Bioequivalence and Bioavailability Forum

Rakesh_20 ☆ India, 2015-01-22 11:26 (3739 d 14:57 ago) Posting: # 14309 Views: 7,115	Two-Stage Design with Fully replicate for CANADA? [Two-Stage / GS Designs] Post reply
	Dear Sir, I have following concern : For Canadian submission, can we consider the Fully replicated crossover design with two-stage sequential approach for BE study? and how TPD will look at this? Thanks and Regards, Rakesh

ElMaestro ★★★ Denmark, 2015-01-22 11:56 (3739 d 14:27 ago) @ Rakesh_20 Posting: # 14310 Views: 6,064	Two-Stage Design with Fully replicate for Canada Post reply
	Hi Rakesh, I am not aware of anyone ever having done it. The theoretical framework is still missing, at least in the public domain, and this is probably because the computational requirements for the necessary simulations could be a serious bottleneck. If you approach regulators and ask about it then I would be very happy to hear about the outcome — Pass or fail! ElMaestro

Helmut
★★★

Vienna, Austria,
2015-01-22 13:37
(3739 d 12:46 ago)

@ Rakesh_20
Posting: # 14311
Views: 5,922

RSABE/TSD-combo

Post reply

Dear Madam,

why are you shouting at us?

❝ For Canadian submission, can we consider the Fully replicated crossover design with two-stage sequential approach for BE study?

ElMaestro correctly wrote above that no method is published yet. Note that currently HC does not accept reference-scaling. In June 2014 the “Scientific Advisory Committee on Pharmaceutical Sciences and Clinical Pharmacology” recommended FDA’s method for implementation but is considering other approaches as well. Let’s see.
If you want to explore RSABE in a TSD with current hard-/software finding and validating an adjusted α would take ~six years. If you believe in Moore’s law (processing power doubles every 18–24 months) and you buy always the fastest available machine it would still take ~1½ years to get a result. Unless you find someone who is able to write some fancy C- or FORTRAN-code for a massive parallel supercomputer (which will cost you a fortune) I don’t see a realistic way out.

If you want to run a fully replicate design for conventional ABE (no scaling) IMHO that should be doable. Start your own simulations.

❝ and how TPD will look at this?

With great interest. Report back their response.

—
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

Rakesh_20 ☆ India, 2015-01-23 07:32 (3738 d 18:51 ago) @ Helmut Posting: # 14313 Views: 5,860	Fully replicated-conventional ABE/TSD-combo Post reply
	Dear Both, Thank you so much for the reply. ❝ If you want to run a fully replicate design for conventional ABE (no scaling) IMHO that should be doable. Start your own simulations. So, if we consider Fully replicated crossover design-conventional ABE with two-stage sequential approach for BE study then, is it acceptable to canadian submission? Thanks and Regards, Rakesh

Helmut
★★★

Vienna, Austria,
2015-01-23 13:41
(3738 d 12:43 ago)

@ Rakesh_20
Posting: # 14316
Views: 5,840

Simulations mandatory

Post reply

Hi Rakesh,

❝ So, if we consider Fully replicated crossover design-conventional ABE with two-stage sequential approach for BE study then, is it acceptable to canadian submission?

Maybe we were not clear.

Nothing is stated in HC’s guidance, but: “Alternate approaches to the principles and practices described in this document may be acceptable provided they are supported by adequate justification. Alternate approaches should be discussed in advance with the relevant program area to avoid the possible finding that applicable statutory or regulatory requirements have not been met.”
(my emphases)
Since nothing is published, you have to set up Monte Carlo simulations for a matrix of possible combinations of stage 1 sample sizes (n₁) and CVs in order to find and validate a suitable adjusted α (i.e., which maintains the overall type I error at 5%). Due to the slow convergence you have to simulate 10⁶ BE-studies at a T/R-ratio of 1.25 (for α) and 10⁵ at the expected T/R-ratio (for power) for every combination of n₁ and CV. Keep in mind that in a fully replicated design CV_WR not necessarily equals CV_WT.

Good luck.

—
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

Two-Stage Design with Fully replicate for CANADA? [Two-Stage / GS Designs]

Two-Stage Design with Fully replicate for Canada

RSABE/TSD-combo

Fully replicated-conventional ABE/TSD-combo

Simulations mandatory