Hello,
I am doing a crossover 2x2 bioequivalence analysis. I was wondering why winnonlin did not compute the parameters for a few subjects. The Final Parameters Pivoted plan is attached, it seems it is using 0 points for estimating excretion regression rate, but this particular subject concentrations are not any different from other (I mean just like others there are a few BQL), so I would like to understand why that is happening. Many thanks.
final parameters.xlsx (12.5 KB)
HI Ailton,
I see just one profile with the ‘problem’ you describe:
formulation subject period Cmax AUClast AUCINF_obs Tmax
R 3 II 2.05 12.31 6.00
I can’t see your whole profile becuase you didn’t post the raw data but it appears that is simply coulnd’ calculate and AUCinf.
DId you let WNL select it’s own Lz points? if so it might be worthwhile reveiwing this profile in particular and seeing if you can do something manually with the data you have. Perhaps you will have to include Cmax which WNL will not do automatically since v 6.3
Also I’d note that the exposure is looking rather low by Cmax 2.05 and AUClast any way when some others have values of 99 etc., so I can guess it’s not BE anyway, sorry.
Simon
Hi Simon, thanks for you reply.
I am attaching a text file with some raw data, including subject 3, who is causing the problem. sample size n=87.
Overall, the treatments are no bioequivalente, but I wanted to understand what winnonlin tried to do and did work.
I let it to select the Lz points.
Ailton
raw data for a few profiles.txt (2.14 KB)
OK looking at your raw data you may well have another problem if that is how you are working with Phoenix, it requires you to be using a decimal point, not a comma, and you should set that up at your OS level before restarting Phoenix.
Take a look at the project I’ve attached - it’s clear that with subject 3’s Cmax occuring at 6h, the penultimate quantifiable timepoint (tlast=8h) it is very difficult to be confident in estimating an apparent rate of elimination. You may need to do some modelling first to get a better understaning before you go to the next design/formulation. WIll you be working with an IVIVC?
Since your compound is also probably 2 compartment, in my opinion I would take extra care with your assay to ensure are capturing this second phase for all profiles if at all possible and select that manually.
Simon.
You may also want to look at taking one of our online courses e.g. NCA;
http://www.certarauniversity.com/lms/index.php?r=course/details&id=43
or the intro to Phoenix has a lot on NCA
http://www.certarauniversity.com/lms/index.php?r=course/details&id=26
Simon.
ailton.phxproj (505 KB)
Simon,
I am not working with an ivivc. I am just doing a bioequivalence study.
I will take a look at the data.
Ailton.
OK - I was just suggesting the process of building an IVIVC can help you and the formualtion scientists better plan your vivo studies. Good luck, Simon.
Hi Simon,
According to the Winnonlin guide pp.32-34, it needs at least three non-zero points after the cmax in order to compute de elimination rate, subject 03 (and others) only has one (or two), hence there is no parameter estimation for these.
Ailton.
Exactly Ailton, that is the ruleset of the automatic Lambda_Z selection algorithm, however if you want/need to, you can overule them, but you should consider carefully the science of what you are trying to do when using NCA to describe your data in this way.
Simon
Dear Simon,
How can I overule it on winnonlin? I have a dataset (it is not a BE study) in which the PK that depends on lambda_z were not computed, we know the reasons pointed above, but how do I force WNL to compute lambda_z?
If you would like to take a look, please see attachment.
Thanks.
New Project.phxproj (266 KB)
Dear Ailton,
I kindly suggest to look into WNL User’s guide, pp.36-37 (Slopes Selector Panel)
There you can find the way to do that manually.
Another approach could be seen in the examples chapter of the guide, pp.302-307
Hope it helps,
Mittyright
