Bioequivalence and Bioavailability Forum

OT: TTT subtleties [PK / PD]

posted by d_labes  – Berlin, Germany, 2013-07-19 11:47 (4353 d 16:56 ago) – Posting: # 11025
Views: 22,374

Dear Yung-jin,

❝ we did some research trying to figure out if different algo really made differences. Basically, we got similar results as you did: no differences could be found with AUC0-∞, but λz and AUC0-t (AUC from time zero to the last measurable conc.) could be significantly different.

How could this be? How is AUC0-t influenced by the estimate of lamdaZ. IMHO in no way!

❝ In this thread, I noticed that Detlew mentioned "TTT rule... only restricted to one-compartment model..."

This is a misunderstanding. We talked in this thread about models, especially obtaining k01. And in a model other than a one-compartiment model you can't get k01 by log-linear regression

❝ It raises some questions here.

❝

1. Does it imply that if we want to choose TTT rule to estimate λ_z we need to prove that the study drugs exhibit (or best fit with) the one-compartment model in the body first?
2. Is it something related to "the inflection point of the curve" (i.e., the second point; the first is Cmax) in one-compartment model?
3. Should we put "warning" in bear for this restriction?
4. Also, if TTT rule is a model-dependent algo for λ_z estimation, is it appropriate to add the algo to the part of NCA?

IMHO the TTT rule in it's strict sense - use all points after two-times-tmax for log-linear regression - is indeed only appropriate for concentration-time curve shapes which resemble those from a one-compartiment model. And yes it has to do with the second inflection point of the curve after which the linear part of log(C) versus time begins in an one-compartiment model.

But as a tool to restrict the upper number of points to consider in the terminal phase it may be also used for other shapes. Hopefully the fit criterion you use (adjR2, AIC or whatever you prefer) will pick the linear part. In that sense I have used the TTT rule and found it giving reasonable results.
To be detailed for my method:

1. Start with the last 3-4 points in the terminal part
   
2. Go back adding additional points until you reach the time point after/at TTT.
   
3. Choose the points which gave the best fit for a log-linar regression to obtain lambdaZ.

The TTT is using ideas derived from models, but itself is not a model-dependent algo because it doesn't use any model. Beside the assumption of a log-linear part of the concentration-time curves of course, which is in itself a model conception.
Thus don't worry to much .

Complete thread:

• 1 compartment model with lag time jag009 2013-07-17 15:58 [PK / PD] 
  • solve for k01? Helmut 2013-07-17 16:08
    • solve for k01? jag009 2013-07-17 16:19
      • solve for k01? ElMaestro 2013-07-17 17:09
      • no closed form Helmut 2013-07-17 23:01
  • 1 compartment model with lag time yjlee168 2013-07-17 20:57
  • Package-free solution ElMaestro 2013-07-17 21:53
    • Package-free solution yjlee168 2013-07-17 22:11
      • Package-free solution ElMaestro 2013-07-17 22:27
        • Package-free solution yjlee168 2013-07-17 22:42
    • Package-free solution ElMaestro 2013-07-17 23:21
      • A new R-King was born d_labes 2013-07-18 08:45
        • Brent ElMaestro 2013-07-18 09:07
        • A new R-King was born yjlee168 2013-07-18 10:21
          • OT: TTT in bear Helmut 2013-07-18 20:28
            • OT: TTT in bear yjlee168 2013-07-18 22:17
              • OT: TTT subtletiesd_labes 2013-07-19 09:47
                • OT: TTT subtleties yjlee168 2013-07-20 19:44
                  • OT: beyond TTT? Helmut 2013-07-21 00:08
                    • OT: beyond TTT? yjlee168 2013-07-22 13:45
                      • OT: keep TTT! Helmut 2013-07-22 14:21
                        • OT: keep TTT! yjlee168 2013-07-22 23:42
                          • OT: EOD (here) Helmut 2013-07-23 01:06
      • Package-free solution yjlee168 2013-07-18 10:27
        • Thank you guys! jag009 2013-07-18 16:37
          • Thank you guys! ElMaestro 2013-07-18 18:57
            • Thank you guys! jag009 2013-07-19 20:40