Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 227
3
BRIEFLY FOLLOWING A CHEMICAL
TH ROUGH AN I LLUSTRATIVE SYSTEM
This chapter presents a simplified overview of the operation of an
illustrative priority-setting system. It assumes that the system has
been fully designed and implemented and is ready for operation. Although
it uses the decision rules from an optimization model, it does not
attempt to show how those rules were derived (see Appendix B).
As currently envisioned, the principal operations of the system occur
in three stages (Stages 1, 2, and 3), with successively increasing costs
per chemical processed and decreasing number of chemicals needing
scrutiny. This chapter also discusses the formation of the universe of
chemicals of concern (Stage 0) and the actual testing of chemicals (Stage
4~. Figures 3 through 5 illustrate the steps in each stage. The reader
is encouraged to refer to the figures while reading the text.
The system can treat a chemical only if it is in the universe of
chemicals of interest (or is nominated for Stage 3 by an agency or
individual). The universe must be assembled as a list containing unique
identifiers for each chemical, such as its number in the CAS Registry.
me universe is made up of several lists, the largest and most accessible
of which is the select universe compiled by the Committee on Sampling
Strategies. Other sources are the environment (which could be
characterized in the sampling programs of EPA); food ingredients,
including natural constituents, contaminants, and conversion products
(which might be available from lists maintained by FDA); and the growing
group of chemicals on which premanufacturing notices have been filed with
EPA. The additional lists must be acquired or constructed, and
identifiers and associated data added; then these new lists must be
merged with the existing lists by eliminating duplication and sorting the
chemicals into the order in which they will be maintained on the master
list.
To illustrate the operation of the remaining stages, it is useful to
have an example chemical, rather than dealing in the abstract with an
unlimited range of possibilities. We have chosen bisphenol A as our
example chemical, although at some places in our analysis we have
speculated about its properties to create the best illustration.
STAGE 1
Bisphenol A is identified by its CAS Registry number, 80057. This
number links the priority-setting system with the various automated data
bases on which it depends. In Stage 1, only a few fragments of
information can feasibly be retrieved and processed. Among them are the
following:
227
OCR for page 228
STAGE
DATA INPUTS
DATA PROCESSING
DECISIONS
OUTPUTS
CSS Select Universe
Pre-Manufacturing Notices
Environmental Occurrence
Food I ngredients, etc.
:-:~:-:-:-:-:-:~:-:~:~:~:-:~:~:~:-:~:
·- ~
.....................................
~ Cunmjpvelrese
· .....
, ..................
:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:~:
.............. -
...................................
.............. ~ ~
- ·.~................
.................
· ~.~.~. - .~.~. - .~.~.~.. - -
· ~ ................
:--~-~------~-~-~-~----~-.-.-.-.-.
............. :~:~:~:~:.
· ~ ....~. - .~. - .-.~. - . - .~. - . - - -
·:-:-:-:~:-:-:-:-:-:-:-:-:-:-:-:::::
............. :.:~:~:~:.
:::::::::::::::::::::::::::::::::::::
Computer
Computer
Bisphenol A
t
...................
·:-:~:-:-:-:-:-:~:~:~:~:-:-:-:-:-:-:
.. OC
·:~: ·. ~- . ·
·....................................
:-:-:-:-:~:-:-:-:-:-:-:-:-:-:-:-:~:~:
.............. ~
..................
:-:-:-:-:~:-:-:~:-:-:-:-:-:~:-:-:-:-:
:-:-:-:~:~:-:-:~:-:-:-:~
.................
·.~...~............................
,..................
~-
Eliminate
Dupl icates
Sort
Universe of
Chemicals
FIGURE 3 Stage 1 of illustrative system.
228
TSCA Inventory
CTDE Data Base
RTECS, EMIC, ETIC
SANSS
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _-_
Computer
~1
STAGE 1
Compile Data
- Production Volume
Use Class
Toxic Effects Reported
Toxic Class (SAR)
Decision
Rules
Stage 1 Processor
L M H
·:-:-: ·:-:-: ·:-:-:
·_ ._....~.
Computer
L M H
_
...-~
.-~
Exposure Toxicity
CAS 80057
Bisphenol A
Prod:>10
Use: Cosmetics -
Effects: M UT, ETA, TE R
SAR Class: Neuro, Skin, Pulmonarv
Micro Dossier
OCR for page 229
~r>~
`'.~
Exposure Needs
and Sources
Toxicity Needs
and Sources
~ 1
Exposure Toxicity
Staff Staff
Data Bases
Compilations
Reviews
l
Literature
Staff
ll
Exposure Toxicity
Staff Staff
FIGURE 4 Stage 2 of illustrative system.
229
STAGE 2
Uses
Chem Interm
Fungicide
Nail Polish
Fate
Solid
I nsol H 2O
Sol var organics
Acute
LD50 4000+2000 mg/kg
Sensitizer
Tera
1500~500 mg/kg
Marc
Mutagen
Bioassay Indef.
Neuro
Phenol (subst)
~ Mini Dossier
1
Confidence
Le e L;. M H
7~ .
..~:~: ,
.~.~.
l
. :.:.:. l
_
or
:~::
_~
e
__
EM
H
Exposu re Tox icity CA R
l , 1,
~ a~
:. Decision:
.. Rules
:~:~:~:~:~:~:~:~:~:~:~:~:~: l
~ 111
~1
Stage 2 Processor
L M H
Exposure Toxicity
CAS 80057 - +, +
Bisphenol A
Uses Toxicity
Fate |
Mini Dossier
OCR for page 230
Exposure Needs
and Sources
Toxicity Needs
and Sources
~ 1 ~
~ I lid
~::'
Committee of
Exposure Experts
Exposure Toxicity
Staff Staff
i
FIGURE 5 Stages
l ~
Specialized Data Bases
Computer Literature Search
Secondary Literature
Primary Literature
Literature
Staff Dossier
CAS 80057
Bisphenol A
Conf idence
L M H
Exposure
Level
L M H
.
. . .
.... ,.. ....
·:-:-:- ·:-:-: 1~:-:~:
:~.. . ~ ~ :::.,
am.- ..-.-. ·...~..
I ...... ......
.:. A...
·:~:~: X :::.
_ _ ~
IM
H
Committee of
Toxicity Experts
Toxicity
R EPRO ~
NTP Scientific and
Policy Review
R EPRO
Bisphenol A
-
_ \ /
-(~
- - 11
Dossier
3 and 4 of illustrative system.
230
STAGE 3 , I STAGE 4
~ 1
fertility Assessment
by Continuous Breeding
Evaluation of Results
Fungicide, wheat & oats
Indirect food additive (packaging)
Anti-corrosive for Coatings
Flame Retardants
Dental Fillings
203,000 Workers Exposed
Teratogenicity Inadequate
Sperm Abnormalities
No Convincing Evidence
for Carcinogenicity
[. Rules ·
L ~ | ~ ~Hi 111 '
M :.:.:: : ::.: X
H a ~ .. ~.............................
Stage 3 Processor
I L M Hi
1
L M H
Exposure Toxicity
1
Reassessment of
Testing Needs
1
Feedback to Data Base
OCR for page 231
· Information on the magnitude of production of the chemical from
the TSCA Inventory US. Environmental Protection Agency, 1979, 1980),
the International Trade Commission's synthetic organic chemicals data
set, the Toxicity Data Bank (TDB), or any other accessible automated data
banks. In our example, bisphenol A is found in the TDB, where it is
shown as being produced at more than 1011 g/yr, or more than
2 x 108 lb/yr.
· Intended-use class in the Committee on Sampling Strategies data
base for the select universe, or an equivalent use code from another data
base. Bisphenol A is classified as a cosmetic ingredient in the
committee's data base, although we shall see that it could well have been
listed in another category.
· Chemical structure, if it has a defined structure and is in an
accessible data base, such as the CAS connection tables or the Chemical
Information System data base. Bisphenol A proves to have two phenol
rings joined by a propyl group at the central carbon, as shown in the
"microdossier" in Figure 3.
· Registry of Toxic Effects of Chemical Substances (RTECS) data on
observed toxic effects. RTECS includes several listings for bisphenol A
in the categories of mutagenic effects (MUT), equivocal tumorigenic
effects (ETA), and teratogenic effects (TER) and has several entries for
acute toxicity.
· Chemical fragments found in the structure by the Structure and
Nomenclature Search System (SANSS) and the corresponding suspected
effects from Table 5. Our current table would show bisphenol A only in
the "phenol (substituted)" category and would show some suspicion of
neurotoxicity, skin sensitization, and pulmonary irritation.
When the data on a given chemical have been collected from the
various automated data bases, the computer implements the decison rules
derived from the optimization model as the "Stage 1 processor." These
computer-generated rules (described in Appendix B) effectively compute a
pair of probability distributions of exposure and toxicity, as indicated
by the two small bar charts in Figure 3. Because the production of
bisphenol A is greater than 108 lb/yr and it is used in cosmetics, the
processor decides that exposure to it is more likely to be high than is
exposure to a random chemical, but still shows that the most probable
exposure is low. Similarly, the occurrence of so many RTECS entries and
the suspicion of even more toxic effects from the structure-activity-
relations (SAR) class suggest with more than usual confidence that the
toxicity is likely to be at least moderate, so moderate toxicity is shown
as having highest probability.
On the basis of these distributions, the computer decides not only
that bisphenol A should be passed to Stage 2, but that it deserves the
231
OCR for page 232
higher level of investigation for both exposure and toxicity, as
indicated on Figure 4 by "+, +," meaning that a moderate amount may be
spent on gathering and assessing data both on exposure and on toxicity.
In practice, the computer need not generate the explicit probability
distributions in making the decision, because the rules used to select
chemicals for further consideration in Stage 2 are always the same, given
the same initial data ranges. However, the distributions can be useful
inputs to Stage 2 and can be included in the microdossier that is the
output of Stage 1. Figure 6 shows the selection rules when
carcinogenicity is the only toxic effect considered. There would be
other rules for suspicion of teratogenicity, neurotoxicity, and so on.
The possibility of all such effects would be mentioned in the
microdossier for consideration in Stage 2.
STAGE 2
The microdossier (if available) is received by NTP staff and examined
by two persons, one expert in toxicity assessment and one in exposure
assessment. The exposure assessor may or may not deal with the toxicity
part of the microdossier, and the toxicity assessor may or may not deal
with the exposure part. Knowledge of the suspected toxic effects can
help guide the exposure search, but can also bias it.
me first task is to check the status of the chemical in the testing
program to ensure that the effects of concern are not currently under
examination and not yet reported in RTECS. Let us assume that no tests
of bisphenol A are in progress.
Each expert scans the microdossier and develops a strategy for
searching the appropriate data bases, compilations, and reviews, noting
what sorts of information are probably most important to retrieve. In
the case of bisphenol A, the exposure assessor calls for a routine
printout from the Toxicity Data Bank and checks the Cosmetic Ingredient
Dictionary (Estrin et al., 1982), the Merck Index (Windholz et al.,
1976), and the Directory of Chemical Producers for further information on
production, uses, and physical and chemical properties. This strategy is
carried out by the literature staff of NTP. In our example, they find
that at least 90% of the bisphenol A produced is used as an intermediate
in the production of other chemicals, especially resins. In cosmetics,
it turns out to be a component of nail polish, but there are other
dispersive uses, e.g., as a fungicide. The melting point implies that
bisphenol A is ordinarily a solid; and it is relatively insoluble in
water, but soluble in many other solvents. The exposure assessor
integrates this information subjectively and decides that the best guess
is still moderate exposure; but confidence in that guess is still low,
because the amounts used in nail polish and fungicides are so uncertain.
Meanwhile, the toxicity assessor requests a full printout of the
RTECS listings and examines the TDB printout. The literature staff
232
OCR for page 233
STAGE 1: EXPOSURE DATA ELEMENT
.
Intended Use and Production
Food chemical, unclassified or pesticide with
unknown production
STAGE 1: TOXICITY DATA ELEMENTS
Part 4
STAGES 2-4
Branch 2
Go to Stage 1
Toxicity Data
Elements Part
-
4
Go To Stages 2-4
via Branch
Member of a chemical group associated
with cancer with "low suspicion
. not in
in RTECS, no mention of MUT or CAR 2
in RTECS, mention of MUT, but not CAR 6
in RTECS, mention of ~
Stage 2: Perform search for exposure data and limited search for
toxicity data
Assessments Based on
Data Gathered in
Stage 2 Stage 3:
Exposure Toxicity Data Gathering
Assessments Based on
Data Gathered
in Stage 3
Stage 4:
Recommended
Testinga
Medium Medium
or high
Dossier
aST = Short-term test; LT = Long-term test
High toxicity
and high exposure
Otherwise
FIGURE 6 Decision rules used in processing example
chemical bisphenol A.
233
LT
ST
OCR for page 234
supplies copies of the data found in such handbooks as Dangerous
Properties of Industrial Materials by Sax (1979) and the Merck Index,
which shed little additional light on the toxicity picture. However, it
becomes evident that bisphenol A is not very acutely toxic (LD50,
around 4,000 mg/kg) and that it has shown some teratogenicity (also at
high doses) and has been judged indefinite in a cancer bioassay. The
cancer hazard is still judged as moderate, but with only moderate
confidence, and the reproductive hazards are considered high, also with
only moderate confidence. No support arises for any of the SAR
suspicions, and they are eliminated as concerns.
Both assessors have filled out the matrix of concern, as shown in
Figure 4 immediately under the minidossier. Using the decision rules
(summarized in Figure 6 for the carcinogenicity datal, they find that a
dossier covering both exposure and toxicity searching is justified in
Stage 3. Hey can also generate the exposure and toxicity probability
distributions (Figure 4, "Stage 2 Processor") either automatically from
the matrix assignment or on the basis of their own judgment. These
estimated probability distributions are passed to Stage 3 with the
augmented minidossier containing all information gathered to date.
STAGE 3
Exposure and toxicity experts (not necessarily the same ones) now
examine the minidossier and determine the Stage 3 search strategy (Figure
5~. Clearly, it is desirable to acquire more information about consumer
exposures to bisphenol A in both cosmetics and fungicides and about
occupational exposures to bisphenol A in its large use as an
intermediate. Searches of the Chemical Economics Handbook (SRI
International), the NIOSH National Occupational Hazard Survey (U.S.
Department of Health, Education, and Welfare, 1974, 1977, 1978), EPA
contractor reports, and so on are ordered. The literature staff finds
additional uses of bisphenol A as an indirect food additive (because of
residue in plastic food containers), as a fire retardant, as an
anticorrosive in coatings, in dental fillings, and in various other minor
applications. The fungicidal application turns out to be for use on
wheat and oats. In the manufacture of bisphenol A and its derivatives,
over 200,000 workers may be exposed.
Considerable effort goes into computer-assisted literature searches
of the secondary and primary literature for citations on the toxicity of
bisphenol A. The teratology experiments are found to be rated as
inadequate, but there has been a finding of sperm abnormalities in rats
after intraperitoneal injection. The "indefinite" cancer bioassay is now
assessed as showing "no convincing evidence of carcinogenicity."
Once the toxicity and exposure parts of the dossier are compiled,
they are submitted to expert committees on toxicity and exposure,
respectively. Again, there are pro's and con's as to whether the
committees should see both dossiers or only the parts pertinent to their
234
OCR for page 235
own expertise. The dossiers are studied by the committee members both
individually and collectively, but the full toxicity committee is asked
to evaluate the evidence and reach a consensus on the toxicity for each
human effect, and the full exposure committee likewise for exposure.
Each committee is also asked to state its members' confidence in the
rating. If there is much disagreement over the rating, there is good
reason to lower the stated degree of confidence. The ratings are
reported on a matrix as in Stage 2, or the committee may prefer to
generate a probability distribution directly, instead of using the
standard one generated by the Stage 3 processor. The important
difference from current procedure, however, is that the committees are
not asked to make testing recommendations, but only to assess the
available data in their field of expertise.
Whether the committees generate their own probability distributions
or simply use matrix rankings, the Stage 3 processor uses the previously
developed decison rules to decide what tests should be recommended to
NTP. In the case of bisphenol A, the committee on exposure has rated the
exposure as still only moderate, but now with high confidence. The
toxicity committee now believes that there is comparatively little
evidence of carcinogenicity, but is highly confident of moderate
teratogenicity. It now believes that other reproductive effects may be
of high concern, but has only moderate confidence in that judgment. The
carcinogenicity rule, reproduced in Figure 6, suggests that no repeat of
the bioassay is advisable now, but that an expansion of the battery of
short-term tests for mutagenic potential may be worth while. We assume
that the rule for reproductive effects would call for a test, as shown in
the box on the final version of the dossier in Figure 5. Before this
recommendation is accepted, however, the dossier and recommendations are
reviewed by the NTP panels for scientific merit--especially to see
whether the recommendation is unrealistic for some reason not perceived
in the formal process--and for such policy considerations as public
concerns, costs and feasibility of controls, or regulatory
responsibility. Recommendations may be overridden on the basis of
results of either type of review.
Assuming that the NTP panels accept the processor's recommendations,
the dossier and test directions are sent on to the testing program in
Stage 4.
STAGE 4
Finally, the substance goes into the specific type of reproductive
study warranted by the particular combination of concerns about exposure
and toxicity and the degree of confidence in each. In the case of
bisphenol A, fertility assessment by continuous breeding of rodents is
the preferred test. When the test is complete, its results are evaluated
with respect to public policy and their relevance to further refinement
of the selection process (Figure 5, "Feedback to Data Based. The test
235
OCR for page 236
results can also be fed back into the specific data base on bisphenol A
to determine whether further tests are justified. Although it is not
shown in Figure 5, the argument for a new bioassay would be much more
compelling if the short-term mutagenicity assays had proved consistently
positive.
236
Representative terms from entire chapter:
exposure toxicity
