Bioequivalence and Bioavailability Forum • Statistical Issue (significant Formulation, Period and seq.)

Imran
☆

Mumbai,
2007-04-20 09:22
(6595 d 21:26 ago)

Posting: # 677
Views: 14,199

Statistical Issue (significant Formulation, Period and seq.) [Regulatives / Guidelines]

Dear sir/Madam

In Bioequivalence study, if formulation (Treatment) effect is significant (since ANOVA p-value is less than 0.05) for both Cmax and AUC and individually for Cmax or AUC, what we can conclude from this significant formulation effect. Similarly, what we can conclude from period and sequence significant effect.
In case of significant formulation effect is that mean number of subject used in the study were more than required.

Regards,

Imran

P.S.
In a statistical analysis of BA/BE data, many time we come across sequence, formulation and period effect in ANOVA (p <0.05). what conclusion should one draw from such data? what would be reason for such effects?

Edit: P.S. attached from new post by Imran of 2007-05-07 15:00 [HS]

—
Dr.Imran Khan

usfda_emea
●

2007-05-05 11:12
(6580 d 19:36 ago)

@ Imran
Posting: # 714
Views: 12,342

Statistical Issue (significant Formulation, Period and seq.)

Post reply

Dear IMRAN,

❝ In Bioequivalence study, if formulation (Treatment) effect is significant (since ANOVA p-value is less than 0.05) for both Cmax and AUC and individually for Cmax or AUC, what we can conclude from this significant formulation effect.

Assuming that the bioequivalence is observed, inspite of significant difference in treatment effect for ANOVA, it can be inferred that the formulations are different but bioequivalent.

❝ Similarly, what we can conclude from period and sequence significant effect.

A significant period effect could reflect on following causes:

positioning,
timing,
degree of physical activity,
composition of food, beverages,
temperature of water administered during dosing,
psychological status of subjects during two periodin turn effecting the bowel transit and drug absorption.

Opinions only, with no binding.
USFDA_EMEA. ;-)

Helmut
★★★

Vienna, Austria,
2007-05-07 16:42
(6578 d 14:06 ago)

@ Imran
Posting: # 715
Views: 13,393

significant Formulation and Period effects

Post reply

Dear Imran!

❝ In Bioequivalence study, if formulation (Treatment) effect is significant (since ANOVA p-value is less than 0.05) for both Cmax and AUC and individually for C_max or AUC, what we can conclude from this significant formulation effect.

❝ In case of significant formulation effect is that mean number of subject used in the study were more than required.

Simple answer: Yes.
But: Your sample size estimation relied on a couple of uncertainties (CV_intra, ∆, …); you should not bother about a significant treatment effect (i.e., 100% not included in confidence interval) unless your drug is a NTID and you have to to deal with Danish Regulators.

❝ Similarly, what we can conclude from period and sequence significant effect.

Bioequivalenve assessment is not influenced by a period effect.
As an example you may use this data, which gives:

┌───────────┬─────────┬──────────┬────────────┬─────────────────┐

│   Effect  │ p-Value │ Estimate │ Confidence │     Interval    │

├───────────┼─────────┼──────────┼────────────┼─────────────────┤

│ Period    │  0.7856 │   98.4%  │     95%    │  87.4% - 110.8% │

│ Sequence  │  0.3239 │  115.4%  │     95%    │  86.0% - 154.8% │

│ Treatment │  0.5429 │   96.5%  │     90%    │  87.5% - 106.5% │

└───────────┴─────────┴──────────┴────────────┴─────────────────┘

If we multiply all values of the second period by 1.25 (i.e., adding a “true” period effect of 25%), we get a highly significant period effect (100% not within the CI); but sequence and treatment effects (and their CIs as well) stay exactly the same:

┌───────────┬─────────┬──────────┬────────────┬─────────────────┐

│   Effect  │ p-Value │ Estimate │ Confidence │     Interval    │

├───────────┼─────────┼──────────┼────────────┼─────────────────┤

│ Period    │  0.0015 │  123.0%  |     95%    │ 109.3% - 138.5% │

│ Sequence  │  0.3239 │  115.4%  │     95%    │  86.0% - 154.8% │

│ Treatment │  0.5429 │   96.5%  │     90%    │  87.5% - 106.5% │

└───────────┴─────────┴──────────┴────────────┴─────────────────┘

Sequence effects in a 2×2 cross-over study are a different cup of tea, and worth an entire lecture ;-)

As an entry point have a look at:

~~http://www.fda.gov/cder/guidance/3616fnl.pdf~~ FDA's guideline, Section VII. B.
D'Angelo G, Potvin D and J Turgeon
Carry-Over Effects in Bioequivalence Studies
J Biopharm Stat 11, 35-43 (2001)

Edit: Link corrected for FDA’s new site. [Helmut]

—
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

Dipesh Jayswal ● 2007-05-10 15:22 (6575 d 15:26 ago) @ Imran Posting: # 718 Views: 12,303	Statistical Issue (significant Formulation, Period and seq.) Post reply
	Dear Imran, u will get answers to ur questions on page no. 23 of below linked pdf document. www.egagenerics.com/doc/zanen-biogenerics.pdf I hope it will help u out to understand BA/BE. Regards, Dipesh

Imran
☆

Mumbai,
2007-05-25 15:08
(6560 d 15:40 ago)

@ Dipesh Jayswal
Posting: # 741
Views: 12,158

normalised Cmax

Post reply

Dear Dipesh

Thank you for the article. It was very informative. I have on querry related to normalised Cmax give on page 10 of the article. Does any guideline suggest to use normalised Cmax for PK and Statistical calculation. I have compared 90% CI data of Cmax and normalised Cmax, and both the results are different. so to what extent it can be use.

Please suggest.

Dr.Imran Khan
Sitec Labs.Pvt. Ltd
Mumbai

Edit: Full quote removed. Please see this post! [HS]

—
Dr.Imran Khan

Helmut
★★★

Vienna, Austria,
2007-06-30 16:57
(6524 d 13:51 ago)

@ Imran
Posting: # 851
Views: 12,268

normalised Cmax

Post reply

Dear Imran!

❝ Does any guideline suggest to use normalised Cmax for PK and Statistical calculation.

Not a suggestion, but an option.
The European Note for Guidance states in Section 3.3:

‘From the primary results, the bioavailability characteristics desired are estimated, namely AUC_t, […] or any other justifiable characteristics […]’

In the recent past some EU-regulators wanted us to include C_max/AUC for MR-products (not as a primary rate parameter, but descriptively).

Some statisticians are concerned about which method should be applied comparing C_max/AUC, since (by convention, PK-reasoning and bioanalytical grounds) both parameters are assumed to follow a lognormal distribution (evaluation by a parametric multiplicative model, e.g., ANOVA). Now what’s the statistical distribution of the ratio of two lognormal distributions?
Most people are pragmatic (simply ignoring the problem), others opt for a nonparametric method.
Unfortunately to my knowledge nobody ever has published on this topic…

❝ I have compared 90% CI data of Cmax and normalised Cmax, and both the results are different.

Different metrics give different results.
In most cases intra-subject variability of C_max/AUC is lower than C_max’s – resulting in a tighter confidence interval.

References:

Endrényi L, Yan W. Variation of C_max and C_max/AUC in investigations of bioequivalence. Int J Clin Pharm Ther Toxicol. 1993; 31(4): 184–9.
Schall R, Luus HG, Steinijans VW. Choice of characteristics and their bioequivalence ranges for the comparison of absorption rates of immediate-release drug formulations. Int J Clin Pharm Ther. 1994; 32(7): 323–8.

—
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