Pär-Ola Bendahl, avd. för Onkologi

Selected statistical problems in cancer research

Abstract:

I will in this talk give examples of statistical problems that I am
facing as a biostatistician at the Department of Oncology here in
Lund. The examples I have chosen come from a wide spectrum of cancer
related projects: clinical trials, epidemiological studies, animal
experiments, linkage studies, genetic councelling and gene expression
studies.

All these seemingly different projects have one important aspect in
common, namely the ultimate goal of learning more about disease
mechanisms and to use this new knowledge in the clinical management of
patients.

Numerous prognostic factors for various cancer types have been
identified during the past decades and many more will me identified in
the future. These factors are useful for allocation of patients to
different risk groups, but it is much more important to find factors
that predict treatment response. Only a handful of such predictive
factors have been identified so far, but recent studies indicate that
SNP-analyses, gene expression profiling, and protein profiling will be
useful tools for optimizing the treatment for each individual cancer
patient. This area of research, known as pharmacogentics, is very
exciting and the kind of data produced is indeed a challenge for us
statisticians.