Science & Technology: Pharmaceuticals
Custom Treatments: This Drug's for You
Genetically tailored treatments could transform medicine
The 9-year-old boy was near death when he arrived at the Mayo Clinic last fall. He had leukemia--but that's not why he was so sick. The cause was his medicine. He had been given one of the so-called thiopurine drugs, which have transformed acute lymphoblastic leukemia from a virtual death sentence to a disease with an 80% survival rate. But 1 in 300 Caucasians possess two copies of an otherwise harmless genetic variation that alters the drugs' action in the body. As a result, the medicines' standard dose destroys vital bone marrow.
Scientists at the Mayo Medical School discovered this genetic variation in 1980. By 1991, researchers had perfected a simple test to spot it. Now, in one of the first glimpses of a potential revolution in medicine, doctors at top medical centers routinely test for the variation. When it's present, physicians know they must drastically drop the standard drug dose. Yet hundreds of children in the U.S. and around the world still aren't being tested before the drugs are given--often with tragic results.
The story of the 9-year-old boy has a happy ending: Under the care of Mayo Clinic doctors, his bone marrow slowly recovered. After three months in the hospital, he was able to go home.
His tale, however, reveals both the vast promise of genetically tailored treatments--and the formidable hurdles that lie in the path of a new era of personalized medicine. Treatments genetically tailored for certain segments of the population could, for instance, shrink the market for existing blockbusters, cutting into drug-company profits. And experts worry that even when such treatments are shown to work, they may take years to become widespread.HIT OR MISS. Still, the new approach will eventually transform medicine. Already, pharmaceutical and biotech companies (table) are investing millions of dollars in the relatively new science of cataloging and exploiting human genetic differences. Those variations, execs hope, will offer an answer to the industry's top challenge: getting better drugs on the market faster. Currently, only about 1 of every 10 drugs that enters clinical trials makes it to market. The rest fall by the wayside because they don't work, are too dangerous, or don't fill a market need. "Nothing will improve in the industry until that 1-in-10 number gets better," says Randal W. Scott, president and chief scientist at Incyte Pharmaceuticals Inc. in Palo Alto, Calif.
But if drug development is combined with genetic analysis--an idea dubbed pharmacogenomics--researchers should be able to identify subsets of people for whom the "failed" drug candidates actually work--and to weed out those who suffer the side effects. Indeed, Pascal Brandys, CEO of Genset in Paris, figures that drug companies could save $1 billion of an estimated $5 billion now spent on drug-development efforts that don't pan out. "Many interesting drugs that haven't passed trials could be rescued," he explains. The approach should speed drug development as well. Clinical trials can be far more efficient and cheaper, since only those people who respond to an experimental drug would be included in the studies. Any pharmaceutical company that does not explore the importance of genetic variation "is sticking its head in the sand," says B. Michael Silber, director of clinical genetics at Pfizer Inc.
The biological concept underlying this commercial gold rush and potential medical revolution is simple: Humans are not quite geneTically identical. The spiraling strands of DNA that spell oUt our genetic instructions contain some 3 billion individual molecules. For any two people, 99.9% of those letters are the same. But the 0.1% difference--about 3 million letters--is crucial. These variations, called polymorphisms, explain why people come in so many different sizes, shapes, colors, and abilities. And "this small amount of information holds clues to why people respond differently to medicine or are susceptible to disease," explains Dr. Elliott Sigal, vice-president for applied genomics at Bristol-Myers Squibb Co.
One example is the genetic variation that makes the standard dose of thiopurine leukemia drugs lethal. Another polymorphism, announced early in 1998 by Dutch and Canadian scientists, predicts whether or not heart-disease patients will benefit from Bristol-Myers' cholesterol-lowering drug pravastatin. And researchers at Myriad Genetics Inc. have found a variation that explains why salt raises blood pressure in some people but not in others.SMART BOMBS. Industry researchers expect to find many more such correlations. Even the best drugs work in only about 80% of patients--and many help as few as 20%. Making the connection between people's differing responses to drugs and specific genetic polymorphisms, therefore, should enable drugs to be used like laser-guided weapons instead of dumb bombs. That could dramatically reduce the problem of adverse drug reactions--a problem with an annual price of more than $5 billion--while boosting health.
The scientific challenge is finding these crucial polymorphisms. There are two fundamentally different approaches. At Variagenics Inc. in Cambridge, Mass., CEO Dr. Fred D. Ledley is betting on variations in biological pathways affected by drugs. "From the time you take a pill, it touches 30 to 40 different proteins before it leaves the body," explains Ledley. Variagenics' researchers have identified more than 6,000 genes involved in drug pathways. And in May, 1998, they began a clinical trial to investigate genetic differences in colon-cancer patients being given the potent drug 5-fluorouracil. "The drug works for some, is toxic for others, and fails in others," Ledley says. The trial is expected to show that the reason for the different responses lies in polymorphisms in the drug's biological pathway.
But scientists don't know all the proteins in all the pathways. Dr. Daniel Cohen, genomics chief at Genset, is taking a different tack--looking at all variations, not just those involved in drug pathways. By the end of the year, the company expects to have a map of some 60,000 polymorphisms spaced throughout the genetic code. And in the industry's first major pharmacogenomics deal, a $22.5 million collaboration with Abbott Laboratories, Genset researchers are looking for polymorphisms that can identify--and weed out--people who would suffer liver damage from Abbott's asthma drug zileuton. "With predictive tests, we can reduce side effects by 75%," says Cohen. Critics of this strategy say Genset might turn up statistical artifacts or miss many predictive variations because it isn't looking at enough polymorphisms.
Other companies are focusing on a different question. Instead of trying to find genetic variations that predict responses to drugs, they are searching for those that foretell disease itself. At Millennium Pharmaceuticals' Predictive Medicine Division, researchers are comparing the genes turned on in various types of cancer cells in order to spot those that can predict the disease's aggressiveness. "We are going to change the diagnosis of cancer--and change the practice of medicine," proclaims division President Kenneth Conway. Clearly, being able to tell if a patient's prostate cancer is slow- or fast-growing has profound implications for treatment. Already, Millennium has found a gene in melanoma cells that acts like a crystal ball. When the gene is there, the cancer doesn't metastasize. But when it's absent, the melanoma kills.LONG WAIT. Eventually, it should be possible to catalog each person's thousands of genetic polymorphisms into a sort of "genetic bar code." The bar code will reveal which drugs are safe and effective, as well as identifying susceptibilities to disease. But don't expect that to happen soon. At some point, there will be a genetic test to identify the best treatment for high blood pressure, predicts hypertension expert Dr. Michael Alderman of Albert Einstein School of Medicine--but he doesn't expect it in his lifetime. "The dream was to have a test that predicts 100% of side effects," concedes Genset's Cohen. "This will never happen, because there are too many genes." And if people get hurt, drugmakers may face thorny liability issues.
There are other business concerns as well. If polymorphic analysis shows that a blockbuster works in only 40% of patients, the drug's market will shrink dramatically. "The downside for industry is that not everyone will be a candidate for a particular drug," explains former Bristol-Myers research-and-development chief Leon E. Rosenberg.
All this means that pharmacogenomics is a technology at an awkward adolescent stage. It's enormously promising but far enough from delivering that investors could lose patience before the approach lives up to its potential. "There is a lot of hype," admits Ledley. "But it is also one of those great technologies that is going to make a difference." The question is, when?By John Carey in Princeton, N.J., and WashingtonReturn to top