Abstract
Background
Traumatic brain injury (TBI) is a leading cause of pediatric mortality and long-term morbidity. Hypoxia-inducible factor-1α (HIF-1α) and its downstream target, BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (BNIP3), are key mediators of the hypoxia response and secondary brain injury. However, their temporal serum dynamics and clinical utility in pediatric TBI remain undefined. This study aimed to measure serum HIF-1α and BNIP3 levels at admission and day 7 post-TBI to evaluate their dynamic changes and relationship with injury severity and neurological outcome.
Methods
In this prospective single-center study, 169 pediatric TBI patients were enrolled. Serum HIF-1α and BNIP3 levels were quantified by ELISA at admission and day 7 post-injury. The relationship between biomarker dynamics (Δ, admission minus day 7), initial injury severity, and 6-month neurological outcome was analyzed.
Results
Admission levels of both HIF-1α and BNIP3 were associated with TBI severity. At day 7 post-TBI, serum HIF-1α decreased significantly in mild TBI, while BNIP3 showed a non-significant trend. In contrast, both biomarkers decreased significantly in the moderate-to-severe TBI group. Among children with moderate-to-severe TBI, a smaller reduction (Δ) in HIF-1α and BNIP3 levels over the first week was independently associated with unfavorable outcome after adjusting for the motor Glasgow Coma Scale score (ΔHIF-1α: aOR = 0.149, P = 0.029; ΔBNIP3: aOR = 0.297, P = 0.046). This association remained robust in a sensitivity analysis of survivors.
Conclusion
The early trajectories of serum HIF-1α and BNIP3 are independently associated with neurological outcomes in pediatric moderate-to-severe TBI. Serial measurement of these biomarkers may enhance risk stratification.
Trial registration
ChiCTR2100048753 (16/07/2021).
Supplementary Information
The online version contains supplementary material available at 10.1186/s12887-026-06528-9.
Keywords: Traumatic brain injury, Hypoxia-inducible factor 1α, Bcl-2/adenovirus E1B19kDa-interacting protein 3
Introduction
Traumatic brain injury (TBI) is a leading cause of mortality and long-term disability in children [1]. Objective biomarkers are needed to overcome the limitations of current clinical and radiological assessments in this population [2, 3]. Yet, relevant research remains scarce for pediatric TBI, which has distinct pathophysiology and management needs compared to adults [4].
Brain tissue hypoxia is a central driver of secondary injury and poor outcomes after TBI [5, 6]. Under hypoxia, the transcription factor hypoxia-inducible factor-1α (HIF-1α) is upregulated, orchestrating cellular adaptation [7]. Elevated serum HIF-1α levels reflect this hypoxic burden and correlate with the magnitude of cerebral injury, as demonstrated in acute stroke where they are associated with infarct volume [8, 9]. Moreover, interventions that reduce cerebral hypoxia and lower HIF-1α levels are linked to improved TBI outcomes [10], highlighting its role within a critical, modifiable pathway. One critical downstream target of HIF-1α is the pro-apoptotic protein, Bcl-2 and adenovirus E1B 19 kDa-interacting protein 3 (BNIP3), which is upregulated in hypoxia and mediates post-TBI neuronal apoptosis [11–13].
Concurrently, astrocytic injury releases S100 calcium-binding protein B (S100B) into the circulation, making it a well-established biomarker that correlates with brain injury severity and is associated with functional outcomes, serving as a clinical benchmark [14–16].
Given that BNIP3 is a prognostic biomarker in oncology [17], we hypothesized that both HIF-1α and BNIP3 might have clinical relevance in pediatric TBI. To contextualize their performance, we included S100B as a reference. This prospective study therefore aimed to serially measure serum HIF-1α and BNIP3 levels in pediatric TBI patients to characterize their early dynamics and evaluate their association with initial injury severity and neurological outcome.
Methods
Study design and participants
This trial was prospectively registered with the Chinese Clinical Trial Registry on 16 July 2021 (Registration number: ChiCTR2100048753). This prospective cohort study was conducted at the Affiliated Xuzhou Children’s Hospital of Xuzhou Medical University between August 2021 and December 2024. We consecutively enrolled pediatric patients presenting with acute TBI as the case group. Based on the Glasgow Coma Scale (GCS) score at admission, patients were stratified into three severity subgroups: mild (GCS 13–15), moderate (GCS 9–12), and severe (GCS 3–8) TBI. A control group of 40 children was recruited from patients admitted for elective inguinal hernia repair. Control subjects were deemed otherwise healthy, with no history of neurological, systemic, or chronic diseases. A comprehensive set of demographic and clinical characteristics was systematically recorded for all participants. This included, but was not limited to age, sex, mechanism of injury, GCS scores (total and motor components), and the occurrence of complications.
Inclusion and exclusion criteria
Inclusion criteria for the TBI group were: (1) age > 28 days and < 18 years at the time of injury; (2) a clear history of trauma with radiologically confirmed TBI; (3) presentation to the hospital within 24 h post-injury; and (4) no prior history of neurological disorders or traumatic brain injury. Exclusion criteria were: (1) presence of underlying diseases such as autoimmune disorders, hematological diseases, endocrine diseases, or malignancies; (2) having undergone surgical intervention for the current injury prior to admission (for TBI cases); or (3) concomitant severe trauma to other organ systems.
Clinical management protocol
All enrolled pediatric TBI patients received standardized management adhering to established national and international guidelines [18, 19]. For patients with moderate-to-severe TBI, the primary goal was to prevent secondary brain injury through measures aimed at avoiding systemic hypotension and maintaining normoxia and normocapnia. Surgical interventions, including evacuation of mass lesions or decompressive craniectomy, were performed based on established neurosurgical criteria.
Sample collection and biochemical analysis
Peripheral blood samples (3 mL) were collected from all participants within 1 h of admission. For the control group, samples were obtained prior to the elective surgical procedure. Samples were centrifuged at 3000 × g for 15 min at 4 ℃ to isolate serum, which was subsequently aliquoted and stored at -80 ℃ until batch analysis to minimize inter-assay variability. Serum levels of HIF-1α, BNIP3, and S100B were quantified using commercial sandwich enzyme-linked immunosorbent assay (ELISA) kits (Catalog Numbers: SEA798Hu, SEJ545Hu, and SEA567Hu, respectively; Wuhan FineBiotech Co., Ltd., China), in strict accordance with the manufacturer’s protocols.
Follow-up and outcome assessment
Serum levels of the biomarkers (HIF-1α, BNIP3, and S100B) were re-measured in the TBI group on day 7 post-TBI. Patients with moderate-to-severe TBI were followed up for 6 months post-injury. Neurological outcomes were assessed at this endpoint using the Glasgow Outcome Scale (GOS). The GOS scores were obtained via structured follow-up, which was conducted via telephone interview or during an outpatient clinic visit, based on family availability and preference. Based on the GOS score, outcomes were dichotomized into favorable (GOS scores 4–5, indicating good recovery or moderate disability) and unfavorable (GOS scores 1–3, indicating severe disability, vegetative state, or death) for primary analysis [20].
Statistical analysis
All statistical analyses were performed using IBM SPSS Statistics (Version 20.0). Data distribution was assessed using box plots and the Shapiro-Wilk test. Continuous data are presented as mean ± SD (normally distributed) or median (IQR) (non-normally distributed). Categorical data are presented as n (%). Group comparisons used t-tests, Mann-Whitney U, Kruskal-Wallis H, or chi-square tests as appropriate. Paired comparisons (admission vs. day 7) used paired t-tests or Wilcoxon signed-rank tests based on data normality. For biomarker comparisons across the four groups (control, mild, moderate, severe TBI), The Mann–Whitney U test with Bonferroni correction was applied after a significant Kruskal-Wallis H test (pair-wise significance threshold α = 0.0083). The biomarker dynamic change was calculated as Δ (admission minus day 7). Logistic regression was used to identify factors associated with unfavorable outcome. Due to the limited events, multivariable modeling was precluded; instead, a series of bivariable models were constructed. Each model assessed a single biomarker (Δ value or admission level) adjusted for motor GCS. Biomarkers were standardized (Z-scores), and model improvement was evaluated using the likelihood ratio test. All tests were two-sided, with P < 0.05 considered significant except for the Bonferroni-adjusted comparisons noted above.
Results
Baseline demographic and clinical characteristics
A total of 169 pediatric patients with TBI were enrolled in this single-center, prospective study. The flow of participants, including details of exclusions and loss to follow-up, is presented in Fig. 1. Specifically, 46 patients in the mild TBI group did not complete the day 7 serum biomarker re-assessment due to early discharge or parental refusal.
Fig. 1.
Flow chart of the study
Table 1 presents the demographic and clinical characteristics by TBI severity. This table includes pertinent variables such as the mechanism of injury, GCS scores, motor GCS, and the occurrence of complications. Clinical characteristics of the moderate-to-severe TBI cohort by 6-month outcome are shown in Table 2. The prevalence of systemic complications, including multi-organ dysfunction syndrome (MODS), is detailed therein.
Table 1.
Baseline demographic and clinical characteristics of the study cohort, stratified by TBI severity (n = 169)
| Mild TBI (n = 88) |
moderate TBI (n = 54) |
Severe TBI (n = 27) |
|
|---|---|---|---|
| Age (years), median (IQR) | 3.00 (1.62, 5.00) | 2.54 (1.23, 5.63) | 2.42 (1.58, 5.00) |
| Gender (male), n (%) | 52 (59.09%) | 33 (61.11%) | 17 (62.96%) |
| Mechanism of injury | |||
| Traffic accidents | 50 | 27 | 14 |
| Falls | 33 | 25 | 12 |
| Blunt trauma | 5 | 2 | 1 |
| Primary neuroimaging findings on admission | |||
| Intracranial hemorrhage | 62 | 54 | 27 |
| Diffuse axonal injury | 0 | 0 | 3 |
| Hypoxic-ischemic encephalopathy | 0 | 0 | 1 |
| No acute intracranial injury | 26 | 0 | 0 |
| Time from injury to initial imaging (hours)*, median (IQR) | 6 (4, 9) | 4 (3, 5) | 2 (2, 3) |
| GCS, median (IQR) | 15 (15, 15) | 12 (10, 12) | 6 (5, 7) |
| Motor GCS, median (IQR) | 6(6, 6) | 5 (4, 5) | 3 (2, 3) |
| Complications | |||
| Cardiac arrest | 0 | 0 | 1 |
| Hypotension/Shock | 0 | 1 | 3 |
| Hypoxemia/Respiratory Failure | 0 | 2 | 7 |
| Coagulation Abnormalities | 5 | 18 | 16 |
| Hyperglycemia | 2 | 12 | 13 |
| Others (Electrolyte imbalances, Acid-base imbalances, Stress ulcer) | 1 | 6 | 7 |
| Disposition | |||
| Home | 88 | 51 | 14 |
| Rehabilitation facility | 0 | 3 | 9 |
| Death | 0 | 0 | 4 |
| Time to death | Not applicable | Not applicable | 8d, 10d, 11d, 14d# |
| Distribution of GOS scores | |||
| 5 | 88 | 51 | 14 |
| 4 | 0 | 0 | 1 |
| 3 | 0 | 3 | 7 |
| 2 | 0 | 0 | 1 |
| 1 | 0 | 0 | 4 |
* The initial neuroimaging study was computed tomography (CT) for the majority of patients. Exceptions were three patients with mild TBI and one patient with moderate TBI, for whom magnetic resonance imaging (MRI) was performed due to parental preference regarding radiation exposure
# Data are presented as individual values due to the small number of events (n = 4)
Table 2.
Baseline demographic and clinical characteristics of children with moderate and severe TBI, stratified by prognosis (n = 81)
| Favorable outcome (n = 66) | Unfavorable outcome (n = 15) | |
|---|---|---|
| Age (years), median (IQR) | 2.88 (1.67, 5.63) | 1.58 (0.95, 2.75) |
| Gender (male), n (%) | 44 (66.67%) | 6 (40.00%) |
| Mechanism of injury | ||
| Traffic accidents | 31 | 10 |
| Falls | 32 | 5 |
| Blunt trauma | 3 | 0 |
| Primary Neuroimaging Findings on Admission | ||
| Intracranial hemorrhage | 66 | 15 |
| Diffuse axonal injury | 1 | 2 |
| Hypoxic-ischemic encephalopathy | 0 | 1 |
| Time from injury to initial imaging (hours)*, median (IQR) | 4 (2.8, 5.0) | 2 (2, 4) |
| GCS, median (IQR) | 11.00 (8.75, 12.00) | 6.00 (3.00, 8.00) |
| Motor GCS, median (IQR) | 4 (3, 5) | 2(1, 4) |
| Complications | ||
| Cardiac arrest | 0 | 1 |
| Hypotension/Shock | 0 | 4 |
| Hypoxemia/Respiratory Failure | 2 | 7 |
| Coagulation Abnormalities | 22 | 12 |
| Hyperglycemia | 13 | 12 |
| Multi-organ Dysfunction Syndrome (MODS) # | 1 | 11 |
| Others (Electrolyte imbalances, Acid-base imbalances, Stress ulcer) | 6 | 7 |
| Disposition | ||
| Home | 65 | 0 |
| Rehabilitation facility | 1 | 11 |
| Death | 0 | 4 |
| Time to death | Not applicable | 8d, 10d, 11d, 14d& |
| Distribution of GOS scores | ||
| 5 | 65 | 0 |
| 4 | 1 | 0 |
| 3 | 0 | 10 |
| 2 | 0 | 1 |
| 1 | 0 | 4 |
* The initial neuroimaging study was CT for all patients except one. This patient (with moderate TBI, in the favorable outcome group) underwent MRI due to parental preference regarding radiation exposure.
# MODS was defined as dysfunction in ≥ 2 organ systems
& Data are presented as individual values due to the small number of events (n = 4)
Association of admission biomarker levels with TBI severity
Serum concentrations of HIF-1α, BNIP3, and S100B in controls and patients with varying severities of TBI are shown in Fig. 2. The Kruskal-Wallis H test confirmed statistically significant overall differences among the four groups for each biomarker (all P < 0.001). Post-hoc pairwise comparisons were performed using the Mann-Whitney U test with a Bonferroni-adjusted significance level of α = 0.0083. The detailed results are provided in Supplementary Table S1. Serum levels of HIF-1α and BNIP3 in patients with mild TBI were not significantly different from those in controls after correction. In contrast, S100B levels were significantly elevated in the mild TBI group compared to controls. For all three biomarkers, a significant stepwise increase in serum concentrations correlated with TBI severity. All pairwise comparisons among the three TBI severity groups remained statistically significant after Bonferroni correction (all P < 0.001).
Fig. 2.
Serum concentrations of HIF-1α, BNIP3, and S100B in controls and patients across TBI severity groups. (A-C) Box plots display the distribution of serum concentrations of (A) HIF-1α, (B) BNIP3, and (C) S100B among the four groups. Circles represent outliers, and asterisks denote extreme outliers. The central line within each box indicates the median, the box represents the interquartile range (25th to 75th percentile), and the whiskers extend to 1.5 times the interquartile range
Temporal changes in serum biomarkers across severity subgroups
In patients with mild TBI, serum HIF-1α and S100B levels showed a significant decrease at day 7, while BNIP3 exhibited a non-significant declining trend (Supplementary Table S2). In contrast, patients with moderate-to-severe TBI exhibited a significant reduction in all three biomarkers by day 7. This pattern was observed in both surgical and non-surgical patients (Supplementary Table S3).
Serum biomarkers and neurological outcomes in moderate-to-severe TBI
Among the 81 patients with moderate-to-severe TBI who completed the 6-month follow-up, 66 (81.5%) had favorable outcomes, while 15 (18.5%) had unfavorable outcomes. Their baseline demographic and clinical characteristics are detailed in Table 2. As illustrated in Fig. 3, admission serum levels of HIF-1α, BNIP3, and S100B were significantly elevated in patients who subsequently developed an unfavorable outcome compared to those with a favorable outcome (HIF-1α: Z = -3.034, P = 0.002; BNIP3: Z = -3.843, P < 0.001; S100B: Z = -2.584, P = 0.010).
Fig. 3.
Admission serum concentrations of HIF-1α, BNIP3, and S100B stratified by 6-month outcome in children with moderate-to-severe. (A-C) Box plots comparing serum concentrations of (A) HIF-1α, (B) BNIP3, and (C) S100B between patients with favorable and unfavorable outcomes. Circles represent outliers, diamonds represent statistically significant intergroup differences (Mann-Whitney U test, all P < 0.05). Center lines represent medians, boxes indicate interquartile ranges, whiskers extend to 1.5×IQR
Dynamic changes in biomarkers are associated with neurological outcome
The absolute change (Δ) for each biomarker from admission to day 7 was calculated. As shown in Fig. 4, patients with unfavorable outcomes exhibited a significantly smaller reduction in serum levels of HIF-1α, BNIP3, and S100B compared to those with favorable outcomes (HIF-1α: Z = 2.507, P = 0.012; BNIP3: Z = 2.626, P = 0.009; S100B: Z = 2.093, P = 0.036). Individual trajectories varied (Fig. 4). A post-hoc review was performed for patients showing a rise in HIF-1α by day 7. Among the subset of severe TBI patients undergoing invasive ICP monitoring, three individuals exhibited this phenotype. All three developed refractory intracranial hypertension and had unfavorable outcomes. One of these patients suffered an in-hospital cardiac arrest; this patient presented with the highest admission HIF-1α level, which increased further, culminating in a fatal outcome.
Fig. 4.
Dynamic changes in serum biomarker concentrations from admission to day 7, stratified by 6-month outcome in moderate-to-severe TBI. (A-C) Box plots display the absolute change (Δ) in serum concentrations of (A) HIF-1α, (B) BNIP3, and (C) S100B between children with favorable and unfavorable outcomes. Circles represent outliers, asterisks denote extreme outliers, and diamonds represent statistically significant intergroup differences (Mann-Whitney U test, all P < 0.05). Center lines represent medians, boxes indicate interquartile ranges, whiskers extend to 1.5×IQR.
Independent association of biomarker dynamics with outcome
In univariate analyses, an unfavorable outcome was significantly associated with a lower motor GCS score, higher admission levels of HIF-1α, BNIP3, and S100B, as well as smaller reductions in their levels over time (lower ΔHIF-1α, ΔBNIP3, and ΔS100B) (Table 3). Given these univariate associations, we next assessed which biomarkers remained associated with outcome independently of the baseline motor GCS score.
Table 3.
Univariate associations between individual variables and unfavorable neurological outcomes
| Variable | OR (95% CI) | P value |
|---|---|---|
| Motor GCS | 0.188 (0.081–0.432) | < 0.001 |
| HIF-1α (admission) | 2.666 (1.366–5.205) | 0.004 |
| BNIP3 (admission) | 3.791 (1.826–7.869) | < 0.001 |
| S100B (admission) | 2.361 (1.298–4.293) | 0.005 |
| Δ HIF-1α | 0.273 (0.081–0.921) | 0.036 |
| Δ BNIP3 | 0.394 (0.165–0.940) | 0.036 |
| Δ S100B | 0.288 (0.087–0.959) | 0.042 |
OR odds ratio, CI confidence interval
Biomarker variables were standardized (Z-scores) to facilitate the interpretation of odds ratios.
Due to the limited number of outcome events, separate bivariable logistic regression models were constructed, each adjusting for motor GCS. After this adjustment, the admission levels of all biomarkers and ΔS100B were no longer significantly associated with the outcome. Crucially, both ΔHIF-1α and ΔBNIP3 retained significant independent associations with unfavorable outcome after adjustment for motor GCS (Table 4). Moreover, the addition of either ΔHIF-1α or ΔBNIP3 to a model containing only motor GCS significantly improved the model fit (likelihood ratio test: for ΔHIF-1α, P = 0.009; for ΔBNIP3, P = 0.013).
Table 4.
Bivariable associations of serum biomarkers with unfavorable neurological outcome, adjusted for motor GCS (forward LR selection)
| Variable | Adjusted OR (95% CI) | P value |
|---|---|---|
| Model 1 | ||
| Motor GCS | 0.155 (0.059–0.408) | < 0.001 |
| Δ HIF-1α | 0.149 (0.027–0.825) | 0.029 |
| Model 2 | ||
| Motor GCS | 0.174 (0.071–0.426) | < 0.001 |
| Δ BNIP3 | 0.297 (0.090–0.977) | 0.046 |
Adjusted OR adjusted odds ratio, CI confidence interval
Biomarker variables were standardized (Z-scores) to facilitate the interpretation of odds ratios.
A parallel analysis using the total GCS score yielded consistent findings. After adjustment for total GCS, the admission levels of all three biomarkers were again not significantly associated with the outcome. Conversely, ΔHIF-1α, ΔBNIP3, and ΔS100B each remained significantly associated with an unfavorable outcome (Supplementary Table S4). Similarly, the addition of ΔHIF-1α, ΔBNIP3, or ΔS100B to a model containing only total GCS significantly improved the model fit (likelihood ratio test: for ΔHIF-1α, P = 0.004; for ΔBNIP3, P = 0.005; for ΔS100B, P = 0.005).
Association of biomarker dynamics with outcome in survivors
A sensitivity analysis was conducted in the subset of patients who survived to the 6-month follow-up (n = 77). In survivors, the Δ values for all three biomarkers were significantly smaller in patients with unfavorable outcomes (all P < 0.05, Supplementary Table S5). In univariate analyses, an unfavorable outcome was associated with a lower motor GCS score, higher admission levels of HIF-1α, BNIP3, as well as lower ΔHIF-1α, and ΔS100B (Table 5).
Table 5.
Univariate associations between individual variables and unfavorable neurological outcomes in survivors
| Variable | OR (95% CI) | P value |
|---|---|---|
| Motor GCS | 0.229 (0.095–0.547) | 0.001 |
| HIF-1α (admission) | 2.402 (1.156–4.991) | 0.019 |
| BNIP3 (admission) | 3.291 (1.551–6.984) | 0.002 |
| S100B (admission) | 1.723 (0.925–3.210) | 0.086 |
| Δ HIF-1α | 0.136 (0.024–0.781) | 0.025 |
| Δ BNIP3 | 0.372 (0.137–1.012) | 0.053 |
| Δ S100B | 0.145 (0.024–0.874) | 0.035 |
OR odds ratio, CI confidence interval
Biomarker variables were standardized (Z-scores) to facilitate the interpretation of odds ratios.
After adjustment for the motor GCS, ΔHIF-1α remained independently associated with outcome (aOR = 0.069, 95% CI: 0.006–0.744, P = 0.028), and the addition of ΔHIF-1α to the model containing only motor GCS significantly improved model fit (likelihood ratio test: χ² = 8.309, P = 0.004). The associations for ΔBNIP3 and ΔS100B did not reach formal statistical significance (for ΔBNIP3: aOR = 0.308, P = 0.056; for ΔS100B: aOR = 0.083, P = 0.052; Table 6).
Table 6.
Bivariable associations of serum biomarkers with unfavorable outcome in survivors, adjusted for motor GCS (forward LR selection)
| Variable | Adjusted OR (95% CI) | P value |
|---|---|---|
| Model 1 | ||
| Motor GCS | 0.207 (0.079–0.542) | 0.001 |
| Δ HIF-1α | 0.069 (0.006–0.744) | 0.028 |
| Model 2 | ||
| Motor GCS | 0.207 (0.080–0.536) | 0.001 |
| Δ BNIP3 | 0.056 (0.092–1.031) | 0.056 |
| Model 3 | ||
| Motor GCS | 0.244 (0.102–0.585) | 0.002 |
| Δ S100B | 0.083 (0.007–1.023) | 0.052 |
Adjusted OR adjusted odds ratio, CI confidence interval
Biomarker variables were standardized (Z-scores) to facilitate the interpretation of odds ratios.
Discussion
The principal finding of our prospective study is that the dynamic changes of serum HIF-1α and BNIP3 over the first week post-injury are independently associated with 6-month neurological outcome in children with moderate-to-severe TBI, after adjustment for the initial motor GCS. This serial measurement approach identified children at highest risk for unfavorable outcomes, even among survivors.
HIF-1α serves as a key hypoxia-responsive signaling molecule that is rapidly upregulated under pathological conditions such as cerebral hypoxia and ischemia [21]. Experimental studies have established its pivotal role in post-TBI pathophysiological processes, including neuroinflammation, apoptosis and vascular regeneration [22–24]. In adults with severe TBI, elevated serum HIF-1α has been linked to prognosis [25]. However, its dynamics and clinical relevance in pediatric TBI remain undefined. BNIP3, a key downstream effector of HIF-1α, is implicated in neuronal death after TBI [13]. While a prognostic biomarker in oncology [17], its role in pediatric neurotrauma is unexplored.
To contextualize our findings, we included S100B, a well-established biomarker in TBI, as a reference biomarker [26]. In our cohort, while ΔS100B showed an association with outcome in univariate analysis, only ΔHIF-1α and ΔBNIP3 remained independently associated with outcome after adjustment for motor GCS in separate bivariable models, underscoring the clinical value of these hypoxia pathway markers.
Our results delineate their context-dependent utility. In mild TBI, admission levels of HIF-1α and BNIP3 did not differ from controls, suggesting specificity for more substantive injury. In moderate-to-severe TBI, admission levels of both biomarkers were elevated and correlated with injury severity. While higher admission levels were associated with unfavorable outcomes in univariate analysis, this association was not independent after adjustment for motor GCS. This finding prompted a deeper investigation into their dynamic changes. We discovered that the Δ values of HIF-1α and BNIP3 were significantly different between outcome groups. Most importantly, ΔHIF-1α and ΔBNIP3 emerged as significant independent factors associated with unfavorable outcome after adjusting for motor GCS in our bivariable regression analyses.
Unlike the admission level, ΔHIF-1α was independently associated with outcome, suggesting it reflects pathological processes that evolve beyond the initial insult. We propose that sustained HIF-1α elevation signifies entry into a maladaptive, self-perpetuating cycle of secondary injury. The initial hypoxic insult following TBI upregulates cerebral HIF-1α, which then drives secondary injury through blood-brain barrier (BBB) disruption, cerebral edema, pyroptosis, and neuroinflammation [27, 28], worsening local hypoxia and creating a positive cerebral feedback loop.
Critically, this cerebral cycle propagates systemically. Through neuroinflammatory spillover and damage-associated molecular patterns (DAMPs) release, TBI initiates a systemic inflammatory response [29, 30]. This response contributes to dysfunction in multiple organs, including the lungs, liver, and the brain itself [31]. This deleterious brain-systemic interplay can culminate in multi-organ dysfunction syndrome (MODS), substantially worsening patient prognosis [30]. Our cohort study revealed that systemic complications, including MODS, were concentrated in the unfavorable outcome group, and these patients exhibited persistently elevated serum HIF-1α levels. This clinical-biological link is mechanistically plausible, as research in systemic inflammation (e.g., sepsis) has established that inflammatory pathways (NF-κB, STAT3, PI3K/Akt) upregulate HIF-1α, which in turn amplifies the production of cytokines like TNF-α, IL-1β, and IL-6, creating a self-reinforcing vicious cycle [32]. Therefore, persistent HIF-1α elevation likely reflects the integrated output of a mutually reinforcing “brain-systemic” circuit, where the cerebral and systemic pathological processes fuel each other in a deleterious feedback loop. BNIP3, as a downstream target, fluctuates accordingly.
Although HIF-1α and BNIP3 are not brain-specific, they are core mediators of key secondary injury processes in TBI, such as cerebral edema, BBB damage, and neuroinflammation. Consequently, their serial serum dynamics provide a unique, integrative window into the intensity of the combined “brain-systemic” pathological burden, which ultimately determines neurological destiny.
Study limitations
Despite the novel findings presented, our study has several limitations. First, its single-center design and moderate sample size may limit the generalizability of the findings. Statistical power, particularly for detailed subgroup analyses (e.g., the sensitivity analysis confined to survivors), was constrained. Second, while our analysis identified multi-organ dysfunction syndrome (MODS) as a key correlate of poor outcome and incorporated systemic inflammatory pathways into our interpretive model, comprehensive, high-resolution hemodynamic (e.g., ICP/CPP) data were not available for all patients, as monitoring was clinically guided. Future studies with protocolized multimodal monitoring are needed to precisely define the physiological drivers of persistent biomarker elevation. Third, the measurement of biomarkers was confined to two time points (admission and day 7). A denser sampling schedule could have better delineated the kinetic profiles. Fourth, the batched analysis of samples, while ensuring assay consistency, precludes real-time clinical application; the development of rapid point-of-care assays is essential for clinical translation. Finally, as an observational study, our findings demonstrate important associations and propose a plausible pathophysiological model but cannot establish causality or confirm predictive utility. External validation in independent, multi-center cohorts is required before clinical implementation can be considered.
Conclusion
In children with moderate-to-severe TBI, the early dynamic changes (Δ) in serum HIF-1α and BNIP3 levels are independently associated with neurological outcome and provide information complementary to the initial clinical assessment. Serial measurement of these hypoxia pathway biomarkers may contribute to early risk stratification.
Supplementary Information
Acknowledgements
The authors extend their sincere gratitude to all the infants and their families for their participation in this study, as well as to the clinical staff for their support. We are also deeply grateful to Professor Wang Ke and to Zeng Ping for their expert guidance and invaluable assistance in the statistical analysis.
Authors’ contributions
HY, LJ M, KX L, and LZ collaboratively conceived and designed the study. LZ formulated the research protocol and supervised the research. KX L assisted in data acquisition, ZW L contributed to the experimental design, and TQ carried out data curation and statistical analysis. The initial draft of the manuscript was prepared by HY and LJ M created the figure and tables. All authors reviewed and approved the final manuscript.
Funding
This study was supported by grants from the Jiangsu Provincial Health Commission Research Project (No. M2021083), the Xuzhou Science and Technology Plan Project (No. KC23148), and the Pengcheng Talents Program-Medical Youth Reserve Talent Project (No. XWRCHT20220021).
Data availability
The datasets generated and/or analyzed during this study are not publicly available in order to protect patient privacy. However, they can be made available from the corresponding author upon reasonable request and with permission from the relevant ethics committee.
Declarations
Ethics approval and consent to participate
This study was reviewed and approved by the Medical Science Ethics Committee of the Affiliated Xuzhou Children’s Hospital of Xuzhou Medical University (Approval No. 2021-S-07-K07). Written informed consent was obtained from all legal guardians of the participants prior to their inclusion in the study.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Hong Yin, Ling-jian Meng and Kai-xun Liu contributed equally to this work.
References
- 1.Chevignard M, Câmara-Costa H, Dellatolas G. Predicting and improving outcome in severe pediatric traumatic brain injury. Expert Rev Neurother. 2024;24(10):963–83. [DOI] [PubMed] [Google Scholar]
- 2.Huang H, Fu G, Lu S, et al. Plasma profiles of inflammatory cytokines in children with moderate to severe traumatic brain injury: a prospective cohort study. Eur J Pediatr. 2024;183(8):3359–68. [DOI] [PubMed] [Google Scholar]
- 3.Oris C, Kahouadji S, Bouvier D, et al. Blood biomarkers for the management of mild traumatic brain injury in clinical practice. Clin Chem. 2024;70(8):1023–36. [DOI] [PubMed] [Google Scholar]
- 4.Manet R, de Courson H, Capel C, et al. Neurosurgical management of the acute phase of adult and pediatric traumatic brain injury: 2025 guidelines of the French society of neurosurgery. Neurochirurgie. 2025;71(4):101686. [DOI] [PubMed] [Google Scholar]
- 5.Meyer LJ, Lotze FP, Riess ML. Simulated traumatic brain injury in in-vitro mouse neuronal and brain endothelial cell culture models. J Pharmacol Toxicol Methods. 2022;114:107159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Meyer LJ, Riess ML. Evaluation of in vitro neuronal protection by postconditioning with poloxamer 188 following simulated traumatic brain injury. Life (Basel). 2021;11(4):316. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Qian Y, Li X, Fan R, et al. MicroRNA-31 inhibits traumatic brain injury-triggered neuronal cell apoptosis by regulating hypoxia-inducible factor-1A/vascular endothelial growth factor A axis. NeuroReport. 2022;33(1):1–12. [DOI] [PubMed] [Google Scholar]
- 8.Cai YY, Zhuang YK, Wang WJ, et al. Potential role of serum hypoxia-inducible factor 1alpha as a biomarker of delayed cerebral ischemia and poor clinical outcome after human aneurysmal subarachnoid hemorrhage: A prospective, longitudinal, multicenter, and observational study. Front Neurol. 2022;13:1072351. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Xue L, Chen H, Lu K, et al. Clinical significance of changes in serum neuroglobin and HIF-1α concentrations during the early-phase of acute ischemic stroke. J NEUROL SCI. 2017;375:52–7. [DOI] [PubMed] [Google Scholar]
- 10.Jia H, Chen Y, Wang Y, et al. The neuroprotective effect of electro-acupuncture on cognitive recovery for patients with mild traumatic brain injury: A randomized controlled clinical trial. Medicine. 2023;102(6):e32885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Madhu V, Boneski PK, Silagi E, et al. Hypoxic regulation of mitochondrial metabolism and mitophagy in nucleus pulposus cells is dependent on HIF-1α-BNIP3 axis. J Bone Min Res. 2020;35(8):1504–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wang X, Wu Z, Zhang Y, et al. Autophagy induced by hypoxia in pulpitis is mediated by HIF-1α/BNIP3. ARCH ORAL BIOL. 2024;159:105881. [DOI] [PubMed] [Google Scholar]
- 13.Mu J, Weng J, Yang C, et al. Necrostatin-1 prevents the proapoptotic protein Bcl-2/adenovirus E1B 19-kDa interacting protein 3 from integration into mitochondria. J NEUROCHEM. 2020;156(6):929–42. [DOI] [PubMed] [Google Scholar]
- 14.Gayger-Dias V, Vizuete AF, Rodrigues L, et al. How S100B crosses brain barriers and why it is considered a peripheral marker of brain injury. Exp Biol Med (Maywood). 2023;248(22):2109–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ma R, Liu Y, Mao Q. Clinical value of serum LINC02446 and S100B in early diagnosis and prognosis assessment of traumatic brain injury. Int J Neurosci. 2024;135(9):998–1005. [DOI] [PubMed] [Google Scholar]
- 16.Godeiro Coelho LM, Teixeira FJP, Koru-Sengul T, et al. Predictive value of nervous cell injury biomarkers in Moderate-to-Severe traumatic brain injury: A network Meta-Analysis. Neurology. 2025;105(5):e213997. [DOI] [PubMed] [Google Scholar]
- 17.Yu Q, Fu W, Fu Y, et al. BNIP3 as a potential biomarker for the identification of prognosis and diagnosis in solid tumours. Mol Cancer. 2023;22(1):143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kochanek PM, Tasker RC, Carney N, et al. Guidelines for the management of pediatric severe traumatic brain Injury, third edition: update of the brain trauma foundation guidelines. PEDIATR CRIT CARE ME. 2019;20(3S Suppl 1):S1–82. [DOI] [PubMed] [Google Scholar]
- 19.Pediatric Craniocerebral Trauma Professional Group, Neurosurgery Branch of Chinese Medical Association; Neurological Injury Professional Group, Traumatology Branch of Chinese Medical Association. Chinese expert consensus on the diagnosis and treatment of pediatric traumatic brain injury. Chin J Neurosurg. 2021;37(12):1200–8. [Google Scholar]
- 20.Munoz Pareja JC, de Rivero Vaccari JP, Chavez MM, et al. Prognostic and diagnostic utility of serum biomarkers in pediatric traumatic brain injury. J Neurotrauma. 2024;41(1–2):106–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lisy K, Peet DJ. Turn me on: regulating HIF transcriptional activity. Cell Death Differ. 2008;15(4):642–9. [DOI] [PubMed] [Google Scholar]
- 22.Tang W, Guo ZD, Chai WN, et al. Downregulation of miR-491-5p promotes neovascularization after traumatic brain injury. Neural Regen Res. 2022;17(3):577–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Xu X, Yang M, Zhang B, et al. HIF-1α participates in secondary brain injury through regulating neuroinflammation. Transl Neurosci. 2023;14(1):20220272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Fang Y, Lu J, Wang X, et al. HIF-1α mediates TRAIL-Induced neuronal apoptosis via regulating DcR1 expression following traumatic brain injury. Front Cell Neurosci. 2020;14:192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Lv QW, Zheng ZQ, Zhang H, et al. Serum hypoxia-inducible factor 1alpha emerges as a prognostic factor for severe traumatic brain injury. Clin Chim Acta. 2021;522:77–82. [DOI] [PubMed] [Google Scholar]
- 26.Wichmann TO, Babaee A, Duch K, et al. A head-to-head comparison of S100B and GFAP/UCH-L1 for detection of traumatic intracranial lesions in a Scandinavian trauma cohort. Scand J Trauma Resusc Emerg Med. 2025;33(1):52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Higashida T, Kreipke CW, Rafols JA, et al. The role of hypoxia-inducible factor-1α, aquaporin-4, and matrix metalloproteinase-9 in blood-brain barrier disruption and brain edema after traumatic brain injury. J NEUROSURG. 2011;114(1):92–101. 10.3171/2010.6.JNS10207. [DOI] [PubMed] [Google Scholar]
- 28.Yuan D, Guan S, Wang Z et al. HIF-1α aggravated traumatic brain injury by NLRP3 inflammasome-mediated pyroptosis and activation of microglia. J CHEM NEUROANAT. 2021; 116 101994. 10.1016/j.jchemneu.2021.101994 [DOI] [PubMed]
- 29.Chaikittisilpa N, Krishnamoorthy V, Lele AV, et al. Characterizing the relationship between systemic inflammatory response syndrome and early cardiac dysfunction in traumatic brain injury. J NEUROSCI RES. 2017;96(4):661–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Akshita Garg N, Khatri BP, Singh, et al. Multi organ dysfunction (MODS): A life-threatening aspect of traumatic brain injury. IJPTher. 2025;6(3):127–39. [Google Scholar]
- 31.Thakur M, Sharma S, Vasudeva N, et al. Eprosartan alleviates the traumatic brain injury-induced multi-organ dysfunction syndrome in mice via AT1R/SNS/HMGB1 Blockade and PDL-1 modulation. MOL CELL NEUROSCI. 2025;135:104035. [DOI] [PubMed] [Google Scholar]
- 32.Liu SY, Ruan H, Li SS. HIF-1α: a Bridge connecting sepsis and acute respiratory distress syndrome. Eur J Med Res. 2025;30(1):827. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated and/or analyzed during this study are not publicly available in order to protect patient privacy. However, they can be made available from the corresponding author upon reasonable request and with permission from the relevant ethics committee.