Change In The Analytical Error Function Depending On The Calibration Curve Range: Implications For Pharmacokinetic Parameter Estimation Of Carebastine.

Modamio P, Jansat JM, Lastra CF, Mariño EL

,1Clinical Pharmacy and Pharmacotherapy Unit, Department of Pharmacy, Faculty of Pharmacy, University of Barcelona, Barcelona, Spain, 2 Department of Pharmacokinetics and Drug Metabolism, Grupo Almirall S.A., Barcelona, Spain..

The error function of a previously validated analytical procedure used as a weighting method in least squares nonlinear regression analysis, could be a good alternative of other classical weighting methods. Nevertheless, the selection of a suitable concentration working range according to the experimental data has to be considered.

In this study, the concentration levels of the carebastine calibration curve were 25, 50, 100, 200, 300, 500 and 1000 ng/ml. The study of the analytical method error function was carried out in two concentration ranges: one from 25 to 500 ng/ml and another from 25 to 1000 ng/ml. The best analytical error functions obtained were

(1) SD = 0.6089 + 0.0438 C – 9.5868 x 10-8 C3 and (2) SD = 4.2949 + 3.6115 x 10-5 C2 respectively.

After that, considering the inverse of the concentration variance (1/Vi) as a weighting method, the corresponding weights (wi) for each concentration value were obtained from both analytical error functions and subsequently compared (Table).

C (ng/ml)

SD1*

SD2**

Wi1 = 1/Vi*

Wi2 = 1/Vi2**

25

1.7024

4.3175

0.3451

0.0536

50

2.7869

4.3852

0.1288

0.0520

100

4.8930

4.6561

0.0418

0.0461

200

8.6020

5.7395

0.0135

0.0304

300

11.1605

7.5453

0.0080

0.0176

500

10.5254

13.3237

0.0090

0.0056

1000

-51.4591

40.4099

0.0004

0.0006

* obtained from 25 to 500 ng/ml calibration curve
** obtained from 25 to 1000 ng/ml calibration curve

After the application of a Student’s t-test, no significant differences (P>0.05) between weights corresponding to the concentration range from 25 to 500 ng/ml were found.

As can be seen from the Table, the greatest concentration (1000ng/ml) has a negative standard deviation when the analytical function obtained up to 500 ng/ml is applied. This fact confirms that it is not possible to extrapolate the analytical error function out of the calibration curve range where it is used as a weighting method.

In conclusion, the results obtained show that when an analytical error function is used as a weighting method, it is especially important to select correctly the concentration working range of the calibration curve according to the drug concentration in real plasma samples.

Reference: PAGE 6 (1997) Abstr 662 [www.page-meeting.org/?abstract=662]

Poster: poster