Bioequivalence and Bioavailability Forum

Hi Kumar,

❝ Actually we got pre-screening query from FDA for one of our study that table 3A is not filled. In that study we had Swr >0.294 for all three parameters thats why we filled table 3B and in table 3A we put NA.

Amazing response. To make the FDA happy, you could give the ratio and 90% CI from Table 3B also in Table 3A. Stupid, IMHO.

❝ After seeing this query some CRO experts suggest that we should also do unscaled approach for filling table 3A.

Experts? The guidance unambiguously states on page 8:

“For PK parameters with a s_WR < 0.294, use the unscaled average bioequivalence approach.”

That’s definitely not the case in your study. Furthermore, the ABE-code given by the FDA some­times fails to converge if the study was done in a partial replicate design (doesn’t matter which software one uses!). This is due to the over-specified model in coincidence with the specification of the covariance structure as TYPE=FA0(2). If that happens one would have to use TYPE=FA0(1) – which is against the guid­ance.
A nice detail: The 90% CI from ABE does not equal the one obtained from RSABE (different effects in the models…).
Example* from Patterson/Jones, 90% CIs:

           AUC             Cmax     
────────────────────────────────────
ABE    96.25–116.15%  118.75–158.55%
RSABE  96.27–116.13%  118.66–158.67%

Would the CROs’ “experts” suggest to confuse the FDA by giving the first row in Table 3A and the second one in Table 3B? I would expect yet another query from FDA’s review staff approaching:

“The 90% Confidence Intervals reported by the applicant in Tables 3A and 3B are inconsistent. Please justify.”

❝ But I think if we are doing by scaled method here because of high variability then what is the purpose of applying unscaled method here.

Exactly.

---

• Patterson SD, Jones B. Viewpoint: observations on scaled average bioequivalence. Pharm Stat. 2012;11(1):1–7. doi 10.1002/pst.498

---

PS: IMHO, the “logic” of FDA’s form is bureaucratic crap.

Table 3A Statistical Summary of the Comparative Bioavailability Data for
Unscaled Average BE Studies
Reference Scaled Average Bioequivalence Approach Used   ☐ Yes  ☐ No
If Yes, then complete Tables 3A and 3B

I see! We tick ☑ Yes for RSABE and have to give in Table 3A the Ratio and 90% CI for Unscaled ABE. Bizarre. I’m fine with the geometric LSMs and Ns, of course.

Table 3A consists of two parts (fasting/fed studies), but there is only one in Table 3B. I beg your pardon? Pros at work…