Benchmark Concentration (BMC) Modeling Versus the NOAEL/LOAEL Approach: A Case Study With Methylene Diphenyl Diisocyanate. M. M. Greenberg and J. Gift, National Center for Environmental Assessment, U.S.E.P.A., Research Triangle Park, NC 27711
Chronic inhalation exposure of male and female rats to three concentration levels of methylene diphenyl diisocyanate (MDI) results in a statistically significant increase in the incidence of nasal olfactory degeneration at the highest concentration and an apparent increased concentration-related trend. Dose-response evaluation was by BMC, a modeling procedure increasingly being used by the U.S.E.P.A. to bypass the limitations of the default NOAEL/LOAEL approach for noncarcinogens. The BMC is defined as the 95% lower confidence bound on the concentration (Maximum Likelihood Estimate or MLE) producing a predetermined level of change (e.g., 1, 5, or 10%) in adverse response compared with the response in untreated animals. Seven different models available from software being developed by the U.S.E.P.A. for public dissemination were applied to the combined data set. Of the seven models, the logistic and probit models failed to compute BMC estimates and both were excluded from further consideration. Of the remaining models, all provided acceptable fits to the data whether parameters were restricted or not. The MLE at the 10% extra risk level ranged from 0.16 to 0.64 mg/cu.m compared to the default-identified NOAEL of 0.18 mg/cu.m. The BMC at this level of extra risk (BMC10) ranged from 0.03 to 0.54. Because (1) the shape of the incidence curve from males may be dissimilar from that of females, and (2) there are only three data points with one widely spaced from the other two, the true shape of the concentration-response curve is unknown. In this instance, it is difficult to determine which model best fits the data. The wide range of the BMC10, because it includes estimates near zero, suggests that the MLE is more biologically realistic than the BMC10. Thus, the use of the MLE derived from the Weibull model (0.16 mg/cu.m) could be recommended over the BMC10 because it provides the most conservative estimate of the NOAEL and is similar to that value identified through the default approach. This analysis suggests that the BMC approach will have more utility when applied to data sets with (1) more data points more evenly spaced over the experimental range and (2) with concentration-response curves that models more clearly differentiate.
Go to . . .