Login to Your Account



How HIV Treatment Failure Occurs Without Resistance

By Anette Breindl
Science Editor

Scientists have developed a mathematical model that they hope will ultimately be able to help determine how to best prevent resistance to currently available HIV drugs – and how to find effective combinations of new drugs early on in clinical development.

Collection of drug metabolism data that the model uses as one of its measurements is a standard part of preclinical development, co-corresponding author Alison Hill, of Harvard University, told BioWorld Today. And in some cases, researchers also try to generate resistance mutations to individual compounds. "Our model is a way to predict what such mutations would mean clinically," in the real-world context of drug combinations and varying levels of patient compliance with their drug regimens.

The model also explained one of the most puzzling features of HIV treatment failure – namely, why it can occur without resistance to the protease inhibitors that are the backbone of most combination drug regimens.

Since the 1980s, HIV infection has been transformed from a universally fatal disease to a chronic one. That transformation is largely due to the advent of highly active antiretroviral treatment, or HAART, combination drug regimens that can suppress viral load to undetectable levels.

HAART does not eliminate the virus, however. If patients do not stick to their treatment schedules, viral levels can bounce back. And given how complicated those treatment schedules can be, adherence is dishearteningly low.

The return of detectable viral levels can take two forms. Such viruses can be resistant to the drugs used to treat them. The current working hypothesis is that HAART is effective because HIV would need to evolve resistance to several drugs simultaneously, which is far less likely than evolving resistance to only one drug.

But in practice, many patients for whom HAART stops working do not have protease inhibitor-resistant HIV. In such cases, when HIV infection returns, it returns in the form of high levels of wild-type virus.

The team developed their model, which they published in the Sept. 2, 2012, advance online issue of Nature Medicine, to understand how wild-type virus can rebound in those cases. They used four parameters to predict the circumstances under which either resistance mutations would evolve, or viral levels would rise in the absence of such mutations.

Those four parameters were drug characteristics such as how fast different drugs are metabolized, drug-resistance mutations, the fitness costs of such mutations – since drug-resistant viral strains will grow more slowly than drug-sensitive ones when there is no drug around – and patient adherence to the treatment regimen.

The model suggested that treatment failure in the absence of resistance mutations occurs due to a mix of drug kinetics and patient adherence. Protease inhibitors, in particular, are metabolized rapidly, and so if patients do not stick to the drug-dosing schedule precisely, there are fairly long periods where blood levels of protease inhibitors are too low to prevent HIV replication.

In contrast, resistance mutations arise in the presence of higher drug levels and are influenced by the fitness cost of the mutations more than by patient adherence.

The work suggested that improving adherence could reverse HAART treatment failure in some patients, when such failure is due to the growth of wild-type virus due to low drug levels.

It also suggested that there are certain single-pill drug combinations that are less sensitive to patient adherence. In general, Hill said, partial adherence – such as when a patient is prescribed two different drugs and alternates them, perhaps to lessen side effects – makes it less likely that wild-type virus will grow, "because there's always some drug around." But it makes resistance mutations more likely, because such a regimen, though meant to be a combination treatment, is an "effective monotherapy" as far as blood levels of drugs are concerned.

The model, Hill said, "needs a few more parameters" before it can be used to make clinically relevant recommendations on what sorts of drug combinations are likely to be most successful clinically. Specifically, Hill and her colleagues are adding parameters to reflect the fact that HIV levels can be different in the blood plasma and in other organs.

Future versions of the model also will take into account the fact that multiple mutations can develop at the same time.

But ultimately, its utility will not be limited to HIV. Hill said in the lab of senior author Martin Nowak, very similar work is being done on cancer drug combinations, as well.