6 replies to this topic

[Tushar Gaonkar]

Dear all,

 

Recently we have upgraded our phoenix WinNonlin software from 6.1 to 6.2.

 

Now after installation, whenever I reopen my earlier calculated project (Done on WinNonlin Version 6.1), each time NCA object in workflow and its respective NCA parameters output displays in red color.

 

If I repeat the same calculation again in new version, it doesn’t show red highlighted background. Results are also similar as earlier one.

 

What would be the reason behind this highlighted results?

 

Just for example purpose, I have attached one project which is performed in Phoenix WinNonlin 6.1, where NCA object is displayed in red color.

 

Can anyone explain me the correct procedure if I am going wrong?

 

Thanks in advance.

Tushar

 [file name=New_Trial.phxproj size=408675]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/New_Trial.phxproj[/file]

Attached Files

•  New_Trial.phxproj   399.1KB   0 downloads

[Simon Davis]

Tushar, when you say 'repeat' the analysis in 6.2 do you mean you 'refreshed' i.e. re-executed the existing red NCA object?. If so then you are doing the correct process.

Essentially Phoenix is suggesting there maybe a change in settings between 6.1 and 6.2 and is indicating you should re-execute the results. For NCA I can't think of anything that would cause that, however in a Phoenix model there are known enhancements that are highlighted to the user when opening a 6.1 project in 6.2.

I did try to look at the project you attached in 6.1 but it was saved in 6.2. There is one unusual thing in your workflow I noted; why have you linked the BE object to a copy of the Final Parameters file in the data folder - this means that the data is now static whereas the 'expected way to build that workflow would be to link the NCA final parameters sheetdirectly to the BE tool

I've attached an example of the suggestion - Can you out of interest open this 6.1 project and tell me what you see in 6.2

Simon. [file name=New_Trial_BE_linked.phxproj size=428456]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/New_Trial_BE_linked.phxproj[/file]

Attached Thumbnails

• 
• 

Attached Files

•  New_Trial_BE_linked.phxproj   418.41KB   401 downloads

[Helmut Schütz]

Dear Tushar and Simon,

 

just my two cents: In 6.2's NCA the variable name of extrapolated AUC is different from earlier versions. Previously the percent character was displayed as an underline. Since you have a new variable, the NCA object is highlighted in pink.

[Tushar Gaonkar]

Dear Sdavis,

 

I am not able to open the attachment provided by you.

can you provide me the password to open that RAR file?

here is my email id if you want-

tushargaonkar@hotmail.com

 

Regards,

Tushar

[Simon Davis]

Tushar, my guess is you are using Internet Explorer as I've heard some other users complain about this too; that attachment is a normal Phoenix project - however it seems some browsers rename the file when downloading, simply rename the extension back to .phxproj and open in Phoenix.

 

Please confirm your browser type and if this works, sorry for the problems -we're looking at how to fix these issues in the forum (I know the latest posts link is also somewhat unresponsive currently and I have a request to address that too).

 

   Simon.

[Tushar Gaonkar]

Dear Sdavis,

From your first reply, where you were asking about repeat calculation, I meant if I completely repeat the study (i.e. from inputting data of concentration into data folder to till BE object) in version 6.2. That time it doesn’t show red highlighted results.
Same way you mentioned, if I re-execute the workflow of earlier studies in new version then also result doesn’t get highlighted.
Is it new feature of WinNonlin 6.2 (to Eexecute workflow)? Because in last version I couldn’t find it any

The attachment you have provided is accessible after changing its extension. Thanks.
In that, NCA and Bioequivalence both objects are red highlighted in new version. But after re-executing the total workflow, it displays as normal.

Finally, As far as that unusual thing in my workflow is concern, actually I had tried my best to connect BE tool directly to NCA object but couldn’t find the way to add period and sequence column in Final parameteer pivoted sheet (in NCA output).
Thanks for guiding me how connect BE object to NCA.
By using this technique provided by you, I got the same results as compared to yours, except AUC values. Because you have utilized “Linear Up and Log Down method” for AUC calculation whereas I have used “Linear Trapezoidal Linear Interpolation” method.

Many thanks to you and Helmut for your kind replies
Regards,
Tushar

[Helmut Schütz]

Dear Tushar!

I have opened almost all of my 6.1 projects in 6.2 now, and can confirm that always the NCA was pinky. Now I'm pretty sure that the reason is the changed variable name of AUC-extr. Nothing to worry about.

I must confess that I haven't looked at Simon's project, but I guess his mapping is essentially the same as my suggestions in the following.

Is it new feature of WinNonlin 6.2 (to Eexecute workflow)? Because in last version I couldn’t find it any

It was in exactly the same place. ;-)

As far as that unusual thing in my workflow is concern, actually I had tried my best to connect BE tool directly to NCA object but couldn’t find the way to add period and sequence column in Final parameteer pivoted sheet (in NCA output).

In the setup of NCA, you did not map Period and Sequence. Map both of them to "Carry".

… you have utilized “Linear Up and Log Down method” for AUC calculation whereas I have used “Linear Trapezoidal Linear Interpolation” method.

Consider revising your SOP. Lin-up/Log-down is a much better choice, especially if you have samples missing within measured ones (problems in sampling, vial broken, etc.). Especially in the elimination phase with the linear trapezoidal you will get a positive bias (=too high AUC), since measured data points are "connected" by a straight line – which does not reflect an exponential decrease.
See this presentation (slide 15). I excluded one measured value from the dataset; lin/lin gives a bias of +3.49& whereas lin/log only +0.14%.

You may use the linear trapezoidal if you are studying the elimination of C₂H₅OH. But that’s another story.

BTW, your dataset pushes the R²adj-method far beyond its limits. I strongly hope you don’t rely on the lambdaz-values?