Aspartate aminotransferase to platelet ratio index as a predictor of mortality in traumatic brain injury

1. Introduction

Traumatic Brain Injury (TBI) is a major cause of morbidity and mortality worldwide and accounts for approximately 30% of all injury-related deaths in the United States [1]. Outcomes after TBI are highly variable, ranging from complete recovery to fatality [2]. Established predictors of poor outcomes include advanced age, severe initial neurological impairment (e.g., low Glasgow Coma Scale (GCS)), and markers of traumatic shock or coagulopathy. In particular, post-traumatic coagulopathy (such as thrombocytopenia) has been linked to worse prognosis in TBI patients [3]. Likewise, TBI can trigger systemic inflammatory responses that affect peripheral organs, as studies show that acute TBI is often accompanied by hepatic injury, evidenced by elevated liver enzymes within hours of injury [1]. These systemic effects of TBI suggest that laboratory biomarkers reflecting multi-organ stress may have prognostic value.

The Aspartate Aminotransferase-to-Platelet Ratio Index (APRI) is a simple index originally developed to assess liver fibrosis in hepatitis patients. APRI is calculated from routine blood tests (aspartate aminotransferase (AST) level and platelet count) and was initially validated for predicting hepatic fibrosis severity. Interestingly, recent evidence indicates that APRI may also reflect the severity of systemic illnesses. It has been identified as an independent risk factor for mortality in various critical conditions such as COVID-19, severe malaria, dengue fever, and sepsis [4]. APRI has been recognized for its potential to reduce reliance on invasive procedures like liver biopsy in staging liver fibrosis, especially in resource-limited settings [5]. In critical care settings, an elevated AST level often signals widespread tissue injury or shock, while a low platelet count can indicate consumption coagulopathy or severe inflammatory response. APRI combines these two parameters, potentially serving as a composite marker of systemic insult. Over the years, APRI’s diagnostic and prognostic applications have expanded beyond its original scope, finding utility across diverse clinical scenarios [4, 6, 7, 8, 9, 10].

TBI presents unique pathophysiological challenges that induce systemic inflammatory responses, coagulopathy, and liver dysfunction, all of which may impact APRI values [11, 12, 13]. While APRI has been used to predict outcomes in liver diseases [14, 15, 16], inflammatory conditions like COVID-19 [7], and after total hip arthroplasty and arthroscopic rotator cuff repair [10, 17], its role in TBI outcomes has not been fully studied. To our knowledge, only limited data exist on liver-related indices in TBI prognosis. A recent study by Wang et al. [18] noted that TBI patients who did not survive had significantly higher APRI values and related liver fibrosis scores compared to survivors. The fibrosis-4 (FIB-4) index, which is based on AST and includes age and alanine transaminase (ALT), became a reliable predictor of death in that group [18]. However, APRI itself was not the primary focus, leaving a knowledge gap as to whether it independently predicted TBI mortality. Given AST’s sensitivity to these systemic effects, it may serve as a valuable biomarker for early intervention in TBI, with further research needed to clarify its prognostic value in TBI and utility in early risk stratification.

The present study, therefore, aimed to determine whether APRI on admission is a significant predictor of mortality in patients with TBI. We conducted a retrospective cohort analysis to evaluate the association of APRI with in-hospital mortality after TBI while controlling for known risk factors. We hypothesized that elevated APRI at hospital admission would be associated with higher mortality, reflecting the burden of systemic injury in TBI.

2. Methods

3. Results

3.1 Patient enrollment

The study evaluated trauma patients from the Trauma Registry System between 1998 and 2023, focusing on those aged 20 years and older with a head or neck injury severity score (AIS) of ≥3. From an initial registry of 54,313 patients, 47,922 met the age criteria. Among these, 7657 patients presented with head or neck injuries classified as AIS ≥3. Further exclusions were applied, removing cases with hanging injuries (n = 5), burn injuries (n = 1), and instances of missing laboratory data (n = 1399). These exclusions led to a final cohort of 6252 patients. Within this cohort, 594 cases resulted in fatalities, while 5658 patients survived, as illustrated in Fig. 1.

Fig. 1.Patient enrollment flowchart for the traumatic brain injury cohort study. AIS: Abbreviated Injury Scale.

3.3 The mortality predictive performance of APRI

The effectiveness of the APRI as a tool for predicting death outcomes among the research cohort is shown in Fig. 2. In the overall cohort, APRI demonstrated only moderate discrimination for mortality (AUC = 0.648 with an optimal cutoff of 0.59, yielding 63.6% sensitivity and 60.3% specificity), whereas the GCS showed a substantially higher predictive accuracy (AUC = 0.835 at a cutoff of 9.5) (Fig. 2B). When stratified by injury severity, the predictive performance of APRI declined with increasing ISS. The AUC for APRI was 0.590 in the mild injury subgroup (ISS 1–15), 0.567 in the moderate ISS group (16–24), and 0.552 in the severe ISS category (ISS ≥25) (Fig. 2C–E), indicating only poor to fair discrimination within these strata.

Fig. 2.Receiver Operating Characteristic (ROC) curve analysis showing the mortality predictive performance of (A) Aspartate Aminotransferase-to-Platelet Ratio Index (APRI) and (B) Glasgow Coma Scale (GCS) in the overall traumatic brain injury cohort and APRI in patient subgroups stratified by injury severity: mild (ISS 1–15, C) moderate (ISS 16–24, D), and severe (ISS ≥25, E). For each analysis, the optimal cutoff value, sensitivity, specificity, and area under the curve (AUC) are displayed. ISS: Injury Severity Score.

4. Discussion

Our study assessed the prognostic value of the APRI in predicting mortality among patients with TBI, with results indicating a moderate predictive performance. We identified an optimal cutoff of 0.59 for APRI, yielding an AUC of 0.648, with a sensitivity of 63.6%, and specificity of 60.3%. When comparing these findings with APRI’s application in other clinical scenarios, APRI demonstrates variability in diagnostic and prognostic efficacy, likely influenced by the distinct pathophysiological processes across diseases.

TBI has a significant impact on systemic physiology, including liver function, due to complex neuro-inflammatory and systemic inflammatory responses. Studies have demonstrated that liver function test abnormalities are common in TBI patients, even in the absence of pre-existing liver conditions. Among these abnormalities, AST levels tend to rise due to hepatic ischemia, oxidative stress, and inflammatory responses triggered by brain injury [19]. The systemic inflammatory response following TBI involves cytokine release (Interleukin-1 beta, Interleukin-6, and Tumor Necrosis Factor alpha), which induces hepatocellular damage, leading to increased AST levels [20]. Histological analysis from experimental models suggests that TBI induces hepatocyte dystrophy, sinusoidal congestion, and hepatocellular necrosis, contributing to AST elevation [21]. The APRI score, is widely used as a non-invasive marker of liver fibrosis and function. In TBI patients, an elevated APRI score may serve as an indicator of multi-organ dysfunction and a predictor of mortality. Since platelet count is often reduced due to systemic inflammatory responses and disseminated intravascular coagulation post-TBI, and AST levels are elevated due to hepatocellular damage, a higher APRI score may reflect the severity of both hepatic injury and systemic complications. Previous studies in intensive care unit (ICU) populations have demonstrated an association between deranged liver function tests and poor prognosis, supporting the hypothesis that a high APRI score in TBI patients correlates with increased mortality risk [19].

In chronic liver diseases, APRI has shown significant diagnostic value in staging fibrosis and cirrhosis, especially in hepatitis B and C contexts. For instance, Jin et al. [15] observed an AUC of 0.79 for predicting significant fibrosis in hepatitis B, with sensitivities reaching 84% at a lower cutoff of 0.5, though specificity was moderate (41%). A systematic review showed moderate accuracy with AUC values of 0.74 for significant fibrosis, 0.73 for advanced fibrosis, and 0.73 for cirrhosis [14]. In hepatitis C, Lin et al. [5] reported an AUC of up to 0.83 for APRI in cirrhosis prediction, demonstrating APRI’s robust application within liver disease. These AUC values exceed our findings in TBI, possibly due to APRI’s closer association with chronic liver dysfunction rather than acute trauma-related responses. In hepatocellular carcinoma (HCC), APRI has shown prognostic significance, with elevated values correlating with poorer overall survival and recurrence-free survival. Zhang et al. [16] reported hazard ratios of 1.62 for overall mortality and 1.83 for recurrence in HCC patients with higher APRI levels, underscoring APRI’s role in identifying high-risk HCC patients. In contrast, our study in a TBI population observed an AOR of 1.67 for mortality in patients with APRI values ≥0.59, indicating a significant association between elevated APRI and increased mortality risk in TBI. This AOR is similar to the hazard ratios reported in HCC studies, suggesting that APRI may serve as a marker of poor prognosis across diverse conditions.

Beyond liver-related conditions, APRI has been applied in COVID-19 and dengue fever. In COVID-19 patients, Çopur et al. [7] reported an AUC of 0.648 for APRI’s mortality prediction, closely mirroring our study’s findings in TBI. Their cutoff of 0.58 produced a sensitivity of 56.4% and specificity of 63.6%, indicating APRI’s moderate predictive power for systemic inflammatory conditions. This result aligns with our findings in TBI, where we observed an AUC of 0.648, sensitivity of 63.6%, and specificity of 60.3% at a similar cutoff value of 0.59. These comparable results between COVID-19 and TBI underscore APRI’s moderate yet consistent performance across diverse inflammatory and injury-related conditions. In contrast, our study’s findings diverge from those of Jamil et al. [8] in complicated dengue fever, where APRI demonstrated a high predictive accuracy for severe complications, with an AUC of 0.93, sensitivity of 84.9%, and specificity of 89.0% at a cutoff of 9.04. This stark contrast may stem from the acute inflammatory and hematological changes in severe dengue, which directly impact platelet counts and liver function, thereby enhancing APRI’s diagnostic relevance. In TBI, however, APRI’s moderate sensitivity and specificity suggest a weaker correlation between liver enzyme and platelet changes and mortality outcomes.

In this study, GCS remains a substantially stronger predictor of TBI mortality than APRI, as evidenced by GCS’s higher discriminative ability (AUC ~0.84 vs. ~0.65 for APRI). APRI’s prognostic performance was only moderate overall, and its utility appeared to depend on injury severity. In stratified analyses, an elevated APRI (≥0.59) did not significantly predict death in patients with mild injuries (ISS 1–15), whereas it was associated with markedly higher mortality risk in moderate (ISS 16–24) and severe (ISS ≥25) trauma. This pattern suggests that APRI’s prognostic signal emerges primarily when the overall injury burden is substantial—consistent with the notion that trauma-induced systemic insult (e.g., coagulopathy from thrombocytopenia) contributes to worse outcomes in more severe TBI [3]. Clinically, while APRI can serve as an adjunct prognostic marker (especially in moderately to severely injured patients), its moderate accuracy and dependency on injury severity warrant a cautious application alongside established indices like GCS, rather than as a standalone predictor.

The variability in APRI’s performance across studies can be attributed to multiple factors. Differences in underlying disease mechanisms likely play a significant role; APRI was initially developed as a liver fibrosis marker, making it directly applicable to liver diseases. In contrast, TBI and other inflammatory conditions may only indirectly influence APRI through systemic inflammation, which limits its predictive strength. Additionally, variations in study populations, APRI cutoff values, and the presence of comorbidities can significantly impact its performance. Moreover, differences in measurement timing (e.g., preoperative, at admission, or longitudinal) and study designs across settings add to the discrepancies in APRI’s predictive capacity. Clinically, APRI can serve as a valuable supplementary tool for predicting mortality in TBI patients, helping to identify those at higher risk early on. Its non-invasive nature and ease of calculation make it particularly useful for rapid assessment in emergency settings. However, due to its moderate sensitivity and specificity, APRI should be interpreted in conjunction with other validated trauma scores, such as the GCS or ISS, to provide a more comprehensive risk evaluation. Rather than relying on a single measurement, tracking APRI trends over time may offer deeper insight into the progression or resolution of liver dysfunction. By incorporating serial APRI assessments into clinical practice, healthcare providers can better detect patients at risk of worsening hepatic dysfunction or coagulopathy, enabling timely intervention with supportive therapies, hemodynamic stabilization, and liver-protective strategies.

This study has several limitations. First, its retrospective design introduces selection bias, particularly due to the exclusion of patients with incomplete data, and limits causal inferences regarding APRI’s role in TBI prognosis. Second, APRI’s predictive performance may be confounded by pre-existing liver conditions or systemic inflammation unrelated to TBI, affecting its reliability. In addition, although APRI has been proven to be a valuable predictive tool in both liver-related trauma/surgical cases and non-liver surgical settings, the incidence and extent of liver injury could be accurately estimated or eliminated from its effect on the APRI expression, thus leading to a limitation in the interpretation of the results. Third, as a single-center study, the findings may not be generalizable to other populations, necessitating external validation in multi-center cohorts. Fourth, while APRI showed moderate predictive accuracy, it was not compared with established critical illness scores and trauma scores such as Acute Physiology and Chronic Health Evaluation (APACHE), or the Sequential organ failure assessment (SOFA) due to data limitations; future studies should integrate these scores for a more comprehensive evaluation. Finally, the lack of longitudinal APRI measurements prevents analysis of its dynamic changes over time, which could provide further insights into its role in monitoring patient recovery and deterioration. Future prospective, multi-center studies with repeated biomarker assessments are warranted to validate APRI’s prognostic value and its integration into clinical decision-making.

5. Conclusions

In conclusion, while APRI demonstrated moderate predictive utility in TBI for mortality risk assessment, it is less effective than in chronic liver disease and other systemic conditions, highlighting the need for further investigation in trauma-specific applications.

Availability of data and materials

Not applicable.

Author contributions

CYH and CHT—wrote the manuscript. WTS—performed the research and validate the data from database. SYH—analyzed the data. CHH—designed the research study.

Ethics approval and consent to participate

With approval number 202401700B0, the procedure was authorized by Chang Gung Memorial Hospital’s Institutional Review Board (IRB) before the research started. The need of informed consent was waived by IRB due to the retrospective design on registered database.

Acknowledgment

We express our thanks to the Biostatistics Center at Kaohsiung Chang Gung Memorial Hospital for assisting us with statistical analyses.

Funding

This research was supported by a grant from CMRPG8P0751 for Ching-Hua Tsai.

Conflict of interest

The authors declare no conflict of interest.