Bioequivalence and Bioavailability Forum • Point Estimate consideration in sample size estimate

kimhuang
☆

China,
2025-10-09 09:21
(252 d 17:49 ago)

(edited on 2025-10-09 09:51)
Posting: # 24449
Views: 3,451

Point Estimate consideration in sample size estimate [Power / Sample Size]

Dear Helmut:
Can we use 10% formulation difference or more for sample size estimate. I have a failed BE study and now we want to plan another BE study with the sample size based on the mean ratio (T/R)=112% (from the failed BE study). Is it appropriate to consider this much difference. In FDA guidance "Statistical Approaches to Establishing Bioequivalence" they have written the below line in sample size consideration section.
Sample sizes for average BE should be obtained using published formulas. Sample sizes for population and individual BE should be based on simulated data. The simulations should be conducted using a default situation allowing the two formulations to vary as much as 5% in average BA with equal variances and certain magnitude of subject-by-formulation interaction. The study should have 80 or 90% power to conclude BE between these two formulations. Sample size also depends on the magnitude of variability and the design of the study. Variance estimates to determine the number of subjects for a specific drug can be obtained from the biomedical literature and/or pilot studies.

Looking forward to your reply!
Best Regards!

Helmut
★★★

Vienna, Austria,
2025-10-14 14:33
(247 d 12:37 ago)

@ kimhuang
Posting: # 24452
Views: 2,910

Point Estimate consideration in sample size estimate

Post reply

❝ Dear Helmut:

↑↑↑↑↑↑ Not interested in the opinion of other members?

❝ Can we use 10% formulation difference or more for sample size estimate.

Sure.

❝ I have a failed BE study and now we want to plan another BE study with the sample size based on the mean ratio (T/R)=112% (from the failed BE study). Is it appropriate to consider this much difference.

Of course, it is. The estimate you got from your own study is better than any assumption you might have used in planning it. However, power – and thus, the sample size – is extremely sensitive to the T/R-ratio (see the examples in this article). It would be risky to assume that in the next study the T/R-ratio will be exactly 112% again. Some statisticians call that the Carved-in-Stone approach. If it deviates even more from 100% you might fail again! On the other hand, if it is ‘better’ you invested more but have an even higher chance of passing than planned.

Don’t follow guidelines blindly (even if from the FDA).

—
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

kimhuang
☆

China,
2025-10-15 02:53
(247 d 00:17 ago)

@ Helmut
Posting: # 24454
Views: 2,865

Point Estimate consideration in sample size estimate

Post reply

Dear Helmut,

Thank you very much for your response and kind reminder. I am still interested in the opinions of other experts. :-D

I still have a concern: when using GMR = 1.1 for sample size estimation, the actual GMR is 1.13 and 90% confidence interval is still within the range of 0.8 – 1.25. If the GMR of other generic drugs is within the ±5% difference to the reference, will the FDA approve product marketing? Is there any risk of ANDA’s review? Is it possible to provide examples where GMR is < 0.9 or > 1.1 and the product still be approved by FDA or other regulator?

Best Regards!

Edit: Full quote removed. Please delete everything from the text of the original poster which is not necessary in understanding your answer; see also this post #5! [Helmut]

Helmut
★★★

Vienna, Austria,
2025-10-15 11:32
(246 d 15:38 ago)

@ kimhuang
Posting: # 24455
Views: 2,924

BE = CI within acceptance range (PE is irrelevant)

Post reply

Hi kimhuang,

❝ I still have a concern: when using GMR = 1.1 for sample size estimation, the actual GMR is 1.13 and 90% confidence interval is still within the range of 0.8 – 1.25.

There are three assumptions in sample size estimation (apart from level of the test α and the acceptance range, which are fixed):

The T/R-ratio
The CV
The anticipated dropout-rate (the estimated sample size should be increased accordingly)

Since you passed BE in the study (the confidence interval lied entirely within the acceptance range), it means that you were lucky. The ‘worse’ PE was compensated by a lower CV and/or more eligible subjects. That’s rare because power is most sensitive to the T/R-ratio but still possible.

»Being lucky is not a crime.« (ElMaestro 2014)

I’m not a friend of post hoc (a posteriori, retrospective) power (see this article why).
However, if you calculate it, you will find that it was possibly lower than what you targeted in sample size estimation. Nevertheless, it was > 50%. Otherwise, the study would have failed.

❝ If the GMR of other generic drugs is within the ±5% difference to the reference, will the FDA approve product marketing? Is there any risk of ANDA’s review?

No (twice). The FDA assesses your study and don’t compare it to others. For decades bioequivalence is based on the confidence interval inclusion approach. Nothing else.*

❝ Is it possible to provide examples where GMR is < 0.9 or > 1.1 and the product still be approved by FDA or other regulator?

Possible yes. Feasible no. You would have to check – thousands of – ANDAs to find examples… Good luck.

Only in reference-scaling of highly variable drugs the PE has to – additionally – lie within 0.80 – 1.25.

—
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, 2025-10-15 12:07 (246 d 15:03 ago) @ Helmut Posting: # 24456 Views: 2,826	BE = CI within acceptance range (PE is irrelevant) Post reply
	Hi kimhuang, here's an example, I think: https://www.fda.gov/media/166738/download Read carefully, sometimes a sentence needs to be read more than once before a meaning transpires – so it is at least for me — Pass or fail! ElMaestro

kimhuang
☆

China,
2025-10-16 11:37
(245 d 15:33 ago)

@ ElMaestro
Posting: # 24462
Views: 2,768

BE = CI within acceptance range (PE is irrelevant)

Post reply

Dear ElMaestro,

❝ here's an example, I think:

❝ https://www.fda.gov/media/166738/download

Thank you very much for sharing this case. It has just completely removed my concerns. :-D

❝ Read carefully, sometimes a sentence needs to be read more than once before a meaning transpires – so it is at least for me :-)

is this sentence "Reformulate the proposed product and conduct additional relative bioavailability studies to demonstrate the bioequivalence to the listed drug or conduct a clinical study to support the effective and safe use of the proposed atorvastatin oral suspension product." meaningful? :-D

Best Regards!

kimhuang ☆ China, 2025-10-16 10:26 (245 d 16:44 ago) @ Helmut Posting: # 24461 Views: 2,770	BE = CI within acceptance range (PE is irrelevant) Post reply
	Dear Helmut, Got it, Thank you for your all nice responses, I have no other questions! Best Regards!