Dear Markus,
You have a lot of minor problems in the controls stream
-FO is outdated now; there are no reasons not to use FOCEI; you may have problems with ALAG1 and FOCEI, but you can either use transit compartments to imitate absorption delay, or ignore ALAG1 variability, or use hybrid method (FO for ALAG1 and FOCE for everything else).
-you defined Y twice; the first one can be removed
-W should not be defined as LOG(F). You have additive error (in log scale) so W=1 (or if you like to normalize, W = SD of corresponding error) -you can use ADVAN5 (since you have linear system) rather than ADVAN6; it will be much faster; - TOL=5 is low; sometimes, this could be problem. I would increase it to 8, 9, or even 10. If the model complains, you may use ADVAN13 instead of ANDAN6;
-C2 etc. in $DES are not used and can be removed.
-fixed volumes V3, V4, V5: why it is fixed to 1? You have K20, so V3 can be fixed to anything, but why 1? Also, fixing V3-V4=V5 assume complete metabolism of these steps and equal volumes of distributions, is this correct? To start, I would remove K20, and allow estimation of V3, V4, V5. If the fit is good, then you can impose assumptions
None of the comments above should affect the fit, except the very last one, about volumes, but even that one would not affect the parent's fit (individual predictions).
If by "early concentrations" you mean absorption phase, you may want to try a more elaborate absorption model. If you see cycle dependence, you may need to introduce time-dependent parameters. Some hint of time-dependency can be found in the literature:
http://www.ncbi.nlm.nih.gov/pubmed/11286326
"Plasma concentrations of fluorouracil increase by 10 to 60% during long term administration..."
Regards
Leonid
--------------------------------------
Leonid Gibiansky, Ph.D.
President, QuantPharm LLC
web:
www.quantpharm.com
e-mail: LGibiansky at quantpharm.com
tel: (301) 767 5566
Quoted reply history
On 12/15/2012 3:24 AM, markus joerger wrote:
> dear NMUSERS,
>
> I am presently working on a capecitabine model (n=12, increasing levels
> of liver dysfunction, 3 with additional renal dysfunction).
> While the model works in FO, early capecitabine-concentration estimates
> are too low, and this does not improve with adding a peripheral
> compartment for capecitabine.
>
> May some of the esteemed readers suggest how to improve the model for
> the parent compound's early time-points?
>
> thanks so much and best regards!
>
> --
> Markus Joerger MD-PhD
> Department of Medical Oncology&Hematology
> Clinical Trial Unit
> Cantonal Hospital
> St. Gallen
> Switzerland
> [email protected] <mailto:[email protected]>
> [email protected] <mailto:[email protected]>
> Phone: +41-765591070
> Fax: +41-714946325