nlpati
☆    

India,
2013-10-10 11:40
(2746 d 18:27 ago)

Posting: # 11641
Views: 6,799
 

 Difference between test and reference [Regulatives / Guidelines]

Dear Sir,
Guidance for Industry Bioavailability and Bioequivalence Studies for Orally Administered Drug Products — General Considerations, mentions as "The drug content of the test product cannot differ from that of the reference listed product by more than 5 percent". Kindly request you to suggest that if there is any formula for finding the difference or is it only difference between assay values of test and reference.

with rgds,
Lakshmipathy
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-10-10 12:46
(2746 d 17:21 ago)

@ nlpati
Posting: # 11642
Views: 6,299
 

 Assayed content

Namaste Lakshmi,

» Dear Sir,

    Not interested in opinions of the female members of the forum?


» […] "The drug content of the test product cannot differ from that of the reference listed product by more than 5 percent". […] if there is any formula for finding the difference or is it only difference between assay values of test and reference.

Only the latter.

Dif-tor heh smusma 🖖
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
jag009
★★★

NJ,
2013-10-10 13:19
(2746 d 16:47 ago)

@ nlpati
Posting: # 11643
Views: 6,159
 

 Difference between test and reference

Yes, only the difference between assay values of test and reference. No complicated formula.

John
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-10-10 14:08
(2746 d 15:58 ago)

@ jag009
Posting: # 11644
Views: 6,502
 

 ∆ <5% ⇒ T/R 95%

Hi John & Lakshimi,

forgot one point: Even if the assayed contents differ by less than 5%, don’t take this value as the “truth”. The analytical method has limited accuracy/precision. For anything <5% I would still assume a T/R-ratio of 95%.

Dif-tor heh smusma 🖖
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
nlpati
☆    

India,
2013-10-11 02:37
(2746 d 03:30 ago)

@ Helmut
Posting: # 11649
Views: 6,196
 

 ∆ <5% ⇒ T/R 95%

» Even if the assayed contents differ by less than 5%, don’t take this value as the “truth”. The analytical method has limited accuracy/precision. For anything <5% I would still assume a T/R-ratio of 95%.

Dear Sir,

We are using formula for calculation of this value is this the right one.

Formula is (difference between assay value of test and reference / assay value of reference) X 100

Is this formula required for afore said difference calculation.
jag009
★★★

NJ,
2013-10-13 19:02
(2743 d 11:04 ago)

@ Helmut
Posting: # 11657
Views: 6,144
 

 ∆ <5% ⇒ T/R 95%

Hi Helmut,

Yes but we usually only have one number (don't present SD or range) on the CofAs (test and ref)

Thanks

John
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-10-14 15:50
(2742 d 14:16 ago)

@ jag009
Posting: # 11662
Views: 6,071
 

 Optimism

Hi John,

» Yes but we usually only have one number (don't present SD or range) on the CofAs (test and ref)

Correct. You could ask the lab for the accuracy/precision of the method. In my previous post I only wanted to wave the warning flag concerning overly optimistic assumptions like “Hey, the difference is just 2%, let’s estimate the sample size based on a ratio of 98%”

I’m not a pessimist,
I’m just a well informed optimist.
(José Saramago)


Dif-tor heh smusma 🖖
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
kumarnaidu
★    

Mumbai, India,
2013-11-26 04:36
(2700 d 01:30 ago)

(edited by kumarnaidu on 2013-11-26 10:08)
@ Helmut
Posting: # 11969
Views: 5,900
 

 T/R 90% to 110% for sample size estimation

Dear all,
If in my earlier studies I am getting T/R ratio= 110% means >5% difference then for planning another study for another market should I go with the T/R=110% or 95% for sample size estimation.

As per EMA guideline (2010) "batches should not differ more than 5%" so taking 10% difference in T/R ratio is considerable.

Also if I am taking the T/R ratio=110% I am getting sample size around 50 (without dropout) but many studies I found for this drug is around 40 even our earlier study had 42. then in that case if am exceeding sample size will it considered as a forced bioequivalence?

Earlier study results used for sample size estimation
LnCmax ratio=110.3 (90%CI- 101.06-120.54) CV=24.3.

Please reply.

Kumar Naidu
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2013-11-30 13:02
(2695 d 17:04 ago)

@ kumarnaidu
Posting: # 11992
Views: 5,748
 

 T/R 110%

Hi Kumar,

» If in my earlier studies I am getting T/R ratio= 110% means >5% difference then for planning another study for another market should I go with the T/R=110% or 95% for sample size estimation.

The former. Although with another reference product the ratio might be better, without a pilot study I would suggest a conservative approach.

» As per EMA guideline (2010) "batches should not differ more than 5%" so taking 10% difference in T/R ratio is considerable.

Yes, why not? The ≤5% refer to measured contents, not the assumed in vivo ratio.

» Also if I am taking the T/R ratio=110% I am getting sample size around 50 (without dropout) …

Correct (given the CV of 24.3%).

» … but many studies I found for this drug is around 40 even …

Some drug, but different formulations.

» … our earlier study had 42.
» LnCmax ratio=110.3 (90%CI- 101.06-120.54) CV=24.3.

Passed by luck. “Power” was 76%.

» then in that case if am exceeding sample size will it considered as a forced bioequivalence?

Don’t think so – unless you go to Denmark* (100% not included in CI).


  • In the opinion of the Danish Medicines Agency, the 90 % confidence intervals for the ratio of the test and reference products should include 100 %.
    Reference

Dif-tor heh smusma 🖖
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
kumarnaidu
★    

Mumbai, India,
2013-12-02 04:35
(2694 d 01:32 ago)

@ Helmut
Posting: # 11995
Views: 5,624
 

 T/R 110%

Thanks Helmut for your guidance.

Kumar Naidu
mittyri
★★  

Russia,
2014-03-14 07:10
(2591 d 22:56 ago)

@ Helmut
Posting: # 12620
Views: 5,050
 

 the basis of T/R in the Pro­tocol

Hi Helmut and All,

» » If in my earlier studies I am getting T/R ratio= 110% means >5% difference then for planning another study for another market should I go with the T/R=110% or 95% for sample size estimation.

» The former. Although with another reference product the ratio might be better, without a pilot study I would suggest a conservative approach.

Do you state in the protocol the reasons why did you choose the exact T/R ratio (not default 95% setting)?

Kind regards,
Mittyri
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2014-03-14 12:53
(2591 d 17:13 ago)

@ mittyri
Posting: # 12625
Views: 5,064
 

 the basis of T/R in the Pro­tocol

Hi Mittyri,

» Do you state in the protocol the reasons why did you choose the exact T/R ratio …

Yes.

» … (not default 95% setting)?

There is no “default” of 95%. The batch release spec’s generally are ±10%. EMA wants to have less than ±5% in measured content (unless it is demonstrated that nor suitable reference batch could be found; in this case a dose-correction is accept­able). But ±5% in content does not necessarily translate into ±5% in vivo. If I have information about a larger deviation from a pilot study I use it.
The GL only mentions an “appropriate sample size calculation”, whereas the old NfG was more specific: “the expected deviation from the reference product”. No idea why EMA boiled down the detailed – and reasonable – section about sample size estimation to plain :blahblah:.
I also include a sensitivity analysis according to ICH-E9, showing how violations of assumptions (larger deviation and/or CV, higher drop-out rate) might affect power. I generally aim for ~90% power (best case) and try to keep power ≥70% (worst case).

For HVDs/HVDPs I use ±10% even if I have seen a smaller deviation in a pilot study. For these kinds of drugs, the ratio tends to jump around between studies. I try to be con­servative.

Note that the FDA requires for NTIDs tighter release spec’s (±5%). For NTIDs I there­fore use ±2.5%. IMHO planning a NTID-study without a reasonably sized pilot is risky.

Dif-tor heh smusma 🖖
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
mittyri
★★  

Russia,
2014-03-14 13:31
(2591 d 16:36 ago)

@ Helmut
Posting: # 12626
Views: 5,025
 

 the basis of T/R in the Pro­tocol

Dear Helmut,

Thank you for the explanation!

» If I have information about a larger deviation from a pilot study I use it.

Do you insert any links to the publications about these pilot studies (may be for another market)?
Would you rest the T/R Ratio in the protocol without explanation?
Is it important to the Assessor to know - where do you get this ratio or why are you assuming such ratio?

Kind regards,
Mittyri
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2014-03-14 14:01
(2591 d 16:06 ago)

@ mittyri
Posting: # 12629
Views: 5,063
 

 the basis of T/R in the Pro­tocol

Hi Mittyri,

» » If I have information about a larger deviation from a pilot study I use it.
» Do you insert any links to the publications about these pilot studies (may be for another market)?

By “pilot” I meant studies the sponsor performed or have access to in an acquired dossier. Only such studies can be assesses for the quality of data (clinical performance, bioanalytics, …). Studies from the literature should be used with caution.

» Would you rest the T/R Ratio in the protocol without explanation?

No.

» Is it important to the Assessor to know - where do you get this ratio or why are you assuming such ratio?

I think we have to distinguish between assessing the protocol (authority, IEC)* and the report.
  • The former is much more important – especially from an ethical per­spec­tive. Only if you give all the data which led you to your decision about the sample size, the IEC could justify whether the study is ethically accept­able.
  • Once the study passes BE, all assumptions in the protocol are not rele­vant any more (therefore, post hoc power is nonsense).

  • Once the Austrian agency AGES did not like the sample size we stated in the protocol: “It is not acceptable to perform the study in only 16 sub­jects when most of the BE studies of XYZ are performed in more than 30.” Hey, it was a pilot study… Case closed. :-D

Dif-tor heh smusma 🖖
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
mittyri
★★  

Russia,
2014-03-14 20:08
(2591 d 09:59 ago)

@ Helmut
Posting: # 12631
Views: 4,984
 

 the basis of T/R in the Pro­tocol

Dear Helmut,

Thank you for the inestimable contribution to my (and others) knowledge in BEQ :cool:

» By “pilot” I meant studies the sponsor performed or have access to in an acquired dossier. Only such studies can be assesses for the quality of data (clinical performance, bioanalytics, …).

If I understand right, you state in the protocol phrases like that:
According to the earlier performed by sponsor study, T/R ratio could be assumed as 0.92%.
How could I prove to the assessor that "pilot" study is really performed? As a sponsor I can write that I performed the study in another market, but assessment report isn't published (no links)

» Once the Austrian agency AGES did not like the sample size we stated in the protocol: “It is not acceptable to perform the study in only 16 sub­jects when most of the BE studies of XYZ are performed in more than 30.” Hey, it was a pilot study… Case closed. :-D

Nice example :-D
In this case, did you "hit the jackpot" (by the luck CI's were in the borders)?

Kind regards,
Mittyri
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2014-03-14 22:02
(2591 d 08:04 ago)

@ mittyri
Posting: # 12632
Views: 5,081
 

 the basis of T/R in the Pro­tocol

Hi Mittyri,

» […] you state in the protocol phrases like that:
» According to the earlier performed by sponsor study, T/R ratio could be assumed as 0.92%.

Something like that.

» How could I prove to the assessor that "pilot" study is really performed? As a sponsor I can write that I performed the study in another market, but assessment report isn't published (no links)

I never was a sponsor. When I worked for a CRO and now as a consultant I refer only to a study which was really performed (i.e., I have the synopsis / report or at least the data on my desk) and not some product of wet fan­ta­sies. I was never asked (neither by an IEC or an authority) to prove that the pilot was not a fake. However, would be easy to satisfy them.
If a sponsor gives me a number in an e-mail – and doesn’t want to show me the data – I immediately quit working for him/her/it.

» » Once the Austrian agency AGES did not like the sample size we stated in the protocol: “It is not acceptable to perform the study in only 16 sub­jects when most of the BE studies of XYZ are performed in more than 30.” Hey, it was a pilot study…
»
» Nice example :-D
» In this case, did you "hit the jackpot" (by the luck CI's were in the borders)?

Can’t remember. But it was not the purpose of the study to show BE. There are many reasons to perform a pilot study (check the design, bioanalytics, the T/R, CV, …) and I’m positive that we stated all of them in the protocol. Prob­ably the reviewer had such a “chain of command” in his mind:
BE-study ⇒ drug XYZ ⇒ most studies I have seen were performed in 32–56 subjects ⇒ the CRO wants to get a lucky punch with 16 ⇒ reject the protocol.
In Austria protocols are not approved (written stuff). The system works the other way ’round: You submit the protocol and receive a statement of receipt. This date starts the clock. If you don’t get an objection letter within four weeks, you can start the study.
It took me a three-minute phone-call to sort the “misunderstanding about the sample size” out. The reviewer suggested to throw his letter into the next wastebasket, which I did.

Dif-tor heh smusma 🖖
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
Activity
 Admin contact
21,419 posts in 4,475 threads, 1,510 registered users;
online 12 (0 registered, 12 guests [including 6 identified bots]).
Forum time: Sunday 06:07 UTC (Europe/Vienna)

Nothing fails like success because you do not learn anything from it.
The only thing we ever learn from is failure.
Success only confirms our superstitions.    Kenneth E. Boulding

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