Development of a Consensus Approach for Botanical Safety Evaluation – A Roundtable Report

Abstract

Botanical safety science continues to evolve as new tools for risk assessment become available alongside continual desire by consumers for “natural” botanical ingredients in consumer products. Focusing on botanical food/dietary supplements a recent international roundtable meeting brought together scientists to discuss the needs, available tools, and ongoing data gaps in the botanical safety risk assessment process. Participants discussed the key elements of botanical safety evaluations. They provided perspective on the use of a decision tree methodology to conduct a robust risk assessment and concluded with alignment on a series of consensus statements. This discussion highlighted the strengths and vulnerabilities in common assumptions, and the participants shared additional perspective to ensure that this end-to-end safety approach is sufficient, actionable and timely. Critical areas and data gaps were identified as opportunities for future focus. These include, better context on history of use, systematic assessment of weight of evidence, use of in silico approaches, inclusion of threshold of toxicological concern considerations, individual substances/matrix interactions of plant constituents, assessing botanical-drug interactions and adaptations needed to apply to in vitro and in vivo pharmacokinetic modelling of botanical constituents

1. Introduction

There is continued and increasing interest in botanicals in a variety of consumer care products. Recognition that the processing of a naturally sourced ingredient and incorporation into a dietary/food supplement can change the chemical composition, and a strong desire to reduce reliance on, and /or eliminate use of animals in toxicity testing, all present unique challenges for this category of ingredients (Brickel et al., 2018). There are various approaches for assessing the safety of botanicals, including the review of adverse events through post-market surveillance and a tiered analysis of existing and new data specifically generated to address safety assessment needs. For example, critical data endpoints that are often lacking include raw material characterization, Absorption, Distribution, Metabolism and Excretion (ADME), 90-day subchronic toxicity to establish “no observed adverse effect levels”, genotoxicity, developmental and reproductive toxicity, and when necessary carcinogenicity. Other endpoints that should be considered include but are not limited to botanical-drug interactions (BDI) and botanical-botanical interactions (BBI).

To expand on the topic of botanical safety assessments, a scientific session entitled, “Botanical safety evaluation in the era of alternatives,” was held at the 53rd Congress of the European Societies of Toxicology (EUROTOX) in Bratislava, Slovakia (September 10–13, 2017) (Rietjens et al., 2017). This session was followed by a multi-stakeholder Roundtable Event (hereafter “roundtable”) (Griffiths et al., 2018), which built on the elements shared by presenters during the scientific session.

The first objective of the roundtable was to review decision tree approaches for botanical evaluation which were described previously by the European Food Safety Authority (EFSA; 2009) and Little et al (2017) and were shared in the EUROTOX session. Focus of the roundtable was on the decision tree approach recently proposed by Little et al (Fig. 1) which starts with accurate chemical identification of the constituents of the botanical raw material, then to understand if those constituents with known structures are commonly consumed in the diet and whether the food/dietary supplement1 exposure is consistent with intake via food for these constituents. For those constituents that are not commonly consumed, or are above dietary intake levels, published safety data are reviewed to determine if an adequate margin of safety can be established to support the use of the botanical in the food/dietary supplement. If sufficient data are not available, a systematic evaluation is then conducted by applying quantitative structure-activity relationships (QSAR). In the absence of toxicological data, and once the chemical structure is known, the exposure can be compared to established Thresholds of Toxicologic Concern (TTC). Where safety endpoint gaps are identified which cannot be resolved without conducting in vitro or in vivo studies, then an understanding of the botanical compositional data is critical to inform on study design, and in test article selection. A similar approach is being applied to the safety evaluation of natural flavour complexes used as ingredients in food (Cohen et al., 2018), which also involves grouping together of constituents with similar ADME profiles and common toxicological properties, and comparing the intake relative to established TTC values.

Figure 1. Decision Tree for Botanical Constituent(s).

Used with permission, from “In silico approach to safety of botanical dietary supplement ingredients utilizing constituent-level characterization,” by Little et al., 2017, Food Chem. Toxicol. Pt A, 418–429.

Review of this decision tree approach (Fig. 1) during the roundtable provided an opportunity for stakeholders to:

• Share perspective on this approach (i.e., is there agreement to the key elements that are needed to build a robust botanical safety evaluation)

• Highlight and debate any vulnerabilities in the decision tree approach

• Share additional perspectives to ensure this end-to-end safety approach is sufficient and actionable

The second objective of the roundtable focused on a number of potential consensus statements (Summarized in Table 1), and provided an opportunity for the group to discuss, amend and align on these statements.

TABLE 1.

Aligned consensus statements from Roundtable.

Statement number*	Aligned Statement
1	Critical to the safety evaluation of a botanical, is clarity on the identity of the plant and plant part, along with an understanding of the constituents, method of botanical preparation and extraction, and final daily dose.
2	When applying a weight of evidence approach, evaluate the relative robustness of a review/evaluation, which could range from high to low dependent on the source and quality of the review and supporting references, the scientific quality of the studies that have been conducted, and data sources supporting history of use.
3	The safety of a botanical cannot be judged based solely on a history of food use unless it can be demonstrated that a comparable composition is ingested on a regular basis across broad geographic and demographic populations.
4	In the assessment of a botanical, it is misleading to assume that a history of human use addresses all aspects of safety.
5	Integration of in silico approaches broadens the dataset and may provide information on structural alerts that suggest the potential for certain biological properties.
6	In the absence of toxicity data, TTC is a worst-case risk assessment tool that can be applied to individual constituents to support their safety.
7	The application of TTC to a botanical product needs to consider dose addition, which is relevant if assessing constituents in the same chemical class and/or with a common mode of action.
8	There is a requirement for guidance on how to generate and interpret ADME-related data on complex botanical mixtures. This guidance will include how to incorporate physiologically-based pharmacokinetic (PBPK), and other kinetic models, to extrapolate to in vivo relevance.

^*Statement number relates to the discussion by statement below

2. Summaries of the presentations given at the roundtable event and key discussion points:

The roundtable event kicked off with four short topical presentations from experts.

As a member of EFSA Scientific Committee involved in the preparation of the Guidance on Safety Assessment of Botanicals and Botanical Preparations Intended for Use as Ingredients in Food Supplements (EFSA, 2009), Professor Corrado Lodovico Galli, from the University of Milan (Italy), opened the roundtable and presented on “Botanical Safety Evaluation in the Era of Alternatives” which revisited the content of the 2009 EFSA guidance document. He noted that there were two main EFSA concerns, and these remain the same today: 1) characterization of the test material and 2) contamination (e.g., chemical, microbiological). He shared the EFSA tiered approach, comprising of Level A – no testing required (assumed presumption of safety), and Level B – further testing and/or data required, and supported this approach with case studies on wheat bran (Triticum aestivum), Common Wormwood (Artemisia absinthium), which may contain the neurotoxin thujone, and fennel (Foeniculum vulgare), which may contain estragole, a potential genotoxic and carcinogenic toxin.

This presentation primed the group with regards to the concepts and the EFSA approach. It raised discussion around the point of “significant” versus “traditional” history of use and what these definitions mean. A significant history of use is used to describe the presumption of safety where there is dietary exposure for a number of generations and in a large, genetically diverse population. This also considers a scientific evaluation of the supporting information. In contrast, traditional use is based on exposure and knowledge in a specific population and may have limited scientific documentation (Roe et al., 2018). On the point on long term history of food use and its application to support a food/dietary supplement, it was noted that food consumption can be a different presentation of the material to the gut, and a concentrated extract may be more like a bolus dose (depending on the dissolution profile). Also, it was noted that there are gaps in exposure, toxicokinetic, and metabolism data.

The second presentation was from Dr. Nigel Walker, Deputy Director for Research for the Division of the National Toxicology Program (NTP) at the National Institute of Environmental Health Sciences (NIEHS) who presented on new approaches addressing the challenge of evaluating the safety of botanical dietary supplements. Botanical dietary supplements are complex, containing multiple constituents responsible for efficacy and/or toxicity. The United States NTP has had a long history investigating the possible safety liabilities associated with exposure to botanical dietary supplements, stemming from a workshop held in 1999 (Matthews et al., 1999). Some of the key recommendations from that workshop were; 1) the need for research on potential toxicity associated with high dose or prolonged use of botanicals, 2) identification and standardization of product ingredients, 3) increased consumer education, 4) identification of BDI and BBI, and 5) further research on potential risk to sensitive subpopulations. These recommendations were further supported by a workshop held in 2016 entitled “Addressing challenges in the assessment of botanical dietary supplement safety” (Shipkowski et al., 2018).

Since that time, NTP has evaluated numerous botanical dietary supplements in robust OECD-compliant long-term rodent-based toxicity studies conducted under GLP (Table 2), including seven of the top 20 supplements used in the United States.

TABLE 2.

NTP’s ongoing and completed studies on botanical dietary supplements.

Botanical Dietary Supplement	Common Use
ONGOING
Black cohosh	Menstrual and menopausal symptoms
Dong Quai	Menopausal symptoms
Echinacea Purpurea	Common cold and other infections
Evening Primrose oil	Eczema, menstrual and menopausal symptoms, breast pain
Garcinia Cambogia	Appetite suppression and weight loss
Gum Guggul	Lowering cholesterol, acne, and weight loss
Usnea Lichen	Weight loss
Valerian Root	Insomnia and other sleep disorders
Vinpocetine	Memory enhancement
COMPLETED
Aloe Vera	Constipation and gastrointestinal disorders
Bitter Orange	Heartburn, congestion, weight loss
Ephedra	Weight loss, energy, performance
Ginkgo Biloba	Brain function and memory
Ginseng	General well-being, improved physical stamina, and concentration
Goldenseal	Skin disorders, ulcers, fevers
Green Tea Extract	Mental alertness, digestive symptoms, headaches, and weight loss
Kava	Anxiety
Milk Thistle	Liver cirrhosis, chronic hepatitis, and gall bladder disorders, in addition to other ailments
Senna	Constipation, irritable bowel syndrome, hemorrhoids, and weight loss.

Rationale for selection for a program of work on a specific botanical is usually driven by a variety of factors based around exposure and potential for toxicity, and the key ones are captured below:

• Estimates of the size of the population using a given supplement/botanical through the use of sales figures can be used as a surrogate for exposure and priority given to test articles that are used to a high degree.

• The use pattern can be used to prioritize if there is potential for exposed populations to have long term use, or if exposure may occur to potentially sensitive populations (e.g., children and women of child bearing age).

• The potential for inherent toxicities can be assessed based on constituent profiles constituent profiles that may suggest specific biological activities which may be indicators for potential health impacts, particularly if the pattern of use differs from that recommended.

Typically, in such toxicology studies, following an initial phase of characterization of various commercial sources, a single test article is selected for evaluation. In this strategy there is an assumption that the chosen test article is representative of other available products with similar product labels. However, it is not clear how compositional differences among botanicals that have the same or similar plant source or product label relate to the biological activity of those samples that may impact on the toxicological results and/or safety.

To address some of these challenges, NTP has developed several case studies of botanicals that cover a range of chemical and biological profiles to explore the use of new approaches for determining phytoequivalence (termed “sufficient similarity”) of botanical dietary supplements. This approach includes targeted chemical analyses of known “actives” (or marker constituents) identified from the literature where authentic standards are available, which may or may not be biologically active, as well as non-targeted chemical “fingerprinting,” using various analytical techniques (e.g., HPLC with charged aerosol detection, LC/ MS, etc.) to identify more robust complex patterns for comparison and qualitative and quantitative identification of components whenever possible. In addition, possible contaminants such as metals, pesticides, mycotoxins, and total microbial counts are identified along with an understanding of the components that could impact the nutritional and safe status of the animal model when used in in vivo approaches (e.g., vitamins, metals, amino acids, fatty acids). These compositional data are then used in case studies to develop statistical approaches for the assessment of sufficient similarity using multidimensional biological similarity, in vitro and in vivo screening assays and genomic benchmark dose response modelling from subchronic rodent studies (Catlin et al., 2018; Shipkowski et al., 2018).

This approach is based on a fundamental assumption derived for the risk assessment of mixtures; that is that two mixtures are “sufficiently similar” if they are very close in composition, such that differences in their components and their proportions are “small.” It is then assumed that the toxicologic consequences of exposure to either of the mixtures will be identical or at least “indistinguishable from one another” (USEPA, 2000). Given these assumptions, then several questions are evident:

• What data should be used to assess similarity?

• Does chemical similarity reflect biological similarity?

• Can non-targeted chemical compositional analysis and informatic analyses be effectively used to understand mixture similarity?

• Are there objective criteria that can be used to assess divergence for “dissimilar” materials?

• Can a standard approach be applied across botanicals, or does it require adjustment on a case-by-case basis?

Initial work using a case study with Ginkgo biloba extracts supports that this approach of using chemical fingerprint similarity coupled with short term assessments of biological similarity can likely be used to predict that “untested” materials have similar safety profiles to those that have undergone more comprehensive toxicological testing.

Dr. Nicholas Oberlies, from the University of North Carolina at Greensboro, presented on characterization methods for natural (botanical) products, and shared perspective from his work in the field of chemical mycology. He presented a published framework (VanderMolen et al., 2017) that outlined a multi-faceted approach to evaluation of the safety of non-culinary fungi. He spoke about the issue of contamination and highlighted that accidental contamination, and intentional adulteration, must be considered. With respect to the European Food Safety Authority (EFSA) approach (Level A), he recommended taxonomic identification (Raja et al., 2017) and confirmation via analytical characterization (VanderMolen et al., 2017) for all botanicals.

This raised discussion about the level of characterization required and attempts to characterize 100% of the material. Various opinions were shared from the participants. For example, there could be a pre-set criterion that if using the raw material at over 1%, then full characterization (i.e., to near 100%) is required, with reduced requirement for level of characterization required at reduced usage amounts. To build on this, another participant raised the TTC approach, and how this exposure-based approach could help with decisions around characterization. Based on the dataset, a certain TTC would be applied for a specific chemical entity, and this would then drive the limit of detection and hence the level of chemical characterization. Dr. Oberlies was supportive of this approach, as it presents a pragmatic way to deal with the chemical complexity of a botanical in the context of potential toxicological and/or ADME concerns, understanding that it can be extremely challenging to characterize and identify every constituent driven by limitations in detection and matrix effects.

Finally, Dr. Amy Roe, provided context on the challenges in assessing ADME of these complex mixtures, including evaluation of BDI and BBI. The potential for natural ingredients to interact with drugs is an important consideration in an overall botanical safety assessment. This is largely because of a few well-known (and well-studied/characterized) interactions (Bailey et al., 1998; Henderson et al., 2002). Much has been made of BDI potential in the lay media and in scientific publications, so both consumers and healthcare professionals have difficulty making informed decisions about co-use of dietary supplements and drugs. In addition, global regulatory agencies are increasingly questioning whether interactions of herbal products with drugs have been taken into consideration.

Dr. Roe presented a published framework developed to assist with assessment of BDI potential (Roe et al., 2016). This framework includes key elements and considerations involved in that process, and it was noted that it encompasses many of the same considerations (history of safety use, existing literature data, analytical characterization of botanicals and constituents and dose performance data such as dissolution/ disintegration measurements) of botanicals and complex mixtures in general (VanderMolen et al., 2017; Roe et al., 2018).

Numerous reports of various BDI can be found in the scientific literature and are mostly from in vitro studies conducted in simple metabolic systems (e.g., hepatic microsomes). These screening-level studies in simplistic metabolic systems such as liver microsomes likely represent overly conservative findings. This is based on our opinion that these systems may overpredict since the concentration of phytochemical constituents at the enzyme active site will be higher than what is likely to occur in vivo (or in cell-based systems such as hepatocytes that have active efflux). Without follow-up studies in more physiologically-relevant models the clinical relevance of these initial findings is largely unknown. An additional problem is that these early screening findings are often conducted with individual constituents of botanical extracts and results are often not taken into context of the whole extract or the food matrix. Lastly, data from these types of studies are often transferred to the safety literature and on-line databases. Subsequent work has primarily focused on development of more physiologically-relevant in vitro liver models to study BDI such as sandwich-cultured human hepatocytes (SCHH). Most recently, a SCHH model was used to successfully predict clinically-relevant changes in hepatic clearance of select drugs when co-administered with the botanical, Schisandra spp. (Jackson et al., 2017). Although BBI are less well studied, likely due to the significant chemical complexity represented, similar approaches may be useful.

ADME-related data are critical to understanding the safety of botanicals. A number of points were shared on why ADME-related data are useful in botanical safety evaluation and included:

• Aiding in the design of toxicology studies (e.g., route of administration, dose selection and dosing paradigm, dose formulation)

• Determining that a lack of toxicological response is not due to lack of systemic exposure

• Providing key data necessary to develop and validate PBPK models to help link external exposure to internal or target site dose and to extrapolate animal data to human safety assessment

• Aiding in design of in vitro BDI studies and prediction of clinically-relevant results

The importance of ADME data for use in botanical safety assessments is well recognized. However, much work is needed to address the challenges of obtaining ADME data on complex mixtures represented by botanicals. Recommendations for assessing disposition parameters have recently been proposed (Waidyanatha et al., 2018).

These four presentations provided a helpful re-cap of the EUROTOX session (“Botanical safety evaluation in the era of alternatives”), introduced some new concepts (such as the ADME considerations) and provided opportunity for further discussion. The key elements of this discussion are described below.

Several questions were raised and briefly discussed regarding the genotoxic potential of botanicals. Where botanical extracts have tested positive in genotoxicity assays, uncertainty was expressed on the impact of trace or multiple minor botanical constituents or the relevance of these results for in vivo exposures to a complex mixture. The application of a threshold approach was considered reasonable in principle where the mechanism of action was determined to be indirect (non-DNA reactive). However, this approach was not considered suitable by some participants as a default if the botanical was simply identified as genotoxic. It was also noted that genotoxicity testing protocols for botanicals are challenging and must be well designed and controlled (e.g., sample extraction; risk of false positives and negatives). The roundtable conclusion was that genotoxicity testing of botanicals is a topic that could benefit from a follow-up workshop to define research needs.

History of safe use was discussed. Focus was on what weight should be put on human use experience along with when is human use experience sufficient? The starting position is that a food is presumed safe based on a broad exposure over a long period of time. However, when making this assumption the assessment also needs to consider how the role of the cooking process and how the food is consumed (i.e., the entire preparation and ingestion scenario must be assessed). This also highlights the need to understand the method of botanical preparation, as constituent content can be altered by the extraction methods applied. In conclusion to this point, it was agreed that there are challenges in how to capture and assess these types of data as they may be region specific or based on anecdotal reports, etc. For future learning, it may be that a new and broader approach (e.g., crowd sourcing), or more reliable historical text/documents are needed. In addition, it was noted that in the marketplace there are situation is widespread. Any approach should aim to help mitigate this challenge, with awareness that it will likely never be fully eliminated.

3. Discussion and consensus statements:

Pre-worked statements related to the decision tree approach for botanical safety evaluation were shared with the participants as noted above (Table 1), with time allocated for discussion and revision. The objective was to discuss and reach consensus on each statement. The final statements are presented below, with discussion points captured for each.

3.1. ALIGNED STATEMENT #1: Critical to the safety evaluation of a botanical, is clarity on the identity of the plant and plant part, along with an understanding of the constituents, method of botanical preparation and extraction, and final daily dose.

This statement recognizes the criticality of adequate chemical characterization but leaves open the question of how thorough this analysis needs to be to address safety. Discussed, but not resolved, was whether a cut-off level could be defined, and a pragmatic approach achieved. When consumer exposure to a botanical occurs, it is considered paramount to be able to identify what the consumer is exposed to and quantify that exposure. Understanding of how and where grown, harvesting and manufacturing processes were considered critical. These factors drive differences between test materials and underscore the importance of characterization and understanding the contribution of minor constituents. A follow-up question focused on the value of testing whole extracts versus a deconstructionist approach (fractionated components of the complex mixture) specifically in relation to BDI/BBI testing but no conclusion was made on this point. Instead, from the discussion there was a recommendation to develop guidance on a tiered-testing approach to build up the knowledge and data on a botanical extract.” Strong concerns were raised about the value or elevance of testing individual, minor constituents. Understanding contamination was considered a critical step in the characterization with evaluation regarded as context-specific depending on the type of contaminant and its source (agricultural, processing, manufacturing).

3.2. ALIGNED STATEMENT #2: When applying a weight of evidence approach, evaluate the relative robustness of a review/evaluation, which could range from high to low dependent on the source and quality of the review and supporting references, the scientific quality of the studies that have been conducted, and data sources supporting history of use.

The use of the term “relative robustness” was pulled out and discussed as there could be bias in how robustness is supported. It was noted that often the evidence for traditional use was anecdotal and safety concerns may involve single case reports. This highlighted an opportunity to encourage more systematic reviews of the peer-reviewed scientific literature, for example Cochrane reviews (CDSR, 2018) or the approach used by NTPs Office of Health Assessment and Translation (NTP, 2018) to include evidence for history of safe use. It was questioned whether the source and data being applied in publications are transparent, an enhancement that would build trust and ownership. A question was raised on the extent and challenges associated with pharmacovigilance process for botanicals (“nutrivigilance” or “safety vigilance”). The typical complexity of botanical food supplement products led to a discussion on the challenges associated with obtaining and trusting adverse event data in this area. This area was captured as a gap and follow up opportunity in Table 3.

Table 3.

Summary of topics for future guidance and/or workshop discussions.

Topic	Questions/challenges/Next steps
Concept of ‘sufficiently similar’.	How is ‘sufficiently similar’ to be defined? Is there an in-silico approach to answer this or will testing be required to determine? Are adequate in vitro/in vivo bioassay tools or other approaches currently available or will new tools need to be developed?
Level of characterization and identification for a botanical/complex mixture.	What level of characterisation is required and how is this determined? Are there more efficient approaches to address the challenges in developing analytical chemistry methods to measure either a large number and/or low levels of constituents. How are “minor” constituents defined? What is the cut off level for “minor” constituents?
TTC; A concept that was originally applied in a regulatory framework to account for individual contaminants in food. However, this discussion highlighted its value in characterizing constituents, regardless of their intent in a botanical material.	What is the role of reapplication of the TTC approach in the safety evaluation of botanicals? When considering dose addition, how is chemical class or common mode of action defined? Guidance would be required. Is there a need to consider synergistic action? If no, then need to document in guidance why not.
ADME; It is readily acknowledged that botanical extracts are complex. The discussion highlighted both the need and the added complexity in decisions to follow individual constituents in vitro and/or in vivo (animal and human clinical).	Which constituent(s) should be monitored intracellularly (in vitro studies) or in biological matrices such as plasma or urine (in vivo animal and/or clinical studies)? What about the potential metabolites of constituents? Need to develop guidance that includes recommendations for which in silico/in vitro models to use at which level of testing (i.e., tiered testing approach), how to include blood sampling as part of study design to characterize ADME parameters (in in vivo animal and/or clinical studies), and how to interpret ADME results in the context of clinical relevance. Dose performance data are often lacking (i.e., dissolution/disintegration of dose form). These data help determine potential gut concentrations and can have implications for predicting absorption and/or local metabolizing enzyme and transporter interactions. Novel formulations, particularly those aimed at enhancing bioavailability should be carefully studied for changes in systemic exposure from more typical in-market formulations.
Genotoxicity testing of botanicals; Testing of complex botanical materials in sensitive in vitro genotoxicity assays often creates false positives and/or uncertainties for extrapolation to humans in vivo.	What modifications are necessary to adapt current genotoxicity protocols to better understand botanicals? What considerations are critical in test sample preparation for botanical raw materials and botanical extracts? Does a botanical mixture modify the toxicokinetics and thresholds ordinarily applied to genotoxins?
History of safe use	Clarity is required on the limits associated with a history of safe use definition (e.g., how consumed, role of cooking), along with regional versus broad geographical and demographic consumption. Does the history of safe use extend to all populations (i.e. pregnant women or children)?
Systematic review of botanicals	Need to build criteria to guide the application of systematic review approaches to the assessment of botanicals; e.g., inclusion of consideration of assessment of botanical characterization, and patterns of use.
Pharmacovigilance/Nutrivigilance	What are the critical elements that are needed to adequately track and characterize adverse events for botanical-containing products? products (e.g. botanical identification, population diversity)?
Botanical constituents	Consider development of a database that captures botanical constituent information. Good opportunity to build knowledge and develop collaborative partnerships. To achieve, would Public/Private partnership be useful? Who would own? Guidance is needed on how to group into chemical class or via a common mode of action.

3.3. ALIGNED STATEMENT # 3: The safety of a botanical cannot be judged based solely on a history of food use unless it can be demonstrated that a comparable composition is ingested on a regular basis across broad geographic and demographic populations.

As a follow-on from the discussion of Statement #2, this discussion highlighted the importance of understanding the power of the data for reapplication to diverse contexts. Most notable in this discussion was attention to the historical context to assure a sufficiently diverse population accounting for genetic diversity and use by a broad enough cross-section of individuals (e.g., age, gender, disease state) to address the product’s intended target population.

3.4. ALIGNED STATEMENT # 4: In the assessment of a botanical, it is misleading to assume that a history of human use addresses all aspects of safety.

Phrased in the negative, this statement was included to underscore the concerns if the investigator chooses to ignore the principles behind Statements #2 and #3. While a history of human use may be documented, the integrity of the “nutrivigilence” data, the specific material used in the population investigated, and the diversity of that population (males vs females; pregnant women; children; etc.), and the consideration that food consumption is a different, lower dose presentation to the gut versus a concentrated bolus dose from a dietary supplement, are all critical factors underpinning the strength of the assumptions.

3.5. ALIGNED STATEMENT # 5: Integration of in silico approaches broadens the dataset and may provide information on structural alerts that suggest the potential for certain biological properties.

A tiered approach that enables systematic progression through a botanical safety evaluation was considered a useful approach for building a strong case of support. Starting with available information on the historical usage of the material and its safety profile coupled with sufficiently robust analytical characterization was considered critical for establishing the input elements for a systematic evolution of understanding with predictive power. For resolving constituent discrepancies, the inclusion of tools to evaluate chemo-toxic structural alerts (where available) was generally agreed to be valuable in decision-making. Where safety concerns remain for the test material (or sub-fractionate elements) there was consensus that test systems were needed for botanical materials that mimic in vivo human biology (physiologically relevant test systems).

3.6. ALIGNED STATEMENT # 6: In the absence of toxicity data, TTC is a worst-case risk assessment tool that can be applied to individual constituents to support their safety.

With this statement, there was a question on when, or at what level, TTC can be applied. Sufficient chemical characterization and identification of constituents coupled with comparator materials with comprehensive toxicology data were considered essential. For individual constituents that lack specific toxicity data, the TTC approach has been shown to be a reliable alternative to the generating of new test data for minor constituents. This conservative approach that assigns worst-in-class assumptions for each toxicity endpoint, can be a powerful tool to eliminate concerns for minor constituents. Another valuable perspective is to use the principles of TTC to establish a universal or chemical class threshold without necessarily establishing defined TTC levels for each material.

3.7. ALIGNED STATEMENT # 7: The application of TTC to a botanical product needs to consider dose addition, which is relevant if assessing constituents in the same chemical class and/or with a common mode of action.

On discussion of this statement, it was highlighted that dose addition will only apply to certain subsets of constituents, and it should be noted that this approach is already being applied in the application of the TTC approach to the safety evaluation of constituents of natural flavour complexes (Cohen et al., 2018). However, this discussion highlighted that more guidance is required to define the chemical classes and which constituents will fall into groups that show structural alert moieties. There was also a point raised about how to consider synergistic action. These questions are also relevant for the common modes of action.

3.8. ALIGNED STATEMENT # 8: There is a requirement for guidance on how to generate and interpret ADME-related data on complex botanical mixtures. This guidance will include how to incorporate PBPK, and other kinetic models, to extrapolate to in vivo relevance.

It was agreed that guidance is needed on how to collect and interpret ADME-related parameters for complex botanical mixtures. The discussion and final statement reflects the many points that should be considered in any guidance developed for this area. For example, how to identify marker constituents to extrapolate across in vitro, in vivo animal and/or human clinical studies, and consideration of interaction potential (BDI and BBI). Such guidance would also need to address the use of advanced intestinal and hepatic models and/or co-cultures, along with the utility of molecular docking models and other in silico approaches that may also prove useful, particularly for assessing interactions.

4. Challenges and/or Next steps:

This roundtable enabled the convening of key stakeholders across different functions and provided an opportunity for an open and productive discussion on assessing botanical safety. The discussion highlighted several key areas that remain challenging or require future guidance or workshop discussions. These are summarized below in Table 3:

Conclusion:

Botanical safety evaluation is a rapidly evolving science. Meetings such as this roundtable provide an opportunity for open exchange among scientists directly engaged in safety assessment of botanicals. The key elements of a robust botanical safety evaluation and application of a decision tree methodology were discussed. This event facilitated scientifically driven discussion which led to consensus statements. The consensus statements reflect a unified and pragmatic approach to evaluate botanicals for safe human use by determining dietary intake levels, sufficient safe margins of exposure from existing safety information, and/or establishing safe exposure thresholds via TTC/QSAR approaches. Additionally, critical areas and data gaps were identified as opportunities for future focus including better context on history of use, application of systematic assessment of weight of evidence, fit-for-purpose animal alternative methods and adequate in vitro tools for botanicals, and guidance on incorporating ADME parameters of botanical constituents for better understanding of extrapolations from in vitro to clinical data.

Abbreviations/Acronyms:

BDI

botanical-drug interaction

BBI

botanical-botanical interaction

ADME

absorption, distribution, metabolism, excretion

EUROTOX

European Societies of Toxicology

EFSA

European Food Safety Authority

QSAR

quantitative structure activity relationships

TTC

threshold(s) of toxicologic concern

PBPK

physiologically-based pharmacokinetics

NTP

National Toxicology Program

NIEHS

National Institute of Environmental Health Sciences