President Obama wants to double the National Cancer Institute’s budget and representatives in Congress on both sides of the aisle are supportive. But before our elected leaders escalate the long-running war on cancer, the Obama administration might want to review the past performance of cancer research, which emphasized finding new cures, not fine-tuning existing oncology treatments.
By the most objective measure of progress—improved overall survival—the results from the hunt for new cures are meager. “The age-adjusted mortality rate for cancer is about the same in the 21st century as it was 50 years ago, whereas the death rates for cardiac, cerebrovascular, and infectious diseases have declined by about two-thirds,” laments Harold Varmus, the Nobel Prize-winning cancer researcher and former head of the National Institutes of Health, in his otherwise upbeat assessment about the future prospects for cancer research.
This is why a redrawn battle plan—one that focuses on turning the treatment system into a research and learning system that can teach oncologists the best use of the weapons they already have—is long overdue. To understand why this shift in focus is essential, let first examine our record to date in hunting for new cures.
First of all, there has been some progress, of course. About 66 percent of people diagnosed with cancer today survive five years, up from 50 percent in 1975. And mortality rates for some cancers are falling.
The outcomes are never gathered. The data is never analyzed. And the findings are never disseminated.
But even that progress is under intense scrutiny. While the drug and biotechnology industries are lavishing increased attention on cancer (861 drugs and vaccines in clinical trials, according to a recent announcement) most of the newly approved agents that squeeze through the pipeline extend the lives of patients for only a few weeks or months, often at great expense.
The evidence suggests the greatest strides in reducing cancer deaths in recent years have come not from better drugs but from the nation’s war on smoking. What’s more, most of the improved longevity for cancer patients is an artifact of early diagnosis rather than improved care.
There’s no shortage of blue-ribbon committees wrestling with the problem of slow progress in the medical fight against cancer. The Institute of Medicine has a panel evaluating the government-funded clinical trial network. The National Cancer Institute has pulled together its own team for an in-house review.
But an obsessive focus on flaws in the government-funded clinical trial system may be missing the core problem. Government-funded trials are a minuscule portion of the overall clinical research enterprise. Most clinical research today is sponsored by industry, and those trials are focused largely on getting new drugs through the U.S. Food and Drug Administration approval process. While firms do test their approved drugs for new indications, it’s usually in limited populations, which may be large enough to generate a medical journal article but rarely reach the size needed to win FDA approval for that new use.
Even rarer are the head-to-head clinical trials that compare different chemotherapy regimens since few companies are willing to risk having their drug proved less effective than a competitor’s drug. Post-approval trials aimed at fine-tuning the best use of new drugs are equally rare since by definition such trials would limit their use to subsets of individuals with a particular form of cancer. Saving money for patients and their insurers—the total health care cost tab for cancer came to an estimated $93.2 billion last year—is not a very high priority for industry-funded research.
The result is that many of the nation’s 30,000 oncologists are engaged in what could be described as an unobserved and uncontrolled science experiment, especially when it comes to treating the 560,000 Americans who die each year from the more than 100 forms of the disease. As these patients’ cancers advance, their physicians try regimens they read about in journals or hear about from colleagues. The outcomes are never gathered. The data is never analyzed. And the findings are never disseminated.
Indeed, an estimated 70 percent of all cancer drugs are used off-label. In other words, most prescribed chemotherapy regimens have not been approved by the Food and Drug Administration for that particular use. Much of the off-label use is supported by the slimmest of evidence, often just a single trial in the medical literature of limited size and duration.
“Most of off-label use is done with good intentions,” observes University of Chicago oncologist Richard Schilsky, the current president of the American Society of Clinical Oncology. But he adds that “no data is collected on whether it’s an effective strategy or not. And most of it gets reimbursed.”
Meanwhile, only 3 percent to 15 percent of adult cancer patients (it depends on whom you ask) are enrolled in clinical trials that would show if those regimens are truly effective or refines their best use (when treatment should be started or stopped, for instance, or in what doses, in combination with what other treatment approaches, and for which patient sub-groups). “The extensive off-label use of drugs allows patients access without having to participate in clinical trials,” Schilsky said. “At some point the treatment crosses over from evidence-based to completely not evidence-based. And when you cross that line, the cost of care runs up and patients become diverted from participating in clinical trials.”
Pediatric oncology serves as the model for cancer researchers who believe enrolling more patients in clinical trials would speed up the process of learning which therapeutic interventions work best for which patients. The NCI-funded Pediatric Oncology Group, which is now part of the non-profit Children’s Oncology Group, over the past five decades enrolled about 90 percent of all children with cancer in clinical trials. This enabled pediatric oncologists to steadily refine the treatment regimens to the point where survival rates today are over 80 percent, up from 20 percent in the 1960s.
“We have made stunning progress” in treating childhood cancers, says Sharon Murphy, a visiting fellow at the Institute of Medicine and former head of the Pediatric Oncology Group. “The trials built over time on previous learning [by] incorporating better scientific understanding of the science of the disease into treatment. It led to risk stratification,” she explains, which is the holy grail of personalized medicine that researchers often talk about in adult oncology but, alas, has proven elusive except in a few situations such as HER2-positive breast cancers.
Yet not all pediatric oncologists believe repeating that approach in adult oncology will lead to similar progress. Adult tumors can take decades to develop and are resistant to treatment. Pediatric cancers, in contrast, usually “arise from embryonic development and develop into cancers that are much more susceptible to chemotherapy and radiation,” explains Joseph V. Simone, director of the University of Florida’s Shands Cancer Center.
In any case, getting more adults patients enrolled in clinical trials won’t be easy. Many cancer patients confronting their life-threatening illness are understandably reluctant to take a 50-50 chance of getting put in the arm of a clinical trial that doesn’t receive the experimental therapy. And that’s only half the problem. Most patients are treated by community oncologists in private offices, not at academic medical centers that get NCI financial support to enroll patients in trials.
“It takes a lot of time and support staff to carry out a clinical trial,” notes John Mendelsohn, director of the M.D. Anderson Cancer Center at the University of Texas in Houston and chairman of the IOM committee evaluating NCI’s clinical trial system. “That requires reimbursement.”
Medicare will reimburse physicians for the cost of patients’ usual care if they are enrolled in a clinical trial. But usual care doesn’t include the cost of setting up a system to solicit informed consent, collect complete pre-trial information, conduct rigorous follow-up, and perform all the other tasks crucial to generating scientifically valid data. The result, a recent study showed, is that just 30 percent of all cancer clinical trials listed in the government’s registry (www.clinicaltrials.gov) recruit all the patients required by the trial’s original protocols.
Given these constraints on rapidly expanding the clinical trial system, some experts say a better approach to rapidly improving cancer care would focus on collecting extensive information about the 1.3 million Americans who are diagnosed and treated for cancer every year. As the nation’s physicians move toward adopting electronic medical records (spurred on by the incentives in the Obama administration’s $767 billion economic stimulus plan), it should enable oncologists to record their patients’ demographic and genetic information (properly protected for privacy), their diagnoses, their treatments, and, eventually, their outcomes. This information could then be analyzed retrospectively to see what works and translated into guidelines for better care. Such a “learning system” would be a major step on the road toward personalized medicine, the proponents say, because the analysis could also identify which subgroups responded best to which treatments and regimens.
“Typically it now takes two or three years and a big investment” to get new information from clinical trials about a particular cancer treatment, says Lynn Etheredge, director of the Rapid Learning Project at George Washington University and an advocate for database-derived retrospective research. “In contrast, if the data is in the database, you can answer that question in a few hours, or maybe even a few minutes . . . This could be like hooking up the steam engine to the spinning jenny.”
The National Cancer Institute recently took a major step in building an information superhighway to serve a cancer learning network. The ambitious goal behind CaBIG, the Cancer Biomedical Infomatics Grid, is nothing less than turning the billions the nation spends each year on cancer care into that “learning system.” The agency has already linked the 60 NCI-funded cancer research centers and a network of community cancer centers where 85 percent of patients receive care.
The big problem now is figuring out how to convince the nation’s oncologists to drive on that highway. Kaiser Permanente, whose 14,000 salaried physicians in nine states and the District of Columbia include about 200 oncologists, is already there, since its medical records are already fully computerized. Kaiser oncologists do not enroll their patients in clinical trials, which would require informed consent and institutional review board review. But each oncologist in the Kaiser system follows a pre-specified protocol for each cancer. They are allowed to deviate from that path if and when the cancer progresses and in ways tailored to the individual’s needs.
Each of those new regimens follow a pre-specified path, too. “The goal is not to restrict physicians from personalizing therapy; it’s to create a framework so it can be tracked back to see if it made sense to use that variation,” explains Paul Wallace, medical director for health and productivity management at Kaiser Permanente.
UnitedHealthcare, which has about 20,000 patients on cancer therapy at any point in time, recently set up a demonstration project made up of six panels of 10-to-25 oncologists. Each group will use pre-specified protocols for their patients. The goal over the next few years is to generate data that will reveal which approach worked best for which patients.
Recruiting physicians in its provider networks wasn’t difficult. “I had more volunteers than I had space,” says Lee Newcomer, senior vice president for oncology at UnitedHealthcare. “Lots of doctors want this kind of approach so they’ll learn something.”
But beyond these pioneering efforts, it is going to take a lot of convincing to get large numbers of oncologists to begin entering their patients’ data into CaBIG. Cancer doctors, like physicians generally, have been taught that prospective clinical trials are the gold standard of medical research, and can’t be replaced by a retrospective analysis of outcomes.
That’s where Medicare comes in. The Center for Medicare and Medicaid Services, which pays for the lion’s share of cancer care, could create new financial incentives to get physicians involved—by underwriting their transition to electronic health records and by beefing up reimbursement for data collection. There have some tentative steps in that direction through the coverage with evidence development process, instituted by former CMS chief Mark McClellan, now at the Brookings Institution. In a few cases, CMS paid for experimental therapies or imaging tests as long as physicians gathered data on outcomes so it could be determined later if those treatments truly deserved permanent coverage.
“Medicare has to play the leading role,” argues Etheredge. “Previous efforts to evaluate therapies came out of the regulatory framework. What we’re talking about is learning as much as possible as quickly as possible about new technologies after they’ve been introduced.”