Benchmark dose calculations for methylmercury-associated delays on evoked potential latencies in two cohorts of children.

Katsuyuki Murata,Philippe Grandjean,Esben Budtz‐Jørgensen

doi:10.1111/0272-4332.00034

Abstract

Delays in evoked potential latencies were observed at increased exposures to methylmercury from seafood in two cohorts of children. Because this outcome parameter appeared to be virtually independent of confounders, including cultural differences, a joint analysis of benchmark doses was carried out. Comparable cohort members included 382 Faroese and 113 Madeiran children without middle ear infection or neurological disease at age seven years. Maternal hair-mercury concentrations at parturition in the Faroese cohort ranged from 0.6 to 39.1 microg/g (geometric average, 4.49 microg/g). In Madeira, mothers who had not changed their diet since pregnancy had current hair-mercury concentrations ranging from 1.1 to 54.4 microg/g (geometric average 10.14 microg/g). The mercury-associated delay in peak III latencies at two frequencies (20 and 40 Hz) showed similar regression equations in the two groups of children, and benchmark dose calculations were therefore carried out for the two groups separately and jointly. For a doubling of a 5% prevalence of abnormal results of the peak III latencies at 40 Hz in a linear dose-response model, the benchmark dose for the maternal hair-mercury concentration was 8.79 microg/g for the Faroese children; 8.04 microg/g for the Madeiran children; and 9.46 microg/g for both groups. Results were similar for the 20 Hz condition. Benchmark dose results were substantially lower using a logarithmic or square root curve function, although the difference in fit between the curves was far from statistically significant. The benchmark results using evoked potential latencies are in close agreement with results based on neuropsychological test performance.

Highlights

IntroductionMethylmercury is a well documented neurotoxicant, which can cause serious adverse effects on brain development, especially when exposures occur prenatally.[1,2] While current risk assessment efforts have emphasized neuropsychological test results as important outcome variables,[2,3] delays in evoked potential latencies recorded from the brain have provided useful supplementary evidence of pathological changes in methylmercury poisoning.[4,5,6,7,8] The neurophysiological response parameters require only minimal cooperation by the subject and are unlikely to be affected by socioeconomic or cultural differences.[9]
Emphasis in risk assessment has been placed on the neurological function of children with developmental exposure to this neurotoxicant
Assessment of evoked potential latencies is one such method, which has been shown to be sensitive to methylmercury[4,5,6,7,8] and other neurotoxicants.[9,28]

Summary

Introduction

Methylmercury is a well documented neurotoxicant, which can cause serious adverse effects on brain development, especially when exposures occur prenatally.[1,2] While current risk assessment efforts have emphasized neuropsychological test results as important outcome variables,[2,3] delays in evoked potential latencies recorded from the brain have provided useful supplementary evidence of pathological changes in methylmercury poisoning.[4,5,6,7,8] The neurophysiological response parameters require only minimal cooperation by the subject and are unlikely to be affected by socioeconomic or cultural differences.[9]. Dose-response modeling for risk assessment purposes often uses benchmark dose calculations,[15] and substantial information is available for mercury-associated effects on neurological development and performance on psychological tests.[2,16,17,18] The current study was carried out to determine benchmark dose levels using similar default conditions for the neurophysiological results and to apply the benchmark approach for joint analysis of results from two separate studies. Because the shape of the dose-response functions for methylmercury neurotoxicity is unknown,[2] different curve functions were applied

Methods

Results

Conclusion