Concluding Remarks and Recommendations

1
Concluding Remarks and Recommendations

• General recommendations
• Consider adaptive dose response designs in
  exploratory development more often
• Whenever possible use an approach that
  incorporates a model for the dose response.
• Model assumptions can be either monotonic or
  umbrella shaped
• That trial specific objectives would determine
  the choice of particular methodology
• Consider several methodologies as AD candidates
  and pick the best-performing one
• Define the dose assignment mechanism
  prospectively and fully evaluate its operational
  characteristics through simulation prior to
  initiating the study
• Relative performance of various adaptive design
  methods is an area of ongoing research (PhRMA
  ADRS WG etc.)

2
Concluding Remarks and Recommendations (cont.)

• Benefits of adaptive designs in exploratory
  development
• Establishing POC exploring D-R can be
  accomplished in one trial
• Often with less time/resources than 2 separate
  trials
• Even if resources are the same, quality of
  information extracted about D-R may be better
  leading to increased probability of success
  (PoS) in subsequent trials
• Benefits of adaptive designs in (Phase I)
  oncology
• Balance between individual and collective ethics
  
• maximum information from the minimal number of
  patients
• Identify MTD more precisely
• limit allocation to extreme doses (above MTD)

Improve chances of success of Phase II-III trials
3
Concluding Remarks and Recommendations (cont.)

• Adaptive trial logistics
• Needs to be workable
• Response observable reasonably quickly relative
  to patient entry
• Allow ample time for planning !!!
• Simulations require substantial time commitment
  from statistician
• Extensive discussion with clinical needed to
  frame the problem
• Simulations often require custom programming
• Limited readyto-use software options exist (none
  of them is perfect!)
• Be aware of dynamic allocation issues
• Drug Supply Labeling more complicated
• Regulatory issues less important in early
  development, however should not be completely
  ignored

4
Adaptive Design Software Options

• CytelSim (in development)
• NOW available only as a Merck in-house tool
• FUTURE (TBD) may become commercially available
• Decimaker (fully supported product)
• developed by ClinBay as R-based product
• http//www.decimaker.com
• D-optimal design software (free)
• http//haggis.umbc.edu/cgi-bin/dinteractive/inna1
  .html
• EWOC software (free)
• http//sisyphus.emory.edu/software_ewoc.php
• MD-Anderson Cancer Center software (free)
• http//biostatistics.mdanderson.org/SoftwareDownlo
  ad
• Variety of methods available, including
• Phase I/II dose-finding based on efficacy and
  toxicity
• CRM

5
The End!

• Comments/Questions/Discussion?