Study design and aims
This randomised, cross-over trial investigated the effect of nicotine mouth spray on urges to smoke in healthy smokers who had abstained from smoking overnight prior to study treatment administration. The primary objective was to compare urges to smoke in the period immediately after administration of either 2 mg nicotine mouth spray, 2 mg nicotine lozenge or 4 mg nicotine lozenge. The study treatments were also compared with respect to the time to predefined reductions of the baseline urges to smoke score, and the proportion of subjects who reported predefined degrees of reduction of their baseline urges to smoke within 1, 3, 5 and 10 min postadministration. The trial was conducted at McNeil AB, Department of Clinical Pharmacology, Lund, Sweden. The study was initiated on 1 February 2010 and completed by 19 May 2010. The trial protocol, developed by McNeil AB, was approved by the local Independent Ethics Committee in Lund, Sweden, and the Swedish Medical Products Agency, and the study was performed in accordance with current International Conference on Harmonisation of Good Clinical Practice Guidelines. All subjects provided signed, informed consent before entering the study.
Healthy adult male and female smokers of at least 10 cigarettes/day, between 19 and 55 years old, who smoked their first cigarette of the day within 30 min of waking. Eligible subjects had to weigh at least 50 kg and have a body mass index of 17.5–32 kg/m2. Intention to stop smoking was not an explicit inclusion criterion.
Subjects who presented with any of the following were excluded: women who were either pregnant or breastfeeding; any pathological oral status that may have interfered with normal oral function or transmucosal absorption; use of any other nicotine replacement medication or bupropion or varenicline, or a quit attempt, within the 3-month period prior to the study; previous regular use of either nicotine mouth spray or lozenge; or regular alcohol consumption that exceeded weekly limits of 2 litre of wine or 5 litre of beer or 0.6 litre of spirits for women, and 3 litre of wine or 7.5 litre of beer or 0.9 litre of spirits for men.
Treatment periods and study medications
The study period comprised three separate treatment visits, with washout periods of at least 36 h between visits. Subjects were to abstain from smoking from 20:00 h in the evening before each treatment visit until the end of the visit at about 16:00 h. All subjects received two doses of each study treatment at each visit, one in the morning and another in the afternoon, 5 h after the first administration. The morning administration and measurements of urges to smoke allowed subjects to get used to the study treatments and rating scales.
The three study treatments administered were nicotine mouth spray 2 mg (Nicorette Mouth Spray 1 mg/spray, McNeil AB), nicotine lozenge 2 mg and nicotine lozenge 4 mg (NiQuitin Lozenge 2 and 4 mg, GlaxoSmithKline, Copenhagen, Denmark). The doses selected were based on the currently recommended single doses for nicotine lozenge and the maximum proposed single dose for nicotine mouth spray. A computer-generated randomisation schedule, produced by the sponsor, was used to randomly allocate the subjects to the different treatment sequences. Randomisation was balanced for the six possible treatment sequences by using randomisation blocks of size 12. Randomisation numbers were assigned sequentially to subjects at their first treatment visit.
After priming, the spray delivers 1 mg of nicotine per metered-dose spray. A 2 mg dose of mouth spray was administered as two consecutive sprays of the solution, sprayed into the mouth. Subjects were advised to avoid swallowing immediately after administration of the spray. The reference product, nicotine lozenge, is a widely used oral NRT product that effectively aids smoking cessation and has been shown to relieve craving.7
Subjects were instructed to place the nicotine 2 or 4 mg lozenge in their mouth and to occasionally move it from one side to the other until the lozenge had completely dissolved. They were instructed to not chew or swallow the lozenge.
A lozenge typically dissolves in 20–30 min. Subjects did not eat or drink from 15 min before until 60 min after administration of each study treatment.
Before entering the trial subjects underwent a physical examination, including vital signs, and provided a medical and smoking history.
Urges to smoke were scored on a 100 mm visual analogue scale (VAS) 10, 6 and 2 min before and 1, 3, 5, 10, 15, 20, 25, 30, 45 min and 1, 1.5 and 2 h after treatment administration. On the VAS, zero represented ‘no urges to smoke’ and 100 mm represented ‘extreme urges to smoke’. Ratings of urges to smoke were collected using electronic subject diaries (eDiaries). The eDiary consisted of a hand-held computer that had been programmed to allow subjects to enter specific data and answer predefined questions. The study personnel demonstrated how to use the eDiary, and were present to assist while subjects used the eDiaries if any queries arose. All adverse events spontaneously reported by subjects were recorded, and subjects were also questioned about adverse events.
The primary study endpoints were the area under the urges-to-smoke-versus-time curve (AUC) from start of administration until 1, 3 and 5 min postadministration (AUC1 min, AUC3 min and AUC5 min, respectively). Secondary endpoints were the AUC from start of the administration until 10 min (AUC10 min), the time to 25%, 50%, 75% and 90% reduction in urges to smoke score versus baseline, and the proportion of subjects who attained 25%, 50%, 75% and 90% reduction of urges to smoke, versus baseline, at 1, 3, 5 and 10 min.
The composite nature of the primary study objective motivated a Bonferroni-adjusted significance level of 2.5% and a target type-II error rate of 5% for each separate comparison. On the basis of previous data from a similar study, the within-subjects SD was assumed not to exceed 120 mm×min for any of the primary endpoints. Under these premises, to be able to detect a mean treatment difference of 50 mm×min with a balanced cross-over design and a two-sided analysis of variance-based test, in total 178 subjects (2×89) were needed. To allow for dropouts, recruitment continued until 200 subjects were included in the study.
The data set analysed for relief of craving comprised the data collected following administration of study treatments after 5 h of witnessed abstinence from smoking (the afternoon administration). All randomised subjects with any urges to smoke data were included in the pharmacodynamic analysis, and all subjects who received at least one dose of study treatment were analysed for safety.
Pair-wise treatment comparisons with respect to AUCn min (where n=1, 3, 5 and 10, that is, the area under the linearly interpolated urges-to-smoke-versus-time curve from time zero until n min) were based on a mixed linear model (SAS V.9.1, PROC MIXED) that included sequence, treatment, period and the baseline urges-to-smoke score as fixed effects, and subject, nested within sequence, as random effect. The baseline urges-to-smoke score was determined by the mean value of the three pretreatment assessments, 10, 6 and 2 min before administration.
Treating subject as a random factor in the model makes efficient use of all available valid data points. A post hoc sensitivity analysis of the primary endpoints based on the subset of subjects with complete and valid data from all three study periods (n=178) was also performed.
The possible effects of carry-over were assumed small relative to main effects of treatments, and in the primary analysis, no attempt was therefore made to test for this or any other treatment-by-period interaction. A post hoc sensitivity analysis of the primary endpoints restricted to valid data from the first session only, and consequently not affected by carry-over, showed similar results to the main findings presented in .
Table 1 Estimated mean (LSmean±SE) average changes in urges-to-smoke scores and corresponding comparisons between treatments (estimated treatment difference (97.5% CI) and Bonferroni-adjusted p value) at 1, 3, 5 and 10 min postadministration by (more ...)
To aid interpretation of the study findings model-based estimates of treatment differences in mean average score changes were presented together with CIs instead of, and equivalently to, model-based estimates of treatment differences in mean AUCs. The average urges-to-smoke change from baseline was defined as
was the endpoint of the time interval.
As the study included two different treatment comparisons (mouth spray 2 mg vs either 2 or 4 mg lozenge), the significance level for the relevant test was adjusted to 2.5% to ensure that the overall type-I error rate was no greater than 5%.
To evaluate the primary study objective and to avoid inflation of the significance level due to multiple testing, for each strength of the reference product (2 or 4 mg lozenge), the statistical evaluation of the primary endpoints was performed in a hierarchical order, starting with the 5-min evaluation. Thus, no additional multiplicity adjustments were needed, that is, all statistical tests relating to evaluation of the primary endpoints were performed at the 2.5% significance level.
To correct for bias due to the discrete sampling scheme, the times to 25%, 50%, 75% and 90% reduction of baseline urges-to-smoke scores were estimated from the assessments up to 120 min postadministration using linear interpolation between assessment time points. Any time-to-event that exceeded 120 min was censored at that time point. Estimated medians of time-to-event and corresponding 95% CIs were calculated by treatment. Pair-wise treatment comparisons of the times to 25%, 50%, 75% and 90% reduction of baseline urges-to-smoke scores were based on the Sign test.
The proportion of subjects who attained 25%, 50%, 75% and 90% reduction of baseline urges to smoke within 1, 3, 5 and 10 min, respectively, were pair-wise compared between nicotine mouth spray and the two comparator lozenges by using the McNemar test for each time point.
All statistical tests relating to evaluation of the primary endpoints were two-sided with a significance level α=2.5%. Corresponding CIs were two-sided and had a nominal confidence level of 97.5%. All other statistical tests were two-sided with a significance level α=5%. Corresponding interval estimates were two-sided and had a nominal confidence level of 95%.