Bioequivalence and Bioavailability Forum • NTI drug issue

yuvrajkatkar
★

Pune, Maharashtra (India),
2010-04-15 16:44
(5910 d 12:59 ago)

Posting: # 5149
Views: 6,393

NTI drug issue [Power / Sample Size]

Dear all,

Please see the below results


Dependent       Ratio[%Ref]  CI_90_Lower  CI_90_Upper  ICV(%)

Ln(AUCINF_obs)  95.80        92.57        99.15        11.58%

Ln(AUClast)     96.23        92.78        99.82        12.33%

Ln(Cmax)        90.20        86.02        94.58        16.04%

The 90% CI for Ln(Cmax) is not within the limits 90% to 111.11%. As per the EMEA guideline on Investigation of BE study for the BE criteria of NTI drugs.

The maximum Ratio[%Ref]= 90.20% and corresponding Intrasubject CV= 16.04% for Ln(Cmax).

If we take T/R ratio = 90.20% and Intrasubject CV= 16.04% then sample size will be greater than 20000 to get 90% confidence intervals of log transformed Cmax, AUC0-t and AUC0-inf for data within 90.00 - 111.11 % to achieve 80 % power at 5% level of significance.(Ref. of formula: Shein-Chung Chow,Jan Shao (sample size calculation in clinical reserch, page 31, 32 ). Instead of ln (1.25)=0.20, I have taken ln(1.11)=0.10.

Is it right way to choose sample size for the next study

If we take T/R ratio = 95% to 105% and Intrasubject CV= 16.04% then sample size will be 109.

Please suggest me.

Edit: Table coded, UTF-8 characters removed. [Helmut]

—
Best Regards,
Yuvraj

Helmut
★★★

Vienna, Austria,
2010-04-15 17:43
(5910 d 12:00 ago)

@ yuvrajkatkar
Posting: # 5150
Views: 5,380

Warning

Post reply

Dear Yuvraj!

I'm bored of editing your posts almost all the time. Don't use tabs, UTF-8 characters, sloppy spelling, etc. You expect answers. Fine. Use the preview before posting.
If you don't comply with the forum's Policy, your account will be blocked.

—
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,
2010-04-16 12:11
(5909 d 17:32 ago)

@ yuvrajkatkar
Posting: # 5152
Views: 5,169

Googol

Post reply

Dear yuvrajkatkar,

❝ The maximum Ratio[%Ref]= 90.20% and corresponding Intrasubject CV= 16.04% for Ln(Cmax).

Is the minimum ratio but maximum CV.

❝ If we take T/R ratio = 90.20% and Intrasubject CV= 16.04% then sample size will be greater than 20000 to get 90% confidence intervals of log-transformed Cmax ...

Exactly 63750 (power=80.0003%) using my super code. Approximate Googol :cool:

!
I think this is a very great opportunity to earn a big lot of money and have occupation until sunset of your years. If you find someone to spend ... :-D

❝ Instead of ln(1.25)=0.20, I have taken ln(1.11)=0.10.

ln(1.25) = 0.2231436, ln(1.11)=0.1043600 on my machine (Intel® Pentium 4, 1 GB RAM; Win XP Professional, SAS9.2 or R10.0).
BE upper acceptance limits with your numbers: exp(0.2)=122.14% and exp(0.1)=110.52%.

In the light of the FDA guidance and the new EMA guidance both asking for CI within 80.00 to 125.00 % it is IMHO not a good idea to use only 1 decimal. :no:

❝ Is it right way to choose sample size for the next study

Since I guess your results came from a BE study with approx. 38 subjects why do you expect that your point estimates are unreliable?
With such a point estimate you never achieve BE within the acceptance range 90 ... 111.11%
So the right suggestion for a next study is: reformulate is the question.

—
Regards,

Detlew