Traumatic brain injury (TBI) is one of the leading causes of death and long-term disability worldwide, imposing a heavy burden on society (1). The pathophysiology of TBI is complex. Following the primary mechanical injury, a series of secondary damages are triggered, including neuroinflammation and excitotoxicity (2, 3). This process can last from hours to days and greatly impacts neurological outcomes. Therefore, identifying biomarkers that can objectively and dynamically reflect the pathophysiological progression and prognosis of TBI is crucial for achieving precise disease assessment and intervention.

Accurate determination of initial brain injury following TBI is essential for establishing neurological prognosis and balancing the risks and benefits of treatment options. Currently, clinical evaluation of TBI mainly relies on the Glasgow Coma Scale (GCS) and computed tomography (CT) imaging. However, GCS scores may be affected by factors such as sedation, alcohol, or shock, while CT, although effective in identifying macrostructural damage, has limited sensitivity for diffuse axonal injury or early secondary damage (4–6). More importantly, the pathophysiological process of TBI is progressive, and accurately assessing and quantifying the severity of the initial and subsequent secondary damage remains challenging (7). Against this backdrop, serum biomarker detection has emerged as a promising supplementary approach due to its non-invasive and dynamic monitoring characteristics. Current research has established several biomarkers associated with TBI. For instance, glial fibrillary acidic protein (GFAP), as a specific marker of astrocyte injury, has been proven to effectively distinguish CT-positive lesions and correlate with injury severity (8, 9). However, there is still controversy over its prognostic predictive efficacy, and most studies have focused on the population with mild to moderate TBI (10–12). Other classic markers such as S100B have their clinical accuracy and practicality limited due to their short half-life and relatively insufficient central specificity (13). Ubiquitin C-terminal hydrolase L1 (UCH-L1), although neuronal specific, may yield false positives in orthopedic trauma and is also expressed in neurodegenerative diseases, which limits its diagnostic specificity (14, 15). Overall, existing biomarkers mainly reflect acute structural damage and have limited ability to identify the dynamic process of secondary injury that drives prognostic outcomes.

Milk fat globule-EGF factor 8 (MFGE8), also known as lactadherin, is a secreted glycoprotein that plays a critical role in apoptotic cell clearance and inflammatory regulation. In the central nervous system, MFGE8 is predominantly expressed by microglia, with lower levels also detected in astrocytes and neurons (16, 17). In animal models of TBI and subarachnoid hemorrhage, the expression of MFGE8 in the brain is significantly upregulated, and exogenous supplementation of MFGE8 has been confirmed to alleviate neuroinflammation and promote the recovery of neurological function (18, 19). Mechanistically, MFGE8 acts as a bridging molecule by binding phosphatidylserine exposed on apoptotic cells or pathological extracellular vesicles, while simultaneously interacting with integrin αvβ3 on phagocytes, thereby facilitating efficient efferocytosis and limiting the accumulation of pro-inflammatory mediators (20–22). In addition, MFGE8 may exert potential anti-apoptotic and neuroprotective effects through binding to integrin-β3 and activating downstream signaling pathways, including the FAK/PI3K/AKT axis (18). Collectively, these findings suggest that endogenous MFGE8 upregulation may represent a crucial compensatory repair response to neural injury. Unlike GFAP, which mainly reflects astrocytic injury, MFGE8 is more likely associated with immune-inflammatory regulatory processes mediated by microglia or macrophages (23). This biological distinction indicates that MFGE8 and GFAP may reflect different dimensions of the pathological process in TBI, offering complementary potential. However, there is still a lack of clinical transformation evidence of serum MFGE8 in human TBI, especially whether its predictive efficacy can match or be complementary to GFAP, which requires systematic studies.

To address these gaps, we hypothesized that serum MFGE8 levels after traumatic brain injury exhibit time-dependent changes, are associated with injury severity, and relate to short-term neurological outcomes. To test these hypotheses, serial blood samples were collected from patients with TBI, and MFGE8 was systematically compared with GFAP. Linear mixed-effects models (LMMs) were applied to characterize longitudinal trajectories and predefined between-group differences based on GCS and Rotterdam CT severity stratifications, as well as their associations with short-term neurological outcomes. In addition, time-specific exploratory ROC analyses were conducted to compare the discriminatory performance of MFGE8 and GFAP at different post-injury time points.

This prospective study systematically compared the dynamic changes in the novel biomarker MFGE8 with the classic biomarker GFAP following TBI, using continuous-time point serological testing. It assessed their predictive value for injury severity and neurological outcomes. Longitudinal analyses based on LMMs demonstrated that serum MFGE8 levels significantly increased after TBI, peaking at approximately 24 h post-injury, and that the dynamic trajectory of MFGE8 was significantly associated with both clinical and radiological severity, as well as with 30-day mortality and 3-month neurological outcomes. These longitudinal associations were comparable to those observed for GFAP. Time-specific ROC analyses further supported the discriminatory ability of MFGE8 for neurological outcomes, with the most favorable performance observed at approximately 24 h post-injury. Collectively, these findings support MFGE8 as a TBI biomarker with considerable potential.

GFAP is a biomarker that has been approved by the U. S. Food and Drug Administration (FDA) in recent years for emergency screening for traumatic intracranial abnormalities (30). It also performs well in distinguishing TBIs of different severity (9). Therefore, this study selected GFAP as a reference for the value evaluation of MFGE8. TBI patients can be treated at different times after injury. When should these biomarkers be used for clinical assessment of TBI patients? First of all, the time distribution of GFAP and MFGE8 in serum should be clearly understood. In this study, the first blood sample was collected within 12 h after injury (median of 4 h), and the last sampling was 72 h after injury, which ensured the reliability of capturing the changes in the acute phase. As a cytoskeletal protein of astrocytes, GFAP levels increased significantly at admission and peaked at 24 h, which is consistent with previous studies (31, 32). GFAP concentration at admission showed the strongest correlation with the initial GCS score, consistent with its characteristics as a marker of acute structural injury. MFGE8, a multifunctional glycoprotein, plays an important role in central nervous system diseases by mediating the clearance of apoptotic cells and regulating neuroimmune inflammation. For instance, in neurodegenerative diseases, microglia clear damaged neurons and their fragments through MFGE8 to maintain the health of the neural network (33). In neuroinflammatory processes, MFGE8 can promote the clearance of apoptotic cells and reduce the inflammatory response by binding to phosphatidylserine and integrin receptors (34). This has also been confirmed in TBI and ischemic stroke models (18, 35). Given that neuroinflammation and apoptosis are core pathological processes in secondary brain injury after TBI, the dynamic changes of serum MFGE8 may reflect the biological functional state related to secondary brain injury after TBI to some extent. This study observed that MFGE8 and GFAP showed similar early elevation and peaking trends, and there was a moderate positive correlation between the two in the acute phase (Spearman’s ρ = 0.424–0.602). On one hand, this suggests that there may be a close link between astrocytic damage (GFAP release) and subsequently activated immunomodulatory response (MFGE8 level changes). On the other hand, it raises the question of the source of serum MFGE8. We speculate that its origin might be twofold. First, it may be directly released from central nervous system cells following blood–brain barrier (BBB) disruption. Second, it might originate from a systemic compensatory response to central nervous system injury. Some studies have found that neuroinflammatory responses can further damage BBB integrity, allowing central signals to enter the peripheral circulation (36). Therefore, the systemic response initiated by the body can be manifested as the compensatory secretion of serum MFGE8, which is likely to reflect the compensation of the need for central clearance.

Interestingly, the dynamic changes in the GFAP/MFGE8 ratio provide a new perspective for understanding the differences in their biological roles after TBI. The ratio peaked at 24 h post-injury and then continued to decline within 72 h. This time window is consistent with the critical period of secondary injury and the peak of death (37). The peak of this ratio may mark a turning point in the injury process. The difference in the ability of GFAP and MFGE8 to predict the severity of injury, namely that GFAP performs best at admission while MFGE8 does so at 24 h, further supports this possibility. GFAP can serve as a marker of early structural damage, while the peak level of MFGE8 more reflects the severity of secondary injury and the intensity of endogenous repair response. The continuous decline of the ratio after 24 h may mean that the repair mechanism gradually dominates. Therefore, continuous monitoring of the GFAP/MFGE8 ratio may provide a potential time window indicator for determining the pathological stage transition after TBI, which is conducive to more precise neuroprotective treatment intervention. This strategy is also consistent with the view of Thelin et al. (38), who argue that the combining biomarkers with different cellular origins and temporal dynamics can significantly improve predictive accuracy compared to an evaluation system relying on a single indicator.

The effectiveness of these biomarkers in distinguishing injury severity provides key insights for defining their clinical application scenarios. However, the severity assessment of TBI itself faces challenges: the accuracy of the initial GCS score of admission may be affected by factors such as sedation, poisoning or co-injury, and it may not match the degree of structural damage seen in imaging. Therefore, in this study, imaging severity defined by Rotterdam CT score was also included in addition to clinical GCS score to obtain more objective and complementary evidence of biological injury when assessing injury severity. We found that the concentration of both biomarkers increased in a severity dependent manner and reached their peaks 24 h after injury. In addition, GFAP could stably distinguish between mild-to-moderate and severe TBI patients, which is consistent with previous reports (9, 39). Notably, MFGE8 also showed this efficient dichotomous risk stratification ability, although both biomarkers could not further distinguish between moderate and severe injuries. This may reflect that when the damage exceeds a specific severity threshold, the release dynamics of biomarkers enter a high-level platform period, causing their concentration to no longer increase linearly with the intensification of clinical severity. Despite this limitation, this ability can still be used to clinically identify the most critical patients for rapid decision-making when traditional methods such as CT scan and GCS scores cannot provide an accurate assessment of the degree of injury. Additionally, the significant association between MFGE8 and the severity of imaging injuries further enhances its value as an objective biological indicator, providing an important basis for supplementing and calibrating the initial GCS assessment that may be interfered with, and thereby more reliably identifying potential severe structural injuries.

By repeatedly measuring GFAP and MFGE8 in the serum of the same patient cohort, we clarified the differences in their temporal spectrum for predicting prognosis in the acute phase of TBI. In this study, GFAP demonstrated moderate to good prognosis prediction ability in the early stage of admission, reflecting its clinical value in early risk assessment. This finding differs from the report by Huebschmann et al. (10). The difference may mainly come from the GCS distribution characteristics of the study population: in our cohort, the proportions of mild, moderate and severe TBI patients are relatively balanced (40, 31.4, and 28.6% respectively), while their study focused on a single mild group, which may lead to dilution or deviation in the predictive efficacy of GFAP. For MFGE8, the data showed that its serum level increased significantly overall in non-survivors and poor prognosis, demonstrating good prognostic discrimination efficacy. Importantly, MFGE8’s predictive ability at 24 h post-injury was better than that of GFAP at all time points, and it remained elevated at 48 h, thereby broadening the clinical monitoring window of TBI biomarkers. Therefore, we speculate that persistently high levels of MFGE8 may imply an unbalanced compensatory repair response: despite the initiation of repair processes in response to severe tissue damage, the response may not be enough to offset the continuous damage, which ultimately leads to the failure to restore neuroimmune homeostasis. This hypothesis is consistent with the temporal findings of this study: the continuous high expression of GFAP indicates the persistence of damage, while the correlation between the peak of MFGE8 and the poor prognosis suggests the failure of the repair mechanism. These findings indicate that MFGE8 is not only a potential prognostic biomarker, but also an important indicator for assessing the dynamic balance of damage and repair after TBI. In the future, multi-center and large-scale studies are needed to verify its clinical value, and to explore its specific mechanism in the pathological evolution of TBI with the help of basic experiments. In addition, established multimodal prognostic models for TBI, such as the IMPACT score, integrate neurological examination, CT findings, and laboratory variables; in this context, MFGE8 may provide complementary biological information related to secondary injury and repair, potentially enhancing the predictive performance of such models.

This study has several limitations. First, as a single-center observational study, the sample size is limited and the population source is relatively single, which may affect the universality of the conclusion. Secondly, there is a lack of long-term follow-up data on neurological function. It should be emphasized that MFGE8 is not a biomarker specific to the central nervous system. Although strict inclusion criteria were adopted (such as excluding patients with severe multiple trauma) and GFAP was used as a reference to indirectly evaluate the relative contribution of peripheral sources to the changes in MFGE8, background signals from the peripheral immune system or endothelial system could not be completely eliminated. Furthermore, all the tests were conducted on serum samples without simultaneous cerebrospinal fluid (CSF) testing, which limited the ability to directly explain specific changes in the central nervous system. Future research should focus on combining CSF testing to clarify the central specificity of MFGE8 and verify its clinical value in a larger population through long-term follow-up.

This study identified the dynamic changes of the novel biomarker MFGE8 after TBI and emphasized its close correlation with the severity of injury and clinical outcomes, especially demonstrating the best prognostic predictive ability within 24 h after injury. Unlike GFAP, which serves as an early marker of structural damage, the temporal characteristics of MFGE8 suggest that it is more indicative of the secondary injury process following TBI. Thus, MFGE8 holds potential as a biomarker for assessing TBI severity and prognosis, providing valuable support for clinical grading and treatment decision-making.