Advancing Botanical Safety: A Strategy for Selecting, Sourcing, and Characterizing botanicals for developing Toxicological Tools

Abstract

Increases in botanical use, encompassing herbal medicines and dietary supplements, have underlined a critical need for an advancement in safety assessment methodologies. However, botanicals present unique challenges for safety assessment due to their complex and variable composition arising from diverse growing conditions, processing methods, and plant varieties. Historically, botanicals have been largely evaluated based on their history of use information, based primarily on traditional use or dietary history. However, this presumption lacks comprehensive toxicological evaluation, demanding innovative and consistent assessment strategies. To address these challenges, the Botanical Safety Consortium (BSC) was formed as an international, cross-sector forum of experts to identify fit-for purpose assays that can be used to evaluate botanical safety. This global effort aims to assess botanical safety assessment methodologies, merging traditional knowledge with modern in vitro and in silico assays. The ultimate goal is to champion the development of toxicity tools for botanicals. This manuscript highlights: 1) BSC’s strategy for botanical selection, sourcing, and preparation of extracts to be used in in vitro assays, and 2) the approach utilized to characterize botanical extracts, using green tea and Asian ginseng as examples, to build confidence for use in biological assays.

1. Introduction:

The use of botanicals, including dietary supplements and herbal medicines, is increasing globally. In the United States, sales of botanical supplements increased by over 17% in 2020, with consumers buying botanicals for reasons including the purported improvement of immune health, stress relief, and COVID-19 prevention and treatment (Smith et al., 2021). The trend continued in 2021, with a further increase of almost 10% (Smith, 2022). Botanicals, which we define to include plant-derived products but also algae and fungi, are complex mixtures containing hundreds to thousands of constituents. Many of these constituents are involved in the normal function of the plant, including growth, development, or reproduction (i.e., primary metabolites). However, some constituents (i.e., secondary metabolites) serve adaptive or protective roles such as having pesticidal, antifungal, antiviral, or antibiotic activity (Carreno-Quintero & Cárdenas, 2021). Many have been utilized by humans as pesticides, pharmaceuticals, or other products. Famously, aspirin is based on the natural product salicin from Salix alba, a willow tree (Desborough & Keeling, 2017).

Products derived from botanicals can range from whole plant material (e.g., aloe vera plant used to treat sunburns) to a purified single chemical (e.g., the pharmaceutical digoxin from Digitalis purpurea). The constituent composition of botanical products can vary for many reasons, including, but not limited to, growing conditions, post-harvest processing, extraction solvent, and purification steps (Belwal et al., 2018; Kriker et al., 2013; Orians et al., 2019; Ryan et al., 2019). The constituent content can also vary by plant variety (i.e., a cultivar) or the part of the plant used (e.g., root or leaves) (Cho et al., 2015; Touati et al., 2022). Botanical products are available in various forms, such as teas, tinctures, capsules, gummies, and salves. The inherent and added complexity of botanicals makes them difficult to assess for toxicity and efficacy (Mitchell et al., 2022). Global regulatory requirements for botanicals vary, but often these products have a presumption of safety based on demonstrating a history of use either as dietary supplements (e.g., green tea leaf) or through traditional medicinal use (e.g., Asian ginseng in traditional Chinese medicine). The removal of a product from the market often requires post-market surveillance evidence of safety issues (e.g., adverse event reporting as in the case of ephedra supplements (FDA, 2004)). In cases where botanicals lack history of use data (which does not consider factors like delayed effects [e.g., cancer] or susceptible subpopulations [e.g., people with diabetes]), additional preclinical toxicity data may be needed. However, testing in mammalian models is resource intensive. Additionally, data generated may not be as useful due to challenges in selecting a representative material for testing because of the variety of products on the market and variable composition of botanicals. Lastly, regulatory agencies have called for action to move away from animal testing to more predictive approaches (FDA, 2017). All these factors lead to the need for new approach methodologies to evaluate toxicity of botanicals (Jităreanu et al., 2023).

Given the need for safer botanical products and the difficulties in assessing their toxicity, the Botanical Safety Consortium (BSC) was formed with experts from various fields, such as natural products chemistry, toxicology, data analysis, and pharmacognosy (Mitchell et al., 2022). This international, cross-sector group of experts is working to identify fit-for-purpose in vitro and in silico assays that can be used to evaluate botanical safety. Ultimately, these assays can be utilized by the manufacturers to develop safer botanical products and promote their responsible use. In this manuscript, we focus on the need for high-quality and fit-for-purpose chemical analysis data to support the use of botanical materials in biological assays.

We describe our strategy for selecting botanicals, preparing and characterizing botanical extracts to build confidence for use in biological assays.

2. BSC Chemical Analysis Working Group Mission

The BSC’s Chemical Analysis Working Group mission is to develop a strategy and methodologies to characterize botanicals for the purpose of enabling safety assessments. The four main objectives are described below (Figure 1).

Figure 1:

Objectives of the Botanical Safety Consortium’s Chemical Analysis Working Group; compile literature, identify methods and approaches, select botanicals, and maintain a library of botanicals.

• Compile existing literature on chemical composition of botanicals selected.

• Identify resource-efficient analytical approaches and partners that can characterize botanicals to the degree required for material qualification and testing.

• Select botanicals for characterization based on the needs of the other Working Groups.

• Maintain a botanical inventory including ensuring material stability to cover study duration.

These objectives allow for the generation of robust data which can be compared widely with other studies reported in the literature. There are several key considerations when selecting a botanical material for testing: 1) it should be authentic, belonging to the correct plant genus and species, 2) it should be prepared in a form suitable for use in bioassays, 3) it should be contaminant and adulterant free (within accepted guidelines), 4) the material should be characterized to the level required by safety considerations and possible by currently available technology, and 5) knowledge of the storage stability of the material and dosing solutions for the duration of the study is needed.

3. STRATEGY

This section describes the overall strategy (Figure 2), from selecting and sourcing botanicals, preparing the materials (i.e., extracts) for biological assays, characterizing materials, and preparing solutions for in vitro assays.

Figure 2:

General strategy of the Botanical Safety Consortium’s Chemical Analysis Working Group. HPTLC, high-performance thin layer chromatography.

3.1. Botanical Selection

To determine the suitability of in vitro toxicological assays for botanicals as complex mixtures, botanicals with well reported evidence of toxicity (or lack thereof) were selected (Table 1) for various endpoints (i.e., neurotoxicity, cardiotoxicity, hepatotoxicity, development & reproductive toxicity, genotoxicity, and botanical-drug interactions). Evidence included a combination of human clinical trials, adverse event reporting, animal data, and mechanistic data. The botanical materials selected (e.g., plant parts) were meant to closely mirror the type of materials tested in those studies – for example, green tea leaf extract (GTE) selected for BSC activities was based on similar (−)-epigallocatechin-3-O-gallate (epigallocatechin gallate, EGCG) content to that tested by the National Toxicology Program in their animal studies and described by USP in their monograph (NTP, 2016; Oketch-Rabah et al., 2020). Additionally, botanicals were typically selected for one or two endpoints, but the status of the ability to induce a response in another endpoint was often unknown (e.g., comfrey has evidence of genotoxicity and hepatotoxicity (Mei et al., 2010) but it has not been evaluated for developmental toxicity). In this effort, the selected botanicals are used as tools to understand the suitability of toxicity assays. We are not making safety determinations on individual botanicals.

Table 1 –

List of botanicals selected by the Botanical Safety Consortium, including their standardized common and scientific names, Distributed Structure-Searchable Toxicity (DSSTox) substance identifier (DTXSID), supplier, and the part(s) of the plant used to derive the botanical extract.

Standardized Common Name	Scientific Name	DTXSID1	Supplier	Plant part(s)	Type of extract
Aconite	Aconitum napellus L.	DTXSID701061676	University of Mississippi National Center for Natural Products Research, Oxford, MS	Mixed parts	95% ethanol extract
Aristolochia fangchi	Aristolochia fangchi Y.C. Wu ex L.D. Chou & S.M. Hwang	DTXSID201349132	China Medical University	Root	95% ethanol extract
Ashwagandha2	Withania somnifera (L.) Dunal	DTXSID201042372	Verdue Sciences	Root	LWS7EP2005K31; Ethanol: Water ~15:1) extract; 1.56% total withanolides
Asian ginseng2	Panax ginseng C.A. Mey.	DTXSID1023780	Euromed S.A.	Root	Ginseng Dry Extract 4% (Quintozene free); 4.6% total ginsenosides
Blue cohosh	Caulophyllum thalictroides (L.) Michx.	DTXSID401042859	Mountain Rose Herb	Root and rhizome	95% ethanol extract
Comfrey	Symphytum officinale L.	DTXSID20274226	Mountain Rose Herb	Root	95% ethanol extract
Ephedra	Ephedra sinica Stapf	DTXSID801018482	China Medical University	Aerial parts	95% ethanol extract
Green tea2	Camellia sinensis (L.) O. Kuntz	DTXSID0031398	Indena S.p.A.	Leaf	Green tea dry decaffeinated extract; 78.5% total catechins (54.6% (−)-epigallocatechin-3-O-gallate)
Goldenseal	Hydrastis canadensis L.	DTXSID40274228	Mountain Rose Herb	Root and rhizome	95% ethanol extract
Kava	Piper methysticum G. Forst.	DTXSID901018742	Mountain Rose Herb	Root and rhizome	95% ethanol extract
Kratom	Mitragyna speciosa (Korth.) Havil.	DTXSID001334842	Ethos Natural Medicine LLC	Leaf	95% ethanol extract
Milk thistle	Silybum marianum (L.) Gaertn.	DTXSID8031657	Euromed S.A.	Seed	Milk thistle dry extract; 90.6% of silymarin isomers calc. as silibinin
Oleander	Nerium oleander L.	DTXSID201042091	University of Mississippi National Center for Natural Products Research	Leaf	95% ethanol extract
Usnea	Usnea spp.	DTXSID701349537	Mountain Rose Herb	Whole lichen	95% ethanol extract
Tripterygium3	Tripterygium wilfordii Hook. f.	DTXSID301349830	University of Mississippi National Center for Natural Products Research	Root	95% ethanol extract
Yohimbe	Pausinystalia johimbe (K. Schum.) Pierre ex Beille	DTXSID4032291	Afrigetics Botanicals	Bark	95% ethanol extract

¹Substance identifier used in Environmental Protection Agency CompTox dashboard (https://www.epa.gov/chemical-research/comptox-chemicals-dashboard). Note, DXTSIDS have been created for all plants but are iteratively added to the EPA site

²Procured as a commercial extract

³Also commonly known as thunder god vine.

3.2. Sourcing

Identifying a reliable source of authentic plant material is essential for biological assay evaluations of botanicals (i.e., validated genus and species). Although plant material or products deriving from them can be obtained from various sources (e.g., grocery stores, health food stores, herbal companies, and online suppliers) authenticity or potential adulteration (i.e., economically motivated or unintentional) of such material is typically unknown. In the BSC, we sought to get high quality, raw materials from growers or procure unformulated materials (e.g., extracts) from commercial sources that match materials used in existing literature. Sources were recommended by experts within the consortium. Table 1 describes the sources of all the botanicals obtained. At this stage (with the exception of ashwagandha root extract and GTE), one lot of each botanical was obtained. For ashwagandha root extract and GTE up to 3 lots were procured and screened to select the final lot for BSC activities. Future work can include comparing multiple lots or even commercial products to understand the variability of materials on the market (as has been done by others (Quiroz-Delfi et al., 2023)), but this is beyond the scope of this phase of work for the BSC.

3.3. Material Preparation and Solubility Determination

The choice of plant part(s) and the processing condition(s) (e.g., extraction method and solvent(s)) depends on the phytochemical composition required for the intended pharmacological properties of a given botanical. Alcohols, ethyl acetate, ethers, water (or a combination thereof), and supercritical carbon dioxide are widely used as extraction solvents.

Except for those purchased as extracts as described in Table 1, the botanical extracts used in the BSC were prepared by the extraction of the ground plant materials in 95% ethanol to maintain a consistent process. As a polar organic solvent, ethanol is capable of extracting phytochemicals with differing chemical properties representing the composition of the botanical. This extraction method was selected based on input from analytical chemists, pharmacognosists, and toxicologists. Although the extract preparation here does not necessarily resemble procedures used for botanical material preparation when guided by their pharmacological properties, it is adequate for the current goal of the BSC, which is to establish bioassays towards different toxicity endpoints. Additionally, as given in subsequent sections, state-of-the-art analytical approaches were used to ensure the appropriate phytochemical composition of the prepared extracts prior to use in bioassays.

While these were prepared as ethanol extracts, many bioassays use dimethyl sulfoxide (DMSO) as a solvent. Therefore, solubility of the extracts was determined in both ethanol and DMSO and the data are presented in Table 2. As evident from the data, in many cases, the materials were equally or more soluble in DMSO when compared to 95% ethanol.

Table 2:

Solubility of botanical extracts in 95% ethanol and dimethyl sulfoxide (DSMO).¹

Botanical Extract	95% Ethanol solubility (mg/mL)	Dimethyl sulfoxide (DSMO) solubility (mg/mL)
Aristolochia fangchi	5	83
Ashwagandha	8	5
Asian ginseng	4	11
Blue cohosh	26	125
Comfrey	6	83
Ephedra	15	632
Goldenseal	46	125
Green tea	167	125
Kava	59	6
Kratom	17	21
Milk thistle	13	167
Usnea lichen	11	832
Yohimbe	19	63

¹Data are given for only 13 botanicals.

²At this concentration material was partially soluble with a layer of insoluble material on top.

3.4. Evidence of Authenticity

To ensure the accuracy and reliability of bioassay evaluation, it is important to ensure that botanical materials used belong to the correct plant and with appropriate phytochemical composition. There are several cases in the literature that illustrate the importance of determining the authenticity and composition of a botanical. One famous example was the confusion of the names gang fang ji (Aristolochia fangchi) with han fang ji (Stephania tetrandra). This led to an outbreak of aristolochic acid-induced kidney disease in over 100 patients in Belgium in the 1990s, with over 70% of them developing end-stage renal disease (Anger et al., 2020; Debelle et al., 2008; Vanherweghem et al., 1993). More recently, “lavender” products allegedly causing endocrine disruption in children were found to be devoid of lavender essential oil (Gafner et al., 2022; Larkman, 2020). Unfortunately, mislabeling of bulk botanical ingredients and botanical dietary supplement, personal care, and cosmetic products appear to be quite common (Ichim, 2019), especially in cases in which the labeled ingredient is purposefully replaced with materials of lower quality (i.e., economically motivated adulteration) (Ichim & Booker, 2021). Hence, the determination of ingredient identity is the first step in the characterization of botanical ingredients prior to any safety or efficacy testing (Patel et al., 2023).

Botanical authentication techniques gives confidence that the correct species has been acquired (van Breemen et al., 2007) and is essential knowledge for the BSC prior to the use in bioassays. When the source material is available there can be evaluations using plant taxonomy, morphology, and genetics. Taxonomic examination includes macroscopic and microscopic methods, and microscopic identification can be used even if the plant material has been milled. Genetic identification uses DNA isolated from intact or milled plants and the application of Polymerase Chain Reaction (PCR) techniques with comparison to authentic voucher material. As an alternative to DNA analysis, immunoassays may be used to detect species-specific proteins. If processing such as solvent extraction has eliminated DNA and proteins, High Performance Thin Layer Chromatography (HPTLC) is widely used as a chemical authentication method for botanical extracts along with authentic voucher material (Reich et al., 2019). The BSC utilized multiple analytical methods as described in subsequent sections to support the identification of the botanicals tested.

The National Center for Complementary and Integrative Health (NCCIH) of the National Institutes of Health (NIH) provided guidelines to ensure the quality and reproducibility of their research involving botanicals by confirming botanical integrity (Simmler et al., 2015). These guidelines include identifying each botanical using scientific taxonomic nomenclature and author citation (e.g., Panax ginseng, C.A. Mey), the parts of the plant used, processing procedures such as extraction solvent, and the botanical source or supplier. In terms of authenticating botanicals, it is important to specify the method used for species identification. Additionally, if raw plant material is involved, the location of a verified voucher specimen should be documented for future reference.

The use of botanical reference materials to enhance the verification of a botanical is becoming popular amongst researchers. However, reference materials are not available for all botanical ingredients and there are limited sources providing these materials. The National Institute of Standards and Technology (NIST, www.nist.gov) and U.S. Pharmacopeia (USP, www.usp.org) offer reference materials for a number of botanicals. ChromaDex (www.chromadex.com) is a commercial source that offers the curated reference materials.

BSC used a combination of analytical approaches (described below) to characterize botanicals along with sourcing from reputable suppliers, and adherence to guidelines to have confidence in the botanicals tested. When available, reference materials were also utilized. In this manuscript, we have selected green tea and Asian ginseng extracts to demonstrate our approach to characterizing these complex mixtures to build confidence in BSC bioassay data.

3.4.1. Characterization of botanicals

Generally, characterization of a botanical can be carried out using a wide array of analytical approaches. These approaches often include chromatographic separations such as liquid chromatography (LC), gas chromatography (GC), or thin-layer chromatography (TLC). Chromatographic separations are paired with spectroscopic/spectrometric detection such as ultraviolet (UV) and mass spectrometry (MS) (van Breemen, 2020). Additionally, LC with evaporative light scattering detection (ELSD) or charged aerosol detection (CAD) methods are sometimes used when profiling botanical extracts or to estimate concentration of botanical constituents (Collins et al., 2020; Sica et al., 2018; Waidyanatha et al., 2022; Waidyanatha et al., 2020). Although ELSD and CAD produce signals for most non-volatile compounds, they don’t provide any structural information. Therefore, when feasible, it is recommended to use a multidetector system such as LC-UV-CAD-HRMS (high resolution mass spectrometry) to characterize these complex mixtures where both identification and estimation of constituents concentrations can be obtained from a single analytical system (Baker & Regg, 2018; Sica et al., 2018). Use of LC-MS techniques to collect the information on the full chemical profile (i.e., non-targeted analysis [NTA]) allows, in addition to confirming the botanical constituents present, for identification of potential adulterants in the material. Nuclear Magnetic Resonance (NMR) spectroscopy also has been successfully used to characterize botanicals (Simmler et al., 2018).

We used a combination of analytical approaches (e.g., HPTLC, LC- coupled with tandem MS (MS/MS), or LC-UV-CAD-HRMS) to authenticate botanical materials used by the BSC. In this manuscript, we have selected green tea and Asian ginseng extracts as examples to demonstrate our approach including the use of a multidetector system (e.g., LC-UV-CAD-HRMS) to characterize these complex mixtures. Asian ginseng and GTE are botanicals where a significant fraction of the extract has been characterized and authentic standards for some of the identified constituents are commercially available. However, as with any other botanical, significant lot-to-lot variability in phytochemical composition is anticipated due to the reasons described above (AbouZid et al., 2016; Ryan et al., 2019). Hence, it is imperative to characterize the botanical comprehensively prior to use to ensure that the material is acceptable and subsequently the data (e.g., chemical or biological data) generated can be compared across studies. Selected data on GTE and Asian ginseng extract from our analysis is given in subsequent sections. Additional information on sample preparation, analysis, and data can be found at NIEHS’s Chemical Effects in Biological Systems Database (CEBS) here.

Green Tea Extract (GTE)

GTE is a popular dietary supplement used in weight loss products (Jurgens et al., 2012). Catechins in GTE make up 50–80% of the extract (Singh et al., 2011). The US Pharmacopeia (USP) published a review of the hepatotoxicity of green tea extracts, which reported a pattern of liver injury from adverse event case reports and other sources. EGCG is a catechin present in GTE that has hepatotoxic potential when ingested at high quantities for long periods of time (EFSA, 2018; Hu et al., 2018). Our aim was to select an unformulated extract (i.e., a bulk extract that serves as source material for a formulated product) with similar composition (i.e., EGCG concentration) to the lot used in the NTP rodent bioassay (NTP, 2016) to allow for comparison between the NTP rodent data and the selected in vitro assays. Towards this, 3 lots were procured from 3 vendors and screened using an LC-UV-CAD-HRMS along with commercially available GTE constituents and NIST GTE Standard Reference Material (SRM, #255).

For the three lots selected, the detector profiles were similar to each other (Figures 3A, 3B, and 3C) and to the NIST SRM, except that a significant caffeine peak was observed in the SRM. The caffeine peak was not detected by CAD due to sublimation of caffeine as the detection requires analytes to remain in solid phase following evaporation of the mobile phase. Based on a single point calibration standard used in this profiling analysis, the concentrations of known GTE constituents (see Figure 3 legend) were estimated. We compared the profiles and estimated constituent concentration with the SRM and NTP rodent bioassay lot and selected lot 20S0293200 for further characterization and to be used in BSC activities. As can be seen from the figure, detector profiles of the selected lot (i.e., test lot) is nearly identical to the SRM, except that the caffeine levels were low in the test lot, as expected for a decaffeinated material (Figure 3).

Figure 3.

Comparison of green tea extract lots from multiple suppliers using LC-UV-CAD-HRMS A) UV profile at 274 nm, B) CAD chromatogram, and C) MS total ion chromatogram. (C, catechin; EC, epicatechin; EGC, epigallocatechin; EGCG, epigallocatechin gallate; ECG, epicatechin gallate; GCG, gallocatechin gallate; GC, gallocatechin; GA, gallic acid.)

HTPLC analyses were performed on the test lot by two independent laboratories. Analysis by Laboratory A using an existing method (Reich, 2007) is shown in Figures 4A (UV detection at 245 nm) and 4B (UV detection at 366 nm following derivatization with vanillin and sulfuric acid reagent). The test lot sample (lane 6) is similar to the testing laboratory’s green tea leaf extracts (lanes 7 −13) with respect to the pattern and retention factors (R_f) of the bands. Intensities of some bands varied between the test lot sample and testing laboratory’s green tea leaf extract samples likely due to natural variability of constituent concentration in different extracts. Based on these observations, testing laboratory A concluded that the test lot sample is Camellia sinensis leaf extract. The absence of a band at the R_f (~ 0.65) (red arrow) of the caffeine standard (lane 5) (Figure 4A) in the test lot sample (lane 6) supports that the test lot is decaffeinated GTE.

Figure 4:

High performance thin layer chromatography (HPTLC) analysis of decaffeinated green tea extract performed by Laboratory A A) UV detection at 254 nm B) UV detection at 366 nm following vanillin/sulfuric acid treatment. Lane 1, epicatechin; lane 2, epicatechin gallate, lane 3, epigallocatechin; lane 4, gallocatechin gallate; lane 5, caffeine; lane 6, test lot sample (lot 20S0293200); lanes 7, 8, 10, 11, 12, Laboratory A’s Camellia sinensis (leaf) extract; lane 9, Laboratory A’s Camellia sinensis (herb) extract; lane 13, Laboratory A’s Camellia sinensis (aerial part) extract.

Laboratory B also performed an HTPLC analysis of the decaffeinated green tea extract using existing methodologies (Do et al., 2021; Reich et al., 2006) (Figure 5). As can be seen from the lanes labelled, reference samples for a variety of tea extracts were included for comparison. The test lot sample (lane 3) is similar to the testing laboratory’s green tea extracts (lanes 4 and 5) with respect to the number, intensity, pattern and retention factors (R_f) of constituents detected (5A) and was somewhat different from other types of teas reference standards used in the analysis (lanes 6–14). Laboratory B concluded that the polyphenol profile of test lot sample is typical of green tea. A chromatographic fingerprint obtained using UV detection at 254 nm (Figure 5B) shows that the band eluting at the Rf for caffeine (lane 2 red arrow) is absent in the test lot sample (lane 3, red arrow) confirming that the test lot sample is a decaffeinated GTE.

Figure 5:

High performance thin layer chromatography (HPTLC) analysis of green tea extract by Laboratory B A) detection under white light following derivatization with natural product/polyethylene reagent (NP) + anisaldehyde (AS) reagents. The black arrow denotes a zone present in all the tested green tea extracts but not the other samples. B) UV detection at 254 nm following derivatization with NP + AS reagents. Lane 2, green tea constituent standards (red arrow caffeine); lane 3, test lot sample; lane 4, Laboratory B’s Camellia sinensis (leaf) reference standard; lane 5, Laboratory B’s Camellia sinensis (leaf) reference standard, decaffeinated.

Although the two laboratories used slightly different approaches and reference materials, both laboratories confirmed that test lot sample is decaffeinated GTE. HPTLC data provided confidence that the material was GTE with chromatographic bands and patterns characteristic of Camellia sinensis. Absence of additional bands in the test lot sample compared with the testing laboratories’ reference samples suggests that the test lot sample does not contain non-GTE constituents that can be detected via these methods. These data with the additional constituent identification and quantification on GTE below are used in a weight-of-evidence to confirm that the material being tested is GTE.

The GTE test lot was analyzed by LC-UV-CAD-HRMS system, along with SRM and available authentic standards to identify constituents. The exact mass, MS/MS spectra of authentic GTE standards and the peaks with matching retention time in the test lot were compared to confirm the identities of constituents in the test lot. As an example, MS/MS spectrum of EGCG (expected M + H, 459.091 Da) for authentic standard (Figure 6A) and for the peak at the same retention time in test lot (Figure 6B) is shown in Figure 6. Identity of the peaks in the lot were confirmed using a combination of retention time, exact mass, and MS/MS spectra. Based on this approach 11 constituents in the test lot were identified with high confidence (Level 1) (Schymanski et al., 2014)— three other constituents were tentatively identified (Data can be found at https://doi.org/10.22427/NTP-DATA-500-007-001-000-3). Figure 7A shows the extracted ion chromatograms of 11 green tea constituents identified with high confidence—corresponding extracted ion chromatograms for authentic standards are given in Figures 7B.

Figure 6.

MS/MS spectrum of A) the EGCG authentic standard and B) the peak at the same retention time in the test lot sample. Expected M +H is 450.093 Da.

Figure 7.

Extracted ion chromatograms for corresponding to GTE constituents A) authentic standards B) test lot sample. C, catechin; EC, epicatechin; ECG, epicatechin gallate; EGC, epigallocatechin; ECGC, epigallocatechin gallate; GC, gallocatechin; GCE, gallocatechin gallate.

Asian Ginseng

Asian ginseng root is a popular herbal dietary supplement purported to improve glucose metabolism and moderate immune responses (Shergis et al., 2013). Given a long history of use, based on many clinical trials (albeit of varying quality), and animal studies, Asian ginseng root is believed to be safe for most people (NCCAM, 2020; NTP, 2011). The main constituents contributing to its pharmacological effects are ginsenosides, which are steroid glycosides and triterpene saponins (Chen et al., 2019). More than 40 ginsenosides, belonging to three main groups, have been isolated from ginseng root (Lü et al., 2009). The Asian ginseng root extract used in our studies was procured commercially and was identified by the manufacturer as a ginseng dry extract containing 4.6% total ginsenosides by weight per certificate of analysis (Table 1).

HTPLC analyses were performed on the test lot by Laboratory B and corresponding chromatograms are shown in Figures 8A (visible light detection following derivatization with sulfuric acid reagent) and 8B (UV detection at 366nm following derivatization with sulfuric acid reagent). The testing laboratory ran a variety of reference samples as shown in the figure. Based on Figures 8A and B, the constituent pattern and the retention factors, R_f, of constituents were similar between the test lot sample (lane 4) and Panax ginseng (Asian ginseng) root reference samples (lanes 2, 5–7) and non-leaf reference samples (lanes 12–13). The difference in intensity of bands between the test lot sample and Panax ginseng root reference samples suggests that the constituent concentrations are different between the test lot sample and reference samples used. Different chromatographic patterns between test lot sample (lane 4) and those of other Panax varieties (lanes 1, 8–11,14, 15) suggest that the test sample does not belong to other Panax varieties included in the analysis. Based on these observations, the Laboratory B concluded that the test lot sample is Panax ginseng root extract.

Figure 8:

High performance thin layer chromatography (HPTLC) analysis of ginseng extract by A) detection by visible light following derivatization with sulfuric acid reagent B) UV detection at 366nm following derivatization with sulfuric acid reagent. Lane 1, Laboratory B’s Panax quinquefolius root reference standard; lanes 2, 5–7, Laboratory B’s Panax ginseng root reference standard; lane 3, ginsenoside Rf and pseudoginsenodise F11 standards; lane 4, test lot sample; lanes 8–9, Laboratory B’s Panax notoginseng root reference standard; lanes 10–11, Laboratory B’s Panax quinquefolius root reference standard; lanes 12–13, Laboratory B’s Panax ginseng leaf reference standard; lanes 14–15, Laboratory B’s Panax quinquefolius leaf reference standard.

In our LC-UV-CAD-HRMS analyses, we included USP SRM of Asian ginseng extract (lot R116W0) along with commercially available ginseng constituent standards. Details of analyses can be found here. Figure 9 shows detector profiles of the Asian ginseng extract test lot sample and USP SRM for comparison. As can be seen from these profiles (Figures 9A, 9B and 9C, three detector profiles of test sample) are similar to the USP SRM. Extracted ion chromatograms for constituents in Asian ginseng extract test lot sample are shown in Figure 10 based on exact mass and retention time matching with authentic standards.

Figure 9:

HPLC-UV-CAD-HRMS analysis of the Asian ginseng extract test lot extract and USP standard reference material (SRM) A) UV detection at 202 nm B) CAD chromatogram C) MS total ion chromatogram (m/z 50–1500). Rg1, Re, Rf, Rg2, Ro, Rb1, Rb2, Rc, Rd are ginsenosides.

Figure 10:

Extracted ion chromatograms of ginseng constituents in the Asian ginseng extract test lot based on exact masses and retention times of authentic standards.

Extracted ion chromatograms for selected exact masses based on constituent standards showed multitude of peaks at varying retention times in the test lot sample demonstrating the presence of multiple isomers of a given constituent further demonstrating the complex phytochemistry of this botanical. Extracted ion chromatograms for ginsenoside Rg3 (expected M + H, 785.505 Da) and ginsenoside Rh1 (expected M + H, 639.447 Da) for standards and corresponding chromatograms for test lot sample are shown in Figures 11 A and 11B, respectively, as examples.

Figure 11:

Extracted ion chromatograms for exact masses of A) ginsenoside Rg3 (expected M + H, 785.505 Da) and B) ginsenoside Rh1 (expected M + H, 639.447 Da) for standard and Asian ginseng test extract lot sample.

Similar to the characterization of GTE, MS/MS spectra of authentic standards and the peaks with matching retention time to standards in the Asian ginseng test lot were compared to confirm the identity of constituents in the test lot for those standards that were available. The MS/MS spectra of authentic standards and for the peak at the same retention time in the test lot sample were compared to definitively identify peaks. As an example, MS/MS spectra for ginsenoside Rb1 in the test lot sample and corresponding authentic standard is shown in Figures 12A and 12B, respectively. Identification of the other peaks in the test lot sample was attempted by comparing the exact mass and/or MS/MS spectrum of a peak to a library spectrum. Based on these approaches 19 constituents in the test lot sample were identified with high confidence (Level 1) (Schymanski et al., 2014): 16 other constituents were tentatively identified (refer to data tables in CEBS here).

Figure 12:

MS/MS spectrum of A) ginsenoside Rb1 authentic standard (expected M+ H, 1109.611) and B) the peak at the same retention time in the Asian ginseng extract test lot.

3.4.2. Quantitation of Constituents

To quantify constituent levels, appropriate methods need to be developed and validated. To improve the accuracy of analyses, reference materials should be used if available (You et al., 2022). For standardization of botanical dietary supplements for human use, it is recommended to implement an analytical method that meet the criteria published by AOAC International (AOAC, 2002), the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH, 2005), and the U.S. Food and Drug Administration (FDA, 2015). For the purpose of the BSC use of these botanicals, fit-for-purpose methods (e.g., qualified or validated methods) using appropriate detectors (e.g., UV, MS, CAD detection) were used depending on the botanical, to quantify constituents. As examples, the methods used and corresponding data for GTE and Asian ginseng extract are summarized in subsequent sections.

Green Tea Extract (GTE)

As mentioned previously, GTE has several marker constituents are commercially available. We obtained 11 authentic standards to be used in the identification and quantitation of this extract. We observed in our initial investigations that constituent profiles were similar between UV, CAD, and MS detectors. Notably, caffeine was not detected in CAD. Additionally, there may be potential matrix effects in MS detection. Therefore, we used UV detection and solvent calibration standards to quantify constituent levels in GTE. The same approach was applied to the NIST SRM. Corresponding data are given in Table 3. This table also includes the concentrations reported for the lot used in the NTP rodent bioassay. The constituent concentration of the current lot was similar to that of NIST SRM except that the caffeine levels were low, as expected for the decaffeinated GTE. Additionally, EGCG concentration in the current test lot (44.0% weight/weight) was similar to that of NTP rodent bioassay lot (48.4% weight/weight) (NTP, 2016).

Table 3.

Green tea extract constituent concentrations.

Constituent (Abbreviation)	DTXSID	Concentration
		Test lot		NIST SRM	NTP lot3
		mg/g (Average ± standard deviation)1	% wt/wt2	% wt/wt	% wt/wt
Caffeine	DTXSID0020232	1.12 ± 0.02	0.112	3.41	4.9
Catechin [C]	DTXSID3022322	6.72 ± 0.68	0.672	0.748	0.51
Epicatechin [EC]	DTXSID4045133	85.1 ± 0.7	8.51	5.19	2.83
Epigallocatechin [EGC]	DTXSID70925231	147 ± 2	14.7	8.57	2.26
Epigallocatechin gallate [EGCG]	DTXSID1029889	440 ± 4	44.0	39.1	48.4
Epicatechin gallate [ECG]	DTXSID1029889	84.5 ± 0.7	8.45	8.50	12.8
Gallocatechin gallate [GCG]	DTXSID701336201	23.8 ± 0.4	2.38	4.71	4.6
Gallocatechin [GC]	DTXSID10914220	14.8 ± 0.2	1.48	1.44	0.52
Gallic acid [GA]	DTXSID0020650	6.21 ± 0.11	0.621	0.389	NA
Theobromine	DTXSID9026132	1.43 ± 0.05	0.143	0.156	NA
Theophylline	DTXSID5021336	0.148 ± 0.043	0.0148	0.0134	NA

¹Average standard deviation for 3 replicates is given.

²% weight constituent per weight extract (% wt/wt) is given for comparison to literature data.

³Data are from NTP Technical Report (NTP, 2016)

Asian Ginseng

For the Asian ginseng extract, we were able to procure approximately 20 constituents known to be present in ginseng for use in identification and quantitation. Analysis was conducted using LC coupled with multiple detectors as mentioned previously. Detector peak areas for each constituent for which authentic standards were analyzed to generate calibration curves. Curves were in general linear for the UV detection while MS and CAD response were quadratic due to these detectors having a narrow linear (MS) or pseudo-linear range (CAD). The amount of constituent in the test lot was estimated using the peak area for each constituent and the corresponding calibration curve. The MS detection had the highest specificity and sensitivity and hence was used for the final quantitation of constituents in the test lot sample. Peaks in the test lot for which an authentic standard was not available were quantitated using the CAD response—the standard curve for a structurally similar constituent was used for each constituent for its quantitation. The constituents identified as malonyl and acetyl esters of ginsenosides used the curve for the same constituent unesterified (for example, the Rb1 curve was used for the malonyl Rb1). Other constituents that were protopanaxadiol-based ginsenosides or protopanaxatriol-based ginsenosides used the calibration curve for one of the other ginsenosides of the same base structure. Based on this approach, total ginsenosides found in the test lot accounted for ~ 5% of the total extract weight and made up most of the constituents quantified in the lot. This value was similar to the value reported in the certificate of analysis from the manufacturer. The total estimated constituents of ~ 5% (weight/weight) suggests that there is either a significant unidentified Asian ginseng-specific fraction and/or the rest of the extract contains other plant materials or other unknowns. The LC void peak in reversed phase chromatography should typically contain highly polar compounds (e.g., in plant materials generally mono- and disaccharides, amino acids, small polar organics). In the ginseng LC profile, based on CAD peaks, the void peak accounted for ~ 81% of the total peak area suggesting that the majority of the unidentified fraction in ginseng test lot are not specific to Asian ginseng constituents. Additionally, weight loss on drying of the test lot was estimated to be ~ 3%, which was similar to that given in manufacturer’s CoA. Taken collectively, these analyses account for ~ 90% of the total extract weight.

3.5. Contaminant Analysis and Adulteration

Similar to other food and pharmaceutical products, botanicals are susceptible to unintentional chemical and microbiological contamination during their production and storage. The common contaminants include heavy metals, pesticides, mycotoxins, residual solvents, and microorganisms that can result in well-recognized adverse health problems. Prior to their use in bioassays, the test material can be tested by appropriate contaminant analyses to ensure their contaminant levels are within the tolerable limits for human exposure. As part of the BSC chemical analysis, the contaminant testing was performed on select botanicals due to limited extract quantity available. Future work could include contextualizing results from in vitro assays by checking for contaminants.

GTE and Asian ginseng extract samples were analyzed for heavy metals, pesticides, and mycotoxins by Eurofins (Madison, WI) using established methods. Of the heavy metals, only antimony (7.80 ppb) and arsenic (24.0 ppb) were found but at low concentrations. No pesticide residues were found above their limits of quantitation (between 0.01 and 0.1 mg/kg depending on the analyte) from the pesticide screen that monitored 310 compounds. A sample was analyzed for aflatoxin B1, aflatoxin B2, aflatoxin G1, aflatoxin G2, ochratoxin A, and zearalenone —none were found above their LOQ (0.5 and 30 ppb depending on the analyte).

Discussion

This manuscript describes the strategy for sourcing, preparing, and analyzing botanicals for the BSC prior to testing the botanicals in bioassays. This includes sourcing from reputable suppliers, extract preparation as needed, determining solubility in solvents used in bioassays, and characterizing materials to ensure that the correct pant material is being used. This work serves as a foundation for understanding and contextualizing the results from the bioassays, with the ultimate goal to develop analytical and biological tools to advance botanical safety. The Chemical Analysis Working Group collaborates closely with other bioassay Work Groups, to support the overall mission to advance botanical safety assessment.

In addition to serving as the foundation for all other BSC work, the data generated can be utilized in various ways, including in silico screening. In silico methods involve computational analysis and modeling, which can help predict toxicity and other biological effects based on chemical structures. By integrating their chemical analysis data with in silico approaches, the BSC can enhance the evaluation of botanical safety and identify potential risk factors associated with specific constituents.

The choice of solvent and other conditions (e.g., temperature and pressure) for extracting botanicals can significantly influence the composition of the resulting extracts. Different solvents have varying affinities for specific chemical constituents present in the plant material, leading to differential extraction efficiencies. Ethanol, commonly used in botanical extractions, is effective at extracting a wide range of polar and non-polar compounds, making it a suitable solvent for preparing botanical extracts. Other solvents, like dichloromethane, is more selective towards non-polar compounds, while water favors the extraction of hydrophilic components. Methanol and acetone are also employed in extractions due to their intermediate solubility characteristics. The BSC used 95% ethanol extraction of botanicals for ease of evaporation and consistency in the preparation of final dried extracts, except for the three commercial botanical extracts procured. Future work in the consortium could include comparing different solvents for extracts, both chemically and biologically. We also determined the solubility of the extracts in both ethanol and DMSO to accommodate the needs of in vitro assays. It is worth noting that solubility may decrease depending on the bioassay medium making the material to precipitate and subsequently affecting the bioassay results. Although it is beyond the scope of our investigation, nephelometry has been successfully used to determine drug solubility by measuring particulate matter in sample (BMG, 2023).

Approaches and recommendations to characterize botanical materials for use in pharmacological and toxicological research is available in the literature to ensure generation of quality data so that both chemical and biological data can be compared across studies (Heinrich et al., 2022; Kellogg et al., 2019; Kuszak et al., 2016; Waidyanatha et al., 2022; Waidyanatha et al., 2020). We used fit-for-purpose chemical analysis such as HPTLC and HPLC-UV-CAD-HRMS to support the identity of the botanicals and to determine the phytochemical composition. The constituents identified, and concentrations determined could be used to help contextualize bioassay results and to extrapolate to human exposures and outcomes. Testing a whole extract in an in vitro assay may lead to results not representative of human exposure, as ADME properties are not reflected in in vitro systems. Constituent information is useful for in silico analyses such as ADME modeling and thresholds of toxicological concern (Kroes et al., 2005) analyses which are based on single constituents. The constituent data are also useful for in vitro to in vivo extrapolations (IVIVEs) to calculate concentrations to be tested in assays or to link the concentrations to in the assays to amounts humans are consuming. While our strategy focused on providing adequate evidence for the authenticity of the materials tested using fit-for-purpose chemical analysis, additional analysis (e.g., volatile analysis, nutritional analysis, inorganic analysis), depending on the intended use, can be done to comprehensively characterize botanical materials and to determine the mass balance as has been reported in the literature (Waidyanatha et al., 2022; Waidyanatha et al., 2020). In conclusion, the Chemical Analysis Working Group plays a pivotal role in the BSC, employing a systematic approach for botanical sourcing, extract preparation, and characterization. Collaboration with other teams ensured the generation of high-quality data to evaluate botanical safety. Additionally, the efforts extend to addressing various safety-related issues, supporting in silico screening, and providing essential support to toxicologists and other teams within the consortium. This manuscript describes our approach and can be adapted by others for chemical analyses of botanicals.

Table 4.

Asian ginseng extract constituent concentration

Constituent [Abbreviation]	DTXSID	Concentration
		(Average ± standard deviation)1	% wt/wt2
Ginsenoside Rg1	DTXSID70945111	2.59 ± 0.44	0.259
Ginsenoside Re	DTXSID30965870	4.70 ± 0.72	0.470
Ginsenoside Rb1	DTXSID401316929	7.10 ± 1.42	0.710
Ginsenoside Rc	DTXSID60911542	5.90 ± 0.81	0.590
Ginsenoside Rb2	DTXSID00911541	5.57 ± 0.88	0.557
Ginsenoside Rd	DTXSID301031221	3.91 ± 0.93	0.391
Protopanaxadiol	DTXSID301028626	0.103 ± 0.000	0.0103
Protopanaxatriol	DTXSID30955635	<0.1	<0.01
Prosapogenin A	N/A	<0.1	<0.01
Ginsenoside Rg2	DTXSID901317065	0.672 ± 0.082	0.0672
Ginsenoside Ro	DTXSID401317177	1.10 ± 0.23	0.110
Ginsenoside Rf	DTXSID90904205	0.995 ± 0.126	0.0995
Ginsenoside Rg3	DTXSID101316982	0.144 ± 0.004	0.0144
Ginsenoside Rh1	N/A	<0.1	<0.01
Kaempferol	DTXSID7020768	<0.1	<0.01
Protocatechuic Acid	DTXSID4021212	<0.1	<0.01
Chlorogenic Acid	DTXSID3024786	0.125 ± 0.001	0.0125
p-Coumaric Acid	DTXSID6064660	<0.1	<0.01
m-Coumaric Acid	DTXSID00891551	ND	ND
Naringenin	DTXSID1022392	<0.1	<0.01

¹Average± standard deviation for 3 replicates is given.

²% weight constituent per weight extract (wt/wt) is given for comparison to literature data.

³ND, not detected.