Relationship between circulating inflammatory factors and glioma risk and prognosis: A meta‐analysis

Abstract Background Inflammatory factors have been considered a significant factor contributing to the development and progression of glioma. However, the relationship between circulating inflammatory factors and glioma risk as well as their prognostic values in glioma patients is still inconclusive. Here, we performed a meta‐analysis to address this issue. Methods Relevant articles were identified through PubMed, EMBASE, the Cochrane Library, Web of Science, Wanfang database, and China National Knowledge Infrastructure (CNKI) from inception to February 2019. The weighted mean differences (WMDs) or standard mean differences (SMDs) with 95% confidence intervals (CIs) were used to describe the predictive ability of the levels of circulating inflammatory factors on glioma risk. To evaluate the prognostic values of the circulating inflammatory factors in glioma, hazard ratios (HRs) with 95% CIs were used. Results Thirty‐one studies comprising 2587 patients were included. The overall analysis showed that increased circulating interleukin‐6 (IL‐6) [SMD 0.81 (95% CI: 0.21‐1.40; P = .008)], interleukin‐8 (IL‐8) [SMD 1.01 (95% CI: 0.17‐1.84; P = .018)], interleukin‐17 (IL‐17) [SMD 1.12 (95% CI: 0.26‐1.98; P = .011)], tumor necrosis factor‐α (TNF‐α) [SMD 1.80 (95% CI: 1.03‐2.56; P = .000)], transforming growth factor‐β (TGF‐β) [SMD 10.55 (95% CI: 5.59‐15.51; P = .000)], and C‐reactive protein (CRP) [SMD 0.95 (95% CI: 0.75‐1.15; P = .000)] levels were significantly associated with glioma risk. On the other hand, our results showed that circulating IL‐6 [HR 1.10 (95% CI: 1.05‐1.16; P = .000)] and CRP [HR 2.02 (95% CI: 1.52‐2.68; P = .000)] levels were highly correlated with a poor overall survival (OS) rate in glioma patients. Conclusion Our results indicate that increased circulating IL‐6, IL‐8, IL‐17, TNF‐α, TGF‐β, and CRP levels are significantly associated with increased glioma risk. Moreover, our meta‐analysis suggests that circulating IL‐6 and CRP may serve as powerful biomarkers for a poor prognosis in glioma patients.


| INTRODUCTION
Primary brain cancer is composed of tumors that originate from within the central nervous system (CNS) and comprises a large number of different kinds of tumors with a benign to malignant status. 1 Gliomas represent approximately 75% of malignant brain tumors in adults 2 and are the most common primary tumor of the CNS. Patients receiving treatment, including aggressive surgery followed by chemotherapy and radiotherapy, still experience an extremely poor clinical outcome due to the highly proliferative and aggressive nature of this tumor. 3,4 Considering the poor clinical outcome after standard treatment, a set of biomarkers or a whole sample profile for the early detection of glioma is currently under intensive investigation. 5 At the same time, using several inflammatory markers to predict patient prognosis has also proven to be a viable strategy. 6 Without a doubt, the identification of these indicators of glioma could provide a variety of advantages, such as the early intervention of treatment, a reduction in morbidity and mortality, and monitoring the progression of treatment. Currently, there are no specific serological biomarkers for gliomas, which suggests that the development of an assessment for the association between circulating indicators and the risk of glioma as well as their prognostic values in glioma is extremely indispensable.
Inflammatory factors have been considered a significant factor contributing to the complexity and lethality of glioblastoma multiforme (GBM), which can evade immune surveillance by creating an immunosuppressive tumor microenvironment. 7 The expression of a few genes in connection with inflammation was previously demonstrated as an important contributing factor in glioma development, including tumor promotion and progression. 8 Therefore, as essential mediators linking inflammation and cancers, inflammatory factors are attracting a large amount interest of researchers who are searching potential therapeutic and prognostic biomarkers for cancers. Previous studies have shown that several circulating markers of inflammation, such as interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and C-reactive protein (CRP), were observably higher in patients with glioblastoma compared to healthy controls. 9 Accumulating evidence has also shown that these markers may be involved in the risk and survival of glioma. However, the association between circulating inflammatory factors and glioma risk or prognosis remains controversial. Some research has suggested that CRP and IL-6 could effectively predict clinical outcomes in patients with glioma. 6,[10][11][12][13] In contrast, other studies showed no association between these factors and the prognosis of patients with glioma. 10,[14][15][16] To elucidate the essential relationship between circulating inflammatory factors and the risk of glioma as well as their prognostic values in glioma in clinical practice, we performed this meta-analysis.

| Retrieval strategy
Potentially relevant articles were searched using electronic databases and manual retrieval. We systematically retrieved PubMed, EMBASE, the Cochrane Library, Web of Science, Wanfang database, and China National Knowledge Infrastructure (CNKI) for articles published up to February 2019. There were no restrictions of languages or sources through an initial search of published articles. The following retrieval terms were used: "cytokines," "interleukins," "C-reactive protein (CRP)," "tumor necrosis factor (TNF)," "transforming growth factor beta (TGF-β)," "blood," "serum," "plasma" and "glioma." We manually searched the reference lists of related major studies and reviews for additional citations. Moreover, the references of the retrieved articles were screened for other possibly eligible reports.

| Selection criteria
Eligible studies had to meet the following criteria: (a) the patients in the studies were diagnosed with glioma by pathologists; (b) the studies aimed to explore the different levels of circulating inflammatory factors between glioma patients and controls or the prognostic value of circulating inflammatory factors in glioma; (c) for studies analyzing the relationship between preoperative circulating inflammatory indicators and the risk of glioma, the mean value and standard deviation (SD) of the circulating inflammatory factors' levels for glioma patients and controls were provided directly or could be converted from the materials given in the articles; for prognostic studies, the hazard ratio (HR) and 95% confidence interval (CI) of the circulating inflammatory factors' levels for overall survival (OS) were supplied directly or could be converted from survival data or the Kaplan-Meier curve given in the articles using the methods of Tierney et al 17 ; and (d) for duplicate data reported by the same author, the most complete and recently published study was reviewed. 18 The exclusion criteria were as follows: (a) studies measured the levels of circulating inflammatory factors after treatment or measured tissue cytokines; (b) comments, reviews, clinical guidelines, letters to the editor, or case reports; and (c) incomplete data in the original studies made it impossible to obtain the mean value, SD, HR, and its 95% CI. Due to the lack of relevant articles (less than two), some inflammatory factors, such as interleukin-2 (IL-2), interleukin-5 (IL-5), interleukin-7 (IL-7), interleukin-16 (IL-16), and interleukin-18 (IL-18), were also excluded from the meta-analysis.

| Data extraction
All information and data were extracted from each relevant article by two investigators independently. Any further inconsistencies were addressed by a joint discussion. The following study characteristics were recorded from each eligible study: name of the first author, publication year, country, number of patients and controls, study participants' age, gender, sample type, WHO grade, and inflammatory factors assay method. Furthermore, we extracted the cutoff point, maximum follow-up time, and type of HR from each survival study. For studies analyzing the relationship between inflammatory indicators and the risk of glioma, we extracted the outcome information, including the mean value and SD of circulating inflammatory factors' levels for glioma patients and controls. For prognostic studies, the HR estimated with its 95% CI for OS was extracted. When the mean value and SD were not supplied directly in the study, we contacted the authors for unpublished data or extracted them using the methods of Hozo et al 19

and Wan et al 20 If
HRs and their 95% CIs were not supplied directly in the studies, we also contacted the authors for additional data or calculated them from the survival curves using the methods of Tierney et al 17

| Quality assessment and statistical analysis
The methodological quality of the eligible studies was assessed using the Newcastle-Ottawa Scale (NOS), 21 which concentrated on patient selection, comparability, and exposure factor/outcomes. The total NOS scores ranged from 0 to 9, and a score ≥ 6 was considered high quality. 22 The weighted mean differences (WMDs) or standard mean differences (SMDs) with 95% CIs were used to describe the predictive ability of the circulating inflammatory factors' levels on glioma risk. To evaluate the prognostic value of circulating inflammatory factors in glioma, HRs with 95% CIs were used in the meta-analysis. Heterogeneity was estimated by Cochrane's Q test and I 2 measurement among the studies, and multiple individual studies were considered to have moderate or high heterogeneity when P ≤ .10 or I 2 ≥ 50%. WMDs, SMDs, and HRs were pooled using a random effects model if heterogeneity was significant (I 2 ≥ 50%). Otherwise, we used the fixed effects model. Subgroup analyses were used to explore the potential sources of heterogeneity. A sensitivity analysis was performed to evaluate the availability and reliability of the results. Funnel plots with Begg's test and Egger's test were used to research publication bias. 23 The effects of circulating inflammatory factors on the risk and survival of glioma patients were considered statistically significant if the two-tailed P-value was < .05. The quantitative meta-analysis was conducted using STATA version 12.0 (StataCorp LP).

| Characteristics and quality evaluation of the included studies
The detailed retrieval processes of relevant articles are demonstrated in Figure 1. Based on the selection criteria, a total of 2763 articles were identified by electronic databases and manual retrieval in the primary search. After excluding duplicate articles, there were 2304 remaining studies. Next, 2185 irrelevant articles were excluded after reading the titles and/or abstracts. A total of 119 full-text articles remained for further evaluation and a final decision, and 88 articles were excluded due to the following reasons: 43 did not assess the relationship between the levels of circulating inflammatory factors with gliomas or their prognostic values in glioma, 33 had no statistical data, five were not full-text articles, three were reviews, two had a limited number of studies (These two articles studied the relationship between IL-7 and IL-18 and gliomas. But there were fewer than two studies about IL-7 or IL-18, the meta-analysis could not be performed), and two were articles reporting duplicate data. Finally, 31 studies that met the requirements were included in the final meta-analysis.
On the other hand, only eight studies concentrating on the prognostic significance of circulating inflammatory factors in glioma patients were included. The prognostic values of IL-6 and CRP in glioma were examined in five 6,10,11,14,15 and four studies, 10,12,13,16 respectively. The characteristics and quality evaluation of these prognosticrelated studies are presented in Table 2. Of these eight articles, one study reported the prognostic value of not only IL-6 but also CRP. In these included studies, all outcome observations were based on OS; four provided the original HR, and four reported the Kaplan-Meier curves. The publication dates ranged from 2011 to 2018, and the maximum follow-up time ranged from 28 to 154 months. For IL-6, five studies with 535 glioma patients were enrolled in the prognostic analysis. Regarding CRP, four studies including 466 glioma patients were selected for the prognostic analysis. On NOS scoring, all eight studies scored at least 7.

| Circulating inflammatory factors and glioma risk
A total of 10 studies reported an association between IL-6 and the pathogenesis of glioma. The pooled SMD was 0.81 (95% CI: 0.21-1.40; P = .008), indicating that the level of IL-6 ( Figure 2A) in the peripheral blood of glioma patients was significantly higher than that of the normal control group. This result showed that glioma was associated with a higher IL-6 level compared with the control. Because of the limited number of articles included in our study, it is not appropriate to conduct a meta-regression analysis, so subgroup analyses are used to explore heterogeneous sources from regions, methods used for measurement, patient age, and tumor types. In the subgroup analyses of the samples described above (Table 3), the following predicted that elevated IL-6 levels were associated with an increased glioma risk: region in Asia (pooled SMD 0.87 [95% CI: 0. 13 Additionally, glioma patients were significantly confirmed to possess higher circulating IL-8 ( Figure 2C) levels compared to controls, with a pooled SMD of 1.01 (95% CI: 0.17-1.84; P = .018). In the subgroup analyses of regions (Table 3), we found that IL-8 levels in the circulating blood of glioma patients were significantly increased in European populations, with a pooled SMD of 0.41 (95% CI: 0.08-0.74; P = .014). However, Asian studies did not find a significant association between IL-8 and glioma patients (pooled SMD 1.96 [95% CI: −0.49-4.40; P = .118]). Based on the methodological subgroup analysis (Table 3), patients with glioma detected using other methods (pooled SMD 0.66 [95% CI: 0.39-0.92; P = .000]) but not ELISA (pooled SMD 1.31 [95% CI: −0.45-3.08; P = .144]) had significantly elevated levels of IL-8. However, we did not find any association between circulating IL-8 levels and glioma risk in the subgroup analyses of tumor types (Table 3).

| Prognostic significance of circulating inflammatory factors in glioma patients
Five studies with 535 glioma patients were included in the prognostic analysis for IL-6 ( Figure 5A). The pooled HR was 1.10 (95% CI: 1.05-1.16; P = .000), indicating that an elevated circulating IL-6 level predicted poor OS in glioma patients. There was no conspicuous heterogeneity between the included studies (I 2 = 48.6%, P h = 0.100). Regarding CRP ( Figure 5B), four studies including 466 glioma patients  were selected. The expression of circulating CRP levels was significantly correlated with poor OS in glioma patients, and the pooled HR was 2.02 (95% CI: 1.52-2.68; P = .000). Additionally, no significant heterogeneity was found between the included studies (I 2 = 0.0%, P h = 0.622). These results provide more credible evidence for the prognostic significance of circulating IL-6 and CRP levels in glioma patients.

PUBLICATION BIAS
We conducted a sensitivity analysis and examined publication bias of the meta-analysis on the relationship between IL-6 ( Figure 6A) and TNF-α ( Figure 6B) expression levels and glioma risk. No individual study markedly affected the pooled SMD based on the sensitivity analysis, which indicates that the SMD estimates are stable and reliable. Publication bias was assessed using Begg's and Egger's tests. For IL-6, Begg's test (z = 0.36, P>|z| = .721) revealed no evidence of publication bias among the 10 studies ( Figure  7A), and Egger's test (t = −0.14, P>|t| = .891) showed the same conclusion. Regarding TNF-α, Begg's test (z = 1.15, P>|z| = .251) revealed no evidence of potential publication bias among the nine studies ( Figure 7B), and Egger's test (t = 1.80, P>|t| = .115) also showed no bias.

| DISCUSSION
In tumor areas, inflammatory cytokines and their receptors form a comprehensive regulatory network, which plays a Abbreviations: CI, confidence interval; ELISA, enzyme-linked immunosorbent assay; NA, not available; GBM, glioblastoma multiforme; HGG, High-grade glioma; SMD, Standard mean differences.

F I G U R E 3
The forest plots of the association between circulating inflammatory factors and glioma risk. A, Association between CRP and glioma risk. B, Association between TGF-β and glioma risk | 7463 FENG Et al.
significant role in the development and progression of cancer. Currently, inflammatory factors are highlighted in cancer diagnosis, prognosis, and therapy. However, the research conclusions of biomarkers related to gliomas remain unclear. The present meta-analysis summarized the evidence to demonstrate the relationship between circulating inflammatory factors and the risk of glioma as well as their prognostic values in glioma from 31 included articles. The results showed that increased circulating IL-6, IL-8, IL-17, TNF-α, TGF-β, and CRP levels before treatment are significantly associated with glioma risk. Furthermore, no significant association was found between preoperative circulating IL-4, IL-10, IL-12, IL-23, and MCP-1 levels and glioma risk. On the other hand, our results showed a poor prognostic outcome in glioma patients expressing high levels of circulating IL-6 and CRP.
In the tumor environment, inflammatory factors are responsible for cell proliferation, tumor invasion, significant angiogenesis, and the suppression of certain immune functions to promote the occurrence and development of tumors. 46 The tight junction between endothelial cells is the main structural and functional basis of the blood-brain barrier. 47  factors (such as IL-6, IL-8, and TNF-α) in glioma patients that were closely linked to brain tumor behavior. 37 IL-6 is a multifunctional cytokine that can induce the proliferation and differentiation of immune cells. The tumor cells themselves and immune cells surrounding the tumor can release IL-6, thereby increasing circulating IL-6 levels. Elevated levels of serum IL-6 associated with tumor size, tumor stage, or disease progression have been found in many neoplastic diseases, such as gastrointestinal tumors, lung cancer, and malignant melanoma. 51 Trikha et al revealed that the serum IL-6 level correlated with clinical extension of the tumor and with outcomes. 52 In our study, we found that an elevated circulating IL-6 level is significantly associated with an increased risk and a poor prognosis of glioma, probably because it is involved in the occurrence and development of glioma. Much evidence suggests that IL-6 plays a significant role in the development of glioma by promoting angiogenesis and cell proliferation and because of its antiapoptosis and antiradiation effects. 9,53,54 In addition, studies have shown that the serum IL-6 level is positively correlated with the expression of IL-6 protein in glioma tissues, as detected by immunohistochemistry, suggesting that IL-6 may participate in the progression of glioma in an autocrine or paracrine manner. 53,55 Based on the evidence described above, it is easy to understand that an elevated IL-6 level is closely associated with the risk and prognosis of glioma patients.
CRP is a classical nonspecific acute-phase protein produced in the liver. 34 Epidemiological studies have suggested that for several types of solid cancer, elevated circulating levels of CRP are associated with poor outcomes. 13 Based on the particular relationship between tumors and inflammation, a sharp increase in CRP content in the presence of inflammation makes it a reliable biomarker to reflect the relationship between inflammation and cancer. The results from our present meta-analysis showed a high risk of developing glioma with elevated levels of CRP and a poor prognostic outcome in glioma patients expressing high levels of CRP. Possible mechanisms for high CRP levels associated with an increased risk and a poor prognosis of glioma have been reported. 12 On the one hand, CRP protects endothelial cells from starvation-induced death by acting on microglia through IL-1β, which contributes to tumor angiogenesis and progression. On the other hand, IL-6 secreted by GBM acts on hepatocytes and may secrete high levels of CRP through the janus kinasesignal transducer and activator of transcription (JAK-STAT) pathway, which reaches the tumor site through blood circulation and then accumulates in tumor tissue. These phenomena also suggest that our conclusions about IL-6 are credible and accurate.
Moreover, we found that increased circulating IL-8, IL-17, TNF-α, and TGF-β levels are significantly associated with glioma risk, which indicates that these inflammatory factors are involved in the pathogenesis of glioma. IL-8 is a potent angiogenic factor for the progression of malignant gliomas and is correlated with the histopathological grade of gliomas. 56 Studies have indicated that GBM cells, which secrete IL-8, promote angiogenesis and microvascular endothelial permeability. 57 Over the past few years, the possible role of IL-17 in tumors has received increasing attention. The function of IL-17 in tumors may include the promotion of angiogenesis by upregulating vascular endothelial growth factor (VEGF) and CD31 58 , and/or the promotion of tumorigenesis through the IL-6-STAT3 signaling pathway. 59 Hu et al reported that IL-17 may influence glioma tumorigenesis and progression through certain signaling pathways. 39 Through in vivo and in vitro studies, IL-17 was shown to induce the proliferation and migration of glioma cells by activating PI3K/Akt1/NF-κB-p65. 60 Accumulating evidence has shown that the pleiotropic functions of TNF-α range from antitumor activity to tumorigenesis. The characteristic of TNF-α inhibition is that it binds to malignant tumor cells, changes gene expression in cells, destroys the cell cycle, and then causes the tumor cells to no longer proliferate indefinitely. 61 TNF-α secretion leads to the promotion of glioma formation and development through angiogenesis. 62 The expression of TGF-β can be elevated in a variety of malignant tumors and is closely related to the invasive growth, metastasis, and other processes of the tumor. 63 In particular, TGF-β was shown to correlate with the poor prognosis of advanced stages of glioma. 64 In conclusion, circulating IL-6, IL-8, IL-17, TNF-α, TGF-β, and CRP levels may be reliable biomarkers describing the risk of glioma. Furthermore, the circulating IL-6 or CRP level may be a reliable prognostic indicator in glioma patients.
Our results showed no significant association between circulating IL-4, IL-10, IL-12, IL-23, and MCP-1 levels and the risk of glioma patients. The primary reason for this finding may be that the number of studies included was limited, and the sample size of each study was not large. On the other hand, the negative conclusions are closely related to the various characteristics of these inflammatory factors involved in tumor progression. Therefore, further studies with large sample sizes are necessary to establish the true correlation as well as the unique pathogenesis.
In this meta-analysis, there are unavoidable limitations in several aspects that should be further considered. First, the regions and populations involved in our study were limited (only Asia and Europe). Second, the cut-off values for inflammatory factors were different between the included studies, which could increase heterogeneity among studies. Third, because of the lack of studies relating the expression of other inflammatory factors and OS (except for IL-6 and CRP), we could not draw an exact conclusion. Fourth, based on the incomplete data about different pathological types of gliomas in original studies, we did only a few subgroup analyses of glioma risk in different pathological types, and cannot draw a complete conclusion. Finally, although we conducted subgroup analyses for IL-6, IL-8, IL-10, and TNF-α with glioma risk to explore the potential sources of heterogeneity, high heterogeneity still existed in the studies of IL-4, IL-12, IL-17, IL-23, TGF-β, and MCP-1 based on the limited number of included studies.

| CONCLUSIONS
In conclusion, our meta-analysis suggested that circulating IL-6 and CRP levels may serve as powerful biomarkers for a poor prognosis in glioma patients. Determination of the levels of circulating IL-6 and CRP might aid in predicting the clinical outcome in glioma patients. Moreover, our results indicated that increased circulating IL-6, IL-8, IL-17, TNF-α, TGF-β, and CRP levels are significantly associated with increased glioma risk. Additionally, based on the results of subgroup analyses of glioma risk in different pathological types, we could better understand the various characteristics of inflammatory factors involved in tumor progression. To investigate this problem deeply, studies with large sample sizes, covering extensive areas, and with complete data about different pathological types of gliomas are required. Considering that the acquisition and measurement methods of blood samples are relatively easy, we believe that the detection of circulating inflammatory factors and an in-depth study of their mechanisms of promoting tumorigenesis could provide a broader prospect for reducing glioma risk and supplementing methods of predicting the outcome in glioma patients.