Log in
Register|
Noha Abdeljawad ☆ Cairo, Egypt, 2012-11-27 09:18 (4958 d 03:24 ago) Posting: # 9598 Views: 27,332 |
|
|
Dear All, In the bioequivalence center I work in, we need to carry out a BE study for two generic drugs containing the same active ingridient in the same strength from two different manufacturers. Since they both require using the same reference drug, we thought we'd do one crossover study of 3 phases, with one week (washout period) between each phase and the next. On each phase, each of the volunteers took test1 (A), test2 (B) or reference (C). By the end of the study, each of the three products will be adminstered to each volunteer. Bioequivalence assessment will be based on comparing PK parameters of A with C and then of B with C. Is it ok to use this study design? How will this affect ANOVA analysis? Is this design referred to as two-way or three-way? Thanks — Best Regards Noha Abdeljawad |
|
d_labes ★★★ Berlin, Germany, 2012-11-27 10:51 (4958 d 01:51 ago) (edited on 2012-11-27 15:39) @ Noha Abdeljawad Posting: # 9599 Views: 26,112 |
|
|
Dear Noha, ❝ Is it ok to use this study design? Yes. Why not? For issues of which sequences to use see this thread and try these searches: 1, 2. Eventually you have to consider multiplicity issues. See the reference given in this post. ❝ How will this affect ANOVA analysis? The ANOVA model is the same as for the classical 2x2 cross-over. But implicit the assumption of the equal intra-subject variability for all three study drugs/formulations has to be taken. The mighty Oracle EMA has its own belief how to evaluate such studies. See the EMA guideline page 14 under Subject accountability. ❝ Is this design referred to as two-way or three-way? In a sloppy language: Yes for the second choice. Exactly: None of both. The terms 'two-way' or 'three-way' names ANOVA with two or three factors in the model, usually crossed. It would be better to call these designs 3-treatment-3(or6)-sequence-3-period design or shortly 3x3x3 or 3x6x3. After reading the posts of ElMaestro and Dan I must confess that I simply overlooked the sentence "from two different manufacturers". If the study is really aimed for two sponsors: This raises doubts for using a 3x3x3 or 3x6x3 crossover. — Regards, Detlew |
|
Noha Abdeljawad ☆ Cairo, Egypt, 2012-11-29 08:43 (4956 d 03:59 ago) @ d_labes Posting: # 9628 Views: 24,708 |
|
|
Thanks Detlew Labes — Best Regards Noha Abdeljawad |
|
ElMaestro ★★★ Denmark, 2012-11-27 13:10 (4957 d 23:32 ago) @ Noha Abdeljawad Posting: # 9601 Views: 25,380 |
|
|
Hi Noha, ❝ In the bioequivalence center I work in, we need to carry out a BE study for two generic drugs containing the same active ingridient in the same strength from two different manufacturers. Since they both require using the same reference drug, we thought we'd do one crossover study of 3 phases, with one week (washout period) between each phase and the next. If I get this right I think this could easily lead to a messy situation re. e.g. confidentiality and interpretation of e.g. ICH E6. Traditionally, there is one sponsor + one protocol + one study report. Here if I get it right we may have 2 sponsors, one protocol and ... what... two study reports (one for each sponsor) due to confidentiality... or one study report in its entirety presented to both sponsors? What about responsibility? Traditionally shared between Sponsor and PI, but what does that mean here? Who pays if a subject gets in trouble after a ref. product adm. in period 3? What if a post-study audit identifies an issue related to test product 1 - what does that imply to data generated with test product 2? — Pass or fail! ElMaestro |
|
Dr_Dan ★★ Germany, 2012-11-27 15:32 (4957 d 21:10 ago) @ ElMaestro Posting: # 9606 Views: 25,222 |
|
|
Hi Noha I totally agree with ElMaestro. To my experience it is always a bad idea to run a pivotal study with more arms than just test and reference. This is what an assessor wants to see. Even if CPMP/QWP/EWP/1401/98 Rev. 1 recommends that "In studies with more than two treatment arms e.g. a three period study including two references, one from EU and another from USA the analysis for each comparison should be conducted excluding the data from the treatments that are not relevant for the comparison in question." it is not clear how to evaluate the study. Two treatments but 3 periods? How to handle this problem? So the easiest way is to stick to an ordinary two way cross-over.... I hope this helps Kind regards Dan — Kind regards and have a nice day Dr_Dan |
|
Noha Abdeljawad ☆ Cairo, Egypt, 2012-11-29 08:52 (4956 d 03:50 ago) @ ElMaestro Posting: # 9629 Views: 24,707 |
|
|
Thanks ElMaestro & Dr Dan — Best Regards Noha Abdeljawad |
|
jag009 ★★★ NJ, 2012-11-28 22:26 (4956 d 14:16 ago) @ Noha Abdeljawad Posting: # 9624 Views: 24,666 |
|
|
Dear Noha, ❝ Is it ok to use this study design? Yes I just finished running a 3 test vs 1 reference submission. FDA ok'd. ❝ How will this affect ANOVA analysis? This the part you need to be aware of --> Common Variance. Refer to a discussion I had with Helmut and folks here a few months. There is a potential danger if the product (or products) barely passes the 90% CI. ❝ Is this design referred to as two-way or three-way? 3-way study. John |
|
Helmut ★★★   Vienna, Austria, 2012-11-30 15:04 (4954 d 21:38 ago) @ jag009 Posting: # 9644 Views: 24,716 |
|
|
Hi John! ❝ Yes I just finished running a 3 test vs 1 reference submission. FDA ok'd. Out of curiosity: Did FDA want you to adjust α to deal with multiplicity (>90% CIs: Bonferroni, Holmes, Hochberg,  )?
— 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, 2012-11-30 16:16 (4954 d 20:26 ago) @ Helmut Posting: # 9645 Views: 24,826 |
|
|
Hi Helmut, ❝ Out of curiosity: Did FDA want you to adjust α to deal with multiplicity (>90% CIs: Bonferroni, Holmes, Hochberg, ❝ ❝ Nope. We asked like 4 FDA people (all the way up to the big one) and they even said don't worry about the common variance (I was surprised). Then again I didn't ask the question, my non-pk colleagues who went to the FDA meeting did (and we also asked them via email). FDA ok'd meaning they have no problem with the 4-way design. Of course, I share the same view with you regarding the common variance issue due to its role in the computation of the 90% CI. But in our case it is not an issue because we nailed the 90% CIs. John |
|
Helmut ★★★ Vienna, Austria, 2012-11-30 17:12 (4954 d 19:30 ago) @ jag009 Posting: # 9648 Views: 24,759 |
|
|
Hi John, I’m less concerned with the common variance, but with multiplicity. If you are performing three simultaneous tests at 0.05 the combined risk will be 1 – (1 – 0.05)3 = 0.1426. ❝ […] But in our case it is not an issue because we nailed the 90% CIs. Bonferroni (well, this is the most conservative one) would require 1 – 2 × 0.05/3 = 96.67% CIs in order to keep the overall risk <0.05: 1 – (1 – 0.05/3)3 = 0.0492.  — 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, 2012-11-30 23:47 (4954 d 12:55 ago) @ Helmut Posting: # 9649 Views: 24,768 |
|
|
Hi Helmut, Zipped some absinthe while refreshing my PK memory this afternoon.  My old boss filed a 4-way nifedpine study and if I recall correctly the statistic was straight forward and FDA didn't come back with questions on multiplicity. I recalled we filed the product back in 1997 (Good old days when I was hired as an university summer intern). John |
|
Helmut ★★★ Vienna, Austria, 2012-12-01 03:00 (4954 d 09:42 ago) @ jag009 Posting: # 9651 Views: 24,783 |
|
|
Hi John, ❝ Zipped some absinthe while refreshing my PK memory this afternoon. Nice stuff. Prefer it in the late evening. ❝ My old boss filed a 4-way nifedpine study and if I recall correctly the statistic was straight forward and FDA didn't come back with questions on multiplicity. I recalled we filed the product back in 1997 (Good old days when I was hired as an university summer intern). Yeah, the mid-1990s – the golden age of MR nifedipine. Analysed ~20,000 samples if I recall it right. — 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, 2012-12-03 22:53 (4951 d 13:49 ago) @ Helmut Posting: # 9664 Views: 24,967 |
|
|
Helmut, ❝ Yeah, the mid-1990s – the golden age of MR nifedipine. Analysed ~20,000 samples if I recall it right. Yup, before the introduction of replicate study design. Sample size =infinity sometime  |
|
ElMaestro ★★★ Denmark, 2012-11-30 16:24 (4954 d 20:18 ago) @ Helmut Posting: # 9646 Views: 25,044 |
|
|
Hi Helmut, ❝ Out of curiosity: Did FDA want you to adjust α to deal with multiplicity (>90% CIs: Bonferroni, Holmes, Hochberg, Some personal observations from the bag of EU/US/other memories:
— Pass or fail! ElMaestro |
|
Helmut ★★★ Vienna, Austria, 2012-11-30 16:52 (4954 d 19:50 ago) @ ElMaestro Posting: # 9647 Views: 24,633 |
|
|
Hi Anders!
— 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 |
|
martin ★★ Austria, 2012-12-03 10:24 (4952 d 02:18 ago) @ Helmut Posting: # 9658 Views: 24,565 |
|
|
dear all! you have several strategies to accound for multiplicity (e.g. Bonferroni, fixed sequence testing procedure with a-priori ordered hypothesis, …). However, I would suggest to have a close look at the intersection-union princple: If all calculated two-sided 90% CI are within the acceptance range – the type I error is controlled strongly. Please refer to Wellek (2010; page 219) for more detail. hope this helps Martin PS.: see also this post IUT principle |
|
kumarnaidu ★ Mumbai, India, 2013-12-12 12:03 (4578 d 00:39 ago) @ ElMaestro Posting: # 12045 Views: 22,522 |
|
|
Hi all, Here I found the same situation. I got one protocol for comparative BA study with two references (one is prodrug and other is metabolite) and one test. Here people suggesting 3 comparisons T Vs R1, T Vs R2 and R1 vs R2. Is multiplicity adjustment required for this kind of studies? — Kumar Naidu |
|
Helmut ★★★ Vienna, Austria, 2013-12-12 16:27 (4577 d 20:15 ago) @ kumarnaidu Posting: # 12050 Views: 22,466 |
|
|
Hi Kumar, ❝ I got one protocol for comparative BA study with two references (one is prodrug and other is metabolite) and one test. Which API is used in the test product? ❝ Here people suggesting 3 comparisons T Vs R1, T Vs R2 and R1 vs R2. Is multiplicity adjustment required for this kind of studies? BE is defined as a comparative BA with the same chemical entity (including different salts, esters). Depending one the API in the test, IMHO only T vs. R1 or T vs. R2 makes sense. Results of other comparisons are only “nice to know”. — 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 |
|
Noha Abdeljawad ☆ Cairo, Egypt, 2012-12-03 09:29 (4952 d 03:13 ago) @ jag009 Posting: # 9657 Views: 24,515 |
|
|
Dear John, Is it a three way if two separate statistical analysis will be carried out; once with test1 against the reference and another with test2 against the reference? What does the word way refer to exactly? Thanks in advance — Best Regards Noha Abdeljawad |
|
jag009 ★★★ NJ, 2012-12-03 16:35 (4951 d 20:07 ago) (edited on 2012-12-03 21:52) @ Noha Abdeljawad Posting: # 9662 Views: 24,439 |
|
|
Hi Noha, ❝ Is it a three way if two separate statistical analysis will be carried out; once with test1 against the reference and another with test2 against the reference? No, you analyze everything in one stat analysis and do not break down into 2 separate analyses. ❝ What does the word way refer to exactly? "way" as in two-way, three-way, four-way. It identifies the # of treatment arms. John |
|
d_labes ★★★ Berlin, Germany, 2012-12-04 10:25 (4951 d 02:17 ago) @ jag009 Posting: # 9667 Views: 24,560 |
|
|
Dear John, ❝ "way" as in two-way, three-way, four-way. It identifies the # of treatment arms. As said above: The X-way nomenclature among statisticians is usually something different. It names ANOVA models containing 2, 3 ... or X factors in the model, usually crossed and each factor with at least 2 or more values. A 2-, 3-way crossover design on the other hand is some totally different thing.  "3-way crossover design" to see.— Regards, Detlew |
|
kvgreddy06 ☆ India, 2013-12-19 16:32 (4570 d 20:10 ago) @ d_labes Posting: # 12079 Views: 22,313 |
|
|
Dear All, kindly provide your views: I want to Design a study: four way crossover study, three treatment, four period. Test Product: T1, Test Product: T2 and Reference Product R (twice) which type of sequence is better, how can I explore statistical comparision, I am at learing stage, kindly provide in detail... Thank... venu |
|
kvgreddy06 ☆ India, 2013-12-30 17:48 (4559 d 18:54 ago) @ kvgreddy06 Posting: # 12114 Views: 22,194 |
|
|
Dear All, kindly provide which one is possible and advantages, disadvantages. design, Test product (A) Test Product (B) Reference Product (C) here i am using reference product twice, * : Three treatment, four period, four sequences crossover reference replicated ![[image]](img/uploaded/image57.jpg) **: Three treatment, four period Three sequences, crossover reference replicated ![[image]](img/uploaded/image58.jpg) Among those which one is suitable, or any other one one of my friend suggesting **: study is a reference replicate study with two test treatment and the analysis will be done separately for both the test product by using the sequences ACC, CAC and CCA for test product A and BCC, CBC and CCB for test product B which is derived from the original sequences ABCC ,CABC & CCAB . However the derived sequences met the criteria of classical reference replicate study design and sequence (TRR, RRT and RTR) and the same number of sequence order will not be able to generate from the four sequences. Thanku.. |
|
intuitivepharma ☆ India, 2013-12-31 07:44 (4559 d 04:58 ago) @ kvgreddy06 Posting: # 12116 Views: 22,185 |
|
|
Dear kvgreddy06, IMHO a better alternative to the three treatment, four period, four sequences crossover reference replicated design you have proposed would be RRT1T2 T1T2RR T2T1RR RRT2T1 But, I would stick to the below mentioned sequence if I were you RT1RT2 T1RT2R T2RT1R RT2RT1 — Thanks & Regards, IP. |
|
kumarnaidu ★ Mumbai, India, 2013-12-31 09:24 (4559 d 03:18 ago) @ intuitivepharma Posting: # 12117 Views: 22,225 |
|
|
Hello everyone, ❝ But, I would stick to the below mentioned sequence if I were you ❝ ❝ RT1RT2 ❝ T1RT2R ❝ T2RT1R ❝ RT2RT1 I am little bit confused with the above sequence. I think there might be some imbalance can occur for the above sequence sequence RT1RT2 for T1 - RT1R T1RT2R for T1 - T1RR T2RT1R for T1 - RT1R RT2RT1 for T1 - RRT1 Here we will get RT1R twice ??. According to me we should go with the three sequence. Can we keep the below sequence. T1T2RR for T1- T1RR and for T2-T2RR RT1T2R for T1- RT1R and for T2-RT2R RRT1T2 for T1- RRT1 and for T2-RRT2 Let's take the opinion from other members. — Kumar Naidu |
|
Helmut ★★★ Vienna, Austria, 2013-12-31 16:02 (4558 d 20:40 ago) @ kumarnaidu Posting: # 12120 Views: 22,278 |
|
|
Hi randomizers, why do you want to perform such a design? Which is your “target” regulation? Do you want to market both T1 and T2 or are the “candidate formulations” and you will drop one of them? In the latter case I would suggest pilot studies and performing the pivotal with the “survivor” in a more conventional design. If you want to submit the study to the FDA, keep RTR (“non-recommended in vivo study”) in mind! To get a balanced design is rather difficult. The conditions are:
Which model/code do you intend to use?
— 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 |
|
kvgreddy06 ☆ India, 2014-01-03 15:46 (4555 d 20:56 ago) @ Helmut Posting: # 12132 Views: 21,953 |
|
|
Dear All, Thanku for your valuble opinoins, here we are planing to conduct pilot study first, based on the results of pilot study we can select best test product among two test products. |
|
kumarnaidu ★ Mumbai, India, 2014-12-24 13:56 (4200 d 22:46 ago) @ Helmut Posting: # 14162 Views: 20,356 |
|
|
Hi all, can we keep this set of sequences for such designs.
— Kumar Naidu |
|
Helmut ★★★ Vienna, Austria, 2014-12-24 17:10 (4200 d 19:32 ago) @ kumarnaidu Posting: # 14164 Views: 20,358 |
|
|
Hi Kumar, ❝ can we keep this set of sequences for such designs. ❝ ❝ ❝ ❝ ❝ ❝ ❝ I asked you already almost one year ago about the purpose of your study. The question ist still open… Your design is extremely imbalanced for period effects:
— 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 |
|
kumarnaidu ★ Mumbai, India, 2014-12-26 06:38 (4199 d 06:04 ago) @ Helmut Posting: # 14176 Views: 20,388 |
|
|
Hi helmut, Like kvgreddy I also encountered with same situation. We want to study the two test product and as per the literature the reference is HVD and some PK experts from CRO suggesting this design for pilot BE study. I suggested them to perform 3-way study with T1 T2 R as in pilot study we only want to know the T/R ratio. and the best test product will be used further for the pivotal (3way or 4way reference replicate) study. But my people also think that 4way pilot study with reference replicate will give them the correct idea. CRO suggested the below sequence T1 T2 R RT2 R R T1R R T1 T2R T1 T2 RHere the condition 2 is satisfied but not 3. R : 2×P1, 2×P2, 2×P3, 2×P4As you rightly said in your previous post that full balance could only be achieved if all 4! = 24 permutations are used which is imposible for the pilot study with 24 subjects. Thanks — Kumar Naidu |
|
Helmut ★★★ Vienna, Austria, 2014-12-27 16:52 (4197 d 19:50 ago) @ kumarnaidu Posting: # 14180 Views: 20,114 |
|
|
Hi Kumar, I think that all these “designs” lead to trouble. I would suggest two fully replicated three-period designs.
A fully replicated design gives you not only the variability of R but also of T. If the variance of T is lower than the one R you will get a reward in the pivotal study (smaller sample size). Decision(s):
— 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 |
Ing. Helmut Schütz