march 15, 2013 ms. wendy cleland-hamnett re: acc ......review plans for the work plan risk...
TRANSCRIPT
americanchemistry.com® 700 Second St., NE | Washington, DC 20002 | (202) 249.7000
March 15, 2013
Ms. Wendy Cleland-Hamnett
USEPA Headquarters
Ariel Rios Building
1200 Pennsylvania Avenue, N. W.
Mail Code: 7401M
Washington, DC 20460
Re: ACC comments on dockets: EPA-HQ-OPPT-2012-0722 (HHCB), EPA-HQ-OPPT-2012-
0723 (TCE), EPA-HQ-OPPT-2012-0724 (ATO), EPA-HQ-OPPT-2012-0725 (DCM and NMP)
Dear Ms. Cleland-Hamnett:
The American Chemistry Council (ACC)1 appreciates the opportunity to comment on the five
draft chemical risk assessments announced in the Federal Register on January 9, 2013. ACC has
a longstanding and strong interest in risk assessments conducted by the Environmental Protection
Agency (EPA), including those conducted under the Toxic Substances Control Act (TSCA).
ACC welcomes the general direction that EPA has taken in the Work Plan Chemicals program to
prioritize chemicals for further review and conduct targeted assessments that may then be used to
consider whether additional regulatory action, if any, is warranted. In general, ACC agrees that
this type of approach, with some important and necessary refinements, can help the Agency
effectively evaluate existing chemicals in commerce.
ACC commends the Agency for conducting targeted quantitative assessments that focus on the
potential risks associated with certain uses and applications of the work plan chemicals, and for
employing a margin of exposure (MOE) approach for human health evaluations. The
assessments are appropriately focused on those risks that have not already been determined to
present minimal or negligible risk for ecological or human health endpoints.
1 The American Chemistry Council (ACC) represents the leading companies engaged in the business of chemistry.
ACC members apply the science of chemistry to make innovative products and services that make people's lives
better, healthier and safer. ACC is committed to improved environmental, health and safety performance through
Responsible Care®, common sense advocacy designed to address major public policy issues, and health and
environmental research and product testing. The business of chemistry is a $760 billion enterprise and a key element
of the nation's economy. It is the largest exporting sector in the U.S., accounting for 12 percent of U.S. exports.
Chemistry companies are among the largest investors in research and development.
The MOE approach used for non-cancer assessments, consistent with the approach used by
EPA’s Office of Pesticide Programs (OPP), the European Union, and Canada, is a robust
methodology that improves transparency and is preferred over approaches that use reference
values (RfCs or RfDs).2 Targeted quantitative assessments and use of the MOE are important
steps for the Agency to take in its risk assessments. When properly implemented, these
methodologies should allow EPA to confidently and consistently focus its resources on uses that
present the greatest potential for concern.
While the overarching approach is solid, ACC has several suggestions to improve the overall
methodology and to ensure a scientifically rigorous approach to evaluating risks. Our
suggestions fall into the following four main areas:
1) The assessments use a screening-level methodology and the results should not be used for
further regulatory action without further refinement and evaluation.
In many instances, EPA’s methodology uses worst-case or high-end assumptions. This approach
is very conservative and consistent with a screening-level risk assessment where health
protective assumptions are appropriately used for parameters employed in calculating exposures
and hazards to assure that potential risks are not underestimated. Screening-level assessments
such as these are not designed to provide true and accurate estimates of risk. When a screening-
level assessment indicates an acceptable level of risk, the Agency has a high degree of
confidence that the potential risks are much lower than the calculation and therefore the true
risks are lower and/or perhaps non-existent. However, when a screening-level risk assessment
indicates a potential concern for a health or environmental effect, this does not mean that the true
risks are significant and warrant action. Rather, it means that the risk evaluation should be
refined using more realistic and accurate parameters in the methodologies to calculate risks.3 The
outcome is then a refined risk assessment that more accurately quantifies actual risks.
Before these risk assessments were publicly released, ACC called for EPA to treat these
assessments as Highly Influential Scientific Assessments (HISAs) in its comments on the Peer
Review Plans for the Work Plan Risk Assessments.4 ACC notes that the depth and quality of
the assessments as published clearly are in the nature of screening-level assessments. If EPA
agrees that the assessments are indeed screening-level assessments, and publicly clarifies that
aspect, ACC would agree that EPA's original characterization of the assessments as "influential"
2 ACC prefers the use of the MOE for these assessments. When reference values are used they incorporate science
policy judgments, in the form of uncertainty factors that are not immediately transparent to risk managers. Using an
MOE approach, the risk manager can more clearly determine if the differential between the exposure level and the
effect level is appropriate considering the populations exposed, endpoints of concern, strength of the evidence and
other key factors that are often built into uncertainty factors but are not immediately transparent to the risk manager. 3 See for example, http://www.epa.gov/oswer/riskassessment/ragsa/pdf/rags_ch6.2.pdf, Page 6-25
4 See ACC September 27, 2012, letter from ACC to EPA
may be appropriate. If EPA does not agree that the assessments are screening-level assessments,
and will base regulatory action on them as drafted, ACC believes that EPA is required to treat
them as HISAs. In either event, EPA should respond to public comments received on the peer
review agendas before commencing with the peer reviews.
2) A clear framework for prioritizing specific use scenarios must be developed and
implemented.
EPA has not clearly articulated why it chose to focus on the use scenarios selected for these five
assessments. In some cases, the specific use scenario evaluated is not consistent with current
practices or existing safety recommendations, e.g., the NMP use scenario assumed that gloves
were not used, despite the fact that the use of gloves is a recommended practice for the use of
this substance. In order to maximize public health protection, EPA should focus on use
scenarios that are plausible and consistent with the product’s directions for use, as well as the
Consumer Product Safety Commission’s (CPSC) current recommendations for use.5 Clear
procedures and protocols should be developed, through an open and transparent process, to
describe how EPA will choose the use scenarios determined to be high priority for each chemical
assessed.
3) The assessments must be based on the best available data and must use reproducible state
of the science approaches to evaluate risk.
The five assessments apply inconsistent standards to existing scientific information, using
methodologies that do not comport with today’s best scientific approaches to evaluate and
integrate scientific information. A systematic evaluation of the quality (including relevance and
reliability) of each study is necessary. When evaluating both hazard and exposure, it is critical
that EPA rely on the studies of the highest quality, not simply those studies that produce the
lowest points of departure or the highest exposure estimates. EPA should develop, through an
open and transparent process, clear procedures and protocols that will promote consistent and
scientifically sound assessments that can be compared and evaluated.
4) The assessments should explicitly address the authority of the Occupational Safety and
Health Administration (OSHA) to regulate occupational exposures and strive to achieve
consistency with OSHA’s approach to assessing safety.
OSHA is the authority to prescribe or enforce regulations affecting occupational safety and
health. It is critically important that EPA policies and scientific approaches are coordinated with
5 See CPSC recommendations at: http://www.cpsc.gov/en/Safety-Education/Safety-Guides/Home-Appliances-
Maintenance-and-Structure/What-You-Should-Know-About-Using-Paint-Strippers/
OSHA’s and are not in conflict. Otherwise, there remains potential for confusion and
inconsistency.
The attached Appendix A elaborates on ACC’s concerns with respect to each chemical-specific
assessment. In Appendix B, we provide specific suggestions to improve the charge questions
that will be used by peer reviewers.
Thank you for considering our comments and suggestions. We look forward to seeing the
Agency’s written response to comments. If we can provide additional information, or if you
have any questions regarding our comments, please feel free to contact Nancy Beck at 202-249-
6417 or Christina Franz at 202-249-6406.
Sincerely,
Senior Director
Regulatory and Technical Affairs
Senior Director
Regulatory and Technical Affairs
Appendix A
ACC Detailed Comments on Chemical Risk Assessments
ACC commends EPA for conducting targeted, quantitative assessments focusing on the potential
risks associated with certain uses and applications of the work plan chemicals, and for employing
a margin of exposure (MOE) approach for human health evaluations. When properly
implemented, these methodologies should allow EPA to confidently and consistently concentrate
its resources and efforts on the uses that present the greatest potential risk.
While the overarching approach is solid, ACC offers the following suggestions to improve the
overall methodology and ensure a scientifically rigorous approach to evaluating risks. Our
suggestions are discussed below in further detail and refer to specific examples from the draft
assessments where appropriate. Adopting the suggestions outlined below should enable EPA to
strengthen and improve its approach to evaluate chemical risks. We believe these
recommendations can be implemented by EPA in a timely and efficient manner.
1. The assessments use a screening-level methodology and the results should not be used
for regulatory action without further refinement and evaluation.
In many instances, EPA uses worst-case or high-end assumptions that lead to an overestimation
of the potential risks. Although EPA acknowledges this conservative bias in some of the
assessments, ACC believes that a further refinement of the assessments is necessary before the
results can be used properly to initiate regulatory decisions. EPA‟s webpage on these
assessments states: “If an assessment of specific uses indicates potential risk of concern, EPA
will evaluate and pursue appropriate risk reduction actions, as warranted. If an assessment
indicates negligible risk, EPA will conclude its current work on assessment of those specified
targeted uses of that chemical.”1 While EPA‟s approach is appropriate to conclude work on an
assessment where negligible risks are identified, it is not appropriate to pursue further risk
reduction measures until refined assessments are conducted.
The schematic below (Figure 1), demonstrates how EPA could use the MOE derived from a
screening-level risk assessment as a trigger to determine when further evaluation, i.e., conducting
a more refined risk assessment to more accurately quantify potential risks, is necessary. In this
particular case, a MOE of 30 is used because it was the default MOE value used in the
trichloroethylene (TCE) and N-methylpyrrolidone (NMP) assessments. However, depending
upon the conservative nature of the screening assessment, it may be appropriate to use a higher
or lower MOE to trigger further evaluation. As illustrated below, additional risk management
actions should only be considered after a refined assessment is concluded. EPA should adopt
this approach and conduct further refined assessments for TCE, NMP and dichloromethane
1 See EPA webpage at: http://www.epa.gov/oppt/existingchemicals/pubs/workplans.html accessed on Feb. 18,
2012.
ACC Appendix A 2 | P a g e
(DCM) for those use scenarios where the MOE derived from screening level assessments fall
below the benchmark value of 30.
Figure 1. Using the Margin of Exposure as a Trigger for Further Evaluation.
Chemical Specific Examples:
Antimony Trioxide (ATO):
Exposure conclusions are contradictory
In the ATO assessment, EPA repeatedly and accurately states: “Use of the generic category,
„antimony compounds‟ and conservative assumptions, regarding modeled releases likely
overestimated exposure potential.”2 Since EPA‟s conservative approach found negligible risks
and that no further work is needed, EPA can reasonably conclude that the uses evaluated are
safe. However, had EPA found risks of concern, the Agency would need to refine the
assessment and develop a more thorough and accurate assessments of risk before considering
risk reduction actions under TSCA.
2 See ATO draft assessment at pages 8 and 47.
ACC Appendix A 3 | P a g e
In addition, EPA states on page 8 of the assessment: “Measured and modeled concentrations of
ATO are assumed to represent reasonable estimates of environmental exposures.”3 However,
EPA also states that the modeling used conservative assumptions and approaches, including in
the use of the E-FAST2 model, which led to overestimates of exposure potential. It is not clear
how overestimates of exposure can be reasonable. These conclusions are inherently
contradictory and we recommend that EPA reconcile this inconsistency before finalizing the
assessment.
Trichloroethylene (TCE):
Overly conservative assumptions and values result in an unrealistic assessment which
should be refined
In the TCE assessment, there are many overly conservative assumptions that skew the
assessment to overestimate risks. Even EPA‟s overall framework for acceptable risk is
conservative. For instance, although other EPA programs, including the Office of Solid Waste
and Emergency Response (OSWER), consider risk ranges of 10-4
to 10-6
as acceptable for cancer
risk, EPA has set a bright line for acceptable consumer risk at 1 x 10-6
for cancer and an
acceptable worker cancer risk at 10-5
. In addition, OSHA considers a risk of 1 in 1000 to be
significant.4 Accordingly, greater clarity is needed regarding EPA‟s rationale for acceptable risk
for occupational exposures.
Most conspicuous however, is EPA‟s use of the 99th
percentile Human Equivalent Concentration
(HEC), for the development of non-cancer points of departure. EPA fails to provide a rationale
or scientific justification for using this conservative range to represent the sensitive human HEC.
It is unclear why this value is used as opposed to the more standard 90th
or 95th
percentile value
for sensitive populations. Indeed, the stability and justification for such a value is questionable.
Further, although there is some limited comparison in the assessment to the HEC 50th
percentile,
those values are not presented transparently. EPA only carries forward the range of 99th
percentile values and uses the most conservative point in the range (the low end of the 99th
percentile) to benchmark the assessment. The full quantitative impact of using a different point
in the range, or using a different percentile, is not clearly presented in the assessment or
appendices. Although some qualitative descriptions are provided, they are not sufficiently robust
to understand the quantitative impact of using different HEC percentiles. While we recognize
that EPA does not want to under-estimate risks, using such a conservative value, in conjunction
with other conservative decisions in the assessment, undermines the value of this assessment,
rendering it unrealistic.
3 When available, the Agency should preferentially use robust environmental monitoring data to determine
whether environmental concentrations of a substance exceed a concentration of concern. 4 See OSHA discussion of significance of risk at:
https://www.osha.gov/pls/oshaweb/owadisp.show_document?p_table=PREAMBLES&p_id=1007.
ACC Appendix A 4 | P a g e
Coupled with the overly conservative use of the lower bound of the HEC99, EPA also
inappropriately uses 30 as an acceptable MOE. The value of 30 accounts for an uncertainty
factor of 10 for intraspecies differences. As the HEC99 already represents the most sensitive
individual in the population, adding an intraspecies uncertainty factor of 10 to the MOE
compounds the conservative risk calculation. Further justification for the use of this factor is
necessary.
More complete characterization of the range of results is needed
In its presentation of the range of results, the TCE assessment is inconsistent with both the EPA
risk characterization handbook and the OMB/OSTP Memorandum on Principles for Risk
Analysis.5,6
It is impartially biased toward high-end conservative estimates. While this is
acceptable for a screening-level assessment, EPA should refine this screening-level assessment
and in doing so must follow the risk characterization handbook and OMB/OSTP memorandum
and evaluate and present the full range of impacts more transparently and objectively. In
particular we suggest that in refining the assessment, EPA conduct the assessment such that
Table 3-26, clearly provides the quantitative values for the 50th
percentile (lower and higher end)
HEC, the 75th
percentile (lower and higher end) HEC, the 90th
percentile (lower and higher end)
HEC and the 95th
percentile (lower and higher end) HEC as well as the 99th
percentile (lower and
higher end) HEC. Presenting this information transparently will allow decision managers to
fully understand the range of risks identified and the impacts major assumptions, like the HEC
percentile, have on the identified risks. A complete understanding of the impact of all major
assumptions will improve the risk managers‟ ability to make informed and science-based
decisions.
Lifetime daily average exposures should be used
Together with the most conservative percentiles for the HEC (which are daily averages
computed for continuous exposures) EPA makes many conservative assumptions related to the
exposure modeling. For instance, EPA uses HECs for daily exposures as a benchmark to judge
the intermittent exposures for degreasers and hobbyists. The appropriate comparison should be
the lifetime daily average exposure for the worker or hobbyist. However, EPA makes no
adjustments for the intermittent exposures (once per week for the clear spray scenario and twice
a month for the degreaser). The EPA approach leads to a large overestimation of the true
5 See EPA Risk Characterization Handbook, 2000 available at: http://www.epa.gov/spc/pdfs/rchandbk.pdf. The
Handbook discusses the need for transparency, including clear presentation of impacts of choices. 6 See OMB and OSTP Memorandum for the Heads of Executive Departments and Agencies, M-07-24, 2007,
available at: http://www.whitehouse.gov/sites/default/files/omb/assets/regulatory_matters_pdf/m07-24.pdf. This memorandum discusses the importance of providing quantitative values to inform a range of policies to reduce risk.
ACC Appendix A 5 | P a g e
xposure. The lifetime average daily doses should be the value compared to the HEC levels.7
These adjustments are necessary for a refined and more accurate assessment of true risks.
Internal dose metrics are necessary
Similarly, as EPA is looking at discontinuous exposures, the use of internal dose metrics would
be more appropriate. EPA‟s approach does not account for the elimination of TCE and its
metabolites from the body. This elimination would occur between exposure periods for the
hobbyists and bystanders and therefore the true level of TCE in the target tissues must be
accounted for. Based on the half-lives provided in Table 3-15 of the draft assessment, it is likely
that, for episodic exposure scenarios, TCE would be completely eliminated from the body
between exposures for the hobbyist and the bystander, and the TCE would be eliminated over the
weekend period for the occupational workers. Using internal dose metrics would provide a more
accurate and refined picture of true TCE levels.
Data and modeling do not represent today’s exposures
For the occupational degreaser exposure scenario, it is not clear that the EPA is using the most
currently available data, particularly when estimating the number of workers exposed. EPA is
using data from the 1980s which likely reflects data from the 1970s. Since that time, Toxics
Release Inventory (TRI) emissions data demonstrates that, TCE releases have decreased
substantially.8 We also know that workplace practices and industrial hygiene standards have
improved. Thus, it is likely that EPA overestimates the worker and bystander exposure numbers.
In addition, the draft assessment contains no discussion of the impacts of the Halogenated
Solvent Cleaning National Emission Standard for Hazardous Air Pollutants (NESHAP) which
EPA finalized in 2007.9
To evaluate worker exposure for the TCE risk assessment, EPA did not use published studies
with workplace sampling data and instead chose to estimate exposures based on calculating an
emission rate. Further, EPA cites Walden 1989 and incorrectly states degreaser emissions as
ranging from 2.57 to 27.29 g TCE/min. Walden 1989 indicates data is for environmental
emissions released externally from ventilation hood exhaust and only 9.5% of the value is
emission escaping into the workplace (Walden 1989). The EPA methodology should not assume
environmental emission data is representative of occupational exposures.
The Agency admits that the E-FAST model leads to “hypothetical yet conservative” exposure
values.10
As such, it would be inappropriate for EPA to use the findings of this screening-level
7 See http://www.epa.gov/oswer/riskassessment/ragsa/pdf/rags_ch8.pdf.
8 http://iaspub.epa.gov/triexplorer/tri_release.chemical
9 See 2007 final rule at 72 FR 25138 available at: http://www.epa.gov/ttn/atw/degrea/fr03my07.pdf.
10 In the TCE draft assessment, at page 32, EPA states: “The resulting exposure estimates are characterized as
hypothetical exposures. As explained more fully in the Supplemental Information these hypothetical exposures are
ACC Appendix A 6 | P a g e
assessment to inform regulatory actions without conducting a more refined assessment that more
accurately reflects actual risks and exposure scenarios which are not hypothetical.11
In addition,
a model should be used to provide a time-weighted average concentration, over 24 hours, to
make an appropriate comparison to the HEC levels. Only one example of an estimated air
concentration is presented in the assessment, making it difficult to understand and reproduce all
the calculations. The conversion EPA makes to go from the mg/kg/day E-FAST output is
confusing and unclear. The inputs to the model should justified and transparently presented to
inform the refinements that are necessary to build the draft assessment beyond the screening
level.
N-methylpyrollidone (NMP):
The minimal effects on fetal body weight do not warrant use of BMD05
In the draft NMP assessment, to evaluate inhalation exposures and developmental toxicity
effects, EPA uses a point of departure (POD) derived from a rat study by Saillenfait et al. (2001,
2003)12
. In this teratology study, no teratogenicity or effects on survival were noted for the
fetuses. Developmental toxicity occurred in the form of a 4.9% decrease in overall fetal body
weight at the high dose of 487 mg/m3 (120 ppm).
13 Based on these results, the maternal NOAEL
was determined to be 122 mg/m3 and the developmental NOAEL was 243 mg/m
3. It should be
noted that a 5% reduction in fetal weight is generally the minimum reduction that would be
considered biologically significant.14
Instead of simply describing this effect as “fetal toxicity”
in summary tables, EPA should be more transparent and describe the effect as “minimal changes
in fetal weight.” Further discussion of the biological significance of this endpoint is needed in a
refined assessment. In addition, for benchmark dose (BMD) modeling, EPA uses a BMD05
instead of the typical BMD10 as recommended in EPA guidance.15
While the guidance discusses
more likely to be high-end exposures than they are to be central tendency exposures; thus they represent conservative exposure values.” 11
Id., at page 70 EPA states: “Furthermore, for the exposure estimations in this risk assessment, the use patterns assumed for the two hobbyist products, including mass of product used per event, duration of event, and events per year, are hypothetical and are not based on consumer product survey data.” 12
The two references cited for the Saillenfait inhalation developmental toxicity study appear to present the same set of inhalation data. The 2003 citation is a peer-reviewed publication, while the 2001 citation is a conference abstract which appears to include the same inhalation data as well as oral developmental toxicity data, which were not used by the EPA in its evaluation. EPA should consider citing only Saillenfait et al., 2003 as the most definitive source for their evaluation. 13
There is a discrepancy between conversion factors (mg/m3 to ppm) for NMP as they relate to the Saillenfait et al.
(2003) study. On p. 46 it is stated that 243 mg/m3 = 120 ppm, while on p. 150 it is stated that 487 mg/m
3 = 120
ppm. The information on p. 46 is incorrect and should be corrected in the final document. 14
See Hayes, W, 1989, Principles of and Methods of Toxicology, Second Edition, Chapter 11, Test Methods for Assessing Female Reproductive and Developmental Toxicology, p. 325-326. 15
See EPA (2012) Benchmark Dose Technical guidance, available at: http://www.epa.gov/raf/publications/pdfs/benchmark_dose_guidance.pdf, which states: “The 10% response level has customarily been used for comparisons because it is at or near the limit of sensitivity in most cancer bioassays and in noncancer bioassays of comparable size.”
ACC Appendix A 7 | P a g e
using a BMD 5% for some developmental effects, EPA should reevaluate whether the minimal
changes found in this study warrant the conservative use of the BMD05.
EPA’s use of scaling to extrapolate from rat dermal to human dermal is not adequately
justified scientifically and should be revised
In the draft assessment, EPA extrapolates dermal absorption in the rat to human dermal doses. In
this cross-species extrapolation, EPA did not take into account the innate difference in absorption
rate of substances across rodent skin versus human skin, as described in the EPA Interim Report
on Dermal Exposure Assessments (EPA, 1992).16
Rat skin has been shown to be consistently
more permeable to a range of chemicals than human skin, and EPA‟s own report on this topic
states that, “it may be reasonable to correct the percutaneous absorption rates from mouse and rat
studies by a factor of 3 to 5 to obtain more realistic estimates of human K values”. This is
consistent with an in vitro skin study conducted by Príborský and Mühlbachová (1990),17
showing that permeability to NMP is approximately four-fold lower in human than in rat skin.
Thus, in order to refine this assessment, it is recommended that the differences in percutaneous
permeability be taken into account when scaling rat-to-human dermal doses.
EPA uses a two-hand dermal exposure model to estimate dermal exposure. The model makes
basic assumptions about dermal absorption for occupational and residential use, including the
weight fraction, surface area of exposed skin, and exposure duration (see section 3.1.6 and Table
3-4, p. 30). The model also assumes that 100% of the applied NMP will be exposed through the
skin, even with aqueous NMP formulations. However, this assumption fails to take into account
the work of Keener et al. (2007), who measured dermal absorption rates for various
concentrations of NMP in human volunteers.18
Absorption rates fell from 6.5 mg/cm2 for neat
NMP to 0.9 mg/cm2 for a 50% aqueous formulation of NMP, and only 16% of the applied dose
was accounted for in the urine after a two-hour exposure period to the 50% formulation,
demonstrating that aqueous formulations are less well absorbed than neat NMP. Using the
absorption rates from the Keener study, the maximum two-hour time allotted for (non-bathtub)
paint stripping in Table D-4 (p. 105) and the absorption rate of 50% NMP, exposure for an 80 kg
individual would be 11 mg/kg rather than the 100 mg/kg dermal ADR estimated in Table 3-4. It
is therefore recommended that EPA refine these estimates of dermal exposure to take into
account information on concentration-dependent absorption rates based on the published human
studies.
16
See EPA (1992) Dermal Exposure Assessment: Principles and Applications, Interim Report. EPA Office of Health and Environmental Assessment, Section 3.4. 17
See Príborský J, Mühlbachová E (1990) Evaluation of in-vitro percutaneous absorption across human skin and in animal models. J Pharm Pharmacol, 42:468-72. 18
See Keener SA, Wrbitzky R, Bader M. Human volunteer study on the influence of exposure duration and dilution of dermally applied N-methyl-2-pyrrolidone (NMP) on the urinary elimination of NMP metabolites (2007) Int Arch Occup Environ Health, 80(4):327-34.
ACC Appendix A 8 | P a g e
In addition, in Chapter 1 of the NMP draft assessment (p. 14) EPA states: “NMP may cause skin
swelling, blistering, and burns after prolonged direct contact with skin.” EPA‟s assumption that
users would endure dermal exposure of both hands over an area of 490 cm2 (residential) or 840
cm2 (occupational) for a prolonged length of time seems highly unlikely.
19 In a revised
assessment, more realistic estimates of exposure time and the use of gloves should be factored
into the assessment. This would be more consistent with the statement in the Executive
Summary of draft assessment (p. 13) that dermal exposure estimates may not represent real
world conditions based on the corrosive properties of NMP.
EPA’s exposure scenarios should be clarified and exposure estimates refined
Finally, EPA‟s use of conservative scenarios does not seem logical and is not clearly articulated
in the assessment tables. For example, the scenarios in Table 3-7 lead to confusing outcomes. In
the upper-end scenario for a non-user (scenario 3) the table shows a higher exposure for a user as
compared to the upper-end user scenario (scenario 2). This outcome does not appear logical.
Similarly, the scenarios for residential use generate high concentrations as compared with the
actual published occupational inhalation exposures shown in Table 3-3. For example, a short-
term peak exposure in professional contractors ranges from 3.3-10 ppm (less than eight hours),
while eight-hour residential exposures are as high as 14 ppm. This does not make sense, given
that professionals are presumed to use a formulation containing 100% NMP, whereas residential
formulations contain 25-40% NMP, and contractors are more apt to use these products for longer
periods throughout the day. Table D-4 also contains some inconsistencies and does not appear to
represent common exposures. Some examples include:
The scenarios for non-users indicate that they will be in the workshop during application
and scraping periods. This is contrary to the text on p. 109, which indicates that non-
users were assumed to be in the rest of the house (ROH) throughout the run model. It
should be verified that the model was run correctly for non-users and the table should be
corrected.
The table also shows non-users being exposed for a longer period of time than users for
both the application and scraping periods. It is unclear why this assumption was made.
It is unclear why the assessment states on p. 109 that some users took breaks outside
rather than in the ROH. It is not clear why this likely exposure scenario was apparently
not included in the modeling.
It is recommended that EPA refine, revise, and clarify these tables, and associated modeling to
ensure that the exposure scenarios are plausible and realistic.
19
It is also noteworthy that Keener (2007) chose to administer NMP via a small 10 cm2 pad (occluded)
administered to the back of the hand in order to mimic a workplace scenario where splattering would cause small patches of NMP-exposed skin to be present under work gloves. Therefore, the assumption that 840 cm
2 of skin
would be exposed in an occupational setting with the use of gloves is likely to be a gross over-estimate.
ACC Appendix A 9 | P a g e
Dichloromethane (DCM):
Greater transparency, clarity and central tendency estimates are necessary refinements
before this assessment can be relied upon for regulatory action
The DCM draft assessment presents exposure information in a conflicting and confusing manner,
thus making it difficult to reproduce and determine where refinements are needed. For instance,
in Appendix E of the draft it is unclear how the exposure concentration ranges in Table 3-3 relate
to the values in the source documents. Appendix E also contains discussion of data from the
OSHA Integrated Management Information System dataset. The text concludes that the
relevance of these data to paint stripping is uncertain; however, there is no doubt that these data
constitute representative exposures from an occupational setting where DCM is used. In refining
the exposure assessment, EPA should consider these data, which indicate significantly lower
exposures than the values EPA calculates in the draft assessment.
Another concern in the draft assessment is EPA‟s use of average daily concentrations (ADCs)
and lifetime average daily concentrations (LADCs), using the minimum and maximum time
weighted averages (TWA). By using only the extremes of the range, there is no sense of the
average values. Use of central tendency estimates would be useful and informative, particularly
for the LADCs which should be based on the average exposure concentration over time, rather
than a single maximum eight-hour TWA. A refined exposure assessment that uses central
tendency values should be conducted to inform true risks. EPA‟s current approach is not
reasonable in that it assumes an individual would be exposed to the maximum eight-hour TWA
for the entire working life.
Greater transparency and clarity is needed to refine the inhalation exposure scenarios. One
interesting observation is reflected in Table 3-7, where the upper-end scenarios for the non-user
(i.e., Scenarios 3 and 6) result in higher exposure concentrations than those in the upper-end
scenarios for the user (i.e., Scenarios 2 and 5). It is not clear why the upper-end scenarios for the
user (i.e., Scenarios 2 and 5) actually do not produce the highest exposure estimates. Further
clarification and refinement is necessary before this assessment can be relied upon as a basis for
regulatory action.
2. A clear framework for prioritizing specific use scenarios must be developed and
implemented.
In each of these five assessments, it is not clear why EPA concentrated on some of the use
scenarios in the assessments. In some cases, the use scenario evaluated is not consistent with
current practices or existing safety recommendations. Improved stakeholder engagement at the
design/problem formulation phase when choosing use scenarios would be tremendously helpful
to improve these and future risk assessments.
ACC Appendix A 10 | P a g e
In addition, in order to maximize public health protection, EPA should focus on use scenarios
that are plausible and consistent with CPSC‟s and OSHA‟s federal recommendations for use.
Clear procedures and protocols should be developed, through an open and transparent process, to
describe how EPA will choose the specific use scenarios that are determined to be high priority
for each chemical.
Chemical Specific Comments:
N-methylpyrollidone (NMP):
EPA’s assessments should be based on intended uses in accordance with label instructions
It is not clear why EPA used a scenario for dermal absorption that has a user failing to wear
gloves. It is clearly stated in Chapter 1 of the NMP draft assessment (p. 14) that “NMP may
cause skin swelling, blistering, and burns after prolonged direct contact with skin.” The
assumption that users would endure dermal exposure of both hands over an area of 490 cm2
(residential) or 840 cm2 (occupational) for a prolonged length of time seems highly unlikely. In
fact, the CPSC‟s current recommendations for use clearly recommend that chemical-resistant
gloves be worn.20
In addition, paint strippers containing NMP are clearly labeled with eye and
skin irritancy warnings and the very first written recommendation is to wear chemical resistant
gloves.21
EPA has not quantified the potential risks for use of the material in accordance with
current labeling instructions. On the contrary, EPA has done the opposite. The Agency has
calculated risks when the use is completely at odds with label instructions. The label on a
product and instructions for use provide critical information about how to handle and safely use a
product to avoid harm to human health and the environment. Therefore, EPA‟s assessments
should be based on intended uses in accordance with label instructions.
In Section 3.4 of the draft assessment, EPA attempts to justify the no-glove scenario by stating,
“it may address accidental exposures that may occur in the workplace.” If the goal of the
assessment is to address accidental situations, that fact should be clarified in the executive
summary and conclusions. However, since it is quite unlikely that it is an objective of this
assessment, additional justification for this scenario is warranted. As stated above, EPA should
conduct a refined assessment that includes a realistic scenario in which a user is wearing gloves
as directed by use instructions.
3. The assessments must be based on the best available data and must use replicable, state
of the science approaches to evaluate risks.
The assessments apply inconsistent standards to existing scientific information and use scientific
approaches that do not comport with current standards for developing new robust hazard and
20
See CPSC recommendations at: http://www.cpsc.gov/en/Safety-Education/Safety-Guides/Home-Appliances-Maintenance-and-Structure/What-You-Should-Know-About-Using-Paint-Strippers/. 21
See, for example, bottles of Citristrip stripping gel found in high quantities at Home Depot.
ACC Appendix A 11 | P a g e
exposure information. In particular, a systematic review approach should be used to evaluate
and integrate scientific information to ensure that EPA is using the best available methodologies
and relying on the best and most relevant scientific data. EPA should develop, publish, and seek
public comment on clear procedures and protocols for the assessments to develop reproducible
and scientifically sound assessments that can be compared and evaluated using similar, rather
than discordant, approaches.
a. Crosscutting Scientific Concerns:
i. Use of “Level of Concern” is Inappropriate
In each assessment where EPA uses the MOE, the term „Level of Concern‟ for the MOE <
composite UFs is inappropriate and should be removed. The WHO and European agencies
(EFSA, 2005; UK, 2010) have used the term „Minimal‟ or „Minimum MOE‟ to indicate what
MOE is considered „safe‟ and EPA‟s Office of Pesticide Programs (OPP) has used the term
“Target” MOE (e.g., EPA, 2011).22
Use of the term „Level of Concern‟ is poor risk
communication and conveys a conclusion about risk (population and/or individual) that is not
consistent with the risk communication associated with the use of the EPA reference values or
other values, such as the National Academies AEGL. EPA should consider substituting
„minimal‟ or „target‟ MOEin the tables or alternatively, report MOE and use the text to discuss
the minimal MOE associated with each guidance value (as the EPA OPP has employed).
ii. Systematic and Consistent Framework is needed
EPA does not apply a systematic approach to evaluating data throughout the five assessments,
thus we see an inconsistent use of hazard values. There does not appear to be consistent criteria
utilized to select the appropriate values across the assessments. While an approach that evaluates
each value on the merits of its underlying scientific basis and relevance is preferred, it would be
helpful for EPA to develop a consistent hierarchy for the selection of relevant hazard values to
use in screening-level assessments. It may not be necessary for the Agency to conduct an in-
depth analysis of the scientific underpinnings of the various available values because these are
screening-level assessments.
iii. Treatment of Uncertainties Should be Improved
The treatment of uncertainties in all of the draft assessments could be significantly improved.
While it is a promising that EPA has included a discussion of uncertainties and limitations in
four of the assessments (ATO does not include a discussion of uncertainties), the discussion
22
See: EFSA. 2005, Opinion of the Scientific Committee on a request from EFSA related to A Harmonised Approach for Risk Assessment of Substances Which are both Genotoxic and Carcinogenic. Available at: http://www.efsa.europa.eu/en/efsajournal/doc/282.pdf; UK 2010, The Development of a Web-Enabled Framework for Probabilistic Exposure Assessments. www.hse.gov.uk/research/rrpdf/rr763.pdf; and EPA 2011, Triclosan Residential Exposure & Illustrative Risk Assessment, Office of Pesticide Programs, Antimicrobials Division.
ACC Appendix A 12 | P a g e
provided is minimal and incomplete. A more expansive discussion of assumptions and
uncertainties would strengthen the draft assessments. For example, in the HHCB assessment, the
discussion of uncertainty is predominantly focused on data gaps in available studies and missing
information. EPA‟s response to these concerns is to use conservative values (i.e., a conservative
LOAEL) to account for this. The NMP draft assessment similarly addresses data gaps with
conservative high-end assumptions and fails to make an attempt to further clarify whether actual
central and high-end exposures lie within the range of exposures estimated in the draft
assessment. Therefore, the final user characterization for consumer inhalation is considered to
be “upper-end to bounding”.
In the DCM draft assessment, in addition to treating data gaps with conservative assumptions,
EPA acknowledges that the term “upper-end” is used rather than the term “high-end” because
“more definitive descriptors would imply an inappropriate level of accuracy.” EPA
acknowledges that it is unclear whether central tendency, as well as high-end exposures, fall
within the range of exposures estimated in the assessment. In each assessment, for both hazard
and exposure, where potential risks are identified, it is critical that EPA more thoroughly attempt
to quantify the cumulative impacts of all of the assumptions and inputs used to compensate for
the data gaps.
b. Chemical Specific Comments:
ATO and HHCB:
Unclear Standards for Study Quality and Acceptable Risk
In both the ATO and HHCB draft assessments, the standards applied for study reliability and
data adequacy are unclear. Greater transparency is needed to ensure that EPA‟s approach relied
on the best available information. In addition, further clarity is needed concerning what EPA
considers to be negligible as opposed to acceptable risk. A priori criteria should be developed.
In the draft ATO assessment, less than 0.4% of 3,643 samples exceeded the chronic
concentration of concern and EPA determined this to present „minimal concern‟. In the HHCB
assessment, 1.5% of 600 samples exceeded the chronic concentration of concern and EPA
determined that the risks were „negligible‟. It would be helpful for EPA to clarify the criteria for
acceptable (negligible, minimal or other) as well as unacceptable levels of concern.
It is also unclear why EPA uses two different approaches to evaluate safety. For the human
health evaluations EPA uses a MOE approach; however, in the ATO and HHCB assessments,
EPA appears to incorporate uncertainty adjustments into the hazard benchmarks (typically a
concentration of concern in the two assessments) and uses a risk quotient (RQ) approach. We
believe the MOE approach is more transparent and should be preferred for all evaluations. 23
23
As noted above, when using the MOE approach, using the label “Level of Concern” is inappropriate. The WHO and European agencies (EFSA, 2005; UK, 2010) have used the term “Minimal” or “Minimum MOE” to indicate what
ACC Appendix A 13 | P a g e
Trichloroethylene (TCE):
(i) A Systematic Review of Study Quality and Relevance is Absent and Must be
Included
In 2011, EPA‟s Office of Research and Development (ORD) developed an Integrated Risk
Information System (IRIS) RfC for TCE derived from oral studies and PBPK modeling to
conduct route-to-route modeling. ORD considered this RfC to be a robust value for inhalation
exposure. The RfC represents an estimate for the continuous inhalation exposure to the human
population (including sensitive subpopulations) that is unlikely to have deleterious effects during
a lifetime. EPA, in conducting this assessment of TCE chose not to use the 2011 IRIS RfC.
Instead, EPA used other inhalation studies contained in the IRIS file to develop new, preferred
points of departure and HEC levels. EPA did not clearly articulate why this approach was
adopted, but the executive summary of the assessment alludes to the fact that EPA chose this
approach based on its desire to use inhalation studies as opposed to oral studies. However, it is
troubling that EPA, in its risk assessment of TCE, discounts the IRIS RfC value solely because
the underlying studies did not use the inhalation exposure route. As TCE is well absorbed and
the PBPK models used were peer reviewed and validated, EPA should offer a clear and complete
justification as to why the IRIS RfC does not represent the best available science to evaluate
inhalation exposures.
The hazard evaluation in the draft TCE assessment is also of concern because it appears to rely
on studies based solely on the fact that the route of exposure was inhalation. Beyond this, EPA
does not appear to evaluate the quality, reliability, and appropriateness of each study, but instead
appears to treat them all as equal. A systematic evaluation of the quality (including relevance
and reliability) of each study is necessary. Without such an evaluation, studies of lower quality
are inappropriately accorded too much weight in the overall assessment, leading to a flawed
evaluation. It is critical that EPA rely on the studies that are of the highest quality, not simply
those studies that produce the lowest points of departure. Examples of data quality concerns,
stemming from an insufficient evaluation of study quality, include the following:
EPA relies on kidney toxicity endpoints to create the lowest HEC values used in the draft
assessment. However the EPA Science Advisory Board (SAB) during its review of the
IRIS assessment in 2011 noted that the kidney effects should not be the primary basis for
the RfC.24
The IRIS document itself mentions concerns with the Woolhiser et al (2006)
MOE is considered “safe” and EPA’s Office of Pesticide Programs (OPP) has used the term “Target” MOE (e.g., EPA, 2011). Use of the term ‘Level of Concern’ is poor risk communication and conveys a conclusion about risk (public and/or individual) that is not consistent with the risk communication associated with the use of the EPA reference values or other values such as the National Academies AEGL. EPA should consider using a ‘minimal’ or ‘target’ MOE instead in the tables or simply report MOE and use the text to discuss the minimal MOE associated with each guidance value (as the EPA OPP has employed). 24
The 2011 SAB report states: “Although recognizing the kidney hazards of TCE, the Panel was concerned about the use of three candidate RfD/RfCs [referring to toxic nephropathy (NTP 1988); toxic nephrosis (NCI 1976);
ACC Appendix A 14 | P a g e
study, which is one of the studies relied upon in the draft assessment to develop values
for the kidney endpoint.25
EPA relies on the Chia et al (1996) study for the determination of male reproductive
effects. However, the findings were within the normal range for the sperm density
endpoint, even in the high exposure group. Consequently, there is uncertainty as to
whether these effects are adverse, and the EPA IRIS file noted the uncertainties in the
characterization of exposure and the adversity of the effect.26
EPA relies on Arito et al. (1994) for the acute hazard value for neurotoxicity. The effect
in this study was time spent in wakefulness. The TCE IRIS document describes this study
as having lower confidence than other neurotoxicity studies. It is unclear why EPA is
relying on this study to generate and HEC value to represent neurotoxicity when ORD
found it be a low confidence study.
EPA relies on Healy et al. (1982) for the acute hazard value for developmental toxicity.
The findings in this study are not consistent with the findings in other large, well-
designed inhalation studies. Unfortunately, EPA relies on this study in isolation without
sufficiently evaluating other inhalation studies that looked at the same endpoint.27
The concerns created by EPA‟s lack of a quality evaluation for each study relied upon are
compounded by EPA‟s use of the same MOE regardless of study quality. A separate MOE
should be used for each hazard value based on the uncertainty in the study from which the hazard
value is derived. Other factors, such as the population of concern, should also be considered
when setting an appropriate MOE.
increased kidney weights (Woolhiser et al. 2006) based on kidney effects as the primary basis for the RfD and RfC because of uncertainties regarding the relative rate of formation of toxic metabolites in humans vs. animals.” 25
The 2011 IRIS file states: “For kidney effects (U.S. EPA 2011, Section 5.1.2.2), there is high confidence in the evidence for a nephrotoxic hazard from TCE. Moreover, the two lowest candidate RfDs for kidney effects (toxic nephropathy [referring to an oral NTP 1988] and increased kidney weight [referring to Woolhiser et al. (2006) here]) are both based on benchmark dose (BMD) modeling and one is derived from a chronic study. However, as discussed in U.S. EPA (2011), Section 3.3.3.3, there remains substantial uncertainty in the PBPK model-based extrapolation of glutathione (GSH) conjugation from rodents to humans due to limitations in the available data. In addition, the candidate RfD for toxic nephropathy had greater dose-response uncertainty since the estimation of its POD involved extrapolation from high response rates (>60%). Therefore, kidney effects are considered supportive but are not used as a primary basis for the RfD.” U.S.EPA 2011, with bracketed text and references added. 26
The 2011 IRIS file states: “For the human study by Chia et al. (1996), as discussed above, there are uncertainties in the characterization of exposure and the adversity of the effect measured in the study.” 27
EPA should also consider the results from Dorfmueller et al. 1979, Westergren et al. 1984, and Carney et al. 2006.
ACC Appendix A 15 | P a g e
N-methylpyrollidone (NMP):
EPA inappropriately uses developmental toxicity endpoints in the acute assessment
In the draft NMP assessment, EPA acknowledges that BMD modeling is the preferred approach
for dose response assessment, but also presents the NOAEL data as a comparison.28
While
presenting alternative approaches typically helps to inform judgments, in this case the EPA
rationale is not clear. The advantage of the BMD approach over the NOAEL approach is that it
uses the entire dose-response curve with respect to the quantitative fetal body weight data and
does not rely on the arbitrary selection of doses by the researcher. Where the data exist, and
where the BMD modeling can be conducted in a manner where model fit is acceptable and
scientifically justified, EPA should focus on using the BMD approach for developing PODs.
More puzzling however, is EPA‟s decision to use developmental toxicity endpoints in the acute
risk analysis. NMP has low systemic toxicity and developmental toxicity (fetal weight
reductions) was determined by EPA to be the most sensitive toxicity endpoint. Therefore, in the
NMP draft assessment, EPA uses developmental toxicity data to calculate MOEs for
occupational and residential exposures. While this approach makes sense for chronic and
subchronic exposures, caution should be exercised when using developmental toxicity to assess
the risk of acute exposures, as that approach may not be valid for all developmental toxicity
endpoints (van Raaij et al., 2003).29
The draft assessment bases the use of developmental toxicity endpoints to evaluate acute
exposures and risks on the fact that a single exposure at a developmentally critical period can
cause developmental effects (Section 3.2.4, p. 48). The concept of a “critical window” for
developmental processes is intended to apply to structural malformations and is underpinned by
the fact that organs form in very discrete time periods and, once formed, their structure is not
easily altered (Hood, 2006).30
Fetal growth, on the other hand, occurs throughout gestation
rather than during a discrete developmental window. Effects on fetal growth are also potentially
reversible or partially reversible with removal of the toxic insult (Chernoff et al., 2008; van Raiij
28
There appears to be an error in the way that the NOAEL value for developmental toxicity is time-scaled in Table 3-10 (p. 53) and Table 3-12 (p. 56). The NOAEL value of 243 mg/m
3 (Saillenfait et al., 2003) for a duration of 6
hrs/day, 7 days/week was scaled to a value of 61 mg/m3 for a chronic 24 hrs/day exposure. However, this value
should have been further scaled to 364.5 mg/m3 in the two tables to represent an acute 4 hr exposure, as was
done for the BMCLHEC value (i.e., 302 mg/m3 for 6 hrs/d, 76 mg/m
3 for 24 hrs/d, and 453 mg/m
3 for 4 hrs per day).
While the NOAEL value is used for comparison purposes only, it should be corrected if maintained in the final document. 29
See van Raaij MTM, Janssen PAH, Piersma AH (2003) The Relevance of Developmental Toxicity Endpoints for Acute Limit Setting. RIVM (Dutch National Institute of Health and the Environment) Report 601900004. 30
See Hood RD (2006) Principles of Developmental Toxicology Revisited, Developmental and Reproductive Toxicology: A Practical Approach, Ed., RD Hood, Taylor and Francis, Boca Raton, p. 7.
ACC Appendix A 16 | P a g e
et al., 2003).31
In that context, an effect on fetal growth could be considered more similar to
maternal body weight change or other chronic toxicity endpoints than to frank malformations in
that it may require repeated dosing to be manifested, particularly in the absence of other
developmental signals (van Raiij et al., 2003).
It is important to note that in the pivotal inhalation developmental toxicity study with NMP
(Saillenfait et al., 2003),32
the main developmental effect was a small decrement (5%) in fetal
weight at the high dose which occurred after 15-days of dosing (gestation day (GD) 6-20). A 5%
reduction in fetal weight is the smallest decrement that can be reliably detected as different from
control weights (Chernoff et al., 2008). The Saillenfait study (2003) also showed decreased
maternal body weight gain at the high- and mid-doses and decreased food consumption, which
could have been a potential contributing factor to decreased fetal body weight in that study
(Chernoff et al., 2008; Fleeman et al., 2005).33
Likewise, in the pivotal dermal developmental
toxicity study with NMP (Becci et al., 1982), a significant decrease in maternal body weight gain
occurred throughout gestation at the highest dose level along with a reduction in fetal body
weights, skeletal changes indicative of immaturity, and a slight increase in resorptions.34
An inhalation study by Lee and colleagues (1987) with a shorter treatment period and a slightly
reduced dose did not show a reduction in fetal weight or maternal body weight, while inhalation
studies of longer duration showed more significant reductions in pup weight extending into
postnatal life (Solomon et al., 1995).35
These observations are consistent with the idea that fetal
weight reductions may require extended dosing periods to develop, and this reason may not be
appropriate to use for acute exposure risk assessments.
The NMP draft cites a publication from the Dutch Institute of Health and the Environment
(RIVM) (van Raiij et al., 2003) to support the use of fetal body weight data to set acute exposure
standards. In fact, the intent of the van Raiij publication is to caution against the use of non-
malformation developmental toxicity data (such as fetal weight) in the setting of acute exposure
limits. The concern stated in the NIVM publication was that an acute exposure limit based on
developmental toxicity data such as fetal weight could lead to an overly conservative ARfD or
AEGL, resulting in, for example, unnecessary emergency responses. It was also noted that
31
See Chernoff N, Rogers EH, Gage MI, Francis BM (2008) The Relationship of Maternal and Fetal Toxicity in Developmental Toxicology Bioassays with notes on the biological significance of the “no observed adverse effect level.” Repro Toxicol, 25: 192-202 and see footnote 29. 32
Saillenfait AM, Gallissot F, Morel G (2003) Developmental toxicity of N-methyl-2-pyrrolidone in rats following inhalation exposure. Food Chem Toxicol, 41:583-588. 33
See footnote 25 and Fleeman TL, Cappon GD, Chapin RE, Hurtt ME. Effects of Feed Restriction during Organogenesis on Embryo-Fetal Development in the Rat. Birth Defects Res B Dev Reprod Toxicol. 2005; 74:442–9. 34
See Becci PJ, Knickerbocker MJ, Reagan EL, Parent RA, Burnette LW (1982) Teratogenicity Study of N-methylpyrrolidone after Dermal Application to Sprague-Dawley rats. Fundam Appl Toxicol, 2:73-76. 35
See Lee KP, Chromey NC, Culik R, Barnes JR, Schneider PW (1987) Toxicity of N-methyl-2-pyrrolidone (NMP): Teratogenic, Subchronic, and Two-Year Inhalation Studies. Fundam Appl Toxicol, 9:222-35 and Solomon HM, Burgess BA, Kennedy GL Jr, Staples RE (1995) 1-Methyl-2-pyrrolidone (NMP): Reproductive and Developmental Toxicity Study by Inhalation in the Rat. Drug Chem Toxicol, 18:271-93.
ACC Appendix A 17 | P a g e
exposure durations in guideline-based developmental toxicity studies cover a large portion of
gestation in experimental animals, but a much smaller percentage of gestation in humans. This
was compared with a single dose exposure, which might cover less than 1% of gestation in
humans (van Raiij et al., 2003).
In order to address the question of whether fetal body weight or other developmental endpoints
were appropriate for assessment of acute exposures, van Raiij and colleagues (2003) compared
published NOAELs and LOAELs for repeated dose vs. single dose administration. For fetal
weight, the data were variable, and they concluded that, “The relevance of fetal body weight as
an endpoint for acute limit setting should therefore be evaluated within the total context of
effects including maternal toxicity and other fetal effects.” This sentiment was echoed by
Billington and Carmichael (2000) who suggest that embryo-fetal survival and structural
malformation data could be relevant to setting acute reference doses for pesticides in food
residues, but that decrements in maternal or fetal body weight data may not be relevant unless
they are shown to be altered after a single dose.36
Given the lack of teratogenic effects elicited by NMP in developmental toxicity studies, the
minor decrement in fetal weight observed after repeated dosing, and the association of fetal
weight decrements with maternal toxicity, developmental NOAELs based on fetal weight
changes are not considered appropriate for setting acute exposure limits for NMP. EPA,
therefore, should not rely on this endpoint to set acute exposure levels.37
This would also be
more consistent with statements in the Executive Summary of the NMP draft assessment which
indicates that there is a low concern for reproductive toxicity with NMP based on recent
information.
Dichloromethane (DCM):
EPA needs to justify its choice of hazard values
In the DCM draft assessment EPA has not clearly justified the use of one hazard value over
another. For example, EPA has chosen to use the CalEPA acute REL instead of its own AEGL-1
36
See Billington, R, Carmichael, N (2000) Setting of acute reference doses for pesticides based on existing regulatory requirements and regulatory test guidelines. Food Additives Contaminants, 17:621-626. 37
We also note that the ECHA Guidance on information requirements and chemical safety assessment Chapter R.8: “Characterisation of dose *concentration+-response for human health” 2012 states at page 8: “The establishment of an acute toxicity DNEL set for effects occurring after a single exposure of a few minutes up to 24 hours is not only cumbersome (there is no established consensus methodology) and resource-intensive but probably unnecessary, as the long-term DNEL is normally sufficient to ensure that these effects do not occur. It is therefore proposed that if an acute toxicity hazard (leading to C&L) has been identified, a DNEL for acute toxicity is only established for the effects of peak exposures as these peaks can be significantly higher than the average daily exposure and the long-term DNEL (to be complied with on average over e.g. a working day) may be insufficient to limit them. Overall, therefore, a DNEL for acute toxicity should be derived if an acute toxicity hazard (leading to C&L) has been identified and there is a potential for high peak exposures, for instance when sampling or connecting/disconnecting vessels.” Document available at: http://echa.europa.eu/documents/10162/13632/information_requirements_r8_en.pdf.
ACC Appendix A 18 | P a g e
values, without providing any scientific justification. It appears that EPA has simply made a
policy choice to select lower (more conservative) exposure guidance values. As a result of this
apparent policy decision, the science basis of the draft risk assessment for DCM suffers. The
CalEPA REL reflects a relatively simplistic evaluation of dose-response assessed in terms of
administered concentration that is more uncertain (net uncertainty factor = 60) than the more
sophisticated application of PBPK modeling used to derive EPA‟s AEGL values (net uncertainty
factors of 1-3). The CalEPA basis for the REL (Putz et al., 1979) was included in EPA AEGL
documentation, but ultimately was not adopted as the primary basis for the AEGL value.
Furthermore, the Putz et al. (1979) study involved human volunteers exposed to a single
inhalation exposure challenge to DCM, almost identical to the study used as the basis of the
AEGL-1. Therefore, the acute REL does not have any more validity for assessing repeat (or
intermittent) acute exposures than the AEGL-1 value. If EPA had applied a systematic review
approach to evaluating the evidence, it is not clear that the outcome would have supported use of
the CalEPA value. We recommend that EPA apply such an approach to justify its choice of
hazard values.
4. The assessments should explicitly address the authority of the Occupational Safety and
Health Administration (OSHA) in regulating occupational exposures and should strive
to achieve consistency with OSHA’s approach to assessing safety.
OSHA is the authority to prescribe or enforce regulations affecting occupational safety and
health. It is critically important that EPA policies and scientific approaches are coordinated with
OSHA‟s and not in conflict. As EPA evaluates occupational exposures scenarios, there is
significant potential for confusion and conflict with OSHA standards if those standards and
OSHA‟s role is not specifically acknowledged in the assessments.
The Agency‟s duty to consult and coordinate with other agencies is outlined in §9(a) of TSCA
and §9(c) stipulates that EPA “shall not be … deemed to be exercising statutory authority to
prescribe or enforce standards or regulations affecting occupational safety and health.”
The assessments for DCM, MNP, and TCE consider potential risks to workers without
addressing the occupational controls in place as a result of a worker standard, Permissible
Exposure Limit (PELs), or for personal protective equipment (PPE) requirement established by
the Occupational Safety and Health Administration (OSHA) under the 1970 OSH Act.
ACC Appendix A 19 | P a g e
Chemical Specific Comments
Trichloroethylene (TCE):
In the TCE assessment at Table 3-21, EPA presents exposure values ranging from 2-63 ppm,
which are then determined to be a potential risk concern. Each of these values is below the
OSHA permissible exposure limit (PEL) of 100 ppm. Further clarity is needed from EPA
regarding how these MOEs should be interpreted in light of the OSHA standards.
Dichloromethane (DCM):
In the draft DCM assessment, EPA should consider using the OSHA PEL for DCM for
evaluating risks for workers exposed to DCM. Since EPA choses to use the Spacecraft
maximum allowable concentration (SMAC) and CalEPA RELs for evaluating risks, it is not
clear why OSHA values are excluded from consideration. In the case of DCM, OSHA adopted a
standard under §6(b)(5) of the OSH Act in 1997 lowering the workplace exposure limit to 25
parts per million (ppm) as an eight-hour time-weighted average (TWA) and establishing a short-
term exposure limit (STEL) of 125 ppm and an action level for concentrations of airborne DCM
of 12.5 ppm.38
More clarity is needed regarding how these assessments intersect with advice and
standards provided by OSHA.
In the draft assessment, EPA includes exposures for commercial paint stripping operations that
predate the OSHA standard, in some cases by as much as 20 years. Unfortunately, the
assessment makes no attempt to eliminate the measurements prior to the implementation of the
OSHA standard and makes only a general statement that exposures may have declined as a result
of the standard.
N-methylpyrollidone (NMP):
Although no exposure limits have been set for NMP, OSHA regulations39
require the use of
appropriate hand protection when employees are exposed to hazards from skin adsorption like
those with NMP. These concerns are discussed more fully under section 2 of this Appendix.
38
29 CFR § 1910.1052; 62 Fed. Reg. 1494 (January 10, 1997). Development of the OSHA standard was initiated following an EPA review of potential methylene chloride risks under TSCA §4(f) in 1985.
39 29 CFR 1910.138, Personal Protective Equipment – Hand Protection
Appendix B
1 | ACC Appendix B
Appendix B (ATO):
ACC appreciates the opportunity to provide comments on the draft charge questions. This is a
very important step in the stakeholder engagement process. A complete and thorough set of
charge questions is critical to ensuring that a high quality peer review is conducted. Well-
constructed charge questions help resolve specific questions and concerns stakeholders have
identified.
The redline edits provided below should help ensure that the peer review process resolves
different scientific perspectives and fosters explicit discussion of identified concerns. Addressing
these concerns is an important part of a robust and rigorous peer review.
OPPT Proposed Draft Charge to External Peer Reviewers for the review of the TSCA Work
Plan Chemical Risk Assessment of ATO
December 2012
This assessment evaluated the environmental risks that may be associated with ATO use as a
synergist in halogenated flame retardants. Human health risks were determined to be of low concern;
the available hazard data are summarized in an appendix.
Issue 1. Overall Clarity of the Assessment. Chapter 1 provides the scope of the assessment and
a brief introduction. Supporting information on chemistry, fate, and uses are provided in
Chapter 2 and the exposure, hazard, and risk characterizations are presented in Chapter 3.
Additional supporting information is available in the appendices.
Question 1-1. Please comment on the clarity and strengths and weaknesses of the risk
assessment and provide specific suggestions regarding how this may be improved. Is the scope
and problem that the assessment addressed clearly defined?
Question 1-2. Please comment on the overall objectivity of the assessment. In particular,
comment on whether the conclusions are presented in an accurate and unbiased manner that
accurately reflects the assessment that was conducted.
Question 1-3. Please comment on the overall approach the agency has taken to evaluate data
adequacy and study reliability throughout this assessment.
Question 1-4. EPA has not characterized this assessment as a screening-level evaluation. Please
comment on whether such a descriptor should be considered. If such a descriptor is appropriate,
please describe the specific areas where further refinement is necessary.
Issue 2. Characterization of Environmental Exposures Based on Release Data. During project
scoping, OPPT identified ATO use as a synergist in halogenated flame retardants as the focus
of this assessment. EPA’s 2010 Toxic Release Inventory (TRI) was used to obtain information
on water releases associated with this end-use scenario. Data collection was refined using the
North American Industry Classification System (NAICS) codes to identify a subset of TRI
facilities (i.e., those indicating production, processing or use of ATO-containing flame
retardants). Because ATO is not specifically listed on the TRI, data reported under the broader
Appendix B
2 | ACC Appendix B
category of ‘antimony compounds’ were used as a surrogate for ATO in this assessment. ATO
surface water concentrations were predicted using a screening level tool (E-FAST2), to model
water releases reported by selected TRI facilities.
Question 2-1: Please comment on EPA’s decision to focus on the use of ATO as a synergist in
halogenated flame retardants. Is EPA’s rationale presented objectively and do you agree that this
use is the appropriate choice for this assessment?
Question 2-2: TRI information reported for ‘antimony compounds’ was used to estimate ATO
releases associated with its use as a synergist in halogenated flame retardants. Please comment on
this approach. In particular, please comment on EPA’s finding that this approach would likely
lead to an overestimation of exposure potential.
Question 2-23: Because TRI does not indicate the number of days associated with reported
releases, two exposure scenarios were developed (assuming the total water releases reported in
the 2010 TRI occurred over a period of 250 days or, a more conservative scenario of 24 days to
provide a range of predicted water concentrations for comparison with hazard benchmarks
(concentrations of concern) identified for aquatic organisms. Please comment on the assumptions
used to develop modeling scenarios using EPA’s E-FAST2 model to assess aquatic exposures to
ATO. Please specifically address whether these assumptions are reasonably expected exposures
or are likely overestimates.
Question 2-34: Please comment on the approach EPA used to estimate environmental releases
resulting from uses of ATO as a synergist in flame-retardant chemicals. Are there other data
sources or approaches that EPA should consider to estimate environmental releases of ATO in
this end-use? If so, please provide citations or data for consideration in further revision of the
draft assessment.
Issue 3. Characterization of Environmental Exposures Based on Monitoring Data. OPPT
used environmental monitoring information obtained from the U.S. Geological Survey-
National Water Information System and EPA's STORET database to evaluate antimony
levels in environmental media. Data collection was limited to states (n=10) with TRI
facilities (n=14) having NAICS codes corresponding to ATO uses as a synergist in
halogenated flame retardants.
Question 3-1: Please comment on the use of these large monitoring data sets to characterize
ecological exposures to ATO use as a synergist in flame retardants and their representativeness
for other locations in the US.
Question 3-2: Are there other major sources of environmental monitoring data (or other pertinent
information) that EPA should incorporate in the exposure assessment? If so, please provide the
necessary citations and/or data for inclusion in the revised document.
Question 3-3: Are there concerns or limitations in these data sets that may impact their utility for
risk assessment?
Issue 4. Fate, Transport and Bioavailability. Information available in the published literature
regarding the chemistry, fate and transport of ATO is used qualitatively to assess
Appendix B
3 | ACC Appendix B
bioavailability to ecological organisms. There is a lack of site-specific data on geochemistry that
would inform specification and availability antimony compounds in its toxic forms for
environmental receptors.
Question 4-1: Please comment on the use of this information in the ATO assessment. Are there
other data sources that EPA should consider?
Issue 5. Environmental Hazard Assessment. The available hazard information was critically
evaluated based on specific test guidelines, accepted endpoints used to assess ecotoxicity, and
the amount of detail provided in each study report. Hazard benchmarks (i.e., concentrations of
concern) were subsequently derived using the most robust ecotoxicity studies and conservative
ecotoxicity values identified in the published literature. Acceptable toxicity data were not
available for ATO for both chronic and acute exposures in all media. For this reason, toxicity
data for antimony trichloride is used to characterize hazards to water, soil, and sediment
dwelling organisms. This is not expected to significantly impact the findings of this assessment
because (1) upon dissolution, antimony compounds release antimony ions, and it is the fate and
toxicity associated with the total antimony ion concentration that is of most importance when
assessing the toxicity of antimony in environmental media and (2) both the oxide and chloride
salts of antimony produce comparable amounts of antimony ions upon dissolution in water.
Question 5-1: Please comment on the approach the agency has taken to evaluate data adequacy,
study reliability, and data acceptability. In particular, are the criteria clearly described and
appropriate?
Questions 5-2: What other factors should EPA consider in evaluating the potential risks of
concern for ecological organisms from antimony species? Please comment on the use of toxicity
data for antimony trichloride to characterize hazards to water, soil, and sediment dwelling
organisms.
Issue 6. Environmental Risk Characterization. The ATO assessment evaluates risks of
concern posed to ecological organisms as a result of ATO use as a synergist in halogenated
flame retardants. Generally speaking, risks are indicated when antimony levels in
environmental media (as indicated by environmental monitoring and industrial release
information) exceed the hazard benchmarks (i.e., concentrations of concern) identified for
ecological organisms in water, soil and sediment. This approach resulted in very few
instances where the concern concentrations for water or sediment dwelling organisms were
exceeded (< 1%). No exceedances of the hazard benchmarks for soil dwelling organisms
were identified. The uncertainties/limitations of this approach are discussed in the ATO
document.
Question 6-1: Please comment on the implicit assumption that antimony concentrations
measured in environmental media reflect many different inputs (i.e., from various end use
applications and other types of antimony compounds in addition to ATO) and how this could
impact the risk estimation.
Question 6-2: The findings reported in this assessment hinge on the assumption that a
‘conservative’ scenario has been presented (based on the assumption that all releases of various
types of antimony compounds have been attributed to ATO use in flame retardants) and as such,
Appendix B
4 | ACC Appendix B
reflect a conservative estimate of risk of exposure to ATO. Please comment on validity of this
assumption and the likelihood that the actual risks of concern posed to ecological organisms have
been over (or under) estimated. Please comment on any other assumptions, including exposure
modeling assumptions, that may impact the likelihood of the actual risks being over (or under)
estimated.
Question 6-3: Please comment on the reasonableness of the final estimates.
Question 6-34: Please comment on the data set used to evaluate exposures to soil-dwelling
organisms and the limitations and/or uncertainties in estimating risk to soil-dwelling organisms.
Please provide comment on additional data sources, surrogate/related chemicals, or approaches to
estimate risks under data poor conditions?
Issue 7. Human Health Hazard Assessment. Human health risks were determined to be of low
concern; the available hazard data are summarized in an appendix.
Question 7-1. : Please comment on the accuracy, objectivity, and utility of the information
presented in the appendix. Please also suggest any studies that should be included in this
appendix.