IBM uses machine learning to tackle cancer
November 22, 2016
IBM is to use machine learning to understand why some cancers become resistant to drugs. In a five-year, $50m project with the Broad Institute of MIT and Harvard, researchers hope to discover the basis of cancer drug resistance.
The project will study thousands of drug resistant tumours and draw on IBM Watson's computational and machine learning methods to help researchers understand how cancers become resistant to therapies. The anonymised data will be made available to the scientific community to catalyse research worldwide.
While a growing number of treatments can hold cancers in check for months or years, most cancers eventually recur. This is in part because they acquire mutations that make them drug resistant. The development of drug resistance is a major cause of nearly 600,000 annual cancer deaths in the USA alone.
In a limited number of cases, scientists have discovered the cause of drug resistance, allowing the development of new drugs to overcome resistance. In most cases, however, the causes of drug resistance are not fully understood.
To help understand how cancers become resistant to specific therapies, Broad Institute will generate tumour genome sequence data from patients who initially respond to treatment but who then become drug-resistant. Broad will use genome-editing methods to conduct large-scale cancer drug resistance studies in the laboratory to help identify tumours' specific vulnerabilities. IBM scientists will use Watson to analyse these data and identify genomic patterns that may help researchers and clinicians predict drug sensitivity and resistance.
This partnership is expected to help lay a foundation for understanding the basis of drug resistance in cancer – especially the genetic mechanisms observed in patients – and accelerate research across the cancer community to turn knowledge of resistance mechanisms into therapies.
"Defeating cancer involves playing a high-stakes game of biological chess," said Eric Lander, founding director of the Broad Institute. “When we make a move with a therapy, cancer often responds with a counter-move by finding a way to become resistant. The key will be learning from clinical experience, so that we know cancer's moves in advance and can plan strategies to cut off its escape routes.”
He said knowing how cancers become resistant would ultimately require learning from hundreds of thousands of patients' experiences.
“We're proud to work with IBM to make an important start towards this goal, and to make the information broadly available to the scientific community," he said.
John Kelly III, senior vice president at IBM Research, added: "The Broad Institute is leading the industry in areas of cancer biology, genomics and computational biology, and we are proud to bring Watson's data prowess to help researchers learn more about one of most important medical challenges that too often stands in the way of effective cancer treatment. Watson is already being used in the clinic to aid clinicians in cancer care. Our hope is that this effort, if successful, could eventually lead to significant breakthroughs.”
He said that someday, patients who would not otherwise have options in their battle against cancer might have reason for hope.
"Currently, cancer researchers have access to genomic information from only a few hundred drug-resistant cancers samples," said Todd Golub, chief scientific officer and founder of the cancer programme at the Broad Institute. “In addition to the goals of this specific study, IBM and Broad are committed to advancing cancer research by sharing the data from thousands of tumour samples with the scientific community to accelerate progress everywhere against cancer. What we and many others will learn with this information will have important implications for the future of cancer precision medicine and cancer diagnostics."
Earlier this month, IBM and Quest Diagnostics launched a service available to oncologists to help advance precision medicine by combining cognitive computing with genomic tumour sequencing. The collaboration with the Broad Institute is the latest application of Watson for genomics to help researchers and clinicians tackle the challenges and opportunities of cancer genomics.
Broad Institute of MIT and Harvard was launched in 2004 to empower creative scientists to transform medicine. It seeks to describe all the molecular components of life and their connections; discover the molecular basis of major human diseases; develop effective approaches to diagnostics and therapeutics; and disseminate discoveries, tools, methods and data openly to the entire scientific community.