Glioblastoma (GBM) is a common primary tumor that can occur anywhere in the central nervous system of adults (1). GBM is marked by profound cellular heterogeneity and diffuse infiltrative growth, characteristics that render it essentially incurable (2, 3). Although the current standard of care—maximal safe surgical resection followed by adjuvant radiotherapy, chemotherapy, and other modalities—can temporarily control tumor burden, intrinsic resistance to these treatments results in a dismal median overall survival of only 14–16 months after diagnosis (1). Consequently, there is an urgent need to develop novel therapeutic strategies to improve patient outcomes. Per established protocols, adjuvant radiotherapy (60 Gy delivered in 30 fractions over 6 weeks) should commence within 3–6 weeks post−surgery, with daily concurrent administration of the oral alkylating agent temozolomide. Emerging approaches now focus on advanced radiation techniques and molecularly targeted therapies to overcome GBM’s treatment resistance. Temozolomide should be resumed 4 weeks after the completion of radiotherapy, usually for 5 consecutive days every 28 days for a total of 6 months in one cycle. In a clinical trial involving 573 participants, compared with radiotherapy alone, this regimen improved survival rates (14.6 months vs. 12.1 months, hazard ratio (HR) 0.63, 95% confidence interval (CI) 0.52-0.75; p<0.001) (4). Drugs targeting immune checkpoints, such as Cytotoxic T-Lymphocyte-Associated Protein 4 (CTLA-4), Programmed Cell Death Protein 1 (PD-1), and Programmed Death-Ligand 1 (PD-L1), can enhance the anti-tumor immune response and enable T cells to more effectively eradicate cancer cells. Given the success in many solid tumors, the potential of immune checkpoint blockade therapy has been actively explored for GBM (5). Gliomas, which arise from glial cells or their progenitors, are predominantly classified as astrocytomas or oligodendrogliomas (6). Under the World Health Organization grading system, gliomas are divided into circumscribed (grade I) and diffuse (grades II–IV) entities, with higher grades indicating greater malignancy. GBM, defined as a grade IV diffuse astrocytoma, represents the most aggressive glioma subtype, hallmarked by pronounced hypercellularity, rapid mitotic activity, extensive microvascular proliferation, and characteristic pseudopalisading necrosis (7, 8).

Several demographic, genetic, and environmental factors have been implicated in GBM pathogenesis. Advanced age and male sex are consistently associated with higher incidence, with risk rising markedly after 50 years and peaking in late adulthood (9). Approximately 5% of gliomas develop in the context of hereditary cancer syndromes such as Li–Fraumeni, Turcot, and neurofibromatosis types 1 and 2, highlighting a genetic predisposition component (10). Epidemiological studies have also reported an inverse association between atopic conditions (e.g., asthma, eczema) and glioma risk, suggesting a role for immune-mediated mechanisms in protection against GBM (11). Outside of high-dose ionizing radiation, which remains the only established environmental risk factor for GBM, associations with chemical exposures, occupational hazards, and non-ionizing radiation have been largely inconclusive. However, most existing investigations rely on retrospective case–control designs with limited cohort sizes, potential recall and selection biases, heterogeneous exposure assessments, and simplistic modeling approaches, impeding the identification of robust, clinically translatable risk factors.

In this study, we conducted an extensive survey of clinical data of oncology patients and non-oncology patients in Guangdong Provincial People’s Hospital from 2016 to 2023. Subsequently, an easy-to-use nomogram was developed using univariate versus multivariate analysis. The primary objective of this study was to analyze risk factors for GBM and create a reliable, non-invasive nomogram to predict the likelihood of GBM using appropriate, validated analytical methods. Our nomogram uses real cases from our hospital to create a clinically relevant predictive tool. There are no known risk factors for glioblastoma other than rare genetic predisposition and irradiation (12, 13). In this study, we aimed to analyze the risk factors for brain metastasis in GBM patients and to establish a valid, noninvasive column-line diagram of the likelihood of brain metastasis in GBM patients using advanced statistical analysis methods. In our nomogram, we can infer the possibility of brain metastasis by simple blood counts and pathology types, and the nomogram is easier to apply in clinical practice than other column charts of the same type.

Regarding our finding that the concentration of chloride (Cl) in blood electrolytes is associated with the occurrence of GBM, this is a very novel and interesting conclusion. It has been previously shown that the presence of Cl is associated with a reduced rate of oral cancer recurrence (24). To dig deeper into the mechanisms involved, we need to have some understanding of the role of Cl. As far as current knowledge goes, Cl plays an important role in maintaining fluid balance, digestive processes, nerve conduction and acid-base balance. We hypothesize that it may be possible that excess Cl due to disturbances in ion channel function, especially chloride intracellular channel 1 (CL IC1), reduces cytoplasmic pH and thus induces apoptosis in tumor cells. Elemental chloride is considered a protective factor for tumor recurrence (25). CI has an important role in cellular homeostasis in both physiological and pathological states. changes in Cl flow regulate cell volume, modulate cellular secretion, and maintain intra- and extracellular pH, all of which are important for the maintenance of enzyme activity and the cell cycle (26–28). Some scientists have pointed out that the concentration of intracellular chlorine is dynamic and plays an irreplaceable role in regulating the activity of a variety of substances including hemoglobin (29). Cl’s roles in cellular physiology are clear; however, their relationship to the pathogenesis of cancer remains unclear. The prominence of Cl channels increased following the revelation that multidrug resistance proteins (MDR/P-glycoprotein) interact with volume-activated Cl channels in the cancer cells of chemotherapy-treated patients (30). Multiple studies have documented the correlation between the expression of chloride channels and the prognosis and survival of patients (31–33). CLIC1 plays an active role as an ion channel or signal transducer in numerous physiological and pathological processes (34, 35). Prior research has provided evidence indicating that CLIC1 plays a crucial role in the advancement of various malignant tumors (36–41).There was a notable increase in CLIC1 expression observed in oral cancer tissues and in the blood of cancer patients. Furthermore, the upregulation of CLIC1 exhibited a significant correlation with clinical and pathological stage, tumor size, and overall survival (42). Alterations in tumor cell gene structure and function result in tumor cells with the following characteristics: insensitivity to growth inhibitory signals, evasion of apoptosis, and unlimited proliferative potential (24). In recent studies, the role of CLIC1 in regulating tumor cell proliferation and apoptosis has been highlighted. Notably, Kobayashi demonstrated that the absence of CLIC1 hindered cell proliferation and triggered apoptosis in esophageal squamous cell carcinoma (ESCC) (39). Similar results were found in gastric cancer cells (40). In hepatocellular carcinoma studies, CLIC1 overexpression increased cell viability (34). Recent research indicates that CLIC1 plays a role in the advancement of cancer, yet the precise mechanism behind this phenomenon has yet to be fully elucidated. Wang P’s study revealed that CLIC1 governs the movement and infiltration of colon cancer cells by modulating the ROS-mediated MAPK/ERK signaling pathway (36). Research focusing on gastric cancer has indicated that CLIC1 might control the expression of ITG family proteins, resulting in the consecutive activation of PI3K/AKT, MAPK/ERK, and MAPK/p38 pathways (40).

Some researchers have discovered that the cell surface costimulatory molecule LFA-1 relies on Mg to adopt an active conformation on CD8+ T cells (43). We acknowledge that the extremely wide 95% confidence interval for serum Mg reflects limited precision likely driven by a small effective sample size at extreme Mg values, a right−skewed distribution with influential outliers, potential analytic variability in the colorimetric assay, and multicollinearity with other ionic predictors; a post−hoc sensitivity analysis excluding the highest and lowest 5% of Mg values confirmed that higher Mg remained associated with increased risk, albeit with only modest narrowing of the interval, underscoring that future studies with larger cohorts should (i) model Mg categorically (e.g., quartiles), (ii) perform formal influence diagnostics such as Cook’s distance to identify and down−weight outliers, and (iii) validate these findings in external datasets to determine whether Mg truly contributes to GBM risk or if the current estimate primarily reflects statistical instability.

This enhances calcium flux, signaling, metabolic reprogramming, and the formation of immune synapses, subsequently boosting specific cytotoxicity. These findings conceptually connect co-stimulation and nutrient sensing, and highlight the Mg-LFA-1 axis as a biological system with potential therapeutic applications (43). Low Mg intake and hypomagnesemia can impact a broad spectrum of diseases and support various disease processes, including infections and cancer (44–48). Mice fed a Mg-deficient diet have been reported to exhibit accelerated metastatic spread of cancer cells (49), and insufficient inducible T-cell kinase (ITK) activity has led to impaired immune responses against influenza in mice due to low Mg intake (50). Extensive epidemiological studies have suggested that Mg intake may be linked to a reduced risk of colorectal cancer (51, 52).

High-density lipoprotein (HDL-C) levels in plasma have been reported to demonstrate an inverse association with cancer risk (53). In a large meta-analysis, lower plasma HDL-C levels were found to be correlated with an increased risk of cancer. Each 10 mg/dL increase in plasma HDL-C levels was found to significantly reduce the risk of cancer incidence by 36% (54). However, conflicting results have also emerged, with some studies proposing that low plasma HDL-C levels may be considered incidental to the presence of cancer (55). Chemotherapy-induced reductions in HDL-C levels have been identified as another link between cancer and HDL, further complicating the relationship between HDL and cancer (56). This dual role of HDL in cancer has also been observed in in vitro studies. For instance, the antioxidant activity of HDL has been found to restrain prostate cancer cell proliferation (57), while HDL can stimulate cell migration in breast cancer (BC) cell lines (58), potentially due to oxidative modification of HDL under oxidizing conditions in BC. In addition to the effects of HDL itself on tumorigenesis and development, the impact of HDL-related enzymes on tumors cannot be overlooked (59).

Serum uric acid (UA) is an antioxidant that is abundant in the blood and has a wide range of roles: antioxidant action; regulation of vascular function; antimicrobial action; regulation of immune function; and maintenance of acid-base balance. It is thus clear that UA plays an important role in maintaining normal physiological functions in the body (60). There are already studies that have already revealed the association between UA and cancer risk. In 2019, a meta-analysis showed that hyperuricemia was associated with a higher risk of cancer in men; hyperuricemia and increased mortality in women were linked (61). Additionally, another study showed that patients with hyperuricemia were at higher risk of developing kidney cancer (62). For the relationship between hyperuricemia and tumors, only a link has been found, and for the time being, no scientists have been able to specifically elucidate the mechanisms involved, and we speculate that it may be related to the following mechanisms: one may be due to impaired renal excretion, rapid cellular renewal, and increased purine metabolism due to the presence of xanthine oxidase and elevated UA levels. In addition, we all know that reactive oxygen species are associated with cellular damage and cancer.UA is able to react with reactive oxygen species to avoid reactive oxygen species damage.1 Moreover, increased SUA has been associated with an attenuated anticancer response. Therefore, because of the relationship between SUA and reactive oxygen species response, it is not difficult to understand that high levels of UA are a predictor of tumor presence.

Granulocytes are leukocytes with specific cytoplasmic granules mainly including eosinophils, basophils and neutrophils (63). Eosinophils are a type of leukocytes that usually increase in acidic environments, and their characteristics include: nuclear morphology: the nucleus of eosinophils tends to be bilobed or multilobed, and the nucleus has granular structures. Cytoplasmic granules: The cytoplasm of eosinophils contains a large number of eosinophilic granules, which mainly contain eosinophilic dyes, such as the eosinophilic dyes basic protein and histamine. Eosinophils are mainly involved in the regulation of parasitic infections and allergic reactions by releasing chemicals within the granules to kill parasites or regulate allergic reactions. Overall, eosinophils are characterized by the special morphology of their nuclei and the large number of eosinophilic granules in their cytoplasm, and their main function is to participate in the regulation of parasitic infections and allergic reactions (64). Basophils are the least numerous granulocytes in the blood. They have a characteristic morphology with a large number of staining granules (63). There are few studies on the number of eosinophils in peripheral blood as a prognostic parameter in patients with tumors. It has been suggested that a decrease in eosinophil count may lead to a shorter overall survival (OS) in patients with stage I colorectal cancer. In this context, eosinophilia is considered an independent risk factor for colorectal cancer in stages II and III (65). On the other hand, allergy was associated with reduced mortality from nodal tumors (66). These observations conclude that blood eosinophil and basophil counts may be associated with the prognosis of colorectal cancer. The active role of these two granulocytes in tumors is most likely related to the secretion of basophil granule contents, including histamine and pro-inflammatory cytokines-such as TNFα, IL-6, and IL-1β-which increase the inflammatory response, recruit cancer-specific CD8+ T-cells into tumors, and increase apoptosis of cancer cells (67).

Including systemic inflammatory parameters such as eosinophil and basophil counts in a glioma risk model is supported by evidence that the peripheral immune signature of glioma differs markedly from that of other solid tumors. Preoperative eosinophil‐based scoring (ENS) has been shown to be an independent prognostic indicator for glioma grade and overall survival: patients with eosinophil counts ≥0.08×10⁹/L demonstrated significantly higher 3-year OS rates (84.0% vs. 80.0%, P=0.043) and ENS positively correlated with tumor grade (r=0.311, P<0.001) (68). Epidemiological data further reveal that atopic conditions—marked by elevated eosinophils—are inversely associated with glioma risk, a protective relationship not observed in colorectal or nasopharyngeal cancers (69). Likewise, preoperative basophil counts ≥0.015×10⁹/L independently predict longer progression-free survival in glioblastoma patients (P<0.05), whereas basophil prognostic value in melanoma or ovarian carcinoma is less consistent (18). These findings likely reflect the unique neuroinflammatory microenvironment of the central nervous system and differential trafficking of innate immune cells across the blood–brain barrier in glioma, thereby justifying the specificity of including eosinophil and basophil parameters in a glioma-focused nomogram.

Moreover, recent high−dimensional profiling studies of the GBM immune microenvironment have begun to uncover mechanistic links for eosinophils and basophils. A 2024 scoping review of single−cell RNA−seq in GBM identified eosinophil−like myeloid clusters enriched for type−2 inflammatory transcripts that inversely correlate with patient survival in TCGA−GBM datasets (70). Single−cell and spatial transcriptomic analysis of grade IV glioma samples further revealed eosinophil hotspots at invasive tumor margins co−localized with CD8⁺ T cells, suggesting an immune−activating niche that may underlie the protective association of peripheral eosinophilia (71). Clinically, pretreatment circulating basophil counts have been shown to independently predict longer progression−free survival in glioblastoma patients, providing direct evidence for basophil−mediated antitumor effects. Together, these cutting−edge findings lend biological plausibility to our nomogram’s inclusion of peripheral eosinophil and basophil counts as protective GBM risk factors.

In this study, we developed a nomogram based on six independent risk factors for predicting glioblastoma. The model showed good discrimination and calibration in both the training and validation cohorts. Furthermore, decision curve analysis demonstrated its clinical utility across a wide range of threshold probabilities. All six independent risk factors retained statistical significance in multivariate analysis (p < 0.05), underscoring their independent contributions to GBM risk. We recognize that circulating eosinophil, basophil and chloride levels can be influenced by a variety of non-neoplastic conditions—eosinophilia and basophilia occur in allergic disorders, parasitic infections or hematologic syndromes (e.g. hypereosinophilic syndrome), while hyperchloremia may reflect dehydration, renal impairment or acid–base disturbances. To mitigate these confounders, we applied stringent exclusion criteria—omitting patients with metabolic diseases (gout, thyroid disorders), cardiovascular comorbidities or recent trauma ( Table 1 )—and collected all blood samples pre-operatively under standardized fasting conditions. Moreover, in our multivariate logistic regression each parameter remained independently significant, indicating that the associations with GBM risk persist after adjusting for other measured clinical and laboratory variables. Nevertheless, we cannot fully exclude residual confounding by unmeasured conditions such as subclinical infection or atopic disease. Future prospective studies with comprehensive comorbidity profiling—and ideally sensitivity analyses excluding patients with documented inflammatory or allergic disorders—are warranted to confirm the validity and specificity of these inflammatory markers in glioblastoma risk prediction.

Although we have identified several risk factors for GBM and provided a new method for predicting the occurrence of GBM, this study has some limitations and shortcomings. Firstly, the factors identified through MR analysis do not seem to be validated in hospital data. Understanding the reasons for this requires a deeper understanding of the underlying mechanisms. Our MR analysis is based on GWAS databases, where over 90% of the data comes from individuals of European descent, and it is unknown whether this could be one of the reasons. Additionally, due to the limited amount of real-world data, the applicability of the conclusions on a large scale remains to be tested. Secondly, our study only explored the relationship between risk factors and GBM, without elaborating on the specific mechanisms, which is also our next goal for further research.

This study provides a better understanding of the risk factors for GBM occurrence. Additionally, we developed a new practical nomogram, greatly expanding the scope of clinical practice to calculate the characteristics of GBM occurrence. This provides theoretical support for the prevention of GBM and further demonstrates the importance of promoting a healthier lifestyle in reducing the incidence of GBM.

Future Directions: While our nomogram demonstrates strong discrimination and calibration, we plan to pursue the following specific research aims to broaden and deepen its clinical applicability:

(1) External Validation in Independent Cohorts. We will assemble and test our model on retrospective datasets from at least two external institutions—e.g., the TCGA/CGGA public GBM cohorts and a prospective, multi−center clinical registry—to evaluate generalizability and recalibrate risk thresholds as needed. (2) Radiomics–Clinical Integration. Using standardized feature extraction pipelines (e.g. PyRadiomics), we will derive quantitative imaging biomarkers from contrast−enhanced MRI (texture, shape, wavelet features) and incorporate these radiomics signatures alongside our six laboratory parameters to build a combined radiomics−clinical nomogram. (3) Multi−Omics Expansion. We will layer on key molecular markers (MGMT promoter methylation, IDH mutation status) and, where available, transcriptomic profiles to develop a multi−omics risk model and assess whether this further improves predictive accuracy beyond clinical and radiomic data alone. (4) Longitudinal Dynamic Modeling. By prospectively collecting serial blood tests and imaging at defined postoperative intervals, we aim to construct a time−dependent risk score that captures dynamic changes in inflammation, metabolism, and radiomics over the disease course. (5) Clinical Implementation Study. Finally, we will integrate the refined nomogram into our hospital’s electronic decision−support system, piloting its use in multidisciplinary tumor boards and measuring its impact on diagnostic timing, treatment selection, and patient outcomes in a feasibility study.