Researchers at University College Dublin have merged biological information and mathematical models to deliver the ultimate in personalised medicine, a computer programme that can predict how an individual will respond to medical treatments.
** See www.ucd.ie/thinkbigger/research **
[Producers: Dominic Martella & Lisa Flannery (University College Dublin). Filmed by Tiny Ark on location at University College Dublin, Ireland.]
The Systems Biology Ireland lab at UCD developed models that are based on a person’s own biochemistry. Once integrated with a computer algorithm the team led by Professor Walter Kolch can predict with a high degree of accuracy how a person will respond to various treatments.
The work has been underway at UCD for five or six years and has focused on picking apart the complex interactions that take place between our cells. “We are trying to understand the communications network in the human body,” he says.
Many diseases occur when a fault occurs in the complex inter-cellular communications system. “We use computational methods to investigate this network and when used with biological information we can simulate a lot of this on the computer and simulate what is going wrong,” Kolch explains. “People are really excited because you can feel that motivation of doing work which nobody has done before.”
“It will revolutionise the way we go about providing medical treatments,” he says.
In effect it creates a computer model that mimics the way an individual’s cells would work. With the bio-silico model it becomes possible to test treatment options, decide on what drugs to use and know when a commonly used treatment might not work.
Studies have shown that the patient response rate for the world’s top 10 drugs only runs at between 10 and 30 per cent, he says. “That means 70 to 90 per cent of those who take those drugs don’t do any better, they just get the side effects,” Kolch says. “We can use computations to make a personal patient model and then make the right drug choices for them.”
The important thing is Kolch and his team already have good data showing that the approach works. They produced a model for a childhood brain tumour called neuroblastoma and then used biological data from 700 patients. “We used this model to find high risk patients who could not previously be identified,” he says. These are the patients who require aggressive high dose chemo, an approach that would have caused serious but unnecessary side effects in those outside this high risk group. His models reached accuracy rates in the 90 per cent range, he says.
The team started working with a clinic in Mainz, Germany, looking at children who had resisted all treatments and had moved into hospice care. But Kolch’s models were able to identify different chemo options that would not have been used in the past in these children. Kolch is now looking to set up a similar collaboration with Our Lady’s Hospital for Sick Children, Crumlin, working with Prof. Owen Smith CBE, Professor of Paediatrics.
A more accurate diagnosis is a key reason for developing this bio-silico approach but it is not the only one. The models can identify the best treatment options for an individual before any drugs are introduced and can help predict what the most likely patient outcome will be. It will also assist in choosing a treatment option with the lowest possible side effects.
It also provides a way to cut the cost of health care, Kolch argues. Many of the compounds used to treat cancer are extremely expensive and are applied generally, even though a person may not gain sufficient benefit from their use.