Serum Amyloid A Is Not Useful in the Diagnosis of Severe Acute Respiratory Syndrome

To the Editor:

In our present study, we aimed to investigate whether the serum concentration of serum amyloid A (SAA), as measured by the surface-enhanced laser desorption/ionization (SELDI) ProteinChip technology or by ELISA, is useful in differentiating the patients with severe acute respiratory syndrome (SARS) from the non-SARS patients who were suspected cases during the SARS outbreak period.

In a recent report from Kang et al. (1), the mean intensity of the protein peak at m/z 11514, identified by SELDI ProteinChip technology, of the SARS patient groups was 8 times greater than the intensity of the corresponding peak in the control patient group. This SELDI peak was observed previously in another proteomic study, under similar experimental conditions, and was identified as SAA (2). In a recent study by Yip et al.(3), the intensity of a SELDI peak at m/z 11695, which was identified as SAA, was significantly higher in the SARS patient group than in the control group. A similar increased peak was also found in pediatric patients with SARS (4). All of these studies suggested that SAA is useful in the diagnosis of SARS.

In these studies, the control cases were either healthy persons or patients with viral infections from other clinics. Unfortunately, the degree of similarity of the symptoms between SARS and control group participants and the time point of blood collection had not been considered in these studies. From the perspective of infectious disease diagnosis, we are not trying to differentiate healthy persons from infected patients; rather, we are trying to identify the disease causing the symptoms in patients presenting with similar symptoms (5). Bearing in mind the above issues, we recently attempted to profile and compare the serum proteomes of 39 adult patients in the early stages of SARS infection and 39 adult non-SARS patients who were suspected cases during the SARS outbreak period (6). We found specific SELDI peaks in the sera of the adult SARS patients; however, the peaks corresponding to SAA were not identified as SARS-specific features. This led us to question whether SAA is a useful biomarker for the diagnosis of SARS.

In our study, the non-SARS patients were those who had symptoms similar to SARS patients at admission. They were admitted to the same hospital as the SARS patients and were later shown to be negative for SARS coronavirus (CoV) infection by an anti-SARS-CoV antibody serology test at least 6 weeks after the onset of symptoms. The SAA concentrations in the serum samples (37 non-SARS cases and 29 SARS cases) that remained from the SELDI study were determined by an anti-SAA ELISA according to the manufacturer’s instructions (BioSource International).

Using WCX2 ProteinChip arrays (also called CM10) and pH 4 binding buffer, Tolson et al. (2) showed that 3 peaks, at m/z 11682, m/z 11526, and m/z 11439, were full-length SAA and des-arginine and des-arginine/des-serine variants at the NH₂ terminus, respectively. In our SELDI dataset, obtained with the same ProteinChip type and binding conditions, there were 3 SELDI peaks with similar m/z values (mean values): m/z 11681, m/z 11526, and m/z 11439. Spearman rank correlation analysis showed that the normalized intensities of these 3 peaks correlated highly with the serum concentration values obtained by ELISA (all correlation coefficients >0.9; all P values <0.0005; Table 1 ). Such high correlations strongly suggested that the SELDI peaks at m/z 11681, m/z 11526, and m/z 11439 were full-length SAA and the des-arginine and des-arginine/des-serine variants at the NH₂ terminus. In contrast to the previous SELDI studies, the normalized intensities of these 3 peaks were significantly lower in the adult SARS patients, instead of higher, than in the adult non-SARS patients (all P values <0.005; Table 1 ).

Table 1.

Summary of the SELDI peaks corresponding to SAA and serum SAA concentrations in adult SARS patients and adult non-SARS patients who were suspected cases during the SARS outbreak period.

A. Results of SELDI analysis
Mean (minimum–maximum) m/z of the SELDI peak	Theoretical average mass	Protein identity1	Mean (SD) normalized intensity of proteomic feature relative to total sum of proteomic features,2 %			P3	Correlation (r)4 with serum concentration of:
							Non-SARS	SARS		SAA	CRP5
11 439 (11 431–11 448)	11 439	SAA-1 (-RS)	0.25 (0.21)	0.10 (0.10)	0.004	0.912 (<0.0005)	0.444 (0.006)
11 526 (11 514–11 541)	11 526	SAA-1 (-R)	0.78 (0.62)	0.36 (0.43)	0.003	0.923 (<0.0005)	0.345 (0.029)
11 681 (11 657–11 689)	11 862	SAA-1	1.38 (1.19)	0.50 (0.58)	0.001	0.930 (<0.0005)	0.402 (0.013)

B. Serum SAA concentrations
	Non-SARS (n = 37)	SARS (n = 29)	P3	Correlation (r)4 with serum CRP concentration
Mean (SD) serum SAA concentration measured by ELISA, mg/L	853 (769)	395 (428)	0.003	0.499 (0.011)

¹-RS, des-arginine/des-serine variant; -R, des-arginine variant.

²Sample size for both non-SARS and SARS patients, n = 39.

³Mann–Whitney test.

⁴Spearman rank correlation test (P in parentheses).

⁵CRP, C-reactive protein.

When analyzing the ELISA data, we found that the serum SAA concentrations were greatly increased in both the SARS and non-SARS patient groups. The mean serum SAA concentrations of the SARS and non-SARS patient groups were 40- and 85-fold higher than the upper limit of the reference interval (<10 mg/L), respectively. Consistent with the SELDI data, the serum SAA concentrations were significantly lower in the SARS patient group (P <0.005; Table 1 ). The results from both the SELDI ProteinChip assays and ELISA indicated that serum SAA by itself was not useful in differentiating the SARS patients from the non-SARS patients who were suspected cases during the SARS outbreak period. Because serum SAA was increased in the SARS patients, however, we could not exclude the possibility that it could be used in combination with other serum markers to develop a classification model for SARS diagnosis.

Serum SAA is an acute-phase reactant (7) that has been shown to increase in various types of viral and bacterial infections (8). Regardless of the types of infection, serum SAA concentrations can increase up to 2000 mg/L. The degree of increase may reflect only the severity of the illness and does not indicate the cause. In the SARS patient group, we found that the SAA peaks and the serum concentration correlated significantly with the serum C-reactive protein concentration (Table 1 ), as in other infectious diseases (8). This suggests that the increases in serum SAA were caused mainly by the inflammatory response to SARS infection.

In conclusion, data from both the SELDI ProteinChip profiling study and an ELISA study do not support the contention that increased serum SAA is indicative for SARS. In contrast, our results strongly suggest that the serum SAA concentration is not useful in differentiating the SARS patients from the non-SARS patients who are suspected cases during the SARS outbreak period.