Kate Robbins was in her mid-forties when she began writing a journal for her children. Knowing she had only a few months before her metastatic lung cancer would take her life, she wanted to leave her young son and daughter words of comfort. Those few months of survival have since turned into years, however, and Robbins’s journal has been stashed in a closet. The reason? An experimental therapy that unexpectedly melted her tumors.
Robbins’s unusual response to treatment provided not only joy to her and her family, but also delivered a startling insight to oncologist Daniel Haber. In the fall of 2003, Haber, who directs the Massachusetts General Hospital Cancer Center, was midway through a bowl of Cheerios and the Boston Globe when he came across a story about Robbins. He learned that she was among a small sample of patients whose non-small-cell lung cancer had successfully responded to the drug gefitinib, marketed as Iressa. Her miraculous reprieve could only be explained by genetics, he thought. He immediately called her doctor.
Haber’s collaboration with that doctor, Thomas Lynch, then an HMS professor of medicine at Mass General, led to the discovery of a gene-linked wrinkle to the drug’s effectiveness. Their research found that Iressa could bind well to a protein known as epidermal growth factor receptor, or EGFR, but that the tightness of this drug–protein bond increased tenfold if the gene that produced the protein had mutated during the development of the tumor, as in Robbins’s case. That work—along with simultaneous research by a team of scientists from the Dana–Farber Cancer Institute and Brigham and Women’s Hospital, led by Matthew Meyerson ’89, an HMS professor of pathology—gave rise to a genomics test that could predict, simply by analyzing lung-tumor tissue DNA, which patients would benefit from the drug.
Today, DNA from lung tumors can be sequenced to determine the genetic mutations that drive a tumor’s growth so that targeted, more effective treatments can be selected. Tarceva, a drug with action similar to that of Iressa, has now succeeded Iressa as a first-line treatment for non-small-cell lung cancer; second-line treatments have since been added to the clinical arsenal as well.
Researchers have dubbed this kind of precision diagnostic and treatment personalized medicine, an umbrella term for a range of approaches that mine genetic data to bring greater resolution to the often murky factors that contribute to disease—and maybe (just maybe!) identify precise, targeted therapies.
Where personalized medicine ends up may well depend on how doctors and scientists define it during this time of rapid discovery and change. And medicine as we know it is changing fast. The first sequencing of a human genome cost nearly $3 billion and took more than a decade to complete. Today the expense runs in the thousands, and it can be done in a matter of months—and even that cost and that timeframe continue to shrink.
Yet amid such change, people define personalized medicine differently, the focus zooming in and out depending on where along the laboratory-to-clinic spectrum the lens is pointed. Personalized medicine could just as easily be called medicine—and the need to link a modifier may spring solely from a wish for something to hang onto during the wild ride.
Some of the confusion about this relatively new field may spring not from the realm of medicine, but from the publicity surrounding companies that will crack your distinctive code for a fee. These direct-to-consumer businesses, such as 23andMe and Navigenics, “offer personalized genotyping without a physician in the loop,” says Mark Boguski, an HMS associate professor in the Center for Biomedical Informatics, leaving people pondering what, exactly, their long sequence of letters really means and sometimes wondering whether they might rather not have known.
Moreover, some of the confusion stems from the diverse ways in which DNA influences health. Well known are the DNA variations we inherit from our parents. Some are rare and confer a deterministic influence on the so-called genetic diseases, such as cystic fibrosis. Others involve many different genetic variants that interact with the environment, resulting in such common disorders as diabetes and Crohn’s disease. And still other mutations are not inherited but simply arise in otherwise normal cells, leading to cancer. Each of these disease categories involves DNA, but the manifestations and the needed therapeutic approaches differ drastically.
In the right settings, however, genetic testing has proved lifesaving. “In these cases,” says Raju Kucherlapati, the Paul C. Cabot Professor of Genetics at HMS, “it’s no longer the promise of personalized medicine. It’s actually happening today.”
Hitting the Target
Over the past decade, cancer has provided a heartening arena for personalized medicine’s promise, demonstrating the clearest success stories thus far. In fact, the Iressa finding was not the first time researchers demonstrated that therapies could target DNA defects arising in tumors.
Several years earlier, for example, the drug Herceptin was developed based on understanding derived from cancer gene mutations. Herceptin targets breast cancers that over-express the HER2 gene, which appears in about 30 percent of those with breast cancer. This drug increases survival rates for those testing positive for HER2 by 33 percent. Another drug, Gleevec, inhibits a malfunctioning enzyme that, based on genetic studies, was discovered to play a role in chronic myeloid leukemia—a diagnosis that essentially amounted to a death sentence before the drug began to extend lives, sometimes dramatically. Gleevec has since been approved for gastrointestinal stromal cancers as well.
Similarly, the Iressa finding represented the first tailored therapy for lung cancers that were otherwise fatal. “Iressa has proven effective in a small fraction of patients,” says Kucherlapati, “but for that fraction, the impact has been dramatic.” The value of such drugs, Kucherlapati adds, is not just that they can be targeted to the right patients—it’s that they target a mutation that is limited to the tumor. Because the mutation doesn’t occur in normal cells, the drug has fewer side effects than traditional chemotherapy.
As the field continues to mature, Mass General’s Haber predicts, therapies will be targeted at underlying gene abnormalities, regardless of the tissue from which the cancer originated. “You can no longer do cutting-edge oncology without genetic tests,” Haber says. “We’re finally able to bring to the clinic decades of breakthroughs in understanding the implications of our genes. With advances in both genetics and chemistry, we can now finally see these advances coming together with a very real impact for patients. It’s the beginning of a true revolution in the way we treat cancer.”
Needles in Haystacks
Slices of the truth of Haber’s assessment are evident today. “New information is coming at us fast and furiously,” says Kucherlapati, and scientists and technicians are scrambling to validate it and put it into practice. Today, for example, Mass General’s cancer center offers molecular fingerprinting—DNA analysis that reveals specific cancer-causing mutations in an individual’s tumor cells—for 130 mutations.
Even though few of these fingerprints have corresponding “smart drugs” yet, molecular fingerprints are changing the way doctors diagnose and treat cancer. In fact, the HER2 mutation, Herceptin’s target, has been identified in some types of lung cancer.
Finding such lowest-common denominators of cancer will take time, and a project now under way, The Cancer Genome Atlas, aims to systematize that process. A consortium of researchers—including Harvard affiliates Matthew Meyerson, Lynda Chin, Raju Kucherlapati, and Stacey Gabriel—is sequencing DNA from samples of more than 20 kinds of cancer to look for critical genetic abnormalities. As the atlas expands, Chin, an HMS dermatology professor at the Dana–Farber Cancer Institute and the Broad Institute of Harvard and MIT, expects it to drive development of new biomarkers, new ways of categorizing cancer, and new treatments.
“This project has already changed the way we look at cancer genetics,” says Chin. “It’s the first step toward a future of personalized medicine, when we no longer lump cancers together by a particular organ site, but treat each as a genetically distinct disease.”
The Heart of the Matter
While personalized medicine will no doubt continue to chip away at a number of mutations that arise in cancer cells, there are many other conditions for which genetic testing is used to evaluate the inherited risk of disease. These rare conditions—including Huntington’s disease, BRCA1- and BRCA2-mutated breast cancers, cystic fibrosis, muscular dystrophy, and hypertrophic cardiomyopathy—tend to be caused by a single, inherited genetic component—one mutation in one gene resulting in one disease.
In hypertrophic cardiomyopathy, for example, a genetic defect causes an abnormal thickening of the walls of the heart. This problem can lead to pumping problems and can cause heart failure or sudden death, accounting for the heartbreak of young athletes collapsing unexpectedly on playing fields. In a series of studies spanning the past two decades, HMS professors Christine Seidman and Jonathan Seidman have identified genetic roots for this disorder. Today, Harvard’s Laboratory for Molecular Medicine, for which the Seidmans serve as advisors, offers tests for 11 genetic mutations associated with the disorder. Treatment options include the lifesaving implantation of a cardioverter-defibrillator, a device similar to a pacemaker.
Since this heart disorder is inherited, genetic testing is recommended not only for people with symptoms of the disease, but also for the parents, siblings, and children of anyone diagnosed with it. In one case, Heidi Rehm, an HMS associate professor of pathology at Brigham and Women’s Hospital and the Laboratory of Molecular Medicine, found that a young patient carried a harmful cardiomyopathy gene. When she discovered that the patient’s father had the same mutation, she asked that all his children be tested as well.
The undertaking was greater than anyone could have imagined. This particular father had been a sperm donor, and he had 22 biological offspring in addition to the two children he and his wife were raising. “We said, ‘Oh my God, this is unbelievable! We have to test all of them,’ ” says Kucherlapati. Of the offspring the researchers tracked down, nine had the mutation. One had died of heart failure as a toddler, and two were showing symptoms of the disease in their teens. One of those teens has since received a cardioverter-defibrillator.
Kucherlapati believes this case illustrates not just the complexity and power of predictive genomics, but also the uniqueness of each patient. Different mutations call for different interventions. “In one subset of patients, you can implant devices to prevent sudden death,” says Kucherlapati, “while another subset has an enzyme deficiency that you can treat with drugs.”
Beyond the Magic Bullet
Despite the growing number of success stories in treating certain cancers and in predicting rare genetic conditions, personalized medicine remains a work in progress for many of the most familiar diseases.
“I’m skeptical about the promise of personalized medicine when it comes to common and complex diseases,” says David Altshuler ’90, HMS professor of genetics and medicine at Mass General and the Broad Institute. “In fact, many diseases are common because they have so many triggers, whether genetic, environmental, or lifestyle factors. No single gene explains diabetes, for example. Instead, scientists have identified dozens of genes associated with the disease. No amount of DNA sequencing will turn these complex conditions into single-gene disorders like Huntington’s disease.”
Even as scientists pinpoint the genetic risk factors for diabetes, Altshuler adds, it isn’t likely that testing results alone would change their treatment recommendations. Regardless of degree of risk, the best bet is to maintain a healthy weight and to eat well. In this instance, Altshuler argues, the clinical value of knowing genetic risk factors is debatable.
For Altshuler, the real value of human genetic analysis lies not in diagnostic testing, but in understanding disease. “What limits our progress in many diseases is that we don’t know their biological causes,” he says. “We need to figure out those causes, then devise interventions that will work for many people because they target the origins of disease.”
Along the way, Altshuler believes, some genetic tests will turn out to guide precision diagnoses and treatments, and they should be employed in the clinic as they are proven effective. Yet he points out a potential snag. “If people end up conflating the hype around personalized medicine with the science of understanding disease and developing better interventions,” he says, “the entire enterprise could lose credibility.”
Although Sir Francis Galton, a nineteenth-century scientist and proto-geneticist, was eclipsed by his more glamorous half-cousin, Charles Darwin, his legacy is notable nonetheless. A forerunner to the biometrics movement and the inventor of regression analysis, Galton studied height variations in populations, correlating parents and children. While the followers of Mendelian genetics insisted that one gene carried one trait, Galton’s adherents saw more of a bell-shaped distribution. The two camps battled it out until Ronald Fisher, widely credited as the founder of modern statistical science, pointed out that the answer was a bit of both: a Reese’s Peanut Butter Cup rather than a collision of peanut butter and chocolate. While we inherit genes following Mendel’s laws, manifesting a particular trait is hardly a binary equation. A six-foot father and a five-foot mother don’t produce children with the exact height of either. Rather, the heights of their offspring range between the two endpoints, even though each child has inherited either the “tall” or the “short” gene.
In many respects, the Mendel–Galton debate continues today between those who hunt for single genes to unlock medicine’s mysteries and those who claim that the answer lies in the messy intersection of genes, lifestyle, and environment—those external influences ranging from toxins to traffic-induced stress. Adherents to the latter approach focus on such initiatives as the International HapMap Project and the more recent 1000 Genomes Project. These projects—spearheaded by HMS researchers, including Altshuler and Mark Daly, an associate professor of medicine at Mass General—aim to register every common genetic variant that is likely to appear in a given population.
“The sheer value of all of this for understanding how DNA variation contributes to disease is mind boggling,” says Altshuler. “And that’s where the true power in genetics lies—not in predicting, but in understanding.” Altshuler believes that clues to the most common diseases will be traced only partly to patterns of genetic variants. Their explanation will never be simple, he says, given the confounding influences of environment and scientists’ ongoing discoveries about how genes orchestrate life.
An Uncertain Future
As data rush in from ever-larger population studies, advances tailored to individuals are struggling to catch up. By 2015, predicts Boguski of the HMS Center for Biomedical Informatics, doctors will be able to sequence and analyze an individual genome for about the cost of a routine imaging study today. And by 2020, he believes, doctors will be able to analyze a patient’s genome during a 15-minute office visit.
“I see a genotype as conceptually no different from a urinalysis or a blood count,” Boguski says. “It’s just another piece of laboratory data. In most cases, it’s relatively meaningless in isolation, but when we combine it with everything else we know about that patient—medical history, current condition, other laboratory data—that’s when we can make sense of it.”
Cardiologist Harlan Krumholz ’85 agrees. “With the advent of more genetic tests, we may be tempted to think we don’t even need to talk to patients anymore,” says Krumholz, the Harold H. Hines Jr. Professor of Medicine at the Yale School of Medicine. “We may think we don’t need a history. Just give me a set of symptoms and a bar code and I can say from a million miles away what a patient needs. But you can never do that in medicine.”
Instead, Krumholz envisions working with the patient to chart a treatment course that is in tune with the patient’s distinctive biology, medical history, and psychology; only patients, after all, can make the required lifestyle adjustments. But given environmental and other factors that cannot be controlled, Krumholz says, “I’m still going to be left walking my patient through a decision fraught with uncertainty. The precision of a genetic test may help me refine the information or be more confident about my recommendations, but uncertainty will remain. And that uncertainty, I believe, is where real personalized medicine emerges.”
The biggest challenge, according to Boguski, will be integrating all the information. “Will primary practitioners need to become geneticists to interpret this? The answer is no,” he says. “In the future, doctors will simply deliver genomic findings to patients as part of the larger diagnostic picture.” That is, software-based information systems will reduce the 3 billion bytes of genomic information into several bytes a doctor can use to guide clinical decisions. “The challenge for specialists will be to build an infrastructure for those tools,” Boguski adds. “For frontline doctors, it’s just another medical test. They’ll learn how to contextualize it and interpret it in light of everything else they know about the patient.”
For all this to happen, Boguski envisions new approaches. He can imagine collaborations, for example, among clinical pathologists, genetic counselors, and molecular medicine experts, and even the emergence of new specialties, such as genetic psychiatry.
More and more patients will likely, out of sheer curiosity, opt to have their genomes sequenced by direct-to-consumer outfits, outside the realm of the clinic, without the support of clinically relevant tools. While many clinicians are wary of potential consequences, Boguski sees the direct-to-consumer opportunity as a “phenomenon of participatory medicine, where patients, or consumers, begin to take increasing responsibility for their health and wellness.”
The challenge will also lie in figuring out how to do it now. “Big changes, societal changes, are coming right away,” says Kucherlapati. Both CVS Pharmacy and Medco, a prescription-drug benefit provider, for example, will soon start offering genetic tests to patients based on relevance to the prescriptions their physicians write for them. And tens of millions of people will soon be offered genetic tests for clinical reasons, not just for entertainment. And that, says Kucherlapati, “is a lot of people.”
Elizabeth Dougherty, a science writer formerly on staff at HMS, is now a freelance writer and novelist living in central Massachusetts