Science approach document - Bioactivity exposure ratio: Application in priority setting and risk assessment - Technical Updates

Health Canada
August 2026

1 Introduction

1.1 Background

In July 2025, Health Canada (HC) and Environment and Climate Change Canada (ECCC) published a strategy document to guide ongoing efforts to replace, reduce, or refine the use of vertebrate animals in toxicity testing wherever possible (that is, to the extent practicable and scientifically justified) under the Canadian Environmental Protection Act, 1999 (CEPA). Aligned with these ongoing efforts, in March 2021, Health Canada published the Science Approach Document (SciAD) entitled Bioactivity Exposure Ratio (BER): Application in Priority Setting and Risk Assessment followed by a 60-day public comment period (Health Canada, 2021). Comments received during the 60-day consultation and the Government's responses are provided in the summary of public comments.

The SciAD demonstrated the utility of in vitro bioactivity data in quantitative risk-based prioritization and assessment for human health. However, the approach is regarded as dynamic, and it continues to evolve as new sources of information become available and research progresses. Provided in this document are the modifications and updates to the overall BER approach since the SciAD was first published in 2021 (Health Canada, 2021). Some updates have been made in response to the public feedback on the approach, while others relate to advances in the interpretation and use of in vitro data to derive a bioactivity threshold. The revisions cover 3 broad areas, namely: version updates to databases and tools; changes to in vitro assay filtering criteria; and consideration of in vitro distribution.

2 Methods

2.1 Overview of general approach

In evaluating the potential for human health effects of a substance, a risk assessment determines a level at which adverse health effects occur (that is point of departure), applies factors to account for areas of uncertainty, and compares human exposure estimates against this level to determine risk. Health effects are adverse if they result in functional impairment or pathological lesions that may affect the lifespan of the organism, its ability to reproduce, or reduce the ability of the organism to respond to an additional challenge (US EPA 2011; Lewis et al. 2002; IPCS 2004). In contrast, the bioactivity exposure ratio approach presented here, uses in vitro assays to derive a molecular-based point of departure (POD_Bioactivity) but does not necessarily determine a level at which specific adverse health effects would occur. Rather, it uses perturbations observed in in vitro assays covering a broad biological range of possible biochemical and cellular targets that may form the basis of events in an adverse outcome pathway but are not indicative on their own of a defined adverse health effect. It is expected that these initial biological perturbations occur at lower concentrations than the doses at which downstream adverse health effects manifest following longer term in vivo exposures (Becker et al. 2015; Honda et al. 2019).

The derivation of POD_Bioactivity requires the selection of a threshold concentration observed within the in vitro test system as well as selection of metabolic parameters of a population for in vitro to in vivo extrapolation (IVIVE). Thus, POD_Bioactivity can be tailored depending on the main objective of the intended use. The underlying premise of the approach presented here is that a minimal concentration, corresponding to a bioactivity threshold observed in a broad range of in vitro assays, can be coupled with in vitro to in vivo extrapolation (IVIVE) to estimate a surrogate point of departure (POD_Bioactivity) suitable for the protection of the general population. Conservative assumptions are used within this approach and thus the POD_Bioactivity is intended to be a protective estimate of effect levels that could be observed in vivo independent of the biological events or adverse outcome pathways involved (Paul Friedman et al. 2020; Paul Friedman et al, 2025). In vitro bioactivity is derived from the ToxCast database and while numerous biochemical and cellular assays are employed, it is acknowledged that it does not cover all biological targets or processes.

The methods for deriving POD_Bioactivity follow the methods outlined in Paul Friedman et al. (2020). A generic workflow was developed that illustrates these broad steps:

1. Extract bioactivity data from the ToxCast database for each chemical of interest
2. Apply filtering criteria to remove AC50 values from curve fits of ToxCast data that may be less quantitatively informative
3. Calculate the 5th percentile from the distribution of AC50 values from active assay endpoints to represent an in vitro bioactivity threshold per chemical
4. Use httk R Package to estimate an administered dose equivalent corresponding to the in vitro bioactivity threshold to represent the POD_Bioactivity

Although the broad steps largely remain the same as published in the original SciAD certain changes have been made in their execution which are described in detail below.

2.2 Updates

2.2.1 Databases and tools

ToxCast database

Major updates were introduced in the latest release of the ToxCast database branded invitroDB version 4.1. The new database improves how data is stored and analyzed for studying chemical effects. A new database schema was developed using improved R packages (tcplFit2, tcpl v3.0.1) that lead to more consistent curve-fitting and hit-calling of multi-concentration response data. In addition to improved dose-response curve fits, the database updates include changes to caution flags and cytotoxicity calculations. Details of the database updates and their impact on the new outputs were investigated in Feshuk et al., 2023.

The most significant change in the update is a notable decrease in the total number of assay endpoints curves from InvitroDB version 3.5 to 4.1. New modelling capability of tcplFit2 to address bidirectional or unidirectional dose response datasets eliminates the need for redundant curve-fitting in the previous database version. Curve-fitting models grew from 2 to 7 to evaluate the concentration-responses dataset, including polynomial models for bidirectional curves. Past modelling efforts required to curve-fit twice bidirectional dose-responses since the old models could only fit unidirectionally. In addition to multiple curve-fitting models estimating biological activity concentration (AC50), Benchmark Concentration (BMC) calculations were also introduced in the latest release of the database.

IVIVE using httk R package

A critical piece of the derivation of POD_Bioactivity is the calculation of administered equivalent dose (AED) using steady state toxicokinetics and IVIVE of the threshold in vitro bioactivity concentration. The approach has been updated to use the latest release of the US EPA httk R package version 2.3 (Pearce et al. 2017, Breen et al. 2021) to extrapolate in vitro bioactivity concentrations into in vivo doses. In addition to corrections made for computational model equations and simulation algorithms, new toxicokinetic values were added for hundreds to thousands of chemicals depending on the type of parameters. In vitro experimental data for hundreds of chemicals were obtained for caco-2 membrane permeability oral uptake, blood or plasma binding (fup) and hepatic intrinsic metabolic clearance (Clint). Moreover, computational regression models and quantitative structure property relationships (OPERA version 2.9, Dawson et al. 2021, Pradeep et al. 2020) help determine toxicokinetic parameters for thousands of chemicals, notably for chemicals with no existing experimental values for fup and Clint.  

All the POD_Bioactivity calculations were generated using the same approach described in the original SciAD document (Health Canada, 2021). In vitro to in vivo concentration scaling factors were obtained by using the ratio between the modeled human steady state blood concentration over a standardized dose of 1 mg/kg/day (Wetmore et al., 2015). For this work, the steady state concentrations were obtained by simulating an orally dosed 3-compartment model from the latest httk R package. While the calculation method remained the same, new POD_Bioactivity values were derived as a result of revisions in computing algorithms updated in the latest httk R package.  These POD_Bioactivity values were then compared to the exposure concentrations to derive BER values.

2.2.2 In vitro assay filtering criteria

A critical aspect of the bioactivity approach is to examine the individual ToxCast chemical-assay concentration response curves and respective AC50 values that make up the overall total distribution of results that are used to calculate the 5th percentile which represents the bioactivity threshold. The calculation of the 5^th percentile of the AC50 values can be influenced by the inclusion of spurious results and concentration response curves that do not represent a true response of a chemical in each assay.

Modification of previous filtering rules

The initial approach used generic filtering criteria for assays with an active hit-call with the aim to eliminate less reproducible or less reliable activity calls and respective AC50 values for quantitative use when deriving the POD_Bioactivity. The first portion of the filtering process removed assays that had 3 or more caution flags and a hit percent (that is, the % of 1,000 bootstrap curve-fits runs that are classified as a hit) of less than 50% (both conditions needed to be met for filtering to apply). The latest version of tcpl does not provide a discrete call of activity in a given assay but rather uses a continuous value which is based on the product of 3 proportional weights: at least one median response exceeds the cutoff, the top of the fitted curve is above the cutoff and whether the winning model is not fit to background noise (that is, Akaike Information Criterion of winning model is less than that of the constant model). As is suggested in the tcpl guidance, where a hit call estimate exceeds 0.9 the chemical is considered “active” in the given assay. Hit percent is no longer used as a method of filtering out assays and the continuous value cut-off of greater than 0.9 is now used. The presence of 3 of more cautionary flags is still used to filter out assays from the distribution of AC50 values to calculate the 5th percentile.

The second step in the original filtering process was to remove the tcpl curve fit categories 36 and 45. Fit category 36 corresponds to Hill model fits where the model top (top of the curve fit) is less than or equal to 1.2 times the threshold cut-off for a positive response and an AC50 value less than or equal to the lower limit of the concentration range screened. In other words, the maximal fitted response (or efficacy) is only slightly above the threshold where an assay is considered active and an AC50 value was estimated to be below the lowest concentration screened. Similarly, fit category 45 indicates the same criteria but for a gain-loss model fit. These fit categories are thought to be less quantitatively informative because the efficacy is borderline and the estimated AC50 is in a concentration range where there are no actual data to inform the slope of the curve. In the latest version of tcpl, the process of deriving fit categories was changed to account for more curve-fitting models. Changes were made to include other models rather than constant, Hill, and gain-loss, and fit category numbering was changed as a result. A more generic approach to fit category is now used across models. Fit category is largely based upon the relative efficacy and, in the case of actives, the location of the AC50 and concentration at 95% activity (an estimate of maximum activity concentration, AC95) compared to the tested concentration range. Using the new fit categories numbering, fit categories 36 and 40 are used to filter out assays as the AC50 is less than or equal to the minimum concentration tested and may indicate AC50 values that are less quantitatively informative than AC50 values within the concentration range screened.

Cytotoxicity filters

In the initial SciAD, cytotoxicity was not considered when filtering out AC50 values from active ToxCast assays. It was possible that some of the bioactivity observed in the distribution of active assays is confounded by cytotoxicity and the “burst” phenomenon, where large numbers of assays begin to show activity near cytotoxic concentrations. It is possible that this could result in an increased calculated value of the 5th percentile AC50. Thus, to mitigate this potential, a tiered approach was devised to filter out AC50 values that are above likely cytotoxic concentrations.

The first step in applying the cytotoxicity filters is to filter out assays from the distribution where the AC50 is above the minimum AC50 from a matching cytotoxicity assay from the same assay source platform. For example, if the AC50 value from the ACEA Biosciences assay that measures androgen receptor antagonism (ACE_AR_antagonist) is above the minimum AC50 value from the ACEA Biosciences assay that measures cytotoxicity in the same assay system (ACEA_AR_agonist_AUC_viability) then that assay is removed from the distribution of active assays that are used to calculate the bioactivity threshold. One for one matching of cytotoxicity assays within an assay source platform is not always available. Thus, for the second-tier filter for cytotoxicity, assays are removed from the distribution where the AC50 is above the median AC50 of the collection of cytotoxicity assays conducted using the same cell line. Finally, where cytotoxicity assay matching based on cell line is not available, assays are removed from the distribution if the AC50 of the given assay is above the median AC50 value derived from all the ToxCast burst assays. The tiered cytotoxicity filters are applied in sequential steps. By using this approach, cytotoxicity is considered using the best available match for a given assay/chemical pair.

Additional filters developed by the National Toxicology Program (NTP)

The Integrated Chemical Environment (ICE) provides an online experience to explore curated toxicity data and computational tools. A robust curation process addresses inconsistency of the experimental data or variance in interpreting the results. These inconsistencies may be present due to differences in reported units, effects using synonymous terms or methods applied due to a lack of standard approaches. These curation processes ensure the data is well-structured and reliable to support chemical risk evaluations for human health.

Databases such as EPA’s ToxCast (invitroDB 4.2) underwent a rigorous curation process by the NTP using automated and expert-driven approaches with the output of the exercise being made available in Integrated Chemical Environment (known as cHTS data). The curation also enhances the data integrity for various analyses within ICE, including external workflows. For the latest update of the Health Canada approach, the following curation process from cHTS was applied to exclude or supplement information for possible suspect data.

InvitroDB active calls were assigned once the best-fit curve model reflected a dose-response relationship and exceeded a minimum activity cutoff threshold. For these active calls, ICE added “Flag-Omit” calls to cHTS where the model fits are questionable due to the subsequent reasons (ICE 4.2 Release Notes, July 2025):

• Any active calls with less than 50% efficacy for NovaScreen (NVS) cell-free biochemical assays
• Any active calls where only the lowest concentration tested exceeded the assay activity cutoff threshold, for NCATS Tox21 assays
• Any active calls where only a single mid-range concentration had activity above the activity cutoff threshold
• Any active calls where the AC50 was extrapolated below the testing concentration range
• Any active calls where the AC50 was extrapolated to a concentration above the tested range
• The direction of the curve does not match the intended direction of analysis, resulting in a negative hit call
• Any active calls with flags from the tcpl pipeline that suggest borderline efficacy, and likely overfitting (flags 11 and 16)

The cHTS data also omitted assays from InvitroDB with Tanguay zebrafish assays, and Tox21 assay endpoints comprising channel readouts. The zebrafish assays were omitted since the IVIVE models were only designed for mammalian physiology. For the revision of the POD_Bioactivity calculation using a recent version of invitroDB (4.1), of the 3000 entries filtered prior to this level of the curation, only 140 were excluded due to the reasons above.

The cHTS data also includes chemical Quality Control (QC) flags matched against the InvitroDB database. QC information from the NCATS Tripod Tox21 Data Browser (accessed March 2021) were compared to each sample or chemical when possible. Poor QC grades were given for low purity (<50%), low concentration, poor analytical detection, or incorrect defined molecular weight. The assays with poor QC grades were then labeled as "QC-Omit" and filtered from the ICE database.

Technological interference flags are used to indicate when assay responses might be influenced by the chemical's interaction with the assay technology rather than true bioactivity. Luminescence or fluorescence readouts between samples can be affected by technological interference, leading to false signals. In cHTS, these flags are distinct from QC-omit and Flag-omit in that they provide supplementary information and do not affect the activity calls.

In our revision of the POD_Bioactivity calculation, the assays with poor QC grades were not excluded since those issues were likely addressed in the latest version of invitroDB (4.1). Among the 3000 entries screened up to that stage, over 700 were associated with poor QC grades according to the ICE curation process. The interference flags were also noted but not excluded from the revised invitroDB database. Among the 3000 filtered entries, 800 were notified for interference. Most of the interference was due to neighboring luciferase signals. While the updated approach does not automatically filter out assay curves based on QC-Omit and Technological interference flags, these flags were added on as supplementary information and will be manually considered when applying the approach in the future when using ToxCast data.

2.2.3 Consideration of In vitro distribution

The approach documented in the original SciAD uses the total assumed (that is, nominal) concentration in the media from the in vitro assays as the metric of bioactivity to carry forward for IVIVE and POD_Bioactivity calculations. Over the past 10 years a significant body of evidence from modelling and data analysis has emerged in the peer-reviewed literature to show that the nominal in vitro concentration may not be the best metric for extrapolation for all substances (Armitage et al., 2014; Fischer et al., 2017; Casey et al., 2018; Fischer et al., 2019; Henneberger et al., 2019; Escher et al., 2020; Fischer et al., 2020; Henneberger et al., 2020; Huchthausen et al., 2020). In vitro test systems are multicomponent environments, and chemicals can be distributed differently in these systems based on their properties and the properties of the test system (NAS, 2017; Armitage et al., 2014; Fischer et al., 2017). For example, it is possible that free chemical concentrations in a test system may be lower because of sorption to components in the test system (for example, plasma proteins or lipids, vessel walls). Moreover, there may be less freely dissolved chemical in the test system as certain chemicals will partition more readily to the headspace based on volatility (Armitage et al., 2014; Fischer et al., 2017). Furthermore, different assay compositions may also confound the interpretation of relative potency and hazard across assays for the same chemical and when comparing different chemicals. These factors can influence the interpretation of a chemical’s bioactive concentration and can ultimately lead to less accurate assessment of potential bioactivity in the population and potential health effects (Casey et al., 2018).

Therefore, it is critical in the development of the BER approach for using in vitro toxicity data that the issues related to in vitro partitioning be analyzed and the potential impact on bioactivity-based potency for risk-based priority and assessment efforts be determined.

To account for in vitro distribution, an In vitro Mass Balance Model (IV-MBM) was developed by Armitage and colleagues in 2014 as a relatively simple and effective tool that can be readily parameterized for a range of organic chemicals and in vitro bioassays. It has been upgraded recently to account for sorption to plastic (vessel walls) and to allow the user to simulate ionizable organic chemicals (IOCs) in addition to neutral chemicals. The model can be parameterized for different bioassays (if these parameters are known) and requires physical-chemical properties and the nominal dose concentration as input parameters. The IV-MBM model calculates the steady-state, equilibrium mass distribution of a chemical in an in vitro assay system including the headspace (air) concentration, the dissolved (free) concentration in the water phase, the cell concentration, and the cell membrane concentration. These initial calculations are then used to calculate a chemical- and assay-specific Depletion Factor (DF), which is a ratio of the nominal concentration to the freely dissolved concentration (DF = C_NOM/C_FREE), and Enrichment Factor (EF), which is the ratio of the cell concentration to the nominal concentration (EF = C_CELL/C_NOM). The model can also inform when volatilization of the chemical into the headspace is high, when administered chemical may exceed the solubility of the system (that is, saturation and precipitation of neat chemical occurs), and when cell membrane concentrations approach or exceed values corresponding to baseline toxicity. C_FREE may provide the best metric for comparing chemicals for relative potency within and across assay systems. These calculated parameters can be used to adjust the calculation of the in vitro bioactivity threshold.

To apply the IV-MBM model certain assay parameters are required such as total lipid content, mass per cell, initial seeding density, total protein content and fetal bovine serum (FBS) volume fraction. Not all assay parameters are available for the ToxCast assays that are used in the BER approach.

To gauge the impact of in vitro distribution on the calculation of the in vitro bioactivity threshold, a pilot project, in consultation with Arnot Research and Consulting, was conducted using a subset of ToxCast assays where activity was shown for 851 chemicals on Canada’s Domestic Substances List (DSL). The general findings for the DSL chemicals examined are consistent with previous peer-reviewed publications in that DFs and EFs can be very large for certain chemicals, and thus the assumption that the nominal concentration is appropriate for comparing chemical hazard (potency) and for extrapolations in risk-based assessments may be less reliable. The DFs can be > 1000 for many DSL chemicals and span over 10 orders of magnitude.

The pilot project identified generalizable findings. First, neutral organic chemicals with log K_OW > 4 and/or log K_AW > -1 are expected to exhibit DFs greater than 10 in a typical ToxCast assay. Moreover, the largest EFs will occur for hydrophobic chemicals (log K_OW > 4) in assays with lipid-rich cells and FBS-poor medium. Other interesting and significant findings were that a fraction of DSL chemicals across examined assays (~5–10%) have reported AC50s where the predicted initial water concentration exceeds the water solubility and approximately <1% to 5% of simulated chemicals are predicted to be present largely in headspace (≥ 2/3 of the administered dose) across the selected assays, not in test medium to which cells are exposed (that is, a “volatility issue”).

Based on the findings of the pilot study, work will continue to collect relevant ToxCast assay parameters to apply the IV-MBM. In the meantime, it is known that greater uncertainty applies to chemicals that have high hydrophobicity or are volatile and points to the importance of chemical characterization and suitability determination for a given test system prior to analysis.

3 Results

After adjusting the previous bioactivity approach published in the original SciAD to account for database updates and new filtering criteria, many chemicals evaluated were only marginally affected (within one log unit) with respect to the calculation of the POD_Bioactivity (Figure 3-1). The largest mg/kg-bw/day increases for POD_Bioactivity occurred for 10-Undecenoic acid; o-cresol; 2-Butanone, oxime; Bis(2-ethylhexyl)-hexanedioate, and N-Vinyl-2-pyrrolidone (that is, less conservative). The largest decreases for POD_Bioactivity occurred for o,p’-DDD, resorcinol and 1,2,3,5-Tetrachlorobenzene (that is,. more conservative).

Comparison of POD_Bioactivity calculated using the original SciAD approach (Version 1) against the updates proposed in this technical update document (Version 2).

Comparison of POD_Bioactivity with POD_Traditional from animal studies.

When comparing the minimum POD_Traditional from existing assessments, the POD_Bioactivity generated using the revised approach was lower than the minimum POD_Traditional (that is, lowest NO(A)EL or LO(A)EL) examined during the risk assessment for ~91% of the 43 chemicals covered under the case study (that is, log₁₀POD ratio > 0) (Figure 3-2). This proportion is in line with the previous findings of the original SciAD. The median value for the log₁₀POD ratio was found to be 1.79 which is slightly lower than the original SciAD (2.24) but is within the same order of magnitude. This median value translates to POD_Bioactivity being ~100-fold lower than the POD_Traditional on an arithmetic scale. Four chemicals were found to have minimum POD_Traditional values lower than the POD_Bioactivity (furfual, bisphenol A, allyl chloride, and diisononyl phthalate). Both bisphenol A and allyl chloride were found to also not have a protective POD_Bioactivity in the original SciAD; details can be found therein.

BERs for 38 substances were determined by comparing the POD_Bioactivity and the maximum exposure values for each chemical extracted from the respsective risk assessments. On a log scale, the metric for the comparison between the 2 values is the log₁₀BER, which is the difference between the log₁₀POD_Bioactivity and the log₁₀Exposure value (that is, log₁₀BER = log₁₀POD_Bioactivity - log₁₀Exposure and BER = POD_Bioactivity/Exposure).

For the 38 substances, the log₁₀BER was found to be less than 0 for 11 substances indicating that exposure exceeds the bioactivity-based point of departure. For 16 substances the ratio was between 0 and 2 indicating that exposure is within 2 orders of magnitude of the bioactivity-based point of departure. The remaining 11 susbtances had log₁₀BER exceeding 2 (Figure 3-3). In some cases, low log₁₀BERs (for example, < 2) were primarily driven by high exposure values (for example, 2-ethylhexanoic acid, whereas in other cases, a low log₁₀BER was driven by a very low POD_Bioactivity (for example, tricresyl phosphate). Six of the 7 substances with exposure data, that were originally concluded as a potential risk to human health under CEPA section 64(c), were identified with a log₁₀BER of less than 2. Quinoline, also concluded to meet criteria 64(c) under CEPA had a very high log₁₀BER, based on a very low exposure value and a high POD_Bioactivity. In this case, the 64(c) conclusion was driven by the genotoxic potential of quinoline and was not based on a quantitative risk characterization. Complementary NAM-based approaches have been proposed to adress genotoxicity and are being explored outside of this SciAD (Beal et al. 2023).

Comparison of Toxcast derived PODBioactivity, PODTraditional, and maximum exposure values.

4 Discussion

Overall, updates to the derivation of the POD_Bioactivity using ToxCast data continues to demonstrate the utility of in vitro bioactivity data in quantitative risk-based prioritization and screening assessment.

Since the publication of the original SciAD, further analysis has been conducted as part of the Accelerating the Pace of Chemical Risk Assessment (APCRA) international working group looking specifically at the concordance and level of protection between in vitro based bioactivity approaches and non-clinical/clinical toxicity data for pharmaceuticals (Weitekamp et al., 2025). Among other goals, one important aspect of the work is to inform the selection of appropriate uncertainty factors that provide similar levels of protection compared to using animal toxicity tests. In the study, authors evaluated the quantitative and qualitative concordance of lowest observed adverse effect levels (LOAELs) and adverse endpoints between in vivo (rat and mouse) and in vitro (ToxCast coupled with IVIVE) models of human health and human clinical trials of pharmaceuticals. One of the key findings suggests that rodent based LOAELs were generally higher than the human LOAEL values (that is, not health protective), but when combined with typical composite uncertainty factors (that is, 100-1000) were protective of >97% of drugs evaluated. Whereas in vitro bioactivity POD values were lower than the human LOAEL values (i.e. health protective) and would require lower composite uncertainty factors to achieve the same level of human health protection. POD_Bioactivity defined by using the 5th percentile from a distribution of ToxCast assays and following a similar IVIVE approach used in the SciAD (that is, selection of the 0.95 (sensitive individual) quantiles from a Monte Carlo simulation of interindividual variability in toxicokinetic parameters) coupled with an uncertainty factor of 10 resulted in protection for >95% of drugs studied (Weitekamp et al., 2025). The study comparisons were conducted using LOAEL values. It is generally the practice of risk assessments conducted under CEPA to compare to the no observed adverse effect level (NOAEL) values and where a NOAEL is not available a LOAEL can be substituted with an additional uncertainty factor of 10. Thus, a composite uncertainty factor of 100, applied to POD_Bioactivity could provide a similar level of protection compared to the standard approaches used in human health risk assessment.

The original SciAD proposed using a default uncertainty factor of 100 to guide the evaluation of the BER to support priority setting and risk-based screening (that is, substance binning). This uncertainty factor is considered protective and is expected to cover the potential gaps in biological space covered by the ToxCast assays along with the uncertainties associated with using cell-based assays and IVIVE methods. Provided there is adequate confidence in the exposure prediction, the BERs for known toxics analyzed in the original SciAD and the update presented here suggest that a BER of less than 100 would indicate that the chemical is a higher priority for further action. The results from adjusting the methods to derive the POD_Bioactivity as presented in this update do not suggest the need to alter the default uncertainty factor at this time. It is currently considered fit for the purposes of priority setting and rapid risk-based screening.

 As outlined in the original SciAD, the BER approach is intended to be dynamic. Technical refinements and updates will continue on an ongoing basis as new information emerges and research advances, ensuring the approach remains scientifically robust and aligned with the latest developments.

5 References

Armitage JM, Wania F, Arnot JA. 2014. Application of mass balance models and the chemical activity concept to facilitate the use of in vitro toxicity data for risk assessment. Environmental Science & Technology 48:9770-9779. 

Beal MA, Audebert M, Barton-Maclaren T, Battaion H, Bemis JC, Cao X, Chen C, Dertinger SD, Froetschl R, Guo X, Johnson G, Hendriks G, Khoury L, Long AS, Pfuhler S, Settivari RS, Wickramasuriya S, White P. 2023. Quantitative in vitro to in vivo extrapolation of genotoxicity data provides protective estimates of in vivo dose. Environ Mol Mutagen. 2023 Feb;64(2).

Becker RA, Ankley GT, Edwards SW, Kennedy SW, Linkov I, Meek B, Sachana M, Segner H, Van Der Burg B, Villeneuve DL, et al. 2015. Increasing scientific confidence in adverse outcome pathways: application of tailored Bradford-Hill considerations for evaluating weight of evidence. Regul Toxicol Pharmacol 72:514-537.

Breen M, Ring CL, Kreutz A, Goldsmith MR, Wambaugh JF. High-throughput PBTK models for in vitro to in vivo extrapolation. Expert Opin Drug Metab Toxicol. 2021 Aug;17(8):903-921.

Casey WM, Chang X, Allen DG, Ceger PC, Choksi NY, Hsieh J-H, Wetmore BA, Ferguson SS, DeVito MJ, Sprankle CS, Kleinstreuer NC. 2018. Evaluation and Optimization of Pharmacokinetic Models for In vitro to In vivo Extrapolation of Estrogenic Activity for Environmental Chemicals. Environmental Health Perspectives 126:097001. 

Dawson DE, Ingle BL, Phillips KA, Nichols JW, Wambaugh JF, Tornero-Velez R. 2021. Designing QSARs for Parameters of High-Throughput Toxicokinetic Models Using Open-Source Descriptors. Environ Sci Technol 55 (9):6505-6517.

Escher BI, Henneberger L, Konig M, Schlichting R, Fischer FC. 2020. Cytotoxicity Burst? Differentiating Specific from Nonspecific Effects in Tox21 in vitro Reporter Gene Assays. Environmental Health Perspectives 128. 

Fischer FC, Abele C, Henneberger L, Kluver N, Konig M, Muhlenbrink M, Schlichting R, Escher BI. 2020. Cellular Metabolism in High-Throughput In vitro Reporter Gene Assays and Implications for the Quantitative In vitro-In vivo Extrapolation. Chemical Research in Toxicology 33:1770-1779. 

Fischer FC, Henneberger L, Schlichting R, Escher BI. 2019. How To Improve the Dosing of Chemicals in High-Throughput in vitro Mammalian Cell Assays. Chemical Research in Toxicology 32:1462-1468. 

Fischer FC, Henneberger L, Konig M, Bittermann K, Linden L, Goss KU, Escher BI. 2017. Modeling Exposure in the Tox21 in vitro Bioassays. Chemical Research in Toxicology 30:1197-1208. 

Feshuk M, Kolaczkowski L, Dunham K, Davidson-Fritz SE, Carstens KE, Brown J, Judson RS, Paul Friedman K. The ToxCast pipeline: updates to curve-fitting approaches and database structure. Front Toxicol. 2023 Sep 21;5:1275980.

Health Canada 2021. Science approach document – Bioactivity exposure ratio: Application in priority setting and risk assessment. Ottawa (ON): Government of Canada.

Henneberger L, Muhlenbrink M, Heinrich DJ, Teixeira A, Nicol B, Escher BI. 2020. Experimental Validation of Mass Balance Models for in vitro Cell-Based Bioassays. Environmental Science & Technology 54:1120-1127. 

Henneberger L, Muhlenbrink M, Konig M, Schlichting R, Fischer FC, Escher BI. 2019. Quantification of freely dissolved effect concentrations in in vitro cell-based bioassays. Archives of Toxicology 93:2295-2305. 

Huchthausen J, Muhlenbrink M, Konig M, Escher BI, Henneberger L. 2020. Experimental Exposure Assessment of Ionizable Organic Chemicals in In vitro Cell-Based Bioassays. Chemical Research in Toxicology 33:1845-1854.

Honda GS, Pearce RG, Pham LL, Setzer RW, Wetmore BA, Sipes NS, Gilbert J, Franz B, Thomas RS, Wambaugh JF. 2019. Using the concordance of in vitro and in vivo data to evaluate extrapolation assumptions. PloS one 14(5):e0217564.

[IPCS] International Programme on Chemical Safety. 2004. IPCS Risk Assessment Terminology. Geneva (CH): World Health Organization.

Lewis R, Billington R, Debryune E, Gamer A, Lang B,Carpanini F. 2002. Recognition of adverse and nonadverse effects in toxicity studies. Tox Pathol 30:66-74.

[NAS] National Academies of Sciences and Medicine. 2017. Using 21st Century Science to Improve RiskRelated Evaluations. The National Academies Press, Washington, DC. 

Paul Friedman K, Gagne M, Loo H, Karamertzanis P, Netzeva T, Sobanski T, Franzosa J, Richard A, Lougee R, Gissi A, et al. 2020. Utility of in vitro bioactivity as a lower bound estimate of in vivo adverse effect levels and in risk-based prioritization. Tox Sci 173(1):202-225.

Paul Friedman K, Thomas RS, Wambaugh JF, Harrill JA, Judson RS, Shafer TJ, Williams AJ, Lee JYJ, Loo LH, Gagné M, Long AS, Barton-Maclaren TS, Whelan M, Bouhifd M, Rasenberg M, Simanainen U, Sobanski T 2025. Integration of new approach methods for the assessment of data-poor chemicals, Tox Sci 205(1): 74–105.

Pearce RG, Setzer RW, Strope CL, Wambaugh JF, Sipes NS. 2017. Httk: R package for high-throughput toxicokinetics. J Stat Softw. 79:1.

Pradeep P, Patlewicz G, Pearce R, Wambaugh J, Wetmore B, Judson R. 2020. Using chemical structure information to develop predictive models for in vitro toxicokinetic parameters to inform high-throughput risk-assessment. Comput Toxicol 16:100136.

[US EPA] US Environmental Protection Agency. 2011. Integrated risk information system glossary. 

Wetmore BA. 2015. Quantitative in vitro-to-in vivo extrapolation in a high-throughput environment. Toxicology 332:94-101.