Clinical Therapeutics/Volume ], Number ], 2014

Predictors of Lapse and Relapse to Smoking in SuccessfulQuitters in a Varenicline Post Hoc Analysis inJapanese Smokers

Masakazu Nakamura, MD1; Akira Oshima, MD2; Masayuki Ohkura, PhD3;Carmen Arteaga, PhD4; and Kiyomi Suwa, BA3

1Department of Health Promotion and Prevention, Osaka Center for Cancer and Cardiovascular DiseasesPrevention, Osaka, Japan; 2Cancer Information Services, Osaka Medical Center for Cancer andCardiovascular Disease, Osaka, Japan; 3Pfizer Japan Inc, Tokyo, Japan; and 4Pfizer Inc, New York,New York

Accepted for publication March 23, 2014.http://dx.doi.org/10.1016/j.clinthera.2014.03.0130149-2918/$ - see front matter

& 2014 The Authors. Published by Elsevier HS Journals, Inc. All rightsreserved.

ABSTRACT

Background: The efficacy of the smoking-cessationagent varenicline has been reported in Asian smokers;however, few studies have investigated factors thatcontribute to lapse and relapse.

Objective: This post hoc analysis aimed to identifypredictors of smoking lapse and relapse.

Methods: This was a post-hoc analysis based on adouble-blind, placebo-controlled, randomized, parallel-group study in which Japanese smokers (aged 20–75 years) who smoked ≥10 cigarettes/day and weremotivated to quit were randomized to receive vareni-cline (0.25 mg twice daily [BID], 0.5 mg BID, 1 mgBID) or placebo for 12 weeks followed by a 40-weeknon-treatment follow-up. For inclusion in this analy-sis, participants must have been nicotine dependent(Tobacco Dependence Screener score Z5) and musthave successfully quit smoking continuously for 4weeks (weeks 9–12). Lapse was defined by answeringyes to Z1 question in the Nicotine Use Inventory.Relapse was defined by participants having smokedfor Z7 days during follow-up measured by theNicotine Use Inventory.

Results: Of the 619 randomized individuals, 515had a Tobacco Dependence Screener score of Z5, and277 quit smoking continuously from weeks 9 to 12.Approximately 75% were male, with a mean (SD)BMI of 23.0 (3.0) kg/m2. Maximum length ofcontinuous abstinence (CA) during treatment andage (both P o 0.0001) were significant predictors oflapse. Maximum CA (P o 0.0001), age (P ¼ 0.0002),Minnesota Nicotine Withdrawal Scale (MNWS) scorefor urge to smoke (P ¼ 0.0019), and having made Z1serious quit attempt (P ¼ 0.0063) were significant

] 2014

predictors of relapse. For participants with a max-imum CA of 4 to 6 weeks versus those with amaximum CA of 10 to 11 weeks, the ORs for lapseand relapse were 4.649 (95% CI, 2.071–10.434) and3.337 (95% CI, 1.538–7.239), respectively. In partic-ipants aged 21–34 years versus those aged 47–72 years, the ORs for lapse and relapse were 3.453(95% CI 1.851, 6.441) and 3.442 (95% CI 1.795,6.597), respectively. Participants with a MNWS urgeto smoke score of 2 to 4 versus 0 had an OR forrelapse of 3.175 (95% CI, 1.166–8.644). Participantshaving made Z1 versus no serious quit attempts hadan OR for relapse of 2.108 (95% CI, 1.168–3.805).

Conclusion: Shorter maximum CA and youngerage at quit attempt were associated with increased riskof lapse and relapse. Higher MNWS urge to smokescore and having made Z1 serious quit attempt wereassociated with increased relapse risk. ClinicalTrials.gov identifier: NCT00139750. (Clin Ther. 2014;]:]]]–]]]) & 2014 The Authors. Published by Elsevier HSJournals, Inc. All rights reserved.

Key words: Japan, lapse, relapse, smoking cessa-tion, varenicline.

INTRODUCTIONVarenicline tartrate is a selective α4β2-nicotinic acetyl-choline receptor partial agonist that is approved world-wide as a smoking cessation aid at a dose of 1 mg BID

1

Clinical Therapeutics

for 12 weeks starting with a 1-week up-titration.1 Theefficacy of varenicline as a smoking cessation aid hasbeen reported in a number of studies. Results from 2double-blind, randomized, placebo-controlled studiesconducted in the United States2,3 have reported thatsmoking cessation success with varenicline is signifi-cantly better compared with other smoking cessationaids (eg, bupropion sustained-release tablets) or pla-cebo. A Cochrane meta-analysis of 12,223 participantsin 15 studies, 8100 of whom used varenicline, revealedthat individuals who had received varenicline were 2.3times more likely to quit at 6 months compared withthose who had received placebo.4

A number of clinical studies have reported the efficacyof varenicline in Asian smokers.5–8 Equivalence of effi-cacy and tolerability between Japanese and US smokershas been described in a study of 618 Japanese smokers.5

This study reported a complete abstinence rate at weeks 9to 12 of 65.4% for varenicline, which compared wellwith the 44% rate reported in the US studies.2,3

Although the efficacy of pharmacologic and behav-ioral treatment for smoking cessation is well established,these methods are not always used in quit attempts. Ithas been reported that 80% of smokers who attempt toquit without assistance relapse within the first month ofabstinence, and only 3% to 5% continue to be abstinentat 6 months.9,10 Results from the US Centers for DiseaseControl and Prevention 2010 National Health InterviewSurvey revealed that although 52.4% of smokers hadmade a quit attempt in the past year, only 6.2% hadsuccessfully quit within the same time frame.11 Results ofa Cochrane review concluded that there is insufficientevidence to support a role of behavioral therapy orextended treatment with bupropion in preventing relapseto smoking. However, the report suggested thatextended treatment with varenicline may preventrelapse.12 Furthermore, the results of a studyinvestigating the use of 12 weeks of maintenancetherapy in smokers who achieved abstinence for atleast 7 consecutive days at the end of standardvarenicline treatment confirmed that individuals whoreceived maintenance treatment had significantly greatercontinuous abstinence during weeks 13 to 24 (70.5% vs49.6%; odds ratio [OR], 2.48; 95% CI, 1.95–3.16; P o0.001) and weeks 13 to 52 (43.6% vs 36.9%; OR, 1.34;95% CI, 1.06–1.69; P ¼ 0.02) compared with thosewho received placebo.13

The focus of health care professionals in treatingtobacco dependence is now directed at reducing the

2

number of patients who relapse after initially success-fully quitting smoking. However, to date few studieshave investigated the factors that contribute to re-lapse, and comparison of factors in Japanese smokersversus those from other countries has not beenconducted. In a post hoc analysis of 626 smokersfrom 2 varenicline studies conducted in the UnitedStates, predictors of relapse from week 13 (end oftreatment) to week 52 (1-year follow-up) were as-sessed for treatment-responding participants whoachieved the primary efficacy end point of 4-weekcontinuous abstinence during weeks 9 to 12. Theresults revealed that those who did not initiate con-tinuous abstinence until the final 4 weeks of thetreatment period were almost 5 times more likely torelapse (OR, 4.92; 95% CI, 2.77–8.97; P o 0.001)compared with those who initiated continuous absti-nence at the end of week 1.14 Another studyconducted to assess whether smokers who achievedabstinence later in the original course of treatmentwould benefit from receiving an extended course ofvarenicline15 confirmed that the 12-month quit suc-cess rates decreased with increasing delay in initialquitting, with quit rates of 54.9% for those quitting atweek 1 of open-label treatment and 5.7% for thosequitting at week 11.

This post hoc analysis was conducted to identifypredictors of smoking lapse or relapse during thenontreatment follow-up period based on a double-blind, placebo-controlled, randomized, parallel-groupstudy (ClinicalTrials.gov identifier: NCT00139750),5

which was initially conducted to assess the efficacyand tolerability of varenicline in Japanese smokers.

Currently, it is not clear whether research con-ducted on smoking cessation with Western individualsalso applies to Japanese individuals. Although thereare no significant differences in the metabolism andefficacy of varenicline in Japanese and Western indi-viduals,8 relapse in Japanese smokers may beinfluenced by other factors. In Japan, unlike in manywesternized countries, smoke-free environments arenot mandated with penalties by law in indoor publicplaces, in workplaces, or on public transport, andthere are no legal restrictions on advertising. Tobaccois also relatively less expensive in Japan comparedwith many Western countries. In addition, there maybe differences in the metabolism of nicotine, theseverity of nicotine dependence, and other problemsrelated to smoking cessation, depending on race.

Volume ] Number ]

M. Nakamura et al.

This investigation was conducted to identify factorsthat could contribute to smoking relapse in Japaneseindividuals. It focused on identifying predictors andcharacteristics of smoking relapse, such as the use of asmoking cessation aid and a longer period of supportduring follow-up, which may help provide an effectivestrategy to prevent future relapse.

PATIENTS AND METHODSStudy Design

This was a post hoc analysis based on a double-blind, placebo-controlled, randomized, parallel-groupstudy, which has been previously reported.5 A total of619 Japanese smokers aged 20 to 75 years, whosmoked at least 10 cigarettes per day and weremotivated to quit smoking (indicated in theinformed consent form), were randomized to receivevarenicline or placebo. The number of days ofcigarette smoking since the previous contact waschecked at each visit or at each telephone contactduring the nontreatment follow-up period or at theearly termination visit if individuals discontinued thestudy early during the follow-up phase.

Objectives and DefinitionsThe primary objective of this analysis was to

identify predictors of smoking lapse and relapseduring the nontreatment follow-up period amongindividuals who reported being continuously abstinentduring the last 4 weeks of the treatment period. Thesecondary objective was to investigate the time fromweek 13 to smoking relapse for individuals whoreported being continuously abstinent during the last4 weeks of the treatment period, stratified by levels ofa predictor. Lapse was defined by having an exhaledcarbon monoxide measurement of 410 ppm (at clinicvisits) during the nontreatment period (ie, fromweek 13 through week 52) or answering yes to atleast one of the questions about smoking in theNicotine Use Inventory (NUI): “Has the subjectsmoked any cigarettes (even a puff) since the lastcontact?” or “Has the subject used any other tobaccoproducts since the last contact?” Relapse was definedby an individual having smoked for Z7 days duringthe nontreatment follow-up period as measured by theNUI. Time to lapse or relapse was defined as the time(in weeks) from the beginning of the nontreatmentfollow-up period (ie, week 13) to the first visit whenthe individual’s lapse or relapse was observed. Those

] 2014

who didn’t lapse or relapse were censored at week 52.For individuals who discontinued follow-up visitsbefore lapsing, time to lapse or relapse was definedas the time from week 13 to the week correspondingto the first missing visit that occurred after the date ofdiscontinuation.

ParticipantsParticipants were Japanese smokers (aged 20–75

years) who had participated in an earlier smokingcessation study investigating varenicline5; had beendiagnosed as being nicotine dependent with a score ofZ5 (range, 0–10) on the Tobacco DependenceScreener (TDS),16 which defines the criteria used todetermine which smokers are permitted access to thesmoking cessation treatment service that is reimbursedby public health insurance in Japan; and hadsuccessfully quit smoking continuously for 4 weeksfrom weeks 9 to 12. Individuals were excluded if theyhad significant cardiovascular disease, uncontrolledhypertension, severe chronic obstructive pulmonarydisease, or a history of cancer. Individuals who hadquit smoking continuously for 4 weeks from weeks 9to 12 of the smoking cessation study were included inthe analysis for the primary and secondary objectives.

Predictors of RelapsePotential predictors of relapse included age, sex,

body mass index (BMI), age at onset of smoking,number of cigarettes smoked per day, longest absti-nence in previous year, living with a smoker, frequentcontact with a smoker, number of serious quitattempts before study entry, pack-years smoked, base-line Fagerström Test for Nicotine Dependence score,baseline serum cotinine level, baseline weekly alcoholconsumption, all domains of the Minnesota NicotineWithdrawal Scale (MNWS; negative affect, insomnia,urge to smoke, restlessness, and increased appetite) atweek 12, maximum length of continuous abstinence(CA) during the treatment phase, and quit pattern,defined as whether an individual could initiate absti-nence on or before the target quit date.

Statistical AnalysisModel building was based on a logistic regression

model with forward selection, backward elimination,and stepwise selection used to select potential predictorsof lapse and relapse.17,18 A model validation based ondata resampling using 70% of randomly selected data

3

Clinical Therapeutics

was performed to assess the robustness of the modelthat resulted from the process described above.

Treatment group (0.25 mg BID, 0.5 mg BID, or1 mg BID of varenicline or placebo) was also consideredfor inclusion in the initial set of predictors and in thereduced model. However, because the treatment groupwas not identified as a predictor during the modelbuilding phase and neither affected the estimates in thereduced models, treatment group was subsequentlydropped from inclusion for further consideration as apredictor. All potential predictors were descriptivelysummarized by status: smoking relapse (lapser or non-lapser or relapser or nonrelapser). Odds ratios (ORs)and 2-sided 95% CIs were calculated for identifiedpredictors. Time to lapse or time to relapse was assessedusing Kaplan-Meier plots.19 Statistical hypotheses werenot tested because of the anticipated small sample sizes.

RESULTSParticipant Characteristics

The disposition of the participants used in thissubgroup analysis is shown in Figure 1. A total of 619individuals were enrolled and randomized to receivevarenicline (0.25 mg BID, 0.5 mg BID, or 1 mg BID)or placebo, and 618 participants received eithervarenicline or placebo in the original clinical trial5;577 (93.4%) completed treatment, and 510 (82.5%)

619 Individuals

1 Not treated

153 Randomized tovarenicline 0.25 mg BID

128 Individuals with TDS ≥5

70 Individuals achieved Week

9–12 CA

71 Individuals achieved

Week 9–12 CA

277 Individuals ithis subgroup

58 Individuals failed Week 9–12 CA

57 Individuals failed Week 9–12 CA

25 Individualswith TDS <5

128 Individualswith TDS ≥5

27 Individualswith TDS <5

156 Randomized tovarenicline 0.5 mg BID

Figure 1. Disposition of the study participants incluabstinence; TDS ¼ Tobacco Dependence Scre

4

completed the study. Most (approximately 75%) weremale, with a mean (SD) BMI of 23.0 (3.0) kg/m2. Atotal of 515 participants had a score of Z5 on theTDS. In total, 277 participants quit smokingcontinuously for 4 weeks from weeks 9 to 12 of thestudy and were included in this subgroup analysis.Twelve participants discontinued the study in thenontreatment follow-up period (8 because of adverseevents, 1 because of death [traffic incident unrelated totreatment], and 3 because of other reasons [no regularfollow-up hospital visits]). These individuals wereclassified as lapsers or relapsers.

Predictors of Lapse or RelapseA summary of potential predictors by lapse and

relapse status is presented in Table I. Nonlapsers andnonrelapsers were generally older and had a longermaximum CA compared with lapsers and relapsers.Maximum CA and age were identified as predictors ofboth lapse and relapse. In addition, MNWS score forurge to smoke and having made Z1 serious quitattempt were identified as predictors of relapse (Table II).When analyzed as categorical variables, shortermaximum CA and younger age at quit attempt wereidentified as being associated with increased risk oflapse and relapse (Table III). In addition, higherMNWS urge to smoke score was associated with

enrolled

85 Individuals achieved

Week 9–12 CA

ncluded inanalysis

51 Individuals achieved

Week 9–12 CA

45 Individuals failed Week

9–12 CA

78 Individualsfailed Week

9–12 CA

130 Individualswith TDS ≥5

26 Individuals with TDS <5

129 Individualswith TDS ≥5

25 Individualswith TDS <5

156 Randomized tovarenicline 1 mg BID

154 Randomized to placebo

ded in the subgroup analysis. CA ¼ continuousener.

Volume ] Number ]

Table I. Predictors of lapse and relapse: descriptive statistics by status

Potential Predictor

Lapse Relapse

Lapsers

(n ¼ 130)

Nonlapsers

(n ¼ 147)

Relapsers

(n ¼ 113)

Nonrelapsers

(n ¼ 164)

Demographic characteristicsAge, mean (SD), y 38.0 (12.2) 44.4 (12.0) 38.0 (12.3) 43.8 (12.0)Male, No. (%) 95 (73.1) 114 (77.6) 82 (72.6) 127 (77.4)Body mass index, mean (SD), kg/m2 23.2 (3.0) 23.2 (3.0) 23.3 (3.0) 23.1 (3.0)

SmokingAge at onset of smoking, mean (SD), y 18.7 (3.1) 18.6 (2.4) 18.7 (2.8) 18.7 (2.8)Length of smoking, mean (SD), y 18.9 (11.3) 25.3 (11.5) 18.9 (11.4) 24.6 (11.6)Cigarettes per day in the past month, mean (SD) 22.8 (7.8) 23.3 (9.6) 22.9 (7.9) 23.2 (9.4)Longest abstinence in previous year, No. (%)

0 days 90 (69.2) 121 (82.3) 77 (68.1) 134 (81.7)1–30 days 37 (28.5) 23 (15.6) 33 (29.2) 27 (16.5)430 days 3 (2.3) 3 (2.0) 3 (2.7) 3 (1.8)

Lives with a smoker, No. (%) 41 (31.5) 54 (36.7) 38 (33.6) 57 (34.8)Frequent contact with smoker, No. (%) 120 (92.3) 131 (89.1) 104 (92.0) 147 (89.6)Made Z1 serious quit attempt, No. (%) 98 (75.4) 93 (63.3) 89 (78.8) 102 (62.2)Pack-years, mean (SD) 23.0 (18.0) 30.6 (21.8) 23.1 (18.0) 29.8 (21.6)FTND score, mean (SD) 5.2 (2.0) 5.3 (2.1) 5.3 (2.0) 5.3 (2.0)Baseline cotinine, mean (SD), ng/mL 261.0 (160.9) 249.9 (155.9) 262.5 (157.7) 250.1 (158.6)

Baseline weekly alcohol intake, mean (SD), U 18.8 (31.3) 16.7 (25.9) 19.0 (31.8) 16.8 (26.0)MNWS and length of abstinence during treatmentphase

MNWS,* mean (SD)Negative affect 0.1 (0.3) 0.1 (0.3) 0.2 (0.4) 0.1 (0.3)Insomnia 0.4 (0.8) 0.3 (0.6) 0.4 (0.8) 0.3 (0.6)Urge to smoke 0.5 (0.8) 0.4 (0.7) 0.6 (0.8) 0.4 (0.6)Restlessness 0.1 (0.4) 0.1 (0.3) 0.1 (0.5) 0.1 (0.3)Increased appetite 1.1 (1.2) 1.1 (1.3) 1.1 (1.3) 1.1 (1.3)

Quit pattern (immediate), No. (%) 51 (39.2) 85 (57.8) 42 (37.2) 94 (57.3)Maximum CA,† mean (SD), wk 8.7 (2.5) 9.9 (1.7) 8.7 (2.5) 9.8 (1.8)

CA ¼ continuous abstinence; FTND ¼ Fagerström Test for Nicotine Dependence; MNWS ¼ Minnesota NicotineWithdrawal Scale.*At week 12.†Maximum CA during treatment phase.

M. Nakamura et al.

increased risk of relapse (Table III). Table III detailsthe ORs generated by analysis using predictors ascategorical values. For participants with a maximumCA of 4 to 6 weeks versus those with a maximum CAof 10 to 11 weeks, the ORs for lapse and relapse were4.649 (95% CI, 2.071–10.434) and 3.337 (95% CI,1.538–7.239), respectively. In participants aged 21–34

] 2014

years versus those aged 47–72 years, the ORs for lapseand relapse were 3.453 (95% CI 1.851, 6.441) and3.442 (95% CI 1.795, 6.597), respectively. Similarly,those with a MNWS urge to smoke score of 2 to 4 hadan OR for relapse of 3.175 (95% CI, 1.166–8.644)compared with those with a score of 0. Individualshaving made Z1 serious quit attempt versus no

5

Table II. Identified predictors of lapse and relapse using a logistic regression model.

Identified predictor* Odds ratio (95% CI) P value

LapseMaximum CA 0.762 (0.673–0.863) o0.0001Age 0.958 (0.938–0.978) o0.0001

RelapseMaximum CA 0.786 (0.696–0.887) o0.0001Age 0.959 (0.938–0.981) 0.0002MNWS urge to smoke 1.798 (1.241–2.604) 0.0019Made Z1 serious quit attempt 2.278 (1.262–4.113) 0.0063

CA ¼ continuous abstinence; MNWS ¼ Minnesota Nicotine Withdrawal Scale.*All predictors were continuous variables.

Clinical Therapeutics

serious quit attempt had an OR for relapse of 2.108(95% CI, 1.168–3.805). Figure 2 presents these resultsas Kaplan-Meier plots. The Kaplan-Meier plots revealthat individuals with high-risk predictors tend torelapse to smoking earlier compared with those withlow-risk predictors.

Table III. Identified predictors as categorical variables.

Identified predictor Lapser

Maximum CA, wk4–6 4.649 (2.071–7–9 2.342 (1.269–10–11 —

Age, y21–34 3.453 (1.851–35–46 1.553 (0.828–47–72 —

MNWS urge to smoke score0 —

1 —

2–4 —

Made Z1 serious quit attemptNo —

Yes —

CA ¼ continuous abstinence; MNWS ¼ Minnesota Nicotine W

6

DISCUSSIONThis post hoc study was conducted to identify pre-dictors of smoking lapse and relapse in a populationof Japanese smokers treated with varenicline. It hasbeen reported that any smoking after the target quitdate is predictive of relapse.20,21 The results of this

Odds ratio (95% CI)

s Relapsers

10.434) 3.337 (1.538–7.239)4.320) 2.474 (1.315–4.654)

—

6.441) 3.442 (1.795–6.597)2.913) 1.845 (0.943–3.609)

—

—

1.975 (1.097–3.556)3.175 (1.166–8.644)

—

2.108 (1.168–3.805)

ithdrawal Scale.

Volume ] Number ]

1.0

Lapse byMaximum CA

Relapse byMaximum CA

0.8

0.6La

pse

Dis

trib

utio

nFu

nctio

n0.4

0.00 10 20 30

Study Week40 50

0.2

1.0

Lapse by Age

0.8

0.6

Laps

e D

istr

ibut

ion

Func

tion

0.4

0.00 10 20 30

Study Week40 50

0.2

1.0

Relapse by MNWS Urge to Smoke

0.8

0.6

Rel

apse

Dis

trib

utio

nFu

nctio

n

0.4

0.00 10 20 30

Study Week40 50

0.2

1.0

0.8

0.6

Rel

apse

Dis

trib

utio

nFu

nctio

n

0.4

0.00 10 20 30

Study Week40 50 60

Relapse by Made > = 1 Serious quit attempt

0.2

1.0

Relapse by Age

0.8

0.6

Rel

apse

Dis

trib

utio

nFu

nctio

n

0.4

0.00 10 20 30

Study Week40 50

0.2

1.0

0.8

0.6

Rel

apse

Dis

trib

utio

nFu

nctio

n

0.4

0.00 10 20 30

Study Week40 50

0.24–67–910–11

21–3435–4647–72

21–3435–4647–72

4–67–910–11

012–4

YesNo

Figure 2. Kaplan-Meier plots by maximum continuous abstinence (CA) (A), age (B), Minnesota NicotineWithdrawal Scale (MNWS) urge to smoke (C), and Z1 serious quit attempt (D).

M. Nakamura et al.

study revealed that when analyzed as categoricalvariables, shorter maximum CA and younger age atquit attempt were identified as being associated withincreased risk of lapse and relapse. In addition, higherMNWS urge to smoke score was associated with anincreased risk of relapse. The risk of lapse for smokerswith a maximum CA of 4 to 6 weeks was 4.6 timeshigher than with a maximum CA of 10 to 11 weeks.These results are in agreement with those from apooled analysis of 2 registration studies for vareni-cline, which revealed that maximum CA during thetreatment phase of the study and the longest period ofabstinence in the year before study entry were pre-dictors of relapse.14 Hajek et al15 reported that thelater in the 12-week treatment period with varenicline

] 2014

smokers obtained abstinence, the less likely they wereto remain abstinent at 1 year.

Individuals whose maximum CA was 4 to 6 weeksstarted their CA nearer the end of the treatment phasethan those whose maximum CA was 10 to 11 weeks.To prevent relapse, longer-term treatment after theinitial quit date would be required. Tonstad et al13

reported that the use of 12 weeks of maintenancetherapy with varenicline in smokers who achievedabstinence for at least 7 consecutive days at the end ofstandard varenicline treatment was associated withsignificantly improved quit rates compared withplacebo. Hajek et al15 reported that extendedtreatment with varenicline prevented relapse in latequitters (OR, 1.7; 95% CI, 1.2–2.4; P ¼ 0.0015) but

7

Clinical Therapeutics

not in early quitters (OR, 1.1; 95% CI, 0.8–1.5; P ¼0.6995). They also concluded that smokers shouldhave the best chance of successfully quitting if theytook varenicline for the full 12-week treatment periodwhile they are abstinent because this would permitsufficient time for neuroadaptation. Another pooledanalysis of the 2 varenicline registration studies inwhich smokers received varenicline, bupropion sus-tained release, or placebo2,3 revealed a positive corre-lation between adherence to treatment and tobaccoabstinence.22 The authors reported that age, numberof cigarettes smoked per day, and week 2 abstinencewere significant predictors of adherence for alltreatment groups. Our study also found thatsmokers aged 21 to 34 years were more likely torelapse compared with older smokers. These resultsare in agreement with a number of other studiesreporting that older age is linked to a positivecessation outcome.23–25

To our knowledge, this is the first analysis toidentify predictors of smoking lapse and relapse inJapanese smokers. This study investigated predictorsnot only for lapse to smoking but also for relapse.Maximum CA and age were identified as predictors ofboth lapse and relapse; however, MNWS urge tosmoke score and having made Z1 serious quitattempt were not identified as predictors of lapse.

Lapse was defined as a break in abstinence, whetheror not it led to a full relapse. However, if the lapseoccurred for 47 days during the follow-up period (ie,280 days), the event was classified as a relapse. In thisstudy, 113 of 130 patients (86.9%) who lapsed alsorelapsed. Therefore, it is not surprising to identify thesame predictors of lapse and relapse. However, urge tosmoke and Z1 serious quit attempt were identified aspredictors of relapse but not of lapse.

Several studies have examined the association be-tween lapse and relapse and the urge to smoke. Forexample, West et al26 found that smokers who hadfrequent urges to smoke were likely to relapse in theshort term. Killen et al27 found that postquit craving wassignificantly associated with abstinence at the 2-monthfollow-up and the intensity of craving was associatedwith the time to relapse. Swan et al28 revealed thatcraving was a robust predictor of relapse and a higherlevel of craving was associated with a shorter time torelapse. Brown et al29 reported that relapsers who hadnever sustained any previous quit attempt for 424hours tended to feel a greater urge to smoke compared

8

with those who had sustained a quit attempt lasting forZ3 months.

The results of this study for the urge to smoke areconsistent with previous studies. Reducing the urge tosmoke at the beginning of smoking cessation isimportant to achieve long-term cessation. Recently,Foulds et al30 reported the results of the effects ofvarenicline on nicotine withdrawal symptoms from apooled analysis of 8 randomized, placebo-controlledclinical trials of varenicline. The urge to smoke waslower for varenicline than placebo.

The effect of the past quit attempt on lapse orrelapse has found conflicting results in earlier studies.These different results may be related to the character-istics of the participants (ie, all current smokers or quitattempters only) and the definition of a past quitattempt used in the various analyses. For example,Hagimoto et al31 found that one of the predictors ofachieving abstinence successfully in a general Japanesepopulation was never having tried to quit before. Inboth this study and the study by Hagimoto et al,31 thestudy populations were limited to smokers who hadmade a quit attempt during the follow-up period (quitattempters). Hyland et al32 investigated predictors ofsmoking cessation in quit attempters and all currentsmokers. They found that a past quit attempt was nota significant predictor in quit attempters, whereas itwas negatively associated with smoking abstinence inall current smokers, including quit attempters. Incontrast, Hymowitz et al23 found that a past quitattempt was a positive predictor of quitting success,but this was in an analysis of current smokers (ie,using the different populations from the previous 2studies and this study).

With regard to the inconsistent effects of a past quitattempt on the results from this study and those ofHagimoto et al31 and Hyland et al,32 it should be takeninto consideration that the definition of a past quitattempt was different among the various studies. A pastquit attempt during a lifetime was defined in thepresent study and in the study by Hagimoto et al,31

whereas in the study by Hyland et al,32 a past quitattempt was defined as being during the previous year.

Several other studies have also reported that a pastquit attempt was positively associated with making afurther quit attempt.33–35 However, a past quitattempt was also negatively associated with makinga further quit attempt in the study by Hyland et al.32

They reported that short previous past attempts

Volume ] Number ]

M. Nakamura et al.

(o1 week) were associated with reduced success,whereas longer attempts (Z6 months) wereassociated with increased success compared with noprevious attempt. These short failure attempts close tothe start of the study might have caused the smokersto lose their confidence, which reduced theirlikelihood of a further quit attempt. As Hagimotoet al31 pointed out, high motivation and past quitattempts were considered to be expressions of thesmokers’ eagerness to quit smoking. A past quitattempt was also a marker of unknown attributes orcircumstances, which made it hard to quit smokingbecause it indicated that the smokers did not succeedin quitting smoking despite their eagerness to stop.

The present study used a subgroup of patients (n ¼277) from an earlier clinical trial.5 As a result, theparticipants may have been subject to bias, and thesample size was small. Therefore, further studies willbe required to confirm these results. Furthermore, thenumbers of lapsers but not relapsers was very small (n¼ 17). Additional analyses might be able to identifymore clearly the number of pure lapsers and the effectof the urge to smoke and past serious quit attempt onsmoking relapse.

CONCLUSIONThe results of this analysis suggest that shortermaximum CA and younger age at quit attempt wereassociated with increased risk of lapse and relapse.Higher MNWS urge to smoke score was also asso-ciated with an increased risk of relapse. A longertreatment duration with varenicline after the initialquit date could help prevent these relapses to smoking.

ACKNOWLEDGMENTSEditorial support was provided by Helen Jones, PhD,and Sarah Knott of Engage Scientific Solutions andfunded by Pfizer Inc. All authors were involved in thedesign of the study, collection and interpretation ofdata, decision to submit the article for publication,drafting the article, reviewing the article for importantintellectual content, and approving the article forsubmission. Drs. Arteaga and Oshima were alsoresponsible for conducting the data analyses.

CONFLICTS OF INTERESTMasayuki Ohkura, Carmen Arteaga, and KiyomiSuwa are employees of and shareholders in PfizerInc. Masakazu Nakamura has a research contract

] 2014

with Pfizer Japan Inc (Tokyo, Japan) for clinical trialof varenicline and has received a Medical EducationGrant from Pfizer Japan Inc for the Japan Stop-Smoking Training Outreach Project (J-STOP) as amember of the Japan Medical-Dental Association forTobacco Control. Akira Oshima has researchcontracts with Pfizer Japan Inc (Tokyo, Japan) forclinical trials of varenicline and, as the president of theJapan Medical-Dental Association for TobaccoControl, has received grants from the PfizerFoundation Global Health Partnership (New York,New York) during 2008 to 2010 and from PfizerJapan Inc (Tokyo, Japan) during 2011 to 2012 toconduct J-STOP.

REFERENCES1. Pfizer Inc. Chantix (varenicline tartrate) US prescribing

information. http://labeling.pfizer.com/ShowLabeling.aspx?id=557. Accessed October 28, 2013.

2. Gonzales D, Rennard SI, Nides M, et al. Varenicline, analpha4beta2 nicotinic acetylcholine receptor partial ago-nist, vs sustained-release bupropion and placebo forsmoking cessation: a randomized controlled trial. JAMA.2006;296:47–55.

3. Jorenby DE, Hays JT, Rigotti NA, et al. Efficacy ofvarenicline, an alpha4beta2 nicotinic acetylcholine recep-tor partial agonist, vs placebo or sustained-release bu-propion for smoking cessation: a randomized controlledtrial. JAMA. 2006;296:56–63.

4. Cahill K, Stead LF, Lancaster T. Nicotine receptor partialagonists for smoking cessation. Cochrane Database Syst Rev.2012(4):CD006103.

5. Nakamura M, Oshima A, Fujimoto Y, et al. Efficacy andtolerability of varenicline, an α4β2 nicotinic acetylcholinereceptor partial agonist, in a 12-week, randomized,placebo-controlled, dose-response study with 40-weekfollow-up for smoking cessation in Japanese smokers.Clin Ther. 2007;29:1040–1056.

6. Tsai ST, Cho HJ, Cheng HS, et al. A randomized, placebo-controlled trial of varenicline, a selective α4β2 nicotinicacetylcholine receptor partial agonist, as a new therapyfor smoking cessation in Asian smokers. Clin Ther.2007;29:1027–1039.

7. Wang C, Xiao D, Chan KP, et al. Varenicline for smokingcessation: a placebo-controlled, randomized study. Respir-ology. 2009;14:384–392.

8. Fagerström K, Nakamura M, Cho HJ, et al. Vareniclinetreatment for smoking cessation in Asian populations: apooled analysis of placebo-controlled trials conducted insix Asian countries. Curr Med Res Opin. 2010;26:2165–2173.

9

Clinical Therapeutics

9. Hughes JR. Effects of abstinencefrom tobacco: valid symptoms andtime course. Nicotine Tob Res.2007;9:315–327.

10. Hughes JR, Keely J, Naud S. Shapeof the relapse curve and long-termabstinence among untreated smok-ers. Addiction. 2004;99:29–38.

11. Centers for Disease Control andPrevention (CDC). Quitting smok-ing among adults–United States,2001–2010. MMWR Morb Mortal

Wkly Rep. 2011;60:1513–1519.12. Hajek P, Stead LF, West R, Jarvis

M, Lancaster T. Relapse preventioninterventions for smoking cessa-tion. Cochrane Database Syst Rev.2009(1):CD003999

13. Tonstad S, Tonnesen P, Hajek P,et al. Effect of maintenance therapywith varenicline on smoking cessa-tion: a randomized controlled trial.JAMA. 2006;296:64–71.

14. Heffner JL, Lee TC, Arteaga C, An-thenelli RM. Predictors of post-treatment relapse to smoking insuccessful quitters: pooled data fromtwo phase III varenicline trials. DrugAlcohol Depend. 2010;109:120–125.

15. Hajek P, Tonnesen P, Arteaga C,Russ C, Tonstad S. Varenicline inprevention of relapse to smoking:effect of quit pattern on responseto extended treatment. Addiction.2009;104:1597–1602.

16. Kawakami N, Takatsuka N, Inaba S,Shimizu H. Development of ascreening questionnaire for to-bacco/nicotine dependence accord-ing to ICD-10, DSM-III-R, and DSM-IV. Addict Behav. 1999;24:155–166.

17. Agresti A. Categorical Data Analy-sis. 2nd ed. New York, NY: JohnWiley & Sons Inc; 2002.

18. Hosmer DW, Lemeshow S. AppliedLogistic Regression. 2nd ed. New York,NY: John Wiley & Sons Inc; 2000.

19. Kaplan E, Meier P. Nonparametricestimation from incomplete obser-vations. J Am Stat Ass. 1958;53:457–481.

20. Higgins ST, Heil SH, Dumeer AM,et al. Smoking status in the initial

10

weeks of quitting as a predictor ofsmoking-cessation outcomes inpregnant women. Drug Alcohol De-pend. 2006;85:138–141.

21. Dale LC, Glover ED, Sachs DP, et al.Bupropion for smoking cessation:predictors of successful outcome.Chest. 2001;119:1357–1364.

22. Hays JT, Leischow SJ, Lawrence D,Lee TC. Adherence to treatmentfor tobacco dependence: associa-tion with smoking abstinence andpredictors of adherence. Nicotine

Tob Res. 2010;12:574–581.23. Hymowitz N, Cummings KM, Hy-

land A, et al. Predictors of smokingcessation in a cohort of adultsmokers followed for five years. TobControl. 1997;6(suppl 2):S57–S62.

24. Velicer WF, Redding CA, Sun X,Prochaska JO. Demographic varia-bles, smoking variables, and out-come across five studies. Health

Psychol. 2007;26:278–287.25. Murray RP, Gerald LB, Lindgren

PG, et al. Characteristics of par-ticipants who stop smoking andsustain abstinence for 1 and 5years in the Lung Health Study.Prev Med. 2000;30:392–400.

26. West RJ, Hajek P, Belcher M.Severity of withdrawal symptomsas a predictor of outcome of anattempt to quit smoking. Psychol

Med. 1989;19:981–985.27. Killen JD, Fortmann SP, Newman

B, Varady A. Prospective study offactors influencing the develop-ment of craving associated withsmoking cessation. Psychopharma-

cology (Berl). 1991;105:191–196.28. Swan GE, Ward MM, Jack LM.

Abstinence effects as predictors

of 28-day relapse in smokers. Ad-dict Behav. 1996;21:481–490.

29. Brown RA, Lejuez CW, Kahler CW,Strong DR. Distress tolerance andduration of past smoking cessa-tion attempts. J Abnorm Psychol.2002;111:180–185.

30. Foulds J, Russ C, Yu CR, et al. Effectof varenicline on individual nicotinewithdrawal symptoms: a combinedanalysis of eight randomized,placebo-controlled trials. Nicotine

Tob Res. 2013;15:1849–1857.31. Hagimoto A, Nakamura M, Mor-

ita T, Masui S, Oshima A. Smokingcessation patterns and predictorsof quitting smoking among theJapanese general population: a 1-year follow-up study. Addiction.2010;105:164–173.

32. Hyland A, Borland R, Li Q, et al.Individual-level predictors of cessa-tion behaviours among participantsin the International Tobacco Control(ITC) Four Country Survey. Tob Con-trol. 2006;15(suppl 3): iii83–94.

33. Borland R, Owen N, Hill D, Scho-field P. Predicting attempts andsustained cessation of smokingafter the introduction of work-place smoking bans. Health Psychol.1991;10:336–342.

34. Hellman R, Cummings KM,Haughey BP, Zielezny MA, O’SheaRM. Predictors of attempting andsucceeding at smoking cessation.Health Educ Res. 1991;6:77–86.

35. Venters MH, Kottke TE, Solberg LI,Brekke ML, Rooney B. Depend-ency, social factors, and the smok-ing cessation process: the doctorshelping smokers study. Am J Prev

Med. 1990;6:185–193.

Address correspondence to: Masakazu Nakamura, MD, Department ofHealth Promotion and Prevention, Osaka Center for Cancer and Car-diovascular Diseases Prevention, 1-3-2, Nakamichi Higashinari-ku,Osaka, 537-0025 Japan. E-mail: masa12masa12@m3.dion.ne.jp

Volume ] Number ]