FINAL
AMMONIUM
PERFLUOROOCTANOATE
(
C8)

ASSESSMENT
OF
TOXICITY
TEAM
(
CATT)
REPORT
August
2002
Department
of
Environmental
Protection
­
promoting
a
healthy
environment
EXECUTIVE
SUMMARY
Pursuant
to
a
consent
order
signed
November
14,
2001
between
the
West
Virginia
Environmental
Protection
and
Health
and
Human
Resources
departments,
and
E.
I.
Du
Pont
de
Nemours,
Inc.
(
DuPont)
the
C8
(
ammonium
perfluorooctanoate)
Assessment
of
Toxicity
Team
(
CATT)
was
established
to:

(
1)
determine
risk­
based
human
health
protective
screening
levels
(
SLs)
for
this
unregulated
chemical
in
air,
water,
and
soil;

(
2)
provide
health
risk
information
to
the
public;
and
(
3)
determine
an
ecological
health
protective
SL
for
C8
in
surface
water.

To
date,
two
public
meetings
have
been
held
in
the
vicinity
of
the
DuPont
Washington
Works
facility
located
near
Parkersburg,
West
Virginia.
Also,
a
team
of
10
expert
toxicologists
have
met
and
determined
human
health
provisional
risk
factors
for
the
oral
and
inhalation
routes
of
exposure,
and
calculated
health
protective
SLs
based
on
these
risk
factors
using
Region
9
U.
S.
Environmental
Protection
Agency
standard
methodology.
The
results
of
the
CATT's
investigation
are
presented
in
summary
below.
The
ecological
SL
for
surface
water
currently
is
still
in
development.
An
addendum
to
this
report
is
expected
to
be
released
in
Fall
2002
presenting
the
surface
water
SL
findings.

The
methodology,
overall
process,
and
rationale
utilized
by
the
CATT
to
develop
these
risk
factors
and
SLs
are
discussed,
the
members
are
listed,
and
a
synopsis
of
the
events
leading
to
the
consent
order
are
presented
herein.
The
intent
of
this
report
is
to
document
the
process
and
conclusions
of
the
CATT
in
an
effort
to
provide
to
the
public
a
record
of
these
activities.
It
is
not
intended
to
be
a
summary
of
all
the
toxicology
information
available
on
C8.

The
risk
factor
or
Reference
Dose
(
RfD)
for
the
oral
route
of
exposure
determined
by
the
CATT
for
C8
was
0.004
milligrams
per
kilogram
of
body
weight
per
day
(
mg/
kg­
day).
A
risk
factor
for
the
inhalation
route
of
exposure
or
the
Reference
Concentration
(
RfC)
of
1
micrograms
per
cubic
meter
of
air
(
µ
g/
m3
)
was
determined.
The
RfD
or
RfC
is
defined
by
EPA
as
an
estimate
(
with
uncertainty
spanning
perhaps
an
order
of
magnitude
or
greater)
of
a
daily
exposure
level
for
the
human
population,
including
sensitive
subpopulations,
that
is
likely
to
be
without
an
appreciable
risk
of
deleterious
effects
during
a
lifetime.
Based
on
the
oral
RfD,
health
protective
SLs
were
calculated
for
water
of
150
parts
per
billion
(
ppb),
and
for
soil
of
240
parts
per
million
(
ppm).
Based
on
the
inhalation
RfC,
a
health
protective
SL
of
1
µ
g/
m3
was
derived
for
air.

2
ACKNOWLEDGEMENTS
The
West
Virginia
Department
of
Environmental
Protection
wishes
to
thank
the
following
agencies
and
organizations
that
joined
us
as
primary
participants
in
this
investigation:
West
Virginia
Department
of
Health
and
Human
Resources;
U.
S.
Environmental
Protection
Agency
(
EPA)
Region
3,
Office
of
Research
and
Development
(
ORD)
and
Headquarters;
E.
I.
Du
Pont
de
Nemours,
Inc.
(
as
well
as
their
employees,
consultants
­
Potesta
&
Assoc.,
Inc.,
laboratory
 
Exygen
Research,
Inc.,
and
attorneys);
Marshall
University;
Toxicology
Excellence
for
Risk
Assessment
(
TERA);
and
Menzie
Cura
&
Assoc.,
Inc.
Specifically,
we
thank
the
following
EPA
personnel
for
their
technical
support
and
camaraderie:
Karen
Johnson,
Janet
Sharke,
Garth
Connor,
Roger
Reinhart,
and
Mary
Dominiak.
We
also
thank
the
following
organizations
for
their
cooperation:
EPA
Region
5,
Ohio
EPA,
and
the
National
Institute
for
Chemical
Studies.

We
thank
all
the
individual
members
of
the
C8
Assessment
of
Toxicity
Team
(
CATT)
for
their
participation
and
cooperation.
In
particular,
we
thank
the
following
CATT
members:

 
James
Becker,
M.
D.,
and
Tracy
Smith,
M.
S.,
of
Marshall
University
for
their
professionalism,
scientific
knowledge,
and
common
sense
approach
to
communicating
environmental
health
risks
to
the
public.

 
The
toxicologists
who
embarked
on
an
expedition
to
find
the
truth,
the
ambition
of
all
noble
scientists:

EPA
John
Cicmanec,
D.
V.
M.,
M.
S.,
USEPA
ORD
Samuel
Rotenberg,
Ph.
D.,
USEPA
Region
3
Jennifer
Seed,
Ph.
D.,
USEPA
Headquarters
TERA
Michael
Dourson,
Ph.
D.
Joan
Dollarhide,
MS,
MTSC,
JD
Andrew
Maier,
Ph.
D.,
CIH
Dan
Briggs,
Ph.
D.,
DABT
(
note
taker)

Agency
for
Toxic
Disease
Registry
John
Wheeler,
Ph.
D.

DuPont
Gerald
Kennedy
John
Whysner,
M.
D.,
Ph.
D.,
D.
A.
B.
T.
(
consultant)

Invited
guests:
John
Butenhoff,
Ph.
D.,
3M
(
study
scientist)
Jim
Sferra,
MS,
OEPA
(
observer)

3
TABLE
OF
CONTENTS
Title
Page
Cover
Page
                               
1
Executive
Summary
                           ..
2
Acknowledgements
...............................................................................................................
3
Table
of
Contents
                            ..
4
List
of
Tables
..                             ...
5
1.0
Introduction
                          
   ..
6
2.0
Development
of
Risk
Factors
and
Screening
Levels
for
C8
       
   .
9
2.1
Pre
CATT
Toxicologists
Meeting
Action
Items
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
9
2.2
CATT
Toxicologists
Meeting
Minutes
   .
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
..
.
.
.
..
10
2.3
Post
Meeting
Action
Items
   ..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
..
36
2.4
Summary
Table
of
Findings
   ..
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
42
3.0
Comparison
of
Screening
Levels
to
Site­
Related
Data
       ..
.
.
.
.
.
.
.
.
.
..
46
Attachments
I
Final
Consent
Order
on
C8
between
West
Virginia
and
DuPont
II
Presentations
from
Public
Meetings
IIa
First
Public
Meeting
IIb
Second
Public
Meeting
III
CATT
Toxicologists
Pre­
Meeting
Information
IIIa
Pre­
meeting
Toxicology
Review
Summary
IIIb
Agenda
IIIc
List
of
Potential
Key
Studies
IIId
Detailed
Summary
Table
of
Key
Studies
IIIe
Summary
Table
of
Key
Studies
(
with
blank
columns
for
uncertainty
factors)
IIIf
Summary
Table
of
Key
Studies
with
TERA's
Suggestions
for
Uncertainty
Factors
IIIg
Original
Signatures
IV
Post
Meeting
Actions
Data
IVa
Liver
Weight
Standard
Deviations
from
DuPont
IVb
Particle
Size
Distribution
from
DuPont
IVc
Benchmark
Dose
Modeling
Results
(
during
and
post
meeting)
IVd
Regional
Dose
Deposition
Ratio
Modeling
Results
V
References
and
Lists
of
Reviewed
Data
VIa
Index
of
Administrative
Record
226
VIb
List
of
Documents
Reviewed
by
TERA
4
LIST
OF
TABLES
1.
Summary
of
NOAELs,
LOAELs,
BMDLs,
and
Critical
Effects
for
Key
and
Supporting
C8
Studies
2.
Panel
Recommendations
of
UF
Selection
for
Oral
pRfD
3.
Comparison
of
pRfDs
Derived
Using
Different
Studies
4.
Preliminary
Conc
HEC
Calculations
from
Kennedy
et
al.,
(
1986)
5.
Tally
of
Panel
Votes
for
UFS
and
UFD
6.
Summary
of
Beigel
et
al.,
2001
Leydig
Cell
Data
7.
Biegel
Study:
Pancreas
Tumors
8.
Riker
Study:
Mammary
Tumors
9.
Factors
Used
to
Describe
Various
Areas
in
the
Development
of
MOEs
for
Cancer
Endpoints
10.
Calculation
of
Human
Equivalent
Concentrations
for
Kennedy
et
al.
(
1986)
11.
Benchmark
Dose
Modeling
Results
for
C8
12.
Summary
of
RfD
and
RfC
Values
for
C8
Determined
by
the
CATT
Toxicologists
5
1.0
INTRODUCTION
The
investigation
described
herein
was
conducted
pursuant
to
the
November
14,
2001
Consent
Order
Number
GWR­
2001­
019
between
the
West
Virginia
Departments
of
Environmental
Protection
(
DEP)
and
Health
and
Human
Resources
(
DHHR),
and
E.
I.
Du
Pont
de
Nemours,
Inc.
(
DuPont).
A
copy
of
this
consent
order
is
included
as
Attachment
I.
These
actions
were
instigated
by
the
presence
of
an
unregulated
chemical,
ammonium
perfluorooctanoate
commonly
called
C8,
in
the
Lubeck,
W.
Va.
public
water
supply
which
is
near
the
DuPont
Washington
Works
(
WW)
facility
in
Washington,
W.
Va.
A
site
map
is
included
in
Attachment
IIc.

The
consent
order
established
two
scientific
teams:
(
1)
the
C8
Assessment
of
Toxicity
Team
(
CATT),
and
(
2)
the
Groundwater
Investigation
Steering
Team
(
GIST).
The
CATT
was
tasked
with
investigating
the
toxicity
of
C8;
developing
provisional
risk
factors
for
the
inhalation,
dermal,
and
oral
routes
of
exposure;
and
establishing
human
health
protective
screening
levels
(
SLs)
for
air,
water,
and
soil;
investigating
the
ecological
toxicity
of
C8
and
determining
an
ecological
health
protective
SL
for
surface
water;
and
with
communicating
health
risk
information
to
the
public.
In
the
consent
order
DuPont
agreed
to
meet
these
SLs
at
their
WW
facility,
once
developed,
and
that
these
SLs
would
remain
in
effect
until
superseded
by
U.
S.
Environmental
Protection
Agency
(
EPA)
guidance.
The
CATT's
activities
and
findings
regarding
the
toxicity
of
C8,
development
of
risk
factors
and
SLs
are
presented
in
detail
in
Section
2
of
this
report.
Slides
presented
at
the
two
public
meetings
held
thus
far
are
provided
in
Attachment
II.
The
investigation
into
the
ecological
toxicity
of
C8
and
surface
water
SL
development
is
scheduled
for
completion
in
Fall
2002.
When
finished,
the
surface
water
will
be
presented
in
an
addendum
to
this
report.

The
GIST
was
established
by
the
consent
order
to
determine
the
extent
and
concentration
of
C8
in
both
groundwater
and
surface
water.
The
activities
of
the
GIST
continue
as
of
the
issuance
of
this
CATT
report.
The
GIST
will
issue
a
report
on
the
C8
analytical
data
for
groundwater
and
surface
water
when
that
work
is
finished,
scheduled
for
early
2003.
Interim
reports
are
available
through
the
DEP
Division
of
Water
Resources
(
DWR).
The
groundwater
investigation
focused
not
only
on
the
WW
plant,
but
also
on
areas
where
C8
had
been
disposed,
including
the
Local
Landfill
(
on
WW
property),
Dry
Run
Landfill
(
near
the
WW
plant),
and
the
Letart
Landfill
(
30
miles
south
of
the
WW
plant).
Maps
of
the
one­
mile
radius
study
area
around
these
locations
are
included
in
the
presentation
of
interim
results
at
the
second
public
meeting
provided
in
Attachment
IIc.

Summarized
findings
to
date
by
the
GIST
are
compared
to
the
health
protective
water
SL
developed
by
the
CATT
in
Section
3.0.
Results
of
air
dispersion
modeling
efforts
thus
far
conducted
by
the
DEP
Division
of
Air
Quality
(
DAQ)
are
compared
to
the
air
SL
in
Section
3.0
as
well.

Background
The
DuPont
WW
plant
is
located
approximately
10
miles
southwest
of
Parkersburg,
W.
Va.
along
state
Route
61
in
the
rural
hamlet
of
Washington,
W.
Va.
This
facility
was
established
in
the
1940s
and
currently
is
one
of
the
largest
DuPont
enclaves
in
the
world.
DuPont
has
used
C8
at
this
facility
for
more
than
50
years
as
a
surfactant
in
various
manufacturing
processes,
including
the
production
of
Teflon.
"
C8"
is
the
3M
trade
name
for
its
product
that
contains
ammonium
perfluorooctanoate
(
APFO)
(
CAS
#
3825­
26­
1).
In
biologic
media,
APFO
quickly
dissociates
to
perfluorooctanoate,
which
is
the
anion
of
perfluorooctanoic
acid
(
PFOA).
The
PFOA
form
has
been
identified
as
potentially
toxic
to
animals.
Throughout
this
report,
C8
is
used
as
terminology
to
include
C8,
APFO,
or
PFOA.

6
The
DEP
became
aware
of
and
began
investigating
the
presence
of
C8
in
the
Lubeck,
W.
Va.
public
water
supply
in
November
2000.
In
Spring
2001,
DEP
received
a
letter
requesting
a
formal
agency
investigation
into
DuPont's
environmental
releases
of
C8
and
the
presence
of
C8
in
the
Lubeck
drinking
water
from
attorneys
representing
a
few
citizens
residing
in
proximity
to
the
WW
plant.
The
Lubeck
public
water
supply
well
field
lies
approximately
3
miles
south
of
the
DuPont
WW
plant.
Also
around
this
time,
DEP
became
aware
that
C8
was
chemically
similar
to
perfluorooctane
sulfonate
(
PFOS),
another
perfluorocarbon
manufactured
by
3M,
and
that
3M
had
recently
removed
their
Scotchguard
product
from
the
marketplace
because
it
contained
PFOS.
From
U.
S.
EPA
Region
3
and
Headquarters,
DEP
learned
that
3M
had
undertaken
a
significant
research
effort
into
the
toxicity
of
perfluorocarbons,
particularly
PFOS
and
including
C8;
that
perfluorocarbons
were
potentially
more
toxic
than
previously
thought;
that
3M
was
submitting
the
new
data
to
EPA
under
the
Toxic
Substances
Control
Act
(
TSCA);
and
that
these
data
were
publicly
available
under
Administrative
Record
226
(
AR226).
Additionally,
DEP
learned
that
DuPont
was
submitting
toxicity
data
on
C8
to
EPA,
as
well.

DEP
gathered
data
and
met
with
DuPont
and
met
with
citizens
attorneys
in
Spring
2001.
The
DEP,
which
regulates
groundwater
in
West
Virginia,
was
joined
in
the
investigation
by
the
DHHR,
which
regulates
drinking
water.
The
DHHR
requested
support
from
EPA
Region
3
to
enforce
the
National
Safe
Drinking
Water
Act.
At
the
request
of
these
agencies,
DuPont
supplied
information
regarding
C8
and
its
use
in
manufacturing
processes,
its
toxicity,
and
emissions.
After
several
months
of
investigation
and
discussions,
a
consent
order
was
signed
in
November
2001.
A
copy
of
the
consent
order
is
provided
in
Attachment
I.
It
describes
the
tasks
and
members
of
the
CATT
and
GIST.
The
DEP
informed
the
public
of
the
consent
order
and
scheduled
a
public
meeting
to
discuss
the
order.

The
DEP
held
it's
first
public
meeting
regarding
C8
on
November
29,
2001
at
Blennerhassett
Junior
High
School
which
is
located
near
the
Lubeck
and
Washington
communities.
The
meeting
was
spearheaded
by
the
CATT
and
the
GIST.
The
purpose
of
the
meeting
was
to
inform
citizens
of:
(
1)
the
requirements
of
the
consent
order;
(
2)
the
members
and
activities
of
the
GIST;
(
3)
their
assistance
was
required
to
fill
out
and
return
a
water
use
survey
if
they
had
groundwater
wells,
cisterns,
or
springs
(
particularly
those
used
for
drinking
water),
and
to
allow
sampling
of
these
water
sources;
(
4)
the
members
and
activities
of
the
CATT;
(
5)
the
available
information
regarding
the
toxicity
of
C8;
and
(
6)
the
known
current
levels
of
C8
in
the
Lubeck
public
water
supply,
which
were
below
1
part
per
billion
(
ppb).
At
this
meeting,
James
Becker,
M.
D.
of
Marshall
University
spoke
regarding
environmental
exposures
and
risks
in
general,
and
Dee
Ann
Staats,
Ph.
D.
(
DEP)
explained
the
CATT
and
GIST
activities,
the
consent
order,
and
known
toxicity
of
C8.
The
slides
from
both
presentations
are
provided
in
Attachment
IIa.

By
the
end
of
January
2002,
contractors
were
in
place
to
assist
the
CATT
and
the
GIST
in
their
tasks.
The
GIST
was
headed
by
DEP
and
had
members
from
DHHR,
EPA,
and
Dupont.
The
CATT
was
headed
by
DEP
and
had
members
from
DHHR,
EPA,
DuPont
and
the
Agency
for
Toxic
Substances
Disease
Registry
(
ATSDR).
The
DEP
contracted
with
the
National
Institute
of
Chemical
Studies
(
NICS),
a
nonprofit
organization,
which
subcontracted
the
human
and
ecological
toxicology
work
to
the
Toxicology
for
Excellence
in
Risk
Assessment
(
TERA)
group,
also
a
nonprofit,
which
subcontracted
the
ecological
toxicology
work
to
Menzie
Cura
&
Assoc.,
Inc.
(
MC).
Both
TERA
and
MC
are
well
respected
in
the
field
of
toxicology.
The
NICS
subcontracted
the
risk
communications
tasks
to
Marshall
University.

In
March
2002,
EPA
Regions
3
and
5
signed
a
consent
order
with
DuPont
requiring
the
provision
of
alternative
water
to
any
resident
in
West
Virginia
or
Ohio
with
C8
in
drinking
water
at
levels
above
14
7
ppb.
The
14
ppb
was
an
interim
value
in
effect
until
the
water
SL
was
developed
by
the
CATT.
This
value
was
taken
from
the
final
report
by
ENVIRON
Int.
Corp.
(
a
consulting
firm
hired
by
DuPont)
titled
"
A
Hazard
Narrative
for
Perfluorooctanoate
(
PFOA)",
January
2002.
An
earlier
draft,
"
A
Review
of
the
Toxicology
of
Perfluorooctanoate
(
PFOA)",
November
2001,
had
proposed
a
drinking
water
value
of
210
ppb.
However,
DEP's
toxicologist,
Dr.
Staats,
expressed
concern
over
some
of
the
assumptions
made
in
the
calculation
of
the
210
ppb
to
DHHR
and
EPA
Region
3.
The
outcome
of
these
discussions
was
a
decision
that
a
very
conservative
approach
should
be
taken
in
the
interim
until
the
CATT
water
SL
was
developed.
Therefore,
14
ppb
was
accepted
as
the
interim
water
SL
for
alternative
water
provision.
Note
that
this
consent
order
was
jointly
signed
by
two
regions
of
EPA
because
West
Virginia
is
in
Region
3
and
Ohio
is
in
Region
5.
During
the
investigation,
C8
had
been
found
in
the
Little
Hocking,
Ohio
public
water
supply.
Also,
note
that
DEP
and
DHHR
invited
Ohio
EPA
to
join
the
CATT
and
GIST
as
observers,
but
not
as
members
because
this
would
have
required
renegotiating
the
consent
order
between
West
Virginia
and
DuPont.

TERA
was
assigned
by
DEP
to
review
and
compile
the
C8
toxicological
information
provided
by
DEP
and
to
prepare
for
and
hold
a
meeting
of
the
CATT
toxicologists
during
which
the
provisional
risk
factors
and
health
protective
SLs
would
be
derived.
The
CATT
toxicologists
panel
was
comprised
of
10
expert
scientists
with
a
collective
span
of
experience
of
over
175
years
and
many
specialties
including
endocrinology,
veterinary
medicine,
cancer,
and
risk
assessment.

TERA's
efforts
are
described
further
in
Section
2.1.
By
mid
April
2002,
TERA
was
prepared
for
the
meeting.
Also,
TERA
helped
prepare
the
other
toxicologists
for
the
meeting
by
providing
toxicity
reports
and
summary
information.
The
CATT
toxicologists
met
on
May
6
and
7,
2002
at
EPA
offices
in
Cincinnati,
Ohio.
The
minutes
of
this
meeting
are
provided
in
Section
2.2.
The
meeting
lasted
approximately
18
hours
with
roughly
one­
third
of
that
time
spent
in
discussions
of
C8'
s
potential
carcinogenicity.
The
oral
provisional
reference
dose
(
pRfD)
risk
factor,
and
the
two
health
protective
SLs
(
for
water
and
soil)
based
on
this
risk
factor
were
developed
at
this
meeting.
The
panel
agreed
that
the
toxicology
database
was
insufficient
to
develop
a
dermal
exposure
pRfD.
The
inhalation
provisional
reference
concentration
(
pRfC)
risk
factor
and
air
SL
developed
at
the
meeting
were
only
interim
because
additional
data
collection
was
necessary
for
their
calculation.
These
data
were
collected
and
provided
to
TERA,
who
calculated
the
final
pRfC
and
air
SL,
wrote
a
report
describing
this
activity
and
forwarded
it
to
the
other
CATT
toxicologists
for
their
approval.
This
document
is
provided
in
Section
2.3
as
the
post
meeting
action
items.
Both
the
meeting
minutes
and
the
post
meeting
action
items
were
reviewed
and
approved
by
the
panel
of
10
highly
qualified
toxicologists.

An
internal
briefing
for
the
DEP,
DHHR,
and
EPA
was
held
on
May
8,
2002
to
discuss
the
water
and
soil
SLs.
Rather
than
withhold
this
information
while
the
meeting
minutes
report
was
prepared,
DEP
released
the
water
and
soil
SLs
so
that
the
public
would
be
informed
of
the
status
of
their
drinking
water,
and
decisions
could
be
made
regarding
the
provision
of
alternative
water
supplies.
In
that
spirit,
DuPont
and
the
public
were
informed
 
via
a
meeting
with
the
above
regulators
and
a
press
release,
respectively
­
of
the
water
and
soil
SLs
on
May
9,
2002.

A
second
public
meeting
was
held
at
Blennerhassett
Junior
High
School
on
May
15,
2002,
to
inform
the
public
of
the
details
of
the
SL
development
and
of
the
groundwater
C8
concentrations
that
had
been
detected
at
that
point.
Dr.
Becker
first
spoke
regarding
environmental
health
risks
in
general.
Dr.
Staats
described
the
process
used
by
the
CATT
toxicologists
to
arrive
at
the
water
and
soil
SLs.
Finally,
David
Watkins
(
DEP,
GIST
chairman)
presented
the
C8
analytical
data
for
private
and
public
water
sources.
Slides
of
the
presentations
given
at
this
meeting
are
provided
in
Attachment
IIb.

8
2.0
DEVELOPMENT
OF
RISK
FACTORS
AND
SCREENING
LEVELS
TERA
was
assigned
to
prepare
for,
host
and
document
the
meeting
of
the
CATT
toxicologists
during
which
the
provisional
C8
risk
factors
(
pRfDs
and
pRfC)
would
be
developed
by
the
group.
The
activities
undertaken
by
TERA
to
prepare
for
the
meeting
are
presented
in
Section
2.1.
The
actual
minutes
of
the
meeting
are
provided
in
Section
2.2.,
and
the
tasks
conducted
by
TERA
to
develop
the
final
air
SL
after
the
meeting
at
the
direction
of
the
panel
are
described
in
Section
2.3.

2.1
Pre
Meeting
Action
Items
TERA
is
a
nonprofit
[
501(
c)(
3)]
corporation
dedicated
to
the
best
use
of
toxicity
data
for
the
development
of
risk
values.
This
organization
is
very
well
known
and
respected
in
the
toxicology
arena
for
their
professionalism,
wealth
of
knowledge,
experience,
and
unbiased
approach
to
deriving
risk
factors.
All
the
non­
TERA
toxicologists
on
the
CATT,
whether
from
government
agencies
or
industry,
were
in
unanimous
support
of
including
TERA
in
this
project.

TERA
was
tasked
with
compiling
and
reviewing
the
available
toxicological
data
for
C8.
A
literature
search
and
review
of
these
data
was
in
draft
by
EPA
Headquarters,
this
document
was
provided
to
TERA.
The
3M
submittals
to
AR­
226
were
provided
to
TERA
by
DEP.
These
data
grew
from
a
total
of
seven
compact
discs
to
10
during
the
time
period
of
this
project.
The
AR­
226
continues
to
grow
with
3M
submittals
currently.
The
index
of
the
first
seven
discs
are
provided
in
Attachment
Va.
Additionally,
DEP
conducted
a
literature
search
of
C8
toxicity
data
on
the
National
Library
of
Medicine's
Medline
and
Toxline
databases
in
June
2001.
The
results
of
these
searches
were
provided
to
TERA
by
DEP
as
well.
Also,
documents
submitted
to
DEP
from
DuPont
in
response
to
the
EPA
Region
3
request
for
information
was
made
available
to
TERA
by
DEP,
first
by
mailing
relevant
toxicology
documents
identified
by
Dr.
Staats,
and
then
by
physically
delivering
all
these
documents
to
their
Cincinnati
office
for
TERA
to
sort
and
identify
those
deemed
relevant
and
necessary
for
their
work.
Therefore,
little
literature
searching
or
data
retrieval
was
required
of
TERA.

After
reviewing
the
existing
C8
toxicology
data,
TERA
selected
studies
that
would
be
suitable
for
derivation
of
risk
factors
for
the
oral,
dermal,
and
inhalation
route
of
exposure.
A
list
of
the
potential
key
studies
was
prepared.
An
indepth
review
of
these
studies
was
then
conducted,
and
the
details
of
the
studies
were
summarized
in
tabular
format.
Next,
TERA
prepared
a
condensed
table
of
these
studies
including
critical
effects
and
exposure
levels
identified
by
TERA,
and
blank
columns
for
the
other
criteria
necessary
in
the
risk
factor
development
process,
such
as
the
uncertainty
factors.
The
documents
listed
below
were
provided
to
the
other
CATT
toxicologists
approximately
two
or
three
weeks
prior
to
the
meeting.
TERA
also
prepared
tables
of
suggested
uncertainty
factors,
risk
factors,
and
resulting
SLs
to
DEP.
These
documents
were
discussed
with
Dr.
Staats
but
were
not
distributed
to
the
other
toxicologists
prior
to
the
meeting
in
an
effort
not
to
influence
their
decisions,
and
not
to
give
the
false
impression
that
the
decisions
on
risk
factor
development
had
already
been
made
and
that
the
panel's
purpose
was
simply
to
review
TERA's
work.
Rather,
TERA's
suggestions
would
be
presented
at
the
meeting
as
a
starting
point
for
panel
discussions
and
the
development
of
the
risk
factors
and
SLs
would
be
done
as
a
group.
The
pre­
meeting
documents
provided
to
the
rest
of
the
panel
by
TERA
and
DEP
are
contained
in
Attachment
III.
Also
in
Attachment
III
is
a
more
detailed
description
of
the
decisions
and
methodology
used
by
TERA
in
suggested
risk
factor
development.

9
2.2
CATT
TOXICOLOGISTS
MEETING
MINUTES
Meeting
of
C8
Assessment
of
Toxicity
Team
(
CATT)
Toxicologists
May
6
and
7,
2002
Andrew
W.
Breidenbach
Environmental
Research
Center,
Cincinnati,
Ohio
Attendees:

Voting
Team
Members
John
Cicmanec,
D.
V.
M.,
M.
S.,
ACLAM,
USEPA
Office
of
Research
and
Development
Joan
Dollarhide,
M.
S.,
M.
T.
S.
C.,
J.
D.,
Toxicology
Excellence
for
Risk
Assessment
(
TERA)
Michael
Dourson,
Ph.
D.,
D.
A.
B.
T.,
TERA
Gerald
Kennedy,
E.
I.
Du
Pont
de
Nemours,
Inc.
Andrew
Maier,
Ph.
D.,
C.
I.
H.,
TERA
Samuel
Rotenberg,
Ph.
D.,
USEPA
Region
3
Jennifer
Seed,
Ph.
D.,
USEPA
Office
of
Pollution
Prevention
and
Toxics
(
may
abstain
from
voting)
Dee
Ann
Staats,
Ph.
D.
(
Chairperson),
West
Virginia
Department
Environmental
Protection
(
DEP)
John
Wheeler,
Ph.
D.,
D.
A.
B.
T.,
Agency
for
Toxic
Substances
Disease
Registry
(
ATSDR)
(
representing
West
Virginia
Department
of
Health
and
Human
Resources
[
DHHR])
John
Whysner,
M.
D.,
Ph.
D.,
D.
A.
B.
T.
(
consulting
for
DuPont)

Invited
Guests
John
Butenhoff,
Ph.
D.,
3M
Company
(
study
director)
Jim
Sferra,
M.
S.,
Ohio
EPA
(
observer)

Note
taker
Daniel
Briggs,
Ph.
D.,
D.
A.
B.
T.,
TERA
Introduction
The
toxicologists
on
the
C8
Assessment
of
Toxicity
Team
(
CATT)
met
on
May
6
and
7,
2002,
to
develop
provisional
reference
doses
(
pRfDs)
and
screening
levels
(
SLs)
for
ammonium
perfluorooctanoate
(
C8)
as
specified
in
Consent
Order
GWR­
2001­
019
between
the
West
Virginia
Department
of
Environmental
Protection,
the
West
Virginia
Department
of
Health
and
Human
Resources,
and
E.
I.
Du
Pont
de
Nemours
&
Co.,
(
DuPont)
dated
November
14,
2001.
These
screening
levels
apply
only
to
DuPont
at
their
West
Virginia
facilities
as
specified
in
this
consent
order.
Any
use
of
these
pRfDs
or
SLs
for
any
other
purpose
or
by
any
other
regulatory
agency
is
solely
their
choice
and
responsibility.

10
The
meeting
opened
with
Dr.
Staats
announcing
that
this
meeting
was
being
held
pursuant
to
the
above­
cited
consent
order
as
part
of
an
enforcement
action
and
was
therefore
closed
to
the
public.
Dr.
Staats
noted
that,
except
for
Dr.
Butenhoff
and
Mr.
Sferra
who
were
invited
guests,
the
panelists
were
named
as
part
of
the
consent
order
and
were
free
to
enter
into
discussions
and
vote
on
issues.
It
was
noted
that
Dr.
Seed
could
abstain
from
voting
at
any
time.
The
rules
for
the
meeting
were
set
forth
as
follows:

 
The
panel
would
strive
for
unanimous
consensus,
but
if
such
consensus
could
not
be
reached,
then
the
majority
of
votes
would
rule.
 
The
panel
was
expected
to
be
cooperative
and
courteous
with
each
other.
 
The
risk
factors
and
screening
levels
would
be
developed
together
as
a
group,
rather
than
simply
by
reviewing
the
work
and
suggestions
of
TERA.
 
Votes
would
be
taken
at
each
decision
point.
After
panel
discussion
on
each
point,
a
motion
would
be
made
on
the
floor.
The
chair
would
then
repeat
the
motion
and
verbally
poll
each
panel
member
individually.
The
chair
would
always
vote
last
in
order
to
not
influence
the
voting.

TERA
recorded
the
official
minutes
for
the
meeting.
However,
the
chair
recorded
supplemental
notes,
which
were
provided
TERA
to
assist
in
the
preparation
of
the
final
Meeting
Minutes
Report.
It
was
noted
that
specific
discussion
comments
or
votes
would
not
be
attributed
to
panel
members
(
i.
e.,
no
names
would
be
used)
in
the
meeting
report
in
order
to
facilitate
full
and
open
discussion
among
the
team.
It
was
also
noted
that
TERA
would
distribute
a
draft
meeting
report
to
the
CATT
panel
for
their
review
and
incorporate
panel
comments
as
appropriate.
Each
panel
member
would
be
asked
to
sign
a
statement
agreeing
that
the
meeting
report
is
an
accurate
representation
of
the
discussion
and
conclusions
of
the
CATT
Team.
The
original
signatures
will
remain
on
file
with
the
DEP.

The
sequence
of
discussion
on
Monday,
May
6
was
oral
noncancer
assessment;
dermal
noncancer
assessment
and
on
Tuesday,
May
7
was
cancer
assessment;
inhalation
noncancer
assessment;
oral
screening
level;
and
interim
inhalation
screening
levels.
(
Note
that
Dr.
Seed
left
the
meeting
at
2:
30
pm
on
Tuesday,
May
7,
2002;
she
was
present
and
joined
in
all
discussions
through
the
cancer
assessment.)
However,
for
clarity,
the
meeting
report
is
organized
according
to
noncancer
(
oral,
dermal,
inhalation)
assessment,
cancer
assessment,
and
screening
levels.
Below,
under
each
heading
is
a
brief
description
of
TERA's
opening
comments,
followed
by
the
panel
discussion,
and
then
the
outcome
of
the
panel
discussion.

Noncancer
Assessment:
Review
of
the
Oral
Studies
Prior
to
the
meeting,
TERA
evaluated
the
available
human
and
animal
health
effects
studies
for
C8.
(
A
list
of
the
documents
and
studies
included
in
TERA's
prior
review
is
provided
in
the
Attachments).
TERA
evaluated
the
pool
of
available
studies
to
identify
the
key
studies
that
could
be
selected
by
the
CATT
panel
as
the
basis
for
the
pRfD.
In
narrowing
the
list
of
available
studies,
the
available
data
were
evaluated
weighing
considerations
such
as
observed
effect
levels,
study
duration
and
quality,
and
applicability
to
human
health.
The
judgments
were
made
in
a
manner
consistent
with
hazard
identification
and
dose­
response
assessment
practices
used
in
current
U.
S.
EPA
risk
assessments.
Studies
were
generally
given
greater
consideration
as
potential
principal
studies
if
they
were
at
least
of
subchronic
duration;
identified
NOAEL/
LOAEL
boundaries
on
the
low
end
of
the
range
provided
by
all
the
data;
and
had
robust
design
(
e.
g.,
diverse
array
of
endpoints,
sufficient
number
of
animals).
From
the
total
pool
of
available
studies,
TERA
developed
detailed
summary
tables
for
each
of
the
key
11
studies
having
potential
for
being
selected
as
the
principal
study
for
derivation
of
the
pRfD.
The
resulting
detailed
summary
table
of
key
studies
was
provided
to
the
panel
members
prior
to
the
meeting
to
facilitate
the
selection
of
the
principal
study
by
the
CATT
panel
and
is
attached.
Therefore,
discussion
of
the
oral
studies
at
the
meeting
focused
on
the
tables
presented
in
the
attachment
which
identified
those
studies
of
sufficient
duration,
content,
and
quality
to
merit
consideration
as
the
bases
for
deriving
a
pRfD.
The
tables
present
TERA's
selection
of
critical
effect
levels,
and
highlight
the
study
data
for
key
parameters
that
showed
treatment­
related
changes.

At
the
opening
of
the
meeting,
the
panel
discussed
whether
all
adequate
studies
had
been
included
and
whether
any
potential
key
studies
were
missing.
One
panelist
asked
why
the
90­
day
Rhesus
monkey
study
(
Goldenthal,
1978b)
had
not
been
included.
TERA
responded
that
the
Rhesus
study
was
not
considered
to
be
as
useful
as
the
cynomolgus
monkey
study
(
Thomford
et
al.,
2001)
because
it
had
fewer
animals
per
group,
and
suggested
a
higher
NOAEL/
LOAEL
boundary;
however,
findings
from
the
Rhesus
study
would
be
discussed
together
with
the
cynomolgus
study
as
supporting
data.
The
panel
confirmed
that,
to
the
best
of
their
knowledge,
the
table
included
all
of
the
toxicity
work
that
should
be
considered
in
selecting
principal
studies
for
deriving
the
pRfD
for
C8.

After
agreeing
that
all
of
the
potential
critical
studies
had
been
identified,
the
panel
then
discussed
the
merits
of
each
of
the
studies,
and
the
appropriate
No­
Observed­
Adverse­
Effect­
Levels
(
NOAELs),
Lowest­
Observed­
Adverse­
Effect­
Levels
(
LOAELs),
and
lower
bounds
on
the
benchmark
doses
(
BMDLs)
for
each
study.

Human
Studies
(
Olsen
et
al.
2000;
Olsen
et
al.
1998;
Gilliland
and
Mandel
1996;
Gilliland
and
Mandel
1993;
Ubel
et
al.
1980)

TERA
initiated
the
discussion
by
providing
a
brief
synopsis
on
the
potential
utility
of
the
available
human
health
effects
studies
for
deriving
the
pRfD.
Two
cohort
mortality
studies
were
available:
(
1)
Ubel
et
al.
(
1980)
reviewed
the
records
of
180
deceased
3M
employees
for
a
period
of
30
years
(
1948­
1978)
and
found
no
significant
difference
between
observed
and
expected
mortality
rates;
(
2)
Gilliland
and
Mandel
(
1993)
found
no
increases
in
mortality
rates
from
liver
cancer
or
liver
disease
in
3,537
(
2,788
males
and
749
females)
exposed
3M
workers
for
35
years
(
1947
 
1983).
Note
that
since
the
CATT
meeting,
a
new
epidemiological
study
on
almost
4,000
(
80%
male)
3M
workers
has
been
completed
which
found
no
increase
incidence
of
cancer
in
C8
exposed
workers.
Several
cross­
sectional
studies
of
3M
workers
(
111,
80,
and
74
males
in
1993,
1995,
and
1997,
respectively)
were
available.
However,
these
studies
were
noted
as
being
limited
for
use
in
deriving
the
pRfD,
since
workers
were
exposed
to
unknown
amounts
of
C8
for
varying
time
periods,
and
no
clear
signs
of
toxicity
(
such
as
elevated
serum
levels
of
liver
enzymes
were
reported).
The
mixed
findings
regarding
changes
in
hormone
levels
were
noted.
It
was
noted
that
many
of
these
studies
provided
data
on
serum
levels
of
C8
(
or
serum
fluorine
levels),
which
could
serve
as
a
measure
of
exposure.
However,
the
current
toxicokinetics
data
were
not
viewed
as
sufficiently
developed
to
conduct
a
quantitative
extrapolation
from
the
reported
serum
levels
to
equivalent
oral
doses
in
humans.
Based
on
this
introduction,
the
panelists
were
asked
to
comment
on
the
human
data
and
its
usefulness
for
deriving
the
pRfD.

Key
Panel
Discussion
Points:
Panelists
noted
that,
although
limited,
the
existing
human
data
are
consistent
with
the
animal
data
when
exposure
levels
are
considered.
Although
weaknesses
in
the
epidemiology
data
were
noted,
one
panel
member
commented
that
the
human
data
are
useful
for
hazard
identification
purposes,
and
provide
some
level
of
comfort
in
conducting
the
assessment
since
they
do
not
identify
adverse
effects
in
chronically
exposed
workers.
It
was
noted
that
a
few
of
the
12
human
subjects
had
C8
serum
levels
comparable
to
those
observed
in
animal
studies
[
20
parts
per
million
(
ppm)
or
greater].
Other
panel
members
described
gaps
in
the
human
studies.
Regarding
the
absence
of
effects
observed
in
the
epidemiology
studies,
the
panel
noted
that
the
small
number
of
female
subjects
and
uncertainties
in
exposure
levels
for
workers
prevents
the
existing
data
from
being
used
to
rule
out
human
toxicity.
For
example,
the
very
small
numbers
of
women
in
the
studies
prevent
drawing
a
conclusion
regarding
female
reproductive
effects.
One
panelist
noted
that
the
increased
blood
level
of
estradiol
reported
in
some
subjects
is
not
clinically
significant.
In
addition,
no
adjustments
were
made
for
body
mass
index
(
BMI)
variations
among
subjects.
Since
BMI
is
known
to
affect
estradiol
levels
and
in
this
study
BMI
was
the
only
parameter
to
correlate
with
hormone
levels,
it
was
noted
that
it
is
unlikely
that
C8
exposure
was
related
to
increased
estradiol
levels.
The
panel
discussed
Gilliland
and
Mandel
(
1986),
which
reported
six
prostate
cancer
deaths
overall
and
four
among
exposed
workers.
One
panel
member
commented
on
the
update
to
this
study
(
no
study
report
was
provided),
which
showed
no
indication
of
increased
risk
of
prostate
cancer.
This
follow
up
study
demonstrated
that
only
one
of
the
four
workers
with
prostrate
cancer
were
determined
to
have
been
exposed
when
work
history
records
and
blood
levels
of
C8
were
examined.

It
was
suggested
that
it
might
be
possible
to
correlate
C8
serum
concentrations
with
lack
of
observed
toxicity
to
estimate
a
human
NOAEL.
However,
it
was
noted
that
the
lack
of
clear
exposure
levels
in
the
human
studies
precluded
this
type
of
analysis.
Although
C8
half­
life
determinations
were
conducted
in
some
of
the
human
studies,
this
information
cannot
be
used
to
determine
exposure
doses
because
some
exposure
to
the
subjects
may
still
be
occurring.
However,
it
is
clear
that
humans
do
not
have
the
major
sex­
related
half­
life
difference
that
exists
in
rats.
It
was
noted
that
a
physiologically­
based
pharmacokinetic
(
PBPK)
model
is
being
developed,
which
may
be
useful
in
estimating
exposure
concentrations
from
human
serum
C8
levels.
However,
a
panel
member
familiar
with
the
status
of
this
current
toxicokinetic
modeling
effort,
noted
that
the
data
are
not
sufficiently
developed
to
use
for
quantitative
risk
assessment
purposes
at
this
time.

Outcome:
The
panel
agreed
unanimously
that
the
human
studies
were
not
adequate
to
be
used
for
quantitative
dose­
response
determinations.
The
human
studies
have
many
substantial
data
gaps,
such
as
low
numbers
of
subjects
and
unknown
exposure
concentrations.
No
LOAEL
was
established
and
the
exposure
uncertainty
does
not
allow
identification
of
a
clear
NOAEL.
In
final
comments
made
during
polling
of
the
panel,
one
panel
member
agreed
with
the
group,
but
noted
that
the
data
could
be
used
to
develop
a
bounding
estimate.
A
second
panel
member
added
that
some
evidence
suggests
the
endocrine
system
as
a
target
for
C8
effects,
and
therefore,
the
human
data
might
support
the
animal
toxicity
studies.

Definition
of
Adverse
Liver
Effect
TERA
noted
that
in
all
experimental
animal
studies
liver
effects
occurred.
For
the
purposes
of
conducting
this
assessment,
TERA
defined
adverse
liver
effects
as
the
presence
of
histopathology
(
moderate
grade
hypertrophy
would
be
considered
sufficient)
in
addition
to
statistically
significant
absolute
or
relative
weight
changes,
or
a
liver
weight
change
of
10%
or
greater.
A
doubling
of
serum
levels
of
liver
enzyme
activity
(
e.
g.,
alkaline
phosphatase
(
ALP),
aspartate
aminotransferase
(
AST),
or
alanine
aminotransferase
(
ALT))
would
also
indicate
an
adverse
liver
effect.
These
adverse
effects
are
used
by
other
health
organizations
as
well.
The
panel
unanimously
agreed
with
this
general
definition
of
adverse
for
liver
effects,
but
noted
that
individual
studies
could
demonstrate
a
continuum
of
liver
effects
that
could
be
considered
biologically
significant.

13
Palazzolo
et
al.
1993
This
is
a
90­
day
study
in
male
rats
in
which
animals
received
C8
at
doses
of
0,
0.05,
0.47,
1.44,
and
4.97
mg/
kg­
day
in
feed.
The
major
finding
in
this
study
was
increased
liver
weight
with
histopathological
findings
such
as
moderate
hypertrophy.
Panelists
were
asked
to
comment
on
the
data
from
this
study;
on
the
selection
of
study
adverse
effect
levels;
and
on
the
usefulness
of
this
study
as
the
basis
for
deriving
a
pRfD.

Key
Panel
Discussion
Points:
The
possible
role
of
peroxisome
proliferation
in
the
observed
liver
effects
was
discussed.
The
panel
discussed
uncertainty
in
the
relevance
of
this
mechanism
to
humans.
One
panelist
stated
that
when
considering
the
relevance
of
peroxisome
proliferation,
it
is
important
to
consider
both
qualitative
and
quantitative
issues.
This
panelist
suggested
that
peroxisome
proliferation
may
potentially
occur
in
humans
because
the
cellular
receptor
that
modulates
this
reaction
in
rodents
has
been
found
in
humans,
but
that
this
mode
of
action
should
be
considered
to
be
only
qualitatively
relevant
to
humans
because
the
receptor
is
far
less
expressed
in
humans,
and
humans
have
not
been
shown
to
manifest
a
peroxisome
proliferation
response.
It
was
noted
that
USEPA
has
an
on­
going
project
to
investigate
the
relevance
to
humans
of
rodent
peroxisome
proliferation
effects,
but
at
this
time
EPA
has
no
official
policy
on
the
significance
of
peroxisome
proliferation
for
humans.
It
was
also
noted
that
IARC
has
also
considered
the
issue
of
peroxisome
proliferation
and
concluded
that
this
mode
of
action
is
not
relevant
to
humans
if
it
has
not
been
demonstrated
to
occur
in
human
cells
or
primates
treated
with
the
chemical
in
question.
(
Note
that
the
panel
discussed
the
role
of
peroxisome
proliferation
as
a
potential
mode
of
action
for
tumor
formation
later
in
the
meeting.
The
results
of
this
discussion
are
documented
in
the
section
on
Cancer
Mode
of
Action)

Discussion
occurred
regarding
the
usefulness
of
relative
versus
absolute
liver
weight
in
determining
adverse
effect
levels.
One
panelist
stated
that
changes
in
both
of
these
parameters
are
preferred
before
designating
a
dose
as
an
adverse
effect
level.
However,
most
panelists
considered
a
change
in
relative
liver
weight
to
be
sufficient
to
designate
a
dose
level
as
an
adverse
effect
level.
It
was
noted
that
liver
weights
in
dosed
animals
in
this
study
were
comparable
to
control
values
after
an
8­
week
recovery
period;
however,
the
panel
agreed
that
this
recovery
should
not
influence
selection
of
the
NOAEL
and
LOAEL
values.

Outcome:
The
panel
agreed
unanimously
that
1.44
mg/
kg­
day
is
the
LOAEL
for
this
study
because
at
this
level
statistically­
significant
increases
in
relative
liver
weight
and
CoA
oxidase
activity
occur.
In
addition,
hepatocellular
hypertrophy
of
minimal
severity
or
greater
is
observed
in
14
of
15
animals
at
this
dose,
and
in
2
of
15
animals
at
grade
2
or
higher.
The
panel
recommended
that
benchmark
dose
modeling
be
performed
for
the
data
based
on
grade
2
or
higher
hepatocyte
hypertrophy.
This
modeling
was
conducted
during
the
course
of
the
meeting,
resulting
in
a
BMDL
estimate
of
1.3
mg/
kg­
day.
It
was
noted
that
this
BMDL
is
essentially
the
same
as
the
LOAEL
found
in
this
study.
Most
panelists
believed
0.47
mg/
kg­
day
is
the
NOAEL
because
at
this
dose
there
are
no
statistically
significant
changes
in
either
absolute
or
relative
liver
weight
and
only
a
"
minimal"
severity
of
hepatocellular
hypertrophy
is
reported
at
this
dose.
However,
one
panel
member
preferred
to
call
this
a
"
minimal
LOAEL"
rather
than
a
NOAEL,
noting
that
dose­
related
changes
in
critical
liver
parameters
had
been
established
at
the
lower
dose
levels
and
suggesting
that
these
could
be
part
of
the
continuum
of
effects
that
might
be
considered
a
minimal
LOAEL.

14
Goldenthal
1978a
This
is
a
90­
day
study
in
male
and
female
rats
in
which
animals
received
C8
in
their
feed
at
doses
of
0,
0.56,
1.72,
5.64,
17.9,
or
63.5
mg/
kg­
day
for
males
and
0,
0.74,
2.3,
7.7,
22.4,
or
76.5
mg/
kg­
day
for
females.
This
study
is
limited
by
the
small
number
of
animals
(
5/
sex)
in
each
dose
group.
Therefore,
this
study
was
not
considered
to
be
a
key
study.
However,
it
was
presented
for
the
panel's
consideration
and
comments
because
it
includes
female
as
well
as
male
animals
and
the
data
on
relative
liver
weights
allow
a
BMD
to
be
calculated.

Key
Panel
Discussion
Points:
One
panelist
noted
that
a
sex
difference
was
observed
in
this
study.
Another
mentioned
that
this
study
demonstrates
the
importance
of
internal
dose
(
C8
serum
level),
as
compared
to
the
administered
dose.

Outcome:
The
panel
agreed
with
the
proposed
NOAEL,
LOAEL,
and
BMDL
as
presented
by
TERA.
However,
the
panel
also
agreed
unanimously
that
the
study
was
not
adequate
to
serve
as
the
basis
for
deriving
a
pRfD
because
of
limitations
in
the
study
(
e.
g.,
the
small
number
of
animals).

York
2002
This
is
a
two­
generation
reproduction
study
in
which
male
and
female
rats
received
C8
doses
of
0,
1,
3,
10,
and
30
mg/
kg­
day
by
gavage
in
distilled
water.
Parental
animals
were
exposed
through
cohabitation
and
gestation
to
weaning
of
F1
animals,
approximately
6
weeks.
F1
animals
were
exposed
from
weaning
until
weaning
of
the
F2
generation.
The
primary
findings
were
increased
liver
weight
and
liver
pathology
in
P
and
F1
generation
male
animals;
however,
it
was
noted
that
histology
was
conducted
only
when
gross
effects
had
been
observed,
and
therefore
liver
histopathology
data
were
not
available
for
the
control
and
low­
dose
F1
generation
males.

Key
Panel
Discussion
Points:
One
panelist
stated
that
this
was
study
was
of
excellent
quality
because
it
was
conducted
according
to
OPPTS
guidelines
for
2­
generation
studies.
Two
panelists
noted
that
the
degree
of
F1
generation
exposure
to
C8
while
in
utero
and
while
nursing
was
uncertain
and
may
not
have
occurred
at
all
because
of
rapid
elimination
of
C8
from
the
systemic
circulation
of
the
female
rats
after
it
was
administered
via
gavage.
Therefore,
the
lack
of
reproductive
toxicity
in
this
study
may
not
be
meaningful.
Other
panelists
agreed,
but
stated
that
the
fact
of
rapid
clearance
resulting
in
decreased
fetal
exposure
may
not
be
relevant
for
humans
because
women
do
not
have
the
same
active
secretory
mechanism
for
C8
that
exists
in
the
female
rat.
Another
panelist
noted
that
rodent
placenta
provides
less
of
an
anatomical
barrier
than
exists
in
primates.
Another
panelist
observed
that
studies
with
radiolabeled
C8
demonstrated
that
C8
could
cross
the
placental
barrier
in
rats.
One
panelist
wondered
whether
female
rat
pups
at
weaning
have
developed
the
active
secretory
mechanism
for
C8
that
exists
in
the
mature
females.
Another
panelist
recalled
data
showing
that
weanling
female
rats
were
able
to
clear
C8
faster
than
males,
but
not
as
fast
as
mature
females.
One
panelist
recommended
that
delayed
sexual
maturation
and
increased
frequency
of
estrous
cycles
be
included
in
the
adverse
effects
noted
for
females
for
this
study.
A
panelist
pointed
out
that
this
study
indicated
a
critical
difference
in
the
toxicity
of
C8
versus
the
structurally
similar
perfluorocarbon
PFOS;
in
that
PFOS
caused
fetal
death
at
birth
in
a
similarly
designed
study,
while
in
this
study
C8
administration
was
associated
with
only
a
slightly
statistically
significant
increase
in
fetal
death
at
the
post­
weaning
timeframe.

Outcome:
The
panel
concluded
that
the
LOAEL
for
males
is
1
mg/
kg­
day.
The
males
showed
statistically­
significant
increases
in
liver
and
kidney
weights
at
1
mg/
kg­
day.
No
histology
was
conducted
on
liver
and
kidney
at
this
dose
level
because
no
gross
lesions
were
seen.
However,
given
15
the
substantial
histopathology
noted
at
the
next
higher
dose
level
(
3
mg/
kg­
day),
the
panel
believed
pathology
does
exist
at
the
1
mg/
kg­
day
level;
therefore
this
level
meets
the
agreed­
upon
definition
of
an
adverse
effect.
The
panel
concluded
that
the
LOAEL
for
females
is
30
mg/
kg­
day.
The
females
showed
several
adverse
effects
at
this
dose
level,
including
increased
mortality
and
decreased
body
weight.
No
NOAEL
was
identified
for
males;
the
NOAEL
for
females
is
10
mg/
kg­
day.
All
of
these
values
apply
to
both
the
P
and
F1
generation
animals.
Two
panel
members
reviewed
the
BMDL
modeling
results,
and
agreed
with
the
selection
of
0.42
mg/
kg­
day
as
the
study
BMDL.

Riker
Laboratories
1983
This
is
a
chronic,
2­
year
study
in
male
and
female
rats
in
which
animals
received
C8
in
feed
at
doses
of
0,
1.3,
and
14
mg/
kg­
day
for
males
and
0,
1.6,
and
16
mg/
kg­
day
for
females.
The
primary
findings
in
this
study
are
liver
effects
in
male
rats.
However,
it
was
noted
that
this
chronic
study
also
reported
non­
hepatic
effects
(
ovarian
stromal
hyperplasia
and
ataxia)
in
female
rats.
Although
this
effect
was
not
found
in
the
subchronic
study
that
included
females
(
Goldenthal,
1978),
the
small
number
of
animals
in
that
subchronic
study
(
n=
5)
may
have
limited
the
power
of
the
study
to
observe
these
effects.

Key
Panel
Discussion
Points:
One
of
the
panelists
identified
some
copying
errors
in
the
tables
(
incidences
of
mammary
fibroadenomas,
Leydig
cell
adenomas,
and
ALT
activity
in
the
control
group)
and
these
values
were
corrected
prior
to
the
panel
discussion
(
the
attached
table
presents
the
corrected
values).
The
panel
disagreed
with
the
study
author's
conclusion
stated
in
the
study
report
that
the
testicular
vascular
mineralization
was
a
"
spontaneous
change
occurring
in
aging
rats"
and
that
the
ovarian
stromal
tubular
hyperplasia
was
"
equivocally
related"
to
C8
administration
because
it
did
not
progress.
The
panel
considered
both
these
effects
to
be
biologically
significant
and
relevant
for
determining
adverse
effect
levels.
One
panelist
stated
that
ovarian
stromal
hyperplasia
is
not
commonly
found
in
rats
and
noted
that
in
this
study
the
incidence
of
ovarian
stromal
hyperplasia
in
the
control
animals
is
zero.
The
panel
discussed
the
relevance
of
the
ataxia
observed
in
females,
but
did
not
reach
any
conclusions
about
its
possible
biological
significance.
One
panelist
noted
that
at
the
time
this
study
was
conducted,
the
term
"
hepatic
megalocytosis"
was
synonymous
with
the
term
"
hepatic
hypertrophy"
currently
in
use.
It
was
noted
that
the
BMDL
of
0.73
mg/
kg­
day
calculated
based
on
liver
effects
in
males
is
consistent
with
the
NOAELs
for
liver
effects
observed
in
other
rat
studies.
In
the
initial
summary
table
from
which
the
panel
was
working
it
was
noted
that
no
BMDL
was
estimated
for
ovarian
stromal
tubular
hyperplasia,
since
an
adequate
fit
to
the
data
was
not
achieved.
One
reviewer
suggested
that
a
model
fit
might
be
possible
using
log­
transformed
data,
since
the
study
results
showed
a
clear
log­
related
response
curve.
This
approach
was
applied
during
the
meeting,
and
resulted
in
a
best
estimate
of
the
BMDL
of
1.6
mg/
kg/
day.

Outcome:
The
panel
agreed
unanimously
to
the
proposed
NOAEL
of
1.3
mg/
kg­
day
for
males,
with
a
corresponding
LOAEL
of
14
mg/
kg­
day
based
on
the
following
adverse
effects:
increased
liver
weight,
hepatic
cystoid
degeneration,
increased
ALT
enzyme
activity,
and
testicular
vascular
mineralization.
The
panel
agreed
that
the
LOAEL
in
females
was
1.6
mg/
kg­
day
based
on
a
statistically
significant
increase
in
the
incidence
of
ovarian
stromal
tubular
hyperplasia,
and
that
this
study
did
not
identify
a
NOAEL
for
females.
The
panel
further
agreed
that
the
estimated
BMDL
from
this
study
is
0.73
mg/
kg­
day
based
on
liver
effects
in
males
as
the
benchmark
response.

16
Thomford
et
al.,
2001
This
is
a
26­
week
study
in
cynomolgus
monkeys,
in
which
animals
received
C8
at
doses
of
0,
3,
10,
or
30/
20
mg/
kg­
day
by
gastric
intubation
of
gelatin
capsule.
Gastric
capsule
intubation
was
chosen
as
the
method
of
C8
administration
to
avoid
emesis,
which
had
occurred
in
the
earlier
Rhesus
monkey
study
(
Goldenthal
et
al.,
1978b).
Even
so,
several
animals
had
problems
tolerating
the
highest
C8
dosing;
as
a
result,
the
high
dose
was
either
reduced
or
in
some
cases,
discontinued.
Afterwards,
time­
weighted
average
doses
were
used
to
approximate
the
C8
dose
given
to
the
high­
dose
group.
One
animal
died
in
the
high
dose
group;
primary
findings
included
clinical
signs
and
altered
liver
weight.
TERA
presented
that
altered
liver
weight
was
not
considered
an
adverse
finding.

Key
Panel
Discussion
Points:
At
least
two
panelists
believed
that
the
degree
of
absolute
liver
weight
increase
(
30%)
noted
at
the
3
mg/
kg­
day
dose
should
be
sufficient
to
identify
this
dose
as
the
LOAEL.
Other
panelists
responded
that
this
weight
increase
resulted
from
mitochondrial
proliferation,
and
therefore
was
an
adaptive
response,
not
an
adverse
effect.
They
also
pointed
out
that,
unlike
laboratory
rodents,
cynomolgus
monkeys
routinely
exhibit
large
genetic
variations.
As
a
result,
large
differences
in
organ
weights
among
these
animals
is
relatively
common
and
a
30%
difference
between
groups
 
especially
small
groups,
as
in
this
study
 
is
not
necessarily
biologically
meaningful.
Some
panelists
attempted
to
compare
this
study
with
the
study
conducted
in
Rhesus
monkeys
in
order
to
help
define
the
LOAEL,
but
this
was
not
possible
due
to
the
uncertainty
of
dosing
caused
by
the
emesis
that
occurred
in
the
Rhesus
study.
One
panelist
asked
if
the
dosing
technique
(
gastric
intubation
of
the
drug
contained
in
gelatin
capsules)
might
have
contributed
to
a
large
range
of
C8
blood
levels
because
of
differences
in
capsule
disintegration
rates.
Another
panelist
responded
that
this
was
unlikely
because,
while
the
data
sometimes
demonstrated
large
inter­
animal
variations
in
blood
levels,
the
intra­
animal
variation
over
several
dose
administrations
was
small.
It
was
noted
that
C8
serum
levels
were
essentially
the
same
in
the
low
and
mid­
dose
groups:
74,
80,
and
120
µ
g/
mL
at
3,
10,
and
30/
20
mg/
kg­
day,
respectively.
The
panel
concluded
that
the
similarities
in
serum
C8
levels
may
explain
the
very
similar
effects
observed
between
the
3
and
10
mg/
kg­
day
dose
groups.
One
panelist
noted
that
protein­
binding
saturation
was
similar
between
the
monkey
and
human.

Outcome:
The
panel
agreed
that
the
LOAEL
is
best
described
as
"
from
3
to
10
mg/
kg­
day"
based
on
30%
increased
absolute
liver
weight,
and
that
a
NOAEL
does
not
exist
for
this
study.
At
all
three
dose
levels,
statistically
significant
increases
in
absolute
and
relative
liver
weights
occurred,
but
without
accompanying
histopathology.
No
clinical
or
histopathological
evidence
of
organ
damage
occurred
at
any
of
the
three
dose
levels.
Dose­
related
trends
toward
lower
T3
and
T4
levels
were
observed,
but
these
failed
to
achieve
statistical
significance,
even
at
the
highest
dose.
The
panel
concluded
that
these
data
are
insufficient
to
identify
any
single
dose
as
a
LOAEL
or
NOAEL.
Since
the
serum
C8
levels
were
essentially
the
same
for
both
the
3
and
10
mg/
kg­
day
doses,
the
panel
believed
that
designating
a
range
of
3
to
10
mg/
kg­
day
for
the
LOAEL
is
the
best
way
to
describe
the
study
results.

Noncancer
Assessment:
Oral
Hazard
and
Dose­
Response
Characterization
(
Note:
Dr.
Seed
abstained
from
voting
during
this
part
of
the
meeting.)

Critical
Study
and
Point­
of­
Departure
The
summary
of
NOAELs,
LOAELs,
and
BMDLs
unanimously
agreed
to
by
the
panel
is
presented
in
Table
1
below.
The
individual
study
adverse
effect
levels
were
discussed
by
the
panel
for
the
purpose
of
selecting
a
critical
study
and
effect
level
for
derivation
of
the
pRfD.

17
Key
Panel
Discussion
Points:
The
primary
target
organ
for
C8
is
the
liver,
and
males
are
clearly
more
sensitive
to
this
effect
than
female
rats.
One
panelist
observed
that
the
liver
effects
in
rats
may
be
related
to
peroxisome
proliferation,
and
therefore
may
not
be
quantitatively
relevant
for
humans.
For
this
reason,
the
liver
effects
in
rats
might
not
be
an
appropriate
critical
endpoint
Another
panelist
responded
that,
because
of
this,
it
was
important
to
note
that
the
monkey
and
rat
LOAELs
are
in
the
same
range,
and
since
the
liver
effects
in
monkeys
may
not
be
related
to
peroxisome
proliferation,
liver
toxicity
might
also
be
a
relevant
endpoint
for
humans.
The
observation
of
ovarian
effects
in
female
rats
at
the
same
LOAEL
as
liver
effects
in
males
was
noted
as
a
second
reason
to
consider
the
rodent
studies
as
an
appropriate
basis
for
deriving
the
pRfD.

Table
1.
Summary
of
NOAELs,
LOAELs,
BMDLs,
and
Critical
Effects
for
Key
and
Supporting
C8
Studies
Species
Sex
NOAEL
LOAEL
BMDL
Critical
Effect
Key
Studies
Palazzolo
et
al.
(
1993)
Rat
M
0.47
1.44
1.3
Liver
York
et
al.
(
2002)
Rat
M
None
1
0.42
Liver
Riker
Laboratories
(
1983)
Rat
F
None
1.6
1.6
Ovary
M
1.3
14
0.73
Liver
Thomford
et
al.
(
2001)
Monkey
M
None
3­
10
None
Liver
Supporting
Studies
Goldenthal
et
al.
(
1987a)
Rat
M
0.56
1.72
0.44
Liver
Goldenthal
et
al.
(
1987b)
Monkey
M,
F
3
10
Not
done
Clinical
signs
Some
panelists
favored
choosing
the
monkey
study
as
the
critical
study,
due
to
the
closer
biological
relationship
with
humans
as
opposed
to
rats.
It
was
also
noted
that
the
observed
increase
in
liver
weight
in
monkeys
may
not
be
related
to
peroxisome
proliferation
and,
therefore,
may
be
more
relevant
for
human
health
risk
assessment.
Other
panelists
disagreed,
pointing
to
the
uncertainties
in
dosing
and
effects,
the
small
number
of
animals
per
dose
group,
and
the
unclear
boundary
between
NOAEL
and
LOAEL
values.
Also,
it
was
noted
that
the
monkey
study
could
not
be
considered
the
critical
study
because
the
90­
day,
two­
generation,
and
two­
year
rat
studies
all
have
LOAEL,
NOAEL,
and
/
or
BMDLs
below
the
LOAEL
range
identified
in
the
monkey
study,
and
therefore
based
on
selection
of
the
critical
study
with
the
lowest
adequate
NOAEL/
LOAEL
boundary
would
support
the
use
of
the
rodent
studies.

The
panel
considered
whether
it
would
be
better
to
base
the
pRfD
on
a
NOAEL
or
on
a
BMDL.
Some
panelists
thought
a
NOAEL
basis
is
a
simpler
concept
and
would
be
easier
to
explain
to
the
public.
Others
responded
that
the
BMDL
captures
more
information
from
the
entire
study
(
e.
g.,
reflects
information
from
the
full
dose­
response
curve,
and
variability
in
the
dose­
response
data)
and
therefore
is
the
better
choice
as
the
basis
for
the
quantitative
dose­
response
assessment.
Another
panel
member
mentioned
that
a
NOAEL
is
not
a
"
no
effect"
level,
rather
it
reflects
the
proportion
of
the
responding
population
that
can
physically
be
observed
in
an
experimental
situation.
Therefore,
the
size
of
the
population
is
important.
The
panel
agreed
to
not
rule
out
using
either
a
NOAEL
or
BMDL,
but
instead
to
focus
on
the
quality
of
each
study
and
the
lowest
critical
effect
level
it
provided.

18
The
panel
noted
the
unusually
good
agreement
of
the
NOAELs
and
LOAELs
from
all
the
studies.
The
lowest
NOAEL
observed
in
one
of
the
potential
key
studies
was
0.47
mg/
kg­
day,
from
the
90­
day
rat
study
by
Palazzolo
et
al.
(
1993).
The
lowest
LOAEL
observed
in
a
key
study
was
1
mg/
kg­
day
from
the
rat
two­
generation
study
(
York
et
al.,
2002).
This
study
did
not
test
doses
low
enough
to
identify
a
NOAEL;
however,
the
BMDL
value
estimated
for
this
study,
0.42
mg/
kg­
day,
was
essentially
the
same
as
the
observed
NOAEL
from
the
90­
day
study.
Therefore,
the
panel
agreed
that
the
BMDL
was
an
appropriate
NOAEL
surrogate
for
the
two­
generation
study.
The
ovarian
stromal
hyperplasia
reported
in
the
chronic
rat
study
(
Riker
Laboratories,
1983),
provided
a
higher
LOAEL
than
the
two­
generation
study,
and
the
BMDL
for
this
effect
resulted
in
the
same
value
as
the
LOAEL.
This
demonstrates
that
the
liver
endpoint
is
the
critical
effect,
because
it
occurs
at
lower
doses.

Outcome:
Because
of
the
consistency
in
NOAELs/
LOAELs
and
critical
effect
in
all
the
key
studies,
the
panel
concluded
that
all
studies
could
be
considered
co­
critical
studies
and
that
all
provide
important
information
for
human
risk
assessment.
However,
the
panel
unanimously
agreed
that
the
NOAEL
surrogate
from
the
two­
generation
study,
a
BMDL
of
0.42
mg/
kg­
day,
should
serve
as
the
point­
of­
departure
for
the
pRfD.
This
value
was
selected
since
it
represented
the
lowest
NOAEL
or
BMDL,
and
provided
the
added
consideration
of
having
evaluated
reproductive
and
developmental
effects.

Uncertainty
Factors
If
adequate
human
data
are
available,
these
data
are
used
as
the
basis
for
noncancer
risk
factor
development.
Otherwise,
animal
study
data
are
used,
along
with
a
series
of
professional
judgments
that
are
incorporated
into
the
risk
factor
as
"
Uncertainty
Factors"
and
account
for
an
assessment
of
the
relevance
and
scientific
quality
of
the
experimental
studies.
There
are
five
different
uncertainty
factors
commonly
used
to
address
issues
of
biological
variability
and
uncertainty.
Two
factors
(
Interspecies
and
Intraspecies)
are
used
to
address
variability
or
heterogeneity
that
exists
between
animals
and
humans,
and
within
different
human
populations.
Three
factors
(
Subchronic,
LOAEL,
Database)
are
used
to
address
lack
of
information.
Typically,
the
maximum
total
uncertainty
factor
that
EPA
will
apply
is
3000.
If
all
five
areas
of
uncertainty/
variability
are
present
warranting
a
total
UF
of
10,000,
then
EPA
generally
concludes
that
the
uncertainty
is
too
great
to
develop
an
RfD.
The
panel
discussed
each
area
of
variability
or
uncertainty
separately.
A
short
introduction
to
each
area
of
uncertainty
is
provided
below
to
aid
the
reader
in
evaluating
the
discussions
of
the
panel.

Intraspecies
Variability
(
UFH):
This
factor
accounts
for
the
natural
differences
that
occur
between
human
subpopulations
and
for
the
fact
that
some
individuals
may
be
more
sensitive
than
the
average
population.
This
factor
is
composed
of
two
subfactors
 
one
to
account
for
toxicokinetic
differences
(
how
the
body
distributes
and
metabolizes
the
chemical)
and
one
to
account
for
toxicodynamic
differences
(
how
the
body
responds
to
the
chemical).
If
no
information
is
available
on
human
variability,
then
a
default
value
of
10
is
used.
If
adequate
information
is
available
on
one
of
the
two
subcomponents,
then
this
information
is
used
along
with
a
default
value
of
3
for
the
remaining
subfactor.
If
data
are
available
to
adequately
describe
human
variability
in
both
subfactors,
then
actual
data
may
be
used
to
replace
default
values.
In
addition,
if
a
RfD
is
based
on
human
data
gathered
in
the
known
sensitive
subpopulation,
a
value
of
1
may
be
chosen
for
this
factor.

19
The
panel
discussed
the
lack
of
available
data
describing
human
variability.
One
panelist
suggested
a
comparison
of
human
C8
blood
levels
and
values
from
the
animal
studies.
The
highest
human
serum
C8
level
reported
was
111
ppm,
but
the
average
was
approximately
5
ppm.
No
effects
were
noted
in
the
human
subject
with
the
highest
blood
level.
Thus,
at
least
some
people
achieved
serum
C8
levels
equivalent
to
those
that
resulted
in
adverse
effects
in
animal
studies.

As
noted
in
the
discussion
of
the
human
data
above,
the
panel
acknowledged
gaps
in
the
data
on
human
variability
and
inability
to
define
the
most
sensitive
subpopulation,
and
therefore
concluded
that
the
default
value
of
10
was
appropriate
for
this
factor.

Interspecies
Variability
(
UFA):
This
factor
accounts
for
the
differences
that
occur
between
animals
and
humans
and
is
also
thought
to
be
composed
of
subfactors
for
toxicokinetics
and
toxicodynamics.
If
no
information
is
available
on
the
quantitative
differences
between
animals
and
humans,
then
a
default
value
of
10
is
used.
If
information
is
available
on
one
of
the
two
subcomponents,
then
this
information
is
used
along
with
a
default
value
of
3
for
the
remaining
subfactor.
If
data
are
available
to
adequately
describe
variability
in
both
subfactors,
then
actual
data
may
be
used
to
replace
default
values.
In
addition,
if
a
RfD
is
based
on
human
data,
then
a
value
of
1
is
appropriate
for
this
factor.

One
panelist
mentioned
that
EPA
has
often
used
a
UFA
value
of
3
in
other
assessments
when
extrapolating
monkey
data
to
humans,
because
the
kinetics
and
dynamics
of
monkeys
are
assumed
to
be
similar
to
humans.
This
assumption
is
based
on
the
fact
that
rhesus
monkeys
and
macaques
share
a
92%
genetic
homology
with
humans
and
because
monkey
studies
are
able
to
detect
a
much
broader
range
of
clinical
findings
and
more
specific
histopathology
than
rodents.
In
addition,
studies
on
other
chemicals
in
which
a
good
database
exists
in
rodents,
monkeys
and
humans
demonstrate
that
results
in
monkey
studies
parallel
the
human
effects
more
closely
than
results
in
rodent
studies.

Another
panelist
agreed
and
said
the
half­
life
of
chemicals
in
monkeys
was
usually
closer
to
humans
than
to
rats.
Other
panelists
responded
that
for
C8,
the
half­
life
in
monkeys
is
about
30
days;
and
this
is
much
less
than
the
C8
half­
life
in
humans,
which
is
estimated
to
be
greater
than
one
year.
It
was
noted,
however,
that
data
on
C8
half­
life
in
humans
is
limited.

Because
no
data
are
available
to
warrant
moving
from
the
default,
the
panel
unanimously
agreed
that
a
UFA
value
of
10
is
appropriate
with
either
the
rat
or
monkey
toxicology
studies.

Subchronic
to
Chronic
Extrapolation
(
UFS):
Because
the
RfD
protects
for
a
lifetime
exposure,
this
factor
is
applied
when
the
database
lacks
information
on
the
health
effects
of
the
chemical
following
a
chronic
exposure.
Two
issues
are
considered
when
making
judgment
on
the
use
of
this
factor
 
are
there
data
demonstrating
that
different
health
effects
are
expected
following
chronic
exposure
than
subchronic
exposure,
and
are
there
data
demonstrating
that
the
observed
health
effects
progress
in
severity
as
exposure
duration
increases?
If
the
database
contains
no
information
on
chronic
exposure,
a
default
value
of
10
is
often
applied,
unless
other
data
suggest
a
lack
of
progression
with
exposure
duration.
If
the
database
contains
adequate
chronic
bioassays,
then
a
value
of
1
is
appropriate.
If
there
are
data
addressing
only
one
of
the
two
issues,
then
a
default
of
3
may
be
applied.

It
was
noted
that
the
database
for
C8
contains
an
adequate
chronic
rat
study
(
Riker
Laboratories,
1983).
In
addition,
a
second
chronic
study
(
Biegel
et
al.,
2001)
was
available,
although
this
study
focused
primarily
on
tumorigenic
mechanisms
in
rats.
In
addition,
for
the
purpose
of
evaluating
uncertainty
factors,
the
human
occupational
studies
were
considered
by
the
panel
to
be
informative
on
the
response
(
or
lack
thereof)
of
humans
following
long­
term
exposure.
The
database
demonstrates
that
liver
20
toxicity
was
the
more
sensitive
endpoint
in
both
subchronic
and
chronic
studies.
In
addition,
the
database
clearly
demonstrates
that
liver
toxicity
does
not
progress
in
severity
following
chronic
exposure.
This
conclusion
is
supported
by
the
observation
that
the
subchronic
studies
identified
lower
NOAELs
for
liver
toxicity
than
the
chronic
studies.
One
panelist
noted
that
the
liver
effect
in
rat
progresses
to
cancer.
However
the
panel
concluded
that
the
cancer
effect
was
due
to
the
peroxisome
proliferation
mechanism
(
as
discussed
below
in
the
discussion
of
the
cancer
risk
assessment).
Based
on
these
considerations,
the
panel
unanimously
agreed
that
a
UFS
value
of
1
is
appropriate
for
the
rat
studies.

The
panel
also
discussed
whether
a
different
value
for
UFS
would
be
appropriate
if
the
monkey
study
had
been
used
as
the
critical
or
co­
critical
study.
One
panelist
observed
that
there
were
no
data
in
monkeys
regarding
the
progression
beyond
26
weeks;
another
responded
that
there
was
no
reason
to
think
the
effects
in
monkeys
would
be
any
more
progressive
than
those
in
rats.
Another
panelist
suggested
that
the
toxicity
of
C8
in
humans
does
not
appear
to
be
progressive.
However,
the
panel
agreed
that
there
was
some
inherent
uncertainty
in
the
monkey
study
to
justify
use
of
the
value
of
3
for
UFS
if
the
monkey
study
were
the
critical
study.

LOAEL
to
NOAEL
Extrapolation
(
UFL):
Because
the
RfD
is
considered
to
be
a
subthreshold
value
that
protects
against
any
adverse
health
effects,
this
factor
is
applied
when
the
database
lacks
information
to
identify
a
NOAEL.
If
the
database
does
not
identify
a
NOAEL,
then
a
default
of
10
is
used
for
this
factor.
If
a
NOAEL
is
used,
a
value
of
1
is
appropriate.
Often,
if
the
database
does
not
identify
a
NOAEL,
but
the
adverse
effects
observed
are
of
minimal
severity,
then
a
default
of
3
will
be
considered
appropriate
for
use
of
a
"
minimal
LOAEL".
1
Several
of
the
studies
considered
as
co­
critical
identified
NOAELs;
the
lowest
NOAEL
is
0.47
mg/
kg­
day
from
the
90­
day
study.
Also,
the
BMDL
estimated
for
the
two­
generation
study
was
essentially
the
same
as
the
observed
NOAEL
from
the
90­
day
study.
These
NOAELs
and
BMDLs
were
based
on
well­
conducted
studies
and
their
use
as
a
basis
of
the
pRfD
is
consistent
with
standard
practice.
Therefore,
the
panel
had
confidence
that
the
C8
database
has
identified
the
threshold
for
toxicity
in
rats,
and
it
unanimously
agreed
a
UFL
value
of
1
is
appropriate
for
the
critical
effect
in
the
rat
studies.

The
panel
also
considered
the
value
of
UFL
that
would
be
appropriate
if
the
monkey
study
were
to
be
used
as
the
critical
study.
Because
there
is
no
clear
NOAEL
value,
the
panel
agreed
that
a
value
of
1
was
not
appropriate.
However,
because
the
effects
seen
at
the
low
dose
were
limited
to
mild
increases
in
liver
weight
without
accompanying
changes
in
histopathology,
or
any
other
effect,
the
low
dose
was
considered
to
be
a
minimal
LOAEL.
Therefore,
the
panel
agreed
that
a
UFL
of
3
would
be
appropriate
if
the
monkey
study
were
to
be
used
as
the
critical
study.

1
EPA
is
currently
discussing
the
application
of
UFL
when
using
a
BMDL.
A
BMDL
value
represents
the
lower
limit
on
the
dose
that
should
cause
10%
of
the
experimental
animals
to
respond
with
the
effect
that
is
being
modeled.
Because
animal
studies
typically
cannot
detect
a
response
less
than
10%,
an
experimentally
derived
NOAEL
also
represents
the
dose
that
causes
10%
of
the
animals
to
respond.
For
this
reason,
EPA
has
historically
considered
a
BMDL
to
be
a
NOAEL
surrogate
and
selected
a
UFL
value
of
1
when
a
BMDL
is
used.
Although
EPA
does
not
have
official
guidance
on
this
issue,
recent
discussions
in
the
agency
suggest
that
if
the
effect
being
modeled
for
the
BMDL
is
adverse,
then
the
BMDL
should
be
considered
as
a
LOAEL.
Currently,
BMDLs
are
being
evaluated
on
a
case­
by­
case
basis,
considering
the
nature
of
the
effect
being
modeled
and
the
relationship
of
the
estimated
BMDL
to
observed
NOAELs.

21
Database
(
UFD):
The
database
for
deriving
a
high
confidence
RfD
includes
two
chronic
bioassays
by
the
appropriate
route
of
exposure
in
different
species,
one
two­
generation
reproductive
toxicity
study,
and
two
developmental
toxicity
studies
in
different
species.
The
minimal
database
required
for
deriving
a
RfD
is
a
single
subchronic
bioassay,
that
includes
a
full
histopathology
examination.
The
database
factor
is
used
to
account
for
the
fact
that
a
potential
health
effect
may
not
be
identified
if
the
database
is
missing
a
particular
type
of
study.
This
factor
may
also
be
used
if
the
existing
data
indicate
the
potential
for
a
heath
effect
that
is
not
fully
characterized
by
the
standard
bioassays,
for
example
neurotoxicity
or
immunotoxicity.
If
the
database
is
complete,
a
value
of
1
is
appropriate.
If
only
the
minimal
database
is
available,
then
a
default
of
10
is
used.
A
value
of
3
may
be
used
if
the
database
is
missing
one
or
two
key
studies.

The
panel
agreed
that
the
oral
database
for
C8
is
complete.
For
the
purpose
of
evaluating
uncertainty
factors,
the
panel
felt
that
the
human
occupational
studies
provided
sufficient
information
on
the
effects
of
long­
term
exposure
in
humans
to
function
as
a
chronic
bioassay.
In
addition,
the
consistency
between
the
monkey
and
rat
subchronic
studies
provides
confidence
that
non­
rodent
species
respond
similarly
to
rats
and
that
liver
is
a
sensitive
target
organ
in
all
species.
Furthermore,
a
developmental
toxicology
study
indicated
that
such
effects
only
occurred
at
high
concentrations,
and
reproductive
effects
were
monitored
in
the
2­
generation
reproductive
study.

Therefore,
the
panel
unanimously
concluded
that
a
UFD
value
of
1
is
appropriate
with
either
the
rat
or
monkey
toxicology
studies
selected
as
the
critical
study.

Outcome:
The
summary
of
the
panel's
unanimous
conclusions
regarding
individual
and
composite
uncertainty
factors
is
presented
in
Table
2
below.
The
composite
uncertainty
factor
is
obtained
by
multiplying
the
individual
factors.
(
Note,
that
following
EPA
convention,
an
uncertainty
factor
of
3
actually
represents
the
log
of
the
halfway
point
between
1
and
10.
Therefore
multiplying
half­
log
values
of
3
results
in
a
full
log
value
of
10,
rather
than
9
as
would
be
expected
for
numeric
multiplication.)

Table
2.
Panel
Recommendations
of
UF
Selection
for
Oral
pRfD
Study
UFH
UFA
UFL
UFD
UFS
Composite
UF
All
Rat
10
10
1
1
1
100
Monkey
10
10
3
1
3
1000
Oral
Reference
Dose
(
RfD)

The
final
value
of
the
RfD
is
obtained
by
dividing
the
point­
of­
departure
by
the
composite
uncertainty
factor.
As
discussed
above,
the
point­
of­
departure
selected
by
the
panel
is
the
BMDL
of
0.42
mg/
kg­
day
estimated
from
the
rat
two­
generation
study
(
York
et
al.,
2002)
and
the
composite
factor
is
100.
Therefore,
the
resulting
pRfD
is
0.42
÷
100,
or
0.0042
mg/
kg­
day.
Because
of
the
lack
of
precision
inherent
in
the
RfD,
only
one
significant
figure
is
appropriate;
therefore,
this
value
is
rounded
to
0.004
mg/
kg­
day.

22
For
comparison
purposes,
the
panel
considered
the
pRfD
values
that
would
result
from
choosing
alternative
NOAELs
or
BMDLs
as
the
point
of
departure.
This
analysis
is
presented
in
Table
3
below:

Table
3.
Comparison
of
pRfDs
Derived
Using
Different
Studies
Study
UF
NOAEL
RfD
BMDL
RfD
Palazzolo
et
al.
(
1993)
100
0.47
0.005
0.72
0.007
Riker
Laboratories
(
1983)
100
1.3
0.01
0.73
0.007
York
et
al.
(
2002)
100
­­­
­­­
0.42
0.004
Thomford
et
al.
(
2001)
1000
3­
10
(
LOAEL)
0.003­
0.01
­­­
­­­

Based
on
this
review
table
developed
by
the
panel,
the
pRfDs
that
could
be
derived
from
the
C8
oral
database
range
from
0.003
to
0.01
 
at
most
a
factor
of
3
separates
the
different
potential
pRfDs.
Considering
that
the
definition
of
the
RfD
states
that
the
RfD
incorporates
uncertainty
spanning
an
order
of
magnitude
(
a
10­
fold
variation),
the
panel
noted
that
close
agreement
of
the
potential
pRfD
values
provides
added
confidence
in
the
derived
pRfD
of
0.004
mg/
kg­
day.

Noncancer
Assessment:
Review
of
the
Dermal
Studies
(
Note:
Dr.
Seed
abstained
from
voting
during
this
part
of
the
meeting)

The
data
on
C8
by
the
dermal
route
of
exposure
are
limited.
Other
than
acute
lethality,
skin
sensitization,
and
irritation
studies,
the
dermal
database
consists
of
only
a
single
2­
week
study.

Kennedy
et
al.
1985
This
is
a
two­
week
study
in
male
rats
in
which
animals
had
C8
applied
to
the
skin
for
6
hours/
day,
5
days/
week
at
doses
of
0,
4.2,
42,
and
420
mg/
kg­
day.
Although
this
is
a
short­
term
study,
it
is
the
only
candidate
for
possible
use
in
determining
a
reference
dose
for
the
dermal
route
of
administration.
The
primary
effects
observed
were
increased
liver
weight
and
liver
pathology.
A
panelist
noted
that
the
study
design
prevented
animals
from
ingesting
the
dermally­
applied
material.
Although
the
amount
of
material
inhaled
was
considered
to
be
low,
some
inhalation
almost
certainly
occurred
in
the
dosed
animal
because
the
control
animals
had
detectable
C8
blood
levels.
It
was
also
noted
that
the
consistency
of
the
material
applied
to
the
animals
varied
among
the
dose
groups,
depending
on
the
concentration
of
C8
in
the
material
matrix.
In
all
instances
the
amount
of
material
on
the
skin
was
considerably
thicker
than
a
monolayer,
and
therefore,
the
applied
doses
might
not
reflect
accurately
the
absorbed
doses
of
C8
in
this
study.

Key
Panel
Discussion
Points:
One
panelist
stated
that
this
study
could
provide
potentially
useful
information
because
systemic
effects
are
observed
at
dose
levels
below
those
which
cause
portal
of
entry
effects
(
skin
irritation).
The
panel
discussed
whether
it
would
be
appropriate
to
extrapolate
the
results
of
this
study
to
longer
durations
in
order
to
derive
a
dermal
pRfD.
The
panel
concluded
that
such
extrapolation
would
not
be
advisable
because
of
the
possibility
of
unpredictable
longer­
term
dermal
effects.
One
panelist
asked
if
route­
to­
route
extrapolation
could
be
done
from
the
oral
studies
23
to
estimate
a
dermal
NOAEL
or
LOAEL.
Other
panelists
thought
this
would
not
be
possible
due
to
uncertainties
in
the
C8
toxicokinetics
by
oral
versus
dermal
exposure
routes.
For
example,
enterohepatic
circulation
is
known
to
occur
following
oral
exposure,
but
would
not
occur
following
dermal
exposure.
Therefore,
the
toxicokinetics
of
C8
is
different
between
the
two
routes
of
exposure.
Regardless
of
the
route
of
entry,
C8
is
not
metabolized.
Furthermore,
no
data
on
the
dermal
absorption
rate
were
identified.
One
panelist
noted
that
if
the
findings
from
this
study
were
used
to
determine
a
reference
dose,
the
resulting
value
would
be
higher
than
the
reference
dose
obtained
from
the
oral
studies.
Therefore,
using
oral
studies
to
set
the
reference
dose
would
be
adequately
protective,
of
systemic
exposure
via
the
dermal
route.
Another
panelist
agreed,
stating
that
no
dermal
reference
dose
should
be
identified
at
all,
and
that
a
specific
reference
dose
for
dermal
exposure
was
not
needed.

Outcome:
The
panel
agreed
unanimously
that
this
study
should
not
be
used
to
determine
a
dermal
pRfD
because
of
uncertainties
inherent
in
the
study
design
as
noted
in
the
discussion.

Noncancer
Assessment:
Review
of
the
Inhalation
Studies
(
Note:
Dr.
Seed
was
absent
during
this
part
of
the
meeting)

The
data
on
C8
by
the
inhalation
route
of
exposure
are
limited.
Other
than
acute
lethality
studies,
the
inhalation
database
consists
of
a
2­
week
study
and
a
developmental
toxicity
study.

Kennedy
et
al.
1986
and
Staples
et
al.
1981
Two
inhalation
studies
were
discussed
as
potential
candidates
for
deriving
the
pRfC.
Kennedy
et
al.
(
1986)
reported
a
two­
week
study
in
male
rats
in
which
animals
were
exposed
head­
only
6
hours/
day,
5
days/
week
to
C8
air
concentrations
of
0,
1,
7.6,
or
84
mg/
m3
.
The
primary
effects
observed
in
this
study
at
the
mid­
concentration
included
increased
absolute
and
relative
liver
weight,
supported
by
clinical
chemistry
and
histopathology
findings.
The
high
concentration
resulted
in
severe
toxicity,
including
mortality
in
one
rat.
Other
findings
at
the
high
concentration
group
were
increased
lung
and
testes
weight.
A
concentration­
dependent
increase
in
the
incidence
of
nasal
and
ocular
discharge
was
noted.

A
second
potential
critical
study
for
deriving
the
pRfC
was
a
developmental
toxicity
study
by
Staples
et
al.
(
1981).
Pregnant
rats
were
exposed
whole­
body
6
hours/
day
on
gestation
days
6
to
15
to
C8
air
concentrations
of
0,
0.14,
1.2,
9.9,
and
21.0
mg/
m3
.

The
panel
agreed
the
Kennedy
two­
week
study
provided
the
highest
quality
data
for
possible
determination
of
critical
effects
and
provided
a
slightly
lower
NOAEL/
LOAEL
boundary,
even
though
both
studies
used
similar
air
concentrations.
In
addition,
the
Kennedy
et
al.
(
1986)
study
evaluated
a
broader
array
of
systemic
endpoints,
and
included
a
histopathology
examination.

In
describing
their
initial
review
of
the
study,
TERA
noted
that
EPA's
RfC
methodology
states
that
the
air
concentrations
to
which
animals
are
exposed
are
to
be
converted
to
"
Human
Equivalent
Concentrations
(
ConcHEC
)"
by
applying
dosimetric
adjustments
(
USEPA,
1994).
Dosimetric
adjustments
account
for
the
different
structure
and
surface
area
of
animal
respiratory
tracts
compared
with
humans.
Different
dosimetric
adjustments
are
applied
depending
on
where
effects
are
observed.
For
example,
a
different
dosimetric
adjustment
will
be
applied
for
liver
effects
than
will
be
applied
for
lung
effects.
TERA
noted
that
the
key
piece
of
data
needed
to
calculate
the
ConcHEC
is
a
description
of
the
particle
size
distribution
(
i.
e.,
the
mass
median
aerodynamic
diameter
and
geometric
standard
24
deviation
or
GSD).
Data
available
from
the
published
study
did
not
provide
complete
information
about
the
mass
median
aerodynamic
diameter
for
the
low­
concentration
group,
or
GSD
for
any
exposure
group.
In
order
to
facilitate
the
discussion
of
the
study,
TERA
presented
human
equivalent
concentrations
for
liver
(
extrarespiratory)
and
lung
(
pulmonary)
effects
from
this
study
assuming
either
a
monodisperse
particle
size
distribution
or
a
polydisperse
particle
size
distribution.
These
results
were
presented
to
the
panel
as
shown
in
Table
4
below.

Table
4.
Preliminary
ConcHEC
Calculations
from
Kennedy
et
al.
(
1986)
Study
Concentrationa
GSD
=
1.3
(
Monodisperse)
GSD
=
3
(
Polydisperse)
Liver
Lung
Liver
Lung
1.0
0.6
0.018
0.5
0.09
7.6
4.6
0.14
4.0
0.70
84
67.7
17.7
46.9
7.4
a.
All
values
are
presented
in
units
of
mg/
m3
.

Key
Panel
Discussion
Points:
It
was
noted
that
the
inhalation
database
does
not
meet
the
minimum
database
requirements
for
determining
an
RfC
of
one
subchronic
90­
day
study
that
includes
histopathology
of
the
respiratory
tract,
but
that
the
consent
order
required
a
pRfC
in
order
to
set
air
screening
levels.
One
panelist
stated
that
it
was
not
appropriate
to
extrapolate
from
oral
studies
to
derive
a
RfC
because
of
the
absence
of
data
on
toxicokinetics
differences
between
these
routes
(
e.
g.,
effects
of
enterohepatic
circulation,
or
absorption).

One
panel
member
indicated
that
the
data
needed
to
calculate
the
ConcHEC
(
i.
e.,
the
mass
median
aerodynamic
diameter
[
MMAD]
and
geometric
standard
deviation
[
GSD]),
but
not
reported,
in
the
published
study
could
be
made
available
to
TERA
after
the
meeting.
The
panel
agreed
that
these
data
should
be
provided
to
TERA,
for
calculation
of
the
appropriate
ConcHEC
following
the
meeting.
The
panel
then
discussed
whether
the
lung
or
the
liver
was
the
critical
organ,
recognizing
that
the
final
designation
of
critical
effect
could
not
be
made
until
the
correct
ConcHEC
is
calculated.
TERA
raised
the
question
of
whether
the
reported
increases
in
the
incidence
of
nasal
and
ocular
discharge
should
be
considered
an
adverse
effect.
It
was
noted
that
this
effect
is
not
uncommon
for
the
exposure
protocol
that
was
used,
and
the
effect
was
seen
in
all
groups.
It
was
further
noted
that
C8
is
not
an
irritant,
and
that
no
nasal
histopathology
was
observed
in
exposed
animals.
In
selecting
critical
study
concentrations
the
panel
discussed
the
lung
effects
at
higher
doses.
One
panel
member
suggested
that
at
the
high
concentration
the
overt
pulmonary
toxicity
was
observed
due
to
the
large
particle
burden.
Uncertainties
in
interpreting
the
lung
effects
were
raised
by
the
panel.
One
panelist
noted
that
the
studies
were
too
short
to
determine
what
effect
chronic
exposure
would
have
on
the
respiratory
tract.
Another
suggested
that
existing
human
data
associated
with
the
human
study
reports
discussed
earlier
(
pulmonary
function
testing
of
workers,
etc.)
might
be
useful
in
determining
NOAEL/
LOAEL
values.
After
this
discussion,
the
panel
considered
the
study
concentration
of
7.6
mg/
m3
to
be
the
NOAEL
for
pulmonary
effects,
with
the
LOAEL
of
84
mg/
m3
.

The
panel
next
discussed
the
liver
effects.
It
was
noted
that
the
observed
increases
in
liver
weight
were
consistent
with
the
effects
observed
in
the
oral
studies.
Another
panel
member
noted
the
increased
alkaline
phosphatase
(
AP)
values
observed
at
the
higher
doses
were
not
necessarily
the
result
of
the
types
of
liver
effects
seen
in
the
oral
and
dermal
studies,
since
increased
AP
levels
often
reflect
disorders
of
biliary
flow.
One
panelist
questioned
the
ability
of
the
study
to
detect
systemic
effects
given
the
short
exposure
period
and
the
kinetics
of
the
compound;
however,
another
panelist
replied
that
the
half­
life
of
C8
in
rats
is
5
to
7
days,
and
the
study
design
would
have
allowed
achievement
of
25
steady­
state
concentrations
in
the
blood.
The
panel
considered
the
study
concentration
of
1.0
mg/
m3
as
the
NOAEL
for
liver
effects.
However,
one
reviewer
suggested
that
if
the
liver
effects
are
found
to
be
the
critical
effect
based
on
the
ConcHEC
,
then
benchmark
concentration
modeling
should
be
conducted
before
assigning
a
critical
effect
level.

The
panel
considered
the
appropriate
uncertainty
factors
for
a
pRfC,
noting
that
the
final
choice
of
an
appropriate
value
for
some
areas
of
uncertainty
may
change
depending
on
whether
lung
or
liver
effects
are
found
to
be
critical.
(
Note
to
the
reader:
Essentially
the
same
areas
of
uncertainty
are
considered
in
developing
a
RfC
as
for
the
RfD.
For
a
full
explanation
of
the
purpose
for
each
factor,
see
the
earlier
discussion.)
For
the
same
reasons
as
discussed
for
the
pRfD,
the
panel
unanimously
agreed
that
a
value
of
10
was
appropriate
for
UFH
.
When
considering
interspecies
extrapolation,
it
is
generally
considered
that
the
dosimetric
adjustments
used
to
derive
the
ConcHEC
account
for
the
toxicokinetic
differences
between
animals
and
humans.
Therefore,
the
uncertainty
factor
only
needs
to
address
the
toxicodynamic
differences.
Since
there
are
no
data
regarding
dynamic
differences
between
rats
and
humans,
the
panel
agreed
that
the
default
value
of
3
was
appropriate
for
UFA
.
Since
the
Kennedy
study
identified
a
NOAEL,
the
panel
unanimously
agreed
that
a
value
of
1
was
appropriate
for
UFL
.

The
panel
considered
that
two
of
the
factors,
UFS
and
UFD
,
were
related
to
the
decision
of
whether
lung
or
liver
is
the
critical
effect.
If
liver
effects
are
determined
to
be
the
critical
effect,
then
at
least
one
panelist
felt
that
UFS
,
could
be
addressed
with
an
uncertainty
factor
of
1
because
the
oral
studies
provided
enough
information
to
be
confident
that
the
liver
effects
would
not
progress
in
severity
following
a
chronic
inhalation
exposure.
However,
other
panel
members
stated
that
there
were
insufficient
data
to
assess
whether
liver
would
continue
to
be
the
critical
effect
or
to
provide
information
on
how
the
respiratory
tract
would
respond
following
longer­
term
inhalation
exposures,
and
that
a
value
greater
than
1
for
UFS
was
needed.
For
the
UFS
and
liver
as
the
critical
organ,
the
panel
votes
were
1,
3,
or
10
with
the
majority
choosing
3.
If
liver
effects
are
determined
to
be
the
critical
effect,
then
panelists
were
split
on
the
value
of
the
uncertainty
factor
for
UFD
,
choosing
values
of
either
3
or
10
with
the
majority
of
the
panel
choosing
3.
No
unanimous
consensus
was
reached
on
these
two
factors;
however,
a
clear
majority
vote
was
reached
on
uncertainty
factors
of
3
each
for
UFS
and
UFD
in
reference
to
liver
as
the
target
organ.

If
lung
effects
are
determined
to
be
critical,
the
panel
was
divided
almost
equally
on
the
appropriate
value
for
UFS
with
opinions
covering
the
full
range
of
options
from
1
to
3
to
10.
Note
however,
that
six
scientists
voted
for
a
factor
less
than
10
(
either
1
or
3)
and
five
scientists
voted
for
a
value
greater
than
1
(
3
or
10).
Similarly,
the
panel
was
divided
on
the
appropriate
value
for
UFD
;
panel
opinions
covered
the
full
range
of
options
from
1
to
3
to
10
with
the
majority
of
panelists
choosing
3.

As
noted
above,
after
each
discussion
votes
were
taken
on
individual
factors.
These
votes
are
shown
in
Table
5.
Note
that
one
scientist
was
reviewing
the
dosimetric
adjustment
calculations
during
this
discussion
and
so
was
unable
to
vote
on
these
UFs;
also
note
that
one
more
vote
at
any
point
in
Table
5
would
not
have
changed
the
final
outcome.
In
addition,
the
panel
did
not
reach
consensus
on
the
confidence
in
the
RfC,
with
opinions
ranging
from
"
none"
to
"
high"
with
the
average
being
medium­
to­
low.

26
Table
5.
Tally
of
Panel
Votes
for
UFS
and
UFD
UFS
UFD
Factor
1
3
10
1
3
10
Liver
as
critical
1
6
1
0
6
2
Lung
as
critical
3
3
2
1
5
2
Outcome:
One
panelist
reminded
the
group
that
the
purpose
of
Kennedy
et
al.,
(
1986)
was
to
identify
the
inhalation
hazard,
not
to
look
closely
at
NOAEL,
LOAEL,
etc.
A
prospective
inhalation
study
designed
to
look
more
closely
at
the
NOAEL/
LOAEL
aspects,
to
evaluate
lesions
as
a
function
of
exposure
time,
and
to
evaluate
tissues
of
the
respiratory
tract
using
up­
to­
date
methodology
would
be
valuable
and
would
allow
a
more
focused
evaluation
of
the
RfC.
Nonetheless,
the
panel
agreed
that
a
pRfC
could
be
developed,
but
this
agreement
was
not
unanimous.
The
panel
also
recommended
that
TERA
obtain
additional
data
on
the
particle
size
GSD
value
to
determine
the
ConcHEC
corresponding
to
the
NOAEL
before
determining
whether
the
pulmonary
or
the
hepatic
effects
are
considered
critical.
If
the
liver
effects
are
determined
to
be
the
critical
effect,
then
BMD
modeling
should
be
done.
The
composite
uncertainty
factor
was
expressed
as
a
range
of
30
to
3,000.
The
final
pRfC
is
presented
in
the
Post
Meeting
Action
Items.

Cancer
Assessment
(
Note:
Dr.
Seed
abstained
from
voting
during
this
part
of
the
meeting)

U.
S.
EPA's
1999
Guidelines
for
Carcinogen
Risk
Assessment
were
used
to
frame
the
discussion
of
C8
carcinogenic
potential.
TERA
opened
the
discussion
with
a
short
introduction
to
these
guidelines,
highlighting
the
recent
focus
on
evaluation
of
the
mode
of
action
data
in
developing
a
weight
of
evidence
characterization,
and
in
deciding
the
most
appropriate
dose­
response
approach,
linear
or
margin
of
exposure
(
MOE).
It
was
noted
that
the
EPA's
1999
guidelines
would
be
used
as
the
basis
for
the
deliberations
of
the
panel.

Cancer
Hazard
Identification
and
Mode
of
Action
The
panel
discussed
the
evidence
for
C8
carcinogenicity
in
humans
and
agreed
that
the
human
carcinogenicity
evidence
is
inconclusive.
Although
four
prostate
tumors
were
reported
in
retired
workers,
three
of
these
four
cases
now
are
known
to
have
had
minimal
or
no
C8
exposure.
(
See
Human
Studies
section
for
more
detailed
discussion.)

The
panel
noted
that
two
animal
carcinogenicity
studies
had
been
conducted.
The
first
study
(
Riker
Laboratories,
1983)
reported
treatment­
related
increases
in
Leydig
cell
adenomas
and
mammary
gland
fibroadenomas.
The
second
study
(
Biegel
et
al.,
2001)
reported
treatment­
related
increases
in
tumors
in
the
liver,
Leydig
cells,
and
pancreas.
Panelists
noted
that
the
tumors
identified
in
the
Biegel
et
al.
(
2001)
study
correspond
to
the
triad
of
tumors
associated
with
some
chemicals
that
cause
peroxisome
proliferation.
Other
panelists
agreed
and
suggested
that
a
further
examination
of
the
data
may
indicate
that
this
triad
of
tumors
can
be
best
addressed
using
a
MOE
approach.
The
panel
also
noted
that
the
mammary
fibroadenomas
may
require
the
default
linear
model
because,
following
U.
S.
EPA
cancer
guidelines,
no
actual
mode
of
action
data
for
C8
and
this
tumor
type
are
available
to
warrant
moving
from
the
default
assumption.
Each
of
the
four
types
of
tumors
found
in
the
two
C8
animal
27
carcinogenicity
studies
was
then
discussed
in
detail
with
regard
to
the
weight
of
the
evidence
for
the
mode
of
action,
and
the
evidence
supporting
a
linear
or
MOE
dose­
response
assessment
approach.
Listed
below
are
the
outcomes
and
discussions
for
each
tumor
type.

Liver
tumors
Key
Panel
Discussion
Points:
The
discussion
on
liver
tumors
focused
on
the
role
of
peroxisome
proliferation
as
the
mode
of
action
for
the
observed
liver
tumors.
In
relating
this
liver
tumor
effect
to
humans,
one
panelist
said
humans
are
much
less
sensitive
to
peroxisome
proliferation
than
rats.
Another
panelist
noted
that
IARC's
approach
for
clofibrate
and
other
non­
genotoxic
peroxisome
proliferation
chemicals
was
to
assume
that
the
mode
of
action
was
not
relevant
to
humans
if
no
evidence
of
peroxisome
proliferation
was
observed
in
humans.
Another
panelist
said
that
although
rats
may
be
more
sensitive
than
humans
from
a
toxicodynamic
standpoint
(
due
to
interspecies
differences
in
receptors),
humans
may
be
more
sensitive
from
a
toxicokinetic
standpoint,
since
they
clear
C8
more
slowly
than
rats.
As
a
result,
the
panel
member
suggested
that
these
two
considerations
would
tend
to
decrease
overall
differences
in
species
sensitivity.
On
the
other
hand,
a
panel
member
noted
that
no
increased
incidence
of
tumors
have
been
found
in
people
taking
clofibrate,
a
known
peroxisome
proliferator,
which
suggests
that
humans
are
much
less
sensitive
to
peroxisome
proliferation
than
rats
and
they
may
have
no
response
at
all.
Based
on
these
data,
the
panel
member
suggested
that
the
lack
of
tumor
development
in
humans
exposed
to
C8
should
not
be
discounted.
The
panel
discussed
differences
in
results
between
the
two
cancer
studies.
One
panelist
noted
the
studies
have
differences
in
their
internal
delivered
doses
because
of
differences
in
the
animal
diets.
This
could
explain
the
difference
noted
in
toxic
effects.

Outcome:
The
majority
of
the
panel
agreed
that
the
data
indicate
peroxisome
proliferation
is
the
mode
of
action
for
the
liver
tumors,
and
that
although
the
liver
tumor
response
is
not
likely
to
be
quantitatively
similar
between
rats
and
humans,
the
use
of
the
liver
tumor
response
data
for
human
health
risk
assessment
cannot
be
totally
discounted.
However,
other
scientists
indicated
that
based
on
the
lack
of
peroxisome
proliferation
in
the
non­
human
primate
studies,
the
rodent
liver
tumors
are
not
relevant
at
all
to
humans.

Leydig
Cell
Tumors
Key
Panel
Discussion
Points:
In
reviewing
the
summary
tables
prepared
for
the
meeting,
one
panelist
noted
that
Leydig
cell
hyperplasia
should
be
evaluated.
In
response,
the
hyperplasia
data
from
Biegel
et
al.
(
2001)
was
reviewed
by
the
panel.
The
panel
developed
Table
6
to
facilitate
the
comparison
on
hyperplasia
and
tumorigenic
outcomes.

Table
6.
Summary
of
Beigel
et
al.,
2001
Leydig
Cell
Data
Pair
fed
controls
300
ppm
Liver
carcinomas/
adenomas
3/
79
10/
76*
Leydig
ademonas
2/
78
8/
76*
Pancreatic
carcinomas/
adenomas
1/
79
8/
76*
Leydig
cell
hyperplasia
26/
78
35/
76
28
The
panel
noted
that
no
significant
increase
in
Leydig
cell
hyperplasia
was
apparent
from
these
data;
however,
due
to
different
survival
times
between
the
two
groups
(
C8
treated
animals
survived
longer)
a
false
positive
effect
could
have
occurred
because
older
animals
would
have
more
time
to
develop
naturally
occurring
tumors.
It
was
noted
that
a
more
formal
analysis
would
be
needed
to
determine
whether
the
incidence
of
Leydig
cell
tumors
would
still
be
increased
after
adjusting
for
differences
in
survival,
but
the
formal
statistical
analysis
was
too
complex
to
complete
during
the
meeting.

The
panel
discussed
the
role
of
peroxisome
proliferation
as
the
mode
of
action
of
Leydig
cell
tumors.
Specifically,
the
panel
discussed
a
workshop
publication
(
Clegg
et
al.
1997)
that
evaluated
the
seven
known
modes
of
action
for
Leydig
cell
tumors.
Most
of
the
modes
of
action
involve
altered
hormonal
response
in
response
to
peroxisome
proliferation,
including
increased
estradiol
via
hepatic
aromatase
and
binding
to
the
TGF
 
receptor
or
elevations
in
leutinizing
hormone
to
compensate
for
the
testes
becoming
less
responsive
to
this
hormone.
One
panelist
emphasized
that
the
monkey
study
(
Thomford
et
al.,
2001)
showed
no
effects
in
the
testes,
even
though
the
animals
were
dosed
at
C8
levels
high
enough
to
cause
major
weight
loss
and
mortality.
This
panelist
suggested
that
this
indicates
the
Leydig
cell
effects
seen
in
rats
are
unlikely
to
occur
in
primates.
This
panel
member
also
noted
that
no
increased
estradiol
was
noted
in
the
monkeys.

One
panelist
observed
that
Leydig
cell
tumors
were
a
classic
response
to
peroxisome
proliferation
but
the
available
studies
do
not
provide
positive
evidence,
such
as
increased
estradiol
levels,
that
peroxisome
proliferation
is
the
operative
mode
of
action.
The
panelists
agreed
that
while
data
gaps
exist,
a
peroxisome
proliferation
mode
of
action
was
a
reasonable
assumption.
One
panelist
stated
that
whatever
the
MOA
was,
it
was
not
genotoxicity.

The
panel
agreed
unanimously
that
for
Leydig
cell
tumors:

­
All
7
possible
mechanisms
for
Leydig
cell
tumors
are
non­
linear;
therefore
a
non­
linear
dose­
response
approach
is
reasonable;
­
Humans
have
a
low
incidence
of
these
tumors;
­
The
monkey
study
did
not
demonstrate
Leydig
cell
pathology
or
increased
estradiol;
­
Leydig
cell
tumors
are
a
known
tumor
type
for
other
peroxisome
proliferators;
­
Humans
do
not
develop
Leydig
cell
tumors
following
exposure
to
other
known
peroxisome
proliferators
such
as
clofibrate;
­
Regardless
of
the
actual
mode
of
action,
it
is
likely
to
be
non­
genotoxic.

Outcome:
The
panel
agreed
that
based
on
the
absence
of
genotoxicity,
the
Leydig
cell
tumors
were
likely
to
be
caused
by
a
non­
genotoxic
mechanism.
The
panel
further
agreed
that
if
sufficient
evidence
were
available
to
show
increased
estradiol
levels
(
i.
e.,
secondary
to
peroxisome
proliferation)
as
the
mechanism
for
the
observed
tumors,
then
the
mechanism
would
be
non­
genotoxic
and
would
not
be
quantitatively
similar
or
possibly
not
relevant
at
all
to
humans.
However,
without
this
evidence
this
effect
can
not
be
totally
discounted.

29
Pancreatic
tumors
Key
Panel
Discussion
Points:
Since
the
tumor
results
from
the
Beigel
et
al.,
(
2001)
were
not
provided
in
the
summary
table
distributed
to
the
panel
prior
to
the
meeting,
the
pancreatic
tumor
data
from
this
study
were
presented
as
a
table
at
the
meeting
(
see
Table
7
below):

Table
7
Biegel
Study:
Pancreas
Tumors
Control
pair­
fed
control
300
ppm
Hyperplasia
14/
80
(
18%)
8/
79
(
10%)
30/
48*
(
40%)
Adenomas
0/
80
1/
79
7/
76*
Carcinomas
0/
80
0/
79
1/
76
One
panelist
described
an
analysis
that
had
been
done
to
compare
the
two
cancer
studies
with
regard
to
the
pancreatic
tumors.
This
panelist
noted
that
although
the
first
study
(
Riker
Laboratories,
1983)
did
not
report
pancreatic
tumors
or
hyperplasia,
the
second
study
(
Biegel
et
al.,
2001)
did.
However,
this
panel
member
also
noted
that
the
studies
were
not
inconsistent
because
of
the
different
definitions
of
adenoma
versus
hyperplasia
based
on
pancreatic
cell
size
used
by
the
respective
investigators.
Also,
the
criteria
for
separating
hyperplasia
from
adenomas
is
based
on
lesion
size.
Both
studies
were
qualitatively
similar
with
a
number
of
larger
lesions
(
adenomas)
found
in
the
Biegel
study.
Another
scientist
commented,
when
the
two
studies
were
recently
compared
by
a
group
of
pathologists
using
current
criteria,
there
was
a
consistency
in
a
pancreatic
response;
however,
there
was
not
an
increased
number
of
adenomas
found
in
the
earlier
study.
Instead,
an
increase
in
hyperplastic
nodules
of
the
acinar
pancreas
was
found,
which
is
consistent
with
the
Beigel
study.
However,
even
though
the
dietary
dose
was
the
same
(
300
ppm),
the
Riker
Laboratories
study
rats
did
not
develop
these
hyperplasias
into
adenomas
to
the
extent
that
occurred
in
the
Beigel
study.

With
regard
to
the
potential
mode
of
action,
one
panelist
suggested
that
the
persistent
increase
seen
in
cholecystokinin
and
increased
bile
acids
may
be
involved
in
the
MOA,
but
the
evidence
in
rats,
monkeys
and
humans
does
not
support
this
hypothesis.
When
a
panelist
asked
if
a
strong
case
could
be
made
that
the
pancreatic
tumors
resulted
from
peroxisome
proliferation,
several
panelists
responded
no.
Another
added
that
while
some
peroxisome
proliferation
agents
cause
the
triad
of
tumors
seen
with
C8,
not
all
do.
Another
panelist
added
that
no
pancreatic,
liver,
or
testes
hyperplasia
was
noted
in
monkeys
at
the
time
of
sacrifice.

Outcome:
The
panel
agreed
that
the
evidence
was
not
sufficient
to
demonstrate
the
MOA
for
pancreatic
tumors,
but
enhanced
cell
proliferation
(
hyperplasia)
was
likely
to
be
involved.
The
MOA
appears
to
be
non­
genotoxic
based
on
the
results
of
genotoxicity
bioassays.

30
Mammary
Fibroadenomas
Key
Panel
Discussion
Points:
The
panel
considered
whether
the
fibroadenomas
observed
in
the
Riker
Laboratories
study
were
a
real
treatment­
related
effect,
or
an
artifact
of
classification,
since
other
mammary
tumor
types
observed
in
this
study
showed
no
clear
relationship
with
dose.
Table
8
below
shows
the
data
for
several
types
of
mammary
tumors
from
this
study:

Table
8.
Riker
Study:
Mammary
Tumors
Control
30
PPM
300
PPM
Adenomas
7%
0
0
Adenocarcinomas
15%
31%
11%
Carcinomas
2%
0
0
Fibroadenomas
22%
42%
48%*

One
panelist
suggested
that
even
though
fibroadenomas
were
statistically
significant,
when
all
mammary
tumor
types
are
combined,
they
are
not
likely
to
be
significant.
It
was
noted
by
the
panel
that
the
individual
incidence
data
from
the
study
would
need
to
be
examined
to
determine
the
combined
incidence
of
all
mammary
tumor
types,
rather
than
adding
the
percentages
from
each
category.
The
panel
discussed
the
histological
basis
for
reporting
separately
fibroadenomas
versus
other
types
of
mammary
adenomas.
A
panelist
suggested
that
since
fibroadenomas
do
not
progress
to
the
other
types
it
is
correct
to
report
them
separately.
Another
said
that
the
National
Toxicology
Program
(
NTP)
reports
fibroadenomas
combined
with
adenomas.

The
panel
also
discussed
potential
modes
of
action
for
mammary
tumors.
Increased
estradiol
was
proposed
as
a
possible
MOA
for
the
induction
of
hyperplasia
and
tumor
formation,
but
the
panel
did
not
believe
the
data
were
sufficient
to
demonstrate
this
proposed
mode
of
action.
A
panelist
asked
if
a
linear
assessment
could
be
done
to
help
decide
the
importance
of
the
effect.
Another
responded
that
the
data
were
not
adequately
fit
by
any
of
the
acceptable
dose­
response
models,
so
a
quantitative
dose­
response
assessment
was
not
reported
for
this
data
set.

Outcome:
The
panel
agreed
the
data
are
not
adequate
to
demonstrate
a
MOA;
however
based
on
the
negative
genotoxicity
assays,
C8
is
unlikely
to
be
genotoxic.
Several
panelists
were
not
convinced
the
data
demonstrated
any
real
tumorigenic
effect.

Cancer
Dose­
Response
Assessment
After
evaluating
the
relevance
of
each
tumor
type
to
humans,
and
the
potential
mode
of
action,
the
panel
members
were
asked
to
recommend
a
dose­
response
approach
for
each
tumor
type.
In
all
cases
the
panel
agreed
unanimously
unless
noted
otherwise.
For
the
liver
tumors,
the
panel
agreed
that
the
MOE
approach
was
most
appropriate.
For
the
remaining
tumor
types,
the
panel
agreed
that
both
linear
and
MOE
approaches
were
appropriate,
since
the
mode
of
action
was
not
considered
to
have
been
adequately
demonstrated
for
any
of
these
three
tumor
types.
All
panel
members
agreed
with
these
conclusions,
except
for
the
Leydig
cell
tumors,
where
one
panel
member
argued
that
only
an
MOE
approach
should
be
used.

31
For
the
liver
tumors,
the
MOE
approach
was
selected.
Since
the
MOE
analysis
often
uses
the
benchmark
response
for
a
precursor
as
the
basis
of
deriving
a
point
of
departure,
the
panel
judged
the
pRfD
for
liver
effects
as
sufficiently
protective
of
potential
liver
carcinogenicity.

For
Leydig
cell
tumors,
benchmark
dose
modeling
was
conducted
to
identify
a
point
of
departure
for
the
linear
and
MOE
assessments.
The
Point
of
Departure
(
POD)
for
Leydig
cell
was
chosen
by
the
panel
from
the
BMD
modeling
output.
The
BMDL
of
0.32
mg/
kg­
day
was
selected
as
the
most
appropriate
basis
for
deriving
the
assessment.

The
panel
discussed
the
appropriate
factors
to
apply
to
the
BMR
for
completing
the
MOE
assessment.
The
panel
noted
that
EPA's
1999
guidelines
have
only
recently
begun
to
be
applied,
and
that
formal
guidance
or
examples
of
the
interpretation
and
default
values
to
use
in
deriving
the
MOE
are
lacking.
In
discussing
the
important
considerations
for
the
MOE,
the
panel
decided
that
the
critical
factors
to
be
considered
were
for
"
Nature
of
Effect",
Intrahuman
sensitivity"
and
"
Animal
to
Human
Extrapolation".
A
summary
of
the
factors
chosen
is
shown
in
Table
9.

For
the
Leydig
cell
tumors,
a
factor
of
3
for
nature
of
effect
was
selected
as
the
most
appropriate
value,
since
the
observed
effect
was
for
benign
tumors.
A
factor
of
10
was
selected
for
Intrahuman
sensitivity.
A
factor
of
3
was
used
for
Animal
to
Human
Extrapolation,
since
dosimetric
adjustments
were
applied
to
the
dose
data
used
for
the
BMD
modeling.
This
composite
factor
of
100
was
further
supported
since
these
types
of
tumors,
although
common
in
rats,
are
found
rarely
in
people.
In
addition,
the
mode
of
action
is
likely
via
peroxisome
proliferation
which
is
quantitatively
much
less
important
in
humans.
The
panel
agreed
that
the
composite
MOE
of
100
was
appropriate.

For
the
linear
dose­
response
assessment
for
Leydig
cell
tumors
the
BMDL
of
0.32
mg/
kg­
day
was
used
to
calculate
an
oral
cancer
slope
factor
as
follows:

Slope
factor
=
risk/
dose
=
0.1/
0.32
=
0.31
per
mg/
kg­
day
(
Note:
risk
is
numerically
expressed
as
0.1
because
the
BMDL
is
the
point
that
represents
a
10%
increased
in
tumor
incidence
in
accordance
with
EPA
guidance.)
BMD
modeling
failed
for
the
tumor
data
for
pancreatic
tumors
and
mammary
gland
fibroadenomas.
Therefore,
the
panel
determined
that
the
data
for
these
two
tumor
types
were
not
adequate
to
conduct
a
quantitative
dose­
response
assessment.

Table
9.
Factors
Used
to
Describe
Various
Areas
in
the
Development
of
MOEs
for
Cancer
Endpoints.
Nature
Intra
Animal
Steepness
Total
Tumor
Model
Of
Effect
Human
to
Human
of
Slope
Exposure
MOE
Liver
MOE
1
10
10
NR
NR
100
Leydig
both
3
10
3
NR
NR
100
Pancreas
both
NA
(
cannot
be
modeled)
Mammary
both
NA
(
cannot
be
modeled)
NR
=
Not
Relevant
based
on
panel
judgment;
NA
=
Not
Applicable
32
The
panel
also
voted
on
confidence
ratings
for
the
cancer
assessment.
TERA
noted
that
according
to
EPA
guidance
"
high
confidence"
suggests
that
the
assessment
is
unlikely
to
change
with
the
availability
of
new
data,
while
"
low
confidence"
indicates
that
the
assessment
is
likely
to
change
with
new
data.
Based
on
these
criteria
the
panel
voted
on
their
confidence
in
the
cancer
assessment
using
either
the
pRfD
for
liver
toxicity
to
adequately
account
for
the
liver
cancer
risk
or
using
the
assessment
based
on
Leydig
cell
tumors.
The
panel
voted
as
follows:

Liver
pRfD
=
high
(
7
votes);
medium­
high
(
2
votes)
Leydig
tumors
=
low
(
7
votes);
low­
medium
(
2
votes)

Therefore,
the
panel
agreed
that
the
oral
pRfD
for
liver
toxicity
would
be
the
basis
for
determining
water
and
soil
screening
levels
(
which
are
based
primarily
on
oral
exposure)
for
the
following
reasons:

 
high
confidence
in
the
pRfD
(
i.
e.,
not
likely
to
change
in
the
future
due
to
additional
data
collection);
 
the
pRfD
would
be
protective
against
the
quantitatively
less
sensitive
and
questionable
relevance
peroxisome
proliferation­
related
liver
cancer
in
humans;
 
low
confidence
in
the
Leydig
tumor
analysis
and
questionable
relevance
to
humans;
 
limitations
in
study
design,
data
quality,
and
data
interpretation
rendered
difficult
the
determination
of
whether
the
reported
increased
incidence
of
pancreatic
tumors
or
mammary
tumors
were
related
to
C8
treatment,
and
did
not
allow
the
modeling
of
a
point
of
departure
that
could
serve
as
the
quantitative
basis
for
risk
value
development.

Screening
Levels
(
Note:
Dr.
Seed
was
absent
during
this
part
of
the
meeting)

The
consent
order
required
that
screening
levels
be
developed
for
drinking
water,
soil,
and
air.
The
panel
followed
the
guidance
provided
by
U.
S.
EPA's
"
Risk
Assessment
Guidance
for
Superfund"
as
further
explained
by
both
Region
3
and
Region
9
risk­
based
concentration
guidance.
In
cases
where
a
conflict
occurred
between
the
guidance
documents,
Region
9
guidance
was
followed
because
it
is
more
conservative,
i.
e.
more
health
protective.
For
drinking
water
and
soil,
only
ingestion
and
dermal
absorption
were
considered
as
routes
of
exposure.
EPA
guidance
indicates
volatilization
from
water
or
soil
should
only
be
evaluated
for
chemicals
with
Henry's
law
constants
greater
than
10­
5
and
molecular
weights
less
than
200.
Since
C8'
s
Henry's
Law
constant
is
10­
11
and
its
molecular
weight
is
431,
volatilization
was
not
evaluated.

As
discussed
above,
the
panel
concluded
that
since
both
liver
and
Leydig
cell
tumors
were
potentially
formed
via
nonlinear
modes
of
action,
and
further
since
greater
confidence
was
placed
in
the
quantitative
assessment
based
on
the
liver
endpoint,
the
pRfD
and
pRfC
for
liver
toxicity
would
be
protective
of
potential
cancer
effects
of
C8.
The
panel
considered
that
the
linear
extrapolation
for
Leydig
cell
tumors
was
too
uncertain
to
be
used
with
confidence
and
that
the
MOE
approach
based
on
the
Leydig
cell
tumors
gave
essentially
the
same
numerical
value
as
that
for
the
liver
endpoint,
but
with
less
confidence.
Thus,
the
pRfD
and
pRfC
for
liver
toxicity,
and
"
noncancer"
equations
were
used
for
calculating
screening
levels.
Screening
levels
are
calculated
following
the
premise
that
if
lifetime
exposure
is
equal
to
or
less
than
the
pRfD
or
pRfC,
then
no
risk
of
deleterious
effects
is
expected.
Mathematically,
this
concept
can
be
expressed
by
the
following
standard
equation;
the
ratio
of
the
measured
or
estimated
exposure
to
the
RfD
is
called
the
Hazard
Quotient.

33
If
Exposure
÷
RfD
=
1
or
less,
then
no
risk
of
deleterious
effects
is
presumed.

Using
this
concept,
it
is
possible
to
estimate
the
concentration
in
media
that
results
in
a
lifetime
exposure
equal
to
the
pRfD
or
pRfC.
These
equations,
from
EPA
Region
9'
s
guidance
on
deriving
risk
based
concentrations,
are
listed
below:

Air
Screening
Level
:
[
]
ug/
m3
=
THQ
x
RfDi
x
BW
x
AT
x
1000
EF
x
ED
x
air
IR
Note:
RfDi
(
mg/
kg­
day)
=
RfC
x
20m3/
d
(
IR)
70
kg
(
BW)

Soil
Screening
Level:
[
]
mg/
kg
=
THQ
x
AT
x
BW
EF
x
ED
x
[
soil
IR
/
RfD
x
10
 
6
+
SA
x
AF
x
ABS
/
RfD
x
10
 
6
]

Water
Screening
Level:
[
]
ug/
L
=
THQ
x
AT
x
BW
x
1000
EF
x
ED
x
[
water
IR
/
RfD]

Where:
THQ
=
Target
Hazard
Quotient,
assumed
to
be
1
RfDi
=
The
RfC
expressed
in
terms
of
dose,
mg/
kg­
day
RfD
=
The
oral
reference
dose
estimated
by
the
panel,
0.004
mg/
kg­
day
RfC
=
The
inhalation
reference
concentration
estimated
by
the
panel,
see
below
BW
=
Body
weight,
assumed
to
be
70
kg
for
adults
and
15
kg
for
children
AT
=
Averaging
time,
10950
days,
the
exposure
duration
expressed
in
days
EF
=
Exposure
Frequency,
350
days/
year,
the
average
number
of
days
each
year
people
are
exposed
ED
=
Exposure
duration,
30
years,
the
average
number
of
years
people
are
exposed
IR
=
Inhalation
rate
for
air
screening
levels,
20
m3
/
day;
Ingestion
rate
for
soil
and,
Water
screening
levels,
200
mg/
day
soil
ingested
based
on
child
exposure
and,
2
L/
day
water
ingested
based
on
adult
exposure
SA
=
Surface
area
of
exposed
skin,
2800
cm2
/
day
AF
=
Adherence
factor,
0.2
mg/
cm2
,
the
amount
of
soil
that
adheres
to
skin
ABS
=
Skin
absorption
factor,
specific
factor
not
available
for
C8,
assumed
to
be
0.1
for
semi­
volatile
chemical
per
EPA
guidance
The
panel
unanimously
agreed
that
the
equations,
assumptions,
and
default
exposure
parameters
described
above
were
the
appropriate
choices
for
calculating
screening
levels
for
air,
soil,
and
water.
The
following
values
are
the
screening
levels
estimated
by
the
equations.

34
For
air:
0.1
 
6.0
micrograms
per
cubic
meter
of
air
(
µ
g/
m3)
ambient
air.
Note
that
the
panel
considered
this
range
to
be
interim
until
the
additional
work
discussed
for
the
RfC
is
completed.
This
range
incorporates
the
range
of
possible
NOAELHECs
estimated
by
TERA
prior
to
the
meeting
as
well
as
the
range
of
composite
uncertainty
factors
recommended
by
the
panel.
The
final
pRfC
is
discussed
in
the
following
section
Post
Meeting
Action
Items.

For
soil:
244
miligrams
per
kilogram
of
soil
(
mg/
kg)
residential
soil,
rounded
to
240
mg/
kg.

For
water:
146
micrograms
per
liter
of
water
(
µ
g/
L),
rounded
to
150
µ
g/
L.

35
2.3
POST
MEETING
ACTION
ITEMS
The
following
activities
were
conducted
after
the
CATT
Toxicologists
meeting.

Derivation
of
the
pRfC
for
C8
The
CATT
panel
could
not
develop
a
final
recommendation
on
the
pRfC
or
the
air
screening
level
during
the
May
6
and
May
7,
2002
meeting.
This
was
due
to
a
lack
of
data
necessary
for
these
calculations.
At
the
meeting,
the
panel
chose
the
key
study
for
risk
factor
derivation
as
the
2­
week
inhalation
study
by
Kennedy
et
al.
(
1986)
and
voted
upon
the
uncertainty
factors.
They
directed
the
author,
panel
member
Kennedy
(
DuPont),
to
(
1)
retrieve
the
standard
deviation
data
for
the
absolute
and
relative
liver
weight
data
sets;
and
(
2)
to
measure
the
particle
size
distribution
in
the
exposure
chamber
and
determine
the
corresponding
standard
deviation;
and
(
3)
to
provide
these
data
to
DEP
and
to
TERA.
The
panel
directed
TERA
to
utilize
these
data
to
develop
the
pRfC
based
on
the
most
sensitive
organ
(
liver
or
lung)
and
the
air
screening
level
based
on
USEPA
Region
9
standard
formulas.

During
the
meeting,
the
CATT
panel
agreed
that
the
Kennedy
et
al.
(
1986)
study
was
the
most
appropriate
basis
for
deriving
the
pRfC,
with
the
developmental
study
by
Staples
et
al.
(
1981)
providing
support
for
the
selected
critical
effect
levels.
The
CATT
panel
identified
a
NOAEL
for
increased
liver
weight
at
the
lowest
study
concentration
of
1.0
mg/
m3
,
with
a
LOAEL
of
7.6
mg/
m3
.
The
NOAEL
for
lung
effects
was
identified
by
the
CATT
panel
as
7.6
mg/
m3
,
with
a
LOAEL
was
84
mg/
m3
.

In
order
to
derive
an
pRfC,
the
reported
study
concentrations
were
converted
to
human
equivalent
concentrations
(
ConcHEC
),
according
to
current
U.
S.
EPA
RfC
methodology
(
USEPA,
1994).
The
calculation
of
the
ConcHEC
requires
two
steps.
First,
the
study
concentration
is
adjusted
from
the
exposure
duration
used
in
the
experiment
to
an
equivalent
continuous
exposure
concentration
(
ConcADJ
).
Animals
in
this
study
were
dosed
for
6
hours
per
day,
for
five
days,
then
not
dosed
for
two
days,
and
dosed
again
for
five
days
and
sacrificed
at
the
end
of
the
12th
day;
hence,
continuous
exposure
duration
adjustment
was
made
as
follows:

Study
concentration
x
(
6
hours/
24
hours)
x
(
10
days/
12
days)
=
ConcADJ
Second,
the
duration­
adjusted
concentrations
(
ConcADJ
)
were
converted
to
human
equivalent
concentrations
(
ConcHEC
)
to
account
for
differences
in
the
respiratory
tract
anatomy
and
physiology
for
the
test
species
versus
humans.
This
conversion
is
made
as
follows:

ConcADJ
x
RDDR
=
ConcHEC
The
RDDR
is
the
Regional
Dose
Deposition
Ratio
calculated
using
U.
S.
EPA's
RDDR
software
program
(
USEPA,
1994).
The
RDDR
depends
on
the
characteristics
of
the
particle
size
distribution
(
e.
g.,
mass
median
aerodynamic
diameter,
and
geometric
standard
deviation),
the
test
species
and
body
weight,
and
the
region
of
the
respiratory
tract
(
or
extrarespiratory
tissue
target
if
applicable)
affected
by
exposure.
Appropriate
particle
size
characteristics
to
use
as
inputs
into
the
RDDR
software
were
obtained
from
a
recent
communication
from
DuPont
(
see
attached).
For
the
Kennedy
et
al.
(
1986)
study,
the
test
sex
and
species
was
male
rats.
Since
body
weight
data
were
provided
in
the
study,
these
data
were
used
directly
in
the
RDDR
program.
The
mean
body
weight
data
on
day
5
of
exposure
was
used
for
this
calculation,
rather
than
the
study­
day
10
body
weight
data.
The
day
5
body
weights
were
36
used
because
there
was
evidence
of
changes
in
body
weight
over
the
12­
day
study
period,
and
therefore,
this
value
was
judged
as
the
best
estimate
of
the
mean
body
weight
over
the
period
of
exposure.

The
CATT
panel
considered
two
potential
critical
effects
for
deriving
the
pRfC;
increased
liver
weight
and
overt
toxicity
secondary
to
pulmonary
toxicity.
The
RDDR
for
extrarespiratory
tissues
was
the
most
appropriate
value
to
use
in
calculating
human
equivalent
concentrations
for
assessing
the
liver
effects.
The
RDDR
program
calculates
values
for
a
variety
of
different
regions
of
the
respiratory
tract.
The
CATT
panel
agreed
that
the
overt
toxicity
of
C8
was
likely
due
to
particle
overload,
as
supported
by
pulmonary
edema
in
the
acute
study
reported
in
the
same
paper
(
Kennedy
et
al.,
1986).
Therefore,
the
RDDR
for
the
pulmonary
region
was
selected
as
most
appropriate
respiratory
tract
region
for
calculating
the
human
equivalent
concentrations.
The
calculation
of
the
human
equivalent
concentrations
used
in
the
dose­
response
assessment
is
summarized
in
Table
10.

Table
10.
Calculation
of
Human
Equivalent
Concentrations
for
Kennedy
et
al.
(
1986)

Extrarespiratory
Pulmonary
Study
Concentrationa
ConcADJ
RDDRb
ConcHEC
RDDR
ConcHEC
1.0
0.21
2.956
0.62
0.513
0.11
7.6
1.6
2.954
4.7
0.512
0.81
84
17
2.973
52
0.521
9.1
a.
All
concentrations
reported
in
the
table
are
in
units
of
mg/
m3
.
b.
The
RDDR
values
are
taken
from
the
EPA
RDDR
Program
Output
provided
in
the
attachment
Benchmark
Concentration
Modeling
The
CATT
panel
further
recommended
that
benchmark
concentration
(
BMC)
modeling
be
performed
for
the
increased
liver
weight
endpoint
from
the
Kennedy
et
al.
(
1986)
study.
The
published
version
of
the
study
did
not
provide
standard
deviations
to
accompany
the
group
mean
data,
and
therefore,
BMC
modeling
could
not
be
performed
at
the
time
of
the
CATT
panel
meeting.
Subsequent
to
the
meeting,
the
individual
liver
weight
data
for
this
study
were
obtained
from
DuPont
(
see
attached).
The
individual
animal
data
were
used
to
calculate
group
mean
and
standard
deviations.
These
data
were
then
employed
for
the
BMC
analyses.

The
modeling
was
conducted
according
to
draft
EPA
guidelines
(
U.
S.
EPA,
2000)
using
Benchmark
Dose
Software
(
BMDS
version
1.3.1),
available
from
the
U.
S.
EPA
website
(
U.
S.
EPA,
2002).
The
endpoints
of
interest
with
respect
to
C8
liver
toxicity
were
continuous
rather
than
quantal
(
e.
g.,
incidence
data)
in
nature.
Therefore
the
absolute
and
relative
liver
weight
data
sets
were
modeled
using
the
linear,
Hill,
power,
and
polynomial
models.
An
acceptable
fit
to
the
data
was
defined
as
a
goodness­
of­
fit
p­
value
greater
than
or
equal
to
0.1,
or
a
perfect
fit
when
there
were
no
degrees
of
freedom
for
a
formal
statistical
test
of
fit.
Choice
of
0.1
is
consistent
with
current
U.
S.
EPA
guidance
for
BMD
modeling
(
U.
S.
EPA,
2000).
Goodness­
of­
fit
statistics
are
not
designed
to
compare
different
models,
particularly
if
the
different
models
have
different
numbers
of
parameters.
Within
a
family
of
models,
adding
parameters
generally
improves
the
fit.
BMDS
reports
the
Akaike
Information
Criterion
(
AIC)
to
aid
in
comparing
the
fit
of
different
models.
When
comparing
the
fit
of
two
or
more
37
models
to
a
single
data
set,
the
model
with
the
lesser
AIC
was
considered
to
provide
a
superior
fit.
The
benchmark
response
(
BMR)
level
used
for
this
analysis
was
set
at
a
standard
deviation
(
SD)
value
of
1.0.
This
value
was
chosen
based
on
EPA
draft
guidelines
for
BMC
analysis
(
U.
S.
EPA,
2000),
in
the
absence
of
a
clear
biological
rationale
for
selecting
an
alternative
response
level.

The
following
guidance
was
followed
with
regard
to
the
choice
of
the
Benchmark
Concentration
Lower
Limit
(
BMCL)
to
use
as
a
point
of
departure
for
calculation
of
the
pRfC.
This
guidance
is
consistent
with
recommendations
in
U.
S.
EPA's
BMC
guidance
(
2000).
For
each
endpoint,
the
following
procedure
is
recommended:

1.
Models
with
an
unacceptable
fit
are
excluded.

2.
If
the
BMCL
values
for
the
remaining
models
for
a
given
endpoint
are
within
a
factor
of
3,
no
model
dependence
is
assumed,
and
the
models
are
considered
indistinguishable
in
the
context
of
the
precision
of
the
methods.
The
models
are
then
ranked
according
to
the
AIC,
and
the
model
with
the
lowest
AIC
is
chosen
as
the
basis
for
the
BMCL.

3.
If
the
BMCL
values
are
not
within
a
factor
of
3,
some
model
dependence
is
assumed,
and
the
lowest
BMCL
is
selected
as
a
reasonable
conservative
estimate,
unless
it
is
an
outlier
compared
to
the
results
from
all
of
the
other
models.
Note
that
when
outliers
are
removed,
the
remaining
BMCLs
may
then
be
within
a
factor
of
3,
and
so
the
criteria
given
in
item
2
would
be
applied.

4.
The
BMCL
values
from
all
modeled
endpoints
are
compared,
along
with
any
NOAELs
or
LOAELs
from
data
sets
that
were
not
amenable
to
modeling,
and
the
lowest
NOAEL
or
BMCL
is
chosen.

The
BMC
results
are
summarized
in
Table
11
and
the
individual
BMDS
model
run
output
is
provided
in
the
attachments.

For
modeling
of
the
absolute
liver
weight
data
set,
a
constant
variance
model
was
appropriate
(
see
test
2
in
the
BMDS
output).
The
power
and
polynomial
models
both
defaulted
to
a
linear
model.
None
of
these
linear
models
fit
the
data
well.
The
Hill
model
provided
an
excellent
fit
to
the
data,
as
indicated
by
visual
inspection
of
the
fit
and
the
comparison
of
the
maximum
likelihood
estimates
for
the
fitted
model
to
the
optimum
model
(
shown
as
model
A1
in
the
BMDS
output).
The
linear
models
failed
to
provide
an
adequate
fit
to
the
full
data
set,
since
they
did
not
accommodate
the
plateau
of
the
concentration­
response
curve
between
the
mid­
and
high­
concentrations.
BMC
modeling
was
redone
using
a
truncated
data
set
(
high
concentration
group
removed)
to
optimize
the
fits
of
these
models.
Removing
the
high
concentration
resulted
in
good
fits
for
the
linear
models
(
the
power
and
polynomial
models
again
defaulted
to
linear)
as
indicated
by
the
AIC
and
goodness­
of­
fit
p­
values.
The
Hill
model
could
not
be
run
with
the
truncated
data
set
since
at
least
four
concentration
groups
are
required
to
provide
a
model
fit.

Adequate
fits
to
the
data
were
achieved
when
the
high
concentration
data
were
removed.
An
argument
could
be
made
for
using
these
results
as
the
best
estimate
for
the
data
set,
since
an
adequate
fit
was
achieved
with
fewer
parameters
than
for
the
Hill
model
using
the
full
data
set.
However,
the
BMCL
estimate
for
the
full
data
set
was
on
the
border
of
3­
fold
lower
than
for
the
truncated
data
set,
which
would
suggest
that
the
lower
BMCL
should
be
selected.
Furthermore,
comparison
of
the
chi
square
residuals
in
the
range
of
the
NOAEL
concentration
suggests
that
the
Hill
model
provided
a
better
fit
of
the
data
in
the
low
concentration
region
than
the
linear
models
using
the
truncated
data.
Finally,
since
38
there
was
no
biological
rationale
for
removing
the
high
concentration
data
from
the
modeling,
an
adequate
model
fit
for
the
full
data
set
is
preferred
over
the
model
fit
for
the
truncated
data
set.
Based
on
these
considerations,
the
BMC
of
0.78
mg/
m3
and
corresponding
BMCL
of
0.33
mg/
m3
are
considered
the
best
estimates
for
the
absolute
liver
weight
data
set.

The
relative
liver
weight
data
displayed
a
similar
plateau
between
the
mid­
and
high­
concentration
groups.
The
linear,
power,
and
polynomial
models
all
failed
to
provide
an
adequate
fit.
As
for
the
absolute
liver
weight
data,
the
Hill
model
provided
an
excellent
fit
to
the
data,
but
in
this
case
failed
to
calculate
a
BMCL.
In
the
absence
of
an
adequate
BMCL
estimate
for
any
of
the
models
using
the
full
data
set,
the
data
were
remodeled
with
the
high
concentration
group
data
removed.
The
power
and
polynomial
models
were
nearly
linear,
as
indicated
by
the
parameter
estimates
in
the
BMDS
output.
The
linear,
power,
and
polynomial
models
all
provided
a
similar,
and
very
good
visual
fit
to
the
data.
The
goodness­
of­
fit
statistic
for
the
linear
model
was
0.9.
Although
BMDS
did
not
calculate
the
goodness­
of­
fit
p­
values
for
the
power
and
polynomial
models,
inspection
of
the
maximum
likelihood
estimates
for
these
fitted
models
as
compared
to
the
optimum
model
(
model
A1
in
the
BMDS
output)
confirmed
the
good
fit.
The
linear
model
provided
a
similar
BMC
and
BMCL
estimate
as
the
power
and
polynomial
models,
but
required
less
parameters
to
do
so
(
i.
e.,
as
reflected
in
the
lower
AIC).
Therefore,
the
BMC
of
1.3
mg/
m3
and
the
corresponding
BMCL
of
0.94
mg/
m3
are
considered
the
best
estimates
for
the
data
set
for
relative
liver
weight.

At
the
time
of
the
meeting
the
CATT
panel
did
not
provide
a
recommendation
on
whether
absolute
or
relative
liver
weight
should
be
considered
more
appropriate
as
the
critical
effect.
Both
of
these
measures
were
significantly
increased
beginning
in
the
7.6
mg/
m3
study
concentration
group.
One
would
not
expect
a
difference
in
the
sensitivity
of
these
two
measures
in
this
case,
because
there
was
no
change
in
body
weight
(
the
basis
for
calculating
relative
liver
weight)
at
the
NOAEL.
Therefore,
both
absolute
and
relative
liver
weight
changes
are
considered
to
be
an
adequate
basis
for
the
critical
effect.
Based
on
this
consideration,
the
lower
of
the
BMCL
estimates
for
the
absolute
and
relative
liver
weight
changes
is
the
most
appropriate
basis
for
deriving
the
pRfC.
The
BMC
of
0.78
mg/
m3
with
the
corresponding
BMCL
of
0.33
mg/
m3
for
increased
absolute
liver
weight
are
the
best
estimates
from
the
BMC
modeling
results.
The
BMCL
of
0.33
mg/
m3
is
the
most
appropriate
choice
as
the
critical
effect
level
for
derivation
of
the
pRfC,
because
the
BMCL
is
lower
than
either
the
NOAEL
of
0.61
mg/
m3
for
liver
effects
or
the
NOAEL
of
0.81
mg/
m3
for
pulmonary
effects
in
this
study.

Selection
of
uncertainty
factors
As
described
in
the
technical
meeting
notes,
the
CATT
panel
unanimously
agreed
on
the
choice
of
3
for
extrapolation
from
an
animal
study
(
UFA
),
a
factor
of
10
to
account
for
variability
in
human
sensitivity
(
UFH
),
and
a
factor
of
1
for
extrapolation
from
study
NOAEL
or
BMDL
(
UFL
).
The
CATT
panel
considered
the
selection
of
U.
S.
EPA's
other
two
factors,
for
extrapolation
from
a
study
of
less­
than­
lifetime
duration
(
UFS
)
and
for
database
insufficiencies
(
UFD
),
to
be
dependent
on
whether
liver
or
lung
was
ultimately
selected
as
the
critical
effect.
The
panel
was
not
unanimous
in
selection
of
the
UFs
or
UFD
for
either
organ,
but
a
clear
majority
vote
was
obtained
for
these
UFs
regarding
liver
toxicity.

39
Based
on
the
liver
as
a
critical
effect,
panel
members
recommended
values
of
either
1
(
one
vote),
3
(
six
votes)
or
10
(
1
vote)
for
UFS
,
and
values
of
3
(
six
votes)
or
10
(
two
votes)
for
UFD
.
Therefore,
based
on
the
liver
as
the
critical
effect,
the
composite
UF
would
range
from
100
to
1000,
depending
on
the
selection
of
the
values
for
UFS
and
UFD
.
The
majority
vote
of
the
CATT
panel
(
Table
5)
supported
a
factor
of
3
for
UFS
and
3
for
UFD
.
Based
on
these
values,
a
composite
UF
of
300
for
liver
effects
was
calculated.

Based
on
the
lung
as
the
critical
effect,
panel
members
recommended
values
of
either
1(
three
votes),
3
(
three
votes)
or
10
(
two
votes)
for
UFS
,
and
values
of
1
(
one
vote),
3
(
five
votes),
and
10
(
two
votes)
for
UFD
.
Therefore,
with
the
lung
as
the
critical
effect
the
composite
UF
would
range
from
30
to
3000.
The
majority
of
the
CATT
panel
supported
a
value
of
3
for
UFD
based
on
lung
effects.
A
clear
majority
vote
was
not
determined
for
any
one
value
for
the
UFS
;
however,
six
votes
were
cast
for
a
value
lower
than
10
and
five
votes
were
cast
for
a
value
higher
than
one;
thus
the
median
value
of
3
would
be
a
reasonable
choice.
Therefore,
values
of
3
for
both
UFD
and
UFS
for
lung
effects
would
also
result
in
a
composite
UF
of
300.

However,
it
is
important
to
note
that
the
panel
could
not
arrive
at
a
consensus
on
the
overall
magnitudes
of
UFS
and
UFD
,
because
of
the
numerous
uncertainties
with
the
inhalation
database.
The
resulting
range
in
the
uncertainty
factor
was
generally
considered
reasonable
by
the
panel,
with
values
falling
within
this
range
being
indistinguishable
from
each
other.

Calculation
of
the
pRfC
Liver
toxicity
was
identified
as
the
critical
effect
because
it
was
more
sensitive
to
C8
than
the
lung
(
i.
e.,
liver
toxicity
had
a
lower
NOAEL
or
BMCL
than
lung),
the
composite
UF
ranged
from
100
to
1000
and
was
300
based
on
the
majority
vote.

The
pRfC
is
calculated
as
follows:

pRfC
(
mg/
m3
)
=
critical
effect
level
/
composite
UF
pRfC
range
=
0.33
/
1000
=
0.00033
mg/
m3
(
or
rounded
to
0.3
µ
g/
m3
)
to
=
0.33
/
100
=
0.0033
mg/
m3
(
or
rounded
3.3
µ
g/
m3
)

pRfC
(
majority
vote)
=
0.33
/
300
=
0.0011
mg/
m3
(
or
rounded
to
1
µ
g/
m3
)

Therefore,
the
recommended
pRfC
based
on
the
majority
vote
for
a
composite
UF
of
300
is
1
microgram
per
cubic
meter
of
air
(
µ
g/
m3
)
with
a
range
from
0.3
µ
g/
m3
to
3.3
µ
g/
m3
.

40
Table
11.
Benchmark
Dose
Modeling
Results
for
C8a
Model/
Data
Set
AIC
P­
value
BMCb
BMCL
Absolute
Liver
Weight
 
All
Data
Modeled
Linear
62.58c
<
0.001d
31
19
Hill
48.67
1.0e
0.78
0.33
Power
62.58c
<
0.001
31
19
Polynomial
62.58c
<
0.001
31
19
Absolute
Liver
Weight
­
High
Concentration
not
Modeled
Linear
38.22c
0.72
1.6
1.1
Power
38.22c
0.29d
1.6
1.1
Polynomial
38.22c
0.72
1.6
1.1
Hill
Insufficient
Number
of
data
points
to
run
model
Relative
Liver
Weight
 
All
Data
Modeled
Linear
­
167.65c
<
0.001
21
15
Hill
­
184.29
1.0e
1.1
Failed
Power
­
167.65c
<
0.001
21
15
Polynomial
­
167.65c
<
0.001
21
15
Relative
Liver
Weight
­
High
Concentration
not
Modeled
Linear
­
137.04c
0.90
1.3
0.94
Power
­
135.05c
Failed
1.5
0.94
Polynomial
­
135.05c
1.0e
1.5
0.94
Hill
Insufficient
Number
of
data
points
to
run
model
a
Modeling
was
performed
based
on
absolute
and
relative
liver
weight
results
reported
in
Kennedy
et
al.
(
1986).

b
BMC
and
BMCL
are
based
on
benchmark
response
of
1SD.
Results
are
presented
in
units
of
mg/
m3.
BMC
and
BMCL
estimates
in
bold
type
are
the
estimates
judged
to
be
the
best
estimates
for
each
endpoint.
"
Failed"
indicates
that
BMDS
was
unable
to
produce
the
estimate
or
the
information
required
to
be
able
to
present
a
value.

c
Corrected
from
erroneous
BMDS
output.
Errors
were
identified
in
the
degrees
of
freedom
(
DF)
provided
in
the
output
for
the
fitted
model
in
several
cases.
For
these
cases,
the
AIC
was
calculated
independently
using
the
log
likelihoods
provided
in
the
output
and
the
correct
number
of
DF.
Similarly,
the
goodness­
of­
fit
p­
values
were
corrected
by
calculating
manually
the
chi
square
p­
value
using
the
appropriate
number
of
DF.

d
This
model
provided
an
identical
fit
to
the
linear
and
polynomial
models.
The
reported
P­
value
reflects
a
difference
in
the
maximum
likelihood
estimate
for
the
comparison
model
(
Model
A1
in
the
BMDS
output)
across
the
three
models.
This
difference
the
maximum
likelihood
estimate
should
be
the
same
for
all
three
models,
since
this
estimate
is
model
independent.

e
A
fit
that
maximizes
the
likelihood
is
assigned
a
p­
value
of
1.0,
even
if
there
were
no
degrees
of
freedom
for
a
formal
statistical
test.
The
maximized
likelihood
is
given
by
model
A1
for
constant
variance
models
and
model
A2
for
non­
constant
variance
models.
Models
A1
and
A2
are
independent
of
the
model
chosen
to
fit
the
data
(
e.
g.,
power,
polynomial,
Hill
model)
and
provide
the
best
match
possible
to
the
mean
and
standard
deviation
for
each
dose
level.

41
Calculation
of
an
Air
Screening
Level
As
described
in
the
technical
meeting
notes,
U.
S.
EPA
Region
9
methodology
was
judged
by
the
CATT
panel
to
be
an
appropriate
basis
for
deriving
the
air
screening
level.
The
following
standard
formula
was
used
to
calculate
the
air
screening
level:

Air
Screening
Level
(
µ
g/
m3
)
=
THQ
x
RfDi
x
BW
x
AT
x
1000
EF
x
ED
x
air
IR
Note:
RfDi
(
mg/
kg­
day)
=
RfC
x
20m3/
d
(
IR)
70
kg
(
BW)

Where:
THQ
=
Target
Hazard
Quotient,
assumed
to
be
1
RfDi
=
The
RfC
expressed
in
terms
of
dose,
mg/
kg­
day
RfC
=
The
inhalation
reference
concentration
(
mg/
m3
)
BW
=
Body
weight,
assumed
to
be
70
kg
for
adults
AT
=
Averaging
time,
10,950
days,
the
exposure
duration
expressed
in
days
EF
=
Exposure
Frequency,
350
days/
year,
the
average
number
of
days
each
year
people
are
exposed
ED
=
Exposure
duration,
30
years,
the
average
number
of
years
people
are
exposed
IR
=
Inhalation
rate
for
air
screening
levels,
20
m3
/
day
Using
this
equation,
the
air
screening
level
ranges
from
0.3
µ
g/
m3
to
10
µ
g/
m3
.
Using
a
reasonable
median
value,
the
air
screening
level
would
be
1.1
µ
g/
m3
(
or
rounded
to
1
µ
g/
m3
).

2.4
SUMMARY
OF
FINDINGS
The
key
studies,
critical
effects
and
levels,
uncertainty
factors,
and
provisional
risk
factors
developed
by
the
CATT
toxicologists
are
summarized
in
Table
12.

42
Table
12.
Summary
of
RfD
and
RfC
Values
for
C8
Determined
by
the
CATT
Toxicologists
Reference
Critical
Effect
Critical
Effect
Levela
UFA
UFH
UFL
UFS
UFD
Composite
UFb
RfD/
RfC
Oral
Studies
Palazzolo
et
al.
(
1993)
c
90­
day
rat
study
Increased
relative
liver
weight
with
histopathology
in
male
rats
0.47
(
NOAEL
in
males)

0.72
(
BMDL)
10
10
1
1
1
100
0.005
0.007
York
et
al.
(
2002)

Two­
Generation
rat
study
Increased
liver
weight
in
male
rats,
supported
by
histopathology
at
higher
doses
(
histopathology
was
not
examined
at
the
lowest
dose,

but
incidence
of
hypertrophy
was
100%
at
next
highest
dose).
0.42
(
BMDL
in
males)
d
10
10
1
1
1
100
0.004
RikerLaboratories
(
1983)

Two­
year
rat
study
Hepatic
megalocytosis
in
male
rats.
0.73
(
BMDL
in
males)
10
10
1
1
1
100
0.007
Thomford
et
al.

(
2001)
e
26­
week
cynomolgus
monkey
study
Decreased
thyroid
hormone
levels
in
male
cynomolgus
monkeys,
and
supported
by
a
NOAEL
at
the
same
dose
for
clinical
signs
of
toxicity
in
the
co­
critical
rhesus
monkey
study
(
Goldenthal
et
al.,

1978b)
3
­
10
(
LOAEL
in
males)
10
10
3
3
1
1000
0.003
­

0.01
43
Inhalation
Studies
Kennedy
et
al.
(
1986)
f
Two­
week
rat
study
Increased
liver
weight
supported
by
histopathology
and
clinical
chemistry
in
male
rats
0.61(
NOAEL
­
HECER
males)

0.33
(
BMCL,

BMC
0.78
absolute
liver
weight)

0.94(
BMCL,

BMC
1.3
relative
liver
weight)
3
10
1
3
3
300
1
Dermal
Studies
Kennedy
et
al.
(
1985)
f
Two­
week
rat
study
Increased
liver
weight
in
male
rats
4.2g
(
LOAEL
in
males)
Data
Inadequate
a.
Oral
and
Dermal
effect
levels
and
RfDs
are
presented
in
units
of
mg/
kg­
day,
while
the
inhalation
critical
effect
level
and
RfC
is
presented
in
units
of
mg/
m3
b.
Areas
of
uncertainty
addressed
by
uncertainty
factors
are:
animal
to
human
extrapolation
(
A);
intrahuman
variability
and
protection
of
sensitive
subpopulations
(
H);
extrapolation
from
a
LOAEL
to
a
NOAEL(
L);
extrapolation
from
a
subchronic
to
chronic
exposure
(
S);
and
lack
of
a
complete
database
(
D)

c.
The
subchronic
study
by
Goldenthal
et
al.
(
1978a)
could
serve
as
a
supporting
study
for
liver
effects
in
rats.

d.
BMDL
is
the
95%
lower
confidence
limit
on
the
dose
corresponding
to
a
response
level
of
10%
or
an
increase
of
1SD
in
the
continuous
endpoint
being
assessed.
Only
modeling
results
that
provided
the
lowest
value
and
provided
an
adequate
fit
to
the
data
are
provided.

e.
The
subchronic
study
in
rhesus
monkeys
by
Goldenthal
et
al.
(
1978b)
is
a
co­
critical
study
for
clinical
signs
of
toxicity
in
monkeys.

f.
These
studies
are
not
adequate
for
derivation
of
an
IRIS
quality
RfD/
RfC
of
even
low
confidence.
The
values
shown
could
be
used
to
derive
a
provisional
value.
Derivation
of
the
RfC
or
RfD
via
route­
to­
route
extrapolation
is
not
supported
by
the
available
toxicokinetic
data.
Consensus
on
the
values
for
UFS
and
UFD
was
not
reached
by
the
panel;
however,
a
majority
vote
was
obtained
for
a
value
of
3
for
both
these
UFs
in
reference
to
liver
as
the
target
organ.
See
text
of
this
report
for
ranges
of
UFs
and
SLs
based
on
the
range
distribution
of
the
votes
for
UFs.

g.
4.2
mg/
kg­
day
reflects
the
study
dose
of
20
mg/
kg
adjusted
for
discontinuous
exposure.

44
I
agree
that
the
notes
as
presented
accurately
reflect
the
panel's
discussion
and
conclusions
during
the
May
6­
7,
2002
C8
Assessment
of
Toxicity
Toxicologists
Panel
Meeting,
and
that
the
post
meeting
actions
taken
to
develop
the
pRfC
and
Air
Screening
Level
are
in
accordance
with
the
instructions
provided
to
TERA
by
the
panel.
(
Original
signatures
are
on
file
at
DEP.)

_______________________________________
_____________________
John
Cicmanec,
D.
V.
M.,
M.
S.,
ACLAM,
USEPA
ORD
Date
_______________________________________
_____________________
Joan
Dollarhide,
M.
S.,
M.
T.
S.
C.,
J.
D.,
TERA
Date
_______________________________________
_____________________
Michael
Dourson,
Ph.
D.,
D.
A.
B.
T.,
TERA
Date
_______________________________________
_____________________
Gerald
Kennedy,
DuPont
Date
_______________________________________
_____________________
Andrew
Maier,
Ph.
D.,
C.
I.
H.,
TERA
Date
_______________________________________
_____________________
Samuel
Rotenberg,
Ph.
D.,
USEPA
Region
3
Date
_______________________________________
_____________________
Jennifer
Seed,
Ph.
D.,
USEPA
Headquarters
OPPT
Date
_______________________________________
_____________________
Dee
Ann
Staats,
Ph.
D.,
DEP
(
Chairperson)
Date
_______________________________________
_____________________
John
Wheeler,
Ph.
D.,
D.
A.
B.
T.,
ATSDR
Date
_______________________________________
_____________________
John
Whysner,
M.
D.,
Ph.
D.,
D.
A.
B.
T.
Date
45
46
3.
0
COMPARISON
OF
SCREENING
LEVELS
TO
SITE­
RELATED
DATA
After
the
SLs
for
air,
water,
and
soil
were
determined,
DEP
compared
these
SLs
to
the
site­
related
data
that
has
been
collected
to
date.
These
comparisons
are
summarized
below.
The
work
of
the
CATT
was
only
one
facet
of
an
investigation
that
continues
beyond
the
issuance
of
this
report.
The
GIST
is
expected
to
issue
a
report
of
the
groundwater
and
surface
water
data
in
early
2003.
The
air
modeling
effort
continues
and
is
currently
focusing
on
determining
the
results
of
the
air
emissions
reduction
efforts
by
DuPont
required
in
the
consent
order
as
a
50%
reduction
in
overall
emissions
(
both
air
and
water)
by
the
end
of
2003.
Upgrades
were
completed
in
June
2002
which
included
the
installation
of
a
new
scrubber
and
increased
height
of
the
primary
C8
emissions
stack.

Water
To
date,
of
the
188
samples
collected
from
private
wells,
cisterns,
and
springs,
50
were
used
for
drinking
water
and
none
exceeded
the150
ppb
health
protective
water
SL
for
C8.
Also
to
date,
nine
public
water
supply
facilities
in
West
Virginia
have
been
analyzed
for
C8,
including
Belleville
Locks
and
Dam,
Blennerhassett
Island,
General
Electric,
Lubeck
Public
Service
District
(
PSD),
Mason
County
PSD,
Parkersburg
PSD,
Racine
Locks
and
Dam,
New
Haven
Water
Department,
and
Ravenswood.
None
of
the
drinking
water
from
these
facilities
contained
concentrations
of
C8
that
exceeded
the
150
ppb
water
SL.
In
fact,
the
concentrations
of
C8
in
public
water
supplies
were
all
below
2
ppb,
below
15
ppb
in
private
non­
drinking
water,
and
below
3
ppb
in
private
drinking
water
wells
in
West
Virginia.
Samples
were
collected
from
Ohio
public
and
private
water
supplies.
Although
C8
levels
in
some
Ohio
private
water
supplies
were
higher
than
those
detected
in
West
Virginia,
none
of
these
samples
contained
C8
concentrations
above
the
water
SL.
These
data
have
been
provided
to
Ohio
EPA
and
DEP
will
continue
to
share
information
with
throughout
the
remainder
of
this
investigation.
The
DEP
notes
that
the
water
SL
is
higher
than
DuPont's
internal
community
exposure
guidelines
for
drinking
water
of
1
or
3
ppb;
however,
these
guidelines
were
developed
in
the
early
1990s
and
based
solely
on
a
two­
week
inhalation
study
from
1986.
Since
then
significant
additional
toxicological
data
have
been
collected
and
the
CATT
water
SL
is
based
on
a
comprehensive
examination
of
all
available
information.
Sampling
of
the
Ohio
River
has
begun;
preliminary
analytical
results
are
expected
from
the
laboratory
in
September
2002.
To
date,
no
analysis
has
been
performed
to
measure
C8
in
soils
in
West
Virginia
on
private
property;
therefore,
no
comparison
can
be
made
to
the
soil
SL.

Air
Mathematical
computer
models
that
incorporate
weather
conditions,
chemical
characteristics,
and
facility
measurements
were
utilized
by
DEP
to
simulate
the
ambient
air
concentrations
of
C8.
Based
on
actual
emissions
data
from
the
DuPont
WW
facility
for
the
year
2000,
the
DEP
modeling
efforts
predicted
a
maximum
C8
concentration
in
air
of
approximately
2.7
µ
g/
m3
at
the
facility
fence
line
along
the
Ohio
River.
The
maximum
modeled
C8
air
concentration
in
the
West
Virginia
residential
area
adjacent
to
the
facility
was
approximately
0.2
µ
g/
m3
annual
average.
Predicted
C8
air
concentrations
across
the
Ohio
River
from
the
WW
facility
in
Ohio
residential
areas
were
greater
than
those
predicted
in
residential
areas
in
West
Virginia.
These
data
have
been
provided
to
Ohio
EPA
and
DEP
will
continue
to
share
information
with
Ohio
EPA
throughout
the
remainder
of
this
investigation.
Results
of
similar
subsequent
air
modeling
efforts
conducted
by
DuPont
are
consistent
with
those
of
the
DEP.
Air
modeling
information
can
be
obtained
from
the
DEP
Division
of
Air
Quality.

The
DEP's
Divisions
of
Water
Resources
and
Air
Quality
are
currently
reviewing
all
relevant
air
and
water
data
to
determine
DuPont's
compliance
with
the
November
2001
consent
order
between
DEP
and
DuPont.
