M.tareq
☆

2016-05-03 11:35
(2091 d 19:07 ago)

Posting: # 16266
Views: 7,779

## Sas proc mixed and 4 period partial replicate Cross over [Design Issues]

Hello, can any one please guide me on performing stat analysis of a 4 period 4 seq 3 treatment partial replicate Cross over study, where the ref product only replicate against t1 and t2 for a bioequivlence study of sofosbuvir?
The FDA recommendation is for full replicate and there are little information found regarding a 4 period Cross over with partial replication?
What other effects to account for beside the usual fixed effects in traditional 2x2 Cross over?
Within subject effect, within formulation variability, and carry over effect?

Newbie in stat...

Helmut
★★★

Vienna, Austria,
2016-05-03 12:43
(2091 d 17:59 ago)

@ M.tareq
Posting: # 16268
Views: 7,156

## What is the purpose?

Hi M.tareq,

» […] performing stat analysis of a 4 period 4 seq 3 treatment partial replicate Cross over study, where the ref product only replicate against t1 and t2 for a bioequivlence study of sofosbuvir?

Can you give us an example how the sequences would look like? What is the purpose of having two Test-products in the study? See also this post and followings.

» The FDA recommendation is for full replicate…

No it isn’t. The FDA recommends two conventional 2×2×2 studies (fasting and fed state).

According to the WHO’s PQTm the reference in fed state is a HVDP (Cmax CVwR 54%) whilst variability of AUC is low (CVwR 10%). In the WHO’s guidance reference-scaling for Cmax according to the EMA’s method is mentioned as an alternative.

» …and there are little information found regarding a 4 period Cross over with partial replication?

I have never heard about such a design.

» What other effects to account for beside the usual fixed effects in traditional 2x2 Cross over?

No idea.

» Within subject effect, within formulation variability, and carry over effect?

The sequence effect is confounded with
• the carry-over effect, and
• the formulation-by-period interaction.
A statistically significant sequence effect could indicate that there is
• a true sequence effect,
• a true carry-over effect,
• a true formulation-by-period interaction, or
• a failure of randomization.
It cannot be handled statistically but only avoided by design (sufficient wash-out). Including carry-over in cross-overs does not make sense.

» Newbie in stat...

You mentioned Sas proc mixed in the subject line. If you want to apply the EMA’s ABEL, a mixed-effects model is not acceptable. If you want to apply the FDA’s RSABE, the partial replicate for the unscaled PK metric (AUC) may fail to converge. Specify the covariance structure with TYPE=FA0(1) instead of TYPE=FA0(2) as given in the progesterone guidance.

Dif-tor heh smusma 🖖
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
M.tareq
☆

2016-05-03 15:06
(2091 d 15:37 ago)

@ Helmut
Posting: # 16270
Views: 6,807

## What is the purpose?

Example of the 4 seq
RT2RT1
T1RT2R
T2RT1R
RT1RT2
It was for cost effective as the ref product wasn't avail in the required amount for 2 separate studies
Ohlbe
★★★

France,
2016-05-03 15:48
(2091 d 14:54 ago)

@ Helmut
Posting: # 16271
Views: 6,852

## Alpha inflation ?

Dear all,

» If you want to apply the EMA’s ABEL, a mixed-effects model is not acceptable. If you want to apply the FDA’s RSABE [...]

Any idea about alpha inflation control with such a design ?

Regards
Ohlbe
d_labes
★★★

Berlin, Germany,
2016-05-03 16:06
(2091 d 14:37 ago)

(edited by d_labes on 2016-05-03 16:30)
@ Ohlbe
Posting: # 16273
Views: 6,876

## Alpha inflation

Dear Ohlbe,

» Any idea about alpha inflation control with such a design ?

If the evaluation follows the same line as the EMA recommends for 3- or 4-period studies, namely forget all the data not involved in the respective comparison, the same alpha inflation as for a usual partial replicate design is expected.

See
Munoz et al.
"Consumer’s risk in the EMA and FDA regulatory approaches for bioequivalence in highly variable drugs"
Stat Med. 2016 May 30;35(12):1933-43.
doi: 10.1002/sim.6834. Epub 2015 Dec 28.

The method to control the TIE is coming soon ...

Regards,

Detlew
Helmut
★★★

Vienna, Austria,
2016-05-03 17:49
(2091 d 12:53 ago)

@ Ohlbe
Posting: # 16275
Views: 6,880

## Not with such a high CV

Dear Ohlbe,

» Any idea about alpha inflation control with such a design ?

Adding to Detlew’s wise words: For a CVwR of 54% I would expect no inflation of the TIE with the FDA’s RSABE and practically none with the EMA’s ABEL.

Dif-tor heh smusma 🖖
Helmut Schütz

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

Berlin, Germany,
2016-05-03 21:07
(2091 d 09:35 ago)

@ Helmut
Posting: # 16276
Views: 6,794

## High CV?

Dear Helmut,

» Adding to Detlew’s wise words: For a CVwR of 54% I would expect no inflation of the TIE with the FDA’s RSABE and practically none with the EMA’s ABEL.

where did this CVwR you mention came from? I couldn't detect such a number in this thread. Did you use a crystal ball to figure it out in M.tareq's post? Or do you have some insider information?

Regards,

Detlew
Helmut
★★★

Vienna, Austria,
2016-05-03 21:58
(2091 d 08:44 ago)

@ d_labes
Posting: # 16277
Views: 6,858

## Acc. to WHO PQTm

Dear Detlew,

» where did this CVwR you mention came from? I couldn't detect such a number in this thread. […] do you have some insider information?

I got it from page 2 of the WHO’s guidance linked in my first post above.

Dif-tor heh smusma 🖖
Helmut Schütz

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

Berlin, Germany,
2016-05-04 11:50
(2090 d 18:52 ago)

(edited by d_labes on 2016-05-04 12:36)
@ Helmut
Posting: # 16281
Views: 6,702

## Sofosbuvir

Dear Helmut!

Sorry, I simply overlooked that M.tareq has mentioned sofosbuvir in his opening post.

Regards,

Detlew