ratnakar1811
★    

India,
2013-10-07 10:07
(3825 d 07:10 ago)

Posting: # 11613
Views: 11,079
 

 Actual time for PK analysis [NCA / SHAM]

Dear All,

Recently I was in terrible situation, where one of my FDA submission study got failed in AUC 0to72, the drug under investigation was long half life drug and therefore blood samples were only collected upto 72 hrs post dose. During PK analysis actual time of blood sample collection was used, and CI limits were for AUC 0to72 were (79.34 -114.22). Sponsor subsequently done PK analysis at his end and came up with CI limits of (89.12 -108.11), after investigation it was realized that sponsor has considered the scheduled time for last ambulatory visit with the explanation that our parameter was AUC 0to72 and if we consider actual time (for some of the subjects 72 hr ambulatory sample was collected up to 74 hrs) the PK parameter becomes AUC 0to74, therefore for truncated study only schedule time should be considered.

Further sponsor provided explanation that since there is a window period of 2 hrs for ambulatory visit, we should consider only schedule time for ambulatory sample (i.e. 72 hr; even if the sample is collected by 74 hrs) and for all the samples which are collected after window period of 2 hrs actual time should be considered.

Expert views are highly appreciated, on the consideration of actual time for PK analysis.

Regards,

Ratnakar


Edit: Category changed. [Helmut]
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-10-07 15:36
(3825 d 01:40 ago)

@ ratnakar1811
Posting: # 11616
Views: 10,054
 

 estimated AUC72

Hi Ratnakar,

I think that your approach is not optimal, and the sponsor’s one is simply wrong. Let’s start with the sponsor’s. If a concentration was measured at 74 hours one cannot simply shift it to 72 hours (the “as-if” approach) ⇒ AUC72 will be biased. Guidelines require the actual time, not the scheduled one.

[image]Your approach is better, but biased as well. Imagine that F=100% and tlast after reference was 72 hours and after test 74 hours. Naturally AUC74 (apples) > AUC72 (oranges). The apples-to-oranges ratio will be >1 and you get a positive bias for F. It is a weak argument that due to randomization this should happen equally often to T and R (~similar number of apples and oranges in the fruits’ basket) and therefore means out. Why not calculate AUC72 for both? In Phoenix/WinNonlin request a partial AUC (Start Time 0, End Time 72). C72 will be lin/log-interpolated between the timepoint preceeding tlast and tlast.
Example: t½,abs 1 hour, t½,el 36 hours, F 90%, lin-up/log-down trapezoidal rule.
  t          R       T
 0.00       0.00    0.00
 0.50      28.33   25.50
 1.00      48.09   43.28
 2.00      71.22   64.10
 3.00      81.89   73.70
 4.00      86.34   77.70
 5.50      87.74   78.97
 7.50      86.00   77.40
10.50      81.63   73.46
14.50      75.64   68.07
20.00      68.04   61.24
27.50      58.89   53.00
38.00      48.11   43.30
52.25      36.57   32.91
72.00      25.00    NA
74.00       NA     21.65
AUC52.25 3142.42 2831.29
AUC72    3744.03   NA
AUC74       NA   3416.11

F at 52.52 hours is 90.10% (bias +0.11%). If one uses AUClast for both formulations (your method) one gets 91.24% (bias +1.38%). The sponsor’s “method” of shifting 21.65 measured at 74 hours forward to 72 hours would give 3362.33 (bias –0.22%). The interpolated concentration of test at 72 hours is 22.50 (note: 22.50/25.00=0.90!) and AUC72 3371.96 – which would give F 90.06% (bias +0.07%). If you don’t have suitable software, the interpolation formula is:

Ĉ = ℯlog(Ci–1)+|ti–ti–1|/(ti+1–ti–1)×log(Ci+1/Ci–1)

Now for the nasty parts. The study failed if evaluated according to protocol. I don’t think that the FDA will accept the sponsor’s “method” – cherry-picking and simply wrong. BTW, I’m surprised to see such a large difference. You can present BE based on AUC72 as a sensitivity analysis and discuss the impact on BE (if the conclusions differ). For the next studies describe what you intend to do in the protocol and stick to it. If you don’t feel comfortable with an estimated AUC72 (i.e., prefer mixed fruits), decrease the time allowance window. As an example EMA requires for studies in steady-state (τ 24 hours) a maximum deviation of 10 minutes.

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
ratnakar1811
★    

India,
2013-10-08 10:20
(3824 d 06:56 ago)

@ Helmut
Posting: # 11621
Views: 10,031
 

 estimated AUC72

Dear Helmut,

Thanks for your prompt reply and explanation with example! I always get prompt solution from forum to all my problems, I really appreciate it.

❝ Guidelines require the actual time, not the scheduled one.


I know it has been mentioned in the EMA guideline but could not locate in any of the FDA guideline.

❝ Why not calculate AUC72 for both? In Phoenix/WinNonlin request a partial AUC (start=0, end=72).


Is the use of partial area, the best solution for the truncated study? If yes, what about the non truncated study having primary PK parameters as Cmax and AUCt, where samples up to 36 hrs were collected, but in 20 % of the subjects last measurable concentration appeared at 18 hrs (in fact in couple of subject it was only 8 hrs) and majority of the subjects got measurable concentration at 24 hrs and again few subjects got concentration at 36 hrs post dose. This situation I have encountered in one of the pMDI study and somewhat similar type of variation in the AUC was observed in the DPI and Nasal Spray studies.

❝ If you don’t have suitable software, the interpolation formula


I do have Phoenix/WinNonlin.

❝ If you don’t feel comfortable with an estimated AUC72 (i.e., prefer mixed fruits), decrease the time allowance window. As an example EMA requires for studies in steady-state (τ 24 hours) a maximum deviation of 10 minutes.


I can very well implement your suggestion of the partial area, provided it is the best approach for truncated study.

Best Regards,

Ratnakar
sam
★    

India,
2013-10-08 15:15
(3824 d 02:02 ago)

@ ratnakar1811
Posting: # 11622
Views: 9,976
 

 estimated AUC72

Dear Ratnakar,

❝ I can very well implement your suggestion of the partial area, provided it is the best approach for truncated study.


The approach of the partial area suggested by Helmut is well accepted by any of the Regulatories body and this is only a correct approach for calculation of AUC72 for the truncated design.
I have experience of n number of studies which has been submitted and got approved by various regulatory body.

Best Regards,
Sam
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-10-08 15:51
(3824 d 01:25 ago)

@ ratnakar1811
Posting: # 11624
Views: 9,993
 

 estimated AUC72

Hi Ratnakar,

❝ ❝ Guidelines require the actual time, not the scheduled one.

❝ I know it has been mentioned in the EMA guideline but could not locate in any of the FDA guideline.


Well, FDA’s guidance is from 2003. Expect assessors to look beyond their own nose.

❝ ❝ Why not calculate AUC72 for both? […]


❝ Is the use of partial area, the best solution for the truncated study?


IMHO, yes. See also the discussion in WHO’s GL, Section 6.11.4. See also this post and follow the linked posts within.

❝ If yes, what about the non truncated study having primary PK parameters as Cmax and AUCt, where samples up to 36 hrs were collected, but in 20 % of the subjects last measurable concentration appeared at 18 hrs (in fact in couple of subject it was only 8 hrs) and majority of the subjects got measurable concentration at 24 hrs and again few subjects got concentration at 36 hrs post dose.


Good question, next question. ;-) Note that “t” is defined as the time point of the last quantifiable concentration. Your primary metric is not AUC36. But it’s true that missing values / <LLOQ will also lead to “apples-and-oranges” (old story). BTW, have you ever seen in a steady-state study Phoenix/WinNonlin reporting AUClast AUCtau? If yes, can you guess why?

❝ I can very well implement your suggestion of the partial area, provided it is the best approach for truncated study.


I’m using it for a couple of years now and like Sam never received a deficieny letter. I always described the method in great detail in the protocol (even a plot similar to that one) and don’t hide it in an SOP.

Don’t forget to give it a nice label. Shines the output.

[image]

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
ratnakar1811
★    

India,
2013-10-10 13:30
(3822 d 03:46 ago)

@ Helmut
Posting: # 11639
Views: 9,860
 

 estimated AUC72

Dear Helmut & Sam,
Thanks a lot!

❝ Good question, next question. ;-) Note that “t” is defined as the time point of the last quantifiable concentration. Your primary metric is not AUC36. But it’s true that missing values / <LLOQ will also lead to “apples-and-oranges” (old story). BTW, have you ever seen in a steady-state study Phoenix/WinNonlin reporting AUClast AUCtau? If yes, can you guess why?


Is it because AUClast considers area up to last quantifiable concentration and AUCtau only considers area between dosing interval?

I am still confused about comparing AUC8hrs of one subject with AUC 36hrs of other subject, what is your suggestion on it?

In fact I would like to add a situation I have faced in couple of studies for pMDI formulation (with and without charcoal treatment), for Salmeterol I got AUC% extrapolation more than 20 % in about 35 % of the population and most of the subjects achieved either zero concentration much before the last concentration or had very slight concentration at last time point, actually a straight line was seen in the elimination phase because of which although samples were collected sufficient time period (up to 18 hrs post dose), AUC% extrapolation was greater than 20%. Following is a profile typically seen in most of the subjects but we got a query from regulator over it for validity of the as per guideline?

[image][image]

Can we have some justification for such issue?


Best Regards,

Ratnakar
jag009
★★★

NJ,
2013-10-13 23:17
(3818 d 17:59 ago)

@ ratnakar1811
Posting: # 11659
Views: 9,600
 

 estimated AUC72

Hi,

❝ In fact I would like to add a situation I have faced in couple of studies for pMDI formulation (with and without charcoal treatment), for Salmeterol I got AUC% extrapolation more than 20 % in about 35 % of the population and most of the subjects achieved either zero concentration much before the last concentration or had very slight concentration at last time point, actually a straight line was seen in the elimination phase because of which although samples were collected sufficient time period (up to 18 hrs post dose), AUC% extrapolation was greater than 20%. Following is a profile typically seen in most of the subjects but we got a query from regulator over it for validity of the as per guideline?


Let me make sure I deciphered your question correctly. 1) Are you talking about BE studies (T vs R)? 2) You said some subjects showed AUClast up to 8 hrs only for both T and R, while some showed AUClast up to a longer collection time? If so, I don't understand your concern. What query would the agency have? I can see the issue if the same subject shows AUC only up to 8 hrs for Test but AUC up to a longer collection time for Reference.

I recently ran a highly variable drug BE study. AUCs were a mess with subjects having
  1. undetectable concentration after 4 hr sampling
  2. undetectable concentrations after 12 hr sampling
  3. all concentrations above LLQ at end of collection time of 48 hrs.
John
ratnakar1811
★    

India,
2013-10-15 08:23
(3817 d 08:54 ago)

@ jag009
Posting: # 11665
Views: 9,577
 

 estimated AUC72

Dear John,

There are two queries from two different studies;

Situation 1: Situation somewhat similar to yours, where samples were collected up to 36 hrs post dose, but for some of the subjects AUCt could be calculated only up to 8 hrs, for some of the subjects it was up to 16 hrs and for some of the subjects up to 36 hrs.

Query 1: What approach we should implement while comparing such a variable AUCt? Is it a correct method to compare AUC8hrs to AUC36 hrs? Or Is there any other statistical method which we should adopt for such situation?

Situation 2: Samples were collected up to 36 hrs post dose and even achieved zero concentration quite before 36 hrs but still AUC% extrapolation was more than 20 % in many cases (e.g. profile is given in my earlier post)

Query 2: Since extrapolation is more than 20% how do we justify such situation?

I hope I am clear on this occasion.

Best Regards,

Ratnakar
Dr_Dan
★★  

Germany,
2013-10-15 14:24
(3817 d 02:53 ago)

@ ratnakar1811
Posting: # 11666
Views: 9,559
 

 estimated AUC72

Dear Ratnakar

❝ Query 1: What approach we should implement while comparing such a variable AUCt? Is it a correct method to compare AUC8hrs to AUC36 hrs? Or Is there any other statistical method which we should adopt for such situation?


As John already pointed out it is the intra-subject comparison that we are talking about when evaluation bioequivalence. Do you compare fast vs. slow metabolizers (i.e. inter-subject comparison)?

❝ Query 2: Since extrapolation is more than 20% how do we justify such situation?


Could you please explain how do you extrapolate AUC0-inf?
This would help
Best Regards,
Dan

Kind regards and have a nice day
Dr_Dan
jag009
★★★

NJ,
2013-10-15 23:53
(3816 d 17:23 ago)

@ ratnakar1811
Posting: # 11667
Views: 9,622
 

 estimated AUC72

Hi,

❝ Situation 1: Situation somewhat similar to yours, where samples were collected up to 36 hrs post dose, but for some of the subjects AUCt could be calculated only up to 8 hrs, for some of the subjects it was up to 16 hrs and for some of the subjects up to 36 hrs.


My opinions (based on what I could gather from your description and the 2nd graph you posted in pre. posting):
  1. You meant you had subjects with AUCt up to 36 hrs for treatment A and AUCt only up to 8 hrs for treatment B (And the AUCt values are very different?)? Differences in t1/2 values? What is the pharm/PK property of the molecule? My crystal ball says the molecule may have auto-induction property such that it "enhances" the metabolism enzyme -> If you do a crossover study then you are cooked because the first period treatment would influence the metabolism capability such that the bioavailability of the 2nd period treatment would get reduced. I faced this issue a while back.

  2. Or you meant some subjects had only AUCt upto 8 hrs for both T and R while some subjects had AUCt upto 36 hrs for both T and R? If so, it's just the variability of the drug. If it's a BE study then what's the issue since we are interested in intra-subject comparison.

One question, how's the solubility and permeability of the drug molecule?

❝ Query 2: Since extrapolation is more than 20% how do we justify such situation?


Like Dr_Dan said "How did you compute AUCinf?"? If I recall correctly it is possible to throw out/conclude AUCinf as "not defined" due to ridiculous half-life value. How different were the t1/2 values for this 20% extrapolation data?

Thanks

John
ratnakar1811
★    

India,
2013-10-17 11:34
(3815 d 05:42 ago)

@ jag009
Posting: # 11670
Views: 9,594
 

 estimated AUC72

Dear All,
The study was for combination of Fluticasone+Salmeterol pMDI with and without charcoal treatment (first two periods were without charcoal and last two periods were with charcoal) and the problem of extrapolation was majorly seen in with charcoal treatment periods for Salmeterol component and the AUC was calculated by Phoenix/WinNonlin using log-linear trapezoidal method.
Regards,

Ratnakar
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,638 registered users;
76 visitors (0 registered, 76 guests [including 10 identified bots]).
Forum time: 16:17 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