RESUMO
The acidic and neutral impurities in heroin samples (46 Purified South-East Asian (PSEA) and 8 non-PSEA samples) were analysed using gas chromatography. Mass spectral detection allowed the construction of a comprehensive library of over 649 impurities. A variety of statistical procedures were used to select 70 viable profiling parameters. Cluster analyses were used to investigate the similarities between samples using organic parameters. A blended PSEA profile was constructed for comparison with samples of unknown origins. Resolution between samples of PSEA/non-PSEA origins was demonstrated with a selection of 25 (18 continuous and 7 dichotomised) discriminating factors. Several predictive logistic models were created using up to 18 parameters, explaining 85-100% of the variation in the experimental data.