Prenatal exposure to polychlorinated biphenyls: Effects on birth size and gestational age

Prenatal exposure to polychlorinated biphenyls: Effects on birth size and gestational age Polychlorinated biphenyls (PCBs), environmental pollutants that concentrate in the fatty tissue o f predatory sports fish, cross the placenta in women exposed to ordinary dietary levels. Two hundred forty-two newborn infants whose mothers consumed moderate quantities o f contaminated lake fish and 71 infants whose mothers did not eat such fish were examined during the immediate postpartum period. PCB exposure, measured by both contaminated fish consumption and cord serum PCB levels, predicted lower birth weight and smaller head circumference. Exposed infants were 160 to 190 gm lighter than controls, and their heads were 0.6 to 0.7 cm smaller. Head circumference was disportionately small in relation to both birth weight and gestational age. Control analyses showed that none o f these effects was attributable to any o f 37 potential confounding variables, including socioeconomic status, maternal age, smoking during pregnancy, and exposure to polybrominated biphenyls. These findings are consistent with earlier reports from studies in laboratory animals. (J PEDtATR 105:315, 1984)

Greta G. Fein, Ph.D., Joseph L. Jacobson, Ph.D., Sandra W. Jacobson, Ph.D., Pamela M. Schwartz, Ph.D., and Jeffrey K. Dowler, M.A. Ann Arbor and Detroit, Mich.

POLYCHLORINATED BIPHENYLS are a group of polycyclic chlorinated hydrocarbons, which achieved widespread industrial use because of their thermal stability, adhesiveness, and resistance to biodegradation. A world-wide pollutant, PCBs are lipophilic and poorly metabolized compounds that tend to concentrate in the fatty tissue of aquatic organisms feeding in contaminated waters. Consumption of PCB-contaminated fish from Lake Michigan has been found to predict PCB concentrations in maternal serum and milk?' 2 Maternal serum values, in turn, predict cord serum PCB levels)-5 Lake Michigan fish have been implicated as a dietary source of PCB exposure in women of childbearing age and a source of transplacental exposure for their infants.

From the School o f Public Health, University o f Michigan, and Wayne State University. Supported by Grant CR808520010 (to G.G.F.) from the Environmental Protection Agency. Submitted for publication June 22, 1983; accepted Feb. 17, 1984. Reprint requests: Greta G. Fein, Ph.D., College o f Education, University o f Maryland, College Park, MD 20742.

Evidence of fetotoxicity comes primarily from laboratory animal studies.6'7 Ingestion of PCBs by rhesus mothers is associated with a greater incidence of abortions and resorptions and lower birth weights among liveborn offspring, even those conceived more than 12 months after maternal ingestion.8'9 Ingestion by mouse dams is associated with smaller litter size and a higher rate of neonatal

PBB PCB

Polybrominated biphenyl Polychlorinated biphenyl

mortality2~ The only evidence of human effects comes from studies of pregnant women in Japan and Taiwan who consumed rice oil accidentally contaminated with high concentrations of PCBs along with other compounds (e.g., polyehlorinated dibenzofurans).~.~2 Infants exposed in utero tended to be small and to be born prematurely. 13"14 Follow-up studies reported growth impairment, an average IQ of 70, sluggishness, clumsy and jerky movement, apathy, and hypotonia. ~ The consequences of intrauterine exposure from ordinary dietary sources have not previously been investigated in humans.

The Journal o f P E D I A T R I C S

315

3 16

Fein et al.

The Journal of Pediatrics August 1984

Table I. Control variables yielding differences for exposed versus nonexposed infants Exposed Overall contaminated fish consumption* Maternal prepregnaney weight (kg) (X +_ SD) Type of delivery (% spontaneous) Alcohol prior to pregnancy (%) Alcohol during pregnancy (%) Caffeinet prior to pregnancy (%) Caffeinet during pregnancy (%) Cold medications during pregnancy (%) Cord serum PCB level:~ Maternal age (yr) (X __+SD) Weight gain during pregnancy (kg) (X +_ SD) Type of delivery (% spontaneous)

Nonexposed

62.0 _+ 11.6 67.4 54.2 22.7 40.1 22.7 28.6

<0.10 <0.05 <0.001 <0.01 <0.10 <0.10 <0.01

66.1 _+ 14.7 77.5 28.2 7.0 28.2 12.7 13.2

27.1 _+ 5.3 12.7 _+ 4.6 63.2

<0.10 <0.10 <0.10

26.0 _+ 4.2 13.8 _+ 4.6 73.5

*Exposed definedas >11.8 kg over 6 years, n = 242; nonexposed n = 71. tEquivalent of >2 cups of coffeeper day. ~Exposed definedas >3.0 ng/mL, n = 75; nonexposed n = 166.

METHODS Study participants were 242 infants born between July 1980 and December 1981 to women who reported moderate consumption of Lake Michigan fish, and 71 infants 'born during this period whose mothers ate no Lake Michigan fish. All but two participants were white, 52% were male, 43% were firstborn, and six were the larger members of twin pairs, none of whom died during infancy. Dietary exposure. All newly delivered mothers in four hospitals located near Lake Michigan were asked to participate in a screening interview on the day after delivery. Women reported the amount of each species of Lake Michigan fish they consumed during the prior year and, if they ingested greater amounts at some time in the past, consumption rates for that period as well. Past consumption was considered significant because PCBs once ingested are difficult to eliminate. A weighted value was assigned to each fish species consumed, based on average contaminant levels reported for that species in Lake Michigan? Overall contaminated fish consumption was defined as the weighted sum of annual Lake Michigan fish consumption in the present or past, whichever was greater. The amount of contaminated fish consumed during pregnancy was also recorded. To ensure the participation of infants likely to have been exposed in utero to a broad range of PCB levels, the exposed group composed about three quarters of the sample and was limited to women who had consumed at least 11.8 kg contaminated fish over 6 years. The controls were selected at random from among the women who did not eat fish, and were distributed among the four participating hospitals in proportion to the number of exposed participants from each hospital. Although the mothers

were not told that PCB exposure was the focus of the study, all assessment procedures were explained in detail, and women agreeing to participate gave comprehensive informed consent. Laboratory analysis. Cord and maternal serum samples were analyzed at the Michigan Department of Public Health by means of packed-column gas chromatography. 1,~6 Quantification was provided by adapting the Webb-MeCall method to a computer data system with Aroclors 1016 and 1260 used as calibration standards? 7. ~s Quality control over time was ensured by addition of a spiked sample of bovine serum with every batch of 10 samples, as well as the use of a glassware blank and a reagent blank. Virtually all of the samples matched to the Aroclor 1016 standard were below the laboratory's detection limit, presumably because the lower chlorinated isomers are more susceptible to degradation in the environment and to more rapid metabolism in the organism? 9 Values matched to the Aroclor 1260 standard were, therefore, used in the analyses reported. Serum PCB values were available for 196 mothers. Cord serum PCB and polybrominated biphenyl values were available for 241 infants. Outcome variables. Birth weight was obtained from the hospital delivery record. Crown-heel length and head circumference were measured by project staff on the second day after delivery. Gestational age was based on the mother's report of her last menstrual period and on the Ballard Examination for Fetal Maturity, which estimates gestational age by assessing physical and neuromuscular maturity? ~ The Ballard examination was administered at 30 to 42 hours after birth to the 209 infants (163 fish-eating mothers, 46 controls) for whom it was possible to obtain maternal consent within this limited time. The

Volume 105 Number 2

P r e n a t a l e x p o s u r e to P C B s

3 17

Table II. Adjusted birth size and gestational age measures by overall contaminated fish consumption and cord serum PCB level Overall contaminated fish consumption*

Birth weight (kg) Head circumference (cm) Gestational age based on last menstrual period (wk) Gestational age (Ballard examination) (wk) Neuromuscular maturity Physical maturity

Cord serum PCB levelf

N o n f i s h eaters (n = 71)

Fish eaters (n= 242)

<3 ng/mL (n = 166)

>--3 ng/mL (n = 75)

3.66 • 0.54 35.48 • 1.36 40.82 _+ 3.07

3.47 + 0.53 34.92 _+ 1.31 40.31 • 2.97

P

<0.05 <0.01

3.57 • 0.54 35.28 + 1.18 41.03 -+ 3.01

3.41 • 0.54 34.63 • 1 . 1 9 39.77 ___3.06

<0.05 <0.001 <0,05

39.85 • 1.42

39.15 ___ 1.40

<0.01

39.41 • 1.40

39.47 • 1.4l

19.96 • 2.48 17.13 • 2.25

18.52 • 2.44 16.67 • 2.19

<0.001

19.95 -+ 2.42 16.96 -+ 2.14

19.00 +-- 2.40 16.94 + 2.15

Values represent mean • SD. *Adjusted for effectsof maternal prepregnancyweight,type of delivery,and consumptionof alcoholand caffeineprior to and during pregnancy and cold remedies during pregnancy. 1"Adjusted for sex of infant, type of delivery, maternal weight gain during pregnancy, and maternal age. staff members who examined the infant and interviewed the mother were unaware of maternal fish consumption and maternal and cord serum PCB levels. Control variables. A comprehensive list of 73 potential confounding variables included data pertaining to demographic background, health history, pregnancy and delivery, obstetrical medication, and fetal exposure to other drugs, such as alcohol, caffeine, and nicotine. Because many of these mothers had been exposed to PBBs, a related environmental contaminant, 2~ cord serum PBB level was also used as a control variable. To ensure sufficient variance in the case of dichotomous variables, such as marital status (married/unmarried) or parity (firstborn/later born), only those variables were included for which incidence in each category exceeded 15%. Given this criterion, it was possible to assess the effects of 37 potential confounding variables. Statistical analyses. Analyses based on overall contaminated fish consumption compared two groups of infants: those whose mothers had consumed more than 11.8 kg over 6 years versus those whose mothers had not consumed contaminated fish. These two groups were first compared using t tests or chi-square tests, as appropriate, to determine whether they differed (at P < 0.10) on any of the 37 potential confounding variables. The two groups were then compared on each of the outcome measures (e.g., birth size, gestational age) using analysis of covariance, 22 controlling simultaneously for those potential confounding variables that had yielded between-group differences. A toxic effect was inferred if the group difference remained significant after adjustment for the effects of the control variables. The same procedure was used to compare two groups of infants determined by each of the other exposure

measures. Infants whose mothers ate contaminated fish during pregnancy were compared with those whose mothers did not, and infants with cord serum PCB levels at or above the laboratory's detection limit (3.0 n g / m L ) were compared with those whose levels were below this limit. RESULTS Degree of exposure. The 242 women who reported eating Lake Michigan fish consumed 6.7 _+ 5.8 k g / y r (mean _+ SD). Median consumption was 5.0 kg/yr, equivalent to about two salmon or lake trout meals per month (range 0 to 4). These women reported having eaten Lake Michigan fish for an average of 15.9 + 9.1 years. During pregnancy their consumption rates declined to 4.1 _+ 4.4 kg/yr. Maternal serum PCB concentrations for the sample as a whole averaged 5.5 _+ 3.7 n g / m L , which is comparable to those for other midwestern American samples. 23 Cord serum PCB levels averaged 2.5 _+ 1.9 n g / m L . Although maternal contaminated fish consumption did not predict cord serum PCB level directly, contaminated fish consumption predicted maternal serum PCB level (r = 0.37, P < 0.001), which, in turn, predicted cord PCB level (r -- 0.41, P < 0.001). Evaluation of sampling bias. About two thirds of the mothers invited to participate in the study agreed to do so. Participants (P) and refusers (R) did not differ significantly in terms of either of the contaminated fish consumption measures, maternal age, marital status, birth class attendance, or sex or parity of infant. The participants were somewhat better educated than the refusers, 13.3 years of school (P) versus 12.7 (R), and somewhat higher in socioeconomic status, 24 39.3 (P) versus 35.7 (R). Because

3 18

Fein et al.

The Journal of Pediatrics August 1984

kg

t.u

cm

,.z, 3.6-

p-

ta_

35.5

3.5~-- 35.0-

-t-

3.4-

9 34.5.

3.3-

L

weeks 40 i.=J

.._1

~

39

(.9

38"3"

I

None

~

Low Medium High EXPOSURE LEVEL

18--

None

Low Medium High EXPOSURE LEVEL

Figure. Dose-response relationships for birth weight, gestational age (Ballard examination), head circumference, and neuromuscular maturity by overall contaminated fish consumption. Group means were derived from analyses of covariance, in which they were adjusted for the effects of maternal prepregnancy weight, type of delivery, and consumption of alcohol and caffeine prior to and during pregnancy and cold remedies during pregnancy. Exposed infants were divided into three approximately equal groups: low exposure 2.0 to 3.4 kg/yr, medium 3.5 to 6.5 kg/yr, high 6.6 to 41.7 kg/yr.

the offspring of better educated and higher status women should be less prone to adverse neonatal outcomes, however, their proportionately greater representation in the exposed group can be assumed not to have biased the data in favor of finding toxic effects. Ninety percent of the infants designated at medical risk by hospital staff were among the offspring of fish-eating women, and all but one of their mothers agreed to participate. There is no evidence, however, that infants at risk were systematically underrepresented among the nonexposed mothers who agreed to participate. Control variables. Table I shows group means or percentages for control variables found to be differentially distributed (P < 0.10) in exposed and nonexposed infants based on overall contaminated fish consumption and cord serum PCB level. The magnitudes of these between-group differencs were consistently small. When calculated in terms of point-biserial or phi coefficients, most of the correlations ranged from 0.09 to 0.15, and none exceeded 0.22. Neonatal outcomes. Overall contaminated fish consumption predicted smaller birth size on two measures, birth weight and head circumference, and a shorter gestational period as estimated by the Ballard examination (Table II). Infants born to women who had consumed contaminated

fish averaged 190 gm less in birth weight, 0.6 cm less in head circumference, and 4.9 days less in gestational age. The lower Ballard scores were related primarily to greater neuromuscular immaturity among the exposed infants. All four outcomes were related to overall contaminated fish consumption in a dose-dependent fashion (Figure). The head circumference of the exposed infants was significantly smaller (P < 0.05), even after birth weight and gestational age were statistically controlled. This finding suggests that the head size of these infants was, on average, disproportionately small in relation to both weight and gestation. Although fish consumption during pregnancy was highly correlated with overall fish consumption (r = 0.79, P < 0.001), consumption during pregnancy did not predict either birth size or gestational age and was related only to neuromuscular maturity (18.6 _+ 2.4 exposed during pregnancy versus 19.3 _+ 2.5, P < 0.05). Even with consumption during pregnancy statistically controlled, overall fish consumption continued to predict lower birth weight (P < 0.05), smaller head circumference (P < 0.005), earlier gestational age on the Ballard test (P < 0.005), and poorer neuromuscular maturity (P < 0.05). Because contaminants other than PCBs could be responsible for the observed relationships between fish consump-

Volume 105 Number 2

tion and birth size or gestational age, cord serum PCB values were used to corroborate the findings based on fish consumption. Sex of infant was found to function as a suppressor variable in the relation of cord serum PCB level to birth weight, because of a tendency for males to be overrepresented among the more highly exposed infants in this sample (57.9% highly exposed versus 50.0%). Because infant boys tend to be heavier than girls (P = 0.020), the sex difference in PCB level masked the effect of cord PCBs on birth weight unless sex was included as a control variable in the analysis. With sex of infant and three potential confounding variables (Table I) statistically controlled, cord serum PCB level predicted lower birth weight, smaller head circumference, and shorter gestation based on reported last menstrual period (Table II). Infants with cord serum concentrations a t or above the laboratory's detection limit averaged 160 gm less in birth weight, 0.7 cm less in head circumference, and 8.8 days less in gestational age. Head circumference was significantly smaller (P < 0.01) among newborn infants with detectable cord PCB levels, even when birth weight and gestational age were statistically controlled. DISCUSSION Despite the widespread distribution of PCBs in the aquatic environment and evidence of placental transfer to the fetus, our study is the first to assess the effects of PCB exposure at ordinary dietary levels on the human infant, in our sample of mothers who consumed contaminated Lake Michigan fish, intrauterine exposure was associated with reduced birth size on both the contaminated fish consumption and cord serum PCB measures of exposure. Exposed infants were 160 to 190 gm lighter than controls, a magnitude comparable to that reported by the Surgeon General for smoking during pregnancy.2s In contrast to smoking, however, which primarily affects birth weight, head circumference was most directly affected by PCB exposure. Evidence of reduced birth size and shortened gestation in this chronically exposed sample is consistent with experimental studies in rhesus monkeys8,9 and with research on acutely exposed humans in Japan? 4 Smaller birth size was predicted from a composite of present and past contaminated fish consumption but not from consumption during pregnancy alone. This finding is consistent with the assumption that the more highly chlorinated PCB isomers (i.e., those matched to the Aroclor 1260 standard tested) are stored in human tissue over relatively long periods. The data suggest that, at low levels of exposure, the threat to the fetus is based more on gradual long-term bioaccumulation than exposure during pregnancy per se. Because ethical considerations do not permit the random

Prenatal exposure to PCBs

3 19

assignment of humans to experimental conditions, our study used a correlational design. Any of 37 potential confounding variables that were differentially distributed (P < 0.10) in exposed and nonexposed individuals were included as covariates in the statistical analyses. Because other control variables, such as socioeconomic status, maternal smoking, PBB exposure, and parity, were unrelated to exposure, it is very unlikely that they account for the adverse outcomes associated with contaminated fish consumption and cord serum PCB level. Acutely exposed Japanese infants with similar adverse neonatal outcomes exhibited growth impairment and motoric and cognitive deficits in later infancy,a5 Follow-up assessments are needed to determine whether the effects reported also have implications for subsequent physical and intellectual development. We thank Mason Barr, Jr., for critical evaluation of early drafts of this report; Joel Ager, Jr., for statistical consultation; Wayland Swain, Nelson Thomas, and Harold Humphrey for assistance in developing the project; Gail Brumitt, Patrieia Yaros, Carol Alexander, and Gwen Hendriekse, who collected the data in the field; Gregory Proulx for assistance in the statistical analyses; and the physicians, hospital personnel, and parents who cooperated in providing the data.

REFERENCES 1. Jacobson SW, Jaeobson JL, Schwartz PM, Fein GG: Intrauterine exposure of human newborns to PCBs: Measures of exposure. In D'Itri FM, Kamrin M, editors: PCBs: Human and environmental hazards. Boston, 1983, Butterworth, pp 311-343. 2. Schwartz PM, Jacobson SW, Fein GG, Jacobson JL, Price HA: Lake Michigan fish consumption as a source of polychlorinated biPhenylsin human cord serum, maternal serum, and milk. Am J Public Health 73:293, 1983. 3. Jacobson JL, Fein GG, Jaeobson SW, Schwartz PM, Dowler JK: The transfer of polychlorinated biphenyls and polybrominated biphenyls across the human placenta and into maternal milk. Am J Public Health 74:378-379, 1984. 4. Kodama H, Ota H: Studies on the transfer of PCB to infants from their mothers. Jpn J Hyg32:567, 1977. 5. Masuda Y, Kagawa R, Kuroki H, et al: Transfer of polychlorinated biphenyls from mothers to foetuses and infants. Bull Environ Contam Toxieol 16:543, 1978. 6. Merson MH, Kirkpatriek RL: Reproductive performance of captive white-footed mice fed a PCB. Bull Environ Contam Toxicol 16:392, 1976. 7. Villeneuve DC, Grant DL, Phillips WEJ, Clark ML, Clegg DJ: Effects of PCB administration on mierosomal enzyme activity in pregnant rabbits. Bull Environ Contain Toxicol 6:120, 1971. 8. Barsotti DA, Marler RJ, Allen JR: Reproductive dysfunction in rhesus monkeys exposed to low levels of polychlorinated biphenyls (Aroclor 1248). Food Cosmet Toxicol 14:99, 1976. 9. Allen JR, Barsotti DA, Carstens LA: Residual effects of

320

10.

11. 12. 13.

14. 15.

16.

17. --

Fein et al.

polychlorinated biphenyls on adult nonhuman primates and their offspring. J Toxicol Environ Health 6:55, 1980. Chou SM, Miike T, Payne WM, Davis G J: Neur0Pathy Of "spinning syndrome" induced by prenatal intoxication with a PCB in mice. Ann NY Acad Sci 320:373, 1979. Higuchi K, editor: PCB poisoning and pollution. New York, 1976, Academic Press. Wong KC, Hwang MY: Children born to PCB poisoned mothers. Clin Med (Taipei) 7:83, 1981. Monographs on the evaluation of the carcenogenic risk of chemicals to humans: Polychlorinated biphenyls and polybrominated biphenyls. Lyon, France, 1978, International Agency for Research on Cancer. Funatsu I, Yamashita F, Ito Y, et al: Polychlorbiphenyls (PCB)-induced fetopathy. Kurame Med J 19:43, 1972. Harada M: Intrauterine poisoning: Clinical and epidemiological studies and significance of the problem. Bull lnst Constit Med, Kumamoto University 25(Suppl), 1976. Needham LL, Burse VW, Price HA: Temperature-programmed gas chromatographic determinants of polychlorinated and polybrominated biphenyls in serum. J Assoc Off Anal Chem 64:1131, 1981. Webb RG, McCall AC: Quantitative PCB standards for electron capture gas chromatography. J Chromatogr Sci 11:366, 1973.

The Journal o f Pediatrics August 1984

18. Sawyer LD: Quantitation of polychlorinated biphenyl residues by electron capture gas-liquid chromatography. Reference material characterization and preliminary study. J Assoc Off Anal Chem 61:272, 1978. 19. Kimbrough RD: The toxicity of polychlorinated polycyclic compounds and related chemicals. CRC Crit Rev Toxicol 2:445, 1974. 20. Ballard JL, Novak KK, Driver M: A simplified score for assessment of fetal maturation of newly born infants. J PEDIATR 95:769, 1979. 21. Well WB, Spencer M, Benjamin D, Seagull E: The effect of polybrominated biphenyl on infants and young children. J PEDIATR 98:47, 1981. 22. Winer BJ: Statistical principles in experimental design. New York, 1971, McGraw-Hill. 23. Humphrey HEB: Population studies of PCBs in Michigan residents. In D'Itri FM, Kamrin M, editors: PCBs: Human and environmental hazards. Boston, 1983, Butterworth, pp 299-310. 24. Hollingshead AB, Redlich FCi Social class and mental illness. New York, 1958, Wiley. 25. Smoking and health: Report of the Surgeon General. US Department of Health, Education and Welfare, Publ. No. (PHS) 79-50066. Washington, DC, 1979, US Government Printing Office.