Calculation of AUC

Dear all,
I do not clearly understand the difference between AUClast and AUCall in the NCA output. Maybe somebody can give me some example with explanation? And If right to use AUClast as data for AUC(0-t)?
Is there some rules in guidelines of using particular method for calculating AUC in PK model? Is there some information of exact difference between linear up log down and linear log trapezoidal?

Looking forward to your kind replies!
Best wishes,
Veronika.

Dear Veronica,

AUCall is not an ideal parameter to use because if you have zeros in your profiles the trapezoids are constructed using the 0 times.  If you don't have 0s then AUClast=AUCall.  In general, I tend to not report AUCall.  Recall that AUClast is to the Last MEASURABLE concentration.

If you have a profile that is already at steady state, AUClast might not be equal to AUCtau, that would depend on your sampling. For example, if you dose every 24 hours for 7 days and after the 7th dose you measure concentrations but you sample past 24 hours  then AUClast will not be equal to AUCtau.  AUCtau is only 0-24 for that last profile  (or times 144 to 168), but you could have sampled past hour 168.

There is a lot of information in the WinNonlin User's guide about calculation of AUC. What you need to keep in mind is that whatever method you used it should be documented.  If you have very rich profiles the AUC methods do not differ greatly.  My preference is always linear up log down.

Hope this helps,

Ana

Dear Veronica,

AUCall is not an ideal parameter to use because if you have zeros in your profiles the trapezoids are constructed using the 0 times.  If you don't have 0s then AUClast=AUCall.  In general, I tend to not report AUCall.  Recall that AUClast is to the Last MEASURABLE concentration.

 

If you have a profile that is already at steady state, AUClast might not be equal to AUCtau, that would depend on your sampling. For example, if you dose every 24 hours for 7 days and after the 7th dose you measure concentrations but you sample past 24 hours  then AUClast will not be equal to AUCtau.  AUCtau is only 0-24 for that last profile  (or times 144 to 168), but you could have sampled past hour 168.

 

There is a lot of information in the WinNonlin User's guide about calculation of AUC. What you need to keep in mind is that whatever method you used it should be documented.  If you have very rich profiles the AUC methods do not differ greatly.  My preference is always linear up log down.

 

Hope this helps,

Ana

Ana, thank you! Your answer is helpful for me!

Dear Veronika,

first of all I agree what Ana wrote. I would never ever use AUCall. For some details see this thread in the BEBA-Forum.

Furthermore, I think that the linear trapezoidal method (with very few exceptions: saturation / Michaelis-Menten PK) should go to the waste-bin of NCA. I recommend the lin-up/log-down method instead.

Dear Helmut,

I've always doubted about using  lin-up/log-down or linear trapezoidal method. Now I know exactly, thank you!

Hi Veronika,

Dear Helmut,
I've always doubted about using lin-up/log-down or linear trapezoidal method. Now I know exactly, thank you!

IMHO, the linear trapezoidal is a remnant of the dark ages when people draw profiles on scale paper – naturally connecting data by straight lines –, cut them out, and weighed them on an analytical balance. Should have been abandoned once pocket-calculators were invented.

The linear trapezoidal gives biased results especially when samples are missing. See this example (slides 10–11).