Back to the Top
Hi All,
For non-compartmental analysis there is option for manual slope range selection for Lambda Z. I did NCA for one data set. The R square value is similar when we are using last 3 time points or 3 time points (by excluding 1 time point from last 4 time points) from elimination phase. But the result for half-life shows two fold variation between the two approaches. Is there any specific guideline according to FDA or any other guidelines for selection of slope for Lambda Z.
Thanks,
Nagsen Gautam
Back to the Top
The following message was posted to: PharmPK
in that case better we go with the best fit option or else you can fix
the regression coefficient value to somewhere value which varies
according to the company SOP and the FDA guidelines.
Back to the Top
The following message was posted to: PharmPK
Nagsen,
There are no specific guidances on how to select data points for calculation of Lambda Z (terminal elimination rate). The FDA recommends at least 3 points. Health Canada recommends at least 4 points (in some cases). You can watch a short instructional video on the topic at: http://www.youtube.com/watch?v=ZIHbhdihcC8 My entire set of training videos is available at http://www.youtube.com/learnpkpd
And my PKPD blog which contains a post about this topic is located at http://blog.learnpkpd.com
Hope that helps,
Nathan Teuscher
Back to the Top
Dear Vijaya: I do not use the lambda_z values unless the adjusted r^2 is =/> than 0.9000. Also I do not use the AUC (0-inf) values in the statistical analysis for bioequivalence assessment if the extrapolated AUC is =/> 20%. I think if you follow these simple rules you will be dealing with credible data.
Hope this helps, Aziz Karim
Back to the Top
The following message was posted to: PharmPK
Dear Nagsen!
As Nathan already mentioned there is no guideline...
R2 (or R2adj) gives you only some hints - setting a arbitrary limit for the acceptability of the fit is nonsense.
We already had a long discussion back in 2002 (
http://www.pharmpk.com/PK02/PK2002227.html
See also a lengthy thread at http://forum.bebac.at/mix_entry.php?id=1572
If you are lucky and the drug can be described by a one-compartment model the "TTT-method" is promising:
Scheerans C, Derendorf H and C Kloft
Proposal for a Standardised Identification of the Mono-Exponential Terminal Phase for Orally Administered Drugs
Biopharm Drug Dispos 29, 145-157 (2008)
If the drug follows a multi-compartment model, quite often the best fit is obtained within the last inflection point and the last sampling point.
We have performed simulations comparing the automated methods (our manuscript is at the bottom of the to-do-pile...). As expected especially flip-flop PK (ka>=kel) is nasty. Though we have found some differences between methods in terms of bias for lambda-z, the differences diminish when it comes to the calculation of AUCinf.
Commercial software (e.g. Phoenix/WinNonlin, Kinetica) use solely max. R2adj.
Five automated methods (max. R2adj [ARS], min AIC, TTT, TTT-max. ARS, and TTT-min. AIC) are available in the freeware package bear (
http://cran.r-project.org/web/packages/bear/index.html
for R (
http://cran.r-project.org/
Automated methods fail quite often for MR formulation ("flat" peaks) and almost ever for pulsed formulations (multiple peaks).
All (!) papers and textbooks emphasize the importance of inspection of automated fits and corrective measures if deemed necessary (=eye-balling).
I would not be so radical like Maria Durisova stating in 2008 (
http://www.pharmpk.com/PK08/PK2008035.html
AUCt to AUCinf ratio is meaningless. The reason for this is that time "t" is selected by a pharmacokinetician, consequently AUCt to AUCinf depends on the pharmacokinetician choice.
BTW, it's interesting that regulators tell us at conferences etc. that they accept only unweighted log-linear regression in NCA. On the other hand, they have no problems in dealing with quite complicated aggregate weighting schemes in PopPK. Maybe that's another department. ;-)
Helmut
P.S.: Oldies, but goodies:
FJ Anscombe and JW Tukey
The Examination and Analysis of Residuals
Technometrics 5/2, 141-160 (1963)
Boxenbaum HG, Riegelman S and RM Elashoff
Statistical Estimations in Pharmacokinetics
J Pharmacokin Biopharm 2/2, 123-148 (1974)
-- Ing. Helmut Schuetz
BEBAC - Consultancy Services for
Bioequivalence and Bioavailability Studies
Neubaugasse 36/11
1070 Vienna, Austria
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Non Compartment Analysis - lambda z" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)