The work provides more evidence that SARS jumps from animals to humans, possibly frequently. And it suggests, says one of the researchers, that prompt control of new cases is crucial, before viral strains have much time to adapt to people.
"If there's any lesson from this, it's stop it early before they know how to spread in humans," said evolution specialist Chung-I Wu of the University of Chicago, who helped China's SARS Epidemiology Consortium analyze the findings.
The consortium's research, which genetically fingerprinted virus samples from several dozen infected people and animals from China and Hong Kong, is being published in Friday's edition of the journal Science.
Severe acute respiratory syndrome emerged in southern China in late 2002, and went on to sicken more than 8,000 people worldwide before subsiding last June. In that time, it killed nearly 800 people.
Civet cats, mongoose-like animals that are sold in live food markets in southern China, are the top suspect for first spreading the disease to people. Thousands have been slaughtered as a preventive measure.
But scientists don't know whether some other animals, perhaps rodents who live in the same markets, are the ultimate source of SARS and infected both civets and people at the same time. Determining SARS' natural reservoir is a top research goal.
The new study doesn't shed light on that question.
However, it does suggest that the outbreak wasn't caused by a single species jump. Instead, the study says 11 people apparently were independently infected in the Pearl River Delta area of Guangdong Province beginning in November 2002 - patients whose virus seems identical to viral samples found in some captive civets.
"The paper is fascinating to me because what it shows, with this careful documentation, is that the virus has jumped from the palm civet into the human more than once," said David Brian, a microbiologist at the University of Tennessee in Knoxville.
Then came what the Chinese researchers dubbed the outbreak's "middle phase," 130 cases beginning in January 2003, most caught in a single hospital.
SARS underwent rapid genetic changes between November and the hospital-based infection, the study found, suggesting the mutations made person-to-person spread easier.
One of those mutated strains came to dominate the outbreak. The Chinese researchers linked that strain's first appearance to a patient who got sick in February - and whose doctor a few days later traveled to Hong Kong and spread the virus in the Metropole
Hotel, the launching pad for global spread.
Then, mutations slowed for the remainder of the outbreak.
But last month, China reported the first of three new suspected SARS cases, and that patient was analyzed in the new study, too. His strain is much more similar to SARS found in civets than to any yet-documented human infection, the Science paper concludes.
That's not proof a civet infected the man, cautioned Dr. Mark Denison, a specialist in pediatric viral infections at Vanderbilt University.
But, "it makes the continual looking for a reservoir host important. It also states that these kinds of transmissions may not be rare," he said.
By Lauran Neergaard