We developed an ELISA for SARS that has synthetic peptide antigens as the solid-phase immunosorbent. Over 200 overlapping peptides, deduced from the Tor2 SARS-CoV genomic sequence (19), were synthesized as candidate antigens from the spike (S), membrane (M), and nucleocapsid (N) proteins. Candidate immunodominant S, M, and N peptides were selected and refined on the basis of serologic reactivities to a panel of serum samples from 13 patients clinically diagnosed with SARS at National Taiwan University Hospital in Taipei and the Xiaotangshan SARS Emergency Hospital in Beijing (20). Epitope mapping by serologic validation has been described for the development of peptide-based ELISA tests for HIV, hepatitis C virus, and foot-and-mouth disease virus (21–23). For the peptide-based SARS ELISA, wells of microtiter plates were coated with 2 µg/mL of a mixture of the S, M, and N protein–derived peptides, and serologic reactivities were determined by a standard ELISA procedure as previously described (23), except that the detector was horseradish peroxidase–conjugated goat anti-human IgG, and the chromaphore was 3,3´,5,5´-tetramethylbenzidine (TMB). In brief, serum samples, including two normal human samples provided as nonreactive controls, were diluted 1:20 in phosphate-buffered saline with carrier proteins and preservative. The diluted serum samples were reacted to the peptide-coated microtiter wells for l h at 37°C. Plates were washed 6 times, reacted to the antibody conjugate, again washed 6 times, and reacted to TMB; reactivity was then determined by A₄₅₀. Assay results were obtained within 3 h. Results were scored on the basis of the signal/cutoff (S/C) ratio, and cutoff absorbance was determined from the mean of the two controls plus 6 standard deviations (SD) from the distribution of normal human samples (Figure 1).

Seroconversion panels were collected as serial serum samples from SARS patients at National Taiwan University Hospital in the course of treatment. These panels tested positive for seroconversion by IFA (24) and are used here to evaluate analytical sensitivity.

A panel of 69 serum samples from convalescent SARS patients were provided by the Center for Disease Control, Department of Health, Taiwan, to evaluate diagnostic sensitivity. These serum specimens were confirmed for SARS-CoV infection clinically by the World Health Organization diagnostic criteria (1) and serologically by whole virus–based ELISA and IFA; some specimens were also confirmed by RT-PCR (24). Samples were drawn with appropriate timing for serologic reactivity (7–25 weeks after onset of symptoms). A panel of 1,390 plasma samples collected from random blood donors in Florida before 2001 was obtained from the Gulf Coast Regional Blood Center (Houston, TX) for specificity evaluation.

Additional specificity studies were conducted with serum that had serologic reactivities for bloodborne pathogens (HIV-1, HIV-2, hepatitis C virus, HTLV 1/II, and syphilis) obtained by United Biomedical Inc. before 2000 from various U.S. sources, an interference panel supplied by Boston Biomedica Inc. (Boston, MA) of serum samples with interference substances commonly found in processed clinical samples (EDTA, acid citrate dextrose, and citrate phosphate dextrose with adenine), and serum supplied by National Taiwan University Hospital from patients associated with typical and atypical respiratory pathogens other than SARS-CoV (influenza, rubella, cytomegalovirus, Epstein-Barr virus, and Mycoplasma pneumoniae). Serum samples were collected from healthy healthcare workers after interviews to confirm lack of signs and symptoms of SARS including fever, respiratory symptoms, and diarrhea.

The peptide-based ELISA was evaluated for sensitivity to seroconversion on eight seroconversion panels obtained from National Taiwan University Hospital (Figure 2, Table 1). In patient 1, seroconversion was detected by day 11 with an A₄₅₀ of 1.638. Absorbance remained at >2 at day 97. In patient 2, from whom blood was drawn on days 0, 6, 16, 27, and 116 (no samples were collected from day 6 to day 16), seroconversion was apparent on day 16 after the onset of fever. On day 116, the A₄₅₀ remained >3. In five of the other six seroconversion panels, from acute to convalescent phases, seropositivity was observed by days 8 to 12, and by day 16 in patient 6, from whom serum had not been collected from days 6 to 16 (Figure 2). The peptide-based ELISA showed an analytical sensitivity to earliest time of detection by week 2 and for duration of detection beyond day 100. The seroreactivities of patients 3 to 8 were also evaluated by a standard IFA method (24) for comparison. Seroconversion was detected in all six patients by the IFA method within 2 days of the peptide ELISA (data not shown).

The diagnostic sensitivity of the peptide ELISA was 100% on a panel of 69 convalescent-phase serum samples from SARS patients provided as a reference panel by the Center for Disease Control, Department of Health, Taiwan. These sera, confirmed for SARS by diagnostic and serologic criteria, displayed a mean S/C ratio of 8.08 (Figure 1).

The specificity of the peptide-based SARS ELISA was tested on plasma obtained before 2001 from 1,390 random Florida blood donors (Gulf Coast Regional Blood Center, Houston, TX). These normal plasma samples with a presumed zero seroprevalence rate gave a mean A₄₅₀ of 0.074 ± 0.0342 (SD). Subsequently, the cutoff value for the peptide-based assay was set as the mean A₄₅₀ for duplicate nonreactive controls plus 0.279 (based on 6 SDs from the mean for these 1,390 normal plasma samples). The distribution of the S/C ratio for the blood bank samples is plotted in Figure 1 with the mean S/C ratio of 0.28. None showed positive reactivity, for a specificity on the normal samples of 100% (Table 2).

The peptide ELISA was further evaluated for specificity with a pre-2000 collection of serum samples from patients with seropositivities for bloodborne pathogens such as HIV-1, HIV 2, hepatitis C virus, HTLV 1/II, and syphilis (various U.S. sources), and with normal serum samples spiked with interference substances heparin, EDTA, ACD, and CPDA-1 (Boston Biomedica). The 52 samples with seroreactivities for the pathogens all tested negative by the peptide-based SARS ELISA, as did the 41-sera interference panel (Table 2).

The peptide ELISA was evaluated for specificity on serum samples drawn from patients associated with typical and atypical respiratory pathogens other than SARS-CoV (National Taiwan University Hospital). These included samples from 1) 10 patients naturally infected with influenza (two sequential bleeds per influenza patient), 2) 10 patients with rubella, 3) 8 patients with cytomegalovirus infection, 4) 9 patients with Epstein-Barr virus, 5) 5 patients infected with Mycoplasma pneumoniae, a bacterial agent for atypical pneumonia, and 6) pre- and postvaccine blood samples from 16 patients given influenza vaccine. All samples were tested in duplicate. The site-specific antigens of the peptide SARS-CoV ELISA were free of cross-reactivities to the other respiratory pathogens (Table 2).

A prospective study was performed to determine asymptomatic infection among primary healthcare workers in hospitals that treated SARS patients. We collected serum samples from 623 healthcare workers without symptoms, not all of whom were in direct contact with SARS patients, who agreed to be tested for antibody to SARS-CoV at Ho Ping, Yang Ming, En Chu Kong, and Hsin Tai Hospitals, approximately 4 weeks after the outbreaks were recognized. Ho Ping and Yang Ming Hospitals had admitted SARS patients before the recognition of SARS and before healthcare workers had implemented control measures. Subsequently, these facilities experienced transmission to healthcare workers. The En Chu Kong and Hsin Tai facilities admitted patients once control measures had been implemented. Neither of these hospitals recorded transmission of SARS to healthcare workers.

ELISA detected three cases out of 383 samples from Ho Ping and one case in 50 blood samples from nursing aides at Yang Ming. These four positive samples, indicative of asymptomatic infection, were confirmed for seropositivity by IFA. None of the 190 serum samples from the two hospitals without nosocomial infection displayed seroconversion.