Bioequivalence and Bioavailability Forum • Add-on study

Priyanka_S
☆

2010-01-22 13:40
(5586 d 15:48 ago)

Posting: # 4642
Views: 4,220

Add-on study [Two-Stage / GS Designs]

Dear Helmut,

We conducted a 2-way crossover BE study for CANADA. But it has failed on upper side (135% for Cmax, 130% for AUCt) and Power is 90% and 95% for Cmax & AUCt respectively.

Now we are going to conduct a Add-on study for this. As per CANADIAN Guidance

If BE is not shown, additional subjects are included.

1) If an F-test (equality of variances) shows no significance, we can go for a pooled analysis.

Does it means (F-Test)

F=Mean Square Error (from ANOVA) of Cmax for Study1 / Mean Square Error of Cmax for Study2 Should be insignificant?

In the same way for AUCt.

Plz let me know the above is correct or not.

—
Best Regards

Priyanka S

Helmut
★★★

Vienna, Austria,
2010-01-22 16:49
(5586 d 12:39 ago)

@ Priyanka_S
Posting: # 4643
Views: 3,558

Add-on study (Canada)

Post reply

Dear Priyanka!

❝ We conducted a 2-way crossover BE study for CANADA. But it has failed on upper side (135% for Cmax, 130% for AUCt) …

OK. Well, not OK…

❝ … and Power is 90% and 95% for Cmax & AUCt respectively.

Please, no more a posteriori power calculations… See this post and others linked there.
BTW, where did you get these numbers from?

❝ Now we are going to conduct a Add-on study for this. As per CANADIAN Guidance

❝ If BE is not shown, additional subjects are included.

Yes. See the guidance:

If the study is run with the appropriate size and the standards are not met, the sponsor may add more subjects (a minimum of 12). If this option is chosen, it must be stated in the study protocol. Two criteria must be met before combining is acceptable:

The same protocol must be used (i.e., same formulations, same blood sampling times, a minimum number of 12 subjects, etc.),

Consistency tests must be met at an ALPHA error rate of five percent.

(my emphases)

❝ 1) If an F-test (equality of variances) shows no significance, we can go for a pooled analysis.

Yes - but remember: add-on only if stated in the protocol.

❝ Does it means (F-Test)

Yes.

❝ F=Mean Square Error (from ANOVA) of Cmax for Study1 / Mean Square Error of Cmax for Study2 Should be insignificant?

❝ In the same way for AUCt.

Yes. Significance test at alpha 0.05. Don't forget to run the F-test with the degrees of freedom of studies 1&2. You also have to test for formulation × study interaction. For details see the guidance pp22:

7.4.2 Consistency Tests for Combining Studies
b) Two consistency tests must be done on the studies to ensure that pooling is
meaningful.
The first test is the test of equality of the residual mean squares. Take the ratio of the residual from the first study to the residual for each of the other studies. (For each ratio, the smaller of the two residuals must be used as the denominator.) This ratio is compared to an F statistic, using the degrees of freedom associated with the residuals. If the ratio is greater than the 5% F value, the studies may not be combined. The second test is the formulation by study interaction and is carried out as shown in Table 7-A.

TABLE 7-A
Pooled ANOVA for Testing Formulation by Study Interaction from S Two-period
Cross-over Studies of Size N_I


                      df         F-Ratio Denominator  P-Value

Study                 S-1        Subject(Study)       -

Subject (Study)       Sum(N_I-1)  Residual             -

Period (Study)        S          Residual             -

Formulation           1          Residual             -

Formulation by Study  S-1        Residual             >.05

Residual              Sum(N_I-1)  -                    -

P.S.: According to the drafts (28 Nov 2009) presented at the workshop in Ottawa HPB/TGD is considering to implement Gould's method, and - sursprisingly - not the method by Potvin et al.

—
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