Mendelian randomization and mediation analysis reveal the role of immune cells in the pathways between inflammatory factors and glioblastoma

Abstract

Objective

Inflammatory factors and immune cells play crucial roles in glioblastoma multiforme pathophysiology. However, the relationships between these factors and the underlying mechanisms are not fully understood. This study employed Mendelian randomization to investigate the effects of inflammatory factors and immune cells on glioblastoma multiforme risk, specifically focusing on the mediating role of immune cells.

Methods

Genetic data on 91 inflammatory factors (e.g. interleukins, CXCL, and fibroblast growth factor; N = 14,824), 731 immune cell phenotypic traits (e.g. CD39⁺ secreting Treg absolute count, IgD⁻CD38⁻ absolute count, and T cell absolute count; N = 3,757), and glioblastoma multiforme risk (6,183 cases and 18,169 controls) were obtained from a genome-wide association study database. All data were derived from individuals of European ancestry. Inverse variance weighting was employed as the primary Mendelian randomization method to estimate causal effects.

Results

Mendelian randomization analysis revealed significant associations between two inflammatory factors and glioblastoma multiforme risk. Furthermore, 22 immune cell phenotypic traits were associated with glioblastoma multiforme risk. Notably, Mendelian randomization mediation analysis identified two significant mediation pathways: (a) double-negative (CD4⁻CD8⁻) T cells mediate the causal effect of transforming growth factor β1 on glioblastoma multiforme risk and (b) human leukocyte antigen-DR⁺ T cells mediate the causal effect of CXCL10 on glioblastoma multiforme risk.

Conclusion

This study provides genetic evidence supporting the complex interplay among inflammatory factors, immune cells, and glioblastoma multiforme risk, highlighting key mediation mechanisms. These findings offer novel insights into the therapeutic potential of targeting inflammatory factors within the tumor microenvironment to regulate immune cell responses.

Keywords

Inflammatory factors immune cells mediation analysis Mendelian randomization glioblastoma multiforme

Introduction

Gliomas are primary central nervous system tumors originating from glial cells and neurons. They represent the most common type of brain tumor, accounting for approximately 81% of all central nervous system malignancies.¹ According to the 2021 World Health Organization classification, gliomas are categorized into three main subtypes: (a) oligodendroglioma, isocitrate dehydrogenase (IDH)-mutant, and 1p/19q-codeleted; (b) astrocytoma, IDH-mutant; and (c) glioblastoma multiforme (GBM), IDH-wildtype.² Among these subtypes, GBM, a grade IV glioma, is the most aggressive and prevalent,³ characterized by high incidence, extensive invasiveness and rapid postoperative recurrence. GBM has a poor prognosis, with a median overall survival of only 15 months, thus presenting a significant challenge in neuro-oncology.⁴

Recent advances in tumor microenvironment (TME) research have provided valuable insights into GBM pathogenesis and potential therapeutic strategies. The TME is a complex and dynamic system in which bidirectional interactions between tumor cells and their surrounding environment play a crucial role in tumor progression through several mechanisms.⁵ Among the various components of the TME, studies have shown that inflammatory factors and immune cells play a critical role in regulating tumor cells directly and indirectly.⁶ Although inflammatory factors are essential for tissue defense and repair, their overactivation can lead to tissue damage and tumorigenesis, highlighting their dual role in tumor biology. For example, it has been shown that long noncoding RNA (lncRNA)-135528 can inhibit tumor progression by upregulating the expression of the inflammatory factor CXCL10 through activation of the JAK/STAT pathway.⁷ In addition, gain-of-function mutations in TP53 can lead to upregulation of the expression levels of C-C motif chemokine ligand 2 and tumor necrosis factor-α (TNF-α), which can exacerbate inflammation and worsen the prognosis of patients with GBM.⁸ Similarly, immune cells are key regulators of antitumor immunity, mediating tumor surveillance and cytotoxic responses. Although studies suggest an association between inflammatory factors and GBM risk, the precise causal relationships among inflammatory factors, immune cells, and GBM risk as well as their underlying mechanisms are not fully understood.

Previous studies on the association between inflammatory factors and GBM have been mostly observational and are susceptible to reverse causation (e.g. the tumor itself triggering inflammation) and confounding factors (e.g. obesity and infections). Mendelian randomization (MR) is a robust epidemiological approach that leverages genetic variants as instrumental variables (IVs) to infer causal relationships between exposure factors and disease outcomes.⁹ By utilizing single-nucleotide polymorphisms (SNPs) as IVs, MR mimics the design of randomized controlled trials, minimizing confounding bias. Recent studies have shown that abnormalities in the development of glial cells lead to gliomas.¹⁰ This provides a theoretical basis for identifying risk factors for GBM through MR. Moreover, genetic variants precede the disease, thereby avoiding reverse causation. This strengthens the reliability of evidence supporting a causal association between inflammatory factors and GBM risk.^11–13 In our study, levels of inflammatory factors may fluctuate with disease progression or treatment. In contrast, MR using genetic variants can provide a stable signal of lifetime exposure. To elucidate the causal relationship among inflammatory factors, immune cells, and GBM risk, this study employed inverse variance weighting (IVW) to identify inflammatory factors and immune cell phenotypic traits associated with GBM risk. Furthermore, mediation analysis revealed the crosstalk among inflammatory factors, immune cells, and GBM risk, highlighting the bridging role of immune cells in this causal pathway. This finding deepens our understanding of glioblastoma pathogenesis and provides potential targets for precise immunotherapy.

Methods

Study design

The aim of this study was to investigate the causal relationship between inflammatory factors and immune cells in the pathogenesis of GBM and further explore the mediating role of immune cells in the relationship between inflammatory factors and GBM risk through mediation analysis, thereby clarifying the inflammatory factor–immune cell–GBM axis. To ensure the validity of the MR analysis, IVs were selected based on the following three fundamental assumptions: (a) IVs must be significantly associated with the exposure; (b) IVs must not be associated with confounding factors or the outcome; and (c) IVs must influence the outcome solely through their effects on the exposure (Figure 1).

Figure 1.

Graphical abstract.

Data source

Data on 91 inflammatory proteins were obtained from the Olink Target-96 inflammation panel across 11 cohorts, encompassing a total of 14,824 individuals of European descent (registration numbers from GCST90274758 to GCST90274848).¹⁴ Genome-wide association study (GWAS) data on immune cell phenotypic traits were derived from a cohort study on the east-central coast of Sardinia, Italy. The study initially included 6602 participants from the general population (57% females and 43% males) with an age range of 18–102 years. These participants were screened for immune profiling, resulting in 3757 volunteers with complete immunophenotypic data for this GWAS analysis (registration numbers from GCST90001391 to GCST90002121).¹⁵ Summary statistics for 731 immune cell phenotypes, including morphological parameters (MPs), absolute cell count (AC), relative cell count (RC), and median fluorescence intensity (MFI), were retrieved from the GWAS catalog. These phenotypic traits were further classified as follows. MFI, AC, and RC included B cells; complement-dependent cytotoxicity (CDC); T cell maturation phases; monocytes; myeloid cells; T cells, B cells, and natural killer cells (TBNK); and Treg cells. Furthermore, MPs included CDC and TBNK. Finally, GWAS data for GBM risk were obtained from a large-scale meta-analysis of 6183 GBM cases and 18,169 European ancestry controls.¹⁶ The abovementioned data are all from European populations, which effectively avoids bias due to population structure.

This study was conducted in accordance with the Declaration of Helsinki (1975), as revised in 2024. All data were derived from publicly available GWAS datasets; therefore, ethical approval and patient consent were not required for this study. The reporting of this study conforms to the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines.¹⁷

Selection of IVs

First, in this study, a genome-wide significance threshold (P < 5 × 10⁻⁸) was used to identify significant SNPs associated with GBM risk and inflammatory factors. As only a small number of SNPs reached significance in the association analysis between certain inflammatory factors or immune cell traits and GBM risk, we applied a more lenient threshold (P < 1 × 10⁻⁵). In addition, the same threshold has been used in recent studies.^18–20 Second, SNP selection criteria were set at an r² value of <0.001 and a genetic distance of 10,000 kb to minimize linkage disequilibrium (LD) and ensure IV independence. Third, the strength of each IV was assessed using the F-statistic, calculated as F = R² × (N − 2)/(1 − R²), which evaluates the correlation between IV and exposure. An F-statistic of >10 was considered indicative of a strong IV, minimizing weak instrument bias.²¹

Statistical analysis

Statistical analyses were performed using R software (version 4.3.1) and the Two-Sample MR package. To evaluate the causal relationships among inflammatory factors, immune cell phenotypes, and GBM risk, five MR methods were applied: IVW, MR-Egger, weighted median, simple mode, and weighted mode. Among these methods, IVW was considered the primary analytical method due to its superior accuracy in estimating causal effects.²² The IVW approach applies a random-effects model. Statistical significance was defined as a P-value of <0.05 for the IVW estimate, with consistent results across the other four methods reinforcing causal inference.

To assess the robustness of the results, multiple sensitivity analyses were performed. Heterogeneity among SNPs was evaluated using Cochran’s Q test, with P < 0.05 indicating significant heterogeneity.²³ Horizontal pleiotropy was evaluated using MR-Egger intercept and MR-PRESSO.^24,25 Leave-one-out analysis was conducted to identify any SNP that disproportionately influenced the MR results.

To control the rate of false-positive results due to multiple testing, the Bonferroni correction method was used in this study.²⁶ According to the correction, the significance thresholds were set at 0.000549 (0.05/91) for inflammatory factors and 0.000068 (0.05/731) for immune cells. Two-sided P values less than this threshold were considered statistically significant, while results with P values between 0.05 and the corrected threshold were considered suggestive of potential associations.

A two-step MR approach was used for mediation analysis to assess whether immune cells mediated the causal relationship between inflammatory factors and GBM risk.²⁷ First, inflammatory factors and immune cell phenotypes exhibiting significant causal associations with GBM risk were identified. Second, the causal relationship between these inflammatory factors and immune cells was examined using the same MR criteria to determine whether immune cells mediated the effects of inflammatory factors on GBM risk (Figure 2). The mediation ratio of immune cells was calculated as the indirect effect divided by total effects (β1 × β2/β-all), where β1 represents the effect of inflammatory factors on immune cells, β2 represents the effects of immune cells on the outcome, and β-all represents the total effects of inflammatory factors on the results. The results were considered logically consistent if β-all was positive when both β1 and β2 were either positive or negative. Conversely, if β-all was negative, either β1 or β2 had to be positive, while the other was negative.

Figure 2.
Study diagram. GWAS: genome-wide association study; SNP: single-nucleotide polymorphism.

Results

Causal effects of inflammatory factors on GBM

MR analysis demonstrated significant causal associations between two inflammatory factors—transforming growth factor (TGF)-β1 and CXCL10—and GBM risk (Table 1). To confirm the robustness of these findings, consistency across all five MR methods was required. The results showed that elevated TGFβ1 levels were associated with an increased risk of GBM (odds ratio (OR) = 1.1949; 95% confidence interval (CI): 1.0070–1.4178; P = 0.0413), whereas higher CXCL10 levels were associated with a decreased risk of GBM (OR = 0.7620; 95% CI: 0.6540–0.8880; P = 0.0005). After applying Bonferroni correction, CXCL10 remained significantly correlated with GBM risk. These causal relationships were visually represented using scatter plots (Figure 3). Sensitivity analyses revealed no significant heterogeneity or horizontal pleiotropy (Table S1), confirming the reliability of these associations. Additionally, leave-one-out analysis (Figure 4) confirmed that no single SNP significantly altered the causal estimates, reinforcing the robustness of the findings.

Table 1.
MR results of the relationship between inflammatory factors and GBM.

Exposure Outcome nsnp β SE P OR OR_lci95 OR_uci95

CXCL10 GBM 23 −0.2718 0.0780 0.0005 0.7620 0.6540 0.8880

TGFβ1 21 0.1781 0.0873 0.0413 1.1949 1.0070 1.4178

CI: confidence interval; GBM: glioblastoma multiforme; MR: Mendelian randomization; nsnp: number of single-nucleotide polymorphisms; OR: odds ratio; TGF: transforming growth factor; lci95: lower limit of the 95% confidence interval; uci95: upper limit of the 95% confidence interval.

Figure 3.
Scatter plot of two inflammatory factors in glioblastoma multiforme. MR: Mendelian randomization; SNP: single-nucleotide polymorphism; TGF: transforming growth factor.

Figure 4.
Leave-one-out sensitivity analysis of inflammatory factors in glioblastoma multiforme. MR: Mendelian randomization; TGF: transforming growth factor.

Reverse MR analysis

To validate the direction of causality, a reverse MR analysis was performed to examine the potential causal effects of GBM on TGFβ1 and CXCL10 levels. This step was crucial for eliminating the possibility of reverse causation and ensuring the reliability of subsequent mediation analyses. The results (Table S2) showed no significant causal effect of GBM on TGFβ1 or CXCL10, reinforcing the initial findings.

Causal effects of immune cells on GBM risk

To investigate the role of immune cells in GBM pathogenesis, MR analysis was performed on 731 immune cell phenotypic traits. Consistency across the five MR methods was again required to confirm the robustness of the findings. The IVW method identified 22 immune cell traits with significant causal associations with GBM risk (Table 2, Figure 5). Among these identified immune cell traits, 8 traits, including CD39⁺ resting Treg% CD4 Treg, human leukocyte antigen (HLA)-DR⁺ T cell, myeloid dendritic cells (DC) %DC, CD4 on secreting Treg, double-positive (DP) (CD4⁺CD8⁺) T cell, CD20⁻ B cell, central memory (CM) double-negative (DN; CD4⁻CD8⁻) %DN, and transitional %lymphocyte, were associated with a decreased risk of GBM, while 14 traits, including HLA-DR on CD14⁻CD16⁺ monocyte, CD33⁺ HLA-DR⁺ CD14dim %CD33⁺ HLA-DR⁺, naive CD4⁺ %T cell, CD80 on CD62L⁺ myeloid DC, CD28⁺ DN (CD4⁻CD8⁻) %T cell, CD33br HLA-DR⁺ CD14dim myeloid leukocyte, CD24 on IgD⁺ CD38br, CD4 on activated and secreting Treg, CD8 on CD28⁺CD45RA⁺ CD8br, CD4 on activated Treg, CD80 on plasmacytoid DC, CD80 on CD62L⁺ plasmacytoid DC, HLA-DR on CD33br HLA-DR⁺CD14⁻, and naive CD8br T cell, were associated with an increased risk of GBM. However, no immune cells remained significantly correlated with GBM after applying Bonferroni correction. Scatter plots corroborated these findings (Figure S1). Sensitivity analyses revealed no significant heterogeneity or horizontal pleiotropy for most immune cell traits, except for naive CD8br T cell (Table S3). Leave-one-out analysis (Figure S2) confirmed that no single SNP significantly altered the causal estimates. The observed heterogeneity and horizontal pleiotropy in naive CD8br T cell were attributed to methodological differences across studies.

Table 2.
MR results of the relationship between immune cell phenotypic traits and GBM.

Exposure Outcome nsnp β SE P OR OR_lci95 OR_uci95

CD20⁻ B cell GBM 15 −0.0940 0.0418 0.0246 0.9103 0.8387 0.9881

CD24 on IgD⁺ CD38br 19 0.0651 0.0314 0.0380 1.0672 1.0036 1.1349

CD28⁺ DN (CD4⁻CD8⁻) %T cell 20 0.0739 0.0347 0.0329 1.0767 1.0060 1.1524

CD33⁺ HLA-DR⁺ CD14dim %CD33⁺ HLA-DR⁺ 17 0.0969 0.0410 0.0182 1.1018 1.0166 1.1940

CD33br HLA-DR⁺ CD14dim myeloid leukocyte 16 0.0667 0.0319 0.0366 1.0689 1.0042 1.1379

CD39⁺ resting Treg% CD4 Treg 21 −0.0312 0.0144 0.0300 0.9693 0.9424 0.9970

CD4 on activated and secreting Treg 20 0.0595 0.0264 0.0239 1.0614 1.0079 1.1177

CD4 on activated Treg 18 0.0568 0.0278 0.0409 1.0584 1.0024 1.1177

CD4 on secreting Treg 16 −0.0821 0.0338 0.0152 0.9212 0.8622 0.9843

CD8 on CD28⁺CD45RA⁺CD8br 15 0.0593 0.0282 0.0352 1.0611 1.0041 1.1213

CD80 on CD62L⁺myeloid DC 18 0.0799 0.0345 0.0205 1.0832 1.0124 1.1589

CD80 on CD62L⁺plasmacytoid DC 16 0.0548 0.0279 0.0492 1.0564 1.0002 1.1157

CD80 on plasmacytoid DC 16 0.0549 0.0279 0.0491 1.0565 1.0002 1.1159

CM DN (CD4⁻CD8⁻) %DN 12 −0.1075 0.0445 0.0157 0.8981 0.8231 0.9799

DP (CD4⁺CD8⁺) T cell 14 −0.0922 0.0401 0.0217 0.9120 0.8430 0.9866

HLA-DR on CD14⁻CD16⁺ monocyte 9 0.0998 0.0433 0.0212 1.1049 1.0150 1.2027

HLA-DR on CD33br HLA-DR⁺ CD14⁻ 15 0.0410 0.0190 0.0312 1.0419 1.0037 1.0815

HLA-DR⁺ T cell AC 25 −0.0524 0.0227 0.0210 0.9490 0.9077 0.9921

Myeloid DC %DC 21 −0.0682 0.0305 0.0253 0.9340 0.8798 0.9916

Naive CD4⁺ %T cell 22 0.0834 0.0322 0.0097 1.0870 1.0204 1.1579

Transitional %lymphocyte 17 −0.1233 0.0488 0.0116 0.8840 0.8033 0.9728

Naive CD8br T cell 15 0.2041 0.0939 0.0298 1.2265 1.0202 1.4744

AC: absolute cell count; CM: central memory; DN: double-negative; GBM: glioblastoma multiforme; MR: Mendelian randomization; nsnp: number of single-nucleotide polymorphisms; OR: odds ratio; SE: standard error; DC: dendritic cells; lci95: lower limit of the 95% confidence interval; uci95: upper limit of the 95% confidence interval; HLA: human leukocyte antigen.

Figure 5.
The forest plot shows Mendelian randomization analysis of the association between immune cell phenotypic traits and risk of glioblastoma multiforme. CI: confidence interval; nsnp: number of single-nucleotide polymorphisms; OR: odds ratio.

Mediation analysis

To identify potential immune-mediated pathways linking inflammatory factors to GBM, the 2 inflammatory factors (TGFβ1 and CXCL10) and 22 immune cell traits demonstrating causal associations with GBM risk were subjected to mediation analysis.

MR analysis revealed causal relationships between these inflammatory factors and specific immune cell phenotypes (Table S4). TGFβ1 significantly influenced CM DN (CD4⁻CD8⁻) %DN and CD39⁺ resting Treg% CD4 Treg levels, while CXCL10 affected HLA-DR⁺ T cell AC levels. However, CD39⁺ resting Treg% CD4 Treg was excluded as a potential mediator due to logical inconsistencies in effect directionality. Consequently, two potential mediation pathways were identified: (a) TGFβ1 increases GBM risk by decreasing CD4⁻CD8⁻ T cell levels, with a mediator effect of 13.3963% and (b) CXCL10 exerts a protective effect against GBM by increasing HLA-DR⁺ T cell levels, with a mediator effect of 2.7588% (Table 3). Sensitivity analysis (Table S5) confirmed the reliability of these mediation effects.

Table 3.
Mediation analysis.

Mediating factors Outcome Total effect (β-all) Direct effect Mediation effect (β1 × β2) Mediation proportion

CM DN (CD4⁻CD8⁻) %DN GBM 0.1781 0.1542 0.0239 0.1340

HLA-DR+ T cell AC −0.2718 −0.2643 −0.0075 0.0276

AC: absolute cell count; CM: central memory; GBM: glioblastoma multiforme; DN: double-negative; HLA: human leukocyte antigen.

Discussion

This study used publicly available genetic data to conduct a two-sample MR analysis combined with mediation analysis. Statistical analyses were performed to determine potential causal relationships among 91 inflammatory factors, 731 immune cell traits, and GBM risk. The results revealed that inflammatory factors and immune cells play causative roles in GBM development and progression. Notably, using two-step MR analysis, this study demonstrated for the first time that immune cells mediate the effects of inflammatory factors on GBM risk.

Inflammatory factors are integral components of the TME and are closely linked to tumor development and progression. Within the GBM TME, key inflammatory mediators, such as TGFβ1, TNF-α, and interleukins, induce tumor invasion through receptor-mediated signaling pathways.²⁸

This study identified CXCL10 as a protective factor against GBM, as elevated CXCL10 levels were significantly associated with a decreased risk of GBM, highlighting its tumor-suppressive properties. CXCL10, a small-molecule protein belonging to the CXC chemokine superfamily, has been implicated in tumorigenesis, tumor progression, treatment responses, and prognosis across various tumors.^29,30 Consistent with these findings, Wang et al.⁷ reported that lncRNA-135528 upregulated CXCL10 expression, promoting glioma regression. This effect was attributed to CXCL10’s ability to promote lymphocyte recruitment and activation within the TME.³¹ This study investigated the direct effects of CXCL10 on GBM risk and elucidated its potential mediation mechanism: CXCL10 exerts a protective effect on GBM (2.7588%) by increasing HLA-DR⁺ T cell levels. HLA-DR expression on T lymphocytes is a key marker of T cell activation, playing a crucial role in immune surveillance against tumorigenesis and metastasis. Studies have shown that preactivation of the immune microenvironment and induction of T cell aggregation by CXCL10 lead to significant expansion of CD8⁺ T cells and effector memory T cells and ultimately result in tumor elimination in GBM in situ mouse models.^32,33 Additionally, CXCL10 has been shown to enhance the efficacy of immune checkpoint inhibitors in GBM.³⁴ Mechanistically, within the TME, CXCL10 upregulates HLA-DR⁺ T cell levels through CXCR3 signaling, regulating immune responses, angiogenesis, apoptosis, cell cycle progression, and cell proliferation.³⁵ These immunomodulatory functions enhance antitumor immunity and inhibit GBM progression. However, a meta-analysis revealed an association between elevated CXCL10 levels and an increased risk of glioma.³⁶ We hypothesize that CXCL10 inhibits initial tumor growth by recruiting immune cells and exerting an immune surveillance function in the early stage of GBM, when tumor aggressiveness is low and the TME is not yet fully established.³⁷ In the advanced stage of GBM, as tumor aggressiveness increases and the TME becomes fully established, CXCL10 may promote tumor growth and progression by stimulating neovascularization in gliomas.³⁸ It is also possible that there are more complex mechanisms in the TME. In conclusion, further experiments are needed for validation and exploration.

In contrast to CXCL10, this study identified TGFβ1 as a tumor-promoting factor, as elevated TGFβ1 levels were significantly associated with an increased risk of GBM. This oncogenic effect was attributed to TGFβ1’s ability to sustain glioma stem cell tumorigenicity and suppress T cell activity. TGFβ1, the predominant isoform of TGFβ in the immune system, is a multifunctional cytokine involved in cell proliferation, survival, and differentiation.^39,40 As an immunosuppressive molecule, TGFβ facilitates tumor growth, invasion, and metastasis.^41,42 Although produced by normal neurons, TGFβ is highly expressed in GBM and contributes to tumor progression and immunosuppression.⁴³ The tumor-promoting effects of TGFβ in GBM have been associated with its ability to sustain glioma stem cell tumorigenicity through Sry-related HMG-box factors, leading to rapid tumor invasion.^44,45 Studies have shown that GBM cells, upon stimulation by cytokines such as TGFβ, can acquire an enhanced invasive phenotype with mesenchymal characteristics.⁴⁶ Reprogramming tumor-specific T cell responses against TGFβ can convert it from an immunosuppressant to an immunostimulant, enhancing antitumor responses in GBM.⁴⁷ Furthermore, TGFβ signaling has been shown to drive epithelial–mesenchymal transition in hepatocellular carcinoma.⁴⁸ Ganapathy et al.⁴⁹ reported that TGFβ promotes distal metastasis in basal-like breast cancer. These studies collectively highlight the critical role of TGFβ in GBM development and progression. In addition to its tumorigenic effects, TGFβ functions as a potent immunomodulator, effectively inhibiting the activation and differentiation of T cells such as CD4⁺ T cells and CD8⁺ T cells.⁵⁰ Consistent with these findings, this study’s mediation analysis revealed that TGFβ promotes GBM progression by decreasing CD4⁻CD8⁻ DN T cell levels, further supporting its role in immune suppression and GBM pathogenesis.

This study provides a new perspective on immunotherapy for GBM. It has been shown that in a mouse model, CXCL10 upregulation—induced by lncRNA-135528 or type 1 polarized dendritic cells—can inhibit the progression of GBM.^7,37 Ye et al.⁵¹ found that miR-4666-3p could target TGFβR1 and block activation of the TGFβ1/Smad pathway, thereby reducing colorectal cancer cell stemness and suppressing tumorigenesis and progression. The findings of this study also suggest that targeting inflammatory factor–related pathways may be a feasible strategy for tumor suppression, offering new directions for the development of tumor vaccines and anticancer agents.

This study identified 22 immune cell traits associated with GBM risk, with 8 traits associated with decreased risk and 14 traits associated with increased risk. Notably, HLA-DR⁺ T cells and DP (CD4⁺CD8⁺) T cells were identified as protective factors against GBM. T cells are critical mediators of antitumor immune responses.⁵² In particular, HLA-DR⁺CD8⁺ T cells play a crucial role in antigen presentation through HLA-DR complexes, facilitating tumor cell recognition and elimination.⁵³ CD4 on activated and secreting Treg has been identified as a risk factor for GBM. Tregs primarily function to suppress inflammation and immune responses. Elevated Treg levels in patients with glioma have been associated with the release of immunosuppressive cytokines, such as interleukin-10 and TGFβ, which promote tumor invasion, migration, and proliferation, leading to poor clinical outcomes in patients with high-grade glioma.^43,54

This study also identified DN (CD4⁻CD8⁻) T cells as protective factors against GBM. DN T cells, a rare subset of peripheral T cells, accounting for 1%–5% of all peripheral blood T cells, express either TCRαβ or TCRγδ.^55,56 These cells secrete interferon-γ and other cytokines similar to helper T cells, exerting immune-modulatory effects.⁵⁷ In various tumors, DN T cells have been shown to eliminate tumor cells through several mechanisms. Chen J et al.⁵⁸ demonstrated that DN T cells inhibit tumor proliferation and exert antiproliferative effects in pancreatic tumor models using BXPC-3 cells in nude mice. Similarly, Lee et al.⁵⁹ demonstrated that human DN T cells effectively suppress acute myeloid leukemia (AML) cells in vivo and in vitro, offering a novel therapeutic strategy for chemotherapy-resistant patients with AML. Consistent with these findings, the current study revealed that DN T cells mediated the interplay between TGFβ1 and GBM. Mechanistically, TGFβ1 inhibited DN T cell activity, impairing their cytotoxic and antiproliferative effects on GBM cells, thereby promoting tumor progression. Although limited research exists on the relationship between DN T cells and GBM, these findings provide a foundation for future investigation into their therapeutic potential in GBM treatment.

To establish and sustain tumor growth, malignant cells must overcome several biological constraints, including disruption of normal signaling pathways within the surrounding tissues to create a TME that promotes tumor progression.⁶⁰ This study revealed that inflammatory factors differentially regulated immune cell populations, thereby influencing GBM risk. CXCL10 promotes T cell infiltration, thereby reducing the risk of GBM and highlighting its role in early immune surveillance during GBM carcinogenesis. Conversely, TGFβ1 reduced T cell populations, highlighting its role in immune evasion and increased GBM risk. Newly formed tumor cells convert the surrounding matrix into a tumor-supporting microenvironment, facilitating tumor proliferation and maturation.⁶¹ These findings underscore the intricate role of inflammatory factors within the TME in regulating GBM development and progression through immune cell–mediated mechanisms.

This study comprehensively examined the causal relationships among inflammatory factors, immune cells, and GBM risk. A previous MR study investigated the causal relationship between inflammatory factors and GBM, demonstrating that metabolites act as mediators.⁶² In contrast, this study used a larger GWAS dataset and was the first to investigate immune cells as mediators in the inflammatory factor–immune cell–GBM axis.¹⁶ These findings provide novel insights into GBM pathogenesis and highlight potential therapeutic targets within the complex TME.

This study has several limitations. First, all genetic instrumental variables and outcome data were derived from GWAS datasets based on populations of European ancestry. Owing to significant differences in LD patterns, population-specific selection pressures, and environmental exposures across different ethnic groups,⁶³ the generalizability of our findings to non-European populations remains uncertain. This limitation reflects a common issue in current human genetics research: approximately 79% of GWAS participants are of European ancestry.⁶³ Future MR analyses should incorporate more diverse cohorts (e.g. African and East Asian populations), which will be essential for validating causal relationships and evaluating their global translational potential. Second, although the data were derived from European populations and genetic instruments were appropriately controlled in the statistical analyses, residual confounding—such as SNP pleiotropy or subtle population structure differences—cannot be entirely excluded. Therefore, further validation through experimental studies or GWAS data from diverse populations will be needed. In addition, the lack of detailed population-level characteristics in the GWAS datasets used for inflammatory factors and GBM limits the depth of interpretation of our findings. Moreover, a relatively lenient P-value threshold (P < 1 × 10⁻⁵) was used to determine statistical significance, which may increase the risk of false-positive results and potentially overestimate some associations.

Finally, the low mediation proportions observed for certain immune cells suggest that other unrecognized mediators may be involved. CXCL10 has been shown to function as a chemotactic factor for cytotoxic T cells, promoting their infiltration into tumor sites and enhancing the antitumor immune response.⁶⁴ This suggests that its tumor-suppressive effects may not only depend on HLA-DR⁺ T cells but also involve the recruitment of a broader range of T cell subsets. Similarly, TGFβ inhibits the antitumor activity of macrophages and neutrophils and induces their polarization into protumorigenic M2-type macrophages and N2-type neutrophils, thereby promoting tumor progression.⁶⁵ These findings suggest that CXCL10 and TGFβ may coregulate the GBM microenvironment through a multicellular, multipathway regulatory network, rather than relying solely on a single type of immune cell. Although the present study has provided initial insights into possible immune-mediated mechanisms, many aspects remain to be clarified. This study is intended to propose new research hypotheses rather than draw definitive conclusions. Follow-up in vivo and in vitro experiments are needed to further validate the causal relationships among inflammatory factors, immune cells, and GBM risk.

Conclusion

This study comprehensively investigated the causal relationships among inflammatory factors, immune cells, and GBM risk, identifying 2 inflammatory factors and 22 immune cell traits with potential causal associations with GBM. Two significant mediation pathways were elucidated, underscoring the crucial role of immune cells as mediators in the inflammatory factors–GBM axis. These findings provide novel insights into the intricate interplay among inflammation, immunity, and GBM progression. Furthermore, the study highlights the therapeutic potential of targeting inflammatory factors within the TME to regulate immune cell responses, offering potential strategies for GBM prevention and treatment.

Supplemental Material

sj-pdf-1-imr-10.1177_03000605251372521 - Supplemental material for Mendelian randomization and mediation analysis reveal the role of immune cells in the pathways between inflammatory factors and glioblastoma

Supplemental material, sj-pdf-1-imr-10.1177_03000605251372521 for Mendelian randomization and mediation analysis reveal the role of immune cells in the pathways between inflammatory factors and glioblastoma by Ruilong Xu, Yubo Wang, Xiaoshan Ma, Liang Zhang and Yunqian Li in Journal of International Medical Research

Supplemental Material

sj-pdf-2-imr-10.1177_03000605251372521 - Supplemental material for Mendelian randomization and mediation analysis reveal the role of immune cells in the pathways between inflammatory factors and glioblastoma

Supplemental material, sj-pdf-2-imr-10.1177_03000605251372521 for Mendelian randomization and mediation analysis reveal the role of immune cells in the pathways between inflammatory factors and glioblastoma by Ruilong Xu, Yubo Wang, Xiaoshan Ma, Liang Zhang and Yunqian Li in Journal of International Medical Research

Exposure	Outcome	nsnp	β	SE	P	OR	OR_lci95	OR_uci95
CXCL10	GBM	23	−0.2718	0.0780	0.0005	0.7620	0.6540	0.8880
TGFβ1		21	0.1781	0.0873	0.0413	1.1949	1.0070	1.4178

Exposure	Outcome	nsnp	β	SE	P	OR	OR_lci95	OR_uci95
CD20⁻ B cell	GBM	15	−0.0940	0.0418	0.0246	0.9103	0.8387	0.9881
CD24 on IgD⁺ CD38br		19	0.0651	0.0314	0.0380	1.0672	1.0036	1.1349
CD28⁺ DN (CD4⁻CD8⁻) %T cell		20	0.0739	0.0347	0.0329	1.0767	1.0060	1.1524
CD33⁺ HLA-DR⁺ CD14dim %CD33⁺ HLA-DR⁺	17	0.0969	0.0410	0.0182	1.1018	1.0166	1.1940
CD33br HLA-DR⁺ CD14dim myeloid leukocyte	16	0.0667	0.0319	0.0366	1.0689	1.0042	1.1379
CD39⁺ resting Treg% CD4 Treg		21	−0.0312	0.0144	0.0300	0.9693	0.9424	0.9970
CD4 on activated and secreting Treg		20	0.0595	0.0264	0.0239	1.0614	1.0079	1.1177
CD4 on activated Treg		18	0.0568	0.0278	0.0409	1.0584	1.0024	1.1177
CD4 on secreting Treg		16	−0.0821	0.0338	0.0152	0.9212	0.8622	0.9843
CD8 on CD28⁺CD45RA⁺CD8br		15	0.0593	0.0282	0.0352	1.0611	1.0041	1.1213
CD80 on CD62L⁺myeloid DC		18	0.0799	0.0345	0.0205	1.0832	1.0124	1.1589
CD80 on CD62L⁺plasmacytoid DC		16	0.0548	0.0279	0.0492	1.0564	1.0002	1.1157
CD80 on plasmacytoid DC		16	0.0549	0.0279	0.0491	1.0565	1.0002	1.1159
CM DN (CD4⁻CD8⁻) %DN		12	−0.1075	0.0445	0.0157	0.8981	0.8231	0.9799
DP (CD4⁺CD8⁺) T cell		14	−0.0922	0.0401	0.0217	0.9120	0.8430	0.9866
HLA-DR on CD14⁻CD16⁺ monocyte		9	0.0998	0.0433	0.0212	1.1049	1.0150	1.2027
HLA-DR on CD33br HLA-DR⁺ CD14⁻		15	0.0410	0.0190	0.0312	1.0419	1.0037	1.0815
HLA-DR⁺ T cell AC		25	−0.0524	0.0227	0.0210	0.9490	0.9077	0.9921
Myeloid DC %DC		21	−0.0682	0.0305	0.0253	0.9340	0.8798	0.9916
Naive CD4⁺ %T cell		22	0.0834	0.0322	0.0097	1.0870	1.0204	1.1579
Transitional %lymphocyte		17	−0.1233	0.0488	0.0116	0.8840	0.8033	0.9728
Naive CD8br T cell		15	0.2041	0.0939	0.0298	1.2265	1.0202	1.4744

Mediating factors	Outcome	Total effect (β-all)	Direct effect	Mediation effect (β1 × β2)	Mediation proportion
CM DN (CD4⁻CD8⁻) %DN	GBM	0.1781	0.1542	0.0239	0.1340
HLA-DR+ T cell AC	−0.2718	−0.2643	−0.0075	0.0276

Footnotes

Acknowledgments

We acknowledge the assistance of ChatGPT (OpenAI) in language polishing.

Author contributions

Ruilong Xu: Writing–original draft, Writing–review & editing. Xiaoshan Ma: Conceptualization, Methodology. Liang Zhang: Formal analysis. Yubo Wang: Resources. Yunqian Li: Funding acquisition, Supervision.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Data availability

All data used in this study are publicly available GWAS data. In addition, the pooled data used and analyzed in this study are available from the article or supplemental materials.

Declaration of conflicting interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Funding

This study was supported by the Natural Science Foundation of Jilin Province (Free Exploration Key Project; Grant No. YDZJ202401398ZYTS) and Jilin Provincial Medical and Health Talent Project (Grant No. JLSWSRCZX2023-12).

Human ethics

All data were derived from publicly available GWAS datasets; therefore, ethical approval and patient consent were not required for this study.

ORCID iD

Yunqian Li

References

1.
Ostrom QT Gittleman H Truitt G , et al. CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2011-2015. Neuro Oncol 2018; 20: iv1–iv86.

2.
Louis DN Perry A Wesseling P , et al. The 2021 WHO classification of tumors of the central nervous system: a summary. Neuro Oncol 2021; 23: 1231–1251.

3.
Louis DN Perry A Reifenberger G , et al. The 2016 World Health Organization classification of tumors of the central nervous system: a summary. Acta Neuropathol 2016; 131: 803–820.

4.
Ostrom QT Price M Neff C , et al. CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2016-2020. Neuro Oncol 2023; 25: iv1–iv99.

5.
Dominiak A Chełstowska B Olejarz W , et al. Communication in the cancer microenvironment as a target for therapeutic interventions. Cancers (Basel) 2020; 12: 1232.

6.
Amedei A Prisco D D’Elios MM. The use of cytokines and chemokines in the cancer immunotherapy. Recent Pat Anticancer Drug Discov 2013; 8: 126–142.

7.
Wang P Peng X Zhang J , et al. LncRNA-135528 inhibits tumor progression by up-regulating CXCL10 through the JAK/STAT pathway. Apoptosis 2018; 23: 651–666.

8.
Ham SW Jeon HY Jin X , et al. TP53 gain-of-function mutation promotes inflammation in glioblastoma. Cell Death Differ 2019; 26: 409–425.

9.
Emdin CA Khera AV Kathiresan S. Mendelian randomization. JAMA 2017; 318: 1925–1926.

10.
Sojka C Sloan SA. Gliomas: a reflection of temporal gliogenic principles. Commun Biol 2024; 7: 156.

11.
Smith GD Ebrahim S. ‘ Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003; 32: 1–22.

12.
Richmond RC Davey Smith G. Mendelian randomization: concepts and scope. Cold Spring Harb Perspect Med 2022; 12: a040501.

13.
Hingorani A Humphries S. Nature’s randomised trials. Lancet 2005; 366: 1906–1908.

14.
Zhao JH Stacey D Eriksson N ; Estonian Biobank Research Teamet al. Genetics of circulating inflammatory proteins identifies drivers of immune-mediated disease risk and therapeutic targets. Nat Immunol 2023; 24: 1540–1551.

15.
Orrù V Steri M Sidore C , et al. Complex genetic signatures in immune cells underlie autoimmunity and inform therapy. Nat Genet 2020; 52: 1036–1045.

16.
Melin BS Barnholtz-Sloan JS Wrensch MR ; GliomaScan Consortiumet al. Genome-wide association study of glioma subtypes identifies specific differences in genetic susceptibility to glioblastoma and non-glioblastoma tumors. Nat Genet 2017; 49: 789–794.

17.
Enhancing the QUAlity and Transparency Of health Research. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies, https://www.equator-network.org/reporting-guidelines/strobe/ (2021, accessed 15 July 2025).

18.
Yang H Wu X Xiao X , et al. Elucidating the causal associations and mechanisms between circulating immune cells and idiopathic pulmonary fibrosis: new insights from Mendelian randomization and transcriptomics. Front Immunol 2024; 15: 1437984.

19.
Yi G Li Z Sun Y , et al. Integration of multi-omics transcriptome-wide analysis for the identification of novel therapeutic drug targets in diabetic retinopathy. J Transl Med 2024; 22: 1146.

20.
Xing J Wang J Han L , et al. Causal association between inflammatory proteins, inflammatory cells, and cauda equina syndrome: a two-sample Mendelian randomization. World Neurosurg 2025; 197: 123826.

21.
Burgess S Thompson SG ; CRP CHD Genetics Collaboration. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 2011; 40: 755–764.

22.
Hartwig FP Davies NM Hemani G , et al. Two-sample Mendelian randomization: avoiding the downsides of a powerful, widely applicable but potentially fallible technique. Int J Epidemiol 2016; 45: 1717–1726.

23.
Hemani G Zheng J Elsworth B , et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife 2018; 7: e34408.

24.
Bowden J Davey Smith G Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015; 44: 512–525.

25.
Verbanck M Chen CY Neale B , et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 693–698.

26.
Noble WS. How does multiple testing correction work? Nat Biotechnol 2009; 27: 1135–1137.

27.
Burgess S Daniel RM Butterworth AS ; EPIC-InterAct Consortiumet al. Network Mendelian randomization: using genetic variants as instrumental variables to investigate mediation in causal pathways. Int J Epidemiol 2015; 44: 484–495.

28.
Wolf KJ Chen J Coombes J , et al. Dissecting and rebuilding the glioblastoma microenvironment with engineered materials. Nat Rev Mater 2019; 4: 651–668.

29.
Giuliani N Bonomini S Romagnani P , et al. CXCR3 and its binding chemokines in myeloma cells: expression of isoforms and potential relationships with myeloma cell proliferation and survival. Haematologica 2006; 91: 1489–1497.

30.
Li G Tian L Hou JM , et al. Improved therapeutic effectiveness by combining recombinant CXC chemokine ligand 10 with cisplatin in solid tumors. Clin Cancer Res 2005; 11: 4217–4224.

31.
Karin N Razon H. Chemokines beyond chemo-attraction: CXCL10 and its significant role in cancer and autoimmunity. Cytokine 2018; 109: 24–28.

32.
Metzemaekers M Vanheule V Janssens R , et al. Overview of the mechanisms that may contribute to the non-redundant activities of interferon-inducible CXC chemokine receptor 3 ligands. Front Immunol 2017; 8: 1970.

33.
Qin S Rottman JB Myers P , et al. The chemokine receptors CXCR3 and CCR5 mark subsets of T cells associated with certain inflammatory reactions. J Clin Invest 1998; 101: 746–754.

34.
Xu X Zhang Z Du J , et al. Recruiting T-cells toward the brain for enhanced glioblastoma immunotherapeutic efficacy by co-delivery of cytokines and immune checkpoint antibodies with macrophage-membrane-camouflaged nanovesicles. Adv Mater 2023; 35: e2209785.

35.
Groblewska M Litman-Zawadzka A Mroczko B. The role of selected chemokines and their receptors in the development of gliomas. Int J Mol Sci 2020; 21: 3704.

36.
Yang S Luo M Yang S , et al. Relationship between chemokine/chemokine receptor and glioma prognosis and outcomes: systematic review and meta-analysis. Int Immunopharmacol 2024; 133: 112047.

37.
Fujita M Zhu X Ueda R , et al. Effective immunotherapy against murine gliomas using type 1 polarizing dendritic cells – significant roles of CXCL10. Cancer Res 2009; 69: 1587–1595.

38.
Xue W Zhang J Tong H , et al. Effects of BMPER, CXCL10, and HOXA9 on neovascularization during early-growth stage of primary high-grade glioma and their corresponding MRI biomarkers. Front Oncol 2020; 10: 711.

39.
Li MO Wan YY Sanjabi S , et al. Transforming growth factor-beta regulation of immune responses. Annu Rev Immunol 2006; 24: 99–146.

40.
Massagué J Sheppard D. TGF-β signaling in health and disease. Cell 2023; 186: 4007–4037.

41.
Santibañez JF Quintanilla M Bernabeu C. TGF-β/TGF-β receptor system and its role in physiological and pathological conditions. Clin Sci (Lond) 2011; 121: 233–251.

42.
Zou J Li C Jiang S , et al. AMPK inhibits Smad3-mediated autoinduction of TGF-β1 in gastric cancer cells. J Cell Mol Med 2021; 25: 2806–2815.

43.
Eder K Kalman B. The dynamics of interactions among immune and glioblastoma cells. Neuromolecular Med 2015; 17: 335–352.

44.
Ikushima H Todo T Ino Y , et al. Autocrine TGF-beta signaling maintains tumorigenicity of glioma-initiating cells through Sry-related HMG-box factors. Cell Stem Cell 2009; 5: 504–514.

45.
Lathia JD Mack SC Mulkearns-Hubert EE , et al. Cancer stem cells in glioblastoma. Genes Dev 2015; 29: 1203–1217.

46.
Joseph JV Conroy S Tomar T , et al. TGF-β is an inducer of ZEB1-dependent mesenchymal transdifferentiation in glioblastoma that is associated with tumor invasion. Cell Death Dis 2014; 5: e1443.

47.
Hou AJ Shih RM Uy BR , et al. IL-13Rα2/TGF-β bispecific CAR-T cells counter TGF-β-mediated immune suppression and potentiate anti-tumor responses in glioblastoma. Neuro Oncol 2024; 26: 1850–1866.

48.
Steinway SN Zañudo JG Ding W , et al. Network modeling of TGFβ signaling in hepatocellular carcinoma epithelial-to-mesenchymal transition reveals joint sonic hedgehog and Wnt pathway activation. Cancer Res 2014; 74: 5963–5977.

49.
Ganapathy V Ge R Grazioli A , et al. Targeting the transforming growth factor-beta pathway inhibits human basal-like breast cancer metastasis. Mol Cancer 2010; 9: 122.

50.
Li MO Flavell RA. TGF-beta: a master of all T cell trades. Cell 2008; 134: 392–404.

51.
Ye J Lei J Fang Q , et al. miR-4666-3p and miR-329 synergistically suppress the stemness of colorectal cancer cells via targeting TGF-β/Smad pathway. Front Oncol 2019; 9: 1251.

52.
Wang H Zhou H Xu J , et al. Different T-cell subsets in glioblastoma multiforme and targeted immunotherapy. Cancer Lett 2021; 496: 134–143.

53.
Osuna-Gómez R Arqueros C Galano C , et al. Effector mechanisms of CD8+ HLA-DR+ T cells in breast cancer patients who respond to neoadjuvant chemotherapy. Cancers (Basel) 2021; 13: 6167.

54.
Ooi YC Tran P Ung N , et al. The role of regulatory T-cells in glioma immunology. Clin Neurol Neurosurg 2014; 119: 125–132.

55.
Zhang ZX Yang L Young KJ , et al. Identification of a previously unknown antigen-specific regulatory T cell and its mechanism of suppression. Nat Med 2000; 6: 782–789.

56.
Fischer K Voelkl S Heymann J , et al. Isolation and characterization of human antigen-specific TCR alpha beta+ CD4(−)CD8- double-negative regulatory T cells. Blood 2005; 105: 2828–2835.

57.
Sundaravaradan V Saleem R Micci L , et al. Multifunctional double-negative T cells in sooty mangabeys mediate T-helper functions irrespective of SIV infection. PLoS Pathog 2013; 9: e1003441.

58.
Chen J Xu H Zhu XX , et al. [DNT cell inhibits the growth of pancreatic carcinoma via NKG2D and MICA pathway in vivo]. National Medical Journal of China 2017; 97: 1265–1269.

59.
Lee J Minden MD Chen WC , et al. Allogeneic human double negative T cells as a novel immunotherapy for acute myeloid leukemia and its underlying mechanisms. Clin Cancer Res 2018; 24: 370–382.

60.
De Visser KE Joyce JA. The evolving tumor microenvironment: from cancer initiation to metastatic outgrowth. Cancer Cell 2023; 41: 374–403.

61.
Pradhan R Kundu A Kundu CN. The cytokines in tumor microenvironment: from cancer initiation-elongation-progression to metastatic outgrowth. Crit Rev Oncol Hematol 2024; 196: 104311.

62.
Chen X Han L Xu W. Genetically predicted 3-methoxytyrosine mediates the causal association between fibroblast growth factor 21 and glioblastoma multiforme. J Cancer 2025; 16: 680–688.

63.
Martin AR Kanai M Kamatani Y , et al. Clinical use of current polygenic risk scores may exacerbate health disparities. Nat Genet 2019; 51: 584–591.

64.
Okada H. Brain tumor immunotherapy with type-1 polarizing strategies. Ann N Y Acad Sci 2009; 1174: 18–23.

65.
Deng Z Fan T Xiao C , et al. TGF-β signaling in health, disease, and therapeutics. Signal Transduct Target Ther 2024; 9: 61.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.24 MB

0.50 MB