Prognostic value of tumor-inltrating lymphocytes in gliomas: A Systematic Review

Background: Glioma is the most common primary brain tumor with poor prognosis. Some studies have learned the prognostic role of tumor inltrating lymphocytes (TILs) in gliomas. But conict conclusions were drawn by these studies. In order to reach an agreement, we systematically performed a meta-analysis. Method: A systematic literature research was conducted on the Web of Science, EMBASE, PubMed, Cochrane Library and China National Knowledge Infrastructure. The eligible articles which met the inclusion criteria were included in our study. The clinical outcomes of included patients were dened as progression-free survival (PFS) and overall survival (OS). The basic characteristics and relevant data were extracted. Hazard ratios (HRs) with 95% condence intervals (CIs) were pooled to evaluate the prognostic role of TILs in gliomas. Newcastle Ottawa Scale (NOS) was used to assess the quality of included studies. Results: Eight articles published from 2008 to 2019 were nally included in our study. And 25 studies were conducted in these articles. We assessed the prognostic role of TILs in gliomas by subgroup analysis according to the subtypes of TILs. The pooled HRs for OS revealed that high density of CD3+ and CD4+ TILs were related to the poor prognosis of gliomas (HR for CD3+ TILs=1.266; HR for CD4+ TILs= 2.128). The pooled HR for PFS indicated that only high density of FOXP3+ TILs were related to poor prognosis (HR=2.785; 95%CI=1.848, 4.197). Conclusion: High density of CD3+ and CD4+ TILs may be a potential candidate for predicting poor OS of gliomas while high density of FOXP3+ TILs may serve as a good biomarker for predicting poor PFS of gliomas.


Background
Glioma is the most common primary brain tumor with poor prognosis 1 . Glioblastoma, which accounts for 16% brain tumors in adults, has a median overall survival of less than 12 months 2 . Medulloblastoma, which is another type of glioma, commonly occurs in children. Over 80% medulloblastoma patients have achieved a ve-year survival with the improvement of treatment 3 . But the quality of survivors is still a major issue because of the substantial neurologic and cognitive sequelae. In addition, ependymoma and neuroblastoma, commonly occurring in children, also have a poor prognosis 3,4 . So it is necessary to identify valid biomarkers for predicting the prognosis of these patients.
Tumor in ltration lymphocytes (TILs), part of complex microenvironments, playan important role in tumor development and growth 5 . Until now, they have been identi ed as good biomarkers in ovarian cancer, breast cancer and non-small cell lung cancer [6][7][8] . Recently, TILs have also gained attentions in the prognosis of gliomas. However, some studies indicated that TILs were good biomarker for gliomas [9][10][11][12][13][14] while some studies revealed that no signi cant correlation was observed between TILs and gliomas 9,10,13,15 . In order to reach an agreement, we systematically reviewed the studies focused on glioma and TILs. And in this study, we assessed the prognostic role of TILs in gliomas by subgroup analysis according to the subtypes of TILs.

Method
Search strategy A systematic literature search was conducted in Web of Science, EMBASE, PubMed, MEDLINE, Cochrane Library and China National Knowledge Infrastructure before May 2020. The key words including tumor-in ltrating lymphocytes, prognosis, glioma and all possible combinations were used to search the study.

Inclusion criteria
The studies were included if they met the following criteria: (1) published as original articles with high quality; (2) tumor tissues were isolated from humans; (3) the prognostic role of TILs in gliomas were studied; (4) containing the essential information to estimate the effects of TILs.

Data extraction
The data was independently extracted by Y.L.S. and J.Z.. The general information, including country, sample number, median age, sex (M/F), stage, diagnosis, TIL subtypes, detective methods, cut-off value, follow-up time, outcomes, HR, 95%CIs, and p-values were extracted from these studies. In addition, the included studies were quanti ed with Newcastle Ottawa Scale (NOS) which was used for the quality assessment in meta-analysis 16 . When HR and 95%CIs were not mentioned in these articles but Kaplan-Meier curves were shown in these articles, the data was extracted and digitized according to Tierney's report 17 . If the results of multivariate analyses and univariate analyses were both shown in the article, the multivariate analyses results were chosen to calculate the pooled HR.
Quality assessment of included studies The Newcastle Ottawa Scale (NOS) was used to assess the quality of included studies. In our study, we assessed the quality of each study included in these article respectively. And the studies with 7 points or more than 7 points were de ned as high quality studies 18 .

Statistical analyses
The pooled HRs were calculated by STATA 11.0 (StataCorp, College Station, TX, USA). And the pooled HR was considered to be signi cant if the p value was less than 0.05 and 95% CI did not overlap 1 19 . The heterogeneity between studies was evaluated by chi-squared test and I 2 statistic, and substantial heterogeneity was de ned as p 0.05 or I 2 50% 20 . The random-effects model was used to calculate the pooled HR whether these included studies had a ne homogeneity because of the differences in treatment strategies and patients characteristics 21 . If the heterogeneity was signi cant, sensitivity analysis was performed to assess the contribution of each study in heterogeneity.

Study selection
Our study was designed, conducted and reported based on the PRISMA statement. The article was developed by the order of guidelines of system reviews.
The process of study selection was shown in Figure 1. Brie y, a total of 323 articles were included for initial evaluation after duplicate remove. And 20 articles were selected by screening the title and abstract. Non-related articles, cases, and reviews were excluded. Eight articles were nally identi ed as eligible studies for further assessment [9][10][11][12][13][14][15]22 .
Basic characteristics of selected studies Eight articles published from 2008 to 2019 were included. And 25 studies were conducted in these articles. The basic characteristics of each study were shown in Table 1

Pooled analysis for FOXP3+ TILs in gliomas
Four studies explored the relationship between FOXP3+ TILs and OS of gliomas, and another four studies explored the relationship between FOXP3+ TILs and PFS of gliomas. The pooled HRs were analyzed respectively and shown in Figure 2 and Figure 3. Different with CD3+ TILs, high density FOXP3+ TILs were associated with short PFS (HR=2.785; 95% CI=1.848-4.197; P 0.001) but was not signi cantly related with OS (HR=1.537; 95% CI=0.961-2.457; P=0.073). No signi cant heterogeneity was observed in the studies which analyzed the relationship between FOXP3+ TILs and PFS of gliomas (I 2 =0%; P=0.861). A signi cant heterogeneity was observed in the studies which explored the relationship between CD3+ TILs and OS (I 2 =78.6%; P=0.003). Different conclusions were draw by sensitivity analysis. And a signi cant heterogeneity still existed no matter which study was excluded.

Pooled analysis for CD8+ TILs in gliomas
Eight studies assessed the prognostic value of CD8+ TILs in gliomas. Among these studies, 5 studies analyzed the relationship between CD8+ TILs and OS and 3 studies analyzed the relationship between CD8+ TILs and PFS. The pooled HRs revealed that no signi cant relationship was observed between CD8+ TILs and prognosis of gliomas (HR for OS= 1.813; 95% CI=0.780-4.214; P=0.786 HR for PFS=1.117; 95% CI=0.501-2.494; P=0.786). And a signi cant heterogeneity were observed in these studies (I 2 = 86.0%; P 0.001). So we performed sensitivity analysis to gure out the origin of heterogeneity. The signi cant heterogeneity still existed after excluding each study, and different conclusions were drawn by the analysis.

Discussion
Glioma is the most common primary brain tumor with poor prognosis 1 . Finding a good biomarker for predicting the prognosis of glioma has always been a hotspot. Recently, more and more attention has been focused on the TILs which were part of complex microenvironments 5,23 . But con ict conclusions were draw by these studies. In order to reach an agreement, we performed this meta-analysis.
Considering the heterogeneity of TILs, we pooled HRs according to the subtypes of TILs. As shown in Fig. 2 and Fig. 3, high density of CD3 + and CD4 + TILs were signi cantly related to poor OS while high density of FOXP3 TILs was related to poor PFS. But CD8 + TILs were not a good predictor for glioma. CD3 which was a marker of most T cells were proved to improve the prognosis in many kinds of tumors [24][25][26] . But in our study, high density of CD3 + TILs was identi ed to be related to poor OS. On the contrary, high density of CD3 + TILs was related to better PFS after excluding the study which contributed to the high inner heterogeneity. The similar conclusion was also drawn by Ding et al. in hepatocellular carcinoma 27 . And the mechanisms were not exactly explained in previous study, which needs to be answered in the future.
CD4 + TILs were considered to be a double-edge sword for tumor development. On the one hand, they helped the activation of cytotoxic T lymphocytes. On the other hand, they dampened the anti-tumor immunity and promoted tumor development 28 . In gliomas, CD4 + FOXP3 + lymphocytes were identi ed to act as a dominant immune escape mechanism 29 . CD4 + FOXP3 + lymphocytes obviously suppressed the activation of naïve T cells. After depleting CD4 + FOXP3 + lymphocytes, the CD8 + anti-tumor cells were detected. In addition, CD4 + FOXP3 + lymphocytes in gliomas were reported to be correlated with angiogenesis which promoted the tumor progression 30 . In agreement with these studies, our study elucidated that high density of CD4 + TILs predicted the poor OS of gliomas. Generally, FOXP3 + TILs were considered to predict poor prognosis because of the immune escape role 29 . Recently, FOXP3 was also veri ed to be expressed in glioma cells by quantitative RT-PCR, immunohistochemistry and FACS analysis 31 . And these might explain the prognostic role of FOXP3 + TILs in gliomas.
Intriguingly, CD8 + TILs were frequently considered to be the pivotal effector of anti-tumor immune. And they were veri ed to improve the OS of several tumors, like breast cancer, colorectal cancer, gastric cancer, and melanoma 20,24−26 . But in this study, no correlation was observed between CD8 + TILs and gliomas. This might be related to blood-brain barriers and the speci city of the tumor environment in brain 32,33 . In addition, the inner heterogeneity of the studies which assessed the role of CD8 + TILs in gliomas was signi cant. So more studies that focused on the role of CD8 + TILs in gliomas are needed in the future.
Although our study gures out three potential markers for predicting the prognosis of gliomas and attracts more attentions on the targets in glioma patients, a signi cant heterogeneity was observed in our study, which might be due to the heterogeneity of gliomas. Previous studies proved that the prognosis of the different gliomas was very different. Limited numbers of studies were included in our study, as well as the subgroup analysis of different subtypes of gliomas. So more studies should be focused on the evaluation of different TILs in different gliomas, which will made the conclusion more convincing.

Conclusions
In conclusion, our study provided evidence for the prognostic role of TILs in gliomas and identi ed CD4 and CD3 for OS of gliomas and FOXP3 for PFS of gliomas. However, due to the heterogeneity among studies, more well-designed studies are warranted in the future.

Abbreviations
TILs, tumor in ltrating lymphocytes; FACS, uorescence activated cell sorting; HR, hazard ratio; CI, con dence interval; OS, overall survival; PFS, progression free survival; Newcastle Ottawa Scale (NOS) Declarations -Ethics approval and consent to participate Not applicable.
-Consent for publication Not applicable.
-Availability of data and materials All data generated or analyzed during this study are included in this published article.

-Competing interests
The authors declare that they have no competing interests.
-Authors' contributions YS and JZ worked together on the conception, data analysis and interpretation, and drafting of the manuscript. They contributed equally to this work. LY contributed to the study design, and manuscript editing. XM supervised development of work, helped to evaluate the manuscript and acted as corresponding author. All authors read and approved the nal manuscript.
-Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional a liations.   Pooled hazard ratios of higher density of TILs for progression free survival in patients with gliomas.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.