Transcript Slide 1

Regulatory Aspects of
Comparative Trials
Russell Katz, M.D.
Director
Division of Neurology Products
CDER
1
Summary of Issues
• Areas in which various aspects of
“comparativity” are raised
• Examples of comparative claims
• Principles for “fair comparisons”
2
Is There a Need for
Comparative Information on
Neurologic Treatments?
• Not from a regulatory standpoint (unless a
comparative claim is being sought)
• From a clinical standpoint, valid
comparative efficacy and safety
information would be valuable
• Some argue that this type of information is
the only useful information and “me too”
drugs should not be approved
3
Differences in Efficacy for Neurologic Drugs
Within Defined Classes
• Generally difficult to demonstrate for members within
various classes
• Claim rarely granted (or even studied; although we
sometimes ask!)
– Second line indication may imply efficacy in a
refractory population, but usually no data to support
this conclusion
– Some approval decisions (or decisions to keep a drug
on the market) rely on “belief” in superiority
– Rarely, we require a showing of superiority (related to
a serious AE)
4
Differences in Safety for Drugs
Within Defined Classes
• Often acknowledged for members within various
classes in clinical practice (but rarely in labeling)
• Common reason for selection of specific drug
• Examples:
– Hypnotics-Fewer residual effects (shorter T1/2)
– Drugs for AD-Fewer adverse events
– Ease of dosing (also better compliance?)
• Once a day (more on this later)
• Patch
• Intranasal
5
Approaches to Comparative Claims
• Efficacy Claims
– Superiority
– Equivalence (always noninferiority, i.e.,
“as good as” a standard)
• Safety Claims
– Superiority
– Equivalence (always noninferiority, i.e.,
“no worse than” a standard)
6
Study Design to Support Superiority Claim
• Standard RCT and hypothesis testing (as for
placebo comparison)
– Null hypothesis: new drug = standard drug
– Alternative hypothesis: new drug > standard
drug
– Goal: Reject null hypothesis, in favor of new
drug
• Ideally have placebo as well
– If not, need to be assured that standard drug
not making patients worse (for efficacy)
7
Study Design to Support
Noninferiority Claim
• Noninferiority design
– Null hypothesis: (standard drug - new drug) >
noninferiority margin
– Alternative hypothesis: (standard drug - new
drug) < noninferiority margin
– Goal: Reject null hypothesis, in favor of
noninferiority hypothesis
• Rarely considered interpretable in neurology
8
Examples of Comparative
Efficacy Claims
• Size of the effect (e.g., % responders)
or mean change
• Time of onset of response
• Duration of effect
• Specific aspect of illness (e.g.,
seizure severity)
9
Examples of Comparative
Safety Claims
•
•
•
•
Absence of AE
Lower incidence of AE
Less severe AE
Shorter duration for AE
10
The Key Question for Comparative
Trial
• Is it a fair comparison?
• Applies whether the comparison
is for safety or efficacy
11
Principles for Fair Comparison
• Equi-effective doses
• Optimal use of comparator agent
– Dose schedules
– Titration schedules
– Use of adjunctive medication if appropriate
• Appropriate comparator(s)
• Sensitive assessment methods
• Population studied
– Unbiased
– Broad sampling
• Comparisons in all important domains
• Comparisons at all relevant times
• Replication
12
Need for Equi-Effective Doses in
Safety Comparison
• Problem Case: Low dose of new drug vs high
dose of active standard
• Example: Drug may have less dyskinesia in PD,
but be less effective than standard (probably
would not require true non-inferiority)
• Remedies:
– Multi-arm Trial (compares dose/response
curves of the two drugs)
– Compare high dose of new drug with wide
dose range of standard comparator
13
14
Examples of Failure to Dose Comparator
Agent in Optimal Manner
• Give comparator in AM when optimally
given in PM
• Use more rapid titration of comparator
agent than is conventional
15
Examples of Bias in Selection of
Comparator Agent
• Using the most sedating AED in class
• Using the hypnotic with the most next-day
residual effects
• Using the atypical antipsychotic with most
prominent weight gain
• Remedy: Requiring comparisons with
representative members of the class
16
Examples of Bias in Selection of
Assessment Method
• Failure to use sensitive instrument for
sleep attacks or ICD
• Failure to use orthostatic testing
• Failure to assess frequently and
sensitively during discontinuation phase
• Failure to obtain ECGs at Tmax
• Remedy: Require the most sensitive
instrument and method for the safety
finding of interest
17
Examples of Population Bias
• Selecting patients who have history of
intolerance to comparator
• Selecting patients who have history of
tolerating the new drug
• Selecting patients who are not vulnerable
to the event of interest
• Remedy: careful attention to population
sampled
18
Importance of Assessing Drugs Across
the Domains of Interest
• Narrow focus on one safety domain
may obscure the “big picture”
• Cannot fairly assess 2 drugs unless
they are compared on adequate
range of safety and efficacy outcomes
19
Importance of Comparing Drugs
at All Relevant Times
• Particularly important for drugs with different timeconcentration profiles
– Example: different anti-spasticity drugs may have very
different PK/PD profiles, and need to be evaluated
across the entire time interval of interest, including
evening and night
• Need to observe for both early and late-emerging
adverse events
– Example: new drug might have early advantage for
certain safety outcomes (e.g., acute headache), but
disadvantage longer-term (e.g., cardiovascular risk)
20
DNP Principles for Fair Comparison
Regarding Efficacy Outcome
• Many of same principles for safety apply
• Points to emphasize:
– Need to select an assessment that is fair for
both agents
• Do the drugs move the same
measurements?
– Need to explore full dose range for both drugs
• Here the typical bias is to select high dose
of new drug and compare to marginal dose
of standard drug
21
Comparative Effectiveness
• Must compare appropriate doses
– Overencapsulation may be necessary to
protect blind
– This may result in decreased bioavailability of
the comparator, resulting in decreased
effectiveness
22
Most Drug Development Programs Have Been
Judged Not to Meet DNP’s
Standard for Comparative Claims
• Only rarely has comparative information been
included in labeling
• No mention of active drug in Clinical Trials
summary
• No safety data from active comparators in
Adverse Events table
• Companies have generally accepted this
approach, since uniformly applied
• Nevertheless, comparative claims are a
possibility, if the methods are sound
23
Prominent Placement of Serious Adverse
Event Information in Labeling Provides an
Indirect Basis for Comparative Claims for
Other Drugs in the Class
• Examples:
– Rash/Lamictal
– Liver Failure/Tolcapone
– PML/Tysabri
24
Summary Comments
• Comparative efficacy and safety
information is potentially useful
• Need careful attention to design of trials,
particularly regarding “fairness” of
comparison
• Higher standard for comparative claims
25
Other comparative situations
• Proposed statements in labeling that imply a
particular adverse event does not exist
– One of many scales assessed (e.g., of
cognition) that are not statistically significantly
different from placebo
– Problems of multiplicity as well as of allowing
an explicit “not different from placebo”
statement
– This is really a non-inferiority question
26
Other comparative situations
• Immediate Release vs extended release
• There is an implicit “understanding” that
equal daily doses result in equieffectiveness (though perhaps fewer AEs)
• We have no evidence that any of this is
true
• Belief further enhanced by formal
switching studies
27
Other comparative situations
• Regarding a fair comparison to dose
– Is it appropriate to compare new drug only to
the approved dose (range) of the comparator,
or the “ideal” dose range?
28
Other comparative situations
• Implicit, cross-study comparisons
Treacherous, true, but…
– How to handle a drug that “wins” on a primary
outcome but is negative on secondary
outcomes on which all other approved
treatments are positive?
– Do we require that secondary outcomes be
assessed?
29
Other comparative situations
• For serious diseases, should we require
an active comparator?
– Is it still acceptable to “tolerate” a less
effective drug in some settings?
30
Other comparative situations
• If superiority is demonstrated in a welldesigned study, is that enough?
• Superiority implies, all other things being
equal, that this drug should be chosen first
• But a trial is just proof of principle, and
doesn’t really generalize to all patients
• Does this imply that, in this case, we
should study a random sample?
31