Financial Math Models May Help Tailor Future HIV Vaccines

Predicting the evolution of HIV surface proteins could be used to develop better vaccines.

Scientists drew inspiration from financial math models—–used to predict changes in stock prices––to accurately predict the evolution of HIV surface proteins, which could improve treatment of the virus.

Particularly, investigators from the University of Iowa used computational tools to focus on how different properties of the envelope glycoprotein (Env) evolved in the HIV-positive population of Iowa over a 30-year period.

“HIV is a highly dynamic virus,” said senior author Hillel Haim, of the study published in PLOS Biology. “It continuously changes, both in an infected individual and, as a consequence of that, in the greater population. When we make a vaccine, we are essentially trying to mimic the virus so that the immune system will learn how to recognize and attack the real virus. The problem we are trying to solve for HIV is how can you design a vaccine to hit a moving and continuously changing target?”

Because Env frequently mutates, it causes an increased number of Env variants in the population. Furthermore, this process contributes to the limited success of HIV vaccines to date.

To develop a vaccine that continuously matches HIV over time, investigators need to know which Env variants are circulating in the patient population to predict how the proteins will change.

For the study, investigators used computational tools and approaches inspired by mathematical models. Using this approach, the investigators accurately predicted how different properties of the Env protein evolved in the population of Iowa over 30 years.

In the 1980s, the investigators established an HIV clinic in Iowa City, where they collected blood samples from several hundred patients over the last 3 decades.

From these samples, the investigators isolated and analyzed hundreds of HIV Envs to examine changes in structural properties of the protein that occurred in the HIV-infected population of Iowa over the last 30 years.

The patterns of change observed reminded the investigators of prior research where they examined the diffusion of viruses through liquids.

“Studying the physical process of virus particle diffusion, I became familiar with that [math],” Haim said. “Zoom forward 10 years and looking at the patterns of change in virus properties, I said ‘Wow, this is diffusion!’”

The investigators compared the Envs of different viruses derived from the same blood sample, and found that some properties were relatively similar and others were highly variable. They defined the characteristic variance as volatility, and the volatility of each property was similar among the patients.

“We found that volatilities of Env properties measured from a few patient samples from the 1980s allowed us to accurately predict how these properties of the virus evolved in the Iowa population over the course of 30 years,” Haim said.

The ability to accurately predict future changes by testing a small number of patients could eventually lead to more tailored vaccines for specific forms of HIV, according to the study authors.

“Fortunately, relative to the financial market models that inspired this work, our predictions of changes in HIV are remarkably accurate due to the highly-conserved nature of randomness in this virus,” Haim said.