Matrix effects in human urine analysis using multi-targeted liquid chromatography-tandem mass spectrometry

Different sample preparation methods were tested for human urine for the subsequent analysis with a LC-MS/MS multimethod to quantify 65 micropollutants (pesticides, pharmaceuticals, personal care products, industrial chemicals and metabolites) within the general population. Direct injection of diluted urine revealed highly variable and often severe signal suppression for nearly all analytes at relevant concentration levels during electrospray ionization. Urine samples were highly variable in their total organic carbon (500-10,000mgL(-1)) and creatinine (0.35-13mM) content as well as in electrical conductivity (3-19mScm(-1)) but these differences did not correlate clearly with the strength of matrix effects. Therefore, matrix removal by solid phase extraction was intended and different mixed-mode sorbents were tested. Results showed poor apparent recoveries likely due to insufficient separation of target analytes and matrix compounds for all tested sorbents. The wide variability and high concentration levels of urine constituents strongly affect electrospray ionization as well as recovery during extraction at the sub-microgram per liter level. Especially, the hydrophobic interaction at the hydrophilic-lipophilic-balanced sorbent was affected. It is concluded that the urine matrix is too strong, too diverse and too variable to allow one sample preparation method for the very diverse analytes of a LC-MS/MS multimethod. Instead dedicated methods appear more promising.