HEALTH-SYSTEM EDITION
CLINICAL PRACTICE

Progress continues on selecting best drugs for patients

The science of pharmacogenomics is becoming an increasingly useful tool for predicting individual patient response to drug therapy, according to William E. Evans, Pharm.D., FCP, deputy director of St Jude Children's Research Hospital.

"Even when multiple polymorphisms are involved, it is possible to find a panel of diagnostics that can predict response to medication," Evans said. He was the keynote speaker at the annual meeting of the American College of Clinical Pharmacology, held recently in San Francisco. In a paper published in The Lancet in 2000, for instance, researchers found that six polymorphisms in four genes could predict the response of schizophrenic patients to clozapine with 75% accuracy.

Evans' own research on children with leukemia has revealed that gene expression profiling can be a powerful diagnostic and treatment tool. In a study of 350 children with leukemia, published in Cancer Cell in March of 2002, his team of researchers found that overexpression or down-regulation of certain genes predicted with 97% accuracy which patients would continue in remission or relapse. "This type of diagnostic tool has great potential in helping us select cancer therapy—and identifying patients who need more intensive therapy," he said.

Evans' research group has also analyzed how genes change in response to cancer treatments. In the study of 350 children, participants were randomized to one of four common medications for leukemia, and changes in 13,000 human genes were monitored before and after treatment. The group found that there were unique changes in 124 of the genes they analyzed.

"Our current technology allows us to look inside medications and see what they're doing," Evans told the audience. "What we now want to know is whether these changes in gene expression forecast response." The answer is they do. In a study not yet published, Evans and his colleagues looked at 60 children with leukemia randomized to methotrexate and found that changes in expression in 87 genes right after treatment predicted who would stay in remission or relapse, even years later. "These are small numbers, but the data are very provocative," he said. "It reveals to us which genes are involved in polymorphism."

Such knowledge is especially useful because the empirical approach to prescribing drugs often doesn't work. "By giving the same medication at the same dose to different patients, we don't take into account individual differences," Evans said. The cost is enormous. Adverse drug reactions are the fourth-leading cause of hospitalization, and the fifth-leading cause of mortality in the United States.

Recent scientific literature reveals that genetic polymorphisms can affect a wide array of proteins that direct drug response. These proteins include 25 drug-metabolizing enzymes, seven drug transporters, and 25 drug targets. Single nucleotide polymorphisms, or SNPs, are the most common type of sequence variant found in the human genome. There is one SNP for every 400 to 1,500 base pairs in the genome. Half of SNPs have the "wrong" coding and half are silent, but both can affect protein function, Evans said.

Finding polymorphisms or SNPs that affect drug reaction used to begin with observation of a certain toxicity in patients. Scientists would then search for the genotype that caused this reaction. Now with the discovery of the human genome, however, it is possible to compare genotypes in different patients to see whether they are linked to a certain phenotype or drug reaction. "This approach takes less time and helps us investigate aberrant drug response more efficiently," Evans observed.

Classification of the human genome has also helped scientists understand the polygenic nature of drug effects, Evans added. Drugs can affect disease pathogenesis genes and host susceptibility genes as well as drug metabolism and transport and drug receptors and targets. These mechanisms together predict drug efficacy and toxicity.

Eventually, pharmacogenomic data will give a precise picture of an individual's reaction to a wide variety of medications throughout his lifetime. Patient genotypes will be stored in a secure on-line database that would be accessible to health practitioners only upon the patient's authorization.

Such future advances will improve individual drug therapeutics tremendously, Evans predicted. "Prescribing medications will become more scientific and less empirical," he said. "Pharmacogenomics could thus help us select medications appropriately, or even distinguish the best doses of medications for individual patients," he concluded.

Barbara Boughton

Barbara Boughton. Progress continues on selecting best drugs for patients.

Drug Topics

2002;20:HSE8.