sam
★    

India,
2013-07-27 09:50
(4305 d 07:39 ago)

Posting: # 11067
Views: 22,204
 

 Regulatory Query for study repeat [Study As­sess­ment]

Post reply
Dear Sir,

Recently we have carried two pilot studies (Fasting and Fed) for a molecule and based on the pilot study data we have calculated the sample size and time point for the Pivotal study. Fortunately the data for the Fed study ratio is totally replica of the Pilot Fed study and our Fed study data qualify the BE criteria of 80-125% with 100 % power. But unfortunately our fasting study ratio unexpectedly lower than the Pilot Fasting study and the study does qualify the 80-125% and the power 100%. So based on the above fact we cannot file the molecule for regulatory approval. Since the Fasting pivotal data has unexpected outcome. So we wanted to repeat the whole study in other CRO from the beginning because of the unexpected ratio in the Pivotal study. In that case kindly suggest if there will be any regulatory hurdles if we repeat the whole study in other CRO or regulatory will easily accept the data of the second study if it qualifies the BE criteria.
Hope for your positive reply.


Regards

Sam

Edit: Category changed. [Helmut]
 
ElMaestro
★★★

Denmark,
2013-07-27 13:01
(4305 d 04:29 ago)

@ sam
Posting: # 11068
Views: 20,633
 

 Repeats

Post reply
Hi Sam,

❝ (...) the Fasting pivotal data has unexpected outcome. So we wanted to repeat the whole study (...)


This could be seen as bad and unethical science, and depending on where you submit for approval you may get a thumbs down.
You should not blindly repeat a study just because it had the 'wrong' outcome. There are situations where you can justify a repeat, notably:
  1. If you have a reason to believe the study conduct was flawed and results not trustworthy, then do an audit and investigate. If your audit truly confirms the presence of a problem then you may be able to repeat the study.

  2. If the study failed because of potential lack of power (wide CIs) then you may be able to repeat with a higher sample size. But this is a tricky matter because you will have to decide which PE to use for the calculation of sample size. If you observed a PE of e.g. 0.87 in the latest study, well then you might use that for the calculation and your sample size could go astronomical. After all, the PE from your latest study is a maximum likelihood estimate. Some companies will plug in a PE which is included in the latest observed CI but which is perhaps closer to unity than the latest PE. For example, if the CI was 0.72-1.01 then the PE was around 0.853, and it can be argued that the true PE might be 0.95 - at least there is not significance against that assumption. But plugging in 0.95 for sample size with 80% power or something like that although tempting is a risk of considerable magnitude; the apparent power of 80% is nice to report to management but is not likely to reflect the chance of approval.
Let's hear your numbers. What were your CIs, and what was the sample size?
—
Pass or fail!
ElMaestro
 
sam
★    

India,
2013-07-27 13:18
(4305 d 04:11 ago)

@ ElMaestro
Posting: # 11069
Views: 20,866
 

 Repeats

Post reply
Thanks for your esteemed reply. The PE of my pilot study was 90% and based on that the sample size of 54 was considered and including the dropout and withdrawn the final sample size of 64 was considered. But in the pivotal study the PE comes out to be 79% with power of 100% which shows that our sample size is very correct but there is some problem may be with study conduct or formulation. We have one more justification that the Pivotal PE of the Fed study is same as that of the Pilot which we carried out and Fed study meets the BE criteria.
On the basis of above reasons we wanted to re-conduct the same fasting study again but worried the regulatory will accept the same justification if the proposed study meets the BE limit.
Since this is our premier molecule we don’t want to miss this opportunity so need your expert views.
 
ElMaestro
★★★

Denmark,
2013-07-27 13:31
(4305 d 03:58 ago)

@ sam
Posting: # 11070
Views: 20,678
 

 Repeats

Post reply
Hi Sam,

❝ The PE of my pilot study was 90% and based on that the sample size of 54 was considered and including the dropout and withdrawn the final sample size of 64 was considered. But in the pivotal study the PE comes out to be 79% with power of 100% which shows that our sample size is very correct but there is some problem may be with study conduct or formulation. We have one more justification that the Pivotal PE of the Fed study is same as that of the Pilot which we carried out and Fed study meets the BE criteria.


Post-hoc power is to me not meaningful. I have no idea how to apply knowledge of that value. And stop up and think: What can you say about the power of a trial in which your PE is close to 0.79 regardless of sample size etc.?

Honestly, I think might be game over for the current formulation unless you have a good reason to cast doubt over the result (and as indicated above, a good reason is not just that the result is unwanted). If the past result is credible then a repeat trial could very well be futile and thus unethical as well as a waste of time and money.
—
Pass or fail!
ElMaestro
 
sam
★    

India,
2013-07-27 13:57
(4305 d 03:32 ago)

@ ElMaestro
Posting: # 11071
Views: 20,644
 

 Repeats

Post reply
Dear ElMaestro,

Thanks for the final conclusions.

If I am not wrong than we have to reformulate the formulation and try to get the ratio above 90% so that the formulation can qualify for the BE criteria.
Here I think we should not go for repeat of the whole study to waste of time and money.
One last suggestions I need from your end, if we go to repeat the whole study without any solid reason except the unexpected PE compared to the pilot study. What will be regulatory queries consequence with the regulatory approval. Will they approve our products based on the new data if it qualifies after repeat.


Regards

Sam
 
ElMaestro
★★★

Denmark,
2013-07-27 14:27
(4305 d 03:03 ago)

@ sam
Posting: # 11072
Views: 20,680
 

 Repeats

Post reply
Hi Sam,

❝ One last suggestions I need from your end, if we go to repeat the whole study without any solid reason except the unexpected PE compared to the pilot study. What will be regulatory queries consequence with the regulatory approval. Will they approve our products based on the new data if it qualifies after repeat.


I don't know how they will react. If you have a PE of 0.79 in one pivotal trial and the next trial passes with flying colors and PE close to 1.00 then I think an eyebrow might be raised here and there.
—
Pass or fail!
ElMaestro
 
sam
★    

India,
2013-07-27 14:45
(4305 d 02:44 ago)

@ ElMaestro
Posting: # 11073
Views: 20,783
 

 Repeats

Post reply
Dear All,

I seek help from you all of you who so ever have exposure related to this case study.

Regards
Sam
 
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-07-27 16:15
(4305 d 01:14 ago)

@ sam
Posting: # 11074
Views: 21,010
 

 Bad science

Post reply
<irony>

Dear Esteemed Madam,1

</irony>

I’m tempted to cut in.

❝ One last suggestions I need from your end, if we go to repeat the whole study without any solid reason except the unexpected PE compared to the pilot study. What will be regulatory queries consequence with the regulatory approval. Will they approve our products based on the new data if it qualifies after repeat.


I completely agree with what ElMaestro already stated. Let me summarize & clarify:
  • Your pilot passed and the pivotal failed. Although asked for, you didn’t give us the PEs and CIs (or CVs), the target power (π) in study planning, and the final sample size(s). Let’s say you planned for 90% power and expected ≥80% power taking drop-outs into account. Producer’s risk β = 1 – π. In other words, with 80% power one out of five studies are expected to fail on pure chance. That’s life!
  • Post hoc power is meaningless and should go straight into the statistical waste bin (remove it from your reports, it’s plain nonsense). The 100% you gave are outright wrong anyway. I strongly suggest to revise your estimation method. 100% power trans­lates into 0% chance of failure. But your study failed. Don’t you think that this is a contradiction? If you insist (for any wacky reason) to estimate post hoc power you must plug in the study’s PE into the formula. You might see that 79% will not even produce a result in some software. By definition power at the limits of the acceptance range will not exceed the level of the test α (patient’s risk). Hence, with any PE at or outside of the acceptance range – and with any CV! – power will be ≤5%.2 You don’t even need soft­ware; this a consequence of the 90% confidence interval or TOSTs per­formed at 5%.
  • Based on the above repeating the study would be both bad science3 and unethical.
  • A PE outside the acceptance range from a properly powered study calls for re­for­mula­tion, IMHO.
  • BTW, in my experience more studies fail in fed state than in fasting state. It is an in­com­prehensible mystery to me why many sponsors perform the fasting study first.
    :confused:
  • Let’s assume you ignore our advice,4 repeat the study, and pass now. You have to sub­mit the synopsis of the failed study to the authority, which – since results are contra­dic­tory – immediately will ask for the complete report of the failed study. Authorities make their decision based on the whole body of evidence. Why should they trust results of the second pivotal study more than the first’s? The results of the pilot are supportive at its best. Or will you try to convince them that something (Namely what? Do you have any evidence?) went wrong in the first study (since you proposed to repeat the study in another CRO), rendering it less credible? In the best (!) case this will trigger an inspec­tion – not only of this study, but all others as well. Potential outcome:
    • No findings. Both studies are credible. Since the authority is interested in pro­tect­ing public health (not in the profit of the generic industry!) they will stay on the safe side and don’t approve the product.
    • Form 483 (FDA) or PSRtPH (EMA) issued – confirming the first study not credible. Maybe the second study will “count more” now.
      But: According to GCP it is the responsibility of the sponsor to guarantee proper study conduct (pre-study audit, qualification and training of personnel, pro­ce­dures, SOPs, QAU, study monitoring, …). Obviously you did not comply with these requirements. In the worst case it will retrospectively affect already accepted studies performed at this CRO.
  1. See also this post.
  2. [image]Since you gave us no suitable numbers of the pivotal study my crystal ball whispered CV=24.5% and n=60. With a PE of 79% post hoc power is 2.70%. That’s not what I call an awful lot.
  3. […] our greatest mistake would be to forget that data is used for serious decisions in the very real world, and bad information causes suffering and death. Ben Goldacre
  4. The combination of some data and an aching desire for an answer does not ensure that a reasonable answer can be extracted from a given body of data. John W. Tukey
—
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
 
sam
★    

India,
2013-07-28 11:05
(4304 d 06:24 ago)

@ Helmut
Posting: # 11082
Views: 20,764
 

 Bad science

Post reply
Dear Helmut:

Totally agree with your suggestions given.
  1. As per your earlier suggestions we have carried out Fed studies before the Fasting study.
  2. We were very happy when our Fed study meets the BE criteria.
  3. But totally depressed by seeming the results of the Fasting Data.
Its also acceptable that the repeating the study will comes out with various doubts in the mind of Regulatory persons and it is also unethical. But as you stated that the chances of variation is more in the Fed compared to the Fasting study. That’s why the questions comes out in our mind that there may be some flaws in the study conduct of the Fasting study.
If we are talking about the formulation problem then why the same is not reflecting in the Fed study.
Also the results of the Fed pilot is similar to the Fed Pivotal.

Few of the results are also provided for your references and give us the final suggestions.

Fed Pivotal Study Results:
Dependent       Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)
Ln(AUCINF_obs)     96          91          102      100     14.5
Ln(AUClast)        96          91          101      100     14.8
Ln(Cmax)           87.91       82           94       99.99  20.3

Fasting Pilot study Data:
Dependent       Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)
Ln(AUCINF_obs)     87.02       69.00      109.75    0.48    37.07
Ln(AUClast)        95.22       77.65      116.76    0.56    32.34
Ln(Cmax)           89.43       77.16      103.66    0.81    23.13

Fasting Pivotal Study results:
Dependent       Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)
Ln(AUCINF_obs)     98          92         103       100     18
Ln(AUClast)        98          92         103       100     18
Ln(Cmax)           79          74          84       100     21


Best Regards
Sam

Edited using BBCodes. Don’t use tabs in your posts. [Helmut]
 
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-07-28 14:32
(4304 d 02:58 ago)

@ sam
Posting: # 11083
Views: 20,650
 

 Posting style & incomplete information

Post reply
Hi Sam!

❝ 1. […] we have carried out Fed studies before the Fasting study.


Oops, I overlooked the order of studies. Sometimes that happens (I didn’t say that products in fed state always perform worse than in fasting state). If I recall it correctly Dan reported another case a while ago.

❝ 3. But totally depressed by seeming the results of the Fasting Data.


No need to be depressed. Get paroxetine or visit your vegetarian and ask for Schützo­mycin. Once again: One out of five studies fail on pure chance. Make it your Mantra. Suggest it to your boss as well.

❝ But as you stated that the chances of variation is more in the Fed compared to the Fasting study.


Chances! Rules of thumb ≠ laws. This information comes from my personal experience of 600+ BE studies.

❝ That’s why the questions comes out in our mind that there may be some flaws in the study conduct of the Fasting study.


Did you monitor the study? Any observations? If not, audit the CRO. Without nailing down a root cause it will not be ethical to repeat the study in another CRO – only “justi­­fied” because you did not like the study’s outcome (murmur the Mantra instead).

❝ If we are talking about the formulation problem then why the same is not reflecting in the Fed study.


I don’t practice reading tea leaves. You should know the formulation best.

❝ Few of the results are also provided for your references and give us the final suggestions.


Why only a few? BTW, it took me ten minutes to edit your post (tabulators are rendered to single spaces in HTML). Please read the Forum’s “Operating Instructions”, go and play around in the Sandbox-category, and always (!) use the [image] before posting. It’s intentional that this button is the first one in the row.
Thank you very much in advance.

In the future please give results in percent to two decimals (as required by FDA & EMA) or with five significant figures. Why do you think we deserve less information than regulators?

I still miss the sample sizes. It seems that the fasting pilot was performed in fifteen sub­jects. For the fed pivotal I guess you evaluated 49 (Cmax), 48 (AUCt), and 36 (AUC∞). For the fasting pivotal I guess 56 for both AUCs and 60 for Cmax. Why did you exclude subjects – especially from the comparison of AUCt?

Are the headings of the tables correct? Which was the outcome of the fed pilot?

Please [image] your original post until Monday 12:35 IST. The output looks winnonlinish to me. Post hoc power is
  1. meaningless (see my last post) and
  2. flawed in WinNonlin for ages and in Phoenix as well (see this thread).
    This column offends my eye.
You want some help? Fine. In the future give us all information you have already in the first post.* This question-and-answer-to-and-fro-game (aka worm information out of you) is wasting our time. Example: From the different sample sizes of metrics in the same studies I suspect problems with the profiles. Don’t you consider it worthwhile answering John’s post?

  • Sometimes the key to an answer is found
    in the way you formulate the question.     David Brin
—
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
 
sam
★    

India,
2013-07-29 08:46
(4303 d 08:43 ago)

@ Helmut
Posting: # 11091
Views: 20,784
 

 Posting style & incomplete information

Post reply
Dear Helmut,

Thanks for your time.

The heading of the Tables are correct as i have not given the Table for Fed pilot in my previous post but yes ratios of all the three parameters were same as per the pivotal study.

Fed Pilot Study Results (Total Sample szie 18):
Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n
Ln(AUCINF_obs)     87.39     69.74      108.72      51.0   32.2  14
Ln(AUClast)        81.09     67.57       97.32      65.0   26.3  14
Ln(Cmax)           87.07     67.52      113.11      41.0   37.8  14

Fed Pivotal Study Results (Total Sample szie 48):
Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n
Ln(AUCINF_obs)     96.45     91.53      101.64     100.0   14.5  44
Ln(AUClast)        95.94     90.95      101.21     100.0   14.8  44
Ln(Cmax)           87.91     81.81       94.47      99.99  20.3  44

Fasting Pilot study Data (Total Sample szie 18):
Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n
Ln(AUCINF_obs)     87.02     69.00      109.75       0.48  37.07 14
Ln(AUClast)        95.22     77.65      116.76       0.56  32.34 14
Ln(Cmax)           89.43     77.16      103.66       0.81  23.13 14

Fasting Pivotal Study results (Total Sample szie 64)
Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n
Ln(AUCINF_obs)     98.26     93.05      103.77     100.0   18.0  61
Ln(AUClast)        98.14     92.84      103.74     100.0   18.3  61
Ln(Cmax)           79.10     74.09       84.45     100.0   21.9  61

I have personally monitored the whole study but there were no observation at that time.

Now the final sample size on which data has been evaluated are given in each tables.

Hope i have given all the information required by you, if not pl let me know so that i can add more if required.

Once again thanks for time and valuable responses.

So pl let me know if i need revisit the CRO and do the Audit. Also suggest me for the main areas where i need to concentrate during the audit so that the root cause of the study failure can be find out and i can prepare the justification for repeating the study.


Regards

Sam
 
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-07-29 09:53
(4303 d 07:37 ago)

@ sam
Posting: # 11092
Views: 20,451
 

 RTFM!

Post reply
Dear Sam,

why didn’t you use the fucking [image] before posting? :angry:
Would you be so kind and format tables as described here? Scroll down to Note 1. Highlight the text and click the button Code to the right.

Please give also data of the fed pilot study. The “ratios of all the three parameters were same” is not exhaustive information + I don’t believe it at all.
—
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
 
sam
★    

India,
2013-07-29 10:40
(4303 d 06:49 ago)

@ Helmut
Posting: # 11094
Views: 20,341
 

 RTFM!

Post reply
Dear Helmut,

Sorry for inconvenience.
Corrected now with added pilot study results.


Regards

Sam
 
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-07-29 13:14
(4303 d 04:15 ago)

@ sam
Posting: # 11098
Views: 20,351
 

 copy & paste error or what?

Post reply
Dear Sam,

I don’t believe in your sample sizes. Based on your CIs, I get different ones. Still I think that you excluded subjects’ AUCs. Why?

Fed Pilot Study
                     yours             mine
Dependent      Power   CV(%)  n  “power”  CV%   n
Ln(AUCINF_obs)  51.0   32.2  14    7.88  32.3  13
Ln(AUClast)     65.0   26.3  14    5.72  26.3  13
Ln(Cmax)        41.0   37.8  14    3.42  37.9  13

Fed Pivotal Study
                     yours             mine
Dependent      Power   CV(%)  n  “power”  CV%   n
Ln(AUCINF_obs) 100.0   14.5  44  100.00  14.5  43
Ln(AUClast)    100.0   14.8  44  100.00  14.8  43
Ln(Cmax)        99.99  20.3  44   69.96  20.3  44

Fasting Pilot study
                     yours             mine
Dependent      Power   CV(%)  n  “power”  CV%   n
Ln(AUCINF_obs)   0.48  37.07 14    5.49  37.07 15
Ln(AUClast)      0.56  32.34 14   18.35  32.34 15
Ln(Cmax)         0.81  23.13 14   33.73  23.13 15

Fasting Pivotal Study
                     yours             mine
Dependent      Power   CV(%)  n  “power”  CV%   n
Ln(AUCINF_obs) 100.0   18.0  61  100.00  18.0  60
Ln(AUClast)    100.0   18.3  61  100.00  18.3  60
Ln(Cmax)       100.0   21.9  61    2.68  21.9  61

Why are you notoriously not answering questions raised by ElMaestro, John, and myself? If you only here to watch for …

❝ the justification for repeating the study

… be prepared for a rather extended waiting period. :sleeping:
—
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
 
sam
★    

India,
2013-07-29 14:46
(4303 d 02:43 ago)

@ Helmut
Posting: # 11107
Views: 20,225
 

 copy & paste error or what?

Post reply
Dear Helmut,

I have not excluded any subjects from AUC.

There may be littile diffrence in sample szie due to diffrent softwares we use.


Regards

Sam
 
jag009
★★★

NJ,
2013-07-29 18:06
(4302 d 23:23 ago)

@ Helmut
Posting: # 11113
Views: 20,218
 

 Calm down!

Post reply
Woo wise one. Calm down. ;-)

John

Point taken. Helmut
 
ElMaestro
★★★

Denmark,
2013-07-29 11:19
(4303 d 06:10 ago)

@ sam
Posting: # 11095
Views: 20,385
 

 Slightly off-topic: The wonders of pilot trials

Post reply
Hi Sam and all,

I think this is extremely interesting:

❝ Fed Pilot Study Results (Total Sample szie 18):

❝ Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n

❝ Ln(AUCINF_obs)     87.39     69.74      108.72      51.0   32.2  14

❝ Ln(AUClast)        81.09     67.57       97.32      65.0   26.3  14

❝ Ln(Cmax)           87.07     67.52      113.11      41.0   37.8  14

❝

❝ Fed Pivotal Study Results (Total Sample szie 48):

❝ Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n

❝ Ln(AUCINF_obs)     96.45     91.53      101.64     100.0   14.5  44

❝ Ln(AUClast)        95.94     90.95      101.21     100.0   14.8  44

❝ Ln(Cmax)           87.91     81.81       94.47      99.99  20.3  44


Note that the upper CI limit for ln(AUCt) is about 97% and PE=81% in the pilot trial and in the pivotal trial the PE comes out at 96%. This shows the practical importance of not trusting too much in pilot trials. In this case, however, it worked in the sponsor's favour.

I think I will mention this example at the IPAC-RS meeting in Orlando in March 2014 where the talk is about two-stage approaches. This story could perhaps exemplify why a two-stage trial with a stopping criterion would have been a potentially interesting alternative.

At this point, however, I am slightly confused:
  1. How did you transit from the pilot trial to the pivotal?
    It looks like you aimed for N=48 in the pivotal, and got N=44 evaluable subjects. But how did N=48 enter the scene?
    With the CV for Cmax being 37.8, and assuming this is a 2,2,2-BE trial (is it not?), N=48 gives a very low power at PE=87.07; if optimism prevails and we think the PE is actually 95% then I get a power of around 71% for N=48; it would need to be N=60 to get 80% power.
  2. Could you tell how much time was in between the pilot and the pivotal fed studies?
Highly, highly interesting this story. And garnished with a classical Hötzi rant, it has truly made my day. :-D
—
Pass or fail!
ElMaestro
 
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-07-29 14:16
(4303 d 03:13 ago)

@ ElMaestro
Posting: # 11101
Views: 20,442
 

 Futility rule?

Post reply
Hi ElMaestro,

❝ I think this is extremely interesting:


PEs of fed study:  pilot  pivotal
Ln(AUCINF_obs)     87.39   96.45
Ln(AUClast)        81.09   95.94
Ln(Cmax)           87.07   87.91

Extremely interesting indeed because Sam wrote in his original post:

❝ […] data for the Fed study ratio is totally replica of the Pilot Fed study

And later

❝ […] Pivotal PE of the Fed study is same as that of the Pilot

And again

❝ […] Fed pilot […] ratios of all the three parameters were same as per the pivotal study


Same? Would you buy such a Bangkok Rolex-replica?*

❝ […] This story could perhaps exemplify why a two-stage trial with a stopping criterion would have been a potentially interesting alternative.


Can you elaborate? I think it was very courageous (pun!) to proceed with a PE of 81% for AUCt to a pivotal study.

❝ 1. How did you transit from the pilot trial to the pivotal?


As Sam said:

❝ We were very happy when our Fed study meets the BE criteria.


With no dropouts and PE/CV from the pilot “carved in stone” I would expect power of 8.2% [sic] in 48 subjects. That’s slightly lower than walking into a casino and go for a street in roulette (chances on a French table 1/11 ~9.1%). No surprise everybody was very happy when the pivotal study passed in 43 subjects.

❝ […] garnished with a classical Hötzi rant,


[image]

  • Wer einmal lügt, dem glaubt man nicht. (German proverb)
—
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,
2013-07-29 14:46
(4303 d 02:43 ago)

@ Helmut
Posting: # 11106
Views: 20,320
 

 Futility rule?

Post reply
Hi Hötzi,

❝ Same? Would you buy such a Bangkok Rolex-replica?


Haha, I don't know what to think. I was mainly addressing the numbers, hoping they reflected the actual and reported study results. Whether or not 81 equals 96 appears to be a subjective matter. But when women call their friends and tell you they are just going to chat for 5 minutes then you can, as a rule of thumb, be very sure that 5 equals 60 or more.

❝ Can you elaborate? I think it was very courageous to proceed with a PE of 81% for AUCt to a pivotal study.


No doubt about that, and you are now almost too generously using terms from the positively charged heap. If my clients get a PE of 81% in a pilot then I would recommend that the plug be pulled. To be honest, I think this was a case of mild dumb luck.
Sam, I am not saying this to offend you, and if you take offence then I apologise and will offer to revise my post. But the decision to move forward after a pilot having PE=81% was simply a bad one from a scientific/ethical perspective, even though it in hindsight worked in your favour.

In the subsequent pilot-pivotal trial pair the opposite phenomenon was observed.

What I meant with the two-stage comment was that sometimes pilots do not reflect pivotals, and there could be different reasons for it. With OIPs, for example, products can change a lot over time and no-one really knows what this implies for the in vivo situation. But it suggests that sometimes the value of a pilot trial can be questionable if there is a large time span between the pilot and the pivotal (assuming same batches were used), and hence my question about this aspect.
—
Pass or fail!
ElMaestro
 
sam
★    

India,
2013-07-29 14:41
(4303 d 02:48 ago)

@ ElMaestro
Posting: # 11104
Views: 20,320
 

 Slightly off-topic: The wonders of pilot trials

Post reply
Dear

Based on the Cmax Data we proceed further for the Pivotal study.

The time between the Pilot and Pivotal in around 2 months.

Regards

Sam
 
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-07-29 14:45
(4303 d 02:44 ago)

@ sam
Posting: # 11105
Views: 20,251
 

 PE of AUC?

Post reply
Hi Sam!

❝ Based on the Cmax Data we proceed further for the Pivotal study.


… by completely ignoring the 81% observed for AUCt. Why? Or do you like gambling?
—
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
 
sam
★    

India,
2013-07-29 14:56
(4303 d 02:33 ago)

@ Helmut
Posting: # 11108
Views: 20,247
 

 PE of AUC?

Post reply
Dear Helmut,

In the last pilot study we have reduced the sampling period upto 72 hours to just check the Cmax not AUC for saving the cost.

We are neglecting the Pilot AUCs for Sample size calculations.

There is very less variation in the AUC that has been evident from the previous study.

Based on that i am requesting you pl give me the conclusions

Regards

sam
 
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-07-29 15:18
(4303 d 02:12 ago)

@ sam
Posting: # 11110
Views: 20,181
 

 EOD

Post reply
Hi Sam,

❝ We are neglecting the Pilot AUCs for Sample size calculations.


Brilliant!

❝ Based on that i am requesting you pl give me the conclusions

  • Have a decent cup of tea, lite a chillum, and meditate on your first sentence.
  • Read my presentations on sample size estimation.
  • Read them again.
  • Attend a course on ethics in science.
EOD from my side.
—
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,
2013-07-29 18:32
(4302 d 22:57 ago)

@ sam
Posting: # 11114
Views: 20,269
 

 Posting style & incomplete information

Post reply
Hi,

❝ Fed Pilot Study Results (Total Sample szie 18):

❝ Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n

❝ Ln(AUCINF_obs)     87.39     69.74      108.72      51.0   32.2  14

❝ Ln(AUClast)        81.09     67.57       97.32      65.0   26.3  14

❝ Ln(Cmax)           87.07     67.52      113.11      41.0   37.8  14

❝

❝ Fed Pivotal Study Results (Total Sample szie 48):

❝ Dependent      Ratio[%Ref] CI_90_Lower CI_90_Upper Power   CV(%)  n

❝ Ln(AUCINF_obs)     96.45     91.53      101.64     100.0   14.5  44

❝ Ln(AUClast)        95.94     90.95      101.21     100.0   14.8  44

❝ Ln(Cmax)           87.91     81.81       94.47      99.99  20.3  44


This really bugs me (even though I am looking at fed data). You have a 15% increase in AUCt ratio between pilot and pivotal.

Question 1) I am sure that you did. But did you use the same lot# of the reference product throughout the program (yes, all pilot and pivotal studies)?

John
 
sam
★    

India,
2013-07-30 08:08
(4302 d 09:21 ago)

@ jag009
Posting: # 11121
Views: 20,156
 

 Posting style & incomplete information

Post reply

❝ Question 1) I am sure that you did. But did you use the same lot# of the reference product throughout the program (yes, all pilot and pivotal studies)?


Hi John,

Yes I use the same batch of reference product throughout the pilot and pivotal for fasting and Fed

Regards

Sam
 
ElMaestro
★★★

Denmark,
2013-07-28 14:43
(4304 d 02:46 ago)

@ sam
Posting: # 11084
Views: 20,482
 

 Bad science

Post reply
Hi Sam,

CV of Cmax in the pilot was lower than CV of AUCt.
This is almost never the true case. My guess is the pilot trial gave you the wrong impression of the product performance; the two products are not equivalent as evident by the last pivotal trial.
—
Pass or fail!
ElMaestro
 
jag009
★★★

NJ,
2013-07-28 00:17
(4304 d 17:13 ago)

@ ElMaestro
Posting: # 11078
Views: 20,619
 

 Repeats

Post reply
Hi ElMaestro,

❝ Post-hoc power is to me not meaningful. I have no idea how to apply knowledge of that value. And stop up and think: What can you say about the power of a trial in which your PE is close to 0.79 regardless of sample size etc.?


Yup I agreed. Just finished a study with n=60 and post-hoc power was 75%. Guess what, study passed 90%CI, intraCV was 55%. T/R ratio was 99%.

John
 
ElMaestro
★★★

Denmark,
2013-07-28 01:11
(4304 d 16:18 ago)

@ jag009
Posting: # 11080
Views: 20,488
 

 Repeats

Post reply
Hi John,

❝ Just finished a study with n=60 and post-hoc power was 75%.


The other day I made strawberry smoothies for my five friends and myself. I thought one cup of berries would suffice for three smoothes, so I used two cups. At the end I was able to make seven smoothies with that amount of berries. Oddly enough, in spite of the fact that I had more berries than necessary for six smoothies, all seven smoothies tasted completely normal. Isn't that just very strange?

❝ Guess what, study passed 90%CI, intraCV was 55%. T/R ratio was 99%.


Good work, John :ok:. Now try and hit PE=99% with an inhaled product :pirate:
—
Pass or fail!
ElMaestro
 
jag009
★★★

NJ,
2013-07-28 00:29
(4304 d 17:01 ago)

@ sam
Posting: # 11079
Views: 20,709
 

 Regulatory Query for study repeat

Post reply
Hi Sam,

❝ Recently we have carried two pilot studies (Fasting and Fed) for a molecule and based on the pilot study data we have calculated the sample size and time point for the Pivotal study. Fortunately the data for the Fed study ratio is totally replica of the Pilot Fed study and our Fed study data qualify the BE criteria of 80-125% with 100 % power. But unfortunately our fasting study ratio unexpectedly lower than the Pilot Fasting study and the study does qualify the 80-125% and the power 100%.


Just want to make sure... You meant fasting study didn't qualify for 80-125% on the 90%CI and the power 100%?

Questions:
  1. Similar to what the other gurus have asked, please provide the numbers to us.
  2. Was there a change in the formulation or process? From pilot batch to scale-up batch... Please check because I have seen a few cases from my previous company (and friend's companies) that the issue was related to the manufacturing process. I work closely with formulators these days (small company) and you will be surprised to hear the things that they educate me on...
  3. Have you checked the PK data in details, both pilot and pivotal. Look for outliers? Don't bother with the post-hoc power business.
  4. Any difference in study population between studies?
  5. Drug has complicated kinetics?
John
 
sam
★    

India,
2013-07-29 14:36
(4303 d 02:53 ago)

@ jag009
Posting: # 11103
Views: 20,291
 

 Regulatory Query for study repeat

Post reply
Dear John,

❝ 1. Similar to what the other gurus have asked, please provide the numbers to us.


Not possible to provide the numbers

❝ 2. Was there a change in the formulation or process? From pilot batch to scale-up batch... Please check because I have seen a few cases from my previous company (and friend's companies) that the issue was related to the manufacturing process. I work closely with formulators these days (small company) and you will be surprised to hear the things that they educate me on...


There was no change in the formulation or process.

❝ 3. Have you checked the PK data in details, both pilot and pivotal. Look for outliers? Don't bother with the post-hoc power business.


I have checked the Pilot and Pivotal data thoroughly.

❝ 4. Any difference in study population between studies?


The study populations are totally different. Pilot are carried out in other CRO and Pivotal in other

❝ 5. Drug has complicated kinetics?


No.

Best Regards
Sam

Edit: Standard quotes restored. [Helmut]
 
luvblooms
★★  

India,
2013-07-30 10:28
(4302 d 07:01 ago)

@ sam
Posting: # 11122
Views: 20,202
 

 Regulatory Query for study repeat

Post reply
Dear Sam,

A few more thngs to look at from formulation point of view

❝ ❝ 2. Was there a change in the formulation or process? From pilot batch to scale-up batch...

❝ There was no change in the formulation or process.


Well, how about
  1. the changes in API particle size? Was the PSD range too wide and both the lots (pilot and pivotal). Normally particle size changes doesnt give major difference in FED condition as drug stays in stomach for longer time and have sufficient fat to get solubilized but these changes are prominent in Fasting condition.
  2. Different form of API?: were the form (polymorphic/amorphous) remains same from pilot to pivotal?
  3. Different lot of API?Again to the first point are there any change in lots of API?
  4. Changes in Inactive ingredient grade or quality? Was there any change in amount or grade of any inactive ingerdient that may affect the release from the product. Some very commomly used excipients behave wildly within different grades.

❝ ❝ 3. Have you checked the PK data in details, both pilot and pivotal.

❝ I have checked the Pilot and Pivotal data thoroughly.


Any differences observed in Tmax or shape of C-t curve? That could give you an idea why the product behave was different.

❝ ❝ 4. Any difference in study population between studies?

❝ The study populations are totally different. Pilot are carried out in other CRO and Pivotal in other


How about the meal compositions? Were they remain the same? In India the food varies a lot between Chennai and Ahmedabad and so thus the performance of the formulations (this is based on my experience of some IR and XR products). Get a hold of that too along with the demographic data. Even that might give you some idea. :cool:


Hope this will help..
—
~A happy Soul~
 
jag009
★★★

NJ,
2013-07-30 18:53
(4301 d 22:37 ago)

@ luvblooms
Posting: # 11131
Views: 20,087
 

 More questions..

Post reply
Hi,

❝ ❝ ❝ 3. Have you checked the PK data in details, both pilot and pivotal.

❝ ❝ I have checked the Pilot and Pivotal data thoroughly.

❝

❝ Any differences observed in Tmax or shape of C-t curve? That could give you an idea why the product behave was different.


One more.. How did the reference behave between pilot and pivotal studies under fasting and fed conditions? How close were the numbers (Cmax, AUC, Tmax)?

If the reference behaves "wildy" between pilot and pivotal studies then maybe you can use that to support the proposal for a repeat study? This is a sort of deadend thought.

Were the study population comparable? ie: males to female ratio.

John
 
UA Flag
Activity
 Admin contact
23,424 posts in 4,927 threads, 1,669 registered users;
134 visitors (0 registered, 134 guests [including 9 identified bots]).
Forum time: 17:30 CEST (Europe/Vienna)

We should not speak so that it is possible
for the audience to understand us,
but so that it is impossible
for them to misunderstand us.    Quintilian

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