During initial phone screens, 82 eligible participants scheduled a medical screening visit, of which 53 attended (64%), and 41 were determined eligible and expressed interest in participating in the study. For the laboratory sessions, 29 participants completed the first laboratory session, and of those, 23 participants returned for the follow up session (80%), which is consistent with previous attrition rates and similar to our projection for this study. Seventeen of the 23 participants (3 female) completed all visits and assessments and are the basis of our analysis below. The majority of participants described themselves as Caucasian (94%), 53% had never been married, more than half (59%) had a college degree, and 53% were employed at least part-time. Participants ranged from 22 to 64 years old (mean=43 years; SD=15) and had an average body mass index (BMI) of 26.1 (SD=5.8). Participants reported that on average they smoked 18 cigarettes per day (SD=5.6), began smoking at age 17 years (SD=7), and were moderately nicotine dependent (FTND mean=5.1, SD=2). At the eligibility screening visit, the average carbon monoxide (CO) breath level was 26 ppm (SD=10.6).
As a result of the high rate of attrition, we conducted preliminary analyses comparing participants who completed the study to those who did not. Compared to those who completed the study, participants who were lost to attrition tended to be younger (mean=35 years, SD=12) and have lower baseline CO levels (mean=20.4, SD=11), F(1,40)=3.6 and 3.1, p=0.06 and 0.09, respectively. The two groups did not differ on any other characteristic.
Effect of treatment on behavior
depicts mean (SE) for cigarettes per day for all study visits by drug phase. For cigarettes per day, there was an overall significant drug × time interaction, Huynh-Feldt F(5,65)=3.6, p=0.006, ηp2=0.22. Separate contrasts revealed that the drug × time linear interaction was significant, F(1,13)=17.4, p=0.001, ηp2=0.57. Post hoc comparisons indicated that in the varenicline phase, there was no change in cigarettes per day between Visits 1 (mean=16.1, SE=0.86) and 3 (mean=15.1, SE=1.3; Days 1 through 9), F(1,13) = 1.1, p=0.35, ηp2=0.16. Cigarettes per day began to decrease at Visit 4 (14.1, SE=1.3) and continued to decrease through Visit 6 (mean=12.8, SE=1.3), F(1,13)=4.8, p=0.02, ηp2=0.57. Overall, there was a 23% reduction in cigarettes per day during the varenicline phase. In contrast, there was a non-significant reduction in cigarettes per day during the placebo phase (mean difference=1.7, SE=1.1), F<1. Similar results were observed when examining cigarettes per day by discrete timepoints (i.e., the final day between each visit). The drug × time interaction was significant, F(1,13)=8.8, p=0.01, ηp2=0.4. The means (SE) for the varenicline period were: 15.2(0.65); 15.5(1.2); 15.2(0.95); 14.8(1.6); 14.5(1.4); 13.3(1.3); and 12.4(1.3). For the placebo period, the means (SE) were: 15.8(1.2); 16.3(1.0); 15.0(1.1); 15.8(1.3); 16.7(1.2); 14.2(1.6); and 13.4(1.7). Despite the fact that participants smoked fewer cigarettes during the second phase compared to the first phase, treatment order × drug interaction F(1,13)=11.5, p=0.005, ηp2=0.47, treatment order had no effect on the critical drug x time interaction, p=0.22.
Cigarettes per day (CPD) across time during the varenicline and placebo phase.
Although there were main effects of age and treatment order on total puff volume, F(1,11)=5.8, p=0.034 and F(1,11)=8.9, p=0.012, respectively, the main effects of drug and time and their interaction were not significant, all ps>0.22. Despite randomizing treatment order across participants, those who received placebo first demonstrated lower total puff volume (mean=523.4, SE=41), regardless of drug or time compared to those who received varenicline first (mean=684.6, SE=33).
Daily smoking behavior
For the composite measure of `daily smoking behavior' (product of cigarettes per day and total puff volume), the Huynh-Feldt adjusted drug × time interaction was significant, F(6,66)=2.4, p=0.036, ηp2=0.18. Similar to cigarettes per day, the drug × time linear contrast was significant, F(1,13)=7.5, p=0.02, ηp2=0.40. Simple effects post hoc tests suggested that during varenicline, there was a significant reduction in total smoking behavior from Visit 1 (Day 1) (mean=9375.4, SE=538) to Visit 6 (Day 21) (mean=6564.4, SE=999), F(1,13)=8.2, p=0.014, ηp2=0.38. In contrast, there was no significant change in total smoking behavior during placebo from Visit 1 (mean=9550.8, SE=1097) to Visit 6 (mean=7892.3, SE=1068), F < 1.
Effect of treatment on craving
depicts QSU Factor 1 and 2 scores by time for each drug phase. Based on our hypothesis that varenicline may reduce smoking behavior through a decrease in the anticipation of pleasure from smoking, we predicted that decreases in urges to smoke would be primarily due to reductions in QSU Factor 1 (`anticipation of pleasure from smoking'). Although the drug × time interaction was not significant, F(1,11)=1.2, p=0.30, QSU Factor 1 tended to decrease across time, F(1,11)=2.9, p=0.11, ηp2=0.21. Unprotected follow-up tests indicated that decreases in `urges to smoke for pleasure' (QSU Factor 1) only occurred in the varenicline phase, F(1,13)=6.0, p=0.03, ηp2=0.32. Between Day 1 and 10, there were no significant changes in QSU Factor 1 scores, ps>0.22. Beginning at Day 14, there were significant decreases in urges to smoke for pleasure in the varenicline phase, F(1,13)=11.8, p<0.01.
(a) Urges to smoke for pleasure (QSU Factor 1) and (b) urges to smoke for withdrawal relief (QSU Factor 2) across time during the varenicline and placebo phase.
There were no effects of drug or changes across time for QSU Factor 2 (`urges to smoke for withdrawal relief'), Fs <1. For the total QSU score, the repeated measures ANCOVA model revealed no significant changes in overall urges to smoke as a function of time or drug, ps>0.17. However, the pattern of means suggested a greater reduction in craving scores during varenicline (mean difference=0.55, SE=0.24) compared to placebo (mean difference=0.32, SE=0.27).
Effect of treatment on biomarkers of nicotine exposure
contains means (SD) for all drug × time cells for nicotine metabolites. For total metabolites, there was an overall effect of time, F(1,12)=5.1, p=0.045, ηp2=0.32, such that the total urinary metabolites decreased from Day 1 to Day 21, irrespective of drug condition, drug and drug × time interaction, Fs <1. For urinary nicotine, the drug × time quadratic interaction was marginal, F(1,12)=3.6, p=0.08, ηp2=0.24. Consistent with self-reported cigarettes per day, there was a significant decrease in urinary nicotine in the varenicline phase, F(1,12)=12.3, p=0.004, ηp2=0.51, but not in the placebo phase, F<1. Although cotinine tended to decrease across time this effect was not statistically significant, F(1,12)=3.1, p=0.10, ηp2=0.21. The overall effect of drug and the drug × time interaction were not significant, Fs <1. There were no significant changes in CO breath levels either as a function of drug or time, Fs <1.
Nicotine metabolites for each drug × time cell (N=17).
There were no significant main effects or interactions for drug and time on the side effects summary score, Fs<1.2, ps>0.30. Nausea ratings were marginally higher in the varenicline phase compared to placebo, F(1,12)=4.3, p=0.06, ηp2=0.27 and this effect appeared to vary across time, drug × time quadratic interaction, F(1,12) = 5.7, p=0.034, ηp2=0.32. However, the increase in nausea appeared to be driven by a small subset of participants. Indeed, the majority of participants reported no nausea during varenicline (n=14). Only three smokers reported mild or moderate nausea and no one reported severe nausea. There were no significant changes in nausea ratings in the placebo phase, F(1,12)=1.2, p=0.31. Importantly, there were no differences in medication adherence between individuals who reported nausea (mean compliance=99%) and individuals who did not (mean compliance=98%), nor were there differences between the placebo phase and the varenicline phase (96% vs 97%), F<1.
Association between changes in urges to smoke and cigarette consumption
Post hoc correlation analyses were conducted to explore the association between the time course of varenicline's effects on urges to smoke and cigarette consumption. We observed a reduction in cigarettes per day beginning at Visit 3 (i.e. Day 10), and so we were interested in whether a reduction in urge to smoke prior to Day 10 was related to changes in cigarette consumption. Thus, we examined the relationship between decreases in urges to smoke from Day 1 to Day 10 and subsequent reductions in cigarettes per day from Day 10 to Day 20. Both QSU factors were included and the varenicline and placebo phases were examined separately. Results revealed that decreases in urges to smoke for pleasure (QSU Factor 1) were significantly associated with greater reductions in cigarette consumption (r=0.60, p=00.01) during the varenicline phase, but not the placebo phase (r=−0.06, p=0.82). Changes in urges to smoke to reduce negative affect and decreases in cigarettes per day were not related in either phase (rs<0.33, ps>0.21).