10.1016/S0140-6736(16)32052-9 nosis and management of patients with stable ischemic heart disease:
10. Buszman PE, Buszman PP, Banasiewicz-Szkróbka I, Milewski KP, a report of the American College of Cardiology/American Heart As-
Żurakowski A, Orlik B, Konkolewska M, Trela B, Janas A, Martin JL, et al. sociation Task Force on Practice Guidelines, and the American Asso-
Left main stenting in comparison with surgical revascularization: 10-year ciation for Thoracic Surgery, Preventive Cardiovascular Nurses Asso-
outcomes of the (Left Main Coronary Artery Stenting) LE MANS Trial. JACC ciation, Society for Cardiovascular Angiography and Interventions, and
Cardiovasc Interv. 2016;9:318–327. doi: 10.1016/j.jcin.2015.10.044 Society of Thoracic Surgeons. Circulation. 2014;130:1749–1767. doi:
11. Park DW, Ahn JM, Yun SC, Yoon YH, Kang DY, Lee PH, Lee SW, 10.1161/CIR.0000000000000095
Park SW, Seung KB, Gwon HC, et al. 10-year outcomes of stents versus 26. Windecker S, Neumann FJ, Juni P, Sousa-Uva M, Falk V. Considerations
coronary artery bypass grafting for left main coronary artery disease. for the choice between coronary artery bypass grafting and percutane-
J Am Coll Cardiol. 2018;72(23 Pt A):2813–2822. doi: 10.1016/j.jacc. ous coronary intervention as revascularization strategies in major cat-
2018.09.012 egories of patients with stable multivessel coronary artery disease: an
12. Thuijs DJFM, Kappetein AP, Serruys PW, Mohr FW, Morice MC, Mack MJ, accompanying article of the task force of the 2018 ESC/EACTS guide-
Holmes DR Jr, Curzen N, Davierwala P, Noack T, et al; SYNTAX Extend- lines on myocardial revascularization. Eur Heart J. 2019;40:204-212. doi:
ed Survival Investigators. Percutaneous coronary intervention versus 10.1093/eurheartj/ehy532.
coronary artery bypass grafting in patients with three-vessel or left 27. Esper RB, Farkouh ME, Ribeiro EE, Hueb W, Domanski M, Hamza TH, Siami
main coronary artery disease: 10-year follow-up of the multicentre ran- FS, Godoy LC, Mathew V, French J, et al. SYNTAX score in patients with
domised controlled SYNTAX trial. Lancet. 2019;394:1325–1334. doi: diabetes undergoing coronary revascularization in the FREEDOM trial. J Am
10.1016/S0140-6736(19)31997-X Coll Cardiol. 2018;72(23 Pt A):2826–2837. doi: 10.1016/j.jacc.2018.09.046
13. Stone GW, Kappetein AP, Sabik JF, Pocock SJ, Morice MC, Puskas J, 28. Farooq V, van Klaveren D, Steyerberg EW, Meliga E, Vergouwe Y,
Kandzari DE, Karmpaliotis D, Brown WM 3rd, Lembo NJ, et al; EX- Chieffo A, Kappetein AP, Colombo A, Holmes DR, Jr., Mack M, et al.
CEL Trial Investigators. Five-year outcomes after PCI or CABG for left Anatomical and clinical characteristics to guide decision making between
main coronary disease. N Engl J Med. 2019;381:1820–1830. doi: coronary artery bypass surgery and percutaneous coronary intervention
10.1056/NEJMoa1909406 for individual patients: development and validation of SYNTAX score II.
14. Holm NR, Mäkikallio T, Lindsay MM, Spence MS, Erglis A, Menown IBA, Lancet. 2013;381:639-650. doi: 10.1016/s0140-6736(13)60108-7.
Trovik T, Kellerth T, Kalinauskas G, Mogensen LJH, et al; NOBLE investi- 29. Park DW, Park SJ. Contemporary state-of-the-art PCI with functional and
gators. Percutaneous coronary angioplasty versus coronary artery bypass imaging concepts: forethoughts on the FAME 3 trial. EuroIntervention.
grafting in the treatment of unprotected left main stenosis: updated 2019;15:e219–e221. doi: 10.4244/EIJV15I3A40
5-year outcomes from the randomised, non-inferiority NOBLE trial. Lan- 30. Morice MC, Serruys PW, Kappetein AP, Feldman TE, Ståhle E,
cet. 2020;395:191–199. doi: 10.1016/S0140-6736(19)32972-1 Colombo A, Mack MJ, Holmes DR, Choi JW, Ruzyllo W, et al. Five-year
15. Ahn JM, Roh JH, Kim YH, Park DW, Yun SC, Lee PH, Chang M, outcomes in patients with left main disease treated with either per-
Park HW, Lee SW, Lee CW, et al. Randomized trial of stents versus by- cutaneous coronary intervention or coronary artery bypass graft-
pass surgery for left main coronary artery disease: 5-year outcomes of ing in the synergy between percutaneous coronary intervention with

taxus and cardiac surgery trial. Circulation. 2014;129:2388–2394. doi: Long-term safety of drug-eluting and bare-metal stents: evidence from a
10.1161/CIRCULATIONAHA.113.006689 comprehensive network meta-analysis. J Am Coll Cardiol. 2015;65:2496–
ORIGINAL RESEARCH
31. Cho MS, Ahn JM, Lee CH, Kang DY, Lee JB, Lee PH, Kang SJ, Lee SW, 2507. doi: 10.1016/j.jacc.2015.04.017
Kim YH, Lee CW, et al. Differential rates and clinical significance of peri- 33. Piccolo R, Bonaa KH, Efthimiou O, Varenne O, Baldo A, Urban P, Kaiser C,
ARTICLE
procedural myocardial infarction after stenting or bypass surgery for mul- Remkes W, Räber L, de Belder A, et al; Coronary Stent Trialists’ Collabo-
tivessel coronary disease according to various definitions. JACC Cardiovasc ration. Drug-eluting or bare-metal stents for percutaneous coronary
Interv. 2017;10:1498–1507. doi: 10.1016/j.jcin.2017.05.051 intervention: a systematic review and individual patient data meta-
32. Palmerini T, Benedetto U, Biondi-Zoccai G, Della Riva D, Bacchi-Reggiani L, analysis of randomised clinical trials. Lancet. 2019;393:2503–2510. doi:
Smits PC, Vlachojannis GJ, Jensen LO, Christiansen EH, Berencsi K, et al. 10.1016/S0140-6736(19)30474-X

ª 2017 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN 0735-1097/$36.00
PUBLISHED BY ELSEVIER https://doi.org/10.1016/j.jacc.2017.10.029
Surgical Versus Percutaneous Coronary

Revascularization in Patients With
Diabetes and Acute Coronary Syndromes
Krishnan Ramanathan, MB, CHB,a James G. Abel, MD,a Julie E. Park, MMATH,b Anthony Fung, MBBS,a
Verghese Mathew, MD,c Carolyn M. Taylor, MD,a G.B. John Mancini, MD,a Min Gao, MD, PHD,b Lillian Ding, MSC,d
Subodh Verma, MD, PHD,e Karin H. Humphries, DSC,a,b Michael E. Farkouh, MD, MSCf
ABSTRACT
BACKGROUND Randomized trial data support the superiority of coronary artery bypass grafting (CABG) surgery over
percutaneous coronary intervention (PCI) in diabetic patients with multivessel coronary artery disease (MV-CAD).
However, whether this benefit is seen in a real-world population among subjects with stable ischemic heart disease
(SIHD) and acute coronary syndromes (ACS) is unknown.
OBJECTIVES The main objective of this study was to assess the generalizability of the FREEDOM (Future REvascularization
Evaluation in Patients with Diabetes Mellitus: Optimal Management of Multi-vessel Disease) trial in real-world practice among
patients with diabetes mellitus and MV-CAD in residents of British Columbia, Canada. Additionally, the study evaluated the
impact of mode of revascularization (CABG vs. PCI with drug-eluting stents) in diabetic patients with ACS and MV-CAD.
METHODS In a large population-based database from British Columbia, this study evaluated major cardiovascular
outcomes in all diabetic patients who underwent coronary revascularization between 2007 and 2014 (n ¼ 4,661, 2,947
patients with ACS). The primary endpoint (major adverse cardiac or cerebrovascular events [MACCE]) was a composite of
all-cause death, nonfatal myocardial infarction, and nonfatal stroke. The risk of MACCE with CABG or PCI was compared
using multivariable adjustment and a propensity score model.
RESULTS At 30-days post-revascularization, for ACS patients the odds ratio for MACCE favored CABG 0.49
(95% confidence interval [CI]: 0.34 to 0.71), whereas among SIHD patients MACCE was not affected by revascularization
strategy (odds ratio: 1.46; 95% CI: 0.71 to 3.01; pinteraction <0.01). With a median follow-up of 3.3 years, the late (31-day
to 5-year) benefit of CABG over PCI no longer varied by acuity of presentation, with a hazard ratio for MACCE in ACS patients
of 0.67 (95% CI: 0.55 to 0.81) and the hazard ratio for SIHD patients of 0.55 (95% CI: 0.40 to 0.74; pinteraction ¼ 0.28).
CONCLUSIONS In diabetic patients with MV-CAD, CABG was associated with a lower rate of long-term MACCE
relative to PCI for both ACS and SIHD. A well-powered randomized trial of CABG versus PCI in the ACS population is
warranted because these patients have been largely excluded from prior trials. (J Am Coll Cardiol 2017;70:2995–3006)
© 2017 by the American College of Cardiology Foundation.
G lobally, the prevalence of diabetes mellitus

(DM) among adults is projected to reach
642 million by 2040 (1). Total deaths from
DM are projected to rise by more than 50% in the
complications important societal and public health
concerns. Patients with DM are prone to a diffuse
and rapidly progressive forms of atherosclerosis,
which increases their likelihood of having multivessel
next 10 years, thereby making DM and its coronary artery disease (MV-CAD) requiring
Listen to this manuscript’s

audio summary by
JACC Editor-in-Chief From the aUniversity of British Columbia, Vancouver, Canada; bBC Centre for Improved Cardiovascular Health, Vancouver,
Dr. Valentin Fuster. Canada; cLoyola University Medical Center and Stritch School of Medicine, Maywood, Illinois; dCardiac Services British Columbia,
Vancouver, Canada; eSt. Michael’s Hospital, Toronto, Canada; and the fPeter Munk Cardiac Centre and the Heart and Stroke
Richard Lewar Centre, University of Toronto, Toronto, Canada. Presented in part at the Annual Scientific Sessions of the American
Heart Association, Orlando, Florida, November 2015. The authors have reported that they have no relationships relevant to the
contents of this paper to disclose. Sripal Bangalore, MD, MHA, served as Guest Editor for this paper.
Manuscript received October 21, 2016; revised manuscript received September 29, 2017, accepted October 10, 2017.
2996 Ramanathan et al. JACC VOL. 70, NO. 24, 2017
Revascularization in Diabetes With Multivessel CAD and Survival DECEMBER 19, 2017:2995–3006
ABBREVIATIONS revascularization (2). Selecting the optimal demographic and clinical data for all coronary
AND ACRONYMS coronary revascularization strategy for pa- angiography and revascularization procedures in
tients with DM and MV-CAD is crucial to British Columbia. This registry was used to define the
ACS = acute coronary
syndrome(s)
reduce the high rate of thrombotic complica- study cohort. Hospitalization separation data for all
tions and improve quality of life. patients hospitalized in British Columbia, including
CABG = coronary artery
bypass grafting The choice between CABG and PCI in DM admission and discharge date, hospital identification
CAD = coronary artery disease remains an area of intense discussion and code, and International Classification of Disease-10th
debate. In 2012, the results of the randomized Revision diagnosis codes, were used to identify the
controlled FREEDOM (Future REvasculariza- following outcomes: myocardial infarction (MI) and
DES = drug-eluting stent
tion Evaluation in Patients With Diabetes stroke. The Vital Statistics Deaths registry provides all
DM = diabetes mellitus
Mellitus: Optimal Management of Multi-vessel death dates in British Columbia and was used to
HR = hazard ratio
Disease) trial demonstrated lower rates of identify all-cause mortality. There are 5 cardiac
major adverse cardiovascular events in patients catheterization sites in British Columbia, all with PCI
IV = instrumental variable
with stable ischemic coronary disease (SIHD) and CABG capabilities. All 5 sites are located in the
LVEF = left ventricular ejection who were assigned to coronary artery bypass southern and central regions of British Columbia and
fraction
grafting (CABG) compared with percutaneous are reliant on a “hub and spoke” referral process and
MACCE = major adverse
coronary intervention (PCI) using drug-eluting an extensive transport network. In British Columbia,
cardiac or cerebrovascular
event(s) stents (PCI-DES), over a median of 3.8 years of approximately 80% of all PCIs are carried out as ad
MACCE(r) = composite of follow-up (3). The results of FREEDOM also hoc procedures with high rates of DES use (80%). We
major adverse cardiac or showed a borderline reduction in all-cause obtained ethics approval from the University of
cerebrovascular event and
mortality (p ¼ 0.049) favoring CABG over PCI- British Columbia-Providence Health Care Research
repeat revascularization
DES; this effect was confirmed in a robust Ethics Board.
meta-analysis of randomized trials (4).
MV-CAD = multivessel COHORT DEFINITION. This is a population-based,
SEE PAGE 3007
coronary artery disease retrospective cohort study of all patients older than
OR = odds ratio Although carefully planned and conducted 20 years of age with DM and angiographically
PCI = percutaneous coronary randomized trials provide the greatest confirmed MV-CAD (stenosis of >70% in 2 or more
intervention major epicardial vessels, excluding the left main
intrinsic validity of the study question under
SIHD = stable ischemic heart coronary artery) who underwent either PCI or isolated
consideration, the generalizability of these
disease
observations in the real world are often CABG between October 1, 2007 and January 31, 2014
limited by the recruitment criteria. Therefore, well- in British Columbia. As in the FREEDOM trial, sub-
conducted population-based analyses that provide jects with severe heart failure (New York Heart As-
complementary extrinsic validation of randomized sociation functional class III or IV), prior CABG or PCI
trials remain important components of overall knowl- within 6 months, prior valve surgery, 2 or more
edge translation efforts and are often the sources of chronic total occlusions, ST-segment elevation MI
supplementary information for key stakeholders. within 72 h, and stroke within 6 months were
The main objective of our study was to assess the excluded (Figure 1).
generalizability of the FREEDOM trial in real-world
OUTCOME DEFINITION. The primary outcome was
practice among patients with DM and MV-CAD in
the first occurrence of a major adverse cardiac or ce-
residents of British Columbia, Canada. Additionally,
rebrovascular event (MACCE), defined as a composite
we evaluated the impact of mode of revascularization
of all-cause mortality, nonfatal MI (International
(CABG vs. PCI-DES) in diabetic patients with acute
Classification of Disease-Tenth Revision [ICD-10]
coronary syndromes (ACSs) and MV-CAD. Previous
codes I21, I22) and nonfatal stroke (ICD-10 codes I60
studies of CABG versus PCI in DM patients with MV-
to I64, H356, H341, H342, and H348) after revasculari-
CAD consisted mostly of patients with stable
zation. Secondary outcomes included the individual
ischemic heart disease (SIHD) (4). The present study
components of MACCE, repeat revascularization post-
allowed us to analyze the ACS subgroup that repre-
discharge (RR), and a composite of MACCE and repeat
sents a large proportion of people with DM who are
revascularization (MACCE[r]). Staged PCI was defined
undergoing revascularization, and where there is
as a nonemergency PCI that was planned and per-
guideline equipoise (5).
formed within 2 months of the previous PCI; these PCIs
METHODS were not included in identifying repeat re-
vascularizations post-discharge. The validity of ICD-10
DATA SOURCES. Cardiac Services British Columbia codes for determination of outcomes has been well
holds a province-wide registry that captures patients’ validated (6).
JACC VOL. 70, NO. 24, 2017 Ramanathan et al. 2997
DECEMBER 19, 2017:2995–3006 Revascularization in Diabetes With Multivessel CAD and Survival
STATISTICAL METHODS. Baseline variables were

F I G U R E 1 Cohort Derivation
summarized as frequency and percentage for categor-
ical variables and as mean and SD for continuous var-
51,203 revascularization cases
iables. The comparisons by revascularization strategy
(40,053 PCI; 11,150 isolated CABG)
(PCI vs. CABG) and clinical presentation (ACS vs. SIHD) in BC from Oct 1, 2007 to January 31, 2014
were tested using the chi-square test for categorical (44,766 patients)
variables and the Student’s t-test for continuous vari-

Exclude non-diabetics
ables. The proportional hazard assumption was not
met over the 5-year follow-up period; therefore the
14,080 procedures
analysis was divided into early (30-day) and late (12,006 patients)
(31-day to 5-year) follow-up post-revascularization,
with results reported separately for each time frame. Exclude single vessel and left main disease
For the first 30-days, event rates, odds ratios (ORs), and
95% confidence intervals (CIs) are reported for CABG 7,843 procedures
(6,830 patients)
versus PCI. For the late period, event rates are reported
as Kaplan-Meier estimates, and Cox proportional haz-
ard models were fitted to obtain hazard ratios (HRs) Exclusion criteria similar to FREEDOM trial
1. Severe CHF (NYHA class III/IV)
and 95% CIs for CABG versus PCI. Data were censored if 2. Prior CABG or PCI within 6-months
a patient did not experience an outcome, had another 3. Prior valve surgery
4. Two or more chronic total occlusions
revascularization procedure within 1 year, or reached 5. STEMI within 72 hours
the end of follow-up, which was March 31, 2014. When 6. Stroke within 6 months
the individual outcomes were studied, death was
additionally censored. To assess the impact of revas- 4,819 procedures
cularization strategy on patients presenting with ACS (4,661 patients)
compared with SIHD, the interaction term of acuity

subtype by revascularization strategy was included in
the model. To account for baseline differences, fully 2,888 1,931
adjusted logistic and Cox models of MACCE were also PCI CABG
constructed. Baseline characteristics with a p value
< 0.20 in the univariate models were considered as
potential confounders; factors with a p value < 0.05
1,966 922 1,051 880
remained in the final model, except for sex, which ACS Stable IHD ACS Stable IHD
remained in the model regardless of its significance.
Given the difference in baseline variables between PCI
n = 3,017 n = 1,802
and CABG groups, a propensity score (PS) method, in-
verse probability of treatment weight (IPTW) (7), was
performed as sensitivity analysis. The list of variables The derivation of the study cohort from the British Columbia (BC) provincial
revascularization database between October 1, 2007 and January 31, 2014 is shown.
that were used to create the PS were age, fiscal year of
The study cohort was designed to replicate the FREEDOM (Future REvascularization
procedure, hospital, presentation acuity (ACS vs. Evaluation in Patients with Diabetes Mellitus: Optimal Management of Multi-vessel
SIHD), DM type, left ventricular ejection fraction Disease) randomized clinical trial as closely as possible by comparing coronary artery
(LVEF), hypertension, history of congestive heart dis- bypass grafting (CABG) versus percutaneous coronary intervention (PCI) with mainly
ease, pulmonary disease, peripheral arterial disease, drug-eluting stents. The proportion of the revascularization procedures received is shown,
as well as whether the patient had stable ischemic heart disease (IHD) or a recent acute
renal insufficiency, liver or gastrointestinal disease,
coronary syndrome (ACS). CHF ¼ congestive heart failure; NYHA ¼ New York Heart
and extent of CAD. The use of stabilized weights pro- Association (functional class); STEMI ¼ ST-segment elevation myocardial infarction.
vides more precise estimation of treatment effect
without losing any observations from incomplete
matching. Standardized differences were calculated to
All analyses were performed using SAS software
assess the balance in baseline variables between
version 9.4 (SAS Institute, Cary, North Carolina).
groups before and after applying IPTW. In addition, a
multilevel hospital-by-procedure year variable was RESULTS
chosen as an instrumental variable (IV); IV analysis on
30-day MACCE was performed to obtain an OR esti- COHORT CHARACTERISTICS. From October 1, 2007
mate for CABG versus PCI while controlling through January 31, 2014, 51,203 coronary revascu-
for selection bias from unmeasured confounders. larization procedures were performed in British
T A B L E 1 Baseline Characteristics by Mode of Revascularization T A B L E 2 Baseline Characteristics by Acuity Subtypes

(PCI vs. CABG) (ACS vs. SIHD)
PCI CABG ACS SIHD

(n ¼ 2,888) (n ¼ 1,931) p Value (n ¼ 3,017) (n ¼ 1,802) p Value
Age, yrs 67.3 10.8 65.2 9.0 <0.01 Age, yrs 66.8 10.6 66.0 9.3 0.01
Female 810 (28.0) 441 (22.8) <0.01 Female 855 (28.3) 396 (22) <0.01
Hyperlipidemia 2,237 (77.5) 1,538 (79.6) 0.07 Hyperlipidemia 2,254 (74.7) 1,521 (84.4) <0.01
Hypertension 2,545 (88.1) 1,772 (91.8) <0.01 Hypertension 2,672 (88.6) 1,645 (91.3) <0.01
Peripheral arterial disease 409 (14.2) 235 (12.2) 0.05 Peripheral arterial disease 406 (13.5) 238 (13.2) 0.82
Pulmonary disease 476 (16.5) 231 (12.0) <0.01 Pulmonary disease 469 (15.5) 238 (13.2) 0.03
Renal insufficiency* 204 (7.1) 133 (6.9) 0.81 Renal insufficiency* 238 (7.9) 99 (5.5) <0.01
Liver gastrointestinal disease 421 (14.6) 216 (11.2) <0.01 Liver gastrointestinal disease 378 (12.5) 259 (14.4) 0.07
Malignant disease 282 (9.8) 153 (7.9) 0.03 Malignant disease 281 (9.3) 154 (8.6) 0.37
3-vessel disease 815 (28.2) 1,241 (64.3) <0.01 3-vessel disease 1,296 (43.0) 760 (42.2) 0.60
LAD <0.01 LAD 0.56
Proximal 683 (23.6) 868 (45.0) Proximal 960 (31.8) 591 (32.8)
Other LAD 1,589 (55.0) 971 (50.3) Other LAD 1,603 (53.1) 957 (53.1)
Type 2 DM 2,840 (98.3) 1,878 (97.3) 0.01 Type 2 DM 2,951 (97.8) 1,767 (98.1) 0.57
ACS (vs. SIHD) 1,966 (68.1) 1,051 (54.4) <0.01 CCS (III/IV) 2,873 (95.2) 398 (22.1) <0.01
CCS (III or IV) 2,075 (71.8) 1,196 (61.9) <0.01 Urgency <0.01
Urgency <0.01 Elective 61 (2.0) 1,597 (88.7)
Elective 864 (29.9) 794 (41.1) Urgent 2,800 (92.8) 202 (11.2)
Urgent 1,923 (66.6) 1,079 (55.9) Emergency 156 (5.2) 2 (0.1)
Emergency 100 (3.5) 58 (3.0) Ejection fraction <0.01
Ejection fraction <0.01 >50% 1,595 (54.4) 1,253 (71.1)
>50% 1,579 (56.9) 1,269 (66.1) 30%–50% 751 (25.6) 306 (17.4)
30%–50% 550 (19.8) 507 (26.4) <30% 184 (6.3) 34 (1.9)
<30% 132 (4.8) 86 (4.5) Not entered† 403 (13.7) 169 (9.6)
Not entered† 514 (18.5) 58 (3)
Values are mean SD or n (%). *Dialysis or creatinine level >200 mmol/l.
†Ejection fraction purposely not entered; combined with <30% group in adjusted
models.
models. Abbreviations as in Table 1.
ACS ¼ acute coronary syndrome; CABG ¼ coronary artery bypass grafting;
CCS ¼ Canadian Cardiovascular Society angina scale; DM ¼ diabetes mellitus;
LAD ¼ left anterior descending coronary artery; PCI ¼ percutaneous coronary
intervention; SIHD ¼ stable ischemic heart disease. and left ventricular dysfunction. The 2 groups did not
differ with respect to triple-vessel disease or left
anterior descending coronary artery involvement
Columbia: 40,053 PCIs (78%) and 11,150 isolated (Table 2), respectively. Among ACS patients, 65.2%
CABG procedures. Of this total, 4,819 coronary underwent PCI; 34.8% underwent CABG.
revascularization procedures (PCI or isolated CABG) EARLY (30-DAY) CLINICAL OUTCOMES. We exam-
were performed in 4,661 patients with DM and CAD ined the impact of revascularization modality on the
who met the inclusion or exclusion criteria outlined outcomes reported in the original FREEDOM trial,
in Figure 1. namely MACCE, as well as each of the components of
The baseline characteristics of the overall cohort MACCE. The outcomes are reported as counts and
are described by revascularization strategy in Table 1. crude event rates in Table 3. The interaction between
Patients who underwent CABG were significantly acuity (ACS, SIHD) and treatment (PCI, CABG) was
younger, less likely to be female, and less likely to significant (p < 0.01). Therefore the early event rates
present with ACS, and they presented with less pe- for MACCE and its components are presented by
ripheral arterial disease, pulmonary disease, and left acuity. The rates were lower in patients undergoing
ventricular dysfunction. Conversely, CABG-treated CABG compared with PCI for MACCE, MI, and death,
patients had more triple-vessel disease and more but these differences were statistically significant
proximal left anterior descending coronary artery only for MACCE and MI. Conversely, stroke rates were
involvement. significantly higher in the CABG group.
When baseline characteristics are described by When expressed as ORs, CABG-treated patients
acuity of presentation—ACS versus SIHD—patients had significantly lower odds of MACCE and nonfatal
presenting with ACS (63%) were slightly older, more MI compared with patients who underwent PCI.
likely to be female and had more renal insufficiency Although the odds of all-cause mortality were also
T A B L E 3 Early (30-Day) Crude Counts and Event Rates
Overall ACS SIHD
PCI CABG PCI CABG PCI CABG

(n ¼ 2,888) (n ¼ 1,931) p Value (n ¼ 1,966) (n ¼ 1,051) p Value (n ¼ 922) (n ¼ 880) p Value
MACCE 176 (6.1) 64 (3.3) <0.01 162 (8.2) 45 (4.3) <0.01 14 (1.5) 19 (2.2) 0.31
Death 47 (1.6) 20 (1.0) 0.10 45 (2.3) 13 (1.2) 0.05 2 (0.2) 7 (0.8) 0.10
MI 130 (4.5) 22 (1.1) <0.01 119 (6.1) 19 (1.8) <0.01 11 (1.2) 3 (0.3) 0.06
Stroke 16 (0.6) 27 (1.4) <0.01 15 (0.8) 17 (1.6) 0.04 1 (0.1) 10 (1.1) <0.01
Values are n (%).

MACCE ¼ major adverse cardiac or cerebrovascular event(s); MI ¼ myocardial infarction; other abbreviations as in Table 1.
lower in the CABG-treated patients, this did not reach nevertheless, for consistency with the short-term
statistical significance. After multivariable adjust- outcomes, we present the results stratified by acu-
ment for baseline differences, the odds of MACCE in ity. MACCE and all the component outcomes,
the CABG group remained significantly lower at 0.60 including repeat revascularization, were all lower in
(95% CI: 0.43 to 0.84). The IPTW method–based odds the CABG-treated patients. These differences were all
of MACCE were consistent with the results of the statistically significant, except for stroke. When
multivariable adjusted model (OR: 0.62; 95% CI: 0.46 expressed as HRs, CABG-treated patients had a
to, 0.84). The ORs for each of the component out- significantly lower risk of MACCE (adjusted HR: 0.63;
comes favored CABG, as shown in Figure 2, except for 95% CI: 0.53 to 0.74; IPTW adjusted HR: 0.69; 95% CI:
stroke. 0.59 to 0.81), MACCE(r) (HR: 0.40; 95% CI: 0.35 to
IMPACT OF PRESENTATION WITH ACS ON EARLY 0.46), and the component outcomes of all-cause
(30-DAY) CLINICAL OUTCOMES. The impact of
revascularization strategy on MACCE varied
significantly by acuity of presentation (ACS vs. F I G U R E 2 Plot of ORs (95% Confidence Intervals) for 30-Day (Early) Outcomes
SIHD; p interaction <0.01) and also by disease (CABG vs. PCI)
severity (3-vessel disease vs. 2-vessel disease;

Favors CABG Favors PCI
p interaction ¼ 0.05). Among ACS patients undergoing
CABG compared with PCI, the adjusted OR for MACCE MACCE Composite endpoint
(unadjusted)
strongly favored CABG (adjusted OR: 0.49; 95% CI:
0.34 to 0.71), whereas among patients with SIHD, the Composite endpoint
(adjusted)
odds of MACCE did not vary significantly on the basis
of revascularization strategy (OR: 1.46; 95% CI: 0.71 to
Death
3.01). The IPTW-based model corroborated the find-
ings from the multivariable adjusted analysis, with a
MI
significant p interaction of 0.02. Among ACS patients,
the OR for MACCE favored CABG (OR: 0.52; 95% CI:
0.37 to 0.73), whereas among SIHD patients, MACCE Stroke
was not affected by revascularization strategy (OR:

1.16; 95% CI: 0.63 to 2.11). 0.1 0.2 0.4 0.6 1 2 4 6
Similarly, among patients with triple-vessel dis- Odds Ratio
ease who were undergoing CABG compared with PCI,

the adjusted OR for MACCE favored CABG (adjusted The early impact of revascularization modality on the primary outcome (major adverse
OR: 0.47; 95% CI: 0.31 to 0.71), whereas the advantage cardiac or cerebrovascular event(s) [MACCE]), which was a composite of all-cause death,
nonfatal myocardial infarction (MI), and nonfatal stroke, and secondary outcomes (the in-
of CABG over PCI in patients with 2-vessel disease
dividual components of major adverse cardiac or cerebrovascular events) expressed as
was not significant (adjusted OR: 0.88; 95% CI: 0.54
odds ratios (ORs). The odds ratios for primary outcome of MACCE was adjusted for age,
to 1.44). sex, presentation (ACS vs. SIHD), urgency (emergent, urgent vs. elective) and LVEF (>50%,
LATE (31-DAY TO 5-YEAR) CLINICAL OUTCOMES. 30–50% vs. <30%). The X axis is on logarithmic scale. The odds ratios for each of the
component outcomes favored coronary artery bypass grafting (CABG), except for stroke.
The late Kaplan-Meier event rates for MACCE and its
The patients who underwent coronary artery bypass grafting had significantly lower odds
component outcomes are presented in Table 4. The of major adverse cardiac or cerebrovascular events and nonfatal myocardial infarction,
interaction between acuity (ACS, SIHD) and treatment compared with patients who underwent percutaneous coronary intervention (PCI).
(PCI, CABG) was not significant (p ¼ 0.39);
T A B L E 4 Late (31-Day to 5-Year) Kaplan-Meier Event Rates
Overall ACS SIHD

MACCE 29.8 16.7 <0.01 33.4 20.8 <0.01 22.8 12.0 <0.01
MACCE(r) 42.7 20.4 <0.01 43.6 24.2 <0.01 40.9 16.1 <0.01
Death 19.0 10.3 <0.01 22.3 12.4 <0.01 12.2 7.8 <0.01
MI 15.5 6.8 <0.01 17.6 9.9 <0.01 11.7 3.3 <0.01
Stroke 5.1 4.1 0.23 5.8 6.2 0.97 3.8 1.7 0.18
RR 24.4 7.8 <0.01 22.6 8.2 <0.01 27.7 7.4 <0.01
Values are %.
MACCE(r) ¼ composite of major adverse cardiac or cerebrovascular event and repeat revascularization; RR ¼ repeat revascularization; other abbreviations as in Tables 1 and 3.
mortality, MI, stroke, and repeat revascularization, The results suggest that the use of DESs had little
compared with PCI-treated patients (Figure 3). Post- impact on the results reported for CABG versus PCI,
30-days, the all-cause mortality rate was lower by both in the early follow-up period and in the late
52% (HR: 0.48; 95% CI: 0.39 to 0.59), MI by 60% (HR: follow-up period. Among ACS patients, CABG versus
0.40; 95% CI: 0.31 to 0.51), stroke by 20% (HR: 0.80; all PCI demonstrated an advantage of CABG (OR:
95% CI: 0.56 to 1.15), and repeat revascularization by 0.49; 95% CI: 0.34 to 0.71); this advantage was
72% (HR: 0.28; 95% CI: 0.23 to 0.35) in patients un- essentially unchanged when restricted to DESs only
dergoing CABG compared with PCI. The results from (OR: 0.51; 95% CI: 0.35 to 0.77). Among SIHD patients,
the IPTW-based model for the component outcomes there was no significant difference between CABG
were consistent with the results reported earlier and versus all PCI on MACCE (OR: 1.46; 95% CI: 0.71 to
were only minimally attenuated. The cumulative 3.01), and this remained unchanged when restricted
incidence curves are plotted in Figure 4, thereby to CABG versus DESs (OR: 1.46; 95% CI: 0.67 to 3.19).
demonstrating the higher risk for all outcomes in the Similarly, in the late follow-up, the advantage of
PCI-treated patients and continuing divergence of the CABG over all PCIs (HR: 0.63; 95% CI: 0.53 to 0.74)
curves over time. The median follow-up time was 3.3 was essentially unchanged when CABG was compared
years (interquartile range [IQR]: 1.8 to 4.9 years). with DESs (HR: 0.74; 95% CI: 0.62 to 0.89).
The late benefit of CABG over PCI no longer With respect to second-generation DES stent use,
varied by acuity of presentation. In patients pre- we examined the impact of year of procedure
senting with ACS or SIHD, the use of CABG was on MACCE by constructing an interaction term—
associated with a significantly lower risk of MACCE procedure year by procedure type—in which proced-
(HR: 0.67; 95% CI: 0.55 to 0.81; and HR: 0.55; 95% CI: ure type was defined as CABG or DESs or bare-metal
0.40 to 0.74, respectively; p interaction ¼ 0.28). Again, stents, and procedure year is a surrogate for type of
the IPTW-based model corroborated these results (HR DES (first- or second-generation stent). Although we
for MACCE in ACS patients: 0.74; 95% CI: 0.62 to 0.89; have aggregate information on type of DESs, we do not
and HR for MACCE in SIHD patients: 0.56; 95% CI: have these data at the patient level. Procedure year
0.41 to 0.77). Figure 5 illustrates the differences be- allows us to determine whether the low use of second-
tween CABG- and PCI-treated patients before and generation DESs before 2010, followed by a rapid up-
after applying PS-IPTW). Importantly, after applying take reaching 93% by 2014, attenuated the advantage
IPTW, all between-group standardized differences of CABG over PCI. There was no further attenuation of
were <10. the advantage of CABG over PCI on outcomes, in the
SENSITIVITY ANALYSES. The study time period en- early or late follow-up period, when procedure year,
compasses a marked change from bare-metal stents to as a proxy for second-generation DES use, was
DESs and among DESs from first-generation to included. For further details, see Online Table 1.
second-generation stents for PCI. We conducted In addition, to assess the extent to which the
further sensitivity analyses to determine whether the treatment estimate could be affected by selection bias
overall advantage of CABG is attenuated with an in- from unmeasured confounders, we performed an IV
crease in PCI with DES use from 28% to 95% and/or analysis. The by-year and by-hospital proportion of
the introduction of second-generation DESs for PCI in CABG procedures, on the basis of the hospital and
2010, with a subsequent rise in use to 93% (Online year of diagnostic catheterization, reflects the treat-
Figure 1). ment selection tendency (CABG vs. PCI) for a given
year at a given hospital and was used as our instru-

F I G U R E 3 Plot of HRs (95% Confidence Interval) for 31-Day to 5-Year (Late)
ment in IV analysis. The assessment of the IV in- Outcomes (CABG vs. PCI)
dicates minimal bias in treatment effect as a result of
potential unobserved confounders (r ¼ 0.04; Favors CABG Favors PCI
p ¼ 0.789), and the strength of our IV, on the basis of a MACCE Composite endpoint
partial F test statistic of 201.2 (p < 0.001) is substan- (unadjusted)
tially greater than the conventional threshold of 10. Composite endpoint

(adjusted)
For the entire cohort, the OR for CABG versus PCI on
MACCE(r) Composite endpoint
30-day MACCE was 0.79 (95% CI: 0.34 to 1.83). When
restricted to the years 2011 through 2014, when
Death
second-generation DES use was >80%, the OR was
0.52 (95% CI: 0.19 to 1.44). MI
DISCUSSION Stroke
In our large, observational, real-world study of CABG RR

versus PCI, among patients with DM and MV-CAD,
0.2 0.4 0.6 0.8 1 1.4 2
MACCE outcomes were lower in those patients who
Hazard Ratio
underwent CABG (Central Illustration). During long-
term follow-up, all individual components of
The late impact of revascularization modality on the primary outcome (major adverse
MACCE were lower in patients undergoing CABG,
cardiac or cerebrovascular event(s) [MACCE]), which was a composite of all-cause death,
notably a statistically significant 52% relative risk nonfatal myocardial infarction (MI), and nonfatal stroke, and secondary outcomes
reduction in all-cause mortality. (the individual components of major adverse cardiac or cerebrovascular events), repeat
The controversy about the most efficacious mode revascularization post-discharge (RR), and a composite of major adverse cardiac or
of revascularization in SIHD patients with DM and cerebrovascular events and repeat revascularization (MACCE[r]) expressed as hazard ratios
(HRs). The hazard ratios for the primary outcome of MACCE was adjusted for age, sex,
MV-CAD was clarified by the results of the FREEDOM
presentation (ACS vs. SIHD) and LVEF (>50%, 30–50%, vs. <30%). The X axis is on
trial (3) and in subsequent meta-analysis of random- logarithmic scale. The hazard ratios for the primary outcome and all aspects of the sec-
ized trials (4). In FREEDOM, among patients receiving ondary outcome significantly favored patients undergoing coronary artery bypass grafting
aggressive evidence-driven medical therapy, the (CABG) compared with patients undergoing percutaneous coronary intervention (PCI).
5-year primary composite outcome occurred in 18.7%

of the CABG group and in 26.6% of the PCI-DES group
(p ¼ 0.005) (3). Importantly, the benefit of CABG was and exclusion criteria as possible and using provincial
driven by highly significant absolute reduction in MI administrative data for clinical events. To maximize
(7.9%; p < 0.001), but with a higher risk of stroke in the reliability of our data and achieve the highest-
the CABG group (5.2% vs. 2.4%; p ¼ 0.03). Our real- quality results, we limited the observational cohort
world observational analysis complements the effec- to include only those patients undergoing CABG or
tiveness of CABG over PCI in the randomized clinical PCI after October 2007, to reflect contemporary
trial findings in SIHD patients. When compared with practices (surgical techniques, early invasive angiog-
FREEDOM, our cohort was older and had more raphy, use of DESs, and use of dual antiplatelet
comorbidities such as left ventricular impairment therapy) most accurately.
(LVEF < 40%) (4.7% vs. 2.5% in FREEDOM) but Overall, PCI (60%) was chosen twice as often as
considerably lower rates of 3-vessel disease (43% vs. CABG in our cohort, similar to what was observed in
83%). Both the observational and the randomized the BARI (Bypass Angioplasty Revascularization
cohorts were similar in incidence of hypertension and Investigation) registry (percutaneous transluminal
in percentage of female sex (3). Most importantly, coronary angioplasty [PTCA] vs. CABG in patients
FREEDOM excluded patients presenting with with MV-CAD) (9). In the BARI registry both the PTCA
ACS within 72 h, whereas our cohort included and CABG cohorts were well matched for clinical
all-comers. characteristics, though anatomically the CABG-
Registries to better inform clinical trial findings are treated patients had more 3-vessel diseases and
often conducted concurrently (8). However, a formal proximal left anterior descending coronary artery
FREEDOM registry was not prospectively funded. involvement. A similar and substantial number in
Within the confines of the Cardiac Services British both groups had a diagnosis of ACS. Importantly, to
Columbia registry we set out to overcome this short- be enrolled in the BARI registry, patients were eligible
fall by using as many of the FREEDOM trial inclusion for both revascularization strategies, which may have
F I G U R E 4 Cumulative Incidence Function Curves for Each Outcome by Revascularization Strategy
50 50
MACCE(r) (%)
p < 0.01 by log-rank test p < 0.01 by log-rank test
MACCE (%)
40 40
Post 30-day 5-year rate: 29.8 vs. 16.7% Post 30-day 5-year rate: 42.7 vs. 20.4%
30 30
20 20
10 10
0 0
1 12 24 36 48 60 1 12 24 36 48 60
N. at risk N. at risk
PCI 2712 2193 1735 1248 798 428 PCI 2658 2023 1564 1097 684 357
CABG 1867 1488 1174 906 634 365 CABG 1861 1458 1142 869 604 346
50 50
40 p = 0.03 by log-rank test 40 p < 0.01 by log-rank test
Death (%)
MI (%)
30 30
20 20
10 10
0 0
1 12 24 36 48 60 1 12 24 36 48 60
PCI 2841 2404 1943 1417 926 510 PCI 2722 2226 1770 1282 827 449
CABG 1911 1568 1249 972 691 395 CABG 1890 1521 1203 930 655 379
50 50
Stroke (%)
RR (%)
30 30
20 20
10 10
0 0
1 12 24 36 48 60 1 12 24 36 48 60
Time from Revascularization (Months) Time from Revascularization (Months)
PCI 2827 2368 1904 1381 895 487 PCI 2756 2138 1689 1197 754 406
CABG 1887 1532 1217 942 664 379 CABG 1902 1515 1194 920 648 369
PCI CABG
The plotted curves are representative of the long-term cumulative incidences for coronary artery bypass grafting (CABG) and percutaneous coronary intervention (PCI)
for the primary and secondary outcomes. The curves demonstrate a higher risk for all outcomes in the patients who underwent percutaneous coronary intervention and
continuing divergence of the curves over time. The median follow-up time was 3.3 years (interquartile range: 1.8 to 4.9 years). Abbreviations as in Figure 3.
not been the case in a real-world clinical database. patients in the PCI group, is entirely consistent with
The BARI registry PTCA-treated patients had better the current practice of generally higher-acuity pa-
outcomes when compared with the registry patients tients and those declined for CABG undergoing high-
on the basis of better patient selection. Similarly, with risk PCI. We attempted to overcome some of the
technological advances over the past 2 decades in PCI clinical differences through statistical adjustment of
(delivery systems, stents, and adjunctive pharmaco- important variables. To minimize differences further,
therapy), there may have been willingness on the part we excluded patients with cardiogenic shock or those
of the interventional cardiologists to undertake more requiring mechanical support. Furthermore, if LVEF
challenging anatomic cases, thereby contributing to assessment was not available, these patients were
the differences we observed in favor of CABG over categorized in the highest risk category in our
PCI. modeling. Conversely, it is also possible that given
There were substantial clinical and angiographic the high percentage of patients undergoing ad hoc
differences between the PCI and CABG groups in our PCI, those patients with the most favorable anatomy
cohort, with many of the unfavorable clinical char- and those with lower disease burden may have un-
acteristics (age, sex, comorbidities, and lower LVEF) dergone PCI without consideration of CABG.
being more common in the PCI group. These impor- Prior studies of CABG versus PCI in patients with
tant differences, combined with higher rates of ACS DM and MV-CAD in the DES era mainly enrolled SIHD
F I G U R E 5 Absolute Standardized Differences in Unweighted and Propensity Score–Weighted Data Sensitivity Analyses
3-Vessel Disease
LAD
ACS vs. SIHD
Urgency
Age
CCS (III or IV)
Ejection Fraction
Pulmonary Disease
Hypertension
Sex
Liver/Gastrointestinal Disease
Type II DM
Malignancy
Peripheral Arterial Disease
Hyperlipidemia
Renal Insufficiency
0 10 20 30 40 50 60 70 80
Absolute Standardized Difference
Unweighted Weighted
Given the difference in baseline variables between coronary artery bypass grafting and percutaneous coronary intervention groups, a
propensity score–based method, inverse probability of treatment weight, was performed as sensitivity analysis. The figure illustrates the
differences between patients undergoing percutaneous coronary intervention and those undergoing percutaneous coronary intervention
before (o) and after (x) applying inverse probability of treatment weight. Importantly, after applying inverse probability of treatment weight,
all between-group standardized differences were <10. ACS ¼ acute coronary syndrome; LAD ¼ left anterior descending coronary artery;
CCS ¼ Canadian Cardiovascular Society; DM ¼ diabetes mellitus; SIHD ¼ stable ischemic heart disease.
patients (4). In contrast, more than 63% of our cohort with variable durations of follow-up are available for
consisted of patients with ACS, thus reflecting an culprit lesion PCI versus multivessel PCI in ACS pa-
interesting practice for DM patients in British tients with non–ST-segment elevation MI (10). Gener-
Columbia: a higher rate of CABG as the initial revas- ally, CABG is preferentially selected over multivessel
cularization strategy compared with U.S. registries. PCI for patients with more extensive disease burden.
The determinants to undergo ad hoc PCI are likely Our cohort also demonstrated that CABG-treated pa-
multiple and complex and include patients’ and tients had a significantly higher burden of CAD
physicians’ preference, upstream use of dual anti- (p < 0.01), whereas PCI-treated patients were older,
platelet therapy, delayed availability of CABG, anat- more likely to be female, and had comorbidities and
omy or comorbidities not suited for CABG, and other lower LVEF (p < 0.01 for all). Currently, there is con-
specific local institutional factors. The ACS patients troversy about the optimal timing of CABG following
compared with the SIHD patients in our cohort were an ACS. The safe time interval between myocardial
slightly older, had more comorbidities, and had injury and CABG was beyond 90 days in the European
similar anatomic variables. System for Cardiac Operative Risk Evaluation (Euro-
In the past, the perceived increased risk with CABG SCORE) (11). Early surgery post-myocardial injury can
in ACS compared with non-ACS patients was a ratio- be associated with edema causing poor visualization of
nale for considering PCI over CABG. However, the the target vessel and arteriotomy site. Furthermore,
choice of revascularization is more complex and in- successful reperfusion of an occluded vessel can result
cludes the severity and extent of disease, the extent of in reperfusion injury (12). Conversely, early CABG
ischemia, procedural risks, durability of the results, may lead to improved left ventricular systolic
and completeness of revascularization; many of these function and decreased arrhythmias. With improved
elements are not assessed in this population-based surgical techniques and periprocedural operative
registry. At present, only post hoc retrospective data care, there is increasing acceptance to carry out CABG
C E NT R AL IL L U STR AT IO N Landscape of Patient With Multivessel Coronary Artery Disease and Diabetes:
Revascularization Strategies
Ramanathan, K. et al. J Am Coll Cardiol. 2017;70(24):2995–3006.
ACS ¼ acute coronary syndrome; CABG ¼ coronary artery bypass grafting; FREEDOM ¼ Future REvascularization Evaluation in Patients with Diabetes Mellitus: Optimal
Management of Multi-vessel Disease; ICD-10 ¼ International Classification of Disease-Tenth Revision; IHD ¼ ischemic heart disease; MACCE ¼ major adverse cardiac or
cerebrovascular event(s); PCI ¼ percutaneous coronary intervention; SIHD ¼ stable ischemic heart disease.
following an ACS. Davierwala et al. (13), in a single- reinforce the critical need for well-powered random-
center study of 758 patients, showed that patients ized trials, most notably in patients with DM.
with CABG performed within 24 h of a non–ST-segment
elevation MI had in-hospital mortality rates and long- STUDY LIMITATIONS. This study has the inherent
term outcomes similar to those having CABG after limitations of an observational registry, with only
3-days. Among the patients undergoing CABG in our selected baseline characteristics gathered, limited
cohort, the procedure was deemed necessary during data on background therapy, and uncertainty on
the index hospitalization by the treating team. The adherence to guideline-directed management of risk
median time from cardiac catheterization to CABG in factors. Strict protocol-driven follow-up is unavai-
the overall cohort was 14.9 days (IQR: 6.7 to 65.0 days) lable outside the confines of a clinical trial, and as
compared with the ACS group of 7.8 days (IQR: 4.8 to such it is possible that our analysis may truly under-
13.9 days), thus reflecting 2 distinct patterns of practice. estimate the number of events. However, because the
The early and late benefits seen with CABG in our nature of the events of interest in the FREEDOM trial
cohort of mainly ACS patients raise the possibility would have resulted in hospitalization, it is unlikely
that in the current era, further gains may be made by that we have missed significant events because each
moving beyond ad hoc PCI as the default procedure in patient has a unique personal health number that
diabetic patients with MV-CAD. Our results need to be allows us to link to the province-wide hospitalization
validated by other large, population-based registries, database. Similarly, linkage with the provincial vital
and a randomized controlled trial of CABG versus PCI statistics database ensures that we capture all deaths,
in the ACS population is needed to inform practice and linkage with the Cardiac Services British
guidelines. Columbia registry ensures that we capture all repeat
In a network meta-analysis of 100 trials and more revascularization procedures.
than 90,000 patients, in which revascularization was There are also some analytical limitations with
compared with medical therapy for SIHD, Windecker missing variables, notably details on kidney function.
et al. (14) found improved survival compared with However, the number of patients with severe renal
medical therapy with CABG as well as with newer- failure in the range of renal replacement therapy was
generation stents, but not older PCI technologies. similar in both groups, and it was small. Therefore,
Indeed, the findings were similar even when the applicability of our findings specifically in pa-
including patients with recent ACS. Similarly, Ban- tients with moderate to severe kidney function
galore et al. (15) reported indirect comparisons of cannot be substantiated. We also have limited
patients undergoing CABG with PCI specifically with anatomic and procedural data and therefore cannot
DM and demonstrated similar mortality rates with comment on the complexity of disease SYNTAX
either strategy. Such analyses continue to raise the (Synergy Between PCI With Taxus and Cardiac Sur-
important question whether advancing PCI technol- gery) scores and completeness of revascularization
ogy will yield a different result from what has been and on the influence of these variables on outcomes.
seen thus far in comparative revascularization trials; Finally, our patient population consisted primarily of
the FAME 3 (A Comparison of Fractional Flow patients undergoing ad hoc PCI compared with pa-
Reserve-Guided Percutaneous Coronary Intervention tients having CABG after a period of stabilization.
and Coronary Artery Bypass Graft Surgery in Patients Therefore, it is possible that sicker patients were be-
With Multivessel Coronary Artery Disease) trial seeks ing selected for PCI for their acuity. Moreover, there
to address this hypothesis in patients with 3-vessel may have been a survival bias because of the delay
disease, by using a newer-generation stent platform between cardiac catheterization and CABG, with the
in concert with fractional flow reserve guidance (16). frailer and older patients dying before their CABG
To date, however, such analyses are limited by vary- procedures (17). Furthermore, because we lack
ing extents of 3-vessel CAD, variable inclusion of ACS patient-level data on type of DESs, we could not
patients, and varying degrees of follow-up. Also provide a robust comparison between CABG and
necessary to consider is that the FREEDOM trial was second-generation DESs.
largely a trial of 3-vessel disease (83% of patients), Despite multivariable adjustment and PS-IPTW
whereas in our study 42% to 43% of the patients based modeling, and the robustness of the findings
3-vessel disease. The fact that our findings were when using both these methods, it is still possible
limited to those patients with 3-vessel disease sug- that important, unmeasured confounders were not
gests that the extent of CAD is an important accounted for. Observational studies can address only
discriminator of benefit from higher-order therapy. measured confounders. If an unmeasured confounder
Taken in total, these hypothesis-generating findings both has a strong effect on our outcomes of interest
and is differentially distributed by mode of revascu- present in patients with and without ACS. For pa-
larization, this could bias the results. To assess this tients with DM who present with MV-CAD, these data
possibility, we carried out an IV analysis, which provide extrinsic validation of the randomized trials
indicated minimal bias in treatment effects as a result for patients with SIHD and represent a call to action
of unobserved confounders. Although the IV analysis for a large definitive randomized trial of patients
results were not statistically significant, this is not presenting with ACS.
surprising because IV analyses tend to generate wider
95% CIs, even when the IV is exogenous and rela- ADDRESS FOR CORRESPONDENCE: Dr. Krishnan
tively strong. However, the consistency of the study Ramanathan, University of British Columbia, 1081
findings with the findings of the randomized control Burrard St – B475, Vancouver, BC V6Z 1Y6, Canada.
trials, the comprehensiveness of the clinical data E-mail: kramanathan@providencehealth.bc.ca.
available for multivariable adjustment and PS-IPTW–
based modeling, and the inclusion and follow-up of PERSPECTIVES
all patients undergoing revascularization during the
study time period strengthen the validity of the study COMPETENCY IN PATIENT CARE: In patients with
findings. DM and MV-CAD stabilized following an ACS, as for
those patients with SIHD, a “heart team” approach is
CONCLUSIONS
recommended to individualize care, but CABG is
generally the preferred method of revascularization.
In a large, contemporary, and validated database, we
provide robust evidence that CABG was associated
TRANSLATIONAL OUTLOOK: Patients with DM
with better outcomes compared with PCI for MV-CAD
and MV-CAD who survive an ACS have been under-
at a population level. Importantly, these benefits are
represented in clinical trials, and further prospective
driven by a marked 37% superiority of CABG (over
studies are needed to define optimum revasculariza-
PCI) in terms of MACCE outcomes and a 52% reduc-
tion strategies.
tion in all-cause mortality in the long term. The long-
term cardiovascular benefits of CABG appear to be
REFERENCES
1. Shaw JE, Sicree RA, Zimmet PZ. Global esti- in nonexperimental research. Circ Cardiovasc Qual 14. Windecker S, Stortecky S, Stefanini GG, et al.
mates of the prevalence of diabetes for 2010 and Outcomes 2013;6:604–11. Revascularisation versus medical treatment in
2030. Diabetes Res Clin Pract 2010;87:4–14. patients with stable coronary artery disease:
8. Hochman JS, Buller CE, Sleeper LA, et al.
network meta-analysis. BMJ 2014;348:3859.
2. Natali A, Vichi A, Landi A, Serveri A, L’Abbate L, Cardiogenic shock complicating acute myocardial
Ferrannini E. Coronary atherosclerosis in type II infarction—etiologies, management and outcome: 15. Bangalore S, Toklu B, Feit F. Outcomes with
diabetes: angiographic findings and clinical out- a report from the SHOCK Trial Registry. J Am Coll coronary artery bypass graft surgery versus
comes. Diabetologia 2000;43:632–41. Cardiol 2000;36 Suppl 1:1063–70. percutaneous coronary intervention for patients
with diabetes mellitus: can newer generation
3. Farkouh ME, Domanski M, Sleeper LA, et al. 9. Feit F, Brooks MM, Sopko G, et al. Long-term
drug-eluting stents bridge the gap? Circ Car-
Strategies for multivessel revascularization in pa- clinical outcome in the Bypass Angioplasty
diovasc Interv 2014;7:518–25.
tients with diabetes. N Engl J Med 2012;367: Revascularization Investigation Registry compari-
2375–84. son with the randomized trial. Circulation 2000; 16. Fearon W. A Comparison of Fractional Flow
101:2795–802. Reserve-Guided Percutaneous Coronary Interven-
4. Verma S, Farkouh ME, Yanagawa B, et al.
tion and Coronary Artery Bypass Graft Surgery in
Comparison of coronary artery bypass surgery and 10. Weintraub WS, Grau-Sepulveda MV,
Patients With Multivessel Coronary Artery Disease
percutaneous coronary intervention in patients Weiss JM, et al. Comparative effectiveness of
(FAME 3). Stanford University. Trial identifier:
with diabetes: a meta-analysis of randomised revascularization strategies. N Engl J Med 2012;
NCT02100722. Available at: clinicaltrials.gov.
controlled trials. Lancet Diabetes Endocrinol 2013; 366:1467–76.
Accessed October 23, 2017.
1:317–28.
11. Nashef SA, Roques F, Gauducheau PM,
17. Antman EM, Cohen M, Bernink PM, et al. The
5. Amsterdam EA, Wenger NK, Brindis RG, et al. Lemeshow S, Salamon R. European system for
TIMI risk score for unstable angina/non–ST eleva-
2014 AHA/ACC guideline for the management of cardiac operative risk evaluation. Eur J Car-
tion MI: a method for prognostication and thera-
patients with non-ST-elevation acute coronary diothorac Surg 1999;16:9–13.
peutic decision making. JAMA 2000;284:835–42.
syndromes: a report of the American College of
12. Bell RM, Yellon DM. There is more to life than
Cardiology/American Heart Association Task Force
revascularization: therapeutic targeting of
on Practice Guidelines. J Am Coll Cardiol 2014;64:
myocardial ischemia/reperfusion injury. Cardiovasc KEY WORDS diabetes, multivessel
e139–228.
Ther 2011;29:e67–79. coronary artery disease, revascularization,
6. McCormick N, Lacaille D, Bhole V, Avina- survival
13. Davierwala PM, Verevkin A, Leontyev S,
Zubieta JA. Validity of myocardial infarction di-
Misfeld M, Borger MA, Mohr FW. Does timing of
agnoses in administrative databases: a systematic
coronary artery bypass surgery affect early and
review. PLoS One 2014;9:e92286. A PPE NDI X For a supplemental figure and
long-term outcomes in patients with non-ST-
7. Brookhart MA, Wyss R, Layton JB, Sturmer I. segment-elevation myocardial infarction? Circula- table, please see the online version of this
Propensity score methods for confounding control tion 2015;132:731–40. paper.
ª 2017 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN 0735-1097/$36.00
PUBLISHED BY ELSEVIER https://doi.org/10.1016/j.jacc.2017.10.029
Surgical Versus Percutaneous Coronary

Revascularization in Patients With
Diabetes and Acute Coronary Syndromes
Krishnan Ramanathan, MB, CHB,a James G. Abel, MD,a Julie E. Park, MMATH,b Anthony Fung, MBBS,a
Verghese Mathew, MD,c Carolyn M. Taylor, MD,a G.B. John Mancini, MD,a Min Gao, MD, PHD,b Lillian Ding, MSC,d
Subodh Verma, MD, PHD,e Karin H. Humphries, DSC,a,b Michael E. Farkouh, MD, MSCf
ABSTRACT
BACKGROUND Randomized trial data support the superiority of coronary artery bypass grafting (CABG) surgery over
percutaneous coronary intervention (PCI) in diabetic patients with multivessel coronary artery disease (MV-CAD).
However, whether this benefit is seen in a real-world population among subjects with stable ischemic heart disease
(SIHD) and acute coronary syndromes (ACS) is unknown.
OBJECTIVES The main objective of this study was to assess the generalizability of the FREEDOM (Future REvascularization
Evaluation in Patients with Diabetes Mellitus: Optimal Management of Multi-vessel Disease) trial in real-world practice among
patients with diabetes mellitus and MV-CAD in residents of British Columbia, Canada. Additionally, the study evaluated the
impact of mode of revascularization (CABG vs. PCI with drug-eluting stents) in diabetic patients with ACS and MV-CAD.
METHODS In a large population-based database from British Columbia, this study evaluated major cardiovascular
outcomes in all diabetic patients who underwent coronary revascularization between 2007 and 2014 (n ¼ 4,661, 2,947
patients with ACS). The primary endpoint (major adverse cardiac or cerebrovascular events [MACCE]) was a composite of
all-cause death, nonfatal myocardial infarction, and nonfatal stroke. The risk of MACCE with CABG or PCI was compared
using multivariable adjustment and a propensity score model.
RESULTS At 30-days post-revascularization, for ACS patients the odds ratio for MACCE favored CABG 0.49
(95% confidence interval [CI]: 0.34 to 0.71), whereas among SIHD patients MACCE was not affected by revascularization
strategy (odds ratio: 1.46; 95% CI: 0.71 to 3.01; pinteraction <0.01). With a median follow-up of 3.3 years, the late (31-day
to 5-year) benefit of CABG over PCI no longer varied by acuity of presentation, with a hazard ratio for MACCE in ACS patients
of 0.67 (95% CI: 0.55 to 0.81) and the hazard ratio for SIHD patients of 0.55 (95% CI: 0.40 to 0.74; pinteraction ¼ 0.28).
CONCLUSIONS In diabetic patients with MV-CAD, CABG was associated with a lower rate of long-term MACCE
relative to PCI for both ACS and SIHD. A well-powered randomized trial of CABG versus PCI in the ACS population is
warranted because these patients have been largely excluded from prior trials. (J Am Coll Cardiol 2017;70:2995–3006)
© 2017 by the American College of Cardiology Foundation.
G lobally, the prevalence of diabetes mellitus

(DM) among adults is projected to reach
642 million by 2040 (1). Total deaths from
DM are projected to rise by more than 50% in the
complications important societal and public health
concerns. Patients with DM are prone to a diffuse
and rapidly progressive forms of atherosclerosis,
which increases their likelihood of having multivessel
next 10 years, thereby making DM and its coronary artery disease (MV-CAD) requiring
Listen to this manuscript’s

audio summary by
JACC Editor-in-Chief From the aUniversity of British Columbia, Vancouver, Canada; bBC Centre for Improved Cardiovascular Health, Vancouver,
Dr. Valentin Fuster. Canada; cLoyola University Medical Center and Stritch School of Medicine, Maywood, Illinois; dCardiac Services British Columbia,
Vancouver, Canada; eSt. Michael’s Hospital, Toronto, Canada; and the fPeter Munk Cardiac Centre and the Heart and Stroke
Richard Lewar Centre, University of Toronto, Toronto, Canada. Presented in part at the Annual Scientific Sessions of the American
Heart Association, Orlando, Florida, November 2015. The authors have reported that they have no relationships relevant to the
contents of this paper to disclose. Sripal Bangalore, MD, MHA, served as Guest Editor for this paper.
Manuscript received October 21, 2016; revised manuscript received September 29, 2017, accepted October 10, 2017.
ABBREVIATIONS revascularization (2). Selecting the optimal demographic and clinical data for all coronary
AND ACRONYMS coronary revascularization strategy for pa- angiography and revascularization procedures in
tients with DM and MV-CAD is crucial to British Columbia. This registry was used to define the
ACS = acute coronary
syndrome(s)
reduce the high rate of thrombotic complica- study cohort. Hospitalization separation data for all
tions and improve quality of life. patients hospitalized in British Columbia, including
CABG = coronary artery
bypass grafting The choice between CABG and PCI in DM admission and discharge date, hospital identification
CAD = coronary artery disease remains an area of intense discussion and code, and International Classification of Disease-10th
debate. In 2012, the results of the randomized Revision diagnosis codes, were used to identify the
controlled FREEDOM (Future REvasculariza- following outcomes: myocardial infarction (MI) and
DES = drug-eluting stent
tion Evaluation in Patients With Diabetes stroke. The Vital Statistics Deaths registry provides all
DM = diabetes mellitus
Mellitus: Optimal Management of Multi-vessel death dates in British Columbia and was used to
HR = hazard ratio
Disease) trial demonstrated lower rates of identify all-cause mortality. There are 5 cardiac
major adverse cardiovascular events in patients catheterization sites in British Columbia, all with PCI
IV = instrumental variable
with stable ischemic coronary disease (SIHD) and CABG capabilities. All 5 sites are located in the
LVEF = left ventricular ejection who were assigned to coronary artery bypass southern and central regions of British Columbia and
fraction
grafting (CABG) compared with percutaneous are reliant on a “hub and spoke” referral process and
MACCE = major adverse
coronary intervention (PCI) using drug-eluting an extensive transport network. In British Columbia,
cardiac or cerebrovascular
event(s) stents (PCI-DES), over a median of 3.8 years of approximately 80% of all PCIs are carried out as ad
MACCE(r) = composite of follow-up (3). The results of FREEDOM also hoc procedures with high rates of DES use (80%). We
major adverse cardiac or showed a borderline reduction in all-cause obtained ethics approval from the University of
cerebrovascular event and
mortality (p ¼ 0.049) favoring CABG over PCI- British Columbia-Providence Health Care Research
repeat revascularization
DES; this effect was confirmed in a robust Ethics Board.
meta-analysis of randomized trials (4).
MV-CAD = multivessel COHORT DEFINITION. This is a population-based,
SEE PAGE 3007
coronary artery disease retrospective cohort study of all patients older than
OR = odds ratio Although carefully planned and conducted 20 years of age with DM and angiographically
PCI = percutaneous coronary randomized trials provide the greatest confirmed MV-CAD (stenosis of >70% in 2 or more
intervention major epicardial vessels, excluding the left main
intrinsic validity of the study question under
SIHD = stable ischemic heart coronary artery) who underwent either PCI or isolated
consideration, the generalizability of these
disease
observations in the real world are often CABG between October 1, 2007 and January 31, 2014
limited by the recruitment criteria. Therefore, well- in British Columbia. As in the FREEDOM trial, sub-
conducted population-based analyses that provide jects with severe heart failure (New York Heart As-
complementary extrinsic validation of randomized sociation functional class III or IV), prior CABG or PCI
trials remain important components of overall knowl- within 6 months, prior valve surgery, 2 or more
edge translation efforts and are often the sources of chronic total occlusions, ST-segment elevation MI
supplementary information for key stakeholders. within 72 h, and stroke within 6 months were
The main objective of our study was to assess the excluded (Figure 1).
generalizability of the FREEDOM trial in real-world
OUTCOME DEFINITION. The primary outcome was
practice among patients with DM and MV-CAD in
the first occurrence of a major adverse cardiac or ce-
residents of British Columbia, Canada. Additionally,
rebrovascular event (MACCE), defined as a composite
we evaluated the impact of mode of revascularization
of all-cause mortality, nonfatal MI (International
(CABG vs. PCI-DES) in diabetic patients with acute
Classification of Disease-Tenth Revision [ICD-10]
coronary syndromes (ACSs) and MV-CAD. Previous
codes I21, I22) and nonfatal stroke (ICD-10 codes I60
studies of CABG versus PCI in DM patients with MV-
to I64, H356, H341, H342, and H348) after revasculari-
CAD consisted mostly of patients with stable
zation. Secondary outcomes included the individual
ischemic heart disease (SIHD) (4). The present study
components of MACCE, repeat revascularization post-
allowed us to analyze the ACS subgroup that repre-
discharge (RR), and a composite of MACCE and repeat
sents a large proportion of people with DM who are
revascularization (MACCE[r]). Staged PCI was defined
undergoing revascularization, and where there is
as a nonemergency PCI that was planned and per-
guideline equipoise (5).
formed within 2 months of the previous PCI; these PCIs
METHODS were not included in identifying repeat re-
vascularizations post-discharge. The validity of ICD-10
DATA SOURCES. Cardiac Services British Columbia codes for determination of outcomes has been well
holds a province-wide registry that captures patients’ validated (6).
STATISTICAL METHODS. Baseline variables were

F I G U R E 1 Cohort Derivation
summarized as frequency and percentage for categor-
ical variables and as mean and SD for continuous var-
51,203 revascularization cases
iables. The comparisons by revascularization strategy
(40,053 PCI; 11,150 isolated CABG)
(PCI vs. CABG) and clinical presentation (ACS vs. SIHD) in BC from Oct 1, 2007 to January 31, 2014
were tested using the chi-square test for categorical (44,766 patients)
variables and the Student’s t-test for continuous vari-

Exclude non-diabetics
ables. The proportional hazard assumption was not
met over the 5-year follow-up period; therefore the
14,080 procedures
analysis was divided into early (30-day) and late (12,006 patients)
(31-day to 5-year) follow-up post-revascularization,
with results reported separately for each time frame. Exclude single vessel and left main disease
For the first 30-days, event rates, odds ratios (ORs), and
95% confidence intervals (CIs) are reported for CABG 7,843 procedures
(6,830 patients)
versus PCI. For the late period, event rates are reported
as Kaplan-Meier estimates, and Cox proportional haz-
ard models were fitted to obtain hazard ratios (HRs) Exclusion criteria similar to FREEDOM trial
1. Severe CHF (NYHA class III/IV)
and 95% CIs for CABG versus PCI. Data were censored if 2. Prior CABG or PCI within 6-months
a patient did not experience an outcome, had another 3. Prior valve surgery
4. Two or more chronic total occlusions
revascularization procedure within 1 year, or reached 5. STEMI within 72 hours
the end of follow-up, which was March 31, 2014. When 6. Stroke within 6 months
the individual outcomes were studied, death was
additionally censored. To assess the impact of revas- 4,819 procedures
cularization strategy on patients presenting with ACS (4,661 patients)
compared with SIHD, the interaction term of acuity

subtype by revascularization strategy was included in
the model. To account for baseline differences, fully 2,888 1,931
adjusted logistic and Cox models of MACCE were also PCI CABG
constructed. Baseline characteristics with a p value
< 0.20 in the univariate models were considered as
potential confounders; factors with a p value < 0.05
1,966 922 1,051 880
remained in the final model, except for sex, which ACS Stable IHD ACS Stable IHD
remained in the model regardless of its significance.
Given the difference in baseline variables between PCI
n = 3,017 n = 1,802
and CABG groups, a propensity score (PS) method, in-
verse probability of treatment weight (IPTW) (7), was
performed as sensitivity analysis. The list of variables The derivation of the study cohort from the British Columbia (BC) provincial
revascularization database between October 1, 2007 and January 31, 2014 is shown.
that were used to create the PS were age, fiscal year of
The study cohort was designed to replicate the FREEDOM (Future REvascularization
procedure, hospital, presentation acuity (ACS vs. Evaluation in Patients with Diabetes Mellitus: Optimal Management of Multi-vessel
SIHD), DM type, left ventricular ejection fraction Disease) randomized clinical trial as closely as possible by comparing coronary artery
(LVEF), hypertension, history of congestive heart dis- bypass grafting (CABG) versus percutaneous coronary intervention (PCI) with mainly
ease, pulmonary disease, peripheral arterial disease, drug-eluting stents. The proportion of the revascularization procedures received is shown,
as well as whether the patient had stable ischemic heart disease (IHD) or a recent acute
renal insufficiency, liver or gastrointestinal disease,
coronary syndrome (ACS). CHF ¼ congestive heart failure; NYHA ¼ New York Heart
and extent of CAD. The use of stabilized weights pro- Association (functional class); STEMI ¼ ST-segment elevation myocardial infarction.
vides more precise estimation of treatment effect
without losing any observations from incomplete
matching. Standardized differences were calculated to
All analyses were performed using SAS software
assess the balance in baseline variables between
version 9.4 (SAS Institute, Cary, North Carolina).
groups before and after applying IPTW. In addition, a
multilevel hospital-by-procedure year variable was RESULTS
chosen as an instrumental variable (IV); IV analysis on
30-day MACCE was performed to obtain an OR esti- COHORT CHARACTERISTICS. From October 1, 2007
mate for CABG versus PCI while controlling through January 31, 2014, 51,203 coronary revascu-
for selection bias from unmeasured confounders. larization procedures were performed in British
T A B L E 1 Baseline Characteristics by Mode of Revascularization T A B L E 2 Baseline Characteristics by Acuity Subtypes

(PCI vs. CABG) (ACS vs. SIHD)
PCI CABG ACS SIHD

(n ¼ 2,888) (n ¼ 1,931) p Value (n ¼ 3,017) (n ¼ 1,802) p Value
Age, yrs 67.3 10.8 65.2 9.0 <0.01 Age, yrs 66.8 10.6 66.0 9.3 0.01
Female 810 (28.0) 441 (22.8) <0.01 Female 855 (28.3) 396 (22) <0.01
Hyperlipidemia 2,237 (77.5) 1,538 (79.6) 0.07 Hyperlipidemia 2,254 (74.7) 1,521 (84.4) <0.01
Hypertension 2,545 (88.1) 1,772 (91.8) <0.01 Hypertension 2,672 (88.6) 1,645 (91.3) <0.01
Peripheral arterial disease 409 (14.2) 235 (12.2) 0.05 Peripheral arterial disease 406 (13.5) 238 (13.2) 0.82
Pulmonary disease 476 (16.5) 231 (12.0) <0.01 Pulmonary disease 469 (15.5) 238 (13.2) 0.03
Renal insufficiency* 204 (7.1) 133 (6.9) 0.81 Renal insufficiency* 238 (7.9) 99 (5.5) <0.01
Liver gastrointestinal disease 421 (14.6) 216 (11.2) <0.01 Liver gastrointestinal disease 378 (12.5) 259 (14.4) 0.07
Malignant disease 282 (9.8) 153 (7.9) 0.03 Malignant disease 281 (9.3) 154 (8.6) 0.37
3-vessel disease 815 (28.2) 1,241 (64.3) <0.01 3-vessel disease 1,296 (43.0) 760 (42.2) 0.60
LAD <0.01 LAD 0.56
Proximal 683 (23.6) 868 (45.0) Proximal 960 (31.8) 591 (32.8)
Other LAD 1,589 (55.0) 971 (50.3) Other LAD 1,603 (53.1) 957 (53.1)
Type 2 DM 2,840 (98.3) 1,878 (97.3) 0.01 Type 2 DM 2,951 (97.8) 1,767 (98.1) 0.57
ACS (vs. SIHD) 1,966 (68.1) 1,051 (54.4) <0.01 CCS (III/IV) 2,873 (95.2) 398 (22.1) <0.01
CCS (III or IV) 2,075 (71.8) 1,196 (61.9) <0.01 Urgency <0.01
Urgency <0.01 Elective 61 (2.0) 1,597 (88.7)
Elective 864 (29.9) 794 (41.1) Urgent 2,800 (92.8) 202 (11.2)
Urgent 1,923 (66.6) 1,079 (55.9) Emergency 156 (5.2) 2 (0.1)
Emergency 100 (3.5) 58 (3.0) Ejection fraction <0.01
Ejection fraction <0.01 >50% 1,595 (54.4) 1,253 (71.1)
>50% 1,579 (56.9) 1,269 (66.1) 30%–50% 751 (25.6) 306 (17.4)
30%–50% 550 (19.8) 507 (26.4) <30% 184 (6.3) 34 (1.9)
<30% 132 (4.8) 86 (4.5) Not entered† 403 (13.7) 169 (9.6)
Not entered† 514 (18.5) 58 (3)
models.
models. Abbreviations as in Table 1.
ACS ¼ acute coronary syndrome; CABG ¼ coronary artery bypass grafting;
CCS ¼ Canadian Cardiovascular Society angina scale; DM ¼ diabetes mellitus;
LAD ¼ left anterior descending coronary artery; PCI ¼ percutaneous coronary
intervention; SIHD ¼ stable ischemic heart disease. and left ventricular dysfunction. The 2 groups did not
differ with respect to triple-vessel disease or left
anterior descending coronary artery involvement
Columbia: 40,053 PCIs (78%) and 11,150 isolated (Table 2), respectively. Among ACS patients, 65.2%
CABG procedures. Of this total, 4,819 coronary underwent PCI; 34.8% underwent CABG.
revascularization procedures (PCI or isolated CABG) EARLY (30-DAY) CLINICAL OUTCOMES. We exam-
were performed in 4,661 patients with DM and CAD ined the impact of revascularization modality on the
who met the inclusion or exclusion criteria outlined outcomes reported in the original FREEDOM trial,
in Figure 1. namely MACCE, as well as each of the components of
The baseline characteristics of the overall cohort MACCE. The outcomes are reported as counts and
are described by revascularization strategy in Table 1. crude event rates in Table 3. The interaction between
Patients who underwent CABG were significantly acuity (ACS, SIHD) and treatment (PCI, CABG) was
younger, less likely to be female, and less likely to significant (p < 0.01). Therefore the early event rates
present with ACS, and they presented with less pe- for MACCE and its components are presented by
ripheral arterial disease, pulmonary disease, and left acuity. The rates were lower in patients undergoing
ventricular dysfunction. Conversely, CABG-treated CABG compared with PCI for MACCE, MI, and death,
patients had more triple-vessel disease and more but these differences were statistically significant
proximal left anterior descending coronary artery only for MACCE and MI. Conversely, stroke rates were
involvement. significantly higher in the CABG group.
When baseline characteristics are described by When expressed as ORs, CABG-treated patients
acuity of presentation—ACS versus SIHD—patients had significantly lower odds of MACCE and nonfatal
presenting with ACS (63%) were slightly older, more MI compared with patients who underwent PCI.
likely to be female and had more renal insufficiency Although the odds of all-cause mortality were also
T A B L E 3 Early (30-Day) Crude Counts and Event Rates
Overall ACS SIHD

MACCE 176 (6.1) 64 (3.3) <0.01 162 (8.2) 45 (4.3) <0.01 14 (1.5) 19 (2.2) 0.31
Death 47 (1.6) 20 (1.0) 0.10 45 (2.3) 13 (1.2) 0.05 2 (0.2) 7 (0.8) 0.10
MI 130 (4.5) 22 (1.1) <0.01 119 (6.1) 19 (1.8) <0.01 11 (1.2) 3 (0.3) 0.06
Stroke 16 (0.6) 27 (1.4) <0.01 15 (0.8) 17 (1.6) 0.04 1 (0.1) 10 (1.1) <0.01
Values are n (%).

MACCE ¼ major adverse cardiac or cerebrovascular event(s); MI ¼ myocardial infarction; other abbreviations as in Table 1.
lower in the CABG-treated patients, this did not reach nevertheless, for consistency with the short-term
statistical significance. After multivariable adjust- outcomes, we present the results stratified by acu-
ment for baseline differences, the odds of MACCE in ity. MACCE and all the component outcomes,
the CABG group remained significantly lower at 0.60 including repeat revascularization, were all lower in
(95% CI: 0.43 to 0.84). The IPTW method–based odds the CABG-treated patients. These differences were all
of MACCE were consistent with the results of the statistically significant, except for stroke. When
multivariable adjusted model (OR: 0.62; 95% CI: 0.46 expressed as HRs, CABG-treated patients had a
to, 0.84). The ORs for each of the component out- significantly lower risk of MACCE (adjusted HR: 0.63;
comes favored CABG, as shown in Figure 2, except for 95% CI: 0.53 to 0.74; IPTW adjusted HR: 0.69; 95% CI:
stroke. 0.59 to 0.81), MACCE(r) (HR: 0.40; 95% CI: 0.35 to
IMPACT OF PRESENTATION WITH ACS ON EARLY 0.46), and the component outcomes of all-cause
(30-DAY) CLINICAL OUTCOMES. The impact of
revascularization strategy on MACCE varied
significantly by acuity of presentation (ACS vs. F I G U R E 2 Plot of ORs (95% Confidence Intervals) for 30-Day (Early) Outcomes
SIHD; p interaction <0.01) and also by disease (CABG vs. PCI)
severity (3-vessel disease vs. 2-vessel disease;

Favors CABG Favors PCI
p interaction ¼ 0.05). Among ACS patients undergoing
CABG compared with PCI, the adjusted OR for MACCE MACCE Composite endpoint
(unadjusted)
strongly favored CABG (adjusted OR: 0.49; 95% CI:
0.34 to 0.71), whereas among patients with SIHD, the Composite endpoint
(adjusted)
odds of MACCE did not vary significantly on the basis
of revascularization strategy (OR: 1.46; 95% CI: 0.71 to
Death
3.01). The IPTW-based model corroborated the find-
ings from the multivariable adjusted analysis, with a
MI
significant p interaction of 0.02. Among ACS patients,
the OR for MACCE favored CABG (OR: 0.52; 95% CI:
0.37 to 0.73), whereas among SIHD patients, MACCE Stroke
was not affected by revascularization strategy (OR:

1.16; 95% CI: 0.63 to 2.11). 0.1 0.2 0.4 0.6 1 2 4 6
Similarly, among patients with triple-vessel dis- Odds Ratio
ease who were undergoing CABG compared with PCI,

the adjusted OR for MACCE favored CABG (adjusted The early impact of revascularization modality on the primary outcome (major adverse
OR: 0.47; 95% CI: 0.31 to 0.71), whereas the advantage cardiac or cerebrovascular event(s) [MACCE]), which was a composite of all-cause death,
nonfatal myocardial infarction (MI), and nonfatal stroke, and secondary outcomes (the in-
of CABG over PCI in patients with 2-vessel disease
dividual components of major adverse cardiac or cerebrovascular events) expressed as
was not significant (adjusted OR: 0.88; 95% CI: 0.54
odds ratios (ORs). The odds ratios for primary outcome of MACCE was adjusted for age,
to 1.44). sex, presentation (ACS vs. SIHD), urgency (emergent, urgent vs. elective) and LVEF (>50%,
LATE (31-DAY TO 5-YEAR) CLINICAL OUTCOMES. 30–50% vs. <30%). The X axis is on logarithmic scale. The odds ratios for each of the
component outcomes favored coronary artery bypass grafting (CABG), except for stroke.
The late Kaplan-Meier event rates for MACCE and its
The patients who underwent coronary artery bypass grafting had significantly lower odds
component outcomes are presented in Table 4. The of major adverse cardiac or cerebrovascular events and nonfatal myocardial infarction,
interaction between acuity (ACS, SIHD) and treatment compared with patients who underwent percutaneous coronary intervention (PCI).
(PCI, CABG) was not significant (p ¼ 0.39);
T A B L E 4 Late (31-Day to 5-Year) Kaplan-Meier Event Rates
Overall ACS SIHD

MACCE 29.8 16.7 <0.01 33.4 20.8 <0.01 22.8 12.0 <0.01
MACCE(r) 42.7 20.4 <0.01 43.6 24.2 <0.01 40.9 16.1 <0.01
Death 19.0 10.3 <0.01 22.3 12.4 <0.01 12.2 7.8 <0.01
MI 15.5 6.8 <0.01 17.6 9.9 <0.01 11.7 3.3 <0.01
Stroke 5.1 4.1 0.23 5.8 6.2 0.97 3.8 1.7 0.18
RR 24.4 7.8 <0.01 22.6 8.2 <0.01 27.7 7.4 <0.01
Values are %.
MACCE(r) ¼ composite of major adverse cardiac or cerebrovascular event and repeat revascularization; RR ¼ repeat revascularization; other abbreviations as in Tables 1 and 3.
mortality, MI, stroke, and repeat revascularization, The results suggest that the use of DESs had little
compared with PCI-treated patients (Figure 3). Post- impact on the results reported for CABG versus PCI,
30-days, the all-cause mortality rate was lower by both in the early follow-up period and in the late
52% (HR: 0.48; 95% CI: 0.39 to 0.59), MI by 60% (HR: follow-up period. Among ACS patients, CABG versus
0.40; 95% CI: 0.31 to 0.51), stroke by 20% (HR: 0.80; all PCI demonstrated an advantage of CABG (OR:
95% CI: 0.56 to 1.15), and repeat revascularization by 0.49; 95% CI: 0.34 to 0.71); this advantage was
72% (HR: 0.28; 95% CI: 0.23 to 0.35) in patients un- essentially unchanged when restricted to DESs only
dergoing CABG compared with PCI. The results from (OR: 0.51; 95% CI: 0.35 to 0.77). Among SIHD patients,
the IPTW-based model for the component outcomes there was no significant difference between CABG
were consistent with the results reported earlier and versus all PCI on MACCE (OR: 1.46; 95% CI: 0.71 to
were only minimally attenuated. The cumulative 3.01), and this remained unchanged when restricted
incidence curves are plotted in Figure 4, thereby to CABG versus DESs (OR: 1.46; 95% CI: 0.67 to 3.19).
demonstrating the higher risk for all outcomes in the Similarly, in the late follow-up, the advantage of
PCI-treated patients and continuing divergence of the CABG over all PCIs (HR: 0.63; 95% CI: 0.53 to 0.74)
curves over time. The median follow-up time was 3.3 was essentially unchanged when CABG was compared
years (interquartile range [IQR]: 1.8 to 4.9 years). with DESs (HR: 0.74; 95% CI: 0.62 to 0.89).
The late benefit of CABG over PCI no longer With respect to second-generation DES stent use,
varied by acuity of presentation. In patients pre- we examined the impact of year of procedure
senting with ACS or SIHD, the use of CABG was on MACCE by constructing an interaction term—
associated with a significantly lower risk of MACCE procedure year by procedure type—in which proced-
(HR: 0.67; 95% CI: 0.55 to 0.81; and HR: 0.55; 95% CI: ure type was defined as CABG or DESs or bare-metal
0.40 to 0.74, respectively; p interaction ¼ 0.28). Again, stents, and procedure year is a surrogate for type of
the IPTW-based model corroborated these results (HR DES (first- or second-generation stent). Although we
for MACCE in ACS patients: 0.74; 95% CI: 0.62 to 0.89; have aggregate information on type of DESs, we do not
and HR for MACCE in SIHD patients: 0.56; 95% CI: have these data at the patient level. Procedure year
0.41 to 0.77). Figure 5 illustrates the differences be- allows us to determine whether the low use of second-
tween CABG- and PCI-treated patients before and generation DESs before 2010, followed by a rapid up-
after applying PS-IPTW). Importantly, after applying take reaching 93% by 2014, attenuated the advantage
IPTW, all between-group standardized differences of CABG over PCI. There was no further attenuation of
were <10. the advantage of CABG over PCI on outcomes, in the
SENSITIVITY ANALYSES. The study time period en- early or late follow-up period, when procedure year,
compasses a marked change from bare-metal stents to as a proxy for second-generation DES use, was
DESs and among DESs from first-generation to included. For further details, see Online Table 1.
second-generation stents for PCI. We conducted In addition, to assess the extent to which the
further sensitivity analyses to determine whether the treatment estimate could be affected by selection bias
overall advantage of CABG is attenuated with an in- from unmeasured confounders, we performed an IV
crease in PCI with DES use from 28% to 95% and/or analysis. The by-year and by-hospital proportion of
the introduction of second-generation DESs for PCI in CABG procedures, on the basis of the hospital and
2010, with a subsequent rise in use to 93% (Online year of diagnostic catheterization, reflects the treat-
Figure 1). ment selection tendency (CABG vs. PCI) for a given
year at a given hospital and was used as our instru-

F I G U R E 3 Plot of HRs (95% Confidence Interval) for 31-Day to 5-Year (Late)
ment in IV analysis. The assessment of the IV in- Outcomes (CABG vs. PCI)
dicates minimal bias in treatment effect as a result of
potential unobserved confounders (r ¼ 0.04; Favors CABG Favors PCI
p ¼ 0.789), and the strength of our IV, on the basis of a MACCE Composite endpoint
partial F test statistic of 201.2 (p < 0.001) is substan- (unadjusted)
tially greater than the conventional threshold of 10. Composite endpoint

(adjusted)
For the entire cohort, the OR for CABG versus PCI on
MACCE(r) Composite endpoint
30-day MACCE was 0.79 (95% CI: 0.34 to 1.83). When
restricted to the years 2011 through 2014, when
Death
second-generation DES use was >80%, the OR was
0.52 (95% CI: 0.19 to 1.44). MI
DISCUSSION Stroke
In our large, observational, real-world study of CABG RR

versus PCI, among patients with DM and MV-CAD,
0.2 0.4 0.6 0.8 1 1.4 2
MACCE outcomes were lower in those patients who
Hazard Ratio
underwent CABG (Central Illustration). During long-
term follow-up, all individual components of
The late impact of revascularization modality on the primary outcome (major adverse
MACCE were lower in patients undergoing CABG,
cardiac or cerebrovascular event(s) [MACCE]), which was a composite of all-cause death,
notably a statistically significant 52% relative risk nonfatal myocardial infarction (MI), and nonfatal stroke, and secondary outcomes
reduction in all-cause mortality. (the individual components of major adverse cardiac or cerebrovascular events), repeat
The controversy about the most efficacious mode revascularization post-discharge (RR), and a composite of major adverse cardiac or
of revascularization in SIHD patients with DM and cerebrovascular events and repeat revascularization (MACCE[r]) expressed as hazard ratios
(HRs). The hazard ratios for the primary outcome of MACCE was adjusted for age, sex,
MV-CAD was clarified by the results of the FREEDOM
presentation (ACS vs. SIHD) and LVEF (>50%, 30–50%, vs. <30%). The X axis is on
trial (3) and in subsequent meta-analysis of random- logarithmic scale. The hazard ratios for the primary outcome and all aspects of the sec-
ized trials (4). In FREEDOM, among patients receiving ondary outcome significantly favored patients undergoing coronary artery bypass grafting
aggressive evidence-driven medical therapy, the (CABG) compared with patients undergoing percutaneous coronary intervention (PCI).
5-year primary composite outcome occurred in 18.7%

of the CABG group and in 26.6% of the PCI-DES group
(p ¼ 0.005) (3). Importantly, the benefit of CABG was and exclusion criteria as possible and using provincial
driven by highly significant absolute reduction in MI administrative data for clinical events. To maximize
(7.9%; p < 0.001), but with a higher risk of stroke in the reliability of our data and achieve the highest-
the CABG group (5.2% vs. 2.4%; p ¼ 0.03). Our real- quality results, we limited the observational cohort
world observational analysis complements the effec- to include only those patients undergoing CABG or
tiveness of CABG over PCI in the randomized clinical PCI after October 2007, to reflect contemporary
trial findings in SIHD patients. When compared with practices (surgical techniques, early invasive angiog-
FREEDOM, our cohort was older and had more raphy, use of DESs, and use of dual antiplatelet
comorbidities such as left ventricular impairment therapy) most accurately.
(LVEF < 40%) (4.7% vs. 2.5% in FREEDOM) but Overall, PCI (60%) was chosen twice as often as
considerably lower rates of 3-vessel disease (43% vs. CABG in our cohort, similar to what was observed in
83%). Both the observational and the randomized the BARI (Bypass Angioplasty Revascularization
cohorts were similar in incidence of hypertension and Investigation) registry (percutaneous transluminal
in percentage of female sex (3). Most importantly, coronary angioplasty [PTCA] vs. CABG in patients
FREEDOM excluded patients presenting with with MV-CAD) (9). In the BARI registry both the PTCA
ACS within 72 h, whereas our cohort included and CABG cohorts were well matched for clinical
all-comers. characteristics, though anatomically the CABG-
Registries to better inform clinical trial findings are treated patients had more 3-vessel diseases and
often conducted concurrently (8). However, a formal proximal left anterior descending coronary artery
FREEDOM registry was not prospectively funded. involvement. A similar and substantial number in
Within the confines of the Cardiac Services British both groups had a diagnosis of ACS. Importantly, to
Columbia registry we set out to overcome this short- be enrolled in the BARI registry, patients were eligible
fall by using as many of the FREEDOM trial inclusion for both revascularization strategies, which may have
F I G U R E 4 Cumulative Incidence Function Curves for Each Outcome by Revascularization Strategy
50 50
MACCE(r) (%)
p < 0.01 by log-rank test p < 0.01 by log-rank test
MACCE (%)
40 40
30 30
20 20
10 10
0 0
1 12 24 36 48 60 1 12 24 36 48 60
PCI 2712 2193 1735 1248 798 428 PCI 2658 2023 1564 1097 684 357
CABG 1867 1488 1174 906 634 365 CABG 1861 1458 1142 869 604 346
50 50
Death (%)
MI (%)
30 30
20 20
10 10
0 0
1 12 24 36 48 60 1 12 24 36 48 60
PCI 2841 2404 1943 1417 926 510 PCI 2722 2226 1770 1282 827 449
CABG 1911 1568 1249 972 691 395 CABG 1890 1521 1203 930 655 379
50 50
Stroke (%)
RR (%)
30 30
20 20
10 10
0 0
1 12 24 36 48 60 1 12 24 36 48 60
Time from Revascularization (Months) Time from Revascularization (Months)
PCI 2827 2368 1904 1381 895 487 PCI 2756 2138 1689 1197 754 406
CABG 1887 1532 1217 942 664 379 CABG 1902 1515 1194 920 648 369
PCI CABG
The plotted curves are representative of the long-term cumulative incidences for coronary artery bypass grafting (CABG) and percutaneous coronary intervention (PCI)
for the primary and secondary outcomes. The curves demonstrate a higher risk for all outcomes in the patients who underwent percutaneous coronary intervention and
continuing divergence of the curves over time. The median follow-up time was 3.3 years (interquartile range: 1.8 to 4.9 years). Abbreviations as in Figure 3.
not been the case in a real-world clinical database. patients in the PCI group, is entirely consistent with
The BARI registry PTCA-treated patients had better the current practice of generally higher-acuity pa-
outcomes when compared with the registry patients tients and those declined for CABG undergoing high-
on the basis of better patient selection. Similarly, with risk PCI. We attempted to overcome some of the
technological advances over the past 2 decades in PCI clinical differences through statistical adjustment of
(delivery systems, stents, and adjunctive pharmaco- important variables. To minimize differences further,
therapy), there may have been willingness on the part we excluded patients with cardiogenic shock or those
of the interventional cardiologists to undertake more requiring mechanical support. Furthermore, if LVEF
challenging anatomic cases, thereby contributing to assessment was not available, these patients were
the differences we observed in favor of CABG over categorized in the highest risk category in our
PCI. modeling. Conversely, it is also possible that given
There were substantial clinical and angiographic the high percentage of patients undergoing ad hoc
differences between the PCI and CABG groups in our PCI, those patients with the most favorable anatomy
cohort, with many of the unfavorable clinical char- and those with lower disease burden may have un-
acteristics (age, sex, comorbidities, and lower LVEF) dergone PCI without consideration of CABG.
being more common in the PCI group. These impor- Prior studies of CABG versus PCI in patients with
tant differences, combined with higher rates of ACS DM and MV-CAD in the DES era mainly enrolled SIHD
F I G U R E 5 Absolute Standardized Differences in Unweighted and Propensity Score–Weighted Data Sensitivity Analyses
3-Vessel Disease
LAD
ACS vs. SIHD
Urgency
Age
CCS (III or IV)
Ejection Fraction
Pulmonary Disease
Hypertension
Sex
Liver/Gastrointestinal Disease
Type II DM
Malignancy
Peripheral Arterial Disease
Hyperlipidemia
Renal Insufficiency
0 10 20 30 40 50 60 70 80
Absolute Standardized Difference
Unweighted Weighted
Given the difference in baseline variables between coronary artery bypass grafting and percutaneous coronary intervention groups, a
propensity score–based method, inverse probability of treatment weight, was performed as sensitivity analysis. The figure illustrates the
differences between patients undergoing percutaneous coronary intervention and those undergoing percutaneous coronary intervention
before (o) and after (x) applying inverse probability of treatment weight. Importantly, after applying inverse probability of treatment weight,
all between-group standardized differences were <10. ACS ¼ acute coronary syndrome; LAD ¼ left anterior descending coronary artery;
CCS ¼ Canadian Cardiovascular Society; DM ¼ diabetes mellitus; SIHD ¼ stable ischemic heart disease.
patients (4). In contrast, more than 63% of our cohort with variable durations of follow-up are available for
consisted of patients with ACS, thus reflecting an culprit lesion PCI versus multivessel PCI in ACS pa-
interesting practice for DM patients in British tients with non–ST-segment elevation MI (10). Gener-
Columbia: a higher rate of CABG as the initial revas- ally, CABG is preferentially selected over multivessel
cularization strategy compared with U.S. registries. PCI for patients with more extensive disease burden.
The determinants to undergo ad hoc PCI are likely Our cohort also demonstrated that CABG-treated pa-
multiple and complex and include patients’ and tients had a significantly higher burden of CAD
physicians’ preference, upstream use of dual anti- (p < 0.01), whereas PCI-treated patients were older,
platelet therapy, delayed availability of CABG, anat- more likely to be female, and had comorbidities and
omy or comorbidities not suited for CABG, and other lower LVEF (p < 0.01 for all). Currently, there is con-
specific local institutional factors. The ACS patients troversy about the optimal timing of CABG following
compared with the SIHD patients in our cohort were an ACS. The safe time interval between myocardial
slightly older, had more comorbidities, and had injury and CABG was beyond 90 days in the European
similar anatomic variables. System for Cardiac Operative Risk Evaluation (Euro-
In the past, the perceived increased risk with CABG SCORE) (11). Early surgery post-myocardial injury can
in ACS compared with non-ACS patients was a ratio- be associated with edema causing poor visualization of
nale for considering PCI over CABG. However, the the target vessel and arteriotomy site. Furthermore,
choice of revascularization is more complex and in- successful reperfusion of an occluded vessel can result
cludes the severity and extent of disease, the extent of in reperfusion injury (12). Conversely, early CABG
ischemia, procedural risks, durability of the results, may lead to improved left ventricular systolic
and completeness of revascularization; many of these function and decreased arrhythmias. With improved
elements are not assessed in this population-based surgical techniques and periprocedural operative
registry. At present, only post hoc retrospective data care, there is increasing acceptance to carry out CABG
C E NT R AL IL L U STR AT IO N Landscape of Patient With Multivessel Coronary Artery Disease and Diabetes:
Revascularization Strategies
Ramanathan, K. et al. J Am Coll Cardiol. 2017;70(24):2995–3006.
ACS ¼ acute coronary syndrome; CABG ¼ coronary artery bypass grafting; FREEDOM ¼ Future REvascularization Evaluation in Patients with Diabetes Mellitus: Optimal
Management of Multi-vessel Disease; ICD-10 ¼ International Classification of Disease-Tenth Revision; IHD ¼ ischemic heart disease; MACCE ¼ major adverse cardiac or
cerebrovascular event(s); PCI ¼ percutaneous coronary intervention; SIHD ¼ stable ischemic heart disease.
following an ACS. Davierwala et al. (13), in a single- reinforce the critical need for well-powered random-
center study of 758 patients, showed that patients ized trials, most notably in patients with DM.
with CABG performed within 24 h of a non–ST-segment
elevation MI had in-hospital mortality rates and long- STUDY LIMITATIONS. This study has the inherent
term outcomes similar to those having CABG after limitations of an observational registry, with only
3-days. Among the patients undergoing CABG in our selected baseline characteristics gathered, limited
cohort, the procedure was deemed necessary during data on background therapy, and uncertainty on
the index hospitalization by the treating team. The adherence to guideline-directed management of risk
median time from cardiac catheterization to CABG in factors. Strict protocol-driven follow-up is unavai-
the overall cohort was 14.9 days (IQR: 6.7 to 65.0 days) lable outside the confines of a clinical trial, and as
compared with the ACS group of 7.8 days (IQR: 4.8 to such it is possible that our analysis may truly under-
13.9 days), thus reflecting 2 distinct patterns of practice. estimate the number of events. However, because the
The early and late benefits seen with CABG in our nature of the events of interest in the FREEDOM trial
cohort of mainly ACS patients raise the possibility would have resulted in hospitalization, it is unlikely
that in the current era, further gains may be made by that we have missed significant events because each
moving beyond ad hoc PCI as the default procedure in patient has a unique personal health number that
diabetic patients with MV-CAD. Our results need to be allows us to link to the province-wide hospitalization
validated by other large, population-based registries, database. Similarly, linkage with the provincial vital
and a randomized controlled trial of CABG versus PCI statistics database ensures that we capture all deaths,
in the ACS population is needed to inform practice and linkage with the Cardiac Services British
guidelines. Columbia registry ensures that we capture all repeat
In a network meta-analysis of 100 trials and more revascularization procedures.
than 90,000 patients, in which revascularization was There are also some analytical limitations with
compared with medical therapy for SIHD, Windecker missing variables, notably details on kidney function.
et al. (14) found improved survival compared with However, the number of patients with severe renal
medical therapy with CABG as well as with newer- failure in the range of renal replacement therapy was
generation stents, but not older PCI technologies. similar in both groups, and it was small. Therefore,
Indeed, the findings were similar even when the applicability of our findings specifically in pa-
including patients with recent ACS. Similarly, Ban- tients with moderate to severe kidney function
galore et al. (15) reported indirect comparisons of cannot be substantiated. We also have limited
patients undergoing CABG with PCI specifically with anatomic and procedural data and therefore cannot
DM and demonstrated similar mortality rates with comment on the complexity of disease SYNTAX
either strategy. Such analyses continue to raise the (Synergy Between PCI With Taxus and Cardiac Sur-
important question whether advancing PCI technol- gery) scores and completeness of revascularization
ogy will yield a different result from what has been and on the influence of these variables on outcomes.
seen thus far in comparative revascularization trials; Finally, our patient population consisted primarily of
the FAME 3 (A Comparison of Fractional Flow patients undergoing ad hoc PCI compared with pa-
Reserve-Guided Percutaneous Coronary Intervention tients having CABG after a period of stabilization.
and Coronary Artery Bypass Graft Surgery in Patients Therefore, it is possible that sicker patients were be-
With Multivessel Coronary Artery Disease) trial seeks ing selected for PCI for their acuity. Moreover, there
to address this hypothesis in patients with 3-vessel may have been a survival bias because of the delay
disease, by using a newer-generation stent platform between cardiac catheterization and CABG, with the
in concert with fractional flow reserve guidance (16). frailer and older patients dying before their CABG
To date, however, such analyses are limited by vary- procedures (17). Furthermore, because we lack
ing extents of 3-vessel CAD, variable inclusion of ACS patient-level data on type of DESs, we could not
patients, and varying degrees of follow-up. Also provide a robust comparison between CABG and
necessary to consider is that the FREEDOM trial was second-generation DESs.
largely a trial of 3-vessel disease (83% of patients), Despite multivariable adjustment and PS-IPTW
whereas in our study 42% to 43% of the patients based modeling, and the robustness of the findings
3-vessel disease. The fact that our findings were when using both these methods, it is still possible
limited to those patients with 3-vessel disease sug- that important, unmeasured confounders were not
gests that the extent of CAD is an important accounted for. Observational studies can address only
discriminator of benefit from higher-order therapy. measured confounders. If an unmeasured confounder
Taken in total, these hypothesis-generating findings both has a strong effect on our outcomes of interest
and is differentially distributed by mode of revascu- present in patients with and without ACS. For pa-
larization, this could bias the results. To assess this tients with DM who present with MV-CAD, these data
possibility, we carried out an IV analysis, which provide extrinsic validation of the randomized trials
indicated minimal bias in treatment effects as a result for patients with SIHD and represent a call to action
of unobserved confounders. Although the IV analysis for a large definitive randomized trial of patients
results were not statistically significant, this is not presenting with ACS.
surprising because IV analyses tend to generate wider
95% CIs, even when the IV is exogenous and rela- ADDRESS FOR CORRESPONDENCE: Dr. Krishnan
tively strong. However, the consistency of the study Ramanathan, University of British Columbia, 1081
findings with the findings of the randomized control Burrard St – B475, Vancouver, BC V6Z 1Y6, Canada.
trials, the comprehensiveness of the clinical data E-mail: kramanathan@providencehealth.bc.ca.
available for multivariable adjustment and PS-IPTW–
based modeling, and the inclusion and follow-up of PERSPECTIVES
all patients undergoing revascularization during the
study time period strengthen the validity of the study COMPETENCY IN PATIENT CARE: In patients with
findings. DM and MV-CAD stabilized following an ACS, as for
those patients with SIHD, a “heart team” approach is
CONCLUSIONS
recommended to individualize care, but CABG is
generally the preferred method of revascularization.
In a large, contemporary, and validated database, we
provide robust evidence that CABG was associated
TRANSLATIONAL OUTLOOK: Patients with DM
with better outcomes compared with PCI for MV-CAD
and MV-CAD who survive an ACS have been under-
at a population level. Importantly, these benefits are
represented in clinical trials, and further prospective
driven by a marked 37% superiority of CABG (over
studies are needed to define optimum revasculariza-
PCI) in terms of MACCE outcomes and a 52% reduc-
tion strategies.
tion in all-cause mortality in the long term. The long-
term cardiovascular benefits of CABG appear to be
REFERENCES
1. Shaw JE, Sicree RA, Zimmet PZ. Global esti- in nonexperimental research. Circ Cardiovasc Qual 14. Windecker S, Stortecky S, Stefanini GG, et al.
mates of the prevalence of diabetes for 2010 and Outcomes 2013;6:604–11. Revascularisation versus medical treatment in
2030. Diabetes Res Clin Pract 2010;87:4–14. patients with stable coronary artery disease:
8. Hochman JS, Buller CE, Sleeper LA, et al.
network meta-analysis. BMJ 2014;348:3859.
2. Natali A, Vichi A, Landi A, Serveri A, L’Abbate L, Cardiogenic shock complicating acute myocardial
Ferrannini E. Coronary atherosclerosis in type II infarction—etiologies, management and outcome: 15. Bangalore S, Toklu B, Feit F. Outcomes with
diabetes: angiographic findings and clinical out- a report from the SHOCK Trial Registry. J Am Coll coronary artery bypass graft surgery versus
comes. Diabetologia 2000;43:632–41. Cardiol 2000;36 Suppl 1:1063–70. percutaneous coronary intervention for patients
with diabetes mellitus: can newer generation
3. Farkouh ME, Domanski M, Sleeper LA, et al. 9. Feit F, Brooks MM, Sopko G, et al. Long-term
drug-eluting stents bridge the gap? Circ Car-
Strategies for multivessel revascularization in pa- clinical outcome in the Bypass Angioplasty
diovasc Interv 2014;7:518–25.
tients with diabetes. N Engl J Med 2012;367: Revascularization Investigation Registry compari-
2375–84. son with the randomized trial. Circulation 2000; 16. Fearon W. A Comparison of Fractional Flow
101:2795–802. Reserve-Guided Percutaneous Coronary Interven-
4. Verma S, Farkouh ME, Yanagawa B, et al.
tion and Coronary Artery Bypass Graft Surgery in
Comparison of coronary artery bypass surgery and 10. Weintraub WS, Grau-Sepulveda MV,
Patients With Multivessel Coronary Artery Disease
percutaneous coronary intervention in patients Weiss JM, et al. Comparative effectiveness of
(FAME 3). Stanford University. Trial identifier:
with diabetes: a meta-analysis of randomised revascularization strategies. N Engl J Med 2012;
NCT02100722. Available at: clinicaltrials.gov.
controlled trials. Lancet Diabetes Endocrinol 2013; 366:1467–76.
Accessed October 23, 2017.
1:317–28.
11. Nashef SA, Roques F, Gauducheau PM,
17. Antman EM, Cohen M, Bernink PM, et al. The
5. Amsterdam EA, Wenger NK, Brindis RG, et al. Lemeshow S, Salamon R. European system for
TIMI risk score for unstable angina/non–ST eleva-
2014 AHA/ACC guideline for the management of cardiac operative risk evaluation. Eur J Car-
tion MI: a method for prognostication and thera-
patients with non-ST-elevation acute coronary diothorac Surg 1999;16:9–13.
peutic decision making. JAMA 2000;284:835–42.
syndromes: a report of the American College of
12. Bell RM, Yellon DM. There is more to life than
Cardiology/American Heart Association Task Force
revascularization: therapeutic targeting of
on Practice Guidelines. J Am Coll Cardiol 2014;64:
myocardial ischemia/reperfusion injury. Cardiovasc KEY WORDS diabetes, multivessel
e139–228.
Ther 2011;29:e67–79. coronary artery disease, revascularization,
6. McCormick N, Lacaille D, Bhole V, Avina- survival
13. Davierwala PM, Verevkin A, Leontyev S,
Zubieta JA. Validity of myocardial infarction di-
Misfeld M, Borger MA, Mohr FW. Does timing of
agnoses in administrative databases: a systematic
coronary artery bypass surgery affect early and
review. PLoS One 2014;9:e92286. A PPE NDI X For a supplemental figure and
long-term outcomes in patients with non-ST-
7. Brookhart MA, Wyss R, Layton JB, Sturmer I. segment-elevation myocardial infarction? Circula- table, please see the online version of this
Propensity score methods for confounding control tion 2015;132:731–40. paper.

Tugas Metaanalisis Kelompok 5

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Tugas Metaanalisis Kelompok 5

Diunggah oleh

Hak Cipta:

Format Tersedia

TUGAS PAPER

MATA KULIAH METODELOGI PENELITIAN DAN STATISTIKA

MATA KULIAH DASAR UMUM

Title Perbandingan angka kematian pasien penyakit jantung koroner

a. coronary artery disease

Tabel 1. Pencarian Literatur

PubMed “coronary artery disease” AND “coronary

Artikel Akhir yang

Identification of studies via databases and registers

Records identified from: Records removed before

Duplicate records removed (n

Records screened Records excluded

Reports sought for retrieval Reports not retrieved

Reports assessed for eligibility

Studies included in review

Authors Ramanathan K, Hideyuki Daniel J F M Gennaro Giustino, Park D, et al Ori Ben-Yehuda, et

PRISMA 2009 Checklist for systematic review

Section/topic # Checklist item Reported on page #

Mortality After Repeat Revascularization

ISSN 1936-8798/$36.00 https://doi.org/10.1016/j.jcin.2019.09.019

PCI CABG Hazard Ratio PCI CABG Hazard Ratio

Values are Kaplan-Meier estimated rate (number of events).

ischemia-driven revascularization at 3 years. Deﬁni- of each type of repeat revascularization event on

analyses were performed using SAS version 9.4

revascularization was stent thrombosis in 8 of 117

(A) All-cause mortality. (B) Cardiovascular mortality.

(A) All-cause mortality. (B) Cardiovascular mortality.

index revascularization strategy is illustrated in DISCUSSION

C E NT R AL IL L U STR AT IO N Repeat Revascularization and Mortality After Percutaneous Coronary Intervention or

7.6% underwent repeated Impact on Mortality of Repeat Revascularization

3-Year Cardiovascular Mortality

Any repeat revascularization p < 0.0001

Giustino, G. et al. J Am Coll Cardiol Intv. 2020;-(-):-–-.

further improve the beneﬁt/risk proﬁle of PCI

Percutaneous coronary intervention versus coronary artery

www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X 1

Prof P Jüni MD), and

2 www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X

consisted of patients with any left main disease, either

The sample size of the SYNTAXES study was based on

www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X 3

the SYNTAX trial from the extended follow-up in the

4 www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X

and angiographic characteristics were well matched 80

Probability of death (%)

0·97–1·41, p=0·092; figure 2A). Landmark analysis 60

in 106 (12%) patients after CABG (HR 1·15 [95% CI 40

www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X 5

A Three-vessel disease B Left main coronary artery disease

PCI group CABG group HR (95% CI) pinteraction

Type of coronary disease

0·5 0·8 1·0 1·25 2·0

Favours PCI Favours CABG

6 www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X

was complete for 94% of randomly assigned patients and

Probability of death (%)

www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X 7

8 www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X

www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X 9

10 www.thelancet.com Published online September 2, 2019 http://dx.doi.org/10.1016/S0140-6736(19)31997-X

ª 2021 THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION.

PUBLISHED BY ELSEVIER. ALL RIGHTS RESERVED.

Mortality 10 Years After Percutaneous or

ISSN 0735-1097/$36.00 https://doi.org/10.1016/j.jacc.2020.11.055

Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039 xxx xxx, 2020 1

2 xxx xxx, 2020 Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039

Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039 xxx xxx, 2020 3

Requiring insulin 10 (3.3) 9 (3.0)

Left main plus 1-vessel disease 50 (16.7) 53 (17.7)

Left main plus 3-vessel disease 122 (40.7) 123 (41.0)

4 xxx xxx, 2020 Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039

Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039 xxx xxx, 2020 5

6 xxx xxx, 2020 Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039

Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039 xxx xxx, 2020 7

8 xxx xxx, 2020 Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039

Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039 xxx xxx, 2020 9

10 xxx xxx, 2020 Circulation. 2020;141:00–00. DOI: 10.1161/CIRCULATIONAHA.120.046039