U.S. Federal Agency Interests and Key Considerations for New Approach Methodologies for Nanomaterials

Engineered nanomaterials (ENMs) come in a wide array of shapes, sizes, surface coatings, and compositions, and often possess novel or enhanced properties compared to larger sized particles of the same elemental composition. To ensure the safe commercialization of products containing ENMs, it is important to thoroughly understand their potential risks. Given that ENMs can be created in an almost infinite number of variations, it is not feasible to conduct in vivo testing on each type of ENM. Instead, new approach methodologies (NAMs) such as in vitro or in chemico test methods may be needed, given their capacity for higher throughput testing, lower cost, and ability to provide information on toxicological mechanisms. However, the different behaviors of ENMs compared to dissolved chemicals may challenge safety testing of ENMs using NAMs. In this study, member agencies within the Interagency Coordinating Committee on the Validation of Alternative Methods were queried about what types of ENMs are of agency interest and whether there is agency-specific guidance for ENM toxicity testing. To support the ability of NAMs to provide robust results in ENM testing, two key issues in the usage of NAMs, namely dosimetry and interference/bias controls, are thoroughly discussed.


Introduction
Engineered nanomaterials (ENMs) are materials with a size range, in at least one dimension, from 1 nm up to 100 nm (ASTM E2456-06, 2006ISO, 2019) or are engineered to exhibit properties or phenomena (chemical, physical, or biological) that are attributable to their dimension(s), even if those dimensions fall outside the nanoscale range, up to one micrometer (1,000 nm) (FDA, 2014b). Compared to larger materials with the same elemental composition, ENMs may have enhanced or novel properties and may exhibit a wide variation in their structure as well as in their physical and chemical properties. These enhanced and novel properties of ENMs have led to their use in a broad range of fields, including agriculture (Adisa et al., 2018;Borgatta et al., 2018;Kah et al., 2019), consumer products 1 , environmental remediation Lowry et al., 2019;Zhang et al., 2019), food production and packaging (Uddin et al., 2016;Szefler, 2018), and nanomedicine (Besinis et al., 2015;Rösslein et al., 2017;Sun et al., 2017). Due to the widespread use of ENMs, it is necessary to ensure that potential environmental (Waissi-Leinonen et al., 2012;Edgington et al., 2014;Mortimer et al., 2016;Lead et al., 2018;Geitner et al., 2020) or human health (Nelson et al., 2013;Grafmueller et al., 2015;Fadeel et al., 2018;Salieri et al., 2020) risks of ENMs are understood and minimized.
facilitates the development, validation, and regulatory acceptance of NAMs and other approaches that replace, reduce, or refine the use of animals for chemical safety testing 4 .
To perform specific tasks for the development or validation of NAMs, ICCVAM establishes ad hoc workgroups 5 . ICCVAM established its Nanomaterials Workgroup (NanoWG) to identify and evaluate ENM-specific testing requirements/recommendations among different U.S. government agencies, to determine whether ENM testing requirements/ recommendations among the different agencies differ from testing requirements/ recommendations specified for other types of substances, and to identify opportunities for NAMs to be used or developed to address agency needs.
This article summarizes the NanoWG's evaluation of U.S. government agency requirements/ recommendations for ENM testing. During this process, the NanoWG identified key considerations that need to be evaluated before NAM-based methods can be used to conduct safety testing on ENMs. Based on the information provided by the agencies on ENM-specific testing requirements/recommendations, we were able to collate references to published documentary standards that have been published relevant to ENM hazard testing. We also discuss key issues regarding control measurements and dosimetry during in vitro testing when evaluating ENMs. This article is not intended to be a comprehensive collection of all test methods used to evaluate ENM toxicity, nor is it a complete compendium of all U.S. agencies, offices, or divisions that utilize ENM testing. The article is intended to provide information to guide future discussion of approaches to advance the use of NAMs for evaluating the hazards of ENMs. Additional information on the regulatory framework for nanomaterials may be found in Ridge (2018), and a recent review by Shaffer et al. (2021) provides an overview of the agencies that perform chemical evaluations for different exposure scenarios. reference materials and standard methods related to ENM testing and evaluation. Tables 1, 2, and 3, respectively, include information on ENMs of agency interest, some test guidelines under which nanomaterials test data are submitted, and ENM-specific guidance documents developed by regulatory agencies.
As mentioned previously, differences in ENM synthesis and handling can alter their toxicological profiles. ENMs also tend to agglomerate/aggregate, settle out of suspension, or react with test media and/or testing components. Consequently, these properties indicate that ENMs have complex dosimetry, and therefore characterization of test media and/or testing components is a critical part of testing. The NanoWG also conducted an additional survey to discuss considerations for ENM characterization and dosimetry for in vitro assays.
Responses were received from CPSC, EPA, FDA, and CDC/NIOSH, and are discussed in Section 4.

Agency needs for ENM testing
Agency responses regarding ENMs of interest, tests used to evaluate ENMs, and agencyspecific guidance documents were compiled and reviewed and are discussed in more depth below. Agencies or divisions that have an interest in ENMs but do not require or conduct testing are the Pacific Northwest National Laboratory of the U.S. Department of Energy, the U.S. Department of the Interior, CPSC, and the U.S. Department of Agriculture National Institute of Food and Agriculture.

Responses relating to materials of interest
Given that the type of ENM and its end use may influence the required testing, the workgroup sought information about what ENMs are of interest to member agencies. The identified materials of interest are presented in Table 1, along with some of the use cases that brought these materials to agency attention. While a broad range of ENMs was represented in responses, almost all the most common ENMs are a focus for at least one agency. Some ENMs, such as carbon nanotubes, graphene family materials, metal oxides, nanoclays, and nanosilvers are a focus for multiple agencies. In addition to providing information about materials of interest, agencies also provided information on why the materials are of interest and indicated which materials are emerging concerns.
CPSC indicated that graphenes and nanoclays are emerging nanomaterials of interest, as well as complex mixtures of carbon nanotubes, metal ENMs, and other particles. They also stated that recently published studies have detected styrene, metals, and carbon nanotubes in the emissions from 3D printers, and carbon nanotubes, nanometals, metal oxides, polycyclic aromatic hydrocarbons, ozone, and carbon dioxide in emissions from laser printers (Kim et al., 2015;Pirela et al., 2019), and that these emissions will require further study.
EPA and CPSC collaborate to evaluate the potential release of free ion or micronized (e.g., formulations consisting of copper carbonate particles ranging in size from a few nanometers to several microns) copper particles from the paint or coating containing nanocopper and nanocopper pressure-treated lumber during their normal use, as well as to evaluate the effects of released metal oxides from treated wood. There is also a potential interest in other forms of nanocopper (e.g., aqueous alkaline copper azole), which has similar use applications and toxicological outcomes to micronized copper.
EPA's Office of Pesticide Programs (OPP) indicated that nanosilica and nanometals bound to nanosilica and mixtures of nanometals were of emerging interest. EPA's Office of Pollution Prevention and Toxics (OPPT) indicated that graphene and graphene oxides are emerging nanomaterials of interest.
FDA's Center for Food Safety and Applied Nutrition (CFSAN) commented that while nanoclays are used in food packaging, they are not expected to migrate into food products. There is potential dietary exposure to ionic copper or silver derived from food contact packaging use of nanoparticulated silver or copper. Because titanium dioxide and silicon dioxide, when used as direct food additives, may contain some particles in the nanoscale range, consumers may also be exposed to nanoparticulate forms of titanium dioxide and silicon dioxide.
CDC/NIOSH's Nanotechnology Research Center (NTRC) is the leading federal agency conducting research and providing guidance on the occupational safety and health implications and applications of advanced materials and nanotechnology. NTRC has a robust field study and laboratory research program that investigates ENM toxicity and conducts exposure assessments and epidemiological studies in the workplace. In addition, the NTRC focuses on critical areas of ENM research including material properties such as dustiness and explosivity behavior, and emissions characteristics of nanomaterials and NM-enabled products that are important in assessing potential toxicity and risk associated with real-world occupational exposures (Bishop et al., 2017). The data suggests that low solubility nano-scaled particles are generally more toxic than larger particles on a mass-tomass basis (Oberdörster et al., 2005;Rothen-Rutishauser et al., 2007;Sager and Castranova, 2009;Zhao et al., 2009;Bakand et al., 2012). There are also strong indications that particle surface area, surface chemistry, and solubility play a role in the observed toxicity of ENMs in cell culture and animal models (Sager and Castranova, 2009;Roberts et al., 2013). In vitro models employing both acute and sub-chronic exposure conditions have been developed and used to predict in vivo toxicological responses (Cho et al., 2013;Manke et al., 2014;Wang et al., 2014). Based on comparable exposure doses, time courses, target cell types, and relevant biological endpoints, consistent results have been obtained from comparable experiments with in vitro vs. in vivo models using similar ENMs (e.g., based on physicochemical properties) such as carbon nanotubes (Mercer et al., 2011;Mishra et al., 2012;Sargent et al., 2014;Siegrist et al., 2014;Snyder-Talkington et al., 2015, metal oxide nanoparticles (Ma et al., 2015;Davidson et al., 2016), boron nitride nanotubes (Kodali et al., 2017;Xin et al., 2020), and end-life cycle (incinerated) nanoclay enabled thermoplastics (Stueckle et al., 2018;Wagner et al., 2018). These results, mainly observed from CDC/NIOSH research projects on the ENMs of agency interest listed in Table 1, support the implementation of in vitro models as a rapid and economical tool to screen and predict the potential in vivo toxicological responses to ENMs for reducing, refining, and replacing animal usage.
The U.S. Department of Agriculture Forest Service Forest Products Laboratory is primarily or partly responsible for the development of many of today's wood-based technologies such as wood science, building structures, building resilience, building materials, pulp and paper, biofuels, performance polymers from wood, and high-value chemicals from wood. In the area of nanotechnology, the laboratory focuses on research into the application of cellulose ENMs, the nanoscale aspects of wood, especially renewable, forest-based nanomaterials, and partners with other organizations on understanding the environmental, health and safety aspects of forest-based nanomaterials.

Responses relating to methods and guidance documents relevant to ENM toxicity testing
One difficulty with the evaluation of ENMs is determining when ENM-specific testing is required. For example, an agency's definition of what may be considered an "ENM" varies between U.S. agencies and may be dependent on end use. This implies the need for a case-by-case ENM-specific safety assessment, based on the material's characteristics, the proposed use of the material, and the route of exposure/administration, among other factors (FDA, 2014b; EPA, 2017). As described in Table 1, there are multiple types of ENMs of interest to U.S. agencies, spanning an array of applications and uses. While testing of ENMs often needs to be evaluated on a case-by-case basis, there are test guidelines, provided in Table 2, that are frequently used for the evaluation of ENMs for use as food additives, new dietary ingredients, pesticides, or as part of pesticide formulations. Table 2 is not intended to be a complete compendium of all test methods used to evaluate ENM hazard, nor should it be implied that these guidelines are only used to test the substances/products indicated in the table. In addition to the guidelines listed in Table 2, some agencies (such as EPA) allow studies to be conducted in accordance with Organisation for Economic Co-operation and Development (OECD) guidance 6 . Moreover, EPA requires or recommends that protocols be submitted prior to study submission if modifications of these methods are proposed for toxicity testing of ENMs. It was often not possible to provide prescriptive suggestions about what specific methods are acceptable for testing ENMs, because the science on this topic is rapidly evolving and decisions are often made on a case-by-case basis. Given this rapid evolution, consensus has not yet been reached within agencies on some topics. EPA OPP regulates the manufacturing and use of pesticides (including insecticides, herbicides, rodenticides, disinfectants, sanitizers, etc.) in the United States and establishes maximum levels for pesticide residues in food. OPP operates under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), which governs pesticide registration, distribution, sale and use. Enacted in 1947, FIFRA sets risk/benefit standards for pesticide registration, requiring that pesticides perform their intended function, when used according to labeling directions, without posing unreasonable risks of adverse effects on human health or the environment (7 U.S.C. § 136 et seq., 1947). In 1972, FIFRA was amended, expanding EPA's authority to strengthen the registration process, enforcement provisions, and broaden the legal emphasis on further protecting health and the environment (7 U.S.C. § 136 et seq., 1972). FIFRA was further amended by the Food Quality Protection Act (FQPA) (7 U.S.C. §136, 1996) and the Federal Food Drug and Cosmetic Act (21 U.S.C. §301 et seq, 2002), under which EPA establishes tolerances or maximum legal limits for pesticides that apply to food. Under FQPA, a collection of pesticide data is necessary to set allowable levels and to conclude that a pesticide is safe. The rule further ensures that no harm will result to infants and children from aggregate exposure to the pesticide chemical residue. As a result, pesticide products, including ENM-containing antimicrobial products, inquiring registration require various data generation to address potential adverse effects to humans and environmental fate.
In evaluating a pesticide registration application, OPP assesses a wide variety of potential toxicological effects associated with the use of the product or active ingredient. In general, for ENMs, OPP requires data generated with the toxicological test guidelines presented in Table 2, but ENMs' physical-chemical product characteristics are evaluated by product chemistry test guidelines and often compared with ENMs reported in toxicology studies.
EPA OPPT administers the Toxic Substances Control Act (TSCA; (15 U.S.C. §2601 et seq., 1976)), which regulates chemical substances and mixtures that are manufactured, imported, processed, distributed, used or disposed of in the United States and that are not regulated under other laws (such as those that apply to pesticides or food and drugs). TSCA was originally enacted in 1976 and serves as the nation's primary chemicals management law. In 2016, TSCA was amended by the Frank R. Lautenberg Chemical Safety for the 21 st Century Act, which included language to encourage alternatives to animal use for testing done under TSCA (15 U.S.C. § 2601 et seq., 2016).
Under TSCA, most nanomaterials are regarded as "chemical substances". New chemical substances manufactured at the nanoscale must be submitted to EPA review before they can enter the marketplace 7 . Although upfront toxicity testing is not required under TSCA for any chemical substance, including ENMs, manufacturers must submit any existing data in their possession or control at the time of the new chemical application in a premanufacture notice. Premanufacture notice submissions for new nanomaterials under TSCA are reviewed and regulated individually. If EPA determines that the available information is insufficient to make a reasoned evaluation as to whether an ENM might produce an unreasonable risk to human health or the environment under the expected conditions of use, the agency may issue a consent order under Section 5(e) of TSCA to the submitter for additional testing. The recommended testing is specific to the area of human health concern. For example, if the concern is about inhalation exposure to various nanoparticles, the recommended testing may include an inhalation toxicity study (OPPTS Test Guideline 870.3465 (EPA, 1998f) or OECD Test Guideline 413 (OECD, 2018b).
The 2016 Lautenberg Chemical Safety Act (15 U.S.C. § 2601 et seq., 2016) requires EPA to develop a plan to "promote the development and implementation of alternative test methods and strategies to reduce, refine, or replace vertebrate animal testing and provide information of equivalent or better scientific quality and relevance for assessing risks of injury to health or the environment of chemical substances or mixtures." As part of this effort, EPA published a strategic plan in 2018 (EPA, 2018) to promote the development and implementation of alternative test methods or NAMs and a list of acceptable NAMs within the TSCA program (EPA, 2021). Even though NAMs presented in this list are not specific to ENMs, EPA expects to consider NAMs for several TSCA ENM decision contexts including hazard identification and characterization. Table 3 lists selected guidance documents that U.S. federal agencies have issued to advise stakeholders on ENM testing. In 2017, EPA issued guidance (Tab. 3) to assist companies to report under the TSCA nanotechnology reporting and recordkeeping requirements rule (EPA, 2017). This rule mandates that manufacturers report information including specific chemical identity, production volume, methods of manufacture and processing, exposure and release information, and existing data on environmental and health effects.
FDA recently released a progress report (FDA, 2020) that shows a steady increase in drug product submissions containing nanomaterials to FDA. These submissions include nanomaterials of differing compositions, sizes, and surfaces, as well as nanomaterials containing therapeutic agents (Farjadian et al., 2019). FDA has issued several guidance documents on topics related to the application of nanotechnology in FDA-regulated products (Tab. 3) as part of ongoing implementation of recommendations from FDA's 2007 Nanotechnology Task Force Report (FDA, 2007). These documents serve to convey FDA's current opinion on a topic rather than to bind the FDA or the public.
In 2014, FDA issued the FDA Final Guidance for Industry -Considering Whether an FDA-Regulated Product Involves the Application of Nanotechnology (FDA, 2014b). This guidance describes an overarching framework for FDA's approach to the regulation of nanotechnology products. FDA has not established a regulatory definition of nanotechnology, nanomaterial, nanoscale, or related terms. In this overarching guidance, FDA identified two "points to consider" that should be used to evaluate whether FDAregulated products involve the application of nanotechnology:

1.
Whether a material or end product is engineered to have at least one external dimension, or an internal or surface structure, in the nanoscale range (approximately 1 nm to 100 nm);

2.
Whether a material or end product is engineered to exhibit properties or phenomena, including physical or chemical properties or biological effects, that are attributable to its diniension(s), even if these dimensions fall outside the nanoscale range, up to one micrometer (1,000 nm).
The FDA Center for Devices and Radiological Health follows this guidance when evaluating new medical devices. A key statement from this document is: "Based on our current scientific and technical understanding of ENMs and their characteristics, FDA believes that evaluations of safety, effectiveness, public health impact, or regulatory status of nanotechnology products should consider any unique properties and behaviors that the application of nanotechnology may impart." In addition to the FDA Final Guidance for Industry -Considering Whether an FDA-Regulated Product Involves the Application of Nanotechnology, the Center for Drug Evaluation and Research also refers to another draft guidance 8 , Drug Products, Including Biological Products, that Contain Nanomaterials -Guidance for Industry (FDA, 2017). This draft guidance "does not address, or presuppose, what ultimate regulatory outcome, if any, will result for a particular drug product that contains nanomaterials." Safety, effectiveness, public health impact, and regulatory status of drug products that contain ENMs are currently addressed on a case-by-case basis using FDA's existing review processes. Current Center for Drug Evaluation and Research guidance documents and requirements for the evaluation and maintenance of quality, safety, and efficacy apply to drug products containing ENMs that fall within their scopes. "As such, this guidance should be viewed as supplementary to other guidances for drug products" (FDA, 2017).
FDA has also issued guidance documents pertaining to ENMs in food (FDA, 2014a). CFSAN has premarket authorization authority over food additives and new dietary ingredients under the United States Federal Food, Drug, and Cosmetic Act (21 U.S.C. §301 et seq, 2002). As both product areas concern potential oral exposure to an ENM, the toxicity testing paradigms generally used to evaluate the safety of food additives or new dietary ingredients primarily comprise repeated oral dosing studies in rodents. Existing test guidelines describing repeated oral dosing and inhalational exposure studies in rodents (EPA, 1998e,f;OECD, 1998b;EPA, 2000;OECD, 2008OECD, , 2009bOECD, , 2018b appear to be appropriate for use with ENMs (OECD, 2009a(OECD, , 2012. To evaluate carcinogenicity of these products, genotoxicity studies, such as the Ames assay or the mouse lymphoma assay, are used to ascertain the mechanism of action of any observed neoplastic effects in rodent bioassays (Kobets et al., 2018). However, for the Ames assay, some ENMs have been shown to be unable to enter the bacterial cells, which would make such test articles incompatible with the test system (Woodruff et al., 2012). It is notable that none of the standard OECD test guidelines on in vitro genotoxicity assays has been validated for use with ENMs, though the guideline describing the in vitro mammalian cell micronucleus test directly acknowledges the requirement for methodological adaptation for ENMs (OECD, 2016). In addition, toxicokinetic studies may be used to inform the safety assessment regarding the potential for systemic exposure to the food additive or new dietary ingredient, for route-to-route extrapolation from the results of non-oral toxicity studies, and for refining the inter-and intraspecies uncertainty factors used in quantitative risk assessment for non-neoplastic endpoints.
CDC/NIOSH leads the federal government health and safety initiative for nanotechnology 9 .
Research and activities are co-ordinated through CDC/NIOSH's NTRC. The contributions of NTRC to the nanotechnology and nanotoxicology fields include the guidance documents of safety programs, guidelines, and design solutions for ENM workplaces (Tab. 3).
The CPSC's regulations do not require testing; the Federal Hazardous Substances Act (15 U.S.C. §1261 et seq., 2008) and its implementing regulations only require that a product be labeled to reflect the hazards associated with that product. Manufacturers, retailers, and distributors of nano-enabled products, as with any consumer product under the CPSC's jurisdiction, must report to the CPSC immediately if they obtain information that reasonably supports the conclusion that their product fails to comply with an applicable consumer product safety rule, contains a defect that could create a substantial product hazard, or creates an unreasonable risk of serious injury or death (CPSC, 2019).
The U.S. Department of Defense generally uses data collected using EPA's guidelines for ENM testing. Some specific tests such as the zebrafish (Danio rerio) embryo test (OECD, 1998a;Haque and Ward, 2018) or Daphnia magna toxicity testing (Xu et al., 2019) are primarily directed at understanding the ecotoxicity of novel ENMs.
In addition to the test guidelines and guidance documents identified in Tables 2 and 3, the NanoWG compiled a list of documentary standards and guidelines designed or evaluated for ENM characterization and/or toxicity testing issued by the American Society for Testing and Materials International (ASTM), the International Organization for Standardization (ISO), and the OECD Working Party on Manufactured Nanomaterials. The compiled list, which contains recommended vocabularies for ENMs, methods for the characterization of ENMs, and some methods for working with and evaluating ENMs, is presented in Table S1 10 . This compilation of methods has been prepared to support scientists with identifying potentially relevant standards. While some of these methods describe toxicity tests designed for use with ENMs (e.g., ASTM E2526 (2013)), many also describe the protocol considerations and measurements that are needed to support toxicity testing such as ENM characterization in the test media and quantification of the ENM concentration. The key issue of dosimetry during in vitro tests with ENMs will be discussed in depth in Section 4.2.

Practical considerations for in vitro toxicity testing of ENMs
Compared to substances that readily dissolve in test medium or other solvents, ENMs pose multiple challenges owing to their unique physicochemical characteristics. It is increasingly realized that commonly used in vitro inhalation toxicity study models where the effects of ENMs on cultured cells are tested under submerged conditions, may not represent real exposure conditions, i.e., inhaled "dry" ENM deposition in the lung. One of the foremost challenges in ENM testing relates to changes in dosimetry occurring during experiments (Teeguarden et al., 2007;DeLoid et al., 2017). Changes in dosimetry can occur as a result of each ENM's effective density in culture medium (DeLoid et al., 2014;Pal et al., 2015), dissolution of particles (e.g., nanosilver particles dissolving and forming silver ions (Liu et al., 2010)), agglomeration of particles (e.g., particles interacting with other particles to form larger agglomerates (Li et al., 2010)), heteroagglomeration of the particles (e.g., particles interacting with, for example, algae or bacterial cells during the assay to form agglomerates (Hartmann et al., 2012;Hanna et al., 2018)), and transformations such as redox changes (e.g., changes in the speciation of particles such as the conversion of AgNPs to silver chloride particles (Ha et al., 2018;Poli et al., 2020)). Dissolution, agglomeration, and/or redox changes can cause the exposure concentration to vary substantially when testing pelagic organisms (i.e., organisms in the water column such as Daphnia magna) or suspended cells. In addition, the results of in vitro assays for some ENMs may vary strongly based on the composition of the test medium, which can impact the dissolution of ENMs, their transformations (e.g., redox changes), or the formation of a protein corona (Drasler et al., 2017;Kaiser et al., 2017). Another key challenge that we discuss in Section 4.3 is the potential for experimental artifacts during toxicity testing of ENMs. This necessitates adequate control experiments to identify and minimize potential artifacts and may reveal additional control experiments required for elucidating mechanisms of toxicity.
One approach that may have more physiological relevance and overcome some of the issues with transformations that can occur during exposure with submerged models is to expose cell culture models having an air-liquid interface to aerosolized ENMs. This exposure approach utilizes cells grown on porous culture inserts, such as 3D models with pseudostratified epithelium and intact mucosa and cilia, which enables direct deposition of nanoparticle powders through aerosol exposure. This approach has been used in numerous recent ENM studies (Polk et al., 2016;Drasler et al., 2017;Barosova et al., 2020;Leibrock et al., 2020).

Dosimetry survey responses
The complexity of ENM dosimetry (i.e., particle agglomeration/aggregations, redox changes, interaction of particles with proteins in media, particle dissolution rate, etc.) led the NanoWG to develop a list of detailed considerations for those using in vitro tests (Tab. Table 4 are suggested based on best practices from the scientific literature. However, it is important to note that standardized methods are not yet available for some potential dose metrics such as particle number concentration or surface area concentration. Additional concerns are described below.

4). The measurements in
Accurate dosimetry measurements, in general, are challenging and may not be technically feasible for all types of ENMs (Johnston et al., 2020). For example, it is substantially more difficult to characterize the agglomeration status of rod-or plate-shaped ENMs than that of spherical nanoparticles. This is because dynamic light scattering, a commonly used agglomeration characterization method, typically determines the hydrodynamic diameter of an ENM based on the size of a sphere that diffuses at the same rate as the particle being measured (Petersen and Henry, 2012;Carvalho et al., 2018). In addition, commonly used in vitro dosimetry models for submerged cells are limited to relatively low-aspect-ratio ENMs (i.e., those with a length similar to their width) (DeLoid et al., 2017).
Another factor that must be accounted for is the effective density of the ENM agglomerate unit, which includes both the particles and the media (DeLoid et al., 2014). The effective density for an ENM can vary greatly from one culture medium to another, thus changing the delivered dose to the cells for the same ENM. The capacity to characterize different concentration dose metrics also varies based on the type of ENM and its agglomeration state (Minelli et al., 2019). For example, a comparison of the number concentration measurements of gold ENMs had substantially worse agreement among techniques for samples which showed substantial agglomeration than for those that remained individually dispersed (Petersen et al., 2019a). The detection limit of analytical methods to quantify ENM mass concentration in test media for in vitro NAMs also varies for different ENMs.
For example, it is difficult to measure the concentration of carbonaceous ENMs in test media with high concentrations of serum Goodwin et al., 2018), while the presence of serum in medium is less problematic for quantification of metal and metal oxide ENMs (Laborda et al., 2016).
The procedure to prepare an ENM suspension at the necessary concentration prior to an assay can vary greatly among laboratories, which may change the experimental outcome. Thus, there is a need to standardize the preparation for each ENM to reduce variability between testing laboratories. For example, most ENMs are sonicated prior to testing, but the level and duration of the sonication can vary, which affects the amount of energy delivered to the material. This variation can affect the agglomeration size, which ultimately affects the dose of material delivered to the cells. A way to minimize variation is to calorimetrically calibrate all sonicators to ensure the exact same energy is delivered to the material each time for consistent dispersion results (Taurozzi et al., 2011). Also, the total delivered sonication energy and the number of sonications needed to disperse ENMs should be reported for each study. Table 4 was circulated within the workgroup to assess the relevance of these considerations on the characterization of ENMs to agencies' information needs. As expected from agencies with very different testing needs, responses to Table 4 varied.
Responses from EPA OPP were that several characterizations (i.e., ENM mean size prior to addition to test media, ENM size distribution prior to addition to test media, ENM mean size in test media prior to exposure period, ENM size distribution in test media prior to exposure period, and ENM dissolution in test media before and after exposure period) are not required as part of toxicity testing, but are requested as part of physicochemical properties of products and environmental fate determinations. Thus, these measurements are not necessarily made in the presence of cell culture or environmental media. ENM mass concentration in test media before exposure period are not required, but OPP typically requests clarification of such information as part of the dissolution kinetic studies when test media are buffer solutions or water. For toxicology studies, if not provided, OPP encourages registrants "to provide nanomaterial mass concentrations in media" under certain circumstances. It is important to note that, if ENM-specific modifications to test methods are needed, a revised protocol submission is recommended for review prior to initiating the study. Such modifications may be needed to generate robust results.
EPA OPPT stated that manufacturers are not required to submit any specific dosimetry characterization data for ENMs. However, manufacturers are encouraged to submit ENM mean size and size distribution before exposure period along with other standard physicochemical characterization data, which may assist with EPA's understanding of the toxicity of an ENM.
For review of engineered nanomaterial food contact substances where consumer exposure to the nanomaterial is expected, FDA CFSAN requires the following ENM-specihc information: particle number or surface-area concentration in test media before exposure period, ENM mean size or size distribution prior to addition to test media, and ENM mean size in test media prior to exposure period (Rice et al., 2009). ENM dissolution in test media after exposure period would be considered a key metric both in assessing test system exposure to the ENM and also in assessing the feasibility of using "read-across" to its non-nano analogs (e.g., a particle with the same composition and shape with all dimensions > 100 nm) in the safety assessment of the ENM. CFSAN indicated that some information such as measurements or modeling of ENM mass concentration associated with cells after the exposure period would be considered key metrics for documenting exposure of the test system to the test article.
Regarding delivered dose, there was discussion about the benefits and limitations of two different particokinetic models: the in vitro sedimentation, diffusion, and dosimetry (ISDD) model (Hinderliter et al., 2010;DeLoid et al., 2017) and the in vitro sedimentation diffusion, dissolution, and dosimetry (ISD3) model (Thomas et al., 2018). NanoWG discussion specifically concerned the models' usefulness in relating a nominal concentration to an estimate of the actual amount of ENMs reaching the cells. Ultimately, the workgroup reached no consensus as to how to use different dose metrics or particokinetic models to understand the results from in vitro studies, although several workgroup members agreed with the CPSC response that, in general, robust studies include hydrodynamic or aerodynamic size distribution data for aqueous dispersions or airborne ENMs before the start of the exposures.
The measurements presented in Table 4 are not necessarily required data for the submission of ENMs to regulatory agencies, and there is still debate within and across agencies as to which data should be required or considered as part of toxicity study requirements for ENMs. Nonetheless, the measurements are still useful for consideration during the development and testing of ENMs. Table 5 lists five main categories of in vitro test exposure systems. The choice of whether to require additional dosimetry measurements for in vitro methods may vary based on the exposure system used.

Dosimetry considerations
While promising research has been conducted on the fourth (airborne exposure to a biological test system located on an air-liquid interface (Lacroix et al., 2018;Barosova et al., 2020)) and fifth (lung-on-a-chip model of inhalation toxicity (Zhang et al., 2018)) exposure systems/categories, there are no standardized methods using these exposure approaches. Thus, this discussion will focus on the first three types of exposure systems.
As described in Section 4.1, dosimetry and dosimetry requirements/recommendations for ENMs can be complex, differing to some extent among agencies, and detailed guidance is not always available. In the absence of such guidance, it can be helpful to consider the dosimetry requirements for testing dissolved substances, which are described in detail for the OECD testing program. For human health testing for either in vivo or in vitro measurements, only the verification of the initial dose is required. However, it is widely known that the exposure concentration of dissolved chemicals can vary due to factors such as volatilization, adsorption to the well sidewalls, and metabolism (Tanneberger et al., 2013). The trade-offs between test method accuracy and the additional costs and workload associated with testing the concentrations in the wells is a topic of ongoing discussion (Natsch et al., 2018). In addition, numerous efforts have been made to move from a nominal to a cellular concentration in in vitro assays using submerged culture exposure conditions and in associated in vitro to in vivo extrapolation modeling ( In NAMs with liquid exposure to suspended molecules or cells (Category 1 of Tab. 5), rapid agglomeration and settling of the ENM in these systems would reduce the suspended exposure concentration to the ENM. Therefore, it may be appropriate to measure the change in the suspended ENM mass concentration across the duration of the assay to evaluate if the concentration is constant, unless the ENM concentration at the bottom of the test container would be expected to have the same effect as the fraction that remains suspended. For Category 2 assays, those in which cells growing in monolayers are submerged in media, it is possible to quantify changes in the suspended concentration during the exposure period and to estimate that the exposure concentration is equivalent to the change in the suspended concentration. For the third exposure approach (Category 3: a liquid, cream, or solid directly applied to a biological test system such as a 3D construct), determining the ENM mass applied to the surface is likely sufficient. The exposure concentration on the biological test article can be determined from the ENM concentration in the formulation or solid and the mass or volume applied to the biological construct. For submerged cell model exposure (Category 2), there have also been extensive efforts to model the expected cellular exposure concentration based on the effective density and size of the ENM, as described above for the ISDD model (Hinderliter et al., 2010;Thomas et al., 2018). However, this approach has not yet been standardized, the reproducibility of effective density measurements has not undergone interlaboratory testing, and the modeled cellular concentration may depend upon the method used to quantify the ENM size (Petersen et al., 2019a). Further dosimetry modelling to model deposition relies upon accurate input parameters, such as dispersant density and viscosity, that are not universally available. This can lead to uncertainty in attaining expected cellular exposure concentrations; therefore, in the absence of parameters published in the literature, the required parameters should be experimentally derived. Lastly, gaps in dosimetry include the impact of physiochemical parameters on ENM behavior in medium during dosing, modeling deposition within the cellular environment for high-aspect ratio fibers (Price et al., 2019) and two-dimensional ENMs, and efficient dosing with buoyant ENMs, such as virgin and nano-enabled composite thermoplastics. Until robust models are developed and validated, secondary analytical techniques presented in Table 4 should be considered to reduce uncertainty in assessing cellular exposure.

Interference/bias controls
One of the foremost challenges in using in vitro test methods with ENMs is the potential for analytical biases or artifacts (i.e., problems that occur during the test leading to an incorrect result or misinterpretation). in vitro ENM studies often either overlook or provide incomplete interference characterization (Ong et al., 2014), because control experiments to detect and characterize ENM-derived artifacts are often not performed. A list of potential control experiments is provided in Table 6 along with assays that could be impacted by each artifact. No specific recommendations or guidelines for the detection and characterization of method specific ENM interference currently exist. Since each test method is performed under a unique set of circumstances, which may include methodspecific reagents, incubation temperature and times, or biological sample matrices, it is necessary to critically review each parameter prior to determining what control experiments may be needed when testing a particular ENM.
If artifactual results are expected or observed, it may be necessary to consider whether mitigation strategies, bias characterization, or complete methodological replacement are warranted. In the case of cytotoxicity, membrane integrity, and proliferation screening assays, no single method is universally robust against interference for all ENMs (Monteiro-Riviere et al., 2009;Kroll et al., 2012). Therefore, each ENM-method pairing should be screened for known sources of interference highlighted in Tables 6 and 7 to determine analytic fitness for purpose and to characterize approximate direction and magnitude of analytic bias, if possible (Han et al., 2011;Holder et al., 2012). In addition to the sources of interference highlighted in the tables, when using methods with indirect measurement endpoints, e.g., colorimetric, fluorometric, luminometric, etc., ENM absorbance, quenching, and autofluorescence should be examined to assess appropriateness of that method. Where applicable, signal inhibition/enhancement and spike-in control experiments may be warranted. Further, measures of cytotoxicity, membrane integrity, and proliferation can be performed using two or more concurrent methods to assess concordance and facilitate result interpretation.
In certain instances, method replacement may not be plausible, and adaptation of an extant method may be required. Here, we use the in vitro cytokinesis-block micronucleus assay using cytochalasin B, which is a standard assay for measuring genotoxicity of a chemical (Fenech, 1997), as an example. In the method, cytochalasin B is added to cultured cells to inhibit cytokinesis, but it also inhibits actin assembly, which can decrease cellular uptake of ENMs (MacLean-Fletcher and Pollard, 1980;Kettiger et al., 2013). Therefore, while not formally adopted, the OECD has proposed methodological adaptation through delayed cytochalasin B treatment after ENM treatment to mitigate potential ENM uptake inhibition for the in vitro cytokinesis-block micronucleus assay (Gonzalez et al., 2011).
Under certain circumstances, artifactual influences on the biological system may be unavoidable. For example, the formation of a proteinaceous ENM corona can lead to immunomodulatory or toxicodynamic effects on in vitro models (Mo et al., 2018). Effects caused by a protein corona during in vitro experiments may not necessarily be translatable to in vivo models or the human milieu, but they cannot be immediately discounted given that the incorporation of nano-enabled medicines may potentially lead to bioavailable serumbound ENMs (Rampado et al., 2020).
In addition to potential analytical artifacts and biases, it is possible to perform additional control experiments to better understand and contextualize the mechanism of toxicity to match inherent properties of a particular ENM and its respective exposure conditions. For example, the addition of a particle dispersant may impart a biological or toxicodynamic effect on the in vitro system that may not translate to the in vivo milieu. Though such controls are typically routine, the potential biological effects due to corona formation in the presence of proteinaceous dispersants, such as serum, should be considered. The toxicodynamic effects of dissolvable ions from ENM and leachable constituents from complex mixtures may warrant investigation with a myriad of methods, including treatments with soluble ion controls and filtrate controls. A list of experiments to understand the mechanism(s) of toxicity is presented in Table 7. For some contexts of use, gaining insight into the toxicity mechanism as well as contributory sources of biological effect may be critical for risk assessment, while for other contexts of use, this infonnation may not be essential but assists in interpreting the assay results. When conducting assays to fulfil regulatory requirements/recommendations, the relevant regulatory agency should be consulted to determine what control experiments are required prior to the submission of in vitro toxicity or efficacy test data.

Conclusions and future directions
The NanoWG surveyed ICCVAM member agencies to request information as to which types of ENMs are of agency interest, which toxicology tests are performed on ENMs, whether there is agency-specific guidance for ENM toxicity testing, and what dosimetry and interference/bias controls are requested for the use of in vitro test methods with ENMs. Based on the responses received, the workgroup determined that there are significant challenges in identifying and clarifying the toxicity testing needs of ENMs across agencies and programs, because the requirements or key considerations at each agency differ based on the products they regulate. Therefore, the NanoWG evaluated two key issues, namely dosimetry and interference/bias controls, which are relevant across a broad range of NAMs when testing ENMs to assist in vitro method developers in understanding the perspectives of different agencies on these topics and to help provide general guidance.
Demonstrating the technical reproducibility and biological relevance of NAMs is the key to supporting their broader use for dissolved and particulate substances such as ENMs. One important topic for future work related to technical reproducibility to support the broader use of in vitro test methods is to provide clear guidance on determining whether a particular method is applicable for use with ENMs. This may require performing the assay with a specific set of ENMs with diverse properties such as different surface charges, elemental compositions, and surface coatings, and clarifying specific control measurements that should be performed simultaneously. If control measurements of a NAM show artifactual results with some types of ENMs, the applicability domain of the NAM may be limited to those ENMs that do not produce such results, or modifications to the NAM to minimize the effect of the artifacts may be needed.
An important topic for future work related to biological relevance is how to correlate in vitro and in vivo test results, and how to evaluate to what extent in vitro responses can be used to predict corresponding in vivo exposures and effects. This is especially important if the in vitro test results will be used for more than just screening and prioritization. As described in the ICCVAM roadmap (2018), it is recommended, when possible, to discuss proposed applications of NAMs with regulatory agencies during the NAM development process to carefully clarify the context of use. To validate the in vitro to in vivo correlation, it would be helpful to collect high-quality data available for different standardized in vivo test methods with different ENMs. These results could then be compared to those obtained using individual NAMs (e.g., lung fibrosis (Barosova et al., 2020)) or combinations of NAMs (e.g., those for skin sensitization (OECD, 2021b)) testing specific key events along an adverse outcome pathway (Halappanavar et al., 2019(Halappanavar et al., , 2020. Suggested priority areas for comparing in vivo results and NAMs are for endpoints that have demonstrated defined approaches (e.g., skin sensitization) for dissolved chemicals and for endpoints that have robust in vivo datasets with ENMs.
Stakeholders place confidence in data from toxicology test methods, i.e., that they are producing the correct result and identifying a potential hazard (or not). Hazard evaluation has historically been accomplished through in vivo approaches. As highlighted above, to establish confidence in NAMs, we compare them to the in vivo test method result, and discordance is viewed as a limitation of the NAM. However, in addition to assessing NAM reproducibility, several studies are now investigating the reproducibility of in vivo methods so that limitations can be taken into consideration in the context of any discordance noted when comparing to NAMs (Luechtefeld et al., 2016;Pham et al., 2020;Rooney et al., 2021).
Other recent work has focused on evaluating traditional in vivo toxicity tests, as well as NAMs, based on their relevance to human biology (Clippinger et al., 2021). With that in mind and given the challenges to implementation of NAMs as complete replacements of animal use for testing single chemicals, it stands to reason that their implementation for testing ENMs has yet to be realized. Therefore, while substantial progress has been made in the testing of ENMs during the past two decades, additional work on these topics is needed to support the increased usage of in vitro test methods with ENMs for regulatory testing. Progress towards this goal will be predicated on federal agencies and stakeholders working together using flexible, robust, and integrated approaches to implement NAMs that both protect human and environment health and reduce or eliminate the need for testing in animals.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

Acknowledgements
We would like to thank Dr World Nieh for his information on U.S. Department of Agriculture interests in nanomaterials and Ms Catherine Sprankle for editorial review of the manuscript. This project was funded in part with federal funds from the National Institute of Environmental Health Sciences, National Institutes of Health under Contact No. HHSN273201500010C to ILS in support of the National Toxicology Program Interagency Center for the Evaluation of Alternative Toxicological Methods. ENM mass concentration in test media after exposure period Determines the ENM concentration after exposure; mass measurements are easier to measure than particle number or surface area concentrations; the information at the beginning and end of the exposure period can enable determining the actual exposure concentration and changes in the ENM (e.g., dissolution) during the test.

ICP-MS b
ENM number or surface area concentration in test media before exposure period Suggested to be more reflective of the toxicological risk than mass based ENM concentration, and thus better enable in vitro to in vivo extrapolation. Modeling approaches include the ISDD and ISD3 models (DeLoid et al., 2017;Thomas et al., 2018) ENM mean size prior to addition to test media Provides fundamental information about the ENM to be tested and is broadly recommended.
c Citations are the same as those used for "ENM number or surface area concentration in test media after exposure period".
d Citations are the same as those used for "ENM size distribution prior to addition to test media".
DLS, dynamic light scattering; ICP-MS, inductively-coupled plasma mass spectrometry; NTA, nanoparticle tracking analysis; spICP-MS, single particle ICP-MS; TEM, transmission electron microscopy 3 A liquid, cream, or solid is directly applied to a biological test system such as a 3D construct In vitro skin irritation: reconstructed human epidermis test method (OECD, 2021a) 4 Airborne exposure to a biological test system located on an air-liquid interface insert Considerations for in vitro studies of airborne nano-objects and their aggregates and agglomerates (ISO, 2020d) 5 Exposure via multiple routes using an in vitro microphysiological system (e.g., eye-on-a-chip, gut-on-a-chip, and lung-on-a-chip devices) Standard methods or guidance documents are not yet published to our knowledge. Potential control experiments to understand toxicity mechanisms and support interpretation of assay results a Potential control experiments Method to perform control experiment Purpose(s)

References
Coating control Perform the assay using the ENM coating at a relevant coating concentration. Test if coating has toxicological or biological effects on organisms or cells. Petersen et al., 2011;Sun et al., 2017 Dispersant control Perform the assay using the ENM dispersant at a relevant dispersant concentration. Test if coating has toxicological or biological effects on organisms or cells. Wang et al., 2010;Youn et al., 2012 Dissolved ion control For ENMs that dissolve, perform the assay using the dissolved ion. Allows for comparison of endpoints between. ENM and constituent dissolved ions. Assess if ENM formation could occur from ions in test media or in cells present during the assay Scanlan et al., 2013 Filtrate only control Filter the ENM suspension and then perform assay with the filtrate. Assess potential toxicity of contaminants, and dissolution from ENMs during the synthesis, storage, and dispersion processes Hanna et al., 2016;Coyle et al., 2020 a This table has been modified and edited with permission from Petersen et al. (2014), © 2014 American Chemical Society.