Log in
Register|
jag009 ★★★ NJ, 2012-06-11 17:01 (4329 d 12:20 ago) Posting: # 8683 Views: 9,737 |
|
|
Hi everyone, I have a question. If there is a FDA a draft BE guidance for a product specifing the type of 2 way crossover studies that need to be conducted:
Would that be doable? The only issue I can see (beside the cost) is the ethical issue because we will (obviously) draw more blood samples per subject, and may have a larger sample size due to potential dropouts since the length of the study is doubled. I believe FDA allows the use of an alternative approach if the approach satisfies the requirements of the applicable statutes and regulations. Do you think the agency will have an issue with the approach? Thanks John |
|
Helmut ★★★   Vienna, Austria, 2012-06-11 18:23 (4329 d 10:59 ago) @ jag009 Posting: # 8684 Views: 8,847 |
|
|
Hi John! ❝ I have a question. If there is a FDA a draft BE guidance for a product specifing the type of 2 way crossover studies that need to be conducted: ❝ ❝ 1. 2-way crossover (Test vs Reference) fasting ❝ 2. 2-way crossover (Test vs Reference) fed Sure, many. One example is our baby methylphenidate XR (+sprinkled, BTW). ❝ What if we end up running a 4-way crossover involving 3 test formulations vs 1 Reference for each study? Everything will be the same except for 2 extra arms. The reason we are doing this is because we want to bypass the pilot study. ❝ ❝ Would that be doable? In principle yes. ❝ The only issue I can see (beside the cost) is the ethical issue because we will (obviously) draw more blood samples per subject, and may have a larger sample size due to potential dropouts since the length of the study is doubled. Right. But there’s another problem: In this type of higher-order XOver you obtain a common (pooled) variance from treatment variances. What if one of the formulations not only shows a PE far away from 100% (which would lead to dropping it from development), but performs ‘bad’ in terms of the CV? Especially if MR formulations are concerned, the differences sometimes are large. This formulation will inflate your CIs and require a larger sample size compared to simple XOvers. We had a similar debate in the EU where EMA in their drafted GL wanted to see a four-way XOver T and R (fed/fasting). In the final GL such a design is not required. ❝ I believe FDA allows the use of an alternative approach if the approach satisfies the requirements of the applicable statutes and regulations. Sure. ❝ Do you think the agency will have an issue with the approach? No idea.  *
— Dif-tor heh smusma 🖖🏼 Довге життя Україна! ![[image]](https://static.bebac.at/pics/Blue_and_yellow_ribbon_UA.png) Helmut Schütz ![[image]](https://static.bebac.at/img/CC by.png) The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
jag009 ★★★ NJ, 2012-06-11 19:25 (4329 d 09:56 ago) @ Helmut Posting: # 8685 Views: 8,741 |
|
|
Hi Helmut, ❝ Right. But there’s another problem: In this type of higher-order XOver you obtain a common (pooled) variance from treatment variances. What if one of the formulations not only shows a PE far away from 100% (which would lead to dropping it from development), but performs ‘bad’ in terms of the CV? Especially if MR formulations are concerned, the differences sometimes are large. This formulation will inflate your CIs and require a larger sample size compared to simple XOvers. We had a similar debate in the EU where EMA in their drafted GL wanted to see a four-way XOver T and R (fed/fasting). In the final GL such a design is not required. Agreed. I forgot to say that a larger sample size will also serve to compensate for the widening of the CIs due to potential increase in variations due to formulation performance differences. Now how much of an increase in sample size is the problem... Like you said we don't know how bad 1 or 2 out of the 3 test formulations will perform. |
|
Helmut ★★★ Vienna, Austria, 2012-06-11 19:35 (4329 d 09:46 ago) @ jag009 Posting: # 8686 Views: 8,767 |
|
|
Hi John! ❝ Agreed. I forgot to say that a larger sample size will also serve to compensate for the widening of the CIs due to potential increase in variations due to formulation performance differences. Now how much of an increase in sample size is the problem... Like you said we don't know how bad 1 or 2 out of the 3 test formulations will perform.  Everybody wants to kill two birds with one stone. What you essentially would need is a full adaptive design (adjusting for observed PEs and CVs). This is one of our member’s specialty. Acceptance? No idea.P.S.: Avoid multiple blanks (a reminiscence of the age of typewriters) in your posts. They will be cut down on a single blank – so the aimed effect does not work anyway. — Dif-tor heh smusma 🖖🏼 Довге життя Україна! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
jag009 ★★★ NJ, 2012-06-11 22:03 (4329 d 07:18 ago) @ Helmut Posting: # 8687 Views: 8,703 |
|
|
Hi Helmut ❝ I thought about adaptive design as well but then timing would become an issue since bioanalytical would be involved during the process? The stupid thing is we want to test all and hit the spot with one (hopefully). It is a resource (outsourcing) vs time issue. I am the planner designer executioner (or whatever, not the one with the sickle), higher up gives the call (timeframe wise)  |
|
ElMaestro ★★★ Denmark, 2012-06-11 22:34 (4329 d 06:47 ago) @ jag009 Posting: # 8688 Views: 8,868 |
|
|
Hi Jag009, ❝ I thought about adaptive design as well but then timing would become an issue since bioanalytical would be involved during the process? The stupid thing is we want to test all and hit the spot with one (hopefully). It is a resource (outsourcing) vs time issue. And on a side note I have to say I think this is becoming almost a syndrome now in the pharma industry. Everything is being LEAN'ed into dysfunctionality where management dictates a "yes we can"-attitude which is basically founded in hot air. Another syndrome is the "fight for it"-order that top management issues when studies have gone so pearshaped that noone (except management) believes it can be accepted by the regulatory authority. Now I've said it. It feels good. I will now take my medicine. Sorry. Back into my cage. — Pass or fail! ElMaestro |
|
Helmut ★★★ Vienna, Austria, 2012-06-12 03:03 (4329 d 02:18 ago) @ ElMaestro Posting: # 8689 Views: 8,710 |
|
|
Hi ElMaestro! ❝ Everything is being LEAN'ed into dysfunctionality where management dictates a "yes we can"-attitude which is basically founded in hot air. Another syndrome is the "fight for it"-order that top management issues when studies have gone so pearshaped that noone (except management) believes it can be accepted by the regulatory authority. Well said.  ❝ Now I've said it. It feels good. I will now take my medicine. Sorry. Back into my cage. Are you still “within pharmacological reach”? If ev’rything fails consider a hint of Schützomycin. Happy dreams! — Dif-tor heh smusma 🖖🏼 Довге життя Україна! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
ElMaestro ★★★ Denmark, 2012-06-12 15:22 (4328 d 13:59 ago) @ Helmut Posting: # 8708 Views: 8,712 |
|
|
Hi HS, ❝ Are you still “within pharmacological reach”? If ev’rything fails consider a hint of Schützomycin. Happy dreams! Thank you, good idea, I'll take a ton of them tablets. And perhaps I need a straitjacket also. — Pass or fail! ElMaestro |
|
jag009 ★★★ NJ, 2012-06-12 21:56 (4328 d 07:25 ago) @ Helmut Posting: # 8719 Views: 8,617 |
|
|
Hi Helmut, ❝ Thinking it over I would expect that the FDA will not like this approach. What if it’s the other way ’round? Three tests; T1 and T2 perform bad in terms of their PEs whilst T3 is fine. Now let’s assume that s²WT1=s²WT2=s²WR ≪ s²WT3. The pooled s²W will be ‘dampened’ by the (later dropped) formulations T1 and T2. In other words you would never be able to demonstrate BE of T1 in a 2-way XOver of the same size. Though theses variances are not accessible in a nonreplicate design I think that the FDA wouldn’t like the idea of approving a product with a potential ‘variance-brake’ in the pivotal study. Totally agree. Could I hypothetically strip the two treatments (successful test formulation and the reference) and re-run stats in a two-way crossover fashion to re-evaulate the data? Would that stand a chance with regulatory? Just a thought... Thanks John |
|
jag009 ★★★ NJ, 2012-06-12 22:00 (4328 d 07:21 ago) @ jag009 Posting: # 8720 Views: 8,652 |
|
|
Just out of curiosity, no I am not happy substances..Would any reg agency accept a bioequivalence study involving 1 Test and 2 reference (meaning 2 difference lot numbers) and claim bioequivalence when the test shows BE to only one lot of reference? Someone from the my lab asked this question. I really thought he was  John |
|
Helmut ★★★ Vienna, Austria, 2012-06-12 22:39 (4328 d 06:42 ago) @ jag009 Posting: # 8721 Views: 8,657 |
|
|
Hi John! ❝ Could I hypothetically strip the two treatments (successful test formulation and the reference) and re-run stats in a two-way crossover fashion to re-evaulate the data? See this goody (slides 60–67). The title ‘Don’t try this at home!’ tells it all. You are not the first one having such an idea (see also here). ❝ Would that stand a chance with regulatory? Close to nil – unless you run sufficiently large simulations demonstrating that the patient’s risk is preserved. Not a trivial job. — Dif-tor heh smusma 🖖🏼 Довге життя Україна! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |
|
d_labes ★★★ Berlin, Germany, 2012-06-13 11:11 (4327 d 18:10 ago) @ Helmut Posting: # 8724 Views: 8,568 |
|
|
Hi Helmut! ❝ Close to nil ... What about "In studies with more than two treatment arms (e.g. a three period study including two references, one from EU and another from USA, or a four period study including test and reference in fed and fasted states), the analysis for each comparison should be conducted excluding the data from the treatments that are not relevant for the comparison in question." CPMP/QWP/EWP/1401/98 Rev. 1, page 14 May this eventually imply that in case of BE to the EU Reference but not to the USA Reference the study would pass? IMHO this is one of the 'rationales' behind that crippled evaluation method that EU regulators are not interested in the BE result to the USA Reference. — Regards, Detlew |
|
jag009 ★★★ NJ, 2012-09-11 00:48 (4238 d 04:34 ago) @ d_labes Posting: # 9169 Views: 8,110 |
|
|
Hi d_labes and Helmut, This really bugs me (from EMEA guidance)... "In studies with more than two treatment arms (e.g. a three period study including two references, one from EU and another from USA, or a four period study including test and reference in fed and fasted states), the analysis for each comparison should be conducted excluding the data from the treatments that are not relevant for the comparison in question." I have a dataset containing 4 treatments (from a 4-way 4-treatmens 4-sequence crossover study) and each treatment has different # of subjects (ie: A(T-Fast)=24, B(Ref Fast)=19, C(T-Fed)=22, D(Ref Fed)=24). I tried doing the stats as per the above statment (A vs D, C vs D, C vs A, D vs B) by removing the treatments that are not of interest but SAS gave me an error message for my SAS code estimat1 = Estimate "A - D" Treatment 1 0 0 -1"WARNING: More coefficients than levels specified for effect TREATMENT. Some coefficients will be ignored. NOTE: A vs D is not estimable." I got the same error message for each comparison. Any reason why? Thanks John |
|
d_labes ★★★ Berlin, Germany, 2012-09-11 10:20 (4237 d 19:01 ago) @ jag009 Posting: # 9174 Views: 7,934 |
|
|
Hi John, ❝ This really bugs me (from EMEA guidance)... ❝ ❝ "In studies with more than two treatment arms (e.g. a three period study including two references, one from EU and another from USA, or a four period study including test and reference in fed and fasted states), the analysis for each comparison should be conducted excluding the data from the treatments that are not relevant for the comparison in question." Me too ❝ I have a dataset containing 4 treatments (from a 4-way 4-treatmens 4-sequence crossover study) and each treatment has different # of subjects (ie: A(T-Fast)=24, B(Ref Fast)=19, C(T-Fed)=22, D(Ref Fed)=24). I tried doing the stats as per the above statment (A vs D, C vs D, C vs A, D vs B) by removing the treatments that are not of interest but SAS gave me an error message for my SAS code ❝ ❝ "WARNING: More coefficients than levels specified for effect TREATMENT. Some coefficients will be ignored. NOTE: A vs D is not estimable." I got the same error message for each comparison. ❝ ❝ Any reason why? Have a look at the "Class levels" output of your Proc GLM evaluation. Especially the levels of treatment  .You will notice that you have only 2 treatments in your EMA crippled evaluation (A and D in the data if the comparison A vs. D is concerned). Thus you have to formulate: Estimate "A - D" Treatment 1 -1to get the 90% CIs of of A-D. — Regards, Detlew |
|
jag009 ★★★ NJ, 2012-06-13 18:08 (4327 d 11:13 ago) @ Helmut Posting: # 8728 Views: 8,593 |
|
|
Hi Helmut, ❝ ❝ Could I hypothetically strip the two treatments (successful test formulation and the reference) and re-run stats in a two-way crossover fashion to re-evaulate the data? ❝ (my emphasis) ❝ ❝ See this goody (slides 60–67). The title ‘Don’t try this at home!’ tells it all. You are not the first one having such an idea (see also here). Don't fly over the atlantic and hit me with a hammer on this ... I am just hypothesizing If I do strip the good test and reference arms from such a 4-way crossover study (3 T vs R), assign the 2 treatments to a 2-treatment randomization scheme according to how the two were arranged in the original 4-treatment randomization such as, A, B, C, D --> assign as A D B, C, D, A --> assign as D A C, D, A, B --> assign as D A B, A, C, D --> assign as A D I then re-run the stats to show that the test is bioequivalent to the reference. The sample size doesn't change. Why would the above be "Don't try this at home"? I ran a simulation last night based the data from a successful 2-way crossover study by:
? Thanks John |
Ing. Helmut Schütz