UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
March 9, 1988
Hon. tee M. Bicmas SA1-EHC-88-019 THe A*J,"£rt?£ATo« 1
Administrator »
U.S. Bnviroiroental Protection j
Agency *
401 M Street, SW }
Washington, D.C. 20460
Dear Mr. Ttocmas!
• Ihe Halogenated Qrganies Subccimittee of the Science Advisory Board's
Environmental Health Cawdttee has completed its independent scientific review
of the Draft drinking water Criteria Document for Polychlorinated Biphenyls
(PCBs) and is pleased to transmit its final report to you. The Subccmnittee
conducted a public review; of the draft criteria document at a meeting held in
Washington, D.C, on November 19-20, 1987..
In general, the Subcomittee concludes that the document suffers fran a
failure to clearly identify its scientific objectives* Subcommittee members
encountered a myriad of facts that were not critically presented in order to
support subsequent regulatory decisions, Etor exanfjle, the document must state
a scientific rationale to enable risk managers to decide whether to regulate
Aroclors or PCBs in the environment. Vtiile sections of the document are im-
proved fran a previous draft, other chapters, discussed in the attached re-
port, require extensive revision before the document as a whole represents a
scientifically adequate statement of existing knowledge for PCBs.
A major recomnendation of the Subccnraittee is that EPA explore whether
the available data on PCS congeners can be developed on a scale of toxicity,
similar to the toxicity equivalency factor that EPA has already prepared
for dioxins. This effort could potentially yield seme scientifically^inter-
est ing insights relating to uncertainties in the PCS data base, even if it
represents only an approximation in which data analysis and scientific
judgment are eembined.
Ihe Subccwwittee appreciates the opportunity to conduct this particular
scientific review. We request that the Agency formally respond to the scien-
tific advice presented in the attached report.
Sincerely,
Norton Nelson, Chairman
Executive Contiittee
Richard A.*T£leserner, Chai
EnvirorertiaJNtealth Ccnmit
'
Halogenated crganics Subcommittee
-------
Review of the Eraft Crinking Water Criteria Document
for polyehlorinated Biphenyls (PCBs)
by the
Halogenated Qrganics Subcommittee
Environmental Health Committee
Science Advisory Board
I. General Comments
Ihe regulation of PCBs in drinking water or, in fact, any environmental
medium, encounters many problems, including several involving the existing
scientific data base. All are described in the criteria document, but virtually
none are resolved. These scientific problems include the complexity provided
by 209 different congeners and iscroers present in ecramercial PCS mixtures, each
with a different toxicity? the presence of highly toxic non-PCB ecnponents such
as polychlorinated dibenzofurans (PEEPs); the differential rates of removal of
PCB congeners from the environment which leads to differences in congener
profiles frcm those of the original pollutant mixtures? and the incomplete
knowledge of individual PCB congener toxicity.
Ihe draft criteria document suffers frcm a failure to clearly identify
its scientific objectives, ihus, as reviewers, the Subcomittee members
encountered a myriad of facts that were not critically applied, either direct-
ly or to support a hypothesis for subsequent regulatory decisions. It must
be initially resolved whether* the objective is to regulate Aroclors or PCBs
in the environment. The Subconmittee believes there is a need to more pre-
cisely define what is meant by the teon "PCBs", in part because Aroelors may
or may not be PCB compounds and because of the varied composition of PCB
mixtures»
Ihe Subcommittee reeotroends that the available data on PCB congeners be
developed on a scale of congener toxicity, even if this is only an approximation
in which data and professional judgment are conbined. Any such scale will be
£ny such scale will be imperfect, however. For example, it is unlikely that
the adverse health consequences of thalidcmide would be detected.
Ihe draft document has improved considerably frcm a previous version.
Ihere are, however, scroe sections that require extensive revisions before it
represents a scientifically adequate statement of existing knowledge. A major
problem with chapters V, VI and VIII is that they are out of date, Fbr
example, in chapter V all of the recent structure activity relationships
(SAR) for PCBs have been ignored even though they are discussed in"detail
in chapter vii. The chapter on human health effects almost completely emits
a series of papers in EnvjLj^arirental Health Perspectives (volumes 59 and 60)
and the American Journal"'of Jnd,_JtedTcijnie"(volume 5> that discuss Yusho/YuCheng
poisoning In Japan" and"Taiwan, and occupational exposures, Ihese emissions
severely compromise the quality of this chapter. In addition, it has been
shown by two studies that the major etiologic agents in Yusho and YuCheng
poisoning are PCEFs.l This does not, however, absolve PCBs as toxic agents
and as contributors to seme of the symptom of Yusho and Yucheng poisining.
-------
II. Specific leehnical Comments and Recommendations
1. Concerning the contributions from contaminant toxicity, the document
needs to wore critically address the claims that PCtFs were absent from seme
Aroclor preparations. In describing a number of studies, it is indicated that
the PCBs {individual congeners or mixtures) are free of dibenzofurans. Examples
of such statements have already been transmitted to EPA. In many instances, it
is stated that there was only "apparently" no dibenzofuran contamination, and in
no instance was the claim carefully and critically documented. The draft
criteria document probably provides sufficient and adequate information to
partially resolve the issue of dibenzofuran contribution, to observed Aroclor
toxicity. Comparisons of studies using PCS mixtures, with and without' dibenzo-
furan contamination, and comparisons of studies with purified individual PCB
congeners and with mixtures should be made in the draft document. Studies
such as the one recently published that indicates that the toxicity of dioxins
can be diminished by co-administered aroelors should be taken into account
in assessing a role for PCEPs in PCB mixture toxicity*^
2. The conclusion (page 111-26, line 10 and page 111-42, paragraph 2,
line 5) that the rat is a good model for human metabolism of PCBs is based
upon a gross oversimplification since only three distinct congeners were
evaluated, and the correlation is more relative than absolute. The statements-
should be modified or deleted.
3. The draft document is somewhat ambiguous concerning a role of metabolism
in PCB toxicity. Ctt the one hand, the less highly chlorinated congeners, and
particularly those with unchlorinated vicinal carbons, are the most highly
metabolized (page II1-37, points 3 and 4, and pages V-53 and 55), This meta-
bolism is erroneously associated with aspects of toxicity (page 111-40, para-
graph 2 and page vii-11). On the other hand, the more highly chlorinated
congeners which are minimally metabolized are reported to be the most toxic.
Metabolism plays a role in PCB toxicity, and cytochromes P-45Q are involved.
Furthermore,, PCBs can differentially induce cytochrtroes P-450 and thus affect
their own metabolism and toxicity during chronic and subchronic exposures—a
fact not discussed in the criteria document. For example, PCB exposure frcm
food could induce hepatic cytochrcmes P-450, which would affect the fate of
the more readily metabolized PCB congeners. The role of different forms of
cytochrcme p-450 in PCB metabolism should be discussed.3
4. Analytical measuranent of PCBs in water provides no indication of an
equivalent Aroclor concentration, and application of Aroclor toxicities to
such equivalent contamination is without any reasonable scientific basis and
cannot be used as the basis for regulatory decision making. Specific congener
analysis is more precise, accurate and well-suited to FCBs exposed to physical,
chemical or biotic forces. Unfortunately, not all specific congener analyses
-------
-3-
are equal, and they are frequently not comparable. The most thorough analysis
(Safe et. al,f 1985) is too ambitious at this time for routine monitoring and
cannot be applied retroactively to the vast data bank needed to determine trends.
Hie inequalities of specific congener analyses can be very misleading, if not
recognized. For example, Ballschmiter and Zell do not consider PCBs 84, 110,
141 and 149 in their 1980 paper, likewise, Schwartz et* al., (1987) emit these
congeners even though Sissons and Welti (1971), Webb and McCall (1972), Bush et.
al., (1983) and Safe et. al., (1985) have confirmed their significant presence.
In addition, PCB 95 is reported at low levels by seme authors and higher levels
are observed in the same PCB by others. Ifrerefore, the problem extends beyond
specific congener analysis verses Aroclor estimates. Ambiguities and disagree-
ments should be cited to a limited extent in any precautionary statement, per-
haps as a repeating footnote.
The task of resolving conflicting reports and selective congener identifi-
cations is monumental. Certainly the current document cannot resolve this
issue. Nevertheless, it appears to be a disservice to state that, "10 of
the 19 congeners were unambiguously from Aroclor 1016" {page W-3). Several
of these monochloro- to tetrachloro-congeners are also found in aroclors 1232,
1242 and 1248 (Webb and McCall, 1972) as well as low levels in 1260 (Safe). This
is further confirmed on page W-6 in Table IV-1.
Also, on pages lv-8 and 9 the congener composition that is "compatible".
with 24, 42 and 34% Aroclors 1242, 1254 and 1260 requires sane qualification
since lower chlorinated congeners might dissolve out and/or be degraded,
and the sediment would be enriched in higher chlorinated congeners because
of solubility and binding to particulates. later in the same paragraph,
it was stated that Lake Superior water contained 37-56% of Aroclor 1242,
while years later the sediments contained 15-21%» Also, Table IV-2 shows
higher 1242 in the aqueous phase ccwpared to particulate precipitation, while
1254 and 1260 are about equal in rain and much higher in snow particulates than
in the aqueous phase.
Ihus, it is known that chlarabiphenyls partition in these ways, and this
affects residue composition. The apparent Aroclor composition can and does
seem to change. Cautioning that these things happen and that scientists must
recognize that Aroclor estimates may be misleading, and subsequently trying to
force a mixed residue to fit the pattern(s) of specific Aroclor(s) is inconsis-
tent and roust be rectified in the document. For example, page V-95, Table v-21
indicates grossly different compositions of Aroclor 1260 which are most likely
a consequence of analytical methodology differences.
Knowing that specific congener analyses are not equal also does little good
if there are not attempts to resolve the inequalities. By throwing oat PCB
data_with seme ambiguities, there would be virtually no data to assess. More-
over,' the vast amount of data available would yield even more information if
attempts to force a convenient fit would be avoided. The ambiguities and forced
fits do not invalidate the data that are vital to the types of considerations
that need to be undertaken.
-------
In summary, the Subcommittee recoroends that a simple two-footnote qualifi-
cation system be developed stating that* 1) when Aroclor estimates are used,
there should be references to an appendix page stressing the probleras, inferences
and possible valuable information? and 2) when specific congener analyses are
used, subsequent pages should present scute of the ambiguities and appropriate
precautions.
5. there is evidence for PCB metabolite binding to protein, CNA and RNA
(page 111-40) and tNA damage by metabolites (page 111-41). Hbwever, the
mutagenieity data presented in the document show that the PCBs tested do not
induce gene mutations in either prokaryote or eukaryote test systems. Similar
studies testing the ability of PCBS to induce clastogenic effects, i.e.
chromosome aberrations resulting from chromosome breakage and re-arrangsnents,
were also negative. Thus, the conclusion, to date, would be that the PCBs
studied provide no evidence for genotoxicity, namely no induced QJA damage.
Or* the other hand, there is a possibly conflicting result in the studies
with orosophila. Here, exposure to purified Clophen-A-50 (at certain doses)
has resulted in significant increases in exceptional XO flies. This means
that either x or Y chromosomes were lost in development of the male germ line*
Such chromosome loss could result either from chrcrooscme breakage or non-
disjuction. Non-disjunction may be induced by damaging the spindle fibers
(tubulin protein), the centrosomes and/or other elements of the mitotic
apparatus such that the chromosomes are improperly segregated in roeiosis.
Thus, a mutagenic, but not genotoxic, endpoint is possibly induced by some
PCBs leading to cellular aneuploidy (unbalanced chromosome ra*nbers). The
Subcommittee reconmends that this phenomenon be discussed because it may
provide additional insight into the mechanism of carcinogenesis.
6. In reaching a regulatory decision for PCBs in drinking water, the
Subcomittee reccnmends that a modified version of Cation I (see attached
issue paper provided by the Office of Drinking Water) be adopted. It is not
necessary to derive criteria for drinking water for each PCB isaner/congener
but, rather, only for the more toxic ones. A scale of toxicities for PCB
isomers/congeners should be prepared and an "equivalency approach" developed,
using the most toxic PCBs as the basis for comparison. All available data on
PCB congener toxicity should be used in deriving a toxicity scale, including
cytochrome p-450 and non-Ah receptor effects and induction data. Although the
Subcommittee recognizes that some of these data cannot be considered "toxicologic,"
it concludes that this will not negate derivation of a reasonable approximation.
A similar approach is currently used by EPA and other governmental agencies for
halogenated dibenzofurans and dibenzcdioxins. Once such a scale of toxicities
is developed it can be used as a guide in selecting congeners of major signifi-
cance for regulatory decisions. This approach will have the advantage over
Option III in that any PCB mixture can be analyzed and decisions made and
defended.
References
1. Bondiera, et. al«, Chemosphere 13 (1984), 507; and Kunita, et. al.,
American Journal of Ind. Med., 5 (1984), 45.
-------
-5-
2, Bulletin of Environmental Cental. TPoxicol., 39 (1987), 791.
3- Pl^cta^?^"13001-' 30 (1981 5 r 577-588, and 29 (1980), 729-736;
and Mo^^Si^S 20 (1984), 7379-7384.
NoteJtG the Reader
Er, Stephen Safe, a consultant to the Halogens ted Organics Suhconmittee, did
not formally vote to approve this Subcommittee report because of a prior in-
volvment in the preparation of the draft criteria document. He did partici-
pate in the public meeting in which the Subcommittee reviewed the draft docu-
ment bat served in the capacity as a resource to clarify specific technical
issues .
-------
U.S. Environmental Protection Agency
Science Advisory Board
Envirormental Health Committee
Halogenated Qrganics Subcommittee
Roster for November 19-20, 1987 Review of the Draft Final
Drinking Water Criteria Document for Polychlorinated Biphenyls
Or. John Cbull, Chairmn^ Professor of Pharmacology and Toxicology, University
of Kansas Medicaf Center, Kansas City, Kansas 66103
Or. Seymour Abrahamson, Vice-chairman, Professor of Zoology and Genetics, Efepart-
ment of Zoology, University of Wisconsin, Madison, Wisconsin 53706
Subcommittee Members and Consultants
Dr. George T. Bryan, tepartnient of Human Oncology, K-4, Roan 528, 60S Clinical
Science Center, 600 Highland &ve,, University of Wisconsin, Madison, Wisconsin
53792
Dr. tarry Hansen, College of veterinary Medicine, University of Illinois, 2001
South Lincoln, Urbana, Illinois 61801
Er. Ronald D. tfood, Professor and Coordinator, Cell and Developmental Biology
Section, Department of Biology, University of Alabama, and Principal Associate,
R. D, Hood and Associates, Consulting Toxicologists, p. Q. Box 1927, University,
Alabama 35486
tr. Larry Kaminsky, Director, Wadsworth Center for laboratories and Research,
New York State Department of Health* Albany, New York 122Q1
Cr. Curtis Klaassen, Professor of pharmacology and Ibxicology, University of
Kansas Medical Center, 39th and Rainbow Blvd., Kansas City, Kansas 66103
Or. Eton E. McMillan, Chairman, Department of Pharmacology, Mail 1638, University
of Arkansas, Medical Sciences, 4301 West Markham, St. Little Rock, Arkansas
72205
Dr. Martha Radike, University of Cincinnati Medical Center, Department of
Environmental Health 3223 Eden Avenue—M. L. *56, Cincinnati, chio 45267
Cr. Iheroas Starr, CIIT, P. 0. Box 12137, Research Triangle Park, North Carolina
27709
Executive Secretary
Dr. C* Richard Cothern, Executive Secretary, Environmental Health Committee,
Science Advisory Board (A-101F), U.S. Environmental Protection Agency 401
M Street, SW, Washington, D.C. 20460
-------
Drinking Water Criteria for Polyehlorinated Biphenyl
I. Background;
(EPA)
Appreciable levels of PCBs were detected in ground water
samples from wells near highly industrialized and landfill
areas of New Jersey. Surface water, sediments and fish from
a good number of rivers in the 0*S. and in the Great Lakes
were found to contain PCBs. PCBs were also detected in tap
water samples of a few communities which obtain their water
from the highly contaminated Hudson River. However, none of
the above ground water and municipal tap water surveys identi-
fied the specific Arochior mixture in drinking water samples.
There is only one published report which identified Aroclor
1016 mixture, at levels as high as 100 ng/1, in samples from
a small ^upstate New fork public water supply system. The two
reservoirs of this waterworks, where Iroclor 1016 'was also
detected, used the Hudson River as their source of water.
Higher chlorinated Iroclor mixtures were present in the
Hudson River and in one of the reservoirs but not detected in
the finished drinking water samples from this community.
Polyehlorinated biphenyls pose special problems with respect
to establishing meaningful drinking water regulations. There
are 209 different PCB isomers and it seems only 100 individual •
isoroers have been identified at significant levels in commercial
mixtures. Although toxicological studies have been performed
on only a small number of the Aroclor mixtures, it is evident
that there ^are significant differences in toxicity between
different isomers and congeners. In particular, toxicity
appears to increase with increasing chlorine content, and
isomers which are axially .substituted (positions 3,4rand 5}
are more toxic than species that are substituted in positions
2 and 6. Consequently, the toxicity of PCB mixtures depends
on the isomer-specific composition of the rtixture, as illustrated
by the differing chronic toxicities of various commercial PCB
formulations. In addition, different PCB preparations may
differ considerably in the content of toxic contaminants,
such as polychlorinated diben^ofyrans.
^ these reasons, toxicity data derived from a particular
commercial PCB formulation are not directly applicable to
other formulations, and may not even be applicable to different
batches of the same formulation. More importantly, toxicity
data from studies of commercial PCB formulations may have
little relevance to the toxicity of PCBs in drinking water
because the composition of environmental mixtures is markedly
changed from the parent contaminant as a consequence of
differing solubility and stability characteristics of the PCB
isomers. Since the toxicity of a mixture could be dominated
by a few relatively minor but highly toxic constituent isoroers
or contaminants, measurement of total concentration is not an
adequate index for assessing the toxicity of PCB mixtures.
-------
-2-
II. Issue;
Based on the information on the toxicity of PCBs (mixtures,
isomers and congeners) and on their solubility and stability in
drinking water, is it possible to develop meaningful acceptable
concentrations of PCBs in drinking water?
III. Options
Option 1
0 Consider evaluating the toxicity of individual isomers
and derive criteria for drinking water for each isomer, if the
data permit. This is a scientifically sound method for addressing
the toxicity of PCB mixtures. This approach is not feasible at
present, both because isomer-specific toxicity data are not
available and because isomer specific analysis of water is not
feasible on a routine basis.
Option II
0 Assume that all PCB mixtures in the environment are
composed entirely of the most highly toxic isomer (3,4,5,3',4',5'
hexachlorobiphenyl) and use this to estimate acceptable level
of PCBs in drinking water. This will be the most conservative,
approach. While this would be certainly protective* this isomer
is a minor component of most formulations and has low solubility
in water. This option may not be protective of the carcinogenic
potential of PCBs because of the lack of data on the carcinogenic
potential of this isomer.
Option III
0 Assume that a mixture of PCBs in water retains an average
toxicity that is not greatly different from that of the parent
formulation. Based on this assumption, acceptable levels could
be derived for commercial formulations as if they were individual
chemicals. One disadvantage to this approach is that it is
usually difficult to identify which specific PCBs formulationCs)
is (are) the source of PCBs in drinking water,
IV. Recommendation;
Option III is recommended. The key assumption upon which
this option rests (that PCBs in water retain an average toxicity
similar to the parent formulation) may not be true. However,
it is unlikely to underestimate risk since the' more toxic
higher chlorinated isomers would have least water transport
potential. Changes in the acceptable levels of PCBs in drinking
water or in the basic regulatory approach may be possible in the
future as additional data and techniques become available.
-------
V. Points Of interest;
1. PCS cancer potency? Comparison with the human evidence
(Discussion to be added to document)
The, Agency's cancer potency for PCBs (calculated from
the Norback and weltman rat study and reported in the May
19S7(Drinking Water Criteria Document to be 7.7 per ntg/kg/d
continuous lifetime exposure) compares favorably to the
number of cancer cases seen following the rice-oil incident
in Japan, Although more cancer cases way be reported in the
future, a rough calculation can be made from currently available
information.
The Drinking Water Criteria Document (page VI-14, attributed
to Kuratsune} reports the average amount of PCBs consumed during
the riceoil incident to be about 2 grams. Dividing this by 70
kg (the weight of a typical adult) and by 25,600 days (the number
of days in a typical 70-year lifespan), the average daily exposure
is estimated to be about 0.0011 mg/kg/d. Multiplying this by
7.7 per mg/kg/d (the Agency's cancer potency) shows the risk
to be 847 per 100,000, In an exposed population of, 1761 (page
Vl-31, attributed to Kuratsune) approximately 14.9 excess cancer
cases are expected. This projection is not inconsistent with
the 7.39 excess liver cancer cases reported to date (page VI-31,
9 observed minus 1.61 expected). . *
Several major sources of uncertainty in this comparison
should be noted:
a. The Agency's potency estimate is a plausible upper
bound, which would tend to overestimate the number
of cancer cases.
b. The rice oil was contaminated with polychlorinated
dibenzofurans at a level 250 times more concentrated
than in the commercial PCB product Kanechlor 500.
To the extent that these dibenzofurans are responsible
for the observed cancer cases, a projection based on
based on the cancer potency of PCSs alone would tend
to underestimate the number of cancer cases.
c. Calculating an average daily exposure prorated over
an entire lifetime is very problematical, since the
exposure was intense but of short duration. In the
absence of evidence to the contrary on PCBs, this
approach is consistent with the Agency's cancer
guidelines. Nevertheless, it remains a substantial
source of uncertainty.
------- |