Several Aromatic amines (AA) have been classified as human carcinogens, and tobacco smoke is one of the most important sources of human exposure. They are metabolized in the body and can ultimately be excreted in urine as metabolites or free AA. They can be analyzed with gas chromatography-mass spectrometry (GC-MS), typically after a complex and labor-intensive sample preparation procedure, involving hydrolysis, extraction, and derivatization steps. The objective of this thesis was therefore to optimize and automate the sample preparation procedure for the analysis of aromatic amines in urine.

As a proof-of-concept, an existing procedure was evaluated and it´s suitability for such analysis studied. Therefore, the relationship between the smoking status of the urine donors and the amount of AA present was studied in 68 samples from 10 smokers (S), 28 past smokers (PS) and 30 never smokers (NS). Furthermore, three different data evaluation approaches were presented: a qualitative analysis, a quantitative analysis, and a quantitative screening. Due to the high variability in concentrations typically observed in biological samples, the quantitative screening was proven a very promising alternative to the quantitative analysis. And a relationship between the smoking status of the donors and the AA present could be established. The method was therefore deemed suitable, and it was further optimized and automated.

The Liquid-Liquid Extraction (LLE) step was the first step studied because of the large volumes of sample and toxic organic solvents needed, and the fact that it is a very time-consuming and labor-intensive step. Two alternatives were evaluated, namely Hollow Fiber - Liquid Phase Microextraction (HF-LPME) and Parallel Artificial Liquid Membrane Extraction (PALME) and relevant extraction parameters were optimized. Because significantly higher recoveries could be observed with PALME when comparing the optimized methods, PALME was further validated. PALME was proven a very promising alternative to LLE, with limits of detection (LOD) of 45-75 ng/L, and repeatability and peak area relative standard deviations (RSD) below 20 %.

The next step was to evaluate different detectors. To this end, GC-MS in single-ion monitoring (SIM) mode with (1) electron ionization (GC-EI-MS) and (2) negative chemical ionization (GC-NCI-MS), and (3) GC-EI-MS/MS in multiple reaction monitoring (MRM) mode using electron ionization were studied. Most analytes showed excellent LOD (50, 3.0-7.3, and 0.9-3.9 pg/L for (1), (2), and (3) respectively), good precision (intra and inter-day repeatability < 20 %) and excellent recoveries (between 80 and 104 %). From the three techniques studied, the most promising one was GC‑EI‑MS/MS. GC-NCI-MS was proven an interesting alternative for qualitative/non-target analysis, since all the derivatized iodinated amines could be easily identified and the significant loss in sensitivity observed over time would not be as detrimental as in quantitative analysis.

Finally, in order to minimize the need for human intervention and the opportunities for errors, and improve the overall greenness of the analytical procedure, the sample preparation was automated. Different problems encountered, like volume limitations or needle penetration depth adjustments, are discussed in detail. And thanks to the less labor-intensive and time-consuming sample preparation procedure, several steps, such as reaction/extraction times or some of the reagents which were not optimal for the automated set-up, could be further optimized. The sample preparation procedure for the analysis of aromatic amines in human urine, could therefore be successfully optimized and automated.

Having an automated and optimized sample preparation procedure would enable the analysis of enough real samples so that the relationship between AA, smoking status and smoking-related diseases could be determined. Furthermore, the analytical method could be used to analyze collective samples, for example, from workers at risk of AA exposure and the surrounding population and could enable the real time monitoring of occupational exposure. Moreover, this method could also be used for wastewater-based epidemiology and could help monitor a population´s exposure to tobacco smoke and its health status.