An Early Pharmacogenomics Application

Pharmacogenomics, the application of genotyping to patient therapy, holds great promise for solving a long-standing problem: differences in individual responses to drug treatments. The ultimate goal is to maximize drug efficacy while minimizing side effects. The time when a report with each person's genetic code will guide doctors in personalized medicine is still far away, but pharmacogenomics already is allowing physicians to make treatment decisions regarding human immunodeficiency virus type 1 (HIV-1). John Mellors Mutations in HIV accumulate and interact with each other and cause resistance to one drug and then others, one of the pivotal problems in treatment. In the past, the complexity of HIV drug-resistance testing and the limited information on its clinical utility made routine application impractical. Recent advances in automated assay technology have allowed rapid characterization of HIV in blood samples, so an increasing number of commercial laboratories now offer phenotypic and genotypic testing. In fact, this past May an international panel of expert HIV/AIDS physicians and scientists recommended HIV-1 drug-resistance testing as part of routine clinical practice.1 "[Since] the value of drug-resistance testing in predicting response to treatment was demonstrated a little over two years ago, people have flocked to the area," says John Mellors, professor of medicine at the University of Pittsburgh. Several studies conducted over the last two years suggest that resistance testing, providing a physician with data about the phenotype or genotype of a patient's HIV, can improve the odds of a treatment's success. "We are asking a lot of a diagnostic: to improve treatment outcome when there is variation in adherence to medications, drug exposure, and drug activities. I find it pretty impressive that any difference was shown in such a complicated system," says Mellors. Based on the results of these studies, expert panels agree that the test is definitely beneficial to some patients but not recommended for all. Physicians are most likely to use the test on patients who fail to suppress virus to undetectable levels as determined by HIV RNA assays during treatment. This treatment failure may indicate that resistance has developed, patients are not taking their medications, or drug levels are low because of poor absorption or rapid metabolism. Resistance testing helps determine if resistance to a treatment has developed and can suggest the next best treatment choice. Testing is also recommended to optimize treatment of pregnant women. Phenotype vs. Genotype Image: Courtesy of The National Institute of Allergy and Infectious Disease HIV budding out of a T cell. Phenotyping a virus involves growing the virus in centralized laboratories and determining which drugs it resists. The labor-intensive, expensive process limits the number of samples that can be tested. Recently, Virco Lab Inc., of Baltimore, and ViroLogic, of South San Francisco, have managed to commercialize and automate the process to increase throughput. For instance, Virco can phenotype 1,000 samples a week, about two orders of magnitude higher than previously possible. Genotyping, actually sequencing a part of the virus genome, is a comparatively simpler and less expensive process. Currently resistance testing has focused on two target proteins in HIV, protease and reverse transcriptase, which encompass about 1,900 nucleotides. The challenge in genotyping will be to get consistently high-quality sequencing and then to properly interpret the sequence, according to Mellors. Virco has created a database that relates genotype to phenotype after initially measuring the phenotypic resistance of HIV to a panel of commonly prescribed retroviral drugs. "We have a rich database because we have combined phenotypic data collected in large volumes, which no one has been able to obtain in this way, with the mutational genetic patterns of the viruses," says Brendan Larder, vice president of research and development at Virco. The database includes approximately 80,000 genotypes and phenotypes, and subsets of those are matched. The relational database can predict the resistance of a patient's virus, also called a virtual phenotype, based on the virus sequences. Software searches a patient's virus sample for any mutations that influence drug resistance. It can then construct patterns of the mutations and search the database for matching patterns. The matching genotype patterns are paired with phenotypes that are in the database. As a result, a virtual phenotype can be reported, for example, as an average 10-fold increase in resistance to drug X based on Y number of samples. Larder indicates that the database will predict with a high degree of confidence what the genotype of the virus actually means in terms of its sensitivity toward drugs. "It has actually been used in a number of clinical trials, and we have done testing on existing data sets where we have clinical trial samples from patients," he says. While Virco is providing resistance testing as a service, Visible Genetics, of Toronto, is in clinical trials with a three-component system for resistance testing, which they plan to market. A chemistry kit allows a physician to take a patient's virus sample through a PCR step to amplify certain regions of the virus, and then instrumentation sequences the genetic material. The sequence data is streamed to a software package, which compares the sequence against a consensus wild-type virus sequence. The output outlines the mutations in the patient sample point by point compared with the wild type. This data set is passed to a second software package, which interprets the mutation pattern to drug resistance. Visible Genetics' genotyping algorithm incorporates information gathered from phenotyping and clinical observations. The genotype report lists whether a virus is resistant, partially resistant, or not resistant to a particular drug. The entire system is going through the Food and Drug Administration regulatory process. Richard Daly "We are trying to decentralize this technology, so within a year there won't be half a dozen, but several hundred places in the [United States] that are doing [resistance testing]," says Richard Daly, president and CEO of Visible Genetics. The availability of Visible Genetics' sequencers, which cost less than $10,000, will make the technology accessible to more labs. "The key problem [has been] that not many doctors understood what the individual mutations meant," says Daly. The software package makes the interpretation easy. "[Virco's] relational database will take the guesswork out of the interpretation by relating the genotype to the phenotype, but interpretation of the clinical implications of the test result should be aided by experts in the field," adds Mellors. A Future Standard of Care The testing is already available, and cost should not be prohibitive. A number of Medicaid and private insurers have already started reimbursement for the test, about $400. "Everybody realizes this is going to become standard of care within the next year, so everybody is getting ready to use it and be reimbursed for it," says Daly. He adds that FDA approval of the Visible Genetics system will make resistance testing more available too. "In the future we want to predict clinical response of an individual patient through use of databases and neural networks," says Larder. Virco is using neural networks to derive patterns, fine-tune pattern recognition, and find new mutations. Mellors agrees: "The interactive database is an important paradigm for the future. The direction of HIV treatment will be toward predicting not only the phenotype of the virus but also the response to drugs through relational databases that incorporate many variables including resistance, CD4+ [T-cell] count, viral load, drug levels, and medication adherence." Nadia S. Halim can be contacted at nhalim@the-scientist.com. 1. M.S. Hirsch et al., "Antiretroviral drug resistance testing in adult HIV-1 infection," JAMA: The Journal of the American Medical Association, 283:2417-26, May 10, 2000.

An Early Pharmacogenomics Application

Interested in reading more?

Become a Member of