Partner und Internationale Organisationen
(Englisch)
|
University of Strathclyde, (UK); National Bureau of Investigation, (FIN); National Laboratory of Forensic Science, (S)
|
Abstract
(Englisch)
|
Development of a harmonised method for the profiling of amphetamine is a multi-task project which began on February 1st, 1999. The ultimate aim of the project is to develop a harmonised, collaboratively tested method for the profiling of amphetamine so that meaningful analytical data, for use in the control of amphetamines, may be distributed on a pan-European basis. Development of this method can then be broken down into a number of distinct phases, namely :-
1. Synthesis of documented standards for use throughout the project;
2. Study of the stability of the impurities in solution;
3. Optimisation of GC and detector systems;
4. Optimisation of extraction protocols;
5. Determination of variability of results;
6. Investigation of numerical classification schemes for amphetamine;
7. Report preparation and completion.
Phases 1 , 2 and part of phase 3 were carried out during the first year of the project.. Phase 3, phase 4 and part of phase 5 were successfully completed during the second year of the project (until September 1st 2001). From then until October 31st 2002, the remaining part of task 5 as well as task6 and task 7 (wiriting up of the final report) were carried out and are now completed.
Phase 5 : this task dealt with the variablity of results within day, between days and between laboratories. For this purpose, samples were prepared by each laboratory and sent to the other partners.
Results showed good results in terms of reproducibility. Relative errors were in the order of 5 % for within day experiments, 8 % for between days and between 8 and 10 % between laboratories.
Phase 6 : Several batches of amphetamine were synthesised via the three defined synthetic routes and analysed with the optimised method to allow a proper evaluation of the numerical comparison methods. The recipes were varied in order to span the experimental domain as much as possible. Batches of amphetamine were also made by carefully repeating a recipe in order to study the variability that is obtained in such a case. Additionally, about 400 street samples were collected and analysed (from Finland, Sweden and Switzerland). Different distance metrics, potentially applicable for comparison analysis, were evaluated. These include the Euclidean distance, the Manhattan distance, the Canberra Index, Pearson correlation, the squared sinus and the quotient method. Moreover, a range of different methods for pre-treatment of data, including normalisation, weighting, the logarithm and the fourth square root, was evaluated. The quality of the comparison was also evaluated with respect to the number and type of impurities used in the calculations. Handling of zeroes and missing data were also paid attention. The number of impurities and the pre-treatment method was optimised for principal component analysis (PCA) used for classification of amphetamine. Finally, the developed method was compared to an existing profiling method, which was considered as a reference method, with respect to distance calculations and PCA modelling.
Finally, MSD (Mass Spectrometer Detector) data were of better quality than FID (Flame Ionization Detector) data, which was demonstrated by the superior discrimination between linked (i.e. same batch) and unlinked distances. This was probably due to the better selectivity provided by the MSD, which therefore easily can handle co-elution that by necessity occurs in amphetamine profiling due to the complex nature of the samples. Furthermore, it was shown that the developed profiling method performed better than a profiling method that currently is in use in some forensic laboratories. This was demonstrated by a better separation of linked and unlinked distances. Although a bit subjective, the different synthetic routes seemed to be separated more efficiently when applying PCA to the data of the developed method.
The developed profiling method was demonstrated to provide information of high quality that can be extracted efficiently when employing proper distance and pre-treatment methods.
|
