Development and validation of a liquid chromatographic - tandem mass spectrometric method for the detection of fumagillin in honey: use in a stability study

Nosemosis, a damaging adult bee disease caused by Nosema apis or N. ceranae was in the past often treated with fumagillin, an antibiotic commercialized under the name Fumidil B®. Despite the fact that the use of Fumidil® is presently not permitted, the product is still available in the United Kingdom. In the past it was shown that residues of antibiotics can be present in honey after treatment. In order to be able to screen honey samples on the presence of fumagillin, a liquid chromatographic tandem mass spectrometric was developed and validated according to Commission Decision 2002/657/EC. After addition of the internal standard roxithromycin, 5 g of honey was dissolved in 10 ml of water and a clean-up was performed on a C18-SPE column. Five µl of the final extract was injected into the LC-MS/MS system on a Waters X-Terra® C8 column kept at 25°C. The instrument used was an Acquity UPLC (Waters) system. Gradient elution was performed with a mobile phase mixture of methanol and 2 mM ammonium formate in 0.01 % formic acid. Mass spectrometric detection was performed on a Quattro Ultima Pt® instrument, following the transition of the precursor ion to three product ions. For the internal standard, one transition was measured. Performance characteristics evaluated were specificity, linearity, recovery, repeatability, intra-laboratory reproducibility, decision limit, detection capability and expanded measurement uncertainty. For specificity, honey was spiked with compounds belonging to the groups of the tetracyclines, sulfonamides, nitro-imidazoles, phenicols and quinolones at 10 µg/kg. No interference was detected at the retention time of fumagillin that could give rise to a false non-compliant result. For linearity, a correlation coefficient of at least 0.9988 was obtained in honey. Eight blank honey samples were spiked at three different concentrations being 3, 4.5 and 6 µg/kg. Quantification was performed making use of a calibration curve in matrix. Recoveries ranged between 102 and 105 % and CVs for repeatability were between 4 and 5 %. To determine the intra-laboratory reproducibility, the above experiment was repeated two times and CVs ranged from 6 to 10 %. Decision limit was determined as the concentration for which a signal to noise ratio of minimum 3 was obtained for the three product ions and was determined at 3 µg/kg. Detection capability of the method is 3.3 µg/kg and the expanded measurement uncertainty is 24.6 %.
The method was used to determine the stability of fumagillin in honey. Honey was spiked at 20 µg/kg with fumagillin and was analysed immediately. The rest of the sample was kept at room temperature in the dark. After two weeks the concentration of fumagillin was reduced with 29 %, while after four weeks concentration was reduced with 52 %. The final results will be discussed at the symposium.
The method was also used to analyse 20 honey samples from 2011 but as fumagillin is unstable in honey, the negative results obtained have no significance. Samples of 2012 will be analysed immediately after the harvest in order to check if fumagillin is no longer used in practice.