48
best fits the actual data Cp ; instead there may be several
exp
solutions with similar minimum sum of squares. A unique solution can be
obtained only if the drug transferred into other compartments or
transformed into metabolites were analysed in their respective
52 ...
compartments. The plasma data of buprenorphine administered
intravenously to 6 dogs were fitted by nonlinear regression to a sum of
three exponentials (In study #1, Dog A, the data were also fitted to a
53
4-compartment model; Fig. 4, top inset). The fitting was effected
54
with the computer program of Yamaoka et al. (See also appendix I),
where 1/Cp, the inverse of the plasma concentration was the weighting
factor (See appendix II for a discussion of different weighting
techniques). The validity of the triexponential equation,
QPcaic = P e1T + A e~a t + B e" 31 Eq. 2
was confirmed by demonstration that regressions of the various studies
of the weighted residuals
( ^exp ^calc ) / ^calc
Eq. 3
against log Cpca^c gave mean residuals £ slopes and intercepts,
55
which were not significantly different from zero (Fig. 10, Table 2).
The residuals were randomly distributed above and below the regression
line, indicating no bias in the fitting of the chosen model. (Fig. 10)
Outliers were defined as those experimental concentrations which had
residual values, 6, greater than 2 (Eq. 3) which corresponds to a
greater than 200% deviation from the presumed best fitted value. One
such outlier in dog B at 2.5632 mg/kg (Study #3) dose of buprenorphine
was not included in the nonlinear least square curve fitting technique