CXCL13 expression in the prognostic assessment of human gliomas was explained from the bioinformatics analysis of the TCGA low-grade glioma and glioblastoma (GBMLGG) dataset, which is composed of TCGA brain low-grade glioma (LGG) and glioblastoma multiforme (GBM) datasets. Astrocytoma, oligodendroglioma, and oligoastrocytoma were included in the LGG group, while the GBM group consisted of patients with glioblastoma multiforme. The level 3 data of exon expression and DNA methylation were downloaded by the University of California Santa Cruz (UCSC) Xena browser (https://xenabrowser.net). Glioma cases without exome sequencing or methylation data were excluded. 695 available cases were included in the CXCL13 exon analysis, and 681 samples were identified as qualified samples for CXCL13 methylation analysis.

CXCL13 exon profiling was measured using the Illumina HiSeq 2000 RNA sequencing platform, provided by the TCGA Genome Characterization Center at the University of North Carolina. Four exons were included in this dataset: chr4:78432907–78432942, chr4:78526978–78527083, chr4:78528857–78528989, chr4:78531768–78531848. The transcription estimate at the exon level is shown by the RPKM value (reads per kilobase of exon model per million mapped reads). The average RPKM value from the four exons is reported as CXCL13 expression. CXCL13 DNA methylation profiling was performed using the Illumina Infinium Human Methylation 450 platform by the Johns Hopkins University and University of Southern California TCGA genome characterization center. Four methylation probes were included in this dataset: cg17001652, cg12020230, cg01134794, and cg06662476. DNA methylation beta value of each array probe is a continuous variable between 0 and 1, recorded through Bead Studio software, that is, the intensity ratio between the methylated beads and the binding site. CXCL13 methylation status was evaluated by the average beta value of four methylation probes.

The analysis was conducted retrospectively. To investigate whether prognosis would be related to the patient outcome, 112 astrocytoma patients were enrolled from the Cancer Center of Kaohsiung Medical University Hospital (KMUH); then, the astrocytoma was classified into grades II, III and IV with clinicopathological parameters, age, gender, WHO grade, tumor size, recurrence, survival rate and IDH1 mutation being selected. The authors reviewed the surgical pathology reports of patients diagnosed with astrocytoma, which included available data for tumor tissue and clinical follow-up while clinical and pathological data were obtained from the cancer registry and medical records. The study protocol was approved by the Institutional Review Board of KMUH (KMUHIRB-E(I)-20190188).

3 μm paraffin-embedded sections were de-paraffinized in xylene and dehydrated through the grading alcohol. Then, antigen retrieval was performed in 0.1 M citrate buffer (pH 6.0) at 121°C for 10 min. 3% hydrogen peroxide (H₂O₂) was used to block endogenous peroxidase activity and incubated for 5 min at room temperature. Sections were incubated with primary antibodies, CD163 (NCL-L-CD163, Leica Biosystems, United Kingdom), CXCL13 (MAB801, R&D Systems, Abingdon, United Kingdom), for 1 h at room temperature. Antigen-antibody complexes were visualized by DAKO REAL Envision detection system, peroxidase/DAB, rabbit/mouse (Dako, Glostrup, Denmark). Finally, sections were counterstained with hematoxylin and mounted. DoubleStain IHC Kit: Mouse and Rabbit on human tissue (DAB and AP/Red) (ab210059, Abcam, Cambridge, MA, United States) was used to assess the levels of two different antigens from the same tissue in immunohistochemical staining.

The proportion of stained cells and the staining intensity of CXCL13 and CD163 were utilized to evaluate immune activity. The expression of CXCL13 and CD163 were evaluated by the degree of immunopositive cytoplasm and plasma membrane. CXCL13 staining was evaluated by the ratio score of positively stained cells (0, none; 1, <10%; 2, 10–50%; 3, >50%), and the intensity score of staining intensity (0, no staining; 1, weak; 2, medium; 3, strong). The product of the proportion of stained cells and staining intensity is the staining index. The total score ranges from 0 to 9, where 0–4 is defined as low expression and 6–9 is defined as high expression (Supplementary Table S1). CD163 expression was assessed by the average of stained cell frequency (0, <10 cells; 1, 10–49 cells; 2, 50–100 cells; 3, >100 cells) and distribution (perivascular or scattered in the parenchyma) in five randomly selected high power fields (×400 magnification) (49). The semi-quantitative scoring system includes four categories: 0, 1, 2 and 3, comparisons were made by two categories of CD163 staining scores: low expression (including the above 0 and 1) and high expression (including the above 2 and 3) (Supplementary Table S1).

Descriptive statistics estimated for the study population included means with corresponding standard deviations (SD), medians with corresponding ranges, and proportions, together with 95% confidence intervals (95% CI). Associations between target proteins (CXCL13 and CD163) and clinicopathological parameters were analyzed by chi-square test; the strength of the relationship between two variables were measured using Pearson’s correlation; survival and hazard functions were illustrated by Kaplan–Meier survival curve; and survival was compared between groups by two-sided log-rank test. Cox proportional hazard model was used to examine risk factors related to survival after adjusting for other factors, with risk factors including gender, age, grade, and recurrence. Statistical analyses were performed with SAS 9.3 (SAS Institute, Cary, NC, United States).

Public database analysis from the TCGA GBMLGG cohort was used to evaluate the CXCL13 expression in varying degree of malignant glioma. The prognostic potential of CXCL13 gene in gliomas was verified by 529 LGGs and 166 GBMs in exon expression profiles and DNA methylation patterns. As described in the method section, the four exons and four methylation probes of CXCL13 are used for gene expression analysis and conducted by UCSC Xena (Figures 1A,B). Bioinformatics analysis revealed that CXCL13 exhibited greater expression in the GBM group (chr4:78526978–78527083:+, chr4:78528857–78528989:+, chr4:78531768–78531848:+, all p < 0.0001, Welch’s t-test; Figure 1A). On the contrary, lower CXCL13 methylation was observed in GBMs (all p < 0.0001, Welch’s t-test; Figure 1B).

Further, the associations of CXCL13 expression and methylation status with overall survival were analyzed by Kaplan–Meier curves. CXCL13 expression is divided into low (−) and high (+) groups, depending on the median of exon expression profiles (0.0755) and methylation profiles (0.8231). CXCL13 as an available prognostic biomarker for glioma outcome was illustrated by Kaplan-Meier curves (exon, p = 0.0002; methylation, p < 0.0001; Figures 1C,D). High level of CXCL13 was strongly correlated with age (p < 0.0001), WHO grade (p < 0.0001), patient survival (p = 0.0012, Supplementary Table S2). CXCL13 methylation was associated with age (p < 0.0001), grade (p < 0.0001), patient survival (p < 0.0001), and IDH1 mutation (p = 0.0309; Supplementary Table S2).

A total of 112 cases of astrocytoma were enrolled in this study, including 45 females and 67 males (Table 1). The average age is 50.61 ± 17.73 years, and the median age is 52.50 years (range, 20–83 years). A subgroup of tumor size is defined relative to 2 cm and the mean tumor size ±SD, 2.85 ± 2.00 cm. The mean (±SD) follow-up time of this cohort was 22.66 (±19.43) months. Forty patients died during the follow-up period. Notably, 45 (40.18%) patients had the highest-grade carcinomas (WHO grade IV), 51 (45.54%) patients had IDH1 mutations, and 57 (50.89%) patients had recurrence.

CXCL13 and CD163 expressions in astrocytoma tissues were displayed by immunohistochemical staining. Immunoreactivity of CXCL13 and CD163 was determined based on cytoplasmic and plasma membrane staining (low or high expression of CXCL13, Figures 2A or B, respectively; low or high expression of CD163, Figures 2C or D, respectively). Significantly high levels of CXCL13 and CD163 have been observed in some malignant tissues (Figures 2B,D). In CXCL13/CD163 double staining, CXCL13 was stained red in the cytoplasm of both tumor cells and TAMs, whereas CD163 was visible in brown plasma membrane staining of M2 macrophages. Intense M2 infiltration was observed in high-grade astrocytoma tissue and surrounding neovasculature by double staining (Figures 2E,F). The significant increases in CXCL13, CD163, and CXCL13/CD163 immunoreactivity of astrocytoma tissues were found in the apparently aggressive GBM subgroups (CXCL13, p = 0.0002; CD163, p < 0.0001; CXCL13/CD163, p < 0.0001; Figures 2G–I). Table 1 shows the associations of CXCL13, CD163, and CXCL13/CD163 phenotypes in 112 astrocytoma patients with various clinicopathological parameters. Increased level of CXCL13 was detected in 43 (38.39%) patients, and it was strongly correlated with grade (p = 0.0002), patient survival (p = 0.0017), and IDH1 mutation (p = 0.0008; Table 1). High expression of CD163 was associated with tumor grade (p < 0.0001) and IDH1 mutation (p = 0.0314; Table 1). There were positive associations between CXCL13/CD163 and grade (p < 0.0001), patient survival (p = 0.0007), and IDH1 mutation (p = 0.0002; Table 1). When CD163 and CXCL13 coexisted, it significantly affected patient survival, which was consistent with the hypothesis that CXCL13 showed enhanced M2 regulation of tumor immune escape. Besides, the results of correlation coefficients indicated CXCL13 had strong correlation with CD163+ M2 distribution (p = 0.0013) and IDH1 mutation (p = 0.0007; Table 2). Pearson correlation also confirmed that CD163 expression also correlated with IDH1 mutation (p = 0.0315; Table 2).

The Kaplan-Meier survival curve was used to evaluate the relationship between CXCL13, CD163 and patient survival of astrocytoma. Poor prognosis of the patients were found in the high performance of CXCL13, CD163 and CXCL13/CD163 (p = 0.0039, p = 0.0227, and p = 0.0002; Figure 3). Univariate and multivariate logistic analyses were used to observe the independent prognostic clinicopathological indicators of survival in astrocytoma (Table 3). The results of univariate logistic analysis showed that high expression of CXCL13 was significantly associated with poor overall survival (HR = 1.897, 95% CI: 1.218–2.954, p = 0.0046; Table 3). Multivariate Cox regression analysis including age, tumor size, gender, and recurrence indicated that high level of CXCL13 may not be a significant predictor of OS (HR = 1.569; 95% CI, 0.974–2.527; p = 0.0642; Table 3). Furthermore, CXCL13/CD163 phenotype revealed significant association between co-expression of these two proteins and patient outcomes in univariate analysis (HR = 2.369, 95% CI: 1.491–3.764, p = 0.0003; Table 3). After adjusting for parameters such as gender, age, grade, and recurrence, CXCL13/CD163 coexpression was also regarded as an independent prognostic indicator of patient survival in astrocytoma (HR = 1.682, 95% CI: 1.032–2.740, p = 0.0368; Table 3).