Imran
☆    

Mumbai,
2007-12-13 11:05
(5950 d 21:23 ago)

Posting: # 1368
Views: 17,282
 

 Sequence effects [General Sta­tis­tics]

Dear HS and PK Friends

Please comment:
What will be significance of Sequence effect (P value less than 0.05) in the Primary variables (Cmax, AUCt and AUCinf) in the estimation of bioequivalence.

What justification or rational can be given if sequence effects are found in AUCs and the study is bioequivalent.

Thanks and Regards,

Dr.Imran Khan
Ohlbe
★★★

France,
2007-12-13 14:45
(5950 d 17:42 ago)

@ Imran
Posting: # 1370
Views: 14,707
 

 Sequence effects

Dear Imran

Did you try and type "sequence effect" in the forum's search box ? This is a rather frequent topic. You will already find some answers this way, and you could then ask any unanswered specific questions.

regards
Ohlbe
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2007-12-13 15:27
(5950 d 17:00 ago)

@ Imran
Posting: # 1371
Views: 15,907
 

 Sequence effects?!

Dear Imran!

❝ What will be significance of Sequence effect (P value less than 0.05)


Are you meaning: “What will be the clinical relevance of a statistical significant effect…”?

Testing for a sequence effect in a 2×2 cross-over study (Grizzle 19651) is statistically flawed and therefore simply futile (Freeman 19892).

❝ … in the Primary variables (Cmax, AUCt and AUCinf) in the estimation of bioequivalence.


Nothing? OK, to be more precise, if:
  • the study was performed in healthy subjects,
  • the drug is not an endogenous entity, and
  • an adequate washout were maintained (no pre-dose concentrations in period II).
As an entry point have a look at the FDA’s guidance, Section VII.B.

❝ What justification or rational can be given if sequence effects are found in AUCs and the study is bioequivalent.


To quote one of the “fathers” of statistical BE testing (Westlake 19883):

Note that the carryover effect is, essentially, the sequence effect, which can be tested against the sum of squares within sequence. If this carryover effect exists, then it confounds the test on formulations. […] My own experience with a large number of comparative bioavailability trials has led me to believe that significant carryover effects (at the 0.05 level) tend to occur in about 5% of the trials; in other words, I believe that carryover effects do not normally exist.

(my emphasis)

In a large metastudy significant effects were found in about ~10% of 2×2 studies (n=324), and ~5% of 96 6×3 studies (D’Angelo et al. 20014), which was actually the level of the tests.

Since no valid statistical method to correct for a true sequence effect exists, and if the conditions of the study design given above hold, even testing for a sequence effect should be abandoned and put into the statistical “trash can” (my personal interpretation of Senn 20025).

For an overview you may also see one of my presentations and a later article.

  1. Grizzle JE. The two-period change over design and its use in clinical trials. Biometrics. 1965;21(2):467–80. doi:10.2307/2528104
  2. Freeman P. The performance of the two-stage analysis of two-treatment, two-period cross-over trials. Stat Med. 1989;8(12):1421–32. doi:10.1002/sim.4780081202
  3. Westlake WJ. Bioavailability and Bioequivalence of Pharmaceutical Formulation. In: Peace KE, editor. Biopharmaceutical Statistics for Drug Development. New York: Marcel Dekker; 1988. p. 336–7.
  4. D’Angelo G, Potvin D, Turgeon J. Carry-Over Effects in Bioequivalence Studies. J Biopharm Stat. 2001;11(1–2):35–43. doi:10.1081/BIP-100104196
  5. Senn S. Cross-over Trials in Clinical Research. Chichester: Wiley; 2002. p. 35–88.

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

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

Mumbai,
2007-12-14 07:58
(5950 d 00:29 ago)

(edited by Ohlbe on 2007-12-14 09:17)
@ Helmut
Posting: # 1375
Views: 14,744
 

 Sequence effects?!

Dear HS and PK Friends,

The same topic I had discussed earlier. Things are clear for Formulation and Period effects, but sequence effects was not discussed up to the level.

Here I mean to say statistical significant Sequence effect. I was trying to rule out the statistical possiblities for the sequence effects.

Please Comment

Regards,

Dr.Imran

--
Edit: Full quote removed. Please see this post! [Ohlbe]
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2007-12-14 12:53
(5949 d 19:35 ago)

@ Imran
Posting: # 1376
Views: 14,722
 

 Sequence effects?!

Dear Imran!

❝ […] but sequence effects was not discussed up to the level.

Really not? :confused:

❝ Here I mean to say statistical significant Sequence effect. I was trying to rule out the statistical possiblities for the sequence effects.


I thought that I already gave you detailed answers aboveWhich are the open questions?

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

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

Poland,
2008-07-16 14:49
(5734 d 18:38 ago)

@ Helmut
Posting: # 2047
Views: 14,353
 

 Sequence effects?!

Dear All!
I'd like to refresh this topic:

Nothing? OK, to be more precise, if:

the study was performed in healthy subjects,

the drug is not an endogenous entity, and

an adequate washout (no predose concentrations in period II) were maintained.


Is meeting these criteria enough to ignore sequence effect?
From [image] reference given previously I can see that also equivalence must be present.
And in study that I'm currently participating in, it turned out that test drug did not meet BQ criteria for Cmax when compared to reference drug and significant sequence effect for Cmax was detected.
Now this publication http://cat.inist.fr/?aModele=afficheN&cpsidt=984872 claims that:

"The sequence effect is confounding with the unequal residual effect and with the formulation by period interaction. Since the existence of the sequence effect questions the quality of the trial, the applicant should provide possible explanations and information on the subjects, the trial conditions, the clinical settings and the assay methodology. An additional statistical analysis on the data from the first period of the trial may support the bioequivalence. If it is proven that the sequence effect is a true effect then the generic may be approved for marketing authorization."

So should we go deeper and try to find explanation for sequence effect? Is it possible at all to find anything if we pass criteria mentioned by HS (cited above)?
Thanks in advance for your opinions.
Łukasz
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2008-07-16 16:53
(5734 d 16:34 ago)

@ lukamar
Posting: # 2048
Views: 14,334
 

 Sequence effects?!

Dear Łukasz!

❝ ❝ OK, to be more precise, if:

❝ ❝ the study was performed in healthy subjects,

❝ ❝ the drug is not an endogenous entity, and

❝ ❝ an adequate washout (no predose concentrations in period II) were maintained.


❝ Is meeting these criteria enough to ignore sequence effect?


Actually this a quote from the FDA’s Guidance and (hopefully) consensual – Freeman’s paper disproving Grizzle’s 2-stage approach was published 19 (!) years ago.

❝ From [image] reference given previously I can see that also equivalence must be present.


This is a doubtful reference. Anyhow, yes, if you want to demonstrate bioequivalence your CI must be entirely within the acceptance range. Statistical significant sequence effect (SE) or not.

❝ And in study that I'm currently participating in, it turned out that test drug did not meet BQ criteria for Cmax when compared to reference drug and significant sequence effect for Cmax was detected.


So the test formulation is not BE to the reference in respect to Cmax.

❝ Now this publication [E Zintzaras, 2000] claims that:

❝ "The sequence effect is confounding with the unequal residual effect and with the formulation by period interaction. […]"


True.

❝ "[…] Since the existence of the sequence effect questions the quality of the trial, the applicant should provide possible explanations …"


Yes, but since a statistical significant SE may have different sources – which can’t be separated in a nonreplicated study – it’s futile to aim at an “explanation”. It’s only possible to speculate, which can’t be the basis of further statistical analyses!

❝ "… and information on the subjects, the trial conditions, the clinical settings and the assay methodology. […]"


True. Only by assessing the side conditions of the study (wash-out, randomization, lack of pre-dose concentrations, …) one can get some insights on the SE. The nasty thing is, that once the study is finished, it’s not possible to get any clues from the data. We can avoid a potential SE only in study planning. If a study is properly planned, any testing for a SE is rubbish.

❝ "[…] An additional statistical analysis on the data from the first period of the trial may support the bioequivalence. […]"


Nonsense. The paper was published in 2000 in a journal – though peer-reviewed – has an impact factor close to zero (about one year ago: 0.3). Obviously both author’s and reviewer’s “state of the art” was with Grizzle’s two stage approach of 1965 [sic]. Since then Freeman’s paper, meta-analyses, a book by Stephen Senn (which in between the lines deals to a good part with rejection of SE testing), chapters on the topic in books by Jones/Kenward, and Patterson/Jones were published. Although the author is with a European regulatory authority (which is known for its bizarre interpretation of the guidelines, like a posteriori power ≥80% needed for BE) I would not take the statement seriously, because:
  • Patient’s risk in Grizzle’s method may be inflated
  • If you are able to demonstrate BE in period 1 as a parallel design, your study was extremely overpowered; so your design is questionable…

❝ "[…] If it is proven that the sequence effect is a true effect then the generic may be approved for marketing authorization."


Only by means of “black magick” it’s possible to ‘prove the sequence effect as a true effect’ in a 2×2 cross-over study.

❝ So should we go deeper and try to find explanation for sequence effect? Is it possible at all to find anything if we pass criteria mentioned by HS (cited above)?


You can look at the design and the performance of the study, but if you are not BE, there’s no statistical “bail-out procedure” to deal with the problem. Sorry.

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,636 registered users;
93 visitors (0 registered, 93 guests [including 3 identified bots]).
Forum time: 08:28 CET (Europe/Vienna)

With four parameters I can fit an elephant,
and with five I can make him wiggle his trunk.    John von Neumann

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5