key: cord-0795299-1nkrt28h
authors: Zinellu, Angelo; Paliogiannis, Panagiotis; Carru, Ciriaco; Mangoni, Arduino A
title: Serum amyloid A concentrations, COVID-19 severity and mortality: an updated systematic review and meta-analysis
date: 2021-03-15
journal: Int J Infect Dis
DOI: 10.1016/j.ijid.2021.03.025
sha: 4daae0f458f9f53462d1dea8a1d6bbe8196ec7d7
doc_id: 795299
cord_uid: 1nkrt28h

BACKGROUND AND OBJECTIVES: An excessive inflammatory response in patients with coronavirus disease 2019 (COVID-19) is associated with high disease severity and mortality. Specific acute phase reactants might be useful for risk stratification. We conducted a systematic review and meta-analysis of studies of serum amyloid A (SAA) in patients with COVID-19. METHODS: We searched PubMed, Web of Science and Scopus, between January 2020 and December 2020, for studies reporting SAA concentrations, COVID-19 severity and survival status. RESULTS: Nineteen studies in 5,617 COVID-19 patients were included in the meta-analysis. Pooled results showed that SAA concentrations were significantly higher in patients with severe disease and non-survivors (standard mean difference, SMD, 1.20, 95% CI 0.91-1.49, P < 0.001). Extreme between-study heterogeneity was observed (I(2) = 92.4%, P < 0.001). In sensitivity analysis, the effect size was not significantly affected when each study was in turn removed (range between 1.10-1.29). The Begg's (P = 0.030), but not the Egger's (P = 0.385), test revealed the presence of publication bias. Pooled SMD values were significantly and positively associated with gender (t = 2.20, P = 0.047) and aspartate aminotransferase (t = 3.44, P = 0.014). CONCLUSIONS: SAA concentrations were significantly and positively associated with higher COVID-19 severity and mortality. This acute phase reactant might assist with risk stratification and monitoring in this group.

A state of excessive local and systemic inflammation and immune activation are strongly associated with oxidative stress, coagulation abnormalities, and multi-organ dysfunction in patients with J o u r n a l P r e -p r o o f coronavirus disease 2019 (Fajgenbaum and June, 2020, Hojyo et al., 2020) . While safe and effective vaccines have been developed and are currently being rolled out effective therapies, e.g., repurposed antiviral and immunosuppressant agents, to mitigate the clinical manifestations of COVID-19 remain limited (Siemieniuk et al., 2020) . In this context, the use of biomarkers of disease severity and clinical progression would facilitate the early identification of patients requiring aggressive management and monitoring and assist with the judicious use of health care resources. Given the key pathophysiological role of inflammation and immunity in the clinical progress of COVID-19, markers that reflect the activation of these pathways might be particularly useful for risk stratification and effective management. Serum amyloid A (SAA) genes and proteins are significantly activated during the acute phase response, which comprises a number of phenomena, e.g., increased temperature and hormonal and metabolic alterations, that occur in the presence of inflammation and infection (Yoo and Desiderio, 2003) . Circulating SAA concentrations, typically low under physiological circumstances (20-50 mg/L), can increase up to 1,000-fold within the first 24-48 hours of an acute phase response. This is the consequence of an increase synthesis in the liver that is triggered by several stimuli, including tumor necrosis factor (TNF), interleukin-1 (IL-1), interleukin-6 (IL-6) and interferon- (Morrow et al., 1981, Uhlar and Whitehead, 1999) . SAA, in turn, can activate the complement system and the nucleotide-binding domain leucine-rich repeat-containing family pyrin-domain containing 3 (NLRP3) inflammasome, further increase the synthesis of TNF, IL-1, and IL-6, and activate other pro-inflammatory cytokines such as interleukin-1 and IL-23 (Ather et al., 2011 , De Buck et al., 2016 , Yuste et al., 2007 . Notably, these mediators have been shown to significantly contribute to the onset of the cytokine storm and its adverse clinical consequences in COVID-19 (Fajgenbaum and June, 2020) . Therefore, it is plausible that the acute increase in SAA concentrations in patients with might not only reflect the presence of an acute phase response but also herald the development of a cytokine storm and, consequently, multi-organ failure and an increased risk of adverse outcomes.

Two systematic reviews and meta-analyses on a relatively limited number of studies, three and five, J o u r n a l P r e -p r o o f respectively, have reported a significant and positive association between SAA concentrations and COVID-19 severity (Akbari et al., 2020 , Zeng F. et al., 2020 . Following the publication of several additional studies, we conducted an updated systematic review and meta-analysis of the available evidence on the clinical implications of SAA concentrations in patients with COVID-19.

We conducted a systematic search, using the terms "serum amyloid A" or "SAA" and "coronavirus disease 19" or "COVID-19", in the electronic databases PubMed, Web of Science, and Scopus, between January 2020 and December 2020, for studies investigating SAA concentrations in COVID-19 patients according to disease severity or survival status. The references of the retrieved articles were also searched to identify additional studies. Inclusion criteria were the following: (a) studies reporting continuous data on SAA concentrations in COVID-19 patients, (b) articles investigating COVID-19 patients with different disease severity or survival status, (c) adult patients, (d) 10 participants, (e) English language, and (f) full-text available. Two investigators independently screened the abstracts. If relevant, the two investigators independently reviewed the full articles. We used the Newcastle-Ottawa scale to assess the quality of each study, with a score 6 indicating high quality (Wells et al., 2013) .

Standardized mean differences (SMD) were calculated to build forest plots of continuous data and to evaluate differences in SAA concentrations between COVID-19 patients with low vs. high disease severity or survivors vs. non-survivors during follow up. A P-value of less than 0.05 was considered statistically significant, and 95% confidence intervals (CIs) were reported. When studies reported concentrations as median and interquartile range (IQR) the mean and standard deviation J o u r n a l P r e -p r o o f were estimated as previously described (Wan et al., 2014) . The Q-statistic was used to test heterogeneity of SMD across studies (the significance level was set at P<0.10). Inconsistency across studies was evaluated using the I 2 statistic where I 2 <25% indicated no heterogeneity, I 2 between 25-50% moderate heterogeneity, I 2 between 50-75% large heterogeneity, and I 2 >75% extreme heterogeneity (Bowden et al., 2011, Higgins and Thompson, 2002) . A random-effects model was used to calculate the pooled SMD and the corresponding 95% CIs in presence of significant heterogeneity. Sensitivity analyses were conducted to assess the influence of individual studies on the overall effect size using the leave-one-out method (Tobias, 1999) . The presence of publication bias was assessed using the of Begg's adjusted rank correlation test and the Egger's regression asymmetry test at the P<0.05 level of significance (Begg and Mazumdar, 1994, Sterne and Egger, 2001 ). The Duval and Tweedie "trim and fill" procedure was also used to investigate the effect of publication bias. This method recalculates a pooled SMD by incorporating the hypothetical missing studies as though they actually existed, to augment the observed data so that the funnel plot is more symmetric (Duval and Tweedie, 2000) . Statistical analyses were performed using Stata 14 (STATA Corp., College Station, TX, USA). The study was fully compliant with the PRISMA statement regarding the reporting of systematic reviews and meta-analyses (Liberati et al., 2009) .

A flow chart describing the screening process is presented in Figure 1 . We initially identified 256 studies. A total of 230 studies were excluded because they were either duplicates or irrelevant. After a full-text revision of the remaining 26 articles, seven were excluded because they did not meet the inclusion criteria. Thus, nineteen studies, all conducted in China, were included in the meta-analysis (Table 1) , Cheng et al., 2020 , Liu Q. et al., 2020 J o u r n a l P r e -p r o o f 6 2020, Mo et al., 2020 , Zeng Z. et al., 2020 , Zhang J. J. et al., 2020 , Zhang Q. et al., 2020 . A total of 5,617 COVID-19 patients were studied, 3,723 (49% males, mean age 53 years) with low disease severity or alive during follow-up and 1,894 (63% males, mean age 64 years) with high severity or not surviving during follow up. Fifteen studies were retrospective , Cheng et al., 2020 , Liu Q. et al., 2020 , Liu S. L. et al., 2020 , Zeng Z. et al., 2020 , Zhang Q. et al., 2020 , one was prospective , whilst no information was available in the remaining three , Mo et al., 2020 , Zhang J. J. et al., 2020 . Endpoints included disease severity, based on current clinical guidelines (16 studies) , Liu Q. et al., 2020 , Liu S. L. et al., 2020 , Mo et al., 2020 , Zeng Z. et al., 2020 , Zhang J. J. et al., 2020 , Zhang Q. et al., 2020 or the occurrence of acute respiratory distress syndrome (ARDS, one study) , and survival status (two studies) , Cheng et al., 2020 .

The overall SMD in SAA concentrations between COVID-19 patients with low vs. high severity or survivors vs. non-survivors is shown in Figure 2 . In 18 studies, patients with high severity or nonsurviving status had higher SAA concentrations when compared to those with low severity or alive during follow up (mean difference range, 0.08 to 3.40) , Liu Q. et al., 2020 , Liu S. L. et al., 2020 , Mo et al., 2020 , Zeng Z. et al., 2020 , Zhang J. J. et al., 2020 , Zhang Q. et al., 2020 , although the difference was non statistically significant in two studies (Zeng Z. et al., 2020 , Zhang J. J. et al., 2020 . By contrast, in the remaining study, SAA concentrations were J o u r n a l P r e -p r o o f slightly higher in patients with low severity or alive during follow up (mean difference -0.33) (Cheng et al., 2020) . The pooled results confirmed that SAA concentrations were significantly higher in patients with high severity or non-surviving status (SMD 1.20, 95% CI 0.91 to 1.49, P<0.001) (Figure 2) . Extreme heterogeneity between studies was observed (I 2 =92.4%, P<0.001).

SAA concentrations remained significantly higher (SMD 1.25, 95% CI 0.87 to 1.63, P=0.001; I 2 =92.7%, P<0.001) in patients with high severity or non-survival status after excluding the large study by Yu et al , which accounted for nearly 58% of the overall sample size. Sensitivity analysis, performed by removing each study in turn and re-assessing the pooled estimates, showed that the magnitude and direction of the effect size were not affected (effect size ranged between 1.10 and 1.29) (Figure 3 ). The Begg's (P=0.03), but not the Egger's (P=0.385), test revealed the presence of publication bias. Accordingly, the trim-and-fill method identified four potential missing studies to add on the left side of the funnel plot to ensure symmetry (Figure 4 ).

The adjusted SMD was attenuated but remained significant (SMD 0.89, 95% CI 0.59 to 1.20, P=0.001).

To explore possible contributors to the between-study variance, we investigated the effects of age, gender, end point studied, study design (retrospective or prospective), aspartate aminotransferase 

Sub-group analysis showed that, in the 16 studies in which patients were characterized according to disease severity, the SMD was significantly higher in severe vs. mild disease patients (SMD 1.39, 95% CI 1.09 to 1.69, P=0.001), albeit with extreme between-study variance (I 2 =90.3%, P<0.001).

Similar results were obtained when considering the 15 retrospective studies (SMD 1.10, 95% CI 0.81 to 1.39, P=0.001), again in the presence of extreme heterogeneity (I 2 =90.9%; P<0.001).

In our systematic review and meta-analysis, SAA concentrations were significantly higher in COVID-19 patients with more severe disease, assessed on clinical grounds or based on the presence of ARDS, and in those who did not survive during follow up when compared to patients with milder forms of the disease or those who survived during follow up. The observed SMD, 1.20, suggest an effect size that is both biologically and clinically relevant (Cohen, 1988) . There was extreme between-study heterogeneity however in sensitivity analysis the overall effect size was not significantly influenced when individual studies were in turn removed. Our analyses, based on the Begg's, but not the Egger's, test revealed the presence of publication bias. Although the trim-and-fill method identified four potential missing studies to add on the left side of the funnel plot to ensure symmetry the adjusted SMD was attenuated but remained significant. The SMD was significantly and positively associated with gender and AST but not with age, end point studied (disease severity, ARDS, or survival status), study design (retrospective or prospective), ALT, DD, Cr, PT, CRP and WBC.

Two previously published systematic reviews and meta-analyses identified a relatively low number of studies investigating SSA in COVID-19 patients. The first, in three studies, reported a weighted mean difference (WMD) between severe and non-severe patients of 43.35 (95% CI 5.85 to 80.85, P=0.020; I 2 =66.7%, P=0.050) (Zeng F. et al., 2020) . The second, in five studies, reported a WMD of 90.45 (95% CI 28.69 to 152.21; I 2 =90.6%, P<0.0001) between severe and non-severe patients J o u r n a l P r e -p r o o f (Akbari et al., 2020) . In both meta-analyses, the small number of studies prevented the conduct of meta-regression analysis to identify clinical and demographic factors potentially accounting for the between-study variance. It is important to highlight that, in contrast with these meta-analyses, we evaluated the effect size by using the SMD instead of the WMD. As previously highlighted by other authors, the SMD is particularly appropriate when different studies investigate separate end points, in our case disease severity and mortality (Faraone, 2008) .

Increases in SAA concentrations can be useful to diagnose inflammatory processes and monitor the response to therapeutic interventions . The magnitude and the speed of the increase in SAA concentrations during the acute phase response is greater than that observed with other inflammatory markers, particularly the CRP. Moreover, SAA concentrations generally return to baseline levels more quickly than the CRP in virtue of the shorter half-life of the former. These characteristics suggest a specific pathophysiological role of SAA, which might complement that provided by other biomarkers, e.g., CRP, in clinical practice (Maury, 1985 , Nakayama et al., 1993 , Takata et al., 2011 . This proposition is further confirmed by the lack of significant associations between the SMD and CRP concentrations in our meta-regression, indicating the potential complementary role of SAA and CRP in informing clinical decisions. An increase in SAA concentrations has been observed in several pro-inflammatory conditions, including infections, particularly viral (Kajiya et al., 2008) , liver disease (Siegmund et al., 2016) , autoimmune disease (O'Hara et al., 2000) , diabetes (Kumon et al., 1994) , obesity (O'Brien et al., 2005) , atherosclerotic cardiovascular disease (King et al., 2011) , amyloidosis (Real de Asua et al., 2014) , and cancer (Biran et al., 1986) . Notably, most of these conditions, particularly obesity, diabetes, cardiovascular disease, liver disease, and cancer, have also been independently associated with significantly worse outcomes in patients with COVID-19 . In addition to its potential role in the pathogenesis of the cytokine storm, it has been recently reported that SAA might also exert procoagulant effects that are mediated by an increase in fibrinogen and a concomitant platelet activation to generate a pro-thrombotic state (Page et al., 2019) . Therefore, pending further J o u r n a l P r e -p r o o f research, acute increases in SAA concentrations might represent an important factor linking proinflammatory and pro-thrombotic pathways. The interplay between inflammation and thrombosis has also been observed in COVID-19, a condition that is often characterized, particularly in patients with the more severe form of the disease, by significant alterations in coagulation and a prothrombotic state (Al-Samkari et al., 2020) .

The extreme between-study heterogeneity and the presence of publication bias represent potential limitations in our study. However, the overall effect size was not significantly influenced in sensitivity analyses. The lack of significant associations between clinical and demographic characteristics, barring gender and AST, and SMD and the persistently high heterogeneity observed in sub-group analyses suggest that other unreported factors might contribute to heterogeneity. Few studies have investigated the association between SAA concentrations and gender. No significant differences between males and females have been reported in healthy populations (Carbone et al., 2020) . In another study, the percent fat mass and the waist-to-hip ratio explained the highest percentage of the variability in SAA concentrations in females and males, respectively (Thorand et al., 2006) . While no studies have specifically investigated the associations between SAA concentrations and AST this relationship might be explained, at least in part, by the involvement of this acute phase reactant in several conditions associated with liver injury (Yuan et al., 2019) .

Potential, unreported, methodological factors contributing to the observed between-study heterogeneity involve the use of different SAA detection methods, based on immuno-based assays, different antibodies against various SAA components, and different calibrators in individual studies . While some of these methodological inconsistencies might have been In conclusion, our updated systematic review and meta-analysis has shown that the presence of relatively high SAA concentrations is significantly associated more severe disease, based on clinical assessment or presence of ARDS, and increased risk of mortality in patients with COVID-19. The measurement of this acute phase reactant, singly or in combination with other clinical and demographic parameters, might be useful for risk stratification and clinical monitoring in this group.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

This study did not require ethical approval as it was a systematic review and meta-analysis of published studies.

The authors declare no conflict of interest.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. 

The role of cytokine profile and lymphocyte subsets in the severity of coronavirus disease 2019 (COVID-19): A systematic review and meta-analysis

COVID-19 and coagulation: bleeding and thrombotic manifestations of SARS-CoV-2 infection

Serum amyloid A activates the NLRP3 inflammasome and promotes Th17 allergic asthma in mice

Operating characteristics of a rank correlation test for publication bias

Serum amyloid A (SAA) variations in patients with cancer: correlation with disease activity, stage, primary site

Quantifying, displaying and accounting for heterogeneity in the meta-analysis of RCTs using standard and generalised Q statistics

Serum amyloid A in healthy subjects: assessment of reference value using ELISA method

The predictive value of serum amyloid A and C-reactive protein levels for the severity of coronavirus disease 2019

Longitudinal hematologic and immunologic variations associated with the progression of COVID-19 patients in China

Prognostic value of serum amyloid A in patients with COVID-19

Statistical Power Analysis for the Behavioral Sciences

The cytokine-serum amyloid A-chemokine network

A novel simple scoring model for predicting severity of patients with SARS-CoV-2 infection

Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis

Interpreting estimates of treatment effects: implications for managed care

The value of serum amyloid A for predicting the severity and recovery of COVID-19

Quantifying heterogeneity in a meta-analysis

How COVID-19 induces cytokine storm with high mortality

Toll-like receptor 2 expression level on monocytes in patients with viral infections: monitoring infection severity

Serum amyloid A in atherosclerosis

Serum amyloid A protein in patients with non-insulin-dependent diabetes mellitus

Serum Amyloid A is a biomarker of severe Coronavirus Disease and poor prognosis

Contribution of acute-phase reaction proteins to the diagnosis and treatment of 2019 novel coronavirus disease (COVID-19)

The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration

Longitudinal characteristics of lymphocyte responses and cytokine profiles in the peripheral blood of SARS-CoV-2 infected patients

Associations between serum amyloid A, interleukin-6, and COVID-19: A cross-sectional study

Expressions of SAA, CRP, and FERR in different severities of COVID-19

Comparative study of serum amyloid A protein and C-reactive protein in disease

Serum amyloid A is a predictor for prognosis of COVID-19

Induction of hepatic synthesis of serum amyloid A protein and actin

Monitoring both serum amyloid protein A and C-reactive protein as inflammatory markers in infectious diseases

Diet-induced weight loss is associated with decreases in plasma serum amyloid a and C-reactive protein independent of dietary macronutrient composition in obese subjects

Acute-phase serum amyloid A production by rheumatoid arthritis synovial tissue

Serum amyloid A binds to fibrin(ogen), promoting fibrin amyloid formation

Systemic AA amyloidosis: epidemiology, diagnosis, and management

Serum Amyloid A Induces Inflammation, Proliferation and Cell Death in Activated Hepatic Stellate Cells

Drug treatments for covid-19: living systematic review and network meta-analysis

Funnel plots for detecting bias in meta-analysis: guidelines on choice of axis

Kinetics of c-reactive protein (CRP) and serum amyloid A protein (SAA) in patients with community-acquired pneumonia (CAP), as presented with biologic half-life times

Sex differences in the relation of body composition to markers of inflammation

Assessing the influence of a single study in the meta-analysis estimate

The kinetics and magnitude of the synergistic activation of the serum amyloid A promoter by IL-1 beta and IL-6 is determined by the order of cytokine addition

Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range

Correlation analysis between disease severity and clinical and biochemical characteristics of 143 cases of COVID-19 in Wuhan, China: a descriptive study

The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses

Suppressed T cell-mediated immunity in patients with COVID-19: A clinical retrospective study in Wuhan

Clinical progression and changes of chest CT findings among asymptomatic and pre-symptomatic patients with SARS-CoV-2 infection in Wuhan, China

Innate and acquired immunity intersect in a global view of the acute-phase response

Novel biomarkers for the prediction of COVID-19 progression a retrospective, multi-center cohort study

Serum amyloid A levels in patients with liver diseases

Serum amyloid P aids complement-mediated immunity to Streptococcus pneumoniae

Association of inflammatory markers with the severity of COVID-19: A meta-analysis

Serum-soluble ST2 as a novel biomarker reflecting inflammatory status and illness severity in patients with COVID-19

Clinical characteristics of 140 patients infected with SARS-CoV-2 in Wuhan, China

Clinical analysis of risk factors for severe COVID-19 patients with type 2 diabetes

Acute phase reactant serum amyloid A in inflammation and other diseases

Serum Iron Level as a Potential Predictor of Coronavirus Disease 2019 Severity and Mortality: A Retrospective Study

Comorbidities and the risk of severe or fatal outcomes associated with coronavirus disease 2019: A systematic review and meta-analysis

J o u r n a l P r e -p r o o f

 (Zeng Z. et al., 2020