-> "i',0
it'll
SJ
REPORT
from the
WORKSHOP ON THE APPLICATION OF
2,3,7,8-TCDD TOXICITY EQUIVALENCY FACTORS
TO FISH AND WILDLIFE
L Chicago Hilton & Towers
K^ Chicago, Illinois
\A
^"' January 20-22, 1998
Submitted to:
Risk Assessment Forum
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
Submitted by:
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02173-3134
March 31, 1998
-------
NOTE
This report was prepared by Eastern Research Group, Inc.: an EPA contractor, as a
general record of discussion during the workshop. As requested by EPA, this report
captures the main points of scheduled presentations, highlights from the group discussion,
and a summary of comments offered by observers attending the workshop; the report is
not a complete record of all details discussed, nor does it embdlish, interpret, or enlarge
upon matters that were incomplete or unclear. This report will be used by EPA as a basis
for additional study and work on the application of toxicity equivalency factors (TEFs) in
ecological risk assessments. Except as specifically noted, none of the statements in this
report represent analyses or positions of EPA.
-------
TABLE OF CONTENTS
Preface i
I. Introduction 1
II. Opening Presentations 2
Synopsis of the WHO Stockholm Meeting 3
Martin van den Berg, Chair of the WHO Working
Group on TEFs
Overview of the Retrospective Case Study 9
Donald Tillitt, EPA/DOI Planning Group
Overview of the Prospective Case Study 12
Steven Bradbury, EPA/DOI Planning Group
Workshop Structure/Summary of Premeeting Comments 18
Charles Menzie, Workshop Chair
Observer Comments 25
III. Workshop Proceedings 26
Review of the Total Maximum
Daily Load (TMDL) Model 26
Plenary Session: Discussion of the
Prospective Case Study 30
Plenary Session: Discussion of the
Retrospective Case Study 42
IV. Conclusions and Recommendations 58
Appendices
A. Workshop Participants A-l
B. Agenda B-l
C. Premeeting Comments C-l
D. Detailed Summaries of Expertise Group Discussions D-l
E. Detailed Summaries of Case Study Discussions E-l
F. Written Comments from Observers F-l
-------
PREFACE
This workshop was developed by a joint planning group from the U.S. Environmental
Protection Agency (EPA) and the U.S. Department of Inferior under the aegis of EPA's
Risk Assessment Forum. One role that the Risk Assessment Forum plays within EPA is to
promote consensus on risk assessment issues and to ensure that this consensus is
incorporated into appropriate Agency risk assessment guidance. In the past, the Forum
has issued guidance on the use of toxicity equivalency factors (TEFs) for assessing the
human health risks associated with exposures to complex mixtures of chlorinated dibenzo-
p-dioxins and dibenzofurans (EPA/625/3-87/012 and EPA/625/3-89/016). This workshop
was convened to examine the applicability of recently developed World Health
Organization TEFs for assessing risks to fish and wildlife from polychlorinated dioxins,
furans, and biphenyls.
-------
I. INTRODUCTION
Many individual members of the* family of chemicals known as polyhalogenated
aromatic hydrocarbons have been shown to produce toxic effects that are similar to those
associated with exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). Among the
classes of environmental contaminants falling into this general category are
polychlorinated biphenyis (PCBs), dibenzofurans (PCDFs), and dibenzo-p-dioxins
(PCDDs), all of which are believed to exert their toxic effects at least in part as a result of
their binding to the aryl hydrocarbon receptor (AhR).
Based both on their mechanistic similarity to TCDD and on the fact that these
chemicals often exist as complex mixtures in the environment, efforts have been made to
derive toxitity equivalency factors (TEFs) that can be used to express the toxicity of
individual PCB, PCDF, and PCDD congeners relative to the toxicity of TCDD. In two
previous workshops, convened by the World Health Organization (WHO) in August
1996 and June 1997, scientific experts reviewed the available relative potency data and
developed consensus TEF values for use in risk assessments involving dioxin-like
compounds. In addition to updating the existing mammalian TEFs, the WHO group
developed consensus TEFs for birds and fish.
To examine issues associated with the application of TEFs and the related toxicity
equivalents (TEQs) to ecological risk assessments, Eastern Research Group, Inc. (ERG), in
consultation with the U.S. Environmental Protection Agency (EPA) and the U.S.
Department of the Interior (DOI), assembled a group of experts to consider two
hypothetical case studies: a prospective case study involving a risk assessment for a
hypothetical point source requiring a water quality permit and a retrospective case study
focusing on a hypothetical freshwater ecosystem in which reproductive effects have been
observed and a remediation effort is being considered.
To examine issues associated with the application of TEFs and the related toxicity
equivalents (TEQs) to ecological risk assessments, Eastern Research Group, Inc. (ERG), in
consultation with the U.S. Environmental Protection Agency (EPA) and the U.S.
-------
Department of Interior {DOI), assembled a group of experts to consider two hypothetical
case studies: a prospective case study involving a risk assessment for a point source
requiring a water quality permit and a retrospective case study focusing on a freshwater
ecosystem in which reproductive effects have been observed .ind a remediation effort is
being considered.
II. OPENING PRESENTATIONS
To begin the workshop, Dr. Menzie introduced Ms. Christine Boivin, who welcomed
workshop participants on behalf of EPA's Risk Assessment Forum, and Mr. John
Blankenship, who extended a welcome on behalf of the DOI Fish and Wildlife Service.
Following these introductions, Dr. Menzie provided an overview of the overarching goal
of the workshop, which he described as exploring the extent to which a TEF/TEQ
approach can be used in risk assessments that have progressed beyond the screening
stage. As such, the focus of the workshop would be on the application and use of this
particular tool rather than on the broader range of issues associated with the performance
of ecological risk assessments. During the course of discussions, the group would attempt
to identify, document, and compare the uncertainties associated with the derivation of
individual TEF values—including both the uncertainties related to statistical variability
and those related to a lack of knowledge—and to assess the impact of these uncertainties
on ecological risk assessments.
Noting that risk assessment almost by definition occupies a position at the interface
between science and policy, Dr. Menzie indicated that it would be most useful for
discussions to remain as focused as possible on the more technical implications of gaps in
the TEF knowledge base. Thus, he anticipated that discussions over the next few days
would center on issues such as the relative contribution of TElF-related uncertainties to
the overall uncertainty of an ecological risk assessment, additional data requirements and
analytical support that might be needed to implement a TEF approach as opposed to
other approaches that might be considered, and the ability and/or need to support a TEF
approach with other lines of evidence. To consider these and related issues in a real-
-------
world context, workshop participants would be asked to apply the TEF/TEQ methodology
to the two case studies developed by the EPA/DOI Planning Group. For each of these
cases, the goal would be to see how application of the TEF/TEQ methodology might
impact the uncertainties associated with the exposure assessment, the effects assessment,
and the overall characterization of risk.
Following Dr. Menzie's opening remarks, the experts heard a series of formal
presentations designed to establish a common frame of reference for subsequent
discussions. Brief summaries of these presentations are provided below.
Synopsis of the WHO Stockholm Meeting
Dr. Martin van den Berg, Chair of the WHO Working Group on TEFs
Dr. van den Berg began by noting that his presentation would provide an overview of
the issues addressed and the decisions agreed to at the WHO-sponsored meeting on the
derivation of TEFs for dioxin-like compounds in humans and wildlife, which was held in
Stockholm, Sweden, in June of 1997. In contrast with earlier TEF meetings, which had
addressed only mammalian and human TEFs, the Stockholm meeting also undertook an
evaluation of TEFs for birds, fish, and wild mammals. Participants included
approximately two dozen experts in wildlife toxicology and/or in the laboratory
determination of TEFs, including several of the experts and Planning Group members
who are also participating in this workshop. The Stockholm meeting was divided into
two sessions—one dealing with human and mammalian TEFs derived from laboratory
experiments, and the other dealing with TEFs for wild mammals, fish, and birds. The
human/mammalian session was chaired by Dr. Linda Birnbaum, who is a member of the
EPA/DOI Planning Group, and the wildlife session was chaired by Dr. Richard Peterson,
who is one of the experts at this meeting. Rapporteurs were Drs. Mark Feeley and Sean
Kennedy, who is also an experts at this meeting. Dr. van den Berg served as organizer
and overall Chair of the Stockholm meeting.
-------
Prior to the Stockholm meeting, criteria for including a compound in the WHO TEF
scheme had already been established. To be included in the TEF scheme, a compound
must:
• be structurally related to PCDDs and PCDFs;
« bind to the Ah receptor;
• elicit dioxin-specific biochemical and toxic responses; and
• be persistent and accumulate in the food chain.
In its deliberations, the WHO group discriminated between TEFs and relative effect
potencies, or REPs. As defined by WHO, a TEF is an order-of-magnitude estimate of the
toxicity of a compound relative to the toxicity of TCDD that is derived using careful
scientific judgment after considering all available data. An REP, in contrast, is derived
from the results of a single in vivo or in vitro study, which ma)' be either a biochemical or a
toxicological study.
In preparation for the Stockholm meeting, the Karolinste. Institute assembled a
database containing the results of thousands of published studies comparing the
biochemical or toxicologic profiles of individual congeners wiiih a reference compound
(either TCDD or PCB 126). When PCB 126 was used as th<: reference compound, a REP
of 0.1 was assumed. To be included in the database, a published study had to meet the
following three criteria:
• At least one PCDD, PCDF, or PCB congener and ;i reference compound must
be included in the study.
• The reference compound and the congener(s) mus: be included in the same
experiment or studied with the same experimental design and by the same
authors in separate experiments.
• The relevant endpoint should be affected by the congener as well as the
reference compound.
-------
Regarding the determination of relative potency values for inclusion in the Karolinska
database, Dr. van den Berg indicated that there were several methods used. If a relative
potency value was reported in a published study, that REP was included in the database
without modification. If no REP was reported, one could be derived by any of the
following methods:
• calculated by comparing dose-response curves or by using linear interpolation
of log doses, comparing the same effect level;
• determined from the ratio of reported EE^0, LD50, or ECSO values; or
• calculated from tumor promotion indices, 1^ values for Ah receptor binding, or
directly estimated from graphs.
The Karolinska database is now part of the public domain and can be accessed by anyone
who applies to use it at the WHO European Center of Environmental Health.
Based on the wide range of REPs reported in the literature, workgroups at the
Stockholm meeting proposed human, wild mammal, bird, and fish TEFs for each
individual congener. These proposed values were then the subject of extensive discussion
during a plenary session, and on the last day of the meeting consensus values were
derived for each compound.
Turning specifically to the derivation of the human and mammalian TEFs , Dr, van
den Berg noted that meeting participants decided that there was no scientific reason to
assign TEFs for wild mammals that would differ from those derived for humans and
laboratory mammals. He then outlined the criteria used to weight different types of
experimental data. In evaluating toxicity data, meeting participants agreed that in vivo
data should be given precedence over in vitro data, which in turn should be given
precedence over data from quantitative structure-activity relationship (QSAR) studies.
When more than one in vivo study was available, those involving chronic exposures were
given the highest priority, and progressively lower priority was given to those involving
subchronic, subacute, and acute exposure scenarios. Among studies using Ah receptor
endpoints, toxicity studies were given greater weight than biochemical studies.
-------
Because mammalian TEFs had previously been assigned by WHO on the basis of
work done by Ahlborg et al. in 1994, participants at the Stockholm meeting had to decide
under what conditions they would incorporate an existing TEF into their scheme. -They
agreed that if the available information was insufficient to warrant a change, the existing
TEF value for PCDDs, PCDFs, and PCBs would be adopted. The major changes to
existing mammalian TEFs agreed to at the Stockholm meeting are summarized in Figure
1. Notably, meeting participants agreed that the di-ortho PCBs,
REVISED MAMMALIAN TEFs
Congener
1,2,3,7,8-
PeCDD
OCDD
OCDF
PCB77
PCB81
PCB 170
PCB 180
Old TEF
0.5
0.001
0.001
0.0005
—
0.0001
0.0001
New TEF
1
0.0001
0.0001
0.0001
0.0001
-•
—
Explanation of Change
CYP1A1/A2, tumor promotion
misinterpretation of earlier data;
versus tissue concentration
similarity to OCDD
EROD induction
similarity to PCB 77
in vivo data (CYI'lAl, repro) do
support in vitro observations
in vivo data (CYI'lAl, repro) do
support in vitro observations
exposure
not
not
Figure 1.
which were assigned TEF values in the earlier WHO effort, should no longer be included
in the TEF scheme. This decision was based on the fact that in vivo data, which includes
both enzyme induction and reproduction studies, do not support the in vitro observations
upon which the initial TEF values were based.
In evaluating the data for fish and birds, the WHO groups used a four-tier approach.
In decreasing priority, the tiers were:
• Tier 1: overt toxicity observed in developing embryos (endpoint = LQ0);
-------
• Tier 2: biochemical effects observed in developing embryos (endpoint =
CYP1A);
• Tier 3: biochemical effects observed in in vitro systems (endpoint = CYP1A);
and
• Tier 4: estimates from QSAR studies.
To simplify matters for risk assessment and management purposes, participants at the
WHO meeting attempted to harmonize the TEFs across the different taxonomic
categories. This was not possible, however, because of clear taxonomic differences in the
effects of various congeners. As an example of these differences, Dr. van den Berg
mentioned the responses of fish and mammals to mono-ortho PCBs.
Another aspect of the harmonization effort involved a decision about whether to
report the consensus TEFs as distinct individual values or to round them as had been
done previously. For conformity with the existing TEF values, some of which were
adopted by the WHO, new TEFs were rounded to a value of either 1 or 5. In this
rounding procedure, Dr. van den Berg said that a conservative approach was used to
provide optimal protection of fish and wildlife.
The consensus TEFs for dioxins, furans, non-ortho PCBs, and mono-ortho PCBs are
listed in Figure 2. In general, fish and birds tend to be less sensitive to hexachloro- and
heptachlorodioxins than are mammals, but there were not enough data to determine the
relative sensitivity of either fish or birds to octachlorodioxins. The most notable
taxonomic distinction for the dibenzofurans is the generally greater sensitivity of birds
than either fish or mammals to TCDF and the two pentachlorodifurans. Among the
planar PCBs, birds tended to be more sensitive than fish, particularly to PCBs 81 and
126. However, PCB 169 was less toxic to fish and birds than to mammals. For the
mono-ortho PCBs, the group felt that it was not possible to establish TEFs for fish; to
accommodate the fact that some regulatory agencies might require some number to be
used, the group decided to assign an upper limit value to the TEFs for fish. In most cases,
these compounds were also determined to be slightly less toxic to birds than to mammals.
-------
WHO CONSENSUS TEFs FOR MAMMALS, FISH, AND BIRDS
2,3,7,8-TCDD
1,2,3,7,8-PeCDD
1,2,3,4,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDD
OCDD
2,3,7,8-TCDF
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,4,7,8-HxCDF
U,3,6,7,8-HxCDF
1,2,3,7,8,9-HxCDF
2,3,4,6,7,8-HxCDF
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
OCDF
3,4,4',5-TCB(81)
3,3',4,4'-TCB (77)
3,3',4,4',5-PeCB (126)
3,3',4,4',S,S'-HxCB (169)
2,3,3',4,4'-PeCB (105)
2,3,4,4',5-PeCB(I14)
2,3',4,4',5-PeCB (1 18)
2',3,4,4',5-PeCB (123)
2,3,3',4,4',5-HxCB (156)
2,3,3',4,4',5'-HxCB (157) .
2,3',4,4',5,5t-HxCB (167)
2,3,3',4,4t,5,5'-HpCB (189)
HUMANS/
MAMMALS
1
1
0.1
0.1
0.1
0.01
0.0001
O.I
0.05
0.5
O.I
0.1
0.1
0.1
0.0 1
- o.o i
0.0001
0.0001
0.0001
0.1
0.01
0.0001
0.0005
0.0001
0.0001
0.0005
0.0005
0.00001
0.0001
FISH
1
1
0.5
0.01
0.01
0.001
-
0.05
0.05
0.5
0.1
O.I
0.1
0.1
0.01
0.01
0.0001
0.0005
0.0001
0.005
0.00005
0.000005
O.000005
<0.000005
O.G00005
O.000005
<0.000005
O.COOOOS
O.C00005
BIRDS
1
1
0.05
0.01
0.1
<0.001
*
1
0.1
1
0.1
0.1
0.1
0.1
0.01
0.01
0.0001
0.1
0.05
0.1
0.001
0.0001
0.0001
0.00001
0.00001
0.0001
0.0001
0.00001
0.00001
Figure 2.
-------
Overview of the Retrospective Case Study
Dr. Donald Tillitt, EPA/DOI Planning Group
Dr. Tillitt began his presentation by thanking the experts for the excellent job they
did in the premeeting comments they had submitted prior to the workshop. The goal of
the workshop exercises, he said, was to apply the TEF methodology to a couple of
hypothetical cases that are broadly representative of situations in which the method
might be applied, and in so doing to gain a more complete understanding of the strengths
and weaknesses of the approach.
Dr. Tillitt acknowledged, as some of the experts had pointed out in their premeeting
comments, that the retrospective case study was not a true risk assessment, in that it did
not address the full range of stressors on the system of interest. This limited focus was
intentional, however, as the Planning Group had tried to confine its description of the
case only to those elements that might be relevant to use of the TEF methodology. For
the same reason, the Planning Group had provided a detailed description of mechanisms
involved in the transfer of contaminants up the food chain. By establishing this type of
information at the outset, the Planning Group hoped to steer participants away from
discussions about what the correct values might be so that they could focus more directly
on issues associated with application of the TEF methodology.
The site for the retrospective case study was Oneofakind Lake, a
mesotrophic/oligotrophic freshwater system located in the northern United States (Figure
3). There are no industrial sources of contamination around the lake. At one time, there
were some eutrophication problems in the lake, but those are now largely resolved. The
source of dioxin-like contamination was a spill that occurred in the Yuckymuck River and
subsequently moved into Oneofakind Lake. Currently, sediments and biota are known to
be contaminated with PCBs and furans from the spill, and temporal sampling of the
sediments has suggested a first-order loss of these compounds which is believed to be
occurring primarily through sediment burial. Dioxin and furan loading to the lake is
believed to occur mainly via atmospheric inputs. Previous logging activity around the
lake included the use of DDT for insect control, but no logging has occurred for 30 years.
-------
Map of One of a Kind Lake
Figure 3.
Components of the aquatic ecosystem include lake trout, Atlantic salmon, largemouth
bass, catfish, crappie, and bluegills; the forage fish are emerald and spottail shiners. The
waterbird population is normal for this type of lake; the species that may be of concern to
state agencies include herons, gulls, and terns. The three types of evidence suggesting
some sort of disruption of the ecosystem are decreased Caspian tern reproduction,
decreased lake trout recruitment, and anecdotal reports from 'trappers that the otter
population is declining. For this case study, the Planning Group selected a tissue residue
assessment approach. The target organ for dioxin-like effects is the developing embryo in
the case of birds and fish, and the developing fetus in the case of mammals.
10
-------
Figure 4 illustrates the simplified food chain model developed for this case study.
Contaminated sediments are the primary load to the system. Biota sediment
accumulation factors (BSAFs) are used to estimate the trophic transfer of contaminants
from the sediments and up through the food chain and to predict tissue concentrations in
the forage and piscivorous fish. Biomagnification factors (BMFs) are used to estimate the
transfer of contaminants from fish to piscivorous birds and mammals, and to predict
tissue and egg concentrations in the piscivorous species. Assessment endpoints for this
study are lake trout recruitment, Caspian tem reproduction, and the size of the otter
population.
Compartmcnial Model and Simplified Pathways of Chemicals in OncofakimJ Lake.
BSAF- and BMF-Relatcd Compartments are Bracketed.
Di.sso I vcd/Part icuiatc
Chemical in Sediment
Figure 4.
Dr. Tillitt concluded his presentation by noting that, in the workshop exercise,
participants were being asked to apply the TEF/TEQ methodology to determine how
contaminant levels in the species of interest compare to hypothetical no-effect thresholds
for fish eggs, bird eggs, and mink liver. In particular, he said, the Planning Group would
11
-------
be interested in the experts' thoughts about how a risk assessment based on the use of the
TEF model would compare with an assessment based either on TCDD alone or on total
PCBs.
Overview of the Prospective Case Study
Dr. Steven Bradbury, EPA/DOI Planning Group
Dr. Bradbury began by noting that both the retrospective and prospective case studies
were designed to explore whether it might be possible to move beyond the traditional use
of TEFs, which has been exclusively for screening-level assessments. In the retrospective
scenario, for example, it has already been established that an AhR agonist situation exists,
and the question is whether the TEF methodology can be used to inform a decision about
remediation. In the prospective scenario, the situation is that dioxins and furans are
going to be released into an environment that already contains some PCBs, and the
question is whether the TEF approach can be used to inform a permitting decision. In
this sense, he noted, one goal of the workshop is to determine whether the state of the
science is sufficiently advanced to support a different application of the TEF methodology
than has been used in the past.
Regarding the specifics of the prospective case study, Dr. Bradbury noted that the
setting for this case is a lake in the northwestern United States (Figure 5). A new paper
mill has been proposed, and the mill is likely to discharge dioxins and furans into the
system. The engineers associated with the plant may have some flexibility in
manipulating the mix of congeners that will be released, but they need to know what
targets they should be aiming to meet. There are already PCBs in the system, due to
atmospheric deposition and other background sources.
12
-------
RIVER
Map of Roundtail Lake
Figure 5.
13
-------
In issuing a permit for the new paper mill, the state has deckled to use a total maximum
daily load (TMDL) approach. Accordingly, the regulators want to determine the total
load of AhR agonists the system can tolerate and still maintain the productivity of fish,
birds, and mammals in the ecosystem. Based on the current loading of the system from
background sources, they will then be able to decide how much the new plant will be
allowed to contribute and how much of the maximum load to set aside both to provide a
margin of safety and to accommodate future demands on the system.
Among the aquatic species present in the ecosystem are s;dmon, lake trout, and bull
trout. The bull trout is of particular concern to the risk managers, since it has recently
become a listed species. A variety of piscivorous birds use this system for foraging, relying
on Lake Roundtail for roughly half of their diet and on other lakes and rivers in the area
for the remainder. River otter and mink are found in the system, but there is some
question as to the home ranges of these populations.
Possible risk assessment endpoints for the prospective scenario include the
productivity of birds, fish, and mammals, and the assessment could focus on the most
representative, the most highly exposed, or the most sensitive species. Although
population-level effects are clearly of concern, the bull trout's status as an endangered
species also introduces a need for at least some attention to individual-level effects. As in
the retrospective case study, the Planning Group provided hypothetical standards for
protection of the species of concern, in this case the bull trou"., bald eagle, and river otter.
As Figure 6 illustrates, the conceptual model for the prospective case study is similar
to that used in the retrospective case, except that it relies on dther freely dissolved or
total concentrations in the water as a predictor of residues in the organisms and therefore
of expected effects. This approach is necessary because of the prospective nature of the
assessment and the fact that loading of the system is the variable for which the permit is
to be written.
14
-------
Demographic | Population
Sooal Impacts • "ode!
Reproduction
Age-ip Uotul
Pr«9 Vulner f „»„„,
Disease Resist
Chemical Transport
Fale Models
Risks
Hazards
Exposures
. Conceptual Model for Risk, A«ws>nwnis and Criu-rij IVvclnnincni Involvini;
Determination of Safe Lcudmp of Hiiuccuniulaim- Ch^nnciib to Aqujlif Sy
Figure 6.
Potential routes of exposure to the contaminants of concern are illustrated in
Figure 7. As in the retrospective case study, movement of these chemicals through the
various trophic levels of the ecosystem will determine the doses received by the
organisms of concern. Thus, also as in the previous case, bioaccumulation and
biomagnification factors will have to be used to work through the various exposure
scenarios.
15
-------
Inputs
Paihways ofTCDD Accumulation in Rtwiitui! Lite
Figure 7.
To conclude his presentation, Dr. Bradbury presented a methodology the Planning
Group had devised to address the various issues likely to arise in a prospective risk
assessment tied to a TMDL model (Figure 8). The first step, he suggested, is to relate the
total concentration of dioxin-like chemicals in the water to the. concentrations of
individual congeners, keeping track of both their TEFs and their bioaccumulation
potential relative to TCDD. By using TCDD to standardize both the effect and exposure
metrics, it should be possible to determine the maximum load an individual congener
could contribute to the system and not exceed the water quality threshold. Repeating this
process for each congener and for each of the species-specific threshold values will
generate a matrix of values that are all normalized to TCDD. When the regulator decides
on the percentage of the maximum load that will be allocated to the plant, the same
fraction can be applied to all elements of the matrix. At the same time, the discharger can
use the matrix both to see which congener is driving the assessment and to determine
whether particular combinations of congeners will or will not
-------
The Basic Relationship:
y,* s [
-------
In response to a question from one of the experts (deFur), Dr. Bradbury indicated that
alternatives to using the TEF methodology are tracking TCDD alone and basing the
permit on that determination, or issuing separate permits for each of the individual
congeners. If workshop participants had other ideas about how to approach the problem,
however, Dr. Bradbury encouraged them to explore these approaches and present them to
the Planning Group.
A member of the Planning Group (Henningsen) questioned the case study's emphasis
on daily loading, when the toxicity of these chemicals is usually more chronic and the
sensitivity of the target organisms varies over different life stages. Dr. Bradbury noted
that the TMDL model has a regulatory underpinning, and indicated that it would be just
as useful for the group to think about total maximum load over some other time frame for
risk assessment purposes.
Workshop Structure/Summary of Premeeting Comments
Dr. Charles Menzie, Workshop Chair
After the two case studies had been presented, Dr. Menzic reviewed the proposed
agenda for the workshop (Appendix B). He noted that the workshop was designed to
follow an iterative process in which small work group meetings would alternate with
plenary sessions at which the group as a whole would have an opportunity to discuss the
various approaches taken and lessons learned in the smaller work groups. To begin this
process, workshop participants had been assigned to one of three expertise groups:
• Toxic Equivalency Factors, chaired by Dr. Richard Peterson;
• Fate and Transport, chaired by Dr. William Adams; or
• Risk Assessment and Population Modeling, chaired by Dr. Menzie.
The purpose of these groups, Dr. Menzie said, would be for individuals with specific
expertise in each of these areas to come to a common understanding of what the issues
are and how they might be addressed in the context of the two hypothetical case studies.
Once this was done, members of the expertise groups would fan out among the three
work groups in which the case studies themselves were to be reviewed. Thus, each work
18
-------
group would contain some individuals from all three expertise groups. In this sense, the
work group portion of the workshop could be thought of as a replication effort to see how
three more or less similar groups might address the issues posed by each of the case
studies. Then, in plenary sessions, the efforts of each work group would be discussed by
the group as a whole to identify areas of agreement where they exist and to illuminate the
reasons for any differences of opinion in areas where agreement could not be reached.
Having provided this overview of the workshop structure, Dr. Menzie noted that in
its charge to the experts, the Planning Group had emphasized that the primary objective
of the workshop was to identify, document, and compare the uncertainties associated
with the use of the TEF/TEQ approach and to consider the impact of these uncertainties
on ecological risk assessments. Toward this end, the Planning Group had posed a series
of questions and issues to focus the experts' deliberations. Prior to the workshop, each of
the experts submitted written comments outlining their individual responses to these
questions (Appendix C). To provide a sense of the range of views experts had coming
into the workshop, Dr. Menzie offered a general summary of the commonalities and
differences he had noticed in his own review of the premeeting comments. His
observations related to selected charge questions are summarized in the paragraphs that
follow.
• Charge Question 1-1: The WHO consensus TEF values are reported as point
estimates and generally rounded off to the nearest order of magnitude. For the
risk assessment case studies, additional background information used in the
derivation of the TEF values is provided. Does this additional information
enhance the means of evaluating uncertainties in the assessments? If so, how?
If not, why?
In general, Dr. Menzie said, most experts agreed that the additional information was
an enhancement. A number of experts indicated that the WHO tier system offers a
useful framework for identifying at least the sources of uncertainty. Some felt that
additional background regarding the derivation of specific TEF values would also be
helpful, in that it would allow uncertainties to be carried along through the risk
assessment in a more quantitative way. One person thought that this information was
particularly important for the compounds that were driving a particular case study, while
19
-------
another suggested that it would be very .useful for someone to take on the task of
developing a single document that addresses the uncertainties associated with the
derivation of each of the consensus TEFs.
With respect to the rounding procedure used by WHO, Dr. Menzie noted that
various opinions were expressed, but most experts agreed that rounding is probably not
an important contributor to the overall uncertainty in the as sessment. The general
feeling seemed to be that the uncertainty associated with rounding would be less than half
of an order of magnitude, and at least one expert noted that this question could be
readily addressed by performing a model sensitivity analysis.
Finally, Dr. Menzie noted, various commenters had offered specific cautions related to
use of the consensus TEF values. One expressed the view that it is not possible to •
quantitatively evaluate the available data and assign valid, comparable uncertainty
rankings, and that qualitative assessment may be possible bui: may also be misleading.
Another suggested that probabilistic methods could be used to examine uncertainties and
limit the illusion of certainty associated with a point estimate.
• Charge Question 1-2: Some TEFs were determined from several studies,
endpoints, and exposure routes, while other TEFs were based on a single study
and endpoint. Given the range of knowledge associated with specific
compounds, should all TEFs be considered to have similar uncertainties?
Why? Or why not?
In reviewing the individual responses to this question, Dr. Menzie noted that the
overwhelming sense of the group was that uncertainties associated with the TEFs should
not be considered similar, and that the level of uncertainty is related to the weight of
evidence used to derive each of the individual TEF values. Several people noted that
uncertainties tended to be largest for the least potent and most easily metabolized
compounds, which are also the compounds least likely to drive a risk assessment. One
expert wondered whether it might be possible to develop a sliding scale to capture the
uncertainty associated with the individual TEF values. Others raised the possibility that
uncertainty in the TEFs could be addressed by adopting an uncertainty factor similar to
those employed to deal with other types of uncertainty in the risk assessment process.
20
-------
Regarding the uncertainty associated with use of the TEF/TEQ approach, some
experts felt that probabilistic methods could be used to determine the impact of TEF-
related uncertainties on the overall uncertainty associated with the assessment, but others
wondered whether even this level of quantification would be possible using the available
data. One person expressed the view that uncertainties associated with individual TEFs
will not be quantifiable until there is a common experimental basis for derivation of these
values, and that attempts to partially quantify uncertainty could impart a false sense of
accuracy. Another expressed particular concern about the TEFs for birds, which were
derived mainly from in vitro assays using endpoints that are only peripherally related to
the effects of interest.
• Charge Question 1-3: The TEF values provided were based on endpoints that
ranged from in vitro biochemical responses (e.g., induction of cytocnrome P450
I Al) to in vivo early life stage mortality. To what extent can these endpoints
be extrapolated to the measures of effects that are relevant for the assessment
endpoint for each case study?
Dr. Menzie noted that in responding to this question a number of experts mentioned
that uncertainty increases as the experimental evidence strays farther from the endpoint
of interest. Many TEFs, however, are based on biochemical effects rather than toxic
injuries, and these endpoints are poorly linked to survival, growth, and reproduction. In
this regard, experts cautioned that particular care should be taken in applying TEFs
derived from in vitro data unless the laboratory endpoint has been closely correlated to a
toxic effect in a relevant species. As an example, one person commented on the
questionable relationship between ethoxyresorufm-o-deethylase (EROD) induction and
mortality in bird eggs, since in vitro enzyme induction assays do not take metabolism into
account, and since the shape of the dose-response curve for EROD induction varies from
one congener to the next. Another factor that may complicate the use of TEFs is the
paucity of information about compensatory mechanisms that may mitigate the effect of
dioxin-like compounds at the population level.
Charge Question II-1. What are the implications, both quantitatively and
conceptually, of assuming no dose-additivity or no interaction among the
components of the mixtures described in the case studies? To what extent
would the risk assessment conclusions differ if stressor response analyses were
based on total PCBs or 2,3,7,8-TCDD alone?
21
-------
Although some experts disagreed, the majority opinion was that the assumption of
non-additivity would require a procedure for evaluating each compound separately. If
such a procedure were used, however, the lack of toxicity data for many compounds
would complicate an assessment of overall risk, which would normally be done by
summing the hazard quotients for individual compounds. Some experts noted that the
assumption of additivity was more likely to result in an oven;stimation of risk than the
TEF/TEQ approach was to result in an underestimation of risk. Also, most experts felt
that assessments based on total PCBs or on TCDD alone would typically give lower
estimates of risk than would the TEQ approach. However, some noted that differences
among the three approaches would largely disappear if the results of the assessment were
to be judged against an established criterion or other benchmark value.
• Charge Question II-2. Many TEFs are based on LQ0 or EC50 values. To what
extent should TEF values derived at a median response level be used in risk
assessments where a no adverse effect level is being employed?
Responses to this question covered a broad range of opinions, most of which had to
do with the shape of dose-response curves for the endpoints of interest. A number of
experts felt that the use of median response values was acceptable, since the goal was to
determine relative rather than absolute potencies. Also, some pointed out that LQ, and
EC50 values tended to be more stable measures within the dose-response curve than either
NOAEL or LOAEL values. Other experts disagreed, however. One suggested using an
effect level that is more relevant to the protection of ecologicil endpoints, and another
suggested that it would be more appropriate to use a no adverse effect level, particularly
for screening-level assessments. A third felt that this issue was relatively unimportant,
since differences between the various metrics would probably be lost in the noise.
• Charge Question II-3. The TEF values provided were typically based on a
single or limited number of mammal, bird, or fish experiments. To what extent
can class-specific TEFs be directly extrapolated to the species identified within
each case study?
The issue of interspecies extrapolation generated a variety of opinions, Dr. Menzie
said, and most experts believe that this is a matter of substantial concern. In general, the
experts felt more comfortable applying TEFs to organisms that are closely related to the
species in which the TEF was derived, and less comfortable as the taxonomic distance
22
-------
between the reference species and the species of interest in the risk assessment increased.
In the prospective case study, for example, most people felt that it was appropriate to
apply the fish TEF to the bull trout, since the data from which the TEF was derived were
from another salmonic species; if largemouth bass had been the species of concern,
however, use of the fish TEF would have been more problematic. A similar situation
arises when TEFs based on data collected in chickens are used to predict the effects of
exposure to dioxin-like compounds on eagles. One expert suggested that if data for the
species of interest were available, those data should be used in lieu of the more generic
TEF values.
Regarding the uncertainty associated with this aspect of the TEF/TEQ approach, some
experts felt that a traditional uncertainty factor could be applied to account for
differences between the reference species and the species of concern. One person pointed
out that interspecies differences in sensitivity to TCDD are so large that they might in
fact dwarf the uncertainties associated with the TEF approach. Dr. Menzie noted that
this observation is particularly germane to the case studies, since the threshold for TCDD
toxicity is itself a variable rather than a fixed value.
• Charge Question III- 1: To what extent does the TEF approach present
challenges, introduce new uncertainties, or modify old uncertainties associated
with modeling the exposure of AhR agonists? To what extent does the
availability and quality of congener-specific physicochemical data limit the
means of employing fate and transport or food chain models?
In general, Dr. Menzie noted, experts were in agreement that the TEF methodology
poses a number of challenges for modeling, most of which are logistical problems that
have to do with ways of accounting for the differing fate and transport properties of
individual congeners and carrying these differences through the modeling effort. Some
experts felt that this problem could be minimized if the model is focused on those
compounds that are driving both the exposure and the risk.
A number of experts cautioned that uncertainties will be magnified in attempts to
model exposure over more than two levels of the ecosystem. As an example, one person
noted that uncertainties would be great in an approach that attempted to model avian
23
-------
exposure on the basis of sediment levels, since contaminants would be moving through
many different trophic levels of the system and uncertainties would be introduced at each
step along this pathway.
• Charge Question III-3: To what extent does the TEF approach require a more
rigorous analytical design in quantifying sediments, soil, and biota AhR agonist
concentrations than is apparent in other methods which aggregate stressors
(e.g., total PCBs)?
In their responses to this question, most experts agreed that the TEF methodology
requires a more rigorous analytical design than other methods, and that analytical costs
would probably be greater as a result of the need to quantify individual congeners.
Others, however, felt that this might not be the case, since congener-based analytical
methods are now routinely used by many agencies and organizations.
• Charge Question IV-1: In evaluating the case studies, are the uncertainties
associated with TEFs more problematic than other uncertainties of the risk
• assessments? Do the uncertainties associated with TEFs limit the means of
performing the assessments, or do the other areas of the effect and exposure
characterization contribute similar or greater level;; of uncertainty?
In general, experts did not feel that uncertainties associated with the TEF
methodology would be any more problematic than other types of uncertainty in the risk
assessment process. Indeed, one person suggested that the TEF-related uncertainties may
actually be less problematic, since people have already worked through them. Others,
however, felt that this question could not be answered a priori, noting that someone would
have to go through a TEF exercise and really think through the issues to make any
reasonable statement about the relative magnitude of the associated uncertainties.
• Biologically-based TEQ assays on environmental samples could be employed as
an alternative to the TEF-based approach. What would the strengths and
weaknesses of such an approach be? To what extent could these approaches be
integrated?
24
-------
Responses to this question were mixed. Several individuals pointed out the
advantages of these methods, which include their ability to focus on an integrated
response to a mix of chemicals in the environment and their lower cost in comparison
with chemical-based approaches. Others, however, focused on the limitations of these
methods: they do not account for metabolism; they can be confounded by other
compounds; and they may not identify the most important compound for control
purposes. In general, biologically-based TEQ assays were viewed primarily as a research
tool at present, with a lack of regulatory acceptance. Some experts felt that these
methods could be very useful, however, particularly as screening tools, and several
suggested that these methods could be used in concert with the TEF/TEQ approach.
Observer Comments
At the end of his presentation, Dr. Menzie opened the floor to comments from those
attending the workshop as observers. The only observer to take advantage of this
opportunity was Dr. Angelique van Birgelen, who identified herself as a lexicologist with
the National Institute for Environmental Health Sciences (NIEHS). Dr. van Birgelen
noted that while it is rewarding to see how much progress has been made in the
development and now the application of TEFs for dioxin-like compounds, it is also
important not to lose sight of other ways in which the TEF approach can be improved.
Toward this end, she suggested that there are three additional compounds or classes of
compounds that should be assigned TEF values and included in the WHO scheme:
3,3',4,4'-tetrachIoroazobenzene (TCAB); hexachlorobenzene (HCB); and several of the
polychlorinated naphthalenes (PCNs).
According to Dr. van Birgelen, all of these compounds have been shown to bind to the
Ah receptor, all have been shown to produce dioxin-like effects, and all have been shown
to accumulate or to have a long half-life in certain species. Moreover, each may account
for a substantial fraction of the total TEQ in some environmental settings.
Dr. van Birgelen provided the group with an extensive body of published data related
to these three compounds/classes of compounds, which she summarized by briefly
25
-------
describing the AhR binding properties, effect profiles, physicochemical characteristics j
and estimated annual discharge for each compound or class. Based on this information,
she urged the group to consider recommending that these compounds be included in the
TEF scheme, and offered to provide further information if that would be useful.
III. WORKSHOP PROCEEDINGS
The second day of the workshop began with concurrent meetings of the Expertise
Groups. Discussions in these groups were organized around question lists assembled by
the Planning Group to raise issues of relevance to the various expertise areas. Each group
included a notetaker from the Planning Group, whose job it was to capture the key points
of the discussion. Appendix D of this report contains a list of Expertise Group
assignments and the discussion summaries prepared by the notetakers.
Review of the TMDL Model
Before adjourning into breakout groups to discuss the prospective case study,
workshop participants heard a brief presentation by Dr. Philip Cook, of the EPA/DOI
Planning Group, who reviewed key aspects of the TMDL model and worked through a
series of calculations related to that model. Dr. Cook began by discussing some elements
of the flow chart originally presented during the opening plenary session by Dr. Steven
Bradbury (see Figure 8, above). He noted that one can set a water quality standard based
on the toxicity of TCDD, and that this standard may be based on effects observed in
birds, fish, or mammals. Such a standard is represented in the uppermost box of the flow
chart, where C represents concentration, the subscript w indicates that water is the
medium of interest, and the superscript t refers to the fact th;it the standard deals with
the total concentration of the contaminant of interest, in this; case TCDD. In the second
box, the same standard is expressed in terms of dioxin toxicity equivalents. Based on the
additivity assumption, this standard can also be expressed in a third way, as the sum of
the toxicity equivalence concentrations of individual congeners.
26
-------
To determine toxicity equivalence concentrations for individual congeners in the
system of interest, each congener's concentration in water must be adjusted to reflect both
its toxicity relative to TCDD and its bioaccumulation potential relative to that of TCDD.
This is done by taking the product of the congener-specific water concentration, the
congener-specific TEF, and the congener-specific bioaccumulation factor, divided by the
bioaccumulation factor for TCDD. When this process is completed for each congener,
the toxicity equivalence concentrations for all congeners can be added together to
determine the total toxicity equivalence concentration for the system, and this value can
be compared with the standard to determine whether the system is or is not in
compliance.
These same relationships underlie the TCDD Toxicity Equivalence Waste Load
Allocation Model selected for the prospective case study. In this model, it is assumed that
the ecosystem has a definable assimilative capacity for chemicals which, if not exceeded,
will provide the desired level of protection. To facilitate waste load allocation for complex
mixtures of AhR agonists, maximum allowable concentrations in water (MAQs) and
maximum allowable loads (MALs) to the water body are calculated on the basis of each
individual chemical's TEF, bioaccumulation factor, and fate/transport properties. Because
each chemical is modeled individually, each MAQ. is equal to the toxicity equivalence
concentration of that chemical in water.
Because of these relationships, the accuracy of the approach depends on how well the
relationships between chemical sources and organisms of interest are modeled for each
individual congener in the ecosystem. An important step in the modeling process, for
example, involves relating the concentration of a contaminant in fish tissues, which can
be measured, to a concentration of concern in water. Ideally, this conversion is achieved
by applying a bioaccumulation factor that is both congener- and organism-specific.
Similarly, fate and transport properties determine the relationship between a mass
loading of the chemical to the system and its ultimate concentration in water, and these
properties, too, are congener-specific.
27
-------
The purpose of the MAQ calculation is to determine the maximum concentration
each congener could have in the water of this system if none of the other congeners were
present, based on its toxicity profile. The MAL, in turn, relates this concentration to the
loading of the congener into the system, based on its fate and transport characteristics.
Because MAQs and MALs are normalized values, they can be manipulated to assess the
combined impact of different mixtures of congeners on the system of interest.
To illustrate the application of this methodology, Dr. Cook worked through an
example that showed how the TMDL approach would be applied to the two-chemical
mixture described in Figure 9, assuming fish to be the organisms of interest.
VARIABLES USED IN A SAMPLE TMDL CALCULATION
FOR A TWO-CHEMICAL MIXTURE
Chemical TEF BAP log Km,
X(TCDD) 1.0 10? 7
Y 0.1 106 65
Projected Load
0.1 g/day
20 g/day
Figure 9.
The two chemicals considered in this example, TCDD and a related congener Y, have
different TEFs, different bioaccumulation factors, and different lipid solubilities. In the
example, the proposed loading of dioxin is 0.1 g/day, and the proposed loading of
congener Y is 20 g/day, and the water quality standard for TCDD has been set at 0.02
pg/L
By definition, the MAQ for TCDD is equal to the standard, or 0.02 pg/L To
determine the MAQ, for congener Y, the standard must be multiplied by the
bioaccumulation factor for TCDD (107) and divided by the congener-specific TEF (0.1)
and bioaccumulation factor (10s). This calculation yields a maximum concentration of 2
pg/L for congener Y, which is, as one would expect given the lower potency of congener Y,
many times higher than the maximum concentration for TCDD. Using a system-specific
. 28
-------
mass balance model, the details of which are irrelevant to this example, the MAQs
convert to MALs of 2 g/day for TCDD and 500 g/day for congener Y.
In the final stage of the TMDL methodology, the total load represented by the two
compounds in the mixture is compared with the load allocated to the discharger under the
permit condition, which in this example is defined as 10% of the total MAL. This is done
by dividing the projected load of each chemical by both the allocation factor and its
individual MAL, and summing the resulting values for all congeners present in the
discharge. As long as this sum is equal to or less than 1, as it is in this case, the discharger
is in compliance. Importantly, this is true regardless of the precise congener composition
of the discharge; as long as the sum of their individual adjusted loads is less than or equal
to 1, the permit condition is being met.
In response to a question from one of the experts, Dr. Cook indicated that the greater
difference between the MALs than MAQs for these two chemicals has to do with
physicochemical differences that affect their individual fate and transport profiles.
Another expert asked whether water quality standards are typically based on dissolved or
i»
total concentrations of TCDD, and Dr. Cook said that there are currently no national
water quality criteria for protection of fish and wildlife from the effects of dioxin. Based
on what he has seen within EPA, however, Dr. Cook said that he would expect such
standards to focus on the total concentration of chemical in the water. A third expert
said that the example made it clear how to determine MAQs for chemicals in the case
study, but that it was not dear how the associated MALs would be derived. Dr. Cook
indicated that this had been a topic of discussion in the Fate and Transport Expertise
Group, and that people from that group would be prepared to address questions about
MAL derivation within the context of each breakout group's analysis of the case.
At the conclusion of Dr. Cook's presentation, workshop participants reported to their
respective breakout groups for discussion of the prospective case study. The three
breakout groups were chaired by Drs. Peter deFur, Janet Burns, and Charles Menzie. On
the final day of the workshop, the same groups met to discuss the retrospective case
study. Appendix E of this report contains a list of breakout group assignments and the
29
-------
detailed summaries prepared by each of the workgroup facilitators at the conclusion of
the workshop. Following each of the individual workgroup meetings, participants met in
a plenary session to discuss the results of their deliberations.
Plenary Session: Discussion of the Prospective Case Study
Group # 1. Dr. deFur noted that his group began its deliberations by addressing the
use of more general as opposed to site-specific bioaccumulation factors (BAFs) for risk
assessment purposes. The group agreed that site-specific BAFs would be a vast
improvement over the more generic BAFs proposed for use in the case study. At a
minimum, the group felt that some effort should be made to determine whether trophic
conditions in the system of interest were or were not similar to those assumed in the
derivation of the generic BAFs. If they were not, various methods could be used to
generate more site-specific values. One method that was suggested was to develop a site-
specific model that would incorporate published data more relevant to the site; another
involved the collection of field data that could be used to devdop more site-specific
values.
Regarding uncertainties associated with the use of BAFs, members of the group
identified numerous sources of variability in these values. In general, the group agreed
that BAFs are most applicable in the system where they were developed, and that their
reliability decreases as they are applied to systems that are progressively more different
from the original system in terms of their size, biological and -physical complexity, and
scope. Indeed, group members felt that the relationship between the bioaccumulative
behavior of TCDD and other congeners was likely to be more stable than the behavior of
TCDD in different systems. As a result, they concluded that it would be more useful to
improve understanding of the bioaccumulative behavior of TCDD than to improve
understanding of the relationships between BAFs for TCDD *md other dioxin-like
compounds.
Throughout their discussions, Dr. deFur's group encountered a number of issues that
highlighted differences in the European and American approaches to assessments of
30
-------
dioxin-like compounds. The most striking of these was the fact that in Europe chemical
analyses are seldom if ever done for a single congener, so it simply would not be the case
that TCDD would be measured alone. As a result, environmental concentrations of the
individual congeners are known, and it is usually possible to determine BAFs for the full
suite of dioxin-like congeners. Given the obvious importance of BAFs to the TMDL
model, the group agreed that wider adoption of the European practice would substantially
reduce the uncertainty associated with TMDL-based regulatory and management
decisions.
Turning to the question of dose-response relationships, the group discussed problems
associated with relying on TEFs that are derived at the cellular or molecular level to
predict effects at the population level. While recognizing that regulatory and
management decisions are often constrained by the legal, policy, or even cultural context
within which those decisions are made, group members felt that the level of uncertainty
associated with these types of extrapolations is large and that this aspect of the
assessment paradigm needs to be addressed. Particularly when attempting to set
regulatory limits such as MACs, information about population dynamics is a critical
component of the knowledge base. Like BAFs and other elements of the TMDL
approach, population data will be most useful if collected on a site-specific basis, focusing
on density-dependent as well as density-independent factors.
Another element of the group's discussion focused on the relationship between TEQ-
and TEF-based approaches. In general, the group felt that these approaches are
complementary, in the sense that TEQ-based bioassays might serve as a reality check for a
TEF-based analyses. If the results obtained via both methods were concordant,
confidence in the TEF-based analysis would certainly increase. Even non-concordance
might be useful in highlighting specific areas where further investigation is needed.
The group also spent a fair amount of time discussing how the uncertainties
associated with application of the TEF methodology compare to those associated with
other elements of the risk assessment process, including the uncertainty in BAFs,
uncertainty in population dynamic models, and uncertainty in environmental
31
-------
measurements. In addition, the group discussed the many places within the TMDL
model that errors were likely to. be propagated and perhaps even magnified. At the end of
this discussion, there was general agreement that no single source always generates the
greatest amount of uncertainty, and that the relative contribution of individual sources of
uncertainty varies from site to site.
At the end of his summary, Dr. deFur asked whether other group members would like
to comment on any additional issues that came up during the group's deliberations. One
member of the group noted that toward the end of the session there had been some
discussion of the need to identify the uncertainties associated with various elements of
the TMDL model, including but not limited to the uncertainties associated with the
derivation of TEFs, and to find appropriate ways of carrying these uncertainties through
the risk assessment process. Although presented as point esti mates, all of the numbers in
the case study exercise have some variance associated with them. To determine the
relative contribution of individual uncertainties, therefore, one could use a Monte Carlo
or other probabilistic method to see how each of these uncertainties affects the values
generated via the TMDL process.
In response to a question from one of the other experts, Dr. deFur elaborated on the
role that bioassay-based approaches might play within the TMDL framework. One way
that bioassays could be useful, he said, was in screening-level analyses—for example, to
see whether contaminants actually do accumulate at the predicted rate. Later in the
process, bioassays could be used to determine how rates of enzyme induction, for
example, compare with those predicted at one level of the TMDL model. In this setting,
observed values should be fairly close to predicted values, or there should at least be some
way of explaining disparities between the two approaches. Hs also noted that the group
recognized the difference between their around-the-table discussion and the
circumstances under which management decisions generally reed to be made. In this
sense, it might not always be possible for confirmatory bioassiys to be run, due to both
resource and logistic constraints. The group nevertheless felt that in some situations
bioassays could provide a useful complement to a TEF-based approach.
32
-------
Group 2^ Ms. Burris began by noting that her group spent a good portion of the
session discussing the uncertainties associated with the derivation of TEFs and the effect
of these uncertainties on their application within the prospective case study. Based on
this discussion, the group agreed that a hierarchical approach should be used to select the
TEFs applied to a particular risk assessment. If a species-specific value is available, for
example, that value should be used in lieu of the WHO consensus TEF. Also preferable
to the consensus TEF would be a value derived for a more closely related species than that
used to derive the WHO value. However, a sensitivity analysis should be performed to
determine whether uncertainty would actually be reduced by the use of species-specific
values.
Group members felt that more information about the methods used to derive
consensus TEFs would have been helpful, since it would have allowed the uncertainties to
be better understood and carried through the analysis. Their impression was that the
process used to derive consensus values was not consistent from one congener to the next,
and that this made it difficult to have even a qualitative sense of the uncertainties
introduced by using the consensus TEFs. Rounding, in particular, seemed to be a
quantifiable source of uncertainty, but information about the rounding process was too
scant to allow a more detailed consideration of this issue.
Despite its shortcomings, the group concluded that the TEF approach is more valid
than approaches using either total PCBs or TCDD alone. However, they thought that
there would still be a need for total PCB-based approaches, since some of the effects of
these compounds are not mediated by the Ah receptor.
Turning to the prospective case study, the group decided to use the consensus avian
TEF for the bald eagle, but to look at the effects of rounding and not rounding the TEF
value. In general, group members were comfortable extrapolating from the endpoint used
in deriving the TEF to the reproductive endpoint in the assessment. For the bull trout,
the group elected to use TEFs derived from rainbow trout data, and they thought that
early life stage mortality was the appropriate endpoint. For the otter, they chose to use
the WHO consensus TEF, but there was some discomfort about extrapolating from the
TEF endpoint to the assessment endpoint.
33
-------
Group members did not feel that the use of median values for deriving TEFs was a
significant source of uncertainty, since the median values tended to be more stable and
were probably more appropriate for looking at relative toxicky.
Moving on to the exposure assessment, the group felt that use of the TEF approach
for this particular fate and transport modeling exercise was really no different than the
use of any other chemical-specific model. The challenge, however, was in modeling the
many different congeners and in having the data available to complete the modeling
exercise.
Looking at the measurements of individual congeners in .sediment and fish tissue, the
group felt that the greatest uncertainties were in water measurements, due mainly to
limit-of-detection issues. From a physicochemical perspective, the group had high
confidence in the log K^, values, but the !<„. data and Henry's Law constants were
considered suspect. Biotransformation and metabolism of the individual congeners were
not as clearly understood; in some cases there was no knowledge, and in others it is
known that there are changes in the composition of congeners as they move between the
different species. PCB 126 is enriched, for example, during transfers from fish to wildlife
species, and this needs to be considered. In general, however, we have a better
understanding of the transfer within fish than we do from fish to wildlife. In order to be
able to appropriately model or understand the fate and transport of various congeners
within the food chain, we need to know more about what the .organisms are consuming,
since the composition of congeners is species-specific and will therefore vary from one
species to another.
In general, group members felt reasonably confident that they would be able to
complete a worthwhile modeling exercise if they had more information about transfers
from sediment to the sediment-water interface and about sediment transport within the
system. Without this information, however, the modeling exercise would be extremely
uncertain. Some members of the group thought that it would be a good idea to advise the
risk managers to substitute a better-characterized model for trie one proposed in the case
study, but there was a divergence of opinion on this issue.
34
-------
In terms of the analytical requirements to implement a TEF approach, group members
agreed that the TEF approach would be more costly than the total PCB or TCDD-only
approaches, since the discharger would have to analyze many different congeners. This
might turn out to be beneficial, however, since a better understanding of the toxicity
associated with specific congeners might give the discharger more flexibility in altering the
composition of the discharge.
Overall, group members agreed that the uncertainties associated with the exposure
profile and with projecting exposures in the future under these conditions were at least as
great and possibly greater than those associated with the stress response profile or the use
of TEFs. To gain a better understanding of relative uncertainties, the group
recommended a sensitivity analysis focusing on TEFs, Koc values, and biomagnification
factors. Regarding the latter, group members parenthetically noted that the same dose
metric should be used for BMFs and TEFs.
Regarding the use of biological assays, group members felt that these really were not
applicable to a prospective case study, since it is not yet dear which- chemicals will be
present in the system. However, biological assays could be used to document background
conditions in the system before the discharge occurs, particularly since it is already known
that PCBs are present.
When the group discussed errors associated with the application of a TCDD-based
water standard, two potential problems were raised: the enrichment of PCB 126 from
fish to wildlife and the observed loss of chlorinated dibenzofurans in some species of
birds.
The group concluded its discussion by talking about ways the assessment for this site
might be done better or differently. Group members agreed that it might be useful to put
together a more site-specific model, but there would be no way of knowing whether such
a model would be predictive. Other existing food chain models could be used, but these
would have to be modified to address metabolism issues. Everyone was more comfortable
using the TEF/TEQ approach than using either of the default approaches, but most
35
-------
thought that the assessment would generate a range of risk estimates that would be
perplexing to the risk manager. It was agreed, however, that 'this may be the best we can
do given the current state of the science.
Following Ms. Burris' presentation of the group's findings and recommendations, there
was considerable discussion of the role that bioassays might play in a prospective case
scenario. In response to a question about how they came to iheir decision that bioassays
would not be useful, a member of the group explained that there was some concern about
how the results of bioassays could be misleading if appropriate extraction and
fractionation steps were not included. Another member of the group mentioned studies
of Canadian paper mills in which bioassays were applied directly to the effluent, resulting
in a gross overestimation of discharge toxicity. The questioner agreed that these issues
need to be taken into account, but suggested that the wording of the group's conclusion
was overly strong. He noted that there are many different types of bioassays, and that
some would be very useful in a prospective setting. As an example, he suggested a
bioassay that is able to predict the relative potencies of various congeners for relevant
endpoints in a fish species of concern. Such a bioassay could be used to test both how
sensitive that system is to different compounds and how the sensitivity of the target
species compares with that of other organisms in the system. This information, in turn,
might be extremely useful in a prospective assessment of the impact that further loading
of the system might have on the species of concern.
In response to a question from the Chair, Ms. Burris confirmed that the group's sense
had been that uncertainties associated with the use of TEFs are no greater than those
associated with exposure or response assessments, although the group did not have
enough information to quantify these different types of uncertainty. The group also felt
that uncertainties were less manageable in the context of a prospective case study, since a
prospective scenario does not lend itself to the sorts of approaches that can be used to
reduce uncertainty in a retrospective assessment.
Group 3. Dr. Menzie indicated that the results of his group's deliberations would
presented by group members Donald Tillitt and Wayne Landis. Dr. Tillitt began by
36
-------
noting that Group 3 had begun its analysis of the case study where the other groups had
left off, in that this group had focused almost exclusively on how the various sources of
uncertainty might be addressed in a risk characterization for the prospective case study.
For purposes of this exercise, the group identified five major sources of uncertainty: the
derivation of TEFs, the derivation and use of BAFs, extrapolation of TEFs between
species, exposure modeling, and derivation of the threshold values themselves. For each
of these sources of uncertainty, the group developed specific criteria that could be used to
rank degrees of uncertainty on a scale of 1 to 4, which was chosen because of its rough
correspondence to the tier system used in the derivation of TEFs at the Stockholm
meeting.
Dr. Landis added that the group's intent in developing these criteria was to move from
"feelings" and "senses" of relative uncertainty to a more quantitative expression. While
recognizing that the ranking system is not quantitative in a statistical sense, it does
provide a way of assigning relative values to the differing degrees of qualitative
uncertainty that most people would agree exist in different interspecies extrapolations or
in different types of gaps in the congener-specific data. In addition, this approach aliows
the uncertainty rankings to be manipulated arithmetically in ways that provide additional
information about the system as a whole.
To illustrate the results of the group's deliberations, Dr. Landis showed the matrix
reproduced as Figure 10. For each cell in the matrix, the group attempted to rank the
uncertainty associated with a particular variable in either species- or congener-specific
terms. For example, they felt that the uncertainty associated with application of a TEF
derived in rainbow trout or lake trout to bull trout was considerably less than the
uncertainty associated with applying a TEF derived in chickens to bald eagles; as a result,
the group gave the TEFs for bull trout an uncertainty ranking of 1 and the TEFs for bald
. eagle an uncertainty ranking of 4. In considering BAFs, the group felt that these were less
uncertain for fish than for either birds or mammals, and rankings were assigned
accordingly. Similarly, because the exposure model was developed around fish, its
application resulted in less uncertainty if a fish rather than a bird or mammal was the
species of concern. Also, because of their migratory potential, birds and mammals are
much more likely to have exposures outside the system than are fish.
37
-------
Relative Uncertainties in the Ecological Risk Assessment Including Us of TEF Values
Ranks for uncertainty
Species/Congener
1
Bull trout
1
2
3
4
s
6
7
BaltfEagl*
1
2
3
4
S
6
7
Rlv>rOtt*r
1
2
3
4
5
6
7
TEFi
1
4
3
BAFs
Species Sens/ExtrapoU
2j 2
Exposure Mod
2
21 11
21 11
2
11
21 1
2
2
.2
1
3
3|
3
3
3
3
3
3
1
4
3
3
2
3i 3
3| 1
3
3
3
3
3
3
1
1
2
2
2
2
4
4
Threshold cui'icfiAli ^tion
2l
4
3
Cntena trc desaiBea in me ten THIS iroroaOi and these values are presnted tor Illustration only
•
£[ecm tpcclfic
Ccngener specific
9
21
19
36
16
32
Total
30
ss
48
Bull Trout
Oild Etdl*
River Otttr
Figure 10.
Once these individual rankings were completed, the group summed all of the species-
and congener-specific values to see how each contributed to overall uncertainty. From
this summation, it became clear that the species-specific uncertainty was greatest for bald
eagle, slightly less for the river otter, and much less for the bull trout. One of the
encouraging conclusions that can be drawn, therefore, is that uncertainty is relatively low
for-the species that is endangered. In addition, the group concluded that the species most
likely to drive the lower limit would be the river otter, for wh:tch uncertainty was the
greatest.
38
-------
Another way the group used this matrix was to identify the sources of greatest
uncertainty in the assessment. To a large extent, Dr. Landis said, overall uncertainty was
driven by uncertainty in the modeling. For individual species, however, it was possible to
identify specific areas in which uncertainty was due to a lack of knowledge about the
properties and effects of different congeners. In this sense, the matrix could also be used
to identify ways of reducing the uncertainty in these assessments. For both the bald eagle
and river otter, for example, additional information about species-specific TEF and BAF
values would substantially reduce the uncertainty of the assessment. In this way, Dr.
Landis suggested, use of this matrix would allow the risk assessor to answer a variety of
questions that are vitally important to stakeholders, including how the situation might be
improved. In addition, the group felt that this matrix might be a useful tool in
communicating the results of the assessment to risk managers.
One caveat that the group identified in considering possible uses of the matrix is that
the relative rankings are specific to the system under consideration. Because the rankings
reflect relative rather than absolute measures of uncertainty, different values would have
to be generated for different systems, and the results of site-specific analyses could not be
directly compared.
Following these presentations, one of the experts from a different work group
expressed some concerns about using a matrix such as this to identify the areas in which
additional research is most needed. The reason for his concern was that the matrix does
not address the relative sensitivity of the model as a whole to specific elements of the
matrix. Depending on the model, it could be more important to reduce the uncertainty in
one variable from 2 to 1 than to reduce the uncertainty in a different variable from 4 to 2.
Dr. Landis agreed with this observation, noting that it would be necessary to combine the
matrix with a more conventional sensitivity analysis to determine precisely where
additional research would have the greatest impact on overall uncertainty. However, he
thought that the matrix enables assessors and managers to better understand those
aspects of the uncertainty problem that are not typically addressed in a sensitivity
analysis. A member of the Planning Group suggested that it might be possible to combine
these two approaches by weighting different cells in the matrix to reflect the results of a
sensitivity analysis.
39
-------
At this point in the discussion, another member of Group 3 noted that the group was
unable to identify any place in the process diagram where this and other information
about relative uncertainties could be incorporated into and carried through the TMDL
process. He thought that this would be an important issue for the modelers to address,
since the ultimate value of quantifying the uncertainties depends on there being a way to
bring this information to bear on the decisionmaking process. One way to do this, he
thought, would be to go back and reframe the question that tJhe model was designed to
answer in a way that includes specific attention to the impact of various types of
uncertainty.
In response to a request from Dr. Menzie to describe the group's thoughts about use
of the TEF approach as opposed to one of the defaults, Dr. Tillitt said that there was an
agreement that the use of TEFs does not contribute disproportionately to overall
uncertainty, and that the TEF approach reveals some useful information that would not
be apparent if other approaches were used. As a result, the group felt that something
important would be lost if one of the defaults were used.
One of the experts noted that it is important to be cautious when using a semi-
quantitative method as a decisionmaking tool. The reason for his concern was that the
weighting of different variables may reflect subjective biases, and this subjectivity could
be obscured by the quasi-mathematical nature of the method. If this occurred, the
method would simply be validating a conclusion that was essentially predetermined. Dr.
Landis agreed, and noted that this is why it is important for the ranking criteria to be
established a priori, before the method is applied to specific shes. Another group member
noted that the ranking criteria themselves would certainly be open to debate, and might
even change over time, as more information becomes available. Continuing along these
same lines, another expert suggested that it would be an interesting test of the method
this group used to see how different groups given the same a priori criteria and the same
data set would rank the relative uncertainties. Finally, a member of the Planning Group
urged that, in the workgroup's more detailed report of its deliberations, members of the
group try to more clearly describe the ranking scheme they used to construct their matrix,
since these a priori criteria represented such a key element of the process.
40
-------
gummary. To conclude the plenary session, Dr. Menzie provided a brief summary of
what he thought were the major conclusions that could be drawn from the group's
consideration of the prospective case study. In general, all three workgroups felt that the
TEF approach could be applied to a prospective case scenario, but that this approach
might be more costly than the other alternatives. All three groups felt that there needed
to be a way to track uncertainties through the risk assessment process, but that
uncertainties associated with the application of TEFs are no greater than those associated
with other elements of the TMDL model, and that they may in fact be smaller. As a
result, all three groups concluded that use of the TEF-based approach is preferable to use
of the traditional TCDD-based methodology, which in comparison might underestimate
risk. There was some discussion of the usefulness of biological assays in supplementing
the TEF approach, and a divergence of opinion regarding the applicability of these
methods to a prospective case scenario. Finally, the group had discussed the need for
better ways of incorporating what we do know about different sources of uncertainty into
the TMDL model and for communicating the results of the assessment to risk managers.
At the end of Dr. Menzie's summary, one of the Planning Group members asked if
any of the groups had addressed the aspect of the TMDL approach that has to do with
issuing a permit that is based at least in part on chemicals that are not in the discharger's
wastestream. One of the experts noted that this had been addressed to some extent in the
comment that a TEF-based approach might in some cases actually turn out to be
beneficial to the discharger, since only the subset of AhR agonists would be driving the
assessment and therefore the permitting process. The questioner noted that this is a
departure from the chemical-specific approach that EPA has traditionally used in
regulating environmental contaminants, since it directs the regulator to mode of action or
ecological effect rather than to chemical identity. One of the experts suggested that if the
goal is truly environmental protection, then this is an appropriate re-focusing of the
regulator's attention. Another expert disagreed, suggesting that further ground-truthing is
needed before TEF-based approaches can reasonably be applied in a regulatory setting.
41
-------
Plenary Session: Discussion of the Retrospective Case Study
Group # 1. Dr. deFur began by noting that his group's approach to the retrospective
case study differed in two important respects from their approach to the prospective case.
First, the group attempted to be as quantitative as possible in addressing the retrospective
scenario, as opposed to the largely conceptual approach they had taken to the prospective
case. In addition, in accordance with guidance the facilitators had been given by
members of the Planning Group, the group agreed to try to make a decision about the site
described in the retrospective case study.
After reviewing the features of the site, the group first talked about what the decision
was that they were trying to make. Rather than a decision about whether to remediate or
not to remediate, the group elected to try and decide whether the data were sufficient to
support a regulatory or management decision. In particular, they agreed to focus on
whether the TEF/TEQ approach offered any advantages over approaches based on total
PCBs or on TCDD alone.
The group's quantitative analysis centered on a graph that one of the members drew
to summarize how the data from the site would look from bcth a TEQ and total PCB
perspective (Figure 11). In this figure, the left-most bars in each graph represent the
species-specific TEQs for the site, broken down to reflect the contribution of various
classes of compounds to the total TEQ. The vertical line to the right of this bar
represents the threshold range for the species of concern. In the right half of each graph,
a similar method is used to depict the site-specific values and threshold ranges for total
PCBs.
Interpretations of this graphic covered a fairly broad range. Some people felt that
conclusions drawn on the basis of the TEQ data would differ from those drawn using total
PCBs, but others felt that there would be no difference in the: bottom-line conclusions as
to whether exposures do or do not reach threshold. The group did not try to reach an
agreement on this issue, since it seemed important to note that these data could be
interpreted one way by some people and differently by others.
42
-------
Fish - Lake Trout Eggs
Z
1
3-3
RLbB
PCBs
Lake Trout Eggs
Bird - Caspian Tem
SCO
too
M50
3
Z
I
'*5
4
3
/
Mammal
. 200
ISO
100
so
Figure 11.
PCDF
43
1.0
0.5)
-------
The group noted that in two of the three species, PCBs were the main contributors to
total TEQj TCDD for the most part made a relatively small contribution to the total
TEQ and furans were similarly minor contributors, except in fish. Clearly, the
contribution of various classes was more obvious using the T5Q approach. Group
members felt that this was important, since it increased people's comfort level about the
range of conclusions that could be drawn about the site. Everyone agreed that the results
of the TEQ analysis were sufficient to support screening-level decisions. Opinions began
to diverge, however, as application of the TEQ approach moved closer to the regulatory
arena.
Group members concluded that the amount of additional information revealed by
application of the TEF approach depends on the mix of congeners present in the system.
In at least one case, moreover, the group agreed that reliance on TCDD alone would alter
the outcome of the risk analysis. In this case as in the prospective case study, group
members who were not accustomed to dealing with the U.S. regulatory system were
surprised that anyone would actually go out and measure TCDD alone, as opposed to the
full suite of dioxin-like congeners, and even more surprised that a regulatory decision
might be based on TCDD alone. Group members agreed that this approach is
scientifically unsound.
The group engaged in an extended discussion of uncertainty, and members agreed
that it is important to identify and put bounds on the various.sources of uncertainty in
the TEQ-based analysis. In particular, it is important to recojmize that some
uncertainties are quantitative, having to do with statistical variability, while others have
to do with gaps in the knowledge base. Different analytical tools should be used to
address these differing types of uncertainty and different analytical approaches are
required to carry them through the assessment.
44
-------
When it came to the actual decision the group had agreed to make, there was a
divergence of opinion about whether the TEF approach is sufficient. Some people felt
that the approach provided enough information to move forward, and others did not.
Everyone agreed that the approach provides useful information about where the key gaps
in the data are, and for that reason alone there was agreement that the approach should
not be turned down. However, some people felt that the results of the TEF approach
would have to be supplemented with more information, on population dynamics and on
the relationship between the biochemical or molecular endpoints on which the TEFs are
based and effects at the population level before the approach could be used to decide
whether to move forward into a regulatory decisionmaking mode.
Differences in opinion about the sufficiency of the TEF approach were based mainly
on the paucity of information about the uncertainties associated with individual TEF
"!^
values. Although group members uniformly felt that the underlying data was probably
very robust, some nevertheless felt that TEF values could not legitimately be used in a risk
assessment until and unless the associated uncertainties were expressed quantitatively and
carried through the analysis. In particular, group members were concerned about
uncertainties associated with the derivation of TEFs, with species differences in
responsiveness to the various congeners, and with the ability of TEF-based methods to
predict population-level effects.
At the end of their deliberations, Dr. deFur's group attempted to identify data gaps
that seemed particularly critical in the context of the retrospective case study. Research
efforts that might be useful in addressing these gaps included:
• testing of the Caspian terns themselves to develop species-specific BAP and
BMF values;
• performing ground-truthing exercises to get a better sense of the relationship
between exposure levels and responses in the tem population;
• gathering population data for the three species of concern;
• examining sediment core samples from the lake as opposed to the river to get a
better sense of the distribution of chemicals in the system as a function of both
time and space;
45
-------
• determining deposition rates and inputs from sources other than the site of the
prior spill; and
• performing ground-truthing exercises to assess the predictive capability of the
TEF/TEQ approach at sites for which there is already a good body of data.
In response to a question from Dr. Menzie, who asked whether the group had
identified any specific types of uncertainty in the TEF approach that were particularly
problematic, Dr. deFur indicated that the three major concerns of the group had to do
with differences between the species used to derive the TEF values and the species of
concern in the risk assessment, with the statistical uncertainty in the derivation of a TEF
from multiple REP values, and with the statistical uncertainty in the REP values
themselves. Another group member pointed out that the reason for this concern was that
group members were unsure whether the uncertainty in TEF/TEQ values was high enough
to impact conclusions about whether observed levels of contaminants did or did not
exceed the threshold value.
Another member of the expert group commented that the group's reticence to
recommend that the results of the TEF analysis be used as a basis for risk management
decisions seemed to include some presumptions about what those decisions might be.
Noting that there was a similar reticence in his own group, thi; expert suggested that
assessors should be sure they are not attempting to do the risk manager's job, since the
decision could just as easily be whether to spend an additional $100,000 on research as to
embark on a $ 1 billion remediation effort. If experts believe the method sufficient to
support the former decision - which most seem to - then it w;is not clear to him why it
wouldn't be sufficient to support the latter, since the validity of the method would not
have changed. The task of the assessor, he noted, is to present the facts and associated
uncertainties in a way that will inform the risk manager's decision, not to determine
which decisions should or should not be made on the basis of the available data. Dr.
deFur responded that there had been some discussion of this in the group, and that no
one wanted to go on the record as recommending remediation even for a fictitious site.
46
-------
• determining deposition rates and inputs from sources other than the site of the
prior spill; and
• performing ground-truthing exercises to assess the predictive capability of the
TEF/TEQ approach at sites for which there is already a good body of data.
In response to a question from Dr. Menzie, who asked whether the group had
identified any specific types of uncertainty in the TEF approach that were particularly
problematic, Dr. deFur indicated that the three major concerns of the group had to do
with differences between the species used to derive the TEF values and the species of
concern in the risk assessment, with the statistical uncertainty in the derivation of a TEF
from multiple REP values, and with the statistical uncertainty in the REP values
themselves. Another group member pointed out that the reason for this concern was that
group members were unsure whether the uncertainty in TEF/TEQ values was high enough
to impact conclusions about whether observed levels of contaminants did or did not
exceed the threshold value.
Another member of the expert group commented that the group's reticence to
recommend that the results of the TEF analysis be used as a basis for risk management
decisions seemed to include some presumptions about what those decisions might be.
Noting that there was a similar reticence in his own group, this expert suggested that
assessors should be sure they are not attempting to do the risk manager's job, since the
decision could just as easily be whether to spend an additional $100,000 on research as to
embark on a $1 billion remediation effort. If experts believe the method sufficient to
support the former decision—which most seem to—then it was not clear to him why it
wouldn't be sufficient to support the latter, since the validity of the method would not
have changed. The task of the assessor, he noted, is to present the facts and associated
uncertainties in a way that will inform the risk manager's decision, not to determine
which decisions should or should not be made on the basis of the available data. Dr.
deFur responded that there had been some discussion of this in the group, and that no
one wanted to go on the record as recommending remediation even for a fictitious site.
47
-------
million cleanup, industry scientists would have a strong motivation to fill some of these
gaps in the understanding of uncertainty, precisely because taey would not want to be
caught in the position of having to comply with management decisions that were based
on back-of-the-envelope risk calculations that failed to take uncertainty into account. He
went on to note that even he and the other people who were calling for better
characterization of the uncertainties like the TEF approach, because it does have the
advantage of bringing different congeners together in an integrated model. The only
question is whether the method is sufficiently well developed to support definitive,
quantitative risk management decisions. Without more precise information about the
error in these values, it is simply not possible to answer this question.
In response to this comment, one of the experts expressed the opinion that
uncertainties in the method do not mean that the method cannot or should not be used.
He noted that decisions are made every day on the basis of incomplete information; if a
decision needs to be made tomorrow, this incomplete method may represent the best that
we can do. Another expert suggested that, at least from a risk management perspective,
the question can also be framed in terms of the need to select between three different
methods that are all incomplete in some way. From this perspective, he thought that
most people would agree that despite its limitations, the TEF methodology offers
important advantages over those based on total PCBs or on TCDD alone.
Group 2. Ms. Burris noted that her group began its deliberations by discussing the
effects portion of the analysis, working through each of the species of concern to
determine which TEF they would use and what level of uncertainty was associated with
these selections.
For lake trout, the group decided to use both a TEF derived from the rainbow trout
data (0.005) and an REP for PCB 126 in lake trout (0.003). The group felt that
extrapolation from the trout data to other, non-salmonic species in the lake would
introduce uncertainty, but that the magnitude of this uncertainty is unknown because the
data needed to quantify it are not available.
48
-------
For the Caspian tern, the group chose to adopt the WHO TEF, mainly because the
information used to derive it was of better quality than the species-specific data that were
available. Based on an EROD assay of PCB 126, the TEF derived for the Caspian tern
using species-specific data was 0.03, and the WHO consensus value was 0.1. Therefore,
use of the WHO value increased the TEQ from 185 to 426.
At this point in their deliberations, the group briefly discussed whether the risk
assessor should be allowed to select a species-specific TEF from the available REPs, or
whether that decision should be left to individuals with a better understanding of the
literature. The group did not reach an agreement on this point, but they did feel that it
was important for the assessor to have the flexibility to use a species-specific value if one
was available.
For the mink, the group elected to use the WHO value. There was some discussion of
the endpoints used in the derivation of this value, but the information needed to resolve
this issue was not available.
Because of the difficulties they had in selecting TEFs for the species of interest, the
group had a general concern about the lack of transparency in the WHO consensus TEF
values. The group also felt that it would be more useful if these values were expressed as
ranges, since management decisions are frequently not based on point estimates. Ranges
would also help to quantify the uncertainty associated with a particular TEF, which would
increase overall confidence in the results of the analysis.
Looking more closely at the issue of using TEFs other than those set forth by the
WHO, the group attempted to develop a TEF selection hierarchy. In decreasing order of
preference, the hierarchy they developed was as follows:
• a TEF derived using the endpoint of interest in the species of concern;
• a TEF derived on the basis of in vivo toxicity data in the species of concern;
• a TEF derived using the endpoint of concern in a related species;
49
-------
• a TEF derived on the basis of in wwtoxicity data in a related species;
• a TEF derived from a Tier 2 REP for the species of interest; and
• the WHO consensus TEF.
The group also discussed whether uncertainty in the asses sment could be reduced by
performing a full food chain modeling exercise. They decided that such an effort would
be problematic both because of the heterogeneity in the system and a possible lack of
equilibrium: Members agreed that a full modeling exercise was probably not necessary,
but that a partial modeling exercise could be useful in developing site-specific BSAFs and
BMFs. These values, in turn, would allow the risk manager to examine the tissue level
reductions that could be expected to occur in target species under different management
scenarios. However, the model could probably not be used to predict concentrations over
time.
The group's approach to the risk characterization was similar to that followed by Dr.
deFur*s group, and they noted that the TEF methodology yielded a higher estimate of risk
that either the total PCB or TCDD-only methodologies.
A question that came up during the group's discussion of this case was how to account
for the fact that, as a migratory species, the terns might be getting some of their exposure
at another site. After some discussion, members agreed that rht assessor could use a
weight of evidence approach to evaluate the relevant scientific literature and develop an
opinion about whether and to what extent tissue concentrations in the birds should be
attributed to the site.
The group developed hazard quotients for individual organisms in each of the species
of interest. In general, these values were borderline. Use of s. TEF for common tern data
as opposed to a TEF derived from the Caspian tern data altered the hazard quotient by
less than an order of magnitude. There was some concern widiin the group about how
hazard quotients should be translated to effects at the population or community level.
Because the stated goal of the assessment was protection at the population level, the
50
-------
group felt that a separate modeling exercise would be required to better understand the
relationship between hazard quotients and the assessment endpoint. Without this
information, some members of the group were concerned about the advisability of basing
a management decision on the results of the TEF-based analysis.
Regarding issues that should be addressed in the risk characterization, one person
suggested that it would be useful to try to describe how the system might look in one,
five, and ten years if no action was taken. Some members of the group thought that PCB
concentrations would decrease over time, eventually reaching a level that is lower than
the action threshold. Others suggested that a hundred-year flood scenario should be
included in the characterization, and that there should be some discussion of the decrease
in reproduction required to produce a population effect. In view of the borderline
condition of the system, some group members also felt that attention should be focused
on the potential effect of additional inputs to the system that might occur in the future.
When a vote was taken, two members of the group voted for action and four voted for
no action. In the event that the risk manager decided to pursue a cleanup, the group
agreed that the otter would be the species of concern in setting cleanup levels. The
reason for this choice had to do with the fact that the otter is considerably more sensitive
to dioxin-like compounds than the reference species, so there is reason to believe that the
true threshold for toxic effects would be at the low end of the range established for the
mink.
To follow up on this latter point, one of the other members of the group noted that
the range in the threshold for fish covers three orders of magnitude, and that this is a
TCDD-based threshold. Given that the uncertainty in the threshold value for a single
congener, particularly TCDD, is so great, this person wondered how much the estimated
order-of-magnitude uncertainty in TEF values would actually add to the overall
uncertainty of the assessment.
Another group member elaborated on the decision not to recommend a food web
model for this system. First, group members had concluded that it would be difficult to
51
-------
obtain credible water concentrations for the individual congeners, since they are present
at such low levels. It would also be difficult to estimate sediment values, since the
distribution of these compounds in sediment was likely to be heterogeneous. As a result,
group members thought that development of species-specific BAFs and BMFs wouid be
sufficient to reduce the uncertainty without introducing such formidable analytic
challenges.
A member of the expert group raised a general issue related to the use of Ah receptor
agonist levels in the liver as a marker of exposure, since there is a tendency for these
chemicals to accumulate in the liver, and accumulation is itself dependent on the level of
exposure. One of the Planning Group members pointed out that studies addressing this
issue have shown no effect on the BMFs for the various congeners.
Another member of the Planning Group questioned the workgroup's use of a 50%
reduction as a more or less universal population effect of concern, rather than tailoring
this threshold to the local population. He thought that for bald eagles or nesting pairs,
for example, a different metric might be more appropriate. The group member who had
originally proposed the 50% value agreed, and said that historical records of reproductive
performance might also be useful if the number of individuals or nesting pairs in the
system was small. A different member of the Planning Group suggested that another way
to approach this issue would be to simply use exceedance of the standard as a surrogate
for population-level effects, since standards are developed to protect the most sensitive
members of a population.
Group 3. Dr. Menzie said that his group began by revisiting a couple of the topics
they had addressed previously, during consideration of the prospective case study. One
member of the group, for example, had developed a concern that the uncertainty
associated with the derivation of TEFs might be greater than v/as reflected in the matrix
the group presented at the previous day's plenary session. The: group therefore decided
that it was important to stress that the matrix was intended to illustrate a conceptual
approach, rather than to present hard and fast descriptions of the uncertainty in this
particular system.
52
-------
The group also revisited the issue of uncertainty in the water quality standards.
Initially, the group had thought about the uncertainty in these values as having mainly to
do with the interspecies extrapolations required in the application of these values.
Subsequently, however, group members realized that there are probably other
uncertainties associated with these values as well. The lesson, Dr. Menzie suggested, is
that it is important to think about uncertainties on the exposure as well as the effects side
of the analysis.
Like the previous group, Dr. Menzie's workgroup was able to trace the origin of the
WHO consensus TEFs for fish and birds, but not for mammals. The group understood
that this information does exist, but for purposes of this risk assessment the associated
uncertainties were not quantifiable. Given the importance of uncertainty information to
the risk assessment process, the group decided to recommend that some organization
make an effort to provide that level of documentation for the consensus TEF values, so
that risk assessors could have a better understanding of where those values come from.
One of the lessons the group learned from the case study exercise had to do with the
availability of site-specific measurements in this case study. The group discussed the
uncertainties associated with the measurements themselves, and concluded the need for
measuring a large number of congeners in the TEF approach did not add appreciably to
the overall uncertainty of the assessment. Assuming that appropriate analytical methods
are used, the group thought that errors in these measurements would fall in the 5% to
30% range. The effect of these uncertainties might be substantial, however, if there was
reason to question the analytical methods themselves.
Another point of discussion had to do with the potential for uncertainties related to
detection limits for the individual congeners. In some situations, the detection limits of
an analytical method might be well above levels of a congener that are of importance for
risk assessment purposes. Because of this, risk assessors involved in a TEF/TEQ analysis
must recognize the importance of achieving detection levels that correspond to the needs
of the assessment process.
53
-------
Dr. Menzie noted that the group talked a little bit about whether there are any
sampling issues that are specific to the TEF/TEQ approach. .Although they recognized
sampling as an important element of the risk assessment process, group members did not
think that sampling issues associated with the TEF/TEQ approach are any different than
those associated with other methodologies.
Group members thought that the cost of the TEF approach would probably be greater
than the cost of other methods, since the need for multiple-congener analysis translates to
a higher price per sample. Some members predicted, however, that the cost of multi-
congener analyses will decline as this methodology becomes more widely used.
The group also discussed how a risk assessor might use the TEF approach in dealing
with a partial data set - for example, one in which data were available only for PCBs.
The group decided that in such a case it would be very valuable to analyze at least some
samples for the full suite of congeners to get some sense of the relative importance of the
different congener groups and to confirm that the compounds for which data are available
are actually the congeners driving the assessment.
4
As a longer-term improvement to the methodology, the group felt that it might be
useful to see if there is a reliable way of identifying, on a site-specific basis, a simpler
measurement that could be used as a surrogate for TEQs. If so, this surrogate could be
used to more cost-effectively monitor the effects of a remediation effort over time.
To address the quantitative aspects of the retrospective case study, Dr. Menzie's group
used a process that was similar to those used in the other two groups, and they arrived at
essentially the same conclusions. One caveat that the group thought it important to
mention, however, is that there could be effects on endpoints other than reproduction
that are not specifically being addressed in the risk assessment, particularly with regard to
PCBs.
Regarding the issue of whether the TEF methodology was robust enough to support a
regulatory decision, the group first agreed that the decision might involve a range of
options rather than simply focusing on whether or not to dredge. In thinking about the
54
-------
Dr. Menzie noted that the group talked a little bit about whether there are any
sampling issues that are specific to the TEF/TEQ approach. Although they recognized
sampling as an important element of the risk assessment process, group members did not
think that sampling issues associated with the TEF/TEQ approach are any different than
those associated with other methodologies.
Group members thought that the cost of the TEF approach would probably be greater
than the cost of other methods, since the need for multiple-congener analysis translates to
a higher price per sample. Some members predicted, however, that the cost of multi-
congener analyses will decline as this methodology becomes more widely used.
The group also discussed how a risk assessor might use the TEF approach in dealing
with a partial data set—for example, one in which data were available only for PCBs.
The group decided that in such a case it would be very valuable to analyze at least some
samples for the full suite of congeners to get some sense of the relative importance of the
different congener groups and to confirm that the compounds for which data are available
are actually the congeners driving the assessment.
As a longer-term improvement to the methodology, the group felt that it might be
useful to see if there is a reliable way of identifying, on a site-specific basis, a simpler
measurement that could be used as a surrogate for TEQs. If so, this surrogate could be
used to more cost-effectively monitor the effects of a remediation effort over time.
To address the quantitative aspects of the retrospective case study, Dr. Menzie's group
used a process that was similar to those used in the other two groups, and they arrived at
essentially the same conclusions. One caveat that the group thought it important to
mention, however, is that there could be effects on endpoints other than reproduction
that are not specifically being addressed in the risk assessment, particularly with regard to
PCBs.
Regarding the issue of whether the TEF methodology was robust enough to support a
regulatory decision, the group first agreed that the decision might involve a range of
options rather than simply focusing on whether or not to dredge. In thinking about the
55
-------
the spill. Given its borderline status, some members felt that a recommendation to
simply monitor the system might be appropriate, while other, thought that it would be
preferable to formally model what the system was likely to look like in years to come.
Additional lines of evidence that the group thought mighi. be brought to bear on the
remediation decision include more extensive field observations of the current state of the
population, with attention to whether effects predicted by the TEF/TEQ. approach are
actually occurring at the individual level. Similarly, they thought that it would be useful
to obtain a more precise understanding of the distribution of contaminants within the
sediments, so that remediation efforts can be directed where xhey are most needed.
A final point of discussion within the group had to do with the need for a top-down,
population-level analysis of this system. In general, Dr. Menzie said, group members'
sense of the urgency of this need tended to reflect their individual areas of expertise and
familiarity with specific tools. Thus, toxicologists were more comfortable with the idea of
collecting and working with toxicity data, while the population biologists were more
comfortable with the use of specific metrics to describe what is going on in the system at a
population level. During the course of this discussion, however, all members of the group
agreed that it will be important to find ways of bringing together the lines of evidence
that come from these different perspectives.
Following Dr. Menzie's summary of the group's deliberations, Dr. van den Berg noted
that several groups had commented on the lack of transparency in the derivation of
WHO consensus TEF values for mammals. He indicated that the authors of the WHO
document had not realized that these values would be useful, and he said that specific
references to the studies driving those TEF values would be aided to the paper, at least
for those TEFs that were changed by the Working Group. Adding this information for
the TEFs that were adopted without modification may be difficult, since documentation
as to how those values were derived is scant.
Regarding the issue of expressing the consensus TEFs as ranges rather than point
estimates, Dr. van den Berg said that participants at the Stockholm meeting had decided
56
-------
against this approach because many of the TEFs were derived from a variety of endpoints
and so may have a range that covers several orders of magnitude. In the past, people have
used this fact to wrongly claim that the TEF system doesn't work. If risk assessors wish to
work with ranges instead of point estimates, Dr. van den Berg suggested that they go back
to the studies from which the TEFs were derived, and develop their own TEF ranges from
the ones that are most appropriate to the site they are assessing.
A member of the Planning Group noted that the 1994 Ahlborg paper does include
histograms describing the studies used to derive mammalian TEFs, and that, contrary to
popular belief, a large number of these values are based on in vivo, Tier 1-level data.
Another member of the Planning Group asked Dr. van den Berg to comment on the
accessibility of the Karolinska database and on how the database would be maintained -
whether anyone had assumed responsibility for keeping it current and/or for assessing the
quality of studies that are included. Dr. van den Berg said that it was his understanding
that the database would be accessible to anyone who wanted to use it, and that the charge
for access would be minimal. Regarding maintenance of the database, he noted that at
the time of the Stockholm meeting the database was two or three months behind the .
calendar. Although he did not know whether the database has been similarly maintained
since the meeting, he indicated that the issue of maintenance is currently being discussed.
There are no plans to review the data from a quality control perspective, but informal
guidelines have been established.
After this exchange, another member of the Planning Group commented on the
Menzie group's discussion of detection limits as they relate to use of the TEF approach,
noting that one way to address this problem is to be sure that the concentrations a lab
provides are accompanied by information about the quantitative limits of the detection
method.
One of the experts questioned the group's suggestion that a surrogate such as total
PCBs might be useful for screening or monitoring purposes. He cautioned that this could
be misleading, as it would be in the retrospective scenario, where dibenzofurans, despite
57
-------
against this approach because many of the TEFs were derived from a variety of endpoints
and so may have a range that covers several orders of magnitude. In the past, people have
used this fact to wrongly claim that the TEF system doesn't work. If risk assessors wish to
work with ranges instead of point estimates, Dr. van den Berg suggested that they go back
to the studies from which the TEFs were derived, and develop their own TEF ranges from
the ones that are most appropriate to the site they are assessing.
A member of the Planning Group noted that the 1994 Ahlborg paper does include
histograms describing the studies used to derive mammalian TEFs, and that, contrary to
popular belief, a large number of these values are based on in vivo, Tier 1-level data.
Another member of the Planning Group asked Dr. van den Berg to comment on the
accessibility of the Karolinska database and on how the database would be
maintained—whether anyone had assumed responsibility for keeping it current and/or for
assessing the quality of studies that are included. Dr. van der. Berg said that it was his
understanding that the database would be accessible to anyone who wanted to use it, and
that the charge for access would be minimal. Regarding maintenance of the database, he
noted that at the time of the Stockholm meeting the database was two or three months
behind the calendar. Although he did not know whether the database has been similarly
maintained since the meeting, he indicated that the issue of maintenance is currently
being discussed. There are no plans to review the data from a quality control perspective,
but informal guidelines have been established.
After this exchange, another member of the Planning Group commented on the
Menzie group's discussion of detection limits as they relate to use of the TEF approach,
noting that one way to address this problem is to be sure that the concentrations a lab
provides are accompanied by information about the quantitative limits of the detection
method.
One of the experts questioned the group's suggestion that A surrogate such as total
PCBs might be useful for screening or monitoring purposes. He cautioned that this could
be misleading, as it would be in the retrospective scenario, where dibenzofurans, despite
58
-------
the evaluation of point source discharges (within the framework of the Clean Water Act)
and the evaluation of contaminated sites (within the framework of the Comprehensive
Environmental Remediation and Compensation Liability Act). The applicability of the
method is situation-specific. As with any method, appropriate caution should be
exercised to avoid misuse or application of the methodology to situations where the
underlying assumptions are known not to be valid. When applying the method, it should
be recognized that there may be effects associated with the chemicals of concern that are
unrelated to AhR and, therefore, may need to be evaluated under a separate methodology.
These possibilities should be considered during the planning stage of an assessment.
2. The TEF/TEQ methodology reduces uncertainties associated with developing dose-
response information for AhR agonists that exist with methods that rely on a single
compound (e.g., TCDD) or on compounds evaluated as an aggregate (e.g., total PCBs).
Specifically, because the method takes into account the possible effects of the suite of
chemicals that act as AhR agonists, it is less likely to underestimate risks than are
methods based on only one of these compounds (i.e., TCDD). Further, because total
PCBs in the environment can be comprised of many compounds that vary in
concentration and potency as AhR agonists, the TEF/TEQ methodology provides a means
for accounting for these variables.
3. The uncertainties associated with using REPs or TEFs are not thought to be larger
than other sources of uncertainty within the risk assessment process (e.g., dose-response
assessment, exposure assessment, and risk characterization.) However, these
uncertainties should be quantified better.
O
4. As is the case with any ecological risk assessment, the nature and magnitude of
uncertainties should be identified and carried through the ecological risk assessment
process (dose-response assessment, effects assessment and risk characterization). This
could involve a number of different approaches, including qualitative analyses,
assignment of ordinal rankings to sources of uncertainty, presentation of ranges, fuzzy
arithmetic, and probabilistic analyses. Information on the sensitivity of the risk estimates
to the uncertainties associated with the TEF approach (as well as other ERA components)
59
-------
should be identified and quantified (if possible). This knowledge can be used to
communicate the range of possible results to the decision maker and to identify what
additional information would be the most useful for decisionmaking. Specific examples of
approaches are provided in the summaries of the workshop breakout group sessions on
the case studies (Appendix E).
5. Workshop participants supported the use of a hierarchical procedure for selecting
REP or TEF values for use in risk assessment. In general, the most appropriate values are
those that are closely related to the taxa and endpoints being evaluated. Workgroup
participants agreed that uncertainties are introduced with increasing taxonomic and
endpoint extrapolation. The workgroups suggested schemes i:or selecting REP and/or
WHO TEF values, as well as schemes for considering how uncertainties associated with
selecting values can be identified and tracked. These are identified in the workgroup
summaries (Appendix E).
6. A database of REP and TEF values should be maintained in order to facilitate the
application of the hierarchical procedure and to enable the conduct of sensitivity and
uncertainty analyses. The appropriate regulatory agencies will need to consider how to
insure the quality of the data in the database, document the values and the procedures
used to derive them, make the database accessible, and provide guidance for its use.
7. The derivation of REP and WHO TEF values needs to be adequately documented
(including specific citations) in order to support the use of these values in regulatory risk
assessments. The WHO TEF document provided to worksho;p participants did not
include documentation for the mammalian TEF values. This was viewed as a major
limitation on the use of the document for risk assessment purposes.
, 8. The TEF/TEQ method requires analytical methods to identify and quantify the
individual dioxin, furan, and PCB compounds. The accuracy .and precision of available
methods are considered acceptable for risk assessment purposes. The analytical
measurement errors are not considered to be a large source of uncertainty within the
assessment. A few of the workshop participants familiar with the analytical methods
reported measurement errors in the range of 5 to 30%.
60
-------
9. The costs for analyzing the suite of individual dioxin, furan, and PCB compounds
are greater than those associated with analyzing an individual compound (e.g., TCDD) or
for measuring "total PCBs." Workshop participants agreed that it may be possible to
focus the analytical effort at different stages of the assessment, thereby reducing costs.
For example, investigations may indicate that risks are due to a few of the compounds or
to a particular class and these may form the basis for subsequent evaluation. Further, it
may be possible to complement detailed analyses of individual compounds with simpler
and cheaper analytical methods (e.g., to provide information on spatial extent of
contamination).
10. Analytical detection levels for congeners need to be lower than concentrations at
which important biological effects might occur. Workshop participants agreed that this
can be achieved with available methods. As with any analytical program where data will
be used in risk assessments, data quality objectives should be specified and care taken to
insure that they are met.
11. Because physical, chemical, and biological properties vary among the individual
dioxin, furan, and PCB compounds, exposure assessments that complement the TEF/TEQ
methodology may require more information and resources (i.e., effort) than exposure
assessments for an individual compound (e.g., TCDD) or a class of compounds (e.g., total
PCBs). Fate and transport models used to support the exposure assessment will need to
account for individual compounds through the various modeled components. In some
cases, it may be possible to model groups of compounds with similar fate and transport
properties.
12. Information on the environmental behavior of individual chemical congeners is
needed to understand and use the congener-specific information in a modeling effort.
With increasing use of a TEF/TEQ approach, gaps in knowledge on chemical-specific
environmental behavior will become evident. Regulatory agencies will need to consider
how best to acquire this information and/or develop exposure assessment tools that can
complement the use of TEF/TEQ for specific regulatory applications.
61
-------
13. Application of a TEF/TEQ method could be considered within the framework of a
"lines of evidence" approach as described within the EPA's guidance for ecological risk
assessment. As such, additional field and laboratory information could corroborate or
improve the results of an assessment that is based, in part, o n the application of the
TEF/TEQ method, analysis. Use and integration of various lines of evidence in ecological
risk assessment can often strengthen the analysis and provid; a greater degree of
confidence in the results than can be achieved from relying cnly on a single line of
evidence. Each piece of information will have inherent strengths and limitations, and the
amount of confidence placed on the information will also reflect the technical background
of the individuals using the method and their experience with it.
14. Several workshop participants stressed the value of applying population-level
assessment tools and obtaining population-level information in support of assessments
(i.e., as a line of evidence). These included methods by which risks to individuals could
be described in terms of potential risks to local populations. In addition, a few
participants gave examples of tools that could be helpful for assessing whether
population-level effects were being manifested (for retrospective assessments.) Examples
included direct observations of hatching success, the condition of fledgling birds, and the
age structure of populations.
15. Participants also discussed the use of bioassay tools to support the assessment.
These methods could complement assessments that rely upon the TEF/TEQ approach.
One participant summarized the strengths and limitations of these tools as follows. In
vitro TEQ bioassays have the advantage of measuring the integrated effects of complex
mixtures of Ah receptor agonists. In addition, such assays have the potential of
identifying compounds that act via the Ah receptor which would not be identified by a
chemical residue approach that measures only dioxins, furans and PCBs. In vitro
bio,assay-derived TEQ concentrations can be obtained at a lower cost than TEQ
concentrations obtained by analysis of chemical residues. One potential problem with in
vitro bioassays is that they can overestimate the toxic potency of compounds which are
rapidly metabolized in vivo (e.g., PCS 77). However, recent research has shown that such
problems can likely be circumvented. Various in vitro bioassays have considerable
potential for predicting TEQs which are relevant to whole organisms.
62
-------
16. Participants adopted the language given in the WHO document cautioning
against the potential misapplication of the TEF/TEQ method to environmental media
(e.g., sediments or soils). Specifically, the participants indicated that it is not appropriate
to derive TEQs for these media. TEQs are relevant only with respect to specific ecological
receptors. The methodology can be used to support decisions concerning the regulation
of point source discharges and environmental clean ups that involve chemicals in
environmental media. However, in these cases, the decision involves identifying
concentrations of chemicals and/or the composition of mixtures that would yield
acceptable TEQ with respect to specified ecological receptors.
Pan 2: Conclusions Related to Charge Questions
I. STRESS-RESPONSE PROFILE RELATIVE TO THE DERIVATION OF
SPECIFIC TEF VALUES
1. The WHO consensus TEF values are reported as point estimates and generally
rounded off to the nearest order of magnitude. For the risk assessment case studies,
additional background information used in the derivation of the TEF values is
provided. Does this additional information enhance the means of evaluating
uncertainties in the assessments? If so, how? If not, why?
Conclusion: Participants found this information useful. However, they
indicated that additional information - beyond that provided - would be important for
risk assessment purposes. This additional information includes better documentation of
the process used to derive TEF values, references for the values employed for mammalian
receptors, and access to the database.
2. Some TEFs were determined from several studies, endpoints, and exposure routes,
while other TEFs were based on a single study and endpoint. Given the range of
knowledge associated with specific compounds, should all TEFs be considered to have
similar uncertainties? Why? Or why not?
Conclusion: All TEFs should not be considered to have similar uncertainties.
Participants discussed several derivation and extrapolation issues that affect the
uncertainty associated with using TEF values. They also provided an example of how
these uncertainties might be tracked.
63
-------
16. Participants adopted the language given in the WHO document cautioning
against the potential misapplication of the TEF/TEQ method to environmental media
(e.g., sediments or soils). Specifically, the participants indicated that it is not appropriate
to derive TEQs for these media. TEQs are relevant only with respect to specific ecological
receptors. The methodology can be used to support decisions concerning the regulation
of point source discharges and environmental clean ups that anvolve chemicals in
environmental media. However, in these cases, the decision involves identifying
concentrations of chemicals and/or the composition of mixtures that would yield
acceptable TEQ with respect to specified ecological receptors.
Part 2: Conclusions Related to Charge Questions
I. STRESS-RESPONSE PROFILE RELATIVE TO THE DERIVATION OF
SPECIFIC TEF VALUES
1. The WHO consensus TEF values are reported as point estimates and generally
rounded off to the nearest order of magnitude. For the risk assessment case studies,
additional background information used in the derivation of the TEF values is
provided. Does this additional information enhance the means of evaluating
uncertainties in the assessments? If so, how? If not, why?
Conclusion: Participants found this information useful. However, they
indicated that additional information—beyond that provided-—would be important for
risk assessment purposes. This additional information includes better documentation of
the process used to derive TEF values, references for the values employed for mammalian
receptors, and access to the database.
2. Some TEFs were determined from several studies, endpoints, and exposure routes,
while other TEFs were based on a single study and endpoint. Given the range of
knowledge associated with specific compounds, should all TEFs be considered to have
similar uncertainties? Why? Or why not?
Conclusion: All TEFs should not be considered to have similar uncertainties.
.Participants discussed several derivation and extrapolation issues that affect the
uncertainty associated with using TEF values. They also provided an example of how
these uncertainties might be tracked.
64
-------
III. EXPOSURE PROFILE
1. To what extent does the TEF approach present challenges, introduce new
uncertainties, or modify old uncertainties associated with modeling the exposure
of AhR agonists? To what extent does the availability and quality of congener-
specific physico-chemical data limit the means of employing fate and transport or
food chain models?
Conclusion: The approach will likely require additional resources to model exposure
because a larger number of chemicals will need to be taken into account. Because these
chemicals vary in their properties, information is needed on various physicochemical
properties in order to support modeling efforts.
2. The route of administered or absorbed dose used to derive TEFs may differ from
those needed to establish exposure profiles in a risk assessment. To what extent do
exposure route differences used in deriving the TEFs affect their application in the
case studies?
Conclusion: This was not discussed at length.
3. To what extent does the TEF approach require a more rigorous analytical
design in quantifying sediments, soil, and biota AhR agonist concentrations than is
apparent in other methods which aggregate stressors (e.g., total PCBs)?
Conclusion: Sampling design issues were judged to be comparable. However, as
discussed in the main conclusions, there will be additional analytical costs and care must
be taken to specify and meet data quality objectives.
IV. RISK CHARACTERIZATION
1. In evaluating the case studies, are the uncertainties associated with TEFs more
problematic than other uncertainties of the risk assessments? Do the uncertainties
associated with TEFs limit the means of performing the assessments, or do the
other areas of the effect and exposure characterization contribute similar or greater
levels of uncertainty?
Conclusion: These uncertainties are not more problematic than other uncertainties
of the risk assessment. They do not limit the means of performing assessments. However,
use of the method places demands on analytical methods and on modeling of exposure.
65
-------
2. Biologically-based TEQ assays on environmental samples could be employed as
an alternative to the TEF-based approach. What would the strengths and
weaknesses of such an approach be? To what extent could these approaches be
integrated?
Conclusion: These assays should not be used as an alternative to the TEF/TEQ
approach. However, they could be used to complement the analyses. They could also be
used as a screening tool. These assays were thought to be most useful in retrospective
assessments. There was not an agreement on how they would be used in a prospective
(i.e., predictive) assessment.
66
------- |