JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY

VOL. 68, NO. 10, 2016

2016 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION

PUBLISHED BY ELSEVIER

ISSN 0735-1097/$36.00
http://dx.doi.org/10.1016/j.jacc.2016.06.024

Outcomes After Percutaneous Coronary

Intervention or Bypass Surgery in Patients
With Unprotected Left Main Disease
Rafael Cavalcante, MD, PHD,a,b Yohei Sotomi, MD,c Cheol W. Lee, MD,d Jung-Min Ahn, MD,d Vasim Farooq, MD, PHD,e
Hiroki Tateishi, MD, PHD,a Erhan Tenekecioglu, MD,a Yaping Zeng, MD, PHD,a Pannipa Suwannasom, MD,c
Carlos Collet, MD,c Felipe N. Albuquerque, MD,a Yoshinobu Onuma, MD, PHD,a Seung-Jung Park, MD, PHD,d
Patrick W. Serruys, MD, PHDf

ABSTRACT
BACKGROUND Currently available randomized data on the comparison between percutaneous coronary intervention
(PCI) and coronary artery bypass graft (CABG) for the treatment of unprotected left main coronary disease (LMD) lacks
statistical power due to low numbers of patients enrolled.
OBJECTIVES This study assessed long-term outcomes of PCI and CABG for the treatment of LMD in specic subgroups
according to disease anatomic complexity.
METHODS We conducted a pooled analysis of individual patient-level data of the LMD patients included in the
PRECOMBAT (Bypass Surgery Versus Angioplasty Using Sirolimus-Eluting Stent in Patients With Left Main Coronary
Artery Disease) and SYNTAX (Synergy Between PCI With TAXUS and Cardiac Surgery) trials. Incidences of major adverse
cardiac events were assessed at 5 years follow-up.
RESULTS Study population comprised 1,305 patients. The incidence of major adverse cardiac and cerebrovascular
events at 5 years was 28.3% in the PCI group and 23.0% in the CABG group (hazard ratio [HR]: 1.23; 95% condence
interval [CI]: 1.01 to 1.55; p 0.045). This difference is mainly driven by a higher rate of repeat revascularization
associated with PCI (HR: 1.85; 95% CI: 1.38 to 2.47; p < 0.001). The 2 strategies showed similar rates of the safety
composite endpoint of death, myocardial infarction, or stroke (p 0.45). In patients with isolated LM or LM 1-vessel
disease, PCI was associated with a 60% reduction in all-cause mortality (HR: 0.40; 95% CI: 0.20 to 0.83; p 0.029) and
67% reduction in cardiac mortality (HR: 0.33; 95% CI: 0.12 to 0.88; p 0.025) when compared with CABG.
CONCLUSIONS In patients with unprotected LMD, CABG, and PCI result in similar rates of the safety composite
endpoint of death, myocardial infarction, or stroke. In patients with isolated LM or LM 1-vessel disease, PCI is
associated with lower all-cause and cardiac mortality when compared to CABG. (J Am Coll Cardiol 2016;68:9991009)
2016 by the American College of Cardiology Foundation.

evascularization for unprotected left main

in most catheterization laboratories across the globe

coronary artery disease (UPLMD) has evolved

following the improvement in PCI outcomes (1).

considerably in the last few years. What was

Currently available randomized data on treatment

once a forbidden territory for percutaneous coronary

of UPLMD lack statistical power due to low numbers

intervention (PCI) has now become common practice

of patients enrolled in randomized controlled trials,

Listen to this manuscripts

audio summary by

From the

JACC Editor-in-Chief

Netherlands; bHeart Institute (InCor), University of So Paulo Medical School, So Paulo, Brazil; cAcademic Medical Center,

Dr. Valentin Fuster.

University of Amsterdam, Amsterdam, the Netherlands; dHeart Institute, University of Ulsan College of Medicine, Asan Medical

Thoraxcenter, Department of Interventional Cardiology, Erasmus University Medical Center, Rotterdam, the

Center, Seoul, South Korea; eManchester Heart Centre, Manchester Royal Inrmary, Central Manchester University Hospitals NHS
Trust, Manchester, United Kingdom; and the fInternational Center for Circulatory Health, Imperial College London, London,
United Kingdom. Dr. Onuma serves on the advisory board of Abbott Vascular. All other authors have reported that they have no
relationships relevant to the contents of this paper to disclose. Drs. Cavalcante and Sotomi contributed equally to this work.
Manuscript received April 19, 2016; revised manuscript received May 25, 2016, accepted June 6, 2016.

1000

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

ABBREVIATIONS

especially for specic subgroup analyses. The

of including only population-level data instead of

AND ACRONYMS

PRECOMBAT (Bypass Surgery Versus Angio-

analyzing individual patient-level data (68). In the

plasty Using Sirolimus-Eluting Stent in Pa-

present study we performed a pooled analysis of in-

tients

Artery

dividual patient-level data from the 2 largest ran-

Disease) trial is the largest randomized

domized populations available so far. Our objective

controlled trial to ever address specically

was to compare long-term clinical outcomes of coro-

this population, and included only 600 pa-

nary artery bypass graft (CABG) surgery and PCI for the

tients (2). The SYNTAX (Synergy Between PCI

treatment of UPLMD in the 1,305 patients randomized

With TAXUS and Cardiac Surgery) trial

in the SYNTAX and PRECOMBAT trials and to assess

included a slightly larger number (n 705),

outcomes across several specic subgroups (9,10).

CABG = coronary artery bypass

graft

CAD = coronary artery disease

CI = condence interval
HR = hazard ratio
LM = left main
MACE = major adverse cardiac
event(s)

UPLMD = unprotected left

Main

Coronary

patients and was subject to the limitations

PCI = percutaneous coronary

main disease

Left

but was not specically designed for UPLMD

MI = myocardial infarction

intervention

With

METHODS

of subgroup analyses, albeit being a prespecied one (3,4). Both of these trials were

The methods and designs of both trials have been

underpowered to denitively answer the

previously described elsewhere (2,3). Some differ-

question of which is the best revascularization

ences between them are worth noting and are sum-

strategy for patients with UPLMD, if there is one (5).

marized as the following;

STUDY

SEE PAGE 1010

POPULATION. The

SYNTAX

trial

was

multicenter randomized controlled trial conducted in

Although several meta-analyses have tried to

17 countries in Europe and the United States that

address this statistical power issue, they have the issue

included 1,800 patients with 3-vessel or left main

F I G U R E 1 Study Flowchart

All Patients Randomized

SYNTAX trial N = 1800
PRECOMBAT trial N = 600

Patients with 3-vessel

disease in SYNTAX trial
N = 1095

Patients with left main disease

N = 1305

CABG
N = 648

PCI
N = 657
with TAXUS Express N = 357
with Cypher N = 300

Withdrew consent N = 12
Lost to follow-up N = 39

5-Year Follow-up
92.1% (597/648)

Withdrew consent N = 3
No longer met criteria N = 1
Lost to follow-up N = 28

5-Year Follow-up
95.1% (625/657)

Illustration depicting the patient selection and allocation to treatment randomization arms. Numbers of patients lost to follow-up and complete
follow-up rates are shown. CABG coronary artery bypass graft; PCI percutaneous coronary intervention; PRECOMBAT Bypass Surgery
Versus Angioplasty Using Sirolimus-Eluting Stent in Patients With Left Main Coronary Artery Disease; SYNTAX Synergy Between PCI With
TAXUS and Cardiac Surgery.

1001

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

(LM) coronary artery disease (CAD) of whom 705

comprised the LM cohort stratied at the time of

T A B L E 1 Baseline Characteristics of Patients With

Unprotected Left Main Disease Randomized to PCI or CABG

randomization. The PRECOMBAT trial was a ranPCI

(n 657)

CABG
(n 648)

p Value

63.8  10.0

64.3  9.9

0.35

Male

485 (73.8)

494 (76.2)

0.31

SYNTAX score

27.3  12.1

28.0  12.2

0.32

domized trial conducted at 13 sites in Korea that

included 600 patients with documented UPLMD.
PROCEDURES. In

both studies, patients deemed

Age, yrs

eligible for both PCI and CABG by an interventionalist

EuroSCORE

and a cardiac surgeon were prospectively enrolled

Body mass index, kg/m2

and randomized to treatment with either strategy. In

Diabetes mellitus

187 (28.5)

the SYNTAX trial, the PCI procedures were done with

Dyslipidemia

416 (63.3)

the use of paclitaxel-eluting TAXUS stents (Boston

Previous MI

Scientic, Natick, Massachusetts). The PRECOMBAT

trial used sirolimus-eluting CYPHER stents (Cordis,
Miami Lakes, Florida).
ENDPOINTS

AND

primary

endpoint of the SYNTAX trial was the 1-year occur-

Previous stroke
Peripheral vascular disease

rence of major adverse cardiac events (MACE), a

composite of all-cause death, myocardial infarction
(MI), stroke, or repeat revascularization. MI was
dened as any MI occurring after randomization.
More detailed denitions of MI were described previously (3). Stroke was dened as a focal central

COPD

0.47
0.26

179 (27.6)

0.74

380/645 (58.9) 0.10

0.72

40 (6.1)

41 (6.3)

0.86

14/355 (3.9)

14/344 (4.1)

0.93

52 (7.9)

47 (7.3)

0.65

81.8  31.6

81.0  27.7

0.66

35 (5.3)

41 (6.3)

0.44

59.3  13.9

59.5  11.1

0.80

153/642 (23.3)

165 (25.7)

0.31

LVEF, %
Current smoking

3.4  2.5
26.3  4.5

114/655 (17.4) 107/642 (16.7)

Previous PCI

Creatinine clearance, ml/min

DEFINITIONS. The

3.3  2.5
26.5  4.4

Values are mean  SD, n (%), or n/n (%). Denominators are provided for variables
with any missing data.
CABG coronary artery bypass graft surgery; COPD chronic obstructive
pulmonary disease; EuroSCORE European System for Cardiac Operative Risk
Evaluation; LVEF left ventricular ejection fraction; MI myocardial infarction;
PCI percutaneous coronary intervention; SYNTAX Synergy Between PCI With
TAXUS and Cardiac Surgery.

neurological decit lasting >72 h resulting in irreversible brain damage or body impairment. Repeat
revascularization was dened as any repeat PCI or
CABG.

T A B L E 2 Outcomes in PCI and CABG Groups Stratied According to SYNTAX Score

The primary endpoint of the PRECOMBAT trial was

PCI
(n 657)

the 1-year rates of the composite of all-cause death,

MI, stroke, and ischemia-driven target vessel revas-

CABG
(n 648)

HR (95% CI)

p Value

Overall cohort

cularization. MI was dened as new Q waves and in-

All-cause death

62 (9.4)

72 (11.1)

0.83 (0.591.16)

0.27

crease

Cardiac death

41 (6.2)

44 (6.8)

0.90 (0.591.37)

0.61

7 (1.1)

16 (2.5)

0.42 (0.171.02)

0.06

34 (5.2)

21 (3.2)

1.58 (0.922.72)

0.10

Repeat revascularization

128 (19.5)

70 (10.8)

1.85 (1.382.47)

cedure, or as new Q waves or an increase in creatine

<0.001

MACCE

186 (28.3)

149 (23.0)

1.25 (1.011.55)

kinase-myocardial band concentration to greater than

Death, MI, or stroke

92 (14.0)

98 (15.1)

0.90 (0.671.19)

in

the

creatine

kinase-myocardial

band

concentration to >5 times the upper limit of the

normal range, if occurring within 48 h after the pro-

the upper limit of the normal range, plus ischemic

Stroke
MI

SYNTAX score 032 (n 878)

(n 457)

0.045
0.45

(n 421)

symptoms or signs, if occurring more than 48 h after

All-cause death

31 (6.8)

42 (10)

0.66 (0.411.04)

0.07

the procedure. Stroke was dened as a sudden onset

Cardiac death

17 (3.7)

27 (6.4)

0.56 (0.311.02)

0.06

of vertigo, numbness, aphasia, or dysarthria resulting

Stroke

from vascular lesions of the brain, including hemorrhage, embolism, thrombosis, or rupturing aneurysm,
and persisting for >24 h. In both studies independent
clinical events committees blinded to group allocation adjudicated all events.
For the current pooled data analysis, we assessed
the incidences of MACE, a composite of all-cause
mortality, MI, stroke (as dened by each study protocol), and any repeat revascularization and each individual component, as well as cardiac death at 5 years

5 (1.1)

8 (1.9)

0.55 (0.181.68)

0.29

MI

18 (3.9)

10 (2.4)

1.63 (0.753.52)

0.22

Repeat revascularization

74 (16.2)

48 (11.4)

1.41 (0.982.03)

0.07

109 (23.9)

89 (21.1)

1.11 (0.841.47)

53 (11.6)

54 (12.8)

MACCE
Death, MI, or stroke
SYNTAX score $33 (n 427)

0.88 (0.61.28)

0.47
0.48

(n 200)

(n 227)

All-cause death

31 (15.5)

30 (13.2)

1.19 (0.721.97)

0.50

Cardiac death

24 (12)

17 (7.5)

1.63 (0.883.02)

0.13

2 (1)

8 (3.5)

0.29 (0.071.35)

0.11

MI

16 (8)

11 (4.8)

1.68 (0.783.61)

0.19
<0.001

Stroke
Repeat revascularization

54 (27)

22 (9.7)

3.09 (1.895.08)

MACCE

77 (38.5)

60 (26.4)

1.59 (1.142.23)

0.007

Death, MI, or stroke

39 (19.5)

44 (19.4)

1.01 (0.661.56)

0.96

follow-up (30 days). For the purpose of the current

analysis, repeat revascularization dened as any

Values are n (%) unless otherwise indicated. Results represent Kaplan-Meier estimates.

repeat PCI or CABG was used in the PRECOMBAT trial

CI condence interval; HR hazard ratio; MACCE major adverse cardiovascular and cerebrovascular
events; other abbreviations as in Table 1.

data, instead of ischemia-driven revascularization.

1002

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

F I G U R E 2 Subgroup Analysis

A
Female

1.24 [0.82-1.88]

Male

1.23 [0.95-1.58]

p value

HR [95%CI]

for interaction

Sex

p value

HR [95%CI]

for interaction

Sex

0.98

Female

1.15 [0.59-2.24]

Male

0.73 [0.49-1.09]

0.26

Age (yrs)

Age (yrs)
<65

1.39 [0.98-1.98]

65

1.18 [0.9-1.56]

0.47

<65

0.63 [0.31-1.29]

65

0.96 [0.65-1.41]

0.31

Renal Failure

Renal Failure
CrCl 60

1.33 [1.04-1.72]

CrCl < 60

0.98 [0.65-1.49]

0.22

CrCl 60

0.66 [0.43-1.02]

CrCl < 60

1.19 [0.68-2.07]

0.11

Ejection fraction

Ejection fraction
EF<50

1 [0.62-1.62]

EF>50

1.27 [1-1.63]

0.37

EF<50

0.73 [0.48-1.09]

EF>50

0.94 [0.49-1.83]

0.52

Diabetes

Diabetes
No DM

1.16 [0.89-1.51]

DM

1.42 [0.97-2.07]

0.38

No DM

0.75 [0.49-1.14]

DM

1 [0.56-1.77]

0.42

COPD

COPD
No COPD

1.24 [0.99-1.55]

COPD

1.29 [0.63-2.63]

0.89

No COPD

0.82 [0.57-1.17]

COPD

0.99 [0.33-2.93]

0.80

PVD

PVD
No PVD

1.35 [1.07-1.7]

PVD

0.61 [0.33-1.12]

0.02

No PVD

0.87 [0.6-1.26]

PVD

0.61 [0.27-1.36]

0.42

Trial*

Trial*
SYNTAX

1.2 [0.93-1.55]

PRECOMBAT

1.28 [0.86-1.89]

0.80

SYNTAX

0.86 [0.57-1.29]

PRECOMBAT

0.73 [0.39-1.37]

0.67

SYNTAX terciles

SYNTAX terciles
0-32

1.11 [0.84-1.47]

33

1.59 [1.14-2.23]

0.72

0-32

0.65 [0.41-1.04]

33

1.19 [0.72-1.96]

0.09

Disease Extent

Disease Extent
LM isolated/LM +1VD

0.96 [0.65-1.42]

LM + 2VD or 3VD

1.33 [1.02-1.72]

0.14

0.10

1.00
Favors PCI

10.00
Favors CABG

LM isolated/LM +1VD

0.4 [0.2-0.83]

LM + 2VD or 3VD

1 [0.67-1.5]

0.029

0.10

1.00
Favors PCI

10.00
Favors CABG

*Test for heterogeneity between the trials

I2 = 0% - p=0.62

*Test for heterogeneity between the trials

I2 = 0% - p=0.90

Hazard ratio (HR) and 95% condence interval (CI) for the occurrence of (A) major adverse cardiovascular and cerebrovascular events (a composite of
all-cause death, stroke, myocardial infarction or repeat revascularization), (B) all-cause death, and (C) cardiac death after PCI and CABG are shown for
several important subgroups of patients. COPD chronic obstructive pulmonary disease; CrCl creatinine clearance; DM diabetes mellitus; EF ejection
fraction; LM left main; PVD peripheral vascular disease; 1VD 1-vessel disease; 2VD 2-vessel disease; 3VD 3-vessel disease; other abbreviations as
in Figure 1.

Continued on the next page

SYNTAX SCORE II. The SYNTAX score II is a multi-

peripheral vascular disease. The score gives an indi-

variate model that was shown to predict long-term

vidual treatment recommendation on the basis of a

all-cause mortality in patients with multivessel or

95% predictive interval of the difference between the

LM CAD and is described in detail elsewhere (11,12). In

predicted 4-year mortality rate if the patient would

brief, it consists of a multivariable Cox proportional

undergo CABG and that of if the patient would un-

hazards model derived from the SYNTAX trial data

dergo PCI. If there is no difference or the difference

that includes independent predictors of 4-year mor-

cannot be separated with 95% condence, the

tality that showed an interaction between CABG and

recommendation is for either strategy or equipoise.

PCI. This score, externally validated in a large popu-

Any signicant difference (p < 0.05) favoring one

lation with UPLMD, takes into account the presence

strategy over the other will result in the recommen-

of UPLMD along with 7 other variables, namely

dation for the strategy associated with the smallest

anatomical SYNTAX score, age, creatinine clearance,

4-year predicted mortality rate (11).

left ventricular ejection fraction, sex, and presence

of

chronic

obstructive

pulmonary

disease

and

We calculated the SYNTAX score II values to

determine the calibration of the model in the

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

PRECOMBAT study and the pooled data. Furthermore, we compared the observed mortality rates according

to

the

SYNTAX

score

II

treatment

recommendations in the as-treated population.

F I G U R E 2 Continued

p value

HR [95%CI]

for interaction

Sex

STATISTICAL ANALYSIS. Continuous variables are

Female

1.38 [0.62-3.03]

presented as mean  SD and compared with Student

Male

0.73 [0.44-1.23]

t test. Binary variables are expressed as counts and

Age (yrs)

percentages

and

compared

using

chi-square

or

<65

0.67 [0.3-1.54]

65

1.06 [0.64-1.74]

Fisher exact tests as appropriate. The outcome anal-

Renal Failure

yses were performed according to the intention-to-

CrCl 60

0.73 [0.43-1.23]

CrCl < 60

1.32 [0.63-2.77]

treat principle and are presented as Kaplan-Meier

EF<50

1.04 [0.48-2.24]

ards models. The assumption of proportional hazards

EF>50

0.81 [0.49-1.37]

was veried for every model using time dependent

Diabetes
No DM

0.68 [0.39-1.18]

DM

1.36 [0.69-2.7]

population was stratied according to SYNTAX score

COPD

terciles and the outcomes were compared in low-

No COPD

0.93 [0.59-1.45]

intermediate (score 0 to 32) and high (score $33)

COPD

0.71 [0.17-2.97]

terciles. The consistency of the effect of treatment in

No PVD

0.97 [0.61-1.55]

PVD

0.57 [0.2-1.6]

testing for the endpoints of MACE, all-cause death,

Trial*

and cardiac death. Heterogeneity between the trials

SYNTAX

1.18 [0.69-2.01]

PRECOMBAT

0.54 [0.26-1.13]

selected

outcomes

was

assessed

with

the

Cochrans Q test and I2 statistic. All reported p values

0.20

0.62

0.12

0.69

PVD

subgroups was assessed with formal interaction

for

0.36

Ejection fraction

estimates and compared with Cox proportional haz-

Cox models using time as a continuous variable. The

0.19

0.35

0.10

SYNTAX terciles
0-32

0.56 [0.3-1.02]

were 2-sided, and values of p < 0.05 were considered

33

1.62 [0.87-3.02]

0.016

statistically signicant. Signicant p values are pre-

Disease Extent

sented with 3 decimal places and nonsignicant p

LM isolated/LM +1VD 0.33 [0.12-0.88]

1.15 [0.69-1.9]
LM + 2VD or 3VD

0.025

values are presented with 2 decimal places. All anal-

0.10

yses were done with SPSS version 21 (IBM, Armonk,

RESULTS

1.00

Favors PCI

New York).

10.00

Favors CABG

*Test for heterogeneity between the trials

I2 = 68% - p=0.08

PATIENTS. The total UPLMD population in the pre-

sent study comprised 1,305 patients. Five years

p 0.61), and MI (5.2% vs. 3.2%; HR: 1.58; 95% CI:

follow-up data were available in 597 (92.1%) of 648

0.92 to 2.72; p 0.10) as well as of the composite

patients randomized to CABG and in 625 (95.1%) of

safety endpoint of all-cause death, MI, or stroke

657 patients randomized to PCI (Figure 1). Overall,

(14.3% vs. 16.8%; HR: 0.90; 95% CI: 0.67 to 1.19;

mean age was 64.0  10.0 years, 75% of the patients

p 0.45) were observed (PCI vs. CABG respectively,

were men, and 28% were diabetic. Baseline charac-

for all). PCI was associated with a numerically lower

teristics were well balanced between the 2 study arms

rate of stroke that did not reach statistical signi-

(Table 1).

cance (1.1% vs. 2.5%; HR: 0.42; 95% CI: 0.17 to 1.02;

OUTCOMES AT 5 YEARS. Clinical outcomes were

p 0.06) (Table 2).

assessed in a time to rst event analysis. MACE at 5

SUBGROUP ANALYSIS AND SYNTAX SCORE TERCILES.

years occurred in 186 (28.3%) patients in the PCI

The results regarding the endpoints of MACE, all-

group and in 149 (23.0%) patients in the CABG

cause death, and cardiac death were consistent

group (hazard ratio [HR]: 1.25; 95% condence in-

among most subgroups (Figure 2). Of note, patients

terval [CI]: 1.01 to 1.55; p 0.045). This difference

with low-intermediate SYNTAX and patients with

was mainly driven by a higher rate of repeat

isolated LM or LM 1-vessel disease had lower

revascularization associated with PCI (19.5% vs.

overall and cardiac mortality in the PCI arm when

10.8%; HR: 1.85; 95% CI: 1.38 to 2.47; p < 0.001).

compared with CABG. Signicant p values for inter-

Similar rates of all-cause death (9.4% vs. 11.1%; HR:

action were observed for disease extent for all-cause

0.83; 95% CI: 0.59 to 1.16; p 0.27), cardiac death

mortality and for disease extent and SYNTAX score

(6.2% vs. 6.8%; HR: 0.90; 95% CI: 0.59 to 1.37;

terciles for cardiac mortality.

1003

1004

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

The outcomes according to anatomic SYNTAX

mainly driven by a higher repeat revascularization

score terciles are presented in Table 2. Figure 3 depicts

rate associated with PCI; 3) the rates of all-cause

Kaplan-Meier curves for MACE, all-cause death,

death and of the composite safety endpoint of

and cardiac death stratied according to SYNTAX

death, MI, or stroke are similar between both treat-

scores

high

ment strategies; 4) patients with low-intermediate

[$33] SYNTAX scores). The MACE rates were similar

SYNTAX scores showed a trend toward lower all-

with CABG and PCI in the low-intermediate (0 to 32)

cause and cardiac mortality when treated with PCI;

(low

intermediate

[0

to

32]

vs.

SYNTAX scores and signicantly higher with PCI in

5) patients with isolated LM or LM 1-vessel disease

patients with high ($33) SYNTAX scores. For all-cause

had signicantly lower all-cause and cardiac mortal-

death and cardiac death, there was a trend toward

ity rates when treated with PCI; and 6) the SYNTAX

lower rates with PCI in the low-intermediate SYNTAX

score II model can be used to help guide the treat-

score group and similar rates in patients in the high

ment decision between PCI and CABG in this patient

SYNTAX score group (Figure 3).

population when it recommends PCI or equipoise.

In patients with isolated LM or LM 1-vessel dis-

Over the past few years, the cardiology community

ease, PCI was associated with a 60% reduction in all-

saw a paradigm change in favor of PCI for the treat-

cause mortality (HR: 0.40; 95% CI: 0.20 to 0.83;

ment of patients with UPLMD (5). After the intro-

p 0.029) and 67% reduction in cardiac mortality

duction of drug-eluting stents and following the

(HR: 0.33; 95% CI: 0.12 to 0.88; p 0.025) when

publication of the initial results of the SYNTAX trial,

compared to CABG (Figure 2).

stenting of the LM stem started to gain rapid and

SYNTAX SCORE II CALIBRATION. The SYNTAX score

widespread popularity among PCI operators, with

II model demonstrated a good calibration in the

consent of surgical colleagues. Several registries

PRECOMBAT trial as well as in the pooled data and

showed comparable results between PCI and CABG,

both the CABG and PCI arms as depicted by values of

whereas the SYNTAX trial demonstrated an even

slopes

lower all-cause mortality with PCI in the subgroup of

and

intercepts

in

the

calibration

plots

(Figure 4). The model also showed fair discrimination

patients

ability for all-cause mortality with a c-index of 0.67 in

SYNTAX score 0 to 32) (1317).

with

lower

anatomic

complexity

(i.e.,

the PRECOMBAT trial, 0.70 in the overall pooled data,

Because this lack of difference in outcomes

0.73 in the pooled PCI arm, and 0.68 in the pooled

between PCI and CABG could be due to the relative

CABG arm (Figure 4). SYNTAX score II model

lack of power of the 2 randomized trials available

discrimination was better when compared to the

(SYNTAX and PRECOMBAT) or selection bias of

anatomic SYNTAX score that had c-indexes of 0.61

observational registries, pooling of randomized data

(95% CI: 0.56 to 0.66; p < 0.001), 0.55 (95% CI: 0.45 to

at the patient level proved of paramount importance.

0.65; p 0.29), 0.66 (95% CI: 0.58 to 0.73; p < 0.001),

Indeed, current European guidelines on myocardial

and 0.56 (95% CI: 0.49 to 0.63; p < 0.001) for the

revascularization give a class I recommendation, level

overall pooled, PRECOMBAT, pooled PCI, and pooled

of evidence B, for patients with low SYNTAX score

CABG populations, respectively.

(0 to 22), but a class IIa recommendation, level of

In patients with a SYNTAX score IIbased recom-

evidence B, for the subgroup of patients with inter-

mendation for PCI, the treatment with CABG led to a

mediate anatomic complexity (SYNTAX score 23 to 32)

signicantly higher long-term mortality rate than

(5). On the other hand, American guidelines give a

treatment with PCI (5.8% for PCI vs. 19.1% for CABG;

class IIa recommendation for a SYNTAX score #22 and

p 0.018) (as-treated population) (Table 3). In pa-

a class IIb recommendation for SYNTAX scores <33

tients in the equipoise group, the observed mortality

(18). These recommendations are based mainly on the

rates for PCI and CABG were similar (9.9% for PCI vs.

results of the SYNTAX trial before the 5 years results

9.6% for CABG; p 0.91) (Table 3). In the CABG

of the PRECOMBAT trial were available, and also on

recommendation group there was no statistically

the fact that these subgroup analyses are underpow-

signicant difference in all-cause death between the 2

ered for stronger conclusions.

treatment strategies.

The present study contributes to this discussion by

increasing the number of patients with long-term

DISCUSSION

follow-up in this subgroup of low and intermediate

SYNTAX score (n 878). The rates of MACE and its

The results of the present analysis can be summarized

individual components were all consistently similar

as follows: 1) in patients with unprotected LM CAD,

between PCI and CABG. Furthermore, when consid-

CABG is associated with a lower overall MACE rate

ering disease extent, patients with isolated LM or

than PCI at 5 years of follow-up; 2) this nding is

LM 1-vessel disease show lower overall and cardiac

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

F I G U R E 3 Incidence of Major Adverse Cardiovascular Events

Low-Intermediate (0-32) SYNTAX scores

High (33) SYNTAX scores

0.5

0.5

0.4

0.4

0.3
Log-rank p=0.47

23.9%

0.2

21.1%

0.1

Estimated MACCE Rate

Estimated MACCE Rate

38.5%
Log-rank p=0.007

0.3
26.4%
0.2

0.1

0.0

0.0
0
Number at risk
CABG 421
PCI 457

365

370
413

730
1095
1460
Days Since Randomization
350
381

340
371

320
344

1825

79
96

Number at risk
CABG 227
PCI 200

Low-Intermediate (0-32) SYNTAX scores

365

195
155

730
1095
1460
Days Since Randomization
185
146

175
134

167
127

1825

61
38

High (33) SYNTAX scores

0.5

0.4

0.3
Log-rank p=0.07
0.2
10.0%

0.1

6.8%

Estimated All-Cause Death Rate

Estimated All-Cause Death Rate

0.5

0
Number at risk
CABG 421
PCI 457

365

396
448

730
1095
1460
Days Since Randomization

Log-rank p=0.50
0.2

15.5%
13.2%

0.1

384
439

377
432

365
412

1825

98
123

Number at risk
CABG 227
PCI 200

365

214
185

Low-Intermediate (0-32) SYNTAX scores

0.5

730
1095
1460
Days Since Randomization
210
183

200
177

195
171

1825

73
62

High (33) SYNTAX scores

0.5

0.4

0.3
Log-rank p=0.06
0.2

0.1

6.4%

Estimated Cardiac Death Rate

Estimated Cardiac Death Rate

0.3

0.0

0.0

0.4

0.4

0.3
Log-rank p=0.13
0.2
12.0%
0.1
7.5%

3.7%

0.0
0
Number at risk
CABG 421
PCI 457

365

396
448

730
1095
1460
Days Since Randomization
384
439

377
432

365
412

0.0

1825

98
123

0
Number at risk
CABG 227
PCI 200

PCI

365

214
185

730
1095
1460
Days Since Randomization
210
183

200
177

195
171

1825

73
62

CABG

Kaplan-Meier curves depicting estimated event rates at 5 years of follow-up in patients with unprotected left main coronary disease. The
graphs are stratied according to SYNTAX score, with low-intermediate SYNTAX score (0 to 32) and high SYNTAX score ($33). (A, B) Major
adverse cardiac and cerebrovascular events (MACCE) (a composite of all-cause death, stroke, myocardial infarction, or repeat revascularization),
(C, D) all-cause death, and (E, F) cardiac death. Abbreviations as in Figure 1.

1005

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

F I G U R E 4 Calibration Plots for the SYNTAX Score II Model

PRECOMBAT (n=600)

25%

25%

20%

20%

Kaplan-Meier Estimates

Kaplan-Meier Estimates

Pooled data (n=1305)

15%

10%

5%

0%
0%

5%

10%

15%

10%

5%

0%
0%

25%

20%

15%

5%

10%

15%

20%

SYNTAX Score II Predicted Mortality

SYNTAX Score II Predicted Mortality

A 1.04 (95%CI 1.03-1.06; p<0.001)

B 0.06 (95%CI 0.05-0.08; p<0.001)
C 0.70 (95%CI 0.66-0.75; p<0.001)

A 1.02 (95%CI 1.01-1.04; p=0.013)

B 0.05 (95%CI 0.04-0.08; p<0.001)
C 0.67 (95%CI 0.58-0.75; p<0.001)

PCI arm (n=657)

25%

20%

20%

15%

10%

5%

0%
0%

5%

10%

15%

25%

CABG arm (n=648)

25%

Kaplan-Meier Estimates

Kaplan-Meier Estimates

1006

25%

20%

15%

10%

5%

0%
0%

5%

10%

15%

20%

SYNTAX Score II Predicted Mortality

SYNTAX Score II Predicted Mortality

A 1.04 (95%CI 1.03-1.05; p<0.001)

B 0.06 (95%CI 0.04-0.08; p<0.001)
C 0.73 (95%CI 0.67-0.80; p<0.001)

A 1.06 (95%CI 1.04-1.08; p<0.001)

B 0.06 (95%CI 0.04-0.09; p<0.001)
C 0.68 (95%CI 0.61-0.75; p<0.001)

25%

Calibration plots for the SYNTAX score II prediction model are shown for the PRECOMBAT trial population, the pooled database population, and
the PCI and CABG arms. Circles and triangles represent median in quintiles of predicted all-cause mortality. Observed all-cause mortality rates
are Kaplan-Meier estimates. For each plot, slope (A), intercept (B), and c-index (C) are reported, with condence interval (CI) and p value.

mortality with PCI, whereas signicant interaction for

T A B L E 3 Kaplan-Meier Estimates of All-Cause Mortality

treatment effect is observed with SYNTAX score ter-

Stratied According to SYNTAX Score II Treatment

ciles and disease extent. These ndings may question

Recommendations in the As-Treated Population

the class IIa guidelines recommendation for the in-

As Treated

termediate SYNTAX subgroup. With this longer-term

SYNTAX Score II
Recommendation

PCI
(n 679)

CABG
(n 614)

Equipoise (n 965)

50/503 (9.9)

44/459 (9.6)

0.91

tion, PCI may now be consider a safer alternative to

4/69 (5.8)

13/68 (19.1)

0.018

CABG in this specic subset of patients.

12/107 (11.2)

9/87 (10.3)

0.85

PCI (n 138)
CABG (n 196)
Values are n/N (%).
Abbreviations as in Table 1.

p Value

equivalent MACE rates in this much larger popula-

Our study is consistent with previously available

data in showing that CABG is associated with a
numerically higher stroke rate and a lower repeat
revascularization rate when compared to PCI in

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

1007

Multivessel Disease Involving Proximal LAD

C ENTR AL I LL U STRA T I O N Selection of the Best Revascularization Strategy for Unprotected LM Coronary Disease

Unprotected left main coronary disease (LMCD)

Patient with low-intermediate SYNTAX scores

Patient with isolated LM or LM + 1-vessel disease

Percutaneous coronary intervention (PCI)

Coronary artery bypass graft (CABG)

Higher rate of repeat revascularization

Higher overall MACCE rate

at 5 years follow-up

Lower rate of repeat revascularization

Similar overall safety

(death/myocardial infarction/stroke)

Lower overall MACCE rate

at 5 years follow-up

Cavalcante, R. et al. J Am Coll Cardiol. 2016;68(10):9991009.

LM left main; MACCE major adverse cardiac and cerebrovascular events; SYNTAX Synergy Between PCI With TAXUS and Cardiac Surgery.

patients with UPLMD. Specically in this population,

STUDY LIMITATIONS. First, it is an analysis of pooled

PCI shows similar results to CABG in patients with up

data from 2 different trials that included different

to intermediate anatomic complexity (SYNTAX score

patient populations in different time periods and this

0 to 32). Furthermore, this is the largest individual

could have introduced bias that might interfere with

patient-level database in the published data so far

our conclusions. In fact, CABG procedures showed

and this feature is what allows for strong conclusions

some signicant differences between the 2 studies. In

the SYNTAX trial 2.6  0.8 conduits were used, 1.3 

derived from it.

Last, the current analysis showed the ability of the

0.6 of which were arterial conduits. The left internal

SYNTAX score II model in predicting long-term

mammary artery was used in 87.6% of patients and

mortality in this patient population. In the present

off-pump surgery was performed in 16.1% of cases. In

study, patients with SYNTAX score II recommenda-

the PRECOMBAT trial 2.8  1.0 conduits were used

tion for PCI were penalized when treated with CABG

(p 0.02), 2.1  0.9 of which were arterial conduits

by a signicantly higher rate of death on long-term

(p < 0.001). The left internal mammary artery was

follow-up; and patients in the equipoise group

used in 84.7% of patients (p 0.07) and off-pump

showed

the

surgery was performed in 55.3% of cases (p < 0.001).

treatment recommendation was CABG, the score did

Despite these differences, in regard to MACE and all-

not show different mortality rates between PCI and

cause death, no heterogeneity was observed between

CABG treated patients. The reasons for the lower

trials (I 2 0%, p 0.90; I 2 0%, p 0.62, respec-

discrimination ability of the score in the CABG pa-

tively). For cardiac death some heterogeneity was

tients are still unclear. Furthermore, these ndings

observed (I 2 68%, p 0.08).

similar

long-term

mortality.

When

have to be interpreted with caution due to low

Furthermore,

denitions

regarding

outcomes,

numbers involved and the subgroup analysis nature.

especially MI and stroke, were not identical in the 2

Nevertheless, they provide some support to the

trials. Another limitation is that in both trials only

current guidelines recommendations for the use of

rst-generation stents were used in the PCI arms

such a tool during the Heart Team assessment to

and results could have been different with current

help guide the revascularization strategy selection

generation devices. Nevertheless, our data suggest

(5,11,12,19).

that even with less safe and efcacious stents, as

1008

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

has been shown by several studies, PCI is associ-

SYNTAX score II is a useful tool to help guide such

ated with a similar safety prole as CABG for this

selection.

subset of patients. The impact of newer generation

devices and PCI techniques on the treatment of

REPRINT REQUESTS AND CORRESPONDENCE: Dr.

UPLMD will be seen after the publication of the

Patrick W. Serruys, Department of Medicine, Erasmus

results

of

University, s-Gravendijkwal 230, 6th Floor, Rotter-

XIENCE Everolimus Eluting Stent Versus Coronary

dam 3000, the Netherlands. E-mail: patrick.w.j.c.

Artery Bypass Surgery for Effectiveness of Left

serruys@gmail.com.

of

the

ongoing

EXCEL

(Evaluation

Main Revascularization) (NCT01205776) and NOBLE

(Coronary Artery Bypass Grafting Vs Drug Eluting

PERSPECTIVES

Stent Percutaneous Coronary Angioplasty in the

Treatment

of

Unprotected

Left

Main

Stenosis)

COMPETENCY IN PATIENT CARE AND

(NCT01496651) randomized trials that will bring

PROCEDURAL SKILLS: For patients with unpro-

more and stronger evidence to this scenario (20,21).

tected LM CAD, revascularization can be achieved

CONCLUSIONS

with bypass surgery (CABG) or PCI with similar overall

rates of death, MI, and stroke. In those with isolated

In patients with UPLMD, CABG is associated with a

reduced

need

for

repeat

revascularization

and

similar rates of the composite safety endpoint of

death, MI, or stroke when compared to PCI. Specifically in the subgroup with low and intermediate
SYNTAX score (0 to 32), our results are reassuring in
regard to equivalent long-term results between the 2
treatment strategies and potentially better outcomes
in patients with isolated LM or LM 1-vessel disease with PCI (Central Illustration). The selection of
the revascularization should take into account not
only

the

coronary

anatomy,

but

also

clinical

comorbidities that impact long-term mortality, along

with

patient

and

physician

preferences.

LM disease or LM disease associated with stenosis of 1

other vessel, PCI is associated with lower rates of
all-cause and cardiac mortality compared with CABG.
Conversely, as reected by the SYNTAX scores, for
those with more diffuse or extensive atherosclerotic
disease, CABG may be preferred.
TRANSLATIONAL OUTLOOK: The results of future
comparative trials that incorporate advances in surgical and stent technologies should help guide clinicians in choosing an optimal revascularization strategy
for patients with specic anatomical patterns of coronary atherosclerotic disease.

The

REFERENCES
1. Park SJ, Ahn JM, Kim YH, et al. Temporal
trends in revascularization strategy and outcomes
in left main coronary artery stenosis: data from
the ASAN Medical Center-Left MAIN Revascular-

the Task Force on Myocardial Revascularization of

the European Society of Cardiology (ESC) and the
European Association for Cardio-Thoracic Surgery
(EACTS). Developed with the special contribution

9. Morice MC, Serruys PW, Kappetein AP, et al.

Five-year outcomes in patients with left main
disease treated with either percutaneous coronary
intervention or coronary artery bypass grafting in

ization registry. Circ Cardiovasc Interv 2015;8:

e001846.

of the European Association of Percutaneous

Cardiovascular Interventions (EAPCI). Eur Heart J
2014;35:2541619.

the synergy between percutaneous coronary

intervention with taxus and cardiac surgery trial.
Circulation 2014;129:238894.

6. Ferrante G, Presbitero P, Valgimigli M, et al.

Percutaneous coronary intervention versus bypass
surgery for left main coronary artery disease: a
meta-analysis of randomised trials. EuroIntervention 2011;7:73846, 1.

10. Ahn JM, Roh JH, Kim YH, et al. Randomized trial
of stents versus bypass surgery for left main coronary
artery disease: 5-year outcomes of the PRECOMBAT
study. J Am Coll Cardiol 2015;65:2198206.

2. Park SJ, Kim YH, Park DW, et al. Randomized trial

of stents versus bypass surgery for left main coronary artery disease. N Engl J Med 2011;364:171827.
3. Serruys PW, Morice MC, Kappetein AP, et al.
Percutaneous coronary intervention versus
coronary-artery bypass grafting for severe coronary artery disease. N Engl J Med 2009;360:
96172.
4. 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;121:264553.
5. Windecker S, Kolh P, Alfonso F, et al. 2014 ESC/
EACTS Guidelines on myocardial revascularization:

7. Jang JS, Choi KN, Jin HY, et al. Meta-analysis of

three randomized trials and nine observational
studies comparing drug-eluting stents versus
coronary artery bypass grafting for unprotected
left main coronary artery disease. Am J Cardiol
2012;110:14118.
8. Gargiulo G, Tamburino C, Capodanno D.
Five-year outcomes of percutaneous coronary
intervention versus coronary artery bypass graft
surgery in patients with left main coronary artery
disease: An updated meta-analysis of randomized
trials and adjusted observational studies. Int J
Cardiol 2015;195:7981.

11. Farooq V, van Klaveren D, Steyerberg EW, et al.

Anatomical and clinical characteristics to guide
decision making between coronary artery bypass
surgery and percutaneous coronary intervention
for individual patients: development and validation of SYNTAX score II. Lancet 2013;381:63950.
12. Campos CM, van Klaveren D, Iqbal J, et al.
Predictive performance of SYNTAX Score II in patients with left main and multivessel coronary artery disease-analysis of CREDO-Kyoto registry.
Circ J 2014;78:19429.
13. Park DW, Seung KB, Kim YH, et al. Long-term
safety and efcacy of stenting versus coronary
artery bypass grafting for unprotected left main

Cavalcante et al.

JACC VOL. 68, NO. 10, 2016

SEPTEMBER 6, 2016:9991009

Multivessel Disease Involving Proximal LAD

coronary artery disease: 5-year results from the

MAIN-COMPARE (Revascularization for Unprotected Left Main Coronary Artery Stenosis: Comparison of Percutaneous Coronary Angioplasty

(Premier of Randomized Comparison of Bypass

Surgery versus Angioplasty Using SirolimusEluting Stent in Patients with Left Main Coronary
Artery Disease) study. J Am Coll Cardiol Intv 2012;

Versus Surgical Revascularization) registry. J Am

Coll Cardiol 2010;56:11724.

5:70817.

14. Chieffo A, Meliga E, Latib A, et al. Drug-eluting

stent for left main coronary artery disease. The
DELTA registry: a multicenter registry evaluating
percutaneous coronary intervention versus coronary artery bypass grafting for left main treatment. J Am Coll Cardiol Intv 2012;5:71827.
15. Chieffo A, Magni V, Latib A, et al. 5-year outcomes following percutaneous coronary intervention with drug-eluting stent implantation versus
coronary artery bypass graft for unprotected left
main coronary artery lesions the Milan experience.
J Am Coll Cardiol Intv 2010;3:595601.
16. Kim YH, Park DW, Ahn JM, et al. Everolimuseluting stent implantation for unprotected left
main coronary artery stenosis. The PRECOMBAT-2

17. Shiomi H, Morimoto T, Furukawa Y, et al.

Comparison of percutaneous coronary intervention with coronary artery bypass grafting in
unprotected left main coronary artery disease - 5Year outcome from CREDO-Kyoto PCI/CABG registry cohort-2 -. Circ J 2015;79:12829.
18. Levine GN, Bates ER, Blankenship JC, et al.
2011 ACCF/AHA/SCAI Guideline for Percutaneous
Coronary Intervention: a report of the American
College of Cardiology Foundation/American Heart
Association Task Force on Practice Guidelines and
the Society for Cardiovascular Angiography and
Interventions. J Am Coll Cardiol 2011;58:e44122.
19. Campos CM, Stanetic BM, Farooq V, et al. Risk
stratication in 3-vessel coronary artery disease:
applying the SYNTAX Score II in the Heart Team

Discussion of the SYNTAX II trial. Catheter Cardiovasc Interv 2015;86:E22938.

20. Campos CM, Christiansen EH, Stone GW,
Serruys PW. The EXCEL and NOBLE trials: similarities, contrasts and future perspectives for left
main revascularisation. EuroIntervention 2015;11
Suppl V:V1159.
21. Campos CM, van Klaveren D, Farooq V, et al.
Long-term forecasting and comparison of
mortality in the Evaluation of the Xience Everolimus Eluting Stent vs. Coronary Artery Bypass
Surgery for Effectiveness of Left Main Revascularization (EXCEL) trial: prospective validation
of the SYNTAX score II. Eur Heart J 2015;36:
123141.

KEY WORDS coronary artery bypass graft

surgery, left main coronary artery disease,
long-term outcomes, percutaneous coronary
intervention, randomized controlled trials

1009