United States       Science Advisory      EPA-SAB-CASAC-99-001 V
Environmental       Board (1400)          . October 19M
Protection Agency      Washington DC        www.epa.9ov/H6



   U.S. Environmental Protection Agency
   Region 5, Library (PL-12J)
   77 West Jackson Bputevard, 12th Floor
   Chicago, II 60604-3590

                                WASHINGTON, D.C. 20460
                                    October 7, 1998

EPA-S AB-C AS AC-99-001

Honorable Carol M.  Browner                                           OFFICE OF THE ADMINISTRATOR
Administrator                                                           SCIENCE ADV.SORY BOARD
U.S. Environmental Protection Agency
401 M Street SW
Washington, DC 20460

              Subject.  Review of the Diesel Health Assessment Document

Dear Ms. Browner:

       The Clean Air Scientific Advisory Committee (CAS AC) of EPA's Science Advisory
Board (SAB), supplemented by expert consultants (together referred to as the "Panel"), met on
May 5-6, 1998 to review the February 1998 draft document, "Health Assessment Document for
Diesel Emissions" (EPA/600/8-90/057C), in a public meeting in Research Triangle Park, NC.
An SAB Subcommittee conducted an initial review of the diesel topic in 1990.  Subsequently,
CAS AC reviewed the 1995 draft and found it wanting. Specifically, the Committee concluded
that the 1995 document was not scientifically adequate for making regulatory decisions
concerning the use of diesel-powered engines.  At the May 1998 meeting and in written
comments provided to EPA staff, the Panel assessed the adequacy of the present draft as an
accurate statement of current knowledge about the health effects of diesel exhaust inhaled in the
environment, and made numerous suggestions for improvement. The determination of the Panel
is summarized below. -The attached report describes the Panel's views in more detail, and
contains its responses to the four specific questions posed by EPA as a charge to the Panel.

       It was the unanimous view of the Panel that the February 1998 draft is not an acceptable
summary of current knowledge of the health effects of diesei exhaust inhaled in the environment,
and thus, does not serve as an acceptable basis for regulatory decision making based on adverse
health effects. The nature and magnitude of the draft's inadequacies precluded the choice of
closing on the document pending revision.

       Sections of the document, and especially the description of diesel engine emissions, are
considerably out of date.  The substantial differences between emissions from engines produced
since the early 1990s and those to which human and animal subjects comprising our present
health database were exposed was not portrayed. The document takes two approaches to using
rat lung tumor data to develop quantitative estimates of human lung cancer risk from low-level
environmental exposures. The majority view of the Panel was that neither  approach is supported
                                                                     Pnm*3 ««
rat lung tumor data to develop quantitative estimates of human lung cancer risk from low-level
environmental exposures. The majority view of the Panel was that neither approach is supported
by present knowledge of the nature and likely mechanisms of the rat response. The Panel noted
that the above two issues repeat the two major criticisms of the 1995 draft; indeed, there has
been no substantive updating of the emissions section since the 1990 draft.

       The document failed to link the potential health effects and likely risks from
environmental diesel soot to the effects and risks of airborne paniculate matter, which were
summarized and extensively reviewed and debated in conjunction with the recent review of the
paniculate matter standard.  Through this lack, the document fails to make a clear case for
treating diesel soot differently from the aggregate environmental paniculate matter to which it
contributes.  Epidemiological data from occupational exposures are considered by the Panel to
present the strongest current evidence for human cancer risk from inhaled diesel exhaust,
although considerable uncertainty remains regarding the most appropriate use of these data. The
present document falls short in its analysis of the exposure-dose-response relationships which are
crucial for extrapolating from occupational to environmental exposure levels of soot and its
potentially carcinogenic constituents.  The absence of a convincing portrayal of the quantitative
basis for extrapolation contributed to a division of opinion among the Panel as to whether a
quantitative, in contrast to a qualitative, assessment can be justified at this time.

       The Panel encourages the Agency to make a serious effort to develop a revised document
that constitutes an acceptable statement of current knowledge regarding the potential health risks
from environmental diesel exhaust. The Panel acknowledges that the task is difficult, but
believes that such a document is within the Agency's grasp if sufficient attention is given to the
above issues,  the numerous written comments from the Panel, and the discussion recorded in the
meeting transcript. The Agency is encouraged to engage CAS AC in a discussion of its proposed
strategy for remedying the document's deficiencies, prior to completing the next revision.  The
Panel looks forward to the opportunity to review and approve an appropriately revised
                                  Dr. Joe L. Mauderly, Chair
                                  Clean Air Scientific Advisory~Cotnrnittee

       This report has been written as a part of the activities of the Science Advisory Board, a
public advisory group providing extramural scientific information and advice to the
Administrator and other officials of the Environmental Protection Agency. The Board is
structured to provide a balanced, expert assessment of scientific matters related to problems
facing the Agency.  This report has not been reviewed for approval by the Agency; hence, the
comments of this report do not necessarily represent the views and policies of the Environmental
Protection Agency or of other Federal agencies. Any mention of trade names or commercial
products does not constitute endorsement or recommendation for use.

       The Clean Air Scientific Advisory Committee (CAS AC) of the EPA Science Advisory
Board (SAB) reviewed the Agency's Health Assessment Document for Diesel Emissions. While
acknowledging the difficulty of the task, the CAS AC encouraged the Agency to revise the
document, which the Committee judged to be not acceptable as a summary of the current
knowledge of the health effects of diesel exhaust inhaled in the environment.  Consequently, in
CASAC's view, it does not serve as an acceptable basis for regulatory decision making, based on
adverse health effects.  The Committee's main concerns are as follows: a) Some of the
information was judged to be considerably out of date. For example, the changes in diesel
engines and their emissions that have occurred in the 1990s is not reflected in the document; b)
Neither of the two approaches  employed by the  Agency to use animal data to generate estimates
of human risks associated with environmental exposure to diesel exhaust was found to be
supported by present knowledge; c) The document fails to distinguish the effects of diesel
exhaust, per se, from the effects of PM2 5 (particulate matter less than 2.5 microns in diameter),
of which it is a constituent; and d) The human epidemiological data from occupational exposures
present the strongest current evidence for human cancer risk from inhaled diesel exhaust.
However,  the Agency's document  does not effectively address ongoing debates about the
existing data. In the end the CASAC could not  reach a consensus on whether a quantitative,
rather than a qualitative, assessment can be scientifically justified at this time.  This marks the
second time that the CASAC has reviewed the Agency's health risk assessment of diesel exhaust.
In its 1995 review, the Committee identified a number of shortcomings,  some of which persist in
the current document.

Keywords:  Diesel Emissi. n, cancer risk, diesel exhaust, particulate matter

                       U.S. Environmental Protection Agency
                                Science Advisory Board
                            Clean Air Scientific Advisory
                     Committee (CASAC) Diesel Review Panel

Dr. Joe Mauderly, Lovelace Respiratory Research Institute, Albuquerque, NM

Mr. John Elston, Office of Air Quality Management, State of New Jersey, Department of Environmental
       Protection and Energy, Trenton, NJ

Dr. Philip K. Hopke, Clarkson University, Department of Chemistry, Potsdam, NY

Dr. Jay Jacobson, Boyce Thompson Institute, Ithaca. NY (did not participate in this review)

Dr. Arthur C. Upton, M.D.,Environmental and Occupational Health Sciences Institute, Piscataway, NJ

Dr. Sverre Vedal, M.D.. University of British Columbia, Vancouver Hospital, Vancouver, BC Canada

Dr. Warren White, Washington University, Chemistry Department, St. Louis, MO

Dr. David Diaz-Sanchez, Department of Medicine, UCLA, Los Angeles, CA

Dr. Eric Garshick, M.D., West Roxbury VA Medical Center, West Roxbury, MA

Dr. Roger O. McClellan, Chemical Industry Institute of Toxicology, Research Triangle Park, NC

Dr. Gunter Oberdfirster, University of Rochester Medical Center, Department of Environmental
       Medicine, Rochester, NY

Dr. William Pierson, Energy & Environmental Engineering Center, Desert Research Institute, Reno, NV

Dr. Leslie Stayner, NIOSH, Risk Evaluation Branch, Taft Laboratories, Cincinnati, OH

Dr. Ron Wyzga, Electric Power Research Institute, Palo Alto, CA

Science Advisory Board Staff
Mr. Robert Flaak, Designated Federal Officer (DFO) and Team Leader, Committee Operations Staff,
       Science Advisory Board (1400), US Environmental Protection Agency, Washington, DC 20460

Ms. Dorothy M. Clark, Management Assistant, Committee Operations Staff, Science Advisory Board
       (1400), US Environmental Protection Agency, Washington, DC 20460

                           TABLE OF CONTENTS

      2.1 Introduction  	4
      2.2 Charge 	5

      3.1 Response to the Charge	,	6
      3.2 Threshold vs. Non-threshold Approaches  	6
      3.2 Developing Estimates of Cancer Risk 	6
      3.3 Using an RfC for Diesel Exhaust Exposure	7
      3.4 Comments by Chapter	8
             3.4.1  Chapter 2 - Diesel Emissions	8
             3.4.2  Chapter 4 - Dosimetric Factors	8
             3.4.3  Chapter 5 - Noncancer Health Effects	9
             3.4.4  Chapter 6 - Derivation of RfC Non-cancer Health Effects	9
             3.4.5  Chapter 7 - Carcinogenicity in Laboratory Animals	10
             3.4.6  Chapter 8 - Epidemiological Studies of Cancer Risk	11
             3.4.7  Chapter 9 - Mutagenicity 	11
             3.4.8  Chapter 10 - Metabolism and Mechanism of Action  	12
             3.4.9  Chapter 11  - Qualitative  and Quantitative Evaluations of
                               Carcinogenicity  	13
             3.4.10 Chapter 12 - Health Risk Characterization	13


APPENDIX A - Detailed Written Comments of Individual Panel Members 	A -1

                           1. EXECUTIVE SUMMARY
       The Clean Air Scientific Advisory Committee (CASAC) of EPA's Science Advisory
Board, supplemented by expert consultants (together referred to as the "Panel"), met on May 5-6,
1998 to review the February 1998 draft document, Health Assessment Document for Diesel
Emissions (EPA/600/8-90/057C), in a public meeting in Research Triangle Park, NC.  This was
the third draft of the document; preceding drafts were reviewed in 1990 (by an SAB
Subcommittee) and 1995 (by an earlier CAS AC Panel). CAS AC found that the 1995 draft "was
not scientifically adequate for making regulatory decisions concerning the use of diesel-powered
engines". At the May 1998 meeting and in written comments, the Panel assessed the
acceptability of the present draft as an adequate statement of current knowledge about the health
effects of diesel exhaust inhaled in the environment, raising several key criticisms and making
numerous suggestions for improvement.

       It was the unanimous view of the Panel that the February 1998 draft is not an acceptable
summary of current knowledge of the health effects of diesel exhaust inhaled in the environment,
and thus, does not comprise an acceptable foundation for regulatory Decision making on the
basis of adverse health effects. The diverse nature and extensive magnitude of the draft's
inadequacies precluded the choice of closing on the document pending minor revision.

       The Panel found four key deficiencies in the draft.  First, sections of the document, and
especially the description of diesel engine emissions, were considerably out of date.  The
substantial differences between emissions from engines produced since the early 1990s and those
to which human and animal subjects comprising our present health database were exposed was
not portrayed.  This had been a major criticism of the 1995 draft, and apparently no serious
attempt has been made to  correct that deficiency.  Other areas needing updating included the
current understanding of likely mechanisms of lung carcinogenesis in rats, and the current status
of knowledge concerning the exposure-dose-response relationships among the epidemiological
data. The Agency has not addressed the much-debated extent to which the current human
epidemiologic database supports an exposure-response relationship between diesel exhaust and
lung cancer.

       Second, despite CAS AC advice to the contrary in 1995, the Agency continues to use rat
lung tumor data to develop quantitative estimates  of human lung cancer risk from low-level
environmental exposures.  The present draft included two approaches, and the majority view of
the Panel was that neither  approach was supported by present knowledge of the nature and likely
mechanisms of the rat response. Current knowledge comprises compelling evidence that the
species-specific, overload-related rat lung tumor response to extremely high level exposures is
not useful for estimating risk at environmental exposure levels, and is of doubtful relevance to


human risk from higher occupational exposures.  The Agency also developed a quantitative risk
estimate from potential rat tumor responses at lower exposure levels (still two orders of
magnitude above environmental levels) using the highest response that might not have been
detected because of the statistical power of the sizes of the individual treatment groups. The
Agency justified this approach on the presumption of effects of the soot-associated organic
compounds at low exposure levels. The Panel found no evidence supporting an effect of organic
mutagens in the rat response at either the low or the high levels, and considerable evidence to the

       Third, the  document failed to attempt any linkage between the potential health effects and
likely risks  from environmental diesel soot to the effects and risks of airborne ambient
paniculate  matter (PM). The effects of ambient PM were summarized and extensively reviewed
and debated in conjunction with the recent review of the paniculate matter standard.  An
important issue is  whether or not diesel  soot should be treated any differently than PM2 5, of
which it is a constituent. By failing to address this issue, the Agency did not make a case for
treating diesel soot differently from PM2 5 from a regulatory viewpoint.

       Fourth, epidemiological data from occupational exposures were agreed by the Panel to
present the strongest current evidence for human cancer risk from inhaled diesel exhaust,
although the Panel noted that considerable uncertainty remains regarding the most appropriate
interpretation and use of these data.  The present draft fell short in its discussion and analysis of
the exposure-dose-response relationships that are crucial for establishing a scientific basis for
extrapolating from occupational to environmental exposure levels of soot and its potentially
carcinogenic constituents.  The continuing, unresolved debate on this topic was hardly
mentioned. The Panel was disappointed that the Agency has not taken a lead role in resolving
this issue, or by suggesting additional research that is needed to resolve it. There was an
inadequate discussion of the amounts of mutagenic and carcinogenic exhaust constituents that
would actually deposit in the respiratory tract during lifetime exposures. In part as a result of the
lack of a convincing argument for a quantitative basis for extrapolation, the Panel remained
divided as to whether a quantitative, rather than a qualitative, assessment can be justified at this
time. No consensus was developed on this critical issue.

       Numerous other important issues and additional more-minor points were raised by  the
Panel, and  are contained in their individual written comments and the transcript of the meeting.
The  staff responsible for revising the document is strongly encouraged to review these sources  of
information in addition to the following summaries to gain  the most complete perspective
possible of the Panel's criticisms,  and to contact Panel Members individually for clarification, if
necessary.  Regardless of the approach taken in its revision of the document, staff must make
several key decisions in the face of continuing uncertainty.  The Panel strongly encourages staff

to engage CASAC in a consultation on the strategy it proposes for remedying the document's
deficiencies, prior to expending substantial effort in actually implementing the revisions.

       Although acknowledging that the task is difficult, the Panel encourages the Agency to
make a serious effort to develop a revised document that constitutes an acceptable statement of
current knowledge regarding the potential health risks from environmental diesel exhaust.

                      2. INTRODUCTION AND CHARGE
2.1 Introduction

       The Clean Air Scientific Advisory Committee (CASAC) convened a Diesel Review
Panel (Members plus expert Consultants) to conduct a review of the Agency's revised draft
Health Assessment Document for Diesel Engine Emissions prepared by the Agency's National
Center for Environmental Assessment (NCEA) - Washington, DC Office. The Committee met
May 5-6, 1998 in Research Triangle Park, NC.

       This effort follows an earlier review in 1995 when CASAC conducted a peer review of
the December 1994 version of the diesel assessment.  As a result of that review, the CASAC
recommendations focused on: a) the use of specific uncertainty factors in deriving the RfC
(reference concentration) value for protecting from adverse noncancer respiratory effects; b) the
minimal scientific support for using rat bioassay data for estimating human cancer risks; and c)
the outdated nature of information in several chapters. The Committee also made numerous
suggestions and recommendations for improving the draft document, asking to review the
revised document when it was ready. These recommendations are covered in detail in the
CASAC report of that review (CASAC, 1995).

       For the present review, NCEA provided CASAC with a listing that identifies the
disposition of the significant recommendations made by the Committee in 1995.  This was
provided to the Committee along with the 1998 version of the diesel assessment.  The CASAC
Diesel Review Panel that was created for this review included a number of Members and
Consultants who served  on the 1995  Panel as well as new panelists to ensure that the
composition of the review panel would be fresh and objective.  This is the standard practice of
the SAB and is consistent with the provisions of the Agency's 1994 Peer Review Policy and the
1998 Peer Review Handbook (EPA,  1998).  Panelists were asked to provided written comments
on the questions in the charge as well as specific chapters that they had been assigned for review.
These comments were submitted during the May 5-6, 1998 meeting and are part of the public
record.  The written comments, along with oral deliberations at the meeting, form the basis for
the recommendations contained in this report. For completeness, we have included the
individual comments of each panelist in Appendix A.  Although a number of the comments are
editorial, we believe that it is valuable to maintain a complete record of the peer review in this

2.2 Charge

       A baseline CASAC review objective is taken for granted by NCEA (i.e., the adequacy of
the assessment in identifying key hazard endpoints and characterizing the dose-response aspects
pertinent to public health exposure according to EPA's guidance on assessing cancer risk and
developing reference concentrations (RfC's).  NCEA also asked that CASAC focus on several
specific questions/issues.

       a)     For carcinogenic hazards and risk estimation purposes a key risk assessment
             choice is to decide whether the available evidence supports a nonthreshold hazard
             - low dose or threshold - higher dose hazard, or in the absence of definitive
             information whether rational inferences are more plausible one way or the other.
             Is NCEA's discussion of the topic and support for the position of an inferred
             nonthreshold - low dose hazard and risk, satisfactory?

       b)     NCEA discusses various approaches (and related uncertainties) in developing
             estimates of cancer risk.

                     1)      Does the equal mixing of approaches and the resulting risk values
                            define a plausible range of risk estimates or is there a scientific
                            case to be made that a subset of the estimates provides a more
                            defensible basis for establishing a risk range?

                    2)      Do you find that the documents's discussion, or other insights the
                            Committee might have, provides a basis for selecting a single or
                            scientifically "best" estimate of cancer risk?

       c)     EPA's approach to characterizing the noncancer health hazards is to develop an
             "RfC" for diesel exhaust exposure. Do you find that our identification of the
             critical effects/studies and the selection of the RfC uncertainty factors (  as
             allowed  in the RfC methodology)  is scientifically supportable and consistent with
             broader  considerations of particle  effects on humans?

                            3.  DETAILED FINDINGS
3.1 Response to the Charge

       On April 20, 1998, the EPA submitted a Charge to the Panel in the form of four questions
concerning its approach to characterizing the potential health risks of diesel exhaust. EPA staff
agreed at the close of the May 5-6, 1998 public meeting that the issues raised by the Charge had
been covered during the discussion; however, there was not a focused attempt to provide
consensus answers to the questions beyond the range of opinion expressed during the view of the
document. The Agency is referred to the summary comments in subsequent sections as the most
useful answers to the Charge.

3.2 Threshold vs. Non-threshold Approaches

The first element of the Charge asks

       For carcinogenic hazards and risk estimation purposes, a kt. risk assessment choice is
       to decide whether the available evidence supports a non-threshold hazard -  low dose or
       threshold - higher dose hazard, or in the absence of definitive information whether
       rational inferences are more plausible one way or the other. Is NCEA 's discussion of the
       topic and support for the position of an inferred non-threshold - low dose hazard and
       risk, satisfactory*?

       The Panel expressed concern that the discussion of threshold was not adequate.  The
Panel recognizes that there is no clear evidence for a threshold in the potential human lung
cancer risk from environmental diesel exhaust. However, some panelists noted that there was
not a sufficient scientific basis for assuming that lung  cancer risk had no threshold; both
regarding extrapolation from occupational exposure levels to the very low environmental
exposure levels, and regarding the plausible dose of mutagenic organic material  from
environmental exposures.  The discussion  of the issue needs strengthening.

3.2 Developing Estimates of Cancer Risk

       The second Charge element asks

       NCEA discusses various approaches (and related uncertainties) in developing estimates
       of cancer risk.

              a)     Does the equal mixing of approaches and the resulting risk values define a
                     plausible range of risk estimates, or is there a scientific case to be made
                     that a subset of the estimates provides a more defensible basis for
                     establishing a risk range ?

              b)     Do you find the document's discussion, or other insights the Committee
                     might have, provides a basis for selecting a single or scientifically "best"
                     estimate of cancer risk?

       The Panel did not consider the different methods for developing quantitative estimates of
cancer risk to be of equal value; thus, it was not comfortable with an "equal mixing" of
approaches.  For example, the Panel considered the estimates derived from rat data to be of
lesser value than those developed using other methods. Both general and specific comments
argued against portraying estimates derived by all approaches as a single range of estimates
having equal  validity.

       Although there was a range of opinion regarding the validity of deriving any form of
quantitative estimate of risk,  as contrasted to a qualitative statement of risk, the Panel expressed
a preference for using the epidemiological data if a quantitative estimate must be derived.

3.3 Using an RfC for Diesel Exhaust Exposure

       The third Charge element asks

              EPA's approach to ^characterizing the non-cancer health hazards is to develop an
              "RfC" for  diesel exhaust exposure. Do you find that our identification of the
              critical effects/studies and the selection of the RfC uncertainty factors (as allowed
              in the RfC methodology) is scientifically supportable and consistent with broader
              considerations of particle effects on humans. ?

       There was considerable discussion about the value of calculating an RfC and the various
uncertainty factors used in the document to derive the RfC.  Although no consensus developed
regarding the number and magnitude of the uncertainty factors, there was unanimous agreement
that the draft  document's  discussion of the uncertainty factors was inadequate. Because of the
lack of clarity about the basis and development of the uncertainty factors, it was not yet clear
whether or not the derivation is scientifically supportable. The Panel noted that in this section,
as throughout the document, there was an inadequate linkage of the information on diesel
exhaust to the information on ambient paniculate matter (PM) in general.  The lack of rationale
for an RfC lower than the 15 |ag/m3 annual PM2 5 standard was noted by the Panel.

3.4 Comments by Chapter

  3.4.1  Chapter 2 - Diesel Emissions

       The Panel did not agree with the Agency's decision not to expend the effort to update this
chapter on diesel emissions. The chapter must be updated in order for the document to be a
credible statement  of current knowledge. The fact that there are still 30-year old engines in use
does not justify this decision. There are three interrelated key reasons, as well as several more
minor ones, why this must be done: a) it is important to consider how changes in emissions
might influence the nature of their toxicity and their potency; b) it is important to portray the
differences between emissions from current production engines (i.e., the ones relevant to future
risk) and those from engines to which the humans and animals comprising the present health
database were exposed; and c) it is important, in th* final analysis, to make a clear statement
about whether or not the differences in emissions affect the value of the epidemiological data for
assessing present and future risk.

       This chapter should also include a discussion of the relevance of the exhaust dilution and
measurement conditions used in the laboratory,  and the resulting data, to the nature of exhaust
actually inhaled in the environment.  It should also include a summary of the diesel emissions
control strategy and schedule that were presented orally at the meeting, and a projection of
environmental exposure levels anticipated in view of the progressive controls.

  3.4.2  Chapter 4 - Dosimetric Factors

       This chapter fails to integrate dosimetric information into a coherent quantitative
exposition of the deposition and disposition of inhaled soot. A quantitative integration would
provide a much needed perspective on the actual amounts of soot and individual soot-borne
compounds and classes of compounds that constitute the "doses" to tissues and cells under
environmental exposure conditions.  This discussion is important to the consideration of the
plausibility of carcinogenesis from environmental exposures.  This chapter should include
linkage to the dosimetry portions of the recent PM Criteria Document (EPA, 1996). The Panel
did not see any  basis for taking a different approach to soot dosimetry than that taken for fine
PM. The discussion  should also include the more recent published models for diesel soot
dosimetry (e.g., Stober and McClellan, 1997).

       The large uncertainty that presently exists in models used to extrapolate dosimetry  from
animals to humans  is not adequately portrayed,  and the discussion of the "particle overload"
phenomenon, and its relevance to the high-dose rat diesel studies is inadequate.  Properly
reviewed, this information comprises a cogent argument against extrapolating high-dose rat lung
tumor response to  human cancer risk at environmental exposure levels.


       The draft is not clear as to why non-soot exhaust constituents, such as volatile and
semi-volatile organics and gases, are not considered in this chapter. Are they considered

  3.4.3 Chapter 5 - Noncancer Health Effects

       As in other chapters, the lack of linkage to the recent PM documents is a significant
deficiency. There is no discussion of the relationship between the potential health effects of
diesel soot and the effects thought to result from exposure to ambient fine PM.  The Panel views
these as interrelated, rather than separate, issues.

       The potential contribution of diesel exhaust to respiratory sensitization, amplification of
allergic responses, and asthma, is very uncertain.  While it is appropriate to mention this issue,
the present draft overstates the present certainty of the relationship. The fact that diesel
emissions have been falling while the incidences of asthma and rhinitis have been increasing is
largely ignored.  The chapter also gives a false impression that this is a recently emerging issue,
by failing to note much of the earlier literature on the topic, including literature on the potential
role of organic compounds.  In addition, the bases and justifications for  selecting the benchmark
concentration  and the interspecies uncertainty factor are not described clearly or argued

  3.4.4 Chapter 6 - Derivation of RfC Non-cancer Health Effects

       The rationale underlying selection of the  reference concentration (RfC) for diesel soot
was not presented clearly or argued convincingly. The basis  for selecting the benchmark effect
level, and why it differed among health endpoints was not clear. The basis for the premise that
humans are more sensitive than rats to non-cancer effects of diesel exhaust is unclear and
unconvincing.  It appears that the Agency changed its mind during the final stages of developing
the document  and gave different uncertainty factors in different chapters, demonstrating the
Agency's own  ambivalence on the issue and helping to fuel the Panel's skepticism.

       Even after extensive discussion at the meeting, the Panel remained somewhat uncertain
about the Agency's derivation of the RfC, and could not come to consensus regarding the most
appropriate RfC.  No clear guidance from the Panel emerged from the discussion. When polled,
three panelists recommended setting the RfC  at 15 (ig/m3, consistent with the annual standard for
PM2 5, three agreed that an RfC of 5 ^g/m* was probably acceptable, but could neither
understand in detail nor justify the method used to derive that value, two recommended giving a
range of RfCs, and the rest abstained.

       The Panel recommends that the Agency review its approach to this chapter and to
calculating the RfC, giving consideration to this report and the individual written and oral
comments of the Panel, and then discuss their proposed approach to the revision with CAS AC
prior to development of the next draft.

  3.4.5  Chapter 7 - Carcinogenicity in Laboratory Animals

       This chapter attempts to catalogue, but fails to integrate adequately, information from the
animal carcinogenicity studies. Most relevant studies are correctly cited, but the information
presented is inconsistent among the studies.  Some very relevant studies are not cited; for
example, neither the most extensive dose-response study of mice (Mauderly et a/., 1996) nor the
most extensive study of DNA adducts in rats (Randerath et a/., 1995) are described.

       There was an inadequate effort to place the exposure material used in the studies in
context.  For example, it is not emphasized that all of the animal studies were conducted using
old technology light- or medium-duty engines, and that no studies have been conducted using
exhaust from railroad or marine engines.  As another example, it is not noted that the titanium
dioxide used in some studies had an ultra-fine particle size, while that used in  other studies had a
much larger particle size.

       There is too strong an emphasis on reconciling the results among species.  The lung
tumor response clearly differs among the animal species tested to date and current evidence
suggests that it may differ between rats and humans. The attempt to synthesize the existing data
into hypotheses that unify the responses among species engendered unsupportable speculations.

       The statement that there are not adequate dose-response studies in mice is erroneous.
The Mauderly et al. (1996) dose-response study of mice was done in parallel to the study which
is cited as one of the most reliable sources of dose-response data from rats, but the negative
mouse study is not cited in the chapter.

       This chapter does  not contain an adequate analysis of the lung tumor data from rats
exposed at the lower levels (still very high compared to environmental levels), nor does it
contain an adequate discussion of the evidence concerning the effect of soot-associated organic
compounds in rats at either high or low levels. These deficiencies lay the foundation for the
questionable risk estimates that appear later in the document.  The Panel viewed the premises
that: a) a small tumor response at low exposure was overlooked due to statistical power; and b)
soot-associated organic mutagens had a greater effect at low than at high exposure levels to be
without foundation. In the absence of supporting evidence, the Panel did not view derivation of
a quantitative estimate of human lung cancer risk from the low-level rat data as appropriate.  The
Panel noted that the aggregate data from several studies provide a useful test  of carcinogenesis at


exposure levels two orders of magnitude above ambient, but give no suggestion of even an
insignificant effect.  The Panel also noted that there is no evidence that the organic fraction of
soot played a role in rat tumorigenesis at any exposure level, and considerable evidence that it
did not.  However, the Panel also noted that the lack of organic effect in rats cannot be taken as
proof that the organic fraction is not relevant to human risk.

  3.4.6 Chapter 8 - Epidemiological  Studies of Cancer Risk

       The majority of the Panel were in general agreement with the final conclusion that there
is limited evidence for a causal association between occupational exposure to diesel exhaust and
lung cancer. The Panel was less supportive of either the utility of, or the basis for, the Agency's
assertion that diesel exhaust was "close to being a known human carcinogen" within the present
risk assessment framework.

       The basis for selecting studies for presentation was not stated clearly.  Several
suggestions were made in the individual comments for presenting the studies in a clearer, more
consistent manner, and some inaccuracies were noted in both the descriptions and the
quantitative data presented. The discussion of the strengths and weaknesses of the individual
studies should be strengthened, including the most likely duration of exposure in the different
studies, the related issue of latency, and the likely importance of confounding in each study.
Overall, the information in this chapter  should be integrated in a more  analytical manner.

       This chapter does not  contain an adequate discussion of the evidence for
exposure-dose-response relationships between inhalation of diesel exhaust and lung cancer.
Confidence that such a relationship exists is requisite for confidence in any extrapolation of
cancer risk from occupational to environmental exposure levels.  Much of the debate concerning
the epidemiological data and their appropriate use has centered on this issue during recent years.
Different investigators have analyzed the same data set and reached very different conclusions
regarding the dose response for cancer  risks from diesel exposure. The Panel found  it
disappointing that the Agency had not taken a lead role in resolving this crucial issue, and
unacceptable that the chapter  does not deal  with this issue at all. The Panel recognizes that the
issue may not be clearly resolvable at this time, but notes that regardless, our confidence in the
quantitative risk assessment is directly proportional to the quality of our understanding of this

  3.4.7 Chapter 9 - Mutagenicity

       The information on mutagenicity from organic compounds is presented well overall,  and
the chapter could be acceptable with attention to the following two issues.

       a)     The chapter needs to include a discussion of the current information from
              laboratory studies of mutagenicity from particles with high doses of poorly
              soluble particles of low cytotoxicity without organic mutagens.  The alternate
              mutagenic pathways, such mutagenicity from oxygen radicals, which are now
              thought to contribute to the lung tumor response of rats to chronic, heavy
              exposures to particles should be discussed. This discussion will help place the rat
              results in their appropriate context.

       b)     The issue of dose is not discussed adequately. The doses of mutagenic material
              applied to bacteria and mammalian cells in the laboratory  must be placed in
              context regarding the deposited doses that might plausibly result from human
              exposure to diesel soot in the environment.

  3.4.8  Chapter 10 - Metabolism and Mechanism of Action

       This chapter fails to pull the relevant information together into a cogent synthesis.  The
effort suffers from an apparent desire to reconcile results from animals and humans into a single,
unified mechanistic framework. The existing evidence does not provide for such a
reconciliation,  and strongly suggests that if carcinogenesis occurs in humans, it occurs by-
mechanisms different from those responsible for the rat response.

       It is considered most plausible that any human cancer risk from inhaled diesel exhaust
would result from the mutagenicity of organic compounds absorbed in the respiratory tract. On
this presumption, the issue of dosimetry of the organic compounds is crucial. The chapter does
not give an adequate discussion of the actual doses of organic material likely to be absorbed
from environmental exposures.

       Present evidence does  not support a role of organic mutagens in the lung tumor response
of rats.  Lung tumor and DNA adduct data from studies of rats exposed to diesel exhaust and
other particles presents compelling evidence that the organics play no significant role at high
exposure levels.  There is no evidence for in vivo mutagenicity or DNA adduct formation in rats
at non-overloading exposure levels. Present evidence suggests that carcinogenesis in rats is
related to the inflammatory response and is likely mediated by oxidant injury.  Discussion of this
mechanistic pathway needs to  be added to the chapter.

       It is stated in this chapter and elsewhere in the document that exposure to diesel exhaust
early in life, and especially from conception, is likely to render individuals more susceptible to
exhaust.  If no evidence can be cited nor a plausible mechanism given to  support this assertion, it
should be deleted.

  3.4.9  Chapter 11 - Qualitative and Quantitative Evaluations of Carcinogenicity

       There was a considerable range of opinion among the Panel regarding the derivation of
quantitative estimates of human lung cancer risk from environmental exposures to diesel
exhaust. That range of opinion is summarized in Section 4 (Conclusions) below. Staff is
encouraged to read the individual written comments and the meeting transcript thoroughly to
assess the many issues that were raised and suggestions for improvement.

       Opinion was divided as to whether a quantitative  risk assessment (derivation of a unit
risk value) is justified at this time, or whether a qualitative statement is a more appropriate
reflection of the current evidence for a likely carcinogenic effect at environmental exposure
levels. In considering this issue, the Panel noted that the  document does not describe the
Agency's need for a quantitative risk assessment, or its intended use of a unit risk value unique to
diesel exhaust paniculate, and recommended that this information be added.

       Consonant with the advice given to the Agency in 1995 (CASAC, 1995), the majority of
panelists felt that the animal data should not be used to derive a qualitative risk estimate for
environmental exposures. Several panelists felt that the laboratory results were useful for
characterizing the carcinogenic hazard, and some felt that the animal data might be used in some
manner to help frame cancer risk. Several other panelists noted that the issue of threshold had
not been adequately discussed, and did not agree that present information supports a
non-threshold linear extrapolation from occupational to environmental exposure levels.

       The derivation and interpretation of the upper and lower bounds of risk need to be
presented more clearly. There was considerable uncertainty  among Panel members regarding the
definitions of the bounds as presented in the chapter. Some preference was expressed for the use
of maximum likelihood estimates rather than  95% confidence intervals for expressing the bounds
of risk.  It was noted that even if estimates from animal data were to be retained, it was not
appropriate to combine estimates derived from human and animal data into a single range of risk.

  3.4.10 Chapter 12 - Health Risk Characterization

       The document states that this chapter is intended as a"lay" summary of the foregoing
information and a summary synthesis of the health risks from diesel exhaust inhaled in the
environment.  The chapter falls short of accomplishing the former purpose. In several instances,
the  "simplified" language remains unnecessarily complex,  and in others, it is misleading. The
individual written comments should be reviewed for editorial suggestions. Figure 12 is too
complex, especially considering the intended  "lay" audience for this chapter.  It was confusing to
several panelists, and would not be useful for most lay readers.

       The chapter does not give a straightforward, accurate view of the present large
uncertainty regarding the cancer risk from environmental exposures. Characterization of
environmental diesel exhaust as a "major" environmental hazard was considered by many
panelists to be an overstatement.  Because the chapter summarized the foregoing material, many
of the criticisms of the preceding chapters were repeated for Chapter 12. Among the repeated
issues were the failure to discuss technology-related changes in exhaust, the failure to tie
diesel-health issues to PM-health issues, the inappropriateness of deriving risk estimates from
the rat data, overstatement of the likely role of diesel exhaust in allergic disease, differences of
opinion regarding the RfC, and lack of support for the assertion that exposures early in life
render individuals more susceptible.

                                 4.  CONCLUSIONS
       It was the unanimous view of the Panel that the February 1998 draft is not an acceptable
summary of the current knowledge of the health effects of diesel exhaust inhaled in the
environment, and thus, does not constitute an adequate basis for regulatory decision making
based on adverse health effects. The nature and extent of the revisions needed are such that the
Panel could not close on (approve) the document pending minor changes.  It was agreed that a
revised document must be re-reviewed by CASAC.

       It was the consensus view of the Panel that the document must be revised to include an
updated description of diesel engine emissions and  the potential implications of changes in
emissions for health risk. It was also the consensus view of the Panel that the document should
link the discussion of the health risks from diesel exhaust to the health risks from PM,
referencing the recent PM Criteria Document in the discussion of several issues. Finally, it was
also the consensus view of the Panel that developing an acceptable document is a task within the
reach of the Agency, but can only be accomplished if given more attention and resources than
were evident from the advances made since the 1995 draft.

       The range of opinions among the Panel on other major issues defy summarization as
consensus views. Clearly, the Agency faces several difficult choices in revising the document.
Although the Panel could not make consensus recommendations on several important points, two
general recommendations are readily extracted from the discussion:

       a)     the Agency's choices regarding the portrayal and estimation of risk must be
             supported b . scientific rationale that is clearly stated and must be defended oh the
             basis of existing knowledge; and

      b)     the Agency would be well served by discussing their proposed approach to key
             issues with CASAC before completion of the next draft. For example, key issues
             might include the approach to discussing changes in exhaust, linkage to PM
             health risks and standards, derivation of the RfC, dose-response among the
             epidemiological data, approach to developing quantitative estimates of cancer risk
             (if done), and portrayal of the range of likely risk. The Panel recognized the
             Agency's desire for clear guidance on these issues, but developing consensus
             guidance was not possible within the framework of the document review and I'/z
             day meeting.

      The range of opinion on certain issues was solicited by polling the 13  panelists at the end
of the meeting.  The following results might provide a useful perspective and illustrate the


uncertainty that exists in certain areas. The results should not be taken out of context as a

       a)      Recommend including some form of quantitative estimate of cancer risk?
                    Yes = 8, No = 3, Abstain = 2

       b)      Recommend using some form of animal data in estimating risk?
                    Yes = 5, No = 8

       c)      Continue to include an estimate based on benzo(a)pyrene?
                    Yes = 4, No = 4, Abstain = 5

       d)      Favor inclusion of comparative potency approach in general?
                    Yes = 7, No = 6

       e)      Develop quantitative estimate of risk from existing epidemiological data?
                    Yes = 8, No = 3, Abstain = 2

       Note: This list includes references suggested by the Panel as well as any references that
             are cited in the body of the CAS AC report.

Baumgard, K.J., and J.H. Johnson. 1996.  The Effect of Fuel and Engine Design on Diesel
       Exhaust Particle Size Distributions. Warrendale, PA: Society of Automotive Engineers;
       SAE technical paper no. 960131. Reprinted from Diesel Exhaust Aftertreatment 1996

Bagley, S.T, Baumgard, K.J., Gratz, L.D.,  Johnson, J.H., and D.G. Leddy.  1996.
       Characterization of Fuel and Aftertreatment Device Effects  on Diesel Emissions. Health
       Effects Institute Research Report Number 76.

Begeman, C.R.  1962.  Carcinogenic Aromatic Hydrocarbons in Automotive Effluents.  Paper
       No. 440C presented January 1962 at the SAE Automotive Engineering Congress.  SAE
       Technical Progress Series Vol. 6, "Vehicle Emissions", Society of Automotive Engineers
       Inc., New York.

Bj0rseth, A., and A.J. Dennis.  1980. The problem of PAH degradation during filter collection
       of airborne particulates —  An evaluation of several commonly used filter media.
       Polynuclear Aromatic Hydrocarbons: Fourth International Symposium on Analysis,
       Chemistry and Biology, Battelle Press, Columbus, OH.

Boorman, G.A., Brockman, M., Carlton, W.W., Davis, J.M.G., Dungworth, D.L., Hann, F.F.,
       Mohr, U., Reichhelm, H.R., Turusov, V.S., and S. Wagner.  1996. Classification of
       cystic keratinizing squamous lesions of the rat lung: Report  of a workshop.  Tox. Path.

CASAC.  1995. Review of the Diesel Health Assessment, EPA-SAB-CASAC-LTR-95-003,
       Clean Air Scieientific Advisory Committee, Science Advisory Board, US Environmental
       Protection Agency, Washington, DC.

Cass, G.R., and H.A. Gray.  1995. Regional emissions and atmospheric concentrations of diesel
       engine particulate matter: Los Angeles as a case study. In Diesel Exhaust: A Critical
       Analysis of Emissions, Exposure, and Health Effects. Health Effects Institute,
       Cambridge, MA p. 126-137.
                                        R- 1

Cuddihy, R.G., Seller, F A., Griffith, W.C., Scott, B.R. andR.O. McClellan.  1980. Potential
       health and environmental effects of diesel light duty vehicles.  DOE Research and
       Development Report, LMF-82. Springfield, VA: National Technical Information Service.

Davies, C.N.  1980.  An algebraical model for the deposition of aerosols in the human
       respiratory tract during steady breathing-Addendum. J. Aerosol Sci. 11:213-224.

Dolan, D.F., Kittelson, D.B., and D.Y.H. Pui. 1980. Diesel Exhaust Particle Size Distribution
       Measurement Techniques.  Society of Automotive Engineers; SAE technical paper no.
       800187, Warrendale, PA.

Driscoll, K.D., Carter, J.M., Howard, B.W., Hassenbein, D.G., Pepelko, W., Baggs, R.B., and G.
       Oberdorster.  1966. Pulmonary inflammatory chemokine, and mutagenic responses in
       rats after subchronic inhalation of carbon black.  Toxicology and Applied Pharmacology,

EPA. 1996. Air Quality Criteria for Paniculate Matter. Office of Health and Environmental
       Assessment.  Office of Research and Development, US Environmental Protection
       Agency, Research Triangle Park, NC.

EPA. 1998. Peer Review Handbook.  EPA 100-B-98-001, Science-Policy Council, Office of
       Research and Development, US Environmental Protection Agency, Washington, DC,
       January 1998

Gertler, A.W., Sagebiel, J.C., Dippel, W.A., and R.J. Farina.  1998.  Measurements of dioxin and
       furan emission fact -rs from heavy-duty diesel vehicles. J. Air & Waste Manage. Assoc.

Hampton, C.V., Pierson, W.R., Schuetzle, D., and T.M. Harvey.  1983. Hydrocarbon gases
       emitted from vehicles on the road.  2. Determination of emission rates from diesel and
       spark-ignition vehicles. Environmental Science and Technology, 17(12V. 699-708.

ICRP.  1994. Human Respiratory Tract Model for Radiological Protection, International
       Commission on Radiological Protection (ICRP66), ICRP Publication 66, Annals of the
       ICRP 24(1/3) 482 pp.

Kittelson, D.B., Kadue, P.A.,  Scherrer, H.C., and R.E. Lovrien. 1988. Characterization of diesel
       exhaust particles in the atmosphere. Final report to Coordinating Research Council AP-2
       Project Group, March 1988.

Maejima, K., Tamura, K., Taniguchi, Y., Nagase, S., and H. Tanaka. 1997. Comparison of the
       effects of various fine particles on IgE antibody production in mice inhaling Japanese
       cedar pollen allergens. J. Toxicol Environ Health, 52:231-248.

Mauderly, J. L., Banas, D.A., Griffith, W.C., Hahn, F.F., Henderson R.F., and R. O. McClellan.
       1996. Diesel  exhaust is not a pulmonary carcinogen in CD-I mice exposed under
       conditions carcinogenic to F344 rats. Fundam. Appl. Toxicol. 30:233-242.

McClellan, R.O.  1996. Lung cancer in rats from prolonged exposure to high concentrations of
       carbonaceous particles.  Implications for human risk assessment. Inhalation Toxicology
       8(SupplementV 193-226.

McClellan, R.O.  1997. Use of mechanistic data in assessing human risks from exposure to
       particles.  Environmental Health Perspectives, IQSCSupplement 5): 1363-1372.

Miguel, A.H., Kirchstetter, T.W., Harley, H.A., and S.V. Hering. 1998. On-road emissions of
       paniculate polycyclic aromatic hydrocarbons and black carbon from gasoline and diesel
       vehicles. Environmental Science and Technology, 32£4):450-455.

Moore, G.E., and M.  Katz. 1960. Polynuclear Aromatic Hydrocarbons in the Particulates of
       Diesel Exhaust in Railway Tunnels and in the Particulates of an Urban Atmosphere.  Int.
       J. Air Poll.2,  221-235.

NFRAQS.  1998.  Northern Front Range Air Quality Study. EM, January 1998, pg 13-19. (See
       http://charon. CIRA. colostate.edu).

Oberdorster, G., Ferin, J., Baggs, R., Pinkerton, K. and Morrow, P.E. 1997. Alveolar
       macrophage cluster formation: A clearance mechanism for large particles in mouse
       lungs. In:  Inhaled Partices VII, ed. Cherry, N. and Odgen, T., Pergamon Press, p.

Okamoto, W.K., Gorse, R.A., and W.R. Pierson.  1983. Nitric acid in diesel exhaust.  J Air Poll
       Control Assoc., 33(11):1098-1100.

PflugerD.H., and C.E. Minder. 1994. A mortality study of lung cancer among Swiss
       professional drivers: accounting for the smoking related fraction by a multivariate
       approach. Soz Praventivmed; 39:372-378.

Pierson, W.R., Gorse, R.A., Szkarlat, A.C., Brachaczek, W.W., Japar, S.M., Lee, F.S.-C.,
       Zweidinger, R.B., and L.D. Claxton.  1983.  Mutagenicity and chemical characteristics of
       carbonaceous paniculate matter from vehicles on the road.  Environmental Science and
       Technology, 17C1V31-44.

Pierson, W.R., Gertler, A.W., Robinson, N.F., Sagebiel, J.C., Zielinska, B., Bishop, G.A.,
       Stedman, D.H., Zweidinger, R.B., and W.D. Ray.  1996.  Real-world automotive
       emissions ~ summary of studies in the Fort McHenry and Tuscarora Mountain tunnels.
       Atmospheric Environment, 30U2}:2233-2256.

Rafhsson V., and H. Gunnarsdottir. 1991. Mortality among professional drivers. Scan J Work
       Environ Health;  17:312-317.

Randerath, K., Putman,  K.L., Mauderly, J.L., Williams, P.L., and E. Randerath.  1995.
       Pulmonary toxicity of inhaled diesel exhaust  and carbon black in chronically exposed
       rats.  Research Report Number 68, Health Effects Research Institute Research Report,
       Cambridge, MA, December, 1995.

Sagebiel, J.C., Zielinska, B., Pierson, W.R., and A. W. Gertler. 1966. Real-world emissions and
       calculated reactivities of organic species from motor vehicles. Atmospheric
       Environment, 30021:2287-2296.

Salmeen, I.T., Gorse, R.A., and W.R. Pierson.  1985.  Ames assay chromatograms of extracts of
       diesle exhaust particles from heay-duty trucks on the road and from passenger cars on a
       dynamometer. Environmental Science and Technology, 19(31:270-273.

Salvi, S.S., Blomberg, A., Rudell, B., Kelly F.J., Frew A.J., Hlgate S.T., and T. Sandstrom.
       1996. Acute inflammatory response in the airways and peripheral blood after exposure to
       diesel exhaust in health human volunteers. Europ Resp J; 9(S.upp 23).415S.

Scheuch, G.  1991.  Die Dispersion, Deposition, und Clearance von Aerosolpartikeln in den
       menschlkichen Atemwegen (Ph.D. thesis), J.W. Goethe-Universitat. Frankfurt am Main,

Scheuch, G., Kreyling, W., Haas, F. and W. Stahlhofen. 1993. Effect of settling velocity on
       particle recovery from human conducting airways  after breath holding. J. Aerosol Med.

Stahlhofen, W.  1989.  Human lung clearance following bolus inhalation of radioaerosols. In:
       Extrapolation ofDosimetric Relationships for Inhaled Particles and Gases, pp 153-166,
       Academic Press, Washington, DC.

Stahlhofen, W., Gebhart, J. and J. Heyder.  1980. Experimental determination of the regional
       deposition of aerosol particles in the human respiratory tract. Am. Ind. Hyg. Assoc. J.

Stahlhofen, W., Gebhart, J., Rudolf, G. and G. Scheuch.  1986.  Measurement of lung clearance
       with pulses of radioactively-labeled aerosols. J. Aerosol Sci. 17:333-336

Stahlhofen, W., Gebhart, J., Rudolf, G., Scheuch, G. and K. Philipson. 1986.  Clearance from
       the human airways of particles of different sizes deposited from inhaled aerosol boli. In:
       Aerosols: Formation and Reactivity, Second International Aerosol Conference, West
       Berlin, Germany, September 22-26, 1986, pp. 192-196, Pergamon Press, Oxford, U.K.
Stahlhofen, W., Gebhart, J., Rudolf, G. and G. Scheuch.  1987. Retention of radiolabeled
       particles in human lungs. In: Deposition and Clearance of Aerosols in the Human
       Respiratory Tract, Second International Symposium, Salzburg, Austria, September 1 8-
       20,  1986, pp 123-128, (ed. Hofrnann, W.) Facultas Universitatsverlag Ges.m.b.H. Vienna

Stahlhofen, W., Gebhart, J., Rudolf, G., Scheuch, G. and M.R. Bailey.  1986. Human lung
       clearance of inhaled radioactively labelled particles in horizontal and vertical position of
     -  the inhaling person, J. Aerosol Sci.  18:741-744.

Stahlhofen, W., Koebrich, R., Rudolf, G. and G. Scheuch. 1990.  Short-term and long-term
       clearance of particles from the upper human respiratory tract as function of particle size.
       J. Aerosol Sci. 21(Suppl. 1):S407-S410.

Stahlhofen, W., Scheuch G.  and M.R. Bailey.  1994.  Measurement of the tracheobronchial
       clearance of particles after aerosol bolus inhalation. In: Inhaled Particles  VII,
       Proceedings of an International Symposium on Inhaled Particles Organized by the
       British Occupational Hygiene Society, 16-22 September 1991 (Eds. Dodgson, J. and
       McCallum, R.I.) Ann. Occup. Hyg. 189.

Stober, W. and R.O. McClellan. 1997. Pulmonary retention and  clearance of inhaled
       biopersistent aerosol particles: Data-reducing interpolation models and models of
       physiologically based systems. Critical Reviews in Toxicology, 27£6}. 539-598.

Swift, D.L., Montassier, N., Hopke, P.K., Karpen-Hayes, K., Cheng, Y-S., Su, Y.F., Yeh, H.C.
       and J.C. Strong.  1992. Inspiratory Deposition of Ultrafine Particles in Human Nasal
       Replicate Casts, J. Aerosol Sci. 21:65-72.

Tebbens, B.D., Thomas, J.F., and M. Mukai. 1963. Paniculate Air Pollutants Resulting from
       Combustion.  From ASTM Special Publication No. 352, Symposium on Air-Pollution
       Measurement Methods, presented at the Fourth Pacific Area National ASTM Meeting,
       Los Angeles, Oct. 5, 1962 (published 1963 by  ASTM).

Truex, T.J., Pierson, W.R., and D.E. McKee.  1980.  Sulfate in diesel exhaust. Environmental
       Science and Technology, 14(9): 1118-1121.

Xu, G.B. and C.P. Yu. 1987.  Deposition of diesel exhaust particles in mammalian lungs. A
       comparison between rodents and man. Aerosol Science and Technology, 7:117-123.

Zaebst, D.D., Clapp D E., Blade L.M., Marlow D.A.,  Steenland K., Hornung R.W., Scheutzle D.,
       and J. Butler.  1991. Quantitative determination of trucking industry workers' exposures
       to diesel exhaust. Am Ind Hyg Assoc; 52:529-41.

Zielinska, B., Sagebiel, J.C., Harshfield, G.,  Gertler, A.W.r and W.R. Pierson.  1996. Volatile
       organic compounds up to C20 emitted from motor vehicles; Measurement methods.
       Atmospheric Environment, 3002):2269-2286.

                                  APPENDIX A
           Detailed Written Comments of Individual Panel Members
      The following are the original, unedited written comments provided by individual
Panelists prior to or at the May 5-6, 1998 meeting. They do not reflect consensus of the Panel
and, in some cases, may have been revised subsequent to the meeting as a result of discussion.
They were provided to the Aqency following the meeting so that Agency staff would have
detailed editorial comments as well as individual responses to the Charge.  The material in this
Appendix, along with the discussions at the May 5-6 meeting form the basis for this written
report. (Note: these comments may contain unconnected typographical errors that result from
electronic translation).

             Panelist                                             Page
      Dr. Joe Mauderly                                           A - 2
      Dr. Philip Hopke                                            A - 24
      Dr. Arthur Upton                                           A - 29
      Dr. Sverre Vedal                                            A - 30
      Dr. Warren White                               '           A - 36
      Dr. David Diaz-Sanchez                                      A - 41
      Dr. Eric Garshick                                           A - 46
      Dr. Roger McClellan                                         A - 56
      Dr. Gunter Oberdorster                                      A - 75
      Dr. William Pierson                                          A - 81
      Dr. Leslie Stayner                                           A - 98
      Dr. RonWyzga                                             A- 108


Overall, this document, "Health Assessment Document for Diesel Emissions", is not
suitable for closure as presented. It requires substantial revision to constitute an
acceptable statement of current knowledge about the potential health risks from
airborne diesel engine emissions. The draft presented for review does not suggest that
the Agency elected to put forth the effort necessary to develop an acceptable
document. The panel's discussion with the Agency at the 5/5-6/98 meeting suggested
that perhaps the effort expended reflects the Agency's own ambivalence about the
purpose of, and thus the need for, the document. Progressive controls on diesel
emissions are already in place, and will extend for several years into the future. It is
not clear how directly this document would influence the Agency's decisions regarding
diesel emissions.

The quality of the document received by CASAC is disappointing, particularly in view of
the comments provided by CASAC in 1995.  At that time, the principal criticisms were:
1) that several sections of the document were badly out of date; and 2) that it was not
appropriate to develop quantitative estimates of human lung cancer risk from
environmental exposures from the rat lung tumor response to high-level exposures.
The approach taken  by the Agency in developing the present draft conflicted directly
with that advice. Although there are other problems with the document, those choices
alone acted to ensure that the revision could  not be approved by CASAC.


P 1-1, last para: The paragraph beginning on line 31 doesn't make sense as written.

P 1-2, L 6-11: First,  EPA has prepared two previous drafts of this document, so stating
that it has never prepared a comprehensive health assessment document on diesel
exhaust is misleading.  This is certainly not a new effort, as the statement implies.
Second, the abbreviation, "DE" has not yet been defined.

      This chapter presents a problem and forces a decision point. EPA has obviously
decided that it is not necessary to update this information. While that position might be
argued, no argument is presented, other than to say that there are still engines on the
road built in the 70s  and 80s. The material presented is dated to pre-1989, which is
about 10 years ago.  Engines and their emissions have evolved substantially during

that period.  While it is true that there are old engines in use, the chapter certainly
doesn't give a current perspective on the composition of diesel exhaust. It is not
apparent that anyone was tasked with an update at all.

P 2-1, L 4-5:  It is an understatement that the chapter does not reflect the most current
literature. There are  few references younger than 10 years old!  No rationale is given
for this choice. If the information  is  not worth updating, why is it worth including at all?

P 2-2, para 2: Why switch to using  the terminology, diesel "oil" for fuel? That term is
not used consistently.

P 2-2, L 12-31: Why include this  information - is it relevant?

P 2-2, L 33-34:  Presumably, diesel emissions also include products from whatever
happened to be in the intake air as well.

P 2-2, L 40:  The "LDD" and "HDD"  abbreviations don't match those used on the
preceding page.

P 2-4, L 10-22: These sections are repeats,

P 2-5, L 25-27: These CO values don't make sense as tailpipe emissions.  Where do
these concentrations exist?

P 2-5, L 31:  Are  the C-10 to C 25 MW  molecules really "unburned diesel fuel"?

P 2-6, L 11-18: Carbon is still carbon, even if it is combined in different molecules.
Carbon doesn't disappear during  combustion. Presumably, the author is talking about
elemental carbon.

P 2-7, L 15:  The value of 0.5 m3/g surface area  seems small by a factor of 10.  Is this a

P 2-7, L 27:  What does "charged" mean here?

P 2-8, L 1:  Give  a reference for this statement.

P 2-8, L 14:  First, state that the "soluble" extract is soluble in organic solvent. Second,
why not call "diesel-generated particulate material" soot?  Is it something besides soot?

P 2-10, L 6-18: First, the CO comparison must be with uncatalyzed engines. Do
diesels produce lower CO than gasoline engines with catalysts?  Second, what does

"average traffic" mean? Third, you are switching abbreviations again with HDD and

P 2-11:  Why would we be interested in data this old? Have gasoline engine emissions
not evolved since 1982?

P 2-12, second para: Are there no data for the U.S.? Why are we talking about
gasoline engines in the U.K. in the 1980s?

P 2-12, L 30-33:  Why use data from a foreign truck? What country did this come
from? Are there no data for U.S. engines?

P 2-13, L 3: Are the evaporative and refueling emissions increasing as the statement

P 2-14, Figure 2-1:  This figure is 16 years old. What do contemporary data look like?

P 2-27, last para:  I have heard Glen Cass talk during the past year about diesei-
specific tracers. This paragraph suggests that there aren't any.

P 2-41, L 4: Not updating the chapter is highly questionable. Calling a 1988 reference
"recent" is ridiculous! This suggests that the chapter hasn't even been read recently by
the author.

P 2-43, L 7: I don't think "FTP" has been defined yet.

P 2-43, L 27: This statement indicates that ozone gets up to 1.5 ppm these days.  Is
that true?

P 2-47, section 2.5.1:  I don't understand the value of this information if the traffic was
only 5% diesel vehicles.

'P 2-49, L 1: Why should we care about any data for which the portion of diesel
vehicles  is unknown?

P 2-56, first 2 paras: This has already been said - why repeat it here?

P 2-56, L 35: It should be made clear whether this is a U.S. average concentration, or
whatever.  The term "integrated" doesn't make it clear.

P 2-57, L 5: First,  "Cal-EPA" has not been defined.  Second, what kind of locations
were the "three locations"?

P 2-57, para 3:  Why mention dioxin? First, dioxin was not discussed previously as a
diesel engine emission.  Second, the cited facts are "unpublished".

P 2-58, para 1: This is not an adequate rationale for the decision not to update the
chapter.  Not only is the question of changing emissions "not rigorously addressed",  it
isn't addressed at all in any meaningful way. This isn't acceptable.  While older
engines are still on the road, newer ones are as well, and in ever-increasing numbers.

P 2-59, L 7-9:  What about hopane as a marker''  Do»you  disagree that it is one?

The use of the rat in justifying the dosimetric approach seems weak at best. You
discard the importance of vapor-phase compounds because they don't cause tumors in
rats. You discard aldehydes because no nasal tumors were seen in rats. You note that
rat lung tumors occurred only under 'overload' conditions, which you didn't define, and
that soot-associated organics don't seem to be important for the rat response.  The
entire rationale based on rats seems circular

Why not just go back to basics.  The present concern for diesel soot-induced
carcinogenesis in humans is based  solely on the soot-associated organics - it's as
simple as that. Who needs the rat to justify soot as the proper dose indicator? Given
that, you  still need a better rationale, and perhaps some further thinking, about why you
don't care about volatile organics. Are you really  prepared to say that there are no
environmental health issues associated with the volatiies?

This text  indicates that, while you acknowledge the lack of utility of the rat lung tumors,
your thinking is still being driven by the rat results to an inordinate degree.

Why not just defer to the recent PM criteria document  for the particle dosimetry
discussion, and just summarize a few key points here? There is no need to repeat it.

Regardless of the comments above arc! below, this chapter isn't in bad shape, and can
be accepted with minor editing.

P 4-1, L 8-9: It is not clear what is meant by 'tumorigenic mechanisms' are 'dosimetric

P 4-1, para 2:  Here you dismiss vapor-phase compounds because they don't cause
tumors in rats.  How do you know they aren't important in humans?

P 4-1, L 22: You'd better define 'particle overload' if you are going to  use it. In line 29,

you switch to the term 'pulmonary overload1. Although neither is defined, you
complicate things further by switching terms.

P 4-2, section 4.2: Why is the recent PM criteria document not cited for dosimetry

P 4-4, L 7: There are more recent reviews of dosimetry.
P 4-5, Figure 4-1: First, these data don't look to me like they represent current
thinking.  Do they? Second, Why use the 'M' symbols in the headings? Just write out
what they mean. Third, the symbols listed in the third line of the footnotes aren't useful
unless they are defined.

P 4-6, para 2:  This isn't clear. What is meant oy the statement that the 'deposition
rate' will 'initiate particle redistribution'?

P 4-7, L 3-4:  Under more or less continuous deposition conditions, the fact that
organics might be released more rapidly from soot than previously thought doesn't
necessarily make them less important.

P 4-11, L 26: Do you really mean that 'deposition' in the alveolar region was 60 mg, or
was that the lung burden? Deposition and lung burden are quite different things.

P 4-14, L 17: I think  you mean macrophage 'pool', not 'tool'.

P 4-15, figure  4-4: This figure is not useful without explaining all the symbols, which
would be more than the Deader wants,  or needs.  Take it out.

P 4-16, L 7:  Do you  really mean to say that the clearance impairments caused by silica
are likely to be caused by surface-associated organics?  If that's really what you meant
to say, you'd better give references.
P 4-27, L 14-15: While the statements in the paragraph  might be true, how does  the
last statement follow from the preceding statements'?

P 4-28, section 4.4.4: Don't forget that soot gets into epithelial cells as well.  In fact, it
is possible that only the soot in epithelial cells is the 'effective dose'. That may not be
true, but you can't summarily dismiss the possibility without mentioning it.

P 4-29, L 21: Capitalize 'Beagle'.

P 4- 31, L 17-18:  Why are you concerned about 'overload' dosing regimes regarding

human risk?

P 4-33, L 4:  Don't you mean 'organic matter dissociated from particles', rather than
'particle-free organic matter'?  The latter would presumably be volatile organic matter,
and you weren't worried about that earlier.

This chapter ranks among the top in terms of being in acceptable shape. It could be
accepted with relatively minor effort.  The only general issue regards the section on
immune responses.  That section doesn't portray the issue of the specificity of the
effect of diesel soot. While it is true that soot has adjuvant properties, it is not clear
that this effect differs from that of other particle types. There is a Japanese study
(Maejima et al. J Toxicol Env Health 52:231, 1997) which addressed that issue, and
found that diesel soot was about average among several diverse particle types. That
study is not referenced.

P 5-1, L 15:  How can 100/*g/m3 be called a 'low' level?

P 5-2, section  While odor is a problem, The information isn't very useful
unless information can be provided comparing detectable odor levels to concentrations
of soot in units used elsewhere in the document.

P 5-6, L 18:  A new paragraph should start in the middle of this line.

P 5-6, L 30:  What was the challenge dose?

P 5-7, L 9-11, and 25:  DAM may act to enhance allergic disease, but it is important to
place DAM in the context of other particles. Some mention should be made of the
issue of specificity.

P 5-18, L 3:  Something is missing in the dosing description.  How do you administer
mg/kg 'for 30 min'?

P 5-18, L 10: What does 'potently' mean? How big  were the increases? Potent
compared to what?

P 5-25-26, Table 5-3:  First the table  lists the high exposure concentration in the
Mauderly et al. 1987 study as 7.0 mg/m3, when it was actually 7.1 (7.08). It is listed
correctly in the text.  Second, the Mauderly et al. FAAT 30:233, 1996 paper on mice is
not included. It has data on body weight and survival.

P 5-31, Table 5-5: Again, the high dose in the Mauderty et al, 1987 study is listed

P 5-35, Table 5-6: Again, the Mauderly et al., 1996 mouse study is missing.

P 5-41, L 25-26:  The sentence beginning 'although the mouse lungs' is not dear.
Perhaps 'high1 is meant to be 'higher1.

P 5-45, para 1: This summary doesn't mention the possibility of soot in epithelial cells
being a problem.

P 5-51, para 3: Here the high dose in the Mauderly et al. 1987 study is listed
incorrectly as 7.0 instead of 7.1 mg/m3.

P 5-76, L 15-16:  Here, you  switch from calling the rats 'young'  to calling them
'neonatal'. Which is correct?

P 5-77, L 22:  It  should be 'aggregation of macrophages in rats'.

P 5-78, para 1:  It is arguable whether or not a 2-yr exposure of monkeys should be
called 'chronic'.  The point is well-taken that 2 yrs is a much smaller portion of the
lifespan in monkeys than in  rodents, and that a 2-yr study in monkeys is not an
adequate carcinogenesis bioassay. However, a 2-yr exposure is still 'chronic' in the
sense of time, in the parlance of bioassays.

P 5-84, L 6:  Again, it is not  at all clear that one would call a study of the same length
'chronic' in rats and 'subchronic' in monkeys. Better to avoid this jargon.

This chapter is in reasonable shape, and can be accepted with attention to the minor
points listed  below.

P 6-1, L 27:  'RfD' should be 'RfC.

P 6-2, L 10:  It hsould  be -'risk values', rather than '-risk assessment values', unless
you mean something that I don't understand.

P 6-3, L 7: There is high confidence that diesel exhaust can be a respiratory hazard at
occupational exposure levels, but one can hardly have 'high' confidence that it is a
hazard at environmental exposure levels.

P 6-5, paras 2-3: Who cares about the 'target concentrations'?  You don't make that
distinction elsewhere.  Just list the actual exposure concentrations as you do

P 6-6, L 20:  I don't recall that the abbreviations for the respiratory function parameters
have been defined.  They aren't used repeatedly, so just spell them out.

P 6- 6, para 4: Again, the high concentrations in this study were 7.1 mg/m3.

P 6-27, L 3:  The concentration in the ITRI study was 0.35, not 0.36 mg/m3.

The descriptions of the studies are highly variable, to the degree that it appears as
somewhat of a orandom walko.  It is true that the reports of the studies varied in their
detail, but it would be desirable to present them is some logical order and with some
uniformity in describing the features of the methods you think are important for the
reader to understand.  The studies are not presented in alphabetical or chronological
order.  The following characteristics are presented for single,studies, even though in
some cases, the information is available for other studies as well.  These comments will
not be repeated below - look through the text and you'd find them:

Hours of day animals were exposed
Temperature and relative humidity in chambers
Rate of diluting airflow
Exposure during darkness
Tumor rate 'beyond  that expected for aging F344 rats'
Number of engines used
'Lifetime' exposures (none were)
      Trade name of carbon black used
      Data for tumors at different interim sacrifice times

      It is stated that positive studies included a postexposure observation period.
Few did.

      Squamous keratin cysts (or whatever you choose to call them) are not
differentiated from other lesions  in this chapter. That was a major criticism raised in the
last review of the document. The type of lesion is  listed for some studies in the tables,
but no distinction is made in the  text.

      It is stated that exposure-response comparisons can't be  made for mice (like
they can for rats) because  the experimental designs differed.  That isn't any more true
for mice than for rats.  The problem with presenting exposure-response plots for mice,

is that there weren't positive responses to plot in the long-term studies.

P 7-1, para 1: Why repeat this material?

P 7-1, L 22:  These emphases are hardly 'recent'.

P 7-9, Table 7-1: No group sizes are given for the Mauderly et al. Mouse study. They
were presented clearly in the publication.

P 7-12, L 19:  While it is true that the results indicated that the organic fraction was not
the osole caused of the tumors, that is a very biased description of the results.  The
results indicated that the organics played little, if any, role at all.

P 7-20, L 19-20:  First, the numbers of animals listed are not the numbers examined for
tumors, which would seem to be the more important number to list. Second, the high
exposure level is incorrectly stated.  It was 7.1 mg soot/m3, not 7.0.

P 7-21, L 1-5:  In three different places  in five  lines, the population was described as
'animals', 'combined males and females', and  'males and  females'. Are these all the
same? Why the different terminologies?

P 7-23, L 31:  True, 1986 is 'more recent' than 1982.  In 1998, however, that hardly
seems worth pointing out.

P 7-28, L 8:  Why mention  that this material was 'lampblack1? Carbon blacks come in
several forms and are made by several processes, such as 'channel black'.  You don't
mention which kind of process was used when you cite other carbon black studies, so
why here 7

P  7-35, L 22-23: Your treatment of this information is not clear. You state that this
study shows that diesel soot extracts are not effective tumor initiators or promoters,  but
you certainly assume they  are in the rest of the document.  It's not clear how you
discard these findings without explanation.

P 7-39:  The negative Mauderly et al. 1996 study is missing here.  Any reason why it
isn't relevant to the discussion?

P 7-43, figure 7-1:  This figure isn't useful without taking the control tumor incidences
into account some way.  For example, why not use the net incidences (exposed minus

P 7-44;  Why are the Heinrich et al. 1995 and the Nikula et al. 1995 studies not

relevant to this discussion?

It is only partially clear why the authors selected the particular studies for presentation.
Some rationale is given for excluding other studies, but studies are still presented an
then judged as useless for answering the issues. It seems like you either have to
include them all, or excluide all except those you consider to present useful

It is not useful to  switch back and forth between the terms 'exhaust' and 'fumes'.
Technically, the two are not the same thing.  Even though the original authors of the
papers cited may have used the term 'fumes', there is no reason EPA has to perpetuate
the mistake and confuse the reader.

The point is made in one. place that some of the 'nonexposed' population might have
nonoccupational  exposures to diesel exhaust. While true, the point as presented is
absurd.  All subjects in the control groups, occupationally-exposed groups, and indeed
in the entire U.S. population are exposed frequently, if not continuously to diesel
exhaust.  Indeed, they are all incurring the full range of effects that EPA proposes might
result from environmental exposures. The studies only'deal with presumed differences
in the level of exposure, not with exposed and unexposed populations.

P 8-1, para 1: Why reiterate this information?

P 8-1, L 17:  No study has been 'definitive'. What is meant here?

P 8-1, L 20:  Exposure is  'uncertain' in all studies. What's the point?

P 8-2, L 8: Why use the term 'fumes'?  I suppose in this case the idea was that the
original authors used that term  and the heading reflects the title of the paper.

P 8-4, L 26:  Do a universal search on the chapter for the term 'dose-response'.  None
of the studies gave 'dose-response' information.  All they gave was exposure-response

P 8-5, L 11: Who-cares that the term 'rolling stock' was used?  Why is this an
important detail?

P 8-9, L 13-14: There are actually no exposure histories for any study.  It is  incorrect to
state that this, or any other, study has 'limited' information on exposure level. None of
them have any information on exposure level, do they?

P 8-30, L 23-25:  It is not at all clear why comments are made as to how the study could
have been improved. One could make that comment about all of the studies in the
entire document, assuming that none were perfect.  Making this point here doesn't
make sense.

P 8-33, L 29:  Should the first word of the line be ocarcinogenso instead of 'cancers'?
How is one 'exposed to cancers"?

P 8-59, L 4: The point about some individuals being nonoccupationally exposed seems
off the mark.  Do the authors believe that anyone in the U.S. is not nonoccupationally
exposed to diesel exhaust? Everyone in all the 'control' groups  used in the
epidemiological studies are exposed frequently to diesel exhaust, and presumably
incurring all the health risks that EPA proposes might be associated with environmental
exposures. The 'exposed' groups only have higher exposures.

P 8-59, L 20:  What is meant by 'extract' here?

P 8-59, L 29:  It is not clear how the original authors could have used the exposure
information more 'quantitatively'. Here again, the authors offer retrospective advice
that is meaningless.

P 8-63, L 11:  Why is it 'interesting' that the main risk factor for lung cancer was
cigarette smoking? That's hardly unexpected.  What's the point?

P 8-66, L  19:  These criteria don't 'define' causality at all, their fulfillment 'suggests'
that causality is an acceptable working assumption.

P 8-67, L 4: Why call the 1995 publication 'recent'?

P 8-67, L 7: The word 'consistently1 doesn't seem appropriate here. Although there is
no question that the preponderance of evidence indicates that occupational exposure
to diesel exhaust is associated with an excess of lung cancer, certainly not all studies
reviewed by HEI showed that. The word suggests that the finding waqs 'consistent1
among all studies, and it was not.

P 8-67, L 14:  l'  m not confident that this is what 'specificity' really means  by these
criteria.  The studies found lung cancer because that is what they looked for. The fact
that cancer as an endpoint was identified doesn't equate to specificity of effect. Cancer
is certainly not specific to diesel exhaust.

P 8-67, L 34-35: Vostal suggested that particles induce lung cancer nonspecifically in
rats.  The sentence implies that Vostal suggested that this effect might occur in
humans, and he did not.
                                     A- 12

P 8-70, L 6: This sounds like EPA is inventing a new category; 'close to human
carcinogen'.  The rationale of the paragraph seems a bit overdone.

This chapter is lacking mention of whether or not the mutagenicity of soot from newer
engines is similar to, or different than, that of soot from the older engines from which
these data were derived. The chapter is unacceptable without some mention of this
issue. Otherwise,  it's in pretty good shape.

P 9-3, L 19: The point is not whether or not the data are 'conflicting'. The point is
whether or not  present data indicate that exposures by inhalation produce urinary
mutagenic activity.  There is not a clear statement on that.

This chapter does not present a very accurate  view of the current understanding of the
most likely mechanisms of diesel exhaust-induced carcinogenesis in humans and
animals.  The treatment of the issue of mutagenesis by particle-associated organics is
especially confusing,  because seemingly conflicting notions are presented.  A large
part of the problem seems to be the author's attempt to reconcile the human and animal
responses. This is a futile task.

It is repeatedly proposed that the longer residence time of particles in humans would
enhance the opportunity for mutagenesis by organics, but it is also stated that
residence time does not influence adduct formation in animals.

Adduct formation information is used to support an action of organic mutagens, but
there is not emphasis on the fact that there is very little information to date showing that
the elevations of adducts observed in exposed animals are any different from those in
controls.  For the most part, elevations of adducts in the animal studies are increases in
the. concentrations of  the same  adducts that occur in controls, which does not support
the idea that the adducts resulted from soot-associated organic compounds.  One of
the most intensive studies of this issue is not even cited in the chapter (Randerath et at.
HEI Report No. 68, 1995).

It is repeatedly acknowledged that increases in lung tumor incidences only occur in rats
under overload conditions, yet it is also repeatedly hypothesized that a  tumor effect
occurred, or must have occurred, at low doses,  but that the studies were not robust
enough to detect it. This ophantomo tumor effect is a strange and unsupportable
hypothesis. The studies were just as robust at low levels as at high levels. If the
aggregate data from the many studies acceptable as bioassays are presented (which is
                                    A- 13

not done), it is clear that the data do not suggest an increase that simply failed to reach
statistical significance.  The data do not suggest an increase at all.  There is no
scientific basis for calculating a phantome low-dose effect, and even less for using that
imaginary effect to calculate quantitative estimates of human lung cancer risk!

The premise that organics play a larger role at low exposure levels is asserted
repeatedly. This hypothesis doesn't make much sense, and is not well-defended. It
makes sense that, if there are two separate phenomena and one operates only at high
doses, then the other would dominate at low doses. However, this doesn't mean that
the second response is larger at low doses,  it only means that it is relatively larger.

The chapter needs to be edited to clarify the species being discussed at each point. As
presently written, information is blended across species carelessly.  This is a major
point, and there is simply no reason for the Agency to confuse it. We get lung tumors
in rats at high doses.  Our present understanding clearly points to the fact that this
signal should not be used to estimate human risk.  We  do not get tumors in animals at
low doses.  We seem to get an increased risk among human exposed occupationally.
It is most.plausible that the increased risk among humans stems from the soot-
associated organics.  The animal response does not seem to be associated with soot-
associated organics.  We don't understand precisely why there is an apparent
difference among species, but there is plenty of evidence indicating that there is.  End
of story. Trying to unify the responses across species at this time is neither necessary
nor wise.

It is stated that tumors only occur in mice exposed from conception, but this does not
agree with the description of mouse responses elsewhere.

P 10-1, L 11: The organic play a relatively greater role at low levels. We have no
evidence that they have an absolutely larger effect at low levels than at high.

P 10-2, para 2: Why give the brand name for the carbon black here? That isn't done
for other studies - even those described on the same page.

P 10-2, L 19:  The Fraunhofer study did not examine the tumorigenicity of the 'carbon
core'  of diesel soot. They used carbon black, not the carbon core of soot.

P 10-2, L 23:  There must be words missing from this sentence - it doesn't make sense.

P 10-2, L 31  and 33:  In two sentences, you state both that the  carbon black did, and
did not, have organics. Care to select one? The truth  is, as reported, that the carbon
black did have organics, but very little and the material that was extracted had very little
mutagenic activity.
                                   •  A-14

P 10-3, L 2:  Soot in lymph nodes was observed in all diesel studies; why mention it
here if you aren't going to mention it elsewhere?

P 10-3, L  30: The term 'overload', while admittedly vague, is not used only in
reference to retained mass or volume.  If you are going to use this term, you'd better
give your definition.

P 10-4, para 1:  Why talk about specific surface area for this material if you aren't going
to mention it for other carbon blacks?

P 10-4, L 20: What is the evidence that the organics play a 'minor' role in overload-
induced carcinogenicityo in rats? Is there evidence that it plays any role?

P 10-4, L 31: The Nikula reference is not a good reference for the composition of
diesel soot - it just quotes other papers. Many other authors have quoted the Opresko
paper - why not list them as well, if you are going to quote Nikula?

P 10-6, L 7:  What is  the basis for the statement that slow release of organics
6preventso overwhelming of activating pathways?  This isn't stated as a hypothesis, it's
stated  as fact. What  is the reference for this fact?

P 10-6, L 9-13:  The rationale here escapes me. What is the link between the fact that
all particles tend to form aggregates in the lung and their ability to 'react1 with
surrounding  lung medium owithout interferenceo from other particles? What point is
the author attempting to make?

P 10-6, L 24-25: First, I don't,think other chapters agree that tumors have only been
observed in mice exposed from conception. Second, how might exposure from
conception make mice omore  sensitiveo"? is there other evidence for that? What
would the mechanism be?

P 10-7, L 5:  What is  the relevance of kinetics in rats exposed to 500 mg/m3 to possible
kinetics in humans?

P 10-8, L 32: The observation is that most of the retained particles are.in
macrophages, not most of the 'deposited' particles.

P 10-9, para 3:  First, what alveolar cells could form tumors other than Type II  cells?
This is the only alveolar type that has been known to form tumors from any treatment.
Second, the  point  about the metabolic potential of Type II cells becomes moot when
you state that the organic mutagens don't seem to be important in rats anyway.

P 10-14, L 33: What  is a 'particle relapse process'?

P 10-15, L 11:  'What is a 'lung-free1 cell? I didn't know that any cells had lungs.

P  10-16, L 2: Type II cell hyperplasia is a common feature of the lung response of
rats. It is either much less, or missing, in other rodents and in primates. The statement
is misleading.

P 10-17, para 2:  There are lots of problems with the generalizations stated in this
paragraph. First, can we presume that the paragraph  refers to animals, or are humans
included? Second, it's not clear what the author would accept as 'conclusive1 or
<5definitiveo proof. That kind of wording seems hyperbolic. It's hard to imagine what
additional data would be required to determine that the tumors in rats occur only at high
doses. There are dozens of studies with many different particle types. Third, the
recent ILSI workshop concluded that inflammation was a prerequisite for the rat lung '
tumor response.  Of course that workshop occurred after this draft was written.  Finally,
the information about particles in epithelial cells seems to be tossed in to support a lack
of threshold (unless I've missed the point). Of course  particles are taken up in
epithelial cells.  Can we presume that, because of this, the author proposes that there
can be no threshold for any particle? Does that make sense?

P 10-18, para 3:  Isn't the claim that retention of 14C,  a/id thus organic, is not
important in conflict with your premise that humans are at greater risk because of
longer clearance times?  How do  you reconcile the two premises?

P 10-22, para 2:  Why is there no mention of the Randerath et al. report on DMA
adducts in rats exposed to diesel  exhaust and carbon  black?  That report contains half
of what we know about the subject at present.

P 10-22, L 28:  It shoulc oe made clear that this statement applies only to rats.

P 10-23, L 3:  Are there words missing between 'lung'  and 'provides'?  If not,  the
sentence doesn't make sense.

P 10-23, L 15:  Not a 'greater role',  but a greater relative role.

P 10-23, L 20-22:  To what information is the author referring when it is stated that DE
'is effective in non-particle-overload conditions'?

This chapter carries forward some of the difficulties in previous chapters, by building on
assertions that have questionable basis.

There is no reason that the comparative potency approach is not still useful as a range-

finding estimate of risk; especially when there is no method at present for deriving risk
estimates from animal data with confidence.  The Agency has had ample opportunity to
conduct updated comparative potency estimates based on current mutagenicity (and
other) data, using contemporary test materials. This should have been done.
Continuing to use comparative potency estimates based on the old Nissan samples is
not sensible, in view of the well-known fact that those samples were collected with the
engine 'de-tuned" to produce more soot than normal.

Continuing to calculate quantitative estimates of human lung cancer risk from low-level
environmental exposures from the high-dose rat data is simply not sensible,  given our
current understanding of those responses. The Agency was advised against that
approach at the last review.  For some reason, the Agency has decided to directly
ignore that advice.  The evidence against that approach was adequate then  and has
been strengthened considerably since then.

Calculating quantitative human risk values from the imaginary tumor response of rats at
low doses is not sensible. It might be sensible if the aggregate data from the many
studies suggested a response that did not reach statistical significance because of
group sizes, but that is not the case. The aggregate data clearly show no response at
all in the low-dose regime.  Therefore, the agency is  only playing 'what if by using this
approach. It is true that such calculations can be done.  In the face of our considerable
information, however, it is not true that such an exercise is warranted. The logic of this
exercise appears to be coupled somehow with the Agency's repeated assertion that the
organics play a greater role at low dose levels. If the organics played a role in the rat
response, it is true that their role must have been relatively greater at low doses.
However, there are lots of data  in the low-dose regime, and the aggregate data proved
a convincing case against the organics having any apparent effect at all.

The Agency does not advance its case by the above approaches.  It is clear that the
aggregate epidemiological data, and the results of the most robust of the studies, both
make a convincing case for a positive, but small, increase in risk from occupational
exposures. It is also clear that our best current hypothesis for this effect is the action of
the soot-associated organic compounds.  Framing our understanding of the quantitative
risk level using the human data and comparative potency comparisons is logical,
although we can't estimate risk with a very high level of confidence. In aggregate, the
toxicological data support the plausibility of an action of the organics.  However, using
the rat lung tumor response, real or imaginary, to estimate human risk is not a
supportable approach in view of our current understanding.

P 11-2, para 2: The negative Mauderly et al. 1996 mouse study is not cited in  this
chapter, but should be.

P 11-4, L 17-18:  While I suppose one can always invoke 'possibility', there is J
substantial evidence against this assertion.

P 11-5, L 26:  It is never mentioned that the Nissan engine was not operated as
recommended by the manufacturer, but was 'de-tuned1 to produce more soot. Using
such samples for risk estimates is not supportable.

P 11-8, L 26:  While 1986 is unarguably 'more recent' than 1983, it is hardly 'recent1.
Our views of the  best approaches for estimating risk have evolved over the past 10
years, yet this chapter does not seem to acknowledge that evolution. The chapter
lumps together reports from the beginnings of our exploration of cancer risk and those
that are more recent.

P 11-12, L 26:  Again, it is asserted that the organic play a 'larger1 role  at low levels.
There is no support for this speculation.  There is support for the conclusion that
doverloado effects play no role at the lower levels

P 11-13, Table 11-2: These studies were not done using a heavy-duty engine, as the
title states.

P 11-15, L 4-7:  We know from the substantial  data that the risks are not linear in rats,
and that the mechanisms causing tumors at high doses in rats most likely do not apply
for humans.

P 11-15,1 20-21: We have substantial data showing that the 'particle' effects do not
operate at low levels.

P 11-19, L 20-22: How does the Agency propose that exposure from conception
increases sensitivity?  This assertion is made repeatedly, without citation of any
supporting rationale at all.

P 11-20, L 12: What is meant by 'generally'? Have 'significant lung tumor increases'
ever been demonstrated at other than .overloading' doses.

P 11-20, L 27: It should be stated clearly that the 'particle-overload hypothesis' has
only been stated for rats.

P 11-21, L 35: EPA has had plenty of time to carry out such a study. The fact that this
hasn't been done is a significant deficit in the Agency's approach to the issue over the
past 10 years, and particularly over the past 5-8 years when the uselessness of the rat
response  has become evident.

P 11-23, L 3-4: I disagree that the estimates derived from rats are useful for placing

bounds on human risk.  The Agency appropriately uses the estimates from human data
to frame their conclusions regarding risk.  However, the animal-based estimates are
carried through to the final paragraphs and used to frame the range of risks.  Based on
our current knowledge, the animal-based risks should not be used at all.

This chapter includes two aims: 1) summarizing the characterization of health risks that
stem from the preceding chapters; and 2) providing a lay description of the diesel-
health issue.  These are both good goals, but would be better presented in two
separate chapters.

The lay version summary falls short of an accurate portrayal of the issues because of
its attempt to simplify language.  As presently written, the simplified  language is
misleading.  The task could be accomplished without being misleading, with a more
careful choice of wording. No attempt to mislead is implied by these comments, it's just
that there are still some technical terms that would be obscure to the (ay reader, and
word choices in some places actually convey miss-impressions.  The section just needs
to be edited again with an eye on this problem.

In both sections,  more effort should be given to portraying our understanding of risk
more accurately by presenting MLEs and distinguishing our 'best' estimates of risk from
the upper-bound estimates.

Figure 12 doesn't help a bit.  In fact, I became confused by it, even though I felt I had
understood the text.

P 12-2, L 4:  This text seems to imply that 'aggregate' hazard is added on top of the
hazard of the individual  constituents.  Such a 'double whammy'  shouldn't be implied
unless the authors have some actual  information or hypotheses they can provide.

P 12-2, L 17:  What kind of cells  might be damaged other than 'biological1 cells?

P 12-2, L 26:  The meaning of this statement is not clear.

P 12-3, L 25-26:  The statement that our data 'show1 that diesel  exposures play a role in
the 'development' of allergic disease is a misleading overstatement of our present

P 12-3. L 33:  Place 'high' in context for the lay reader.  I think you mean levels that are
2-3 orders of magnitude above human environmental exposures, but the reader can't
tell this.

P 12-5, para 1:  You should be careful to state that the information you are relating
comes from rats, not all animals.

P 12-5, L 19:  It is not clear what you mean by 'short-term basis'  The effects are seen
over work shifts, but the people are mostly exposed daily over long periods.

P 12-6, L 23:  Again, put 'markedly' in perspective.  For example, the exposures are
about 1000 times environmental levels.

P 12-7, L 1-4:  Why use this 1989 Japanese reference when there are U.S. data?

P 12-7, L 21:  What 'science policy' is the author referring to? It's regulatory policy, not
science policy.

P 12-8, L 7-12:  This dissertation on Printex and its surface area doesn't make sense.
First, why use the trade name?  Second, why state that surface area 'makes up' for its
lack of organics? There is no basis in the foregoing chapters for this statement.  Third,
why not mention the other study (Nikula) that makes up half of what we know about
carbon black effects.  Is it because the carbon black in that study produced the same
effects at a much lower surface area  (selected to be like diesel soot in that regard), and
that screws up the surface area trade-off hypothesis?  Overall, this section just doesn't
make sense.

P 12-9, L 3-5:  It is true that multiple tumor types were observed in animals.  However,
multiple tumor types are also observed in humans, and the fact that the tumors in rats
differ from those in humans says nothing about whether or not multiple mechanisms are
at play.  The logic here is-unclear.

P 12-9, L 13:  Studies of rats that included both males and females consistently found a
greater response in females. What is meant by this statement?

P 12-9, L 19-21: What is the point of waving the children's susceptibility flag as a
speculation here?  What would make them more susceptible? The only data we have
on this issue says just the opposite.  Is this an obligatory statement?

P 12-10, L 33: Which did you use, 29 or 23?  Why state the difference?

P 12-11, L 18-20: This statement is confusing, and its meaning Is not clear. You state
that all criteria apply 'well', yet go on to state that others may come to a different
conclusion. What is the reader supposed to derive from this?

P 12-12, L 16: What does 'nominally' mean - some studies had over 200 subject per

P 12-12, L 18: There have been no attempts to produce carcinogenicity in cats or
monkeys.  Surely you don't think that the investigators seriously thought that their 2-yr
studies were a test of carcinogenicity. There were no 'unsuccessful' attempts in these
species, because there were no attempts at all.

P 12-14, L 8:  What are the 'two components'?

P 12-15, L 24: Stating that there is evidence that a 'majoro hazard is presented by
inhalation of diesel exhaust is misleading.  One might envision a 'major' hazard from an
occupational exposure. Based on present data, it would be a real stretch to imagine a
omajoro hazard from environmental exposures.

P 12-17, L 16-20: The comparison to PM2.5 doesn't have much merit.  Diesel soot is
included within PM2.5, so you aren't contrasting two different materials, they overlap.
Overall, it's not clear that this paragraph adds'much.

P 12-21, L 6-15:  Why not label this material as a 'tabular summary1 as done on P 12-

P 12-23, L 6:  This statement isn't useful, since there was no response at the lower

P 12-25, L 13-14: It is not clear what 'rationalizing lower limits on  risk' means.

P 12-26, L 17-18: That's just the problem. The present document does not discount
use of the rat data, and it should ( as CASAC has stated previously).

P 12-26, L 31: This statement is confusing. Upper bound risk estimates bound only
the upper limits of risk.

P 12-27, L 5: 'Illustrates' is misspelled.

P 12-27, L 9-12:  These rat data are from the overload dose regime.  This section
explicitly intermingles that response with human environmental exposures.  It is not
clear what useful margin of exposure (MOE) can result.

P 12-27, L 17-19, and P 12-28, Figure 12-1:  I don't find that this figure helps at all.  I
find it confusing,  even after some study. A naive reader wouldnt have a chance of
getting anything meaningful from it.

P 12-29, L 1-2: Although not all endpoint have been examined, studies have been
done in developing rats, and all the data we have indicates that they are less, not more,

P 12-29, para 2:  First, I though the Cal-EPA draft was marked 'not to be cited or
quoted'. Second, where is mention of HEl's assessment document?

P 12-30, L 7-12:  Doesn't the recent PM Criteria Document provide a more recent
estimate of Pm levels?

P 12-30, L 16: Does ointegratedo estimate mean a nationwide average, or what?

P 12-30, L31:  Define TEQ.

P 12-30, L 34-25:  What does this statement about 'ruling out the possibility of
exposures of interest' mean?  It isn't clear.

P 12-31, para 2:  This paragraph is far too light a treatment of the evolving emissions
issue.  It is a gross understatement that the data on engine emissions are 'pre-1998'! -
I can understand that it's embarrassing to admit how outdated the emissions
information in this document actually is. This document requires  more information on
the nature of emissions from contemporary engines, and the trends that are expected in
coming years.  There should be mention of fleet turnover times, and projections of
emissions into the future.

P 12-32, L 18:  It is not clear what 'background means here.

P 12-32, L 30:  Can you think of a situation in which exhaust is not 'breathable'? What
do you mean here?

P 12-33, L 16:  Are you certain that inhalation is the major pathway? I haven't seen a
recent estimate, but  Cuddihy et al. did  an estimate several years  ago, and said that
more soot would be ingested than inhaled.  Is this assertion an assumption, or do you
have a methodical estimate you can reference to back it up?

P 12-33, L 19:  The statement should be explicit that only a portion of the inhaled
particles and gases deposit on lung tissues - in fact it's a minority of the inhaled

P 12-33, L 27:  I'm not sure what you mean by stating that the symptoms of exposure to
diesel exhaust are like the 'onset of a common cold'.  The  onset of a cold is almost
always signaled by a runny nose accompanied by aches and tiredness. Are these the
symptoms of 'episodic' exposure to diesel exhaust? Do you mean environmental or
occupational exposures? I've been exposed 'episodically' (I think), and I sure haven't
had 'cold-like' symptoms.

P 12-33, Last para:  I don't understand the context of this paragraph.  Remember that

everybody in the U.S. is exposed to diesel exhaust every day

P 12-34, L 25-26: These statements need to be put in context. Are you really
concerned about a 'very high one-time' environmental exposure? I can see this
section quoted in a lawsuit over permanent injury resulting from the passage of a
smoking bus! This whole section needs to be re-thought.

P 12-35, L 1-3:  This wording makes it clear that there is a threshold, and that only
exposures that are 'too much' increase the likelihood of cancer or noncancer effects. I
might agree, but I doubt that's what you intended to state.

P 12-35, L 11: What is a 'test' exposure'?  This term hasn't been  used in any of the
animal study sections.

P 12-35, L 24-25: Here again we have the issue of increased susceptibility from
exposures in early life.  With no evidence given for such an effect, repeating this
speculation so many times seems unreasonable.

P 12-36, L 12: Shouldn't 'many'  be 'most"? If not, then the value is not useful.

P 12-37, L 22: Here and throughout, it should be made clear that the real risk is
expected to be less than the upper-bound estimate. The wording here and elsewhere
implies that the risk may be lower, but that's not an accurate view of the estimates.
Actual risks are always expected to be less than the upper bound estimate.

P 12-37, L 33-34: Of course a carcinogenic hazard is 'indicated'  for ambient PM
exposures.  What do you think the  6-cities study indicated?  Cancer risk was equal to
cardiorespiratory risk for long-term mortality (Hfespan shortening). Indeed, you  ought to
be pointing this out clea'rly  , because it is perhaps the strongest supporting evidence for
a cancer effect from low-level exposures to mutagenic material in the environment.  If it
wasn't for that finding, concern for low-level cancer risk would be pure speculation.

P 12-38, para 1:  This paragraph doesn't help much. For example, where in the
foregoing material was the issue of eutrophication discussed''  Is the naive reader
supposed to know about this and understand the statement?
          Mauderly 5/98


General Comments:

I found this document to be lacking in a clear focus and direction. Although we had the list of
items from the Office of Mobile Sources as to why they wanted it, I think it suffers from the lack of
a clear set of needs specifications to the NCEA personnel writing it as to what it is intended to do
and thus, why it has utility.  It seems it has suffered from lack of a clear purpose and hence has
not been given much attention over the years and certainly very little attention to the modification
of the 1995 version before presenting it to CASAC. I suggest that a chapter 1 that clearly lays
out what the purpose of the document is and how it is expected to accomplish this goal(s) would
provide both the reader and the writers a clearer view of why we are going though this exercise
and aim them towards providing the basis for whatever decisions are likely to be based on this

I can envision several reasons for the document, but none that fit the document as it currently
sits.  It appears it was started when the assessment of the 1980s technology would have been
appropriate for making decisions regarding emissions controls.  However, those controls are now
set and are being put into operation. Thus, the purposes of this document could either be to
assess the  impact of diesel emissions on ambient paniculate matter concentrations and the
toxicity thereof or to look forward to the effect of the reduced emissions that arise from changes
in the combustion conditions and the input fuel. In either case, the'2  should be a view to using
what we know about diesel emissions  and their specific health effects to inform our thinking
about ambient PM2 5 and vice versa.  It is clear that diesel emissions  are an important
component of PM2.5 and yet there are no data to suggest that diesel emissions are significantly
different in their toxicity and carcinogenicity from the general ambient aerosol. Thus, we should
not be separating the discussion of these two clearly interrelated materials.

Estimating the effects of the existing and planned  controls on emissions would seem to be an
important albeit difficult task.  The changes in emissions has reduced paniculate matter with
some increase in NOx which would lead to an increased secondary acidic species concentration.
The reduction in emitted particle mass may be the result of processes leading to larger numbers
of much smaller particles and we are now concerned that the number of ultrafine  particles may
be playing an important role in PM toxicity.  The changes in combustion that increases NOx could
also then increase nitrated organic compounds, some  of which are known carcinogens.  Thus,
changes in  the emitted organics, both gaseous and sorbed, could substantially change the risk
per unit mass of emissions.  Some effort to examine this question and at least the
acknowledgment that the question exists is important since the controls may be failing to improve
public health even if they appear to be lower in concentration.  The document clearly requires
major revision to give a purpose and then provide the information needed to fulfill that purpose.

Chapter 2
If there have been substantial changes in both engine  design and analytical methodology, are
any of the results presented relevant to the discussion at hand?  The statement at the bottom of
page 2-8 lines 35-36) suggests  that the rest of the document may be irrelevant to any
assessment of new diesel engine emissions and only relevant to the  assessment of those older
engines that continue in service. It also does not  speak to how the older engines might change
when they are periodically remanufactured as is common with larger  diesel engines. Since can
be the particle size,  number or nature of the sorbed combustion by-products that  are the
biologically important emissions components, it appears there is a need for up-to-date sampling
and analyses to properly characterize the emissions from current vehicles. There are  apparently


no studies of the effects of current diesei emissions on animal models and the epidemiology
applies to older engines and in particular those used on trains.  Thus, there needs to be a case
made as to how similar or different are the emissions from those engines used to generate
particles for the inhalation studies.  There are published studies of the emissions from new
engines. The California diesei document quotes several of them.  (Bob or Joe, I have some
others that I need to get the references for when I get back home). I do not know if there are
studies of the emissions from  remanufactured engines, but the question should be  asked.  If they
are qualitatively similar, then old results can be scaled by new emission factors.  If there are
important qualitative differences, then it needs to be  made clear that the old data are not
particularly relevant to the further regulation of diesei emissions and that it is critically important
to obtain relevant data.

However, diesei vehicles generally have a  fairly long functional lifetime.  There is some value in
understanding the effects of old vehicles in the context of ambient PM.  Is there an inventory of
vehicles that permits an estimation of how much emissions come from "old" vehicles verses how
much from  "new" ones?

The nature of chain aggregates also provides the opportunity for more material to absorb since a
liquid layer between two spheres will produce a concave meniscus. This is also  relevant for
chapter 4 since it  reduces the  volatility of the semivolatile sorted material  (negative Kelvin effect)
(Mariow reference). It would also reduce the solubility of these compounds in polar solvents like
water and related biological fluids.

Page 2-16 line 7 and 2-17 lines 1-3, how do you know? Reference or speculation?

Page 2-30 lines 5-7, OH measurement and modeling are better now that 1988.  This  is really
pretty far out of date and could be replaced with more modern references  particularly the 1994
Atkinson report on arene nitration reactions.

Lin  et al. (J. Air Waste Manage. Assoc. 42:1057-1062,  1992) show that one can  add  tracers to
diesei emissions.  There are patterns of hydrocarbons that permit assessment of the amounts of
gasoline and diesei emissions of modern receptor modeling methods  to be applied.  See the
California diesei document for references to at least 2 published receptor modeling studies and
there are others that they missed.

Chapter 4
Tracheobroncial deposition should not be .anored particularly as particle size moves smaller with
the  newer engine  emissions.  Although there are some cancers in the peripheral areas of the
lung, one should not discount  more central locations as a point of focus for lung cancer initiation.

The deposition of particles was further refined by the International Commission for Radiological
Protection (ICRP66, 1994) based on considerable new information. The ICRP's  new respiratory
tract deposition model incorporates more accurate values for the filtration  efficiency of the nasal
passages (Swift et al., 1992).  One important question is the rate of clearance of deposited
material. The assumption that the tracheobronchial region is completely cleared by a rapid
process has been challenged by a number of investigators beginning with  Davies (1980).  The
main evidence for slow clearance in humans comes from a series of experiments by  Stahihofen
and coworkes (Stahihofen era/., 1980; 1986a,b; 1987a,b; 1990, 1994; Stahihofen,  1989;
Scheuch (1991); Scheuch et al.,  1993).  The results of these studies are discussed in detail in
ICRP66 (1994) in which they conclude that a slow clearance phase cannot be excluded from a
complete lung dose model.  Thus, calculations based on the ICRP66 model include a fraction of


slow clearance. The presence of a slowly cleared fraction could be important in fully understanding the
health effects of deposited airborne particles. The deposition modeling in Appendix C is badly out of date
and the whole dosimetry should be repeated with current dosimetry.

Davies, C.N. (1980) An algebraical model for the deposition of aerosols in the human respiratory tract
       during steady breathing-Addendum. J. Aerosol Sci.  11, 213-224.
International Commission on Radiological Protection (ICRP66) (1994) Human Respiratory Tract
       Model for Radiological Protection, ICRP Publication 66, Annals of the /CftP 24(1/3) 482
Scheuch, G. (1991) Die Dispersion, Deposition, und Clearance von Aerosolpartikeln in den
       menschlkichen Atemwegen (Ph.D. thesis), J.W. Goethe-Universitat. Frankfurt am Main,
Scheuch, G., W. Kreyling, F. Haas, and W. Stahlhofen (1993) Effect of settling velocity on
       particle recovery from human conducting airways after breath holding. J. Aerosol Med.
       6(Suppl.) 47.
Stahlhofen, W. (1989) Human lung clearance following bolus inhalation of radtoaerosols. In:
       Extrapolation of Dosimetric Relationships for Inhaled Particles and Gases, pp 153-166,
       Academic Press, Washington, D.C.
Stahlhofen, W., J. Gebhart, and J  Heyder (1980) Experimental determination of the regional
       deposition of aerosol particles in the human respiratory tract. Am. Ind. Hyg. Assoc. J.
Stahlhofen, W., J. Gebhart, G. Rudolf, and G. Scheuch (1986.a) Measurement of lung clearance
       with  pulses of radioactively-labeled aerosols. J. Aerosol Sci. 17:333-336
Stahlhofen, W., J. Gebhart, G. Rudolf, G. Scheuch, and K. Philipson (1986b) Clearance from the
       human airways of particles of different sizes deposited from inhaled aerosol boli. In:
       Aerosols: Formation and Reactivity, Second. International Aerosol Conference, West
       Berlin, Germany, September 22-26, 1986, pp. 192-196, Pergamon Press, Oxford, U.K.
Stahlhofen, W., J. Gebhart, G. Rudolf, and G. Scheuch (1987a) Retention of radiolabeled Fe2O3-
       particles in human lungs. In: Deposition and Clearance of Aerosols in the Human
       Respiratory Tract, Second International Symposium, Salzburg, Austria, September 18-20,
       1986, pp 123-128,  ed. Hofmann, W.) Facultas Universitatsverlag Ges.m.b.H. Vienna
Stahlhofen, W., J. Gebhart, G. Rudolf, G. Scheuch, and M.R. Bailey (1986b) Human lung
       clearance of inhaled radioactively labelled particles in horizontal and vertical position of
       the inhaling person, J. Aerosol Sci.  18:741-744.
Stahlhofen, W., R. Koebrich, G. Rudolf, and G. Scheuch (1990) Short-term and long-term
       clearance of particles from the upper human respiratory tract as function of particle size.
       J. Aerosol Sci. 21(Suppl. 1):S407-S410.
Stahlhofen, W., G. Scheuch and M.R. Bailey (1994) Measurement of the tracheobronchial
       clearance of particles after aerosol bolus inhalation. In: Inhaled Particles VII, Proceedings
       of an International Symposium on Inhaled Particles Organized by the British Occupational
       Hygiene Society, 16-22 September 1991 (Eds. Dodgson, J. and McCallum, R.I.) Ann.
       Occup. Hyg. 189.
Swift, D.L, N. Montassier, P.K. Hopke, K.  Karpen-Hayes, Y-S. Cheng, Y.F. Su, H.C. Yeh, and
       J.C.  Strong (1992) Inspiratory Deposition of Ultrafine Particles in Human Nasal Replicate
       Casts, J. Aerosol Sci. 23:65-72.

The diesel particle is a chain aggregate and as such undergoes collapse of the fractal pattern
when exposed to  humidities typical of the respiratory tract (Weingartner et a/., Atmospheric
Environ. 31:2311-2327, 1997).


Chapter 5
Since there has been extensive studies of the individual effects of CO, SO2, and NOX, was there
any effort to compare diesel exhaust with comparable exposure to individual gaseous pollutants?
Are these gaseous species providing additive or multiplicative risks. Can we see evidence of

       There is evidence that the new engines produce smaller particles. What is the potential
implications of the smaller size based on what we know about the effects of uitrafine? We
cannot yet say anything definite, but it is critical to point out that the new emissions may be more
dangerous even at lower mass emission rates.

Chapter 6
How would the changes in emissions going from old engines to new engines affect the
estimation of the RfC. I understand there are no exposure response data to work with, but from
an understanding of the differences in the nature of the emissions can anything be said
qualitatively about the. impact of those changes on non-cancer health effects and hence on the
resulting regulatory considerations.  What evidence is there that DE is 3 times more toxic than
ambient PM2.5?  If it isn't, how can we have a NAAQS which is not protective of pubic health
with an adequate margin of safety?  If it is  more hazardous, make the case.

Chapter 10:
What is the take home message from the carbon core section? Since the evidence for effects is
only in rats and only under the overload conditions, does the carbon core really have any
importance for human cancer risk?  If this chapter is to make a mechanistic case for the
possibility that the core is important to cancer risk, it never makes the case and thus, does not
serve any purpose as written.

For the sorbed organic species, if we know certain compound that are carcinogens are there, are
they present in concentrations that would like produce lung cancer. There needs to be order of
magnitude estimates that suggest that the presence of these compounds is relevant to
estimating risk. Given low concentrations and low deposition, is enough material deposited to
provide a significant chance of tumorogenesis?  That estimate can be made for B(a)P and some
of the nitroarenes and again see if any of this discussion is relevant.

Diesel particles are NOT spheres with a uniform coating of organics.' They are chain aggregates.
The physical chemical behavior of chain aggregate aerosols seems to be a total mystery to the
report writers.  The presence of 5 to 10 nm primary particles in a chain leads to sorption at the
interfaces between primary particles.  The accumulated layer will accumulate with a negative
meniscus leading to a diminished vapor of the material above the particles (Calculations of the
Equilibrium Vapor Pressure of Water over Adhering 50--200-nm Spheres, Crouzet, Y. and Marlow, W.H.,
Aerosol Sci Tech. 22:43, 1995). In an analogous manner, dissolution of sorbed material on these
hydrophobic particles will be inhibited and the bioavailability will be diminished as was shown by the early
data of Siak et al, 1980 as shown in the presentation of Dr. Vostal.

Chapter 11,  page 11-22, line 19 says the animal data are "adequate" to support carcinogenesis.
Since only one species (rats) show an effect and only under overload conditions that are in no
way relevant to ambient exposures, ! do not see how the animal data can be thought of as
supportive of carcinogenic particularly given the lack of observed proliferation in the monkey
studies. Thus there are no animal data that can reasonably be extrapolated to human exposure
conditions that demonstrate that the exhaust is carcinogenic


Chapter 12, Figure 12-1, the x-axis is labelled "dose" when it is clearly "exposure".

This document needs a major effort to property reflect the nature of current engine emissions, an
estimation of what the changes in engine emissions are likely to be and what implications those
changes have for public health. Then the document could be used as part of the evaluation of
the ongoing mandated changes in diesel engine emissions.

                                                                     Art Upton

Review of EPA Draft Health Assessment Document for Diesel Emissions

Chapter 10 Metabolism and Mechanism of Action in Diesel Emission-Induced Carci.nogenesis

This chapter provides a detailed and appropriately inclusive review of the existing data on the
metabolism of diesel exhaust emissions and on the mechanisms that may be implicated in their
carcinogenic effects on the lung.

The chapter notes that although the respective roles of the particulate and organic fractions of
diesel emissions remain to be established, mechanisms can be envisioned through which either or
both fractions may be postulated to exert carcinogenic effects on cells of the respiratory tract

In the case of the particulate fraction, such mechanisms include the release of reactive oxygen
species from activated macrophage and leukocytes, resulting in genotoxic and cytotoxic effects on
neighboring cells, enhancement of the metabolic activation of procarcinogens, and depression of
immunological surveillance.

In the case of the organic fraction, which includes more than 100 carcinogenic or potentially
carcinogenic components many of which are mutagenic and have been shown to form DNA
adducts in diesel-exposed humans and laboratory animals, genotoxic mechanisms may be
postulated to be involved.

In discussing the metabolism and possible carcinogenic mechanisms of diesel emissions, the
chapter properly notes the limitations in the available data and the extent to which interpretation
of the data must therefore be qualified. Linkage to relevant issues in other chapters needs to  be
improved, however, especially as concerns the possibility of a threshold in the dose-response

Chapter 11.  Qualitative and Quantitative Evaluation of the Carcinogenicity of Diesel Engine

This chapter summarizes the weight of the evidence, presented in preceding chapters of the
document, for the carcinogenicity of diesel engine emissions, and it draws on the relevant human
and animal dose-response data to arrive at estimates of the carcinogenic potency of diesel
emissions for humans. The resulting upper-bound estimates of the risk of human lung cancer
attributable to lifelong exposure to diesel exhaust emissions range from 1 x 10"5 /ug/m3 to 200 x
10'5 /ug/m3-

The data and rationale underlying the risk  estimates, and the uncertainties inherent in the estimates
are presented clearly and documented appropriately; however, the possible existence of a
threshold in the dose-response deserves farther attention (ie, lower-bound of risk estimates).

Sverre Vedal


1.     The available evidence provides no support for a non-threshold hazard due
to  diesel exhaust exposure.

2.     a.  The mixing of the epidemiological and animal experimental approaches to
deriving upper and lower bounds for risk is admittedly somewhat strange.  This
reflects doubts about the validity of the occupational epidemiological data and the
applicability of these  data to the ambient exposure setting.  At this time, the large
range of risk suggests to me that we  are able to only make a qualitative statement
about lung cancer risk.

      b.  There is no basis for selecting a single estimate of cancer risk due to
ambient diesel exhaust exposure.  The epidemiological  data provide the best
estimates  of risk due to occupational  exposure concentrations, but these cannot
be extrapolated to  the ambient setting.

3.     I do not find that the selection  of uncertainty factors for the non-cancer
outcomes associated with diesel exhaust exposure is scientifically supportable.
First,  most exposure  concentrations associated with adverse effects, even
expressed as human  equivalent concentrations (HECs), that were  used in the
animal exposure experiments were several orders of magnitude higher than those
present in ambient air and several fold higher (and sometimes over an order of
magnitude higher) than those present in the occupational setting.  There is no
evidence that analogous effects will be present in settings with markedly lower
concentrations.  Second, whatever precedent may be present for  the practice of
using a default factor of 10 to derive  a reference concentration, it is difficult to
follow the logic in this setting.  Somehow this factor is intended to account for
uncertainty in  extrapolating from animals to humans and accounts for the very
susceptible subgroups in a population, although  only the latter was the justification
in  this case. In chapter 12, a factor of 3 was used to account for interspecies
differences, and a factor of 10 for susceptibles,  for a product of 30 and a  resultant
reference  concentration of 5 /ug/m3 (155/30).  While this number is consistent with'
the recently promulgated PM25 annual standard of 15 ^g/m3, this  is merely
happenstance.  The margin of safety  for the PM25 standard was minimal.  In short,
I do not favor  recommending a reference concentration for diesel  exhaust based
on non-cancer outcomes in addition to the PM standards.

Sverre Vedal


      The main issues that are relevant at this point to the epidemiological studies
on the association between diesel exhaust and lung cancer are: 1)  is there a
realistic possibility that the association is still due to residual confounding?  2) is
the time course (latency) between exposure and cancer onset observed in the
studies plausible?  My comments below in general argue that a  more thorough
discussion of the  controversies engendered by the epidemiological  findings is
needed in the Document.

1.    Concern over residual confounding  is exacerbated by the relative small size
of the estimate of effect.  With an effect estimate of only  1.3 (I would  therefore
take exception with the claim on p. 8-66 that the data on  diesel exhaust and  lung
cancer meet the Hill criterion of a strong effect estimate),  one must be  concerned
that inadequate control for confounding might well result in bias sufficient to
produce such an effect estimate.

2.    If so, what confounding factor(s) could do  thjs? Obviously,  cigarette
smoking is the factor of most concern.  Many studies, in particular the
retrospective cohort studies, did  not have the data to allow control for  cigarette
smoking.  However, studies that did have individual data on smoking, and in which
some control of smoking was therefore possible,  have produced effect  estimates
comparable in size to those in which control of smoking was not possible. This
suggests that smoking was not an important confounder in studies that did not
have smoking data, even though smoking was clearly a  strong predictor of lung

3.    Confounding due  to something other than cigarette  smoking is also possible.
However, it is not clear  what such a factor might be, since it would need to be a
reasonably strong risk factor for lung cancer to produce this consistent bias.
Concurrent exposure to  asbestos might be one possibility, but when it  has been
possible to account for asbestos  exposure, no change in effect  estimates has been
observed. It seems, then, that no compelling  argument  can be  put forward to
support residual confounding as an explanation for the observed effects.

4.    Since smoking is such a strong predictor of lung cancer,  another possibility
is inadequate specification of cigarette smoking that would still  allow residual
confounding due to smoking.  The Garshick case-control study  presented several
different approaches to specifying cigarette smoking, and  there  was no meaningful
change in the effect estimate resulting from changing the specification  of smoking.

5.    It could be  argued that, apart from incorrect specification of  the  effect  of
cigarette smoking in the statistical models, that more general  problems  with the

statistical models is possible and could result in erroneously observing an effect
due to diesel exhaust exposure when none in fact was present.  For example, the
proportional hazards assumption, required for valid inference based on the
proportional hazards model used in the cohort studies, may not be met.
Alternatively, misspecification of covariates other than smoking, or inclusion of
highly collinear model terms could result in wrong conclusions. Some insurance
against all of these potential criticisms is provided by the relative consistency
across studies.  Although the same mistake could potentially have been made in
all of the positive studies with all of the population samples, this seems  unlikely.
However, more flexible analytic techniques are now  available for analyzing  cohort
data, such  as those that address non-linear dose-response.  The fact that the
effect estimates from the case control studies are also consistent suggests that
alternative  modelling of the cohort studies is not going to change the findings
substantially.  A study from Germany presented in abstract form this year
(Brueske-Hohlfeld, Am J Respir Crit  Care Med,  1998} observed an odds ratio of
1.4, further adding to the consistency.

6.    Another interesting feature of the epidemioiogical data is the marked
consistency between studies of the  effect estimate size.  Although it is still
possible that consistent confounding by an unidentified factor could produce such
consistent  bias,  it seems unlikely. It could be argued that this consistency  is itself
evidence for confounding, since diesel exhaust concentration exposures must have
been substantially different across the many studies  (although this is not known),
which should have resulted in differences in effect estimates based only on crude
exposure categories.

7.    The  review of the epidemioiogical data in the Health Assessment Document
is reasonably complete, with at least one  exception.  It is not clear to me why
"hypothesis-generating" studies were not included in the review (p.  8-1).  While
such studies are less valuable in isolation when no confirmatory studies have been
performed, when a  number of subsequent studies have been performed, the
logical primacy of "hypothesis-confirming" studies is lost.

8.    The  issue of adequacy of length of the latency period from the time of onset
of diesel exhaust exposure to lung cancer death is potentially a thorny one. In
almost all of the studies, the latency period is relatively short.  For example, in the
Garshick case-control study (1987), cases were defined as lung cancer deaths in
1981-82 with  exposure deemed to have started in 1959, although some workers
would have had significant diesel exposure before that time. In the Garshick
cohort study (1988), only deaths that occurred through 1980 were included. A
latency period of slightly over 20 years is short when, for example, one compared
the latency period for cigarette  smoking.  I cannot recall the potential latency
period in the Lloyd report on lung cancer  in coke oven workers (1971) to
determine whether an equally short latency period was present in  that setting.  It
is plausible that an increased risk of lung  cancer with such a relatively short


latency period could be observed if what is being detected is the early portion of a
latency frequency distribution.  A similar finding would also likely occur when the
effect of cigarette smoking is examined.  Therefore, the relatively short latency
period  does not necessarily make the observed associations implausible.
However, further discussion of this topic is needed.

9.    Does the fact that the epidemiological data address the effect of diesel
exhaust and its components from "old" diesel engines significantly detract from
the value of the epidemiological studies? That is, are they relevant for addressing
the potential carcinogenic effects of today's ambient diesel exhaust?

10.   Finally/ the epidemiological findings and the toxicology findings seem to be
somewhat inconsistent.  This is a relatively unusual situation where  the toxicology
is suggesting that diesel exhaust is less toxic (in terms of carcinogenicity) than
does the epidemiology.  More typically the toxicology suggests carcinogenicity and
the epidemiology is either absent or reassuring.  Some potential reasons for this
anomaly could be residual confounding in the epidemiology studies  or inadequacy
of the toxicologic models. Some discussion of this is needed.  One possibility it
that the BaP comparative potency  procedure ignores any pathogenic effect due to
the particle component, an issue that is  discussed in the Document.

      In short, I am in agreement with the general  conclusion that the
epidemiological evidence is "highly suggestive of a  causal association between
lung cancer and occupational" diesel exhaust exposure.  I would like to see a more
thorough discussion of the potential for confounding and the latency issue.

Sverre Vedal


      Given that epidemiological data are available on the association between
lung cancer and diesel exhaust exposure, one would like to be able to make use of
primarily these data in evaluating lung cancer risk.  Apart from concern that the.
estimates of effects from these data are still due to confounding  bias, the primary
limitation in these data is the lack of good exposure data corresponding to the
periods of  relevant exposure.  Nevertheless, realistic estimates of the range of
mean exposures have been made,  which in turn were  used to estimate exposure-
response relationships.

      The suggested  upper hnunri risk based on an average exposure
concentration of  1 25  M9/m3 was 2 x 10"3 per Mg/m3 This appears to be the upper 95%
confidence bound for the  effect estimate  (see Table 11-1). If so, it is my opinion that this has little
meaning. If I understand  it correctly, this bound would then be  dependent on the precision of the
effect estimate from the study utilized. This seems quite arbitrary, and possibly irrelevant, for the
purpose of setting bounds for risk  For example, one could use  the'orecision based on a meta-
analysis and obtain a much different bound. If this bound is not  based on the upper 95%
confidence limit, but rather based on the  actual effect estimate, fhen I think it has some meaning.
Alternatively, an upper bound could be based on some uncertainties in the assumptions (other
than the precision of the effect estimate)  made to make the risk calculation.   Such a bound may
also have meaning, and if this is the case, then I might favor it (if I understood what uncertainties
were being incorporated in the  calculations). The expression of bounds needs to be clarified, as
evidenced by my obvious  confusion.  Chapter 12, the summary,  is clearer than this chapter in
identifying what the bounds are actually derived from. My overall preference is for an upper
bound of 1 x 10~3 per /ug/m3, which is the maximum likelihood estimate itself for an estimated
exposure of 125
       Regardless, use of occupational epidemiological data to set bounds to risk, either upper or
lower bounds, has a significant limitation in that these data may not be relevant to estimating risk.
at lower exposure concentrations.  That is, even though one can calculate bounds based on the
epidemiological data, these may have relevance only to settings experiencing relatively high diesel
exhaust concentrations.  It is not known whether the exposure-response relationship extends in a
linear manner to much lower exposure concentrations. To express risk as lifetime risk due to
exposure to 1 /ug/m3 of diesel paniculate matter based on the occupational data is therefore a big

       The estimated lower bound risk was not based on the epidemiological data,  in which case
one might have expected a bound that was lower than the upper bound by a factor of four (based
on the estimated upper exposure concentration of 500 ^g/m3), or an estimate that also
incorporated some uncertainties.  The bound based on such a calculation naively  should not be
less by a factor of four, since non-occupational (off work) exposures in the  occupational cohorts
would be expected to be relatively similar regardless of exposure at work. This would tend to

                                        A -34

make total exposure more similar. Nevertheless, the epidemiological data were not used to set
the lower bound.  This clearly reflects some discomfort in basing the risk estimates entirely on the
epidemiological data.  Of the alternatives to the epidemiological data to setting a lower bound, I
find those based* on human data more compelling than those based on animal data. Therefore, I
prefer the work using BaP as a dosimeter in which the maximum likelihood estimate of risk was
2.6 x 1CT6 per yug/m3.  It is argued that the main problem with the use of this estimate is that it
ignores any contribution of the insoluble carbon particle core of diesel particles to risk. The
animal data suggest that the carcinogenic effect of diesel exhaust is due to the particles, or
carcinogens bound to the particles. However, this mechanism may only be relevant under
overload conditions. But there is evidence  of human carcinogenicity of combustion products
without requiring a carbon core particle. Therefore I am not clear that the dosimeter approach is
irrelevant. In my  opinion, given the human data that are available, the assumptions required to
apply the animal data result in relatively insecure estimates of risk.

       Having said that, my preferred upper bound lung cancer risk is 400-fold greater than my
preferred lower bound (1 x 10'3 vs. 2.6 x 10"6 per Mg/m3>, which greatly diminishes the utility of
these risk estimates.  I therefore do not feel there is much merit to attempting a quantitative risk
assessment for diesel exhaust at this time.  Extrapolation to ambient concentrations is also

From:         '
Date:         5/4/98 17:24
Subject:       diesel comments


Comments by Warren H. White, 5/4/98


This is easily the least accessible document I have yet had to review for
CAS AC.  If it is intended to communicate anything to physical and mathematical
scientists like me, it must offer:
a)  an introduction that lays out the specific policy questions confronting
    the Agency, the questions this document is presumably intended to
b)  a road map outlining the structure of the discussion to follow,
    explaining to a non-MD how various physiological perspectives relate to
    each other.

It would have helped me enormously to have read section 12 2.2 (Toxicity Mode
of Action) at the beginning rather than the end. By this time I had more or
less figured out what AMs do, but here - finally — it was spelled right out,
in clear and simple language.  This is an example of the kind of writing
needed earlier in the document.

The "plain-language over, •sw of key information" (section  12 5) is very
disappointing.  I have a hard time extracting any information of the response
to "How does exposure affect human health and how certain are we about these
effects?" (P  12-33), for example. "Another aspect of the exposure event is to
realize that in a general sense TOTAL CUMULATIVE EXPOSURE AND THE RATE AT
UNIQUE TO DE." [P 12-34, L 1-4: emphasis in original]  Is this anything other
than double-talk? What does it mean to say (P 12-37, L 33-34) "Though diesel
particulates are associated with a carcinogenic hazard, this is not indicated,
per se, for ambient PM exposure."


Post-emission atmospheric transformation is explicitly excluded from
consideration in this document, but "primary" emissions are in fact a matter


of measurement convention. Diesel emissions in a hot exhaust pipe differ from
diesel emissions diluted ten-fold, and slowly diluted emissions differ from
diesel emissions almost instantly diluted a thousand-fold. The descriptions
of the inhalation studies in section 7.2 recognise the importance of these
methodological details, but the discussion of primary emissions in section 2.3
does not. It is only in section 2.4 that we get a "by-the-way" mention of
typical sampling conditions, and an acknowledgment that, for example "more
particles in the [ultrafine] may be' expected under typical roadway
conditions."  Discussions of the distinction between carbon core and organic
coating are also difficult to interpret without some understanding of sampling
temperature and concentration history.

A significant new set of diesel exhaust measurements, specifically designed to
support CMB modeling, is available from the Northern Front Range Air Quality
Study (NFRAQS). The final report is available at HTTP:
//charon.CIRA. colostate. edu.

To:            DCFCHO1 .DCFCHPO1 (FLAAK-ROBERT),DCWICO 1 .IN("jmauder.
Date:          5/15/98 19:09
Subject:       detailed comments on diesel document

Detailed comments on Diesel Health Assessment, EPA 600/8-90 057C

by Warren H. White

page 2-7; lines 9,10: In aerosol science, "agglomeration" normally refers to
aggregation between particles: it is thus inherently not a "gas-to-particle"

2-8; 1  What are sulfate emissions as a fraction of fuel sulfur9

2-9; 3: The distinction between gaseous and particulate "primary" emissions
is operational, a matter of how these emissions are measured. These details
are scattered about: e.g., "collected on a filtering medium at a temperature
of 52  deg C or less" (2-13), and "a dilution factor of 10 is typical of many
dilution tunnels" (2-26).  They need to be collected here, and their
implications for relevance to ambient particle characteristics discussed.

2-13;  29: As an example of the preceding comment, the "typical" size
distribution in Figure 2-1 needs some experimental context:  how was it
determined? Alternatively, it might be presented as a "conceptual model" of
such a distribution.

2-14;  1. The assertion that about 70% of TC is EC in Table 2-4 is supported
by the Pierson and Brachaczek data, but the Williams et al. data put the EC
fraction at 42% - 64%. And what's the point of "so-called"?

2-58,  6,7:  What is the point of putting all this work into a health
assessment that is designed to give the Agency no guidance as to whether it
should welcome or fear the new technology?

2-59;  7+: This discussion should note the recent NFRAQS study.  See

12-2;  6: The exhaust particles are formed through the AGGREGATION of smaller
particles: "condensation" is a gas-to-particle process.

12-2;  7: Change "They" to "The emitted particles".

12-2;  10: Organic compounds (the whole) should  be distinguished from organic
carbon (a part).


12-2; 22: HD and off-road engines "have the largest U.S. paniculate
emissions" in what sense?  Are we speaking in the aggregate over the inventory
category, or on some per-mile or per-engine basis?

12-2; 29: The magnitude of extra hazard is "unknown", it might be discernable
if someone looked hard enough.

12-3; 13: Change "shorter and longer" to "shorter to longer".

12-8; 18: Change "rigid" to "strict".

12-26;  11,12: The claim that  1 E-5 "seems to be a floor of all estimates and
therefore, establishes the low  end for a range of risks" is clearly
inconsistent  with the earlier (12-22; 22) characterization of 1 E-5 as "The
95% upper bound" of the biomarker estimate, and with the later observation
"that as these [sic] all of these risks are upper bound, the true risk is
unlikely to be greater and may well be less."

12-32;  4,5:  Change to "more than 75% of the material carried by particles
smaller than 1 um."

12-33;  4: Note that 200 ug/m3 is over three times the peak 24h health
standard for ambient fine particles:  should we worry every time we smell this
familiar odor? Maybe not, but it's a natural question to ask. The fact that
we smell diesel  for just a few  minutes at a time isn't wholly reassuring,
given the speed of olfactory fatigue.

12-33;  19,20: Plain language is not the same as wordiness. This is a good
example:  Replace "The  inhale-exhale pattern of breathing results in some
exhalation of the particles and gases;" by "Some of the particles and gases
are exhaled;".

12-33;  32+:  This whole paragraph  seems meaningless blather

12-34;  12,13: The preceding  paragraph assumed the reader didn't know the
difference between "acute" and "chronic", but here you expect him to
appreciate a subtle distinction between "toxic" and "hazardous"  And  "Taken
individually, both" is hardly "plain language".

12-36;  30: If "the true estimate is undefinable", then how "could [it] be
much lower"? You  mean "undetermined".

12-36;  32,33: "The  risk range is thought to bracket the upper limits of
possible risk" is hardly plain language.  As evident from the discussion of


12-26 above, the very concept of a range of upper bounds invites confusion.

12-37; 6: Change "the probability of cancer" to "the increase in lifetime
probability of cancer", and move up in paragraph so reader understands what
the numbers at the bottom of 12-36 mean.

12-37; 28,29: Change to "Diesel particles are small: particles of diameter
less than 1 um carry more than 75% of the total mass,".

12-37; 33,34: If "diesel particulates are associated with a carcinogenic
hazard" and are a ubiquitous component of ambient PM, must we not consider
ambient PM a carcinogenic hazard?

 Chapter 2

 Two main problems seem to be evident in this chapter. The first is that it seems out of date. Although
 it is stated that this chapter has been updated this is not apparent.  Secondly, it is stated (page 2-8,
 line 33) that "detailed chemical characterization of diesel engine emissions was performed in the late
 1970s and early 1980s". Since new technology has occurred since that time, it is to be presumed that
 the composition of the emissions will have changed. This is important even if one does not believe
 that these chemicals have a role in cancer induction in humans. Several studies have cited the
 possibility of PAHs  and other chemicals to  alter  immunological effects in vitro and in vivo.  A
 qualitative or quantitative change in chemical composition would, therefore, be of considerable
Chapter 5

In this chapter and throughout the document, it seems incredible that virtually no mention is made
of the extensive work done examining the role of particulates (PM2.5; PM10, etc.) on cancer and
non-cancer health effects. For example: epidemiological studies have linked particulate pollution and
asthma/bronchitis/respiratory disease. From the studies presented here and others showing a role for
carbon black in the inflammatory response and in lesion formation, it would seem that general
inflammatory changes, release of inflammatory mediators and lesion formation is mainly due to the
particulate nature of diesel exhaust.  Immunological changes, however, seem to result from other
factors such as PAH although particles may still be needed to deposit them in the right location..
Although it is understandable that the particulate studies do not focus on diesel  exhaust and may be
due to other factors, it should be noted that DEPM is considered one of the main  contributors of
urban particulate matter (from  30 to 80%, depending on the study).  To ignore the  PM studies
completely is such an obvious omission that it suggests perhaps a political/policy decision rather than
a scientific one.

There is an obvious discrepancy between the animal studies that indicate that diesel exhaust can lead
to chronic inflammation, lesions and hypertension and epidemiological studies that show little or no
link between diesel exhaust exposure and disease. What are the reasons for this?

a) Have the correct human and epidemiological studies not been done? Do we simply need better,
more controlled studies?

b) Are the epidemiological studies not examining the correct endpoint? For example,  too few studies
seem to focus on asthma, atopy or emphysema, the logical endpoints suggested by the animal studies.

c) Is there no connection between human and animal studies? That is, do studies in rodents not
translate into the human condition.

From the studies presented here, it would seem that the answer lies in a) and b). Regardless of the
reason,  the epidemiological studies are not reliable or consistent enough to be used as measures to
calculate an inhalation reference concentration.
Many of the studies regarding asthma/allergic sensitization or induction by diesel exhaust have been
overlooked. The knowledge that diesel exhaust particles will act as an adjuvant in mice co-immunized
with a bystander protein was shown by Muranaka et al., in the  1970s. Since this time several other
groups have confirmed this. In vitro and in vivo experiments have shown that the PAH associated
with diesel exhaust have similar properties (see Suzuki et al; Tsien et al). To present these studies
as Overy recentO is somewhat misleading.
Chapter 6

This chapter was obviously written before the uncertainty factor of 3 was reintroduced. Chapter 12
and this chapter must be rewritten to synchronize with one another. An obvious concern is that the
studies used to derive the NOAEL are all from studies using rodents, predominantly rats. The work
performed on monkeys (Lewis 1986,89) show no effects on monkeys. Is it reasonable to assume that
this  is due to lifespan. There is an obvious need for more studies in higher mammals to verify this
question. However, with the available data, I believe that the Rfc calculation is reasonable with one
exception.  The use of an uncertainty value of 3 to account for interspecies sensitivity needs to be
justified, there is no evidence that humans are more sensitive than rats. In the letter by Dr. Farland
it is suggested that allergic hypersensitivity resulting from DEPM exposure occurs in humans but not
rats, however, this is not correct since the issue  has not been studied in rats  Since the 19700s
experimental studies, exposing different strains of mice to diesel exhaust, have shown that DEP will
act as an adjuvant, that is induce allergic antibody formation to bystander allergen. This effect will
vary between mouse strains.
In regards to subpopulations with increased susceptibility, it  would be predicted that at least for
asthma and allergy, allergic/asthmatic individuals would be more at risk.  Additionally, h; should be
noted that in controlled chamber studies on asthmatics only a subpopulation reproducibly showed a
decrease in lung function when exposed to secondhand smoke (which  includes a high concentration
of particulate matter). A similar group of OresponsiveO individuals may exist in response to diesel
Definition of the benchmark concentration is an obvious critical question. How this was chosen needs
to be better justified.

Chapter 11

There are two separate issues in this chapter: first, whether the evidence points to a qualitative
evaluation of diesel emissions as a potential carcinogen and second whether  it is possible to quantify
this risk. Overall the diesel assessment performs an adequate job in defining the target organ as the
lung.  However, estimation  of cancer hazards and risk is hampered by the paucity of reliable
mechanistic studies.  Therefore, despite the large number of published studies, the exact mechanism
(if any) between lung damage,  carcinogenicity and diesel exhaust  exposure  is still  a matter of

speculation.  This becomes important when one needs to choose between a nonthreshold or threshold
dose hazard. There seems to be little or no proof that the particle overload effect occurs in humans
or indeed in any other species other than rats; the absence of lung cancer in pneumoconiosis afflicted
coal miners and particle  deposition patterns certainly  argues  against  this.  Yet, the human
epidemiological data suggest that occupational exposures at concentrations that would in any case
occur at concentrations too low for particle overload will increase the risk for lung cancer  Although
there are inherent problems with these human studies in terms of defining exposure assessment and
eliminating other factors, nevertheless there is agreement between nearly all studies that diesel exhaust
is associated with an increased risk of lung cancer. Therefore, the qualitative statement that diesel
engine emissions are probable human carcinogens is accurate
The question of quantifying this risk is more problematic. It seems clear from carcinogencity studies
in the rat using carbon black and other particulate compounds that the panicle overload phenomenon
is an important toxic mode-of-action in the rat. Evidence for a similar mechanism in other species or
at low dose levels in unconvincing. The lack of reproducibility in other species is indeed worrying and
begs the question of whether these results are species-specific. The use of this model for establishing
low dose human risk estimates is, therefore, suspect. Therefore, it is important to have methods other
than just extrapolation from the rat model to determine risk ranges. The use of the model postulated
by  Chen and Oberdorster is welcome  as it has the important advantage of incorporating the
carcinogenic  potential of both the carbon particle and the associated organics (e.g. PAH, quinones
etc.). However, the data used by this model is questionable. Indeed, no one approach seems more
valid than another. The use of the human epidemiological data suffers from a range of problems
including control  of other factors,  lack of biomonitoring methods for detecting tissue  doses of
exhaust, lack of adequate measurement of exact biological effect and an absence of a dose-response
effect.  The comparative potency method makes significant assumptions that do not necessarily hold
true. The use of a biomarker would seem useful, except for the difficulty in obtaining a reliable and
unique biomarker for exposure.  B(a)P will vary from different diesel emissions and additionally may
be found in tobacco smoke and other sources. No one methodology seems superior over the others
and any preference  would stem primarily from the bias of the reviewer. As a consequence, one can

either reject all these approaches or accept their limitations and compromise with a range of risk
estimates.  To my mind, the limitations are so great that one should take the first option and not
attempt to quantify the cancer risks at all.


To: Robert Flaak, Designated Federal Officer, CASAC

From: Eric Garshick, M.D., M.O.H.
Comments on EPA Health Assessment Document for Diesel Emissions

Chapter 2

This chapter should be updated in order to allow the reader to understand how diesel
exhaust has changed over the years. In order to give the reader a sense of typical
exposures, exposure levels in different occupational settings might be included and
contrasted to estimated environmental levels. Relating occupational exposures would
be useful because the epidemiologic studies have been conducted in occupational
settings. Recent work by Cass and coworkers on source apportionment might be
included (see Health Effects Institute 1995 Report).

Chapter 5

Page 5-1, lines 17-19:  There  is a study published in abstract form in 1996 that reported
the results of bronchial biopsies obtained in normal volunteers following diesel
exposure. A influx of polymorphonuclear leukocytes and an upregulation of endothelial
adhesion molecules were reported (Salvi SS; Eur Resp J 9(S23): 415S, 1996. If this
study has been published subsequently, then it might be useful in understanding the
relationship between exposure in humans and subtle inflammatory' changes.

Page 5-5, bottom of page: Three cases of asthma are listed here under "Immunologic
effects". However, these cases of asthma were not caused by an caused by an
immunologic mechanism, but  occurred after a short-term exposure to high  levels of
exhaust (see later discussion  of this).  It would also be useful to update this chapter on
the behavior of diesel particles acting as an adjutant for pollen.

Page 5-79, top page: These comments seem speculative.

Page 5-80, lines 30-36: The rat data do not seem relevant to assess human non-
cancer effects since the effects notee occur in conditions of particle. The statement
made in lines 3-31 does not seem accurate.

Page 5-82: The short term effects of diesel  exhaust exposure can be summarized as
an increase in respiratory symptoms characterized by cough, phlegm,  and  wheeze. In
some studies, pulmonary function decrements have been noted across a work shift.
Long term effects on pulmonary function and respiratory symptoms are unknown
because of only studying active workers and lack of longitudinal data.

Chapter 8

Introduction: Page 8-1, Line  10: In the railroad industry, the change from steam to
diesel locomotives generally started after World War II such that by 1946, 10% of the
locomotives in service were diesel, by 1952 55% were diesel, and by 1959, 95% of the
railroads were diesel. By stating that the transition to diesel started in 1935 implies that


 most of the population was exposed for many more years than actually occurred.
 Since many epidemiologic studies were done in truck drivers, it would be useful to state
 that the trucking industry changed to diesel trucks by the 1960's with some companies
 not using mostly diesel vehicles until the late 1960's. A point can be made in this
 paragraph is that the date diesel engines were introduced in an occupational setting will
 influence when an increase in lung cancer might occur if diesel exposure is
 responsible. Therefore, it is reasonable to exclude studies where the population had a
 very short duration of exposure and follow-up  Since lung cancer has a long latency, it
 would be worth noting that a weakness of the diesel literature in general is lack of a
 cohort of workers  with a very long duration of follow-up and well-defined exposure (>20
 to 30 years)

 Page 8-1, Lines 17-18: The results of the studies not specifically discussed due to a
 variety of limitations might be included in a table. Some of these studies (Siemiatycki et
 al., 1988, Swanson et al. 1993) have been included in other reviews, such as the 1995
 Health Effects Institute (HEI) report and the meta-analysis written by Bhatia et al.,
 1998. Studies  by Guberan et al., 1992 in professional  drivers in Geneva and by
 Gustafsson et al., 1986 could also be included in the chapter. Two additional studies
 that can be added to the papers discussed are a study of truck drivers in Iceland
 (Rafnsson and Gunnardottir, 1991) and a study in Swiss professional drivers (Pfluger
 and Minder, 1994). These studies support the general conclusions of the chapter.

 Waller (1981): Page 8-2, Lines 11-12: These lines refer to the study by Waller, 1981. In
 this study, employed London transport workers aged 45 to 64 at any time between
 1950 and 1974 were included in the cohort and followed until death or retirement,
 whichever occurred first. The sentence in the document suggests that 20,000
 employees were followed for 25 years which was not true.  Later, on page 8-3, lines 5-7
 it is asked whether the ages included refer to the entire period or the mid-point of the
 25 years period, or refer to the time period between 1950 and 1964. The author of the
 article states the study was limited to men aged 45-64 during the period 1950-1974 so
 these lines can be rewritten to reflect this. The use of 1964 in these sentences should
 be replaced with 1974,

 Page 8-2, Lines 22-23: The major finding of the study  is summarized in this one line,
 and can be written to give additional detail. The engineers in the bus garages had the
 highest  mortality ratio for lung cancer (90%) compared to bus drivers and conductors
 (75%) and other engineers in a more central repair facility (66%). The bus garage
 engineers would have had the greatest exposure to diesel exhaust, but motormen and
 guards without bus-related diesel exposure had a mortality ratio of 87%, similar to the
 mortality ratio obtained for the bus garage workers.

 Howe et al. (1983):  Pages 8-3 to 8-5: A limitation of this study not discussed is that
 only deaths among retired workers were included in the cohort. It is not clear if  a worker
who developed lung cancer while working would be included in the cohort if disability
 rather than retirement benefits were requested. Therefore, it is possible that workers
with the most diesel exposure were excluded from the study. The details of how the
exposure categories were generated were not presented. Nevertheless,  an elevated
risk of dying of lung cancer in workers probably exposed to diesel exhaust was
obtained. Although lines 1 and 2 on page 8-4 indicate  that  SMR's were calculated in
the usual way,  an  internal comparison among exposure groups was also  presented and
not indicated here. The potential confounding factor not discussed in this paper or in


this section is exhaust fumes from coal-fired engines. However, death due to lung
cancer was not elevated in workers who retired before 1950 who would have had
primarily exposure to coal combustion products. The limitations noted are those that
can accompany most cohort studies and the effects of these limitations should be
discussed. Although it is possible that cause of death may be miscoded, lung cancer
tends to be accurately coded in death records. If other cancers are coded as lung
cancer, it would diminish the ability of the study to detect an effect of exposure.
Although smoking information is not available, the use of an internal comparison group
would tend to minimize potential confounding due to differences in smoking among
exposure groups. A generic discussion of the effect of smoking and the use of death
certificates in the diesel literature might be included in the chapter.

Rushton et al. (1983):  Pages 8-5 to 8-6: Comments about this study can be limited to
noting there is insufficient information regarding exposure (even based on job title) to
draw any conclusions about this study and it seems that the duration of exposure of the
cohort was very short. All workers with at least 1 year of work were included, and the
duration of "follow-up" was a mean of 5.9 years. It was not clear from reading the paper
whether this referred to time of entry into the cohort or time since the first year of work.

Wong et al. (1985):  The general discussion about the study seems adequate. The
summary can be improved by eliminating such general statements  such as in line 13-
14: "One has to make do with job histories, which  provide limited information on
exposure level." The extent that this is true depends on the quality of the job history
and how exposure in each job is characterized and should be assessed separately for
each study. Although the SMR for lung cancer increased with length of union
membership, the greatest limitation in this study is the lack of smoking histories since
long term union employees may have been the heaviest smokers when compared to
the U.S. general population.

Boffetta and Stellman (1988):  Page 8-11,  Line 7: The number of males should be
461,981 with known smoking habits, not 46,981 males. However, later in the paragraph
it is stated that there were data  on 476,648 subjects. These numbers should be
reconciled. A strength of this study is the ability to adjust for cigarette smoking, and
once this was done, a residual effect of diesel exposure was still noted, with only a
modest decrease in  relative risk from 1.41 (95%CI 1.19-1.66) to 1.31  (95% Cl 1.10-
1.54). It would be reasonable to make this point so that estimates of the effect of
smoking corrected and crude rates can be compared.

Garshick et al. (1988): Page 8-15, lines 7-9 are duplicates of lines 29-31 on page 8-
14. Although the relationship between years of exposure was reported as written in this
document, we now appreciate that the relationship (slope) between years of exposure,
when adjusting for attained age (rather than age at entry into the cohort) and calendar
year, is flat to negative depending on modeling methods. These comments have been
made previously to EPA regarding these results. In the determination of the relationship
between years of exposure starting in 1959, the slope is influenced by the workers with
the longest duration of exposure and occur in a limited number of cells where there is
under ascertainment of death. In the years 1977-1980 we now recognize that death
ascertainment was not complete with 20% to 70% missing deaths depending on the
year. The use of years of exposure starting in 1959 also excludes exposure before
1959. Before 1959 there could have been up to 10 years or more of additional
exposure by some members of the cohort at a time when the intensity of exposure was


 likely to highest. When analysis of this cohort based on job title in 1959 is limited to
 deaths occurring through 1976, the youngest workers still have the greatest risk of
 dying of lung cancer.

 Gustavsson et al. (1990): Page 8-17, lines 8-9: The statement that there no dose
 response relationship is confusing. Later, a dose response relationship is presented
 based on an exposure index based on job duties and work practices.  In line 12,
 reference is made to a weighted logistic regression analysis; more detail can be given
 to clarify this since it is not clear what is meant by this statement, or the line can be
 deleted and the overall results presented. In lines 25-26, the comment is made that
 cigarette smoking might confound these results. I think it is unlikely because the cases
 and controls are from the same cohort, and thus  the distribution of smoking is not likely
 to differ between workers exposed and unexposed to diesel exhaust.

 Hansen (1993): Page 8-18, lines 26-31: The statement is made that the lack of
 smoking data and a 36% rural population confound the lung cancer results. However,
 the authors present an analysis that make this unlikely to be true and should be
 mentioned.  Although the mortality follow-up of this cohort is short, just 10 years
 between 1970 and 1980, the utility of this study in assessing the effects of using diesel
 trucks depends on when the Danish trucking industry started using large numbers of
 diesel trucks. Unfortunately, this is not clearly presented  but is suggested to be in the
 1950's.  This uncertainty should be more clearly stated.

 Tables, pages 8-19 to 8-21: Can  be modified to  more clearly represent the strengths
 and weaknesses of the studies based on the above comments.

 Williams et al. (1977): Pages 8-22 to 8:23: The main weakness of this study is the lack
 of information linking job title to actual diesel exhaust exposure, and this should be

 Hall and Wynder (1984): Page 8-24, lines 31-32: It is stated that since job titles were
 not validated with work records that recall bias could influence the results. In this study,
 occupational exposure to diesel exhaust was based on job title rather on self-report of
 exposure. Based on job title, each subject was classified in very broad exposure
 categories that would have made a true effect of  exposure harder to detect rather than
 overestimating the effects of exposure. It would be unlikely that recall bias would affect
 the report of job title in this setting. The study summary can be modified to reflect this.

 Damber and Larsson (1987): Page 8-25, line 33. The odds ratio using dead controls
 for professional drivers with 20 or more years of employment was  1.2 with a 95% Cl of
 0.9-2.6, not 0.6-2.2. This is smoking adjusted, and is in the range that  might be
 expected given the lack of detail in exposure history. On  page 8-26, line 7, the results
 are summarized as simply saying the study did not find an increased risk of lung cancer
 in professional drivers and should  be modified to  reflect these  results.

 Garshick et al (1987): Page 8-28, lines 12-13: It  is written that this case-control study
was designed to evaluate the feasibility of conducting a large  retrospective cohort study
which was not the case. Our group conducted an earlier pilot study to examine the
feasibility. Page 8-30, line 3: Instead of post, insert "past".

 Hayes et al. (1989): Page 8-33, line 28-29: The sentence ending with the words


...exposures to other lung cancers...should be fixed. In line' 31, the effect of job
misclassification on the estimates of effect should be discussed, not just that
misclassification could occur. The effect would be to make an effect of exposure harder
to detect since broad job categories are used.

Steenland et al. (1990): Page 8-35, line 19-20. Although there were no exposure data
available for the period of the study, an industrial hygiene survey was done later
(Zaebst et al., 1991) that indicated that the mechanics had the highest level of
exposure,  and the short haul and long haul drivers had similar exposure levels
(approximately 25 ng/m3). The odds ratio for the mechanics was 1.69 (95% Cl=0.92-
3.09), whereas the odds ratios for the long haul drivers was 1.31 (95% Cl=0.81-2.11),
and for the short haul drivers was 1.27 (0.83-1.93). The long haul drivers drove mainly
diesel trucks, whereas the short haul drivers drove gas powered trucks. The similarity in
odds ratios and exposure levels between the short haul and long haul drivers suggests
that much  of the driver's exposures comes from the roadway.  These results can be
added to the section, and serve to link the air pollution and diesel literature.

Boffetta et al. (1990):  Page 8-36, lines 20-23. The comment that no effort was made to
verify exposure is difficult to justify here because it was a hospital-based case-control
study. Verifying exposure in this setting would have been nearly impossible.

Emmelin et  al. (1993): Page 8-38, line 19. change the word "futile" to imprecise.
Bladder Cancer Section: 7 studies are included in this section; studies by Silverman
et al. (1983 and 1986), Vineis and Magnani (1985), Risch et al. (1988) are excluded,
although the overall  conclusions of the  chapter will not change. These studies relate
work as a truck driver to bladder cancer risk.

Discussion: page 8-59, bottom of the page,  discussion about studies in miners.  There
are 2 studies in miners that have recently become available, one from Australia and
one from Germany that was published in a German journal that had an odds ratio of 1.9
(95% Cl= 0.6-6.2) in a  cohort of 5536 Potash miners. These studies may be available
from the Health Effects Institute.

Page 8-62: Discussion about exposure-response in the Garshick et al. (1988)
retrospective cohort study: The evidence for dose-response comes from the
observation  that the younger workers with potentially the longest duration of exposure
had the greatest risk of dying of lung cancer. The current line in the document
comments that the study found increasing risk with increasing duration of exposure.
This statement can be made more precise regarding the findings of the study.

Page 8-63, line 2: The study by Lerchen et al. (1987) was also limited because very few
in the cohort reported exposure,  not only due to  the limitations noted in the paragraph
about the study.

Page 8-63, line 8: The study by Boffeta et al. (1990) reported  a slightly elevated risk of
lung cancer based on  self-reported exposure, but not based on occupational
classification. The reasons for this include the use of very broad exposure categories
that would make it difficult to observe a relationship between exposure and lung
cancer, and there were few subjects with exposure. The comments on lines 10-11  "it is
interesting to note that he leading risk factor for lung cancer is cigarette smoking" and


'The exposure was not measured"  does not fit into the structure of the paragraph and
detracts from the discussion.

 Relevant Methodologic Issues

Page 8-65: This section discusses relevant methodologic  issues. The section on
cigarette smoking should include a discussion that although cigarette smoking is a
cause of lung cancer, it is only a confounder if there is differential smoking rates among
workers exposed and unexposed to diesel exhaust. If a cohort includes workers of
similar socioeconomic class, it is unlikely there will be  substantial confounding by
cigarette smoking.

Other limitations relate to exposure assessment, both in intensity and knowing how
long an individual was actually exposed. Given when diesel engines were introduced in
the US trucking and railroad industry, there is no study with large numbers of workers
with well-defined exposure and mortality ascertainment over 20 to 25 years, although
the railroad worker studies come closet to achieving this. It might be useful to state
what is lacking in the human epidemiologic literature regarding the relationship between
human exposure to diese! and lung cancer. What would be required to permit EPA to
declare diesel exhaust a definite human lung carcinogen? This would help guide
recommendations for future research.

Page 8-69. line 9: There is a comment about a lack of an elevated risk of bladder
cancer in cohort studies. In this document, no cohort study of bladder cancer is
reviewed in this document.  In the HEI document, several cohort studies are noted
without an increase in risk noted.  A major limitation in the bladder cancer literature is
knowing the type of vehicle exhaust fumes the transportation workers studied were
exposed to.

 Chapter 11

Page 11-1, bottom paragraph: The results of the epidemiologic studies can be
summarized in a more concise fashion, with the limitations listed and their impact on
the estimates of risk rather than just stating there are "major" limitations in line 33). A
summary might include the points that the literature is unlikely to be confounded by
cigarette smoke; the effect of using job title would be to underestimate the effect of
exposure. Measurements of actual exposure are lacking, as does the study of a cohort
with years of well-defined exposure and follow-up, given the long latency typically
observed in lung cancer.

1986 Guidelines, page 11-1:1 agree with the wording of how carcinogenicity is
described given the constraints of these guidelines.

1996 Guidelines, page 11-3:1 agree with the qualitative statement made.  I think that
confounding factors are not the major concern in the diesel literature, but relate to the
assessment of exposure duration and intensity, as well as the lack of studies with long
term follow-up of exposed workers. Given the time period of the introduction of diesel
engines in the railroad and trucking industries (late 1950's-1960's), until recently it has
not been possible to design such studies.

 Page 11-4, lines 2-3: The statement is made that the  results of inhalation experiments


in mice were equivocal, but in the previous section (page 11-2, lines 10-11) positive
results in two experiments in mice are noted.  This should be clarified.

Page 11-10, lines 16: 1945 should read 1959.

Page 11-10, lines 18-21: As noted earlier, the relationship between years of exposure
starting in 1959 and lung cancer mortality in the cohort study was influenced by
additional analysis that adjusted for age at death rather than age at entry into the
cohort, and by the missing deaths after 1976. The relationship might also be influenced
by considering exposure before 1959. In the analysis presented in the p~aper, exposure
pre-1959 was not considered.

Page 11-10, line 27: The comment is made that the differences between the results
obtained by Dr. Dawson and Dr. Crump cannot yet be explained. A summary might
indicate that the differences are related to how background exposure is used in the
modeling, the use of different markers of exposure in some models, and how age and
calendar year are handled in the regression results.

Page 11-10, bottom of page:  The mean values obtained by Woskie et at. ranged from
roughly 50 ng/m3 to 100 ng/m3 for the train riders and 120 ng/m3 to 160 |j.g/m3 for
the shop workers as presented in the second of the exposure papers.  Exposure varied
based on the period of exposure since the earlier engines historically weren't as dean
as later engines.  The equipment on the four smaller railroads that were sampled was
weighted towards early equipment (1950's) rather than later generation diesel engines
(built in the 1960's and 1970's). We estimate that the actual historic values of exposure
were likely 2 to 4 times the values we measured, and there is some indication the
historic values could have been 10 times as great. Part of the uncertainty of performing
a risk assessment due to estimating  these historical levels and applying them in any
model selected. The relationship between personal exposure to diesel exhaust and
lung cancer occurrence in the railroad worker database is therefore uncertain.

The railroads gradually replaced steam locomotives with diesel locomotives throughout
the 1950's. The date that this happened for each worker was not available, although
industry average data are known. Some workers may have used diesel equipment
starting in the late 1940's and early 1950's, and others 10 years later. Due to this
uncertainty, the regression results using years of exposure was based on exposure
starting in 1959 because 95% of the locomotives in service were diesel by 1959.
Therefore, the use of the regression results from the case-control study doesn't take
into account pre-1959 exposure that was more intense. The regression coefficient
overestimates the risk per year.

Since we know which railroad each worker last worked on and railroad employment
was very stable, it is possible to refine estimates of pre-1959 exposure by using
information from railroad rosters to estimate the conversion date  from steam to diesel
for each worker, and estimate the conversion from first to second generation engines.
In my opinion, analyses could be done that consider the uncertainty in past exposures.
These analyses could be done for both the case-control and retrospective cohort study,
and would give insight into the uncertainties in using these data for risk assessment
rather than using one slope based on the case-control study in this document. In the
cohort study, this uncertainty would be reduced by obtaining additional information on
deaths after 1976 given the short duration of follow-up at present time (1959-1976). In


 my opinion, analysis of the current retrospective cohort database indicates that the
 exposure-response relationship cannot be described by a single slope.

 If we use these data to estimate the risk per ^g of inhaling ambient diesel over a
 lifetime it also indicates that we are confident that the type of exhaust generated is
 similar to more recent exposure. Additional research is needed in this area.these data
 seem to be more relevant to estimating risk for workers with higher level occupational
 exposures. Use of these data are also limited by not knowing if we are using the right
 marker of exposure since we do not know the biologic mechanism of how diesel
 exposure results in lung cancer in humans. These points need to be made qualifying
 the results  of a quantitative risk assessment.

 If a quantitative risk assessment is needed to regulate diesel exhaust, particularly given
 new technology, it should be acknowledged that the best information would be
 obtained from a study that better able to link personal exposure to health outcome, and
 is designed to determine a dose-response relationship. Given the uncertainties in  the
 risk assessment regarding extrapolation to low dose exposure at this time, I am
 concerned  that any number calculated will be taken out of context.

 Page 11-11, line 5: It is not clear what is meant by maximum likelihood estimates of
 mortality. In looking at the calculations  done by Dr.  McClellan in the original reference,
 it is  not clear to me how the values used in the current EPA report based on the
 human data are derived and these calculations should be shown.

 Page 11-11: Use of the animal bioassay data. My opinion is that the animal bioassay
 data should not be used to estimate human risk, particularly at low exposures.

 Page 11-22: The conclusion that diesel is a probable human carcinogen is valid.

 Chapter 12

 This chapter is largely based on previous chapters, and its content reflects the
 knowledge  and uncertainties of the past chapters. In general, it is poorly written and
 contains material that is speculative regarding mechanisms and health outcomes.  It
 needs to be rewritten once the other chapters are rewritten.

 Page 12-2, lines 23-25: The relationship between more recent and past exposure  is
 mentioned. There is some information available regarding this rather than simply saying
 'There is no single  answer to this question". The type of missing information needed
 can then be described.

 Page 12-3, lines 26-27: This section suggests that hypersensitivity to diesel exhaust
 occurs and is based on an immunologic reaction. At this time, I don't  think this
 conclusion  is justified. There were 3 cases of asthma that were noted following high
 level exposure. The mechanism of this is likely not to be immunologic since the case
 reports share similarities with patients who develop asthma following a single massive
 exposure to an irritant gas, fume, vapor or smoke. The authors of the paper note that
the mechanism needs further clarification and speculated on the role of allergy. On line
27 the phrase "other ambient contaminants" occurs. It is not clear what this refers to.

 Page 12-5:  The sentence "Because of the key role alveoli play in the exchange of


gases, these changes may inhibit the efficiency of pulmonary function"  is written
referring to episodes of pulmonary edema in animals exposed to high levels of diesel
exhaust acutely. The terminology "pulmonary function" usually refers to pulmonary
function test results. It would be more precise that this could lead to hypoxia.

Page 12-5:, lines 19-29: The language describing the health effects here is written
poorly, such as the sentence "Several studies of workers occupationally exposed to  .
diesel exhaust on a short-term basis have monitored pulmonary function at the
beginning and end of work shifts to see if this marker of respiratory distress have been
impaired by exposures". This should be rewritten.

Page 12-5, bottom: The findings of a chronic respiratory impairment due to diesel
exposure are  overstated here, as is the comment that there is immunologic based lung
disease at the top of page 12-6.  The literature of long term pulmonary function
changes in humans is quite limited and inconsistent, and it is not clear that reported
changes in pulmonary function were due to exposure in the studies.

Page 12-6: Animal results are noted here regarding non-cancer health effects. It is not
clear that this  is relevant to human disease.

Page 12-7, bottom: The statements made here suggest that diesel exhaust can cause
emphysema. Has this been supported by experimental data? The sequence of events
noted in this section seems to apply only to animals exposed under particle overload

Page 12-9, line 16-17: A comment is made about "toxicological wisdom". This should
be made  more precise.

Page 12-10, line  17: A comments about a 20% to 70% elevation in risk is made based
on the exposure-response relationship from the cohort study. This sentence is based
on the relationship wjh years of exposure, and should  be deleted.

Page 12-11: The language selected to describe the epidemiologic studies is poorly
chosen. This section should mention also that studies with workers exposed for a
longer duration and with more years of follow-up  are needed.

Page 12-19, line  32-33: The case control study was not nested within the retrospective

Page 12-20, line  30: The comment is made that the use of 500 ng/m3 has "no
particular support, though it can't be ruled out"...I believe more informative statements
can be made  about historical exposures in the railroad industry.

Page 12-21: As noted earlier, it is not clear how these calculations are related to the
original calculations presented. I  also question whether 500 ng/m3 is likely to be in the
overload range (lines 23-24).

Page 12-25: As noted earlier, the rat data do not seem adequate for human risk

Page 12-27: I question whether much of the paragraph on susceptible subgroups is


speculative, particularly at ambient diesel concentrations. This section implies that
ambient diesel can worsen pulmonary fibrosis.

Page 12-37:1 am very concerned about quoting cancer risks of 1/100 due to ambient
diesel.  It is very likely that the risk is much less. Is it necessary to use quantitative
methods to assess health risk? Is a qualitative description sufficient?

Chemical Industry Institute of Toxicology
Roger O. McCIellan, D.V.M.
                                      May 11, 1998
        P.O. Box 12137
         6 Davis Drive
     Research Triangle Park,
      North Carolina 27709
        (919) 558-1200
      FAX (919) 558-1300
  Mr. Robert Flaak
  Designated Federal Officer
  Clean Air Scientific Advisory Committee
  U.S. Environmental Protection Agency
  Washington, DC 20460

  Dear Mr. Flaak:

        Enclosed are my written  review comments  on the "Health Assessment
  Document for Diesel Emissions" (February 1998 SAB Review Draft). These comments
  are intended to complement the oral comments I made at the meeting of the Diesel
  Review Panel on May 5-6, 1998.

        As I noted. at the meeting, the present document does not represent an update
  review and synthesis of the available data on diesel emissions and, thus, is not an
  adequate document for  regulatory decision-making.  Although  some portions of the
  document are  improved  over the 1995 draft, the present draft is highly uneven in its
  coverage of the published literature,  Moreover, the document is seriously flawed in its
  interpretation of the extensive body of data available on lung cancer in rats chronically
  exposed to high levels of diesel exhaust.  The  mechanistic data  now available
  indicates these findings  are not relevant for assessing human lung cancer risks from
  ambient environmental exposures.  In  addition, contrary to  statements in the report
  there is  no compelling evidence that -the organic fraction of inhaled diesel exhaust
  particles causes lung cancer in  rats. And  finally, the document overstates the potential
  utility of the existing epidemiological data  for quantitatively  estimating human lung
  cancer risks from ambient environmental exposures to diesel exhaust.

        Several portions of the document would have benefited from cross-referencing
  the recently completed criteria document and staff  paper on airborne  particulate
  matter.  This is especially the case regarding derivation of an RfC. The report did not
  present compelling evidence to support  an RfC 5  ng/m3 as opposed to adapting  a
  value of 15 (j.g/m3 based on  the recently promulgated  PM2.5 annual standard of
        I  urge the Agency to proceed with revision and updating of the document at an
   early date.  Delays in updating the document will result in some sections which are

Mr. Robert Flaak
May 11, 1998
Page Two
now nearly current becoming out-moded.  Alternatively, the updating and revisions
could be done concurrently with preparation of the next criteria document for airborne
particulate matter.


                                              &\   /I-   Sit / * „
                                   Roger O. McClellan, D.V.M.


Enclosure:  Review Comments

cc:    Dr. J.L. Mauderty, Chair
      Clean Air Scientific Advisory Committee

                                                             Roger O. McClellan
                                                                   May 4, 1998
Chapter 2.  Diesel Emission. Transport, and Transformation
      The present chapter is seriously deficient in two areas. First, the chapter does
not clearly describe in quantitative or semi-quantitative terms, the substantial changes
in diesel engine technology, and fuel quality that have occurred over the past two
decades. Second, the chapter does not adequately describe the most recent research
findings on the influence of exhaust control technologies or emissions.

                                                             Roger O. McCleilan
                                                                    May 4, 1998
Chaoter 4.  Dosimetric Factors
      This chapter reviews the relevant literature on the disposition of inhaled diesel
paniculate material (DPM) as it existed in 1994. The chapter is deficient in not
reviewing substantial literature published since 1994.  Some of the newer literature
and analyses were included in the most recent Particulate Matter Criteria document
compiled by EPA.
      A major deficiency of the chapter is the failure to integrate and synthesize the
information presented.  Thus, the chapter in its present form is an exposition on
dosimetric factors.  What is needed is an integrative treatise on the "disposition of
inhaled diesel exhaust." The chapter appropriately concentrates on the fraction of
DPM deposited in the lower respiratory tract. The chapters coverage of the key
concept of "particle overload" in the rat should be expanded (See McCleilan 1996).
The coverage of this issue could  be enhanced by including two figures based on the
work of Wolff et al. (1987) (attached).  The text should  more  clearly describe the inter-
relationship between particle burden, inflammation and impaired clearance and how
each builds on the other.
      The chapter could be improved by providing some quantitative estimates of
DPM deposition and retention in the human respiratory tract especially the
tracheobronchial and alveolar regions. Such information  would aid the reader in
appreciating the small quantities  of DPM including particle associated organics
predicted to be deposited under the highest levels of plausible human exposure.
      Quantitative estimates can be developed using  the information presented in
Table 4.1 and the paper by Xu and Yu (1987) that is the source of Table 4.1. Table 1
from Xu and Yu (1987) is attached and should  probably be included in the chapter. As
an aside, I believe this paper was included in a report by C.P. Yu to the Health Effects

                                                             Roger O. McClellan

Institute.  The Health Effects Institutes report includes a critique of the Xu and Yu work
which should be considered by the author(s) of the chapter in citing Xu and Yu.
      From the paper of Xu and Yu the following may be calculated for an individual
continuously exposed to diesel exhaust particles at 1 ng/m3 with 20% organic content
and methylanthracene (as an example) as 10^3 of the total organics and  1,6-
dinitropyrene (as an example) present as 10*6 of the organic fraction.

                                         Annual Deposition  in Lung
       Total Diesel Particulate Matter               420 u,g
       Total organics                             84 u,g
       Methylanthracene                        84 x 10'3  jig
       1,6-dinitropyrene                         84 x 10'6  p.g

      These values for continuous exposure to 1 |j.g of DPM/m3 can readily be scaled
to higher or lower air concentrations.  Although the precise estimates may be  open to
criticfsm in may opinion they are accurate to within a factor of 2-3.
      Information such as discussed above is useful in understanding the apparent
contradiction between diesel exhaust particles containing known or suspected
carcinogens yet their being no clear carcinogenic signal  in humans and  no clear
signal that the organics are causing lung cancer in rodents.  At high concentrations of
even a few hundred fig of DPM/m3 the amount of carcinogenic material delivered to
the lung is apparently well within the capacity of the  lung to detoxify the organic
compounds. The delivery of huge quantities of these chemicals to isolated cells in in
vitro studies on the other hand results in clear signals for mutagenicity.
      A recent review paper on pulmonary retention and clearance of inhaled
biopersistent aerosol particles by Stober and McClellan (1997) will be useful to the
author in updating the chapter.  A recent review by McClellan (1997) on the use of

                                                            Roger O. McCIellan

mechanistic data in assessing human risks from exposure to particles provides
guidance for reconciling the in vitro and in vivo findings.

McCIellan, R.O. (1997). Use of mechanistic data in assessing human risks from
      exposure to particles.  Environmental Health Perspectives 105 (Supplement
      5) :1363-1372.
McCIellan, R.O. (1996). Lung  cancer in rats from prolonged exposure to high
      concentrations of particles: implications for human risk assessment.  Inhalation
      Toxicology 8 (Supplement): 193-226.
Stober, W. and McClellan, R.  O. (1997). Pulmonary retention and clearance of inhaled
      biopersistent aerosol particles: data-reducing interpolation models and models
      of physiologically based system.  Critical Reviews in Toxicology 27(6):539-598.
Xu, G. B. and Yu, C. P. (1987). Deposition of diesel exhaust particles in mammalian
      lungs. A comparison between rodents and man. Aerosol Science and
      Technology 7:117-123.






a. co

                                                            Roger O. McClellan
                                                                   May 4, 1998
Chapter 7.  Carcinoaenicitv
      This chapter is a comprehensive compilation of all the studies conducted in
laboratory animals to evaluate the carcinogenicity of exposure to diesel exhaust and
organic extracts of particles.  However, the author has not adequately integrated and
interpreted the data. This deficiency is apparent early in the chapter as the section on
inhalation studies in rats starts immediately with a detailed description of a single
study. The same flawed approach is true of other sections. An introductory paragraph
could be  used to good advantage to alert the reader to what will follow.
      The author has apparently not critically read all of the referenced studies. For
example, the  author appropriately notes on page 7-40, lines 27-28 that most rat
studies indicate a trend of increasing "lung" tumor incidence at exposures exceeding
1 x 104 mg • hr/m3.  This is followed by the erroneous statement —"A similar
comparison could not be adequately made for mice, because experimental designs
were not  comparable." The author is reminded that the rat study of Mauderly et al.
(1987) showing the trend of increased lung tumor incidence at high exposure
concentrations was conducted concurrently with the negative  mouse study reported by
Mauderly et al. (1996).  Why does the author not view these two designs as
comparable, indeed identical.  The reason a similar trend is not seen for mice is that
the studies are generally negative.
      A significant deficiency in the chapter is the failure to note the substantial
changes  in diesel engine technology and fuel quality over the past two decades. The
result has been substantially reduced diesel paniculate matter emissions.  The data
available on diesel  exhaust carcinogenicity was obtained with engines that fall far
short of today's technology. For example, the Lovelace studies with rats and mice
were conducted using a 1980 General Motors V-8 diesel engine with the exhaust for

                                                            Roger O. McClellan

the high level animals diluted about 1 to 10 with clean air to achieve a concentration
for the high level animals of 7 mg/m3.  I suspect if the study were repeated today it
would be necessary to use undiluted exhaust to get 7 mg/m3 and 02 would have to be
added to sustain the animals.
      The chapter is also deficient in not more clearly pointing out that most of the
carcinogenicity data was obtained in exhaust from light duty vehicles with only limited
data available for exhaust from heavy duty engines. There is a shortage of information
available on DPM from locomotives or boats/ships powered by diesel engines.
      The summary is excessively long and is somewhat misleading.  The last two
paragraphs definitely need to be rewritten. The author needs to make it clear that
positive lung tumor findings in rats are high exposure concentration dependent and
are very likely species dependent. It should also be emphasized that the data on
carcinogenicity of the organic fraction of DPM is based on high dose skin painting
studies. There is no. evidence for the organic fraction contributing to lung cancer in the
      The statement  (page 7-44, lines 17-18) that the relative importance of the
absorbed organics remains to be elucidated is misleading.  The present evidence is
compelling for the adsorbed organics having no role in the pathogenesis if the lung
tumors observed in the rats.

                                                            Roger 0. McClellan
                                                                   May 4, 1998
Chapter 9.  Mutagenicity

      The body of this chapter is a well-written review of the substantial amount of
data available on the mutagenicity of DPM in bacteria mammalian cells and intact
      However, the chapter is deficient in not including the results reported by
Driscoll, Oberdorster and colleagues (1996) on mutations observed in lung cells of
rats exposed to high concentrations of carbon-black. Admittedly, carbon  black is not
DPM; however, carbonaceous material is the major constituent of DPM.  Because
carbon black used by Driscoll and Oberdorster was free of any significant quantity of
PAHs, it is generally accepted that the mutations in the carbon black-exposed rats
arise as a result of  damage from oxygen  radicals produced by chronic particle-induced
inflammation.  There is strong circumstantial evidence linking the high lung binders  of
particles with chronic  inflammation with increased lung cell mutations with increased
lung cancer in  rats exposed to DPM, carbon black, and several other kinds of
particulate matter (PM). This effect appears to be rat specific and not  relevant for
assessing human cancer results of PM.
      A second deficiency in this chapter is the failure to note that the vast majority  of
the mutagenicity data was obtained on DPM (and extracts) collected from engines
manufactured a decade or more ago. The advances in engine technology and fuel
quality over the last two decades has been substantial.  Particulate emissions have
been substantially reduced. It would be surprising if the chemical composition of the
organic fraction had not changed, thus influencing mutagenicity.  The old data may  be
qualitatively relevant to the new engines but very likely do not have quantitative
      The first two paragraphs of the summary are appropriate.  The third paragraph
overstates the uncertainty in DPM as be&gstf concern for producing veritable genetic

                                                           Roger O. McClellan

risks.  In this reviewer's professional opinion, chronic exposure to DPM at the highest
plausible levels of human exposure does not pose any veritable genetic risk.
      The fourth paragraph of the summary borders on the incoherent and needs to
be rewritten.

Driscoll, K.D., Carter, J.M., Howard, B.W., Hassenbein, D.G., Pepelko, W., Baggs, R.B.,
      and Oberdorster, G. (1996).  Pulmonary Inflammatory, Chemokine, and
      Mutagenic Responses in Rats after Subchronic Inhalation of Carbon Black.
      Toxicology and Applied Pharmacology 136:372-380.

                                                             Roger O. McClellan
                                                                   May 4, 1998
Chapter 12.  Health Risk Characterization for Diesel Engine Emissions

      This chapter fails to meet the stated objective of integrating and summarizing
the key findings about the health hazards and risk potential for humans exposed to
ambient levels of diesel exhaust.  The failure to meet the stated objectives relates to
deficiencies in the earlier chapters.  Many of these chapters do not contain information
published since 1994 and, thus, provide an incomplete basis for developing a
summary chapter.  In addition, many of the chapters while prefacing extensive
compilations of published work fail to provide a crisp summary of key conclusions.
This absence of critical summarization in individual chapters makes it difficult to
prepare an overall integrated summary chapter.
      A critical deficiency in Chapter 12 is the failure to adequately convey
information on the marked changes in diesel  engine technology and fuel quality that
have occurred over the last two decades.  As noted earlier in my comments, it is
important to recognize that most of the data available on health effects of diesel
exhaust in laboratory animals were derived using diesel exhaust from engines that are
now out-moded and the fuels are  of inferior quality as compared to the fuels used
today.  It is crucial that these differences be considered when risks are being estimated
for current technologies.
      It would also be helpful if the chapter were to more clearly distinguish between
information obtained with light duty diesel engines versus heavy duty engines.
      The chapter fails in many areas to place observations within a quantitative
framework. For example, the descriptions of  acute, short-term and chronic exposure
on pages 12-3 to  12-7 make no reference to the  exposure concentrations that elicit the
various endpoints observed.  Likewise, the description of loxicity mode  of action" on
pages 12-7 to 12-9 is presented without adequate discussion of how the exposure

                                                             Roger 0. McClellan

level influences the "mode of action."  See the papers by McClellan (1996) and
McClellan (1997) on the importance of considering exposure level specific effects.
      Finally, on page 12-12, quantitative data are presented with a concluding
statement that responses were not detected in rats or mice at lower exposure
concentrations of 350-2,200 (ig/m3 in contrast to the lung tumor responses observed in
rats exposed to higher concentrations.  The author then proceeds to hypothesize that
positive results would have been observed at low concentrations if this study had used
more animals. What the authors should have noted is that in a pooled analysis by
Valberg (1998) he determined that 1212 rats had been exposed to average lifetime
diesel exhaust concentrations from 50 to 550  p.g/m3 without an excess lung tumor risk
having been observed compared to 1135 controls. These exposure concentrations
did not elicit an overload effect (inflammation, impaired clearance, mutations and lung
tumors) so the potential existed for tumors to be observed due to the diesel exhaust
particle organic constituents. An excess incidence was not observed.
      This absence of an effect is totally compatible with the low quantities of
individual organic compounds that are actually  inhaled and this apparent low potency
for eliciting mutagenic and carcinogenic responses when detoxification mechanisms
are not overloaded.
      The section 12.2.4 (Weight of Evidence Summary) overstates the current status
of knowledge on the human carcinogenic risks of diesel exhaust.  I agree with the
conclusion that the human epidemiological evidence is "limited."  I strongly disagree
with characterization of the rodent data or "sufficient evidence" absent qualification as
to the requirement for high level chronic diesel exposures being required to elicit
"positive" effects in  rats.
      I would characterize our present knowledge from epidemiological, laboratory
animal and mechanistic studies as supporting a classification of diesel exhaust as a

                                                             Roger O. McClellan

"possible" to "probable" human carcinogen with likely low potency at plausible
environmental levels of exposure to diesel exhaust. The data are not suitable for
deriving quantitative exposure lung tumor response relationships.
      I disagree with the derivation of an RfC value of 5 ng diesel paniculate matter
per m3.  The use of an uncertainty factor of 10 and an RFC of 15 }o.g/m3 would be
appropriate. This value is compatible with the Agency's recent promulgation of an
Annual PM2.5 standard of 15 p.g/m3. In my opinion, the PMa.s annual standard arrived
at through "expert judgmenf and "policy calls' should take precedent over the use of a
process that uses some arbitrarily selected uncertainty factor.
      The chapter then proceeds to relate some quantitative estimates of this risk
potency of diesel particulate matter using four approaches: epidemiological studies,
Benzo(a)pyrene as a biomarker, animal studies, and a comparative potency approach.
The epidimiologic approach builds on a "back of the envelope" approach I advanced a
decade ago.  Much additional information has been developed over the  last decade
causing me to reassess my own earlier statements as well as earlier statements of
other individuals. In my opinion today, none of the approaches are fully satisfactory for
providing quantitative estimates of human lung cancer risk.  They all have serious
limitations which should be more clearly articulated in this chapter.
      The epidemiological data are not sufficiently rodust for developing quantitative
estimates of risk. Moreover, our knowledge  of the exposures clearly precludes
development of quantitative estimation of potency for lung cancer induction by diesel
particulate matter.
      The benzo(a)pyrene biomarker approach is flawed because it does not appear
that benzo(a)pyrene has any role in diesel particulate matter causing lung cancer.
The animal studies do not provide a basis for estimating human lung cancer risk

                                                             Roger O. McClellan

because the mechanisms by which diesel participate matter causes lung tumors in rats
is not relevant for humans exposed at plausible environmental levels of exposure.
      And finally, the comparative potency method is not supported by our current
knowledge of how the several agents or technologies cause lung cancer. The
differences are likely substantial.
      Thus, as a bottom line, it is not possible to derive quantitative estimates of
human cancer risks for DPM.  It would be appropriate to note that chronic low level
exposure to DPM may increase the human lung cancer risk particularly at high levels
of exposure in excess of 100 ^ig/m3.
      One can use the findings of Garshick et al. (1987) to give perspective to the
potential lung cancer risks of exposure to diesel exhaust. They observed a relative
risk of 1.41 for railroad workers exposed to diesel exhaust.  They observed relative
risks of 3.29 and 5.68 for the same population smoking less than 50 pack-years and
over 50 pack-years, respectively (See attached table). Thus, diesel exhaust was only
very weakly associated with lung cancer as compared to cigarette smoking.
      The Garshick et al. (1987) data can, with trepidation, be used to derive relative
risks for environmental exposures to  diesel exhaust. One adjustment is for intensity of
exposure. If it is assumed that the railroad workers were exposed to 500 p.g/m3 during
work hours, this value can be normalized to a continuous exposure concentration of
120 n.g/m3.  A second adjustment may be appropriate for duration of exposure. One
approach is to assume all exposures throughout life are equally effective in their
association with lung cancer. Using this approach, the excess relative risk of 0.41 is
adjusted to 1.43 excessive risk for 70 years of exposure (versus 20 years for the
railroad workers). The 1.43 excess risk for 120 jig/m3 continuous exposure can be
extrapolated downward, again with trepidation, to an excess relative risk of
approximately 0.01 for a nationwide average concentration of 1.1 ng/m3.  Because a

                                                            Roger O. McClellan

relative risk model is used, the vast majority of the excess cases attributed to diesel
exhaust exposure are estimated to occur in smokers or former exposures; very few of
the excess cases are estimated to occur in nonsmokers.
      Figure 12-1 comparing cancer risks, RfC and Margin of Exposure could be
modified to more clearly depict the risks observed in epidemiological studies and then
being estimated. (See attached Figure 12.1 with hand-drawn additions). The revised
figures show estimated values for the lifetime risk of lung cancer in smokers and non-
smokers. The estimated risk for individuals smoking and exposed to diesel exhaust is
also  shown based on the Garshick et al. (1987) data. I suspect that his estimate of the
odds ratki for diesel exposure was largely based on smokers since i suspect he did
not have sufficient lung cancer cases in either diesel exposed or nonexposed
nonsmokers to derive statistically meaningful estimates of odds ratio.
      One point clear from the revised graph is that extrapolations from real data
estimates must be made by more than two orders of magnitudes to reach the exposure
levels of concern for ambient environmental exposures.  As an aside, in examining
Figure 12.1, it is not clear to me why the "cancer risk line" at 2 x 10'3 is more than 2
orders of magnitude displaced from the cancer risk line of 1  x 10~4.

Garshick et al. (1987) Am. Rev. Resp. Dis.
McClellan, R.O. (1996).  Lung cancer in rats from prolonged exposure to high
      concentrations of particles: implications for human risk assessment.  Inhalation
      Toxicology 8 (Suppl): 193-226.
McClellan, R.O. (1997). Use of mechanistic data in assessing human risks from
      exposure to particles.  Environmental Health Perspectives 105 (Suppl. 5):1363-


HO d




LJ-I  -1
coco I


^ *°
?? AI

S *
g S
O vi








§ %
- £
T- D)
1 -g
^ CD





If- •





    ,ui/flrt (OQS) Ida - ainsodxg UBUOTH
    .uu/Brt (096) JWUBO JQJ Tivoi teg
tiu/6rt (szU Xpms K^ - aunaxfcq
                               tuu/6rt (5) luetqmv UBIUHH
                               Cuj/6r1 (3) luaiquiy UBUJPH

P?ge.4r.l -    line 26: Replace "these findings" with "the exceptional findings"
              line 33: Replace "total" with "retained"
Page.4;3 -    line 8-14: Deposition in the alveolar region is mentioned here,  implying that the
    small diesel particles deposit preferentially in this area. However, tracheobronchial deposition
    if expressed per unit surface area may even be greater, which may be of significance especially
    in situations (humans) where mucociliary clearance is impaired.
Page.fM -     line 34
Page.4-6 -     line 1: Table 4-1 does not list data on a volume basis which makes this sentence
    hard to understand unless the deposited mass is also expressed per lung volume.
Page.4^6 -     lines 13 and 24: Replace "highly insoluble" ('.'relatively insoluble") with "poorly
Page^lp -   line 6:  Replace "individuals" with "coal miners" and add after pneumoconiosis
    "with presumably high lung burdens of coal mine dust"
Page.47l2 -   line 11:  It sounds as if only the high dose exposure group resulted in significant
    lung burdens, but not the other ones, which I am sure is not correct.  It  would be best to
    indicate the amount actually found being retained in all dose groups.

              Line 4-10: The data described here in this chapter should be summarized in a short
    table which would  make it much easier for  the reader to evaluate  prolongation of alveolar
    clearance with exposure concentrations and lung burdens.
Page,4:_13 -   line 18: Delete the word "generally"
P.age..4r.l.4 -   line 35: Replace in "aerosols" with "poorly soluble particles of low toxicity"
Page.4:16 -   lines 4-9:  It should be more clearly discussed that the term "panicle  overload"
    relates only to poorly soluble particles of low cytotoxicity and that cytotoxic particles such as

    crystalline SiO2 and effects caused by them in the  lung are not  part  of the overload
    phenomenon.  Also in line 7, it is not clear what surface associated organics in the case of silica
    mean. Silica particles induce their effects when administered as highly pure particles without
    any adsorbed material.
?ase.4:.1.6 -    line 16:  In the study by Freedman and Robinson (1988) no attempts were made to
    diagnosing lung cancers. Thus, this sentence needs to be changed or deleted.
Page.4;19 -    line 6: Replace "lymphocytes" with "neutrophils"
Page.^l? -    line 9:   Delete the word "significantly" since  only  0.001% of the particles were
    translocated to the lymph nodes within 24 hrs.
Page.4;26 -    line 13  and  14:  The identical elimination rates of  BaP and particles  also could
    indicate that the desorption for BaP is the same as the clearance of particles. This conclusion is
    also used in lines 21 and 22 of the same page.  However,  this study was only done over a 7-
    day period and cannot be used to predict long-term clearance of adsorbed BaP.
              Line 30 and 31: It has not been shown by Snipes and Gem and by OberdSrster et
    al. that 15 Jim particles can be phagocyn'zed by alveolar macrophages.   On the contrary,
    Oberdorster et al. (1997, Inhaled Panicles VIII) have shown that in mice even  10 pm particles
    are not phagocytized by alveolar macrophages. Mice were used  in the study by Creasia et al.
Page.4:28 -    line 11:  Type I cells are not target cells  for long-term effects by  diesel particles.
    Type n cells,  however,  are.
Page.4:28 -    line 25,26: Retention halftimes for humans can be even higher up  to two years as
    shown by studies in humans by Bailey et al.  I would also suggest here (line 26) and in other
    parts of this document to use the term retention in the context of halftime rather than clearance,
    and to use the term clearance when the actual rate is given or discussed,  but not clearance
Page.4r32-    line 26:  Change "inhibited" with "altered"
                                        * * * * *
       This dosimetry  chapter basically contains the same information as the previous version of
the document.  However, it should be updated in at least two areas: One is the  model that was

described by St5ber et at. (POCK model) has now been described as an updated advanced version
by Stober and McClellan (1997) and should be incorporated here. The other potentially important
development is the change in diesel engines, at least for trucks,  resulting in a  significantly
decreased output by mass of particles, but at the same time a significant increase in particles of the
ultrafine mode.  Discussion on the potential health effects of ultrafine particles is ongoing now and
it would be useful to incorporate dosimetric aspects of ultrafine particle deposition in  this chapter.
These particles according to the ICRP model can have an extreme high deposition efficiency in the
alveolar region, and  their disposition after deposition is also quite different from that of larger
particles. This information would be useful to include in a chapter written in 1998.
Chapter  7:   Carcinogenicity of Diesel Emissions in Laboratory Animals
PageJ.:3.and.fpllowingt Table .7-1:  Add "inhalation" to the title of this table.
Page_7rlp -    lines 1 and 2: The issue of the squamous cyst may need to be discussed again in
    view of the publication by Boorman et al. (Toxicologic Pathology 24:  564-572, 1996) which
    is a summary report of a meeting of international pathologists describing a consensus of how to
    classify squamous cysts.
              lines 8 and 9: The logistic regression model described here is presumably that by
    Mauderly et al. (1987) which needs to be referenced.  It may also be useful to show a figure of
    this regression model in the document.
Page_7:12 -    line 9: The TiO2 used in  this study was ultrafine TiO2.  The animals  exposed to
    these high concentrations of ultrafine  TiO2 and carbon black showed also lower  survival
    compared  to the diesel-exposed animals which the  authors attributed to the  high inhaled
    concentration of these particles resulting in increased general toxicity.
Page.7:.13 -    lines 4-13: The effects described here on prolongation of alveolar clearance should
    also be presented in Chapter 5.

Page.7:17 -   line 15: The development of mesotheliomas in the diesel-exposed rats is mentioned
   here.  This is not indicated in Table 7-1, and if significant, would be a finding  worth
   discussing. Was there any fiber exposure ongoing at that time in this contract laboratory?
Page.7:.19 -   lines 31-34:  TiO2 used in this study was ultrafine:  Increase of the ultrafine TiO2
   concentration and carbon black concentrations resulted in high toxicity in the mice such that die
   exposure duration  of the whole study had to  be shortened to a total of 13.5  months of
   exposures. This should be mentioned here in the report since it reflects the high exposure
   concentrations and lung burdens used in this study.
Page.7:26 -   Table 7-2: Include in the tide that treatment was done by surgical lung implantation.
Page.7:28 -   line 6: Replace "MMAD" with "surface area"
Page.7:29 -   Table 7-3: The source should be indicated for the study, Dasenbrock et al., 1996.
       If available, a characterization of the different carbon black used in the different studies
should be included with respect to particle size and surface area.
                                       * * *  *  *
       A general comment for the whole document relates to the lack of including newer data. For
example, in Chapter 5, animal studies on effects of hypersensitivity reactions after diesel exposure
are missing; Chapter  10  should  include  a more  detailed description  and discussion on  the
inflammatory mechanism of carcinogenicity for which there are a number of  recent relevant
publications by Driscoll et al.

Chapter 6: I suggest to include a range for the RfC (5-16 mg/m ) because of the questionable use
   of the Data Base Uncertainty factor.
Chapter 9:   The studies listed on  pages 9-3 and 9-4  should all include  the doses or inhaled
   concentrations that were used. This would put the results into perspective, and it would be
   helpful to point out the high doses  (in vitro or injected)  which resulted in positive effects.
   Missing  in this chapter are our studies reported by Driscoll et al.,  (1996) with carbon black
   showing secondary genotoxicity (HPRT mutations) after  subchronic inhalation  in rats;  and
   additional studies with in vitro exposures of epithelial cells to inflammatory cells from in vivo
   exposed animals resulting in HPRT mutations; and the  blocking of  these  mutations  with
   antioxidants (all studies by Driscoll et a/.). Taken together these studies strongly suggest an
   inflammatory mechanism of mutagenesis in particle overload situations.   This information
   should also be presented and included in the Mechanism Chapter 10.
Chapter 10:   Page.lO-4A lines.9:17: Include Driscoll et ai. studies mentioned above, in vivo, in
   vitro and ex vivo studies which are key for plausible mechanism.
             ?.?SS..10:.17A..line.5:  Inflammation,  however,  appears to  be a prerequisite for
                           Line 6/7: Point out the high dose in Riebe-Imre study.
                           Line 9:  PTFE particles are highly toxic and not a surrogate for diesel
   particles, a cautionary note should be added.
             ?£§£. l.P^2j3AAineJ9/20: The slower particle clearance in humans allows for greater
   efficiency of extraction:  This statement is in contract to what is discussed in the dosimetry
   chapter on page 4-28.
                           Line.21/22:  DE is effective in non-overload conditions: Which study
   is that?
                           Lines.25-29: The "low" doses in the Riebe-Imre study are not that
   low, the fact that no cytotoxicity was seen does not mean that the in vitro dose was low.

Chapter 11:    Page.l 1:.16ATable .1.1-.6:  I suggest to express risk estimates per ng organics on the
   diesel particles.  This estimate then can be compared to cancer risk estimates derived from a
   good data base of humans exposed  to PAH, such as coke oven workers.  This will allow a
   comparison between  diesel-associated organics and others and  provide a test about how
   reasonable the diesel-derived risk estimate is.
Chapter 12:  I suggest to define the estimated risk as "hypothetical risk", indicating the uncertain
   nature. In addition, a listing of the uncertainties of using the high dose rat data for predicting
   low dose human risk should be included as an explanatory note for the hypothetical risk.
G. Oberdfirster

Bill Pierson
Chapter 1

Is it true that the fraction less than 2.5 urn that are implicated in the cancer and noncancer effects
observed from diesel particulate emissions exposure?  Where is the support for this statement
(lines 1-1-33 to 1-2-2) ?

Is it really true that "companion comprehensive characterizations	" are not included in the
assessment? I thought some of them were, in Chapter 2.
Chapter 2

1) A glaring weakness is that there is no mention of the work of Baumgard and Johnson (1996)
or of Bagley et al. (1996). In that work they claim to have found with present-day engines there
is a  103-fold increase in fine particulate matter compared to past engines.   It is as if current
engines,  supposedly cleaner than in the past, give huge amounts  of ultrafine (0.0075-0.46)
particles.  The nature of these particles is not identified but Baumgard and Johnson think it may
be H2SO4. Kittelson, on the other hand, believes that the same is true of the older engines as well
depending on how the sampling is carried out, and he cites Kittelson et al. (1988) and the work
of Paul Harrison in support.  Currently this is an intense issue among  the diesel manufacturers
and a Coordinating Research Council initiative is getting underway.  None of this is mentioned
in Ch. 2  except indirectly perhaps at the bottom of p. 2-8 and top of p. 2-9 and the end of the
exposure perspective (section 2.6, p. 2-57, 1. 20 to p. 2-58, 1. 8), and there they say that basically
they recognize  that they do not want to get  into it.  But I think they have to, discussing at least
the papers mentioned above.

They do acknowledge the dilution of diesel exhaust under roadway conditions as opposed to
dilution in a  dilution tunnel is  an important factor to consider.  As  stated on p. 2-26,  1.  7-10
"This  discrepancy  leads  to  slightly different particle  size distributions under real  driving
conditions than those predicted from laboratory data (Kittelson and Dolan, 1980); for example,
because of slower coagulation processes, more particles in the Aitken  nuclei range (< 0.08 um
diameter) may be expected under typical roadway conditions". See also pp. 2-45, 1.23-30.  Also
see Dolan, Kittelson, and Pui (1980).

2) In fact, all but about 5 references are dated 1990 or before.  In the next few days I will be
sending reprints and reprint  lists that in my opinion should be included.  And yet, on the other
hand, there seems to be no awareness ( on p. 2-8, for example), of the early work by Kotin, Falk
and Thomas (1955) - see complete ref. in Ch. 11-26, line 36 - and others listed at the  end of my

 3) The material on pp. 2-38 to 2-43 (see sec., and and 2.5.3) are important for
the rest of the document and yet seem to have been overlooked by all of the writers of the rest of
the document.  In sec. it is described how, in the gas phase,  2- to 4-ring PAH's emitted


Bill Pierson
by diesels into the ambient air are attacked by a 2-step process involving OH or N205 followed
by NO2 to produce airborne mutagens in the gas phase not found in exhaust emissions, such as 2-
nitrofluoranthene and 2-nitropyrene (instead of 3-aitrofluoranthene and 1-nitropyrene); see Table
2-18 (p. 2-52). Eventually, these compounds, though formed in the gas phase, will find their way
to the particle phase (sec. would probably be a good place to point this out).

Thus, "a knowledge of diesel emissions at or near their sources is not sufficient to fully assess the
impact of these emissions on human health ... However, data on how diesel exhaust contributes
to exposure levels for these secondary pollutants  are currently lacking" (p. 2-56).  The point is
also made on p. 2-43 that "the formation of nitro-PAHs during sampling may be an important
problem for diesel particulate matter because of the presence  of NO2 and HNO3. In fact, Gorse,
as mentioned on p.  2-47, found in the Allegheny Tunnel that 1-nitropyrene was much lower than
it is in dilution-tube sampling.  Differences listed on p. 2-45 should serve as a warning on this
•   dilution ratio
•   temperature
•   residence times
•   mixing with other vehicle exhausts
•   mixing with ambient pollutants

In summary this whole section of Ch. 2 (and the Exposure Perspective and the Summary) carries
the message  that as one moves away from the dilution tube  and especially as one moves
downwind and into the real world and allows processes (see pp. 2-28 and 2-40);
•   photolysis
•   reaction with OH
•   reaction with ozone
•   reaction with HO2 and H202
•   reaction with NO3, N2O5, NO^ HN03, HNOa,  H2SO4
we enter an area that could be important but we actually don't know - and the other chapters of
the criteria document do not reflect any of this at all.

4)  However, there is another problem in that gasoline-powered vehicles emit PAHs also. If we
start down the path of downwind effects, then the problem suddenly becomes one of much more
than diesel effects.

5)  What do the nitro-PAHs transform into inside the body?  This is discussed hi Ch. 10 Sec. 2
which fact should be mentioned in Ch. 2.

There are a number of minor comments:

Beginning in this chapter and continuing all the  way through the document,  DE and DPM are
used. I saw DE defined as diesel exhaust and DPM defined  as diesel particulate matter but they

Bill Pierson
are often used to refer to what is apparently the same thing. I suggest that the document be gone
through to fix this.

p. 2-2,  11. 4-36 and p. 25,  1. 35:  The same lubricating oil is used in gasoline engines.  Should
this not be pointed out?  Why is lubricating  oil not as much of a problem in SI engines; the
consumption rates are similar between SI and diesel.

Sec. 2-2. Somewhere in this section might be a better place than in Chapter 12 (p. 12-38,  1. 1-4)
that most engine modification steps to reduce particulate emissions  increases NOX emissions
(with its own set of problems as listed in Chapter 12), and vice versa.

Also in Chapter 2, probably in Sec. 2-2, there should be some discussion of the remanufacturing
process for diesel trucks.   Because of remanufacturing and also the long lifetime of a diesel
relative to gasoline engines, the  time between emission-reduction (or any other) step is much
longer than we are used to thinking of based on gasoline engines. And I do not know how much
freedom there  is to implement  steps that  would  improve  emissions  at each  remanufacture.
Obviously this relates to the recent discussions from Johnson et  al. about newer engines putting
out completely different particles.  It affects how long it  takes for steps taken to  improve
emissions to  show up on the road, or whether if there is a big change how long we can expect
Chapters 4-12 to have any bearing on anything.

p. 2-3, define HEI.

p. 2-6, line 25, why is carbon never formed at < 1900° K?

p. 2-7,  11. 16-21: Hampton et al. (1983) had a treatment of this,  showing that the gas/particle
phase apportionment essentially followed Raoult's Law (see Fig.  8 of Hampton et al.).

p. 2-7,  11. 24 to 2-8, 1. 3:   Truex et al.  (1980) showed that 1) the particulate sulfate  in diesel
exhaust is H2SO4 and 2) as  the S content of the fuel goes down the % H2SO4/SO2 ratio goes up.
This paper is discussed on p. 2-10, lines 10-13, but it is not mentioned in the latter case that the
% H2SO4/SO2 goes up as fuel S goes down.1

p. 2-8, lines 4-11: Shouldn't this have a " Oxidation  of Nitrogen Oxides" title on it?  It
does not belong under "Oxidation of Sulfur Oxides".  Also the Harris et al. measurements  of
HNO3  are mentioned but those of Okomoto et al. are not; but on p. 2-10, 11. 14-18 they are all
mentioned.  Since NOX and HNO3 are gases, perhaps all of the discussion on p. 2-8, 1.  4-11
should be moved to, and combined with, that on p. 2-10,  11.14-18.

p. 2-8, line 25, can you give an idea of how much mutagenicity has been found?

p. 2-8, line 35 to 2-9, line 3: If the diesel emissions currently used are different  from those that
may have occurred in the past, this is a real quandary for the  whole document (which I think we
all recognize).

Bill Pierson

Table 2-3: It is very difficult to figure out which numbers come from where.  What is ref. h - it
must be one of f and g. What does the [g/km] mean at the top, since only one set of numbers is

The numbers in Table 2-3:  There is something seriously wrong. The original references for n-
decane and a-dodecane (n-undecane is left out), toluene, ethylbenzene and naphthalene  give
numbers between  11 and 8666 times the numbers given here.  The value for benzene is a factor
of at least several hundred off from any reasonable number.

Table 2-3: Text at bottom of p.2-10 and top of p. 2-12 implies that all data in Table 2-3 were
obtained on the Federal Test Procedure.  The data from the tunnels were not on the Federal Test

Table 2-3: I am not sure why U.K. data for noncatalyst cars were used.  There should be plenty
of data from the U.S.

p. 2-13, 1.5: "Table 2-3 lists the emission rates for exhaust pipe emissions only". But this is not
true since Table 2-3 lists data from vehicle tunnels. If it is true (1. 5-7) that 30-60 % of total HC
emissions from passenger gasoline vehicles are from fuel evaporation,  then in Table 2-3 there
should be a note and the tunnels should be segregated from the others.

p. 2-13,11. 22-23:  Carey and Cohen are cited as if they are included in Table 2-3 but they  are

p. 2-12,11. 5-29: Subsequently we have carried out an intensive tunnel experiment and now have
considerable data  to add o this discussion and to Table 3, if you want at this late date to add it.
In particular we now have the data for C<8 which would make up for the deficiency mentioned
on lines 27-29.  (See Pierson et al, Atmos. Environ. 30, 2233-2256;  Sagebiel et al,  Atmos.
Environ. 30,2287-2296.) However, see the comment below on p. 2-47,11. 11-31.

Table 2-4:  The g/mi numbers for heavy and light-duty diesels from Williams et al. sound a bit
high.  Is it appropriate to use data from Australia?  Do we not have data from SWRI (indeed
sponsored by EPA) that would serve?

Table 2-4:  I think ref. d had numbers for more than just OC.  In particular, we had 84 ± 14%
carbon in the total particulate mass from diesels, 67 ± 42% from SI, which should be added to
your row labeled TC (% w/w) hi Table 2-4.

p. 2-17,11. 1-2: How does this statement affect the statements on p. 2-13,1. 36 and p. 2-14,  1. 1?

p. 2-19,11. 22-24: There is much older work, starting with Kotin, Falk and Thomas (see Chapter
7 refs., p. 7-46,1.  50) which is not cited here.

Bill Pierson
p. 2-23, 1. 19-20: What is the significance of phenazine and phthalic anhydride being identified,
or is it just an observation?

p. 2-23, 11. 4-7:  The paper by Salmeen, listed in your refs. (p. 2-67, 1. 23), is far more pertinent
than the Schuetzle paper cited here.  The latter paper was dealing with material recovered after
indefinite times from the wall of a dilution tube and may or may not represent diesel exhaust.

p. 2-24: There is another much more serious problem than blow-off. There is also degradation.
Far more serious is the problem of degradation into other products.  This was shown in a vehicle
tunnel by Lee et al., ref. 33 (see your list of refs., p. 2-64, 11. 33-36).

Fig. 2-2:  Since  presumably the vapor/particle distribution depends critically on the amount
present, could we draw the Figure to show the amounts present in each case, or  if that is too
much trouble,  label the right end of each bar with the amount present? That would convey a lot
more information.

p. 2-25, 11.9-12:  I believe that Wm.  Wilson has some work on this.  You should check with him
and, if appropriate, add a reference.

p. 2-32: Fix 1. 11. The letters on 11. 16-19 and below on the page are too little to read

p. 2-30 to 2-34: Where does that leave us? Is all of this necessary?

p. 2-33,1. 16:  What does radical (a) refer to?

p. 2-33,1. 23:  What does Table 2-3 refer to here? I do not recognize anything in Table 2-3 (p. 2-
11) that has to do with what we are talking about here.

p. 2-34: Table 2-13 is out of place. It is not introduced in the text until p. 2-40,1. 14.  The ref. to
Table 2-13 on p. 2-33,1. 12 is an error since there are no alkenes listed in Table 2-13.

p. 2-36,11. 21;  p. 2-37,11. 3,7, 28 ff - all too little to read.

p. 2-40,1. 23-33:  Why should this list and the list on pp. 2-28,1. 8-24 be any different?

p. 2-41 to p. 2-42 or p. 2-43:  It is beginning to seem as if there is  a lot of attention being given to
PAHs - and maybe not enough to other constituents.

p. 2-43, 11.  4-7:  Since Gorse  got much lower 1-NP in the vehicle tunnel, as  you  mention
elsewhere, I am skeptical of the < 10-20 % cited here. Also, where did the 43° C come from; in
dilution tube sampling the temperature can be as high as 52° C?

Bill Pierson
p. 2-43, 1.  12-36:  You can  get the same sort of thing happening under much  less serious
conditions.  This has already been mentioned on the comment on p. 2-24. Again, the Lee reprint
(your ref. on p. 2-64,11. 33-36) shows this.

p. 2-45, 11. 28: Five seconds sounds too long. With a flow of 450 cfin (typical for a dilution
tube) with a 1 ft2 cross section and 25 ft. from entrance to sampling point, residence time would
be more like 0.05 seconds (check my math).

p. 2-47, 1.  11-31 and Table 15: You many not at this late date want to include the references
mentioned above in p. 2-12,11. 5-29 but I mention it again here. In any case, there seems to be
some overlap between Sec. 2.3.1 and Sec. 2.5.1.

p. 2-49, 11.  15-16:  There  is a recent paper by Miguel et al., Env. Sci. Technol. 32, 450-455
(1998) which could be cited here.

Table 2-18:  Is this the gas-phase only?

p. 2-53,1. 11: I do not see anything in either trace on p. 2-54 labeled 3-NF (probably too low?).

p. 2-53,1. 18-19: Doesn't the 2-NF/2-NP ratio  depend on the concentrations of fluoranthene and

p. 2-57,1. 9-16: This work has now been published. [GertlerA-W.; Sagebiel,J.C.; Dippel, W.A.;
Farina,R.J.  (1998). Measurement of dioxin and furan emission factors from heavy-duty diesel
vehicles. JAWMA, 48: 276-278J.  But it seems out of place to be reporting it here; it should be
reported somewhere in sec. 2.2 or 2.3. Also the subject is taken up on p. 12-30,1. 26 ff. These
should be combined into one discussion. However, p. 12-30 has several misstatements. It was
the Fort McHenry Tunnel  (not Baltimore  Harbor Tunnel)  conducted  by  the Desert  Research
Institute (not Desert Research Laboratory).

p. 2-59, 11.  11-14:  This statement may  be untrue once the results of the NFRAQS (Northern
Front Range Air Quality Study) come out. Stay tuned.

In general  I believe that Chapter 2 has a great deal of material in it and is definitely worth
updating, if for no other  reason than to serve as a warning about how lacking  the research
covered in all the subsequent chapters is.

References to insert in 1. 2

Baumgard,K.J.; JohnsonJ.H. (1996). The Effect of Fuel and Engine Design on Diesel Exhaust
       Particle Size Distributions.  Warrendale, PA:  Society of Automotive Engineers; SAE
       technical paper no. 960131.  Reprinted from Diesel Exhaust Aftertreatment 1996 (SP-


Bill Pierson
Bagley,S.T; BaumgardJCJ.; Gratz^L.D.; JohnsonJ.H.; Leddy,D.G. (1996).  Characterization of
       Fuel and Aftertreatment Device Effects on Diesel Emissions.  Health Effects Institute
       Research Report Number 76.

Begeman,C.R. (1962).  Carcinogenic Aromatic Hydrocarbons in Automotive Effluents.   Paper
       No. 440C presented January 1962 at the SAE Automotive Engineering Congress. SAE
       Technical  Progress  Series  Vol.  6,  "Vehicle  Emissions", New York:    Society  of
       Automotive Engineers Inc., 1964.

Dolan,D.F.; KittelsonJD.B.; PuiJXY.H.  (1980).    Diesel Exhaust Particle  Size  Distribution
       Measurement  Techniques.  Warrendale, PA:  Society of Automotive Engineers; SAE
       technical paper no. 800187.

Kittelson,D.B.;  Kadue,P.A.; Scherrer,H.C.; Lovrien,R.E. (1988).  Characterization of diesel
       exhaust particles hi the atmosphere. Final report to Coordinating Research Council AP-2
       Project Group, March 1988.

Moore,G.E.; Katz^I.  (1960).  Polynuclear Aromatic Hydrocarbons in the Particulates of Diesel
       Exhaust hi Railway Tunnels and in the Particulates of an Urban Atmosphere.  Int. J. Air
       Poll. 2,221-235.

Tebbens,B.D.; ThomasJ.F.;  Mukai,M. (1963).   Paniculate Air  Pollutants Resulting from
       Combustion.  From ASTM  Special Publication No.  352, Symposium on Air-Pollution
       Measurement  Methods, presented at the Fourth Pacific Area National ASTM Meeting,
       Los Angeles, Oct. 5,1962 (published 1963 by ASTM).
Chapter 4

p. 4-2,1. 29: It is interesting that electrostatic precipitation is mentioned. This is an area that has
not been thought of as much as it should be,  given that there have been many  references
reporting considerable + and - charges on diesel exhaust.  In fact it might be useful to include a

p. 4.4, i. 9: v/hat are the obligatory nose breathers (name).

Table 4-1: I do not understand

p. 4-7, 1. 26: 98Tc has no metastable state (98mTc). Something wrong.

Fig. 4-3:  Redraw. Lines too faint.

Bill Pierson
Sec. seems all to have to do with rat data although it is called "alveolar clearance hi
animals" and I thought there is something wrong with rat data (see p. 10-2 11.11 -1 and 2)

p. 4-14,1. 7 and 9:  AM has been introduced without any definition (Alveolar microphages).

p. 4-18,1. 9 and several places on p. 4-20: What are Type 1 cells?

p. 4-18,1. 23 to 33: The word endocytosis is used three times.  I went to the dictionary and then
to the medical dictionary and I still don't know what it is. Put the cookies on the lower shelf.

p. 4-26,15: I do not see any evidence about extractability etc., of the 2-nitropyrene formed in
air from diesel exhaust (see comments "in Chapter 2). Is there no such work?

p. 4-27  all of Sec. 4.4.3:  How does extraction of nitro-pyrene etc. compare with something like
B(a)P; how soluble and how strongly bound to elemental carbon?

p. 4-28, line 8: What is a target site?

p. 4-29,1. 31: What is solvent green?

p. 4-31,1. 10-11 and 15-16: The  idea of the effect being proportional to the surface area of the
particles is very worth mentioning.

p. 4-31, I. 25-27: The lack of dosimetric factors for the gas phase seems a weakness of studies
conducted so far; after all, there  are carcinogens in the gas phase.  How do you get at  them?
Particularly (see Ch. 2) the 2-NP  and 2-NF in the ambient air (not present in diesel exhaust but
made from diesel exhaust in the ambient air)? Eventually these compounds da find their way to
the particle phase.  (This is more  a criticism  of the state of the science being reviewed than it is
of the reviewer.)

p. 4-32,  11. 28-29:  What animals are being discussed - more likely to be awake etc.
Chapter 5

p. 5-3,11.  11 to 13 are virtual repetitions of 11. 8 to 10.

p. 5-7,11.  19 to 21: "The ...exposure" -1 do not see the support for this.

p. 5-9, 1. 7:  How  is it possible to expose stevedores to diesel  exhaust particulate  without
exposing them to NO2?

Bill Pierson
p. 5-11, 11. 15-25:  What was the difference between the two Edling studies that such different
results were found?

p. 5-11,11. 33-34: Epidemiology is discussed in Ch. 8. Why then here?  Has there been no effort
to edit these chapters and see that they conform (this is not the first time that I see evidence that
this has not been done).  Also see p. 5-82, 1. 8.

p. 5-13:  I very much like the use of Tables here and throughout the document, to pull together
all of the health effects of one kind or another. These are extremely helpful.

p. 5-17, 1.13: Do we need Appendix A?

p. 5-18, 1. 6: vascular, not vacular.

p. 5-19,  1. 8-11:  The statement seems too strong since diesel paniculate is only a small part of
the paniculate matter to which the average person is exposed.

p. 5-23, 11. 34 ff:  This is important to emphasize.

p. 5-24, 1. 4: What does frank mean?

p. 5-24, 1. 12: Resjiiratory, not resoiratory.

p. 5-33, 11. 25-26:  What are "treatment-related"?

p. 5-45, 1. 3: What is chemotactic?

p. 5-53, 11. 7, 12,17,19, 20,23: What is BALF - defined on p. 5-42, 1. 21  but I think that is too
far away.

p. 5-86,  1. 34 (and also in Chapter 4 p. 4-1,1. 31 and in Chapter 6 p.  6-3,1. 32) we speak of a
"carbonaceous core".  This is quite misleading and may not be very helpful as we think of diesel
exhaust extractive processes hi the  lung. The particle is not constructed like an onion with an
organic skin around a carbonaceous core of carbon.   Rather, the carbon is linked in chains or
aggregates like a bunch of grapes, with the organic material filling hi the interstices.

Chapter 6

p. 6-2, 1. 1:  Take out the comma after research.

p. 6-3,  11. 6-7: "The weight of evidence from the available toxicological data on diesel exhaust
indicates with high confidence that inhalation of diesel exhaust can be a respiratory hazard, based


Bill P'.erson
on findings in multiple controlled laboratory animal studies in several species with suggestive
evidence from  human occupational  studies."   From context, this refers to the two previous
chapters, or else this chapter is out of place.  So I re-studied these chapters. Chapter 4 is largely
about rats only, usually in an overload condition.  Chapter 5 gave data for humans exposed to
gross diesel exhaust (usually without  isolation of the particulates); I see a statement, "The overall
conclusion of these studies is that reversible changes in pulmonary function in humans can occur
in relation to diesel exhaust exposure", without offering any support; I see nothing in the long-
term exposures, and in fact quite the  opposite. It is stated (p. 5-11,  11. 26-27), "The absence of
reported noncancerous human health effects, other  than infrequently  occurring effects related to
respiratory symptoms and pulmonary function changes, is notable."  Table 5-1 (Human studies
of exposure to  diesel exhaust) showed conflicting  evidence of short-term respiratory reactions
(cough, phlegm, etc.) to diesel exhaust, but the patterns were not consistent or in some cases
attributed to other causes. The animal studies (Table 5-2. Short-term effects of diesei exhaust on
laboratory animals) definitely show effects (largely attributable to CO and NO2 - see bottom of p.
5-23 ff.)  But "Little evidence exists, that subchronic exposure to diesel exhaust impairs lung
function" (p. 5-24, 11.  11-12).  Chronic exposures  (Table 5.3 and 5.4) often indicate  no effect.
Effects on pulmonary function (Table 5.5) again are conflicting.  Histopathological effects (Table
5-6) seem to  agree on inflammatory changes, otherwise results are all over the place.  Effects of
exposure to  diesel exhaust on the pulmonary defense mechanisms of laboratory animals (Table
5-7) are very mixed.   The effects of exposure to diesel exhaust  on the immune  system of
laboratory animals (Table 5-8) seems to show nothing.  Same for effects of exposure to  diesel
exhaust on the liver of laboratory animals (Table 5-9).  Same for effects of exposure to  diesel
exhaust on the hematological  and cardiovascular systems of laboratory animals (Table  5-10).
Same for effects of chronic exposures to diesel exhaust on serum chemistry (Table 5-11).  As for
the effects of chronic exposures to diesel exhaust on microsomal enzymes of laboratory animals
(Table 5-12), again the results appear to be mixed.   There does seem to be an effect for chronic
exposures to diesel  exhaust on behavior and  neurophysiology  (Table 5-13).  For  effects of
chronic exposures to diesel exhaust on reproduction and development (Table 5-14) there appears
to be little or no effect. "The most readily identified acute noncancer health effect of diesel
exhaust on humans  is its ability  to elicit subjective complaints of eye, throat and bronchial
irritation  and  neurophysiological symptoms  such as headache, lightheadedness,  nausea,
vomiting, and numbness and tingling of the extremities" (p. 5-81,  11. 7-10).  "Studies on the
acute health  effects of exposure to diesel exhaust  in humans, experimental and epidemiologic,
have failed to demonstrate a consistent pattern of  adverse effects on respiratory morbidity; the
majority of cases offer, at best, equivocal  evidence for an exposure-response relationship."  My.
question is:  In view of all this, how  can one honestly make the statement  at the too of this

p. 6-5,  1. 14: I raise again the question of whether this chapter is out of place. Carcinogenicity
does not get discussed until Chapter 7 and 8.

p. 6-13,11. 19-20:  Again the same question.

p. 6-26,  11. 30-32: Again the same question. This chapter seems to belong after Chapter 8.

 Bill Pierson'
 Chapter 7

 p. 7-1, 11. 19-21: How does this square with statements elsewhere that the unextracted core is
 the only active part of the PM? (See for example p. 7-28,  11. 14-18.)

 p. 7-2, 11.7-10: What health assessment documents are referred to here? (giveref.)

 p. 7-2, 11. 13 ff:  These are all rat studies. Therefore, is the statement on p. 7-10,  11.  16-21 any
 good?  Is the whole section of any import?  I keep hearing that the rat data should be rejected
 because of overload.  Is this true or not?

 p. 7-18, 11.30-32:  This is just nonsense. You do not decide to accept a finding as statistically
 significant because the  signal  is supposed to be low anyway.  If it isn't statistically significant,
 then you can't talk as if it is.

 p. 7-44, 11. 24-26:  I believe the statement, "In summary, based on positive inhalation exposure
 data in rats and mice, intratracheal instillation in rats, and injection or skin painting in mice and
 supported by positive mutagenicity studies, the evidence for carcinogenicity of diesel  exhaust is
.considered to be  adequate." But at this point, mutagenicity studies have not been discussed yet
 (they come in Chapter 9). So the statement, which is important and much more to the  point than
 some  of the stuff in Ch. 12,  is a statement that  is worth repeating at a prominent point near
 (perhaps in Ch. 12), and perhaps at the end, ofT the criteria document
 Chapter 8

 Beginning with this chapter and sprinkled throughout all of the subsequent chapters, there are 1Z
 (!) references to  Garshick and his work.  I believe that this  represents in part a deficiency in
 organization of the chapters.  I have  already referred to other instances which indicate an
 organizational problem.

 Nonetheless, I do like in this chapter the organization by each set of authors (Sec. 8.2, and  also
 Sec. 8.3, and Sec. 8.4.

 p. 8-33,  1. 5,7,8,10,11,12,13,14,17,21: What does (do?) MERmean?

 p. 8-52,  1.5:  What are dorsopathies? Back problems?

 p. 8-54,  11. 30-32:  "Animal data suggest that diesel exhaust  is a pulmonary carcinogen among
 rodents exposed  by inhalation..." This Chapter  8 is a review of Cohort studies, case-control
 studies of (human) lung cancer,  case-control studies of (human) bladder cancer. I do not see

Bill Pierson
what the statement in quotes is doing here - even if it is true as in another form was discussed
earlier.  But I agree with what it leads into (p. 8-54,11. 32-35).

p. 8-62, 11. 24-32: I agree.

p. 8-63, 11. 2-3: "Among the 10 lung-cancer case-control studies reviewed in this chapter, only 2
studies did not find any increased risk of lung cancer."  This is a distortion. The fact is that 2
found no evidence and indeed opposite, evidence. 7 found no evidence one way or the other, and
3 found positive evidence. The statement in quotes leaves the impression that 8 out of 10 studies
found a statistically significant increase in lung cancer.

p. 8-63, 1. 20: And in addition the document says the results in these studies are underestimated
at best.  I do not understand this and I think there should be an explanation

p. 8-64,11. 6 to 26: I agree that the best study was that of Garshick. I would not put Steenland et
al. (1990) into the same category.

p. 8-64,11. 33 ff:  I have the same problem with the discussion of bladder cancer,-of which you
say 4/7 of them found increased risk in occupations with high potential  exhaust exposure.  I
believe your ending sentence on p. 8-65,  11.9-13 takes care of bladder cancer.

p.8-69,1. 41 to p.8-70,1. 6:  What does the equivocation mean?

Chapter 9

My only  comment is that several papers from Ford have been cited but perhaps the most
important one (Pierson et al., Mutagenicity and chemical characteristics of carbonaceous
participate matter from vehicles on the road, Env. Sci. Tech. 17, 31-44 [1983]) probably should
be included because it is on-road.
Chapter 10

p. 10-1 and elsewhere: DE and DPM seem to be used interchangeably.  If there is a difference, it
should be clarified; if not, then use one or the other.

p. 10-1 and following: I thought the rat data were to be ignored (because of reasons stated on p.

p. 10-1, lines 14-15:  "Epidemiologic data suggest that there is a small increased cancer risk in
humans following long-term exposure to diesel  exhaust". Again, in Chapter 12 - line 19-20:
"Epidemiologic data are  strongly suggestive  of a carcinogenic  hazard to the  lung under


Bill Pierson
occupational exposure conditions". These statements seem inconsistent.  But now look at Ch. 8
which deals with the epidemiologic data.  Some of these studies turned up negative results,
others had some very serious limitations (like not controlling for smoking), some had too little
power to  mean anything, etc.; only the Garshick  1988 study among the cohort  studies was
convincing.  The case-control studies again come down to a study by Garshick (1987). Finally,
the case-control studies for bladder cancer, sec. 8.4, had many limitations; the statement (p. 8-63,
1. 2-3) that "among the 10 lung  cancer case-control studies reviewed in this chapter, only 2
studies did not find any increased  risk of lung cancer" is a distortion.  According to  my reading,
only 2 showed any support, and the rest were either negative or produced nothing significant one
way or the other.  Similarly, "of the seven bladder-cancer case-control studies, four studies found
increased risk in occupations with a high potential diesel exhaust exposure". This is  a distortion.
The fact is that of 7 cases given, five gave no support, one gave positive support, and one appears
to be positive or mixed.  Finally, on p. 8-70 it is stated in summary, "based  on the human
evidence alone, diesel exhaust is close to being a known human carcinogen".  I have a hard time
reconciling these various quotes.

When these problems  were  discussed at the meeting, some  felt  that statistically insignificant
numbers all lying on the same side of zero can be taken to signify a positive effect. I confess to a
good deal of discomfort with that. I would rather put my faith in the  studies by Garshick which
are significant and in any case, I  still maintain that the statements, "among the 10  lung cancer
case-control studies reviewed in  this chapter, only 2 did not find any increased risk of lung
cancer," and "of the seven bladder-cancer case-control studies, four studies found increased risk
in occupations with a high potential exhaust exposure," are both distortions for the reasons given
above -1 would even say grqss distortions.

p. 10-3, line 9: What are athmyic mice? (Missing the thymus gland?  What for?)

p. 10-4, 1. 1: Add a reference. Also, tie it to the material on p. 2-15, 1.  15  to p.  16, 1. 2 and
conform.  They should all say the same numbers and have the same refs., etc. in both places.

p. 10-4,1.  25: How do you know it is surface-area-associated?

p. 10-11,11.  1-36  and following page:  See p. 2-52;  2-NF is probably more appropriate to worry
Chapter 11

p. ll-l, 11. 2-4: The presence of known carcinogens such as B(a)P had been known since 1955.
What Huisingh et al. did, in 1978, was to show that diesel exhaust contained substances that did
not require S9 activation to cause a positive Ames  result.  The compounds known up to then
would not have done that.  So therefore there were more compounds that had not been identified.
Some of them eventually proved to be  very "hot".


Bill Pierson
p. 11-1,11. 11-12: Forgot to mention Chapter 9 mutagenicity.

p. 11-1,11. 31-33: Is it not true that many, even most, of the positive studies also suffered from
the same methodo logic limitations?

p. 11-2, 1. 30  to p.- 11-3, 1.  19:  Apparently we have IARC concluding that the evidence for
carcinogenicity of whole engine exhaust in experimental animals was adequate but the evidence
in humans is limited. This is said to be in agreement with the conclusion of EPA. Under EPA's
Proposed Guidelines diesel exhaust is considered a likely carcinogen by the inhalation route of
exposure and is considered to be at the upper end of this grouping.

p. 11-4, line 9:  such as dinitropyrene

p. 11-6, 11. 1-2:  "The potency of the other diesel  emission samples was not estimated directly
because of the weak response in the skin tumor initiation test. I do not understand this statement,
or what follows.  On the face of it, it sounds as if the potency test was rigged.  The material
immediately following does not help (or does not help me).

p. 11-10:  Why is the Garshick data being discussed here?  It has already been taken up in Ch. 8
and used in Ch. 10.  Is there a plan to resolve the Dawson-Crump disagreement on p. 11-10, line

p. 11-17, 11. 25-26:   If cancer induction in the rat bioassays was observed only under particle-
overload conditions, why are we continuing to use the rat data? See p. 11-19,11. 11-28; p. 11-20,
11. 3-10; and 11-21,11. 3-",

p. 11-22, 1. 20:  "DE is considered to be a probable human carcinogen." This conflicts with
Chapter 10 p. 1,1. 14.

p. 11-23,11. 9-10: Something wrong with the sentence.

p. 11-23,1. 10: What is a MLEs? (See also p. 12-26,11. 6 and 7).

p. 11-23,1. 22: I think we decided this is upper-bound, not lower-bound.
Chapter 12

p. 12-1,1. 19: Is the word "strongly" too strong?

p. 12-2,1. 15: What sulfur compounds are carcinogenic?


Bill Pierson
p, 12-2,1. 22-24:  This will always be a bug-a-boo as long as the technology is changing rapidly
but certainly is true right now.

p. 12-2,1. 27 ff: Needs to be always kept in mind in all of this.

p. 12-5,11. 17: "Current data do not support confident identification of health hazards other than
for the respiratory system."  I think this is a fair statement and it certainly reflects where I come
out on all of this.

p. 12-8,  1. 19:  You have  already MOA as standing for mode of action (see many times on
previous page). You should not write it out here.

p. 12-10,11. 8-9:  I agree that "the most convincing evidence that exposure to DE can induce lung
cancer in humans comes from case-control and cohort studies among U.S. railroad workers and
truck drivers".  But I also agree  that it does not deserve to be classed as a "known" human
carcinogen (p. 12-11,1. 23) - in fact I might quibble with "highly suggestive" on p. 12-11,1. 22.
(How about just suggestive?)

p. 12-13,1. 33 and p. 12-14,1. 5: I don't see the difference between "very likely" and "probable".
(I would see a difference between "likely" and "probable".)'

p. 12-17,1. 16: Delete to note. It is interesting that what you say is true but not interesting to
note it.

p. 12-20,1. 16: 17 ug/m3

p. 12-24, 11.  9-22: I gather that the problem is that there is a lot of rat data but they have the
overload problem. If you throw out the rat data then you have mice  and hamsters left (which I
read somewhere in the document have very much less extensive data).  So you wish you didn't
have to throw out the rat data. But the  argument has nothing to do with the amount  of data you
have; it is all either compromised  by the overload problem or it is not.  If it is, then you have to
throw it out without looking over your shoulder to see what you are left with.  In particular the
negative findings on rat and mouse data (p. 12-24, lines 12-14) should  have  nothing to do with
the decision on the rat data.  You argue for keeping the rat data because they show a response
and you know that humans show a  response in  epidemiologic data.  But you are getting the
human and the rat to agree for the wrong reasons; the rat  is doing it because it is  in overload
which is a situation that is inapplicable for the human case.  I think the only scientifically
defensible procedure (if I understand the situation correctly) is to throw out the rat data  if it is
compromised by overload and to make your case with what you are left with.  If this raises the
estimated risk (p.  12-24,1. 34), then so be it

p. 12-26,11. 6-7:  What are MLE?  (Same question as earlier  in Ch. 11 p.  11-23,1. 10).

Bill Pierson
p.  12-27,1. 22: I did not initially recall any discussion about exacerbation of asthma.  Perhaps it
should be noted here where that is discussed (p. 5-6, 11.  1-2; p. 5-18, 11. 27 ff; p. 12-34, 1.
anywhere else?)
p. 12-33  ff:  This is pure speculation and should be left out.  In the present document we are
supposed to be dealing with what is known, not what is not known.

p. 12-29, 11.  14-18: As I understand it, Cal - EPA has subsequently decided that diesel exhaust
(or emissions?) is indeed a toxic air contaminant.  The measured or more careful response in the
present document is a long way from that. I think that there is a world of difference between the
two readings and therefore I would question the characterization on lines 17 and 18.

p. 12-30, 11.  26-27:  Please  see p. 2-57, 1. 9-16 and  my review of that section earlier.  I see no
reason to have this is 2 places and actually I think that neither is the correct place but rather in
Sec. 2.2  or 2.3.  But in any case 1. 26 should read Fort McHenry Tunnel (not the Baltimore
Harbor Tunnel).  And  line 27 should read  Desert Research Institute (not Laboratory).  Also,
though dioxins an4 furans (not mentioned here)  are emitted from Diesels, these measurements
indicate that it is only 0.28 ±0.13 ng of TEQ equivalent per vehicle-mi, which may be worth

p. 12-30,11.  33 ff:  This really sounds like a very big stretch, in view of your estimate of 60 ng
TEQ from trucks vs. U.S. emissions from all sources of 3000 ng.  I would stop the paragraph
after the word source on p. 12-30,1. 33.

p. 12-30,1. 31: Since we are now in May, what do you want to do about the April 1998 date?
p. 12-32,1. 9-11: I do not agree with the wording which indicates that after aging you don't have
to worry  about  diesel  exhaust as much because it is diluted and has been transformed  into
compounds of less activity. In the first place, however dilute, people are exposed all the time to
aged diesel exhaust and not just near the roadway.  Second, some of these compounds may be
more active  than what you  started with as you acknowledge on p. 2-56,1. 5-6 and on p. 2-58,1.
29.  Thus, as Chapter 2 points out and as I have pointed out in my review, you get a whole set of
different compounds such as 2-nitropyrene (not found hi fresh  exhaust)  instead of  the 1-
nitropyrene found in fresh exhaust.  The two are about comparable in mutagenic activity.

p. 12-33, 1.  2-3:  This statement is not necessarily true since we are  talking here  about .black
smoke which derives most of its effect from absorption. The absorption cross section, unlike
scattering, is essentially particle size independent.  So that seeing smoke or soot says nothing
about the size of the particles.

p. 12-33, 11. 26-27: Where did the onset of a common cold come from?  I never in the whole
document saw any comparison to the effects of episodic DE exposure to the onset of a common


Bill Pierson
p. 12-34, 1. 16:  Where did the number 19 come from?  I never saw it in the carcinogenicity
discussion. This is supposed to be a summary, not a place to introduce new material - without a
reference (!)

p. 12-34, 11. 28-29:  I have no idea what this refers to.  Where in the document is  there any
support for this? Again, this is not the place to be bringing in further information.  This should
be expanded and moved somewhere else.

p. 12-35,11. 17-19:  "EPA takes the position..."  I have a very strong objection to this wording.
This document is not the place for EPA to be taking a position on this or any other matter.  The
document is a place where the science bearing on a given question is pulled together. I think I
can agree with a conclusion but not offered as a position.  This document is not the place for

p. 12-35,11. 22-26: This is pure speculation, has not been discussed anywhere in the document.
The sentence should be dropped.

p. 12-36; p. 29-30: Could the risk be,  say, zero? If so, or whatever lower-bound risk there is, it
should be stated,  I still do not understand when you get to p. 12-37, 1. 23 you cannot give the
centroid of the risk and a ± c and be done with it.

p. 12-37,1. 22: Are we being consistent in upper/lower between here and p.  11-23,1. 22 which
we had decided should be lower-bound.

p. 12-37,11. 30-33: I thought that the gap in the particle deposition spectrum was below 1 urn, in
which case I am not sure diesel exhaust will necessarily deposit deeper in the lung (however, I
am not  sure of this).  But the statement about their small size  does not give half of the story
because these particles are assemblages of small spheres; see the comment above on p. 5-86,1.

p. 12-37,11. 30-33: In any case you don't know that because you don't know what properties are
responsible for the alleged effect of small particles.

p. 12-38, 1. 1-4:  It should be emphasized that most engine modification  steps to reduce NOX
increase the particulate emission rate and vice versa.  A more appropriate place for this is Sec.
2.2, as mentioned in the Chapter 2  review.

p. 12-38, 1. 7-8:   "These special  population subgroups are difficult to enumerate,  but they do
exist."  This sounds like conjuring spooks.  I am sure you are right but it sounds very odd.

From:        "Stayner, Leslie T." 
To:          ROBERT FLAAK 
Data:        5/21/98 18:55
Subject:     RE: Preparation for the May 5-6 Meeting

 Dear Robert

Dear Robert:

I am sending this E-mail to formally express my comments on the EPA
draft risk assessment on diesel exhaust.  I hope it is okay for me to do
so using this medium rather than regular mail.  Please let know if it is
not and I will send you a letter.

I have limited my review primarily to the qualitative and quantitative
health analyses, and particularly those concerning the epidemiologic
data.   I am attaching a WordPerfect document containing detailed
comments that I have on the document.   I will try to summarize my more
general comments that I expressed at our meeting.    I have divided my
comments into those pertaining to the non-cancer and cancer qualitative
and quantitative sections.

Non-Cancer Qualitative Assessment

Overall, this section was well written and I only have a few comments.
First, the review should give greater emphasis to the fact that the
epidemiologic studies that have been performed were h»nhly unlikely to
be able to detect an effect on the respiratory system, since they"were
cross-sectional studies that only included active workers.  It is well
recognized that this type of study may lead to a serious bias due to
workers with the disease leaving employment.

Second, the review really must discuss the implications of the studies
of PM2.5 and PM10 for diesel exhaust.   Diesel is a major contributor to
PM and thus most likely contributes to the increased mortality and
morbidity that has been associated with environmental PM exposures.

Third, the new research on asthma induced by exposure to diesel exhaust
is treated unevenly in this document.  It seems that this was added to
the document at a late stage, and that this is an important issue that
needs to be fully evaluated.   In some sections, it is not discussed at
all.   I think it also may be a bit of exaggeration to suggest that
diesel exhaust may be responsible for the increasing trend in asthma

Non-Cancer Quantitative Risk Assessment

The benchmark dose section seems to be totally illogical.   This may
reflect a lack of clear EPA agency policy on how to apply this procedure
to continuous outcomes.   Why was a 10% change used as the benchmark for
some outcomes (e.g. lung weight) and a 200% change for others  (e.g.
bronchiolar lavage enzymes)?   I like to the think of the benchmark for
categorical variables as a replacement for a NOAEL.  A 10 or 1 % risk
level  is essentially the level of risk at which a toxicologic or
epidemiologic study is unlikely to detect an effect.   However, the same
does not hold true for a continuous variable and it is unclear to me how
one picks a benchmark in this case.

Regarding the issue of whether or not to use a safety factor of 3 for
interspecies variability, I believe that this  is justified and that one
could possibly even justify a higher factor.   I do not believe that we
know enough about the differences in sensitivity of the rats versus
humans to argue that the rat is more sensitive than the humans.  In
fact, the epidemiology suggests  just the opposite is true!   The risk


estimates that we derive from using the human data are generally much
higher than those derived using the animal data.

Cancer - Qualitative Review

The description of the strengths and weaknesses of the individual
studies is not well written and at time unbalanced.   The reader is left
with a false impression of which studies are weak and which are strong.
The strongest studies by far are the studies or railroad workers and
truckers, since these are the populations with the best defined

The review overemphasizes the importance of smoking in explaining the
findings of these studies.   Cigarette smoking has been demonstrated by
several investigators to be unlikely a large source of bias in
occupational studies.  Furthermore, in the epidemiologic studies that
did control for smoking it was not found to have much effect.

The review also grossly overemphasizes the potential bias related to the
use of death certificate information for lung cancer.   This is truly a
non-issue, since lung cancer is one cause of death that has been found
to be generally reliably diagnosed on death certificates.   Furthermore,
since there is no treatment for lung cancer the difference between using
mortality and incidence data is minimal.

Finally, -the silence of this document on the issues related to the
exposure-response analysis for the Garshick cohort is puzzling.  The EPA
ought seriously consider the debate between Stan Dawson and Kenny Crump
and discuss these issues in the next draft of the document.

Cancer Quantitative Risk Assessment

In general, I liked the approach of presenting risk estimates from many
different data sets and analytic methods.   I agree that the greatest
emphasis should probably be given to the results from using the
epidemiologic data, but I would not at this time dismiss the results
from using the toxicologic data either.

I realized that there is a problem with the approach that was used by
McClellan and by the EPA in this document.    This approach uses a slope
for the relationship of. lung cancer and duration of exposure from the
case-control study by Garshick et al.    However, the model that this
was taken from does include exposures prior to 1959.  Thus exposure was
underestimated in this model, which would result in an overestimation of
risk-   I have the following alternative suggestions for future
epidemiologic risk analyses:

1) Use the overall relative risk and the range of exposure estimates
from the case-control and/or cohort studies by Garshick et al. to
compute a slope.  This slope of course would need to be adjusted for
differences between occupational and environmental exposures.

2) Similarly,  use the overall relative risk estimates from the
meta-analysis or from other epidemiologic studies to compute a slope.

3) Use the results from the NIOSH truckers study to conduct a risk
assessment.   Kyle steenland and I have a paper in press in which he
uses this approach to estimate occupational risks.

The only other option I can see is to wait at least five more years
until the results from the NIOSH study of diesel exposed miners is
completed, or until the study that the Health Effects Institute is
initiating is completed.   I think clearly that this option is
unacceptable from a public health viewpoint.


These are all my thoughts and comment a.  z hope they are helpful to you
and the authors of this document.   Thank you for allowing me to
participate in this very important scientific review.

Leslie Stayner

 «R-D IES-9. EPA»

CC:          "Stayner, Leslie T." 
                                    A- 100

                        Specific Comments
                           Leslie Stayner

Page 4-1,  1st paragraph, 2nd sentence - The statement that the "tumorigenic
response... is the result of pulmonary overloading" is far too strong. The
overload mechanism is still just a hypothesis and other mechanisms may play a
role as the document discusses in subsequent sections.  I would suggest
modifying  the statement using words such as "may be primarily the result of.

Page 4-1,  2nd paragraph, 1st sentence - The statement is not entirely true since
one of the studies did report an increase in lung tumors among mice exposed to
filtered exhaust (Heinrich et al. 1986a).

Page 4-3,  2nd paragraph, 2nd sentence - although this statement is supported
by the toxicologic data, I don't think it can be supported by the epidemiologic
studies. Most lung cancers in humans are bronchogenic in origin and its not
clear from the studies performed to date if the excess of lung cancer observed
are from tumors of the  bronchus or alveoli.   Based on species differences in
lung deposition and clearance one might argue that  the greatest dose in humans
will be at the bronchii.

Page 4-16, 2nd paragraph - A recently published doctoral dissertation by Eileen
Kuempel which modelled lung clearance rates in coal miners should be included
in this discussion.  This model revealed nearly complete shutdown of clearance
in these miners, and a  very different pattern of lung dust accumulation in
humans than in rodents.

The last sentence needs modification.  An excess of lung cancer was noted
among workers with pneumocionosis in a study reviewed later in this document.
Also a recent paper by Dr. Peter Morfeld from Germany appears to indicate an
increased  risk of lung cancer among German coal miners (see abstract from the
last Inhaled Particles Meeting or the 1998 ICOH meeting in Zimbabwe).

Page 4-19, 2nd paragraph, lines 4-6 - The review shold note that there is
evidence than in the role of interstitialization or particles is greater in primates
than in rodents. The dissertation by Kuempel (1997) discussed above makes
this point for coal dust, and there is I believe also evidence from experimental
studies of  monkeys.

Page 4-20, 4th paragraph - macrophages may not be the primary  reservoir in
primates since interstitialization may be more important as discussed above.

Section 5 - It seems really odd that this section contains no discussion of the PM
2.5/10 problem. Diesel is a major contributor to PM2.5 and thus the studies of


PM (e.g. the six city studies) are clearly relevant to the discussion of the
potential  health effects of diesel.

Page 5-11, lines 33-34 - it seems that several of the studies discussed in this
section did attempt to evaluate chronic as well as acute respiratory effects.
Thus I am not sure why this statement was made, unless you mean that the
studies were too small or inadequate for this purpose (which I would agree with).

Page 5-17, lines 4-6 - The fact that these studies only included active workers is
a major limitation that should be stressed in this review. Workers who develop
symptoms or severe respiratory disease are obviously not going to stay in these
jobs and thus these studies are likely to have had a substantial "Healthy Worker
Survivor Effect".  The  effect of this is clearly to bias these studies towards not
observing an increased risk of respiratory disease, which is likely to explain-the
negative and inconsistent findings from these studies particularly for pulmonary
function and chronic effects.

Page 5-19, lines 9-11 - The suggestion that DPM may be involved in the rising
epidemic of asthma in  this country seems to be a big stretch.  Have diesel
exposures in the general population increased over the last decade or so
paralleling the increase in asthma?  A better case needs to be made for this
statement or else it should  be modified or dropped.

Page 5-30, line 30 -31 - These sentences suggest that the pulmonary function of
these monkeys was "increased".  Is this a typo and shouldn't it be decreased if
this evidence of obstructive lung disease?

Section 6.2 - The human and primate studies seem not have been used at all in
attempting to derive an RFC.  Couldn't some attempt have been made to
identify NOAELs or LOAELs from these studies?

Page 6-4, 2nd paragraph, lines 9-10 - The model developed by Yu and Yoon
(1982) has some serious limitations and associated uncertainties that need to be
discussed in this section.  In particular, the human model has not been
adequatetly tested and is largely based on scaling parameters from rats to

Page 6-15, 2nd paragraph - The text says that a uncertainty factor of 1 was used
for interspecies variability;  however, in the cover letter from Mr. Flaack and in
the summary it indicates that an uncertainty factor of 3 was used.

The justification for choosing an uncertainty factor of 1 is very weak.  First of all,
while some allowance should be made for the use of a dosimetric model for
interspecies scaling, as already mentioned the model that is being currently
used is subject to large uncertainties.  Furthermore, the argument that primates


may be less sensitive than rats based on the study by Lewis et at. (1989) seems
totally unjustified.  This study only had 1 exposure level which showed a
response.  It is unclear how one can based such a statement on this database?

Page 6-19, Table 6-2 - What is the basis for the different benchmark definitions
used for these different endpoints?  How can you justify using a 10%
benchmark for some outcomes and a 200% benchmark for others?  The whole
approach used here for these continuous outcomes is confusing.  What is the
objective here? Is this an attempt to identify a NOAEL with modeling? If so I
am not sure that this is the right way to go about this. If this is the objective then
perhaps the benchmark should be set a level corresponding to a level at which
the study was unlikely to have the statistical power to detect an effect.

Page 7-25, line 19-20 - This sentence is illogical. How  can a single monkey
demonstrate a significant lung tumor response?  Do you mean that overall there
was no significant increase in lung tumors among this group of monkeys relative
to controls?
Page 8-1, 1st paragraph - Similar statements were made in previous sections.
Thus I would suggest dropping this paragraph since it is redundant.

Page 8-4 - The issue of exposure to coal dust is brought up as a potential
confounder of the study by Waller et al. (1981). The review rightly points out
that coal dust is not generally believed to be a lung carcinogen, which makes
this issue moot.  However, isn't the issue exposure to combustion products of
coal rather than to coal dust exposure itself?   This may be a  more complex
issue since combustion of coal has been associated with lung cancer in some
studies as this document discusses at a later point.

Page 8-5, lines 4-5 - The document here and in several other places
overemphasizes the possibility of disease misclassification due to the reliance
on death certificates.  For lung cancer this is truly  a non-issue, since studies
comparing death certificate information to pathologic studies  have demonstrated
a high degree of concordance.

Page 8-6, lines 5-6 and 15-16 - These 2 sentences are redundant.
Furthermore,  the document overemphasizes the potential role smoking might
have played in these studies.. First of all, it should  be made clear that smoking is
only a potential confounder and in fact several authors (ie. Axelson, Siemyatiki)
have demonstrated that it is a best a weak confounder in occupational studies.
It should also be noted that in several studies where smoking was controlled for
(i.e., Steenland and Garshick) it did not change the findings.

Page 8-7, lines 7-8 -1 suggest changing "confirmed the" to "is consistent" with


the healthy worker effect.  One can't really confirm whether there is a healthy
worker effect or not.

Page 8-9, 20-22 - Again this review overemphasizes potential confounding by
smoking.  The fact that the vital status of 1,765 (5%) of the cohort was unknown
is a trivial issue that is not worth mentioning. This is a very small percentage,
which is typical of studies of this type. The impact of this would be
inconsequential regardless of how the person-years are handled.

Page 8-9, lines 28-34 - What were the results from this smoking survey that is
discussed in this paragraph?

Page 9-11, lines 2-3 - This last sentence should be dropped.  The fact that
asbestos and smoking were not controlled for is irrelevant for this negative

Page 8-12, lines 32-33 - Again, death certificate information is fine for lung

Page 8-14, lines 1-2 - You should mention that the time dimension for the Cox
model was time since first entry into the cohort, and that the model controlled for
birth year and calendar time.  This is important since it turns out that if you
model using age as the time dimension, you do not see an exposure-response

Page 8-14, lines 32-25 - In discussing the exposure-response analysis it should
be noted that the investigators only included exposure after 1959 and that the
duration of expor,jre to diesel prior to that time was unknown.

Page 8-15 lines 7-9 - This sentence may be dropped, since it was already stated
on the previous page (lines 29-31).

Page 8-15, lines 24-25 - change may have been to was.

Page 8-15, lines 29-30 - This study did in fact examine the effect of years of
exposure; however, this analysis did include years of exposure  prior to 1959.

Page 8-15, line 33-34 - The word demonstrate is too strong, I would change this
to suggest.  Also I would change risk to relative risk.  A relative risk of 1.5 for
lung cancer is actually a large risk on an absolute scale (about  2%).

Page 8-15 - This review should at least mention the problems with
underascertainment in deaths after 1976 in  this cohort, that were discovered by
K. Crump and acknowledged by the authors of the Garshick study. It should
also at least briefly discuss the controversy over whether there  is or isn't an


exposure-response relationship in this study.

Page 8-17, lines 26-30 -This is not an overstatement and in fact is probably an
understatement.  Axelson has shown that, in general, confounding by smoking
could only cause a relative risk of 1.5 for lung cancer.

Page 8-18, lines 29-30 - Add the  word "potentially" before confound.

Page 8-22 -1 would think this might be considered a "hypothesis generating"
study, which would not meet the criteria  for inclusion in this review.

Page 8-23, line 15 - The review of this study should note that the greatest
limitation of this study was its lack of information on whether or not individuals
were actually exposed to diesel exhaust.

Page 8-25, line 13 - add the word cell to after the word small.

Page 8-31, line 22 - Change relative risk to odds ratio.

Page 8-31, line 35 - Put the word significant before excess.
Page 8-26, line 14 - A reference needs to be added to support this statement.  I
would think that the use of surrogates for interviews could bias the results either

Page 8-28, line 33 - Actually the transition from steam to diesel engines began in
the 1940's.

Page 8-30, line 11 - Change relative odds to odds ratio

Page 8-33, line 29 - Change cancers to carcinogens,

Page 8-35, lines 15-21 - This paragraph makes this study sound very weak.  In
fact, this is one of the best studies that have been performed.  The strengths of
this study need to be emphasized, including: the  measurements of exposures,
the availability of smoking data. I don't think you can say that the limitations of
this study resulted in  an underestimation of risk.

Page 8-49, lines 3-4 - The fact that this study demonstrates the utility of city
directories would seem to be irrelevant to this review.  .

Page 8-59, lines 34-35 - There is a very recently  released study of German
miners exposed to diesel exhaust that observed a 2 fold excess of lung cancer.

Page 8-60, lines 9-13 - NIOSH and NCI  have in fact  been able to identify a

                               A- 105

cohort of miners exposed to diesel exhaust with virtually no exposure to diesel
exhaust.  The latency period (length of followup) of this cohort is not too short to
study at this time.

Page 8-63, lines 10-11.  Why is it interesting to note this here?  Everyone
knows this already.

Page 8-67, lines 2-3 -You should note here that the studies that control for
smoking had little effect in the studies did have information on smoking.

Page 8-67, lines 21-22 - The study by Steenland was not a study of railroad
workers as implied by this sentence.  Modify the sentence by adding "and
truckers" after railroad workers.

Page 10-4, lines 35-36 - It is incorrect to suggest that epidemiologic data do not
support a mutagenic effect.  In fact, the fact that an excess of lung cancer has
been observed in studies of working population who are not exposed to high
enough levels to induce overload of the lungs would provide indirect evidence
that mutations may be involved.

Page 11-2, line 2  - Substitute "control for" for "eliminate most". One can really
never eliminate the potential for confounding.

Page 11-7, line 22-23 - It should be noted that these potency estimates are for
occupational exposures and not for environmental exposures. I suspect this is
also true of several of the other potency estimates cited.
Page 11-7, lines 29-32 - Hattis et al. presented separate risk estimates for
smokers  and non-smokers.

Page 11-10, lines 9-10.  Again, one can not "eliminate" confounding in
epidemiologic studies.  It is  only possible to attempt toxontrol for it.

Page 11-10, lines 27-31  - It seems that this document should more directly
discuss the issues in the debate between Dawson and Crump on the exposure-
response analysis, and if possible offer insight  into this issue.

Page 11-11  - Substitute unit risk for mortality.

Page 11-11, 2nd paragraph - The possibility of the bladder as a target site
should be at least discussed based on the epidemiologic studies that were
reviewed in this document.

Page 11-11, last paragraph - This paragraph suggests that the primary site is
assumed to be the alveoli.  This would be an appropriate assumption for rats,


but it is not as obvious for humans. Most human lung cancers arise from
bronchii and not the alveoli. At least some discussion of this seems warranted.

Page 11-17, lines 2-3 - How is the fact that the upper bound estimate using this
approach being greater than the unit risks derived from the complete animal
bioassays a disadvantage? I wouldn't think that the unit risks derived from the
complete animal bioassays should be regarded as the gold standard.

Page 11-17, lines 14-15 - How can you estimate an upper bound when you can't
derive one?  This needs some further explanation.

Page 11-20, line 2 - There is also a recent study of German coal miners  in which
an excess of lung cancer has been observed (P. Morfeld).

Page 11-21, line  19 - How do we know that this approach yields reasonably
good estimates of risk? We don't know what the true risk is so I can't see how
one can establish whether or not the risk estimates are "good".

Page 12-6, lines 20-35 - The issue of PM and its implications for diesel should
be more thoroughly discussed in this document.

Page 12-9, last line - The  summary should note the association between bladder
cancer and diesel exposure that has been noted in several epidemiologic

Page 12-11, lines 1-4 - Suggest replacing "showing the trend" with summarizing.
This sentence raises an important alternative method for exposure-response
analysis, which is to use the meta-analysis results rather than individual  studies.
I think this is a useful approach, and that it should be used in future drafts of this

                                     Comments on Chapter 12

                                           Ron Wyzga

page 12-1, lines 19-23:  This is misleading. The mouse studies are not convincing, and the rat studies are
controversial with respect to interpretation. On the other hand, diesel emissions contain known carcinogens.

page 12-2, lines 19-26:  This presents a very important point which is lost in much of the remainder of the

page 12-3, line 31:  at high doses; perhaps the exposure levels should be given to provide context.

page 12-10, lines 16-18:  Should  it mentioned that an independent analyses of these data reached alternative

page 12-12: It is important to note that diesel exhaust does contain known carcinogens.

pages 12-16-7: This was discussed in other sections at the meeting. In general I do not support an interspecies
factor of 3; EPA Staff said that this was not used, but rather a "data base" adequacy factor of 3 was used in this
calculation. First of all, the language and description must be clarified. I'm ambiguous about the latter factor
and suggest EPA provide a range for the RfC based upon the inclusion and non-inclusion of this factor yielding
a range of 5-15 ug/m3.  The monkey study cited was never carefully reviewed in the chapter; hence  I am
uncomfortable with its use here as supportive to the argument made.  If I understood EPA staff correctly, use
of these data would lead to a lower value for the RfC. I am disturbed, however, that an essentially negative
2-year study in monkeys at 20QOug/m3 leads to an RfC of 1 ug/m3.

page 12-17, line 17:1-year average

page 12-21:  I was among those who expressed concern about using the human data to derive dose-response.
In particular, analyses of the retrospective cohort study data, based on the same population, showed conflicting
results depending upon the analysts undertaken. This suggest to me that the range of estimates is greater than
that indicated by the document.

page 12-22:  I don't like the use of the term biomarker for B(a)P, but calculations based upon a component of
diesel exhaust are meritorious as lower limits of risk; in such cases, however, the calculation of risk from B(a)P
data should be reviewed carefully.  It is my understanding that no committee of EPA's SAB has ever reviewed
the basis for the B(a)P risk estimates. If that is the case, it should be noted that the estimates here are not
based upon EPA's  estimates.

page 12-23, lines 26-32:  I am deeply troubled by this analysis. If I understand correctly, the analysis is based
upon a net increase of one tumor in a population of several hundred rats. If data from other rat studies had
been included, the result would have  been negative.  So would have the results from an analysis of mouse
data. I believe this  calculation should  be removed from the report.

page 12-25, lines 1-29:  I have severe problems with this section and approach.  First of all, the risk calculation
here is not in accordance with EPA guidelines. Secondly, choice of an alternative endpoint or of response at
an alternative dose level would have lead to very different results.  If this section is to be retained, sensitivity
analyses 'must be undertaken to demonstrate whether the results are robust.

page 12-27: lines 22-34: Much of this speculative and should be denoted as such or be deleted.

page 12-31, lines 6-17: very well stated

section 12.5 pages 12-31  -12-38: I applaud these efforts.

                                             A- 108

                          Comments on Chapter 8

  This chapter provides a fair summary of the recent and most relevant
epidemiology literature.  It is very similar to the earlier version of the document.
Relatively few new studies are presented; these appear to be fairly summarized
and critiqued.

   There are places in the chapter where better articulation could aid the reader;
for example, on page 8-14, middle paragraph, from the description of the
Garshick et al study results, it is not clear whether the relative risks presented
are for total neoplasms or for neoplasms of the lung.  In other cases, particularly
in the discussion of bladder cancer studies,  more discussion could have been
given to the presence of potential confounders  Dietary and chemical exposures
could be very important for these cancers.

   One issue that is dismissed very readily in this document, but  covered in more
detail in the earlier version,  are the analyses of the Garshick et al. study by
Crump et al. for EPA and by Dawson for CaliforniaEPA.  The current document
limits the discussion of these to less than 10 lines on page 11-10, concluding
that "Until they [the differences in the Crump et al. and Dawson analyses] are
resolved, utilization of the Garshick et al. (1988) study to quantitate cancer risk
from DE  is not anticipated."   It is disappointing that the Agency has not been
able to resolve these differences in the past three years. As a minimum these
differences should be discussed in some detail in the document so that this
panel can determine whether it agrees with the document's conclusion,

   The key issue is how to summarize and weight the epidemiology evidence.
The document walks a fine line here, and probably ends up making the correct
call of "limited"  epidemioloic evidence for carcinogenicity, but in  its discussion
there are several factors that need to be highlighted.  As the document points
out, because of necessity all of the epidemiology studies are based upon
occupationally-exposed individuals; hence there are difficulties in extrapolating
results.  The document discusses the issue of the "healthy worker" effect,  the
document also notes that occupational exposures are significantly higher than
would be expected in the general population. The magnitude of  this difference
need be  presented and discussed.  There is little discussion about how the
nature of diesel emissions has changed over time (and geographically as well).
Are studies based upon exposures twenty or more years ago relevant for
contemporary exposures? I would like to see some discussion of this in the text.
The document also discusses the possibility  of confounders in an occupational
study with emphasis on concurrent asbestos exposure or on smoking. Other
lifestyle factors (e.g., alcohol,  itinerancy) could also influence the results for the
workers generally studied here.

                         Comments on Chapter 11  -

   I think this chapter and the classification of diesel participate matter may end
up in  the right place, but I have problems with the logic and would change some
of the discussion.  Personally I am more troubled by risk estimates derived from
the animal research than by the epidemiology.  That fact that positive tumor
results essentially exist only for the rat in a situation associated with particle
overload troubles me.   I would like to listen to what my colleagues have to say
about how to interpret the results of the animal data before giving a final
decision on interpretation and on the content of this chapter as well as on the
merits of deriving quantitative risk assessment estimates from these data.

   I have some specific comments about some of the content of the chapter:
pages 11-4-11-7:   I am not a fan of comoarative potency estimates.  I find
them to be interesting academic exercises, out hold little stock in their results
because rankings are rarely preserved for different toxic endpoints.  if the
endpoint were a definitive one, e.g., tumors, I would be more comfortable, but
even then we would lose a lot of information associated with different responses
for alternative exposures. For that reason, I cannot assign equal degrees of
reasonableness and credibility to this method as to the others. The credibility of
this method is further undermined by the fact that trie process somehow converts
a negative result (for the Caterpillar engine) to a positive risk; how this is
accomplished is unclear, but alternative reasonable methods could have
resulted in different results.

page  11-7:  I would like to learn the reactions  of the animal toxicology experts on
the panel before commenting upon the assessments based on lung tumor
induction data for rats.  The Hattis and Silver model could be particularly
interesting because it purports to incorporate lung burden.

pages 11-8-11-10: I am more comfortable using the epidemiology data with all
of their uncertainties than the animal data, but again there are several problems
which cause me to hesitate advocating that the risk results be published in this
document.  First of all, the exposure data are poor and considerable
assumptions need  be made in order to derive estimates.  Secondly I am deeply
troubled by the non-resolution of the Crump et al. and Dawson analyses of the
Garshick et al. data. It is imperative that this contrast be resolved because, in
the absence of any resolution if one gives equal weights to both studies, the
range in risk estimates goes from essentially zero to a significant risk level. If
the uncertainty is genuine, it need be reflected in the document; if it is not, we
need  to learn that.  The current document provides a disservice by not
elaborating upon these analyses and, if possible,  deriving conclusions about
their relative strengths.

pages 11-11 -11-16: Again I would like to learn more from my colleagues
before making any final judgment about using the rat bioassay data to derive
risk estimates. I applaud the efforts by Chen and Oberdorster; this model
appears to respond to concerns about the effects of paniculate overloading, but
it requires subjective judgment about the threshold levels at which particle
effects occur.  It would be useful to poll several experts about the parameter
inputs and to see how the model performs under these assumptions.  Using only
data associated with low exposures in animal studies is one way to address the
particle overload problem, but none of the responses at these levels reflected
statistically significant increases in tumors over background levels. The total
difference in tumors was minimal; indeed  if one had undertaken a similar
analysis for the mouse data, it would have suggested a beneficial effect of diesel
particulate matter.

page 11-17:  I am  intrigued by the results of the Pike and Henderson using B[a]P
as a surrogate for diesel emissions. These results could be interpreted to
explain a fair amount of the carcinogenicity associated with diesel emissions.
The document should discuss this issue further and speculate upon the
incremental benefits of restricting exposure to diesel emissions beyond
restricting exposure to B[a]P. In other words, is a risk management strategy
which focuses upon B[a]P adequate to protect the public from diesel emissions?

pages 11-17 -11-22: A good discussion, which could be augmented given some
of my above comments. The discussion need emphasize the facts that all
diesel emissions are not the same in characterization as well  as in
concentration.  Anything that can be said  about these changes over time would
be particularly useful here. This could aid in the interpretation of various studies
and analyses as well

pages 11-22 -11-23: Changes made in response to the above should be
incorporated here.


   I have reviewed the EPA responses to earlier CASAC comments and some
concerns about the response to Comment 6:  The noncancer concerns for
which the RfC was based were those associated with chronic exposures.  In my
opinion it is not appropriate to adjust  this RfC because of allergic
hypersensitivity.  That argument would be relevant for an RfC based upon acute
exposures and responses.

   Comment 9: Habeas law is exercised over a considerable range  of
exposures and times in the document. I believe that it is incumbent upon  the
authors of the report to demonstrate that Haber's law is relevant for the full
range of dose -exposure time periods for which it is applied. This is of particular

concern when results are presented across experiments with very different
exposure protocols. For example, Table 5-2 presents C x T data without
qualification for exposure periods varying from 19 weeks to 3 hours (a factor of
600), and concentrations ranging from 0.21 to 6.8mg/m3  (a factor of 30); other
tables are similar. As a minimum I would like to see a footnote addressing this
concern in the various tables.
                                                  Ronald E. Wyzga, Sc. D.
                                                              May 1, 1998
                                  A- 112

                       DISTRIBUTION LIST
Deputy Administrator
Assistant Administrators
EPA Regional Administrators
EPA Laboratory Directors
EPA Headquarters Library
EPA Regional Libraries
EPA Laboratory Libraries
Library of Congress
National Technical Information Service
Office of Technology Assessment
Congressional Research Service
                                        U.S. Environmental Protection Agency
                                        Region 5, Library (PJ.-12J)
                                        77 West Jackson Boulevard, 12th Fkw
                                        Chicago, II 60604-3590