Dose-response assessment for developmental toxicity. II. Comparison of generic benchmark dose estimates with no observed adverse effect levels.

Developmental toxicity risk assessment currently relies on the estimation of reference doses (RfDDTS) of reference concentrations (RfCDTS) based on the use of no observed adverse effect levels (NOAELS) divided by uncertainty factors (UFs). The benchmark dose (BMD) has been proposed as an alternative basis for reference value calculations. A large database of 246 developmental toxicity experiments representing 1825 endpoints related to dead implants or malformed fetuses has been compiled for use in evaluating alternative approaches to developmental toxicity risk assessment. Using this database we have compared two approaches for BMD estimation with each other and with corresponding statistically derived NOAELS. Comparisons have been based on proportion of affected litters (litters with one or more affected offspring, a quantal response variable) and on the proportion of affected offspring within each litter (a continuous response variable). A quantal Weibull model was used to calculate generic BMDs for the quantal response variable (QBMDs) and a continuous power model was used to calculate generic BMDs for the continuous response variable (CBMDs) at three levels of additional risk (10, 5, and 1%). CBMD05s (continuous benchmark doses for 5% risk) and CNOAELs (statistically derived NOAELs based on the continuous response variable) were similar, with over 98% of the data subsets having CBMD05 and CNOAEL values that differed by less than an order of magnitude. In contrast, QNOAELs tended to be greater than corresponding QBMD10s. The observed conservatism of the QBMD values relative to the corresponding CBMD values was attributed to two factors, lower maximum likelihood estimates for the quantal model and wider confidence intervals around the maximum likelihood estimates, compared to the continuous model.(ABSTRACT TRUNCATED AT 250 WORDS)