Determination of Concentration-Time-Mortality Relationships to Replace LC50 Values

Abstract

To determine concentration-time-mortality relationships of ammonia (NH₃), chlorine (Cl₂), hydrogen sulfide (H₂S), and phosgene (COCl₂) in acute inhalation toxicity studies with rats and mice, groups of five males and five females each were exposed for different periods of time to different concentrations of the respective test atmospheres. The consequences of a decrease in the number of animals per group on the accuracy of the LC50 values calculated form the estimated relations were studied by analyzing mortality rates in new sets of data obtained by removing one, two, three, or four animals per sex from the original group results in a random fashion, 500 times for each test compound.

LC50 values for different durations of exposure were calculated with the newly estimated concentration-time-mortality relationships and the 500 LC50 values were characterized by their fifth, fiftieth, and ninety-fifth percentiles. Furthermore, the mean and standard deviations of the coefficients of the calculated relationships were determined. Within the range of exposure times used in these studies, the fiftieth percentiles were scarcely influenced by the number of animals per sex per group, whereas the fifth and ninety-fifth percentiles covered a larger range when decreasing the number of animals, reaching about ±10% when four animals per sex per group were removed. In the latter situation a small number of draws showed no convergence during estimation of the concentration-time-mortality relationship. Standard deviations of the coefficients of the relationships increased considerably when the number of animals per sex per group was decreased from two to one due to a loss of information on the heterogeneity in some draws.

It is concluded that LC50 values in the range of duration of exposure applied could have been estimated with one animal per sex per group. The resulting fifth and ninety-fifth percentiles in that case compare favorably with the 90% confidence limits when determining an LC50 according to OKD guideline 403. When extrapolation to low mortality rates is needed, two animals per sex per group seem to determine the lower limit of animal use.