New approaches for the risk assessment of pyrrolizidine alkaloids and their N-oxides present in teas and plant food supplements

Pyrrolizidine alkaloids (PAs) and pyrrolizidine alkaloid N-oxides (PANOs) are natural toxins produced by plants. They can be present in food from direct use of PA-producing plants and/or through accidental contamination. The latter is especially concerning as these toxins have been reported in many products including teas and plant food supplements (PFS). In nature, these compounds exist in two forms: PAs and PANOs. Although PANOs are considered less toxic than their corresponding parent PAs, PANOs can undergo metabolic reduction to PAs via intestinal microbiota and liver enzymes, and thus, can induce similar effects mediated by their parent PAs. Understanding such conversion process can help with establishment of the so-called relative potency (REP) value of PANO relative to the corresponding PA. Deriving these REP values using animal studies is unethical and unrealistic given the large number of these compounds. Given that and the fact that these compounds have reported in plant-based food products made from non-PAs-producing plants, the aims of the current thesis are twofold. The first was to employ new approach methodologies (NAMs) to predict the in vivo REP values of a series of PANOs relative to their parent PAs using in vitro and in silico-based approaches including physiologically based kinetic (PBK) modelling focusing on compartments relevant for PANOs’ reductions. The second was to perform a combined risk assessment of PAs/PANOs in (herbal) teas and PFS using the margin of exposure (MOE) approach for lifetime exposure, while for less than lifetime (LTL), Haber’s rule and the threshold of toxicological concern (TTC) approach were employed. Moreover, the risk assessment was conducted with and without applying REP values corrections.

On one hand, the predicted REP values seemed to be influenced by many factors, some of which are the species differences, structural characteristics, endpoints and dose. However, at doses relevant to human dietary intake, the REP values showed to be only influenced by the structural differences of the examined compounds and became independent of other factors. At low dose scenarios, the obtained data showed that the REPPANO to PA values decrease following the order: macrocyclic diester 7R-retronecine-type > monoester 7R-retronecine-type > open chain diester 7R-retronecine-type > monoester 7S-heliotridine-type = open-chain diester 7S-heliotridine-type. On the other hand, although none of the examined samples contained PA-producing plants, about 16% of (herbal) teas and 31% of the PFS resulted in MOE values below 10,000 when assuming lifetime consumption. When considering LTL scenarios, several teas and PFS showed concerning MOE values and/or exceeded the TTC values. Overall, the present thesis provided tools to predict the in vivo REP PANO-to-PA values at low but realistic dose levels without the need for animal studies, thus contributing to establishing more robust REP PANO-to-PA values that can be used for the risk assessment of PANOs. Using these values in the combined risk assessment provides a refined and optimized way of defining the samples of concern. The outcomes reveal that (herbal) teas and PFS even when not containing PA/PANO producing botanicals do raise concerns and indicate a need for risk management actions.