Study Could Improve Treatment of HIV Patients

Doctors may now be able to prescribe more effective treatments for HIV-positive patients due to a new study from several Harvard researchers.

Assistant Professor of Biostatistics Gregory DiRienzo and colleagues recently identified specific regions in the HIV genome which allow the virus to resist certain drugs.

Using information from DiRienzo’s study, physicians can now distinguish HIV with genetic coding that is resistant to certain drugs and prescribe alternative medication.

Currently, there are three categories of commercially available drugs being produced to combat HIV infection at various places in the virus’ replication cycle: nucleoside reverse transcriptase inhibitors, non-nucleoside reverse transcriptase inhibitors and protease inhibitors.

While the first two inhibitors work to prevent the virus from infecting a host cell, the protease inhibitor class of drugs works to disable protease, an enzyme occurring naturally in every living organism that works to digest proteins.

HIV protease cuts the virus’ proteins and enzymes into smaller pieces which then go on to infect new cells.

By blocking HIV protease, amprenavir—the focus of DiRienzo’s study—and other drugs keep the virus from making copies that can infect cells. HIV protease inhibitors slow virus production at the end stage of virus replication.

DiRienzo said he worked to uncover the link between mutations in the protease-coding region of HIV, and their ability to resist protease-inhibitor antiretroviral drugs.

To do this, he took a data set provided by Virco Group—a Belgium-based biotech company—containing both sequences of the protease genome and measurements of the amount of amprenavir necessary to inhibit replication of HIV by 50 percent for 2,747 patients.

DiRienzo said in order to conduct the study he had to devise a new non-parametric method for analyzing data.

Traditional statistical methods for handling unordered data require that there are more observed data points than possible realizations of data, he said.

Normally, ordered data values may be parameterized, establishing a relationship between genotype and phenotype. However, according to DiRienzo, with the HIV drug resistance genotype, not only are amino acid values unordered, but there are almost an infinitessimal number of possible combinations of amino acids and thus an even higher number of possible genotype patterns. Thus, the only practical way of analyzing the data is through non-parametric methods.

Methods used to interpret data were designed specifically for the setting where the genotype of drug resistance provides a basis for predicting a phenotype of drug resistance, according to the published paper.

The analysis identified specific patterns of mutations that are associated with resistance to amprenavir.

“Patients with these patterns should thus be taking drugs other than amprenavir, and possibly, drugs not in the protease inhibitor class—rather, nucleoside or non-nucleoside reverse transcriptase inhibitors or the new class of drugs called fusion inhibitors,” said DiRienzo.