Bioequivalence of Highly Variable (HV) Drugs: Clinical Implications Why HV Drugs are Safer Leslie Z. Benet, Ph.D. Professor of Biopharmaceutical Sciences University of California San Francisco FDA Advisory Committee for Pharmaceutical Science Rockville, MD April 14, 2004

The Current U.S. Procrustean Bioequivalence Guidelines The manufacturer of the test product must show using two one-sided tests that a 90% confidence interval for the ratio of the mean response (usually AUC and C max ) of its product to that of the reference product is within the limits of 0.8 and 1.25 using log transformed data. (Procrustean  marked by an arbitrary, often ruthless disregard for individual differences or special circumstances.) Note: BCS is a non-Procrustean advance We should consider other non-Procrustean advances

Bioequivalence Issues What are we trying to solve? For all drugs, but particularly for NTI drugs, practitioners need assurance that transferring a patient from one drug product to another yields comparable safety and efficacy (switchability). For wide-therapeutic index, highly variable drugs we should not have to study an excessive number of patients to prove that two equivalent products meet preset (one size fits all) statistical criteria. To give patients and clinicians confidence that a generic equivalent approved by the regulatory authorities will yield the same outcome as the innovator product.

Why is meeting bioequivalence criteria a relatively minor concern for drugs with narrow therapeutic indices? By definition, approved drugs By definition, approved drugs with narrow therapeutic with narrow therapeutic indices exhibit small indices exhibit small intrasubject variability. intrasubject variability. If this were not true, patients If this were not true, patients would routinely experience would routinely experience cycles of toxicity and lack of cycles of toxicity and lack of efficacy, and therapeutic efficacy, and therapeutic monitoring would be useless. monitoring would be useless.

NTIDrugs Frequently Proposed NTI Drugs Frequently Proposed to Limit Generic Substitution CV% CV% Inter Intra Inter Intra Subject Subject Subject Subject Carbamazepine38 Conjugated Estrogens Digoxin52 Levothyroxine sodium20<20 Phenytoin sodium Theophyllin sustained release Warfarin sodium536-11

(µ T - µ R ) 2 +  D 2 + (  WT 2 -  WR 2 ) <  (µ T - µ R ) 2 +  D 2 + (  WT 2 -  WR 2 ) <   WR 2 Individual Bioequivalence (IBE) Initial Promises for IBE Initial Promises for IBE Addresses the correct question (switchability) Addresses the correct question (switchability) Considers subject by formulation interaction (  D ) Considers subject by formulation interaction (  D ) Incentive for less variable test product Incentive for less variable test product Scaling based on variability of the reference product Scaling based on variability of the reference product both for highly variable drugs and for certain both for highly variable drugs and for certain agency-defined narrow therapeutic range drugs agency-defined narrow therapeutic range drugs Encourages use of subjects more representative of Encourages use of subjects more representative of the general population the general population

Re-examination of the Initial Promises for IBE Re-examination of the Initial Promises for IBE Addresses the correct question (switchability)— Necessity questionable Addresses the correct question (switchability)— Necessity questionable Considers subject by formulation interaction— Unintelligible parameter Considers subject by formulation interaction— Unintelligible parameter Incentive for less variable test product—ABE with scaling could also solve this issue Incentive for less variable test product—ABE with scaling could also solve this issue Scaling based on variability of the reference product both for highly variable drugs and for certain agency- defined narrow therapeutic range drugs– ABE with scaling could also solve this issue Scaling based on variability of the reference product both for highly variable drugs and for certain agency- defined narrow therapeutic range drugs– ABE with scaling could also solve this issue Encourages use of subjects more representative of the general population—Failed Encourages use of subjects more representative of the general population—Failed

Highly Variable Drugs (CV>30%) For wide-therapeutic index highly variable drugs we should not have to study an excessive number of patients to prove that two equivalent products meet preset (one size fits all ) statistical criteria. This is because,by definition, highly variable approved drugs must have a wide therapeutic index, otherwise there would have been significant safety issues and lack of efficacy during Phase 3 Highly variable narrow therapeutic index drugs are dropped in Phase 2 since it is not possible to prove either efficacy or safety.

Progesterone The Poster Drug for High Variability A repeat measures study of Prometrium® 2x200 mg capsules in 12 healthy post-menopausal females yielded: Intrasubject CV for AUC of 61% Intrasubject CV for C max of 98% A generic company calculated that a 2 period crossover BE study for Progesterone Capsules, 200 mg would require dosing in 300 postmenopausal women to achieve adequate statistical power

Recommendations of the FDA Expert Panel on Individual Bioequivalence to this Advisory Committee on November 21, 2001 Sponsors may seek bioequivalence approval using either ABE or IBE (with SxF deleted) Scaling of ABE should be considered. If an IBE study is carried out and the test product fails, the data or a subset of the data may not be reanalyzed by ABE for approval A point estimate criteria on mean AUCs of ±15% and on mean C max of ±20% should be required for both ABE and IBE. Consideration should be given for narrower point estimate criteria for NTI drugs (e.g., AUC ±10%, C max ±15%)

Recommendations Methodology should be developed to allow approvals based on weighting of average bioequivalence analyses for highly variable drugs (i.e.,  WR > 30%). A point estimate criteria on mean AUCs of ±10% and on mean C max of ±15% should be required for NTI drugs where  WR  20%. A point estimate criteria on mean AUCs of ±15% and on mean C max of ±20% should be required for all other drugs, including NTI drugs where  WR > 20%.