Bioequivalence and Bioavailability Forum

Hi Naghma,

without information about the country/regulation you are aiming at, it is difficult to answer. The primary purpose of the WHO’s guidelines is to support regulatory agencies of countries which don’t have their own ones so far – not applicants. Hence, the WHO’s GLs are often ambiguous (listing different approaches to select from). Until you find out which path “your” regulator will follow, picking out something from the WHO’s GLs is risky. The GL you are referring to is a good example.

❝ 1. Bioequivalence will be plan in fed state or fasting?

See what is stated in the section “Pharmacokinetics of Sofosbuvir” about the food effect, especially

[…] administration is recommended with food in the Summary of Product Characteristics approved by the EMA, although it can be taken irrespective of meals according to the Product Labelling approved by the FDA.

(my emphases)
You were quoting the (August 2015) label of the FDA’s RLD. Since for IR products the fasting state is considered more sensitive to detect differences between formulations – and the reference product could be administered regardless the pranial state – you would be right. However, the FDA’s product-specific guidance (June 2015) recommends two studies (fasting, fed).*
If your “target regulation” follows the WHO’s product-specific GL or the EMA’s general BE-GL (administration according to the SmPC), only one study in fed state.*

❝ 2. should the sofosbuvir metabolite also analyzed?

No (acc. to WHO, FDA, EMA).

❝ 3. What will be the minimum sample size?

Depends on which minimum power is acceptable for you. Many GLs recommend 80 – 90%.
The WHO’s GL states that C_max might be highly variable (CV_wR 54%) in fed state. If your target regulation accepts reference-scaling (widening of the acceptance range) you could follow the EMA’s approach given in the BE-GL (Average Bioequivalence with Expanding Limits, ABEL). If not, the common acceptance range of 80 – 125% for ABE is applicable. For HVD(P)s a T/R-ratio of not “better” than 0.90 should be assumed (regardless whether you intent to design the study for evaluation by ABE or ABEL). If you plan for 80% power this translates into the following sample sizes:

  design     ABE  ABEL
──────────────────────
RT|TR        230   NA 
RTRT|TRTR    116   28 
RTR|TRT      172   44 
RRT|RTR|TRR  174   42 

If your regulation accepts the FDA’s Reference-scaled Average Bioequivalence (RSABE) that would mean:

  design     RSABE
──────────────────
RTRT|TRTR     22  
RTR|TRT       34  
RRT|RTR|TRR   30  

Note that AUC is not highly variable (CV_wR 10%) and you have to evaluate it for ABE. In some cases the FDA’s mixed effects model fails to converge in the partial replicate design (RRT|RTR|TRR) – you have done the study, but the software cannot give you a result. Therefore, avoid this design.

---

• BTW, this makes an interesting story. Obviously the originator showed in one study lacking food effects (which lead to the FDA’s label) and in another one a 1.8fold increase in AUC (which lead to the EMA’s SmPC). Strange, politely speaking. The FDA seems to be aware of that, since asking in the product-specific guidance for two studies. If the FDA trusts in the study which lead to approval why are they asking for a fed study as well? If they have other information (which contradicts the approved label) why they don’t ask the originator for the data? Given the low variability of AUC no statistics is needed to see a significant food effect with a PE of ~180%. Regulators are a strange bunch of people sometimes.
  According to the Australian PAR the originator performed two food effect studies:

  The findings of both were in agreement; compared with the fasted state, food, specifically a high-fat meal, resulted in a slower rate of absorption of SOF [Sofosbuvir] (high-fat meal vs. fasted; prolonged T_max: 1.5 vs. 0.5 hrs) with no substantial alteration in the extent of absorption (high-fat meal vs. fasted; mean AUC_inf increased 67–91%). […] The equivalence criterion for a lack of food effect was not met; however, C_max decrease was not considered clinically significant. Moreover, whilst SOF dosing in Phase 2 and 3 clinical studies was recommended without regard to food, in reality, SOF when co-administered with RBV [Ribavirin] in Phase III studies, was dosed with food, as required in the RBV prescribing information.

  (bold in the original)
  Fascinating that an increase of AUC by ~80% was judged Down Under to be not substantial…