A critical review of the 2020 EPA risk assessment for chrysotile and its many shortcomings

Abstract

From 2018 to 2020, the United States Environmental Protection Agency (EPA) performed a risk evaluation of chrysotile asbestos to evaluate the hazards of asbestos-containing products (e.g. encapsulated products), including brakes and gaskets, allegedly currently sold in the United States. During the public review period, the EPA received more than 100 letters commenting on the proposed risk evaluation. The Science Advisory Committee on Chemicals (SACC), which peer reviewed the document, asked approximately 100 questions of the EPA that they expected to be addressed prior to publication of the final version of the risk assessment on 30 December 2020. After careful analysis, the authors of this manuscript found many significant scientific shortcomings in both the EPA’s draft and final versions of the chrysotile risk evaluation. First, the EPA provided insufficient evidence regarding the current number of chrysotile-containing brakes and gaskets being sold in the United States, which influences the need for regulatory oversight. Second, the Agency did not give adequate consideration to the more than 200 air samples detailed in the published literature of auto mechanics who changed brakes in the 1970–1989 era. Third, the Agency did not consider more than 15 epidemiology studies indicating that exposures to encapsulated chrysotile asbestos in brakes and gaskets, which were generally in commerce from approximately 1950–1985, did not increase the incidence of any asbestos-related disease. Fourth, the concern about chrysotile asbestos being a mesothelioma hazard was based on populations in two facilities where mixed exposure to chrysotile and commercial amphibole asbestos (amosite and crocidolite) occurred. All 8 cases of pleural cancer and mesothelioma in the examined populations arose in facilities where amphiboles were present. It was therefore inappropriate to rely on these cohorts to predict the health risks of exposure to short fiber chrysotile, especially of those fibers filled with phenolic resins. Fifth, the suggested inhalation unit risk (IUR) for chrysotile asbestos was far too high since it was not markedly different than for amosite, despite the fact that the amphiboles are a far more potent carcinogen. Sixth, the approach to low dose modeling was not the most appropriate one in several respects, but, without question, it should have accounted for the background rate of mesothelioma in the general population. Just one month after this assessment was published, the National Academies of Science notified the EPA that the Agency's systematic review process was flawed. The result of the EPA’s chrysotile asbestos risk evaluation is that society can expect dozens of years of scientifically unwarranted litigation. Due to an aging population and because some fraction of the population is naturally predisposed to mesothelioma given the presence of various genetic mutations in DNA repair mechanisms (e.g. BAP1 and others), the vast majority of mesotheliomas in the post-2035 era are expected to be spontaneous and unrelated in any way to exposure to asbestos. Due to the EPA’s analysis, it is our belief that those who handled brakes and gaskets in the post-1985 era may now believe that those exposures were the cause of their mesothelioma, when a risk assessment based on the scientific weight of evidence would indicate otherwise.

Keywords:

• Asbestos
• EPA assessment
• mesothelioma
• environmental regulation
• occupational health
• auto mechanics
• encapsulated asbestos
• toxicology
• occupational disease

Introduction

In June 2016, the Frank R. Lautenberg Chemical Safety for the 21st Century Act amended the Toxic Substances Control Act (TSCA), the United States’ primary chemicals management law (EPA 2016). Under the amended statute, the EPA was required to conduct risk evaluations to determine if a chemical substance presented an unreasonable risk of injury to health or the environment under typical use conditions without considering costs or other non-risk factors (EPA 2016). This includes an unreasonable risk to potentially exposed or susceptible subpopulations identified as relevant to the risk evaluation process.

In December 2016, the TSCA Science Advisory Committee on Chemicals (SACC) identified asbestos as one of the “First 10 Chemicals” to undergo the amended risk evaluation process. The SACC is an advisory committee operating in accordance with the Federal Advisory Committee Act and is part of the EPA’s Science Advisory Board (SAB). The SACC's purpose is to examine the quality and relevance of the scientific and technical information being used by the EPA or proposed as the basis for Agency regulations and to advise the Agency on broad scientific matters. This panel was tasked with providing independent advice and recommendations to the EPA on the scientific basis for the chrysotile risk assessment. Subsequently, the EPA decided to examine the import, processing, distribution, disposal, and use of chrysotile asbestos in the United States.

In March 2020, the Draft Risk Evaluation for Asbestos (DRE) was issued by the EPA Office of Chemical Safety and Pollution Prevention (OCSPP) and Office of Pollution Prevention and Toxics (OPPT) (EPA 2020a). In the 310-page report, the EPA assembled, reviewed, and evaluated numerous published and unpublished studies, datasets, and risks for various conditions of use (COUs) for chrysotile asbestos. The EPA reviewed 32 COUs, including the “… use of diaphragms in the chlor-alkali industry, sheet gaskets in chemical production facilities, oilfield brake blocks, aftermarket automotive brakes/linings, other vehicle friction products, and other gaskets” (EPA 2020a). The Agency reportedly used “reasonably” available information to develop a risk evaluation report for chrysotile asbestos that relied on the “best” available science and was based on the weight of the “scientific evidence.” One of the key aspects of this regulatory activity is that the EPA only has responsibility to discuss current and future hazards, not past hazards. Therefore, for regulatory action to be appropriate, the products must be currently sold in the United States, presumably in some appreciable quantity.

Shortly after the EPA released the report to the public, roughly 100 outside scientists responded to the DRE with comments and reviews. On 2 June 2020, the EPA closed the public comment period, and the SACC held a meeting shortly thereafter. The virtual public meeting was held on 8–11 June 2020, and hosted by the SACC panel, which reviewed the Draft Risk Evaluation for Asbestos. Approximately 20 outside scientists gave testimony through presentations via zoom, due to the COVID-19 pandemic. The senior author, Dr. Paustenbach, participated in this meeting and gave oral comments regarding the shortcomings in the Agency’s risk evaluation.

In December 2020, the EPA summarized the external peer review and public comments that the OPPT received for the Draft Risk Evaluation for Asbestos (EPA 2020a). It also provided the EPA’s response to the comments received from the peer review panel and the public. The EPA said that they “appreciated” the vital input provided by the peer review panel and the public and stated that the information resulted in numerous revisions to the DRE (EPA 2020c). A careful review of the final document shows very little sensitivity to the seriousness of the scientific shortcomings identified during the review process and the key conclusions of the original draft document remained functionally unchanged in the final version (EPA 2020b).

In its comments and final document, the Agency repeatedly said that they relied on the views of the SACC committee (EPA 2020b, 2020c). However, this panel did not include many acknowledged asbestos experts or individuals with significant experience with encapsulated asbestos products. For example, no participants of the 2017 Monticello Conference, an event where 60 of the top asbestos researchers gathered for three days in Charlottesville, Virginia, were on the SACC (Weill 2018). In addition, none of the roughly 15 exposure scientists who have published papers on exposure to encapsulated asbestos products or any of the ten or more epidemiologists who have published original work on these asbestos products were on the SACC committee. Beyond that, at the time of the SACC meeting in July 2020, the EPA acknowledged that the committee had reviewed less than 25% of the comments that were submitted by outside scientists (Paustenbach 2020). The Agency indicated that they needed to read all the comments before sharing them and that the COVID-19 pandemic had diminished the number of available staff (Paustenbach 2020).

Risk Evaluation for Asbestos Part 1: Chrysotile Asbestos was promulgated by the EPA on 30 December 2020 (EPA 2020b). It is noteworthy that the EPA’s management must have been aware that in February 2021, the National Academies of Sciences, Engineering, and Medicine (NAS) was to release a report critical of the systematic review process that the EPA had been using since 2016, including the process used for the chrysotile asbestos risk evaluation (National Academies of Sciences 2021). Regarding the NAS report, the EPA stated that:

EPA is not using, and will not again use, the systematic review approach that was reviewed by the Academies. The Application of Systematic Review document released in 2018 represented EPA’s practices at that time. As acknowledged in the 2018 document, the agency’s intent was to update the document based on the experience gained from the first 10 risk evaluations and stakeholder input. To that end, EPA has already begun to develop a TSCA systematic review protocol in collaboration with the agency’s Office of Research and Development to incorporate approaches from the Integrated Risk Information System (IRIS) Program, which the Academies’ report strongly recommends. (EPA 2021)

The EPA did not comment on if it planned to revisit its risk assessments conducted between 2016 and 2020. The Agency additionally stated that Part 2 of the asbestos risk evaluation, which discusses legacy uses of asbestos, was currently “under examination” and that the COUs and fiber types were to be examined in a “scoping” document that is currently under development (EPA 2020b).

What the scientific community knows about chrysotile asbestos

The term “asbestos” encompasses six chemically and physically diverse types of asbestiform mineral fibers that are characterized as either serpentine (chrysotile asbestos) or amphibole (crocidolite, amosite, anthophyllite, actinolite, or tremolite asbestos). Currently, the term “asbestos” only refers to these six fibrous minerals and excludes other fibrous minerals (i.e. erionite, wincherite, or richterite) that may possess an asbestiform habit, but do not exhibit all of the properties of asbestos.

Chrysotile asbestos fibers form large parallel sheets but are curly and pliable due to an incomplete fit between the two layers (Shukla et al. 2003). In contrast to chrysotile, amphiboles are arranged in long, linearly organized chains, forming straight, inflexible, rod-like, and relatively acid-resistant fibers that have more tensile strength than chrysotile (Bernstein and Hoskins 2006). Chrysotile fibers are quickly depleted of critical components of their structure (e.g. magnesium and other cations) within the acidic environment of a macrophage, which facilitates the breakdown of fiber bundles and their subsequent clearance from the lung and pleura following human exposure (Jaurand et al. 1977; Roggli and Brody 1984). Amphibole fibers are far more resistant to this type of leaching and therefore have a much longer residence time in the lung and pleura (Jaurand et al. 1977; Roggli and Brody 1984; Hesterberg et al. 1998; Korchevskiy et al. 2019). The biological half-life of inhaled amphibole fibers is in the range of years to decades, whereas the half-life of chrysotile fibers is only days to weeks (de Klerk et al. 1996; Finkelstein and Dufresne 1999; Bernstein and Hoskins 2006). Because of these and other physicochemical differences, the risk of developing an asbestos-related disease is generally recognized to be far higher following exposure to amphibole asbestos than chrysotile asbestos (McDonald and McDonald 1980; Churg 1998).

The weight of scientific evidence in the last 20 years indicates that chrysotile asbestos, following very high lifetime doses, can cause lung cancer and perhaps mesothelioma. Much of the scientific literature suggests that it takes lifetime doses of 89–168 fibers/cc-years (f/cc-years) for chrysotile to increase the risk of lung cancer and lifetime doses of 208–415 f/cc-years to increase the risk of mesothelioma (Pierce et al. 2016). It has been known for at least 50 years that chrysotile asbestos is significantly less potent than the amphibole fibers for producing both lung cancer and mesothelioma (ACGIH 1978; Berman and Crump 2008a; Garabrant and Pastula 2018; Korchevskiy et al. 2019). For example, Lacquet and van der Linden (1980) reported that among asbestos cement workers exposed to chrysotile asbestos, there were no cases of asbestosis detected at doses less than 100 f/cc-years and 16 cases at doses of 100–800 f/cc-years. There were no cases of mesothelioma reported below 800 f/cc-years (Lacquet and van der Linden 1980). Ferrante et al (2020) reported on asbestosis and mesothelioma among mine Tworkers at a chrysotile asbestos mine in Balangero, Italy. They found that for mine workers with cumulative chrysotile asbestos doses less than 27 f-cc/years, there were no cases of asbestosis reported and one case of pleural cancer, and for workers with exposures between 27 f/cc-years and 345 f/cc-years, there were eight cases of asbestosis and two cases of pleural cancer (Ferrante et al. 2020).

As these studies suggest, even if chrysotile asbestos does have the potency to cause mesothelioma, the evidence suggests that the doses are in the vicinity of those that cause asbestosis (i.e. > 100 f/cc-years), and the fibers probably must be much longer than 5 µm (likely closer to 25–40 µm in length) and have greater than a 3:1 aspect ratio (Berman and Crump 2003, 2008a, 2008b; Pierce et al. 2016).

Due to OSHA regulations, the vast majority of the high-dose occupational exposures to chrysotile asbestos, in the United States ended approximately 40 years ago (OSHA 1994). Therefore, it is worth asking why regulatory action is needed at this time or why information regarding the high doses that workers in the textile mills experienced in the 1940s–late 1970s (Dement et al. 2009; Loomis et al. 2009; Elliott et al. 2012; Loomis et al. 2012, 2019) are being used to characterize risks at lifetime doses of chrysotile asbestos that were around 2–5 f/cc-years, such as for auto mechanics in the 1950s–1985 era (Paustenbach et al. 2003, 2004; Finley et al. 2007).

Why did the EPA tackle this thorny issue?

On March 13, 2017, Arthur Frank, MD, a plaintiff expert in asbestos litigation, and an unidentified plaintiffs’ attorney petitioned EPA Administrator Scott Pruitt to consider asbestos as a toxic material that the EPA should ban (Frank 2017). Dr. Frank submitted to the EPA his standard 216-page opinion letter which has been submitted to courts dozens of time, a document that he admitted he assembled with assistance from a plaintiff attorney, along with his petition (Weglarz et al. 2020). He and another asbestos plaintiffs’ expert, Dr. Barry Castleman, met with an EPA deputy administrator to promote the selection of asbestos for this type of evaluation and to share their views on the risks of asbestos (Weglarz et al. 2020). The Asbestos Disease Awareness Organization (ADAO), a lobbying group that advocates for a “global asbestos ban,” and plaintiff experts Drs. John Dement, Richard Lemen, Jacqueline Moline, and Christine Oliver submitted written statements similar to Dr. Frank’s during the comment period (Dement 2020; Lemen 2020; Moline 2020; Oliver 2020; Reinstein and Sussman 2020; Weglarz et al. 2020). Their comments were grounded in the notion that the EPA should attempt to promulgate stronger regulations for current exposures to chrysotile asbestos (mostly encapsulated), which they claimed still exist.

Beyond opining on the hazards of having asbestos-containing products in commerce, the EPA decided to offer views on the toxicology and cancer potency of pure chrysotile asbestos, as well as resin encapsulated chrysotile asbestos in gaskets and brakes. The only way the EPA had any authority over such products was to claim that they were still routinely sold in the United States which, due to the abundance of asbestos litigation in the United States, is not accurate (Maines 2012). Interestingly, the EPA failed to demonstrate that (1) these products were still entering commerce, except in the chlor-alkali industry where use is tightly regulated, (2) if new gaskets or brakes actually still contain chrysotile asbestos, and (3) that more than a few persons are currently exposed to these allegedly asbestos-containing items. Had the EPA collected more information, the authors believe that the Agency would have found that only a handful, if any, chrysotile asbestos-containing gaskets or brakes are imported into the United States each year. Had the EPA done that, one might expect that the Office of Management and Budget (OMB) would not have supported this initiative due to the small numbers of potentially exposed persons. In the United States, data indicates that only the chlor-alkali industry imports raw chrysotile asbestos (USGS 2019), and even the EPA report indicates that exposures associated with this use are well controlled (EPA 2020b).

Equally interesting is the speed with which this risk evaluation was rushed through the regulatory process. Often, when a large and complicated document similar to the chrysotile asbestos risk evaluation is advocated by the EPA, it takes two to four years and many comment periods before it is approved, if it is approved at all. In this case, the time from the release of the draft to its finalization was nine months, even during the COVID-19 pandemic when few face-to-face meetings were held. A review of the documents obtained through a Freedom of Information Act (FOIA) request by our firm indicates that even Agency employees were surprised that management was going to issue the final risk evaluation by 30 December 2020, because of the many rather complicated scientific matters that the document needed to resolve (FOIA 2021).

It was noted by some commenters to the EPA that the SAB panel not only lacked enough respected and published researchers of chrysotile asbestos, including encapsulated asbestos-containing products, but that the panel was not balanced (Commerce 2020). In recent years, experts who are active in litigation are rarely asked to participate in these panels due to concerns by EPA ethics lawyers or because lawyers for one side or the other complain to EPA that these experts are not impartial. Sometimes, when plaintiff experts serve on SAB panels, the EPA attempts to balance out (person for person) with experts who normally testify for the defense. None of this appears to have happened with this panel. For example, Drs. Anderson, Markowitz, and Kanarek, who served on the SACC panel, have long held views about the toxicity of chrysotile asbestos that were often inconsistent with the weight of scientific evidence. For example, Dr. Anderson’s views regarding “any exposure” to asbestos as a cause of asbestos disease have been rejected by the courts (In re W.R. Grace & Co. 2007), as have Dr. Markowitz’s views regarding encapsulated chrysotile products as a cause of mesothelioma (Matter of New York City Asbestos Litig 2015).

Beyond this, Dr. Loomis was a subcontractor to SRC and contributed to the development of the Inhalation Unit Risk (IUR) in the asbestos risk evaluation (EPA 2020b). Although he is not known to testify on these matters in court, he clearly held strong views about studies that served as the foundation for the IUR, instead of the vehicle mechanic epidemiology studies that were not relied upon by the EPA. Indeed, the Agency discarded these studies, in part, because of comments by the plaintiff experts on the panel who claimed that they were too weak to be relied on. This is an area where experts for the plaintiff and defense have disagreed for almost 15 years, and it would seem that a healthy debate among experts regarding study inclusion should have occurred before such an important decision was reached by the Agency.

The SACC report

The meeting minutes and Final Report for the TSCA SACC meeting on 8–11 June 2020, were released in August 2020 (EPA 2020d). In the authors’ assessment, a key limitation of the EPA meeting was that the questions that the panelists were asked to address, termed “charge questions,” did not focus on the most pertinent aspects of the document. Thus, by asking questions that avoided the thorny topics regarding chrysotile asbestos which were often poorly focused, the EPA failed to obtain relevant topical insight from the advisory panel. Readers only need to look at the wide diversity of the comments from the panel to the EPA to understand the magnitude of the differing views (EPA 2020c). It is unclear from reading the public record that the panel members were asked to review and submit comments on the EPA staff summary of the SACC meeting to ensure that the summary properly reflected the diversity of views of the panel members.

Since the SACC panel was asked to address a wide diversity of poorly focused topics, it is not surprising that the report contains contrasting views and factual errors. While some reviewers wanted the report to address legacy issues of chrysotile use, others thought such a scope was unnecessary, was outside the purview of the EPA, and would be better addressed by OSHA. Others sought the inclusion of amphiboles in the risk evaluation. A single cancer potency factor for all asbestos forms was proposed by some reviewers, despite the weight of scientific evidence regarding relative potency differences for mesothelioma causation across various fiber types (Berman and Crump 2003; Langer 2003; Garabrant and Pastula 2018; Korchevskiy et al. 2019; Bernstein et al. 2020a, 2020b). Some panelists thought the report should address more industries and products than what were ultimately included in the assessment.

The EPA report on the SACC deliberations sent to the Director of the Office of Pollution, Prevention, and Toxics, was 265 pages in length, 284 pages with references, and contained over 230 suggestions for improvement (EPA 2020c). In the authors’ opinion, it is difficult to imagine that the EPA’s Final Risk Evaluation report properly reflects all the wisdom submitted by the multitude of commenters and the comments of the SACC panel members, given that the SACC panel was convened in a rushed manner and the final risk evaluation was prepared in a remarkably short time span. If the Agency had responded to the more than 100 written letters (containing several hundred suggestions) submitted by non-panel members and the 230 comments of the panel members, it likely should have taken more than one year to generate a new version of this risk evaluation. A convening of a follow-up meeting of outside experts was surely necessary and consistent with the EPA’s historical actions in related matters (EPA 2020c).

Where are these chrysotile asbestos-containing products that are supposedly entering the U.S.?

If the EPA had attempted to quantify the number of chrysotile asbestos-containing brakes brought into this country, we believe that they would have found that it is nearly impossible to purchase an asbestos-containing brake originating from Canada, India, Russia, or China. A number of researchers, including the senior author of this paper and Dr. Charlie Blake, both of whom have been studying this topic in depth for more than 20 years, have not been successful in purchasing a single new asbestos-containing brake or gasket after multiple attempts via the internet over the past 10–15 years. On numerous occasions, even when a supplier claimed that their parts contained asbestos, subsequent post purchase testing found that they did not contain any asbestos.

The EPA relied on some comments by the SACC members that claimed to have found advertisements on the internet for brakes that allegedly contained asbestos. For that reason, panel members did not understand why the numerous commenters questioned their availability. However, these panel members seemed unaware that these internet advertisements may be years old and no longer have the brakes available for sale, even if they did contain asbestos at one time. Just this month, our firm attempted to contact suppliers from outside the United States and any who claimed to be selling asbestos-containing brakes on the internet; we received no response from any of them. Before spending millions on this regulatory initiative, it would have been prudent for the EPA to have purchased 20 − 30 sets of allegedly asbestos-containing brakes from these websites and assay them to confirm that, indeed, the underlying assumptions about asbestos exposure were valid. Depending on what was learned, then a careful study of the number of potentially exposed persons should have been conducted. At that point, the EPA and the Office of Management and Budget (OMB) would have been equipped to decide if these allegedly asbestos-containing brakes warranted regulatory action.

Another weakness in the EPA's analysis is that, even if chrysotile-containing brakes could be obtained, there is a lack of significant inhalation exposure associated with changing brakes as reflected in numerous NIOSH studies (Dement 1972; Johnson et al. 1979; Roberts 1980a, 1980b; Roberts and Zumwalde 1982; Sheehy et al. 1989). The preponderance of data cited in the assessment shows that the airborne concentrations to which persons were exposed 40+ years ago, when brakes contained between 30%–80% chrysotile asbestos, was approximately 0.04 f/cc as an eight-hour time-weighted average (TWA) (Paustenbach et al. 2004). The EPA reached a similar conclusion as Paustenbach et al. (2004) with respect to the exposure of mechanics; that is, 0.04 f/cc, as an eight-hour TWA (Weil 1985).

Shortcomings in the EPA’s risk assessment for asbestos

One of the key issues of which readers deserve to be reminded is that the EPA has no authority to regulate workplace exposures – this is the domain of OSHA. For the EPA to weigh in on the topic of “asbestos exposures to workers,” they are obligated to evaluate “future exposures” to products in commerce. Therefore, one of the largest hurdles that the EPA should have faced was to prove that the belief that asbestos-containing gaskets and brakes continue to be easily purchased in the United States was grounded in fact. In the draft risk evaluation, the Agency claimed that approximately 759,900 persons may be exposed annually to chrysotile asbestos-containing brakes. After the EPA was questioned during the review process about the accuracy of this claim, they adjusted the number of potentially exposed persons to 375 (EPA 2020b).

In the EPA’s final risk evaluation, they stated:

According to U.S. Census Bureau data, the average annual value of imports in HTS 6813.20.15 during the period from 2010 to 2019 was $1,949,006.37 According to the web page of a market research group, the demand for aftermarket automotive brake is approximately $4.3 billion per year for all of North America. Based on this data, asbestos brakes may represent approximately 0.05% of aftermarket automotive brakes. Assuming that the number of potentially exposed individuals is equal to the apparent market share of asbestos brakes and applying a 0.05% adjustment factor to the estimates of 749,900 yields a value of 375 for both workers and ONUs. (EPA 2020c, p. 225)

Like everything else in science, the question is more complicated than the EPA risk evaluation conveys. Asbestos in brakes was only a possible health hazard during the era when drum brakes were commonly used, which was approximately from 1940 to the early 1980s for the majority of automobiles in the United States. It is the drum brake that kept the brake dust in the drum and the drum brake sometimes needed to be arced (e.g. grinded) prior to installation. Both disassembly and grinding offered the possibility of exposure to brake dust, which could have contained asbestos, depending on the car manufacturer and year that the part was installed.

However, it should be noted that brake wear debris was at least 99% converted to forsterite due to heating during use (Lynch and Ayer 1968; Hatch 1970; Hickish and Knight 1970; Anderson 1973; Davis and Coniam 1973; Jacko et al. 1973; Rowson 1978; Williams and Muhlbaier 1982; Cha et al. 1983; Sheehy et al. 1989; Spencer 2003; Paustenbach et al. 2004; Boelter et al. 2007; Madl et al. 2009), and forsterite does not have asbestos-like characteristics (Langer 2003).

Importantly, the EPA failed to discuss in the final risk evaluation the vast difference between drum and disk brakes, as well as the differences in plausible exposure to brake wear debris between the two (EPA 2020b). Beginning in the late 1950s, European car manufacturers began switching to the use of disk brakes, which use brake pads, not large drum brakes, which were designed to be open to the ambient air. Similar to drum brakes, the wear debris from disk brakes was also converted to forsterite; however, unlike with drum brakes, the brake wear debris fell onto the roadway and there was no plausible exposure to asbestos for the mechanic. In the United States, a large fraction of the cars that were manufactured had disk brakes beginning in the early- to mid-1970s, which was also when asbestos was being phased out in brakes. Interestingly, the EPA’s risk evaluation acknowledged that there are both types of brakes in the marketplace, but they fail to note that a significant fraction of all European and American cars have had only disk brakes since the late 1990s. Today, some cars still use drum brakes for the rear wheels, however, since asbestos is not being used in these brakes, there is no potential for exposure to mechanics.

In the risk evaluation, the EPA stated that “While EPA has verified that U.S. automotive manufacturers are not installing asbestos brakes on new cars for domestic distribution, EPA has identified a company that is importing asbestos-containing brakes and installing them in their cars in the United States. These cars are exported and not sold domestically” (EPA 2020c, p. 113). What the EPA failed to say was that since these brakes were not manipulated for installation, there was no plausible exposure to asbestos and they failed to be transparent by not identifying the company.

There is evidence from a series of NIOSH studies conducted from 1972 through 1989 that exposure to asbestos was steadily decreasing during vehicle brake servicing operations in the United States during this time period (Dement 1972; Johnson et al. 1979; Roberts 1980a, 1980b; Roberts and Zumwalde 1982; Sheehy et al. 1989). Additionally, OSHA conducted compliance inspections for asbestos from 1984 to 2011, and for the samples collected for the automobile industry (i.e. sales, service, and repair), they showed a consistent decrease in exposure levels throughout the 1980s until 1994, when the last OSHA compliance sample was collected (Cowan et al. 2015). As stated in Cowan et al. “In the 1990s, a total of 172 personal air samples were collected, and of those, 170 samples were non-detectable for asbestos. The two personal samples, 0.03 and 0.056 f/cc, collected in the 1990s were well below the contemporaneous occupational exposure limits.” These facts put into question not only the EPA estimates of exposure for chrysotile asbestos, which were all below the contemporaneous OSHA Permissible Exposure Limits (PEL) for asbestos–0.1 f/cc on an eight-hour time-weighted-average (TWA), but the fact that that there will be minimal exposure to chrysotile asbestos in the future, even if some small fraction of asbestos-containing brakes might be available on the internet.

Among several shortcomings in the final version of the EPA’s risk evaluation, many of the scientists who commented on the draft risk evaluation noted that the EPA did not use most of the relevant published and unpublished literature that was important to conducting a correct analysis (Carbone 2020; Case 2020; Garabrant 2020; Hardin 2020; Mezie et al. 2020; Mossman 2020; Price 2020; Williams 2020). Two authors of this paper (Drs. Paustenbach and Brew) identified nearly 100 papers relevant to the EPA’s chrysotile asbestos risk evaluation that the Agency did not cite and brought this shortcoming to attention in May 2020 (Paustenbach and Brew 2020); which allowed more than enough time for them to be considered in the final risk evaluation. It seems inexplicable that studies which showed the lack of biologic activity of resin-soaked fibers, the conversion of chrysotile to forsterite during braking, the epidemiological studies of mechanics, the published meta-analyses of mechanics, and the studies describing the low potency of chrysotile compared to amphiboles were not mentioned in the EPA's Draft Risk Evaluation for Asbestos. This was despite the claim that the EPA conducted a first-rate systematic review.

In the final response to comments, the Agency stated that the SACC found the EPA analysis was adequate; however, a March 2021 National Academies of Sciences, Engineering, and Medicine (NAS) report that addressed the EPA’s Draft Risk Evaluation for Trichloroethylene (TCE) and the Risk Evaluation for 1-Bromopropane (n-Propyl Bromide) (1-BP) (NAS 2021, p. 2) stated that:

The committee finds that the process outlined in the 2018 guidance document, and as elaborated and applied in the example evaluations, does not meet the criteria of ‘comprehensive, workable, objective, and transparent.’ The committee’s evaluation was made difficult by the incomplete and hard-to-follow documentation of many details of the process—adequacy of documentation is requisite for achieving transparency, objectivity, and replicability. In the committee’s judgment, the specific and general problems in TSCA risk evaluations are partially due to the decision to develop a largely de novo approach, rather than starting with the foundation offered by approaches that were extant in 2016. OPPT was further challenged by the statutory schedule for completing assessments. As a general finding, the committee judged that the systematic reviews within the draft risk evaluations considered did not meet the standards of systematic review methodology. (National Academies of Sciences 2021, p. 7)

The NAS report concluded that:

The committee found that the systematic reviews within the draft risk evaluations considered did not meet the standards of systematic review methodology. The committee applied the critical appraisal tool for systematic review ‘assessment of multiple systematic reviews’ (AMSTAR-2) to the hazard assessment in the draft TCE risk evaluation and found the appraisal process to be unnecessarily complicated due to insufficient and unclear documentation. Despite this barrier to applying the AMSTAR-2 instrument, the committee found that the TCE hazard assessment did not perform positively on the vast majority of AMSTAR-2 questions. Hence, the committee concluded that the hazard assessment within the TSCA TCE risk evaluation was of critically low quality, meaning that the review had ‘more than one critical flaw and should not be relied on to provide an accurate and comprehensive summary of the available studies’”. (Shea et al. 2017, p. 6)

Consequently, the committee suggests that the OPPT team comprehensively reevaluate its approach so as to achieve the state of the practice for systematic review. In the committee’s judgment, the specific and general problems in TSCA risk evaluations are partially due to the decision to develop a largely de novo approach, rather than starting with the foundation offered by approaches that were extant in 2016. OPPT was challenged by the statutory schedule for completing assessments. Nonetheless, looking forward, the committee strongly recommends that OPPT reconsider its overall strategy. (National Academies of Sciences 2021, p. 52–53)

While the NAS report did not explicitly examine the EPA’s chrysotile asbestos risk evaluation, the Agency followed the same methodology used in the TCE and 1-BP risk evaluations when examining asbestos (EPA 2020b). Therefore, it is not surprising that there were significant shortcomings in the asbestos risk evaluation. Many of the issues identified by the NAS regarding the methodology that the EPA used were also identified by many outside reviewers (Carbone 2020; Case 2020; Garabrant 2020; Hardin 2020; Mezie et al. 2020; Mossman 2020; Price 2020; Williams 2020). In spite of the NAS’ sharp criticisms of the methodology used, the EPA’s chrysotile asbestos risk evaluation is being relied on in litigation (Weglarz et al. 2020).

EPA’s chrysotile risk evaluation has already influenced the litigation

Dr. Edwin Holstein submitted an amended opinion letter (on behalf of plaintiffs) in January 2021, less than 30 days after the EPA released its final document, where he stated that his purpose in doing so was to “… summarize a recent reliable and authoritative document published by the Office of Chemical Safety and Pollution Prevention of the United States Environmental Protection Agency dated December, 2020” (Holstein 2021). He concluded his supplemental report by stating “Essentially, the EPA’s conclusions, like the opinions I stated [in his report], is that the Garabrant study fails to establish its claim because the underlying studies on which it relies are methodically incapable of doing so. The same conclusion applies to any other expert who would rely upon the Garabrant et al. publication or its underlying studies to make a similar claim” (Holstein 2021).

Similarly, Dr. Murray Finkelstein submitted an amended opinion letter (on behalf of plaintiffs) in January 2021, that relied upon the “Supplemental File for Epidemiologic Studies of Automotive Mechanics” section of the Risk Evaluation for Asbestos Part 1: Chrysotile” (Finkelstein 2021). Relying upon the EPA’s risk evaluation, Dr. Finkelstein concluded that “The opinions stated by the EPA in their December 2020 Final Risk Evaluation for Chrysotile Asbestos are thus essentially similar to the opinions stated … [in his] report” (Finkelstein 2021).

Comments submitted to EPA during the review process by Drs. Paustenbach and Brew

Two authors of this manuscript (Drs. Dennis Paustenbach and David Brew) identified more than 20 significant shortcomings in the EPA’s Draft Risk Evaluation for Asbestos, which were discussed in 139 pages of comments submitted to the Agency on 26 May 2020 (Paustenbach and Brew 2020). The following are 11 of the most important shortcomings and how the Agency responded to them in the Final Risk Evaluation for Asbestos Part 1: Chrysotile Asbestos (EPA 2020b, 2020c).

1. Comment submitted to the Agency: It is nearly impossible to find an imported asbestos product in the United States, despite the few products advertised on the internet. In our experience, the asbestos products on the internet are either no longer available when you attempt to order them or do not contain asbestos despite being labeled as such (Paustenbach and Brew 2020).

   1. EPA’s Response: “As stated in the draft risk evaluation, EPA reviewed published literature, online databases, government and commercial trade databases. EPA also reviewed company websites of potential manufacturers, importers, distributors, retailers, or other users of asbestos. EPA consulted with USGS and Customs and Border Protection (CBP) staff. EPA reviewed data from CBP’s Automated Commercial Environment (ACE) system, which provided information for 26 companies that reported the import of asbestos-containing products between 2016 and 2018. Also, EPA confirmed that chrysotile brakes are still imported and installed on vehicles that are then exported” (EPA 2020c, p. 116–117).

   2. Note: The authors have asked the EPA for its documentation regarding this statement as it is not consistent with research that we have conducted on this topic. To be specific, in 2005 and 2015, Drs. Dennis Paustenbach and Charlie Blake attempted to purchase allegedly asbestos-containing brakes from the internet, and they were only able to find two suppliers of allegedly asbestos-containing brakes in 2005. However, when these brakes were purchased and tested, they were found to not be asbestos-containing.

      In 2020 and 2021, as a result of the EPA’s risk evaluation, Drs. Paustenbach and Brew searched the internet for brakes that allegedly contained asbestos from Canada, India, Russia, and China (i.e. the countries that the EPA alleged asbestos-containing brakes in the United States were sourced from) (EPA 2020b). If any can be purchased and if we receive them, the authors will have them tested to determine if they in actuality contain asbestos.

2. Comment submitted to the Agency: EPA claims that one million Americans annually, in the upcoming years, might be exposed to new chrysotile-containing products, but we find the basis for this claim to be lacking a factual foundation (Paustenbach and Brew 2020).

   1. EPA’s Response: “EPA has been unable to identify reasonably available information to determine the number of potentially exposed workers and the proportion of vehicles with asbestos-containing brakes or clutches” (EPA 2020d, p. 86). The Agency also stated that “the EPA estimated potentially exposed individuals (both workers and ONUs) by applying this factor (0.05%) to the universe of automotive service technicians and mechanics (749,900), EPA’s estimate of potentially exposed workers is 375” (EPA 2020c, p. 108).

   2. Note: It is the authors’ view that there is no evidence that even dozens of persons could be exposed to chrysotile asbestos-contaminated brakes in the coming years, especially not the hundreds of thousands of automotive workers mentioned in the draft risk evaluation, or even the hundreds stated in the final risk evaluation (EPA 2020b). This is especially true since the Agency’s analysis is based on cascading assumptions rather than facts. As documented in Cowan et al. 2015, which was cited but mischaracterized in the EPA’s final risk evaluation (EPA 2020d, p. 109), OSHA has not detected asbestos in personal breathing zone (PBZ) samples at automotive repair services and parking facilities since 1990. It is the authors’ opinion that, had the EPA conducted a thorough systematic review and correctly characterized the data within the identified studies, the weight of evidence would not have supported the misleadingly high estimates of the current number of likely exposed persons to chrysotile asbestos from performing brake work.

3. Comment submitted to the Agency: Starting about 50 years ago, when OSHA promulgated its earliest asbestos regulations, exposures to asbestos in the workplace began to drop dramatically. Today, virtually no one is exposed to measured quantities of asbestos in new products. This is noted in the EPA’s Draft Risk Evaluation for Asbestos, where identified exposures were generally far below the current OSHA PEL of 0.1 fibers/cc except for a couple of “one-off” situations (Paustenbach and Brew 2020).

   1. EPA’s Response: The agency failed to address this comment.

   2. Note: We do not understand how this comment does not directly bear on the EPA’s risk evaluation for chrysotile asbestos, since we believe the data indicates that there is no appreciable exposure or any measurable exposure to workers in the identified conditions of use in the EPA’s evaluation (EPA 2020b). Normally, an appreciable number of persons are expected to be exposed to unacceptable airborne concentrations of a toxicant for regulatory action to be warranted.

4. Comment submitted to the Agency: About 40–50 years ago, most firms told their purchasing departments not to purchase asbestos-containing products, including gaskets, packing, and brakes. This remains a fact, and it makes sense considering the massive liabilities associated with asbestos litigation (Paustenbach and Brew 2020).

   1. EPA’s Response: “All studies used in the Risk Evaluation, including industry submissions, are evaluated using the same data quality criteria under the TSCA Systematic Review process described in the document, Application of Systematic Review in TSCA Risk Evaluations” (EPA 2020c, p. 118).

   2. Note: The systematic review process that the EPA used was found to suffer from serious flaws based on a report issued by the National Academies of Sciences, Engineering, and Medicine (NAS) in March 2021 (National Academies of Sciences 2021). First, it is logical to infer that the EPA knew when it issued the final risk evaluation of chrysotile asbestos on December 30, 2020, that its process was to be harshly criticized by the NAS. Second, the Agency had to have known, based on all the comments submitted by interested parties, that its systematic review process was unsound long before the NAS report was issued (EPA 2020c).

   As the NAS stated in its report:

   The OPPT approach to systematic review does not adequately meet the state of the practice. The committee suggests that OPPT comprehensively reevaluate its approach to systematic review methods, addressing the comments and recommendations of Chapter 2.

   With regard to hazard assessment for human and ecological receptors, the committee comments that OPPT should step back from the approach that it has taken and consider components of the OHAT, IRIS, and Navigation Guide methods that could be incorporated directly and specifically into hazard assessment.

   The committee finds that OPPT’s use of systematic review for the evidence stream for which it has not been previously adapted to be particularly unsuccessful. Given these novel applications of systematic review, the committee suggests that OPPT elaborate plans for continuing the refinement of methods, ideally, in collaboration with internal and external stakeholders. The committee alsosuggests that OPPT evaluate the ways that existing OHAT, IRIS, and Navigation Guide methods could be modified for the other evidence streams. In addition, OPPT should use existing guidance within the agency such as the Guidelines for Human Exposure Assessment, the Guidelines for Ecological Risk Assessment, and the operating procedures for the use of the ECOTOXicology knowledgebase, as following existing guidelines would improve transparency of the assessments.

   The committee recommends that a handbook for TSCA review and evidence integration methodology be put together that details the steps in the process. Throughout this report, the committee points to problems of documentation. The committee believes that the effort of developing and publicly vetting a handbook will pay off in the long run by making the process more straightforward, transparent, and easier to follow. (National Academies of Sciences 2021, p. 56)

   In response to the NAS’ report, the EPA issued a press release that stated:

   EPA is not using, and will not again use, the systematic review approach that was reviewed by the Academies. The Application of Systematic Review document released in 2018 represented EPA’s practices at that time. As acknowledged in the 2018 document, the agency’s intent was to update the document based on the experience gained from the first 10 risk evaluations and stakeholder input. To that end, EPA has already begun to develop a TSCA systematic review protocol in collaboration with the agency’s Office of Research and Development to incorporate approaches from the Integrated Risk Information System (IRIS) Program, which the Academies’ report strongly recommends. (EPA 2021)

   However, the Agency did not comment on if they planned to revisit the risk assessments conducted between 2016 and 2020 (EPA 2021).

5. Comment submitted to the Agency: The EPA based its cancer potency factor for chrysotile on the erroneous assumption that a textile mill in the 1950s–1970s (Dement et al. 2009, 2011; Elliott et al. 2012) could be a surrogate for describing the potency of fibers from gaskets and brakes. For the past 20 years, research has indicated that chrysotile fibers need to be closer to 15–40 µm in length to pose a significant health hazard (Berman and Crump 2003, p. 2003, 2008a, p. 2008). Such fiber lengths were generally found only in textile mills, and they are virtually absent from brake dust or any dust associated with brake or gasket manipulation today (Paustenbach and Brew 2020).

   1. EPA’s Response: “EPA did not find sufficient data that showed chrysotile asbestos effects experienced by textile workers would be different from the effects of exposure to chrysotile asbestos in other asbestos products. There was no quantitative exposure data of adequate quality to derive cancer potency values for chrysotile asbestos based on studies of automobile mechanics; they were all unsuitable for use in quantitative risk assessment” (EPA 2020c, p. 137).

   2. Note: The Agency seems to be unaware of the significant quantity of information about the exposure of mechanics to chrysotile which were collected by the United States government in the 1970s and 1980s (Dement 1972; Johnson et al. 1979; Roberts 1980a, 1980b; Roberts and Zumwalde 1982; Sheehy et al. 1989). The Agency seemed not to understand that the North and South Carolina textile mill cohorts that they relied on in the final risk evaluation (Hein et al. 2007; Dement et al. 2009; Loomis et al. 2009; Elliott et al. 2012; Loomis et al. 2019) were not relevant to characterize the exposure to products containing encapsulated chrysotile asbestos, and that most of the cohorts that they cited also had some exposure to amphiboles.

      For example, for the North Carolina textile cohorts, approximately 96% of the person-time accrued was from workers at plants three and four (Garabrant 2020), which also had documented use of amosite and crocidolite asbestos (UNARCO 1954; UNACRO 2012). Due to these exposures to amphibole asbestos, the Dement et al. (2009), Loomis et al. (2009), Elliott et al. (2012) and Loomis et al. (2019) articles are of limited usefulness when examining the lung cancer and mesothelioma risks from exposure to chrysotile asbestos alone.

6. Comment submitted to the Agency: The EPA assumed in its risk assessment that textile workers in four mills in North Carolina and the GARCO mill in South Carolina were exposed only to chrysotile, which is untrue (Paustenbach and Brew 2020).

   1. EPA’s Response: “SACC did consider all these limitations raised in public comments to the SACC and this and other SACC recommendations establish that NC and SC datasets have the best exposure data and are appropriate to use for chrysotile asbestos IUR derivation. The SACC concurred that the NC and SC cohorts (Burdorf and Heederik 2011; Lenters et al. 2012) have the best exposure data and that this was a good reason to focus on these cohorts’ (see 4.13 for more detail on this SACC recommendation). The submitted comments provided undisputed information that UNARCO owned the NC Marshville plant for a period of time. They also provided evidence that some other plants from the several that UNARCO owned did produce amosite products in those other plants. EPA did not find evidence in the comments or court depositions that the Marshville plant was using amosite in its productions and was ever anything other than a textile plant that used commercial chrysotile asbestos (EPA 2020c, p. 140).

   2. Note: In our view, the EPA’s comments are not responsive to the many comments submitted to the EPA by experts on this topic. Also, the authors’ understanding of the SACC panel is not consistent with the EPA’s response. Indeed, the panel had members who made it clear that the North Carolina and South Carolina textile mills should not be used in the risk evaluation (Paustenbach 2021). In North Carolina, there is no available record that any type of amphibole asbestos was used at Plant 1 (Davidson, North Carolina) or Plant 2 (Charlotte, North Carolina), while only chrysotile asbestos was used (Garabrant 2020). There is evidence that amosite asbestos was carded, twisted, and woven between 1963 and 1976 at Plant 3 (Charlotte, North Carolina) (Loomis et al. 2009) and amosite and crocidolite asbestos was used at Plant 4 (Marshville, North Carolina) while UNARCO owned the mill from 1947 to 1963 (UNARCO 1954; UNACRO 2012). This was confirmed by researchers who assayed worker's lungs from these mills and found that amphiboles were present (Roggli et al. 1997; Pavlisko et al. 2020). If the SACC panel had the opportunity to thoroughly review the comments submitted by Dr. Garabrant, and other reviewers, they would have been aware that there were numerous documents available from the Asbestos Claims Research Facility (ACRF) in Aurora, Colorado, that conclusively demonstrated that the Marshville, North Carolina, textile mill worked with, and had in inventory, crocidolite, amosite, and chrysotile asbestos (Garabrant 2020. p. 5–6). Furthermore, in comments submitted to the SACC panel, Dr. Garabrant identified 12 other important points regarding the North Carolina textile plants:

   1. “All 8 cases of pleural cancer and mesothelioma arose in plants 3 and 4, where there were mixed exposures to chrysotile and commercial amphiboles (amosite and crocidolite). Because of mixed exposures, plants 3 and 4 cannot contribute to the DRE for chrysotile asbestos.

   2. “Plant 1, where only chrysotile was used, contributed little to the person-time experience of the North Carolina asbestos textile cohort. Among subjects employed 30 days or more, only 3,933 person years occurred in plant 1 out of 100,742 person years (3.9%) in plants 1, 3, and 4 combined, and only 3.7% of the person years (3,933/105,077) among subjects employed for 30 days or more in plants 1, 2, 3, and 4 combined. Because roughly 96% of the person-time experience of the North Carolina textile cohort was derived from plants 3 and 4, the papers by Loomis, Elliott, and Dement (Dement 2009; Loomis 2009; Elliott 2012; Loomis 2012, 2019) are essentially uninformative on the risks of lung cancer and mesothelioma among textile workers exposed to only chrysotile.

   3. “In spite of numerous publications from the NC cohort (Dement 2009; Loomis 2009; Elliott 2012; Loomis 2012, 2019), the authors never mentioned how many of the 8 cases of pleural cancer and mesothelioma arose in workers who had less than 30 days of employment, or 30 or more days of employment. This omission demands an explanation. Workers with less than 30 days of employment contributed 42% of the total person years of observation [(181,640–105,077)/181,640 = 0.42]. If cases arose in workers with < 30 days of employment this would substantially affect the interpretation of the findings from these papers. Among short-term workers (<30 days of employment), the overwhelming majority of their careers must have been spent in jobs elsewhere, about which nothing (including asbestos exposure) was known.

   4. “In calculating the rates (in Table 1, above) of lung cancer, pleural cancer, and mesothelioma, no adjustment for differences in age or calendar period among plants was possible (only Loomis and Elliott have the data that would allow such adjustments). In calculating pleural cancer and mesothelioma rates, I have assumed (in a manner that provides the high-end estimate of mesothelioma risks) that no cases occurred among workers with < 30 days of employment. If this assumption is incorrect, and cases arose among workers with < 30 days of employment, the person years in each plant would be increased on average by 73% (181,640/105,077 = 1.73), and the pleural cancer and mesothelioma rates would be reduced inversely. This does not alter the fundamental conclusion that all the cases arose where commercial amphiboles were used, and no cases arose in plants 1 and 2, where only chrysotile was used.

   5. “In plant 2, where only chrysotile was used to “produce friction products and other finished goods from purchased asbestos yarn and tape” (Loomis 2009), there were 0 deaths due to pleural cancer and mesothelioma in 8,780 person-years of observation. This observation is directly relevant to EPA’s DRE for AAABL, particularly regarding whether the chrysotile in asbestos-containing brakes is associated with increased risk of mesothelioma. Although the cohort in plant 2 was small, the available evidence does not support any such conclusion.

   6. “In Plant 3, where chrysotile and amosite were used, there were 3 deaths due to pleural cancer and mesothelioma in 80,428 person-years of observation (rate = 3.7/100,000).

   7. “In Plant 4, where chrysotile, crocidolite, and amosite were used, there were 5 deaths due to pleural cancer and mesothelioma in 16,381 person-years of observation (rate = 30.5/100,000).

   8. “If the calculations are restricted to mesothelioma (ignoring pleural cancer deaths), the overall pattern of rates across plants is not appreciably different. Plant 4 has at least an order of magnitude higher risk of mesothelioma than any of the other plants.

   9. “If the pleural cancers and mesotheliomas occurred among workers with >1 day of employment in the textile plants (rather than among workers with > 30 days of employment as I have assumed) and the person years were increased by an average factor of 1.73, this would not appreciably change the overall pattern of rates across plants.

   10. “The mesothelioma plus pleural cancer rate in Plant 4 (Marshville) is approximately an order of magnitude higher than in the Plant 3 (Charlotte) where amosite and chrysotile were known to have been used. It is also an order of magnitude higher than in the Charleston, SC textile plant studied by Loomis (2009) and Elliott (2012). Even after adjusting for cumulative exposure (f/ml-yrs), the mesothelioma plus pleural cancer rate in Plant 4 (Marshville) is about 3.5 times higher than the rate in Plant 3 (Charlotte). This is consistent with a conclusion that there is a difference in the potency of the fibers between plant 4 and all other North Carolina plants (1, 2, and 3).

   11. “It is inappropriate that Loomis (2009) and Elliott (2012) combined results across the North Carolina and South Carolina cohorts without addressing the heterogeneity of pleural cancer and mesothelioma risks across plants.

   12. “It is unexplained how EPA regarded the North Carolina plant exposure data to be “of higher quality than those utilized in other studies of occupational cohorts exposed to chrysotile” (page 141 lines 5200–5201), when Loomis and Dement were completely unaware of the evidence of amphibole asbestos use in plant 4 (Marshville), failed to recognize the heterogeneity in pleural cancer and mesothelioma risks among the North Carolina Plants, and failed to recognize that the mesothelioma risk in the Marshville plant was incompatible with all other evidence from occupational cohorts exposed to only chrysotile” (Garabrant 2020, p. 10–11).

       It is our view that the SACC’s belief that the North Carolina and South Carolina textile mill cohorts “… have the best exposure data and are appropriate to use for chrysotile asbestos IUR derivation” is flawed and invalidates the EPA’s entire risk evaluation. As discussed above, members of this cohort were known to have been exposed to amphiboles (Pooley and Mitha 1986; Sebastien et al. 1989), and this was confirmed by lung fiber analysis (Sebastien et al. 1989; Green et al. 1997; Roggli et al. 1997; Case et al. 2000; Pavlisko et al. 2020).

       Also, the fibers that were used in these textile mills were exceedingly long (i.e. >20 μm), and therefore, were vastly different in length and aspect ratio from the short, encapsulated asbestos fibers that were used in brakes and gaskets. Historically, friction products used grade 7 chrysotile asbestos as a filler or binding material (Cossette and Delvaux 1979; Mann 1983); while the textile industries used longer asbestos fibers (Grades 1–3) that could be spun and woven into finished products (Mann 1983; Pigg 1994). It is virtually unique to the textile industry that chrysotile fiber lengths are routinely longer than 25 μm.

7. Comment submitted to the Agency: For decades, the airborne concentrations of chrysotile associated with handling brakes or gasket have been very low or nonexistent, and we now know that the fibers likely lacked significant biologic activity since they were either degraded during use (Paustenbach et al. 2004) or were soaked in phenolic resin, which appears to eliminate the toxicity (Bernstein et al. 2003, 2018; 2020a; 2020b).

   1. EPA’s Response: The Agency failed to address the comment directly; however, they did state: “In response to the comment on longer vs shorter fibers, the Bayesian analysis by Hamra et al. (2017) shows that shorter fibers also have cancer risk.” As the SACC noted in other comments (See Section 4.12) “… most asbestos fibers in textile plants were, in fact, short fibers, i.e. less than 5 microns in length” (EPA 2020c, p. 137).

   2. Note: We do not understand how the EPA believed that the exposure of textile workers in the 1940–1970s, during which these populations were exposed to fibers which were often 20–50 µm in length (and may have been contaminated with amphiboles), is in any way similar to exposure to brake wear debris or short fiber chrysotile soaked in resin. The EPA failed to discuss the work of Bernstein et al. (2018, 2020a, 2020b), which showed that fibers soaked in resin seem to lack biological activity. Beyond ignoring those studies and the meta-analyses of the epidemiology studies involving mechanics (Wong 2001; Goodman et al. 2004; Garabrant et al. 2016), the EPA also failed to cite many historically important studies that show that short fiber chrysotile lacks biologic activity (Davis and Coniam 1973; Stanton 1973; Stanton et al. 1977; Platek et al. 1985; Berman and Crump 2003; Berman and Crump 2008a; Berman and Crump 2008b; Lippmann 2014). The Agency relied on the Hamra et al. (2017) paper to respond to public comments regarding the lack of toxicity of shorter fibers versus longer asbestos fibers (EPA 2020c, p. 137). The premise of the Hamra article was that unregulated asbestos fibers, described as those not meeting the definition of an OSHA fiber - i.e. a diameter greater than 0.25 µm, a length greater than 5 μm, and having an aspect ratio of 3:1 or greater - posed a lung cancer hazard in the North Carolina and South Carolina textile mill cohorts described in Elliott et al. (2012). The author’s claimed that “A persistent concern is that the specified dimensional criteria may not be sufficient for protecting the health of exposed workers because they are not based solely on health concerns” (Hamra et al. 2017).

      Several shortcomings of the Hamra et al. (2017) article appeared to have been overlooked by the Agency in its chrysotile asbestos risk evaluation. In our opinion, the primary scientific issue with this article is that the Bayesian model was developed from two textile cohorts that were known to be exposed to both amphiboles and chrysotile asbestos (see comments #5 and #6 and Garabrant 2020), yet the authors stated that “… we note that the vast majority of asbestos used in these plants was raw chrysotile. Small amounts of crocidolite yarn were used in the SC plant (1950 − 1975) and some amosite was used in one of the North Carolina plants (1963–1976). Thus, results are with regard to chrysotile asbestos” (Hamra et al. 2017), without describing the methodology for distinguishing between exposures to differing fiber types.

      Additionally, the authors stated that they relied on three studies to describe the pooled North and South Carolina cohorts (Loomis et al. 2009; Loomis et al. 2010; Elliott et al. 2012); however, as stated in the Loomis et al. (2009) cohort “Information on smoking was available for <15% of the cohort … Adjustment for smoking was not feasible because of the very limited data available for the cohort.” Loomis et al. (2010) stated that “The epidemiologic data have several limitations, which have been discussed previously [10]. These include: inadequate information on smoking …” Elliott et al. (2012) does not describe any smoking details of these cohorts. This is despite the fact that cigarette smoking remains the leading cause of preventable chronic disease, disability, and death in the United States, accounting for more than 480,000 deaths every year, or approximately one in five deaths (HSS 2020).

      Approximately 90% of lung cancers in the United States are caused by cigarette smoking (Alberg and Samet 2003). Interaction between cumulative asbestos exposure and smoking and their combined effect on lung cancer risk has been studied since the 1960s (Selikof et al. 1968; Saracci 1977; Hammond et al. 1979; Selikoff et al. 1979; Selikoff et al. 1980; Berry et al. 1985; Erren et al. 1999; Liddell and Armstrong 2002; Yano et al. 2010). Markowitz et al. (2013) conducted a follow-up study on a subset of the cohort of insulation workers previously studied by Selikoff et al. (1979) and Hammond et al. (1979). The authors compared lung cancer mortality rates among those that underwent clinical examination between 1981 and 1983 (n = 2,377) to lung cancer mortality rates from “blue collar” workers studied in a separate, but contemporaneous, cohort with no reported asbestos exposure (Markowitz et al. 2013). According to Markowitz et al. (2013), asbestos exposure, in the absence of asbestosis, resulted in a rate ratio of 3.6 (95% CI: 1.7–7.6); smoking in the absence of asbestos exposure resulted in a lung cancer rate ratio of 10.3 (95% CI: 8.8–12.2). However, most importantly, in the absence of asbestosis, asbestos exposure in combination with smoking resulted in a lung cancer rate ratio of 14.4 (95% CI: 10.7–19.4). This would have severe implications for any study that attempted to correlate asbestos exposures with lung cancer mortality, without correcting for smoking rates.

      Therefore, by not having an adequate understanding of and not correcting for smoking in the Hamra et al. (2017) article, in our opinion, this article suffers from severe methodological flaws that should have prevented the EPA from relying on it to draw conclusions regarding the alleged carcinogenic effects of short asbestos fibers.

      Additionally, the EPA failed to acknowledge the findings of articles going back decades that demonstrate that short fiber chrysotile asbestos lacks biologic potency to cause mesothelioma or lung cancer. It has been known since the early 1970s that long, thin fibers present a greater risk of disease than shorter fibers (Stanton 1973; Stanton et al. 1977; Lippmann 1988; ATSDR 2002; Eastern Research Group 2003a, 2003b).

      The ‘Stanton hypothesis’ states that the optimum fiber morphology for inducing intrapleural tumors by injection and implantation in animal models is a diameter of ≤ 0.25 µm and length of > 8 µm (Stanton et al. 1981). Upon inhalation, asbestos fibers interact with macrophages and pulmonary epithelial cells (Mossman and Churg 1998); with both short and long asbestos fibers stimulating the release of O⁻₂ and H₂O₂ from alveolar macrophages and neutrophils (Kamp and Weitzman 1999). Macrophages are able to phagocytize short asbestos fibers efficiently, while longer asbestos fibers results in frustrated phagocytosis, which leads to a continuous overproduction of O⁻₂ (Pott 1978; Mossman and Churg 1998; Kamp and Weitzman 1999). Retention of the long asbestos fibers in the lung is believed to lead to the development of disease (ATSDR 2002; Eastern Research Group 2003b; Berman and Crump 2008a; Berman and Crump 2008b).

      In 2002, The Agency for Toxic Substances and Disease Registry sponsored the convening of an expert panel to address the influence of fiber length on asbestos-related health effects (ATSDR 2002). These panelists were asked to comment on the physiological fate, as well as non-carcinogenic and carcinogenic health effects associated specifically with asbestos and synthetic vitreous fibers less than 5 μm in length. Overall, the ATSDR-sponsored “panelists agreed that there is a strong weight of evidence that asbestos and short vitreous fibers shorter than 5 μm are unlikely to cause cancer in humans” based on “findings from epidemiologic studies, laboratory animal studies, and in vitro genotoxicity studies, combined with the lung’s ability to clear short fibers” (Eastern Research Group 2003b).

      As noted previously, the EPA contracted the preparation of a technical support document to establish a protocol to assess asbestos-related disease versus fiber length (Berman and Crump 2003). Based on modeling results presented in the technical support document, Berman and Crump concluded that the best estimate of risk for mesothelioma for fibers between 5 and 10 μm in length was one three-hundredth of the risk assigned to fibers longer than 10 μm. The authors explained that “results from [their] review of the supporting literature suggest that the optimum cutoff for increased potency occurs at a length that is closer to 20 μm [rather] than 10 μm.” Further, Berman and Crump (2003) reported that the best estimate of the potency of fibers shorter than 5 μm for mesothelioma is zero. A second expert panel convened by the U.S. EPA agreed that the risk associated with fibers less than 5 μm in length is “very low and could be zero” (Eastern Research Group 2003b).

      In conclusion, there are decades of asbestos, smoking, and lung cancer research that put into question the claims by Hamra et al. (2017) regarding the evidence that unregulated asbestos fibers have a meaningful effect on lung cancer mortality.

8. Comment submitted to the Agency: The EPA failed to recognize or consider 16 epidemiology studies and three meta-analyses of auto mechanics. They happen to be an ideal cohort for evaluating the risks of exposure to brake dust and gaskets. If they had considered the studies, they would have found, based on a weight of the evidence approach, that no increased risk of mesothelioma is present in these workers (Garabrant et al. 2016; Paustenbach and Brew 2020).

   1. EPA’s Response: “Following the SACC recommendation, EPA evaluated the quality of the studies of auto-mechanics for both lung cancer and mesothelioma and found deficiencies in exposure assessment and other aspects of the design of these studies (see Supplemental File to this Summary of Comments and Disposition). Therefore, those studies (and published meta-analyses based on these studies) are not appropriate to be included in Part 1 of the Risk Evaluation for Asbestos” (EPA 2020c, p. 153–154).

   2. Note: The majority of comments indicating deficiencies in the exposure data or epidemiologic studies regarding vehicle mechanics appear to have been offered by persons who did not understand the quality of those studies. These data were sufficiently abundant and important to this EPA analysis that a separate committee of qualified experts should have been formed to evaluate them. Up until this risk evaluation, the EPA had embraced the government’s papers on this topic (Dement 1972; Johnson et al. 1979; Roberts 1980b; Roberts 1980a; Roberts and Zumwalde 1982; Sheehy et al. 1989) and they were not particularly concerned about the toxicology of brake wear debris, as it was converted to forsterite (Jacko et al. 1973; Rowson 1978; Williams and Muhlbaier 1982).

      American cars converted at least 99% of the asbestos in the brake pads and linings to forsterite or related compounds, due to the intensity of the heat and pressure from braking (Lynch and Ayer 1968; Hatch 1970; Hickish and Knight 1970; Anderson 1973; Davis and Coniam 1973; Rowson 1978; Williams and Muhlbaier 1982; Cha et al. 1983; Sheehy et al. 1989; Spencer 2003; Paustenbach et al. 2004; Boelter et al. 2007; Madl et al. 2009).

      Dr. Langer of NYU conducted a chemical analysis of brake wear debris. He noted that "Using heating studies and milling as an approximation of thermal and mechanical shear stress that chrysotile is subjected to on a brake lining, biological blunting is shown to begin much earlier than the olivine transformation process. Minimal degradation of the chrysotile surface structure imparts a disproportionately great effect on its biological activity" (Langer 2003). Dr. Langer’s hypotheses were confirmed in two studies conducted in 2014 and 2015 (Bernstein 2014; Bernstein et al. 2015). It was reported that "These results provide the support that brake-dust derived from chrysotile containing brake drums would not initiate a pathological response in the lung or the pleural cavity following short term inhalation" (Bernstein et al. 2015). Thus, the available information is that even the few fibers that were sometimes found in the breathing zone in various studies of mechanics in the 1960s and 1970s, likely had little or no carcinogenic potency.

9. Comment submitted to the Agency: It seemed unnecessary and inappropriate to have EPA once again conduct this risk assessment since they have commissioned two evaluations previously, and there have been no less than a dozen published papers that support the conclusions of those panels regarding the virtual lack of potency of chrysotile to produce mesothelioma. EPA should not use the word asbestos as if it were a family of similar fibers, as they are not remotely similar in potency (Berman and Crump 2003, p. 2003; Langer 2003; Korchevskiy et al. 2019; Bernstein et al. 2020a, 2020b). Chrysotile, amosite, crocidolite, tremolite, and anthophyllite all have different chemical structures, all have different iron content, and all have different solubilities in lung tissue fluids (Korchevskiy et al. 2019; Paustenbach and Brew 2020).

   1. EPA’s Response: “EPA clarified that Libby Amphibole Asbestos IUR and 1986 IUR are not directly comparable to each other or with IUR for commercial chrysotile derived in this Risk Evaluation. Footnote 4 states ‘It is important to mention that the methodology involved in risk characterization has evolved over time and the existing EPA IURs for other asbestos types (EPA (2014, 1986) estimated risks of cancer mortality and did not account for the risk of other cancers, and the 1986 IUR did not adjust for mesothelioma under-ascertainment

      “EPA evaluated quality of cohorts and used only cohorts that are exposed to commercial chrysotile asbestos. It should be noted that SACC supported use of these cohorts and did not suggest use of amphibole-exposed cohorts in this assessment” (EPA 2020c, p. 165).

   2. Note: The Agency did not directly respond to our comment.

10. Comment submitted to the Agency: The serpentine fibers have a biological half-life in the lung that is dramatically shorter than the amphibole fibers (Bernstein 2005). Amphibole fibers, depending on length, can have a half-life in the lung as long as one decade (Bernstein et al. 2011). Many believe that this is yet another explanation for the negligible cancer potency of chrysotile in both animals and humans (Paustenbach and Brew 2020).

    1. EPA’s Response: The agency failed to address this comment. Instead, they stated that “the Bayesian analysis by Hamra et al. (2017) shows that shorter fibers also have cancer risk.” The agency also clarified that “for the purposes of the asbestos risk evaluation, EPA adopted the TSCA Title II definition of asbestos which is the “asbestiform varieties of six fiber types–chrysotile (serpentine), crocidolite (riebeckite), amosite (cummingtonite-grunerite), anthophyllite, tremolite or actinolite. EPA is only evaluating the asbestiform varieties of these mineral fibers” (EPA 2020c, p. 246).

    2. Note: The data in the toxicology literature regarding the lack of biologic activity of short fiber chrysotile has been known for decades (Stanton 1973; Platek et al. 1985; Lippmann 1988; Eastern Research Group 2003b; Berman and Crump 2008a; Berman and Crump 2008b; Pierce et al. 2016; Barlow et al. 2017). In 1974, Wagner and colleagues published their classic study exposing rats to amosite, anthophyllite, crocidolite, Canadian chrysotile, and Rhodesian chrysotile for one day, three months, six months, 12 months, and 24 months (Wagner et al. 1974). While they demonstrated that all of these fibers could produce asbestosis and lung tumors in rats, the exposure concentrations were approximately 100,000 times higher than the current OSHA PEL of 0.1 f/cc on an eight-hour time-weighted-average (TWA) basis.

       Further, the chemical differences among the fiber types and their half-life in the lung are markedly different (Garabrant and Pastula 2018; Korchevskiy et al. 2019). In our view, this did not receive adequate consideration by the EPA. The EPA repeatedly relied on the paper by Hamra et al (2017), which contained a large amount of interesting information, the vast majority of which is irrelevant to characterizing the hazard to those exposed to encapsulated chrysotile fibers like those associated with brake or gasket work of the 1940–1980s. In our view, and as discussed above, the conclusions in the Hamra et al. (2017) paper are not relevant to short chrysotile fibers filled with phenolic resin.

11. Comment submitted to the Agency: The agency relied on the linearized multistage model to derive the IUR for chrysotile. The approach involved fitting a risk model to available exposure-response data from epidemiological studies of workers exposed to asbestos. The Agency seems to have performed its dose extrapolation using the method of Maximum Likelihood Estimation (MLE), assuming that the observed number of cases in a group is a random variable described by the Poisson distribution. Given the likely mode of action of chrysotile asbestos and the fact that there is mesothelioma in background populations, it seems inappropriate to have used the LMS model, which assumes no risk at zero dose, to identify the cancer slope factor (Paustenbach and Brew 2020).

    Beyond this, the Agency stated in its draft risk evaluation that “Once the cancer-specific lifetime unit risks are obtained, the two are then combined. It is important to note that this estimate of overall potency describes the risk of mortality from cancer at either of the considered sites and is not just the risk of an individual developing both cancers concurrently. Because each of the unit risks is itself an upper bound estimate, summing such upper bound estimates across mesothelioma and lung cancer mortality is likely to overpredict the upper bound on combined risk. Therefore, following the recommendations of the Guidelines for Carcinogen Risk Assessment (EPA 2005), a statistically appropriate upper bound on combined risk was derived as described below. To the best of our knowledge, no regulatory agency or scientific body has ever proposed a cancer potency factor for asbestos by combining different cancer types (EPA 2020b, p. 151).

    1. EPA’s Response: “Following SACC recommendation, EPA used linear model fit from North Carolina cohort in its chrysotile asbestos IUR derivation based on NC cohort. EPA provided all modeling results (linear and exponential fit for lung cancer data) for comparison in Table 3–12 illustrating changes in PODs and risks. EPA was not able to obtain data for pooled analysis of mesothelioma data. EPA does not agree with commenters proposing relative risk model for mesothelioma, because the SACC recommended that EPA use absolute risk model and EPA followed EPA cancer guidelines in its low-dose extrapolation, as indicated by expanded MOA discussion (SACC recommendation 94), which indicated appropriateness of linear extrapolation” (EPA 2020c, p. 175).

    2. Note: Based on the available information, the authors believe it is inaccurate to say that “Following SACC recommendation, EPA used linear model fit from North Carolina cohort in its chrysotile asbestos IUR derivation based on NC cohort,” as based on the senior author’s personal communication with a member of the SACC panel. There were several panel members who felt strongly that the LMS model was inappropriate to use with chrysotile and this cohort (Paustenbach 2021).

       We urge the Agency to be cautious about applying any variation of the linearized multistage (LMS) model to estimate the risks of low dose exposure for short chrysotile fibers. Currently, the specific mechanism of carcinogenesis from exposure to chrysotile asbestos is not firmly established, and an association between exposure and a recurrent signature for pleural mesothelioma has not been identified (Bueno et al. 2016; Hylebos et al. 2016; Casalone et al. 2018; Hmeljak et al. 2018; Lorenzini 2021). There is also evidence that both pleural mesothelioma and lung cancer can undergo epigenetic modifications (i.e. those that are not related directly to nucleotide sequences (DNA) of the genome) (Toyooka et al. 2008; Christensen et al. 2009; Brzeziańska et al. 2013; Reid 2015). Further, there is ample evidence that the excess lung cancer from exposure to asbestos occurs with fibers longer than 20 μm (Lippmann 2014).

       Based on the weight of scientific evidence, there is a strong belief that the mechanism of action for asbestos involves chronic inflammation, combined with cellular toxicity and repair that leads to the generation of reactive oxygen species and DNA damage, rather than direct interaction with DNA (Huang et al. 2011a; Barlow et al. 2013). This implies that there exists a threshold and that cellular damage only occurs at exposure concentrations high enough to overwhelm cellular defense mechanisms. The linear model that the Agency used in its risk evaluation is primarily intended for genotoxic carcinogens and not those that act through a mechanism similar to asbestos (Huang et al. 2011b; Barlow et al. 2013). If the Agency had followed its own guidelines for assessing carcinogens, it is the author’s view that the Agency would not have relied on a linear model, but would have utilized a threshold model instead (EPA 2005).

Select comments made by scientists questioning the agency’s views - affiliations in parentheses

1. Comment by Drs. Mezie, Moolgavkar, and Mowat (Exponent, Inc): “The major omission of relevant scientific data is the failure to consider the numerous epidemiologic studies of occupations in which low-level exposure to chrysotile asbestos occurs. We note that risk assessments for low-level exposures in specific occupations based on mathematical extrapolations from studies at high levels of exposure are of dubious interest when there are multiple, well-conducted epidemiologic studies of the occupations in question. The EPA has ignored the numerous epidemiologic studies reporting no increased risk of either lung cancer or mesothelioma in occupations exposed to low levels of chrysotile asbestos” (Mezie et al. 2020).

   1. EPA’s Response: “Following the SACC recommendation, EPA evaluated the quality of the studies of auto-mechanics for both lung cancer and mesothelioma and found deficiencies in exposure assessment and other aspects of the design of these studies” (EPA 2020c, p. 153).

   2. Note: The Agency and SACC were incorrect regarding the perceived inadequacies of these epidemiology studies. The EPA seem to have relied on the comments of two panel members, who have served as plaintiff experts on this matter, in concluding the exposure and epidemiology studies were fatally flawed. There are at least 20 epidemiology studies that uniformly report no increased risk of developing mesothelioma among vehicle mechanics and others who historically worked with chrysotile asbestos containing brakes (McDonald and McDonald 1980; Teta et al. 1983; Spirtas et al. 1985; Järvholm and Brisman 1988; Hansen 1989; Gustavsson et al. 1990; Spirtas et al. 1994; Woitowitz and Rodelsperger 1994; Teschke et al. 1997; Agudo et al. 2000; Hansen et al. 2003; Hessel et al. 2004; Rolland et al. 2005; Welch et al. 2005; Rake et al. 2009; Aguilar‐Madrid et al. 2010; Merlo et al. 2010; Rolland et al. 2010; Borre and Deboosere 2015; Tomasallo et al. 2018). Additionally, when smoking was considered and controlled among mechanics, there are 12 case-control studies (Williams et al. 1977; Lerchen et al. 1987; Benhamou et al. 1988; Vineis et al. 1988; Morabia et al. 1992; Swanson et al. 1993; Matos et al. 2000; Richiardi et al. 2004; MacArthur et al. 2009; Consonni et al. 2010; Corbin et al. 2011; Guida et al. 2011) and two cohort studies (Hrubec et al. 1992; Hrubec et al. 1995) that report no increased risk of lung cancer among these workers, compared to the general population. A careful review of these data sets shows that they were far superior to using the textile worker data (Dement et al. 1994; Hein et al. 2007; Loomis et al. 2009; Elliott et al. 2012; Loomis et al. 2019) to predict low-level exposure to encapsulated chrysotile that the EPA alleged was occurring in 2020. This also reflects the inadequacy of the EPA’s systematic review process that the NAS report identified (National Academies of Sciences 2021).

2. Comment by Dr. Goodman, Mr. Dodge, Dr. Bailey, Dr. Prueitt, Mr. Peterson, Dr. Beck, and Ms. Engel (Gradient): “US EPA derived an IUR for chrysotile asbestos by applying a linear-no-threshold (LNT) model from the point of departure (1% benchmark risk) from two chrysotile occupational epidemiology studies (i.e. Elliott et al. 2012; Loomis et al. 2019) and exposure-response models with the best fit. The LNT model likely considerably overestimates the cancer potency of chrysotile asbestos. Regarding DNA-reactive substances, many recent scientific papers indicate that the small increase in DNA damage that might occur from very low exposures, in addition to the already high levels of endogenous DNA damage, should not overwhelm DNA repair capacities (Cardarelli and Ulsh 2018). This indicates a practical threshold exists even for DNA-reactive substances.

   For substances that do not directly interact with DNA, an LNT model is even less biologically plausible. Although the specific mechanism of chrysotile-asbestos-induced carcinogenesis is not established, the evidence is generally supportive of a mode of action involving chronic inflammation and cellular toxicity and repair that leads to the generation of reactive oxygen species and DNA damage, rather than direct interaction with DNA (Huang et al. 2011a). This threshold mechanism can only occur at exposure concentrations high enough to overwhelm cellular defense mechanisms.

   “Pierce et al. (2016) derived ‘best estimate’ chrysotile no-observable-adverse-effect levels (NOAELs) of 208 to 415 f/cc-years for mesothelioma and 89 to 168 f/cc-years for lung cancer that can be applied as thresholds in chrysotile cancer risk evaluations. US EPA did not discuss Pierce et al. (2016) or acknowledge a possible threshold mode of action for chrysotile” (Goodman et al. 2020)

   1. EPA’s Response: “EPA does not agree with commenters proposing relative risk model for mesothelioma, because the SACC recommended that EPA use the absolute risk model and EPA followed EPA cancer guidelines in its low-dose extrapolation, as indicated by expanded MOA discussion (SACC recommendation 94), which indicated appropriateness of linear extrapolation” (EPA 2020c, p. 175).

   2. Note: This topic is sufficiently important that a new and smaller panel of experts should have been convened to evaluate it. As noted previously, it is not scientifically adequate for the EPA to repeatedly say, “per the SACC recommendation,” we adopted their suggestion. First, that is not a proper characterization of what the SACC said. Second, the SACC did not issue their own report; we have only the EPA summary. There were many different views expressed by the panel, so it is not accurate to say that the panel reached a conclusion on this matter (Paustenbach 2021). Third, it would have been more appropriate, given the likely mechanism of action for chrysotile asbestos (Huang et al. 2011a; Barlow et al. 2013), to have presented several different results of various models, including those with a threshold, and have the recommended smaller panel of experts discuss the merits of each model.

   As NIOSH stated in its recent policy on chemical carcinogens:

   “In general, previous guidelines were premised on an assumption that it was not scientifically possible to predict safe levels of exposure to carcinogens; therefore, risks at low doses (i.e. below the observable range) have been estimated using linear extrapolation. However, there is increasing scientific evidence that the response at low doses of some carcinogens may be nonlinear and may include a threshold. In these situations, simple linear extrapolation at low doses may result in an overestimation of cancer risk” (NIOSH 2017).

   This is particularly true in scenarios where occupational exposures occurred at low levels, such as those involving mechanics working with chrysotile asbestos containing products, and where there is a large body of epidemiologic data that provides essential information relevant to risk.

3. Comment by American Chemical Council: “EPA Needs to Clarify Certain Aspects of Its Data Screening and Data Evaluation Decisions in the Asbestos Risk Evaluation to Demonstrate Use of Best Available Science. Several times in the Systematic Review (section 1.5 of the asbestos draft risk evaluation), EPA distinguishes certain aspects of this review from those conducted in previous TSCA draft risk evaluations. While for the most part, EPA justifies these differences by references to the still-evolving nature of its approach to a systematic review (as noted in its Application of Systematic Review in TSCA Risk Evaluations document), some of these differences need further clarification to demonstrate that EPA is basing its decisions on best available science” (Walls 2020).

   1. EPA’s Response: No acknowledgment or response from the EPA on this comment.

   2. Note: While the Agency did not address this comment, the NAS report did identify this as a major shortcoming of the EPA’s systematic review process (National Academies of Sciences 2021).

4. Comment by Mr. Beckett, Mr. Miller, Dr. Pierce, and Dr. Finley (Cardno ChemRisk): “EPA considered the central tendency, short-term exposure value (0.006 f/cc) to be representative of the “central tendency” eight hour TWA exposure value for occupational brake repair work (EPA 2020, p. 95). To justify this approach, EPA indicated that this ‘median short term exposure level could persist for an entire workday’ based on the assumption that brake repair mechanics ‘may conduct 40 brake repair jobs per week,’ or an average of 8 brake jobs per workday” (Beckett et al. 2020).

   1. EPA’s Response: “EPA has updated Section 2.3.1.7.5 to clarify how central tendency exposure estimates were derived for the aftermarket automotive parts condition of use. That text reads: ‘The central tendency short-term TWA exposure value for workers is based on the seven studies found to include relevant measurements. For each study, EPA identified the central tendency short-term exposure, which was either reported by the authors or inferred from the range of data points, and the value in Table 2 15. (0.006 fibers/cc) is the median of those central tendencies from those seven studies. Thus, three of the studies reported central tendency concentrations lower than 0.006 fibers/cc, one reported a central tendency concentration of 0.006 fibers/cc, and the other three studies reported higher exposure concentrations” (EPA 2020c, p. 89–90).

   2. Note: The EPA’s approach is inaccurate. Extrapolating an eight-hour time-weighted average (TWA) exposure from short-term exposure values is imprecise and unnecessary since numerous eight-hour TWA exposure values for occupational brake work have been published in the literature (Hickish and Knight 1970; Dement 1972; Johnson et al. 1979; Roberts 1980b; Roberts 1980a; Rödelsperger et al. 1986; Moore 1988; Sheehy et al. 1989; Plato et al. 1995; Yeung et al. 1999; Weir et al. 2001; Blake et al. 2003; Paustenbach et al. 2003; Madl et al. 2008; Richter et al. 2009). The seven studies (Cooper et al. 1987; Godbey et al. 1987; Sheehy et al. 1987a; Sheehy et al. 1987b; Cooper et al. 1988; Blake et al. 2003; Madl et al. 2008) that the EPA noted to have “relevant measurements” were, in fact, entirely unrepresentative of the body of literature on the topic, as they excluded a multitude of relevant data sets.

      It appears that the Agency misunderstood how long it takes to perform a brake job and the percentage of the week a mechanic normally does them. For a typical mechanic, we believe that a high-end number for brake jobs per week is 4–5 and that 40 brake jobs per week is implausible (discussed in Finley et al. 2007).

5. Comment by Mr. Beckett, Mr. Miller, Dr. Pierce, and Dr. Finley (Cardno ChemRisk): “In its draft risk evaluation, EPA applied a multiplier of 1.39 to adjust the risk of deaths from mesothelioma based on the assumption that the deaths would be undercounted (EPA 2020, p. 150–151). The 1.39 value comes from Table 3 of Kopylev et al. (2011), which reports the mean, median, 5th, and 95th percentile of the ‘Distribution of Ratio of Posteriors with Adjustment and without Adjustment for Under ascertainment of Mesothelioma’ when the specific misdiagnoses of mesothelioma approach is used.

   “While the mean ratio of the posterior distribution is 1.39, it was not statistically significantly different than 1. In fact, it was estimated with 90% probability that the true ratio is between 0.80 and 2.17. Thus, the possibility of over-counting mesothelioma cases also exists. Furthermore, Loomis et al. (2019) attempted to reduce undercounting of mesothelioma by examining death certificate data for any mention of mesothelioma and for codes often used before there was a specific ICD code for mesothelioma. Therefore, without specific evidence of undercounting of mesothelioma among the textile studies, EPA has no basis to assume this occurred” (Beckett et al. 2020).

   1. EPA’s Response: “According to other SACC recommendations, EPA changed its approach to use North Carolina data in chrysotile IUR calculations and used post-1999 data in the analysis. For reasons articulated by this SACC recommendation, mean under-ascertainment factor (1.39) is more appropriate for this situation, and EPA continued to use this factor.

      As to suggestion by a public commenter of a relative risk model, EPA has used an absolute risk model for mesothelioma since 1986, and SACC found this model appropriate to use (see comment 4.31).

      “EPA clarifies that the perception of lack of significance is a misunderstanding that this is an adjustment factor and testing of a statistical hypothesis is thus, not appropriate” (EPA 2020c, p. 187).

   2. Note: The EPA’s note is mainly non-responsive to this comment and references “other SACC recommendations,” which remain unspecified. The Agency does not justify the approach of using a 1.39 multiplier for mesothelioma deaths when the literature suggests there is also a possibility of over counting. Further, as noted in the original comment, even the authors from which the EPA derived its value of 1.39, found that the mean distribution ratio was not statistically different from 1, with the estimated 90% probability that the true ratio was between 0.80 and 2.17 (Kopylev et al. 2011).

      Additionally, as Dr. Garabrant stated in his comments,

      EPA’s method for correcting for underascertainment of mesothelioma, in which it multiplied the mesothelioma mortality in the North Carolina and South Carolina asbestos textile cohorts by a factor of 1.39, was derived from a study of asbestos workers from Libby, Montana (Kopylev 2011) who were exposed to amphibole asbestos. The method for deriving the correction factor of 1.39 was highly dependent on assumptions about the number of missed mesothelioma cases, particularly peritoneal mesotheliomas. Insofar as peritoneal mesotheliomas are rarely, if ever, seen among chrysotile-exposed cohorts, this correction factor is inappropriate for use in chrysotile asbestos risk assessment. EPA provided no justification for using it in this DRE. Kopylev provided another, smaller correction factor based solely on underestimation of mesothelioma cases on death certificates, and not on the number of missed peritoneal mesothelioma cases. EPA chose not to use that smaller correction factor, even though, according to its own reliance materials (Kopylev 2011), it would have been more appropriate.

      More importantly, Kopylev cautioned “The mean ratio of 1.39 implies that the estimated risk of exposure to the Libby amphibole asbestos would be approximately 1.4 times larger if all decedents for whom medical information and pathology samples would have supported a diagnosis of mesothelioma had been identified.” (Kopylev 2011) Herein lies a grave error in EPA’s methods: it invoked an implausible correction factor to deal with possible mesothelioma underascertainment in the North Carolina and South Carolina textile cohorts, while it ignored the studies of the Quebec chrysotile asbestos miners (Liddell 1997) and Balangero chrysotile asbestos miners (Ferrante 2020; Pira, 2017) in which all decedents for whom medical information and pathology samples supported a diagnosis of mesothelioma were identified, and in which underascertainment of mesothelioma cases was not an issue. (I have discussed EPA’s inappropriate exclusion of the Quebec and Balangero chrysotile asbestos miners and refuted EPA’s decision above.) Thus, EPA’s correction for mesothelioma underascertainment was not only unnecessary, it was erroneous.

      EPA’s correction for potential underascertainment problems could have been obviated if EPA had simply used a relative risk model for mesothelioma, as it did for lung cancer. No correction factor for mesothelioma underascertainment would have then been needed (Garabrant 2020).

6. Comment by Drs. Roggli and Sporn (Duke University Medical Center): “The EPA draft assessment notes that the estimated risks of death are conservative because they have not considered non-neoplastic asbestos-related diseases in their analysis. It is well recognized that there is a threshold of exposure to asbestos associated with the disease asbestosis. There is no convincing evidence that the low levels of exposure considered in the current document cause or contribute to any level of asbestosis, let alone the levels that result in fatal pulmonary fibrosis. With respect to benign asbestos-related pleural diseases, only diffuse visceral pleural fibrosis has been associated with mortality, and then only rarely. There is no evidence that the levels of exposure considered in this document cause or contribute to fatal diffuse visceral pleural fibrosis” (Roggli and Sporn 2020)

   1. EPA’s Response: The EPA was not directly responsive to this comment

7. Comment by The Chlorine Institute: “EPA notes in lines 2319–2323 of the Asbestos Notice an estimate of potentially exposed janitorial staff. Only separate, non-process area breakrooms or office space, which are designated clean areas, are cleaned by non-operator janitorial staff, at most once a day. The 100 ONUs stated in the draft Risk Evaluation overestimates janitorial staff for the 10 plants (Asbestos Notice p. 66; line 2320–2323). EPA’s assumption that ONUs are exposed to asbestos for eight-hours/day, 250 days/year is inaccurate for the reasons discussed here and greatly overestimates any potential time spent around a restricted area. All janitorial-like housekeeping activities in asbestos process areas within the restricted areas are handled by the workers who perform the asbestos-handling tasks with appropriate PPE in compliance with site procedures” (Brooks 2020)

   1. EPA’s Response: “For the Chlor-alkali ONU exposure estimate in Part 1 of the Risk Evaluation for Asbestos, EPA made several assumptions including the assumption that workers not directly handling asbestos may access the work area and ONU exposure concentration is comparable to the area monitoring result. It is possible that workers not related to Chlor-Alkali production may be trained on asbestos so that they could access the work area for tasks unrelated to Chlor-Alkali” (EPA 2020c, p. 68–69).

   2. Note: The authors recommend that the EPA should review its own data regarding the alleged exposures of workers in the chlor-alkali industry. As stated in the Risk Evaluation for Asbestos Part 1: Chrysotile Asbestos, workers in this industry wear respiratory protection (EPA 2020b, p. 26; 79) and based on the data presented the measured exposures were de minimus compared to the OSHA PEL (EPA 2020b, p. 83–85).

8. Comment by Dr. Garabrant (University of Michigan, Ann Arbor): “It is inappropriate to base the DRE for Aftermarket Auto Brakes and Clutches on high exposure to long fiber asbestos used in textile manufacturing while ignoring risks to workers engaged in brake and clutch repair who have low exposures to short fiber asbestos.

   “EPA bases its DRE on the inhalation unit risk (IUR) derived from studies at high exposure levels that are extrapolated to low exposure levels using theoretical models. Moreover, asbestos textile manufacturing, such as in the North Carolina and South Carolina textile plants, preferentially used long asbestos fibers, whereas friction product manufacturing (brakes and clutches) preferentially used short asbestos fibers. It is inappropriate to base the AAABL DRE on risks derived from long fiber exposures rather than short fiber exposures, and on mixed chrysotile/amphibole exposures rather than on only chrysotile exposures” (Garabrant 2020)

   1. EPA’s Response: “EPA agrees with SACC that N.C. and S.C. cohorts have best exposure data, and it is appropriate to focus on these cohorts. Following SACC and public comments recommendations, EPA evaluated the quality of studies of auto-mechanics suggested by SACC and public comments for both mesothelioma and lung cancer and found deficiencies in exposure assessment and other aspects of the design of these studies” (EPA 2020c; p. 149).

   2. Note: As mentioned previously, the SACC members that commented on this matter are entirely misinformed about the quality of the exposure and epidemiology studies of the vehicle mechanics. In the authors’ opinion, for the exposure assessment studies to be inadequate, one would have to discount the National Institute for Occupational Safety and Health (NIOSH) studies that were conducted from 1972 through 1989 that investigated vehicle brake servicing operations in the United States (Dement 1972; Johnson et al. 1979; Roberts 1980b; Roberts 1980a; Roberts and Zumwalde 1982; Sheehy et al. 1989), which was a relevant time period for when chrysotile asbestos exposures could plausibly have occurred.

      The consistency among the independent 8-h TWA airborne asbestos concentrations among these studies indicates substantial similarities in brake repair operations and the asbestos content in brake linings. However, it is unclear if the aforementioned NIOSH studies considered all of the activities typically associated with brake repair work (e.g. handling the brake box; various machining activities; sweeping/cleaning the work area after brake work is completed). As such, several researchers have since assessed the potential for airborne asbestos exposure in brake service workers, resulting from specific tasks related to brake maintenance and the handling of friction products (Weir et al. 2001; Blake et al. 2003; Paustenbach et al. 2003; Madl et al. 2008; Richter et al. 2009).

      A major component of any possible exposures of brake mechanics is acknowledging that at least 99% of the chrysotile asbestos in brake pads and linings is converted to forsterite or related compounds during braking (Lynch and Ayer 1968; Hatch 1970; Hickish and Knight 1970; Anderson et al. 1973; Davis and Coniam 1973; Jacko et al. 1973; Rowson 1978; Williams and Muhlbaier 1982; Cha et al. 1983; Sheehy et al. 1989; Spencer 2003; Boelter et al. 2007; Madl et al. 2009), and studies have shown that the chrysotile fibers found in the air from brake wear debris lack the biologic activity of asbestos (Langer 2003; Bernstein 2014; Bernstein et al. 2015; Boyles et al. 2019). Further, research has shown that chrysotile asbestos, the only type of asbestos used in brakes, is rapidly cleared from tissue compared to other forms of asbestos, suggesting that it can be more effectively removed by the body’s defense mechanisms (Bernstein 2005; Bernstein and Hoskins 2006; Bernstein et al. 2010; Bernstein et al. 2011; Mossman et al. 2011; Bernstein et al. 2013; Bernstein 2014; Bernstein et al. 2015; Bernstein et al. 2018; Bernstein et al. 2020a; Bernstein et al. 2020b; Bernstein et al. 2021). Additionally, multi-dose 90-day exposure rat inhalation studies have demonstrated fundamental differences in the lack of biopersistence and effective clearance and dissolution of brake dusts and chrysotile asbestos from the lungs, as opposed to more biologically persistent amosite and crocidolite asbestos (Bernstein et al. 2020a; Bernstein et al. 2020b; Bernstein et al. 2021). Thus, the available information indicates that even the small quantity of chrysotile fibers sometimes found in worker breathing zones in studies of mechanics in the 1970s and 1980s had little or no carcinogenic potency.

      In terms of the epidemiology studies, the SACC panel decided to narrowly focus on epidemiologic studies that provided quantitative exposure estimates that would enable the calculation of an inhalation unit risk (IUR), rather than focus on the weight of evidence from epidemiologic studies that show that there is no increased risk of mesothelioma (McDonald and McDonald 1980; Teta et al. 1983; Spirtas et al. 1985; Järvholm and Brisman 1988; Hansen 1989; Gustavsson et al. 1990; Spirtas et al. 1994; Woitowitz and Rodelsperger 1994; Teschke et al. 1997; Agudo et al. 2000; Hansen et al. 2003; Hessel et al. 2004; Rolland et al. 2005; Welch et al. 2005; Rake et al. 2009; Aguilar‐Madrid et al. 2010; Merlo et al. 2010; Rolland et al. 2010; Borre and Deboosere 2015; Tomasallo et al. 2018; Thomsen et al. 2021) or for lung cancer, when it was controlled for smoking (Williams et al. 1977; Lerchen et al. 1987; Benhamou et al. 1988; Vineis et al. 1988; Hrubec et al. 1992; Morabia et al. 1992; Swanson et al. 1993; Hrubec et al. 1995; Matos et al. 2000; Richiardi et al. 2004; MacArthur et al. 2009; Consonni et al. 2010; Corbin et al. 2011; Guida et al. 2011), in vehicle mechanics.

      Thomsen et al. (2021) recently evaluated the long-term risk of mesothelioma, lung cancer, asbestosis, and other lung diseases and mortality in Danish motor vehicle mechanics who had been working since 1970. This study included 138,559 Danish vehicle mechanics (median age 24 years; median follow-up 20 years (maximum 45 years)), and found that compared to other Danish workers, vehicle mechanics had a lower risk of morbidity due to mesothelioma/pleural cancer (n = 47 cases) (age-adjusted and calendar-year- adjusted HR = 0.74 (95% CI 0.55 to 0.99)), a slightly increased risk of lung cancer (HR = 1.09 (95% CI 1.03 to 1.14)), increased risk of asbestosis (HR = 1.50 (95% CI 1.10 to 2.03)), and a chronic obstructive pulmonary disease risk close to unity (HR = 1.02 (95% CI 0.99 to 1.05)) (Thomsen et al. 2021). The mesothelioma results of this study were no surprise, as it supported decades of epidemiological studies of vehicle mechanics which found similar results (Wong 2001; Goodman et al. 2004; Garabrant et al. 2016).

      Regarding the findings of increased incidence of asbestosis among mechanics, the authors were aware that this was illogical given the magnitude of exposure and the fact that it has never been previously reported. The authors offered several thoughts about why this anomaly occurred including (1) misdiagnosis of idiopathic fibrosis as asbestosis, (2) diagnostic bias by occupational physicans beginning in the mid-2000s, (3) sufficient cumulative exposures during the 1950s–1970s to develop pathological asbestosis, and/or (4) confounding factor of smoking being associated with an increased risk of interstitial fibrosis (Thomsen et al. 2021). Also, it would require a special study to determine if the diagnostic criteria used in Denmark had changed over the past several decades, during which a tendency to label a questionable case as “possibly asbestosis” could have varied.

      Per personal communications between Dr. Paustenbach and a member of the SACC panel, the SACC did not speak with one voice on nearly any topic that they evaluated (Paustenbach 2021). As such, it is inappropriate for the Agency to continually state that they were relying on SACC comments because the panelists either did not know the material, did not have adequate time to carefully consider the public comments, or they failed to reach an agreement on certain aspects of the risk evaluations.

      In the authors’ opinions, to give credibility to the EPA’s chrysotile risk evaluation, the Agency needs to address the shortcomings identified in the 2021 NAS report (National Academies of Sciences 2021), and special panels should be convened which contain published experts in the various specialties relevant to this risk evaluation. When experts involved in litigation join these panels, they should make their published views well known, as well as their potential conflicts of interest. The current EPA guidelines suggest that the three plaintiff experts should not have served on the SACC panel due to conflicts of interest.

9. Comment by Dr. Garabrant (University of Michigan, Ann Arbor): “EPA assumed the background risk of malignant mesothelioma is zero. There is reliable evidence that there are other causes of malignant mesothelioma besides asbestos, and that malignant mesothelioma occurs in the absence of known causes (i.e. there is a non-zero background rate)” (Garabrant 2020)

   1. EPA’s Response: “EPA followed SACC recommendation by accounting for variable exposure for North Carolina cohort. The new results are presented in Tables 3 and 4. Use of Peto model (without background exposure term) for mesothelioma modeling is a longstanding Agency framework (U.S. EPA 1986, 2014). In addition, use of variable exposure obviates EPA modeling of NC mesothelioma data as variable exposure modeling are presented in Loomis et al. (2019). SACC agreed with the EPA that the use of extra risk models is recommended by various Agency guidance (e.g. BMDS Technical Guidance)” (EPA 2020c, p. 181).

   2. Note: The authors did not find the EPA's comment to be responsive to Dr. Garabrant’s comment regarding the background incidence rate of mesothelioma. The EPA’s assumption that there is a zero risk of mesothelioma when there is zero asbestos exposure is outdated with respect to the current scientific understanding of cancer development and ignores evidence that all cancers, including mesothelioma, can and do develop spontaneously and without exposure to environmental carcinogens. There is convincing evidence that cancer incidence is strongly correlated with the number of cellular divisions that a tissue has undergone (Tomasetti and Vogelstein 2015; Tomasetti et al. 2017). Similar to any other cancer, mesothelioma occurs due to the accumulation of mutations during cellular division. As this process occurs spontaneously and continuously throughout a person’s life, it is logical to assume that there is a background rate of mesothelioma in the population. Price and Ware (2009) analyzed trends in the National Cancer Institute’s (NCI) SEER data, and estimated that for men in the United States, approximately 70–75% of their mesotheliomas were attributable to asbestos exposure, while for women, it was only 3–10% (Price and Ware 2009). Moolgavkar et al. (2009) estimated that the background lifetime risk of pleural and peritoneal mesothelioma, after adjusting for other-cause mortality, is approximately three and one per 10,000 population, respectively (Moolgavkar et al. 2009). Moreover, as stated in Carbone et al. (2019), “As for any other cancer, irrespective of exposure and of inherited mutations, some mesotheliomas may occur because of the inevitable accumulation of spontaneous mutations, as observed in mesotheliomas developing in lions, cats, horses, dogs, birds, clams, sharks, etc.” (Carbone et al. 2019).

      Additionally, the weight of evidence demonstrates that there are other causes and risk factors for mesothelioma, including age (La Vecchia et al. 2000; Clements et al. 2007; Cree et al. 2009; Moolgavkar et al. 2009; Girardi et al. 2014; Schonfeld et al. 2014; Mensi et al. 2016), erionite exposure (Maltoni et al. 1982; Poole et al. 1983; Johnson et al. 1984; Wagner et al. 1985; Carthew et al. 1992; Fraire et al. 1997; Baumann and Carbone 2016; Attanoos et al. 2018; Korchevskiy et al. 2019), genetic predisposition (Wassermann et al. 1980; Brenner et al. 1981; Fraire et al. 1988; Huncharek 2002; Kayatta et al. 2013; Kanbay et al. 2014; Attanoos et al. 2018), and therapeutic ionizing radiation (van Kaick et al. 1999; Travis et al. 2003; Travis et al. 2005; Tward et al. 2006; Deutsch et al. 2007; Hodgson et al. 2007; Teta et al. 2007; De Bruin et al. 2009; Goodman et al. 2009; Berrington de Gonzalez et al. 2010; Farioli et al. 2013; Gibb et al. 2013; Schaapveld et al. 2015; Schubauer-Berigan et al. 2015; Chang et al. 2017).

10. Comment by Dr. Garabrant (University of Michigan, Ann Arbor): “EPA adopted inappropriate methods for inferring values when measured fiber counts are below the limit of detection and the shape of the fiber count distribution is unknown. In these instances, EPA’s inferred values systematically overestimate fiber counts” (Garabrant 2020).

    1. EPA’s Response: The Agency failed to address this comment.

Select comments made by other individuals and organizations - affiliations in parentheses

1. Comment by Dr. Lemen (Emory University): “There is no safe way to use asbestos, all fibers cause both nonmalignant disease and cancer and there is no known safe exposure below which there is no risk for all” (Lemen 2020).

   1. EPA’s Response: “Under TSCA section 6(b)(4)(D), EPA is required to determine whether a chemical substance presents unreasonable risks for the conditions of use without consideration of costs or other non-risk factors. As such, TSCA does not require EPA to make a finding of whether there is ‘no safe level’ of asbestos. As stated in Section 5.1 of Part 1 of the Asbestos Risk Evaluation, for the purposes of making risk determinations for the conditions of use, EPA uses 1×10⁻⁶ as the benchmark for consumers (e.g, do-it-yourself mechanics) and bystanders.

      “In addition, consistent with the 2017 NIOSH guidance, EPA uses 1×10⁻⁴ as the benchmark for individuals in industrial and commercial work environments subject to Occupational Safety and Health Act (OSHA) requirements. It is important to note that 1×10⁻⁴ is not a bright line, and EPA used discretion to make risk determinations based on other considerations including other risk factors, such as severity of endpoint, reversibility of effect, or exposure-related considerations” (EPA 2020c, p. 170–171).

   2. Note: The comment that there is no safe level of exposure to asbestos is lacking in scientific merit. It is the author’s opinion that there is a safe dose for chrysotile asbestos, as there is for every other chemical known to mankind. This has been a bedrock principal of toxicology since before the 16th century, and, despite claims made over the years, asbestos is not a magical fiber that breaks the known principles of toxicology. The claim that there is no known safe dose is based on policy, not science. The exposure data and the epidemiology studies of populations exposed to chrysotile asbestos (see references above) suggests that even if chrysotile asbestos does have the capacity to cause mesothelioma, the required doses are likely > 100 f/cc-years, and the fibers probably must be much longer than 5 µm (likely closer to 25–40 µm in length) and have greater than a 3:1 aspect ratio (Berman and Crump 2003; Berman and Crump 2008a; Berman and Crump 2008b; Pierce et al. 2016). Due to OSHA regulations, the vast majority of the high-dose occupational exposures to chrysotile asbestos, in the United States, ended approximately 50 years ago. It is the author’s opinions that in the modern era, due to OSHA regulations (OSHA, 1994), that there is no evidence that even dozens of persons are currently exposed to chrysotile asbestos in new products, much less the hundreds that the Agency stated in the final risk evaluation (EPA 2020b)

2. Comment by Dr. Dement (Duke University): “As part of the risk assessment process EPA took into consideration possible reductions in inhaled dose due to use of respirators. As a general rule of industrial hygiene practice, respirators are considered to be minimally effect, to be used as the last line of worker protection, and generally to be used as a short-term measure pending institution of more effective engineering and/or administrative controls” (Dement 2020).

   1. EPA’s Response: “EPA agrees that there are challenges associated with use of PPE; they are described in Section 5.1. By providing risk estimates that account for use of PPE, EPA is not recommending or requiring use of PPE. Rather, these risk estimates are part of EPA’s approach for developing exposure assessments for workers that relies on the reasonably available information (including information from the industry) and expert judgment. When appropriate, EPA will develop exposure scenarios both with and without engineering controls and/or PPE that may be applicable to particular worker tasks on a case-specific basis for a given chemical. EPA did assess the risk to workers in the absence of PPE, and those risks are presented in Section 4 Risk Characterization under Table 4–55, Summary of Risk Estimates for Inhalation Exposures to Workers and ONUs by COU” (EPA 2020c, p. 236–284).

   2. Note: For short-term use respirators have been found to be highly effective, until effective engineering or regulatory controls can be implemented. The National Institute for Occupational Safety and Health (NIOSH) studies that were conducted from 1972 through 1989 (Dement 1972; Johnson et al. 1979; Roberts 1980b; Roberts 1980a; Roberts and Zumwalde 1982; Sheehy et al. 1989) that investigated vehicle brake servicing operations in the United States, and OSHA compliance studies that were conducted from 1984 through 2011 (reviewed in Cowan et al. 2015), showed that asbestos exposures to mechanics were generally well controlled and under the contemporaneous PEL on a TWA basis. This demonstrates that the engineering and regulatory controls that have been in place for the past 40 years at these facilities were generally protective of worker health.

3. Comment by Dr. Moline (Northwell Health): “The omission of talc-based consumer and industrial applications from the draft evaluation is a significant gap because of the likelihood that some grades of talc used in these applications are contaminated by asbestos, putting consumers and workers at risk. EPA should include these conditions of use in its risk evaluation. To assure that their risks are assessed fully, it should also expand the evaluation to include other known health impacts of asbestos like ovarian cancer and address all fiber types, not simply chrysotile” (Moline 2020).

   1. EPA’s Response: “EPA is aware that talc originating from certain sources and used in consumer and industrial applications may contain asbestos. However, considering the significant scope of evaluating the potential risks posed to individuals from exposure to talc, it would be more appropriate to evaluate talc (and any known or reasonably foreseen co-located asbestos therein) in a separate and subsequent risk evaluation focused on talc” (EPA 2020c, p. 25–26).

   2. Note: We agree with the EPA that the topic of talc is outside the scope of the EPA’s chrysotile asbestos risk evaluation. Regarding the EPA’s response that this should be addressed in a future risk evaluation, the authors believe that it should be conducted after the Agency addresses the methodological shortcomings identified by the NAS (National Academies of Sciences 2021).

4. Comment by Ms. Linda Reinstein and Mr. Robert Sussman (Asbestos Disease Awareness Organization - ADAO): “The draft evaluation does not address the risks of legacy asbestos products despite a US court of appeals decision requiring EPA to evaluate these risks. Legacy asbestos is pervasive in US buildings and is a significant contributor to ongoing asbestos-related death and disease. A comprehensive assessment of the risks of asbestos to the US population is impossible without accounting for exposure to legacy asbestos” (Reinstein and Sussman 2020).

   1. EPA’s Response: “As a result of the court decision in Safer Chemicals Healthy Families v. EPA, 943 F.3d 397 (9th Cir. 2019), the Agency will evaluate legacy asbestos uses and associated disposals of those uses in Part 2 of the Risk Evaluation for Asbestos. For Part 2, EPA plans to issue the following documents: a draft supplemental scope document for public comment, a final supplemental scope document, a draft risk evaluation document for public comment and a final risk evaluation document. Prolonging finalization of the risk evaluation for chrysotile asbestos (Part 1 of the Risk Evaluation for Asbestos), by expanding the document to also evaluate legacy uses (where only use and associated disposal is present) would significantly delay needed risk management to address COUs where unreasonable risk is present for chrysotile asbestos” (EPA 2020c. p. 22).

   2. Note: Legacy uses of pure chrysotile asbestos in brakes and gaskets, while of greater magnitude in the decades prior to the 1970s, presented relatively limited, if any, risk to human health based on the epidemiology data for mesothelioma (McDonald and McDonald 1980; Teta et al. 1983; Spirtas et al. 1985; Järvholm and Brisman 1988; Hansen 1989; Gustavsson et al. 1990; Spirtas et al. 1994; Woitowitz and Rodelsperger 1994; Teschke et al. 1997; Agudo et al. 2000; Hansen et al. 2003; Hessel et al. 2004; Rolland et al. 2005; Welch et al. 2005; Rake et al. 2009; Aguilar‐Madrid et al. 2010; Merlo et al. 2010; Rolland et al. 2010; Borre and Deboosere 2015; Tomasallo et al. 2018; Thomsen et al. 2021) and lung cancer (when smoking was considered and controlled among mechanics) (Williams et al. 1977; Lerchen et al. 1987; Benhamou et al. 1988; Vineis et al. 1988; Hrubec et al. 1992; Morabia et al. 1992; Swanson et al. 1993; Hrubec et al. 1995; Matos et al. 2000; Richiardi et al. 2004; MacArthur et al. 2009; Consonni et al. 2010; Corbin et al. 2011; Guida et al. 2011).

      Regarding the EPA’s response addressing “legacy asbestos uses and associated disposal of those uses in Part 2 of the Risk Evaluation for Asbestos” (EPA 2020b), the authors believe that it should be conducted after the Agency addresses the methodological shortcomings identified by the NAS (National Academies of Sciences 2021).

The EPA’s use of the linear no threshold (LNT) model of carcinogenesis

The linear no threshold (LNT) model of carcinogenesis contends that every exposure to a carcinogen contributes to a cumulative, linear increase in the risk of developing cancer. That is, the LNT model holds that cancer risk is directly proportional to exposure dose, even at extremely low levels of exposure.

The LNT model is built upon five guiding scientific principles and assumptions, i.e. (1) One or two changes in cells can transform and lead to cancer, (2) no population-based threshold due to heterogeneity, (3) a transformed cell is irreversibility propagated, (4) If the mode of action (MOA) involves mutation no threshold assumed; if no MOA is identified and cancer occurs, mutation is assumed, and (5) a single or few molecules can cause mutation (Golden 2019). In the authors’ view, that while these assumptions might have been viable when they were first proposed in 1928, none of them are currently valid based on today’s expanded scientific knowledge. For more detailed information regarding the limitations of the LNT model, the authors refer readers to a 2019 special edition in Chemico-Biological Interactions by Drs. Edward Calabrese and Robert Golden (Golden 2019).

Despite public comments highlighting research showing a threshold almost certainly exists for chrysotile asbestos (Beckett et al. 2020; Goodman et al. 2020; Mezie et al. 2020; Mossman 2020; Paustenbach and Brew 2020; Price 2020; Roggli and Sporn 2020), below which asbestos exposure is unlikely to cause mesothelioma, the EPA rejected the notion of a threshold response for asbestos based on “accepted models for cancer” (EPA 2005) and the Agency’s perception of the linear dose response model as being applicable to chrysotile asbestos (EPA 2020b).

Plaintiff’s attorneys and experts use the LNT model to argue that there is no safe level of exposure to chemical carcinogens and that there is no threshold below which exposure will not increase an individual’s risk of developing cancer. This claim has been strongly asserted for all forms of asbestos by experts for the plaintiffs. Lawyers repeatedly claim that “there is no safe dose with respect to asbestos and U.S. regulatory agencies agree with this position.” However, in the authors’ views, this is an outdated understanding of molecular biology that ignores decades of research demonstrating that subcellular, cellular, organ, and whole-body defenses exist that protect against endogenous and exogenous attacks on DNA and physiological function. Logically, the application of the LNT model to the real-world would mean that a single exposure to any carcinogen (e.g. a single exposure to sunlight) has the capacity to cause cancer, which, if this was a valid dose-response model, would have made the evolution of life on Earth unlikely (Golden et al. 2019).

Given the insight obtained in recent years about the abundance of mutations that occur each minute in humans and the myriad of DNA repair mechanisms that exist, the EPA’s application of the LNT model to chrysotile’s effects on lung cancer and mesothelioma seems unnecessarily conservative and likely to seriously over-estimate true risk. With respect to the chrysotile risk assessment, the EPA would do well to convene a panel of asbestos experts who have expertise in the mechanisms of cancer and the various repair mechanisms so they can improve their approach for estimating the risks associated with low level exposure.

The impact of the EPA assessment on society

The authors generally support a ban on nearly all uses of asbestos. However, we do not believe that the totality of scientific information suggests that working with historical brakes and gaskets that contained resin-soaked chrysotile “poses an unreasonable risk to the health of workers.” Most importantly, though, utilizing a poorly conducted risk assessment methodology that overlooked much of the published scientific literature and did not draw on the vast expertise of experts in the field who have studied encapsulated asbestos for decades sets a poor precedent for future risk evaluations. These shortcomings were recently identified in a NAS report (National Academies of Sciences 2021), and the authors hope that the Agency fixes the identified problems in its risk assessments conducted since 2016.

On a practical level, it is the authors’ opinion that the final EPA document opens the doors for billions of dollars of scientifically unwarranted litigation in the coming years. Even though substantial exposures to chrysotile asbestos generally ended approximately 40 years ago, there are convincing arguments that the background rate of pleural mesothelioma in the population is approximately three in 10,000 (Moolgavkar et al. 2009; Price and Ware 2009) and that a substantial fraction of mesotheliomas cannot be attributed to asbestos exposure (Spirtas et al. 1994; Rake et al. 2009; Aguilar‐Madrid et al. 2010; Lacourt et al. 2014; Offermans et al. 2014). Such mesotheliomas are unrelated to asbestos exposures. As such, we have to ask ourselves, is it in the best interests of the nation to award millions of dollars to these persons, even if their cancer was spontaneous or due to smoking? In the authors’ views, the weight of scientific evidence regarding chrysotile asbestos exposures makes this redistribution of wealth seem inappropriate.

Our recommendations

The authors of this manuscript have spent over 300 h studying the draft and final versions of the EPA’s risk evaluations for chrysotile asbestos, various supporting documents, public meeting records, and asbestos literature. In our view, if the Agency adopted our nine suggestions below, and incorporated the recommendations from the NAS (National Academies of Sciences 2021), the EPA would be equipped with the information needed to conduct a proper, scientifically sound risk evaluation of chrysotile asbestos. These proposed actions also incorporate requests made by some panelists to the Agency based on issues deemed unaddressed by the EPA, along with problems raised by various public commenters.

1. The regulated community should attempt to order from the internet brakes that allegedly contain asbestos, which are supposedly manufactured in Canada, India, Russia, and China. The brakes should then be assayed to determine if they contain chrysotile asbestos, with the results being published in a peer-reviewed journal.

2. If any of these purchased brakes can be obtained and shown to actually contain chrysotile asbestos, then a concerted effort should be made to determine how many might be entering the country and the number of potentially exposed persons. This is important information to the Office of Management and Budget (OMB), which is generally tasked with deciding if the monies spent on regulatory efforts by the EPA make economic sense.

3. Assemble all of the relevant toxicology studies on chrysotile asbestos, including those which evaluated the toxicity of fibers soaked in resins. This was identified by the NAS report as a major shortcoming of the methods that the EPA used in its recent risk evaluations (National Academies of Sciences 2021), and we identified approximately 100 articles that should have been included (Paustenbach and Brew 2020). These studies need to be incorporated into future iterations of this risk assessment – which will probably be resolved by the courts.

4. Assemble all the relevant epidemiology studies of exposed populations and create a table that identifies all of the strengths and weaknesses of these studies. There are approximately 20 applicable epidemiology studies of relevantly exposed populations, even though the EPA relied on only two cohorts of North and South Carolina textile workers, that were, in fact, not applicable to the goals of the EPA’s risk evaluation.

5. Assemble all of the various science panel views and the published papers of the past 30–40 years that are relevant to understanding the cancer potency of chrysotile asbestos and the amphibole fibers and the various low dose extrapolation models that are appropriate based on the most likely mechanism(s) of action for chrysotile asbestos.

6. Critically evaluate the above topics and then, using the current EPA Guidelines for Carcinogen Risk Assessment (EPA 2005), decide if the threshold approach is more appropriate than the linear no threshold (LNT) for chrysotile asbestos for identifying a virtually safe dose or reference concentration (RFC), then conduct the appropriate analyses.

7. Collect additional data, if necessary, to be certain that an adequate industrial hygiene database is available for understanding the exposure to asbestos associated with changing brakes and interacting with gaskets in the 1950s–1980s-time frame. Although the SAB panel seemed to believe the data from that time period were inadequate, hundreds of samples have been collected, and most appear in published, peer-reviewed papers and have been identified in this article and public comments to the Agency. This dataset, not surprisingly given the time constraints on the panel, was clearly not fully understood by all members of the SAB panel.

8. Importantly, be sure that the revised or updated chrysotile asbestos risk evaluation focuses entirely on the future probable exposure to chrysotile asbestos in brake dust and gasket particles, rather than assume that such exposures are likely to occur or are currently occurring in the United States.

9. Lastly, conduct a cost-benefit analysis, as is normally expected for regulatory initiatives, to evaluate whether a risk assessment of encapsulated chrysotile asbestos is necessary, given the small number of persons who could plausibly be exposed to chrysotile asbestos.

Closing

In the authors’ opinion, when one assembles all of the published scientific information that was not cited nor given serious consideration by the EPA, it is evident that its chrysotile asbestos risk evaluation lacks scientific rigor. This position is also supported by the NAS report that identified numerous and serious flaws in the methodology that the EPA used in its risk evaluation process (National Academies of Sciences 2021).

Unfortunately, it appears that the process was rushed by the EPA in an apparent attempt to have it approved before the new administration took office. For example, at the time of the SACC meeting in July 2020, the EPA acknowledged that the committee, at the time of the meeting, had received less than 25% of the comments that were submitted by outside scientists (Paustenbach 2020). In addition, the speakers at this public meeting were only allowed to talk to the panelists for 5 min. Further, the poor systematic review and involvement of special interest groups in this assessment is apparent when one tabulates the number of papers that were not considered and the manner in which some flawed papers were the foundation for the EPA’s risk evaluation (EPA 2020b; EPA 2020c).

The scientific shortcomings and the Agency’s flawed review process will probably not be rectified by scientific debate since the document has been finalized. Hopefully, the EPA will follow the recommendations stated in the NAS’ 2021 report (National Academies of Sciences 2021) and will remedy the myriad issues that have been identified in the chrysotile risk evaluation. As of July 2021, the risk evaluation has already impacted asbestos litigation (Finkelstein 2021; Holstein 2021).

Thus, in our opinion, it will be up to the private sector to bring this matter before the courts, or it will be used to mislead judges and jurors about the weight of scientific evidence that suggests that if chrysotile asbestos has the potency to cause mesothelioma, the required doses are likely > 100 f/cc-years, and the fibers probably must be much longer than 5 µm (likely closer to 25–40 µm in length) and have greater than a 3:1 aspect ratio (Berman and Crump 2003; Berman and Crump 2008a; Berman and Crump 2008b; Pierce et al. 2016).

It is the author’s opinions that in the modern era, due to OSHA regulations (OSHA, 1994), that the vast majority of exposure to chrysotile asbestos ended nearly 40 years ago in the United States and that exposure to asbestos in new products does not occur.

Abbreviations
1-BP	=	1-Bromopropane (n-Propyl Bromide)
AAABL	=	Aftermarket Automotive Asbestos-Containing Brakes/Linings
ACRF	=	Asbestos Claims Research Facility
ADAO	=	Asbestos Disease Awareness Organization
AMSTAR-2	=	assessment of multiple systematic reviews
ACE	=	Automated Commercial Environment
BMDS	=	Benchmark Dose
COUs	=	conditions of use
CBP	=	customs and border protection
DRE	=	draft risk evaluation for asbestos
f/cc-years	=	fibers/cc-years
FOIA	=	Freedom of Information Act
IUR	=	inhalation unit risk
IRIS	=	integrated risk information system
LNT	=	linear no threshold
LMS	=	linearized multistage
MLE	=	maximum likelihood estimation
NAS	=	National Academies of Sciences, Engineering, and Medicine
NCI	=	National Cancer Institute
NIOSH	=	National Institute for Occupational Safety and Health
NC	=	North Carolina
ONUs	=	Occupational Non-User
OSHA	=	Occupational Safety and Health Act
OCSPP	=	Office of Chemical Safety and Pollution Prevention
OMB	=	Office of Management and Budget
OPPT	=	Office of Pollution Prevention and Toxics
PEL	=	Permissible Exposure Limits
PBZ	=	personal breathing zone
PPE	=	personal protective equipment
RFC	=	reference concentration
SC	=	South Carolina
SAB	=	Science Advisory Board
SACC	=	Science Advisory Committee on Chemicals
TWA	=	time-weighed average
TSCA	=	Toxic Substances Control Act
TCE	=	trichloroethylene
US	=	United States
EPA	=	United States Environmental Protection Agency

Acknowledgements

This paper received numerous suggestions from five different peer reviewers selected by the journal’s editor and these comments improved the quality of our submission.

Declaration of interest

The authors work within Paustenbach and Associates, which is a consulting firm headquartered in Jackson, Wyoming. The firm specializes in conducting risk assessments of occupational and environmental hazards, as well as contaminated foods, sediments and soils, ambient air, radiological sites, and consumer products. Dr. Paustenbach and those who have worked with him have published nearly 40 peer-reviewed papers on asbestos over the past 20 years. He has served as an expert witness for defendants and has testified in over 400 depositions, as well as more than 30 trials involving the alleged health hazards of brakes, gaskets, phenolic molding compounds, mastics, and other materials having encapsulated asbestos. This paper received no funding from any entity and its development was not requested by any lawyer. No lawyer had input into the text, which was entirely written by the authors. The senior author, Dr. Paustenbach, gave oral comments regarding the inadequacies of the chrysotile asbestos risk evaluation to the EPA SACC panel during the summer 2020 public hearing. The PowerPoint presentation that he presented during this meeting is publicly available from the EPA. Additionally, Dr. Paustenbach has served on science advisory panels, for both Democratic and Republican presidents.