The only clinical trial data on the experimental Ebola drug ZMapp shows it is 100% effective in monkey studies, even in later stages of the infection.

The researchers, publishing their data in Nature, said it was a "very important step forward".

Yet the limited supplies will not help the 20,000 people predicted to be infected during the outbreak in West Africa.

And two out of seven people given the drug, have later died from the disease.

ZMapp has been dubbed the "secret serum" as it is still in the experimental stages of drug development with, until now, no public data on effectiveness.

Doctors have turned to it as there is no cure for Ebola, which has killed more than 1,500 people since it started in Guinea.

Cocktail

Researchers have been investigating different combinations of antibodies, a part of the immune system which binds to viruses, as a therapy.

Previous combinations have shown some effectiveness in animal studies. ZMapp is the latest cocktail and contains three antibodies.

Trials on 18 rhesus macaques infected with Ebola showed 100% survival.

This included animals given the drug up to five days after infection. For the monkeys this would be a relatively late stage in the infection, around three days before it becomes fatal.

Scientists say this is significant as previous therapies needed to be given before symptoms even appeared.

One of the researchers, Dr Gary Kobinger from the Public Health Agency of Canada, said this was a huge step up from previous antibody combinations.

"The level of improvement was beyond my own expectation, I was quite surprised that the best combination would rescue animals as far as day five, it was fantastic news.

"What was very exceptional is that we could rescue some of the animals that had advanced disease."

Human implications?

However, there is always caution when interpreting the implications for humans from animal data.

A Liberian doctor, one of three taking the drug in the country, and a Spanish priest both died from the infection despite ZMapp treatment.

William Pooley, the first Briton to contract Ebola during this outbreak, has been given the experimental drug ZMapp as were two US doctors who recovered.

The course of the infection is slower in humans than macaques so it has been cautiously estimated that ZMapp may be effective as late as day nine or 11 after infection.

But Dr Kobinger said: "We know there is a point of no return where there is too much damage to major organs, so there's a limit."

The group wants to start clinical trials in people to truly assess the effectiveness of the drug.

Commenting on the findings, Prof Jonathan Ball, a virologist at the University of Nottingham, told the BBC: "Before, ZMapp was a total mystery.

"This is an incredible improvement on those earlier cocktails, to have 100% clearance and most importantly that clearance when they've started to show outward signs of infection."

Referring to the seven treated patients, he added: "Clearly there is the caveat that all evidence in humans is anecdotal and no hard evidence has been released on what happens to the virus in those patients."

Prof Peter Piot, the director of the London School of Hygiene and Tropical Medicine, said: "I never thought that 40 years after I encountered the first Ebola outbreak, this disease would still be taking lives on such a devastating scale.

"This welldesigned trial in non-human primates provides the most convincing evidence to date that ZMapp may be an effective treatment of Ebola infection in humans.

"It is now critical that human trials start as soon as possible."