jag009
★★★

NJ,
2013-09-09 23:12
(4111 d 13:38 ago)

Posting: # 11464
Views: 11,130
 

 Boneheads! [Study Per­for­mance]

Hi guys,

If a blood draw has a time deviation which pushes it to overlap with the next blood draw time point, how do you deal with PK if you need to calculate partial AUC0-4?

3.5 hr blood draw has time deviation of +41 mins.
4 hr blood draw has time deviation of +2 mins.

$&#%! Don't ask how and why.

Thanks
John
Ohlbe
★★★

France,
2013-09-09 23:26
(4111 d 13:24 ago)

@ jag009
Posting: # 11465
Views: 9,487
 

 WTF ?

Hi John,

❝ 3.5 hr blood draw has time deviation of +41 mins.

❝ 4 hr blood draw has time deviation of +2 mins.


How did they manage to collect the 3.5 hr sample after the 4 hr sample :surprised: ?

❝ how do you deal with PK if you need to calculate partial AUC0-4 ?


With PK, no idea. With the CRO, I have a pretty good idea :stop:

Unless the staff were kept busy by a serious adverse event and had more urgent things to do, I can't see how they managed to f... up that much. Even with a blocked canula they should be able to collect the sample via a fresh venipuncture. Or did they have no spare phlebotomist at hand ?

OK, I know it doesn't help, sorry...

Regards
Ohlbe
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-09-09 23:55
(4111 d 12:56 ago)

@ Ohlbe
Posting: # 11466
Views: 9,538
 

 Snafued!

Hi Ohlbe & John,

Unbelievable snafu:
3:30 ⇒ 4:11
4:00 ⇒ 4:02

I hope they had no barcode-/CDMS in place. Such data might foul up the entire data base. At least in our validation we did not challenge the system with a later sample harvested earlier than the ‘previous’ one. Wacky.

❝ With the CRO, I have a pretty good idea :stop:


:ok: What was going on in their wetware? Lights on, nobody at home?

❝ ❝ how do you deal with PK if you need to calculate partial AUC0-4 ?


In Phoenix no need to worry. Like in this post.

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
jag009
★★★

NJ,
2013-09-10 18:11
(4110 d 18:40 ago)

@ Helmut
Posting: # 11469
Views: 9,454
 

 Snafued! WTF! &%$@!

Hi Ohlbe & Helmut,

Not a joke. The reply from the project manager:
Subject 143 – the 3.5 hour sample should not have been collected due to the delay in resticks of 41 minutes. Unfortunately, both were collected, shipped. I have reached out to PK group for input regarding this.

Restick = Difficulty in drawing blood. WTF for 41 mins. How many holes did they poke? I don't know. The subject should've been dropped out since it is inhuman to poke so many holes (it hurts!).

P.S. Helmut,
From our previous discussion, you wrote:

❝ BTW, if you ask Phoenix/WinNonlin for partial AUCs (and sampling wasn’t performed exactly at the cut-off time point) an interpolation between two adjacent time points (before/after) is used; linear if C2 ≥ C1 and log/linear if C2 < C1. Makes sense, IMHO.


I don't think you can make interpolation to obtain the concentration at the desire time right (example: interpolate 4 hr concentration from 3.5 hr and 4.25 hr concentration)? I thought you have to use the time point as is?

Thanks
John
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-09-11 23:01
(4109 d 13:49 ago)

@ jag009
Posting: # 11482
Views: 9,736
 

 pAUCs

Hi John,

❝ ❝ BTW, if you ask Phoenix/WinNonlin for partial AUCs (and sampling wasn’t per­formed exactly at the cut-off time point) an interpolation between two adjacent time points (before/after) is used; linear if C2 ≥ C1 and log/linear if C2 < C1. Makes sense, IMHO.


❝ I don't think you can make interpolation to obtain the concentration at the desire time right


Why not? As you know we work on the same drugs (pAUCs standard). Never had a problem. Of course I describe the procedure in the protocol and even give an example plot similar to that one …

[image]


… showing how pAUCs, t75%, and HVD are calculated. Samples denoted with an orange circle were not taken at scheduled time points – they were either interpolated (at t=0 and t=4) or extrapolated (t=24).

❝ (example: interpolate 4 hr concentration from 3.5 hr and 4.25 hr concentration)?


I thought that you don’t have the 3.5 sample (delayed till 4:11)? But you have the one at 4:02. Interpolate between this one and the one before. If you feel that you are reaching too far “to the left”, drop the AUCs and keep the Cmax.

❝ I thought you have to use the time point as is?


Who told you that? :-D
Did you never have small time deviations close to the truncation time point and asked Phoenix/WinNonlin for partial AUCs? If yes, I’m surprised. If no, guess what was going on behind the veil (or look it up in the User’s Guide).

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
jag009
★★★

NJ,
2013-09-12 22:48
(4108 d 14:03 ago)

@ Helmut
Posting: # 11489
Views: 9,236
 

 pAUCs - the SEQUEL

Hi Helmut,

❝ Why not? As you know we work on the same drugs (pAUCs standard). Never had a problem. Of course I describe the procedure in the protocol and even give an example plot similar to that one …


That's the beauty of it. It's not described on the protocol and CRO's SOP doesn't have this. Of course I can do a supplmental analysis myself...

❝ I thought that you don’t have the 3.5 sample (delayed till 4:11)? But you have the one at 4:02. Interpolate between this one and the one before. If you feel that you are reaching to far “to the left”, drop the AUCs and keep the Cmax.


Oh something NEW has popped up (I was looking at 50% of the subject, now I have 100%). They have the following:

3.5 hr with a time deviation of 45 mins; Reason - Restick
4 hr with a time deviation of 15 mins; Reason - Restick
The concentration values are different.

@$^% !!!!! How the hell do you tackle this??? Should I take the average of the two concentrations and use only the 4 hr time point?

John
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-09-13 02:55
(4108 d 09:56 ago)

@ jag009
Posting: # 11490
Views: 9,312
 

 A Nightmare on Elm Street

Hi John,

❝ It's not described on the protocol and CRO's SOP doesn't have this.


So in the protocol they simply copypasted from the guidance “AUC0-4, AUC4-t, AUC0-∞, and Cmax, where AUC0-4 is the area under the plasma-concentration vs. time curve from 0 to 4 hours, AUC4-t is area under the curve from 4 hours to the last measurable time point” without stating the algo? Brilliant. Which software are they using?

❝ They have the following:

❝ 3.5 hr with a time deviation of 45 mins; Reason - Restick

❝ 4 hr with a time deviation of 15 mins; Reason - Restick


Splendid. As you posted before I thought I have seen everything – but this?!

❝ The concentration values are different.


Sure. AP hits.

❝ How the hell do you tackle this??? Should I take the average of the two concentrations and use only the 4 hr time point?


The latter (OK, the second one at 4:15). Set the first one to “missing”, “not reportable“, or “oops”.

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
jag009
★★★

NJ,
2013-09-13 17:26
(4107 d 19:25 ago)

@ Helmut
Posting: # 11491
Views: 9,263
 

 A Nightmare on Elm Street

❝ So in the protocol they simply copypasted from the guidance “AUC0-4, AUC4-t, AUC0-∞, and Cmax, where AUC0-4 is the area under the plasma-concentration vs. time curve from 0 to 4 hours, AUC4-t is area under the curve from 4 hours to the last measurable time point” without stating the algo? Brilliant. Which software are they using?


Exactly... Software wise I think they use both WinNonlin and SAS. I think they can do the interpolation by doing a protocol amendment? I have seen this before and FDA accepted it (not for partial AUC but for other things).
A lot of companies do this because they don't know whether their Kel elimination coding in SAS is "safe" so they need something that is officially validated.

❝ ❝ They have the following:

❝ ❝ 3.5 hr with a time deviation of 45 mins; Reason - Restick

❝ ❝ 4 hr with a time deviation of 15 mins; Reason - Restick


❝ Splendid. As you posted before I thought I have seen everything – but this?!


YES! I told you I haven't seen EVERYTHING :-(

❝ The latter (OK, the second one at 4:15). Set the first one to “missing”, “not reportable“, or “oops”.


Why the latter and not the first? If they have both samples and the numbers are different? I think they should take the average?

Thanks
John
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-09-13 18:00
(4107 d 18:50 ago)

@ jag009
Posting: # 11493
Views: 9,245
 

 A Nightmare on Elm Street

Hi John,

❝ Exactly... Software wise I think they use both WinNonlin and SAS. I think they can do the interpolation by doing a protocol amendment? I have seen this before and FDA accepted it (not for partial AUC but for other things).

❝ A lot of companies do this because they don't know whether their Kel elimination coding in SAS is "safe" so they need something that is officially validated.


Again: If you ask Phoenix/WinNonlin for partial AUCs they will be calculated up to/from the requested truncation time point – irrespective of time deviations (like in the plot above). I don’t think that you need an amendment.

❝ ❝ The latter (OK, the second one at 4:15). Set the first one to “missing”, “not reportable“, or “oops”.


❝ Why the latter and not the first?


Gut feeling. I prefer a 15 min deviation over 45. I’m asking myself how they got two samples at the same time point. Two nurses sticking needles in the left and right arms of the subject competing on who finishes first?

❝ If they have both samples and the numbers are different?


Due to analytical variability the values are expected to be different.

❝ I think they should take the average?


Yes and no. Although the average should be more close to the “true value” you might open a can of worms. Some nitpicking assessor might tell you this value carries undue weight (since all other values are single determinations). Duno.

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
jag009
★★★

NJ,
2013-09-13 21:57
(4107 d 14:54 ago)

@ Helmut
Posting: # 11496
Views: 9,174
 

 A Nightmare on Elm Street

Hi Helmut,

❝ ❝ Why the latter and not the first?


❝ Gut feeling. I prefer a 15 min deviation over 45. I’m asking myself how they got two samples at the same time point. Two nurses sticking needles in the left and right arms of the subject competing on who finishes first?


Like Formula 1 GP. Photo finish... I don't know what they were thinking and I don't know why they persistent in collecting the 3.5 hr sample if they were unsuccessful after a few poke on the subject's arm...

John
jag009
★★★

NJ,
2013-09-10 18:13
(4110 d 18:38 ago)

@ Ohlbe
Posting: # 11470
Views: 9,473
 

 WTF ?

Hi Ohlbe,

❝ Unless the staff were kept busy by a serious adverse event and had more urgent things to do, I can't see how they managed to f... up that much. Even with a blocked canula they should be able to collect the sample via a fresh venipuncture. Or did they have no spare phlebotomist at hand ?


❝ OK, I know it doesn't help, sorry...


For 41 mins??? I still want to know how many holes they punched the poor subject with!

John
ElMaestro
★★★

Denmark,
2013-09-10 20:31
(4110 d 16:20 ago)

@ jag009
Posting: # 11472
Views: 9,360
 

 monitor

Hi John,

❝ For 41 mins??? I still want to know how many holes they punched the poor subject with!


Did the monitor have any remarks in the report? Did (s)he visit on the day this happened?

Pass or fail!
ElMaestro
jag009
★★★

NJ,
2013-09-11 00:48
(4110 d 12:03 ago)

@ ElMaestro
Posting: # 11473
Views: 9,347
 

 monitor

❝ ❝ For 41 mins??? I still want to know how many holes they punched the poor subject with!


❝ Did the monitor have any remarks in the report? Did (s)he visit on the day this happened?


Waiting for their "explanation"... Study info to be released in upcoming weeks. So far the explanation is "restick". The clinic thought subjects were not hydrated adequately before dosing in Period 1 so they ask them to drink more the night before period 2. Problem solved? Sort of but nope, still had long time deivations...

In my opinion the subject should've been dropped. Its not ethical to keep on poking for blood for so long... It occured at 3.5-4 hr so the P.I. should've made a decision (A DOCTOR!)

John
UA Flag
Activity
 Admin contact
23,336 posts in 4,902 threads, 1,666 registered users;
45 visitors (1 registered, 44 guests [including 5 identified bots]).
Forum time: 11:51 CET (Europe/Vienna)

Biostatistician. One who has neither the intellect for mathematics
nor the commitment for medicine but likes to dabble in both.    Stephen Senn

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