Vortrag

am Donnerstag, 19. Juni 2008
16:00 Uhr
Institut für Medizinische Informatik, Statistik und Dokumentation / Medizinische Universität Graz
Auenbruggerplatz 2/II, 8036 Graz
SZ KW.21




Phase II designs in oncology

von Hans Ulrich Burger
F. Hoffmann-La Roche AG, 4070 Basel, CH


Abstract

The classical development of new treatments in oncology followed in the past a very specific path: In phase I, patients were treated with a new molecular entity in small cohorts of patients to increasing doses, until side effects started to become intolerable. In the next phase, patients out of a more homogenous population then received the new treatment with the selected dose from phase 1. As soon as one or more tumor responders were observed, the new treatment was regarded as promising enough to enter phase III to explore the new treatment against the current standard.
In such a development program phase II was limited in quality – usually only non-randomized studies – as well as in size. With the increasing number of biologic entities in oncology today this development paradigm starts now to change. Defining an adequate dose just by early safety readings is no longer adequate and proof of concept designs today seem to be complicated with the increasing use of biologics or combination therapies. In consequence phase II designs in oncology start to change as well. This talk will present some alternative designs for randomized phase II studies and stress especially the necessity to look into the operating characteristics for each design. Only when knowing the type I and II errors for each design, a conscious decision about the use of a specific design and its adequate sample size can be taken. This is important for proof of concept studies as well as for any dose evaluation study and it is important when limiting the subsequent risks in phase III.

Zurück zur Seite der biometrischen Sektion Steiermark-Kärnten