Subjects

In the summer of 1985, we conducted the first field survey of all living children who had been born to mothers poisoned by PCBs and their derivatives in the 1978-1979 incident. A total of 118 children, born between June 1978 and March 1985, entered this study after their parents had given permission. One hundred and eighteen unexposed children were selected as a paired control group, individually matched by gender, age (age difference less than 15 days for those under 1 year of age and less than 1 month for those over 1 year). The difference in the mothers' ages in each pair was less than 3 years. We also matched for socio-economic status and residential location of the parents. The socio-economic status ratings for the children were based on the educational levels of both parents and the occupational status of the parent with the higher occupational level. The five-point system used to score educational level was: 1 point for 6 years or less of education; 2 points for 7-9 years; 3 points for 10-12 years; 4 points for college level (13-16 years); and 5 points for postgraduate level (more than 16 years). The five-point system for occupational status was: unskilled, 1 point; semi-skilled worker, 2 points; skilled worker, 3 points; semi-professional, 4 points; and professional, 5 points. The total scores of these three variables can range from 3 to 15. As expected, there was no difference in age and socio-economic status scores between Yucheng children and the matched control children. One field study has been conducted each year since 1985.

Measurements

Four cognitive measurement tools, according to the different ages of the subjects, were used in this study. These were the Bayley Scales for Infant Development (BSID) for those aged from 6 months to 30 months, the Chinese version of the Stanford-Binet IQ Test for those aged 30 months to 6 years, Raven's Coloured Progressive Matrices (CPM) for those aged 5-9 years, and Raven's Standard Progressive Matrices (SPM) for those aged above 9 years. These tests were administered annually to all the children at their homes. Because these cognitive measurement tools have their own age-range limitations, each subject's score was individualised according to the subject's actual age on the date of testing. The Yucheng child and the control child were tested on the same day by the same tester, and random cross-validations were carried out by a senior clinical psychologist (N.W.G.). The examiners were unaware of the children's history of exposure. All examiners held bachelor's degrees in either special education or psychology and were trained and supervised by our senior clinical psychologist.

Analysis

Student's t-test was used to compare the difference in scores at each age level between each of the 118 Yucheng children and their matched controls.

Effects of PCBs on children

The PCBs had been partially heat-degraded, and were further contaminated with high levels of PCDFs and polychlorinated quarterphenyls (Reference Masuda, Kuroki and HaraguchiMasuda et al, 1986). In 1979, over 2000 persons in Taiwan were intoxicated by heat-degraded PCBs (Reference Hsu, Ma and HsuHsu et al, 1985). Children exposed prenatally to PCBs/PCDFs (Yucheng children) had higher prenatal mortality and lower birth-weight; dysmorphic features at birth such as hyperpigmentation, neonatal teeth, and deformed nails; and delays of developmental milestones (Reference MillerMiller, 1985; Reference Rogan, Gladen and HungRogan et al, 1988). These symptoms were caused not only by PCBs but by their heat-degraded products, PCDFs (Reference Kashimoto, Miyata and FukushimaKashimoto et al, 1985). In humans PCBs, PCDFs and PCDDs have long half-lives, and these compounds have been found to have significant developmental toxicity in animals (Reference Tilson, Jacobson and RoganTilson et al, 1990). They are known to cross the placenta and affect the foetus (Reference Kodama and OtaKodama & Ota, 1980).

Cognitive development in children prenatally exposed to PCBs and their contaminants has been carefully studied in four cohorts of humans.

1. (a) Japanese investigators followed 11 Japanese Yusho (the same two Chinese characters as Yucheng, i.e. oil disease, as pronounced in Japanese) children living on an isolated small island and found that 6 years later they appeared to be apathetic and inactive, with IQs averaging 70 (Reference HaradaHarada, 1976). Harada did not specify the measurement tool used and no control children were studied. Rogan et al (Reference Rogan, Gladen and Hung1988), who cross-sectionally examined 117 children born between 1978 and 1985 to Taiwanese Yucheng women, reported that congenitally exposed children showed delayed developmental milestones and deficits on formal developmental testing.

2. (b) A Michigan cohort included 313 newborns, 242 of whose mothers had consumed Lake Michigan fish presumed to be contaminated by PCBs. These exposed infants showed impaired visual recognition memory at age 7 months, which persisted up to age 4 years (Reference Jacobson, Fein and JacobsonJacobson et al, 1985; Reference Jacobson, Jacobson and HumphreyJacobson et al, 1990).

3. (c) In North Carolina, children in an exposed cohort were found to have neurological anomalies at birth (Reference Rogan, Gladen and McKinneyRogan et al, 1986) and developmental delays in gross motor function during infancy (Reference Gladen, Rogan and HardyGladen et al, 1988).

4. (d) In Taiwan, previous long-term follow-up studies had noted a consistent tendency which indicated that Yucheng children scored lower on each kind of measurement tool at each age level (Reference Chen, Guo and HsuChen et al, 1992; Reference Lai, Guo and YuLai et al, 1994).

Findings in auditory event-related potentials

Chen & Hsu (Reference Chen and Hsu1994) compared the auditory event-related potentials (P300) in 27 pairs of Yucheng and control children in 1991. This method offers considerable promise as a means of testing children with disorders of cognitive function. Prolonged latencies and reduced amplitude were shown with P300 in Yucheng children, indicating slow cognitive processing and attention deficits. The P300 latencies were also found to be inversely correlated with Wechsler Intelligence Scale for Children-Revised (WISCR) full-scale IQs. These findings suggest that the reduced developmental quotient and IQ scores in the Yucheng children may have a neurophysiological basis.

Other factors that could affect the results

Yucheng children and control children were similar with respect to socio-economic status, educational level and parental ages. None of the exposed families had any psychiatric history, none of the exposed mothers had ever smoked, and only two exposed mothers had drunk alcohol regularly before — though not after — marriage.

So far as the possibility of learning effects caused by repeated measures are concerned, it appears that the learning increments by children of the two groups were similar.

Serum levels of PCDFs were not previously measured for the mothers of our subjects. Kashimoto et al (Reference Kashimoto, Miyata and Fukushima1985) reported mean blood levels of PCBs and PCDFs in a group of 113 Yusho patients as 39 parts per billion (p.p.b.) and 0.076 p.p.b., respectively. In 1991, analysis of serum levels in a randomly selected subgroup of Yucheng children in this study (Reference Guo, Lin and YaoGuo et al, 1994; Reference Ryan, Hsu and BoyleRyan et al, 1994) showed median levels of 106 ng/kg lipid, 160 ng/kg lipid and 1.3 ng/kg whole weight for 2,3,4,7,8-PnCDF, 1,2,3,4,7,8-HxCDF and total PCBs, respectively: levels much higher than those of a pooled serum sample of the matched control children (19 ng/kg, 23 ng/kg and 0.17 ng/kg, respectively).

Raven's Progressive Matrices are considered to be relatively culture-fair and have been used to evaluate the general intelligence factor, i.e. Spearman's g factor (Reference Raven, Irvine and BerryRaven, 1983; Reference Raven, Court and JavenRaven et al, 1985). These matrices assess a person's ability to form comparisons and to reason by analogy as well as to show to what extent the subject is capable of organising spatial perceptions into systematically related wholes. Our findings of reduced CPM and SPM scores were consistent with results of the previous assessments in the Yucheng children. A tendency towards a ceiling effect in the SPM scores has been noted since the 1998 field survey, and we are now trying to use another measurement tool, Advanced Progressive Matrices (APM), to minimise this effect.

There may be several reasons why the cognitive functions of Yucheng children were more severely affected than those of the Michigan and North Carolina cohorts. The main reasons may be that Yucheng exposure involved PCBs with PCDFs, terphenyls and quarterphenyls, and that the total quantity of exposure to PCBs and related chemicals was much higher than in the US cohorts.

Although many studies have been done on the toxicity and mechanisms of PCBs and related compounds, the two Asian out-breaks represent the only known cases of clinically obvious PCB/PCDF poisoning in large groups of adults and children. This particular study is the only prospective longitudinal follow-up study on more than 100 prenatally exposed children. Findings from studying the vicissitudes and possible correlates of the adverse effects on Yucheng children as they grow older will provide important information about the toxicity, health effects and mechanisms of prenatal exposure to PCBs, CDFs and PCDDs.