Project Leader: Dr. Wayne Phillips
Peter Mac Laboratory Head: Prof. R Thomas.
Project Status: Step 1 & 2: expanded sample collection and initial discovery modeling.
One of the most difficult and clinically important questions facing clinicians treating advanced cancer is deciding which patients will, and which will not, benefit from chemotherapy and/or radiotherapy. This is particularly true for clinicians treating locally advanced oesophageal cancer. In this project, the aim is to use microarray technology to profile the gene expression patterns of oesophageal tumors to determine if the expression patterns can be used to predict tumor response to chemoradiotherapy. This will open the path to the development of a clinically important test that will significantly improve the management of advanced cancer patients and provide a rational basis upon which to tailor individualized therapeutic strategies.
NICTA’s role will be to develop and test novel learning algorithms to identify patterns of gene expression that are predictive of tumour response.
Preliminary findings obtained for the squamous cell carcinoma subclass of oesophageal tumours are encouraging and clearly demonstrate proof-of-principle for the approach. Results for the other major subclass, adenocarcinoma of the oesophagus, have demonstrated unusual and counter-intuitive behaviors that we have called “anti-learning”. A partial theoretical explanation for this phenomenon has been developed, and new class of supervised learning algorithms capable of coping with such anomalies should be practical soon.
Current plans aim to substantially expand sample numbers, which will facilitate more reliable prediction and enable validation of findings.