DRAFT
FINAL
REPORT
on
COMPARATIVE
EVALUATION
OF
VITELLOGENIN
METHODS
FOR
MEDAKA
AND
ZEBRA
FISH
EPA
CONTRACT
NUMBER
68­
W­
01­
023
WORK
ASSIGNMENT
2­
26
July
2003
Prepared
for
LES
TOUART,
PH.
D.
WORK
ASSIGNMENT
MANAGER
U.
S.
ENVIRONMENTAL
PROTECTION
AGENCY
WASHINGTON,
DC
20460
BATTELLE
505
King
Avenue
Columbus,
Ohio
43201
OPPT­
2003­
0027­
0004
RECEIVED
OPPT
NCIC
2003
August
4
1:
26PM
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
ii
LIMITED
DISTRIBUTION
NOTICE
This
document
is
transmitted
in
advance
of
patent
clearance.
This
document
is
not
to
be
published
nor
its
contents
otherwise
disseminated
or
publicly
used
until
approval
for
such
release
or
use
has
been
obtained
from
Battelle
Intellectual
Property
Services,
Pacific
Northwest
National
Laboratory,
Richland,
Washington
99352.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
iii
TABLE
OF
CONTENTS
Page
ACKNOWLEDGMENTS
............................................................................................................................................
iii
1.0
INTRODUCTION..................................................................................................................................................
1
2.0
SAMPLE
PREPARATION
AND
HANDLING.....................................................................................................
2
3.0
SAMPLE
METHODS............................................................................................................................................
3
4.0
LABORATORY
ANALYTICAL
METHODS
......................................................................................................
6
5.0
PARTICIPATING
LABORATORIES...................................................................................................................
7
6.0
DATA
ANALYSIS
...............................................................................................................................................
7
7.0
SUMMARY
OF
RESULTS
...................................................................................................................................
9
8.0
DISCUSSION.......................................................................................................................................................
22
9.0
REFERENCES.....................................................................................................................................................
28
APPENDIX
A.
PARTICIPATING
LABORATORY
PROTOCOLS
.....................................................................
A1
APPENDIX
B.
DATA
FROM
PARTICIPATING
LABORATORIES
..................................................................
B1
APPENDIX
C
RESULTS
......................................................................................................................................
C1
APPENDIX
D
DESCRIPTIVE
STATISTICS
OF
THE
WITHIN­
RUN
VTG
RESULTS
....................................
D1
APPENDIX
E
DESCRIPTIVE
STATISTICS
OF
THE
INTRALABORATORY
RESULTS
..............................
E1
APPENDIX
F
DESCRIPTIVE
STATISTICS
OF
THE
INTRA­
ASSAY
RESULTS.
...........................................
F1
LIST
OF
TABLES
1.
Estrogen
exposure
schedule
for
zebra
fish
and
medaka
..........................................................................................
4
2.
VTG
spike
concentration
in
positive
controls..........................................................................................................
5
3.
Summary
of
reporting
laboratories
and
their
VTG
ELISA
methods........................................................................
8
4.
Species
analyzed
by
participating
laboratories
......................................................................................................
10
5.
Summary
of
codes
for
relative
VTG
concentration,
standards,
and
ELISA
methods
for
zebra
fish
......................
11
6.
Summary
of
standards
employed
by
participating
laboratories
analyzing
zebra
fish
............................................
11
7.
Summary
of
codes
for
relative
VTG
concentration,
standards,
and
ELISA
methods
for
medaka
.........................
17
8.
Summary
of
standards
employed
by
participating
laboratories
analyzing
medaka
................................................
17
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
iv
LIST
OF
FIGURES
Figure
1.
Diagram
of
the
organization
of
one
set
of
zebra
fish
samples
for
analysis.................................................
6
Figure
2.
Mean
of
zebra
fish
triplicate
VTG
values
for
each
of
three
samples
per
concentration
in
the
standard
series,
plotted
by
laboratory:
(
a)
liver;
(
b)
whole
body........................................................
14
Figure
3.
Mean
of
medaka
triplicate
VTG
values
for
each
of
three
samples
per
concentration
in
the
standard
series,
plotted
by
laboratory:
(
a)
liver;
(
b)
whole
body........................................................
20
Figure
4.
VTG
measurement
by
laboratory
for
each
concentration
code
averaged
over
standard
and
assay
for
(
a)
zebra
fish
liver;
(
b)
zebra
fish
whole
body;
(
c)
medaka
liver;
and
(
d)
medaka
whole
body
samples
...........
25
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
v
ACKNOWLEDGMENTS
This
document
was
prepared
by
the
following:

Susan
A.
Thomas,
Valerie
Cullinan,
Roy
Kropp,
Ann
Skillman,
and
Michael
L
Blanton
Battelle's
Pacific
Northwest
Division
1529
West
Sequim
Bay
Road
Sequim,
Washington
98382
USA.

This
report
would
not
be
possible
with
out
the
participation
of
the
following
laboratories,
which
donated
materials
and
time
in
the
application
of
their
methods
to
the
samples
in
this
study:

Biosense
Laboratories,
Norway
Centre
for
Fish
and
Wildlife
Health,
University
of
Bern,
Switzerland
Department
of
Pathology,
Vet.
Medicine,
Swedish
University
of
Agricultural
Sciences,
Sweden
EnBioTec
Laboratories,
Ltd.,
Japan
Environmental
and
Symbiotic
Sciences,
Prefectural
University
of
Kumamoto,
Japan
Institute
of
Biology,
University
of
Southern
Denmark,
Denmark
Los
Angeles
County
Sanitation
District,
USA
National
Institute
for
Environmental
Studies
(
NIES),
Japan
Notox
Safety
&
Environmental
Research,
the
Netherlands
Phylonix
Pharmaceuticals,
Inc.,
USA
Unité
d'Evaluation
des
Risques
Ecotoxicologiques
(
INERIS),
France.

The
authors
gratefully
acknowledge
review
comments
from
Les
Touart
(
U.
S.
Environmental
Protection
Agency
[
EPA]),
Gary
Ankley
(
EPA
Mid­
Continent
Ecology
Division
[
EPA­
MED]).
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
1
1.0
INTRODUCTION
The
U.
S.
Environmental
Protection
Agency
(
EPA)
is
implementing
an
Endocrine
Disruptor
Screening
Program
(
EDSP)
composed
of
a
battery
of
Tier
1
screening
assays
and
Tier
2
tests.
An
international
effort
is
also
underway
to
develop
and
coordinate
screens
and
tests
appropriate
for
use
in
investigating
potential
endocrine
disrupting
chemicals.
The
Organization
for
Economic
Cooperation
and
Development
(
OECD)
has
established
an
Endocrine
Disrupter
Testing
and
Assessment
task
force
(
EDTA)
to
oversee
the
coordination
of
this
effort.
One
of
the
Tier
1
assays
under
development
is
a
short­
term
screening
assay
designed
to
detect
substances
that
interact
with
the
estrogen
and
androgen
systems
of
fish.
It
is
thought
that
the
inclusion
of
the
fish
screening
assay
in
Tier
1
is
important
because
estrogenic
and
androgenic
controls
on
reproduction
and
development
in
fish
may
differ
significantly
from
that
of
higher
vertebrates,
such
that
mammalian
screening
methods
may
not
identify
potential
endocrine
disruptor
chemicals
(
EDCs)
in
this
important
class
of
animals.
The
measurement
of
a
biochemical
marker,
vitellogenin
(
VTG),
in
oviparous
vertebrates
is
generally
agreed
to
be
a
good
indicator
of
estrogenic
and
antiestrogenic
effects,
and
it
is
proposed
as
one
of
several
endpoints
in
the
fish
screening
assay.
VTG
is
a
phospholipoglycoprotein
precursor
to
egg
yolk
protein
that
normally
occurs
in
sexually
active
female
oviparous
fishes,
but
can
be
induced
to
occur
in
males
in
response
to
estrogenic
substances.
Different
methods
are
available
to
assess
VTG
induction
in
fishes,
such
as
measurement
of
the
VTG
protein
with
enzyme­
linked
immunosorbant
assays
(
ELISA)
or
matrix­
assisted
laser
desorption/
ionization­
mass
spectrometery
(
MALDI­
MS),
and
messenger
ribonucleic
acid
(
mRNA)
detection.
Plasma,
liver,
and
whole
body
measurements
have
been
proposed.
The
Validation
Management
Group
for
ecotoxicity
(
VMGeco)
of
the
EDTA
recommended
that
a
survey
of
existing
VTG
analytical
methods
be
undertaken
to
assess
their
relative
comparability.

The
purpose
of
this
study
was
to
coordinate
an
interlaboratory
comparison
of
existing
ELISA
VTG
methods
for
analysis
of
zebra
fish
(
Danio
rerio)
and
medaka
(
Oryzias
latipes)
for
suitability
in
a
routine
screening
program.
This
comparison
was
not
intended
to
be
a
validation
of
a
given
method,
but
an
evaluation
across
methods
to
ascertain
the
qualitative
and/
or
quantitative
comparability
of
the
variety
of
methods
currently
available.

The
objectives
of
the
study
were
the
following:

A.
Prepare
a
standard
evaluation
of
two
tissue
homogenates
(
liver
and
whole
body)
and
plasma
taken
from
each
of
two
species,
zebra
fish
(
Danio
rerio)
and
medaka
(
Oryzias
latipes),
to
provide
a
range
of
VTG
concentrations
produced
in
male
and
female
fish
that
are
either
exposed
or
not
exposed
to
an
estrogen
compound.
The
series
for
each
species
was
to
be
produced
with
1)
uninduced
male,
2)
uninduced
female,
3)
induced
male,
and
4)
induced
female
fish.
In
addition
to
the
standard
series,
a
set
of
positive
control
samples
was
to
be
prepared
using
uninduced
male
tissue
spiked
with
purified
VTG
from
the
appropriate
species.
After
shipment
of
samples
to
participating
laboratories,
an
archive
of
the
standard
evaluation
series
and
controls
was
to
be
created
and
maintained.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
2
B.
Identify
laboratories
to
participate
in
the
analysis
of
the
standard
evaluation
series,
coordinate
transfer
of
the
samples
to
the
participating
laboratories,
and
collect
analytical
results.
Each
laboratory
was
directed
to
employ
the
specific
analytical
technique
it
routinely
uses
to
measure
VTG,
to
report
the
results
of
the
analysis,
and
to
provide
a
detailed
analytical
protocol.
C.
Statistically
compare
the
data
derived
from
the
variety
of
analytical
methods
applied
to
the
standard
series
by
the
participating
laboratories,
and
prepare
a
final
report
that
presents
and
discusses
the
data
provided
by
the
study
participants,
and
the
variability
of
the
results.
D.
Prepare
a
Quality
Assurance
Project
Plan
(
QAPP)
supporting
this
task
to
identify
all
applicable
procedures
and
quality
requirements.

2.0
SAMPLE
PREPARATION
AND
HANDLING
The
two
VTG
standard
evaluation
series
were
to
be
prepared
from
two
homogenates
(
liver
and
whole
body)
and
from
blood
plasma
of
each
species
of
fish.
1
Fish
were
acquired,
a
subset
of
fish
was
exposed
to
estrogen,
and
both
exposed
and
unexposed
fish
were
used
to
prepare
the
standard
series
under
an
animal
care
protocol
reviewed
and
approved
by
the
Battelle's
Pacific
Northwest
National
Laboratory
Animal
Care
Committee
(
accredited
by
the
Association
for
Assessment
and
Accreditation
of
Laboratory
Animal
Care
[
formerly
American
Association
for
the
Accreditation
of
Laboratory
Animal
Care]).
In
the
preparation
of
all
materials
for
the
standard
series,
several
steps
were
employed
to
aid
in
preserving
the
integrity
of
the
samples,
such
as
the
use
of
a
protein­
inhibitor
(
aprotinin),
cold
processing,
and
quick­
freezing
to
stabilize
the
VTG
in
the
samples.

To
generate
samples
of
each
tissue
with
zero
to
low
levels
of
VTG
representing
uninduced
background
concentrations
for
the
standard
series,
a
set
of
adult
male
and
female
fish
of
each
species
was
maintained
without
exposure
to
17
 ­
estradiol.
To
generate
samples
of
each
tissue
with
high
levels
of
VTG
for
the
series
representing
induced
concentrations,
a
set
of
adult
male
and
female
fish
of
each
species
was
exposed
to
a
nominal
concentration
of
300
ng/
L
17
 ­
estradiol
in
the
laboratory
in
a
7­
day
static
renewal
treatment
to
stimulate
the
production
of
VTG.
After
a
1­
week
exposure,
when
maximal
VTG
protein
levels
were
anticipated,
all
of
the
fish
from
both
the
exposed
and
unexposed
sets
were
sacrificed
and
processed
as
necessary
to
obtain
the
required
tissues,
which
were
then
pooled
by
species,
gender,
and
tissue
type,
and
quick­
frozen.
At
a
later
date,
each
composite
was
quickly
thawed
and
used
to
prepare
the
standard
series.
In
summary,
the
approach
outlined
in
more
detail
below
and
in
Section
3
resulted
in
four
samples
within
the
series:
unexposed
(
uninduced)
male,
unexposed
female,
exposed
(
induced)
male,
and
exposed
female.
In
addition,
a
positive
control
was
prepared
from
unexposed
male
tissue
spiked
with
a
known
quantity
of
purified
VTG,
as
the
fifth
sample
to
be
included
in
each
analysis.

 
One
portion
of
male
and
female
fish
from
the
exposed
and
unexposed
groups
was
sacrificed
for
the
preparation
of
whole
body
homogenate.
The
fish
in
each
category
(
e.
g.,

1
However,
blood
plasma
was
subsequently
deleted
from
this
study.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
3
female­
exposed,
or
male­
unexposed,
etc.)
were
individually
weighed,
then
pooled
and
homogenized.
The
composites
were
subsampled
to
prepare
aliquots
of
each
category
for
the
standard
series
to
be
provided
to
each
participating
laboratory
for
analysis.

 

From
another
portion
of
exposed
and
unexposed
male
and
female
fish,
blood
plasma
and
liver
tissue
samples
were
collected.
Due
to
the
small
size
of
zebra
fish
and
medaka,
there
was
a
limited
amount
of
material
available.
Livers
from
individual
fish
were
pooled
and
homogenized
to
create
a
large
composite
sample
from
which
subsamples
were
taken
to
prepare
individual
aliquots
for
the
standard
series.
The
plasma
from
individual
fish
was
pooled
to
create
a
composite
plasma
sample.

 
A
positive
control
for
each
tissue
type
for
each
species
was
prepared
from
unexposed
male
tissue
spiked
with
a
known
quantity
of
purified
VTG
from
the
appropriate
species,
and
was
subsampled
to
prepare
a
set
of
aliquots.
The
purified
VTG
for
each
species
was
purchased
from
Biosense
Laboratories.

 
For
each
of
the
two
fish
species,
multiple
aliquots
of
each
of
the
four
samples
in
the
whole
body
and
liver
standard
series
and
of
a
positive
control
for
each
tissue
were
prepared
as
described
above,
and
stored
at
­
80
°
C
until
required
for
analysis.

 
Shipment
was
coordinated
to
the
participating
laboratories,
each
of
which
assessed
the
level
of
VTG
in
the
samples
by
various
ELISA
methods.
Samples
were
shipped
on
dry
ice
in
a
package
that
included
appropriate
chain­
of­
custody
documentation
and
instructions
for
storage,
sample
handling,
analysis,
and
reporting.
Instructions
and
forms
were
also
sent
electronically
to
each
lab.
The
entire
activity
was
carefully
documented
to
ensure
that
sample
integrity
was
not
compromised.
One
of
11
shipments
for
ELISA
analysis
was
allowed
to
thaw
after
the
shipment
had
been
received,
and
a
replacement
sample
set
was
provided
for
analysis,
following
the
standard
shipping
procedure.

3.0
SAMPLE
METHODS
Approximately
400
adult
zebra
fish
and
400
adult
medaka
were
used
in
this
study.
Both
species
are
small
fish,
and
consequently,
limited
amounts
of
liver
tissue
and
plasma
can
be
collected
from
individuals.
Therefore,
to
generate
sufficient
tissue
for
analysis
by
multiple
laboratories,
the
plan
was
to
collect
samples
from
at
least
20
individuals
of
each
gender
for
each
category
of
tissue
type,
exposure
or
nonexposure
to
estrogen,
and
species.
Because
the
determination
of
the
sex
of
zebra
fish
based
on
morphology
alone
is
not
reliable,
sex
ratios
were
confirmed
by
examination
of
the
gonads
after
sacrifice
of
the
fish.
For
the
treatment
of
live
fish,
the
number
of
male
and
female
zebra
fish
was
estimated
based
upon
an
anticipated
normal
50:
50
sex
ratio
of
the
fish
(
Braunbeck
et
al.
2003).
The
medaka
were
sexed
prior
to
exposure
based
on
external
morphology.
Table
1
summarizes
the
schedule
of
exposure
to
estrogen
for
both
species;
the
staggered
initiation
dates
facilitated
fish
handling.
Fish
were
exposed
to
300
ng/
L
17
 
estradiol
in
a
7­
day
static
renewal
treatment.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
4
Table
1.
Estrogen
exposure
schedule
for
zebra
fish
and
medaka
Species
Initiation
Date
Termination
Date
(
Day
7)
Number
of
Fish
Sex
Type
31
Mar
03
7
Apr
03
40
20
unknown
unknown
whole
body
WQ(
a)

1
Apr
03
8
Apr
03
20
unknown
whole
body
2
Apr
03
9
Apr
03
60
unknown
plasma/
liver
Zebra
fish
3
Apr
03
10
Apr
03
60
unknown
plasma/
liver
27
Mar
03
3
Apr
03
20
20
20
20
male
female
male
male
whole
body
whole
body
whole
body
WQ
9
Apr
03
16
Apr
03
60
female
plasma/
liver
Medaka
10
Apr
03
17
Apr
03
60
male
plasma/
liver
Total
400
(
a)
WQ
Water
quality;
these
fish
were
used
as
control
to
monitor
water
quality,
not
for
generation
of
tissue
samples.
They
were
therefore
not
sacrificed
on
Day
7.

On
Day
7
of
exposure,
approximately
40
exposed
fish
of
each
species
were
anesthetized
with
tricaine
methane
sulfonate
(
MS­
222),
examined
to
determine
sex
in
the
case
of
zebra
fish,
quick­
frozen,
and
stored
at
­
80
°
C
to
be
used
to
prepare
whole
body
induced
homogenate
(
Brion
et
al.
2002).
A
set
of
60
unexposed
fish
of
each
species
was
similarly
collected,
sexed,
and
sacrificed
for
the
whole
body
uninduced
homogenate
preparation.
Whole
body
tissue
homogenate
was
prepared
as
follows:
fish
were
placed
in
ice­
cold
ELISA
assay
buffer
in
a
1:
2
ratio
by
weight
and
homogenized
on
ice
with
a
hand­
held,
ground­
glass
homogenizer.
After
the
resulting
material
was
centrifuged,
the
supernatant
was
harvested,
subsampled
to
20­
µ
L
aliquots,
quick­
frozen
on
liquid
nitrogen,
and
stored
at
­
80
°
C
(
Kang
et
al.
2002).

Similarly,
one
set
of
60
exposed
fish
and
one
set
of
60
unexposed
fish
of
each
species
were
processed
on
Day
7
for
the
liver
and
plasma
standard
series.
From
the
sacrificed
fish,
blood
was
collected
from
the
caudal
vein
via
microcapillary
tubes
(
Kordes
et
al.
2002;
Van
den
Belt
et
al.
2002).
Following
centrifugation
of
the
plasma,
the
supernatant
was
harvested
and
frozen
at
­
80
°
C.
Following
the
collection
of
blood,
livers
were
removed
from
the
same
fish.
The
liver
homogenate
was
prepared
by
placing
the
weighed
livers
in
ice­
cold
ELISA
assay
buffer
in
a
1:
2
ratio
by
weight
and
homogenizing
on
ice
with
a
hand­
held,
ground­
glass
homogenizer.
The
resulting
homogenate
was
centrifuged,
the
supernatant
was
harvested,
subsampled
to
20­
µ
L
aliquots,
quick­
frozen
on
liquid
nitrogen,
and
stored
at
­
80
°
C
(
Kang
et
al.
2002).

Positive
controls
were
created
from
whole
body
and
liver
homogenates
of
unexposed
male
fish
of
each
species
spiked
with
a
known
quantity
of
purified
VTG
from
the
corresponding
species.
All
work
was
conducted
on
ice
to
maintain
the
cold
temperature
of
materials.
The
spiking
rate
for
each
control
depended
on
the
actual
weight
of
lyophilized,
purified
VTG
available
in
each
commercially
prepared
vial
purchased
from
Biosense
Laboratories;
the
specific
concentrations
were
6.25
µ
g
VTG/
mL
homogenate
for
the
zebra
fish
whole
body,
medaka
whole
body,

Table
2.
VTG
spike
concentration
in
positive
controls
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
5
Tissue
VTG
Concentration
(
µ
g/
mL
homogenate)
Zebra
fish
whole
body
homogenate
6.25
Zebra
fish
liver
homogenate
6.25
Medaka
whole
body
6.25
Medaka
liver
6.85
and
medaka
liver
positive
controls,
and
6.85
µ
g
VTG/
mL
homogenate
for
the
zebra
fish
liver
positive
control.
(
Table
2).
The
resulting
positive
controls
were
subsampled
to
20­
µ
L
aliquots,
quick­
frozen
on
liquid
nitrogen,
and
stored
at
­
80
°
C
(
Kang
et
al.
2002).
Care
was
taken
in
each
of
the
sample
preparation
steps
to
collect
and
process
the
samples
in
a
timely
manner
under
cold
conditions
followed
by
quick­
freezing
to
limit
the
time
from
collection
to
storage
and
to
avoid
repeated
freezing
and
thawing.

The
total
number
of
samples
prepared
as
aliquots
for
analysis
was
determined
by
the
number
of
participating
laboratories.
The
volume
of
the
aliquots
was
based
on
the
amount
of
material
available
and
the
analytical
requirements;
20
µ
L
was
determined
to
be
sufficient
and
appropriate
for
all
homogenate
samples.
Cryovials
had
been
selected
as
the
appropriate
containers
for
the
aliquotted
samples.
In
addition
to
the
samples
for
analysis
by
the
participating
laboratories,
the
excess
material
of
every
category
of
the
standard
series,
including
blood
plasma
and
positive
controls,
was
placed
in
long­
term
storage
at
­
80
°
C,
assigned
a
unique
code,
and
entered
into
an
archive
management
system.

Eleven
laboratories
participated
in
the
VTG
ELISA
survey
for
the
analysis
of
the
tissues
of
one
or
both
species.
Based
on
the
requirements
of
each
laboratory,
a
complete
set
of
either
zebra
fish,
medaka,
or
both
was
assembled,
packaged
on
dry
ice
in
well­
insulated
containers,
and
shipped
via
Federal
Express
to
the
testing
facilities
of
the
participating
laboratories.
Also
included
in
the
shipping
boxes
were
chain
of
custody
documents,
sample
receipt
questionnaire,
and
complete
instructions/
information.
To
limit
the
need
to
thaw
and
refreeze
samples,
three
aliquots
of
each
sample
in
the
series
(
e.
g.,
three
vials
of
induced
male
whole
body
homogenate)
were
provided.
Because
the
samples
were
analyzed
fully
blind,
each
vial
was
labeled
with
a
unique
code
that
did
not
reveal
to
the
analyst
the
identity
of
the
samples
in
the
series,
nor
their
corresponding
low
to
high
expected
concentrations
of
VTG.
Only
the
fish
species
and
the
type
of
homogenate,
either
whole
body
or
liver,
were
defined.
The
laboratories
were
instructed
to
analyze
the
contents
of
each
sample
cryovial
in
triplicate,
and
each
cryovial
contained
sufficient
volume
of
material
to
apply
to
three
wells
on
an
ELISA
plate.
Figure
1
illustrates
diagrammatically
the
organization
of
one
set
of
zebra
fish
samples
as
they
were
provided
to
each
laboratory
for
analysis.
One
large
zip­
lock
bag
labeled
by
species
held
three
smaller
bags
containing
the
whole
body
standard
series
and
positive
control
(
15
cryovials
organized
as
5
sets,
each
consisting
of
3
replicate
samples),
liver
homogenate
standard
series
and
control
(
15
cryovials
organized
as
5
sets,
each
consisting
of
3
replicate
samples),
and
purified
VTG
(
1
vial
of
commercially
prepared,
lyophilized
VTG
of
the
appropriate
species).
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
6
4.0
LABORATORY
ANALYTICAL
METHODS
Several
methods
have
been
developed
for
the
quantification
of
VTG
in
blood
plasma,
liver
tissue,
or
whole
body
homogenates;
they
differ
in
sensitivity,
specificity,
and
technical
difficulty.
A
comprehensive
survey
of
the
literature
and
information
from
experts
in
the
field
of
induction
of
VTG
in
fish
(
Battelle
2002)
revealed
that
the
technique
of
ELISA
is
currently
the
most
widely
developed
and
applied
technique,
with
multiple
methods
that
have
been
developed
for
specific
application
to
zebra
fish
and
medaka.
The
various
ELISA
methods
employ
enzyme­
linked
antibodies
and
an
adsorbent
surface
to
detect
specific
antigens
in
solution,
typically
in
one
of
three
general
assay
formats:
competitive,
sandwich,
and
direct
assessments.
Competitive
ELISAs
incorporate
a
step
in
which
the
samples
and
antibody
(
antibody­
capture)
or
labeled
antigen
(
antigen­
capture)
are
incubated
together
prior
to
adding
the
sample
on
the
test
plate.
This
nonequilibrium
design
is
often
used
to
enhance
sensitivity,
and
it
counteracts
potential
preferential
binding
(
Edmunds
et
al.
2000).
Sandwich
ELISAs
employ
two
antibody
preparations
to
detect
the
antigen.
The
antigens
can
recognize
different
epitopes
on
the
target
analyte,
thereby
providing
a
large
degree
of
specificity
and
sensitivity.
In
a
direct
antibody­
capture
ELISA,
the
sample
and
standards
are
adsorbed
directly
on
the
surface
of
the
microwell
plate.
After
incubation,
the
wells
are
blocked
and
anti­
VTG
antibody
is
added
to
bind
to
the
VTG
attached
to
the
well.
As
in
other
ELISAs,
subsequent
steps
culminate
in
the
development
of
color
indicator
that
is
reflective
of
the
amount
of
antigen
present
in
the
sample.

Specific
protocols
employed
by
the
participating
laboratories
(
Appendix
A)
were
applied
to
the
identical
sets
of
zebra
fish
and/
or
medaka
homogenate
samples
supplied
for
the
present
study.
Although
the
majority
of
laboratories
used
the
same
method
(
a
commercially
available
kit),
there
were
four
distinct
ELISAs
applied
to
zebra
fish
homogenates,
and
three
to
medaka.
Further,
in
the
medaka
survey,
one
laboratory
applied
all
three
medaka
methods
using
three
sets
of
samples,
thereby
providing
a
special
opportunity
to
make
a
comparison
in
which
variability
due
to
laboratory
precision
or
accuracy
could
be
reduced.
ZEBRA
Zebra
Fish
W2
Zebra
Fish
L2
Zebra
Fish
S2
Figure
1.
Diagram
of
the
organization
of
one
set
of
zebra
fish
samples
for
analysis
(
codes:
S
indicates
purified
VTG,
W
indicates
whole
body,
L
indicates
liver,
and
2
is
the
laboratory
identification
number;
each
vial
was
also
individually
labeled
with
a
unique
code)
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
7
5.0
PARTICIPATING
LABORATORIES
The
final
list
of
participants
consisted
of
11
laboratories
selected
because
of
their
previous
experience
in
the
measurement
of
VTG
protein,
and
their
willingness
and
ability
to
commit
to
completing
the
analysis
on
a
volunteer
basis.
Each
laboratory
had
established
protocols
in
routine
use.
The
10
laboratories
that
analyzed
VTG
in
the
whole
body
and
liver
homogenate
standard
series
and
contributed
their
results
to
the
survey
are
listed
in
Table
3,
with
the
methods
used.
One
laboratory
was
unable
to
complete
the
analysis
within
the
required
timeline
for
inclusion
in
the
statistical
comparison;
therefore,
this
laboratory's
method
is
described
briefly
in
a
footnote
to
Table
3
and
its
original
data
sheets
are
presented
in
Appendix
B
along
with
the
results
from
the
other
participants.

6.0
DATA
ANALYSIS
Data
analysis
was
intended
to
provide
descriptive
statistics
and
plots
that
allow
a
general
assessment
of
the
objectives
of
the
study.
Statistically,
the
first
objective
was
to
determine
whether
an
increasing
concentration
of
VTG
was
produced
by
the
standard
series.
This
series
was
represented
in
order
of
concentration
from
low
to
high
by
1)
uninduced
male,
2)
uninduced
female,
3)
induced
male,
and
4)
induced
female
fish.
The
second
statistical
objective
was
to
determine
the
analytical
results
and
variation
for
the
set
of
control
and
spiked
VTG
samples
for
each
species.
The
third
statistical
objective
was
to
compare
the
analytical
results
and
variation
of
each
laboratory's
analytical
method,
including
the
standard
and
assay
used.

Analysis
of
the
data
yielded
descriptive
statistics,
including
the
number
of
samples,
means,
standard
deviations
(
SD),
medians,
first
and
third
quartiles,
and
the
coefficient
of
variation
(
CV).
Simple
linear
regression
of
the
ranked
average
VTG
concentration
(
mean
of
the
within­
run
analyses)
and
plots
of
the
analytical
results
against
the
concentration
series
were
used
to
assess
the
strength
of
the
VTG
concentration
trend,
ignoring
the
positive
control.
Tukey's
Honestly
Significant
Difference
(
HSD)
multiple
comparison
test
was
conducted
on
the
ranked
average
VTG
concentrations
to
specifically
determine
whether
neighboring
means
in
the
series
were
significantly
different
(
i.
e.,
the
blank
mean
compared
with
the
uninduced
male
mean,
the
uninduced
male
mean
compared
with
the
uninduced
female
mean,
and
so
on).
Linear
regression
for
each
laboratory
was
also
conducted
on
the
average
VTG
concentrations
observed
for
the
blank
and
the
uninduced
male
data.
The
regression
results
allow
a
test
of
the
null
hypothesis
that
the
slope
equals
0
and
provides
a
measure
of
the
strength
of
the
trend.
The
multiple
comparison
testing,
which
is
less
powerful
than
the
regression
analysis
due
to
the
smaller
degrees
of
freedom
for
testing,
provides
a
test
of
how
quickly
differences
can
be
detected
in
the
series.
Excel
spreadsheet
software
(
Microsoft
Excel)
and
Minitab
statistical
software
(
Minitab
Inc.)
were
used
for
this
analysis.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
8
Table
3.
Summary
of
reporting
laboratories
and
their
VTG
ELISA
methods(
a)

Lab
ID
Participating
Laboratory
ELISA
Method
Reference
zebra
fish
liver
MDL
zebra
fish
whole
body
MDL
medaka
liver
MDL
medaka
whole
body
MDL
1
University
of
Kumamoto
Kumamoto,
Japan
Sandwich
ELISA(
b)
Biosense
Laboratories
(
2002,
2003)
0.49
ng/
mL
0.49
ng/
mL
0.24
ng/
mL
0.24
ng/
mL
2
Biosense
Laboratories
Bergen,
Norway
Sandwich
ELISA
Biosense
Laboratories
(
2002,
2003)
determined
for
two
standards
using
whole
body
homog
(
e.
g.,
151
and
76
ng/
mL
at
1:
300
dilution)
determined
for
the
two
STDs
using
whole
body
homog;
(
e.
g.,
151
and
76
ng/
mL
at
1:
300
dilution)
determined
for
two
standards
using
liver
homog
(
e.
g.,
75
and
1.5
ng/
mL
at
1:
300
dilution
detemined
for
two
standards
using
liver
homog
(
e.
g.,
75
and
1.5
ng/
mL
at
1:
300
dilution)

4
EnBioTec
Laboratories,
Ltd.
Tokyo,
Japan
Sandwich
ELISA,
monoclonal
antibody(
c)
EnBio
(
2002);
Nishi
et
al.
(
2002)
8.2
ng/
mL
(
n=
8)
determined
by
ELISA
(
0.82
using
EnBio
VTG
standard)
16.4
ng/
mL
(
n=
8)
determined
by
ELISA
(
0.82
using
EnBio
VTG
standard)
9.8
ng/
mL
(
n=
8)
determined
by
ELISA
(
0.49
using
EnBio
VTG
standard)
9.8
ng/
mL
(
n=
8)
determined
by
ELISA
(
0.49
using
EnBio
VTG
standard)
5
Notox
Safety
&
Environmental
Research
Hertogenbosch,
The
Netherlands
Sandwich
ELISA
Biosense
Laboratories
(
2002,
2003)
0.12
ng/
mL
determined
based
on
Lab
5'
s
standard
0.12
to
0.24
ng/
mL
determined
based
on
Lab
5'
s
standard
0,4357
ng/
mL,
from
previous
experiments
0,4357
ng/
mL,
from
previous
experiments
7
Prefectural
Universität
Bern
Bern,
Switzerland
Sandwich
ELISA
Biosense
Laboratories
(
2002)
0.5
ng/
mL,
routine
using
Biosense
kit
0.5
ng/
mL,
routine
using
Biosense
kit
8
Phylonix
Pharmaceuticals,
Inc.
Cambridge,
Massachusetts,
USA
Sandwich
ELISA
Biosense
Laboratories
(
2002)
1.999991544
ng/
mL,
determined
by
phylonix
(
n=
8)
1.999991544
ng/
mL,
determined
by
phylonix
(
n=
8)

9
Institute
of
Biology,
University
of
Southern
Denmark
Odense,
Denmark
Direct
noncompetitive
sandwich
ELISA(
d)
Holbech
et
al.
(
2001)
determined
for
two
standards:
16
ng/
mL
(
n=
12)
and
35.6
ng/
mL
(
n=
12)
determined
for
two
standards:
16
ng/
mL
(
n=
12)
and
35.6
ng/
mL
(
n=
12)
10
Department
of
Pathology,
Faculty
Veterinary
Medicine
Swedish
University
of
Agricultural
Sciences
(
SLU)
Uppsala,
Sweden
Modified
direct
noncompetitive
sandwich
ELISA(
d)
Borg
(
unpublished)(
e)
40
ng/
mL
determined
for
Lab
10'
s
standard
40
ng/
mL
determined
for
Lab
10'
s
standard
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
9
Table
3.
Contd.
Lab
ID
Participating
Laboratory
ELISA
Method
Reference
zebra
fish
liver
MDL
zebra
fish
whole
body
MDL
medaka
liver
MDL
medaka
whole
body
MDL
11
National
Institute
for
Environmental
Studies
(
NIES)
Ibaraki,
Japan
Sandwich
ELISA,
monoclonal
antibody;
Direct
sandwich
ELISA,
monoclonal
and
polyclonal
antibodies;(
f)

Sandwich
ELISA
EnBio
(
2002);
Nishi
et
al.
(
2002)

Transgenic
(
2002)

Biosense
Laboratories
(
2002)
2.0
ng/
mL
published
in
protocol
0.488
ng/
mL
minimum
of
working
range,
published
Transgenic
2.0
ng/
mL
published
in
protocol
0.488
ng/
mL
minimum
of
working
range,
published
Transgenic
12
Los
Angeles
County
Sanitation
Districts
Whittier,
California,
Sandwich
ELISA
Biosense
Laboratories
(
2002,
2003)
0.12
ng/
mL
matrix­
specific
determination
0.12
ng/
mL
matrix­
specific
determination
0.49
ng/
mL
matrix­
specific
determination
0.49
ng/
mL
matrix­
specific
determination
a)
Laboratory
6,
Unité
d'Evaluation
des
Risques
Ecotoxicologiques
(
INERIS),
Verneuil­
en­
Halatte,
France,
used
a
competitive
binding
assay
(
Brion
et
al.
2002).
Because
the
results
from
Laboratory
6
were
received
too
late
for
inclusion
in
the
comparative
statistical
analysis,
the
original
data
sheets
are
presented
in
Appendix
B.
b)
Sandwich
enzyme
immunoassay,
anti­
zebra
fish
or
anti­
medaka
VTG
capture
antibody
and
detecting
antibody.
c)
Sandwich
ELISA,
anti­
zebra
fish
VTG
monoclonal
antibody
produced
in
mouse;
mono­
mono.
d)
Direct,
noncompetitive
sandwich
ELISA,
anti­
zebra
fish
lipovitellin,
polyclonal
antibody
produced
in
rabbit.
e)
D.
Borg,
2003
(
unpublished),
ELISA
protocol
for
the
detection
of
vitellogenin
in
zebrafish,
Department
of
Biology,
Odense
University,
Denmark;
received
as
personal
communication,
4
July
2003.
f)
Direct
sandwich
ELISA,
anti­
medaka
VTG
monoclonal
antibody
and
biotinylated
polyclonal
antibody;
poly­
poly.

7.0
SUMMARY
OF
RESULTS
Sets
of
zebra
fish
and/
or
medaka
homogenates
of
liver
and
whole
body
were
supplied
to
11
participating
laboratories
for
VTG
analysis;
not
every
laboratory
analyzed
both
species
(
Table
4).
It
was
intended
that
one
laboratory
would
also
analyze
blood
plasma
from
each
species;
however,
due
to
extenuating
circumstances,
the
particular
laboratory
was
unable
to
join
the
study
as
planned,
and
plasma
was
therefore
not
analyzed.
All
laboratories
contributed
their
services
without
compensation,
and
Battelle
made
every
attempt
to
accommodate
and
to
assist
the
laboratories
in
performing
their
task.

The
majority
of
the
participating
laboratories
(
6
of
11)
employed
exclusively
a
commercial
enzyme
immunoassay
(
EIA)
kit
for
zebra
fish
or
medaka,
which
is
a
sandwich
ELISA
using
specific
binding
between
antibodies
and
VTG
(
Biosense
2002,
2003)
(
see
Table
3).
One
additional
lab
used
the
same
kit
along
with
two
other
methods
for
comparison:
a
sandwich
ELISA
using
monoclonal
antibodies
(
a
commercial
kit,
Nishi
et
al.
2002)
and
a
direct
sandwich
ELISA
using
monoclonal
antibodies
and
biotinylated
polyclonal
antibodies
(
a
commercial
kit,
Transgenic
2002).
Still
others
used
a
direct
noncompetitive
sandwich
ELISA
(
Holbech
et
al.
2001),
or
a
modification
thereof
(
personal
communication,
Daniel
Borg,
2003,
Department
of
Biology,
Odense
University,
Denmark,
unpublished
ELISA
protocol
for
the
detection
of
vitellogenin
in
zebrafish).
The
eleventh
laboratory,
not
included
in
the
statistical
comparison,
used
a
competitive
binding
assay
(
Brion
et
al.
2002).
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
10
Table
4.
Species
analyzed
by
participating
laboratories(
a)

Lab
ID
Participating
Laboratory
Zebra
Fish
Medaka
1
University
of
Kumamoto
X
X
2
Biosense
Laboratories
X
X
4
EnBioTec
Laboratories,
Ltd.
X
X
5
Notox
Safety
&
Environmental
Research
X
X
7
Prefectural
Universität
Bern
X
8
Phylonix
Pharmaceuticals,
Inc.
X
9
Institute
of
Biology,
University
of
Southern
Denmark
X
10
Swedish
University
of
Agricultural
Sciences
(
SLU)
X
11
National
Institute
for
Environmental
Studies
(
NIES)
XXX(
b)

12
Los
Angeles
County
Sanitation
Districts
(
LACSD)
X
X
a)
Laboratory
6,
Unité
d'Evaluation
des
Risques
Ecotoxicologiques
(
INERIS),
analyzed
both
zebra
fish
and
medaka
tissues;
however,
it
reported
its
results
too
late
for
inclusion
in
the
comparative
statistical
analysis.
The
original
data
sheets
presented
in
Appendix
B.
b)
NIES
requested
and
analyzed
three
sets
of
medaka
samples,
using
three
different
methods.

Protocols
appear
in
detail
in
Appendix
A
and
are
listed
in
Table
3
with
a
brief
description
along
with
the
method
detection
limit
(
MDL).
Original
data
sheets
from
all
the
participating
laboratories
are
found
in
Appendix
B.
A
full
and
detailed
presentation
of
results,
including
text,
data
tables,
and
figures,
comprises
Appendix
C,
and
the
descriptive
statistics
are
available
in
Appendix
D,
E,
and
F
for
the
within­
run
VTG
results,
intralaboratory
results,
and
intra­
assay
results,
respectively.

Zebra
Fish
Nine
of
the
participating
laboratories
analyzed
zebra
fish
tissue
homogenates.
A
summary
of
the
codes
indicating
concentration,
standards,
and
assay
method
for
zebra
fish
statistical
analysis
is
presented
in
Table
5.
The
concentration
codes
1
through
5
are
based
on
the
exposure
history
and
sex
of
the
fish
used
to
generate
the
samples
in
the
series,
plus
the
positive
control.
The
standard
codes
identify
the
two
different
calibration
standards.
However,
in
the
text
that
follows,
the
term
"
homologous"
refers
to
the
standard
routinely
employed
by
the
individual
laboratory
with
respect
to
the
assay
in
use,
and
is
prepared
or
purchased
by
the
laboratory;
the
term
"
purified"
refers
to
the
standard
that
was
supplied
by
Battelle
to
the
participating
laboratories
 
that
is,
the
purified
VTG
standard
commercially
prepared
by
Biosense
Laboratories.
The
four
method
or
assay
code
numbers
define
the
particular
commercial
kits
or
unique
method
employed.
These
groupings
were
used
to
analyze
the
variability
of
the
reported
results.

Not
all
laboratories
used
both
standards:
six
of
the
nine
laboratories
that
analyzed
zebra
fish
homogenates
employed
the
two
standards,
as
requested;
two
used
only
the
purified
zebra
fish
VTG
supplied
by
Battelle,
2
and
one
used
only
its
own
homologous
standard
(
Table
6).

2
The
two
laboratories
commented
that
the
purified
standard
was
identical
to
that
they
typically
used
with
the
method
employed
for
this
study
(
see
original
data
sheets,
Appendix
B).
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
11
Table
5.
Summary
of
codes
for
relative
VTG
concentration,
standards,
and
ELISA
methods
for
zebra
fish
Description
Code
CONCENTRATION
Uninduced
Male
1
Uninduced
Female
2
Induced
Male
3
Induced
Female
4
Positive
Control
5
STANDARD
Homologous
1
Purified
2
ZF­
METHOD
AND
REFERENCE(
a)

1
(
Biosense
2003)
1
2
(
EnBio
2002;
Nishi
et
al.
2002)
2
3
(
Holbech
et
al.
2001)
3
4
(
Borg
2003,
unpublished)
4
a)
ZF
Zebra
Fish;
see
description
of
assay
methods
in
Table
3.

Table
6.
Summary
of
standards
employed
by
participating
laboratories
analyzing
zebra
fish
Lab
ID
Participating
Laboratory
Homologous
standard
Purified
zebra
fish
standard(
a)

1
University
of
Kumamoto
X
X
2
Biosense
Laboratories
X
X
4
EnBioTec
Laboratories,
Ltd.
X
X
5
Notox
Safety
&
Environmental
Research
X(
b)

7
Prefectural
Universität
Bern
X(
b)

8
Phylonix
Pharmaceuticals,
Inc.
X
X
9
University
of
Southern
Denmark
X
X
10
Swedish
University
of
Agricultural
Sciences
X
12
County
Sanitation
Districts
of
Los
Angeles
County
X
X
a)
Commercially
prepared,
purified
VTG
supplied
by
Battelle
to
participants.
b)
The
standard
supplied
by
Battelle
was
the
same
as
that
used
by
the
laboratory;
accordingly,
the
laboratory
elected
to
employ
only
one
standard,
providing
one
set
of
data
for
its
sample
set.

Trends
of
Standard
Series:
Measured
VTG
Concentrations
A
goal
of
the
study
was
to
generate
a
range
of
concentrations
of
VTG
in
male
and
female
liver
and
whole
body
homogenates,
beginning
with
zero
to
low
levels
in
uninduced
males,
increasing
in
the
following
order:
uninduced
male<
uninduced
female<
induced
male<
induced
female
zebra
fish
VTG
concentrations.
The
VTG
concentration
in
the
precisely
spiked
positive
control,
which
was
6.25
µ
g
(=
6250
ng)
VTG/
mL
liver
and
whole
body
homogenate
for
zebra
fish,
could
have
been
expected
to
fall
between
the
zero­
to­
low
uninduced
male
and
the
uninduced
female
levels.
In
consideration
of
all
of
the
reported
data,
the
general
trend
for
the
liver
samples
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
12
observed
for
each
laboratory,
averaged
over
antibodies,
standards,
and
assays,
differed
from
the
expected
increasing
concentration
series
of
VTG
specifically
in
the
similarity
of
VTG
levels
in
uninduced
female
and
induced
males
(
Figure
C4
in
Appendix
C).
The
whole
body
homogenate
samples
were
more
consistent
with
expected
results
with
respect
to
the
order
of
induced
females
with
VTG
levels
greater
than
those
for
induced
males
in
over
71%
(
5
of
7
labs)
of
the
reported
data
sets;
however,
the
VTG
concentrations
in
whole
body
tissue
of
induced
and
uninduced
females
were
reversed
with
respect
to
the
expected
gradient
in
over
half
(
4
of
7
labs)
of
the
laboratories'
data
sets
reporting
for
this
analysis
(
Figure
C5
in
Appendix
C.)
The
divergence
from
the
expected
series
points
out
the
need
for
further
examination
of
the
potential
sources
of
variation.
This
analysis
included
all
reported
results,
and
additional
analysis
was
conducted
based
on
laboratory
and
type
of
method
to
examine
this
variability.

Difference
in
Standards
A
critical
aspect
of
the
performance
of
an
analytical
method
is
the
use
of
standards
and
controls
in
an
assay.
Two
standards
were
used
for
the
present
study:
purified
standard
was
purchased
and
supplied
to
the
participating
laboratories
by
Battelle;
the
homologous
standard
was
the
standard
routinely
used
by
each
laboratory
with
their
preferred
method.
The
percentage
difference
between
the
average
replicate
VTG
concentrations
obtained
with
the
homologous
(
H)
and
the
purified
(
B)
standard
was
calculated
as
(
H­
B)/
B
X
100%.
Negative
values
represented
greater
VTG
concentrations
obtained
with
the
purified
standard.
Six
of
the
nine
labs
analyzing
zebra
fish
homogenates
used
both
standards;
two
used
only
the
purified
standard
because
it
was
identical
to
their
homologous
standard;
and
one
used
only
its
homologous
standard.
In
general,
the
differences
were
relatively
small
for
all
concentrations
for
zebra
fish
(
ZF)
Methods
1
and
3,
which
represent
a
sandwich
ELISA
and
a
direct
noncompetitive
sandwich
ELISA,
respectively,
but
quite
large
in
the
results
of
ZF­
Method
2,
a
monoclonal­
antibody­
based
sandwich
ELISA,
for
all
concentrations
(
see
Tables
C9­
C12
and
Figures
C6­
C10
in
Appendix
C;
Appendix
F).
VTG
concentrations
were
about
700%
to
800%
lower
for
the
purified
than
for
the
homologous
standard.

There
was
a
wide
range
of
values
measured
in
both
liver
and
whole
body
samples
analyzed
in
the
various
laboratories
with
calibration
to
the
two
different
standards;
the
variability
depended
primarily
on
the
analytical
method
and
secondarily
on
the
number
of
laboratories
running
the
analyses.
VTG
values
tended
to
be
highest
for
the
ZF­
Method
1
and
lowest
for
ZF­
Method
2
for
liver.
For
example,
for
the
homologous
standard,
VTG
concentrations
measured
by
ZF­
Method
1
for
liver
homogenate
were
24
times
to
3277
times
greater
than
those
measured
by
ZF­
Method
2.
Similarly,
for
the
purified
standard,
the
concentrations
were
98
to
2099
times
greater
measured
by
ZF­
Method
1
than
by
2.
The
great
variability
between
methods
is
illustrated
by
the
detection
of
VTG
in
uninduced
males:
not
detected
by
ZF­
Method
2,
but
measured
at
136,117
ng/
mL
by
ZF­
Method
1.

For
whole
body
homogenate,
homologous
standard,
the
measured
VTG
values
were
highest
for
ZF­
Method
4,
a
modified
direct
noncompetitive
sandwich
ELISA,
from
13
to
19,951
times
higher
than
those
measured
by
ZF­
Method
2.
For
the
purified
standard,
the
values
were
generally
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
13
highest
for
ZF­
Methods
1
and
3,
which
were
16
to
17,589
times
greater
than
those
measured
by
ZF­
Method
2.

Multiple
laboratories
applied
the
same,
commercially
available
method
(
ZF­
Method
1),
and
a
high
interlaboratory
variability
(
CVs
from
63%
to
229%
for
liver;
26%
to
82%
for
whole
body)
could
be
seen
in
the
values
attained
with
the
homologous
standard,
with
a
surprisingly
high
variability
(
the
highest)
for
the
positive
controls
for
the
liver
homogenate,
and
in
the
uninduced
male
samples
for
whole
body.
For
the
purified
standard,
the
highest
variability
was
likewise
for
the
positive
controls
in
liver,
but
for
induced
male
samples
in
whole
body
homogenate
samples.

Within­
Run
Variability
The
three
analytical
replicates
provide
a
measure
of
the
within­
run
variability
(
Table
C7
and
Figures
C2,
C3
in
Appendix
C;
Appendix
D).
When
all
of
the
methods
were
applied
to
the
full
range
of
samples,
both
the
zebra
fish
liver
and
whole
body
showed
a
wide
range
of
variability
of
absolute
values
(
CVs
0%
to
72.9%;
mean
9.4%,
and
0%
to
76.1%,
mean
8.6%,
respectively)
among
analytical
replicates,
but
a
consistency
of
trend.
However,
because
75%
of
the
CVs
were
<
10%
for
both
homogenates,
and
both
tended
to
be
<
30%
for
uninduced
females
and
induced
males,
this
comparison
indicates
relatively
low
variability
in
the
expected
mid­
and
higher
VTG
concentrations,
but
a
somewhat
higher
variability
in
bottom
ranges.
The
latter
can
be
graphically
seen
to
be
explained
by
low
outlier
values
for
concentration
Codes
1
and
2
for
Labs
4
and
5
for
the
liver
(
Figure
2a),
and
for
Codes
1
and
3
in
particular
for
Lab
4
in
whole
body
(
Figure
2b).
Otherwise,
the
results
of
the
balance
of
the
laboratories'
results
appear
to
hold
relatively
tight
grouping,
especially
at
the
higher
values.
This
plot
also
emphasizes
the
general
trend
of
measured
VTG
values,
which
follows
the
expected
concentration
curve
of
the
standard
series.
The
detection
of
low­
level
VTG
is
a
critical
component
of
a
method
for
use
in
a
screening
assay
to
detect
the
induction
of
the
VTG
protein;
the
zebra
fish
results
appear
to
exhibit
sufficient
consistency
within
that
range
for
most
of
the
laboratories.

Intra­
Assay
Variability
The
analysis
of
sample
triplicates
provided
a
measure
of
intra­
assay
variability
(
Table
C8
in
Appendix
C;
Appendix
E).
The
range
of
CVs
was
broad
for
both
liver
and
whole
body
homogenates
(
0.3%
to
173.2%),
but
75%
of
the
intra­
assay
CVs
were
<
29%
for
both
types
of
tissue.
This
level
of
intra­
assay
variability
indicates
that
when
a
sample
is
provided
to
multiple
laboratories
employing
a
variety
of
methods
(
the
results
using
multiple
standards
are
also
included
in
this
sample
set)
a
quarter
of
the
methods
provide
a
moderately
high
degree
of
variability
in
absolute
VTG
measured
value
when
replicate
samples
are
analyzed,
but
three
quarters
showed
low
variability.
The
general
trends
(
in
contrast
with
the
focus
on
absolute
VTG
quantities)
tracked
the
expected
concentration
series
fairly
well.
To
further
examine
this
type
of
variability,
which
is
critical
to
the
application
of
ELISA
to
a
screening
assay,
the
data
were
further
examined
by
individual
laboratory.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
14
0.1
1
10
100
1000
10000
100000
1000000
10000000
100000000
0
1
2
3
4
5
6
Concentration
Code
Average
Within­
Run
VTG
Concentration
(
ng/
mL)
Lab
1
Lab
2
Lab
4
Lab
5
Lab
7
Lab
8
Lab
9
Lab
10
Lab
12
1
10
100
1000
10000
100000
1000000
10000000
100000000
1000000000
0
1
2
3
4
5
6
Concentration
Code
Average
Within­
Run
VTG
Concentration
(
ng/
mL)
Lab
1
Lab
2
Lab
4
Lab
5
Lab
7
Lab
8
Lab
9
Lab
10
Lab
12
Figure
2.
Mean
of
zebra
fish
triplicate
VTG
values
for
each
of
three
samples
per
concentration
in
the
standard
series,
plotted
by
laboratory:
(
a)
liver;
(
b)
whole
body
(
Code
1
=
Uninduced
Male;
Code
2
=
Uninduced
Female;
Code
3
=
Induced
Male;
Code
4
=
Induced
Female;
Code
5
=
Positive
Control)
A:
Zebra
Fish
Liver
A:
Zebra
Fish
Whole
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
15
Comparison
by
Method
The
ranked
average
(
natural
log­
transformed)
VTG
concentrations
of
the
samples
were
compared
using
Tukey's
HSD
to
determine
whether
methods
vary
in
their
ability
to
detect
differences
between
pairs
of
treatment
concentrations
(
see
Tables
C13­
C16
in
Appendix
C).
For
the
liver
samples
and
homologous
standard,
the
direct
noncompetitive
sandwich
ELISA
(
ZF­
Method
3)
failed
to
detect
differences
among
any
of
the
six
possible
pairwise
comparisons;
the
other
three
methods
did
only
slightly
better,
each
distinguishing
one
of
the
six
pairs.
In
contrast,
for
whole
body
samples,
all
four
methods
detected
differences
in
VTG
values
in
at
least
four
of
the
six
possible
pairwise
comparisons
and
could
distinguish
differences
between
uninduced
males,
which
represent
the
low
end
of
the
concentration
spectrum,
and
the
three
other
treatments,
and
between
uninduced
females
and
induced
males,
which
represent
the
middle
ranges.
One
significant
difference
was
detected
by
ZF­
Method
1
between
uninduced
males
and
induced
females,
which
should
represent
the
maximum
spread
of
concentrations,
zero
to
high,
respectively,
for
liver
samples
using
the
purified
standard.
With
the
purified
standard,
ZF­
Methods
2
and
3
failed
to
detect
differences
among
any
of
the
pairwise
comparisons
for
liver
samples.

For
whole
body
samples
and
homologous
standard,
all
three
analytical
methods
detected
differences
in
VTG
values
for
at
least
three
of
the
six
pairwise
comparisons.
All
three
methods
distinguished
uninduced
males
from
the
other
treatments,
again
with
focus
on
the
low
end
of
the
concentration
range.
Only
ZF­
Method
2
could
detect
the
difference
between
uninduced
females
and
induced
females,
and
none
of
the
methods
saw
a
difference
between
induced
males
and
induced
females
at
the
high
end
of
the
concentration
gradient.
The
strongest
pattern
that
emerges
from
the
detailed
outcomes
in
this
test
is
the
contrast
between
liver
and
whole
body
tissue
results.

Within­
Method
1
Comparison
 
Multiple
Laboratories
Applying
the
Same
Method
Four
laboratories
used
ZF­
Method
1,
the
sandwich
ELISA
with
anti­
zebra
fish
VTG
capture
antibody
and
detecting
antibody,
using
the
homologous
standard
for
liver
and
whole
body
(
Tables
C17,
C18
in
Appendix
C).
The
detection
of
differences
among
treatment
concentrations
depended
first
on
the
laboratory
conducting
the
analysis,
and
next
on
the
sample
type
and
standard
used.
For
liver
homogenate,
three
laboratories
distinguished
the
concentration
differences
for
five
of
the
six
possible
pairwise
comparisons.
No
laboratory
detected
the
expected
midrange
values
(
uninduced
females,
induced
males).
Three
of
the
four
laboratories
reported
results
for
whole
body
samples,
but
the
results
were
variable:
for
each
of
the
six
pairwise
comparisons,
at
least
one
laboratory
detected
a
significant
difference.

For
the
purified
standard
for
liver
homogenate,
the
results
were
similar
to
those
above.
Six
laboratories
used
the
same
method
(
ZF­
Method
1),
and
the
detection
of
differences
depended
on
the
same
factors
in
the
same
order
as
in
the
case
of
homologous
standard
use.
Five
of
the
six
laboratories
failed
to
detect
differences
for
all
of
the
six
possible
pairwise
comparisons
(
Table
C19
in
Appendix
C);
one
laboratory
detected
significant
differences
in
two
pairs.
Only
four
of
the
six
laboratories
reported
results
for
all
pairwise
comparison
for
whole
body,
and
the
ability
to
detect
pairwise
differences
varied
considerably
(
Table
C20
in
Appendix
C).
No
laboratory
detected
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
16
differences
between
uninduced
and
induced
females,
but
did
detect
significant
differences
between
three
other
pairs.

Positive
Control
Results
Each
lab
analyzed
a
positive
control
precisely
spiked
with
purified
zebra
fish
VTG,
the
concentration
of
which
was
unknown
to
the
analysts.
As
prepared,
the
concentration
of
the
positive
controls
was
6.25
µ
g/
mL
(=
6250
ng/
mL)
liver
and
whole
body
homogenate
(
see
Tables
C21­
C23
in
Appendix
C).
Although
the
within­
laboratory
variability
was
relatively
low
(
75%
of
CVs
were
below
45%
for
liver
samples
and
below
20%
for
whole
body),
the
interlaboratory
variability,
and
that
among
laboratories
and
methods
was
high
in
the
percentage
recovery
of
the
spiked
concentration
from
liver
samples.
For
example,
percentage
recovery
from
liver
homogenates
was
56%
to
20,073%
for
the
homologous
standard,
and
from
7%
to
6647%
for
the
purified.
The
variability
for
whole
body
analysis
was
lower,
but
nonetheless,
>
100%
recovery
was
seen
in
most
samples.
It
appears
that
the
outcome
of
tests
could
be
influenced
by
the
choice
of
method
in
particular,
as
well
as
by
the
performing
laboratory
and
other
factors
considered
here.

Medaka
Six
laboratories
analyzed
medaka
tissue
homogenates
by
three
different
methods;
one
among
these
participants
analyzed
three
sets
of
medaka
samples
by
all
three
methods,
bringing
the
number
of
data
sets
to
eight,
and
providing
a
unique
opportunity
for
a
comparison
of
assays
that
reduced
interlaboratory
variability.
The
medaka
analyses
were
similar
to
and
concurrent
with
performance
of
those
of
zebra
fish
tissues,
and
two
of
the
ELISA
methods
followed
the
same
or
parallel
protocols
specifically
tailored
to
medaka.
See
Table
7
for
the
codes
assigned
to
categorize
relative
VTG
concentrations
in
the
standard
series,
the
two
standards
used
for
calibration,
and
the
three
methods
applied
to
medaka.
The
methods
are
described
and
associated
with
specific
participants
in
Table
3
and
presented
in
full
in
Appendix
A.
Not
all
laboratories
used
both
standards:
five
of
the
six
laboratories
that
analyzed
medaka
homogenates
employed
the
two
standards,
as
requested;
one
used
only
the
purified
medaka
VTG
supplied
by
Battelle
(
Table
8).

Trends
of
Standard
Series:
Measured
VTG
Concentrations
It
was
anticipated
that
male
fish
that
were
not
induced
by
exposure
to
estrogenic
compounds
would
provide
minimal
levels
of
VTG
in
their
tissues,
and
that
uninduced
female
fish,
induced
male,
and
induced
female
fish
would
generate
increasing
levels
of
VTG
in
their
respective
systems.
In
consideration
of
all
reported
results,
the
general
trend
for
liver
samples
fit
the
expected
serial
increase
in
general:
uninduced
male<
uninduced
female<
induced
male<
induced
female
medaka
VTG
concentrations
(
Figure
C13
in
Appendix
C).
(
The
VTG
concentration
in
the
precisely
spiked
positive
control,
which
was
6.85
µ
g
[=
6850
ng]
VTG/
mL
liver
homogenate,
and
6.25
µ
g
[=
6250
ng]
VTG/
mL
whole
body
homogenate
for
medaka
[
Table
2],
could
have
been
expected
to
fall
between
the
zero­
to­
low
uninduced
male
and
the
uninduced
female
levels.)
Exceptions
to
this
consistent
performance
were
as
follows:
three
of
the
eight
analyses
from
six
laboratories
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
17
Table
7.
Summary
of
codes
for
relative
VTG
concentration,
standards,
and
ELISA
methods
for
medaka
Description
Code
CONCENTRATION
Uninduced
Male
1
Uninduced
Female
2
Induced
Male
3
Induced
Female
4
Positive
Control
5
STANDARD
Homologous
1
Purified
2
M­
METHOD
AND
REFERENCE(
a)

1
(
Biosense
2003)
1
2
(
Transgenic
2002)
2
3
(
EnBio
2002;
Nishi
et
al.
2002)
3
a)
M
medaka;
see
description
of
assay
methods
in
Table
3.

Table
8.
Summary
of
standards
employed
by
participating
laboratories
analyzing
medaka
Lab
ID
Participating
Laboratory
Homologous
standard(
a)
Purified
medaka
standard(
b)

1
University
of
Kumamoto
X
X
2
Biosense
Laboratories
X
X
4
EnBioTec
Laboratories,
Ltd.
X
X
5
Notox
Safety
&
Environmental
Research
X(
c)

11
National
Institute
for
Environmental
Studies
X
X
12
Los
Angeles
County
Sanitation
Districts
X
X
a)
The
standard
routinely
employed
by
the
individual
laboratory
with
respect
to
the
assay
in
use.
b)
Commercially
prepared,
purified
VTG
supplied
by
Battelle
to
participants.
c)
The
standard
supplied
by
Battelle
was
the
same
as
that
used
by
the
laboratory;
accordingly,
the
laboratory
elected
to
employ
only
one
standard,
providing
one
set
of
data
for
its
sample
set.

found
VTG
concentrations
in
induced
female
livers
to
be
lower
than
those
of
induced
males.
This
is
contrary
to
the
logic
of
the
standard
series;
however,
it
will
be
valuable
to
explore
further
the
possible
explanations,
based
perhaps
on
condition
of
the
fish
 
for
example,
the
medaka
females
were
reported
in
the
laboratory
notebook
to
be
actively
laying
eggs
at
the
time
of
exposure
 
or
possibly
on
dose­
related
response
to
estrogen
exposure.
The
exposure
concentration
varies
from
study
to
study;
here,
300
ng/
L;
Nishi
et
al.
(
2002)
used
10
ng/
L
(
medaka);
Brion
et
al.
(
2002)
tried
a
range
from
0.1
µ
g/
L
to
100
µ
g/
L
(
zebra
fish);
and
Holbech
et
al.
(
2001)
used
10
ng/
L
exposure
(
zebra
fish).
For
the
whole
body
results,
the
pattern
was
as
would
be
expected
for
six
of
the
eight
reports:
the
VTG
concentration
sequence
was
uninduced
female<
induced
males<
induced
females.
These
results
appear
to
be
a
more
promising
indication
of
utility
of
medaka
over
zebra
fish
for
screening
of
VTG
induction.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
18
Difference
in
Standards
The
comparison
of
data
by
standard
used
offers
a
perspective
on
method
specificity
and
standardization
in
a
screening
assay.
The
percentage
difference
between
the
average
replicate
VTG
concentrations
obtained
with
the
homologous
(
H)
and
the
purified
(
B)
standard
was
calculated
as
(
H­
B)/
B
X
100%.
Negative
values
represented
greater
VTG
concentrations
obtained
with
the
purified
standard.
Five
of
the
six
laboratories
employed
both
standards;
one
used
only
the
purified
medaka
standard.
This
comparison
showed
relatively
small
differences
in
the
results
generated
with
use
of
the
two
standards
for
all
three
medaka
methods.
For
both
liver
and
whole
body
homogenates
results,
there
was
generally
<
60%
difference
(
see
Tables
C28­
C31
and
Figures
C13­
C19
in
Appendix
C;
Appendix
F).

The
VTG
values
measured
in
both
liver
and
whole
body
samples
for
the
homologous
standard
varied
primarily
with
analytical
method,
and
secondarily
on
the
number
of
laboratories
conducting
the
analyses.
For
both
tissue
homogenates,
VTG
values
tended
to
be
highest
for
the
M­
Method
1,
which
represent
a
sandwich
ELISA
with
anti­
medaka
VTG
capture
antibody
and
detecting
antibody,
for
all
concentrations,
and
lowest
measured
by
M­
Method
2,
which
is
a
monoclonal
and
polyclonal­
antibody­
based
direct
sandwich
ELISA.

For
liver,
M­
Method
1'
s
measured
concentrations
were
four
to
seven
times
greater
than
those
by
M­
Method
2,
and
1.4
to
9
times
greater
than
those
measured
by
M­
Method
3,
which
is
a
monoclonal­
antibody­
based
sandwich
ELISA.
When
the
purified
standard
was
used,
VTG
values
measured
in
liver
and
whole
body
samples
varied
widely
by
analytical
method,
primarily,
and
secondarily
by
number
of
laboratories
conducting
the
analyses
or
by
laboratory,
respectively.
For
both
tissue
homogenates,
M­
Method
1
yielded
higher
VTG
concentration
measures
( 
7%
for
liver,
 
13%
for
whole
body)
than
did
either
of
the
other
methods.

Multiple
laboratories
applied
the
same,
commercially
available
medaka
methods
(
M­
Methods
1,
3),
and
a
high
interlaboratory
variability
(
CVs
from
62%
to
126%
for
liver;
63%
to
195%
for
whole
body)
could
be
seen
in
the
values
attained
with
the
homologous
standard.
The
highest
variability
was
seen
for
the
positive
controls
for
liver,
and
for
uninduced
male
in
whole
body
homogenate.
Similarly
for
the
purified
standard
with
liver
and
whole
body
homogenate,
the
CVs
ranged
from
10%
to
108%
(
liver)
and
12%
to
184%
(
whole
body),
with
the
highest
variability
occurring
in
positive
controls
for
liver,
and
in
induced
male
samples
for
whole
body.

Within­
Run
Variability
The
three
analytical
replicates
provide
a
measure
of
the
within­
run
variability
(
Table
C26
and
Figures
C11,
C12
in
Appendix
C;
Appendix
D).
When
all
of
the
methods
were
applied
to
the
full
range
of
samples,
both
the
medaka
liver
and
whole
body
showed
a
wide
range
of
variability
(
CVs
0.8%
to
100.4%;
mean
9.1%,
and
04%
to
125.7%,
mean
11.9%,
respectively)
among
analytical
replicates.
However,
because
75%
of
the
CVs
were
<
14%
for
both
homogenates,
and
both
tended
to
be
<
30%
for
uninduced
females
and
induced
males,
this
comparison
indicates
relatively
low
variability
in
the
expected
midrange
VTG
concentrations,
but
a
somewhat
higher
variability
in
the
top
and
bottom
ranges.
However,
because
75%
of
the
CVs
were
<
10%
for
both
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
19
homogenates,
and
both
tended
to
be
<
30%
for
uninduced
females
and
induced
males,
this
comparison
indicates
relatively
low
variability
in
the
expected
mid­
and
higher
VTG
concentrations,
but
a
somewhat
higher
variability
in
bottom
ranges.
The
variability
can
be
graphically
seen
in
Figure
3:
most
of
the
values
are
grouped
tightly
for
the
within
run
results
for
each
laboratory,
particularly
for
the
medaka
liver
samples.
The
exception
is
for
a
single
low,
outlier
result
from
Laboratory
12
in
Code
4
the
liver
samples
(
Figure
3a).
In
the
whole
body
samples,
there
is
a
wider
range
for
Laboratory
5'
s
Code
1
results
(
two
moderately
high
outlier
values)
(
Figure
3b).
This
plot
also
emphasizes
the
general
trend
of
measured
VTG
values,
which
follows
the
expected
concentration
curve
of
the
standard
series
particularly
well
in
the
medaka
liver
samples,
and
nearly
as
well
in
the
whole
body.

Intra­
Assay
Variability
The
analysis
of
sample
triplicates
provided
a
measure
of
intra­
assay
variability
(
Table
C27
in
Appendix
C;
Appendix
F).
The
range
of
CVs
was
broad
for
both
liver
and
whole
body
homogenates
(
0.7%
to
158.8%),
and
75%
of
the
intra­
assay
CVs
were
<
41%
for
both
types
of
tissue.
This
level
of
intra­
assay
variability
follows
a
trend
parallel
to
that
of
zebra
fish
results,
indicating
that
when
a
sample
is
provided
to
multiple
laboratories
employing
a
variety
of
methods
(
the
results
using
multiple
standards
are
also
included
in
this
sample
set)
the
methods
provide
a
relatively
high
degree
of
variability
when
replicate
samples
are
analyzed.
To
further
examine
this
type
of
variability,
explored
here
by
analysis
of
all
results,
the
data
were
further
examined
by
individual
laboratory.

Comparison
by
Method
The
ranked
average
(
natural
log­
transformed)
VTG
concentrations
of
the
samples
were
compared
using
Tukey's
HSD
to
determine
whether
methods
vary
in
their
ability
to
detect
differences
between
pairs
of
treatment
concentrations
(
see
Tables
C32­
C35
in
Appendix
C).
For
the
liver
and
whole
body
samples
and
both
standards,
the
results
were
identical:
M­
Methods
1
and
3
could
detect
differences
on
the
low
end
of
the
concentration
gradient,
namely,
between
the
lowest
VTG
concentration
group,
the
uninduced
males,
and
each
of
the
following:
uninduced
females,
induced
males,
and
induced
females.
These
are
important
distinctions
for
discovery
of
induction
in
males,
as
would
be
applied
in
endocrine
disrupter
screening
programs.
None
of
the
methods
could
detect
differences
at
the
high
end
to
distinguish
induced
females
from
induced
males,
and
in
the
midrange,
uninduced
females
from
induced
males
and
induced
females.

Between­
Method
Comparison
 
Single
Laboratory
Applying
Multiple
Methods
Laboratory
11
used
all
three
methods,
allowing
a
comparison
of
the
methods
under
relatively
consistent
conditions
(
Figures
C20,
C21
and
Tables
C36,
C37
in
Appendix
C).
For
both
sample
types,
the
VTG
concentrations
averaged
over
standards
were
the
highest
measured
by
M­
Method
1,
and
lowest
by
M­
Method
3,
and
the
variability
was
small
(
most
CVs
<
20%).
There
were
no
differences
between
the
two
standards
by
any
analytical
methods
for
the
measurement
of
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
20
1
10
100
1000
10000
100000
1000000
10000000
0
1
2
3
4
5
6
Concentration
Code
Average
Within­
Run
VTG
Concentration
(
ng/
mL)

Lab
1
Lab
2
Lab
4
Lab
5
Lab
11­
M1
Lab
11­
M2
Lab
11­
M3
Lab
12
100
1000
10000
100000
1000000
10000000
100000000
0
1
2
3
4
5
6
Concentration
Code
Average
Within­
Run
VTG
Concentration
(
ng/
mL)

Lab
1
Lab
2
Lab
4
Lab
5
Lab
11­
M1
Lab
11­
M2
Lab
11­
M3
Lab
12
Figure
3.
Mean
of
medaka
triplicate
VTG
values
for
each
of
three
samples
per
concentration
in
the
standard
series,
plotted
by
laboratory:
(
a)
liver;
(
b)
whole
body
(
Code
1
=
Uninduced
Male;
Code
2
=
Uninduced
Female;
Code
3
=
Induced
Male;
Code
4
=
Induced
Female;
Code
5
=
Positive
Control)
B:
Medaka
Whole
Body
B:
Medaka
Liver
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
21
VTG
in
liver
homogenates.
In
contrast,
for
whole
body,
the
VTG
concentrations
did
not
vary
between
standards
for
M­
Method
1,
varied
for
some
concentrations
in
the
standard
series
by
M­
Method
2,
and
differed
for
each
and
every
concentration
by
M­
Method
3.
The
pairwise
comparisons
analyzed
by
Tukey's
HSD
on
natural
log­
transformed
VTG
levels
showed
that
all
three
methods
could
detect
differences
in
liver
and
whole
body
VTG
levels
for
at
least
four
of
the
six
possible
pairs
for
the
homologous
standard,
and
for
five
of
the
six
possible
pairs
for
the
purified
standard.
This
comparison
demonstrates
the
potentially
large
degree
of
variability
among
laboratories
that
could
account
for
some
of
the
observed
disparity
of
results.

Within­
Method
Comparison
 
Multiple
Laboratories
Applying
the
Same
Methods
Two
analytical
methods,
M­
Method
1
and
M­
Method
3
were
exercised
by
more
than
one
laboratory
in
the
medaka
study,
providing
the
opportunity
to
evaluate
within­
method
variability
using
the
homologous
standard
for
liver
and
whole
body
(
Tables
C38­
C43
in
Appendix
C).
Five
laboratories
used
M­
Method
1
to
determine
VTG
concentrations
in
liver
and
whole
medaka
samples
using
the
homologous
standard.
The
ability
of
the
assay
to
detect
differences
among
the
treatment
concentrations
depended
on
the
laboratory
performing
the
analyses,
the
sample
type,
and
the
standard
that
was
used.
Four
of
the
five
laboratories
that
reported
liver
results
for
all
six
comparisons,
and
all
five
that
reported
whole
body
results
failed
to
detect
differences
between
uninduced
females
and
either
induced
males
or
induced
females,
and
between
induced
females
and
induced
males
in
the
middle
and
high
range
of
the
concentration
gradient.
For
liver,
four
of
five
laboratories'
results
distinguished
uninduced
males
from
induced
males,
and
for
whole
body,
all
five
could
distinguish
not
only
induced
males,
but
also
uninduced
females
and
induced
females
from
the
uninduced
males,
at
the
low
range
of
the
gradient.

In
comparison,
when
purified
standard
was
used,
the
ability
to
discern
differences
in
VTG
concentrations
in
both
liver
samples
depended
on
the
laboratory
performing
the
analysis,
the
sample
type,
as
well
as
the
standard
used.
Notably,
one
laboratory
detected
significant
differences
in
VTG
concentrations
only
between
uninduced
males
and
induced
males,
at
the
low
concentration
range,
and
the
list
of
failed
detection
corresponds
to
that
resulting
from
the
homologous
standard,
above.
All
six
laboratories
that
used
M­
Method
1with
purified
standard
to
analyze
whole
body
samples
detected
differences
in
the
same
pairs
as
the
whole
body,
homologous
standard
results
listed
above:
between
uninduced
males
and
uninduced
females,
between
uninduced
males
and
induced
males,
and
between
uninduced
males
and
induced
females
(
Table
C41
in
Appendix
C).
All
six
laboratories
similarly
failed
to
detect
differences
between
uninduced
females
and
induced
males,
between
uninduced
females
and
induced
females,
and
between
induced
males
and
induced
females.

Two
laboratories
(
4
and
11)
used
M­
Method
3
to
analyze
VTG
concentrations
in
medaka.
Whether
the
homologous
or
the
purified
standard
was
used,
both
laboratories
detected
differences
both
in
liver
and
in
whole
body
VTG
concentrations
for
five
of
the
six
possible
pairwise
comparisons
(
Tables
C42,
C43
in
Appendix
C).
In
the
case
of
both
standards
applied
to
whole
body
analysis,
Laboratory
11
did
not
detect
a
difference
in
VTG
concentration
between
uninduced
females
and
induced
males,
in
the
midrange
of
the
concentration
gradient.
Nonetheless,
this
is
the
most
positive
result
for
any
of
the
methods
with
respect
to
within­
method
variability.
Positive
Control
Results
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
22
Each
lab
analyzed
a
positive
control
precisely
spiked
with
purified
medaka
VTG,
the
concentration
of
which
was
unknown
to
the
analysts.
The
VTG
concentration
in
the
precisely
spiked
positive
control
was
6.85
µ
g
(=
6850
ng)
VTG/
mL
liver
homogenate,
and
6.25
µ
g
(=
6250
ng)
VTG/
mL
whole
body
homogenate
for
medaka
(
Table
2).
Although
the
within­
laboratory
variability
was
relatively
low,
75%
of
CVs
were
below
42%
for
liver
samples
and
below
41%
for
whole
body,
the
interlaboratory
variability,
and
that
among
laboratories
and
methods
was
moderate
(
lower
than
the
comparable
zebra
fish
figures)
in
the
percentage
recovery
of
the
spiked
concentration
from
liver
samples
(
Tables
C44,
C45
in
Appendix
C).
For
example,
percentage
recovery
from
liver
homogenates
was
52%
to
863%
for
the
homologous
standard,
and
from
76%
to
882%
for
the
purified.
The
variability
for
whole
body
analysis
was
greater,
from
>
100%
up
to
1873%
(
homologous)
and
up
to
3418%
(
purified)
recovery
(
Table
C46
in
Appendix
C).
The
variability
was
in
the
orders­
of­
magnitude
range
for
the
positive
control
test
results.
It
appears
that
the
outcome
of
tests
could
be
influenced
primarily
by
the
choice
of
method,
and
secondarily
by
the
performing
laboratory
and
other
factors
considered
in
these
statistical
comparisons.

8.0
DISCUSSION
Zebra
fish
and
medaka
are
among
the
fish
under
consideration
to
be
test
species
for
endocrine
disrupter
research.
Both
species
are
well
characterized,
sensitive
to
exposure
to
hormones
and
other
endocrine
disrupting
compounds,
and
suitable
for
laboratory
handling.
The
present
study
surveyed
existing
ELISA
methods
that
are
currently
available
to
detect
the
protein,
VTG,
in
liver
and
whole
body
homogenate
of
zebra
fish
and
medaka.
It
was
initially
intended
that
blood
plasma
would
also
be
analyzed
by
one
laboratory.
Although
the
plasma
samples
were
collected
and
preserved,
circumstances
precluded
the
participation
of
the
particular
group
that
could
have
conducted
the
analysis,
and
plasma
was
therefore
not
included
in
the
survey.

Methods
routinely
performed
by
the
participating
laboratories
were
applied
to
a
standard
series
of
samples
of
one
or
both
species
of
fish.
Each
series
consisted
of
tissue
homogenates
in
four
categories
by
gender
and
treatment
of
fish
expected
to
yield
a
gradient
of
VTG
concentration
from
low
(
zero)
to
high.
In
addition,
a
positive
control
was
prepared
from
unexposed
male
tissue
spiked
with
a
known
quantity
of
purified,
species­
specific
VTG
as
the
fifth
sample
to
be
included
in
each
analytical
series.
All
of
the
samples
were
provided
blind­
coded
with
respect
to
the
concentration
series
and
control,
but
clearly
identified
as
zebra
fish
or
medaka,
liver
or
whole
body
homogenate.
Because
the
samples
were
blind­
coded
and
represented
a
potentially
wide
range
of
concentrations,
it
required
multiple
dilutions
of
the
sample
to
ensure
a
response
within
the
working
range
of
the
assays.
Accordingly,
a
significant
investment
of
time
and
resources
was
needed,
which
was
kindly
donated
by
the
participating
laboratories
to
aid
in
reaching
the
goals
of
the
study.
The
contributions
of
the
participants
are
gratefully
acknowledged.

The
study
was
not
intended
to
validate
a
given
method,
protocol,
system,
or
technique,
but
rather,
it
was
meant
to
survey
the
current
methods
and
to
discern
the
relative
variability
among
them.
The
results
obtained
from
the
use
of
the
particular
method,
by
circumstance,
by
a
statistically
valid
number
of
laboratories
should
not
be
used
to
assess
the
strength
or
weakness
of
this
method
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
23
compared
with
others
within
or
outside
of
the
present
survey.
Rather,
it
should
be
assumed
that
the
variability
encountered
in
results
would
be
found
with
the
application
of
any
one
of
the
methods
in
this
study
by
multiple
laboratories.
The
use
of
trade
names,
identification
of
laboratories,
and
methods
described
in
this
do
not
constitute
endorsement
by
EPA
or
Battelle
Memorial
Institute.

A
standard
series
of
liver
and
whole
body
homogenate
samples
representing
a
range
of
VTG
concentrations
in
male
and
female
zebra
fish
and
medaka,
and
a
set
of
positive
controls
were
generated
for
this
study.
A
sample
repository/
archive
was
created
and
maintained.
Sets
of
aliquots
representative
of
the
standard
series
and
positive
controls
were
prepared
for
shipping
to
11
laboratories
around
the
world.
Along
with
a
commercially
prepared,
purified
VTG
standard
for
each
species,
documentation,
and
information,
and
they
were
shipped
in
coordination
to
the
participating
labs.
The
integrity
of
the
samples
was
maintained
at
all
times
by
proper
cold­
temperature
control
(
maintained
at
­
80
°
C),
both
in
storage
at
the
preparatory
laboratory
and
in
transit
in
special,
well­
insulated
boxes
to
the
participating
laboratories,
and
by
instruction,
during
storage
and
use
at
the
participating
laboratories.
Ten
of
the
laboratories
returned
results
that
were
included
in
the
statistical
comparison;
the
eleventh
laboratory
was
unable
to
complete
the
analysis
in
time
for
inclusion
in
this
analysis,
but
the
original
data
sheets
and
protocol
are
nonetheless
presented
in
the
appropriate
appendices
of
this
report.

Nine
of
the
laboratories
analyzed
zebra
fish
homogenates,
and
six
laboratories
analyzed
medaka;
however,
the
number
of
data
sets
for
medaka
was
brought
to
eight,
because
one
of
the
participants
analyzed
three
sets
of
medaka
samples
by
three
different
methods.
Four
ELISA
methods
were
applied
to
zebra
fish
homogenates,
and
three
to
medaka.
The
methods
could
be
generally
grouped
as
various
sandwich
ELISAs
based
either
on
monoclonal,
polyclonal,
or
both
types
of
antibody;
and
two
direct,
noncompetititve
sandwich
ELISAs
using
anti­
zebra
fish
lipovitellin
and
polyclonal
antibody.
The
eleventh
laboratory,
the
results
of
which
were
not
included
in
the
statistical
analysis,
used
a
competitive
binding
assay.
Three
of
the
ELISA
methods
were
commercially
available
kits;
the
others
were
unique.

The
statistical
analyses
in
this
report
address
the
within­
run
variability,
the
intra­
assay
variability
based
on
the
mean
triplicate
result,
and
the
general
trend
of
the
ELISA
VTG
results
associated
with
the
standard
evaluation
series
of
fish
liver
and
whole
body
homogenates.
This
series
was
represented
by
1)
uninduced
male,
2)
uninduced
female,
3)
induced
male,
and
4)
induced
female
zebra
fish
or
medaka,
respectively.
In
addition
to
the
standard
series,
a
set
of
positive
control
VTG
results
are
summarized.
The
distribution
of
CVs
of
the
resulting
triplicate
mean
VTG
concentrations
are
summarized
for
a
given
concentration,
laboratory,
and
standard;
across
laboratories,
standards,
and
assays
for
a
given
concentration;
and
by
method
for
a
given
laboratory,
standard,
and
concentration.

There
was
substantial
variation
in
the
reported
quantitative
ELISA
results
in
this
study
for
both
fish
species.
However,
in
spite
of
a
range
in
the
absolute
values
measured,
the
trends
for
the
concentration
series
values
in
both
liver
and
whole
body
of
both
species
tracked
the
expected
values
fairly
well
and
were
generally
consistent
(
of
similar
slope)
within
each
series,
namely,
liver
or
whole
body
for
each
species
(
Figure
4).
The
curving
slopes
of
the
medaka
liver
most
precisely
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
24
track
the
standard
series
expected
slope,
with
the
lowest
VTG
value
for
uninduced
males
(
Code
1),
increasing
with
each
category
to
peak
at
induced
females
(
Code
4),
and
to
drop
lower
again
for
the
positive
control
(
Code
5).
The
"
M"­
shaped
tendency
of
the
slopes
in
the
whole
body
curves
for
both
species
would
indicate
that
in
these
cases,
the
uninduced
females
exhibited
higher
VTG
concentrations
than
did
induced
males,
and
that
there
was
a
small,
or
no
difference
between
the
values
in
females,
whether
induced
or
not.
There
are
indeed
biological
factors
that
could
account
for
these
results:
for
example,
the
females
of
both
species
were
laying
eggs
at
a
high
rate
during
the
period
of
the
experiment,
and
could
therefore
have
had
elevated
VTG
levels
in
any
case,
or
have
been
more
resistant
to
effects
from
exposure
to
estrogen­
like
compounds
in
the
environment
(
Irv
Schultz,
personal
communications,
July
2003).
Brion
et
al.
(
2002)
reported
substantial
difference
in
VTG
levels
(
from
3.97
±
2.7
µ
g/
mL
to
442.5
±
180
µ
g/
mL)
of
unexposed
females
depending
on
their
state
of
maturity.

The
degree
to
which
one
must
focus
on
the
absolute
value
of
the
VTG
measurements
measured
by
ELISA
must
be
tempered
by
the
understanding
that
there
are
sources
of
variation
within
the
tests
that
can
influence
the
precision
of
quantification.
For
example,
as
the
natural
degradation
of
VTG
occurs
in
the
samples,
the
large
molecule
may
be
broken
down
into
several
smaller
components,
and
the
VTG
antibody
might
detect
all
of
the
"
pieces"
and
count
each
as
a
molecule
of
VTG,
resulting
in
higher
than
expected
values.
Similarly,
the
VTG
antibody
could
cross­
react
with
other
proteins
in
the
sample,
and
detect
them
as
VTG
when
they
are
not
actually
the
specific
target.

There
was
a
wide
range
of
within­
run
variability,
but
75%
of
the
coefficients
of
variation
were
low,
<
10%
and
<
14%
for
zebra
fish
and
medaka,
respectively.
That
is,
it
was
typically
one
of
the
methods
out
of
four
that
accounted
for
some
outlier
values
that
increased
the
coefficient
of
variation,
whereas
the
others
showed
very
low
variability,
and
their
within­
range
values
were
closely
clustered.
The
overview
of
within­
run
variability
shown
in
Figures
2
and
3
above
(
see
7.0
Results)
graphically
demonstrates
this
observation,
and
confirms
the
fit
to
the
general
trend
of
concentrations
in
the
standard
series.

The
range
of
intra­
assay
variability
among
the
methods
in
the
study
was
also
broad
for
both
zebra
fish
and
medaka,
with
coefficients
of
variation
almost
three
times
higher
(<
30%
for
zebra
fish,
and
<
41%
for
medaka)
than
those
in
the
within­
run
tests,
but
still
moderate.
Although
the
absolute
VTG
concentrations
measured
by
the
various
methods
varied
by
orders
of
magnitude,
the
objective
of
applying
ELISA
methods
to
detect
the
standard
series
of
concentrations
in
liver
and
whole
body
homogenates
was
generally
met.
In
general,
the
better
fit
was
shown
by
the
medaka
applications
than
by
those
used
for
zebra
fish,
and
the
medaka
liver
concentrations
most
closely
represented
the
expected
concentration
series
as
analyzed
by
all
methods
for
medaka.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
25
1
10
100
1,000
10,000
100,000
1,000,000
10,000,000
100,000,000
0
1
2
3
4
5
6
Concentration
Code
Mean
VTG
Concentration
Lab
1
Lab
2
Lab
4
Lab
5
Lab
7
Lab
8
Lab
9
Lab
10
Lab
12
Max
Detection
Limit
A:
Zebra
fish
liver
B:
Zebra
fish
whole
body
C:
Medaka
liver
D.
Medaka
whole
body
Figure
4.
VTG
measurement
by
laboratory
for
each
concentration
code
averaged
over
standard
and
assay
for
(
a)
zebra
fish
liver;
(
b)
zebra
fish
whole
body;
(
c)
medaka
liver;
and
(
d)
medaka
whole
body
samples
(
see
Figures
C4,
C5,
C13,
and
C14
in
Appendix
C)
(
Code
1
=
uninduced
male;
Code
2
=
uninduced
female;
Code
3
=
induced
male;
Code
4
=
induced
female;
Code
5
=
positive
control)

A
recent
EPA
study
compared
VTG
methods
for
fathead
minnow
(
Pimephales
promelas)
whole
body
homogenate
and
blood
plasma
(
EPA
Work
Assignment
WA
2­
19,
Battelle
2003),
and
reported
that
the
trend
of
the
standard
series
followed
the
expected
increasing
levels
of
VTG
for
all
methods
in
the
comparison,
even
though
there
were
large
variations
in
the
actual
quantities
of
VTG
measured.
That
is,
one
method
consistently
reported
higher
actual
values
than
the
others;
however,
the
methods
under
consideration
varied
more
dramatically
in
the
types
of
antibodies
used
(
one
method
used
carp
instead
of
fathead
minnow
antibody,
for
example),
than
did
the
methods
used
in
the
present
study.
In
the
present
zebra
fish
and
medaka
study,
there
were
more
participating
laboratories
that
used
the
same
method
for
comparison;
up
to
six
laboratories
applied
a
single
method
for
zebra
fish,
for
example.

The
comparison
of
data
by
standard
offers
a
perspective
on
method
specificity
and
standardization
in
a
screening
assay.
The
percentage
difference
between
the
average
replicate
VTG
concentrations
obtained
with
the
homologous
and
the
purified
standard
was
calculated.
The
comparison
showed
little
variation
between
the
two
for
medaka
for
all
three
methods
applied;
however,
for
zebra
fish,
the
differences
were
small
for
all
concentrations
for
two
methods,
but
quite
1
10
100
1,000
10,000
100,000
1,000,000
10,000,000
100,000,000
0
1
2
3
4
5
6
Concentration
Code
Mean
VTG
Concentration
Lab
1
Lab
2
Lab
4
Lab
5
Lab
11
Lab
11
Lab
11
Lab
12
1
10
100
1,000
10,000
100,000
1,000,000
10,000,000
0
1
2
3
4
5
6
Concentration
Code
Mean
VTG
Concentration
Lab
1
Lab
2
Lab
4
Lab
5
Lab
11
Lab
11
Lab
11
Lab
12
1
10
100
1,000
10,000
100,000
1,000,000
10,000,000
100,000,000
0
1
2
3
4
5
6
Concentration
Code
Mean
VTG
Concentration
Lab
1
Lab
2
Lab
4
Lab
5
Lab
7
Lab
8
Lab
9
Lab
10
Lab
12
Max
Detection
Limit
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
26
large
in
the
results
of
a
third
method.
In
the
latter,
VTG
concentrations
were
about
700%
to
800%
lower
for
the
purified
than
for
the
homologous
standard.

The
comparable,
commercially
available
sandwich
ELISA
kit
used
by
the
majority
of
laboratories
for
both
species
yielded
the
highest
VTG
values.
For
example,
for
zebra
fish,
VTG
concentrations
measured
by
this
method
were
up
to
3277
times
greater
than
those
measured
by
another
method
for
liver.
The
most
dramatic
example
is
perhaps
the
measurement
of
VTG
in
uninduced
male
zebra
fish
by
different
methods:
VTG
was
undetected
by
one
method,
but
was
measured
as
136,117
ng/
mL
by
another,
the
aforementioned
commercial
ELISA
kit.
Nonetheless,
there
was
a
high
degree
of
variability
in
the
results
from
the
multiple
laboratories
that
applied
the
same
method,
both
for
zebra
fish
and
for
medaka.
For
both
the
purified
and
homologous
standard
in
the
zebra
fish
and
medaka
liver
homogenate,
the
highest
variability
was
seen
in
the
positive
controls,
and
for
whole
body,
in
the
uninduced
male
(
zebra
fish)
or
induced
male
(
medaka)
samples.
This
demonstrates
that
particularly
for
zebra
fish,
the
quantification
of
VTG
is
not
absolute,
but
rather,
depends
primarily
on
the
method
used
for
measurement.

Statistical
comparison
by
method
to
determine
whether
the
various
ELISAs
vary
in
their
ability
to
detect
differences
between
pairs
of
treatment
concentrations
yielded
a
moderately
strong
distinction
between
liver
and
whole
body
homogenates
in
the
zebra
fish.
With
one
method,
differences
in
the
low
end
of
the
concentration
spectrum
were
more
readily
discerned
for
liver
samples
used
with
the
homologous
standard,
whereas
with
the
purified
standard,
only
the
highest
and
lowest
values
were
significantly
distinguished.
For
whole
body
samples,
all
four
methods
could
detect
differences
in
VTG
values
in
at
least
two­
thirds
of
the
concentration
pairs,
and
could
distinguish
between
uninduced
males,
which
represent
the
low
end
of
the
concentration
spectrum,
and
the
three
other
treatments
that
represent
the
middle
concentration
range.
Analysis
of
whole
body
samples
used
with
purified
standard
found
that
all
three
methods
distinguished
uninduced
males
from
the
other
treatments,
again
with
focus
on
the
low
end
of
the
concentration
range.

In
medaka,
for
the
liver
and
whole
body
samples
and
both
standards,
the
results
were
identical:
two
methods
could
detect
differences
on
the
low
end
of
the
concentration
gradient,
namely,
between
the
lowest
VTG
concentration
group,
the
uninduced
males,
and
each
of
the
other
three
treatment
concentrations.
These
are
important
distinctions
for
discovery
of
induction
in
males
that
would
be
the
focus
in
endocrine
disrupter
screening
programs,
for
example.
None
of
the
methods
could
detect
differences
at
the
high
end
to
distinguish
induced
females
from
induced
males,
nor
in
the
midrange,
to
distinguish
uninduced
females
from
induced
males
and
induced
females.

Because
one
laboratory
used
all
three
methods
for
medaka
analysis,
a
comparison
of
the
methods
under
relatively
consistent
conditions
could
be
made.
For
both
liver
and
whole
body
samples,
very
high
VTG
concentrations
averaged
over
the
two
standards
were
obtained
by
application
of
one
method,
and
very
low
concentrations
by
another
method.
There
were
no
differences
shown
by
any
analytical
method
when
one
or
the
other
of
the
two
standards
was
used
for
the
measurement
of
VTG
in
liver
homogenates.
In
contrast,
for
whole
body,
the
VTG
concentrations
did
not
vary
between
standards
for
one
method,
varied
for
some
concentrations
in
the
standard
series
by
a
second
method,
and
differed
for
each
and
every
concentration
by
a
third
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
27
method.
All
three
methods
could
detect
differences
in
liver
and
whole
body
VTG
levels
for
at
least
four
or
five
of
the
six
possible
pairs
in
the
pairwise
comparison
using
both
standards.
Altogether,
the
one­
laboratory
three­
method
comparison
demonstrates
by
contrast
the
potentially
large
degree
of
variability
among
laboratories
that
could
account
for
some
of
the
disparity
of
results
reported
here.

Two
analytical
methods
were
exercised
by
more
than
one
laboratory
in
the
medaka
study,
providing
the
opportunity
to
evaluate
within­
method
variability.
For
example,
five
laboratories
used
the
same
method
to
determine
VTG
concentrations
in
liver
and
whole
body
medaka
samples
using
the
homologous
standard.
The
ability
of
the
assay
to
detect
differences
among
the
treatment
concentrations
depended
primarily
on
the
laboratory
performing
the
analyses,
secondarily
on
the
sample
type
and
by
the
standard
that
was
used.
Two
laboratories
used
a
second
method
for
analysis,
by
which
both
laboratories
detected
differences
both
in
liver
and
in
whole
body
VTG
concentrations
for
five
of
the
six
possible
pairwise
comparisons,
regardless
of
which
standard
was
used.
Again,
the
ability
of
the
assay
to
detect
differences
among
the
treatment
concentrations
depended
primarily
on
the
laboratory
performing
the
analyses,
and
next
on
the
sample
type
and
the
standard
that
was
used.
Nonetheless,
this
is
the
most
positive
result
for
any
of
the
methods
with
respect
to
within­
method
variability.
For
zebra
fish,
up
to
six
laboratories
used
a
single
method
for
liver
and
whole
body,
using
one
or
the
other
standard.
As
in
the
previous
comparisons,
detection
of
differences
among
treatment
concentrations
depended
first
on
the
laboratory
conducting
the
analysis,
and
next
on
the
sample
type
and
standard
used.

In
both
the
zebra
fish
and
medaka
studies,
a
positive
control
precisely
spiked
with
purified
VTG
was
analyzed.
The
concentration,
which
was
toward
the
lower
end
of
the
concentration
gradient
of
the
standard
series,
was
unknown
to
the
participating
laboratories.
As
an
example,
although
the
within­
laboratory
variability
in
the
percentage
recovery
of
the
spiked
concentration
from
liver
samples
was
relatively
low,
the
variability
among
laboratories
and
among
methods
was
moderate
(
medaka)
to
high
(
zebra
fish).
The
variability
was
in
the
range
of
several
orders
of
magnitude
for
these
positive
control
test
results.
However,
similar
results
are
seen
in
other
studies.
For
instance,
Brion
et
al.
(
2002)
reported
that
in
a
similar
ELISA
study,
the
control
group
yielded
VTG
concentrations
that
were
in
some
fish
less
than
the
practical
detection
limit
for
the
assay
(
i.
e.,
<
40
ng/
mL),
whereas
in
others,
as
high
as
560
ng/
mL.
The
relative
increase
of
VTG
in
males
with
exposure
to
an
estrogen
compound,
however,
is
a
clearly
marked
trend,
in
spite
of
the
variability
among
controls,
up
to
an
18,000­
fold
increase
in
VTG
for
exposed
males
Brion
et
al.'
s
study
(
2002).
Holbech
et
al.
(
2001)
saw
a
200­
fold
increase
in
VTG
levels
in
males
exposed
for
7
days
to
a
nominal
concentration
of
ethinylestradiol,
compared
with
levels
in
controls.
Of
course,
the
compound
and
the
concentration
thereof
used
for
the
exposures,
along
with
other
experimental
conditions
and
techniques
vary
from
one
study
to
another,
and
differ
from
the
specifications
of
the
present
study.
The
increases
seen
among
induced
compared
with
uninduced
males
in
the
present
study
were
typically
in
the
range
of
one
to
more
than
five­
hundredfold
among
medaka,
and
less
than
one
hundredfold
among
zebra
fish.
It
appears
that
the
quantitative
result
in
the
case
of
positive
controls
in
this
study
could
be
influenced
or
confounded
primarily
by
the
choice
of
method,
and
secondarily
by
the
performing
laboratory
and
by
the
selection
of
standard
for
calibration,
with
respect
to
the
two
species
of
fish
that
were
the
focus
of
this
survey.
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
28
In
the
categories
compared
statistically
in
this
study,
the
sources
of
the
variability
could
not
be
precisely
defined
based
on
the
present
results.
Due
to
the
design
of
the
study,
not
all
factors
could
be
addressed;
as
an
example,
potential
confounding
factors
associated
with
storage,
shipping,
and
handling
of
samples
were
not
evaluated.
The
possible
consequences
of
time
and
resource
restrictions
were
also
not
considered.
Because
all
of
the
participating
laboratories
donated
their
labor,
materials,
and
other
resources
to
conduct
the
analysis,
it
is
possible
that
some
portion
of
the
interlaboratory
variability
could
have
grown
from
consequent
limitations
 
that
is,
for
example,
perhaps
the
number
of
dilutions
that
could
be
run
for
each
sample
varied
among
laboratories,
depending
on
their
available
time
and
materials.
Differences
could
arise
from
any
number
of
other
potential
sources,
such
as
technical
issues
in
the
protocols,
or
particular
characteristics
of
a
laboratory's
performance
of
the
tests.
There
could
be
concern
about
the
range
of
method
detection
limits,
or
the
methods
by
which
they
are
determined.
Other
issues
could
include
instability
of
samples,
reagents,
or
other
materials
that
are
stored,
or
inherent
variability
among
the
source
populations
of
fish
used
in
the
study,
as
well
as
other
biological
factors
affecting
the
fish,
among
others.
Understanding
the
factors
capable
of
contributing
to
the
high
degree
of
variability
observed
in
the
present
survey
would
be
a
valuable
contribution
to
scientific
progress
in
this
area.

Various
researchers
have
demonstrated
the
utility
of
VTG
as
a
biomarker
for
providing
evidence
of
endocrine
disruption
in
fish.
Based
on
the
results
of
this
study,
it
can
be
said
that
most
of
the
laboratories
and
methods
considered
are
capable
of
distinguishing
changes
in
VTG
levels
in
zebra
fish
and/
or
medaka.
However,
there
are
issues
still
to
be
resolved
before
VTG
measurements
could
be
used
as
a
reliable
tool
in
screening
and
testing.
It
is
recommended
that
greater
effort
be
given
to
developing
specific
performance
criteria
for
VTG
analytical
methods.
It
is
also
recommended
that
a
single,
standardized
protocol
for
each
fish
species
be
used
in
quantifying
VTG
in
the
interlaboratory
validation
trials.

9.0
REFERENCES
Battelle.
2002.
Draft
Detailed
Review
Paper,
Fish
Screening
Assays
for
Endocrine
Disruption.
Prepared
for
the
U.
S.
Environmental
Protection
Agency
by
Battelle
Pacific
Northwest
National
Laboratory,
Sequim,
Washington.

Battelle.
2003.
Draft
Final
Report
on
Comparative
Evaluation
of
Vitellogenin
Methods.
Prepared
for
the
U.
S.
Environmental
Protection
Agency
by
Battelle
Pacific
Northwest
National
Laboratory,
Sequim,
Washington.

Brion,
F.,
B.
M.
Nilsen,
J.
K.
Eidem,
A.
Goks yr,
and
J.
M.
Porcher.
2002.
Development
and
validation
of
an
enzyme­
linked
immunosorbent
assay
to
measure
vitellogenin
in
the
zebrafish
(
Danio
rerio).
Environmental
Toxicology
and
Chemistry
28(
8):
1699­
1708.

Braunbeck,
T.,
H.
Holbech,
L.
Norrgren,
and
G.
Petersen.
2003.
Zebrafish
as
a
test
organism
in
endocrine
disrupter
research,
pp.
9­
14.
In
Proceedings
of
the
3rd
International
Medaka
Symposium,
Development
of
Test
Methods
with
Medaka
to
Detect
Endocrine
Disrupting
Chemicals,
Session
1:
Battelle
Draft:
May
be
subject
to
change
following
QA,
editorial,
and
managerial
review
July
2003
29
Each
fish's
activity
of
evaluation
of
endocrine
disrupting
chemicals,
27­
28
February
2003,
Okazaki,
Japan.
Available
online:
http://
www.
env.
go.
jp/
chemi/
end/
medaka_
sympo.
html.

Edmunds,
J.
S.,
R.
A.
McCarthy,
and
J.
S.
Ramsdell.
2000.
Permanent
and
functional
male­
to­
female
sex
reversal
in
d­
rR
strain
medaka
(
Oryzias
latipes)
following
egg
microinjection
of
o,
p;­
DDT.
Environmental
Health
Perspectives
108(
3):
219­
224.

Holbech,
H.,
L.
Andersen,
G.
I.
Petersen,
B.
Korsgaard,
K.
L.
Pedersen,
and
P.
Bjerregaard.
2001.
Development
of
an
ELISA
for
vitellogenin
in
whole
body
homogenate
of
zebrafish
(
Danio
rerio).
Comparative
Biochemistry
and
Physiology
Part
C
130(
2001):
119­
131.

Kang,
I.
J.,
H.
Yokota,
Y.
Oshima,
Y.
Tsuruda,
T.
Oe,
N.
Imada,
H.
Tadorokoro,
and
T.
Honjo.
2002.
Effects
of
bisphenol
A
on
the
reproduction
of
Japanese
medaka
(
Oryzias
latipes).
Environmental
Toxicology
and
Chemistry
21(
11):
2394­
2400.

Kordes,
C.,
E.
P.
Rieber,
and
H.
O.
Gutzeit.
2002.
An
in
vitro
vitellogenin
bioassay
for
oestrogenic
substances
in
the
medaka
(
Oryzias
latipes).
Aquatic
Toxicology
58(
3­
4):
151­
164.

Nishi,
K.,
M.
Chikae,
Y.
Hatano,
H.
Mizukami,
M.
Yamashita,
R.
Sakakibara,
and
E.
Tamiya.
2002.
Development
and
application
of
a
monoclonal
antibody­
based
sandwich
ELISA
for
quantification
of
Japanese
medaka
(
Oryzias
latipes)
vitellogenin.
Comparative
Biochemistry
and
Physiology
Part
C
132(
2002):
161­
169.

Transgenic.
2002.
Medaka
ELISA
kit
information.
Available
online:
http://
www.
metocean.
co.
jp/
new/
inet/
vol2/
vol2_
information.
htm
(
in
Japanese).

Van
den
Belt,
K.,
P.
W.
Wester,
L.
T.
M.
Van
der
Ven,
R.
Verheyen,
and
H.
Witters.
2002.
Effects
of
ethynylestradiol
on
the
reproductive
physiology
in
zebrafish
(
Danio
rerio):
Time
dependency
and
reversibility.
Environmental
Toxicology
and
Chemistry
21(
4):
767­
775.
