1. 1 Informative Studies of New Therapeutic Agents in Major Depression, GAD & Panic W Z Potter, M.D., PhD. Merck Research Laboratories

2. 2 Overview For depression, current approaches deliver data necessary to drug use in broad population Greatest need is to find novel antidepressants that may provide broader efficacy, faster action and/or better risk benefit ratio for acute active phase of illness Recent research into achieving optimal yields with current designs should be applied to the above need Whether alternate studies supporting registration would yield benefit is a matter for research GAD and Panic might benefit from formal consensus discussion among experts as to best studies to support registration

3. 3 Goals of Treatment Studies Safety over period of likely treatment Establish meaningful drug efficacy –Clinically significant relief –Cure –Maintenance of Response –Prevention of Recurrence if episodic –Cost effectiveness from societal point of view

4. 4 How well do the Current Requirements for Submission and Approval Meet these Goals? Major Depression: Remarkably well despite many failed studies and reasonable standard for novel agents given that we are beginning to understand sources of variability GAD: Not well studied as primary indication but no published evidence on clinically detrimental gap in knowledge PANIC: Even less studied as primary indication – open question as to best study designs for providing most clinically relevant information

5. 5 Selected Potential Antidepressants in 1997: Wrong Targets for Better Responses or Poor Signal Detection? CompoundMechanismCompany R-FluoxetineSSRISepracor/Lilly MK-742694NK-1/Sub PMerck EMD 688435HT1a agonMerckKGaA /SSRI Flibanserin5HT1a agonBoehringer /2a antag Ingleheim YKP-10ANMDA antag?SK Corp NS 2389SNDRINeurosearch Sunepitron5HT1a agonPfizer /alpha 2 antag Robalzotan5HT1a antagZeneca CGP 49823NK-1/ sub PNovartis antag ORG 129625HT2c agonistOrganon BefloxatoneMAO- A inhSynthelabo

6. 6 Signal Detection: A Major Problem Using Standard Designs with Proven ADs Proportion of Failed Antidepressant Trials in FDA SBA Data Sets Khan A, et al. J Clin Psychopharmacology 2002;22(1):40-45. RCTs (#)Drug = PBO (%) New AD 44 23 (52%) Standard AD 22 8 (36%)

7. 7 Drug CNS entry? Pharmacologic effect? Dosing? Design Complexity Instrumentation Lead-in Duration Blinding Flexible/fixed Data analysis Patient Source Severity Diagnosis Comorbidity Expectations Behaviour Investigator Affiliation Training Specialty Interaction Studying Signal Detection

8. 8 Similar Problems in GAD starting with Drug Placebo Differences * * * * * * * * * * * * * *

9. 9 Venlafaxine –FDA SBA: Variation Across Doses & Studies in Effect Size

10. 10 Researched Factors Contributing to Variance in Signal Detection with Current Designs Marked US vs OUS differences in terms of efficacy and safety – Different Patients & Practices “Ski Slope” phenomenon from Single Blind Lead-In – Tests of Double-Blind Alternative Evidence of systematic bias -- Constriction of severity assessment around entry criteria Rating Scales – Relevant items and mode of administration evolving

11. 11 Enhanced Signal Detection with Current Designs vs Other Issues for AD Development Impact of wide availability of SSRIs – Why enter a study? –Partial or non-responder to SSRI? Lack of novel agents -- Can drive over investment in single mechanism vs assessing many –Expectation around and wish for novel treatments can drive premature studies in large numbers of patients prior to acute efficacy being established (Substance P Antagonist Experience)

12. 12 Depression: Value of Current Study Design: Acute & Maintenance Heterogenous disease and population of patients available for studies affect trial designs supporting registration: –Establishing subacute efficacy for a novel mechanism would itself constitute a breakthrough –Large drug effect size (twice that in acute designs) in discontinuation studies after 2-3 months on drug (only 50-60% of treated reach this point) has consistently predicted long-term efficacy -- 6 months on drug would mean more drop outs, have restrictive effects on patient selection and be less informative at this stage of development –2-3 month design captures both those who truly require drug and those who are out of episode by 6 months

13. 13 NIMH Funded Classic Study of Relevant Drug-Placebo Difference within 12 weeks of Discontinuation after 12 weeks on Drug

14. 14 GAD & Panic: Efficacy Chronic conditions which when treated with benzodiazepines risk attendant dependence and “rebound” after acute discontinuation Recent data in GAD with SSRIs conform remarkably to pattern seen in depression Little informative data in panic -- last primary indication studies with alprazolam

15. 15 Double Blind Discontinuation Results for an SSRI in GAD* *Stocchi, F et al, J Clin Psych, 2003

16. 16 Conclusions For depression, current approaches deliver data necessary to drug use in broad population Greatest need is to find novel antidepressants that may provide broader efficacy, faster action and/or better risk benefit ratio for acute active phase of illness Recent research into achieving optimal yields with current designs should be applied to the above need Whether alternate studies supporting registration would yield benefit is a matter for research GAD and Panic might benefit from formal consensus discussion among experts as to best studies to support registration