Introduction

TBI has attracted much attention for decades due to the high prevalence and unfavorable outcome [1]. Although much progress has been made on medical treatments for improving prognosis of TBI patients, the poor outcome caused by TBI has not been completely eliminated. It is necessary to evaluate risk of possible adverse progress in early stage after suffering initial brain injury, which could help clinicians to adjust treatment strategies and provide intensive care. Exploring potential prognostic marker for prognosis of TBI patients could be helpful for risk assessment of these patients.

As one kind of cysteine protease inhibitor proteins, the cystatin C could be continuously transcribed and expressed at a constant rate in all nucleated cells without tissue specificity [2]. In addition, the serum concentration of cystatin C could not be easily influenced by inflammation, infection, tumor and liver function. Since cystatin C in the blood is excreted solely through glomerular filtration, it is widely acknowledged as a marker reflecting changes of glomerular filtration rate and is confirmed effective in predicting adverse renal events such as AKI in many clinical settings [3,4,5,6,7]. In addition to the clinical significance of renal function, cystatin C has also been explored to evaluate risk of mortality in various kinds of patients including those received intensive care, diagnosed with acute respiratory distress syndrome, coronavirus disease 2019, cirrhotics, pulmonary arterial hypertension or heart failure [8,9,10,11]. Several research confirmed that cystatin C was useful in predicting mortality of brain injury patients including those suffered ischemic or hemorrhagic stroke [12,13,14,15]. While the prognostic value of cystatin C in TBI patients has not been verified. We designed this study to testify whether serum cystatin C level measured on admission could be valuable in predicting mortality of TBI patients. Besides, we analyzed the correlation between cystatin C, AKI, brain injury severity and the outcome of TBI patients.

Materials and methods

Study patients

Patients received medical treatments in the ward of West China hospital including the ICU ward for TBI between January 2015 and June 2019 were eligible for this study. Patients were excluded if they: (1) admitted to hospital 6 h after suffering injury; (2) transferred from other hospitals; (3) lacked in included variables (4) complicated with chronic renal diseases. The flowchart of exclusion is shown as Fig. 1. Finally, 415 TBI patients were included in this study. This study obtained approval from the ethics committee of West China hospital and performed abiding by Declaration of Helsinki. Informed consent forms about joining observational research of each patient were regularly signed by patients themselves or their legally authorized representatives once admitted.

Fig. 1
figure 1

Flowchart of patients’ inclusion

Full size image

Data collection

Clinical variables including injury mechanism, blood pressure on admission, Glasgow Coma Scale (GCS) on admission, injury severity score (ISS), Sequential Organ Failure Assessment (SOFA), laboratory tests, intracranial injury types and AKI were collected. Results of laboratory tests were obtained by analyzing the first blood sample collected once patients admitted to emergency department. Outcome of this study was in-hospital mortality in included TBI patients. And AKI was diagnosed based on the KDIGO guidelines using creatinine criteria.

Statistical analysis

Normality of included variables was testified using Kolmogorov–Smirnov test. Normally distributed and non-normally distributed variables were presented as mean ± standard deviation and median (interquartile range), respectively. And categorical variables were shown as numbers (percentage). Independent Student’s t test and Mann–Whitney U test were, respectively, utilized to verify the difference between two groups of normally distributed and non-normally distributed variables. χ2 test or Fisher test was utilized to analyze the difference of categorical variables. Univariate logistic regression and multivariate logistic regression with forward stepwise method were sequentially performed to explore risk factors of mortality. Factors significant in multivariate logistic regression were then combined to construct prognostic model. Next, we conducted mediation analysis using linear regression to examine whether AKI and brain injury severity mediated the relationship between serum cystatin C level and mortality in included TBI patients. Mediation analysis is used to explore the potential mechanism between two variables and to examine whether mediating variables affect the relationship [16]. The β coefficients and corresponding statistical significance of relationships between cystatin C, AKI, GCS and mortality were calculated and then confirmed the contribution proportion of mediating effects of AKI and GCS on the relationship between serum cystatin C level and mortality. Finally, receiver operating characteristic (ROC) curves were drawn and area under the ROC curve (AUC) was calculated to evaluate the prognostic value of single serum cystatin C level and the constructed model. Z test was utilized to compare the AUC difference between single serum cystatin C and the prognostic model.

Two-sided p value < 0.05 was considered to be of statistical significance. SPSS 22.0 Windows software (SPSS, Inc, Chicago, IL) was used for all statistical analyses and figures drawing.

Results

Baseline characteristics of included TBI patients

Among 415 included TBI patients, 203 suffered death with mortality of 48.9%. Compared with survivors, non-survivors had lower GCS (5 vs 8, p < 0.001), higher ISS (25 vs 16, p < 0.001) and SOFA (7 vs 4, p < 0.001) (Table 1). Results of laboratory test presented that white blood cell (15.49 vs 13.66, p = 0.011), glucose (12.39 vs 7.88, p < 0.001), BUN (7.04 vs 5.66, p < 0.001) serum creatinine (83 vs 64, p < 0.001) and cystatin C (0.92 vs 0.71, p < 0.001) were all higher in non-survivors. While platelet (80 vs 139, p < 0.001) and albumin (28.2 vs 34.6, p < 0.001) were lower in non-survivors than survivors.

Table 1 Baseline characteristics of included TBI patients
Full size table

Logistic regression analysis for risk factors of mortality in TBI patients

Univariate logistic regression showed ISS (p < 0.001), SOFA (p < 0.001), WBC (p = 0.013), glucose (p < 0.001), BUN (p < 0.001), cystatin C (p < 0.001), subdural hematoma (p < 0.001), subarachnoid hemorrhage (p = 0.001), diffuse axonal injury (p = 0.009) and AKI (p < 0.001) were potential risk factors of mortality in included TBI patients (Table 2). While GCS (p < 0.001), albumin (p < 0.001), hemoglobin (p < 0.001) and platelet (p < 0.001) were negatively correlated with mortality in included TBI patients. After adjusting confounding effects, multivariate logistic regression indicated GCS (p < 0.001), glucose (p < 0.001), albumin (p = 0.009), cystatin C (p < 0.001) and subdural hematoma (p = 0.042) were independent risk factors of mortality in included TBI patients.

Table 2 Univariate and multivariate logistic regression analysis of risk factors for mortality in TBI patients
Full size table

Association between cystatin C, GCS, AKI and mortality in TBI patients

To explore whether correlation between cystatin C and mortality of included TBI patients is mediated by AKI and brain injury, mediation analysis was adopted by performing linear regression. The β efficient of correlation between cystatin C and mortality was 0.323 (p < 0.001) (Table 3). Cystatin C was also correlated with GCS and AKI with β efficient of − 0.164 (p = 0.001) and 0.524 (p < 0.001), respectively. Additionally, GCS and AKI were correlated with mortality of included patients with β efficient of − 0.516 (p < 0.001) and 0.249 (p < 0.001), respectively. The contribution rate of mediating effect of AKI to the correlation between cystatin C and mortality was 40.40% (0.524*0.249/0.323). And the contribution rate of mediating effect of brain injury severity to the correlation was 26.20% (0.164*0.516/0.323). The chart of mediation analysis process is presented as Fig. 2.

Table 3 Correlations between cystatin C, GCS, AKI and mortality
Full size table
Fig. 2
figure 2

Association between cystatin C, GCS, AKI and mortality in TBI patients by mediation analysis

Full size image

Prognostic value of cystatin C and predictive model in TBI patients

The AUC of GCS and cystatin C alone for predicting mortality was 0.814 and 0.731, respectively (Table 4) (Fig. 3). The AUC of combining GCS with cystatin C was 0.862, which was higher than that of GCS alone (Z = 1.7354, p < 0.05). Composed of GCS, glucose, albumin and cystatin C, the predictive model had the highest AUC of 0.899, though without statistical significance when compared with 0.862 of combining GCS with cystatin C (Z = 1.5791, p > 0.05).

Table 4 Prognostic value of cystatin C and the constructed predictive model in TBI patients
Full size table
Fig. 3
figure 3

Prognostic value of cystatin C and predictive model in TBI patients

Full size image

Discussion

Among included TBI patients, 203 (48.9%) patients suffered death and 82 (19.8%) developed AKI. The serum cystatin C level was significantly higher in non-survivors than survivors. And GCS, glucose, albumin, cystatin C and subdural hematoma were finally found as significant risk factors. The reference level of serum cystatin C is 0.51–1.09 mg/L in clinical practice. Therefore, the cystatin C level of patients in survivors was mainly within the normal range, whereas level in non-survivors was 0.92 (IQR: 0.73–1.22) which indicated a part of non-survivors had increased cystatin C level above normal physiological range.

The increased cystatin C in TBI patients might be attributable to multiple mechanisms. The most apparent cause of correlation between cystatin C and outcome of TBI patients is the AKI development which reflected by increased cystatin C level. The linear regression in the study indicated the β coefficient between cystatin C and AKI was 0.524 with statistical significance (p < 0.001). And AKI was also associated with mortality with β coefficient of 0.249 (p < 0.001). Actually, AKI has been verified as a risk factor of poor prognosis and commonly occurred in TBI patients with incidence of 9.2–24% [17,18,19,20,21]. The development of AKI in TBI may indicate the severity of initial and secondary brain injury and systemic inflammation. And AKI may aggravate the brain edema which in turn promote the progression into poor prognosis. In our study, the contribution proportion of the mediating effect of AKI to the correlation between cystatin C and mortality was 40.40%, indicating AKI plays a major role in the process of increased cystatin C exerting effect on mortality of TBI patients. While non-renal mechanism may be another explanation for the correlation between cystatin C and mortality. In our study, the linear regression revealed that cystatin C was associated with GCS (β = − 0.164, p < 0.001) and GCS was correlated with mortality (β = − 0.516, p < 0.001). And the contribution rate of mediating effect of brain injury severity (GCS) to the correlation between cystatin C and mortality was 26.20%, which indicated increased cystatin C level might also reflect more severe brain injury. This conclusion was consistent with findings of one previous study [22]. This study discovered that serum cystatin C concentration was negatively related with TBI severity and positively related with levels of inflammatory mediators including hsCRP (high sensitive C-reactive protein), white blood cells, and IL-6 (interlukin-6) and TNF-α (tumor necrosis factor-α) in TBI patients. Summarizing findings of ours and this study, we concluded that increased cystatin C might also contributing to poor outcome of TBI patients through reflecting brain injury severity and systemic inflammation level. Actually, the increased production of cystatin C in brain injury is considered as a kind of automatic neuroprotective responses [23]. Cystatin C could prevent oxidative injury and inactivate cathepsins, which is a cysteine-proteases released after ischemic brain injury and could cause increased neuronal death [24,25,26]. Therefore, we make a reasonable assumption that more severe traumatic brain injury which included a serious of pathophysiological process including local cerebral inflammation and oxidative stress would provoke more intense neuroprotective response of cystatin C.

Multivariate logistic regression in our study showed that GCS, glucose, albumin, cystatin C and subdural hematoma were independent risk factors of mortality in included TBI patients. The AKI, statistically significant in univariate logistic regression, however, did not show statistical significance in multivariate regression. This fact may be caused by the stronger predictive value of cystatin C which also indicate the possible AKI development and consequently cover the relationship between AKI and mortality. The AUC comparison revealed that integrating cystatin C with GCS could elevate the predictive accuracy and sensitivity for mortality of TBI patients. The prognostic model we constructed incorporating cystatin C may be more effective in evaluating the risk of poor outcome of TBI patients in early stage.

There were several limitations in this study. Firstly, this observational study was conducted in a single medical center with moderate sample size. Inevitable selection bias might limit the reliability of our conclusion. Future studies with larger sample size are worthwhile to be conducted to verify our conclusions. Secondly, initial cystatin C level was only recorded on admission so that the optimal prognostic value of cystatin C in other timepoint and the effects of cystatin C fluctuation on prognosis cannot be further explored. Thirdly, factors influencing the baseline level of cystatin C such as diabetes mellitus, hypertension, stroke and heart failure were not collected in this study. Though a rough screening of the included patients showed the extremely low prevalence of these factors. Finally, common biomarkers of TBI severity such as S100-β, glial fibrillary acidic protein, neurofilament light chain were not routine laboratory tests in our hospital so that we could not show the level of biomarkers for TBI severity. Future studies including these markers may better reflect the mediating effect of brain injury severity on the association between the cystatin C and mortality of TBI patients.

Conclusion

The correlation between serum cystatin C and mortality of TBI patients is not completely mediated by AKI. As a marker of injury severity, cystatin C is effective in predicting possible poor outcome of TBI patients.