Relaxation
★

Germany,
2022-12-02 16:53
(65 d 11:21 ago)

Posting: # 23376
Views: 2,186

## New guidance on Sta­tis­tical Approaches to Establishing Bio­equi­va­lence [BE/BA News]

Good afternoon everybody.

Although I assume that most of us have subscribed to the FDA newsletter, I wanted to post that FDA just notified that the new "Statistical Approaches to Establishing Bioequivalence" was made available.
From a quick once over interesting topics included .

Helmut
★★★

Vienna, Austria,
2022-12-02 17:43
(65 d 10:31 ago)

@ Relaxation
Posting: # 23377
Views: 2,023

## A first look

Hi Relaxation,

THX for noticing us.

❝ From a quick once over interesting topics included .

Indeed.
• 3-period 2-sequence full replicate as an example (is the bloody partial replicate design finally landing in the dustbin of history where it belongs?)
• Adaptive designs (yes, also simulation-based; EMA, are you reading this?)
• Sparse sampling – wow! I hope by ‘mean’ the FDA means the geometric mean.
• Sensitivity of sample size estimation. THX, now in line with ICH E9 (of 1998!)
• The dreadful estimands enter the scene.
• ‘for a replicated PK BE study, if reference-scaled average BE is used, the applicant should ensure that the calculated intra-subject variability is not inflated due to extreme values or situations.’ That’s new but inline with all others (except ANVISA, still not requiring it)
’In the Random statement, TYPE=FA0(2) could possibly be replaced by TYPE=CSH or UNR.’ THX for CSH. Why not TYPE=FA0(1)?
Instead of Satterthwaite’s degrees of freedom Kenward-Roger could be used (what does ‘possibly’ mean here?)
• Sorry folks, the CI inclusion approach TOST. Only the conclusions are the same.
• What’s the purpose of reporting arithmetic means additionally to geometric means?
• Covariates in parallel designs. THX!
• Details about studies in multiple groups or sites. Why equal sizes / group?
• How to deal in Higher-Order Crossover Designs with references from two regions (i.e., Two at a Time).

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
jag009
★★★

NJ,
2022-12-02 18:58
(65 d 09:15 ago)

@ Helmut
Posting: # 23378
Views: 1,999

## A first look

Hi Helmut,

❝ ❝ From a quick once over interesting topics included .

Almost a complete overhaul. Very detailed. I notice something, they removed the min sample size requirement of n=12 (Unless it's written in later part of the doc)

J
BRB
☆

2022-12-02 19:20
(65 d 08:54 ago)

@ jag009
Posting: # 23379
Views: 2,004

## A first look

Hi J,
No, it's still there. See lines 260-261:

The number of evaluable subjects in a PK BE study should not be less than 12. For highly variable drug products, a minimum of 24 subjects are recommended for BE assessment.

❝ Almost a complete overhaul. Very detailed. I notice something, they removed the min sample size requirement of n=12 (Unless it's written in later part of the doc)

mittyri
★★

Russia,
2022-12-02 21:38
(65 d 06:35 ago)

@ Helmut
Posting: # 23380
Views: 2,056

## SAS strikes back

Dear Helmut, Dear All!

what bothers me (1095-1096):
Alternative software could also be used if same results are generated as in PROC MIXED in SAS.

what? How could I know if I am not equipped with Power To Know God Blessed Software®?

Kind regards,
Mittyri
Helmut
★★★

Vienna, Austria,
2022-12-02 23:30
(65 d 04:44 ago)

@ mittyri
Posting: # 23381
Views: 1,968

## SAS strikes back

Hi Mittyri,

Alternative software could also be used if same results are generated as in PROC MIXED in SAS.

❝ what? How could I know if I am not equipped with Power To Know God Blessed Software®?

Three options.
1. Hire a consultant with a SAS license.
2. AFAIK, the FDA recalculates all studies. Cross fingers.
3. Wait till someone publishes a paper about cross-validation.
As an aside: How to assess ‘outliers’ in RSABE?

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
mittyri
★★

Russia,
2022-12-02 23:52
(65 d 04:21 ago)

@ Helmut
Posting: # 23382
Views: 1,959

## Outliers?

Hi Helmut,

❝ Three options.

1. Hire a consultant with a SAS license.

2. AFAIK, the FDA recalculates all studies. Cross fingers.

3. Wait till someone publishes a paper about cross-validation.

I'll cross fingers to have that omitted from the final guidance (even if I do not plan to submit to FDA anything; FDA guidelines are getting used to penetrate across the world).

❝ As an aside:

How to assess ‘outliers’ in RSABE?

the only one method I can recall is described by Mr.Schall with coauthors
But I know that your question is not trivial and deeper than that.

PS: what is wrong with the format of cited list?

Kind regards,
Mittyri
Helmut
★★★

Vienna, Austria,
2022-12-03 00:11
(65 d 04:03 ago)

@ mittyri
Posting: # 23383
Views: 1,956

## Outliers?

Hi Mittyri,

❝ I'll cross fingers to have that omitted from the final guidance (even if I do not plan to submit to FDA anything; FDA guidelines are getting used to penetrate across the world).

Somebody at the FDA might have pushed the panic button. Behind the curtains possibly GBHI and ICH M13 are driving forces. Harmonization comes at a price.

❝ ❝ How to assess ‘outliers’ in RSABE?

❝ the only one method I can recall is described by Mr.Schall with coauthors

❝ But I know that your question is not trivial and deeper than that.

Yep. At least the EMA considered Schall et al. as too ‘statistical’ [sic] and recommended my joke of box plots. Not the slightest idea what the FDA expects.

❝ PS: what is wrong with the format of cited list?

As it ever was. Sorry. Christian Seiler’s PHP StringParser class is of 2008 and can’t handle lists properly. The only other I found is not compatible with PHP ≥7.0…

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
mittyri
★★

Russia,
2022-12-07 17:26
(60 d 10:47 ago)

@ Helmut
Posting: # 23389
Views: 1,638

## model-based BE?

Hi Helmut and All!

regarding

FDA encourages generic and new drug applicants to propose and discuss novel methodologies (e.g., model-based BE…)

(lines 56-58)

are you reading this as a projection of VBE or some broader successful trends exist?

Kind regards,
Mittyri
Helmut
★★★

Vienna, Austria,
2022-12-10 11:13
(57 d 17:01 ago)

@ mittyri
Posting: # 23397
Views: 1,540

## model-based BE?

Hi Mittyri,

❝ FDA encourages generic and new drug applicants to propose and discuss novel methodologies (e.g., model-based BE…)

❝ are you reading this as a projection of VBE or some broader successful trends exist?

Perhaps. See this post refering to the EMA’s Methodology Working Party. Aren’t these points contradictory?
• EMA does not exclude use of model informed BE, especially in cases when in vivo BE studies cannot be simply conducted in healthy subjects.
• When model-based BE is used, model evaluation should include technical and clinical validation of the model as well as assessment of its applicability.

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
mittyri
★★

Russia,
2023-01-11 16:32
(25 d 11:42 ago)

@ Helmut
Posting: # 23421
Views: 585

## Hooker, Moellenhoff et al. are moving MBBE forward

Hi Helmut and all!

just want to share model based approaches FDA is working on with different scentists.
Uppsala Universitet raises the flag:
Hooker #1
Hooker #2
He claims that model averaging should help in case of troubles with model building.

Kathrin Moellenhof with colleagues (another giant there: France Mentre) are attacking classical BE approaches from other side.
Interesting results of TIE estimation. Simulation-based only.

In PubMed you can see multiple papers referencing this article.

Kind regards,
Mittyri
SMA
☆

Europe,
2023-01-17 10:41
(19 d 17:32 ago)

@ Helmut
Posting: # 23425
Views: 455

## model-based BE?

Dear all

EMA does not exclude use of model informed BE, especially in cases when in vivo BE studies cannot be simply conducted in healthy subjects.

When model-based BE is used, model evaluation should include technical and clinical validation of the model as well as assessment of its applicability.

Please note this interesting and very recent publication by some European regulators (incl members of MWP), which IMHO does not contain solutions yet, but establishes the principle that there is some openness to discuss this topic with EMA:

Manolis, E, García-Arieta, A, Lindahl, A, Kotzagiorgis, E, Limberg, J, Holte, Ø, Paixao, P, Versantvoort, C, Tshinanu, FM, Blake, K, Van Den Heuvel, M. Using mechanistic models to support development of complex generic drug products: European Medicines Agency perspective. CPT: pharmacometrics & systems pharmacology. 2023-Jan-11. Link
PharmCat
★

Russia,
2022-12-21 20:30
(46 d 07:44 ago)

@ Helmut
Posting: # 23405
Views: 1,237

## A first look

Hi all!

❝ ’In the Random statement, TYPE=FA0(2) could possibly be replaced by TYPE=CSH or UNR.’ THX for CSH. Why not TYPE=FA0(1)?

❝ Instead of Satterthwaite’s degrees of freedom Kenward-Roger could be used (what does ‘possibly’ mean here?)

Well it finally happened. It took 20 years to find out that FA0(2), CSH and UN is the same for this model for 2 formulations (but possibly means that not entirely sured about this). May be possibly means that in is not true for 3 formulation?

I don't understand what FA0(1) means for this model? No correlation? Same as diagonal cov structure?

❝ Alternative software could also be used if same results are generated as in PROC MIXED in SAS.

It is very funny. Because if percept it exactly - there is no software that give you exactly same DDoF (Satterthwaite’s or freedom Kenward-Roger). All of them (SPSS, WinNonlin ets...) give you slightly different results for DDoF.
Rayhope
☆

India,
2023-01-06 07:56
(30 d 20:18 ago)

@ Helmut
Posting: # 23420
Views: 767

## A first look

Very enriching and expedient discussion on this new guidance. Thanks all guys putting your thoughts. I would like to know more about "Model based BE designs". if any one can provide Helpful articles (Really practical one with examples) on this topic will be great.
Helmut
★★★

Vienna, Austria,
2022-12-03 14:09
(64 d 14:04 ago)

@ Relaxation
Posting: # 23384
Views: 1,907

## Groups, sites…

Hi Relaxation & all,

❝ From a quick once over interesting topics included .

Since I’m working on a paper on group-effects in BE (still collecting data; see this post) I had a closer look at Section III.D.

The relevant parts:

[…] sponsors should minimize the group effect in a PK BE study as recommended below:

1. Dose all groups at the same clinic unless multiple clinics are needed to enroll a sufficient number of subjects.
2. Recruit subjects from the same enrollment pool to achieve similar demographics among groups.
3. Recruit all subjects, and randomly assign them to group and treatment arm, at study outset.
4. Follow the same protocol criteria and procedures for all groups.
5. When feasible (e.g., when healthy volunteers are enrolled), assign an equal sample size to each group.

Bioequivalence should be determined based on the overall treatment effect in the whole study population. In general, the assessment of BE in the whole study population should be done without including the treatment and group interaction(s) term in the model, but applicants may also use other pre-specified models, as appropriate. The assessment of interaction between the treatment and group(s) is important, especially if any of the first four study design criteria recommended above are not met and the PK BE data are considered pivotal information for drug approval. If the interaction term of group and treatment is significant, heterogeneity of treatment effect across groups should be carefully examined and interpreted with care. If the observed treatment effect of the products varies greatly among the groups, vigorous attempts should be made to find an explanation for the heterogeneity in terms of other features of trial management or subject characteristics, which may suggest appropriate further analysis and interpretation.

It is important that statistical methods and models for the primary BE analysis are fully pre-specified in the protocol or SAP (e.g., in an ANDA study, the applicant should pre-specify detailed statistical criteria and models to be used if the interaction term of group and treatment is applicable). In addition, the statistical model should reflect the multigroup nature of the study. For example, if subjects are dosed in two groups in a crossover BE study, the model should reflect the fact that the periods for the first group are different from the periods for the second group, i.e., the period effect should be nested within the group effect.

(my emphases)

The terrible ‘model (I)’ \eqalign{Y&|\;\text{Group},\,\text{Sequence},\,\text{Treatment},\\ &\phantom{|}\;\text{Subject}(\text{Group}\times \text{Sequence}),\,\text{Period}(\text{Group}),\\ &\phantom{|}\;\text{Group}\times \text{Sequence},\,\text{Group}\times \text{Treatment}}\tag{I} as a mandatory pre-test – which was not stated in any of the previous guidances but in deficiency letters – is gone with the wind. Very good, because it inflated the Type I Error and compromised power.
Now the FDA recommends ‘model (II)’ \eqalign{Y&|\;\text{F},\,\;\text{Sequence},\,\text{Subject}(\text{F}\times \text{Sequence}),\\ &\phantom{|}\;\text{Period}(\text{F}),\;\text{F}\times \text{Sequence},\,\text{Treatment}\small{\textsf{,}}}\tag{II} where $$\small{\text{F}}$$ is the code for $$\small{\text{Group}}$$ or $$\small{\text{Site}}$$, respectively. Whilst important in a parallel design, I fail to understand why ‘similar demographics’ are of any importance in a crossover design. Equally sized groups are recommended but not necessary.
If any of the first four conditions are not met, ‘model (I)’ should be used to assess the Group-by-Treatment interaction. If $$\small{p(G\times T)<0.1}$$, ‘vigorous [sic] attempts’ should be made to find an explanation. It is an open question what to do if treatment effects vary between groups. Base the BE assessment on the largest one or on the one with smallest variability?

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,
2022-12-04 07:34
(63 d 20:40 ago)

@ Relaxation
Posting: # 23385
Views: 1,836

## SxF

Hi all,

the new draft is interesting and I am still digesting it.

I see that the SxF interaction is given some consideration here, but I am not sure what is meant now. One thing is the outlier aspect and that is ok (Line 387 etc). But it enters the scene again in Line 478 and that part is not so clear. As you read it, do they recommend estimation of SxF whenever the design allows? What do we do with the info?

Muchas gracias.

Pass or fail!
ElMaestro
Helmut
★★★

Vienna, Austria,
2022-12-10 11:37
(57 d 16:37 ago)

@ ElMaestro
Posting: # 23398
Views: 1,543

## SxF

Hi ElMaestro,

❝ I see that the SxF interaction is given some consideration here, but […] it enters the scene […] in Line 478 and that part is not so clear. As you read it, do they recommend estimation of SxF whenever the design allows?

If we follow the recommended mixed-effects model (lines 1068–1096), yes.

❝ What do we do with the info?

Attach the software’s output to the report.

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
Mahmoud
☆

Jordan,
2022-12-10 15:08
(57 d 13:05 ago)

@ Relaxation
Posting: # 23399
Views: 1,545

## New guidance on Sta­tis­tical Approaches to Establishing Bio­equi­va­lence

Dear All

The FDA Generic Drugs Advisory Committee recommended in 1991 that the primary comparison of interest in a BE study is the ratio, rather than the difference, between average PK parameter data from the T and R formulations. Using logarithmic transformation, the general linear statistical model employed in the analysis of BE data allows inferences about the difference between the two means on the log scale, which can then be retransformed into inferences about the ratio of the two averages (geometric means) on the original scale. Logarithmic transformation thus achieves a general comparison based on the ratio rather than the differences.

Under lognormal data for PK paramertes ratio of geometric means the same as ratio of medians

However, situations with different within-subject variances of drugs may often occur in practice. BE tests cannot be easily transformed back from the log scale to the original scale under unequal within-subject variances.