1 Introduction
Degenerative central nervous system (CNS) diseases (Xu et al., 2022), such as Alzheimer's disease (AD), Parkinson’s disease (PD), multiple sclerosis (MS), and amyotrophic lateral sclerosis (ALS), as well as neurological tumors, particularly glioblastoma (GBM) (Broekman et al., 2018), represent the most challenging diseases in the field of neuroscience. These diseases not only impose a heavy burden on patients and their families, but also pose significant pressures on society and the healthcare system (Wareham et al., 2022).
Despite the significant differences in clinical presentation and pathology between degenerative CNS diseases and GBM, they exhibit remarkable similarities in certain key biological processes, such as ferroptosis, oxidative stress, and neuroinflammation.
Ferroptosis is a form of cell death dependent on iron and lipid peroxidation, shown to play a critical role in both degenerative CNS diseases and GBM. For example, iron accumulation and the resulting oxidative damage are key pathological processes in Parkinson’s disease (Yan et al., 2024). Similarly, iron metabolism abnormalities have been observed in GBM, suggesting that ferroptosis could be a common pathological mechanism in these two types of diseases (Soo et al., 2024). Oxidative stress is important in both degenerative CNS diseases and GBM. Excess reactive oxygen species (ROS) can cause cellular damage and death. In AD and PD, oxidative stress is considered a major cause of neuronal damage (Isidro, 2018). In GBM, oxidative stress not only promotes tumor cell proliferation and migration but may also influence tumor progression by modulating interactions within the tumor microenvironment (Lauryn E et al., 2017). Neuroinflammation is common in degenerative CNS diseases. Inflammation plays a key role in the progression of AD and PD (Pei-Lun et al., 2024). In GBM, the inflammatory environment not only promotes tumor cell growth and survival but may also affect the tumor’s response to treatment by modulating immune cell function (Ana Helena Larangeira et al., 2024).
Research indicates significant overlap in gene and epigenetic regulation between degenerative CNS diseases and GBM. For example, a study (Lu et al., 2023) analyzed the expression of different isoforms (alternative splicing variants) of genes in brain tissue and identified multiple gene regulatory sites associated with neurological traits and diseases. These regulatory sites may play similar roles in different degenerative CNS diseases and GBM. Notably, isoform-ratio quantitative trait loci have been shown to regulate gene expression in brain tissue, with implications for various CNS conditions such as AD, mood fluctuations, and sleep duration. Understanding these regulatory mechanisms across degenerative CNS diseases and GBM could provide new insights into shared molecular pathways and identify novel therapeutic targets (Ashraf et al., 2023). Understanding these common regulatory mechanisms helps elucidate the shared genetic regulation in different diseases. In both brain tumors and degenerative CNS diseases, cell-cell interactions are crucial in disease development and progression. For instance, the activation of astrocytes and microglia is critical in AD progression, where microglial activation contributes to amyloid-β plaque clearance or neuroinflammation, as reported in recent study (Teresa et al., 2024). Similarly, abnormal activation of astrocytes is a key driver of GBM proliferation and invasion, as demonstrated in preclinical research (Jian and Shiwei, 2024). Despite these shared mechanisms, degenerative CNS diseases and GBM exhibit significant differences. Degenerative CNS diseases, such as AD and Parkinson’s disease, are chronic progressive conditions characterized by gradual neuronal damage and death, leading to cognitive and motor function decline (Andrea J. et al., 2020; M. et al., 2024).
In contrast, GBM is a highly invasive malignant brain tumor characterized by rapid growth and extensive damage to surrounding brain tissue. Therapeutically, the treatment of degenerative CNS diseases mainly focuses on symptom management and slowing disease progression, such as using dopamine replacement therapy in Parkinson’s disease. In contrast, GBM treatment includes surgical resection, radiotherapy, and chemotherapy, although these methods often cannot completely eradicate the tumor and it tends to recur.
Despite significant progress in understanding the molecular mechanisms underlying these diseases over the last decades, these findings are far from being fully translated into effective treatments. Therefore, the search for novel biomarkers that can be used as early diagnostic tools, indicators of disease progression, and monitoring means of treatment response, as well as the identification of new therapeutic targets, has become a cutting-edge topic in neuroscience research.
In recent years, the rapid advancements in bioinformatics (Gauthier et al., 2019), genomics (Gangfuß et al., 2022), and proteomics (Cui et al., 2022) have significantly deepened our understanding of the molecular underpinnings of neurological disorders. The application of high-throughput gene expression analyses (Hrdlickova et al., 2017) and differential expression analysis techniques (McDermaid et al., 2019) has emerged as a powerful tool for identifying disease-associated genes and pathways. By examining the gene expression patterns in degenerative CNS diseases and GBM, we aim to uncover potential common molecular mechanisms and cross-disease biomarkers between these conditions. These cross-disease biomarkers not only offer a new perspective for understanding these seemingly disparate diseases, which may share certain fundamental biological traits, but also hold promise for the development of therapeutic approaches that transcend specific diseases.
4 Discussion
Nearly one-sixth of the world's population is affected by CNS diseases (Zhou et al., 2021). These disorders range from mild neurological impairment (which may manifest as motor, sensory, visual, speech, cognitive impairment, or a combination of these symptoms) to more severe conditions, such as coma and brain death. The central nervous system, including the brain and spinal cord, carries important tasks for processing sensory information, controlling movement, and coordinating all higher functions of the body, such as thinking, memory, and affect (Capani et al., 2016).
In many CNS diseases, ongoing progressive neurological damage is often caused by primary neuronal dysfunction. However, not all central nervous system disorders are degenerative. For example, although brain trauma or stroke affects the CNS, they are usually acute events and differ from chronic degenerative processes (Campbell and Khatri, 2020). In contrast, some CNS diseases such as Alzheimer's disease, Parkinson’s disease, and multiple sclerosis are indeed degenerative, and their main features are gradual deterioration in the structure and function of nerve cells, particularly neurons (Fan and Huo, 2021). These degenerative diseases progress over time, and damage and death of nerve cells lead to various dysfunctions. For example, cognitively function-associated neurons are particularly affected in Alzheimer's disease (Tiwari et al., 2019). Pathological mechanisms of these diseases may include factors such as abnormal protein folding, mitochondrial and energy metabolism disorders (Ferrer, 2018), oxidative stress (Singh et al., 2019), inflammation (Voet et al., 2019), and imbalances in neurotransmitter systems (Mahalakshmi et al., 2020). These complex pathological processes not only lead to the decline and/or death of neuronal function, but also further affect the overall function of the brain and spinal cord.
In 2014, GBM represented 16% of all primary brain and central nervous system tumors. The age-adjusted mean incidence rate was 3.19 per 100,000 (Thakkar et al., 2014). By 2023, however, the incidence of malignant brain tumors reached an alarming 7 cases per 100,000 and increased with age. The 5-year survival rate is approximately 36%. Older patients have a lower survival rate than younger patients. Survival rates ranged from approximately 71.5% in patients aged 15–39 years and 21% in patients aged 40 years and older. About 49% of these primary malignant brain tumors are glioblastomas (Schaff and Mellinghoff, 2023). Glioblastoma is a very aggressive brain tumor classified as grade IV in malignant glioma, which is the most severe grade. This tumor is characterized by rapid growth and high invasiveness to surrounding brain tissue. Symptoms of GBM include headache, seizures, neurocognitive impairment, and focal neurological deficits (Ostrom et al., 2022).
Recent research has unveiled a novel therapeutic strategy against cancer, drawing inspiration from treatments for neurodegenerative disorders (Soragni et al., 2016). This approach involves the development of ReACp53, a designed inhibitor that targets the aggregation of the p53 protein, a common pathological feature in various cancers, notably ovarian carcinomas. By preventing p53 aggregation, ReACp53 restores its tumor-suppressive function, leading to a reduction in tumor growth and an increase in cancer cell apoptosis. This breakthrough highlights the potential of leveraging neurodegenerative disease therapies to combat cancer, offering a promising avenue for future cancer treatment research.
Our study employed a novel approach that integrates WGCNA, multiple machine learning algorithms, and functional enrichment analyses to identify biomarkers shared between GBM and degenerative CNS diseases. This integrated pipeline offers several advantages over conventional methods.
First, traditional biomarker discovery methods often rely on a single dataset or apply univariate statistical methods to identify significant genes. These approaches may overlook subtle but biologically important interactions and fail to generalize across datasets. By contrast, our method combines data from multiple independent datasets, ensuring robustness and minimizing potential biases. The WGCNA framework, in particular, enables the identification of gene modules based on co-expression patterns, providing a systems-level understanding of disease-associated networks that traditional differential expression analyses cannot capture (Oluwatosin A et al., 2025).
To further refine the identification of key candidate genes from the WGCNA results, we integrated three machine learning techniques: SVM-RFE, LASSO, and Random Forest. Each method contributes unique strengths to the selection process. SVM-RFE is well-known for its ability to select relevant features while minimizing the risk of overfitting, making it particularly useful for analyzing datasets with a high number of variables (Hector et al., 2018). LASSO is effective in both feature selection and regularization, promoting models that are sparse, interpretable, and generalizable (Maike and Andreas, 2024). Random Forest, on the other hand, is highly precise in handling large datasets, managing data imbalance effectively, and providing valuable insights into feature importance (Ibrahim et al., 2024). By combining these methodologies, we capitalize on the unique strengths of each: LASSO’s simplicity and interpretability, SVM-RFE’s adaptability to complex datasets, and Random Forest’s comprehensive evaluation of feature relevance. This integrated strategy ensures thorough and reliable identification of crucial genes, enhancing the predictive capabilities and robustness of our analysis.
Traditional methods often neglect downstream functional validation at the pathway level, limiting the understanding of the broader biological context of identified biomarkers. In this study, we addressed this limitation by incorporating functional enrichment analyses (GO and KEGG) to link the identified biomarkers with specific biological processes and pathways. This approach not only strengthens the biological relevance of the biomarkers but also provides insights into their potential roles in disease mechanisms, paving the way for future experimental validation (Nguyen et al., 2024).
Building on these analyses, we further explored the biological context of the identified hub genes using the NetworkAnalyst platform, an online tool for comprehensive data analysis (Guangyan et al., 2019). This platform provided access to resources such as the ChEA database to predict TFs associated with the hub genes and the TarBase database to identify miRNAs linked to these genes. By integrating these analyses, our approach enhances the understanding of the pathways and regulatory networks in which the hub genes operate, shedding light on the potential mechanisms through which they influence disease processes.
When we delve into diseases of the central nervous system, we find unexpected associations between GBM and degenerative CNS diseases. Although there are clear differences in clinical presentation and pathogenesis between these two types of diseases, they show striking similarities in certain critical biological processes. Ferroptosis, as a key mode of cell death, plays a crucial role in both diseases. In degenerative neurological diseases such as PD, iron deposition and consequent oxidative damage are among the critical pathological processes (Martin-Bastida et al., 2017; Mitre et al., 2022). Further studies have shown that intervening in the metabolic pathway of iron can not only bring new therapeutic strategies for Parkinson’s disease (Wang et al., 2023), but also provide new ideas for the treatment of glioblastoma.
Lipid rafts serve as dynamic hubs for integrating signaling events in both tumor and neuronal cells, including those associated with RELN and Wnt signaling (Yunden and Ven, 2024). These lipid-rich microdomains play a pivotal role in organizing and coordinating the interaction of various signaling pathways, including the Wnt pathway, which intersects with RELN signaling through LRP6. This intersection is particularly significant in both tumor progression and neurodevelopmental processes (Lenz et al., 2024). In the case of tumors, including GBM, lipid rafts enable the integration of multiple pro-survival and growth-promoting signals, enhancing cellular responses to external stimuli (Reza et al., 2023; Pampa et al., 2024). Similarly, in neuronal cells, lipid rafts facilitate the RELN-mediated signaling crucial for processes like neuronal migration and synaptic plasticity, by coordinating the activation of downstream molecules such as Dab1 (Matthew et al., 2022). Additionally, oxidative stress is modulated within these lipid domains, influencing cellular fate decisions by affecting both Wnt and RELN pathways (Akira et al., 2024). Alterations in lipid raft composition or function have been shown to disrupt this integration, contributing to both tumorigenesis (Momoko et al., 2023) and neurodegeneration (Yi Sak et al., 2024). Therefore, lipid rafts are indispensable for coordinating complex signaling networks in these vital biological processes.
While glioblastoma and degenerative central nervous system diseases differ fundamentally in their molecular mechanisms and pathological outcomes—such as cell death in degenerative CNS diseases (Salem et al., 2025) versus cell proliferation and resistance to cell death in cancer (Talvard-Balland et al., 2024)—our findings suggest that certain molecular pathways, such as those involving iron metabolism, may be shared (Brown et al., 2024). This observation broadens our understanding of these distinct neurological disorders and suggests potential overlapping therapeutic targets. Based on this, we investigated markers common to degenerative CNS diseases and GBM, identifying RELN and GSTO2 as key candidates.
RELN is a large extracellular matrix protein first identified during brain development, crucial for regulating neuronal migration and cortical stratification. It involves processes like neuronal migration, dendritic outgrowth, dendritic spine formation, synapse generation, and synaptic plasticity (Khialeeva and Carpenter, 2017). Research has shown that RELN is expressed in the adult brain and plays roles in neuronal plasticity, memory formation, and various neurological diseases. Additionally, RELN impacts the immune system, liver fibrosis, and cancers (Canet-Pons et al., 2018).
In addition to their general role in cellular signaling, lipid rafts are essential for the precise regulation of the RELN/ApoER2 signaling pathway, which is crucial for neuronal migration, synaptic plasticity, and neurodevelopment (Ling Xiao et al., 2024). These lipid-enriched microdomains serve as platforms that facilitate the clustering of signaling receptors such as ApoER2, enhancing the signal transduction efficiency. The interaction between RELN and ApoER2 within lipid rafts has been shown to modulate the activation of downstream signaling molecules, including Dab1, a key kinase involved in neuronal migration (Mitsuki et al., 2024). Recent study had demonstrated that lipid rafts play a pivotal role in ensuring that this signaling cascade is both robust and specific, thereby facilitating the proper formation of neural networks (Harald et al., 2005). Moreover, lipid rafts may influence the trafficking and internalization of ApoER2 receptors, which is necessary for effective signal initiation.
Disruption of lipid rafts can lead to a failure in receptor clustering and signaling efficiency, impairing essential neurodevelopmental processes like synaptogenesis and axon guidance. Emerging evidence suggests that such disruption could contribute to various degenerative CNS diseases by altering synaptic integrity and neuronal survival. Notably, research on AD and other neurodegenerative conditions has revealed that the perturbation of lipid raft domains can impair RELN/ApoER2 signaling, resulting in deficits in memory, learning, and neuroplasticity (Christina M et al., 2024). Furthermore, alterations in lipid raft composition and function are thought to affect the balance between neuroprotection and neurodegeneration, implicating lipid rafts as a key player in both disease onset and progression. These findings underscore the critical role of lipid rafts not only in the RELN/ApoER2 pathway but also in the overall maintenance of neuronal function and resilience (S Sakura et al., 2011).
RELN expression and function are crucial in degenerative CNS diseases such as AD, PD, and Huntington’s disease. Research indicates that RELN expression is typically decreased in these diseases, contributing to pathological processes such as the loss of neuronal connectivity, neuronal death, and cognitive decline. These findings suggest that RELN’s functional activity may be inhibited, further exacerbating disease progression. (Lidón et al., 2020). RELN’s mechanism involves neurodevelopment and neuronal plasticity. Core components of the RELN pathway include VLDLR, ApoER2, Src family kinases, and Dab1 (Jossin, 2020). RELN binds to its receptors (LRP1/2 and ApoER2/VLDLR), activating downstream pathways like Dab1. This activation is essential for regulating intracellular calcium levels, reorganizing the cytoskeleton, and maintaining synapses (Bock and May 2016). In AD and other neurodegenerative disorders, reduced RELN signaling can alter synaptic structure and function, affecting neural network stability and plasticity, leading to cognitive decline. Reduced RELN signaling may impair NMDA receptor function, crucial for learning and memory (Joly-Amado et al., 2023).
The role of RELN in tumors is complex. Research shows that it plays a role in regulating the invasion and proliferation of cancer, and inhibits the migration and invasion of pancreatic cancer cells (Ramaker et al., 2017). The expression of RELN decreased in breast cancer, colorectal cancer and pancreatic cancer, but increased in retinoblastoma, myeloma and prostate cancer (Ndoye et al., 2021; Biamonte et al., 2021; Seigel et al., 2007; Qin et al., 2017). Some studies have found that the expression level of RELN in GBM is different from that of normal brain tissue or other types of brain tumors, which may be related to tumor invasiveness, growth rate, and treatment response. (Perrone et al., 2007). Moreover, recent studies have suggested that inhibiting heparanase, an enzyme involved in the remodeling of the extracellular matrix (Minghong et al., 2025), may significantly affect the regulation of autophagy and apoptosis in glioblastoma (Valeria et al., 2023). Heparanase inhibition has been shown to alter the balance between cell survival and cell death by modulating key signaling pathways involved in these processes (Vy M et al., 2017). By interfering with the autophagic flux and promoting apoptosis, heparanase inhibitors may enhance the sensitivity of glioblastoma cells to therapeutic agents, providing a promising approach for treatment strategies (Carolyn G and Renato V, 2022).
In GBM, the expression of RELN and its main downstream effector molecule Dab1 is inhibited, and mRNA expression is inversely proportional to the degree of malignancy (Talebian et al., 2019). In addition, RELN expression was positively associated with survival in patients in two large independent clinical annotation datasets. Silencing of RELN occurs through promoter hypermethylation (Serrano-Morales et al., 2017). At the functional level, RELN regulates GBM cell migration in a Dab1 (tyrosine phosphorylation) dependent and non dependent manner, depending on the substrate provided. Activation of RELN signaling significantly reduces the proliferation of GBM cells, a phenotype dependent on RELN stimulation of Dab1. Mutants lacking all RELN induced tyrosine phosphorylation sites (DAB1-5F) failed to induce growth arrest. Proteomic analysis shows that these effects are mediated by reducing the dephosphorylation of E2F targets and ERK1/2. RELN may also indirectly affect tumor growth and spread by affecting astrocytes and microglia in the tumor microenvironment. This interaction is achieved by altering the secretion of cytokines and chemokines, thereby affecting the interaction between tumor cells and their microenvironment (Jandial et al., 2017; Zeng et al., 2023).
GSTO2 is a member of GST family, belonging to the Omega subfamily (Peng et al., 2023). This family plays a critical role in cellular detoxification processes. GST enzymes, found in animals, plants, and microorganisms, protect cells from oxidative stress and chemical damage by facilitating the conjugation of glutathione with various electrophilic compounds, including drugs, environmental toxins, and metabolic by-products, thereby promoting their excretion (Zhang et al., 2013).
Degenerative CNS diseases are often associated with increased oxidative stress and inflammation (Schmuck et al., 2005). Oxidative stress during these diseases can lead to cellular damage and death, accelerating disease progression. GSTO2, through its antioxidative activity, may help neutralize excess free radicals and reactive oxygen species, alleviating oxidative stress-induced damage to neurons and other brain cells. Studies suggest a significant correlation between GSTO2 and the age of onset in AD and PD (Ozturk et al., 2005). Cha et al. (2023) found that knocking down the GSTO2 gene in fruit flies leads to an age-related increase in Cabeza protein levels in neurons. Cabeza, a homolog of the human FUS protein associated with degenerative CNS diseases like ALS and Frontotemporal Dementia, shows increased mislocalization and aggregation in the cytoplasm of neurons and reduced solubility in aging neurons when GSTO2 is knocked out. This suggests GSTO2 plays a crucial role in regulating Cabeza localization and aggregation, potentially impacting neurodegenerative disease development (Machamer et al., 2018).
Abnormal expression of GSTO2 is also related to the occurrence and development of tumors. Interestingly, GSTO2 exhibits a bidirectional regulatory role in tumors. Several studies (Peng et al., 2023; Masoudi et al., 2009; Djukic et al., 2015) have discussed the association of GSTO2 with cancers. Sumiya R (Sumiya et al., 2022) revealed that GSTO2 is uniquely expressed in various lung stem cells but silenced in lung squamous cell carcinoma (LSCC) due to DNA hypermethylation. GSTO2 regulates cell growth, β-catenin expression, and mitochondrial respiration through p38 phosphorylation. Activation of p38 MAPK by GSTO2 leads to the downregulation of β-catenin, likely via ubiquitination, contributing to lung tissue homeostasis and preventing malignant transformation. Loss of GSTO2 in LSCC disrupts this mechanism, allowing β-catenin to avoid degradation, leading to increased mitochondrial oxidative phosphorylation and enhanced energy production, promoting LSCC growth. These findings highlight the potential importance of GSTO2 in preventing LSCC and provide insights into lung carcinogenesis pathways. However, the specific role and mechanism of GSTO2 in GBM require further investigation. It is hypothesized that GSTO2, by regulating oxidative stress responses and metabolic processes, could influence GBM cell survival, proliferation, and treatment response. Upregulation of GSTO2 might enhance tumor cells’ ability to clear chemotherapeutic drugs, reducing treatment efficacy. Conversely, reducing GSTO2 activity or expression could increase oxidative stress, promoting cancer cell death, and potentially serving as a therapeutic strategy.
Although RELN and GSTO2 have different functions and regulatory mechanisms, studies suggest they may have potential interrelated mechanisms in certain diseases. RELN plays a crucial role in neural development and synaptic plasticity, and its abnormal expression is associated with degenerative CNS diseases. RELN combats oxidative stress by regulating synapse formation and stability (Sajjad et al., 2014), while GSTO2 is important in the cellular antioxidant response, helping to neutralize excess free radicals and ROS, thereby reducing neuronal damage (Lauren E et al., 2019). RELN and GSTO2 may jointly regulate oxidative stress and inflammatory responses in neurons, protecting them from damage. Additionally, RELN affects the stability and plasticity of neural networks by regulating synapse structure and function (Murat S et al., 2009), whereas GSTO2, through its antioxidant activity, reduces apoptosis and autophagy induced by oxidative stress, thus protecting neuronal survival. They may co-operate in the pathways of neuronal apoptosis and autophagy, maintaining normal neuronal function and survival (Xifeng et al., 2022). The signaling pathways involving RELN and GSTO2 may interact at certain points; for instance, RELN influences intracellular calcium levels by regulating synaptic function and stability (Yehui et al., 2019), while GSTO2 maintains cellular homeostasis by mitigating oxidative stress (Tatjana et al., 2022). They may influence each other in cellular signal transduction, jointly maintaining neuronal function and survival.
Clinically, our findings indicate that RELN and GSTO2 show significant expression changes in GBM, suggesting their potential utility as biomarkers for this disease. Specifically, the observed differential expression of these genes in GBM samples compared to controls highlights their diagnostic relevance. However, we acknowledge that further experimental and clinical validation is essential to substantiate their roles in early diagnosis, disease progression monitoring, and treatment response evaluation.
Regarding neurodegenerative CNS diseases, we did not evaluate RELN and GSTO2 in relevant cohorts; thus, their potential roles in neurodegenerative CNS diseases remain speculative. While shared molecular mechanisms between neurodegenerative CNS diseases and GBM suggest possible connections, we refrain from making definitive claims about their biomarker or therapeutic potential in these diseases without supporting evidence. Future research should validate these findings in disease-specific cohorts and investigate whether RELN and GSTO2 could serve as progression markers or therapeutic targets for GBM and other CNS diseases.