PB84-188887
Intrauterine Exposure of Humans to
PCBs (Polychlorinated BiphenyIs >
Newborn Effects
Michigan Univ., Ann Arbor
Kay 8«
'111 feapsteert §f Commerce
^Sisari Tedsscsl Information Service
mwm
-------
INTRAUTEKru EXPOSURE OF .HUMANS. TO..PCBsr
Newborn Effects
I o O 1
Grtta Fein1, Joseph L. Jacobson , Sandra Vf. Jacobson and Pamela Schwarz
1, School of Public Health 2, Department of Psychology
University of Michigan Wayne State University
Ann Arbors, MI 48109 Detroit, MI 48202
PS-CR-808S2Q010
PROJECT OFFICER:
Nelson Thomas
ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
DULUTH, MD 55804
-------
TECHNICAL REKJfiT DATA
/Ro^f^/iuawrioMaN
1, RBPttsr KO, 12. fi. REofisSlT'S
EPA-600/3-04-060 1 | PB8 4 18 8 8 3 7
S. TITLE AMD SUfcTITIB J
Intrauterine Exposure of Husaaa to PCBs: Newborn
Effects
9. RGFOUT DATS
May 1984
&KMFONMSNS OKOAM.ZATIOM COOS
>. MTMdmei
G. Fein, J. L. Jacobean, S. W. Jacobson, and P. Schwarz
3. PCRPOAUINQ ORSAMIZATIOM R6P0*T ISO.
9. PCfi^QRH-.N& OK8AKIZATIO* NAM* AMD ATOMS*
School of Public Health
University of Michigan
Ann Arbor* MI 48109
16. fftosSRAsi iii.ku&tcV7i6.
It, COKTRA'CtfBttHlY HQ.
CR808520-01-0
is, sMxeoftma mimcv imm amd
Environmental Research Laboratory
Office of Research and Developaent
U.S. Environmental Protection Agency
Dul uth, MH S5S04
it. surri.iM(i»T*iiv motes
1ft. abstract
The effect of low-level chronic exposure to polychlorinated biphenyls (PCBs) from
consumption of Lake Michigan fish was assessed in pregnant women and their newborn
offspring. Low levels of PCBs remain in the htmm body for some time, and caused, in
this sample, decreases in birch weight, head circumference, and gestational age of
the Tiewoorn. PCBs appeared to be transmitted to the infants prior to birth through
the maternal serum, and after birth through breast tailk. Behavioral deficiencies were
observed in the infants exposed to PCSs both in autonomic immaturity and depressed
responsiveness.
KkV *ORDl ANO OOCUUIKT AMAVVSI*
L DESCRIPTOR*
b.lOCNTirilRSyOrEN CND80 TIRUI
c COSATl Ffcld/Grovp
•
I*. OISTMIIUTION STATEMENT
1>. SiCURITV CLAM ITKU R,fan)
UNCLASSIFIED
31. NO. Of PAQIi
88
RELEASE TO PUBLIC
s», stcumf v cuiia (ixup^i
UNCLASSIFIED
za. mict
IP* Pm 3BS-1 (K>*. 4-77) nifieui CDiTiOm >• oiioliti
i
liTOI OP R SPORT ANOPSHtOO CO
VENEO
1«. M-ONSOMINO AGfHCY COOS
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
nent or recommendation for use.
ii
-------
Intrauterine Exposure of Humans to PCBs:
Newborn Effects
1 2
• Greta G. Fein , Joseph L. Jacobson ,
2 1
Sandra W. Jacobson , and Pamela W. Schwartz
Pinal report Submitted to the U.S. Environmental Protection Agency-
February, 1983
This research was funded under grant #C1808S20010 from the
United States Environmental Protection Agency. We wish to thank
Waylar.d R. Swain and Nelson A. Thomas of the EPA Large Lakes
Research Station and Harold E« B. Humphrey of the Michigan
Department of Public Health for their generous assistance in the
development of this project. We also wish to thank Harold A.
Price, Michael D. Mullin, and Janet Eyster for their many help-
ful suggestions regarding the analysis of biological data and
Joel Ager for his suggestions regarding the statistical analysis
of the data. We greatly appreciate the assistance of Carol
Alexander, Gail A. Brumitt, Gwen Hendrickse, and Patricia Yaros
in the collection of the data? Jeffrey R» Dowler and Gregory A.
*Proulx in data analysis,- and Carol Stanik and Maria Petkoff
in preparation of this manuscript. Finally, without the "cooper-
ation of physicians, hospital personnel, and parents in Grand
Rapids and Muskegon, this study would not have been possible.
1
-------
Polychlorinated biphenyls (PCBs) are organochlorine com-
pounds. Their chemical properties, including a high dielectric
constant and the ability to withstand high temperatures for long
periods, have resulted in their utilization in a wide variety of
manufacturing processes. A world wide pollutant, PCBs have been
found in air, water and soil samples of .industrialized countries
(Carey & Gowen, 1976; Jelinek & Corneliussen, 4976? Veith & Lee,
1971; Young, McDermott & Heesen, 1976). Detcctible levels of
PCBs have also been found in tissue and blood samples of the
general human population (Biros, Walker & Medbery, 1970; Yobs,
1972) , and relatively high levels have been reported ill human
breast milk samples collected in Canada and the United States
(Grant, Mes, & Frank, 1976; Musial, Hutzinger, Zitko & Crocker,
1974; Savage, Tessari, Malberg, Wheeler & Bagley, 1973).
Evidence from several sources indicates that PCBs cross
th*j placenta and are teratogenic in human infants exposed at high
levels. PCBs are found in umbilical cord serum at birth when
exposure results from ordinary dietary sources, and maternal
serum levels are moderately related to fetal serum levels (Kodama
et al.,
& Ota, 1980; Masuda, Kagawa, Kuroki,A1978; Polishuk, Wasserman,
Cucos & Ron, 1977). In separate incidents in Japan and "Taiwan,
infants of pregnant women who consumed rice oil accidentally
contaminated with sizable amounts of PCBs showed elevated PCB
serum levels e»nd a set of physical characteristics (e.g., brown
pigmentation of the skin, swollen eyelids) known in the litera-
ture as Yusho (rice oil) disease (Karada, 1976; Hicuchi, 19 76;
Wongs Hwang, 1981). Infants exposed in utero were small-for-
2
-------
gestational age and tended to be born prematurely (Funatsu,
et al.,
Yamashita, Ito,^1972). Follow-up studies of exposed infants
reported an average IQ of 70; sluggish, clumsy and jerky move-
ment; apathy and hypotonia; autonomic disturbances; and growth
impairment (Harada, 1976/. These disturbances in responsiveness
and neuromusclar organization are consistent with reports of
slowed sensory and motor nerve conductance in exposed adults
(Chia, Su, Chen, Wu £. Chu, 1981; Harada, 1976) . Although PCB
blood levels diminish once exposure ceases, infants born several
years after the exposure of their mothers showed elevated PCB
blood levels, reduced birth weight, and brown pigmentation of
the skin (Abe, Inoue & fakamatsu, 1975; Harada, 1976).
Placental passage and fetopathy have also been demonstrated
in experimental studies with mammals such as rabbits, nice, rats,
and rhesus monkeys (Allen, Barsotti & Carstens, 1980; Bowman,
Ksironimus & Allen, 1978; Chou, Miike, Payne? & Davis, 1979;
et al., ¦
Masuda, Kagawa, Kuroki ,^1979Grant, Villeneuve, McCully &
Phillips, 1971; Tilson, Davis, McLachlan & Lucier, 1379). PCBs
may cross the placenta more freely as gestation progresses. In
mice, higher concentrations are found in the fetus, placenta,
and amniotic fluid as parturition approaches (Torok & Weber, 1981).
In rhesus monkeys, prenatally exposed infants weighed lets
at birth than controls and by two months of age developed Vushc-
like symptomatology. Behavioral tests given between 6 and 24
months of age revealed hyperlocomotor activity and learnin 5
retardation (Bowman, Heironimus & Allen, 1978), but when tested
at 44 months of age, these animals were hypoactive relative to
controls (Bowman 6 Heironimus, 1981) . In a follow-up study, adult
female monkeys were removed from their PCB diet for a one-year period
3
-------
prior to conception (Alien et al., 1980), Although the physical
condition of these animals improved and PCB levels in maternal
adipose tissue declined markedly, four of the IS offspring were
stillborn. The birth weight of the liveborn infants was reduced
relative to nonexposed infants, and four infants died after
weaning at four months of age. In these studies, animals were
fed diets containing only 2.5 or 5.0 mg/kg of PCBs, levels
thought at the time to be low. In rhesus monkeys, as in humans
(Harada, 1976), teratogenic effects appear in infants whose
mothers had consuxnd PCB prior to, but not during, pregnancy.
When mice were fed relatively low levels of PCBs during
pregnancy (32 mgAg on days 10-16 post-conception), some pups
subsequently developed the "spinning syndrome" (Chou et al.,
1979; Tilson et al., 1979), More subtle signs of neurobehavioral
dysfunctioning appeared in prenatally PCB-expcsed mice which did
not develop obvious clinical signs of neurotoxicity. Assessments
of subtle behavioral functions indicate that exposed but
apparently asysptomtic mice have learning deficits and depressed
neuromuscular reflexes even though PCBs are no longer detectable
in their body tissues (Tilson et al,, 1979). A sluggish response
m
to a novel situation and less pronounced habituation have been
observed in asymptomatic mice exposed perinatally to PCB doses
below those associated with obvinu«j neuromotor disorders (Storm,
Hart 4 Smith, 1981),
Table 1 summarizes the nsyr assessment categories used in
teratogenic assessment (Spyker, 1975) and the results of the
studies reviewed above. While to date no morphological anomalies
4
-------
have been found to result from fetal exposure to PCBs, such expo-
sure has been associated with clinical symptoms, maturations^ and
growth retardation, and a variety of subtle behavioral deficits.
Consumption of PCB-contaminated fish constitutes a major source
of human exposure to PCBs, with PCB body burden directly related to
amount of fish consumed (Humphrey, 1976? Kreiss, Zack, Kimbrough,
Needham, Smrek & Jones, 1981). Although epidemiological studies of
acutely exposed h'cans indicate that PCBs are injurious,, there are
problems in extrapolating from studies of humans acutely exposed at
high levels to those exposed chronically at low to moderate levels.
Even at relatively high levels of consumption of PCB-contasinated
fish, short-term exposure is considerably less than that experienced
in the rice oil incidents. In contaminated fisheaters, obvious and
distinctive Yusho symptoms have not been noted, and infants born
to mothers who consume contaminated fish are likely to be clini-
cally asymptomatic. Analyses of the consequences are complicated
because the effects of low level chronic exposure are likely to
be subtle and may occur in the absence of symptoms typically
associated with acute exposure.
Although subtle behavioral effects appear in animals exposed
to low doses of PCBs, extrapolation from animals to humans is
also risky. For example, species differ in their ability to
eliminate PCBs, and as a result, the dose required to produce an
effect will vary among species. Animal models are an essential
aspect of teratogenic testing; however, the findings of this type
of research must be corroborated in studies of humans.
5
-------
Tab!* 1
Assessment Cntagorlas In the (valuation
of Teratcgontc Effects
Assoflsnent category
11 lustration
I. Morphological character I at »CS
3. Physical charactarlsl lea
3. Matur at lonal landmarks
4. Growth
5. Re?taxes
8, Activity level*
7. Neuromuscular and sensory motor
capacItles
B. Sonsory and attentlonat functions
S. Learning ability
tO, Autonomic regulation
1 I. SoxuaI doveIopp«mt
*no«iial Ins not found In PCS studies
Chtoraene; brown pigmental Ion of akin. Hps,
gums; parchnenl-lIks sk in; swotI en eyelids. faca
premature erupt fori of leslh
Preteri* birth (neurofnuacular and phya teal
Imno turI
Small birth sIza; depressed postnatal growth
¦Spinning syndrome* In nice
Hypo- and hyper-act Ivlty In an opon-f fold;
clinical QGEOGsmant of apathy
Tost® of coordination (swimming In mIce)
OoffcHl In visual, auditjry or olfactory
a 11 nr t lona V funct lon.t not f ound; nunhness
report»d In humana exposed In adul thrvod
Alteration and revernaI learning deficits In
monkeys
Inapproprlata response to stress: d«f
-------
The present study involves two groups of hmeans who are
particularly at risk; the pregnant woman and her newborn
offspring* The study, v'nich utilizes a prospective longitudinal
design, investigates the model shown ir. Figure 1. In this model,
consumption of contaminated lake fish, assumed to be a major
source of PCB exposure, results in elevated asternal PCB levels.
Elevated maternal body burden has both direct and indirect
implications for the developing fetus. Since PCBs cross the
placenta, the developing fetus is exposed directly to some part
of the mother's body burden. Maternal body burden may also
affect maternal health and reproductive capacity, which in turn
may influence fetal development. PCBs are also transferred from
maternal tissue to breast milk, which provides an additional
source of direct exposure for breast-fed infants.
In this report, relations among various measures of exposure
and newborn outcomes are of special interest. First, measures of
exposure are examined to validate linkages between contaminated
fish consumption and elevated maternal PCB body burden, and
between PCBs in maternal serum and PCBs in cord serum. Then,
three measures of exposure (fisfe. ffcmsumption and PCBs in maternal
and cord sera) are analyzed in relation to measures of newborn
status.
Method
Subjects
7
-------
Figure 1. Model of pre- and postvital exposure to polychlorlnated b1phenyls via ingestion
Of contaminated fish.
-------
The sample, comprised of 313 newborn infants and their
mothers, was drawn from three hospitals in Grand Rapids and one
4
in Muskegon. All mothers delivering at these hospitals were
considered eligible to participate with the following exceptions:
mothers had to be at least 18 years of age and have completed
10th grads in high school.
The infant sample consisted of 164 males and 149 females.
The proportion of males (52%) did not differ significantly from
expected (51%). All infants were white, with the exception of
two black infants. Forty-three percent were fiist-borns. Six
were the larger members of twin pairs, none of whom died during
infancy. Anthropometric measures and demographic characteristics
are presented in Table 2. Only ore infant was below the tenth
percentile on v/eight-for-gestational age (Lubchenco, Hansnsan,
Dressier & Boyd, 1963). Cne severely premature newborn (GA =
26.9 weeks, birth weight - &Q9 gru) died. When the health status
of the sample was evaluated on the Postnatal Complications Scale
(Littman & Parmelee, 1974b),^ the mean score was 94.1 percent
optimal. Only five infants (2.1%) had a 5-minute Apgar score
below 7.
Socioeconomic status (SES) JfjPthe study sample, based on
a weighted sum of father's education and occupational status
(Hollingshead, n.d.), ® was pSpfiominantly middle-middle class.
Most of the fathers' occupations ranged from semi-skilled
9
-------
Table 2
Characteristics of the Sample at Birth (N=313)
Mean
m
Range
Infant characteristics
Birth weight (gtn)
3502.7
(560.3)
909.1 - 5028.9
Crown-heel length (cm)
51.1
(2.5)
28.5 -
56.0
Head circumference (cm)
35.1
(1.4)
30.5 -
41.3
Fonderal index3
2.7
(0.3)
1.9 -
5.6
Gestational age (weeks)'5
40.4
(3.0)
24.3 -
45.9
Apgar scores
1-minute
7.9
(1.3)
1.0 -
9.0
5-minute
8.9
(0.8)
2.0 -
10.0
c
Postnatal Conplicatioas score
0.9
(0.1)
0.2 -
1.0
Fanily characteristics
Socioeconomic status (SES)^
39.3
(13.5)
14.0 -
66.0
Maternal age
26.1
(4.6)
18.0 -
42.0
Maternal education (years)
13.3
(1.9)
10.0 -
19.0
Marital status (married)
283.0
(90.4)
—
Obstetrical Complications scoref
84.0
C0.1)
0.6 -
1.0
a_ . , /UJ.. . „ , ,. birth weight (rat) x 100
Based on equation (Miller & Has:>anein, 1971): —1— ; 7—: —rr
^ crown-nec1 length (cm)3
bBased on reported last menstrual period.
c
Based on Postnatal Complications Scale (Littnan & Panne lee, 1974b).
Eased on Hollingshead Index (n.d.).
e
_N is presented rather than nean; percent of women married is in parentheses.
e
"Based on Obstetrical Complications Scale (Llttnan & Paraelee, 1974a).
10
-------
laborers to middle-level management, with only 4.5 percent
classified as menial or unskilled. Almost all of the women were
married and living with the father of their newborn infant.
The sample was also homogeneous in terms of maternal
education and age. Only 7,7 percent of the mothers failed to
complete high school, while 18.2 percent completed college.
Eighty-one percent- were between 18 and 30 years of age and only
3.2 percent over 35. Fifty-seven percent attended prenatal birth
classes; 95.8 percent began prenatal medical care by the fourth
month of pregnancy, and the remaining 13 women by the sixth
month. Their scores on the Obstetrical Complications Scale
(Littman & Parmelee, 1974a )^ ranged from a minimum of 63 percent
to a maximum of 98 percent optimal, with a median of 85 percent
optimal, suggesting little variability in obstetrical history.
The sample was selected to overrepresent women who had
consumed Lake Michigan fish during some time in their lives. The
majority (77%) reported consuming moderate to large quantities of
Lake Michigan fish, while the rest (23*) reported consuming none
of these fish. This design permits two types of analyses.
First, adverse effects can be identified by comparing women who
consume Lake Michigan fish with these who do net. Additional
analyses can be performed to determine the statistical magnitude
of the toxic effects and the particular dose-response
relationships.
Procedure
Preliminary Screening
11
-------
To ensure participation of infants likely to have been
exposed i_n utero to a broad range of PCB levels, women were
screened for past and present consumption of Lake Michigan fish
on the morning following delivery. A questionnaire was used to
Q
ascertain the source, amount, and kind of fish consumed in the
year prior to the infant's birth, and the number of years of fish
consumption at that rate. Women were also asked if there was a
time in the past {dating back to 1966} when they ate greater
amounts of Lake Michigan fish (e.g., the woman grew up in a
family in which the father was a sports fisherman). If so, the
amount and kind of fish consumed in the past, as well as the
number of years of consumption at that rate, were recorded.
Sampling for contaminants in Lake Michigan fish has shown
that most species tested contain detectable levels of PCBs, with
especially high levels in fatty species, such as salmon and lake
trout (International Joint Commission, 1978; Swain, 1981,*
Willford, Hesseburg B Nicholson, 1976) . These fish, although hig
contamination, may be eaten in small quantities, while other less
contaminated fish, such as perch or smelt, may be eaten in large
quantities. To obtain a more accurate measure of contaminated
f ish consumption, a weighted value was assigned to each fish
species based on average contaminant levels for that species
(Table 3), The range of values Has normalized to lake trout,
salmon, carp, and catfish, species containing the highest
concentrations of PCBs, e.g., a fish having half the average PCB
level of saloon or lake trout would be weighted 0.5.
12
-------
Table 3
Relative PCB Values of Lake Michigan Fish
Type of fish PCB weights
Chinook salmon
1.0
Coho salmon
1.0
Lake trout
1.0
Carp
1.0
Catfish
1.0
Bullhead
0.6
Pike or pickerel
0.6
German brown trout
0.4
Perch
0.4
Walleye
0,4
White sucker
0.4
Alevi fe
0.3
Burbot (lawyer)
0.2
Steelhead or rainbow trout
0.2
Brook trout
0.2
Menominee (round whitefish)
0.1
Smelts
0.1
Whitefish and chubbs (herring)
0.1
Note, Derived from data supplied by' the Large Lakes
Research Station, U.S. Environmental Protection Agency, Grosse
lie, Michigan.
13
-------
The following equation was used to estimate the mother's
annual rate of consumption for each type of fish:
RELATIVE NUMBER OF FISH 0.23 KG PCB-RG
PCB FISH X MEALS/YEAR X (1 PISH MEAL) = PER YEAR
WEIGHT
This measure is summed across species for each individual to
determine a woman's annual fish consumption rate. To estimate
cumulative exposure, annual PCB-fish consumption rate was
multiplied by the number of years the mother reported eating
contaminated fish at that rate. If she reported eating
contaminated fish at a higher rate in the past, the number of
years at the current and past rates were determined separately
and then multiplied by their respective consumption rates to
yield a measure of cumulative exposure as follows;
PCI NUMBER OF PCB NUMBER OF CUMULATIVE
PRESENT KG/ X TEARS AT + PAST KG/ X YEARS AT = CONSUMPTION
YEAR PRESENT RATE YEAR PAST RATE FCB-KG
Sampling Criteria: Exposed and Unexposed
Two criteria were used to determine inclusion of women in
the study: (1) If cumulative exposure within a 6-year period
reported by the mother was equal to 11.8 kg (26 pounds) or more
of contaminated fish, the mother qualified for inclusion in the
study. This criterion was based on evidence that PCBs are
retained in body tissues over relatively lopg periods of time
14
-------
(Humphrey, 1976). (2) In addition, a control group was selected
comprised of newly-delivered mothers who reported consuming no
Lake Michigan fish in the past or during pregnancy.
Of the 8402 women screened (i.e., 96 percent of all
maternity patients in the four participating hospitals), about 4
percent or 343 ate PCB-fish in sufficient quantities to qualify
Q
for inclusion in the exposed group. About 29.3 percent of the
women screened reported eating no Lake Michigan fish; 4.6 percent
or 114 of these women were invited to serve as controls. As a
check on whether women invited to participate in the study,
either as fish-eaters or as controls, were similar
demographically to women who were not asked to participate, a
demographic comparison group was selected by matching each
potential participant with the next woman whose screening
interview followed her own. This procedure assured a comparison
group matched for hospital and approximate date of delivery with
the potential sample of exposed and nonexpcsed mothers. Women
invited to participate did not differ significantly from others
in their community in socioeconomic status, maternal age,
education, marital status, or sex of infant (Table 4). Among
those women invited to participate, fish-eaters did not differ
from control women on these variables (Table 5).
15
-------
Table 4
Comparison of Qualifiers and
Demographic Comparison Groups (N=914)
Qualifiers? Demographic
Exposed and comparison
control groups group
(N=457) (N=457)
Mean
N
(%)
2
X
Marital status (married)
419
(91.7)
413
(90,4)
0.18
Sex of infant (male)
224
(49.6)
241
(53.9)
1.71
Note. Data on SSS, maternal age, and sex of infant were unavailable as follows:
SES was unavailable for eight qualifiers and ten comparison women} sex of infant for
five qualifiers and ten comparison women; maternal age for one comparison woman.
g
Based on Hollingshead index (n.d.).
-------
Table 5
Comparisons of Qualifying Exposed and Control Croups (N*457)
Qualifying Qualifying
Exposed Group Control Group
(N=345) (N=112)
Mean
(SD)
Mean
(SD)
i
Socioeconomic status (SES)a
38.3
(13.1)
39.4
(13.2)
-0.76
Maternal age
26.0
(4.6)
25.4
(4.3)
1.33
Maternal education (years)
13,1
(1.9)
13.2
(1.9)
-0.44
N
(!)
N
(1)
2
X
Marital status (married)
319
(92.5)
100
(89.3)
1.12
Sex of infant (male)
167
(49.4)
54
(49.5)
0.00
Note. Data were unavailable on SES lor six exposed and two control womenj aex for
seven exposed and three control infants.
3Based on Hoilingshead index (n.d.).
-------
01 the 457 women invited to participate, 144 (32%) refused.
Mothers who consented to participate were more educated than
refusers, suggesting that greater education may be associated
with greater interest in research. Among women who agreed to
participate, exposed and non-exposed control groups did not-
differ in socioeconomic status, maternal age or education, number
of offspring (parity), or sex of infant (fable 6). While
slightly more of the exposed women were married, this difference
was not statistically significant (jj > ,05).
Measures of Exposure
PCB-fish consumption. The principal measures of
contaminated fish consumption were: (1) highest annual rate of
PCB-fish consumption; (2) cumulative PCB-fish consumption; and
(3) rate of PCB-fish consumption during pregnancy. These
measures were weighted for the contaminant levels of the species
reported by the mothers. The first measure (highest annual rate)
reflects findings from previous studies that the highest level of
contaminated fish consumption fji^dicts subsequent PCB serum
levels (Humphrey, 1976), findings consistent with the tendency of
PCls to be poorly metabolized. But PCBs tend to accumulate in
body tissues over time, and prediction might be improved when
18
-------
Table 6
Study Sample: Exposure and Control Groups (N=»313)
Exposure
group
(N-242)
Mean (SD)
Control
group
(N-71)
Mean (SD)
Socioeconomic status (SEj)'
Maternal age
Maternal education {years)
39,3 (13,6)
26.2 (4.7)
13.3 CI.9)
39.5 (13.4)
25.7 (4.5)
13.3 (2.0)
-0.14
0.81
0.15
N
(%)
N (%)
Marital status (married)
Sex of infant (male)
Parity (first-borns)
223 (92.1)
124 (51.2)
100 (42.2)
60 (04.5)
40 (56.3)
32 (45.1)
3.70
0.39
0.18
SES, based on Hollingshead index (n.d.), was unavailable for three exposed women,
-------
reported values and fears at these levels are taken into account.
The measure of cumulative PCB-f ish consumption, which sums rates
over years after normalizing each component of the measure to z
scores, was constructed to assess this possibility. The third
measure, rate of PCB-fish consumption during pregnancy, was used
to assess a somewhat different possibility, i.e., that the impact
on the infant is especially pronounced when the mother consumes
contaminated fish during pregnancy. PCB-kg refers to a fish
consumption score weighted by the contaminant levels of the
species reported by the mother. Mean annual rate of PCB-fish
consumption at the highest levels reported by the exposed mothers
was 6.7 PCB-kg (14,7 PCS-pounds) per year (SD ¦ 5.8 PCB-kg, range
- 1.2-41.7), This rate is equivalent to about 0.6 PCB-kg per
month or 2-3 salmon or lake trout meals (at 0.2 kg per meal).
The women reported eating Lake Michigan fish for an average cf
16.1 years (SD - 9.0, range ¦ 1.0-40.0). In the year preceding
the infant's birth, mean level of fish consumption was 4.4 PCB-kg
(9.7 PCB-pounds; SD • 4.4 PCB-kg, range ¦ 0.0-26.5) . Therefore,
the selection procedure described earlier yielded a range of
exposure levels with specific values depending on the particular
measure of maternal fish consumption used. The fish consumption
variables were positively skewed and were normalized by means of
log transformations. One outlier, a woman who reported ingesting
54.5 kg of contaminated fish in one year, was eliainated from the
analyses.
Laboratory Analyses
20
-------
Hospital personnel collected samples of umbilical cord blood
for every infant born in the participating hospitals at the time
of delivery. Cord blood specimens for newborns in the sample
were available, therefore, for laboratory analysis, while those
of non-participants were discarded shortly thereafter. Samples
of maternal blood were collected on the second day following
delivery, as part of routine blood tests. Sixty-one percent of
the infants in the sample were breast-fed, and of those, 70.3
percent of the mothers provided breast milk samples between 1-16
weeks after delivery. A second sample of breast milk was
collected from 45 women who continued breast-feeding at 5 months.
Analyses of cord serum, maternal serum, and breast milk
samples for PCBs were performed at the Michigan Department of
Public Health by means of parked column gas chromatography (Burse,
Needham, Liddle, Bayse & Price, 1980). ^ 4-raL sample of serum was
mixed with 2 mL methanol. This mixture was extracted with 5 mL
hexane/diethyl ether (1:1, v/v) three times by agitating on a
rotary mixer 15-20 minutes at 50-55 rpm. The extracts were
combined in a Rentes concentrator tube, reduced to a small volume
(about 0.5 mL), and transferred quantitatively to a miniature
Florisil cleanup column from which PBBs,PCBs, and chlorinated
hydrocarbon pesticides were eluted with 6 percent diethyl ether/
petroleum ether mixture (v/v). T&e eluate was concentrated to a
small volume (1.0 mL), transferred with three x 1.0 mL rinsings
of hexane to a column of Siliq^ Gel 60, and eluted with hexane to
separate PCBs from the chlorinated hydrocarbon pesticide group
(Picer 6 Ahel, 1978). The fraction containing PCBs, PBBs, and DDE
21
-------
was concentrated to 2 mL ana analyzed by electron-capture gas
chromatography with a 3 percent SE-30 column. Injector, initial
column, final column, and detector temperatures were 240°C»
160°C, 210°C, and 32Q°C, respectively. Quantification was
provided by adapting the Webb-McCall method (Sawyer, 197$; Webb &
HcCallr 1973) to a computer data system with Aroclors 1016 and
1260 used as calibration standards. Estimation of PCB levels in
sample extracts was based on comparison of peaks on the
chronatogram with respective peaks of known concentrations in the
reference standards. The Aroclor 1016 portion of the
chromatogram was used to estimate the quantity of lower
chlorinated congeners, while the 1260 portion was applied to
measure levels of more highly chlorinated congeners and isomers,
Limits oi detection were 5 ng/ciL for Aroclor 1016 anu 3 ng/aL for
Aroclor 1260 in the sample extracts. Quality control over time
was ensured by addition of a spiked sample of bovine serum with
every batch of ten samples, as well as the use of a glassware
blank and a reagent blank. Serum lipids* were determined
colorimetrically using the Dade Total Standard for plotting the
calibration curve of concentration vs. absocbence (Fringe,
Ferdley, Dunn & Queen, 1972),
A similar procedure was used in analyzing breast milk
samples. A 5 g sample of breast milk. in a screw-capped {teflon-
lined) culture tube was mixed with 0.1 of potassium oxalate and 5
mL methanol. This mixture was extracted in the same manner as
serum samples. Extracts were combined in a micro distilling
assembly and concentrated to a small volume. Evaporation of
22
-------
solvent from the concentrated extract was conducted in a small
preweighed aluminum dish and the residue dried under vacuum in a
desiccator. In contrast to serum analysis, wherein lipid levels
were estimated, lipid content of the breast milk sample was
determined graviaetrically. The lipid residue was then
redissolved in a snail amount "of petroleum ether and transferred
quantitatively to a miniature Florisil cleanup column.
Subsequent procedures for serum were then applied to the breast
milk analyses.1^ (For a review of technical problems associated
with these analytical procedures, see I&RC Monograph, 1978;
(Safe, Pochini, McCrindie & Rotries
Middle, in press; Mullin, Saf^T^in P^ss; Humphrey"]! T57BTT
All serum samples containing lipid levels below 200 mg/dL
were omitted from the data analyses, since measurement techniques
were judged to be insufficiently sensitive to detect PCB levels
at lipid values of this magnitude. The removal of three maternal
serum and 43 cord serum samples for this reason did not alter the
fundamental nature of the PCB distribution. Twenty-seven
additional cord blood samples could not be analyzed because
insufficient quantities had been collected, the hospital had
mistakenly discarded them, or because of equipment failures in
^the laboratory. Included in the following analyses were 198 cord
Infant Measures
There were four primary considerations in selecting infant
outcome measures. First, there AsP to be evidence from studies
of humans and animals that particular clinical symptoms, and
%
maturatior.al end other subtle functional disorders were
associated with pre- or postnatal exposure to PCBs.
23
-------
Second, measures selected had to havi proven sensitive in
discriminating various at-risk conditions and in detecting subtle
behavioral effects.
Third, measures had to be comprehensive. Spyker's model of
teratogenic effects suggests that in contrast to acute massive
exposure which is associated with dramatic clinical disorders,
lcngterm low-level exposure to a toxin produces subtle functional
disturbances. Depending on genetic predisposition, the chemical
involved, the doss and time of exposure, a variety of
disturbances may occur. Assessment? need to be comprehensive and
include the various domains in which teratogenic deficits appear
in different species.
Fourth, measures of infant and maternal health and
background have to be complete enough to permit identifying
correlates of PCS exposure so thet potential confounding
variables, such as maternal health, social class, parity,
exposure to other drugs and anesthesia, can be statistically
controlled.
The principle dependent measures include: (a) three
anthropometric measures—birth weight, crown-heel length, and
head circumferencer (b) two estimates of gestational age, one
using an observational technique developed by Ballard, Novak, and
Driver ( 1979) and the other based on the mother's report of her last
menstrual period prior to pregnancy; and (c) a behavioral assessmer
of the newborn based on the Neonatal Behavioral Assessment Scale
(NBAS) {Brazelton, 13735.
24
-------
Birth weight was obtained from the infant medical record.
Because hospital measures of crown-heel length and head circum-
ference were less strongly correlated with birth weight than
our own assessments (r = .66 vs. .75 and .29 vs. .68, respec-
tively) , these measures were obtained from our examination
performed during the second day following delivery.
The Ballard Examination for Fetal Maturity was administered
to 220 infants at 20-53 hours of age. The remaining 93 infants
were not tested because maternal consent could not be obtained
within the limited time period during which the Ballard is
considered reliable. The Ballard Scale correlates highly (r -
.9?) with the longer Dubowitz examination (Dubowits, Dubowitz
& Goldberg, 1970), on which it was based (Ballard, Novak £
Driver, 1979). The Ballard, which takes an average of
three as compared to the Dubowitz* nearly seven minutes to
administer, omits items found to be unreliable, especially for
sick newborns.
The Brazelton Neonatal Behavioral Assessment Scale (NBAS)
evaluates the infant's response to increasing levels of external
stimulation. Although new, the NBAS has emerged as a useful
technique for assessing neonatal functioning and detecting a
variety of at-risk conditions, such as pre- and post-maturity
(Field, Hallock, Ting, Dempsey, Dabiri & Shuman, 1978), low
birth weight (Als, Tronick, Lester & Brazelton, 1979), exposure
to obstetrical medications (Aleksandrowicz & Aleksandrowicz,
1974), and methadone treatment (Strauss, Lessen-Firestone,
Starr & Ostrea, 1975). It has also proven effective in predicting
25
-------
interactive and attachment behavior during the first year (Field,
Dabiri, Hallock & Shuman, 1977; DiVitto & Goldberg, 1979; Vaughn,
Taraldson, Crichton & Egeland, 1980) and performance on measures
of subsequent intellectual functioning (Powell, 1974; Sameroff,
Krafchuk & Bakow, 1978), especially when used in connection with
other risk factors (Field, Hallock, Ting, Dempsey, Dabiri & Shuman,
1978; Field, Demsey & Shuman, 1981).
The NBAS was administered to 287 newborns. In all but three
instances (98% of the cases), testing was performed on the third
day after birth (i.e., at 48-72 hours). In addition, the testing
of one premature was delayed until the infant reached the con-
ceptual age of 37 weeks, the earliest age at which the NBAS is
considered valid.
The NBAS consists of 27 nine-point scales and 17 reflexes
(Brazelton, 1973; Als, Tronick, Lester & Brazelton, 1979). In
order to enhance reliability of measurement and to decrease the
likelihood of chance findings resulting from an excessive number
of comparisons, the 44 NBAS items were reduced to seven summary
clusters. These clusters were derived by synthesizing the
results of factor analyses from studies of six independent
samples, including our own (Aleksandro$icz & Akelsandrowicz,
1976; Kaye, 19 78; Osofsky & O'Connell, 1977; Sameroff, Krafchuk,
St Bakow, 1978; Streissguth, Martin & Barr, 1977; see Jacobson,
Fein, Jacobson & Schwartz, in prfe^, for a more detailed review of
procedures used in deriving these clusters). The clusters
A
described m Table 7, which generally resemble the seven clusters
suggested by Lester, Als, and Brazelton (1982), regroup those
26
-------
Table 7
HBAS Items included in the Seven Clusters
Clusters
Response decrement
Orientation
Tonicity
Range of state
Regulation of state
Autonomic maturity
Reflexes
Light
Rattle
Bell
Pin prick
Inanimate visual
Inanimate auditory
Inanimate visual and auditory
Animate visual
Animate auditory
Animate visual and auditory
General tonus
Pull-to-si t
Defensive movements
Activity
Peak of excitement
Rapidity of build-up
Irritability
Lability of states
Self-quieting activity
Hand-to-mouth facility
Tremulousnessb b
Amount of startle
Motor maturity
Total number of abnormal
reflex scores
aAdded ir. the present study from the NBAS-K {Horowitz, Sullivan,
and Linn, 1978}.
Receded to reverse the scale's direction, so that a high
degree of tremulousness or startle receives a low score.
27
-------
items which did not co-occur in these samples as initially pre-
dicted. High scores on the response decrement, orientation,
tonicity, regulation of state, and autonomic maturity clusters
represent more optimal behavior. A high score on the range of
state cluster indicates a combination of greater lability and
more intense arousal. A high score on the reflex cluster indicates
a greater number of abnormal reflexes.
Control Variables
k comprehensive list of 68 potential confounding variables
was assembled to test whether the observed relationships are
spurious and attributable to factors other than PCB exposure
(Table 8). Based on information obtained from the maternal
interview and maternal medical records, the list contained data
pertaining to demographic background, health history, pregnancy
and delivery, obstetrical anesthesia, and exposure to other drugs,
such as caffeine, nicotine, and alcohol. Cord blood levels of
polybrorcinated biphenyls (PBSs) wore provided by the Michigan
Department of Public Health laboratory. Multiple scores were used
to verify certain information, e.g., gestational age, perinatal complica-
tions, while health-related data were included if they appeared
in either source, the medical record or maternal interview. In
addition, an Obstetrical Complications Scale COCS) (Littman &
Parrnelee, 1974a), modified to omit any items (e.g., infections
during pregnancy) which could be.consequences of PCB exposure,
28
-------
Tal»le 8
Potential Confounding Variables
Background
~Socioeconomic status (SES)a
Marital status (married/unmarried)
Race
*Maternal age
*Matcrnal education (years)
Maternal weight (prior to pregnancy)
*Maternal height
*Maternal body mass^
*Sex of Infant
*Parity (first born/later born)
*Gravidi ty (prlnmgravlda/nultigravida)
Multiple birthC
Pregnancy
*Strass (e.g., family illness, narital probleas, unemployment^
^Prenatal care (month began)
*Birth classc
*Diet (milk and cheese consuned during pregnancy)
Surgery during pregnaneyC
Exposure to radiation during prggpg3cyc
*Weight gained during pregnancy (net weight gain)
*Weight gained during last 20 v^eks of pregnancy (aet weight gain)
*Welght loss during pregnancy (total weight lost)
29
-------
fable 8 (continued)
Reproductive hisCory
Infertility problemsc
Number of previous multiple births (i.e., twins)
familial birth defects or anomalies c
Delivery
Cephalopelvic disproportionc
Rh antagonism or other ABO incompatibilityc
•Time membranes ruptured (hours prior to delivery)
~Length of labor (hours)
~Mode of delivery (spontaneous/Caesarian section)
, *Meconiua present in amniotic fluid or meconium staining0
Fetal presentation (vertex/other)
Abnormal fetal heart-rate during laborc
~Cord around neckc
Placental infarction, placenta previa, or abruptioc
Hypoxia or anoxiac
~Obstetrical complications score (modified)d
Drugs during pregnancy
~Alcohole
~Caffeine ^
~Nicotine (packs of cigarettes per day)
Marijuana
Hashish
Cocaine
LSD
30
-------
Table 8 (continued)
Drugs during pregnancy (continued)
Amphetamines
Heroin
Drugs prior to pregnancy
~Alcohol6
~Caffeine*
~lieotine (packs of cigarett es per day)
Anesthesia and analgesia during labor and delivery
Meperidine (intramuscular) (ng)
Meperidine (intravenous) (eg)
~Viataril (intramuscular) (mg)
Vistaril (intravenous) (mg)
~Carbocatne (pudendal)0
~Xylocaine (pudendal)c
Carbocaine (uterosacral)C
Xylocaine (uterosacral)c
#
Spinal0
Epiduralc
General anesthesia0
Carbocaine (local)c
~Xylocaine (local)c
~Oxytocin (hours)
Folybroainated biphenyl (HBB) (cord seruo sample)(ng/mL)
Procedural
1. Infant age at Ballard examination
2. Infant age at NBAS examination
31
-------
Table 8 (continued)
Note: Items with an asterisk met the criterion that no more than 85
percent of the sample fell In any stogie category.
a8ased on Hollingshead index, n.d.
bRQCo, „ oni,_,nri. weight (kg)
Based on equation. (_J 1.5
c
Recorded as present or absent.
^Modified at-risk score base! on 24 OCS items which are not likely to be
affected by PCBs (e.g., marital status, sode of delivery, chronic illnesses).
®Mean daily intake of absolute alcohol (AA) was estimated from the re-
ported rate e£ consumption of beer, wine, and hard liquor and calculated
according to the method described by Kuzoa and Kissinger (1981). All
servings of a given alcoholic beverage were assumed to be the same size
(i.e., 12 oz. for beer, 5 oz. for wine, and 1.25 oz. for hard liquor) and
to contain a constant proportion of alcohol by volume (i.e., 0.035 for beer,
0,15 for wine, and 0.45 for h«rd liquor).
^Caffeine consumption was based on a weighted sua of reported consumption
of coffee, tea and cola; decaffeinated coffee and herbal tea were not in-
cluded. It was assumed that the average size of a cup of coffee is about
225 n>L. Since previous research has reported that consumption is divided
roughly evenly among instant, percolated, and drip coffee (Barr, Streiss-
guth, Martin, & Korst, 1981), the average tag of caffeine per cup of coffee
was estinated by taking the oean for instant (74 mg)» percolated (74 rag),
and drip or filter (1)2 mg> (Gilbert, Marsboan, Schwieder, A Berg, 1976),
Since tea and cola contain roughly one third of the average caffeine con-
tent of coffee, total daily caffeine consumption was estimated by assigning
coffee three Ciaes the weight of the other two beverages.
32
-------
provided a summary score assessing the degree to which the infant
was medically at-risk. The modified OCS, based on 24 items
relating to maternal age, parity, mode of delivery, drugs, etc.,
assigns a score to each mother, with higher scores reflecting
fewer complications.
Interval scale variables, such as socioeconomic status (SES),
maternal age, number of chronic illnesses, amount of caffeine
consumed daily, presented no problems for the control analyses.
However, to ensure sufficient variance in the case of dichotomous
variables, such as marital status {married/unmarried} or parity
(first born/later born), only those variables were included for
which incidence in each category exceeded 15 percent. Given this
criterion, it was possible to assess the effects of 36 potential
confounds.
The relation of each potential confounding variable with each
independent variable (reported fish consumption or biological
measure) was examined in a Pearson product moment correlation
analysis. Wherever a potential confound was significantly related
33
-------
to a given independent variable at an alpha level less than or
equal to .10, the confound was held constant in all analyses
performed on that independent variable. For each multiple
regression analysis we report the results of PCB
exposure after the influences of all potential confounds have
been partialled out. Similarly, the means reported in the
dose-response analyses have been adjusted for the effects of
the relevant potential confounds by using analysis of covariance.
Reliability
Staff members were tested for reliability in filling out
maternal and infant medical forms with information obtained from
the hospital records. Median reliabilities were 90 percent for
the maternal and 96 percent for the infant medical records.
These estimates are conservative because three examiners
had to agree for an item to be counted as an "agreement."
The coding of data into numerical categories was checked
by a second coder who recorded all disagreements. These dis-
agreements were then resolved by investigator J. J. After being
entered on the computer, data were checked further with a pro-
gram from the University of Michigan's OSIRIS IV computer
package which inspects each variable in the file and flags
illegal codes that may have bean entered by mistake.
Variable Trans formations
Each variable was checked for normality of distribution.
Twelve variables were highly positively skewed (skew >2.0):
PCB-fish consumption (highest rate); PCB-fish consumption during
9
34
-------
pregnancy; weight loss during pregnancy; time membranes ruptured;
length of labor; alcohol, caffeine, and nicotine consumption
during pregnancy; alcohol and caffeine consumption prior to
pregnancy; oxytocin; and age at NBAS examination. These vari-
ables were normalized by means of log transformations. Biologi-
cal measures of FCB and FBB exposure were also logged in all
analyses except those presenting means and standard deviations.
Results
Measures of Exposure
Laboratory results for cord serum, maternal serum,
and milk samples are presented in Table 9, The column labeled
"nondetectable" refers to PCB values of 0.1 ng/mL to five x the
noise level on the chromatogram; the right hand column refers to
PCB levels that are detectable, but still below the laboratory's
quantification limits. The majority of the PCB levels matched
to the 1016 standard were below quantification limits. This was
particularly true for the cord and maternal sera analyses.
Among these samples, only one of 195 and five of 190 cases,
respectively, were quantifiable. Since comparisons with the
1260 standard yielded considerably fewgr values below the quanti-
9 fication limit, 1260 values may provide a more reliable measure
of PCB body burden. Arithmetic means for the more highly
chlorinated isomers exceeded th^ ksver chlorinated isomers by a
ratio of 2,7 to 1 for the cord samples and as much as 5,5 to 1
for the neonatal breast milk^sjpr.ples. This differential presum-
ably results from the tendency of lower chlorinated isomers
35
-------
Tabla 9
PolychlorInatod Blphtnyli In Cord Sarum,
Maternal Seriwi. and Maternal Breast Ml lb (vholt Basil)
Moan (SPl
Median
Nondatactrhlo Detectable but
nonqusnt I r labia
m
(*}
Cord coron (ng/nL)
1016 standard
1360 standard
H«t«rna 1 mrun (ng/«L)
10 IS b tandnrd
1360 standard
Brenst milk * noonataI (ng/g)
1016 standard
1260 R tftndnrd
Breast milk - 5 months !ng/g)
lOtfi standard
1260 standard
195
198
too
IS6
138
130
49
<9
0.9 (0.9)
3.5 (1.9)
1.6 (4.5)
5.5 (3.7)
4.1 (4.5)
23.5 (12.3)
3.7 (3.4)
17.4 (13.1)
0.6
2.0
o.a
4.6
3.9
20.4
3.8
14. I
33 ( 11 .8)
4 (2.0)
16 (8.4*
0 (0.0)
•11 (8.0)
0 (0.0)
(6.7)
(0.0)
171 (87.7)
t30 (65.7)
169 (88.9)
44 (22.4)
85 (61.6)
2 (1.4)
30 (66.7)
4 (8.0)
Nondetectaole refors to values of 0.1 ng/nl to f 1 ve x no IS* level of the chroaatogra*
Nonquant I Mcble refers to detectable lava It below the quantifiable limits of S ng/ioL for Aroclor 101B and 3 rtg/»L
for Arorlor 1260.
-------
tc be less stable, more susceptible to degradation in the
environment, and metabolized more rapidly in the organism
(Matthews, in press).
PCB levels for a large proportion of the cord serum samples
were not quantifiable, even when the 1260 standard was used. A
pairwise comparison of PCB 1260 serum levels on a whole basis
indicates that mean PCB levels in maternal serum were
substantially higher than in cord serum (Table 10), confirming
Fujitani, Ohi &
previous findings (Akiyama,/Vagyu, T975; Kodama & Ota, 1977).
Masuda et al. (1978) found a similar differential in a recent
Japanese sample and concluded that the placenta say act as a
partial barrier to PCBs. to alternative explanation may be
related to the mixture of arterial ar.d venous blood in the cord
sample. If lipids and lipophilic substances such as PCBs are
retained by the fetus, then arterial blood will contain only
residual amounts of these substances, and PCI levels derived from
cord serum will under-estimate fetal exposure. Moreover, lipid
levels in maternal serum are twice thos® in cord serum and when
1>CB concentrations are calculated on a fat basis, the
differential is no longer significant. Whichever explanation
prevails, the evidence supports inference that PCBs cross the
placenta, thereby exposing the fetus.
37
-------
Table to
Palrwito ConporIsons of PCBs and Lipids In Cord Serut*,
Maternal Sarin, end Maternal Breast Milk (1260 Standard)
Whole basis (ng/«L or ng/g) Fat basts (ng/g) Lipids
H
Mean
i
Mean
(SO)
A
*
t
Cord flerum
Maternal sorum
148
2.A
5.3
( t .9)
(3,7)
-10.IB*
788, 1
635.3
(696.5)
(557,1)
-0.02
0,32
0.64
-25.69*
Cord serum
Breast milk * neonatal
95
2.4
32.8
11-01
(13.0)
-15.45*
C24 .2
813.9
(714.2)
(377.9)
O. 13
0.31
2.89
-20.45*
Maternal serum
Breast milk - neonate 1
9t
5.6
22 * 7
(3,2)
( 12,3)
-13.36*
932.2
827.8
(579.1)
(380.*)
1.55
0.63
2.85
-17.06*
Breast mtIk - neonatal
Breast mlTk - 5 month
4a
2Q + Q
18.0
(9.3)
( 12.2)
1.31
750.5
772.3
(40&.8)
(671.2)
-0. 19
2.96
2.82
O.S1
Cord and mstarnal »«ra values on a whole Da*Is aro given In ng/mL end breast Milk values In ng/g.
*E < .0001
-------
Table 11
Intercorrelations of tog PCis (1260 Standard)
in Cord Serum, Maternal Serum, and Maternal Breast Milk
Cord
Maternal
Breast milk
Breast milk
serum
serum
neonatal
5-month
Cord serum
.41***
.09
-.06
(148)
(95)
(25)
Maternal serum
.43***
__
.23*
.01
ft
(145)
(91)
(30)
(?rennt tfiillk - neonatal
.16
.28**
--
,61***
V
(93)
(91)
(42)
Breast milk - 5 month
-.01
-.10
.30*
...
•
(24)
(30)
(42)
Note, Rs above the main diagonal are values calculated on a whole bapis, while
those below are on a fat basis. N appears in parentheses. Lipid values were
missing for three cord serum, two neonatal breast milk, and one 5-month breast milk
samples.
< .05
**£ < .01
*•*2 < .0001
-------
Correlations between maternal and cord sera were similar in
magnitude, whether calculated on a whole serum or fat basis, r «=
.41 and .43, respectively (Table 11). while the whole and fat
basis findings were also similar for correlations between
maternal serum and breast milk, the two approaches yielded
different results for the neonatal to 5-month breast milk
correlation.
Consistency between analyses derived on a whole and fat
basis was also indicated by the pattern of the intercorrelations
shown in Table 12. PCB levels in the cord and maternal sera on a
whole serum basis were highly correlated with levels calculated
on a fat basis (_r ¦ .93). Thus, for serum, it seems to make
little difference whether measures are derived on a whole or fat
basis. Although cord and maternal sera are low in lipids, lipid
levels tend to be relatively stable. Because breast milk lipids
fluctuate considerably, the same consistency does not appear for
breast milk values derived by the two methods. This apparent
instability suggests that in the analyses of breast milk,
measures that take 1ipid fluctuation into account are likely to
be more reliable.
41
-------
Table 12
IrstercorrolatlefW of Lag PCBt (I2RO Standard)
In Cord Sorum, HimirnsI Serum,
and MatornaI Breast Milk
(Whole and f-t Baals)
Cord
Borum
Vhol • basil
HotornaI
serum
Broost >1Ik-
neonatal
Breast lit I k-
5 month
ro
Fat basis
Cord BBrufi
Ma t ernaI amrvm
Breast ml ik-
neonatal
Breast »llk-
5 month
.93***
(194)
. 43*••
( 140 J
, is
(95)
-.06
(S3)
.37*••
(145)
,83***
( 196)
.35*••
(91)
- .08
(30)
.01
(93)
.20*
(SI )
,63*»*
( 138)
.08
(4S!
• OS
(24)
.01
(30)
. 30*
(42)
. 39**
(45)
Not.». N appears In parenthasaa.
*1 < ,10
•e < .05
**g * 01
***e < .001
-------
These findings were further supported by an analysis of the
degree to which PCB levels could be predicted from lipid levels
in serum or breast milk, while the PCS levels in the cord and
maternal sera were unrelated to corresponding cord and maternal
lipid levels, there was a significant association between PCB and
lipid levels in the neonatal and 5-month breast milk {£ « .58, j>
< ,0001, and £ ¦ .63, g < .0001, respectively).
Women and their offspring were selected for the study on the
basis of reported fish consumption patterns. Levels of fish
consumption reported by the exposed and control mothers for whom
the different biological measures were available are shown in
Table 13. When women who breast-fed were compared with those who
did not, no difference was found for highest annual rate or
cumulative consumption measures. While their rate during
pregnancy was slightly lower than the rest of the sample (M - 2.3
vs. 3.8 PCB-kg/year), the difference failed to reach conventional
levels of statistical significance (t -1.90, g < ,10).
Correlations between maternal reports of contaminated fish
consumption and the four biological measures of PCB body burden
are presented in Table 14. No^relationship was found between
cord serum PCB levels and maternal fish consumption, possibly
43
-------
Toble 13
Invals or r.fjortmd Contaminated Flah Consumption (PCB-kg)
Of Vorjttn Providing Blood ni*J 5r««"t HI lk Samplaa
N
Mean
Median
* lingo
Cord ••rum onslydBS
Conmjmpt Ion at
Mghost rata
190
4,3
<5.5)
3.0
0.0 - 41.7
Copaurspt ion
durlnjj prsgnaney
189
2.8
(3.9)
1.9
O.O - 22.4
Cuoulst1v« consumption
198
61.3
(37.«)
27.8
O.O - 716.«
Matarna1 larun analyse*
Consumption at
hlghost rata
• Si,
4.4
(5.3)
3.2
0.0 - 41.7
Consumption
during prognoncy
190
2.8
( 3 . <5)
2.2
0.0 - 22.4
Cumulative consumption
196
G2.4
(31.5)
28.6
0.0 - 716.6
Breast milk - neonate)
ConBumptIon at
highest rots
138
4,7
(5.3)
3.1 '
O.O - 41.7
Co ¦"•sump t Ion
during prugnancy
133
2.9
<3 5)
3.2
0.0 - 20.0
Cumulnllva consuwpt ton
US
(2.9
(0O.9)
38,4
0.0 - 716.6
Brcnnt milk - 8 nonthi
Consumption at
highest rate
45
S.S
C8.0)
3.8
O.O - 41.7
Consumption
during pregnancy
43
a.#
(2.9)
2.2
0.0 - S.S
Cu»ul«tlv» consumption
45
82.5
((35.3)
43.1
O.O - 716.6
Note. Moan consumption love)s Include dntn on «xpos«d and control wonnn.
-------
because of the detection problems in cord serum analysis
described earlier. In contrast, a positive correlation was found
between PCB levels in maternal serum and all three reported fish
consumption measures. Mean PCB level in maternal serum was 6.1
ng/mt {SD « 3.7) among women who ate contaminated fish, as
contrasted to 4,1 ng/mL (SD = 2.7) among those who abstained (t =
3.83f £ < .0001). These data replicate in a sample of women of
child-bearing age Humphrey's {1976) finding that Lake Michigan
fish consumption predicts PCI body burden in Michigan residents.
The data extend this finding by showing a significant correlation
between fish consumption and PCB levels in neonatal breast milk.
Mean PCB residues were higher in breast milk samples of exposed
than in samples of nonexposed women (K ¦ 865.6 vs. 622.2 ng/g on
a fat basis, _t = 3.21, g < .005), The lack of a consistent
relationship between maternal report obtained at birth and breast
milk levels at 5 months may be due in part to postpartum
ingestion of contaminated fish, which will be examined in future
analyses. Figure 2 shows a scatterplot demonstrating the linear
relationships between maternal serum 1260 and fish consumption at
the highest annual rate. For every pound of contaminated fish
»
ingested by the mother, the PCBs in her serum increase by about
0.15 r,g/nL on a whole basis or 23.6 ng/g on a fat basis.
45
-------
Table 14
Correlations batman Log Contaminated F tsh Consumption and
Loo PCBs In Cord Serum, Hatarnal Scrum, and Breast Ml IK
(12SO SlMcfard)
Regression
N r cquat ion »
F 1 ah consumption at
tifgitost annual rate with
Cord (whole
-------
20
a
m
<
o
z
iS
15
*•
S
3
K
Sf
z
cc
ui
H-
<
s
10
III*
t •
I « *J«*
I
*
* * • • •
•J • • t •
2*33 *
*3*1*
•••21• 2 •
* j •• •
u-
o
10
15 20 23 30
35 40
45
PCB~hg/ye#r
FISH CONSUMPTION AT HIGHEST RATI
Figure 2.' Scatter plot of maternal serum (1260 standard)by fish consumption at Mphost rato.
-------
Infant Outcomes
Infant outcomes were examined Cor two measures of exposure—reported
maternal PCB-Hsh consumption at the highest race and cord serum PCB level
(1260 standard). As the best predictor of maternal serum and breaat milk
P.?Bs, fiaii consumption at the highest rate may provide a useful estimate of
intrauterine exposure. Although PCB levels in cord senna are a mora direct
aeasure of infant exposure, most values are below the quantification limit
and, therefore, this measure my lack the reliability needed to detect subtle
effects. Since toxins other than PCBs are also found in Lake Michigan fish,
however, cord serum was used to corroborate findings obtained from tb#
analyses of maternal fish consumption.
Table 15 presents the correlations of 36 potential confounding variables
with each of these exposure measures, indicating those at an alpha level less
than ot equal to .10. Maternal PCS-fish consumption is correlated with
caffeine consumption before and during pregnancy and with alcohol consumption
before and during pregnancy. Cord serum PCB level is associated witn alcohol
and caffeine consumption during pregnancy, weight gain during pregnancy, and
maternal age. Polybroninated biphenyls (PBBs), a chemically related compound
*
which entered the Michigan food chain through the accidental contamination of
cattle feed, is not correlated with either measure of PCB exposure. Serum
PBB levels are consistently low in this sample, H - 0.4 jig/oL, SO » 0.7.
Since eight correlations at the £_ < .10 significance level is about what
would be predicted by chance, it seems unlikely that the correlations In
Table 15 will replicate beyong the present sample. It is necessary to
control for these potential confounds in the present analyses, however, in
48
-------
fable 15
Correlations of PCB Exposure Measures
with Potential Confounding Variables
PCB-fish Cora
Consumption Serus
(highest rate) PCB level
Background
Socioeconomic status - -.08 ,07
Maternal age .01 .12+
Maternal education -.05 .01
Maternal weight -.09 .11
Maternal height -.02' .06
Maternal body mass -.08 .09
Sex of infant .0? -.07
Parity .02 .04
Gravidity .09 -.01
Pregnancy
Stress -.01 -.OS
Prenatal care -.05 -.02
Birth class -.00 -.04
Diet .04 -.01
Weight gain (total) -.03 -.14+
Weight gain (last 20 weeks) .06 -.09
Weight loss .01 .08
49
-------
Table 15 {continued)
PCB-fish Cord
consumption serua
(highest rate) level
Delivery
Tine membranes ruptured -.04 -.08
Length of labor (hours) -.04 -.01
Mode of delivery .02 -.04
Mecoaiua -.00 -.09
Cord around neck -.05 -.00
Obstetrical complications -.06 -.04
Drugs during pregnancy
Alcohol .14* .14+
Caffeine .10+ .12+
Nicotine .04 ,09
Drugs prior to pregnancy
Alcohol .17** .10
Caffeine ,12* .02
Nicotine .02 .11
Anesthesia and analgesia
Vistaril (intramuscular) -.01 .09
Carbocaine pudendal -.06 ,01
Xylocaine pudendal -.00 ,02
Xylocaine local -.08 .05
Oxytocin, .03 -.06
50
-------
Table.15 (continued)
PCB-fish
Cord
consumption
serum
(highest rate)
PCB level
Polybrooinated biphenyls (EBBa)
.08
.04
Procedure
Age at Ballard examination
-.03
-.00
Age at NBAS examination
.09
-.02
Notes More detailed definitions of these variables are presented in
Table 8.
+£ < -10
* £ < .05
** £ < .01
51
-------
order Co insure that PCI* effects found in the present sample are not
attributable to these related variables.
Effects on Birth Size and Gestational Age
The relation between maternal PCB-fish consumption and birth size is
shown in Table 16. The F-ratios test the statistical significance of each
predictor after the effects of the other four predictors have been partialle
out, two measures of newborn sise—birth weight and head circumference—are
significantly correlated with asternal PCB-fish consumption, with higher
levels of contaminated fish consumption predicting smaller birth size.
In order to examine dose-response relationships, PCB-fish consumption
was divided into four categories. Mothers who reported having eaten Lake
Michigan fish were divided into three approximately equal-sized groups:
(a) high exposure (greater than 6.5 kg per year); (b) moderate exposure
(3.5-6.5 kg); and (c) low exposure (1,9-3,4 kg). The fourth group consisted
of mothers who did not eat Laki Michigan fish. Birth weight decreases in a
dose-related fashion at each level of additional PCB exposure (Figure 3).
Even after adjusting for the effects of the four potential confounds, the
birth weight of the most highly exposed infants averages 245 g less than tha
of the nonexposed controls. Head circuaferenc«s also decreases in a dose-
-------
Table 16
Maternal Contaminated Fish Consumption and Cord Serum
PCB Level; Effects on Birth Size and Gestational Age
Head Gestational
Birth Weight Circumference Am Ballard
Contaminated Fish Consumption
Controlling for potential confounds
•Contaminated fish consumption
-.21
13.01***
-.20
10.67**
-.15
5.67*
-.30
16.05***
Alcohol prior to pregnancy
,05
<1
.15
4. PI*
.06
1.38
.06
<1
Alcohol during pregnancy
.01
<1
-.05
<1
i
*
o
o*
<1
-.14
2.20
Caffeine prior to pregnancy
-.13
3.17
-.07
<1
-.14
3.04
.14
2.24
Caffeine during pregnancy
.02
<1
-.02
<1
.11
2.37
-.05
<1
Controlling for gestational agea
and potential confounds^
Contaminated fish consumption -.16 7,69* -.19 7#97*
Cord Serum PCB Level
Controlling for potential confounds
Cord serum PCB level
-.16
5.25*
-.18
6.53*
-.19
5.16*
-.07
<1
Alcohol during pregnancy
.03
<1
,04
<1
-.07
<1
.11
<1
Caffeine during pregnancy
-.04
<1
.01
<1
.09
1.27
.03
<1
-------
Table 16 (continued)
Birth Weight
6
Head
Circumference
Bb FC
Gestational
a
Age
b c
0 F
Ballard
Controlling for potential confounds
(continued)
Weight gain during pregnancy
.23
11.11**
.16
5.30*
.12
1.98
.04
<1
Mother's age
.03
<1
.02
<1
-.02
<1
-.14
1.47
Mother's prepregnancy weight
.27
15.18***
.15
4.85*
.12
2.06
.07
<1
Sex of infant
-.14
4.21*
-.36
26.51***
.04
<1
-.08
<1
Controlling for gestational age.
and poterfiHal confounds^
Cord serum PCB level
-.09
1.73
-.15
3.86
Based on reported last menstrual period.
^Standardized regression coefficient from the final step of the regression.
c
Tests the unique variance associated with each predictor, i.e., the degree to which the predictor
adds significantly tc the explained variance when entered last in the regression.
dRegression coefficients for the potential confounds in this analysis are available on request.
*j> < .05
**£ < .01
***£ < .001
-------
3600
rr, 3500
3400
as
3300.
weeks
o
"g
0>
-------
related fashion, with the head circumference of the most higlily exposed
infanta averaging 0.7 cm less than the controls. Head circumference is not
disproportionately small, given the low weight of these infants. With birth
weight held constant, the effect of fiin consumption on head circumference
drops below conventional levels of significance, JF (1, 272) m 2.92, £ > .05.
Contaminated fish consumption also predicts a shorter gestational period
on both the last menstrual period measure and the Ballard examination (Table
16). Greater exposure is associated with greater immaturity on both sub-
scales of the Ballard examination: F_ (1, 167) » 10.67, < .001 for
neuromuscular maturity; J? (1, 164) ¦ 8.44, _£ < .01 for physical maturity.
Gestational age decreases in a dose-related fashion, with the most highiy
exposed infants averaging gestational periods that are 6.3 days shorter than
the nonexposed controls on both gestational age measures (Figure 3).
Since birth weight and gestational age are moderately intercorrelated
(r * .36 and .46 for the last menstrual period and Ballard measures,
respectively), it is possible that the effects shown in Table 16 are not
independent. For example, the smaller birth size fowid to be associated with
PCB exposure could actually be a consequence of the exposed infants' earlier
birth. Conversely, the apparent immaturity indicated by the Ballard examina-
tion may merely be a by-product of the infant*1 smaller birth size. These
hypotheses were tested in multiple regression analyses in which the effect of
fish consumption on each of these variables was examined after controlling
-------
for the effects of the others. With gestational age based on last menstrual
period statistically controlled, fish consumption continues to predict both
lower birth weight and smaller head circumference (Table 16), indicating that
the affected infants are smaller-for-date. Fish consumption also predicts
lower Ballard scores when birth site is held constant, £ (1, 203) " 4,96,
£< .05, suggesting that the exposed infants' poorer Ballard acores are not
attributable to their smaller size.
Because contaminants other than PCBs could be responsible for the
observed relations between fish consumption and birth size or gestational
age, we attempted to corroborate the findings in Table 16 using the cord
aerun PCB measure (1260 standard). Two variables, maternal prepregnancy
weight and sex of infant, were found to function as suppressor variables in
the present sample due to a tendency for cord P"B levels to be higher in the
offspring of heavier mothers and in male infants. Since the infanta of
Heavier mothers tend to be larger Or » .24, £ < ,001, and _r • .13, £ < .10,
with birth weight and head circumference, respectively) and since male
infants tend to be larger than female infants (_t (192) ¦ 2,34, £ < ,025, and
t_ (192) ¦ 5.31, £ < .001, for birth weight and head circumference, respec-
tively), these variables initially masked the effect of cord PCB level on
0 birth size, With maternal prepregnancy weight, sex of infant, and four
potential confounding variables statistically controlled, cord serum PCB
level predicts lower birth weight, smaller head circumference, and a shorter
period of gestation based on reported last menstrual period (Table 16).
Infants with cord serum levels greater than 5 ng/mL are 194 g lighter on the
average when compared with infants whose levels are below the laboratory's
-------
quantification limit (3 ng/mL). The head circumference of the more highly
exposed infants averages 0.3 cm less, and they are born on an average of 13,3
days earlier.
Effects on Neonatal Behavior
In order to assess the overall impact of PCB exposure on neonatal
behavior, PCB-fish consumption (adjusted for the effects of four potential
confounds) was examined ii, relation to the NBAS clusters. Fish consumption
was adjusted in the following manner. First, it was regressed on its four
potential confounding variables. The regression coefficients from that
analysis were then used to calculate a predicted fish consumption score for
each mother, given her scores on the confounding variables. The predicted
score was subtracted from her actual score in order to remove the variance
attributable to the confounds. This adjusted score was then regressed on six
of the HAS clusters. The seventh cluster, response decrement, was omitted
due to a large number of missing observations.
PCB-fish consumption (adjusted for the four potential confounds) pre-
dicts a linear combination of five NBAS clusters, JF (5, 193) - 3.18, j> < .01,
with a multiple correlation coefficient (R) of .28 (see Table 17). The
strongest relationships are with autonomic maturity, reflexes, and range of
state. Each of the items in the autonomic and range of state clusters was
examined in a separate regression analysis similar to those in Table 16.
PCB-fish consumption predicts two of the three autonomic items: motor
maturity, J? (1, 280) * 6.93, £ < .01; and amount of startle j? (1, 280) «
10.07, £ < .005; and one of the four range of state items; lability of
states, J_ (1, 279) ¦ 14.46, £< .001, Motoric immaturity (Scale points 1, 2,
58
-------
fable 17
Maternal PCB-Fish Consumption: Effects on Neonatal Behavior
Raw
regression,
coefficient
Standardized
regression
coefficient
L
Autonomic, maturity
-0.2
-.17
5.85**
Reflexes
0.1
.14
3.57*
Range of state
-0.2
-.14
3.55+
Orientation
-0.0
-.06
<1 •
Regulation of state
-0.0
-.04
<1
g
Tonicity
Constant
1.7
Note; Based oa a multiple regression analysis in which fish consumption
(adjusted for the effects of caffeine consuaptioc before and during pregnancy
and alcohol consumption before and during pregnancy) was regressed on six
NBAS clusters.
aF-to-enter at the last step was too small to permit entry into the regression.
*£ < .07
*£ < .05
**£ < .025
59
-------
and 3) is defined as jerky, unbalanced, cogwheel-like movement in which
flexor and extensor muscles mm, to be competing. Eleven percent of the
highly exposed infanta are classified as immature in this sense, as compared
with only 3Z of the nonexposed controls. It has been suggested that fewer
than six state changes during the NBAS examination indicates deficient
lability of state (Ala et al., 1979, fable 6-1). Given this criterion, 42%
of the highly exposed infants are hyporesponsive compared with only 17% of
the nonexposed controls.
The 17 reflex items were not examined individually. Although the reflex
cluster exhibit® moderate internal consistency reliability (J. Jacobson et
al., in press), individual itecs administered in a single examination are not
likely to be reliable. Th? reflex items were therefore factor analyzed in
order to assess whether particular subgroups of reflexes might be influenced
by PCB exposure. Three factors each account for at least 10X of the variance
in the reflex cluster scores: (a) reflexes associated with the legs:
standing, walking, placing; (b) reflexes associated with the mouth; rooting,
# sucking; and (c) a third group consisting of Babinski, passive movements, and
tonic deviation. PCB-fish consumption is related to only one of these
aubgroupsj reflexes associated with the mouth, _F (1, 281) » 7.22, j» < .01.
A second analysis differentiated reflexes which are abnormally strong oi
hyperactive from those which are abnormally weak or difficult to elicit.
P3B-fish consumption is significantly related to number of abnormally weak
reflexes, F (1, 281) - 9.63, £ < .005,-
60
-------
Als et al. (1979) have proposed a set of a priori clusters which classi-
fy neonatal behavioral performance qualitatively as being either optimal,
aoroal, or worrisome. The "worrisome" category describes a behavioral
pattern that i» sufficiently deviant to warrant the concern of the experien-
ced clinician. The Als et al. criteria for classifying each of the NBAS
items (1979, Table 6-1} were applied to the seven clusters used in the
present study.^ As indicated in Table 18, maternal fish consumption is
highest uraang infants classified as worrisone on the three clusters
associated with PCB exposure. On the range of state cluster, where two
categories of worrisome behavior are defined, exposure is highest among the
infants classified as flat or depressed. Tables 19-21 compare the aost
highly exposed infants with the non-exposed controls. High levels of
contaminated fish consumption increase the likelihood of worrisome
performance on all three clusters.
Since small birth size and shortened gestation nay themselves be associ-
ated with neonatal behavioral deficits, it is possible that the behavioral
r
effects found here are not direct consequences of exposure to PCS but merely
by-products of PCB effects on birth siae and gestational age. This hypothe-
sis implies that, when birth sine and gestational age are statistically
controlled, the effect of exposure on behavioral outcome will no longer be
significant (Kenny, 1975). Partial correlation analyses (Table 22) indicate
that the behavioral effects of contaminated fish consumption are not
61
-------
Table 18
Maternal PCB-fish Consumption (kg/yr.) for Infants
Classified as "Worrisome" on Three NBAS Clusters
Worrisome
Optimal
Normal
Worrisome
Excessively Flat,
labile depressed
F
Autonomic
maturity
__a
5.0
9.3
4.88*
(277)
(10)
Reflexes
3.7
4.7
7.7
8.07**
(47)
(175)
(65)
Range of
state
5.6
4.5
__b
2.9 10.4
7.02**
(95)
(159)
(12) (20)
Note: N's are given in parentheses.
aNo optimal criteria are designated "by Als et al, (1979) for autonomic
maturity.
^Two categories of worrisome behavior are defined for range of state.
* 2. < -05
** n « .001
62
-------
Table 19
Worrisome Performance oa the Autonomic Maturity
Cluster by PCB-fish Consumption
Nonexj-osed High exposure
controls (>6,5 kg/yr.)
Normal 64 67
Worrisome 1 6
(1) » 3.18, £ <.0fl
63
-------
Table 20
Worrisome Performance on the Reflex Cluster
by PCl-fish Consuasption
Nonexposed High exposure
controls (>6.5 kg/yr.)
Optimal or normal 44 35
Worrisome 21 38
X2 CD - 5.48, 2. <-025
64
-------
Table 21
Worrisome Performance (Flat, Depressed) oa the
Range of State Cluster by PCB-fish Consumption
Nonexposed High exposure
controls (>6.5 kg/yr.)
Optimal, normal, or
excessively labile 64 65
Worrisome {flat,
depressed) 1 7
X2 (1) - 4.16, £ <.05
65
-------
Table 22
Partial Correlations of Adjusted Contaninated Fish Consumption
with Neonatal Behavior, Controlling for Birth Size and
Gestational Age
Zero-
Control
Variable
order
relation-
Birth
Head
Gestational
•Ballard
ship
weight
circumference
a
age
score
Behavioral outcome
(1-287)
(N-282)
(S>272)
(N-226)
(N-121)
Autonomic maturity
-.16**
-.16**
-.17**
-.15*
-.09
Reflexes
.17**
.IS**
.16**
.17**
.15
Range of state
O
»
J
-.08
-.08
-.09
-.09
Lability of states
-.23***
-.21***
-.23***
-.24***
-.23***
Based on reported last menstrual period
*£ < ,05
**E < -01
***£ < .001
§6
-------
predicted by either birth siae or gestational age based on last menstrual
period. Controlling for the size and gestational age of the newborn does not
appreciably alter the relation of contaainated fish consumption to HBAS
performance.
The pattern Is less clear for the Ballard examination. Although
controlling for the Ballard has little effect on the relation of contaminated
fish consumption with lability of states or reflexes, the effect on autonomic
maturity is substantially reduced when the Ballard is held constant (Table
22), Nevertheless, it does not seem appropriate to conclude that this effect
is mediated by shortened gestation. Unlike gestational age based on last
menstrual period, it is possible that the Ballard scores are themselves
influenced by neonatal behavioral deficits. Ballard et al. (1979) note that
certain disease states and intrauterine environmental factors may retard
physical and neurological maturation, thereby causing affected infants to
appear gestationally younger than they are. Thus, the newborn who exhibits
motoric immaturity or a propensity to startle is likely to be scored
behaviorally or physically younger on the Ballard exaa. The partial
correlation analysis should, therefore, not- be interpreted as evidence that
gestational age mediates these behavioral outcomes, since low Ballard scores
and poor autonomic functioning easy both be due to neurobehavioral immaturity
resulting from toxic exposure.
67
-------
Evidence that the physical and behavioral effects of PCBs are indepen-
dent suggests that some affected infants are small while others exhibit only
behavioral deficits. The total impact of PCB exposure in the present sample
can be estimated most completely by examining the relation between adjusted
fish consumption and a linear combination of birth sice, gestational age, and
behavioral variables. The multiple correlation between fish consumption and
neonatal outcome emerging from this analysis is .34, JF (7, 152) " 2.80,
£ < .01.
An attempt was made to corroborate the behavioral effects associated
with sttternal PCB-fieh consumption in analyses based on the cord serum
Measure. Serum PCI level (adjusted for its four potential confounds) is not
significantly correlated either with a linear combination of six NBAS clus-
ters, F (5, 118) ¦ 1.13, or with any of the individual clusters. Further
examination of the data did not reveal any control variables that might be
suppressing the relation between cord serum PCB levels and neonatal behavior.
Discussion
Measures of Exposure
Despite their widespread distribution in our environment and concern
about huaan ac:umulatiem» few studies are currently available on the serum
PCB levels of individuals chronically exposed to PCBs via their diet. More-
over, the toxicology of PCBs is little understood in comparison with that of
other xenobiotic chemicals, such as DOT and BHC (Hirsysoa, 1976; Yoshtmura &
Toshioura, 1976). The present longifucrinai study was designed to investigate
effects of PCB exposure from contaminated fish consumption on two potential
at-risk groups, pregnant women ae3 their newborn offspring. While several
68
-------
reports have appeared on the transfer of PCBs from mother to child via breast
milk (e.g., Itodama & Ota, 1980; Kuwabara, Yakoshiji, Vatanabe, Yoshida,
Koyama 4 Kuni ta, 1979), only a few studies examined the transfer of PCBs in
otero (Akiyama et al., 1975; Rodana & Ota, 1980; Masuda et al,, 1978). These
studies used either relatively snail unexposed samples or women and infants
exposed during the Yusho incident. No study to date has examined the pattern
of transfer from mothers to offspring exposed via the food chain.
The data in this report provide empirical support for linkages between
source of contamination and body burden shown in the analytic model (Figure
1). They also provide evidence of the preponderance of more heavily chlori-
nated isomers in the serum and breast milk of women whose exposure Co PCBs is
due in part to the ingestion of contaminated fish. The small amounts of 1016
present in the biological specimens and the poor reliability of their measure-
merit indicates that values matched to the 1260 standard are more likely to
yield meaningful findings.
The present study evaluated the linkages from fish consumption to
maternal body burden and subsequently to the developing fetus and breast
milk. Even though exposure levels are only moderate in the women studied,
reported contaminated fish consumption predicts PCI residues in two measures
of body burden, maternal serum and breast silk. These results confirm prior
studies which found an association between fish consumption and PCB residues
in serum specimens (Kreiss et al., 1981; Humphrey, 1976), while also demon-
strating that the relationship holds for younger women, an age group not
previously studied. The particular significance of this finding is that PCBs
in the environment may pose a risk to the fetus and newborn, since the women
in this age group are in the middle of their child-bearing years.
-------
Transplacental passage by the PCS molecule was indicated by the
significant positive correlation between maternal and cord sera (£ • .41, jj <
.0001), confirming two prior studies which yielded similar fir.dings (r » .24,
Masuda et al., 1978; jr ¦ ,45, Kodama & Ota, 1977). The transfer of PCBa to
bresst aiIk wag confirmed by a similar positive correlation (jr * .27, £ <
.01, for maternal serum on a whole basis to neonatal breast milk on a fat
basis), confirming Masuda et al.'s finding (r ¦ .29, 1978). While cord and
maternal sera yield similar results regardless of whether the analysis is
perforwd on a whole serum or fat has*#, breast milk residues yield varying
results due presumably to fluctuations in lipid levels in breast milk.
The present data are consistent with the notion that PCBs move through
the environment and the human body into systems which may potentially
threaten the infant both before and after birth (Swain, 1981). At this time
it is not clear which of these exposures—prenatal (intrauterine) or post-
natal (via breast milk)—has the greater itapact on infant development. While
the absolute quantity of PC3 residues is substantially lower io cord serum
than in breast milk, the fetal organism is particularly vulnerable during the
prenatal period. One factor overlooked in this debate (e.g., Kasuda et al.,
1978) is the size o£ the fetus. When PCB-exposure is calculated on the basis
of body weight, the infant's prenatal exposure is substantial. In addition,
the values for PCBs in cord serum may underestimate actual levels of exposure
for reasons discussed earlier. Moreover, different biological systems in the
fetus and newborn may be affected depending on the developmental period
during which exposure occurs.
70
-------
Methodological Coastderations
As measures of intrauterine exposure, reported contaminated fish
consumption and cord serum PCS levels each have disadvantage#. Since toxic
chemicals other than FCSs are also found in Lake Michigan fish, the
association of fish consumption with adverse neonatal outcones needs to be
corroborated by a more direct measure of PCI body burden. However, at the
moderate exposure levels exaatned in the present etudy, the majority of cord
serum samples yield PCI values below the laboratory's qualification limit.
Given the uncertain reliability of the cord serum measure, reported fish
consumption nay provide a more accurate measure for moderately-exposed
infants. It aay also provide a better measure of chronic exposure, since
cord aerua is saspled only at birth.
Since ethical considerations do not permit the random assignment of
humans Co experimental conditions, the present study used a correlational
design. Thirty—five potential confounding variables were tested for an
association with fish consumption and cord serum PCBs. In the analysis of
infant outcomes, variables with £ values <.10 were controlled using either
multiple regression or analysis of covariance. Other potential confounds,
such as socioeconomic statue, maternal Slacking, parity, obstetrical complica-
tions, and exposure to Plls are unrelated to maternal fish consumption and
cord serum PCB levels. These variables, therefore, cannot account for the
adverse neonatal outcomes associated with PCB exposure. With relevant
confounding variables controlled, maternal fish consumption predicts adverse
outcomes in three areas; birth size, gestational age, and neonatal behavior.
71
-------
Birth Site and Gestational Age
Intrauterine exposure to PCBs is aost clearly associated with reduced
birth sise and a shorter period of gestation. The oast highly exposed
Infants in this sample are about 200-250 g lighter than the nonexposed
controls, whether exposure is measured hy maternal fish consumption or cord
serum PCB levels. Similar effects hold for head circumference and gestation
al age based on reported last menstrual period. All three measures decrease
in a dose-related fashion with increasing levels of FCiJ exposure. Evidence
of reduced birth size and shortened gestation in this chronically-exposed
sample is consistent with research on acutely-exposed humans in Japan
(Fuaatsu et al., 1972) and with experimental studies using rhesus monkeys
(Allen et al., 1980),
The birth weight effect found here is of the same order of magnitude
reported by the Surgeon General (1979) for smoking during pregnancy; the
infants of smokers are on the average 200 g lighter than those of nonsniokers
In contrast to smoking, however, there is no evidence that abstaining from
fish consumption during pregnancy will reduce this effect, since PCBs
(especially the more highly chlorinated isomers that are matched to the 1260
standard) accumulate in human tissue over tine and are difficult to
eliminate.
Neonatal Behavior
PCB-fish consumption also predicts neonatal behavioral deficits,
including (a) autonomic iwsaturity, as indicated by a greater propensity to
startle and poorer motoric, reflex, and neuromuscular functioning; and (b)
depressed responsiveness, as indicated by a greater number of hypoactive
72
-------
reflexes and more limited lability of state. Analyses based on the Ala et
al. (1979) criteria suggest that these behavioral effects are of sufficient
magnitude to warrant the concern of the experienced clinician. These
findings are consistent with reports of depressed reflexes and sluggishness
in the animal literature (Storm et al., 1981; Tilson et al., 1979) and with
Japanese reports of sluggish, clumsy, and jerky movement, apathy, and
hypotonia in somewhat older acutely exposed human infants (Harada, 1976).
Because the chronic low level exposure in the present population is not
associated with chloracne and other rashes that nark the Yusho syndrome,
these neonatal behavioral deficits are less likely to be noticed, and
behavioral disturbances emerging later in development are less likely to be
traced to intrauterine PCS exposure.
The association of contaminated fish consumption with neonatal behavior-
al deficits is not corroborated by the cord serun measure of PCB exposure.
This discrepancy may be due, in part, to the poor reliability of the cord
serum measure. In addition, cord samples were available for only 48,7% of
the infants with the highest fish consumption exposure levels (> 10 kg/yr.),
compared with 65.3% for the remainder of the sample, so that the range of
exposure levels available for the cord PCB analyses is mere restricted.
Alternatively, it is possible that the behavioral deficits associated with
fish consumption are due to toxins other than PCBs that are present in the
same contaminated fish. Toxaphene, for example, has recently been detected
in fatty Lake Michigan fish in quantities comparable to those of PCBs
(Everett, 1982).
73
-------
According to Spyker's (1975) model, a toxin may be associated with
morphological, maturational, reproductive, or behavioral disorders. The
symptoms exhibited ay a particilar individual depend on such variables as
genetic predisposition, period of development when exposed, and when the
assessment is made. Spyker suggests that at high levels of exposure a toxin
may be sufficiently potent to produce a coherent syndrome of effects, but at
lover levels individuals are likely to exhibit disparate symptoas. The case
of PCBs confirms these predictions. Path analytic models show that the birth
size and behavioral effects are la-gely independent tn the present sample.
Thus, in contrast to high level exposure, which produced the Yusho syndroms
in Japan, chronic low level exposure may be associated with only one or
another of Che neonatal deficits. The total impact of a toxin at these
relatively low levels of exposure is, therefore, best estimated in a multiple
regression analysis which includes a variety of potential adverse outcomes.
In the preoent study, such an analysts indicates that 11.42 of the variance
in contaminated fish consumption is associated with measureable neonatal
deficits. Follow-up assessments of the affected infants are necessary to
establish whether these deficits are transitory or whether they signal the
onset of a toxic process that may impair subsequent cognitive and eootional
development.
74
-------
Footnotes
^School of Public Health, University of Michigan, Ann Arbor,
Michigan 46109.
2
Department of Psychology, Wayne State University, Detroit,
Michigan 48202.
^Since the fish consumption interviews were not conducted by
the same persons who subsequently collected medical and behavioral
data, those engaged in assessment activities were blind with re-
spect to the mothers' pattern of fish consumption and infants'
potential exposure levels.
4
Results of a state-wide survey of breast milk found PCB
levels exceeding 2 mg/kg (fat basis) in 27 percent of the counties
in the western part of tl.e lower peninsula, .in contrast to only
5 percent of the eastern counties (Wickizer, Brilliant, Copeland
6 Tilden, 1981). Grand Rapids is the largest metropolitan area
located near Lake Michigan. The Grand River connects it to the
lake and a fish ladder has made Grand Rapids a popular location
for lake trout and spawning salmon, Muskegon is located directly
on lake Michigan.
5Data from the infant medical records were used to assign
infants at-risk scores based on questions from the Postnatal
Complications Scale {PCS) (Littman £ Panr.elee, 1974b). The PCS
is based on 10 items considered to reflect increased risk of
mortality and morbidity to newborns, such as respiratory
distress, infection, hyperbilirubinemia, and metabolic and
temperature disturbances. The PCS used an optimal scoring
technique, such that high scores indicate few complications.
75
-------
6If the father was a student, his father's education and
occupation were used to estimate the family's socioeconomic
status. In those cases.in which the mother was unmarried,
socioeconomic status was based on the occupation and education of
the individual from whom the family was receiving most of its
financial support.
1
Data from the maternal interview and aedxcal records are
used to assign mothers at-risk scores based on 41 questions in
the Obstetrical Complications Scale (OCS) (Littman & Parmelee,
1974a). The CCS score is derived from data relating to
pregnancy, such as maternal infection, bleeding, parity, age;' the
reproductive history, such as previous premature births or
stillbirths? and the delivery, such as obstetric medication,
duration of labor, type of delivery, and birth weight. Like the
PCS, the OCS is based on optimal scoring with high scores
reflecting few complications.
i
Lake Michigan was the primary source of fish. In addition,
coho and Chinook salmon and lake trout from rivers and lakes
connected to Lake Michigan were included since these fish arp
normal inhabitants from the lake populations.
9 . .
The qualification rate of 4 per cent of the population screened,
coupled with demographic data on the 18 Michigan counties adjacent bo lake
Michigan, suggests that almost 100,000 v* • • m this region might qualify
as moderate fishr-eaters. These figures do not include women in nearby
Wisconsin, or from the Chicago area.
76
-------
fOyjjg sujnmary 0f the laboratory techniques used to analyze
the serum and breast milk samples were written in conjunction
with Dr. Harold A. Pr_ce of the Michigan Department of Public
Health.
stricter criterion was applied to the reflex cluster.
More than four abnormal reflexes were required, rather than three.
77
-------
References
Abe, S., Inoue, Y., & Takamatsu, M. Polychlorinated biphenyl
residues in plasma of Yusho chldren born to mothers who had
consumed oil contaminated by PCB. Acta Medica Fukuoka.
1975, 66, 605-609. ^ ^
Akiyama, K., Fujitani, K., Ohi, G., & Yagyu, H, Polychlorinated
biphenyl residues in maternal and cord blood in Tokyo
aetroppolitan area. Bulletin of Environaental Contamination
and Toxicology, 1975,"Til 588-592.
Aleksandrovicz, N. R, & Aleksandrovicz, D. R. Obstetrical pain-
relieving drugs as predictors of infant behavior
variability. Child Development, 1974, 45, 935-245.
Aleksandrovicz, K. K., i Aleksandrovicz, 0. S. Precursors of ego
in neonates: Factor analysis of Brazelton Scale data.
Journal of the American Academy of Child Psychiatry, 1976,
V5, 257-268.
Allen, J. R., Barsotti, D. A., & Carstens, L, A. Residual
effects of polychlorinated biphenyls on adult nonhuman
primates and their offspring. Journal of Toxicology and
Environmental Health, 1980, 6, 55-66.
Als, K., Tronick, E., Lester, 1. M., & Brazelton, T. B. Specific
neonatal measures: The Brazelton Neonatal Behavioral
Assessment Scale. In J. Osofsky (Ed.), Handbook of Infant
Development. New fork; Wiley, 1979,
Ballard, J. L., Novak, K. K., & Driver, M. A simplified score
for assessment of fetal maturation o£ newlv born infants.
The Journal of Pediatrics, 1979, 95(5), 769-774.
Barr, H. M., Streissguth, A. P., Martin, D. C., 4 Horst, T. E,
Methodological issues in assessment of caffeine intake: A
method for quantifying consumption and a test-retest
reliability study. In L. F. Soyka I S. P. Redmond (Eds.),
Drug metabolism in the immature human. New fork: Raven,
1981.
Biros, F. J., Walker, A. C., L Medbery, A. Polychlorinated
biphenyls in human adipose tissue. Bulletin of
Environmental Contamination and Toxicolcoy, 1970, 5,
317-323.
Bowman, R. E., ft Heironintus, M. P, Hypoactivity in adolescent
monkeys perinatally exposed to PCBs and hyperactive as
juveniles. Neurobehavioral Toxicology and Teratology, 1981,
3, 15-18.
78
-------
Bowman, R. E., Heironimus, M. P., & Allen, j. R. Correlation of
PCB body burden with behavioral toxicology in monkeys.
Pharmacology, Biochemistry, and Behavior, 1978, 9, 49-56.
Brazelton, T. E. Neonatal Behavioral Assessment Scale.
Philadelphia: Lippincott, 1973.
Burse, V. W. , Needham, L. L., Liddle, J. A., Bayse, D. 0,, *
Price, H. A. Interlaboratory comparison for results of
analyses for polybrotninated biphenyls in human serum.
Journal of Analytical Toxicology, 1980, 4, 22-26.
Carey, A. E., I Gowen, J. A. PCB's in agricultural and urban
soil. In: Proceedings of the National Conference on
Pol yc hi or inert ed Biphenyls, Chicago, 1975,BPA-56O/6-75-Q04.
Washington D.C., Environmental Protection Agency, 1976,
tSS-198,
Chia, L., Su, H., Chen, R., Wu, Z., I Chu, F. Neurological
manifestations in polychlorinated biphenyls (PCB) poisoning.
Clinical Medicine (Taipai), 1981, J, 45-61.
Chou, S. M., Miike, T., Payne, W. M., & Davis, G. J.
Neuropathology of "spinning syndrome" induced by prenatal
intoxication with a PCB in mice. Annals of the New York
Academy of Sciences, 1979, 320, 373-395.
Divitto, B., & Goldberg, S. The effects of newborn medical
status on early parent-infant interaction. In T. Field,
A. Sostek, S. Goldberg, & H. Shuman (Ids.), Infants Born at
Risk. Hoiliswooa, Hf: Spectrum, 1979.
Dubowitz, L. M., Dubowitz, V., & Goldberg, C. Clinical
assessment of gestational age in the newborn infant. The
Journal of Pediatrics, 1970, 77(1), 1-10.
Everett, D. IPA to limit toxaphene use. Detroit Free Press,
October 19, 1982, 3; 8. ~ '
Field, T. M., Dabiri, C., Hallock, N., £ Shuman, H. H.
Developmental effects of prolonged pregnancy and the
postmaturity syndrome. Journal of Pediatrics, 1977, 90,
836-839.
Field, T. M., Dempsey, J, R., & Shuman, H. H. Developmental
follow-up of pre- and post- fefrn infants. In S. L. Friedman
& M. Sigman (Eds.), Preterm birth and psychological
development. New fork: Academic, .981.
79
-------
field, T., Hallock, N., Ting, G., Dempsey, 3., Dabiri, C., *
Shuman, H. H. A first-year follow-up of high-risk infants;
formulating a cumulative risk index. Child Development,
1978, 49, 119-131.
Fringe, C. S. , Fendley, T. W., Dunn, R. T., & Queen, C. A.
Improved determination of total serum lipids by the sulfo-
phospho-vanillin reaction. Clinical Chemistry, 1972, 18,
673-674.
Punatsu, I., Yamashita, F., I to, Y., Tsugaina, S., Funatsu, T.,
Yoshikane, T., Hayashi, M., Kato, T., Yakushiji, M.,
Okamoto, G., Yamasaki, $,, Arima, T., Xuno, T., Ide, H., k
Ibe, I, Polychlorbiphenyls (PCB)-induced fetopathy. Kurume
Medical Journal, 1972, IS, 43-51.
Gilbert, R. M., Marshman, J. A., Schwieder, M., & Ber§, R.
Caffeine content of beverages as consumed, Canadian Medical
Association Journal, 1976, , 205-208,
Grant, D., Mes, J., & Frank, R. PCB residues in human adipose
tissue and milk. National Conference on Polychlorinated
Biphenyls, November 19-21, 1975, Chicago, Illinois. NTIS
PB-253-248, 1t<76, 144-146.
Grant, D. L., Villeneuve, D. C., McCully, K. A., & Phillips, W.
Placental transfer cf polychlorinated biphenyls in the
rabbit. Environmental Physiology, 1971, 2# 61-66.
Harada, M. Intrauterine poisoning. Clin.'eal and epidemiological
studies and significance of the problem. Bulletin of the
Institute of Constitutional Medicine. Kumamoto University,
1976, 25(Suppl.). ^ ~ ^
r
Higuchi, K. (Ed). PCB Poisoning and Pollution. New York;
Academic Press, 1576.
Hirayama, C. clinical aspects of PCB poisoning. In R. Hiquchi
(Ed.), PCS Poisoning and Pollution. Hew York; Academic
Press, 1976.
Hollingshead, A. B. Four factor index of social status.
Unpublished manuscript, Yale Universi ty, New Haven, n.d.
Humphrey, H, E. B, Evaluation of changes of the level of
polychlorinated biphenyls (PCB) in human tissue.
Unpublished manuscript, Michigan Department of Public
Health, Lansing, 1976.
Horowitz, F. D., Sullivan, J. W., s Linn, P. Stability and in-
stability in the newborn infant: The search for elusive
threads. In A. J. Sameroff (Ed,), Organization and stability
of newborn behavior: A Commentary on the Brazelton Neonatal
Behavioral Assessment Scale, Monographs of the Society for
Research in Child Development, 1978, 43(5-6, Serial No. 1777.
80
-------
International Agency for Research on Cancer, Polychlorinated
biphenyls and polybrominated biphenyls. IARC Monographs on
the Evaluation of the Carcinogenic Risk of Chemicals to
Humans, 1978, 18.
International Joint Commission, Great Lakes Water Quality Board:
Appendix E. Status report on organic and heavy metal con-
taminants in the Lakes Erie, Michigan, Huron and Superior
basins, 1978.
Jacobson, J. L., Fain, G. G., Jacobson, S. W., & Schwartz, P. M.
Factors and clusters for the Brazelton Scale: An investiga-
tion of the dimensions of neonatal behavior. Developmental
Psychology, in press.
Jacobson, S. W., Jacobson, J. L., Schwartz, P. M., & Fein, G. G.
Intrauterine exposure of human newborns to PCEs: Measures
of exposure. In F. M. D'ltri & M. Kamrin (Eds.), PCBs:
Human and environmental hazards. Ann Arbor: Ann Arbor
Science, in press. ^ ^
Jelinek, C. F,, & Corneliussen, P. E, Levels of PCBs in the US
food supply. In: Proceedings of the National Conference
on Polychlorinated Biphenyls, Chicago, November 19-21, 1975,
EPA-560/6-75 -004. Washington, D.C.: Environmental Protection
Agency, .1976, 147,
Kaye, X. Discriminating among normal infants by multivariate
analysis of Brazelton scores: Lumping and smoothing. In
A. J. Sameroff (Ed,), Organization and stability of newborn
behavior; A commentary on the Brazelton Neonatal Behavioral
Assessment Scale. Monographs of the Society of Research in
Child Development, 1978, 43 (5-fe, Serial No. 177).
Kenny, D. A. Correlation and causality. New York; Wiley, 1975.
Kodama, H., & Ota, H. Studies on the transfer of PCB to infants
from their mothers. Japanese Journal of Hygiene, 1977, 32,
567-573.
Kodama, H., & Ota, H. Transfer of polychlorinated biphenyls to ^
infants from their mothers. Archives of Environmental Health,
1980, 35, 95-100.
Kreiss, K., Zack, M. M. , Kimbrough, R. D., Needham, L. L., Srnrek,
A. L., & Jones, B. T. Association of blood pressure and
polychlorinated biphenyl levels. Journal of the American
Medical Association, 1981, 245, 2505-2509.
Kuwabara, K., Yakushiji, T., Watanabe, I., Yoshida, S., Koyama,
K., & Kunita, N. Levels of polychlorinated biphenyls in
blood of breast-fed children whose mothers are non-
occupational iy exposed to PCBs. Bulletin of Environmental
Contamination and Toxicology, 1979, 21, 458-462.
81
-------
Kuzma, J. W. , & Kissinger, D. G. Patterns of alcohol and cigar-
ette use in pregnancy. Meurobehavioral Toxicology and -
Teratology, 1981, 3, 211-221.
Lester, B. M., Als, H., & Brazelton, T. B. Regional obstetric
anesthesia and newborn behavior: A reanalysis toward
synergistic effects. Child Development, 1982, 53, 687-692.
Liddle, J, A, Interlaboratory comparability of serum PCI
analyses. In F. K. D'ltri i M. Ramrin (Eds.), PCBs: Human
and environmental hazards. Ann Arbors Ann Arbor Science,
in press.
Littman, 1., & Parmelee, A. H. Manual for obstetrical
complications. Unpublished manuscript. University of
California, Los Angeles, 1974a.
Littman, B., & Parmelee, A. H. Manual for postnatal
complications. Unpublished manuscript, University of
California, Los Angeles, 1974b.
Lubchenco, L. O., Hansman, C., Dressier, M., & Boyd, E.
Intrauterine growth as estimated from live-born birthveight
data at 24 to 42 weeks of gestation. Pediatrics, 1963, 32,
793-800.
Masuda, Y., Ragawa, R., Ruroki, H., Kuratsune, M,, foshimura, T.,
Taki , I. , Kusuda , M., Taniashita , F., t Hayashi , M, Transfer
of polychlorinated biphenyls from mothers to foetuses and
infants. Bulletin of Environmental Contamination and
Toxicology. 1978, JI6, 543-546.
Masuda, Y., Kagawa, R., Ruroki, B., Tokudome, S., 6 Kuratsune, M.
Transfer of various, polychlorinated biphenyls to the
foetuses and offspring of mice. Food and Cosmetic
Toxicology, 1*79, J_7, 623-62"'.
Matthews, H. B. The metabolism of PCBs in mammals. In
F. M. D'ltri & M. Ramrin (Eds.), PCBs; Human and
environmental hazards. Ann Arbor; Ann Arbor Science, in
press.
Miller, H, C. , & Hassar.ein, K. Diagnosis of impaired fetal
~ growth in newborn infants. Pediatrics, 1971, 48(4),
511-522.
Mull in, M., Safe, S., Safe, L., Pochini , C., McCrindie, S., 6
Romkes, M. Analysis of PC3s using specific isomer high
resolution capillary gas c hepatography. In F. M. D'ltri &
M. Ramrin (Eds.) , PCBs: Human and environmental hazards.
Ann Arbor; Ann Arbor Science, in press.
32
-------
Musial, C, J., Hutzinger, 0., Zitko, v., & Crocker, J. Presence
of FCB, DDE and DDT in human milk in tee provinces of'New
Brunswick and Nova Scoti3, Canada, Bulletin of
Environmental Contamination and Toxicology, 1974, JJ2, 258.
Needham, L. L., Burse, V. w., 6 Price, H. A. Temperature-
programmed gas chromatographic determinations of
polychlorinated and polybrominated biphenyls in serum.
journal of Association of Official Analytical Chemists. •
1981 , 64, 1131-1137.
Osofsky, J. D., & O'Connellr E. J. Patterning of newborn
behavior in an urban population. Child Develoment, 197?,
' 48, 532-536.
Picer, M., i Ahel, ft. Separation of polychlor inated biphenyls
from DDT and its analogues on a miniature silica gel column.
Journal of Chromatography, 1978, ISC, 119-127.
Polishuk, 2, W., Kasserman, D., Wasserman, K.Cucos , S., 6 Ion,
M. Organochlorine compounds in mother and fetus during
labor. Environmental 'Research, 1977, jy3# 278-204.
Powell, L. F, The effect of extra stimulation and maternal
involvement on the development of low-birth-weight infants
and on maternal behaviors. Child Development, 1974, 45,
106-113.
Sameroff, A. J., Krafchuk, E. E., & Bakov, H. A, Issues in
grouping items from the Neonatal Behavioral Assessment
Scale. In A. J, Sameroff (Ed.), Organization and stability
of newborn behavior: a commentary on the Brazelton Neonatal
Behavioral Assessment Scale. Monographs of the Society for
Research in Child Development, 1978, 43(5-6, Serial "
No. 177).
Savage, E. P., Tessari, J. D., Kalberg, J. W., Wheeler, H. W., &
Bagley, J. R. A search for polychlorinated biphenyls in
human milk in rural Colorado. Bulletin of Environmental
Contamination and Toxicology, 1973, 12, 250.
Sawyer, L. D. Quantitation of polychlorinated biphenyl residues
by electron capture gas-liquid chromatography. Journal of
Association of Official Analytical Cheaists, 1978, 61,
272-291.
Spyker, J. M. Assessing the impact of low-level chemicals on
development: Behavioral and latent effects. Federation
'Proceedings, 197a, 34(9), 1835-1044.
Storm, J. E., Hart, J. L., & Smith, R» F. Behavior of mice after
pre- and postnatal exposure to Arochlor 1254.
Neurobehavioral Toxicology and Teratology. 1981, 3, 5-9.
83
-------
fobs, A. R. Levels of polychlorinated biphenyls in adipose
tissue of the general population of the nation.
• Environmental Health Perspectives.' 1972, 79-81.
Yoshimura, H., & Yoshimura , S. Toxicological aspects II: the
metabolic fate of PCS' s and their toxicological evaluation.
• In K. Higuchi (Ed»), PCB Poisoning and Pollution. New York: '
Academic Press, 1976.
Young, D. 1,, McDermott, D. J. , & Heesen, 7. C. Marine inputs of
polychlorinated biphenyls off Southzrn California." In:
Proceedings of the National Conference on Polvchlorinatc*
Biphenyls, Chicago, EPA-560/6-7 5-004. Washing tea
D.C., Environmental Protection Agency, 1976, 199-208-
85
------- |