Biomedical Engineering Reference
In-Depth Information
steady-state human blood concentrations generated with the linear
kinetics of the Bouchard et al. (2001) model.
6.5 SHOULDN'T THE POSSIBILITY OF SYSTEMATIC BIAS
BE CONSIDERED CAREFULLY DURING THE MODEL
FITTING AND PARAMETER ESTIMATION PROCESS?
As noted previously, significant dose-related effects of methanol on the
frequency of cervical rib malformations in mouse pups from pregnant
dams exposed to methanol 7 hours/day via inhalation were observed in
the Rogers et al. (1993) study. The key exposure levels were 1000, 2000,
and 5000 ppm, and the grand mean measured blood methanol concen-
trations immediately after selected 7-hour exposure periods (days 6, 10,
and 15) were 97
16, 537
77, and 1690
160mg/l (mean
SEM),
respectively (see Starr and Festa (2003), Table 1). The corresponding
Battelle (2006) model-predicted blood concentrations (C
max
) are 134,
757, and 2000mg/l, respectively (Battelle, 2006, Table 2), values that
are 38, 41, and 18 percent higher, respectively, than the measured mouse
blood methanol concentrations. There is no reason to believe that the
extent of this bias toward overprediction does not carry over to the
steady-state situation, so it is reasonable to expect that the discrepancies
between Battelle (2006) model-predicted human and measured mouse
blood methanol levels at steady state might be even greater than those
indicated in Table 6.2 and Figure 6.1.
Figure 3 of the Battelle (2006) report (not shown) compares pre-
dicted and observed mouse blood methanol levels during and after
inhalation exposures using data from Dorman et al. (1995) and Perkins
et al. (1995). The model underpredicts substantially the Dorman et al.
(1995) data for 15,000 ppm at the exposure termination time point and
beyond. It underpredicts substantially the Perkins et al. (1995) data from
the 5000 ppm exposure group at all time points.
Figure 4 of the Battelle (2006) report (not shown) compares pre-
dicted and observed mouse blood methanol levels following oral
exposures using data from Dorman et al. (1995) and Ward et al.
(1997). The data for oral exposures of 2500mg/kg are substantially
