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 the21 l(b) testing program must be provided to EPA for review, even
if they are negative or are not used in the statistical analyses.	Data
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

	In general, the API response to the independent peer reviewers'
comments appears to adequately address the reviewers' comments; however,
except for the deletion of Table 4-4, the study report available for
review has not yet been revised in accordance with the API response.
Although nonessential, it would be preferable to re-instate tables (such
as Table 4-4) into the text that accurately summarize fetal examination
data. On a related note, it would be helpful, though not essential, to
also include in the text summary tables of other developmental toxicity
data, showing some of the data from pages F-1 and F-2.	Table 4-4 deleted
at Sponsor’s request

	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 needed

	We regard the statistically significant decreases in maternal weight
gain and food consumption on GD 8-1 1 as evidence of slight maternal
toxicity. Weight gain is often a more sensitive indicator of maternal
toxicity than body weights alone; the lack of effect on GD-11
bodyweights is not sufficient reason to dismiss the effect on decreased
weight gains.	Abstract, summary, and discussion changed	i, 1-1, and 4-5

The significant reductions of maternal weight gain and food consumption
should be included in the Summary.	Added to summary	1-1

The statistical analyses 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.)	Fetal examination data reanalyzed	3-9 – 3-11 and 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: sternebral deficits
such as asymmetric, bifid, hypoplastic, and dumbbell shaped; rudimentary
and well-formed lumbar ribs; and ossification and cartilaginous
variations of thoracic centra su.ch as bifid, hypoplastic, and dumbbell
shaped.	Fetal examination data reanalyzed	3-9 – 3-11 and Appendix K

Some findings listed in Appendix H, as listed, do not clearly denote an
abnormality. The study authors should clarify what is meant by these
descriptors. Examples of such findings include:

- Individual ossification variations - sternebrae: ossification site (is
this referring to antra ossification site between sternebra 5 and the
xyphoid?)

- Individual ossification variations - vertebrae: cervical presacral

- Individual ossification variations - vertebrae: lumbar presacral

- Individual cartilaginous variations - rib adage: ossification site
Revised appendix H	Appendix H

Peer Reviewer – Schlesinger



The study was conducted in a scientifically sound manner and followed
appropriate protocols.  The conclusions are sound.  There were no
significant deviations from these protocols that would have affected the
outcome of the study.  There are, however, some generic concerns related
to statistical issues.	No change required

	When the Bartlett’s test indicated nonhomogeneity of variance the
investigators used a nonparametric analysis of variance.  An alternative
approach would be to use some transformation that would have resulted in
variance homogeneity and then a parametric analysis of variance could
have been used. This would have made the statistical analysis of all
data sets much more internally consistent and would not present any
potential for different types of tests having different degrees of
conservativeness in detecting any significant changes from air control.
In any case, the investigators should indicate which specific
statistical tests were used for which datasets.	No change; the
statistical symbols used indicate the type of analysis that was
conducted.

	The investigators also indicate that they performed statistical testing
with both transformed and nontransformed percentage data but that the
former where not reported since they were statistically significant. 
This implies that whichever procedure resulted in significance would be
reported.  This is an improper use of statistical techniques.  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.	Transformed data reported	F-2

It is also 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.	Justification added for
reporting the different levels of statistical significance	3-10

Specific Comments



p. 4-1. Gestation Body Wt. It is noted that there was a significant
decrease in mean body weight at the highest exposure level. However, it
is not clear whether this decrease was relative to control or to other
levels of exposure.	Clarified	4-1

p. 4-1. Food Consumption. Comment similar to that above related to mean
food consumption change.	Clarified	4-1

p. 4-6. Fetal Observations. Table 4-4 does not seem to present the data
on dumbbell shaped thoracic vertebral centra anlage described in the
text.	Table 4-4 deleted	4-3

p. 4-7. Exposure Data. In the paragraph on particle size, it is stated
that, ". . .there was no aerosol component to the chamber concentrations
and preventing calculation of the equivalent aerodynamic diameter and
the geometric standard deviation." The meaning of this statement is
unclear.  Particles would not be expected in the vapor-only atmosphere.
The term "preventing" suggests that particles were expected.	Revised	4-4

p. 4-8. Discussion. It is noted herein that, ". ..the incidence of this
finding in this study in all groups including the control was much
greater than the incidence in the historical control data." This
supports comments made on previous reports about the danger in using
historical control data to conclude issues of statistical significance.
No change required

	

Peer Reviewer – Goldsworthy



The study was properly conducted and scientifically sound.  None of the
protocol ‘exceptions’ noted adversely effected study results or
integrity.  Experimental procedures and exposures were appropriate and
adequate to test the potential developmental toxicity of baseline
gasoline.  Maternal toxicity was not observed and all dams survived to
scheduled terminal sacrifice.  There were no statistically significant
differences between control and BGVC-MTBE treated groups for uterine
implantation data, fetal body weights, and external, visceral and
skeletal observations.  No increase in dead or reabsorbed fetuses was
observed.  The most frequently noted observations during fetal
examinations were bifid centra of the thoracic vertebrae, rudimentary
lumbar ribs, and hypoplastic anlage of the thoracic vertebral centra. 
These reported changes are summarized on page4-6, table 4-4.  However,
the values listed need to be cross checked with appendix H-6 as some of
the values do not match up i.e. page H-6 lists different values for
thoracic centra hypoplastic.  The high incidence observed for thoracic
centra bifid is unknown but is most likely not related to treatment. 
Rudimentary lumbar ribs were frequently noted; the incidence was within
historical control range and thus not considered to be a
treatment-related finding.  However, it should be noted in text that
this finding exhibited a dose response trend by the Armitage test.  The
observed incidence of thoracic centra hypoplastic was unusual and noted
in text as not observed previously in the laboratory’s historical
control database.  This finding was higher in all treatment groups,
though not statistically significant.  Additionally, the incidence of
dumbbell-shaped thoracic vertebrae centric anlage was statistically
significant in the 2000 treatment group on both a fetus and litter
basis.  These data should also be included in Table 4-4; the fact that
this finding was also noted in the other exposed groups and not the
control group should be stated in the text (abstract, summary, and
discussion sections). Where as the lack of a dose response tends to
limit

an emphasis on these two endpoints noted above, these rare findings may
need to be further assessed in the context of the entire database API is
generating in their testing initiative.	Table 4-4 deleted.  Linear trend
test results added to the results.	4-1 and 4-2

Text/Editorial



Abstract (page i), last sentence:  add “with MTBE Vapor Condensate”
after “baseline gasoline”.	Added	i

Summary (page 1-2): it is incorrect to state “no other observations…
were statistically or biologically different” – the incidence of
dumbbell-shaped thoracic vertebral centra anlage were statistically
different (even though text deemed it not biologically significant).  It
is noteworthy to add that, although not dose related, this finding only
occurred in treatment groups.	Revised	1-1

211 (b) Research Group Reviewer 



General Comments



- For the final report, we will need sequential numbering in addition to
the A-1, A-2, etc. system so that we can be certain that we have all
report pages.	Repaginated

	-This report has historical control data, which is a great improvement
over not receiving it. Below, under Appendix K, there are several
suggestions that will make the data set more readily understandable.
There are many parameters relevant to the interpretation of historical
control data that are not shown here.	Historical control data revised
Appendix L

- The QA statement, even at the Draft Report stage, should list all of
the audited phases of the study and the audit dates. The report should
not be finalized without sponsor review of the QA statement.	QA
statement added	ix

- Tables should indicate the N from which statistics were derived. Even
with a nested analysis, the number of litters still represents the
statistical power of concern to reviewers.	N revised	G-1

- The statistical programs include linear trend analysis, but not a
criteria for ignoring statistically significant linear regression, which
appears to be the process in practice. There need to be explanations for
why it is run when intergroup differences are not statistically
significant and why it is ignored in these circumstances.	Linear trend
results added to the texts	4-1 and 4-2

- The interpretation of the study results is the responsibility of the
study director. In this study, there was a transient effect upon body
weight gain and food consumption early in gestation that isn’t
considered maternal toxicity in this report; this was quite similar in
duration and degree to that which= considered to be maternal toxicity in
one of the earlier reports from EMBSI. Separate from the report, it is
appropriate for API to request that the study director provide a
response for the difference in interpretations, since ultimately the
entire program will be reviewed, and API may be expected to explain the
findings as a whole.	Abstract, summary, and discussion revised	i, 1-1,
and 4-5

Specific Comments (by page and section)



Page i, Abstract:

Line 8-10: The abstract should focus only upon the information necessary
for interpretation of the study's purpose. Frequently noted findings
that are not associated with exposure should be left out of the
abstract, although discussed in the results. Please delete the sentence
beginning, "The most frequently noted ..." and the word "However"
starting the subsequent sentence.	Sentence removed	i

Line 13: Typographical: delete second period after "response."	Removed	i

Page ii, Table of Contents

"Justification of Dose Selection" is on page 2-2, not 2-1.	Corrected	ii

Page iii, Table of Contents	No action required

	"Visceral Observations" is page 4-5, not 4-4.	Table of contents revised
iii

"Skeletal Observations" is page 4-6, not 4-4.	Table of contents revised
iii

"Exposure and Chamber Conditions" is page 4-7, not 4-5.	Table of
contents revised	iii

"Discussion" is page 4-8, not 4-6.	Table of contents revised	iii

"Protocol Exceptions" is page 4-9, not 4-7.	Table of contents revised
iii

Perhaps the lab can indicate if these page differences are due to
different print settings. If so, let's standardize them to whatever
margin settings are required or APl's preference for submissions.	No
action required

	List of Tables:

Table 4-2 & 4-3 are in error; these descriptions fit Tables 4-5 & 4-6.
Tables 4-2 through 4-4 are missing from this list.	Tables renumbered	iv

Page viii, QA Statement

As noted above, this should list the phases and dates of inspection.	QA
statement added	ix

Page 1-1, Summary 

First paragraph, line 5: There should be a comma after "(air control)."
Added	1-1

First paragraph, lines 6-7:

"The Sponsor selected these doses." sounds like blame and does not
provide adequate justification, since the study director and performing
laboratory still have responsibility to run a valid study. Better
wording would indicate what the Sponsor based these exposure levels
upon; this is written in the following sentence regarding safety
considerations, but there is no comment about why the 2K level was
considered a likely NOAEL, or other completed/ongoing tox studies, or
the testing program.	Wording revised to that supplied by the Sponsor	1-1

Last paragraph, end:

Please add in whether these increases were within the historical control
range.	Added	1-1

Page 1-2, Summary

There is no reason to include a table of fetal weights in the summary
when this is not a parameter that is interpreted as affected by
exposure. This table should be deleted.	Table removed	1-2

Page 2-1, Justification for Selection of Test System:

I believe that the testing requirement for this program is the EPA 1994
reference, not 1996. There is no requirement for the rat in the
developmental toxicity guideline, although in practice one would need to
provide very strong justification for not using it.	Section revised	2-1

Page 2-2, Justification of Dose Selection:

Same comments as for the summary: adequate justification for dose
selection should provide the reasoning for all of the exposure levels
selected.	Wording revised to that supplied by the Sponsor	2-2

Page 3-1, Characterization of the Test Substance:

All of this information as described in the first paragraph should be
provided with the report as an appendix (Appendix J).	No change.  A
separate report is being prepared for the characterization.	3-1

Page 3-4, Feed and Water:

Copies of the feed and water analyses should be included in the report
as appendices.	Feed and water analyses added	Appendix M

Page 3-7, Euthanasia:

The method of euthanasia for viable fetuses must be included.	Method of
euthanasia added	3-8

Page 3-8, Statistical Analysis:

Second paragraph: If transformations are performed, they need to be
reported whether statistically significant or not.	Transformations
reported	3-9, 

Third paragraph: The lab needs to provide explanations for: (a) why
linear regression is performed when there isn't a difference between
groups; and (b) what criteria is used to ignore statistically
significant linear trend performed in such situations.	Linear regression
runs automatically every time the data is analyzed. It is also part of
the protocol.

Linear regression added to the results.	No change in report

4-1 and 4-2

Page 4-1, Gestation Body Weight:

Although not statistically significant, it is of interest that one of
the females in the 10K group, IGK434F, lost 47g during GD 5-8, with a
subsequent rebound of 66g during GD 8-1 1. The animal next in line on
page C-5 and C-9, No. IGK437, gained 50g during GD 5-8 and lost 279
during GD 8-1 1. Although there were other substantial weight gains,
there were no other weight losses anywhere near these amounts. This kind
of weight loss is usually associated with some sort of observation,
since 479 can't really disappear so quickly without some dehydration,
reduced stool, reduced food consumption, increased food spillage,
problems with teeth being noted, yet there are no observations for
either animal. Food consumption for each animal was normal during both
intervals. Please check the individual body weight data to confirm that
the correct entries are shown in Appendix C; if the body weights were
inadvertently reversed, both animals would have typical results.	The
weights in Appendix C agree with the raw data.  It appears that the body
weights were recorded for the wrong animals in the data but there is not
enough evidence to support switching the weights in the data. 
Therefore, I wrote a memo to file stating that these weights were
dubious and would not be used in the statistical analyses.

	GD 0-21C is shown on page C-2 as being statistically significant for
trend analysis; this should be addressed here.	Added	4-1

Page 4-1, Gestation Food Consumption:

Food consumption for GD 5-8 is shown as "AL+" for statistical
significance on page D-1, the same symbols as for GD 8-1 1. The
statistical significance for this interval needs to be discussed and
evaluated or discounted with justification.	Added a description of the
statistical significance among the treated groups	4-1

Page 4-2, Uterine Implantation Data:

Resorptions, Fetus/lmplantation, and Resorption/lmplantation are all
shown as "L" on page F-1; these need to be addressed in the text.	Linear
trend results discussed	4-2

Page 4-3, Fetal Body Weight, continued:

The number of litters per group needs to be represented in the table.
This is repetitive of the data in Appendix G, page G-1, and really
doesn't need to be shown in the results except in situations where there
is a need for a detailed explanation of the findings. That isn't the
case here.	Table deleted	4-2

Page 4-4, Table 4-2:

This table should be deleted, since none of the findings are attributed
to exposure, nor do they need detailed explanation.	Table deleted	4-2

Page 4-5, Visceral Observations:

See previous comment; recommend deleting this table also.	Table deleted
4-3

Page 4-6, Skeletal Observations:

Text, first paragraph: The statistically significant trend analysis for
rudimentary lumbar ribs must be addressed. Please also compare the data
to the laboratory historical control range for "dumbbell-shaped thoracic
vertebral centra anlage," for dismissal. Also, "dumbbell” is
misspelled.	New statistical analyses does not have linear trend
analyses.  Comparison to historical control added.	4-3

Table 4-4:

None of these entries are displaying results attributed to exposure,
therefore this table should be deleted. If kept, it must be accurate and
correspond exactly with the findings in Appendix H. The entries do not
appear to be correct for any exposure level for "Thoracic Centra Bifid"
and do not appear to be correct the 2K and 10K groups for "Thoracic
Centra Hypoplastic." If kept, it must also include the sternabrae, rib
(rudimentary lumbar, statistically significant dose-response trend) and
the discounted dumbbell findings.	Table 4-4 deleted	4-3

Page 4-8, Discussion:

Second paragraph: More detail than is needed for findings that were not
statistically significant and were not attributed to exposure. If
nonsignificant findings are discussed here, it would be appropriate to
also discuss the statistically significant findings that were not
attributed to exposure.	Removed the non-statistically significant
observatiuons	4-5

Page 4-9, Protocol Exceptions:

Chamber temperature and humidity: The actual values need to be presented
here, as well as the protocol specified ranges. This level of detail is
provided for the chamber concentrations and animal room
temperature/humidity. It is nice to have an explanation of why the
chamber conditions were out of range, but it is important that the
actual information be provided.	Section revised and all values are
presented in Apppendix I	4-6, I-24 – I-31

Pages 6-1 & 6-2, References:

Was it the author's intent to have the references print out in landscape
instead of portrait format? Portrait format, as for the prior text,
would be preferable.	Problem corrected.  References print in portrait
format.	6-1, 6-2

Appendices



Appendix B

Page B-8: Animal No. IGK363F is missing the GD 21 observation: Is this a
typo, or does this need to be footnoted as not taken? This was not
listed as a protocol deviation, but should be if it was inadvertently
not performed and/or not recorded. Also, if this is missing data, then
the GD 21 "No observable abnormalities" summation of 24 for the 20K
group needs to be corrected to 23.	Observation added to report	B-8

Appendix C

Page C-1, C-2: Delete "Run 2" on each page.

As requested above, please confirm the data for animals no. IGK434F and
IGK437F.	“Run 2” removed.

Note added to IGK 434 and IGK437 data	C-1, C-2

Appendix D

Page D-1: Delete "Run 2."	“Run 2” removed	D-1

Appendix G

Page G-1: The number of litters per group needs to be included in this
table. Even in a nested design, providing the number of fetuses in the
data set is inappropriate.

Pages G-2 onwards, Individual data tables: The mean female fetal weight
and mean male fetal weight need to be provided for each litter. The
information that is provided, Mean Fetus Weight, is not the parameter
used for calculating the values that are compared for intergroup
differences.	Number of litters added

Mean male and female weights provided for each litter.  Mean litter
weights deleted	G-1





Appendix H

Page H-16: An explanation needs to be provided for why fetal numbers 11
and 13 were arbitrarily assigned.

Page H-19: An explanation needs to be provided for why fetal numbers 7
and 13 were arbitrarily assigned.

Page H-21: An explanation needs to be provided for why fetal numbers 1,
3, 7, 9 and 13 were arbitrarily assigned.

Page H-31: An explanation needs to be provided for why fetal numbers 8,
12, and 14 were arbitrarily assigned.

Page H-33: An explanation needs to be provided for why fetal numbers 1,
5, and 7 were arbitrarily assigned.

Page H-44: An explanation needs to be provided for why fetal numbers 6,
12 and 14 were arbitrarily assigned.

Page H-48: An explanation needs to be provided for why fetal numbers 1
and 7 were arbitrarily assigned.

Page H-55: An explanation needs to be provided for why fetal numbers 5,
7, 11 and 15 were arbitrarily assigned.

Page H-59: An explanation needs to be provided for why fetal numbers 3,
9, and 13 were arbitrarily assigned.

Page H-63: An explanation needs to be provided for why fetal numbers 3,
5, 7, 9, and 15 were arbitrarily assigned.

Page H-69: An explanation needs to be provided for why fetal numbers 2
and 10 were arbitrarily assigned.

Page H-72: An explanation needs to be provided for why fetal numbers 4
and 8 were arbitrarily assigned.

Page H-83: An explanation needs to be provided for why fetal numbers 2,
6, 12, and 14 were arbitrarily assigned.

Page H-90: An explanation needs to be provided for why fetal numbers 5,
7, 9, 13 and 17 were arbitrarily assigned.

	Footnotes revised	H-16, H-19, H-21, H-31, H-33, H-44, H-48, H-55, H-59,
H-63, H-69, H-72, H-83, and H-90



Appendix H

Page H-91: An explanation needs to be provided for why fetal numbers 1,
7, 11, 13, 15, and 17 were arbitrarily assigned.

Page H-92: An explanation needs to be provided for why fetal numbers 3,
7,9, and 11 were arbitrarily assigned.

Page H-94: An explanation needs to be provided for why fetal numbers 2,
4, 6, 8, 10, 12, 14, and 16 were arbitrarily assigned.

Page H-95: An explanation needs to be provided for why fetal numbers 2,
12, 14, and 16 were arbitrarily assigned.

Page H-96: An explanation needs to be provided for why fetal numbers 2,
4, 6, 8, and 10 were arbitrarily assigned.

Page H-99: An explanation needs to be provided for why fetal numbers 5,
7, and 15 were arbitrarily assigned.	Footnotes revised	H-91, H-92, H-94,
H-95, H-96, H-99



Appendix K

First of all, the font should be no smaller than size ten.

Secondly, the reader will need to place the historical control data into
the context in which the data were collected, requiring the following:

strain and source of animals

number of animals in each control group (number of litters and number of
fetuses)

dosing route (if oral, a vehicle should be included. If no dosing,
indicate if the animals were sham-treated or not)

date of the data collection (month/year)

since EMBSI is in a new facility, the studies conducted in the old
facility must be asterisked or otherwise distinguished. Facility affects
environment.

It needs to be clear if blank spaces indicate that the observation
wasn't found, or wasn't evaluated.

Viable litter size, implantations, corpora lutea, % pregnant should all
be in the Historical Control appendix.

The data should be organized according to external, visceral, and
skeletal, as well as variations and malformations.

There are a number of $ signs under study numbers on pageK-1, without a
footnote to define the use of this symbol. Same for the use of F and L.

Page K-2: Cyst - F, Cyst - L: what organs are these cysts on?

Page K-5: The number of litters per study needs to be included. Also,
the study numbers can't be seen on this page.

The individual historical control data should be preceded by a summary
of the historical control data that includes the average value for the
historical control data and the range (maximum, minimum) for each
endpoint.  If this is too difficult for the individual skeletal
observations, then it should at least be provided for total
malformations, variations, fetal weight, and the reproductive parameters
noted above.	

Font size change

Historical control data revised

Historical control data revised

Footnotes added

Cyst is no longer relevant in the revised historical control data

Number litter added, study numbers added

Historical control data revised

	Appendix L for all comments

211 (b) Research Group QA/QC Reviewer



The following items require further consideration:



Page i, Abstract, lines 10 – 12:

The text states that the incidence of dumbbell-shaped thoracic vertebral
centra anlage was statistically significantly increased in the 2000
mg/m3 group on both a fetus and litter basis. The statistics printouts
reviewed at the time of this audit indicated that this finding was
significant on a fetus basis, but not on a litter basis and that the
10,000 mg/m3 group was significant on a litter basis for this finding.
This should also be addressed on Page 1-1, Summary and Page 4-6,
Skeletal Observations. It should be noted that corrections to the
reported skeletal findings are required, as discussed later in this
audit report. These changes may further affect statistics, summary
tables and text.	Abstract, summary, and results revised based on the new
statistics	i, 1-2, 4-3

Page vii, 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.	Space for Sponsor signature added	vii

Compliance Statement:

Since it was the sponsor’s responsibility to maintain the method of
synthesis, fabrication, or derivation of the test fuel, and this has not
been completed, it must be included in the sponsor’s compliance
statement.	This is not a compliance issue.  The Sponsor has provided the
location of the information and a statement has been added in the test
substance section	3-1

Page vii, Compliance Statement:

As noted in item #1 under Additional Data Issues, of this report, some
analytical data were not collected in a GLP consistent manner and a GLP
consistent data trail was not kept. This needs to be included and
detailed in the compliance statement.	GLP deviation added	vii

Page viii, Personnel:

Is a compound preparation supervisor an appropriate person to have
listed under personnel? Did this person actually work on this study?	No
change.  It is an SOP requirement.

	Page ix, QA Statement:

The QA statement needs to be completed.	QA statement added	ix

Page 1-2, Summary:

“SUMMARY (CONT’D)” appears twice on the top of this page. Also,
Table 1-1 Mean Fetus Weights is presented in the summary. Since there
were no test substance-related effects on Fetal Weight, is it necessary
to include this table in the Summary? The table title indicates Mean
Fetus Weights and Mean Litter Weights. The values reported appear to be
mean fetal weights, only.	Extra “SUMMARY (CONT’D)” removed,

Table 1-1 removed

	1-2

Page 3-1, Test Substance:

The test substance is described as a clear amber liquid in the report.
The description documented on the Test Material Use Log is “clear
colorless liquid. Please address this apparent discrepancy	Description
revised	3-1

Page 3-1, Test Substance:

The first date of test substance receipt should be December 11, 2000
(not 2001).	Corrected	3-1

Page 3-1, Test Substance:

A total of 16 small tanks were received in the two shipments. Only 11 of
these are indicated in this section of the report text. If these tank
numbers are intended to reflect those actually used on this study, the
text should indicate tank nos. 1, 3, 6, 7, 11 – 14 and 16. Please
recheck and clarify. Tank 10 was also used for particle size recheck.
Tank numbers changed	3-1

Page 3-1, Test Substance:

There was also a big tank (#17) received 6/14/01. Should this be
included in the test material inventory reported?	No change.  Tank not
used for this study.

	Page 3-1, Test Substance:

According to Chevron, tanks 1-8 were received 12/11/00, then 1-8 were
received 4/9/01 and EMBSI labeled them, 1-8 and 9-16. Please clarify. As
well, for the first shipment, Chevron had labeled the tanks 1 of 9, 2 of
9, etc., yet only sent 8 tanks. There needs to be documentation of this
in the file to clarify exactly how many were sent, received, etc.	No
change to our data or report.  Our data is clear that we received 8
tanks each on 12/11/00 and 4/9/01.

	Page 3-1, Test Substance:

Expiration dates of December 2005 and April 2005 are given for the two
batches received. Are there stability data to support these dates?	The
expiration dates are per EMBSI SOPs.

	Page 3-1, Materials and Methods, Test Substance Characterization:

Is the stability of the test material being tested by EMBSI as part of
the characterization analysis? This should be indicated as being
reported separately. The analyses performed by EMBSI as part of this
developmental study were originally designed to determine stability for
the test period and under the conditions of use. This should be
indicated.	Added to the report	3-1

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. The number of components identified should
probably be indicated for reproducibility purposes.

The next paragraph of this section should indicate that this analysis
was done to determine component proportions of the test material
atmosphere compared to the liquid test material.	Number of components
added

Section revised	3-2

Page 3-2, Analytical concentration, second paragraph:

It appeared that at least some of the sorbent tube samples were frozen
in the inhalation lab and transported to the analytical lab at a later
date for analysis. These sample handling procedures should be indicated
in the methods.	Section revised	3-2

Page 3-5, Environmental conditions:

These are the protocol specified ranges, not the actual ones. This
should be clarified.	Actual ranges added as applicable	3.5

Page 3-7, The Test Atmosphere:

The last sentence needs to indicate that the daily mean exposure
concentrations were within the + 10% of target exposure levels, “with
the exceptions noted in the deviations section of this report.”
Sentence revised	3-8

Page 4-3, Table 4-1:

The table title indicates “Mean Fetus Weights and Mean Litter
Weights”. The data presented are only mean fetus weights.	Table 4-1
deleted	4-2

Page 4-6, Table 4-4:

The reported incidences for Vertebrae, Thoracic Centra Bifid (all
groups) do not agree with those presented in Appendix H Incidence Table.
Please verify and correct as needed. Also, the incidences of Vertebrae
Anlage, Thoracic Centra Hypoplastic for the 2000 and 10,000 mg/m3 groups
need to be rechecked, as they do not agree with those presented in
Appendix H incidence table. It should be noted that several corrections
to the Appendix H individual data may be needed (see items 32-45 below)
and the text tables may need further corrections after the individual
tables are revised. Please verify all text tables after individual
animal table corrections are made.	Table 4-4 deleted	4-3

Page 4-7, last paragraph:

Oxygen levels in the chambers ranged from 20.7 to 20.9.	Corrected	4-4

Page 4-6, particle size paragraph:

Should the text indicate that particle size determination for the
control and 20,000 mg/m3 chamber detected NO particles? The rest of the
paragraph needs clarification.	Section revised	4-4

Page 4-9, Protocol Exceptions:

For chamber temperature and humidity deviations, should ranges be
included to define the extent of the temperature and humidity
excursions? Also it would be helpful to indicate how many days the
values were outside of the protocol-specified ranges in each test group.
Protocol exceptions revised	4-6, I-24 – I-31

Page 6-2:

The Staples reference should be moved so it is below the section header
of “References (Cont’d)”	Corrected	6-2

Appendix B, Page B-8 for Animal No. 363F, a “+” should appear in the
day 21 column.	Added	B-8

Appendix B, Page B-9:

Under Animal No. 441F, “Discharge” is mistyped.	Corrected	B-9

Appendix C, Page C-2:

For GD 0-21, the statistical notation should be “A-L” not “K-J”
(parametric analysis was appropriate).	Corrected	C-2

Appendix D, Pages D-1 through D-5:

Means and Standard Deviations are reported on the summary table (page
D-1) for Food Consumption intervals GD 5-20 and GD 0-21. No individual
values for these intervals are presented in the Individual Data Tables.
Should these values be reported?	Individual values added	D-2 – D-5

Appendix F, Page F-5:

For 10,000 mg/m3 animal no. 361F, the number of implants was greater
than the number of corpora lutea. These counts were not verified by
another individual at the time of examination as is required by EMBSI
SOP. This should be addressed/acknowledged as an SOP deviation in the
study records.	Deviation acknowledged in raw data

	Appendix H, Page H-1:

For the 10,000 mg/m3 group, it appears that the number of fetuses
examined externally should be 369 (187 males and 182 females), not 370.
Corrected	H-1

Appendix H, Page H-6:

According to the statistical printouts reviewed at the time of the
audit, the litter incidence of “Vertebrae Anlage, Thoracic Centra
Dumbbell Shaped” (i.e., 3 affected litters) for the 2000 mg/m3 group
was not significantly different than control by Fisher’s Exact test.
This finding is noted as being significant on the incidence table. The
statistics printouts did indicate that the litter incidence for this
finding in the 10,000 mg/m3 group (i.e., 5 affected litters) was
significantly greater than control. This statistical significance is not
denoted on the incidence table. Please verify and correct table as
appropriate.

See also, item no. 1 above, for text corrections related to these
findings. It should be noted that data entry errors for skeletal
findings were noted in the individual animal data tables presented in
Appendix H. These findings are discussed below in this audit report.
Statistical analyses may need to be redone and report text and tables
corrected accordingly.	Table revised based on the revised statistics	H-6

Appendix H, Page H-10:

For animal 332F, Fetus 3, it cannot be determined from the skeletal exam
raw data whether finding “c” (Vertebrae T11-12, bifid centra) was
crossed out or not. Please verify and correct/clarify report and data as
appropriate. Additionally, there appears to be a notation on the bottom
of the raw data page that indicates that finding “b” is considered a
malformation. This notations appears to be erroneous and should be
corrected on the raw data.	Data clarified

	Appendix H, Page H-17:

Please verify skeletal findings for animal 343F, fetus nos. 2. and 10.
It appears from the skeletal exam data that a finding of “Vertebrae
T-11 bifid centra” should also be reported for these fetuses. For
fetus no. 12, it appears that a finding of “Vertebrae T12, bifid
centra” should be reported.	Table corrected	H-17

Appendix H, Page H-29:

For animal 435F, fetus 4, according to the skeletal exam raw data, it
appears that finding “a” should be for thoracic vertebrae T11, 12.
Table corrected	H-29

Appendix H, Page H-33:

For animal 338F fetus 3, it appears that finding “b” should indicate
thoracic vertebrae T11, 12, not T8, 9. For fetus nos. 9 and 11, the
skeletal findings need to be verified. The reported findings do not
appear to agree with what is documented in the raw data.	Table corrected
H-33

Appendix H, Page H-37:

For animal 364F fetus nos.5 and 11, the skeletal exam raw data show that
vertebral findings “d” and “e” involved T11, 12. Only T12 is
indicated in the report appendix.	Table corrected	H-37

Appendix H, Page H-38:

For animal 358F fetus 5, the skeletal exam raw data show that the
vertebral findings “b” and “c” involved T11, 12. Only T11 is
indicated in the report appendix.	Table corrected	H-38

Appendix H, Page H-61- For animal 357F, “anlage” is mistyped in the
notation for finding “g”.	Corrected	H-61

Appendix H, Page H-66:

For animal 409F, all reported skeletal findings should be verified. The
reported findings do not agree with what is documented in the skeletal
exam raw data.	Table corrected	H-66

Appendix H, Page H-67:

For animal 371F, “asymmetric” is mistyped in the notation for
finding “b”.	Corrected	H-67

Appendix H, Page H-75:

For animal 434F fetus 7, the head was inadvertently removed and no
skeletal exam was performed. This should be documented as a protocol
deviation in the study data and report since the fetus did not receive
the protocol-required examination.	Memo added to the data

	Appendix H, Page H-84:

For animal 325F, finding “a” should be reported for fetus 7, not
fetus 8.	Corrected	H-84

Appendix H, Page H-85:

For animal 375F fetus 6:

an additional finding of “Vertebrae, T11, bifid centra” should be
reported.	Corrected	H-85

Appendix H, Page H-90:

For animal 340F fetus 13, skeletal finding “b” and “c” should be
deleted from the table. A March 27, 2002 data correction on the skeletal
exam sheet showed that these findings were recording errors.	Corrected
H-90

Appendix H, Page H-99:

For animal 436F, “asymmetric” is mistyped in the notation for
finding “b”.	Corrected	H-99

Appendix H, General Comment:

Due to the corrections discussed above in findings 32-45, statistical
analyses for skeletal findings will need to be redone and all relevant
summary tables, text tables and text sections will need to be revised
accordingly to reflect the corrected data.	Data reanalyzed	Appendix K

Appendix I, Page I-2, Test Atmosphere Generation:

The schematic of the test atmosphere generation and exposure system is
presented in Table “I-1”.	Corrected	I-2

Appendix I, Page I-2, Chamber Environmental Conditions:

Should other environmental conditions such as light, noise or oxygen
levels be discussed in this appendix?	No change; described in the main
body of the report

	Appendix I, Page I-3, Analytical Procedures:

The sample chromatograms, identified as Figure I-4 do not appear to be
included in the appendix. Also, the GC operating conditions are
presented in Table I-2, not I-4.	Sample chromatogram removed from list
I-3

Appendix I, Page I-4:

Should particle size results be discussed in this appendix?	No change. 
The particle size data is discussed in the results.

	Appendix I, Table 1-2:

Some of the GC operating conditions do not agree with what is documented
in the analytical data. Please verify Supelco no, inlet temperature,
oven temperature and inlet flow cc/min and revise table as necessary.
Table I-2 corrected	I-9

Appendix I, Figure I-3:

“Response” is mistyped in figure title.	Corrected	I-10

Appendix I, Table I-3, Page I-12:

“Summary is mistyped in table title. Also, should the mean relative
humidity for the control chamber on September 1, 2001 be changed to 73%?
This difference may be due to rounding variation.	No change; the
unrounded value is 71.8

	Appendix I, Table I-3, Page I-14:

For the 2000 mg/m3 on September 5, 2001, the hour 5 analytical
concentration was considerably high (3897 mg/m3). There did not appear
to be any discussion or acknowledgement of this excursion in the study
data or draft report. Please address.	Added to the protocol exceptions
4-6

Appendix I, Table I-3, Page I-16:

Footnote “a” should be indicated for the September 7, 2001 hourly
concentration data rather than September 6. Also a footnote explanation
should be added to the bottom of the page to indicate that this refers
to the suspected problem with the tank dip tube.	Corrected and footnote
added	I-17

Appendix I, Table I-4:

“Approximately” is mistyped in the notation at the bottom of the
page.	Corrected	I-19

Appendix I, Page I-21, Should there be an explanation for the wide range
of light intensity readings in Room PE103 (ranges from 7.8 to 43.6 fc).
No change.  We have no documented reason for the range.  The room
condition data shows no abnormalities.

	Appendix I, Table I-6, Page I-22:

Sample flow rate and sample duration are reported as 4 lpm and 5
minutes, respectively. I found documentation of 3 lpm and 3 min. Please
verify.	Corrected. 	I-21 and I-22

Appendix I, Table I-6, Page I-23:

The reported particle size data for the 20,000 mg/m3 chamber appears to
be that from the particle size recheck performed on January 31, 2002,
rather than the original analysis performed on August 9, 2001. The
reason for the performance of the repeat analysis should be discussed in
the report.	Added to the results	4-4

Appendix I, Table I-t, Page I-23:

Sample duration is reported as 5 minutes. I found documentation of 15
minute sampling duration. Please verify.	Corrected	I-22

Appendix J:

Pagination of the analytical chemistry report begins with page J-23?
Corrected	J-1 – J-3

Appendix J:

There is not description of how the charcoal tube samples were taken and
handled from the time of collection to the time of analysis. This
procedure needs to be described somewhere in the report. It appears that
some samples were desorbed and then frozen until analysis. If this is
true, this needs to be stated in the report. A description of how they
were stored prior to receipt by the Analytical Chemistry Lab is needed
as well.	The collection and storage until transfer to Analytical
Chemistry is described in the test Substance section of the report.  

It does not appear that the aliquots were frozen by chemistry before
analysis.	J-1 and 3-2

Appendix J, Table J-1:

The table should indicate that these are area percent values.	Added to
table	J-3

Additional Data Issues:



A GLP consistent data trail was not maintained for the charcoal tube
samples collected and stored in the animal area before being transferred
to the analytical lab.	Deviation added to the data and the report

	Since some charcoal tube samples from chambers were apparently frozen
prior to analysis, a stability study should be done to verify that these
samples are stable in the freezer for this period of time. Also, since
the analytical laboratory sometimes desorbs samples with CS2, then
freezes them until analysis, the stability of these samples needs to be
shown.	No change.  These are standard methods for handling these types
of samples.  Additionally, the data from these samples shows that the
samples were stable.

	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.

	The Miran was used on September 16, 2001 to check the concentration in
the 2000 mg/m3 chamber at the first hourly interval, since an extremely
high concentration was reported by the GC. A memo to the study records
indicates that the Miran data indicated that the chamber concentration
was acceptable. However, the actual Miran data which supports this
statement was not found in the study records. If possible this data
should be located and placed in the study file. If the Miran data are
not available, then this also needs to be addressed.

Additionally, complete documentation of what was done to the GC to
resolve the problem should be included in the study data.	No data exists
for the MIRAN.  Memo added to the data.

Nothing was done to resolve the problem.  The problem resolved on its
own.  The Groups 3 and 4 analyses were fine.

	Maternal Sacrifice Sheet for Group 1 animal 402F:

The balance identification is not checked off for uterine weight.	Note
added to data

	Maternal Sacrifice Sheet for Group 3 animal 356F:

Balance identification for pup weights is not checked off.	Note added to
data

	Uterine Examination Data for Group 1 animal 388F, fetus 10:

The sex of the fetus is changed from female to male without proper GLP
error correction.	There is a footnote in the data that indicates the sex
was re-evaluated on September 14, 2001 by GT

	Maternal Sacrifice Sheet for Group 3 animal 361F:

The number of implants (13) was greater than the number of corpora lutea
(12), but these counts were not verified by a second person at the time
of collection as required by EMBSI SOP. This is not acknowledged as an
SOP deviation, although it is noted as an oversight on the data
collection form.	SOP deviation acknowledged on the data sheet.

	Particle Size Data:

The pre sample weight for Impactor stage 4 was recorded as 845100. It
appears that this should have been 84510.	Corrected

	Particle Size Data:

The initial particle size analysis for the 20,000 mg/m3 chamber is not
reported. This analysis was redone on January 31, 2002 to recheck sample
no. 1. An memo which details the reason for the repeat analysis should
be included in the study records for clarification.	Memo added to the
data

	Clean Air Act Program Book, memo dated August 2, 2001, regarding test
material handling:

Incorrect MRD numbers are documented for BGVC and MTBE. These numbers
should be 00-695 and 00-713.	Corrected

	Test Material Usage Records:

The test material used for method development is not documented on the
usage logs. Therefore, a complete inventory of test material used is not
available. Also confirmation of test material description was performed
on April 18, 2002 by dispensing 10 mL samples from “out of service”
cylinders. This dispensation was also not documented on the usage logs.
Method development/trial usage of test substance is documented for tanks
11, 13, and 16.

Memo added to file regarding the 10 mL sample.

	Skeletal Examination Data:

For 2000 mg/m3 animal 452F, fetus 17, the finding of vertebrae12, bifid
centra should be documented for thoracic vertebrae, not cervical.
Corrected

	Skeletal Examination Data:

For control animal 332F fetus 3, notation 1 at the bottom of the page
says “considered a malformation”, yet all findings are variations.
This notation should be deleted or otherwise clarified. Also, was the
finding of “9B” (bifid centra) for T11-12 supposed to be deleted as
per the March 27, 2002 re-examination? This finding is still included in
the report table.	Note added to data

Bifid centra removed from report

	

211(b) Toxicology Research Group

Gas+MTBE Rat Developmental Toxicity Study Report 

Reviewer Checklist

  PAGE  1 	  DATE \@ "M/d/yyyy"  4/4/2008 

