Specifications for the Nonconformist

1
Specifications for the Non-conformist

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

2
Specifications for the
Non-conformist

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

3
Specifications for the
Non-conformist

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

4
Specifications for the
Non-conformist

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

5
Non-conformist
Specifications for the

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

6
Non-conformist
Specifications for the

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

7
Non-conformist
Specifications for the

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

8
Non-conformist
Specifications for the

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

9
Non-conformist
Specifications for the

• Philip Krause, M.D.
• Deputy Director
• Division of Viral Products
• FDA/CBER/OVRR

10
Non-conformist One who does not conform to, or
refuses to be bound by, accepted beliefs,
customs, or practices

• Appeal to people who are interested in new ways
  to think about specifications
• Imply some insight into how to handle results
  that dont conform to initial expectations
• Implied disclaimer Opinions are not FDAs, but
  are based on my experiences as a reviewer, as
  influenced by many smart people

11
Purposes of specifications

• Confirm absence of critical contaminants
• Confirm that released lots are similar to lots
  tested in clinical trials
• Confirm that manufacturing process is under
  control

12
Specifications and limits imply that a comparison
is taking place
13

• Depending on the reason for setting any given
  specification, the comparator may be different

14
Purpose of specifications

• Confirm absence of critical contaminants
• Confirm that released lots are similar to lots
  tested in clinical trials
• Confirm that manufacturing process is under
  control

15
Confirm absence of critical contaminants

• Examples sterility, adventitious agent tests
• Generally, all released lots need to be tested
• Comparator is an ideal standard
• Products should be free of extraneous agents
• Its impossible to prove a negative
• Ability to support assertion that a large sample
  is free of adventitious agents depends on the
  quantity being tested and on the sensitivity of
  the assay

16
For absence specifications

• The most important issue isnt setting of a
  release specification (by definition, zero or
  non-detectable), it is determining how much
  sample must be shown to be negative using a given
  assay
• This depends on the risk associated with false
  negatives
• Process validation also plays a major role in
  this type of assurance

17
Purpose of specifications

• Confirm absence of critical contaminants
• Confirm that released lots are similar to lots
  tested in clinical trials
• Confirm that manufacturing process is under
  control

18
Confirm that released lots are similar to
clinical lots

• Example Potency testing
• Generally, all released lots need to be tested
• To compare current product with clinical lots,
  one needs to understand the material tested in
  clinical trials (including post-marketing
  experience)
• Potency specs are inextricably linked to
  stability data

19
Investigation of stability

• Perform regression analysis to determine the
  rate constant, k

20
Setting the shelf life
U.L.
Potency
95 lower bound on potency estimate
L.L.
Time
Shelf life
21
Setting the shelf life
U.L.
Potency
L.L.
Time
Shelf life
22
Setting the shelf life
U.L.
Potency
L.L.
Time
Shelf life
23
Setting the shelf life
U.L.
Potency
L.L.
Time
Shelf life
24
Setting the shelf life
U.L.
Potency
L.L.
Time
Shelf life
25
Setting the shelf life
U.L.
Potency
L.L.
Time
Shelf life
26
Using Release Specifications to Protect the
Customer

• Lower potency release specification Should
  assure that doses will have a minimum potency
  shown to be effective
• Upper potency release specification Should
  assure that doses will not exceed a potency
  demonstrated to be safe
• Ideally, the release model will account for all
  the uncertainties associated with providing these
  assurances

27
Accounting for uncertainty in data used to assign
release potencies

• Release Potency Minimum Potency expected
  loss due to instability error term
• Error term encompasses 95 confidence in
• Uncertainty in potency determination
• Uncertainty in stability determination
• Uncertainty in interpretation of clinical data

28
Are there other uncertainties that should be
considered?

• Other uncertainties that might influence ideal
  release potencies
• Uncertainty introduced by Normal approximations
• Uncertainty associated with stability modeling
• Uncertainty introduced by potential drift in
  assays
• Uncertainty in relevance of potency assay to
  safety/efficacy
• Is it possible to quantify all these
  uncertainties?
• If not, can we place bounds on them?

29
Purpose of specifications

• Confirm absence of critical contaminants
• Confirm that released lots are similar to lots
  tested in clinical trials
• Confirm that manufacturing process is under
  control

30
Confirm that manufacturing process is under
control

• Examples Tests to identify clinically irrelevant
  variants, bulk potencies, mock production runs
• You need prior manufacturing experience as a
  comparator
• Comparison may be dynamic, as prior manufacturing
  experience keeps changing
• Need to think about which prior experience yields
  the most relevant comparisons
• May be achieved with limits, as opposed to
  specifications
• All tests dont necessarily need to be done on
  all lots

31
Monitoring Product StabilityAgainst
Specifications

• What is the purpose of stability monitoring?
• What measures of stability can reasonably be
  monitored?
• Can feasible measures of stability fulfill the
  purpose of stability monitoring?

32
A Previous Paradigm for Shelf Life
U.L.
Potency
L.L.
33
A Previous Paradigm for Shelf Life
U.L.
Potency
L.L.
34
A Previous Paradigm for Shelf Life
U.L.
Potency
L.L.
35
A Previous Paradigm for Shelf Life
U.L.
Potency
L.L.
Shelf Life
36
A Previous Paradigm for Shelf Life
U.L.
Potency
L.L.
Shelf Life
37
A Previous Paradigm for Shelf Life
Problems Not necessarily a good
estimate Confidence in value is not
defined Conceptually flawed more determinations
increase the risk of failure
U.L.
Potency
L.L.
Shelf Life
38
Is stability monitoring meant to show
acceptability of a single lot, similarity to
previous manufacturing experience, or something
else?
39
If goal is to show acceptability of the tested
lot

• If it is important for all lots to meet
  specifications at all time points, then testing
  only a subset of lots would be inappropriate
• It is difficult to decide how to extrapolate
  results from stability monitoring studies to
  other lots
• Using multiple tests over time to evaluate
  individual lots has the same flaws as using these
  tests to assign shelf life
• Its not appropriate to test into compliance
  why would we want to test into failure?

40
If goal is to show similarity to previous
experience

• Compare current stability with previous stability
  estimate that was used to determine shelf life
• Best comparison may be between regression lines
  (assuming first order kinetics)
• Need to use statistics to show that regression
  lines are similar
• Need to define similarity
• Need some assurance that meaningful differences
  would be detected (beta error)
• Need to confirm assumptions (e.g., kinetics)

41
If goal is something else

• According to 21CFR211.166
• There shall be a written testing program
  designed to assess the stability characteristics
  of drug products. The results of such stability
  testing shall be used in determining appropriate
  storage conditions and expiration dates.
• Implies that one should use data from stability
  monitoring to refine stability estimates, and
  thus, release potency specs

42
Are there other uncertainties that should be
considered?

• Other uncertainties that might influence ideal
  release potencies
• Uncertainty introduced by Normal approximations
• Uncertainty associated with stability modeling
• Uncertainty introduced by potential drift in
  assays
• Uncertainty in relevance of potency assay to
  safety/efficacy
• Potential for drift in stability, as detected by
  stability monitoring studies
• Is it possible to quantify all these
  uncertainties?
• If not, can we place bounds on them?

43
Conclusion

• Purpose of specifications
• Confirm absence of critical contaminants
• Confirm that released lots are similar to lots
  tested in clinical trials
• Confirm that manufacturing process is under
  control
• Depending on purpose of any given specification,
  derivation of the most appropriate acceptance
  criterion is different