UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON D.C. 20460
OFFICE OF THE ADMINISTRATOR
SCIENCE ADVISORY BOARD
March 17,2011
EPA-SAB-11-004
The Honorable Lisa P. Jackson
Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, N.W.
Washington, D.C. 20460
Subject: SAB Review of EPA's "Development of a Relative Potency Factor (RPF)
Approach for Polycyclic Aromatic Hydrocarbon (PAH) Mixtures (February
2010 Draft)"
Dear Administrator Jackson:
EPA's current approach to assessing cancer risk for polycyclic aromatic hydrocarbon
(PAH) mixtures uses the relative potency factor (RPF) approach, which estimates the cancer risk
of individual PAHs relative to benzo[a]pyrene (BaP). In 1993, EPA published RPF values for 6
PAHs. EPA's Office of Research and Development (ORD) has updated the RPF values for these
6 PAHs and developed new RPF values for 18 additional PAHs, using recent studies from the
published literature, as described in Development of a Relative Potency Factor (RPF) Approach
for Polycyclic Aromatic Hydrocarbon (PAH) Mixtures (February 2010 Draft).
ORD requested the Science Advisory Board (SAB) to peer review the PAH Mixtures
document, focusing on: rationale for recommending an RPF approach, discussion of previously
published RPF approaches, evaluation of the carcinogenicity of individual PAHs, methods for
dose-response assessment and RPF calculation, selection of PAHs for inclusion in the RPF
approach, derivation of RPFs for selected PAHs, and uncertainties and limitations associated
with the RPF approach. The SAB convened the PAH Mixtures Review Panel to provide advice
to the Agency. The key points and recommendations of the Panel are detailed in the report.
Below is a brief highlight of the major comments and recommendations.
Overall, the SAB finds the document to be logical, clear, and concise. The SAB
recognizes the pragmatic need for the RPF approach. Based upon the currently available data,
the SAB supports EPA's use of the RPF approach for assessing carcinogenic risk from PAH
mixtures. The SAB, however, provides recommendations to strengthen the scientific rationale
for the RPF approach, the selection of studies, methods for dose-response modeling, and
calculations of final RPFs.
-------
Although the SAB supports the use of benzo[a]pyrene (BaP) as the index compound for
the RPF approach, the cancer slope factor for BaP is outdated and it is essential that EPA
expeditiously update the cancer slope factor for BaP.
The SAB also recommends that EPA consider developing a whole mixtures approach for
PAHs. This approach could validate the RPF approach and in the future, could replace the RPF
approach. The Agency should set this as a strategic initiative, with a specific timeline and
benchmarks, that lays the foundation for an underlying concerted research program. The SAB
recommends that the Agency seek support from the National Toxicology Program (NTP) and/or
other entities to conduct testing of an appropriate portfolio of different complex PAH mixtures.
These complex PAH mixtures should represent a diverse array of mixtures, but also represent the
most important PAH mixture classes of concern to EPA.
We appreciate the opportunity to provide EPA with advice. We look forward to
receiving the Agency's response.
Sincerely,
/Signed/
Dr. Deborah L. Swackhamer, Chair
EPA Science Advisory Board
/Signed/
Dr. Nancy K. Kim, Chair
SAB PAH Mixtures Review Panel
-------
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 Web site at:
http://www.epa.gov/sab.
in
-------
U.S. Environmental Protection Agency
Science Advisory Board
Polycylic Aromatic Hydrocarbon (PAH) Mixtures Review Panel
CHAIR
Dr. Nancy K. Kim, Senior Executive, Health Research, Inc., Troy, NY
MEMBERS
Dr. Shantu Amin, Professor, Department of Pharmacology, Penn State Hershey Cancer
Institute , Penn State College of Medicine, Hershey, PA
Dr. Frederick A. Beland, Director, Division of Biochemical Toxicology, National Center for
Toxicological Research, U.S. Food and Drug Administration, Jefferson, AR
Dr. James Chen, Senior Biomedical Research Service/Senior Mathematical Statistician,
National Center for Toxicological Research, U.S. Food and Drug Administration, Jefferson, AR
Dr. John DiGiovanni, Professor and Coulter R. Sublett Chair in Pharmacy, Division of
Pharmacology and Toxicology and Department of Nutritional Sciences, Dell Pediatric Research
Institute, The University of Texas at Austin, Austin, TX
Dr. Marilie Gammon, Professor, Epidemiology, Gillings School of Global Public Health,
University of North Carolina at Chapel Hill, Chapel Hill, NC
Dr. David Gaylor, President, Gaylor and Associates, LLC, Eureka Springs, AR
Dr. Nicholas Geacintov, Professor, Chemistry, New York University, New York, NY
Dr. Chris Gennings, Professor, Department of Biostatistics, Virginia Commonwealth
University, Richmond, VA
Dr. Joshua Hamilton, Chief Academic and Scientific Officer; Senior Scientist, Bay Paul Center
for Comparative Molecular Biology and Evolution, Marine Biological Laboratory (MBL),
Woods Hole, MA
Dr. Edmond LaVoie, Professor and Chair, Department of Pharmaceutical Chemistry, College of
Pharmacy, Rutgers, The State University of New Jersey, Piscataway, NJ
Dr. Aramandla Ramesh, Assistant Professor, Biochemistry and Cancer Biology, School of
Medicine, Meharry Medical College, Nashville, TN
Dr. Benjamin Rybicki, Senior Scientist, Department of Research Epidemiology and
Biostatistics, Henry Ford Hospital, Detroit, MI
IV
-------
Dr. Paul Strickland, Professor, Environmental Health Sciences, Bloomberg School of Public
Health, Johns Hopkins University, Baltimore, MD
Dr. Emanuela Taioli, Professor, Department of Epidemiology and Biostatistics, School of
Public Health, State University of New York (SUNY) Downstate Medical Center, Brooklyn, NY
SCIENCE ADVISORY BOARD STAFF
Mr. Aaron Yeow, Designated Federal Officer, U.S. Environmental Protection Agency,
Washington, DC
v
-------
U.S. Environmental Protection Agency
Science Advisory Board
BOARD
CHAIR
Dr. Deborah L. Swackhamer, Professor and Charles M. Denny, Jr., Chair in Science,
Technology and Public Policy and Co-Director of the Water Resources Center, Hubert H.
Humphrey School of Public Affairs, University of Minnesota, St. Paul, MN
SAB MEMBERS
Dr. David T. Allen, Professor, Department of Chemical Engineering, University of Texas,
Austin, TX
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 Buckley, Associate 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, Department of Health Policy and Management, 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
Dr. Otto C. Doering III, Professor, Department of Agricultural Economics, Purdue University,
W. Lafayette, IN
VI
-------
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, Vice President for Research, Office of the Vice President for Research,
Texas Tech University, Lubbock, TX
Dr. Elaine Faustman, Professor, Department of Environmental and Occupational Health
Sciences, School of Public Health and Community Medicine, 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 Griffiths, Associate 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. 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. Catherine Kling, Professor, Department of Economics, Iowa State University, Ames, IA
Dr. Kai Lee, Program Officer, Conservation and Science Program, David & Lucile Packard
Foundation, Los Altos, CA (affiliation listed for identification purposes only)
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. Lee D. McMullen, Water Resources Practice Leader, Snyder & Associates, Inc., Ankeny,
IA
Dr. Judith L. Meyer, Professor Emeritus, Odum School of Ecology, University of Georgia,
Lopez Island, WA
vn
-------
Dr. James R. Mihelcic, Professor, Civil and Environmental Engineering, State of Florida 21st
Century World Class Scholar, University of South Florida, Tampa, FL
Dr. Jana Milford, Professor, Department of Mechanical Engineering, University of Colorado,
Boulder, CO
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, Grants Facilitator, Ernest Mario School of Pharmacy, Rutgers University,
Piscataway, NJ
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. Arden Pope, 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, 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. Kathleen Segerson, Philip E. Austin Professor of Economics , Department of Economics,
University of Connecticut, Storrs, CT
Dr. Herman Taylor, Director, Principal Investigator, Jackson Heart Study, University of
Mississippi Medical Center, Jackson, MS
Vlll
-------
Dr. Barton H. (Buzz) Thompson, Jr., Robert E. Paradise Professor of Natural Resources Law
at the Stanford Law School and Perry L. McCarty Director, Woods Institute for the Environment,
Stanford University, Stanford, CA
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. Thomas S. Wallsten, Professor and Chair, Department of Psychology, University of
Maryland, College Park, MD
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
Dr. Thomas Armitage, Designated Federal Officer, U.S. Environmental Protection Agency,
Washington, DC
IX
-------
ACRONYMS
BaP Benzo[a]pyrene
BMD Benchmark Dose
BMDL Benchmark Dose (Lower Confidence Limit)
BMR Benchmark Response
CSF Cancer Slope Factor
EPA Environmental Protection Agency
IARC International Agency for Research on Cancer
IRIS Integrated Risk Information System
NCEA EPA's National Center for Environmental Assessment
ORD EPA's Office of Research Development
PAH Polycyclic Aromatic Hydrocarbon
RPF Relative Potency Factor
SAB Science Advisory Board
x
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 5
3. RESPONSE TO EPA CHARGE QUESTIONS 6
3.1. Charge Question 1 - Overall Scientific Soundness of the RPF Approach 6
3.2. Charge Question 2 -Rationale for Recommending an RPF Approach 6
3.3. Charge Question 3 - Discussion of Previously Published RPF Approaches 10
3.4. Charge Question 4 - Evaluation of the Carcinogenicity of Individual PAHs 11
3.5. Charge Question 5 - Methods for Dose-Response Assessment and RPF Calculation 12
3.6. Charge Question 6 - Selection of PAHs for Inclusion in the Relative Potency Approach .. 20
3.7. Charge Question 7-Derivation of RPFs for Selected PAHs 23
3.8. Charge Question 8 - Uncertainties and Limitations Associated with the RPF Approach.... 27
3.9. Charge Question 9 - Adequacy of Appendices for Independent Verification 29
4. REFERENCES 30
APPENDIX - CHARGE QUESTIONS A-l
XI
-------
1. EXECUTIVE SUMMARY
In 1993, EPA developed the document Provisional Guidance for Quantitative Risk
Assessment of PAH that recommends a Relative Potency Factor (RPF) approach for assessing
PAH mixtures. EPA's Office of Research and Development (ORD) has developed a draft
technical document, Development of a Relative Potency Factor (RPF) Approach for Polycyclic
Aromatic Hydrocarbon (PAH) Mixtures, hereafter called "PAH Mixtures document", to update
the 1993 document by expanding the number of PAHs assessed and including recent studies
from the published literature.
EPA's Office of Research and Development (ORD) requested that the Science Advisory
Board (SAB) review the PAH Mixtures document. There were nine charge questions, which
focused on the overall scientific soundness of the approach, on the specific chapters of the
document, and the adequacy of the appendices to allow for independent verification. These
charge questions are included in the Appendix and the responses to the charge questions are
detailed in the report. The recommendations from the Panel for the major charge questions are
highlighted below. Detailed responses to all the charge questions are presented in the body of
the report.
General Comments
Overall, the Panel finds the document to be logical, clear, and concise. The Panel
recognizes the pragmatic need for the RPF approach, and based upon the currently available data,
recommends that EPA continue to use the RPF approach for assessing cancer risk for PAH
mixtures. The Panel agrees with EPA's decision to update the 1993 approach by increasing the
number of compounds in the approach, and including more recent data in calculating and
expanding the RPF values for PAHs. The Panel recommends that the Agency finalize the
document based upon the Panel's comments and recommendations.
Rationale for Recommending an RPF Approach
EPA's document presents the scientific rationale for recommending an RPF approach for
PAH mixtures. The Panel has several recommendations for strengthening the rationale for using
the RPF approach. Additional historical perspective should be added, since it is an important
component in, and justification for the agency's practical decision to continue with the RPF
method. EPA indicated at the meeting that they had previously considered implementing a
whole mixtures approach, but decided against it due to significant data gaps in available
information. The Panel recommends including a discussion of these previous considerations and
evaluation of data gaps, which would add to the rationale to continue with the RPF approach.
The Panel recommends strengthening the rationale by discussing that the RPF approach
relies on a direct comparison between dose-response curves from actual cancer bioassay data
between BaP as the index compound and the target PAH. The Panel finds that the choice of BaP
as the index chemical is well justified and is appropriately described for this RPF approach. The
Panel is aware that a revised Integrated Risk Information System (IRIS) assessment for BaP is
under concurrent development, and urges the Agency to quickly finalize that assessment.
-------
The Panel finds that EPA's assumption that interactions among PAH mixture
components do not occur at low levels of environmental exposure is not well justified in the
document; however, in the absence of data that support a specific interaction (additive, sub- or
super-additive, etc.), a default assumption of additivity is reasonable for the purposes of the RPF
analysis.
Concurrent with the continued use of the RPF approach, the Panel recommends that EPA
pursue developing a whole mixtures approach for PAHs to potentially validate the RPF approach
and to serve as a possible replacement for the RPF approach in the future.
Discussion of Previously Published RPF Values
EPA presents a background on how RPFs have been derived in the past and a qualitative
comparison between the previous RPF approaches and studies testing whole mixtures of PAHs.
The Panel believes that the document adequately summarizes the previous RPF approaches, but
could be improved by providing more quantitative information on the comparison between
cancer risk estimates derived from the previous RPF approaches and those estimates derived
from the whole mixtures approach. The Panel also recommends editing Table 3-1 to use a
standardized approach for reporting values (same significant figures, scale, etc.).
Evaluation of the Carcinogenicity of Individual PAHs
EPA discusses the development of a database of primary literature and the criteria used to
include or exclude studies. Based upon the initial literature search, a list of 74 PAHs was
evaluated. The Panel finds that the list of 74 PAHs is reasonable and that the database of
primary literature appears adequate, but recommends that a recently published IARC Monograph
on PAHs, Volume 92, be added to the database as an additional resource (IARC, 2010).
One of EPA's study selection criteria is the stipulation that BaP must be tested
concurrently with the target PAH being considered. This restriction raises the concern that
animal bioassay data from studies of high quality may be dismissed. The Panel recommends that
EPA consider exploring an approach where a target PAH that was tested with BaP could serve as
a surrogate for BaP in studies where BaP was not tested concurrently. This may allow for
additional data from studies of high quality to be included. However, in considering this
alternative approach, EPA should also take into account factors that could potentially outweigh
the benefits in the establishment of a RPF for a specific PAH, such as cross-study and cross-
laboratory comparability issues.
The Panel believes that a quality assessment should be done for each individual study.
The Panel recommends including information such as sample size, dosing, mortality (prior to
tumor development), test compound purity, and whether or not the data used are derived from
tumor incidence or multiplicity data.
-------
Methods for Dose-Response Assessment andRPF Calculation
EPA presents the selection of dose-response data and methods for dose-response
assessment and RPF calculation. For quantal data (i.e., tumor incidence), EPA used the multi-
stage cancer model. The Panel agrees with EPA's use of the multi-stage cancer model for
quantal data, but has specific recommendations on the parameterization of the model. The Panel
also recommends that EPA provide further detail on the assumptions regarding the distribution
of data and further detail on the parameterization of the model.
For continuous data (i.e. tumor counts), EPA used a linear model to calculate the
benchmark dose (BMD). The Panel finds that the justification for using a linear model for multi-
dose continuous data is insufficient and recommends that EPA provide further justification on
the use of a linear model. In addition, the Panel recommends that the modeling strategy for
continuous data include polynomial models or nonlinear models (e.g., the Hill model) that are
flexible enough to fit the data and would also adequately approximate a linear relationship.
Selection ofPAHsfor Inclusion in the Relative Potency Approach
EPA describes the selection of PAHs for inclusion in the RPF approach. The Panel finds
that the method for selecting the PAHs appears to be scientifically justified, but several issues
such as individual study quality and study design variability across studies are incompletely
considered. The Panel recommends that a list of quality criteria be defined, articulated, and
applied a priori, prior to the weight of the evidence evaluation. Only studies of sufficient quality
should be considered in the weight-of-evidence evaluation. The Panel recommends that once a
study is considered to have sufficient quality, the variability in study design characteristics
among studies be carefully considered prior to inclusion in the RPF calculation. Differences
among studies in some of these design characteristics may significantly affect the dose-response
within each study, which in turn, will affect the RPF calculation.
Derivation ofRPFsfor Selected PAHs
EPA describes various methods (e.g., prioritization of studies) and different averaging
approaches for deriving final RPFs. The Panel has several reservations regarding the RPF
calculation approach. The Panel is concerned about calculating RPFs based upon a single
experiment as well as calculating RPFs using studies with only a single-dose level of BaP and/or
the target PAH, particularly if it was a high dose or if, only single doses of both the target PAH
and BaP are available. The Panel does not make a recommendation on whether or not to
calculate RPFs for PAHs with these data characteristics, recognizing that the Agency will need
to apply professional judgment based on analyzing the actual available data. However,
calculating an RPF from only a single dose of BaP is reasonable if the BaP tumor incidence is in
the low-dose range and when adequate dose response data are available for the target PAH.
The Agency is encouraged to continue evaluating other methods for combining RPFs
across studies, such as using a geometric mean instead of an arithmetic mean. Using a geometric
mean would give less weight to outlier values.
-------
The Panel strongly believes that use of cancer bioassay data is essential for determining
the RPF for a given PAH. Cancer-related endpoint data are useful as supporting data but the
Panel does not recommend the use of only cancer-related endpoint data for determining the RPF.
Therefore, the Panel does not recommend calculating an RPF for dibenz[a,c]anthracene and
recommends that it be removed from Table 7.2 until further bioassay data become available.
Uncertainties and Limitations Associated with the RPF Approach
EPA discusses the uncertainties and limitations associated with using the RPF approach
for PAH mixtures risk assessment. The Panel finds that the uncertainties in the methodology of
deriving RPFs are well described. The major methodological uncertainties are clearly defined
and discussed so that little doubt remains about the methods that were used and the limitations of
the final RPF values. The Panel has the following recommendations to strengthen this section of
the document:
• Include comparisons of cancer risk estimates of complex mixtures using the RPF
approach and bioassay data.
• Include a discussion on bioavailability.
• Include a discussion of the uncertainty that arises from the difficulty and limitation of
completely characterizing mixtures.
Adequacy of Appendices for Independent Verification
The appendices in the document include dose-response data for potency calculations,
benchmark dose modeling outputs, and calculation of RPFs to allow independent verification of
the calculated RPFs. The Panel finds the appendices to be generally useful for verifying the
calculations of the RPFs, but has the following recommendations:
• Reorganize the appendices by chemical (with each identified in the Table of
Contents). This would include the corresponding BaP data for each study within each
chemical section which may be repeated across PAHs.
• Revise the plots from the BMD software output to be based on BMDs instead of the
lower confidence limits of the BMDs (BMDLs).
-------
2. INTRODUCTION
In 1993, EPA developed the document Provisional Guidance for Quantitative Risk
Assessment of PAH that recommends a Relative Potency Factor (RPF) approach for assessing
PAH mixtures. EPA's Office of Research and Development (ORD) has developed a draft
technical document, Development of a Relative Potency Factor (RPF) Approach for Polycyclic
Aromatic Hydrocarbon (PAH) Mixtures (February 2010 Draft), hereafter called "PAH Mixtures
document", to update the 1993 document by expanding the number of PAHs assessed and
including recent studies from the published literature.
PAHs are a class of chemicals that have variously been defined to include organic
compounds containing either two or more, or three or more, fused rings made up of hydrogen
and carbon atoms (WHO, 1998). The number of chemicals that comprise the PAH class is not
known, but hundreds of PAHs are thought to be present in complex mixtures (WHO, 1998).
PAHs do not occur in the environment as isolated entities; they primarily occur in complex
mixtures generated from the incomplete combustion or pyrolysis of substances containing
hydrocarbons. Some of the complex mixtures containing PAHs that are typically found in the
environment include coal tar, manufactured gas plant (MGP) residues, coke oven emissions,
diesel and gasoline exhaust, and coal plant emissions. Many PAHs are demonstrated to be
carcinogenic in animal bioassays.
EPA's PAH Mixtures document presents a component-based approach to assessing the
cancer risk of PAH mixtures by summing doses of component PAHs after scaling the doses
relative to the potency of the selected index PAH, benzo[a]pyrene (BaP). The cancer risk is then
estimated using the dose-response curve for the index PAH.
The PAH Mixtures document is limited in focus to analyzing only unsubstituted PAHs
with three or more fused aromatic hydrocarbon rings. The analysis evaluated 74 PAHs, and final
RPFs were calculated for 24 of the PAHs. Six of these PAHs have updated RPFs from the 1993
guidance, and 18 of these PAHs have new RPF values. Additionally, 3 PAHs were assigned an
RPF of zero.
ORD has requested that the Science Advisory Board (SAB) conduct a review of the
document. In response to ORD's request, the SAB Staff Office solicited nominations of experts
and formed the SAB PAH Mixtures Review Panel. The Panel held a public teleconference on
June 8, 2010, and a public meeting on June 21-23, 2010, to review EPA's Development of a
Relative Potency Factor (RPF) Approach for Polycyclic Aromatic Hydrocarbon (PAH) Mixtures
document and to deliberate over the charge questions. The Panel discussed its draft report during
a subsequent conference call on September 30, 2010. The Panel's draft report was approved by
the Chartered SAB on December 16, 2010, on a public teleconference call. There were nine
charge questions, which focused on an overview of the document, on the specific chapters of the
document, and the appendices. These charge questions are included in the Appendix and the
responses to the charge questions are presented below.
-------
3. RESPONSE TO EPA CHARGE QUESTIONS
The charge questions are presented below in italics, followed by the responses from the
Panel in normal text.
3.1. Charge Question 1 - General Charge Questions
la. Please comment on whether the report is logical, clear and concise. Please comment on
whether EPA has clearly synthesized the scientific evidence for the derivation of relative potency
factors for individual PAHs.
Overall the Panel finds the PAH Mixtures document to be logical, clear, and concise.
EPA has clearly synthesized the scientific evidence for the derivation of relative potency factors
for individual PAHs. The Panel recognizes the pragmatic need for the RPF approach. Based
upon the currently available data, the Panel recommends that EPA continue to use the RPF
approach for PAH mixtures. The Panel agrees with the Agency that in order to continue with the
RPF method, it is important to expand the number of compounds in the 1993 guidance, and for
the most part the candidate compounds for this expanded list are appropriate (see Chapter 4
discussion). The Panel also agrees that it is important to include more recent data for these
compounds (since 1993) in calculating and expanding the RPF values for PAHs, since many of
the values used in the current RPF method are based on older data.
Ib. Please comment on whether the report provides adequate context for how the proposedRPF
approach could be used in a PAH mixtures risk assessment.
The Panel finds that the PAH Mixtures document does not provide an adequate context
for how the proposed RPF approach could be used in a PAH mixtures risk assessment. The
Panel recommends that more discussion is needed to provide this context, including moving
relevant portions from Chapter 7 (sections 7.3 and 7.4) into earlier sections of the document and
providing an example (also see response to Charge Question 7e).
3.2. Charge Question 2 - Rationale for Recommending an RPF Approach
Chapter 2 presents the rationale for recommending an RPF approach. In an RPF approach,
doses of component chemicals that act in a toxicologically similar manner are added together,
after scaling the doses relative to the potency of an index chemical. Benzo[a]pyrene (B[a]P) is
selected as the index compound for this RPF approach. The RPF approach involves two key
assumptions related to the application of a dose-additivity model: (1) PAH components in the
mixture act in a similar toxicological manner; and (2) interactions among PAH mixture
components do not occur at low levels of exposure typically encountered in the environment.
-------
2a. Please comment on whether the report provides adequate justification for using an RPF
approach as a scientifically defensible method to assess the cancer risk associated with exposure
to PAH mixtures.
At the face-to-face meeting, the Panel discussed this issue in considerable detail, and
concluded that this charge question actually represents two distinct questions: first whether,
based on available literature, there is adequate justification for use of the single-agent relative
potency factor (RPF) approach, particularly with respect to the two core assumptions of this
rationale that were proposed in the PAH Mixtures document; and second, whether there is a
reasonable practical consideration in using the RPF approach at this time, independent of the
justification and underlying assumptions. The rationale for this dichotomy is outlined below.
With regard to the first question, the Panel concludes that the rationale for the proposed
RPF approach is not well justified in the current document. There are two basic assumptions that
are proposed in the document as the basis for considering the RPF approach specifically for
PAHs: first, that the chemicals of comparison are all assumed to act in a similar toxicological
manner as the reference compound (i.e., benzo[a]pyrene - BaP); and second, that their effects are
additive by assuming no significant interactions at low, environmentally relevant doses.
The Panel considered the PAH Mixtures document, the studies cited within, as well as
other data. The document discusses studies that call into question both of the underlying core
assumptions and further elaborates on a number of other uncertainties, some of which cannot
currently be validated or dismissed, that further undermine the logical basis for the assumptions
on which the RPF method is based. These are discussed in more detail in response to charge
questions 2c and 2d below, but are briefly summarized here. It is not clear that the first
assumption - i.e., that the other PAHs under consideration act in a similar toxicological manner
as BaP - is required as a foundation for the RPF method, since for these particular PAHs the
method is based on the outcomes of cancer bioassays, rather than the underlying toxicological
action. There are also results, some of which are discussed in the document (page 23, lines 11-
19; page 39, lines 3-12 and Table 2-2), that call into question the second assumption - i.e., that
there are no significant low-level interactions of PAHs in a mixture beyond simple additivity,
and therefore that the effects (cancer risks) of a mixture of agents are the simple sum of the
individual risks.
Despite these concerns about the underlying justification for the RPF method and the
logic of the two core assumptions, the Panel concludes that there is adequate practical
justification for continuing to use this approach in the near term to assess cancer risk of PAH
mixtures in the absence of a good alternative. In particular, although this Panel and previous
expert panels have strongly suggested that the EPA move toward a whole mixtures-based
approach, the fact remains that the regulatory and scientific communities do not have sufficient
information to adopt a whole mixtures approach for risk assessment at this time. Therefore, the
Panel recommends the continued use of the component-based RPF approach as the most
practical choice but recommends that this should proceed in parallel with continued development
of one or more whole mixtures-based approaches that could potentially validate the RPF
approach or possibly replace it.
-------
Given these conclusions, the Panel has several recommendations for revising the
document and moving forward with the RPF approach. First, additional historical perspective
should be included in the revised document, since it is an important component in, and
justification for the agency's practical decision to continue with the RPF method. EPA indicated
that they had previously considered implementing a whole mixtures approach, but decided
against it due to data gaps in available information. The Panel recommends including a
discussion of these previous considerations and evaluation of the data gaps. This would help
with the rationale for continuing with the RPF approach.
In parallel with the continued use of the RPF approach, the Panel recommends that EPA
begin developing a whole mixtures approach to achieve two goals: (1) to potentially validate the
RPF approach, and (2) to explore as a possible replacement for the RPF approach in the near
future. The Panel recommends that the Agency set these goals as strategic initiatives, with
specific timelines and benchmarks. This would lay the foundation for an underlying concerted
research program to achieve these goals.
The Agency should seek support from the National Toxicology Program (NTP) or other
entities to test a portfolio of 12-15 different complex mixtures in animal studies. These mixtures
should represent a diverse array of mixtures but also represent the most important mixture
classes of concern to EPA (based on the level of health concerns and/or extent of exposure) such
as coal tar, manufactured gas plant (MGP) residues, coke oven emissions, diesel and gasoline
exhaust, coal plant emissions, etc. With these data in hand, one could then potentially validate
the RPF approach and also compare an environmental mixture to this portfolio of standardized
mixtures and be able to adequately estimate risk.
These mixtures could also be compared to a surrogate mixture (e.g., a mixture
representing the ca. two dozen compounds being assessed in the RPF method) as well as BaP as
a single agent. This would provide a direct validation of the RPF method and link these results
to previous data on environmental samples for which RPF compound values are known.
2b. Please comment on whether the choice ofbenzofajpyrene as the index compound is
scientifically justified and appropriately described. Please identify and provide the rationale for
any alternative index compound(s) that should be considered.
The choice of BaP as the index compound is well justified and is appropriately described
for this RPF approach. It is the best studied PAH and meets the criteria for an index compound
for an RPF assessment. The Agency noted that a revised IRIS assessment of BaP is undergoing
a parallel review that will likely lead to a revised cancer slope factor (CSF) as well as separate
values for oral, dermal and inhalation BaP exposures. An up-to-date estimate of the CSF for BaP
is central to the validity of the RPF method since this is the index compound, and the Panel urges
the Agency to quickly finalize the BaP assessment.
-------
2c. Please comment on whether the weight of evidence indicating that PAHs, as a chemical class,
have a similar mode of carcinogenic action has been adequately described and is scientifically
justified.
There is some evidence that a subset of closely related PAHs have a similar mode of
action in inducing cancer as described in the document. This is not unexpected since the
compounds in question have already been defined in large part by their comparison to BaP.
However, although these compounds have similar modes of action, available data indicate that
they each act via different precise mechanisms when examined at a more detailed level. For
example, even though many PAHs are metabolized to reactive intermediates that then form DNA
adducts at guanine residues, their potency for conversion of DNA adducts to mutations varies
among compounds. Moreover, the pattern of guanine mutations within specific DNA sequences
varies among these adducts. By definition, these adducts are therefore acting by slightly
different mechanisms. Since cancer risk can be related to mutation rate and to specific mutations
within certain DNA sequences, this will result in different risks even though these compounds
share a similar mode of action.
Additionally, there are hundreds of other PAHs and PACs that may not act by these
modes of action and that likely, particularly in complex mixtures, contribute in positive or
negative ways to the overall carcinogenicity of the mixture. These compounds should also be
considered in the RPF method if adequate cancer bioassay data are available.
Also of importance, other PAHs in a mixture may alter the risk for known PAHs in that
mixture in more complicated ways. For example, through mass action a complex mixture may
contain total PAHs that collectively overwhelm the levels of an individual PAH such as BaP,
perhaps by 1000:1 or greater. These may collectively interfere with the overall metabolism of
BaP, or ratios of specific metabolites, or the capacity to repair DNA adducts from BaP, etc., such
that one cannot predict the cancer risk from BaP solely from its concentration in the mixture.
Therefore, this assumption is not well justified.
In addition, there is a question as to whether similar modes of action are sufficient to
predict in vivo carcinogenicity. As discussed in the PAH Mixtures document (e.g., page 35,
section 2.6), mutagenicity, genotoxicity and similar short-term assays are relatively poor
predictors of in vivo carcinogenesis. Yet a basic assumption of the document is that this
information is sufficient to predict their relative carcinogenicity. There are PAHs that are
positive in short-term in vitro assays but negative or weak in animal bioassays, and vice versa,
further undercutting this basic assumption.
The document also discusses the role of the Ah receptor (AhR) in detail as another
potential unifying mode of action for carcinogenic PAHs. However, elsewhere in the document,
EPA also acknowledges that interaction with and activation of the AhR is not a good indicator of
promotion or in vivo tumorigenesis for PAHs (as opposed to dioxins). The Panel agrees with
this latter assessment.
Taken together, these points argue that there are weaknesses in using the assumption of
PAHs having a similar mode of action as a rationale for using the RPF approach. The Panel
-------
recommends strengthening the rationale by including a discussion about the RPF approach
relying on a direct comparison between dose-response curves from actual cancer bioassay data
between BaP and the target PAH.
2d. Please comment on whether the assumption that interactions among PAH mixture
components do not occur at low levels of exposure typically encountered in the environment has
been adequately described and is scientifically justified.
The assumption that there are not significant interactions among PAHs in complex
mixtures at low doses is not well justified in the document. As discussed in the document (page
23, lines 11-19) coal tar behaved differently in in vivo carcinogenesis assays than would be
predicted from studies with BaP as a single agent, or what would likely be predicted from a RPF
approach based on BaP as the index chemical. Likewise, as discussed in the document (page 39,
lines 3-12 and Table 2-2), the complex and unpredictable results to date of simple binary
combinations of PAHs that do not follow simple additivity also undercuts the assumption of
interactions between PAHs not occurring at low environmental levels. However, in the absence
of consistent data that support a specific type of interaction (additive, sub- or super-additive, etc.)
that could be used for a variety of PAH mixtures, a default assumption of additivity is a
reasonable assumption for the purposes of the RPF analysis.
It should be noted, however, that complex mixtures such as coal tar, MGP residues,
creosote, diesel exhaust and other PAH mixtures contain hundreds of other compounds, not
included in this RPF assessment, that likely contribute to the overall biological effects of the
mixtures. Other contributing mechanisms may include: induction or suppression of specific
metabolic pathways; competition for metabolism through mass action at active sites; epigenetic
effects; promotion and progression effects; endocrine disruption, and neurological and
immunological effects that contribute to cancer risk. Other classes of potentially potent
carcinogens including substituted PAHs, volatile organic compounds (VOCs), metals, as well as
other compounds may also be present. Collectively, these mechanisms and compounds may
contribute in complicated ways to the overall cancer risk of a complex mixture. This uncertainty
can be reduced by directly testing mixtures in cancer bioassays. As discussed above the Panel
recommends the Agency test 12-15 complex mixtures of concern to EPA.
3.3. Charge Question 3 - Discussion of Previously Published RPF Approaches
This chapter presents a discussion of previously published RPF approaches. Due to the
evolution of the state of the science and an increased understanding of PAH toxicology, EPA is
reevaluating the RPF approach for PAHs in this analysis.
3. Please comment on whether the discussion provides a meaningful background on how RPFs
have been derived in the past, and the advantages and disadvantages of previous methods.
This chapter provides a summary of previous RPF approaches and provides a qualitative
discussion and comparison of cancer risk estimates based on RPF approaches with estimates
obtained from testing whole mixtures. The Panel finds that this chapter adequately summarizes
10
-------
the previous RPF approaches, but could be improved by providing more quantitative information
on the comparison between the previous RPF approaches and the whole mixtures approach. The
Panel also recommends editing Table 3-1 to use a standardized approach for reporting values
(same significant figures, scale, etc.).
3.4. Charge Question 4 - Evaluation of the Carcinogenicity of Individual PAHs
This chapter discusses the development of a database of primary literature on PAH
carcinogenicity and cancer-related endpoints and the criteria used to include or exclude studies
from the database.
4a. Please comment on whether the list of 74 PAHs (Table 2-1) included in the initial literature
search is complete. Please comment on whether the rationale for the choice of PAHs included in
the literature search has been appropriately described. Please identify other databases or
resources that should be included.
Chapter 4 of the PAH Mixtures document details the basis for the selection criteria that
was used to develop the database related to PAH carcinogenicity and cancer-related endpoints.
The list of 74 PAHs provided in Table 2-1 is believed by the Panel to be reasonable in view of
the criteria of having three or more fused rings and not containing heteroatoms, alkyl or nitro
substituents. The development of the database of primary literature on PAH carcinogenicity and
cancer-related endpoints and the criteria used to include or exclude studies from the database are
described in detail within this chapter. The database appears adequate, with the recommendation
that a recently published LARC Monograph on PAHs, Volume 92, be included as an additional
resource (IARC, 2010).
4b. Chapter 4 includes a description of how studies were selected for use in dose-response
assessment. Please comment on whether the choices and assumptions in making the selection
have been adequately described. Please comment on whether the information in Tables 4-1
through 4-14 provides adequate information to inform how decisions were made. Please
comment on whether studies were rejected or included appropriately. Please comment on
whether positive and nonpositive studies have been considered appropriately.
The basis for selecting which studies were used in dose-response assessment is clearly
delineated. The information in Tables 4-1 through 4-14 provides adequate information related to
whether certain studies were rejected or included in this document. However, the criteria for
including or rejecting a study should be revised to include only studies that are deemed to be of
sufficient quality using a priori standards as described in the response to charge question 6a.
Given this revision of including only studies of high quality, the EPA approach inappropriately
discards data that do not achieve statistical significance. Please see the response to charge
question 6a for further detail.
11
-------
4c. The methodology for the choice of studies to use in the derivation ofRPFs includes studies
where at least one PAH was tested at the same time as B[a]P. Studies where individual PAHs
were tested without concurrent testing ofB[a]P were not included in the quantification ofRPFs.
Please comment on the scientific rationale for this approach. Please comment on whether the
advantages and disadvantages of excluding certain data from the derivation ofRPFs have been
adequately described.
Chapter 4 of the document stipulates that BaP had to be tested concurrently to include a
study on the carcinogenicity or other cancer-related endpoints of one or more of these 74 PAHs.
This restriction raises a concern that data from carcinogenicity studies of high quality might be
dismissed. The Panel recommends that EPA consider whether a PAH other than BaP, with a
RPF that has a comparatively narrow range, might be able to serve as the surrogate for the BaP
index compound in those instances where BaP was not included in a bioassay. This approach
offers the possibility that additional data from studies of high quality could be included in the
development of a RPF for a given PAH. The Panel recommends that this be examined especially
in those instances where limited animal bioassay data were used to establish a RPF value.
However, in considering this alternative approach, EPA should also take into account factors that
could potentially outweigh the benefits in the establishment of a RPF for a specific PAH, such as
cross-study and cross-laboratory comparability issues.
The Panel has a few recommendations that relate to the evaluation of the carcinogenicity
studies for individual PAHs. These recommendations include providing some quality
assessment to individual studies, such as a tabulation for various studies, including information
on: 1) sample size, 2) dosing, 3) mortality (prior to tumor development), 4) defined test
compound purity, and 5) whether or not the data used are tumor incidence or multiplicity data.
In addition, the Panel recommends incorporating or reiterating some of the discussion
about alternatives for ranking RPFs provided in Appendix G into the discussions on individual
PAHs in Chapter 4 as well as in Chapter 6. For example, the Panel considers the discussion
about the influence of the route of administration on the RPF calculations to be particularly
informative.
3.5. Charge Question 5 - Methods for Dose-Response Assessment and RPF Calculation
This chapter describes the selection of dose-response data and methods for dose-response
assessment and RPF calculation from the selected datasets. The methodology for estimation of
the RPFs varied depending on the characteristics of the datasets, however, the general equation
was the ratio of the slope of the dose-response curve for the subject PAH to the slope of the dose-
response curve for B[a]P.
12
-------
5a. Please comment on whether the scientific rationale for the dose-response modeling
approaches used in the derivation ofRPFs is adequately described. Please comment on whether
there are other appropriate modeling approaches for estimating the relative potencies of PAH s.
Please describe alternative approaches (e.g., other model forms) that could be considered.
The Panel finds that the scientific rationale for the dose-modeling approaches is
adequately described. The panel does have recommendations on additional modeling approaches
that could be considered. The modeling approaches described in Chapter 5 of this document for
multi-dose studies are based on whether the data are quantal (binary) or continuous. The quantal
endpoints considered in this document include tumor incidence or incidence of cancer-related
endpoints, including frequency of mutations per number of cells interrogated. The continuous
endpoints include tumor counts (number of tumors per animal) or cancer-related endpoints of a
continuous-variable nature (e.g., number of sister chromatid exchanges, number of
morphologically transformed colonies).
When modeling quantal data, the mean model is for the probability of response (e.g.,
tumor incidence) and is generally assumed to follow a sigmoid-shape. Commonly used models
that could be used include the logistic, probit, multi-stage, and Gompertz models. Since the
multi-stage cancer model has a biological basis, it is the standard model used for cancer
incidence and is considered sufficiently flexible to accommodate the dose-response data for these
PAHs. Specifically, the multi-stage cancer model for the probability of a tumor is parameterized
based on the number of dose groups (g) with the polynomial assumed to equal g-2:
It should be noted that a model for data with g dose groups will exactly track the sample
means (here, sample proportions) if the degree of the polynomial is g-1. However, a variation of
this general model, typically used in risk assessment, assumes a monotonic relationship and
constrains all parameters to be non-negative. The Benchmark Dose (BMD) Software used in the
document makes such an assumption as the default analysis. With quantal data, assumed to be
independent across and within-dose groups, it is generally assumed that the data are binomially
distributed with binomial variance (i.e., with n subjects evaluated at a dose group, the variance in
the number "responding" is assumed to be n/u(\ - JLI) ). Alternatively, the data may follow hyper-
or hypo-binomial variability, i.e., greater than or less than binomial variability. These
assumptions are not specified in the document and should be. The BMD Software used to
estimate unknown model parameters uses a maximum likelihood estimation criterion and
standard iterative algorithms for estimation. However, these distributional assumptions and the
parameterization of the multi-stage cancer model should be clearly stated in the document. It is
not clear whether the assumption of binomial variability was verified; the assumption of
binomial variability should be verified and the document should include information about the
verification. Instead, the model checking was based on the goodness-of-fit of the mean model
and did not assess the assumptions regarding variability.
For continuous endpoints, a nonlinear dose-response shape may be expected from the
data. However, the analysis plan for continuous endpoints is to use a linear model (i.e., a linear
function). The justification for using the linear model for the multi-dose continuous data is
13
-------
insufficient and additional justification should be added. Although the linear model is the
simplest model, there are other models such as the Hill model or polynomial model that are
commonly used. EPA's justification for using the linear model is that there are a small number
of dose groups. This is an inadequate explanation.
The modeling strategy for the continuous endpoints should include polynomial models or
nonlinear models (e.g., the Hill model) that are flexible enough to fit the data and would also
adequately approximate a linear relationship. In some cases, the variance in response is assumed
to be constant over the dose range of observed data. A least-squares (or nonlinear least squares)
criterion is used to estimate unknown model parameters. In contrast, the sample variance may
change with the mean. For example, the responses in the low-dose region may have lower
variance than that observed as the dose (and response) increase. Such data may be estimated
using a quasi-likelihood estimation criterion.
For the continuous data included in this document, the assumption about whether the
variance changes across the dose groups is not addressed and the potential for a nonlinear shape
is not allowed. Only a linear model was used to estimate the mean response. A goodness-of-fit
criterion was used; if the model did not provide adequate fit, high-dose groups were sequentially
eliminated in an effort to achieve adequate fit. This strategy is arbitrary and should be avoided.
A more flexible model should be used instead that accommodates the nonlinearity of the data.
Selection of Benchmark Response (BMR)
Since the RPFs are going to be used to estimate cancer risks at generally low
environmental exposures, the calculation of RPFs should be applicable to the low-dose range,
preferably excess risks <0.10 for quantal data. Similarly for continuous data, the calculation of
RPFs should preferably be based on changes in the mean of less than or equal to one standard
deviation (and certainly less than two standard deviations) in order to remain in the low-dose
region of interest. For normally-distributed data, a change in the mean from the control mean of
two standard deviations will result in approximately 50% of the subjects in the abnormal range.
The RPF can increase or decrease substantially as dose (incidence or response) changes.
The analysis strategy described in Chapter 5 (with the suggested changes included)
should be specifically followed. Deviations from the planned analysis strategy should be clearly
explained.
To illustrate the use of a nonlinear model, the in vitro clastogenicity dose-response data
of Tong et al (1981) (Table C-19, page C-85 of PAH Mixtures document) is reanalyzed. For
convenience, the data table is reproduced in Table 1. The data clearly follow a nonlinear
relationship, which is particularly evident, considering the two highest concentrations of
benz[a]anthracene (BaA) that have similar responses with a log change in concentration.
14
-------
Table 1: Data from Tong et al, 1981, for sister chromatid exchange summary data
(Record number: 21710; Table C-19, page C-85). The BMR was set to the control
mean from the predicted Hill model + SD of the control group. The BMDs are
estimated from the Hill model using the specified BMR.
PAH
Control
BaP
BaP
BaP
Control
BaA
BaA
BaA
Concentration
(Molar)
0
10-b
io-5
io-4
0
To13
io-4
io-j
Mean Sister
Chromatid
Exchange/cell
11.15
16.15
59.75
103.3
15.75
21.2
29.15
26.2
Standard
Deviation
(SD)
3.81
3.83
16.96
22.75
5.18
9.59
9.93
6.96
Benchmark
Response
(BMR)
13.7
20.9
Benchmark
Dose
(BMD)
(Molar)
4xlO'7
7xlO'6
Instead of fitting a linear model to these data, a 3-parameter Hill model is selected, which
can accommodate an asymptotic response for large concentrations, i.e.,
" x+e'
where x is the concentration of the PAH, a is the response for the control group, y is the range of
response such that a+y is the asymptote for large x. Since only sample means and standard
deviations are available at each concentration level, a weighted analysis is imposed, with weights
set to the inverse of the sample standard deviation at each concentration. Unknown parameters
are estimated using a weighted least squares criterion in a Gauss-Newton iterative algorithm
using PROC NLIN in SAS (version 9.2). The resulting predicted models for BaP and BaA are
provided in Figure 1. Using all of the data, a Hill model adequately fits the observed sample
means for both PAHs.
15
-------
Figure 1: Observed sample means and predicted response from a Hill model. Sample means are
denoted with dots and +/- one standard deviation from each mean is denoted by the error bars.
A
PAH s BaA
40
_ 30
20
10
B
140
120
100
80
60
40
20
PAH = BaP
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010
PAH Concentration (M)
0.00000 0.00005
PAH Concentration (M)
0.00010
The specified BMR for continuous data is one standard deviation (SD) above the control
mean as predicted from the Hill model (shown in Table 1). For BaP, the estimated BMDiso is
4x 10"7 and for BaA, the estimate is ?x 10~6. However, in Table E-14 (page E-31), the BMR and
BMD values are blank and the point estimate responses are 92 and 13 for BaP and BaA,
respectively; and the point estimate dose is IxlO"4 for both compounds. It is not clear how the
point estimate responses were calculated. This is an example where the described analysis plan
does not seem to be followed without any explanation of why it was not followed.
5b. For each individual dataset considered in the assessment, the B[a]P dose-response was
calculated from the study-specific data. Please comment on whether this approach has been
appropriately described. If there are additional approaches using the available data that should
be considered, please describe how the approach could lead to a better estimate of cancer risk.
The strategy of using study-specific data for the BaP dose-response with PAH dose-
response is advantageous since downstream calculations are intra-study and avoid comparisons
that would require accounting for possible study-specific effects. This strategy has been
appropriately described and the Panel does not have other approaches to suggest.
It should be noted that the estimates of BaP slope across studies with different
characteristics are very different. The range of the estimates can be more than 1,000 fold. This
supports the idea of using study-specific estimates for calculating the RPFs.
16
-------
5c. The point of departure for slope estimation that has been used for the derivation ofRPFs is
the benchmark dose (BMD) estimate rather than the lower confidence limit on the benchmark
dose (BMDL). Please comment on whether this approach is scientifically justified and
adequately described. Please comment on whether alternative approaches should be considered.
It is correct to base the derivation of the RPFs on the estimate derived from the BMD,
rather than the lower confidence limit on the benchmark dose (BMDL), in order to obtain an
estimate of the total exposure for a mixture (expressed as the total BaP equivalent dose). Due to
chance experimental variation, some of the RPFs will be overestimated and some will be
underestimated. These biases will tend to cancel each other for the total exposure of a mixture.
On the other hand, when the study sizes are similar, the BMDLs between the BaP and PAH may
be stable. But when the two studies have different precision, the ratio of BMDLs is tenuous.
Therefore, the ratio of BMDs is advisable. The Panel does not believe that any alternative
approaches are necessary.
5d. Please comment on the methodology used for the RPF calculations for multidose and single
dose datasets. Please comment on whether the process for calculating RPFs from the various
datasets is scientifically justified and adequately described. Please comment on the utilization of
high response levels in some instances as the point of comparison. Please describe alternative
approaches that could lead to a better estimate of cancer risk that should be considered using
the available data. Please comment on whether the considerations for RPF calculation as
outlined in Sections 5.6 and 5.7 are scientifically justified and adequately
described.
When multiple doses are available for dose-response modeling, all of the data should be
used with a sufficiently flexible model, e.g., the multi-stage cancer model or a polynomial model
for continuous endpoints. An example of such an analysis strategy is given in 5a above. In the
Appendix, there are cases where single-dose data were used when multiple doses were available;
this should be explained.
Generally, the Panel is concerned about using high-BMR values to calculate the RPFs in
single-dose studies. If the dose-response curves were parallel across PAHs, then the choice of
BMR would not impact the estimation of a relative potency factor. However, as discussed in
earlier chapters, it is generally assumed that the chemicals are not dilutions of one another, so
their dose-response curves will generally not be parallel. Thus, the choice of the BMR should be
in the low dose-region. However, in some special cases, the RPF calculation is not dependent on
the response level. For example, consider the data from a BaP single-dose study and multi-dose
comparison PAH for benzo[k]fluoranthene (BkF) (LaVoie et al, 1982). For convenience, the
data from Table C-l, page C-4 of the PAH Mixtures document are reproduced below in Table 2.
17
-------
Table 2: Data from LaVoie et al, 1982 for dermal bioassay data
(Record number: 630; Table C-l, page C-4) - primarily squamous cell
papilloma in female mice. The data include a single-dose study for BaP
and multiple-dose study for the PAH, BkF.
PAH
Control
BaP
BkF
BkF
BkF
Dose
(Hg/mouse)
0
30
30
100
1000
Number of
Animals in
Group
20
20
20
20
20
Number of
Animals
with
Tumors
0
17
1
5
15
%
Tumor-
bearing
animals
0
85
5
25
75
Suppose a one-stage model is used for analysis of the single-point BaP study, i.e.,
where /?0 =0, x is the dose of BaP, and yfl, is the unknown parameter associated with the slope.
Assuming a zero background response rate (i.e., /?0 =0), the BMD(IO) is estimated as
BMD(IG) = -log(0.9)//?! and the BMD(85) is estimated as BMD(S5) = -log(0.15)//?j . Since
there are four dose groups for BkF, a multi-stage model is used, parameterized with linear and
quadratic terms (i.e., g-2= 2 for a second-degree model):
where x is the dose of BkF and again we assume /?0 =0. However, in the EPA document, (32 was
set to zero and the one-stage model was used due to convergence problems. Therefore the same
parameterization is used for both BaP and BkF. The fitted dose-response curves are provided in
Figure 2. Notice the predicted response from the single-dose study is the sample mean (here,
observed sample proportion).
18
-------
Figure 2: Single-dose data for BaP (red) and multi-
dose data for BkF (LaVoie et al, 1982).
200 400 600 800 1000 1200
Dose (meg/mouse)
When the one-stage model is used for both chemicals, the choice of BMR is not relevant
in the calculation of the RPF. Consider the following algebraic manipulations to demonstrate for
th-
a general BMR=|j,o and for a general j PAH:
/LiJBMD(/ii0)BaP
-\og(\-jU0)/j3BaP
Thus, the RPF is not a function of the BMR when a one-stage model is used for both the BaP and
comparison PAH. To illustrate from the LaVoie (1982) data, the results for a BMR of 10% and
85% (the observed response from BaP) are given in Table 3. The resulting RPFs are identical.
19
-------
Table 3: Illustration with BaP single dose study and multi-dose comparison PAH, here BkF from
LaVoieetal, 1982.
LaVoieetal 1982 data
BaP
BkF
slope PAH
slope BaP
BMD10
estimates
G*g)
1.7
64.6
Slope =
0.1/BMD10
0.060
0.0015
0.025
BMD85
estimates
(ng)
30
1163
Slope =
0.85/BMD85
0.028
0.0007
0.025
This illustration demonstrates that in a single-dose study, a one-stage model can be fit,
which will exactly predict the observed mean response. In this case, the ratio of slopes for
calculating the RPF is not dependent on the BMR. However, with the single-dose studies, there
is no way to verify the prediction where data are not available. Therefore the result is based on a
lack of information rather than evidence that both the BaP and PAH dose-response data are
adequately approximated with one-stage models.
Although the use of single-dose study data may be helpful in informing the risk
assessment, these studies are clearly less informative than multi-dose studies. When single-dose
studies are used to calculate the RPF, the Panel recommends describing the impact on the RPF
calculation. For example, in Table 7-1, the Panel recommends including the number of studies
per RPF calculation based on a one-dose study.
For section 5.7 of the document, the Panel recommends the use of a (g-l)-degree
polynomial in the multi-stage model (page 111, lines 31-36) instead of reducing the degree of the
polynomial. This model will exactly track the observed sample means.
3.6. Charge Question 6 - Selection of PAHs for Inclusion in the Relative Potency
Approach
This chapter describes the selection of PAHs for inclusion in the RPF approach. The evaluation
focuses on whether the available data were adequate to assess the carcinogenic potential of each
compound. If the data were not considered adequate, then the PAH was excluded.
6a. Please comment on whether the rationale for the weight-of-evidence evaluation is
scientifically justified and adequately described. Please comment on whether the approach
adequately considers the available information. Please comment on whether other information
(e.g., additional structure-activity) could contribute further to the weight-of-evidence evaluation
and how this information could be utilized in the analysis.
The Panel believes that the method for selecting the PAH appears to be scientifically
justified, but has recommendations about several issues that are incompletely considered. These
20
-------
issues include: (1) the quality of the individual studies considered and (2) the variability of other
design characteristics among studies, and how this may weigh on their evaluation prior to
inclusion in the weight-of-evidence evaluation or calculation of an RPF.
Regarding the quality of individual studies being considered, the Panel recommends that a
list of quality criteria should be defined, articulated, and applied a priori (e.g., methodologically
robust, such as inclusion of an adequate control group, sample size, dose level, number of doses,
number of PAHs measured, purity of the compounds considered, etc.) prior to the weight-of-
evidence evaluation. This information should be illustrated in the form of tables or individual
graphs. Only studies of sufficient quality, as defined a priori, should be considered in the
weight-of-evidence evaluation.
The Panel recommends that once a study is considered to have sufficient quality, the
variability of other study characteristics among studies should be carefully considered prior to
their use in the calculation of the RPF. Some of these study characteristics include: species,
strain and sex of animal model, route of exposure, form of exposure (injection, implantation,
etc.), frequency of administration, exposure duration, location of tumors, types of tumors
(papillomas, adenomas, carcinomas etc.), and stage of tumors (benign, malignant). Differences
among studies in some of these characteristics may significantly affect the dose-response within
each study, which in turn, will affect the RPF calculation. For example, for a given PAH one
may have one study that used skin tumorigenesis and another that used implantation of solid
material intratracheally. The latter study, if positive, might add weight to the overall
determination that the PAH is tumorigenic in animals, but may be a poor study from which to
calculate dose-response or relative potency. Or one could be comparing one study that has a
physiological exposure route such as inhalation, ingestion, or dermal application, with one with a
non-physiological exposure such as an intraperitoneal or implantation study. Likewise, there are
tumor models, such as the A/J mouse, where tumor multiplicity counting is required due to the
high penetrance of tumor response. It is not a simple task to reconcile these studies with other
studies of tumor incidence for the purposes of quantitatively assessing dose-response.
There is no simple formulaic method for determining, a priori, how to include or exclude
such studies or how to weight them, since this will vary depending on the individual PAH and
also depending on what studies are available; it also requires a measure of expert scientific
judgment. Instead, the Agency should clearly articulate the quality criteria (e.g., expand and
articulate the characteristics listed in Table 7.1), and then only use studies with adequate quality
to calculate RPFs. Weighting factors may be required for inclusion of some studies for RPF
calculations, or they may only be valuable as a qualitative, weight-of-evidence assessment of
carcinogenicity rather than for quantitative RPF calculations. The criteria for how such decisions
are made for each study and each PAH should be clearly defined and described by the Agency as
part of its assessment.
The EPA approach inappropriately discards data that do not achieve statistical
significance. Lack of statistical significance does not necessarily mean that an effect is zero. It
could be that there is an effect with biological relevance, but the sample sizes were too small to
achieve statistical significance. Using a cutoff P-value of 0.05 for inclusion of data in the
weight-of-evidence assessment is arbitrary. It can create a scenario where there are two studies
21
-------
of equally high quality and one study is included because it has a P-value = 0.04 and the other
study is not included because it has a P-value = 0.06. A study of high quality that produces a
low statistically non-significant RPF is relevant and must be included in calculating the best
(weighted average) estimate for an RPF. Discarding values in the lower tail of a statistical
distribution, solely due to lack of statistical significance, results in a biased estimate of the effect.
6b. The weight-of-evidence analysis does not include data related to Ah-receptor binding,
cytotoxicity or tumor promotion. Please comment on whether the scientific rationale for this
decision is appropriate. If these data should be considered in the derivation of RPFs, please
describe how they should be incorporated into the analysis.
The Panel finds that the rationale for omission of Ah-receptor data is well justified.
Additional information is not necessary. The Panel also agrees with EPA's decision that once
information demonstrating tumor formation is obtained, additional information on cytotoxicity
and tumor promotion is not necessary. However, the document should clearly state the reasons
for the omission of these data.
6c. The analysis uses an RPF detection limit as a means of comparing positive and nonpositive
(or negative) bioassays. Please comment on whether this method is scientifically justified and
adequately described.
EPA employed the use of an "RPF detection limit" to evaluate the results of positive and
nonpositive results in the same test system. The "RPF detection limit" was defined as the RPF
determined by the lowest response that would have been statistically significant for the subject
PAH and the actual benzo[a]pyrene response. The Panel did not find this definition to be clear
nor did the Panel find the description of the use of "RPF detection limits" to be clear. The Panel
does not recommend utilizing statistical significance as a means to determine which studies to
include or exclude. As discussed above, the Panel recommends assessing study quality to
determine which studies to include or exclude from the weight-of-evidence evaluation. It is
scientifically incorrect to discard data of sufficient high quality that do not achieve statistical
significance and therefore, the Panel does not recommend using "RPF detection limits" for that
purpose.
6d. Graphic arrays of the calculated RPF s (Figures 6-2 through 6-35) are presented as a means
of representing the variability in RPF s from different data sources, the weight-of-evidence for
carcinogenic potential, and the basis for the selected RPF. Please comment on whether the
figures are informative and adequately described. Please comment on whether there is other
information that should be included in the figures. Please comment on whether the narratives
are informative and complete.
The Panel finds that Figures 6-2 through 6-35 provide a good summary of the individual
studies considered and the variability of individual RPF estimates across studies. However, the
Panel recommends clearly indicating which studies were used to estimate the final RPF. This
would make the figures much more informative.
22
-------
With respect to the presentation of RPFs for individual studies, the Panel proposes that
rather than graphically displaying the RPF for each individual study as a bar, it can be shown as
a point estimate coupled with information on variability (e.g., standard error, standard deviation,
confidence interval, and range). The information on variability in the study is viewed as key, to
help the reader interpret the study findings.
The Panel recommends that, for ease of reading and to ensure completeness, the
narratives be presented in a consistent structure and format, both in terms of the information
presented, as well as the order in which they are presented. The Panel also recommends
integrating information provided in Appendix G into the narratives that correspond to Figures 6-
2 through 6-3 5.
3.7. Charge Question 7 - Derivation of RPFs for Selected PAHs
This chapter describes various methods (e.g. prioritization of studies) and different approaches
for deriving final RPFs (e.g., arithmetic mean). Final RPFs were derived by averaging the
individual study RPFs (across all exposure routes) calculated from bioassay data for PAHs that
had at least one RPF based on a bioassay. The exception was dibenz[a, c]anthracene, where the
RPF was calculated from cancer-related endpoint data.
7a. Please comment on the scientific justification for the approach for deriving the final RPFs
and the discussion of alternative options for the estimation of the final RPFs. Please comment
on the reporting of the range of RPFs as a measure of variability instead of a confidence interval.
Please comment on whether the data are adequate to support more (or less) precision in
deriving the RPFs.
The Panel believes that presenting the range instead of a confidence interval is
appropriate. The Panel does have reservations regarding several aspects of the RPF calculation
approach. First, the Panel has concerns regarding calculating RPFs based upon a single
experiment (e.g., 1 lH-benz[b,c]aceanthrylene, benzo[g,h,i]perylene, benzo[e]aceanthrylene,
benz[j]aceanthrylene, dibenzo[a,h]pyrene, indeno[l,2,3-c,d]pyrene and naphtho[2,3-e]pyrene).
Second, there is concern regarding the use of data for calculating RPF values in which there was
only a single-dose level of BaP and/or the target PAH being evaluated (e.g., benz[a]anthracene,
1 lH-benz[b,c]aceanthrylene, benzo[e]aceanthrylene, naphtho[2,3-e]pyrene and fluoranthene),
particularly if it was a high dose, and/or particularly if only single doses of both the target PAH
and BaP are available. An RPF can be calculated from only a single dose of BaP when dose
response data are available for the PAH. Since the RPF varies with the level of the tumor
incidence, RPFs should only be calculated from a single dose of BaP if the tumor incidence at
the dose of BaP is in the low-dose range; certainly, only if the BaP tumor incidence is less than
50%. Finally, there is concern about calculating the arithmetic mean for PAHs that have
markedly divergent individual RPFs (e.g., benzo[c]fluorene). The Panel does not make a
recommendation on whether or not to calculate RPFs for PAHs with these data characteristics,
recognizing that the Agency will need to apply professional judgment based on analyzing the
actual available data.
23
-------
The Agency is encouraged to continue evaluating other methods, such as using a
geometric mean instead of an arithmetic mean. Where sufficient data are available, the use of a
geometric mean would give less weight to outlier values. Further, examination of the variability
for estimates of RPFs in several of the figures indicates that a log-normal distribution may be
appropriate to describe the variability of RPFs. Hence, the geometric mean would be a better
estimate of central tendency. Also, the best central estimate would be a weighted geometric
mean where the weights are inversely proportional to the square of the standard errors. That is,
RPFs with large standard errors would receive less weight. The Panel believes that calculating
RPFs to one significant figure is appropriate.
7b. Please comment on whether the scientific rationale for consideration ofbioassay data
versus cancer-related endpoint data has been adequately described. Please comment on
whether the cancer-related endpoint data could be used in a more quantitative manner. Please
comment on the justification of the final RPF derived for dibenzfa,cj'anthracene. Please
comment on the use of tumor multiplicity data in the weight-of-evidence evaluations and for the
determination of the P^PFs.
The Panel believes that the scientific rationale for considering bioassay data versus
cancer-related endpoint data has been adequately described. The Panel strongly believes that the
use of cancer bioassay data is essential for determining the RPF for a given PAH. Cancer-related
endpoint data are useful as supporting data, but the Panel does not recommend the use of only
cancer-related endpoint data for determining the RPF. As such, the Panel does not have
recommendations on how to use cancer-related endpoint data in a more quantitative manner.
The Panel does not recommend calculating an RPF for dibenz[a,c]anthracene and recommends
that it be removed from Table 7.2 until further bioassay data become available.
The Panel recommends that additional information and justification be provided for the
inclusion or exclusion of cancer bioassay data for PAHs that did not give significant tumor
responses in well-designed studies. One suggestion is to include the IARC classification for
those PAHs where a classification exists in Table 7.1 or perhaps in Table 7.3. The Panel
believes that there is a need for some additional measure of the quality of individual studies used
in determining the final RPF values. This is important in addition to the confidence ratings
provided in Table 7.3 (see also further discussion below). The Panel also strongly believes that
more cancer bioassay data with mixtures would be extremely helpful in further validating the
RPF approach.
Tumor multiplicity (continuous data; average number of tumors per mouse) and tumor
incidence (quantal data; percentage of mice with tumors) represent different measures of
tumorgenicity/carcinogenicity. In the document, RPFs calculated from tumor multiplicity data
are combined with other RPFs calculated from tumor incidence data to calculate final RPFs. An
example of the problem is benzo[c]fluorene. The divergent RPFs used to calculate the final RPF
value for benzo[c]fluorene in Table 7.1 come from averaging a study where multiplicity data
were used (RPF of 50) and one where incidence data were used (RPF of 1). It is recommended
that RPF values not be averaged from these two different measures without sufficient
justification for using the multiplicity data. In this regard, accurate assessment of differences in
potency using tumor multiplicity requires adequate dose-response data. For accurate
24
-------
comparisons, at least 3 doses of each PAH should be available for comparison. In addition, these
doses should be distributed across the dose-response range and not be clustered on the high or
low end of the dose response range. In lieu of adequate dose-response data for tumor
multiplicity, the Panel recommends that only tumor incidence data be used to calculate final
RPFs. Additionally, if calculated RPFs for a given PAH still remain divergent across multiple
well-designed studies due to multiple factors (e.g., combining incidence and multiplicity,
combining data from different organs, combining data from different routes of exposure, etc) the
Agency may wish to consider use of the geometric mean in place of the arithmetic mean as
discussed above.
7c. Please comment on 'whether the recommendation to apply the proposed RPFs across all
routes of exposure is adequately described. Please comment on whether there is additional
scientific information that would inform this recommendation. Please comment on whether the
available data are adequate to recommend exposure route-or target organ-specific RPFs.
The Panel finds that the recommendation to apply the proposed RPFs across all routes of
exposure is adequately described. The Panel does not believe that there would be much value in
providing route- or target organ-specific RPFs at the present time because a significant
proportion of the studies used to calculate the final RPFs involved dermal
application/carcinogenesis (approximately 60% of the studies involve dermal application to mice
and >90% of the studies were conducted in mice). Additional studies and data using different
routes of exposure and tumor data from other organ sites would be necessary to calculate such
RPFs. Although the Panel agrees with the decision to not calculate separate RPF values for
different routes of exposure, the route of exposure may be an issue of concern for generating
RPF values for compounds where the available data are only via non-physiological routes (e.g.,
benzo[g,h,i]perylene, lung implantation in rat only; benzo[j]aceanthrylene, intra-peritoneal only;
fluoranthrene, intra-peritoneal only; indeno[l,2,3-e]pyrene, lung inplantation in rat only).
Additional dermal or oral tumor studies may be warranted in these cases since the route of
exposure can play an important role in bioavailability and toxicokinetics that may alter the
relative potency of the test compound as compared to BaP, when tested via a more standard route
of exposure. A sensitivity analysis should be performed to determine, in those cases where there
are data from several routes of exposure, whether these alternative routes cause a particular bias
or greater variability in the RPF values. It is interesting to note in this regard, that some
compounds, such as benzo[c]fluorene, demonstrate widely divergent RPFs in studies using
different routes of exposure (in this case, oral versus interperitoneal, with values of 1 and 50)
(see also dibenz[a,h]anthracene and dibenzo[a,l]pyrene). Without additional supporting data or
justification, the Panel does not recommend developing RPFs for compounds with data only
from studies using non-physiological routes of exposure.
25
-------
7d. Please comment on whether the scientific rationale for the assignment of an RPF of zero for
some PAHs is adequately described. Please comment on whether there are other data that
should be considered to assess whether an RPF of zero is appropriate. Please comment on
whether the scientific rationale for assigning no RPF based on inadequate data for some PAHs
is adequately described. Please comment on whether there are alternative methods for assigning
RPFs to these PAHs. Please comment on whether the text provides adequate distinction between
PAHs with RPFs of zero and PAHs with no selected RPF and whether this distinction is useful
for describing uncertainty in determining the cancer risk associated with PAH exposure.
The Panel generally believes that the scientific rationale presented in the document for
assignment of an RPF of zero, the assignment of no RPF, and the distinction between them is
adequately described. However, the Panel does have concern regarding the quality of the data
used to assign an RPF of zero for some studies and also regarding the inconsistent use of studies
with RPFs of zero in calculating the final RPFs. The Panel recommends that a consistent
approach be adopted for using RPFs of zero for all compounds for which final RPFs are
calculated. In addition, the Panel recommends that the Agency continue to evaluate how RPFs
of zero are calculated as well as the rationale for assigning no selected RPF values. In addition,
the Panel recommends the Agency discontinue assigning a value of zero to studies of high
quality that have non-statistically significant results. See the response to charge question 6a for
further details. The Panel did not identify alternative methods for assigning RPFs to PAHs that
had inadequate data.
7e. The final RPFs are characterized with confidence ratings. Please comment on whether the
rationale for the confidence ratings is appropriately described. Please comment on whether
there are other approaches for describing confidence using the available data that could be
applied in either a qualitative or quantitative manner that would be more useful for risk
assessment.
In general, the Panel believes that characterizing the final RPFs with confidence ratings is
a good idea and finds that the rationale for confidence ratings is appropriately described.
However, the confidence ratings do not appear to give any indication of the overall quality of the
data being assessed and used for the RPF calculation. Based on the information provided in
Table 7.3, confidence ratings appear to be related to the number of studies used, data from more
than one route of exposure, presence of non-cancer endpoint supporting data to calculate the
RPFs, etc. The Panel strongly believes that there needs to be some measure of the quality of the
individual studies used to generate the RPFs. In this context, quality refers to study
characteristics, such as sample size and statistical power, presence or absence of non-lethal
toxicity, unusual mortality, and other potential confounding factors. Also, the Panel's comments
and recommendations in the response to Charge Question 7c may be useful in developing
confidence ratings.
Chapter 7 also includes a description of how the RPF method is used to calculate relative
cancer risk from exposure to PAH mixtures (section 7.3). In addition, there is a section (section
7.4) dealing with the use of age-dependent adjustment factors (ADAFs) to adjust for differences
in susceptibility during early life (i.e., <16 years of age). The Panel believes that these two
sections are extremely important to the overall presentation of the document and are somehow
lost by inclusion at the end of Chapter 7. It is strongly recommended that the information on
26
-------
cancer risk assessment (sections 7.3 and 7.4) be moved to the beginning of the document as a
separate section.
3.8. Charge Question 8 - Uncertainties and Limitations Associated with the RPF
Approach
This chapter discusses the uncertainties and limitations associated with using the RPF approach
for PAH mixtures risk assessment. Many of the general uncertainties related to chemical-
specific risk assessment are also applicable to the proposed RPF approach for PAHs. In
addition, uncertainties exist regarding the selection of data and dose-response assessment
methodology, the selection of PAHs for inclusion in the analysis, the derivation of the final RPF,
the assumption of a common mode of action and dose additivity, and the extrapolation ofRPFs
across exposure routes.
8. Please comment on whether, overall, the document describes the uncertainties and limitations
in the methodology used to derive RPFs in a transparent manner. Please comment on whether
the most important uncertainties and limitations are identified. Please comment on whether
there is existing information that could be used to evaluate the accuracy or validity of the RPF
values to predict the cancer risk associated with exposure to PAH mixtures.
The uncertainties in the methodology of deriving RPFs are described quite well in the
PAH Mixtures document. The major methodological uncertainties are clearly defined and
discussed so that there is little doubt about the methods that were used and the limitations of the
final RPF values reported.
More data dealing with the comparisons of the RPF approach and estimates of cancer risk
derived from complex mixtures are needed, which would reduce some of the uncertainties
associated with the RPF approach described in the document. The feasibility of directly studying
complex mixtures is illustrated by the limited pair of existing data sets. Chronic bioassays in
mice for two synthesized coal tar mixtures were conducted at the National Center for
Toxicological Research, Food and Drug Administration (Gulp et al., 1998). The RPF approach
applied to these data were reported in the Electric Power Research Institute (EPRI) public
comments (Rohr, 2010). Comparisons of cancer risk observed in the chronic animal bioassays
for the two coal tar mixtures were within a factor of two to four (lower) of the cancer risks based
on the RPF approach. This is an encouraging result for use of the RPF approach, albeit for only
two mixtures. Additional comparisons, such as those submitted by EPRI, should be added to the
document as it provides very useful information about the RPF approach. Statistical variation of
cancer risk estimates between chronic animal bioassays on the order of three to four is expected
(Gaylor et al., 2000). More data dealing with the comparisons of the RPF approach and
estimates of cancer from tested mixtures are needed.
Additional mixtures of PAHs need to be studied in chronic animal bioassays in order to
compare the observed cancer risk of a mixture with the risk estimated from the RPF approach.
Section 3.1 of the PAH Mixtures document discusses the availability of several studies on
mixtures that provide data for comparing cancer risk estimates using the RPF approach with
27
-------
direct estimates of risk from the mixtures. Unfortunately, no quantitative information was
presented in the document to indicate the potential size of uncertainty for the RPF approach.
This quantitative information needs to be added to the document in order to evaluate the
accuracy and precision of the RPF approach from existing examples.
The cancer slope factor for BaP is multiplied by the RPFs in order to obtain cancer unit
risk factors for each of the PAHs. Hence, the cancer unit risk factor for BaP is critical to the
calculation of the cancer risk estimate for a mixture using the RPF approach. Based on old
studies, the upper limit of the cancer unit risk factor for lifetime oral exposure to BaP is
7.3 x 10"3 per [ig/ kg per day listed in the EPA Integrated Risk Information System (IRIS), 1994.
Based on a Good Laboratory Practice (GLP) study the upper limit of the cancer unit risk factor
for BaP is 1.2 x 10~3 per ng/kg per day (Gaylor et al., 2000). Because of the relatively large
uncertainty in the cancer unit risk factor for BaP, this value needs to be updated before reliable
estimates of cancer risk can be derived for mixtures of PAHs.
Extending the classes of PAH should be considered by incorporating other PAH
derivatives, e.g., PACs that occur in mixtures, particularly where bioassays exist such as for
nitro-aromatics and alkylated PAHs.
The relevance of high doses in animal studies to the much lower doses experienced by
humans is not discussed in the document. The Panel recommends that additional information or
discussion of the uncertainty that arises from extrapolating from high animal doses to low human
doses be added to the document.
The state of a single PAH administered to animals in bioassays may be different from the
state of the same PAHs in mixtures where they may not be easily desorbed from solid particles.
The bioavailability to humans for PAHs in a mixture needs to be compared to the bioavailability
in animal bioassay experiments that utilize purified PAH compounds. Cancer risk estimates
based on the RPF values and the total concentration of PAH in mixtures may be overestimated.
Using measured concentrations of PAHs in mixtures, sensitivity analyses can indicate
which uncertainties in individual RPFs have a significant impact on the total BaP equivalents for
a mixture. EPA should consider adding this to the document, perhaps by using the mixtures
discussed in the EPRI comments.
The composition for each individual mixture must be adequately determined, otherwise
uncertainty is added to the RPF approach. Completely characterizing mixtures is difficult, and
this limitation and uncertainty should be discussed. For example, different PAHs may have
different effects on the induction phase I and/or phase II enzymes that might affect the metabolic
activation or deactivation of other potentially highly tumorigenic PAHs, i.e., a non-additive
effect as mentioned in the PAH Mixtures document. Various PAHs may inhibit each other.
Mixtures may or may not contain substances that act as promoters of tumorigenesis rather than
as genotoxic initiators. Without adequately characterizing mixtures, these effects may not be
considered.
28
-------
3.9. Charge Question 9 - Adequacy of Appendices for Independent Verification
9. Please comment on whether the information in the Appendices is adequate to allow
independent verification of the calculated RPFs. If not, please comment on what additional
information would be useful.
There are 7 appendices in the document and the information contained in them include: a
bibliography of secondary sources reviewed for identification of primary literature, a
bibliography of studies without BaP as a reference compound, dose-response data for potency
calculations, benchmark dose modeling outputs, calculation of RPFs, an example calculation of
an RPF detection limit, and evaluation of alternatives for ranking RPFs.
The appendices are generally useful for verifying the calculations of the RPFs. However,
the Panel recommends reorganizing the appendices by chemical (with each identified in the
Table of Contents). This would include the corresponding BaP data for each study within each
chemical section which may be repeated across PAHs.
The plots from the Benchmark Dose Software output are useful but it should be noted
that the linear extrapolation to the origin is based on BMDLs instead of BMDs. The calculation
of the multi-stage cancer slope factor is also given based on the BMDL instead of the BMD. The
Panel recommends that the slope factors be added to these appendices based on the BMD -
which is the approach taken in the document.
29
-------
4. REFERENCES
Gulp SJ, Gaylor DW, Sheldon WG, Goldstein LS, and Beland FA. 1998. A comparison of the
tumors induced by coal tar and benzo[a]pyrene in a 2-year bioassay. Carcinogenesis, 19: 117-
124.
Gaylor DW, Gulp SJ, Goldstein LS, and Beland FA. 2000. Cancer risk estimation for mixtures
of coal tars and benzo[a[pyrene. Risk Analysis, 20: 81-85.
IARC (International Agency for Research on Cancer). 2010. World Health Organization. IARC
Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 92. Some Non-
heterocyclic Polycyclic Aromatic Hydrocarbons and Some Related Exposures. Available at:
http://monographs.iarc.fr/ENG/Monographs/vol92/index.php.
LaVoie EJ,Amin S, Hecht SS, Furuya K, Hoffmann D. 1982. Tumour initiating activity of
dihydrodiols of benzo[b]fluoranthene, benzo[j]fluoranthene, and benzo[k]fluoranthene.
Carcinogenesis, 3:49-52.
Rohr AC. 2010. Comments Prepared for Consideration by Science Advisory Board PAH
Mixtures Review Panel. Electric Power Research Institute (EPRI). Available at:
http://yosemite.epa.gov/sab/sabproduct.nsf/E10147F622E092188525773E0069A262/$File/EPRI
+Comments-SAB+PAH+Mixtures+Review+Panel.pdf
long C, Laspia MF, Telang S, Williams GM. 1981. The use of adult rat liver cultures in the
detection of the genotoxicity of various polycyclic aromatic hydrocarbons. Environmental
Mutagenesis, 3:477-487.
WHO (World Health Organization). 1998. Selected non-heterocyclic polycyclic aromatic
hydrocarbons Environmental health criteria. Vol. 202. International Programme on Chemical
Safety, Geneva, Switzerland.
30
-------
APPENDIX - EPA CHARGE QUESTIONS
National Center of Environmental Assessment Charge to External Reviewers for the
Development of a Relative Potency Factor (RPF) Approach for Poly cyclic Aromatic
Hydrocarbon (PAH) Mixtures for the IRIS Program
February 2010
U.S. EPA's IRIS Program is seeking an external peer review of the scientific basis supporting
the document titled Relative Potency Factor (RPF) Approach for Polycyclic Aromatic
Hydrocarbon (PAH) Mixtures that will appear on the Agency's online database, the Integrated
Risk Information System (IRIS). IRIS is a human health assessment program that evaluates
quantitative and qualitative 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.
PAHs do not occur in the environment as isolated entities; they primarily occur in complex
mixtures generated from the incomplete combustion or pyrolysis of substances containing carbon
and hydrogen. Many PAHs are demonstrated tumorigenic agents in animal bioassays and are
active in cancer-related in vivo or in vitro tests. In addition, PAHs exhibit noncancer effects that
may be of concern to public health. The analysis presented in the document under review
represents an RPF approach for estimating cancer risk and is characterized as one approach to
assessing cancer risk from exposure to PAH mixtures.
In concordance with U.S. EPA (2000, 1986) guidance for health risk assessment of chemical
mixtures, assessment of the cancer risk from human exposure to a particular PAH mixture would
best be conducted with quantitative information on the dose-response relationship for the mixture
of concern. When data for the mixture of concern are not available, the recommendation is to
use toxicity data on a sufficiently similar mixture. However, quantitative cancer dose-response
information exists only for a few complex PAH-containing mixtures. Component-based
approaches, involving an analysis of the toxicity of components of the mixture, are
recommended when appropriate toxicity data on a complex mixture of concern, or on a
sufficiently similar mixture, are unavailable. The RPF analysis under review is not a
reassessment of individual PAH carcinogenicity, but rather provides an approach for estimating
cancer risk for PAH mixtures by summing doses of component PAHs after scaling the doses
(with RPFs) relative to the potency of an index PAH (i.e., benzo[a]pyrene). The cancer risk is
then estimated using the dose-response curve for the index PAH.
Below is a set of charge questions that address general and scientific issues in the document.
Please provide detailed explanations for responses to the charge questions.
A-l
-------
General Charge Questions
la. Please comment on whether the report is logical, clear and concise. Please comment on
whether EPA has clearly synthesized the scientific evidence for the derivation of relative
potency factors for individual PAHs.
Ib. Please comment on whether the report provides adequate context for how the proposed RPF
approach could be used in a PAH mixtures risk assessment.
Chapter 2. Rationale for Recommending an RPF Approach
Chapter 2 presents the rationale for recommending an RPF approach. In an RPF approach, doses
of component chemicals that act in a lexicologically similar manner are added together, after
scaling the doses relative to the potency of an index chemical. Benzo[a]pyrene (B[a]P) is
selected as the index compound for this RPF approach. The RPF approach involves two key
assumptions related to the application of a dose-additivity model: (1) PAH components in the
mixture act in a similar toxicological manner; and (2) interactions among PAH mixture
components do not occur at low levels of exposure typically encountered in the environment.
2a. Please comment on whether the report provides adequate justification for using an RPF
approach as a scientifically defensible method to assess the cancer risk associated with
exposure to PAH mixtures.
2b Please comment on whether the choice of benzo[a]pyrene as the index compound is
scientifically justified and appropriately described. Please identify and provide the rationale
for any alternative index compound(s) that should be considered.
2c. Please comment on whether the weight of evidence indicating that PAHs, as a chemical class,
have a similar mode of carcinogenic action has been adequately described and is
scientifically justified.
2d. Please comment on whether the assumption that interactions among PAH mixture
components do not occur at low levels of exposure typically encountered in the environment
has been adequately described and is scientifically justified.
Chapter 3. Discussion of Previously Published RPF Approaches
This chapter presents a discussion of previously published RPF approaches. Due to the
evolution of the state of the science and an increased understanding of PAH toxicology, EPA is
reevaluating the RPF approach for PAHs in this analysis.
3. Please comment on whether the discussion provides a meaningful background on how RPFs
have been derived in the past, and the advantages and disadvantages of previous methods.
Chapter 4. Evaluation of the Carcinogenicity of Individual PAHs
This chapter discusses the development of a database of primary literature on PAH
carcinogenicity and cancer-related endpoints and the criteria used to include or exclude studies
from the database.
A-2
-------
4a. Please comment on whether the list of 74 PAHs (Table 2-1) included in the initial literature
search is complete. Please comment on whether the rationale for the choice of PAHs
included in the literature search has been appropriately described. Please identify other
databases or resources that should be included.
4b. Chapter 4 includes a description of how studies were selected for use in dose-response
assessment. Please comment on whether the choices and assumptions in making the selection
have been adequately described. Please comment on whether the information in Tables 4-1
through 4-14 provides adequate information to inform how decisions were made. Please
comment on whether studies were rejected or included appropriately. Please comment on
whether positive and nonpositive studies have been considered appropriately.
4c. The methodology for the choice of studies to use in the derivation of RPFs includes studies
where at least one PAH was tested at the same time as B[a]P. Studies where individual
PAHs were tested without concurrent testing of B[a]P were not included in the quantification
of RPFs. Please comment on the scientific rationale for this approach. Please comment on
whether the advantages and disadvantages of excluding certain data from the derivation of
RPFs have been adequately described.
Chapter 5: Methods for Dose Response Assessment and RPF Calculation
This chapter describes the selection of dose-response data and methods for dose-response
assessment and RPF calculation from the selected datasets. The methodology for estimation of
the RPFs varied depending on the characteristics of the datasets, however, the general equation
was the ratio of the slope of the dose-response curve for the subject PAH to the slope of the
dose-response curve for B[a]P.
5a. Please comment on whether the scientific rationale for the dose-response modeling
approaches used in the derivation of RPFs is adequately described. Please comment on
whether there are other appropriate modeling approaches for estimating the relative potencies
of PAHs. Please describe alternative approaches (e.g., other model forms) that could be
considered.
5b. For each individual dataset considered in the assessment, the B[a]P dose-response was
calculated from the study-specific data. Please comment on whether this approach has been
appropriately described. If there are additional approaches using the available data that
should be considered, please describe how the approach could lead to a better estimate of
cancer risk.
5c. The point of departure for slope estimation that has been used for the derivation of RPFs is
the benchmark dose (BMD) estimate rather than the lower confidence limit on the
benchmark dose (BMDL). Please comment on whether this approach is scientifically
justified and adequately described. Please comment on whether alternative approaches
should be considered.
5d. Please comment on the methodology used for the RPF calculations for multidose and single
dose datasets. Please comment on whether the process for calculating RPFs from the various
A-3
-------
datasets is scientifically justified and adequately described. Please comment on the
utilization of high response levels in some instances as the point of comparison. Please
describe alternative approaches that could lead to a better estimate of cancer risk that should
be considered using the available data. Please comment on whether the considerations for
RPF calculation as outlined in Sections 5.6 and 5.7 are scientifically justified and adequately
described.
Chapter 6: Selection of PAHs for Inclusion in the Relative Potency Approach
This chapter describes the selection of PAHs for inclusion in the RPF approach. The evaluation
focuses on whether the available data were adequate to assess the carcinogenic potential of each
compound. If the data were not considered adequate, then the PAH was excluded.
6a. Please comment on whether the rationale for the weight-of-evidence evaluation is
scientifically justified and adequately described. Please comment on whether the approach
adequately considers the available information. Please comment on whether other
information (e.g., additional structure-activity) could contribute further to the weight-of-
evidence evaluation and how this information could be utilized in the analysis.
6b. The weight-of-evidence analysis does not include data related to Ah-receptor binding,
cytotoxicity or tumor promotion. Please comment on whether the scientific rationale for this
decision is appropriate. If these data should be considered in the derivation of RPFs, please
describe how they should be incorporated into the analysis.
6c. The analysis uses an RPF detection limit as a means of comparing positive and nonpositive
(or negative) bioassays. Please comment on whether this method is scientifically justified
and adequately described.
6d. Graphic arrays of the calculated RPFs (Figures 6-2 through 6-35) are presented as a means of
representing the variability in RPFs from different data sources, the weight-of-evidence for
carcinogenic potential, and the basis for the selected RPF. Please comment on whether the
figures are informative and adequately described. Please comment on whether there is other
information that should be included in the figures. Please comment on whether the narratives
are informative and complete.
Chapter 7: Derivation of RPFs for Selected PAHs
This chapter describes various methods (e.g. prioritization of studies) and different approaches
for deriving final RPFs (e.g., arithmetic mean). Final RPFs were derived by averaging the
individual study RPFs (across all exposure routes) calculated from bioassay data for PAHs that
had at least one RPF based on a bioassay. The exception was dibenz[a,c]anthracene, where the
RPF was calculated from cancer-related endpoint data.
7a. Please comment on the scientific justification for the approach for deriving the final RPFs
and the discussion of alternative options for the estimation of the final RPFs. Please
comment on the reporting of the range of RPFs as a measure of variability instead of a
confidence interval. Please comment on whether the data are adequate to support more (or
less) precision in deriving the RPFs.
A-4
-------
7b. Please comment on whether the scientific rationale for consideration of bioassay data versus
cancer-related endpoint data has been adequately described. Please comment on whether the
cancer-related endpoint data could be used in a more quantitative manner. Please comment
on the justification of the final RPF derived for dibenz[a,c]anthracene. Please comment on
the use of tumor multiplicity data in the weight-of-evidence evaluations and for the
determination of the RPFs.
7c. Please comment on whether the recommendation to apply the proposed RPFs across all
routes of exposure is adequately described. Please comment on whether there is additional
scientific information that would inform this recommendation. Please comment on whether
the available data are adequate to recommend exposure route- or target organ-specific RPFs.
7d. Please comment on whether the scientific rationale for the assignment of an RPF of zero for
some PAHs is adequately described. Please comment on whether there are other data that
should be considered to assess whether an RPF of zero is appropriate. Please comment on
whether the scientific rationale for assigning no RPF based on inadequate data for some
PAHs is adequately described. Please comment on whether there are alternative methods for
assigning RPFs to these PAHs. Please comment on whether the text provides adequate
distinction between PAHs with RPFs of zero and PAHs with no selected RPF and whether
this distinction is useful for describing uncertainty in determining the cancer risk associated
with PAH exposure.
7e. The final RPFs are characterized with confidence ratings. Please comment on whether the
rationale for the confidence ratings is appropriately described. Please comment on whether
there are other approaches for describing confidence using the available data that could be
applied in either a qualitative or quantitative manner that would be more useful for risk
assessment.
Chapter 8. Uncertainties and Limitations Associated with the RPF Approach
This chapter discusses the uncertainties and limitations associated with using the RPF approach
for PAH mixtures risk assessment. Many of the general uncertainties related to chemical-
specific risk assessment are also applicable to the proposed RPF approach for PAHs. In addition,
uncertainties exist regarding the selection of data and dose-response assessment methodology,
the selection of PAHs for inclusion in the analysis, the derivation of the final RPF, the
assumption of a common mode of action and dose additivity, and the extrapolation of RPFs
across exposure routes.
8. Please comment on whether, overall, the document describes the uncertainties and limitations
in the methodology used to derive RPFs in a transparent manner. Please comment on
whether the most important uncertainties and limitations are identified. Please comment on
whether there is existing information that could be used to evaluate the accuracy or validity
of the RPF values to predict the cancer risk associated with exposure to PAH mixtures.
A-5
-------
Appendices
9. Please comment on whether the information in the Appendices is adequate to allow
independent verification of the calculated RPFs. If not, please comment on what additional
information would be useful.
A-6
------- |