Potential diagnostic value of pleural fluid cytokines levels for tuberculous pleural effusion

Patients with tuberculous pleural effusion (TPE) or malignant pleural effusions (MPE) frequently have similar pleural fluid profiles. New biomarkers for the differential diagnosis of TPE are required. We determined whether cytokine profiles in the PE of patients could aid the differential diagnosis of TPE. 30 patients with TPE, 30 patients with MPE, 14 patients with empyema (EMP) and 14 patients with parapneumonic effusion (PPE) were enrolled between Dec 2018 and 2019. The levels of interleukin (IL)-6, IL-18, IL-27, CXCL8, CCL-1 and IP-10 were determined in PE by ELISA along with measurements of adenosine deaminase (ADA). The best predictors of TPE were combined ADA.IL-27 [optimal cut-off value = 42.68 (103 U ng/l2), sensitivity 100%, specificity 98.28%], ADA [cut off value 27.5 (IU/l), sensitivity 90%, specificity 96.5%] and IL-27 [cut-off value = 2363 (pg/ml), sensitivity 96.7%, specificity 98.3%, p ≤ 0.0001]. A high level of IL-6 [cut-off value = 3260 (pg/ml), sensitivity 100%, specificity 67.2%], CXCL8 [cut-off value = 144.5 (pg/ml), sensitivity 93.3%, specificity 58.6%], CCL1 [cut-off value = 54 (pg/ml), sensitivity 100%, specificity 70.7%] and IP-10 [cut-off value = 891.9 (pg/ml), sensitivity 83.3%, specificity 48.3%] were also predictive of TPE. High ADA.IL-27, ADA and IL-27 levels differentiate between TPE and non-TPE with improved specificity and diagnostic accuracy and may be useful clinically.


Introduction
Mycobacterium tuberculosis (Mtb) is one of the oldest and most important human pathogens and infection with Mtb has a high rate of mortality worldwide (1). Nearly one-third of the world's population is asymptomatically (latently) infected with tuberculosis, and about 3 to 10 percent of these people progress to active disease throughout their life. In 2018, ten million people became infected with tuberculosis and 1.5 million died which including 0.3 million people con-infected with HIV. (https://www.who.int/tb/global-report-2019).
Tuberculosis has two forms in human being based on affected organs, one pulmonary and second is extra-pulmonary tuberculosis. In extra-pulmonary form many organs may involve but importantly pleural involvement is an important manifestation of the disease (2). Pleura are divided into a parietal layer and visceral layer. The parietal layer lines the inner aspect of the chest wall whilst the visceral layer covers the interlobar ssures. These two layers are separated by a cavity that contains 1-10 ml of uid (3). Pleural effusions (PEs) are an accumulation of uid between the pleural layers (4) and are a clinical problem induced by several etiologies such as local diseases of the pleura and resulting from increased pressure to the lung, organ dysfunction, systemic diseases, pulmonary infections, pleural tumor metastasis and tuberculous pleurisy (5).
Differential diagnosis of TPE from other pleural effusions, especially malignant pleural effusion (MPE), is challenging clinically (6). TPE and MPE are both of lymphocytic source (2). The gold standard for differentiating TPE from other pleural effusions with different etiologies is the isolation Mtb from either pleural uid or pleural biopsy (100% speci city) (7). Although culturing of sputum has a diagnostic value with 100% speci city, it is time consuming and delays the diagnosis. Manifestations of granuloma (~95%), provided that other causes of granulomatosis are discounted, is also used to diagnose tuberculous pleurisy but is considered as an invasive approach (8).
TPE is a delayed hypersensivity reaction to Mtb and mainly the result of pathological immune response associated with increasing in the cytokines, including interleukins (ILs) and chemokines (9)(10)(11). Based on previous studies (12)(13)(14)(15)(16) we hypothesized that cytokine and chemokine levels in TPE may differentiate this disease from other causes of PE. Thus, in the current study, we measured the levels of adenosine deaminase (ADA), IL-6, IL-18, IL-27, CXCL-8, CCL-1 and IP-10 in the pleural uid and shown that ADA, IL-6, IL-27, combined ADA-IL-27, CXCL-8, CCL-1 and IP-10 levels are signi cantly higher in TPE than in other PE containing groups. In addition, we report that levels of ADA, IL-27 and combined ADA.IL-27 have the best sensitivity and speci city to predict the diagnosis of TPE.

Patient selection
The study protocol was approved by the Institutional Review Board for human studies of clinic center from Massih Daneshvari Hospital, Tehran, Iran.
The study was carried out in accordance with the approved Ethics (Ethic code: IR.SBMU.MSP.REC.1397.584). From December 2018-December 2019, 190 consecutive patients with pleural effusions of unknown causes were enrolled. Patients were referred to the Infection wards of the Massih Daneshvari Hospital from across Iran.
Inclusion criteria included: a) no invasive procedures to the pleural cavity, b) not receiving anti-tuberculosis therapy, and c) not suffering from lung trauma for three months prior to hospitalization. At the time of sampling, none of the patients received antibiotic therapy, anti-tuberculosis drugs, anti-malignancy treatments, corticosteroids, or non-steroidal anti-in ammatory drugs. Patients without a clear diagnosis and patients with more than one possible etiology of effusion, and with a pleural transudate, hemothorax or chylothorax were excluded from study.
Thirty HIV negative patients, aged 18-84 yr., with a positive Mtb test in biopsy specimens and pleural tissue granuloma were enrolled as TPE. 30 patients, aged 32-80 yr., newly diagnosed with malignant cells in their pleural uid and/or on pleural biopsy specimen were enrolled in the MPE group. Based on histologically analysis, 15 cases were adenocarcinoma, 10 patients had squamous cell carcinoma (SCC) and 5 patients suffered from nonsquamous cell carcinoma (NSCC). Empyema (EMP) was con rmed in 14 patients, aged 20-75yr by the presence of frank pus in their pleural effusions or smear or positive bacterial or fungal culture of pleural uid (except for Mtb). 14 patients with parapnemonia (PPE) were also included based on PE ndings including glucose <60 mg/dl; pH<7, LDH>1000 and no organisms found in staining culture of effusion.

Sample collection and processing
The pleural uid (5 mL) was collected in heparin containing tubes, by thoracocentesis within 24h of hospitalization and immediately placed in ice.
Tubes were centrifuged at 1200 x g for 5 min and mononuclear cells isolated by Ficoll-Hypaque gradient (Pharmacia, Uppsala, Sweden) within 1h. Total and differential cell counts, protein, lactate dehydrogenase (LDH), ADA, glucose, cytology, and bacterial examination were evaluated in the biochemistry laboratory of the Masih Daneshvari Hospital. In addition; the cell-free supernatants of pleural uid were frozen at -80 °C immediately after centrifuge for later determining concentrations of cytokines by ELISA.

Measurement of cytokines and chemokines
The concentrations of IL-6 and IL-27 were measured by enzyme linked immunosorbent assay (ELISA) (R&D SYSTEM, Minneapolis, MN, US). The concentrations of IL-18, CCL-1, and IP-10 were measured by ELISA (Invitrogen by Thermo Fisher Scienti c, Vienna, Austria). CXCL-8 was measured by ELISA (BD Biosciences, CA, USA) according to the manufacturer's protocol.

Statistical analysis
Analysis was performed using SPSS version 16.0 (SPSS, Inc. Chicago, USA) and GraphPad Prism software (version 6; 07 GraphPad Software, Inc.). Non-parametric Mann-Whitney U test (Median, 95% con dence intervals (CI) was used for the non-normally distributed variables and a t-test (Mean ± SEM) used for normally distributed variables. Receiver operating characteristic (ROC) curve analyses were used to evaluate the capacity of ADA and other biomarkers to differentiate TPE from non-TPE. The area under the ROC curve (AUC) was calculated, and 95% con dence intervals (CIs) were used to test the hypothesis that the AUC is 0.5. An optimum cut-off value was established by using Receiver operating Curve (ROC). P-values < 0.05 were considered as statistically signi cant.

Results
190 patients were enrolled in this study with 145 subjects providing PE exudates and 45 subjects with PE transudate (Fig.1A, 1B). 102 samples were excluded due to failure to meet the diagnostic criteria (30), transudate effusions (45) and exudates with miscellaneous etiology (27) (Fig. 1B). A total of 88 patients were included in the current study and classi ed in 4 diagnostic groups: TPE, MPE, PPE and EMP. The demographic data of the patients and their biochemical characteristics are given in Tables 1-3. The distribution of the cytokines and chemokines in each group of subjects are summarized in Table 4.
Cytokine levels in tuberculous and non-tuberculous pleural uid.
The concentrations of ADA, IL-27, combined ADA.IL-27, IL-6, CXCL-8, CCL-1and IP-10 in the TPE and non-TPE (MPE, EMP, PPE) groups were detected in the pleural uid. We found that these cytokines in the TPE group were higher than the respective levels in the non-TPE groups ( Table 4). The concentration of IL-18 in the TPE group was not signi cantly different from non-TPE groups ( Table 4).

Discussion
In this study, we demonstrated that the levels of ADA, IL-27, ADA.IL-27, IL-6, CXCL-8, CCL-1 and IP-10 in PE were elevated in patients with TPE compared to non-TPE subjects including empyema, parapneumonia, and malignancy. In contrast, IL-18 levels were similar between TPE and non-TPE subjects.
We also described the predicted value of the former cytokines and chemokines for differentiating between TPE and non-TPE diagnosis. We identi ed speci c cut-off levels that gave good-excellent discrimination between causes of PE and among the markers studied, ADA.IL-27 had the highest and IP-10 had the lowest diagnostic accuracy. The predictive ability of IL-27 was similar to that of ADA. The sensitivity and speci city of ADA.IL-27 was higher than that of IL-27 and ADA alone and improved the diagnostic capability.
TB, malignancy, cardiovascular disease and infections can also result in PE. Although several methods including biochemical tests, cytology, bacterial culture and biopsy examination are used to determine the cause of PE in many cases the etiology remains ill-de ned (17). Current diagnostic tests for Mtb within pleural uid have variable results. For example, the positive rate for a smear test for bacillus tuberculosis is about 10% but can be increased to 20-30% by cultivation. The diagnostic value increases further to 70-80% using pleural biopsies (8). Thus, developing a simple test that enables the differential diagnosis of PE-causing diseases.
Various analytes within PE are currently being studied as potential biomarkers of disease etiology including ADA, LDH, CRP and IFN-g. For example, 40 IU/ml ADA in PE has a sensitivity (81-100%) and speci city (83-100%) for TPE (18). Reducing the ADA cut-off value to >35 U/mL results in a lower sensitivity (93%) and speci city (90%) for diagnosing TPE (19). In addition, several studies have shown that IFN-γ levels are elevated in TB pleurisy (20)(21)(22). In a recent meta-analysis the sensitivity and speci city of IFN-γ for the diagnosis of TPE were 89% and 97% respectively (22). In addition, PCR methods have been used to track tubercle bacilli in pleural effusion at the early stage. However, the range reported for a positive test is between 12-100% (23). Due to these diverse results and the time taken to culture Mtb it is clear that new discriminatory biomarkers are needed.
IL-27 is a member of the IL-12 cytokine family. TB enhances the production and secretion of IL-27 from antigen presenting cells (APCs) increasing its local concentration (24). Previous studies have shown the high diagnostic value of IL-27 cytokine for the diagnosis of TPE (2,16,25) with higher levels of IL-27 in TPE than in non-TPE patients (16,26). Wu and colleagues reported that the diagnostic accuracy of IL-27 was even higher than seen with IFN-γ or with ADA with a sensitivity of 95% and speci city of 97.6% (16). A meta-analysis of IL-27 studies in TPE gave a pooled sensitivity of 0.92 (95% CI, 0.90-0.95) with a speci city of 0.90 (95% CI, 088-0.92) (26). In the current study, we found an equally high sensitivity and speci city of IL-27 for the diagnosis of TPE. IL-6 is a pleiotropic cytokine with broad-ranging immune effects (27). Importantly, IL-6 promotes IFN-γ production during TB infection in mice (28).
Increased levels of IL-6 have previously been reported in TB pleurisy compared to non-TB pleurisy subjects (29) which supports our ndings. In that earlier study the sensitivity and speci city of IL-6 in predicting TPE compared to non-TPE were 90.6% and 76.5% respectively (29). Furthermore, Xirouchaki and colleagues reported that IL-6 levels were signi cantly higher in TPE patients than in PPE subjects (30). In the current study, the level of IL-6 in TPE group was higher than in non-TPE groups, and sensitivity and speci city were 100% and 67.24% respectively.
IL-18 is an IFN-g factor (31) which, often acts in concert with IL-12 (32). Increased pleural concentrations of IL-18 have been reported previously in TPE (31). In a meta-analysis study, the sensitivity and speci city of IL-18 for the diagnosis of TPE were 87% and 92% respectively (14). However, in our study we found no difference in the levels of IL-18 in TPE and non-TPE patients and subsequently a much lower sensitivity and speci city which may suggest genetic differences between Chinese and Iranian subjects or the clinical de nitions used. Interestingly, IL-18 was able to differentiate between TPE and MPE.
CXCL-8 causes chemotaxis and localization of neutrophils and lymphocytes within the pleural space in man (33). CXCL-8 was found previously to be higher in patients with empyema and parapneumonia compared to cases of non-infectious effusions (MPE and TPE) (34)(35)(36). For example, Ceyhan and colleagues reported that CXCL-8 levels were signi cantly higher in empyema/parapneumonic effusions groups rather than TB (34). However, Dlugovitzky and co-workers showed statistically higher CXCL-8 levels in TPE patients compared with PPE patients (37). In the current study, CXCL-8 levels were statistically higher in the TPE group compared to non-TPE patients. This re ects the important role of CXCL-8 in stimulating T cells and participating in the granulomatous formation process (37) but also the need to get better quanti cation and validation in larger cohorts. CCL1 or I-309 is a monocyte attractant (38) and has been implicated in the formation and maintenance of granuloma following Mtb infection (39).
CCL-1 levels were signi cantly raised in the plasma of patients with pulmonary tuberculosis (PTB) (40). In this study, we show for the rst time that this chemokine can be used for the diagnosis of TPE with a sensitivity and speci city of 100% and 70.69%.
IFN-γ inducible protein (IP-10) is produced by antigen presenting cells in response to Mtb-speci c antigens in patients with active TB (41,42). IP-10 levels in TPE patients were reported previously to be signi cantly higher than in non-TPE patients groups with a sensitivity (80%) and speci city (82%) for the diagnose of TPE using a cut-off point of 28170pg/ml (43). We report similar ndings in the current study with IP-10 levels in TPE patients being higher than those in non-TPE subjects.
This study shows for the rst time, that CCL-1 was a good predictor of TPE compared to non-TPE subjects with 100% sensitivity and a speci city of 70.7%. However, its diagnostic value was less than that of ADA, IL-27 and combined ADA.IL-27. There were some limitations to this study including the restricted number of patient samples and the analysis of only four PE disease groups. Our study indicates an excellent predictive value for IL-27 (97.72%) and ADA (94.3%), a high sensitivity and speci city for the differential diagnosis of TPE. In addition, we show an even greater accuracy of ADA.IL-27 as a predictive marker of TPE (98.86%). Our data demonstrated that soluble mediators obtained from PE samples may provide high levels of sensitivity and speci city for pleural diseases which may be applicable for the development of a rapid and non-invasive diagnostic test.
In summary, we have demonstrated that CCL-1 is signi cantly higher in TPE patients compared to non-TPE patients and is a good differentiator between clinical groups. Furthermore, combining ADA.IL-27 improves the predictive value of ADA and IL-27 on their own and could be useful as a supporting biomarker in the differential diagnosis of TPE.

Declarations
Ethics approval and consent to participate: The study was approved by the Ethics Committee of the Dr. Masih Daneshvari Hospital, and all patients gave signed informed consent (ethic code: IR.SBMU.MSP.REC.1397.584).

Consent for publication:
All authors have read the manuscript and consent to publication in the Journal Availability of data and material:

Not applicable
Competing interests: The authors con rm that there are no competing interests.

Funding:
This study was supported by internal funding. Values are presented as mean ± standard error of the mean (SEM). TPE = tuberculous pleural effusion, MPE = malignant pleural effusion, EMP = Empyema, PPE = Parapneumonic. †Comparisons of data between TPE, MPE, EMP and PPE effusions were performed using one-way analysis of variance.   ¥Comparisons of data between TPE and the other groups were performed using Mann-Whitney U test. † Values are shown as mean ±SEM. Comparisons of data between TPE and the other groups were performed using student t-test.