2014-03-19

Co.Exist

IBM’s Watson Tackles The Tumor Genome, On The Way To Personalized Cancer Treatments

To develop a specific treatment for each person's cancer, doctors would need to learn their tumor's genetic code. It's a task too difficult for humans to do for every cancer patient. But it might not be too difficult for a problem-solving supercomputer.

IBM’s advanced artificial intelligence program Watson has been a food truck chef, a doctor’s assistant, a personal shopper, and of course, a Jeopardy contestant. Its newest job as a cancer researcher at the New York Genome Center could be an opportunity for its biggest feat yet: helping to save brain cancer patients’ lives.

Cancer has always been diagnosed and treated based mostly on where it shows up in the body. But in reality, every tumor is different, and with genome sequencing getting cheaper and faster, oncologists are starting to dream about personalized cancer treatments tailored to the DNA mutations that are actually causing the disease.

In a trial with the New York Genome Center, a consortium of hospitals and medical schools in the NYC region, Watson will be scanning thousands of mutations found in the tumors of 20 brain cancer patients who enrolled in a clinical trial and trying to match the mutations to cancer drugs available today. Doctors, then, could use this information to tailor better treatments in the short timeline the aggressive disease demands.

While it’s technically possible today for doctors to sequence a tumor and prescribe a personally tailored treatments, this feat is far too difficult and time-consuming to do in clinical practice with today’s tools. First, doctors must sequence the DNA in the tumor multiple times and then comb through billions of data points to determine all where the cancer cell differs from a normal cell. These steps themselves are a barrier today, but the costs of the gene sequencing and computing power needed are dropping every year.

But then comes the even more onerous part--doctors combing through what may be thousands of mutations found in the tumor cell and pinpointing the ones that are important. This is where Watson comes in.

“What we currently do is essentially go through the mutations by hand with a team of doctors ... going to the medical literature to figure out what they really mean and then going to see what’s on the pharmaceutical shelf. One by one, we’re trying to connect up which mutations might match which drugs,” says New York Genome Center CEO and scientific director Bob Darnell. “Rather than trying to do all of this informatics by hand, Watson can be used, instead of winning at Jeopardy!, to beat cancer,” he says.

The New York Genome Center is enrolling 20 to 25 patients in a clinical study with Watson approved by institutional review boards. All have glioblastoma, the most common and particularly aggressive kind of brain cancer that kills some 13,000 people in the U.S. each year. Their goal is to train Watson and evaluate its aptitude at speeding up the development of personalized treatments. In the long-term, says Darnell, it hopes to build a model that can be scaled up to all cancer patients based on the “prototype” being tested.

Steve Harvey, an executive in IBM’s Global Technology Services group, says that this process of screening all tumor mutations that today takes weeks or months is what Watson can reduce to mere minutes. “The amount of time it takes today really prevents you from scaling it to a large amount of patients,” he says.

The work is at the research stage, but it comes as IBM is pushing Watson out into commercial applications in a wide range of fields, including health care, education, and consumer mobile apps. In January, IBM announced a $1 billion investment in a new business group for Watson, based in New York City, and a $100 million venture capital investment fund. IBM’s Harvey sees that Watson, the genetics scientist, could be a lucrative role, too--one day in use not just in clinical practice at hospitals, but also in fields like agriculture.

“What’s really exciting is that we’re seeing the intersection of technology and biology,” he says. “It’s not just something that people talk about anymore.”

[Image: Brain scans via Shutterstock]

Add New Comment

1 Comments