HIV is adept at evading attack by the immune system
An experimental HIV vaccine has for the first time cut the risk of infection, researchers say.
HIV researchers say the results from a large trial in Thailand offer renewed optimism that an effective HIV vaccine is possible.
What is different about this vaccine?
For more than 20 years, the field of HIV vaccine research has been dogged by disappointment.
Most notably, in 2007, Merck halted trials on a vaccine which had been regarded as one of the most promising to date, after finding no difference in rates of HIV infection between vaccinated and unvaccinated volunteers.
The latest trial combined two vaccines which on their own had been shown not to be effective.
In all 8,000 18-30 year olds in Thailand were given a primer dose with one vaccine followed by a booster dose of the other and 8,000 given a dummy vaccine.
After three years, the rate of HIV infection was a third lower in the vaccine group with 51 people infected compared with 74 unvaccinated people infected.
What do the results tell us?
The numbers of people in the trial who became infected with HIV were very small and there is always the possibility that the difference between the groups is down to chance.
But Dr Adriano Boasso, an HIV vaccine expert at Imperial College London, said the numbers were statistically significant.
"Tests will have been done to verify that the difference is unlikely to have occurred by chance and I have no trouble believing the figures."
Scientists are anxiously awaiting the publication of the full trial data so they can judge how significant the findings are.
But because this is the first time that an HIV vaccine has been shown to be effective - at least partially - the general view in the vaccine community is that the results are an important step forward.
The HIV strains used in the vaccines were B and E - the B strain predominates in North America and Europe - so the results are not immediately relevant to Africa where the main strain is C.
What do the experts think?
Some 33 million people around the world have HIV and a vaccine would have a huge impact especially in areas where the infection is endemic, such as sub-Saharan Africa.
But HIV appears to be remarkably good at fooling the immune system hence the failure to produce a vaccine that can protect against the disease.
The latest study offers renewed optimism for HIV vaccine researchers after years of depressing findings.
Scientists are cautiously hopeful that it will be possible to build on the positive results to boost the chance of eventual success.
Dr Boasso said the findings were a "breath of fresh air".
"It suggests there is a chance for a HIV vaccine and we shouldn't give up just because there have been some failures.
"Vaccine development is a long process and this is a step forward."
Lisa Power, head of policy at the Terrence Higgins Trust, said the results gives experts a good idea of where to concentrate research in the future.
And Dr Seth Berkley, president and CEO of the International AIDS Vaccine Initiative, added: "Until now, we've had evidence of feasibility for an AIDS vaccine in animal models.
"Now, we've got a vaccine candidate that appears to show a protective effect in humans, albeit partially."
So what now?
There is much work to be done before a vaccine becomes available.
It is highly unlikely that any HIV vaccine would get licensing approval with an efficacy of 30% and researchers will be aiming for a success rate in the region of 70-80%.
They will need to build on the findings and tweak the vaccine to produce a stronger response.
"The researchers will probably now look at the type of immune response they have induced to try and figure out what kind of effect they need to induce in volunteers to get a certain degree of protection," Dr Boasso said.
"That could also teach us quite a lot about other vaccines."
It is also possible that further studies could be done in high-risk volunteers, where rates of infection would be higher, although those kinds of studies are hard to do and results can be difficult to interpret.