Serum amyloid A‐to‐albumin ratio as a potential biomarker to predict the activity, severity, and poor prognosis of systemic lupus erythematosus

Abstract Objectives To evaluate the predictive value of serum amyloid A‐to‐albumin ratio (SAR) for active systemic lupus erythematosus (SLE), severe active SLE, and poor prognosis of SLE. Methods One hundred and eighty‐six patients with SLE undergoing treatment in our hospital were selected. The demographic characteristics, clinical data, and disease prognosis of all patients were collected and analyzed. Results There were significant differences in SLEDAI, total glyceride (TG), serum amyloid A (SAA), SAR, urinary microalbumin‐to‐creatinine ratio (ACR), erythrocyte sedimentation rate (ESR), albumin (ALB), complement 3 (C3), anti‐dsDNA, anti‐Sm positive rate, and anti‐dsDNA positive rate between active SLE and stable SLE patients. TG, SAR, C3, ACR, and positive anti‐dsDNA were independent influencing factors of active SLE, and the odds ratio (OR) values were 2.342, 10.921, 0.832, 1.451, and 2.476, respectively. The area under curves (AUCs) of SAA, ALB, and SAR for predicting active SLE and severe active SLE were 0.743, 0.724, 0.787, 0.711, 0.686, and 0.733, respectively. The AUC of SAR for predicting the poor prognosis of active SLE was 0.719. High SAR, high ACR, low C3, and positive anti‐dsDNA were high risk factors for poor prognosis. Kaplan–Meier (K‐M) survival analysis showed that patients with high SAR, high ACR, low C3, and positive anti‐dsDNA had shorter continuous remission time than that with low SAR, low ACR, high C3, and negative anti‐dsDNA. Conclusion SAR had high predictive value for active SLE, severe active SLE, and poor prognosis of SLE. High SAR may be a potential marker for predicting the activity and prognosis of Chinese patients with SLE.


| INTRODUC TI ON
Systemic lupus erythematosus (SLE) is a chronic, inflammatory, and autoimmune disease, 1 which damages multiple organs and tissues, including the kidney, liver, and nervous system. 2,3 According to previous studies, SLE patients with high disease activity index have more active immune systems and more disordered inflammatory indexes, and as a result, they suffer from more serious tissues damage. 4 Previous studies also showed that the disease activity indexes are closely related to the types and doses of hormones 5 ; hence, timely and effective evaluation of activity indexes is very important in disease treatments. At present, the SLE Disease Activity Index (SLEDAI) and the British Isles Lupus Assessment Group Index 6,7 which mainly depend on laboratory test indexes and clinical symptom indexes are commonly used to evaluate disease progression and the prognosis of patients with SLE.
However, both scoring methods are complex and time-consuming.
An important clinical feature of autoimmune diseases is the disorder of inflammatory factors, such as serum amyloid A (SAA), C-reactive protein (CRP), albumin (ALB), erythrocyte sedimentation rate (ESR), and neutrophil-to-lymphocyte ratio, 8,9 which have been proved to be related with the severity and prognosis of the disease, and the predictive value of different indicators are different. Hwang YG et al. 10 showed that with the increase of disease activity in rheumatoid arthritis (RA), the levels of CRP and SAA increased, but SAA could better respond to this trend. Shen C et al. 11 showed that compared with CRP, SAA can better reflect the disease activity score for 28 joints (DAS28) of RA, and the correlation index with DAS28 was higher than that of CRP.
The ideal predictor of disease activity should have high degree of sensitivity and specificity and have certain predictive value for the severity and prognosis of the disease; in addition, the detection process needs to be simple and fast. Serum amyloid A (SAA), as an acute phase response protein, is widely used in the diagnosis of infectious diseases and the evaluation of the therapeutic effect. 12 Wang CM et al. 13 found that SAA was not only positively correlated with SLEDAI, but also an independent influencing factor of active SLE.
Yip J et al. 14 found that ALB in patients with active SLE was significantly lower than that in stable SLE, and ALB was negatively correlated with SLEDAI. C-reactive protein-to-serum albumin (CAR), as a new inflammatory marker, has been proved to have good predictive value for the diagnosis or prognosis of RA 15 and SLE, 16 and so on. At present, there is no report on the predictive value of SAR for active SLE, severe active SLE, and poor prognosis of SLE. This study compared the predictive value of SAA, serum albumin (ALB) and SAR for active SLE, and severe active SLE, in order to provide a new predictive biomarker for the disease activity and prognosis.

| Study population
This is a prospective study on the predictive value of SAR in ac-

| Statistical analysis
Spss20.0 was performed to establish a database. Categorical variables were presented as counts and compared by chi-squared test.
Normal distribution data were presented as the mean ± standard deviation (SD) and compared with Student's t-test. Non-normal distribution data were presented as the median and interquartile range (IQR) and compared with Mann-Whitney U-test. Binary logistic regression analysis was used to find the independent influencing factors for active SLE, and Spearman's correlation analysis was used to analyze the correlation between two continuous variables. The receiver operating curve (ROC) was used to analyze the predictive value of different indicators for active SLE, severe active SLE, and poor prognosis of SLE, and the optimal clinical cutoff value was determined by the maximum Youden index (Youden index =sensitiv-ity+ specificity-1). Medcalc software was used for area under curve (AUC) comparison, and the Z-test was used to compare the predictive ability of different indicators. Potential risk factors which identified in a univariate model were included in a multivariate model.
Kaplan-Meier (K-M) analysis was used to estimate the survival curve of sustained remission between different groups, and logrank test was used to analyze the differences between two groups. p < 0.05 means the difference was statistically significant.

| Clinical characteristics of participants
From January 2018 to March 2020, 230 patients were diagnosed with SLE in Traditional Chinese Medicine Hospital of Taihe. Among them, 3 patients were with cancer, 7 patients were with RA, 8 patients were with pregnancy, 4 patients were with severe liver disease, 10 patients were unwilling to participate in the study, and 12 patients lost follow-up after discharge, above of them were excluded from the study. Finally, 186 SLE patients were included in the study ( Figure 1). Table 1 shows the clinical characteristics of all participants. The average age of participants was 38.05 years old, ranging from 20 to 56, and the ratio of male to female was 1: 9.94. There was no significant difference in age, sex, and disease duration between active SLE and stable SLE.

| Analysis on influencing factors of active SLE
All SLE patients were divided into active SLE group and stable SLE group according to whether SLEDAI ≥5. Testing the baseline clinical data between two groups, 11 factors shown differed significantly (Table 1)

| Predictive value of SAA, ALB, and SAR for active SLE
The AUCs of SAR, SAA, and ALB for predicting active SLE were 0.787, 0.743, and 0.724, respectively. Compared with SAA and ALB, SAR had the highest predictive value. The optimal cutoff value of SAR was 0.43 mg/g, and the prediction sensitivity and specificity were 67.20% and 79.70%, respectively. The optimal cutoff value of SAA was 16.05 mg/L, and the prediction sensitivity and specificity were 63.90% and 75.00%, respectively. The optimal cutoff value of ALB was 38.50 g/L, and the prediction sensitivity and specificity were 66.40% and 71.90%, respectively (Figure 2).  (Figure 3).

| Correlation analysis between SAA, ALB, SAR, and SLEDAI
The correlation analysis results showed that SLEDAI was significantly and positively correlated with SAA ( Figure 4A, r = 0.409, p = 0.000), but negatively correlated with ALB ( Figure 4C, r = −0.368, p = 0.000). A stronger positive correlation was observed between SAR and SLEDAI ( Figure 4B, r = 0.440, p = 0.000).   Figure 5A) or high C3 (C3>0.62 g/L, Figure 5C). It also showed patients with high ACR (ACR≥64.34 mg/g) or positive anti-dsDNA were more like to have shorter continuous remission time than patients with low ACR (ACR<64.34 mg/g, Figure 5B) or negative anti-dsDNA ( Figure 5D), and the differences were statistically significant.

| Predictive value of SAR for poor prognosis in active SLE
ROC curve analysis revealed that the AUC of SAR was 0.719, the optimal cutoff value was 0.53 mg/g, and the prediction sensitivity and specificity were 72.10% and 63.30%, respectively ( Figure 6).

| DISCUSS ION
To the best of our knowledge, this is, thus far, the first study for SAR in SLE. The purpose of this study was to investigate the predictive value of SAR in active SLE, severe active SLE, and poor prognosis of SLE. Our findings showed that SAR had high predictive value for active SLE, severe active SLE, and poor prognosis of SLE. High SAR, high ACR, high TG, low C3, and positive anti-dsDNA were demonstrated to be independent influence factors for active SLE. In addition, high SAR, high ACR, low C3, and positive anti-dsDNA were demonstrated to be independent prognostic factors for shorter con- in serum and inflammatory lesions in glomerulonephritis. 28 The fact that circulating antibody levels are usually associated with active SLE and renal involvement has strengthened the assumption of pathogenetic importance of anti-dsDNA. 29 In our study, we found that SAR, C3, TG ACR, and anti-dsDNA were independent influencing factors of active SLE. In addition, SAA and SAR were positively correlated with SLEDAI, but ALB was negatively correlated with SLEDAI. These three indicators had high predictive value for active SLE and severe active SLE, and the predictive value of SAR was significantly higher than SAA and ALB. The reasons may be that SAR combines the positive correlation factor and the negative correlation factor, which expand the inflammatory difference between active SLE and stable SLE, severe active SLE, and non-severe active SLE. Therefore, SAR prediction value was obviously higher than a single index.
Many studies proved that laboratory indicators not only had important reference value for SLE diagnosis, but also could judge the activity, recurrence, and treatment effect, including age, sex, race, economic, and organ damage. 30 According to previous research, the prognosis of Caucasians was better than Blacks, 31 the prognosis of men was worse than women, 32 and the prognosis of patients with superior family economic conditions and high educational level was better than patients with poor economic conditions and low educational level. 33 However, there are few studies on which laboratory indexes and clinical symptoms related to the prognosis of SLE.
Feng XN reported that the decreasing of CD4 + T lymphocytes and the increasing of ESR were risk factors for poor prognosis. 34 Pang J

Variables
The univariable Cox regression analyses  35 In our study, we found that the poor prognosis rate of active SLE was 35.25%, which was higher than that reported above.
The reason may be related to the longer follow-up of this study than above. We also found that the increasing of SAR, ACR, the decreasing of C3, and the positive of anti-dsDNA were risk factors for poor prognosis in active SLE. K-M survival analysis further showed that patients with high SAR, high ACR, low C3, and positive anti-dsDNA had shorter continuous remission time than that with low SAR, low ACR, high C3, and negative anti-dsDNA. The above results were reported for the first time.
However, this study had several limitations. Firstly, there was a small sample size included in this study, resulting in obvious break points of ROC curve analysis. Secondly, the influence of treatment factors on active SLE after discharge was not analyzed. In the future, multicenter, big data and multi-index prospective research may help

| CON CLUS ION
This study aimed to investigate the predictive value of SAR for active SLE, severe active SLE, and poor prognosis of SLE. Our data indicated that high SAR may be a potential biomarker for predicting the activity and poor prognosis of Chinese patients with SLE. This potential indicator can help doctors predict the disease severity and prognosis of SLE in time. SAA and ALB should be detected for such patients, and the treatment plan should be adjusted according to SAR to prevent further progress of the disease.

ACK N OWLED G M ENT
Thanks to all colleagues who conducted telephone follow-up and provided statistical analysis in this study.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data are available upon reasonable request.

Feng He
https://orcid.org/0000-0002-6847-3467 F I G U R E 6 ROC curve analysis of SAR for poor prognosis in active SLE