DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
90
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
4.0
RESULTS:
TRENBOLONE
4.1
EPA
14­
Day
Assay
for
Trenbolone
The
EPA
14­
Day
Trenbolone
assay
was
conducted
from
February
4,
2003
to
February
11,
2003
(
prevalidation
test)
and
from
February
11,
2003
to
February
25,
2003
(
validation
test).

4.1.1
Survival
All
males
in
all
treatments
and
all
females
in
the
Control
and
High­
concentration
treatments
survived
the
EPA
14­
Day
Trenbolone
assay.
Two
females
in
the
Low­
concentration
treatment
died
during
the
assay
(
88%
survival).

4.1.2
Vitellogenin
Vitellogenin
concentrations
in
Control
treatment
females
used
during
the
EPA
14­
Day
Trenbolone
assay
ranged
from
1,385,500
ng/
mL
to
3,722,500
ng/
mL
(
Figure
4.1).
Among
females
exposed
to
the
two
trenbolone
concentrations,
vitellogenin
concentrations
ranged
from
0
ng/
mL
(
not
detected)
to
5,847,500
ng/
mL.
Significant
differences
in
the
mean
vitellogenin
concentration
per
treatment
(
Table
4.1)
were
detected
(
Kruskal­
Wallis,
H
=
31.72,
p
=
<
0.001,
df
=
2).
Vitellogenin
concentrations
in
Controltreatment
females
and
Low­
concentration
females
were
significantly
greater
than
those
in
females
exposed
to
the
High
trenbolone
concentrations.
The
achieved
power
for
this
endpoint
was
100%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
4
(
Table
4.1).

Table
4.1.
Summary
statistics
and
power
estimates
for
female
vitellogenin
concentrations
(
ng/
mL)
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
16
2,608,969
592,805
23%
100%
4
Low
14
2,237,993
1,798,862
80%
High
16
40,910
79,868
195%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
14.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
91
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
6000000
5000000
4000000
3000000
2000000
1000000
0
treatment
Trenb­
VTG
Figure
4.1.
Boxplot
of
female
vitellogenin
concentration
(
ng/
mL)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
the
asterisk
represents
a
probable
outlier.

Vitellogenin
concentrations
in
Control
treatment
males
used
during
the
EPA
14­
Day
Trenbolone
assay
ranged
from
0
ng/
mL
(
not
detected)
to
8,813
ng/
mL
(
Figure
4.2).
Among
males
exposed
to
the
two
trenbolone
concentrations,
vitellogenin
concentrations
ranged
from
0
ng/
mL
to
2,333
ng/
mL.
No
significant
differences
in
the
mean
vitellogenin
concentration
per
treatment
(
Table
4.2)
were
detected
(
Kruskal­
Wallis,
H
=
1.26,
p
=
0.532,
df
=
2).
The
achieved
power
for
this
endpoint
was
14%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
66
(
Table
4.2).

Table
4.2.
Summary
statistics
and
power
estimates
for
male
vitellogenin
concentrations
(
ng/
mL)
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
1,811
3,061
169%
14%
66
Low
8
774
1,033
133%
High
8
447
226
51%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
8.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
92
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
treatment
Trenb­
VTG
Figure
4.2.
Boxplot
of
male
vitellogenin
concentration
(
ng/
mL)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
the
asterisk
represents
a
probable
outlier.

4.1.3
Appearance
/
Secondary
Sex
Characteristics
All
of
the
Control­
treatment
females
used
during
the
trenbolone
EPA
14­
Day
Trenbolone
assay
exhibited
typical
female
morphology
(
no
fat
pad,
no
tubercles,
no
vertical
banding,
ovipositor
present).
Five
of
the
15
females
exposed
to
the
Low
concentration
had
vertical
banding.
Among
the
16
females
exposed
to
the
High
concentration,
13
had
tubercles,
1
had
a
dorsal
fat
pad,
and
12
had
vertical
banding.
Thus,
trenbolone
appeared
to
have
a
dose­
related
effect
on
female
morphology.

All
of
the
males
used
during
the
EPA
14­
Day
Trenbolone
assay
exhibited
typical
male
morphology
(
fat
pads,
tubercles,
vertical
banding,
no
ovipositor
present).

4.1.4
Gonadosomatic
Index
The
range
of
GSI
values
calculated
for
females
in
the
all
treatments
varied
from
two­
to
three­
fold
(
Figure
4.3),
and
the
overall
variability
within
the
treatment
was
moderate
(
CVs
=
23%
 
27%;
Table
4.3).
GIS
values
for
fish
in
the
Control
treatment
ranged
from
6.4
to
14.3.
The
highest
female
GSI
value
was
about
24
(
one
fish
in
the
High
concentration).
No
other
fish
had
a
GSI
value
greater
than
18.6.
A
significant
difference
in
the
mean
GSI
value
per
treatment
(
Table
4.3)
was
detected
(
Kruskal­
Wallis,
H
=
9.84,
p
=
0.007,
df
=
2).
The
test
indicated
that
the
mean
GSI
for
females
in
the
Control
treatment
was
less
than
those
for
the
other
two
treatments.
However,
the
mean
GSI
values
calculated
for
the
Low
and
High
trenbolone
concentrations
were
not
statistically
different.
The
achieved
power
for
this
endpoint
was
78%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
15
(
Table
4.3).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
93
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.3.
Summary
statistics
and
power
estimates
for
female
gonadosomatic
index
data
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
16
10.5
2.6
25%
78%
15
Low
14
13.7
3.1
23%
High
16
14.3
3.8
27%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
14.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
High
Control
25
15
5
treatment
GSI
Figure
4.3.
Boxplot
of
female
GSI
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.

The
range
of
most
GSI
values
calculated
for
males
during
the
EPA
14­
Day
Trenbolone
assay,
was
small,
ranging
from
0.7
to
1.8
(
Figure
4.4),
which
approximates
the
typical
range
for
reproductively­
active
male
fathead
minnows.
The
highest
and
lowest
male
GSI
values
were
2.2
(
one
fish
the
High
concentration)
and
0.7
(
two
fish
in
the
Control
treatment),
respectively.
A
significant
difference
in
the
mean
GSI
value
per
treatment
(
Table
4.4)
was
detected
(
Kruskal­
Wallis,
H
=
7.66,
p
=
0.022,
df
=
2).
The
test
indicated
that
the
mean
GSI
for
males
in
the
High
concentration
was
greater
than
those
for
the
other
two
treatments.
The
achieved
power
for
this
endpoint
was
71%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
10
(
Table
4.4).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
94
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.4.
Summary
statistics
and
power
estimates
for
male
gonadosomatic
index
data
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
1.04
0.28
27%
71%
10
Low
8
1.02
0.33
33%
High
8
1.56
0.39
25%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
8.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
High
Control
2.0
1.5
1.0
treatment
GSI
(
y
)

Figure
4.4.
Boxplot
of
male
GSI
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.

4.1.5
Female
Gonad
Histology
General
Ovary
Staging
 
Statistical
analysis
of
the
mean
ovarian
staging
from
12
microscopic
fields
per
female
in
the
EPA
14­
Day
Trenbolone
assay
revealed
no
significant
difference
among
treatments
(
Kruskal­
Wallis,
H
=
2.21,
p
=
0.331,
df
=
2).

Quantitative
Ovarian
Staging
 
One
hundred
cells
in
each
of
three
sections
per
female
were
examined
to
quantitatively
determine
the
developmental
stage
of
the
ovaries.
Ova
from
fish
in
all
treatments
ranged
from
Stage
1a
to
Stage
5
(
see
Methods
for
a
description
of
the
stages
(
Figure
4.5).
Variability
within
treatments
for
each
stage
was
very
high
as
indicated
by
CVs
that
ranged
as
high
as
400%
(
Table
4.5).
Although
statistical
analyses
showed
that
there
was
a
significant
difference
among
treatments
in
the
proportion
of
cells
in
developmental
Stage
3
and
Stage
4,
there
were
no
significant
differences
among
treatments
in
the
proportion
of
cells
in
the
developmental
Stages
1a,
1b,
2,
and
5
(
Table
4.5).
The
proportion
of
cells
in
developmental
Stage
3
in
the
High
concentration
was
significantly
greater
than
that
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
95
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
in
the
Control
treatment
and
the
Low
concentration.
The
proportion
of
cells
in
developmental
Stage
4
in
the
High
concentration
was
significantly
lower
than
that
in
the
Control
treatment
and
the
Low
concentration.
Thus,
it
can
be
concluded
that
the
trenbolone
High
concentration
inhibited
the
development
of
ova
from
the
early
to
late
vitellogenic
stages
(
i.
e.
from
stage
3
to
stage
4).

Table
4.5.
Descriptive
statistics
of
the
proportion
of
ovarian
cells
in
each
developmental
stage
for
females
from
the
EPA
14­
Day
Trenbolone
assay
and
results
of
the
Kruskal­
Wallis
Test
(
df
=
2)
comparing
treatments.

Control
(
n
=
16)
Low
(
n
=
15)
High
(
n
=
16)
Kruskal­
Wallis
Stage
Mean
Stdev
CV
Mean
Stdev
CV
Value
Stdev
CV
H
p
1a
0.104
0.037
36%
0.096
0.033
35%
0.082
0.023
28%
2.31
0.315
1b
0.321
0.064
20%
0.304
0.057
19%
0.294
0.059
20%
1.56
0.459
2
0.194
0.050
26%
0.174
0.044
25%
0.161
0.042
26%
5.77
0.056
3
0.138
0.037
27%
0.158
0.051
32%
0.291
0.056
19%
29.47
<
0.001*
4
0.215
0.091
42%
0.241
0.079
33%
0.130
0.068
52%
14.39
0.001**
5
0.003
0.011
339%
0.001
0.003
300%
0.001
0.003
400%
0.42
0.809
*
p
<
0.001
**
p
<
0.01
0.0
0.1
0.2
0.3
0.4
0.5
1a
1b
2
3
4
5
Ovarian
Stage
Proportion
of
Cells
Control
Low
High
Figure
4.5.
Frequency
histogram
showing
the
quantitative
developmental
staging
of
ovaries
for
each
treatment
of
the
EPA
14­
Day
Trenbolone
assay.
For
each
treatment,
the
columns
represent
the
grand
mean
proportion
of
cells
in
each
stage
and
the
bars
represent
the
standard
deviation.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
96
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Atretic
Follicles
 
The
mean
proportion
of
atretic
follicles
per
300
follicles
(
counted
per
fish)
ranged
from
0.002
for
females
in
the
Control
treatment
to
0.03
follicles
for
females
in
the
High
concentration
(
Figure
4.6).
There
was
a
significant
difference
in
the
proportions
of
atretic
follicles
among
treatments
(
Kruskal­
Wallis,
H
=
23.99,
p
=
<
0.001,
df
=
2).
The
proportion
of
atretic
follicles
in
females
from
the
High
concentration
was
significantly
greater
than
that
in
Control
treatment
and
Low
concentration
females.
This
appears
to
be
related
to
the
inhibition
of
development
of
ova
from
stage
3
to
stage
4
in
the
High­
concentration
treatment.

Low
High
Control
0.2
0.1
0.0
treatment
atretic_
follicles
Figure
4.6.
Boxplot
of
the
proportion
of
atretic
follicles
per
300
follicles
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
asterisks
represent
probable
outliers.

Corpora
Lutea
 
The
mean
proportion
of
corpora
lutea
per
300
follicles
(
counted
per
fish)
ranged
from
0.01
for
females
in
the
High
concentration
to
0.02
for
females
in
the
Control
treatment
and
the
Low
concentration
(
Figure
4.7).
There
was
a
significant
difference
in
the
proportions
of
corpora
lutea
among
treatments
(
Kruskal­
Wallis,
H
=
7.04,
p
=
0.030,
df
=
2).
The
value
for
the
High
concentration
was
significantly
lower
than
that
of
the
Low­
concentration
treatment.
This
may
be
related
to
the
inhibition
of
ovarian
development
in
the
High
concentration,
so
that
over
time,
while
ovarian
development
tends
to
be
stalled
between
stages
3
and
4,
the
corpora
lutea
in
the
High
dose
females
are
resorbed.

Observations
 
One
female
in
the
Control
treatment
had
abnormal
Stage
3
ova
that
were
more
typical
of
Stage
2,
but
had
vitellogenin
granules
in
the
cytoplasm.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
97
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
treatment
corpora_
lutea
(
y
)

Figure
4.7.
Boxplot
of
the
proportion
of
corpora
lutea
per
300
follicles
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
asterisks
represent
probable
outliers.

4.1.6
Male
Gonad
Histology
Testes
Staging
by
Microscopic
Field
 
Testes
from
males
exposed
to
trenbolone
during
the
EPA
14­
Day
Trenbolone
assay
were
examined
to
determine
the
general
developmental
condition.
Males
in
all
treatments
had
well­
developed
testes
with
most
showing
Stage
4
and
Stage
5
development
(
see
Methods
for
description
of
developmental
stages).
All
of
the
96
microscopic
fields
examined
in
the
8
Control
treatment
males
showed
Stage
4
(
91
fields)
or
Stage
5
(
5
fields)
development.
All
of
the
96
fields
examined
in
the
8
Low­
concentration
treatment
males
showed
Stage
4
(
90
fields)
or
Stage
5
(
6
fields)
development.
All
of
the
96
microscopic
fields
examined
in
the
8
High­
concentration
treatment
males
showed
Stage
4
(
58
fields)
or
Stage
5
(
38fields)
development.
Statistical
analysis
of
the
mean
staging
from
12
microscopic
fields
per
fish
revealed
a
significant
difference
among
treatments
(
Kruskal­
Wallis,
H
=
12.36,
p
=
0.002,
df
=
2).
The
mean
testes
stage
for
males
in
the
High
concentration
(
4.40,
sd
=
0.29)
was
significantly
greater
than
that
in
the
Control
treatment
(
4.05,
sd
=
0.09)
and
Low­
concentration
treatment(
4.06,
sd
=
0.10).

Quantitative
Testicular
Staging
 
One
hundred
cells
in
each
of
three
sections
per
male
were
examined
to
quantitatively
determine
the
developmental
condition
of
the
testes.
The
developmental
stage
of
the
testes
from
fish
in
all
treatments
ranged
from
Stage
2a
to
Stage
5
(
Figure
4.8).
Variability
within
treatments
for
each
stage
was
very
high
as
indicated
by
CVs
that
ranged
as
high
as
129%
(
Table
4.6).
Although
statistical
analyses
showed
that
there
was
a
significant
difference
among
treatments
in
the
proportion
of
cells
in
developmental
Stage
3a,
there
were
no
significant
differences
among
treatments
in
the
proportion
of
cells
in
developmental
Stages
2a,
2b,
3b,
4,
and
5
(
Table
4.6).
Therefore,
there
was
no
consistent
pattern
of
significant
difference
associated
with
trenbolone
dose.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
98
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.6.
Descriptive
statistics
of
the
proportion
of
testes
cells
in
each
developmental
stage
for
males
from
the
EPA
14­
Day
Trenbolone
assay
and
results
of
the
Kruskal­
Wallis
Test
(
df
=
2)
comparing
treatments.

Control
(
n
=
8)
Low
(
n
=
8)
High
(
n
=
8)
Kruskal­
Wallis
Stage
Mean
Stdev
CV
Mean
Stdev
CV
Value
Stdev
CV
H
p
1
0
0
­­
0
0
­­
0
0
­­
 
 
2a
0.006
0.005
72%
0.003
0.004
129%
0.003
0.003
113%
3.49
0.175
2b
0.018
0.008
41%
0.010
0.006
62%
0.013
0.012
92%
3.24
0.198
3a
0.167
0.060
36%
0.186
0.062
33%
0.045
0.037
80%
12.32
0.002*
3b
0.250
0.084
34%
0.204
0.095
47%
0.181
0.089
49%
2.22
0.329
4
0.173
0.056
33%
0.156
0.044
29%
0.127
0.111
88%
2.27
0.322
5
0.386
0.178
46%
0.441
0.184
42%
0.632
0.237
38%
4.75
0.093
*
p
<
0.01
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1
2a
2b
3a
3b
4
5
Testicular
Stage
Proportion
of
Cells
Control
Low
High
Figure
4.8.
Frequency
histogram
showing
the
quantitative
developmental
staging
of
testes
for
each
treatment
of
the
EPA
14­
Day
Trenbolone
assay.
For
each
treatment,
the
columns
represent
the
grand
mean
proportion
of
cells
in
each
stage
and
the
bars
represent
the
standard
deviation.

Tubule
Diameter
 
The
average
diameter
of
the
seminiferous
tubules
of
males
from
the
Control
treatment
ranged
from
103.6
µ
m
to
136.4
µ
m
(
Figure
4.9).
Tubule
diameters
of
males
from
the
two
test
concentrations
ranged
from
90.8
µ
m
to
172.8
µ
m.
A
significant
difference
in
the
mean
tubule
diameter
per
treatment
(
Table
4.7)
was
detected
(
Kruskal­
Wallis,
H
=
6.86,
p
=
0.032,
df
=
2).
However,
the
pattern
of
significance
(
Low
<
Control
<
High)
was
not
consistent
with
the
trenbolone
dose.
The
achieved
power
for
this
endpoint
was
24%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
33
(
Table
4.7).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
99
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.7.
Summary
statistics
and
power
estimates
for
male
seminiferous
tubule
diameter
data
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
125.4
13.6
11%
24%
33
Low
8
112.1
20.7
18%
High
8
138.6
23.4
17%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
8.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
180
170
160
150
140
130
120
110
100
90
treatment
diameter
p
y
Figure
4.9.
Boxplot
of
male
seminiferous
tubule
diameter
(
µ
m)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
asterisk
represents
a
probable
outlier.

Observations
 
No
interstitial
Sertoli
cell
proliferation
and
no
Leydig
cell
proliferation
was
observed
for
any
treatment.
No
testicular
atrophy
was
recorded
and
no
ovatestes
were
observed
for
any
treatment.

4.1.7
Plasma
Steroid
Concentrations
Estradiol
 
Estradiol
concentrations
in
Control­
treatment
females
used
during
the
EPA
14­
Day
Trenbolone
assay
ranged
from
1,434
pg/
mL
to
4,553
pg/
mL
(
Figure
4.10).
Among
females
exposed
to
the
two
trenbolone
concentrations,
estradiol
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
5,109
pg/
mL.
A
significant
difference
in
the
mean
estradiol
concentration
per
treatment
(
Table
4.8)
was
detected
(
Kruskal­
Wallis,
H
=
24.55,
p
<
0.001,
df
=
2).
The
mean
estradiol
concentration
in
females
from
the
High
concentration
was
less
than
that
in
females
from
the
Low
concentration
and
the
Control
treatment.
The
achieved
power
for
this
endpoint
was
98%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
6
(
Table
4.8).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
100
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.8.
Summary
statistics
and
power
estimates
for
female
estradiol
concentrations
(
pg/
mL)
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
15
2,907
972
33%
98%
6
Low
10
2,512
1,740
69%
High
15
388
620
160%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
10.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
5000
4000
3000
2000
1000
0
treatment
Estradiol
(
y
)

Figure
4.10.
Boxplot
of
female
estradiol
concentration
(
pg/
mL)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisks
represent
probable
outliers.

Estradiol
concentrations
in
Control
treatment
males
used
during
the
EPA
14­
Day
Trenbolone
assay
ranged
from
188
pg/
mL
to
698
pg/
mL
(
Figure
4.11).
Among
males
exposed
to
the
two
trenbolone
concentrations,
estradiol
concentrations
ranged
from
184
pg/
mL
to
513
pg/
mL.
No
significant
differences
in
the
mean
estradiol
concentration
per
treatment
(
Table
4.9)
were
detected
(
Kruskal­
Wallis,
H
=
1.55,
p
=
0.460,
df
=
2).
The
achieved
power
for
this
endpoint
was
7%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
99
(
Table
4.9).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
101
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.9.
Summary
statistics
and
power
estimates
for
male
estradiol
concentrations
(
pg/
mL)
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
322
166
52%
7%
99
Low
4
331
143
43%
High
6
352
62
18%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
700
600
500
400
300
200
treatment
Estradiol
(
y
)

Figure
4.11.
Boxplot
of
male
estradiol
concentration
(
pg/
mL)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.

Testosterone
 
Testosterone
concentrations
in
Control­
treatment
females
used
during
the
EPA
14­
Day
Trenbolone
assay
ranged
from
492
pg/
mL
to
3,047
pg/
mL
(
Figure
4.12).
Among
females
exposed
to
the
two
trenbolone
concentrations,
testosterone
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
691
pg/
mL.
Significant
differences
in
the
mean
testosterone
concentration
per
treatment
(
Table
4.10)
were
detected
(
Kruskal­
Wallis,
H
=
19.78,
p
<
0.001,
df
=
2).
The
mean
testosterone
concentration
in
females
from
the
High
concentration
was
less
than
that
of
females
from
the
Low
concentration
and
the
Control
treatment.
The
achieved
power
for
this
endpoint
was
70%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
4
(
Table
4.10).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
102
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.10.
Summary
statistics
and
power
estimates
for
female
testosterone
concentrations
(
pg/
mL)
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
1,515
911
60%
70%
4
Low
3
484
184
38%
High
12
93
121
129%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
3.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
3000
2000
1000
0
treatment
Testosterone
(
means
are
indicated
by
solid
circles)

Figure
4.
12.
Boxplot
of
female
testosterone
concentration
(
pg/
mL)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.

Testosterone
concentrations
in
Control
treatment
males
used
during
the
EPA
14­
Day
Trenbolone
assay
ranged
from
1,765
pg/
mL
to
4,349
pg/
mL
(
Figure
4.13).
Among
males
exposed
to
the
two
trenbolone
concentrations,
testosterone
concentrations
ranged
from
1,038
pg/
mL
to
5,824
pg/
mL.
No
significant
differences
in
the
mean
testosterone
concentration
per
treatment
(
Table
4.11)
were
detected
(
Kruskal­
Wallis,
H
=
0.70,
p
=
0.703,
df
=
2).
The
achieved
power
for
this
endpoint
was
8%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
113
(
Table
4.11).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
103
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.11.
Summary
statistics
and
power
estimates
for
male
testosterone
concentrations
(
pg/
mL)
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
7
2,899
809
28%
8%
113
Low
6
3,227
1,668
52%
High
7
2,787
2,002
72%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
6.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
6000
5000
4000
3000
2000
1000
treatment
Testosterone
Figure
4.13.
Boxplot
of
male
testosterone
concentration
(
pg/
mL)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

11­
ketotestosterone
 
11­
ketotesosterone
was
not
detected
in
females
from
the
Control
treatment
(
4
individuals),
from
the
Low
concentration
(
4
individuals),
or
from
the
High
concentration
(
11
individuals).

11­
ketotesosterone
concentrations
in
Control
treatment
males
used
during
the
EPA
14­
Day
Trenbolone
assay
ranged
from
14,810
pg/
mL
to
33,093
pg/
mL
(
Figure
4.14).
Among
males
exposed
to
the
two
trenbolone
concentrations,
11­
ketotesosterone
concentrations
ranged
from
2,090
pg/
mL
to
120,800
pg/
mL.
No
significant
differences
in
the
mean
11­
ketotesosterone
concentration
per
treatment
(
Table
4.12)
were
detected
(
Kruskal­
Wallis,
H
=
4.97,
p
=
0.083,
df
=
2).
The
achieved
power
for
this
endpoint
was
31%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
22
(
Table
4.12).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
104
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.12.
Summary
statistics
and
power
estimates
for
male
11­
ketotesosterone
concentrations
(
pg/
mL)
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
24,652
5,927
24%
31%
22
Low
7
34,349
19,619
57%
High
8
23,835
40,136
168%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
7.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
100000
50000
0
treatment
11­
keto
(
y
)

Figure
4.14.
Boxplot
of
male
11­
ketotesosterone
concentration
(
pg/
mL)
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

4.1.8
Fecundity
Total
Fecundity
 
Variability
among
treatments
in
the
total
number
of
eggs
produced
during
the
EPA
14­
Day
Trenbolone
assay
was
very
high
(
Figure
4.15).
Total
counts
in
the
Control
treatment
ranged
from
1,951
eggs
to
3,227
eggs.
Total
counts
for
three
replicates
of
the
High
trenbolone
concentration
treatment
were
similar,
ranging
from
280
eggs
to
346
eggs,
whereas
the
total
number
of
eggs
produced
in
the
fourth
replicate
was
666.
It
is
also
notable
that
the
primary
egg
production
in
three
High
trenbolone
concentration
replicates
occurred
only
on
Day
1
(
one
replicate
yielded
30
eggs
on
Day
12).
Egg
production
in
the
fourth
High
trenbolone
concentration
occurred
only
on
Days
2
and
3.
The
minimum
and
maximum
production
among
replicates
in
the
Low
trenbolone
concentration
ranged
from
2339
eggs
to
3140
eggs.
Statistical
analysis
of
square­
root
transformed
egg
counts
showed
significant
amongtreatment
differences
in
the
mean
cumulative
number
of
eggs
produced
per
treatment
(
Table
4.13)
(
1­
way
ANOVA,
F
=
63.99,
p
<
0.001,
df
=
2,
9).
Dunnett's
comparison
identified
significant
differences
in
mean
egg
production
between
the
High
trenbolone
concentration
and
the
Control
treatment,
and
also
between
the
High
and
Low
concentrations.
The
Low
concentration
and
the
Control
treatment
fecundity
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
105
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
values
were
not
statistically
different.
The
achieved
power
for
this
assay
was
100%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
2
(
Table
4.13).

Table
4.13.
Summary
statistics
and
power
estimates
for
fecundity
data
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
4
2654
600
23%
100%
2
Low
4
2689
348
13%
High
4
400
180
45%
1
Calculated
from
square­
root
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
square­
root
transformed
data.

0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
0
2
4
6
8
10
12
14
16
Day
Cumulative
Number
of
Eggs
Control
Low
High
Figure
4.15.
Total
egg
production
by
replicate
per
treatment
for
the
EPA
14­
Day
Trenbolone
assay.

Fecundity
per
Female
Reproductive
Day
 
During
the
EPA
14­
Day
Trenbolone
assay,
the
maximum
number
of
female
reproductive
days
was
achieved
for
the
Control
treatment
and
the
High
concentration,
whereas
55
female
reproductive
days
were
achieved
in
the
Low
concentration
(
Table
4.14).
The
number
of
eggs
produced
per
female
reproductive
day
in
the
Control
treatment
varied
from
34.8
eggs
to
57.6
eggs
and
from
41.8
eggs
to
58.1
eggs
in
the
Low
concentration
(
Figure
4.16).
For
the
High
concentration,
the
number
of
eggs
produced
per
female
reproductive
day
ranged
from
5.0
eggs
to
11.9
eggs,
with
fish
in
three
of
the
replicates
producing
fewer
than
7
eggs
per
day
(
5.0,
5.5,
6.2).
A
significant
difference
in
the
mean
number
of
eggs
produced
per
female
reproductive
day
among
treatments
was
detected
(
Kruskal­
Wallis,
H
=
7.42,
p
=
0.024,
df
=
2).
Fewer
eggs
were
produced
per
female
reproductive
day
in
the
High
concentration
than
in
either
the
Control
treatment
or
the
Low
concentration.
There
was
no
difference
in
the
mean
number
of
eggs
laid
per
female
reproductive
day
between
the
Control
treatment
and
the
Low
concentration.
The
achieved
power
for
this
assay
was
100%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
2
(
Table
4.14).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
106
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.14.
Summary
statistics
and
power
estimates
for
fecundity
per
female
reproductive
day
for
the
EPA
14­
Day
Trenbolone
assay.

Level
Mean
Number
of
Reproductive
Days
1
N
Mean
Stdev
CV
Achieved
Power
2
Sample
Size
Required
3
Control
56.0
4
47.4
10.7
23%
100%
2
Low
55.0
4
48.5
7.1
15%
High
56.0
4
7.2
3.2
45%
1
Maximum
number
=
56.
2
Calculated
from
natural
log
transformed
data;
with
sample
size
=
4.
3
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
60
50
40
30
20
10
0
treatment
eggs/
ReproDay
Figure
4.16.
Boxplot
of
the
number
of
eggs
produced
per
female
reproductive
day
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

Eggs
on
Tiles/
Dishes
 
The
mean
number
of
eggs
laid
on
the
tiles
among
the
treatments
during
the
EPA
14­
Day
Trenbolone
assay
varied
from
350
eggs
for
the
High
concentration
to
2526
eggs
for
the
Low
concentration
(
Appendix
E,
Table
2.3).
The
number
of
eggs
on
dishes
ranged
from
50
eggs
for
the
High
concentration
to
222
eggs
for
the
Control
treatment.
Because
of
the
variability
in
the
total
number
of
eggs
laid
per
treatment,
the
proportional
difference
in
the
number
of
eggs
on
dishes
versus
those
on
tiles
[
1 (#
eggs
on
dishes
÷
#
eggs
on
tiles)]
was
calculated.
The
proportional
difference
ranged
from
0.78
(
one
High
concentration
replicate)
to
0.96
(
one
Low
concentration
replicate)
(
Appendix
E,
Figure
2.3).
There
were
no
significant
differences
in
this
mean
proportional
difference
among
treatments
(
Kruskal­
Wallis,
H
=
3.23,
p
=
0.199,
df
=
2).

4.1.9
Fertilization
Success
Total
Fertilization
 
The
total
(
tiles
+
dishes)
fertilization­
success
rates
for
most
treatment
replicates
during
the
EPA
14­
Day
Trenbolone
assay
were
high,
ranging
from
0.981
(
one
High
concentration
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
107
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
replicate)
to
1.00
(
one
Control
treatment
replicate)
(
Figure
4.17).
A
significant
difference
in
mean
fertilization­
success
rates
(
Table
4.15)
among
treatments
was
detected
(
Kruskal­
Wallis,
H
=
7.65,
p
=
0.022,
df
=
2).
The
fertilization­
success
rate
was
lower
in
the
High
concentration
than
in
the
Control
treatment
or
the
Low
concentration.
There
was
no
difference
in
the
fertilization­
success
rate
between
the
Control
treatment
and
the
Low
concentration.
The
achieved
power
for
this
assay
was
89%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
4
(
Table
4.15).

Table
4.15.
Summary
statistics
and
power
estimates
for
the
proportion
of
eggs
fertilized
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
4
0.999
0.001
0%
89%
4
Low
4
0.998
0.002
0%
High
4
0.990
0.007
1%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
High
Control
1.00
0.99
0.98
treatment
Total
Prop­
fert
(
y
)

Figure
4.17.
Boxplot
of
the
proportion
of
eggs
fertilized
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

Fertilization
of
Eggs
on
Tiles
and
Dishes
 
The
fertilization­
success
rates
for
all
treatment
replicates
for
eggs
laid
on
tiles
during
the
EPA
14­
Day
Trenbolone
assay
were
high,
ranging
from
0.978
(
one
High
concentration
replicate)
to
1.00
(
two
Control
treatment
replicates,
one
High
and
one
Low
concentration
replicate)
(
Appendix
E,
Figure
2.3).
No
significant
differences
in
mean
fertilization­
success
rates
for
eggs
laid
on
tiles
(
Appendix
E,
Table
2.4)
among
treatments
were
detected
(
Kruskal­
Wallis,
H
=
3.16,
p
=
0.206,
df
=
2).
The
fertilization­
success
rates
for
all
treatment
replicates
for
eggs
laid
on
dishes
during
the
assay
were
high,
ranging
from
0.933
(
one
High
concentration
replicate)
to
1.00
(
several
replicates;
all
treatments)
(
Appendix
E,
Figure
2.4).
No
significant
differences
in
mean
fertilization­
success
rates
(
Appendix
E,
Table
2.4)
among
treatments
were
detected
(
Kruskal­
Wallis,
H
=
0.04,
p
=
0.981,
df
=
2).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
108
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
4.1.10
Hatchability
and
Larval
Development
Eggs
were
collected
during
the
14­
day
pre­
exposure
period
for
the
evaluation
of
hatchability.
The
mean
proportion
of
fertilized
eggs
that
hatched
was
0.98
(
sd
=
0.03)
for
Control
treatment,
0.94
(
sd
=
0.03)
for
the
Low
concentration,
and
0.97
(
sd
=
0.03)
for
the
High
concentration.
Among
the
tanks
evaluated
during
the
pre­
exposure
period,
but
that
were
not
used
in
the
14­
day
assay,
the
mean
proportion
of
fertilized
eggs
that
hatched
was
0.75
(
sd
=
0.23),
which
was
significantly
lower
than
those
for
the
other
treatments
(
Kruskal­
Wallis,
H
=
9.59,
p
=
0.022,
df
=
3).

Eggs
were
collected
during
the
EPA
14­
Day
Trenbolone
assay
for
hatchability
analysis.
Fish
in
the
High
concentration
did
not
produce
eggs
during
that
period,
therefore,
the
High
concentration
was
not
included
in
the
analysis.
The
proportion
of
fertilized
eggs
that
hatched
ranged
from
0.96
to
1.00
in
the
Control
treatment
and
was
1.00
for
all
replicates
of
the
Low
concentration
(
Figure
4.18).
There
were
no
significant
differences
between
the
two
treatments
in
the
proportion
of
eggs
that
hatched
(
Kruskal­
Wallis,
H
=
2.29,
p
=
0.131,
df
=
2).
The
achieved
power
for
this
endpoint
was
11%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
13
(
Table
4.16).

Table
4.16.
Summary
statistics
and
power
estimates
for
the
proportion
of
fertile
eggs
that
hatched
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
4
0.98
0.02
2%
11%
13
Low
4
1.00
0
0%
High
0
 
 
 
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
Control
1.00
0.99
0.98
0.97
0.96
treatment
Prop­
Hatch
(
y
)

Figure
4.18.
Boxplot
of
the
proportion
of
fertile
eggs
that
hatched
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
109
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Eggs
were
collected
during
the
pre­
exposure
period
for
the
evaluation
of
larval
development.
The
mean
proportion
of
larvae
that
developed
normally
(
i.
e.,
that
showed
no
morphological
abnormalities)
was
0.98
(
sd
=
0.02)
for
Control
treatment.
The
mean
proportion
of
normal
larvae
in
the
remaining
treatments
was
0.98
(
sd
=
0.02)
in
the
Low
concentration
and
0.99
(
sd
=
0.01)
in
the
High
concentration.
Among
the
tanks
evaluated
during
the
pre­
exposure
period,
but
that
were
not
used
in
the
21­
Day
assay,
the
mean
proportion
of
normal
larvae
was
0.98
(
sd
=
0.03).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
eggs
that
hatched
(
Kruskal­
Wallis,
H
=
0.84,
p
=
0.840,
df
=
3).

Eggs
were
collected
on
Days
7
through
10
during
the
EPA
14­
Day
Trenbolone
assay
for
the
evaluation
of
larval
development.
Fish
in
the
High
concentration
did
not
produce
eggs
during
that
period,
therefore,
the
High
concentration
was
not
included
in
the
analysis.
The
proportion
of
larvae
that
developed
normally
(
i.
e.,
that
showed
no
morphological
abnormalities)
ranged
from
0.96
to
1.00
in
the
Control
treatment
and
from
0.78
to
1.00
for
the
Low
concentration
(
Figure
4.19).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
eggs
that
hatched
(
Kruskal­
Wallis,
H
=
0.79,
p
=
0.375,
df
=
2).
The
achieved
power
for
this
endpoint
was
8%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
29
(
Table
4.17).

Table
4.17.
Summary
statistics
and
power
estimates
for
the
proportion
of
normal
larvae
for
the
EPA
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
4
0.98
0.02
2%
8%
29
Low
4
0.92
0.10
11%
High
0
 
 
 
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
Control
1.0
0.9
0.8
treatment
Prop­
Norm
Figure
4.19.
Boxplot
of
the
proportion
of
normal
larvae
by
treatment
for
the
EPA
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
110
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
4.1.11
Body
Weight
The
body
weight
of
females
used
in
the
EPA
14­
Day
Trenbolone
assay
ranged
from
1.0
g
to
2.9
g.
A
significant
difference
in
the
mean
female
body
weight
per
treatment
was
detected
(
Kruskal­
Wallis,
H
=
10.77,
p
=
0.005,
df
=
2).
The
test
indicated
that
the
mean
female
body
weight
for
females
in
the
Lowconcentration
treatment
was
less
than
those
for
the
other
two
treatments.
The
body
weight
of
males
used
in
the
EPA
14­
Day
Trenbolone
assay
ranged
from
2.4
g
to
4.7
g.
There
were
no
significant
differences
in
mean
body
weight
among
treatments
(
Kruskal­
Wallis,
H
=
3.72,
p
=
0.156,
df
=
2).

4.2
EPA
21­
Day
Assay
for
Trenbolone
The
EPA
21­
Day
Trenbolone
assay
was
conducted
from
January
20,
2003,
to
February
3,
2003
(
preexposure
test),
and
from
February
3,
2003,
to
February
24,
2003
(
exposure
test).

4.2.1
Survival
All
males
in
all
treatments
and
all
females
in
the
Control
treatment
survived
the
EPA
21­
Day
Trenbolone
assay.
One
female
in
the
High
concentration
and
one
female
in
the
Low­
concentration
treatments
died
during
the
21­
day
exposure
(
94%
survival
in
each
case).

4.2.2
Vitellogenin
Vitellogenin
concentrations
in
most
Control
treatment
females
used
during
the
EPA
21­
Day
Trenbolone
assay
ranged
from
329,100
ng/
mL
to
1,434,000
ng/
mL
(
Figure
4.20).
Two
Control­
treatment
females
had
vitellogenin
concentrations
of
3,443,500
ng/
mL
and
4,162,000
ng/
mL.
Among
females
exposed
to
the
two
trenbolone
concentrations,
vitellogenin
concentrations
ranged
from
0
ng/
mL
(
not
detected)
to
3,184,500
ng/
mL.
Significant
differences
in
the
mean
vitellogenin
concentration
per
treatment
(
Table
4.18)
were
detected
(
Kruskal­
Wallis,
H
=
21.37,
p
=
<
0.001,
df
=
2).
Vitellogenin
concentrations
in
Control­
treatment
females
and
Low­
concentration
females
were
significantly
greater
than
those
in
females
exposed
to
the
High
trenbolone
concentrations.
The
achieved
power
for
this
endpoint
was
99%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
7
(
Table
4.18).

Table
4.18.
Summary
statistics
and
power
estimates
for
female
vitellogenin
concentrations
(
ng/
mL)
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
13
1,289,831
1,164,016
90%
99%
7
Low
14
1,221,589
1,093,574
90%
High
15
113,444
270,367
238%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
13.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
111
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
4000000
3000000
2000000
1000000
0
treatment
Trenb­
VTG
Figure
4.20.
Boxplot
of
female
vitellogenin
concentration
(
ng/
mL)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
asterisks
represent
probable
outliers.

Vitellogenin
concentrations
in
Control
treatment
males
used
during
the
EPA
21­
Day
Trenbolone
assay
ranged
from
0
ng/
mL
(
not
detected)
to
8,368
ng/
mL
(
Figure
4.21).
Among
males
exposed
to
the
two
trenbolone
concentrations,
vitellogenin
concentrations
ranged
from
0
ng/
mL
(
not
detected)
to
185,700
ng/
mL.
No
significant
differences
in
the
mean
vitellogenin
concentration
per
treatment
(
Table
4.19)
were
detected
(
Kruskal­
Wallis,
H
=
0.65,
p
=
0.722,
df
=
2).
The
achieved
power
for
this
endpoint
was
8%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
140
(
Table
4.19).

Table
4.19.
Summary
statistics
and
power
estimates
for
male
vitellogenin
concentrations
(
ng/
mL)
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
2,919
3,607
124%
8%
140
Low
8
43,439
76,533
176%
High
7
6,983
11,964
171%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
7.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
112
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
200000
100000
0
treatment
Trenb­
VTG
Figure
4.21.
Boxplot
of
male
vitellogenin
concentration
(
ng/
mL)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

4.2.3
Appearance
/
Secondary
Sex
Characteristics
All
of
the
Control­
treatment
females
used
during
the
EPA
21­
Day
Trenbolone
assay
exhibited
typical
female
morphology
(
no
fat
pad,
no
tubercles,
no
vertical
banding,
ovipositor
present).
One
of
the
15
females
exposed
to
the
Low
concentration
had
a
dorsal
fat
pad.
Five
had
vertical
banding.
Among
the
16
females
exposed
to
the
High
concentration,
6
had
tubercles,
8
had
dorsal
fat
pads,
7
had
vertical
banding,
and
3
lacked
ovipositors.
Thus,
trenbolone
appeared
to
have
a
dose­
related
effect
on
female
morphology.

All
of
the
males
used
during
the
EPA
21­
Day
Trenbolone
assay
exhibited
typical
male
morphology
(
fat
pads,
tubercles,
vertical
banding,
no
ovipositor
present).

4.2.4
Gonadosomatic
Index
The
range
of
GSI
values
calculated
for
females
in
the
all
treatments
varied
from
4­
to
22­
fold
(
Figure
4.22),
and
the
overall
within­
treatment
variability,
as
indicated
by
CVs,
ranged
from
34%
 
52%;
Table
4.20).
Several
fish
had
relatively
high
GSI
values.
The
highest
value
was
obtained
for
a
female
exposed
to
the
Low
trenbolone
concentration
had
a
GSI
value
of
29.8.
The
GSI
values
for
several
other
fish
exceeded
21.0,
which
is
much
greater
than
the
typical
upper
limit
for
reproductively­
active
female
fathead
minnows
(
GSI
~
13.0).
The
total
body
weight
of
these
females
were
well
within
the
normal
range
for
the
test
animals
and
it
appears
the
high
GSI
value
resulted
from
each
fish
having
unusually
high
gonad
weights.
The
gonad
weights
of
these
fish
ranged
from
0.41
g
to
0.76
g
(
average
=
0.59
g),
greater
than
the
average
values
for
the
remaining
fish
(
0.2
g).
A
significant
difference
in
the
mean
GSI
value
per
treatment
(
Table
4.20)
was
detected
(
Kruskal­
Wallis,
H
=
8.45,
p
=
0.
015,
df
=
2).
The
test
indicated
that
the
mean
GSI
for
females
in
the
Control
treatment
was
less
than
those
for
the
other
two
treatments.
However,
the
mean
GSI
values
calculated
for
the
Low
and
High
trenbolone
concentrations
were
not
statistically
different.
The
achieved
power
for
this
endpoint
was
58%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
24
(
Table
4.20).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
113
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.20.
Summary
statistics
and
power
estimates
for
female
gonadosomatic
index
data
for
the
EPA
21­
Day
Trenbolone
assay
Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
16
10.4
5.4
52%
58%
24
Low
15
15.9
5.4
34%
High
15
15.9
7.7
48%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
15.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
High
Control
30
20
10
0
treatment
GSI
(
y
)

Figure
4.22.
Boxplot
of
female
GSI
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisks
represent
probable
outliers.

The
range
of
most
GSI
values
calculated
for
males
during
the
EPA
21­
Day
Trenbolone
assay,
was
small,
ranging
from
0.4
to
1.7
(
Figure
4.23).
Most
estimates
were
between
about
1.0
and
1.7,
which
approximates
the
typical
range
for
reproductively­
active
male
fathead
minnows.
There
were
no
significant
differences
in
mean
GSI
values
(
Table
4.21)
among
treatments
(
Kruskal­
Wallis,
H
=
0.97,
p
=
0.617,
df
=
2).
The
achieved
power
for
this
endpoint
was
19%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
42
(
Table
4.21).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
114
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.21.
Summary
statistics
and
power
estimates
for
male
gonadosomatic
index
data
for
the
EPA
21­
Day
Trenbolone
assay
Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
1.12
0.47
42%
19%
42
Low
8
1.20
0.30
25%
High
8
1.35
0.27
20%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
8.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
High
Control
1.5
1.0
0.5
treatment
GSI
(
y
)

Figure
4.23.
Boxplot
of
male
GSI
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

4.2.5
Female
Gonad
Histology
General
Ovary
Staging
 
Statistical
analysis
of
the
mean
ovarian
staging
from
12
microscopic
fields
per
female
in
the
EPA
21­
Day
Trenbolone
assay
revealed
no
significant
differences
among
treatments
(
Kruskal­
Wallis,
H
=
4.71,
p
=
0.095,
df
=
2).

Quantitative
Ovarian
Staging
 
One
hundred
cells
in
each
of
three
sections
per
female
were
examined
to
quantitatively
determine
the
developmental
stage
of
the
ovaries.
Ova
from
fish
in
all
treatments
ranged
from
Stage
1a
to
Stage
5
(
see
Methods
for
a
description
of
the
stages)
(
Figure
4.24).
Variability
within
treatments
for
each
stage
was
very
high
as
indicated
by
CVs
that
ranged
as
high
as
412%
(
Table
4.22.
Although
statistical
analyses
showed
that
there
was
a
significant
difference
among
treatments
in
the
proportion
of
cells
in
developmental
Stages
1a,
1b,
2,
and
3,
there
were
no
significant
differences
among
treatments
in
the
proportion
of
cells
in
the
developmental
Stages
4,
and
5
(
Table
4.22).
There
was
no
consistent
pattern
of
significant
difference
associated
with
trenbolone
dose.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
115
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.22.
Descriptive
statistics
of
the
proportion
of
ovarian
cells
in
each
developmental
stage
for
females
from
the
EPA
21­
Day
Trenbolone
assay
and
results
of
the
Kruskal­
Wallis
Test
(
df
=
2)
comparing
treatments
Control
(
n
=
16)
Low
(
n
=
13)
High
(
n
=
17)
Kruskal­
Wallis
Stage
Mean
Stdev
CV
Mean
Stdev
CV
Value
Stdev
CV
H
p
1a
0.079
0.034
44%
0.099
0.021
22%
0.077
0.019
25%
6.94
0.031*
1b
0.365
0.065
18%
0.327
0.067
20%
0.265
0.089
34%
8.36
0.015*
2
0.202
0.057
28%
0.178
0.029
16%
0.250
0.096
38%
8.03
0.018*
3
0.102
0.052
51%
0.122
0.031
26%
0.192
0.096
50%
14.09
0.001**
4
0.215
0.096
45%
0.240
0.068
28%
0.156
0.106
68%
5.47
0.065
5
0.003
0.008
278%
0.001
0.003
361%
0.002
0.006
412%
2.76
0.252
*
p
<
0.05
**
p
<
0.01
0.0
0.1
0.2
0.3
0.4
0.5
1a
1b
2
3
4
5
Ovarian
Stage
Proportion
of
Cells
Control
Low
High
Figure
4.24.
Frequency
histogram
showing
the
quantitative
developmental
staging
of
ovaries
for
each
treatment
of
the
EPA
21­
Day
Trenbolone
assay.
For
each
treatment,
the
columns
represent
the
grand
mean
proportion
of
cells
in
each
stage
and
the
bars
represent
the
standard
deviation.

Atretic
Follicles
 
The
mean
proportion
of
atretic
follicles
per
300
follicles
(
counted
per
fish)
ranged
from
0.007
for
females
in
the
Control
treatment
to
0.05
for
females
in
the
High
concentration
(
Figure
4.25).
There
were
significant
differences
in
the
proportions
of
atretic
follicles
among
treatments
(
Kruskal­
Wallis,
H
=
14.88,
p
=
0.001,
df
=
2).
The
proportions
of
atretic
follicles
in
the
High
concentration
were
greater
than
those
in
the
Control
treatment,
but
not
greater
than
those
in
the
Lowconcentration
treatment.
Therefore,
there
was
no
consistent
pattern
of
significant
difference
associated
with
trenbolone
dose.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
116
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
0.2
0.1
0.0
treatment
atretic_
follicles
(
y
)

Figure
4.25.
Boxplot
of
the
proportion
of
atretic
follicles
per
300
follicles
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
asterisks
represent
probable
outliers.

Corpora
Lutea
 
The
mean
proportion
of
corpora
lutea
per
300
follicles
(
counted
per
fish)
ranged
from
0.01
for
females
in
the
High
concentration
to
0.025
for
females
in
the
Low
concentration
(
Figure
4.26).
There
were
significant
differences
in
the
proportions
of
corpora
lutea
among
treatments
(
Kruskal­
Wallis,
H
=
11.43,
p
=
0.003,
df
=
2).
The
mean
proportion
of
corpora
lutea
in
the
High
concentration
was
less
than
those
in
the
Control
treatment
and
the
Low
concentration.

Low
High
Control
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
treatment
corpora_
lutea
(
y
)

Figure
4.26.
Boxplot
of
proportion
of
corpora
lutea
per
300
follicles
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
asterisks
represent
probable
outliers.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
117
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
4.2.6
Male
Gonad
Histology
Testes
Staging
by
Microscopic
Field
 
Testes
from
males
exposed
to
trenbolone
during
the
EPA
21­
Day
Trenbolone
assay
were
examined
to
determine
the
general
developmental
condition.
Males
in
all
treatments
had
well­
developed
testes
with
most
showing
Stage
4
and
Stage
5
development
(
see
Methods
for
description
of
developmental
stages).
All
of
the
96
microscopic
fields
examined
in
the
8
Control
treatment
males
showed
Stage
4
(
79
fields)
or
Stage
5
(
17
fields)
development.
All
of
the
120
microscopic
fields
examined
in
the
10
Low­
concentration
treatment
males
showed
Stage
4
(
99
fields)
or
Stage
5
(
21
fields)
development.
All
of
the
72
microscopic
fields
examined
in
the
6
High­
concentration
treatment
males
showed
Stage
4
(
48
fields)
or
Stage
5
(
24
fields)
development.
Statistical
analysis
of
the
mean
staging
from
12
microscopic
fields
per
fish
revealed
no
significant
differences
in
testicular
development
among
treatments
(
Kruskal­
Wallis,
H
=
3.59,
p
=
0.166,
df
=
2).

Quantitative
Testicular
Staging
 
One
hundred
cells
in
each
of
three
sections
per
male
were
examined
to
quantitatively
determine
the
developmental
condition
of
the
testes.
The
developmental
stage
of
the
Control
treatment
testes
ranged
from
Stage
1
to
Stage
5,
whereas
testes
in
males
from
the
Low
and
Highconcentration
treatments
showed
Stage
2a
to
Stage
5
development
(
Figure
4.27).
Variability
within
treatments
for
each
stage
was
very
high
as
indicated
by
CVs
that
ranged
as
high
as
283%
(
Table
4.23).
Although
statistical
analyses
showed
that
there
was
a
significant
difference
among
treatments
in
the
proportion
of
cells
in
developmental
Stage
2a
and
Stage
3b,
there
were
no
significant
differences
among
treatments
in
the
proportion
of
cells
in
developmental
Stages
1,
2b,
3a,
4,
and
5
(
Table
4.23).
Therefore,
there
was
no
consistent
pattern
of
significant
difference
associated
with
trenbolone
dose.

Table
4.23.
Descriptive
statistics
of
the
proportion
of
testes
cells
in
each
developmental
stage
for
males
from
the
EPA
21­
Day
Trenbolone
assay
and
results
of
the
Kruskal­
Wallis
Test
(
df
=
2)
comparing
treatments.

Control
(
n
=
8)
Low
(
n
=
10)
High
(
n
=
6)
Kruskal­
Wallis
Stage
Mean
Stdev
CV
Mean
Stdev
CV
Mean
Stdev
CV
H
p
1
0.0004
0.001
283%
0
0
 
0
0
 
2.00
0.368
2a
0.007
0.003
38%
0.004
0.005
125%
0.002
0.004
182%
6.05
0.049*
2b
0.016
0.009
54%
0.016
0.014
88%
0.014
0.016
114%
0.19
0.910
3a
0.117
0.068
58%
0.156
0.107
69%
0.102
0.077
76%
1.19
0.550
3b
0.280
0.063
22%
0.202
0.112
55%
0.114
0.065
57%
9.44
0.009**
4
0.213
0.081
38%
0.147
0.097
66%
0.115
0.053
46%
4.38
0.112
5
0.367
0.170
46%
0.475
0.252
53%
0.652
0.191
29%
4.84
0.089
*
p
<
0.05
**
p
<
0.01
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
118
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
2a
2b
3a
3b
4
5
Testicular
Stage
Proportion
of
Cells
Control
Low
High
Figure
4.27.
Frequency
histogram
showing
the
quantitative
developmental
staging
of
testes
for
each
treatment
of
the
EPA
21­
Day
Trenbolone
assay.
For
each
treatment,
the
columns
represent
the
grand
mean
proportion
of
cells
in
each
stage
and
the
bars
represent
the
standard
deviation.

Tubule
Diameter
 
The
average
diameter
of
the
seminiferous
tubules
of
males
from
the
Control
treatment
ranged
from
78.6
µ
m
to
160.3
µ
m
(
Figure
4.28).
Tubule
diameters
of
males
from
the
two
test
concentrations
ranged
from
88.1
µ
m
to
215.8
µ
m.
No
significant
differences
in
the
mean
tubule
diameter
per
treatment
(
Table
4.24)
were
detected
(
Kruskal­
Wallis,
H
=
3.33,
p
=
0.190,
df
=
2).
The
achieved
power
for
this
endpoint
was
33%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
17
(
Table
4.24).

Table
4.24.
Summary
statistics
and
power
estimates
for
male
seminiferous
tubule
diameter
data
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
114.4
28.0
24%
33%
17
Low
10
121.4
28.1
23%
High
6
147.9
37.0
25%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
6.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
119
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
200
150
100
treatment
diameter
Figure
4.28.
Boxplot
of
male
seminiferous
tubule
diameter
(
µ
m)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
asterisk
represents
a
probable
outlier.

Observations
 
One
male
in
the
Control
treatment
showed
early
stage
cells
in
the
lumen
of
some
tubules.
No
interstitial
Sertoli
cell
proliferation
and
no
Leydig
cell
proliferation
was
observed
for
any
treatment.
No
testicular
atrophy
was
recorded
and
no
ovatestes
were
observed
for
any
treatment.

4.2.7
Plasma
Steroid
Concentrations
Estradiol
 
Estradiol
concentrations
in
Control­
treatment
females
used
during
the
EPA
21­
Day
Trenbolone
assay
ranged
from
357
pg/
mL
to
5,112
pg/
mL
(
Figure
4.29).
Among
females
exposed
to
the
two
trenbolone
concentrations,
estradiol
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
4,567
pg/
mL.
A
significant
difference
in
the
mean
estradiol
concentration
per
treatment
(
Table
4.25)
was
detected
(
Kruskal­
Wallis,
H
=
11.62,
p
=
0.003,
df
=
2).
The
mean
estradiol
concentration
in
females
from
the
High
concentration
was
less
than
that
in
females
from
the
Low
concentration.
The
achieved
power
for
this
endpoint
was
39%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
22
(
Table
4.25).

Table
4.25.
Summary
statistics
and
power
estimates
for
female
estradiol
concentrations
(
pg/
mL)
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
12
1,361
1,380
101%
39%
22
Low
9
1,696
1,553
92%
High
15
396
393
99%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
9.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
120
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
5000
4000
3000
2000
1000
0
treatment
Estradiol
(
y
)

Figure
4.29.
Boxplot
of
female
estradiol
concentration
(
pg/
mL)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisks
represent
probable
outliers.

Estradiol
concentrations
in
Control
treatment
males
used
during
the
EPA
21­
Day
Trenbolone
assay
ranged
from
0
pg/
mL
(
not
detected)
to
645
pg/
mL
(
Figure
4.30).
Among
males
exposed
to
the
two
trenbolone
concentrations,
estradiol
concentrations
ranged
from
153
pg/
mL
to
732
pg/
mL.
No
significant
differences
in
the
mean
estradiol
concentration
per
treatment
(
Table
4.26)
were
detected
(
Kruskal­
Wallis,
H
=
4.50,
p
=
0.105,
df
=
2).
The
achieved
power
for
this
endpoint
was
22%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
16
(
Table
4.26).

Table
4.26.
Summary
statistics
and
power
estimates
for
male
estradiol
concentrations
(
pg/
mL)
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
6
323
266
82%
22%
16
Low
7
299
121
40%
High
4
548
188
34%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
121
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
800
700
600
500
400
300
200
100
0
treatment
Estradiol
(
y
)

Figure
4.30.
Boxplot
of
male
estradiol
concentration
(
pg/
mL)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

Testosterone
 
Testosterone
concentrations
in
Control­
treatment
females
used
during
the
EPA
21­
Day
Trenbolone
assay
ranged
from
464
pg/
mL
to
4,534
pg/
mL
(
Figure
4.31).
Among
females
exposed
to
the
two
trenbolone
concentrations,
testosterone
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
667
pg/
mL.
Significant
differences
in
the
mean
testosterone
concentration
per
treatment
(
Table
4.27)
were
detected
(
Kruskal­
Wallis,
H
=
15.24,
p
<
0.001,
df
=
2).
The
mean
testosterone
concentration
in
females
from
the
Low
concentration
was
less
than
that
of
females
from
the
High
concentration
and
the
Control
treatment.
The
achieved
power
for
this
endpoint
was
60%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
8
(
Table
4.27).

Table
4.27.
Summary
statistics
and
power
estimates
for
female
testosterone
concentrations
(
pg/
mL)
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
1,332
1,330
100%
60%
8
Low
5
180
126
70%
High
12
475
137
29%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
5.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
122
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
5000
4000
3000
2000
1000
0
treatment
Testosterone
(
y
)

Figure
4.31.
Boxplot
of
female
testosterone
concentration
(
pg/
mL)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.

Testosterone
concentrations
in
Control
treatment
males
used
during
the
EPA
21­
Day
Trenbolone
assay
ranged
from
893
pg/
mL
to
2,592
pg/
mL
(
Figure
4.32).
Among
males
exposed
to
the
two
trenbolone
concentrations,
testosterone
concentrations
ranged
from
481
pg/
mL
to
4,709
pg/
mL.
No
significant
differences
in
the
mean
testosterone
concentration
per
treatment
(
Table
4.28)
were
detected
(
Kruskal­
Wallis,
H
=
2.89,
p
=
0.236,
df
=
2).
The
achieved
power
for
this
endpoint
was
12%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
44
(
Table
4.28).

Table
4.28.
Summary
statistics
and
power
estimates
for
male
testosterone
concentrations
(
pg/
mL)
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
1,414
600
42%
12%
44
Low
8
2,656
1,459
55%
High
5
2,528
1,766
70%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
5.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
123
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
5000
4000
3000
2000
1000
0
treatment
Testosterone
(
y
)

Figure
4.32.
Boxplot
of
male
testosterone
concentration
(
pg/
mL)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

11­
ketotestosterone
 
11­
ketotesosterone
was
not
detected
in
females
from
the
Control
treatment
(
2
individuals)
or
from
the
Low
concentration
(
three
individuals).
No
females
from
the
High
concentration
were
available
for
11­
ketotestosteron
analyses.

11­
ketotesosterone
concentrations
in
most
Control
treatment
males
used
during
the
EPA
21­
Day
Trenbolone
assay
ranged
from
0
pg/
mL
(
not
detected)
to
13,938
pg/
mL
(
Figure
4.33).
One
male
from
the
Control
treatment
had
a
11­
ketotestosterone
concentration
of
56,704
pg/
mL.
Among
males
exposed
to
the
two
trenbolone
concentrations,
11­
ketotesosterone
concentrations
ranged
from
481
pg/
mL
to
40,953
pg/
mL.
No
significant
differences
in
the
mean
11­
ketotesosterone
concentration
per
treatment
(
Table
4.29)
were
detected
(
Kruskal­
Wallis,
H
=
4.25,
p
=
0.120,
df
=
2).
The
achieved
power
for
this
endpoint
was
26%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
26
(
Table
4.29).

Table
4.29.
Summary
statistics
and
power
estimates
for
male
11­
ketotesosterone
concentrations
(
pg/
mL)
for
the
EPA
21­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
12,925
18,174
141%
26%
26
Low
8
26,128
14,008
54%
High
7
14,128
11,455
81%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
7.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
124
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
60000
50000
40000
30000
20000
10000
0
treatment
11­
keto
(
y
)

Figure
4.33.
Boxplot
of
male
11­
ketotesosterone
concentration
(
pg/
mL)
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

4.2.8
Fecundity
Total
Fecundity
 
Two
pre­
exposure
evaluations
of
total
egg
production
were
performed.
One
examined
the
total
egg
production
at
14
days,
which
is
within
the
protocol
guidelines;
the
other
evaluated
egg
production
at
day
7
of
the
pre­
exposure
assay
to
determine
whether
or
not
the
shorter
pre­
exposure
period
would
be
sufficient
to
evaluate
the
potential
reproductive
success
of
the
test
fish.
Total
14­
day
counts
among
the
tanks
that
were
eventually
used
for
the
three
treatments
in
the
exposure
assay
(
individual
tank
values
summed
for
each
treatment)
ranged
from
about
12,000
eggs
to
14,000
eggs
(
Figure
4.34).
No
significant
differences
in
the
mean
14­
day
egg
production
among
the
groups
of
replicates
that
would
be
used
in
the
trenbolone
exposure
assay
were
detected
(
1­
way
ANOVA,
F
=
0.11,
p
=
0.953,
df
=
3,
13).
Production
at
Day
7
of
the
pre­
exposure
test
among
the
12
replicates
that
were
eventually
used
in
the
exposure
assay
ranged
from
503
eggs
to
1,923
eggs.
No
significant
differences
in
the
mean
7­
day
egg
production
among
the
groups
of
replicates
that
would
be
used
in
the
trenbolone
exposure
assay
were
detected
(
1­
way
ANOVA,
F
=
0.11,
p
=
0.955,
df
=
3,
13).
No
significant
differences
were
detected
in
the
mean
number
of
eggs
laid
per
day
per
replicate
group
at
7
days
versus
14
days
(
Two­
Sample
t­
Test,
t
=
 
0.85,
p
=
0.401,
df
=
25).
However,
daily
within­
treatment
variation
(
indicated
by
fluctuating
coefficient
of
variation
values)
was
high
through
Day
9.
Therefore,
variability
during
a
7­
day
preexposure
assay
is
likely
to
be
much
greater
than
that
during
longer
pre­
exposure
assay,
which
reduces
the
likelihood
of
reliably
choosing
replicates
with
successful
reproduction
histories
for
use
in
the
exposure
assay.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
125
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
0
5,000
10,000
15,000
20,000
25,000
30,000
­
14
­
12
­
10
­
8
­
6
­
4
­
2
1
3
5
7
9
11
13
15
17
19
21
Day
Cumulative
Number
of
Eggs
Control
Low
High
Figure
4.34.
Total
egg
production
per
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
vertical
line
at
Day
0
denotes
the
start
of
the
exposure
period.

During
the
EPA
21­
Day
Trenbolone
assay,
total
counts
in
the
Control
treatment
were
reasonably
consistent
among
replicates,
varying
from
2,741
eggs
to
4,334
eggs
(
Figure
4.35).
Variability
in
total
egg
production
among
Low
concentration
replicates
was
somewhat
greater,
ranging
from
1,873
eggs
to
4,017
eggs.
Total
counts
among
the
High
concentration
replicates
varied
about
five­
fold,
ranging
from
215
eggs
to
1,095
eggs.
Statistical
analysis
of
square­
root
transformed
egg
counts
showed
significant
amongtreatment
differences
(
1­
way
ANOVA,
F
=
23.14,
p
<
0.001,
df
=
2,
10)
in
mean
cumulative
numbers
of
eggs
produced
(
Table
4.30).
Tukey's
pairwise
comparison
identified
significant
differences
in
mean
egg
production
between
the
High
trenbolone
concentration
and
the
Control
treatment,
and
also
between
the
High
and
Low
concentrations.
The
Low
concentration
and
the
Control
treatment
fecundity
values
were
not
statistically
different.
Daily
within­
treatment
variation
(
indicated
by
fluctuating
coefficient
of
variation
values)
was
much
reduced
after
Days
8
 
10,
except
for
the
High
concentration.
Egg
production
in
the
other
four
High
concentration
replicates
was
sporadic.
Egg
production
stopped
after
Day
3
in
two
replicates,
after
Day
7
in
a
third
replicate,
and
occurred
only
on
Day
9
in
the
fourth
replicate.
The
achieved
power
for
this
assay
was
100%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
3
(
Table
4.30).

Table
4.30.
Summary
statistics
and
power
estimates
for
fecundity
data
for
the
EPA
21­
Day
Trenbolone
assay
Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
4
3672
795
22%
100%
3
Low
4
2818
914
32%
High
4
590
426.5
72%
1
Calculated
from
square­
root
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
square­
root
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
126
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
0
2
4
6
8
10
12
14
16
18
20
22
Day
Cumulative
Number
of
Eggs
Control
Low
High
Figure
4.35.
Total
egg
production
by
replicate
per
treatment
for
the
EPA
21­
Day
Trenbolone
assay.

Fecundity
per
Female
Reproductive
Day
 
During
the
14­
Day
pre­
exposure
evaluation,
the
mean
number
of
eggs
produced
per
female
reproductive
day
ranged
from
55.6
eggs/
day
for
the
tanks
that
would
be
used
for
the
Control
treatment
to
58.5
eggs/
day
for
the
tanks
that
would
be
used
for
the
High
concentration
during
the
21­
day
exposure
assay.
There
were
no
significant
differences
among
treatments
in
the
numbers
of
eggs
produced
per
reproductive
day
during
the
pre­
exposure
period
(
Kruskal­
Wallis,
H
=
0.99,
p
=
0.804,
df
=
3).

During
the
EPA
21­
Day
Trenbolone
assay,
the
maximum
number
of
female
reproductive
days
was
achieved
for
the
Control
treatment,
whereas
80
and
79.3
female
reproductive
days
were
achieved
in
the
Low
concentration
and
High
concentration,
respectively
(
Table
4.31).
The
number
of
eggs
produced
per
female
reproductive
day
varied
from
32.6
eggs
to
51.6
eggs
in
the
Control
treatment
and
from
27.5
to
47.8
in
the
Low
concentration
(
Figure
4.36).
For
the
High
concentration,
the
number
of
eggs
produced
per
female
reproductive
day
ranged
from
2.6
eggs
to
13.0
eggs.
A
significant
difference
in
the
mean
number
of
eggs
produced
per
female
reproductive
day
among
treatments
was
detected
(
Kruskal­
Wallis,
H
=
8.38,
p
=
0.015,
df
=
2).
The
mean
number
of
eggs
produced
per
female
reproductive
day
was
significantly
smaller
in
the
High
concentration
than
in
either
the
Control
treatment
or
the
Low
concentration.
The
mean
number
of
eggs
produced
in
the
Low
concentration
did
not
differ
from
that
produced
in
the
Control
treatment.
The
achieved
power
for
this
assay
was
100%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
3
(
Table
4.31).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
127
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.31.
Summary
statistics
and
power
estimates
for
fecundity
per
female
reproductive
day
for
the
EPA
21­
Day
Trenbolone
assay
Level
Mean
Number
of
Reproductive
Days
1
N
Mean
Stdev
CV
Achieved
Power
2
Sample
Size
Required
3
Control
84.0
4
43.7
9.5
22%
100%
3
Low
80.0
4
34.8
9.3
27%
High
79.3
4
7.2
4.8
67%
1
Maximum
number
=
84.
2
Calculated
from
natural
log
transformed
data;
with
sample
size
=
4.
3
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Low
High
Control
50
40
30
20
10
0
treatment
eggs/
ReproDay
Figure
4.36.
Boxplot
of
the
number
of
eggs
produced
per
female
reproductive
day
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

Eggs
on
Tiles/
Dishes
 
The
mean
number
of
eggs
laid
on
the
tiles
during
the
14­
day
pre­
exposure
assay
varied
from
2,101
eggs
for
the
tanks
that
would
be
used
for
the
Control
treatment
to
2,561
eggs
for
the
tanks
that
would
be
used
for
the
High­
concentration
treatment.
The
mean
number
of
eggs
laid
on
the
dishes
during
the
14­
day
pre­
exposure
assay
varied
from
733
eggs
for
the
tanks
that
would
be
used
for
the
High­
concentration
treatment
to
1,005
eggs
for
the
tanks
that
would
be
used
for
the
Control
treatment.
Because
of
the
variability
in
the
total
number
of
eggs
laid
per
treatment,
the
proportional
difference
in
the
number
of
eggs
on
dishes
versus
those
on
tiles
[
1 (#
eggs
on
dishes
÷
#
eggs
on
tiles)]
was
calculated.
There
were
no
significant
differences
in
the
mean
proportional
difference
among
treatments
during
the
14­
day
pre­
exposure
assay
(
Kruskal­
Wallis,
H
=
1.39,
p
=
0.707,
df
=
3).

The
mean
number
of
eggs
laid
on
the
tiles
among
the
treatments
during
the
EPA
21­
Day
Trenbolone
assay
varied
from
500
eggs
for
the
High
concentration
to
3,384
eggs
for
the
Control
treatment
(
Appendix
E,
Table
2.7).
The
number
of
eggs
on
dishes
ranged
from
90
eggs
for
the
High
concentration
to
288
eggs
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
128
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
for
the
Control
treatment.
The
proportional
difference
ranged
from
 
0.10
(
one
High
concentration
replicate
had
more
eggs
on
the
dish
than
on
the
tile)
to
0.96
(
one
Control­
treatment
replicate)
(
Appendix
E,
Figure
2.6).
There
were
no
significant
differences
in
this
mean
proportional
difference
among
treatments
(
Kruskal­
Wallis,
H
=
2.46,
p
=
0.292,
df
=
2).

4.2.9
Fertilization
Success
Total
Fertilization
 
Eggs
were
collected
during
the
14­
day
pre­
exposure
period
for
the
evaluation
of
fertilization­
success
rate.
The
mean
proportion
of
eggs
fertilized
in
the
Control
treatment
was
0.981
[
standard
deviation
(
sd)
=
0.028],
0.997
(
sd
=
0.002)
in
the
Low
concentration,
and
0.985
(
sd
=
0.023)
in
the
High
concentration.
The
mean
proportion
of
eggs
fertilized
in
the
replicates
that
were
not
used
in
the
21­
day
validation
assay
was
0.977
(
sd
=
0.041).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
eggs
that
were
fertilized
(
Kruskal­
Wallis,
H
=
2.42,
p
=
0.490,
df
=
3).

The
total
(
tiles
+
dishes)
fertilization­
success
rates
for
all
treatment
replicates
during
the
EPA
21­
Day
Trenbolone
assay
were
high,
ranging
from
0.991
(
one
Low­
concentration
replicate)
to
1.00
(
one
Highconcentration
replicate)
(
Figure
4.37).
No
significant
differences
in
mean
fertilization­
success
rates
(
Table
4.32)
among
treatments
were
detected
(
Kruskal­
Wallis,
H
=
0.04,
p
=
0.981,
df
=
2).
The
achieved
power
for
this
assay
was
5%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
764
(
Table
4.32).

Table
4.32.
Summary
statistics
and
power
estimates
for
the
proportion
of
eggs
fertilized
for
the
EPA
21­
Day
Trenbolone
assay
Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
4
0.997
0.001
0.1%
5%
764
Low
4
0.996
0.004
0.4%
High
4
0.997
0.004
0.4%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
4.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
129
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Low
High
Control
1.000
0.999
0.998
0.997
0.996
0.995
0.994
0.993
0.992
0.991
treatment
Total
Prop­
fert
Figure
4.37.
Boxplot
of
the
proportion
of
eggs
fertilized
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

Fertilization
of
Eggs
on
Tiles
and
Dishes
 
During
the
14­
Day
pre­
validation
assay,
there
were
no
significant
differences
in
the
fertilization­
success
rates
among
treatments
for
eggs
laid
in
tiles
(
Kruskal­
Wallis,
H
=
4.51,
p
=
0.211,
df
=
3)
or
on
dishes
(
Kruskal­
Wallis,
H
=
2.67,
p
=
0.446,
df
=
3).

The
fertilization­
success
rates
for
all
treatment
replicates
for
eggs
laid
on
tiles
during
the
EPA
21­
Day
Trenbolone
assay
were
high,
ranging
from
0.990
(
one
Low­
concentration
replicate)
to
1.00
(
two
Highconcentration
replicates)
(
Appendix
E,
Figure
2.7).
No
significant
differences
in
mean
fertilizationsuccess
rates
(
Appendix
E,
Table
2.8)
among
treatments
were
detected
(
Kruskal­
Wallis,
H
=
0.35,
p
=
0.841,
df
=
2).
The
fertilization­
success
rates
for
all
treatment
replicates
for
eggs
laid
on
dishes
during
the
assay
were
high,
ranging
from
0.958
(
one
Control­
treatment
replicate)
to
1.00
(
several
replicates;
including
all
treatments)
(
Appendix
E,
Figure
2.8).
No
significant
differences
in
mean
fertilizationsuccess
rates
(
Appendix
E,
Table
2.8)
among
treatments
were
detected
(
Kruskal­
Wallis,
H
=
0.62,
p
=
0.733,
df
=
2).

4.2.10
Hatchability
and
Larval
Development
Eggs
were
collected
during
the
pre­
exposure
period
for
the
evaluation
of
hatchability.
The
mean
proportion
of
fertilized
eggs
that
hatched
was
0.97
(
sd
=
0.02)
for
Control
treatment,
0.98
(
sd
=
0.02)
for
the
Low
concentration,
and
0.94
(
sd
=
0.06)
for
the
High
concentration.
Among
the
tanks
evaluated
during
the
pre­
exposure
period,
but
that
were
not
used
in
the
21­
Day
assay,
the
mean
proportion
of
fertilized
eggs
that
hatched
was
0.92
(
sd
=
0.03).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
eggs
that
hatched
(
Kruskal­
Wallis,
H
=
6.91,
p
=
0.075,
df
=
3).

Eggs
were
collected
during
the
EPA
21­
Day
Trenbolone
assay
for
the
evaluation
of
hatchability.
The
proportion
of
fertilized
eggs
that
hatched
ranged
from
0.26
to
1.00
in
the
Control
treatment
and
from
0.82
to
1.00
for
the
two
test
concentrations
(
Figure
4.38).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
eggs
that
hatched
(
Kruskal­
Wallis,
H
=
0.07,
p
=
0.964,
df
=
2).
The
achieved
power
for
this
endpoint
was
5%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
185
(
Table
4.33).
One
of
the
Control
treatment
replicates
had
a
very
low
proportion
of
the
eggs
hatch
(
0.26)
during
the
second
week
of
the
hatch
evaluation.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
130
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Because
the
data
sheet
for
that
replicate
indicated
that
the
eggs
may
have
been
abnormal
at
the
initiation
of
the
evaluation,
the
analyses
were
rerun
excluding
those
data.
The
resulting
mean
proportion
hatch
for
the
Control
treatment
was
0.95
(
sd
=
0.08),
however
there
still
were
no
significant
differences
among
treatments
in
the
proportion
of
eggs
that
hatched
(
Kruskal­
Wallis,
H
=
0.59,
p
=
0.746,
df
=
2).

Table
4.33.
Summary
statistics
and
power
estimates
for
the
proportion
of
fertile
eggs
that
hatched
for
the
EPA
21­
Day
Trenbolone
assay
Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
0.86
0.26
30%
5%
185
Low
8
0.94
0.06
7%
High
2
0.94
0.08
9%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
2.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
High
Control
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
treatment
Prop­
Hatch
(
y
)

Figure
4.38.
Boxplot
of
the
proportion
of
fertile
eggs
that
hatched
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
asterisks
represent
probable
outliers.

Eggs
were
collected
during
the
pre­
exposure
period
for
the
evaluation
of
larval
development.
The
proportion
of
larvae
that
developed
normally
(
i.
e.,
that
showed
no
morphological
abnormalities)
was
1.00
for
all
replicates
of
the
Control
treatment.
The
mean
proportion
of
normal
larvae
in
the
remaining
treatments
was
0.96
(
sd
=
0.01)
in
the
Low
concentration
and
0.95
(
sd
=
0.07)
in
the
High
concentration.
Among
the
tanks
evaluated
during
the
pre­
exposure
period,
but
that
were
not
used
in
the
21­
Day
assay,
the
mean
proportion
of
normal
larvae
was
0.93
(
sd
=
0.05).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
normal
larvae
(
Kruskal­
Wallis,
H
=
6.26,
p
=
0.099,
df
=
3).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
131
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Eggs
were
collected
on
Days
7
 
10,
14
 
16,
and
21
during
the
EPA
21­
Day
Trenbolone
assay
for
the
evaluation
of
larval
development.
The
proportion
of
larvae
that
developed
normally
(
i.
e.,
that
showed
no
morphological
abnormalities)
ranged
from
0.08
to
1.00
in
the
Control
treatment
and
from
0.98
to
1.00
for
the
two
test
concentrations
(
Figure
4.39).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
normal
larvae
(
Kruskal­
Wallis,
H
=
2.77,
p
=
0.250,
df
=
2).
The
achieved
power
for
this
endpoint
was
9%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
16
(
Table
4.34).
As
for
the
hatchability
analyses,
the
larval
development
analysis
was
rerun
excluding
the
one
Control
treatment
replicate
that
may
have
had
abnormal
eggs
at
the
initiation
of
the
evaluation.
The
resulting
mean
proportion
of
larvae
that
developed
normally
for
the
Control
treatment
was
0.98
(
sd
=
0.22),
however
there
were
no
significant
differences
among
treatments
in
the
proportion
of
normal
larvae
(
Kruskal­
Wallis,
H
=
1.89,
p
=
0.390,
df
=
2).

Table
4.34.
Summary
statistics
and
power
estimates
for
the
proportion
of
normal
larvae
for
the
EPA
21­
Day
Trenbolone
assay
Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
0.78
0.35
45%
9%
16
Low
8
0.997
0.01
1%
High
2
0.99
0.02
2%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
2.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.

Low
High
Control
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
treatment
Prop­
Norm
Figure
4.39.
Boxplot
of
the
proportion
of
normal
larvae
by
treatment
for
the
EPA
21­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
132
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
4.2.11
Body
Weight
The
body
weight
of
females
used
in
the
EPA
21­
Day
Trenbolone
assay
ranged
from
1.1
g
to
3.7
g.
A
significant
difference
in
the
mean
female
body
weight
per
treatment
was
detected
(
Kruskal­
Wallis,
H
=
16.34,
p
<
0.001,
df
=
2).
The
test
indicated
that
the
mean
female
body
weight
for
females
in
the
Highconcentration
treatment
was
greater
than
those
for
the
other
two
treatments.
The
body
weight
of
males
used
in
the
EPA
21­
Day
Trenbolone
assay
ranged
from
2.6
g
to
6.0
g.
There
were
no
significant
differences
in
mean
body
weight
among
treatments
(
Kruskal­
Wallis,
H
=
0.97,
p
=
0.616,
df
=
2).

4.3
Non­
spawning
Adult
14­
Day
Assay
for
Trenbolone
The
Non­
spawning
Adult
14­
Day
Trenbolone
assay
was
conducted
from
February
10,
2003
to
February
24,
2003.

4.3.1
Survival
Total
survival
in
the
Control
and
all
three
test
concentrations
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
was
100%.

4.3.2
Vitellogenin
Vitellogenin
concentrations
in
Control
treatment
females
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
1,409,500
ng/
mL
to
4,440,000
ng/
mL
(
Figure
4.40).
Among
most
females
exposed
to
the
three
trenbolone
concentrations,
vitellogenin
concentrations
ranged
from
0
ng/
mL
(
not
detected)
to
915,750
ng/
mL.
Two
females
exposed
to
the
Low
trenbolone
concentration
had
vitellogenin
concentrations
of
1,549,500
ng/
mL
and
3,821,000
ng/
mL.
Significant
differences
in
the
mean
vitellogenin
concentration
among
treatments
(
Table
4.35)
were
detected
(
Kruskal­
Wallis,
H
=
23.23,
p
=
<
0.001,
df
=
2).
Mean
vitellogenin
concentrations
among
Control
treatment
females
were
significantly
greater
than
the
mean
concentrations
for
females
exposed
to
the
all
three
trenbolone
concentrations.
The
achieved
power
for
this
endpoint
was
96%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
6
(
Table
4.35).

Table
4.35.
Summary
statistics
and
power
estimates
for
female
vitellogenin
concentrations
(
ng/
mL)
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
2,810,125
1,019,160
36%
96%
6
Low
10
907,905
1,123,918
124%
Medium
9
125,801
241,178
192%
High
9
112,255
145,135
129%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
8.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
133
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
4000000
3000000
2000000
1000000
0
treatment
Trenb­
VTG
Figure
4.40.
Boxplot
of
female
vitellogenin
concentration
(
ng/
mL)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
asterisks
represent
probable
outliers.

Vitellogenin
concentrations
in
Control­
treatment
males
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
0
ng/
mL
(
not
detected)
to
7,672
ng/
mL
(
Figure
4.41).
Vitellogenin
was
not
detected
above
the
MDL
for
four
of
the
eight
males
in
the
Control
treatment.
Among
most
males
exposed
to
the
three
trenbolone
concentrations,
vitellogenin
concentrations
ranged
from
0
ng/
mL
(
not
detected)
to
1,020
ng/
mL.
Vitellogenin
concentrations
were
not
detected
above
the
MDL
for
18
of
the
20
males
exposed
to
the
Low
and
Medium
trenbolone
concentrations.
One
male
exposed
to
the
High
trenbolone
concentration
had
a
vitellogenin
concentration
of
11,390
ng/
mL.
Significant
differences
in
the
mean
vitellogenin
concentration
per
treatment
(
Table
4.36)
were
detected
(
Kruskal­
Wallis,
H
=
16.81,
p
=
0.001,
df
=
2).
Vitellogenin
concentrations
in
males
form
the
Control
treatment
and
the
High
concentration
ranked
higher
than
those
from
males
in
the
Low
and
Medium
concentrations.
The
achieved
power
for
this
endpoint
was
41%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
20
(
Table
4.36).

Table
4.36.
Summary
statistics
and
power
estimates
for
male
vitellogenin
concentrations
(
ng/
mL)
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
9
1,337
2,561
192%
41%
20
Low
10
102
323
317%
Medium
10
30
96
320%
High
10
1,449
3,500
242%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
9.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
134
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
10000
5000
0
treatment
Trenb­
VTG
(
y
)

Figure
4.41.
Boxplot
of
male
vitellogenin
concentration
(
ng/
mL)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
asterisks
represent
probable
outliers.

4.3.3
Appearance
/
Secondary
Sex
Characteristics
All
of
the
Control­
treatment
and
Low­
concentration
females
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
showed
normal
female
morphology.
Six
of
10
females
from
the
High
concentration
had
tubercles
typical
of
males.
One
female
each
from
the
Medium
and
High
concentrations
had
dorsal
fat
pads.
Four
females
from
the
High
concentration
had
vertical
banding.
Therefore
trenbolone
likely
had
a
dose­
related
effect
on
female
morphology.

Morphological
development
among
males
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
varied
among
treatments
(
Figure
4.42).
Five
males,
three
from
the
Control
treatment
and
two
from
the
Low
concentration,
had
female
body
shapes.
Three
males
for
the
same
two
treatments
lacked
tubercles.
Two
males
from
the
Control
treatment
lacked
fat
pads.
Twelve
of
the
40
males
used
during
the
assay
lacked
vertical
banding.
There
was
no
consistent
dose­
related
pattern
to
these
variations
in
morphology.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
135
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
Tubercles
Abs
Tubercles
Prs
Fat
Pad
NP
Fat
Pad
P
Fat
Pad
Pro
Vert
Banding
NP
Vert
Banding
P
Vert
Banding
Pro
Proportion
Control
Low
Mid
High
Figure
4.42.
Secondary
sex
characteristics
of
males
used
during
the
flutamide
Non­
spawning
Adult
14­
Day
assay.

4.3.4
Gonadosomatic
Index
The
range
of
GSI
values
calculated
for
females
in
the
all
treatments
varied
from
two­
to
ten­
fold
(
Figure
4.43),
and
the
overall
within­
treatment
variability
was
moderate
(
CVs
=
22%
 
50%;
Table
4.37).
The
highest
female
GSI
value
was
24.9
(
one
fish
in
the
High
concentration),
but
several
fish
had
GSI
values
>
20.
There
were
no
significant
differences
in
mean
GSI
values
(
Table
4.37)
among
treatments
(
Kruskal­
Wallis,
H
=
3.05,
p
=
0.384,
df
=
3).
The
achieved
power
for
this
endpoint
was
18%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
55
(
Table
4.37).

Table
4.37.
Summary
statistics
and
power
estimates
for
female
gonadosomatic
index
data
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
13.2
5.0
38%
18%
55
Low
10
13.8
3.5
25%
Medium
10
16.4
3.6
22%
High
10
14.8
7.4
50%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
10.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
136
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
25
20
15
10
5
0
treatment
GSI
(
y
)

Figure
4.43.
Boxplot
of
female
GSI
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

The
range
of
most
GSI
values
calculated
for
males
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay,
varying
from
0.8
to
2.0
(
Figure
4.44),
which
approximates
the
typical
range
for
reproductivelyactive
male
fathead
minnows.
Within­
treatment
variability
was
moderate,
as
indicated
by
CVs
that
ranged
from
24%
to
59%;
Table
4.38).
The
highest
and
lowest
male
GSI
values
were
2.2
 
2.3
(
for
one
fish
each
in
the
Low
and
High
concentrations)
and
0.2
 
0.3
(
one
fish
each
in
the
Low
and
Medium
concentrations),
respectively.
However,
there
were
no
significant
differences
among
treatments
in
mean
GSI
values
(
Table
4.38)
among
treatments
(
Kruskal­
Wallis,
H
=
4.60,
p
=
0.203,
df
=
3).
The
achieved
power
for
this
endpoint
was
16%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
64
(
Table
4.38).

Table
4.38.
Summary
statistics
and
power
estimates
for
male
gonadosomatic
index
data
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
1.21
0.41
34%
16%
64
Low
10
1.11
0.66
59%
Medium
10
1.34
0.56
42%
High
10
1.60
0.38
24%
1
Calculated
from
arcsine
square­
root
transformed
data;
with
sample
size
=
10.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
arcsine
square­
root
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
137
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
2
1
0
treatment
GSI
Figure
4.44.
Boxplot
of
male
GSI
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

4.3.5
Female
Gonad
Histology
General
Ovary
Staging
 
Statistical
analysis
of
the
mean
ovarian
staging
from
12
microscopic
fields
per
female
in
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
revealed
a
significant
difference
among
treatments
(
Kruskal­
Wallis,
H
=
9.13,
p
=
0.028,
df
=
2).
The
mean
ovarian
stage
of
females
in
the
Control
treatment
was
greater
than
that
of
females
in
the
other
three
treatments.

Quantitative
Ovarian
Staging
 
One
hundred
cells
in
each
of
three
sections
per
female
were
examined
to
quantitatively
determine
the
developmental
stage
of
the
ovaries.
Ova
from
fish
in
the
Control
treatment,
Medium
concentration,
and
High
concentration
ranged
from
Stage
1a
to
Stage
5
(
see
Methods
for
a
description
of
the
stages),
whereas
ova
from
females
from
the
Low­
concentration
treatment
showed
Stage
1a
to
Stage
4
development
(
Figure
4.45).
Variability
within
treatments
for
each
stage
was
very
high
as
indicated
by
CVs
that
ranged
as
high
as
316%
(
Table
4.39).
Although
statistical
analyses
showed
that
there
was
a
significant
difference
among
treatments
in
the
proportion
of
cells
in
developmental
Stages
3
and
5,
there
were
no
significant
differences
among
treatments
in
the
proportion
of
cells
in
the
developmental
Stages
1a,
1b,
2,
and
4
(
Table
4.39).
Therefore,
there
was
no
consistent
pattern
of
significant
difference
associated
with
trenbolone
dose.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
138
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.39.
Descriptive
statistics
of
the
proportion
of
ovarian
cells
in
each
developmental
stage
for
females
from
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
and
results
of
the
Kruskal­
Wallis
Test
(
df
=
2)
comparing
treatments.

Control
(
n
=
10)
Low
(
n
=
10)
Medium
(
n
=
10)
High
(
n
=
10)
Kruskal­
Wallis
Stage
Mean
Stdev
CV
Mean
Stdev
CV
Mean
Stdev
CV
Mean
Stdev
CV
H
p
1a
0.107
0.034
32%
0.086
0.029
33%
0.082
0.041
51%
0.069
0.047
68%
5.65
0.130
1b
0.370
0.085
23%
0.325
0.080
25%
0.257
0.080
31%
0.291
0.128
44%
7.70
0.053
2
0.135
0.048
35%
0.137
0.046
33%
0.108
0.043
40%
0.151
0.052
34%
3.31
0.346
3
0.096
0.047
48%
0.202
0.093
46%
0.236
0.072
31%
0.238
0.088
37%
15.03
0.002*
4
0.211
0.154
73%
0.233
0.116
50%
0.251
0.131
52%
0.197
0.134
68%
1.41
0.704
5
0.034
0.052
151%
0.000
0.000
 
0.002
0.006
316%
0.002
0.004
179%
9.28
0.026**
*
p
<
0.01
**
p
<
0.05
0.0
0.1
0.2
0.3
0.4
0.5
1a
1b
2
3
4
5
Ovarian
Stage
Proportion
of
Cells
Control
Low
Medium
High
Figure
4.45.
Frequency
histogram
showing
the
quantitative
developmental
staging
of
ovaries
for
each
treatment
of
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
For
each
treatment,
the
columns
represent
the
grand
mean
proportion
of
cells
in
each
stage
and
the
bars
represent
the
standard
deviation.

Atretic
Follicles
 
The
mean
proportion
of
atretic
follicles
per
300
follicles
(
counted
per
fish)
ranged
from
0.01
follicles
for
females
in
the
Low
concentration
to
0.05
follicles
for
females
in
the
High
and
Medium
concentrations
(
Figure
4.46).
There
were
significant
differences
in
the
proportions
of
atretic
follicles
among
treatments
(
Kruskal­
Wallis,
H
=
10.32,
p
=
0.016,
df
=
2).
The
value
for
females
in
the
Low
concentration
was
lower
than
those
for
females
in
the
other
three
treatments.
However,
there
was
no
consistent
pattern
of
significant
difference
associated
with
trenbolone
dose.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
139
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
treatment
atretic_
follicles
Figure
4.46.
Boxplot
of
the
proportion
of
atretic
follicles
per
300
follicles
by
treatment
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
asterisk
represents
a
probable
outlier.

Corpora
Lutea
 
The
mean
proportion
of
corpora
lutea
per
300
follicles
(
counted
per
fish)
ranged
from
0.001
for
females
in
the
High
concentration
to
0.011
for
females
in
the
Control
treatment
(
Figure
4.47).
There
were
significant
differences
in
the
proportions
of
corpora
lutea
among
treatments
(
Kruskal­
Wallis,
H
=
10.16,
p
=
0.017,
df
=
2).
The
mean
value
for
fish
from
the
High
concentration
was
less
than
that
for
the
Low
concentration,
but
not
than
that
for
the
Control
treatment.

Mid
Low
High
Control
0.04
0.03
0.02
0.01
0.00
treatment
corpora_
lutea
Figure
4.47.
Boxplot
of
the
proportion
of
corpora
lutea
per
300
follicles
by
treatment
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
asterisks
represent
probable
outliers.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
140
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
4.3.6
Male
Gonad
Histology
Testes
Staging
by
Microscopic
Field
 
Testes
from
males
exposed
to
trenbolone
during
the
Nonspawning
Adult
14­
Day
Trenbolone
assay
were
examined
to
determine
the
general
developmental
condition.
Males
in
all
treatments
had
well­
developed
testes
with
most
showing
Stage
4
and
Stage
5
development
(
see
Methods
for
description
of
developmental
stages).
All
of
the
120
microscopic
fields
examined
in
the
10
Control
treatment
males
showed
Stage
4
(
58
fields)
or
Stage
5
(
62
fields)
development.
All
of
the
108
microscopic
fields
examined
in
the
9
Low­
concentration
treatment
males
showed
Stage
4
(
74
fields)
or
Stage
5
(
34
fields)
development.
All
of
the
120
microscopic
fields
examined
in
the
10
Medium
concentration
treatment
males
showed
Stage
4
(
50
fields)
or
Stage
5
(
70
fields)
development.
All
of
the
120
microscopic
fields
examined
in
the
10
High­
concentration
treatment
males
showed
Stage
4
(
76
fields)
or
Stage
5
(
44
fields)
development.
Statistical
analysis
of
the
mean
staging
from
12
microscopic
fields
per
fish
revealed
no
significant
differences
among
treatments
(
Kruskal­
Wallis,
H
=
4.24,
p
=
0.236,
df
=
2).

Quantitative
Testicular
Staging
 
One
hundred
cells
in
each
of
three
sections
per
male
were
examined
to
quantitatively
determine
the
developmental
condition
of
the
testes.
The
developmental
stage
of
the
testes
from
fish
in
all
treatments,
except
the
Low
concentration,
ranged
from
Stage
2a
to
Stage
5
(
Figure
4.48).
The
developmental
stage
of
the
testes
from
fish
in
the
Low
concentration
ranged
from
Stage
2b
to
Stage
5.
Variability
within
treatments
for
each
stage
was
very
high
as
indicated
by
CVs
that
ranged
as
high
as
211%
(
Table
4.40).
There
were
no
significant
differences
among
treatments
in
the
proportion
of
cells
in
developmental
Stages
2a,
2b,
3a,
3b,
4,
and
5
(
Table
4.40).
Therefore,
there
was
no
consistent
pattern
of
significant
difference
associated
with
trenbolone
dose.

Table
4.40.
Descriptive
statistics
of
the
proportion
of
testes
cells
in
each
developmental
stage
for
males
from
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
and
results
of
the
Kruskal­
Wallis
Test
(
df
=
2)
comparing
treatments.

Control
(
n
=
10)
Low
(
n
=
9)
Medium
(
n
=
10)
High
(
n
=
10)
Kruskal­
Wallis
Stage
Mean
Stdev
CV
Mean
Stdev
CV
Mean
Stdev
CV
Mean
Stdev
CV
H
p
1
0
0
 
0
0
 
0
0
 
0
0
 
 
 
2a
0.001
0.002
175%
0
0
 
0.002
0.003
194%
0.001
0.001
211%
3.44
0.329
2b
0.009
0.013
150%
0.010
0.012
121%
0.005
0.008
159%
0.007
0.008
125%
1.10
0.777
3a
0.027
0.032
119%
0.066
0.055
84%
0.046
0.076
165%
0.055
0.031
56%
5.69
0.128
3b
0.116
0.111
95%
0.149
0.082
55%
0.152
0.162
107%
0.117
0.093
79%
1.62
0.654
4
0.082
0.084
102%
0.090
0.061
67%
0.064
0.074
115%
0.087
0.066
76%
1.77
0.621
5
0.765
0.224
29%
0.685
0.166
24%
0.731
0.305
42%
0.700
0.170
24%
1.79
0.618
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
141
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1
2a
2b
3a
3b
4
5
Testicular
Stage
Proportion
of
Cells
Control
Low
Mid
High
Figure
4.
48.
Frequency
histogram
showing
the
quantitative
developmental
staging
of
testes
for
each
treatment
of
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
For
each
treatment,
the
columns
represent
the
grand
mean
proportion
of
cells
in
each
stage
and
the
bars
represent
the
standard
deviation.

Tubule
Diameter
 
The
average
diameter
of
the
seminiferous
tubules
of
males
from
the
Control
treatment
ranged
from
85.3
µ
m
to
236.7
µ
m
(
Figure
4.49).
Tubule
diameters
of
males
from
the
three
test
concentrations
ranged
from
87.2
µ
m
to
250.0
µ
m.
No
significant
differences
in
the
mean
tubule
diameter
per
treatment
(
Table
4.41)
were
detected
(
Kruskal­
Wallis,
H
=
2.54,
p
=
0.468,
df
=
2).
The
achieved
power
for
this
endpoint
was
13%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
78
(
Table
4.41).

Table
4.41.
Summary
statistics
and
power
estimates
for
male
seminiferous
tubule
diameter
data
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
159.3
54.8
10
13%
78
Low
9
133.3
41.8
9
Medium
10
162.2
42.4
10
High
10
160.3
39.5
10
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
9.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
142
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
250
200
150
100
treatment
diameter
Figure
4.49.
Boxplot
of
male
seminiferous
tubule
diameter
(
µ
m)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
asterisk
represents
a
probable
outlier.

Observations
 
Two
males
in
the
Medium
concentration
treatment
showed
disorganized
tubular
components.
No
interstitial
Sertoli
cell
proliferation
and
no
Leydig
cell
proliferation
was
observed
for
any
treatment.
No
testicular
atrophy
was
recorded
and
no
ovatestes
were
observed
for
any
treatment.

4.3.7
Plasma
Steroid
Concentrations
Estradiol
 
Estradiol
concentrations
in
Control­
treatment
females
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
234
pg/
mL
to
3,382
pg/
mL
(
Figure
4.50).
Among
females
exposed
to
the
three
trenbolone
concentrations,
estradiol
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
3,913
pg/
mL.
A
significant
difference
in
the
mean
estradiol
concentration
per
treatment
(
Table
4.42)
was
detected
(
Kruskal­
Wallis,
H
=
11.87,
p
=
0.008,
df
=
2).
The
mean
estradiol
concentration
in
females
from
the
Medium
concentration
was
less
than
that
in
females
from
the
Low
concentration
and
the
Control
treatment.
The
achieved
power
for
this
endpoint
was
59%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
11
(
Table
4.42).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
143
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.42.
Summary
statistics
and
power
estimates
for
female
estradiol
concentrations
(
pg/
mL)
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
7
1,346
1,013
75%
59%
11
Low
10
1,730
1,272
74%
Medium
10
378
461
122%
High
9
475
430
91%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
7.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Mid
Low
High
Control
4000
3000
2000
1000
0
treatment
Estradiol
Figure
4.50.
Boxplot
of
female
estradiol
concentration
(
pg/
mL)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisks
represent
probable
outliers.

Estradiol
concentrations
in
Control
treatment
males
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
0
pg/
mL
(
not
detected)
to
566
pg/
mL
(
Figure
4.51).
Among
males
exposed
to
the
three
trenbolone
concentrations,
estradiol
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
520
pg/
mL.
A
significant
difference
in
the
mean
estradiol
concentration
per
treatment
(
Table
4.43)
was
detected
(
Kruskal­
Wallis,
H
=
8.02,
p
=
0.046,
df
=
2).
Estradiol
concentrations
in
males
from
the
Low
concentration
consistently
ranked
lower
than
those
form
the
Medium
concentration.
The
achieved
power
for
this
endpoint
was
17%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
33
(
Table
4.43).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
144
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.43.
Summary
statistics
and
power
estimates
for
male
estradiol
concentrations
(
pg/
mL)
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
6
209
210
101%
17%
33
Low
6
78
121
156%
Medium
8
276
81
30%
High
7
320
168
52%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
6.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Mid
Low
High
Control
600
500
400
300
200
100
0
treatment
Estradiol
Figure
4.51.
Boxplot
of
male
estradiol
concentration
(
pg/
mL)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.

Testosterone
 
Testosterone
concentrations
in
Control­
treatment
females
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
790
pg/
mL
to
1,145
pg/
mL
(
Figure
4.52).
Among
females
exposed
to
the
three
trenbolone
concentrations,
testosterone
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
526
pg/
mL.
Significant
differences
in
the
mean
testosterone
concentration
per
treatment
(
Table
4.44)
were
detected
(
Kruskal­
Wallis,
H
=
9.56,
p
=
0.023,
df
=
2).
The
mean
testosterone
concentration
in
females
from
the
High
concentration
was
less
than
that
of
females
from
the
Low
and
Medium
concentrations
and
from
the
Control
treatment.
The
achieved
power
for
this
endpoint
was
10%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
14
(
Table
4.44).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
145
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.44.
Summary
statistics
and
power
estimates
for
female
testosterone
concentrations
(
pg/
mL)
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
4
975
156
16%
10%
14
Low
8
217
245
113%
Medium
2
138
195
141%
High
4
134
156
116%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
2.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Mid
Low
High
Control
1000
500
0
treatment
Testosterone
Figure
4.52.
Boxplot
of
female
testosterone
concentration
(
pg/
mL)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

Testosterone
concentrations
in
Control
treatment
males
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
1,125
pg/
mL
to
6,094
pg/
mL
(
Figure
4.53).
Among
males
exposed
to
the
three
trenbolone
concentrations,
testosterone
concentrations
ranged
from
226
pg/
mL
to
4,796
pg/
mL.
Significant
differences
in
the
mean
testosterone
concentration
per
treatment
(
Table
4.45)
were
detected
(
Kruskal­
Wallis,
H
=
11.14,
p
=
0.011,
df
=
2).
The
mean
testosterone
concentration
in
males
from
the
Medium
and
Low
concentrations
were
less
than
that
in
males
from
the
Control
treatment.
The
achieved
power
for
this
endpoint
was
76%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
8
(
Table
4.45).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
146
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.45.
Summary
statistics
and
power
estimates
for
male
testosterone
concentrations
(
pg/
mL)
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
7
3,487
1,649
47%
76%
8
Low
8
1,018
955
94%
Medium
9
1,283
1,075
84%
High
9
1,818
1,514
83%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
7.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Mid
Low
High
Control
6000
5000
4000
3000
2000
1000
0
treatment
Testosterone
Figure
4.53.
Boxplot
of
male
testosterone
concentration
(
pg/
mL)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

11­
ketotestosterone
 
11­
ketotesosterone
was
not
detected
in
females
from
the
Control
treatment
(
3
individuals),
the
Low
concentration
(
3
individuals),
the
Medium
concentration
(
2
individuals),
or
the
High
concentration
(
6
individuals).

11­
ketotesosterone
concentrations
in
Control
treatment
males
used
during
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
8,913
pg/
mL
to
64,045
pg/
mL
(
Figure
4.54).
Among
males
exposed
to
the
three
trenbolone
concentrations,
11­
ketotesosterone
concentrations
ranged
from
0
pg/
mL
(
not
detected)
to
57,115
pg/
mL.
A
significant
difference
in
the
mean
11­
ketotesosterone
concentration
per
treatment
(
Table
4.46)
was
detected
(
Kruskal­
Wallis,
H
=
8.04,
p
=
0.045,
df
=
2).
The
mean
11­
ketotesosterone
concentration
in
males
from
the
Low
concentration
was
less
than
that
in
males
from
the
Control
treatment.
The
achieved
power
for
this
endpoint
was
53%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
14
(
Table
4.46).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
147
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.46.
Summary
statistics
and
power
estimates
for
male
11­
ketotesosterone
concentrations
(
pg/
mL)
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
8
28,598
17,106
60%
53%
14
Low
9
7,926
11,034
139%
Medium
10
13,943
16,314
117%
High
10
16,500
22,096
134%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
8.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Mid
Low
High
Control
70000
60000
50000
40000
30000
20000
10000
0
treatment
11­
keto
Figure
4.54.
Boxplot
of
male
11­
ketotesosterone
concentration
(
pg/
mL)
by
treatment
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represent
probably
outliers.

4.3.8
Body
Weight
and
Length
The
body
weight
of
females
used
in
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
1.4
g
to
3.3
g
(
Figure
4.55).
There
were
no
significant
differences
in
mean
body
weight
(
natural
log
transformed)
among
treatments
(
Kruskal­
Wallis,
H
=
4.05,
p
=
0.256,
df
=
3).
The
achieved
power
for
this
endpoint
was
35%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
26
(
Table
4.47).

The
body
(
fork)
length
of
females
used
in
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
47
mm
to
63
mm
(
Figure
4.56).
There
were
no
significant
differences
in
mean
body
length
(
natural
log
transformed)
among
treatments
(
Kruskal­
Wallis,
H
=
5.90,
p
=
0.117,
df
=
3).
The
achieved
power
for
this
endpoint
was
33%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
28
(
Table
4.48).
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
148
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Table
4.47.
Summary
statistics
and
power
estimates
for
female
body
weight
(
g)
data
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
1.9
0.3
16%
35%
26
Low
10
2.1
0.4
19%
Medium
10
2.3
0.5
22%
High
10
2.1
0.5
23%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
10.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.

Mid
Low
High
Control
3.5
2.5
1.5
treatment
fish_
wgt_
whole
(
y
)

Figure
4.55.
Boxplot
of
female
body
weight
(
g)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

Table
4.48.
Summary
statistics
and
power
estimates
for
female
body
length
(
mm)
data
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
51.8
3.9
7%
33%
28
Low
10
54.0
3.1
6%
Medium
10
55.2
3.3
6%
High
10
52.8
4.9
9%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
10.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
149
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
65
60
55
50
treatment
fish_
length_
fork
Figure
4.56.
Boxplot
of
female
body
length
(
mm)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.

The
body
weight
of
males
used
in
the
EPA
14­
Day
Trenbolone
assay
ranged
from
2.4
g
to
6.5
g
(
Figure
4.57).
There
were
no
significant
differences
in
mean
body
weight
among
treatments
(
Kruskal­
Wallis,
H
=
5.34,
p
=
0.148,
df
=
3).
The
achieved
power
for
this
endpoint
was
48%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
19
(
Table
4.49).

The
body
(
fork)
length
of
males
used
in
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay
ranged
from
51
mm
to
79
mm
(
Figure
4.58).
There
were
no
significant
differences
in
mean
body
length
(
natural
log
transformed)
among
treatments
(
Kruskal­
Wallis,
H
=
6.72,
p
=
0.081,
df
=
3).
The
achieved
power
for
this
endpoint
was
48%,
and
the
sample
size
required
to
detect
a
significant
difference
from
the
Control
treatment
at
80%
power
was
19
(
Table
4.50).

Table
4.49.
Summary
statistics
and
power
estimates
for
male
body
weight
(
g)
data
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
3.7
0.8
21%
48%
19
Low
10
4.0
0.8
21%
Medium
10
4.7
1.1
23%
High
10
4.4
0.9
21%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
10.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
150
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
6.5
5.5
4.5
3.5
2.5
treatment
fish_
wgt_
whole
(
y
)

Figure
4.57.
Boxplot
of
male
body
weight
(
g)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
the
circle
is
the
mean
value,
and
the
asterisk
represents
a
probable
outlier.

Table
4.50.
Summary
statistics
and
power
estimates
for
male
body
length
(
mm)
data
for
the
Nonspawning
Adult
14­
Day
Trenbolone
assay.

Level
N
Mean
Stdev
CV
Achieved
Power
1
Sample
Size
Required
2
Control
10
59.9
5.4
9%
48%
19
Low
10
65.4
4.8
7%
Medium
10
65.8
7.1
11%
High
10
64.8
3.2
5%
1
Calculated
from
natural
log
transformed
data;
with
sample
size
=
10.
2
To
detect
a
significant
difference
from
Control
treatment
based
on
maximum
achieved
absolute
difference;
calculated
on
natural
log
transformed
data.
DRAFT
EPA
WA
3­
8
(
Report
of
WA
2­
18
Study)
151
Preliminary
Data:
May
be
subject
to
change
following
QA
and
Management
review
July
30,
2003
Mid
Low
High
Control
80
70
60
50
treatment
fish_
length_
fork
(
y
)

Figure
4.58.
Boxplot
of
male
body
length
(
mm)
by
treatment
for
the
Non­
spawning
Adult
14­
Day
Trenbolone
assay.
The
box
represents
the
interquartile
range,
whiskers
represent
the
data
range,
the
horizontal
line
is
the
median
value,
and
the
circle
is
the
mean
value.
