United States
              Environmental Protection Agency
              Risk Assessment Forum
Workshop on the Application  of
2,3,7,8-TCDD Toxicity Equivalency
Factors to Fish and  Wildlife
Chicago Hilton & Towers .
Chicago, IL
January 20-22, 1998

Expert Reviewers
William Adams
Director, Environmental Science
Kennecott Utah Copper Corporation
8315 West 3595 South
Magna, UT 84044-6001
Fax: 801-252-3083
E-mail: adamsw@kennecott.com

Bjorn Brunstrom
Associate Professor
Department of
Environmental Toxicology
Uppsala University
Norbyvagen 18A
S-752 36 Uppsala
E-mail: bjom.brunstrom@etox.uu.se

Janet Burris
Senior Health Scientist
McLaren Hart/ChemRisk
109 Jefferson Avenue - Suite D
Oak Ridge. TN 37830
Fax: 423-482-9473
E-mail: oakridge_office@
Steven Bursian
Department of Animal Science
Michigan State University
132 Anthony Hall
East Lansing, Ml 48824
Fax: 517-432-1518
E-mail: bursian@pilot.msu.edu

Peter deFur
Stewardship Concepts
11223 Fox Meadow Drive
Richmond, VA 23233
Fax: 804-360-7935
E-mail: pldefur@igc.org

Joseph DePinto
Great Lakes Program
State University of New York
at Buffalo
202 Jarvis Hall
Buffalo, NY 14260-4400
Fax: 716-645-3667
E-mail: depinto@eng.buffalo.edu
Lev Ginzburg
Professor      ;*
Department of
Ecology andiEMolution
State University of New'York
Setauk^t NY
Fax: 516/751*3435
E-mail: lev@rtmgs.com-
Jay William Gooch
Senior Scientist
Professional and
Regulatory Services
Gambte Company
VvTntoti;M»rTeo»Kiidal Center
Mark Hahn
Associate Scientist
Biology Department
Woods Hole
Oceanographic Institution
45 Water Street (MS #32)
Redfield 338
Woods Hole, MA 02543-1049
Fax: 508-457-2169
E-mail: mhahn@whoi.edu

Sean Kennedy
Research Scientist
National Wildlife Research Centre
Wildlife Toxicology Division
Environment Canada
100 Gamelin Boulevard
Hull, Quebec  K1A OH3
Fax: 819-953-6612
E-mail: seah.kennedy@ec.gc.ca

Wayne G. Land is
Director, Institute of Environmental
Toxicology and Chemistry
Huxley College of
Environmental Studies
Western Washington University
516 High Street (MS-9180)
Bellingham, WA. 98225-9180
Fax: 360-€J50-7284
E-mail: landis@henson.cc.wwu.edu

Lynn McCarty
L.S. McCarty Scientific
Research & Consulting
280 Glen Oak Drive
Oakville, Ontario L6K 2J2,
Fax: 905-842-6526
United States
Environmental Protection Agency
Risk Assessment Forum
    Workshop on the Application of
    2,3J,8-TCDD Toxicity  Equivalency
    Factors to Fish and Wildlife
    Chicago Hilton & Towers
    Chicago, IL
    January 20-22, 1998

    EPA/DOI Planning  Group
     Linda Birnbaum (not in attendance)
     National Health and Environmental
     Effects Research Laboratory
     Experimental Toxicology Division
     U.S. Environmental Protection Agency
     Research Triangle Park, NC 27711
     E-mail: bimbaum.linda@epamail.epa.gov

     Stev.e Bradbury
     Regional Scientist
     U.S. Environmental Protection Agency
     999 18th Street-Suite 500
     Denver, CO 80202-2466
     Fax: 303-312-6067
     E-mail: bradbury.steven@epamail.epa.gov

     Pat Cirone
     Office of Environmental Assessment
     U.S. Environmental Protection Agency
     1200 Sixth Avenue
     Seattle, WA 98101
     E-mail: cirone.patricia@epamail.epa.gov
                        Philip Cook
                        Acting Chief, Ecological Toxicology Branch
                        National Health and Environmental
                        Effects Research Laboratory
                        Mid-continent Ecology Division
                        U.S. Environmental Protection Agency
                        6201 Congdon Boulevard
                        Duluth, MN 55804
                        E-mail: cook.philip@epamail.epa.gov

                        Mike Devito
                        National Health and Environmental
                        Effects Research Laboratory
                        Experimental Toxicology Division
                        U.S. Environmental Protection Agency
                        86 TW Alexander Drive (MD-74)
                        Research Triangle Park, NC 27711
                        E-mail: devrto.mike@epamaihepa.gov

                        Gerry Henningsen
                        Regional Senior Toxicologist
                        U.S. Environmental Protection Agency
                        999 18th Street -Suite 500
                        Denver, CO 80202-2466
                        Fax: 303-312-6065
                        E-mail: henningsen.gerry@epamail.epa.gov
        Printed on Recycled Paper

Tim Kubiak
Division of Environmental Contaminants
U.S. Fish and Wildlife Service
4401 North Fairfax Drive (ARLSQ 330)
Arlington, VA 22203
E-mail: tim_kubiak@mail.fws.gov

Cynthia Nolt
Office of Science Policy
U.S. Environmental Protection Agency
401 M Street, SW (8104R)
Washington, DC  20460
E-mail: nolt.cynthia@epamail.epa.gov

Robert Pepin
U.S. Environmental Protection Agency
77 West Jackson Boulevard (WT-16J)
Chicago. IL 60604-3590
Fax: 312-886-0168
E-mail: pepin.robert@epamail.epa.gov

Donald Tillitt
Biological Resource Division
U.S. Geological Survey
4200 New Haven Road
Columbia, MO 65201
E-mail: donald_tillitt@nbs.gov
Steve Wharton
U.S. Environmental Protection Agency
726 Minnesota Avenue (SUPR/FFSE)
Kansas City, KS 66101
E-mail: wharton.steve@epamail.epa.gov

Lisa Williams
U.S. Fish and Wildlife Service
2651 Coolidge  Road
East Lansing, Ml  48823
Fax: 517-351-1443
E-mail: lisa_williams@mail.fsw.gov

Bill Wood (not in attendance)
Risk Assessment Forum
National Center for Environmental Assessment
U.S. Environmental Protection Agency
401 M Street SW (8601)
Washington, DC 20460
Fax: 202-260-3955
E-mail: wood.bill@epamail.epa.gov

              United States
              Environmental Protection Agency
              Risk Assessment Forum
Workshop  on  the Application of
2,3,7,8-TCDD  Toxicity Equivalency
Factors to  Fish and  Wildlife
Chicago Hilton & Towers
Chicago, IL
January 20-22,  1998

John Blankenship
Assistant Regional Director
U.S. Fish and Wildlife Service
Henry Whipple Federal Building
1 Federal Drive
Fort Snelling, MN 55111-4056
612-725-3536, Ext.: 201
Fax: 612-725-3526

Douglas Beltman
Hagler Bailly Services, Inc.
1881 Ninth Street - Suite 201
Boulder, CO 80302
Fax: 303-443-5684
E-mail: dbeltman@habaco.com

John Bleiler
35 Nagog Park
Acton, MA 01720
978-635-9500, Ext.: 3050
Fax: 978-635-9180
    Printed on Recycled Paper
Christine Boivin
Risk Assessment Forum
National Center for
Environmental Assessment
U.S. Environmental
Protection Agency
401 M Street, SW(MC: 8601)
Washington, DC 20460
Fax: 202-260-3955
E-mail: boiven.chris@epamail.epa.gov

Eddie Buxton
Project Scientist
General Engineering
P.O. Box 30712
Charleston, SC 29417
Fax: 803-769-7397
E-mail: jeb2@gel.com

Rhona Compton
Cantox, Inc.
111 5th Avenue. SW
Suite 1160
Calgary, Alberta  T2P 3Y6
Fax: 403-237-0291
E-mail: rcompton@cantox.com
Bruce Diel
Section Manager, Environmental &
Analytical Chemistry
Midwest Research Institute
425 Volker Boulevard
Kansas City, KS 64110
816-753-7600, Ext.: 1631
Fax: 816-753-5359
E-mail: bdiel@mriresearch.org

Arunas Draugelis
U.S. Environmental Protection Agency
77 West Jackson Boulevard (SR-6J)
Chicago, IL 60604
Fax: 312-353-5541 .
E-mail: draugelis.arunas@epamail.epa.gov

Steve Ellingson
Geraghty and Miller
105 5th Avenue, South - Suite 350
Minneapolis, MN 55401
Fax: 612-336-4538
E-mail: sellings@gmgw.com

William Enriquez
Office of Resource Conservation
and Recovery Act
U.S. Environmental
Protection Agency
77 West Jackson Boulevard
Chicago, IL 60604-3590

Jay Field
National Oceanic & Atmospheric
U.S. Department of Commerce
7600 Sand Point Way, NE
Seattle. WA 98115
Fax: 206-526-6865
E-mail: jay.field@hazmat.noaa.gov

Brent Finley
Principal Health Scientist
1135 Atlantic Avenue
Alameda, CA  95404
Fax: 707-526-6705

Richard Fox
Sediment Quality Specialist
Hart Crowser
6250 River Road - Suite 3000
Rosemont, IL  60018
Fax: 847-292-0507

Michael Gilbertson
International Joint Commission
P.O. Box 32869
Detroit, Ml 48232
Fax: 519-257-6740
E-mail: gilbertsonm@ijc.wincom.net

Paul Goettlich
Board of Directors,  Hoosier
Environmental  Council
P.O. Box 6854
South Bend, IN 46660-6854
E-mail: gottlich@sbt.infi.net
Steve Johnson
Pesticides and Toxic
Substance Branch
U.S. Environmental
Protection Agency
77 West Jackson Boulevard
(DRT 8J)
Chicago, IL  60604
Fax: 312-353-4342

Russ Keenan
Vice President
and Chief Health Scientist
1685 Congress Street
Portland, ME 04102
Fax: 207-774-8263
E-mail: russell_keenan@mclaren-

Edward Knapick
Director of Research
Marcal Paper Mills, Inc.
One Market Street
Elmwood Park, NJ  07407-1451
Fax: 201-703-6227

Carol-Ann Manen
Chief,  Injury Assessment
Damage Assessment
National Oceanic and
Atmospheric Administration
U.S. Department of Commerce
1305 East-West Highway
Silver Spring, MD 20910
301-713-3038, Ext: 196
Fax: 301-713-4389

Afif Marouf
U.S. Environmental
Protection Agency
77 West Jackson Boulevard
Chicago, IL  60604
Fax: 312-353-5541
E-mail: marouf.afif@epamail.epa.gov
Stacy McAnulty
Project Manager
RMT, Inc.
744 Heartland Trail
Madison, Wl 53717
Fax: 608-831-3334
E-mail: stacy@rmtmsn.nntinc.com

John McCarty
Program Manager
Site Assessment and Remediation
740 Pasquinelli Drive
Westmont, IL  60559
Fax: 630-850-5307
E-mail: jmccart@ensr.com

Margaret McDonough
Environmental Scientist
U.S. Environmental Protection Agency
JFK Federal Building
Boston,  MA 02203
Fax: 617-573-9662

Michael Moore
Senior Toxicologist
PTf Environmental Services
15375 Southeast 30th-Place - Suite 250
Bellevue, WA  98007
Fax: 206-643-9827
E-mail: ptisvcs@halcyon.com

Terry Quill
Beveridge and Diamond
1350 I Street,  NW
Washington, DC 20005
Fax: 202-789-6190
E-mail: tquill@ddlar.com

William Ruoff
Project Risk Assessor
Woodward-Clyde International
Stanford Place 3 - Suite 1000
4582 South Ulster Street
Denver,  CO 80237
Fax: 303-694-3946

Daniel Smith
Conestogo Rovers, Inc.
559 West Uwchlan Avenue
Suite 120
Exton, PA 19341
Fax: 610-280-0278
E-mail: dsmith@phi.rovers.com

David Soong
Environmental Engineer
U.S. Environmental
Protection Agency
77 West Jackson. Boulevard
Chicago, IL 60604

Kirsti Sorsa
Project Scientist
RMT, Inc.
744 Heartland Trail
Madison, Wl 53717
608-831-1989, Ext.: 3338
Fax: 608-831-3334
E-mail: kirsti@rmtmsn.rmtinc.com

Thomas Starr
Environmental Corporation
7500 Rainwater Road
Raleigh, NC 27615-3700
Fax: 919-876-0201
E-mail: tbstarr@mindspring.com

Ken Stromborg
Environmental Contaminant
U.S. Fish & Wildlife Service
1015 Challenger Court
Green Bay, Wl 54311
Fax: 920-465-7410
E-mail: ken_stromborg@mail.fws.gov

Katherine Super
Risk Assessor
ICF Kaiser Engineers
1600 West Carson Street
Pittsburgh, PA  15219
Fax: 412-497-2212
E-mail: ksuper@icfkaiser.com
Dan Thomas
Senior Editor
Great Lakes Basin Publications
P.O. Box 297
Elmhurst, IL 60126
E-mail: dan@great-lakes.org

Angelique van Birgelen
National Institute of Environmental
Health Sciences
P.O. Box 12233 (MD-B307)
Research Triangle Park, NC 27709
E-mail: vanbirge@niehs.nih.gov

William van der Schalie
Office of Research and Development
National Center
for Environmental Assessment
U.S. Environmental
Protection Agency
401 M Street, SW (8623)
Washington, DC 20460
Fax: 202-260-6370

Jack Walker
Project Manager
Construction Operations
U.S. Army Corps of Engineers
334 Meeting Street (919)
Charleston, SC  29402-0919
Fax: 803-727-4445
E-mail: jwalker@sac.usace.army.mil

Jim Warchall
Sidley and Austin
First National Plaza - 53rd Floor
Chicago, IL 60603
Fax: 312-853-7036

    Appendix B

United States
Environmental Protection Agency
Risk Assessment Forum
    Workshop on the Application of

    2,3,7,8-TCDD Toxicity Equivalency

    Factors to Fish and Wildlife

    Chicago Hilton & Towers

    Chicago, IL

    January 20-22, 1998


    T U E S D A  Y ,'J A N U A R Y 2 0 ,  1  9 9 8

     3:OOPM   Registration

     4:OOPM   Welcome	 Dr. Chris Boiven
                                                   Risk Assessment Forum,
                                     U.S. Environmental Protection Agency (U.S. EPA),
                                                        Washington, DC

                                     Mr. John Blankenship, Assistant Regional Director
                                            U.S. Fish and Wildlife Service (FWS),
                                                        Fort Snelling, MN

     4:10PM   Scope and Charge for the Workshop	 Dr. Charles Menzie, Workshop Chair
                                               Menzie-Cura & Associates, Inc.,
                                                        Chelmsford, MA

     4:30PM   Synopsis of the World Health Organization Workshop
            Held in Stockholm	 Dr. Martin van den Berg
                                                     University of Utrecht,
                                                   Utrecht, The Netherlands

     5:OOPM   Presentation of Prospective Case Study and Discussion	Dr. Steve Bradburg
                                                            U.S. EPA,
                                                          Denver, CO

     5:30PM   Presentation of Retrospective Case Study and Discussion	Dr. Donald Tillitt
                                                    U.S. Geological Survey,
                                                         Columbia, MO
     6:OOPM   BREAK

       Printed on Recycled Paper

TUESDAY,  JANUARY   20,  1998

 6:15PM    Review Structure of Workshop and
            Goals and Objectives of Breakout Groups	Dr. Charles Menzie

 6:45PM    Observer Comments



 8:30AM    Expertise Group Sessions:

            Toxicity Equivalency Factors (TEFsj Experts	Dr. Richard Peterson, Facilitator
                                                                         University of Wisconsin,
                                                                                   Madison, Wl

            Fate & Transport and Bioaccumulation Experts	Dr. William Adams, Facilitator
                                                              Kennecott Utah Copper Corporation,
                                                                                    Magna, UT

           ^ Risk Assessors and Population Modelers	  Dr. Charles Menzie, Facilitator

10:30AM    BREAK

10:45AM    Breakout Group Session I: Apply TEFs to Case Study 1

            Group 1	Dr. Peter deFur, Chair
                                                             Environmental Stewardship Concepts,
                                                                                 Richmond, VA

            Group 2	 Ms. Janet Bums, Chair
                                                                        McLaren Hart/ChemRisk,
                                                                                 Oak Ridge, TN

            Group 3	Dr. Charles Menzie, Chair

            LUNCH (at discretion of individual groups)

3:45PM     BREAK

4:OOPM     Plenary Session
            Breakout groups report on Case Study 1 and discuss commonalities and differences among
            their groups


8:OOPM     Plenary Session
            Complete reports on Case Study 1 and continue plenary group discussion


THURSDAY,  JANUARY  22,   1998

8:30AM      Breakout Group Session II: Apply TEFs to Case Study 2
            Same breakout groups as Wednesday

            BREAK (at discretion of individual groups)

12:30PM     LUNCH

 1:30PM     Plenary Session
            Breakout groups report on Case Study 2 and discuss commonalities and differences among

 3:OOPM   .  BREAK

 3:15PM     Overall Meeting Conclusions and Wrap-Up	Dr. Charles Menzie

Note to Observers:
We are aware that many of you did not have the opportunity to review the
materials prior to the workshop. We encourage you to submit written
comments to the workshop and discussion group chairs, so that your
comments can be considered during the writing of the workshop summary

       Appendix C

 Workshop on the Application of 2,3,7,8-TCDD
Toxicity Equivalency Factors to Fish and Wildlife
               Premeeting Comments
                    Chicago, Illinois
                  January 20-22, 1998
                Prepared and compiled by:
              Eastern Research Group, Inc.
                  110 Hartwell Avenue
               Lexington, MA 02173-3134


                            Table of Contents

Charge to Reviewers	 C-5

Peer Reviewer Comments

William Adams	  C-12
Bjorn Brustrom	  C-20
Janet Burris	  C-26
Steve Bursian	  C-32
Peter deFur	-.	  C-40
Joseph DePinto 	  C-46
Lev Ginzburg  	  C-54
Jay Gooch	  C-58
Mark Hahn	  C-70
Sean Kennedy	  C-82
Wayne Landis	  C-92
Lynn McCarty	  C-102
Charles Menzie	'.	'.	  C-116
Chris Metcalfe	  C-122
Michael Meyer 	  C-130
Patrick O'Keefe	:	  C-136
Richard Peterson	  C-146
Mark Servos	  C-158
Martin van den Berg	  C-136
Bert van Hattum	  C-170



   It is reasonable to assume that the proposed WHO TEFs are appropriate for risk assessments
associated with permitting discharges, attributing causality to specific compounds, and establishing
remediation goals for AhR agonists. These risk assessment situations are the primary focus of the
workshop. The major issue to be addressed in the workshop is the extent to which a TEF/TEQ
approach can be used in risk assessments that have progressed beyond the screening stage.

   The primary objective of the workshop is to identify, document, and compare uncertainties  (lack
of knowledge and variability) in TEF development and their impact in ecological risk assessments.
To achieve this goal, two case studies that represent hypothetical situations for prospective and
retrospective risk assessments have been prepared. For each case study,  a series of questions
and issues are raised that will help focus the panels' deliberations. The majority of  issues/questions
raised are directed towards effect characterization topics.  However, it is recognized that assessing
the exposure of PCDD, PCDF,  and PCS mixtures is also a significant challenge for implementation
of a TEF/TEQ approach in a risk assessment. Therefore, issues and questions concerning
exposure characterizations are also provided to highlight important concepts that can not be
excluded from the risk assessment process.


   The major objective of the workshop is to address uncertainties associated with using a
TEF/TEQ approach in effects characterizations for ecological risk assessments. These
uncertainties need to be identified, documented, and to the extent possible, quantified.  For
example, there are gaps in the  TEF knowledge base for mammalian wildlife, avian wildlife, and
aquatic life in terms of interspecies, exposure route, and endpoint extrapolations. A challenge  to the
participants of this workshop is to evaluate the relative contribution of TEF-related uncertainties in
relation to other effect characterization uncertainties found within an ecological risk assessment
(e.g., uncertainties in identifying 2,3,7,8-TCDD dose levels of concern; extrapolating effects from the
individual to the population). To place the effect characterization uncertainties associated with the
use of TEFs in perspective, TEF analyses in the case studies can, for example, be compared to
analyses based on total PCBs or 2,3,7,8-TCDD alone.  Application of a TEF approach to an
ecological risk assessment also requires additional information for parameters in the exposure
characterization for the mixture. A critical need is the documentation of additional data
requirements for use of a TEF approach (e.g., K^, K<,s, BAFs, BMFs, BSAFs, biotic and abiotic
degradation rates, etc.). The extent to which these exposure issues can contribute to risk
assessment uncertainties needs to be estimated.

   The following questions are generally organized around components of the draft U.S. EPA
Ecological Risk Assessment Guidelines (U.S. EPA, 1997). It is understood that not everyone will
answer every question.  Please prepare responses to the questions appropriate to your area of


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?

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?

3.  The TEF values provided were based on endpoints that ranged from in vitro biochemical
   responses (e.g., induction of cyp1A1) 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?


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 where based on total PCBs or 2,3,7,8-TCDD alone?

2.  Many TEFs are based on LC50 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?

3.  The TEFs 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?

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?

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?

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)?


    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?

    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?

    3.   Assume that site-specific data or additional research could be gathered or performed to
       generate more information for the case study assessments. Provide a list of specific
       investigations/studies and rank them from highest to lowest priority. What is your rationale
       for the ranking?

Additional Questions Specific to the Prospective Case Study:


    1.   The state adopted BAF^s used by the GLWQG. What improvement in the accuracy of
       maximum allowable concentrations for individual congeners in water, (MAC^, can be
       expected through use of BAF"s determined from Roundtail Lake data?

    2.   What errors are associated with the state's application of the GLWQG TCDD water quality
       standards for birds and mammals without consideration of congene~r-specific differences in
       biomagnification factors from fish to tissues in wildlife relevant to the effects of concern?


    3.   How should the uncertainties associated with the available fish, avian, and mammalian  TEFs
       be incorporated into decisions about which TCDD water quality standard should be chosen
       for setting a TEqTMDL for regulating chemical discharges into Roundtail Lake?

Additional Questions Relative to the Retrospective Case Study:


    1.   Would TEQ sediment cleanup goals be the same for each vertebrate group? If not, why
    .   would there be a difference? If the vertebrate group with the most certainty is not the group,
       with the most restrictive sediment cleanup goal, how would you council the risk manager's
       concerns for the other vertebrate groups?

    2.   Would the TEF/TEQ-based sediment remediation goals be the same as those determined for
       total PCBs for the identical vertebrate class?  Assume that a simple ratio of total PCB
       sediment concentration goal to TEQ sediment concentration goals was formulated to allow
       for the use of total PCBs to monitor  cleanup efforts based on TEQs.  What exposure and
       effect issues would need to  be evaluated before using the less costly total PCB analysis to
       support the TEQ-based sediment remediation goal?

Table 1
Parameters for PCBs, PCDDs, PCDFs

Total PCB






(Eisler & _
Belisle. 1996)2




NA "

(MacKay, Shiu &
Ma. 1992)3

5.8 - 6.3
6.1 - 6.8
6.3 - 7.5









Lake Trout BSAF
(Oliver & Niimi)'
(g OC/g lip)




(g OC/g lip)





3.31 E+08

6.71 E+05


(g fish/g egg)




BMF be, I7
(g lip/ g lip)




(g lip/ g lip)




1  US EPA, 1995 (EPA-82O-B-95-005).
2.  Eisler, R., and A.A. Belisle. 1996. Planar PCB Hazards to Fish, Wildlife, and Invertebrates: A Synoptic Review. National Biological Service Biological Report 31, 75pp.
3.  Mackay, Shiu & Ma. 1992. Illustrated Handbook of Physical and Chemical Properties lor Organic Chemicals. Boca Raton, FL Lewis Publishers.
4  Values from Eisler & Belisle (1996) or MacKay. Shiu & Ma (1992).
S.  Mean BAF ", for salmonids from Table 10 of US EPA (1995), with the exception of total PCBs which are from Appendix F.
6.  BMFbe.w is the BMF from forage fish to bird eggs on a wet weight basis from Braune, B.M., and R.J. Norstrom. 1989. Dynamics of Organochlorine Compounds in Herring
   Gulls: III. Tissue Distribution and Bioaccumulation in Lake Ontario Gulls.  Environ. Toxicol. Chem. 8:957-966.
   BMFs for PCB congeners 77.126, and 169 are from the same samples, but reported in Hoffman et al.(1996).
7.  BMF be,I is the BMF from forage fish to bird eggs on a lipid basis calculated from the % lipid in the fish and bird eggs from Braune and Norstrom (1989).
   BMFs for PCB congeners 77,126, and 169 are from the same samples but reported in Hoffman et al., (1996).
8.  BMFdl.l is the BMF from diet to mink liver on a lipid basis from Tillrtt et al. (1996). The BMFs were normalized to feed consumption which differed among treatment groups.  The BMFs
    in this column are the means, among treatment groups, of the BMFs for which both values (diet and liver concentrations) were above the limit of quantitation unless noted by an *.

NA = Not available
BMF = Biomagnification factor
BAF = Bioaccumulation factor
EPA = Environmental Protection Agency
Ko, = Octanol water partition coefficient
GU = Great Lakes Water Quality Initiative
OC = Organic carbon
Up = lipid
g = grams

Peer Reviewer Comments


                                                   William J. Adams, Ph.D.
                                                 Director, Environmental Science
                                              Kennecott Utah Copper Corporation
                                                         8315 West 3595 South
                                                                P.O. Box 6001
                                                        Magna, UT 84044-6001
                                                            Fax: 801-252-3083
                                                 E-mail: adamsw@kennecott.com
An expert in bioaccumulation factors, Dr. Adams received both his Ph.D. in aquatic
toxicology and an M.S. degree in wildlife toxicology at Michigan State University, and
his B.S. in biological sciences at Lake Superior State University. In his current position
as the director of environmental science at Kennecott Utah Copper Corporation, Dr.
Adams directs environmental research in the areas of toxicology and environmental risk
assessment.  In addition to aquatic toxicology, he has expertise in environmental fate
and environmental risk assessment. He has extensive experience with both metals and
organics and has published 50 articles in  his areas  of study. Dr. Adams is also an
editor/author of two books and two book chapters. He is a member of the EPA Science
Advisory Board, Committee on Environmental Processes and Effects.

                                                            William J. Adams

1.  It was not clear.to me that the additional information provided  reduced the
    uncertainty associated with the TEFs.  Perhaps this will become clearer at the
    meeting.  The rounding of the TEFs to the nearest order of magnitude introduces
    uncertainty in the final calculation of risk and it reflects the uncertainty associated
    with the individual values. Additional discussion on how this uncertainty should be
    dealt with in a risk assessment context is needed at the workshop.

2.  Intuitively, I would say that all TEFs should not be considered to have similar
    uncertainties.  This is based on both weight of evidence and lines of evidence for
    those chemicals which have been studied the most. However, it is fair to ask the
    question, can we quantify the uncertainty through rigorous statistical assessment
    of the available data on TEFs? The question posed is somewhat similar to asking
    the question, would a single acute toxicity test with Daphnia magna have the same
    uncertainty in deriving a water quality criterion as a genus mean acute value based
    on the average of several Daphnia magna studies as well as several other daphnid
    species.  The answer is, of course, that we would have less uncertainty with a
    genus mean acute value than with a single acute toxicity test.

3.  This question gets to the heart of the entire risk assessment approach for TCDD
    and other HOHs and  deserves in depth  review at the workshop. The TEF
    approach is one that has found favor because it provides a way forward  for
    numerous chemicals with a similar mode of action.  The complexity of assessing
    all PCB, Furan and Dioxin isomers is monumental and is somewhat simplified by
    this approach.  However, care has to be taken in the use of the "model" results as
    measurement endpoints for the purpose of evaluating key assessment endpoints
    (i.e., the valued resource) in risk assessments.  The data seem to indicate that the
    use of in vitro measurements and QSARs introduce additional uncertainty into the
    measures of effects that are ultimately used to estimate risk. A measurement of
    selenium in the egg of a black-necked stilt, for example, provides a reasonable

                                                                William J. Adams
      estimate of the potential  for reproductive effects at the individual  level.   A
      measurement of selenium  in the diet of the birds can be used to estimate egg
      concentrations and reproductive effects, but the uncertainty becomes greater.
      Measuring the selenium in  the sediments where the dietary species lives as an
      indicator of potential for reproductive effects introduces  even more error. The
      same analogy applies here. As a general rule, the further away you get from the
      a direct measure of the assessment endpoint the uncertainty becomes greater.  I
      would add, that this does not necessarily imply that as the uncertainty increase
      there is a need for use of additional safety factors. The inappropriate use of safety
      factors has been shown to increase the conservatism and decreases the accuracy
    -  of the risk estimate. Ultimately, what has to be answered is, can TEFs be used to
      accurately predict population effects in aquatic ecosystems?  This can only be
      answered by the careful use of both laboratory  and field data.


   1.  When assessing chemicals  with the same mode of action and the same receptor,
      the literature strongly supports the use of an additive  model.  According to
      Konemann,  when  there is no interaction between the chemicals they  can  be
      assumed to be additive.  If you do not consider dioxin isomer effects to be additive
      then one must assume they are either antagonistic or synergistic or unpredictable.
      The consequence is that  you have to assess each  isomer independently and
      determine its potential  to  cause effects (calculate separate hazard quotients).
      Adding individual hazard quotients to  assess  the overall potential for risk has
      serious limitations. The use of total PCBs, which  has limitations unique to itself due
      to the environmental degradation of the constituents, or  the use of just 2,3,7,8-
      TCDD provides a  single point estimate of the potential for risk, but does not
      consider the cumulative potential for risk from similar compounds co-located in the
      environment.  The fundamental basis  for using an additive model exists, what
      hasn't been determined accurately is when does it over predict the potential for
      effects? The potential for  antagonism appears to be somewhat greater than for

                                                                 William J. Adams

   2.  The use of TEFs based on central tendency values such as LC50 or EC50 values
       can be justified even though most in-depth assessment typically use chronic no-
       effect concentrations or threshold values. The selection of a very sensitive EC50
       value can in many cases be more sensitive than .some chronic threshold values.
       Risk assessments  with atrazine, diazinon, copper and  cadmium have shown
       where there were lots of acute and chronic data that the water concentrations
       selected as protective of aquatic species (95%) using sensitive acute endpoints
       were nearly the same as the values selected using chronic no-effect levels.

   3.  My response to this question comes not from extensive experience with TEFs, but
       with  having performed risk assessments where laboratory to field relationships
       have been examined. As a general rule, when one has to extrapolate within a
       class the best approach is to use the  same value for the species of interest as
       obtained from the toxicity test. The use of safety factors in this situation provides
       protection, but sacrifices accuracy and predictability.


   1.  I don't see the use of the TEF approach introducing significant new uncertainties
       into the exposure assessment.  However, in regards to the second part of this
       question,  the  lack  of congener-specific  physico-chemical  data  provides
       considerable issues relative to accurately modeling or predicting the fate and
       transport of these materials both within the physical environment  and the biota.
       Transport estimates within the food chain can be developed without these data if
       one chooses to rely only upon sediment to biota and biota to biota accumulation

   2.  The route of administration is always important in an overall risk assessment. The
       key is to match  the route of "exposure" in the effects characterization with that
       which actually occurs in natural systems. TEFs derived from in vitro biochemical
       measurements will  have greater uncertainty associated with them because the

                                                                William J. Adams
       potential for metabolism to occur in the body is removed. This ultimately translates
       to an increase in the uncertainty  associated with the final risk estimate.

   3.  Interesting question.  Is more error introduced via the analytical techniques used
       when multiple chemicals are measured and quantified than when a  class of
       chemical are measured as a group.  I would think so. I'll leave this question to the


   1.  Regarding uncertainty introduced into the risk assessment by the use of TEFs,  I
       don't think the use of TEFs necessarily introduces additional uncertainty into the
       risk assessments that other approaches would not.  However, not all TEFs are
       equal (i.e., some are based  on  QSARs, in vitro biochemical measures, in vivo
       effects measurements etc.), therefore, depending on-what is actually used as the
       set of TEFs for the risk assessment may have more or less uncertainty.  Clearly,
       extrapolating  across  species and  perhaps  classes introduces uncertainty.
       Additionally, estimating exposure from  sediment using BSAFs has considerable
       uncertainty when one considers all the compounds of interest. So, does the use
       of TEFs introduce more uncertainty than already exists?  Who  knows? The
       question which should be asked is, can we measure and quantify the uncertainty
       in each part of  the risk assessment?  If so, this would  be a  useful research

   2.  The use of residue based approaches for deriving  water  quality criteria and
       performing risk assessments is gaining favor for both organics and metals.  I favor
       the approach and believe the uncertainty associated with the risk estimate would
       be reduced. This approach can be applied in several ways including evaluation of
       TEQs within a given tissue (liver) for a valued species (otters) or alternatively by
       comparing the TEQs in the diet of a given species against a known dietary effect
       concentration. I  prefer the latter because it gets directly at the issue of exposure
       for HOHs.

                                                                 William J. Adams

       The process of risk characterization typically looks at several lines of evidence to
       help assess the uncertainty, therefore I don't see the various approaches as
       mutually exclusive. Why not perform the assessment using both TEFs and TEQs
       for comparative purposes (cost aside)?  Especially if you are locating a new
       industry as proposed in the prospective case study.

   3.  A.   Gather additional data on the species or resource to be protected. I would not
           assume, for example, that Bull trout are as sensitive as lake trout and then
           divide by a factor of 10  to account for  species to species  extrapolation.
           Perform the necessary early life stage test or egg exposure study to obtain the
           information. At each stage of the assessment I would gather as much site-
           specific data as  possible on the species of interest.  This will  reduce

       B.   Collect additional field data at the population level at each of the sites used in
           the risk assessment cases. Risk assessments performed at the species level
           and extrapolated  to the population or community level  tend to be  overly
           conservative. They typically assume constant dietary and water exposure, for
           example, and this is rarely true. Individual level assessments rarely consider
           the behavioral aspects of populations, e.g. migration, feeding behavior, habitat
           selection, etc. all  of which effect the exposure regime.  Additional on-site
           evaluation of the populations of interest (in the retrospective case) will provide
           additional information on whether or not actual effects occurring at the site.

Additional Questions Specific to the Prospective Case:


 .  1.  No comment at this time.

   2.  No comment at this time.

                                                                William J. Adams

   3.  Relative to the question as to how uncertainty should be handled in setting water
       quality standards - this is an area where the state-of-the-science is improving.
       Probabilistic techniques are emerging using Bayesian theory and Monte Carlo
       calculations to account for uncertainty and to predict a range of values that might
       be protective. The advantage of this approach is that it also provides an estimate
       of the confidence along the range of values identified such that one can select a
       value with a given level of confidence (say 90%). The approach can be used to
       include site-specific parameters and can be used at the population level if sufficient
       data  are available.. We  recently completed such an approach for  selenium to
       assess levels in water that are protective of bird egg concentrations to prevent
       teratogenic effects.  .

Additional Questions Specific to the Retrospective Case:


   1.  I would think that the TEQ sediment cleanup goals would not be the same for each
       vertebrate group. There are differences in sensitivity of different vertebrate species
       to TCDD and similar compounds (consider the variability that exists just for trout
       species to TCDD) and this should be evaluated and discussed as part of the
       effects characterization.

       Providing the risk manager with an assessment of the uncertainty associated with
       each of the risk estimates in the overall risk assessment is the job of the risk
       assessor.  Hence the statements identifying the uncertainty with a given risk
       estimate become very important. A decision, in fact, could be made by the risk
       manager to set a level of protection based on a less sensitive species when the
       data are well characterized as opposed to using a more sensitive species, but with
       an uncertainty level so large that the confidence in the estimate is very low. Risk
       management is not a quantitative science and often involves personal judgement

                                                              William J. Adams
   and personal/societal values.  The risk assessor must provide sufficient information
   so the manager can make an informed decision. In short, if you do not have much
   confidence in your risk estimate the selection of a level of protection is very difficult
   and often tends to be on the conservative side.

2. Relative to the second part of this question - what exposure and effects issues
   would have to be evaluated before using the less costly PCB  analysis as an
   alternative to TEQ-based sediment remediation goal-1 suggest the following might
   be important:
   1.  Exposure - the quantitation of the total  PCBs has to be matched to the
       congeners used to perform the TEF/TEQ assessment other wise there will not
       be a match between the exposure and effects estimates.

   2.  There needs to be in vivo laboratory evaluation between effects observed
       using the TEF/TEQ approach and that obtained using the total PCB approach.
       This approach tests the additivity model, reproducibility and provides the first
       level  of field verification.  .

                                                       Mark E. Hahn, Ph.D.
                                                            Associate Scientist
                                                    Biology Department, MS #32
                                            Woods Hole Oceanographic Institution
                                                   Woods Hole, MA 02543-1049
                                                            Fax: 508-457-2169
                                                       E-mail: mhahn@whoi.edu
Dr. Hahn received his B.S. in biological sciences from the State University of New York
at Binghamton and a Ph.D. in toxicology from the University of Rochester, New York.
His doctoral thesis  was  titled  "Studies  on the Role of the Ah Receptor in
Hexachlorobenzene-lnduced Porphyria." He received the New Investigator Award from
SETAC as well as the Individual National  Research Service Award from NIEHS. His
research interests include chemical-biological interactions, especially receptor-mediated
mechanisms governing interactions of halogenated aromatic hydrocarbons with fish
and other aquatic wildlife, and his major focus is on the comparative biochemistry and
molecular biology of the Ah receptor.  Dr. Hahn is a reviewer for more than a dozen
related publications, as well as a frequent contributor to journals and other publications
in the fields of  aquatic toxicology, marine environmental  research, biochemistry,
pharmacology, and chemistry. Dr. Hahn is a member of the American Association for
the Advancement of Science, the Society of Toxicology, SETAC, the American  Society
for Biochemistry  and Molecular Biology, and the Society for Marine Mammology.

                            Workshop on the Application of
                   2,3,7,8-TCDD Toricity Equivalency Factors (TEFs)
                                  to Fish and Wildlife

                                 Pre-meeting comments

                                  Associate Scientist
                              Biology Department, MS-#32
                          Woods Hole Oceanographic Institution
                             Woods Hole, MA  02543-1049
                                email: mhahn@whoi.edu
                   WWW site: http://www.whoi.edu/biology/hahnm.html
                                 Phone: 508-289-3242
                                  Fax: 508-457-2169

I. Derivation of specific TEF values.
1.  Does the additional information provided enhance the evaluation of uncertainties in the
   Documents such as the draft WHO report identify some of the data gaps, sources of variability
and uncertainty, and possible shortcomings of the TEFs used and thus are helpful in  evaluating
uncertainties in the assessments.  Rather than the many papers and reports provided, a single
document summarizing all  the TEF values in the literature might be more useful.  Such  a
document could also review the major sources of uncertainty and perhaps even provide estimates
of the magnitude of each.

2.  Should all TEFs be considered to have similar uncertainties?
   Theoretically, the degree of uncertainty associated with each "consensus  TEF"  should be
compound-specific.  This is because certain compounds may be more strongly affected by the
variables that  lead  to uncertainty.  For example, a  compound that is broadly resistant  to
metabolism may show less variability (uncertainty) in TEF  values obtained in different systems
than a  compound that  exhibits differential metabolism  among  systems.   Similarly,  some
compounds (e.g. 2,3,7,8-PCDDs) will have high affinity for the AHR in most species, while for
other compounds  (e.g. mono-ortho-PCBs) there may be substantial species-specific variation in
their ability to bind the AHR.
   Whether this is true can be evaluated by looking at the range of TEF values for each compound
in a variety of systems.  As an example, Figure 1 shows a  comparison of all published relative

                                                   Bjorn Brunstrom, Ph.D.
                                                           Associate Professor
                                          Department of Environmental Toxicology
                                                            Uppsala University
                                                              Norbyvagen 18A
                                                      S-75236 Uppsala, Sweden
                                              E-mail: bjorn.brunstrom@etox.uu.se
Dr. Brunstrom received a Ph.D. from Uppsala University in zoophysiology/ecotoxicology
and continues to perform postdoctoral research at Uppsala.   He is currently an
associate professor in ecotoxicology. He participated in the World Health Organization's
"Meeting on the Derivation of Toxic Equivalency Factors for PCBs, PCDDs, PCDFs,
and Other Dipxin-like Compounds for Humans and Wildlife" held on June 15-18, 1997.
Dr. Brunstrom is also a member of the working group preparing the report, "Endocrine
Disrupting Substances: Impairment of Reproduction and Development" for the Swedish
EPA.  Dr. Brunstrom's research involves experimental studies on the reproductive
toxicity of persistent environmental contaminants in mammals and birds, and he has
approximately 65 publications related to that issue.

                                                                        Bjorn Brunstrom
  Workshop on the Application of 2.3,7.8-TCDD Toxicity Equivalency Factors (TEFs) to Fish and
                                          Wildlife •
 Pretneeting comments
 I.I. When carrying out a risk  assessment based on. a TEF/TEQ approach, it is important  to be
 aware of the limitations of such an approach and the uncertainties associated with the  fish
 and wildlife TEFs for a  certain congener. The uncertainties in the TEF values add to the other
 uncertainties in the assessment and the background information used  in the derivation of
 the TEFs is important in the  evaluation of these uncertainties. The extent of the TEF value
.uncertainties in relation to the other  assessment uncertainties partly depends on which
 congeners that are of concern.
      The TEF values provided are order of magnitude 'estimates based on the presently
 available information and  future research data will  result in revaluation of these values. TEF
 values for certain .compounds have been estimated from  a single study and relative potencies
 have generally been determined only in a few species. However, it should be remembered
 that a conservative.approach was used when deriving the TEF values. The currently available
 data  used as a basis for TEF  development and the major uncertainties in this  development
 are discussed in  the recent WHO report on proposed TEFs for mammals, birds, and fish.
      The very large interspecific differences in sensitivity to Ah receptor agonists that exist
 within animal classes contribute significantly to the  total uncertainty in a risk assessment.
 The WHO document only deals with the relative potencies of various Ah receptor agonists and
 LOAEL and NOAEL values for different species are  not discussed. Other background  documents
 give information  about LOAEL and NOAEL values and the uncertainties in these values for
 mammalian and avian wildlife.

 1.2. Since a tiered approach was used  when setting the TEFs, it is obvious  that some TEFs
 should be considered more uncertain  than others. Some  of the TEF values were estimated by

                                                                        Bjorn Brunstrom
using a QSAR model based  on enzyme induction data and these values of course are less
reliable than those based on data from a carefully conducted reproduction study.
Uncertainties appear to be largest for the least potent compounds since their TEFs are
frequently based on biochemical effects observed in-//? n'tro systems or  on estimates from
QSAR studies. Values also tend to be more uncertain for easily metabolized compounds, such
as PCB 77, since these compounds show different relative potencies in acute and subchronic
studies.                      '                                         .            •
      Concerning the  exposure routes, it  should be kept in mind that mammalian  TEFs are
mainly based on studies where the compounds were administered via the food and the
effects were related to concentrations in  the diet! In contrast, the fish and bird TEFs are
based on egg injection studies in which the effects were related to egg concentrations.

1.3. Any Ah  receptor-mediated response may,  principally, be used when determining  relative
potency values. The rationale for using a  tiered approach when developing TEF values is
nevertheless that certain endpoints are considered more useful than others. It should  also be
kept in mind that  metabolism is  largely overlooked in in n'tro assays and in acute studies.
Also, the shapes of dose-response curves in enzyme assays may  differ between congeners
which leads  to difficulties in the  interpretation.
      The most relevant compounds in the case studies  all were  designated fish TEF values
that are based on  early life stage mortality in rainbow trout. For protection of bull trout  and
lake trout these values  should be relevant. Several of the bird TEFs are  based on EROD
induction studies in chicken embryos. For these values  there are  uncertainties  associated
with the interpretation  of differently  shaped dose-response curves and also with the
extrapolation from the chicken to the bald eagle and the Caspian tern.

II.1.  When 2,3,7,8-TCDD is not the major  contributor to total TEQs, a response analysis based
on TCDD alone would significantly underestimate the impact of the chemical stressors

                                                                        Bjorn Brunstrom
present. In the prospective study, TCDD is one of the major compounds of concern and an
assessment based on a TEF/TEQ approach would only decrease the permitted TCDD toxicity
equivalent load from the effluent  a few times.
      The impact of a'PGB mixture depends on the relative concentrations of the congeners
in the mixture. Only if the relative concentrations  of different congeners were  determined in
some samples, and could be predicted to be similar across the lake, would total  PCS
determinations be sufficient. The  relative concentration of  PCB  126 seems to be crucial in
the retrospective case .study.

112. The problem with using LC50  or EC50 values for determination of relative potencies is-
that the shapes of the dose-response curves may  differ for different congeners..Th'is
primarily seems to be a problem  involving the least active  congeners.

11.3. Only few comparative studies addressing the relative potencies of various Ah receptor
agonists across species have been carried out. Whether class-specific TEFs are valid for
different wildlife  species is a matter of concern. Most  data  suggest similarities but some.
studies indicate that there may be relative potency differences across species within an
animal  class. For the species identified in the case studies, the use of the new class-specific
TEFs should give  better estimations than the .old. TEFs but extrapolations between species,
e.g., from the chicken to the bald eagle and the Caspian tern, are rather uncertain. However,
the contribution  from interspecific differences in sensitivity to TCDD to total uncertainties
may be as large  or larger as the  contribution from differences in relative potencies across

III.2. The mammalian TEFs are mainly based on food intake whereas fish and bird TEFs are
based  on egg  concentrations. When the models used predict levels in eggs of fish and birds
there is no contradiction. Uncertainties are  introduced when models describe the relationship

                                                                         Bjorn Brunstrom
between the concentrations in sediment and those in avian diet  or mammalian tissue. For
instance, the high metabolic transformation of  PCB 77 is accounted for in  the TEF value for
mammals and this means that the contribution by this compound to total TEQs will be
underestimated if its TEF value is applied for a tissue concentration. However, the
concentration of  PCB 77 appears  to be  low  in the retrospective case study.

IV. 1. I think that  uncertainties other than those associated with  the TEF values contribute to
a similar or even greater level to the total  uncertainty. Major  problems are uncertainties in
the sensitivities of the wildlife species to TCDD and uncertainties in exposure
characterization.  Also, it should be remembered that the TEF values assigned are
conservative estimates.

IV.2: In biologically-based TEQ assays, the -total effects of Ah receptor agonists  and
antagonists in a sample are measured.  Preparing  extracts from fish eggs and bird  eggs and
injecting these extracts into eggs of laboratory species gives an  opportunity to study
chemical interactions and relevant end-points.
      In certain in vitro systems, the relevant species may be studied. Disadvantages with
using in vitro systems include that they do not accurately'model all the interactions that
occur in vivo, and that the  biochemical end-points usually measured  are more or  less
connected to adverse effects.
      By combining bioassays with chemical analysis and a TEF approach, the contribution
from the analyzed congeners and from  non-analyzed compounds to total effects can be
IV.3. The highest priority should be given to  clarifying the extent of species differences in
sensitivity to TCDD for the relevant species and the basis for such differences. Are any
piscivorous bird species as sensitive as the gallinaceous birds? Sensitivities are difficult to

                                                                        Bjorn Brunstrom
determine  for relevant species but the use of  in ^//^assays and receptor studies may give
some information about those species not available for  in v/vostudies. Second, studies  of the
relative potencies of the congeners of concern in terms of various end-points should be
carried out in relevant species. For instance, the relative potency of 1,2,3,4,7,8-HxCDF in bull
trout should be examined  since the  mill effluent was predicted to contain high
concentrations of this compound.
      Both the relative potency value  of 2,3,7,8-TCDF in bull trout and the BAF of this
congener in Roundtail lake would be important information for the  prospective case study

Additional Questions Specific to the Prospective Case Study:
2.  The tentative water quality standards (WQSs) for TCDD can not be used as WQSs for TCDD
equivalents. 2,3,7,8-TCDF would be the major contributor to  the water TEQ  concentration
when using  the TEFs for mammals or  birds without consideration of the low biomagnification
factor for this congener.

3.  An uncertainty factor including uncertainties in the BAFs, BMFs, and TEFs for the different
congeners should  be considered.

Additional Questions Relative to the Retrospective Case Study:
1/2. The relatively low  potency of PCB 126 in fish (TEF value of 0.005) means that the
Caspian tern and the otter are more  likely to  be affected than the lake trout. A PCB
sediment concentration goal to protect the Caspian tern and the otter should be related to
PCB 126 as  the major contributor to  total TEQs.  If the concentration of  PCB 126 in  relation
to total PCB concentrations would be similar in sediments across the lake, then cleanup
efforts may be monitored  by  total PCB analysis.

                                                            Janet A. Burr is
                                                         Senior Health Scientist
                                                       McLaren Hart/ChemRisk
                                                  109 Jefferson Avenue - Suite D
                                                         Oak Ridge, TN 37830
                                                            Fax: 423-482-9473
Ms. Burris received her B.A. in zoology from Depauw University and her M.S.P.M. in
aquatic toxicology from the University of North Carolina at Chapel Hill.  Ms. Burris is a
senior health scientist for McLaren Hart/ChemRisk. Ms. Burris has over ten years of
experience in the successful design, management, implementation, and completion of
risk assessments for CERCLA, RCRA, and other types of hazardous waste sites. Ms.
Bum's specializes in ecological risk assessments and has completed assessments for
several federal facilities, private clients, state programs,  and the EPA.  Ms. Burris has
designed ecological risk assessment methods and guidelines for the EPA Superfund
Program and provided risk assessment training for both clients and peers.  She also
has  experience in  the design  and implementation of biological  sampling (fish,
invertebrates, plants) bioassays (groundwater, soil, and sediment), tissue analyses
(fish, invertebrates,  and plants),  toxicity  reduction evaluations (TREs), toxicity
identification evaluations (TIEs), macroinvertebrate studies, and wetlands assessments.
She is a member of the Society of Environmental Toxicology and Chemistry and the
American Society for Testing  and Materials.

                                                                           Janet Burns

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?

        Toxicity Equivalency Factors (TEFs) for mammals, fish and birds should not be
        considered to have similar uncertainties.  There is greater uncertainty in the derivation of
        some TEFs versus others and these uncertainties should be understood in the application
        of the TEFs as part of an ecological risk assessment (ERA).  The uncertainties are
        expressed in part in the tiered approach used to derive the World Health Organization
        (WHO) TEFs for fish and birds.  The tiered approach provides for preferential use of the
        more "certain" data, if available.  For example, several of the WHO TEFs for fish for
        furans and mono-ortho polychlorinated biphenyls (PCBs) are based on no testing data
        and are estimated based on structural similarity assumptions and/or Quantitative
        Structure activity relationships (QSARs). There is obviously less certainty in these TEFs
        compared to TEFs derived from LD50 data on overt toxicity in developing embryos {in
        vivo) studies.

        Uncertainties in the TEF value directly results in  associated uncertainty in the ERA. The
        amount of uncertainty should be assessed qualitatively or quantitatively in order to
        understand the influence of the uncertainty on risk assessment results. The stakeholders
        m the ERA should have an accurate understanding of the confidence in the risk
        estimates.  The greater the confidence the greater the certainty that actions will result in
        actual reduction of risks and attainment of the assessment goals.

        Probabilistic techniques could be used to examine quantitatively the uncertainties
        associated with the TEFs. Probability density functions could be used to represent TEF
        values (as well as TEQs) in place of the existing point estimates. The stakeholder would
        then have a quantitative understanding of the uncertainty. The current presentation of
        TEF values as point estimates provides the illusion that all of derived values are "equal"
        in their predictive ability of dioxin-like toxicity.

The TEF values provided were based on endpoints that ranged from in vitro biochemical
responses (e.g., induction ofcyplAl) 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?

        Certain endpoints used in the derivation of the WHO consensus TEFs may not be
        relevant to the selected assessment endpoints for the case studies and ecological risk in
        general. For example, maximum enzyme induction levels, tumor promotion, and
        increased organ weight are used as endpoints in the derivation of TEFs. However, these
        toxic effects may not have consequences on the survival, growth, development, and
        reproduction of individuals, and the sustainability of populations and communities
        (typical assessment endpoints for an ERA). Some of the toxic endpoints used to derive

                                                                          Janet Burns
       TEFs are not toxic responses but instead represent biochemical effects (binding affinity
       or induction of cytochrome P4501A) that may be in some way associated with .
       subsequent toxic responses (WHO, 1997). Other toxic effects used to derive TEFs (aryl
       hydrocarbon hydroxylase (AHH) or ethoxyresorufin o-deethylase (EROD) activity) have
       been reported to not directly correlate with toxic injury (Stegeman et al., 1992). Without
       a clear association between the toxicity endpoint used to derive the TEF and the
       assessment endpoint for a specific ERA extrapolation may either impossible or
       extremely uncertain.

       One of the primary questions that should addressed in reviewing the application of the
       TEF values to the ERA process concerns endpoints.  As with the case studies, each ERA
       will have specific assessment endpoints that reflect site-specific risk management goals.
       The WHO consensus TEF values, however, represent "fixed" toxicity endpoints.  Are
       these  TEFs appropriate for use in effects characterization for all ERAs? Are the toxic
       effects used to derive these TEFs reliable indicators of the toxic effects of concern (those
       relevant to the assessment endpoint)? Should there be some site-specific flexibility in
       the selection of TEFs for use in an ERA? Should TEFs be derived that are species-
       specific and/or endpoint-specific?

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 ifstressor response analyses were based on
total PCBs or 2,3,7,8-TCDD alone?

       The  TEF approach inherently assumes dose additivity and this is considered in the case
       studies. Possible interactions among mixtures of congeners, however are not addressed.
       The  assumptions of additivity and no interaction could result in overestimation of risks.
       Noh-dioxin like PCBs and metabolites may be antagonistic to TCDD-like response
       (Zhao et al., 1997; Biegel et al., 1989; Haake et al., 1987). PCB 153, a reported TCDD
       antagonist, is the predominant congener in the tissue and eggs of a number of avian
       species (Focardi et al., 1988; Elliott et al., 1989; Borlakogul  et al., 1990; Ormerod and
       Tyler,  1994; Van den Berg et al., 1994; and Mora, 1996).

       In other cases the assumptions of additivity may underestimate risks. Non-dioxin like
       PCBs and metabolites may induce toxic effects not addressed in the TEF (Safe, 1990 and
       McFarland and Clarke, 1989). Aryl hydrocarbon receptor (AhR) TEFs may be poor
       predictors of PCB reproductive toxicity (Battershill, 1994).

Many TEFs are based on LC50 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

       For screening level ERAs, a no adverse effect level is preferable to a median response
       value, as the goal is to identify potential risks under conservative conditions.
       Application of a toxicity equivalency approach, however, requires the use of response

                                                                             Janet Burns
        data to calculate relative potencies. TEFs derived based on median responses can still be
        used in risk assessments employing no-adverse effect levels, if the uncertainties are
        addressed quantitatively or qualitatively. The use of probabilistic methods to derive
        distributions of TEFs and/or TEQs in the ERA (in place of point estimates) could be
        used to address this uncertainty in quantitative identification of margins of "safety".

 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 TEF values provided represent the selection of the most  sensitive test species and
        endpoint. As such the TEFs may over represent risks for less sensitive ecological
        receptors. The WHO fish TEFs are based on testing of one fish species, the rainbow
        trout. Use of these TEFs to characterize potential toxicity for the fish species of concern
        in the case studies (cold water fisheries including lake trout and rainbow trout) is entirely
        appropriate due to similarity in the specific species and sensitivity.  However, in other
        applications outside of the case studies for warm water fisheries, these fish TEFs may
        not be directly applicable. Available data indicate that the relative risk of TCDD to early
        life stage survival for seven freshwater fish species are from  16 to 180 fold less than that
        for lake trout (Spehar, 1998?).  Existing information on relative toxicity could be used to
        derive interspecies extrapolation factors to predict species-specific TEFs for non cold-
        water fish species.

        The possible problems in extrapolation between WHO TEFs  for mammals and the
        specific species of interest in the case studies is difficult to discern. More information
        on the specific derivation of the WHO TEFs for mammailian species is required above
        that provided in the distributed materials.


 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?

        Exposures for risk assessment for mammals are typically expressed as oral exposures
        (dietary, water and/or sediment). These exposure routes are often not equal to the
        exposure route used to establish potency of congeners (interperitoneal injections and in-
        vitro exposures). As the exposure routes are not directly comparable between exposure
        estimate (in the risk assessment) and the TEF, resulting TEQs are not accurate and
        introduce uncertainties into the risk analyses. The potency of congeners can vary by
        exposure route (intake orally with transfer and absorption through the gastrointestinal
        tract versus direct injection into peritoneum.

 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)?

        The TEF approach requires a more rigorous and expensive analytical program compared
        to the traditional analyses of aggregate stressors (total PCBs). In a practical sense this is

                                                                           Janet Burns

        one of the more important questions in the general application of the TEF approach.  The
        data that exists for most contaminated sites is in the form of total PCB measurements.
        NPDES permit and other regulatory monitoring requirements may not traditionally
        require  congener specific analyses?


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?

        The uncertainties associated with the TEFs are primarily related to the relevancy of the
        toxicity endpoirits used to establish potency [see previous endpoint discussion]. Without
        the ability to complete an effect assessment specific to the unique assessment endpoints
        that is directly comparable to the exposure assessment data, the TEFs are more
        problematic than other uncertainties. Attaining the smallest difference in the laboratory
        (or field) measurements and the assessment endpoints (species and exposure route)
        minimizes minimizes uncertainties in the effect and exposure characterization
        (extrapolation error). Use of the TEFs limit the means and scope of assessments in
        setting forth the measurement endpoint (the toxic effect) and specifying the
        measurements of exposure that need to be performed  (egg tissue concentrations in birds
        and fish).

        Use of the TEFs also introduces uncertainty as it requires evaluation of risks for fish and
        birds based on egg tissue exposures. Prediction of egg tissue concentrations based on
        maternal exposures will often be necessary (due to analytical data constraints). This
        process is probably less certain than other established procedures to estimate oral doses
        for avion receptors.
Additional Questions Specific to the Prospective Case Study

The state adopted BAFj s used by the GL WQG.  What improvement in the accuracy of maximum
allowable concentrations for individual congeners in water, (MACt,w)ij, can be expected
through use ofBAFfdws determined from Roundtail Lake data?

What errors are associated with the state's application of the GLWQG TCDD water quality
standards for birds and mammals without consideration of congener-specific differences in
biomagnification factors from fish to tissues in wildlife relevant to the effects of concern?

How should the uncertainties associated with the available fish, avion, and mammalian TEFs be
incorporated into decisions about which TCDD water quality standard would be chosen for
setting a TEQTMDLfor regulating chemical discharges into Roundtail Lake?

                                                                            Janet Burns
Additional Questions Relative to the Retrospective Case Study

Would TEQ sediment cleanup goals be the.same for each vertebrate group? If not, -why -would
there be a difference? If the vertebrate group with the most certainty is not the group with the
most restrictive sediment cleanup goal, how would you council the risk manager's concerns for
the other vertebrate groups?
        The TEQ sediment cleanup goals would not be the same for each vertebrate group as the
        TEFs represent different sensitivities across the general classes (mammals, birds and
        fish). The TEQs for each vertebrate are also based on different exposures (oral for
        mammals and egg tissue of birds and fish) which whould result in different cleanup

        My general advice to the risk manager's concern would be somewhat practical. The
        vertebrate group with the most certainty in the risk results represents the most certain
        clean up option with the greatest chance of attaining the management goals.  Specifically
        I would substantiate recommendations with quantitative information on the uncertainties
        in the assessment including the effect of the uncertainties on risk results and clean up
        concentrations.  Clean up options for the protection of the different vertebrate classes
        would be represented geographically. A cost-benefit analysis would also be completed
        to identify for the various clean up concentration the amount of risk reduction per unit
        cost. Uncertainties would be considered in the cost-benefit analyses. The primary goal
        of risk assessment in most regulatory applications is to identify how to reduce the most
        risk for least amount of cost. This type of quantitative analyses would be used to
        demonstrate the most effective and protective options.

                                                  Steven J. Bursian, Ph.D.
                                                  Department of Animal Science
                                                             132 Anthony Hall
                                                      Michigan State University
                                                       East Lansing, Ml 48824
                                                  E-mail: bursian@pilot.msu.edu
Dr. Bursian received a B.S. in experimental biology from the University of Michigan -
Dearborn, an M.S. degree in ecology and behavioral biology from the University of
Minnesota, and a Ph.D. in physiology from North Carolina State  University. He is
currently a professor in the Animal Science Department at Michigan State University.
Dr. Bursian's current research focus is organophosphate-induced delayed neurotoxicity
and assessment of animal exposure to environmental contaminants.  Professional
memberships include the Society of Toxicology, Sigma Xi, the Society of Experimental
Biology and Medicine, the Society for Environmental Toxicology and Chemistry, and
Gamma Sigma Delta.   Dr. Bursian has authored or co-authored  more than 85
publications on the assessment of animal exposure to environmental contaminants.
He is the chair of the All University Committee on Animal Use and Care, a member of
the Department of Animal Science Graduate Committee, and the Institutional Biosafety
Subcommittee on Animals and Animal Pathogens.

                                                                Steve Bursian

1.    The additional background information which was provided enhances the
process of evaluating uncertainties. The supplementary material provides details
related to experimental design which can account for differences between studies
using the same species for determination of Lethal Dose 50 (LD50), lowest
observable adverse effect level (LOAEL), or no observable adverse effect level
(NOAEL).  These  differences may be due to the type of compound(s) administered to
a particular species [eg. commercial polychlorinated biphenyl (PCB) mixture as
compared  to weathered PCBs/dioxins/furans provided through fish collected from a
contaminated site]; the method by which the  compound is administered to the animal
(eg. injection into air cell vs. yolk; injection on day 0 vs. day 4); the endpoint(s)
which are  chosen to assess LOAELs or NOAELs; the time at which endpoints are
assessed (eg. 18 days of incubation vs. hatch); whether the NOAEL is actually
determined from  the dose-response curve or if it is estimated by dividing the LOAEL
by 10;  and differences in doses used between studies which could result in
differences in LOAELs and NOAELs.  The additional material is also helpful in terms
of assessing  differences between species in terms of LD50 values for specific
chemicals  (eg. PCB 126) that could influence  toxic equivalency factor (TEF) values.

2.    In the process of deriving a TEF based  on several studies, a number of
variables would be taken into consideration and the resulting TEF could be more
accurate than one derived from a single study. For example, if one considers the
data derived  from studies in which PCB 126 and 2,3,7,8-tetrachlorodibenzo-p-dioxin
(TCDD) have been injected into eggs of different species, a consensus TEF for PCB
126 could be established which would reflect considerations made for differences in
methodology and species.  A TEF for a particular PCB congener that has been
derived using mortality data from a cormorant egg injection study (air cell on day 4)
and the chicken LD50 value for TCDD (yolk on day 0) could be very different from a
TEF in which the same species and methodology was used for both TCDD and the
PCB congener. It was apparent from the background material that depending on  the

                                                                Steve Bursian
data chosen, TEFs for a specific congener could be different by an order of
magnitude, thus, a TEF derived from a single study introduces more uncertainty than
one which is based on a number of different studies.

3.    TEF values based on egg injection studies utilizing embryo mortality as an
endpoint would be  the most relevant in terms of the avian species to be protected,
particularly if the chicken was used as the animal model. In those situations in which
a TEF for a particular chemical has been developed using in vitro induction of
cyp1A1, for example, it could be applied to the present case studies with the
awareness that in vitro induction of cyp1A1  may occur at a different-concentration
than an  increase in embryo mortality in bald eagles. However, the variability in TEFs
derived in different studies for a particular chemical and species using similar
endpoints appears to be just as great in  many cases as the variability in TEFs for the
same chemical based on different endpoints such as in vitro enzyme induction and
embryo  mortality. Thus, it would be preferable to utilize a TEF based on the relevant
endpoint, if available. If not, a TEF based on another endpoint could be applied if it
was thought that the value  was conservative and would protect the species in


1.    If there is no dose-additivity or interaction among the components of the
mixtures described in the case studies, then each chemical would have  to be
assessed individually. The risk assessment decision would have to  be based on the
chemical judged to have the potential of causing the most harm to each targeted
species  based on its potency and its environmental concentration.  If stressor
response analyses were based on total PCBs or 2,3,7,8-TCDD alone in the
retrospective study, one could come to different conclusions concerning the risk.
The data presented in the table are from the retrospective case study and relate to
the concentration of total PCBs, TCDD,  and toxic equivalents (TEQ) detected in
Caspian tern eggs (Table 2) and otter livers (Table 3) as well as the NOAEL

                                                                Steve  Bursian
thresholds for avian eggs and mink livers (Table 4).
                            Bird Egg                    Mammalian Liver
                 Concentration      NOAEL      Concentration      NOAEL
                   Detected       Threshold       Detected        Threshold
5667 ng/gm
4.5 pg/gm
445 pg/gm
5000 ng/gm
100 pg/gm
100 ng/gm
1001 ng/gm
1 .43 pg/gm
144 pg/gm
2000 ng/gm
60 pg/gm
60 pg/gm
       In the case of the avian species, the concentration of total PCBs detected in
the egg is slightly higher than the NOAEL threshold, whereas the TCDD
concentration in the egg is 22 times lower than the NOAEL threshold.  The TEQs
present in the egg are 4.5 times greater than the NOAEL threshold. Thus, one could
conclude that an analysis based on TCDD only would suggest little risk, an analysis
based oh total PCBs would suggest a risk, and an analysis based on TEQs would
strongly support the notion that a risk exists. In the case of the otter,  an analysis
based on total PCBs might indicate  concern, since the concentration detected in the
liver is half of the NOAEL threshold based on mink studies.  The TCDD  NOAEL
threshold is 42 times higher than the concentration of TCDD detected in  the liver
suggesting that there is little risk while the TEQ concentration in the liver is twice the
NOAEL threshold. For both species, an analysis based on TEQs would  suggest that
the contaminants present in the environment are posing a risk.
       In the prospective study, the use of total PCBs would provide little protection
since the proposed paper mill is expected to generate dioxins and furans  only.  The
use of TCDD would certainly be an improvement but it would  not provide the
protection afforded by the TEQ approach.  If the relative mass concentration ratios of
each of the 7 dioxins and furans expected to be in the mill effluent are multiplied by
their respective TEF values, then TCDD contributes 26% of the total TEQs while
2,3,7,8-TCDF contributes 51 %.

                                                               Steve Bursian
2.    The NOAEL is dependent upon the doses employed in a particular study.  For
example, two studies are designed to assess the effect of PCB 126 on chick bursa
weights. One study employs doses of 0, 0.625, 1.25, 2.5, 5.0, and 10 /ug/kg egg
while a second study uses doses of 0, 1.0, and 10 M9/kg egg.  In both studies, the
bursa weight is reduced at 10 M9/kg egg but not at the next lowest dose. Thus, in
the first study the NOAEL is determined  to be 6.4 M9/kg egg while in the second
study the NOAEL is 1.0 ^g/kg egg. While the difference between the two values is
relatively small, it  would seem that considerations of the entire dose-response curve
in determining an LD50 or ED50 value is more accurate than designating the dose at
which no effect is observed  in that particular study as the true no effect level.

3.    The avian TEF for PCB 126 is 0.1 as indicated in the 1997 World Health
Organization (WHO)  report.  Egg injection studies in our laboratory which have
involved assessing the effects of PCB 126 and 2,3,7,8-TCDD in the chicken and
double-crested cormorant suggest that the PCB 126 TEF values for both species are
reasonably close to the 0.1 value.  In the chicken, the LD50 value for PCB 126 was
2.3 M9/kg egg and the LD50 value for TCDD was 0.15 M9/kg egg.  Thus, the TEF
value derived in  this study was 0.07. In the cormorant, the PCB 126 LD50  value
was  177 M9/kg egg while the TCDD LD50 value was 4.2 ^9/kg egg.  Based on these
data, the TEF for PCB  126 in the double-crested cormorant is 0.02. The consensus
avian TEF for PCB 126 and the two values established  in our laboratory are  within
the same order of  magnitude despite the marked difference in sensitivity between the
chicken and  cormorant to PCB 126 and TCDD. Thus, it would seem acceptable to
apply class-specific TEFs to  the avian.species identified.


2.    In egg injection studies, the site of injection (yolk vs. air cell) and the time of
injection (day 0 vs. day 4 of incubation)  influence the concentration of the chemical
at which effects occur. Typically, yolk injections yield a lower LD50 value than air
cell injections and  injection on day 4 of incubation precludes exposure of the embryo
during its first 96 hours of development.  However, differences in LD50 values

                                                                Steve Bursian
because of injection site are relatively small and should not prevent the use of egg
injection-derived TEFs for environmental risk assessments.  For avian species, the use
of egg injections is easier and probably more accurate than feeding the contaminant
in question to laying hens and then assessing the effects of the compound on egg
production and hatchability. Feeding contaminants to non-domesticated avian
species such as cormorants or terns would be  considerably more difficult if not
impossible, while the injection of eggs collected from relevant species is feasible.

2.     Giesy et al. (1994) summarized the advantages and disadvantages of using
the H4IIE assay for determination of TCDD-EQ as compared to the chemical
analysis/TEF approach. The bioassay is rapid and considerably less expensive than
congener-specific analysis.  Since the bioassay is a mechanistically-based
determination of an integrated biochemical response (induction of ethoxyresorufin O-
deethylase activity), it is more biologically relevant.  The bioassay accounts for
interactions between the  poly chlorinated hydrocarbons and other types of
compounds that may be present in the mixture.  In a comparison study to determine
the TEQs by instrumental and H4IIE  bioassay analysis, the bioassay determined a
higher number of TEQs in an environmental mixture when compared to the chemical
analysis/TEF approach. It was possible that components of the mixture were acting
synergistically or there were components in the mixture which were not quantified.
If feasible,  both approaches could be used and in those cases where one method
offers greater protection than the other, the risk assessment would be based on the
most conservative approach.

3.     Perform egg injection studies with the chicken as the experimental animal
(more sensitive than species of interest) to determine TEF values for each relevant
chemical.  If such studies could be done utilizing consistent techniques throughout,
some of the uncertainty associated with the derivation of the TEFs would be

                                                               Steve Bursian
      Conduct a mink reproduction trial in which animals would be exposed to
relevant concentrations of TCDD from 3 months prior to mating through weaning of
the young.  Mink are extremely sensitive to PCBs and TCDD.  While reproductive
trials utilizing commerical PCB mixtures, PCB congeners, and environmentally derived
PCBs have been run, no such trial has been conducted with TCDD.
      Conduct a feeding study with otter in which they would be fed diets
containing TCDD, specific PCB congeners, or environmentally derived PCBs.  While
otter are more difficult to work with than mink, it could be done.


                                                      Peter L. deFur, Ph.D.
                                            Environmental Stewardship Concepts
                                                      11223 Fox Meadow Drive
                                                         Richmond, VA  23233
                                                            Fax: 804-360-7935
                                                        E-mail: pldefur@igc.org
Dr. deFur received B.S. and M.A. degrees in biology from The College of William and
Mary, Williamsburg, Virginia.  He holds a Ph.D. in biology from The  University of
Calgary in Canada. Currently, Dr. deFur is affiliate associate professor at the Center
for Environmental Studies, Virginia Commonwealth University, Richmond.  He also
owns and manages Environmental Stewardship Concepts. Professional experience
includes projects for the Environmental Defense Fund in Washington, D.C. and an
investigatorship at the Smithsonian Environmental Research Center. He is active in
SETAC workshops and is currently a member of the Endocrine Disrupter Screening
and Testing Advisory Committee to the EPA. He participated in the National Research
Council's Study Committee on Risk  Characterization as well as being part of the
drafting team for a chapter of the EPA  dioxin reassessment;  Recent research includes
projects on the effects of medical waste disposal, the effects of hormone disrupting
chemicals on children's health, and a number of risk assessment programs in aquatic
ecosystems. He has numerous published articles and reports.

                                                                       P L deFur
                                                                          Page 1
Comments of Peter L. deFur on die Application of 2,3,7,8 TCDD TEF's to fish and Wildlife
November 17, 1997
General comments on the use of TEF's for wildlife and the two case studies
TEF's: The concept of TEF's and the application of TEF's is neither new nor is it entirely
novel in biological sciences. Fundamentally, there is an abundant literature from the fields
of endocrinology, toxicology, pharmacology, neurobiology and other areas of physiology
supporting the concept of equivalencies in cellular/molecular biology.  The specific
development of TEF's for TCDD is also well established, although the literature may not be
as old.  I strongly support the effort to extend the use of TEF's from rodents and humans to
fish and wildlife.  This should prove fruitful in applying basic information to environmental
control and clean-up and should provide important insight into the comparative aspects of
environmental biology.

That said, the one point that I see limiting TEF's in this context is the metabolic differences
among animals, a point also made in the present papers. The present work notes that some
Ah active compounds are not metabolized in marine mammals as in other species (e.g.
rodents) with subsequent different accumulations. If we accept the fact that the molecular
events of Ah receptor binding are common to all Ah active compounds, then mere are two
major steps where substantial interspecific differences are likely to  occur. The first is in the
cellular events following Ah binding; the second is the process by which Ah active
compounds or products of Ah activity are metabolized (either the upstream or downstream
metabolic pathways).  So far, most of the vertebrates examined for  Ah activity, excepting
marine mammals, are temperate to northern boreal animals  Few, if any show extremes in
life history (hmgfish), evolutionary development (platypus) or environmental adaptation
(e.g. desert reptiles).  I would expect to see the most remarkable differences in metabolic
processing among these types of animals, as observed in marine mammals

The use of TEF's in risk assessment or any other regulatory program or plan should pose no
more or less problematic than any other analytical tool. TEF's seem to apply to fish and
mammalian wildlife,  but have not been attempted or well demonstrated in amphibians and

Both cases are based  on well studied situations with rich databases  and numerous examples.
Data on Great Lakes  fish and on pulp and paper mills discharging TCDD are abundant in the

                                                                        P L deFur
                                                                           Page 2
EPA files and the literature.  The advantage of using such familiar types of cases will be an
easier application of the method. After mis exercise is completed and any modifications
included, I urge EPA to consider a follow-up case that draws 6n a poorly studied type of

Specific Review Questions:

I.      Stress-response
       1.  Point estimates of TEF's still should include reference to me background
           information from which the points were calculated. As the number and variety
           of applications of TEF's increases, so shall the need to consider additional
           species less similar to the ones for which the original data were developed. The
           background information should enable the users to determine the extent, if any,
           to which new applications to other species require further modification.
       2.  It is clear from the literature mat not all TEF's have the same or even similar
           levels of experimental data in the development Yet, mere seems to be no
           apparent reason why one compound should behave fundamentally different
           from those compounds for which there is a substantial database.
       3.  The TEF's for biochemical and cellular effects should be usable for whole
           organism effects. The mechanism whereby enzyme induction (or other
           molecular event) is related to whole organism effects, e.g. reproductive
           impairment, has not been elucidated in full. This information should be usable
           in the future, but should not prevent the application of TEF's now.

n.     Stress-response and Application of TEQ's
       1.  I will have to give this more consideration.  Part of the answer to mis is found hi
           the answer to V.I.
       2.  The TEF's derived at some dose to  achieve median effect (response) are clearly
           useful in the range in which they were developed, and for the effect or
           mechanism for which they were developed.  But the usefulness has not been
           challenged or tested at low doses or perhaps not at high doses for wildlife and
           aquatic life (presumably the high dose exposures in mammals and some rodents
           may confirm the applicability in this end of the range).  The low dose research
           so far has focused on enzyme induction and similar biochemical events. Has

                                                                         P L deFur
           anyone confirmed or refuted the applicability of TEF's in very low doses in
           these groups of animals?
        3.  Class-specific TEF's should be more applicable than are more general TEF's
           (e.g. vertebrates). Thus far, the experimental evidence supports the class  basis,
           even for the exceptions (marine mammals).

ffl.     Exposure Profile
        1.  To what extent does the TEF approach limit exposure analysis? The challenges
           associated with the TEF approach are those of increasing the complexity of
           exposure modeling, including fate and transport. The congeners should not be
           collapsed together in exposure models, but should be treated individually so mat
           congeners with dramatically different TEF's can be accounted for in the
           exposure, rather than assume that all congeners are similar.  The physico-
           chemical data suggest (or more) mat congeners will act differently in the
           physical environment Because these congeners would likely (or certainly) have
           quite different TEF's and hence toxic effects, their exposures should be treated
           separately. Models that do not now treat the congeners separately will not
           suffice for use in TEF specific risk assessments.
        2.  Exposure route differences used in deriving TEF's may alter the final outcome
           if (and only if?) the route of exposure alters the absorbed and tissue dose, and if
           this alteration is not accounted for in the final calculation. Efficiency of uptake
           is high in digestive tracts of most, if not all animals; this  is the primary route of
           exposure and of administration in laboratory work.  I cannot see mat this would
           be a problem in the derivation and use of TEF's.
        3.  The TEF approach will prompt or require a more detailed analysis of TCDD
           (and PCB) sources such as sediments, water and soils than is the case in which
           the congeners are aggregated. The aggregation approach simply assumes that
           all congeners in the total are the same, or mat the aggregate can be treated in a
           simple, single model approach.  The different toxicity  of each congener will
           affect the final toxicity of the mixture; hence it will be much more accurate to
           know the real mixture toxicity based on the sum of the TEF's.
IV.     Risk Characterization
        1.  Sources of uncertainty in an entire risk assessment are not so quantitatively
           predictable as to make any conclusion as to the relative contribution a priori.

                                                                       PL deFur
                                                                         Page 4
           Uncertainties in TEF's are largely from the variation in the results of
           experimental outcomes, whereas the uncertainties in a "field7"' risk assessment
1           include fate and transport, exposures, endpoint sensitivities and population
           dynamics. I doubt this can be determined on a generic basis.
        2.  Comparing the TEF approach with a site-specific TEQ analysis, there are a few
           ways to approach this question. I think the first question is why would one do
           both? Or one or the other? I imagine mat most scientists would want the
           congener specific TEF analysis, based on a congener analysis of the source.
           Given the acceptance of TEF's, and the confidence in them, as well as an
           exposure analysis mat incorporated individual congeners, a simple analysis of
           the source of contamination (air, water, sediment, etc) would provide a
           straightforward method for determining the dose to the target species. In the
           cases where the TEF's are most likely to apply tot he target (endpoint) species,
           the TEF approach would likely provide the most accurate approach. But, in
           cases where tile TEF's are not as likely applicable to the target species, men the
           total sample TEQ approach would circumvent the lack of applicability.
           The TEQ approach is likely to be more difficult
        3.        *
V.     Prospective Case
        1.  This question about improvement in BAF Hw's makes a comparison and h is not
           deartowhatmenewBAF'sarecximparedorifhistoanyah^rnative. The
           improvement at increased accuracy is in targeting for lower MAL's those
           congeners that are more accumulative.  I do not see me safety m permitting the
           relaxation or raising of MAL's for congeners with lower BAF's in a single
           species or in a class. In mis latter case, me permitting of more discharge of any
           congeners of TCDD assumes mat the congeners will not ever pose a toxichy
           problem in the lake.  If at some later time me congener with the higher MAL
           does pose a problem, men the loading over time will present future
           contamination problems. The temporal lag between discharge and effect, control
           and response is problematic for TCDD and congeners mat have such a long
           half-life in the environment.
        2.  Without using congener specific data, the state may have to treat all congeners
           in an approximately similar fashion, erring in exposure, dosnnetry and in
           toxicity.  Alternatively, the state may chose to ignore all congeners for which

           there are not she-specific data, as lias been done in the past when only TCDD
           was analyzed. Both approaches have the potential for underestimating toxicxry;
           the latter approach having been carried oat by many states for years (and likely
           still practiced). The former error could treat all lower toxitity congeners as
           more toxic, thereby overestimating the effect Unless, of coarse the state
           chooses to "average" the ttutichy along with lumping the dosimetry, thereby
           taking some sort of average toxitity to use with a total dose of TCDD's,
VI.    Retrospective  Case
       1.  I have not calculated the numeric clean-up goals for each vertebrate species.
           Seldom are the clean-up goals the same in such cases. Frequently one endpoint
           drives the clean-up because of greater BAF, greater sensitivity (more toxic in
           one species than the others), or because of a different target level in the clean-up
           goals, as is the case with some endangered species, or one for which there is a
           specific population recovery plan.

                                                  Joseph V. DePinto, Ph.D.
                                           Professor of Environmental Engineering
                                                          Great Lakes Program
                                            State University of New York at Buffalo
                                                                202 Jarvis Hall
                                                        Buffalo, NY 14260-4400
                                                            Fax: 716-645-3667
                                                 E-mail: depinto@eng.buffalo.edu
Dr. DePinto has a B.A. in physics from Miami University of Ohio, an M.S. in physics and
environmental engineering, and a Ph.D. in environmental engineering all from the
University of Notre Dame.  Dr. DePinto is a professor in the  Department of Civil,
Structural, and Environmental Engineering and the director of the Great Lakes Program
at the State University of New York at Buffalo.  One of Dr. DePinto's major research
interests is the mathematical modeling of transport and fate  of pollutants in aquatic
systems.  Other areas of interest are aquatic chemistry, lake eutrophication, aquatic
ecosystem modeling, and the application of geographic information systems to water
quality modeling. He is the author of over 150 reports and scientific publications. Dr.
DePinto is a member of the American  Chemical Society, the American Society of
Limnology and Oceanography, and the International Association  of Great  Lakes
Research. Dr. DePinto serves on several Great Lakes councils and advisory groups.

                                                                   Joseph V. DePinto
              Pre-meeting Comments in Response to Charge Questions for

   Workshop on the Application of 2,3j7,8,-TCDD Toxicity Equivalency Factors to
                              Aquatic Life and Wildlife
       Since my expertise is limited to modeling the transport, fate and bioaccumulation of
contaminants in aquatic systems, my comments in response to the pre-meeting questions will be
confined to those questions relating to exposure and risk characterization.  I will leave comments
on toxicity questions to those with much more expertise in that area.
1.      To what extent does the TEF approach present challenges, introduce new uncertainties,
       or modify' old uncertainties associated with modeling the exposure ofAhR agonoists? To
       •what extent does the availability and quality of congener-specific phyisco-chemical data
       imit the means of employing fate and transport or food chain models?

       Until the Green Bay Mass Balance Study and the modeling work conducted in that
study, fate and transport models of hydrophobic organic chemicals (HOCs) were not applied to
specific congeners. Having been one of the modeling team working on that project, I can state
that one of the significant outcomes of that study was that "once an accurate model  for the
dynamics of sorbents (solids and non-settleable organic material) hi a specific system has been
developed, we have enough knowledge of and appropriate formulations for the transport and fate
of HOCs in surface waters that we can accurately model the concentrations in water and
sediments of specific congeners of these compounds merely by having good congener-specific
physico-chemical data (e.g., K^, H,, biotic and abiotic degradation rates)." This result was
demonstrated for PCBs in the Green Bay study by successfully modeling PCB congeners
spanning a wide range of hydrophobicity and volatility using the same sorbent dynamics and
only changing the respective physico-chemical properties. This development has made it
possible to use the TEQ approach as proposed in the Prospective Case Study. However, as the
statement above suggests, we will be limited in this approach by the availability of accurate
congener-specific physico-chemical parameters. As indicated in Table 1 of the Charge
Questions, there are gaps in these data and potentially order of magnitude or more uncertainties
in some of the properties estimated for the more hydrophobic congeners.  In my opinion,
considerably more work needs to be done  in measuring or calculating (based on structure-
activity models) these chemical properties.

                                                                    Joseph V. DePinto
       The status of bioaccumulation model is slightly more problematic because of the
uncertainties of in model formulation and parameterization of food chain bioenergetics and
predator-prey dynamics and because of the large system-to-system variability of these ecosystem
dynamics. Measurements of BAFs and BSAFs on a congener-specific basis can obviate the need
for the more mechanistic food chain bioaccumulation models, but extrapolation of site-specific
measurements carries with it a significant uncertainty in terms of two things: 1) a different
system with a different food web will exhibit a different BAF or BSAF; and 2) the measurement
in a given system is representative of a specific point in time and there may be a lag between a
change in the concentration in the water column or sediments and the response of the
concentration up the food chain (i.e., the measurement may not have been at bioaccumulation

2.     To what extent do exposure route differences used in deriving the TEFs affect their
       application in the case studies?
       This could be problematic because of the fact that we tend to see a decrease in BAF as a
function of log K^ for the super-lipophilic congeners (log K^ z 6.5).  It is not known whether
this is because the congener is so insoluble that it cannot transport as effectively across the gut
wall or whether the kinetics of the bioaccumulation process is so slow that the organism  cannot
respond to a given exposure-level in a reasonable length of time.  In any event, there is likely to
be a big difference between the BAF for one of these compounds if .the exposure is via food
intake versus direct injection.

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?
       If I understand the  approach, projecting a AhR-based toxicity in fish or wildlife based on
will require a chemical measurement of the concentration of each  congener in each stressor
source. In other words, if contaminated sediments are the source of toxicity, then the initial
concentration of each relevant congener will have to be quantified. Of course, with current
analytical methods, measurements like total PCBs are actually made by appropriately summing
individual congener concentrations. This may be somewhat problematic using EPA's accepted
standard method because detection limits for individual congeners are often too high to give an
accurate TEQ for comparison against a effects standard.

                                                                    Joseph V. DePinto
1.      In evaluating the case studies, do uncertainties in the TEFs limit the assessment or are
        other aspects of effect and exposure characterization contributing similar or greater
        levels of uncertainty?
        It is clear to me that uncertainties TEFs will increase the overall uncertainty of a risk
assessment, simply because we are propagating more error through the calculation as we
increase the number of parameters to specify. However, one must weigh uncertainty against the
information obtained — or utility — in a given calculation.  In my opinion, there is a potential
to gain much more information using the TEF approach; therefore, it is worth using even though
the error might be somewhat higher. I feel confident that, over time, experience with the
approach and more empirical data will reduce the uncertainty.

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?
        With regard to exposure modeling, if the biologically-based TEQ assays approach were
used, we would have to develop models for the fate and transport of whatever it was that the
TEQ assay was measuring.  If we did this we would not only find it virtually impossible to
parameterize such a model, but we would have no way of guaranteeing that a TEQ assay level at
a source would be transported and transformed through the aquatic system in such a way that
made it directly comparable to the same assay conducted on the receptor (some fish, bird or
mammal).  In other words, TEQ for multiple sources (including background sources) would not
necessarily be additive at the receptor.  Put another way, modeling the fate and transport of
"toxicity" as a single constituent is fraught with problems and uncertainties that may indeed
exceed the errors introduced by making the analysis more complex by using the TEF approach.

3.      Provide a list of specific research or site-specific data that would improve the analyses
        in the case studies.
        In my opinion,  there are three primary areas of uncertainty associated with the type of
regulatory analysis described in both case studies:
1.  Quantifying the exposure distribution in water and sediments of the system of interest as a
    function of the various sources.
2.  Confirming that BAF or BSAF measurements in one system are applicable to another;  in

                                                                     Joseph V. DePinto
    other words, understanding the ecosystem factors that control bioaccumulation and hence
    these measurements.
 3.  Reducing uncertainty in TEF values by building an empirical database over time.
 All three of these research/data acquisition areas are very important in my opinion. Suffice it to
 say that application of the TEF-based for risk management requires continued research
 associated with its application in order to build a better experience and knowledge base.

 Additional Questions Specific to the Prospective Case Study
 1.  The state adopted BAf^^. used by the GLWQG. What improvement in the accuracy of
    maximum allowable concentrations for individual congeners in water can be expected
    through use ofBAFfdv determined from Roundtail Lake data?
        The GLWQG BAFs were determined largely using Lake Ontario measurements using
 lake trout. Measurement error and time-variability aside, these BAFs could easily vary by 1-2
 orders of magnitude between Lake Ontario and another system with a different food web (e.g.,
 more benthic versus pelagic or having a different number of trophic levels). To the extent that
 the food webs of Lake Ontario and Roundtail Lake are similar, I would not expect to gain much
 accuracy in measuring site-specific BAFs  in Roundtail Lake. However, if the food webs are
 significantly different or if there had been  a significant perturbation (e.g., one or more very bad
 recruiting years) in one of the trophic levels (such the prey fish), then site-specific measurements
 would certainly be advisable.

 4.  What errors are associated with application ofTCDD water quality standards for birds and
    mammals without consideration of congener-specific differences in biomagnification factors
    from fish tissues?                                                         .
        If I understand this question correctly, there is potentially significant error involved in
 not accounting for BMF from fish to wildlife. The BMF could be significant on a congener-
 specific basis, which in turn might have a  significant effect on the standard. Good data on the
 fraction of the wildlife diet coming from fish and the age and size offish in their diet is crucial
 and perhaps an even greater source of error in many cases.
, 5.  How should uncertainties associated with the available fish and wildlife TEFs be
    incorporated into decisions for setting TeqTMDL?
     ,   We must attempt to quantify the uncertainty in TEFs for each target group and then

                                                                    Joseph V. DePinto
propagate that error through the calculation of TeqTMDL for each.  Then the actual TMDL
allocated to the discharger should be based on the target group yielding the allowable loading
that is statistically lowest without making a Type I error.

Additional Questions Relative to the Retrospective Case Study
1. Would the TEQ sediment cleanup goals be the same for each vertebrate group? If not, why
would there be a difference? How would you handle a situation in which the group with the most
certainty is not the group with the most restrictive sediment cleanup goal?

       I would think that sediment cleanup goals would vary from group to group; because,
even though the source may be the same and may have the same congener distribution for each
group, the pathway to each group is very likely to be different and each trophic level in those
pathways may be subject to different exposure distributions and may bioaccumulate and
metabolize that exposure differently. Therefore, I would not be surprised at all that computing a
cleanup goal based on a TEQ would yield different values.
       Refer to my response to the last question for my opinion on how to handle a situation
where unequal certainty exists among groups.

2.  Would the TEF/TEQ-based sediment remediation goal be the same as those determined for
    total PCBsfor the identical vertebrate class? Assume that a simple ratio of total PCB
    sediment concentration goal to TEQ sediment concentration goal was formulated to allow
   for the use of total PCBs to monitor cleanup. What exposure and effect issues would need to
    be evaluated before using the less costly total PCB analysis to support the TEQ-based
    sediment remediation goal?
       This is an excellent question, but it is difficult to answer without going through
significant calculation and modeling. But most importantly, the concentration of PCBs (and
other chemicals of interest) in the sediments of Yuckymuck River have not been measured (or at
least not specified); therefore, it is impossible to know if the two goals will differ. But given the
total PCBs in the lake surface sediments are 110 ppb, I would venture an educated guess that the
PCBs in the river sediments are still well over 1 ppm. Therefore, a goal based simply on getting
total PCBs down below 1 ppm would probably require removal of more sediment than the TEQ-
based remediation goal.
       Using a simple ratio of goals to permit measuring total PCBs as a means of monitoring
cleanup progress is fraught with error. I can think of no in-place or removal-treat-and-replace
sediment remediation process that would not be congener-specific in its removal efficiency.

                                                                   Joseph V. DePinto
Even using a simple dredging and disposal approach would not necessarily work.  This is
because the congener distributions of PCBs and PCDFs would no doubt change with depth in the
sediments; therefore, a goal and ratio based on surface sediments would not necessarily be
constant through the full treatment depth of the sediments.  If the spill occurred 30 years ago and
loss is by burial, the spill chemicals may have penetrated quite deep into the sediments.

       I am strongly in favor of applying the TEF/TEQ approach for risk management offish
and wildlife; however,  we must move forward by maintaining a concurrent research and data
acquisition program that will allow us to continue to reduce uncertainty in decision-making.

                                                                         Joseph V. DePinto
Gong, Yuyang, J.V. DePinto, G-Y. Rhee, X. Liu. 1997. Desorption rates of two PCB congeners from
    suspended sediments: (I) Experimental results. In press: Water Research (October, 1997).
Gong, Yuyang and J.V. DePinto. 1997. Desorption rates of two PCB congeners from suspended sediments:
    (II) Model simulation.  In press: Water Research (October, 1997).
Velleux, M, S. Burns, J.V. DePinto, and J.P. Hassett. 1995. A  screening-level mass balance analysis of
    mirex transport and fate in the Oswego River. J. Great Lakes Res. 21 (1 ):95-111.
Cheng, C-Y., J.F. Atkinson, and J.V. DePinto. 1995. Desorption during resuspension events: kinetic vs.
    equilibrium model. Australian Journal of Marine and Freshwater Research. 46:251 -256.
DePinto, J.V., M. Morgante, J. Zaraszczak, T. Bajak, and J.F. Atkinson. "Application of Mass Balance
    Modeling to Assess Remediation Options for the Buffalo River (ARCS/RAM Program)."  Final
    Techincal Report for Cooperative Agreement CR-X995915 to U.S. EPA, Great Lakes National
    Program Office, Chicago, IL. Report No. EPA 905-R95^007. (April, 1995).
Bierman, V.J. Jr., J.V. DePinto, T.C. Young, P.W. Rodgers, S.C. Martin, and R.Raghunathan.  1992.
    Development and validation of an integrated exposure model for toxic chemicals in Green Bay, Lake
    Michigan.  Final Report for EPA Cooperative Agreement CR-814885, ERL-Duluth, Large Lakes and
    Rivers Research Branch, Grosse He, MI, 350 pp. (September, 1992).
DePinto, J.V., T. Fiest, R. Raghunathan, D. Smith.  1997. Analysis of Toxaphene Behavior in the Great
    Lakes. Organochlorine Compounds, 33:280-284 (Proceedings  of 17th International Symposium on
    chlorinated dioxins and related compounds).
DePinto. J.V.. R. Raghunathan, P. Sierzenga, X. Zhang, V.J. Bierman, Jr., P.W. Rodgers, T.C. Young,
    "Recalibration of GBTOX: An Integrated Exposure Model for Toxic Chemicals in Green Bay, Lake
    Michigan." Final Technical Report to U.S. EPA, Large Lakes and Rivers Research Branch, Grosse lie,
    MI, March 1, 1994.
DePinto, J. V.,  R. Raghunathan, V.J. Bierman, Jr., P.W. Rodgers, T.C. Young, and S.C. Martin,  1993.
    Analysis of organic carbon sediment-water exchange in Green  Bay, Lake Michigan. Water Science
    and Technology, 28(8-9): 149-159.

                                                     Lev R. Ginzburg, Ph.D.
                                             Department of Ecology and Evolution
                                         State University of New York, Stony Brook
                                                           11 Crane Neck Road
                                                        Stony Brook, NY 11794
                                                        E-mail: lev@ramas.com
Dr. Ginzburg is a professor of ecology at the State University of New York, Stony Brook
since 1977 and the president of Applied Biomathematics since 1982.  His expertise
focuses on risk analysis and population  modeling.  He currently has the following
articles in press: Inertial growth: population dynamics based on maternal effects and
Assymmetry of population cycles: abundance - growth representation of hidden causes
of ecological dynamics.  Dr. Ginzburg has had over 100 articles and 5 books published
during his career.

                                                                            Lev Ginzburg

0. General and preliminary comments
The scope of the workshop's discussion is confined by the organizers to considerations of direct
effects only.  While perhaps useful in limiting the scope of the discourse for the sake of
manageability, this seems dangerously restrictive in the context of ecological risk assessments.
After all, indirect effects such as trophic cascades certainly do play a very important role in the
ecotoxicology of TCDD and its congeners. Perhaps this restriction should be relaxed somewhat.

Like many, I have serious and strong reservations about the use of the "hypothesis-testing"    o
approach in environmental risk assessment and management, including use of the hazard
quotient, no observed effects levels, and their ilk. The conceptual difficulties with EPA's
approach are many and have been widely discussed (e.g., Barnthouse et al. 1986; Landis and Yu
1995; inter alia).  Whether or not to regulate or remediate should be framed as a decision
problem, not a hypothesis testing problem.  Much of the use of TEFs (toxicity equivalency
factors) has heretofore been embedded in hypothesis-testing approaches which I find barely
intelligible.  It is heartening, however, that the TEFs should be of use beyond the rarefied  context
of hazard quotients. I think it will be important for the workshop discussion to consider how
TEFs will continue to be  useful when hazard quotients are replaced by probabilistic methods of
decision analysis.

n.3. Extrapolating class-specific information  to particular species
Although one might hope that TEFs will provide a means of freely translating toxicity
information within the big matrix of chemical congeners and biological species, there appears to
be a considerable amount of interspecific variability in toxicity of TCDD itself (and presumably
this cannot be erased in the TEF method). It might be very interesting to explore the available
information about TEFs for the existence of allometries in which this residual variation might be
at least partially explained by a species' body size, typical egg size, or other easily measured
species-level variable. If even crude allometric relationships exist, they may be very useful in
making the TEF method more accurate with little additional effort.

IV.l. Are uncertainties of TEFs more problematic?
I doubt that the uncertainty about TEFs is any more problematic than that of the other sundry
inputs to a quantitative risk characterization. The magnitudes of these uncertainties may be

                                                                            Lev Ginzburg

fairly similar to those we've seen in other inputs, and even if they're considerably bigger, it
shouldn't necessarily lead to any fundamental incompatibility. It is important to understand,
however, that the uncertainty in TEFs will likely primarily be lack of knowledge (i.e., incertitude
or ignorance), rather than variability.  We have argued that it may be necessary to use different
uncertainty propagation techniques to handle this kind of uncertainty (Person and Ginzburg
1996). In particular, the indiscriminant application of Monte Carlo techniques in this case can
lead to erroneous conclusions that underestimate the risks involved.

The task of identifying, and quantifying, the uncertainties associated with TEFs belongs
primarily to the empiricists who collect the original toxicity data and the synthesizers who
collate this information and compute the TEF values. The former must report their measurement
errors in full detail; the latter must propagate these uncertainties using appropriate techniques.
Reviewers can help by checking that the results seem reasonable and by guessing at what
possible mistakes or omissions may occur, but they cannot be expected to develop
characterizations of uncertainty if the requisite underlying details are missing.

IV.3. Further empirical investigations for the case studies
Most of the  documents focus on effects on juvenile survivorship. Are there known to be no
effects from common environmental concentrations of TCDD etc. on other demographically
important variables? Possibilities include time to reproductive maturity, onset of adult
senescence,  growth rate, reproductive investment, among others. Since the toxicological effects
are believed to be additive, I would supposed they are likely to also be cumulative in time with
iterated exposures.  Thus one might expect to see effects in later life stages. Unless it's clear that
no effects on such variables are possible (via the Ah receptor mechanism or otherwise), I think it
would be very important that further specific empirical and synthetic studies be conducted to
extend the TEF method to such variables. It seems doubtful that a TEF value for one life stage is
really general for all life stages.

Often a biochemical response (e.g., induction of cyplAl) is observed in lieu of measuring effects
on juvenile survivorship. It is harder and harder to justify regulations based merely on
measurable  biochemical effects in non-human species.  Unless this biochemical effect has an
obvious and direct consequence on some population-level vital rate (reproduction, mortality,

                                                                          Lev Ginzburg

growth rates), or perhaps on some organismal-level variable related to individual health of
humans or a listed endangered species, we should expect to encounter "so what?" questions from
the regulated communities and the public.

Extra Note:
The word 'congener', like 'species', has both a meaning in chemistry and another meaning in
biology.  The documents are consequently rather confusing.

Person, S. and L.R. Ginzburg. 1996. Different methods are needed to propagate ignorance and
       variability. Reliability Engineering and Systems Safety 54:133-144.
Landis, W.G. and M.-H. Yu.  1995. Introduction to Environmental Toxicology. Lewis Publishers,
       Boca Raton.
Bamthouse, L.W., G.W. Suter II, S.M. Bartell, J.J. Beauchamp, R.H. Gardner, E. Linder, R.V.
       O'Neill and A.E. Rosen. 1986. Users's Manual for Ecological Risk Assessment. Oak
       Ridge National Laboratory, ORNL-6251. National Technical Information Service,
       Springfield, Virginia.

                                                       Jay W. Gooch, Ph.D.
                                                               Senior Scientist
                                               The Procter and Gamble Company
                                               Professional & Regulatory Services
                                                        Paper Products Division
                                                     Winton Hill Technical Center
                                                        6100 Center Hill Avenue
                                                     Cincinnati, OH 45224-1788
                                                       E-mail: gooch.jw@pg.com
Dr. Gooch received both his B.S. and M.S. degrees in fisheries and wildlife, and a
Ph.D. in environmental toxicology, all from Michigan State University. He is currently
a toxicologist at The Procter and Gamble Company, Cincinnati, as well as adjunct
assistant professor in the Zoology Department at Miami University in Ohio. Within the
past 11 years, he has held positions with the Chesapeake Biological Laboratory in
Maryland and at the Woods Hole Oceanographic Institution in Massachusetts. His
expertise for the task at hand, toxic equivalency factors, is in toxicology, especially of
fish in marine and estuarine settings. He has received the SEA Grant and National
Fisheries Institute Award for Outstanding Applied Marine Research, and he is active in
three professional societies: the Society of Environmental Toxicology and Chemistry
(SETAC), the American Chemical Society, and the International Society for the Study
of Xenobiotics. Dr. Gooch is an ad hoc reviewer for 10 publications and has organized
and chaired professional meetings for SETAC. Pertinent publications by Dr. Gooch are
concerned  with environmental fate and transport,  PCBs, and environmental risk

                                                                 Jay W. Gooch

Responses to General Questions

Because I often had difficulty understanding exactly what was being asked in some of
these questions, my responses below contain my paraphrase of the question prior to the


1.   I was very perplexed by this question. Was there any information in the descriptions
of the case studies which was useful for reducing the uncertainties in the derivation of
the WHO TEFs and their application to each particular assessment?

No. I  believe the greatest uncertainties in the application of the WHO TEF values are in
the reliability of the extrapolations to other untested species, and extrapolations to
endpoints biologically or biochemically distant from the endpoints used to derive the

2.  Should all TEFs be considered to have similar uncertainties?  Obviously, no. The
uncertainties associated with each of the TEFs have multiple sources. As referred to
above, these are most importantly the cross species extrapolations, and the cross-
endpoint extrapolations.  As pointed in several of the documents provided, the
uncertainties in the application of any given point estimate of a TEF increase the more
distant the endpoint on which the TEF is based is from the endpoint of interest in the
assessment. I believe this is nothing  more than common sense and toxicology 101.
There are several examples which exemplify these  uncertainties; as eloquently pointed
out in document 6F by Cook, et al..  In Table 2 of that document TEFs are listed across
endpoints. By looking at the TEFs for a given chemical (across endpoints), it is
apparent that the relative difference in the TEFs is not the same across chemicals (see

                                                                  Jay WiGooch
Table 1.
                                            TEF Ratio
                    In vivo RET liver EROD/            In vitro RBT liver EROD/
                    RBT ELS Egg Mortality	RBT/ELS Egg Mortality
PCB 126
PCB 77


Looking down the row of each ratio, it is notable that relative difference in the TEFs
derived from the biochemical endpoint of EROD induction to the more ecologically
relevant endpoint of ELS egg mortality is clearly non constant across the chemicals and
spans a range of nearly  200 fold. Also of interest is the lack of concordance in the
relative ratio for any given chemical (i.e.  looking across the rows).  As pointed out by
Cook et. al., the difference, hence the best example of uncertainty, is for PCB congener

Since uncertainty is also a general function of the information richness of the data set,
the trout data serves as the basis for another point.  In general, the rainbow trout and
lake trout data sets are the most information rich data sets we have for formulating and
evaluating TEFs.  And this data reveals the uncertainties above. It is very difficult to
say whether these same observations about uncertainty would hold if the data set were
on another species entirely, for example with a bird species. In other words, it is
difficult to say if the differences would be greater or less?

                                                                  Jay W. Gooch

In addition to this source of uncertainty, there is also uncertainty derived from cross
species extrapolations. For example, the LC50/NOEC values derived from the data of
Spehar et al. are as follows.
Fathead Minnow
Channel Catfish
Lake Herring
White Sucker
Northern Pike
LC50/NOEC Ratio
                          Avg   ~3
Note that the NOEC values included non-lethality (non-acute) measures such as growth.
Also the range (estimated) of egg LC50 values is 35 fold (lake trout to zebrafish).

Clearly, there are significant uncertainties in the risk benchmark  values for any given
species.  Would this same pattern hold true for the other PCDDs and PCDFs of
interest? Would the magnifude of the differences among species be more or less?
3.  To what extent can the TEFs be extrapolated to the measures of effect that are
relevant to the assessment endpoints?  The best case, i.e. the one with the least
uncertainty, is the case where the extrapolation is biologically proximate—as is the case
when the risk endpoint is focused on early life stage mortality and the TEFs are based
on the same endpoint. So, in both cases, the fish assessments are on the firmest
footing.  The bird assessments, because the TEFs are based primarily on biochemical
endpoints, are the least certain, and potentially the most conservative—if the lessons
from the fish data set are ultimately applicable.

                                                                 Jay W. Gooch


1. What if we didn't do a total TEQ approach? Quite simply, you would miss important
information in the estimation of the overall risk. It is clear from all of the information
provided, and the other literature in this area, that a total TEQ approach to the risk
assessment is a rationale one. I agree that additivity at the cellular level is a reasonable
assumption for these chemicals.

2. Are we erring in our assessments by using TEFs based on median response levels?
Probably not.  The toxicology 101 answer to this question is "Not as long as the slopes
of the dose response curves and the magnitudes for the range of the responses are
similar across the chemicals of interest. From the data I've seen so far, it appears as
though, at least forTCDD, the slopes of the various dose-response curves are generally
similar (cf. The Spehar et. al. data).   It also turns out that for this particular  class of
chemicals the dose response curves are quite steep, with ratios of EC50 values to
NOEC values on the order of 2-3X. The downside to this type of dose'response curve is
that in the effective range small changes in exposures can result in large changes in
levels of effect. The positive aspect is that below this narrow range, no measurable
effects are likely.

3. To what extent can class-specific TEFs be extrapolated within each case study? As
per the data in my answer to question 2 of Section I above, there are indeed significant
uncertainties associated with applications across biological endpoints or species.   For
the cases provided, these uncertainties are probably less problematic because it is
evident from the datasets that we are likely to have TEF values and toxicologic
benchmark values for species that are the most sensitive. For example, because the
TEFs for fish are derived from the .lake trout  and rainbow trout datasets, and because all
of the other data on toxicity of TCDD suggests that these are the most sensitive species,
one can have confidence that extrapolating across all other species of fish is not under
conservative (i.e under protective).

                                                           Jay W. Gooch

I would be willing to make similar conclusions for estimates of risk to wildife which are
derived from TEFs based on mink. I think the extrapolation with.the least certainty is
that with birds.


1.  Does the TEF/TEQ approach make it more difficult to assess "exposure" than if the
assessments were focused on one chemical alone?  Clearly yes, if only for the simple
reason that this approach deals with the aggregate uncertainty associated with dealing
with a larger number of chemicals. While we have built a fairly reasonable data set on
which to describe (or model) the likely fate profile of TCDD and some of the PCBs, it is
clear that we do not have the same level of understanding for all of the compounds
which are included in these TEQ calculations. We have what I would consider
"adequate" physico-chemical data on which to estimate fate and trophic level exposure if
we are willing to assume little or no biodegradation through metabolism at any (every)
trophic level.  However, we clearly know this is not the case for many of these
chemicals. The conservative modeling  approach is to assume no appreciable  losses at
any given level  of the food  chain.  The application of BSAFs and BAFs derived from the
literature are an improvement in realism, but are still subject to uncertainty depending on
the difference between the trophic structure of the system on which the BSAFs/BAFs
are derived versus the system in which  these factors are being applied.  As pointed out
in the GLI, the strongest  application is where site specific BSAFs and BAFs are

In addition to the complexity of estimating general exposure to a large suite of
chemicals, there is also the uncertainty  introduced by assuming that the internal kinetics
and dynamics of these chemicals are the same among species (and across the
chemicals).  For example, there is data  to suggest that the ratio of the whole fish or
muscle tissue concentration to the estimated gonadal tissue concentration of TCDD in
trout is approximately 3.  Do we really know how valid this assumption is for all the
other Ahr agonists?

                                                           Jay W. Gooch

2. How much uncertainty is introduced because TEFs are often derived from exposure
routes that do not simulate realistic exposure and tissue deposition? Given all the other
areas of uncertainty and other data gaps, I don't believe this is an area of great concern.
While I have often wondered about whether slow accumulation over long periods of time
(for example over the lifespan  of a lake trout) leads to significant tissue pools (or
"compartments") which are not bioavailable and effectively "sequestered", I haven't seen
any data which address this.

3. Does the TEF approach require that the analytical data be  more rigorous than with
aggregated measures like total PCBs? Clearly the TEF approach requires very specific
data, and the concentrations of many of the analytes are often close to the limits of
detection.  In addition, in most cases relative amounts of the various analytes varies
over several decades of concentration. And finally, the TEFs themselves vary over
several orders of magnitude. The net  effect is that the very  low concentration, high
potency, analytes generally contribute most to the TEQ calculation. Because these low
concentration analytes are often the least certain from a general analytical standpoint, a
more rigorous analytical design is generally necessary.  All of the standard QA/QC rigor
associated with stable isotope spiking, blanks, etc. etc.  become paramount.

1. Are the TEFs the dominant source  of uncertainty in these assessments? I'm not
expert in uncertainty analyses, but would guess that the uncertainties in the TEFs are of
similar magnitude to the other uncertainties.

2. Should cellular assays of TEQ content in extracts be used to make these
assessments? These assays  can provide valuable data, particularly for screening
purposes.  They  have high throughput, are standardizable, and are relatively simple.
However,  because they are subject to potential interferences (depending on how
"refined" the extract is), they are best utilized as an exploratory tool. When a full and
specific assessment is required, the specific analytical data  should be used.

                                                          Jay W. Gooch


1.   Will site specific BAFWif
                                                            Jay W. Gooch

3. I believe the assessment makes a prudent choice in selecting TEqTMDL estimated
from the fish data.  As stated in my response to an earlier question, the best data set we
have for relative TEFs is the fish data set. The bird data set is not sufficiently robust yet.
Since we know very little about the River Otter, and it's clear that there is a very large
degree of difference among the mammals in sensitivity to dioxins, it is difficult to say
exactly just how conservative (or overly conservative) the numbers might be.  While I do
not dispute the logic and the scientific underpinnings to the calculations that have been
used to derive the wildlife values of approximately 3 fg/l, I remain eager to see an
example where this calculation has been supported by field data.  It is difficult to
imagine that it is not significantly overly conservative.

One additional comment on the Prospective study. I  believe it is appropriate to consider
the form in which the allowable TEqTMDL enters Roundtail Lake via the discharge.  If
we assume that much of the measured mass loading that is contained in an effluent
(particularly a pulp mill effluent) enters in a form that is largely already associated with
organic material, have these assessments adequately attempted to account for the'
possibility that most of the mass of the material will never become freely dissolved or
otherwise bioavailable. The models that are used generally assume that whatever is
discharged is all discharged in a freely dissolved form, is instantaneously well mixed
throughout the system, and then partitions to equilibrium (again instantaneously) based
on affinities for organic carbon (living or dead). I contend that while this  approach is  in
many ways "necessary", it is likely a very conservative one from an exposure
standpoint. Since there are so many other uncertainties, should this source of
conservatism (i.e uncertainty) also be articulated and dealt with? Is it reasonable to
propose that a significant amount of the mass flux of material that would be associated
with a pulp mill  never partitions to a bioavailable form?

                                                           Jay W. Gooch

I think the 503 regulations for sludge application to land are a good example of an
attempt to deal with estimating allowable loadings taking into account the form and
availability of chemicals, in this case metals, as they enter the environment. The
analyses that did not take this into account, i.e. the ones that tried to establish
acceptable loadings based on total metal concentrations, ended up producing estimates
of acceptable loading that were low and impractical.  My point here is that the form of
entry of the chemical, particularly ones that are poorly soluble and carbon reactive, can
potentially be an important factor to take into account when trying to estimate
"acceptable" loadings.


1. Pardon my sarcasm, but is this a rhetorical question? The answer has direct
parallels with the calculated WQS' in the prospective study.  In that example, the values
were different for the three vertebrates groups, the fish, the birds and the bald eagles.
This is  expected because of the input values that go into the derivation; the variable
TEFs and the variable BAFs,  BSAFs and FCMs for each group. The common
convention for risk managers is to use the lowest, ostensibly most restrictive, value.
However, this is generally judged against the amount and quality of the input data
relative to the estimated risk and protection goals. In the case of the WQS derived to
protect the otter in the prospective study, I assume from the information given that it was
considered prudent, given the very low number, to get additional data to better
understand the exposure of otters in that system.

                                                          Jay W. Gooch

2.  No, because the TEF/TEQ based sediment goals are based on chemicals which
have a potency/unit mass (or mole) which is much higher than for total PCBs. I interpret
the rest of this question to be "Can total PCBs, or any other co-occuring contaminant
with similar properties,  be used as a proxy measure for TEQs for determining the
progress of a remediation attempt? Unless you believe that the PCDDs, PCDFs and
coplanar PCBs will behave differentially to the bulk PCBs during the clean-up process,
such that the ratio of these compounds to the bulk total will change,  then I can't think of
a reason why you couldn't use totals to monitor.  I would  presume that a confirmatory
analysis would  be done to confirm that the TEQ clean-up goal had been met.


                   -,    •
                   * -SARA 3£3  ECOLOGICAL FACT SHEET

                 U.S.  Environmental Protection Agency
                 ^ „„.*-. vpff ice  of Toxic Substances
                 r->-'*>' **•'
       Acetaldehyde (CAS,,  No.  75-07-0)  is  a flammable liquid with  a
       characteristic pungent odor.   It is.• .used to make paraldehyde,
       acetic  acid,  butanol,   perfumes,   flavors,   aniline= dyes,
       plastics,  and synthetic rubber.   It is  also ;used; in silvering
       mirrors and  in hardening  gelatin  fibers.   It ^ari; enter  the,,
       environment through manufacuturing  effluerits/prlspilis«"//.>

   ACUTE fSHORT-TEpIp  ECOLOGICAL EFFECTS f^, •       -—S,C '   :  ,   -,' .^ '

       Acute toxic effects  may include'''the;death  of aftiinais:, birds•'£
       or fish,  and  death or low  growth 'rate  in, plants.   Acute
       effects are  seen  two  to  four days after  animals  or plants
       come  in contact with a toxic chemical:substance.       .T".-

       Acetaldehyde has high  acute toxicity to  aofuatiq-life. : ;f€ill
.;       in plant  info.]   No-data are  available  on  t^e
\'•••':     effects of acetaldehyde to terrestr/iai-aninials.
•}'     •••••'      • i'   '""''•       ''•'*.••.   •"'  '?•''..'...-"' '.-."-:'.'   '-•'-.•.'"".'  ;.•"-..  .    V •/.
;  CHRONIC (LONG-TERM) ECOLOGICAL EFFECTS   ^                t     :

       Chronic   toxic  effects   may  include   shortened  lifespan,
       reproductive  problems,   lower  fer1t|.lity,   and  changes  in
       appearance or  behavior.   Chronic  effects  can be  seen long
    ,?r:  after first exposure (s) to 'a* toxic? chemical.  "-          "-.. -

       Acetaldehyde has high  chronic toxicity to aquatic  life.   No,J
      .data  are  available on the  long-term effects  of acetaldehyde
   , .  , to plants, birds; or land animals.      "I   ' •£
*"(.'.                        '              .*'-'•-''      '
   WATER SOLUBILITY     ;    .'   . ":"'' .'  V.- '•': " • "'-'••.'. '   '-..;/:  "    ' ;- '',-:• •^•:.\,

       Acetaldehyde is highly soluble in water.   Concentrations  of
       1,000 milligrams and more will••.jnix,,^w^th»;a; liter;pf water.

      •Acetaldehyde is modektely persis^tvjji  water,  with a half-
       life of between 2  to::20^ days^;' .Tn^-'h'^fef-life  of a pollutant
       is  the amount  of  time it  takes  for one-half  of  the chemical
       to  be degraded.   About  73% of  acetaldehyde  will eventually
       end up in air;  the rest will-end  up in the water. : "


    Some substances  increase  in  concentration,  or bioaccumulate,
    in living  oranisms as  they  breathe contaminated  air,  drink
    contaminated  water/   or  eat  contaminated   food.     These
    chemicals can become concentrated in the tissues and internal
    organs ofanimals and humans.

    The concentration  of  acetaldehyde  found in  fish  tissues  is
    expected to be about the same as the average concentration of
    acetaldehyde in the water from- which the fish was taken.
                   Phytotox database

                 SARA 313 ECOLOGICAL FACT SHEET  .

              U.S. Environmental Protection Agency
                   Office of Toxic Substances

            . -4                                        " •'

GENERAL INFORMATION                  '                  r.'.

    Acetamide (CAS No.  60-35-5)  is a colorless crystalline solid.
    It  is used  as  a  general  solvent  for both  inorganic  and
    organic compounds,  a solubilizer,  a plasticizer,  an  antacid
    in  the  lacquer,  explosives  and  cosmetics  industries,   a
    stabilizer in  peroxides,  and  in  the  synthesis of  organic
    chemicals such  as  methylamine and  thioacetamide.   It  may
    enter the environment from industrial discharges or spills.


    Acute toxic effects  may include the  death  of animals,  birds,
    or  fish,  and  death or  low growth  rate  in plants.    Acute
    effects are  seen two to  four days  after animals  or  plants
    come in contact with a toxic chemical substance.

    Acetamide has  slight acute  toxicit to aquatic life and high
    acute tpxicity to birds.  It has caused germination decrease
    and .  size   decrease  in   several  agricultural  crops.
    Insufficient data  are available to  evaluate or predict  the
    short-term effects of acetamide.to land animals.


    Chronic   toxic  effects  may   include   shortened   lifespan,
    reproductive  problems,   lower  fertility,   and  changes   in
    appearance or  behavior.    Chronic  effects can  be  seen long
    after first exposure(s)  to a toxic chemical.

    Acetamide  has   slight  chronic  toxicit   to aquatic  life.
    Insufficient data  are available to  evaluate or predict  the
    long-term effects  of  acetamide to  plants,   birds, or land


    Acetamide is highly soluble  in  water.   Concentrations  of
    1,000 milligrams and more will mix with a liter of water.


    Acetamide is slightly persitent in water, with a half-life of
    between 2 to  20  days.  The half-life of a  pollutant  is  the
    amount of time it takes  for  one-half of the chemical  to be

             degraded.   Virtually  100%  of acetamide  will end  up in the
             water. •..     t< -


             Some substances increase in concentration, or bioaccumulate,
             in  living. oranisms  as  they breathe  contaminated air, drink
             contaminated   water,  or   eat  contaminated  food.    These
             chemicals can  become concentrated in the tissues  and  internal
             organs of animals and humans.

             The  concentration  of acetamide  found  in  fish  tissues  is
             expected to be about the same  as the average concentration of
             acetamide in the water from which the fish was taken.
         SUPPORT DOCUMENT:  AQUIRE Database, ERL-Duluth, U.S. EPA.
                            phtotox; eeb/birds

                                                                         Mark Hahn

potency values for HAH in fish. This graph appears to illustrate different degrees of variability in
the estimates, indicating different degrees of uncertainty depending on the compound chosen.
                         HAH relative potencies in fish

ES-« NB-B-

relationship between the chosen TEF endpoint and the "endpoint of concern" in other systems
where it is known, and attempt to quantify or predict the degree of uncertainty introduced by
using the surrogate endpoint.
   With regard to the use of CYP1A induction, there are often some misconceptions about the
relationship between this biochemical response and toxicity. Although the mechanism of toxicity
of dioxin-like compounds is not completely known, available evidence suggests that it involves
changes in the expression of genes involved in the regulation of cell growth and differentiation.
CYP1A induction is relevant as an endpoint for TEF determination for two reasons.
   a)  In  a general sense,  induction of CYP1A occurs in  parallel  with the changes in gene
expression that are responsible for dioxin toxicity. CYP1A induction signals activation of the Ah
receptor (AHR), which  is the common initial step in toxicity. In this way, CYP1A induction is a
surrogate for toxicity.
   b)  In addition  to acting as a surrogate for AHR-dependent toxicity, induction of CYP1A can
also be directly responsible for some forms of toxicity.  This may occur through the generation of
reactive oxygen species, for example. Such a mechanism could be important for some endpoints
of concern, such as cardiovascular toxicity involved in early-life stage mortality in fish (15-17).
   The correlation between potency to induce CYP1A and toxic potency is often strong (e.g. 18,
19), but is not  perfect.  CYP1A induction is usually measured as an acute effect, whereas effects
of concern may occur only  after chronic or subchronic exposure.  Thus, some compounds may
induce CYP1A acutely but—because of rapid metabolism,  for example (e.g. PAH)—may not
produce the sustained activation of the AHR that appears to be important for toxicity (20, 21).

   Another endpoint that is  sometimes considered for TEF determination is the accumulation of
highly carboxylated porphyrins.  This effect is AHR-dependent and also appears to be  linked
mechanistically to induction of CYP1A.  However, it also appears to involve two additional
steps—induction of aminolevulinic acid synthase and  binding of HAH to  the induced CYP1A
(12, 22,  23)—that complicate the determination or  interpretation  of relative porphyrogenic

n. Application of the TEQ Approach
1.  What are the implications of assuming no dose-additivity (??) or no interaction among the
components of mixtures? How would risk assessment conclusions differ if analyses were based
on total PCBs or TCDD alone?


2. Should TEF values derived using median response levels (LC50 or EC50) be used in risk
assessments where a "no adverse effect" level is being employed?
   According to receptor theory, the relative potencies for full agonists should be independent of
the location on the dose-response curve where effects measurements are made.  In the real world,
parallel dose-response curves are not always seen because of a) antagonism and partial agonism,
and, b) artifacts introduced by additional phenomena such as enzyme inhibition.
   (a) Partial agonism occurs in situations where there are differences in the intrinsic efficacy of
compounds and where other factors (such as  receptor number) are such  that compounds with
lower  intrinsic efficacies  are incapable  of producing the same  maximal tissue response as
compounds with higher intrinsic efficacies (e.g. see reference 24). (Intrinsic efficacy refers to the
inherent property of a chemical that determines the activity of the chemical-receptor complex (24,
25).  Intrinsic efficacy is distinct from affinity, which is the probability of  a chemical binding to
the receptor.)  There is evidence for partial agonism of some PCB congeners in some systems
(26). Because of their lower intrinsic efficacy, partial agonists will antagonize full agonists under
certain conditions (25).
   (b)  Compounds may  appear to be partial agonists or have non-parallel dose-response curves
as a result of secondary effects on the endpoint measured.  For example,  in some systems
compounds that induce CYP1A protein can also bind to and inhibit the activity of the enzyme
(27). This inhibition will result in reduced levels of maximally-induced CYP1A activity (EROD)
and an underestimate of the EC50 for CYP1A induction.  This will lead  to an overestimate of
relative potency (TEF) values (11, 13, 19).
   For risk assessments  in which a "no adverse effect level" is being employed, it may make
sense to use TEFs derived from lower level responses to avoid the potential problems discussed
above. Such lower response levels might include "threshold responses" (6), initial slopes ("slope-
ratio" methods) (28), or EC values based on  25%  (29)  or 10% (19) of the maximal response
caused by TCDD.

3. To what extent can class-specific TEFs be used?
   Whether "class-specific ,TEFs"  exist  is  an important  question.  There seem to be some
differences that are characteristic of a vertebrate class (e.g. low activity of mono-ortho PCBs in
fish) but there has not yet been a systematic attempt to compare within-class and among-class
variability  in relative potencies.  It might be  possible to address this question by  determining
relative potencies of several HAHs in several species in each of several vertebrate classes,  and
then using multivariate  statistical techniques to evaluate  the within-class  and among-class


IV. Risk Characterization
1.  Are the uncertainties associated with TEFs more problematic than other uncertainties of the
risk assessments?
   It would be useful to quantify the degree of uncertainty associated with each step of the risk
assessment process, and then to focus on the steps for which the uncertainties are greatest.

2.  Strengths and weaknesses of biologically based TEQ assays?  Integration?
   The strengths of bioassays for determining  TEQs include:  a) relatively  low cost, b) the
response integrates additive effects plus any non-additive interactions that may occur between
components of mixtures, c) the  responses reflect the presence of all AHR agonists, including
compounds that may not have been identified by chemical analysis.  For acute bioassays, and
depending on the source of the extract (e.g. sediment vs. tissue), a possible disadvantage is that
rapidly metabolized compounds  such as PAH may contribute more  significantly to the bioassay
response than they would  likely contribute to  toxicity in the  target species.   Because of the
advantages inherent  in each approach—i.e.  bioassay-derived TEQs and TEQs calculated  from
chemical data and TEF values—a combined approach is desirable.

3.  Additional data or research for use in the risk assessments?
   Species-specific TEF values and relative sensitivities for the species of concern, i.e. bull trout,
river otter, bald eagle, Caspian terns would be helpful.  It is important to characterize both the
relative potencies  of HAHs (TEFs) as well as the absolute "dioxin sensitivity" of the target
species relative to species used to determine the levels of concern (e.g. no-effect thresholds). For
example, in  the retrospective scenario , risk assessment  for Caspian terns uses TEFs (and no-
effect thresholds?) based  on  chicken  data.  Common  tems  are  approximately 80-fold  less
sensitive to TCDD than chickens (based on EROD induction in embryo hepatocyte cultures (30))
and exhibit different TEF values (e.g. for PCB-126, which drives the  TEQ in this assessment).
   In  both scenarios, it would be useful  to  have  long-term data on  population  structure,
productivity, etc. for the species of concern so  that possible population-level effects of current
chemical burdens (in the retrospective study) or  future increases (in  the prospective study) could
be evaluated.  For example, in the retrospective scenario, TEQ levels in Caspian tern eggs are
well above the  level of concern established using data from other species of birds. What is the
reproductive success of Caspian  tems at this site now, in comparison to past success at this site
and current success at less contaminated sites?

Retrospective Case  Study
2.  Use of total PCB analyses to monitor cleanup efforts?


   Total PCB levels could provide a useful surrogate for TEQs in monitoring cleanup. However,
to the extent that (a) PCB congener composition or (b) concentrations of PCBs relative to other
HAH classes change with time or depth in the sediment, the ratio of total PCB to total TEQ could
change. Why not use bioassay-derived TEQs to monitor cleanup?

Miscellaneous comments on WHO (1997) Draft Report on Derivation of TEFs for humans
and wildlife
   1.  An important question is  raised in this document (p. 9-10):  To what extent do relative
potencies for lethality mirror relative potencies for sublethal effects? Direct comparisons of lethal
and sublethal endpoints  are scarce.  In mammals, the huge difference in TCDD LD50 values
(guinea pig 1 ug/kg to hamster  5000 ug/kg) is not necessarily reflected to  the same extent in
potencies for sublethal effects (e.g. see ref. 31).  In birds, the correlation may be stronger (19).

   2. With regard to the  molecular basis for TEFs across species, it is noted that homologs of the
AHR and ARNT exist in the nematode C.  elegans. These homologs  have not yet been isolated,
but are predicted based on computer-predicted coding regions (exon structure) of genomic DNA
sequences (32). However, in the putative C. elegans AHR, the "PAS-B domain", which has been
associated with ligand-binding in the mammalian AHR, is not well  conserved (32).  Based  on
this, it has been hypothesized (32) that the ligand-binding characteristics of the C. elegans AHR
homolog may be different than those of vertebrate AHRs.
   An  additional  complication in  understanding  the molecular basis of dioxin action is the
identification of a second AHR in fish (32) and mammals  (33). The functions of the second
AHR, including its ligand-binding properties, are not yet known.
   In the discussion of the species differences in the AHR (p. 28), it should be noted that an AHR
gene has been identified  in lamprey as well as in cartilaginous and bony fish  (32). Interestingly,
however, adult lamprey appear to be non-responsive to AHR agonists, as CYP1A is not inducible
in lamprey treated with  3,3',4,4'-tetrachlorobiphenyl  (34).  The comparative biochemistry and
molecular biology of the  AHR has been reviewed recently (35).

   3.  Antagonistic effects (pp. 36-37).  According to receptor theory, antagonistic properties do
not result from differences in receptor-binding affinity, but rather from differences in intrinsic
efficacy  (see  discussion  above).  This is  an important distinction because  it means  that low-
affinity compounds will not necessarily act as AHR antagonists.  But compounds with lower
intrinsic efficacies may act as partial agonists, and partial agonists will antagonize full agonists
under  certain conditions  (25).  In addition, because receptor  number influences whether

                                                                              Mark Hahn

compounds with lower intrinsic efficacy will act as full or partial agonists, there will be tissue-
and species- differences in antagonistic properties of a given chemical.

   4.  Hexachlorobenzene (p. 41) is a low-affinity AHR  agonist in rat (36).  It has a relative
potency (in rat) of approximately 0.0001 based on receptor-binding affinity and approximately
0.0005 based on porphyrogenicity (36).   HCB has a similar relative potency of approximately
0.0001 for EROD induction and uroporphyrin accumulation in chicken embryo hepatocytes (23).
References cited
1.   Walker, M.K. and Peterson, R.E. (1991) Potencies of polychlorinated dibenzo-p-dioxin,
dibenzofuran, and biphenyl congeners, relative to 2,3,7,8-tetrachlorodibenzo-p-dioxin, for
producing early life stage mortality in rainbow trout (Oncorhyhchus mykiss), Aquat. Toxicol.,2\:
2.   Zabel, E.W., Cook, P.M. and Peterson, R.E. (1995) Toxic equivalency factors of
polychlorinated dibenzo-p-dioxins, dibenzofurans, and biphenyl congeners based on early life
stage mortality in rainbow trout (Oncorhynchus mykiss), Aquat. Toxicol. 31: 315-328.
3.   Zabel, E.W., Cook, P.M. and Peterson, R.E. (1995) Potency of 3,3',4,4',5-
pentachlorobiphenyl (PCB 126), alone and in combination with 2,3,7,8-tetrachlorodibenzo-p-
dioxin (TCDD), to produce lake trout early life stage mortality, Environ. Toxicol. Chem. 14:
4.   Janz, D.M. and Metcalfe, C.D. (1991) Relative induction of aryl hydrocarbon hydroxylase
by 2,3,7,8-TCDD and two coplanar PCBs in rainbow trout (Oncorhynchus mykiss), Environ.
Toxicol. Chem. 10: 917-923.
5.   Harris, GJE., Kiparissis, Y. and Metcalfe, CD. (1994) Assessment of the toxic potential of
PCB congener 81 (3,4,4',5-tetrachlorobiphenyl) to fish in relation to other non-ortho-substituted
PCB congeners, Environ. Toxicol. Chem. 13: 1405-1413.
6.   Parrott, J.L., Hodson, P.V., Servos, M.R., Huestis, S.L. and Dixon, D.G. (1995) Relative
potency of polychlorinated dibenzo-p-dioxins and dibenzofurans for inducing mixed function
oxygenase activity in rainbow trout, Environ. Toxicol. Chem. 14: 1041-1050.
7.   van der Weiden, M.E.J., de Vries, L.P., Fase, K., Celander, M., Seinen, W. and van den
Berg, M. (1994) Relative potencies of polychlorinated dibenzo-p-dioxins (PCDDs),
dibenzofurans (PCDFs) and biphenyls (PCBs), for cytochrome P450 1A induction in the mirror
carp (Cyprinus carpio), Aquat. Toxicol. 29: 163-182.
8.   Newsted, J.L., Giesy, J.P., Ankley, G.T., Tillitt, D.E., Crawford, R.A., Gooch, J.W., Jones,
P.D. and Denison, M.S. (1995) Development of toxic equivalency factors for PCB congeners and

                                                                           Mark Hahn

the assessment of TCDD and PCB mixtures in rainbow trout, Environ. Toxicol. Chem. 14: 861-
9.   demons, J.H., van den Heuvel, M.R., Stegeman, J.J., Dixon, D.G. and Bols, N.C. (1994)
Comparison of toxic equivalent factors for selected dioxin and furan congeners derived using fish
and mammalian liver cell lines, Can. J. Fish. Aquat. Sci. 51: 1577-1584.
10.  demons, J.H., Lee, L.E.J., Myers, C.R., Dixon, D.G. and Bols, N.C. (1996) Cytochrome
P4501A1 induction by polychlorinated biphenyls (PCBs) in liver cell lines from rat and trout and
the derivation of toxic equivalency factors (TEFs)., Can. J. Fish. Aquat. Sci. 53: 1177-1185.
11.  Hahn, M.E., Woodward, El., Stegeman, J.J. and Kennedy, S.W. (1996) Rapid assessment
of induced cytochrome P4501A (CYP1A) protein and catalytic activity in fish hepatoma cells
grown in multi-well plates: Response to TCDD, TCDF, and two planar PCBs, Environ. Toxicol.
Chem. 15: 582-591.
12.  Hahn, M.E. and Chandran, K. (1996) Uroporphyrin accumulation associated with
cytochrome P4501A induction in fish hepatoma cells exposed to Ah receptor agonists,
including 2,3,7,8-tetrachlorodibenzo-p-dioxin and planar chlorobiphenyls, Arch. Biochem.
Biophys. 329: 163-174.
13.  Hahn, M.E. (1996) Overestimation of toxic equivalency factors (TEFs) resulting from
inhibition of EROD activity by cytochrome P450 1A inducers in cultured cells., Proceedings of
the 22nd Annual Aquatic Toxicicity Workshop: Oct. 2-4, 1995, St. Andrews, New Brunswick.
Canadian Technical Report of Fisheries and Aquatic Sciences No. 2093 132-134.
14.  Zabel, E.W., Pollenz, R. and Peterson, R.E. (1996) Relative potencies of individual
polychlorinated dibenzo-p-dioxin, dibenzofuran, and biphenyl congeners and congener mixtures
based on induction of cytochrome P4501A mRNA in a rainbow trout gonadal cell line (RTG-2),
Environ. Toxicol. Chem. 15: 2310-2318.
15.  Cantrell, S.M., Lutz, L.H., Tillitt, D.E. and Hannink, M. (1996) Embryotoxicity of 2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD): The embryonic vasculature is a physiological target for^
TCDD-induced DNA damage and apoptotic cell death in Medaka (Orizias latipes), Toxicol. Appl.
Pharmacol. 141: 23-34.
16.  Guiney, P.O., Smolowitz, R.M., Peterson, R.E. and Stegeman, JJ. (1997) Correlation of
2,3,7,8-tetrachlorodibenzo-p-dioxin induction of cytochrome P4501A in vascular endothelium
with toxicity in early life stages of lake trout, Toxicol Appl Pharmacol 143:t256-273.
17.  Stegeman, J.J., Miller, M.R. and Hinton, D.E. (1989) Cytochrome P450IA1 induction and
localization in endothelium of vertebrate (teleost) heart, Mol. Pharmacol. 36: 723-729.
18.  Safe, S. (1987) Determination of 2,3,7,8-TCDD toxic equivalent factors (TEFs): support for
the use of in the vitro  AHH induction assay, Chemosphere 16: 791-802.

                                                                            Mark Hahn

19.  Kennedy, S.W., Lorenzen, A., Jones, S.P., Hahn, M.E. and Stegeman, JJ. (1996)
Cytochrome P4501A induction in avion hepatocyte cultures: a promising approach for predicting
the sensitivity of avian species to  toxic effects of halogenated aromatic hydrocarbons, Toxicol.
Appl. Pharmacol.  141: 214-230.                                                   ,
20.  Devito, MJ. and Bimbaum, L.S. (1995) The importance of pharmacokinetics in determining
the relative potency of 2,3,7,8-tetrachlorodibenzo-p-dioxin and 2,3,7,8-tetrachlorodibenzofuran,
Fund Appl Toxicol 24: 145-148.
21.  DeVito, MJ., Maier, W.E., Diliberto, JJ. and Bimbaum, L.S. (1993) Comparative ability of
various PCBs, PCDFs, and TCDD to induce cytochrome P450 1A1 and 1A2 activity following 4
weeks of treatment, Fundam. Appl. Toxicol. 20: 125-130.
22.  DeMatteis, F., Harvey, C., Reed, C. and Hempenius, R. (1988) Increased oxidation of
uroporphyrinogen by an inducible liver microsomal system.  Possible relevance to drug-induced
uroporphyria, Biochem. J. 250: 161-169.
23.  Sinclair, P.R., Walton, H.S., Gorman, N., Jacobs, J.M. and Sinclair, J.F. (1997) Multiple
roles of polyhalogenated biphenyls in causing increases in cytochrome P450 and uroporphyrin
accumulation in cultured hepatocytes, Toxicol. Appl. Pharmacol. 146: 000-000.
24.  Kenakin, T. (1993) Pharmacologic Analysis of Drug-Receptor Interaction (Raven Press,
New York).
25.  Goldstein, A., Aronow, L. and Kalman, S.M. (1974) Principles of Drug Action: The Basis
of Pharmacology,  2nd Edition (Wiley, .
26.  Richter, C.A., Tieber,  V.L., Denison, M.S. and Giesy, J.P. (1997) An in vitro rainbow trout
cell bioassay for aryl hydrocarbon receptor-mediated toxins, Environ. Toxicol. Chem. 16: 543-
27.  Gooch, J.W., Elskus, A.A., Kloepper-Sams, P.J., Hahn, M.E. and Stegeman, JJ. (1989)
Effects of ortho and non-ortho substituted polychlorinated biphenyl congeners on the hepatic
monooxygenase system in scup (Stenotomus chrysops), Toxicol. Appl. Pharmacol. 98: 422-433.
28.  Ankley, G.T., Tfflitt, J3.E., Giesy, JP., Jones, P.D. and Verbrugge, D.A. (1991) Bioassay-
derived 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalents in PCB-containing extracts from the flesh
and eggs of Lake Michigan Chinook salmon (Oncorhynchus tshawytscha) and possible
implications for reproduction, Can. J. Fish. Aquat. Sci. 48: 1685-1690.
29.  Engwall, M., Broman, D., Ishaq, R., Naf, C., Zebuhr, Y. and Brunstrom, B. (1996) Toxic
potencies of lipophilic extracts from sediments and settling particulate matter (SPM) collected ina
PCB-contaminated river system, Environ. Toxicol. Chem. 15: 213-222.
30.  Lorenzen, A., Shutt, L. and Kennedy, S.W. (1997) Sensitivity of common tem (Sterna
hirundo) embryo hepatocyte cultures to CYP1A induction and porphyrin accumulation by

                                                                             Mark Hahn

halogenated aromatic hydrocarbons and common tern egg extracts, Arch. Environ. Contain.
Toxicol. 32: 126-134.
31.  Pohjanvirta, R. and Tuomisto, J. (1994) Short-term toxicity of 2,3,7,8-tetrachlorodibenzo-p-
dioxin in laboratory animals:  effects, mechanisms, and animal models, Pharmacol. Rev. 46: 483-
32.  Hahn, M.E., Karchner, S.I., Shapiro, M.A. and Perera, S.A. (1997) Molecular evolution of
two vertebrate aryl hydrocarbon (dioxin) receptors (AHR1 and AHR2) and the PAS family, Proc.
Natl. AcatLSci.  U.S.A.94: in press.
33.  Fujii-Kuriyama, Y., Kobayashi, A., Ema, M., Mimura, J., Morita, M. and Sogawa, K. (1997)
Transcription regulation by Ah receptor, ARNT, and their related transcription factors, FASEB J.
11: A780 (Abstract P56).
34.  Hahn, M.E., Woodin, B.R., Stegeman, J.J. and Tillitt, D.E. (1997) Aryl hydrocarbon
receptor function in early vertebrates: Inducibility of cytochrome P4501A in agnathan and
elasmobranch fish, Comp. Biochem. Physiol.: in revision.
35.  Hahn, M.E. (1998) The Aryl Hydrocarbon Receptor A Comparative Perspective, Comp.
Biochem. Physiol.: submitted.
36.  Hahn, ME., Goldstein, J.A., Linko, P. and Gasiewicz, T.A. (1989) Interaction of
hexachlorobenzene with the receptor for 2,3,7,8-tetrachlorodibenzo-p-dioxin in vitro and in vivo.
Evidence that hexachlorobenzene is a weak Ah receptor agonist., Arch. Biochem. Biophys. 270:


                                                  Sean W. Kennedy, Ph.D.
                                                            Research Scientist
                                                     Wildlife Toxicology Division
                                                National Wildlife Research Centre
                                                          Environment Canada
                                                        100 Gamelin Boulevard
                                                 Hull, Quebec, Canada  K1A OH3
                                                 E-mail: sean.kennedy@ec.gc.ca
In addition to his work as a research scientist with Environment Canada, Dr. Kennedy
is currently adjunct professor of biology at the University of Ottawa in Ottawa.  He
received his Ph.D. and B.Sc. degrees in chemistry and biochemistry, respectively, from
Carleton  University.  His  research focus is  to gain knowledge of the effects of
environmental contaminants on wildlife and human health, with a current emphasis on
the development and application of  novel biochemical  techniques to measure the
effects of dioxins, PCBs, and structurally-related chemicals whose affects are meditated
by the Ah receptor.  He has worked  in programs internationally and was an invited
participant at the World Health Organization's (WHO) meeting in Stockholm in 1997 on
the derivation of TEFs for PCBs, PCDDs, PCDFs, and other dioxin-like compounds.
Dr. Kennedy is a member of the Canada-Germany Agreement on Environmental Health
as well as being part of  assessment panels in  Canada, the United  States,  the
Netherlands, and other locations in Europe. His toxicology studies are often related to
avian subjects. He is currently a primary author of 40 publications.

                                                                       S. W. Kennedy


1.  Does additional information on TEFs enhance the means of evaluating uncertainties in the

To answer this question, I examined Tables 1 and 2 in the Retrospective Scenerio. While there
are some changes (e.g. fish TEQ1 for PCB 77 is 0.031 and fish TEQ2 is 0.62), such differences
have very little effect on the relative contributions of total PCBs, total PCDDs or total PCDFs to
total TEQ concentrations. Therefore, I do not think that the additional information is valuable,
particularly when one considers all of the other uncertainties which go into  a risk assessment.
2.  Should all TEFs be considered to have similar uncertainties?

I do not think that all TEFs should be considered to have similar uncertainties. The  WHO
meeting in Stockholm established the use of a tiered approach for ranking studies from which
TEFs could be derived. I think this approach is reasonable, and TEFs obtained from in vivo
studies should be (and were, at the  WHO meeting) ranked higher than other types of studies. For
example, in fish, TEF values that are based on mortality following egg injections are more likely
to be "accurate!' (I use the word "accurate" to mean that they are more likely to be predictive of
the relative potency in ovo than are values obtained from in vitro studies or from methods that
use QSAR. This statement is not meant to indicate that in vitro and/or QSAR derived TEFs are
of no value - they certainly are. They can be particularly useful for helping one decide what in
vivo studies are required.  For example, several studies with avian hepatocytes have shown
TCDF to be either equipotent or more potent than TCDD at inducing EROD activity.  For the
time-being, a TEF for TCDF in birds of 1.0 seems reasonable, but in vivo studies are warranted
to test this prediction.

TEFs that were derived from several studies (rare for fish and birds, common for mammals)
should be considered to have less uncertainty than TEF values obtained from single studies.

                                                                       S. W. Kennedy

3. To what extent can different types of endpoints that were used to derive TEFs be extrapolated
   to effects that are relevant for the assessment endpoint for each case study?

Retrospective Scenario:

       In general, one should be cautious when using TEFs that have only been derived from in
vitro biochemical responses. However, it should be noted that compounds which contribute the
most to total TEQ concentrations (see below) have been tested for overt toxicity in ovo both in
fish and birds (albeit in a limited number of species).

Fish   Approximately 93% of the total TEQ concentrations in both shiners and lake trout was
obtained from the following compounds: PCB 126, 1278-TCDD,  12378-PCDD, 2378-TCDF,
1,2,3,7,8-PCDF and 23478-PCDF. The TEFs for all of the these compounds were obtained from
studies which determined mortality in rainbow trout following injection of compound into the
egg (ie. a Tier 1 study).  In my opinion, the total TEQ is highly relevant to the assessment
endpoint of interest, despite the fact that lethality-based TEFs have, to date, only been reported
in one species offish.

Birds  Approximately 85% of the total TEQ concentration in Caspian tern eggs was obtained
from the following  compounds: PCB 77, PCB 126 and PCB 105.  TEFs for all of these
compounds were derived from egg injection studies which measured lethality.  In my opinion,
the total TEQ is highly relevant to the assessment endpoints of interest for birds. Despite the fact
that TEFs for many other compounds were only obtained from studies that measured either
EROD induction in ovo or in cultured hepatocytes ,or were from QSAR estimates, these
compounds contribute, in total, only approximately 15% to the total TEQ.

Mammals      For mammals, the TEFs for most of the compounds were derived from several
studies, and TEQ estimates are likely to be relevant.

It should also be noted, that there are generally quite good correlations between relative
potencies of compounds as EROD inducers and their respective toxic potencies (as long as one
considers some of the "problems" with in vitro assays - such as differences in efficacy and

                                                                      S. W. Kennedy

metabolism - see Bastien and Kennedy, Organohalogen Compounds (1997) 34, 215 - 220. In
vitro derived REPs can be very useful for predicting in vivo TEFs.
Prospective Scenario:                                  ,
Above comments for the Retrospective Scenario are of relevance for this case.


1. What are the implications of assuming no dose-additivity or no interaction among the
   compounds? To what extent would the risk assessment conclusions differ if the stressor
   response analyses were based on total PCBs or 2,3,7,8-TCDD alone?

In general, is dose-additivity is not assumed, then the risk assessments need to be based solely on
TCDD and total PCBs. In the following, differences between using TCDD and total PCBs vs.
the TEQ approach is examined. No-effect thresholds indicated in Table 5 of the Retrospective
Scenario were used hi all cases.

Retrospective Scenerio:

Shiners  and Lake Trout

The concentration of TCDD in shiners and lake trout is much lower (230-fold and 55-fold,
respectively) than the no-effect threshold for fish (30 pg/g).  PCB concentrations in these species
offish are 14-fold and 5-fold lower than the no effect level (5 ug/g), shiners and lake trout,
respectively. However total TEQ concentrations of 1.3 pg/g and 4.2 pg/g in shiners and lake
trout, respectively are at a level which approaches levels which might be expected to have some
effect in sensitive species (eg. lake trout). I say this because the lowest value for no-effect
threshold in fish indicated in table 5 is 3 pg/g. PCB 126  and two PCDFs are major contributors
to total TEQ concentration.  It should be noted, however, that 4.2 pg TEQ/egg is much lower
than the reported LD50 for TCDD in lake trout (74 pg/g).

                                                                       S. W. Kennedy

Caspian Tem

The concentration of total PCBs in Caspian tern eggs is 5.7 ug/g, which is higher than the no-
effect threshold for birds indicated in table 5 (1-20 ug/g). Thus, sensitive species might be
expected to have some effects using a risk assessment that is based on total PCBs alone. A risk
assessment that used TCDD alone would conclude that levels of TCDD were much below the
no-effect level (concentration of TCDD is 4.5 pg/g and the the no-effect threshold is 100 pg/g).
In contrast, if one were to use the TEQ approach, the total TEQ concentration (426 pg/g) exceed
the no-effect theshold of 100 pg/g by approximately 4-fold. Thus, the TEQ approach certainly
indicates more reason for concern than does a risk assessment that is based on TCDD alone.  In
addition, the TEQ approach might indicate more reason for concern than would an assessment
that is based on total PCBs.
The concentration of TCDD in otter liver of 1.4 pg/g is much lower than the no-effect level
indicated in table 5 for mammals (60 pg/g).  The concentration of total PCBs of 1 ug/g is
approximately 1A of the no-effect level (2 ug/g). However when the TEQ approach is used, a
much different conclusion is reached.  The total TEQ concentration is 144 pg/g, which is higher
than the no-effect threshold  of 60 pg/g.  PCB 126 contributes the most to the total TEQ.

Prospective Scenario:

I did not to make the type of detailed analysis for this that I did above for the Retrospective
Scenario because residue levels in eggs were not given - my reasoning would be the same,

2. To what extent should TEF values derived at median response levels be used in risk
   assessments where no adverse effect level is being employed?

I not see a serious problem at all with using LC50 or EC50 values (assuming one carefully
considers, and accounts for situations where compounds are not full agonists at eliciting a
particular effect; eg. EROD  induction, in some cases).

                                                                        S. W.  Kennedy
3. To what extent can class-specific TEFs be directly extrapolated to the species identified in
   each case study?

Given that only a limited number of species have been tested, one cannot be absolutely certain.
However, in my opinion, large errors are not likely to be made.  For example, PCB 126 has been
assigned a TEF of 0.1  in birds, based on Tier 1 studies with chickens. REPs for PCB 126 as an
EROD inducer have been determined to be very close to this TEF in hepatocyte cultures
prepared from a large number of avian species. My conclusion, is that the TEF of 0.1 for PCB
126 is likely to be reasonable across avian species, including those of interest in the present
scenarios.  Further studies are required to determine relative potencies of compounds in different
species offish, but the values derived from egg injection studies in rainbow trout are likely to be
relevant to fish in general (based on studies we and others are seeing in hepatocyte cultures) and,
almost certainly, to be relevant to bull trout (Prospective Scenario).


1. To what extent does the TEF approach present challenges, introduce new uncertainties, or
   modify old uncertainties with modeling exposure to AhR agonists?

The modeling of exposure to contaminants, including AhR agonists is beyond my area of
expertise. However, based on the data provided in Table 1  in the document entitled, "Charge
Questions and Physico-Chemical Properties Table",  it would seem that BAFs have come from a
very limited number of studies, and I would question how reliable these are across species. In
addition, one needs to have a lot of information on feeding patterns of the species being  studied.

2. To what extent do exposure route differences used in deriving the TEFs affect their
   application in the case studies?

For fish and birds, the exposure route used to derive TEFs for the most important contributors to
total TEQs (see above) was from egg injections. It is my opinion that these values of definitely
of relevance to the species of interest in the case studies. Values for mammals are also relevant.

                                                                       S.W. Kennedy
3. To what extent does the TEF approach require a more rigorous analytical design?

The TEF approach requires the measurement of dibenzo-p-dioxins, dibenzo furans and non-ortho
PCBs by GC-MS. This increases the analytical costs over the costs of total PCBs\


1. Are the uncertainties associated with TEFs more problematic than other uncertainties of the
   risk assessments?

In my opinion, the uncertainties associated with TEFs are no more problematic than other
uncertainties which are associated with the risk assessment for the tow scenarios.

2. What are the strengths and weakness of using biologically-based TEQ assays, and to what
   extent could these approaches be integrated?

Biologically-based TEQ assays (I refer her to in vitro assays) 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 measuring only dioxins, furans and PCBs. Some of
these assays are considerably less expensive than chemical residue analysis (particularly where
measurement of dioxins, furans and no-ortho substituted PCBs is required).

 One potential problem with such in vitro assays is that they can over estimate the toxic potency
of compounds which are rapidly metabolized in vivo (eg. PCB 77) but recent research has shown
that such problems can likely be circumvented. For example, Bastien and Kennedy
(Organohalogen Compounds (1997) 34, 215-220) and others have reported that the REPs of
rapidly metabolized compounds are dependent on the length of time between the addition of the
compounds to the cells and analysis. Thus, various bioassays under development have
considerable potential for predicting TEQs which are relevant to whole organisms. For the two
case scenarios, I would recommend the incorporation of in vitro bioassays.

                                                                        S. W. Kennedy

compounds to the cells and analysis. Thus, various bioassays under development have
considerable potential for predicting TEQs which are relevant to whole organisms. For the two
case scenarios, I would recommend the incorporation of in vitro bioassays.

I would recommend incorporation of at least one in vitro bioassay into the risk assessments for
both scenarios.  This might either be the H4IIE bioassay or an assay which uses a reporter gene.
In addition, I would consider using primary hepatocyte cultures for species of interest (eg.
Caspian tern and bull trout).  Such methods can be very useful in predicting the sensitivity of
species of concern to complex mixtures of compounds that elicit effects which are mediated by
the Ah receptor.

3. What additional research do you recommend?

I would recommend incorporation of a study that would include the addition of extracts from soil
and tissues to hepatocytes cultures prepared from species of concern. For example, this could be
done for the bull trout in the Prospective Scenario and for Caspian terns for the Retrospective
Scenario.  Such methods are now routine in my laboratory and others, and show considerable
promise for risk assessment purposes. If bull trout cannot be obtained from any location (due to
their endangered status), then rainbow and/or Lake trout could be used. A small number
(approximately 10 eggs) of Caspian tern eggs could be obtained from another location,
incubated, and primary hepatocyte cultures could be prepared.  The advantage of doing such
studies is that one can obtain important information regarding species sensitivity to complex
mixtures of compounds that elicit effects which are mediated by the Ah receptor which would
not be identified by the chemical-based TEQ approach.

I would also recommend inclusion of other biologically-based TEQ (eg. H4IIE, CALUX) assays
into the assessments.

                                                                      S. W. Kennedy
Additional Questions

Prospective Case Studv:

The questions asked here go beyond my area of expertise.

Retrospective Case Study

Sediment cleanup goals would be the same for birds and mammals since PCBs are, by far, the
most important contributors to total TEQ concentrations in these taxa.  However, in fish, PCDFs
are major contributors. Total PCBs could be used to monitor the results of clean-up efforts
providing a good correlations were found between major PCB congeners and  the following:
PCDFs, TCDD and PCB 126.


                                                    Wayne G. Landis, Ph.D.
                                 Institute of Environmental Toxicology and Chemistry
                                           Huxley College of Environmental Studies
                                                  Western Washington University
                                                          Bellingham, WA 98225
                                                             Fax: 360-650-7284
                                               E-mail: landis@henson.cc.wwu.edu
Dr. Landis's expertise is in environmental toxicology, ecological risk assessment, and
population biology.  He holds a  bachelor's degree in biology from  Wake Forest
University, a master's in biology from Indiana University, and a doctorate in zoology
from Indiana University.  He is currently director and professor at the Institute of
Environmental Toxicology and Chemistry at Huxley College, Western Washington
University.  He belongs to six professional societies, including SETAC, ASTM, Sigma
Xi, and the Genetics Society of America.   Dr.  Landis is currently the principal
investigator in a project entitled "Novel Models for the Evaluation and Interpretation of
Ecological  Datasets Applied to the Ecological Risk Assessment of Biotechnological
Products" funded by EPA. His current research includes regional risk assessment, the
application of metapopulation dynamics in estimating the impacts of toxicants, and
microcosm/mesocosm research.  Dr. Landis also is co-developer of the Community
Conditioning hypothesis, a non-equilibrium description of the impacts of toxicants to
populations and communities. He serves on a variety of advisory committees.

                                                             Wayne G. Landis
Workshop on the Application of 2,3,7,8 -TCDD toxicity Equivalency Factors
(TEFs) to Fish and Wildlife
Answers to Charge Questions
Wayne G. Landis, Director
Institute of Environmental Toxicology and Chemistry
MS 9180, Western Washington University
Bellingham, WA 98225
General comments:
  The assessment of compounds that have modes of action similar to that of TCDD and
yet also have estrogenic type interactions is challenging.  Many of my comments that
are specific to the charge questions are based on two factors. First, the risk assessment
process stated here is based on the derivations of LCSOs or NOELs (no observed
effects levels),  methods that misrepresent the toxicity of the compounds. Second, in
both case examples, the risk assessments are purely toxicological not ecological.
  In my reading of the material supplied to us only one paper, the draft by Elonen et al,
used the dose-response curves in order to judge the relative toxicity of TCDD to a group
of organisms. The other papers used a median lethal dose, a no-effects level or a
lowest observed level to compare toxicity. The failings of the NOEC and LOEC (lowest
observed effect level) approaches have been discussed (Stephan and Rodgers 1985,
Chapman et al. 1996, Chapman and Chapman 1997) although debate continues
(Dhaliwal et al 1997). A summary of the problems of NOECs and LOECs can be also
found in Landis and Yu (1995).
  Basically, NOECs and LOECs are artifacts of the hypothesis testing process and  the
concentrations selected by the researcher. While they may be of some interest within a
set of experiments conducted under identical conditions with similar experimental
variance, replication and statistical  power, they can not be compared in a strict sense
between laboratories or species because the statistical power of the experiments
change. As the statistical power changes so does the results of the NOEC and LOEC.
As the statistical power decreases, the NOEC and LOEC will increase even without a
real change in the concentration-response curve.

                                                               Wayne G. Landis

  An alternative approach using regression techniques and curve fitting have been
proposed (Stephan and Rodgers 1985, Moore and Caux 1997).  Specific points along
this curve can then be compared (an ECX) in order to determine relative potencies at
concentrations that correspond with acceptable effects. In this instance we can
compare numbers with similar units. The uncertainty in the comparisons can also be
quantified since the error in the estimates will also be available. This is a much better
situation than comparing statistical artifacts.
  The second failing of the ecological risk assessments provided to us as examples is
that they are still toxicological assessments. Only direct toxicity is considered,  as is
appropriate for determining effects upon a particular receptor. However, the goals of
these example assessments is to attain the same fish populations as before.
Oneofakind lake has depressed fish and tern populations.. It is claimed to have healthy
pelagic and benthic invertebrate communities, but since health is  undefinable
ecologically I have no idea what this means.  Roundtail lake has seen the introduction of
mysids that have drastically altered the food web and the bull trout populations. The
introduction of paper mill effluent would constitute another stressor, with the impacts
partially controlled by the other introductions. Observed alterations in the fish dynamics
could be due to historical impacts, the rates of migration due to landscape structure, or
the toxicity of the effluent.

Caux, P-Y and D. R. J. Moore. 1997. A spreadsheet program for estimating low toxic
   effects.  Environ. Toxicol.  Chem. 16:802-806.
Chapman, P. M., R. S.  Caldwell and P. F. Chapman.  1996. A warning: NOECs are
   inappropriate for regulatory use. Environ. Toxicol. Chem. 15:77-79.
Chapman, P. F. and P. M. Chapman.  1997. Author's reply: Environ. Toxicol. Chem.
Dhaliwal, B. S., R. J. Dolan, C. W. Batts, J. M. Kelly, R. W. Smith, S. Johnson.  1997.
   Warning: Replacing NOECs with point estimates may not solve regulatory
   contradictions.  Environ. Toxicol. Chem. 16:124-125.
Landis, W. G. and M.- H. Yu.  1995. An Introduction to Environmental  Toxicology:
   Impacts of Chemicals on Ecological Systems. Lewis Publishing, Boca Raton, FL.
Moore, D. R. J.  and P-Y Caux. 1997.  Estimating low  toxic effects. Environ. Toxicol.
   Chem. 16:764-801.
Stephan, C. E. and J. R. Rodgers.  1985. Advantages of using regression analysis to
   calculate results of chronic toxicity tests.  In Aquatic Toxicology and Hazard
   Assessment: Eighth Symposium.  R.C. Bahner and D.J.H. Hansen, eds., American
   Society for Testing and Materials, Philadelphia, pp. 328-339.

                                                               Wayne G. Landis

I. Stress-Response profile relative to the derivation of specific TEF values.
3, The TEF values provided were based on endpoints that ranged from in vitro
biochemical responses to in vivo early life stage mortality. To what extent can these
endpoints be extrapolated to the measures of the effects that are relevant for the
assessment endpoint for each case study?

  The more the test is run under conditions similar to the exposure in the field, the
easier and more confident the extrapolation. Biochemical responses observed from in
vitro tests are more  like bioassays for exposure to specific concentrations than
indications of toxicity.  Early life stage mortality tests are more useful, but rarely does the
dosing correspond to situations typical of the field. Each test allows more confidence in
the prediction,  and the greater the number of endpoints measured the better the
characterization of concentration-effects. However, laboratory tests can not take the
place of properly designed field studies or taking advantage of natural experiments
(spills, prior contamination etc.).

II. Stress-Response profile relative to the application of the TEQ approach.
2. Many TEFs are based on LC50 or EC50 values.  To what extent should TEF values
derived at a  median response level be used in risk assessments where no adverse
effect level is being employed?

  in keeping with my introductory comments, the use of LC50 and EC50 values is
inappropriate,  but no more than the use of a no adverse effect level for the risk
assessment. The use of the LC50 and EC50 for TEFs uses a part of the concentration-'
response curve that is of relatively little interest for the protection of ecological
endpoints. No adverse effect level is a statistical artifact at best, at worst it is trying to
prove a negative and that can not be accomplished  scientifically. A more appropriate
alternative would be to settle on an acceptable effect, even one as small as an EC10.
Then use the EC10  values from the LC50 or EC50 data to calculate the TEFs. Once
the risk assessment goal is quantified, then the appropriate endpoints for the
computation of the TEFs is trivial.

                                                               Wayne G. Landis

  The concentration-response curves illustrated in the Elonen et al. manuscript
demonstrate the variability of the slopes. Given the same EC50, the compound with the
shallowest slope will have greater effects at lower concentrations.

3. The TEFs 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?

  In the Elonen et al. manuscript (Table 5), the range of LC^IO and LC^SO both have
a five-fold range in toxicity for seven teleost fish. The data presented in Figure 3 show a
twenty-five fold range from lowest to highest LC^SO values among the fish. Without
comparable data for other Ah receptor compounds It is not possible to tell if the ratios
between TCDD and other compounds shows a comparable interspecific variability. Do
comparable data exist for the ratios and can that be used to examine the range of
TEFs? Getting more data would answer that specific question, otherwise it simply is

III. Exposure profile
IV. Risk Characterization
1. In evaluating the case studies, are the uncertainties associated with the 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

  Given the current methods of estimating the TEFs, reliance on NOECs  and LC50
values, the uncertainty in the  estimates of these values at realistic levels of impacts is
high. Without the  basic biological effects data, the basic yardstick by which to judge
impact is uneven and bent. It is like measuring a centimeter with only a meter stick
marked in meters. It does not matter than there is uncertainty is the other factors as
much because they are not the yardstick by which impacts are measured.

                                                                Wayne G. Landis

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?

  Data from well designed experiments from environmental samples is always a
preferred approach for several reasons. 1) It provides data for sediments and water
conditions that will  be found at the site of interest. 2) Field work can provide a measure
of the temporal and spatial heterogeneity of the environment and the fate and
bioavailability of the contaminants.  3) Data from field samples can provide a  measure of
uncertainty provided  by the laboratory studies and the TEF approach.  4) Site specific
data forces the investigators to pay close attention to the site and reality instead of
laboratory tests and models.

3. Assume that site-specific data or additional research could be gathered or performed
to generated more information for the case study assessments. Provide a list of specific
investigations/studies and rank them from highest to lowest priority. What is your
rationale for the ranking?

  Highest to lowest ranking; assuming that this is a prospective  risk assessment.
1) Obtain as much data as possible on the spatial and temporal distributions of the
species of interest, their supporting food web, and the organisms that alter the physical
structure of the habitat. This information will eliminate a lot of the guesswork about
exposure and population effects. Particularly  important are data about other stressors,
patch distribution and landscape form that may confound  predicted impacts.

2) Simulate the dosing of the system using a model multispecies system that includes
fish as a receptor.  Have specific questions and predictions in mind to guide the
experimental design.  If the models and toxicity data can not effectively predict the risk
to a model system  there is little hope that it can predict risk to the ecological system of
interest. It should also be possible to obtain correlations between biomarkers,
reproductive success and population and community alterations that should allow the
answering of so what type questions.

                                                               Wayne G. Land is

3) Get reliable concentration-response data that actually includes accurate estimates of
effective levels of concern, not NOELs (not real) and LCSOs (too high).  For
bioavailablity studies use sediments and water from the site of interest in order to gain
site-specific data. These studies should allow the elimination of a great deal of the
uncertainty in the toxicological and exposure aspects of the risk assessment.

Additional questions specific to the prospective case study:
1. The sate adopted BAFs used by the GLWOG.  What improvement I the accuracy of
the maximum allowable concentrations for individual congeners in water (MAC) can be
expected through the use of BAFs determined from Roundtail lake data?

  This is a crystal ball, not a scientific question. The accuracy in indeterminable without
doing the experiment.   The important fact is that it is the BAFs from the Roundtail lake
data that should be the most relevant to a risk estimation since they can provide a range
of values assisting in the quantification of the variance, and data on spatial and temporal
variability. This type of data will not be available using model results. After all, models
produce output, not data.

3. How should the uncertainties associated with the available fish, avian and
mammalian TEFs be incorporated into decisions about which TCDD water quality
standards should be chosen for setting a TEqTMDL for regulating chemical discharges
into Roundtail Lake?

  Tell me how much uncertainty the decision maker can live with. The uncertainties
need to reported  fairly and as accurately as possible.  How the decision  is made is more
a political issue when such unspecified and indeterminable criteria such  as no adverse
effect are used.

                                                               Wayne G. Landis

Additional questions specific retrospective Case Study:
1. Would TEQ sediment cleanup goals be the same for each vertebrate group? If not,
why would there be a difference? I the vertebrate group with the most certainty is not
the group with the most restrictive sediment cleanup goal, how would you council the
risk manager's concerns for the other vertebrate groups?

  Of course the clean up goals will be different for each vertebrate group depending
upon the route of exposure. Terrestrial mammals will be exposed in a very different
fashion compared to sediment dwelling fish. Seed eating birds are likely to have little
concern about sediment concentrations compared to fish eating birds. Reptiles and
amphibians that burrow in the mud during parts of the year will have a direct exposure to
the sediment for prolonged periods. Amphibians have to breed in the water, mammals
and birds do not and so have different exposure routes and sensitive stages.
  The second part of the question is amusing.  For the most vertebrate groups are not
represented by any toxicity data and when they are for only a few species. Given the
lack  of representation of the different vertebrates the level of uncertainty is going to
relatively high no matter what. Considering the problems with estimates of exposure,
lack  of tissue data for most species, and the lack of truly comparative toxicology, I do
not hold out much hope for reducing uncertainty for vertebrate groups, only the few well
studied species.
  How about uncertainty factors for extrapolation across vertebrate types? Considering
the reported 25 fold difference in TCDD toxicity in teleost fish,  how much more
uncertainty is there between vertebrate groups. I suspect the answer is species specific
given the precise mode of action of the TCDD and similar compounds. Very subtle
alterations in biochemistry may give rise to big differences in realized toxicity in a largely
stochastic fashion.

2. Would the TEF/TEQ-based sediment remediation goals be the same as those
determined for total PCBs for the identical vertebrate class? Assume that a simple ratio
of total PCB sediment concentration goal to  TEQ sediment concentration goals was
formulated to allow for the use^of total PCBs to monitor cleanup efforts based on TEQs.
What exposure and effect issues would need to be evaluated before using the less
costly total PCB analysis to support the TEQ-based  sediment remediation goal?

                                                              Wayne G. Landis

  No, total PCBs are comprised of many compounds that work with very different modes
of action compared to the TCDD like PCBs. The proportion of the various PCB types
will be important in estimating the likely toxicity resulting from the mixture.  Why not a
TEF for estradiol mimics as well as TCDD mimics?
  I am generally against clean up goals set on chemical concentration alone. Chemistry
does not estimate toxicity very well,  and when have been so caught up in numerical
analytical goals that toxicity prevention can get lost.


                                                    Lynn S. McCarty, Ph.D.
                                     L.S. McCarty Scientific Research & Consulting
                                                           280 Glen Oak Drive
                                               Oakville, Ontario, Canada L6K 2J2
                                                            Fax: 905-842-6526
                                                  E-mail: lmccarty@interlog.com
Dr. McCarty received B.Sc. and M.Sc. degrees from Brock University and a Ph.D. from
the University of Waterloo. He has spent over 19 years as an environmental scientist
in both  business and government positions and currently operates a consulting
business.  In these positions he has been involved in a wide variety  of projects
examining environmental impacts and/or human health effects. This included the
production or critical review of a number of air and water quality guidelines, as well as
work on risk assessments in Canada and the USA.  He  has been involved in the
preparation of over several  dozen scientific papers, many presentations/posters  at
scientific meetings, and numerous proprietary reports for a variety of clients.  As well
he is a coauthor of two chapters in the second edition of the "Fundamentals of Aquatic
Toxicology" (Rand, 1995). A particular interest is the theory and practice of toxicity test
design and interpretation and application to risk management and assessment. Dr.
McCarty has been an invited  expert at a number of workshops dealing with  human and
environmental health issues sponsored by the Canadian Forestry Service, Environment
Canada, Canadian Network of Toxicology Centres, SETAC, US EPA, and US Army
Corps of Engineers^   He currently serves  on the editorial boards  of Human and
Ecological Risk Assessment and Journal of Aquatic Ecosystem Stress and Recovery.

L.S. McCarty Scientific Research & Consulting               1156868 Ontario inc.
280 Glen Oak Drive, Oakville
Ontario. Canada L6K 2J2	905/842-6526 (phone & fax)	lmccarty@interlog.com
TO:            U.S. EPA TCDD TEF Workshop
FROM:        L.S. McCarty
DATE:        Novembers, 1997
TOPIC:        Answers to Questions/Issues for the Workshop on the Application of 2,3,7,8-
               TCDD Toxicity Equivalency Factors (TEFs) to Fish and Wildlife

I have an objection with a statement in the opening paragraph: "It is reasonable to assume that
the proposed WHO TEFs are appropriate for risk assessments ... " I agree that this is the basis
for the subsequent questions on refinements of the TEF approach to assessments beyond the
screening stage, but do not believe that it is a universally agreed upon assumption for either the
initial application or the refinements being considered by the workshop. In fact, it should be
made clear that such an assumption clearly establishes this workshop as a policy-based exercise.
The workshop is a means of obtaining the best professional judgement of scientific experts on
how, in their opinion, to most suitably apply available but incomplete scientific facts to serve
policy objectives.  Without explicit clarification there is a danger that such deliberations may be
perceived by many as being a purely scientific discussion when it is not. I both understand and
support the general need for some degree of precautionary activity, but strongly object to
dressing it up as science. Good, reasonable policy incorporates input from a variety of sources
and does not need a scientific aura for respectability.

Rather than stating that it is reasonable, I believe that a list of the assumptions required to enable
the TEF process to be used in risk assessment be presented, both for the screening and advanced
cases. Any reader can then judge the degree of reasonableness for themselves.  This is
consistent with the concerns which prompted the Levin-Thompson bill currently under debate in
the U.S. Senate. This bill illuminates the need for identification and clarification of both the
scientific and policy basis of assumptions used in risk assessment. Such a separation of science
and policy in should make the risk assessment process more transparent and understandable.
The recent Presidential/Congressional Commission on Risk Assessment and Risk Management

                                        L.S. McCarty, TCDD TEF Workshop Comments
(1997) has made a call for improved risk communication and a clear identification of science and
policy aspects of the TEF approach would also contribute to achieving such a goal. I have also
commented on the confusion of policy for science in risk assessment (Power and McCarty,

It is my opinion that the TEF approach as currently constituted is not sufficiently rigorous or
comprehensive to be employed in other than screening level risk assessments for aquatic, avian,
and mammalian wildlife. The approach represents a reasonably founded policy for screening that
also serves as a useful guide for directing additional scientific research. However, the limitations
and restrictions specified in the meeting description and charge questions represent little more
than a detailed list summarizing why, at this time, it should not be used beyond an initial
screening risk assessment.

The method addresses only Ah-receptor mediated effects.  This provides only an illusion of full
protection since non-Ah-receptor-mediated effects associated with the dioxin-like chemicals may
still cause adverse effects by other modes of action. Thus, the overall goal of environmental
protection may not be achieved using a TEF risk analysis alone. The method assumes strict
additivity and, although a reasonable assumption, cases of over- and under-protection are
possible for a variety of reasons. The TEF approach is strictly a lexicological approach dealing
only with direct effects and ignoring indirect and nondirect (induced) effects. Nondirect or
induced effects are the result of changes in physical/ecological conditions which are not either a
direct or indirect biological response of an organism to a chemical stressor, but may be a sequela.
Examples of this would be changes in benthic communities associated with changes in sediment
texture or quality resulting from biological or physical/chemical events associated with the
contaminant of concern, or loss of habitat associated with ecological or anthropogenic events
related to the chemical contamination of concern. In the traditional toxicological sense, no
pharmacological dose of a chemical can be described to model the situation, but such effects
may combine with or dominate the direct toxicological effect (Munkittrick and McCarty, 1995).

Ecological dynamics in the field are not considered. Population (both intra-species and
interspecies) and community level compensating factors can have substantial influences on the
nature and degree of response in natural field populations are ignored.  This issue is particularly
problematic as empirical information questions the  validity, or at least the accuracy, of the

                                        L.S. McCarty, TCDD TEF Workshop Comments
extrapolation method:  "However harmful effects (e.g. effects on survival, growth and
reproduction) of dioxin-like chemicals are often difficult to detect at the population level..
Therefore, methods to assess and predict effects on individuals are required" (WHO, 1997).
Also, the method does not address interactions, both positive and negative, with other stressors
and factors in the real environment that is being assessed.

Despite the problems noted above and my views of them, I will attempt to include answers to the
supplied questions which are in the context of the question asked.

VALUES                                                                       ,
1.  The additional background information available for some TEFs provides an attractive, but
illusionary means of evaluating uncertainties. I am not aware of a comprehensive list of possible
sources of uncertainties, with a quantitative ranking of the possible contribution of each. Thus, it
is not possible to quantitatively evaluate the data that is  available and assign valid, comparable
uncertainty rankings. Although some qualitative assessment may be carried out, it too is prone
to being misleading since it is a evaluation of only the uncertainty information known. It is be
quite possible that influencing factors for which there is currently no information could
dramatically alter any uncertainty evaluation made with incomplete knowledge.  As well, since
the overall uncertainty level is not quantified, the relative magnitude and significance of any
uncertainty reduction cannot be determined.

By my estimate about 25% of the proposed WHO TEFs (Table 5) are rounded to the nearest 1/2
order of magnitude (significant digit is 5 rather than 1).  I think that the statement in this question
that TEFs are generally rounded  to the nearest order of magnitude is overstating the case. With
25% rounded to the nearest 1/2 order of magnitude I think that is more representative statement
of the actual rounding practice. I note that this is stated correctly in Tables 1-3 in the
retrospective case study.

2.  All TEFs should not be considered to have similar uncertainties. As noted, a variety of studies,
endpoints, and exposure routes have been employed. Until such time as either there is a common
experimental basis for the TEF scheme or there is quantitative knowledge of the toxicokinetic

                                        L.S. McCarty, TCDD TEF Workshop Comments
and toxicodynamic relationships between various tests, endpoints, and exposure routes, the
uncertainty associated with derivation remains problematic.  Although there are greater amounts
of background information for some congeners, the information base for all is insufficient or
incomplete. Therefore, all uncertainties associated with each TEF are not quantifiable and the
similarities in the uncertainties associated with each TEF are unknown.  Although TEF estimate
uncertainties may lie in a similar range, or be of modest influence compared to other
uncertainties, partial quantification at this time would impart a false sense of accuracy.

3. There is a question as to whether any of the TEFs derived from in vitro and in vivo laboratory
testing can be reliably extrapolated to the effects relevant to the chosen assessment endpoints.
Assessment endpoints are usually clear goals related to the maintenance of populations of certain
valued or threatened/endangered species or, more specifically, the maintenance of reproduction
and protection of sensitive life stages in these species. Protection of the community is assumed
to be accomplished when the sensitive or sentinel species are protected. This is a very broad,
unfounded assumption.

Success in protecting a community/ecosystem is closely related to population modellers
knowledge of the system being examined and their ability to employ the toxicological data in
their models to address the assessment endpoints selected.  Currently toxicological data that are
or can be"quantitatively related to growth, reproduction, and survival (mortality) are most likely
to be of use, since these are the effects that current models have been developed for.  Any other
endpoints are likely to be of little use for extrapolation modelling and of little use for risk
assessment purposes.

The following has been noted on page 3 of the WHO Draft Report (WHO, 1997) "However
harmful effects (e.g. effects on survival, growth and reproduction) of dioxin-like chemicals are
often  difficult to detect at the population level. Therefore, methods to assess and predict effects
on individuals are required."  It appears that most TEFs based on laboratory tests are likely to be
unreliable or at least unvalidated for prediction of populations/communities of organisms in field
situations at the current state of the knowledge. If relatively dramatic effects of dioxin-like
chemicals, such as survival, growth, and reproduction, are difficult to detect at the population
level it suggests that compensating mechanisms are active. Without a good knowledge of the
number, types, and effectiveness of such compensating mechanisms it will not be possible to

                                        L.S. McCarty, TCDD TEF Workshop Comments
reliably extrapolate laboratory data. Since a variety of effects observed in laboratory testing are
more subtle or less clearly linked to survival, growth, and reproduction, they will be of even
lesser utility in predicting effects in the field.

In summary, since many TEFs are based on effects that are poorly linked to survival, growth,
and reproduction, and since it appears that compensating mechanisms in  field
populations/communities are poorly understood for the effects of dioxin-like chemicals, accurate
extrapolation using current TEFs to protect selected populations in the field is unlikely.

1. Conceptually, no additivity or mixture interaction would result in a lower estimation of risk.
Risk would be based on the extrapolated effect of only the most toxic congener, that is to say the
chemical present with the an expected or observed ambient concentration closest to or most in
excess of an estimated or regulated adverse effect level. This is the opposite situation to that
where simple,  non-potency adjusted mixture additivity is employed and a higher estimation of
the risk would result. The degree of underprotection or overprotection of these different
approaches to mixture toxicity compared to the TEF approach cannot currently be assessed
quantitatively since considerably more toxicology and ecology knowledge  and data would be
required. Furthermore, there are insufficient data to perform a qualitative evaluation. The TEF
approach, although clearly based on current scientific  understanding and principles, is best
viewed as a policy based on good judgement, and should not be presented as having strong
empirical support for risk assessment extrapolation.                                         .

The risk assessment conclusions for the retrospective case study would not be completely
different if based on total PCBs or 2,3,7,8-TCDD alone. For this the TEF approach is assumed
to be used to adjust potency but only the single most potent congener is used to assess risk
relative to the  proposed guideline.  Based on Table 1 it can be seen that neither the total TEQs
nor the PCB TEQs exceed the provisional fish guidelines of 30 ng/g. Similarly, the TEQs from
PCDDs and PCDFs as groups or any individual congener alone does not exceed the guideline.
Also, the total PCB concentrations do not exceed the provisional guideline of 5,000,000 ug/g.

The original TEQ analysis for birds (Table 2) finds exceedences of the provisional guidelines for

                                        L.S. McCarty, TCDD TEF Workshop Comments
total TEQ (100 ug/g) by PCBs, but not for the TEQs from PCDD or PCDF. Using a non-mixture
approach'the total TEQ guideline is exceeded by PCB-126 . No other individual congener
exceeds the it. The total PCB concentration in Caspian tern eggs exceeds the 5,000,000 ug/g
limit by a relatively small amount. Thus, the conclusion of a modest adverse effect on birds can
be obtained from either the detailed TEQ analysis or the total PCB analysis. The analysis for
mammals (Table 3) is different. The original TEQ analysis finds exceedences of the provisional
guidelines for total TEQ (60ug/g) by both the TEQ total and  the PCBs, but not for the TEQs
from PCDD or PCDF. Using a non-mixture approach the total TEQ guideline is exceeded only
by PCB-126. The total PCB concentrations do not exceed the provisional guideline of 2,000,000
ug/g. In this case the exceedence estimated by the TEQ analysis is not confirmed by the total
PCB analysis.
Although the results of the risk assessment do change somewhat, the general conclusion drawn
from them would not change substantially with an alteration  from mixture additivity to
consideration of only the most significant single congener or to consideration of total PCB
concentration alone. The conclusion is that, in this watershed, there are levels of certain
organochlorine chemicals present in organisms above the proposed effect levels and the
dominant source is PCBs, in particular PCB-126.  Of course, the conclusion depends on the
nature of the residue levels present in the study and the above conclusion would not be universal
for all cases. However, the multiple receptor approach with foodchain considerations does appear
to be robust and appears to provide more certainty than a less diverse examination would

2. Given the uncertainties and variability in the data on which TEFs are based, any differences
caused by the use of median response level versus no adverse effect level data is likely within the
considerable noise associated with the TEF estimation process. However, an estimate of the
contribution can be made. There are empirical data to suggest that differences between acute and
chronic responses in conventional aquatic toxicity data is usually of the order of a factor of 10 or
less (see Rand et al, 1995). Also, some fish TCDD TEFs calculated at the threshold of EROD
induction were about four to five times larger than international TEFs (I-TEFs), while being
similar to I-TEFs when conventional ED50 data were employed (Parrott et al. 1995).  This
suggests that at low concentrations typical of environmental exposures, fish TEFs may be
different from mammalian-based TEFs and/or there may be a difference between TEFs

                                         L.S. McCarty, TCDD TEF Workshop Comments
calculated at median response levels versus those calculated from information closer to no effect
levels. If the latter is the primary source of the difference, then it supports the contention that
TCDD TEF toxicity estimates are affected by differences in endpoint response proportion, that
such differences may be as great as the order of a factor of 5, and that such differences represent
nonconservative errors in the risk assessment process using TEFs since congeners appear to be
more toxic compared to TCDD than when compared a median response levels.  It should be
noted that the opposite appears to be true for TEFs for PCBs since they are often smaller that I-
TEFs when estimated away from median response levels.
3. Extrapolation of class-specific TEFs (e.g., primarily based on single or limited mammal, bir
or fish data) to species identified in the case studies is currently a matter of policy rather than
science. It is not uncertainty, but rather ignorance, that is the main controlling factor.  In
addition to the general laboratory-to-field extrapolation problems discussed in the response to
question 1.3, there is now the differences between the species used in class TEF development and
the species selected in a given risk assessment. There are exposure and toxicokinetic
differences. These include differences composition and timing in exposure routes (e.g., water,
diet (sediment, foodchain)), lifestage and other seasonal factors, and metabolic handling
differences.  Toxicodynamic factors such as differences in Ah-receptor density in target tissues,
as well as possible differences in receptor character, also complicate extrapolation.

In addition, the choice of assessment-specific species is not based on a rigorous scientifically-
based method, and it is clearly not optimized for lexicological extrapolation.  For example, in the
prospective study bull trout "... as a potentially very sensitive species (probably as sensitive as or
more sensitive than lake trout), was chosen because of its status as a threatened species." while
bald eagle and the river otter were chosen as "representative bird and mammal species" without
any detailed technical justification being supplied. Knowledge concerning TEF extrapolation is
largely qualitative, semi-quantitative at best, and if TEFs are to be used it should be clear that
such use is based on professional judgement and is a policy-based assumption rather than a
scientific  fact. At the moment TEF extrapolation should be considered as good policy but
inadequate, incomplete science.

1 . The exposure modelling uncertainties associated with TEFs are those common to modelling

                                        L.S. McCarty, TCDD TEF Workshop Comments
the fate of any chemical contaminant or contaminant mixture.  The TEF approach has an
advantage that, unlike the case where a mixture of chemicals may contain a diverse group of
chemicals with differences in mode of toxic action, dose additivity is an integral part of the
approach. The ranking of the potency of various congeners does provide an advantage since the
degree of accuracy on the ambient level estimation  can be adjusted relative to potency. For
congeners not on the TEF list, chemical analysis can be avoided. For low potency congeners,
analysis can be less rigorous as their contribution is likely modest anyway. Analytical efforts ca
then focus on for high potency congeners, since these have the greatest contribution and should
be determined most accurately. A similar logic applies to fate/transport and foodchain models,
since the level of effort and degree of accuracy can be tailored to the potency of the congener.

Although there are some differences in the availability and quality of congener-specific physico-
chemical data I believe that any deficiencies here are less significant than in the knowledge of
physical, chemical, biological, and ecological processes and relationships used in fate/transport
and foodchain models.
2. Exposure route differences between the data used to derive the TEFs and the exposure
profile(s) in a particular case study can be of great importance and effort are required to address
this issue. The closer or more representative the dose surrogate is to the dose at the site of toxic
action, the more useful and more readily interpretable it is likely to be from a lexicological point
of view. Parrott et al. (1995) provide a useful example. Liver concentrations of PCDD/F
congeners were better predictors of EROD activity than oral doses. There were some differences
in the ranking of potencies of the PCDD/Fs between fish and mammalian data  As well, fish
TEFs calculated at the threshold of EROD induction were about four to five times larger than
international TEFs, suggesting that at low concentrations typical of environmental exposures,
TEFs may be different from mammalian-based TEFs which are often based on median response
levels. This suggests another twist related to different exposure routes. Since an estimate of the
received dose is not usually obtained in exposure-based dosing, some of the differences in TEF
estimates reported in different species or endpoint testing may be simply related to differences in
the amount of the received dose.

In summary, estimates of received doses are more readily interpreted from a lexicological point
of view. However, if only received dose data are available information on bioayailability,

                                        L.S. McCarty, TCDD TEF Workshop Comments
partitioning, and metabolic breakdown differences may be missing. This is the^very data needed
to facilitate risk assessment which is commonly focused on concentrations of dioxin-like
chemicals in environmental media.  Thus, unless bioavailability, partitioning, and metabolic
breakdown differences between organisms, congeners, and test endpoints are available, along
with either an exposure or received dose estimate, application of TEFs in risk assessments will
be difficult and potentially misleading.

3. In all regulatory approaches based on comparison with a critical effect or no-effect level it is
important to minimize measurement and manipulation errors and uncertainties to the extent
reasonably possible. The simple total  PCB approach relies on summing PCB data and
comparing the result to a guideline level.  In the TEQ approach congener-specific measurements
are manipulated by equations containing several parameters and the errors/variability increases
as a result. The greater the uncertainty in the parameters the greater the uncertainty in the
product which is the basis of the comparison. Thus, in the interests of keeping uncertainty down,
and perhaps comparable to the simple  total PCB approach, chemical analysis of AhR agonists
should be more rigorous and thereby produce less uncertain estimates that will allow the TEQ
product to exhibit a similar uncertainty.

The above comments are based largely on mathematical considerations. On the other hand both
methods have substantial, but unquantified errors and uncertainties associated with lexicological
and ecological aspects. Thus, the overall extent to which any additional analytical efforts would
substantially reduce TEF methodological uncertainty is unknown.

1. The uncertainties associated with TEFs are not more problematic than other uncertainties
associated with case study risk assessments. In fact, given their relatively narrow focus and
comparatively detailed examination, they are likely less uncertain than some of the other aspects
of the risk assessment process.  With the TEF approach at least some attempt has been made to
quantify the differences in toxic potency. On the other hand, as noted elsewhere, assumptions
required to project populations, communities, and ecosystem effects from controlled toxicity
testing results are rather broad and, for the moment, little quantification of the influence of
current practice has been attempted. Also, bioavailability directly from the environment, as well
as at various stages in the foodchain (direct bioavailability from dissolved water phase,  dietary

                                       L.S. McCarty, TCDD TEF Workshop Comments
absorption efficiency from ingested sediment and prey organisms), is a major source of
variability. Although addressed in some degree in the current BSAF, BAF, BMF, and FCM
approaches, detailed consideration would allow for better understanding and quantification of
this likely important source of variability. I expect that it would be at least a significant a source
of variability as the TEF toxicity scheme.

2. At this time I do not believe that biologically-based TEQ assays with environmental samples
represent a useful or viable extension to the current TEQ screening approach to regulation.
Certainly such activities would be useful in the examination of the validity and accuracy of TEQ
screening, and should provide useful insights helpful to further refinement of the scheme.
However, it is premature and unwise to use research tools in a regulatory process.

3. For regulatory purposes I would not desire any further site-specific data.  As I noted earlier I"
do not believe that the TEF approach should be used for anything other than a screening risk
assessment. Although there can be debate about what constitutes a screening risk assessment
and a detailed site specific risk assessment, the case studies provided certainly tend more
towards the latter. I do not believe that there is enough understanding of the toxicology and,
especially, ecology to further refine such regulatory approaches at this time. Even the current
status is providing a false sense of scientific validity and I would not wish to have it go any
further. Additional work in basic research is needed to better understand the toxicology and
ecology in a field situation to aid in better understanding extrapolation.  Only then would
additional site-specific data be of substantially greater utility.

Additional Questions Specific to the Prospective Case Study
1.1 trust the question refers to BAFH, rather than BAFfdw since the latter does not appear in the
GLWQG, Table  1 in the Charge Questions, or Figure 5 of the prospective study. Improvements
in the accuracy of congener-specific MACs using site-specific data for BAFfl, determination will
be a function of how different the site-specific values would be compared to those values used in
the GLWQG determination process. It will also depend on whether the first or second most
preferred method of deriving baseline BAFs is followed (see GLWQG, 1995, page 2). Since the
values used in the GLWQG consider all routes of exposure and all aspects of environmental fate,
including metabolism, a very thorough extensive sampling and analysis program on Roundtail •*

                                      L.S. McCarty, TCDD TEF Workshop Comments
sediment cleanup goals from the various methods, the scientists should offer a best professional
judgement ranking the values and the manager should consider additional non-scientific (i.e.,
economic, technological etc.) factors in the choice of a final project cleanup value.

2. Not answered.
References Cited

Munkittrick, K.R. and L.S. McCarty, 1995. An integrated approach to ecosystem health
management: top-down, bottom-up or middle-out? J. Aquat. Ecosys. Health 4:77-90.

Parrott, J.L., P.V. Hodson, M.R. Servos, S.L. Huestis, and D.G. Dixon, 1995. Relative potency
of polychlorinated dibenzo-p-dioxins and dibenzofurans for inducing mixed-function oxygenase
activity in rainbow trout. Environ. Toxicol. Chem. 14:1041-1050.

Power, M and LS McCarty, 1997. Fallacies in Ecological Risk Assessment Practices. Environ.
Sci. Technol. 31(8):370A-375A.

Presidential/Congressional Commission on Risk Assessment and Risk Management, 1997.
Volume 1: Framework for Environmental Health Risk Management. Volume 2: Risk Assessment
and Risk Management in Regulatory Decision-Making.  Commission on Risk Assessment and
Risk Management, Washington DC.

Rand G.M, P.G. Wells, and L.S. McCarty, 1995.  Chapter 1: Introduction to Aquatic Toxicology.
In:  G.M. Rand (ed.), Fundamentals of Aquatic Toxicology II: Effects, Environmental Fate, and
Risk Assessment. Taylor and Francis, Bristol PA. pp. 3-67.


                                      L.S. McCarty, TCDD TEF Workshop Comments
Lake would be required to improve the estimates. However, even given that, the low to
nondetectable levels of Ah-receptor stressors currently in the system make it unlikely that
improvements could be made in a prospective study since non-detect data points would confound
the analysis, especially for PCDD/F.

2. Not answered.         .                       '

3. As presented in the prospective case study the water quality standard estimates of 0.032,
0.028, and 0.021 pg TCDD/L have too many significant digits. The equations used (e.g., 1 or 2)
employ parameters with various significant digits. However, the TEF estimates which are used
in the equations are declared to be a single significant digit which is rounded to the nearest order
or 1/2 order of magnitude, depending on the source of the statement. Thus, values with 2
significant digits, such as are presented, represent a serious distortion of the actual precision of
the output of the formulas. Conventionally, the output of such an equation is presented with a
level of significance no greater than that of the least precise parameter. -In this case it is the TEF.
Thus, rather than a choice of 3 values the choice should be between either 0.01 or 0.05 pg
TCDD/L if 1/2 order of magnitude precision is used. If the precision is at the order of magnitude
level there is only one estimate: 0.01 pg TCDD/L. Given the uncertainty and lack of precision in
the other input parameters of these equations I am inclined to go with the order of magnitude
estimate. This represents a more realistic consideration of the uncertainties in the estimation

Additional Questions Relative to the Retrospective ("!ase Study
1. It is very unlikely that the sediment cleanup goals would be the same for each vertebrate
group, although I cannot confirm this without doing the detailed calculations.  The reason for the
expected difference is that the three formulas used to estimate fish, bird egg, and mink TEQ
relationships to sediment use differing BASF/BMF and TEF values, as can be seen from the
information in the included tables. These differences are appropriate and expected since the
target organisms occupy different locations in the food chain. Given the variety of data sources
and limitations, and numerous assumptions required I feel it will be difficult to quantify
meaningful differences in certainty of clean-up goals. If there are substantial differences in the

                                                     Charles Menzie, Ph.D.
                                                  Menzie-Cura & Associates, Inc.
                                                    2 Courthouse Lane - Suite 2
                                                            Fax: 978-970-2791
                                                    E-mail: charliemen@aol.com
Dr. Menzie specializes in assessing environmental risks of toxics in aquatic and marine
systems. He has many years of experience working with EPA on projects related to the
development of the ecological risk assessments, case studies, and guidelines.  Dr.
Menzie has a good working knowledge of EPA's risk assessment guidance and has
used this in risk assessments for several aquatic and marine environments. He has
also used a risk or hazard assessment framework to identify research needs related to
the fate and effects  of toxics in estuaries.  He  has performed  cross-media risk
assessments for ocean-dumped  wastes.  He  investigates marine and  estuarine
environmental problems on all coastal areas of the United States, including Alaska and
Hawaii, and provides multi-media assessments of various remedial action alternatives.

Dr. Menzie has chaired multiple workshops and colloquia related to the development
of the EPA ecological  risk assessment guidelines. Additionally, he has prepared case
studies and process diagrams, performed peer reviews, and other tasks for EPA related
to ecological risk  assessment.  Dr.  Menzie will serve as the chair of the EPA workshop
on shrimp virus issues related to ecological risk assessment. Dr. Menzie acted as the
facilitator for a series of public stakeholder meetings on shrimp virus issues in July of
1997.  He was chosen as the chair based on his extensive experience with ecological
risk assessment,  environmental risk in marine and estuarine systems, and his skill as
a facilitator and chair.

                                                                        Charles A. Menzie
To: Eastern Research Group
From: Charles Menzie
Topic: Pre-meeting comments on TEF Charge Questions

1.       Does the additional information enhance the means of evaluating uncertainties in the
        assessments? If so, how? If not, why?
        The additional background information is useful for evaluating the uncertainties in the
        assessments primarily because these give insight into the methodology used to derive the
        estimates. The "uncertainties" probably have more to do with the methodology than to
        rounding issues.

2.       Should all TEFs be considered to have similar uncertainties?
        No. Because TEFs are "models" based on empirical data, the amount and quality of data
        affects the level of confidence that can be given to each value. The derivation of TEFs is
        commonly based ona weight-of-evidence approach. Therefore, as the weight of evidence
        increases, there is greater certainty about the TEFs as well as the variability of these

3.       To what extent can endpoints be extrapolated to the measurers of effects that are relevant
        for the assessment endpoint for each case study?
        The different measured endpoints are related to the endpoint of interest. As long as the
        same type of related endpoint is used to develop relative measures of effects,
        extrapolation is possible. There  is greater uncertainty associated with using endpoints
        that are surrogate measures of the effects of interest than endpoints that are more directly
        related. This source of uncertainty is difficult to quantify. However, where data sets exist
        for several endpoints, it may be  possible to quantify the extent to which relative
        measures diverge from one another.

                                                                        Charles A. Menzie
       APPROACH                               .

1.      What are the implications of assuming no dose additivity or no interactions among the
       components of the mixtures?
       Most environmental exposures of consequence occur at relatively low doses. Most
       available information suggests that an additivity (i.e., non-synergistic and non-
       antagonistic) model is appropriate under such circumstances. The use of such a model is
       consistent with our knowledge of effects under low dose exposures. Alternatively, it is
       unlikely that sufficient information would be obtained in the near future to  support an
       alternative model. Assuming additivity is probably the most appropriate approach and is
       more likely to overestimate than to underestimate effects.

2.      To what extent should TEFs derived at a median response level be used in risk
       assessments where a no adverse effect level is being  employed?
       The question suggests that there is a potential "apples and oranges" problem associated
       with mixing these different types of information. This is not the case. Median response
       data are selected because they provide useful - and more stable values - of relative
       measures than do data at the tails of dose-response curves (e.g., NOAEL values).
       However, these relative measures can still be combined with absolute toxicity data at the
       tails of a distribution for the purpose of estimating risks. In such cases, there would be
       uncertainty associated with the selected toxicity data but the relative measures would
       still be appropriate.

3.      To what extent can class-specific  TEFs be directly extrapolated to the species identified
       within each case study?

       It would be useful to have measures of variability among species within a class for both
       toxicity and relative measures of toxicity. Without such information, it is difficult to
       comment on the uncertainties  associated with extrapolation.

                                                                        Charles A. Menzie


1 a.     To what extent does the TEF approach present challenges ....?
        The approach reduces uncertainties associated with estimating risks associated with
        mixtures because it makes greater use of the information available on the relative
        toxicities of the compounds within the mixture. Because the mixture is variable in
        composition, a method that accounts for such variability is likely to provide a better
        estimate of effects than a method that  assumes a specific composition.

Ib.     How does the approach affect fate and transport modeling considerations?

        The approach does require more detail to be included in fate and transport models. For
        simple models, the impact will be small. However, for large models with extensive
        computations, the additional effort (models runs and times) can become demanding.
        Modeling these complex mixtures will require the same types of considerations that have
        been given to models of petroleum hydrocarbons. The recent work of the Total
        Petroleum Hydrocarbon Workgroup (TPHCWG) is a good example. This group has
        divided the complex mixture of petroleum hydrocarbons into manageable fractions for
        the purpose of modeling and for risk assessment.

2/      To what extent do exposure route differences used in deriving the TEFs affect their
        application in the case studies?

        TEFs are relative measures of effects.  However, it is possible that the relationships
        between administered, absorbed, and effective doses could vary depending on route of
        exposure and that these do not vary consistently among compounds. Thus, there is
        greater uncertainty with using TEFs that are based on routes of exposure different from
        those being evaluated in the risk assessment.

        To what extent does the TEF approach require a more rigorous design...?

                                                                         Charles A. Menzie

        The TEF approach will require greater analytical costs. Based on experience, the
        analytical cost may be higher by a factor of two to ten as compared to total PCB
        measurements. The TEF approach will also require greater efforts to perform QA/QC,
        data validation, and data management.


1.       Are the uncertainties associated with TEFs more problematic than other uncertainties?

        No. Use of TEFs does not introduce greater uncertainties into the analysis in cases where
        toxicity data are based on literature values (as compared to direct measures of toxicity.)
        These uncertainties do not limit the analysis.

2.       What would be the strengths and limitations of a biologically-based TEQ approach?

        The major strength is that such an approach provides a better measure of the effects of
        the mixture and avoids having to rely upon a reconstruction of the effects from an
        estimated "sum of the parts." The major disadvantage has to do with having an
        acceptable approach and the analytical costs associated with implementing that

3.       Provide a list of investigations and rank them.

I would rely upon a weight-of-evidence approach. This would consist of three components: a)
field observations of effects using an ecoepidemiological approach, b) laboratory exposures
using extracts of sediment, water, or fish, and c) an assessment of effects based on chemical
measurements. All of these contribute to an overall understanding of effects. I place greater
reliance on field observations for retrospective studies and on laboratory toxicity tests for
prospective analyses.

                                                                         Charles A. Menzie

Additional Questions for Prospective Case Study


1.      The Roundtail Lake data are more relevant for site-specific evaluation. Therefore, MAC
       based on these data should be more appropriate than GLWQG.
2.      The approach should be internally consistent. If a TEF approach is being applied to
       assess toxicity, then it should also be used to evaluate exposure. Otherwise, the
       improvements gained on effects may be offset by uncertainties and errors associated
       with modeling exposure.
3.      I suggest that a Monte Carlo approach be used. The approach should adhere to recent
       EPA policy concerning the use of probabilistic methods. A policy decision will need to
       be made concerning level of protection. Typically, this is selected as a value at the tail of
       the distribution (e.g., 95th percentile.) In lieu of Monte Carlo analyses, other probabilistic
       methods may be helpful.

                                             Christopher D. Metcalfe, Ph.D.
                                          Chair, Environmental & Resource Studies
                                                               Trent University
                                           Peterborough, Ontario, Canada K9J 7B8
                                                     E-mail: cmetcalfe@trentu.ca
Dr. Metcalfe is  currently  chair of Environmental and  Resource Studies at Trent
University in Peterborough, Canada.  He received a B.Sc.  degree at University of
Manitoba in zoology/chemistry, a M.Sc.  in  biology from  the  University of New
Brunswick, and a Ph.D. in biochemistry from McMaster University.  He is a recent
winner of the "Excellence in Research and Technology" award from the Ontario Ministry
of the Environment.  Dr. Metcalfe has a range of experience in international projects
concerning aquatic contaminants. Particular research interests are in  determining fate
and  toxic  effects  of halogenated  aromatic hydrocarbons, polynuclear aromatic
hydrocarbons, and alkylphenol ethoxylate surfactants in the aquatic environment.  Dr.
Metcalfe has been involved  in international  projects  in Belize, Mexico,  Ecuador,
Argentina, and Indonesia, as well as conducting several research projects on toxicity
in the Great Lakes. He has published more than 70 journal articles and other refereed

                                                             C.D. Metcatfe

                  Response to Charge Questions
                        from Chris Metcalfe
          Trent University, Peterborough, Ontario, Canada

                           November 14,1997

I. Stress-response profile related to the derivation of TEF values:

Question 1. The additional information for derivation of TEFs provided in the case
studies informs the reviewers that the toxic endpoints of interest  in these case
studies are reproductive success and recruitment within the populations of exposed
organisms.  This informed me to place greater emphasis on TEFs that have been
derived using toxic endpoints that affect recruitment, such  as early life stage

Questions 2 and 3. TEFs will vary in level of certainty. There is a good toxicity data
base with in vivo and in vitro mammalian models from which TEFs for wild mammals
can be derived. For fish, there is a comprehensive data base for TEFs that are
based  upon early life stage mortalities in salmonids, but data for other in vivo
endpoints are incomplete. I  am particularity concerned about TEFs derived for
birds, which are mainly based on in vitro assays  using endpoints that are only
peripherally related to effects that are relevent to the assessment endpoints in the
case studies (i.e. recruitment).

II. Stress-response profile relative to the application of the TEQ approach:

Question 1.  The implications of assuming no dose-additivity or no  interactions in
the case studies are a major leap of faith for the risk assessment process.  The

                                                            C.D. Metcatfe
limited information on this subject  indicates that other non-toxic halogenated
aromatic hydrocarbons (HAHs) exert a modulating effect upon the toxicity of planar
HAHs; hence a TEQ-based risk assessment based upon an assumption of no
interactions will over-estimate the toxic risk to fish and wildlife. However, having
said this, risk assessments based upon concentrations of TCDD or total PCBs
would  offer no major advantages  over the TEQ approach.  Basing  toxicity
assessments upon TCDD concentrations would be problematic in the retrospective
case study where planar PCBs contribute to a large percentage of the total TEQ,
and in the prospective  study where chlorinated dibenzofurans  are major
contributors to the total TEQ. Estimates based upon total PCBs do not take into
account the changes in congener proportions that take place through a process of
partitioning in the environment and biomagnification through food-webs.

Question 2.  For me, estimates based  upon EC50 or LC50 values are not a problem
for calculating  NOAELs.  As stated in the documentation for this exercise, the
dose-response curves for planar HAHs tend to be so steep that there are not likely
to be large differences in ECSOs and NOAELs. The use of median response levels
for risk assessment based on NOAELs will add a safety factor that partially
compensates for the uncertainties that are inherent in the TEF estimates.
Question 3.  There are some problems in extrapolating TEFs based upon tests with
a limited number of test species to an entire taxonomic group.  In the case of fish,
TEFs based upon early life stage  mortalities with salmonids are particularily
appropriate for assessing risk to salmonid species of esthetic or economic value;
a situation that is common for assessing risk in temperate lakes. However, these
TEFs may be of limited value for risk assessment in warm-water environments with
species such as bass and channel catfish. The mustelids appear to be particularily
sensitive to the toxic effects of planar HAHs, so risk may be underestimated for

                                                             C.D. Metcalfe
these mammals when using TEFs based upon rodent models. The limited amount
of  data available on TEFs for birds indicates that interspecies differences in
sensitivity are large, so applications of TEFs to the avian species identified in the
case studies may be inaccurate; either under- or overestimating the toxic risk.

III.  Exposure Profile:

Question 1: The TEF approach presents challenges for modeling the environmental
distribution and  exposure dynamics  of planar HAHs.  I  am not particularity
concerned with the quality of the physico-chemical data for these compounds.  In
most cases, there are adequate data for Kow, Koc, H, etc. for each of the toxic
compounds,  and where there is not,  estimates can  be made from empirical
relationships or structure-activity relationships.  However, I am concerned that there
are few data on the relative rates of biodegradation of these compounds. There
may be a tendency to model bioaccumulation and biomagnification of planar HAHs
solely on the basis of ability to partition into lipids (or fugacity); forgetting the effect
of biotransformations and excretion on this process. We particularity do  not
understand the relative biotransformation capabilities of various taxa,  since it
appears that different groups of organisms (e.g. fish-eating birds;  marine mammals)
may have different mertabolic capabilities for PCB congeners, PCDDs and PCDFs.

Question 2: It is difficult to assess the effect of differences in exposure routes on
estimates of TEFs.  For instance, injections of eggs in studies with fish and birds
may not reflect the normal toxicokinetics and partitioning of contaminants that occur
in eggs as a result of parental transfer of contaminants. More work  is needed to
assess this problem.

                                                            C.D. Metcalfe
Question 3:  The methods required for analysis of specific congeners of PCDDs,
PCDFs and PCBs, (in particular, coplanar PCBs) are definately more rigorous, time-
consuming and expensive than methods for aggregate stressors. This means that
only a small number of analytical labs with appropriate technical expertise and
analytical instrumentation (e.g. high resolution GC-MS) will be able to provide the
data that is appropriate for risk assessment, and tight research budgets will limit the
number of samples that can  be analyzed.  In  addition, some of the analytes
identified in these risk assessment scenarios are often not routinely analyzed (e.g.
PCB congener 81).        ;

.IV. Risk Characterization:

Question 1: In  my opinion, uncertainties  in modeling the  bioaccumulation and
biomagnification of planar HAHs are a limitation of the risk assessment process that
may exceed the uncertainties associated with calculating the TEFs.
Question 2:  In vitro or in vivo biological assays to determine TEQs may be a useful
approach. However, protocols must be developed to define the degree of sample
fractionation prior to the assays. For instance,  typical in vitro assays for EROD
induction with H4IIE cell lines have been conducted with environmental samples
that have undergone considerable fractionation to isolate  planar HAHs.  Use of
these samples may result in overestimates of biological responses. Use of a more
crude fraction also containing non-toxic  PCBs, for instance, may significantly
modulate the degree of EROD induction.

Question 3: Studies are needed to address:
       i)  The appropriateness of an additive approach for estimating total TEQs.
       ii) The relative rates of transformation and elimination of planar HAHs in

                                                            C.D. Metcaife
      different taxa, and the effects upon bioaccumulation and biomagnification.
      iii) The influence of exposure route on estimates of TEFs.

Additional Questions:
Questions Specific to Prospective Case Study:
Questions 1 and 2: No comments until I can further examine the basis of the BAFs
used by the GLWQG.

Questions Specific to Restrospective Case Study:.
Question 1: I would council a risk manager to develop TEQ sediment cleanup goals
that ensure protection of the vertebrate group with the most certainty in TEQ
estimates. In my opinion, the lack of certainty in TEQ estimates (e.g. 10-20 fold?)
would probably exceed the differences in sediment cleanup goals calculated for the
various vertebrate groups.

Question 2: I do not consider a ratio of total TEQs to total PCBs to be an effective
method  for setting TEQ-based sediment remediation goals.  The reason for this
opinion is illustrated in the attached figure (from Metcaife and Metcaife, 1997, Sci.
Total Environment) that shows variations in total TEQs for coplanar PCBs relative
to total PCBs in different components of the Lake Ontario food web. The ratio varies
among environmental compartments and groups of biota; probably as a result of
differences in metabolism and bioaccumulation of coplanar PCBs relative to other
PCB congeners. This is especially noticeable when comparing TEQ/PCB ratios in
biotic and abiotic compartments.

                                                                                       C.D. Metcalfe
                 T.L. Metcalfe, C.D. Metcalfe / The Science of the Total Environment 201 (1997) 245-272

                                              TEQ/Total  PCB
                     Water   Sed.    Plank Diporeia  Mysid  Sucker' Sculpin   Alwf   Smelt   Trout   Gull
Fig. 11. Ratio of Toxic Equivalent Quantities (TEQs) calculated for total mono-wife) and non-ortho PCB congeners relative to

total PCB concentrations in water, sediment and biota from the L?kf Ontario food-web.


                                                   Michael W. Meyer, Ph.D.
                                                            Wildlife Toxicologist
                                            Bureau of Integrated Science Services
                                       Wisconsin Department of Natural Resources
                                                       107SutliffRoad, Box 818
                                                         Rhinelander, Wl 54501
                                                 E-mail: meyerm@dnr.state.wi.us
Dr. Michael Meyer received his B.S. in biology from the University of Wisconsin -
Stevens Point, his M.S. in animal science from Texas A&M University System, formerly
known as Texas A&l University, and his Ph.D. in wildlife ecology from the University of
Wisconsin - Madison. Dr. Meyer is a wildlife toxicologist at the Bureau of Research for
the Wisconsin Department of Natural Resources.  Dr. Meyer has studied and written
many articles on bald eagles and loons in Northern Wisconsin, including Patterns of
common loon (Gavia  \mmer) mercury exposure,  reproduction,  and survival  in
Wisconsin; and A geographic trend in mercury exposure measured in common loon
feathers and blood.

                                                              Michael W. Meyer
Response to Charge Questions for Workshop on the Application of TEFs to Fish and

Section I.  .
1.      Because of the variety of species and endpoints used in developing TEFs,
additional information describing TEF derivation is required for uncertainty analysis.
Ideally, standard protocols would be established for congener specific TEF
derivation (same species, same endpoint) and a TEF profile be established for,each
class (mammal, fish, bird) Unless or until this is established, additional information
should be provided for all TEFs which comprise > 10% of a calculated TEQ,
including endpoint, species,  and study citation. The effect of TEF rounding on the
risk assessment process should be investigated via model sensitivity analysis.

2.      If the TEF is derived from an enzyme induction endpoint, from QSAR studies,
or  if multiple TEFs have been calculated for the same congener in different studies,
an uncertainty value should  be assigned to the TEF. Perhaps a "sliding scale" of
uncertainty could be assigned to all TEFs comprising  > 10% of a calculated TEQ
(i.e.  zero uncertainty assigned to TEFs that are derived from embryo toxicity studies
using the "target" species, with more uncertainty added incrementally as
quantitative rigor diminishes).

3.      TEF values developed using in vivo early life stage endpoints for relevant
species can be directly used to  predict a stress response in a risk assessment.
However I am skeptical of using biochemical responses unless they have been
closely correlated to a toxic endpoint in a relevant species.

Section II.

1.      The implications are  that no antagonistic or synergistic effects are occurring
between a complex mix of congeners as they compete to bind with the AhR
receptor - if synergistic effects  do occur the risk assessment would be too
permissive, if antagonistic effects occur it would  be too conservative.  If one where

                                                             Michael W. Meyer
to use the total PCS and 2,3,7,8-TCDD no-effect thresholds presented in the
Retrospective Case Study (i.e. 5 ug PCB, 100 ppt TCDD/g bird egg), the Caspian-
tern eggs collected from Oneofakind Lake would be close to the no-effect threshold.
However, the calculated egg total TEQ for the same Caspian tern eggs exceeds a
reported total TEQ avian'egg no-effect level (calculated by Giesy et al.1995 Arch.
Env. Com. 29:309-321) by a factor of nearly 60.  This reflects the author's
establishment of a 2,3,7,8-TCDD no effect level of 7 ppt vs. 100 ppt in  the
Retrospective Study.

2.     Calculating a NOAEL from the slope of the LC50 or EC50 dose-response
curve may not protect the most sensitive individuals in a population.  This could be
permissive if the risk assessment targets an endangered or declining species.

3.     The existing data for birds and mammals indicates that use of TEFs derived
from chicken or mink studies will provide highly protective, conservative calculated
TEQs. Chicken and mink are nearly an order of magnitude more sensitive to TCDD
TEQs than other species within their respective classes, and wild mink may
consume a limited amount of contaminated fish in their natural diet. Establishment
of conservative TEQ standards is desirable from the perspective of the risk assessor
and the resource, but will predictably result in controversy amongst the regulated
community.  If a permitting process uses the most conservative calculated TEQ to
establish effluent discharge, and achieving that new discharge goal requires
substantial capital investment by the regulated parties, you can expect litigation and
delay in implementation of the new rule. The cost-benefit of this trade off should be
addressed from a policy perspective.

Section III.

2.     One will need to assume that assimilation  efficiency and detoxification/
metabolism routes are similar when  one pools TEFs derived from various dosing
(injection, oral gavage, dietary) experiments.  One also then needs to assume that
wildlife contaminant exposure in the natural environment will result in similar

                                                             Michael W. Meyer
assimilation, metabolism, and effects patterns.   These assumptions should be kept
in mind when establishing a TEF.

3.     Sediment, soil, and biota will likely have differing congener patterns within
the same environmental system due to differential metabolism/degradation of the
various PCB/TCDD congeners present in the parent contaminant. An understanding
of these differences may be necessary to  predict risk within the various biotic and
abiotic compartments of an ecosystem, requiring additional sampling and analysis
costs. " A biologically based TEQ assay may be the preferred route to travel (see
Section IV #2).

Section IV.

1.     To be quite frank, it is difficult to answer this question without simulating the
risk assessment for the various contaminants and species of concern.  In most risk
assessments there is a great deal of uncertainty in describing exposure (limited diet
studies for few species, few prey items characterized  to congener content, etc.) and
effect (species sensitivity, endpoint characterization, etc.).  How that uncertainty
compares to that generated by extrapolation of TEFs between species and endpoints
is beyond the capability of my hand calculator.

2.     Biologically-based TEQ assays are by far the  best conceptual approach and
most economical means of using TCDD TEQs in the regulatory process. The cost
associated with collecting the data required to conduct a calculated TEQ risk
assessment may prohibit a meaningful characterization of exposure in most
scenarios.  Therefore a bioassay would be cost effective. Furthermore, the bio-
based TEQ would theoretically account for antagonism/synergism between
congeners in complex mixtures. Unfortunately, none of the existing bioassays
appears ready to go on line for routine screening in risk assessment exercises. For
instance, though bioassay TEQ values (using rat liver hepatoma cell line H4IIE) and
congener specific calculated TEQs  were very similar in an experiment where  mink
were fed diets containing Saginaw Bay carp, suggesting the additive assumption to

                                                             Michael W. Meyer
be correct (Tillet et al. 1996. Env. Science Tech. 30: 283-291), large discrepancies
exist between bioassay TEQ values and congener specific calculated TEQs values in
birds (Tillet et al. 1991. Arch. Env. Tax. Chem. 21:91-101). An understanding of
these inter-class differences and receptor binding mechanisms is necessary before a
bioassay can be implemented.

3.     Suggested research/site specific data
       a.     Establish a standard protocol for deriving TEFs (early stage mortality
       endpoint) and determine  the TEFs for the most relevant congeners for all 3
       b     Establish standard bio-based TEQ assay and conduct research to
       understand mechanisms  responsible for different results between classes.
       c.     Establish protocol for quantifying site-specific
       biomagnification/bioaccumulation factors to quantify TEQ exposure including
       dietary habit studies, prey base contaminant characterization, magnitude of
       trophic level  biomagnification, etc.
       d.     Conduct additional research to provide scientifically defensible TEQ
       effect levels  for mammals, birds, and fish if current data is insufficient.
       Investigations of potential TCDD TEQ interactions with residual DDE  in Great
       Lakes systems is also desirable.

Prospective Case Study

 2.     It has been shown that non-ortho PCB congeners are more readily
bioaccumulated and are more resistant to metabolism when compared to ortho-
substituted PCB congeners. It follows that wildlife tissues may contain a larger
proportion of dioxin-like PCB congeners/g total PCB, enriching the toxic potency of
the total PCBs measured in their tissues.  While not firmly established, it is also
likely that species differ in their ability to assimilate/metabolize the various PCB and
TCDD congeners.

3.     A risk assessment model should be developed which simulates exposure and

                                                             Michael W. Meyer
effect thresholds under a ranges of values which reflect the uncertainty inherent in
the model and its parameters.  This model output should then produce a range of
possible TeqTMDL with associated risk attached  (zero risk for the lowest value,
"x"risk for the greatest value).  Once this range has been established, a final rule
can be developed which is most protective of the ecological concerns while utilizing
the best available technology.

Retrospective Case Study

1.     It seems obvious that the variability in BSAFs, as well as thresholds of effect
between vertebrate groups, will result  in different sediment clean up goals. In
addition, it does not seem possible to provide a scientifically defensible TCDD TEQ
threshold of effect for Caspian terns and river otters as their sensitivity to these
compounds has not been experimentally established. Indeed, it seems that Caspian
terns are insensitive to the embryo toxic effects of PCBs (hatching success was not
depressed despite eggs PCB levels of 19-40 ug total PCB/g wet weight; Struger and
Weseloh,  1985, Colonial Waterbirds 8:142-149).  No data is currently available on
the relative sensitivity of otters to TCDD TEQs  as compared to mink though rumor
has it such work is  underway. I would therefore council the risk manager to go with
the sediment TCDD TEQ value that protects lake trout, a species whose TCDD TEQ
for early life stage mortality is well characterized. I would then request that the
responsible party support a dose-response study for river otters/TCDD-TEQs, or, at a
minimum, a study which compares river otter TCDD sensitivity to that of the mink.
I'd exclude the Caspian tern from the risk assessment because of their insensitivity
to the toxic effects of PCBs.

                                                  Patrick W. O'Keefe, Ph.D.
                                                             Research Scientist
                                         Laboratory of Organic Analytical Chemistry
                                                             Wadsworth Center
                                                  New York Department of Health
                                                            Empire State Plaza
                                                                 P.O. Box 509
                                                             Albany, NY 12201
                                                  E-rnail: okeefe@wadsworth.org
Dr. Patrick O'Keefe received his B.S. in chemistry from University College in Dublin, his
M.S. in food science from Cornell University, and a Ph.D. in food science from Oregon
State University. Dr. O'Keefe is an assistant professor at the school of public health
and a research scientist for the Wadsworth Center for the State University of New York
in Albany.  Prior to his current work, Dr. O'Keefe was a  research fellow for Harvard
University, a food scientist for ITT Continental Baking, and a research scientist for
Battelle Northwest Laboratories. He is a member of the American Chemical Society
and  is a consultant to the Science Advisory Board.  Dr. O'Keefe has  published
numerous  articles on PCDDs, PCDFs,  and  related compounds  in contaminated
sediments and biota.

                                                              Patrick W. O'Keefe, Ph.D.
       USEPA Workshop on the Application of TEFs to Aquatic Life and Wildlife
                 January, 20-22,1998, Chicago, Illinois
                    Answers to Premeeting Questions
                      Patrick W. O'Keefe, Ph.D.
                        Wadsworth Center
                New York State Department of Health
                 PO Box 509,  Albany, NY 12201
1.      In addition to uncertainties in the TEFs themselves, there are numerous challenges and
uncertainties associated with the application of TEFs to environmental risk situations. In the
prospective study the consultant choose to calculate permitted concentrations in water (TEqCw
values) based on the initial premise that each compound contributed alone to the TCDD toxic
equivalence. In the final step it appears that the mass loadings of each compound were then
distributed on the basis of their relative mass distribution in the effluent. This is a complicated
process and since I did not have access to the modeling program I was only able to carry out the
initial calculation for allowable water concentrations for fish as shown:
                                            )   2.49JS"+07x0.08xl
                        / 'pecdd^d'pecdd^    pecdd'

The fd values were determined using DOC , POC and K^ values as described in EPA-820-B-95-
005. The complete set of TEqCw values are shown below (pg/L):

                    Fish                Avian              Wildlife
1,2,3,7,8 PeCDD     0.019               0.021               0.0029

1,2,3,4,7,8 HxCDD   0.115               0.106               0.018

                                                             Patrick W. O'Keefe, Ph.D.
2,3,7,8 TCDF        1.2                 1.29                0..183

1,2,3,7,8 PeCDF      1.9                 2.05                0.289

2,3,4,7,8 HxCDF     0.026               0.028               0.004

1,2,3,4,7,8 HxCDF   2.28                2.45                0.348

It is apparent that with the DOC and POC values quoted in the prospective study there would be
considerable errors if the BAFd/ values from the background literature were applied without
converting them to BAF', values using fd values. For instance the TEqC'w values for PeCDD
would be increased by nearly a factor of 3.
       When the WASP4 model is applied to the calculation of maximum allowable loads
(MALg) for each congener, several other parameters are required hi addition to maximum
allowable water concentrations (MACW). These are sediment related parmaters (settling flux,
respusension flux, log KOC etc.) and two important physico-chemical parameters, the Henry's
Law constant for vapor/water partitioning and the photolysis rate constant. In the Lake Ontario
TCDD study it was determined under steady-state conditions, that for a given annual load to the
lake  from the Niagra River, 6% would be transported out of the lake via the St. Lawrence River,
25% would be incorporated into the bottom sediments, 31% would volatilize and 38% would
undergo photolysis. The Henry's Law constants have been determined by a number of
investigators and are known with reasonable accuracy for several PCBs and PCDDs/PCDFs,
although only a very small number of P«CB values are shown in the physico-chemical parameters
table (Table 1). Since PCBs have absorption maxima at wavlelenghts below the lowest sunlight
waylelenght (less than 300 nm) photolysis may not be as important a removal process for these
compounds as it is for PCDDs/PCDFs which absorb light in the UV-B region (300-340 nm).
Based on studies carried out in the laboratory using 50/50 acetonitrilerwater solutions a value of

                                                              Patrick W. O'Keefe, Ph.D.
0.002 was selected for the quantum yield ((J>) for TCDD in the Lake Ontario study. Our studies
confirmed this value for both pure water and acetonitrile/water solutions photolyzed at 300 nm in
the laboratory. However for reasons that are not completely clear quantum yields for PCDDs and
PCDFs are an order of magnitude higher in sunlight than at 300 nm, a finding corroborated by
work carried out in the laboratories of Derek Muir and Barrie Webster. Furthermore Dung and
O'Keefe (Environ. Sci. Technol. 28: 549-554, 1994) and Friesen et al (Environ. Sci. Technol. 24:
1739-1,744, 1990) have shown that dissolved organics potentiate the photodegradation of
PCDDs/PCDFs. Taken together these studies show that photodegradation should be given
serious consideration in any studies modeling the transport and fate of PCDDs/PCDFs in the
aquatic environment. A major uncertainty at the present time is the lack of knowledge on the
extent to which PCDDs/PCDFs photodegrade when they are bound to suspended sediment

2. In mammalian and avian species the extent of absorption and the tissue distribution of toxic
compounds do not appear to differ significantly between i.p. and oral routes of administration.
However if toxicity to fish eggs is used as an endpoint for risk characterization and the water
quality standard is based on tissue residue measurements in whole fish,  then it must be rioted
that concentrations in fish eggs are 2 to 3 times lower than maternal tissue concentrations.
3. Since measurements of individual congeners are required for the TEF approach analytical
methods must be much more rigorous than those used in the determination of total compound
type concentrations. In the case of PCBs, the non-ortho (coplanar) congeners are the most toxic
but they may only constitute 1% of the total PCBs. However PCBs are generally analyzed using
GC/EC instrumentation, a relatively nonspecific analytical technique. Under these circumstances
the trace signals from the coplanar congeners may be obscured by coeluting diortho congeners.
Therefore relatively complex cleanup methodologies based on carbon chromatography must be
used to separate the coplanars from the noncoplanars prior to GC/EC analysis. In PCDD/PCDF

                                                              Patrick W. O'Keefe, Ph.D.
analysis individual congener identity is accomplished more readily since analyses are carried out
by GC/MS and isomer identity within a given chlorine group is simplified by the fact that most
biota accumulate predominantly 2,3,7,8-substituted isomers. Rigorous cleanup methodologies
are still required since PCBs are generally present in environmental samples at higher
concentrations than PCDDs/PCDFs (ng/g - ug/g vs. pg/g) and certain PCS congeners can
interfere with the analysis of PCDDs/PCDFs by low resolution GC/MS.


1. As pointed out above in answer to Question 1 on Exposure Profiles, there are some major
uncertainties associated with some of the physico-chemical parameters used for modeling the
fate and transport of PCBs and PCDDs/PCDFs. On the other hand there are uncertainties
associated with the TEFs determined for avian and mammalian species. In the case of both avian
and mammalian species an uncertainty factor of 10 was used to adjust NOAEL levels from
subchronic exposure studies conducted in the laboratory to chronic exposures which biota would
experience in the field. Is this appropriate ? Another uncertainty relates to the nature of the diets
consumed by predatory avian and mammalian species. If a diet contains a high proportion of
fish-eating biota then food chain biomagnification can be significant and the Great Lakes Water
Quality Criteria may severely overestimate MAC'W values. On the other hand MACW values can
be underestimated if plant-eating terrestrial biota are consumed.

2. If the TEFs are indeed additive then biologically-based TEQ assays may not provide any
additional information from a risk assessment viewpoint. However there  are several literature
citations in the Interim 1993 Report on TCDD Risks in Aquatic Life and Wildlife (p 4-4)
suggesting that an additive model may not always be appropriate. Recent studies conducted at
the Wadsworth Center, NYSDOH have shown that PCBs  126 and 169 inhibit TCDD-induced
estradiol (Ej) metabolism by hydroxylation at both the 2 (CYP1 Al activity) and 4 (CYP1B1

                                                               Patrick W. O'Keefe, Ph.D.
activity) positions in certain human cancer cell lines (Shaokun Pang, Ph.D. Thesis, 1997). In the
absence of TCDD the two PCB congeners induced E2 metabolism. Under these circumstances a
biologically-based assay would provide more definitive information on the risks associated with
a defined mixture of compounds. On the other hand a nonadditive model would make it
extremely difficult to regulate the compounds on an individual basis. Perhaps the most useful
approach would be use a bioassay to adjust TEF values. As pointed out in the Interim Report the
bioassay would also need to be calibrated against a biological endpoint of environmental

3. Since the total PCB concentrations in gull eggs from Roundtail Lake approach 3 u.g/g a major
concern would relate to the coplanar and mono-ortho PCB concentrations in the gull eggs and
also in sensitive mammals such as mink and otter. In the retrospective study a total PCB
concentration of 5.7 ug/g in Caspian tern eggs results in a TEQ value of 400 pg/g, which is xlO
higher than the no-effect threshold. In conjunction with these monitoring studies the state should
determine the status of the river otter population in the area. Results from the population survey
may indicate that the more stringent WQS for river otters should be adopted in the risk
assessment. The next research priority would be the determination of BAF1/ values for those Ah
active PCBs which are present in high enough concentration to be measured in water samples.
Since the fish currently in Roundtail Lake have no detectable concentrations of PCDDs or
PCDFs additional monitoring of biota or sediments for these compounds is probably not
warranted. However it would be appropriate to determine PCDD/PCDF residue levels in biota
from other lakes where there are discharges of known magnitude from pulp and paper mills.
       In the retrospective study, the fact that the TEQ concentrations in the Caspian tern eggs
and the otter livers exceed the NOAELs for these species is a major concern. Three research
questions  arise from this concern: (1) Could the low success rate in Caspian terns be explained
by interactions between the DDE residues in the tern eggs and the PCBs/doxins ? Field
observations coupled with residue measurements might useful information, (2) The manager of

                                                              Patrick W. O'Keefe, Ph.D.
the area should determine if the anecdotal accounts of low numbers of mink and otter are valid,
and (3) the proportions offish and fish-eating biota in the diet of mink and otter should be
assessed. Consumption offish-eating biota could have a considerable impact on the
biomagnification of PCBs and dioxins by these mammals.

Additional Questions on the Prospective Study

1 .Since there is no information on either dissolved or total aqueous concentrations of the
chemicals, field derived BAFs cannot be derived. However field-derived BSAFs can be
determined for PCB 77 and PCB  126 lake trout and sediment data and it is possible to use this
information to obtain a ratio of the BAFd/ using the following equation:
This value can then be compared to the ratio determined from the GLWQG document. With / =
PCB 126 and r = PCB 77 the ratio was determined to be 5.7 compared to a ratio of 37.5
determined from Table 1. Therefore we might expect some errors if the GLWQG values are

2. The bald eagle data are most suitable for this type of comparison since the data can also be
anajyzed by the BMP approach described on pages 33 through 44 of the USFWS Critique of the
GLWQG Document. Basically the established water standard for bald eagles needs to be divided
by 21, the BMP for forage fish to bird eggs. This will adjust all the dietary components for

                                                              Patrick W. O'Keefe, Ph.D.
biomagnification from forage fish, using the equation on page 3-13 of the GLWQI Criteria
Document for TCDD (the BMFs for forage fish to piscivorus fish and piscivorus fish to bird eggs
have already been taken into consideration by using two BAFs and the factor of 30,
respectively). The BMF-adjusted WQS then becomes 0.0013 pg/L instead of 0.028 pg/L. Using
the chicken embryo NOAEL of 100 pg/g for bald eagles together with a safety factor of 10 the
same WQS can be determined by the USFWS approach as shown below:
           NOAEL              lOOxlO3
                                                - =TargetDietaryConcentration
           totalBMF  0.736x21 +0.184x10+0.056x659
              TargetDietaryConcentration   184Qpg/kgforagefish
              	=	0.001 IpgIL
                        BAF            172,100kgforagefish/l
       It is more difficult to determine a BMF-adjusted WQS for mink since the diet-to-mink
BMF is presented on a lipid basis in Table 1. If we assume that the lipid concentration in the
mink is 4% compared to a lipid forage fish concentration of 8% then the wet weight BMF for
TCDD would be 5.5 and the WQS value should be divided by this number to give a BMF-
adjusted WQS of 0.0005 pg/L rather than the value of 0.00292 pg/L. However if the USFWS
approach is used assuming a NOAEL of 60 pg/g, as per the retrospective study, and a diet
composed exclusively of forage fish then the WQS would be 0.0063 using the xlO safety factor.
The discrepancy between the two approaches is partly related to the fact that the EPA method
uses a daily toxic dose (TD) whereas the USFWS uses the NOAEL body burden. The

                                                             Patrick W. O'Keefe, Ph.D.
NOAEL/TD ratio is 7 for the bald.eagle compared to 60 for the mink.

Additional Questions Relating to the Retrospective Study:

1. The cleanup goals would not be the same for each vertebrate group since the prder of
sensitivity of the groups is mammals>birds>fish. As shown in the retrospective study document
TEQs can be directly linked to sediment concentrations of the chemicals via BASFs. Therefore
the most restrictive sediment cleanup standard would be based on the otter TEQs and the extent
of cleanup required would depend on the extent to which the TEQs exceed the NOAELs.

2. This question can be addressed by considering the equation for calculating TEQs in birds and
mammals on page 8 of the retrospective study. In addition to the organic carbon-normalized
congener concentration in the sediment (Coc) and the appropriate TEF, this equation involves the
use of two partition coefficients, a BSAF and a BMP. When the shiner BSAFs were determined
from the field  data and the BMFs from Table 1 were used TEQs were obtained which can be
compared with the TEQs listed in Tables 2 and 3:

Caspian Tern Eggs
                          TEQCalc.          TEQTable2

      PCB77             29                 54
      PCB 126            232                275
      2,3,7,8-TCDD       2.3                 4.5
      2,3,4,7,8-TCDF      4.1                 9.58

                                                             Patrick W. O'Keefe, Ph.D.
       PCB 105
       PCB 126
TEQ Calc.
TEQ Table 3
       It is apparent that there is considerable agreement between the calculated TEQs and the
TEQs derived from tissue concentrations. However these data were obtained using field-derived
spottail shiner BSAFs. As shown in the table below the agreement  would have been much lower
if the EPA lake trout BSAFs had been used:
       PCB 77
       PCB 105
       PCB 118
       PCB 126
BSAF shiners
Oneofakind Lake
BSAF lake trout
Oneofakind Lake
14   '
Table 1 Questions
       Consequently if generic BSAFs cannot be used, field-derived BSAFs must be determined
for individual congeners using state-of-the art analytical methods. If this were the case there
would be no savings in analytical costs by analyzing sediments for total PCBs and then
determining individual Coc values by a ratio calculation.

                                                Richard E. Peterson, Ph.D.
                                                       425 North Charter Street
                                                           School of Pharmacy
                                                        University of Wisconsin
                                                           Madison, Wl  53706
                                                            Fax: 608-265-3316
Dr. Peterson received his B.S. degree from the University of Wisconsin, and his Ph.D.
in pharmacology from Marquette University Medical School (now The Medical College
of Wisconsin).  He is currently a professor of toxicology and pharmacology at the
University of Wisconsin School of Pharmacy, as well as being a toxicology consultant
on the Science Advisory Board of EPA. He was a recipient of a Research Career
Development Award from the National Institutes of Health and sits on the  editorial
board of Toxicology and Applied Pharmacology.  His fields of specialization include
toxicology of halogenated aromatic and halogenated aliphatic hydrocarbons in fish,
avian,  and  mammalian species; the  neurotoxicology of halogenated  aromatic
hydrocarbons; and   bioaccumulation,  tissue distribution, biotransformation,  and
elimination of halogenated aromatic hydrocarbons.  He is frequently invited to
participate on review committees and projects concerned with halogenated aromatic
hydrocarbon toxicology  issues by NIH, EPA,  NOAA, and  other major research
organizations. Dr. Peterson was the chair of the wildlife workgroup at the World Health
Organization's "Meeting  on the Derivation of Toxic Equivalency Factors for PCBs,
PCDDs, PCDFs, and Other Dioxin-like Compounds for Humans and Wildlife" held on
June 15-18,1997. Dr. Peterson has more than 150 publications in his fields of study.

                                                                Richard E. Peterson
Specific Questions/Issues

1. Inclusion of the WHO draft report (July 30, 1997) on derivation of toxic equivalency factors
(TEFs) for polychlorinated biphenyls (PCBs), dibenzo-p-dioxins (PCDDs), and dibenzofurans
(PCDFs) for humans and wildlife is useful in evaluating uncertainties in the fish-, bird-, and
mammal-specific TEFs as they relate to the retrospective and prospective case studies.
       In the case studies the fish receptors of concern are lake trout and bull trout.  By
consulting Table 3 of the WHO report one finds that fish-specific TEFs for PCBs, PCDDs, and
PCDFs were based on early life stage mortality in rainbow trout. This is useful information
because the fish receptors of concern are species of trout.  So there is a relatively Idw level of
uncertainty in extrapolating the fish-specific TEFs to lake trout and bull trout because they are a
closely related fish species to rainbow trout. In addition, the endpoint upon which the fish-
specific TEFs were based, early life stage mortality,  is relevant to recruitment which is the
assessment endpoint proposed for fish in the case studies.  Thus, in assessing the risk to
recruitment of lake trout and bull trout caused by exposure to PCBs, PCDDs, and PCDFs there
is a relatively low level of uncertainty in using the internationally agreed upon TEFs for fish.
       In the retrospective and prospective risk assessments the wild bird receptors of concern
are the Caspian tern and bald eagle, respectively, and the assessment endpoint is also recruitment
as it was for fish. In consulting Table 4 of the WHO report for bird-specific TEFs one observes
that the TEFs for PCDDs and PCDFs are based on induction of ethoxyresorufin-O-deethylase
(EROD) activity in the chicken embryo. This is important, because there is greater uncertainty
in these  EROD induction-based TEFs for PCDDs and PCDFs, with respect to the assessment
endpoint of recruitment, than there would be if they had been based on embryo mortality.  In this
context, it is useful to find in the WHO report that the TEFs for essentially all of the
environmentally relevant PCBs were based on the LD50 for embryo mortality in the chicken.
Thus, the bird-specific TEFs for PCDDs and PCDFs are more uncertain than those for the PCBs
when assessing the risk to recruitment of Caspian terns and bald eagles caused by the presence of
complex mixtures of PCDDs, PCDFs, and PCBs in the eggs.. In addition, it is important to note

                                                                  Richard E. Peterson
that all of these TEFs whether they were based on EROD induction or embryo mortality tended
to be based on results obtained in chicken which is the most sensitive of all bird species to aryl
hydrocarbon receptor (AhR) agonist toxicity. Yet the wild bird receptors of concern in the case
studies, the Caspian tern and bald eagle, are not closely related to the chicken. This may be
significant because it is uncertain to what extent TEFs determined in a highly sensitive species
like the chicken can be extrapolated to more TCDD insensitive and distantly related bird species
like the Caspian tern and bald eagle. It is concluded, in assessing the risk to recruitment in lake
trout and bull trout versus Caspian terns and bald eagles, due to exposure of the fish or bird
embryo to TCDD and related compounds, that there is more uncertainty in using the
internationally agreed upon bird-specific TEFs than the fish-specific TEFs.  Another point is that
of the two case studies, there is more uncertainty in estimating TCDD equivalents (TEQs) from
bird-specific TEFs in the prospective study. This is because the mill effluent of concern in this
case study is predicted to contain only PCDDs and PCDFs and bird-specific TEFs for these
classes of AhR agonists are the most uncertain for birds.
       In the two case studies the mammal receptor of concern is the river otter and the
assessment endpoint, recruitment, is the same for both case studies.  In consulting Table 2 of the
WHO report no information is provided on which species were used in the derivation of the
mammal-specific TEFs, however, from the text of the report it is clearly stated that the majority
of these TEFs were based on studies in laboratory rodent species. Furthermore, it is stated in the
report that relative  potencies of the PCBs, PCDDs, and PCDFs toward mink reproductive
toxicity are not different from those of the rodent models from which most of the data to derive
the TEFs were obtained. This interpretation is useful, because in the case studies the mammal-
specific TEFs will be used to determine TEQs in river otter liver and there is uncertainty in the
extent to which TEFs can be extrapolated across species as well as across endpoints. Another
way in which the WHO report was useful is that it demonstrates that the most rigorously
determined TEFs among the three vertebrate classes are those for mammals. In fish and birds a
TEF might be based on one study whereas in mammals the results of several studies using
different routes and durations of exposure are available for consideration in the derivation of
TEFs.  This leads to the conclusion that the  mammal-specific TEFs probably have less
uncertainty than those for fish and birds.

                                                                 Richard E. Peterson

2. Within and between the three vertebrate classes (fish, birds, and mammals) there is a range of
uncertainty in the TEFs determined for individual PCDD, PCDF, and PCB congeners. However,
if TEFs are acknowledged to be order of magnitude estimates of actual relative potencies for
AhR-mediated responses in a particular species it might not be necessary to apply an additional
uncertainty factor to the TEQs that are generated by the TEF approach in order to acknowledge  •
the uncertainty that exists in these estimates. In general the mammal-specific TEFs would
appear to have the least uncertainty because they are derived from a larger number of studies
than is the case for fish and birds. In the latter two vertebrate classes, it would seem that the
degree of uncertainty associated with the TEFs for PCBs is probably similar because the studies
that were relied upon to derive the TEFs for PCBs used an egg injection route of exposure and an
LD50 for embryo mortality as the basis for deriving the TEFs. However, for PCDDs and PCDFs
there is a significant difference between the endpoint used to derive the TEFs for fish and birds.
In the case offish the TEFs were based on embryo mortality whereas for birds they were based
on EROD induction which is more uncertain because it is an adaptive rather than a toxic
response.. Thus, the degree of uncertainty in the TEFs varies across vertebrate classes and would
appear to be less in mammals than fish and birds. The uncertainty in the TEF for a particular
PCDD, PCDF, or PCB congener is influenced by a number of factors including whether it was
based on an in vivo or in vitro study, species, route and duration of exposure, endpoint assessed,
and reproducibility of the results in  similarly designed studies.

3. The measures of effect in the case studies pertain to reproductive success as measured by
effects on egg production and viability and/or larval and offspring survival. Until evidence is
presented to the contrary for each vertebrate class (fish, birds, and mammals) it would seem to be
more uncertain to extrapolate TEFs based on cytochrome P4501A1 induction (determined in
vivo or in vitro) to these measures of effect, than to extrapolate from TEFs based on clearly
adverse developmental  and reproductive toxicity endpoints such as early life stage mortality in
fish, embryo mortality in birds, or a reduction in litter size in mammals.
        If one is going to rely on EROD induction based TEFs for PCDDs and PCDFs in birds,
which is currently the situation for bird-specific TEFs, then it would be prudent to show that

                                                               Richard E. Peterson
relative potencies (REPs) for a few of the most environmentally relevant PCDD and PCDF
congeners following injection of graded concentrations into bird eggs give rise to REPs for
EROD induction and embryo mortality that are within an order of magnitude of one another.
Also to the extent such information is available for PCBs in bird embryos it would be helpful to
include the findings in the WHO report.  If REPs for these two endpoints are similar for PCBs in
birds it would suggest this will probably also be the case for the PCDDs and PCDFs.

1. Assuming no additivity of the PCDD, PCDF, and PCB congeners that are AhR agonists would
underestimate the risk which exposure to this mixture of chemicals poses to recruitment of the
fish and wildlife receptors of concern in the two case studies.  Furthermore, if an alternative
method of ecological risk assessment, based on total PCBs relative to an Aroclor standard or
TCDD alone, were applied to the prospective and retrospective scenarios they would both
probably underestimate the risk to the fish, bird and mammal receptors of concern.
       In the prospective scenario, mill effluent will contain, in addition to TCDD, three
PCDDs and four PCDFs which are AhR agonists, but no PCBs.  Since the mill is not a source of
PCB contamination, the measurement of total PCBs would be inappropriate. Relying on the
concentration of TCDD alone in fish and wildlife tissues as the exposure metric has the problem
of neglecting the potential contribution to AhR-mediated toxicity of the other PCDD and PCDF
co-contaminants in the effluent which have the potential to bioaccumulate in the fish and wildlife
receptors of concern. For example, if TCDF which is present in the mill effluent at a 20 times
higher concentration than TCDD was found to bioaccumulate in bald eagle eggs to a higher
concentration than TCDD,  then TCDF would contribute more TEQs to the eggs than TCDD
(bird-specific TEF for TCDF = 1.0). However, this potentially greater contribution to egg TEQs
by TCDF would be  missed if the exposure analysis were based solely on TCDD. Thus, the risk
assessment conclusions reached from relying on total PCBs or TCDD alone would underestimate
the actual ecological risk posed by the discharge from this particular mill and would result in
higher concentrations of PCDDs and PCDFs being permitted in the effluent than would be
justified from an ecological risk perspective.

                                                                  Richard E. Peterson
         For the retrospective, study, involving the PCB spill it would not be appropriate to assess
 exposure offish and wildlife to these halogenated aromatic hydrocarbons by measuring TCDD
 alone because it was not a contaminant of the used hydraulic fluid. The major source of PCBs in
 the used fluid was Aroclor 1248.  However, it would also be inappropriate to monitor the impact
 of this spill on fish and wildlife by measuring total PCBs in the tissues of such animals. This is
 because Aroclor 1248 is contaminated with PCDFs which are not detected by measuring total
 •PCBs. Also the used hydraulic fluid might be actually enriched in PCDFs when compared to
 Aroclor 1248 if it was used at high temperatures that could result in PCDF formation. Thus, the .
 main point is that certain PCDFs that would be expected to be present in significant
 concentrations in the used hydraulic fluid that was spilled into the Yuckymuck River would not
 be detected by measuring total PCBs in fish and wildlife inhabiting the river.
         Another point is that weathering of the PCBs that were spilled into the river would result
 in PCB concentration profiles in the fish and wildlife receptors  of concern (lake trout, Caspian
 terns, and otter) that are different  from both that of the PCBs spilled and an Aroclor 1248
 standard that might be used to quantify total PCBs. In this regard it is possible that lake trout
 and Caspian tern eggs and otter liver will have greater concentrations of PCB 126 than are
 present in Aroclor 1248. This enrichment of these tissues in PCB 126, which is a major
 contributor to TEQs in this particular case study, might be missed if exposure to AhR agonists is
 based on total PCBs.  Thus, if TCDD alone or total PCBs were used to assess exposure in the
 retrospective case study the results obtained would underestimate, retrospectively, the risk to
 recruitment offish and wildlife caused by the spill.

2. There is less variability in the LC50 and EC50 on a dose response curve than there is in the
LC1 or EC 1 which are closer to the NOAEL.  Therefore, REPs based on the LC50 and EC50
should be more accurate than those based on a certain percent response at the lower end of the
dose response curve near the NOAEL in deriving TEFs. In my judgement TEFs derived in this
manner can be used in risk assessments where a NOAEL is being employed. The only exception
to this generalization is if TEFs are based on REPs for EROD induction in cell culture systems
where full dose response curves are unable to be generated for certain congeners. In those cases  it
has been recommended that REPs  for this particular response be based on ED 10 values.

                                                                  Richard E. Peterson

3. The assumption in deriving vertebrate class-specific TEFs (WHO, 1997) was that they could
be used to determine TEQs in fish, bird, and mammalian species, respectively, with less
uncertainty than if a single set of TEFs was used. However, significant uncertainty still remains
in extrapolating these new TEFs across species.  This is reflected in the new TEFs still being
referred to as "order of magnitude estimates". This certainly applies in directly extrapolating the
fish-specific TEFs (determined in rainbow trout) to lake trout and bull trout, the bird-specific
TEFs (based on studies in chickens) to Caspian terns and bald eagles, and the mammal-specific
TEFs (determined in laboratory rodents) to the river otter.'
       Those TEFs that are the most uncertain are the ones derived solely from either QSAR or
AhR binding affinity studies followed by TEFs that are based on CYP1A1 induction in vitro and
in vivo. Bird-specific TEFs are the most problematic in this latter regard because all PCDD and
PCDF TEFs for birds are based solely on EROD induction in the chicken embryo.  Thus, for the
Caspian tern and bald eagle there is more uncertainty in directly extrapolating the bird-specific ,
TEFs for PCDDs and PCDFs.
       Nevertheless there is such a paucity of studies, particularly in fish and birds, on the
magnitude of species differences in REPs for the same endpoint that it seems prudent to assume
for each vertebrate class  that the TEFs can be extrapolated across species with a one order of
magnitude uncertainty until there is evidence to the contrary. In support of this assumption sets of
TEFs for PCDDs, PCDFs, and PCBs, determined by various authors/agencies, were recently used
to determine TEQ concentrations in lake trout eggs from the Great Lakes. It was found that the
TEQs so determined varied by less than one order of magnitude in spite of the different sets of
TEFs that were used (Cook et al.,  1997). In light of these findings, and recognizing that TEFs are
one order of magnitude estimates, it would appear that class-specific TEFs can be directly
extrapolated to the fish and wildlife receptors of concern in the case studies.
       The REP determined for PCB 126 in rainbow trout eggs, based on the endpoint of early
life stage mortality, the Peterson laboratory has shown to be accurate in predicting the egg dose of
PCB 126 that caused early life stage mortality in lake trout eggs. Thus, between these two closely
related fish species, rainbow trout and lake trout, the REPs for PCB 126 were quite similar.
Whether this will be the case for fish species that are not as closely related is not known, but is an

                                                                  Richard E. Peterson
important area for future research.

1.  The TEF approach, in and of itself, does not present new uncertainties or modify old
uncertainties associated with modeling the exposure of AhR agonists, because it is not applied
until after the concentration of an AhR agonist has been estimated in a particular tissue for the
receptors of concern (i.e., fish egg, bird egg, or mammal liver).

2.  The TEF of a particular congener is based on its potency for producing a particular response,
relative to that of TCDD, when both compounds are administered by the same route.
For fish and birds, the majority of TEFs were based on the egg injection route of exposure. This
is significant because the concentrations of PCDDs, PCDFs, and PCBs in eggs when multiplied
by such TEFs should give rise to a TEqC that has a greater level of certainty associated with it
than if a different route of exposure had been used. Also in  lake trout it has been shown that the
potency of TCDD in causing early life stage mortality is essentially identical irrespective of
whether TCDD is transferred naturally from the female to the oocytes prior to spawning, is
directly injected into the egg, or is taken up by the egg following waterbome  exposure to TCDD.

3.  No comment.

1.  Uncertainties associated with the TEFs are not more problematic than the  other sources of
uncertainty in the ecologic risk assessment nor do they limit the means of performing the
assessment.  In my judgement uncertainties associated with  estimating exposure to the various
PCDD, PCDF, and PCB congeners, retrospectively and prospectively, are greater than  those
associated with the TEFs.

2.  The H4IIE bioassay, has the advantage over the TEF-based approach of assessing interactive
effects of AhR agonists. The endpoint of such a bioassay is that a "net" AhR-mediated response
in cell culture, such as induction of cytochrome P4501A1 activity, is determined. TEQ

                                                                 Richard E. Peterson
concentrations are then estimated by comparison to a TCDD standard curve after appropriate
corrections for dilution of the tissue extract are made,  Another strength of the TEQ bioassay is
that it is relatively' inexpensive when compared to congener-specific GC/MS, and can be used,
therefore, to screen a large number of samples for high concentrations of TEQs in a more cost
effective manner. The weakness of the method is that it will detect other AhR agonists that are not
PCDDs, PCDFs, and PCBs, such as polyaromatic hydrocarbons and several other classes of
compounds, and can lead to false positives.
       The two approaches could be integrated if the H4IIE bioassay, the recently developed
CALUX bioassay, or an equivalent, validated, TEQ bioassay were used to screen large numbers
of environmental samples for TEQs. Congener-specific GC/MS which is more cost prohibitive
could then be reserved for confirming PCDD, PCDF, and PCB congener related AhR agonist
activity in only the most highly contaminated samples and for confirming reduced AhR agonist
activity in designated "cleaned up" media such as lake or river sediments in the case studies.

3. List of specific studies that would reduce uncertainty in the case study assessments (ranked
from highest to lowest priority)
1.     There is a need for determining a NOAEL and LOAEL for effects of TCDD in reptiles
       (snakes and turtles) and amphibians as well as TEFs that can be used to determine TEQs
       for these species.
2.     There is a need for a laboratory-conducted, dose response, developmental and
       reproductive toxicity study in mink exposed in utero and via lactation to TCDD or PCB
        126 alone. Such a study does not exist causing uncertainty in the NOAEL and LOAEL
       used for TCDD in piscivorous mammals.
3.     There is a need to determine the NOAEL and LOAEL for TCDD and PCB 126 in bull
       trout. The bull trout is related to the lake trout, the most sensitive fish species to TCDD-
       induced early life stage mortality.  Given its threatened status, it might be significantly
       more sensitive than lake trout to TCDD-induced early life stage mortality. If so,
       determination of the NOAEL and LOAEL for bull trout might change the conclusion of
       an ecological risk assessment which otherwise would  have relied on the higher NOAEL
       and LOAEL for TCDD in lake trout eggs.

                                                                 Richard E. Peterson
4.     For fish and birds there is a need to conduct cross-species comparisons of REPs based on
       a population relevant endpoint such as embryo mortality. This should be done for those
       PCDD, PCDF, and PCS congeners that are generally considered to be the major
       contributors to the TEQ concentrations in fish and bird eggs in North America.  The
       question to be addressed is: for each individual congener tested, in fish and bird species of
       widely differing sensitivity to TCDD-induced embryo mortality, will the REPs vary by
       more than one order of magnitude?
5.     There is a need to determine, for a wide variety of environmentally relevant egg or body
       burden mixtures of AhR agonists in fish and wildlife, if in ovo exposure (fish and birds)
       and in utero and lactational exposure (mammals) causes  population relevant signs of
       toxicity (i.e., developmental and/or reproductive) by an additive interaction.
6.     TCDD embryotoxicity studies need to be conducted in long-lived aquatic species that live
       in close contact with contaminated lake and river sediments such as snapping turtles and

Additional Questions Specific to the Prospective Case Study

1. No comment.

2. No comment.

3. A source of uncertainty in applying TEFs across species of the same vertebrate class is not
knowing to what extent TEFs vary between species. Until the scientific literature clearly
demonstrates (within the same vertebrate class for those PCDD,  PCDF, and PCS congeners that
are generally considered to be the major contributors to TEQs) that the TEFs determined for one
species are consistently more than one order of magnitude different from TEFs determined for the
same endpoint in a different species - it is recommended that an uncertainty factor not be applied
to the TEFs.

                                                                  Richard E. Peterson
Additional Questions Relative to the Retrospective Case Study

1.  The TCDD equivalents sediment clean up goal would be determined by which wildlife
receptor of concern would have its recruitment adversely affected at the lowest concentration of
TCDD equivalents in eggs (lake trout or Caspian tern) or liver (otter).  This sediment clean up
goal, because it is the most restrictive, would also be protective of recruitment in the other two
       If the vertebrate group with the most certainty is not the group with the most restrictive
sediment clean up goal it might still play a useful role in directing the sediment clean up.  That is,
the sediment clean up goal for the "less sensitive but more certain group" could represent an
"upper bound" for clean up whereas the sediment clean up goal for the "more sensitive but less
certain group" would represent the "lower bound" for clean up. By bracketing the sediment clean
up goal in this way would set limits on what is acceptable.

2.  No comment.


                                                      Mark R. Servos, Ph.D.
                                             Aquatic Ecosystem Protection Branch
                                                 National Water Research Institute
                                          Canada Center for Inland Waters (CCIW)
                                                           867 Lakeshore Road
                                                                 P.O. Box 5050
                                             Burlington, Ontario, Canada L7R 4A6
                                                             Fax: 905-336-4420
                                                    E-mail: mark.servos@cciw.ca
Dr. Mark Servos received a B.S. in fisheries biology, an M.S. in aquatic science from
the University of Guelph,  and a Ph.D. in philosophy from the University of Manitoba.
Dr. Servos is the project chief of the aquatic ecosystems protection branch of the
National Water Research Institute  of Environment Canada.  He  leads a team of
scientists in investigations on the fate, exposure, bioavailability, and effects of priority
contaminants in effluents and aquatic environments.  Previously, Dr. Servos was a
research scientist for the Great Lakes Laboratory for Fisheries and Aquatic Sciences
where  his focus of research was  on  the environmental chemistry and biological
responses of fish to organic contaminants in effluents; the isolation and identification
of toxic components of effluents, pesticides, and environmental, samples. Dr. Servos
is currently on the board of directors of the Society of Environmental Toxicology and
Chemistry. He  has published many relevant articles such as Confirmation ofchloro-
nitro-trifluoromethyl-substituted dibenzo-p-dioxins in  lampricide formations  of 3-
trifluoromethyl-4-nitrophenol (TFM): Assessment to induce P450A1 activity; Evaluation
of temporal and age related trends of chemically biologically  generated 2,3,7,8-
tetrachlorodibenzo-p-dioxin equivalents in Lake Ontario lake trout; and Evidence for a
reduction of 2,3,7,8-TCDD toxic equivalent concentration in white sucker (Catastomus
commersoni)  exposed to bleached kraft mill effluent, following process and treatment

                                                                                      Mark Servos

TEF Workshop: Responses to questions

Stress response profile

1.  There is very little gained by using the exact values for TEFs. The values are derived from information
for which there is a lot of variability with the experimental data, approaches, species, etc.  Using the actual
values would in many ways be misleading because it gives more credibility to the number than is justified.

2.  No. There is certainly more data available for some chemicals. There should be more credibility give to
studies that use whole organism responses as an endpoint or that have been validated in field studies. Only
a few congeners generally contribute to the REQ and these are the ones for which the uncertainty is most
critical. Emphasis should be on estimating the uncertainty of these.

3.  The farther you move from a whole organism response the less faith we can have on the predictive
ability of the measure. The whole organism responses integrate the many complex responses in the fish.
Many factors can alter, inhibit or modify biochemical responses at the cellular level dramatically alter the
interpretation of the relative toxicity. Early life stage mortality is an endpoint which we can apply with
some certainty. However, there are numerous mechanism by which these chemicals can interact with
organisms and cause adverse effects. The early life stage mortality is a very well studied and sensitive
endpoint for fish. In birds most of the studies are at lower levels of organization. Of particular interest in
birds would be the validation of the TCDF TEF of < 1.

Stress-response profile relative to the application of the TEQ Approach

1.  A single chemical approach would seriously underestimate the potential Ah-receptor mediated toxicity.
The TCDD alone would not be considered a major problem at any of the trophic levels in the northern lake
scemario. Looking at t-PCBs alone would also be very misleading as the t-PCBs are  less than would be
expected to cause any responses in the trout eggs or otter liver. In contrast the t-PCBs are in the range that
would raise concern in the terns; 5.657 compared to 5 (1-20) ug/g no effect threshold. If we assume
additiviry, the conclusions are very different. The total-TEQs in trout eggs is below the threshold for
effects (30) but approaching the range of reported values. The total-TEQs in the terns is 4 times higher than
the threshold values. This driven by the high  levels of PCB 126,77 and 81 with a relatively small amount
of 2,3,4,7,8-PCDF. Mink liver also have high values (although less than terns) but the total-TEQs are
driven by PCB  126 and 2,3,4,7,8-PCDF. By assuming additivity we have ignored potential antagonistic
and synergistic effects which could alter the expression of toxicity.

                                                                                      Mark Servos
 2. The threshold values may differ from the ECSOs leading to a misinterpretation of the relative toxicity of
 congeners. The assumption that the dose-response slope are parallel is often valid and is testable although it
 may differ according to species and response measured. Caution should be used when making conclusion
 based on this type of data.

 3.  In the spill case study, it would seem reasonable to apply the TEFs to all of the species within each class
 of biota with some caution. There will be considerable variability within each class but if caution is used
 and the limitations recognized this approach will be very useful. The closer the phylogenetic relationship
 the higher the level of confidence in extrapolating the results. The lake trout TEF can be applied with
 considerable confidence to the lake trout and salmon but less so to the carp and sturgeon. Many fish species
 such as fathead minnows show a much reduced response (EROD) to exposure to various chemicals which
 may affect the application of the TEF developed from trout, etc.

 III. Exposure profile

 1.  The need to model numerous chemical presents a  challenge. The weakness of the physical/chemical and
 biaccumulation data for specific congeners introduces considerable uncertainty. This leads to many
 assumptions or simplified approaches being employed. The particularly important weakness is knowledge
 as to the extent of bioaccumulation and the changes in the relative composition of congeners at different
 trophic levels resulting from differential metabolism  and/or biomagnification. This can lead to very
 different relative importance of each congener in different organisms. The chemical focused for
 remediation may differ depending on the trophic level that is at risk. For example, in the spill senario, the
 PCDFs are the most important (more than half) the lake trout while the terns are driven by the  PCBs
 especially PCB 126. PCB 126 is the dominant congener of concern for otter but PCS 77 and 81 are
 important for terns.

 2. For the persistent slowly metabolized or excreted compounds this would not be a significant problem.
 For other compounds it could be if the actual dose is  not considered. There is  some concerns about
 continuous exposure to congeners that do not bioaccumulate (this could be happening in a pulp mill
 discharge). A tri-substituted dioxin found in the pesticide TFM caused induction of MFO enzymes in fish
 even though it was relatively water soluble and easily degraded. A constant exposure to low levels may
 result in responses which would not be seen in experiments where a single dose is administered and the
 chemical is quickly metabolized.

 3. Congener specific analysis is difficult and expensive. It require additional steps in the clean-up and high
 resolution GC-MS detection.

                                                                                    Mark Servos

IV. Risk characterization

2. The biologically based TEQs could be employed as an alternative to address some specific questions.
Biologically derived TEQs could be used as a surrogate for more expensive chemical analysis to monitor
the success of remediation or to detail the distribution of the contamination. However, the concentrations
would have to be validated and it would have to be demonstrated that the chemicals of concern were
causing the biological response in the environmental samples being monitored. The biologically derived
TEQs would differ based on the type of cell line used (fish vs mammals) so the appropriated procedure
would be important. Biologically derived TEQs have been used to demonstrate that chemicals other than
PCDD/Fs at pulp and paper mills were present and contributing to the MFO induction response. In a case
such as this the biologically derived TEQ would respond to the other chemicals and fail to demonstrate a
reduction in the chemicals of concern (e.g. PCB 126, PCDF).

3.  site specific studies, etc.

•   Relative contribution of items in the diet of the birds and otter. Are the PCBs associated with
    bioaccumulation through the aquatic system in the lake?
•   The role of other sources to the diets. To determine the relative role of the lake and determine if
    remediation will reduce levels. The calculation of the concentration of chemicals in the birds and mink
    did not consider that only part of the diet comes from the lake.
•   The seasonal contribution of the  lake to loadings and reproductive success hi tems.
•   Reproductive success of the terns and hatchability of lake trout. To determine if there is a predicted

Risk Characterization

The clean-up goals would not be the same for each vertebrate group. The fish are just approaching the
threshold values so there would be no apparent need for remediation. The birds have the highest TEQs but
also a higher thresholds than the otters. The birds and the mammals have about the same BMP  for PC 126
but the mamamals have a BMP of 54 compared to only 1.6 for 2,3,4,7,8-PCDFin birds. The otter
metabolized some of the PCBs, most notably the PCB 77 and 81 which changes the relative composition of
the congeners at this trophic level. PCDF contributed the most to the PCDD/PCDF TEQ value  in terns and
otters. However, when considered alone or with the PCDD/PCDF totals they are predicted to not be high
enough to cause the effects. There is a huge difference between  the TEFs of PCB 77 and  81 in  birds and
mammals but the value for PCB 126 which is the major contributor in both is the same. The focus should
be on the PCBs in the higher trophic levels but these goals would have to be translated into sediment

                                                                                    Mark Servos

concentrations goals. The sediment cleanup goals should be set for the group with the most restrictive
values. However, if the uncertaintity is high the values may be too low and unnecessary and expensive
cleanup goals may be set. If the terns are getting only a small dose from the lake when we have assumed it
all comes from the lake then remediation may not result in  the desired goal. On the other hand the poor
prediction of the BMFs or other factors may set a value to low which is not adequate to protect the most
sensitive species. If possible the various scenarios should be presented with some level of confidence (or
lack of) for the risk manager to use as needed. To use the  t-PCBs to guide remediation we would need to
ensure that the key toxic congeners are changing in proportion to the total-PCBs and the relative
bioavailability of the congeners is not changing with the remediation.


                                                Martin van den Berg, Ph.D.
                                   Associate Professor of Environmental Toxicology
                                                  Research Institute of Toxicology
                                                           University of Utrecht
                                                               P.O. Box 80176
                                               Utrecht,  3508 TD, The Netherlands
                                            E-mail: m.vandenberg@ritox.dgk.ruu.nl
Dr. Martin  van den Berg  received his MS. and his Ph.D. in environmental and
toxicological chemistry from the University of Amsterdam.  Currently, Dr. van den Berg
is  an associate professor of environmental  toxicology at the Research  Institute of
Toxicology (RITOX) of the University of Utrecht in the Netherlands. He is the leader of
the research group studying toxicokinetics and dynamics of persistent environmental
contaminants.  Dr. van den Berg's areas of research interest include: toxicokinetics and
reproductive effects of halogenated polyaromatics, interactions of xenobiotics on steroid
hormone metabolism, and interactions between dioxin-like compounds and PAHs with
respect to genotoxicity.  He conducts these studies on fislvbirds, and mammals.  Dr.
van den Berg  has published over 120 scientific articles  and papers on this area of
research. Dr.  van den Berg was the chairman of the World Health Organization's
"Meeting on the Derivation of Toxic Equivalency Factors for PCBs, PCDDs, PCDFs,
and Other Dioxin-like Compounds for Humans and Wildlife" held on June 15-18,1997.

                                                                Martin van den Berg

 Comments on the questions related to the Workshop on the Application of TEFs to Fish
 and Wildlife.

 With respect to additional information provided, I do not think that it will provide the
 workshop with more information regarding the use of TEF values for wildlife, as it mostly
 concerns guidelines for TCDD only. During the WHO Stockholm meeting all available
 material has been evaluated for "Eco-TEFs". Therefore it does not enhance the means
 of evaluating the uncertainties more.

 The uncertainties in "Eco-TEFs" are much larger than those obtained from mammalian
 studies due to the limited information available. Nevertheless, by using a tiered approach
 (see WHO document) and the rounded off procedure the most protective way was
 chosen which was possible.

 At the TEF Stockholm meeting a tiered approach was followed for "Eco-TEFs" in which
 priority was given to more classical toxic parameters, e.g. ELS mortality, above
 biochemical effects or QSARs.
 If it is assumed that no additivity or no interactions exist, each compound involved
 should be evaluated seperately and the basic information for this process is than lacking,
 because it was simply not available. Actually, because we acknowledge the Ah-receptor
 mechanism and derived TEF concept we are worrying about these compounds. In
 addition, there are sufficient in vivo and in vitro studies which support this TEQ/TEF
 approach. In general non-additive interactions which have been reported are general
 within one of a magnitude or even much lower. For ecotoxicology the largest
 uncertainties do not seem to be these non additive interactions but large differences in
 species sensitivity.

                                                              Martin van den Berg

From a practical point of view there is not much choice as at best usually the EC50 or
LC50 values were reported. As the efficacy of a dose respons curve varies, especially
for PCBs, another value maybe better. An EC10 or or even lower might be better.
However I feel that the amount of information now available in literature does not permit
such an approach. With respect to differences in slope of the dose respons curve I am
not convinced that the statements which have been made by the critics that this
phenomenon makes the TEF concept impossible to work with are that valid. Opponents
using this "difference in slope" argument have to my knowledge this argument not solidly
supported by statistical analysis.

I think that the differences observed in species specific TEFs are less of a problem than
the value which is actually used as a LOAEL or NOAEL for the species of concern. In
the environment we have a huge amount of species variation and we know from lab
studies with different taxa that sensitivities towards these compounds, usually TCDD,
can vary more than three orders of a magnitude. Therefore the right choice of
LOAEL/NOAEL seems to much more essential to the process.

One of the challenges within the present approach of the TEF concept comes from the
fact that the efficacy (Ymax)  of the response varies a lot, especially with  PCBs. This
difference in efficacy strongly influences the TEF values we work with. I would like to
know how this problem should be approached in future and what the fundamental
reasons are behind this phenomenon. At my lab we have developed some ideas, but
sofar have not come up with a good solution or suggestion (Maybe this topic belongs to
another question).
I think food chain models will work pretty nicely as long as the compounds are highly
resistant against biotransformation. As soon as compounds are more effectively
metabolized the modeling becomes more difficult due to species differences in metabolic
capacities. Luckily for most of the PCDDs, PCDFs and PCBs which accumulate in the
food chain this seems not to play a dramatic role. I think other fate and transport models
will also be a good approach for these compounds as long as you stay with the more
hydrophobic compounds. In  addition, I think that physico-chemical data can be much


                                                              Martin van den Berg

easier obtained for these compounds than biological data, while further more physico-
chemical data could be easier estimated using e.g. log Kow.

The TEF concept could certainly be strongly improved if values are derived in the future
from e.g. tissue levels instead of administered dose.  Using PB-PK modeling in
combination with the right toxicodynamic models would improve the risk assessment for
these compounds significantly and even bridge the differences between species. The
present TEF concept has always been  presented as  an "interim" method for risk
assessent during the last 10-15 years,  but in fact nobody ever came up with a better
method. In addition, governmental agencies have not put much effort in improving the
TEF-concept either. So for the time being we are just stuck with  this interim method if we
want to do (eco)risk assessment for these compounds.

I do not think that the present TEF concept requires a more rigid design of analysis as
most labs which are involved  already analyze the non and mono ortho PCBS in addition
to the 2378-PCDDs and PCDFs already measured. Measuring total PCBs is rather risky
for sediments etc. as you might miss geographical and temporal changes in the most
toxic PCBs in the matrix easily. As these are the congeners you  are interested in, this
information should not be ignored. However, it should be noted that TEQ values from
e.g. sediments have more a comparative meaning than an actual toxicological one. As is
illustrated nicely with both risk assessment exercises much more information is
necessary before e.g. a risk assessment can be done for a top predator. In other words
when ecological risk assessment is done on these compounds models should work as
long as possible with the congener specific approach. The combination between TEQs
and toxicity should be done in the final  step of final food. In general the use of TEFs and
TEQs should not exceed that of a single trophic level.

                                                               Martin van den Berg

IV. 1.
I think that the problems with the TEFs and associated TEQs are in general greater than
those for a number of other groups of environmental contaminants eg. OP-esters,
a) It involves an exceptionally high number of compounds with a specific mode of action.
b) These compounds show a large variability in physico-chemical properties.

However, if one would develop an appropriate TEF model for the polyaromatic
hydrocarbons and genotoxicity, I am sure similar problems as with the dioxin TEF model
would be encountered.
See also earlier comments I made regarding the uncertainties of the present TEF

I think that biological based assays, e.g. Caluc Ah-receptor, could serve very well as a
prescreening method for selecting abiotic environmental samples for further chemical
analysis. In fact they could also serve as a way of measuring TEQ tissue levels from
target species. However, these bioassays could never be used in biological samples as
long as the species specific sensitivity is unknown. Alternatively, a  large general safety
factor for eco risk assessment in combination with these bioassays might do it also. I do
believe that these bioassays can save us a lot of money on expensive chemical analysis
as long there limitations are acknowledged.

I think that the present amount of information is adequate for the two case  studies
presented. From  my expertise (TEFs, toxicity and pharmacokinetics) the recent WHO
evaluation tried to incorporate as much as possible all available scientific information.
More information was simply not available.  It might be desirable to have the WHO TEF
database available at the meeting for consultation, If necessary I can bring it with me.


                                                    Bert van Hattum, Ph.D.
                       Program Manager, Environmental Chemistry and Ecotoxicology
                                               Institute for Environmental Studies
                                              Free University, De Boelelaan 1115
                                                          1081 HV Amsterdam
                                                              The Netherlands
                                                     Fax: 011-31-020-444-9553
                                                    E-mail: bvanhattum@sara.nl
Dr. van Hattum received a Ph.D. in biology from Vrije University in Amsterdam and a
master's degree in chemistry at the State University of Utrecht. At Utrecht his major
areas of study were analytical chemistry and chemical oceanography. Most recently
he has researched food chain transfer and effects of planar PCBs in top predators,
analysis and bioaccumulation  of  organic compounds, risk assessment of water
discharges from  offshore installations,  design  of biomonitoring programs,  and
sustainability and environmental quality in river basins. Dr. van Hattum has participated
in several  advisory committees for the Dutch government on secondary poisoning in
mammals and birds and on the development of environmental quality criteria for PCBs.
He is currently a member of the board of the Environmental Toxicology Section of the
Dutch Society of Toxicology, a member of IAWQ, SETAC, and.the editorial board of
Environmental Pollution journal. He has published more than 40 articles and reports,
4 books or chapters, and he has made presentations at a number of conferences in
Europe and Canada.

                                                          Bert van Hattum
                           Premeeting Comments
With much pleasure I have read the extensive set of documentation. I was impressed
by the profound and critical approach laid down in the different documents. Below I
have tried to answer some of the charge questions in my field of expertise. Some of
the citations include recent technical reports on Dutch and Danish otter studies, which
can be made available on request.

L      Stress-response profile relative to the derivations of specific TEF-values.
Ql,2,3         The material provided for the workshop and the references cited in the
case studies, provide further experimental corroboration of the validity of the
approach laid down hi the WHO-document for the evaluation of the hazards of AhR
agonists, such as PCBs, diobenzodioxins and dibenzofurans. Although many questions
remain unanswered (WHO-1997), it helps to identify critical compounds, pathways,
species at risk, and to focus emission reduction programmes. Especially the material
from recent experimental or review studies on mink (Tillitt et al.  , 1996; Leonards et
al., 1995) provides substantial evidence for the extreme sensitivity of this species, the
cause-effect linkage between contaminants in the diet and reproductional effects, and
the soundness of the TEF-approach as a framework to account for the joint toxicity of
mixtures of contaminants. The uncertainty of the proposed TEQ-based no-effect
concentrations (NOEC) for mink probably is much lower than  for other chemicals, for
which NOECs usually are being extrapolated using safety-factors from experimental
studies with 'surrogate' laboratory species (Luttik et al. ,1993).

n.    Stress-repons profile relative to the application of the TEQ-approach
Ql.    Rejecting the additive dose-interaction model of the-TEQ-approach, would
imply separate risk-assessments for all potentially active individual congeners. In that
case the focus should be directed towards the compounds that are most likely to
induce effects. Depending on the target organism, a significant proportion of the toxic
potency may be left out of the evaluation. Based on the TEQ-values calculated for the
Oneofakind Lake case study, it can be hypothesised that individual congeners may

                                                           Bert van Hattum

contribute represent at most iip to 30-40% (23478-PCDF) of the total dioxin-
equivalent conentration for fish, and 60-70% (PCB-126) for birds and mammals.
Reasoning further along this line, an evaluation based solely on total PCBs would
probably underestimate fish early life stage mortality, and an evaluation based solely
on 2378-TCDD would underestimate potential reproductional effects in mammals or
Q2.    I am not familiar with the discussions held during the preparation of the
WHO-TEF document. Based on analogy of risk assessments conducted for other
compounds, chronic NOEC (no observed effect concentration) or NOAEL-based
values (no observed adverse effect level) if available, should be preferred.
Relationships between NOECs and other endpoints values have been studied
systematically for several compounds and test-species (Slooff et al.  1986), and have
resulted in specific extrapolation factors, which are used in the Dutch risk-assessment
protocols (Luttik et al., 1993). I  have no knowledge if similar surveys have been
made for AhR-agonists.
Q3.    If available, data from eco-epidemiological studies for potentially affected
species should be used to evaluate the feasibility of the lethal body-burden concept
and within-class extrapolations of toxicity endpoints for AhR-agonists. In Leonards
(1997) a comparison was made of no-effect and critical levels for otter, mink and
seal, expressed as concentrations in the liver of the target-species. Hepatic TEQ-based
NOECs for mink (0.4 -9 ng/g lipid wt.) and otter (1 - 2 ng/g lipid wt.) were
comparable. A lower NOEC was found for seals (0.1 ng/kg lipid wt.). Due to the
large differences in toxicokinetics and biotransformation between the species,  fish
diet-based NOECs for TEQs (ng/kg wet wt.) exhibited a much different ranking,
ranging from 0.7 ng/kg wet wt for otters, and 1-50 ng/kg  for mink and 8 ng/kg wet wt
for seals. This demonstrates that within-class extrapolations of toxicity endpoints
should be dealt with only with great care.

HI.     Exposure Profile
Ql.     The proposed risk-modelling framework requires the  input of high-quality
data for calibration of currently used parameter estimates (BAFs, BSAFs, BMFs, rate
constants) in chemical fate and bioaccumulation modelling. Special  attention should
be given to habitat- or target species-specificity of parameter estimates, in order to

                                                            Bert van Hattum

judge the validity of the application of generic parameter-estimates. In the studies
conducted in food chains of European river otters (Lutra lutra) in Danish and Dutch
habitats (Smit et al. , 1996; Leonards et al., 1997), we observed extremely high otter-
fish BMP values (indicated in Table 1) especially for some non-ortho substituted
PCBs (126 and 169), which are much higher than the mink-diet based BMFs reported
by Tilliitt et al., (1996) and applied in the case-studies. As concentrations in otters
and fish are influenced by factors such as age, sex, reproductional activity, and other
species-specific factors, BMP estimates may be highly sensitive to sampling design
and experimental methodological choices made in the various studies.
Especially with respect to the role of non- and mono-ortho substituted PCBs, the
availability of high-quality analytical exposure data is a limiting factor. The analytical
procedure (pre-separation followed by HRGC-MS) is costly, and a rigorous analytical
quality control is required in order to produce accurate and precise exposure data.
Most of the currently involved laboratories in OECD countries have participated in
round robin exercises or proficiency testing-programmes. There is a need for
development of low-cost analytical techniques for AhR agonists, which e.g. can be
used in combination with AhR-responsive bioassays.
Most of the old exposure assessment data for PCBs are expressed as equivalent
technical-mixture concentrations, as concentrations of selected dominant di-ortho
substituted congeners,  e.g. PCB-153, or as total concentrations of individual
congeners. In most European monitoring programmes  and regulatory practices
attention is focused mainly on di-ortho PCBs. Therefore, there also is a need to
develop  and evaluate the feasibility of generic or habitat-specific extrapolation
algorithms to derive TEQ-exposure profiles from e.g.  PCB-153 concentration data.
Q2.   Exposure route differences may have a profound influence on the actual
dosage at receptor sites, due to variations in e.g. bioavailability, feeding-preferences,
toxicokinetics and biotransformation of contaminants in the target-species and in
species from lower trophic levels. The differences in susceptibility to dietary PCBs
between mink and otter, as discussed previous section, seem to be related to variation
in toxicokinetics and lipid-metabolism between both species.

                                                           Bert van Hattvm

TV.    Risk Characterization
Ql.    Many methodological, habitat-related and biological factors contribute to the
uncertainty of parameter estimates used in risk assessment models. Complementing
the angle taken in the case studies, with evaluations based on trials with probabilistic
models would provide insight in the effect of the uncertainties on the extent to which
target species are protected.
Q2.    The added value of some of the recently developed bioassays and biomarkers,
is that they provide insight hi the total quantities of AhR-responsive compounds. In
studies conducted by Murk et al. (1996) and (1997) good correlations were found
between CALUX-based TEQs and values derived with TEF-values from measured
concentrations. Care should be taken to account for confounding response of other
potentially active compounds, such as e.g. PAHs in sediments. In the study of Smit et
al. (1996) a combined approach was applied, in which the low-cost and sensitive
CALUX-assay was used for screening purposes and selection of samples for further
extensive chemical analysis.
Q3.    With respect to gathering  of site-specific data my recommendation would be:
•   exposure concentrations in sediments, fish and tissues of predatory species species
    to examine if generic BSAF and BMP values can be used, or that site-specific
    values should be applied. For  the prognostic case-study the predicted water
    column partitioning (dissolved, DOC-bound and POC-bound) should be
    corroborated with experimental data.
•   ecological assessment of status of targeted species, assessment of influence of
    other natural or anthropogenic stress-factors
•   confirmation of predicted risks with relevant laboratory-bioassays
•   uncertainty analysis with probabilistic modelling

V     Prospective case-study
Ql.    Due to variable hydrodynamic conditions, large variations may be expected  hi
the transport, distribution and bioavailability of contaminant in the water column. The
applicability of generic BAF  values needs to be investigated. Additional studies
could contribute to the accuracy and precision of the proposed water quality

                                                          Bert van Hattum

Q2.    Based on the high biomagnification of some non-ortho substituted PCBs
(PCB-126, PCB 169) in the food chain of the otter, an approach which ignores
congener-specific biomagnification,  may result in an underestimation of risks of these
congeners to sensitive predators.

VI     Retrospective case-study
Q2.    In the study of Smit et al.  (1996) TEQ-based NOECs and critical levels (for
Vitamin A reduction) in otter-liver, were extrapolated with congener-specific BMFs
and BSAFs to equivalent critical levels and quality objectives in fish-diet and
sediments. Significant double-logarithmic correlations were observed - for sediments
and biota - between concentrations of I TPGBs (summation of 7 selected congeners,
which usually account for 50% of total PCBs : 28, 52, 101,118,  138, 153, 180) or
indicator congeners (PCB-153) and TEQ levels. Although the relative contribution of
individual congeners to the  total TEQ-based concentration appeared to be species-
specific, extrapolation factors could be derived to express the proposed critical levels
on the basis of I 7PCBs and of PCB-153. As a rule of thumb,  a one order of
magnitude range of uncertainty may be introduced due to this extrapolation.
Nonetheless, as most of the European fish and sediment-monitoring data are based on
these standard congeners, this provided a framework to evaluate the quality of
potential otter habitats  in the Netherlands. Similar relationships also have been
observed hi recent (unpublished) studies planar PCBs in sediments and cormorant
food chains in the Rhine-Meuse estuary.
As correlations between total PCBs and TEQ-based concentrations may be site-and
species-specific, my recommendation would be to validate such  extrapolations with
measurement data. Some of the low-cost biomarker techniques also could have
potential for screening purposes in this context.
Ahlborg U.G. Becking, G.C. Bimbaum, L.S. Brouwer, A. Berks, H.J.G.M. Feeley, M.
   Color, G. Hanberg, A., Larsen, J.C. Liem, A.K.D. Safe, S.H. Schlatter, C. Waern, F.
   Younes, M. Yrjanheikki, E. 1994. Toxic Equivalency Factors for Dioxin-Like PCBs •
   Report on a WHO-ECEH and IPCS Consultation, December 1993. Chemosphere.

                                                             Bert van Hattum

Hoffinan, D.J., C.P.Rice, and T.J.Kubiak (1996): PCBs and Dioxins in Birds. In:
   Environmental Contaminants in Wildlife - Interpreting Tissue Concentrations, edited
   by W.N. Beyer, et al, pp. 165-20CRC Press, Boca Raton, FL, USA.
Leonards P.E.G., T.H. de Vries, W. Minnaard, S. Stuijfzand, P. de Voogt, W.P. Cofino,
   N.M. van Straalen and B. van Hattum (1995). Assessment of experimental data on
   PCB-induced reproduction inhibition in mink, based on an isomer- and congener-
   specific approach using 2,3,7,8-tetrachlorodibenzo-p-dioxin toxic equivalency.
   Environ. Toxicol. Chem. 14, 639-652.
Leonards P.E.G., Y. Zierikzee, U.A.Th. Brinkman, W.P.C. Cofino, N.M. Van Straalen and
   B. Van Hattum (1997). The selective dietary accumulation of planar polychlorinated
   biphenyls in the otter (Lutra lutra). Environ. ToxicoLChem. 16):1807-1815.
Leonards, P.E.G. (1997). PCBs in mustelids - analysis, food chain transfer and critical
   levels. Thesis. Vrije Universiteit, Amsterdam.
Luttik, R., Romijn, C.A.F.M., and Canton, J.H. (1993). Presentation of a general algorithm
   to include secondary poisoning in effect assessment Sci. Total Environ. (Supplement,
   Part-2), 1491-500.
Murk AJ., Legler J., Denison MS., Giesy JP., Vandeguchte C., Brouwer A. (1996).
   Chemical-activated luciferase gene expression (CALUX) - a novel in vitro bioassay for
   ah receptor active compounds in sediments and pore water. Fund. Appl. Toxicol.
Murk AJ., Leonards PEG., Bulder AS., Jonas AS., Rozemeijer MJC., Denison MS.,
   Koeman JH., Brouwer A. (1997). The CALUX (chemical-activated luciferase
   expression) assay adapted and validated for measuring tcdd equivalents in blood
   plasma. Environ. Toxicol. Chem.  16(8): 1583-1589.
Slooff, W., J.A.M. Oers, and D. de Zwart (1986). Margins of uncertainty in
   ecotoxicological hazard assessment Environ. Toxicol. Chem. 5, 841-852.
Smit, M., P.E.G. Leonards, AJ. Murk, A.WJ.J. de Jongh, B. van Hattum, (1996).
   Development of Otter-based Quality Objectives for PCBs (DOQOP). ISBN-90-5383-528-
   8. IVM-R96/11, Institute for Environmental Studies, Vrije Universiteit Amsterdam, 170 p
Tillitt, D.E., Gale, R.W., Meadows, J.C., Zajicek, J.L., Petennan, P.H., Heaton, S.N.,
   Jones, P.D., Bursian, S.J., Kubiak, T.J., Giesy, J.P., and Aulerich, R.J. (1996): Dietary
   exposure of mink to carp from saginaw bay. Environmental Science & Technology,
WHO (1997). Draft report on the meeting on the derivation of toxic equivalency factors
   (TEFs) for PCBsm PCDDs, PCDFs and other dioxin-like compounds for humans and
   wildlife. June 15-18,1977, Stockholm, Sweden.

                                                               Bert van Hattum
Table 1.
Average fish diet-based biomagnification factors (lipid weight basis) of PCBs
for otters in Danish (Lymfjord) and Dutch (Lakes Oude Venen) habitats.

E TEQs***
Oude Venen

0.47 (0.002 -

0.54 (02-2.3)
7.9 (0.7 - 50)


37 (3 - 505)
84 (2 - 2086)

13 (2 - 83)

63 (5 - 442)

2.5 (3 - 7.9
130 (4.2 - 900)
. . 108(3-1700)
36 (2.9 - 209)
95 (3.5 - 640)
 * Geometic mean BMFs for 5 otters from Leonards et al.  (1997); ** From Smit et al. (1996),
 geometric mean values (di-ortho PCBs n=20; non/mono-ortho PCBs n=9) and minimu to
 maximum ranges of BMFs between brackets; *** calculated with TEFs from Ahlborg et al.
 (1994). BMFs are expressed on a lipid normalized basid and calculated for an average diet

          Appendix D



Facilitator: William Adams

Group Members: Joseph DePinto, Lynn McCarty, Christopher Metcalfe, Patrick
O'Keefe, Mark Servos, Phil Cook, Cynthia Nolt, and Lisa Williams (notetaker)

Discussion started with the questions distributed.

Question 1:   How well do we know the uncertainties associated with accuracy and
precision of analytical chemistry data, including measurement of BAFs, BSAFs, and

BAFs may vary from one aquatic system to another.  Modeling of these aquatic
systems may allow estimates of BAFs to be made. Modeling BAFs requires parameters
like K^s and Henry's Law constants for individual congeners.  I^s are known
reasonably well, but Henry's Law constants are usually calculated themselves and may
be uncertain to within a couple of orders of magnitude.

BAFs can be determined empirically by measuring concentrations of congeners in
biological tissues and in water. Concentrations of some compounds, including TCDD
itself, may be near the detection limit in water, especially in the dissolved phase.
Water concentrations may be calculated from other partitioning coefficients.

Precision and accuracy of the compound-specific measurements needed to determine
BAFs and BSAFs vary among matrices.  The sources of uncertainty in these
measurements include analytical variability, the extent to which sampling protocols
represent the real heterogeneity in the system, and magnitude of the real
heterogeneity. Analytical variability increases as concentrations approach the limits of
quantification (LOQJ and detection (LOD).  In sediments and tissues, concentrations
of individual PCBs, PCDDs, and PCDFs can currently be determined to within ± 30%
for most samples. As concentrations approach the LOD, concentrations can be
determined to within a factor of 5 to 10.  Concentrations in ambient water samples are
near LOQ and LODs in most samples, so determinations are generally accurate to
within a factor of 10. Sampling protocols need to be designed using power analyses.
The real heterogeneity of concentrations of these compounds may be huge  for
sediments within a given aquatic system because of the heterogeneity of sediment
types. Concentrations vary spatially and with organic carbon type and amount,
particle  size distribution, and other sediment characteristics. Concentrations of these
compounds in individual fish within a population may vary by an order of  magnitude.
Heterogeneity in water samples within a system is not well  studied, but varies with
solids dynamics in the system.

Analytical techniques are available for all of these compounds.  Many commercial
laboratories are currently analyzing 2,3,7,8-substituted PCDDs and PCDFs. Fewer are
regularly analyzing non-ortho-substituted PCBs, but more would likely add these to
their available analysis if regulations began requiring quantification of these

compounds. No capital expenditures would be required as the methods currently
being used for analysis of PCDDs/PCDFs are very similar to those used for the non-
ortho-substituted PCBs. The most likely method to be used for these analyses is a
carbon column separation step followed by gas chromatography/mass spectrometry
with isotope dilution internal standards.

The uncertainty in sampling and analysis of the individual congeners is similar to that
introduced by sampling and analyzing total PCBs or 2,3,7,8-TCDD alone. The
samples required would be the same for any of these analyses.  Measurements of total
PCB concentrations may have greater uncertainty than those for individual congeners.
Analysis of total PCBs is generally done by analyzing individual congeners and then
summing their concentrations. Laboratories may use a different number of congeners
to quantify total PCBs and  may use Aroclor mixtures or individual congeners as
standards. Analysis of total PCBs is less expensive than for congener-specific analysis
which includes the non-ortho-substituted PCB congeners.  The difference  in the cost of
analyzing 2,3,7,8-TCDD and analyzing all of the 2,3,7,8-substituted PCDDs and
PCDFs is negligible.

Question 2:  Are chemical fate and transport properties (hydrophobicity, volatility,
photolysis, biodegradability, etc.) well characterized for all chemicals with  TEFs? If
not, what uncertainties are  introduced in exposure predictions?

Hydrophobicity (K^,) and volatility are better known and generally more important in
determining fate and exposure than photolysis and biodegradability. I^s are known
reasonably well.  Using K^s to predict K^.5 introduces an uncertainty of about an
order of magnitude (based on a 95% confidence interval) because of inherent
differences in the nature of organic matter. Henry's Law constants for some congeners
may only be accurate to within a couple of orders of magnitude, so in systems in which
volatilization is a significant fate pathway, this uncertainty could have a significant
impact on predicting water  concentrations.  Biodegradation is system-, congener-, and
concentration-specific, so generalizability of this process is very poor. The time scale
for this  process is very long, so this is a less significant process in determining overall
fate than partitioning, sorption, volatilization, and many other processes. Photolysis
is also generally of minor importance to an overall mass balance for these compounds,
but sensitization could result in this process being important for some compounds in
some systems. In mass balance exercises in major rivers and bays, approximately 80%
of the accuracy of the model was determined by the accuracy of the modeling for the
solids dynamics of the system.

Question 3: What degree of uncertainty is associated with biotransformation/
metabolism in the food chain?

Biotransformation, metabolism, and differential absorption patterns alter  congener
patterns more significantly  in birds and mammals than they do in fish and other biota.
Congener patterns among fish species and fish tissues are relatively homogeneous.
This is not true for birds and mammals that eat those fish. Relative to the pattern in
fish, concentrations of PCB 77 decrease and those of PCB 126 increase  as a percentage
of total PCBs  (see Figure D-l).

Figure D-l.
                 £ 0.008
                 e 0.004

                                       Food Web Group
                                       Food Web Group
                            L. Ont. food web
L. Zandemeer food web

These shifts, especially for PCB 77, may be quite species-specific. Changes in patterns
among tissues within a bird or mammal are not well characterized.  Most
environmental data on birds is for eggs and most mammalian data is for livers.

Biomagnification factors (BMFs; ratio of concentration of a given compound in the
tissue of the predator to the concentration in the tissue of its prey)  are needed for the
species of concern, especially for fish-eating birds, for a given assessment.  BMFs for a
given species in one system could be applied in another system if the dietary
composition is known. The basic ecology (food web structure and dietary
composition) of a system is usually less certain than the BMFs themselves.

The TEF approach is critical in estimating risk to birds and mammals in particular
because of the change in congener patterns from source Aroclors and through the food
chain. For birds and mammals at the top of the food chain, these changes in pattern
appear to be even more species-specific or class-specific than they are source-specific
(in chronic exposures). In lakes with only atmospheric sources of PCBs and compared
to those with local sources with varying patterns of PCBs, the patterns of congeners in
the top of the food chain are similar although significant differences in patterns are
observed low in the food chain. In the Great Lakes, analysis of archived samples of
fish and bird eggs have shown little or no change in congener patterns within species
over a time period, while absolute concentrations dropped by an order of magnitude.

Question 4: For these classes of chemicals (PCDDs, PCDFs, and PCBs), what are the
greatest sources of inter-ecosystem variability in bioavailability and  bioaccumulation?
Are there any unique considerations for exposures in marine ecosystems?

The greatest sources of inter-system variability in bioavailability and bioaccumulation
are solids dynamics and food web structure.  This variability is most important in
predicting absolute concentrations of PCBs,  PCDDs, and PCDFs rather than in
predicting the relative proportions among them.  The solids dynamics is perhaps the
most difficult to determine, especially in a marine system. Systems can be compared
by knowing the distribution of contaminants between sediment organic carbon and the
freely dissolved phase (p^^; Cook and Burkhard, National Sediment Bioaccumulation
Conference, September 1996). This relationship can be measured for one congener
and generalized to the others.  Differences in benthos structure may have significant
influences on absolute concentrations at the top of the food chain for compounds with
KOW greater than 6. At the top of the food chain, the source of the contaminants to
critical tissues is important because birds and mammals are mobile.  For example, a
migratory bird may arrive on a breeding ground and begin feeding locally and then
transfer lipid from the bloodstream to the developing egg. In that case, the
contaminants in the eggs would reflect local  sources and diet. If the eggs are
developing during migration or if the bird arrives on the breeding ground and must
rely on stored lipid to produce eggs, then the contaminants in the eggs would reflect
other sources of contamination.

Question 5:  From the standpoint of TEF applications, what are alternatives to, or
improvements for, the waste load allocation process model described in Figure 5 of the
prospective scenario?

The waste load allocation process model needs to be validated, but it is expected that
this process is more likely to change the values of the parameters than the tool itself.
Overall, this model appears to be a good approach for dealing with this complex
mixture of compounds with additive toxicity and which are bioaccumulated and
exhibit chronic toxicity. The model would need to be more complex for an acutely
toxic substance or condition.  The assumption in this model of the existence of an
assimilative capacity for these persistent compounds can be questioned on
philosophical grounds. Loss processes from a given system are dominated by physical
movement within or from the specific system rather than by chemical destruction.

A critical component of this waste load allocation process model is the system level
mass balance model. A mass  balance model can be very complex, so a hierarchy of
mass balance models may need to be  developed. Different mass balance models could
be developed for different types of systems and with varying degrees of complexity and
number of input parameters required. There is nothing unique about mass balance
modeling for this application. If one  can model TCDD for a given system, then one
can model other PCBs, PCDFs, and PCDDs. The modeling will become more accurate
as more parameters are measured in more systems.

The overall uncertainty in the waste load allocation model is unknown, but is largely
related to knowledge of the system, not knowledge of congener-specific information.
For example, ratios of BAFs among congeners across various systems are fairly constant
(varying by less  than a factor of 2 or 3), whereas absolute BAFs among systems are less
certain. The uncertainty can be reduced with more measurements and can be explored
using Monte Carlo simulations. Regulators may be given guidance on an amount of
the waste load to be allocated to the uncertainty. Overall uncertainty in calculated
final MALs may currently be  a couple of orders of magnitude.

                           TEF EXPERTISE GROUP

Facilitator: Richard Peterson

Group Members: Bjorn Brunstrom, Steve Bursian, Jay Gooch, Mark Hahn, Bert van
Hattum, Sean Kennedy, Martin van den Berg, Steve Bradbury, Mike DeVito, Don
Tillitt, and Tim Kubiak (notetaker)

Dr. Richard Peterson opened the session by describing basic terminology for the
discussion. Two terms, Toxicity Equivalency Factors (TEFs) and Relative Potency
(REP), were defined to provide clarity. In the group discussion, Toxicity Equivalency
Factors (TEFs) were defined as consensus values derived from multiple REPs. A REP
pertains to the relative potency of a dioxin-like congener to TCDD in a single study.
Additionally, it was discussed that a receptor could be defined as both a species under
assessment and for pharmacological use of Ah receptor interactions. It was
recommended that "target species" be used for the species and "receptor" be reserved
for pharmacological use.

Dr. Peterson  reviewed and handed out copies of the nine questions to be considered by
the workgroup.

An issue was  raised pertaining to the specific  purpose of the World Health
Organization TEFs. Group members questioned how broadly they are to be applied,
and whether these values are for screening only.  These questions were subsequently
addressed in the  case studies and are addressed elsewhere.

The following discussions pertain to the numbered questions prepared for the expertise
group. It should be noted that six of the twelve individuals in this group also attended
the WHO TEF meeting in Stockholm, where the WHO TEFs were developed for non-
human mammals, birds, and fish.

Question 1. Are taxa-specific TEFs a reasonable approach given current scientific
understanding of relative differences between fish, birds, and non-human mammals?

Discussion centered on the use of TEFs for the various classes and their use for site-
specific analysis to be protective. There was considerable discussion about the
meaning of the term "protective."  It was  generally agreed that the term was misused in
the sense that there was no numerical uncertainty factor included to provide a any
margin of safety  due to uncertainty. Derivation was by a tiered  approach (WHO
1998). Site-specific use of the TEFs was thought to provide predictive value in the
interpretation of dioxin-like exposure, risk, and effects. While the WHO TEFs were
rounded up or down to the nearest half or whole order of magnitude, the differences
were not large. Fish and  bird TEFs were  considered predictive because the
documentation was available for specific REP to TEF conversion and these values
represented a toxicological endpoint for the most important congeners. There was a
dear indication that a statistical validity  assessment of the different TEFs was not
performed.                                     .

Question 2. What are the sensitivity analysis arguments for .using the WHO TEFs
that have been rounded to provide maximally harmonized TEFs across taxa versus the
unrounded TEFs that are more taxa-specific?

The discussion resulted in a finding that, despite rounding, the TEFs are not less but
more predictive. The relative contribution of PCDFs, PCDDs, and PCBs to total
TEQs does not change. Differences in species sensitivity is a larger source of

Question 3. How would you distribute uncertainty in the methodology among the
following categories:

   A.  High dose to low dose extrapolation
   Data indicate that using REPs from EQ0 versus EC10 does not result in much
difference.  In vitro CYP1A protein induction is less consistent and there are problems
associated with in vitro tests.  The greatest confidence in the data occurs at the EC50
level of relative  potency.  Therefore, greater uncertainty involves the extreme ends of
the dose-response curves, such as in the EQ.,0 range. Inflection points vary in dose-

   B.  Species differences in relative  potency

   There was general agreement that there is a large degree of uncertainty associated
with the comparative data across species when looking at REP data sets.  This includes
all tiers of the WHO database. The uncertainty is not associated with major variation
but with a lack of interspecies comparative data for the same endpoint.  For example,
the chicken response to embryo toxicity is  70X more than double-crested cormorant.
For PCB 126 the difference is 40X. REPs between the two are 0.07 and 0.02 for PCB
126. The question is what are minor differences for REPs? There should be some
evolutionary differences.  CYP1A1 has little difference in REPs and EQ0s are not
statistically different. As  long as dose-response curves are parallel, it works. There was
some concern that there may be an overestimation of relative potency. While there is
a lack of toxicity data for many REPs, those based on mortality data show some

   C.  Species differences in the sensitivity to TCDD

   For birds, the data are limited. There is a need to be able to explain the
mechanism of action at the receptor  level relative to different species' sensitivity. It
was mentioned  that receptor occupancy determines toxicity, but receptor populations
in tissues and organs are not routinely measured.  There are some limited data on four
species that supposedly can explain this but they were not identified.  There is still
high uncertainty in this area due to lack of information, that is equal  to B above and
greater than A.

    D. Experimental versus environmental exposures

    There is not a high degree of uncertainty.  Target tissue doses are important in the
environment.  Routes of exposure in laboratory are different from environmental
measurements in some cases, but for bird eggs there appears to be consistency between
lab egg injected doses and maternally deposited doses.  Egg injection can
underestimate responses based on methodology.

    E.  Differences in the relative potency across endpoints

There was general agreement that there is some uncertainty, as with items B and C,
above.  For fish and birds, REP/TEF data rely on ecologically relevant population
assessment endpoints through a reproductive endpoint (LQ0) for many congeners.
Some other TEFs in mammals may have greater uncertainty.

    F.  Use of the assumption of additivity

    TCDD &. 153 are a limited data set showing antagonism at certain relative
concentrations.  Generally, data for dioxin-like congeners are consistent with
additivity. Others present problems, because there is not much information. There
appears to be low uncertainty relative to other sources. Mink data clearly support the
additive model.  Also low uncertainty based on chickens, lake trout, and brook trout.
Environmental mixtures have a factor of 2 or 3 for variation.  For fish, the departure
factor from additivity is small (2-5). Methods to verify additivity have not yet been

Question  4. Are there different uncertainties due to chemical classification of the
chemicals (i.e., dioxins versus dibenzofurans versus biphenyls)? If so,  is there a
biological explanation for this difference?

Mono-ortho PCBs produce mixed effects and induce more than  CYP1A1. This is
problematic in fish, since fish are not responsive to mono-ortho-substituted PCBs.
Among vertebrates, there are class-specific differences.  For humans, PCBs 118 &. 105
give high tissue TEQs. The respective class differences are four orders of magnitude.
PCB 126  drives TEQ calculation across classes and results in reduced uncertainty.
Uncertainty is generally lower than  for items B and C, above.

Question  5. If we decide that a given species is the most sensitive species in the
ecosystem of study and relative potency values are available for that species for all of
the congeners present, should we use the relative potencies specific for that species or
should we use the WHO TEF values for all chemicals?

The purpose of the TEF is for risk assessment. WHO TEFs were rounded to
half/whole order of magnitude.  While there are great data gaps across species, there  is
also uncertainty that the assessed species is the most sensitive. Species-specific data
could be used.

Question 6. If we chose to use relative potencies specific for the species of interest
instead of the WHO TEF values, will we decrease the uncertainty in the TEF

Use of species-specific values will reduce uncertainty.

Question 7. How would a surrogate species be identified given that relative
sensitivities among species cannot be easily predicted?

It is common sense to use similar species.  There may not be a need for surrogates, if a
species of interest can be used. Uncertainty associated with extrapolations across
species is greater than that associated with extrapolations across eridpoints.

Question 8. Given the toxic endpoints used to establish each set of TEFs, how far-
ranging should the assessment endpoint diverge before the TEF predictive uncertainty
is high? How would uncertainty be valued to account for these other endpoints when
communicating risk?

The answer to this involves seeing data on species-related differences in REPs. Having
not seen this sort of data makes the question difficult to address.  At a minimum,  tier
consistency for a congener reduces uncertainty.

Question 9. What are the advantages and disadvantages of using the TEF
methodology over that for total PCBs?  Based on the type of AhR agonists that are
present at a contaminated site is one approach preferable to the other?

Pharmacodynamic differences between species result in a need for better exposure
assessment, which is  important for ecosystems assessment.  TEFs should be applied
only to biotic matrices, not to abiotic.  Different commercial mixtures vary in congener
composition, resulting in further uncertainty associated with total PCB assessment. It
may be possible to use indicator congeners to track and extrapolate from known-
composition mixtures.

An observer addressed the need for feedback on the global background of
contamination. For fish and wildlife, other Ah-active compounds, such as
azobenzenes, hexachlorobenzene,  some chloronapthalenes,  should be receiving

                             EXPERTISE GROUP
Facilitator: Charles Menzie

Group Members:  Janet Burris, Peter deFur, Lev Ginzburg, Wayne Landis, Mike
Meyer, Pat Cirone, Robert Pepin, and Steve Wharton (notetaker)

CM: provided an overview of goals of the workshop and reviewed key issues:

    •  we should ask ourselves how we as a group will help the other groups understand
      the risk asssessment process
    •  risk assessment terminology is important and should be clarified (three handouts
    •  screening level vs. baseline ERA; we should have a common working definition

WL:  For screening, the CCME has defined a Tier 1 assessment (not probabilistic, uses
an array of risk quotients, done quickly, some lab/field data).
JB: In Superfund, screening level ecological risk assessments (SLERAs) are used to
identify whether there is a problem, and the results are then used to design or scope a
more in-depth ecological risk assessment (ERA), which is likely probabilistic.
CM:  Typical characteristics might include identification of possible receptors, use of
literature values, limited additional investigation, benchmark comparisons ("quick and
easy" or "off-the-shelf comparisons).
MM: They are often used as a measure of exposure to identify the need for additional
SW:  The number of versions of SLERAs  is increasing, with states and non-
governmental professional organizations generating their own approaches (e.g., ASTM,
Soil Screening Levels).
LG: It is important that we not restrict our definition to only non-probabilistic risk
assessments; this is not necessarily the case for all SLERAs.

CM:  How do we characterize SLERAs in the context of TEFs?
WL:  All  TEFs are already a screening approach; they are an expression of a relative
value (i.e., high-med-low); this is Tier I.
PdF:  TEFs go beyond screening by virtue of the congener-specific nature of their
analysis;  TEFs shift the boundary between screening and full-blown risk assessment
(toward more detailed).
CM:  We have the potential to move to more  complicated evaluations using the TEF
approach. What are the next couple of steps in this process, and do the TEFs lend
themselves to taking these steps?  In taking these steps, do  we move toward more or
less uncertainty?
JB: ERAs will typically address specific receptors, pathways, and endpoints beyond the
limits used to derive the TEFs. In some ways TEFs may restrict your approach,
because some assumptions are inherent in the TEF approach. The decisions are not
transparent in the WHO document and these may affect their applicability.  One


should track the toxicity, assumptions, and uncertainties used in the TEF approach.
CM:  Is there a way to expose these? Would we need to collect additional data to
complete the SLERA and move beyond it to the ERA? Yes; examples include:

    •  site-specific data
    •  more complex data
    •  field data
    •  relate TEFs to field conditions (chicken versus eagle)
    •  try to reduce uncertainty of lab-derived TEFs

MM: Screening is setting up plausibility of effects, baseline allows conversion of TEF
to actual pg/g values.

Uncertainty Terminology
LG: There is another choice between deterministic and probabilistic. There is a
problem with the probabilistic description of uncertainty. The mode associated with
each distribution assumes independence (e.g., triangular distribution results from
algebraic extension of a normal distribution), but not all environmental data are
actually independent; rather, they are sometimes dependent. A "tight" mode may
underestimate extremes owing to too much central tendency.  If variables A and B are
dependent or correlated, we often don't know their actual relationships due to lack of
adequate data.
WL:  Correlation matrices developed in the field lead us to focus on dependent
LG: Probability bonds allow expression of uncertainties through generalized
application of full algebra.  Alternatively, "fuzzy arithmetic," where you do not assign a
specific uncertainty to distributions, allows description relationships.  Ignorance versus
variability should be identified so that we apply the correct tools to reduce uncertainty
when possible.  Fuzzy arithmetic does not consider dependent/independent terms. We
are more in the ignorance mode rather than the variability mode when extrapolating
between species.
CM:  Variability is measurable, but not reducible; lack of knowledge (ignorance) is
PdF:  It is important that everyone agrees with these concepts.
CM:  Which of these are we tracking?  There are two different techniques to propagate
these forms of uncertainty.  How do the risk assessments document elements of
uncertainties?  Should the details of each uncertainty be carried forward throughout
the risk assessment all the way to the decisionmaker (risk manager)?

[At this point, the group discussion turned to the issues raised in the charge questions
raised by the Planning Group; the comments/conclusions are listed under specific
questions where possible, otherwise, they are grouped at the end of this section.]

Question 1.  How does one characterize a risk assessment with total PCBs and
congener-specific PCBs?
PC: Do you lump the endpoints, or do you carry through the individual endpoints?
Congener  endpoint may be behavior whereas the endpoint for total PCBs may be

lethality.  Is this a piece of information that should be provided to the risk manager?
WL:  It depends on the nature of the site.
SW:  It also depends on the nature of the risk management decision to be made and
the level of uncertainty deemed acceptable in the decisionmaking process.

Question 2. How does one treat uncertainties associated with variable detection limits
for individual PCB, dioxin,  and furan congeners?
PdF:  Do we agree that where mixtures of compounds are at fairly high levels, those
ERAs are more certain than those where the levels are near the detection limits?
Generally, yes.
WL:  It may depend on the nature of the risk management decisions that need to be
PdF:  This also raises the multiple stressor issue.

Question 3. How does one characterize a risk assessment where the risks are primarily
due to the PCB or dioxin congener that has the lowest toxicity limit but the highest
concentration, or the highest toxicity but the lowest concentration?

Question 4. How does one distinguish the risk characterization of a "screening
assessment" from a "final risk assessment" where both rely on TEFs?
CM:  The differences in uncertainties between screening and final risk assessments
may be based on policy decisions related to screening level assessments. For a SLERA
(minimum case), what information do you carry on through the assessment—a simple
narrative providing a qualitative uncertainty analysis?
LG: If you use any numbers, you should fully disclose the associated uncertainties.
One should have some idea of the spread.
WL:  Be honest about your uncertainties, and if you need to, go out and take'some
MM:  The site-specific ERA can build on the SLERA.

Question 5. Describe the lines of evidence that should be included in all TEF risk

Question 6. Is a quantitative uncertainty analysis appropriate for the TEF toxicities?
(Toxicological uncertainty)
WL:  There is some quantification available, although exposure models (fate and
transport) are problematic in that they are not necessarily empirically derived (e.g.,
K^,) and may create the perception of "falsely precise" values..
PdF:  They also may be based on lab as opposed to field data.
LG: TEFs are not the only  component of uncertainty in the risk analysis.
PdF:  We  are willing to use  available data to do screening, including TEFs.
MM:  We should ask ourselves does your risk estimate bound the realistic value?

GROUP CONSENSUS: There was a general desire to provide the basis for a numeric
estimate even in the SLERA,

Question 7. Do you need to present all toxicity data when you do the risk assessment
or is a reference to an EPA or WHO summary document on toxicity sufficient?

Question 8. How do you describe the severity of effect? By individual congener or by

Miscellaneous Issues
LG: We should carry forward some form of uncertainty analysis.
PC: Do congener-specific ERAs leave out information that total PCBs ERAs capture or
presents more realistically?
CM: This issue should be dealt with in Problem Formulation - up front endpoints.

Exercise in  Cataloging Uncertainties
CM:  Derivation of TEFs by WHO included multiple uncertainties:  lab studies,
rounding up, limited information, lack of knowledge (including technical limitations in
LG:  Uncertainty is associated with the level of aggregation (i.e., only three classes of
animals are used in the ERAs and many chemicals versus using many species and only
a few chemicals).
CM:  Only some information is available; confidence is given qualitatively. There may
be a need to access the original data to quantify uncertainties.
PdF:  Each TEF should come with a set of information regarding the uncertainties
associated with the derivation of these values.
CM:  For the compounds that drive the risk, they should be based on higher quality
data/less uncertainty.
MM: For the regulators to adopt the risk assessment, they should be given the
information regarding uncertainty so that they may use it with the responsible parties
to reduce the risk.
WL:  EROD versus toxicity relationship is not expressed in the TEFs.  Are there data
for a higher-tiered TEF, or does the toxicity data have the lowest uncertainty?
CM:  If this is important, we don't have the information at hand.
WL:  There has not been a plot of the correlation of the tiered approach and the
associated reduction in uncertainties.  This would be helpful.

RECOMMENDATION: Justify the tiered approach qualitatively.  There may be
knowledge  not presently available in the WHO report.
Application of TEFs
CM:  What thoughts do you have on other uncertainties associated with application of
PdF:  There are uncertainties related to sources: Is the list of chemicals correct? What
form are they in? Are there similarities or differences in the chemicals in the
environment brought about through biological, chemical,  and physical transfers?
WL and PdF: Biotic transformations prior to entering aquatic systems are important.

CM:  Compare the uncertainties of this approach with modeling total PCBs.
WL:  We should ask ourselves several questions: How well do these models work?

Are they predictive? Are they applicable to all media/endpoints?  Are model
uncertainties constant across all compounds?
WL:  TEFs have been described as order of magnitude estimates.  When modeling
populations, order of magnitude changes are very important.
LG: It is foreseeable that we may likely have to apply standard food-chain models in
the future (to reduce modeling uncertainties).
WL:  Detection limits are a source of uncertainty
PdF:  Whether the risk assessment is based on a few congeners as opposed to many
congeners is important in determining heterogeneity in exposure.
WL:  There are spatial and temporal issues associated with sampling; aggregation of
samples destroys heterogeneity of environmental data.
CM:  Would you expect greater variability in tissue with congeners than an aggregate
(total PCBs)?
WL:  At lower concentrations, variability would likely be greater with .congeners.

RECOMMENDATION: Variability in tissue concentrations for congeners vs.
aggregate Ah receptor agonists may present significant uncertainty. Data addressing
this issue should be reviewed/compiled.

Effects-Related Issues         •                                        /
CM:  What are the sources of uncertainty?
PdF:  The greater the phylogenetic extrapolation, the greater the uncertainty. Also,
uncertainty associated with life stage sensitivities may be as great (due to metabolic
changes related to age/life stage).
LG: Simplistic description of populations using extreme endpoints (e.g., death) is
meaningless. We should use more meaningful endpoints. Resilience should also be
PdF:  There may be genetic determinants of population effects (e.g., Fundulus).
WL:  Patch dynamics present uncertainties (breeding occurring in one area versus
adults living in another).
LG: Endpoints derived for one life stage applied to another life stage (e.g., survival of
adults versus larva)
MM: When you get down to dietary habits, you need precise data for TEF
development.  Preferential exposure by congener may be different than total PCBs.
WL:  Compounds are distributed preferentially into food items, and BAF for exposure
may be determined by log P.
MM: You can differentiate dietary habits by age.
PdF:  Gender, life stage, and overall condition will  contribute uncertainties.
WL:  The shape and slope of the dose-response curve (TCDD) may add to uncertainty
with respect to measured values in environmental media.
CM:  Consider other dose-response interspecies differences (Spehar's paper). There
are significant interspecies differences and there may be a broad range within a species
by compound.
WL:  Source(s) of population regulation are important.
MM: When you look at the overall population versus the impacted (or studied)
population, the level of risk may be different.
CM:  Social considerations (values) come into play when determining level of

protectiveness (extinction of protected species versus prevention of population
WL: The shape of the dose-response curve will impact what happens on a meta-
population level.

CONSENSUS: Congener-specific information is important for exposure and effect
characterization, predicting severity of adverse effects, magnitude (number of
individuals affected), and subtle effects. The details of uncertainties associated with
congener-specific ERAs should be retained throughout the risk assessment and
interpreted for the decision maker.  Congener-specific risk assessments, are actually a
form of multiple stressor risk assessment.  These points should be discussed during
problem formulation and in the selection of appropriate assessment endpoints.

Charge Questions Related to Prospective Case Study
CM: Are the TEF uncertainties greater than or less than other approaches?

CONSENSUS: They are no worse than other approaches (ignorance exists in both
areas), they are not the only source of uncertainties, and TEFs may present

CM: Biologically based TEF assays, what would be their strengths?
SW: Generally, assays are cheap, and they provide an aggregate response.
WL: They allow direct testing (validation) of risk hypotheses (risk estimates).
WL: If TEF data contain bioassay (toxicity) results, then you may make direct
extrapolations to risk management decisions.

CM: What would be their disadvantages?
LG: The metabolic processes are undefined.
MM:  We are extrapolating from assays on a limited number of organisms to
SW: There is an inability to know which  compound to regulate.

CONSENSUS: Bioassays provide another line of evidence. Biocriteria versus
chemical criteria present difficulties in implementation.  Testing methods need

Responses to Charge Question IV-3 . (discussion also relevant to Question 5, above)
LG: Additional demographics on species are necessary, both historical and present
MM:  Precise dietary habits are needed, including ecological implications.
WL: Abiotic characterization of ecosystems is needed (e.g., physical, chemical,
LG: Additional information on patch dynamics is needed for interpreting population
effects, especially for statutorily protected species.
PC: We should exercise caution discounting risk when laboratory assays or modeling
fails to demonstrate predicted risks. They represent one line of evidence.

        Appendix E


Breakout Group Assignments
Wednesday, January 21,1998—10:45AM-3:45PM
Thursday, January 22,1998—8:30AM-12:30PM
Group 1
Chair: Peter DeFur
Group 2
Chair: Janet Bum's
Group 3
Chair Charles Menzie
TEFs Experts
Jay Gooch (fish)
Martin van den Berg (mammals)
Sean Kennedy (birds)
Mark Hahn (fish)
*(See EPA/DOI Planning Group)
Bjorn Brunstrom (birds)
Richard Peterson (fish)
Bert van Hattum (mammals)
Steven Bursian (birds)
Fate & Transport/BAF Experts
William Adams
Joseph DePinto
Patrick O'Keefe
Christopher Metcalfe
Mark Servos
Lynn McCarty
Population Modeler
Lev Ginzburg
Mike Meyer Wayne Landis
Risk Assessor
Peter deFur
Janet Burris Charles Menzie
EPA/DOI Planning Group
Gerry Henningsen
Lisa Williams
Robert Pepin

*Mike Devito
Tim Kubiak
Steve Wharton
Steve Bradbury
Phil Cook
Pat Cirone
Cynthia Nolt
Don Tillitt

  Printed on Recycled Paper

                          WORKGROUP #1
                       Facilitator:  Peter deFur

Prospective Case Study

       The discussion focused on the prospective case study, which involved
permitting a new pulp/paper mill on a generic lake in the midwestern United
States. The point of the discussion was to review the uncertainties and issues
raised by using TEFs to evaluate permit application and discharge conditions for
the hypothetical facility. The facilitator briefly reviewed the case, going over the
basic facts and issues. The group discussed the elements of the case: the complex
effluent, the multiple chemicals with dioxin-like activity, the ambient atmospheric
input of dioxin-like materials, and the level of scientific knowledge and
uncertainty  surrounding the case.  The group agreed that several issues were
critical: bioaccumulation, ambient inputs, and species variability.

       The group opened with a discussion of the bioaccumulation factors used in
this case and the issues associated with application of BAFs to TEFs. The charge
questions included several for this case that the group felt were a good starting
point, specifically:   "What errors are associated with the use of BAFs, given the

       The group discussed the scientific nature and derivation of the BAFs , and
compared these with the TEFs for wildlife.  The BAF issues are largely ones of
application,  and do not so much directly affect the TEFs as they influence the final
outcome of the technical analysis in which the toxicity of a "mixture"—in this
case, a complex effluent—is assessed. The TEF approach requires the use of BAFs
because of the nature of dioxin-like compounds.

       The group agreed that there is significant uncertainty in "generic" BAFs
determined  for one site with the intent of using the BAFs elsewhere. However, the

species used in deriving the BAFs and the basic similarity of the systems (both
northern freshwater systems) provides greater usability in the application to the
system in the case study.  Such is not always the case for other applications. The
group agreed that the greatest certainty in BAFs exists in systems where the
species and the aquatic conditions are the same. The group recommended that
research efforts to improve the understanding and use of TEFs focus on
applicability to the same system under different conditions rather than on
expanding to new systems. The group felt that there was less variability in the
system than in the variation among systems, and that the need for site-specific
factors was greatest when moving across widely different systems. The
interspecies variation largely came up when addressing the species that had no
counterparts in other systems. One great source of variability among ecosystems
derived from the differences in trophic structure, especially the number of trophic
levels and the nature of the top level.

       The use of a Monte Carlo or other probabilistic approach generated some
discussion,  but most agreed that the Monte Carlo approach could give greater
insight into the variations in the system, providing that the Monte Carlo was not
misapplied. Additional information on this issue is provided at the end of this
summary of the workgroup discussion and in the summary of discussions in the
Risk Assessment Expertise Group (Appendix D).

       At least one member felt that the use of probabilistic approaches was
largely over-rated. He noted that everyone had a propensity for using the same
independent variable approach with the same distributions, mostly normal (in
statistics). This member noted that the normal functions used to predict the
distributions of these things do not follow the way in which the effects actually
occur in the natural world. In fact, he noted, real-world variables are often
dependent and not normal in distribution, but either non-normal or stochastic.
Thus, some of the conditions cannot be predicted in the Monte Carlo approach
because the wrong formulas are being used.  Other statistical approaches will offer
a different analysis of those dependent and  non-predictable events. Two important

points are that the variables are often linked, that is they are dependent, and that
the distributions are not normal, in a statistical sense.  Additional discussion of
this topic is provided later in this summary.

       In discussing this case, the group decided that it would be better to
understand the behavior of a single congener (i.e., TCDD) across many different
systems than to understand the behavior of all congeners in a single system and
then expand to the next system.  The reason for this is that the behavior of
congeners in relation to one another is more constant than other variables. Thus,
by knowing how one congener performs, the others can be extrapolated with
greater certainty than other extrapolations. The group agreed that acquiring BAF
data for a suite of chemicals would be ideal, but the utility of such data is directly
related to its specificity.

       The data should be coherent among data types and forms for calculated
versus measured values and for field measurements versus estimates. Different
approaches (e.g., TEFs for individual congeners versus TEQs for a whole effluent)
should reveal concordance or lack thereof.  These approaches then turn out to be
data checks.

       The present case offers  a scenario with multiple sources and multiple
chemicals and the requirement to conduct a TMDL for the water body.  In this
scenario, the TEF makes it possible to examine options and compare data from
widely disparate sources. This is a positive feature of the approach.

Retrospective Case Study

       This case, which involves contaminated sediments from a spill, was
handled somewhat differently in the group discussion than the case study from the
previous day.  For this case, after discussion with the steering committee, the
facilitator urged the group to reach a decision  regarding the use of the TEF
approach in applying the available data to a decision. The group was asked to

determine if a decision could be reached, and specifically if the TEF approach
would or would not affect the outcome of the decision.

      The facilitator summarized the case as follows. A chemical spill took place
previously, resulting in contaminated sediments in an upstream segment of a river
that flowed into a lake. Despite some time passing, populations of several species
of wildlife seemed to remain at some level of risk or impairment.  Some of the data
on the populations may be more qualitative and observational than quantitative.  <
Three species have already been identified as species of concern: lake trout, otter,
and Caspian tern.  Data were provided to assess the toxic load from the PCBs
using two alternative approaches, either based on total PCB levels in the tissues
(and sediments), or based on the TEF approach for all dioxin-like compounds
summed across individual congeners.  The case provided previously determined
decision reference values for action for each of the three species, using either
approach. These reference values were given as single values and as ranges,
indicating that if the predicted exposures exceeded these reference values, some
sort of action would be recommended.

      The case description included the notes that atmospheric deposition is a
source of TCDD, that there are indications of population effects on wildlife, that
the analysis already includes a conceptual model and selection of endpoints, and
that the exposure pathway has some predictions, but uncertainties as well. There
is knowledge of prior eutrophication, with unknown consequences. The group
agreed, reluctantly, to treat the source(s) as constant, without degradation or
recalculation or loss. This assumption makes the "no-action" option less viable.

      The group could not make a recommendation for the site to be cleaned up,
but attempted to determine whether the data warranted some type of conclusion
about the, risks that might lead to an action, which might be to leave the system
alone, study it more, or identify management options to reduce/control toxicity).

       A brief analysis and summary of the data on exposure and reference dose
 revealed that the two approaches gave somewhat different conclusions. The TEF
 approach yielded exposure concentrations higher than the reference values, for the
 most part, while the total PCB-estimated exposures yielded values that were
 somewhat lower, and less clearly exceeding the reference values.  This observation
 was in a general form, recognizing that the magnitude of the ranges made it
 difficult to determine whether or not the predicted values overlapped the ranges in
 Table 5 of the case study.

       The group discussed how to interpret the data on exposures from the two
 estimation methods. The resulting discussion revealed that individual members of
 the group did not view the results the same, in that some members did not see
 such a great difference in the interpretation based on the two approaches. The
 group members who thought that results did not differ between the two
 approaches saw both as giving borderline "positive" (exceeding the reference dose)
 results. Everyone in the group considered the data to be at least in the borderline
 category for making a decision, but everyone agreed that there was room for
 interpretation of the meaning and significance of the impacts of the exposures.
 One concern expressed by several members  of the group was the  lack of data on
 ranges, or on variability of the data. The ranges of values given for the NOAEL in
 Table 5 was not sufficient. The group wanted data on TEF variability, BAF
 variability, toxicity, and so on.  There was a general discussion of the need for
 species-specific dose-response functions for  the biological processes in this
.ecosystem, and several members wanted to  see the measures or other expressions
 of variability in the data, including the TEFs.

       The group had a range of opinions regarding the quality and sufficiency of
 the data in supporting a decision on the case using the TEF approach. Basically,
 all thought more data would  make the decision easier, but only part of the group
 were satisfied enough with the available data to make a decision.  Others wanted
 to see site-specific data before concluding that a decision could be made, and still
 others thought that population data were needed for the three species before
 decisions could begin.

       All in the group were satisfied that the TEF approach provided a useful way
to understand the data, the case, and the situation. Also, the available data were
sufficient to support use of TEF approach at least for screening level analyses.

Discussions Relevant to Both Case Studies

       A number of points were raised in the course of the two days that the
group agreed were important to include in the summary report of the workshop.
The following paragraphs were drafted by individual members of the workgroup to
reflect the nature of the discussion and the agreement or lack thereof among the
members of the group.

       Need for Ground-Truthing of the TEF Approach. Field verification of
TEFs and the resulting  TEQs is highly desirable, in the least, and was considered
essential by some. The TEQ approach has been examined in the field for birds
(Tillitt et al., 1992; Murk et al.; Kubiak et al., 1989; Harris et al.), fish (Cook and
Peterson's retrospective work on lake trout in the Great Lakes), and mammals
(otter in Europe). Effects observed in the  field are generally consistent with what
would  be predicted from exposures expressed on a TEQ basis.  Differences in
TEQs among colonies of double-crested cormorants in the North American Great
Lakes explained more of the differences in hatching success among colonies than
did differences in total PCBs (Tillitt et al.). Among-species variability in absolute
sensitivity to dioxin-like compounds is greater than among-species within-class
variability in TEFs.

       In the ecological risk assessment process, exposure and effects data are
integrated and the potential for risk is characterized. As a general rule, when
exposure levels exceed the effects level (threshold), expressed as a risk quotient
greater than 1.0, excess risk is expressed. When excess risk is calculated (e.g.,
when the summation of TEQs exceeds a threshold effects value), it is important
that the potential for effects to occur in natural environments  (i.e., at the
population or community level) be assessed.  There is a need to ground-truth the
                                    E-7                                    .

TEF/TEQ approach such that when this approach is used to demonstrate risk that
measured effects at those exposure levels have been observed in field populations.

       Status of the TEF Approach as a Screening-Level versus Decisiohmaking
Tool. As a general rule, screening level assessment/ranking/scoring tools should
have an accuracy in the range of a factor of 5-10. For more definitive/quantitative
risk assessment, the accuracy of the assessment tool should be less than a factor of
5, preferably between 2 and 3. In general, the uncertainty associated with the
derivation of a TEQ based on class-specific TEFs (i.e., TEFs for fish, mammals, or
birds) is on the order of a factor of 5-10.  In light of this, we conclude that the risk
ratios derived using the application of a TEQ value based on TEFs are best viewed
as a screening tool, to provide direction in determining additional data needs. We
recognize that in some cases, for example where the species and endpoints are in
fact the species and endpoints on which the TEF values are derived, the TEQ
portion of the risk quotient may be sufficiently accurate to justify more confidence
in the estimation of risk inferred from the hazard quotient. In these cases,
obtaining a better estimate of the TEQ would be less of a priority than obtaining a
better estimate of exposures.

       This level of certainty in the TEF-derived estimates was apparent in the
interpretation of data from the two case studies. One of the reasons the group
members held somewhat different interpretations was the  understanding of the
data that support the TEFs used in these cases.  The members with greater
familiarity with this data were more satisfied with the final outcomes.

       Uncertainty. Throughout the two case study discussions, group members
urged the incorporation of variability and/or uncertainty in the numerical
expressions and elsewhere. This expression of uncertainty may take different
forms, only one of which is a Monte Carlo or probabilistic analysis. The group
acknowledged the dual nature of uncertainty, which includes both statistical
variability and the unknown. The group was confident that the former type of
uncertainty was adequately, if not always, expressed in statistical ways, often as

standard error or mean, as confidence limits, or other such numerics.  However,
the latter type of uncertainty, unknowns (also called ignorance), is less well
expressed and is not conveyed through  traditional statistics or through
probabilistic approaches.

       Both no-effect levels (NOAELs) and computed values, whether of risk or of
variables such as BAFs, have to explicitly incorporate error.  No meaningful
comparison of values (e.g., benchmark versus NOAEL) is possible otherwise.  In
the retrospective case study, the benchmark values did come with a range, but the
computations did not include explicitly propagated errors in the input values,
which are significant.

       Uncertainty does not have to be described in a probabilistic framework.
Lack of knowledge, or ignorance, does not easily lend itself to a frequentist's
interpretation. We have to clearly distinguish natural variability (heterogeneity)
from ignorance-based uncertainty. The first can be described probabilistically; the
second, by either range or in terms of fuzzy arithmetic (i.e., using formulas,that do
not recognize frequentist views but attempt to  reflect subjective uncertainty).
Reference to such techniques (Person and Ginzburg, 1996) can be found in Dr.
Ginzburg's pre-meeting comments (Appendix C).

       Grounding of the TEF Approach in Ecology.  The idea of a so-called "no-
effect" concentration certainly provides a useful first cut, but it effectively
disconnects toxicologically-based decisions from ecological considerations. In
practice, and in our case study, the "no-effect" concentration is exceeded, and
another level of analysis (i.e., population or ecosystem dynamics) is need to make
ecologically-based judgments.  Specifically, we need information about:

   •    the slope or, preferably, the shape of the dose-response curve above the
       "no-effect" level; and
   •    demographic characteristics of the target species, including an  idea of the
       strength of population regulation (density-dependence).

       In the retrospective case study, even though a no-effect level is exceeded,
population level consequences are uncertain and may not warrant immediate
action but rather a closer look at population dynamics.  The group was confident
that many, if not most, cases would exhibit some ambiguity in the  interpretation
of effects of dioxin-like chemicals on animal populations. Thus, data on and
evaluation of population dynamics is the next step.

       Measurement Issues.  The group agreed that current knowledge of dioxin,
Ah receptors, the dioxin-like compounds, and the mechanism of action of the
dioxin-like compounds necessitates measuring all of the relevant congeners when
taking environmental samples.  It makes no sense scientifically to measure only
the one congener, TCDD.

                         WORKGROUP #1
                      Facilitator: Janet Burris
Prospective Case Study

      The following sections summarize the discussions completed by Group 2
during its review of the prospective case study. The summary is organized
according to the primary issues discussed, and indicates which members
contributed to the related text.

      TEF Derivation and Application in Ecological Risk Assessment (Janet
Burris and Mark Hahn). The group reached an agreement that the TEFs used in
ecological risk assessment should be selected using a hierarchical approach.
Species-specific values should be used if available or a value for a closely
(phylogenetically) related species.  The WHO consensus TEFs would be used as a
default.  The proposed system also considered in vivo results preferable to in vitro
in selection of the most appropriate endpoint. See the section entitled "Selection of
TEF values for use in a TEQ-based Ecological Risk Assessment)," in the Group 2
summary of the retrospective case study, below, for a more complete description of
the proposed hierarchy.

      The group expressed the need to have more information on the derivation
procedures used to identify the WHO TEFs, as well as the underlying data.  This
information is necessary to understand the uncertainties in the values and to carry
those uncertainties through the risk assessment.

      The group felt that the rounding procedures used in the WHO TEFs do
introduce some uncertainty in the ecological risk assessment. This uncertainty is,
however, quantifiable and can be evaluated in a sensitivity analysis where results
are calculated using both rounded and unrounded values. A sensitivity analysis
can also be performed to evaluate uncertainties associated with the use of species-
specific TEFs versus the WHO TEFs and endpoint-specific TEFs versus the WHO

TEFs.  Further quantification of the uncertainties in the derivation of the WHO

TEFs may be limited as the TEFs were n6t derived in a systematic manner. The
exact process used to derive the WHO TEFs is not specified in the current draft

report.           -
       Discussion of Appropriate TEF Values for Species of Interest in the Case

Study (Janet Burris). The group's selection of TEFs for the species of interest was:

• Bald Eagle:  The group recommended using the WHO avian TEF values. The
  rounded and unrounded values should both be used to perform a sensitivity
  analysis. The group was comfortable with extrapolating from the endpoints
  used to derive the TEF to the reproductive  assessment endpoint, but "this needs

• Bull Trout:  The group recommended using the rainbow trout values and the
  endpoint of early life stage mortality.

• Otter: The group recommended using the WHO TEFs as a default, since values
  for the otter or more closely related species are not known to be available.
  Uncertainties in the extrapolation from the TEF endpoint to the assessment
  endpoint are not  large, as these values (the  TEFs)  represent relative potency

       TEF Approach Compared with TCDD or Total PCBs (Bjorn Brunstrom).

The group agreed that the TEQl approach provides significantly more information

compared with an assessment based on TCDD alone, total PCDDs, or total PCBs.
However, use of the TEQ approach should not replace or exclude risk assessment

based on total PCBs, since non dioxin-like effects may also be important. In the

prospective case study,-the TEF/TEQ approach provides information on the
potential toxicity of the mixture of congeners  in the effluent and identifies the
specific congeners that may make the largest contribution to toxicity.

       Using a TEQ approach in the modeling process is principally similar to
modeling a single compound. The challenge is to get valid fate and transport-
related parameters for a number of compounds.                          0

      The Use of Median Values to Derive TEFs (Mike DeVito and Janet
Burris).  The group discussed the issue of the use of median values (EQ0 and LC50
data) to derive the TEFs and the implications of this on their application in risk
assessments where no effect levels are used to identify risks.  The group decided
that median values for the derivation of TEFs is appropriate and does not effect
their application in risk assessment.

      Derivation of REP values typically use the ratio of EDj0s, EC50s, LOELs or
NOELs of the test chemical compared to TCDD. A question often asked about
the use of the TEF methodology is whether the TEF should be based on the ratio
of the lower end of the dose-response curve as opposed to the EQ0s. The relative
potency for full agonists should be independent of the level of response where the
measurements are determined.  One advantage  of using ED^s is that the ED50 can
be determined with greater accuracy and precision than the NOEL, LOEL, or
either the ED01 or ED10. The increased  precision and accuracy are related to the
greater ease of measuring a 50% response above background. In comparison,
determination of LOELs, NOELs, or ED^s and  ED01s are fraught with
uncertainty.  The ability to accurately detect NOELs and LOELs, EQ0s, or ED01s
is dependent upon the magnitude of the maximal response compared to the
controls and the variability in the measurement of the control and lower-dose
groups.  Estimates of these low-dose parameters are also highly dependent upon
study design and dose selection. The uncertainty of the EDj, or EDIO estimate is
much greater than the uncertainty of the estimated EQ0 (DeVito et al., 1997).
The increased uncertainty of the estimate of the low-dose parameters would
increase the uncertainty of the REP. While REPs may be dependent upon where
on the dose-response curve they were derived, the greater accuracy and precision
of the ED50 determination provides a significant advantage for its use in estimating
the REP.

      While the relative potencies of full agonists are independent of the
measured response, the  same is not true for partial agonists. Partial agonists do
not have the same intrinsic efficacy of full agonists and may antagonize the effects

of full agonists under certain conditions. Assigning REPs or TEFs for these
chemicals is problematic.  Under certain conditions (predominately low-dose
exposures), the interactions of full agonists and partial agonists may be additive.
Under high-dose conditions, the interactions of full and partial agonists maybe
antagonistic. The use of the TEF methodology for partial agonists should be
viewed with caution.  The interactions may be additive in the low-dose region but
non-additive in the high dose region of the dose-response curves.

       Use of Bioassay-derived TEOs (Christopher Metcalfe).  The group
discussed the potential for using bioassay-derived TEQs in the prospective risk
assessment scenario. Reservations were expressed over the value of this technique
for monitoring total TEQs discharged in wastewater. The potential for generating
"false positive" responses in in vitro assays was considered high. For instance, in
pulp mill effluents,  there are high concentrations of potent EROD-inducing
compounds (e.g. retene) which are not AhR-agonists.  PAHs will also give a
response in some in vitro assays if they are not removed from complex
environmental mixtures by fractionation. The biologically-based TEQ assays are
not chemical-specific and therefore do not show causality.

       Discussions with Sean Kennedy indicated that bioassay-derived TEQs
could be applied to estimating burdens of planar HAHs  in fish and wildlife in the
lake system, both before construction of the mill and for monitoring purposes after
the mill is operational. This was discussed by the group and was considered a
valid application for these techniques.

       Challenges and New Uncertainties Associated with Modeling the  Exposure
of AhR Agonists (Christopher Metcalfe). The group concluded that the use of the
TEF/TEQ approach in this case study introduces no uncertainties in exposure
estimation that are  not also common to other chemical-specific assessments.  The
challenge of the TEF/TEQ method is that it requires the modeling of individual
congeners, as well as modeling of the fate and transport of the many congeners, in
comparison to traditional  modeling for individual or lesser numbers of chemicals.

       Measurements of specific congeners may be problematic, considering
congener-specific detection limits; and the problem will be greater for analyses of
water, due to the higher volumes required than for sediment or tissue samples.
When the detection limits are high, uncertainty is high, but the uncertainties are
offset by the advantages of using congener-specific as opposed to aggregate (i.e.,
total chemical) values.

       The group further discussed the fate and transport issues in two parts:
physical/chemical and metabolic parameters. Regarding the former, there were
concerns expressed about the quality of chemical and physical data that could
drive the mass balance models involved in the risk assessment process for the
prospective scenario. The quality of log K^,w data was considered relatively high,
but there was concern over the lack of .empirically-derived I^,. data, and over the
quality of !<„,. values derived from K^. There was also concern over the lack of
credible Henry's Law constants (H) and photolysis data for planar HAH
congeners, although it was acknowledged that this may not be a problem for
hydrophobic compounds that bind readily with the paniculate phase. Finally,
there was discussion of and general agreement with a point made by Joe DePinto
(in the Wednesday morning meeting) and by Phil Cook, that the partitioning of
the planar HAHs between sediment and water and the particle dynamics in the
aquatic system are the most important processes that will drive the mass balance
model. There was concern expressed that "getting these processes right" will
require an extended and expensive research effort, a luxury that may not be
feasible for the risk manager. For further discussion of this topic, see the section
of this  summary dealing with generic mass balance modeling, below.
       Regarding metabolic parameters, concern was expressed over the lack of
information on the metabolic capacity of target organisms for planar HAHs.  This
may have implications for estimating BAFs and BSAFs. It was pointed out that
many fish-eating birds and mammals appear to "enrich" PCB congener 126 and
appear to metabolize PCDFs relative to the levels in the fish they are eating.
Therefore, a knowledge of the metabolic capacities of fish-eating target vertebrates

for the various planar HAHs is essential for generating BAFs in the prospective

       Requirements and Considerations in Analytical Design  Associated with
TEF-Based as Compared to Aggregate Analyses (Patrick O'Keefe). The TEF
approach requires quantitative analytical data on a large number of PCB and
PCDD/PCDF congeners. Consequently, more complex and costly analytical
methods must be selected compared to those used for the measurement of total
PCBs. In general terms, sample extracts will need to be analyzed by gas-
chromatography/mass spectrometry (GC/MS) using isotope-labeled internal
standards-rather than by GC with electron capture (EC) detection, the current
method-of-choice for total PCBs. While many laboratories engaged in PCB
analysis do not have the equipment or expertise to carry out these procedures,
those laboratories currently involved in PCDD/PCDF analysis do have the
appropriate equipment and expertise.

       Since compound identification is more specific using GC/MS methods, the
possibility of false positives is considerably reduced compared to using GC/EC.
Quantitation is also improved when isotope-labeled internal standards are
included rather than the surrogate compounds used in GC/EC.  However, it should
also be understood that, in the TEF approach, the PCB and PCDD/PCDF
congeners with the largest contributions to the TEQs are usually present at very
low concentrations relative to the total PCB concentration. Consequently, rigorous
quality control procedures will be required to ensure accuracy and precision in the
analytical data. In addition to the generally accepted internal quality control
samples (blanks, duplicates and spikes), standard reference materials should be
used for calibration purposes. Currently, fish tissue and soil sample standard
reference materials are available for 2,3,7,8-substituted PCDDs/PCDFs and for
cpplanar (non-ortho) PCBs. Similar materials are not currently available for mono-
ortho PCBs. Round-robin studies using selected samples from different matrices
represent an alternative method for comparing the results from different

       In the prospective case study, there are five matrices of concern: avian eggs
(bald eagles), fish tissue (bull trout and lake trout), mammalian tissue (otter),
sediment, and water. In this scenario, a fish tissue TEQ residue level of 9 pg/g was
judged to be the level of concern. Since the proposed pulp mill would release
PCDDs/PCDFs but not PCBs, the discussion can be limited to the two former
compound classes. Laboratories with proficiency in PCDD/PCDF analysis can
achieve detection limits of 1 pg/g for individual PCDD/PCDF congeners. With this
detection limit, it should be possible to obtain reliable data near the level of
concern for fish, especially since the major contributor to the TEQ value (2,3,7,8-
TCDF) has a TEF value of 0.05. Accuracy of ± 30% should be achievable with a
signal-to-noise value of 10,  but would be reduced for data near a detection limit of
3:1, the minimum detection limit used in many laboratories.
       It is difficult to determine the residue levels of concern in the bald eagles
and otters, since the data are discussed in the scenario using water quality
considerations which are in turn related to ingestion levels leading to toxic
responses. However, if it is assumed that the no-effect threshold levels in the
retrospective study (100 pg/g for bird eggs and 60 pg/g for mink liver) are also
appropriate for the prospective study, then laboratories that are capable of
analyzing the fish tissues should have no problem meeting the higher detection
limits of the avian and mammalian species. The same considerations apply to
sediment samples, since the biota-sediment accumulation factors are all less than
1 and therefore the sediment concentrations will exceed the fish tissue

       When the tissue levels of concern are translated into water quality
guidelines, using biota accumulation factors, the maximum allowable total water
concentrations  (MACW) will all  be less than 2 pg/L, and in many cases they are
below 0.1 pg/L. Currently, there are no routine laboratory procedures available
that are capable of meeting these detection limits.

       This discussion of the prospective scenario assumes that all of the
permissible discharge will be allocated to the pulp mill.  However, in the

description of the risk assessment scenario it was proposed that only 25% of the
maximum allowable load (MAL) would be allocated to the pulp mill. If the mill is
required to measure this increment at the levels of concern for the three species
described above, .the data will be very close to the 1 pg/g detection limit, and
accuracy will probably not be better than a factor of 2 and possibly even a factor
of 10.  As a final note, the group realized that, in this scenario, use of the TEF
approach will require  the plant to assume responsibility for producing state-of-the
art analytical data.  However, by doing this the plant management will have
considerable flexibility in controlling the mix of PCDDs/PCDFs in the plant

       Challenges in the Modeling the Food Chain Transfer of AhR Agonists
(Janet Burns and Mike DeVito). The group concluded that  sound modeling could
be completed if the transfer of congeners from the sediment/water interface and
sediment transport within the lake are both well understood.  However, the group
observed several challenges in the modeling of food chain transfer.

       First, there is concern that the poor understanding of biodegradation and
metabolism of specific congeners may limit modeling. Biodegradation and
metabolism rates are absent or incomplete.  Second, composition of the diet and
metabolism affect the transfer of congeners. There are large mixture composition
changes of congeners  from plankton to fish and from fish to  fish-eating birds.
Third, the food chain transfer of congeners is species-specific. To address this
factor, knowledge of the composition of the diet of species within the food chain
needs to be clearly understood and considered in the modeling exercise. Fourth,
the data available for  estimates and projection of food chain transfer are good for
fish but are not adequate for wildlife. Measured biomagnification factors (BMFs)
are better for fish, but are much less certain  than those for fish-to-wildlife

       The group observed that BMFs should be consistent with the dosimetric
basis of the TEF. In ecological risk assessments of dioxin-like chemicals, BMFs

and TEFs are important parameters in exposure and toxicity assessment,
respectively. The BMP is a function of the physical/chemical and pharmacokinetic
parameters of the individual chemical. The TEF of a chemical is related to its
binding affinity to the Ah receptor and its pharmacokinetic parameters compared
to TCDD.  Because BMFs and TEFs are both dependent upon pharmacokinetics,
differences in pharmacokinetics of a chemical between species may alter the BMF
and the TEF in the same direction. For example, if the BMF increases between
two different species for a test chemical, while the BMF remains constant for
TCDD, the TEF may increase across these species as well, since retention or
accumulation of the chemical has increased relative to TCDD across the two
species. A note of caution is that the relationship between BMF and TEF is not
direct. Chemicals that have high BMFs may have low or no TEFs. However, if
BMFs for a chemical change dramatically between species, the TEF may also
change dramatically between  species, Hence large changes in BMF between
species warrant further examination of the TEF for that congener.

       Bioaccumulation Factors (BAFT Biomagnification Factors ('BMF'). and
Toxic Equivalency Factors (TEF) (Mike Meyer). Site-specific BAFs for PCB and
dioxin congeners will provide greater accuracy in prediction of wildlife and fish
tissue congener concentrations than will extrapolation of the BAF derived for the
GLWQG. The GLWQG BAF was derived from Lake Ontario data that predicted
tissue concentrations of total PCBs and 2,3,7,8-TCDD in lake trout. The trophic
structure of water bodies can differ greatly, as a function of the complexity and
structure of the food web. Stressors other than chemical contamination (e.g.,
climate, nutrient loading, introduction of exotic species, and so on) can create
perturbations in the recruitment at various trophic levels (such as prey fish).
Elimination, reduction, or additions of organisms to the food  chain  can affect the
trophic transfer of chemical contaminants, thus altering BAFs from site to site.
When using the TEF approach for risk assessment, calculations of BAFs become
more complex, as the individual congeners will have unique BAFs, and those BAFs
can vary between sites. If this level of precision and accuracy is desired, sampling,
modeling, and database management efforts will become more complex and costs

will increase. However, the additional information will allow risk managers to
provide congener-specific discharge allowances or remediation goals.

       It has been shown that non-ortho PCB congeners are more readily
bioaccumlated and are more resistant to metabolism when compared to ortho-
substituted PCB congeners. It follows that wildlife tissues may contain a larger
proportion of dioxin-like PCB congeners per gram of total PCBs than do fish,
reflecting the increased toxic potency of the total PCBs measured in their tissues.
Therefore, direct extrapolation of the TCDD BMP from the GLI will provide
erroneous risk estimates.  A recent study in the Netherlands demonstrated the
effect of this error. In that study, a diet-specific BMP of 14 was calculated from
fish to otter on a total PCB basis, however the BMP for total TEQs was 41
(Leonards et al.,  1997, Env. Tox. Chem 16: 1807-1815).  This was mainly due to
the high BMP of PCB 126.  When incorporating BMFs into the risk model, it is
essential that congener-specific BMFs be used, and, when possible, species-specific
BMFs should be measured at the risk assessment site.

       Possible Errors in the Application of the TCDD Water Standard (Janet
Burris). The group identified  two possible errors in the application of the TCDD
water standard to the prospective scenario.  The standard does, not consider the
enrichment of PCB 126 from  fish to wildlife or the loss of chlorinated
dibenzofurans in some species of birds. Some members of the group observed that
underestimation of effects may result.

       Other Possible Approaches:  Generic Mass-Balance Modeling (Christopher
Metcalfe). The group explored the idea of alternate or improved approaches to
the solution of the problem presented in the prospective case study.  How could
we do it better?  Several group members acknowledged that a model for the
system could be constructed but it was not  clear whether such a model would be
predictive given the uncertainties. Other existing food chain models could be used
if they could handle the metabolism issues.  Another idea that was put forth
involved the use of a generic mass-balance model.

       A potential problem with the prospective scenario for TEQ-based risk
assessment is the complexity of the mass-balance modeling exercise. Discussions
at the workshop indicated that knowledge of sediment-water partitioning and
particle dynamics within an aquatic system are essential for accurate prediction of
assimilative capacity. Our experience with mass-balance models indicates that it
may take several years of research effort to obtain the necessary information to
develop an accurate model.  In a prospective scenario, this level of research would
not be possible. Therefore, several members of Group 2 felt that another
approach to this situation would be to develop "generic" mass balance models for
different types of ecosystems that could be used by risk managers to make
decisions. Generic models could be developed for:

       •      Small eutrophic and small oligotrophic lakes;
       •      Small embayments that connect to larger lake systems;
       •      Large embayments that connect to large lake systems;
       •      High flow and low flow rivers; and
       •      Marine or estuarine systems with high or low tidal flushing.

Risk managers could use these generic models to make initial decisions on the
siting of industrial facilities, and so  on.  More complex site-specific mass-balance
models could be used later, at the discretion of the risk manager or siting applicant
affected by the decision. In the latter case, it may be beneficial to adopt a
"polluter pays" policy, in which the applicant is responsible for paying for the
more complex modeling exercise.

       Overall Conclusions Concerning the Use of the TEF/TEO Approach  in the
Prospective Case Study and Associated Uncertainties (Janet Burris). The group
observed that uncertainty is less manageable in the prospective case study (risk
assessment) than in the retrospective case. In a retrospective case, actual
measurements of congeners in the environment or within the food chain could be
used to decrease uncertainties. In a prospective application, such measurements
are not possible.


       Uncertainties in the exposure profile appear to be equal to or greater than
those associated with stressor-response (effects) assessment.  A sensitivity analysis
would be beneficial to evaluate the various uncertainties in the risk estimates.
Suggested parameters for sensitivity analyses include: TEFs, 1^,, K^, P^^ , and
BMFs (location- and species-specific)

       The group observed that risk managers are attempting to  "titrate the
system" to permit the release of the last increment of chemical into the system,
based on its full assimilative capacity. Use of the TEF/TEQ approach reduces
uncertainty in the assessment, as there is not a better approach to be applied, and
use of a TEQ is more appropriate than the current TCDD  standard. However, the
uncertainty in the exposure profile may be result in a high  enough uncertainty that
the risks of loading the system beyond capacity are much greater than the manager
is willing to face.

       Measuring Uncertainty at the Population Effect Level (Mike Meyer).  At
present, little effort has been made to assess the impact of  "threshold levels of
effect" on target wildlife populations.  In most cases, the assumption is that early
life stage mortality measured in laboratory studies  (with species such as rats and
chickens) translates into population level effects in wildlife (such as otters and
bald eagles). This extrapolation is not supported by correlational data from the
field nor with laboratory studies using relevant wildlife, species (the exception
being feeding studies in mink).  This lack of knowledge produces a level of
uncertainty that dwarfs any presented by the TEF/TEQ approach. The current
ecological risk assessment process is seriously compromised by the inability of the
"best available science" to accurately predict effects (see Meyer, 1998, Env. Tox.
Chem. 17: 137-138). Until this uncertainty is addressed, the effectiveness of the
TEF approach to establish water quality guidance is suspect.

Retrospective Case Study

       The following sections summarize the discussions completed by Group 2
during its review of the retrospective case study. The summary is organized
.according to the primary issues discussed.

       Selection of TEF Values for Use in a TEO-Based Ecological Risk
Assessment (Mark Hahn). As part of a TEQ-based ecological risk assessment
(ERA) for PCBs, PCDDs, PCDFs, and other compounds that act via the Ah
receptor (AhR), toxic equivalency factors (TEFs) or relative potencies (RPs or
REPs) are used to convert congener-specific chemical residue data into 2,3,7,8-
TCDD toxic equivalents (TEQs). Although the TEQ approach is based on the
broad similarities in relative potencies that exist across different endpoints and
species (Safe, 1990), specific REP values can vary between species and across
endpoints within a species.  In an ideal ERA situation, congener-specific relative
potencies would be known for the species of concern (e.g., lake trout) and the
endpoint of concern (e.g., early-life-stage mortality).  Often, however, such data
are not available.  The use of REP values determined in a different species or for a
different endpoint, or use of a "consensus TEF," represents an important source of
uncertainty in a TEQ-based ecological  risk assessment.  This uncertainty is
separate from the uncertainty occurring as a result of species differences in
sensitivity to TCDD, which affects the choice of the "threshold" or action level to
which the calculated TEQ is compared.

       In the absence of species-specific REP values for the endpoint of concern,
a decision must be made as to which REP or TEF values provide the most accurate
measure of relative potency for use in calculating TEQs from congener-specific
residue data.  In essence, the decision involves choosing between the uncertainty
introduced by species differences in relative potencies (for the same endpoint) and
endpoint-dependent differences in relative potencies (in the same species). In
some cases, both types of uncertainty may be present. Common  sense suggests
that one should select the REP or TEF value that represents the best (i.e., most

accurate) information available.  However, since the uncertainty or "potential
error" inherent in a given REP/TEF choice is not always known (i.e., quantifiable),
the choice is often not clear.

       The approach described in Figure E-1 provides a framework for thinking
about the different kinds of REP or TEF values that may be available, and the
types of uncertainty inherent to  each. Using this matrix, selection of a REP or
TEF value is based on a hierarchical approach involving use of the best available
information, relative to the ideal choice—a species-specific REP for the endpoint
of concern.
                 Framework for choosing relative potency values for use in
        Ecological Risk Assessment for fish, birds, and mammalian wildlife.

la (endpoint of
1 b (other in vivo toxic
Tier 2
Tier 3
(in vtrrn CYPU)
Tier 4
Some species

Related species
(e.g. same genus
or family)

("Class-specific") •

Figure E-l.

       In the first column, four tiers reflecting and prioritizing the various in vivo
and in vitro endpoints used to determine REP values are listed. These categories
are based on the tiered approach used by WHO in deriving TEFs for fish and birds
(van den Berg et al., 1997). The first tier has been subdivided to differentiate in
vivo data for the endpoint of concern (Tier la) from other in vivo toxic endpoints
(Tier Ib). As with the WHO TEF approach, the highest priority is given to REPs
determined for the in vivo endpoint of greatest concern. Lower priorities are
assigned to REP values determined using endpoints more distantly related to the
assessment endpoint.

       The top row in the matrix indicates the phylogenetic relatedness of the
species of concern to the species in which REPs were determined.  It is divided
into  three levels, reflecting different degrees of uncertainty. If REPs are available
for the species of concern, there is interspecies extrapolation and so no uncertainty
associated with species extrapolation (although there could.be differences between
populations within a species).  If REP data are available for a closely related
species—a species within the same genus or family, for example—uncertainty is
higher due to potential species differences, but not as high as when REP data are
from a more distantly related species within the same class or when "consensus"
TEF values (such as the WHO TEFs) are used.

       The matrix might be used to consider and select among the types of REP
data available, by comparing the relative position of each set of REP data to the
ideal. An example is provided later in this discussion.

       The rationale behind this hierarchical approach is a mechanistic
understanding of AhR-mediated toxicity as well as empirical data that support
such the extrapolation of relative potency data across endpoints and/or species.
There is abundant evidence that most effects (endpoints) of dioxin-like
compounds, whether biochemical effects such as induction of CYP1A1 or toxic
effects such as ELS mortality, occur through the same initial  step—binding to the
AhR. The structure-activity relationships are similar across various endpoints,

including receptor binding, CYP1A induction, and various forms of toxicity (Safe,
1987; Safe, 1990).  The basic AhR-dependent mechanism of toxicity is the same
in most vertebrate species.  Most vertebrate taxa express an AhR (Poland and
Glover, 1987; Hahn etal, 1994; Hahn etal.,  1997) and are sensitive to dioxin
toxicity (Poland and ICnutson, 1982; Cook et al, 1991), although there may be
exceptions  (Jung and Walker, 1997). Despite the commonality of the basic
mechanism, however, there may be species- or endpoint-dependent variation in
specific details of the mechanism that result in different REPs.

       The basis of the four-tiered approach used to derive the "class specific"
WHO TEFs for fish and birds has been described (van den Berg  et al., 1997).
This approach involves weighting REP values based on the endpdint for which
they were derived, with preference to REPs determined for in vivo toxicity in
developing embryos.

       The basis for the phylogenetic approach reflected in the top row of the
matrix in Figure E-l is both theoretical and empirical. It assumes that two species
that are more closely related phylogenetically will have REP values (determined
for the same endpoint) that are similar or identical. This approach is supported by
data such as that showing that the REPs for CB-126 to produce ELS mortality in
lake trout and rainbow trout are similar (Zabel et al,  1995).  However, it is clear
that a more systematic effort to test this assumption will be needed.  Moreover,
although it is expected that closely related species will in general exhibit similar
REPs, exceptions to this assumption for certain species and/or congeners may be
revealed as additional data are collected.

       As stated earlier, it is important to keep in mind that the issue of species-
or endpoint-specific differences in REP values is separate from that of species
differences in sensitivity to TCDD.  In fact, there may be little or no relationship
between the two issues.  Two species that differ widely in their sensitivity to
TCDD can have similar REP values for most congeners. For example, chickens are
119-fold more sensitive than Pekin ducks to in vitro effects of TCDD, yet for

TCDF and PCB congeners 126 and 81 the REPs differ less than 5-fold between

these species (Kennedy et al, 1996).

       The matrix in Figure E-1 is intended to provide a framework for thought

and discussion concerning the selection of REPs for ecological risk assessments.

There are a number of practical questions that arise when considering this

approach;                                                           ,
  Often, REP data sets are incomplete.  Is it appropriate to draw REPs from
  multiple data sets to calculate TEQs for a given species? For example, in
  performing a risk assessment for lake trout, the only 'Tier 1 a" REP value that
  exists for lake trout is for CB-126.  For other congeners, REPs exist only for
  rainbow trout or other fish species. A "best available information" approach
  would lead one to choose the lake trout REP for CB-126, and rainbow trout
  REPs for the other congeners.

  The phylogenetic approach assumes that closely related species will exhibit
  similar REP values. But how close is "close"? Can we expect species within the
  same family to show greater similarities in REPs than occur between families, or
  must the species be within the same genus before such similarities are evident?
  Again, more data are needed to resolve this question.

  One of the most difficult questions concerns choosing between uncertainties
  based on species "differences versus endpoint differences, in the absence of data
  that would allow one to quantify the uncertainty in each. For example, suppose
  a risk assessor is performing an assessment for Caspian terns, using  measured,
  congener-specific concentrations of PCBs, PCDDs, and PCDFs in tern eggs.
  There  are no data on REPs for ELS mortality in Caspian terns, but  let us
  suppose that there are REP values (A) for in vitro CYP1A induction in Caspian
  terns, and (B) for in vivo ELS mortality in domestic chickens (the latter used to
  establish the WHO "consensus TEFs").  Perhaps there are also data for in vivo
  CYP1A induction in embryos of common terns, a closely related species (C).
  Figure E-2 illustrates the positions these three types of data would have in the
  matrix. Which of these three sets of REP data would provide the most accurate
  estimate of TEQs in Caspian tems? One option when confronted with such a
  decision might be to perform the TEQ calculations with each set of REPs; a
  comparison of the  resulting TEQ values might provide a measure of the
  uncertainty in selecting any one of the REP sets.

                Matrix showing position of three choices of REP values
                     for the scenario described above.

Tier 1
la (cndpoinlof
1 b (other in v/w
toxic cndpoint)
Tier 2
(In vivo CYP1 A)
Tier 3
(In vilrv CYP1A)
Tier 4
Same species


Related species
(e.g. same genus
or family)



Figure E-2..
       An example related to the last of these practical questions can befound in the
retrospective scenario discussed at the workshop.  In this case, the avian species of
concern is the Caspian tern. Figure E-3 shows TEQ values and STEQ determined
using two different sets of REPs. The first set of values are based on the WHO TEFs,
which are derived largely from chicken embryo data (van den Berg et al., 1997).  The
second set of values are  based on REPs from in vitro CYP1A induction in embryo
hepatocytes from common  terns,Sterna hirundo (Lorenzen et al, 1997), a species
closely related to the Caspian tern (Sterna caspia). There is a substantial difference
in the STEQ calculated using each set 'of values; the difference is due largely to the
3-and 17-fold lower relative potencies for CB-126 and CB-77 in common terns as
compared to the WHO TEFs.

           TEQ values and 2TEQ, determined using two different sets of RHPs.
                for Caspian terns in the retrospective scenario.

STEQ for nil
congeners (including
those not shown)

Caspian Tern
Common Tern

Caspian Tcm
 1 WHO TEFs (van den Berg et al., 1997).
 - REP values for in vitro CYP1A induction determined for common tern (LorenTen et
 al., 1997).
Figure E-3.

       The WHO TEFs, based largely on chicken embryo mortality, are thought
to be preferable because the endpoint used is more relevant to the effect of
concern. However, the differences between WHO TEFs and common tern REPs
could indicate some fundamental difference between terns and chickens in the
relative potencies of these congeners. The comparison is useful in providing an
indication of both the magnitude and source of the uncertainty (in this case, two
PCB congeners). Thus, this type of analysis contributes to the risk assessment
itself as well as identifying additional data that might help to reduce the

       When confronted with a lack of REP data for the species and endpoint of
concern, alternative REP values must be chosen. This choice involves the
introduction of uncertainty based on species differences and/or endpoint

differences in relative potencies. There is currently insufficient data to determine

which type of uncertainty is greater, and thus to guide the selection of particular

values. A best available information approach is recommended; this may involve

use of multiple REP values and sensitivity analysis of the resulting TEQs.

       Selection of Appropriate TEF Values for Species of Interest  in the Case
Study (Janet Burris). The group recommended construction of a hierarchy for
selection of TEFs for use in the retrospective risk assessment. The group could not
complete the hierarchy in consideration of the time and the absence of data and
derivation procedures for the TEFs in the WHO report. The group recommends
expressing consensus TEF values as a range instead of point estimates, since risk
management decisions are often not point estimates. Use of a range could provide
the manager with and understanding of the uncertainty and confidence in the

       In its review of the retrospective case study, the group selected TEFs for the

species of concern as follows:
  Lake trout: The group recommended using the rainbow trout REPs for all
  congeners except PCB 126, for which the lake trout REP was selected. Lake
  trout is the species of concern, and it was decided that a species-specific value
  would result in less uncertainty than a REPs for another species.  Both REPs
  (lake trout = 0.003 and rainbow trout = 0.005) are for early life stage
  mortality. The group agreed that uncertainty is introduced into the assessment
  by extrapolation from rainbow trout to other non-salmonid species (e.g.,
  largemouth bass).  This uncertainty could not be quantified, since it represented
  a lack of knowledge about relative REPs  between fish species.

  Caspian tern: The group recommended using the WHO TEF of 0.1 (embryo
  mortality) and a common tern in vitro EROD REP of 0.03.  The common tern
  value represents data for a more closely related species (compared to the
  chicken data used for the WHO TEFs).  However, the common tern REP is
  based on an in vitro endpoint, which is less useful. In this case, the group
  decided to run the analysis with both REPs to provide a sense of uncertainty by
  giving  a range of reasonable risk estimates. This would be one way of avoiding
  the use of a single point estimate in the risk assessment.

  Otter:  The group recommended using the TEFs in the WHO report, as values
  could not be identified for the otter or another closely related species. The
  group could not fully evaluate the uncertainties in the mammalian wildlife
  TEFs, since the derivation of these values is not fully described in the WHO
  report. The group observed that the values seem consistent with mink exposure
  studies of AhR agonist mixtures.

       The group would have liked to use the same TEF selection hierarchy for
mammals as was recommended for birds and fish. However, no description of the
derivation procedures for the mammalian TEFs or the underlying data is included
in the present WHO report.  Questions raised by the group included:

•  What endpoints were used?  Were in vivo endpoints used?
•  What species-specific data was available?
•  Can cellular effects, tumor promotion, and other endpoints be extrapolated to
   reproductive effects?
•  Were the endpoints used in the derivation applicable to reproductive and
   population level effects?
•  How were consensus values selected? Were these the most conservative values?
•  What rounding procedures were used?
The group agreed that the risk assessor would need this information to document
the values and assumptions for the risk assessment, to examine the uncertainties,
and for "transparency" requirements.
       Exposure Issues (Janet Burris). The group concluded that the exposure
assessment is driven by the analytical measurements used to determine
concentrations of congeners in sediment and tissues.  Several members of the
group observed that if sediment remedial goals were needed, these could be easily
identified based on the linear relationship between sediment concentration and
receptor tissue concentrations, as depicted in the equations in the retrospective
case study. In other words, to reduce tissue concentrations in the organisms of
concern, sediment concentrations would need to be decreased proportionately.

       The group concluded that use of the TEF/TEQ approach in the fate and
transport modeling is not different from traditional chemical-specific methods.
Use of a congener-specific approach in the retrospective case study does not create
a new problem for the exposure assessment.  It could, however, change the
approaches engineers would use in developing remediation plans.

       Analytical Considerations (Patrick O'Keefe).  In the retrospective scenario,
the analytical problems are not as difficult as those associated with the
prospective study. In the first place, data on concentrations of chemicals in the
water are not required to carry out the risk assessment. Second, the TEQ values
for the avian and mammalian species of concern—the Caspian tem and river otter,
respectively—are determined by relatively high concentrations of certain PCB
congeners, primarily PCB 126. Using the detection limit of 1 pg/g discussed above,
regulatory agencies should be able to assess the effects of an order of magnitude
reduction in sediment concentrations on tissue residues in the species of concern.

       Sampling programs need to be carefully designed in order to answer the
questions posed in a risk assessment.  For instance, fish samples should be
collected in the same vicinity as sediment samples for calculation of BASFs, and
sediment samples should be collected along an appropriate grid to monitor
chemical input from an effluent, as in the case of the prospective scenario.  In
addition to analytical precision, sample heterogeneity will play a major role in
determining the number of samples required to obtain data with an acceptable
variance. For statistical purposes, it may be necessary to carry out a preliminary
sampling program to obtain information on the variability of the chemical
concentrations with respect to sampling location. This is especially true in the case
of sediments where organic carbon concentrations can have a major influence on
residue concentrations. For tissue samples, other variables such as sex and age are
important factors in determining contaminant concentrations.

       Food Chain Modeling (Janet Burris).  The group concluded that a full food
chain model was not necessary. Such a model would be difficult to construct due
to lack of equilibrium in system, heterogeneity, and detection limit issues. It
would be especially difficult to build a model from sediment to the water interface
to fish.  A partial model would, however, be useful. Under assumption of steady-
state could use the linear relationship between sediment and biota levels to assess
various sediment remediation options without having to deal with lower trophic
levels. The partial model would use site-specific BSAF/BMFs. Such a partial
model would allow the risk manager to  examine reductions of chemicals in target

model would allow the risk manager to examine reductions of chemicals in target
species under given remediation scenarios, but would be difficult to use to predict
chemical movement over time.

       If not all of the exposure data were available as provided in the case study,
then the group recommended obtaining sediment samples in a transect of
depositional zones to get a sense of gradient, obtaining measurements of
congeners in prey species, and relating these to the dietary composition and forage
habits of the predators.

       Risk Characterization (Janet Burris). The prospective assessment is set up
to provide a point estimate of exposures and risks that can be identified as a
hazard quotient for an individual organism. The real question is how or whether
the adverse effect to individuals is reflected in the population.

       Risks are. not identified for the lake trout. The risks for the Caspian tern
are summarized below and were identified by the group as being "on  the edge."
                            Threshold      Exposure       Hazard Quotient
    (WHOTEF)               100pg/g       426 pg/g                4
    Common tern             100 pg/g       185 pg/g                2
    Total PCBs                5 jig/g         4.5 /ig/g               < 1
       The group discussed the likelihood of population-level effects.  Based on
the hazard quotients for embryo mortality, the group did not expect population
effects, but acknowledged that not all possible endpoints were assessed. The
potential effect of species-specific TEFs is noted here. The common tern exposure
is 2 times the threshold. Exposure based on the WHO TEF is, however, 4 times
higher than the threshold. Basing the assessment on TCDD or any other single
compound would in most cases underestimate the potential risk in comparison to
the TEF/TEQ approach.

       The group also noted that Caspian terns are not year-round residents and
could be getting exposures elsewhere. The group advised that further data would
be required to evaluate the origin of exposure including possible reference samples
and a weight-of-evidence evaluation.

       Risks for the otter were also identified as being "on the edge:"

                             Threshold      Exposure       Hazard Quotient
    TEF                      60pg/g         144pg/g              ,2
    TCDD                   60pg/g         l-4pg/g               <1
    Total PCBs               2.0/ig/g        1-0/ig/g               <1

       Action Decision (Janet Burris). When asked what they would do next, the
group concurred they would leave the site alone. They recommended monitoring
trends in TEQs with time to illustrate declining risk probabilities to the
population. No further loading of the system should be permitted. The
monitoring should include further diagnostic studies to better characterize risks,
including expression of risks in the context of population-level effects.  Group
members felt that the risk characterization should also include a description of
what the system may look like in 10 years, an estimate of effects of a 100-year
flood and redistribution of contaminated sediments, and  an estimate of habitat
destruction that could result in more risks. In order to understand if there is a
population level effect, we would need to understand the level of decreased
reproduction associated with a meaningful reduction in the population.

       When the group voted on action versus  no action  in the retrospective case
study, two  members voted for action and eight  for no .action.

       If the risk manager chose to proceed with remediation, the otter would be
the primary species of concern. The otter is related to the mink, but is known to
have greater sensitivity to AhR agonists. The group  would provide scenarios to the
risk manager and discuss  population level effects in  the context of adverse effects
associated with the remedial alternatives.


Cook, P.M., Kuehl, D.W., Walker, MIC, and Peterson, RE. (1991) Bioaccumulatioi
and toxicity of TCDD and related compounds in aquatic ecosystems, mBanbury
Report 35: Biological Basis for Risk Assessment ofDioxins and Related Compounds, Gallo,
M.A., Scheuplein, R.J., andHeijden, KAV.d., Editor., Cold Spring Harbor Press: p.

DeVito, MJ, Diliberto, JJ, Ross, DG, Menache, MG, Bimbaum, LS (1997).  Dose-
response relationships for polyhalogenated dioxins and dibenzfurans following
subchronic treatment in mice. I. CYP1A1 and CYP1A2 enzyme activity in liver,
lung and skin. Toxicol. Appl. Pharmacol. 147: 267-280.

Hahn, M.E., Karchner,  S.I., Shapiro, M.A.,  and Perera, S.A. (1997)  Molecular
evolution of two vertebrate aryl hydrocarbon (dioxin) receptors (AHRl and AHR2)
and the PAS family. Proc. Natl. Acad. Sci. U.SA. 94: 13743-13748.
Hahn, M.E., Poland, A, Glover, E., and Stegeman, J.J.  (194) Photoaffmity labeling
of the Ah receptor: Phylogenetic survey of diverse vertebrate and invertebrate species
Arch. Biochem. Biophys. 310:218-228.

Jung, RE. and Walker, M.K (1997) Effects of 2,3,7,8-Tetrachlorodibenzo-p-dioxin
(TCDD) on development of anuranamphibians. Environ. Toxicol. Chem.  16:230-240.

Kennedy,  S.W., Lorenzen, A., Jones, S.P., Hahn, M.E., and Stegeman, J.J.  (1996)
Cytochrome P4501A induction in avian hepatocyte cultures: a promising approach
for predicting the sensitivity of avian species to toxic effects of halogenated aromatc
hydrocarbons. Toxicol. Appl. Pharmacol  141: 214-230.

Lorenzen, A, Shutt, L., and Kennedy, S.W. (1997) Sensitivity of common tern
(Sterna hirundo)  embryo hepatocyte cultures to CYP1A induction and porphyrin
accumulation by halogenated aromatic hydrocarbons and ©mmon tern egg extracts.
Arch. Environ. Contam.  Toxicol. 32: 126-134.

Poland, A. and Glover, E. (1987) Variation in the molecular mass ofthe Ah receptor
among vertebrate species and strains of rats. Biochem.  Biophys. Res. Commun.  146:

Poland, A and Knutson, J.C. (1982) 2,3,7,8-Tetrachlorodibenzo-p-dioxin and relatd
halogenated aromatic hydrocarbons: examination of the mechanism of toxicityAnnu.
Rev. Pharmacol. Toxicol. 22: 517-554.

Safe, S. (1987) Determination of 2,3,7,8-TCDD toxic equivalent factors (TEFs):
support for the use of in the vitro AHH induction assay. Chemosphere  16: 791-802.

Safe, S. (1990) Polychlorinated biphenyls (PCBs),  dibenzo-p-dioxins  (PCDDs),
dibenzofurans (PCDFs), and related compounds: environmental and mechanistic
considerations which support the development of toxic equivalencyfactors (TEFs).
CRC Crit. Rev. Toxicol 21:51-88.


van den Berg, M. and et al. (1997) Draft Report of Meeting on the derivation of
Toxic Equivalency Factors (TEFs) for PCBs, PCDDs, PCDFs and other dioxin-like
compounds for humans and wildlife. World Health Qganization (WHO), June 15-
18, Stockholm, Sweden (manuscript).

Zabel, E.W.,  Cook, P.M., and Peterson,  R.E.  (1995) Potency of 3,3',4,4',5-
pentachlorobiphenyl (PCB  126),  alone and in  combination  with  2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD), to produce lake troutearly life stage mortality.
Environ. Toxicol. Chem.  14:2175-2179.

                    Facilitator:  Charles Menzie

Prospective Case Study

       The group initiated their discussions by outlining and reviewing the
features of the case study.  This provided a basis for understanding the approach
that would be taken. Key aspects of the analysis included identifying the
organisms of concern, the pathways of exposure (conceptual model), and the
target concentrations in water that are judged "acceptable" for the various

       Our subsequent discussions are organized around two categories. The first
involves points that were made relative to the case study.  The second  involves a
discussion of how to identify and track uncertainties in an ecological risk
assessment process that includes application of the TEF/TEQ method.

       Part 1:  Points Drawn from the TMDL Case Study

       With respect to the case study, the group discussed various  issues. Each of
these issues is described further in the paragraphs that follow.

       Rounding and Significant Digits.  Some participants noted that care should
be taken to not overstate (via calculation) the number of significant digits.  Some
numbers presented in the case study appear to be presented at a level of precision
that is  unlikely to have been achieved.  By presenting the numbers to two or more
significant numbers, a false sense of precision is given.

       Application of REP versus TEF Values.  The participants discussed how
values  should be selected for use in the TEF/TEQ methodology. The group
concluded that, where available, REP values should be selected over generic TEF

values. The group believed that uncertainties were reduced if taxa-specific
unrounded values were utilized in lieu .of the "order of magnitude" values
presented in the WHO TEF report.  A hierarchical approach was suggested, within
which the best and most appropriate values were selected first. The REP values
are not rounded and have their own level of significance.  When asked if the group
would usually elect to use REP values even when they needed to extrapolate to
another species (e.g., trout to largemouth bass), the group favored use of REP over
TEF. values. In part, this view reflected concerns over the rounding done when the
WHO TEF values were developed. A few members of the group did not agree
with this, because of the variability that might exist around REP values within a
class of animals. When asked if they felt that the results would be compromised if
they were required to use TEFs, the participants said that they would not, but that
there would be additional uncertainty in the estimate.

      Uncertainty in TCDD Toxicity Values and Target Concentrations.  The
group acknowledged that there was uncertainty in the target numbers used in the
analysis.  Underlying this uncertainty was the toxicity data as well as the models
used to derive the concentration of TCDD in water.

      Selection of Chemicals and Percentage Allocation. The group noted that
for this exercise, we did  not have to pick the chemicals. However, we should
recognize that atmospheric PCBs also add to the Waste Load Allocation.  For this
exercise,  it had already been decided that the site would be given 25% of total
allocated load.  This  allows us to focus on this smaller set of chemicals.

      Uncertainty Analysis. There was considerable discussion concerning
uncertainty analysis. A more detailed account of this discussion is provided later
in this workgroup report.  However, workgroup participants discussed the need to
track and document  uncertainties in the ecological risk assessment process. At a
minimum, this includes a narrative discussion. It is also possible to talk about the
sources and potential magnitudes  of uncertainties. One member of the workgroup
noted that we could  make a semi-quantitative attempt to estimate the  magnitude;
for example, we can say REPs have uncertainty of about 2 or 5.

       Part 2: Identifying and Tracking Uncertainties

       The workgroup discussed procedures that might be used to identify and to
take into account the various uncertainties in the assessment. We worked through
a process that is based on the assessment of multiple stressors. A key aspect of
this approach is writing down the criteria used to judge the uncertainties
associated with each aspect of the analysis. This information is also useful for
helping clarify the issues of concern in discussions with managers and with

       After much discussion, the group agreed upon an ordinal ranking system
for uncertainty that reflected our level of confidence regarding the relative
uncertainty in the information.  We chose values of 1 (most confidence) to 4
(least confidence). We selected a range of 1 to 4 in part because this is consistent
with the number of categories in the WHO TEF document,

       A quantitative aspect of uncertainty which the group did not include but
acknowledged as important was the magnitude of the error around the values.
Such information would be important for sensitivity analyses. Our group did not
address this because of time and data constraints. However, we  believe that this is
important to consider.

       The group then identified the areas of uncertainty within the ecological
risk assessment process.  These do not represent all  the possible areas, and some
areas could have been broken  into smaller components. Our purpose here was to
illustrate the  concept rather than arrive at a definitive approach. The areas that
were evaluated include:

       •     uncertainty criteria for TEF values;
       •     uncertainty in comparing TEFs to target water levels;
       •     uncertainty factors for BAF values;

       •      uncertainty factors associated with species extrapolations; and
       •      uncertainty factors associated with the exposure model.
Each of these areas is described in more detail below.

       Uncertainty Criteria for TEF Values. Our group had two separate
discussions concerning these criteria. The first was on our first day; on the second
day we revisited the issue and made some modifications. The criteria we came up
with on our first attempt were:

       •      Level 1:  egg injection with mortality endpoint;
       •      Level 2:  whole organism with other endpoints;
       •      Level 3:  in vitro studies (e.g., enzyme induction); and
       •      Level 4:  QSARs  (from in vitro data).
Based on this set of criteria, we initially assigned the bull trout REP an uncertainty
level of 1, the value for eagle an uncertainty level of 2, and the value for river otter
an uncertainty level of 1. These assignments reflect the levels of confidence given
in the WHO report.  When we revisited these assigriments, it was noted that some
participants had more certainty in the fish values than in those for mammals and
birds and that the initial assignments did not capture this. The following ranking
procedure was subsequently proposed:

       •      Level 1:  REP with population-relevant endpoint;
       •      Level 2:  TEF from in vivo study with toxicological endpoint;
       •      Level 3:  TEF based on biochemical response;  and
       •      Level 4:  TEF based on QSAR or enzyme induction.
       Individuals familiar with the derivation of the TEF values noted that there
was more uncertainty associated with the bird values than with the mammal
values  (which are based on a rich body of data). These individuals had greatest
confidence in the fish values.  Based on our reassessment of uncertainty levels, the


uncertainty for the bull trout remained at 1, but uncertainty for the bald eagle
increased from 2 to 3 or 4, and for the river otter from 1 to 2 or 3.

       Uncertainty in Comparing TEFs to Target Water Levels. The group
assigned uncertainty factors based on taxonomic extrapolation from the available
toxicity databases.  It was noted that there were other sources of uncertainty in
these values, including the models used to derive the water concentrations in the
criteria documents. Considering only the taxonomic extrapolations, the following
criteria were developed:

       •     Level 1: same species;
       •     Level 2: same genera;
       •     Level 3: same family; and
       •     Level 4: same class.
Because bull trout are in the same genus as the reference species, comparison of
the TEF with the water quality criterion was assigned an uncertainty factor of 2.
Because bald eagles are raptors and the bird standard is based on galliforms, this
comparison was assigned an uncertainty of 4.  Finally, since river otter and mink
are in the same family, the comparison for mammals was given an uncertainty
ranking of 3.

       Uncertainty Factors for BAF Values.  The following criteria were proposed:

       •     Level 1: site-specific measurement;
       •     Level 2: lab validated with field;
       •     Level 3: field data with no lab, or lab with no field  corroboration;
       •     Level 4: BAF based on !<„„ (prediction).
The group agreed that the BAF for 2,3,7,8-TCDD would be assigned an
uncertainty value of 2. We also decided that the value should be 2 for the relative
BAFs of all congeners.


       Uncertainty Factors Associated with Species Extrapolations. The group
noted some of the differences that can occur among taxa. We decided to use the
same taxonomic extrapolation uncertainty criteria:

       •      Level 1: same species;
       •      Level 2: same genera;
       •      Level 3: same family; and
       •      Level 4: same class.
The assigned uncertainty factors were both congener- and taxa-specific.  For the
eagle, the three dioxin congeners (Congeners 1-3) were assigned an uncertainty of
1, the first furan (Congener 4) was assigned an uncertainty of 4, the next two
furans (Congeners 5 and 6) were assigned an uncertainty of 3, and Hx (Congener
7) was assigned an uncertainty of 2. For the bull trout, all congeners were
assigned an uncertainty of 1. For the river otter, dioxins were assigned an
uncertainty of 1 and furans were assigned an uncertainty of 2, based on  metabolic

       Uncertainty Factors Associated with the Exposure Model.  We noted that
this was a simplified TMDL model that incorporated 1^, and Henry's Law
constants, and that there are chemical- and environment-specific factors added
into these models. We noted that the model assumes equilibrium or at least
steady state; in the real world, however, non-equilibrium conditions are likely to
be present.  Further, we noted that there were uncertainties associated with the
various physicochemical parameters used to predict the behavior of the chemicals.

       For fish, we assigned the exposure model an uncertainty of 2 or 3, while for
eagle and otter we assigned an uncertainty of 3 or 4. The higher uncertainty
associated with the eagle and otter models reflect the anticipated increase in
uncertainties associated with relating exposures at these higher trophic levels to
sources within the lake. The workgroup noted that any exposure  model is
expected to have uncertainty associated with it, given all the simplifying
assumptions that need to be made.

       Representation of Uncertainties

       The different sources and "levels" of uncertainty can be displayed in a table
and are amenable to mathematical representation and analysis. As an example,
the results of our analysis are illustrated in Figure E-4.  The table indicates which
areas of uncertainty are potentially most important.  It also provides an indication
of the levels of uncertainty that accompany results of the risk analysis for specific
organisms of interest.
           Relative Uncertainties in the Ecological Risk Assessment Including Us of TEF Values
Ranks for uncertainly

Bull trout
Bald Eagle
Rlvtr Otter





Species Sens./Ex1rapola

Exposure Mod
2! 2
2| 1
3 3

31 2




Threshold concentration



Criteria are described in me text This approach and these values are presnted tor illustration onryj

Species specific
Congener specific





30|Bull Trout



Bald Eagle

River Otter

Figure E-4.

Retrospective Case Study

       TEF/TEQ Issues Related to Measurements

       The retrospective case study is based on measurements of individual
PCDDs, PCDFs, and PCBs in tissues and environmental media. The group
considered several issues related to using such information in ecological risk
assessments that rely, in part/on a TEF/TEQ approach.

       Accuracy and Precision of the Measurements.  Because the TEF/TEQ
approach relies on information related to individual compounds, we discussed the
ability of available methods to provide accurate and precise results.  Individuals
familiar with the methodologies used to identify and quantify individual PCDDs,
PCDFs, and PCBs in tissues and environmental media believed that the accuracy
and precision of the measurements was good. Reanalysis of samples gives similar
results. There are certified reference standards. For the concentrations provided
in the case study, one individual noted that the measurement error was probably
about 30%. Another reported that he observed coefficients of variation on the
order of 100%. Measurement error tends to increase with decreasing
concentration.  Overall, the group concluded that analytical error was not a large
source of uncertainty in the overall analysis.

       Variability in data can arise from biological variability.  In particular, it was
noted that different size or sex of fish would be a source of variability. Further,
analytical variability in lipid measurements could be  a source of variability in data
that are reported on a per lipid basis. Variability in analysis of environmental
media can also result from variability in the physical  and chemical characteristics
of these media (e.g., grain size and organic carbon content of sediments.)

       Detection levels.  The group discussed detection levels with respect to the
use of concentration data for risk assessment purposes.  The group concluded that
available analytical methods could achieve detection  levels low enough to support

the TEF/TEQ approach as it is applied to ecological risk assessment. However, it
was noted that detection levels should be stated as part of the Data Quality
Objectives and that the laboratory should be informed of the detection levels they
will need to meet. When considering the detection levels that need to be met, one
participant noted that it was important to be aware of the dose response curve.  It
is also helpful to consider which compounds contribute most to the toxicity of the
mixture and to be sure that detection levels are adequate to quantify ecologically
significant concentrations of these compounds.

       Design issues. The workgroup concluded that sampling design issues were
comparable between the TEF/TEQ method and other methods used to evaluate
risks associated with PCDDs, PCDFs, and PCBs.  The case study included sample
sizes of 12. Participants noted that they would want to know values for eggs from
12 different terns or 2 nests.  Such information would be important to
understanding uncertainty.

       Costs. Analytical costs associated with congener analyses are higher "than
for  total PCBs or analysis of an individual compound.

       Other Effects not Captured in TEF/TEQ. Participants noted that some
PCB compounds could affect the species of concern via toxic mechanisms other
than binding to the AhR receptor. Care must be taken to identify effects that may
be important during planning stages of the analysis.

       Working with Partial Data Sets. The TEF/TEQ methodology involves an
assessment of PCDDs, PCDFs, and PCBs.  However, for some situations, only one
of these groups may be important. The workgroup concluded that partial data
sets for one of these three groups would be adequate for evaluation if available
information indicated that this was the only group of importance at the site or for
the application.  If available information indicates that background concentrations
of other groups contribute significantly to the TEQ estimate, then those groups
would have to be included in  the analysis because the TEF/TEQ approach involves

comparisons to TEQ benchmarks or dose-response curves.  In such cases, it would
be inappropriate to consider the effects of any one group alone.

       Utilizing Surrogate Methods in Concert with the TEF/TEQ Methodology

       The workgroup concluded that there were opportunities to complement the
TEF/TEQ methodology with surrogate analytical approaches. It was noted that
once you have this information, and have calibration between individual
congeners and total PCBs, you could use total PCBs as well. This would involve
validating and calibrating as you go. Surrogate methods could be employed during
investigations or in helping guide remedial measures. Workgroup participants
noted that surrogate approaches work for the Great Lakes.  It is not known how
well these methods might work in other systems.

       Comparison of TEQ vs. Traditional Total PCB Approach

       The workgroup concluded that the traditional PCB approach would have
missed important aspects of the problem. The traditional total PCB method
would have underestimated risks as compared to the TEF/TEQ method.

       Adequacy of Available. Information for Decisionmaking Purposes

       The workgroup identified several pieces of information that would be
desirable for supplementing the information already at hand. These are discussed

       Background Conditions. The case study did not provide information on
background conditions. Therefore, even though body burdens of PCBs could be
explained in terms of exposure to PCBs in the lake, there is the possibility that
PCBs in the lake and in the tissues is comparable to those found in other lake
systems.  This  could be evaluated by examining other lake systems or by
evaluating conditions upstream of the spill.  The group felt that a better

understanding of this issue would be needed before proceeding with a
recommendation concerning management options.

       Reasonableness of Association Among Concentrations in the Species of
Interest.  The workgroup concluded that it was important to examine the data
with respect to the underlying conceptual model and relationships among media
and receptors from other systems.  Body burdens may vary among clutches of
eggs. Typically, there are higher concentrations in the second clutch.

       System and Food Web Issues.  Workgroup participants concluded that
additional information on food web relationships would be valuable. For example,
it was noted that relationships within the case study were being inferred by
assuming simple food chains. However, food chains could be much more complex
and quite different from those assumed or inferred from the available data.

       Vertical Distribution of'Compounds in Sediment Cores. Information on
concentrations in sediment cores would provide insight into the history of
deposition, including pre-spill conditions. Such information could also be used to
judge the rate of recovery.

       Deriving "Acceptable" Target Levels for Environmental Media

       Mechanics of Back Calculating Target Levels.  The workgroup concluded
that this  would involve working the exposure equations backward.  This would
involve beginning with "acceptable" TEQ levels in ecological receptors and
deriving "acceptable" target levels in sediments or water.  The major challenge
here is that the TEF/TEQ methodology involves tracking a number of compounds.
This is primarily a logistical challenge. However, back-calculating will require
information on the environmental behavior of the individual compounds.
However, it may be possible to limit back-calculation  to those compounds that
contribute most to the TEQ levels in ecological receptors. Back-calculation would
involve applying appropriate mass-loading models as well as biological uptake

models.  These could be simple or complex.  One participant noted that the
analysis should extend beyond simple calculations of average concentrations.
Individual animals do not experience the average, but rather the overall

       Regression analyses. Apart from bioassay-based approaches (e.g., cell
lines), use can be made of regression relationships between individual congeners,
which can easily be measured with low cost GC-ECD techniques, such as PCB 153
                                             11        -7
and concentrations of toxic congeners (e.g., PCB 126) or total TEQ
concentrations.  In^several ecosystem studies, relationships have been observed for
PCB 153 and total TEQ concentrations in various fish species, otter, invertebrates,
and cormorant eggs covering several orders of magnitude in PCB TEQ
concentration.  As a rule of thumb, a 0.5 to  1.0 order of magnitude range of
uncertainty may be involved in extrapolations based on this relationship. As the
regression relationships may be species- or site-specific, a preliminary validation
may be required.  Members of the workgroup recommend further exploration of
this regression approach, using available data from monitoring studies, and further
assessment of the feasibility of this potentially cost-effective approach.

       Body burdens in some animals are size- or age-dependent. In the case of
otters, for example, a recommendation was made to sample young carcasses.

       Risk Management Options

       The workgroup concluded that the decision "to clean up or not clean up"
was one of several possibilities. The workgroup discussed several possible risk
management  options that could be explored using technical information.

       Evaluating the Future Potential and Time Course for "Recovery". This is
the "no action" or "limited action" option. Essentially, this option would involve
providing the risk manager with information concerning how the system may
change in the future. With respect to the TEF/TEQ approach, this will involve

understanding how concentrations of individual compounds in media and tissues
will change in the future. Processes that could be involved include burial in
sediments, degradation rates, metabolism, dissolution, and loss rates via
evaporation and advection from the system.

       Developing Additional Lines of Evidence.  Workgroup participants
discussed additional lines of evidence that could support decisionmaking.  These
included direct observations of effects on populations and bioassays. An example
of observations that could be made on birds is to look for scelma (Great Lakes
edema/mortality syndrome) in tern eggs.

       Identifying Alternative Remedial Strategies. Remedial options could vary
in type and magnitude. .The efficacy of these alternatives could be judged by
applying "what if" scenarios utilizing the TEF/TEQ methodology.

       Integrating Lines of Evidence from Different Levels of Ecological

       The workgroup discussed the strengths and limitations of different lines of
evidence that could be used to complement the TEF/TEQ approach. These
discussions underscored differences in perspective related to "bottom up"
approaches represented by applying the TEF/TEQ concentration-based
methodology and "top down" approaches represented by making direct
observations on populations.  Bioassay methods fall in between. The group
acknowledged that these different approaches had various strengths and
limitations. The group concluded that it would be useful to explore how these
different lines of evidence could be brought together to provide an overall
assessment. With respect to organisms and population biology, it would be
beneficial to foster exchanges between scientists working with the various
bioassays, conservation biologists (working with populations), and ecological risk
assessors.  It would be helpful, for example, to have these groups work on a model
for the first year of life for salmonids.

       For this case study, participants noted the importance of having
information for several trophic levels. In the present case, the fish populations do
not appear to be at risk.  However, species at even higher trophic levels—the tern
and the otter—exhibit levels that indicate potential risks.

            Appendix F

                                                                 National Institutes of Health
                                                                 National Institute of
                                                                 Environmental Health Sciences
                                                                 P. O. Box 12233
                                                                 Research Triangle Park, NC 27709


                 January 16, 1998

                 Angelique P.J.M. van Birgelen, Ph.D.

                 3 ,3 ',4,4'-Tetrachloroazobenzene, hexachlorobenzene, 1,2,3,4,6,7-
                 hexachloronaphthalene, 1,2,3,5,6,7-hexachloronaphthalene, and 1,2,3,4,5,6,7,-
                 heptachloronaphthalene as additional dioxin-like compounds for inclusion in TEF

                 Chair of the Workshop on the Application of 2,3,7,8-TCDD Toxic Equivalency Factors
                 to Fish and Wildlife

                 Drs. K. Abdo, J. Bucher, and G. Lucier
 Inclusion in TEF concept
 A dioxin-like compound is a compound that binds to the aryl  hydrocarbon receptor, results in
 dioxin-like effects, and bioaccumulates. These are the three factors for inclusion of dioxin-like
 chemicals  in the  TEF  scheme (Ahlborg et  al,  1992,  1994).  3,3',4,4'-Tetrachloroazobenzene
 (TCAB),  hexachlorobenzene (HCB), 1,2,3,4,6,7-hexachJoronaphthalene (PCN 66),  1,2,3,5,6,7-
 hexachloronaphthalene  (PCN  67),  and  1,2,3,4,5,6,7-heptachloronaphthalene  (PCN  73)  are
 compounds which bind  to the Ah-receptor, result in dioxin-like effects, and have been shown to.
 bioaccumulate and should therefore be included in the TEF concept.

 Sources of TCAB, HCB, and PCNs
 TCAB is present  as a  contaminant  of 3,4-dichloroaniline (DCA) and  its  herbicidal  derivatives
 Propanil, Linuron, and Diuron (Poland et al., 1976; Sunstrom et al, 1978; Bunce et al., 1979; Hill
 et al, 1981). In addition, environmental contamination by TCAB occurs from the degradation of
 chloranilide herbicides  (acylanilides, phenylcarbamates, and phenylureas)  in soil by peroxide-
 producing microorganisms (Bartha et al, 1968; Bartha and Pramer, 1969; Lay and Dnicki, 1974). It
 is also formed by the photolysis and biolysis of 3,4-dichloroaniline (Mansour et al, 1975; Miller et
 al, 1980).

 HCB was used as a fungicide for crops such as wheat, barley, oats, and rye to prevent growth of
 fungi. In the mid seventies the application of HCB as a fungicide was discontinued due to concerns
 about adverse health effects. In Tunisia however, HCB  was still used as a fungicide, seed-dressing,
 and scabicide in sheep in 1986 (IPCS, 1997).
 In industry, HCB has been used in the manufacture of pyrotechnics,  tracer bullets, and as a fluxing
 agent in the manufacture of aluminum. HCB has been used as a wood  preserving agent, a porosity
 control agent in the manufacture of graphite anodes, and as a peptizing agent in the production of
 nitroso and styrene rubber for tires (IPCS, 1 997).
HCB is generated as a  by-product in various chemical  processes  such as  thermal chlorination,
 oxychlorination, and pyrolysis operations in the manufacture of chlorinated solvents such as carbon

 tetrachloride, trichloroethylene, and tetrachloroethylene (PCS, 1997). HCB is a by-product during
 the  manufacture  of pesticides,  such  as  pentachloronitrobenzene,  chlorothalonil,  dacthal,
 pentachlorophenol, atrazine, simazine, propazine,  and maleic hydrazide (IPCS, 1997). In the
 herbicide Propanil it  has been found in concentrations up to 10-14% (IPCS, 1997). Furthermore,.
 HCB is released into the environment by waste incineration. The release of HCB from all municipal
 incinerators  in the U.S.  was estimated by  the EPA to be between 57 and 454 kg per year as
 documented  in 1986 (PCS, 1997).

 Polychlorinated naphthalene formulations  have  been used  in industry as dielectric fluids  in
 capacitors, transformers,  and cables.  The products containing technical PCNs are still in use or
 disposed  in  landfills. PCNs  are also  formed during  production  of technical mixtures  of
 chlorobiphenyls and can be found in  concentrations up to  1% in various polychlorinated biphenyl
 formulations (Falandysz et al., 1996).

 Binding to the Ah-receptor
 TCAB and HCB have an affinity for the Ah-receptor  5-  and  10,000-fold lower that  TCDD,
 respectively (Hahn et al, 1976; Poland et al., 1976; Schneider et al, 1995). Preliminary results
 from an Ah  receptor binding assay indicate a relatively high binding activity  of the hexa- and
 heptachlorinated naphthalenes (Hanberg et al, 1990).

 Dioxin-like effects
 TCAB exposure results in typical dioxin-like effects in rodents which include chloracne and dermal
 lesions, body weight  loss, thymic  atrophy,  hepatotoxicity, developmental toxicity, induction of
 cytochrome P4501A1, anemia, and an increase of porphyrins in chick embryo liver cell cultures-
 (Hsia et al, 1980, 1981, 1982; Hill et al, 1981; Mensink and Strik, 1982; Schrankel et al, 1982;
 Hsia and Kreamer, 1985; McMillan etal, 1990).

 HCB results  in dioxin-like effects, such as induction of hepatic CYP1A1 and CYP1A2 activities,
 hepatic  porphyrin  accumulation  and  excretion,  alterations  in thyroid hormone  levels  and
 metabolism,  alterations hi  retinoid levels, liver damage (hepatocellular enlargement,  bile duct
 proliferation,  necrosis), reduction in reproduction, splenomegaly, increase in mortality, neurological
 alterations  (such  as  tremors, paralysis,  weakness,  hyperexcitability), teratologic  effects,  and
 immunotoxic effects (PCS,  1997).  HCB is a carcinogen in rodents (PCS, 1997). HCB exposure
 also results in phenobarbital-like effects, such as induction of hepatic CYP2B activity (PCS, 1997).

 PCN 73, and a  mixture .of PCN 66  and  PCN 67 induced EROD  and AHH activities in a rat
 hepatoma H-4-H cell line (Hanberg etal, 1990,1991).


TCAB has a log octanol/water partition coefficient of 5.53 to 6.69 (US-EPA, 1985; Hashimoto et
al, 1994). The solubility in water is  calculated to be 1 ug/1  (US-EPA, 1985). In male Sprague
Dawley rats administered radiolabeled  TCAB by gavage, 66% of the dose was excreted in urine and

feces after 24 hours (Burant and Hsia, 1984).  The pattern indicated  a  biphasic elimination,
consisting of an early rapid phase with a half-life of 18 hours and a slow terminal phase with a half-
life greater than 20 days.
HCB has a log/octanol water partition coefficient of 5.5 (EPCS, 1997). The solubility in water has
been reported to range from 0.005 (mg/L at 25°C) to 0.035 ppm (PCS, 1997; Kenaga, 1980). The
(whole body) half-life of HCB in male Wistar rats has been reported to be 20 days (Yamaguchi et
al, 1986). In male Sprague Dawley rats and male white rabbits, the half-life was calculated to be 24
days and 32 days, respectively (Scheufler and Rozman, 1984). The major route of excretion was via
the feces in both rats and rabbits. In rhesus monkeys, the half-life has been estimated to be 2.5 to 3
years (Rozman et al., 1975).

PCN 66 and PCN 67 were selectively retained in the liver of rats exposed to Halowax 1014, a
commercial mixture of PCNs (Asplund et al, 1986, 1994;  Jacobsson et al., 1992). In marine
mammals such as harbour porpoise a BMP value greater than 1 was observed only for the pah- of
PCN 66/PCN 67  (Falandysz, 1997). PCN 66/PCN  67 and  PCN 73 have been found at high
concentrations in cod liver samples from Southern Norway (Schlabach et al., 1995).

Relative potency estimates for TCAB, HCB, and PCNs

Table 1. Relative potency estimates for 3,3',4,4'-tetrachloroazobenzene (TCAB).
Binding affinity to the Ah
receptor (nM)
EC50 for binding to the
mouse hepatic Ah receptor
ED50 (nmol/kg) for
induction of aryl
hydrocarbon hydroxylase in
chicken embryos
LD50 (ng/egg) in chicken
Cytochrome P4501A1
induction in the skin in a
90-day oral gavage study in
female B6C3F1 mice with


Relative potency
for TCAB
Poland etal., 1976.
Schneider et al., 1995.
Poland et al, 1976.
Higginbotham et al,
1968; Schrankel et al,
Hebert etal, 1993.

Table 2. Relative potency estimates for hexachlorobenzene (HCB).
Binding affinity to the Ah
receptor (nM)
EC50 for EROD induction in
chicken hepatocytes (nM)
EC50 for accumulation of
uroporphyrin in chicken
hepatocytes (nM)
Hepatic porphyrin
accumulation in female rats


Relative potency
for HCB
Habnetal., 1976.
Sinclair et al, 1997.
Sinclair et al, 1997.
Cantoni et al., 1981.
 Table 3. Relative potency estimates for polychlorinated naphthalenes (PCNs).
AHH activity in a rat
hepatoma H-4-n cell line
CYP1A1 activity in a rat
hepatoma H-4-II cell line
Relative potency
Relative potency
Hanberg et al.,
Hanberg et al.,
1990, 1991.
 Impact on TEQ
 Based on an annual production volume of 10 million pounds  of Propanil in the US  and the
 concentration of TCAB in Propanil ranging from 1,000 to 2,700 ug/g, this could lead to an annual
 release of 12,000 kg of TCAB into the environment due to Propanil alone (Sunstrom et al., 1978;
 Bunce et al., 1979; Hill et al, 1981; US-EPA, 1987 as cited in McMillan et al, 1991). With an
 annual production volume of 0.1 to 1 million pounds of DCA and a concentration of TCAB in DCA
 ranging from 9 to 8,600 ug/g, this could lead to a production of 3,900 kg of TCAB per year hi the
 US (Sunstrom et al, 1978; Bunce et al., 1979; Hill et al., 1981; US-EPA, 1985). Analyses of a rice
 plot treated with 6.7 kg Propanil/hectare indicated a TCAB concentration of 0.09 ppm (Kearney et
 al., 1970). Six of 99 soil samples from the rice-growing states of Arkansas, California, Louisiana,
 Mississippi, and Texas contained 0.01  to 0.05 ppm TCAB, whereas no residual concentration of
 Propanil was detected (Carey et al., 1972). Assuming TCAB is three orders of magnitude less
 potent than TCDD (to pick  a  number), this indicates that the  concentration of TCAB in the
 mentioned soil samples, calculated as TEQ could be as high as 90 ng TEQ/kg soil. For comparison,
 the mean level of dioxin-like compounds (PCDDs and PCDFs only) has been estimated to be 8 ng
 TEQ/kg soil (US-EPA, 1994). Using the same  calculation for the production of TCAB due to
 Propanil and DCA, this could lead to an annual release of 16 kg TEQ in the environment.

 HCB                                ,
Levels of HCB measured in bald eagle eggs from the British Columbia coast from 1990 to 1992
ranged from 0.012 to 0.025 mg/kg wet weight (Elliott  et al., 1996). Assuming HCB has a relative
potency of 0.0001 (to  pick a number), this could  be  as high as 25 ng TEQ/kg wet weight. For
comparison, the concentration of PCDDs, PCDFs, and PCBs (planar and mono-ortho substituted)
ranged from 120 to about 320 ng TEQ/kg in bald eagle eggs from the same areas (Elliott et al,

 The concentration of PCN 66/PCN 67 in cod liver samples from Southern Norway ranged from 927
 to 123,000 pg/g wet weight (Schlabach et al, 1995). Using a relative potency value of 0.003 this
 equals to 2.8 to 369 pg TEQ/g wet weight. In the same study, the TEQ based on PCDDs, PCDFs,
 and non-ortho  PCBs was  calculated to range from 175 to 2000 pg TEQ/g  wet weight. In the
 samples, up to 37% of the total TEQ was derived from PCN  66/PCN 67 and  1,2,3,4,5,6,7-
 heptachloronaphthalene (PCN 73), with 25% derived from PCN 66/PCN 67.

 Awareness of limited data
 I am very well aware of the limited  data sets available to derive a TEF value. The ones available
 include chicken embryos and in vitro systems in chicken hepatocytes,  binding assays to the Ah
 receptor, in vitro studies with a rat hepatoma cell line, and in vivo studies in rodents. However, TEF
 values  are interim values which will change until more data  become available. By setting  TEF
 values  for the mentioned congeners  and using these preliminary values for the calculation of the
 total TEQ in selected samples, scientists and regulatory agencies can be made aware of the need for
 designing robust studies to derive relative potency values and continue - or even start - measuring
 the mentioned compounds in biota.

 In summary, TCAB, HCB, and PCNs should be included in the TEF concept based on binding to
 the Ah-receptor, their dioxin-like effects, and their bioaccumulation. The limited data available on
 environmental levels of TCAB, HCB, and PCNs suggest that these  compounds could considerably
 add to the total TEQ in environmental samples.

 I would like to thank Dr. John Bucher for giving me the challenge in the National Toxicology -
 Program to be involved in the technical report on the toxicity of 3,3',4,4'-tetrachloroazobenzene,
 Dr. George  Becking for  the opportunity to participate in the  IPCS task group meeting on
 hexachlorobenzene, and Prof. Bo Jansson and Dr. Jerzy Falandysz for their efforts to keep me
 updated on chlorinated naphthalenes.

 Ahlborg,  U.G., Brouwer, A., Fingerhut,  M.A, Jacobson, J.L., Jacobson,  S.W., Kennedy, S.W., Kettrup, AA.F.,
 Koeman, J.H.,  Poiger, H., Rappe, C., Safe, S.H., Seegal, R.F., Tuomisto, J., and Van den Berg, M (1992). Impact of
 polychlorinated dibenzo-p-dioxins,  dibenzofurans, and biphenyls on human and environmental health, with special
 emphasis on application of the toxic equivalency factor concept Eur. J. Phormacol. 228,179-199.
 Ahlborg, U.G., Becking, G.C., Bimbaum, L.S., Brouwer,  A., Derks, H.J.G.M., Feeley, M, Golor, G., Hanberg, A.,
 Larsen, J.C.  Liem, A.K.D., Safe, S.H.,  Schlatter, C., Waern, F., Younes, M, and Yijanheikki, E. (1994). Toxic
 equivalency factors for dioxin-like PCBs. Chemosphere 28,1049-1067.

 Asplund,  L., Jansson, B., Sundstrom, G., Brandt, I., and Brinkman, U-A.. (1986). Characterisation of a strongly
 bioaccumulatinghexachloronaphthalene. Chemosphere'15,  619-628.

 Asplund, L., Jakobsson, E., Haglund, P., and Bergman, A.  (1994). 1,2,3,5,6,7-Hexachloronaphthalene and 1,2,3,4,6,7-
 hexachloronaphthalene - selective retention in rat liver and appearance in wildlife. Chemosphere 28,2075-2086.

Bartha, R., Linke, H.A.B., and Pramer, D. (1968). Pesticide transformations: production of chloroazobenzenes from
 chloroanilines. Science 161,582-583.

Bartha, R., and Pramer, D. (1969). Transformation of the herbicide methyl-n-(3,4-dichlorophenyl)-carbamate (Swep) in
soil. Bulletin of Environmental Contamination and Toxicology 4,240-245.


 Hsia, M.T.S., and Kreamer, B.L. (1985). Delayed wasting syndrome and alterations of liver gluconeogenic enzymes in
 rats exposed to the TCDD congener 3,3',4,4'-tetrachloroazoxybenzene. Toxicol. Lett. 25,247-258.

 IPCS.  (1997). Environmental  Health  Criteria  195:  Hexachlorobenzene.  Geneva, World  Health  Organization,
 International Programme on Chemical Safety.

 Jacobsson,  E.,  Eriksson,  L.,  and  Bergman,  A.   (1992).   Synthesis  and  crystallography  of  1,2,3,5,6,7-
 hexachloronaphthalene and 1,2,3,4,6,7-hexachloronaphthalene.^c/a Chem. Scand. 46, 527-532.

 Kearney, P.C., Smith, R.J.J., Plimmer, Jr., and Guardia, F.S. (1970). Propanil and TCAB residues in rice soils. Weed
 Sci.  18, 464-466.

 Kenaga, E.E. (1980) Predicted bioconcentration factors and soil sorption coefficients of pesticides and other chemicals.
 Ecotoxicol. Environ. Saf. 4, 26-38.

 Lay, M.M., and Ilnicki, R.D. (1974). Peroxidase activity and propanil degradation in soil. Weed Res. 14, 111-113.

 Mansour, M., Parlar, H., and Korte, F. (1975). Ecological chemistry. I. Reaction behavior of 3,4-dichloroaniline and
 3,4-dichlorophenol in solution as a solid and in gas phase during UV radiation. Chemosphere 4,235-240.

 McMillan, D.C., Leakey, J.E.A., Arlotto, M.P., McMillan, J.M., and Hinson, J.A. (1990). Metabolism of the arylamide
 herbicide Propanil. Toxicol. Appl. Pharmacol 103,102-112.

 McMillan, D.C., Bradshaw, T.P., Hinson, JA.,  and Jollow, DJ. (1991).  Role of metabolites in  Propanil-induced
 hemolytic anemia. Toxicol. Appl. Pharmacol. 110,70-78.

 Mensink, J.A.,  and Strik, J.J.T.W.A. (1982). Porphyrinogenic action of tetrachloroazobenzene. Bull Environ. Contam.
 Toxicol. 28, 369-372.

 Miller, G.C., Zisook, R., and Zepp, R. (1980). Photolysis of 3,4-dichloroaniline in natural waters. J. Agric. Food Chem.

 Poland, A., Glover, E., Kende, A.S., DeCamp, M., Giandomenico, and C.M. (1976). 3,4,3',4'-Tetrachloro azoxybenzene
 and azobenzene: potent inducers  of aryl hydrocarbon hydroxylase.Sc/ewce 194,627-630.

 Rozman, K., Mueller, W., latropoulos, M., Coulston, F.,  and Korte, F.  (1975). Ausscheidung, Koerperverteilung und
 Metabolisierung von Hexachlorbenzol nach oraler Einzeldosis in Ratten und Rhesusaffen. Chemosphere 4,289-298.

 Scheufler, E., and Rozman, K.K. (1984). Comparative decontamination of hexachlorobenzene-exposed rats and rabbits
 by hexadecane. J. Toxicol. Environ. Health 14,353-362.

 Schlabach, M., Biseth, A., Gundersen, H., and Knutzen, J.  (1995).  Congener specific determination and levels of
 polychlorinated naphthalenes in cod liver samples from Norway. Organohalogen Compounds 24,489-492.

 Schneider, U.A., Brown, M.M.,  Logan, RA., Millar, L.C., and Bunce, N.J. (1995). Screening assay for dioxin-like
 compounds based on competitive binding to the urine hepatic Ah receptor.  1. Assay development. Environ. Sci.
 Techno!. 29, 2595-2602.

 Schrankel, K.R.,  Kreamer, B.L., and Hsia,  M.T.S. (1982).  Embryotoxicity of 3,3',4,4'-tetrachloroazobenzene and
 3,3',4,4'-tetrachloroazoxybenzene in the chick embryo.^rc/z. Environ. Contam. Toxicol. 11,195-202.

 Sinclair, P.R., Walton, H.S., Gorman, N., Jacobs, J.M., and Sinclair, J.F. (1997). Multiple roles of polyhalogenated
 biphenyls in causing increases in cytochrome P450 and uroporphyrin accumulation in cultured hepatocytes.  Toxicol.
Appl. Pharmacol.  147,171-179.

 Sundstrom,  G.,  Jansson,  B.,  and  Renberg,  L.  (1978).  Determination  of  the toxic  impurities  3,3',4,4'-
tetrachloroazobenzene and 3,3',4,4'-tetrachloroazoxybenzene in commercial Diuron, Linuron and 3,4-dichloroaniline
samples. Chemosphere 12,973-979.

US-EPA. (1985). Health and environmental effects profile for TCAB, TCAOB and TCHB. EPA/600/X-85/394.

US-EPA (1987). Pesticide Fact Sheet No. 149. Office of Pesticides and Toxic Substances, Washington, DC.

US-EPA. (1994). Estimating exposure  to dioxin-like compounds. Volume II: Properties, sources, occurrence and
background  exposure.  EPA/600/6-88/005Cb. External review draft.  U.S.  Environmental  Protection Agency,
Washington, DC.

Yamaguchi, Y., Kawano, M., and Tatsukawa, R. (1986). Tissue distribution and excretion of hexabromobenzene (HBB)
and hexachlorobenzene (HCB) administered to rats. Chemosphere 15, 453-459.

Paul Goettlich, Board of Directors
Hoosier Environmental Council
P.O.Box 6854 South Bend IN 46660-6854
 Council           www.envirolink.org/orgs/becweb
 22 January, 1998
 The Comments of an Observer to the Expert Reviewers and EPA at the Chicago Hilton
 Workshop on the Application of 2,3,7,8-TCDD Toxicity Equivalency Factors to Fish and Wildlife

 Thank you for allowing me the opportunity to be a part of this workshop as an observer. It has been an educational event
 for me and something that I will build upon, especially the acronyms.
 Thank you also for all your hard work and persistence in your pursuit of truth. I will feel safer knowing you are all
 working so hard on the subject of this workshop. I have had conversations with many of you. Some of my comments may
 have seemed vague and generally out of place. But they are not so out of place, for the questions I raise must be asked at
 each step that we take in our work. We must all develop a long range outlook on our home the earth because it is the only
 environment that most will have for a very long time. To look ahead 3 generations is definitely not enough. Even 30
 generations is not enough. I am not certain what is the appropriate amount of foresight when it comes to the degradation
 of our environment.

 The minimum time span that should be included is the useful  life span of the chemical in question. How long does it take
 for this chemical to break down into something that aJLscientists are certain is safe? Whether the chemical in question is a
 PCB, dioxin or one of the latest supposedly safe  chemical compounds such as glyphosate. Even that is looking to have
 the characteristics of an endocrine disrupter. As long as I have your attention I'd like to throw nuclear waste into the
 arena, as well, because there are agencies hard at work presently attempting to set limits on the amount of nuclear waste
 that can be recycled into consumer products.

 The reason I believe such prudence is due is to ensure that we are not leaving our descendants an intolerable
 environmental mess.  It is important that industry include the full costs of a product in its price to the consumer. Almost
 never included are the costs of production waste, environmental pollution, habitat loss, health problems for factory and
 field workers, health problems for the consumers and long term storage of toxic wastes.

 We must stop passing these problems and costs on to future generations of unsuspecting people,  many of whom will
 never have any use for the products or byproducts that will degrade their bodies and environment. When we increase a
 pollutant into our environment at any level it has some effect,  whether it is presently measurable or not. When a permit is
 granted to an industry to add additional pollutants because our legal system has deemed it safe, the effects are felt around
 the world, especially when it comes to such things as dioxin and PCBs. You may think of these effects as social or
 economic problems, and therefore not in your field.  It is then  safe to ignore the obvious problems that we have saddled
 other neighborhoods, states or countries with. Maybe we cannot prove that a particular PCB or dioxin was made by a
 particular factory, but we did add  to the whole ecosystem by granting that permit to increase pollutants production. As
 such we are just as liable as the next polluter or discharger, as  the legal people would say.

 In conclusion, I must urge you all to do as much interdisciplinary communication as possible. Get really wild and speak
to a psychologist or psychiatrist about their patient that has been chemically injured and has behavioral problems. It is
with these inventive communications'that your new ideas will  form. Remain open at all times to new ideas. Most of all,
think in terms of many hundreds of generations in an attempt to slow down the constant degradation.

Paul Goettlich

Date:        February 25, 1998

To:          Susan Brager Murphy, TEF Workshop Coordinator

cc:           Charles Menzie, Linda Birnbaum

From:       Brent Finley, Workshop Observer

Subject:     Expanding the TEF Approach to include Hexachlorobenzene (HCB)


As part of the effort to address potential human health risks posed by chlorinated  dibenzo-p-
dioxins (CDDs) and chlorinated dibenzofurans (CDFs) in the environment, the U.S. EPA adopted
an interim procedure in 1987 based on dioxin "toxicity equivalence" factors (TEFs) for estimating
the hazard and dose-response  of complex mixtures containing CDDs and  CDFs in  addition to
2,3,7,8-TCDD (USEPA,  1989).  The adoption of TEFs for CDD/CDF congeners was explicitly
stated and recognized by researchers to be an interim science  policy measure.  The technical
subcommittee that was gathered  to derive and periodically update the TEF scheme noted that a
general (order of magnitude) approach was needed to  characterize potential risks posed by the
209 CDD/CDF congeners other than 2,3,7,8-TCDD because of the lack of detailed toxicity data
on almost all of these congeners.  With the development of updated TEFs in 1989 (i.e., I-TEFs), it
was again noted by the subcommittee  that the TEF approach was an "interim" approach and
should be replaced as soon as  practicable with a bioassay method.  Over the past several years,
several efforts have been  made  to expand the TEF approach to include mixtures of coplanar
polychlorinated biphenyls (PCBs) (Ahlborg et al. 1994; Safe et al. 1994), but a consistent set of
TEF values has yet to be adopted by USEPA for PCBs.

In the draft  1997 W.H.O. document,  it was suggested that the health risks associated with
exposure to HCB (and other chemicals) could"be assessed using the 2,3,7,8-TCDD-based TEF
scheme.  This suggestion was based on an apparent concern that: 1) HCB might possess "dioxin-
like" properties and, 2) it is important to understand the "total TEQ body burden" of humans and
ecological species.  This suggestion was further detailed by Angelique  van Birgelen at the
workshop in  Chicago.  During her presentation, she elaborated on the information she  had written
for the W.H.O. document. It was not clear to me whether any of the other workshop participants
supported or disagreed with her position.  In general, I  have some serious reservations about the
practicality and applicability of a "TEF" for a chemical which exists not as a mixture but as a
single compound (such as HCB), and I specifically have some misgivings about applying a TEF to
HCB, a compound whose toxicity is already  well characterized.  I have detailed some of my
observations and concerns below.

1.     A TEF for HCB is not Warranted Because There are no HCB "Mixtures"

The  TEF approach  was originally developed as an interim approach for complex mixtures  of
GDD/Fs. Because it is obviously impractical to conduct long-term bioassays on each and every
one of the CDD/F congeners, the TEF approach was developed as a "short-cut" that would allow
for assessments of complex  mixtures using existing congener data (in vitro studies,  etc.). As
discussed in the W.H.O. report (1997),  there are a large number of compounds  which could
contribute to the total  concentration  of TCDD TEQs.  These chemicals include  polycyclic
aromatic compounds such as biphenyls, 2- and 3-ring polycyclic aromatic hydrocarbons and other
heterocyclic compounds. Depending on the degree of substitution, many congeners of polycyclic
compounds can exist.   For  most of these  compounds,  little or no  toxicity data  exists  to
characterize a dose-response relationship. If in fact ircan be demonstrated that some or all of the
various congeners in a chemical class are likely  to possess significant "dioxin-like" activity, then
perhaps development of TEFs for this chemical class is appropriate.

On the contrary, dose-response relationships have been developed for HCB using the results  of
several bioassays  to characterize effects from subacute to chronic exposures.  Attempting  to
incorporate HCB into a TEF  scheme is therefore inconsistent and contrary to  the TEF concept
because it ignores the more  accurate assessment techniques  that have already been applied  to
HCB toxicity data for characterizing adverse effects.  More importantly,  because there are no
congeners or isomers of HCB, its inclusion in the TEF scheme seems counterintuitive.

2.     There  is Insufficient Evidence to Conclude HCB Meets  W.H.O.'s Definition  of

As defined by the W.H.O.(1997) in their  draft workshop proceedings,  there are four criteria that
W.H.O. has proposed to determine whether a chemical might be evaluated using the TEF scheme.
Each of these criteria, as they apply to HCB, is discussed below.

> Structural  Similarity to TCDD - As is shown in the  figure below, HCB is a monocyclic
   (single-ringed) aromatic  compound  with—full chlorine substitution, whereas  TCDD is  a
   coplanar, polycyclic compound with  chlorine substituted at four locations. As such, HCB
   lacks the structural dimensions (a 6.8x13.7 Angstrom box of chlorine substitutions) required
   for TCDD-like toxicity (Hahn et al. 1989; McKinney and Singh,  1981).

                    I-9.8A-I                              I	13.7A—-I

            Hexachlorobenzene                            2,3,7,8-TCDD

   Extensive evidence exists on CDDs/CDFs to show that small configurational changes such as
   chlorine substitution on a specific carbon atom drastically affects the potency of a compound.
   For example, 2,3,5,7-TCDD is not considered as a candidate for a TEF because of the subtle
   difference in the placement of a single chlorine  atom.  The significant structural  and
   configurational  differences between the polycyclic CDDs/CDFs  and monocyclic HCB is
   compelling evidence that support the conclusion it does not meet the first criterion for a
   dioxin-like compounds.

>  Binds to the Ah Receptor - The evidence supporting HCB binding to the Ah receptor has
   been described as "equivocal" and "at best a very w.eak competitor" (Goldstein et al. 1986;
   Linko et al. 1986).  The binding affinity of HCB is  far less than that of every 2,3,7,8-
   substituted CDD/F, and in fact, is  much less than that observed for  naturally occurring
   aromatic compounds such as polycyclic aromatic hydrocarbons and indole carbazoles (Ames
   et al. 1990; Kleman and Gustafsson,  1996).  Therefore, it  would seem that HCB fails on this

>  Dioxin-Like Toxic/Biochemical Responses - HCB can induce several responses that can also
   be induced by TCDD,  including CYPIA1/IA2 induction, thyroid hormone alterations,  hepatic
   retinol  depletion,  porphyrin  accumulation,  hepatic  hypertrophy,  and   immunotoxicity.
   However, the HCB doses required to elicit these effects are orders of magnitude higher than
   TCDD doses required to elicit the same degree of effect. In addition, the mechanism by which
   HCB  produces toxic  effects may be quite different than  that for TCDD.  For instance,
   oxidative  metabolites   of HCB  (pentachlorophenol,  tetrachJorohydroquinone)  have been
   implicated in the manifestation of hepatic porphyria and other effects (Rietjens et al. 1995;
   Schielen et al. 1995; Van Ommen et al. 1989).  Conversely, the toxicity of TCDD is generally
   attributed to the interaction of the parent compound with the Ah receptor.

>  Persistence - In humans, the half-life of HCB has been estimated to be approximately 215
   days (Freeman et al. 1989), which is less than a tenth of the half-life reported for TCDD of
   approximately  7.5  years (Needham et  al.  1994).  Clearly, the pharmacokinetics of HCB is
   vastly different from that of TCDD.  Indeed,  one of the  deficiencies  of the TEF approach,
   particularly with TEFs based on in vitro or acute in vivo responses, is that it does not account
   for differences in kinetics, an important determinant of toxicity. In summary, it is questionable
   whether HCB can be considered "persistent"  relative to TCDD.

3.     Numerous Chemicals Meet W.H.Q.'s Criteria of "Dioxin-Like"

As noted above, there are several  chemical classes which  could be interpreted to meet the
somewhat arbitrary criteria of "dioxin-like," even  as  defined by the use  of W.H.O criteria.
However, it is unreasonable to suggest that TCDD-based TEFs will be derived for each isomer of
each of these chemical classes.  Even if sufficient resources existed to establish such TEFs over
the next 5-10 years, the result would be an unwieldy collection of hundreds or thousands of TEF
values. This is unlikely to occur, and therefore  I believe that simply because a chemical meets a
definition of "dioxin-like"  is insufficient reason to  attempt to establish a TEF for that compound.
This is particularly true for HCB which, as noted above, does not satisfy W.H.O.'s criteria.

A TEF is Not Warranted Because Sufficient Toxicological Criteria Exist for HCB
The TEF approach was originally developed for PCDD/F congeners since data to characterize
toxicity was limited or not available.  This is not the case for HCB.  Rather, the toxicity of HCB
has been extensively studied, and the dose-response relationships for various  health effects have
been well characterized.  Toxicity values  for HCB and TCDD used  for risk assessment  are
compared in the table below.
Toxicity Value
Cancer Slope Factor •
Oral Reference Dose
Based on liver tumors in
female rats (Eturk, 1986)
Based on liver effects in
rats (Arnold et al. 1985)
Based on survival and
reproductive effects in
mink (Rush et al. 1983;
Bleavins et al. 1984)
Based on survival in
several aquatic species
(USEPA, 1988)
Based on survival and
reproductive effects in
several species of birds
(Vosetal. 1971)
Based on liver tumors in
female rats (Kociba et al.
Not available
Based on reproductive
effects in rats (Murray et al.
Based on survival in
rainbow trout and northern
pike (USEPA, 1993)
Based on survival,
reproductive &
developmental effects in
several species of birds
(Nosek et al. 1992; Hudson
et al. 1984)
Currently, the cancer slope factor and reference dose for HCB are derived without prejudice to
the mechanism by which adverse effects were produced.  As such, any "dioxin-like" activity
imparted by HCB in  the critical lexicological studies is already accounted for in the existing slope
factor and reference dose.

For ecological risk assessment endpoints, there are no promulgated toxicity values as there are for
human health. However, the quantity and quality of available toxicolpgical studies for ecological
receptors of potential concern for HCB is as good or better as that for TCDD. Controlled studies
of subchronic and chronic HCB exposures have been conducted with 10 species of mammals and
five  species of birds.  These include wild species such as the mink and the kestrel which are
frequently identified  as receptors of interest for ecological risk assessment.   These studies have
assessed the dose-related effects of HCB on the survival, growth, reproduction and development
of these species; all  endpoints clearly related to risks to exposed populations. The studies with

Japanese quail indicate that hatchability of eggs and survival of chicks are sensitive endpoints for
birds and provide a sound basis of a toxicity reference value for this taxonomic group (Vos et al.,
1971; Schwetz et al.,  1974).  Similarly, the studies with mink show that reductions  in litter size
and survival of kits are sensitive endpoints and  provide  a documented basis for developing  a
toxicity reference value for mammals (Bleavins et al.,  1984; Rush et al., 1983).  Because these
studies are generally of good quality and have assessed the dose-related effects of HCB exposure
on relevant endpoints in a relatively large number of species, there are substantial data from direct
tests to evaluate HCB toxicity and little justification, if any, for abandoning these data for this
individual compound for "an order of magnitude" TEF.

The dose-related health  effects of HCB observed in long-term animal feeding  studies are a far
more accurate and direct measure of HCB toxicity than an indirect "TEF-estimate" which  is based
on the potency of an entirely different chemical.  Chemical-specific information, based on chronic
bioassays for all endpoints of concern, is clearly a more preferable basis for risk assessment.

5.     The Relative Potency (REP) of 0.0001 is Overestimated

Although an REP of 0.0001 can be calculated for HCB  based on relative binding affinity to the Ah
receptor in vitro (Hahn et al. 1989), this is likely to overestimate the carcinogenic potency. Using
the relative carcinogenic potencies  of HCB [(1.6 (mg/kg-day)"1] and TCDD [(156,000  (mg/kg-
day)"1] in female Sprague-Dawley rats following  lifetime exposures (Eturk,  1986; Kociba et al.
1978), an REP of 0.00001 may be viewed as a conservative upper bound (under the  unlikely
assumption that all  HCB-induced liver tumors are attributable to a "dioxin-like" mechanism of
action). Because HCB-induced tumors are primarily attributable to a non-dioxin-like mechanism
of action, the REP for  carcinogenic effects is likely to be much less than  0.00001, and in all
likelihood is closer to (if not equal to) zero.


The TEF approach was developed and adopted specifically to address the potential risks posed by
related constituents (i.e., congeners) with similar structural features that might elicit a response or
toxic  effect under an identical mechanism  of action.  HCB is a well-studied chemical for which
current  risk assessment methods are  superior to the TEF approach.  Given  that there is no
discernible benefit in  adding HCB  to the TEF  scheme, I strongly recommend that the TEF
approach not be applied  to HCB for characterizing potential risks.


Alhborg UG, Becking GC, Birnbaum LS, et al.  1994. Toxic equivalency factors for dioxin-like
PCBs. Chemosphere 28(6): 1049-1067.

Ames BN, Gold LS.  1990.  Natural chemicals, synthetic  chemicals, risk assessment, and cancer.
Princess Takamatsu Symp 21:3 03 -14.

Arnold DL, et al.  1985.  Long-term toxicity of hexachlorobenzene in the rat and the effect of
dietary vitamin A. Food Chem Toxicol 23(9):779-93.

Bleavins MR, Aulerich  RJ, Ringer RK.   1984.  Effects of chronic dietary exposure on the
reproductive performance and survivability on mink and European ferrets.  Arch Environ Contain
Toxicol 13:357-365.

Eturk E, Lambrecht RW, Peters HA,  et al.  1986.   Oncogenicity  of hexachlorobenzene.  In:
Hexachlorobenzene:  Proceedings of an international symposium. IARC Sci Pub 77:417-423.

Freeman RA, Rozman  KK, Wilson AG.    1989.   Physiological  pharmacokinetic  model  of
hexachlorobenzene in the rat. Health Phys 57 Suppl 1:139-47.

Goldstein JA, et al.  1986.  Structure-activity relationships of chlorinated benzenes as inducers of
hepatic cytochrome P-450 isozymes in the rat. IARC Sci Publ, 77:519-26.

Hahn ME,  et al.    1989.   Interaction of  hexachlorobenzene  with the receptor  for 2,3,7,8-
tetrachlorodibenzo-p-dioxin in vitro and in vivo. Evidence that hexachlorobenzene is a weak Ah
receptor agonist. Arch Biochem Biophys 270(l):344-55.

Hudson RH, Tucker RK, Haegele MA.  1984.  Handbook of toxicity of pesticides to wildlife.
U.S. Fish and Wildlife Service Resource Pub. 153.

Kleman M, Gustafsson  JA.  1996.  Interactions of procarcinogenic  heterocyclic amines and
iridolocarbazoles with the dioxin receptor. Biol Chem 377(11):741-62.

Kociba RJ, Keyes DG, Beyer JE, et al. 1978. Toxicol Appl Pharmacol 46:279-303.

McKinney JD, Singh P.  1981. Structure-activity relationships in halogenated biphenyls: unifying
hypothesis for structural specificity. Chem Biol Interact 33:271-283.
Murray FJ,  et  al.  1979.   Three generation reproduction study  of rats  given 2,3,7,8-
tetrachlorodibenzo-p-dioxin (TCDD) in the diet.  Toxicol Appl Pharmacol 50:241-252.

Needham LL, Gerthoux PM, Patterson DG,  et al. 1994. Organohalogen Compd 21:81-85.

Nosek JA, Craven SR,  Sullivan JR, et al. 1992.  Toxicity and reproductive effects of 2,3,7,8-
tetrachlorodibenzo-p-dioxin in ring-necked pheasants.  J Toxicol Environ Health 35:187-198.

Rietjens IM, et  al.   1997.  Cytochrome P450-catalyzed oxidation of halobenzene derivatives.
Chem Res Toxicol 10(6):629-35.

Rose JQ, et al.   1976.  The fate of 2,3,7,8-tetrachlorodibenzo-p-dioxin following  single and
repeated oral doses to the rat. Toxicol Appl Pharmacol 36(2):209-26.

Rush GF,  Smith JH, Malta K, et al. 1983.   Perinatal hexachlorobenzene toxicity in the mink.
Environ Res 3 1:1 16- 124.

Safe SH.   1994.  Polychorinated biphenyls (PCBs): environmental impact, biochemical and toxic
responses, and implications for risk assessment. Crit Rev Toxicol 24:87-149.

Scheufler E, Rozman KK.  1984.  Comparative decontamination of hexachlorobenzene-exposed
rats and rabbits by hexadecane. J Toxicol Environ Health 1 4(2-3) :3 53-62.

Schielen P, et al: 1995. Immune effects of hexachlorobenzene in the rat: role of metabolism in a
13 -week feeding study. Toxicol Appl Pharmacol 131(l):37-43.

Schwetz  BA, Norris  JM,  Kociba  RJ.   1974.   Reproduction study in Japanese quail  fed
hexachlorobutadiene for 90 days.  Toxicol Appl Pharmacol 30:255-265.

USEPA.   1988.  Ambient aquatic life water quality criteria for hexachlorobenzene.  Office of
Research and Development, U.S. Environmental Protection Agency.

USEPA.    1993.    Interim  report  on  data  and  methods  for  assessment  of  2,3,7,8-
tetrachlorodibenzo-p-dioxin risks to aquatic life and associated wildlife.  Washington, DC:  U.S.
Environmental Protection Agency.
    .Ommen  B, et al.   1989.   The relation between the oxidative  biotransformation  of
hexachlorobenzene and its porphyrinogenic activity. Toxicol Appl Pharmacol 100(3): 5 17-28.

Vos JG, Botterweg PF, Strik JJ, Koeman JH.  1971.  Experimental studies with HCB in birds.
TNONieuws 27:599-603.

W.H.O. 1997. Meeting on the derivation of toxic equivalency factors (TEFs) for PCBs, PCDDs,
PCDFs and other dioxin-like compounds for humans and wildlife (Draft).