Cardiovascular Safety of Varenicline: Patient-Level

Meta-Analysis of Randomized, Blinded,

Placebo-Controlled Trials
James H. Ware, PhD,
1
* George W. Vetrovec, MD,
2
Alan B. Miller, MD,
3
Andrew Van Tosh, MD,
4
Michael Gaffney, PhD,
5
Carla Yunis, MD,
5
Carmen Arteaga, PhD,
5
and Jeffrey S. Borer, MD
6
Smoking is a major modifiable risk factor for cardiovascular (CV) disease. Varenicline is a pharmacological
aid for smoking cessation. To explore the CV safety of varenicline, we investigated the incidence of CV
events in varenicline-treated subjects across all phase 24 randomized placebo-controlled clinical trials of
$12-week treatment duration conducted in smokers aged $18 years and sponsored by the drug manu-
facturer. This manuscript reports a subject-level meta-analysis of time to major adverse cardiovascular
events (MACE; defined as CV-related death, nonfatal myocardial infarction, nonfatal stroke) and time to
MACE+ (defined as MACE plus worsening or any procedure for peripheral vascular disease, hospitali-
zation for angina, or performance of coronary revascularization). All events were adjudicated by an
independent adjudication committee, blind to treatment assignment. Events were assessed during treat-
ment and up to 30 days after the last treatment dose. The primary analytical method was a stratified
logrank time-to-event analysis; secondary analyses were meta-analyses of incidence rate ratios and rate
differences. Overall, 7002 subjects were included (varenicline: 4190; placebo: 2812) from 15 studies. MACE
were reported by 13 varenicline subjects (0.31%) and 6 placebo subjects (0.21%) [hazard ratio, 1.95; 95%
confidence interval (CI): 0.794.82; P 5 0.15; risk difference, 0.006 events per subject-year; 95% CI: 20.003,
0.015, P 5 0.19]. MACE+ were reported by 26 varenicline subjects (0.62%) and 12 placebo subjects (0.43%)
(hazard ratio, 1.74; 95% CI: 0.913.34, P 50.10; risk difference, 0.010; 95% CI: 20.002, 0.022, P 50.11). This
subject-level meta-analysis of MACE or MACE+ up to 30 days posttreatment in placebo-controlled clinical
trials of varenicline found a trend toward increased incidence of these events in varenicline-treated patients
that did not reach statistical significance. The overall number of events was low and the absolute risk of
CV events with varenicline was small.
Keywords: varenicline, cardiovascular, adverse events, safety, risk
1
Department of Biostatistics, Harvard School of Public Health, Boston, MA;
2
Division of Cardiology, VCU Medical Center, Richmond, VA;
3
Department of Medicine, Division of Cardiology, University of Florida-Jacksonville, Jacksonville, FL;
4
Division of Cardiovascular Research,
St. Francis Hospital, Roslyn, NY;
5
Pfizer Inc., 235 East 42nd St, New York, NY; and
6
Division of Cardiovascular Medicine, Department of
Medicine, State University of New York Downstate Medical Center, Brooklyn and New York, NY.
This analysis and all clinical trials included in the analysis were sponsored by Pfizer Inc., the manufacturer of varenicline. J. H. Ware
received remuneration for travel expenses from Pfizer Inc. to attend a meeting to discuss this analysis. G. W. Vetrovec received remuneration
from Pfizer Inc. for consultancy and adverse event adjudication related to this analysis. G. W. Vetrovec also reports receiving financial
support for various activities, unrelated to this manuscript, from Pfizer Inc. and other pharmaceutical companies. A. B. Miller received
remuneration from Pfizer Inc. for consultancy and adverse event adjudication related to this analysis. A. B. Miller received financial support
for various activities unrelated to this manuscript from Pfizer Inc. and other pharmaceutical companies. A. Van Tosh received remuneration
from Pfizer Inc. for consultancy and adverse event adjudication related to this analysis. J. S. Borer received remuneration from Pfizer Inc. for
consultancy related to this analysis. J. S. Borer also reports receiving financial support for various activities unrelated to this manuscript
from Pfizer Inc. and other pharmaceutical companies. M. Gaffney, C. Yunis, and C. Arteaga are current employees of Pfizer Inc. and
currently hold Pfizer stock and stock options. None of the authors were paid in relation to developing this manuscript.
*Address for correspondence: Department of Biostatistics, Harvard School of Public Health, Boston, MA 02115. E-mail: ware@hsph.harvard.edu
American Journal of Therapeutics 20, 235246 (2013)
10752765 2013 Lippincott Williams & Wilkins www.americantherapeutics.com
INTRODUCTION
Since 1950, when the first report of the detrimental
effects of tobacco smoking on human health identified
smoking as a cause of lung cancer,
1
innumerable stud-
ies have documented the pervasive and serious conse-
quences for health of tobacco smoking, including more
than 30 Reports of the US Surgeon General published
since 1964 on the harms of smoking and environmen-
tal tobacco smoke. Cigarette smoking, which causes
almost 500,000 premature deaths each year in the
US,
2,3
is a major modifiable risk factor for cardiovas-
cular disease (CVD).
4
Varenicline is a partial agonist of a
4
b
2
nicotinic acetyl-
choline receptors
5
approved by the US Food and Drug
Administration (FDA) for marketing as a pharmacologi-
cal aid for smoking cessation. The only study of vareni-
cline for smoking cessation in subjects with stable CVD
was performed in approximately 700 patients by Rigotti
et al.
6
The results indicated that subjects who received
varenicline were more likely to quit smoking and remain
abstinent than subjects who received placebo.
6
The inci-
dence of cardiovascular (CV) mortality, all-cause mortal-
ity, CV events, or serious adverse events (AEs) did not
differ significantly between the varenicline and placebo
groups in that study; however, certain CV events were
reported more frequently in the varenicline arm.
6
Two
meta-analyses using somewhat different approaches
have also examined the occurrence of CV events in pub-
lished trials of varenicline.
7,8
Both of these analyses sug-
gested an increased incidence of CV events in patients
treated with varenicline compared with placebo but the
association was of borderline statistical significance in 1
analysis and not significant in the other. In 2011, the FDA
issued a Drug Safety Communication indicating that var-
enicline may be associated with a small increased risk of
certain CVevents in these patients
9
(ie, those with stable
CVD) and added the information about the small
numerical differences observed in the study by Rigotti
et al
6
to the prescribing information in the warning and
precautions section.
10
The FDA also requested a patient-
level meta-analysis of serious CV AEs observed in
manufacturer-sponsored, phase 24, randomized
placebo-controlled clinical trials of varenicline treatment
in smokers aged $18 years of $12 weeks duration. The
protocol for the meta-analysis was constructed in consul-
tation with the FDA. This article reports the results of
this meta-analysis, which have already been added to the
prescribing information. The strengths of our meta-
analysis are that all serious CV events were adjudicated
by an independent adjudication committee blind to treat-
ment assignment and that access to subject-level data
allowed a time-to-event analysis.
MATERIALS AND METHODS
Studies
This meta-analysis includes all Pfizer-sponsored phase
24 clinical trials of varenicline completed as of June
30, 2011 that were randomized, blinded, parallel arm,
and placebo controlled with at least 12-week treatment
duration in smokers aged $18 years. Trials conducted
in nonsmoking healthy volunteers or to obtain primar-
ily clinical pharmacology data were excluded from this
analysis.
Fifteen studies were included in the primary
analysis.
6,1124
The study by Rigotti et al
6
was con-
ducted in smokers with stable documented CVD
(other than hypertension alone) that had been diag-
nosed for at least 2 months. The other studies had
the following exclusion criteria for CVD: 1 study
excluded uncontrolled, unstable clinically signifi-
cant medical conditions including CVD;
12
2 studies
excluded any history of clinically significant CVD
and uncontrolled hypertension;
17,18
and the remain-
ing studies
11,1316,1924
excluded such history in the
past 6 months. All studies had a 12-week random-
ized treatment period except a single study that had
a 52-week treatment period.
24
The total duration of
follow-up in these studies was 4 months,
13
24
weeks,
11,12,19,2123
or 52 weeks.
6,1418,20,24,25
An additional study with a double-blind random-
ized withdrawal design
25
was included in a sensitivity
analysis. All subjects in this study received a 12-week
course of varenicline in an initial open-label phase.
Subjects successful in quitting were randomly assigned
to varenicline or placebo in double-blind fashion for an
additional 12 weeks.
25
CV events that occurred during
the blinded treatment phase were included in the sen-
sitivity analysis. Patients with any history of clinically
significant CVD in the past 6 months were excluded
from this study. Table 1 summarizes all 16 studies.
Outcome measures
CV events of interest were defined as CV death, myo-
cardial infarction (MI), stroke, unstable angina, perfor-
mance of revascularization, new onset of clinically
evident peripheral vascular disease, or worsening or
procedure for such disease. These CV events were iden-
tified prospectively in the study by Rigotti et al
6
of
patients with CVD and retrospectively for all other stud-
ies based on reported serious adverse events (SAEs),
defined as events that resulted in death, hospitalization,
disability, or birth defect, or were considered life-threat-
ening. All CV events and deaths were adjudicated by an
independent adjudication committee, which was com-
posed of 3 of the authors of this article G. W. Vetrovec,
236 Ware et al
American Journal of Therapeutics (2013) 20(3) www.americantherapeutics.com
Table 1. Characteristics of randomized placebo-controlled trials of varenicline included in the analysis.
Varenicline study
Duration of
treatment (weeks)
Duration of
study (weeks) Drug and dose
Number of
subjects treated
Mean age
(years)
Mean years
smoking
Oncken et al
18
(dose-finding study) 12 52 Varenicline 1 mg BID
titrated
129 42.2 24.1
Varenicline 1 mg BID
nontitrated
124 43.9 25.9
Varenicline 0.5 mg BID
titrated
129 43.6 25.2
Varenicline 0.5 mg BID
nontitrated
124 43.1 26.1
Placebo 121 43.3 25.6
Niaura et al
17
(flexible-dose study) 12 52 Varenicline flexible
dosing:
0.52.0 mg total daily
dose
157 41.5 24.9
Placebo 155 42.1 25.7
Gonzales et al
14
(pivotal study) 12 52 Varenicline 1 mg BID 349 42.5 24.3
Bupropion 150 mg BID 329 42.0 24.1
Placebo 344 42.6 24.7
Jorenby et al
15
(pivotal study) 12 52 Varenicline 1 mg BID 343 44.6 27.2
Bupropion 150 mg BID 340 43.0 25.5
Placebo 340 42.3 24.3
Williams et al
24
(long-term safety study) 52 52 Varenicline 1 mg BID 251 48.2 30.7
Placebo 126 46.6 29.9
Tsai et al
21
(Korea and Taiwan study) 12 24 Varenicline 1 mg BID 126 39.7 20.2
Placebo 124 40.9 22.1
Nakamura et al
16
(Japan study) 12 52 Varenicline 1 mg BID 156 39.2 20.7
Varenicline 0.5 mg BID 155 39.1 20.3
Varenicline 0.25 mg BID 153 40.4 21.0
Placebo 154 39.8 20.8
Rigotti et al
6
(CVD study) 12 52 Varenicline 1 mg BID 353 57.0 40.0
Placebo 350 56.0 39.1
Tashkin et al
20
(COPD study) 12 52 Varenicline 1 mg BID 248 57.2 40.4
Placebo 251 57.1 40.6
Wang et al
22
(China study) 12 24 Varenicline 1 mg BID 165 39.0 20.5
Placebo 168 38.5 19.6
(Continued on next page)
M
e
t
a
-
A
n
a
l
y
s
i
s
o
f
C
V
S
a
f
e
t
y
o
f
V
a
r
e
n
i
c
l
i
n
e
2
3
7
w
w
w
.
a
m
e
r
i
c
a
n
t
h
e
r
a
p
e
u
t
i
c
s
.
c
o
m
A
m
e
r
i
c
a
n
J
o
u
r
n
a
l
o
f
T
h
e
r
a
p
e
u
t
i
c
s
(
2
0
1
3
)
2
0
(
3
)
A. B. Miller, and A. Van Tosh. The committee used
prespecified criteria based on the Standardized Defini-
tions for End Point Events in Cardiovascular Trials
26
and the universal definition of MI.
27
Two composite end points were defined prospec-
tively: major adverse cardiovascular events (MACE)
that comprised CV death, nonfatal MI, and nonfatal
stroke and MACE+ that comprised MACE plus new
onset, worsening or any procedure for peripheral vas-
cular disease, hospitalization for unstable angina, and
performance of coronary revascularization. In addi-
tion, all-cause mortality was assessed.
Five of the studies,
6,11,16,19,21
including the study by
Rigotti et al in patients with CVD,
6
collected SAEs
occurring beyond 30 days after the end of blinded
treatment until the end of the study. However, only
the study by Rigotti et al reported any serious CV
events during this extended follow-up period. In the
other 10 studies, these events were collected beyond 30
days after treatment only if considered to be treatment
related by the site investigators. None of these 10 stud-
ies reported treatment-related serious CV events in the
follow-up phase. The primary analysis in this report
includes events occurring during treatment plus 30
days. The additional CV AEs observed by Rigotti
et al
6
after this interval are included only in a sensitiv-
ity analysis. The half-life of varenicline is approxi-
mately 24 hours, and it clears from the body in
approximately 1 week
28
; thus, it is supposed that var-
enicline is no longer present 30 days after the last dose,
making it unlikely that later events are varenicline
related.
Baseline CV risk
To assess the effect of baseline CV risk on the overall
results, subjects were classified as having lower or
higher CV risk. Higher CV risk was inferred if 1 or
more of the following CV risk factors was present (in
addition to smoking, an important risk factor com-
mon to all participants) at study initiation based on
the medical history and/or physical examination:
body mass index .30 kg/m
2
; a medical condition
included in the standard Medical Dictionary for Reg-
ulatory Activities (MedDRA) System Organ Classes
cardiac disorders or vascular disorders; or in the
high-level group terms cardiac and vascular disor-
ders congenital, cardiac therapeutic procedures, vas-
cular therapeutic procedures, central nervous system
vascular disorders, or high-level terms of diabetes
mellitus (excluding hyperglycemia), cardiovascular
complications of diabetes, elevated cholesterol, ele-
vated cholesterol with triglycerides, elevated trigly-
cerides, or hyperlipidemias. The remaining subjects
were classified as being at lower CV risk. T
a
b
l
e
1
.
(
C
o
n
t
i
n
u
e
d
)
C
h
a
r
a
c
t
e
r
i
s
t
i
c
s
o
f
r
a
n
d
o
m
i
z
e
d
p
l
a
c
e
b
o
-
c
o
n
t
r
o
l
l
e
d
t
r
i
a
l
s
o
f
v
a
r
e
n
i
c
l
i
n
e
i
n
c
l
u
d
e
d
i
n
t
h
e
a
n
a
l
y
s
i
s
.
V
a
r
e
n
i
c
l
i
n
e
s
t
u
d
y
D
u
r
a
t
i
o
n
o
f
t
r
e
a
t
m
e
n
t
(
w
e
e
k
s
)
D
u
r
a
t
i
o
n
o
f
s
t
u
d
y
(
w
e
e
k
s
)
D
r
u
g
a
n
d
d
o
s
e
N
u
m
b
e
r
o
f
s
u
b
j
e
c
t
s
t
r
e
a
t
e
d
M
e
a
n
a
g
e
(
y
e
a
r
s
)
M
e
a
n
y
e
a
r
s
s
m
o
k
i
n
g
W
i
l
l
i
a
m
s
e
t
a
l
2
3
(
s
c
h
i
z
o
p
h
r
e
n
i
a
s
t
u
d
y
)
1
2
2
4
V
a
r
e
n
i
c
l
i
n
e
1
m
g
B
I
D
8
4
4
0
.
2
2
3
.
7
P
l
a
c
e
b
o
4
3
4
3
.
0
2
4
.
9
B
o
l
l
i
g
e
r
e
t
a
l
1
1
(
A
f
r
i
c
a
,
M
i
d
d
l
e
E
a
s
t
,
L
a
t
i
n
A
m
e
r
i
c
a
s
t
u
d
y
1
2
2
4
V
a
r
e
n
i
c
l
i
n
e
1
m
g
B
I
D
3
9
0
4
3
.
1
2
5
.
0
P
l
a
c
e
b
o
1
9
8
4
3
.
9
2
6
.
8
R
e
n
n
a
r
d
e
t
a
l
1
9
(
f
l
e
x
i
b
l
e
q
u
i
t
d
a
t
e
s
t
u
d
y
)
1
2
2
4
V
a
r
e
n
i
c
l
i
n
e
1
m
g
B
I
D
4
8
6
4
3
.
9
2
6
.
0
P
l
a
c
e
b
o
1
6
5
4
3
.
2
2
4
.
6
F
a
g
e
r
s
t
r
o

m
e
t
a
l
1
2
(
s
m
o
k
e
l
e
s
s
t
o
b
a
c
c
o
s
t
u
d
y
)
1
2
2
6
V
a
r
e
n
i
c
l
i
n
e
1
m
g
B
I
D
2
1
3
4
3
.
9
2
0
.
3
*
P
l
a
c
e
b
o
2
1
8
4
3
.
9
2
1
.
7
*
G
a
r
z
a
e
t
a
l
1
3
(
n
i
c
o
t
i
n
e
w
i
t
h
d
r
a
w
a
l
s
t
u
d
y
)
1
2
1
6
V
a
r
e
n
i
c
l
i
n
e
1
m
g
B
I
D
5
5
3
3
.
4
1
6
.
9
P
l
a
c
e
b
o
5
5
3
3
.
8
1
6
.
8
T
o
n
s
t
a
d
e
t
a
l
2
5
,

(
r
a
n
d
o
m
i
z
e
d
w
i
t
h
d
r
a
w
a
l
s
t
u
d
y
)
2
4
5
2
V
a
r
e
n
i
c
l
i
n
e
1
m
g
B
I
D
6
0
2
4
5
.
4
2
8
.
2
P
l
a
c
e
b
o
6
0
4
4
5
.
3
2
8
.
1
*
N
u
m
b
e
r
o
f
y
e
a
r
s
o
f
s
m
o
k
e
l
e
s
s
t
o
b
a
c
c
o
u
s
e
.

S
t
u
d
y
i
n
c
l
u
d
e
d
i
n
a
s
e
n
s
i
t
i
v
i
t
y
a
n
a
l
y
s
i
s
o
f
1
6
s
t
u
d
i
e
s
.
C
O
P
D
,
c
h
r
o
n
i
c
o
b
s
t
r
u
c
t
i
v
e
p
u
l
m
o
n
a
r
y
d
i
s
e
a
s
e
;
B
I
D
,
t
w
i
c
e
d
a
i
l
y
.
238 Ware et al
American Journal of Therapeutics (2013) 20(3) www.americantherapeutics.com
Smoking cessation
Subjects in all studies were classified as quitters or
nonquitters based on the primary efficacy end point,
as reported in most of the smoking cessation studies
included. A quitter was defined as a subject who was
continuously abstinent during the last 4 weeks of treat-
ment (weeks 912 in all studies but 1 and weeks 4852
for the 1 study that treated smokers for 1 entire year
24
).
Analytical methods
For MACE and MACE+, the time to the first event was
analyzed by a stratified logrank test. Study was included
in the model as a fixed categorical variable. The relative
treatment effect was measured by the estimated hazard
ratio (HR) in the analysis of the aggregated data.
The combined estimate of the risk difference (RD) was
estimated by a weighted average of the individual study
RDs with the weights equal to the MantelHaenszel
weights. Studies with no events could not be included
in the analyses of HRs but were included in the analyses
of RD. Identical methods were used for the meta-analy-
ses within the subgroups described above.
All confidence intervals (CIs) are at the 95% level.
Heterogeneity among studies was assessed by the I
2
statistic. The primary analysis included CV events
occurring during the treatment phase or within 30
days of the end of the treatment period.
Two sensitivity analyses were conducted. The first
included MACE and MACE+ that occurred more than
30 days after the last dose of treatment up to 52 weeks
Table 2. Distribution of first MACE+ occurring in 38
unique subjects during and 30 days after blinded
treatment.
Varenicline, n Placebo, n
n (subject-years) 4190 (1314) 2812 (838)
Event
CV death 1 1
Nonfatal MI 8 3
Nonfatal stroke 4 2
Unstable angina 6 2
Peripheral vascular disease 4 1
Coronary revascularization 3 3
Total 26 12
All-cause mortality 1 2
MACE+, major CV events plus worsening or any procedure for
peripheral vascular disease, hospitalization for angina, or per-
formance of coronary revascularization.
FIGURE 1. RD analysis for MACE for treatment plus 30 days.
Meta-Analysis of CV Safety of Varenicline 239
www.americantherapeutics.com American Journal of Therapeutics (2013) 20(3)
and that were reported in the study of patients with
CVD by Rigotti et al.
6
The second expanded the anal-
ysis of MACE and MACE+ during blinded treatment
plus 30 days to include the randomized double-blind
phase of the study by Tonstad et al
25
that used a with-
drawal design.
RESULTS
A total of 4190 varenicline subjects (1314 subject-years)
and 2812 placebo subjects (838 subject-years) from 15
placebo-controlled studies (Table 1) were included in
the primary analysis.
In the 15 placebo-controlled studies, 173 potential CV
events of interest were sent for adjudication (84 prospec-
tively from the study of Rigotti et al
6
and 89 retrospec-
tively from the remaining 14 studies
1124
). Ninety-three
of these events (all SAEs) were adjudicated as MACE+.
Twenty-one subjects experienced more than 1 MACE+
event, in which case only the first event was counted in
this analysis, yielding a total of 65 subjects with 1 or
more MACE+, of which 28 were MACE. Among the
65 MACE+ subjects, 38 (26 treated with varenicline; 12
treated with placebo) experienced the first MACE+
during treatment plus 30 days and were included in
the primary analysis. For 19 of these 38 subjects (13 var-
enicline and 6 placebo), the first CV event was MACE.
The types of MACE+ events and all-cause mortality are
displayed by treatment group in Table 2.
The results of the time-to-event meta-analyses of
MACE+ and MACE are presented in Figures 14.
The summary estimates of the HR were 1.74 (95%
CI: 0.913.34; P 5 0.10) for MACE+ and 1.95 (95%
CI: 0.794.82; P 5 0.15) for MACE. There was no indi-
cation of heterogeneity across the studies (I
2
5 0 for
each end point). However, the power of the test for
heterogeneity was low because the number of events
in each study was small. The meta-analysis for abso-
lute risk difference in MACE+ is depicted in Figure 2.
Although studies with 0 events do not contribute to
the estimation of the HR, they do contribute to the
estimates of incidence rates and RDs. The estimate of
the RD for MACE+ was 0.010 events per subject-year
(95% CI: 0.002, 0.022; P 5 0.11). The corresponding
result for MACE (Figure 1) was an RD of 0.006 events
per subject-year (95% CI: 0.002, 0.015; P 5 0.16).
The study by Rigotti et al
6
contributed 17 (10 vare-
nicline and 7 placebo) of the 38 MACE+ events and 9
FIGURE 2. RD analysis for MACE+ for treatment plus 30 days.
240 Ware et al
American Journal of Therapeutics (2013) 20(3) www.americantherapeutics.com
(6 varenicline and 3 placebo) of the MACE events
included in the primary analysis. The study by Rigotti
et al
6
is the only one in which CV events were reported
beyond 30 days after the end of treatment. The 27
MACE+ and 9 MACE events occurring after the treat-
ment period plus 30 days were included in the first
sensitivity analysis. In this analysis, the overall estimate
of the HR for MACE+ was 1.56 (95% CI: 0.962.56; P 5
0.08) and the RD was 0.012 events per subject-year (95%
CI: 0.001, 0.025; P 5 0.08). The corresponding results
for MACE were an HR of 1.66 (95% CI: 0.793.49; P 5
0.18) and an RD of 0.006 events per subject-year (95%
CI: 0.003, 0.015; P 5 0.19).
The randomized withdrawal study by Tonstad
et al
25
had 2 adjudicated MACE in the varenicline
group and 0 in the placebo group. Including this study
in a second sensitivity meta-analysis yielded an HR of
1.87 for MACE+ (95% CI: 0.993.53; P 5 0.05) and 2.21
for MACE (95% CI: 0.93, 5.21; P 5 0.07).
As only 3 deaths were observed during and within
30 days of treatment, a formal meta-analysis was not
performed for this period. The mortality rate (all
cause) was higher in the placebo group [n 5 2
(0.07%)] compared with the varenicline group [n 5 1
(0.02%)] during this time period and also for the entire
study period [n 5 6 (0.14%) for varenicline and n 5 7
(0.25%) for placebo]. Of these 13 deaths, 4 were CV
deaths, with 2 occurring during the treatment phase
plus 30 days and 2 during the posttreatment follow-up
phase (1 in each treatment arm for both phases).
Baseline CV risk
Subjects were classified into lower and higher baseline
CV risk subgroups based on whether the subject had 1
or more CV risk factors in addition to smoking. Most of
the MACE+ events occurred in subjects in the higher
CV risk subgroup. Only 5 subjects in the lower CV risk
subgroup experienced MACE+ (3 in the varenicline
group and 2 in the placebo group), whereas 33 subjects
experienced MACE+ in the higher CV risk subgroup
(23 varenicline and 10 placebo). Of note, 30 of these
33 subjects had a history of CV disease (20 on vareni-
cline and 10 on placebo). In the lower CV risk group
(Table 3), the overall estimate of the HR was 1.30
(95% CI: 0.218.04; P 5 0.78) and of RD was 0.001
(95% CI: 0.007, 0.010; P 5 0.77). For the higher CV risk
group, the HR was 1.78 (95% CI: 0.883.60; P 5 0.11)
and the RD was 0.018 (95% CI: 0.005, 0.042; P 5 0.12).
Although the treatment by CV risk interaction was not
statistically significant (P 5 0.75 in the time-to-event
FIGURE 3. HR analysis for MACE for treatment plus 30 days.
Meta-Analysis of CV Safety of Varenicline 241
www.americantherapeutics.com American Journal of Therapeutics (2013) 20(3)
analysis), these stratified results indicate that the results
of the primary meta-analysis were driven by the sub-
group with higher baseline CV risk.
Smoking cessation
In a meta-analysis by quitting status, there was no evi-
dence of an association between smoking status during
the last 4 weeks of treatment and the occurrence of
MACE+. Although smoking status was determined by
the last 4 weeks of treatment, the meta-analysis
included all CV events for the entire treatment period
plus 30 days. A meta-analysis of time-to-MACE+ in
quitters and non-quitters subgroups was used to
investigate the effect of varenicline on the incidence
of MACE+ in these 2 subgroups. The HRs for subjects
who quit smoking and those who did not are given in
Table 4. The HRs indicate a higher MACE+ incidence
in the varenicline group compared with the placebo
group for subjects who did not quit but a lower
MACE+ incidence in subjects who did quit smoking
(HR 5 0.77 in quitters and 2.10 in non-quitters). Nei-
ther the subgroup HRs nor the treatment by quit
status interaction was statistically significant, but
sample sizes were small, affording limited power
for rigorous intergroup comparisons.
DISCUSSION
This pooled analysis of patient-level data from 15
placebo-controlled clinical trials of varenicline found
a nonsignificant trend toward an increased incidence
rate of both MACE and MACE+ in varenicline-trea-
ted patients. The estimated risk differences per
subject-year in the primary analysis were quite small,
0.010 for MACE+ and 0.006 for MACE, and were not
significantly different from 0. Our results are similar
to those of the previous studies, including the 2 meta-
analyses reported by Singh et al
7
and Proschaska
et al
8
as well as the study by Rigotti et al
6
in patients
with stable CVD. Only the analysis of Singh et al
7
resulted in a borderline statistically significant asso-
ciation between treatment and incidence of CV events.
The relatively low rates of CV events, small sample sizes,
and short trial durations limited the statistical power of
the studies to detect small differences in the risk of CV
outcomes. Based on the upper bound of the CI for the
RD (putatively the "worst case" outcome), the current
meta-analysis suggests that any increase in the risk of
CV events in varenicline-treated patients is unlikely to
exceed 2.2 events per 100 subject-years or approximately
FIGURE 4. HR analysis for MACE+ for treatment plus 30 days.
242 Ware et al
American Journal of Therapeutics (2013) 20(3) www.americantherapeutics.com
0.7 events per 100 patients during 12 weeks of treatment
and a 30-day posttreatment period.
Aspects of the design of the earlier studies may shed
light on appropriate interpretation of the differences
between the 3 analyses. Singh et al
7
included all serious
CV events in their analysis, including those that
occurred more than 30 days after the end of study treat-
ment and concluded that varenicline was associated
with a 72% greater risk of CV events than placebo based
on a Peto odds ratio (OR) analysis. However, the dif-
ference in CV event rates in that analysis of 14 studies
was small, with 52 of 4908 events (1.06%) in the vareni-
cline group versus 27 of 3308 events (0.82%) in the pla-
cebo group (Peto OR 5 1.72; 95% CI: 1.092.71).
7
The
authors commented that the CV risk estimates are
imprecise owing to the low event rates.
7
In letters pub-
lished in the same journal issue,
2932
writers noted that
the Peto method can perform poorly when the data are
sparse or follow-up is dissimilar in the treated and con-
trol groups, that the statistical significance of the results
was sensitive to the choice of statistical methods, and
that the absolute differences in event rates between
the varenicline and placebo groups were very small.
Also, since dropout rates were higher in the placebo
groups compared with the varenicline groups in most
of these trials, bias may have been introduced into
a meta-analysis based on number of subjects at risk.
In a second meta-analysis of 22 trials, Prochaska et al
8
found that rates of serious CV events during or up to 30
days after the end of study treatment were 34 of 5431
(0.63%) events on varenicline and 18 of 3801 events
(0.47%) on placebo. The summary estimate for the RD
of 0.27% (95% CI: 0.10, 0.63; P 5 0.15) based on all
trials was not statistically significant.
8
Furthermore,
the relative risk (1.40, 95% CI: 0.822.39; P 5 0.22),
MantelHaenszel OR (1.41, 95% CI: 0.822.42; P 5
0.22), and Peto OR (1.58, 95% CI: 0.902.76; P 5
0.11), based on 14 trials with at least 1 event, again
indicated a nonsignificant difference between the var-
enicline and placebo groups.
8
Including events occur-
ring only during the treatment phase is justified in
light of the half-life of the drug and the potential for
bias in the analyses of longer follow-up induced by
differential loss to follow-up.
In the present meta-analysis, there was no evidence of
heterogeneity of HRs between component studies. How-
ever, a stratified analysis of patients at high versus low
CV risk demonstrated that the trend toward increased
risk in the primary analysis was almost entirely attribut-
able to a difference in the event rate between varenicline-
and placebo-treated patients in the higher CV risk group.
The estimated RD for MACE+, for example, was 0.001
(95% CI: 0.007, 0.001) in the lower-risk group and 0.018
(95% CI: 0.005, 0.042) per subject-year in the higher-risk T
a
b
l
e
3
.
H
a
z
a
r
d
r
a
t
i
o
a
n
d
r
i
s
k
d
i
f
f
e
r
e
n
c
e
s
t
r
a
t
i
f
i
e
d
b
y
b
a
s
e
l
i
n
e
C
V
r
i
s
k
f
o
r
M
A
C
E
+
f
o
r
t
r
e
a
t
m
e
n
t
p
l
u
s
3
0
d
a
y
s
.
V
a
r
e
n
i
c
l
i
n
e
P
l
a
c
e
b
o
H
a
z
a
r
d
r
a
t
i
o
(
9
5
%
C
I
)
,
P
R
i
s
k
d
i
f
f
e
r
e
n
c
e
(
S
E
)
(
9
5
%
C
I
)
,
P
M
A
C
E
+
S
u
b
j
e
c
t
y
e
a
r
s
I
n
c
i
d
e
n
c
e
r
a
t
e
M
A
C
E
+
S
u
b
j
e
c
t
y
e
a
r
s
I
n
c
i
d
e
n
c
e
r
a
t
e
L
o
w
e
r
r
i
s
k
3
6
6
6
0
.
0
0
5
0
2
4
3
4
0
.
0
0
3
7
1
.
3
0
(
0
.
2
1

8
.
0
4
)
,
0
.
7
8
0
.
0
0
1
(
0
.
0
0
4
)
(

0
.
0
0
7
,
0
.
0
1
0
)
,
0
.
7
7
H
i
g
h
e
r
r
i
s
k
2
3
6
4
7
0
.
0
4
0
6
1
0
4
0
3
0
.
0
2
2
2
1
.
7
8
(
0
.
8
8

3
.
6
0
)
,
0
.
1
1
0
.
0
1
8
(
0
.
0
1
2
)
(

0
.
0
0
5
,
0
.
0
4
2
)
,
0
.
1
2
C
I
,
c
o
n
f
i
d
e
n
c
e
i
n
t
e
r
v
a
l
;
M
A
C
E
+
,
m
a
j
o
r
c
a
r
d
i
o
v
a
s
c
u
l
a
r
e
v
e
n
t
s
p
l
u
s
w
o
r
s
e
n
i
n
g
o
r
a
n
y
p
r
o
c
e
d
u
r
e
f
o
r
p
e
r
i
p
h
e
r
a
l
v
a
s
c
u
l
a
r
d
i
s
e
a
s
e
,
h
o
s
p
i
t
a
l
i
z
a
t
i
o
n
f
o
r
a
n
g
i
n
a
,
o
r
p
e
r
f
o
r
m
a
n
c
e
o
f
c
o
r
o
n
a
r
y
r
e
v
a
s
c
u
l
a
r
i
z
a
t
i
o
n
;
S
E
,
s
t
a
n
d
a
r
d
e
r
r
o
r
.
Meta-Analysis of CV Safety of Varenicline 243
www.americantherapeutics.com American Journal of Therapeutics (2013) 20(3)
group. Although the interaction between CV risk group
and study treatment did not reach statistical significance,
the relatively low background rate of CV events in the
lower-risk groups indicates that an increase in risk of CV
events during varenicline treatment would be of greater
clinical significance in patients at higher background risk,
particularly those with known CV disease. It is also note-
worthy that both all-cause mortality and deaths,
although sufficiently infrequent to preclude firm inferen-
ces, nominally occurred at a lower rate per subject-year
of exposure in the varenicline arm.
An important unresolved issue is the effect of nicotine
withdrawal on CVrisk and the possibility of confounding
due to the higher proportion of quitters in the varenicline
group. Regarding changes in risk after smoking cessation,
a recent study of ex-smokers contemplating bypass sur-
gery found that abnormal levels of gene expression of
matrix metalloproteinase (MMP) subtypes MMP-2 and
MMP-9 in saphenous veins typically associated with cig-
arette smoking persisted up to 1 year after smoking ces-
sation, suggesting increased risk of graft failure in recent
ex-smokers.
33
We attempted to address the effect of
smoking cessation on risk by stratifying subjects by smok-
ing cessation status during the last 4 weeks of the treat-
ment period. This analysis did not indicate an association
between smoking status and CV events. However, this
stratification is an imprecise measure of nicotine exposure
because of the limitations of the data on repeated quit
attempts and lapses. Because of the small numbers of
CV events and the limitations of the analysis, the issue
is not resolved by these assessments. Moreover, the stud-
ies involved in this and other meta-analyses included few
patients who underwent vein bypass grafting for coro-
nary artery disease and might have particularly benefited
from judiciously timed smoking cessation.
Although this analysis was based on 15 placebo-
controlled studies of varenicline, 48% of the meta-
analysis weight was assigned to the study by Rigotti
et al.
6
To date, that study remains the most important
individual study of the possible AEs of varenicline.
In summary, the analysis reported here provides more
robust evidence about association between varenicline
treatment and CV events than either of the previous
meta-analyses because (1) the study end points were pre-
specified; (2) the composite end point focused on ische-
mic events; (3) all serious CV events were adjudicated by
an independent adjudication committee blind to treat-
ment assignment; and (4) subject-level data were avail-
able, making it possible to conduct a time-to-first-event
analysis rather than a meta-analysis of summary data.
Nevertheless, because of the relative rarity of CV events
in these studies, of the order of 1% per subject-year, the
analysis had limited power to characterize the relative
hazard of CV events in patients taking varenicline. The
95% CI for the HR for MACE+, for example, was 0.93.3,
indicating that the data were consistent with both no
increase in risk and a tripling of risk. On the RD scale,
the 95% CI was 20.002 to 0.022. Thus, the totality of the
available evidence leaves considerable uncertainty about
the existence and magnitude of increased risk of CV
events in patients taking varenicline. However, the data
from this and earlier studies suggest that the absolute risk
of CV events in patients taking varenicline is relatively
modest. Thus, the utility of this therapy in support of
efforts at smoking cessation must be determined in each
individual patient based on the benefits of smoking ces-
sation versus the risk of CV events which, as noted, is
relatively small and may not ultimately prove to be
increased with varenicline treatment. A large (n 5 8000)
ongoing randomized controlled trial is monitoring CV
safety in smokers treated with the 3 main smoking ces-
sation pharmacotherapies, varenicline, bupropion, and
nicotine replacement therapy (www.clinicaltrials.gov
identifiers: NCT01456936 and NCT01574703) and will
provide additional data about CV risk associated with
these smoking cessation therapies when completed. More
definitive information regarding the CV risk of these
agents must await the results of that study.
ACKNOWLEDGMENTS
The authors acknowledge and thank Cristina Russ,
MD, Larry Samuels, PhD, and Theodore C. Lee, MD,
of Pfizer Inc. for their contributions to this manuscript.
Table 4. Hazard ratio for MACE+ for treatment plus 30 days in quitters and non-quitters during weeks 912.
Varenicline Placebo
Hazard ratio (95% CI) MACE+ n MACE+ n
Quitters 12 2027 4 550 0.77 (0.232.63)
Non-quitters 14 2163 8 2262 2.10 (0.894.94)
CI, confidence interval; MACE+, major cardiovascular events plus worsening or any procedure for peripheral vascular disease, hos-
pitalization for angina, or performance of coronary revascularization.
244 Ware et al
American Journal of Therapeutics (2013) 20(3) www.americantherapeutics.com
Most of the analyses were performed by Li-Jung Tseng,
PhD, Ching-Ray Yu, PhD, and David Lawrence, PhD,
of Pfizer Inc. All clinical trials were sponsored by Pfizer
Inc. Editorial support in the form of creating tables and
figures, adding and formatting references, collating
author comments for early drafts, proofreading, editing,
and formatting for journal style was provided by Alex-
andra Bound, PhD, and Abegale Templar, PhD, of UBC
Scientific Solutions and funded by Pfizer Inc.
REFERENCES
1. Doll R, Hill AB. Smoking and carcinoma of the lung;
preliminary report. Br Med J. 1950;2:739748.
2. Fiore MC, Jaen CR, Baker TB, et al. Treating Tobacco Use
and Dependence: 2008 Update. Clinical Practice Guideline.
Executive Summary. Available at: http://www.ahrq.
gov/path/tobacco.htm#clinic. Accessed January 18, 2013.
3. Centers for Disease Control and Prevention (CDC).
Annual smoking-attributable mortality, years of potential
life lost, and productivity lossesUnited States, 1997-
2001. MMWR Morb Mortal Wkly Rep. 2005;54:625628.
4. Centers for Disease Control and Prevention. The Health Con-
sequences of Smoking: A Report of the Surgeon General. Avail-
able at: http://www.cdc.gov/tobacco/data_statistics/sgr/
sgr_2004/chapters.htm. Accessed January 18, 2013.
5. Coe JW, Brooks PR, Vetelino MG, et al. Varenicline: an
alpha4beta2 nicotinic receptor partial agonist for smok-
ing cessation. J Med Chem. 2005;48:34743477.
6. Rigotti NA, Pipe AL, Benowitz NL, et al. Efficacy and
safety of varenicline for smoking cessation in patients
with cardiovascular disease: a randomized trial. Circula-
tion. 2010;121:221229.
7. Singh S, Loke YK, Spangler JG, et al. Risk of serious
adverse cardiovascular events associated with vareni-
cline: a systematic review and meta-analysis. CMAJ.
2011;183:13591366.
8. Prochaska JJ, Hilton JF. Risk of cardiovascular serious
adverse events associated with varenicline use for tobacco
cessation: systematic review and meta-analysis. BMJ.
2012;344:e2856.
9. Food and Drug Administration (FDA). FDA Drug Safety
Communication: Chantix (Varenicline) May Increase the Risk
of Certain Cardiovascular Adverse Events in Patients with Car-
diovascular Disease. Available at: http://www.fda.gov/
Drugs/DrugSafety/ucm259161.htm. Accessed January 18,
2013.
10. Food and Drug Administration (FDA). FDA Drug Safety
Communication: Chantix (Varenicline) Drug Label Now Con-
tains Updated Efficacy and Safety Information. Available at:
http://www.fda.gov/Drugs/DrugSafety/ucm264436.
htm. Accessed January 18, 2013.
11. Bolliger CT, Issa JS, Posadas-Valay R, et al. Effects of
varenicline in adult smokers: a multinational, 24-week,
randomized, double-blind, placebo-controlled study.
Clin Ther. 2011;33:465477.
12. Fagerstrm K, Gilljam H, Metcalfe M, et al. Stopping
smokeless tobacco with varenicline: randomised dou-
ble blind placebo controlled trial. BMJ. 2010;341:
c6549.
13. Garza D, Murphy M, Tseng LJ, et al. A double-blind
randomized placebo-controlled pilot study of neuro-
psychiatric adverse events in abstinent smokers treated
with varenicline or placebo. Biol Psychiatry. 2011;69:
10751082.
14. Gonzales D, Rennard SI, Nides M, et al. Varenicline, an
alpha4beta2 nicotinic acetylcholine receptor partial ago-
nist, vs sustained-release bupropion and placebo for
smoking cessation: a randomized controlled trial. JAMA.
2006;296:4755.
15. Jorenby DE, Hays JT, Rigotti NA, et al. Efficacy of vare-
nicline, an alpha4beta2 nicotinic acetylcholine receptor
partial agonist, vs placebo or sustained-release bupro-
pion for smoking cessation: a randomized controlled
trial. JAMA. 2006;296:5663.
16. Nakamura M, Oshima A, Fujimoto Y, et al. Efficacy and
tolerability of varenicline, an alpha4beta2 nicotinic ace-
tylcholine receptor partial agonist, in a 12-week, ran-
domized, placebo-controlled, dose-response study with
40-week follow-up for smoking cessation in Japanese
smokers. Clin Ther. 2007;29:10401056.
17. Niaura R, Hays JT, Jorenby DE, et al. The efficacy
and safety of varenicline for smoking cessation using
a flexible dosing strategy in adult smokers: a random-
ized controlled trial. Curr Med Res Opin. 2008;24:
19311941.
18. Oncken C, Gonzales D, Nides M, et al. Efficacy and
safety of the novel selective nicotinic acetylcholine recep-
tor partial agonist, varenicline, for smoking cessation.
Arch Intern Med. 2006;166:15711577.
19. Rennard S, Hughes J, Cinciripini PM, et al. A random-
ized placebo-controlled trial of varenicline for smoking
cessation allowing flexible quit dates. Nicotine Tob Res.
2012;14:343350.
20. Tashkin DP, Rennard S, Hays JT, et al. Effects of vareni-
cline on smoking cessation in patients with mild to mod-
erate COPD: a randomized controlled trial. Chest. 2011;
139:591599.
21. Tsai ST, Cho HJ, Cheng HS, et al. A randomized,
placebo-controlled trial of varenicline, a selective alpha4-
beta2 nicotinic acetylcholine receptor partial agonist, as
a new therapy for smoking cessation in Asian smokers.
Clin Ther. 2007;29:10271039.
22. Wang C, Xiao D, Chan KP, et al. Varenicline for smoking
cessation: a placebo-controlled, randomized study. Re-
spirology. 2009;14:384392.
23. Williams JM, Anthenelli RM, Morris CD, et al. A ran-
domized, double-blind, placebo-controlled study eval-
uating the safety and efficacy of varenicline for
smoking cessation in patients with schizophrenia or
schizoaffective disorder. J Clin Psychiatry. 2012;73:
654660.
24. Williams KE, Reeves KR, Billing CB Jr, et al. A double-
blind study evaluating the long-term safety of varenicline
Meta-Analysis of CV Safety of Varenicline 245
www.americantherapeutics.com American Journal of Therapeutics (2013) 20(3)
for smoking cessation. Curr Med Res Opin. 2007;23:
793801.
25. Tonstad S, Tnnesen P, Hajek P, et al. Effect of main-
tenance therapy with varenicline on smoking cessa-
tion: a randomized controlled trial. JAMA. 2006;296:
6471.
26. Hicks KA, Hung HMJ, Mahaffey KW, et al. Standardized
Definitions for End Point Events in Cardiovascular Trials. Draft
guidance. Available at: http://www.clinpage.com/images/
uploads/endpoint-defs_11-16-2010.pdf. Accessed January
18, 2013.
27. Thygesen K, Alpert JS, White HD, et al. Universal definition
of myocardial infarction. Circulation. 2007;116:26342653.
28. Faessel HM, Gibbs MA, Clark DJ, et al. Multiple-dose
pharmacokinetics of the selective nicotinic receptor
partial agonist, varenicline, in healthy smokers. J Clin
Pharmacol. 2006;46:14391448.
29. Samuels L. Varenicline: cardiovascular safety. CMAJ.
2011;183:14071408.
30. Squire EN. Varenicline: quantifying the risk. CMAJ. 2011;
183:14041405.
31. Takagi H, Umemoto T. Varenicline: quantifying the risk.
CMAJ. 2011;183:1404.
32. Woods DJ, Caswell MD. Varenicline: quantifying the
risk. CMAJ. 2011;183:1404.
33. Yongxin S, Wenjun D, Qiang W, et al. Heavy smoking
before coronary surgical procedures affects the
native matrix metalloproteinase-2 and matrix metal-
loproteinase-9 gene expression in saphenous vein con-
duits. Ann Thorac Surg. 2013;95:5561.
246 Ware et al
American Journal of Therapeutics (2013) 20(3) www.americantherapeutics.com