Comments	Action/Response	Page #

EPA 



ORD finds that., in general, the six developmental toxicity studies were
conducted according to the EPA test guidelines. At this time, however,
we cannot recommend approval of the study reports as submitted, because
we have identified several deficiencies in the statistical analyses and
reporting of data. The attached memoranda provide specific comments on
each of the study reports.	No action required

	Additionally, in our review of the developmental toxicity study
reports, we noticed that certain data on dams that delivered early were
omitted from the appendices because the data were excluded from the
statistical analyses. Please note that all of the data generated as part
of the 21l(b) testing program must be provided to EPA for review, even
if they are negative or are not used in the statistical analyses.
Gestation observations, body weight and food consumption data for the
non-pregnant animals added to the report

	We strongly encourage the RG to address our comments before revised
versions of the reports are submitted to EPA. The RG should also
adequately respond to the comments raised byte independent peer
reviewers and provide the additional information requested. To help our
reviewers complete the reviews of the revised study reports, we also
request that a summary of the changes or marked-up copies be provided.
Checklist completed for changes.  Report provided with track changes
function enabled

	In general the NHEERL reviewers concluded that the studies were
conducted in accordance with the testing guidelines; however, several
issues were raised that warrant further attention. 	No action required

	

Interpretation of the data in this study is confounded by significantly
different numbers of corpora lutea (CL,) in the treatment groups. As
noted in the API response to the independent reviewers (but not in the
study report), the number of CL is clearly not a treatment effect
(because ovulation occurs well before the initiation of exposure), but
does influence the number of implantation sites, litter size, and litter
weight. In view of the CL data, we recommend that the study authors
re-analyze the relevant data using the number of CL as a covariate.	Data
reanalyzed using the CL counts as a covariate	Appendix K

Uterine weights were significantly decreased at 20,000 mg/m3. Because
this endpoint can reflect effects on litter size and fetal weight, this
finding represents developmental toxicity, not maternal toxicity. The
report conveys this finding in the context of endpoints reflecting
maternal toxicity. The study authors should report this finding in the
context of developmental toxicity.

In view of this finding, the study authors should analyze litter weight
as a more direct measure of the combined influence of fetal weight and
litter size. (As noted above, because of significant differences in the
numbers of CL in this study, CL should be used as a covariate in this
analysis.)

Pending re-analysis of uterine weight and litter weight, with CL as a
covariate, we regard the: significant reduction in uterine weight to
reflect developmental toxicity at 20,000 mg/m3.	Data reanalyzed

Data reanalyzed	Appendix K

Given the significant reduction in extrauterine weight gain on GD 0-21,
there appears to be a lasting effect of exposure at 20,000 mg/m3; thus,
it is questionable to describe the maternal toxicity evidenced here as
"transient".	Summary and discussion revised	1-2 and 4-8



The statistical ana1yses of the fetal examination data are inappropriate
and should be revised. In the analyses (chi-square and Fisher's exact
test), the study authors used the fetus, rather than the litter, as the
experimental unit. Since fetuses from the same litter are not
independent of each other, a fetus-based analysis inappropriately
inflates the degrees of freedom. (The study authors also did chi-square
and Fisher's exact test on litter incidences; this is statistically
valid, but is inadequate without additional statistics because it
doesn't consider within-litter incidences and can lead to false
]negatives.)	Data reanalyzed	Appendix K

Analyses of fetal examination data should combine incidences of related
findings to better show trends that may be treatment related. Examples
of related findings that could be combined include:

vertebral centra variations (including thoracic and lumbar; including
unossified, bifid, and dumbbell-shaped)

supernumerary thoracic/lumbar ribs (including rudimentary and
well-formed ribs on L1 and T14)	Data reanalyzed with related
observations combined	Appendix K

The statistically significant increased fetal incidences of skeletal
variations and dumbbell-shaped thoracic vertebral anlage were not
regarded as evidence of developmental toxicity by the study authors
because the litter incidence was not statistically significant. True,
the litter is the appropriate unit of analysis; however, as stated
above, the litter analyses (chi-square and Fisher' exact) presented in
the report do not consider within-litter incidences and can lead to
false negative results. 'Thus, pending re-analysis, we tentatively
regard these increased incidences to reflect deve1opme:ntal toxicity at
20,000 mg/m3.	Data reanalyzed and the incidence of these observations
were not statistically significant.	Appendix K



In several passages of the report, terminology is used improperly:

The abstract states that "The extent of these decreases [in body weight
change] also produced ... decreases in food consumption ..." Although
the body weight data and food consumption data clearly were associated
with each other, it is inappropriate to speculate -which effect caused
the other. (I.e., decreased food consumption may have led to reduced
weight gain).

The abstract and summary state that "reduced body weight gain produced
... reduced gravid uterine weight." Although body weight gain can be
influenced by an effect on uterine weight, it was not demonstrated to
affect uterine weight, as implied in the report.

In the discussion, the report states "body weight change data correlated
well with food consumption data ..." The word "correlated" should be
avoided unless a correlation was statistically evaluated.	Abstract
revised

Abstract revised

Discussion revised	i

i

4-7

Peer Reviewer – Schlesinger



The study was conducted in a scientifically sound manner and followed
the appropriate protocols. There were no significant deviations from
these protocols that would have affected the outcome of the study.	No
action required

	The reported conclusion of a NOAEL of 2,000 mg/m3 for maternal toxicity
may be supported by the results of the study, with the caveat noted
below. However, it is not clear that the NOAEL of 20,000 mg/m3 for
developmental toxicity is justified. This is due to the apparent
conflicting results of statistical tests, noted below, used to evaluate
fetal skeletal deformation. Furthermore, there was a significant
decrease in aspects of uterine implantation data, namely the number of
live male fetuses, implantation sites and corpora lutea, in the 2,000
mg/m3 exposure group. However, if these changes are interpreted as
maternal toxicity rather than developmental toxicity, then the NOAEL for
the former toxicity as stated in the report is too high.	No change

	Some other concerns are as follows:



It is stated that, "Statistical treatment of the results was conducted
where appropriate." What determined whether analysis was appropriate or
not? The report should clearly indicate which specific statistical tests
were used for which datasets.	Statistical analyses section rewritten
3-9, 3-10, and 3-11

The investigators indicate that they performed statistical testing with
both transformed and nontransformed data. If data need to be transformed
because they are not normally distributed as raw data then the
statistical tests with these transformed data are the ones that should
be reported. There cannot be a "picking and choosing" of data sets to
report. This could result in confusion in deciding which dataset is
"biologically significant" if results of one analysis are statistically
significant and the results of the other are not.	Statistical analyses
section rewritten	3-9, 3-10, and 3-11

It is stated that statistical tests were conducted at both the 5% and 1%
significance levels. The justification for this is not apparent. One of
the a priori decisions that are to be made before the study is conducted
involves choice of the appropriate level of significance. Only one
should be selected based upon whatever criteria the study director
chooses and only this level should be reported.	Statistical analyses
section rewritten	3-9, 3-10, and 3-11

In the reported results of skeletal observations in the fetus, it is
noted that rudimentary lumbar limbs were significantly different from
control in the high dose group using the Chi Square analysis, but no
such difference was noted with the Fischer Exact Test. This begs the
question of how is the reader to interpret the biological significance
of this effect. As noted, the type of statistical testing should be
determined before the study is performed and testing of the same dataset
by different techniques should not be performed.	Data reanalyzed with
different statistical methods	Appendix K

The investigators should indicate why there was a fairly high
concentration of particles in the air control chamber. Was incoming air
filtered? This reviewer does not consider a level of 0.5 mg/m3 as
"minimal," as indicated in the report.	No change.  The air supply to the
chamber room was filtered.  However, the make-up air for the chambers
was drawn from the chamber room where animal dander and household-type
dust may have been present.  0.5 mg/m3 is minimal in an atmosphere of
2000 to 20000 mg/m3.  Additionally, the particulates were considered to
be animal dander or household type dust and not from the test substance.

	Peer Reviewer – Goldsworthy



The study including exposure methodology was properly conducted and
scientifically sound.  None of the protocol exceptions noted adversely
affected study results or integrity.	No action required

	Although not significant, at every treatment and interval, mean
gestation body weight and body weight change was decreased in the 2000
mg/mg3 group; this observation may be significant in trend analysis.  No
clinical or gross observations were attributable to treatment.	No
change.  Much of this difference is probably due to the difference the
control group uterine weight  and the 2000 mg/m3 group uterine weight. 
The GD 21 body weights corrected for the uterine weight are very similar
in the control group and the 2000 mg/m3 group, 314.2 and 312.9 grams ,
respectively.

	Results (page 4-3 and Appendix F-1) note statistically decreases in
number of live fetuses, implantation sites and corpora lutea in the
2,000 m/m3 treatment group. In the 20,000 mg/m3 treatment group, a
statistically significant decrease in the number of corpora lutea was
observed. Although not significant, the number of corpora lutea was also
decreased in the 10,000 mg/m3 group. The abstract and summary (1-2)
state that there were not treatment-related differences for uterine
implantation data, presumably because of the qualifier 'on either a
fetus or litter basis'. Results section should show and present the data
in all farms. The abstract and summary should be reworded to be more
explicit and comprehensive. The discussion section incorrectly states
that no differences were observed for this endpoint and needs to be
changed. Additionally, this finding needs to be considered and discussed
in the

setting of the NOAEL.	Abstract, summary, and discussion revised	1, 1-2,
and 4-8

A statistical increase in total fetuses with skeletal variations in the
high dose group compared to controls was observed. Similarly, then was a
statistical increase in dumbbell-shaped thoracic vertebrae adage on a
fetus basis in the 20,000mg/m3 treatment ~ u p . These findings were not
considered biologically relevant due to the lack of significance on a
litter basis, the basic unit of analysis; thus these two finding were
not considered in the NOAEL setting for developmental toxicity. Text
should provide a table of this data in results section. Also, discussion
should note that an increase was also seen in the incidence of
dumbbell-shaped thoracic vertebrae anlage in Gasoline with MTBE (20,000
mg/m3 pup) study. The importance of these findings in the overall
context of the API testing initiative may need to be further examined.
No change.

	Text/Editorial Comments



Summary, page 1-2:

Fifth paragraph “not not biologically significant”; change to “not
considered to be biologically significant”.	Revised	1-2

Same change to abstract (line 20).	Revised	i

Discussion, apge 4-6:

First paragraph is not a comprehensive discussion of the changes noted
in the body weight, body weight gains and food consumption.  The last
sentence in the first paragraph should delete “thus indicating that
toxicity was no long evident”; this is too broad of a statement.
Revised	4-7

211 (b) Research Group Reviewer 



General Comments



- When the report is final, it should have sequential numbering in
addition to the A-1, A-2, etc. system so that any reader can be certain
that all pages are present.	Report repaginated	All pages

- The QA page for this report, as for the previous ones, needs to state
what was audited. This needs to be distributed and by the sponsor before
the report can be finalized. Providing a blank QA page with the audited
report is not acceptable. The BGV+DIPE study included a draft QA
statement; we should see the draft QA statement for each study.	QA
statement added	ix



- Statistical transformations were performed and therefore need to be
reported, whether or not the results are statistically unlike earlier
reports, there is no statement indicating that nonsignificant
transformations would not be reported; however, just like the earlier
reports, only summary transformation values (group means in Appendix F)
are included, and the individual values be included in the appendix as
well. Secondly, this report differs from previous reports in that this
one does not indicate that statistically significant linear regression
will be disregarded wherever the multi-group comparison is not
statistically significant. should have a consistent approach for the
entire set of 211(b) developmental studies. Since the statistically
significant linear trend GD 21C body weight was ignored in the text
table on p. 4-1, 1 presume that the same philosophy as used previously
has been applied, but no longer described in the methods section. The
better method would be to describe the use of the statistics and justify
the approach.	Transformations reported in Appendix F

Linear trend results added to the results for body weight, food
consumption, and uterine implantation data.	Appendix F

4-1, 4-2, 4-3

- The Abstract and Summary provide interpretation of findings, but the
results section itself is strikingly missing similar evaluation. 
Duplicating table data in the results section doesn't provide the
interpretation that should be provided in the results.	Revised as needed
	4-1 – 4-6

- Appendix J provides information about the composition of the vapor to
which the animals are exposed. This information is pertinent and should
be discussed in the text. In the current draft, there is no mention of
these results anywhere in the text.	Added a reference to Appendix J and
summarized the analytical results.	4-7

- The Historical Control Data, presented as Appendix K, and seen in the
BGV+ETBE draft, is a great improvement over the earlier version. Several
recommendations, mentioned in 6/03 in comments to another 211(b) draft
report:	Historical control data revised	Appendix L



- Inclusion of maternal body weight and food consumption interval data
is not highly useful, and my opinion is that it doesn't enhance the
report. If it remains in Appendix K, the studies should be put into
order, to footnote each page to distinguish the E.  Millstone data from
the Annandale data.	Maternal body weight data and food consumption data
removed	Appendix L

- I also recommend that the lab lighten the background on lines within
the data, since it makes the information difficult to	Tables reformatted
Appendix L

- The usual use of a historical control database, other than to
demonstrate the lab's recent experience, is to aid in of questionable
findings. To make it easier for the data to serve this purpose, it
should be organized by external, visceral, and skeletal, as the lab has
done, and then by parts: i.e., adrenals, discolored, and adrenals,
hypoplastic, should be in order rather than in a group of diverse
discolored and/or diverse hypoplastic findings. This is in part because
certain observations related due to having a similar embryologic cause:
there's not much difference between a dumbbell shaped and a bifid
centra, may simply be more ossified than the other.	Tables reformatted
Appendix L

Rudimentary rib incidence in this study should be compared to the
Historical Control range for this endpoint.	Added to the results and
discussion	4-4, 4-5, and 

4-8



Specific Comments (by page and section)



Page i, Abstract:

Statistically significant decreases in body weight change do not produce
decreases in food consumption, although the reverse be true. The report
has the appropriate information in this section, but not organized
optimally. All food consumption comments should be stated, then body
weight/change findings followed by uterine data. Lack of effect on fetal
body weight, if correct, be mentioned in the abstract as has been done
for no effects upon external or visceral findings. The lack of
biological significance for skeletal findings based upon lack of
statistical significance on the litter basis, although statistically
significant on the fetus not strongly compelling. Control litters in
this study and historically have a much higher incidence of affected
litters for skeletal endpoints than for other parameters such as
external observations, with a consequence that only a large difference
can be detected. The litter incidences for total skeletal variations
from the historical control data for other Annandale studies range from
50% up to 76%. Given that this is based almost completely upon
rudimentary ribs in this study, which have been associated with maternal
stress, there should be a more thorough justification for why total
skeletal variations as well as dumbbell-shaped cartilage should not be
taken as developmental toxicity.	Abstract rewritten to reflect new
statistics, etc.	i

Page ii:

Table of Contents: Page 2-1, there are two headers on the page between
Study Initiation and Inlife Test Period. All headers be listed, or
Experimental Start Date and Experimental Termination Date should be
single-line entries within another section. The same applies to Nominal
Concentration, Analytical Concentration, Chamber Homogeneity, and
Particle Size Analysis.	Added Experimental Start Date and Experimental
Termination Date to Table of Contents.  The subheaders under Analysis of
Mixtures were not added as they are not at the same level as the other
items on the table of contents.	ii



Page viii:

Quality Assurance Statement: This should have been virtually complete by
the time that the draft report was issued. The of what was audited, and
when, will need to be provided to the sponsor prior to the finalization
of the report.	Draft Quality Assurance Statement added	ix

Page 1-1:

Summary, first paragraph: The lab hasn't indicated why 2000 mg/m3 was
expected to be a NOAEL.	Justification of dose selection revised to
statement supplied by the Sponsor.	1-1

Page 1-2:

Summary, first paragraph on this page: Reduced gravid uterine weight
can't be solely chalked up to maternal effects, since so of the weight
belongs to the litter. Although there was no statistical reduction in
live litter size, and litter size is within normal limits, litter size
does associate well with the difference found for gravid uterine weight,
and this should be mentioned as a possibility or as text in the
results/discussion section. If reduced gravid uterine weight is not
considered to be a treatment-related effect, then should either not be
mentioned in the summary (which should focus upon findings attributed to
exposure), or should be provided a "disclaimer."

"The only fetal observations that were considered related to treatment
..." If there are fetal findings attributed to exposure at 20000 mg/m3,
then this level is NOT a NOAEL. My own opinion is that the reduced
uterine weight and increased number of rudimentary ribs suggest marginal
developmental effects likely to be encountered in the presence of
maternal toxicity particularly during the half of exposure. Since the
lab interprets this differently, there should be a stronger
justification for discounting developmental findings as not related to
exposure, such as the reduced gravid uterine weight being related to
normal differences in litter size, or that the dumbbell shaped cartilage
for the thoracic vertebrae occurred without a corresponding finding in
the ossification Although the report is absolutely correct that the
litter should be the standard basis for comparison in a developmental
study, incidences amongst fetuses isn't entirely ignored, or it wouldn't
be statistically evaluated.	Summary revised	1-2

Fifth paragraph on this page: typographical error: delete the redundant
"not" in the final sentence in this paragraph.	Corrected	1-2

Page 2-1:

Justification for selection of test system: I believe that the rat was
specified in the vehicle emissions testing document (1994). It is not a
required species in the developmental testing guideline, it is a
recommended species.	Statement revised	2-1

Page 2-2:

Justification of dose selection: The high dose4 one-half of the LEL.
There's no indication of why the low dose was expected to produce a
NOAEL.	Justification of dose selection revised to statement supplied by
the Sponsor.	2-2

Page 3-1:

Test Substance, Supplier: The correct name may be Chevron Products,
Richmond Refinery OR Chevron Research and Company, but not Chevron
Research. Please review the shipping records; if these state Chevron
Research, there will need to be correction.	Supplier name revised	3-1

Characterization of the test substance: At a minimum, this section
should mention the analytical report in Appendix J.	Statement added to
the characterization	3-1

Page 3-3:

The age of the male rats at the time of their use for breeding should be
provided. The report should also state whether the males are the same
age as the females to clarify that they cannot be littermates.	Age of
the males added as well as a statement about the males and females not
being siblings	3-2

Page 3-4:

Feed, Water: Both the feed and water analyses should be included in the
report as appendices.	Added as Appendix M	Appendix M

Page 3-8:

Statistical Methods: Individual transformation values should be included
in Appendix F.	Individual transformation values added to Appendix F	F-4,
F-6, F-8, F-10

Page 3-9:

Statistical Methods, continued: As with previous 211(b) developmental
draft reports, there should be explanations provided for: (a) why linear
regression is performed when there isn't a difference between groups;
(b) what criteria was used to ignore statistically significant results
from linear trend tests in these circumstances. What was the criteria
for performing a lack of fit test?	Linear regression results added to
the results.  Linear regression and the lack of fit test are performed
every time the statistics are run.  They are part of the software
package.	4-1, 4-2, and 4-3

Page 4-1:

Table 4-1 is redundant to the information provided in Appendix C. What
is missing in this section is the interpretation of the such as provided
in the Summary. Please see comments above.	Table 4-1 deleted. 
Interpretation of the data added	4-1

Page 4-2:

Gestation food consumption: The same remarks apply here as for body
weight information. There is a virtual absence of interpretation,
although this was handled well in the Summary. Table 4-2 reiterates
information already presented clearly in Appendix D, and should not take
the place of writing out an interpretation.	Table 4-2 deleted. 
Interpretation of the data added	4-2

Page 4-3:

Uterine Implantation Data: The text accurately states statistical
findings, then provides no interpretation of them and does not address
this in the discussion. Corpora lutea should be discounted as an event
that occurs prior to the onset of exposure, and the number of
implantation sites as well as the total number of males are likely to be
a result of the lower CL count. The report should also state that there
were no effects upon a number of important endpoints, such as live
litter size and the incidence of	Terine implantation data section
revised	4-3

Skeletal observations, first sentence: "The only treatment-related
statistically significant difference.. ." Either there were related
developmental effects, or not: if this sentence correctly conveys the
study director's interpretation, then the NOAEL is not 20000 mg/m3.
Treatment –related removed.  	4-3



Total skeletal variations appears to be due mostly to an increase in the
number of fetuses with rudimentary ribs; this should be stated. The text
should continue on to indicate if the incidence in the 20000 mg/m3 group
is within their normal range as indicated by their historical control
data. For the dumbbell-shaped thoracic vertebral cartilage, it is
surprising that this would be elevated without an increase in split or
dumbbell-shaped ossification of the centra, which may be an indication
that the finding is spurious.	Section revised as well as the Discussion
4-4 and 4-7

Page 4-4:

Exposure Data and Chamber Conditions: The analytical report in Appendix
J should be referenced and discussed here.	Analytical report referenced
and summarized in this section	4-6

The first paragraph subsequent to Table 4-4 has an incomplete sentence.
Revised	4-5

Page 4-5:

Discussion: The discussion should be in the following order: (1)
maternal toxicity findings; (2) uterine data findings; (3) developmental
findings; (4) NOAELs.	Discussion revised	4-7

The second and third paragraphs should be reordered to put the results
in better order.	Discussion revised	4-7

Page 4 of 4:

First paragraph: I believe that the body weight and food consumption
data are being over analyzed. Both indicate maternal during -first week
of exposure. I don't think that the report is well-served to dwell upon
the lack of perfect correlation between the statistically significant
intervals for these measurements.	Revised	4-7

Second paragraph: Same comment as previously - if there are fetal
observations related to treatment, there are developmental effects at
20000 mg/m3.	Revised the statement	4-8

Page 4-6:

Protocol Exceptions, chamber temperature and humidity: The values and
dates for the deviations should be listed herein, all of them.	Section
revised	4-9, I-23 – I-30



Page 5-2:

Fetal external and visceral examinations: Does the laboratory have
criteria for what observations constitute malformations and which are
variations? Criteria needs to be provided for why an observation
wouldn't belong in one of these two categories.	Criteria added	5-2

Appendix I, page 1-3:

Third paragraph, second sentence: "...injecting a weighed amount of the
test substance.. ."	Corrected	I-3

Appendix K, pages K-2 to K-5:

The studies are misordered.	Historical control data revised	Appendix L

211 (b) Research Group QA/QC Reviewer



The following items in the report and associated raw data require
further consideration:



Page i, Abstract, lines 18-19; Page1-2, Summary, paragraph 5, lines 2-3;
Page 4- 4, Skeletal Observations, paragraph 1, lines 3-4; Page 4-6,
Discussion, paragraph 2, lines 5-6; and page H-5, Appendix H, Incidence
of Fetal Observations:

The report text and table indicate that the incidence of fetuses with
dumbbell-shaped thoracic vertebral centra anlage was statistically
significantly increased in the 20,000 mg/m3 group. However, the
statistical analysis printouts reviewed at the time of the audit do not
appear to support this conclusion (statistical results showed no
significant differences for this parameter). Please verify and revise
report text and tables as needed.	Report revised to reflect the new
statistical analyses	i, 1-2, 4-4, 4-5, H-5 and Appendix K

Page vi, Compliance Statement:

The sponsor also needs to sign a compliance statement. It can be a
separate one from the Testing Facility’s, but there must be one signed
by the sponsor.	Compliance statement revised	vii



Page vi, Compliance Statement:

Since it was the sponsor’s responsibility to maintain the method of
synthesis, fabrication, or derivation of the test fuel, an this has not
been completed, it should be included in the sponsor’s compliance
statement.	We do not view this as a compliance issue.  The Sponsor has
responded that they have the information and we have added that
statement to the characterization section of the report.	3-1

Page viii, Quality Assurance Statement:

The QA statement needs to be completed.	Quality Assurance Statement
added	ix

Page 3-1, Test Substance:

An expiration date of March 18, 2007 (5 years after receipt) is given
for the test substance. Are there stability data to support this expiry
date?	No.  This date was assigned as per EMBSI’s SOP

	Page 3-2, Test Substance, Analytical Concentration:

It is indicated that chromatographic analyses showed major components of
the test atmosphere and was used to assess the stability of the test
substance over the duration of the study. For reproducibility purposes,
the number of components identified in the analyses should probably be
stated.	Number of components added	3-2

Page 3-5, Environmental Conditions:

The ranges given for animal room temperature and humidity appear to be
the protocol-specified ranges. The actual measured ranges should be
reported.	Corected	3-5

Pages 4-1 and 4-2, Table 4-1:

The word “Gestation” is mistyped in the second line of table titles.
Table 4-1 deleted	4-1 and 4-2

Page 4-4, Skeletal Observations and Page H-4, Appendix H, Incidence of
Fetal Observations:

The statistical analysis printouts indicate that the fetal incidence of
unossified sternebrae was significantly increased for the 2000 mg/m3
group. This significant difference is not indicated in the report text
or table. Please verify and revise report as needed.	Report revised to
reflect the new statistical analyses.  This observation was not
statistically significant in the new analyses	4-4, and Appendix K



Page 4-5, Table 4-3, Mean Exposure Concentrations:

For 2000 mg/m3 chamber, the mean and standard deviation for nominal
concentration should be 1874 and 118.6, respectively.	Corrected	4- 6,
I-10, and I-11

Page 4-7, Protocol Exceptions:

The deviation for the below-range concentration in the low dose chamber
on September 1, 2002 is included in the report. However, no
corresponding deviation report was found in the study records. A
deviation memo should be prepared and filed with the data.	Memo added to
data

	Page 5-2, Abbreviation List for Fetal External and Visceral Exams:

“Visceral” is mistyped in line 6.	Corrected	5-2

Page C-9, Appendix C, Individual Gestation Weight Changes:

For animal IGL934 (10,000 mg/m3), the body weight changes are
incorrectly reported for GD 0-5, GD 5-8, GD 8-11, GD 11-14, GD 14-17, GD
17-20 GD 20-21 and GD 5-21. My calculations indicate that these values
should be 21, 4, 11, 12, 32, 43, 8, and 110, respectively. The summary
table and statistics for the group appear to be correct, however.
Corrected	C-9

Page D-4, Appendix D, Individual Gestation Food Consumption:

For Animal IGL858 (10,000 mg/m3), the study data indicates that GD 11
feeder weight was not taken due to oversight. The explanation given for
the missing values in the report appendix (GD 8-11 and GD 11-14) is
excess feed spillage. The report should be changed to reflect the actual
reason for the missing data.	Footnote corrected	D-4



Page E-1, Appendix E, Incidence of Gross Postmortem Observations:

For the 10,000 mg/m3 group, the urinary bladder finding is reported as
“abnormal contents”, whereas the individual table reports a finding
of “white, hard object”, which is consistent with the study data. It
is recommended that the description given in the incidence table be
changed to reflect what is recorded in the data.	Revised	E-1

Pages E-2 and E-3, Appendix E, Individual Gross Postmortem Observations:

For the 0 mg/m3 group, necropsy findings are not reported for the
following animals:

IGL953, IGL869 and IGL921. All were described as having no observable
abnormalities.	Observations added	E-3

Page F-2, Appendix F, Mean Uterine Implantation Data:

The following items should be addressed:

For “Total Malformations”, According to the statistical analysis
printouts, the reported statistical significance notation should be
changed to K-J- (BARALF = 2.0, therefore, nonparametric analyses were
appropriate).

For “Total Variations”, the statistical notation should be A-L- (not
AL-).

For 2000 mg/m3 group, the mean and standard deviation for “Total
Malformations” should be 0.17 and 0.83, respectively.

For 2000 mg/m3 group, the standard deviation for “Total Affected”
should be 1.1.	a.  Corrected

b.  AL- is correct.  No change

c.  Corrected

d.  Corrected	a.  F-2

B.  No change

c.  F-2

d.  F-2



Pages H-1 through H-6, Appendix H, Incidence of Fetal Observations:

The following items should be addressed:

Pages H-1 through H-6: “Appendix” is mistyped at the top of each
page.

Page H-1: According to the statistical analysis printouts, the reported
statistical notation for “Total Fetuses with Skeletal Variations”
should be XX+ (not XX++), since dose response trend was significant at
p<0.05.

Page H-4: According to the statistical analysis printouts, the
statistical notation for the incidence of rudimentary lumbar ribs should
be XX++ (not X+).

Page H-4: According to the statistical analysis printouts, the number of
fetuses with unossified sternebrae in the 2000 mg/m3 group (observed in
6 fetuses) was significantly increased at p<0.05 by Fisher Exact Test.
An asterisk should be reported in the incidence table.

Page H-5: According to the statistical analysis printouts, the number of
fetuses with dumbbell-shaped thoracic vertebral centra anlage in the
20,000 mg/m3 group was not statistically significant (see item 1 of this
report). Please verify and delete asterisk from table, if appropriate.
a.  Corrected

b.  Statistical notations deleted from Appendix H.  Results of the
statistical analyses are now in Appendix K

c.  Statistical notations deleted from Appendix H.  Results of the
statistical analyses are now in Appendix K

d.  Statistical notations deleted from Appendix H.  Results of the
statistical analyses are now in Appendix K

e.  Statistical notations deleted from Appendix H.  Results of the
statistical analyses are now in Appendix K

	a.  H-1 – H-6

b.  H-1 – H-6

c.  H-1 – H-6

d.  H-1 – H-6

e.  H-1 – H-6

Page H-50, Appendix H, Individual Fetal Observations:

For IGL974 Fetus #1, it is recommended that a notation be added to
clarify that skeletal evaluation was performed on the head only.	No
change.  A full skeletal examination was conducted.

	Page H-78, Appendix H, Individual Fetal Observations:

For IGL866, the word “Visceral” is mistyped in the notes at the
bottom of the page.	Corrected	H-78

Page I-1, Appendix I, Inhalation Exposure Report:

The header on this page should be corrected to indicate TBA (currently
indicates MTBE).	Corrected	I-1

Page I-15, Table I-3, Summary of Exposure Data, 2000 mg/m3 Chamber:

The following items should be corrected/addressed:





September 23, 2002:

According to the data, the nominal concentration was 1715 (not 1757).
The table should be corrected. This changes the group mean and standard
deviation to 1874 and 118.6, respectively. These corrections are also
needed on page I-11 (overall summary table).

September 16, 2002:

At the 3-hour interval, the analytical concentration was observed to be
673 mg/m3. There was no explanation provided in the data for the low
concentration at this interval. Although the daily mean concentration
for this exposure was acceptable, exposure was not consistent. The
reason for the variation and any possible effects should be discussed in
the report.

September 17, 2002:

For the 5-hour interval, an analytical concentration of 0 mg/m3 was
obtained (rotameter was stuck due to air bubble). When calculating the
mean concentration for this exposure, the 5-hour result of 0 was
included, giving a mean concentration of 1738 mg/m3,which was below the
protocol-specified limits of +/- 10% of target. Is it appropriate to
include the “0” value in the mean calculation, or should it be
excluded from the mean since there was a technical problem with the
exposure equipment? If the 0 value is to be excluded, the mean
concentration for the daily exposure and/or the overall mean for this
chamber will need to corrected on this page as well as on pages 4-5, I-8
and I-11. If the 0 value is to be included, a protocol deviation is
needed to address the mean concentration being below protocol
specifications.

For the occurrence discussed in item c, above, the only documentation of
this event appears to be a notation on the GC summary sheet. This is a
summary spreadsheet and not raw data. Where is the initial documentation
of this occurrence? Please clarify in data.	a.  Corrected

b.  Memo added to data and note added to the report

c.  The 0 value should be used in the calculation as it was the actual
chamber concentration.  A protocol deviation will be added to the file
and will be reported.

d.  This is the raw data.  This sheet is populated and the
concentrations calculated as the GC prints out the data.  The note was
added after the 6-hour sampling completed or about an hour after the
problem was discovered. 	a.  I-15

b.  4-9

c.  4-9

d.  No change

Page I-20, Table I-4, Summary of Distribution Samples:

The word “approximately” is mistyped in the note at the bottom of
the page.	Corrected	I-20

ADDITIONAL DATA FINDINGS



Charcoal tube samples were stored frozen prior to transfer to the
analytical laboratory for desorption and analysis. Are stability data
available to support that the samples are stable under these conditions
for this period of time?	No change.  These are standard methods for
handling these types of samples.  Additionally, the data from these
samples shows that the samples were stable.

	Analytical Data, General Comment:

There needs to be a clear indication of what the analytical lab
considers raw data. When data are not printed out until a week after the
analysis, it is not appropriate to consider the paper printout to be
data. Since the GC program is storing data, the computer system must be
completely validated and follow all of the requirements of an on- line
data collection system (including change-control procedures, limited
access, complete maintenance of a data trail, etc.).	The printout is the
raw data.  We have tested the security of the chromatogram in the system
and it is secure.  All other operations around the chromatogram are
documented.

	Inhalation Exposure Records, August 31 – September 2, 2002:

For Groups 2, 3 and 4, the headers at the top of the forms incorrectly
indicate “Trials” rather than actual study exposures.	Corrected

	Inhalation Exposure Records, 10,000 mg/m3, September 19, 2002:

The entry for grams of test article recorded at the bottom of the page
is incorrectly written as 714.4. The correct value is 717.4 g. The
nominal concentration is correctly calculated.	Corrected

	

GC Daily Summary and GC Summary Sheets, General Comment:

Spreadsheets should be checked to ensure that review signatures and
dates are documented on all pages.	Reviewed

	GC Daily Summary Sheet, 2000 mg/m3, September 23, 2002:

The nominal concentration recorded at the bottom of the form should be
1715 (per the Inhalation Exposure Record) not 1757.	Corrected

	GC Daily Summary Sheet, 2000 mg/m3, September 16, 2002:

Total peak area mean should be 3063.933 (not 3177.13093).	Corrected

	GC Summary Sheets, General Comment:

In some instances the GC Daily Summaries were corrected/revised, but the
GC summary sheets, that summarize the results for all chambers, were not
corrected and replaced in the books. This was noted during the audit for
exposure dates of September 1, 3, 5, 7, 11 and 17, 2002, but all days
should be checked to ensure that the corrected spreadsheets are included
in the data books.	Summary sheets reviewed and revised as needed

	GC Printouts, General Comment:

Throughout the GC data, the individual performing the analysis is
identified only as “Inhalation Staff”. Since more than one
technician can perform the exposures, the responsible technician must be
identified in the data. It was noted that there is a dated signature on
the chromatogram cover sheet that is identified as a review signature.
Is the reviewer also the operator in all cases? Please address/clarify
in data.	Memo added to the Section 211 notebook

	GC Data, Chromatograms for September 3, 2002, Room Air (GC File
#7180001.D):

The chromatograms indicate that a test article concentration of 175
mg/m3 was found in the room air sample. Please address/clarify in data.
Note added to the GC printout that clarified that this was residual
material from the butane standard check.

	GC Data, Chromatograms for September 20, 2002, GC File # 71800117.D:

The sample name on the chromatogram is blank. It should indicate that
this was the low dose chamber.	Name added to the beginning of the
printout.  It already was in the methods information.

	GC Calibration Data (Book #1026):

For Method Development and Chamber Trials, Injection #1, please
clarify/define the asterisk that appears for peak #4 in the summary
table (no peaks at a retention time of 2.198).	Note added

	Distribution/Homogeneity Data, August 27, 2002:

Chromatograms for the 10,000 mg/m3 chamber appear to be mislabeled. The
Study/Sample Identification on the chromatograms indicates “Method for
Sampling Room Air in Lab PE102” The identification should indicate
that these were the mid-dose chamber.	Corrected

	Particle Size Data, August 28, 2002:

On both data sheets, the dose level units are incorrectly given as
mg/kg. This should be changed to mg/m3.	Corrected

	Particle Size Data, August 28, 2002, Control Chamber:

For impactor stages 6 and 5, the post sample filter weights appear to be
erroneous entries, based on the pre sample filter weights as follows:

Stage 6:

Pre weight = 85130 Post weight = 08130

Stage 5:

Pre weight = 85240 Post weight = 08230	Corrected

	Randomization Records, August 30, 2002:

At the top of the randomization printout, the animal number series
should indicate “IGL” (not “IFA”).	Corrected

	Animal Receipt Form:

In the first line, the date for uncrating and double-housing of animals
is documented as August 14, 2002. Wasn’t this actually performed on
August 13, 2002 (date of animal receipt)? Please correct/clarify as
appropriate.	Note added to data

	Documentation of Extreme or Unusual Animal Observations:

Animal IFA979 was sacrificed on August 28, 2002, apparently because it
was found out of its cage on the previous day. The reason for sacrifice
should be documented in the data.	Note added to data.  Animal number
corrected to IGL979

	Documentation of Extreme or Unusual Animal Observations:

There is an entry indicating that Animal IFA860 was sacrificed on August
28, 2002 due to its physical condition. Should the animal number be
IGL860?	Corrected

	Maternal Terminal Sacrifice Sheet for IGL974:

NOA should not be checked off on the form for dam necropsy observations.
A necropsy finding was noted for this animal. Also, the number recorded
for left corpora lutea count is not clearly written (can’t tell if it
is a 0 or 6). Please clarify in data.	Corrected 

CL count clarified

	Fetal Examination Data for IGL974, fetus #1:

It appears that due to the external observations noted for this fetus,
the head only was submitted for skeletal examination, whereas the
abdomen and thorax were used for visceral examination. This is not clear
on the Fetal Examination Record, since head is left blank, with no
explanation. Please include a notation in the data to clarify the
disposition of this fetus with respect to visceral and skeletal
evaluations.	Note added to the data.  A complete skeletal evaluation was
performed.

	Skeletal Examination Forms, General Comment:

Pagination of front and back of forms is inconsistent throughout. In
some cases, the front and back are both paginated as Page 1 of 1,
whereas in other instances, the front is paginated as Page 1 of 1 and
the back as Page 1 of 2. Some pages have been corrected to properly
indicate the front as Page 1 of 2 and the back as Page 2 of 2.
Pagination corrected

	Test Material Transfer and Usage Records:

The data indicate that the small tanks used in this study were labeled
1A, 2A, 3A and 4A. The SOP for small tank filling indicates that the
small tanks will be assigned the large tank number (i.e., 1), followed
by a sequential letter each time they are filled or re-filled (i.e., 1a,
1b, 1c, etc.). Since tank labeling for this study did not follow the
SOP, this should be addressed as an SOP deviation in the study records.
Note added to the memo on numbering the cylinders for this study that we
deviated from the Sponsor’s procedure.

	Test Material Transfer/Usage Log, Small Cylinder 1A, September 5, 2002:

The date is incorrectly written as September 4, 2002 in the
initials/date column.	Corrected

	Test Material Transfer/Usage Log and Small Cylinder Filling Procedure,
Small Cylinder 2A:

The amount dispensed is incorrectly recorded as 1180.1 on both forms.
The amount should be 11180.1.	Corrected

	Test Material Transfer/Usage Log, Small Cylinder 2A, September 26,
2002:

The change in weight after Group 2 exposure should be 132 g (not 13277.5
g).	Corrected

	Test Material Transfer/Usage Logs, General Comment:

Expiration date is not filled in on many of the forms.	Expiration dates
added

	Test Article Records, General Comment:

Reviewed by signatures and dates are not documented on many of the forms
including the Test Material Transfer/Usage Logs and forms for the
Preparing and Filling/Re-Filling New Small Gas Cylinders. Please make
sure that review is documented on all pages.	Forms reviewed

	

211(b) Toxicology Research Group

Gas+TBA Rat Developmental Toxicity Study Report 

Reviewer Checklist

  PAGE  25 	  DATE \@ "M/d/yyyy"  6/19/2009 

