News Release

Smart statistics for cancer diagnosis

Peer-Reviewed Publication

CSIRO Australia



Dr David Mitchell, CSIRO Mathematical and Information Sciences

Full size image available through contact

CSIRO statisticians have developed a new way to subtype paediatric Acute Lymphoblastic Leukaemia (ALL), the most common form of cancer in children under 12.

CSIRO's new molecular classifier is being validated on clinical samples by the Children's Cancer Institute Australia for Medical Research (CCIA).

"Accurate classification of ALL into its different subtypes is vital for effective treatment of this disease," says Associate Professor Murray Norris, Deputy Director of CCIA. "We are very excited to be working with CSIRO to evaluate their classifier."

"We used gene expression microarray data from ALL patients to develop our classifier," says Dr David Mitchell of CSIRO Mathematical and Information Sciences. "We have developed several proprietary statistical techniques that enable us to use this type of data to develop simpler and more cost effective diagnostic tests."

"These powerful new statistical techniques have enabled us to develop a classifier for subtyping ALL that uses only nine genes, whereas less statistical techniques result in classifiers that use tens or hundreds of genes," says Dr Mitchell. "The fewer genes that are required to diagnose a disease, the simpler the diagnostic tests can be."

Access to simple technology for accurate subtyping of ALL could raise the cure rate in developing countries from the present level of around 30% to that of developed countries, around 80%.

Additionally, as current typing methods involve direct examination of bone marrow biopsies, CSIRO researchers plan to develop a classifier that only requires a blood sample, not a bone marrow biopsy.

"Our new statistical techniques, known as GeneRave, have applications in diagnostics, toxicogenomics and pharmacogenomics," says Dr Mitchell. "GeneRave enables us to rapidly sift through the very large numbers of gene expression measurements that are generated by microarray experiments in order to identify the smallest set of genes that form the best predictive set."

"We are looking at better ways of analysing microarray data to enable faster drug discovery, personalised medicine and development of simpler clinical diagnostic tests," he says.

###

CSIRO is Australia's leading national science agency and one of the world's largest and most diverse scientific research organisations. CSIRO's capabilities in life sciences and biotechnology span human health, agricultural, environmental and industrial applications.

More information:
Dr David Mitchell
CSIRO Mathematical and Information Sciences
61-402-304-801
David.Mitchell@csiro.au

Dr Glenn Stone
CSIRO Mathematical and Information Sciences
61-2-9325-3259
Glenn.Stone@csiro.au

Media assistance:
Andrea Mettenmeyer
CSIRO Mathematical and Information Sciences
61-415-199-434
61-2-6216-7157
Andrea.Mettenmeyer@csiro.au

Images available from: http://www.cmis.csiro.au/mediapics.htm

If you would like to be removed from this mailing list, please contact CSIROMedia@csiro.au, or telephone us on 2-6276-6479.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.