3 replies to this topic

[Tushar Gaonkar]

Dear All,

 

I have Phoenix WinNonlin software Version 6.2.

Currently I am struggling with calculation of reference replicated approach.

Is there any way to perform the reference replicated calculation (3 way and 4 way) using Phoenix WinNonlin 6.2?

As per US FDA guidelines, in reference replicated approach we need to deal with intra-subject variability of Reference, 95% upper confidence bound etc.

Can I calculate all these stuff using phoenix WinNonlin Software?

I tried some customized data (based on sequence RRT, RTR, TRR) on WinNonlin under population/Individual section, but I am getting ERROR 11107.

 

Thanks in advance.

 

Regards,

Tushar

 

 [file name=Ref_Replicated_Project.phxproj size=465501]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/Ref_Replicated_Project.phxproj[/file]

Attached Files

•  Ref_Replicated_Project.phxproj   454.59KB   0 downloads

[Simon Davis]

Hi Tushar, when you get an error message like that it might be better to log a Support issue; the reference decodes as;

"11107 Fewer than 2 sequences had complete design. Program terminating.", as detailed in the Warnings message."

Unfortunately at this time we have not written a direct routine for RSABE, Helmut Schuetz has done some considerable work on replicating the necessary steps in a Phoenix workflow but it is not a simple matter at this time.

http://forum.bebac.a...p?id=7871#p7931

In 2012 we hope to spend some time reviewing and updating the BE Wizard, the RSABE enhancement is QC_PHX6269, however this will be after the 6.3 release which is in final testing now.

However, please do continue to discuss and share ideas as please be assured we are looking at this board to note feature requests etc.

Simon

PS As a side note I would recommend that for NCA you;

1) Use LinLog or LinUpLog down AUC methods
2) Review the WNL auto fit of Lz, especially in data with high variability for consistency etc.
3) Specifically in your case, investigate the 48 h samples that are showing significantly higher in many profiles than the 36h time point. are you sure they are not contaminated by the subsequent dose period?

The above graph is easy to replicate using the Summary Table as Source and mapping both predicted (to get Lz line) and Conc to get observed points. Different sorts and groupings can help you visually compare your groups.

/********************************* Begin Mappings *********************************/
XY Data : Ref Replicated Project.Workflow.NCA 9 Vol.Summary Table
X : Time  
Y : Predicted [ng/mL], Concentration [ng/mL]
Y2 :
Group : Subject
Data Label :
Row :
Column :
Page (Sort) : Formulation, Period, Sequence

Attached Thumbnails

• 

[Helmut Schütz]

Dear Tushar,

I've attached the project mentioned by Simon. I used EMA's two datasets (4 period, 2 sequence full replicate and 3 period, 3 sequence partial replicate). As you can see in the linked thread in the BEBA-Forum results match the ones obtained by FDA's SAS code give in the progesterone guidance. The workflow is quite complicated (suggestions for improvement welcome) – use at your own risk!

[file name=FDA_Replicate.phxproj size=3209778]http://www.pharsight.com/extranet/media/kunena/attachments/legacy/files/FDA_Replicate.phxproj[/file]

Attached Files

•  FDA_Replicate.phxproj   3.06MB   475 downloads

[Helmut Schütz]

Hi Tushar,

I reviewed your project and agree with Simon. Never perform a parametric analysis on discrete data (e.g. t_max)! BTW, both the FDA and EMA do not require a statistical analysis. Some hints in this thread in the BEBA-Forum.