A team of researchers are analysing past data in
order to predict how children will cope with cancer.

It’s the worst nightmare of most parents: having a child diagnosed with a potentially terminal disease. Just 40 years ago, fewer than three in ten children survived cancer. Today, with advances in treatment and management, that figure has reversed, with seven out of ten now surviving the disease.¹

That figure is small comfort to the families of the three out of ten who don’t make it, and a group of researchers led by Dr Paul Kennedy (UTS:IT) is using some innovative cross-disciplinary techniques to help reduce childhood cancer deaths still further. The most common form of childhood cancer is acute lymphoblastic (or lymphocytic) leukaemia (ALL). Between 1995 and 1999 it accounted for 24 per cent of all new cancer diagnoses in children.

Nearly all of these children are treated successfully and go into remission, but if the child relapses and the disease returns, survival is dramatically reduced. Current research seeks to find genes to explain tumour growth, characteristics and response to therapy. This knowledge is useful in understanding ALL, but does not always help front-line clinicians to choose the right therapy for each patient. The vast quantity and extreme complexity of the clinical data collected makes it difficult to translate the data into practical knowledge that can be applied to the treatment of new patients. Dr Kennedy’s project aims to rectify that, and it’s a task that requires IT and medical science to put their heads together.

Dr Kennedy’s team includes biologist Dr Daniel Catchpoole and (from The Children’s Hospital Westmead), and UTS:IT data mining expert Professor Simeon Simoff (University of Western Sydney and UTS:IT Adjunct Professor), and computer scientist Professor David Skillicorn (Queen’s University, Canada).

They believe they can discover patterns in the mountain of existing data on ALL patients that will help identify the children most likely to relapse. Different aggressive treatment can then be given to those children to increase their chance of survival and, equally, children who are unlikely to relapse can be given less aggressive and therefore less toxic treatment. “The quest is to develop personalised medicine — tailor-made therapies,” says Dr Catchpoole. “What makes each patient an individual? What drives their disease? We would like this information to be able to be used by clinicians to make decisions about how each patient should be treated.”

“Cancer is caused by lots of little changes,” says Dr Kennedy, “so we need to try to look at all the information rather than just a few pieces. There are a million pieces of DNA per person, 20 000 bits of information for the gene expression, so we need to use new data-mining techniques to find patterns in the data and to be able to compare patients.”

Dr Catchpoole agrees. “Looking at the genetic activity of the tumour and also developing techniques so that we can capture all the information about the genetic background of the patient allows us to form a much more complete picture,” he says. “We can look at all this information to get a complete biological picture of each individual patient and compare them based on their underlying biology. We can then use this information as a computational tool to data mine existing clinical data to learn the best ways to treat future patients.”

Data-mining expert Professor Simoff says that from an IT point of view, it’s a big ask — and not just because of the vast quantity and complexity of the data. “The technology used to collect the data has changed, and to compare data from different machines is a challenge!” He says.

Over the last four to five years the team has generated the data and have published some early findings. They have also engaged two PhD students, a Masters student and two honours students. UTS PhD student Ahmad Al-Oqaily has an Australian Rotary Health Research Fund scholarship for the project, and the team also won an internal UTS grant. The team continues to seek funding, so that the research can be brought to the front-line.

The next challenge is to test the data and confirm that it will enable clinicians to tailor cancer treatments based on how patients with similar underlying biological systems responded to treatment, and so bring hope to thousands of young cancer patients and their families.

¹ Australian Cancer Research Foundation