|SOFT - TIAFT 1998||Poster Session 4||Friday October 9, 1998|
SYSTEMATIC TOXICOLOGICAL ANALYSIS OF BIOSAMPLES BY GC-MS. METHODS FOR AUTOMATED PEAKS DETECTION AND MASS SPECTRAL PURIFICATION|
Aldo Polettini, Angelo Groppi, Cristiana Stramesi, Giorgio Marrubini Bouland and Maria Montagna
Department of Legal Medicine and Public Health, University of Pavia, Via Forlanini 12, 27100 PAVIA, Italy
|Different methods for automated peaks detection and mass spectral purification were comparatively evaluated in the processing of full scan GC-MS data obtained for the Systematic Toxicological Analysis of blood and urine samples. The method of Transformed Chromatograms (TC) is based on the transformation of the total ion chromatogram (TIC) into two chromatograms by subtracting from the intensity of each scan both the intensities of the preceding and of the following scan. The two TCs are then integrated and the mass spectra of the negative and positive peaks detected are library searched after background subtraction (apex-start and apex-end). The Peak Purity (PP) method is based on the detection of peaks by integration of the TIC and the application of the Hewlett-Packard Peak Purity algorithm for peaks deconvolution. Scans corresponding to the maxima of each supposed component of unresolved GC peaks are library searched with and without subtraction of adjacent scans. The method of Searching All Scans (SAS) is based on the library search of all S-(S-1) and S-(S+1) subtracted scans. In all cases library search was carried out using the Probability Based Matching algorithm against an implemented version of the PMW_TOX2 library. A macro was prepared using the macro language of the HP Enhanced Chemstation software in order to automatically process a whole TIC up to the printout of the search report. Filters (match quality and retention index) were adopted in order to reduce the number of false positives.
Two different sets of GC/MS data were used in the evaluation: a) GC/MS data from urine samples spiked with different barbiturates and benzodiazepines at the 0.2 mg/l (sample preparation consisted of SPE and trimethylsilylation); b) GC-MS data for blood and urine samples from intoxication cases (samples were submitted to solvent extraction and acetylation; preliminary detection/identification of unknowns was manually carried out by selected ions retrieval).
The performance of PP in terms of percentage of positive identifications (%ID) was substantially improved by subtracting adjacent scans to maxima (%ID from 72.7 to 96.4%), thus approaching the performance of TC (98.2%) and SAS (100%). Usually, methods based on peaks detection by integration of chromatograms (TC, PP) performed satisfactorily even in the case of very dirty TICs provided that a fair GC resolution was maintained; on the contrary when matrix interference strongly affected GC repartition the %ID decreased (TC, 84.1%; PP, 87.8%). The SAS method was in all cases the best performing (%ID, 98.1-100%), although it gave the highest number of false positives (32-52 per TIC compared to 16-29 for TC and 16-30 for PP) and it had the slowest processing speed (2.2 scans/sec, compared to 5 and 4.5 scans/sec).