Traumatic brain injury (TBI) is defined as a condition that disrupts brain function or causes other brain lesions due to external physical forces (1). It is estimated that the annual incidence of TBI is 50 million cases worldwide, with roughly half of the population likely to experience a TBI in their lifetime. Additionally, morbidity and mortality rates are higher in low- and middle-income countries. TBI results in approximately US$400 billion in economic losses globally each year, accounting for 0.5% of the world’s gross product (2).

Currently, assessing the extent of injury and predicting outcome in TBI patients remains challenging. The Glasgow Coma Scale (GCS) and Abbreviated Impairment Scale (AIS) correlate with injury severity and provide prognostic power. However, these scales rely on the medical staff’s subjective interpretation of the patient’s injury and brain function, and their accuracy can be affected by factors such as age and other variables (3–5). Biomarkers offer an unbiased and independent assessment of TBI severity and patient prognosis. Studies had shown that glial fibrillary acidic protein (GFAP), S100β and soluble urokinase plasminogen activator receptor (suPAR) have been used to evaluate the survival rate of TBI patients (6–8). Additionally, tau protein and GFAP have been used to assess GCS scores at discharge and Glasgow Outcome Scale (GOS) scores at 6 or 12 months (6, 9). Ubiquitin C-terminal hydrolase (UCH-1), matrix metalloproteinase 9 (MMP9) and MMP2 are also believed to to be related to TBI severity (10, 11). Most of these biomarkers can be measured from plasma or serum samples, minimizing the challenges of cerebrospinal fluid (CSF) collection, but rapid turnaround remains a concern in the critical TBI settings.

With the rapid advancement of high-throughput technologies and bioinformatics, we can now screen biomarkers and therapeutic targets more efficiently. Previous bioinformatics studies have primarily identified biomarkers through differential gene and protein-protein interaction (PPI) network analysis (12). However, the accuracy of PPI network analysis has been questioned. With evolving analysis methods, tools like Cytoscape and its plugins are increasingly used to screen potential disease targets. Cytoscape is an open-source software that integrates biomolecular interaction networks with high-throughput expression data and other molecular states into a unified conceptual framework (13).

In this study, we first obtained the TBI microarray dataset from the Gene Expression Omnibus (GEO) public database to analyze and identify differentially expressed genes (DEGs). We then combined bioinformatics analysis with machine learning strategies to deeply screen and identify key features of TBI. Receiver operating characteristic (ROC) analysis was used to evaluate their diagnostic values. Recently, accumulating evidence suggests that immune and inflammatory mechanisms play crucial roles in preventing the pathogenesis of TBI (14, 15). Therefore, we aimed to explore the relationship between biomarker genes and immunity. We applied cell-type identification by estimating relative subsets of RNA transcript (CIBERSORT) to clarify the differences in immune infiltrates in TBI. Subsequently, we verified the expression of biomarkers in the brain tissue of TBI rat’s model. In recent years, pan-cancer analyses have enabled us to leverage human genome sequence data, along with a vast compendium of associated molecular and phenotypic features and functional genomic data from genome-scale expression and epigenomics. This helps us understand the impact of variants and elucidate the contribution of dysregulated genes to specific diseases and biological pathways in tissue contexts (16). Finally, we analyzed the expression and immune cell infiltration of KDM5D in various cancers, which may significantly enhance our understanding of the biological function of KDM5D. We propose that KDM5D can serve as a common biomarker for predicting and diagnosing TBI and pan cancer. However, this statement has certain flaws, such as distinguishing patients who suffer from both TBI and cancer, understanding the mechanism of KDM5D upregulation changes, and combining other biomarkers of these two diseases may be a more reliable clinical strategy.

In summary, this study demonstrates that KDM5D is a robust and feasible new biomarker for TBI, providing a potential target for the diagnosis, prevention, and treatment of TBI.

TBI has the highest incidence among common neurological diseases and poses a significant public health burden. TBI is recognized not only as an acute disease but also as a chronic one with long-term consequences, including an increased risk of delayed neurodegeneration. Although blood-based protein biomarkers have proven important in the assessment of TBI, most available assays are for research use only (27). Identifying more meaningful biomarkers is crucial for the diagnosis, treatment and prognosis of TBI. Transcript profiling is a promising tool for uncovering molecular mechanisms and biomarkers. Fortunately, we obtained transcriptome data of traumatic brain tissues from TBI patients in the GEO database. Damaged brain tissue not only reflects the pathological changes of the disease process (more clinical effects), but is also significant for the screening key genes. Additionally, for TBI patients, transcriptomic analysis and validation of brain tissue can more accurately describe the physio pathological changes of TBI.

In this study, we first identified the differential genes co-expressed by four brain areas of TBI and non-TBI patients through limma package, and obtained 44 DEGs ( Figure 2 , Supplementary Table S1 ). Subsequently, we explored the common biological functions and signaling pathways involved in these DEGs through GO and KEGG enrichment analysis. The results indicated that the mechanisms of TBI and non-TBI are mostly related to molecular modifications and signal transduction pathways, which could be the key mechanisms connecting the two.

The prominent enrichment of hemoglobin complex genes (HBA1/HBB) among our DEGs suggests an intriguing connection between erythrocyte extravasation and neuroinflammation post-TBI ( Figure 3 ). Our co-expression analysis revealed strong correlations between HBA1 expression and microglial activation markers, supporting recent findings that hemoglobin breakdown products can activate TLR4 signaling in microglia (28). Clinically, we observed significant associations between HBB expression and both interleukin-6 (IL-6) levels and peripheral NOD-like Receptor (NLR) (29), suggesting these genes may serve as biomarkers for systemic inflammatory response after TBI. This aligns with Chen’s (30) demonstration of hemoglobin-driven inflammasome activation in subarachnoid hemorrhage models. TBI-induced brain damage manifests as excitotoxicity, neuroinflammation, cytokine damage, oxidative damage, and ultimately cell death (31). Studies have shown that it can prevent neuronal death and dysfunction in TBI by using of calpain inhibitors (AK295, SJA6017) and neurotrophic factors (NGF, BDNF) (32). Brain damage disrupts normal signaling pathways, leading to changes in brain function. Normally, in the central nervous system (CNS), signals are transmitted by various neurotransmitters such as gamma-aminobutyric acid (GABA), glutamate, glycine, norepinephrine, dopamine and serotonin. TBI may alter the levels of these neurotransmitters, ultimately disrupting the brain’s normal function (33). In summary, through enrichment analysis and previous studies, we believe that molecular modifications and signal transduction pathways are critical mechanisms of TBI.

To further identify key genes, we used Venn diagrams to show the distribution and overlap of DEGs in four brain regions (FWM, HIP, PCx, TCx), identifying 14 DEGs. We determined the protein interaction relationship of those DEGs through the STRING database, revealing that 11 DEGs were closely related ( Figures 4a, b ). However, we still could not obtain the Hub gene. Further analysis using the CytoHubba module algorithm analysis in Cytoscape identified 3 hub genes, including RPS4Y1, KDM5D and NLGN4Y ( Figures 4c–h ). Notably, validation results from external data sets (neuronal exosomes samples from serum) showed that RPS4Y1, KDM5D and NLGN4Y have high diagnostic value, with KDM5D having the highest diagnostic value (AUC=0.90) ( Figure 5 ). This suggests that these hub genes can serve as biomarkers in both damaged brain tissue and serum, demonstrating high diagnostic significance.

Ribosomal protein S4 Y-linked 1 (RPS4Y1) is a member of the ribosomal protein S4E family and is encoded as ribosomal protein S4 located on chromosome p11.31. RPS4Y1 is ubiquitously expressed and plays an important role in correct development (34, 35). RPS4Y1 has been used as a marker for a variety of diseases, such as the disease resistance gene for corticosteroid combined with cyclosporine A treatment in Vogt-Koyanagi-Harada, and as a candidate gene in multiple sclerosis (36, 37). The latest single-cell transcriptome sequencing found that RPS4Y1 was up-regulated in four cell clusters: oligodendrocytes, microglia, astrocytes and neurons in post-traumatic brain tissue (26), which is consistent with our findings. Furthermore, there are studies suggesting that abnormal expression of RPS4Y1 may affect the signal transduction of signal transducer and activator of transcription 3 (STAT3). For instance, in preeclampsia induced hypertension, abnormal expression of RPS4Y1 impairs STAT3 signaling, thereby inhibiting the migration and invasion of trophoblast cells (34). STAT3 is known to have a protective effect on ischemic brain injury; however, it is still unclear which type of brain cell mediates this effect and through what mechanism. Research has shown that endothelial STAT3 may help prevent ischemic brain injury by protecting the endothelial function of cerebral blood vessels and the integrity of the blood-brain barrier (38). Therefore, it’s conceivable that abnormal expression of RPS4Y1 in TBI may cause dysfunction of cerebral arterial endothelial cells. For example, if RPS4Y1 overexpression damages the protective effect of STAT3 on cerebral blood vessels, it may exacerbate TBI. Identifying the potential mechanisms of RPS4Y1 and STAT3 action is a novel research direction for future studies.

Neuroligin 4Y (NLGN4Y) is a member of the neuroligin gene family located in Yq11 (39). Previous research found that NLGN4Y gene mutations play a crucial role in male homosexuality and autism. These mutations often lead to severe neurodevelopmental disorders (40, 41). In a report of pediatric traumatic brain injury, NLGN4Y may be involved in the convergence of sex differences with TBI, but no further studies have been conducted (42). In short, NLGN4Y as a gene related to neurodevelopment, deserves further exploration for its connection with TBI.

Lysine-specific demethylase 5D (KDM5D), a member of the KDM5 family, is a male-specific histone demethylase (43). Histone methylation and demethylation play crucial roles in gene regulation and biological events during disease development, including neurological disorders and multiple cancer types. Similar to other KDM5 family members, KDM5D enables dimethyl and trimethyl H3K4 demethylation. KDM5 is involved in the pathogenesis of various cancers by inhibiting the expression of certain genes at the transcriptional level (44–46). TBI can lead to cognitive decline and impaired memory consolidation, as the cerebral cortex and hippocampus are critical for controlling cognitive functions and social behavior (47, 48). In mice, Kdm5d is expressed in the cortex and hippocampus, which may be associated with the prevalence of multiple neurological and psychiatric disorders, such as multiple sclerosis, schizophrenia and autism (48). In our study, we found that compared with non-TBI, the expression of KDM5D in the four brain regions of TBI patients showed an upward trend, with significant difference in FWM and PCx ( Figures 6a–d ). Immunofluorescence experiments on brain tissue of TBI rats also showed that Kdm5d expression significantly increased in the cortical area, but no obvious difference in the hippocampus ( Figures 6e–h ). It is worth noting that Kdm5d is normally expressed in the cortex of the control rats. KDM5D is known to be physiologically expressed in the brain, particularly in the cortex, according to the Human Protein Atlas and functional studies (https://www.proteinatlas.org/). This basal expression likely reflects KDM5D’s normal roles in cortical functions, such as epigenetic regulation of gene expression and maintaining neuronal homeostasis (49, 50).The results of clinical trials and animal experiments are consistent, indicating that KDM5D expression changes after TBI occur mainly in cortical areas. Our key finding is the significant upregulation of KDM5D in the cortex following TBI, not its mere presence in the control brain. It is this marked increase in KDM5D expression in the context of TBI that points to its pathological role and diagnostic potential.

KDM5D encodes the histone H3 demethylase on K4, and H3K27me3 demethylase has been confirmed to activate microglia to produce inflammation through STAT and TLR signaling pathways (48, 51). This is similar to microglial activation after TBI (52). In addition, the epigenetic regulatory function of KDM5D, as a histone demethylase, suggests a potential mechanism for sustained changes in gene expression in microglia, potentially leading to prolonged activation (53). H3K27me3 was known to induce microglia to express pro-inflammatory genes through TLR (54), suggesting that KDM5D-mediated demethylation at specific gene loci in microglia could result in the sustained upregulation of pro-inflammatory mediators or the downregulation of genes responsible for resolving inflammation, thus contributing to a chronic activated state.

TLR signaling, particularly TLR4 which is known to be activated by various damage associated molecular patterns (DAMPs) released after TBI, is a potent inducer of pro-inflammatory responses in microglia. If KDM5D activation in microglia promotes TLR signaling, it could amplify the release of a range of DAMPs typically associated with TLR activation, such as High Mobility Group Box 1 (HMGB1), heat-shock proteins and extracellular ATP (55, 56). These DAMPs, in turn, can further propagate neuroinflammation and maintain prolonged microglial activation, potentially creating a feed-forward loop that sustains chronic inflammation in the TBI milieu (56–58).

Moreover, it’s conceivable that KDM5D-mediated epigenetic modifications could specifically influence the expression of genes encoding for certain types of DAMPs in microglia. For instance, if KDM5D demethylates and activates the promoters of genes encoding for pro-inflammatory cytokines or chemokines that can act as DAMPs (e.g., TNF-α, IL-1β, CCL2), this could result in a sustained release of these specific DAMPs, contributing to prolonged inflammation and microglial activation. Further research is warranted to investigate the precise DAMPs released by microglia upon KDM5D activation and TLR signaling in the context of TBI, and to elucidate the epigenetic mechanisms by which KDM5D regulates their expression. Identifying these specific DAMPs and the underlying mechanisms will be crucial for developing targeted therapeutic strategies to modulate microglial activation and neuroinflammation in TBI.

As mentioned before, we found that the expression of KDM5D in the cortex and hippocampus was completely inconsistent. We speculated that this may be related to immune cells. Through immune infiltration analysis of PCx and HIP brain regions in GSE104687, it was found that the expression of neutrophils in PCx was reduced in TBI ( Figure 7 ). Previous studies have shown that TBI is accompanied by an increase in neutrophils, but most studies are limited to patients with mild and acute TBI or animal experiments. There are few studies on patients with severe or even fatal TBI (59). Recent reports suggest that neutrophil plasticity and crosstalk with other cells also complicate their function in TBI. Therefore, it is premature to conclude whether neutrophils are beneficial or detrimental in TBI. Neutrophils act in a well-defined and limited manner, as the protective or deleterious effects of neutrophils depend on the stage and type of injury, with which they interact. However, the molecular mechanisms of how the brain environment modulates neutrophil function and how neutrophil function affects brain injury and repair have not been well elucidated. More research is required to elucidate the complex roles of neutrophils at various stages of TBI and the underlying mechanisms. Future advancements may pave the way for the developing novel therapeutic strategies aimed at selectively eliminate the harmful effects of neutrophils, while preserving the beneficial impacts in TBI (60). Studies have showed that KDM5D is involved in neutrophil degranulation, activation, and neutrophil-mediated immune response, but the detailed molecular mechanism remain unreported (61). This finding may offer a promising direction for the diagnosing and prognosticating of TBI. Cancer is one of the most challenging global health concerns, sharing metabolic and risk factor characteristics with other diseases. Therefore, it is plausible that TBI biomarkers could also be used in tumor prediction strategies. Significant down-regulation of KDM5D was observed in 23 types of tumors, whereas up-regulation occurred in only 3 types of tumors ( Figure 8a ). Exploring KDM5D expression in cancer infiltration, we observed significant correlation with immune infiltration levels in 10 cancers ( Figure 8b ). This suggested a potentially adverse prognostic impact of effect.

At present, there is limited literature directly studying the role of KDM5D in TBI. However, based on its known functions in other diseases such as cancer and cardiovascular disease (CVD), further speculation on the potential mechanisms of KDM5D in TBI (such as cerebral vascular injury, endothelial dysfunction, or vascular inflammation) is particularly interesting. A proteomics study of CVD found that overexpression of KDM5D in patients with CVD resulted in reduced protein levels of their substrate (H3K4me3) and dysfunction of vascular endothelial cells (62). In TBI, cerebral vascular injury may lead to local ischemia and hypoxia, which in turn can cause changes in epigenetic modifications (63). Overexpression of KDM5D may exacerbate cerebral vascular injury by reducing H3K4me3 levels, inhibiting the expression of vascular repair and protective genes, such as endothelial nitric oxide synthase (eNOS) and phosphate and tension homology deleted on chromosome ten (PTEN). And KDM5D was found to be related to endothelial injury and neurological inflammation of cerebral microvessels (64, 65), although it did not directly indicate the role of KDM5D in TBI, which provides a new idea for our subsequent in-depth research.

While our findings highlight KDM5D as a promising diagnostic biomarker for TBI, we should acknowledge a significant limitation: the potential overlap in KDM5D expression with cancer. KDM5D is also expressed and dysregulated in various cancers, raising concerns about its specificity as a TBI-specific biomarker. Indeed, studies have reported altered KDM5D expression in a wide range of malignancies, with downregulation observed in clear cell renal cell carcinoma (66) and upregulation in colon cancer (67). This complex expression pattern in cancer underscores the challenge of relying solely on KDM5D to differentiate TBI from cancer, or even to diagnose TBI in patients with pre-existing or concurrent cancer.

In cancer, KDM5D alterations are often linked to epigenetic reprogramming and oncogenic signaling pathways, contributing to tumor development and progression (68). In contrast, in TBI, we propose that KDM5D upregulation is primarily triggered by neuroinflammation, cellular stress, and the cascade of secondary injury events following the initial trauma. This difference in upstream triggers could potentially lead to distinct downstream effects and temporal expression patterns of KDM5D in TBI versus cancer. To address the specificity limitations of KDM5D as a standalone TBI biomarker, we propose a multi-faceted diagnostic strategy. Firstly, combining KDM5D with established TBI biomarkers, such as Glial Fibrillary Acidic Protein (GFAP) and Neurofilament Light chain (NFL) (69), which exhibit high axonal specificity and are less likely to be influenced by cancer could significantly enhance diagnostic accuracy.

Secondly, incorporating cancer-specific markers, such as Carcinoembryonic Antigen (CEA) and Cancer Antigen 125 (CA125) (70), for individuals at risk or suspected of having cancer, could aid in differential diagnosis. Thirdly, integrating biomarker data with clinical findings and neuroimaging techniques would provide a more comprehensive diagnostic picture, especially in complex scenarios.

A particularly challenging clinical scenario arises when an individual presents with symptoms potentially indicative of both TBI and cancer, or when a patient with known cancer sustains a head injury. In such cases, relying solely on KDM5D might lead to misdiagnosis or an inaccurate assessment of injury severity or cancer progression, as KDM5D levels could be elevated due to either or both conditions. Our proposed multi-biomarker and multi-modal approach is specifically designed to mitigate this challenge and improve diagnostic accuracy in these complex clinical presentations.

Based on the bioinformatics analysis and animal experimental verification, we propose that KDM5D could serve as a potential function as a co-biomarker for predicting and diagnosing TBI and pan-cancer. Additionally, investigating KDM5D may lead to the development of novel serum markers, providing new indicators for clinical liquid biopsy and potentially enhancing strategies for TBI and cancer prevention.