Serum samples collected in May 2001 from 938 healthy Chinese adults in Hong Kong and 48 confirmed SARS patients diagnosed in February and March 2003 in Guangdong were studied. All serum samples were aliquoted and stored at −20°C. The healthy adults were totally asymptomatic persons randomly recruited after a telephone interview concerning hepatitis B virus. The signs and symptoms of the SARS patients met the World Health Organization’s definition for surveillance, and SARS-CoV infection had been confirmed virologically.

All serum samples were heated at 56°C for 30 minutes. Specific antibodies for SARS-CoV and SARS-CoV–like virus were tested by using immunofluorescence (IF) assay at 1:10 dilution on FRhK-4 cells infected with either a human SARS-CoV strain (GZ50) (5) or an animal SARS-CoV–like virus (SZ16) (6), as reported (1). For sera positive for anti–SARS-CoV or anti–SARS-CoV–like virus, the antibody titer was further determined by serial titration. The IF-positive serum samples were serially diluted from 1:20 to 1:640 and then mixed with 100 50% tissue culture infective dose (TCID)₅₀ of the representative human or animal virus strains for a serum neutralization assay. After incubation for 1 hour at 37°C, the mixture was inoculated in triplicate onto 96-well plates of FRhK-4 cell cultures. The results were determined after 3-day incubation at 37°C.

Seventeen (1.8%) archived samples from healthy adults showed IF antibodies against the human virus, animal virus, or both (titer range 1:20 to 1:1,280) and were confirmed by serum neutralization assay. An additional six samples were IF-antibody positive at a 1:10 dilution to either animal or human viruses, but they were negative in neutralization assay and were treated as negative. The positive rate was highest in the group ages 51 to 60 years and appeared to be more prevalent in female (13/561, 2.3%) than male patients (4/377, 1.1%) (Table). Of the 17 seropositive serum samples, 10 were from housewives, retired, or unemployed persons; 6 were from clerks, unskilled workers, or students; and one was from a professional (Table). Most of the seropositive persons (13/17) had a higher IF or neutralization antibody titer to the animal virus than the human virus (Figure). By contrast, the control group, comprising convalescent-phase sera from 48 confirmed SARS patients recruited from hospitals in Guangdong, all showed positive antibody results for both human SARS-CoV and animal SARS-CoV–like viruses, but they invariably exhibited higher IF and neutralization antibody levels against the human virus than the animal virus (Figure).

While the exposure history and symptoms of study participants were unavailable for assessment, our results suggest that a small portion of Hong Kong adults had acquired a SARS-CoV–related virus infection at least 2 years before the 2003 SARS outbreak. Cross-reactivity of the antibody to human SARS-CoV and the animal SARS-CoV–like virus must have occurred, in view of the marked similarity between the two viruses. Recently, we reported that the very similar sequences differed only by 60 to 80 nt, including an additional 29 nt in the animal virus (6). We speculate that the viruses that affected the 17 healthy persons >2 years ago were antigenically closer to the recently isolated animal SARS-CoV–like virus than human SARS-CoV, but interspecies transmission from animal to human was probably inefficient as the viruses might not have adapted in the new host. This hypothesis would explain why only a few persons became infected and why they were likely to be asymptomatic. Avian influenza is another example of a virus appearing first in animals before causing a human disease. While approximately 3%–10% of healthy persons who were in close contact with farm or market chicken or fowls showed positive antibody to avian influenza viruses at the time of the H5N1 outbreak in humans in 1997, none of them had symptoms of influenza (7).

Although human SARS-CoV and animal SARS-CoV–like viruses are related to the three families of coronaviruses that cause respiratory and gastrointestinal diseases in animals, phylogenetic analysis has shown that they are different enough to make up their own, fourth group. The number of members in this new group is not clear. Important factors in the emergence of novel infectious diseases from animal sources include extensive exposure and rapid virus evolution (8), which facilitate human-to-human transmission. The growth of the demand for wildlife in markets in Guangdong in the past 15 years has provided an ideal platform to facilitate interspecies virus transmission from animals to humans. Such factors could even directly trigger a zoonotic disease outbreak. Our observations distinguished two distinct serologic patterns. The high ratio of antibodies to the animal virus compared to the relatively low ratio of antibodies to the human virus in a small proportion of healthy adults >2 years ago signifies the circulation of a SARS-CoV–like virus and its ineffective propagation in the human population. Following rapid virus evolution and in the presence of an unknown trigger, the novel SARS-CoV may have effectively adapted to the human host, as illustrated by a second pattern characterized by a higher human-to-animal virus antibody titer in infected persons. Although this pilot study was limited by an unstandardized design of sample collection, our preliminary findings suggest that the occurrence of SARS might not be due to an isolated cross-species transmission event, but rather to the rapid evolution of a related virus that has taken root in the human population. This implies an expected pattern of potential SARS recurrence. Measuring the prevalence of the two antibodies in different species of animals and persons who had close contact with the animals is important to improve our understanding of SARS-CoV transmission dynamics.