people

• Overview
• Grant Central
• Medvane
• Profiles

learning

• Overview
• Colloquium Series
• Conferences and Seminars
• Dean's Symposium
• Leder Human Biology
• Biomedical Informatics Training
• Biostatistics Training
• Clinical Investigator Training Program
• Scholars in Clinical Science
• KL2 Medical Research Investigator Training
• Introduction to Clinical Investigation
• Biostatistics Seminar Series
• Genetics Courses
• Imaging Courses

resources

• Overview
• Pilot Funding
• AEGIS Influenza
• Atlas
• Community-Based Participatory Research
• Community Connect to Research
• Consulting Programs
• Core Facilities
• HarvardTrials
• Health Disparities Research Program
• In/Outpatient Resources
• Laboratory for Innovative Translational Technologies
• Law and Health Policy
• Medications and Society
• Pathology Specimen Locator
• SHRINE
• Webdash

Biostatistical Science Program Colloquium Series: Targeted Clinical Trials

Posted October 15, 2009

“Targeted Clinical Trials”

Stephen L. George, PhD
Professor, Department of Biostatistics and Bioinformatics
Director, Department of Cancer Center Biostatistics
Director, Department of CALGB Statistical Center
Duke University School of Medicine

October 27, 2009, 3:30pm - 5:00pm
Pavilion Dining Room
Children’s Hospital Boston
300 Longwood Avenue, Boston

A light reception will follow.

Abstract:

The proliferation of predictive and prognostic biomarkers in oncology and the development of agents targeting specific biologic pathways require careful consideration of the types of clinical trials appropriate in this setting. A traditional randomized clinical trial (RCT) is designed to compare the effects of competing therapies “on average” in a population of patients with a given disease and with specified eligibility requirements. There is no assumption that the population is homogeneous; the intent is to reach conclusions that are appropriate for the population of patients that might receive such therapies. But if the population of patients under study is very heterogeneous, in particular if there is only a small percentage of patients who might potentially benefit from the new therapy, such clinical trials are not likely to reach the correct conclusion and are certain to treat a large percentage patients who cannot benefit from treatment.

In a “targeted” RCT, interest centers on a select group of patients, defined by biomarkers and other patient characteristics, considered most likely to benefit from the new therapy under study. The therapy itself may or may not be a targeted therapy. In this talk, we will compare and contrast the statistical characteristics of various potential designs in this setting, including a traditional design, a targeted design, a biomarker-stratified design, an “enrichment” design, and a strategy design. These designs will be compared with respect to the types of inferences and with respect to the total number of patients required, the number of events required, the time to completion of the trial, and the cost of the trial. Examples of trials of each type will be used to highlight the issues.

For more information about this talk, contact Shaina Andelman (E-mail).