Presented by Eugene Laska, Ph.D. at the Arthritis Advisory Committee meeting 07/30/02

1. Presented by Eugene Laska, Ph.D. at the Arthritis Advisory Committee meeting 07/30/02

2. The goal of a RCT is to characterize the properties of a treatment • Onset of pain relief • Peak effects • Duration of pain relief • Dose response relationship • Dosing intervals • Adverse effects NB: These are all dependent on the severity of pain NB:Two measures of efficacy, pain intensity and pain relief are not necessary. They are highly correlated and add no new information

3. Characterizing ONSET

4. How to measure onset of analgesia • One stopwatch method: meaningful captures fact that patient obtained meaningful pain relief (responder) and the time it occurred (onset) • Two stopwatches method: perceptible,meaningful • Time to perceptible pain relief less clinically relevant and susceptible to noise • Complicates the study for no gain • Artificial data changes (adjusting the time of the first click if there is no second click) are of questionable statistically validity

5. How to quantify onset of analgesia • Cure model: H(t) = (1-p) + pS(t) • Population comprised of two groups- responders and non responders • Estimate the proportion p of patients who respond (i.e., obtain onset of meaningful pain relief) • Estimate the survival distribution and median time to meaningful pain relief among the responders LASKA EM, SIEGEL C, and SUNSHINE A. Commentary - onset and duration: Measurement and analysis. Clin Pharm & Ther, 49(1):1-5, January 1991.

6. Regulatory requirement for registration of drug D - onset • The proportion of responders on drug D must be significantly larger than the proportion of responders on placebo: pD> pP • Median onset among responders on drug D < t min NB: Choice of t may depend on pain intensity, model, and setting NB: Drug D may not have “faster” onset than placebo NB: The criterion drug D sig > placebo on PIDt is too dependent on sample size and variability

7. Characterizing DURATION to establish the DOSING INTERVAL

8. How to measure and quantify duration of analgesia • Time to offset (end of meaningful pain relief) using a stopwatch or Time to rescue medication • Estimate the survival distribution and median time to rescue (tR) among responders. Do not impute a value for those without onset • Report the proportion of responders that rescue at e.g. 6 hr, 12 hr, and 24 hr • Median time to rescue, among responders who rescue • Use standard survival methods and cure models

9. Regulatory requirement for claiming a dosing interval of x hr • The proportion of responders on drug D who rescue within x hr  0.5; • thus x  median time to rescue • At x hr, the proportion of responders on drug D who rescue is significantly < the proportion of responders on placebo who rescue NB: Can be evaluated only in models in which the pain can be controlled with a single agent

10. Evaluating the DOSING REGIMENin MULTI-DAY models Old bioassay rule Hold doses fixed - study outcome Hold outcome fixed - study doses

11. How to demonstrate efficacy of dosing regimen in acute pain after day 1 • Mild to moderate pain models- no prn narcotic allowed • The proportion of patients who require rescue on Days 2, 3 etc. is < x% • Demonstrating superior pain relief vs. placebo should not be required because patients who remain are likely to be “placebo responders.” Imputing values for patients who dropped out on Day 1 statistically questionable

12. How to demonstrate efficacy of dosing regimen in acute pain after day 1 • Moderate to severe pain models - prn narcotic allowed • Contrast the amount of prn narcotic used by patients on drug D vs. placebo • The proportion of patients on drug D + narcotics that dropout for lack of pain relief after Day 1 is < x% NB: amount of narcotic use for dropouts must be handled statistically

13. How to demonstrate efficacy of regimen in chronic pain at week W • Patients with severe chronic pain on placebo often drop out so comparisons to placebo are not likely to be valid • Imputation methods are crude at best and do not solve problem • At least x% of the patients on drug D (possibly plus prn narcotic) must still be in the trial by week W • Conduct a withdrawal trial i.e., at week W, half of the patients on Drug D are randomly reassigned to placebo -demonstrate superiority of drug D to placebo (in efficacy or narcotic sparing) in week W+1

14. Key points – onset and duration • Onset • Prob (responder|D) > Prob (responder|placebo) • Median onset among responder on D < t • Duration: Dosing Interval of x hr • Prob (responders who rescue by x hr|D) < Prob (responders who rescue by x hr|placebo) • Prob (responders who rescue by x hr|D)  0.5

15. Key points – demonstrating efficacy • Efficacy after day 1 – acute pain - no prn narcotic • Prob (rescue|D) < y (e.g., 0.15) • Efficacy after day 1 – acute pain - with prn narcotic • Prob (dropout|D+narcotic) < y • Mean narcotic on D < mean narcotic on placebo • Efficacy at week W – chronic pain - with prn narcotic • Prob (cumulative dropout by W|D+narcotic) < y • Conduct a withdrawal trial and show mean narcotic on D < mean narcotic on placebo