A meta-analysis of Th1 and Th2 cytokine profiles differentiating tuberculous from malignant pleural effusion

To clarify the predominance of Th1 or Th2 immune responses in malignant and tuberculous pleural effusion (MPE and TPE, respectively), we performed a meta-analysis of previously published results of the levels of Th1/Th2 cytokines associated with these two types of pleural effusion to evaluate the use of Th1/Th2 cytokine profiles in distinguishing TPE from MPE. We searched the PubMed and EMBASE databases for studies indexed from 2000 to March 2021. We included studies that (a) diagnosed TPE and MPE based on culture or pleural tissue biopsy and that (b) compared levels of Th1/Th2 cytokines between TPE and MPE. Pooled data based on a random-effects model or fixed-effects model and standardized mean differences (SMDs) across studies were used to compare TPE and MPE. We also performed Egger’s test to assess publication bias. Of 917 identified studies, a total of 42 studies were selected for the meta-analysis. Compared with MPE subjects, TPE subjects had a significantly higher level of TNF-α [2.22, (1.60–2.84)], an elevated level of IFN-γ [3.30, (2.57–4.40)] in pleural effusion, a situation where the Th1 immune response dominated. Conversely, the levels of interleukin-4 (IL-4) and IL-10 (Th2 cytokines) were higher in the MPE subjects than in the TPE subjects, showing statistically nonsignificant tiny effects [−0.15, (−0.94 to 0.63) and −0.04, (−0.21 to 0.12), respectively]. We confirmed that TPE, a situation in which the Th1 cytokines are predominant. The slight preponderance of Th2 cytokines in MPE, which is not convincing enough to prove.

Immune responses mediated by either the Th1 or Th2 subset dominate depending on different types of pleural effusion. The Th1-dominated immune response is considered an important factor in the containment of mycobacteria, and the main immune effector mechanism involves classically activated macrophages that arise in response to Th1 cytokine signals 8 . Th1 cytokines have been shown to predominate in immunity to lung tuberculosis (TB) 9,10 . Notably, TPE is considered a common localized form of extrapulmonary TB. However, whether the mechanism of localized immune response in TPE is dominated by the Th1 or Th2 subset remains to be investigated. MPE, on the other hand, has been associated with Th2 cytokine predominance 11,12 . Although some reports have shown a bias towards Th2 predominance in MPE, some reports have not 13 .
Therefore, in this study, we performed a meta-analysis of all available studies to quantitatively evaluate the Th1/Th2 cytokine profiles in TPE and MPE as well as to assess the potentially diagnostic value of these cytokines in discriminating TPE from MPE.

Statistical analysis.
The outcomes were all continuous variables: Th1/Th2 cytokine levels in TPE and MPE.
Data provided as the mean and SD were extracted. Data provided as the median and IQR were transformed to the mean and SD according to the method of S.P. Hozo (2005) 15 . We used standardized mean differences (SMDs) and 95% confidence intervals (CIs) to assess the levels of Th1/Th2 cytokines in TPE compared to MPE. The Cochran's Q test, I 2 statistic and P value were used to assess the heterogeneity. A P ≥ 0.1 or I 2 ≤ 50% was considered as no significant heterogeneity, and the fixed-effects model (FEM) was applied; otherwise, the randomeffects model (REM) was used. We used the sensitivity analysis to assess the robustness. Subgroup analyses were used to stratify the studies by covariables including country, study design, etiology, publishing year and quality scores based on meta-regression, in order to explore potential sources of heterogeneity. The software Rev-Man 5.2 and STATA version 14.0 were used for the statistical analyses and images. Begg's test was used to assess the asymmetry of the funnel plot. p < 0.05 was considered publication bias, and if present, the trim-and-fill method was adopted to determine the influence of publication bias on the results.

Results
Characteristics of the included studies. A total of 917 potentially relevant studies identified from the database were retrieved, of which 43 studies were included in the final meta-analytical processes. The study selection process is detailed in Fig. 1. In addition, we provide the details of the selected studies in Table 1. Considering the large number of included studies with information on TNF-α and IFN-γ, we only included the ELISA results, but other measurement methods [including cytometric bead array (CBA), chemiluminescent enzyme immunoassay (CLEIA), radioimmunoassay (RIA)] were included for assessments of the remaining cytokines. One study was excluded due to CBA measurements of IFN-γ levels 56 . The quality ratings for each study and characteristics of the included studies are presented in Supplementary Methods. Th1/Th2 cytokine levels in TPE and MPE. Analysis of TNF-α. In a pooled analysis of all 19 trials, the results revealed that there was a significant increase in the pleural effusion levels of TNF-α in TPE subjects compared to MPE subjects [2.22, (1.60-2.84, p < 0.00001)] (Fig. 2). However, there was statistically significant heterogeneity among the studies (I 2 95%). Subsequently, sensitivity analysis ( Fig. S1) was conducted by excluding studies one by one at a time, but neither the magnitude nor the direction of the effect size was substantially altered. We showed that the pooled SMD was stable and reliable, implying higher TNF-α levels in TPE. To explore the sources of the heterogeneity, we performed meta-regression analysis based on subgroup stratification by variables, including country, design, etiology, publishing year and quality scores. In the stratified analyses (Table S1), there was no reliable evidence that the above variables represented the main source of heterogeneity. Therefore, we considered that the heterogeneity might be due to differences in ELISA manufacturers and the severity of the diseases selected for study. In this analysis, there was publication bias based on Begg's test (p < 0.05).   S2), with statistically significant between-study heterogeneity (I 2 97%). Subsequently, we performed sensitivity analysis, which indicated that there was considerable variation in one trial 47 . After removing that trial, the remaining results were stable through sensitivity analysis, the REM of pooled SMDs [3.30, (2.57-4.40, p < 0.00001)] (Fig. 3). Our pooled data suggested higher IFN-γ levels in TPE. However, despite a decrease in heterogeneity (I 2 from 97 to 95%) with the removal of that trial, substantial heterogeneity still existed. To explore possible sources of heterogeneity, we further performed meta-regression analysis based on subgroup stratification by country, design, etiology, publishing year and quality scores. In the stratified analyses (Table S2), there was no reliable evidence

Analysis of IL-10.
A meta-analysis of the data from eight trials on IL-10 was performed as shown in Figure  S3a, the REM of pooled SMDs [0.17, (−0.12 to 0.46)]. However, there was statistically significant heterogeneity among the above results (I 2 66%, p < 0.00001). We excluded two studies 48,50 according to the results of the Galbraith radial plot (Fig. S3b), and the heterogeneity was decreased appreciably (I 2 from 66 to 0%) as shown in Fig. 6. On the basis of the 6 remaining trials, the FEM of pooled SMDs [−0.04, (−0.21 to 0.12)] suggested higher IL-10 levels in MPE subjects.
Analysis of IL-6 between TPE and MPE. IL-6 was the most commonly measured inflammatory cytokine. On the basis of 12 trials, the REM of pooled SMDs [3.53, (2.16-4.90)] indicated substantial heterogeneity between studies (I 2 = 98%, p < 0.00001) as shown in Fig. 7. The pooled results confirmed that the levels of IL-6 in TPE patients were higher than those in MPE patients. To assess the stability of the pooled results of the meta-analysis of IL-6, sensitivity analysis was conducted by excluding studies one by one at a time. Neither the magnitude nor the direction of the effect size was substantially altered, indicating that our results were robust. Furthermore, we excluded two trials 22,53 based on the sensitivity analysis (Fig. S4a). After the deletion of these two trials, the remaining results were stable through sensitivity analysis (Fig. S4b) and caused no appreciable change in the pooled SMDs [1.92, (1.13-2.71)], but the heterogeneity was decreased (I 2 from 98 to 94%) (Fig. S4c). To test the robustness of our results, we perform stratified analyses to detect the sources of the heterogeneity in this study as shown in Table S3. Among the 12 trials, 7 presented ELISA measurements, and the remaining five reported other measurements (2 with CBA results, 2 with RIA results, and 1 with CLEIA results). Consequently, we considered  www.nature.com/scientificreports/ that the main heterogeneity arises from differences in measurement methods. However, nonsignificant differences in between-group heterogeneity were found in some stratified analyses, and measurement methods did not indicate additional sources of heterogeneity. In this analysis, there was no publication bias based on Egger's test (p = 0.150, Supplementary Methods).

Discussion
To the best of our knowledge, this is the first large-scale meta-analysis to systematically investigate the levels of Th1/Th2 cytokines in MPE and TPE. Our results showed that the Th1 cytokine profile was predominant in TPE compared to MPE. We confirmed that the immune response in TPE is dominated by Th1 polarization. Although significant heterogeneity was detected, trials with extreme SMDs were removed based on sensitivity analysis and Galbraith radial plot assessment, which diminished the heterogeneity and provided more stable results. Despite the slight bias toward Th2 cytokines in MPE subjects, the pooled SMD was not statistically significant. Exudative pleural effusion results from a vicious loop of interactions between the immune system and a pleural or parenchymal disease like TB infection, malignancy or inflammation. It results in abnormal accumulation of fluid in the pleural space via altering the permeability of the pleural membranes to enhance plasma extravasation. TB and malignant disease are among the most frequent causes of exudative pleural effusions. Th cells are known to play important roles in the pathogenesis of exudative pleural effusion and host defense.
Th1 cells have been observed to have an anti-infectious role in TB 57 . Th1-mediated host immunity inhibits further multiplication of Mycobacterium tuberculosis (Mtb) via the secretion of IFN-γ and other Th1 cytokines, which activate macrophages, promote phagosome maturation to stimulate phagocytosis, and stimulate the production of reactive nitrogen intermediates 58 . TNF-α is essential for stimulating the chemotaxis of inflammatory cells to sites of infection and leads to granulomatous response to containment disease progression 59 . Normally, Figure 6. Forest plot of 6 studies comparing IL-10 levels in TPE and MPE subjects. The standardized mean difference (the pleural effusion levels of IL-10 in TPE subjects minus that in MPE subjects) was estimated by meta-analysis. www.nature.com/scientificreports/ IL-6 is a proinflammatory cytokine, inducing the production of acute phase proteins to promote inflammation. Nevertheless, macrophages infected with Mtb induce the production of IL-6, which inhibits the response of macrophages to IFN-γ, resulting in the inability to eradicate Mtb infection 60 . The increased IL-6 levels also contribute to increased expression of suppressor of cytokine signaling (SOCS) in TB 61 . SOCS1 is regarded as an important mediator that inhibits IFN-γ secretion by macrophages, which in turn hampers the early clearance of Mtb by macrophages 62 . Consistent with the results of other studies, our study found higher levels of IL-6 in TPE subjects, which suggested that IL-6, as an important cytokine, may be involved in the formation of TPE and play an important role in the occurrence and development of pulmonary tuberculosis (PTB). Different from Th1 cytokines, IL-10 is considered an inhibitory and anti-inflammatory Th2 cytokine that negatively regulates IFN-γ-mediated host immunity, limits pathogen clearance in the early immune response to Mtb and mediates long-term chronic infection 63 . The IFN-γ/IL-10 ratio is a useful objective marker of the clinical severity of TB 64 . Additionally, IL-4 is a classic Th2 cytokine that may subvert mycobacterial containment in macrophages by perturbating Th1-related pathways and regulatory T cells (Tregs) 65 .
Early studies have suggested that Th1-and Th2-cell-producing cytokines play important roles in the immune microenvironment of tumors; Th1 cytokines relevant to antitumor immunity have been reported, whereas Th2 cytokines are related to tumor invasiveness and metastasis 66 . Notably, in advanced cancer, a shift from Th1 to Th2 cells in the tumor environment is often observed, and impaired Th1 cell-mediated immunity has been reported to be associated with cancer progression 67,68 . Anthony et al. reported that patients with advanced squamous cell carcinoma had a diminished Th1 antitumor immune response but stronger underlying Th2 immune response 69 . This shift was further demonstrated in malignant effusions 70 . Accordingly, these studies collectively indicate that a shift from a Th1 to a Th2 cell response in the microenvironment of tumors may play a crucial role in the development and progression of tumors. Moreover, evidence indicates that tumor-derived TGF-β-induced overexpression of IL-10 may drive the shift in the Th1/Th2 balance toward a Th2 response and inhibit the Th1 response 71 . The balance of Th1/Th2 cytokines in MPE subjects remains controversial. We speculate that the shift from Th1 to Th2 along with tumor progression may explain the discrepancy regarding whether Th2 cytokines predominate in MPE subjects, but this needs to be studied further.
As future studies clarify the roles of Th1/Th2 cytokine profiles and their effect on the local immune response, new therapeutic approaches to immunotherapy may be developed. Currently, a useful cytokine/anti-cytokine therapy involves supplementation of anti-inflammatory recombinant cytokines as well as neutralizing cytokines or antagonizing receptors by monoclonal antibodies (mAbs). Recombinant IL-2 has been used successfully for cancer therapy and was approved by the FDA for the treatment of metastatic renal-cell carcinoma in 1992 and for metastatic melanoma in 1998. Infliximab, a chimeric anti-TNFα monoclonal antibody, has been demonstrated to be an effective therapy for rheumatoid arthritis, Crohn's disease, ankylosing spondylitis and other autoimmune diseases. Despite these beneficial effects of infliximab treatment in autoimmune diseases, it carries the risk of TB occurrence and latent TB reactivation 72 . Based on the immunosuppressive ability of IL-10 in tumor immunity, therapeutic strategies targeting the IL-10 signaling pathway have evolved considerably in cancer immunotherapy 73 . In addition, overexpressed IL-4 and IL-13 receptors on cancer cells are specific targets for receptor-directed cancer immunotherapy 74 . Our study confirms changes in the Th1/Th2 cytokine profiles of patients with TPE and MPE, which may be helpful for future investigations of cytokine therapies.
Moreover, combining multiple biomarkers has previously been shown to enhance diagnostic accuracy. The combination of TNF-α and adenosine deaminase 2 (ADA 2 ) has been shown to improve the specificity and accuracy of diagnosis for the differentiation of TPE from MPE 29 . Future studies are needed to determine if Th1/Th2 cytokine profiles confer supplementary diagnostic value for pleural effusions.
There are some unavoidable limitations of our study that are deserving of attention. First, significant heterogeneity was observed in some of our results; although stratified analyses were done, we did not identify the source of heterogeneity. Second, some of the trials provided only the median and IQR levels of the cytokines, which we converted to the mean and SD by Hozo et al. 's method, which might lead to slightly skewed data. Third, the sample sizes included in studies on Th2 cytokines were relatively small, and only six trials were included in the pooled results. Fourth, we did not examine the effect of the stages of cancer on cytokine levels in the included studies, which might lead to heterogeneity and deviation in this analysis. Fifth, few studies have reported the primary tumor causing MPE and the histological type of the primary tumor; therefore, future studies should investigate its effect on cytokine levels.

Conclusions
In summary, this systematic review and meta-analysis confirms that pleural effusions caused by TB, a situation in which the Th1 cytokines are predominant. Although MPE appears to be slightly bias toward Th2 cytokines, this finding was not sufficiently convincing to prove. Our findings suggest that these cytokines are likely to serve supplementary diagnostic value for the differentiation of TPE from MPE. Moreover, it is important to know the role of cytokines derived from Th cells on the pathogenesis of diseases and host defense, which may provide new clues for cytokine immunotherapies.

Data availability
The data sets supporting the results of this article are included within the article and its supplementary material.