Ravuri sindhura
☆    

India,
2019-09-18 21:16
(1652 d 16:48 ago)

Posting: # 20614
Views: 6,724
 

 Crossover: samples of p1 and p2 in different batches? [Bioanalytics]

As per FDA Bioanalytical method validation Guidance for Industry "All study samples from subject should be analysed in a single run, especially for studies designed with repeated measures from individual subjects(eg; crossover or sequential design required for BE studies)".
Can the study samples of same subject in more than single run for cross over design can be analysed i'e analysis of P1 samples and P2 samples can be carried out separately in two different runs? If yes, how it need to be justify to regulatory?


Edit: Category and subject line changed; see also this post #1, #2. Please follow the Forum’s Policy[Helmut]
dshah
★★  

India/United Kingdom,
2019-09-19 11:16
(1652 d 02:48 ago)

@ Ravuri sindhura
Posting: # 20616
Views: 5,854
 

 Crossover: samples of p1 and p2 in different batches?

No. As per the guideline- all the study samples from subjects should be analyzed in single run. If we choose different run, then there is high possibility of getting slight different calibration curve equation and thus finally concentration for the unknown samples. Thus it could have direct impact on BE.


Edit: Please follow the Forum’s Policy[Helmut]
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2019-09-19 17:33
(1651 d 20:31 ago)

@ dshah
Posting: # 20619
Views: 5,829
 

 Leads to a (pseudo-) period effect

Dear Ravuri and dshah,

I’m the last person to stick to guidelines (if scientifically doubtful) but the EMA’s, FDA’, and ICH’s are pretty good ones.

❝ Can the study samples of same subject in more than single run for cross over design can be analysed i'e analysis of P1 samples and P2 samples can be carried out separately in two different runs? If yes, how it need to be justify to regulatory?


As always the guidance states “should be” not “have to be”. OK, why do you think it is necessary to analyze periods in different batches? Long washout and problems with stability? More information please.

❝ If we choose different run, then there is high possibility of getting slight different calibration curve equation and thus finally concentration for the unknown samples.


Correct so far.

❝ Thus it could have direct impact on BE.


Here you err, IMHO. Don’t get me wrong, I would analyze all samples in the same batch if ever possible as well (even staggered sampling times – not one period after the other). However, if Ravuri would have different responses in his approach, they would show up in the ANOVA as a period effect – which is automatically corrected for and thus not relevant in assessing the treatment effect.

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
Ohlbe
★★★

France,
2019-09-19 17:50
(1651 d 20:14 ago)

@ Helmut
Posting: # 20620
Views: 5,770
 

 Leads to a (pseudo-) period effect ?

Dear Helmut,

❝ However, if Ravuri would have different responses in his approach, they would show up in the ANOVA as a period effect – which is automatically corrected for and thus not relevant in assessing the treatment effect.


Would they really ? I would say yes if he had systematically a negative bias for period 1 and a positive bias for period 2, or vice-versa. As there will not be a single run for Period 1 samples and a single run for period 2 samples, that will not necessarily be the case: some subjects will get artificially a positive bias in P1, others in P2.

Regards
Ohlbe
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2019-09-19 17:59
(1651 d 20:05 ago)

@ Ohlbe
Posting: # 20621
Views: 5,762
 

 Leads to a (pseudo-) period effect ?

Dear Ohlbe,

❝ Would they really ? I would say yes if he had systematically a negative bias for period 1 and a positive bias for period 2, or vice-versa. As there will not be a single run for Period 1 samples and a single run for period 2 samples, that will not necessarily be the case: some subjects will get artificially a positive bias in P1, others in P2.


Now you confused me (even reading Ravuri’s OP again did not help). If periods are analyzed in single, separate batches it should not matter in a crossover due to the randomization. Prerequisite: Always keep the order of the batches for all subjects p1 → p2 → …

What would not work:
  • A paired design.
  • Periods split into different batches.

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
nobody
nothing

2019-09-19 18:21
(1651 d 19:43 ago)

@ Helmut
Posting: # 20623
Views: 5,784
 

 Leads to a (pseudo-) period effect ?

Dear all

I think you should define which effects (random, i.e. variability by chance or systematic, i.e. bias due to e.g. degrading analytical performance, stock solution etc.) you assume and which of these you can rule out by looking at results for QCs, comparing calibration functions etc. That is what all the quality control for your method is done for.

Without that defined it's hard to jump to conclusions regarding the concentrations you determine, let alone the pk parameters you calculate from these.

If the day-to-day performance of your analytical method is that bad, I wouldn't do any volunteer samples, but go back to the lab and improve your method. No doubt, the least variability/bias will be introduced by the setup Helmut lined out. But I don't think that a good analytical method and a good analytical chemist can't handle the exception from this rule (if necessary and justified)...

Kindest regards, nobody
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2019-09-19 19:22
(1651 d 18:41 ago)

@ nobody
Posting: # 20625
Views: 5,803
 

 Almost entirely OT

Hi nobody,

❝ Without that defined it's hard to jump to conclusions regarding the concentrations you determine, let alone the pk parameters you calculate from these.


Plus: Mickey-Mouse-Pharmacokineticists (© Nick Holford) at work who trust in the “Best Fit” Al Gore Rhythm of Phoenix WinNonlin. A goody I received today (usual half-life 13 hours).

[image]


Was a 4-period full replicate. By chance the other half-lives of those subjects were better fitted and in the range of other subjects.

❝ […] But I don't think that a good analytical method and a good analytical chemist can't handle the exception from this rule (if necessary and justified)...


Cannot agree more. “Good analytical chemists” are an endangered species in those days of guidances, SOPs, and follow-the-book-believers.

To propose that poor design can be corrected by subtle analysis techniques
is contrary to good scientific thinking.
    Stuart J. Pocock

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
nobody
nothing

2019-09-19 19:46
(1651 d 18:18 ago)

@ Helmut
Posting: # 20626
Views: 5,747
 

 Almost entirely OT

...defining terminal phase for determination of terminal half-life is hard, you know. Soooo many things to think of. Let it simply do the software and we're done. No questions from auditors, you know.

If the study fails: Bad luck.

Kindest regards, nobody
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2019-09-19 20:08
(1651 d 17:56 ago)

@ nobody
Posting: # 20627
Views: 5,716
 

 Completely OT

Hi nobody,

❝ […] Soooo many things to think of.


When I saw things like that, I’ve tried a lot to set my brain into idle mode. Hard-core meditation, chanting, C2H5OH, extended walks, SM, a power nap. Schützomycin helped though not sufficiently long enough. Increased dose gave me a blurred vision and visions.

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
dshah
★★  

India/United Kingdom,
2019-09-20 09:52
(1651 d 04:12 ago)

@ Helmut
Posting: # 20629
Views: 5,685
 

 Leads to a (pseudo-) period effect ?

Dear All!
Thanks for the discussion!

❝ Now you confused me (even reading Ravuri’s OP again did not help). If periods are analyzed in single, separate batches it should not matter in a crossover due to the randomization. Prerequisite: Always keep the order of the batches for all subjects p1 → p2 → …


❝ What would not work:

  • A paired design.

  • Periods split into different batches.

I believe that ultimately separate batches in a run is not an issue. Over here the calibration curve equation used will be the same. But when we are running P1 and P2 in different run (not batches), then definitely it could have direct impact on BE Out come.

Regards,
D Shah
nobody
nothing

2019-09-20 10:00
(1651 d 04:04 ago)

@ dshah
Posting: # 20631
Views: 5,710
 

 Leads to a (pseudo-) period effect ?

❝ I believe that ultimately separate batches in a run is not an issue. Over here the calibration curve equation used will be the same. But when we are running P1 and P2 in different run (not batches), then definitely it could have direct impact on BE Out come.


Which (non-random) influence of "different runs" do you expect, that might influence the BE outcome?

Kindest regards, nobody
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2019-09-20 11:42
(1651 d 02:22 ago)

@ dshah
Posting: # 20632
Views: 5,697
 

 Run ≠ batch?

Hi D Shah,

can you please elaborate on the difference between run and batch? I thought that they are synonyms.

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
Ohlbe
★★★

France,
2019-09-20 12:37
(1651 d 01:27 ago)

@ Helmut
Posting: # 20634
Views: 5,665
 

 Run ≠ batch

Dear Helmut,

❝ can you please elaborate on the difference between run and batch? I thought that they are synonyms.


An analytical run can include samples extracted in several batches due to limitations in equipment capacity. Not a problem here IMHO.

Regards
Ohlbe
dshah
★★  

India/United Kingdom,
2019-09-23 09:43
(1648 d 04:21 ago)

@ Helmut
Posting: # 20637
Views: 5,547
 

 Run ≠ batch?

❝ can you please elaborate on the difference between run and batch? I thought that they are synonyms.


Hi All!

I would try to explain to the best of my knowledge.

As per USFDA BMV guideline for Industry in May'2018 on page 16-
" If the bioanalytical method necessitates separation of the overall analytical run into distinct processing batches (e.g., groups of samples processed at distinctly different times or by different analysts), each distinct batch should process duplicate QCs at all levels (e.g., low, middle, high) along with the study samples. Examples might include when the number of samples exceeds the capacity of a 96-well plate or when a solid phase extraction manifold cannot accommodate all samples. See Table 1 for what constitutes an acceptable run based on QC acceptance criteria. A distinct batch or batches in an analytical run may be rejected when it fails to meet QC acceptance criteria, but the remaining batches may pass provided that the analytical run meets the overall QC acceptance criteria."
Acceptance criteria for In-Study Analysis Requirements includes "If the analytical runs consist of distinct processing batches, the QC acceptance criteria should be applied for the whole run and for each distinct batch within the runs."


Thus- in a run- a single batch can pass the acceptance criteria and other may fail too.

Coming to my original point of calibration curve equation-
Say for e.g. the calibration curve equation for a single run (having two batches) can be: f(x)=0.00948370*x-0.00222207 and one run fails to meet the acceptance criteria.
This will require a repeat analysis, and the Calibration curve equation will be not the same for most of the time.
Thus- concentration of unknown samples would be different than the original run.
Over here- @ravuri is asking to run the period in to separate runs- which will definitely have different equation--which may have different concentration than the original concentration -- which may have ultimately impact on final BE statistics.

Regards,
D Shah
nobody
nothing

2019-09-23 10:36
(1648 d 03:28 ago)

@ dshah
Posting: # 20638
Views: 5,508
 

 Run ≠ batch?

A slightly different calibration function not necessarily leads to "different" (from what, btw. We simply don't know the correct concentration of any given analyte in a volunteer sample...) concentrations calculated, as the signal for the volunteer samples might change accordingly.

Again: Which systematic, non-random effects between batches do you assume here?

Kindest regards, nobody
Ravuri sindhura
☆    

India,
2019-09-23 15:04
(1647 d 23:00 ago)

@ Helmut
Posting: # 20640
Views: 5,516
 

 Leads to a (pseudo-) period effect

❝ […] why do you think it is necessary to analyze periods in different batches? Long washout and problems with stability? More information please.


Dear Helmut,

Because of time constraints we wanted to analyze periods in different batches.

Regards
Ravuri


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 #5[Helmut]
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,639 registered users;
73 visitors (0 registered, 73 guests [including 5 identified bots]).
Forum time: 13:04 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