Bioequivalence and Bioavailability Forum 05:33 CET

Main page Policy/Terms of Use Abbreviations Latest Posts

 Log in |  Register |  Search

Jérôme Martin
Junior

France,
2013-03-19 17:42
(edited by Jérôme Martin on 2013-03-20 15:42)

Posting: # 10238
Views: 9,674
 

 R, packages (e.g. PowerTOST) and regulatory issues (e.g.FDA) [R for BE/BA]

I’m preparing a clinical ANDA BE study for the company I work with and for that purpose I’m using R and the package PowerTOST (I’m computing power and sample size and also I use CVfromCI and CVpooled). I plan to perform both the statistical analysis plan (SAP) and the data analysis.

I’m regarding regulatory issues (FDA mainly and/or EMA) and I’m asking if using R and dedicated package (especially PowerTOST) has already be done and accepted at this stage (clinical ANDA)? If it’s routinely done by other statisticians? If R and this package are known from regulators?

In any cases, did you have advice how to manage those regulatory issues (special letter to regulators? presentation of software or data?...) for a clinical ANDA BE trial with R and packages such as PowerTOST?

I’m aware that FDA made recent clear statement about statistical software, but it’s to have your opinions and feedback.
ElMaestro
Hero

Denmark,
2013-03-20 08:11

@ Jérôme Martin
Posting: # 10241
Views: 8,496
 

 R, packages (e.g. PowerTOST) and regulatory issues (e.g.FDA)

Hello Jérôme,

» I’m preparing a clinical ANDA BE study for the company I work with and for that purpose I’m using R and the package PowerTOST (I’m computing power and sample size and also I use CVfromCI and CVpooled). I plan to perform both the statistical analysis plan (SAP) and the data analysis.

» I’m aware that FDA made recent clear statement about statistical software, but it’s to have your opinions and feedback.

Power is usually the applicant's problem so I woudln't expect FDA to reject any dossier just because power has been calculated with a package that the FDA finds dubious. This is of course not to say that the powerTOST package is dubious; in fact this package is anything but.
Unless the package you use may calculate sample size so wrongly that a study is futile (unethical) there should not be too much trouble.

For evaluation of BE in R you need e.g. Yung-Jin Lee's excellent BEAR or some codelines of your own making and then the issue becomes more important. On basis of FDA's message I would say you can certainly validate an R script and use that.

if (3) 4

x=c("Foo", "Bar")
b=data.frame(x)
typeof(b[,1]) ##aha, integer?
b[,1]+1 ##then let me add 1



Best regards,
ElMaestro

"(...) targeted cancer therapies will benefit fewer than 2 percent of the cancer patients they’re aimed at. That reality is often lost on consumers, who are being fed a steady diet of winning anecdotes about miracle cures." New York Times (ed.), June 9, 2018.
Ohlbe
Hero

France,
2013-03-20 10:18

@ ElMaestro
Posting: # 10242
Views: 8,525
 

 R, packages (e.g. PowerTOST) and regulatory issues (e.g.FDA)

Dear Jérôme and El Maestro,

» For evaluation of BE in R you need e.g. Yung-Jin Lee's excellent BEAR or some codelines of your own making and then the issue becomes more important.

Doesn't the FDA systematically repeat the PK and stats calculations in ANDA BE studies ? If this is indeed the case, there is no reason they should raise any difficulty as long as they manage to reproduce your results.

The discussion may unfortunately be somewhat different in Europe, where repeating the stats is not common practice.

Regards
Ohlbe

Regards
Ohlbe
Jérôme Martin
Junior

France,
2013-03-20 10:32
(edited by Jérôme Martin on 2013-03-20 15:43)

@ Ohlbe
Posting: # 10243
Views: 8,483
 

 R, packages (e.g. PowerTOST) and regulatory issues (e.g.FDA)

Thanks to Ohlbe and El Maestro for their quick responses!

» For evaluation of BE in R you need e.g. Yung-Jin Lee's excellent BEAR or some codelines of your own making and then the issue becomes more important.

I will certainly have a closer look to this package that I have discovered in this forum. But we have also WinNonlin Phoenix 1.3 at our compagny, so we plan to use it for BE outcome analysis. I will be able to do some hands-on comparisons with my own dataset.

» Doesn't the FDA systematically repeat the PK and stats calculations in ANDA BE studies ? If this is indeed the case, there is no reason they should raise any difficulty as long as they manage to reproduce your results.

This question is very interesting in the path of submiting an ANDA to the FDA, because it could put less pressure on software validation for this project?
ElMaestro
Hero

Denmark,
2013-03-20 11:41

@ Jérôme Martin
Posting: # 10245
Views: 8,434
 

 FDA recalculation

Hi Jérôme,

» This question is very interesting in the path of submiting an ANDA to the FDA, because it could put less pressure on software validation for this project?

WHile it is true that FDA routinely recalculate trial outcomes I don't think this in itself relaxes any validation requirements.

Also note that for some re-calculations, if done in SAS, it may not be possible to verify exactly the results from R. The best case I can think of is mixed models with imbalance where the SAS Satterthwaite option for denominator DF's cannot be reproduced in R. There is simply no option for that at present if I remember correctly.

if (3) 4

x=c("Foo", "Bar")
b=data.frame(x)
typeof(b[,1]) ##aha, integer?
b[,1]+1 ##then let me add 1



Best regards,
ElMaestro

"(...) targeted cancer therapies will benefit fewer than 2 percent of the cancer patients they’re aimed at. That reality is often lost on consumers, who are being fed a steady diet of winning anecdotes about miracle cures." New York Times (ed.), June 9, 2018.
Helmut
Hero
avatar
Homepage
Vienna, Austria,
2013-03-20 15:06

@ ElMaestro
Posting: # 10246
Views: 8,464
 

 validation

Dear all!

» » This question is very interesting in the path of submiting an ANDA to the FDA, because it could put less pressure on software validation for this project?
»
» WHile it is true that FDA routinely recalculate trial outcomes I don't think this in itself relaxes any validation requirements.

Yep. When it comes to PowerTOST the scripts in the \tests\ folder make validation an easy task.

» Also note that for some re-calculations, if done in SAS, it may not be possible to verify exactly the results from R. The best case I can think of is mixed models with imbalance where the SAS Satterthwaite option for denominator DF's cannot be reproduced in R. There is simply no option for that at present if I remember correctly.

Yessir. Same with Phoenix/WinNonlin, where the ‘Partial Tests’ are expected to agree with SAS’ Type III LSMs in most of cases (whereas the ‘Sequential Tests’ ≡ SAS’ Type I). BTW, the FDA uses not only SAS and R, but also MATLAB and Phoenix themselves.

Cheers,
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
yjlee168
Senior
avatar
Homepage
Kaohsiung, Taiwan,
2013-03-20 22:01

@ ElMaestro
Posting: # 10251
Views: 8,506
 

 R, packages (e.g. PowerTOST) and regulatory issues (e.g.FDA)

Dear all,

The R Foundation for Statistical Computing has an official document (Dec. 28, 2012) about "R: Regulatory Compliance and Validation Issues A Guidance Document for the Use of R in Regulated Clinical Trial Environments". Looks like this document is updated from time to time. Also, Soukup of FDA had slides about "Using R: Perspectives of a FDA Statistical RevieweR". Hope these two documents can help the discussion. The content of Soukup's slide may need to be updated.

» » I’m preparing a clinical ANDA BE study for the company I work with and for that purpose I’m using R and the package PowerTOST (I’m computing power and sample size and also I use CVfromCI and CVpooled)...

PowerTOST is a great tool in sample size and power estimation, especially in RSABE/scABE. I can bet that Detlew Labes must spend a lot of time and efforts to develop it and improve it. I used to download its tar file (containing all source codes; welcome to the open source world!) from CRAN and study its coding skills. I do learn a lots from it.

» ...
» For evaluation of BE in R you need e.g. Yung-Jin Lee's excellent BEAR or ...

Thanks for compliments. Hope the next release will get much better.

All the best,
---Yung-jin Lee
bear v2.8.3-3:- created by Hsin-ya Lee & Yung-jin Lee
Kaohsiung, Taiwan http://pkpd.kmu.edu.tw/bear
Download link (updated) -> here
Jérôme Martin
Junior

France,
2013-03-27 11:59

@ yjlee168
Posting: # 10293
Views: 8,270
 

 R, packages (e.g. PowerTOST) and regulatory issues (e.g.FDA)

Thanks for all the responses adn links! It will help me to prepare internal meetings.

To follow the discussion, I'm wondering what are the views about R from local agency such as ANSM in France for example. In other words, FDA made statment about using R, but what are the opinions of local agencies? For example when companies uses strategy of filling a clinical study in early phase in some country and then move to FDA/EMA, you have to write a SAP to a local agency, and what are your experience with filing an R based SAP in these cases?
Helmut
Hero
avatar
Homepage
Vienna, Austria,
2013-03-27 14:23

@ Jérôme Martin
Posting: # 10296
Views: 8,277
 

 PowerTOST and BfArM

Hi Jérôme!

» […] you have to write a SAP to a local agency, and what are your experience with filing an R based SAP in these cases?

So far I used PowerTOST for the intermediate power analysis and sample size estimation in Two-Stage designs. Protocols were accepted by the German BfARM without any questions; one product approved a while ago.

Cheers,
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. ☼
Science Quotes
Activity
 Thread view
Bioequivalence and Bioavailability Forum |  Admin contact
19,188 posts in 4,084 threads, 1,308 registered users;
online 7 (0 registered, 7 guests [including 6 identified bots]).

The aim of science is not to open the door to infinite wisdom,
but to set a limit to infinite error.    Bertolt Brecht

The BIOEQUIVALENCE / BIOAVAILABILITY FORUM is hosted by
BEBAC Ing. Helmut Schütz
HTML5 RSS Feed