PASFOOD’s first objective was to report occurrence levels of pyrrolizidine alkaloids (PAs) and N-oxide (PANOs) in targeted food items on the Belgian market in order to perform a dietary intake assessment to these natural hepatotoxic protoxins for the Belgian population. Indeed, these contaminants can be transferred to food of vegetal/ animal origin and could become a potential Public Health issue. Therefore, the development of efficient analytical methods was required to detect the presence of PAs at ppb or even ppt levels.
Analytical methods based on analytes extraction and SPE purification, followed by UPLC separation and MS/MS detection were developed at WIV-ISP for the analysis of 16 PAs and 14 PANOs. The sample preparation and instrumental methods were optimized to obtain the best method sensitivity. Nine analytical methods were validated in-house and the validation parameters met the critical validation criteria allowed by Directive 2002/657/CE and SANTE/11945/2015 in the vast majority of cases. Moreover, the LOQs achieved were often lower than those required in the project call.
A structured sampling plan has been established for more than 1100 samples sold on the Belgian market covering 7 food groups (700 samples were initially foreseen). The targeted food commodities were broadly diversified: honey, honey based snacks & sweets, milk, dairy products, (processed) meat, liver products, pre-packaged salads, mixes of herbs & spices, (herbal) teas, ice-teas and beehive products based – & plant based food supplements. It is noteworthy that we have had the opportunity to access all honeys produced in Belgium.
The generated PAs/PANOs occurrence data were compared to the one published by EFSA. Retail honeys were more contaminated than Belgian samples, and high levels of europine were detected for the first time in this matrix. The PAs observed in positive samples among the Belgian honeys were in agreement with the recurrent flora in the country. Honey based sweets & snacks were barely contaminated. In contrast with the data reported by EFSA, the contamination in food items of animal origin (milk, dairy products & meat products) was dominated by N-oxides, which calls in question the results reported in previous studies that stated that PANOs would be degraded or converted to PAs in the course of digestion of the plant material by rumen. However, the levels detected in these matrices were low (ppt range). Consistently with the main target organ of PAs/PANOs, liver products were the most contaminated amongst the meat samples, and mainly products based on duck meat. Surprisingly, very high concentrations of heliotrine-type compounds (ppm range) were detected in herbs mixes for pizza & spaghetti. Some salad mixes were also contaminated with unexpected PA producing botanicals. To the best of our knowledge, only very few studies have focused on the detection of PAs and PANOs in salads and herbs/spices. In light of the results, these matrices should be further investigated. Teas & herbal teas frequently contained high levels of a wide range of PA contaminants. An original spiking experiment enabled to highlight the fact that the transfer of PAs/PANOs from dry (herbal) tea to the infusion in the course of the brewing process was not total: only 16 to 26% of the contamination was transferred to the infusion. This new approach to quantify PAs/PANOs in (herbal) infusions contrasts significantly with the conventional strategy used by EFSA, that only apply a dilution factor from the concentration in the dry plant product to evaluate the contamination of the infusion. As a result, major discrepancies with previously reported concentration values appeared: the contamination levels are lower but much more realistic. Finally, plant based food supplements contained massive concentrations of PA contaminants, but their concentration is highly dependent of the presence of PA producing plants. Indeed, some illegal botanicals according to AR 29/08/97 were found in food supplements bought online and FASFC has been notified for exceeding PA tolerated levels. The contamination in beehive products based food supplements was dominated by pollen samples; other formulations were largely less contaminated. For all studied matrices, the major contributors were senecione-type and lycopsamine-type PAs/PANOs.
The dietary intake assessment was performed by coupling the analytical concentration levels with the data on Belgian dietary intake. When using a refined scenario for the classification of food items, the PAs/PANOs intake fell for a factor 4 compared to a more global scenario. The estimates of the mean intake and the 95th percentile fell within the calculated range of EFSA: between the median and the maximal estimates for all age groups. Despite very low concentrations detected, milk and dairy products were the main contributor to the PAs/PANOs intake, whereas food supplements were anecdotal but presented the higher concentration levels.
