The PASFOOD project delivered unexpected results related to the contamination of aromatic herbs with pyrrolizidine alkaloids (PA’s), which are toxic natural compounds to humans at very low doses. It was thus important to further investigate the possible role of aromatic herbs for the exposure of the Belgian consumer to the carcinogenic PAs in the course of a follow-up project.
The PASFOOD project yielded many interesting results and highlighted a recurrent contamination of herb mixes, sometimes with very high PAs/PANOs concentrations (ppm levels). The number of herbs samples was limited, as the PASFOOD project involved many food groups and herbs were not foreseen in the initial workflow. Further research on herbs was thus very desirable.
Consequently, the PASHERBS project aimed to generate new PAs/PANOs occurrence data in herbs (fresh, frozen, dry, individual or mixed) and related foodstuffs (prepared dishes, herbs based sauces, salted snacks…), using the methods developed in the course of the PASFOOD project. The analytical results showed that presence of pyrrolizidine alkaloids in these food commodities is widespread (151 samples). No PAs/PANOs were detected in individual fresh herbs in pot or cut. In other words, contamination in frozen herbs, dry herbs and meals with herbs could be attributed to PA plants accidental co-harvesting or intended adulteration. Individual dry herbs and mixes of herbs were the most contaminated herb types (up to 5240 ng/g in an Italian mix). Oregano was by far the most problematic aromatic herb regarding PAs contamination: up to 2200 ng/g in an individual oregano sample and the 10 most contaminated herbs mixes samples were all comprised of oregano.
Heliotrine-type compounds (europine, heliotrine, lasiocarpine and related N-oxides) were the most recurrent contaminants in the present study (64% of the overall contamination). Five herbs samples were in common in both PASFOOD and PASHERBS projects (same brands, same ingredients) with sampling respectively performed in August 2016 and October 2017. The PAs concentrations for both identical samples sometimes differed with time, but the contamination profiles were in very good agreement. This clearly showed that PA contamination in culinary herbs is recurrent, independently of time and lot number.
The concentration data generated in both projects for more than 1350 food items available on the Belgian market were linked to the actual consumption data obtained in the 2014 Belgian food consumption survey (FCS2014) for different sub-population with a special focus on children to assess their dietary intake to those natural contaminants.
For the semi-probabilistic exposure by food using mean concentration levels of the sum of 30 PAs/PANOs for each food category, the mean intake was of 14.71 ng/kg bw/day, 15.30 ng/kg bw/day and 16.72 ng/kg bw/day, respectively for children, adolescents and adults. The observed median and 95th percentiles were within the calculated ranges of EFSA in 2016 including data reported for all Member states: between the median and maximum for all age groups. Compared with the EFSA assessment for the Belgian FCS2004, we note a substantial increase in the exposure. It could be explained by different methodologies (more PAs/PANOs and more food groups included in the present study). It appeared that milk and dairy products were by far the main contributors to the overall PAs/PANOs intake for each age category, even though the concentration levels detected in this food group were the lowest. Senecionine-type and lycopsamine-type compounds were respectively the 1st and 2nd contributors to the PAs/PANOs dietary intake. The introduction of analytical concentration data from prepared dishes (mainly pizzas and lasagna), sauces and snacks affects adolescents more than children and adults. This is not a surprise, as this age group is more exposed to these types of foods.
Because of lacking consumption data for these food items, it should be noted that a separate deterministic scenario was used to evaluate the exposure to PAs by dry herbs; its results suggest an increase with 15% (adults) up to more than 50% (children) of the semi-probabilistic scenario. With this conservative approach, the mean dietary exposure (by adding the semi-probabilistic and deterministic results) would still be within the EFSA ranges for all ages. For the dry herbs, the pattern of PAs type present is noticeable: heliotrine-type PAs/PANOs contribute more than half to the estimated exposure to all 30 PAs/PANOs by dry herbs.
For the chronic risk assessment, the EFSA CONTAM Panel concluded that all 1,2-unsaturated PAs share a common metabolic pathway leading to the formation of genotoxic and carcinogenic reactive pyrroles. Therefore, it was decided to apply a Margin of Exposure (MOE) approach for genotoxic carcinogens to the sum of 1,2-unsaturated PAs, assuming equal potency. The CONTAM Panel selected the BMDL10 of 237 µg/kg bw/day derived for the incidence of liver haemangiosarcoma in female rats exposed to riddelliine as reference point for the chronic risk assessment of PAs. The EFSA Scientific Committee concluded that, for substances that are both genotoxic and carcinogenic, a MOE of 10000 or higher, based on a BMDL10 from an animal study, would be of low concern from a public health point of view. The risk assessment performed with this MOE approach yielded insufficient values for the 3 targeted age groups for the highly exposed population (≥ P95), using mean analytical levels for the sum of 30 compounds. The MOEs values for lycopsamine and senecionine subgroups indicate a possible concern for the health of the highly exposed population at P97.5 and P99, whereas monocrotaline and heliotrine subgroups are of low concern. It should be highlighted that dry herbs could not be included in the risk assessment because of the use of a separate deterministic approach to evaluate the exposure. This leads to an underestimation of the risk, particularly for the heliotrine subgroup, which was the most abundant in these food items.
Based on recommendations from the CONTAM Panel, EFSA recently proposed to reduce to 17 the number of relevant PAs and PANOs to be monitored in food, excluding europine, europine-N-oxide, heliotrine and heliotrine N-oxide. The concentration levels detected in the present study would have been drastically lower if these 4 PAs/PANOs would not have been included. As a conclusion based on both PASFOOD and PASHERBS projects results, we advise to keep targeting these 4 compounds for further PAs/PANOs monitoring in food. In line with the EFSA scientific opinion, the monocrotaline-type PAs/PANOs are not relevant in further risk assessments because of very low exposure levels.