Helmut
★★★
avatar
Homepage
Vienna, Austria,
2011-07-26 18:03
(4679 d 21:22 ago)

Posting: # 7226
Views: 11,398
 

 Puzzle: Estimation of half life [Surveys]

Dear all!

From a [image] consultant’s diary:

[image]

  • Study performed in 2011 | IR formulation | very common drug | large CRO.
  • Of course, the report stated “correlation coefficient mandatory >0.95” and “PK parameters were calculated in a blinded manner”…
  • Maybe the pharmacokineticist was blind?
  • PK followed a 2- or 3-compartment model in all subjects.
  • No plots of the estimation of λz in the report.
    I generated the one above by trial and error to match tabulated half lives.
  • No table showing how many and which data points were used.
What do you think?

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
ElMaestro
★★★

Denmark,
2011-07-27 12:58
(4679 d 02:27 ago)

@ Helmut
Posting: # 7227
Views: 9,547
 

 Puzzle: Estimation of half life

Hi HS,

❝ What do you think?


A wise guy at the CRO probably got the brilliant idea to write a script that does the linear regression from the first points after Tmax and onwards.
This is a brilliant idea for several reasons:
  1. All molecules, across all therapeutic classes in all humans subjects are eliminated in a first-order fashion. We all know that. Higher order elimination is free fantasy, like gnomes or eskimoes.
  2. All molecules are polite. They wait patiently until all their fellow molecules have been absorbed into the blood stream before they commit themselves to elimination.
  3. A script saves time and money; it takes a computer 2/10000000 of a second to do the calculation and you do not have to apply any degree of common sense or other processes that involve thinking or other resource-intensive activity.
Probably the wise guy has already been promoted as we speak?

Pass or fail!
ElMaestro
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2011-07-27 17:10
(4678 d 22:15 ago)

@ ElMaestro
Posting: # 7228
Views: 9,551
 

 Puzzle: Estimation of half life

Dear ElMaestro!

Great ideas!
But I guess it’s even worse. The half-life of this drug is reported in the literature with 1–4 hours. The CRO has a pretty good analytical method – with a worse LLOQ you wouldn’t see the slower phase(s). He/she/it didn’t run a script, since in some subjects with lower concentrations he/she/it was happy with just the last 3–4 points. I think that he/she/it tried to tweak the results in such a way that results are not too far away from the literature.
According to the report (54 profiles): x̃ 4.61 h (2.49 – 8.34 h), while my eyeballing PK gives 5.05 h (2.78 – 8.34 h). Since the CRO had no problems with extrapolations (Clast >LLOQ in all pro­files; max. residual AUC 6.2%) probably he/she/it wanted to avoid trouble in reporting a ‘long’ t½
Perhaps a case of anticipatory obedience (vorauseilender Gehorsam). No, the CRO was not in Germany).

❝ Probably the wise guy has already been promoted as we speak?


Oh goodness!

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
d_labes
★★★

Berlin, Germany,
2011-07-28 10:44
(4678 d 04:42 ago)

@ Helmut
Posting: # 7230
Views: 9,604
 

 Puzzle: Estimation of half life

Dear Helmut!

❝ What do you think?


The more interesting question is: What do regulators think?
I never have seen deficiency questions concerning the fit of the terminal phase of concentration time courses in my ~30 years career. Even if the 'fit' was done with only 2 points :-P.

Although I find the fit a little bit strange, I have the opinion it doesn't matter much in the case shown.
  • Half live is only a secondary/additional metric analysed by summary statistics only.
    "... The mean half life was found for Test as 5.018 hours and for Reference as 5.021 hours with a standard deviation of ..." :blahblah:
  • Extrapolations to infinity for AUC will obviously affected only to a minor extent. And this is the main reason to calculate the half life within a BE study.
OK, OK not so very scientific, if one has that pretension.

BTW: What is Your eyeball solution for this curve?

Regards,

Detlew
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2011-07-28 13:38
(4678 d 01:47 ago)

@ d_labes
Posting: # 7232
Views: 9,681
 

 Puzzle: Estimation of half life

Dear D. Labes!

❝ The more interesting question is: What do regulators think?


According to our Master the average time for assessing a BE report is one hour – therefore, nothing.

❝ I never have seen deficiency questions […]



Me not either. :-D

❝ Although I find the fit a little bit strange, […]



Right. Actually the sponsor wants to perform another study and felt that they had too many data points in the previous one (little bit tricky, three drug combo). First I asked for spaghetti plots. Nil. Next for raw data. Received an M$-DOC (actually part of the Appendix to the study report) con­tain­ing SAS-output like the following (one page per profile):
                                                               Plasma
         Age/                               Actual  Blood   concentration
 Site/   Sex/                        Sample  time    vol.      (ng/mL)
Subject  Race   Per. Treat. Sched.   taken  (hh:mm)  (mL)   XXXX    YYYY  Comment
-------------------------------------------------------------------------------------
AA/BBBB 22/M/Ca  1   Tablet pre      Yes    07:37   10.0      BLQ    BLQ
[…]
                            16 h pd  Yes    23:54   10.0   121.33  12.10  late sampl.
[…]

Really handy to be imported into PK software. Next: An Excel-file converted from the SAS-binary. Finally something to work with.

❝ BTW: What is Your eyeball solution for this curve?


Last three data-points. t½ 8.00 h, R² 0.751 (R with 0.867 below the CRO’s cut-off 0.95). Before I start to fit individuals I use to have a look at spaghetti plots. Subjects with higher concentrations consistently showed such a slow phase (t½ >6 h). The quality of fits was sometimes pretty good (R²adj 0.998) and mainly >0.9… Taking the overall-PK into account, I would say that the bad fit of this subject is still justified (haha – I don’t have a R-cutoff…).

I this study residual AUCs were not an issue, but I’m concerned about careless fitting – which might tweak residual AUCs to <20%.

  The only people who see the whole picture
are the ones who step out of the frame.
Salman Rushdie


Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
ElMaestro
★★★

Denmark,
2011-07-28 12:24
(4678 d 03:02 ago)

@ Helmut
Posting: # 7231
Views: 9,435
 

 Puzzle: Estimation of half life

Hello again HS,

For my own clarification:
In the description above do you mean to say that the CROs internal requirement is an r squared of at least 0.95 but that data like this are reported even though -in this case at least- the r sq is less than 0.95?

Pass or fail!
ElMaestro
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2011-07-28 13:53
(4678 d 01:33 ago)

@ ElMaestro
Posting: # 7233
Views: 9,464
 

 Puzzle: Estimation of half life

Hi ElMaestro!

❝ In the description above do you mean to say that the CROs internal requirement is an r squared of at least 0.95…


Not R² – R!

❝ … but that data like this are reported even though -in this case at least- the r sq is less than 0.95?


Well, their fit’s R was 0.9610 (R² 0.9235, R²adj 0.9172). They didn’t report any of those at all.
For my fit see above.

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
ElMaestro
★★★

Denmark,
2011-07-28 18:38
(4677 d 20:48 ago)

@ Helmut
Posting: # 7235
Views: 9,386
 

 100% QC?

Hi HS,

I would do an audit.
Clearly some practice is bad here but it has to be said that there might not be a QA issue involved. Does the CRO have 100% QC, and if yes, how did the QC people get around this one?

Pass or fail!
ElMaestro
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2011-07-28 19:03
(4677 d 20:23 ago)

@ ElMaestro
Posting: # 7236
Views: 9,351
 

 No idea

Hi ElMaestro!

❝ I would do an audit.


Me too.

❝ Clearly some practice is bad here but it has to be said that there might not be a QA issue involved. Does the CRO have 100% QC, and if yes, how did the QC people get around this one?


No idea (as I said: big CRO).
Actually I got this issue like the virgin a child. :-D I was asked by the sponsor for support in design­ing another study (see above). I wanted to run a PopPK model and play around with different sampling schedules – and, hocus-pocus! – the rabbit jumped out of the hat.*

I forwarded a lot of stuff to the (very, very big) sponsor and they are definitely not amused.


  • Rabbit = λz, hat = report.

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
UA Flag
Activity
 Admin contact
23,029 posts in 4,834 threads, 1,641 registered users;
40 visitors (0 registered, 40 guests [including 3 identified bots]).
Forum time: 15:26 CEST (Europe/Vienna)

The most erroneous stories are those we think we know best–
and therefore never scrutinize or question.    Stephen Jay Gould

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5