UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
          °                       WASHINGTON D.C. 20460
                                                            OFFICE OF THE ADMINISTRATOR
                                                             SCIENCE ADVISORY BOARD
                                   January 20, 2010

EPA-C AS AC-10-005

The Honorable Lisa P. Jackson
Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, N.W.
Washington, D.C.  20460

       Subj ect:  Review of Integrated Science Assessment for Carbon Monoxide (Second
                External Review Draft)

Dear Administrator Jackson:

       The Clean Air Scientific Advisory Committee (CASAC) Carbon Monoxide (CO) Review
Panel met on November 16-17, 2009 to review EPA's Integrated Science Assessment (ISA) for
Carbon Monoxide (Second External Review Draft, September 2009).  The Panel's report was
reviewed and approved by the chartered CASAC on a December 22, 2009 public teleconference.
This letter begins with CAS AC's overall comments and evaluation. We highlight the most
important issues which need to be addressed as the ISA is finalized. The Panel and CASAC
membership is listed in Enclosure A. The Panel's consensus responses to the Agency's charge
questions are presented in Enclosure B. Finally, Enclosure C is a compilation of individual panel
member comments.

       We appreciate the responsiveness of EPA staff to our previous comments. The issues
that we targeted as important in our previous review were addressed. Both the revised
document, as well as statements by staff in response to our comments at the meeting, was
responsive to our concerns.  CASAC commends the EPA staff for the development of a
comprehensive and readable second draft of the Integrated Science Assessment for Carbon
Monoxide.  The document integrates relevant evidence from the past decades while emphasizing
newer evidence and a deeper understanding of mechanisms by which CO affects health.  The
extensive literature on CO is thoughtfully summarized and presented effectively in tables and
appendices.

-------
       We are comfortable with the process used by the EPA to produce this document.  The
EPA has implemented a process that is consistent with current approaches to evidence review
and synthesis.  It has progressively refined this process in recent NAAQS reviews.  The process,
which is being applied across the criteria pollutants, will enhance the quality and transparency of
CASAC's reviews.

Some additional major comments follow:

     •  CASAC previously noted that the terms "sensitive, susceptible, and vulnerable" are
        sometimes used interchangeably in EPA's various review documents.  We reiterate the
        need for these terms to be used in a consistent manner. We recommend that the EPA
        develop a glossary of terms to be used across documents related to all criteria
        pollutants. Such a glossary would promote consistency in the ISAs and REAs.  The
        definitions presented in the recently released final ISA for PM (December 2009,
        EPA/600/R-08/139F) may address some of CASAC's concerns.

     •  The Panel expresses concern about the existing CO monitoring network, both for its
        spatial coverage and for its utility in estimating human exposure.  CO exposures may
        not be adequately characterized for populations that may be exposed to higher CO
        levels because of where they live and work.  Moreover, because CO levels are often
        below the limit of detection of current monitors, both exposure assessment and model
        calculations may be limited.  The Panel recommends that monitoring for CO should be
        improved.

     •  The Panel approves the broadening of the evidence base considered in the ISA. For
        example, the discussion of CO in relation to atmospheric chemistry and climate change
        is useful.  Although such considerations do not drive the current standard, it is
        important to acknowledge climate change. While we agree that this topic merits
        discussion, the influence of climate change on a secondary standard is minimal because
        of the high level of uncertainty at present.

     •  Although EPA regulations for most pollutants are weighted heavily by information
        provided from epidemiologic studies, in the case of CO, information from well-
        designed clinical exposure studies has received emphasis. We agree with the weight
        that they are given in the current document.

     •  The problem of co-pollutants serving as potential confounders is particularly
        problematic for CO.  Since exposure levels for CO are now low, consideration needs to
        be given to the possibility that in some situations CO may be a surrogate for exposure
        to a mix of pollutants generated by fossil fuel combustion. A better understanding of
        the possible role of co-pollutants is relevant to regulation and to the design, analysis,
        and interpretation of epidemiologic studies on the health effects of CO.

-------
     •  Some of the challenges in interpreting and reviewing the evidence on CO in the ISA
        (and Risk and Exposure Assessment) reflect the great progress in reducing ambient
        concentrations of CO.  Control measures taken over the past decades have led to greatly
        reduced emissions and concentrations. Notable progress has been made in reducing
        exposure of the public to CO. CASAC notes that the ISA documents a substantial
        decline in CO levels in urban areas over the past two decades, which has undoubtedly
        benefited public health.

       CASAC agrees that the Draft CO ISA will be adequate for rulemaking with the
incorporation of changes in response to the Panel's major comments and recommendations.
We thank the Agency for the opportunity to provide  advice on the Draft CO ISA.
                                 Sincerely,
      /Signed/
      /Signed/
Dr. Joseph D. Brain, Chair
Clean Air Scientific Advisory Committee
   Carbon Monoxide Review Panel
Dr. Jonathan M. Samet, Chair
Clean Air Scientific Advisory Committee
Enclosures

-------
                                    Enclosure A

                     Rosters of the CASAC CO Panel and CASAC

                     U.S. Environmental Protection Agency
  Clean Air Scientific Advisory Committee Carbon Monoxide Review Panel

CHAIR
Dr. Joseph Brain, Cecil K. and Philip Drinker Professor of Environmental Physiology,
Department of Environmental Health, Harvard School of Public Health, Harvard University,
Boston, MA
MEMBERS
Dr. Paul Blanc, Professor and Chief, Department of Medicine, Endowed Chair, Occupational
and Environmental Medicine, Division of Occupational and Environmental Medicine, University
of California San Francisco, San Francisco, CA

Dr. Thomas Dahms, Professor and Director, Anesthesiology Research, School of Medicine, St.
Louis University, St. Louis, MO

Dr. Russell R. Dickerson, Professor and Chair, Department of Meteorology, University of
Maryland, College Park, MD

Dr. Laurence Fechter, Senior Career Research Scientist, Department of Veterans Affairs, Loma
Linda VA Medical Center, Loma Linda, CA

Dr. H. Christopher Frey, Professor, Department of Civil, Construction and Environmental
Engineering, College of Engineering, North Carolina State University, Raleigh, NC

Dr. Milan Hazucha, Professor, Department of Medicine, Center for Environmental Medicine,
Asthma and Lung Biology, University of North Carolina - Chapel Hill, Chapel Hill, NC

Dr. Joel Kaufman, Director, Occupational and Environmental Medicine Program, University of
Washington, Seattle, WA

Dr. Michael T. Kleinman, Professor, Department of Medicine, Division of Occupational and
Environmental Medicine, University of California, Irvine, Irvine, CA

Dr. Francine Laden, Professor, Channing Laboratory, Harvard University, Boston, MA

Dr. Arthur Penn, Professor LSU School of Veterinary Medicine, Department of Comparative
Biomedical Sciences, Louisiana State University, Baton Rouge, LA

-------
Dr. Beate Ritz, Professor, Epidemiology, School of Public Health, University of California at
Los Angeles, Los Angeles, CA

Dr. Paul Roberts, Executive Vice President, Sonoma Technology, Inc., Petaluma, CA

Dr. Armistead (Ted) Russell, Professor, Department of Civil and Environmental Engineering,
Georgia Institute of Technology, Atlanta, GA

Dr. Anne Sweeney, Professor of Epidemiology, Department of Epidemiology and Biostatistics,
School of Rural Public Health, Texas A&M Health Science Center, College Station, TX

Dr. Stephen R. Thorn, Professor, Institute for Environmental Medicine, University of
Pennsylvania, Philadelphia, PA
SCIENCE ADVISORY BOARD STAFF
Ms. Kyndall Barry, Designated Federal Officer, Washington, DC

-------
                     U.S. Environmental Protection Agency
                    Clean Air Scientific Advisory Committee
CHAIR
Dr. Jonathan M. Samet, Professor and Flora L. Thornton Chair, Department of Preventive
Medicine, University of Southern California, Los Angeles, CA
MEMBERS
Dr. Joseph Brain, Cecil K. and Philip Drinker Professor of Environmental Physiology,
Department of Environmental Health, Harvard School of Public Health, Harvard University,
Boston, MA

Dr. H. Christopher Frey, Professor, Department of Civil, Construction and Environmental
Engineering,  College of Engineering, North Carolina State University, Raleigh, NC

Dr. Donna Kenski, Data Analysis Director, Lake Michigan Air Directors Consortium,
Rosemont, IL

Dr. Armistead (Ted) Russell, Professor, Department of Civil and Environmental Engineering,
Georgia Institute of Technology, Atlanta, GA

Dr. Helen Suh, Associate Professor, Department of Environmental Health, School of Public
Health, Harvard University, Boston, MA

Dr. Kathleen Weathers, Senior Scientist, Gary Institute of Ecosystem Studies, Millbrook, NY
SCIENCE ADVISORY BOARD STAFF
Dr. Holly Stallworth, Designated Federal Officer, Washington, DC

-------
                                        NOTICE
This report has been written as part of the activities of the EPA's Clean Air Scientific Advisory
Committee (CASAC), a federal advisory committee independently chartered to provide extramural
scientific information and advice to the Administrator and other officials of the EPA. CASAC
provides balanced, expert assessment of scientific matters related to issues and problems facing the
Agency. This report has not been reviewed for approval by the Agency and, hence, the contents of
this report do not necessarily represent the views and policies of the EPA, nor of other agencies
within the Executive Branch of the federal government. In addition, any mention of trade names of
commercial products does not constitute a recommendation for use. CASAC reports are posted on
the EPA website at http://www.epa.gov/CASAC.

-------
                                      Enclosure B

               CASAC's Consensus Responses to EPA's Charge Questions

    1.  Chapter 1 has been revised in response to comments from the CO Panel, as well as
       related comments from the CASAC PM Panel, to add information regarding criteria for
       study selection and evaluation, to add more CO-specific information to the framework for
       causal determination, and to more clearly describe the process of integrating evidence
       from various disciplines to classify the overall weight of evidence relating to causality.
       What are the views of the Panel on the extent to which this revised Chapter 1 provides
       necessary and sufficient background information for review of the subsequent chapters of
       the CO ISA?

Chapter 1 is an important but challenging chapter.  It is responsive to the comments and
suggestions provided previously by CASAC.  The chapter establishes a solid background to
subsequent chapters of the CO ISA.  The integration of Tables  1-1 and 1-2 provides a concise
summary of the aspects used in epidemiology to assess causality. Section 1.6, EPA Framework
for Causal Determination, now incorporates a detailed description of the criteria for causal
determination. The introductory sentence to Section 1.6.3 clearly describes the process of
moving from association to causation, requiring the elimination of alternative explanations for
the association. In order to illustrate the criteria used to assess the quality of a study, it would be
helpful to include: the definitions of confounding and effect measure modification and the
criteria for determining if a factor (covariate) is a confounder; the process utilized to identify
confounders and effect measure modifiers; available methods to control for confounding in the
design and analysis phase of a study; and the most appropriate ways to interpret effect measure
modification. In general, the ISA would greatly benefit from an improved presentation of
epidemiologic concepts.

More detail on the scope of the critical review of ecological effects (in the sense of effects on the
ecosystem, not ecological associations) is requested. Specifically, what literature databases were
searched, using what keywords, for what time period, and for what geographic scope? Given the
scant literature on the ecological effects of CO, EPA should nonetheless comment on hypotheses
for ecological effects and identify key data gaps.  Such information would be useful for setting a
research  agenda to inform the next revision of the CO NAAQS.

The terms "sensitive," "susceptible," and "vulnerable" are often used interchangeably or at least
with potentially overlapping meanings.  In regard to the final CO ISA, CASAC reiterates its
expectation that the terminology will be consistent with the definitions in the final PM ISA and
used uniformly throughout the entire document.  EPA should develop a glossary of terms that are
used across criteria pollutants, to ensure consistency of terminology in the IS As and REAs for all
NAAQS  reviews. The term "sensitive" is in the statutory  language (see footnote 1 on p. 1-3)  and
thus may have special regulatory significance, which should be explained. The role of
identifying "susceptible" and "vulnerable" groups with respect to characterization of "sensitive"
groups should be  explained as well. Although a table giving definitions of these terms appears
later in the document, clarification earlier would be better.

-------
Finally, given that the epidemiologic literature on CO has grown over the past decade, the Panel
recommends that EPA assess the applicability of performing quantitative meta-analyses when
appropriate. Such analyses would better inform quantitative effect estimates, and allow the
Agency to refine further its inferences from the scientific literature.

2.  Chapter 3 has been revised and expanded in response to Panel comments regarding climate,
monitoring, spatial variability, and exposure.

          a.  Evidence reviewed in Chapter 3 of the ISA indicates that the direct contribution of
          CO to greenhouse warming is very small, while the role of CO in atmospheric
          chemistry cycles involving other species makes a larger contribution to radiative
          forcing.  This combined evidence leads to the conclusion in Chapter 2 that a causal
          relationship exists between current atmospheric concentrations of CO and effects on
          climate.  What are  the Panel's opinions related to this causal statement and the
          evidence provided to support it?

          b. Additional detail has been provided regarding the detection limits of CO monitors
          in the regulatory network, the number of monitors reporting at each horizontal
          spatial measurement scale and comparison of monitoring data at each scale, and
          spatial variability of CO concentrations near major sources, particularly roadways.
          Please comment on the usefulness of these revisions in characterizing the information
          provided by the CO monitoring network.

          c. The section on exposure assessment has been reorganized to provide information
          on exposure assessment at different spatial scales and to create a subsection
          containing information regarding exposure error and its implications for
          interpretation of epidemiologic studies.  Does the Panel consider that the sources of
          exposure error have been appropriately characterized, and agree with the revised
          conclusions regarding the impact of exposure error due to spatial variability and the
          presence of CO as part of a combustion-related mixture on health effect estimates
          from time-series epidemiologic studies?

Substantial information has been added to both Chapters 2 and 3 as well as to Annex A.  The ISA has
been strengthened by these additions.   The review of the literature appears to be thorough, and the
analysis of the scientific evidence systematic. In discussion of non-anthropogenic CO emissions (pages
3-3 to 3-5), emissions from biogenic sources, and CO generation from the oxidation of volatile organic
compounds (VOCs), isoprene, and other biogenic VOCs should be added to the  caption of Figure 3-1.
Additionally, the discussion would benefit from the inclusion of information on  the range of motor
vehicle operations which favor CO emissions, such as operation under high load, emissions during cold-
starts, and emissions from gross-polluting vehicles.

The addition of information on the potential impact of CO on climate is very helpful.  It appears
that the direct impact of CO is small, but the indirect impact of CO may be substantial.
However, the estimates of the  impacts of CO on climate are uncertain.  This current high level of
uncertainty does not favor the development of a secondary standard.

-------
The expanded discussions on CO detection limits, monitoring details, and spatial CO characteristics
better characterizes the information provided by the CO monitoring network and helps to qualify the
data used in exposure estimations.  Specifically, the expanded discussions on monitor detection limits
and monitoring locations in Section 3.4 and the respective Annex figures and tables are critical to
understanding CO concentrations. They are an important addition to the ISA. Although the limitations
of insensitive existing monitors are provided in the text (page 3-43, lines 13-18), these limitations need
to be added at numerous other places in the ISA that also address CO concentrations.  The relaxation of
CO monitoring requirements and the continued use of older, less sensitive monitors with high detection
limits the use of monitoring data for current exposure assessments.

In general, the expanded discussions of sources of exposure and resulting exposure assessment in
Section 3.6 are a great improvement and are useful in characterizing the potential impacts of exposure
error. The section on Land Use Regression Models (page 3-94, lines 6-23) is limited in  scope. It does
not represent the wide range of modeling methods nor exposure results from the literature.  Additional
discussions should help characterize the spatial concentrations of CO between monitoring locations and
particularly near roads where concentrations are usually much higher than in general area locations.
Admittedly, much of the modeling work in the literature is  on pollutants other than CO,  but the
conclusions regarding what methods work and how they relate to estimating pollutant concentrations
should be directly applicable to CO.

   3.  In response to comments from the CASAC CO Panel, material has been added to Chapter
       4 describing comparisons among predictive COHb  models, the relative influence of
       differing exposure scenarios on COHb concentration, and endogenous CO production
       rates in individuals with various diseases and conditions. Please comment on the
       usefulness of this information in illustrating the factors influencing COHb kinetics and
       potential COHb levels under various scenarios.

Generally, we found the revised and substantially expanded Chapter 4 of the second draft ISA to
be comprehensive and very useful in illustrating various physiologic factors and disease states
that influence blood levels of COHb and/or their potential adverse affects.  Section 4.2 describes
in adequate detail various COHb predictive models. However,  despite the addition of section
4.2.3 on Model Comparison and its discussions of the respective models' strengths and
limitations, questions remain. It is unclear to the Panel: (1) how these different models will
perform under the same simulated temporal exposure scenario of 30-60 minute duration with
occasional peak CO concentrations, and (2) which model is the most accurate in predicting
COHb levels? Several models seem to be most suitable for an inter-model comparison and
evaluation, e.g., Smith et al., 1994, Bruce and Bruce, 2008, Gosselin et al., 2009, as well as the
non-linear Coburn-Forster-Kane equation (CFKE) used by  EPA in the Agricultural Policy
Extender (APEX ) model.  This comparison may help to establish whether the CFKE is the best
model given that activity levels are evaluated on a minute-by-minute basis and ambient CO
shows transient peaks. The physiological parameters (e.g., VA and DLCO) used  in both Denver
and Los Angeles COHb calculations should be spelled out, since DLCO and ventilation rates
vary by altitude.
                                           10

-------
There are some inconsistencies among the tables, figures and text presented in the discussion of
the Quantitative Circulatory Physiology (QCP) model in section 4.2.4, which need to be
reconciled. The section could be shortened by dropping less relevant material.

The addition of section 4.3.4 COHb Analysis Methods in this draft is very helpful in pointing to
limitations and inaccuracies of some of the instruments used to measure COHb. Since the
differences in COHb determination among methods may be substantial, we suggest indicating
the method/instruments used to determine COHb in the tabulated studies as well as in other key
studies discussed in the text.

In addition to people with cardiovascular disease there are other large population groups
potentially at-risk from CO exposure, such as those having various forms of anemia or COPD.
Although there are no experimental studies on the effects of CO exposure on these groups at
ambient concentrations, these individuals may be more vulnerable to CO because their disease
state amplifies the action of CO.  The application of COHb predictive models with the inclusion
of appropriate pathophysiological parameters representing such disease states, if feasible, might
be helpful in determining the extent of risk in such populations. Especially, if available,
additional details should be provided and discussed regarding the fetus as an at-risk individual.

Identification of CO-specific associations with health endpoints in epidemiologic studies often
requires complex, multivariate analyses. Utilization of appropriate COHb predictive models can
help in interpreting the biological plausibility of associations of CO with health endpoints and
will likely provide further insight into the pathophysiological basis of responses.

       4. The cardiovascular effects section has been expanded to:
          •   evaluate key uncertainties in the health evidence, particularly regarding the
              biological plausibility of effects at low ambient CO concentrations and
              distinguishing independent effects of CO in multipollutant ambient mixtures;
          •   provide more detail on the design and findings of a multicenter controlled human
              exposure study to clarify the levels at which effects were observed;
          •   add description of new epidemiologic studies, including a large U.S. multicity
              study and studies on associations between blood markers and ambient CO
              concentrations; and
          •   more clearly describe the integration of controlled human exposure and
              epidemiologic evidence to reach a causal determination.

       Please comment on these  revisions to Chapter 5  and the conclusions for each of the
       health outcomes evaluated in this chapter. In particular, we are requesting CASAC
       comment on the interpretation of the evidence and the causal determination for short-
       term exposure to CO and cardiovascular morbidity.

The EPA staff is to be commended for the expanded, wide-ranging and comprehensive
presentation in Chapter 5 of the ISA. They have  added relevant material to the earlier version of
this chapter and included updates from articles that appeared as late as September, 2009. The
Panel offers the following suggestions for Chapter 5 of the final ISA: additional Forest plots to
                                            11

-------
summarize CO effects on blood markers and heart rate variability, and a meta-analysis of health
effects studies for selected outcomes.

Cardiovascular Morbidity.  The most compelling CO-related cardiovascular results remain those
from the controlled human exposure studies of Allred et al; Kleinman et al; and Sheps et al.  The
1991 Allred report contains dose-response information, including responses at COHb
concentrations <2%, based on the air-exposed COHb levels. More recent epidemiological
studies of morbidity at ambient CO levels, including data on hospital admissions, are consistent
with and reinforce the observations from the earlier controlled human studies. A large at-risk
population includes people with CVD who have not yet been formally diagnosed with this
condition. This undiagnosed group will likely grow in size and importance as our population
ages.  Many individuals that have had acute myocardial infarctions do not have already
diagnosed coronary artery disease (CAD).

Although those with diagnosed CAD are the largest CVD group and may represent the most
easily quantifiable highly-susceptible group for CO-related outcomes, CAS AC notes that EPA's
singular focus on CAD will underestimate the at-risk population. Other CVD patients,
regardless of whether they carry a diagnosis of CAD or not, are at increased risk for  CO-related
hospital admissions. Further, limited data from people suffering myocardial infarctions (Mis),
who had recently experienced high acute CO exposures, indicate that those Mis were associated
with vasospasm,  rather than with complications of CAD.  This further indicates that  CAD need
not be a final common pathway to designate adverse CV outcomes.  Finally, the association of
stroke with small increases in ambient CO levels also supports a more broadly defined risk group
going beyond those with established CAD.

In conclusion, it is CASAC's recommendation that EPA broaden the definition of the at-risk
population beyond people with CAD.  The Panel members concur with the ISA's conclusion that
a causal relationship is likely to exist between relevant short-term CO exposure and CV
morbidity.  We note that data are inadequate to establish a relationship between either chronic
CO exposure or transient elevations in ambient CO and morbidity.

Stroke.  The Panel finds there is a strong association between elevated ambient CO levels and
hospital admissions for stroke, distinct from other neurological outcomes (see above).
Consideration might be given to changing the title of the neurological outcomes section to "CO
and non-stroke CNS morbidity".

Respiratory Morbidity.  Positive associations have been demonstrated between short-term
exposure to CO and respiratory-related outcomes including effects on pulmonary function,
respiratory symptoms, medication use, hospital admissions, and emergency department visits.
However, there were no convincing data in which these relationships were consistently observed
after adjusting for multiple co-pollutants, which are also risk factors for adverse respiratory
outcomes.  The Panel was divided, and while the majority view was in favor of "evidence
suggestive of a causal relationship" others favored a designation of "inadequate to infer a causal
relationship."  The evidence on associations between long-term exposure to CO and  respiratory-
related outcomes is even more uncertain than for short-term exposures and were appropriately
categorized as "inadequate." Finally, the point was made that allergy and allergic responses
                                           12

-------
should be considered separately from respiratory outcomes and that this distinction should be
noted in the final ISA.

"Therapeutic " Applications of CO.  There is a growing literature regarding possible therapeutic
applications of CO at levels of-250 ppm. These studies have been carried out in some animal
models and in cell culture.  CO is a pro-oxidant and has profound extended pro-inflammatory
effects. However, in specific scenarios with distinct organ systems or specific cell types, CO
may have short-term anti-inflammatory effects.  Clinical trials thus far have not supported health
benefits of CO administration. Further, there is no evidence that the hypothetical therapeutic
results provide any insight into health effects of acute or chronic ambient exposures in the
general population, and especially in subpopulations susceptible to CO effects.

    5.  The section on susceptible populations has been revised substantially in response to
       comments from the CASAC CO Panel and in consideration of similar comments from the
       CASAC PM Panel.  The definition of a susceptible population has been clarified, and
       each subsection describing a susceptibility characteristic has been  revised to emphasize
       specific evidence from controlled human exposure studies of individuals with underlying
       disease, epidemiologic studies that conducted stratified analyses to examine effect
       modification, and toxicological studies using animal disease models.  Does this revised
       section provide appropriate characterization of populations potentially susceptible  to
       CO-induced health effects?

The discussion of populations susceptible to carbon monoxide has  been greatly improved.  The
data are now presented in a logical framework providing a clear and concise summary. Section
5.7 begins with Table 5-25, which provides  a useful context for understanding the historical use
of the terms vulnerability and susceptibility.  A question was raised during the Panel's
deliberations regarding whether level of exposure should be considered a "vulnerability" factor.
It was suggested that the term "at risk" may be useful  in some instances to  distinguish
individuals who are truly more susceptible owing to some specific  subject  characteristic rather
than to a difference in level of exposure.

Cardiovascular Disease.  As indicated in the previous charge question, the Panel finds that the
discussion of vulnerable subpopulation is focused too narrowly on  coronary artery disease
(CAD). It has been noted that arrhythmias and congestive heart failure should also be discussed
in this section. Further, CAD represents a continuous progression with many more individuals  at
risk than those carrying a doctor diagnosis.

Anemia.  A susceptible subpopulation is individuals with anemia from a diverse range of disease
states.  A key mechanism by which anemia may put individuals at  increased risk is the reduction
in the capacity to carry oxygen. Hemoglobinopathies, including sickle cell disease, should be
distinguished from anemia in general, as these disorders are likely to have  a different
susceptibility relationship with carbon monoxide.

Diabetes. Diabetes was identified as another factor condition might increase susceptibility to
carbon monoxide. During the Panel's deliberations, it was noted that the ISA does not mention
the high rate of obesity and metabolic syndrome as risk factors in addition  to diabetes itself or
                                            13

-------
the high correlation between diabetes and cardiovascular disease. In particular, the Panel cited
the results of a recent South Korean study (not included in the report) in which the effects of
carbon monoxide were investigated in individuals having both diabetes and cardiovascular
disease1.

Gestational Development.  The focus on altered gestational development included both the
mother and fetus. Limited data suggest the possibility of paternally mediated effects of carbon
monoxide owing to altered sperm production2. These effects cannot be ruled out as a potential
contributor to the effects of carbon monoxide in gestational development, but there is currently
no compelling evidence for this concern.

   6.  Chapter 2 has been revised and expanded to provide more information on atmospheric
       science and exposure assessment, policy relevant considerations, and integration of CO
       health effects.

          a.  The section on policy-relevant considerations was revised to present additional
              detail on the concentration-response relationship observed in a multi-center
              controlled human exposure study, present results from a new U.S. multicity
              epidemiologic study investigating the potential presence of a threshold and
              departure from linearity, and summarize the evidence for susceptible populations.
              Please comment on these revisions.

          b.  A section and summary figure have been added to the end of Chapter 2 to
              summarize the main conclusions of the ISA regarding the health effects of CO and
              the range of concentrations at which effects are observed, along with
              uncertainties that complicate the interpretation of the evidence. We would
              appreciate CASAC comment on the material in this section and its effectiveness in
              presenting the conclusions of the ISA.

The inclusion and analysis of data from a multi-site epidemiological study 3 is commendable,
given that it was published only recently. In addition, there are multiple points in which these
data could have been presented in Chapter 5, which is the basis of the presentation in Chapter 2
(see also comments to that charge question). Greater detail should be provided here in Chapter 2
because the data are particularly relevant. For example, limiting the analysis to those days with 1
ppm values or less resulted in the point  estimate for the increased hospitalization actually
increasing to approximately 1.75%.

Figure 2-1 is new to this revision, and is appropriate and helpful. The Panel offers the following
suggestions for improving Figure 2-1. The effect estimate metric as presented in the far right
1 Min JY, Pack D, Cho SI, and Min KB (2009). "Exposure to environmental carbon monoxide may have a greater
negative effect on cardiac autonomic function in people with metabolic syndrome," Science of the Total
Environment, 407(17), 4807-4811.
2 Rubes J, Rybar R, Prinosilova P, Vesnick Z, Chvatalova I,  Solansky I, and Sram RJ. 2009. Genetic polymorphisms
influence the susceptibility of men to sperm DNA damage associated with exposure to air pollution. Mutat. Res. Oct.
2.
 Bell ML, Peng RD,Dominici F, Samet JM (2009). "Emergency admissions for cardiovascular disease and ambient
levels of carbon monoxide: Results for 126 U.S. urban counties, 1999-2005," Circulation, 120 (11), 924-927.


                                             14

-------
column of Figure 2-1 should be more understandable. For example, because the lower bound of
the CI and the point estimate are of far more interest than the upper bound, adjusting the scale is
appropriate and would help the presentation visually.  Also, the effects could be grouped by
endpoint, not by study, with total CVD top, then IHD, CHF, and stroke.  Finally, the 99th
percentile of exposure may be of less interest than the 95th.

Although the CVD endpoint of Bell is included in Figure 2-1, the other specific endpoint of Bell
is left out. This is related to the tree plot consisting of unadjusted CO effects only.  A clarifying
note with text emphasizing that there  are co-pollutant adjusted values from Bell et al. should be
added. These critical endpoints include ischemic heart disease (IHD), congestive failure, and
stroke. The data in Figure 2-1 could be summarized with a formal meta-analysis or, if this
proves inappropriate due to heterogeneity, a comment as to the rationale  for not having such an
analysis should be included. The ISA should also explain why the data in Figure 2-1 are limited
to findings from North America.

We draw attention to the concluding paragraph in Chapter 2.  Before addressing uncertainties
that remain, it might be more straightforward to first catalogue the uncertainties that have been
substantially addressed since the 2000 CO AQCD.  (In the present text this is stated first in the
negative: "some of these uncertainties remain.") The argument with regard to the lack of
biological plausibility runs counter to the rich series of recent studies that indicate the potential
modulatory effects of CO at low levels on a number of systems (pages 5-5 to 5-17).  A separate
section should summarize this central point of improved data that has reduced much of the
uncertainty previously encountered. Moreover, the phrase in the last sentence "biological
plausibility provided by CO's role in limiting O2 availability" seems to diminish the rich data on
other mechanisms and their possible role (e.g., cell signaling independent of heme moiety
binding).

The concluding paragraph refers to the "many new epidemiological  studies adding to the body of
evidence showing associations."  The EPA could modify this  statement to include the adjective
convincingly, to be consistent with the next sentence regarding definitive cardiovascular effects
in controlled exposures. At various points, but most importantly in the very last sentence, the
phrase "relevant...  exposures" is used. If this means exposure at or  below the current EPA CO
NAAQS, this should be explicitly stated. Relevant could imply that other exposure levels are
irrelevant to the assessment of health effects, which of course is not intended as they may be
relevant through the mechanistic insights they provide. The Panel found the focus on
cardiovascular endpoints in the concluding paragraph appropriate. Nonetheless, an additional
sentence acknowledging that there is at least a suggestive relationship with several other
endpoints is warranted.  Also, a restatement of the association with global warming would be
appropriate in this concluding paragraph.

There is agreement among the Panel with the ISA's conclusion that there is a likely causal
association between acute ambient CO exposures in the range of the current air quality standard
and adverse cardiovascular endpoints. In contrast, there is scant evidence in the health effects
section related to chronic CO exposure effects on cardiovascular morbidity. Even though these
data may be categorized as "inadequate," this category appears to have been omitted from Table
                                            15

-------
2-1 and should be added. Also, the 2006 publication by Hedblad et al. 4 seems to be missing
from the discussion.  It is relevant to Chapter 5.

There is heterogeneity of views among the Panel regarding the summary statement that acute CO
exposure has a suspected association with adverse respiratory outcomes. The Panel recommends
tempering the narrative by explicitly indicating that for this association the evidence borders
between suggestive and inadequate.  In particular,  the epidemiological evidence was limited by
an absence of co-pollutant data. We lack studies showing substantive attenuation of CO risk
estimates when co-pollutant modeling is performed.

There is a consensus of the Panel that the evidence on the relationship between chronic CO
exposure and mortality should be categorized as inadequate. This view is based in large part on
the difficulty in epidemiologically differentiating between mortality due to multiple acute effects
compared to prolonged lower level effects without peaks. Also,  we believe that chronic
outcomes in myocardial infarction and stroke can be presumed to include excess mortality.
 Hedblad B, Engstrom G, Janzon E, et al. COHb% as a marker of cardiovascular risk in never smokers: results
from a population-based cohort study. Scand J Public Health 2006; 34:609-615
                                            16

-------
                                  Enclosure C

  Compendium of Review Comments from CASAC Carbon Monoxide Review Panel on
 EPA's Integrated Science Assessment for Carbon Monoxide: Second External Review Draft
                                (September 2009)

Comments received:

Dr. Paul Blanc	18
Dr. Thomas Dahms	23
Dr. Russell Dickerson	27
Dr. Laurence Fechter	31
Dr. H. Christopher Frey	32
Dr. Milan Hazucha	37
Dr. Michael Kleinman	40
Dr. Francine Laden	41
Dr. Arthur Penn	42
Dr. BeateRitz	45
Dr. Paul T. Roberts	48
Dr. Armistead Russell	53
Dr. Anne Sweeney	54
Dr. Stephen Thorn	56
                                      17

-------
                                    Dr. Paul Blanc

Comments on Charge Question #6

a.  Please comment on these revisions: The section on policy-relevant considerations was
revised to present additional detail on the concentration-response relationship observed in a
multi-center controlled human exposure study, [to] present results from a new U.S. multicity
epidemiologic study investigating the potential presence of a threshold and departure from
linearity, and [to] summarize the evidence for susceptible populations.

   It is appropriate to include the Allred concentration (exposure) response data and to point out
(as is done in the text) that this study was far larger and more powerful than any previous
controlled exposure study. In terms of exposure (concentration) response, the key Allred study
appeared in 1991 (the current document refers the reader to the 1989 Allred paper, although the
subsequent 1991 paper provides details of the exposure [concentration]; clarifying this point in
the text would be advisable). The Allred data could be presented to better effect if the point is not
only the linear relationship, but also the presence of a threshold. In this regard, the Allred
analysis of time until angina (from the actual paper,  Figure 12, page 112, 1991 and related text)
indicates that there was an intercept value (at 0% carboxyHb)  of a 1% decrease (± 2.1%; not
different from zero) in time until angina onset. This  is based on an analysis of room air not as
"zero," but as the actual post-exercise room air COHb value (which varied by site and among
subjects). The ST depression intercept, of note, was  significantly in the positive direction, which
could be argued in favor of a threshold for that endpoint. This can be discussed more explicitly
than as it appears in the current draft.  Also in regard to threshold, the analysis by Somoli, in
which the deviance from linearity was associated with a p value of > 0.9, should not be described
as "weak evidence in favor of a threshold" - this would be better described as a finding that does
not support the presence of a threshold. Moreover, there is an overextended discussion here as to
why this analysis was poorly powered to observe a threshold, including wording such as "an
inability to draw conclusions." This could give the appearance of trying too hard to leave open
the possibility of a threshold effect where none was  no threshold observed.

   No attempt was made to perform an integrated analysis of the experimental data for low-level
CO exposure and time until onset of angina from multiple studies (for example, using the data in
Table 27 of Allred of multiple studies on this subject, taking into account baseline room air
carboxyhemoglobin and post exposure levels). It may be that the data, ultimately, so not permit
this.  If so, a brief statement in this regard in the text  would nonetheless be useful.

   The concluding statement of this section reads: "Although the C-R relationship has not been
explicitly evaluated in human clinical studies with exposures resulting in COHb concentrations
< 2.0%, the findings of Allred et al. provide some evidence of a significant C-R relationship over
a range of COHb concentrations relevant to the NAAQS. " This sentence is overly weak,
somewhat confusing, and inexact. Allred, in fact, did explicitly analyze the concentration
response including those resulting from air (ambient CO +metabolism). Many of these
observations included in that regression were less than 2% COHb (see above, re: threshold). The
wording "some evidence" operates to undermine the findings - it is "evidence," which could be
argued to be substantial or strong (as opposed to the indeterminant "some"). If "significant" as
                                           18

-------
used in this text means statistically significant, then this should be explicit as well. Also on this
topic of concentration response, a comment voiced in the meeting was that, in addition the
potential differences between concentration response and delivered dose response based on
biological monitoring should be acknowledged, either here or elsewhere in the document.

   The summary of evidence regarding "susceptible" populations (2.6.1) actually precedes the
discussion of concentration response (2.6.2). This section does a fair job of summarizing lengthy
text elsewhere in the document, but also suffers from the organizational issues of that text. The
most substantive issue here is the usage of the term "susceptible" to refer to two entirely different
concepts, operationally. The first usage is consistent with the way in which susceptible is
typically applied: subgroups in which an exposure identical to the general population could be
expected to have a greater adverse effect. Examples of this include, and are documented: those
with pre-existing cardiovascular disease, diabetics, those with pre-existing anemia, those with
pre-existing hypoxemia, and the fetus. In contradistinction to this classic construct of
susceptibility, the document lumps together with this individuals who are susceptible because
either they are more likely to experience higher ambient exposures (living near roads, greater
commute times) or because the ambient exposure they receive will be superimposed on a higher
baseline value secondary to greater than average exposure to exogenous CO or due to greater
metabolic production of CO. Susceptibility by both routes are important, but the presentation
would be more lucid if the distinction were spelled out explicitly.

   The concluding sentence "Overall the controlled human exposure, epidemiologic, and
toxicological  studies evaluated in this assessment provide evidence for increased susceptibility
among various populations" is overly weak. By saying that the evidence for those with CAD is
"strongest" in the next sentence, the implication could be drawn that the other "evidence" is
somehow weak. It could be argued that the strongest evidence of susceptibility to CO, per se, is
for fetal exposure, on kinetic grounds of a longer half-life. The evidence of susceptibility is
certainly strong to convincing in a number of other scenarios. Also in the sentence in question, if
toxicological  means animal toxicology this should be stated and a fourth category of human
clinical toxicology cases added, or toxicology be clarified to mean both..

   Overall in the document there seems to have been little use made of human toxicology case
reports insofar as the implications that might be drawn from such data. One specific example:
human case reports clearly have shown that coronary artery spasm appears to mediate CO-
induced MI in some individuals (post CO-caused MI coronary vessels without underlying CAD
consistent with MI). [See for example: Marius Nunez AL, Myocardial infraction with normal
coronary arteries after acute exposure to carbon monoxide. Chest 1990; 97:491-4 and related
case reports]. There was, however, another minority view presented by a panel member that
argued against consideration of any data in which high levels  of exposure had occurred as being
irrelevant to lower level scenarios.

   The various scenarios of susceptibility seem to ignore indoor air sources of supplemental CO
exposure. Most glaringly, secondhand smoke exposure is missing [it can also be argued that this
can be an outdoor ambient issue in areas with heavy concentrations of smokers at the threshold
of edifices. This is also relevant to secondhand smoke exposure to vehicular passengers in
automobiles, already exposure to higher roadway levels of CO. Related to this issue, term
                                           19

-------
secondhand smoke (SHS) should be substituted for "ETS" where currently used in the document.
Paralleling the SHS issue, deficient home heating which may also be a risk and is likely to run
with lower socioeconomic status and or living in colder parts of the US in the winter months,
also relevant to susceptibility. In the same vein, occupational exposures superimposed on
ambient exposure should be taken into account as a potential susceptibility factor. Further, the
metabolism of "dihalomethanes" is mentioned as an enteric  source of CO, but this would be
better stated a predominantly methylene chloride (which has also been in some consumer
products). In regard to metabolism, there could be further clarification of data gaps in the overlap
or non-overlap of similar systemic levels of carboxyhemoglobin from internal metabolism
compared to extrinsic exposure. Finally, diabetics are mentioned, but a recent relatively large
study (n=986) from Korea (Min PY et al, Sci Total Environ  Aug 2009) with effect modification
for autonomic dysfunction [decreased heart rate variability] from CO by fasting blood glucose
was not cited or discussed. Many of these points are also relevant to the more detailed
presentation of susceptibility in Chapter 5.
b We would appreciate CASAC comment on the material in this section and its
effectiveness in presenting the conclusions of the ISA:  "A section and summary figure have
been added to the end of Chapter 2 to summarize the main conclusions of the ISA regarding
the health effects of CO and the range of concentrations at which effects are observed, along
with uncertainties that complicate the interpretation of the evidence."

   Figure 2-1 (page 2-21) is new to this revision. The Figure, in principal, is appropriate and
helpful, but it could be improved upon in ways delineated in the points below. The effect
estimate (far right of Figure 2-1) should have the metric presented more  clearly graphically. For
example, because the lower bound of the CI and the point estimate are far more of interest than
the upper bound - scaling so that the scale is bigger would help the presentation visually. Also,
the effects could be grouped by endpoint, not by  study, with total CVD top, then IHD, CHF and
stroke. Finally, the 99th percentile of exposure may be of less interest than the 95th.

   The inclusion and analysis of data from the multi-site epidemiological study (Bell et. al.) is
commendable in the text and to a limited extent in the Figure, given that it was published only
recently. There are multiple points in which these data could have been presented in Chapter 5
(which is the basis of the presentation in Chapter 2 - see also comments  to that charge question)
beyond the limited places where the paper is cited). Greater detail should be provided here in
Chapter 2 because the data are so relevant (for example: that limiting the analysis to those days
with 1 PPM values or less, the point estimate for the increased hospitalization actually increased
to approximately 1.75% [95% interval excludes 0)] or that a re-analysis excluding any days over
the 1 hour 35 ppm standard  had no impact on the estimated of 0.55%).

   In terms of the new Figure, although the CVD endpoint of Bell is included the other specific
endpoint of Bell are left out. This is related to the tree plot being of unadjusted  CO effects only,
but this could be addressed with clarifying notes  and with test emphasizing  that there  are co-
pollutant adjusted values from Bell et al. These critical endpoints include IHD, congestive
failure, and stroke.
                                           20

-------
    The data in Figure 2-1 could benefit from a formal meta-analysis or, if this proves
inappropriate (for example, due to heterogeneity) a comment as to the rationale to forgo the
presentation of such an analysis should be included. (Also in regard to Figure 2-1, a compelling
rationale as to why the data are limited to findings from North America is not provided and
should be inserted).

    The concluding paragraph (pages 2-24 and 2-25) is hampered by the logic of its presentation.
Before addressing uncertainties that remain, it might be more straightforward to first catalogue
the uncertainties that have now been substantially addressed since the 2000 CO AQCD. (In the
present text this is stated first in the negative: "some of these uncertainties remain.") The
argument re: lack of biological plausibility runs counter to the rich series of recent studies
indicating the potential modulatory effects of CO at low levels on a number of systems (see
pages 5-5 to 5-17). Indeed, a separate section as part of the policy section should summarize and
address this central point of an improved data that has reduced much of the uncertainty
previously encountered. Moreover the phrase "biological plausibility provided by CO's role in
limiting O2 availability" [last sentence] basically cuts out the rich data on other mechanisms
from any plausible mechanistic role [e.g., cell signaling independent of heme moiety binding).

    Earlier in the text, this same paragraph refers to the "many new epidemiological studies
adding to the body of evidence showing associations.." This could be modified to include the
adjective convincingly., consistent with the next sentence  re:  definitive cardiovascular effects in
controlled exposures.

    At various points, but most importantly in the very last sentence, the phraseology
"relevant... .exposures" is used. Where this means exposure at or below the current EPA CO
NAAQS, this should be so stated. "Relevant" could imply that other exposure levels are
irrelevant to the assessment of health effects, which of course is not intended (they are relevant
through mechanistic insights they provide, etc).

    In summary,  there was a consensus of the Panel agreeing with the conclusion that there was a
likely causal association between acute ambient CO exposures in the range of the current air
quality standard and adverse cardiovascular endpoints. In contrast to this, there is only scant
evidence in the health effects section re: chronic CO exposure effects on cardiovascular
morbidity. Even though these data may be categorized as "inadequate," this category appears to
have been dropped from Table 2-1 and this row should be added [note: Hedblad B. et al, ScandJ
Public health,  2006 seems to be missing from that discussion and seems on topic; this would be
relevant to Chapter 5].

    In terms of the summary that  acute CO exposure has a suspected association with adverse
respiratory outcomes, there was a heterogeneity of views on the Panel, with a suggestion that this
might be tempered in the narrative with explication indicating that this association was
borderline between suggestive and "insufficient data." In particular, the epidemiological
evidence was limited by an absence of co-pollutant data or was marked by CO risk estimates
were substantively attenuated when co-pollutant modeling was performed.
                                           21

-------
    In terms of chronic CO exposure and mortality, there was a consensus view that this would
be better categorized as "Insufficient data" rather than "unrelated." This view is based in large
part on the difficulty in epidemiologically differentiating between mortality due to multiple acute
effects compared to prolonged lower level effects without peaks and the fact that, logically,
chronic outcomes in myocardial infarction and stroke can be presumed to include excess
mortality.

    In summary, it is appropriate to focus on cardiovascular endpoints in the concluding
paragraph as written. Nonetheless, an additional sentence acknowledging that there is at least a
suggestive relationship with several other endpoints is warranted and also a restatement of the
association with global warming would be appropriate in this concluding paragraph.
                                            22

-------
                                Dr. Thomas Dahms

Statement: In response to comments from the CASAC CO Panel, material has been added
to Chapter 4 describing comparisons among predictive COHb models, the relative
influence of differing exposure scenarios on COHb concentration, and endogenous CO
production rates in individuals with various diseases and conditions.

Q: Please comment on the usefulness of this information in illustrating the factors influencing
COHb kinetics and potential COHb levels under various scenarios.

General comments: This section provides an excellent review of the modeling of CO uptake and
release. It provides the essential information needed to understand most of the variables involved
in relating CO exposure and CO dose. What follows are suggestions/questions that may lead to
further improvement and clarification of the material  presented.

I. With the increasing amount of epidemiology data being considered in this database changes in
atmospheric CO levels with various adverse health effects, the exposure models need to provide
guidance to the reader regarding likely levels  of exposure in some of these studies. The evidence
from the atmospheric data demonstrates a steady fall in monitored levels of atmospheric levels of
CO yet significant relationships with  seemingly small changes in environmental CO continue to
be identified. How can the exposure models provide insight into what might be occurring? I
presume that this would include a discussion of the limitations of the use of the current
atmospheric monitoring data to estimate exposure? I realize that there is data in the RFA and in
the 2000 CO ACQD pertaining to this situation but it is scattered and it would help the reader if
the salient issues were summarized as they pertain to the epidemiologic studies.

II. The modeling discussion in most of chapter 4 is based on factors influencing equilibrium
values for COHb given different exposure conditions. In Section 4.2.3 (Model Comparison), the
brief mention of the Bruce and Bruce model for predicting COHb levels with transient CO
uptake conditions , page 4.9 Iinesl4-20 or the QCP model deserves much greater consideration
based upon what we know from real life exposure scenarios. If the primary exposures to CO
occur during periods of commuting, which model more accurately predicts the CO uptake during
the 30 to 60 minutes of exposure? Section 4.2.3.  mentions the value of the Bruce and Bruce
model but then proceeds to use the  QCP model in the following section 4.2.4.without discussion
or examples as to how ithe QCP compares to the other models. If the models in section 4.2.3
were all compared to observed data, this distinction needs to be made. Otherwise the  use of
untested mathematical modeling in section 4.2.4. does not make sense.

III. Given that adverse health effects  have been demonstrated  at 2% COHb, the discussion on
page 4-5 lines 23-26 report that application of unspecified scenarios in some form of the CFK
model yield ranges of exposure levels required to reach 2% COHb. For the 1 hour (transient)
exposure, these atmospheric levels  of CO are  24-48 ppm which encompases the 35 ppm  hourly
criteria. However for the 8 hour exposure (equilibrium) the required exposure values  are 11 to 13
ppm which is above the 9 ppm standard. This data needs to be better referenced since it applies
so directly to the standards.
                                          23

-------
IV. Use of modeling information:
   1.   With the paucity of actual measurements of COHb distributions in the population
       (nothing since NHANES II), modeling is proposed to provide data relevant exposure
       data.
   2.   Since there are other pieces of missing data from the ideal data base from which to make
       assumptions regarding risks from CO exposure, I would propose that modeling be used
       to provide guidance for identified at risk groups for which there is little or no data. These
       groups would include those frequently mentioned:
   £L   anemia.
   For the past 30 years patients with anemia has been identified as being an at risk group for
   adverse health effects due to CO exposure. It is discussed again in this document in Section
   5.7.1.3. This would be a particularly sensitive subset of patients with CAD since both
   elevated COHb and reduced hemoglobin concentrations reduce oxygen delivery to the
   myocardium. It should be noted that anemia is a significant risk factor for development of
   angina. Yet there appear to be no studies available addressing this issue. The extent of
   exposure risk for this sizable group of people (approximately 4 million over 65 with anemia)
   needs to be addressed.

   The treatment of anemia in this document focuses on the increased risk due to elevated
   endogenous production of CO. It is unclear what influence the elevated endogenous rates
   have on adverse health effects. One would suppose that in the four-element (Section 4.2.
   page 4-2 lines 28-29) CFK model that when the largest element changed would be the
   storage compartment (total body hemoglobin) that exposure conditions would be reduced in
   order to result in the same measures of effective dose (%COHb). One would expect an
   increase in the transfer interface with the hyperdynamic state due to the anemia, but the
   impact of this component would be less clear. It is likely that the lack of a sizeable storage
   compartment in anemic individuals would result in reaching levels of COHb of concern at
   lower atmospheric levels during 1 hour or 8 hour exposures. The relative importance of
   endogenous production, reduced storage capacity and increased transfer rates could be
   determined through the use of modeling.

   (The number of individuals in the USA with anemia is significant. According to NHANES
   III, 10-12% of the population over 65 yrs of age (40 million) has anemia.
   The number of individuals with CAD and anemia is more difficult to estimate but the
   numbers range from 8-15% of those patients with CAD also have anemia.)

   b.  COPD and Emphysema
    According to NHANESIII there are 24 million individuals in the US with some amount of
   COPD. This is such  a sizable at risk group that application of various models of CO exposure
   using the impaired pulmonary function  parameters would be helpful in determining the
   extent of risk in this population.
                                          24

-------
V. Section 4.5.
Whenever COHb is mentioned the method of analysis should also be indicated otherwise the
reader would be misled assuming that all of the values were equivalent when they are not. This is
particularly relevant when discussion the impact of endogenous CO production because the
resultant COHb levels are very low.

The limitations of the easy to use and reproducible CO-oximeter data was outlined in section
2.6.1 of the 2000 CO AQC D. There are many assays with sufficient sensitivity available for use
as used by Coburn et al to produce the data shown in Figure 4-12. However much of the other
data in this Figure was collected with instruments not designed for accurate measurements of low
levels of COHb (De las Heras et al used a CO-Oximeter).

Additional major concern:
   Section 5.7.1.3. The primary concern for individuals with anemia when exposed to CO is that
the tissue hypoxia due to the anemia will be exacerbated by the additional reduction in oxygen
delivered to the tissues due to COHb. This should be the common theme for many of the pre-
existing diseases. Insufficient oxygen delivery making the heart tissue more susceptible to any
increase in oxygen demand as occurs during exercise from the underlying disease should be the
primary reason for concern. This is the case for the current state of our knowledge in the area of
tissue effects of CO as stated multiple times in this document. The only reference to the
pathophysiology of anemia is in line 31 on page 5-170 and the information is not correct. By
convention hypoxia implies a reduced oxygen supply. The blood does not have the reduced
oxygen supply in the lungs in anemia, only the tissues have a reduced supply of oxygen. The
information provided could be:  ... .and result in a reduced arterial oxygen content due... The
focus of this section should not be on the etiology of anemias but on the combined effects of two
different causes of tissue hypoxia.
                                          25

-------
Editorial and minor comments:
    1.  The use of deoxyhemoglobin is probably a carry over from the assumptions used in the
    CFK modeling of McCartney (013162) which should be ignored because it is not correct. .
    2.  Section 4.2. page 4-2, line 17. altitude should read exposure time and altitude.
    3.  Section 4.2.1. page 4-3, line 29. Vco is not shown in Figure 4-1.
    4.  Section 4.2.1 page 4-4, lines 11-14. The discrepancy between arterial and venous blood
    CO levels is mentioned without any interpretation as to why this is important. Also in this
    section the absolute errors in COHb are mentioned without providing any sense of what the
    mean increase in COHb was under these conditions.
    5.  Section 4.2.1. page 4-5, lines 23-26. A reference is needed.
    6.  Section 4.2.4, page 4-5.,lines 27-28. No explanation is given for reduced uptake by
    babies which appears to contradict information given in section 4.1. lines 20-22.
    7.  Section 4.2.3. page 4-9, Iine7. 'differ ±0.5%' needs clarification. 0.5% COHb or of the
    value obtained?
    8.  Section 4.2.4, page 4-9, line 21. Population data for COHb are available in (Radford and
    Drizd, 1982) so this statement needs to be clarified.
    9.  Section 5.7.1.1. In this section the distinction between the terms CAD and IHD needs to
    be spelled out probably according to ICD-9 codes since these disease codes  are the basis for
    most of the epidemiology studies. The term CHD should be dropped or noted as being of
    historic value only.
                                           26

-------
                               Dr. Russell Dickerson
Carbon monoxide, as the major sink for OH in the global troposphere has a substantial role in the
oxidizing capacity of the atmosphere.  For example Shindell et al. (2006) and [Isaksen etal.,
2009] show that the lifetime of methane can change by a factor of two depending on the range of
tropospheric CO mixing ratios. Uncertainties in the budget of OH are such that the current state
of the science is insufficient to establish the safe level of CO based for example on a 1°C
temperature rise. The ISA should reflect this uncertainty and point out the need for further
experiments and theory to inform the EPA.  Because CO (like SO2 and NOx) is both a local
pollutant and contributor to global climate change, a standard based only on the local maximum
concentration is inappropriate for protecting welfare.  Reduction of total emissions is appropriate
for pollutants such as CH4 and CO2 with adverse effects on a global scale. The ISA should
discuss the scientific basis for emissions-based standards or guidelines for CO.
Comments on ISA Charge Question 2a

In reference to ISA Chapters 2&3 that discuss a causal relationship between current atmospheric
concentrations of CO and effects on Climate.  "What are the Panel's opinions related to this
causal statement and the evidence to support it?"

Substantial additional information has been added to both Chapters 2 and 3 as well as in Annex
A, and the ISA is much stronger for it.  The review of the literature appears to be thorough, and
the analysis of the science systematic. One substantive comment I would make is that the
evidence all points to the need for new regulations for the climate effects of CO. The current
ambient concentration-based standards are not appropriate for large-scale global atmospheric
concentration concerns aimed at protecting welfare. This will have to be emissions-based
regulations similar to those being planned for CO2.  I suspect that the state of the  science not yet
adequate to establish a specific CO emissions cap, and if that is the judgment of the EPA authors
then the Integrated Science Assessment should clearly state that further research is needed to
establish a numerical value for American CO emissions. Do we know what the safe level of CO
in the atmosphere is?  If not then the ISA should so state.

The review of satellite measurements for establishing  PRB concentrations is fair -existing
instruments lack sensitivity in the PEL.  Remote sensing is already useful for model evaluation
and may some day be helpful for low-altitude measurements, and is

There is one more relevant paper that came out in Science after the draft was finished;  it shows
gas/aerosol interactions can  amplify the effects of non-CO2 trace gases on radiative forcing
[Shindell etal., 2009].

Comments on ISA Charge Question 2b
                                          27

-------
In reference to Chapters 2&3 that additional detail has been added on detection limits, number
and spatial variability of CO monitors etc.  "Please comment on the usefulness of theses
revisions...."
Table A-l is a great addition. This shows that highly sensitive instruments are commercially
available.  Page 3-20. The LOD is given as 0.04 ppm, but Table A-l shows 0.02 ppm. The ISA
should say the replacement monitors should have the lower LCD's.

Page 3-12 Figure 3-8 is hard to read, perhaps a scatter plot.

Page 3-22 The ISA should state the revoking the CO monitoring requirements impedes our
scientific understanding of air quality and climate.  The paragraph on NCORE is a great addition.

Page 3-33. The bar has a black stripe on top that looks like it should be a red stripe.

The additional detail in 3.5.1.2 is great. Page 3-45.  The tale on top with E C A B D does not
seem to correspond to the columns below.

Comments on ISA Charge Question 6b
A section and summary figure have been added to the end of Chapter 2. "We would appreciate
CAS AC comments...."

Figure 2.1 gives a good demonstration of the morbidity risks associated with CO, and is
understandable by non-specialists in epidemiology.

General Comments on ISA Chapters 2 & 3.
There is some redundancy between Chapters 2 & 3 as well as within the chapters that could be
eliminated without loss of coherence.

Section 3.2 There is a need for a bottom line here: substantial uncertainties in emissions
continue to exist.  On page 3-4 is states that the reviewed literature is consistent in determining a
decrease of 5% per year in on-road CO emissions.  Does that agree with Figure 3.2?  It might be
but it would be nice to see it explicitly compared.

Page 3-10. CH3OOH is not really soluble; the Henry's Law coefficient is about 300 M/atm,
much less that H2O2.

Page 3-13. OH  does not react with the major CFC's that are fully halogenated (such as CFC-11
and 12). There needs to be a hydrogen atom bound to the carbon somewhere.

The Summary and Conclusions should state that:

1. There are substantial uncertainties in the emissions inventories.
                                          28

-------
2.  The current state of the science is insufficient to determine what level of CO emissions is
adequate to protect welfare from adverse changes in global or local climate and in the oxidizing
capacity of the atmosphere.

Minor points on ISA
    1.  Page 2-20 line 14 space.

    2.  Page 3-14 line 22 semicolon where a comma should be.

    3.  The caption to Figure 3-10 and other similar figures should say that the circles indicate
       the position of the monitors.

    4.  AADT is not in the table of acronyms.

    5.  The word 'fraught' on page 3-85 seems odd to my ear.
                                          29

-------
Reference and some additional papers that may be of value to EPA.
[Clements et al., 2009; El-Fade I and Abi-Esber, 2009; Saide et al., 2009; Tomlin et al., 2009;
Wang and Zhang, 2009; Zhu etal, 2009]

Clements, A. L., Y. L. Jia, A. Denbleyker, E. McDonald-Buller, M. P. Fraser, D. T. Allen, D. R.
       Collins, E. Michel, J. Pudota, D. Sullivan, and Y. F. Zhu (2009), Air pollutant
       concentrations near three Texas roadways, part II: Chemical characterization and
       transformation of pollutants, Atmospheric Environment, 43, 4523-4534.
El-Fadel, M. and L. Abi-Esber (2009), In-vehicle Exposure to Carbon Monoxide Emissions from
       Vehicular Exhaust: A Critical Review, Critical Reviews in Environmental Science and
       Technology, 39, 585-621.
Isaksen, I. S. A., C. Granier, G. Myhre, T. K. Berntsen, S. B. Dalsoren, M. Gauss, Z. Klimont, R.
       Benestad, P. Bousquet, W. Collins, T. Cox, V. Eyring, D. Fowler, S. Fuzzi, P. Jockel, P.
       Laj, U. Lohmann, M. Maione, P. Monks, A. S. H. Prevot, F. Raes, A. Richter, B.
       Rognerud, M. Schulz, D. Shindell, D. S. Stevenson, T. Storelvmo, W. C. Wang, M. van
       Weele, M. Wild, and D. Wuebbles (2009), Atmospheric composition change: Climate-
       Chemistry interactions, Atmospheric Environment, 43, 5138-5192.
Saide, P., R. Zah, M. Osses, and M. O. de Eicker (2009), Spatial disaggregation of traffic
       emission inventories in large cities using simplified top-down methods, Atmospheric
       Environment, 43, 4914-4923.
Shindell, D. T., G. Faluvegi, D. M. Koch,  G. A. Schmidt, N. Unger, and S. E. Bauer (2009),
       Improved Attribution of Climate Forcing to Emissions, Science, 326, 716-718.
Shindell, D.T., et al. J. Geophys., Res., (2006) 111 D19306. doi :10.1029/2006JD007100.
Tomlin, A. S., R. J. Smalley, J. E. Tate, J.  F. Barlow,  S. E. Belcher, S. J. Arnold, A. Dobre, and
       A. Robins (2009), A field study of factors influencing the concentrations of a traffic-
       related pollutant in the vicinity of a complex urban junction, Atmospheric Environment,
       43, 5027-5037.
Wang, Y. J. and K. M. Zhang (2009), Modeling Near-Road Air Quality Using a Computational
       Fluid Dynamics Model, CFD-VIT-RIT, Environmental Science & Technology, 43, 7778-
       7783.
Zhu, Y. F., J. Pudota, D. Collins, D. Allen, A. Clements, A. DenBleyker, M. Fraser, Y. L. Jia, E.
       McDonald-Buller, and E. Michel (2009), Air pollutant concentrations near three Texas
       roadways, Part I: Ultrafme particles, Atmospheric Environment, 43, 4513-4522.
                                          30

-------
                               Dr. Laurence Fechter

Comments on ISA question 5

The discussion of susceptible populations to carbon monoxide has been dramatically improved.
The data are now presented in a logical framework providing a clear and concise summary. The
only minor change I would propose on page 5-167 2nd complete sentence is to revise as follows:
     "These analyses require the proper identification of confounders and their subsequent
        adjustment in statistical models, which helps eliminate spurious associations."

-------
                              Dr. H.  Christopher Frey
Review of Carbon Monoxide Second Draft of Integrated Science Assessment

Charge Question 1:  Chapter 1 has been revised in response to comments from the CO Panel, as
well as related comments from the CASAC PM Panel, to add information regarding criteria for
study selection and evaluation, to add more CO-specific information to the framework for causal
determination, and to more clearly describe the process of integrating evidence from various
disciplines to classify the overall weight of evidence relating to causality. What are the views of
the Panel on the extent to which this revised Chapter 1 provides necessary and sufficient
background information for review of the subsequent chapters of the CO ISA?

Response to Charge Question 1:

Chapter 1  is generally very good.

The chapter should define the terms "sensitive," "susceptible," and "vulnerable" when they are
first introduced.  The term "sensitive" seems to be in the statutory language (see footnote 1 on p.
1-3) and thus may have special regulatory significance.  This should be explained.  The role of
identifying "susceptible" and "vulnerable" groups with respect to characterization of "sensitive"
groups should be explained. Furthermore, these terms should be used consistently throughout
the chapter.  Moreover, EPA should develop a glossary of terms that are used across criteria
pollutants, just to ensure consistency of terminology for IS As and REAs for each criteria
pollutant.

Figure 1-1 should be revised.  The current figure is unclear with respect to what it is depicting. It
would be helpful if this figure follows the flow of an individual study or paper that is identified
in the literature review,  (will provide an alternative diagram).

EPA has explained the criteria for study selection and evaluation.  However, some additional
explanation as to why the focus of the literature review is on studies conducted in the U.S.  and
Canada is needed. In particular, given the scarcity of literature on welfare effects, have studies
from other countries been considered?

In Section 1.4, the topic of welfare effects should be discussed more fully. EPA should
explicitly  comment on welfare effects or lack of information about welfare effect, so that the
reader can understand the decision process that leads to lack of treatment of this topic in the ISA,
and that the omission is intentional and well-reasoned.

In Section 1.5, the text on page 1-10, line 22 refers to "the extensive body of literature" but only
four references are cited. The text could be more clear as to the scope of the literature review
and how it was narrowed to the four references cited.

The scope of the critical review of ecological effects needs more detail.  Specifically, what
literature databases were searched, using what keywords, for what time period, geographic
                                           32

-------
scope, and so on.  Given the scarcity of literature on ecological effects of CO, can EPA
nonetheless comment on hypotheses for ecological effects and identify what are the key data
gaps. Such information would be useful for setting a research agenda to inform the next revision
oftheCONAAQS.

Page 1-11, line 4, the term "necessarily" seems out of place.

The discussion of the framework for causality determination is much improved from the first
draft, and nicely addresses CO-specific examples.

Page 1-12, line 7, what is meant by "assessment?"  Does this refer to "endpoint"?

Page 1-13, line 3, the term "susceptible" is used. Here is unclear as to whether this is meant to
inform a determination of "sensitive" groups.

Section 1.6.3.  The term "measure" is unclear. Does this refer to an empirical quantity that is
measured, estimated, or predicted?  Or does it refer to a metric for a  quantity?  Suggest that the
term "measure" should be replaced with more specific or descriptive terms.

Should avoid use of "etc." (e.g., p. 1-13, line 16) and attempt to enumerate all items in a list.

P. 1-13, line 22.  An "assumption" is essentially an untested hypothesis.  For example, an
assumption that an interior indoor space is well-mixed is a hypothesis. More critical discussion
of assumptions would be helpful.

p. 1-13, line 25. Earlier, the term uncertainty "characterization" is defined as qualitative, but
here it is implied to be quantitative. Use terminology consistently.

p. 1-13, line 27. "assessing the evidence from across studies" - does "evidence" here refer to
evidence for causality, or does it refer to empirical  information from which scenarios, models,
and model inputs are inferred?

p. 1-16, line 5, please define "transfer of effects"

Table 1-2 on p. 1-20 is very useful. Another table would also be useful.  Recommend that a table
be added that relates "aspects" to the "Weight of Evidence" categories. Example:
Aspect
Consistency
Coherence
Biological
Plausibility
Biological
Gradient

Causal





Likely to be
Causal





Suggestive of
Causal





Inadequate to
Infer





Not Likely to
be Causal





                                           33

-------
Strength of
observed
association
Experimental
Evidence
Temporal
Resolution
Specificity
Analogy

























The entries in each row could either be text descriptions specific to each case, or some
combination of graphics and text. A table such of this could be used in ISAs for all criteria
pollutants.

Page 1-21, line 9-10.  Missing here  is "exposure-response."  Dose and exposure are not the same
thing, nor are exposure and concentration. Some discussion on these points would be helpful.

Page 1-21, line 13-14. Here again, terms "susceptible" and "vulnerable" are used but not
defined.  How do these relate to "sensitive"?

Page 1-21, line 29-31. Should also  mention the role of exposure misclassification if ambient
concentration is used instead of exposure.

Page 1-22, line 5:  it is not entirely self-evident that averaging will "linearize" a signal, and
assumption such as this might introduce error. If the goal of a model is to predict individual
incidences of adverse health effects (e.g., number of individuals affected), then averaging as
discussed here might be problematic.

Charge Question 6:  Chapter 2 has been revised and expanded to provide more information on
atmospheric science and exposure assessment, policy relevant considerations, and integration of
CO health effects.
    a.  The section on policy-relevant considerations was revised to present additional detail on the
       concentration-response relationship observed in a multi-center controlled human exposure
       study, present results from a new U.S. multicity epidemiologic study investigating the potential
       presence of a threshold and departure from linearity, and summarize the evidence for
       susceptible populations. Please comment on these revisions.

    b.  A section and summary figure  have been added to the end of Chapter 2 to summarize the main
       conclusions of the ISA regarding the health effects of CO and the range of concentrations at
       which effects are observed, along with uncertainties that complicate the interpretation of the
       evidence. We would appreciate CASAC comment on the material in this section and its
       effectiveness in presenting the conclusions of the ISA.

Response to Charge Question 6:

There should be more clear discussion and justification of the absence of treatment of ecological
effects.
                                            34

-------
Page 2-2, line 11-12; it may not be entirely correct to state that CO is formed by photochemical
reactions.  While there is a role of photochemistry in secondary CO formation, CO can also be
formed from chemistry involving radical attack on various hydrocarbon species. Hence, suggest
splitting this sentence into one for primary emissions of CO, and one for secondary formation of
CO.

Page 2-3, line 7, please state what is the inferred PRB for CO for CONUS.

Page 2-5, it should be stated that correlation in ambient CO concentration between monitors may
not imply the same spatial correlation in CO exposure.

Page 2-5, line 15. Exposure assessment is not complicated by multipollutant mixtures that
include CO. The epidemiological inferences may be.

Page 2-5, line 18, does "spatial and temporal variability" refer to exposure here?

Page 2-5, line 31, define "pCO"

Page 2.6, line 4, lack of definition of "susceptibility" in Chapter 1 leads to lack of clarity as to
what are the various categories of susceptibility that are not listed here.

Page 2-8, line 21-23.  A policy question is whether NAAQS should be protective of incremental
health effects to smokers from exposure to ambient pollution, and whether the concentration-
response, exposure-response, or dose-response relationship for effects associated with ambient
CO are linear or not.  These points should be clarified.

Page 2-9, line 1: do the increases  refer to smokers, or nonsmokers?

Page 2-14, line  6-7. Could clarify that the interaction is for CO as part of a mixture.

Page 2-14, should bring up exposure misclassification issues here and how they affect the weight
of evidence discussion and inferences regarding possibility of health effects.

Section 2.6, policy-relevant considerations.

Please add a table that defines and lists attributes  of "susceptible," "vulnerable," and "sensitive"
Page 2-16, line  21, is the 10-15% increase on a relative basis or in terms of COHb percentage
points? Reader infers the former,  but this could be more clear.

Section 2.6.2  concentration-response
The chapter would benefit from a  discussion somewhere of the difference between
concentration, exposure, and dose. Terms should  be used consistently. For example, p. 2-18, line
32 refers to "dose-response" but might actually be based on potential dose or exposure.
Similarly, top of page 2-19, isn't it the case that clinical studies deal with potential dose and not
merely concentration?
                                           35

-------
Table 2-1, label the number scale at the bottom of the last column - i.e. define "effect estimate."

Page 2-22 seems repetitive of Section 2.5

Page 2-22 and 2-23.  There seems to be contradictory text to the effect that exposure
misclassification leads to bias  (see p 3-113, lines 2-4) and then later that it would only widen
confidence intervals (p-23, lines  17-18).

Page 2-25, line 14 - seems to presume a linear dose-response relationship. This should be stated
and discussed.

What about ecological effects? Health effects associated with climate change?
                                            36

-------
                                Dr. Milan Hazucha

Revised Comments on Chapter 4: Dosimetry and Pharmacokinetics of Carbon Monoxide
of the Second External Review Draft of the ISA for Carbon Monoxide

Charge: "In response to comments from the CASAC CO Panel, material has been added to
Chapter 4 describing comparisons among predictive COHb models, the relative influence of
differing exposure scenarios on COHb concentration, and endogenous CO production rates in
individuals with various diseases and conditions. Please comment on the usefulness of this
information in illustrating the factors influencing COHb kinetics and potential COHb levels
under various scenarios ".

This chapter of the Second Draft is much more comprehensive in discussing the respective
material. The Chapter has been expanded by more than one third. New subsections were added
(4.2.3, 4.3.4, 4.4.3.1) and most of the old subsections were expanded, some substantially (4.2.4,
4.5). This is mostly to the benefit by facilitating better understanding of the section topics.
In general, the authors adequately addressed CAS AC's CO panel comments by appropriate
revisions and addition of relevant material discussed in sufficient detail.  One question, however,
which in my view was not satisfactorily answered, is "Which COHb model is the best in
estimating venous COHb"?

Section 4.2.1 The Coburn-Forster-Kane and Other Models
The discussion of various models has been slightly expanded and a most recent model  by
Gosselin et al, 2009 is discussed as well. This model has been developed for and commissioned
by Health Canada, Air Health Effects Division. It is a comprehensive model based on CFKE and
it seems to estimate experimental data very well under a variety of environmental  and
occupational conditions.
Section 4.2.3 Model Comparison"
This is a new very helpful section. It discusses strengths and weaknesses of various models
reviewed in previous sections. However, at the end, there is no conclusion, no recommendation
as to which model is the best in estimating venous COHb. With so many different COHb
prediction models it will be difficult for most of the readers to select the best model. If not here,
maybe section 4.6 Summary and Conclusions could be more specific.
Section 4.2.4 Mathematical Model Usage
This is a substantially expanded section by discussing comprehensively The Quantitative
Circulatory Physiology  (QCP) model supported by several plots. Extensive discussion of this
model seems to suggest that this is another "preferred" model for COHb estimation. So it
appears that we now have two "preferred" models, Gosselin et al, 2009 and QCP.  Again, which
one gives the best estimation of venous COHb? Since these models have been described in a
considerable detail, why not to compare COHb estimates utilizing one of exposure profiles, e.g.,
like in fig. 4-2. Moreover, all of the discussed models are predicting venous COHb. Is it possible
to use these two or any other models to estimate transient arterial COHb level?  It would be
helpful to have a one paragraph discussion of utility of these models, if any, in estimating
transient arterial COHb  level if such data exist.
                                          37

-------
Section 4.3.2.4 Other Tissues
Although this is only a page long section with two tables, I was (in the first draft) and am still
struggling with presented material. There is a substantial discussion of animal studies. However,
the data were based on CO exposures with COHb levels as high as 80%. I do not think that these
data are relevant. Maybe, table 4-2 showing human data, though some at very high COHb would
be sufficient, and drop table 4-3.
Section 4.3.4 COHb Analysis Methods
This new section gives a very good discussion of current methods used for COHb analyses. It
discusses advantages and limitations of various methods which is helpful in interpretation of
data.
Section 4.4.3.1 Fetal Pharmacokinetics
Short and concise new subsection with a figure, pointing out to maternal-fetal differences in
COHb buildup and elimination. However, there are more recent studies published on
maternal/fetal COHb correlation that should be briefly discussed as well (Hayde et al., Early
Human Development 58:205-212, 2000 and other articles from this group., Ziaei et al, Paediat
Perinat Epidemiol 19:27-30, 2005). Although these studies are concerned with specific diseases
they have used healthy controls.
Section 4.5 Endogenous CO Production and Metabolism
Substantially expanded and quite comprehensive. The authors went beyond CAS AC's CO panel
suggestions for revisions and discuss in detail, including very helpful tables, various health
conditions and diseases that can increase endogenous CO production and subsequently elevated
COHb. This is all supported with abundance of references. It is an excellent review.
More specific comments:
Reference list needs to be updated.
Page 4-3, lines 1-9: It would be easier to follow parameter and variable description if they were
       listed in two columns.
Page 4-3, line 17 and p.4-9, line 7: Clarify. Do you mean ±0.5% of the nominal value?
Page 4-5,1. 9: Which two parameters? Be more specific.
Page 4-9,1.13 Clarify. Is it Gosselin's model?
Page 4-9,1.14: Clarify. Is it linear or non-linear CFK model?
Page 4-10,1. 6: There is no 4 ppm value in table 4-1.
Page 4-10, table: increase font size for VA
Page 4-10,1. 16: This study was done in police cars which are regularly maintained and tuned.
       So the real CO value is somewhere between 5 and 50 ppm.
Page 4-14,1.14: insert after "interface" the words "into plasma and subsequently into RBC"
Page 4-15, fig. 4-7. Unusual referencing of the source. Why not simply say that the source is
       U.S.EPA2000.
Page 4-16,1. 22: The value for Haldane constant M is reported to be 218. However, some
       sections report the use of other values, like 230. The M value should be used uniformly,
       whenever possible.
Page 4-17,1. 17: Suggest replacing "quickly" with "2-10 min".
                                          38

-------
Page 4-20, table 4-3; I am not sure that we need this table. For most of exposure conditions listed
       in the table COHb levels are well beyond the scope of this document. Suggest deleting.
       The three sentences in the text (linel 1-14) are sufficient.
Page 4-21,1. 2-15: Similarly, the discussion of rodent's data does not seem to be too relevant.  In
       some referenced studies, though not on the list, %COHb levels were as high as 80%.
Page 4-23,1.17: "distribution" might be a better word than "uptake".
Page 4-28,1. 27-29:  Delete, not relevant.
Page 4-31,1 6: Suggest replacing "processes" with "function"
Page 4-31,1.8: Suggest replacing "combat" with "compensate for"

Chapter 5: Integrated Health Effects of the Second External Review Draft of the ISA for
Carbon Monoxide  (sections 5.1 and 5.5).

Section 5.1 Mode of Action of CO toxicity.  This revised section covers in adequate depth
various mechanisms of CO effects at a cellular level, including NO and CO signaling, redox
status and modulation of kinase activity. I would highlight very important but easily overlooked
determination stated on p.5-16, line 7-8 which says that "...the  situation of increased
endogenous CO production and of exogenous CO exposure are not equivalent." This
distinction is critical to understanding the cellular mechanisms  of action of CO from different
sources. Thus, exogenous CO tissue effects at low concentration are more general and the
pathways of action are not necessarily the same as that of endogenous CO.

In contrast, the summary statement on p. 5-31,1.25-26, over interprets the reviewed studies in
this subsection (5.2.1.8). Considering all the caveats these studies report, there is a lack of
coherence between the endpoints and the evidence of the effects is of uncertain significance.

Section 5.5 Respiratory Effects. The author(s) of this section should have been more critical
evaluating the studies  discussed in this section, particularly when summarizing the findings. For
example, on p.5-118,1.13-17 how can the Asthma study findings, to quote "suggest a potential
effects of CO on lung  function at relatively low CO concentration.." when CO is 3.8 ppm? This
is a concentration which will result in <1%  COHb. Moreover, in asthmatics the endogenous
production is higher than 3.8 ppm (section 4.5)! At this level CO has no effects on lung function!

Similarly, contrary what is stated on p.5-120,1.7-9 European studies do not provide stronger
evidence than the US studies.  Their findings are also full of caveats which make the conclusions
uncertain.

Page 5-143,1.13-14 state that "epidemiologic studies provide evidence of positive
association	" However, statements on 1.18-23 which is a correct summary of available
evidence contradict  this assertion.

Page 5-143,1.29-30 statement should be reconciled with subsequent statement on 1.31-32 which
correctly summarizes the available evidence.
                                           39

-------
                               Dr. Michael Kleinman

Charge question 3:  Material has been added to Chapter 4 describing differences among models
that predict COHb concentrations as a function exposure and physiological parameters. This is
useful however the summary and conclusions do not make it apparent which model will be
preferred for health risk assessments and why.

Specific Comments: The new illustrative material could be better coordinated. Table 4-1 for
example demonstrates that COHb concentrations increase with increasing ventilation rates after
1- and 8-hr of exposure but begin to decrease at 24 hr. The explanation may come later in the
chapter but it would be useful to mention the rationale in the description of the Table.

Figure 4-4 does not seem to agree numerically with Table 4-1 for the higher exposure
concentrations. Also COHb levels for exposures at 20 ppm seem to be increasing during the first
8 hours and those at 50 ppm are decreasing.  The curves appear to be show them approaching the
same concentration if the subject continued to sleep.  It would be helpful to add a graph of Va
used for the model keyed to the right Y axis.

Figure 4-5 data for endogenous production 0.007 does not appear to be consistent with Table 4-
1.

Figure 4-6 seems unnecessary since the scenario it presents is not related to any real-world case
and its importance is not explained in the text.

The Bruce and Bruce model is claimed to better predict COHb levels when inspired CO levels
change rapidly, as might occur during start-up conditions in  some combustion emission scenario
and is said to better predict CO washout than  does the CFK.  However the previous examples
seem to have been calculated using the CFK.  Some reason for why the Bruce and Bruce model
is not selected would be useful.
                                          40

-------
                                 Dr. Francine Laden
Section 5.1.3.2. Recent Studies of Non-Hypoxic Mechanisms, is an excellent summary and I
agree with its conclusions. The multicenter controlled human exposure study is well described
and the levels at which effects were observed are now clear.

One editorial comment, the definitions of Hb and Mb should be repeated at the beginning of each
chapter

The descriptions of the epidemiologic studies and the studies  of the associations between blood
markers and ambient CO concentrations are good. The following reference should be added to
5.2.1.8:
       Delfmo RJ. Staimer N. TioaT. Gillen PL. Polidori A. Arhami M. Kleinman MT. Vaziri
       ND, Longhurst J, Sioutas C. Air pollution exposures and circulating biomarkers of effect
       in a susceptible population: clues to potential causal component mixtures and
       mechanisms. Environ Health Perspect. 2009 Aug; 117(8): 1232-8.

A table or figure summarizing the results from the blood markers studies (much like the ones
included later in the chapter) would be very helpful.

A similar table or figure summarizing the results of the HRV, ECG abnormalities, arrhythmias,
blood pressure (Sections 5.2.1.1. through 5.2.1.7. would be helpful as well.

In Figure 5.6. it should be made clearer that the other pollutants, used as separators, are included
as co-pollutants in the  models of CO with the different cardiovascular outcomes.

    I agree with the conclusions of each of the other health outcomes: CNS, birth outcomes,
    respiratory effects, mortality , and of susceptible  populations.
                                           41

-------
                                  Dr. Arthur Penn

Initial response to CO ISA, 2nd external review draft

Many of the conclusions presented in the 2nd external review of the ISA, are retained
(understandably) from the 1st ISA, My focushere is on areas of CVD-related outcomes on which
less emphasis has so far been placed and which I believe deserve additional attention since they
deal with biological plausibility of CVD outcomes in response to elevations in low daily ambient
CO levels. These outcomes also are consistent with the statement at the top of p. 5-67 "It is
conceivable that the most sensitive individuals respond to levels of COHb lower than 2%" as
well as with the "causal relationship" statement at the bottom of that page.

Ambient CO Effects on CVD

The most impressive CO-related CVD results remain the 20+-year-old controlled human
exposure studies  of Allred et al; Kleinman et al; Sheps et al; however, a direct
connectionbetween these results and the predictions for CO effects on CVD morbidity/mortality
at CO levels close to ambient has yet to be made. The effective CO exposure levels in those 3
studies were > 2 orders of magnitude above ambient levels and resulted in COHb levels of 2-
4%.With ambient CO levels at 0.5-0.6 ppm and associated COHb levels well below 2%, the gap
between a) the controlled studies with  small numbers of high-risk volunteers exposed
to>100ppm CO and b) real-life,large population exposures to small increases (<1 ppm) in daily
max [CO] is too large to discount at present. Further, other studies (Adir et al, 1999; Kizakevich
et al, 2000) with healthy volunteers suggest little or no major responses to elevated (as high as
3000 ppm) CO exposure levels. In those studies there were no reported arrhythmias, no changes
in lactate/pyruvate, no effects on ST-segment changes or on cardiac rhythm.
The focus on possible CO effects in patients with major artery occlusion and MI history  is
understandable from the perspective of a potentially highly susceptible population, but moves
attention away from other populations that may be more likely at risk to elevations in ambient
CO.

Alternative populations meriting attention are the groups suffering from CHF (pp. 5-43 to 5-45)
and arrhythmias (pp. 5-24 to 5-26). While most of the evidence here is carried over from the 1st
ISA Draft, some of the reports summarized in this section + others noted in other sections of
Chapters are worthy of further consideration. In addition I have added some studies from the
past 12 years that were not mentioned  in the 2nd ISA Draft.

Results reported by Yang(JTEH, 2008) on Taipei data for the years 1996-2004-while CHF
hospital admissions (HAs) were associated with all 5 major air pollutant groups for warm days,
the only association on cold days was withincreases in ambient CO.

Mann et al (EHP, 2002) reported that a Ippm increase in 8-hr average CO in So. California was
associated with a 3.6% increase in same-day IHD HAs for patients with a 2°diagnosis of CHF
and 2.99% increase for those with a 2° diagnosis of arrhythmias.
                                          42

-------
Peel et al, (Am J Epidemiol, 2007) reported an association between a 1 ppm elevation in 1 hr
max CO and HAs for patients with dysrhythmias and CHF who had hypertension as a co-morbid
condition. This was an 8-year study with > 4.4 million patient visits to 31 Atlanta area hospitals.

Other relevant CO/CHF studies include:

a) Morris et al (AJPH, 1995)—elevated ambient CO levels in 7 US cities were associated with
increased HAs for CHF in elderly patients;

b) Burnett et al (Epidemiology, 1997)—daily high hour ambient CO levels on day of HA had the
strongest association of any of the 5 major air pollutants with HAs for CHF;

c) Morris and Naumova (EHP, 1998)-HAs in Chicago for CHF were most strongly associated
with increases in ambient CO-effect was strongest at lowest temperature (see Yang, above);

d) Stieb et al (Environ Hlth, 2004)~in a multicity study in Canada (1980s & early 1990s), for
every 0.7 ppm increase in 24-hr mean [CO], there was a 2.6% increase in ED visits for
Ml/angina,  but a 3.8% increase in visits for CHF;

e) most recently, Bell et al (EHP, 2009- in ISA reference list, but not discussed??) in a study of
emergency  HAs for CVD and their association with 1 hr max. CO levels in 126 US urban
counties (av. max CO level=1.6 ppm) found the  highest % increase in CO-related risk for HAs
(-1%) was  for heart failure in patients > 65 yrs of age. HAs for 9.3 million patients over 7 years
were examined.

The downside of these studies-that they are association/correlation studies—is countered by the
large #s of patient records screened in each of these independent studies and the similarity of the
findings for urban CHF/arrhythmia patients in the US, Canada & Taiwan.
Blood Markers of CO Exposure-Coagulation (but not inflammation)
A few recent studies (Baccarelli et al, 2007; Delfmo et al, 2008; Rudez et al, 2009) point to
increased platelet activation and pro-coagulation effects associated with elevations in ambient
CO.  In these and other studies (Ruckerl et al, 2006, 2007; Steinvil et al, 2008) elevations in
fibrinogen in response to elevated CO are largely absent. Many of these studies note that
elevations in ambient CO were not associated with any inflammatory responses (see question to
Panel members below). One exception was the recent report of Ljungman et al, (EHP, 2009).
Among 955  MI survivors, the 16% with specific polymorphisms in both IL-6 & fibrinogen genes
showed larger IL-6 responses to elevated CO than did MI survivors without these
polymorphisms.

Q. for Panel members: In light of the EPA's interest both in controlled human studies with
responses to exposures to> 100 ppm CO and responses of large populations to 1 ppm increases
in peak ambient CO, are there any Panel members concerned (intrigued?) by the growing interest
in therapeutic uses of CO as an anti-inflammatory agent? A number of recent studies on animal
models of injury/disease (sickle cell disease, I/R injury, lung injury associated with
                                           43

-------
cardiopulmonary bypass) have reported on the therapeutic value of treatment with "low", i.e.,
250 ppm, doses of CO.
                                          44

-------
                                    Dr. Beate Ritz

    1.  The framework for causal determination presented in Chapter 1 was developed and
       refined in other ISAs (e.g., the PM ISA). During previous reviews, CASAC generally
       endorsed this framework in judging the overall weight of the evidence for health effects.
       Please comment on the extent to which Chapter 1 provides necessary and sufficient
       background information for review of the subsequent chapters of the CO ISA.

    This chapter has improved but still does not adequately address and present methodologic
concepts in epidemiology and, thus, lacks clarity in how epidemiologic studies are evaluated and
determined to be "high or low quality studies" as necessary for applying the criteria listed in
Table 1.2 (i.e. for assessing the weight of evidence for causal determination).

    A minor point:  I previously recommended using the more appropriate term 'effect measure
modification' instead of 'effect modification' but the wording in this chapter has not been
corrected. More importantly, however, while there is some improvement, the authors of this
chapter still  seem to not be fully understanding nor formulating adequately some of the issues
involved in confounding and confounder control. They claim on page 1-15 that "deciding which
variables to control for in a statistical analysis of the association between exposure and disease or
health outcome depends on knowledge about possible mechanisms and the distribution of these
factors in the population under study. Identifying these mechanisms ...". Knowledge of
'mechanisms' may help, but such knowledge is not needed to decide whether a covariate is a
potential confounder neither is it necessary to know mechanisms to assess confounding. It is
furthermore completely obscure what the authors mean by the following sentence on page 1-15
"adjustment for potential confounders can be influenced by differential exposure measurement
error"; here they seem to confuse error in measuring confounding variables with error in
exposure assessment? Finally on page 1-13 the second sentence under  1.6.3. "Uncertainty  can be
defined...."  seems  to confuse precision and validity or at least does not acknowledge that these
are two different concepts that have a different place in judging study results. These confusions
of concepts does not instill much confidence in the ability of staff who wrote this chapter to
judge epidemiologic studies adequately according to established criteria for study validity  and
precision (both contributing to accuracy); this is further confirmed by the chapter 5 qualitative
reviews that are still grossly lacking in consistency and interpretation of epidemiologic results.

    The criteria for causal determination detailed in table 1-2 are similar to those used by the
IOM and the International Agencies for Research on Cancer. Yet, they leave open what the
criteria are for deciding that a study is high quality (for example, confounding is a bias, so why
list bias and confounding apart from chance?) and it is also unclear what is meant by "replicated"
results and why this would be a criterion. Again, without a standardized approach to the review
of epidemiologic studies or a quantitative meta-analysis based review,  these criteria remain
ambiguous.  Since the epidemiologic literature on criteria air pollution health effects has
multiplied greatly in the past decade, it would be appropriate if EPA staff abandoned qualitative
reviews in favor of quantitative effect estimates based on meta-analytic procedures to draw
inferences about the scientific literature and used standardized and transparent rules for data
abstraction.  Such a systematic and quantitative procedure requires making the authors'
                                           45

-------
assumptions explicit rather than allowing authors to emphasize studies they agree or disagree
with and to pick the results they like to emphasize over others. Such quantitative reviews could
be contracted out to entities that are able to conduct meta- or pooled analyses.
   5.  Chapter   5  presents   information   on  cardiovascular,   central  nervous   system,
       developmental, respiratory, and mortality outcomes following exposure to CO.  To what
       extent are the discussion and integration of toxicological,  clinical, and epidemiologic
       evidence for these health effects scientifically sound, appropriately balanced, and clearly
       communicated?   Are  the  tables  and figures presented in  Chapter  5 appropriate,
       adequate, and effective in advancing the interpretation of these health studies?

   In Chapter 5, the qualitative description of epidemiologic studies improved somewhat but is
still inadequate; the level of detail devoted to each study in the text seems still arbitrary and the
information provided in tables and figures selective without being  systematic; for example why
did the authors decide to present in Figure 5-8  the citywide and negative associations for the
Australian CO study of PTB (Jalaludin) and not the positive associations for births within a 5km
radius of a monitor. The review of birth weight and air pollution is lacking a discussion of the
difference between LEW and term LEW sorely needed since LEW includes preterm birth
outcomes that are then discussed separately and studies examining LEW are possibly more
comparable in their results to those examining PTB; only term LEW is a mutually exclusive
outcome. Also the measure of birthweight as a continuous outcome compared to the
dichotomous variables LEW and PTB deserve some more general  introduction about their
general value (similar to the discussion of SGA versus IUGR), i.e. do we really expect the whole
birthweight distribution to shift according to ambient air pollution  exposures or only the most
susceptibly infants to be affected.

   Surprisingly, there is still a lot of information I requested in my first review missing from this
new  draft. This includes the following: no information is provided in the tables concerning the
type  of study  design employed (e.g. Table 5-12).  I also already mentioned previously that many
of tables report mean CO levels and mention 24 hrs or 8 hrs in brackets; this is misleading for
pregnancy outcome studies in which the averages are for trimesters, weeks, or months (e.g. the
Ritz  et al. (2000) study of PTB is listed in table 5-12  as having a Mean CO of 2.7 ppm for the 6-
9 am period - however this mean represents a mean over the whole first month of pregnancy and
the Wilhelm and Ritz (2005) study mentions a 1.4 ppm mean for 24 hrs but this is in fact & first
trimester mean of 24 daily measurements; the way this data is shown now the bracketed 24 hour
mention seems to imply  similar  averaging period and comparability in effect estimates. I also
mentioned already previously that while the Ritz et al 2007 study is listed in table 5-12 no results
for this study are presented in figure 5-6.1 had also recommended  to rescale quartiles to a
continuous  estimates rather than leaving results from important papers out of a figure that gives
an overview over all study results.

   Also I mentioned previously that according to the text accompanying the figures, the
estimated increase in CO presented have been  'standardized',  however, how this might have
been done across so many  different study types and averages for differing exposure periods
(rather than 24 hour averages as the authors of these chapters seem to imply) has not been
                                           46

-------
explained. Also, in figure 5.1 the title says that the effect estimates have been standardized to a
Ippm increase in ambient CO for 1-hr max CO concentrations, 0.75 ppm for 8-h max CO
concentrations and 0.5 ppm for 24 hrs avg CO concentrations, but the figure does not tell us
which scale has originally been used in which study and it might be  questionable whether effect
estimate sizes based on these different scales and based on different length lag periods are
comparable to each other (indicating which study used which scale would be informative.

   There are also sentences in this review chapter that are plainly wrong, e.g. on page 5-71 and
OR of 1 (95% CI 0.96-1.04) is called a positive association.
                                           47

-------
                                 Dr. Paul T. Roberts

Revised Comments on 2nd draft ISA

ISA Charge Question 2. Chapter 3 has been revised and expanded in response to Panel comments
regarding climate, monitoring, spatial variability, and exposure.

2b. Additional detail has been provided regarding	  Please comment on the usefulness of these
revisions in characterizing the information provided by the CO monitoring network.

In general, the expanded discussions in the 2nd draft ISA Chapter 3 on CO detection limits, monitoring
details, and spatial CO characteristics are very useful in characterizing the information provided by the
CO monitoring network and in qualifying the data for use in exposure estimations. My detailed
comments are provided below.

In discussion of non-anthropogenic CO emissions on page 3-5: it is confusing in the first paragraph
(starting on line 3) to have fire emissions of about 13% (14.5 MT) shown in Figure 3-1, but biogenic
emissions of about 5% not shown in Figure 3-1, and the text implies that the geogenic emissions are
included (in the miscellaneous category?).  This is confusing to the reader and makes  it difficult to
compare these smaller, but still important sources. Please add biogenics to Figure 3-1 and make it clear
what is included.

Comments on discussion of Hudman et al and Figures 3-3 and 3-4, starting on page 3-5 at line 21:
First, I suggest that this be a new paragraph; it is a different topic from the non-anthropogenic emissions.
In addition, the CO from oxidation of VOCs, of isoprene, and of other biogenic VOCs (see lines 27-29),
which are apparently huge in this simulation, relative to the anthropogenic CO emissions, have not been
discussed before. These secondary emissions sources needed to be discussed  in the overall context of
CO emissions first (or put the general discussion in with the Climate text, Chapter 3.3.1).  In addition,
the potential influence of this huge source of CO on the results of the simulation and the conclusion
needs to be discussed.  I also suggest that Figure 3-3 be dropped, since I find it hard to compare these
colored spatial plots.  In contrast, Figure 3-4, as an example of the results, gets the point across that
reducing the anthropogenic emissions by 60 % made significantly better comparisons with
measurements. Maybe also  add a statement saying that other results in the paper support this general
conclusion.

Additional comment on the paragraph on page 3-5, lines 21-29: I suggest that a sentence be added
translating the Tg amounts to MT, so that these emissions can be placed in context with the rest of the
emissions discussion.

Comments on page 3-7, lines 13-15:  I suggest that a comment be added (either in the text or the figure
caption) about general transport winds being from west to east at this latitude, thus carrying the
emissions from the Alaska fires across Canada and the northern US and  into the north Atlantic, as shown
in Figure 3-5 (assuming the data support this).
                                          48

-------
Page 3-11, lines 9-10: The comment on the significant quantities of aromatics in gasoline is likely no
longer true, since regulations have significantly reduced aromatics in gasoline (and these references are
old). I suggest that these comments be modified to say that this used to be the case, but less so with
current fuel content.

Page 3-12, lines 8-9: The comments on limited mixing between the hemispheres would benefit from an
additional comment that northern hemisphere CO emissions are significantly larger than the emissions
in the southern hemisphere, plus a representative reference.

In general, the expanded discussions on monitor detection limits and monitoring locations in
Chapter 3.4 (pages 3-18 to 3-31 and  associated Annex figures and tables) are critical to
understanding CO concentrations and are an important addition to the ISA.  However, these
limitations are still often left out of the discussions on CO concentrations in Chapter 3.5.  In
general, the relaxation of CO monitoring requirements and the continued use of older, less
sensitive, monitors with poor levels of detection impedes the use of monitoring data for exposure
assessments and climate, especially in the future as CO concentrations decrease. See my detailed
comments below.

Page 3-19, lines 20-23: This discussion on the needs for trace-level CO measurements should include
the use of low-level CO data for improved exposure estimates at current ambient concentrations in many
locations in the US.

Table 3-2: Please fix the first row of the table. I think there should be a header row labeling the
columns as "Parameter" and "Specification", for example, plus the current first row should be part of the
body of the table and left justified in each column.

Page 3-20 lines 3,12, and 18:  The LODs listed on these lines are not the same as listed in the
referenced Table A-l in the Annex, which lists the LOD of the trace-level monitor as 0.02, not 0.04.
Thus, the value listed as 50% of the LOD on line 6 should be 0.01.  Note also that several of the LOD
levels listed in slide 10 of the presentation should be 0.02,  not 0.04.

Page 3-20, line 5:  I suggest that this line start with "When the monitored value is below the LOD, some
states...

Page 3-20, line 14, page 3-21 and Figure 3-8: This discussion on a comparison of older and newer
monitors with specific quantifications  is a good one, and important to include here. However, the last
sentence of the paragraph (lines 6-8) does not make sense to me.  Also, the CO axis of the figure needs
to be labeled and the units of the time axis needs to be added (hours since some start time, or?). The
similar figure in slide 10 of the presentation is much better at showing the data, plus the axis are labeled;
these are good modifications and address my comments on the figure.

The limitation of the LOD issues discussed in Chapter 3.4  need to be added in several places in Chapter
3.5, including at the beginning of the sentence that starts on line 11  of page 3-36, in Figures 3-7, 3-8,
and 3-9.  In particular, statements similar to the paragraph  at lines 13-18 on page 3-43 (good job there!)
could be added to address this issue at these locations in the text and in the conclusions (3.7.3) and the
summary of conclusions (2.1).
                                          49

-------
Page 3-41, lines 1-3: What was the cause of the 10.9 ppm CO measured at the Newkirk, OK site? This
seems like an unusual concentration.

Page 3-43, lines 13-18: This is a good qualifying paragraph which is needed here and other places.  On
line 15,1 suggest the following wording: "...in large part very near or below the detection ..."

Figure 3-18 on page 3-45 (and Figure 3-20): How is the data below LOD treated for this and similar
figures? Might this influence the lower ends of the box-whisker plots?  I suggest that a comment on this
be added.

Chapter 3.5.1.2: Add a note that all of these monitors in LA and Denver are older, higher LOD,
monitors, and add some comments  similar to the comments at lines 13-18 on page 3-43.

Page 3-60, lines 14-15: I don't see how the data shown in Figure 3-24 can lead to a statement that
includes the following words "...near-road CO concentrations.." .

Page 3-76, lines 5-6: I suggest this sentence be moved to the end of the paragraph at lines 12-21 on
page 3-77.  In addition, the sentence should read "....analogous to Figures 3-36 and 3-37 for ....in Annex
A, Figures A.44 to A.48."

Page 3-77, line 11:  I suggest these are meteorological, not micrometeorological factors.
2c. The section on exposure assessment has been reorganized to provide information on	 Does
the Panel consider that the sources of exposure error have been appropriately characterized, and
agree with the revised conclusions regarding the impact of exposure error due to spatial
variability and the presence of CO as part of a combustion-related mixture on health effect
estimates from time-series epidemiological studies?

In general, the expanded discussions in the 2nd draft ISA Chapter 3.6 on sources of exposure and
exposure assessment are a great improvement and are very useful in characterizing the potential impacts
of exposure error. My detailed (minor) comments are provided below.

Page 3-93, lines 18 and 29: Please explain or modify the terms "driven cavity" and "posterior
probability distribution function".

Page 3-94, lines 6-23, section on Land Use Regression Models:  This section is still very limited in
scope and does not represent the wide range of results from the literature. Admittedly, much of the LUR
work in the literature is on pollutants other than CO, but the types of conclusions regarding what
methods work and how they relate to estimating pollutant concentrations are directly applicable to CO.
See the list of references I suggested last time (re-listed at the end of my comments), plus there must be
many more than I could easily find.
                                          50

-------
Page 3-98, line 11: How can a regression coefficient be 1.99 (greater than 1.0)? Also, the results in
lines 9-11 and in lines 13-14, although from the same reference, seem inconsistent; please explain how
they are consistent or different.

Page 3-114, line 26:  Table A-l says the LOD for trace-level FRMs is 0.02, not 0.04 as stated here.

Page 3-115, lines  13-17: Please add the limitation statement on LOD to this section regarding
characteristic concentrations.

Page 3-117, line 15:  I suggest that the word "nearby" be added, so that the sentence would read "...at a
location with few nearby CO sources could...".
In summary, Chapter 3 of this 2nd External Review Draft of the ISA clearly conveys and
appropriately characterizes the atmospheric science and air quality analyses. The information
provided regarding CO source characteristics, CO chemistry, policy-relevant background CO,
and spatial and temporal patterns of CO concentrations accurate are relevant to the review of the
CO NAAQS.
Minor edits and typos in the 2nd draft ISA:
       page 3-2, line 22: word near end of the line should be "inherent"
       Page 3-41, line 16: suggest that "medians" be replaced with "median correlation coefficients (r)"
       Figures 3-17 and 3-19: I can barely make out the lines for the highways (whereas the ones in the
       Annex are fine); lease make darker.
       Make bolder the lines separating the scales  in Tables 3-10 and 3-11; it is currently difficult to
       read.
       Page 3-72, line 2 should read:  "...as shown in Figure 3-6." not in Figure 3-32.
   -   It is very hard to see the 95th and 5th percentile lines in Figures 3-33 and 3-34; please make
       darker or bolder.
       Page 3-85, lines 17 and 24-25: I suggest that you use words other than "fidelity" in line 17 (and
       line 4 of page 3-86) and "fraught" in line 25; maybe "accurately" and "are difficult". Also, add a
       comma after troposphere in line 24.

Selected, easy for me to find, references for Land Use Regression and spatial mapping (see above
discussion on Chapter 3.6.3):

Gauderman, Avol, Lurmann, Kuenzli, Filliland, Peters, and McConnell "Childhood Asthma and
Exposure to Traffic and Nitrogen Dioxide, Epidemiology 2005; 16, 737-743.

Ross, Jerrett, Ito, Tempalski, and Thurston "A land use Regression for  predicting fine particulate matter
concentrations in the New York City  region", Atmospheric Environment 41  (2007) 2255-2269.

Hoek, Beelen, Hoogh, Vienneau, Gulliver, Fischer, and Briggs "A review of land-use regression models
to assess spatial variation of outdoor air pollution"  Atmospheric Environment 42 (2008) 7561-7578.
                                          51

-------
Henderson, Beckerman, Jerrett, and Brauer "Application of Land Use Regression to Estimate Long-
Term Concentrations of Traffic-Related Nitrogen Oxides and Fine Particulate Matter ES&T 2007, 41,
2422-2428.

Molitor, Jerrett, Chang, Molitor, Gauderman, Berhane, McConnel, Lurmann, Wu, Winer, and Thomas
"Assessing Uncertainty in Spatial Exposure Models for Air Pollution Health Effects Assessment EHP
vol 115,no 8, August 2007.

Popawski, Gould, Setton, Allen, Su, Larson, Henderson, Brauer, Hystad, Lightowlers, Keller, Cohen,
Silva, and Buzzelli "Intercity transferability of land use regression models for estimating ambient
concentrations of nitrogen dioxide" J Exposure Science & Environmental Epidemiology (2008), 1-11.
                                          52

-------
                               Dr. Armistead Russell

                               Review of CO ISA 2nd Draft

In general, I am pleased with the modifications to the ISA, and believe that the 2nd draft is
stronger in general.  It provides the level of information needed to support the REA and policy
analyses.
In response to specific Charge Questions:
2. Chapter 3 has been revised and expanded in response to Panel comments regarding
climate, monitoring, spatial variability, and exposure.

I appreciate the substantial information added in regards to the potential impact of CO on
climate. As noted, the impact is likely small, and highly uncertain, though the physics are such
that it almost has to have an impact, even if unknown or not soon knowable.  While this lack of
certainty may inhibit developing a related secondary standard, it should motivate the appropriate
research to assess the likely magnitude of the impact. The section on monitoring and
instrumental capabilities is likewise strengthened.  This section should continue to stress the
utility of CO as an indicator for gasoline-powered automobile emissions, and the CO monitoring
network has tremendous value beyond just demonstrating attainment.

       a. Evidence reviewed in Chapter 3 of the ISA indicates that the direct contribution of CO
       to greenhouse warming is very small, while the role of CO in atmospheric chemistry
       cycles involving other species makes a larger contribution to radiative forcing.  This
       combined evidence leads to the conclusion in Chapter 2 that a causal relationship exists
       between current atmospheric concentrations of CO and effects on climate. What are the
       Panel's opinions related to this causal statement and the evidence provided to support it?

       A causal statement is appropriate, though it should also note that the extent of the impact
       is highly uncertain which inhibits using a causal determination to develop a secondary
       standard at this time.

       b. Additional detail has been provided regarding the detection limits of CO monitors in
       the regulatory network, the number of monitors reporting at each horizontal spatial
       measurement scale and comparison of monitoring data at each scale, and spatial
       variability of CO concentrations near major sources, particularly roadways. Please
       comment on the usefulness of these revisions in characterizing the information provided
       by the CO monitoring network.

       This addition strengthens the document. Further discussion of the adequacy of the
       current monitoring network and the potential implementation of a near-road network
       should be considered.
                                           53

-------
                                 Dr. Anne Sweeney
Legislative Requirements: Page 1-3. Lines 25-28 (selecting a margin of safety): include a
reference to the EPA's Supplemental Guidance for Assessing Cancer Susceptibility from Early-
Life Exposure to Carcinogens (2005)

Identification of studies for inclusion in the ISA: Page 1-7, line 4: "included approaches to
evaluate issues related to confounding and effect modification by other pollutants"—add "and/or
host characteristics"; i.e., the level of control for all potential confounding/effect measure
modification.
Page 1-7, line 5:  "Addressed health points and populations not previously extensively
researched": a word of caution in that new evidence can alter previously accepted results.

Scientific Evidence Used in Establishing Causality (1.6.1) Page 1-12, lines 10-13: Add case
control design to types of observational studies

Page 1-7, lines 7-10 (and elsewhere): need to discuss volunteer bias in human clinical studies

Application of framework for causal determination: (1.6.4) Page 1-17: Add to line 21:
Strength of the association"

Effects on Human Populations: (1.6.5.1) Page 1-22, lines 26-32:  Include references to articles
published regarding critical windows of susceptibility (from the 2000 EPA workshop on same)
I strongly agree with Dr. Ritz's suggestion to conduct meta-analyses on the growing number of
studies assessing air pollution and adverse human health effects.

Additional references on Gestational Development

Jensen TK, Bonde JP, and Joffee M. 2006. The influence of occupational exposure on male
reproductive function. Occup. Med.  (Lond.) 56(8):544-553.
Hauser R, Meeker JD, Duty S, Silva MJ, and Calafat AM. 2006. Altered semen quality in
relation to urinary concentrations of phthalate monoester and oxidative metabolites.
Epidemiol. 17(6): 682-691.
Hauser R. 2006.  The environment and male fertility:  recent research on emerging chemicals and
semen quality. Semin. Reprod. Med. 24(3): 156-167.

Hauser R. 2008. Urinary phthalate metabolites and semen quality: a review of a potential marker
of susceptibility.  Int. J. Androl. 31:112-117.

De Rosa M, Zarrilli S, Paesano L, Carbone U, Boggia B, Petretta M, Maisto A, Cimmino F, Puca
G, Colao A,  and Lombard! G. 2003. Traffic pollutants affect fertility in men. Hum. Reprod.
18(5):1055-1061.

Guven A, Kayikci A, Cam K, Arbak P, Balbay O,  and Cam M. 2008. Alterations in semen
                                           54

-------
parameters in toll collectors working at motorways: does diesel exposure induce detrimental
effects on semen? Andrologia 40:346-351.

Hammoud A, Carrell DT, Gibson M, Sanderson M, Parker-Jones K, and Peterson CM. 2009.
Decreased sperm motility is associated with air pollution in Salt Lake City. Fertil. Steril.
[Epub ahead of print]: 19217100.

Hsu PC, Chen I-Y, Pan CH, Wu KY, Pan MH, Chen JR, Chen CJ, Chien G-C, Hsu CH, Liu CS,
and Wu MT. 2006. Sperm DNA damage correlates with polycyclic aromatic hydrocarbons
biomarker in coke-oven workers. Int. Arch. Occup. Environ. Health 79(5):349-356.

Rubes J, Rybar R, Prinosilova P, Vesnick Z, Chvatalova I, Solansky I, and Sram RJ. 2009.
Genetic polymorphisms influence the susceptibility of men to sperm DNA damage
associated with exposure to air pollution. Mutat. Res. Oct. 2.

Sokol RZ, Kraft P, Fowler IM, Mamet R, Kim E, and Berhane KT. 2006. Exposure to
environmental  ozone alters semen quality. Environ. Health Perspect 114(3):360-365.

Xia Y, Han Y,  Zhu P, Gu A, Wang L, Lu C, Fu G, Song L, and Wang X. 2009. Relation between
urinary metabolites of polycyclic aromatic hydrocarbons and human semen quality. Environ.
Sci. Technol. 43(12):4567-4573.
                                          55

-------
                                  Dr. Stephen Thorn

Modifications in the second draft are well done and improve the Integrated Science Assessment
(ISA) for carbon monoxide. It was a good idea to include sections that integrate health effects
risks, but there seem to be some internal contradictions in reviews of pulmonary injury outlined
in chapters 2 (starting on page 2-12) and 5 (starting on page 5-144). The ISA may be open to
criticism because conclusions pertinent to short term and long-term CO exposures differ, but the
discussions outline similar limitations in the data. The statements below are mostly excerpts
taken directly from the ISA, but they were put in a  different order than in the actual document.

Morbidity assessments for short and long term CO  exposure:

Animal toxicological studies provide evidence that short-term exposure to CO (50-100 ppm) can
cause oxidative injury and inflammation and alter pulmonary vascular remodeling. Controlled
human exposure studies have not extensively examined the effect of short-term exposure to CO
on respiratory morbidity. Positive associations between short-term exposure to CO and
respiratory-related outcomes include effects on pulmonary function, respiratory symptoms,
medication use, hospital admissions, and ED visits. The problem is that the majority of this
literature does not report results of extended analyses to examine the potential influence of model
selection, effect modifiers, or confounders on the association between CO and respiratory
morbidity. In particular, the lack of co-pollutant models prevents assessment of which effects are
due to CO versus other combustion-related pollutants. Yet, the ISA conclusion is that evidence is
suggestive of a causal relationship between short-term exposure to relevant CO concentrations
and respiratory morbidity.

The ISA outlines limitations in studies that have  examined the association between long-term
exposure to CO and respiratory morbidity including the lack of replication and absence of
validation studies to evaluate some  of the epidemiological statistical methodologies, whether
health effects observed can be explained by the known biological mechanisms and an absence of
co-pollutant analyses to disentangle the respiratory effects from CO versus other combustion-
related pollutants. The conclusion was that  the evidence is inadequate to conclude that a causal
relationship exists between long-term exposure to relevant CO concentrations and respiratory
morbidity.

Mortality assessments for short and long term CO exposure:

Epidemiological evidence was reviewed from multi- and single-city studies which suggest that
there is an association between short-term exposure to CO and mortality. The limitations in the
data were highlighted along with the observation that CO risk estimates were attenuated in co-
pollutant models. Despite the uncertainty as to whether CO was acting alone or as an indicator of
effects related to other combustion-related pollutants the ISA concluded that evidence suggests
there is a causal relationship between  short-term exposure to relevant CO concentrations and
mortality from respiratory disorders (page 5-158).
                                           56

-------
With regard to pulmonary-related mortality from long-term CO exposure, the ISA outlines the
consistent null and negative associations observed across epidemiologic studies which included
cohort populations encompassing potentially susceptible subpopulations. The discussion includes
an assessment that there is a lack of evidence for respiratory and cardiovascular morbidity
outcomes following long-term exposure to CO (Note that page 5-56 discusses long-term
cardiovascular effects observed in epidemiological studies. A conclusion is offered that there is a
direct effect of short term exposure and cardiovascular disease morbidity - see page 5-67, but no
summary statements are made regarding long term exposures). These assessments, along with an
absence of specific mechanisms to explain the progression from morbidity to mortality, are used
to conclude that it is unlikely that there is a causal relationship between long-term exposure to
CO and mortality (page 5-166).

To conclude, my impression is that similar limitations exist in the data for  pulmonary effects
from short term and long term CO exposure. Despite this similarity, short term effects on
morbidity and mortality are given a stronger summary assessment of risk (evidence suggests
there is a causal relationship) whereas long-term CO exposure is said to be unlikely to be
causally linked to respiratory morbidity and mortality.
                                            57

-------