The Promise of Pharmacogenomics
December 6, 2004

By Nusrat Khaleeli and Dennis Fernandez

Why do some drug therapies work well with some patients but not with others? Why do some patients experience side effects and others don't? The answers may lie in our genes.

The field of pharmacogenomics combines medicine, pharmacology, and genomics to develop drug therapies that compensate for genetic differences that cause varied responses in patients.

Pharmacogenomics stems from a related field, pharmacogenetics, and the two terms are often used interchangeably. Pharmacogenetics is the decades-old study of differences in drug absorption, metabolism, elimination or response, and then examines a few candidate genes for variations underlying the observed phenotypes. In contrast, pharmacogenomics casts a wider net to capture complicated patterns of genetic variation and attempts to correlate these patterns to different drug response phenotypes. The challenge is to identify genetic differences that influence drug metabolism and response, and to correlate that data with drug efficacy and safety information. The goal is to weave all this information together into something that has enough predictive value to be used reliably.

Single nucleotide polymorphisms (SNPs pronounced "snips") are the most prevalent genetic variations in the human genome. They are single base pair differences that occur in 1% of the human population. The human SNP map shows 1.42 million differences, a majority of which occur in coding regions.

Pharmacogenomics is the study of how these sequence differences affect the ways in which people respond to drugs. Variations in the disease-causing genes, drug targets or the enzymes that metabolize drugs influence the drug's potency and efficacy. Also, genetic differences between patients may explain why some patients but not others suffer from harmful drug side effects.

The primary goal of pharmacogenomics is to reduce the time and cost of drug development. Choosing patient candidates for a clinical trial based on pharmacogenomic knowledge and the patients' genotype is hoped to eliminate sub-populations for whom drugs are predicted to be ineffective. This would justify smaller and fewer trials, likely generate more consistent trial results, and make it easier to gain FDA approval.

Another goal of pharmacogenomics is to identify patients who are likely to suffer drug related adverse events. A 1998 study of hospitalized patients published in the Journal of the American Medical Association reported that in 1994, there were more than 2.2 million adverse drug reactions and 100,000 drug-related deaths, making adverse drug reactions one of the leading causes of hospitalization and death in the United States. Moreover, the ability to pre-test patients may have prevented certain high profile drug withdrawals, including the former Warner-Lambert Rezulin (troglitazone) and Glaxo Wellcome's Lotronex (alosetron).

Pharmacogenomics can be used to identify how quickly a patient will metabolize a drug, and therefore, ensure appropriate dosing. Up to 30% patients do not respond optimally to certain drugs, this can often be addressed by merely changing the dose. If these problems were identified and remedied early in clinical trials, results would be more convincing and, therefore, approval would be faster and less costly.

Pharmacogenomics will allow the differentiation of a company's product from others in the marketplace (e.g., by identifying patients by a genotype who will respond to product X but not to product Y). One further benefit to patients is that pharmacogenomic knowledge will also allow identification of those patients in the population who will derive no clinical benefit from a prospective treatment. A look at data from clinical trials in 14 major drug categories reveals that this "non-responder" subset may be 20-75% of the general population. Additionally, pharmacogenomic knowledge from association studies (SNP to disease links) will allow for preventative screening and preventative treatment.

Currently, costs limit the widespread use of pharmacogenomics. For example, it costs approximately one dollar to identify one SNP in a patient sample. It is estimated that it will require the screening of 100,000 SNPs per patient to construct an accurate picture of a patient's response to a drug; this translates to 100,000 dollars per patient.

For this technology to become practicable, the cost must be reduced to a penny per SNP. Further, narrowing down a large number of genetic variations to a number that is amenable to application in a clinical trial would also prove useful. In this regard, computation methods to categorize and prioritize SNPs or haplotyping, the identification of closely associated polymorphisms that tend to occur in clusters, are being developed.

Other limitations in the progress of pharmacogenomics include tools used for collecting, archiving, organizing and interpreting the huge amount of data generated in a pharmacogenomics study so that data from diverse experiments can be compared. Also, drug dosage and treatment schedules need to be standardized in order to accurately compare patient data. Successful interpretation of data also requires comparison of enormous quantities of data such as the publicly available databases, Pharmacogenetics and Pharmacogenomics Knowledge Base (PharmGKB) and the SNP Consortium.

Drug patent holders in pharmaceutical industry have many incentives to use pharmacogenomic knowledge to develop genotyping diagnostic tests to be used with a drug. They have a vested interest in having shorter, less expensive clinical trials, identifying patients who are expected to have adverse drug reactions and those requiring tailored dosages of drug. However, the anticipated loss of sales revenue by identification of the "non-responders" serves as a strict disincentive for the development of genotyping diagnostic tests.

Nusrat Khaleeli and Dennis Fernandez are with Fernandez & Associates, LLP (www.iploft.com).

Go to the Biotechnology section of Larta Institute's Research Archives

Return to this week's issue of VOX