Serum activin A as a prognostic biomarker for community acquired pneumonia

Abstract

BACKGROUND

It is essential to develop new biomarker with effective prognostic roles because of the unclear clinical use of the current community-acquired pneumonia (CAP) predictors.

AIM

To evaluate the association between serum activin A levels and prognosis in CAP patients.

METHODS

A total of 168 CAP individuals grouped according to the severity and prognosis of illness condition, and 48 healthy individuals as the control group were enrolled in this study. Circulating concentrations of activin A were measured using enzyme-linked immunoassays. The interaction between activin A levels and etiologies of CAP was determined. Based on the severity of CAP, 110 patients (65.48%) were categorized into group-I, 42 (25%) cases were grouped into group-II, and 16 (9.52%) cases were categorized into group-III.

RESULTS

Serum activin A levels were higher in patients with CAP than controls, but independent of etiology. Moreover, the scores of Pneumonia Severity Index (PSI) and CURB-65 positively correlated with the increasing levels of serum activin A, and were at their highest peak in individuals in group-III (P < 0.001). Combining activin A with CURB-65 or PSI was more effective in improving predictive property (P < 0.01). According to Cox proportional regression analysis, after adjusting clinical parameters, we confirmed that activin A showed a powerful predictive property for hospital mortality in CAP patients (P < 0.001).

CONCLUSION

Higher level of serum activin A was associated with poor prognosis of CAP. Activin A can be used as a more valuable biomarker of prognosis in CAP patients.

Key Words: Community-acquired pneumonia; Activin A; Etiology; Prognosis; Etiology

Core Tip: Serum activin A levels were significantly higher in community-acquired pneumonia (CAP) patients. Higher level of serum activin A was associated with the severity and poor prognosis of CAP. Activin A can be used as a more valuable biomarker of prognosis in CAP patients. Future studies are justified to explore the precise biomolecular functions of activin A in CAP.

INTRODUCTION

Community-acquired pneumonia (CAP) remains to be the most common infectious disease, leading as the fourth cause of death worldwide. CAP is associated with different prognosis ranging from mild symptoms to death, with extreme use of health resources. CAP-related morbidity and mortality is still increasing despite the progress in diagnosis and treatment[1]. Therefore, early detection and assessment of CAP severity promotes reasonable and effective site of care. Pneumonia Severity Index (PSI) and CURB-65 (confusion, uremia, respiratory rate, blood pressure, age ≥ 65 years) were used to determine the severity of CAP[2,3]. However, neither of these scoring criteria was entirely reliable in predictive ability in intensive care unit admission[4]. It is necessary to establish effective biomarkers which can improve the diagnostic power of CURB-65 or PSI. Although C-reactive protein (CRP), a biomarker of inflammation, has value in predicting pneumonia severity, and the non-specific characteristic is regrettable[5]. Procalcitonin (PCT) is also a marker of bacterial inflammation in CAP severity and progression[6]. The prognostic accuracy of PCT is still affected by many factors even though PCT has better sensitivity and specificity than other inflammatory factors. Chun et al[7] found that PCT measurement was influenced in pneumonia patients with acute kidney injury. A single test for PCT is inadequate to fully assess severity and prognosis of pneumonia[8]. Therefore, there’s increasing need of new biomarkers for predicting more accurate severity of CAP in order to improve proper treatments and prognosis.

Serum activin-A, is a major member of the transforming growth factor-β superfamily. Serum activin-A has roles on cell proliferation and differentiation, immune response and angiogenesis[9]. There were evidences that activin-A regulated biological functions in various pulmonary diseases, such as asthma, chronic obstructive lung disease and lung cancer[10-12]. An in vitro study confirmed the proinflammatory role of activin A-induced levels of interleukin 6 (IL-6), and tumor necrosis factor-α (TNF-α)[13]. Conversely, Meade et al[14] reported anti-inflammatory role of activin A mediated suppression of IL-6 signaling. Moreover, increasing levels of activin A in serum and bronchoalveolar fluid were measured in patients with sepsis and acute respiratory distress syndrome (ARDS)[15]. However, the clinical significance of Activin in serum in CAP has not been elucidated.

In this study, we aimed to investigate clinically informative data of serum activin A to assess the prognosis of CAP patients. We measured the concentrations of serum activin A in CAP patients. We further investigated the relationship between activin A and other biomarkers to indicate whether serum activin-A concentration has any diagnostic or prognostic value for CAP.

MATERIALS AND METHODS

Patients and study design

The study was performed between January 2017 and April 2020 in Yixing People’s Hospital. This study was approved by the ethical committee of Yixing People’s Hospital. A total of 168 patients diagnosed with CAP based on criteria, the presence of respiratory symptoms, including a fever > 37.5 °C, purulent sputum, cough, breathing difficulties and abnormal chest radiograph and 48 aged-matched healthy cases were enrolled in this study. The patients were excluded from participating in this study based on the following criteria, as: (1) Age < 18 years; (2) pregnancy; (3) immunosuppressive conditions (prednisone or equivalent dose of > 20 mg daily for > 2 weeks or any immunosuppressants therapy); (4) malignant tumor; and (5) end-stage renal or serious heart or liver disease. In this study, patients with non-severe CAP who had been discharged from the hospital were categorized into group-I (n = 110). Group-II (n = 42) was composed of patients with severe CAP who required noninvasive or/and invasive mechanical ventilation who had been discharged successfully from the hospital. Furthermore, for those patients who died as a result of CAP in the hospital were categorized as group-III (n = 16). The control group included 48 healthy volunteers.

Data collection

Related clinical characteristics and laboratory parameters were collected from peripheral venous blood before antibiotic treatment at standardized conditions including white blood cell count, CRP, PCT, liver and kidney function etc. The CAP etiology was analyzed using the report of sputum bacterial cultures and next-generation sequencing technology with bronchoscope. The PSI and CURB-65 scores according to the international criteria were calculated for all patients.

Serum biomarker measurements

The concentrations of activin A in peripheral venous blood were measured using enzyme-linked immunoassays kits (R&D Systems, United States) according to the recommendations of the manufacturer.

Statistical analysis

Data was presented as the mean ± SE. Comparison of data concerning unpaired parameters between the groups were compared through Student’s t test using Sigma Stat software. A P-value < 0.05 was considered to be of significant difference. The prognostic accuracy of mortality in hospital was predicted using receiver operating curve (ROC) curve. Areas under the curve (AUC) were used to calculate the diagnostic accuracy of biomarkers alone or in combination. Cox survival analysis was used to investigate the independent effects of serum Activin levels on mortality in the hospital.

RESULTS

Characteristics of the enrolled patients

Initially, 168 CAP patients who did not meet the exclusion criteria were enrolled in this study. A total of 110 (65.48%) participants with improved condition met the criteria for group-I. Of the enrolled cases, parts were admitted to mechanically noninvasive or/and invasive ventilated. In between, 42 (25%) patients who were safely discharged from the hospital were categorized as group-II, while 16 (9.52%) patients with poor outcomes, such as death were categorized as group-III. Moreover, 48 healthy volunteers were recruited. Common characteristics of the overall case were shown in Table 1.

Table 1 Characteristics of study patients.

Characteristics	Group-I	Group-II	Group-III	P value
	n = 110	n = 42	n = 16	I vs II	II vs III
Age (years; mean ± SD)	51.36 ± 19.72	58.15 ± 16.91	61.81 ± 22.65	0.1741	0.2013
Sex (M/F)	56/54	27/15	11/5	0.5975	> 0.9999
Smoking, n (%)	72 (65.45)	27 (64.28)	11 (68.75)	0.6012	0.5812
WBC (× 109/L)	10.87 ± 4.02	11.13 ± 4.38	13.01 ± 4.18	0.0673	0.2015
CRP (mg/L)	36.76 ± 28.11	81.23 ± 51.18	132.77 ± 64.83	0.0057	0.0013
PCT (μg/L)	0.77 ± 1.42	1.01 ± 2.12	2.98 ± 3.43	< 0.0001	< 0.0001
PSI score	76.12 ± 28.94	101.3 ± 40.51	121.8 ± 48.21	< 0.0001	0.0002
CURB-65 score				0.0082	0.0013
0-1	73	20	2
2	32	13	5
3-5	5	9	9
Pathogen checked
Bacteria	28	19	6
Virus	11	7	5
Atypical pathogen	31	1	0
Mixed pathogen	15	4	1
Unknown	25	11	4
Activin A (ng/mL)	11.92 ± 8.21	17.61 ± 5.35	25.33 ± 7.37	0.0006	< 0.0001

WBC: White blood cell; CRP: C-reactive protein; PCT: Procalcitonin; CURB-65: Confusion, urea, respiratory rete, blood pressure, and age ≥ 65 years old; PSI: Pneumonia Severity Index.

Activin A in serum and CAP pathogen

Serum activin A levels were significantly higher in CAP individuals (P < 0.01) compared to the control group. Moreover, Figure 1A showed that the expression of activin-A in serum was markedly elevated in group-III than that in group-I and group-II (P < 0.001).

Figure 1 Serum levels of activin A in enrolled individuals. A: Serum activin A level distribution in different multiple groups and healthy volunteers, ^aP < 0.05; B: Serum activin A level distribution in different causative etiologies of community-acquired pneumonia in the study patients.

We classified pathogens into several categories: bacteria, virus and atypical pathogens, mixed pathogens and unknown pathogens to enhance a more detailed understanding of serum activin A in CAP patients. Clinical data of pathogenic distribution in CAP were showed in Table 1. We investigate the levels of activin A in serum with the causative pathogens in order to gain better understanding of the clinical value of activin A in CAP patients. Our data shown in Figure 1B suggests that there were no significant differences of serum activin A in CAP patients with specific etiologies.

Prognostic analysis

We investigated the relevance of activin A concentration changes to illness severity in order to fully recognize the prognostic value of activin A in CAP patients. Results of Table 2 suggest that activin A concentrations were associated with hospital mortality of CAP patients. Moreover, with the increasing of activin A concentration, the incidence of severe CAP (group-II and III) was added (P < 0.001).

Table 2 Areas under the curve and thresholds for predicting hospital mortality in patients with community-acquired pneumonia.

	Threshold	Sensitivity (%)	Specificity (%)	AUC	P value	95%CI
CRP	> 84	80.65	59.12	0.722	0.0008	0.587-0.811
PCT	> 2.31	57.36	87.91	0.682	0.0005	0.601-0.824
CURB-65	> 1	73.66	81.02	0.846	< 0.0001	0.725-0.914
PSI	> 108	76.08	80.36	0.85	< 0.0001	0.748-0.937
Activin A (ng/mL)	> 19	83.27	81.44	0.854	< 0.0001	0.787-0.941

AUC: Areas under the curve; CRP: C-reactive protein; PCT: Procalcitonin; CURB-65: Confusion, urea, respiratory rete, blood pressure, and age ≥ 65 years old; PSI: Pneumonia Severity Index.

PSI and CURB-65 were used as pneumonia severity indicators[16,17]. Our study also indicated the correlation between the severity of CAP patients with the increase of PSI and CURB-65 index (Table 1). It was necessary to analyze whether activin A was relevant to scores of PSI and CURB-65 to better reveal clinical value of activin A-predicting severity of CAP. Our results suggested that concentrations of activin A in serum were positively associated with increasing PSI and CURB-65 (Figure 2) (P < 0.001).

Figure 2 Correlation between serum activin A levels and biomarkers in different groups of community-acquired pneumonia. A: Positive correlation between activin A and CURB-65 scores in patients; B: Positive correlation between activin A and Pneumonia Severity Index scores in patients.

Prognostic effect of activin A in hospital mortality of CAP patients

Our study confirmed that activin A showed the highest AUC for predicting hospital mortality in CAP patients. The AUC of CURB-65 and PSI were 0.846 and 0.85, respectively; CRP and PCT were 0.722 and 0.682. Both CURB-65 and PSI had clinically higher AUC power for hospital mortality than PCT (Figure 3A). Moreover, compared with CURB-65 or PSI alone, the higher AUC of scoring system combined with activin A indicated that it promoted the effectiveness in predicting hospital mortality in CAP patients (Figure 3B and C).

Figure 3 Receiver operating curve analysis for serum activin A concentrations to predict hospital mortality. A: Area under the receiver operating curve [areas under the curve (AUC)] for the biomarkers evaluated in this study; B: AUC of combination of activin A and CURB-65 or CURB-65 alone, P = 0.0005; C: AUC of combination of activin A and Pneumonia Severity Index (PSI) or PSI alone, P = 0.0094.

We used cox proportional regression analysis to evaluate the value of activin A concentrations on hospital mortality in CAP patients. Table 3 showed that activin A concentrations, PSI and CURB-65 scores were associated with hospital mortality. Furthermore, after adjustment of clinical parameters, activin A level in serum played a significantly independent predictor for hospital mortality in CAP patients.

Table 3 Prognostic effect of activin A for hospital mortality in Cox proportional hazards regression analysis.

Variables	Univaritate analysis		Multivaritate analysis
	HR (95%CI)	P value	HR (95%CI)	P value
Age	1.112 (1.000-1.064)	0.0624
Sex	0.881 (0.343-2.106)	0.5751
Smoking	1.113 (0.551-2.471)	0.9931
WBC	1.127 (0.931-1.202)	0.3428
CRP	1.103 (1.012-1.206)	0.0661
PCT	1.045 (1.023-1.057)	0.0232
CURB-65	2.157 (1.512-3.104)	< 0.0001
PSI	1.018 (1.005-1.037)	< 0.0001	1.024 (1.012-1.047)	< 0.0001
Activin A	1.024 (1.011-1.028)	< 0.0001	1.028 (1.014-1.031)	< 0.0001

WBC: White blood cell; CRP: C-reactive protein; PCT: Procalcitonin; CURB-65: Confusion, urea, respiratory rete, blood pressure, and age ≥ 65 years old; PSI: Pneumonia Severity Index.

DISCUSSION

In this study, we found that activin A concentrations in serum are significantly higher in CAP patients compared to those in healthy controls. Further study found that the variation of activin A concentrations in serum was not related to the etiology of CAP. Moreover, patients with serious medical conditions who suffered with noninvasive or/and invasive ventilation or death had increased activin A levels in serum than relatively mild individuals. Additionally, we revealed that the elevated circulating activin A, positively correlated with PSI and CURB-65 scores. This showed the predictive ability of activin A in hospital mortality. The study confirmed increased activin A in serum as a biomarker in CAP patients, predicting poor prognosis.

Currently, CAP-induced mortality in ambulatory and inpatients is approximately 5% and 15%, respectively[17]. Therefore, early assessment of disease severity is critical in lowering CAP-related mortality[18,19]. Several researches were performed to investigate the clinical value of multiple biomarkers, particularly in assessment of clinical severity and prognosis[20,21]. Some contradictory conclusions were proposed due to the sizes of the samples. In our study, we first confirmed that patients with CAP had higher levels of activin A in serum than healthy controls. Additionally, we found that CAP patients with severe cases displayed significantly activin A concentration than that in patients with mild disease. Moreover, dead patients with CAP showed top level of circulating activin A. The etiologies of pneumonia are considered to be closely related to the curative effect. The present study verifies that there is no statistical difference in activin A levels in CAP with different pathologies. These data demonstrate that increased activin A level in serum is linked to the illness severity.

CURB-65 and PSI scoring systems were used to enhance the prediction of the severity and prognosis of CAP in order to facilitate the choice of clinical selection[22]. Elevated scores of CURB-65 and PSI were presented in critical patients with CAP. These scoring systems also had weaknesses. However, these were also some defects of scoring systems in clinical evaluation. Comorbidities of CAP patients were not fully considered using CURB-65 scoring system[23]. Therefore, Fan et al[24] developed an improved CURB-65 scoring system to identify CAP at high risk. Liu et al[25] expanded CURB-65 scoring system including Albumin and platelet to enhance evaluation of CAP severity and predictive efficiency. PSI is relatively time consuming although it is commonly used in clinical evaluation of CAP practice[26]. PSI performs less well in critical treatment in severe patients with CAP because of great demand of parameters[27]. Therefore, proper biomarkers that can effectively predict the severity of CAP and clinical outcome would guide clinical therapy. In this study, we revealed that the activin A levels in serum were positively correlated with PSI and CURB-65 scores in predicting clinical severity of patients with CAP. We introduced a combined evaluation approach to create a multi-biomarker tool in order to modify the unsatisfactory sensitivity and specificity of these scoring systems. Additionally, ROC analysis suggests that the higher AUCs value of combined activin A with scoring systems showed the more valuable capacity of predicting hospital mortality than PSI and CURB-65 scores alone. However, it is more realistic to introduce activin A in serum as a valid biomarker for identifying high-risk initial patients into the severity of CAP. Results of cox proportional hazards regression analysis show that activin A has proved to be an independent predictor of hospital mortality of CAP patients. Based on these results, serum activin A level is a promising biomarker for severity and hospital mortality of CAP, especially in an emergency department setting.

Studies have confirmed that activin A plays an active role in inflammation and tissue repair processes[28,29]. An in vitro study found that activin A-induced inflammation fibrosis was presented in pathological pulmonary tissues[30]. We previously demonstrated that the up-regulated concentration of activin A in serum correlated with TNF-α expression was shown in patients with chronic obstructive pulmonary disease[31]. Activity of activin A may promote the development of respiratory pathology in several aspects. Overexpression of activin A in bronchoalveolar fluid was confirmed in ARDS patients who had the higher mortality possibility[32,33]. Activin A was proved to be a regulator of adaptive immune response in inflammatory actions[34,35]. An in vivo study revealed that increased serum activin A levels were presented before lipopolysaccharide (LPS)-lead TNF-α, interleukin-1β and IL-6 growth in acute inflammation models[36]. Activin A acts as a producer of proinflammatory cytokines. Further research demonstrated that the unsatisfactory outcomes in patients with acute respiratory failure may be related to activin A-caused apoptosis of hepatocytes and B lymphocytes and production of nitric oxide[37]. Apostolou et al[32] found that therapeutic neutralization of Activin-A successfully attenuated ARDS-like pathology induced either by ectopic activin A or by LPS Currently, the exact mechanism by which serum Activin levels are increased in CAP patients remains unclear. It is necessary to explore the precise functions of activin A as a useful biomarker in CAP patients.

CONCLUSION

Serum activin A levels are significantly increased, especially in serious illness of patients with CAP. The levels of serum activin A are not influenced by etiology. Variation trends of serum activin A are positively correlated with CURB-65 or PSI scores. It is more accurate to predict hospital mortality and prognosis by united evaluation approach of serum activin A and CURB-65 or PSI scores.

ACKNOWLEDGEMENTS

We thank Meng-Jie Rui for providing biostatistics support.