Bioequivalence and Bioavailability Forum

Hi Mittyri,

❝ ❝ I don’t see how you could perform a dose-proportionality study with just two levels (zero degrees of freedom).

❝

❝ But we have more than one observation for each level, so regression is possible, isn't it?

Yeah, well, cough…

Hi Mikkabel,

❝ I would also opt for a simple Pk study (T2 versus 2*T1) but do you have any idea to which guidelines I can refer to justify the rationale of the study and in particular the choice of the comparator (T1 instead of the originator).

See Annex I to EC 1234/2008 about Type II variations …

2. Changes to strength, pharmaceutical form and route of administration:
    (c) change or addition of a new strength/potency;

… and the Q&A about extensions of marketing authorisations.
In the line extensions I was involved in (example) we always used the approved strength of our product as the comparator (and not the innovator’s product).¹

❝ Regarding the dose proportional study, I meant a PK study (T1 vs T2) and the results corrected by the dose should be compared.

I did it once in 2003.² The German BfArM told me “OK, we accept it this time but please don’t do it in the future…”. Makes perfect sense indeed (I learned): Demonstration of BE requires administration of the same molar dose(s).³

Let T₂ = ƒ×T₁, F the absolute BA, and R the response in any given PK metric of interest (AUC, C_max, …).

• If (!) F_T2 ≡ F_T1 and CL_T2 ≡ CL_T1 (a general assumption in BE) then R_T2/(ƒ×R_T1) = 100%. Brilliant.
  
• Now F_T2 < F_T1 but at the same time we run into saturation of a metabolizing enzyme with the high strength. R will be higher since the drug is slower eliminated (CL_T2 < CL_T1). If we perform the dose-cor­rection (in the worst case) we get also 100% since the lower F of T₂ is “masked” by its lower clearance.
  If we would have administered the same doses we would have seen the lower F of T₂.

An example:
D_T1 1 (F 100%) ▬▬, D_T2 2 (F 90%) ▬▬, and at the high dose elimination is prolonged by ~12% due to saturation.

If we dose-correct T₁ ▬▬, we get for T₂ – false – GMR estimates of 100% for AUC_0–t, 103% for AUC_0–∞, and 94% for C_max. We wouldn’t adjust individual profiles. I expect that the rather small prolongation of the elimination would not be evident in the data (hidden by biological noise). And no, we would never notice the seven minutes delay in t_max.

However, if we administer the same doses, elimination should be prolonged to a similar extent (CL_T2 ≈ CL_T1).

Hence, GMR estimates for all PK metrics would be unbiased (≈90%)⁴ reflecting the true relative BA.

---

1. I’ve heard rumors that in MENA states the originator is preferred by some assessors. If possible, opt for a scientific advice.
   
2. It was an intermediate strength and strict dose-proportionality was already established.
   
3. BE-GL, Section 1.1:
   Two medicinal products […] are considered bioequivalent if […] their bioavailabilities (rate and extent) after administration in the same molar dose lie within acceptable predefined limits.
   
4. Due to F_T2 < F_T1 (less drug in the circulation after T₂) we challenge the enzyme less than with 2×T₁. Hence, the estimates might be slightly higher than 90%.