- Email: [email protected]
The Smoker’s Paradox Revisited
JACC: CARDIOVASCULAR INTERVENTIONS
VOL. 12, NO. 19, 2019
ª 2019 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION PUBLISHED BY ELSEVIER
The Smoker’s Paradox Revisited A Patient-Level Pooled Analysis of 18 Randomized Controlled Trials Mayank Yadav, MD,a Gary S. Mintz, MD,b Philippe Généreux, MD,b,c,d Mengdan Liu, MS,b Thomas McAndrew, PHD,b Björn Redfors, MD, PHD,b Mahesh V. Madhavan, MD,e Martin B. Leon, MD,b,e Gregg W. Stone, MDb,e
ABSTRACT OBJECTIVES This study examined the smoker’s paradox using patient-level data from 18 prospective, randomized trials of patients undergoing percutaneous coronary intervention (PCI) with stent implantation. BACKGROUND Studies on the effects of smoking and outcomes among patients undergoing PCI have reported conflicting results. METHODS Data from the RAVEL, E-SIRIUS, SIRIUS, C-SIRIUS, TAXUS IV and V, ENDEAVOR II to IV, SPIRIT II to IV, HORIZONS-AMI, COMPARE I and II, PLATINUM, and TWENTE I and II randomized trials were pooled. Patients were stratified by smoking status at time of enrollment. The 1- and 5-year ischemic outcomes were compared. RESULTS Among 24,354 patients with available data on smoking status, 6,722 (27.6%) were current smokers. Smokers were younger and less likely to have diabetes mellitus; hypertension; hyperlipidemia; or prior myocardial infarction (MI), PCI, or coronary artery bypass grafting. Angiographically, smokers had longer lesions, more complex lesions, and more occlusions, but were less likely to have moderate or severe calcification or tortuosity. At 5 years, smokers had significantly higher rates of MI (7.8% vs. 5.6%; p < 0.0001) and definite or probable stent thrombosis (3.5% vs. 1.8%; p < 0.0001); however, there were no differences in the rates of death, cardiac death, target lesion revascularization, or composite endpoints (cardiac death, target vessel MI, or ischemic target lesion revascularization). After multivariable adjustment for potential confounders, smoking was a strong independent predictor of death (hazard ratio [HR]: 1.86; 95% confidence interval [CI]: 1.63 to 2.12; p < 0.0001), cardiac death (HR: 1.68; 95% CI: 1.38 to 2.05; p < 0.0001), MI (HR: 1.38; 95% CI: 1.20 to 1.58; p < 0.0001), stent thrombosis (HR: 1.60; 95% CI: 1.28 to 1.99; p < 0.0001), and target lesion failure (HR: 1.17; 95% CI: 1.05 to 1.30; p ¼ 0.005). CONCLUSIONS The present large, patient-level, pooled analysis with 5-year follow-up clearly demonstrates smoking to be an important predictor of adverse outcomes after PCI. (J Am Coll Cardiol Intv 2019;12:1941–50) © 2019 by the American College of Cardiology Foundation.
S
moking is one of the strongest modifiable risk
after percutaneous coronary intervention (PCI), a
factors for coronary artery disease (CAD). Thir-
phenomenon often termed the smoker’s paradox.
ty percent of CAD deaths in United States are
Although most of these studies were completed in
attributed to smoking, with the risk being dose
the thrombolytic therapy and balloon angioplasty
related (1,2). Smoking cessation constitutes the single
eras, and even though it has been 3 decades since
most important preventive measure for CAD (1). Para-
the first description of this paradox (3–9), the reasons
doxically, many (3–9), but not all (10–13), studies in
underlying this enigma remain unclear.
the past 3 decades have reported that smoking is associated with better short- and long-term outcomes
To further investigate the relationship between smoking
status
and
5-year
outcomes
in
the
From the aDepartment of Medicine, Division of Cardiology, Bronx Lebanon Hospital Center, New York, New York; bClinical Trials Center, Cardiovascular Research Foundation, New York, New York; cGagnon Cardiovascular Institute, Morristown Medical Center, Morristown, New Jersey; dHôpital du Sacré-Coeur de Montréal, Montréal, Canada; and the eDepartment of Medicine, Division of Cardiology, NewYork-Presbyterian Hospital/Columbia University Medical Center, New York, New York. This investigatorsponsored study was funded by Abbott Vascular (Santa Clara, California). The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Manuscript received February 21, 2019; revised manuscript received May 16, 2019, accepted June 11, 2019.
ISSN 1936-8798/$36.00
https://doi.org/10.1016/j.jcin.2019.06.034
Yadav et al.
1942
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
ABBREVIATIONS
contemporary era of PCI with stent implan-
AND ACRONYMS
tation, we examined this issue from an indi-
Research Consortium definition) (32), and target
vidual patient data (IPD) pooled analysis of
lesion failure (TLF) (a composite of cardiac death,
18 large-scale, multicenter, prospective, ran-
target vessel MI, and ischemic TLR). If TLF was not
domized controlled trials.
reported in the original trials, it was derived based on
ACS = acute coronary syndromes
CAD = coronary artery disease
individual components. TLF was evaluated in this
SEE PAGE 1951
CABG = coronary artery bypass
manner in 6 studies (N ¼ 6,808). Overall, TLF was
grafting
METHODS
IPD = individual patient data
definite or probable stent thrombosis (ST) (Academic
available in 12 of the 18 studies (19,349 of 24,354 patients) included in the present analysis. Survival time
MI = myocardial infarction PCI = percutaneous coronary intervention
STUDY PROTOCOL. The study population
was calculated as days from procedure date to first-
was drawn from an IPD pooled database
event date.
consisting of 18 multicenter, prospective,
ST = stent thrombosis
single- or double-blinded, randomized clin-
TLF = target lesion failure
ical trials with a follow-up period up to 5
TLR = target lesion
years. The study protocols and trial designs
revascularization
have been previously described in detail and are summarized in Table 1 (14–31). The main purpose of the present study was to evaluate the impact of smoking on 1- and 5-year outcomes among patients undergoing PCI in the stent era. Current smoking status at baseline was used to stratify the patient population.
STATISTICAL METHODS. Continuous data are pre-
sented as mean SD and were compared using the Student’s t-test or Wilcoxon rank sum test, as appropriate. Categorical variables are presented as frequency and percent and were compared using the chi-square test or the Fisher exact test (when any cell’s expected frequency fell below 5 counts). Clinical endpoints were presented using Kaplan-Meier estimates and compared using the log-rank test. Data were also analyzed stratified by study (CochranMantel-Haenszel test for categorical variables, 2-way
STUDY ENDPOINTS. The primary (adjudicated) study
analysis of variance for continuous variables).
endpoints were death, cardiac death, myocardial
A Cox proportional hazards model associated
infarction (MI), target lesion revascularization (TLR),
smoking with 5-year clinical endpoints adjusting for
T A B L E 1 Characteristics of Multicenter Randomized Trials Included in the Patient-Level Pooled Analysis
Trial (Ref. #)
RAVEL (14) E-SIRUS (15)
n
Study Cohort
Comparison
Randomization
Antithrombotic
238
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
Primary Endpoint
In-stent late luminal loss
352
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
In-stent minimum lumen diameter
1,058
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
Cardiac death/MI/TVR
C-SIRUS (17)
100
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
In-stent minimum lumen diameter
TAXUS IV (18)
1,314
Stable/unstable angina, SI
PES vs. BMS
1:1
UFH
Ischemia-driven TVR
TAXUS V (19)
1,156
Stable/unstable angina, SI
PES vs. BMS
1:1
UFH
Ischemia-driven TVR
ENDEAVOR II (20)
1,197
Stable/unstable angina, SI
ZES vs. BMS
1:1
UFH
Cardiac death/MI/TVR
ENDEAVOR III (21)
436
Stable/unstable angina
ZES vs. SES
3:1
UFH
SPIRIT II (22)
300
Stable/unstable angina, SI
EES vs. PES
3:1
UFH, BIV
SPIRIT III (23)
1,002
Stable/unstable angina, SI
EES vs. PES
2:1
UFH, BIV
HORIZONS-AMI (24)
3,006
STEMI
PES vs. BMS
3:1
UFHþGPI, BIV
COMPARE (25)
1,800
All-comers
EES vs. PES
1:1
UFH
ENDEAVOR IV (26)
1,548
Stable/unstable angina, SI
ZES vs. PES
1:1
UFH, BIV
Cardiac death/MI/TVR
SPIRIT IV (27)
3,687
Stable/unstable angina, SI
EES vs. PES
2:1
UFH, BIV
Cardiac death/target vessel MI, or TVR
TWENTE (28)
1,391
All-comers*
ZES vs. CoCr-EES
1:1
UFH
Cardiac death/target vessel MI/TVR
COMPARE II (29)
2,707
All-comers
BES vs. EES
2:1
UFH
Cardiac death/nonfatal MI, TVR
PLATINUM (30)
1,530
Stable/unstable angina, SI
PtCr-EES vs. CoCr-EES
1:1
UFH, LMWH, BIV
TWENTE II (31)
2,371
All-comers
CoCr-ZES vs. PtCr-EES
1:1
UFH
SIRIUS (16)
In-segment late lumen loss In-stent late luminal loss In-segment late lumen loss TVR and MACE (death/MI/stroke/ST) Death/MI/TVR
Cardiac death†/MI/†TVR Cardiac death/target vessel MI/TVR
*Except STEMI within 48 h. †All events related to target vessel. BIV ¼ bivalirudin; BMS ¼ bare-metal stent; BES ¼ biolimus-eluting stent; CoCr-EES ¼ cobalt-chromium everolimus-eluting stent; EES ¼ everolimus-eluting stent; GPI ¼ glycoprotein IIb/IIIa inhibitor; LMWH ¼ low-molecular-weight heparin; NSTE-ACS ¼ non–ST-segment elevation acute coronary syndrome; PES ¼ paclitaxel-eluting stent; PtCr-EES ¼ platinum-chromium everolimus-eluting stent; SES ¼ sirolimus-eluting stent; SI ¼ silent ischemia; ST ¼ stent thrombosis; STEMI ¼ ST-segment elevation myocardial infarction; TLR ¼ target lesion failure; TVR ¼ target vessel revascularization; UFH ¼ unfractionated heparin; ZES ¼ zotarolimus-eluting stent.
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
T A B L E 2 Baseline and Procedural Characteristics Stratified by Smoking Status
Smokers (n ¼ 6,722)
Nonsmokers (n ¼ 17,632)
p Value
56.5 9.9
65.1 10.3
<0.0001
5,056/6,722 (75.2)
12,426/17,632 (70.5)
<0.0001
28.2 5.2
29.2 5.4
<0.0001
1,183/6,719 (17.6)
4,605/17,623 (26.1)
<0.0001
Age, yrs Male Body mass index, kg/m2 Diabetes mellitus
330/6,719 (4.9)
1,337/17,623 (7.6)
<0.0001
3,629/6,715 (54.0)
11,942/17,610 (67.8)
<0.0001 <0.0001
Insulin treated Hypertension Hyperlipidemia
3,852/6,675 (57.7)
11,694/17,507 (66.8)
Prior myocardial infarction
1,499/6,683 (22.4)
3,976/17,483 (22.7)
0.60
Prior percutaneous coronary intervention
1,009/6,690 (15.1)
3,707/17,574 (21.1)
<0.0001
203/6,720 (3.0)
1,477/17,630 (8.4)
<0.0001
4,226/6,283 (67.3)
7,678/16,163 (47.5)
<0.0001
2,062/6,722 (30.7)
2,323/17,630 (13.2)
<0.0001
670/6,722 (10.0)
1,333/17,630 (7.6)
<0.0001
Unstable angina
1,494/6,283 (23.8)
4,022/16,163 (24.9)
0.08
Stable coronary artery disease
2,057/6,283 (32.7)
8,485/16,163 (52.5)
<0.0001
1,895/6,283 (30.2)
7,798/16,163 (48.2)
<0.0001
162/6,283 (2.6)
687/16,163 (4.3)
<0.0001
Prior coronary artery bypass grafting Acute coronary syndromes STEMI NSTE-ACS
Stable angina SI Lesion length, mm
17.4 10.8
16.9 11.2
0.006
ACC type C lesion
2,803/6,685 (41.9)
6,237/17,559 (35.5)
<0.0001
1,718/6,343 (27.1)
5,281/16,508 (32.0)
<0.0001
197/3,370 (5.8)
669/8,099 (8.3)
<0.0001
1,118/6,692 (16.7)
1,551/17,587 (8.8)
<0.0001
30.8 21.1
30.5 21.4
BMS
1,033/6,722 (15.4)
2,379/17,632 (13.5)
Drug-eluting stent
5,689/6,722 (84.6)
15,253/17,632 (86.5)
0.0002
First generation
2,370/6,722 (35.3)
5,331/17,632 (30.2)
<0.0001
Second generation
3,319/6,722 (49.4)
9,922/17,632 (56.3)
<0.0001
Calcification (moderate/severe) Tortuosity (moderate/severe) Any occlusions Total stent length
0.25 0.0002
Pre-procedure Reference vessel diameter, mm
2.83 0.86
2.73 0.58
<0.0001
Minimum lumen diameter, mm
0.70 0.49
0.76 0.44
<0.0001
Diameter stenosis, %
76.3 17.0
73.2 15.6
<0.0001
TIMI flow grade 0/1
1,533/6,689 (22.9)
2,333/17,574 (13.3)
<0.0001
Post-procedure Diameter stenosis, %
16.8 11.1
16.8 10.9
0.93
Minimum lumen diameter, mm
2.35 0.92
2.28 0.75
<0.0001
31/6,613 (0.5)
77/17,407 (0.4)
57.1 11.3
57.8 10.9
TIMI flow grade 0/1 Left ventricular ejection fraction, %
0.78 0.005
Values are mean SD or n/N (%). ACC ¼ American College of Cardiology; TIMI ¼ Thrombolysis In Myocardial Infarction; other abbreviations as in Table 1.
age (10-year increments), sex, diabetes mellitus,
performed that accounted for possible confounders
insulin-treated diabetes, hypertension, hyperlipid-
by trial; these include the Cox proportional hazards
emia, prior PCI, prior coronary artery bypass grafting
model: 1) not including the study; 2) stratified by
(CABG) surgery, prior MI, acute coronary syndrome
study; 3) including study as a random effect; and 4)
(ACS) versus stable CAD, any moderate or severe
including the study as a covariate.
calcification, lesion length, stent generation, any occlusion, any American College of Cardiology class C
RESULTS
lesion, and study (as a random effect). All p values are 2-sided, and a p value < 0.05 was considered statis-
BASELINE PATIENT AND ANGIOGRAPHIC CHARAC-
tically
TERISTICS. Smoking status was available in 24,354 of
significant.
Sensitivity
analyses
were
1943
1944
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
American College of Cardiology type C lesions, as
T A B L E 3 Outcomes at 1 and 5 Years in Patients Stratified by
well as any occlusions, but were less likely to have
Smoking Status
moderate or severe calcification and tortuosity. Smokers (n ¼ 6,722)
Nonsmokers (n ¼ 17,632)
p Value
Death
111 (1.7)
290 (1.7)
0.95
to have pre-procedural Thrombolysis In Myocardial
Cardiac death
67 (1.0)
178 (1.0)
0.94
Infarction flow grade 0 or 1, but there was no sig-
Myocardial infarction
226 (3.4)
529 (3.0)
0.15
nificant difference in post-procedure TIMI flow
Target lesion revascularization
321 (5.8)
902 (6.5)
0.06
grade 0 or 1 between the 2 groups. On quantitative
<0.0001
Nevertheless, the total stent length was similar between the 2 groups. Smokers were also more likely
1-yr outcomes
Stent thrombosis (definite/probable) Target lesion failure*
117 (1.8)
148 (0.8)
282 (6.0)
1,042 (7.2)
430 196 450 589 211
1,251 583 892 1,446 275
0.006
5-yr outcomes Death Cardiac death Myocardial infarction Target lesion revascularization Stent thrombosis (definite/ probable) Target lesion failure*
(8.6) (3.9) (7.8) (12.0) (3.5)
(9.0) (4.1) (5.6) (11.4) (1.8)
0.59 0.44 <0.0001 0.45 <0.0001
coronary angiography, smokers were more likely to have a lower pre-procedure minimal lumen diameter and a larger post-procedure minimal lumen diameter. CUMULATIVE 1-YEAR AND 5-YEAR OUTCOMES. Median
follow-up was 3.9 (interquartile range: 3.0 to 5.0) 601 (14.8)
1,892 (14.4)
0.74
Values are n (%) based on Kaplan-Meier estimates. *Cardiac death, target vessel myocardial infarction, and ischemic target lesion revascularization.
years. Patients lost to follow-up were censored at the last known follow-up date. At 1 year post-PCI, smokers compared with nonsmokers had a lower rate of TLF (6.0% vs. 7.2%; p ¼ 0.006), and a trend toward less TLR (5.8% vs.
24,633 (98.9%) patients enrolled in the 18 randomized trials, 6,722 (27.6%) of whom were current smokers. Table 2 summarizes baseline patient clinical characteristics and angiographic findings stratified by smoking
status.
Compared
with
nonsmokers,
smokers were more likely to be men, were younger (by w9 years), and were less likely to have other cardiac risk factors including diabetes mellitus, hypertension, and hyperlipidemia. Smokers were also less likely to have prior PCI or prior CABG surgery.
6.5%; p ¼ 0.06), but a greater rate of definite or probable ST (1.8% vs. 0.8%; p < 0.0001). There were no significant differences in the rates of death, cardiac death, or MI between smokers and nonsmokers at 1 year (Table 3). At 5 years post-PCI, smokers had significantly higher rates of MI (7.8% vs. 5.6%; p < 0.0001) and definite or probable ST (3.5% vs. 1.8%; p < 0.0001); however, there were no significant differences in the rates of death, cardiac death, TLR, or TLF (Table 3).
Baseline characteristics stratified by study are pre-
MULTIVARIABLE ANALYSIS. Table 4 and the Central
sented in Online Table 1.
Illustration present the risk of adverse outcomes ac-
By angiographic core laboratory analysis smokers were more likely to have longer lesions and more
cording to smoking status before and after adjustment
for
potential
confounders.
Notably,
the
T A B L E 4 Unadjusted and Adjusted HRs for Adverse Events at 5 Years in Patients Stratified by Smoking Status
Death
Smokers (n ¼ 6,722)
Nonsmokers (n ¼ 17,632)
Unadjusted HR (95% CI)
430 (8.6)
1,251 (9.0)
0.97 (0.87–1.08)
p Value
Adjusted HR (95% CI)†
p Value
0.59
1.86 (1.63–2.12)
<0.0001
Cardiac death
196 (3.9)
583 (4.1)
0.94 (0.80–1.10)
0.44
1.68 (1.38–2.05)
<0.0001
Myocardial infarction
450 (7.8)
892 (5.6)
1.37 (1.22–1.53)
<0.0001
1.38 (1.20–1.58)
<0.0001
Target lesion revascularization
589 (12.0)
1,446 (11.4)
1.04 (0.94–1.14)
0.45
1.06 (0.94–1.19)
211 (3.5)
275 (1.8)
2.09 (1.75–2.50)
<0.0001
1.60 (1.28–1.99)
<0.0001
601 (14.8)
1,892 (14.4)
0.98 (0.90–1.08)
0.74
1.05 (1.01–1.10)
0.005
Stent thrombosis (definite/ probable) Target lesion failure*
0.33
Values are n (%) based on Kaplan-Meier estimates. *Cardiac death, target vessel myocardial infarction, and ischemic target lesion revascularization. †Univariate variables included for multivariable analysis were current smoker, age (10-yr increments), sex, diabetes, insulin-treated diabetes, hypertension, hyperlipidemia, previous percutaneous coronary intervention, previous coronary artery bypass grafting, previous myocardial infarction, acute coronary syndromes vs. stable coronary artery disease, any moderate/ severe calcification, lesion length, bare metal stents vs. second-generation drug-eluting stents, first vs. second-generation drug-eluting stents, any occlusion, any American College of Cardiology class C lesion. CI ¼ confidence interval; HR ¼ Hazard ratio.
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
C ENTR AL I LL U STRA T I O N Forest Plots Showing Unadjusted and Adjusted HRs for Adverse Events at 5 Years in Patients According to Smoking Status
Unadjusted vs. Adjusted Hazard Ratio (95% Confidence Interval) Death
Cardiac Death
0.97 (0.87–1.08), p = 0.59 1.86 (1.63–2.12), p < 0.0001 0.94 (0.80–1.10), p = 0.44 1.68 (1.38–2.05), p < 0.0001
Myocardial Infarction
1.37 (1.22–1.53), p < 0.0001
Target Lesion Revascularization
1.04 (0.94–1.14), p = 0.45 1.06 (0.94–1.19), p = 0.33
Definite/Probable Stent Thrombosis
2.09 (1.75–2.50), p < 0.0001 1.60 (1.28–1.99), p < 0.0001
Target Lesion Failure
1.38 (1.20–1.58), p < 0.0001
0.98 (0.90–1.08), p = 0.74 1.17 (1.05–1.30), p = 0.005 0.50
0.75
1.00
Risk Decreases
1.50
2.00
2.50 3.00
Risk Increases
Unadjusted HR
Adjusted HR
Yadav, M. et al. J Am Coll Cardiol Intv. 2019;12(19):1941–50.
The true risk among smokers after percutaneous coronary intervention (PCI) emerges only after adjustment of risk factors as evidenced by the marked changes in hazard ratios for death, cardiac death, and target lesion failure. These findings likely underlie the smoker’s paradox often reported in smokers in the published reports. HR ¼ hazard ratio.
association between smoking and the occurrence of
smoking status and hypertension (p int ¼ 0.02),
death, cardiac death, TLR, and TLF were neutral
hyperlipidemia (p int ¼ 0.01), and ACS (pint ¼ 0.0002)
before adjustment, but after adjustment, smoking
for death (Figure 1), but no interactions were present
was strongly and significantly associated with higher
for definite or probable ST (Online Figure 1). The
5-year rates of death, cardiac death, and TLF. Smok-
hazard ratio for death among smokers with ACS and
ing continued to be associated with higher rates of MI
stable CAD were 1.61 (95% confidence interval: 1.37 to
and definite or probable ST after adjustment for po-
1.94) and a 2.19 (95% confidence interval: 1.82 to 2.65)
tential confounders. In sensitivity analysis, there
respectively (p int ¼ 0.02), after adjusting for cova-
were no significant differences in outcomes between
riates (Online Figure 2).
the 4 test methodologies (Online Table 2). Table 5 shows the independent predictors of death
DISCUSSION
and other adverse ischemic events. Smoking was an independent predictor of death, cardiac death, MI,
The principal findings of the present pooled analysis
definite or probable ST, and TLF. In subgroup anal-
of IPD from 18 large-scale, multicenter, randomized
ysis, significant interactions were present between
controlled stent trials are as follows: 1) smokers
1945
1946
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
smoking was a strong and independent predictor of
T A B L E 5 Independent Predictors of Adverse Outcomes at 5 Years
death, cardiac death, MI, definite or probable ST, and
HR (95% CI)
p Value
TLF. Given decades of evidence related to the detri-
Death Current smoker
1.86 (1.63–2.12)
<0.0001
mental effect of cigarette smoking on health in general (33,34), the role of smoking in relation to
Age (10-yr increase)
2.19 (2.06–2.32)
<0.0001
Insulin-treated diabetes
1.68 (1.40–2.02)
<0.0001
Previous coronary artery bypass grafting
1.45 (1.23–1.71)
<0.0001
Diabetes
1.41 (1.23–1.61)
<0.0001
Previous myocardial infarction
1.32 (1.17–1.49)
<0.0001
Male
1.24 (1.10–1.39)
0.0003
First- vs. second-generation drug-eluting stent
1.16 (1.03–1.32)
0.02
Hyperlipidemia
cardiovascular outcomes has been controversial at best since the first description of the so-called smoker’s paradox (4,5). Some studies have supported its existence (i.e., smoking improves cardiovascular outcomes in the context of an acute event such as an MI) (3–9), whereas others have refuted its
0.82 (0.73–0.92)
0.0007
Current smoker
1.68 (1.38–2.05)
<0.0001
Age (10-yr increase)
1.89 (1.73–2.06)
<0.0001
these studies were registry analyses or were con-
Diabetes
1.81 (1.49–2.19)
<0.0001
ducted in the era of thrombolytic therapy and balloon
Insulin-treated diabetes
1.74 (1.35–2.24)
<0.0001
angioplasty; this limits their applicability to modern
Prior coronary artery bypass grafting
1.62 (1.28–2.04)
<0.0001
PCI. To the best of our knowledge, our study is by far
Any occlusion
1.53 (1.17–2.02)
0.002
the largest patient-level pooled analysis of random-
Prior myocardial infarction
1.42 (1.19–1.70)
0.0001
ized controlled trials with blinded core laboratory
Any moderate/severe calcification
1.38 (1.17–1.62)
0.0001
analyses and independent event adjudication inves-
Hyperlipidemia
0.76 (0.64–0.90)
0.001
tigating
existence (i.e., smoking is not associated with better
Cardiac death
Myocardial infarction
cardiovascular outcomes) (10–12). Moreover, many of
the
smoker’s
paradox
concept
in
the
contemporary stent era.
Current smoker
1.38 (1.20–1.58)
<0.0001
First- vs. second-generation drug-eluting stent
1.75 (1.47–2.07)
<0.0001
Insulin-treated diabetes
1.58 (1.26–2.00)
BMS vs. second-generation drug-eluting stents
1.54 (1.24–1.91)
<0.0001
Prior myocardial infarction
1.39 (1.21–1.60)
<0.0001
0.0001
Prior percutaneous coronary intervention
1.31 (1.13–1.51)
0.0004
Previous coronary artery bypass grafting
1.27 (1.03–1.56)
0.02
Hypertension
1.23 (1.07–1.40)
0.003
Any moderate/severe calcification
1.19 (1.05–1.36)
0.007
Smokers
undergoing
PCI
were
significantly
younger, were predominantly men, and had a lower prevalence of known CAD risk factors including diabetes
mellitus,
hypertension,
hyperlipidemia,
prior PCI, or prior CABG surgery. These baseline differences were also noted in most other largescale studies including SYNTAX (Synergy between PCI
with
Taxus
and
Cardiac
Surgery)
trial
Any ACC class C lesion
1.18 (1.02–1.37)
0.03
(3,4,10,13,35). This finding underlines the fact that
Lesion length
1.01 (1.01–1.02)
<0.0001
smokers are prone to more rapidly develop CAD or present
ST (definite/probable)
with
plaque
rupture
and
thrombosis,
Current smoker
1.60 (1.28–1.99)
<0.0001
necessitating PCI almost a decade earlier compared
First- vs. second-generation drug-eluting stent
1.97 (1.45–2.67)
<0.0001
with nonsmokers, even with a lower prevalence of
Insulin-treated diabetes
1.73 (1.17–2.57)
0.006
coronary risk factors (10). After proper adjustment
BMS vs. second-generation drug-eluting stents
1.65 (1.12–2.42)
0.01
for potentially relevant confounders, smokers had
Prior myocardial infarction
1.52 (1.19–1.94)
0.0007
an w95% increased risk of death, w85% increased
Prior percutaneous coronary intervention
1.35 (1.05–1.75)
0.02
risk of cardiac death, and w70% increased risk of
Any moderate or severe calcification
1.31 (1.07–1.62)
0.01
definite or probable ST at 5 years after coronary
Lesion length
1.01 (1.01–1.02)
0.0005
0.87 (0.79–0.96)
0.006
Age (10-yr increase)
Continued on the next page
stent
implantation
compared
with
nonsmokers.
Thus, smoking results not only earlier presentation with CAD, but also a poor prognosis after contemporary
percutaneous
revascularization
therapy,
emphasizing the long-term deleterious effects of undergoing PCI were significantly younger and had
smoking.
less coronary risk factors compared with nonsmokers;
Potential mechanisms to explain the increase in
2) at 5 years post-PCI, smokers had similar unadjusted
mortality and adverse ischemic events after stenting
rates of death and cardiac death but higher rates of MI
in patients who are active smokers is likely multi-
and ST compared with nonsmokers; and 3) after
factorial. Smokers have increased platelet aggrega-
multivariable adjustment for potential confounders,
tion (36,37), increased fibrinogen, and decreased
OCTOBER 14, 2019:1941–50
fibrinolytic activity compared with nonsmokers (37), creating a state of hypercoagulability that pre-
Smoker’s Paradox
T A B L E 5 Continued
HR (95% CI)
disposes to acute thrombosis (38). Moreover, in patients with obstructive CAD, smoking may cause
1947
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019
p Value
Target lesion revascularization
coronary vasoconstriction (39). Smoking also in-
Current smoker
1.06 (0.94–1.19)
creases myocardial workload by increasing heart rate,
BMS vs. second-generation drug-eluting stents
2.59 (2.20–3.05)
<0.0001
myocardial contractility, and blood pressure (38). The
Prior coronary artery bypass grafting
1.44 (1.21–1.71)
<0.0001
higher carboxyhemoglobin level in smokers also de-
First- vs. second-generation drug-eluting stent
1.32 (1.14–1.52)
0.0001
Prior percutaneous coronary intervention
1.31 (1.16–1.48)
<0.0001
Diabetes
1.22 (1.08–1.39)
0.001
Insulin-treated diabetes
1.21 (1.00–1.46)
0.05
creases blood oxygenation. Smoking induces catecholamine surge with cardiac ischemia, which may be a cause of cardiac arrhythmias and sudden cardiac death (40). Moreover, smoking decreases highdensity lipoproteins and increases oxidized lowdensity lipoproteins (41,42), induces endothelial dysfunction (43) and neutrophil activation (38), causes oxidant injury (43), increases fibrinogen levels,
0.33
Hypertension
1.11 (1.00–1.24)
0.05
Lesion length
1.01 (1.01–1.02)
<0.0001
Age (10-yr increase)
0.91 (0.87–0.96)
0.0002
Male
0.89 (0.80–1.00)
0.04
Prior myocardial infarction
0.88 (0.78–1.00)
0.04
TLF*
and causes platelet activation (37), all of which in-
Current smoker
1.17 (1.05–1.30)
0.005
crease the rate of atherosclerosis and plaque pro-
BMS vs. second-generation drug-eluting stents
2.01 (1.72–2.35)
<0.0001
gression by direct or indirect effects (38).
Previous coronary artery bypass grafting
1.62 (1.42–1.85)
<0.0001
First- vs. second-generation drug-eluting stent
1.39 (1.23–1.57)
<0.0001 <0.0001
STUDY STRENGTHS AND LIMITATIONS. The present
Insulin-treated diabetes
1.38 (1.18–1.62)
analysis has several strengths. IPD pooling of more
Diabetes
1.23 (1.10–1.37)
0.0003
than 24,000 patients from 18 randomized trials, each
Any moderate or severe calcification
1.19 (1.08–1.30)
0.0004
of which was monitored and had careful event adju-
Any ACC class C lesion
1.19 (1.07–1.32)
0.002
Hypertension
1.13 (1.02–1.25)
0.02
Age (10-yr increase)
1.05 (1.01–1.10)
0.02
Lesion length
1.01 (1.00–1.01)
<0.0001
dication, helped to identify differences that were hard to ascertain in previous underpowered studies. A recent post hoc analysis of the SYNTAX trial at 5 years also showed similar results. Smokers had worse clinical outcomes due to a higher incidence of recurrent MI in both the PCI and CABG arms, and smoking was an independent predictor of the composite endpoint of death or MI or stroke and the
Variables included for multivariable analysis were current smoker, age (10-yr increments), sex, diabetes, insulintreated diabetes, hypertension, hyperlipidemia, prior percutaneous coronary intervention, prior coronary artery bypass grafting, prior myocardial infarction, acute coronary syndrome vs. stable coronary artery disease, any moderate/severe calcification, lesion length, BMS vs. second-generation drug-eluting stents, first- vs. secondgeneration drug-eluting stents, any occlusion, any ACC class C lesion. *Includes cardiac death, target vessel myocardial infarction, and ischemic target lesion revascularization. Abbreviations as in Tables 1 and 2.
composite endpoint of death or MI or stroke or target vessel revascularization (13). The IPD approach also allows for multivariable analysis to adjust for baseline imbalances, and subgroup analyses. There were
the number of packs of cigarettes smoked, we
no significant interactions between smoking status
could not assess a dose-dependent relationship to
and numerous variables, except for hypertension,
outcomes. Definitions of MI were slightly different
hyperlipidemia, and ACS at 5 years for death as an
across trials, potentially reducing precision. Several
outcome, implying that smoking with hypertension
angiographic
or with ACS combined have a particularly poor long-
small vessels, long lesions, and overlapping stents
term prognosis.
were not systematically collected in all randomized
As in most previous studies of this subject, smoking status was defined at baseline; data on smoking status during follow-up were not available. We cannot
and
procedural
variables
such
as
trials. Similarly, data on access site, left ventricular function,
and
medication
adherence
were
not
available.
therefore evaluate the impact of smoking crossovers after treatment (from nonsmokers or smokers, or vice versa), although the deleterious impact of smoking
CONCLUSIONS
identified in the present report may have been even greater were we able to account for such effects. In
Results from the present large-scale analysis largely
addition, we were unable to differentiate ex-smokers
dismisses a smoker’s paradox after PCI with coronary
from nonsmokers. Also, owing to the lack of data on
stent
implantation.
Even
before
multivariable
1948
Yadav et al. Smoker’s Paradox
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
F I G U R E 1 Forest Plots Showing Subgroup Analysis for Death According to Smoking Status
Subgroup analysis shows significant interaction of smoking status with hypertension, hyperlipidemia and acute coronary syndrome. ACC ¼ American College of Cardiology; BMS ¼ bare-metal stent; CABG ¼ coronary artery bypass graft; CI ¼ confidence interval; DES ¼ drug-eluting stent; KM ¼ Kaplan-Meier; MI ¼ myocardial infarction; PCI ¼ percutaneous coronary intervention.
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
adjustment, active smokers had higher 5-year rates of MI and ST after stenting. When baseline imbalances were also taken into account, smoking was also associated with increased long-term rates of cardiac and all-cause mortality. Given the doubleedged sword of earlier presentation by nearly a
PERSPECTIVES WHAT IS KNOWN? Some studies have shown that smokers have better outcomes after PCI, a phenomenon widely known as the smoker’s paradox.
decade and increased long-term adverse outcomes
WHAT IS NEW? The smoker’s paradox may be explained by the
after PCI, the strongest possible individual and
younger age and lower prevalence of risk factors among
societal-based efforts should be undertaken to pre-
smokers; these trends reverse after adjustment of risk factors
vent smoking and strongly encourage its cessation
such that early and late outcomes are substantially worse in
once begun.
smokers after PCI.
ADDRESS FOR CORRESPONDENCE: Dr. Gregg W.
Stone, Columbia University Medical Center, Cardiovascular Research Foundation, 1700 Broadway, 8th Floor, New York, New York 10019. E-mail: gs2184@
WHAT IS NEXT? Further randomized studies are needed to study the impact of smoking on heart disease in general and specifically after PCI. Greater emphasis on smoking cessation strategies is warranted in patients with CAD.
columbia.edu. Twitter: @greggwstone.
REFERENCES 1. Ockene IS, Miller NH. Cigarette smoking, cardiovascular disease, and stroke: a statement for health care professionals from the American Heart Association. American Heart Association Task Force on Risk Reduction. Circulation 1997;96: 3243–7. 2. Health Quality Ontario. Intravascular ultrasound to guide percutaneous coronary interventions: an evidence-based analysis. Ont Health Technol Assess Series 2006;6:1–97. 3. Andrikopoulos GK, Richter DJ, Dilaveris PE, et al. In-hospital mortality of habitual cigarette smokers after acute myocardial infarction; the “smoker’s paradox” in a countrywide study. Eur Heart J 2001;22:776–84. 4. Barbash GI, Reiner J, White HD, et al. Evaluation of paradoxic beneficial effects of smoking in patients receiving thrombolytic therapy for acute myocardial infarction: mechanism of the “smoker’s paradox” from the GUSTO-I trial, with angiographic insights. Global Utilization of Streptokinase and Tissue-Plasminogen Activator for Occluded Coronary Arteries. J Am Coll Cardiol 1995;26:1222–9. 5. Barbash GI, White HD, Modan M, et al. Significance of smoking in patients receiving thrombolytic therapy for acute myocardial infarction. Experience gleaned from the International Tissue Plasminogen Activator/Streptokinase Mortality Trial. Circulation 1993;87:53–8. 6. Molstad P. First myocardial infarction in smokers. Eur Heart J 1991;12:753–9. 7. Gourlay SG, Rundle AC, Barron HV. Smoking and mortality following acute myocardial infarction: results from the National Registry of Myocardial Infarction 2 (NRMI 2). Nicotine Tob Res 2002;4: 101–7. 8. Ruiz-Bailen M, de Hoyos EA, Reina-Toral A, Torres-Ruiz JM, Alvarez-Bueno M, Gomez Jimenez FJ. Paradoxical effect of smoking in the Spanish population with acute myocardial
infarction or unstable angina: results of the ARIAM Register. Chest 2004;125:831–40.
lesions in small native coronary arteries (C-SIRIUS). J Am Coll Cardiol 2004;43:1110–5.
9. Elosua R, Vega G, Rohlfs I, et al. Smoking and myocardial infarction case-fatality: hospital and population approach. Eur J Cardiovasc Prev Rehabil 2007;14:561–7.
18. Stone GW, Ellis SG, Cox DA, et al. A polymerbased, paclitaxel-eluting stent in patients with coronary artery disease. N Engl J Med 2004;350: 221–31.
10. Hasdai D, Garratt KN, Grill DE, Lerman A, Holmes DR Jr. Effect of smoking status on the
19. Stone GW, Ellis SG, Cannon L, et al. Compari-
long-term outcome after successful percutaneous coronary revascularization. N Engl J Med 1997; 336:755–61.
son of a polymer-based paclitaxel-eluting stent with a bare metal stent in patients with complex coronary artery disease: a randomized controlled trial. JAMA 2005;294:1215–23.
11. Himbert D, Klutman M, Steg G, White K, Gulba DC. Cigarette smoking and acute coronary syndromes: a multinational observational study. Int J Cardiol 2005;100:109–17.
20. Fajadet J, Wijns W, Laarman GJ, et al. Randomized, double-blind, multicenter study of the Endeavor zotarolimus-eluting phosphorylcholineencapsulated stent for treatment of native
12. Berger JS, Bhatt DL, Steinhubl SR, et al. Smoking, clopidogrel, and mortality in patients with established cardiovascular disease. Circulation 2009;120:2337–44. 13. Zhang YJ, Iqbal J, van Klaveren D, et al. Smoking is associated with adverse clinical outcomes in patients undergoing revascularization with PCI or CABG: the SYNTAX trial at 5-year follow-up. J Am Coll Cardiol 2015;65:1107–15. 14. Morice MC, Serruys PW, Sousa JE, et al. A randomized comparison of a sirolimus-eluting stent with a standard stent for coronary revascularization. N Engl J Med 2002;346:1773–80. 15. Schofer J, Schluter M, Gershlick AH, et al. Sirolimus-eluting stents for treatment of patients with long atherosclerotic lesions in small coronary arteries: double-blind, randomised controlled trial (E-SIRIUS). Lancet 2003;362:1093–9. 16. Moses JW, Leon MB, Popma JJ, et al. Sirolimus-eluting stents versus standard stents in patients with stenosis in a native coronary artery. N Engl J Med 2003;349:1315–23. 17. Schampaert E, Cohen EA, Schluter M, et al. The Canadian study of the sirolimus-eluting stent in the treatment of patients with long de novo
coronary artery lesions: clinical and angiographic results of the ENDEAVOR II trial. Circulation 2006; 114:798–806. 21. Kandzari DE, Leon MB, Popma JJ, et al. Comparison of zotarolimus-eluting and sirolimuseluting stents in patients with native coronary artery disease: a randomized controlled trial. J Am Coll Cardiol 2006;48:2440–7. 22. Serruys PW, Ruygrok P, Neuzner J, et al. A randomised comparison of an everolimuseluting coronary stent with a paclitaxel-eluting coronary stent:the SPIRIT II trial. EuroIntervention 2006;2:286–94. 23. Stone GW, Midei M, Newman W, et al. Comparison of an everolimus-eluting stent and a paclitaxel-eluting stent in patients with coronary artery disease: a randomized trial. JAMA 2008; 299:1903–13. 24. Stone GW, Lansky AJ, Pocock SJ, et al. Paclitaxel-eluting stents versus bare-metal stents in acute myocardial infarction. N Engl J Med 2009; 360:1946–59. 25. Kedhi E, Joesoef KS, McFadden E, et al. Second-generation everolimus-eluting and paclitaxeleluting stents in real-life practice (COMPARE): a randomised trial. Lancet 2010;375:201–9.
1949
1950
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
26. Leon MB, Mauri L, Popma JJ, et al. A randomized comparison of the Endeavor zotarolimus-eluting stent versus the TAXUS paclitaxel-eluting stent in de novo native coronary lesions 12-month outcomes from the ENDEAVOR IV trial. J Am Coll Cardiol 2010;55:543–54. 27. Stone GW, Rizvi A, Newman W, et al. Everolimus-eluting versus paclitaxel-eluting stents in coronary artery disease. N Engl J Med 2010;362: 1663–74. 28. von Birgelen C, Basalus MW, Tandjung K, et al. A randomized controlled trial in secondgeneration zotarolimus-eluting Resolute stents versus everolimus-eluting Xience V stents in realworld patients: the TWENTE trial. J Am Coll Cardiol 2012;59:1350–61. 29. Smits PC, Hofma S, Togni M, et al. Abluminal biodegradable polymer biolimus-eluting stent versus durable polymer everolimus-eluting stent (COMPARE II): a randomised, controlled, noninferiority trial. Lancet 2013;381:651–60. 30. Meredith IT, Teirstein PS, Bouchard A, et al. Three-year results comparing platinum-chromium PROMUS element and cobalt-chromium XIENCE
PEERS): a randomised, single-blind, multicentre, non-inferiority trial. Lancet 2014;383:413–23. 32. Cutlip DE, Windecker S, Mehran R, et al. Clinical end points in coronary stent trials: a case for standardized definitions. Circulation 2007;115: 2344–51. 33. Levitz JS, Bradley TP, Golden AL. Overview of smoking and all cancers. Med Clin North Am 2004; 88:1655–75, xiii. 34. Bartecchi CE, MacKenzie TD, Schrier RW. The human costs of tobacco use (1). N Engl J Med 1994;330:907–12. 35. Weisz G, Cox DA, Garcia E, et al. Impact of smoking status on outcomes of primary coronary intervention for acute myocardial infarction–the smoker’s paradox revisited. Am Heart J 2005;150: 358–64. 36. Levine PH. An acute effect of cigarette smoking on platelet function. A possible link between smoking and arterial thrombosis. Circulation 1973;48:619–23.
V everolimus-eluting stents in de novo coronary artery narrowing (from the PLATINUM Trial). Am J Cardiol 2014;113:1117–23.
37. Benowitz NL, Fitzgerald GA, Wilson M, Zhang Q. Nicotine effects on eicosanoid formation and hemostatic function: comparison of transdermal nicotine and cigarette smoking. J Am Coll Cardiol 1993;22:1159–67.
31. von Birgelen C, Sen H, Lam MK, et al. Thirdgeneration zotarolimus-eluting and everolimuseluting stents in all-comer patients requiring a percutaneous coronary intervention (DUTCH
38. Benowitz NL, Gourlay SG. Cardiovascular toxicity of nicotine: implications for nicotine replacement therapy. J Am Coll Cardiol 1997;29: 1422–31.
39. Nicod P, Rehr R, Winniford MD, Campbell WB, Firth BG, Hillis LD. Acute systemic and coronary hemodynamic and serologic responses to cigarette smoking in long-term smokers with atherosclerotic coronary artery disease. J Am Coll Cardiol 1984;4:964–71. 40. Escobedo LG, Zack MM. Comparison of sudden and nonsudden coronary deaths in the United States. Circulation 1996;93:2033–6. 41. Craig WY, Palomaki GE, Haddow JE. Cigarette smoking and serum lipid and lipoprotein concentrations: an analysis of published data. BMJ 1989; 298:784–8. 42. Harats D, Ben-Naim M, Dabach Y, Hollander G, Stein O, Stein Y. Cigarette smoking renders LDL susceptible to peroxidative modification and enhanced metabolism by macrophages. Atherosclerosis 1989;79:245–52. 43. Pittilo RM. Cigarette smoking, endothelial injury and cardiovascular disease. Int J Exp Pathol 2000;81:219–30.
KEY WORDS coronary artery disease, coronary artery disease outcomes, smoker’s paradox, smokers
A PPE NDI X For supplemental tables and figures, please see the online version of this paper.
VOL. 12, NO. 19, 2019
ª 2019 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION PUBLISHED BY ELSEVIER
The Smoker’s Paradox Revisited A Patient-Level Pooled Analysis of 18 Randomized Controlled Trials Mayank Yadav, MD,a Gary S. Mintz, MD,b Philippe Généreux, MD,b,c,d Mengdan Liu, MS,b Thomas McAndrew, PHD,b Björn Redfors, MD, PHD,b Mahesh V. Madhavan, MD,e Martin B. Leon, MD,b,e Gregg W. Stone, MDb,e
ABSTRACT OBJECTIVES This study examined the smoker’s paradox using patient-level data from 18 prospective, randomized trials of patients undergoing percutaneous coronary intervention (PCI) with stent implantation. BACKGROUND Studies on the effects of smoking and outcomes among patients undergoing PCI have reported conflicting results. METHODS Data from the RAVEL, E-SIRIUS, SIRIUS, C-SIRIUS, TAXUS IV and V, ENDEAVOR II to IV, SPIRIT II to IV, HORIZONS-AMI, COMPARE I and II, PLATINUM, and TWENTE I and II randomized trials were pooled. Patients were stratified by smoking status at time of enrollment. The 1- and 5-year ischemic outcomes were compared. RESULTS Among 24,354 patients with available data on smoking status, 6,722 (27.6%) were current smokers. Smokers were younger and less likely to have diabetes mellitus; hypertension; hyperlipidemia; or prior myocardial infarction (MI), PCI, or coronary artery bypass grafting. Angiographically, smokers had longer lesions, more complex lesions, and more occlusions, but were less likely to have moderate or severe calcification or tortuosity. At 5 years, smokers had significantly higher rates of MI (7.8% vs. 5.6%; p < 0.0001) and definite or probable stent thrombosis (3.5% vs. 1.8%; p < 0.0001); however, there were no differences in the rates of death, cardiac death, target lesion revascularization, or composite endpoints (cardiac death, target vessel MI, or ischemic target lesion revascularization). After multivariable adjustment for potential confounders, smoking was a strong independent predictor of death (hazard ratio [HR]: 1.86; 95% confidence interval [CI]: 1.63 to 2.12; p < 0.0001), cardiac death (HR: 1.68; 95% CI: 1.38 to 2.05; p < 0.0001), MI (HR: 1.38; 95% CI: 1.20 to 1.58; p < 0.0001), stent thrombosis (HR: 1.60; 95% CI: 1.28 to 1.99; p < 0.0001), and target lesion failure (HR: 1.17; 95% CI: 1.05 to 1.30; p ¼ 0.005). CONCLUSIONS The present large, patient-level, pooled analysis with 5-year follow-up clearly demonstrates smoking to be an important predictor of adverse outcomes after PCI. (J Am Coll Cardiol Intv 2019;12:1941–50) © 2019 by the American College of Cardiology Foundation.
S
moking is one of the strongest modifiable risk
after percutaneous coronary intervention (PCI), a
factors for coronary artery disease (CAD). Thir-
phenomenon often termed the smoker’s paradox.
ty percent of CAD deaths in United States are
Although most of these studies were completed in
attributed to smoking, with the risk being dose
the thrombolytic therapy and balloon angioplasty
related (1,2). Smoking cessation constitutes the single
eras, and even though it has been 3 decades since
most important preventive measure for CAD (1). Para-
the first description of this paradox (3–9), the reasons
doxically, many (3–9), but not all (10–13), studies in
underlying this enigma remain unclear.
the past 3 decades have reported that smoking is associated with better short- and long-term outcomes
To further investigate the relationship between smoking
status
and
5-year
outcomes
in
the
From the aDepartment of Medicine, Division of Cardiology, Bronx Lebanon Hospital Center, New York, New York; bClinical Trials Center, Cardiovascular Research Foundation, New York, New York; cGagnon Cardiovascular Institute, Morristown Medical Center, Morristown, New Jersey; dHôpital du Sacré-Coeur de Montréal, Montréal, Canada; and the eDepartment of Medicine, Division of Cardiology, NewYork-Presbyterian Hospital/Columbia University Medical Center, New York, New York. This investigatorsponsored study was funded by Abbott Vascular (Santa Clara, California). The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Manuscript received February 21, 2019; revised manuscript received May 16, 2019, accepted June 11, 2019.
ISSN 1936-8798/$36.00
https://doi.org/10.1016/j.jcin.2019.06.034
Yadav et al.
1942
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
ABBREVIATIONS
contemporary era of PCI with stent implan-
AND ACRONYMS
tation, we examined this issue from an indi-
Research Consortium definition) (32), and target
vidual patient data (IPD) pooled analysis of
lesion failure (TLF) (a composite of cardiac death,
18 large-scale, multicenter, prospective, ran-
target vessel MI, and ischemic TLR). If TLF was not
domized controlled trials.
reported in the original trials, it was derived based on
ACS = acute coronary syndromes
CAD = coronary artery disease
individual components. TLF was evaluated in this
SEE PAGE 1951
CABG = coronary artery bypass
manner in 6 studies (N ¼ 6,808). Overall, TLF was
grafting
METHODS
IPD = individual patient data
definite or probable stent thrombosis (ST) (Academic
available in 12 of the 18 studies (19,349 of 24,354 patients) included in the present analysis. Survival time
MI = myocardial infarction PCI = percutaneous coronary intervention
STUDY PROTOCOL. The study population
was calculated as days from procedure date to first-
was drawn from an IPD pooled database
event date.
consisting of 18 multicenter, prospective,
ST = stent thrombosis
single- or double-blinded, randomized clin-
TLF = target lesion failure
ical trials with a follow-up period up to 5
TLR = target lesion
years. The study protocols and trial designs
revascularization
have been previously described in detail and are summarized in Table 1 (14–31). The main purpose of the present study was to evaluate the impact of smoking on 1- and 5-year outcomes among patients undergoing PCI in the stent era. Current smoking status at baseline was used to stratify the patient population.
STATISTICAL METHODS. Continuous data are pre-
sented as mean SD and were compared using the Student’s t-test or Wilcoxon rank sum test, as appropriate. Categorical variables are presented as frequency and percent and were compared using the chi-square test or the Fisher exact test (when any cell’s expected frequency fell below 5 counts). Clinical endpoints were presented using Kaplan-Meier estimates and compared using the log-rank test. Data were also analyzed stratified by study (CochranMantel-Haenszel test for categorical variables, 2-way
STUDY ENDPOINTS. The primary (adjudicated) study
analysis of variance for continuous variables).
endpoints were death, cardiac death, myocardial
A Cox proportional hazards model associated
infarction (MI), target lesion revascularization (TLR),
smoking with 5-year clinical endpoints adjusting for
T A B L E 1 Characteristics of Multicenter Randomized Trials Included in the Patient-Level Pooled Analysis
Trial (Ref. #)
RAVEL (14) E-SIRUS (15)
n
Study Cohort
Comparison
Randomization
Antithrombotic
238
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
Primary Endpoint
In-stent late luminal loss
352
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
In-stent minimum lumen diameter
1,058
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
Cardiac death/MI/TVR
C-SIRUS (17)
100
Stable/unstable angina, SI
SES vs. BMS
1:1
UFH
In-stent minimum lumen diameter
TAXUS IV (18)
1,314
Stable/unstable angina, SI
PES vs. BMS
1:1
UFH
Ischemia-driven TVR
TAXUS V (19)
1,156
Stable/unstable angina, SI
PES vs. BMS
1:1
UFH
Ischemia-driven TVR
ENDEAVOR II (20)
1,197
Stable/unstable angina, SI
ZES vs. BMS
1:1
UFH
Cardiac death/MI/TVR
ENDEAVOR III (21)
436
Stable/unstable angina
ZES vs. SES
3:1
UFH
SPIRIT II (22)
300
Stable/unstable angina, SI
EES vs. PES
3:1
UFH, BIV
SPIRIT III (23)
1,002
Stable/unstable angina, SI
EES vs. PES
2:1
UFH, BIV
HORIZONS-AMI (24)
3,006
STEMI
PES vs. BMS
3:1
UFHþGPI, BIV
COMPARE (25)
1,800
All-comers
EES vs. PES
1:1
UFH
ENDEAVOR IV (26)
1,548
Stable/unstable angina, SI
ZES vs. PES
1:1
UFH, BIV
Cardiac death/MI/TVR
SPIRIT IV (27)
3,687
Stable/unstable angina, SI
EES vs. PES
2:1
UFH, BIV
Cardiac death/target vessel MI, or TVR
TWENTE (28)
1,391
All-comers*
ZES vs. CoCr-EES
1:1
UFH
Cardiac death/target vessel MI/TVR
COMPARE II (29)
2,707
All-comers
BES vs. EES
2:1
UFH
Cardiac death/nonfatal MI, TVR
PLATINUM (30)
1,530
Stable/unstable angina, SI
PtCr-EES vs. CoCr-EES
1:1
UFH, LMWH, BIV
TWENTE II (31)
2,371
All-comers
CoCr-ZES vs. PtCr-EES
1:1
UFH
SIRIUS (16)
In-segment late lumen loss In-stent late luminal loss In-segment late lumen loss TVR and MACE (death/MI/stroke/ST) Death/MI/TVR
Cardiac death†/MI/†TVR Cardiac death/target vessel MI/TVR
*Except STEMI within 48 h. †All events related to target vessel. BIV ¼ bivalirudin; BMS ¼ bare-metal stent; BES ¼ biolimus-eluting stent; CoCr-EES ¼ cobalt-chromium everolimus-eluting stent; EES ¼ everolimus-eluting stent; GPI ¼ glycoprotein IIb/IIIa inhibitor; LMWH ¼ low-molecular-weight heparin; NSTE-ACS ¼ non–ST-segment elevation acute coronary syndrome; PES ¼ paclitaxel-eluting stent; PtCr-EES ¼ platinum-chromium everolimus-eluting stent; SES ¼ sirolimus-eluting stent; SI ¼ silent ischemia; ST ¼ stent thrombosis; STEMI ¼ ST-segment elevation myocardial infarction; TLR ¼ target lesion failure; TVR ¼ target vessel revascularization; UFH ¼ unfractionated heparin; ZES ¼ zotarolimus-eluting stent.
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
T A B L E 2 Baseline and Procedural Characteristics Stratified by Smoking Status
Smokers (n ¼ 6,722)
Nonsmokers (n ¼ 17,632)
p Value
56.5 9.9
65.1 10.3
<0.0001
5,056/6,722 (75.2)
12,426/17,632 (70.5)
<0.0001
28.2 5.2
29.2 5.4
<0.0001
1,183/6,719 (17.6)
4,605/17,623 (26.1)
<0.0001
Age, yrs Male Body mass index, kg/m2 Diabetes mellitus
330/6,719 (4.9)
1,337/17,623 (7.6)
<0.0001
3,629/6,715 (54.0)
11,942/17,610 (67.8)
<0.0001 <0.0001
Insulin treated Hypertension Hyperlipidemia
3,852/6,675 (57.7)
11,694/17,507 (66.8)
Prior myocardial infarction
1,499/6,683 (22.4)
3,976/17,483 (22.7)
0.60
Prior percutaneous coronary intervention
1,009/6,690 (15.1)
3,707/17,574 (21.1)
<0.0001
203/6,720 (3.0)
1,477/17,630 (8.4)
<0.0001
4,226/6,283 (67.3)
7,678/16,163 (47.5)
<0.0001
2,062/6,722 (30.7)
2,323/17,630 (13.2)
<0.0001
670/6,722 (10.0)
1,333/17,630 (7.6)
<0.0001
Unstable angina
1,494/6,283 (23.8)
4,022/16,163 (24.9)
0.08
Stable coronary artery disease
2,057/6,283 (32.7)
8,485/16,163 (52.5)
<0.0001
1,895/6,283 (30.2)
7,798/16,163 (48.2)
<0.0001
162/6,283 (2.6)
687/16,163 (4.3)
<0.0001
Prior coronary artery bypass grafting Acute coronary syndromes STEMI NSTE-ACS
Stable angina SI Lesion length, mm
17.4 10.8
16.9 11.2
0.006
ACC type C lesion
2,803/6,685 (41.9)
6,237/17,559 (35.5)
<0.0001
1,718/6,343 (27.1)
5,281/16,508 (32.0)
<0.0001
197/3,370 (5.8)
669/8,099 (8.3)
<0.0001
1,118/6,692 (16.7)
1,551/17,587 (8.8)
<0.0001
30.8 21.1
30.5 21.4
BMS
1,033/6,722 (15.4)
2,379/17,632 (13.5)
Drug-eluting stent
5,689/6,722 (84.6)
15,253/17,632 (86.5)
0.0002
First generation
2,370/6,722 (35.3)
5,331/17,632 (30.2)
<0.0001
Second generation
3,319/6,722 (49.4)
9,922/17,632 (56.3)
<0.0001
Calcification (moderate/severe) Tortuosity (moderate/severe) Any occlusions Total stent length
0.25 0.0002
Pre-procedure Reference vessel diameter, mm
2.83 0.86
2.73 0.58
<0.0001
Minimum lumen diameter, mm
0.70 0.49
0.76 0.44
<0.0001
Diameter stenosis, %
76.3 17.0
73.2 15.6
<0.0001
TIMI flow grade 0/1
1,533/6,689 (22.9)
2,333/17,574 (13.3)
<0.0001
Post-procedure Diameter stenosis, %
16.8 11.1
16.8 10.9
0.93
Minimum lumen diameter, mm
2.35 0.92
2.28 0.75
<0.0001
31/6,613 (0.5)
77/17,407 (0.4)
57.1 11.3
57.8 10.9
TIMI flow grade 0/1 Left ventricular ejection fraction, %
0.78 0.005
Values are mean SD or n/N (%). ACC ¼ American College of Cardiology; TIMI ¼ Thrombolysis In Myocardial Infarction; other abbreviations as in Table 1.
age (10-year increments), sex, diabetes mellitus,
performed that accounted for possible confounders
insulin-treated diabetes, hypertension, hyperlipid-
by trial; these include the Cox proportional hazards
emia, prior PCI, prior coronary artery bypass grafting
model: 1) not including the study; 2) stratified by
(CABG) surgery, prior MI, acute coronary syndrome
study; 3) including study as a random effect; and 4)
(ACS) versus stable CAD, any moderate or severe
including the study as a covariate.
calcification, lesion length, stent generation, any occlusion, any American College of Cardiology class C
RESULTS
lesion, and study (as a random effect). All p values are 2-sided, and a p value < 0.05 was considered statis-
BASELINE PATIENT AND ANGIOGRAPHIC CHARAC-
tically
TERISTICS. Smoking status was available in 24,354 of
significant.
Sensitivity
analyses
were
1943
1944
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
American College of Cardiology type C lesions, as
T A B L E 3 Outcomes at 1 and 5 Years in Patients Stratified by
well as any occlusions, but were less likely to have
Smoking Status
moderate or severe calcification and tortuosity. Smokers (n ¼ 6,722)
Nonsmokers (n ¼ 17,632)
p Value
Death
111 (1.7)
290 (1.7)
0.95
to have pre-procedural Thrombolysis In Myocardial
Cardiac death
67 (1.0)
178 (1.0)
0.94
Infarction flow grade 0 or 1, but there was no sig-
Myocardial infarction
226 (3.4)
529 (3.0)
0.15
nificant difference in post-procedure TIMI flow
Target lesion revascularization
321 (5.8)
902 (6.5)
0.06
grade 0 or 1 between the 2 groups. On quantitative
<0.0001
Nevertheless, the total stent length was similar between the 2 groups. Smokers were also more likely
1-yr outcomes
Stent thrombosis (definite/probable) Target lesion failure*
117 (1.8)
148 (0.8)
282 (6.0)
1,042 (7.2)
430 196 450 589 211
1,251 583 892 1,446 275
0.006
5-yr outcomes Death Cardiac death Myocardial infarction Target lesion revascularization Stent thrombosis (definite/ probable) Target lesion failure*
(8.6) (3.9) (7.8) (12.0) (3.5)
(9.0) (4.1) (5.6) (11.4) (1.8)
0.59 0.44 <0.0001 0.45 <0.0001
coronary angiography, smokers were more likely to have a lower pre-procedure minimal lumen diameter and a larger post-procedure minimal lumen diameter. CUMULATIVE 1-YEAR AND 5-YEAR OUTCOMES. Median
follow-up was 3.9 (interquartile range: 3.0 to 5.0) 601 (14.8)
1,892 (14.4)
0.74
Values are n (%) based on Kaplan-Meier estimates. *Cardiac death, target vessel myocardial infarction, and ischemic target lesion revascularization.
years. Patients lost to follow-up were censored at the last known follow-up date. At 1 year post-PCI, smokers compared with nonsmokers had a lower rate of TLF (6.0% vs. 7.2%; p ¼ 0.006), and a trend toward less TLR (5.8% vs.
24,633 (98.9%) patients enrolled in the 18 randomized trials, 6,722 (27.6%) of whom were current smokers. Table 2 summarizes baseline patient clinical characteristics and angiographic findings stratified by smoking
status.
Compared
with
nonsmokers,
smokers were more likely to be men, were younger (by w9 years), and were less likely to have other cardiac risk factors including diabetes mellitus, hypertension, and hyperlipidemia. Smokers were also less likely to have prior PCI or prior CABG surgery.
6.5%; p ¼ 0.06), but a greater rate of definite or probable ST (1.8% vs. 0.8%; p < 0.0001). There were no significant differences in the rates of death, cardiac death, or MI between smokers and nonsmokers at 1 year (Table 3). At 5 years post-PCI, smokers had significantly higher rates of MI (7.8% vs. 5.6%; p < 0.0001) and definite or probable ST (3.5% vs. 1.8%; p < 0.0001); however, there were no significant differences in the rates of death, cardiac death, TLR, or TLF (Table 3).
Baseline characteristics stratified by study are pre-
MULTIVARIABLE ANALYSIS. Table 4 and the Central
sented in Online Table 1.
Illustration present the risk of adverse outcomes ac-
By angiographic core laboratory analysis smokers were more likely to have longer lesions and more
cording to smoking status before and after adjustment
for
potential
confounders.
Notably,
the
T A B L E 4 Unadjusted and Adjusted HRs for Adverse Events at 5 Years in Patients Stratified by Smoking Status
Death
Smokers (n ¼ 6,722)
Nonsmokers (n ¼ 17,632)
Unadjusted HR (95% CI)
430 (8.6)
1,251 (9.0)
0.97 (0.87–1.08)
p Value
Adjusted HR (95% CI)†
p Value
0.59
1.86 (1.63–2.12)
<0.0001
Cardiac death
196 (3.9)
583 (4.1)
0.94 (0.80–1.10)
0.44
1.68 (1.38–2.05)
<0.0001
Myocardial infarction
450 (7.8)
892 (5.6)
1.37 (1.22–1.53)
<0.0001
1.38 (1.20–1.58)
<0.0001
Target lesion revascularization
589 (12.0)
1,446 (11.4)
1.04 (0.94–1.14)
0.45
1.06 (0.94–1.19)
211 (3.5)
275 (1.8)
2.09 (1.75–2.50)
<0.0001
1.60 (1.28–1.99)
<0.0001
601 (14.8)
1,892 (14.4)
0.98 (0.90–1.08)
0.74
1.05 (1.01–1.10)
0.005
Stent thrombosis (definite/ probable) Target lesion failure*
0.33
Values are n (%) based on Kaplan-Meier estimates. *Cardiac death, target vessel myocardial infarction, and ischemic target lesion revascularization. †Univariate variables included for multivariable analysis were current smoker, age (10-yr increments), sex, diabetes, insulin-treated diabetes, hypertension, hyperlipidemia, previous percutaneous coronary intervention, previous coronary artery bypass grafting, previous myocardial infarction, acute coronary syndromes vs. stable coronary artery disease, any moderate/ severe calcification, lesion length, bare metal stents vs. second-generation drug-eluting stents, first vs. second-generation drug-eluting stents, any occlusion, any American College of Cardiology class C lesion. CI ¼ confidence interval; HR ¼ Hazard ratio.
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
C ENTR AL I LL U STRA T I O N Forest Plots Showing Unadjusted and Adjusted HRs for Adverse Events at 5 Years in Patients According to Smoking Status
Unadjusted vs. Adjusted Hazard Ratio (95% Confidence Interval) Death
Cardiac Death
0.97 (0.87–1.08), p = 0.59 1.86 (1.63–2.12), p < 0.0001 0.94 (0.80–1.10), p = 0.44 1.68 (1.38–2.05), p < 0.0001
Myocardial Infarction
1.37 (1.22–1.53), p < 0.0001
Target Lesion Revascularization
1.04 (0.94–1.14), p = 0.45 1.06 (0.94–1.19), p = 0.33
Definite/Probable Stent Thrombosis
2.09 (1.75–2.50), p < 0.0001 1.60 (1.28–1.99), p < 0.0001
Target Lesion Failure
1.38 (1.20–1.58), p < 0.0001
0.98 (0.90–1.08), p = 0.74 1.17 (1.05–1.30), p = 0.005 0.50
0.75
1.00
Risk Decreases
1.50
2.00
2.50 3.00
Risk Increases
Unadjusted HR
Adjusted HR
Yadav, M. et al. J Am Coll Cardiol Intv. 2019;12(19):1941–50.
The true risk among smokers after percutaneous coronary intervention (PCI) emerges only after adjustment of risk factors as evidenced by the marked changes in hazard ratios for death, cardiac death, and target lesion failure. These findings likely underlie the smoker’s paradox often reported in smokers in the published reports. HR ¼ hazard ratio.
association between smoking and the occurrence of
smoking status and hypertension (p int ¼ 0.02),
death, cardiac death, TLR, and TLF were neutral
hyperlipidemia (p int ¼ 0.01), and ACS (pint ¼ 0.0002)
before adjustment, but after adjustment, smoking
for death (Figure 1), but no interactions were present
was strongly and significantly associated with higher
for definite or probable ST (Online Figure 1). The
5-year rates of death, cardiac death, and TLF. Smok-
hazard ratio for death among smokers with ACS and
ing continued to be associated with higher rates of MI
stable CAD were 1.61 (95% confidence interval: 1.37 to
and definite or probable ST after adjustment for po-
1.94) and a 2.19 (95% confidence interval: 1.82 to 2.65)
tential confounders. In sensitivity analysis, there
respectively (p int ¼ 0.02), after adjusting for cova-
were no significant differences in outcomes between
riates (Online Figure 2).
the 4 test methodologies (Online Table 2). Table 5 shows the independent predictors of death
DISCUSSION
and other adverse ischemic events. Smoking was an independent predictor of death, cardiac death, MI,
The principal findings of the present pooled analysis
definite or probable ST, and TLF. In subgroup anal-
of IPD from 18 large-scale, multicenter, randomized
ysis, significant interactions were present between
controlled stent trials are as follows: 1) smokers
1945
1946
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
smoking was a strong and independent predictor of
T A B L E 5 Independent Predictors of Adverse Outcomes at 5 Years
death, cardiac death, MI, definite or probable ST, and
HR (95% CI)
p Value
TLF. Given decades of evidence related to the detri-
Death Current smoker
1.86 (1.63–2.12)
<0.0001
mental effect of cigarette smoking on health in general (33,34), the role of smoking in relation to
Age (10-yr increase)
2.19 (2.06–2.32)
<0.0001
Insulin-treated diabetes
1.68 (1.40–2.02)
<0.0001
Previous coronary artery bypass grafting
1.45 (1.23–1.71)
<0.0001
Diabetes
1.41 (1.23–1.61)
<0.0001
Previous myocardial infarction
1.32 (1.17–1.49)
<0.0001
Male
1.24 (1.10–1.39)
0.0003
First- vs. second-generation drug-eluting stent
1.16 (1.03–1.32)
0.02
Hyperlipidemia
cardiovascular outcomes has been controversial at best since the first description of the so-called smoker’s paradox (4,5). Some studies have supported its existence (i.e., smoking improves cardiovascular outcomes in the context of an acute event such as an MI) (3–9), whereas others have refuted its
0.82 (0.73–0.92)
0.0007
Current smoker
1.68 (1.38–2.05)
<0.0001
Age (10-yr increase)
1.89 (1.73–2.06)
<0.0001
these studies were registry analyses or were con-
Diabetes
1.81 (1.49–2.19)
<0.0001
ducted in the era of thrombolytic therapy and balloon
Insulin-treated diabetes
1.74 (1.35–2.24)
<0.0001
angioplasty; this limits their applicability to modern
Prior coronary artery bypass grafting
1.62 (1.28–2.04)
<0.0001
PCI. To the best of our knowledge, our study is by far
Any occlusion
1.53 (1.17–2.02)
0.002
the largest patient-level pooled analysis of random-
Prior myocardial infarction
1.42 (1.19–1.70)
0.0001
ized controlled trials with blinded core laboratory
Any moderate/severe calcification
1.38 (1.17–1.62)
0.0001
analyses and independent event adjudication inves-
Hyperlipidemia
0.76 (0.64–0.90)
0.001
tigating
existence (i.e., smoking is not associated with better
Cardiac death
Myocardial infarction
cardiovascular outcomes) (10–12). Moreover, many of
the
smoker’s
paradox
concept
in
the
contemporary stent era.
Current smoker
1.38 (1.20–1.58)
<0.0001
First- vs. second-generation drug-eluting stent
1.75 (1.47–2.07)
<0.0001
Insulin-treated diabetes
1.58 (1.26–2.00)
BMS vs. second-generation drug-eluting stents
1.54 (1.24–1.91)
<0.0001
Prior myocardial infarction
1.39 (1.21–1.60)
<0.0001
0.0001
Prior percutaneous coronary intervention
1.31 (1.13–1.51)
0.0004
Previous coronary artery bypass grafting
1.27 (1.03–1.56)
0.02
Hypertension
1.23 (1.07–1.40)
0.003
Any moderate/severe calcification
1.19 (1.05–1.36)
0.007
Smokers
undergoing
PCI
were
significantly
younger, were predominantly men, and had a lower prevalence of known CAD risk factors including diabetes
mellitus,
hypertension,
hyperlipidemia,
prior PCI, or prior CABG surgery. These baseline differences were also noted in most other largescale studies including SYNTAX (Synergy between PCI
with
Taxus
and
Cardiac
Surgery)
trial
Any ACC class C lesion
1.18 (1.02–1.37)
0.03
(3,4,10,13,35). This finding underlines the fact that
Lesion length
1.01 (1.01–1.02)
<0.0001
smokers are prone to more rapidly develop CAD or present
ST (definite/probable)
with
plaque
rupture
and
thrombosis,
Current smoker
1.60 (1.28–1.99)
<0.0001
necessitating PCI almost a decade earlier compared
First- vs. second-generation drug-eluting stent
1.97 (1.45–2.67)
<0.0001
with nonsmokers, even with a lower prevalence of
Insulin-treated diabetes
1.73 (1.17–2.57)
0.006
coronary risk factors (10). After proper adjustment
BMS vs. second-generation drug-eluting stents
1.65 (1.12–2.42)
0.01
for potentially relevant confounders, smokers had
Prior myocardial infarction
1.52 (1.19–1.94)
0.0007
an w95% increased risk of death, w85% increased
Prior percutaneous coronary intervention
1.35 (1.05–1.75)
0.02
risk of cardiac death, and w70% increased risk of
Any moderate or severe calcification
1.31 (1.07–1.62)
0.01
definite or probable ST at 5 years after coronary
Lesion length
1.01 (1.01–1.02)
0.0005
0.87 (0.79–0.96)
0.006
Age (10-yr increase)
Continued on the next page
stent
implantation
compared
with
nonsmokers.
Thus, smoking results not only earlier presentation with CAD, but also a poor prognosis after contemporary
percutaneous
revascularization
therapy,
emphasizing the long-term deleterious effects of undergoing PCI were significantly younger and had
smoking.
less coronary risk factors compared with nonsmokers;
Potential mechanisms to explain the increase in
2) at 5 years post-PCI, smokers had similar unadjusted
mortality and adverse ischemic events after stenting
rates of death and cardiac death but higher rates of MI
in patients who are active smokers is likely multi-
and ST compared with nonsmokers; and 3) after
factorial. Smokers have increased platelet aggrega-
multivariable adjustment for potential confounders,
tion (36,37), increased fibrinogen, and decreased
OCTOBER 14, 2019:1941–50
fibrinolytic activity compared with nonsmokers (37), creating a state of hypercoagulability that pre-
Smoker’s Paradox
T A B L E 5 Continued
HR (95% CI)
disposes to acute thrombosis (38). Moreover, in patients with obstructive CAD, smoking may cause
1947
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019
p Value
Target lesion revascularization
coronary vasoconstriction (39). Smoking also in-
Current smoker
1.06 (0.94–1.19)
creases myocardial workload by increasing heart rate,
BMS vs. second-generation drug-eluting stents
2.59 (2.20–3.05)
<0.0001
myocardial contractility, and blood pressure (38). The
Prior coronary artery bypass grafting
1.44 (1.21–1.71)
<0.0001
higher carboxyhemoglobin level in smokers also de-
First- vs. second-generation drug-eluting stent
1.32 (1.14–1.52)
0.0001
Prior percutaneous coronary intervention
1.31 (1.16–1.48)
<0.0001
Diabetes
1.22 (1.08–1.39)
0.001
Insulin-treated diabetes
1.21 (1.00–1.46)
0.05
creases blood oxygenation. Smoking induces catecholamine surge with cardiac ischemia, which may be a cause of cardiac arrhythmias and sudden cardiac death (40). Moreover, smoking decreases highdensity lipoproteins and increases oxidized lowdensity lipoproteins (41,42), induces endothelial dysfunction (43) and neutrophil activation (38), causes oxidant injury (43), increases fibrinogen levels,
0.33
Hypertension
1.11 (1.00–1.24)
0.05
Lesion length
1.01 (1.01–1.02)
<0.0001
Age (10-yr increase)
0.91 (0.87–0.96)
0.0002
Male
0.89 (0.80–1.00)
0.04
Prior myocardial infarction
0.88 (0.78–1.00)
0.04
TLF*
and causes platelet activation (37), all of which in-
Current smoker
1.17 (1.05–1.30)
0.005
crease the rate of atherosclerosis and plaque pro-
BMS vs. second-generation drug-eluting stents
2.01 (1.72–2.35)
<0.0001
gression by direct or indirect effects (38).
Previous coronary artery bypass grafting
1.62 (1.42–1.85)
<0.0001
First- vs. second-generation drug-eluting stent
1.39 (1.23–1.57)
<0.0001 <0.0001
STUDY STRENGTHS AND LIMITATIONS. The present
Insulin-treated diabetes
1.38 (1.18–1.62)
analysis has several strengths. IPD pooling of more
Diabetes
1.23 (1.10–1.37)
0.0003
than 24,000 patients from 18 randomized trials, each
Any moderate or severe calcification
1.19 (1.08–1.30)
0.0004
of which was monitored and had careful event adju-
Any ACC class C lesion
1.19 (1.07–1.32)
0.002
Hypertension
1.13 (1.02–1.25)
0.02
Age (10-yr increase)
1.05 (1.01–1.10)
0.02
Lesion length
1.01 (1.00–1.01)
<0.0001
dication, helped to identify differences that were hard to ascertain in previous underpowered studies. A recent post hoc analysis of the SYNTAX trial at 5 years also showed similar results. Smokers had worse clinical outcomes due to a higher incidence of recurrent MI in both the PCI and CABG arms, and smoking was an independent predictor of the composite endpoint of death or MI or stroke and the
Variables included for multivariable analysis were current smoker, age (10-yr increments), sex, diabetes, insulintreated diabetes, hypertension, hyperlipidemia, prior percutaneous coronary intervention, prior coronary artery bypass grafting, prior myocardial infarction, acute coronary syndrome vs. stable coronary artery disease, any moderate/severe calcification, lesion length, BMS vs. second-generation drug-eluting stents, first- vs. secondgeneration drug-eluting stents, any occlusion, any ACC class C lesion. *Includes cardiac death, target vessel myocardial infarction, and ischemic target lesion revascularization. Abbreviations as in Tables 1 and 2.
composite endpoint of death or MI or stroke or target vessel revascularization (13). The IPD approach also allows for multivariable analysis to adjust for baseline imbalances, and subgroup analyses. There were
the number of packs of cigarettes smoked, we
no significant interactions between smoking status
could not assess a dose-dependent relationship to
and numerous variables, except for hypertension,
outcomes. Definitions of MI were slightly different
hyperlipidemia, and ACS at 5 years for death as an
across trials, potentially reducing precision. Several
outcome, implying that smoking with hypertension
angiographic
or with ACS combined have a particularly poor long-
small vessels, long lesions, and overlapping stents
term prognosis.
were not systematically collected in all randomized
As in most previous studies of this subject, smoking status was defined at baseline; data on smoking status during follow-up were not available. We cannot
and
procedural
variables
such
as
trials. Similarly, data on access site, left ventricular function,
and
medication
adherence
were
not
available.
therefore evaluate the impact of smoking crossovers after treatment (from nonsmokers or smokers, or vice versa), although the deleterious impact of smoking
CONCLUSIONS
identified in the present report may have been even greater were we able to account for such effects. In
Results from the present large-scale analysis largely
addition, we were unable to differentiate ex-smokers
dismisses a smoker’s paradox after PCI with coronary
from nonsmokers. Also, owing to the lack of data on
stent
implantation.
Even
before
multivariable
1948
Yadav et al. Smoker’s Paradox
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
F I G U R E 1 Forest Plots Showing Subgroup Analysis for Death According to Smoking Status
Subgroup analysis shows significant interaction of smoking status with hypertension, hyperlipidemia and acute coronary syndrome. ACC ¼ American College of Cardiology; BMS ¼ bare-metal stent; CABG ¼ coronary artery bypass graft; CI ¼ confidence interval; DES ¼ drug-eluting stent; KM ¼ Kaplan-Meier; MI ¼ myocardial infarction; PCI ¼ percutaneous coronary intervention.
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
adjustment, active smokers had higher 5-year rates of MI and ST after stenting. When baseline imbalances were also taken into account, smoking was also associated with increased long-term rates of cardiac and all-cause mortality. Given the doubleedged sword of earlier presentation by nearly a
PERSPECTIVES WHAT IS KNOWN? Some studies have shown that smokers have better outcomes after PCI, a phenomenon widely known as the smoker’s paradox.
decade and increased long-term adverse outcomes
WHAT IS NEW? The smoker’s paradox may be explained by the
after PCI, the strongest possible individual and
younger age and lower prevalence of risk factors among
societal-based efforts should be undertaken to pre-
smokers; these trends reverse after adjustment of risk factors
vent smoking and strongly encourage its cessation
such that early and late outcomes are substantially worse in
once begun.
smokers after PCI.
ADDRESS FOR CORRESPONDENCE: Dr. Gregg W.
Stone, Columbia University Medical Center, Cardiovascular Research Foundation, 1700 Broadway, 8th Floor, New York, New York 10019. E-mail: gs2184@
WHAT IS NEXT? Further randomized studies are needed to study the impact of smoking on heart disease in general and specifically after PCI. Greater emphasis on smoking cessation strategies is warranted in patients with CAD.
columbia.edu. Twitter: @greggwstone.
REFERENCES 1. Ockene IS, Miller NH. Cigarette smoking, cardiovascular disease, and stroke: a statement for health care professionals from the American Heart Association. American Heart Association Task Force on Risk Reduction. Circulation 1997;96: 3243–7. 2. Health Quality Ontario. Intravascular ultrasound to guide percutaneous coronary interventions: an evidence-based analysis. Ont Health Technol Assess Series 2006;6:1–97. 3. Andrikopoulos GK, Richter DJ, Dilaveris PE, et al. In-hospital mortality of habitual cigarette smokers after acute myocardial infarction; the “smoker’s paradox” in a countrywide study. Eur Heart J 2001;22:776–84. 4. Barbash GI, Reiner J, White HD, et al. Evaluation of paradoxic beneficial effects of smoking in patients receiving thrombolytic therapy for acute myocardial infarction: mechanism of the “smoker’s paradox” from the GUSTO-I trial, with angiographic insights. Global Utilization of Streptokinase and Tissue-Plasminogen Activator for Occluded Coronary Arteries. J Am Coll Cardiol 1995;26:1222–9. 5. Barbash GI, White HD, Modan M, et al. Significance of smoking in patients receiving thrombolytic therapy for acute myocardial infarction. Experience gleaned from the International Tissue Plasminogen Activator/Streptokinase Mortality Trial. Circulation 1993;87:53–8. 6. Molstad P. First myocardial infarction in smokers. Eur Heart J 1991;12:753–9. 7. Gourlay SG, Rundle AC, Barron HV. Smoking and mortality following acute myocardial infarction: results from the National Registry of Myocardial Infarction 2 (NRMI 2). Nicotine Tob Res 2002;4: 101–7. 8. Ruiz-Bailen M, de Hoyos EA, Reina-Toral A, Torres-Ruiz JM, Alvarez-Bueno M, Gomez Jimenez FJ. Paradoxical effect of smoking in the Spanish population with acute myocardial
infarction or unstable angina: results of the ARIAM Register. Chest 2004;125:831–40.
lesions in small native coronary arteries (C-SIRIUS). J Am Coll Cardiol 2004;43:1110–5.
9. Elosua R, Vega G, Rohlfs I, et al. Smoking and myocardial infarction case-fatality: hospital and population approach. Eur J Cardiovasc Prev Rehabil 2007;14:561–7.
18. Stone GW, Ellis SG, Cox DA, et al. A polymerbased, paclitaxel-eluting stent in patients with coronary artery disease. N Engl J Med 2004;350: 221–31.
10. Hasdai D, Garratt KN, Grill DE, Lerman A, Holmes DR Jr. Effect of smoking status on the
19. Stone GW, Ellis SG, Cannon L, et al. Compari-
long-term outcome after successful percutaneous coronary revascularization. N Engl J Med 1997; 336:755–61.
son of a polymer-based paclitaxel-eluting stent with a bare metal stent in patients with complex coronary artery disease: a randomized controlled trial. JAMA 2005;294:1215–23.
11. Himbert D, Klutman M, Steg G, White K, Gulba DC. Cigarette smoking and acute coronary syndromes: a multinational observational study. Int J Cardiol 2005;100:109–17.
20. Fajadet J, Wijns W, Laarman GJ, et al. Randomized, double-blind, multicenter study of the Endeavor zotarolimus-eluting phosphorylcholineencapsulated stent for treatment of native
12. Berger JS, Bhatt DL, Steinhubl SR, et al. Smoking, clopidogrel, and mortality in patients with established cardiovascular disease. Circulation 2009;120:2337–44. 13. Zhang YJ, Iqbal J, van Klaveren D, et al. Smoking is associated with adverse clinical outcomes in patients undergoing revascularization with PCI or CABG: the SYNTAX trial at 5-year follow-up. J Am Coll Cardiol 2015;65:1107–15. 14. Morice MC, Serruys PW, Sousa JE, et al. A randomized comparison of a sirolimus-eluting stent with a standard stent for coronary revascularization. N Engl J Med 2002;346:1773–80. 15. Schofer J, Schluter M, Gershlick AH, et al. Sirolimus-eluting stents for treatment of patients with long atherosclerotic lesions in small coronary arteries: double-blind, randomised controlled trial (E-SIRIUS). Lancet 2003;362:1093–9. 16. Moses JW, Leon MB, Popma JJ, et al. Sirolimus-eluting stents versus standard stents in patients with stenosis in a native coronary artery. N Engl J Med 2003;349:1315–23. 17. Schampaert E, Cohen EA, Schluter M, et al. The Canadian study of the sirolimus-eluting stent in the treatment of patients with long de novo
coronary artery lesions: clinical and angiographic results of the ENDEAVOR II trial. Circulation 2006; 114:798–806. 21. Kandzari DE, Leon MB, Popma JJ, et al. Comparison of zotarolimus-eluting and sirolimuseluting stents in patients with native coronary artery disease: a randomized controlled trial. J Am Coll Cardiol 2006;48:2440–7. 22. Serruys PW, Ruygrok P, Neuzner J, et al. A randomised comparison of an everolimuseluting coronary stent with a paclitaxel-eluting coronary stent:the SPIRIT II trial. EuroIntervention 2006;2:286–94. 23. Stone GW, Midei M, Newman W, et al. Comparison of an everolimus-eluting stent and a paclitaxel-eluting stent in patients with coronary artery disease: a randomized trial. JAMA 2008; 299:1903–13. 24. Stone GW, Lansky AJ, Pocock SJ, et al. Paclitaxel-eluting stents versus bare-metal stents in acute myocardial infarction. N Engl J Med 2009; 360:1946–59. 25. Kedhi E, Joesoef KS, McFadden E, et al. Second-generation everolimus-eluting and paclitaxeleluting stents in real-life practice (COMPARE): a randomised trial. Lancet 2010;375:201–9.
1949
1950
Yadav et al.
JACC: CARDIOVASCULAR INTERVENTIONS VOL. 12, NO. 19, 2019 OCTOBER 14, 2019:1941–50
Smoker’s Paradox
26. Leon MB, Mauri L, Popma JJ, et al. A randomized comparison of the Endeavor zotarolimus-eluting stent versus the TAXUS paclitaxel-eluting stent in de novo native coronary lesions 12-month outcomes from the ENDEAVOR IV trial. J Am Coll Cardiol 2010;55:543–54. 27. Stone GW, Rizvi A, Newman W, et al. Everolimus-eluting versus paclitaxel-eluting stents in coronary artery disease. N Engl J Med 2010;362: 1663–74. 28. von Birgelen C, Basalus MW, Tandjung K, et al. A randomized controlled trial in secondgeneration zotarolimus-eluting Resolute stents versus everolimus-eluting Xience V stents in realworld patients: the TWENTE trial. J Am Coll Cardiol 2012;59:1350–61. 29. Smits PC, Hofma S, Togni M, et al. Abluminal biodegradable polymer biolimus-eluting stent versus durable polymer everolimus-eluting stent (COMPARE II): a randomised, controlled, noninferiority trial. Lancet 2013;381:651–60. 30. Meredith IT, Teirstein PS, Bouchard A, et al. Three-year results comparing platinum-chromium PROMUS element and cobalt-chromium XIENCE
PEERS): a randomised, single-blind, multicentre, non-inferiority trial. Lancet 2014;383:413–23. 32. Cutlip DE, Windecker S, Mehran R, et al. Clinical end points in coronary stent trials: a case for standardized definitions. Circulation 2007;115: 2344–51. 33. Levitz JS, Bradley TP, Golden AL. Overview of smoking and all cancers. Med Clin North Am 2004; 88:1655–75, xiii. 34. Bartecchi CE, MacKenzie TD, Schrier RW. The human costs of tobacco use (1). N Engl J Med 1994;330:907–12. 35. Weisz G, Cox DA, Garcia E, et al. Impact of smoking status on outcomes of primary coronary intervention for acute myocardial infarction–the smoker’s paradox revisited. Am Heart J 2005;150: 358–64. 36. Levine PH. An acute effect of cigarette smoking on platelet function. A possible link between smoking and arterial thrombosis. Circulation 1973;48:619–23.
V everolimus-eluting stents in de novo coronary artery narrowing (from the PLATINUM Trial). Am J Cardiol 2014;113:1117–23.
37. Benowitz NL, Fitzgerald GA, Wilson M, Zhang Q. Nicotine effects on eicosanoid formation and hemostatic function: comparison of transdermal nicotine and cigarette smoking. J Am Coll Cardiol 1993;22:1159–67.
31. von Birgelen C, Sen H, Lam MK, et al. Thirdgeneration zotarolimus-eluting and everolimuseluting stents in all-comer patients requiring a percutaneous coronary intervention (DUTCH
38. Benowitz NL, Gourlay SG. Cardiovascular toxicity of nicotine: implications for nicotine replacement therapy. J Am Coll Cardiol 1997;29: 1422–31.
39. Nicod P, Rehr R, Winniford MD, Campbell WB, Firth BG, Hillis LD. Acute systemic and coronary hemodynamic and serologic responses to cigarette smoking in long-term smokers with atherosclerotic coronary artery disease. J Am Coll Cardiol 1984;4:964–71. 40. Escobedo LG, Zack MM. Comparison of sudden and nonsudden coronary deaths in the United States. Circulation 1996;93:2033–6. 41. Craig WY, Palomaki GE, Haddow JE. Cigarette smoking and serum lipid and lipoprotein concentrations: an analysis of published data. BMJ 1989; 298:784–8. 42. Harats D, Ben-Naim M, Dabach Y, Hollander G, Stein O, Stein Y. Cigarette smoking renders LDL susceptible to peroxidative modification and enhanced metabolism by macrophages. Atherosclerosis 1989;79:245–52. 43. Pittilo RM. Cigarette smoking, endothelial injury and cardiovascular disease. Int J Exp Pathol 2000;81:219–30.
KEY WORDS coronary artery disease, coronary artery disease outcomes, smoker’s paradox, smokers
A PPE NDI X For supplemental tables and figures, please see the online version of this paper.