mittyri
★★  

Russia,
2013-10-23 11:53
(3809 d 09:07 ago)

Posting: # 11730
Views: 6,985
 

 Using sampling “windows” for PK blood samples [Design Issues]

Dear all,

I would like to discuss the post by Nathan Teuscher
you can find it here: http://learnpkpd.com/2013/10/15/using-sampling-windows-for-pk-blood-samples/
What do you think about this consideration?
does sampling “windows” really not affect the PK parameters?

Kind regards,
Mittyri
jag009
★★★

NJ,
2013-10-23 18:50
(3809 d 02:10 ago)

@ mittyri
Posting: # 11739
Views: 5,993
 

 Using sampling “windows” for PK blood samples

Hi,

❝ does sampling “windows” really not affect the PK parameters?


Are we talking about exploratory PK or bioequivalence? I agree for AUC, but for Cmax as it depends on what your purpose is.

John
mittyri
★★  

Russia,
2013-10-24 17:35
(3808 d 03:25 ago)

@ jag009
Posting: # 11750
Views: 5,885
 

 Using sampling “windows” for PK blood samples

Hi, John,

❝ Are we talking about exploratory PK or bioequivalence? I agree for AUC, but for Cmax as it depends on what your purpose is.


Nathan Teuscher does not explain a type of trial. So I'm interested in opinion of the colleagues who are working with BEQ protocols.

Kind regards,
Mittyri
jag009
★★★

NJ,
2013-10-25 07:12
(3807 d 13:48 ago)

@ mittyri
Posting: # 11763
Views: 5,909
 

 Using sampling “windows” for PK blood samples

Hi Mittyri,

❝ Nathan Teuscher does not explain a type of trial. So I'm interested in opinion of the colleagues who are working with BEQ protocols.


His conclusion was based on AUC being similar between the two sampling time schemes which I agree. For BE studies we need to determine both AUC and Cmax equivalence between test and reference. Therefore the PK time point selection is important (if not critical).

John
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-10-28 16:48
(3804 d 03:12 ago)

@ jag009
Posting: # 11798
Views: 6,057
 

 Disagree with Nathan

Hi John & Mittyri,

❝ His conclusion was based on AUC being similar between the two sampling time schemes which I agree.


Well, these are simulated data () of a fast bolus – though in the real world we never would see the highest concentration at t=0*. Note that Nathan calculated the AUC by the lin/log-trapezoidal rule (which mimics the exponentional decrease; see also another blog-post). Only with this algo there is no impact of time deviations on AUCs. If someone still uses the linear trapezoidal (why the heck?) the AUC8 of subject 2001 () would be 0.55% larger than the one of subject 1001 ().

[image]


❝ For BE studies we need to determine both AUC and Cmax equivalence between test and reference. Therefore the PK time point selection is important (if not critical).


Agree. ‘Catching’ Cmax is very important in BE. The same is true if truncated/partial AUCs come into play. In steady state studies EMA requires sampling time deviations of ≤ ±5’ for the pre-dose sample and ≤ ±10’ for the sample at the end of the dosing interval.

Overall I disagree with Nathan’s conclusions, that

[…] setting sampling “windows” does not affect the PK parameters and only leads to extra work for the clinical and data management teams as they verify the time of the blood sample versus the acceptable window, record any deviations, and then supply justification for the deviations.
[…] with such an intense focus on getting a sample at a specific time, often clinical staff record incorrect blood draw times to stay “in window” rather than report the actual blood draw times.


I would still set time allowance windows and require a justification if sampling takes place outside. For Nathan’s second remark I would suggest training the clinical staff.

See also John’s weird experiences (#1, #2).


  • Injecting into the subjects’ one arm’s vein and sampling from the other at the same time? Even for a quick bolus turnaround time of the circulation is up to two minutes. Therefore, any sample taken earlier than two minutes does not make any sense. You might end up with strange low concentrations (due to the intermixing phase) which even for a one-compartment model don’t fit well into the rest of the profile.

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,
2013-10-28 21:36
(3803 d 22:24 ago)

@ Helmut
Posting: # 11799
Views: 5,906
 

 Disagree with Nathan

Hi all,

❝ Well, these are simulated data () of a fast bolus – though in the real world we never would see the highest concentration at t=0*. Note that Nathan calculated the AUC by the lin/log-trapezoidal rule...


Agree with Helmut. Depending on the product the plasma concentrations may fluctuate (bounce up and down) over the time course of PK sampling window such that AUC value may vary between different time sampling schemes.

Take a look at a nifedipine XL blood level profile for example. If I remember correctly, it was bouncy.

I think from a NDA pt of view then it might not be super critical if one just wants to delineate the PK. In the world of BE (ANDA or 505(b)2) then yes it is important.

Partial AUC is another beast. Don't even get me started on that recent fiasco.

John
teuscher
☆    

US,
2013-11-15 04:55
(3786 d 15:05 ago)

(edited by Ohlbe on 2013-11-15 10:16)
@ jag009
Posting: # 11888
Views: 5,562
 

 Using sampling “windows” for PK blood samples

I appreciate the comments and thoughts on my blog post. Perhaps my comments were taken out of context. The context of my comments was whether or not we need to define a "window" during which we must draw a blood sample. And if the sample is outside of that window we declare it a protocol deviation.

As this group stated time point selection is very important to accurately capture the characteristics of a pharmacokinetic profile. We often use optimization algorithms to choose the best time points.

However, once the nominal time points are chosen, does it really matter if a sample is drawn 5-15 minutes early or late? Unless that deviation overlaps a subsequent nominal sampling time, I still posit that it does not matter. As an example, if your sampling times are 0.5, 1, and 1.5 hours. The sample at 1 hour can be taken at 0.75, 1, or 1.25 hrs and still give reasonable results.

Interested in your thoughts ...

Nathan Teuscher


Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post! [Ohlbe]
Ohlbe
★★★

France,
2013-11-15 11:32
(3786 d 08:27 ago)

@ teuscher
Posting: # 11889
Views: 5,770
 

 Using sampling “windows” for PK blood samples

Dear Nathan,

I agree a number of things you wrote in your blog post, but not all. Most comments on this forum were in connection with bioequivalence trials. There, I agree with statements such as "Because PK is the primary endpoint, the blood draws for PK samples should take priority over all other activities and should occur on the scheduled time point before any other procedure". In a BE trial I couldn't care less if "when PK blood sampling is given priority, it can negatively affect other assessments such as vital signs, ECG measurements, and behavioral measures", unless there are PD parameters to be measured too. I prefer to see the blood sample collected at the correct time and vitals delayed by a few minutes, than the reverse.

❝ I still posit that it does not matter. As an example, if your sampling times are 0.5, 1, and 1.5 hours. The sample at 1 hour can be taken at 0.75, 1, or 1.25 hrs and still give reasonable results.


Sounds reasonable in a BA trial with PK modelling. But the PK analysis in BE trials is Mickey Mouse PK: model independent, and most people use the linear trapezoidal rule to calculate AUC (which makes Helmut regularly scream). The difference in AUC usually remains minimal, but you may miss Cmax.

Lastly, deviations in the sampling time in the last collected sample will affect AUC0-t, which is one of the primary BE parameters. Unless of course you use an estimated Clast at the nominal sampling point.

I remain of the opinion that blood samples should be collected as close as possible to the nominal sampling time. Whether sampling time deviations should be recorded as protocol deviations, and the amount of associated paperwork, remains another discussion. Some people indeed overdo it.

Regards
Ohlbe
teuscher
☆    

US,
2013-11-15 16:16
(3786 d 03:44 ago)

@ Ohlbe
Posting: # 11891
Views: 5,604
 

 Using sampling “windows” for PK blood samples

Ohlbe,

Thank you for your thoughtful response. I appreciate that there are differences in PK analysis to support specific projects such as bioequivalence studies. I have to say that I share in Helmut's frustration about using linear trapezoidal methods; however, I also understand that regulators want to standardize analyses. (Although they could standardize on linear up - log down, which would be more accurate than the current standard!)

Warm regards,
Nathan
Ohlbe
★★★

France,
2013-11-15 16:55
(3786 d 03:04 ago)

@ teuscher
Posting: # 11893
Views: 5,614
 

 Using sampling “windows” for PK blood samples

Dear Nathan,

❝ [...] I also understand that regulators want to standardize analyses. (Although they could standardize on linear up - log down, which would be more accurate than the current standard!)


No guideline says that the linear trapezoidal rule should be used for BE. The method should be stated in the protocol and in the report, that's all. Lin up – log down is perfectly acceptable. Sponsors and CROs are just keeping that habit for whatever reason.

Regards
Ohlbe
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-11-17 15:55
(3784 d 04:05 ago)

@ Ohlbe
Posting: # 11902
Views: 7,336
 

 Lin/log trapezoidal

Hi Ohlbe & Nathan,

❝ No guideline says that the linear trapezoidal rule should be used for BE. The method should be stated in the protocol and in the report, that's all.


Exactly. The WHO stated in 2006:

The method of calculating AUC-values should be specified. In general AUC should be calculated using the linear/log trapezoidal integration method.

(my emphasis)

❝ Sponsors and CROs are just keeping that habit for whatever reason.


Why the wacky linear trapezoidal rule is still of such widespread use is a complete mystery to me. Its bad performance was shown decades ago. I’m happy that in bear the lin-up/log-down is the default since v2.5.6 (June 27, 2013) and Pharsight will follow in the next release of Phoenix/WinNonlin (expected in Q I / 2014).

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
22,957 posts in 4,819 threads, 1,636 registered users;
80 visitors (0 registered, 80 guests [including 12 identified bots]).
Forum time: 20:00 CET (Europe/Vienna)

Nothing shows a lack of mathematical education more
than an overly precise calculation.    Carl Friedrich Gauß

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