6 replies to this topic

[liwangatptgx]

Dear Certara team,

 

Need to calculate the accumulation factor after multiple dosing. The dosing regimen is QD, but there are 0s for later sampling time-points after each dosing. Then which one would be an appropriate calculation, the ratio of AUClast or the ratio of AUC0-24?

 

The second question is how the AUC0-24 calculated? For example, if the concentrations at 12 and 24 hrs are BLQ,

 

AUC0-24 = AUClast + Clast/lambda z (AUCinf, if Lambda z available)

 

How would it be calculated if Lambda z cannot be estimated?

 

or

 

AUCtau = AUClast + Clast/slope (AUCall, extrapolate to 12 or 24 hrs?)

 

Thank you very much for your help!

 

 

 

[Simon Davis]

Please have a look at this thread, in particular the links Helmut gives; https://support.cert...ulation-of-auc/

[liwangatptgx]

Please have a look at this thread, in particular the links Helmut gives; https://support.cert...ulation-of-auc/

Hi Simon,

 

Thank you for your reply. I have read the thread you referred to. I understand that AUClast may not be equal to AUCtau if you sample past 24 hrs while AUCtau is only 0-24 for QD. But I do not quite understand if there are 0s for later sampling time-points (such as 12 and 24 hrs), should AUCtau or AUClast be used for accumulation factor calculation. If it is AUCtau, how would it be calculated in the cases of Lambda z available or unavailable, respectively?

 

Thanks,

Li

[Simon Davis]

Li and you read the User manuals? specifically "Partial areas" and  "AUC calculation and interpolation formulas"

 

I would think as you are looking at multiple dosing you would select AUCtau; the rules for extrapolation/interpolation are consistent between this and AUC last.

 

If you have these questions, why not construct a set to explore what happens; in the attached project I have 2 subjects where the first has a Lambda_z calculated and the second not.  Then for each subject I have the same profile except if the last 2 points are left blank i.e. missing (or indeed any text value) or set to 0.

 

  You can see in this table how that affects them;

 

so to answer your second question "how the AUC0-24 calculated? For example, if the concentrations at 12 and 24 hrs are BLQ" it will depend on how you code BLQ, if you set them to 0 that will be interpereted as a 'real' value.  If you leave them as BLQ then....

 

if Lambda_z is calculated it will beable to extrapolate out.

 

if Lambda_z is *not* calculated then it cannot extrapolate beyond Clast, if they are all BLQ.  if you have replaced with zeros then it will simply add a triangle at the end from Clast to the next sampled timepoint (i.e. like AUCall)

 

I wouldn't necessarily recommend this as it will be highly dependent on interval between samples.  Maybe you should consider using an AUC to last common timepoint?

 

  Simon.

 

Attached Thumbnails

• 

Attached Files

•  testing_pAUC.phxproj   240.2KB   465 downloads

[liwangatptgx]

Thank you Simon for your explanation.

[Helmut Schütz]

Hi Simon & Li,

 

Maybe you should consider using an AUC to last common timepoint?

 

That’s an excellent idea and the only one (!) which will give unbiased estimates if comparing two treatments after a single dose (see this presentation (slides 19–24) and the paper by Fisher et al.).

 

Estimating accumulation can be tricky (see this post in the BEBA-Forum). To assess a potential deviation from linear PK one would compare AUC_0–τ (steady state) with AUC_0–∞ (single dose). In this case I would use PHX/WNL’s AUCtau (which should be identical to pAUC0-tau) and exclude subjects from the comparison where a reliable estimate of λ_z is not possible.

 

[Helmut Schütz]

Hi all,

 

attached an example showing the setup to calculate AUC up to the common t_last. Simulated data (acc. to: Csizmadia F, Endrényi L. Model-Independent Estimation of Lag Times with First-Order Absorption and Disposition. J Pharm Sci. 1998;87(5):608–12. doi:10.1021/js9703333). I set ƒ of the reference to 1 and of the test to 0.95. LLOQ was 1% of the theoretical C_max of the reference (185.75). If you have deviations from the planned times you need both the scheduled and actual sampling times.

 

Steps:

1. NCA based on the scheduled (!) sampling times.
   Disable Curve Stripping and all Plots. Include in the workbook only Tlast.
2. Final Parameters Pivoted → Split Worksheets (sort by subject and Tlast).
   Subjects with Count < the number of periods have different t_last-values for R and T.
3. Unique Values → Descriptive Stats. In the column Min you get t_last (Common).
4. Join (or merge) with the original data (sort by subject).
5. Result → Data Wizard / Filter: Include only scheduled times which are ≤ Min.
6. Result → NCA as usual (with the actual sampling time points).

That’s it. As shown by Fisher et al. there is a difference in AUC_0–last but AUC_0–∞ is pretty similar.

The project was developed in PHX7.0. In later releases you have even more options to play around with partial AUCs & the like.

 

 

Attached Files

•  Tlast_common.phxproj   1.06MB   413 downloads