After controlling for socioeconomic differences, we failed to detect any deficit in problem-solving abilities, motor development, or behavior in tests of 7-month-old infants who were prenatally exposed to a wide range of arsenic concentrations. This was a large population-based study, involving 1,799 infants, and we had individual measures of maternal urinary arsenic concentrations at two periods in pregnancy. In addition, we measured a large number of variables that could affect child development and took them into account in the analyses. The developmental tests had good test–retest and interobserver reliabilities and appeared to be sensitive. In the multiple regressions, household assets, fathers’ education, maternal nutritional status, and child’s birth size, gestational age, current nutritional status, sex, and age were significantly independently related to one or more of the child development outcomes. It is therefore reasonable to conclude that the lack of any marked effect of arsenic exposure in pregnancy on infant motor development and problem solving is probably a valid finding.
The lack of detectable effect is encouraging considering the high levels of arsenic exposure. However, we need to be cautious when interpreting the findings because other unmeasured cognitive functions might have been affected. Furthermore, we examined these children at only one age, and it remains possible that effects could appear at a later age when new functions develop. We were unable to find another prospective study that examined the effect of intrauterine exposure of arsenic on infant development.
A recent study took a retrospective history of prenatal exposure to arsenic-contaminated drinking water and found no effect on school-age children’s cognition (von Ehrenstein et al. 2007
). However, cognitive deficits in schoolchildren currently exposed to arsenic-contaminated drinking water have been found in different countries (Rosado et al. 2007
; Wang et al. 2007
; Wasserman et al. 2004
), although the specific cognitive functions affected have varied and the effect size was generally small. It may be that duration of exposure is important. Previous studies in Bangladesh found a smaller effect of arsenic exposure in 6-year-olds (Wasserman et al. 2007
) than in 10-year-olds (Wasserman et al. 2004
). In the present study, arsenic concentration in urine was related to a wide range of socioeconomic variables. The association between arsenic exposure and circumstances of poverty is a common finding (Rahman et al. 2006
), and the effect of poverty on child development is well documented and generally increases with age (Grantham-McGregor et al. 2007
). It is difficult to measure all factors associated with poverty that affect child development. For example, few studies of arsenic exposure in school-age children have measured stimulation in the home, which is an important determinant of child development. However, the previous Bangladeshi study of 6-year-olds attempted to take stimulation into account and still found a small deficit in cognitive function associated with arsenic exposure (Wasserman et al. 2007
Arsenic exerts its toxic effects at least partly via oxidative effects (Chattopadhyay et al. 2002
). In the present study, all mothers received folate in pregnancy, and a possible explanation for the lack of a detectable effect is that there was some protection due to folate, which has antioxidant properties (Mukherjee et al. 2006
). Another protective mechanism is the induced metabolism of arsenic in pregnancy (Vahter 2007
). Inorganic arsenic, the main form in the drinking water, is methylated via one-carbon metabolism, producing methylarsonic acid (MMA) and dimethylarsinic acid (DMA), which are the main metabolites excreted in urine (Vahter 2002
). Although a high percentage of MMA in urine is considered a risk factor for a wide range of toxic effects of arsenic (e.g., Li et al. 2008
; Tseng 2007
), a high percentage of DMA is associated with increased rate of excretion and fewer health risks. We found that the mothers of the studied infants had remarkably efficient methylation of arsenic already in early pregnancy (Li et al. 2008
), and it is likely that further improvement occurred with advancing gestation (Concha et al. 1998
; Hopenhayn et al. 2003
The present findings are likely to reflect mainly intrauterine exposure, because arsenic readily crosses the placenta and breast milk excretes very little arsenic (Concha et al. 1998
). In a subsample in this cohort, the median concentration of arsenic in breast milk was 1 μg/L (Fangstrom et al. 2008
). In this cohort, exclusive breast-feeding lasted 4 months on average, and essentially all children were partly or predominantly breast-fed at the time of the testing at 7 months (Saha et al. 2008
). However, around half the children were taking semisolids or solids at 6 months of age, mostly begun after 5 months. Therefore, a limitation to the study is that the children would have had some small exposure to other sources of arsenic.