Researchers Find Way to Predict Antiviral Activity of Three-drug Combinations in HAART

Highly active antiretroviral therapy, or HAART, typically combines three antiretroviral drugs (ARVs) as a first line treatment for HIV infection. It is a very effective way to keep HIV replication under control. So effective, in fact, that the best drug combinations suppress HIV to undetectable levels in most patients.   

What remains relatively unclear is why some but not other drug combinations work. Further, there is still no systematic method for determining which drugs to combine to get the most effective treatments.   

Robert Siliciano, a professor of medicine at Johns Hopkins University School of Medicine, and his colleagues seem to have solved both problems. In a paper published in Nature Medicine, they describe a model that predicts how much HIV inhibition might be achieved by any three-drug combinations of the 19 most commonly prescribed ARVs (Nat. Med. 18, 446, 2012).   

To create this model, they first measured in an in vitro assay how much each of the 19 ARVs inhibit a single round of HIV replication. They did this for several different concentrations of each drug, and plotted out their results in a graph—generating a dose-response curve that describes how viral inhibition varies with each concentration of every studied drug. They found that the shape of the curves differed between the different ARVs. This highlights the importance of using dose-response curves in determining antiviral activity, Siliciano says, because these ARV-specific differences would likely have been missed by the way researchers have previously used to describe antiviral activity: By only focusing on one drug concentration, the so-called IC50, which results in 50% viral inhibition.   

Next, the researchers determined the dose-response curves for almost all possible two-drug combinations of those 19 drugs, and compared them with the predictions of the relatively simple mathematical models researchers currently use to predict the potential efficacy of drug combinations. They found that, in more than 40% of the cases, these existing models of combined drug effects were unable to correctly predict their experimental results.   

According to Siliciano, these existing models are too simplistic, because they are based on two extreme scenarios. One model assumes that the drugs to be combined target the same process in the HIV life cycle, in which case their pairing would have only an additive effect; the other assumes that two drugs target completely different processes in the HIV life cycle, in which case their combined effect should be multiplicative. Siliciano and his colleagues found, however, that in more than 40% of the cases the drug pairs they measured had effects that did not precisely fit either of these scenarios. “The existing models were so bad that we developed our own,” says Siliciano.   

To do so, he and his colleagues used the results of their dose-response measurements to determine just where between the extremes the effects of various drug-pairs really fell. They then used this information to calculate the inhibitory potential of three-drug combinations, by pulling together what they knew about the inhibitory effects of the three drugs in any given three-drug combination alone and their pairwise combinations.   

When they tested their predictions for 10 three-drug combinations in their in vitro assay, they found that this predicted HIV inhibition much better than the existing models. Their predictions also correlated very well with the outcomes of 47 clinical trials of three-drug combination regimens.   

Before this study, there was no good way to predict the activity of three-drug combinations, says Ruy Ribeiro, a research scientist at the Los Alamos National Laboratory who models HIV treatment and wrote a commentary on the study in the same issue of Nature Medicine. “This gives some sign posts to what drug combinations to try in clinical trials,” he says. The new approach could also make it easier for researchers to identify less expensive drug regimens for use in developing countries, Ribeiro adds, allowing them to devise regimens that cost less but have potentially equivalent effects.   

Siliciano, for his part, hopes to mainly use the new approach to improve management of patients with drug resistance. This should now be possible, he says, by taking into account another recent finding by his group that suggests that resistance mutations in HIV can alter the shape of the dose-response curve (Proc. Natl. Acad. Sci. 108, 7613, 2011).