In early May, IBM Watson Health announced new collaborations with 14 of the nation’s leading cancer institutes. The new partnerships give oncologists access to the supercomputer via a hybrid-cloud system that allows them to more effectively match their patients with available clinical trials and analyze genomic data to ensure patients get the most effective personalized treatments.
IBM’s boosted efforts around cancer research represent a unique effort in both the realm of cognitive computing and cancer research, while bolstering the trend of declining costs in personalized medicine that will someday allow all patients to harness contemporary knowledge of the human genome.
The newly participating medical centers include: Ann and Robert H. Lurie Children’s Hospital of Chicago; BC Cancer Agency; City of Hope; Duke Cancer Institute; Fred & Pamela Buffett Cancer Center in Omaha, Neb.; McDonnell Genome Institute at Washington University in St. Louis; New York Genome Center; Sanford Health; University of Kansas Cancer Center; University of North Carolina Lineberger Cancer Center; University of Southern California Center for Applied Molecular Medicine; University of Washington Medical Center; and Yale Cancer Center.
The aforementioned institutions join early partners like Memorial Sloan Kettering Cancer Center, University of Texas MD Anderson Cancer Center, the Cleveland Clinic and the Mayo Clinic, where Watson first demonstrated how high-performance computing, artificial intelligence and personalized medicine collide for patient benefit.
Sean Hogan, vice president and general manager for healthcare at IBM, says Watson and cognitive computing represent a fundamental shift in how the world understands and uses technology today.
“What we’re in the middle of right now is there’s all of the sudden information becoming available digitally about individuals and their heath and we can start to tap into and put together with medical knowledge,” Hogan said. “The vision that Mayo has is they want to make the best treatments available to the patients and include in the treatment options any clinical trials that may be available and appropriate to that individual. And that is a not-insignificant challenge.”
Cancer patients usually want to explore all their treatment options, including anything new, but with about 8,000 clinical trials conducted at the Mayo Clinic alone each year, identifying suitable trials can be a tedious, manual process. Furthermore, once a patient begins down one path of treatment, it closes the door on other potential options. By consulting a vast database of medical and clinical trial knowledge, Watson can assist clinicians in identifying which trials a patient might take advantage of early in the diagnostic process. Even better, Hogan said, the computer returns qualitative data that explains why a patient is or isn’t eligible for a given trial, allowing for intelligent treatment.
“There’s constant advancement in the development of new drugs that can treat cancers, and there’s also an associated complexity of sometimes combining cancers,” Hogan said. “It’s a complex disease and has many variations to it, and it even evolves in the course of an individual’s experience with it, so the treatment can evolve as well.”
IBM reports that by the end of 2015, a broader set of patients will gain access to personalized medicine, thanks to developments made by earlier Watson cancer research. The human genome is still mysterious in many respects today, but there are also known markers that allow clinicians to identify which treatments will work best for a given patient. IBM’s work with the New York Genome Center has assisted in some of these revelations.
“What they look for is combinations of variants, the variants between the protein makeup and the DNA makeup of the mutant gene and the normal gene,” Hogan explained. “So that variant brings you to the proteins that are behaving differently, then you can start to say ‘OK, well what are the known chemotherapies or drugs that address or target that protein?’ There may be several drugs associated with treating each mutant gene, and that’s where the massive amounts of information start to come into play, because if you’re going to combine those treatments, you have to also understand how it might cause an adverse drug reaction that could cause harm.”
Clinicians will use the cloud-based Watson Genomic Analytics program for genomic analysis. Watson consults its stores of treatment guidelines, research, clinical studies, journal articles and patient information and returns to the user a list of relevant medical literature, along with a list of any drugs identified therein. Watson can be especially beneficial in the treatment of fast-acting cancers like glioblastoma, a type of brain cancer, which typically means a life sentence of about 12 months. Spending less time choosing treatment options and waiting for results means more time for actual treatment and ultimately results in better care for patients, Hogan said.
IBM is accompanied in the world of personalized medicine by several other leading companies, which include Nanthealth and SCP, along with 23andMe, a consumer-focused genomic service, but IBM’s effort is unique, said Anurag Gupta, a medical doctor and research vice president at Gartner.
“What IBM is trying to do is to create an ecosystem of players,” Gupta said. “It’s basically an advanced data mining initiative for all the medical literature and then suggesting the possible treatment for the doctors. That was the initial Watson phase. Now what has changed in the last two or three months is IBM has decided to make an ecosystem play.”
In April, IBM acquired Explorys, a healthcare intelligence cloud company, and in May, IBM acquired Phytel, an integrated population health management software company. Along with investments in the device market, IBM’s efforts point toward a broad approach unlike what anyone else in the industry is going, Gupta explained.
While cancer patients using IBM’s partner institutions for treatment may gain access to physicians and facilities harnessing genomic data, personalized medicine has a few barriers to overcome before it becomes part of the popular medicine culture, Gupta said. Those barriers to popular personalized medicine include cost of analysis, the amount of time is takes to return results and integration with the physician’s interface, which in many cases today means electronic medical records (EMR).
The cost of genomic analysis started at billions of dollars and is now down to the thousands. When the price of a genomic analysis device comes down to $100, Gupta predicted personalized medicine will have greater opportunity to become widespread, not just for cancer but for less serious illnesses too. Some companies are promising genomic data analysis results within a few hours, which would also make the technology more practical, Gupta said.
The genes are tricky, Gupta said, because sometimes genes don’t express themselves and no one understands all the mechanisms of how, why and when that happens. “It could be that that gene expression is being suppressed by an entirely different gene on a different chromosome,” Gupta guessed. “What that means is that when we try to do personalized medicine, we are going down a maze in a certain direction, but that maze may lead us nowhere.”
More than 585,000 Americans died of cancer in 2014. Though cancer remains common and often incurable, the disease has also become increasingly treatable in recent years. The risk of dying from cancer has dropped by 20 percent since 1991, and it’s technological advancement and innovative medical research like that conducted by IBM and its partners that continues to save lives and transform what was once a complete mystery into an understood and treatable illness.