UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON D.C. 20460
OFFICE OF THE ADMINISTRATOR
SCIENCE ADVISORY BOARD
January 3 0,2013
EPA-SAB-13-001
The Honorable Lisa P. Jackson
Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, N.W.
Washington, DC 20460
Subject: Review of EPA' s Draft Assessment entitled Toxicological Review of Libby Amphibole
Asbestos (August 201 1)
Dear Administrator Jackson:
EPA's Office of Research and Development (ORD) requested the Science Advisory Board (SAB) to
conduct a peer review of EPA's draft Integrated Risk Information System (IRIS) assessment, entitled
Toxicological Review of Libby Amphibole Asbestos (August 2011). The draft document is the first IRIS
assessment specific to Libby Amphibole asbestos (LAA), a term used to refer to the mixture of
amphibole mineral fibers identified in the Rainy Creek complex near Libby, Montana. The SAB was
asked to comment on the scientific soundness of the hazard and dose-response assessment of LAA-
induced cancer and non-cancer health effects.
The SAB finds the EPA's draft assessment to be comprehensive and generally clear, logical and well-
written. There are several areas that need more consideration, and we provide recommendations to
further enhance the clarity and strengthen the scientific basis for the conclusions presented. The SAB
responses to the EPA's charge questions are detailed in the enclosed report. The SAB's major comments
and recommendations are provided below:
Localized pleural thickening is an appropriate health endpoint for the derivation of the inhalation
reference concentration (RfC). It is an irreversible structural, pathological alteration of the pleura
and is generally associated with reduced lung function. The SAB has identified additional
references and recommends that the agency include a more detailed review of the literature to
further support this conclusion.
The SAB supports the derivation of an RfC for LAA based on radiographic evidence of
localized pleural thickening in an occupationally exposed Marysville, Ohio, cohort. However,
the SAB recommends that the EPA conduct additional analyses to substantiate the RfC (to the
extent data permit) of pleural abnormalities using the recently published studies on two other
cohorts.
-------
The SAB recommends that more justification be provided for the selection of the "best" model
for non-cancer exposure-response analysis. The SAB also recommends examining other
exposure metrics besides the simple cumulative exposure, such as time-weighting of exposures.
In addition, more justification is needed for the selection of 10 percent extra risk as the
benchmark response since it is not consistent with the guideline for epidemiological data in
EP'A''s Benchmark Dose Technical Guidance.
A composite uncertainty factor of 100 was applied to the point of departure to obtain the RfC.
EPA applied an uncertainty factor of 10 to account for human variability and sensitive
subpopulations, and a database uncertainty factor of 10 to account for database deficiencies in
the available literature for the health effects of LAA. The SAB recommends that the EPA re-
evaluate the use of a default database uncertainty factor of 10 as part of the consideration of
additional studies; additional data (e.g., Minnesota cohort and data on other amphiboles) might
support a lower value, such as 3, for the database uncertainty factor. In addition, the SAB
recommends EPA re-visit its judgement of a subchronic-to-chronic uncertainty factor and a
LOAEL-to-NOAEL uncertainty factor of 1-fold.
The SAB agrees that the weight of evidence for LAA supports the descriptor "Carcinogenic to
Humans by the Inhalation Route" in accordance with EPA's Guidelines for Carcinogen Risk
Assessment. The SAB views the mode of carcinogenic action of LAA as complex, and
recommends that the agency conduct a formal mode of action analysis in accordance with EPA's
Guidelines for Carcinogen Risk Assessment. Based on this formal analysis, the agency may still
conclude that the default linear extrapolation at low doses is appropriate.
The SAB supports the selection of the Libby worker cohort for the derivation of the inhalation
unit risk (IUR) and agrees that the use of the subcohort post-1959 for quantification may be
reasonable due to the lack of exposure information for many of the workers in earlier years. The
SAB has suggested sensitivity analyses that would explore the implications of the selection of
the subcohort. The SAB finds it appropriate to use lung cancer and mesothelioma as endpoints
for the derivation of the IUR. The SAB recommends a more detailed discussion and justification
of how the use of mortality data rather than incidence data may have resulted in an undercount
of cases of lung cancer and mesothelioma and what implications, if any, it may have for the
derivation of the IUR.
The draft assessment clearly described the methods selected to conduct the exposure-response
modeling for lung cancer and mesothelioma. However, the SAB recommends that the agency
provide more support for its choice of statistical models for the exposure-response analysis. The
SAB also recommends consideration of several models in addition to the Poisson and Cox
models used in the draft assessment.
The agency has been overly constrained by reliance on model fit statistics as the primary
criterion for model selection. The SAB recommends graphical display of the fit to the data for
both the main models and for a broader range of models in the draft document to provide a more
complete and transparent view of model fit. The SAB also recommends that the EPA consider
literature on epidemiological studies of other amphiboles for model selection for dose-response
assessment, since the size of the Libby subcohort used in the exposure-response modeling is
small.
-------
The EPA has summarized many sources of uncertainty, sometimes quantitatively, as well as the
direction and magnitude of the likely impact of each source of uncertainty. The SAB
recommends that model uncertainty be evaluated by estimating risks using a more complete set
of plausible models for the exposure-response relationship. This sensitivity analysis, while not a
full uncertainty analysis, would make explicit the implications of these key model choices.
Finally, the SAB has identified critical research needs for epidemiological studies, mode of
action, and measurement methods for LAA to strengthen future LAA assessment.
The SAB appreciates the opportunity to provide the EPA with advice on this important subject. We
look forward to receiving the agency's response.
Sincerely,
/signed/
Dr. David T. Allen, Chair
Science Advisory Board
/signed/
Dr. Deborah L. Swackhamer, Immediate Past Chair
Science Advisory Board
/signed/
Dr. Agnes Kane, Chair
SAB Libby Amphibole Asbestos Review Panel
Enclosure
-------
NOTICE
This report has been written as part of the activities of the EPA Science Advisory Board, a public
advisory committee providing extramural scientific information and advice to the Administrator and
other officials of the Environmental Protection Agency. The Board is structured to provide balanced,
expert assessment of scientific matters related to problems facing the Agency. This report has not been
reviewed for approval by the Agency and, hence, the contents of this report do not necessarily represent
the views and policies of the Environmental Protection Agency, nor of other agencies in the Executive
Branch of the Federal government, nor does mention of trade names or commercial products constitute a
recommendation for use. Reports of the EPA Science Advisory Board are posted on the EPA website at
http ://www. epa.gov/sab.
-------
U.S. Environmental Protection Agency
Science Advisory Board
Libby Amphibole Asbestos Review Panel
CHAIR
Dr. Agnes Kane, Professor and Chair, Department of Pathology and Laboratory Medicine, Brown
University, Providence, RI
MEMBERS
Dr. John R. Balmes, Professor, Department of Medicine, Division of Occupational and Environmental
Medicine, University of California, San Francisco, CA
Dr. James Bonner, Associate Professor, Toxicology, North Carolina State University, Raleigh, NC
Dr. Jeffrey Everitt, Director, Department of Laboratory Animal Science, GlaxoSmithKline
Pharmaceuticals, Research Triangle Park, NC
Dr. Scott Person, Senior Scientist, Applied Biomathematics, Setauket, NY
Dr. George Guthrie, Focus Area Leader, Geological and Environmental Sciences, National Energy
Technology Laboratory, U.S. Department of Energy, Pittsburgh, PA
Mr. John Harris, Principal, LabCor Portland, Inc, Portland, OR
Dr. Tom Hei, Professor and Vice-Chairman, Radiation Oncology, College of Physicians and Surgeons,
Columbia University Medical Center, New York, NY
Dr. David Kriebel, Professor and Chair, Dept. of Work Environment, School of Health & Environment,
University of Massachusetts Lowell, MA
Dr. Morton Lippmann, Professor, Nelson Institute of Environmental Medicine, New York University
School of Medicine, Tuxedo, NY
Dr. John Neuberger, Professor, Preventive Medicine and Public Health, School of Medicine,
University of Kansas, Kansas City, KS
Dr. Lee Newman, Professor of Medicine, Division of Environmental and Occupational Health
Sciences, School of Public Health, University of Colorado, Aurora, CO
Dr. Michael Pennell, Assistant Professor, Division of Biostatistics, College of Public Health, Ohio
State University, Columbus, OH
Dr. Julian Peto, Professor, Department of Epidemiology and Population Health , London School of
Hygiene and Tropical Medicine, London, UK
Dr. Carrie Redlich, Professor of Medicine, Internal Medicine, School of Medicine, Yale University,
New Haven, CT, United States
ii
-------
Dr. Andrew G. Salmon, Senior Toxicologist, Office of Environmental Health Hazard Assessment,
California Environmental Protection Agency, Oakland, CA
Dr. Elizabeth A. (Lianne) Sheppard, Professor, Biostatistics and Environmental & Occupational
Health Sciences, School of Public Health, University of Washington, Seattle, WA
Dr. Randal Southard, Professor of Soils, AES Dean's Office, University of California at Davis, Davis,
CA
Dr Katherine Walker, Senior Staff Scientist, Health Effects Institute, Boston, MA
Dr. James Webber, Research Scientist, Wadsworth Center, New York State Department of Health,
Albany, NY
Dr. Susan Woskie, Professor, Work Environment, Health and Environment, University of
Massachusetts Lowell, Lowell, MA
SCIENCE ADVISORY BOARD STAFF
Dr. Diana Wong, Designated Federal Officer, U.S. Environmental Protection Agency, Washington, DC
in
-------
U.S. Environmental Protection Agency
Science Advisory Board
FY 2012
CHAIR
Dr. David T. Allen, Gertz Regents Professor of Chemical Engineering and the Director of the Center
for Energy and Environmental Resources, Department of Chemical Engineering, University of Texas,
Austin, TX (Chair since November 2012)
Dr. Deborah L. Swackhamer, Professor and Charles M. Denny, Jr. Chair in Science, Technology and
Public Policy, Hubert H. Humphrey School of Public Affairs and Co-Director of the Water Resources
Center, University of Minnesota, St. Paul, MN (Chair until September 2012)
MEMBERS
Dr. George Alexeeff, Acting Director, Office of Environmental Health Hazard Assessment, California
Environmental Protection Agency, Oakland, CA
Dr. Pedro Alvarez, Department Chair and George R. Brown Professor of Engineering, Department of
Civil & Environmental Engineering, Rice University, Houston, TX
Dr. Joseph Arvai, Svare Chair in Applied Decision Research, Institute for Sustainable Energy,
Environment, & Economy, Haskayne School of Business, University of Calgary, Calgary, Alberta,
Canada
Dr. Claudia Benitez-Nelson, Full Professor and Director of the Marine Science Program, Department
of Earth and Ocean Sciences, University of South Carolina, Columbia, SC
Dr. Timothy J. Buckley, Professor and Chair, Division of Environmental Health Sciences, College of
Public Health, The Ohio State University, Columbus, OH
Dr. Patricia Buffler, Professor of Epidemiology and Dean Emerita, Department of Epidemiology,
School of Public Health, University of California, Berkeley, CA
Dr. Ingrid Burke, Director, Haub School and Ruckelshaus Institute of Environment and Natural
Resources, University of Wyoming, Laramie, WY
Dr. Thomas Burke, Professor, Professor and Jacob I and Irene B. Fabrikant Chair in Health, Risk and
Society Associate Dean for Public Health Practice, Johns Hopkins Bloomberg School of Public Health,
Johns Hopkins University, Baltimore, MD
Dr. Terry Daniel, Professor of Psychology and Natural Resources, Department of Psychology, School
of Natural Resources, University of Arizona, Tucson, AZ
Dr. George Daston, Victor Mills Society Research Fellow, Product Safety and Regulatory Affairs,
Procter & Gamble, Cincinnati, OH
Dr. Costel Denson, Managing Member, Costech Technologies, LLC, Newark, DE
iv
-------
Dr. Otto C. Doering III, Professor, Department of Agricultural Economics, Purdue University, W.
Lafayette, IN
Dr. Michael Dourson, President, Toxicology Excellence for Risk Assessment, Cincinnati, OH
Dr. David A. Dzombak, Walter J. Blenko, Sr. Professor of Environmental Engineering , Department of
Civil and Environmental Engineering, College of Engineering, Carnegie Mellon University, Pittsburgh,
PA
Dr. T. Taylor Eighmy, Senior Vice President for Research, Office of the Vice President for Research,
Texas Tech University, Lubbock, TX
Dr. Elaine Faustman, Professor and Director, Institute for Risk Analysis and Risk Communication,
School of Public Health, University of Washington, Seattle, WA
Dr. John P. Giesy, Professor and Canada Research Chair, Veterinary Biomedical Sciences and
Toxicology Centre, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
Dr. Jeffrey K. Griffiths, Professor, Department of Public Health and Community Medicine, School of
Medicine, Tufts University, Boston, MA
Dr. James K. Hammitt, Professor, Center for Risk Analysis, Harvard University, Boston, MA
Dr. Barbara L. Harper, Risk Assessor and Environmental-Public Health Toxicologist, and Division
Leader, Hanford Projects, and Program Manager, Environmental Health, Department of Science and
Engineering, Confederated Tribes of the Umatilla Indian Reservation (CTUIR), West Richland, WA
Dr. Kimberly L. Jones, Professor and Chair, Department of Civil Engineering, Howard University,
Washington, DC
Dr. Bernd Kahn, Professor Emeritus and Associate Director, Environmental Radiation Center, Georgia
Institute of Technology, Atlanta, GA
Dr. Agnes Kane, Professor and Chair, Department of Pathology and Laboratory Medicine, Brown
University, Providence, RI
Dr. Madhu Khanna, Professor, Department of Agricultural and Consumer Economics, University of
Illinois at Urbana-Champaign, Urbana, IL
Dr. Nancy K. Kim, Senior Executive, Health Research, Inc., Troy, NY
Dr. Cecil Lue-Hing, President, Cecil Lue-Hing & Assoc. Inc., Burr Ridge, IL
Dr. Floyd Malveaux, Executive Director, Merck Childhood Asthma Network, Inc., Washington, DC
Dr. Judith L. Meyer, Professor Emeritus, Odum School of Ecology, University of Georgia, Lopez
Island, WA
-------
Dr. James R. Mihelcic, Professor, Civil and Environmental Engineering, University of South Florida,
Tampa, FL
Dr. Christine Moe, Eugene J. Gangarosa Professor, Hubert Department of Global Health, Rollins
School of Public Health, Emory University, Atlanta, GA
Dr. Horace Moo-Young, Dean and Professor, College of Engineering, Computer Science, and
Technology, California State University, Los Angeles, CA
Dr. Eileen Murphy, Director of Research and Grants , Ernest Mario School of Pharmacy, Rutgers
University, Piscataway, NJ
Dr. James Opaluch, Professor and Chair, Department of Environmental and Natural Resource
Economics, College of the Environment and Life Sciences, University of Rhode Island, Kingston, RI
Dr. Duncan Patten, Research Professor, Hydroecology Research Program , Department of Land
Resources and Environmental Sciences, Montana State University, Bozeman, MT
Dr. Stephen Polasky, Fesler-Lampert Professor of Ecological/Environmental Economics, Department
of Applied Economics, University of Minnesota, St. Paul, MN
Dr. C. Arden Pope, III, Professor, Department of Economics, Brigham Young University, Provo, UT
Dr. Stephen M. Roberts, Professor, Department of Physiological Sciences, Director, Center for
Environmental and Human Toxicology, University of Florida, Gainesville, FL
Dr. Amanda Rodewald, Professor of Wildlife Ecology, School of Environment and Natural Resources,
The Ohio State University, Columbus, OH
Dr. Jonathan M. Samet, Professor and Flora L. Thornton Chair, Department of Preventive Medicine,
Keck School of Medicine, University of Southern California, Los Angeles, CA
Dr. James Sanders, Director and Professor, Skidaway Institute of Oceanography, Savannah, GA
Dr. Jerald Schnoor, Allen S. Henry Chair Professor, Department of Civil and Environmental
Engineering, Co-Director, Center for Global and Regional Environmental Research, University of Iowa,
Iowa City, IA
Dr. Gina Solomon, Senior Scientist, Health and Environment Program, Natural Resources Defense
Council, San Francisco, CA
Dr. Daniel O. Stram, Professor, Department of Preventive Medicine, Division of Biostatistics,
University of Southern California, Los Angeles, CA
Dr. Peter Thorne, Professor and Head, Occupational and Environmental Health, College of Public
Health, University of Iowa, Iowa City, IA
VI
-------
Dr. Paige Tolbert, Professor and Chair, Department of Environmental Health, Rollins School of Public
Health, Emory University, Atlanta, GA
Dr. John Vena, Professor and Department Head, Department of Epidemiology and Biostatistics,
College of Public Health, University of Georgia, Athens, GA
Dr. Robert Watts, Professor of Mechanical Engineering Emeritus, Tulane University, Annapolis, MD
Dr. R. Thomas Zoeller, Professor, Department of Biology, University of Massachusetts, Amherst, MA
SCIENCE ADVISORY BOARD STAFF
Dr. Angela Nugent, Designated Federal Officer, U.S. Environmental Protection Agency, Washington,
DC
vn
-------
TABLE OF CONTENTS
Abbreviations and Acronyms ix
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 7
3. RESPONSES TO EPA'S CHARGE QUESTIONS 8
3.1. GENERAL CHARGE QUESTIONS 8
3.1.1. Overall Clarity 8
3.1.2. Additional Literature 9
3.2. SPECIFIC CHARGE QUESTIONS 10
3.2.1. Mineralogy 10
3.2.2. Toxicokinetics 12
3.2.3. Noncancer Health Effects of Libby Amphibole Asbestos 14
3.2.4. Carcinogenicity of Libby Amphibole Asbestos 17
3.2.5. Inhalation Reference Concentration (RfC) 21
3.2.6. Inhalation Unit Risk (IUR) 31
4. LONG-TERM RESEARCH NEEDS 38
4.1. EPIDEMIOLOGY 38
4.2. MODE OF ACTION 38
4.3. FUTURE DEVELOPMENT OF A TEM METHOD FOR PCM EQUIVALENCY 38
REFERENCES 40
APPENDIX A: EPA'S CHARGE QUESTIONS A-l
Vlll
-------
Abbreviations and Acronyms
AIC
ADAF
ATS
ATSDR
BMC
BMCL
BMD
BMDL
BMR
BW
CHEEC
CI
COPD
DPT
EDS
EPA
FEV1
IARC
ICD
ILO
IRIS
IUR
LAA
LOAEL
LPT
MCMC
MOA
NAS
NCI
NIOSH
NOAEL
NRC
NTP
OR
ORD
PCM
PCME
POD
RfC
ROS
RR
RTW
SAED
SEER
SEM
Akaike Information Criteria
age-dependent adjustment factor
American Thoracic Society
Agency for Toxic Substances and Disease Registry
benchmark concentration
lower 95% confidence limit of the benchmark concentration
benchmark dose
lower 95% confidence limit of the benchmark dose
benchmark response
body weight
cumulative human equivalent exposure for continuous exposure
confidence interval
chronic obstructive pulmonary disease
diffuse pleural thickening
energy dispersive spectroscopy
Environmental Protection Agency
forced expiratory volume in one second
International Agency for Research on Cancer
International Classification of Diseases
International Labor Organization
Integrated Risk Information System
inhalation unit risk
Libby Amphibole asbestos
Lowest Observed Adverse Effect Level
Localized Pleural Thickening
Markov Chain Monte Carlo
mode of action
National Academy of Sciences
National Cancer Institute
National Institute for Occupational Safety and Health
No Observed Adverse Effect Level
National Research Council
National Toxicology Program
odds ratio
Office of Research and Development
phase contrast microscopy
phase contrast microscopy equivalent
point of departure
reference concentration
reactive oxygen species
relative risk
residence time-weighted
Selected Area Electron Diffraction
Surveillance, Epidemiology, and End Results
scanning electron microscopy
IX
-------
SMR standardized mortality ratio
SIR standardized incidence ratio
TEM transmission electron microscopy
Thl T Helper Cell Type 1
Th2 T Helper Cell Type 2
TSFE time since first exposure
UCL Upper Confidence Limit
UF uncertainty factor
Database uncertainty factor
Human inter-individual variability uncertainty factor
LOAEL-to-NOAEL uncertainty factor
UFS subchronic-to-chronic uncertainty factor
WDS wavelength dispersive spectroscopy
XRD X-ray diffraction
-------
1. EXECUTIVE SUMMARY
EPA's Office of Research and Development (ORD) requested the Science Advisory Board (SAB) to
conduct a peer review of EPA's draft Integrated Risk Information System (IRIS) assessment, entitled
Toxicological Review of Libby Amphibole Asbestos (August 2011). The draft document is the first IRIS
assessment specific to Libby Amphibole asbestos (LAA), a term used to refer to the mixture of
amphibole mineral fibers identified in the Rainy Creek complex near Libby, Montana. The SAB was
asked to comment on the scientific soundness of the hazard and dose-response assessment of LAA-
induced cancer and non-cancer health effects (see Appendix A).
The SAB finds the EPA's draft assessment to be comprehensive and generally clear, logical and well-
written. There are several areas that need more consideration, and we provide recommendations to
further enhance the clarity and strengthen the scientific basis of the analyses. The SAB's major findings
and recommendations are summarized below.
Mineralogy
The SAB notes that the section on mineralogy provides an important foundation for understanding the
properties of Libby Amphibole asbestos (LAA) as related to the evaluation of its potential toxicity and
carcinogenicity. The SAB recognizes that physical-chemical characteristics of asbestos (e.g., mineral
composition, fiber dimensions) have not typically been available in toxicity studies of LAA. The SAB
encourages a more rigorous and accurate description of LAA in the document, while acknowledging the
potential ambiguities in the use of mineral-species names in toxicity studies.
Fiber Toxicokinetics
The SAB finds the section on fiber toxicokinetics does not distinguish between chrysotile and amphibole
fibers. Since the focus of the draft document is on LAA fibers, it would be better to limit most of the
literature reviews and discussion to those dealing with the family of amphibole asbestos fibers. The
authors of this section should draw on more authoritative and comprehensive reviews in the literature to
correctly specify and clarify issues on deposition and dosimetry.
Noncancer Health Effect
Selection of Critical Studies and Effects
The SAB supports the EPA's selection of the Marysville, Ohio, cohort for development of the RfC. The
SAB finds it reasonable to select the subcohort for the main analysis (118 workers who began work in
1972 or later when exposure data were available and who had X-rays from the 2002-2005 exam), with
the full cohort of 434 workers used for additional substantiating analysis. However, the SAB
recommends additional analyses/cohorts to strengthen and support the RfC since the size of the
Marysville subcohort is small. In addition to localized pleural thickening (LPT), the SAB suggests that
the EPA consider any X-ray abnormalities as the outcome: LPT, diffuse pleural thickening (DPT), or
asbestosis. The SAB also suggests that the EPA conduct analogous analyses (to the extent the data
permit) of pleural abnormalities among the Libby workers cohort and the Minneapolis Exfoliation
Community cohort.
-------
The SAB agrees that the radiographic evidence of LPT in humans is the appropriate adverse critical
effect for the derivation of the RfC. LPT has the appropriate specificity and is not confounded by
cigarette smoking. It is a permanent structural, pathological alteration of the pleura and is generally
associated with reduced lung function. The reported findings are compatible with the animal data
showing tissue injury and inflammation. The SAB has identified additional relevant publications and
recommends that the agency include a more detailed review of the literature to further support this
conclusion.
Use of Animal and Mechanistic Studies
In general, the SAB finds the laboratory animal studies identified in Tables 4-15 and 4-16 and
summarized in Appendix D of the EPA draft report to be appropriate and complete. Laboratory animal
studies using a variety of non-inhalation routes of exposure have been used to ascertain the potential
fibrogenic and carcinogenic potential of LAA. While inhalation is regarded as the most physiologically
relevant means of fiber exposure in animals, there is no published study using this route of exposure for
delivery of LAA to experimental animals. Therefore, the deposition and clearance of LAA has not been
adequately assessed in experimental animals. However, inhalation studies have been conducted with
tremolite, an asbestiform amphibole that is a component of LAA. The potency of inhaled LAA from
epidemiology studies should be compared with that of tremolite fibers in rodents to add new information
for refining the RfC for LAA.
Carcinogenicity
Weight of Evidence Characterization
The SAB supports the EPA's conclusion that the weight of evidence for LAA is "Carcinogenic to
Humans by the Inhalation Route," in accordance with EPA's Guidelines for Carcinogen Risk
Assessment. The occupational studies showed dose-related increased risks of lung cancer and
mesothelioma among workers exposed by inhalation. Effects from short-term intra-tracheal instillation
studies in mice and rats include altered gene expression, collagen induction, and inflammatory
responses, and are consistent with the early-stage pathological change induced by other amphibole
fibers. The EPA also has provided supporting evidence of the carcinogenic potential of LAA from
studies with tremolite fibers, in light of LAA being about 6 percent tremolite by composition.
Mode of Action
The SAB finds the weight of evidence for the mode of action (MOA) of LAA based on laboratory
studies to be weak. However, there are abundant MOA data for other amphiboles such as crocidolite and
tremolite that are likely similar to the MOA for LAA. The SAB views the mode of action of LAA as
complex, and recommends that a formal mode of action analysis of LAA be conducted in accordance
with EPA's Guidelines for Carcinogen Risk Assessment. Based on this formal analysis, the agency may
still conclude that the use of the default linear extrapolation at low doses is appropriate.
Selection of Critical Study andEndpoint
The SAB concludes that the EPA's selection of the Libby cohort for the derivation of the inhalation unit
risk (IUR) is scientifically supported and clearly described. This cohort has been studied thoroughly,
-------
with detailed work histories and a job exposure matrix. This cohort had elevated asbestos exposure, a
wide range of measurements of asbestos exposure, and available cancer mortality data.
The SAB finds the use of the subcohort post-1959 may be reasonable due to the lack of exposure
information in many of the workers in earlier years; out of 991 workers hired before 1960, 706 had all
department and job assignments listed as unknown.
The SAB supports the use of lung cancer and mesothelioma as endpoints for derivation of the IUR.
Since determining the cancer outcome from mortality rather than incidence data may have resulted in an
undercount of both cancer outcomes, the SAB recommends more detailed discussion on how the use of
mortality data could impact the derived IUR. It also would have been useful to know other major
categories of mortality in this cohort.
Use of Laboratory Animal and Mechanistic Studies
The SAB agrees that the database of laboratory animal and mechanistic studies pertaining to LAA is
appropriately presented in the report and its Appendices for support of its analysis of the human effects
observed. However, the SAB finds the body of the document deficient in not utilizing what is known
about the dimensions of the administered fibers from Appendix D. It is generally accepted that
differences in biological potency among the various amphibole fiber types are due primarily to
differences in dimensions, especially in fiber length distributions. The SAB also recommends that
Section 4.6.2.2 be modified to reflect that there are insufficient data to determine the mode of action for
LAA.
Inhalation Reference Concentration (RfC)
Estimates of Human Exposure Concentration
The approach described (in Appendix F of the EPA document) for exposure reconstruction is detailed
and specific. Due to large uncertainties associated with the unmeasured pre-1972 exposures, the SAB
agrees that the draft document appropriately eliminates this set of estimates and adheres only to
exposure estimates based on measured concentrations for the derivation of the RfC.
With regard to the exposure metric, the SAB recommends that the EPA re-evaluate the raw exposure
data and review pertinent sampling documentation to bolster its use of the geometric mean to represent
the job group exposures, rather than an estimate of the arithmetic mean. The agency should consider
whether a sensitivity analysis using the minimum variance unbiased estimator (MVUE) of the mean is
warranted in the development of the cumulative exposure metric.
Exposure-Response Modeling
EPA's approach to the primary exposure-response modeling was generally appropriate, but the SAB
recommends that the procedure be refined and the document should provide a clearer description of how
the "best" model was chosen, in accordance with EPA's 2012 Benchmark Dose Technical Guidance.
Since the Marysville cohort does not support precise estimation of the plateau, the EPA should consider
fixing the plateau level based on a study of highly exposed asbestos insulation workers.
-------
The SAB suggests examining other exposure metrics besides the simple cumulative exposure, such as
time-weighting of exposures. In addition, the document uses a 10% Extra Risk (ER) as the benchmark
response level (BMR) which is not typically used for human quantal response data. The SAB
recommends that EPA explain what features of the dataset or outcome variable led the agency to choose
a BMR that is considerably greater than the norm for epidemiological data.
Alternative Modeling Approach
The SAB agrees that the rationale for performing additional analyses of the full Marysville cohort is
scientifically justified; the analysis of the entire cohort increases the number of cases of LPT available
for analysis and substantiates the RfC estimated using the subcohort. However, the SAB recommends
that the EPA revise its modeling approach and remove "time since first exposure" (TSFE) from the
model of the plateau. EPA should determine whether it is appropriate to use TSFE in the linear predictor
alongside cumulative exposure and/or use an alternative exposure metric that incorporates TSFE. The
SAB also recommends the revised procedures for the subcohort analysis be followed, such as fixing the
plateau using literature values.
Evaluation of Potential Confounders and Covariates
The SAB recommends a revised strategy for evaluation of confounders and covariates. Since the
quantity of interest in the analyses of the Marysville cohort is the point of departure (POD), the
evaluation of the various covariates should be made with respect to this quantity. The SAB suggests that
the covariates fall into two classes: exposure-related covariates (various exposure metrics and TSFE)
and non-exposure-related covariates [age, body mass index (BMI), gender, and smoking status]. For
non-exposure related covariates, no additional primary analyses are needed. For exposure-related
covariates, the SAB recommends that additional work be done to refine the models to consider
alternative exposure metrics, as well as the inclusion of TSFE or other time-related variables in the
analyses of the full cohort.
Conversion from Cumulative Occupational Exposure to Lifetime Exposure
The modeled POD is based on cumulative exposure estimates for the worker cohort examined. The SAB
recommends using the full 70-year lifetime when converting cumulative to continuous exposure rather
than 60 (70 minus the lag of 10 used for exposure in the POD derivation); i.e., do not correct for the lag
of 10 for a 10-year lagged exposure, since the time of disease onset is not known in prevalence data.
Selection of Uncertainty Factors
The uncertainty factors deserve additional consideration and analysis. A composite uncertainty factor of
100 (an intraspecies uncertainty factor of 10 to account for human variability and sensitive
subpopulations; and a database uncertainty factor of 10 to account for database deficiencies) was applied
to the POD for derivation of the RfC. Although it may be difficult to identify specific data on LAA to
support departure from the default value of 10 for human variability, concern for the impact on
susceptible subpopulations, especially women and children, remains an issue. Consideration of
additional data (Minnesota cohort and data on other amphiboles) might support a lower value, such as 3,
for UFo. In addition, a subchronic-to-chronic uncertainty factor higher than 1 may be used, given that
the mean and maximum exposure duration in the study are well below the lifetime exposure of interest.
-------
There also is concern that the BMR of 10% for a severe endpoint is not reflected by the choice of a
LOAEL- to-NOAEL uncertainty factor (UFL) of 1.
Characterization of Uncertainties
Overall, the SAB found that while the discussion on uncertainties in the methodology and approach on
the derivation of the RfC was thorough, detailed, and logical, the uncertainty assessment can be
strengthened. The SAB recommends that additional work be done to substantiate the RfC estimate
through additional sensitivity analyses and discussion of results and insights from other datasets and
studies.
Inhalation Unit Risk (IUR)
Exposure-Response Modeling
The SAB supports the agency's reliance on the Libby worker subcohort for derivation of the IUR
because of its focus on good quality exposure data that are specific for LAA. However, it is important to
acknowledge that this small subcohort may have its own limitations as a basis for modeling exposure-
response relationship that might be expected in a larger population exposed over a lifetime. The SAB
had particular concern about adequate characterization of early life exposures and the potential time
dependence for development of disease.
The SAB agrees that the agency clearly described the methods used to conduct the exposure-response
modeling for lung cancer and mesothelioma. However, given limitations in the subcohort and other
statistical considerations, the SAB made a number of recommendations for providing greater support for
this choice of modeling approach and for characterizing model uncertainty.
Having made these points, the SAB recognizes that the agency did conduct extensive sensitivity
analyses of their chosen models in various ways to characterize exposure in the Libby cohort. However,
the analyses rely on essentially the same underlying models. They do not address the fundamental
question of model uncertainty - that is, whether any one model can or should be assumed to represent
the exposure-response relationship for LAA. This issue is of particular concern for the estimation of
risks from partial lifetime exposure where risk is essentially assumed to be independent of when in the
course of a lifetime exposure occurs. Recommendations for addressing model uncertainty are discussed
under response to charge question 5 in Section 3.2.6.5.
Approach for Quantification of Inhalation Unit Risk
In order to derive an IUR that represents the combined risk of mortality from lung cancer and
mesothelioma, a cancer-specific unit risk for each tumor type was calculated according to the Guidelines
for Carcinogen Risk Assessment (USEPA, 2005) by linear extrapolation from the corresponding POD.
The IUR was then determined as a combined upper bound risk estimate for mortality considering both
cancers. The SAB considers the approach to be consistent with the agency's own guidance, and found
the description of the procedure used to be clear. However, the SAB recommends that EPA
acknowledge that the assumption of independence is a theoretical limitation of the analysis and should
provide a fuller justification for this assumption.
-------
Potential Confounding by Smoking
The SAB agrees that the agency's use of the Richardson (2010) method for exploring possible
confounding for smoking was appropriate. However, the SAB finds the statement that there is no
evidence of confounding by smoking is too strong, and suggests modifications to the discussion that
would be more compelling.
Adjustment for Mesothelioma Mortality Under-ascertainment
The number of mesothelioma deaths was adjusted for under-ascertainment stemming from inadequate
coding in death certificates. The procedure is not described in any detail, but can be found in Kopylev et
al. (2011). The EPA method appears to be scientifically supported, but is not clearly described. The SAB
recommends that this section be expanded to provide a more detailed statement of how the numbers
were calculated.
Characterization of Uncertainties
The SAB commented that the EPA has summarized the many sources of uncertainty and has evaluated
qualitatively, and sometimes quantitatively, the direction and likely magnitude of their impact on
uncertainty in the IUR. However, the SAB notes that an important source of uncertainty, that of model
uncertainty, might not be accounted for either in the sensitivity analyses conducted to date or in the use
of the 95% upper confidence limit (UCL). The SAB recommends that a more straightforward and
transparent treatment of model uncertainty would be to estimate risks using a more complete set of
plausible models for the exposure-response relationship. This sensitivity analysis would make more
explicit the implications of these key model choices for uncertainty in the IUR.
Long-Term Research Needs
The SAB identifies long-term research needs for epidemiological studies, mode of action, and
measurement methods for LAA.
The National Institute for Occupational Safety and Health (NIOSH) and Agency for Toxic
Substances and Disease Registry (ATSDR) should continue to monitor mortality among Libby
workers and residents of Libby and Troy.
The SAB recommends future research on mode of action on LAA to focus on biomarkers that
are more clearly and specifically related to non-cancer endpoints (i.e., asbestosis) or cancer
endpoints (e.g., mesothelioma). Inhalation studies in animal models that can provide both
quantitative as well as mechanistic insight should be included.
EPA should develop a TEM method that provides equivalent data to PCM for LAA.
-------
2. INTRODUCTION
EPA's Office of Research and Development requested the Science Advisory Board (SAB) to review the
Draft IRIS Toxicological Review ofLibby Amphibole Asbestos (hereafter referred to as the draft
document). The draft document is based on a comprehensive review of the available scientific literature
on the health effects ofLibby Amphibole asbestos (LAA), a term used to refer to the mixture of
amphibole mineral fibers of varying elemental composition (e.g., winchite, richerite and tremolite) that
have been identified in the Rainy Creek complex near Libby, Montana. The draft document provides the
scientific and quantitative basis for toxicity values that will be entered into EPA's online Integrated Risk
Information System (IRIS) database. Specifically, this draft IRIS assessment provides an overview of
sources of exposure to LAA, and characterizes the hazard posed by exposure to LAA for carcinogenicity
and noncancer health effects based on the available scientific evidence. The assessment includes the
derivation of a chronic inhalation reference concentration (RfC) and an inhalation unit risk (IUR) that
can be combined with exposure information in a risk assessment to estimate noncancer hazard and
carcinogenic risk, respectively, in humans. The assessment does not address oral exposure to LAA.
In response to the agency's request, the SAB convened an expert panel (the Libby Amphibole Asbestos
Review Panel) to conduct the review. The SAB panel discussed its responses to the EPA's charge
questions during a February 6-8, 2012 face-to-face meeting and on public teleconferences on May 1,
May 8, and July 25, 2012. The SAB Panel's draft report was discussed and approved with clarifying
edits by the chartered SAB on a public teleconference on September 25, 2012. Oral and written public
comments were considered throughout the advisory process. The SAB Panel and the chartered SAB
considered and discussed these comments as part of the process of developing this report.
EPA's charge questions consist of two general charge questions on the organization, presentation, and
clarity of the draft document, as well as specific charge questions that focus on: mineralogy and
toxicokinetics, hazard assessment of non-cancer and cancer health effects, exposure-response
assessment for derivation of an RfC for non-cancer endpoints, cancer weight of evidence classification,
mode of action of LAA carcinogenicity, and exposure-response assessment for derivation of an IUR for
LAA (see Appendix A). The Executive Summary highlights the SAB's major findings and
recommendations. The SAB's full responses to the EPA charge questions are detailed in Section 3 and
brief recommendations on long-term research needs are provided in Section 4.
-------
3. RESPONSES TO EPA'S CHARGE QUESTIONS
3.1. General Charge Questions
3.1.1. Overall Clarity
Question 1. Is the Toxicological Review logical, clear, and concise? Has EPA clearly, and in sufficient
detail, presented and synthesized the scientific evidence for health hazards from Libby Amphibole
asbestos?
In general, the SAB finds the toxicological review to be well-written, logical and appropriately
referenced relative to the health hazards and exposure response of Libby Amphibole asbestos (LAA).
However, the SAB has identified sections where extraneous and repetitive materials could be deleted.
Examples include the following:
For Section 3, since the focus of the draft document is on Libby amphibole fibers, it would be
better to limit the literature reviews and discussions to those dealing with the family of
amphibole asbestos fibers. Chrysotile asbestos fibers are very different from amphibole fibers in
terms of their airborne concentration measurement errors and uncertainties, much lower
biopersistence, faster clearance, and different translocation pathways.
There are a large number of studies discussed in Section 4 nine community studies (4.1.4) and
two case reports (4.1.5)that offered no detailed exposure information and cannot be used in the
risk assessment.
Discussions that offer little or no new insights into the toxicology of asbestos should be briefly
summarized.
Some sections are repetitive (e.g., Section 5.4.4 and 5.4.5).
Regarding clarity and sufficient detail in the presentation and synthesis of the scientific evidence for
health hazards from LAA, the SAB finds the scientific evidence for health effects of LAA to be
reasonably well presented. However, the SAB has identified areas where the draft document could be
clarified and some aspects of the EPA's analysis that require more explanation and justification, as
provided in the responses to specific questions in subsequent sections. In addition, the SAB has
comments on the following areas:
Relevance of Other Literature Related to Amphiboles
The toxicological review does not make clear the relevance of the extensive literature on the
health effects of other amphibole fibers. There are numerous publications on the mode of action
of other amphiboles, inhalation studies in rodents, and epidemiological studies of populations
exposed to amphiboles environmentally. Literature on epidemiological studies of other
amphiboles is particularly useful for model selection for dose-response assessment of LAA due
to the small size of the Libby subcohort used in exposure-response modeling.
Early Lifestage Susceptibility
There is inconsistency in the tone of the conclusions in Section 4.7.1.1 (Lifestage Susceptibility)
and in Section 6.3.3 (Applications to Early Lifetime and Partial Lifetime Environmental
Exposure Scenarios for IUR) to either support or refute early lifestage susceptibility. The SAB
recognizes that no firm conclusion can be drawn about differential risk of adverse health effects
after early life stage exposure to LAA compared to exposure during adulthood, due to the limited
-------
and inconclusive studies on other forms of asbestos. However, the available limited evidence
pointing to excess risk for exposures during childhood needs to be considered when considering
a margin of safety.
Recommendations
An overall summary set of tables or figures describing the major cohorts (Libby workers,
community, Marysville plant), the types/timelines of exposure, and the studies associated with
each would help orient the readers of the document.
The draft document would benefit from greater usage of graphs and figures to highlight
conclusions.
The section on susceptible populations could be better organized and more succinctly
summarized. The section should especially focus on childhood asbestos exposure, the asbestos
susceptibility issue most relevant to this EPA document, as this is the topic where there is at least
some (albeit limited) data.
The draft document could be enhanced with quantitative comparison of the environmental
exposures that have taken place in other geographic regions of the world (i.e., the Anatolia
region of Turkey and Greece) (Sichletidis et al., 2006; Constantopoulos, 2008; Gogou et al.,
2009; Carbone et al., 2011; Metintas et al., 2008, 2010, 2012) with the Libby, Montana,
community with regard to airborne tremolite. This comparison should include numbers of fibers
and fiber size distribution in relation to health effects.
A table comparing these results with the results from the earlier 1988 EPA analysis (USEPA,
1988) on asbestos would be helpful.
3.1.2. Additional Literature
Question 2. Please identify any additional peer-reviewed studies from the primary literature that should
be considered in the assessment ofnoncancer and cancer health effects of Libby Amphibole asbestos.
The SAB has identified additional studies to be considered in the assessment:
Adgate, JL; Cho, SJ; Alexander, BH; Ramachandran, G; Raleigh, KK; Johnson, J; Messing, RB;
Williams, AL; Kelly, J; Pratt, GC. (2011). Modeling community asbestos exposure near a vermiculite
processing facility: Impact of human activities on cumulative exposure. J Expo Sci Environ Epidemiol
21: 529-535.
Alexander, BH; Raleigh, KK; Johnson, J; Mandel, JH; Adgate, JL; Ramachandran, G; Messing, RB;
Eshenaur, T; Williams, A. (2012). Radiographic evidence of nonoccupational asbestos exposure from
processing Libby vermiculite in Minneapolis, Minnesota. Environ Health Perspect 120: 44-49.
Antao,VC; Larson,TC; Horton, DK.(2012). Libby vermiculite exposure and risk of developing asbestos-
related lung and pleural diseases. Curr. Opin. Pulmonary Med. 18:161-167, PMJJD: 22139761.
Berman, DW (2011). Apples to apples: The origin and magnitude of differences in asbestos cancer risk
estimates derived using varying protocols. Risk Analysis 31: 1308-1326.
CyphertJM; Padilla-Carlin, DJ; Schladweiler, MC; Shannahan, JH; Nyska, A; Kodavanti, UP; Gavett,
SH. (2012). Long-term response of rats to single intratracheal exposure of libby amphibole or amosite. J
ToxicolEnviron Health A 75: 183-200. http://dx.doi.org/10.1080/15287394.2012.641203.
9
-------
Marchand, LS; St-Hilaire,S; Putnams, EA., et al.(2012). Mesothelial cell and anti-nuclear autoantibodies
associated with pleural abnormalities in an asbestos exposed population of Libby MT. Toxicology
Letters 208: 168-173.
Shannahan, JH; Nyska, A; Cesta, M; Schladweiler, MC; Valiant, BD; Ward, WO; Ohio, AJ; Gavett, SH;
Kodavanti, UP. (2012). Subchronic pulmonary pathology, iron overload, and transcriptional activity
after libby amphibole exposure in rat models of cardiovascular disease. Environ Health Perspect 120:
85-91.
Shannahan, JH; Ohio, AJ; Schladweiler, MC; Richards, JH; Andrews, D; Gavett, SH; Kodavanti, UP.
(2012).Transcriptional activation of inflammasome components by Libby amphibole and the role of
iron. Inhalation Toxicology 24:60-69, PMJD: 22168577
Webber, JS; Blake, DJ; Ward, TS; Pfau, JC. (2008). Separation and Characterization of Respirable Amphibole
Fibers from Libby, Montana. Inhal. Toxicol. 20: 733 - 740.
Zeka A; Gore R; Kriebel D (2011). The two-stage clonal expansion model in occupational cancer
epidemiology: results from three cohort studies. Occupational and Environmental Medicine 68: 618-24.
3.2. Specific Charge Questions
3.2.1. Mineralogy
Question la. Please comment on whether the presentation of the available data on the mineralogy of
Libby Amphibole asbestos is clear, concise and accurate.
Section 2, Geology and Mineralogy of Libby Amphibole Asbestos, provides a discussion of the
mineralogical and geological aspects of Libby Amphibole. In general, the SAB finds that this section
provides an important foundation for understanding the nature of Libby Amphibole asbestos (LAA) as
related to evaluation of potential exposures. There are places where the clarity and accuracy of the
section can be improved, and these are detailed below.
There is a mismatch between the mineral ogical detail embodied in the definition of mineral species and
the detail available relative to specific exposures in Libby. Specifically, mineral species define a very
specific structure (e.g., amphibole) and a specific composition or range of compositions (e.g., winchite
or tremolite). Given that these factors affect a mineral's physical and chemical behavior, they may in
principle be factors to consider for potential hazard. The SAB recognizes that this level of detail is not
typically available for toxicity studies to allow its application to the evaluation of LAA/>er se. In
general, however, the observed unique aspects of amphibole asbestos support the evaluation of LAA
through comparison with other amphiboles based on particle morphology and amphibole designation.
Nevertheless, the SAB encourages a rigorous and accurate description of LAA in Section 2, perhaps
while noting the potential ambiguities in the use of mineral-species names in other studies.
10
-------
Comments on the subsections follow:
Discussions of mineralogy and morphology in Sections 2.2.1.1 and 2.2.1.2 are good, with
appropriate discrimination between methods/definitions that are applied to mineral field samples
collected from the site versus terms/definitions that are applied to environmental samples
collected via air monitoring (line 16 of page 2-9 and lines 4 and 5 of page 2-10).
Section 2.1 is generally sufficient for providing a background on historical aspects of the mining
operations in Libby, Montana.
Section 2.2 needs modification. This section should lay a foundation for understanding the nature
of Libby Amphibole (e.g., mineralogical characteristics such as composition and morphology),
information on how the material may vary spatially and temporally (with respect to mining
operations), and other factors that may impact exposures. The section does contain much
relevant information. There are parts of the section that are incorrect and misleading;
recommendations to address these issues include:
o Consistent use of terminology associated with par tide morphology. The section mixes a
number of terms that address particle morphology, and these are critically important in
assessing potential exposures and subsequent impacts. As an example, "fibers (e.g.,
acicular...)" implies fibrous and acicular are the same, when in conventional usage they are
different (e.g., see Veblen and Wyllie, 1993). A tight use of terms that are defined up front
should be followed in the EPA document even when a lax use of terms may exist in the
literature cited. A partial attempt is provided in Section 2.2.1.2, but it could be expanded and
carefully vetted with respect to accepted terminology. The four most important terms to lay
out clearly are fibrous, acicular, prismatic and asbestiform. If the report's intent is to note
differences in these terms, they should be discussed; if the conclusion is that there are poorly
defined distinctions, that topic also should be discussed. One specific example of inaccurate
usage is the term "prismatic," which by definition is "prism"-shaped (meaning parallel sides;
it is incorrectly used in multiple places).
o Double-check all mineral formulae. There are numerous incorrect compositions in the report;
although some of these may be typographic errors (which, of course, should be fixed), some
may be incorrectly reported. An example of one incorrect formula is that attributed to
vermiculite, which is listed incorrectly as: [(Mg,Fe,A)3(Al,Si)2Oio(OH)2*4H2O].
o Double check that all mineral-species definitions used are accepted mineralogical standards.
Mineral species are fundamental terms that describe a material with a specific structure and a
specific composition or range of compositions; both factors are primary determinants of a
material's properties. Indeed, at the heart of this report is the definition of likely exposures to
(and risks from) inhaled particles and other fibers based on the use of mineral-species names.
The problems in this category are probably most widespread in Section 2.2.1.1, which details
amphibole mineralogy (which is central to the report). For example, anthophyllite is not a
Libby amphibole.
The SAB appreciates the discussions that highlighted the complexity and variability of LAA in
the context of compositional solid solutions, emphasizing that even the use of mineral-species
names for LAA may mislead readers to believe that LAA is represented by a few discrete
materials as opposed to a mixture of materials with varying compositions. Overall, the
11
-------
mineralogy section could benefit from some technical editing. It presents some irrelevant
material (e.g., section 2.2.1, which is a general description of silicate mineral hierarchy), omits
some critical information (e.g., section 2.2.1.1 does not provide the mineralogical definitions of
key minerals like winchite or richterite), and presents some erroneous and irrelevant
characterizations (e.g., some of the vermiculite-mineralogy descriptions in section 2.2.2).
The report provides a good summary of available information on the LAA. One specific
observation that could be added is one reported by Sanchez et al. (2008), namely that they
observed no correlation between morphology (fibrous vs. prismatic) and major-/minor-element
chemistry. Webber et al. (2008) similarly concluded that there was no correlation between
mineral species and fiber width for respirable fibers. In other words, this is consistent with the
implication that the large set of compositional data from Meeker et al. (2003) shown in the report
reflects the range of compositions associated with inhaled-fiber exposures.
Discussion on page 2-10 glosses over a serious shortcoming of phase contrast microscopy
(PCM); namely, its inability to detect fibers narrower than -0.25 um. These thin fibers are
among the most biologically potent according to the Stanton-Pott hypothesis. The fact that only a
third of the Transmission Electron Microscopy (TEM)-visible Libby fibers were PCM-visible is
buried in McDonald et al. (1986). Furthermore, Text Box 2-2 does not adequately contrast the
capability of EM versus PCM. EM's capability to yield elemental composition via Energy
Dispersive Spectroscopy (EDS) and Wavelength Dispersive X-ray Spectroscopy (WDS)
provides information to identify different asbestos types. PCM, in contrast, cannot even
determine if the fiber is mineral. Furthermore, the Selected Area Electron Diffraction (SAED)
capability of TEM allows determination of crystalline structure, e.g., amphibole versus
serpentine. Finally, Box 2-2 incorrectly states that scanning electron microscopy (SEM)
"produces three-dimensional (3-D) images". Rather, SEM produces 2-D images that reveal
surface structure of particles.
The electron microscopy section on page 2-11 could be clarified. SEM and TEM provide higher
resolution to allow better particle morphological analysis. Electron diffraction allows
mineralogical assessment. Energy dispersive X-ray analysis allows elemental composition
determination, which can corroborate the mineralogical determination. X-ray diffraction (XRD)
mentioned in this section is useful for bulk sample mineralogy measurements.
3.2.2. Toxicokinetics
Question Ib. In the absence of toxicokinetic information specific to Libby Amphibole asbestos, the draft
assessment contains a general summary description of fiber toxicokinetics. Please comment on whether
this overview of general fiber toxicokinetics is clear, concise and accurate.
The discussion of general fiber toxicokinetics is not clear, nor concise, especially since it fails to
distinguish between chrysotile and amphibole fibers. Furthermore, it is inaccurate in many places, as
noted below.
In view of the fact that the focus of the document is on Libby Amphibole fibers, it would be
better to limit most of the literature reviews and discussions to those dealing with the various
kinds of amphibole asbestos fibers. Chrysotile asbestos fibers, which are not a significant
complication in exposures to Libby vermiculite, are very different from amphibole fibers in
terms of their: (a) airborne concentration measurement errors and uncertainties (HEI-AR, 1991);
12
-------
(b) much lower biopersistence (Bernstein et al., 2004; 2005a; 2005b); (c) clearance and
translocation pathways and rates (Bernstein et al., 2004; 2005a; 2005b).
There are some misstatements on fiber deposition and dosimetry in the document.
o The authors should draw on more authoritative and comprehensive reviews in the literature
(e.g., Lippmann, 2009; Mossman et al., 2011). One misstatement in the draft is that
impaction is affected by fiber length. Another is that interception is affected by aspect ratio.
The document should cite the work by Sussman et al. (1991a; 1991b) that demonstrates that
interception of amphibole (crocidolite) fibers is only demonstrably in excess when fiber
lengths are >10 jim. Also, the report should cite the work of Brody and colleagues (Brody et
al., 1981; Brody and Roe, 1983; Warheit and Hartsky, 1990) on chrysotile fiber deposition in
the alveolar region in rodents. In terms of deposition sites, there should be no significant
difference between chrysotile and amphibole fibers.
o Another misstatement is that mucociliary clearance is complete within minutes or hours
rather than the true time frame of hours to a few days (Albert et al., 1969). The authors also
need to acknowledge that particles depositing in the alveolar region can reach the
tracheobronchial tree in two ways: (a) on surface fluids drawn onto the mucociliary escalator
by surface tension, and (b) by passing through lymphatic channels that empty onto the
mucociliary escalator at bronchial bifurcations. The report also should acknowledge that
macrophage-related clearance of fibers is only applicable to short fibers that can be fully
phagocytosed. Nearly all of the references to chrysotile in the discussion of translocation
should be deleted. The Libby asbestos fibers are essentially all amphibole fibers, and there is
very little commonality among serpentine and amphibole fibers in terms of translocation or
long-term retention.
o There are also toxicokinetic misstatements in Section 4.2 describing cancer bioassays in
animals. The section should cite the inhalation study of Davis et al. (1985) with fibrous
tremolite, which is very similar to Libby amphibole. Also, this section should discuss the
tremolite inhalation study of Bernstein et al. (2003, 2005) that is cited in Table 4-16, as well
as the more recent study by Bernstein et al. (2011) that demonstrated pleural translocation in
rats using non-invasive means following airborne amosite asbestos exposure. The study
examined animals for up to one year following a short one-week exposure to amphibole and
characterized the size of fibers that were present in parietal pleura. Non-cancer inflammatory
pleural changes were demonstrated associated with fiber translocation. This paper shows
rapid translocation of fibers to the pleura (at least of rodents) and it should be referenced for
completeness on toxicokinetic issues. Furthermore, the results of the various studies cited in
Section 4.2 are almost all very difficult to interpret with respect to the toxic effects that were,
or were not, reported, since no information was provided in Tables 4.15 and 4.16 on the key
dosimetric factor of fiber dimensions. There were comprehensive summaries of available
information on fiber dimensions of materials administered in the bioassays in Appendix D,
including numbers of long fibers, but Section 4.2.5 is deficient as a summary of animal
studies for LAA and tremolite because it does not discuss how the content of long fibers in
the administered materials had an influence on the effects observed.
13
-------
3.2.3. Noncancer Health Effects of Libby Amphibole Asbestos
3.2.3.1. Selection of Critical Studies and Effects
Question 1. An occupational cohort of workers in a Marysville, OH facility exposed to Libby Amphibole
asbestos (Lockey et al, 1984; Rohs et al, 2008) was selected as the basis for the derivation of the
reference concentration (RfC). Please comment on whether the selection of this study population is
scientifically supported and clearly described. If a different study population is recommended as the
basis for the RfC, please identify this study and provide scientific support for this choice.
The rationale for the use of the Marysville, Ohio, cohort for development of the RfC was well described
and scientifically supported. However, there are clear drawbacks to this cohort due to the lack of
exposure sampling prior to 1972 when most of the cohort began work, the use of self-reported work
histories, the end of Libby vermiculite use in 1980 and the mixture of vermiculite sources used
throughout the life of the plant. These drawbacks are offset by the solely occupational exposure of this
cohort, the use of better quality radiographs taken for research purposes, the use of 2000 ILO standards
for reading radiographs, and a cohort with exposures closer to environmental levels. The selection of
the subcohort for the main analysis has a clear and strong rationale. (There were 118 workers who began
work in 1972 or later when exposure data were available, and who had X-rays from the 2002-2005
exam.) The full cohort of 434 workers was used for analyses to substantiate the subcohort findings.
Although the SAB agrees that the Marysville subcohort represents the best population upon which to
base the RfC, there was discussion about the need for additional analyses/cohorts to strengthen and
support the RfC since the size of the Marysville subcohort was small. One suggestion is to use the
Marysville cohort but include any X-ray abnormalities as the outcome [LPT, diffuse pleural thickening
(DPT), or asbestosis]. In addition, cause of death might be assessed for those who died between the two
exams. Another suggestion for providing support and perspective to the Marysville findings is to
conduct analogous analyses (to the extent the data permit) of pleural abnormalities among the Libby
workers cohort (Larson et al., 2012) and among the Minneapolis exfoliation community cohort (Adgate
et al., 2011; Alexander et al., 2012). The Libby workers have higher, well characterized occupational
exposures compared to the Marysville cohort. The Minneapolis cohort of non-workers generally had
estimated exposures at the lower end of the Marysville cohort but included women and children, thus
providing a cohort more representative of the general population. However, because the Minneapolis
cohort had estimated, not measured exposures, it would not be suitable for the primary RfC analysis.
Similarly, because the Libby workers have both environmental and occupational exposures, this cohort
should not be used for primary RfC analysis.
Question 2. Radiographic evidence of localized pleural thickening in humans was concluded by EPA to
be an adverse effect and was selected as the critical effect for the derivation of the RfC. Pleural
thickening is associated with restrictive lung function, breathlessness during exercise and, for some
individuals, chronic chest pain. Please comment on whether the selection of this critical effect and its
characterization is scientifically supported and clearly described. If a different health endpoint is
recommended as the critical effect for deriving the RfC, please identify this effect and provide scientific
support for this choice.
Radiographic evidence of localized pleural thickening (LPT) in humans is the appropriate adverse and
critical effect for the derivation of the RfC. This is clearly described and well supported by the lines of
evidence presented in section 4.1.1.4.2. However, the SAB believes additional evidence is available to
further support this view and should be reported.
14
-------
While other health endpoints (such as diffuse pleural thickening and small opacity profusion) might
have been considered candidates for the critical effect for deriving the RfC, the use of LPT is
appropriate and well supported. LPT is a permanent, structural, pathological alteration of the pleura.
LPT is found at a significantly elevated prevalence in exposed individuals, has the appropriate
specificity and is not confounded by cigarette smoking. LPT also is associated with reduced lung
function. Furthermore, the findings reported in this section are compatible with the animal data showing
tissue injury and inflammation.
It is important to provide a more detailed review of the literature to support the use of LPT as the
appropriate endpoint, including studies addressing the relationship between LPT and both pathologic
and physiologic abnormalities. Published studies that address the relationship between LPT and lung
function suggested by the SAB include Lilis et al., 1991b; Paris et al., 2009; Clin et al., 2011; Sichletidis
et al., 2006; Whitehouse, 2004; and Wilken et al., 2011, along with those referenced in the American
Thoracic Society (ATS) Statement entitled, Diagnosis and Initial Management ofNonmalignant
Diseases Related to Asbestos: Official Statement of the American Thoracic Society (ATS, 2004) (Ohlson
et al., 1984; 1985; Jarvolm and Sanden, 1986; Hjortsberg et al., 1988; Oliver et al., 1988; Bourbeau et
al., 1990; Schwartz et al., 1990; Miller et al., 1992; Van Cleemput et al., 2001; Miller, 2002; ).
Consistent with that ATS Statement, the SAB concludes that cohort studies have shown significant
reduction in lung function, including diminished diffusing capacity and vital capacity associated with
LPT. To help clarify the difference between "clinically significant" effects of plaques in a given patient
vs. epidemiological studies evaluating the effects of asbestos exposure in an exposed population, the
SAB suggests that the EPA clarify in the assessment the range of endpoints that generally can be used to
derive an RfC.
In addition to localized pleural thickening, the SAB also suggests that the EPA consider looking at LPT,
DPT and small opacity profusion score together as an outcome. There is evidence that LPT is not always
the first adverse effect that is detected on chest radiographs, and some individuals with LAA exposure
can develop either DPT or increased profusion of small opacities without developing evidence of LPT.
Combining outcomes is appropriate, since DPT and small opacity profusion also are effects of asbestos
exposure and the goal is to define an exposure level below which LAA is unlikely to have adverse health
effects.
Recommendations:
The SAB suggests the EPA assessment clarify the range of endpoints that generally can be used
to derive an RfC.
The agency should include a more detailed review of the literature to support the selection of
LPT through detailing the studies that show the relationship between LPT and both pathologic
and physiologic abnormalities, and also risk of other non-cancer asbestos-related diseases.
In addition to LPT, the document should include an analysis that uses all radiographic outcomes
(LPT, DPT and small opacities), recognizing this change may have little impact on the current
analysis.
15
-------
3.2.3.2. Use of Animal and Mechanistic Studies
Question 3. The database of laboratory animal and mechanistic studies of Libby Amphibole asbestos is
summarized in the draft assessment (see Section 4.2 and 4.3, details in Appendix D) to inform the
mechanisms of the biological response to Libby Amphibole asbestos and support the epidemiology
studies used for derivation of the RfC. Please comment on whether the laboratory animal and
mechanistic information presented is used appropriately in the draft assessment.
The EPA draft document discusses the different types of minerals present in LAA and it is uncertain
how the various components relate to adverse health effects. LAA contains -6% tremolite and there
is clear evidence from human and animal studies that tremolite causes adverse health effects in
humans and experimental animals. However, since LAA also contains winchite (84%) and richterite
(~11%), it would be prudent to determine whether these mineral forms contribute to the adverse
health effects of LAA or whether there are interactive effects of winchite or richterite that modify the
toxicity of tremolite. The SAB recommends that this issue be highlighted since it is well-known that
tremolite is highly fibrogenic and causes malignant mesothelioma (MM). However, the contribution
of winchite or richterite to adverse health effects is apparently unknown.
In general, the listing of the laboratory animal studies in Tables 4-15 and 4-16 and the underlying data
summary in Appendix D are appropriate and complete. However, Tables 4-15 and 4-16, and the
summary data in Appendix D do not include the distributions of fiber lengths, and Section 4.2.5 is
therefore deficient as a summary of animal studies for LAA and tremolite, in terms of not discussing
how the content of long fibers in the administered materials had an influence on the effects observed.
The report text in Section 4.2.5 also is deficient in not discussing how the contents of long fibers in the
administered materials had an influence on the effects observed. Therefore, the issue of the influence of
fiber dimensions, and especially of fiber length, needs to be strengthened. The LAA fiber dimensions,
listed in Table D-5 (page D-6) should be moved to the main text in Section 4.4, Mechanistic Data and
Other Studies in Support Of the Mode of Action. A recent paper by Berman (2011), which was not cited
in the draft report, suggests that cancer risk coefficients for various amphiboles are more consistent
when fiber length was taken into consideration. Berman (2011) also suggests that the health risks
presented by amphibole are greater than those of chrysotile.
Laboratory animal studies utilizing various stocks and strains of mice and rats as well as hamsters,
by a variety of non-inhalation routes of exposure, have been used to ascertain the potential
fibrogenic and carcinogenic potential of LAA. While inhalation is regarded as the most
physiologically relevant means of fiber exposure in animals, there is no published study with the
LAA mixture with this route of fiber administration in experimental animals. However, there has
been intratracheal instillation of LAA in short-term studies with mice and rats that resulted in airway
inflammatory change consistent with earlier changes seen in tremolite-exposed animals. The lack of
any inhalation data in rats or mice is an important issue, since the deposition of particles and fibers
cannot be adequately addressed using intratracheal instillation of a bolus of fibers delivered in
aqueous suspension. For example, the development of pleural lesions may be quite different when
comparing fibrogenic or carcinogenic fibers or other particles by inhalation versus instillation. Since
inhalation studies have been conducted with tremolite fibers (e.g., Bernstein et al., 2005) and the fact
that 6% of LAA are tremolite, the potency of inhaled LAA from epidemiology studies should be
compared to that of tremolite fibers in rodents. This relative potency of LAA could add new
information for refining the RfC for LAA. Additional parameters in better defining the RfC can be
found in Section 3.2.5.1.
16
-------
In vitro assay systems utilizing both primary cells and established human and mammalian cell lines
have been used to provide mechanistic insights on the potential mode of action of LAA. These
limited in vitro studies have demonstrated the importance of fiber-cell interactions, the ability of
LAA to induce reactive radical species, inflammatory gene expression, and micronuclei, a marker of
genomic instability. Unfortunately, with the exception of the latter, most of these endpoints are non-
specific and can be demonstrated with any particles including glass fibers in short-term assays.
Similarly, Section 4.4.1 (page 4-63) mentions increases in Thl and Th2 cytokines that are not
specific to the effects of LAA or other types of asbestos, but rather generalized mediators of non-
allergic or allergic inflammatory responses. Likewise, pro-inflammatory cytokines (e.g., interleukin-
8), enzymes (e.g., cyclooxygenase-2) and oxidative stress markers (e.g., heme oxygenase) are
biomarkers of a wide variety of cellular stress and inflammation responses that will probably not
shed much light on the mechanisms of LAA-induced disease. It would be valuable for future
research on LAA mode of action to focus on biomarkers that are more clearly and specifically
related to non-cancer endpoints (i.e., asbestosis) or cancer endpoints (e.g., mesothelioma). Critical
genotoxicity studies including mutagenesis and chromosomal aberration studies have not been
reported/ examined with LAA.
3.2.4. Carcinogenicity of Libby Amphibole Asbestos
3.2.4.1. Weight of Evidence Characterization
Question 1. Under EPA 's Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005;
www.epa.gov/iris/backgrd.html), the draft IRIS assessment characterizes Libby Amphibole asbestos as
"carcinogenic to humans " by the inhalation route of exposure. Please comment on whether the cancer
weight of evidence characterization is scientifically supported and clearly described.
Human epidemiological data supersede animal and other laboratory studies in the identification of a
human carcinogen/toxicant. For LAA, the SAB agrees with the EPA that, while there are no concrete
laboratory studies that unequivocally demonstrate carcinogenicity of the fiber mix, there are strong
epidemiological data that support the notion that LAA fiber is closely linked to cancer incidence and
mortality in humans under occupational settings. The occupational studies appeared most persuasive
at showing dose-related increased risks of lung cancer and mesothelioma among workers exposed by
inhalation. However, the number of mesothelioma cases is small. The case series in the community,
while supportive, do not provide the same level of evidence for an association or for the strength of
the association. Nonetheless, the epidemiologic evidence from the occupational studies does support
the choice of descriptor "carcinogenic to humans by the inhalation route" for LAA under the
conditions of exposure in those studies.
On the other hand, the only solid evidence that the LAA is carcinogenic to animals is in hamsters
injected intraperitoneally with a single 25-mg dose of the fiber mix, which is not a physiologically
relevant route of exposure in humans. Although inflammation of the lung has been demonstrated
using both mice and rats exposed to LAA by intra-tracheal instillation, these short-term studies
failed to demonstrate any cancer induction. The SAB, however, concurs with the EPA report that
these findingswhich include altered gene expression, collagen induction, and inflammationare
consistent with the early-stage disease process induced by other amphibole fibers. As such, the EPA
has derived additional supporting evidence for the carcinogenic potential of LAA from studies with
tremolite fibers. Although the SAB recognizes that these studies provide circumstantial, supporting
evidence of the carcinogenic potential of LAA in light of its -6% tremolite by composition, the
17
-------
limited data base on LAAper se cannot provide a well defined mode of action for either lung cancer
or mesothelioma induction, as will be discussed in the following section.
3.2.4.2. Mode of Carcinogenic Action
Question 2. Due to the limitations of the data available, the draft assessment concludes that there is
insufficient information to identify the mode of carcinogenic action ofLibby Amphibole asbestos. Please
comment on whether this determination is appropriate and clearly described. Note that in the absence of
information to establish a mode of action, a linear low dose extrapolation is recommended by the
Guidelines for Carcinogen Risk Assessment (U.S., EPA, 2005; Section 3.3). If it is judged that a mode of
action can be established for Libby Amphibole asbestos, please identify the mode of action and its
scientific support (i.e., studies that support the key events, and specific data available to inform the
shape of the exposure-response curve at low doses).
A formal mode of action analysis in accordance with EPA's Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 2005) has not been conducted in the draft assessment. The mechanisms by
which amphibole fibers produce malignancy and fibrosis are complex and likely to be multifactorial
in nature. The induction of reactive radical species through persistent interaction of fibers with target
cells, the involvement of chronic inflammatory response, the activation of certain oncogenes and
inactivation of yet-to-be-identified suppressor gene(s), have been proposed as possible mechanisms.
In addition, various in vitro and in vivo studies have shown that fiber dimensions, surface properties,
shape and crystallinity, chemical composition, physical durability, and exposure route, duration, and
dose are important determinants of the biological potency of fibers.
With the LAA, neither the fairly limited amount of research conducted using in vivo as well as in
vitro assays that are described in the review, nor the more extensive body of published work on other
asbestiform minerals, which is also summarized, lead to clear conclusions as to a single mode of
carcinogenic action. The SAB agrees with the EPA conclusion that the laboratory-based weight of
evidence for the mode of action of LAA is weak. Given the limited data base available in the
literature and some limited support from data on carcinogenesis by other amphiboles, the EPA's
conclusion that there is insufficient information to identify the mode of carcinogenic action of LAA
may be justified. However, there are extensive data suggesting multiple mechanisms of carcinogenic
action of other amphibole asbestos fibers (IARC, 2012). The SAB finds that, given the available
information, the default linear extrapolation at low doses may be appropriate.
Recommendation:
A formal mode of action analysis for LAA should be conducted in accordance with EPA's
Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005).
3.2.4.3. Selection of Critical Study and Endpoint
Question 3. An occupational cohort of workers from Libby, MT exposed to Libby Amphibole asbestos
(i.e., the Libby worker cohort) was selected as the basis for the derivation of the inhalation unit risk
(1UR). Please comment on whether the selection of this study population is scientifically supported and
clearly described. If a different study population is recommended as the basis for the IUR, please
identify this study and provide scientific support for this choice.
The selection of the Libby cohort is scientifically supported and clearly described. It appears to be the
best cohort available for cancer outcomes. This cohort has been thoroughly studied previously, had
18
-------
detailed work histories with a job exposure matrix available, had elevated asbestos exposure, had a wide
range of measurements of asbestos exposure (covering a range of two orders of magnitude), was large,
and had cancer mortality data available. Limitations of this cohort include the possible environmental
exposures to asbestos and limited smoking information available, especially given that smoking is an
important risk factor for lung cancer (but not for mesothelioma) and also may have a synergistic effect
with asbestos exposure. Also, outcomes are based on death certificates, which could undercount
incidences of relevant endpoints.
Libby Amphibole asbestos is the only possible source of the asbestos measured in the air samples (i.e.,
there are no other sources of asbestos at the mine and associated facilities). It should be noted, however,
that this study population may not be representative of the larger population, since most of its members
are white males, exposed as adults, and it contains a higher proportion of cigarette smokers than the
larger population. If a residential study is ever completed that includes a larger proportion of women,
other races, and those exposed as children, the derivation of the IUR should be revisited. Additionally, it
is noted that the endpoints are based on cancer mortality noted on death certificates. While this could
lead to an undercounting of actual cases of lung cancer, it seems less likely that lung cancer in a heavily
asbestos-exposed population would either be missed on a death certificate or would significantly
undercount incidence more so than in the comparison population. Mesothelioma cases might not have
been fully accounted for using death certificates, as mesothelioma did not have a distinct classification
(International Classification of Diseases, or ICD) code prior to ICD-10, implemented in 1999. However,
death certificates were manually reviewed, as noted, and possible under-ascertainment of mesothelioma
cases was addressed in the modeling.
Use of the subcohort post-1959 seems reasonable due to the lack of exposure information for many of
the workers in earlier years. Out of 991 workers hired before 1960, 811 had at least one job with an
unknown job assignment and of these 706 had all department and job assignments listed as unknown. It
would seem highly problematic to include workers with limited or no job information in the model.
However, at least some information existed for the remaining 285 workers.
The EPA should strengthen the analysis to calculate an overall Standardized Mortality Ratio (SMR) for
the Libby worker full- and sub- cohorts for lung cancer, using both Montana and U.S. data for
comparison. The later cohort also had lower levels of exposure to asbestos, which would be closer to the
lower levels found in the environment.
Question 4. Mortality from lung tumors and mesothelioma in the Libby worker cohort was selected to
serve as the basis for the derivation of the IUR. Please comment on whether this selection is
scientifically supported and clearly described. If a different health endpoint is recommended for
deriving the IUR, please identify this endpoint and provide scientific support for this choice.
Lung cancer and mesothelioma are entirely appropriate endpoints for derivation of the IUR. They are
scientifically supported and clearly described. Mesothelioma is caused by asbestos exposure. While it is
possible to consider an alternative model focused on mesothelioma alone to derive the IUR, the number
of deaths from mesotheliomas is small and this would likely understate the overall cancer risk.
Since determining the cancer outcome from mortality rather than incidence data may have resulted in an
undercount of both cancer outcomes, the discussion would benefit from more detail on how the use of
incidence data could impact the derived IUR. In addition, the mesothelioma outcome may be
underrepresented because the cohort has been followed for 25 to 46 years, and lag times from exposure
19
-------
to detectable disease onset range from 15 to greater than 60 years. Mesothelioma also may have been
underreported on death certificates. Under-represented outcomes could lead to an underestimated IUR.
While there is sufficient information for derivation of the IUR, revisiting derivation of the IUR after
additional follow-up is warranted. Additional follow-up of both the occupationally and environmentally
exposed populations should be encouraged.
The report mentions laryngeal (n = 2) and ovarian (n = 0) cancer deaths in the text. The number of
deaths from these two cancer types should be included in Table 5-6. The International Agency for
Research on Cancer (IARC) concluded that there was sufficient evidence in humans that some types of
asbestos were causally associated with cancer of the larynx and the ovary as cited in the publication by
Straif et al. (2009). However, no ovarian cancer deaths and only 2 laryngeal cancer deaths were
observed in the Libby worker cohort.
Recommendation:
Tables 5-6 and 5-8 are mis-titled, since the tables include the number of deaths from
mesothelioma and lung cancer as well as demographic and exposure data. The titles should either
be changed and additional causes of death included in the tables or new tables should be created
that focus on the causes of death. Provision of data on other major categories of mortality,
including numbers of COPD, cardiovascular, colorectal cancer and other cancer deaths, could
provide useful information on the representativeness of the mortality experience of these cohorts.
3.2.4.4. Use of Laboratory Animal and Mechanistic Studies
Question 5. The database of laboratory animal and mechanistic studies of Libby Amphibole asbestos is
summarized in this draft assessment (see Section 4.2 and 4.3, details in Appendix D) to inform the
mechanisms of the biological response to Libby Amphibole asbestos and support the epidemiology
studies used for derivation of the IUR. Please comment on the use of laboratory animal and mechanistic
information in the draft assessment.
The SAB agrees, with some exceptions, that the database of laboratory animal and mechanistic
studies pertaining to LAA is appropriately presented for support of the analysis of the human effects
observed. These studies are informative in identifying similar mechanism and progression of
pathological changes in animals as are seen in humans, and help in establishing that similar
pathological endpoints are seen with other amphibole fibers. Although the mechanistic studies fall
short of delineating a complete mode of action, they are useful in identifying some common themes
and potential key mechanisms in asbestos toxicity and will undoubtedly be valuable in guiding
future research on this topic.
It is generally accepted that the toxicity and carcinogen!city of mineral and synthetic vitreous fibers
are governed by fiber dimensions, in vivo durability, and dose, and that all long amphibole fibers are
very durable in vivo. Thus, the differences in biological potency among the various amphibole fiber
types are due primarily to their differences in dimensions, especially in their fiber length
distributions (Berman, 2011). The SAB noted that the text in Sections 4.2 and 4.3, and the tables
cited therein, are deficient in not citing all that is known about the dimensions of the administered
fibers.
20
-------
Recommendations:
Section 4.2 should start with a discussion of the relevance of routes of exposure, and then should
proceed to discuss inhalation data, followed by a discussion of data from other, less relevant
routes of exposure.
Areas of needed improvement in the report include: (1) a discussion on known determinants of
fiber toxicity; and (2) the differences in fiber size distributions between LAA and other known
amphiboles.
Section 4.6.2.2 should be modified to reflect that there are insufficient data to determine if a
mutagenic mode of action for LAA is supported.
3.2.5. Inhalation Reference Concentration (RfC)
3.2.5.1. Estimates of Human Exposure Concentration
Question 1. Exposures to Libby Amphibole asbestos for workers in the Marysville, OH facility were
reconstructed based on industrial hygiene data collected in the facility from 1972 to 1994. Exposures
from 1957 to 1971 were estimated based on extrapolation from the available industrial hygiene data.
The information used for the exposure reconstruction was based on employee interviews, court and
company records, and the expert judgment of the researchers. Is the methodology used for the exposure
reconstruction reported in Appendix F and the subsequent development of exposure estimates used in
the analyses scientifically supported and clearly described?
The approach described in the Appendix F of the EPA document is detailed and specific. The strengths
and weaknesses of the approach are clearly laid out. Large uncertainties are associated with the
unmeasured pre-1972 exposures: subjectivity of workers' estimating relative concentrations, and
unsupported weighting of Libby/South Carolina fiber concentrations. Hence the report appropriately
eliminates this set of estimates and adheres to only measured exposures for its derivation of RfC.
The development of cumulative exposure estimates for the workers in a retrospective study has as its
goal the estimation of the area under the curve of the plot of each individual worker's annual exposure
concentration vs. time (calendar year), producing a summary metric of cumulative fibers/cc-years. In
Appendix F of the EPA document, the authors report using the natural-log-transformed exposure data to
calculate the geometric mean for the job groups for use in developing the cumulative exposure metric.
This approach could introduce bias by decreasing the significance of the highest exposures if the
sampling data represent a random sample of the true underlying distribution of exposures. However,
most company industrial hygienists historically have focused sampling on evaluating compliance using a
methodology that targets the worst case or "most exposed" workers (NIOSH, 1977; Mulhausen and
Damiano, 1998). In such a case, use of the mean of the unlogged data, or preferably the minimum
variance unbiased estimator (MVUE) of the mean (Attfield and Hewett, 1992), would overestimate the
most likely exposure of the average worker. The EPA should re-evaluate the raw exposure data and
review pertinent sampling documentation to bolster its use of the geometric mean to represent the job
group exposures, rather than an estimate of the arithmetic mean, and consider whether a sensitivity
analysis using the MVUE of the mean is warranted in the development of the cumulative exposure
metric.
There should be a table summarizing the changes in proportion of each type of vermiculite used (South
Carolina, Libby and African) at the Marysville plant throughout the time frame represented by the
cohort. This section should explicitly discuss the fact that Libby vermiculite usage ended in 1980, and
21
-------
that the fiber counts used in the cumulative exposure calculation for the production workers, though
small, are generally 1.5 to 6.3 times higher than background. These fibers are presumably from
combinations of African/Virginia/South Carolina vermiculite that were used from 1980 to 2000.
Likewise, the description of the calculation of the cumulative human equivalent exposure concentration
(CHEEC) in Section 5.2.3.1 would benefit by addition of a version of the material on page F-19 to
clarify the correction factors and breathing rate adjustments made due to extended work hours during
some seasons. The approach used has the typical drawbacks of oversimplification of breathing rate (one
size fits all) but is consistent with previous EPA approaches.
The SAB recommends that the EPA consider sensitivity analyses of additional exposure metrics such as:
no exposure since 1980 in any cohort members (based on end date of processing of Libby vermiculite),
and alternative weighting schemes [particularly ones weighting earlier life exposures more heavily given
the importance of time since first exposure, e.g., residence time weighting (RTW)]. These sections also
could be enhanced by showing relationships between the exposure metrics, such as by scatter plots of
unlagged CHEEC vs. other measures (separately by cohort) and by adding more explanation about the
effects of lagging.
Recommendations:
Re-evaluate the raw exposure data and review pertinent sampling documentation to bolster the
use of the geometric mean to represent the job group exposure. Consider whether a sensitivity
analysis using the MVUE of the mean is warranted in the development of the cumulative
exposure metric.
Add a table summarizing the changes in proportion of each type of vermiculite used at the
Marysville plant over time. Clarify the correction factors and breathing rate adjustments made
due to extended work hours.
Consider sensitivity analyses of additional exposure metrics, particularly those weighting earlier
life exposures more heavily.
3.2.5.2. Exposure-Response Modeling
Question 2. Exposure-response modeling was conducted using the incidence of localizedpleural
thickening in workers and cumulative exposure to estimate the point of departure (POD) for derivation
of the RfC. EPA 's estimate of the POD is based upon a Michaelis-Menten model applied to the
subcohort of workers examined in 2002-2005 and first exposed to Libby Amphibole asbestos in 1972
(when measurements of fiber levels in the workplace began) or later with cumulative exposure as the
explanatory variable. Is the selection of the model scientifically justified and clearly described? Has the
modeling and the choice of a benchmark response (BMR) for the POD of 10% extra risk of localized
pleural thickening been clearly described and appropriately conducted according to EPA 's Draft
Benchmark Dose Technical Guidance (U.S. EPA, 2000b)?
This response focuses on the primary analysis of the Marysville subcohort. Additional comments on the
analysis of this cohort can be found in response to Question 4 in Section 3.2.5.4. The SAB found that the
various exposure-response models that were examined were reasonably well described. However, the
SAB recommends a clearer description of how the "best" model was chosen. It appears that EPA fits a
series of quantal response models, retained models with adequate fit according to the Hosmer-
Lemeshow test (presumably based on p > 0.1, but, if so this should be stated). Then, among the retained
models, the authors selected the model with the lowest Akaike Information Criteria (AIC). From a
22
-------
statistical standpoint, this methodology can be justified. However, it is not clear how well aligned it is
with the guidance for selection of the POD in the updated version ofEPA's Benchmark Dose Technical
Guidance (USEPA, 2012). Thus the SAB recommends the EPA revise the approach to be better aligned
with the Benchmark Dose Technical Guidance document.
Consistent with the tone of the Benchmark Dose Technical Guidance (USEPA, 2012), the SAB
recommends that a thoughtful approach to model selection be used, including consideration of
biological/epidemiological plausibility, and desirable model features, combined with careful
examination of the data, model fit, and application of the AIC. The SAB highlights the following points:
Model fit (visual comparison of model predictions to data and/or local smoother estimates from
data) in the region of the benchmark response rate (BMR) should play a role in model selection.
The fitted Michaelis-Menten model has an upper plateau of 60% LPT incidence, while a study of
highly exposed asbestos insulation workers reported a prevalence of 85% (Lilis et al., 1991a).
The Marysville cohort does not support precise estimation of the plateau. Thus, EPA should
consider fixing the plateau at a level justified by the literature.
Other exposure metrics besides the simple cumulative exposure, such as time weighting of
exposures, should be considered. The dichotomous Hill model is attractive because it allows
estimation of an exposure parameter (b in Table 5-4), allowing the exposure effect to scale as
covariates are added, the exposure metric changed, or the plateau fixed.
The authors explain that their choice of a 10% Extra Risk (ER) as the BMR is in line with the EPA's
Benchmark Dose Technical Guidance. However, that rate is generally considered to apply specifically to
the analysis of quantal datasets from animal studies, which is the context in which it was developed. In
the EPA's Benchmark Dose Technical Guidance, it is mentioned that a BMR of 1% ER is typically used
for human quantal response data because epidemiologic data often have greater sensitivities than
bioassay data. The authors should explain what features of the dataset or outcome variable led them to
choose a BMR that is considerably greater than the norm for epidemiologic data.
Recommendations:
Consider model features and balance plausibility, localized fit, and EPA's 2012 Benchmark Dose
Technical Guidance (USEPA, 2012) when choosing the best model and explain decisions in
more detail.
In conjunction with updating and better justifying the primary analysis, evaluate the impact of
different time weightings of the exposure metric.
Either lower the BMR to be more consistent with common practice for epidemiological data or
provide more justification for the 10% BMR used to calculate the POD.
23
-------
3.2.5.3. Alternative Modeling Approaches
Question 3. EPA 's assessment also provides the results of alternative modeling approaches to derive a
POD for localized pleural thickening. This modeling used the full Marysville worker data set with
exposures from 1957 and later and a Cumulative Normal Michaelis-Menten model that incorporates
both cumulative exposure and time from first exposure as explanatory variables. Please comment on
whether EPA 's rationale for presenting these alternative approaches is scientifically justified and
clearly described. Please identify and provide the rationale if a different approach for identifying the
most appropriate population within the cohort of Marysville workers is recommended as the basis for
estimating a POD.
The SAB notes that this question applies to the full Marysville cohort. The SAB agrees that the rationale
for performing additional analyses of the full Marysville cohort is scientifically justified and that the
analysis of the entire cohort increases the number of cases of LPT available for analysis and
substantiates the primary RfC estimate derived from the subcohort.
However, the SAB does not find the rationale for the analysis approach to be well justified and it
recommends that the full cohort analysis be redone. With respect to the approach:
It is not clear that the scientific basis of using time since first exposure (TSFE) is well founded.
EPA should consider what TSFE is supposed to be measuring and how it is related to other
variables in the dataset (specifically age and exposure). There is some suggestion in the draft
document that in this dataset it is a surrogate measure of intensity since people with larger TSFEs
would be more likely to have been exposed to higher levels of LAA present during the early time
periods. This perspective should help identify modeling options.
The SAB also finds that the method for incorporating TSFE into the full cohort analysis is not
well justified. Currently, the EPA uses TSFE as a predictor for the plateau in the Cumulative
Normal Michaelis-Menten model. No biological justification is given for why this maximum
proportion would vary with TSFE.
Regarding revisions to the analysis, the SAB recommends that in this dataset a more natural way to
incorporate TSFE into the model would be to allow TSFE to affect the rate of change in the probability
of LPT by: (1) including it directly in the linear predictor portion of the model alongside cumulative
exposure; and/or (2) using an alternative exposure metric such as residence time weighting (RTW) that
more heavily weights exposure in the distant past. The functional form of TSFE could then be selected
using standard approaches (e.g., comparing AICs). Since adding TSFE to the model should affect the
coefficient of cumulative exposure, the EPA should consider a dichotomous Hill model which allows an
exposure parameter (b in Table 5-4) to be estimated, as an alternative to the Michaelis-Menten model.
Finally, the SAB recommends that other changes to the analysis follow the approaches used for the
subcohort analysis, such as fixing the plateau using literature values as recommended in the response to
charge question 2 in Section 3.2.5.2 of this report.
The SAB notes that in principle it may be preferable to base the RfC on an analysis of incidence rather
than prevalence data. Because of the nature of the dataset, the Marysville cohort does not support a
direct analysis of incidence. While it may be possible to fit an alternative model derived from integration
of a plausible incidence model (e.g., see Berry et al., 1979; Berry and Lewinsohn, 1979; Paris et al.,
2008), this approach will require a number of untestable assumptions, particularly given the small size
24
-------
of the Marysville cohort. In lieu of conducting such an analysis, the SAB recommends that an explicit
acknowledgement be added to the report regarding the implications of various model alternatives.
Recommendations:
Improve the scientific justification for using TSFE in the full cohort analysis; this justification
will include an explanation of its meaning in the context of this dataset.
Revise the full cohort analysis to change the approach to incorporating TSFE, removing it from
the model of the plateau. As part of the revision, the SAB suggests assessments be made to
determine whether it is appropriate to use (a) the dichotomous Hill model, (b) TSFE in the linear
predictor alongside cumulative exposure and/or use an alternative exposure metric that explicitly
incorporates TSFE, and (c) the approaches recommended for the subcohort such as a fixed
plateau. As appropriate, such analyses should include assessment of the functional form of
TSFE.
The SAB recommends that the EPA present the lower 95% confidence limit of the benchmark
concentration (BMCL) estimates from a set of reasonable and plausible models, and selections of
data, which will both inform selection of a preferred model and illustrate the range of model
uncertainty.
3.2.5.4. Potential Confounders and Covariates
Question 4. EPA has evaluated potential confounders and covariates where data are available.
Specifically, EPA has explored the influence of age, body mass index, smoking status, time since first
exposure, gender, and alternative exposure metrics on model fit and evaluated their association with the
modeled health outcomes (see Section 5.3). Are these analyses clearly described and appropriately
conducted? Are the results of these analyses appropriately considered in the RfC derivation?
Additionally, there is a possibility of exposure-dependent censoring in participant selection for the
update of the Marysville cohort (Rohs et al, 2008) but no evidence of selection bias. Does the SAB have
any specific recommendations for evaluating and, if appropriate, quantitatively addressing exposure-
dependent censoring in these analyses?
The SAB recommends a revised strategy for evaluation of covariates. The target of inference for the
analyses of the Marysville cohort is the POD, which in this case is the BMCL. The evaluation of the
various covariates should be made with respect to this target of inference. The SAB suggests the
covariates fall into two classes: exposure-related covariates (various exposure metrics and TSFE) and
non-exposure-related covariates [age, body mass index (BMI), gender, and smoking status]. We provide
recommended revised strategies for considering these two classes of covariates that follow directly from
consideration of the target of inference.
Non-exposure-related covariates: A decision on whether to control for the non-exposure-related
covariates should account for how the EPA wishes to determine and apply the RfC. The SAB suggests a
BMCL most directly applicable to all members of the general population is most appropriate. This
implies that the BMCL should be estimated from a model that includes exposure covariate(s), but that is
otherwise unadjusted. This is the same approach used in the current draft document; only the rationale
for the approach is different. The SAB suggests it would be informative to conduct sensitivity analyses
to examine how the BMCL varies across subgroups defined by covariate values (e.g., older males or
smokers). Because the Marysville subcohort is a small dataset, it is difficult to conduct this evaluation
exclusively in the subcohort. Therefore the SAB suggests that the EPA use the full cohort for the model
selection and parameter estimation components of sensitivity analyses incorporating these covariates.
25
-------
For this activity the EPA would use its selected final model after excluding all exposure variables (e.g.,
the dichotomous Hill model with fixed background, fixed plateau, and after dropping exposure
variables). After fitting a model with a specific set of non-exposure-related covariates in the full cohort,
one can estimate a "risk score" (i.e., the linear predictor for the non-exposure-related covariates). This
risk score would be included as a single term (as either an unsealed offset or scaled by its estimated
coefficient) in the subcohort analysis. Similar to the approach presented in Table E-5, these analyses
can be used to produce a new table of subgroup-specific conditional BMCLs; these values will give
some evidence of how the target of inference varies by subgroup. In addition, weighted averages of the
conditional BMCLs can be computed to reflect population average BMCLs for specific covariate
distributions in target populations. For instance, Gaylor et al. (1998) gives a formula for the upper tail of
a 95% confidence interval and this formula can be extended to obtain BMCLs for weighted averages.
Exposure-related covariates: The inclusion of exposure-related covariates in the model is fundamental to
the inference. The EPA has done excellent preliminary work, and the SAB has provided
recommendations in Sections 3.2.5.2 and 3.2.5.3 of this report about how to revise the approach. In
addition the SAB recommends that the EPA consider taking several further steps. First, alternative
exposure metrics should be assessed directly in the subcohort dataset to determine whether they fit the
data better. In particular, alternative metrics (such as residence time weighted exposure) that more
heavily weight more distant exposure may be more biologically plausible because individuals exposed at
an earlier age might be more susceptible to the damaging effects of asbestos. Second, TSFE should be
considered for addition to the model. Since TSFE is complete and equally well estimated across all
members of the cohort, the full cohort can be used to determine how to model this variable. Similar to
the approach recommended for the sensitivity analyses discussed above, this would be done using the
model intended for the subcohort, but omitting exposure variables other than TSFE. Then, the functional
form of TSFE selected using the full cohort can be added to the subcohort analysis, either as an unsealed
offset term or as a scaled covariate. Given biological understanding of the disease process, for models
with both estimated exposure and TSFE included, it would be appropriate to report the BMCL
conditional on a large TSFE.
Additional comments on covariates:
BMI: In section 5.2.3.3.1., it would be helpful if the justification for considering BMI as a
covariate were briefly explained. It is included elsewhere, but readers may have missed it.
TSFE:
o TSFE deserves careful consideration for both biological and dataset-specific reasons. It is
an important determinant of LPT both because individuals' lung tissues exposed at an
earlier age might be more susceptible to the damaging effects of asbestos and because
asbestos' effect over time is increasingly damaging. It is correlated with exposure in this
dataset since subjects with the longest TSFE were exposed in the early years of the cohort
when exposures were higher. It is also more accurately estimated than exposure.
o The SAB does not agree with the use of the Cumulative Normal Michaelis-Menten
model to adjust for TSFE because it makes the assumption that the TSFE only affects the
plateau. This has not been justified biologically or in the context of features of this
particular dataset. Instead, the SAB recommends that EPA consider alternative
approaches to account for TSFE.
26
-------
Smoking:
o Smoking is included in the follow-up by Rohs et al. (2008). However, the ever/never
categorization of smoking is much less informative than the pack-year analysis of
smoking used in the earlier study by Lockey et al. (1984).
o There is an important discussion of the evidence linking pleural changes and smoking in
footnote 34 on page 5-46. This information could be moved into the body of the report,
and amplified somewhat. A table summarizing the relevant studies (irrespective of type
of amphibole asbestos) summarizing the evidence regarding the role of smoking would
be useful.
Gender: There is little discussion of gender, except in places where the number of females is
listed as too few to analyze in any detail. The SAB did not regard this as a serious concern
because it is reasonable to assume that females and males have similar probabilities of
developing LPT.
The SAB recommends that a table be included summarizing the results of the various sensitivity
analyses and how they change the POD.
Exposure-dependent censoring: The exposure-dependent censoring discussion is based on results from
Rohs et al. (2008) that inappropriately separated deceased non-participants from the remaining non-
participants. Once all non-participants are combined there is no evidence of exposure-dependent
censoring. Furthermore, exposure-dependent sampling by itself does not lead to bias in risk estimates.
The important issue for bias is whether two individuals with the same exposure, one diseased and the
other not, are equally likely to participate in screening. There has been no strong rationale presented that
would indicate that such differential selection has occurred in this cohort.
Recommendations:
Revise consideration of covariates to focus on their impact on the target of inference.
o For non-exposure-related covariates, this only alters the presentation; no additional primary
analyses are needed. Sensitivity analyses conditional on subgroups defined by covariates can
be added.
o For exposure-related covariates, additional work is needed to refine the models to consider
alternative exposure metrics, as well as the inclusion of TSFE or other time-related variables
in analyses of the full cohort. The SAB encourages the EPA to either fully justify analyses
based on the Cumulative Normal Michaelis-Menten model in the context of this particular
dataset, or replace them.
Revise this discussion of Rohs et al. (2008) to make note (perhaps in a revised table) that the
dose distribution in participants is similar to the overall dose distribution of the original full
cohort. Furthermore, revise the discussion of exposure dependent sampling to distinguish this
from bias differential sampling in the sense above.
27
-------
3.2.5.5. Conversion from Cumulative Occupational Exposure to Lifetime Exposure
Question 5. The modeled POD estimate is based on cumulative exposure estimates for the worker cohort
examined. For the derivation of the RfC, this cumulative exposure is prorated over the period of
environmental exposure (lifetime or shorter duration chronic exposure when appropriate). The RfC is
provided in units of continuous air concentration. Is the basis of this conversion clearly explained and
scientifically justified?
The SAB agrees that the conversion is clearly explained and follows standard practice. However, the
SAB recommends a revision to use the full 70-year lifetime in the conversion rather than 60 (70 minus
the lag of 10 used for exposure in the POD derivation). Given that the cumulative exposure metric
(concentration in fiber/cc x years of exposure) is arbitrarily related to the prevalence data, lagging does
not have real meaning as it would in an incidence data analysis since the time of disease onset is not
known. Using a divisor of 60 instead of 70 in deriving the RfC is less protective.
Recommendations:
Use the full 70-year lifetime when converting cumulative to continuous exposure; i.e., do not
correct for the lag of 10 for a 10-year lagged exposure.
The SAB recommends EPA indicate more clearly in Section 5.2.3.1. that "year" is in the
numerator in the exposure metric "fibers/cc-year", and to describe more clearly how cumulative
exposure is derived.
3.2.5.6. Selection of Uncertainty Factors
Question 6. Please comment on the rationale for the selection of the uncertainty factors (UFs) applied to
the POD for the derivation of the RfC. Are the UFs appropriate based on A Review of the Reference
Dose and Reference Concentration Processes (U.S. EPA, 2002; Section 4.4.5) and clearly described? If
changes to the selected UFs are proposed, please identify and provide scientific support. Specifically,
please comment on the rationale for the selection of the database uncertainty factor (UFp) of 10 applied
in the derivation of the RfC. The database uncertainty factor accounts for the lack of data on effects
other than in the respiratory system, including other effects observed in community and laboratory
animal studies (cardiovascular disease and autoimmune effects) that have not been well-studied (See
Section 5.2.3 of the Toxicological Review); and lack of health data assessed at later time points. Is the
rationale for the UFp appropriate and clearly described? Please provide the rationale if a change in
the UFo is proposed.
Uncertainty factors were selected in accordance with the usual procedures laid out in EPA risk
assessment guidelines. A value of 10 was selected for UFn (human inter-individual variability) and
(database uncertainty), with a value of 1 for all others.
Use of a UFn of at least 10 is standard in considering health protective levels based on effects in the
workforce, which is generally healthier and less diverse than the general population. In fact,
publications are available that discuss whether a factor of 10 is sufficient to cover all sensitive sub-
populations, especially children (Hattis et al., 1999; Miller et al, 2002; Scheuplein et al., 2002; Dourson
et al., 2002; OEHHA, 2008). Some treatment of the question of inter-individual variability is offered in
the later summary of conclusions (Section 6 of the EPA document). There is no specific evidence on the
relative sensitivity of children to the non-cancer effects of Libby asbestos, although some indications
with other amphiboles suggest the possibility of enhanced effects following exposure at younger ages
(Haque et al., 1996; 1998; Isaacs and Martonen, 2005; Bennett et al., 2008). Overall, it seems unlikely
28
-------
that a departure from the default guideline value of UFn =10 could be justified within the existing
guidelines, but concerns remain for the impact on susceptible subpopulations, especially women and
children.
EPA explains and justifies the selection of a UFo of 10 based on the limited number of studies of
exposure to Libby asbestos (Libby workers, ATSDR community study and Marysville workers) and the
lack of evaluation of potentially more sensitive alternative endpoints. The SAB finds that this
uncertainty factor would not be reduced even if improved exposure estimates allowed consideration of
the full cohorts (or a larger fraction thereof).
However, some additional data have recently been published for the community surrounding a
Minnesota expansion plant (Adgate et al., 201 1; Alexander et al., 2012). Although there appears to be a
rationale for at least an initial consideration of LAA as a unique material (to provide an unbiased
comparison with other amphiboles), this SAB review has identified very substantial grounds for
considering this material as having composition, physical properties, and biological effects that are very
similar to those seen for other amphiboles. The most relevant comparison would be to tremolite, since
Libby Amphibole is -6% tremolite, an amphibole that is known to cause cancer and non-cancer effects
in human populations. However, it is uncertain how other components of Libby Amphibole (richerite
and winchite) interact as a mixture with tremolite to modify toxicity. This consideration of data on other
amphiboles is particularly pertinent to discussions of the mode of action, as well as the exposure-
response relationships, for Libby Amphibole. In light of this similarity it appears reasonable, and indeed
necessary, to at least debate the question of whether the available data on non-cancer health effects of
amphiboles are sufficient to mitigate the acknowledged data shortage for Libby Amphibole itself.
Therefore, the SAB considers that additional data (e.g., the Minnesota cohort and data on other
amphiboles) might support a lower value, such as 3, for
On the other hand, there are substantial remaining uncertainties that are not addressed by these
additional data, including those raised by consideration of the severity of the endpoint and the selection
of the BMR (see below). This uncertainty should also be revisited by EPA in its judgement of an
uncertainty factor of 1-fold for a LOAEL-to-NOAEL uncertainty factor (UF^. It can also be argued that
a subchronic-to-chronic uncertainty factor (UFs) higher than 1 should be used, given that the mean and
maximum exposure duration in this study are both well below the lifetime exposure of interest. This
uncertainty should also be revisited for EPA in its judgement of an uncertainty factor of 1-fold for UFs.
It may be appropriate for EPA to select a value of 10 for UFo, or a similar uncertainty spread across
several factors, but EPA needs to re-evaluate selection of this factor explicitly once all the additional
information has been incorporated in the discussion.
There is a concern that the BMR of 10%, which was chosen for a severe endpoint, is not reflected by the
choice of a UFL of 1 . The SAB finds that it is appropriate for EPA to consider either a lower BMR, or
the application of a larger UFL for this endpoint. An argument could be made that some allowance has
been made for this concern in the choice of the UFo, but it is debatable whether this is sufficient, given
the other matters to which that UF is also assigned. In any case, UFo is not the appropriate place to
consider this uncertainty. At the very least, this question deserves more consideration and analysis that it
receives in the draft assessment report.
29
-------
Recommendations:
Review additional data, in particular the exposure-response relationship for non-cancer endpoints
in the Minneapolis community cohort.
Determine whether this new analysis supports the existing analysis based on the Marysville data,
and if so whether this warrants reduction of the value of UFo since the limited data basis for the
original analysis has been expanded.
Reassess the selection of the BMR to reflect the severity of the chosen endpoint in the Marysville
cohort and the precision available in the data. Whether or not the chosen BMR is changed,
present this analysis in the document rather than simply asserting that a "default" value for the
BMR was chosen. Similar consideration should be applied to the Minneapolis cohort to provide
a valid comparison. This consideration needs to be linked to discussion of the selection of a
value for UFL as noted below.
Review additional sources of uncertainty:
o Timescale of cohort coverage, normally addressed by UFs if this is a significant concern
rather than including this as a component of UFo which already has several major issues to
account for.
o Additional uncertainty resulting from target population diversity (including women and
children, specific sub-populations of concern not represented in the cohort), and endpoint
severity.
Consider adjusting UFo, UFs or UFL if necessary to accurately reflect the overall uncertainties in
these categories: provide specific justification for the choices made rather than claiming
unsupported use of default values.
3.2.5.7. Characterization of Uncertainties
Question 7. Please comment on whether the document adequately describes the uncertainties and
limitations in the methodology used to derive the RfC and whether this information is presented in a
transparent manner.
In the report there are two sections on uncertainty for the RfC: an application of uncertainty factors
following standard EPA practice (Section 5.2.4), and a discussion of the uncertainties in the overall
methodology and approach (Section 5.3). This response focuses on the latter. Overall the SAB found the
discussion to be thorough, detailed and logical. The document can be improved by harmonizing the full
set of uncertainty discussions, including both the discussion of RfC uncertainty and the related
discussion of the IUR uncertainty (see the SAB response to question 5 under Section 3.2.6.5 below). In
addition, the RfC uncertainty assessment can be strengthened. A key consideration of any assessment is
whether the estimated RfC is adequately protective of public health. The SAB recommends that
additional work be done to substantiate the RfC estimate through additional sensitivity analyses and
discussion of results and insights from other datasets [e.g., cause of death for the deceased non-
participants in Rohs et al. (2008) and the Minneapolis exfoliation community cohort (Alexander et al.,
2012)].
In considering other studies, the appropriate assumption is that LAA fibers have the same mechanisms
of toxicity and quantitative risk relations as that of other asbestos fibers. In sensitivity analyses, consider
alternative exposure metrics (prioritizing residence time weighted metrics and excluding exposures after
1980), methods to fine-tune the RfC estimate from the subcohort (particularly fixing rather than
estimating the plateau, allow the slope parameter to be estimated, use a lifetime of 70 regardless of the
exposure metric), and added sensitivity analyses in the full cohort using suggestions from the SAB.
30
-------
With respect to exposure assessment, analytical methods and environmental conditions are substantial
contributors to uncertainty because of differences between the 1970s and today. As discussed throughout
the report, PCM was the only generally accepted method for measuring airborne fiber concentrations
used until the 1980's. PCM's limitations are well-detailed in the report: an inability to detect fibers
smaller than 0.25 jim, an inability to differentiate asbestos fibers from other fibers, and a limitation to
counting only fibers longer than 5 jim. Today, TEM can easily detect and positively identify airborne
asbestos of all sizes. But, because the RfC is based on 1970's PCM analyses, the RfC must be
implemented in a way that most closely replicates analysis in the 1970's. At the 1970's study site, the
vast majority of measured fibers were almost certainly LAA, so PCM's inability to identify asbestos did
not create much uncertainty. Today, even ambient air will yield fiber concentrations that exceed the
RfC. The culprit fibers will likely be cellulose fibers from cotton, wood, paper or synthetic fibers, rather
than asbestos. Hence, today's PCM counts will be from fibers that are unrelated to the RfC. Thus it is
important that TEM be used to identify and count asbestos fibers in air samples for RfC purposes.
Finally, Page 5-118, Lines 22-33 of the EPA's draft document discuss the two-fold under-reporting of
fibers because of PCM's poorer resolution in the 1970's, 0.44 jim versus 0.25 jim today. Because
today's PCM analysts have no capability for discriminating fibers > 0.44 jim, the need for TEM analysis
of samples collected for implementation of the RfC is even more important. A TEM protocol for PCM
equivalent fibers wider than 0.44 jim could be easily developed.
Recommendations:
Harmonize the uncertainty discussions across the document.
Substantiate the RfC estimate through
o Additional sensitivity analyses of the subcohort;
o Discussion of results from other studies;
o Additional sensitivity analysis of the full cohort; and
o Summarizing in tabular form the results of the various sensitivity analyses and model
alternatives, to show how they affect the POD.
Use TEM to identify and count asbestos fibers longer than 5, 10, and 20 jim in air monitoring
samples for implementation of the RfC.
3.2.6. Inhalation Unit Risk (IUR)
3.2.6.1. Exposure-Response Modeling
Question 1. Exposure-response modeling was conducted separately for lung cancer and mesothelioma
mortality. The POD estimates for these endpoints are based upon analysis of the subcohort of workers
first exposed after 1959 when the exposure data were judged to be better characterized. The exposure-
response modeling included consideration of a variety of exposure metrics that varied with time and
incorporated different lag and decay parameters. Based on the results of the exposure-response
modeling, a life table analysis was used to determine the PODsfor each type of cancer for the various
exposure metrics. Have the exposure-response modeling and determination ofthePODsfrom life table
analysis been appropriately conducted and clearly described? If a different approach to exposure-
response analysis is recommended as the basis for estimating the IUR, please identify the recommended
methods and provide a rationale for this choice.
31
-------
In general, the EPA clearly described the methods it had selected to conduct the exposure-response
modeling for lung cancer and mesothelioma. The risk calculations in the life tables appeared correct but
would benefit from clearer explanations. Some suggestions for clarifications are noted below.
The agency was overly constrained by reliance on model fit as the primary criterion for model selection
and the SAB recommends a broader discussion of biological and epidemiological criteria as well. For
the mesothelioma data, for example, the Peto model was disregarded due to a poorer fit than the Poisson
model. The results for this analysis are not shown and, given the particular interest in this model, should
have been. A parametric survival model (e.g., Weibull) could have also been used to obtain estimates of
absolute risk. It would also be appropriate to compare the results of the final model against those from
fitting a two-stage clonal expansion (TSCE) model. Use of the TSCE model would allow for a more
direct evaluation of, and possibly justification for, age-dependency of the IUR. The Richardson (2008)
paper provides a publicly available and transparent approach to application of the TSCE. Ultimately,
there are many competing models that could have been used instead of the Poisson and Cox models
(e.g., parametric survival models, accelerated failure time models, additive models) that could have
provided very different estimates of risk, but they were not discussed.
Data exist that suggest that the lifetime risk of developing the mesothelioma increases the earlier in life
that exposure is first received. The Peto model (Peto, 1979; Peto et al., 1982) was developed to explain
such observations in the empirical data. While the Peto model has been more widely used for risk
assessment, most notably in the previous IRIS summary for asbestos, it has also only been formally
fitted to data in a limited number of cohorts (HEI-AR, 1991). Ongoing analysis of incidence of
mesothelioma appears to be consistent with the exposure-response relationship described in the Peto
model. The draft report needs to do a more complete job of justifying why this and other epidemiologic
evidence should be excluded as a basis for selection of a plausible model for predicting mesothelioma
risk. Chapters 2 and 3, for example, consider toxicological and other evidence developed with exposures
to asbestos that are not strictly LAA. The cohorts used in the development of the Nicholson/Peto model
and the exposures they experienced should provide information about the time course of the
development of disease.
The SAB recognizes that the agency's effort to focus on good quality exposures specific to LAA has led
to reliance solely on the Libby worker subcohort. This rationale is understandable, but at the same time,
it is important to acknowledge that this small subcohort may have its own limitations as a basis for
modeling exposure-response relationships for a larger population over a lifetime. As a sensitivity
analysis to evaluate the potential impact of omitting the Libby workers hired before 1959, the SAB
recommends analyzing the entire Libby cohort using interval statistics (Nguyen et al., 2012; Manski
2003; inter alia) or other traditional approaches for data censoring in predictors (cf. Kiichenhoff et al.,
2007). It can be misleading to use midpoint substitution (as described in Section 5.4.6.1.2) that assumes
poorly measured or missing predictors have some constant value. Interval statistics and traditional
censoring approaches to measurement uncertainty would, in essence, replace point values with interval
ranges. When the intervals are narrow, as they might be for 21% of the early hires for which jobs titles
are available, there might be a good deal of recoverable information present. When the intervals are
much wider, there would be accordingly less information. Whatever empirical information may be
present, it is worth evaluating whether its inclusion is better than leaving out the data entirely, which in
principle amounts to replacing them with intervals that are completely vacuous, from zero to infinity.
This approach can produce an interval range for the final outputs, which would provide the explicit
quantitative uncertainty statement as recommended by previous National Academy of Sciences reviews.
32
-------
The SAB recognizes that the agency did conduct sensitivity analyses with several analyses of the Libby
cohort data, including those that used different models (Tables 5-20 for lung cancer and 5-21 for
mesothelioma). A limitation of these analyses is that they all rely on the assumption that the effect of
exposure can be modeled as a function of cumulative dose. This assumption is consistent with the
agency's Guidelines for Carcinogen Risk Assessment (USEPA, 2005), which state that "unless there is
evidence to the contrary in a particular case, the cumulative dose received over a lifetime, expressed as
an average daily exposure prorated over a lifetime, is recommended as the appropriate measure of
exposure to a carcinogen." EPA therefore did not address the fundamental question about whether any
one model can or should be assumed to represent the exposure-response relationship for LAA.
Therefore, one cannot be confident that the "true" exposure-response relationship for LAA is really
"accounted for" by use of the upper confidence limit (UCL) on the slope (per fiber/cc) or, ultimately, the
combined IUR from mesothelioma and lung-cancer mortality (see related discussion in response to
question 3 and 5 in Section 3.2.5).
This issue is of particular concern for the estimation of mesothelioma risks from partial lifetime
exposures, where risk is essentially assumed to be independent of when in the course of a lifetime
exposure occurs. For example, one year of exposure to a given concentration in childhood yields the
same lifetime average daily dose as one year of the same exposure in adulthood. This assumption is not
consistent with the relevant body of evidence on the development of asbestos-related disease. Therefore,
there is some probability not well characterized that this approach underestimates the relative
effect of early exposure, but exaggerates the effect of exposure later in life.
Recommendations:
Two types of recommendations have been made. The first set is asking for simple explanations in the
text that the SAB thinks will clarify the rationale for analytic choices made by the EPA. The next set
includes requests for additional presentations of data or analyses, roughly in order of priority, that the
SAB concludes are important to provide some quantitative perspective on the analytic choices made.
Clarifications:
Poisson regression analyses: the mathematical form of the regression function should be given,
and discussion of whether the potential for over-dispersion was assessed.
Cox proportional hazards modeling: the reasons should be given for not conducting a Bayesian
analysis as was done for the Poisson regression model for mesothelioma.
Life-table analysis: the method used to estimate the hazard function for the exposed population
should be clearly spelled out in the text. Was it based on a nonparametric estimate of the baseline
hazard from the sub-cohort? Given that the SEER data were used to calculate the background
incidence of lung cancer, it would seem more appropriate to use those data to estimate the
baseline hazard and then to use the regression coefficient obtained from the Cox model applied
to the sub-cohort data to obtain the hazard of the exposed group. Thus, the reasons for not using
the SEER data to estimate the baseline hazard should be explained.
Expand the discussion of model selection to explain the reliance on model fit criteria for model
selection. In particular, why should the broader epidemiologic evidence on the time course of
disease not argue at least for the presentation of more than one statistical model?
33
-------
Provision of additional data or analysis:
In a tabular form, summarize the fit results, POD estimates, and IUR estimates from the full
range of models considered in order to show the dependence of the IUR estimate on model
selection.
Present the fit to data graphically for both the main models and for a broader range of models,
including the Peto model. This step would provide a more thorough and transparent view of fit,
particularly in the region of the BMR, than is allowed by examining summary statistical values
alone.
Provide in an appendix the details of the Nicholson/Peto model fit for which the text currently
states "data not shown".
Allow evaluation of the time dependence of disease by providing tabulations of mesothelioma
mortality rates and lung cancer SMRs by time since first exposure, duration of exposure and
period of first exposure (for both the full and sub-cohorts of Libby workers).
Evaluate the feasibility of conducting an ancillary analysis of the full Libby data set, including
hires before 1959, using interval statistics or other traditional censoring methods (not simple
midpoint substitution). At a minimum, discuss the possible quantitative uncertainties associated
with using the smaller subcohort.
3.2.6.2. Potential Confounding by Smoking
Question 2. Smoking is a strong independent risk factor for lung cancer and may be an important
confounder of the lung cancer mortality analysis. Data on individual smoking habits and history were
largely missing and could not be used to control for potential confounding in regression analyses.
However, EPA used three approaches to evaluate the confounding issue, including restriction of the
cohort and two analytic evaluations of the potential for confounding by smoking (see Section 5.4.3.6.5).
Please comment on whether the methods and analyses are clearly presented and scientifically justified.
If additional analyses are recommended, please identify the methods and scientific rationale.
The SAB recognizes the challenges in controlling for smoking given the lack of data on smoking
histories for the cohort. The agency has taken reasonable steps to identify the potential for confounding
using independent approaches. However, statements in the document (on p. 5-96 and again on p. 5-127)
that because the proportional hazards assumption is satisfied in the subcohort there is no evidence of
confounding by smoking, are too strong. Reaching this conclusion requires some strong assumptions,
including one that the decline in smoking prevalence observed in the general U.S. population also
occurred in the Libby cohort.
The agency's use of the Richardson (2010) method for exploring possible confounding for smoking was
appropriate. However, the conclusion that there is no evidence for confounding by smoking relies more
heavily on the ^-values, which are marginally non-significant, than it needs to. More compelling is the
observation of a negative association with COPD in their analyses. The fact that the coefficients for
exposure in the COPD Cox models were negative is strong evidence against positive confounding;
smoking is positively related to COPD risk and thus if positive confounding is occurring, then one
would also expect the relationship between asbestos exposure and COPD risk to be positive.
Recommendations:
The numbers of COPD deaths (n) in the sub-cohort that were the basis for the analysis should be
presented in the text.
34
-------
The statements about the evidence against confounding by smoking given by restriction of the
cohort should be qualified by the assumptions required to justify them, or deleted.
The SAB had no recommendations for further analyses.
The reference to three methods is confusing. There are actually only two, the restricted cohort
and the Richardson analysis for which two exposure metrics are explored.
3.2.6.3. Quantification of Inhalation Unit Risk
Question 3. In order to derive an IUR which represents the combined risk of mortality from lung cancer
or mesothelioma, a cancer-specific unit risk for each tumor type was calculated according to the
Guidelines for Carcinogen Risk Assessment (U.S., EPA, 2005; Sections 3.2 and 3.3) by linear
extrapolation from the corresponding POD (i.e., the lower 95% confidence limit on the exposure
associated with 1% extra risk of lung cancer or 1% absolute risk of mesothelioma mortality). The IUR
was then determined as a combined upper bound risk estimate for mortality considering both cancers.
Has this approach been appropriately conducted and clearly described?
The SAB found the description of the procedure used to be clear but considered the justification for the
independence assumption to be lacking in depth. The EPA should provide a discussion of the potential
consequences of assuming that the estimated lURs for mesothelioma and lung cancer mortality are
independent, noting the possibility that the upper bound on the IUR may be understated if the risks are
positively correlated. The document may refer to the 1994 NRC report, which suggested that treating
different tumor occurrences as independent is "not likely to introduce substantial error in assessing
carcinogenic potency". However, the document should acknowledge that this statement was made in the
context of animal bioassays and that human populations are more heterogeneous in risk factors related to
mesothelioma and lung cancer mortality. If any risk factors are shared across outcomes and not
accounted for in the modeling, the risk estimates generated by the different models are likely correlated.
Given the small size of the data set, and lack of an appropriate statistical method, this correlation cannot
be estimated reliably. One approach might be to undertake bounding analysis on the lifetime risk
estimates using, for example, the Frechet inequality for disjunctions (Frechet, 1935) that makes no
assumption about the nature of the dependence. This analysis could reveal how large the impact of
dependence might be. At the very least, the restrictive assumption of independence must be mentioned
and the potential consequences of a violation of this assumption must be discussed.
Recommendations:
The EPA should acknowledge that the assumption of independence is a theoretical limitation of
the analysis, and should provide a fuller justification for this assumption. EPA has cited the NRC
(1994) analysis as suggesting the impact of this issue is likely to be relatively small. This view is
also echoed in the EPA's (2005) Guidelines for Carcinogen Risk Assessment. These provide the
basis for a default assumption. However, it would be preferable if this assessment discussed the
evidence base and rationale for lung cancer and mesothelioma specifically.
As a sensitivity analysis, the EPA should consider quantitatively accounting for dependence in
the risks of mesothelioma and lung cancer mortality either using a method that models the
dependence explicitly, or a bounding study that evaluates the numerical consequences of the
assumption of independence.
35
-------
3.2.6.4. Adjustment for Mesothelioma Mortality Under-ascertainment
Question 4. Please comment on the adjustment for mesothelioma mortality under-ascertainment. Is this
adjustment scientifically supported and clearly described? If another adjustment approach is
recommended as the basis for the IUR, please identify that approach and provide the scientific
rationale.
The number of mesothelioma deaths was adjusted for under-ascertainment stemming from inadequate
coding used in death certificates. The procedure used is not described in any detail, but can be found in
the Kopylev et al. (2011) reference. A total of 18 mesotheliomas were observed in the Libby cohort
from 1980 to 2006. The estimated number of 24 mesotheliomas was obtained after using a Monte Carlo
analysis. The ratio of 24 to 18 yields the median of 1.33. The Kopylev manuscript also provides a figure
of 1.39 in Table 3, which is the mean later reported in the EPA report. The EPA method appears to be
scientifically supported, but is not clearly described. This section should be expanded and a much more
detailed statement of how the numbers were arrived at should be provided.
No additional adjustment approach is described in the EPA report. The authors should provide an
additional estimate using the 37% figure mentioned on page 46 of the Kopylev et al. (2011) reference.
This is the percentage of mesothelioma cases that would be missed using previous histopathological
analyses of cancer registry data. Using 37% would yield an estimate of about 29 mesothelioma cases
instead of 24. The median ratio would then be 1.61 instead of 1.33. This number, and its related mean,
should be utilized to provide a separate analysis of unit risk for comparison purposes.
3.2.6.5. Characterization of Uncertainties
Question 5. Please comment on whether the document adequately describes the uncertainties and
limitations in the methodology used to derive the IUR and whether this information is presented in a
transparent manner.
The SAB commends the EPA for summarizing (in Section 5.4.6.1 of the draft document) the many
sources of uncertainty considered in the course of this document and evaluating, at least qualitatively,
and sometimes quantitatively, the direction and magnitude of the likely impact of each source of
uncertainty.
However, the SAB noted that most of what the document has accomplished is through targeted
sensitivity analyses that examine one assumption at a time, while holding all others more or less
constant. For example, the agency has indeed done a thorough job of exploring sensitivity of the lURs to
a range of investigator analyses of lung cancer (Table 5-20) and mesothelioma (Table 5-21) for the
Libby worker subcohort, and to a wide range of assumptions about the exposure metrics to be used in
the basic models (e.g., Table 5-9). The basic underlying models chosen for lung cancer and for
mesothelioma are the same.
The sensitivity analyses in the document are individually well described, appear well-done and provide
reassurance, under the assumptions of the basic models and approaches chosen to estimate the IUR, that
the particular exposure metric and lag, for example, do not appear to make a big difference in the value
of the IUR. However, they are currently presented somewhat in isolation, and thus do not take into
account the magnitude and likelihood of multiple sources of uncertainty in the same analysis or address
the overall distribution of uncertainty in the IUR. Consequently, the SAB did not think that the
following statement had been fully justified:
36
-------
.. .the EPA's selected combined IUR of mesothelioma and lung-cancer mortality accounts for
both the demonstrated cross-metric uncertainty as well as several additional uncertainties,
which could have resulted in underestimates of the mesothelioma and lung-cancer mortality
risks (p 5-105, lines 1-5).
As noted in response to question 1 in Section 3.2.6.1 above, the SAB identified that model uncertainty is
an important source of uncertainty that might well not be accounted for by using the 95% UCL on the
IUR and the combined IUR or at least that had not been represented by the sensitivity analyses
provided.
Recommendations:
The SAB recommends that a more straightforward and transparent treatment of model uncertainty
would be to estimate risks using a more complete set of plausible models for the exposure-response
relationship (discussed in response to question 1 in Section 3.2.6.1), including the Poisson models.
This sensitivity analysis would make the implications of these key model choices explicit.
The SAB recommends that, as an initial step in conducting an integrated and comprehensive
uncertainty analysis, the agency provide a tabular presentation and narrative evaluation of the IUR
estimates based on a reasonable range of data selections (e.g., all or part of the earlier hires as well as
the "preferred" subcohort), model forms and input assumptions (as discussed, in the response to
question 1 in Section 3.2.5). These input assumptions should include inter alia exposure metrics and
externally defined parameters, as discussed in the response to question 1 in Section 3.2.5. As noted
in the current cancer risk assessment guidelines (EPA, 2005, page 3-29):
The full extent of model uncertainty usually cannot be quantified; a partial characterization can
be obtained by comparing the results of alternative models. Model uncertainty is expressed
through comparison of separate analyses from each model, coupled with a subjective probability
statement, where feasible and appropriate, of the likelihood that each model might be correct
(NRC, 1994).
The SAB notes that ideally, the agency would develop a quantitative characterization of the overall
uncertainty in its IUR estimates by incorporating the major sources of uncertainty the agency has
identified in its evaluation. However, the SAB recognizes the challenge of conducting such an analysis,
and is not recommending that it be undertaken at this time.
37
-------
4. LONG-TERM RESEARCH NEEDS
4.1. Epidemiology
It would be informative and very important for NIOSH and ATSDR to continue monitoring mortality
among Libby workers (including those residing in Libby and nearby towns such as Troy, Montana) and
residents of Libby and nearby towns, respectively, to determine the number of new lung cancers,
mesotheliomas, and non-malignant pulmonary diseases (i.e., asbestosis) in these two populations.
The last occupational ascertainment was through 2006; an additional five years of data should now be
available. In addition to a dose-response evaluation, an overall SMR should be calculated for lung
cancer in this population by comparison to both the Montana and U.S. populations.
The previous ATSDR community SMR mortality survey was from 1979-1998. It should now be
extended through 2011 and should include an analysis specific for community, non-occupationally
exposed, individuals. Early-life exposure to LAA could possibly be obtained from surrogate interview
information from the community population. Smoking, occupational, and residential histories should be
obtained for the lung cancer, mesothelioma, and non-malignant respiratory disease (i.e., asbestosis)
categories. Data concerning previous Libby residents who had moved away (and died in other states)
would need to be obtained by means of a special effort of ATSDR.
A community cross-sectional respiratory health screening was conducted in Libby by ATSDR in 2000
and 2001. A non-malignant respiratory health update since then would be useful. The appropriate
smoking, occupational, and residential histories should be included.
4.2. Mode of Action
It would be valuable for future research on LAA mode of action to focus on biomarkers that are
more clearly and specifically related to non-cancer endpoints (i.e., asbestosis) or cancer endpoints
(e.g., mesothelioma). Critical genotoxicity studies including mutagenesis and chromosomal
aberration studies have not been investigated with LAA. Inhalation studies in animal models that can
provide mechanistic and dose-response relationship should be conducted.
4.3. Future Development of a TEM Method for PCM Equivalency
EPA needs to develop a transmission electron microscopy (TEM) method that provides equivalent data
to phase contrast microscopy (PCM). This TEM method development must first recognize fundamental
differences between TEM and PCM analysis. Areas that need better definition include differences in
analyzable areas, changes in PCM resolution over time, measuring complex fibrous structures,
measuring obscured fibers, defining TEM analysis parameters more succinctly, recognition of several
other measurement characteristics of importance (such as surface area), defining inter-laboratory
variations and their causes, as well as other areas related to analysis.
Other areas of analysis may include but not limited to: differences between PCM reticule areas and TEM
grid opening areas that create biases; TEM rules with regard to fibers obscured by grid bars which create
positive bias in TEM results; measurement of obscured, complex arrangements of fibers by TEM that
differ from PCM counts; TEM measurement errors associated with fibers of various widths; differences
between laboratories with interpretation of TEM counting rules; differences in magnification and
orientations used for analysis; and other issues which create variation between analyses.
38
-------
39
-------
REFERENCES
Albert, RE; Lippmann, M; Briscoe, W. (1969). The characteristics of bronchial clearance in humans and
the effects of cigarette smoking. Arch. Environ. Health 18:738-755.
Amandus, HE; Wheeler, R. (1987). The morbidity and mortality of vermiculite miners and millers
exposed to tremolite-actinolite: Part II. Mortality. Am JIndian Med 11:15-26.
ATS (2004). Diagnosis and initial management of nonmalignant disease of asbestos. American J
Respiratory Critical Care Med 170:691-715.
ATSDR (2000). Health consultation: mortality from asbestosis in Libby, Montana. Atlanta, GA.
ATSDR (2001). Year 2000 medical testing of individuals potentially exposed to asbestiform minerals
associated with vermiculite in Libby, Montana: A report to the community. Atlanta, GA.
Ates,G; Yildiz, T; Akyildiz, L; Topcu, F; Erturk, B. (2010). Environmental asbestos-related pleural
plaque in southeast of Turkey. Arch Environ Occup Health. 65(l):34-7.
Attfield, MD; Hewett, P. (1992). Exact expressions for the bias and variance of estimators of the mean
of a lognormal distribution. Am IndHygAssoc J 53:432-435.
Below, JE; Cox, NJ; Fukagawa, NK.; Hirvonen, A.; Testa, JR. (2011). Factors that impact susceptibility
to fiber-induced health effects. J ToxicolEnviron Health, Part B 14:246-266.
Bennett, WD; Zemean, KL; Jarabek, AM (2008). Nasal contribution to breathing and fine particle
deposition in children versus adults. J Toxicol Environ Health A 71:227-237.
Berman, DW. (2011). Apples to apples: The origin and magnitude of differences in asbestos cancer risk
estimates derived using varying protocols. Risk Analysis 31:1308-1326.
Bernstein, DM; Chevalier, J; Smith, P. (2003). Comparison of Calidria chrysotile asbestos to pure
tremolite: Inhalation biopersistence and histopathology following short-term exposure.
Inhalation Toxicology 15:1387-1419.
Bernstein, DM; Rogers, R; Smith, P. (2004). The biopersistence of Brazilian chrysotile asbestos
following inhalation. Inhal. Toxicol. 16:745-761.
Bernstein, DM; Chevalier, J; Smith, P. (2005a). Comparison of calidria chrysotile asbestos to pure
tremolite: final results of inhalation biopersistence and histopathology examination following
short-term exposure. Inhal. Toxicol. 17:427-449.
Bernstein, D; Rogers, R; Smith, P. (2005b). The biopersistence of Canadian chrysotile asbestos
following inhalation: final results through 1 year after cessation of exposure. Inhalation
Toxicology 17:11-14.
40
-------
Bernstein, DM; Rogers, RA; Sepulveda, R; Donaldson, K; Schuler, D; Gaering, S; Kunzendorf, P;
Chevalier, J; Holm, SE. (2011). Quantification of the pathological response and fate in the lung
and pleura of chrysotile in combination with fine particles compared to amosite-asbestos
following short-term inhalation exposure. Inhalation Toxicology 23(7):372-391.
Berry, G; Lewinsohn, HC. (1979). Dose-response relationships for asbestos-related disease: implications
for hygiene standards. Part 1: morbidity. Annals New York Academy of Sciences 330:185-194.
Berry, G; Gilson, JC; Holmes, S; Lewinsohn, HC; Roach, SA. (1979). Asbestosis: a study of dose-
response relationships in an asbestos textile factory. British J of Industrial Medicine 36: 98-112.
Bourbeau, J; Ernst, P; Chrome, J; Armstrong, B; Becklake, MR. (1990). The relationship between
respiratory impairment and asbestos-related pleural abnormality in an active work force. Am Rev
RespirDis 142:837-842.
Broaddus, VC; Everitt, JI; Black, B; Kane, AB. (2011). Non-neoplastic and neoplastic pleural endpoints
following fiber exposure. J ToxicolEnviron Health, Part B 14:153-178.
Brody, AR; Roe, MW. (1983). Deposition pattern of inorganic particles at the alveolar level in the lungs
of rats and mice. Am. Rev. Respir. Dis. 128:724-729.
Brody, AR; Hill, LH; Adkins, B Jr; O'Connor, RW. (1981). Chrysotile asbestos inhalation in rats:
deposition pattern and reaction of alveolar epithelium and pulmonary macrophages. Am. Rev.
Respir. Dis. 123:670-679.
Brody, AR; Liu, JY; Brass, D; Corti, M. (1997). Analyzing the genes and peptide growth factors
expressed in lung cells in vivo consequent to asbestos exposure and in vitro. Environ Health
Perspect. 105 Suppl 5:1165-71
Carbone, M; Baris, YI; Bertino, P; Brass, B; Comertpay, S; Dogan, AU; Gaudino, G; Tube, S; Kanodia,
S; Partridge, CR; Pass, HI; Rivera, ZS; Steele, I; Tuncer, M; Way, S; Yang, H; Miller, A. (2011).
Erionite exposure in North Dakota and Turkish villages with mesothelioma. Proc NatlAcadSci
USA. 108(33): 13618-23. Epub 2011 Jul 25.
Clin, B; Paris, C; Ameille, J; Brochard, P; Conso, F; Gislard, A; Laurent, F; Letourneux, M; Luc, A;
Schorle, E; Pairon, JC. (2011). Do asbestos-related pleural plaques on HRCT scans cause
restrictive impairment in the absence of pulmonary fibrosis. Thorax 66:985-991.
Constantopoulos, SH. (2008). Environmental mesothelioma associated with tremolite asbestos: lessons
from the experiences of Turkey, Greece, Corsica, New Caledonia and Cyprus. Regul Toxicol
Pharmacol.52(\ Suppl):SI 10-5. Epub 2007 Nov 13.
Davis, JMG; Addison, J; Bolton, RE; Donaldson, K; Jones, AD; Miller, BG. (1985) Inhalation studies
on the effects of tremolite and brucite dust in rats. Carcinogenesis 6:667-674.
41
-------
Dourson, ML; Charnley, G; Scheuplein, R. (2002). Differential sensitivity of children and adults to
chemical toxicity: II. Risk and Regulation. Regulatory Toxicology and Pharmacology 35:448-
467.
Frechet, M. (1935). Generalisations du theoreme des probabilites totales. FundamentaMathematica 25:
379-387.
Gaylor, D; Ryan, L; Krewski, D; Zhu, Y (1998). Procedures for calculating benchmark doses for health
risk assessment. Regulatory Toxicology and Pharmacology 28:150-164.
Gogou, E; Kerenidi, T; Chamos, V; Zintzaras, E; Gourgoulianis, KI. (2009). Mesothelioma mortality in
Greece from 1983 to 2003. IntJClin Pract.63:944-948. Epub 2007 Jun 15.
Haque, AK; Vrazel, DM. (1998). Transplacental transfer of asbestos in pregnant mice. Bull Environ
Contam Toxicol 60:620-625.
Haque, AK; Vrazel, DM; Burau, KD; Cooper, SP; Downs, T. (1996). Is there transplacental transfer of
asbestos? A study of 40 stillborn infants. Pediatr PatholLabMed 16:877-892.
Hattis, D; Banati, P; Goble, R. (1999). Distributions of individual susceptibility among humans for toxic
effects. How much protection does the traditional tenfold factor provide for what fraction of
which kinds of chemicals and effects? Ann NY AcadSci 895: 286-316.
HEI-AR, (1991). Asbestos in Public and Commercial Buildings: A Literature Review and Synthesis of
Current Knowledge. Health Effects Institute-Asbestos Research, Cambridge, MA.
http://pubs.healtheffects.org/view.php?id=13
Hjortsberg, U; Orbaek, P; Arborellius, M; Ranstam, J; Welinder, H (1988). Railroad workers with
pleural plaques. I. Spirometric and nitrogen washout investigation on smoking and nonsmoking
asbestos-exposed workers. American J. of Industrial Medicine 14: 649-656.
IARC (International Agency for Research on Cancer). (2012). Asbestos (Chrysotile, Amosite,
Crocidolite, Tremolite, Actinolite, and Anthophyllite, in A Review of Human Carcinogens:
Arsenic, Metals, Fibres, and Dusts, IARC Monographs - 100C, pg 219-309.
Isaacs, KK; Martonen, TB. (2005). Particle deposition in children's lungs: Theory and experiment. J
Aerosol Med 18:337'-3 53.
Jarvolm, B; Sanden, A. (1986). Pleural plaques and respiratory function. American J. Industrial
Medicine 10:419-426.
Kamp, DW; Weitzman, SA. (1999). The molecular basis of asbestos induced lung injury. Thorax
54(7):638-52.
Klein, JP; Moeschberger, ML. (2003). Survival Analysis: Techniques for Censored and Truncated Data.
New York: Springer.
42
-------
Kopylev, L; Sullivan, PA; Vinikoor, LC; Bateson, TF. (2011). Monte Carlo analysis of impact of
underscertainment of mesothelioma cases on underestimation of risk. The Open Epidemiology
Journal 4:45-53.
Kiichenhoff, H; Bender, R; Langner, I. (2007) Effect of Berkson measurement error on parameter
estimates in Cox regression models. Lifetime Data Analysis 13(2):261-72.
Larson, TC; Antao, VC; Bove, FJ; Cusack, C. (2012). Association Between Cumulative Fiber Exposure
and Respiratory Outcomes Among Libby Vermiculite Workers. J Occup EnvironMed54:56-63.
Leake, BE; Woolley, AR; Arps, CES; Birch, WD; Gilbert, MC; Grice, JD; Hawthorne, FC; Kato, A;
Kisch, HJ; Krivovichev, VG; Linthout, K; Laird, J; Mandarine, JMaresch, WV; Nickel, EH;
Rock, NMS; Schumacher, JC; Smith, DC; Shephenson, NCN; Ungaretti, L; Whittake, EJW;
Youzhi, G. (1997). Nomenclature of amphiboles: Report of the Subcommittee on Amphiboles of
the International Mineralogical Association Commission on New Minerals and Mineral Names.
Mineral Mag 61:295-321.
Lilis, R; Miller, A; Godbold, J; Chan, E; Selikoff, U. (199la). Radiographic abnormalities in asbestos
insulators: effects of duration from onset of exposure and smoking: relationships of dyspnea with
parenchymal and pleural fibrosis. AmJIndMed20:1-15.
Lilis, R; Miller, A; Godbold, J; Chan, E; Selikoff, U. (1991b). Pulmonary function and pleural fibrosis:
quantitative relationships with an integrative index of pleural abnormalities. Am JIndMed29:
145-161.
Lippmann, M. (2009). Asbestos and other mineral fibers. In: M. Lippmann, Ed., Environmental
Toxicants: Human Exposures and Their Health Effects, 3rd Ed.., John Wiley, New York, NY,
pp. 395-458.
Lockey, JE; Brooks, SM; Jarabek, AM; Khoury, PR; McKay, RT; Carson, A; Morrison, JA; Wiot, JF;
Spitz, HB. (1984). Pulmonary changes after exposure to vermiculite contaminated with fibrous
tremolite. Am Rev Respir Dis 129:952-958.
Manski, CF. (2003). Partial Identification of Probability Distributions. Springer, New York.
McDonald, JC; McDonald, AD; Armstrong, B; Sebastien, P (1986). Cohort study of mortality of
vermiculite miners exposed to tremolite. Occup Environ Medical 43:436-444.
McDonald, JC; Harris, J; Armstrong, B. (2004). Mortality in a cohort of vermiculite miners exposed to
fibrous amphibole in Libby, Montana. Occup Environ Med 61:363-366.
Meeker, GP; Bern, AM; Brownfield, IK; Lowers, HA; Sutley, SJ; Hoefen, TM; Vance, JS. (2003) The
composition and morphology of amphiboles from the Rainy Creek Complex, near Libby,
Montana. American Mineralogist 88:1955-1969.
Metintas, M; Metintas, S; Ak G, Erginel S; Alatas F; Kurt E; Ucgun I; Yildirim H.(2008). Epidemiology
of pleural mesothelioma in a population with non-occupational asbestos exposure. Respirology
17-21.
43
-------
Metintas, M; Hillerdal, G; Metintas, S; Dumortier, P. (2010) Endemic malignant mesothelioma:
exposure to erionite is more important than genetic factors. Arch Environ Occup Health.
65(2):86-93.
Metintas, S; Metintas, M; Ak ,G; Kalyoncu C. (2012). Environmental asbestos exposure in rural Turkey
and risk of lung cancer. Int J Environ Health Res. Feb 2. [Epub ahead of print]
Miller, A. (2002) Pleural plaques and lung function. Am JRespir Crit Care Med. 165(2):305-6.
Miller, A; Lilis, R; Godbold, J; Chan, E; Selikoff, U. (1992). Relationship of pulmonary function to
radiographic interstitial fibrosis in 2,611 long term asbestos insulators: an assessment of the
International Labour Organization profusion score. Am Rev Respir Dis 145:263-270.
Miller, MD; Marty, MA; Arcus, A; Brown, J; Morry, D; Sandy, M. (2002). Differences between
children and adults: Implications for risk assessment at California EPA. Int J Toxicol 21(5):403-
418.
Mossman, BT; Lounsbury, KM; Reddy, SP. (2006). Oxidants and signaling by mitogen-activated
protein kinases in lung epithelium. Am J Respir CellMol Biol. 34(6):666-9.
Mossman, BT; Lippmann, M; Hesterberg, TW; Kelsey, KT; Barchowsky, A; Bonner, JC. (2011).
Pulmonary endpoints (lung carcinomas and asbestosis) following inhalation exposure to
asbestos. J Toxicol Environ Health, Critical Reviews, Part B 14:76-121.
Mulhausen, J; Damiano, J. (1998)^4 Strategy for Assessing and Managing Occupational Exposures. 2nd
edition., AIHA Press.
Neri, M; Ugolini, D; Dianzani, I; Gemignani, F; Landi, S; Cesario, A; Magnani, C; Mutti, L; Puntoni, R;
Bonassi, S. (2008) Genetic susceptibility to malignant pleural mesothelioma and other asbestos-
associated diseases. Mutation Research 659:126-136.
NIOSH (1977). Leidel, Bush & Lynch Occupational Exposure Sampling Strategy Manual. NIOSH 77-
173 CDC.
Nguyen, HT; Kreinovich, V; Wu, B.; Xiang, G. (2012) Computing Statistics under Interval and Fuzzy
Uncertainty. Springer, Berlin.
NRC (1994). Science and Judgement in Risk Assessment. Washington, DC: National Academy Press
[Chapter 11, Appendix 1-1, Appendix 1-2] http://www.nap.edu/catalog.php?recird_id=2125
OEHHA (2008). Air Toxics Hot Spots Risk Assessment Guidelines Technical Support Document for
the Derivation of Noncancer Reference Exposure Levels. Available online at:
http ://www. oehha. ca. gov
Ohlson, G; Rydman, T; Sundell, L; Bodin L; Hogstedt, C. (1984). Decreased lung function in long-term
asbestos cement workers: across-sectional study. American JIndustrial Medicine 14:649-656.
44
-------
Ohlson, CG; Bodin, L; Rydman, T; et al. (1985) Ventilatory derangements in former asbestos cement
workers: a four year follow up. Brit JIndMedicine 42:612-616.
Oliver, LC; Eisen, EA; Greene, R; Sprince, NL (1988) Asbestos-related pleural plaques and lung
function. American J. of Industrial Medicine 14:649-656.
Paris, C; Martin, A; Letourneux, M; Wild, P. (2008). Modelling prevalence and incidence of fibrosis and
pleural plaques in asbestos-exposed populations for screening and follow-up: a cross-sectional
study. Environ Health Global Sci Source 7: 30. http://dx.doi.org/10.1186/1476-069X-7-30.
Paris, C; Thierry S; Brochard P, et al. (2009) Pleural plaques and asbestosis: dose- and time-response
relationships based on HRCT data. Eur Respirj. 34:72-79.
Peto, J. (1979). Dose-response relationships for asbestos-related disease: Implications for hygiene
standards. Part II: Mortality. AnnNYAcadSci 330:195-204.
Peto, J; Seidman, H; Selikoff, U. (1982). Mesothelioma mortality in asbestos workers: implications for
models of carcinogenesis and risk assessment. Br J Cancer 45:124-135.
Reid, A; Berry, G; Heyworth, J; de Klerk, NH; Musk, AW. (2009). Predicted mortality from malignant
mesothelioma among women exposed to blue asbestos at Wittenoom, Western Australia. Occup
Environ Med 66:169-174.
Reid, A; Heyworth J; de Klerk NH; Musk B. (2008). Cancer incidence among women and girls
environmentally and occupationally exposed to blue asbestos at Wittenoom, Western Australia.
IntJCancer 122(10):2337-44.
Reid, A; Berry, G; de Klerk, N; Hansen, J; Heyworth, J; Ambrosini, G; Fritschi, L; Olsen, N; Merler, E;
Musk, A. (2007) Age and sex differences in malignant mesothelioma after residential exposure
to blue asbestos (crocidolite). Chest 131:376-382.
Richardson, DB. (2010). Occupational exposures and lung cancer: Adjustment for unmeasured
confounding by smoking. Epidemiology 21:181-186.
Richardson, DB. (2008). Multistage Modeling of Leukemia in Benzene Workers: A simple Approach to
fitting the 2-stage Clonal Expansion Model. Am J. Epi. DOI: 10.1093/aje/kwn284
Robledo, R; Mossman, B. (1999). Cellular and molecular mechanisms of asbestos-induced fibrosis. J
CellPhysiol. 180(2): 158-66.
Rohs, A; Lockey, J; Dunning, K; Shukla,R; Fan, H; Hilbert,T; Borton,E; Wiot, J; Meyer, C; Shipley, R;
Lemasters, G; Kapil, V. (2008). Low-level fiber-induced radiographic changes caused by Libby
vermiculite: A 25-year follow-up study. Am. JRespir Crit Care Med 177:630-637.
Sanchez, MS; Gunter, ME; Dyar, MD. (2008) Characterization of historical amphibole samples from the
former vermiculite mine near Libby, Montana, U.S.A. European Journal of Mineralogy 20:
1043-1053.
45
-------
Schwartz, DA, Fuortes, LJ; Galvin, JR; Burmeister, LF; Schmidt, LE; Leistikow, BN; LaMarte, FP;
Merchant, JA. (1990). Asbestos-induced plueral fibrosis and impaired lung function. Am Rev
RespirDis. 141(2):321-6.
Scheuplein, R; Charnley, G; Dourson, ML. (2002). Differential sensitivity of children and adults to
chemical toxicity: I. Biological Basis. Regulatory Toxicology andPharmacolog 35:429-447.
Sichletidis, L; Chloros, D; Chatzidimitriou, N; Tsiotsios, I; Spyratos, D; Patakas, D. (2006). Diachronic
study of pleural plaques in rural population with environmental exposure to asbestos. Am JInd
Med. 49(8):634-41.
Straif, K; Benbrahim-Tallaa, L; Baan, R; Grosse, Y; Secretan, B; El Ghissassi, F; Bouvard, V; Guha, N;
Freeman, C; Galichet, L; Cogliano, V. (2009). A review of human carcinogens: Part C: Metals,
arsenic, dusts, and fibres. Lancet Oncol 10:453-454.
Sussman ,RG; Cohen, BS; Lippmann, M. (1991a). Asbestos fiber deposition in a human
tracheobronchial cast. I. Exp Inhal Toxicol 3:145-160.
Sussman, RG; Cohen, BS; Lippmann, M. (1991b). Asbestos fiber deposition in a human
tracheobronchial cast. II. Empirical model. Inhal Toxicol 3:161-179.
Testa, JR; Cheung, M; Pei, J; Below, JE; Tan, Y; Sementino, E; Cox, NJ; Dogan, AU; Pass, HI, Trusa
S.; Hesdroffer, M,; Nasu, M; Powers, A; Rivera, Z; Comertpay, S; Tanji, M; Gaudino, G.; Yang,
H; Carbone, M (2011) Germline BAP1 mutations predispose to malignant mesothelioma. Nature
Genetics 43:1022-1025.
USEPA (1988). IRIS summary for Asbestos (CASRN 1332-21-4), Washington, DC.
http://www.epa.gov/iris/subst/0371.htm.
USEPA (1994). Methods for derivation of inhalation reference concentrations and application of
inhalation dosimetry. EPA/600/8-90/066F.
USEPA (2002).^4 review of the reference dose and reference concentration processes. Risk Assessment
Forum. EPA/630/P-02/002F.
USEPA (2005). Guidelines for Carcinogen Risk Assessment. Risk Assessment Forum. EPA/630/P-
03/001B.
USEPA (2012). Benchmark Dose Technical Guidance. Risk Assessment Forum, EPA/100/R-12/001.
Van Cleemput, J; De Raeve, H; Verschakelen, JA; Rombouts J; Lacquet, LM; Nemery, B. (2001),
Surface of localized pleural plaques quantitated by computed tomography scanning: no relation
with cumulative asbestos exposure and no effect on lung function. Am J Respiratory Crit Care
Medicine 163:705-710.
Veblen, DR; Wylie, AG. (1993). Mineralogy of amphiboles and 1:1 layer silicates. In, G.D. Guthrie Jr.
and B.T. Mossman, Eds., Health effects of mineral dusts, Vol 28, pp 61-137. Reviews in
Mineralogy, Mineralogical Society of America, Washington, DC.
46
-------
Warheit, DB; Hartsky, MA. (1990). Species comparisons of alveolar deposition patterns of inhaled
particles. Exp. Lung Res. 16:83-99.
Webber, JS; Blake, DJ; Ward, TJ; Pfau, JC. (2008) Separation and characterization of respirable
amphibole fibers from Libby, Montana. Inhal Toxicol 20:733-740.
Weill, D; Dhillon, G; Freyder, L; Lefante, J; Glindmeyer, H. (2011). Lung function, radiological
changes and exposure: analysis of ATSDR data from Libby, Montana. USA. Eur Respir J38:
376-383.
Weiner, SJ; Neragi-Miandoab, S. (2009) Pathogenesis of malignant pleural mesothelioma and the role of
environmental and genetic factors. J Cancer Res Clin Oncol 135:15-27.
Whitehouse, AC. (2004) Asbestos-related pleural disease due to tremolite associated with progressive
loss of lung function: Serial observations in 123 miners, family members, and residents of Libby,
Montana. Am JIndustr Med46:219-225.
Wilken, D; Garrido, MV; Manuwald, U; Bauer, X. (2011). Lung function in asbestos-exposed workers,
a systematic review and meta-analysis. J OccupMed Tox. 6:21-37.
Zeka, A; Gore, R; Kriebel, D. (2011). The two-stage clonal expansion model in occupational cancer
epidemiology: results from three cohort studies. Occupational and Environmental Medicine
68:618-24.
47
-------
APPENDIX A: EPA'S CHARGE QUESTIONS
EPA Charge to the SAB for the IRIS Toxicological Review
of Libby Amphibole Asbestos
August 2011
Introduction
The U.S. Environmental Protection Agency (EPA) is seeking an external peer review of the scientific
basis supporting the draft Toxicological Review of Libby Amphibole asbestos that will appear on the
Agency's online database, the Integrated Risk Information System (IRIS). IRIS is prepared and
maintained by the EPA's National Center for Environmental Assessment (NCEA) within the Office of
Research and Development (ORD). An existing IRIS assessment for asbestos which includes a
carcinogenicity assessment was posted on IRIS in 1988. The draft on which EPA is now seeking review
is the first IRIS assessment specific to Libby Amphibole asbestos1.
IRIS is a human health assessment program that evaluates qualitative and quantitative risk information
on effects that may result from exposure to specific chemical substances found in the environment.
Through the IRIS Program, EPA provides quality science-based human health assessments to support
the Agency's regulatory activities. Combined with specific exposure information, government and
private entities use IRIS to help characterize public health risks of chemical substances in site-specific
situations in support of risk management decisions.
Libby Amphibole asbestos, found in vermiculite ore deposits near Libby, Montana, is comprised of a
mixture of related mineral forms of amphibole asbestos: primarily winchite, richterite and tremolite with
trace amounts of magnesioriebeckite, edenite, and magnesio-arfvedsonite. Health effects from exposure
to Libby Amphibole asbestos are a potential concern for Libby residents, as well as workers and others
who may have handled vermiculite mined in Libby, Montana. Additionally, vermiculite from Libby,
Montana was incorporated into various consumer products, some of which may remain in place (e.g.,
vermiculite attic insulation in homes).
The external review draft Toxicological Review of Libby Amphibole asbestos is based on a
comprehensive review of the available scientific literature on the health effects of Libby Amphibole
asbestos and was developed in adherence with general guidelines for risk assessment set forth by the
National Research Council in 1983 (NRC, 1983)2 and numerous guidelines and technical reports
published by EPA (see Section 1 of the assessment)3. Specifically, this draft IRIS assessment provides
an overview of sources of exposure to Libby Amphibole asbestos, characterizes the hazard posed by
exposure to Libby Amphibole asbestos for carcinogenicity and noncancer health effects based on the
available scientific evidence, and presents a qualitative and quantitative health assessment, including the
1 The term "Libby Amphibole asbestos" is used in this document to identify the mixture of amphibole mineral fibers of
varying elemental composition (e.g., winchite, richterite, tremolite, etc.) that have been identified in the Rainy Creek complex
near Libby, Montana.
2 NRC (1983). Risk Assessment in the federal government: managing the process. Washington DC: National Academy Press.
3 http://www.epa.gov/iris^ackgrd.html
A-l
-------
derivations of a chronic inhalation reference concentration (RfC) and an inhalation unit risk (IUR) that
can be combined with exposure information in a risk assessment to estimate noncancer hazard and
carcinogenic risk, respectively, in humans. The assessment does not address oral exposure to Libby
Amphibole asbestos.
Charge Questions
Below is a set of charge questions that address scientific issues in the draft human health assessment of
Libby Amphibole asbestos. Please provide detailed explanations for responses to the charge questions.
EPA will also consider the Science Advisory Board reviewer SAB comments on other major scientific
issues specific to the hazard identification and dose response assessment of Libby Amphibole asbestos.
Please identify and provide the rationale for approaches to resolve the issues where possible. Please
consider the accuracy, objectivity, and transparency of EPA's analyses and conclusions in your review.
General Charge Questions:
1. Is the Toxicological Review logical, clear, and concise? Has EPA clearly, and in sufficient detail,
presented and synthesized the scientific evidence for health hazards from Libby Amphibole asbestos?
2. Please identify any additional peer-reviewed studies from the primary literature that should be
considered in the assessment of noncancer and cancer health effects of Libby Amphibole asbestos.
Chemical-Specific Charge Questions:
I. Background
A. Mineralogy and Toxicokinetics
1. In order to inform the hazard identification and dose response of Libby Amphibole asbestos,
background material is included in the document briefly describing the mineralogy and toxicokinetics of
asbestos and related mineral fibers (Section 2 and 3):
a. Please comment on whether the presentation of the available data on the mineralogy of Libby
Amphibole asbestos is clear, concise and accurate.
b. In the absence of toxicokinetic information specific to Libby Amphibole asbestos, the draft
assessment contains a general summary description of fiber toxicokinetics. Please comment on whether
this overview of general fiber toxicokinetics is clear, concise and accurate.
II. Hazard Identification of Libby Amphibole Asbestos
A. Noncancer Health Effects:
1. An occupational cohort of workers in a Marysville, OH facility exposed to Libby Amphibole asbestos
(Lockey et al., 1984; Rohs et al., 2008) was selected as the basis for the derivation of the reference
concentration (RfC). Please comment on whether the selection of this study population is scientifically
supported and clearly described. If a different study population is recommended as the basis for the RfC,
please identify this study and provide scientific support for this choice.
A-2
-------
2. Radiographic evidence of localized pleural thickening in humans was concluded by EPA to be an
adverse effect and was selected as the critical effect for the derivation of the RfC. Pleural thickening is
associated with restrictive lung function, breathlessness during exercise and, for some individuals,
chronic chest pain. Please comment on whether the selection of this critical effect and its
characterization is scientifically supported and clearly described. If a different health endpoint is
recommended as the critical effect for deriving the RfC, please identify this effect and provide scientific
support for this choice.
3. The database of laboratory animal and mechanistic studies of Libby Amphibole asbestos is
summarized in the draft assessment (see Section 4.2 and 4.3, details in Appendix D) to inform the
mechanisms of the biological response to Libby Amphibole asbestos and support the epidemiology
studies used for derivation of the RfC. Please comment on whether the laboratory animal and
mechanistic information presented is used appropriately in the draft assessment.
B. Carcinogenicity:
1. Under EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005;
www.epa.gov/iris/backgrd.html), the draft IRIS assessment characterizes Libby Amphibole asbestos as
"carcinogenic to humans" by the inhalation route of exposure. Please comment on whether the cancer
weight of evidence characterization is scientifically supported and clearly described.
2. Due to the limitations of the data available, the draft assessment concludes that there is insufficient
information to identify the mode of carcinogenic action of Libby Amphibole asbestos. Please comment
on whether this determination is appropriate and clearly described. Note that in the absence of
information to establish a mode of action, a linear low dose extrapolation is recommended by the
Guidelines for Carcinogen Risk Assessment (U.S., EPA, 2005; Section 3.3). If it is judged that a mode
of action can be established for Libby Amphibole asbestos, please identify the mode of action and its
scientific support (i.e., studies that support the key events, and specific data available to inform the
shape of the exposure-response curve at low doses).
3. An occupational cohort of workers from Libby, Montana exposed to Libby Amphibole asbestos (i.e.,
the Libby worker cohort) was selected as the basis for the derivation of the inhalation unit risk (IUR).
Please comment on whether the selection of this study population is scientifically supported and clearly
described. If a different study population is recommended as the basis for the IUR, please identify this
study and provide scientific support for this choice.
4. Mortality from lung tumors and mesothelioma in the Libby worker cohort was selected to serve as the
basis for the derivation of the IUR. Please comment on whether this selection is scientifically supported
and clearly described. If a different health endpoint is recommended for deriving the IUR, please
identify this endpoint and provide scientific support for this choice.
5. The database of laboratory animal and mechanistic studies of Libby Amphibole asbestos is
summarized in this draft assessment (see Section 4.2 and 4.3, details in Appendix D) to inform the
mechanisms of the biological response to Libby Amphibole asbestos and support the epidemiology
studies used for derivation of the IUR. Please comment on the use of laboratory animal and mechanistic
information in the draft assessment.
III. Exposure-Response Assessment
A. Inhalation Reference Concentration (RfC):
A-3
-------
1. Exposures to Libby Amphibole asbestos for workers in the Marysville, OH facility were reconstructed
based on industrial hygiene data collected in the facility from 1972 to 1994. Exposures from 1957 to
1971 were estimated based on extrapolation from the available industrial hygiene data. The information
used for the exposure reconstruction was based on employee interviews, court and company records, and
the expert judgment of the researchers. Is the methodology used for the exposure reconstruction reported
in Appendix F and the subsequent development of exposure estimates used in the analyses scientifically
supported and clearly described?
2. Exposure-response modeling was conducted using the incidence of localized pleural thickening in
workers and cumulative exposure to estimate the point of departure (POD) for derivation of the RfC.
EPA's estimate of the POD is based upon a Michaelis-Menten model applied to the subcohort of
workers examined in 2002-2005 and first exposed to Libby Amphibole asbestos in 1972 (when
measurements of fiber levels in the workplace began) or later with cumulative exposure as the
explanatory variable. Is the selection of the model scientifically justified and clearly described? Has the
modeling and the choice of a benchmark response (BMR) for the POD of 10% extra risk of localized
pleural thickening been clearly described and appropriately conducted according to EPA's Draft
Benchmark Dose Technical Guidance (U.S. EPA, 2000b)?
3. EPA's assessment also provides the results of alternative modeling approaches to derive a POD for
localized pleural thickening. This modeling used the full Marysville worker data set with exposures
from 1957 and later and a Cumulative Normal Michaelis-Menten model that incorporates both
cumulative exposure and time from first exposure as explanatory variables. Please comment on whether
EPA's rationale for presenting these alternative approaches is scientifically justified and clearly
described. Please identify and provide the rationale if a different approach for identifying the most
appropriate population within the cohort of Marysville workers is recommended as the basis for
estimating a POD.
4. EPA has evaluated potential confounders and covariates where data are available. Specifically, EPA
has explored the influence of age, body mass index, smoking status, time since first exposure, gender,
and alternative exposure metrics on model fit and evaluated their association with the modeled health
outcomes (see Section 5.3). Are these analyses clearly described and appropriately conducted? Are the
results of these analyses appropriately considered in the RfC derivation? Additionally, there is a
possibility of exposure-dependent censoring in participant selection for the update of the Marysville
cohort (Rohs et al., 2008) but no evidence of selection bias. Does the SAB have any specific
recommendations for evaluating and, if appropriate, quantitatively addressing exposure-dependent
censoring in these analyses?
5. The modeled POD estimate is based on cumulative exposure estimates for the worker cohort
examined. For the derivation of the RfC, this cumulative exposure is prorated over the period of
environmental exposure (lifetime or shorter duration chronic exposure when appropriate). The RfC is
provided in units of continuous air concentration. Is the basis of this conversion clearly explained and
scientifically justified?
6. Please comment on the rationale for the selection of the uncertainty factors (UFs) applied to the POD
for the derivation of the RfC. Are the UFs appropriate based on A Review of the Reference Dose and
Reference Concentration Processes (U.S. EPA, 2002; Section 4.4.5) and clearly described? If changes
to the selected UFs are proposed, please identify and provide scientific support. Specifically, please
A-4
-------
comment on the rationale for the selection of the database uncertainty factor (UFD) of 10 applied in the
derivation of the RfC. The database uncertainty factor accounts for the lack of data on effects other than
in the respiratory system, including other effects observed in community and laboratory animal studies
(cardiovascular disease and autoimmune effects) that have not been well-studied (See Section 5.2.3 of
the Toxicological Review); and lack of health data assessed at later time points. Is the rationale for the
UFD appropriate and clearly described? Please provide the rationale if a change in the UFD is
proposed.
7. Please comment on whether the document adequately describes the uncertainties and limitations in
the methodology used to derive the RfC and whether this information is presented in a transparent
manner.
B. Inhalation Unit Risk (IUR):
1. Exposure-response modeling was conducted separately for lung cancer and mesothelioma mortality.
The POD estimates for these endpoints are based upon analysis of the subcohort of workers first
exposed after 1959 when the exposure data were judged to be better characterized. The exposure-
response modeling included consideration of a variety of exposure metrics that varied with time and
incorporated different lag and decay parameters. Based on the results of the exposure-response
modeling, a life table analysis was used to determine the PODs for each type of cancer for the various
exposure metrics. Have the exposure-response modeling and determination of the PODs from life table
analysis been appropriately conducted and clearly described? If a different approach to exposure-
response analysis is recommended as the basis for the estimating the IUR, please identify the
recommended methods and provide a rationale for this choice.
2. Smoking is a strong independent risk factor for lung cancer and may be an important confounder of
the lung cancer mortality analysis. Data on individual smoking habits and history were largely missing
and could not be used to control for potential confounding in regression analyses. However, EPA used
three approaches to evaluate the confounding issue, including restriction of the cohort and two analytic
evaluations of the potential for confounding by smoking (see Section 5.4.3.6.5). Please comment on
whether the methods and analyses are clearly presented and scientifically justified. If additional analyses
are recommended, please identify the methods and scientific rationale.
3. In order to derive an IUR which represents the combined risk of mortality from lung cancer or
mesothelioma, a cancer-specific unit risk for each tumor type was calculated according to the Guidelines
for Carcinogen Risk Assessment (U.S., EPA, 2005; Sections 3.2 and 3.3) by linear extrapolation from
the corresponding POD (i.e., the lower 95% confidence limit on the exposure associated with 1% extra
risk of lung cancer or 1% absolute risk of mesothelioma mortality). The IUR was then determined as a
combined upper bound risk estimate for mortality considering both cancers. Has this approach been
appropriately conducted and clearly described?
4. Please comment on the adjustment for mesothelioma mortality under-ascertainment. Is this
adjustment scientifically supported and clearly described? If another adjustment approach is
recommended as the basis for the IUR, please identify that approach and provide the scientific rationale.
5. Please comment on whether the document adequately describes the uncertainties and limitations in
the methodology used to derive the IUR and whether this information is presented in a transparent
manner.
A-5
------- |