by Byron Connolly

How hospitals are using big data analytics to research paediatric cancer

Feb 11, 20156 mins
Big Data

Researchers in the United States are turning to big data analytics to gain a much better understanding of paediatric brain cancer and other diseases with a similar genetic process.

US-based charity organisation, Dragon Master Foundation, has partnered with five specialist hospitals to build awareness around this disease to gather more biological specimens and other data necessary for ongoing research.

Genomic data associated with cancer is growing rapidly, but for rare childhood brain tumours, there has been a lack of advocacy around the need to use genetic and demographic information to develop new treatments and a potential cure.

Dragon Master founder, Amanda Haddock told CIO Australia last April that she wanted to change this. The organisation has been holding community events around the US to raise awareness and create a database containing at least 50,000 human genomes – our genetic material – where researchers can add and share molecular, genetic, clinical, and environmental data related to the disease. This information could be accessed in the cloud.

Haddock and her husband Richard Haddock – a former software developer – created the foundation in 2013 after their son lost his battle with glioblastoma multiforme,a highly aggressive brain cancer.

Since then, five hospitals specialising and conducting research in paediatric cancers have come on board with the organisation. They include The Children’s Hospital of Philadelphia (CHOP), Ann Robert H. Lurie Children’s Hospital of Chicago, the Children’s Hospital of Pittsburgh, the Seattle Children’s Hospital, and Meyer Children’s Hospital in Florence, Italy.

The hospitals and their researchers and clinicians have formed the Children’s Brain Tumor Tissue Consortium, a research program using open source data to study childhood brain tumours.

Speaking to CIO Australia, Adam Resnick, PhD and assistant professor of neurosurgery at The Children’s Hospital of Philadelphia, said the most widely used practice for sharing data is to literally ship a hard drive to another location for researchers to compare results.

“That ends up being a slow and non-efficient process because you need to identify who to send the data to and collaborate with. This created a need to build a space for collecting, interrogating and studying the data.”

Read: Monash University expands cloud infrastructure to support researchers

There are 1,100 paediatric brain tumour tissue specimens from four institutions available to the research community. Hospitals also collect normal blood-derived genomes, which are not affected by cancer, for comparison. This includes blood from a child’s parents to discover any genetic connections.

“This is important because each one of us is very different from the other. It’s very challenging to figure out what is different between people that causes the cancer, and what is different between people just because of normal variations.

“That’s unique to the repository infrastructure that we have created because for a long time, people were just saving tumours – not recognising how important it was to also collect blood specimens,” he said.

Resnick said databases need to be very scalable. Around 100 terabytes of data is stored for use in pilot projects and he expects that number to rise to petabytes or even larger as infrastructure is used to do studies across other diseases.

“For us that’s another level of partnership – what we do in paediatric cancer research of brain tumours can be leveraged for neuroblastoma, leukaemia, or any other paediatric disease because they face the same challenges that we do,” he said.

Currently, Dragon Master and the hospitals are trying to encourage researchers who are not interested in paediatric brain tumours to work on the data to make it more accessible and expand its reach.

“It turns out that when you begin to look at paediatric brain tumours, you identify mutations or alterations that identify certain processes … which may be relevant to other cancers and disease settings in ways that we couldn’t have predicted,” said Resnick.

“We want to be able to recruit those who are interested in those processes to work alongside us.”

The organisations are currently working on a pilot project that analyses and compares mutations in diffuse intrinsic pontine glioma (DIPG), an incurable childhood brain stem tumour, and fibrodysplasia ossificans progressiva (FOP), a disease where muscle, tendons and ligaments are gradually replaced by bone.

Mutations that occur in DIPG also occur in FOP, said Resnick.

“Somehow the pathology between these diseases overlap. They have the exact same mutation – one of them occurs in the germ line, meaning the patient is born with the mutation, or the mutation is acquired in the process of the cancer forming for reasons that we don’t know,” he said.

Resnick said this finding highlights the tremendous need for data dissemination and access, and collaborative platforms that enable people to share and work on data together when they are studying the genomics of cancer.

Working with Dragon Master supports opportunities to make these discoveries, he said. But data analysis needs to happen at a much more rapid pace globally, he said.

“It’s clear that the analysis of data is still in its infancy – meaning there are still many ways to analyse the data and not one universal right way to do it,” he said.

He argues that researchers should gravitate towards paediatric data – which is not as accessible as adult data – to figure out how to analyse data better.

“I think if we do that people will make discoveries that they would not have made otherwise,” he said. “Again, it comes back to the need to create infrastructure that makes data more accessible.”

Dragon Master also wants to attract people outside the medical field, ‘data geeks’ who are working with big data in other industries to see if analysis techniques they are using could be applied to the initiative, said Haddock.

“We are hoping that will be have a think-tank in the next year or two to explore those different avenues,” Haddock said.

Resnick agrees that there’s merit in in this non-standard mode of thinking about heatlhcare and disease-related research. For example, when NASA discovers new planets or stars, the organisation makes the data available to the do-it-yourself astronomer to analyse the information, he says.

Read: Data analysis and mathematics used to predict Earth-like planets

“I’m not suggesting that do-it-yourself biology will solve cancer, but we have no ego in the process as scientists, and recognise that sometimes providing access to data provides opportunities for the individual to make discoveries,” he said.

“Scientists within CHOP or any other hospital don’t have primacy or unique, sacred access to data that other people should not. There’s an opportunity to think about genomic data in multiple different ways and there’s no validated way that is the answer yet.

“Maybe there will be in the near future but currently there is not and we would love people to work with paediatric data,” he said.

Follow CIO Australia on Twitter and Like us on Facebook… Twitter: @CIO_Australia, Facebook: CIO Australia, or take part in the CIO conversation on LinkedIn: CIO Australia

Follow Byron Connolly on Twitter:@ByronConnolly