Drugs Get Smart
On the day Dr. John P. Kane began his residency at the University of California at San Francisco Medical Center in 1959, he got a phone call. His father, Paul -- a retired Army general who had survived combat in both World Wars, never smoked, and had no apparent health risks -- had just died of a heart attack at 66. The news propelled Kane into the field of cardiology. Nearly a half-century later, his discoveries may alter the treatment of the No. 1 killer in the industrialized world: heart disease.
Kane spent much of his first two decades as a doctor doing research that helped link high cholesterol to heart attacks. In 1985, he began collecting samples of DNA, believing that genes must play a role as well. Now 72 and still logging 11-hour days, seven days a week, Kane has scrutinized 10,000 genes, nearly half the human genome. So far he has identified 20 variations that seem to mark the people who carry them for heart attacks.
Working with his longtime lab partner and wife, Dr. Mary J. Malloy, at the UCSF Cardiovascular Research Institute, Kane has reached some surprising conclusions: At least half of the genetic variations linked to heart attacks bear no obvious relation to cholesterol levels, blood pressure, or any of the other usual suspects in heart disease. Instead, they strongly suggest mechanisms such as inflammation, which can be caused by an infection or by a haywire immune system. That implies that there may be several forms of heart disease, just as there are multiple forms of breast and lung cancers.
If Kane is right, doctors might be able to screen patients' DNA and predict not only whether they are likely to have a heart attack but which type of heart disease to expect and which drug or procedure might work. Instead of prescribing a cholesterol-lowering pill such as Lipitor reflexively, they might choose an anti-inflammatory, for example. Other experts share Kane's conviction, including executives at Celera Diagnostics, an Alameda (Calif.) outfit co-founded by genomics pioneer J. Craig Venter. Celera hopes to have a genetic test for heart disease on the market in a few years, based largely on Kane's research. "We're embarking onto the open sea," says Kane, "and we're discovering things that nobody knew about."
Historians have called the late 19th century "the great flowering" of medicine. Thanks to advances in microscopy and a deeper understanding of germs and human physiology, scientists were able to identify the cause of one infectious disease after another. Today, despite headline-grabbing fiascos like Vioxx, medicine is on the verge of a second flowering. This time the findings may improve the treatment of everything from heart disease to cancer, depression, Alzheimer's, and autoimmune disorders. Doctors are calling it the age of "personalized medicine" as they close in on genetic variations that hint at why one person responds to a drug and another doesn't, why some people are prone to strokes or heart disease, why one person's cancer is more aggressive than another's.
The shift to drugs tailored to a specific genetic profile will bring massive changes to a pharmaceutical industry long wedded to treatments that can be taken by millions upon millions. Most prescription medicines are now effective for fewer than half of the people who take them -- and the side effects can be worse than the illnesses. "From a strategic standpoint, of meeting the needs of our customers, the current blockbuster model doesn't work," concedes Sidney Taurel, chairman and CEO of Eli Lilly & Co. (LLY )
Drugmakers suffer mightily when a blockbuster fails -- as demonstrated by Merck & Co. (MRK ) On Aug. 19 a Texas jury returned a $253 million verdict against Merck for the 2001 death of a patient who had been taking the painkiller Vioxx. And that's only the first of thousands of Vioxx lawsuits, which could leave the company reeling.
Merck pulled Vioxx from the market last fall after it was linked to heart attacks and strokes. Although scientists suspect that Vioxx is safe for the vast majority of patients, the drug can dramatically increase the risk of blood clots for a few. But there is no test to reveal who can take the pill safely and who can't.
Well before the Vioxx troubles, Big Pharma and biotech were beginning to explore personalized medicines. Those efforts should pick up momentum, says Robert Goldberg, director of the Center for Medical Progress, a public-policy think tank. "If drug companies do not go the route of understanding and responding to the tremendous genetic variations in how we react to medicine," he says, "you'll see more litigation, not less."
Getting to that understanding is still a challenge because illness is never a simple phenomenon. Researchers have shown that there are many varieties of breast cancer and leukemia, and the same is surely true of other cancers, coronary disease, and neurodegenerative ailments -- with each variety afflicting different subsets of people based on genetic vulnerability. Doctors are seeking to identify differences among groups through DNA tests, hoping that treatments can be tailored as well. Ultimately, each of us should receive the drugs that work best, based on the subset dictated by our genes.
Personalized medicine is set for a natural progression. First, genetic screening will yield more accurate diagnoses. Next, doctors will prescribe medications that will help, not harm. Finally, there will be an explosion of treatments for specific populations. That doesn't mean doctors will handcraft pills for each patient, but there will be more effective treatments, less waste -- and much less Vioxx-style collateral damage.
A milestone was the completion in 2003 of the mapping of the entire human genome -- the 25,000 genes coiled in the chromosomes of every cell. Since then, databases of DNA have been assembled with samples from tens of thousands of people. By comparing genes, researchers can link variations and mutations to all sorts of illnesses. Kane, for instance, has a database of specimens from 20,000 people. He starts with samples from people who have had heart attacks or shown signs of heart disease at early ages. He then compares their genes with those of exceptionally healthy people, such as 935 elderly athletes at last year's Huntsman Senior Games in Utah.
Certain farsighted drugmakers have also mounted deep-pocketed efforts to develop tailored medicine. Abbott Laboratories (ABT ), Johnson & Johnson, (JNJ ) and Switzerland's Roche (RHHVF ) Holding are learning how to examine tissue samples or blood from a patient and spot relevant genetic and biochemical variations, often referred to as biomarkers. Pfizer Inc. (PFE ) and Bristol-Myers Squibb Co. (BMY ), among others, are using these tools to fashion pharmaceuticals for targeted groups.
There have been a few early triumphs. One of Genentech Inc.'s (DNA ) most lucrative drugs is Herceptin, which has been used to treat 175,000 metastatic breast cancer patients who share a particular genetic variation. Doctors have also grasped how genes regulate liver enzymes that break down drugs, thus helping to avoid adverse side effects in some patients.
Such success stories are still rare. Personalized medicine resembles a weather forecast, based on probabilities and interpretations of data. The presence of a single gene or combination of genes makes it likely that a person will develop or avoid a particular disease -- but the outcome is almost never certain. DNA, RNA, proteins, and chemical signals among cells all play a role in diseases, as do higher-level structures such as the human immune system. And most of these complex interactions are still only dimly understood. Kane predicts that by the time he has reviewed the entire genome, he'll have fingered 40 to 100 different biomarkers in heart disease alone. Says Roche CEO Franz B. Humer: "In 20 years' time, when people look back, they will consider us now as if we were in the Dark Ages."
Not only is the field still immature but it is also beset with concerns about public policy and privacy. Experts fear individuals may be denied life insurance, health insurance, or even a job if they're known to harbor genes for a debilitating illness. Also, there is a debate about whether personalized medicine will reduce or increase health-care spending. Better diagnostics would let doctors intervene more quickly, avoiding some costly procedures. But hospitals may also order more and more tests indiscriminately to cover themselves against possible lawsuits for not detecting diseases before it's too late. And those tests can be expensive: A test for abnormalities on the BRCA1/2 genes implicated in certain breast and ovarian cancers costs $3,000 apiece.
Advocates counter that if the money saves lives, it will be well spent. But the point is far from resolved -- and it's just one of many debates about costs, benefits, and timetables. Amid all the contention, doctors, drug-company execs, and patients continue to march forward, drawing on the resources of the world's top academic institutions, hospitals, and government agencies.
While cancer treatments are in the vanguard, heart disease looks as if it might be next. Scientific progress in this area may not help Kane solve the mystery of his own father's unexpected death. But Kane and Malloy have three grown children. If there are markers lurking in their DNA, his work could suggest new courses of action. It also could help millions of others at risk of dying before their time. In this fashion, one research effort at a time, the revolution in personalized medicine is moving forward. Here are some dispatches from the front lines:
The Chief Executive
A trained mathematician, Heino von Prondzynski, the 55-year-old CEO of Roche's diagnostics unit, seems more like a schoolmaster than a crusader. But get him going on the power of diagnostics to transform medicine, and the usually taciturn German exudes an almost missionary zeal. Taking a small silicon chip from his breast pocket, von Prondzynski claims the device -- no bigger than a postage stamp -- and others like it will lay the foundation for a new era of health treatment, enabling earlier and more accurate diagnosis and more customized care.
Launched in the U.S. in January, Roche's device -- called the AmpliChip -- is the first genetic test approved by the Food & Drug Administration to identify responses to a wide array of medications. It does this by detecting genetic variations that control two liver enzymes responsible for how patients metabolize up to 25% of prescription drugs. For some people, known as ultra-rapid metabolizers, there are too many such enzymes in the bloodstream. That causes the body to clear the drugs out of the system so fast they get no benefit. About 3% to 10% of people lack the enzymes altogether and cannot break down medication, so the drugs build up too quickly in the bloodstream, making even a standard dose toxic. Devices like the AmpliChip, says von Prondzynski, "will address the two biggest challenges facing the health-care system today: cost and safety."
That could save society enormous sums. Each year 2.2 million Americans suffer adverse reactions to prescription drugs. Of them, more than 100,000 die, making side effects a leading cause of death. Often blockbusters like Vioxx are found to be the culprits. The cost of treating adverse drug reactions in the U.S. alone totals $4 billion annually, according to a Journal of the American Medical Association study. All told, Roche estimates that the U.S. could shave $21 billion from its health-care bill by 2020 if doctors used tests such as AmpliChip to select the right medicine at the right dose, based on a patient's genetic makeup.
The payoff for Roche could be huge. Swiss consultancy Jain PharmaBiotech estimates that Roche's sales of molecular diagnostic kits and devices could grow from $6.5 billion a year now to $12 billion by 2010. Today such gene-based tests are a small part of Roche's diagnostics sales, which made up more than 25% of the company's $23 billion sales for 2004. But gene tests are growing the fastest by far.
Roche got a head start in molecular diagnostics. Against the advice of its own scientific advisers, the company laid out $300 million in 1991 for the rights to a tool called polymerase chain reaction from Cetus Corp. (CHIR ) in Emeryville, Calif. Invented in the early 1980s by Cetus scientist Kary B. Mullis, PCR was then a largely untested technology that allowed traces of genetic material to be amplified for analysis. Few at the time thought it had much commercial potential. But the doubters were wrong. In 1993, Mullis was awarded the Nobel Prize in chemistry, and Roche has gone on to use PCR as the foundation of cutting-edge diagnostic tests to detect everything from hepatitis and HIV to cancer.
Von Prondzynski believes a series of new cancer diagnostics will further improve treatments. In 2006 the company plans to launch a DNA chip to pick up the p53 gene. In healthy people, p53 plays a role in suppressing tumors, and many people who harbor variations of it wind up with cancer. Roche's chip not only catches the variations but also is designed to show how fast or slowly a tumor is likely to grow, enabling doctors to decide how aggressively the cancer needs to be treated. By 2008, Roche hopes to have a test available to predict the likelihood of relapse in patients with colorectal cancer. The following year it plans to field screening tests for early signs of breast, prostate, and colorectal cancers. "In 10 years' time," says von Prondzynski, "I could envision having three DNA chips for breast cancer alone: One for early detection, one for drug selection, and another for therapy monitoring."
Anneke Westra has experienced the power of Roche's diagnostic technology. Eleven years ago, at 30, she was a promising scientist with a PhD in biotechnology who already had a patent to her name. Despite a lifelong struggle with depression, she became an expert in biosensors, traveling the world to present scientific papers. That changed in September, 1994, when the Londoner was diagnosed with bipolar disorder.
It was the start of a decade-long debacle. Westra's treatment involved a dozen psychiatrists and 18 pharmaceuticals -- each drug, it seemed, with worse side effects than the one before. Her mental and physical health deteriorated. Frequently hospitalized and unable to work, she eventually tried to kill herself. "I lost 10 years of my life from the drugs I was given," she says.
Westra's case is extreme but not unique, and it explains why doctors and patients both are eager for personalized medicine. One in five people at some point suffer depression severe enough for medication. Many spend months trying a variety of drugs, at different dosages, before hitting on a prescription that works -- if they're lucky. For 25% of patients, the most common antidepressants -- selective serotonin reuptake inhibitors such as Prozac -- are ineffective. Millions more are wracked by side effects. Until Roche launched its AmpliChip, there was no reliable means of monitoring the family of enzymes found mainly in the liver, known as CYP450, that dictate how our bodies break down medication. That left trial and error -- and a trail of tears.
After her diagnosis, Westra spent the next few years in and out of the hospital, but no matter what drugs doctors prescribed, terrible side effects set in almost immediately. Promazine knocked her unconscious for five days, while within four days of taking Eli Lilly's Zyprexa she was hearing voices in a drug-induced state of psychosis. "No one believed me when I told them I was being poisoned by the drugs," she says.
Finally she found a psychiatrist who prescribed a mild tranquilizer, and she improved. Rested and thinking clearly for the first time in years, Westra deduced that her troubles had something to do with the way her body metabolized medication. In just three hours of exploring medical Web sites, "I worked out what it might be," she recalls. "I was crying with relief."
Her research led her to Dr. Katherine J. Aitchison at the Institute of Psychiatry at King's College London, who confirmed Westra's suspicions. Using the AmpliChip test, Aitchison found Westra had too little of two drug-metabolizing enzymes in her liver, making her acutely sensitive to numerous medications.
Millions of people may be like Westra, according to Dr. José de León, an associate professor of psychiatry at the University of Kentucky and medical director of the mental health research center at Eastern State Hospital in Lexington. In a study published in the January issue of The Journal of Clinical Psychiatry, de León found that patients without the CYP2D6 enzyme -- one that Westra had too little of -- were three to six times more likely to experience severe side effects when placed on the antipsychotic risperidone. Now he's conducting a study on 4,000 psychiatric patients at three state hospitals in Kentucky to determine if testing patients for these genetic variations before prescribing is cost-effective.
Westra, 41, finally has her life back. She takes minuscule amounts of three psychiatric drugs, and uses homeopathic remedies. "I used to fall into a a suicidal depression that would last nine months. Now the depression lifts in two weeks. That never happened before." In addition to advising mental health practitioners, she's the co-founder of Britain's No Force Campaign, which lobbies against forced medication of the mentally ill. And while she has testified calmly before Parliament on the topic, she's still angry at the health-care system that let her down. "I have been through a nightmare, only to discover it never should have happened," she says. "The information was out there."
Four years ago top FDA official Dr. Janet Woodcock met with drug industry representatives to discuss the promise of personalized medicine. "It was a 'call to light' kind of meeting," recalls Woodcock, now deputy commissioner. But the pharmaceutical folks weren't ready to sign on. "People stood up and said: 'We are terrified."'
Their reaction wasn't surprising. Using drugs in targeted ways raises scientific, financial, and regulatory issues. For companies, it can mean smaller markets for drugs and the risk that they will not win FDA approvals unless they can prove that the target is proper, as well as the drug.
Faced with such concerns, Woodcock is a passionate advocate: She has a vision of the FDA working with scientists and companies to bring better drugs to market faster, and to use existing drugs only for those patients who will benefit from them. The vision is fueled by fresh evidence that patients' varying response to drugs are hardwired in their DNA. Recently, researchers have found variations in two genes that predict how much blood thinning will occur with a given dose of warfarin. Testing for those variations, then adjusting the starting dose accordingly, could be a huge medical advance -- preventing cases of bleeding from too much of the drug or clots from too little. "It isn't that there are bad drugs and good drugs," she says. "It's that some drugs run into bad problems with a small subset of people."
In addition, the medical community is becoming accustomed to the idea of pairing diagnostics and drugs, especially when introducing new treatments. For a cancer drug that has only a 10% response rate in the general population, "we can make it a 100% response if we can figure out who actually benefits," says Woodcock. "Then the other 90% don't have to be exposed to the drug, and everyone is happy." That should also mean fewer Vioxx-like debacles, and fewer crippling liability payouts.
There's also the hope that developing drugs and diagnostics together will simplify FDA approval. By focusing clinical trials more tightly on people who are likely to benefit, the trials could be smaller, they would take less time, and so would the approval process. And by weeding out potential victims of side effects at the trial stage, companies would stand a better chance of getting their products greenlighted.
That's the rosy scenario, anyway. The darker view is that companies will discover that it's expensive -- and scientifically challenging -- to identify biomarkers that predict a patient's response and then to develop the necessary diagnostic tests. Once the tests are available, skeptics contend, doctors still may not use them before prescribing. And because the FDA has no authority over the practice of medicine, it can't insist.
The agency can issue stern warnings, but experience is not reassuring. When the diabetes drug Rezulin started causing liver failure in some patients, the FDA ordered doctors to monitor patients' liver enzymes to spot early warning signs. Yet few did. Patients continued to die, and the agency had to withdraw the drug from the market.
What's more, as companies probe the genetic impact of powerful new drugs, they may be in for unpleasant surprises. What if, for instance, a drug that seems to cause no side effects happens to turn on a gene linked to cancer or to liver repair? Would the FDA then demand years of testing to show that the drug doesn't cause cancer or harm the liver?
Since her meeting with pharmaceutical executives four years ago, Woodcock and her team at the FDA have worked hard to assure drugmakers that they won't be penalized for doing these so-called pharmacogenetic studies. The agency has ruled that disclosing genetic data is voluntary, not mandatory. And it has created a "safe harbor" for companies to discuss findings with FDA experts without triggering regulatory issues or delays.
"I give Janet a lot of credit for continuing to pound on this," says Mikhail L. Gishizky, chief scientific officer at drug researcher Entelos Inc. in Foster City, Calif. And her efforts are slowly having an impact. "We are seeing companies moving this technology into the clinic in the next year or so," says Woodcock. So tomorrow's drugs should come with a far better understanding of how they should be used and in which groups of people. "It is a big scientific endeavor," she says. "But the payoff is so tremendous that it must be accomplished."
When doctors, drugmakers, patients, and regulators all push in the same direction, few barriers are likely to stand in the way.
By Kerry Capell and Michael Arndt, with John Carey