TABLE 4 Regression of Upper Bounds on Low Dose Slopes on the Maximum Tolerated Dosea
|
Regression Parameter |
Extrapolation Method |
|
|
Multistage Model |
Model-Free Extrapolation |
|
|
Intercept ± SE |
0.01 ± 0.05 |
0.11 ± 0.04 |
|
Slope ± SE |
-1.05 ± 0.03 |
-1.07 ± 0.02 |
|
Correlation |
0.944 |
0.961 |
|
Root Mean Square (s) |
0.462 |
0.386 |
|
Factor 102s 95% Prediction Intervalb |
8.4 |
5.9 |
|
aBased on simple linear regression of log slope on log MDT. bUpper limit is 102s x MDT; lower limit is 10-2s x MDT. |
||
approximate 95% prediction intervals for the low dose slope encompass a range of about 8 × 8 = 64-fold about the MDT with the LMS model, and a range of about 36-fold for MFX. Given an upper bound on the low dose slope ß, the corresponding 10-6 RSD is simply 10-6/ß.
6. Interspecies Extrapolation
Since mammalian species share many common physiological characteristics it is expected that they may respond in a somewhat similar manner to toxic substances. While many differences exist between species (Oser, 1981), allometric relationships among physiological parameters have suggested different metrics for quantitative interspecies extrapolation: heat loss, for example, appears to be proportional to the surface area of mammals, whereas metabolism is related to body weight to the 3/4 power (Schmidt-Nielsen, 1984). Such considerations have led to suggestions that allometric equations of the form
