CHINA.- A new case of human-to-human transmission doesn't prove that H5N1 bird flu is learning to spread among humans, but it reinforces fears that there could be many more undiagnosed human infections in China. It may also point to a potential cure.

Last November, a salesman hospitalised in Nanjing, China, with fever, diarrhoea and pneumonia was given antibiotics for suspected bacterial infection, but tested positive for H5N1 shortly before he died. The next day his father fell ill with a nearly identical virus, and was given plasma from a woman who had received an experimental whole-virus H5N1 vaccine. He recovered (The Lancet, DOI: 10.1016/S0140-6736(08)60493-6).

The month before, an H5N1 patient in Shenzhen also recovered after receiving plasma from someone who had survived the infection, suggesting that antibodies from such survivors are a promising approach to treating H5N1, and should be investigated further.

However, poultry in Chinese markets are required to be vaccinated, and the fact that the Nanjing salesman had visited a live poultry market shortly before he fell ill reinforces fears that many Chinese could be getting H5N1 from vaccinated poultry, which carry the virus but remain healthy. Like the salesman, such people could be misdiagnosed because they have not been near sick birds.

Jeremy Farrar of the University of Oxford's Clinical Research Unit in Ho Chi Minh City, Vietnam, says scientists should study family clusters of H5N1 to learn what allows the virus to infect humans. Clusters in Thailand, Indonesia, Pakistan and possibly Vietnam have involved only genetically related people. Generally, the virus remains hard to catch - none of other 91 people who'd had contact with the salesman or his father were infected.

This suggests that "there may be a rare genetic predisposition to the virus", Farrer says.