Roberto Gomeni, Vincenzo Teneggi, Laura Iavarone, Lisa Squassante, and Alan Bye.
GlaxoWellcome, Medicine Research Center, Via A. Fleming 4, 37135 Verona (Italy).
Objective: A placebo controlled, randomised, double blind, three periods, crossover trial was designed to investigate the relationship between craving scores and circulating nicotine concentrations in a population of 24 smokers. In two periods, smoke was forbidden to subjects who received NRT (a patch with a Nicotine Replacement Treatment) and placebo then, in the third period, the subjects were allowed to smoke. A specific questionnaire (the Tiffany rating scale) was used to quantify the degree of smoking urges.
Methods: Two independent modelling methods were used. The first one was based on a population indirect Pk/Pd model describing the nicotine-induced changes on craving scores as a process affected by a production (kin) and dissipation rate (kout). Inspection of placebo changes on craving scores over time indicates that there was circadian variability; therefore a cosine function was included in the model. A semi-parametric approach was used to model the nicotine concentration. A sigmoidal process stimulating the kout rate best described the effect of nicotine while the kin rate accounted for the placebo circadian process. The analyses were performed using NONMEM (Version V).
The second modelling approach was based on the estimate of the probability of a reduction of at least 20% on craving scores as a function of nicotine concentration. A dummy binary variable taking the value 0 for clinical failure and 1 for success was derived from the observed craving scores and used with the corresponding nicotine levels as a predictor variable in a logistic regression using SAS (Version 6.12).
Results: The placebo response model was able to properly fit the circadian changes on craving scores during a cycle of 24 hours. The average craving score reductions observed after NRT and smoke were similar (0.7 and 0.9 respectively) but the average nicotine concentration was more than twice larger in the smoke period (383 ng/ml) than in the NRT period (156 ng/ml). The Emax values after NRT and smoke were very close (0.783 and 0.696 respectively) and were in agreement with the average craving score reductions. Moreover, the estimated nicotine concentrations giving 50% of the maximal effect for NRT and smoke were in same ratio (0.43) as the average nicotine concentration measured after the two treatments (0.41). The logistic analysis showed that the nicotine concentration was a significant predictor of the pre-defined outcome in the smoke period: the probability of a craving score reduction of at least 20% was estimated to 0.5 for a nicotine concentration of 523 ng/ml and to 0.8 for a nicotine concentration of 1900 ng/ml.
Conclusion: The indirect modelling approach adequately described the onset, extent, and duration of the pharmacological response after placebo, NRT and smoke while the logistic model was able to estimate the probability of a given outcome in respect of a nicotine circulating level.
Reference: PAGE 9 (2000) Abstr 84 [www.page-meeting.org/?abstract=84]
Poster: oral presentation