| Comment Record|
Mr. Charles DiLiberti ||
2002-01-28 15:45:59 |
Barr Laboratories, Inc. |
| Comments for FDA General |
1. General Comments
Because of the significant impact that the 90% confidence interval requirement would have on how fed BE studies are powered, it is critical that if the Agency does indeed implement the confidence interval criteria, the effective date of this particular requirement be stated explicitly, and that the Agency provide for a grace period, during which firms may still apply the current (point estimate) BE criteria. It is also critical to specify what event (IRB approval, first dosing, etc.) qualifies a BE study to be accepted under the current (point estimate) BE criteria. The Agency should consider issues of fairness where one generic firm obtains approval under the old (point estimate) criteria, and another firm must meet the new confidence interval criteria simply because it conducted its fed BE study later.
In our view, the fairest solution to this problem would be (1) to give firms sufficient advance notice of the implementation of the confidence interval criteria (i.e., a grace period of at least 6 - 12 months), (2) to provide a date certain (grace period expiration date) on which the new confidence interval criteria will become mandatory, and (3) to specify the qualifying event (preferably, administration of the first dose) which must occur before the expiration of the grace period if the study is to be accepted with the current (point estimate) criteria.
At this point, the industry does not know whether the confidence interval criteria will or will not be implemented. Nevertheless, every week, the industry continues to run food effect BE studies, and must be allowed to proceed with FDA's assurance that these studies, powered for point estimate criteria, will be accepted. The industry cannot afford to power food effect studies to meet confidence interval criteria in anticipation that the Agency might implement the confidence interval criteria. To exemplify the increase in cost required by the confidence interval approach, a 12 subject fed study will provide about 80% power to meet the current 80 - 125% point estimate criteria for a drug with an intrasubject CV of 42% and a true test/reference ratio of 0.95. In order to achieve 80% power to meet the proposed 80 - 125% confidence intervals for the same drug, about 72 subjects would be required. This translates directly to a six-fold increase in study cost.
If FDA simply were to implement the confidence interval acceptance criteria in the draft food effect guidance without a grace period, some firms may be caught short, and find that they are required to conduct much larger studies than their competitors that happened to have conducted their studies earlier. Some firms with studies in progress or planned for the near future might also be caught short and find themselves having to rerun much larger fed studies.
Barr's approach would allow firms to conduct food-effect BE studies powered for point estimate criteria for a grace period, and then provide for an orderly transition to the new confidence interval criteria. The entire industry would know precisely how to design fed biostudies based on when the study is to be dosed. Firms that are seeking approval on products for which generic competitors have already filed point estimate-powered fed studies would be given a window within which to conduct similar point estimate-powered fed studies, and thus receive equal treatment.
Clearly, if the Agency chooses to retain the current point estimate criteria permanently, the preceding comments do not apply.
Lines 190 - 197
Subjects that are unwilling or unable to consume all or part of the test meal for whatever reason (e.g., religious preference, dietary preferences, food allergy, lactose intolerance, etc.) should be excluded from enrollment in a food-effect BA or fed BE study. Barr believes that the proposed guidance should state this explicitly.
Lines 203 - 204
While it is generally true that the same strength test product used for the fasting BE study should be used for the fed BE study, there are a variety of circumstances under which it may be difficult or impossible to use the same lot of test or reference product for both the fasting and fed BE studies. For example, production delays, a desire to conduct stability studies before dosing, business delays, or simply a lengthy fasting BE study design may result in the expiration of the test or reference lots used in a fasting BE study before a fed BE study can be undertaken. There might also be situations in which a generic firm might need to conduct the fasting BE study against one strength of reference product and the fed BE study against a different strength of the same reference product (e.g., for an ANDA suitability petition for a new strength which is later also added by the innovator). While it may often be possible to conduct fasting and fed BE studies using the same lot of test and reference products, requiring firms to do so would result in unnecessary human experimentation for no valid scientific reason.
Barr believes that modifying the language in the draft guidance to the following would clarify the point:
For ANDAs, the same strength of test product used in the fasting BE study should be used as the test product in the fed BE study.
Barr has concerns that the proposed test meal will prove to be more variable in composition than the current test meal. In particular, bacon can vary tremendously in fat content, depending on the extent to which it is cooked. Canadian bacon, on the other hand, appears to be much more consistent in composition. Consequently, Barr supports the use of the standard breakfast specified in the October 1997 version of the draft food effect study guidance.
lines 220 - 221
Barr is aware of several products for which the innovators conducted their fed BA studies in the lightly fed state (e.g., following a low fat breakfast or a light snack). In such cases where the innovator has not assessed BA following a high-fat meal, it is unfair to require generics to conduct their food effect studies following the proposed high fat test meal. In such cases, generics should be permitted to demonstrate BE in the fed state using the same type of meal or snack as was used by the innovator.
Lines 236 - 237
Because of the practical timing difficulties in linking the administration of the dose to the completion of the test meal, Barr believes that the dose should be administered 30 minutes after the initiation of the test meal.
The test meal should be consumed within 30 minutes, with administration of the drug product 30 minutes after initiation of the test meal.
Barr opposes the requirement to calculate lag times because of their high variability and dependence on pharmacokinetic model assumptions.
Lines 271 - 272
To avoid the possibility that Tmax might be subjected to inappropriate log transformation, Barr recommends that the text be changed as follows:
Log-transformation of AUC and Cmax prior to analysis is recommended.
Lines 284, 303, 316
Barr believes that the term population geometric means is technically incorrect and may be misleading. Barr proposes the use of the term least-squares geometric means instead.
Lines 335 - 338
Barr believes that conducting fasting and fed BE studies brackets the state in which a sprinkle BE study as described is conducted. The sprinkle study amounts to a fed BE study with a small amount of food rather than a standard high-fat meal, and thus, falls between the fasting and fed state. Demonstrating BE in the fasting state as well as following a standard high-fat meal is therefore sufficient to demonstrate BE for any intermediate state, such as that proposed for a sprinkle BE study. Requiring the conduct of sprinkle BE studies as proposed, therefore, should be deleted, as it amounts to requiring unnecessary human experimentation. Barr believes that sprinkle BE studies should only be required for drug products whose labeling permits administration in the fasting state or sprinkled on a small amount of soft food, but not with meals. Demonstrating BE for such products would then require conducting a fasting BE study and a sprinkle BE study, but not a fed BE study with a standard high-fat test meal.