Cancer weight of evidence for three lower acrylates: Conclusions and recommendations from an expert panel

.


Introduction
Acrylates are esters of acrylic acid with carbon chains of varying length.As a group, acrylates are important industrial chemicals used in the synthesis of polymers that in turn are used in a variety of consumer products, including adhesives, cosmetics, paints and coatings, plastics, synthetic flavors, and textiles (IARC, 2019).Human populations can be exposed to acrylates by multiple pathways.These exposures are primarily via inhalation and dermal contact, while ingestion is generally considered negligible, and include occupational exposures and through use of consumer products that contain them.Occupational exposures to acrylates are generally low (<5 ppm), although peak measurements that exceed 10 ppm have been reported for specific tasks, which were accompanied by use of personal protective equipment by workers (Suh et al., 2018).Trace levels of some acrylates (e.g., ethyl acrylate<20 ppm; 2-ethylhexylacrylate<100-1000 ppm) have been detected in copolymers used for cosmetics.The potential for environmental exposures to these acrylates via air, water and soil are generally considered low.Natural sources of methyl and ethyl acrylates include certain fruits (e.g., pineapples) (Suh et al., 2018).Acrylates have strong odors with low odor thresholds thereby limiting occupational exposures (Suh et al., 2018), and as reactive chemicals produce irritation and toxicity at the point of contact (Brüning et al., 2014).Once absorbed, acrylates are rapidly metabolized through two pathways: (1) hydrolysis via carboxylesterase to form acrylic acid and an alcohol corresponding to the carbon side chain; and (2) conjugation with glutathione (Suh et al., 2018;IARC, 2019).Metabolism of acrylates is considered to be detoxifying, and due to its rapid nature (several minutes), systemic doses of acrylates are generally expected to be much lower than those experienced at the point of contact.
This assessment was defined to consider the cancer weight of evidence and most likely modes of action (MOA) for three lower molecular weight acrylates: (1) methyl acrylate (MA); (2) ethyl acrylate (EA); and (3) 2-ethylhexyl acrylate (2EHA).All three acrylates are generally considered to be non-genotoxic (Suh et al., 2018).Although epidemiology data are largely lacking for this group of chemicals, a variety of well conducted toxicology studies are available on these acrylates.A summary of the chronic studies is provided in Table 1.Multiple rodent cancer bioassays have been conducted for lower acrylates, including those for MA by the inhalation route (Reininghaus et al., 1991;JBRC, 2017), EA by the inhalation, oral, and dermal routes (DePass et al., 1984;Miller et al., 1985;NTP, 1986;Ghanayem et al., 1994), and 2EHA by the dermal route (DePass et al., 1985;Wenzel-Hartung et al., 1989;Mellert et al., 1994).
Low-to-moderate increases in tumor incidences have been noted in some cancer bioassays (Table 1).Point of contact tumors, including nasal tumors following inhalation (JBRC, 2017), forestomach tumors following oral gavage administration (NTP, 1986; but not via drinking water exposure; Borzelleca et al., 1964), and skin tumors following dermal contact (DePass et al., 1985), are noted in some studies.However, there are also multiple studies reporting no increase in point of contact tumors following chronic exposures to acrylates (JBRC, 2017;Reininghaus et al., 1991;Miller et al., 1985;Mellert et al., 1994).Several studies have reported weak increases in the incidence in systemic tumors, including combined soft tissue sarcomas (male rats; Reininghaus et al., 1991), combined hematopoietic/lymphoid cancers (male rats; Reininghaus et al., 1991), pituitary adenoma (female rats; Reininghaus et al., 1991), adrenal pheochromocytoma (female rats; JBRC, 2017), and thyroid follicular cell adenoma (female mice and male rats; Miller et al., 1985).These reports of systemic tumors are generally limited to a single study in one sex and rodent species/strain, with weak or non-monotonic dose-response relationships, are not statistically significant (e.g., p > 0.05) when incidence is not combined across tissue sites (e.g., soft tissue sarcomas), and/or the reported incidences are within or near historical control ranges.Furthermore, due to the rapid and extensive metabolism of acrylates (as reviewed in Suh et al., 2018), appreciable systemic doses of the parent chemicals are not expected to arise that could result in systemic tumor increases.Several of these tumor types have been evaluated more extensively in the published literature.
• Rat Nasal tumors Produced by MA -Wibbertmann et al. (2021) concluded that, based on the effects observed in the subchronic and chronic toxicity studies (JBRC, 2017), the highest test concentration of MA tested in rats of 160 ppm exceeded the maximum tolerated concentration (MTC).For this reason, the nasal tumors results observed at this concentration are of questionable relevance to human health risk assessment.Given that these exposure levels are much higher than would be tolerated by humans, the author concluded that the potential to pose a substantial cancer risk is implausible.
• Mouse and Rat Forestomach Tumors Produced by EA -Proctor et al. (2018) developed an adverse outcome pathway (AOP) for EA-induced forestomach tumors in rodents.Based on available evidence for EA, a sustained glutathione depletion is defined as a pre-molecular initiating event, epithelial cytotoxicity is defined as the critical initiating event, with supporting key events identified as increased cell proliferation resulting in sustained hyperplasia, ultimately resulting in the adverse outcome of forestomach tumors.Thompson et al. (2018) concluded that the weight of evidence supporting the MOA combined with the lack of a human tissue homologue makes it highly unlikely that EA exposure poses a cancer risk to human populations.• Mouse and Rat Thyroid Tumors Produced by EA -Rosol and Witorsch (2021) assessed the data for thyroid tumors in mice and rats exposed to EA.The authors noted that the increased incidences while statistically significant (p < 0.05), were small and within the range of historical controls, and did not exhibit a consistent dose-response trend.Based on these observations, along with the fact that most of the thyroid tumors were benign, the authors concluded that these tumors were most likely random observations that were of questionable relevance to humans.• Mouse Skin Tumors Produced by 2EHA -Elmets and Yusuf (2020) reviewed the carcinogenicity data for 2EHA reported that there are no reports of cancer in humans exposed to 2EHA.Furthermore, the authors concluded that the evidence from one strain of mouse (C3H/HeJ; DePass et al., 1984) was of questionable relevance for several reasons: (1) this strain of mouse has a mutation in Toll-like receptor 4 (TLR4) that impairs its immune responses; (2) there was a lack of rigorous histopathologic characterization of tumors that developed; (3) the doses of 2EHA tested in the study were high (i.e., above the maximum tolerated dose); and ( 4) the results of this study were not observed when in a second mouse strain (Mellert et al., 1994).Additionally, 2EHA was found to be not mutagenic in Chinese hamster V79 cells and did not induce micronuclei in human lymphocytes (Murphy et al., 2018).
Studies on the potential genotoxicity for these three acrylates are generally negative (as summarized in IARC, 2019; Suh et al., 2018), with limited positive findings that are associated with cytotoxic test concentrations.
The cancer weight of evidence based on the rodent cancer bioassay data for these three acrylates has been considered by several regulatory and health agencies (Table 2).IARC now considers all three acrylates to be "possibly carcinogenic to humans" (IARC, 2019).In contrast, NTP initially considered EA to be "reasonably anticipated to be a carcinogen" but was delisted in 1998 because (1) the forestomach tumors were seen only at high gavage doses that caused local irritation and subsequent cellular proliferation, (2) chronic human exposures to high concentrations are unlikely, and (3) no other tumors or systemic effects were observed (NTP, 1998).USEPA considered MA to be "not classifiable as to human carcinogenicity" (USEPA, 1990), and EA provisionally as "not likely to be carcinogenic to humans", but has not evaluated 2EHA with respect to its carcinogenicity (USEPA, 2014).These conclusions generally mirror those made by other agencies (e.g., ACGIH, 2023;Health Canada, 2011;European Chemicals Bureau, 2005).
This article presents the results of an expert panel-based review of the cancer WOE for three lower acrylates.The recommendations from this panel will be used to support quantitative risk assessment decisions.The methods and results of this expert panel engagement are described below.

Panel engagement
An expert panel was recruited and engaged utilizing the methods described in Kirman et al. (2019) as modified in Appendix A, with specific roles defined for the review sponsor, review manager (SciPinion; authors CRK, SMH), and independent expert panel members.Multiple design elements were included in this review to minimize potential sources of bias and groupthink, and to improve transparency of the review.These elements include the following: (1) a hybrid-blinding process, between single-and double-blinded was adopted for panel recruitment and engagement to minimize potential participation bias (2) the identities of experts were masked (e.g., labeled as Expert 1, Expert 2, etc.) during all online deliberations, (3) a multi-round, modified Delphi format was adopted to collect both independent and deliberative input from the topic experts in an effort to minimize potential groupthink; (4) panelists were specifically asked if there were any issues in the review material that warranted attention and discussion by the panel, in an attempt to minimize potential scope bias associated with charge questions that are too narrowly focused; (5) individual responses and comments from the panelists were recorded and are provided in their entirety (Appendix B) to provide precise measurements for the degree of consensus, ensure transparency, and minimize potential reporting bias; and (6) although individual responses are provided, they are attributed to panelists' anonymous display names (e.g., to Expert 1, Expert 2, etc.) rather than to specific panelist identities in an effort to provide psychological safety (i.e., scientists should feel free to express their scientific opinions without fear of negative repercussions).Based on the robustness of the SciPinion review process, it was recently used to support a cancer weight of evidence decision by USEPA for an unrelated chemical (1,3-dichloropropene; USEPA, 1990).
Review material was defined by the review manager to include several recent reviews/publications as primary review material (IARC, 2019;Suh et al., 2018;Wiench et al., 2022), as well as underlying cancer bioassays and supporting material for the selected acrylates for the panel to consult as needed.Panelists were also given the opportunity to request access to additional publications/reports to support their review as needed.

Cancer weight of evidence framework
Multiple cancer weight of evidence classification frameworks are available from regulatory and health agencies (e.g., IARC, NIOSH, NTP, UN, USEPA), as summarized in ATSDR (2020).For this review, USEPA's Guidelines for Carcinogen Risk Assessment (USEPA, 2014) was adopted as the primary framework as this work is intended to support follow-up work on quantitative dose-response assessments within the United States.USEPA's guidance includes five possible weight of evidence classification descriptors.
• Inadequate Information to Assess Carcinogenic Potential • Not Likely to Be Carcinogenic to Humans • Suggestive Evidence of Carcinogenic Potential • Likely to Be Carcinogenic to Humans • Carcinogenic to Humans In addition, the panelists were asked to consider a 2nd classification framework, United Nations' Globally Harmonized System for Classification and Labelling of Chemicals (UN 2011), which yielded similar conclusions of "carcinogen labeling not required" (see Appendix B).

Consensus analysis
For charge questions on cancer weight of evidence classification, the degree of consensus among reviewers was assessed with precision using the consensus metrics of Tastle and Wierman (2007): where p is the probability (frequency) for the answer option i, μ x is the mean index value, X i is an index for graded or Likert answer option i (e. g., "inadequate evidence" = 0, "not likely carcinogenic" = 1, "suggestive evidence" = 2, "likely carcinogenic" = 3, "carcinogenic" = 4), d x is the width of the index values (X max -X min ), n is the number of answer options.
The value for the consensus metric can range from a value of "0" when there is a complete lack of consensus (i.e., when equal numbers of panelists select the two extreme answer options) to a value of "1" when there is complete consensus (all panel members select the same answer option).All summary statistics, calculations, and figures were prepared using Microsoft Excel (version 15.67).

Panel composition
The panel consisted of seven scientists with expertise in cancer bioassays, mode of action, risk assessment, and weight of evidence.Demographics, affiliations, and expertise metrics for this panel are as follows.

Mode of action
Mode of action (MOA) information is important to key decisions made in cancer risk assessment including those related to human relevance including identification of potentially susceptible populations, calculation of chemical specific adjustment factors/dose measures, and low-dose extrapolation methods used to estimate cancer potency (USEPA, 2014).For the three lower acrylates, confidence in the weight of evidence for potential cancer MOAs was rated on a scale of "-5" (i.e., there is strong refuting evidence) to "0" (i.e., evidence is equivocal) to "+5" (i.e., there is strong supporting evidence) (Fig. 1).
Overall, panel confidence was highest for an MOA involving "Increased cell replication by cytotoxicity" (mean confidence scores ranging from +2.3 to +3.0, reflecting a moderate degree of support).These results reflect the strong evidence of non-neoplastic effects (e.g., irritation, inflammation, hyperplasia, metaplasia) occurring at the point of contact for all three routes of exposure.These non-neoplastic local lesions were found to occur at lower (and similar) concentrations than were the observed tumors (Wibbertmann et al., 2021;Wenzel-Hartung et al., 1989).In addition, observations of reversibility of non-neoplastic effects and lack of tumor response in stop-exposure studies (Miller et al., 1985;Wenzel-Hartung et al., 1989;Ghanayem et al., 1994) were noted as an important consideration.Confidence in the other MOAs ranged from equivocal to strongly refuted, with the following relative ranking of mean scores.An MOA involving "Indirect genotoxicity" (mean scores of − 1.0 to − 0.8) was noted as being at least biologically plausible given the ability of the acrylates to deplete tissue glutathione at high concentrations.Evidence supporting either "Other" (mean scores of − 2.0 to +0.3), "Immune system mediated" (mean scores of − 2.8 to − 1.7), and "Increased cell replication by receptor binding" (mean scores of − 3.5 to − 2.4) are generally negative although based on limited data (e.g., high throughput screening assays).Because of their toxicological profiles, chemical properties, as well as, biological plausibility, the relative weight of evidence for the cytotoxicity/regenerative MOA is considered with more certainty to explain the primary effects of the acrylates.Direct genotoxicity (mean scores of − 3.9 to − 4.1) is considered moderately-to-strongly refuted by the panelists based on largely negative results obtained for genotoxicity studies, with limited positive results associated with cytotoxic test concentrations, or non-relevant exposure routes (e.g., injection).It is recognized that this categorization of individual MOAs here represents a simplification of the underlying biology as they may not be independent of one another, may operate in combination, and/or operate in different tissue sites and/or dose levels.The reader is referred to Appendix B for explanatory text provided by the panelists for charge questions on MOA.

Use of available tumor datasets in risk assessment
As discussed above (see Introduction) there are multiple data sets that could potentially be used to support human health risk assessment for the three lower acrylates (Table 1).Information on MOA can be used to inform how these data should be used (e.g., qualitatively, quantitatively, or not at all).The panelists' conclusions on this topic are summarized in Table 3.
For most tumor data sets, a majority of the panelists indicated that they should either not be used in human health risk assessment or should only be used qualitatively.Factors that influenced this decision include.
1.There is no human tissue counterpart (e.g., rodent forestomach).2. Toxicokinetic factors, such as gavage exposures that yield tissue dose rates that are much greater than expected for drinking water or other exposures.This difference may explain the lack of forestomach tumors in rats exposed to methyl methacrylate or ethyl acrylate via drinking water (Borzelleca et al., 1964).In addition, longer residence times are expected for chemicals in rodent forestomach compared to human esophagus.3. The dose-response relationships for forestomach tumors exhibited non-monotonic behaviors and/or within historical control ranges for common rodent tumors.4. The magnitude of the dose/concentration (e.g., exceeding the maximum tolerated dose) was also a factor considered for human relevance.Some panelists felt this factor should be considered separately from human relevance determination.
The only tumor data sets that garnered minimal support (e.g., from a single panel member) for use in quantitative potency estimates was for data sets that reported no increase in tumors at the highest dose tested.Review of the explanatory text for pertinent charge questions (1.4,2.4, and 3.4 in Appendix B) revealed a strong preference by many panel members to rely upon data sets for non-neoplastic endpoints/precursor lesions to support quantitative, human health risk assessment for the three acrylates.

Cancer weight of evidence
Panelists were asked to classify the cancer weight of evidence for the three lower acrylates as well as indicate their degree of confidence in their conclusion (Fig. 2).
For all three chemicals, a majority of the panelists concluded that the WOE supported a classification of "Not likely to be carcinogenic to humans".The basis for this conclusion includes: (1) observations for irritation effects specifically at the point of entry; (2) evidence supporting a non-genotoxic MOA consistent with cytotoxicity/regenerative proliferation; (3) observations of concentration-dependent responses consistent with a nonlinear/threshold dose-response relationship; and (4) expectations that humans would not be exposed to the levels causing chronic inflammation and cytotoxicity.Specific rationales from individual panel members for supporting this conclusion are provided in Appendix B. The degree of consensus was high, with consensus scores ranging from 0.84 to 0.91 (out of a maximum score of 1), as was the degree of confidence in this classification (mean confidence scores ranging from 7.7 to 8.7 out of a maximum score of 10).Consensus and confidence scores were slightly higher for EA compared to corresponding values for the other two acrylates.A panelist in the minority who indicated "Suggestive evidence of carcinogenic potential" noted that this classification should possibly include the qualifier "at high doses" in one case.In another case a panelist felt that the increases in systemic tumors (Table 1) could not be completely rejected.A similar cancer WOE classification (i.e., carcinogen labeling not required) was obtained when the panelists were asked to consider an alternative framework (United Nations' Globally Harmonized System for Classification and Labelling of Chemicals, UN 2011; see Appendix B).

Need for cancer potency estimation
One of the goals of this work was to collect input from the panel to support decisions made in quantitative human health risk assessments for the lower acrylates (in prep).Along these lines, the panelists were asked whether cancer potency estimates/safety values should be derived for the three lower acrylates (Fig. 3).
A clear majority of panelists (71-86%) do not feel a cancer potency/ safety value is required for these acrylates.These conclusions generally mirror those from the panel for cancer weight of evidence classification (i.e., if a conclusion of "Not likely to be carcinogenic to humans" is reached, then a cancer value is not needed).

Discussion and conclusions
An expert panel was engaged to evaluate the cancer WOE for three lower acrylates, MA, EA, and 2EHA.Genotoxicity studies for this group of chemicals are generally negative, with limited positive associated with cytotoxicity.Per OECD guidelines (OECD, 2015), the highest test concentration should not exceed 55 ± 5%, and care should be taken in interpreting positive results only found in the higher end of this 55 ± 5% cytotoxicity range.Observations for carcinogenicity from rodent cancer bioassays are limited.Weak increases in systemic tumors (Table 1) may not be treatment related and appear to be of questionable relevance to human health.Increases in tumors observed at the point of contact in some studies (Table 1), while clearly treatment related, also appear to be of questionable relevance to human health based on MOA considerations.The overall conclusions of this expert panel can be summarized as follows.
(1) The MOA for point of contact tumors observed in rodent cancer bioassays that is best supported by available data involves increased cell replication by cytotoxicity and regenerative proliferation; (2) The WOE supports a cancer classification of "Not likely to be carcinogenic to humans" based upon USEPA (2014) classification framework, and that carcinogen labeling is not required under the UN Global Harmonization System (UN 2011); (3) A quantitative estimate of cancer potency based on rodent tumor data is not required for these chemicals; and (4) Human health risk assessment for these chemicals should instead rely on non-cancer, precursor endpoints at the point of contact (e. g., hyperplasia).
The degree of consensus across panel members in these conclusions is considered high.The relevance of some tumor types to human health e.g., rodent forestomach tumors, adrenal pheochromocytomas) was questioned by some panel members, a conclusion that is supported by other researcher (Proctor et al., 2007;Adams et al., 2008).A similar conclusion has been reached in the published literature for rodent adrenal pheochromocytoma (Greim et al., 2009), although this issue was not specifically discussed by the panel.
The evidence for each of the three structurally similar acrylates considered individually is strengthened when the totality of the evidence is considered as a group, which is consistent with read-across approaches (OECD, 2014).Read-across is considered appropriate for these acrylates since they share common metabolic pathways (i.e., extensive and rapid carboxylesterase hydrolysis to a common metabolite, acrylate that is not considered to be carcinogenic), and mode of action (i.e., increased cell replication due to cytotoxicity).In addition to conclusions on WOE and MOA for these chemicals, the panelists identified potential data needs (charge questions 1.7, 2.7, and 3.7 in Appendix B), which could be used to increase confidence in their assessments.For example, one panel member suggested that additional studies to provide data that could be used to address modified Bradford-Hill criteria (e.g., essentiality/reversibility, dose-response concordance, biological coherence, biological plausibility, strength of association, temporal concordance, analogy) for hypothesized key events could improve confidence in the proposed MOA.Panelists noted some inconsistencies in the evidence for the role of GSH depletion.
Although these data are not considered essential for their conclusions on MOA, additional data on the role of GSH depletion in the point of contact effects and resolution of potential inconsistencies could improve MOA confidence.Additional toxicokinetic data and modeling of the residual levels of acrylates reaching systemic tissues (i.e., surviving first-pass metabolism at the point of contact) would be helpful for interpreting the biological plausibility of systemic tumor formation.There was some discussion among panelists on the use of exposure information in human relevance and cancer weight of evidence classification decisions.USE-PA's guidelines do specifically permit consideration for exposure information in classification determinations in stating that "convincing evidence that carcinogenic effects are not likely below a defined dose range can be considered for the classification Not likely to be carcinogenic to humans" (USEPA, 2014, p.2-58).While limited information indicates that human exposures to acrylates (Suh et al., 2018;IARC, 2019) are well below those eliciting a tumor response in the cancer bioassays, updated data on the magnitude of human exposures (occupational and in the general population) would serve to bolster this conclusion.
Although the size of the panel (n = 7) does not permit a rigorous statistical analysis, some limited comparisons can be made across panel subgroups.For example, WOE indices averaged across the three acrylates can be compared across panel subgroups (note, a higher mean index value would reflect a shift to the right in Fig. 2 when compared to lower values).Based on this comparison, no significant differences were noted between U.S. panelists (n = 4; mean = 1.2 ± 0.4) vs. non-U.S.panelists (n = 3; mean = 1.1 ± 0.4), or between panelists with current/past industry employment (n = 3; mean = 1.2 ± 0.4) vs. panelists with no industry employment (n = 4; mean = 1.1 ± 0.5).The latter observation is consistent with those of Kirman et al. (2019), which did not find a significant difference in panelist responses when stratified by sector of employment.
The goal of this work is to provide an independent assessment of the cancer weight of evidence data available for three lower acrylates.A panel of topic experts was engaged in reviewing summary material, consulting primary studies as needed, and providing their conclusions and recommendations.The panel's conclusions are in sharp contrast to those of IARC, and are more in line with a classification of Group 3 (Not classifiable as to its carcinogenicity to humans) for these acrylates.The panel's conclusions are consistent with those made previously by other agencies (USEPA, 1990;NTP, 1998;ACGIH, 2023;Health Canada, 2011;European Chemicals Bureau, 2005; Table 2).Confidence in the panel conclusions is considered high, and will serve as valuable input to decisions made in the quantitative dose-response assessments for these chemicals.

Funding body information
This work was funded by BASIC ACRYLIC MONOMER

Fig. 1 .
Fig. 1.Panel Confidence in Potential MOAs for the Carcinogenic Effects of Three Lower Acrylates in Rodent Studies (bars indicate arithmetic mean, error bars indicate standard deviation; This refers to questions 1.3, 2.3, 3.3 in Appendix B).

Table 1
Summary of rodent cancer bioassays for three lower acrylates.
C.R.Kirman et al.

Table 2
Cancer classifications for three lower acrylates.
in Appendix B) a .