Big Hopes for Big Data: Digital Information Focus Underpins Cancer Moonshot Goals

Daniel F. Hayes, MD

Daniel F. Hayes, MD

The passage of the 21st Century Cures Act¹ sparked unusually robust holiday celebrations throughout the cancer research community, which harbors high hopes for both the newly funded initiatives and future budget increases.

The act will provide the National Cancer Institute (NCI) an extra $1.8 billion to fund the Cancer Moonshot, an effort to make 10 years of research gains in just 5 years by directing new money to 10 potentially transformative research areas (Chart, Table). It also gives the FDA the ability to consider real-world clinical data in determining drug indications.

The Moonshot money is the first major boost in federal spending on cancer research in many years, but researchers see reason to hope that it won’t be the last. The surge in new cancer drugs², the improvement of Big Data analysis³, and the often incredible results from immunotherapy trials seem to have increased enthusiasm for research spending among politicians and the public.

“As with everything in cancer research, there are some caveats. The vote in December only allocated the first year of funding, and the extra money could go away next year. All that duly noted, the 21st Century Cures Act is very exciting,” said Daniel F. Hayes, MD, president of the American Society of Clinical Oncology (ASCO). “It should provide both an increase in funding for the National Institutes of Health, and therefore the NCI, as well as additional funding for [former] Vice President Biden’s Cancer Moonshot initiative.

How Big Data Links Plans

“NCI funding has been flat for a decade now, and inflation has decimated real spending to the point that only a tenth of research proposals get funded. If this new initiative is a harbinger of steady increases, then it’s very good news for cancer researchers trying to get support and, more importantly, for our patients who will benefit from it,” said Hayes, who is also co-director of the Breast Oncology Program at the Michigan Comprehensive Cancer Center. “Even if it turns out to be a onetime deal, at least it directs a significant amount of money to many areas that promise big returns.”Asked to name the most promising of the Moonshot’s 10 research foci, Hayes replied that doing so would be as hard as naming the most promising of his children. Asked what tied the various aspects of the Moonshot together, he answered with a single word: data. Most of the specific research recommendations envision the aggregation of huge amounts of data and the extraction of important discoveries from those collection points.

More than 50 years have elapsed since data analysis produced one of the biggest ever breakthroughs in the war against cancer: indisputable proof that smoking dramatically increases cancer risk.⁴ Observers have been predicting for decades now that increasing computer power and digital storage would enable many more analytical triumphs. Although those predictions have yet to come true, many see more promise than ever in Big Data.

The importance of information analysis to the Moonshot becomes apparent in the very first focus area: offering comprehensive tumor profiling to a large number patients who agree to participate in a national network that will track what strategy works on what tumor and provides researchers an easy way to instantly find people who qualify for particular clinical trials.

Other goals that explicitly seek to use Big Data analysis or to encourage such studies include the creation of a national data ecosystem that allows sharing between many existing and all future datasets, the retrospective analysis of biospecimens from patients treated with existing standards of care, the generation of 3-D tumor atlases that map the development of every tumor, and the creation of a bank of tumors categorized by responses generated with various forms of immunotherapy.

The raw technology for most of these initiatives has been available for years now, but a lack of digital data and the inability—or unwillingness—to share what data existed have always limited the impact of data mining on health research.

Many industries had switched entirely to electronic records by the mid-1990s, but less than 30% of all office-based physicians were using electronic health record (EHR) systems just a decade ago.⁵ The percentage of US physicians using some sort of EHR didn’t break 80% until 2014. Even then, most doctors used only basic systems. Nearly 40% of doctors had yet to start storing and viewing imaging results electronically.⁶

That information that was stored electronically, moreover, has been mostly unavailable to researchers hoping to mine it for hidden treasure. Privacy laws, turf wars, and the technical incompatibility of different collection and storage systems have made it nearly impossible to aggregate truly huge amounts of information in ways that would maximize its value for would-be analysts.⁷

The federal government and private organizations have been working for several years now to increase the ability to securely share data, often information that has been anonymized, across multiple platforms. The government has attempted to create standards with programs such as the Health Information Exchange.⁸

Private organizations have sought to build datasets that lend themselves to research, datasets such as ASCO’s CancerLinQ program, which already has enrolled more than 1500 physicians and accumulated more than 1 million patient records.⁹ Hayes voiced particular enthusiasm about CancerLinQ, predicting that it “will be a game changer, both for individual practice quality control and for highly granular observational research in the future.”

Supporters of the Moonshot hope that the new money, along with the publicity surrounding it, create a sense of urgency about breaking down the remaining barriers to data aggregation, all while protecting patient privacy and intellectual property.

“We must agree to a radical new approach to sharing data. Not only should there be more team scientific research, but those researchers should share their data and learnings more broadly with outside researchers and the patients involved in the clinical trials for new products,” said Jeffrey Bluestone, PhD, president and CEO of the Parker Institute for Cancer Immunotherapy, speaking in a statement about the 21st Century Cures Act. “It is time to break down the silos and barriers to advancing knowledge for the best interests of patients who are waiting for new cures.”

Bluestone, a professor and noted immunologist at the University of California San Francisco, is charged with implementing the collaborative mission of the institute, which technology entrepreneur Sean Parker launched last year with a $250 million grant. The institute brings together more than 300 scientists at 6 of the nation’s top cancer centers to share in immunotherapy studies including the administration of intellectual property. One early project calls for a first-in-human trial using CRISPR/Cas9 gene-editing technology.

Practical and Research Uses

The data mining advocated by the new act isn’t just about facilitating the development of new cures. It’s also about helping clinical physicians and their patients make better use of existing treatments by giving them unprecedented amounts of information about how people with genetically similar tumors have responded to various treatment options.

The explosion in treatment options for many cancer types has already made it financially impossible to run trials large enough to compare all the possible combinations to determine which therapies should be used in which order against tumors with particular genetic mutations. Analyzing outcomes among thousands of real-world patients, even if it doesn’t provide the sort of definitive answers that randomized trials do, may well be the only way to gather data on a variety of regimens for different tumor types. Such analysis might narrow the field enough to make randomized clinical trials for the top contenders feasible, or it might determine standards of care on its own.

Under provisions contained in a different portion of the 21st Century Cures Act, data from real-world patients can also be weighed by the FDA in expanding the indication of an already approved medication. Regulators can use such information to supplement trial results, or, if they find the real-world outcomes compelling enough, they can approve new indications based on that data alone.

This provision, which applies to all types of medication, not just cancer drugs, is probably the most controversial component of the new act. Opponents say it is a dangerous gift to drug makers that will lead to unwarranted indications. Proponents say that, used properly, it may be the best part of the act: a scientifically valid method to greatly speed the rollout of drugs with multiple uses.

Existing clinical data may be enough to win expanded approvals for some medications in the relatively near future, but opportunities will multiply for developers of cancer drugs as the Moonshot gets underway and exponentially more clinical data become available to them. These data will come not only from the government, academia, and clinical practitioners but also, the designers of the Moonshot believe, from pharmaceutical companies themselves.

“Pharmaceutical companies have historically been reluctant to share data. Their model has been to use their considerable financial resources to generate proprietary information and use that information to develop drugs. As we move to precision medicine, though, they’re going to need more data to make new targeted medications than any single company can collect. Pharma has already experimented some with data sharing, and I think it will jump in with both feet when it becomes clear how much the benefits of seeing outside data outweigh the costs of sharing their own data,” said Elizabeth M. Jaffee, MD, who co-chaired the blue-ribbon panel that chose the Moonshot’s 10 areas of concentration.

Indeed, Jaffee believes that the unexpected success of much less ambitious efforts to harness Big Data will lead to widespread public, private, and academic participation in the new efforts.

“There was a controversy a few years ago about sequencing cancer genes. Some people argued that it wasted money to compile data that no one would ever use, and it turns out that everyone uses it. If you want to know what percentage of pancreatic cancers express a particular pathway, you can look it up in 30 seconds, and that’s hugely valuable for testing hypotheses,” said Jaffee, the deputy director of the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins.

“Thanks to that success, there’s overwhelming support for the Moonshot plan to move from these one-dimensional tumor models that focus only on genes to 3-dimensional models that provide a complete view of tumor proteins and immunology as well,” she said. “It’s a similar story with the other data initiatives. There seems to be widespread support for finally getting these things done.”

Many of the Moonshot’s 10 areas of concentration are very broad but there are several more targeted initiatives besides the development of the 3-D cancer atlas. Jaffee’s panel specifically recommended more research into childhood cancers. The panel believes that a better understanding of the onco-proteins that produce childhood cancers will uncover potential pathways for inhibiting them.

Another area of specific focus for the Moonshot is early detection and prevention, which currently gets a very small share of cancer research money. Many types of projects could qualify for funding, ranging from the development of more effective antismoking campaigns, to experiments designed to determine what genetic profiles warrant extra cancer screening to the development of cancer vaccines.

Big Gain for Immunotherapy

All the additional funding will not go to research topics that were not attracting big money before the passage of the 21st Century Cures Act. A sizeable chunk of the money will go directly to the research area that’s already attracting billions of dollars of investments per year: immunotherapy.

“As an immunotherapy specialist who remembers how difficult it was to get any research money just a decade ago, I’m obviously delighted that this field is now getting resources and making breakthroughs. I also believe that, for all the money immunotherapy is getting now, it will still be possible to generate major returns with this additional investment, if NCI takes the rather vague language of the legislation and creates a system that awards the grant money judiciously,” said Howard L. Kaufman, MD, assistant director for Clinical Science at the Rutgers Cancer Institute of New Jersey.

“Also—and this is going to sound odd coming from a guy who focuses on immunotherapy—we need to make sure that this one area doesn’t get too big a percentage of the research dollars. No one strategy is the answer and there’s a part of me that thinks the pendulum may have swung a bit too far away from targeted therapies,” Kaufman said. “I’d like to see more research into how targeted therapies and immunotherapies can be better used in conjunction because that is, I suspect, where treatment is headed for a number of tumor types.”

Kaufman, along with everyone else interviewed for this story, had no forecasts about whether the new legislation would prove to be an exception in the ongoing trend of declining public research dollars or whether it would prove a harbinger of better times ahead. All three expressed the feeling that technological advances, along with an improved understanding of disease, have created the opportunities for significant breakthroughs in the battle against many maladies, especially cancer. Increased investment in medical research at this particular moment would therefore create exponential returns, if politicians, philanthropists, pharmaceutical executives, and the public decide to make such an investment.

“The mortality numbers don’t show it yet, but I think we’re already in a period of very significant gains in oncology, sort of like cardiovascular medicine was when statins were first introduced. They knew they had something pretty important, but it took almost 20 years for the mortality figures to show the full extent of the impact,” Kaufman said. “There are just so many conditions in cancer now where people are living dramatically longer, such as advanced melanoma where they used to live 6 to 9 months and currently live for 24 months or more... And significant discoveries keep coming. It’s a good time to invest.”

References

1. H.R.34—21st Century Cures Act. Congress.gov database. https://www.congress.gov/bill/114th-congress/house-bill/34?q=%7B%-22search%22%3A%5B%22HR34%22%5D%7D&r=2. Passed December 13, 2016. Accessed January 6, 2017.
2. Buffery D. The 2015 oncology drug pipeline: innovation drives the race to cure cancer. Am Health Drug Benefits. 2015;8(4):216-222.
3. Lu J, Hales A, Rew D, et al. Data mining techniques in health informatics: a case study from breast cancer research [Lecture Notes in Computer Science]. In: Renda ME, et al eds. Information Technology in Bio- and Medical Informatics. Basel, Switzerland: Springer International Publishing; 2015(9267):56-70. doi:10.1007/978-3-319-22741-2_6.
4. Proctor RN. The history of the discovery of the cigarette—lung cancer link: evidentiary traditions, corporate denial, global toll [published correction appears in Tob Control. 2012;22(1):62]. Tob Control. 2012;21(2):87-91. doi:10.1136/tobaccocontrol-2011-050338.
5. Centers for Disease Control and Prevention. Use and characteristics of electronic health record systems among office-based physician practices: United States, 2001—2013. https://www.cdc.gov/nchs/ data/databriefs/db143.htm#x2013;2013%3C/a. Published January 17, 2014. Accessed January 6, 2017.
6. Heisey-Grove D, Patel V. Any, certified, and basic: quantifying physicians HER adoption through 2014. Washington, DC: Office of the National Coordinator for Health Information Technology. ONC Data Brief No. 28. https://www.healthit.gov/sites/default/files/briefs/oncdatabrief28_certi ed_vs_basic.pdf. Published September 2015. Accessed January 6, 2017.
7. Jensen PB, Jensen LJ, Brunak S. Mining electronic health records: towards better research applications and clinical care. Nat Rev Genet. 2012;13(6):395-405. doi:10.1038/nrg3208.
8. Health information exchange (HIE): standards and interoperability. HealthIT.gov website. https://www.healthit.gov/providers-professionals/standards-interoperability. Updated February 27, 2014. Accessed January 6, 2017.
9. CancerLinQTM platform hits milestone of having more than one million patient records in system [news release]. Alexandria, VA: American Society of Clinical Oncology; October 20, 2016. http://www.asco.org/about-asco/press-center/news-releases/cancer-linq-platform-hits-milestone-having-more-one-million. Accessed January 6, 2017.