&lt;number&gt; The requirement to demonstrate safety and efficacy before drugs are marketed stems from two Congressional actions. The safety requirement comes from the Federal Food and Drug and Cosmetic Act of 1938. The efficacy requirement stems from a 1962 amendment. The FDA has interpreted the 1962 amendment to require in most cases at least two trials for drug approval and to require claims to represent clinical benefit. For drug and indication to receive approval sufficient information must be available to define an appropriate patient population for drug use and enable safe and effective use.The product label represents a licensing agreement between the sponsor and the Federal government. Marketing and advertising claims are derived from the product label.
Survival is the gold standard in demonstrating clinical benefit. Defined as the time from randomization to death, this endpoint is unambiguous and is not subject to investigator interpretation making it an attractive endpoint especially in unblinded, randomized trials.
Although a preferred endpoint, survival has drawbacks. Since most active agents in oncology are associated with small incremental survival benefits, large sample sizes may be required to demonstrate comparatively small differences. An increasing concern is the effect of cross-over therapy in randomized trials which may obscure a survival effect. This issue was discussed in a June, 1999 ODAC meeting with a consensus that survival should remain the primary endpoint for first-line breast cancer trials. Despite significant cross-over, trials have demonstrated survival advantages. An example of this is use CPT-11 in 5-FU refractory patients. Nevertheless, the issue of cross-over and its effect on survival remains an issue of concern and will be reviewed in subsequent advisory committees.
Patient benefit can be described as improved survival or a non-inferior survival after consideration of toxicity and the magnitude of benefit.A survival duration is comprised of a component of a natural history effect plus a treatment effect. A non-inferiority analysis ensures that a survival advantage, “the control effect” associated with an approved drug will not be lost with a new agent.
Time to progression is defined as the time from randomization to time of progressive disease or death. This endpoint has been used for the approval of hormonal therapies and certain biologics because of the more favorable toxicity profile. Differences in TTP will not be obscured by secondary therapies if a cross-over effect exists. We have been interested in improving this endpoint by potentially correlating radiographic changes with a delay of new symptoms or of worsening of symptoms.
Most oncology trials are unblinded potentially introducing bias into decision-making regarding TTP. Patients must be evaluated on a regular basis on all treatment arms for this endpoint to be meaningful and must evaluate all sites of possible disease and ideally measure all lesions. There must be complete ascertainment of all sites of disease at baseline and follow-up. The same assessment technique must be used at each follow-up and the same evaluation schedule should be used.
Other problems with the TTP endpoint is the consideration of the magnitude of difference which would constitute clinical benefit. Since most measurement are performed every 3 months and the differences in TTP may be of similar benefit, the clinical relevance of relatively small differences, especially in unblinded trials, may be questioned. In addition, prolongation of TTP may not impact survival. Other problems with this endpoint include the handling of missing data and censoring issues, especially for long periods between loss to follow-up and death or the introduction of new therapies in the absence of documented progression.
Whereas, the endpoints of survival and TTP are comprised of the effect of the natural history plus an increment produced by treatment, tumor size reduction can usually be attributed entirely to treatment. Duration of the response must be included. Response rates are comprised of partial and complete responses. Stable disease and minor response patients are excluded. Although several criteria exist for response rate measurement, a single criteria should be adopted and uniformly applied. We have accepted the RECIST criteria for registration trials.
The evaluation and comparison of response rates may be difficult. Methodological questions include --Were lesions to be measured prospectively identified? -- Was the same imaging techniques including scan and types of cuts used? In comparing RR, one must analyze the relative number of CRs vs PRs, duration of responses, location of responses, and correlation with symptom imporvment, and extent or bulk of metastatic disease.
Palliation or relief of symptoms are traditional endpoints corroborating clinical benefit. These endpoints are usually reported as patient reported outcomes. Credibility of these endpoints can be enhanced by blinding and an association with a biologic effect of the drug--e.g. response rate. Instruments used to measure these endpoints should be relatively simple, prospectively designed and tailored around an expected outcome. Examples include relief of dysphagia, and pain
Health-related quality of life are included in many registration trials and may provide the patient’s perspective on the treatment. However, their interpretation is complicated by the lack of blinding, missing data leading, multiple endpoints and multiple comparisons to baseline which must be adjusted in the statistical analysis plan. In addition, clinical significance of score changes may be unclear.
This above are recent examples of products approved under the accelerated approval program. The most recent example of an accelerated approval is our approval this week of Gleevec or STI 1571 in patients with CML who are either in chronic phase failing alfa interferon, in accelerated phase or in blast crisis. The review of this NDA application which was approved on Thursday, May 10 was done in a record 2 1/2 months.
Novel agents pose new challenges. Rather than conventional histological criteria for diagnosis, we may need to re-define how we classify cancers based on the mechanism of pathogenesis and therapy. This may allow greater efficiency in the clinical trial process by selecting populations with a greater liklihood of responding to treatments. Novel surrogates will need validation and acceptance by the scientific community. Starting doses of agents may need to be aimed at inhibiting or interacting with a target rather than the MTD. Dose ranging studies, accepted as a norm in most other therapeutic areas, may be accepted by oncologists to avoid toxicities and optimize the therapy’s interaction with targets.
With a greater number of candidate drugs being developed, selection of agents to demonstrate clinical benefit may need to be examined on the basis of surrogate endpoints, such as target inhibition. Nevertheless, we as a community must address patient accrual issues since only a limited number of eligible patients are entered on trials. These patients should reflect the population who will eventually use the drug rather than an selected population with normal laboratory findings and excellent performance status. Since most drug will have international use, regulatory agents must accept international data and communicate more effectively on accepted dosing and approval criteria. At this year’s ASCO we have organized a meeting of physicians involved in oncology drug regulation from the United States, Japan, Australia, Canada, and Europe (EMEA) to initiate this dialog.
Defining Success in Oncology
Richard Pazdur, MD
The views expressed are the results of independent
work and do not necessarily represent the views or
findings of the United States Food and Drug
Administration or the United States
Basis for NDA Approval
• Demonstration of efficacy with acceptable
safety in adequate and well-controlled
• Ability to generate product labeling that
– Defines an appropriate patient population
– Provides adequate information to enable safe
and effective use
– Approval for an indication, not drug
• Accelerated Approval--serious or life-
threatening disease, benefit over available
therapy. Use of surrogate; mandated
phase IV trials
• Fast Track--life-threatening disease,
potential to address unmet medical need.
Rolling NDA, meetings
• Priority review--drug would be a significant
improvement compared to available drugs.
Review of NDA in 6 months
Activity vs. Benefit
• Biologic Activity--screening of a
compound, phase II trial endpoint, an
indication for further study
• Clinical benefit--what is meaningful to a
• The approval process is not a screening
process for drug activity
Should Oncology Drug
Regulation Be Different?
• Life-threatening nature of diseases--patient
access vs necessary data for approval
• Drugs multiple action modes; combinations
• Risk/benefit ratio--different perspective on
serious adverse events; highly trained
specialists using drugs rather than GP
• Product label and off-label uses
Should Oncology Drug
Regulation Be Different?
• Investigational nature of discipline--Cancer
Centers, Cooperative Groups, NCI
• Wide variety of products used by
devices, supportive care, diagnostics
• Multidisciplinary approaches
• Represents over 100 diseases/indications
Risks in Developing Oncology
• Indication--lack of predictive models
• “Creative Indications”--progressively more
refractory patient, market share
• Two trials versus one trial
• Dose ranging studies--moving away from
Oncology Trial Concerns
• Minimize bias
– Blinding trials (few)
– Endpoints that minimize bias
– Internal consistency of subgroups, endpoints
• Magnitude of change of endpoint
– Clinical significance
– Underpowered trials--guessing treatment effect
• Isolating effect of drug
Endpoints for traditional
• Defined as the time from randomization to
• Unambiguous endpoint that is not subject to
investigator interpretation or bias from
• Assessed daily
Traditional Endpoints: Survival
– Requires large sample size and long follow-up
– Confounder--Cross-over therapy may “wash
out” a survival effect
Traditional Endpoints: Survival
• Non-inferior or improved survival
constitutes “patient benefit” after
consideration of toxicity and the magnitude
of the benefit
• Non-inferior outcome ensures that a
survival advantage associated with an
approved drug will not be lost with a new
Time to Progression--Advantages
• Could use a smaller sample size and shorter
follow-up than trials that require a survival
• Differences will not be obscured by
secondary therapy if cross-over effect exists
• “Time to symptomatic progression”
• Unblinded trials introduce bias
• Must evaluate all patients on a regular basis
– Must evaluate all sites of possible disease
– Complete ascertainment of all sites at baseline
and follow-up (i.e., look for new sites)
– Same type of assessment tool at each follow-up
– Should use same evaluation schedule
• How much improvement constitutes
• Unique endpoint--treatment is “entirely”
responsible for tumor reduction
• In contrast, survival and TTP have an effect
of the natural history PLUS treatment effect
• Must consider duration of response
• Does not include stable disease
• Pick your criteria and stick with it
Complicated Picture of RR
• Number of CRs vs PRs?
• Duration of responses?
• Location of responses (e.g., liver vs skin)?
• Association with symptom improvement?
• Extent or bulk of metastatic disease?
Palliation and Patient Reported
• Blinding and associated antitumor effects
(response rates) lend credibility
– Use simple instruments
– Avoid multiple endpoints
– Example: Photofrin PDT and dysphagia
Potential palliative endpoint:
Health-related quality of life
• Pro: Patient’s perspective on treatment
– Blinding is essential, but difficult to do
– Careful serial assessments
• Missing data makes interpretation problematic
• Multiple endpoints and comparisons to baseline must be
adjusted for in the statistical analysis plan
– Clinical significance of score changes may be unclear
– Is additional information gained, compared to a careful
recording of toxicity/symptom data?
Accelerated Approval- Subpart H
• For serious or life-threatening diseases
• Where the drug appears to provide benefit
over available therapy
• Approval based on a surrogate that is
reasonably likely to predict clinical benefit
• Subject to the requirement that the applicant
verify and describe benefit
• Post-marketing studies would usually be
• The applicant shall carry out such studies
with due diligence
Issues Related to the AA
Program As a Whole
• The importance of confirmatory trials being
underway at the time of AA
• The approach of studying slightly different
populations in the confirmatory setting than
the AA population
• Relative merits of different trial designs
– single arm in refractory populations
– randomized trials in less refractory patients
Challenges for Oncology Drug
• New “targeted therapies”
– Re-define definitions of diseases
– Greater efficacy in selected population may
result in smaller patient populations
– Novel surrogates to be validated
– Dosing aimed at target rather than MTD
– Dose studies, chronic administration
• Greater number of candidate drugs
– Careful selection of agents to demonstrate
clinical benefit by oncology community
– Patient accrual to trials need to be increased
– Patients entering trials should reflect the patient
population which will eventually use the drug
– International studies, international agreement of
endpoints and study design and approval