Research

Original Investigation

Coronary Artery Bypass Grafting vs Percutaneous Coronary

Intervention and Long-term Mortality and Morbidity
in Multivessel Disease
Meta-analysis of Randomized Clinical Trials
of the Arterial Grafting and Stenting Era
Ilke Sipahi, MD; M. Hakan Akay, MD; Sinan Dagdelen, MD; Arie Blitz, MD; Cem Alhan, MD

Editor's Note page 231

IMPORTANCE Recent trials of percutaneous coronary intervention (PCI) vs coronary artery Related article page 232
bypass grafting (CABG) for multivessel disease were not designed to detect a difference in
mortality and therefore were underpowered for this outcome. Consequently, the Supplemental content at
jamainternalmedicine.com
comparative effects of these 2 revascularization methods on long-term mortality are still
unclear. In the absence of solid evidence for mortality difference, PCI is oftentimes preferred
over CABG in these patients, given its less invasive nature.

OBJECTIVES To determine the comparative effects of CABG vs PCI on long-term mortality and
morbidity by performing a meta-analysis of all randomized clinical trials of the current era that
compared the 2 treatment techniques in patients with multivessel disease.

DATA SOURCES A systematic literature search was conducted for all randomized clinical trials
directly comparing CABG with PCI.

STUDY SELECTION To reflect current practice, we included randomized trials with 1 or more
arterial grafts used in at least 90%, and 1 or more stents used in at least 70% of the cases that
reported outcomes in patients with multivessel disease.

DATA EXTRACTION Numbers of events at the longest possible follow-up and sample sizes
were extracted.

DATA SYNTHESIS A total of 6 randomized trials enrolling a total of 6055 patients were
included, with a weighted average follow-up of 4.1 years. There was a significant reduction in
total mortality with CABG compared with PCI (I2 = 0%; risk ratio [RR], 0.73 [95% CI,
0.62-0.86]) (P < .001). There were also significant reductions in myocardial infarction
(I2 = 8.02%; RR, 0.58 [95% CI, 0.48-0.72]) (P < .001) and repeat revascularization
(I2 = 75.6%; RR, 0.29 [95% CI, 0.21-0.41]) (P < .001) with CABG. There was a trend toward
excess strokes with CABG (I2 = 24.9%; RR, 1.36 [95% CI, 0.99-1.86]), but this was not
statistically significant (P = .06). For reduction in total mortality, there was no heterogeneity
between trials that were limited to and not limited to patients with diabetes or whether Author Affiliations: Department of
Cardiology, Acibadem University
stents were drug eluting or not. Owing to lack of individual patient-level data, additional
Medical School, Istanbul, Turkey
subgroup analyses could not be performed. (Sipahi, Dagdelen); Harrington Heart
and Vascular Institute, University
CONCLUSIONS AND RELEVANCE In patients with multivessel coronary disease, compared with Hospitals Case Medical Center, Case
Western Reserve University School of
PCI, CABG leads to an unequivocal reduction in long-term mortality and myocardial Medicine, Cleveland, Ohio (Sipahi,
infarctions and to reductions in repeat revascularizations, regardless of whether patients are Blitz); Department of Cardiovascular
diabetic or not. These findings have implications for management of such patients. Surgery, Acibadem University Medical
School, Istanbul, Turkey (Akay,
Alhan).
Corresponding Author: Ilke Sipahi,
MD, FESC, Department of Cardiology,
Acibadem University Medical School,
Acibadem Maslak Hospital,
JAMA Intern Med. 2014;174(2):223-230. doi:10.1001/jamainternmed.2013.12844 Buyukdere Cad 40, 34457 Istanbul,
Published online December 2, 2013. Turkey ([email protected]).

223

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024

 Research Original Investigation CABG vs PCI in Multivessel Disease

D
espite advances in medical, surgical, and percutane- Study Selection
ous therapies, coronary artery disease (CAD) remains All of the 102 publications retrieved from the PubMed-
a leading cause of death in the Western world as well MEDLINE search were reviewed carefully for exclusion crite-
as many in developing countries. One of every 6 deaths in the ria. Studies were excluded if they (1) were not randomized, (2)
United States is caused by CAD. Approximately every 25 sec- did not have a dedicated CABG and PCI arm, (3) did not report
onds, an American will have a coronary event, and approxi- mortality, (4) did not report outcomes in patients with multi-
mately every minute, someone will die of one.1 vessel disease, (5) had an average follow-up duration shorter
The optimal treatment approach for patients with multi- than 1 year, (6) did not use at least 1 arterial graft in at least 90%
vessel coronary disease remains unclear despite a myriad of of the patients receiving CABG, and (7) did not use stents in at
randomized clinical trials performed in the last several least 70% of the patients in the PCI arm. The last 2 exclusion
decades. Several contemporary trials comparing coronary criteria were chosen to assure that the included clinical trials
artery bypass grafting (CABG) with percutaneous coronary reflected the current clinical practice.
intervention (PCI) have reported similar mortality rates with
the 2 treatment techniques. 2 -1 1 These trials have also Data Extraction
reported similar myocardial infarction (MI) rates with the 2 Data from studies meeting the selection criteria were ex-
treatment approaches, 4 , 6 - 1 1 although a longer-term tracted and verified independently by 2 of us (I.S. and M.H.A.).
follow-up of one of these trials suggested a reduction in MIs Information on inclusion criteria, duration of follow-up, pro-
with CABG.12 Given no clear superiority of surgical treat- cedural characteristics, and baseline patient characteristics
ment with regard to mortality and MIs, and given an were collected. Subsequently, number of events and total
increase in early strokes with CABG,9 PCI is often preferred sample size for the outcomes of interest according to treat-
in patients with multivessel CAD. Accordingly, between ment arms at the longest possible follow-up were extracted for
2001 and 2006, the number of PCIs performed annually each trial. If the actual numbers of events were not stated,
for multivessel disease inc reased by 56%, and the Kaplan-Meier estimates were used.
total number of CABG surgeries decreased by 24% and
continued to decline at a rate of approximately 5% per year Statistical Analysis
subsequently.13 Statistical heterogeneity was tested by the Cochran Q statis-
Despite the large number of clinical trials comparing tic and was reported as I2. To obtain meta-analytic risk ratios
CABG with PCI for multivessel disease, all of these trials (RRs) and 95% CIs, fixed effects models using number of events
were underpowered to detect a difference in all-cause mor- and total sample size were used, unless there was heteroge-
tality, the most important outcome of cardiovascular neity among the included trials. In cases of heterogeneity (de-
trials.2,6,7,9-11 Similarly, these trials were also underpowered fined as I2 > 40%), random effects models were used. Sensi-
to detect differences in MI, a major cause of morbidity in tivity analyses were performed according to whether trials were
these patients. Consequently, the current practice regarding limited or not limited to diabetics, whether bare-metal or drug-
treatment of multivessel coronary disease is not evidence eluting stents were used, and by using the one-study-out
based for hard end points. Therefore, our aim was to over- method. To address the issue of publication bias, the Begg-
come the power limitation of the existing data sets by per- Rank correlation method was used.14 The reported P values
forming a meta-analysis all randomized trials directly com- with this method are 2-tailed, with continuity correction. Ad-
paring CABG with PCI in the current era of high arterial graft ditionally, funnel plots were generated to further examine pub-
and stent use and examine the comparative effects of these lication bias. Comprehensive Meta Analysis software, ver-
procedures on long-term mortality and morbidity in sion 2.2.064 (Biostat Inc) was used for all analyses.The PRISMA
patients with multivessel disease. checklist for this meta-analysis can be found in the eTable in
the Supplement.

Methods
Results
Literature Search
A systematic search was made of MEDLINE using PubMed Search Results
through December 2012 to retrieve all published “random- The results of the literature search are shown in Figure 1. Of
ized controlled trials” comparing CABG and PCI in multives- the 102 results, 6 clinical trials without the exclusion criteria
sel coronary disease. The search term was [(bypass or enrolling a total of 6055 patients (3023 CABG, 3032 PCI) were
by-pass) and (PCI or stent) and (multi-vessel or multivessel or included in the meta-analysis. Supplementary searches of Sco-
three-vessel or three vessel or two vessel or two-vessel)]. The pus and Cochrane Central Register of Controlled Trials did not
search was limited to “randomized controlled trials,” and there reveal any additional relevant data.
was no time limit used in the search criteria. Supplementary
searches were made using Scopus (covering MEDLINE, Embase, Study and Patient Characteristics
and several other databases from a variety of disciplines) and The characteristics of the included trials are listed in Table 1.
Cochrane Central Register of Controlled Trials using similar The duration of longest follow-up varied between 1 and 6 years,
search terms. with a weighted average of 4.1 years. The CARDia 7 and

224 JAMA Internal Medicine February 2014 Volume 174, Number 2 jamainternalmedicine.com

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024

 CABG vs PCI in Multivessel Disease Original Investigation Research

FREEDOM16 trials were limited to patients with diabetes, and or 3-vessel coronary disease in all trials except the SYNTAX trial
the remaining 4 trials enrolled mostly nondiabetic patients multivessel group,9 where all patients had 3-vessel disease.
(77% nondiabetic). The SYNTAX9 and FREEDOM16 trials used No evidence of publication bias was detected when this
only drug-eluting stents; ARTS,11 MASS II,6 and SoS2 used only issue was examined by the Begg-Rank correlation method. The
bare-metal stents; and CARDia7 used both. Use of off-pump test statistic for the Begg approach, the Kendall τ, was non-
CABG was 0% to 31% in the included trials. Baseline patient significant for reporting of mortality and MI (P > .80 for both
characteristics are listed in Table 2 and Table 3. These were simi- mortality and MI). Funnel plots examining publication bias are
lar in the CABG and PCI arms of the individual studies, as ex- presented in Figure 2.
pected in large randomized trials. Left ventricular systolic func-
tion was preserved in most patients. Patients had either 2-vessel Quantitative Data Synthesis
The comparative effect of CABG vs PCI on total mortality is
Figure 1. Flowchart of Trials Included in the Meta-analysis shown in Figure 3. There was a significant 27% reduction in
total mortality with CABG compared with PCI (I2 = 0%; RR, 0.73
Literature Search [95% CI, 0.62-0.86]) (P < .001). There were numerically fewer
102 Search results identified at MEDLINE (search myocardial infarctions in all of the included trials (Figure 4).
limited to randomized clinical trials)
On meta-analysis there was a significant 42% reduction in MI
with CABG compared with PCI (I2 = 8.02%; RR, 0.58 [95% CI,
96 Excluded 0.48-0.72]) (P < .001). There was a trend toward excess strokes
33 Substudies with no additional information
3 Reviews, editorials, meta-analysis with CABG (I2 = 24.9%; RR, 1.36 [95% CI, 0.99-1.86]), but this
5 Study protocol descriptions was not statistically significant (P = .06) (Figure 5). Repeat re-
3 Nonrandomized studies
36 Did not present clinical outcomes of vascularizations (I2 = 75.6%; RR, 0.29 [95% CI, 0.21-0.41])
interest (P < .001) and major adverse cardiac and cerebrovascular events
6 Not a coronary intervention trial
2 Early reports of clinical trials without (MACCE) (I2 = 33.0%; RR, 0.61 [95% CI, 0.54-0.68]) (P < .001)
relevant data
8 Not multivessel disease
were significantly reduced with CABG compared with PCI
(Figure 6 and Figure 7).
The number needed to treat was calculated using the ob-
6 Clinical trials included
ARTS10,11 tained meta-analytic RRs and observed cumulative event rates
MASS II6 in the PCI arms of the trials. Accordingly, CABG had to be pre-
SoS2,15
CARDia7 ferred over PCI in 37 patients to save 1 life and in 26 patients
SYNTAX multivessel disease9,12 to prevent 1 MI for the weighted average duration of fol-
FREEDOM16
low-up of 4.1 years. The number needed to treat was 7 for re-
peat revascularizations and 10 for MACCE. The number needed
For study acronym expansions, see the cited references.
to harm was 105 to cause 1 excess stroke with CABG.

Table 1. Characteristics of Randomized Trials of CABG vs PCI in Patients With Multivessel Disease Included in The Meta-analysis
Study
(Publication Patients Assigned to
Year of Each Arm, No. ≥1 Arterial ≥1 Stent Off-Pump
Longest Patients, Follow-up, Outcomes of Interest Graft Used Used in CABG Type of
Follow-up) Total No. CABG PCI Median, y Assessed in CABG, % PCI, % Rate, % Stent Used
10,11
ARTS, 1174 584 590 5 Death, MI, stroke, repeat 93 99 0 BMS
(2005) revascularization,
MACCEa
MASS II6 408 203 205 5 Death, MI, stroke, repeat >92 72 0 BMS
(2007)b revascularization
2,15
SoS 988 500 488 6 Death, repeat revascular- 93 >78 3 BMS
(2008) ization (at second year
only)
CARDia7 490 242 248 1 Death, MI, stroke, repeat 94 100 31 69% DES
(2010) revascularization, (sirolimus);
MACCEa 31% BMS
SYNTAX 1,095 547 548 3 Death, MI, Stroke, 97.3 100 15 100% Paclitaxel
multivessel9,12 Repeat revascularization, DES
(2011)c MACCE
FREEDOM16 1900 947 953 3.8 Death, MI, stroke, repeat 94.4 >94 18.5 Any DES
(2012) revascularization (at first (51% sirolimus;
year only), MACCEa 43% paclitaxel
(at first year only)
b
Abbreviations: CABG, coronary artery bypass grafting; BMS, bare-metal stent; MASS II trial had 3 arms (medical therapy vs CABG vs PCI). Only the CABG and
DES, drug-eluting stent; MACCE, major adverse coronary and cerebrovascular PCI arms were included in this meta-analysis.
events; PCI, percutaneous coronary intervention. c
Number assigned to each arm were calculated assuming equal 1:1 distribution.
a
MACCE is the combined end point of death, non-fatal MI, non-fatal stroke and Data for trial characteristics are from the whole SYNTAX population.
repeat revascularization.

jamainternalmedicine.com JAMA Internal Medicine February 2014 Volume 174, Number 2 225

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024

 Research Original Investigation CABG vs PCI in Multivessel Disease

Sensitivity Analysis There was also no evidence of heterogeneity according to type

The magnitude of risk reduction in mortality was similar in of stent used (heterogeneity P = .56 for bare-metal vs drug-
trials limited to patients with diabetes and those not limited eluting stents). The findings of the meta-analysis remained
to patients with diabetes (P = .80 for heterogeneity) (Table 4). stable with the one-study-out method ruling out the possibility

Table 2. Characteristics of Patients Enrolled in Randomized Trials of CABG vs PCI in Patients With Multivessel Disease Included in the Meta-analysisa
Mean or
Mean Age, y Male Diabetes Hypertension Hyperlipidemia Median EF, % Smoker Previous MI
Source CABG PCI CABG PCI CABG PCI CABG PCI CABG PCI CABG PCI CABG PCI CABG PCI
ARTS10,11 61 61 76 77 16 19 45 45 58 58 60 61 26 28 42 44
MASS II6 60 60 72 67 29 23 63 61 NA NA 67 67 32a 27a 41 52
SoS2,15 62 61 78 80 15 14 47 43 50 53 57 57 14 16 47 44
CARDia7 64 64 71 78 100 100 77 81 93 82 60 59 23 25 NA NA
SYNTAXb 65 65 79 76 35 36 64 69 77 79 NA NA 22 19 34 32
multivessel9,12
FREEDOM16 63 63 70 73 100 100 85c 84c 67 66 17 15 25 26
b
Abbreviations: CABG, coronary artery bypass grafting; EF, ejection fraction; NA, Data for patient characteristics are from the whole SYNTAX population.
not available; PCI, percutaneous coronary intervention. c
Rates for the whole FREEDOM cohort.
a
Unless otherwise noted, data are reported as percentage of participants.

Table 3. Disease Types in Patients Enrolled in Randomized Trials of CABG vs PCI in Patients With Multivessel Disease Included in the Meta-analysisa

2-Vessel Disease 3-Vessel Disease Unstable Angina Stable Angina

Source CABG PCI CABG PCI CABG PCI CABG PCI
ARTS10,11 67 68 33 30 35 37 NA NA
MASS II6 42 42 58 58 0 0 100 100
SoS2,15 52 62 47 38 NA NA NA NA
CARDia7 35 28 60 65 ‘‘Mostly stable coronary artery disease’’
SYNTAXb multivessel9,12 0 0 100 100 28 29 57 57
FREEDOM16 16 18 84 82 31c 69c
b
Abbreviations: CABG, coronary artery bypass grafting; NA, not available; PCI, Data for patient characteristics are from the whole SYNTAX population.
percutaneous coronary intervention. c
Of the whole FREEDOM cohort, 31% had acute coronary syndrome, and 69%
a
All data are reported as percentage of participants. had stable coronary artery disease.

Table 4. Sensitivity Analyses for the Outcome of Mortality

Comparison of
2 Groups for
Characteristic I 2, % Model RR (95% CI) P Value Heterogeneity
Diabetes Status
Trials limited to patients with diabetes 0 Fixed 0.75 (0.58-0.97) .03
(CARDia,7 FREEDOM16) (n = 2390) effects
Trials not limited to patients with diabetes 13.3 Fixed 0.72 (0.58-0.89) .003 0.80
(ARTS,10,11 MASS II,6 SoS,2,15 SYNTAX effects
multivessel9,12) (n = 3665)
Type of Stent Used
Trials using only drug-eluting stents (SYNTAX 0 Fixed 0.69 (0.55-0.87) .001
multivessel,9,12 FREEDOM16) (n = 2995) effects
0.56
Trials using only bare-metal stents (ARTS,10,11 19.9 Fixed 0.77 (0.59-0.99) .04
MASS II,6 SoS2,15) (n = 2570) effects
Excluded Study in 1-Study-Out Model
ARTS10,11 0 Fixed 0.69 (0.57-0.83) <.001 NA
effects
MASS II6 0 Fixed 0.74 (0.62-0.88) <.001 NA
effects
SoS2,15 0 Fixed 0.75 (0.63-0.90) .002 NA
effects
CARDia7 0 Fixed 0.72 (0.61-0.86) <.001 NA
effects
SYNTAX multivessel9,12 0 Fixed 0.76 (0.63-0.91) .003 NA
effects
FREEDOM16 0 Fixed 0.73 (0.59-0.91) .004 NA
effects Abbreviation: NA, not applicable

226 JAMA Internal Medicine February 2014 Volume 174, Number 2 jamainternalmedicine.com

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024

 CABG vs PCI in Multivessel Disease Original Investigation Research

of a single clinical trial dominating the results of the meta- dered obsolete much of the surgical outcomes data from the
analysis. There was also no statistically significant heteroge- clinical trials published before the turn of the century.18-21 As
neity for risk reduction in MI according to whether trials were PCI methods continue to evolve and surgical outcomes im-
limited to patients with diabetes or the type of stent used
(heterogeneity P > .10 for both). For the outcome of stroke, Figure 2. Funnel Plots Examining Publication Bias for Mortality (A)
there was again no heterogeneity in results according to dia- and Myocardial Infarction (B)
betes status or the type of stent used (heterogeneity P > .10 for
both). For repeat revascularizations, there was significant 0.0

heterogeneity according to the type of stent used (P = .002),

with greater risk reduction in this outcome with CABG if bare- 0.1

metal stents were used (RR, 0.27 [95% CI, 0.22-0.34] with bare-
metal stents vs RR, 0.45 [95% CI, 0.36-0.56] with drug-

Standard Error
0.2
eluting stents). There was no significant heterogeneity
according to diabetes status for repeat revascularizations 0.3
(P > .10).
One well-known clinical trial from South America did not 0.4
meet the study inclusion criteria because the frequency of the
use of arterial grafts in the CABG arm and the frequency of stent
0.5
use in the PCI arm of this trial were too low.17 A sensitivity analy-
–2.0 –1.5 –1.0 –0.5 0.0 0.5 1.0 1.5 2.0
sis adding this trial did not change the statistically significant
Log Risk Ratio
reduction in mortality with CABG compared with PCI.
0.0

Discussion 0.1

This meta-analysis of the contemporary era shows that in pa-

Standard Error

tients with multivessel CAD, CABG reduces long-term mortal- 0.2

ity by 27% compared with PCI, regardless of whether the study
population is limited to patients with diabetes or not. Regard-
ing major morbidity, a 42% risk reduction in MI was observed 0.3

in patients randomized to CABG. There was a trend for excess

strokes with CABG, probably related to an increase in peripro-
0.4
cedural strokes. However, the absolute risk increase in stroke
was small compared with the absolute risk reduction in mor- –2.0 –1.5 –1.0 –0.5 0.0 0.5 1.0 1.5 2.0
Log Risk Ratio
tality and MI, as demonstrated by the numbers needed to treat.
Although CAD is a leading cause of death worldwide, the
Log risk ratios less than 0 favor coronary artery bypass grafting; those greater than
optimal treatment strategy for this disease remains to be well
0 favor percutaneous coronary intervention. These funnel plots represent a mea-
defined. There have been important advances in nonsurgical sure of study size on the vertical axis as a function of effect size on the horizontal
therapies, including drug-eluting stents, newer anticoagulant- axis. Large studies appear toward the top of the graph, and tend to cluster near the
antiplatelet drug regimens, and aggressive lipid-lowering treat- mean effect size. Smaller studies appear toward the bottom of the graph and (since
there is more sampling variation in effect size estimates in the smaller studies) will
ment, all of which have led to improved outcomes in nonsur- be dispersed across a range of values. In the absence of publication bias, as is dem-
gically treated patients with multivessel CAD. Additionally, onstrated in these funnel plots, the studies, represented by pale dotted circles, are
improvements in surgical techniques including nearly univer- distributed symmetrically about the combined effect size. The dashed diamond
appearing below the x-axis represents the summary effect.
sal arterial graft use and better postoperative care have ren-

Figure 3. Mortality According to Treatment Arm

Statistics for Each Study Death/Total

Source RR (95% CI) Z Value P Value CABG PCI Favors CABG Favors PCI
ARTS10,11 0.97 (0.66-1.43) –0.16 .87 46/584 48/590
MASS II6 0.67 (0.37-1.23) –1.29 .20 16/203 24/205
SoS2,15 0.63 (0.41-0.95) –2.23 .03 34/500 53/488
CARDia7 1.02 (0.39-2.69) 0.05 .96 8/242 8/248
SYNTAX multivessel9,12 0.60 (0.39-0.92) –2.36 .02 31/547 52/548 Total number of patients, 6055
FREEDOM16 0.73 (0.56-0.95) –2.31 .02 86/947 118/953 (I2 = 0% for the fixed effects model).
Meta-analysis 0.73 (0.62-0.86) –3.69 <.001 221/3023 303/3032 CABG indicates coronary artery
bypass graft; PCI, percutaneous
0.5 1.0 2.0 coronary intervention; RR, risk ratio;
RR (95% CI) for expansion of all study name
acronyms, see the cited references.

jamainternalmedicine.com JAMA Internal Medicine February 2014 Volume 174, Number 2 227

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024

 Research Original Investigation CABG vs PCI in Multivessel Disease

prove, it has become increasingly difficult to answer the ulti- making the results difficult to interpret. This is partially be-
mate question: “What is the best revascularization method for cause, unlike in previous eras, in the present era, the annual
the patient with multivessel CAD?” mortality of these patients is very low, under 2% to 3% in most
The more recent stent era trials comparing CABG with PCI studies. Consequently, the composite primary end point of
have been underpowered for mortality and major morbidity, MACCE was introduced to overcome the power limitation of

Figure 4. Myocardial Infarctions (MIs) According to Treatment Arm

Statistics for Each Study MI/Total

Source RR (95% CI) Z Value P Value CABG PCI Favors CABG Favors PCI
ARTS10,11 0.77 (0.52-1.15) –1.26 0.21 39/584 51/590
MASS II6 0.75 (0.41-1.35) –0.96 0.34 17/203 23/205
CARDia7 0.57 (0.31-1.08) –1.73 0.08 14/242 25/248 Total number of patients, 5067
SYNTAX multivessel9,12 0.46 (0.27-0.80) –2.77 0.01 18/547 39/548 (I2 = 8.02% for the fixed effects
FREEDOM16 0.49 (0.35-0.68) –4.23 <.001 48/947 99/953 model). CABG indicates coronary
Meta-analysis 0.58 (0.48-0.72) –5.17 <.001 136/2523 237/2544
artery bypass graft; PCI,
percutaneous coronary intervention;
0.5 1.0 2.0 RR, risk ratio; for expansion of all
RR (95% CI) study name acronyms, see the cited
references.

Figure 5. Strokes According to Treatment Arm

Statistics for Each Study Stroke/Total

Source RR (95% CI) Z Value P Value CABG PCI Favors CABG Favors PCI
ARTS10,11 0.92 (0.52-1.65) –0.27 0.79 21/584 23/590
MASS II6 1.73 (0.70-4.31) 1.18 0.24 12/203 7/205
CARDia7 7.17 (0.89-57.87) 1.85 0.06 7/242 1/248 Total number of patients, 5067
SYNTAX multivessel9,12 1.14 (0.56-2.32) 0.38 0.71 16/547 14/548 (I2 = 24.9% for the fixed effects
FREEDOM16 1.69 (1.01-2.85) 1.98 0.05 37/947 22/953 model). CABG indicates coronary
Meta-analysis 1.36 (0.99-1.86) 1.91 0.06 93/2523 67/2544
artery bypass graft; PCI,
percutaneous coronary intervention;
0.5 1.0 2.0 RR, risk ratio; for expansion of all
RR (95% CI) study name acronyms, see the cited
references.

Figure 6. Repeat Revascularizations According to Treatment Arm

Repeat
Revascularization/
Statistics for Each Study Total
Source RR (95% CI) Z Value P Value CABG PCI Favors CABG Favors PCI
ARTS10,11 0.29 (0.22-0.39) –8.45 <.001 53/584 182/590
MASS II6 0.11 (0.05-0.23) –5.80 <.001 7/203 66/205
SoS2,15 0.29 (0.20-0.43) –6.26 <.001 30/500 101/488
CARDia7 0.17 (0.07-0.43) –3.72 <.001 5/242 30/248 Total number of patients, 6055
SYNTAX multivessel9,12 0.52 (0.38-0.70) –4.23 <.001 55/547 106/548 (I2 = 75.6% for the random effects
FREEDOM16 0.37 (0.26-0.51) –5.83 <.001 43/947 118/953 model). CABG indicates coronary
Meta-analysis 0.29 (0.21-0.41) –7.00 <.001 193/3023 603/3032
artery bypass graft; PCI,
percutaneous coronary intervention;
0.1 1.0 10 RR, risk ratio; for expansion of all
RR (95% CI) study name acronyms, see the cited
references.

Figure 7. Major Adverse Cardiovascular and Cerebrovascular Events (MACCE) According to Treatment Arm

Statistics for Each Study MACCE/Total

Source RR (95% CI) Z Value P Value CABG PCI Favors CABG Favors PCI
ARTS10,11 0.53 (0.45-0.64) –6.95 <.001 132/584 250/590
CARDia7 0.59 (0.38-0.90) –2.44 .01 28/242 49/248 Total number of patients, 4659
SYNTAX multivessel9,12 0.65 (0.53-0.81) –3.83 <.001 103/547 158/548 (I2 = 33.0% for the fixed effects
FREEDOM16 0.71 (0.57-0.89) –2.95 <.001 112/947 158/953 model). CABG indicates coronary
Meta-analysis 0.61 (0.54-0.68) –8.55 <.001 375/2320 615/2339 artery bypass graft; PCI,
percutaneous coronary intervention;
0.5 1.0 2.0 RR, risk ratio; for expansion of all
RR (95% CI) study name acronyms, see the cited
references.

228 JAMA Internal Medicine February 2014 Volume 174, Number 2 jamainternalmedicine.com

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024

 CABG vs PCI in Multivessel Disease Original Investigation Research

these studies, and CABG almost always led to lower MACCE ity of patients with multivessel disease are nondiabetic.24
rates.7,9,11 However, MACCE is driven mainly by the soft end Our results strongly suggest that CABG should be the revas-
point of repeat revascularizations, which can be acceptable cularization method in patients with multivessel CAD,
for many physicians and patients wishing to avoid cardiac regardless of their diabetic status. However, it should be
surgery. This is especially important in the absence of evi- remembered that the included trials enrolled patients
dence for significant improvement in mortality or major mostly with stable or unstable angina and excluded patients
morbidity such as MI with CABG.3-5,7-10 For example, in the with acute MI. Therefore, our findings do not apply to the
seminal first report of one of the landmark trials comparing type of patients who were systematically excluded from
CABG with PCI, mortality and MI rates were similar, while these trials.
the stroke rate was higher in the CABG arm.9 As a result of Our results must be interpreted in light of several limita-
this and other previous trials, as well as very large observa- tions. This was necessarily a trial-level meta-analysis be-
tional data sets reporting no mortality or morbidity benefit cause we did not have access to individual patient-level data.
with CABG,22 practice patterns shifted toward stenting.13 To Therefore, we were not able to perform subgroup analysis to
overcome the limitations of the underpowered studies, we see whether the superiority of CABG over PCI for mortality
performed this meta-analysis pooling data from multiple reduction was limited to certain subgroups (eg, those with in-
studies including a total of more than 6000 patients. Our termediate to high SYNTAX scores or those with 3-vessel dis-
analysis demonstrates that both long-term mortality and ease). Also, it may be argued that newer generation drug-
MIs are reduced significantly with CABG compared with eluting stents that are now commonly used during PCI such
PCI, regardless of whether drug-eluting or bare-metal stents as the everolimus- or the zotarolimus-eluting stents were not
are used. The validity of our findings are supported by a tested in the trials included in this meta-analysis. In this con-
recent propensity-matched analysis of over 100 000 text, it should be noted that the newer generation drug-
patients reporting superior survival and lower MI rates with eluting stents did not improve mortality compared with the
multivessel CABG compared with multivessel PCI. 23 It is sirolimus- or paclitaxel-eluting stents25,26 or the bare-metal
notable that the results of the clinical trials included in this stents27 in randomized controlled trials. Another argument
meta-analysis were homogeneous for all of the outcomes could be that CABG may still not be the best approach for the
studied (ie, I2<40%) except for the outcome of repeat revas- management of patients with multivessel disease because our
cularization. For this outcome, the effect size was relatively meta-analysis compared CABG only to PCI and not to medi-
smaller for the SYNTAX9 and FREEDOM16 trials (RR, >0.35 in cal therapy, and CABG may not be superior to medical therapy
both), where there was universal use of drug-eluting stents alone. In this context, there are 2 major contemporary ran-
that reduce in-stent restenosis, compared with the other domized trials comparing CABG with medical therapy.6,28 The
trials, which used bare-metal stents (RRs <0.30). MASS II trial,6 which was also included in our meta-analysis,
It has been long debated whether the presence of diabe- is one of these trials. This trial, primarily enrolling nondia-
tes should dictate the revascularization method in patients betic patients, had 3 arms, namely, CABG, PCI, and medical
with multivessel CAD. Traditionally, surgery has been pre- therapy arms. In MASS II, the 5-year mortality was 12.8% with
ferred over PCI for this population. Evidence for this is CABG and 16.2% with medical therapy, although the differ-
largely based on the BARI study19 and comes from the plain ence was not statistically significant. Risk of acute MI was sig-
balloon era. While the BARI study did not show an overall nificantly reduced with CABG compared with medical therapy
mortality benefit between the 2 revascularization methods, (RR, 0.41 [95% CI, 0.18-0.94]). The BARI 2D trial28 including
a post hoc subgroup analysis of diabetic patients showed a patients with diabetes is the other relevant trial. In BARI 2D,
long-term mortality of 34.5% for balloon angioplasty and within the CABG stratum, MIs were significantly less fre-
19.4% for surgery (P = .03). Very recently, the results of the quent in CABG plus intensive medical therapy vs intensive
FREEDOM study 16 enrolling only diabetic patients con- medical therapy alone groups (10.0% vs 17.6%) (P = .003), and
firmed the mortality and morbidity benefit of CABG over the composite end point of death or MI (21.1% vs 29.2%) (P = .01)
PCI in this population. Therefore, it may be argued that the was also less frequent. These data, along with other data show-
benefit of CABG over PCI is limited to patients with diabetes ing equivalence of stenting with medical therapy in stable mul-
and that the mortality benefit of CABG seen in our meta- tivessel coronary disease, suggests that CABG is not only su-
analysis is driven by diabetic patients. In this regard, among perior to PCI but also to medical therapy for at least prevention
the trials included in our meta-analysis, 2 of them were lim- of MI. Nevertheless, an appropriately sized randomized trial
ited to patients with diabetes alone, and 4 of the trials examining the effect of CABG vs medical therapy on total mor-
included primarily nondiabetic patients. On further analy- tality in patients with preserved ejection fraction does not ex-
sis, there was no heterogeneity in reduction of mortality ist for the current era.
and MIs among the trials limited to and not limited to dia-
betic patients. The effect size for mortality reduction was
very similar in trials enrolling only diabetic patients (25%)
and the trials enrolling primarily nondiabetic patients
Conclusions
(28%). While the FREEDOM trial is a landmark study that In patients with multivessel coronary disease, CABG does
will consolidate the approach to revascularization in not only lead to a dramatic reduction in repeat revascular-
patients with diabetes and multivessel CAD, the vast major- ization and MACCE but also leads to a 27% reduction in

jamainternalmedicine.com JAMA Internal Medicine February 2014 Volume 174, Number 2 229

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024

 Research Original Investigation CABG vs PCI in Multivessel Disease

long-term all-cause mortality and a 42% reduction in MIs drug-eluting stents did not alter the mortality benefit. Given
compared with PCI. The benefits were not only observed in these meaningful benefits, CABG should be the preferred
trials of diabetic patients but also in trials where the great revascularization method for most patients with multives-
majority of patients were nondiabetic. Use of bare-metal or sel coronary artery disease.

ARTICLE INFORMATION Artery Revascularization in Diabetes) trial. J Am Coll coronary bypass surgery with angioplasty in
Accepted for Publication: May 7, 2013. Cardiol. 2010;55(5):432-440. patients with multivessel disease. N Engl J Med.
8. Mercado N, Wijns W, Serruys PW, et al. One-year 1996;335(4):217-225.
Published Online: December 2, 2013.
doi:10.1001/jamainternmed.2013.12844. outcomes of coronary artery bypass graft surgery 20. Henderson RA, Pocock SJ, Sharp SJ, et al.
versus percutaneous coronary intervention with Long-term results of RITA-1 trial: clinical and cost
Authors Contributions: Dr Sipahi had full access to multiple stenting for multisystem disease: a comparisons of coronary angioplasty and
all the data used in the study and takes meta-analysis of individual patient data from coronary-artery bypass grafting. Randomised
responsibility for the integrity of the data and randomized clinical trials. J Thorac Cardiovasc Surg. Intervention Treatment of Angina. Lancet.
accuracy of the analysis. 2005;130(2):512-519. 1998;352(9138):1419-1425.
Study concept and design: Sipahi, Akay, Dagdelen,
Alhan. 9. Serruys PW, Morice MC, Kappetein AP, et al; 21. King SB III, Lembo NJ, Weintraub WS, et al. A
Acquisition of data: Sipahi, Akay. SYNTAX Investigators. Percutaneous coronary randomized trial comparing coronary angioplasty
Analysis and interpretation of data: Sipahi, Blitz, intervention versus coronary-artery bypass grafting with coronary bypass surgery. Emory Angioplasty
Alhan. for severe coronary artery disease. N Engl J Med. versus Surgery Trial (EAST). N Engl J Med.
Drafting of the manuscript: Sipahi, Akay. 2009;360(10):961-972. 1994;331(16):1044-1050.
Critical revision of the manuscript for important 10. Serruys PW, Ong AT, van Herwerden LA, et al. 22. Benedetto U, Melina G, Angeloni E, et al.
intellectual content: Sipahi, Dagdelen, Blitz, Alhan. Five-year outcomes after coronary stenting versus Coronary artery bypass grafting versus drug-eluting
Statistical analysis: Sipahi. bypass surgery for the treatment of multivessel stents in multivessel coronary disease. A
Administrative, technical, or material support: disease: the final analysis of the Arterial meta-analysis on 24,268 patients. Eur J
Sipahi. Revascularization Therapies Study (ARTS) Cardiothorac Surg. 2009;36(4):611-615.
Study supervision: Sipahi, Dagdelen, Alhan. randomized trial. J Am Coll Cardiol. 23. Hlatky MA, Boothroyd DB, Baker L, et al.
Conflict of Interest Disclosures: None reported. 2005;46(4):575-581. Comparative effectiveness of multivessel coronary
11. Serruys PW, Unger F, Sousa JE, et al; Arterial bypass surgery and multivessel percutaneous
REFERENCES Revascularization Therapies Study Group. coronary intervention: a cohort study. Ann Intern
1. Roger VL, Go AS, Lloyd-Jones DM, et al; Comparison of coronary-artery bypass surgery and Med. 2013;158(10):727-734.
American Heart Association Statistics Committee stenting for the treatment of multivessel disease. 24. Shahian DM, O’Brien SM, Filardo G, et al;
and Stroke Statistics Subcommittee. Heart disease N Engl J Med. 2001;344(15):1117-1124. Society of Thoracic Surgeons Quality Measurement
and stroke statistics—2012 update: a report from 12. Kappetein AP, Feldman TE, Mack MJ, et al. Task Force. The Society of Thoracic Surgeons 2008
the American Heart Association. Circulation. Comparison of coronary bypass surgery with cardiac surgery risk models: part 1--coronary artery
2012;125(1):e2-e220. drug-eluting stenting for the treatment of left main bypass grafting surgery. Ann Thorac Surg.
2. SoS Investigators. Coronary artery bypass and/or three-vessel disease: 3-year follow-up of the 2009;88(1)(Suppl):S2-S22.
surgery versus percutaneous coronary intervention SYNTAX trial. Eur Heart J. 2011;32(17):2125-2134. 25. Wei G, Fang Y, Yaqi R, Lin C, Ningfu W. Clinical
with stent implantation in patients with multivessel 13. Riley RF, Don CW, Powell W, Maynard C, Dean outcomes of zotarolimus-eluting stents versus
coronary artery disease (the Stent or Surgery trial): LS. Trends in coronary revascularization in the the first generation sirolimus-eluting stents
a randomised controlled trial. Lancet. United States from 2001 to 2009: recent declines and paclitaxel-eluting stents: a meta-analysis
2002;360(9338):965-970. in percutaneous coronary intervention volumes. of randomized trials. Int J Cardiol.
3. Bravata DM, Gienger AL, McDonald KM, et al. Circ Cardiovasc Qual Outcomes. 2011;4(2):193-197. 2012;157(1):152-156.
Systematic review: the comparative effectiveness 14. Begg CB, Mazumdar M. Operating 26. de Waha A, Cassese S, Park DW, et al.
of percutaneous coronary interventions and characteristics of a rank correlation test for Everolimus-eluting versus sirolimus-eluting stents:
coronary artery bypass graft surgery. Ann Intern publication bias. Biometrics. 1994;50(4):1088-1101. an updated meta-analysis of randomized trials. Clin
Med. 2007;147(10):703-716. 15. Booth J, Clayton T, Pepper J, et al; SoS Res Cardiol. 2012;101(6):461-467.
4. Daemen J, Boersma E, Flather M, et al. Investigators. Randomized, controlled trial of 27. Eisenstein EL, Leon MB, Kandzari DE, et al;
Long-term safety and efficacy of percutaneous coronary artery bypass surgery versus ENDEAVOR III Investigators. Long-term clinical and
coronary intervention with stenting and coronary percutaneous coronary intervention in patients economic analysis of the Endeavor
artery bypass surgery for multivessel coronary with multivessel coronary artery disease: six-year zotarolimus-eluting stent versus the cypher
artery disease: a meta-analysis with 5-year follow-up from the Stent or Surgery Trial (SoS). sirolimus-eluting stent: 3-year results from the
patient-level data from the ARTS, ERACI-II, MASS-II, Circulation. 2008;118(4):381-388. ENDEAVOR III trial (Randomized Controlled Trial of
and SoS trials. Circulation. 2008;118(11):1146-1154. 16. Farkouh ME, Domanski M, Sleeper LA, et al; the Medtronic Endeavor Drug [ABT-578] Eluting
5. Hlatky MA, Boothroyd DB, Bravata DM, et al. FREEDOM Trial Investigators. Strategies for Coronary Stent System Versus the Cypher
Coronary artery bypass surgery compared with multivessel revascularization in patients with Sirolimus-Eluting Coronary Stent System in De
percutaneous coronary interventions for diabetes. N Engl J Med. 2012;367(25):2375-2384. Novo Native Coronary Artery Lesions). JACC
multivessel disease: a collaborative analysis of Cardiovasc Interv. 2009;2(12):1199-1207.
17. Rodriguez AE, Baldi J, Fernández Pereira C,
individual patient data from ten randomised trials. et al; ERACI II Investigators. Five-year follow-up of 28. Chaitman BR, Hardison RM, Adler D, et al;
Lancet. 2009;373(9670):1190-1197. the Argentine randomized trial of coronary Bypass Angioplasty Revascularization Investigation
6. Hueb W, Lopes NH, Gersh BJ, et al. Five-year angioplasty with stenting versus coronary bypass 2 Diabetes (BARI 2D) Study Group. The
follow-up of the Medicine, Angioplasty, or Surgery surgery in patients with multiple vessel disease Bypass Angioplasty Revascularization Investigation
Study (MASS II): a randomized controlled clinical (ERACI II). J Am Coll Cardiol. 2005;46(4):582-588. 2 Diabetes randomized trial of different
trial of 3 therapeutic strategies for multivessel treatment strategies in type 2 diabetes mellitus
18. First-year results of CABRI (Coronary with stable ischemic heart disease: impact of
coronary artery disease. Circulation. Angioplasty versus Bypass Revascularisation
2007;115(9):1082-1089. treatment strategy on cardiac mortality and
Investigation): CABRI Trial Participants. Lancet. myocardial infarction. Circulation.
7. Kapur A, Hall RJ, Malik IS, et al. Randomized 1995;346(8984):1179-1184.
2009;120(25):2529-2540.
comparison of percutaneous coronary intervention 19. The Bypass Angioplasty Revascularization
with coronary artery bypass grafting in diabetic Investigation (BARI) Investigators. Comparison of
patients. 1-year results of the CARDia (Coronary

230 JAMA Internal Medicine February 2014 Volume 174, Number 2 jamainternalmedicine.com

Copyright 2014 American Medical Association. All rights reserved.

Downloaded from jamanetwork.com by Leo Barros on 03/27/2024