**1. Introduction**

Pyrrolizidine alkaloids (PAs) are toxic substances that exist naturally in plants [1]. To date, over 660 types of PAs and PA *N-*oxide have been identified in the estimated six thousand plants [2]. 1,2-unsaturated PAs are particularly of concern as they are hepatotoxicants and genotoxic carcinogens [3]. 1,2-unsaturated PAs can be subdivided by the type of esterification, including monoesters, open chained diesters, and cyclic diesters. In addition, cyclic diester PAs with an azacyclooctanone, instead of a 1,2-dehydropyrrolizidine ring system, represent a special class. The main human exposure route to PAs is consuming plant-derived foods, such as (herbal) teas. PAs can induce hepatotoxicity both in humans and animals [1]. Human poisoning and even deaths from the consumption of PAs have been reported in several countries [4]. Severe outbreaks of the PA contamination once occurred in Afghanistan [5] and central India [6]. In addition, a few cases were reported in Hong Kong [7], Switzerland [8,9], Austria [10] and Tajikistan [11]. As a result, the use of the PA-containing plants as food products or supplements has been restricted in several countries [12]. However, there is still a lack of global consensus in regulatory measures regarding PAs in plant-derived products so far. Possibly, conducting a risk assessment for foodstuff containing PAs may contribute to the development of such a consensus.

**Citation:** Chen, L.; Zhang, Q.; Yi, Z.; Chen, Y.; Xiao, W.; Su, D.; Shi, W. Risk Assessment of (Herbal) Teas Containing Pyrrolizidine Alkaloids (PAs) Based on Margin of Exposure Approach and Relative Potency (REP) Factors. *Foods* **2022**, *11*, 2946. https:// doi.org/10.3390/foods11192946

Academic Editor: Oscar Núñez

Received: 19 August 2022 Accepted: 15 September 2022 Published: 21 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

(Herbal) teas are a type of PA-containing plant-derived product. The concentration of PAs can vary enormously among (herbal) teas. For instance, the Federal Institute for Risk Assessment (BfR) analyzed 274 types of (herbal) teas and found that the levels of PAs ranged from below the level of detection (LOD) to 5647.2 μg/kg dry material (d.m.) [2]. Later, Mulder et al. (2015) reported that the concentration of PAs could reach up to 4804.5 μg/kg d.m. based on the analyzed 22 types of (herbal) teas [13].

Due to the omnipresence of PAs in (herbal) teas and their detrimental effects, the safety evaluation of PAs associated with the consumption of teas is crucial. Multiple studies have performed risk assessments for PAs in (herbal) teas in the recent years [3,14,15]. Given that 1,2-unsaturated PAs are genotoxic and carcinogens, the risk assessment was conducted based on the margin of exposure (MOE) approach [16]. The MOE is defined as the ratio between the benchmark dose level with a lower confidence limit associated with a 10% extra risk on a cancer incidence above background levels (BMDL10) and the estimated daily intake (EDI). To date, the BMDL10 values for only two PAs, lasiocarpine and riddelliine, have been derived, whereas the values for the other PAs remained unavailable due to the lack of appropriate animal carcinogenicity studies [17,18]. The European Food Safety Authority (EFSA) used to adopt a BMDL10 of 0.07 mg/kg bw/day of lasiocarpine as the point of departure (PoD) for the MOE calculation [19]. A cut-off value of 10,000 for the MOE is usually applied, which incorporates factors including the inter-species and intra-species differences in toxicokinetic and toxicodynamics, the inter-individual human variability in cell cycle control and DNA repair as well as the potential discrepancy between the BMDL10 serving as a reference point and a NOAEL (No Observed Adverse Effect Level) [16]. The risk assessment suggested that the long-term consumption of several (herbal) teas may pose a potential health risk in humans, especially when considering a lifetime exposure [3,14,15]. It is worthwhile to mention that the MOE values were calculated based on the mean levels of total PAs in these studies, assuming that the metabolism and toxic potencies of PAs were the same with lasiocarpine. This, however, may result in an overestimation of potential risks from the exposure, since the toxic potencies of individual PAs are distinct and most of them could be lower than lasiocarpine. In addition, the obtained BMDL10 value for lasiocarpine (70 μg/kg bw/day) was affected by a high degree of uncertainty [20]. Instead, the EFSA has proposed the BMDL10 of riddelliine, which was 237 μg/kg bw/day, for the combined risk assessment of PAs by dose addition. The relative potency factor (REP) correction serves as an approach for the risk assessment for a mixture of chemicals that exhibit a common mode of action. To derive a REP, the potency of each component in a mixture is compared to that of a reference chemical generating a measure of potency for each component with respect to the toxicity of the index chemical [12]. It is more rigorous to perform risk assessments for PA-containing botanicals by determining the REP factor for each PA contained and then adjusting the concentration of each PA in the mixtures for the assessments.

To date, little is known about risk assessments for PAs in botanical samples when taking into account the REP factor of each PA. Additionally, the BMDL10 of riddelliine was proposed as the new PoD for the MOE calculation. Considering these facts, the actual exposure and related risk assessment of PAs due to the consumption of (herbal) teas need to be reevaluated. Therefore, the aim of the present study is to determine the PA levels in 21 types of (herbal) teas with and without the correction of REP factors, and to perform risk assessments based on the BMDL10 of riddelliine derived MOE approach. To achieve this, in total 68 individual 1,2-unsaturated PAs were analyzed, including cyclic diesters and heliotridine-type (7S) open diesters (e.g., monocrotaline, retrorsine, riddelliine, senecionine, seneciphylliine, senkirkine, heliosupine and lasiocarpine), heliotridine-type (7S) monoesters (e.g., echinatine and heliotrine), retronecine-type (7R) open diesters (e.g., echimidine and symphytine), and retronecine-type (7R) monoesters (e.g., indicine, intermedine and lycopsamine) monoesters, open chained diesters and cyclic diesters. The chemical structure for these PAs and their *N-*oxide form, as well as their corresponding REP factors, has

been previously reported in detail [12]. To clarify the rationale for this study, a graphical workflow is shown as Scheme 1.

**Scheme 1.** Graphical workflow for the whole analytical procedure in this study.
