"This throws into question a lot of the notions that have been accepted about the evolution of the virus" within a typical infected human, explained study co-author Dominik Wodarz, associate professor of biology at the University of California, Irvine.

He and another researcher, David Levy, of New York University, published their findings in the July 31 issue of the Proceedings of the Royal Society B.

Since its first recorded appearance nearly three decades ago, HIV infection has followed the same deadly path: a short, weeks-long period of acute flu-like symptoms followed by years of asymptomatic dormancy, and then symptoms of immune system breakdown that herald the emergence of AIDS.

But what is it that tips asymptomatic, low-level infection into AIDS?

The common dogma among scientists has long been that various strains of HIV battle a silent war within the body over time until the fittest -- defined as the strain that reproduces itself the most -- wins. That strain then goes on to overwhelm the body's immune cells and destroy the host's defenses against disease.

To test that theory, Wodarz and Levy constructed a complex mathematical model that took into account two factors about HIV: how fast the various strains replicate and how fast they kill cells (not always the same thing, the researchers noted). They also factored in human immune system responses to HIV.

What the two scientists found surprised them. According to the new model, AIDS actually begins when a less fit variety of HIV wins the day. This strain kills immune system cells extremely widely and quickly, but, in doing so, also limits the number of copies of itself it can produce. "It basically kills its own habitat, its house," Wodarz explained.

However, because this form of HIV is very good at quickly killing large numbers of immune cells, "once these less-fit strains emerge, they can plunge the patient into AIDS," Wodarz said.

In many cases, two or more strains of the virus can co-infect the same immune system cell, he added. If a fast-killing variety is one of those strains, it kills the cell before slower -- but better-replicating -- versions can go to work making millions of new viral particles.

"But without this ganging up on the same cell, the killer virus [that leads to AIDS] would go extinct, because evolution would select against it -- because it is less fit and replicates less," Wodarz explained.

That means that -- according to the model -- one way of keeping AIDS at bay might be to make sure that only one type of HIV invades a cell at any given time.

Specific cellular mechanisms do allow a second or third viral particle to enter a cell, and a medicine that thwarted these "party crashers" might keep the deadliest form of HIV from ever emerging, Wodarz speculated.

He pointed to wild monkeys that are infected throughout their lives with HIV-like simian immunodeficiency virus (SIV) but never get sick.

"Some of them have a lot of the virus, and it evolves a lot, but it does not cause AIDS, ever," Wodarz said. He suspects the monkey's immune cells may have evolved to block secondary viral entry and thereby keep the most dangerous strain of SIV at bay.

Not everyone is convinced by the new model, however.

Dr. Benigno Rodriguez is assistant professor of medicine at Case Western Reserve University in Cleveland, and a specialist in the evolution of HIV disease. He called Wodarz and Levy's paper "an interesting concept," but said it contained a few significant flaws.

First of all, he said, most of the available data suggests that HIV does get better at forming copies of itself as AIDS progresses. And Rodriguez believes the two scientists have left another important factor out of their model -- the fact that most AIDS patients' immune cells are not killed off by the virus directly but are destroyed by so-called "bystander" mechanisms that accompany AIDS.

"In an individual with advanced disease, if you look at the number of cells that are actually infected [with HIV], we are talking less than 1 percent," he said. "But, in reality, that individual may have lost 20, 30, 50 percent of his immune cells."

Rodriguez also questioned the importance of multiple strains of HIV infecting the same immune cell. "The data that we already have in hand shows that multiple infection is relatively infrequent," he said.

The bottom line, according to the Cleveland expert: As with any mathematical model, this one needs to be tested out in the laboratory.

Wodarz agreed that experimental verification is necessary, but he said mathematical disease models more often than not prove to be right.

In fact, he said, it was just such a model that led scientists to discover that HIV never stops evolving in the body -- even during infection's years-long asymptomatic phase.

"In HIV, mathematical models have led to great progress before," Wodarz said.

More information

To find out more about HIV/AIDS, head to the U.S. National Institute of Allergy and Infectious Disease