Helmut
★★★

Vienna, Austria,
2018-11-09 18:01

Posting: # 19558
Views: 2,983

## EMA: Mahalanobis distance is dead... [Dissolution / BCS / IVIVC]

Cheers,
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
ElMaestro
★★★

Belgium?,
2018-11-09 19:12

@ Helmut
Posting: # 19559
Views: 2,587

## EMA: Mahalanobis distance is dead...

Thanks Hötzi,

I applaud that move.
Bootstrapping has enormous potential, not just for dissolution. Sometimes the idea behind bootstrapping to me appears so simple and naïve that I am thinking it must be wrong . Especially considering how often we complicate simple issues with fancy maths.

So due to this step in the right direction I'll put flags on the table tonight and celebrate.

I could be wrong, but...

Best regards,
ElMaestro

"Pass or fail" (D. Potvin et al., 2008)
Helmut
★★★

Vienna, Austria,
2019-10-06 15:27

@ ElMaestro
Posting: # 20667
Views: 1,470

## Is ƒ2 dead as well – always bootstrapping?

Hi ElMaestro and all,

I’ve read the Q&A again. Here the last paragraph in all of its beauty:

Any approach based upon confidence intervals for ƒ2 would, however, be considered appropriate whether the validity criteria outlined in CHMP guidance are met or not [CPMP/EWP/QWP/1401/98 Rev. 1/Corr**]. Similarity could then be declared if the confidence interval for ƒ2 were entirely above 50. However, regardless of whether the conditions to adequately apply ƒ2 in a dissolution experiment are fulfilled or not, the properties of the ƒ2 sampling distribution do not allow the derivation of exact confidence intervals to adequately quantify the uncertainty of the ƒ2 estimate. To address this, bootstrap methodology could be used to derive confidence intervals for ƒ2 based on quantiles of re-sampling distributions, and this approach could actually be considered the preferred method over ƒ2 and MD.

(my emphases)

Do I get this right? Bootstrapping is preferred over ƒ2 even if the validity criteria* are met?
I expect ‘regulatory creep’. Preferred will turn into mandatory. Will it be required retrospectively? I know some minor variations supported by ƒ2, where it was say, only 53 in one pH. Will not pass the lower confidence limit.
IMHO, the only way to decrease the CV – and hence, the width of the bootstrapped CI – is to substantially increase the number of units.

CV units ──────── ♫    12 ¾    21 ⅔    27 ½    48 ⅓   108

No risk, no fun.

• ≥12 units of T and R, CV ≤20% at ≤15 minutes, CV ≤10% at >15 minutes, not more than one mean value of >85% dissolved for each formulation.

Cheers,
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
ElMaestro
★★★

Belgium?,
2019-10-06 17:54

@ Helmut
Posting: # 20668
Views: 1,439

## Is ƒ2 dead as well – always bootstrapping?

Hi Hötzi,

» Do I get this right? Bootstrapping is preferred over ƒ2 even if the validity criteria* are met?

I think yours is the right interpretation.
And it is free of cost, right? You have free software to do the calculation albeit the validation steps may be a bit cumbersome if you decide to go for that. It is quite uncomplicated: once you have your dissolution datasets you do the f2 and you do some bootstapping with one extra click, wait a few minutes or hours (or just split seconds if you are using my software, made in C of course) and Bob's your uncle.

I could be wrong, but...

Best regards,
ElMaestro

"Pass or fail" (D. Potvin et al., 2008)
Helmut
★★★

Vienna, Austria,
2019-10-06 18:14

@ ElMaestro
Posting: # 20670
Views: 1,464

## Ahem – number of units?

Hi ElMaestro,

» » […] Bootstrapping is preferred over ƒ2 even if the validity criteria are met?
»
» I think yours is the right interpretation.

Was also the opinion of some experts at BioBridges. Left the audience speechless.

» And it is free of cost, right? [etc. etc.]

Sure. The problem is the ‘regulatory crep’ I suspected in my post. Jiři showed nice simulations about the false positive rate. To squeeze the lower CL ≥50 is not another league but another sport. How many units will we have to test in the future? 24, 48, 96?

Cheers,
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
ElMaestro
★★★

Belgium?,
2019-10-06 22:45

@ Helmut
Posting: # 20671
Views: 1,439

## Ahem – number of units?

Hi Hötzi,

» Sure. The problem is the ‘regulatory crep’ I suspected in my post. Jiři showed nice simulations about the false positive rate. To squeeze the lower CL ≥50 is not another league but another sport. How many units will we have to test in the future? 24, 48, 96?

Am thinking loud here: You can bootstrap a principal answer to such questions as well.
You generate a dissolution dataset with n chambers for T and R; then you resample with replacement to generate datasets with m chambers (m > n, presumably) per product, to find out at which level of m you get an f2 limit >50.

This is quite simple, actually, to do the programming and simulation. But yes, if you end up with 326 chambers per product then I guess you are royally screwed.

We could do a publication about this. We'd need a candidate dataset, I don't think I have one for this which I could put into the public domain.

Could this even be done as a two-stage approach? You first look at bootstrap f2 levels in the initial n-chamber scenario. If you are passing, stop. Else, find your sample size m and run the (m-n) chambers on top, then re-bootstrap the whole thing. This would be a "two-treatment, m-chamber, tripple bootstrap, two-stage dissolution trial", and you may need to work with increased coverage intervals for alpha preservation.

Man, I should win a Nobel prize for this idea.

I could be wrong, but...

Best regards,
ElMaestro

"Pass or fail" (D. Potvin et al., 2008)
wienui
★

Germany, Oman,
2019-10-07 19:11

@ ElMaestro
Posting: # 20680
Views: 1,381

## Ahem – number of units?

Hi ElMaestro,

» We could do a publication about this. We'd need a candidate dataset, I don't think I have one for this which I could put into the public domain.

» to work with increased coverage intervals for alpha preservation.

This could be a great idea and let Diane Potvin take part.

» Man, I should win a Nobel prize for this idea.

Why not, but under the condition, you could be able to reduce enormously (Biowaiving) the number of in VIVO conducting studies,

Best regards,
Osama

Cheers,
Osama