A Cancer 'Moonshot' Needs Big Data
Analyzing vast genetic and clinical data from hospitals and doctors would lead to revolutionary advances.
In his State of the Union address on Tuesday, President Obama called for America to become “the country that cures cancer once and for all.” As a three-time cancer survivor (metastatic colon, metastatic melanoma and metastatic prostate), I can tell you that this “moonshot,” as Vice President Joe Biden first called it, is a bold goal—but one within our grasp.
Last week’s report from the American Cancer Society shows that cancer mortality is down more than 20% over the past 20 years. Many patients are living longer thanks to better treatments and earlier detection. Science is tipping the odds of survival in favor of patients.
Ironically, we’re handicapping ourselves in the war on cancer, in part because of a web of privacy regulations like the Health Insurance Portability and Accountability Act. HIPAA makes it difficult for researchers to tap into large caches of clinical and genomic data shared across multiple institutions or firms, and then share their findings more broadly.
The law allows some research uses, but only if the uses (and informed patient consent) are specified in advance. As one analyst put it, “because obtaining [consent] from huge numbers of people or [institutional review board] waivers ranges from the impracticable to the impossible, important research has gone undone and important findings unshared.”
Harnessing that information—“big data”—would allow us to personalize prevention and treatment based on the genetic characteristics of a patient’s tumor, family history and personal preferences, while minimizing unwanted side effects. But today cancers are often fought “off the grid.” Patients whose cancers resist standard treatment, or whose tumors reappear years later, are medical puzzles. Their doctors cobble together treatments through intuition, experience and case studies scattered in the medical literature.
The clinical trials that pharmaceutical companies rely on for FDA approval and drug labeling capture too little of the information patients and physicians need. The trials only enroll 3% of cancer patients and can take years and tens of millions of dollars to finish. Many trials never enroll enough patients to get off the ground.
Big-data analysis could help tell us which cancer patients are most likely to be cured with standard approaches, and which need more aggressive treatment and monitoring. Harnessing the genetic and clinical data routinely generated by hospitals and physicians would also accelerate drug development, by rapidly matching targeted treatments sitting in companies’ research pipelines with the patients who are most likely to respond. This could save lives, streamline drug research and reduce ineffective care.
Insurance companies also could use big data to design contracts with drug companies that link payments to patient outcomes in specific groups of patients, such as fewer side effects or reduced hospital visits. This would ensure that drug prices reflect their real value to patients.
Better metrics for measuring patient outcomes are vital, but current debates on cancer drug pricing largely miss the mark. The total share of U.S. health spending on cancer—on all care, not just drugs—has remained constant at 5% for decades. New cancer drugs are roughly 1% of total health spending, or $42.4 billion in 2014, according to IMS Health.
The costs of cancer to the U.S. economy, patients, families and caregivers are far higher. A 2008 study in the Journal of the National Cancer Institute estimated the total value of lives lost due to U.S. cancer deaths in the year 2000 exceeded $900 billion, rising to $1.47 trillion by 2020.
Putting more cancers into remission—letting patients go back to work and keeping them out of hospitals, and lifting burdens on family members—will reduce those costs. One study in the journal Blood found that when a newly diagnosed patient with multiple myeloma was treated with effective drug therapy and went into remission, costs spiked for the first few months of drug treatment, then fell rapidly by nearly 70%.
Multiple myeloma is a case study on the need for big data in routine cancer care. Better, and better tolerated, oral therapies are keeping patients healthier, longer. The FDA has approved at least 12 new treatments since 2000, including three in November. The new treatments are already being tested in combination with other drugs that unleash patients’ own immune systems to attack tumors. But we’ll learn the results largely through trial and error unless we can figure out who should get which treatments and in which sequence to put, and keep, patients in long-term remission.
Giving patients rather than regulators power to control their own data, deciding whom to share it with, and when, would help break down the barriers that prevent data sharing. The next step is financial incentives.
Here, real-world data on clinical outcomes, including safety and efficacy, should be used to expand a cancer drug’s FDA-approved label. These labels heavily influence insurance reimbursement and how drugs are used. Expanding the labels would allow physicians to rapidly identify patients who could benefit from new uses.
Even better would be shifting from organ- and disease-based labeling to a continuing process that could be updated as we learned more about a medicine’s effects in patients with similar clinical and genetic profiles. A similar provision is already in the 21st Century Cures Act, passed by the U.S. House of Representatives in July.
Congress should unleash big data in America’s moonshot on cancer, giving patients and physicians the tools they need to gain the most precious commodity of all: more time.
This piece originally appeared in The Wall Street Journal
______________________
Dr. Tom Coburn is an advisor to the Manhattan Institute's Project FDA and a former two-term U.S. Senator from Oklahoma. Follow him on Twitter here.
This piece originally appeared in The Wall Street Journal