Sponsoring Section/Society: ASA-Biopharm
Session Slot: 10:30-12:20 Tuesday
Estimated Audience Size: 200-250
AudioVisual Request: two overheads
Session Title: Formal Design Considerations for Phase I Clinical Trials
Theme Session: No
Applied Session: no
Session Organizer: Rosenberger, William F. University of Maryland
Address: Department of Mathematics, University of Maryland 1000 Hilltop Circle Baltimore, MD 21250
Phone: 410-455-2433
Fax: 410-455-1066
Email: Billr@math.umbc.edu
Session Timing: 110 minutes total (Sorry about format):
Opening Remarks by Chair - 5 minutes First Speaker - 25 minutes Second Speaker - 25 minutes Third Speaker - 25 minutes Discussant - 15 minutes Floor Discusion - 15 minutes
Session Chair: Rosenberger, William F. University of Maryland
Address: Department of Mathematics, University of Maryland 1000 Hilltop Circle Baltimore, MD 21250
Phone: 410-455-2433
Fax: 410-455-1066
Email: Billr@math.umbc.edu
1. Designing Studies for Dose-Response
Wong, Weng Ke, University of California, Los Angeles
Address: Department of Biostatistics UCLA School of Public Health Los Angeles, CA 90095-1772
Phone: 310-206-9622
Fax: 310-267-2113
Email: wkwong@sunlab.ph.ucla.edu
Abstract: 'Dose resonse' refers to the regression of a response on a stimulus. We review a number of options for dose-response designs, and compare various designs which may be used in practice. We start with two group designs. Next we introduce basic optimal approximate design theory for simple linear and quadratic regression illustrating different criteria of optimality and their effect on the allocation of the levels of the dose. Then we obtain the efficiencies of these optimal approximate designs and some simple designs which have intuitive appeal (symmetry, equal spacing of treatments, reduced numbers of observations at the highest and lowest doses).
2. Ehrenfest Urn Designs for Quantile Estimation
Flournoy, Nancy, American University
Address: Department of Biostatistics American University Washington D.C. 20016-8050
Phone: 202-885-3127
Fax: 202-885-3155
Email: flournoy@american.edu
Durham, Stephen, University of South Carolina
Abstract: Consider a binary random variable for which the probability of success is a monotone increasing function (called the response function) of some control variable, such as the dose of a drug. Except in the extreme tails, the asymptotic optimal design for estimating a particular quantile of the response function puts all observations at one point which is the unknown target quantile. We describe a randomized version the Ehrenfest urn model that provides a useful appproximation to the asymptotic optimal design of a target quantile. Subjects are sequentially treated according to the color of the ball drawn from the urn; after each treatment a ball is added to the urn whose color depends on the treatment outcome according to an algorithm that centers the treatment disribution around the unknown target quantile. Parametric assumptions regarding the response function are not required. Several estimators of the target quantile are presented. Exact distributional results are obtained. Whem blinding treatments is important, the Randomized Ehrenfest Urn Design may be preferred to the unrandomized procedures such as stochastic approximation, continual reassessment and up and down designs.
3. Dose Escalation Schemes for Cancer Phase I Trials
Zacks, Shelemyahu, Binghamton University
Address: Department of Mathematical Sciences
Phone: 607-777-6035
Fax: 607-777-2450
Email: shelly@math.binhamton.edu
Abstract: One of the objectives of Phase I clinical trials is to give patients the largest possible dose they can tolerate in order to test the reaction of patients to a new drug. Patients entering phase I trials have a certain unknown tolerance distribution for the new drug. The maximum tolerated dose, MTD, is a certain quantile of this tolerance distribution. The problem with clinical trials is that we have to design the doseages under the ethical constraint that the probability of giving a dose greater than the MTD is not greater than a small risk level, alpha. Most of the dose escalation schemes in the literature do not satisfy this constraint. Efficient Bayesian and non-Bayesian escalation schemes satisfying the above constraint are discussed and some results are presented of clinical trials. In particular we discuss modeling which can incorporate information given by covariates.
Discussant: Ting, Naitee Pfizer, Inc.
Address: Central Research Division Pfizer, Inc Groton, Ct 06340
Phone: 860-441-4871
Fax: unknown
Email: tingn@pfizer.com
List of speakers who are nonmembers: None