[Cowystats] CO/WY ASA Seminar - Wednesday November 9th

Matt Pocernich pocernic at rap.ucar.edu
Fri Oct 7 16:09:20 MDT 2005


As you may recall, when we recently surveyed chapter members 
concerning presentation topics of interest for the upcoming year, 
there were numerous requests for talks that would better familiarize 
members with some of the general areas within biostatistical 
research.  In response to these requests, we have invited Dr. Gary 
Clark of OSI Pharmaceuticals, to speak abut his recent experiences 
going through the clinical trial process required by the FDA for 
approval of a new drug.   While this talk focuses on the topic of 
biostatistics, we  also hope that it will have a broader appeal to members 
in other areas of statistics.  An introduction will be included so that 
the context of the medical problem and the statistical problem will be
understandable to those who are not bio-statisticians, just
statisticians.  Also, this talk is open to everyone, not just members
of the local ASA chapter.  Details about parking and the room location
will be provided later in the month.

Title: Pearls of Wisdom Learned from Clinical Trials of a New
Anticancer Therapy

by Gary Clark, Ph.D., Vice President, Biostatistics & Data
Management, OSI Pharmaceuticals Inc., Boulder, CO

When: Wednesday, November 9, 2005 7:30 p.m. *(coffee, tea and
refreshments will be served at 7 p.m.)

Where: CU Denver Math Department - 14th and Larimer, Denver.

Talk Outline:

Tarceva® is a new, oral, anticancer agent. It is approved for the
treatment of patients with non-small cell lung cancer after failure of
at least 1 prior chemotherapy. Recently, the FDA asked its Oncologic
Drugs Advisory Committee (ODAC) to review results of a randomized,
double-blind, placebo-controlled clinical trial of Tarceva® in
combination with standard chemotherapy (gemcitabine) for patients with
untreated, advanced pancreatic cancer. During the preparation for the
presentation to ODAC, several unusual statistical issues were
encountered, including: 
-	The optimal dose was not known at the beginning of the study,
         so two doses were evaluated in subsets of patients 
-	The sample size was changed twice during the study
 	Prior to unblinding, the FDA requested a change in the
         Statistical Analysis Plan and offered two options:  either stratified
         log-rank test for survival or a univariate log-rank test 
-	After unblinding, the FDA identified a subset of patients who
         were not technically eligible for the study 
-	After unblinding, the FDA requested that follow-up information
         be updated

The impact of these issues on the interpretation of the study results
and the trials and tribulations of estimating the treatment benefit
will be discussed.

If you have any other question or suggestions, contact me at 
pocernic at ucar.edu.

Matt Pocernich
NCAR - Research Applications Program

More information about the Cowystats mailing list