The news was a major disappointment for AIDS research. It was not clear why the gel did not work in this trial, since it had seemed to work surprisingly well in a previous one.
Finding a vaginal gel that protects women against the virus that causes AIDS but still allows them to get pregnant has long been sought by AIDS researchers, because it can be used secretly by women who fear being refused or even beaten if they ask their sexual partner to use a condom.
The first trial, reported in South Africa in the summer of 2010, found that a vaginal gel containing the drug tenofovir protected 39 percent of the women who used it, and that those who used it most regularly reduced their chances of infection by 54 percent.
It was hoped that the new trial, nicknamed Voice (for Vaginal and Oral Interventions to Control the Epidemic), would confirm that earlier trial, called Caprisa after the clinic in Durban that ran it.
The Voice trial, which began in 2009, enrolled more than 5,000 women in South Africa, Uganda and Zimbabwe. It was divided into three experiments, or arms, comparing three different products against a placebo — the gel, a tenofovir pill or a Truvada pill (tenofovir and a booster drug).
The trial of the tenofovir pill was canceled in September because it, too, did not appear to be working.
But because part of the study is still continuing, all the collected data — meaning, in particular, who was on the gel and who was on the placebo — cannot be “unblinded” yet, so the researchers cannot try to figure out why it did not work.
“Even when we have more information available to us, understanding why our results differed from the Caprisa results may not be clear,” said Sharon L. Hillier, a lead researcher for the Microbicide Trials Network, which is based at the University of Pittsburgh medical school and oversees many trials.
She said was “surprised and disappointed” by the cancellation.
In a statement, she and Dr. Ian McGowan, another researcher for the network, speculated that the problem might have been that too few women used the gel regularly, that the dosing schedule was wrong, or that it somehow caused inflammation that led to easier entry by the virus. But, Dr. Hillier added, it was unlikely that they would be able to assess that until later next year.
Ethics of modern clinical trials require that at various midway points, enough data be revealed to a panel of outside experts so they can assess whether the intervention being tested is safe for the participants and whether it is working.
In this case, 6 percent of women using the tenofovir gel and 6 percent of those using the placebo had become infected by the time the outside panel looked at the data. It was found to be safe but not effective, which ethically requires the cancellation of the trial to keep any more women from becoming infected.
The trial is expected to go on until mid-2012 and the data are to be released in early 2013. Other trials of gels at different formulations and dosing are planned or under way.