By Krista Conger
Illustration by Anita Kunz
Sylvia Plevritis was excited. It was December 2003, and she had just learned that the National Cancer Institute was offering millions of dollars to researchers in a variety of non-biological fields to study how cancerous tumors behave and grow. She told her boss, Gary Glazer, MD, chair of Stanford’s radiology department, “This is my Christmas present. They are talking to me.”
So Plevritis, who has a PhD in electrical engineering, emailed mathematicians, computer scientists, engineers and biochemists across the campus — anyone she thought would be interested in pursuing the grant. Her message went viral as recipients sent it on to others who might want to be involved.
At the time, Plevritis was deep into her second career: a public-health-focused effort to optimize magnetic resonance imaging technology to diagnose breast cancers. But she wasn’t a laboratory scientist, or a physician.
“At first I wasn’t focused on the molecular biology of the tumor, but on how to deliver data that a health policy-maker would need to recommend new clinical guidelines,” she says. “But as I grew more curious about cancer progression, I’d read papers discussing the effect of one gene on one pathway in a cell, and I’d think, ‘We’re looking at this disease in a very narrow way. We need to think methodically of an underlying network of hundreds or thousands of interactions that drive a cell to divide.’” In other words, a kind of a circuit.
With that, Plevritis faced another conundrum. “I spent four years building a funded research program to study breast cancer screening, but I was increasingly interested in the natural history of the disease. There was only so much I could squeeze out of the clinical data to understand cancer progression, and I felt I needed to dig deeper. I needed to learn about the molecular world of cancer, and that put me at another crossroads in my career. I could either continue with public health, or change direction again.”
In the end, she decided to make the switch. But she wasn’t on her own. Increasingly, investigators are realizing the advantages of viewing cancer as a system, rather than individual mutant cells. Combined with the many advances we’ve made in the past few years in understanding the molecular biology of healthy and cancerous cells, we stand a chance of finally making significant inroads against a disease that has captured the fear and imagination of our nation for four decades.
“Revolutions in cell and molecular biology, biomedical engineering and information sciences are creating the potential for a new generation of more effective and less toxic cancer therapies,” says Beverly Mitchell, MD, who directs the Stanford Cancer Institute. “We can now systematically identify many of the molecular roots of disease and employ powerful computational systems to interpret a dizzying amount of data.”
Since President Richard Nixon declared a “war on cancer” on Dec. 23, 1971, we’ve seen a slow and steady decline in the death rates for several common forms of the disease, but many cancers remain deadly. With the exception of a few precious triumphs, the diagnosis still often feels like nothing less than a heartbreakingly punctuated death sentence: Treatments can lead to disease-free intervals, but about 50 percent of adult cancer patients can still expect to die from their disease.
As Mitchell points out, this is poised to change. In the decade since the first drafts of the Human Genome Project were completed, eye-opening discoveries have become commonplace. As we’ve burrowed into the complexities of our genetic material, it has become easier to identify abnormalities in tumor DNA. We’ve learned about corrupted molecular signaling pathways, cell cycle missteps and drug resistance. We can watch in real time as cancer-killing cells home in on tumors in a human patient and even calculate the phenomenal physical force faced by the cells as they leave the shelter of the tumor and enter the bloodstream.
“We’re now prepared to ask some game-changing questions,” says Jerry Lee, PhD, deputy director of the National Cancer Institute’s Center for Strategic Scientific Initiatives. “We have an amazing amount of data. Now we need insight.”
But more than scientific insight is needed to eliminate some of the biggest obstacles to treatments and cures: a dysfunctional cancer clinical trial system, disastrous drug shortages and a health-care system unable to deliver cancer care at an affordable price. Without meaningful change at the policy level, many experts worry that we will squander the opportunity to bring these discoveries from the lab to patients.
“It is clear that our nation is at an important crossroads, where the science before us presents unprecedented opportunities to create new and better medical products and promote better health for the public,” writes Food and Drug Administration commissioner Margaret Hamburg, MD, in an October 2011 FDA report. “But we must act now and work together to capitalize on this groundbreaking science in order to bring safer and more effective treatments to American families and keep our position as the global leader in scientific innovation.”
“Basic scientists have opened a fire hose of information,” agrees biostatistician Phil Lavori, PhD, who chairs Stanford’s Department of Health Research and Policy. “There are many, many good ideas. But there are real problems in the ways we test these ideas and bring the resulting therapies to patients. If we can’t resolve these, we’re risking an incredible opportunity to make progress.”
Indeed, optimism about the scientific side of cancer research is pervasive. Concepts such as cancer stem cells, immunotherapy and oncogene addiction (a tumor’s dependence on the activity of a single gene) have exploded on the scene in the past 15 years. They’ve left us teetering on the precipice of finally understanding what makes a good cell go bad, and how we can mitigate the damage when it does.
Emboldened by these advances, researchers have begun to tackle the bigger picture of who gets cancer and why, and what can be done about it. For example, the National Cancer Institute recently launched the Provocative Questions endeavor, seeking answers to “important but non-obvious” questions, such as the role of obesity in cancer incidence, how an organism’s life span affects the molecular mechanisms of cancer development, and whether it’s possible to enhance survival — not by killing a tumor, but simply by keeping its growth static. Then the institute invited researchers to submit proposals by mid-November this year to answer one of 24 of these questions; about $17.5 million is up for grabs.
“We’re looking to change the research paradigm,” says NCI’s Lee, “by looking beyond the horizon. And we can do that by not only asking these types of questions, but also by providing the framework for robust data sharing that will allow researchers to ask and answer their own questions.”
The net effect of these advances is an exciting mix of getting up close and personal with the genes and molecules that drive uncontrolled cell growth while also taking a deliberate step back and striving for a more global understanding of the behavior of the many, varied disorders we call cancer. This wide-angle approach is embodied in an emerging concept called systems biology — precisely what the NCI was seeking to encourage in 2003 with the grant announcement that sparked Plevritis’ imagination.
Systems biologists often talk of an effort to crack the “black box” of a cancer cell — in effect mapping a complex circuit of cell growth and death honed by millions of years of evolution. The goal, explains Plevritis, now an associate professor of radiology and a Stanford Cancer Institute member, is to “look at cancer as an integrated system. In the past, people have studied metabolism, or cell death, or oxygen use, or any number of other events in cancer cells. We want to look at all these processes simultaneously and how they regulate one another.”
For example, taking such a long view of cancer can identify parallel gene signaling pathways, or networks that contribute to drug resistance. Knocking out one pathway with a targeted therapy may slow the cancer’s growth, but the cell can still use the alternative route. Blocking both simultaneously takes advantage of a concept known as synthetic lethality and can be a way to specifically kill cancer cells while sparing normal tissue.
“This is a powerful approach that is just in its infancy,” explains Amato Giaccia, PhD, professor and director of radiation oncology research at Stanford. “In this scenario, neither of the two drugs alone will have much effect, but together they can be lethal to the cancer cell.”
Furthermore, the cancer cells may have established unique pathways rarely used in normal cells. “I believe that the number of ways that a cell can become cancerous is not infinite,” says Plevritis. “There are specific drivers, or gene mutations, that orchestrate this transition. If we can identify these drivers and then target and manipulate them, we may be able to move cancer cells into a more benign state.”
Some of these are oncogenes, genes that when mutated, drive uncontrolled growth. Others are genes that govern processes like differentiation that funnel cells into developmental dead ends, where they can do little harm.
Plevritis speculates that there may be as few as 100 drivers. “It should be manageable to understand the system. Once we uncover a network of relationships on a molecular level, when we perturb the system, we can understand how the network responds and design more effective therapies.”
Systems biology may be one of the newest kids on the cancer therapy block, but it owes its existence to a more familiar concept — personalized medicine. Stanford immunologist and oncology division chief Ron Levy, MD, pioneered the technique in the mid-1980s when he hit upon the idea of using monoclonal antibodies to attack cancer-specific molecules on the surface of lymphoma cells. The technique required generating a new batch of unique antibodies for every patient, thus personalizing their treatment.
“For some types of cancer, monoclonal antibodies totally changed the game,” says Levy. “However, although this type of treatment can put people in remission, it only cures a few of them. We need to take them up another notch in potency, to make them work better, and to take advantage of these remissions we’re inducing.” Levy believes the key lies in further activating the immune system against the cancer.
While some personalized medicines, like the prostate cancer therapy Provenge, follow Levy’s original model of generating a unique treatment for each patient, others, like the breast cancer drug Herceptin, work by sorting patients into smaller subgroups and assigning treatment based on the pattern of gene expression in their tumors.
“Personalized medicine is an age-old idea,” says Levy. “Physicians have always been thinking about how best to treat individual patients. But advances in technology and understanding about tumor biology have allowed us to add a whole other dimension. Now we can test the gene expression levels and DNA sequences in each person’s tumors and match them with a therapy most likely to work for them. The potential is amazing.”
Health economists are quick to point out that this potential sometimes comes with a staggering price tag: Three rounds of Provenge, marketed by the Seattle-based company Dendreon, costs about $90,000. Roche’s Zelboraf — a new treatment for advanced melanoma patients whose tumors carry a specific mutation — costs $56,000 for six months, and Seattle Genetics’ Adcetris, which links a tumor-targeted antibody with a cell-killing drug, costs about $108,000 per year.
“I wonder if we’ve now entered a mode where $100,000 per year is an acceptable price for a cancer drug,” says Stanford associate professor of pediatrics and bioinformatician Atul Butte, MD, PhD. “As drugs become more and more targeted to subsets of patients, the price tends to increase. I think the real challenge of personalized medicine now is to figure out how to get costs down.”
Levy disagrees. “It’s not automatic that the overall course of treatment will be more expensive,” he says. “There is a real possibility that we can drive down costs by making better treatment choices for patients. If we pick a treatment most likely to work, while also delivering fewer side effects, that can be cost-saving. Non-targeted cancer drugs can also be expensive, especially if they are associated with hospitalization for side effects. Then the cost to the system can be much higher than the new targeted agents.”
‘I wonder if we’ve now entered a mode where $100,000 per year is an acceptable price for a cancer drug.’
The outpouring of human genetic information over the past decade is being tapped as a resource for cancer treatments. Some of Butte’s work is an example of this. Butte, chief of the division of systems medicine in pediatrics, has used information from publicly accessible databases of gene expression patterns and genetic changes in cancer cells to identify which signaling pathways are likely to be disrupted in specific types of cancer and other diseases. He then pairs the diseases with existing drugs that target the same pathways. In this way, he’s found some unlikely combinations: a widely available, inexpensive medication for ulcers that seems to work in an animal model of a lung cancer called adenocarcinoma, for example. “It’s as if we’re panning for gold,” says Butte. “If I can find these types of associations in my bioinformatics lab, without first doing a wet experiment, we start to think, ‘Wow, how the world has changed.’”
Yet with cancer research going gangbusters, you can’t help but wonder: where are the cancer treatments we were promised? Many are stuck at the human testing stage, it turns out. According to a 2010 Institute of Medicine report, “the system for conducting cancer clinical trials in the United States is approaching a state of crisis. … If the clinical trials system does not improve its efficiency and effectiveness, the introduction of new treatments for cancer will be delayed and patient lives will be lost unnecessarily.” The report estimated that only about 60 percent of NCI-sponsored trials are completed and published — a figure it called “a terrible waste of human and financial resources.”
On the face of it, the concept of a comparative clinical trial is simple: Give one group of patients a proposed treatment and not the other group. Watch to see who improves. Variations on this theme are described in ancient texts, and a similar technique was used in the 1700s to show that citrus fruits can cure scurvy. To make the comparison fair and unbiased, randomly assign people to treatment groups and, to make it more precise, start with a homogenous group of patients.
Until now, we’ve had only a few ways to assess the variability among patients and their tumors. But in an age of personal genomics, clinicians can make many more distinctions. Unless you have an identical sibling, no one else can lay claim to the particular assortment of nucleotides that makes up your DNA. (Even if you have an identical twin, genetic modifications in utero are likely to confer subtle yet important regulatory differences in how your cells respond.)
As the depth of our understanding grows, it becomes clear that the old model of one condition, one drug, one large group of — theoretically homogenous — patients, and the luxury of years in which to come up with an answer is simply inadequate. The pace of medicine today, the need to tailor treatments to the genetic background of a patient and his or her tumor, and the concept of dynamically responding to changes in cancer growth all demand new clinical trial designs. We need rigor without rigidity.
Of course, any changes to the system must still lead to an outcome that will help patients: new treatments or drugs approved by the Food and Drug Administration.
“We are very interested in the possibility that some people will respond better to a drug than others,” says Robert Temple, MD, the deputy director for clinical science at the Center for Drug Evaluation and Research at the FDA. “We also support trial designs that include adaptive elements if a tumor marker predicts survival. If you find a marker that predicts no response, you could drop those people out of the study, for example. But the basic requirement for getting a drug to the marketplace will always be to show that it works in a defined population.”
Adaptive design allows researchers to alter participants’ treatment plans during the study in response to ongoing measurements of efficacy. Another concept, enrichment, involves enrolling only those patients who are most likely to benefit from the treatment tested. Doing so increases the likelihood of generating statistically significant, useful results.
“We want to learn with each patient we treat,” says Stanford Cancer Institute associate director Branimir Sikic, MD. “Right now, we markedly overuse chemotherapy drugs. Maybe half of all patients with common cancers are resistant and don’t benefit from the treatment. But they all get toxic side effects. Now we’re slowly but surely matching up what we know about key mutations and critical pathways in cancer with drugs targeted to those vulnerabilities.” Sikic’s upcoming ovarian cancer trial capitalizes on this knowledge to predict which of five chemotherapies commonly used to treat the disease will work best for each patient.
“We’re trying to harness the power of genomics to answer questions about drug sensitivity,” says Sikic. “If we predict that a patient’s tumor will be sensitive to a particular chemotherapy, and we’re right, the next patient with a similar genomic background will have a higher likelihood of being assigned that drug. With a reasonably accurate test, we hope to get an answer about the effectiveness of treatment after treating fewer than 100 patients. This is very different from the traditional clinical trial model that looks for performance in large groups of patients over long periods of time.”
Well-designed trials are useless, however, without patients to enroll in them, or drugs to use in them. “We’re going to have to get around the problem that less than 5 percent of adult cancer patients who could be participating in clinical trials are enrolled in one,” says Lavori. “The rate of participation is abysmal.”
The problem escalates rapidly when willing patients are stratified into smaller and smaller subgroups based on their particular combinations of biomarkers. Often many patients must be screened to find one who fits the necessary profile, and this process must be repeated for each enrollee.
In addition, chronic, worsening drug shortages are hampering the completion of even adequately enrolled trials. According to an October Scientific American article, 15 of the nearly 200 drugs in short supply in 2011 are cancer drugs required for clinical research; more than 150 NCI-sponsored trials involve medications with limited availability.
The drug shortage problem, which is considered by many physicians and patient advocacy groups to be a national emergency, has been worsening steadily since 2006. According to the FDA, about 75 percent of the shortages result from problems with product quality or delays in production or capacity. What’s not clear is what to do about the crisis. In February, U.S. Sens. Amy Klobuchar, D-Minn., and Bob Casey, D-Penn., introduced S296 — the Preserving Access to Life-Saving Medications Act.
The act requires prescription drug manufacturers to notify the FDA at least six months before any planned discontinuation or interruption in the production of critical drugs, or as soon as possible when unexpected events appear likely to cause a drug shortage. It would also require the FDA to notify the public of shortages and the plan to address them. However, as the agency’s Center for Drug Evaluation and Research’s Edward Cox, MD, noted in a public workshop in September, the FDA has no authority to require that a company produce certain types or amounts of drugs. Cox is the coordinator for CDER’s Drug Shortage Program.
“Manufacturing capacity is not something that we control,” says Cox.
Inadequate adult cancer trial participation is another problem without an obvious solution. In June, the Institute of Medicine’s Forum on Drug Discovery, Development and Translation held a workshop at the Mount Sinai School of Medicine to explore the promise of adaptive trial designs and how to increase public engagement in clinical trials. Participants noted many factors that may impede enrollment: a lack of awareness by physicians and patients of relevant trials, the reluctance of patients to agree to the time commitment and travel required for complex trials, and the difficulty of overcoming pre-existing preferences for certain medications or interventions, among others.
Lack of institutional support for those conducting clinical trials was also a concern. Several programs at Stanford aim to address these issues. Their success is evident in the numbers.
“Currently at Stanford more than 50 percent of children with cancer are in a clinical trial, and about 14 percent of adults,” says Sikic, who directs Stanford’s Clinical and Translational Research Unit and co-directs its Spectrum program, which helps researchers and faculty members better understand how to design effective clinical trials. “Nationwide only about 4 percent of adults with cancer are enrolled.”
The Stanford Cancer Institute typically has about 330 clinical trials open for enrollment at any one time, both adult and pediatric — which have much higher participation. As a result, pediatric cancers, unlike adult cancers, have seen some astounding clinical successes during the past 20 years.
“Until recently, we’ve acted as if the patient doesn’t have to be a partner in their care,” says biostatistician Lavori. “I think we need to reach out to patients and explain why it’s so important to participate in clinical trials.” Lavori is testing a new concept called point-of-care randomization in a pilot study at the national Veterans Affairs Cooperative Studies Program, which allows patients and physicians to quickly and easily participate in clinical research.
“Many times there are three or four possible treatments for a patient’s condition,” says Lavori. “There really isn’t a compelling reason for a physician to pick one over the other. At that point we would randomize the choice of treatments, and use an adaptive trial design to quickly feed back which approach is preferable for which patients. If there’s a winner emerging, we should be able to see it relatively quickly.”
He sees this as a way to increase clinical trial participation among adult cancer patients that will quickly and efficiently drive the identification of effective new therapies. According to Lavori, it’s absolutely critical to do so.
“We need to use our interactions with patients to their full advantage, and tell them, ‘If you don’t want your children dying of cancer in the coming decades, you need to demand and participate in clinical trials.’”
A dysfunctional clinical trial system. A nation struggling to manage a crisis in our drug supply line. And a health-care system that doesn’t blink at offering new treatments that far exceed in price most people’s annual income. What’s to be done?
Many experts agree that within the next five years, a cancer patient’s ideal course of care will look very different from just 10 years ago. Dependence on chemotherapy, radiation and surgery will be reduced; instead treatment will likely emphasize sequencing of both the patient’s genome and the tumor, and analysis of the tumor’s potential drug resistance. Treatment will be tailored to each patient and carefully calibrated to avoid promoting the survival of the strongest, most resistant disease-causing cells. The patient’s immune system cells may even be trained to specifically attack the tumor cells. Principles of evolution, population dynamics and mathematics might be incorporated to predict how the cancer cell population will respond to treatment, which will be monitored with unheard-of imaging capabilities.
But revolutionizing cancer care clearly also hinges on making big changes outside the laboratory. It will require support and innovation at academic medical centers around the country. “New opportunities inevitably create new challenges,” says Stanford Cancer Institute director Mitchell. “Some critical research technologies are too expensive for individual laboratories and instead require core facilities to be shared by many investigators. And although our ability to divide major categories of cancer, such as breast cancer, into many subtypes offers the potential for highly specific therapies, it also creates new demands on our clinical trials system.”
Programs like Stanford’s Spectrum help by teaching clinicians and researchers how to design clinical trials to test their laboratory-inspired ideas. And the Cancer Clinical Trials Research Office works to increase accrual of cancer patients and to lessen the administrative burden of ongoing trials.
But progress will also require many people like Sylvia Plevritis, who fight in the trenches and direct troops. When Plevritis realized she needed to know more about health care, she returned to school to earn a master’s degree in health services research to complement her PhD in electrical engineering. And when she heard about the NCI announcement, she brought together people who hadn’t tackled cancer before — mathematicians, engineers and even statisticians — with experts in genetics, cancer and medicine to brainstorm new ways to think about cancer. Their work paid off: In October of 2004 Stanford was one of nine institutions to receive funding to plan a full-scale center, and in 2010, Stanford received $12.8 million over five years to establish a Center for Cancer Systems Biology. “We’re in a very good place,” says Plevritis, who directs the new center. “We finally have the information and the structure to make a significant contribution to cancer care.”
At the national level, there have also been signs of progress. Although legislative action addressing drug shortages and spiraling health-care costs has been stymied by political and budgetary wrangling, President Barack Obama issued an executive order on Oct. 31 aimed at reducing prescription drug shortages by broadening the requirement that drug companies report possible shortages and expediting review of new drugs — much like the pending Klobuchar-Casey legislation. The executive order, however, also calls on the FDA to investigate possible instances of stockpiling or price gouging. Meanwhile, organizations like the FDA, the Institute of Medicine and the National Cancer Institute are striving to work together to increase the efficiency, timeliness and enrollment in cancer clinical trials. Despite promising signs, however, significant change will likely take time — time stolen from people fighting cancer.
“If we don’t rectify these issues, we may not make the progress we could be making,” says Lavori. “I want my children and grandchildren to be as free of the threat of cancer as we are now in this country of the fear of polio paralysis that gripped our country during the first half of the 20th century. Right now we are falling woefully short of that goal.”
A Breathalyzer-style test for lung cancer? It’s not your usual cancer diagnostic test, but if you think about it, it makes a lot of sense. Breath carries molecular clues about your health, especially about your lungs. And unlike chief alternatives for lung diagnostics — biopsy and CAT scans — a breath test causes no harm.
Daya Upadhyay, MD, who runs Stanford’s lung nodule clinic, has used her molecular biology skills to make just such a test, which she hopes will give patients a head start on fighting cancer. But does her test work? A seed grant from the Stanford Cancer Institute will help Upadhyay find out.
In addition to funding the seed grants ($850,000 to 17 investigators this year), the institute manages and monitors Stanford’s cancer clinical trials system; holds seminars, lecture series and other meetings to bring together potential collaborators from multiple fields; sponsors cancer educational programs; and organizes core facilities where researchers can share critical technologies that are too expensive for individual labs to buy.
As a National Cancer Institute-designated cancer center, the Stanford institute integrates multidisciplinary cancer research, training and education with comprehensive clinical care. It is made up of over 300 experts — investigators and clinician-researchers from more than 30 academic departments throughout the medical school.
“Patients receive comprehensive personalized cancer care from an integrated team of specialists, in an environment of discovery and learning,” says institute director Beverly Mitchell, MD. Patients also have the potential to contribute to the study of cancer by participating in clinical trials.
“The Stanford Cancer Institute builds on a long tradition of cancer discoveries and new treatments developed at Stanford,” says Mitchell. “It also taps Stanford’s significant intellectual resources in academic departments outside the School of Medicine. Simply stated, the institute’s goal is to mobilize the enormous potential of Stanford University in the fight against cancer.”