1. Highly Variable Drugs – Bioequivalence Issues: FDA Proposal Under Consideration
Barbara M. Davit, J.D., Ph.D.
Deputy Director, Division of Bioequivalence (DBE)
Office of Generic Drugs (OGD)
Office of Pharmaceutical Sciences (OPS)/CDER
Advisory Committee for Pharmaceutical Science
October 6, 2006

2. Discussion topics Characteristics of highly variable (HV) drugs
Present bioequivalence (BE) study approach used by the OGD for all drugs
Disadvantages of using current approach for HV drugs
Reference-scaled average BE approach under consideration by FDA
Advantages/concerns
Questions for committee

3. HV Drug Characteristics
Within-subject variability (CVWR) in BE parameters AUC and/or Cmax > 30%
Non-narrow therapeutic index
Represent about 10% of drugs studied in vivo and reviewed by OGD

4. Some reasons for high variability in BE parameters
Drug substance
Variable absorption rate
Low extent of absorption
Extensive presystemic metabolism
Drug product
Inactive ingredient effects
Manufacturing effects
Bioanalytical assay sensitivity
Suboptimal PK sampling
Impractical to identify mechanism in each case

5. Characteristics of HV drugs evaluated by OGD
Can use Root Mean Square Error (RSME) to estimate within-subject variability in two-way crossover studies
Conclude drug is HV if RMSE > 0.3
Using this criterion, about 10% of drugs evaluated by OGD are HV drugs; of these
55% are consistently HV
20% are “borderline” cases
For the remaining 25%, high variability occurs sporadically (not HV in most BE studies)

6. BE issues with HV Drugs
High probability that BE parameters will differ when same subject receives a HV drug on more than one occasion
Because of the high variability, a HV drug that is truly therapeutically equivalent to the reference may not meet BE acceptance criteria

7. Present FDA approach used for BE studies of HV drugs
Generally, firms submitting ANDAs for HV drugs use the same study design as for drugs with lower variability
Two-way crossover study
Replicate-design study
HV drugs must meet same acceptance criteria as drugs with lower variability
90% CI of AUC and Cmax test/reference ratios must fall between limits of 0.8 to 1.25 (80-125%)

8. Disadvantages of present FDA approach for HV drugs
Enroll adequate # of subjects (N) to show BE in 2-way crossover study
Study may require larger N
avg. N = 47 for HV drugs *
avg. N = 33 for other drugs *
If study underpowered, must do new study
Replicate-design (4-period) study
High dropout rate; may need to enroll larger N
Group sequential-design study
Must have protocol in place a priori;
Statistical adjustment
* Based on BE studies submitted to OGD in

9. Evolution of new proposal for BE studies of HV drugs
Pharmaceutical Sciences Advisory Committee in 2004 suggested reference-scaled average BE approach
OGD Science Team studied approach by simulating outcome of BE studies of HV drugs
FDA is considering using this approach for BE studies of HV drugs

10. New FDA proposal: scaled average BE for HV drugs
Three-period BE study
Provide reference product (R) twice
Provide test product (T) once
Sequences = TRR, RRT, RTR
BE criteria scaled to reference variability
(µT - µR)2 - Өσ2WR < 0 ; Ө = upper BE limit
Both AUC and Cmax should meet BE acceptance criteria

11. Advantages of reference-scaled average BE
If T variability < R variability, will benefit test product
If T variability > R variability, no benefit for test product

12. Use of scaled average BE for borderline HV drugs
Our simulations confirmed that, for a true “borderline” HV drug*, either scaled or unscaled average BE approach is suitable
Outcome of a 3-way crossover BE study will be similar whether a reference-scaled average BE analysis or unscaled average BE analysis is conducted
* We define a borderline HV drug as one for which, in individual studies, within-subject variability in BE parameters is generally slightly > or < 30%, and the average within-subject variability is about 30%

13. When scaled average BE approach is unsuitable
HV due to generic product or study conduct
If due to effects of generic formulation, will not benefit from scaled average BE approach
If T variability > R variability
Studies poorly performed
Burden on applicant to prove to OGD that drug substance is HV
OGD can conclude that scaled average BE approach is unacceptable

14. Concerns about reference-scaled average BE
Proposed solution
Firms will conduct a replicate-design study, then submit results with both scaled and unscaled BE analyses
If CVWR > 30%, FDA will use the reference-scaled average BE approach
If CVWR < 30%, FDA will use the unscaled average BE approach

15. Concerns about reference-scaled average BE
Proposed solution
Scaling can allow the resulting AUC and Cmax geometric mean ratios to be either unacceptably low or high
Acceptance criteria can include a point estimate constraint

16. Concerns about reference-scaled average BE
Proposed solution
What should be an appropriate number of subjects for a BE study that uses this approach?
Should the FDA recommend a minimum number of subjects?

17. Acknowledgements OGD Working Group DBE Research Group Mei-Ling Chen
Dale Conner
Sam Haidar
Lai Ming Lee
Rob Lionberger
Fairouz Makhlouf
Devvrat Patel
Don Schuirmann
Lawrence Yu
DBE Research Group
Beth Fabian-Fritsch
Sheryl Gunther
Xiaojian Jiang
Devvrat Patel
Keri Suh
Christina Thompson