Cells that harbour HIV could be picked off by a drug developed by US scientists.
HIV infecting an immune system cell
One of the major obstacles to efforts to treat HIV is the fact that "reservoirs" of the virus can persist in certain types of human cell.
Doctors at the University of Texas Southwestern Medical Center found a toxin which targets them directly.
Early laboratory tests suggest that infected cells could be "purged" by a short course of treatment, they say.
It is not known yet whether the treatment will work in actual patients - or whether it will improve long-term outcomes for them.
It is hoped, however, that whittling down the number of cells harbouring HIV infection could help keep the virus under control for longer periods.
Even when patients are successfully treated using anti-retroviral therapy, and tests can detect no sign of HIV in the blood, it can find hiding places where drugs cannot destroy it, and become active again at a later point.
The scientists are aiming their treatment at a particular type of immune cells called "T" cells - and more specifically, a sub-type called "memory helper T cells".
It is these that hold onto to information about infections so that the immune system can mount a response if the same virus or bacterium comes around again.
They are also well-known for harbouring HIV.
The US researchers, whose work is published in the journal Proceedings of the National Academy of Sciences, used an "immunotoxin" designed to attack this cell type.
This is an antibody - designed to lock onto this type of T cell - coupled with a toxin molecule, in this case derived from the potent nerve toxin ricin.
They took blood samples from 24 HIV patients whose blood tests showed no sign of the virus, and extracted T cells from them.
Treatment with the immunotoxin significantly reduced the numbers of the key cell type believe to act as a reservoir for the virus.
The researchers said: "Our results suggest this immunotoxin should be tested for ability to purge one of the critical latent viral reservoirs in patients."
They issued a call for other HIV research teams to help them conduct a bigger trial into the immunotoxin.
However, some UK experts are more cautious about the prospects of success.
Professor Frances Gotch, from Imperial College London, told BBC News Online that while the laboratory tests suggested that the purge could eliminate any cells found in the bloodstream, to be fully successful, the immunotoxin would have to gain entry to the lymph nodes - a far more difficult proposition.
She said: "Approximately 95% of these cells are found outside the peripheral blood system. We can't say whether something that works in the laboratory would work in humans."
She pointed out that the net effect of the treatment might be to make the body more vulnerable to infection by destroying its immune system "memory".
Professor Gotch expressed worry about the use of a potent toxin such as ricin, even for a short course of treatment.
"Rather than moving straight to human trials, perhaps the researchers should consider working in non-human primates," she said.