The role of serum Dickkopf‐1 in predicting 30‐day death in severe traumatic brain injury

Abstract Objective Dickkopf‐1 (DKK‐1), an inhibitor of the canonical/‐catenin cascade of the Wnt pathway, was upregulated in brain tissues of hemorrhagic stroke rats, and its rising circulating levels were associated with poor prognosis of acute ischemic stroke patients. We attempted to ascertain the relationship between serum DKK‐1 levels and 30‐day death after severe traumatic brain injury (sTBI). Materials and methods Serum DKK‐1 levels were gauged in a total of 94 sTBI patients and 94 healthy controls. Trauma severity was assessed using Glasgow Coma Scale (GCS) and Rotterdam classification based on head computerized tomography scan. Prognostic variable was 30‐day death. Results Compared with controls, serum DKK‐1 levels were substantially elevated in patients (median value, 3.7 versus 1.0 ng/ml). Area under receiver operating characteristic curve was 0.802 (95% confidence interval (CI), 0.708–0.877) for predicting 30‐day death. Adjusted logistic regression showed that serum DKK‐1 levels above 3.7 ng/ml remained as an independent marker of 30‐day death (odds ratio, 8.573; 95% CI, 1.386–53.020) and overall survival (hazard ratio, 7.322; 95% CI, 1.320–40.622). An intimate correlation existed between DKK‐1 levels and GCS scores (r = −.649) in addition to Rotterdam classification (r = .664). Conclusions High serum levels of DKK‐1 are closely associated with increasing severity and rising short‐term mortality of sTBI.

brain diseases (Albanna & Ahmed, 2016;Caricasole et al., 2004;Dun et al., 2012;Matrisciano et al., 2011;Moors et al., 2012;Scott, Zhang, Han, Desai, & Brann, 2013). Moreover, induction of DKK-1 was a key factor for the development of ischemic neuronal death, and blocking DKK-1 could protect neurons against ischemic damage and decreased infarction volume in rat model with focal brain ischemia (Cappuccio et al., 2005;Mastroiacovo et al., 2009). Also, DKK-1 levels in rat brain tissue were obviously enhanced early after intracerebral hemorrhage; inhibition of DKK-1 could markedly ameliorated blood-brain barrier disruption and brain edema, lessened neurological deficits, increased the transcription of Wnt-1 mRNA, and upregulated the expression of tight junction protein zonula occludens-1 (Li et al., 2017). Clinically, serum DKK-1 levels were substantially elevated in acute ischemic stroke, stable angina, or myocardial infarction patients (He et al., 2016;Pérez Castrillón et al., 2010;Seifert-Held et al., 2011;Ueland et al., 2009). In addition, DKK-1 was an independent predictor for long-term poor prognosis of acute ischemic stroke (Zhu et al., 2019). However, there are currently no data available regarding circulating DKK-1 levels in patients with sTBI. In our study, we intended to investigate the change of DKK-1 in peripheral blood of sTBI patients and further determine the association between serum DKK-1 levels and prognosis of head trauma among sTBI patients.

| Study participants
This study was a prospective, observational study conducted at our hospital in Taizhou, China, from August 2014 to May 2018. The inclusion criteria were to meet all of the following: (a) age ≥ 18 years, (b) sTBI (Glasgow Coma Scale (GCS) score < 9 points, not under the influence of pharmacologic agents or alcohol), (c) time from head trauma to admission ≤ 6 hr, and (d) injury severity score ≤ 9 points in noncranial aspects. The exclusion criteria were to meet one of the following: (a) surgery or trauma with recent 4 weeks, (b) previous neurological disease, such as cerebral infarction, intracerebral hemorrhage, subarachnoid hemorrhage, and brain tumors, (c) specific medication usage, such as antiplatelet and anticoagulant medication, and (d) severe diseases in other organs, such as uremia, liver cirrhosis, and malignancy. Healthy volunteers constituted controls. The study adhered to the ethical conduct of research involving human subjects by World Medical Association Declaration of Helsinki.
The study was approved by the ethics committee at our hospital.

Controls wrote informed consent by themselves, while patients'
written informed consent was obtained from their relatives.

| Data collection
Baseline data on demographic characteristics (age and gender), history of smoking and alcohol consumption, medical history (dyslipidemia, hypertension, and diabetes mellitus), traumatic causes (automobile/motorcycle, fall/jump, or others), and details of drug usage were collected at the time of entry into emergency department. Former smokers who had quit smoking more than 2 years ago and sporadic alcohol consumers were excluded from the smoking and alcohol analysis. Trauma severity was evaluated using the postresuscitation GCS score (Nik et al., 2018). Pupillary reactivity was observed. Upon arrival at emergency department, each patient underwent head computerized tomography (CT) scan. We recorded the following trauma-related radiological information: abnormal cisterns, midline shift, skull-cap fracture, skull-base fracture, epidural hematoma, subdural hematoma, subarachnoid hemorrhage, intraventricular hemorrhage, cerebral hematoma, brain contusion, and pneumocephalus. Radiological severity was classified in light of the Rotterdam CT classification (Fujimoto, Miura, Otsuka, & Kuratsu, 2016). In addition, progressive hemorrhagic injury, acute lung injury, and post-traumatic cerebral infarction were defined based on the previous studies (Allard et al., 2009;Bernard et al., 1994;Mirvis, Wolf, Numaguchi, Corradino, & Joslyn, 1990). All patients were followed up until death or the completion of 30 days. A 30-day death was regarded as the study endpoint.

| Immune analysis
Blood samples of all patients were obtained at entrance into emergency department, while those of controls were acquired at study entry. Serum C-reactive protein levels, blood glucose, blood cell counts, and blood coagulative function were measured using the conventional methods. Acute traumatic coagulopathy was defined according to the previous reports (Franschman et al., 2012;Greuters et al., 2011). All serum samples were separated and at once frozen at −70°C until assayed. Serum DKK-1 levels were determined in duplicate samples with a commercially available enzyme-linked immunosorbent assay kit (R&D Systems, Inc). A standard curve was constructed from which DKK-1 concentrations of unknown samples were quantified. Laboratory technicians who detected serum DKK-1 were inaccessible to the clinical characteristics and outcomes of the study participants.

| Statistical methods
Baseline characteristics of study participants were presented.
Continuous variables were summarized as median (the upper-lower quartiles), and categorical variables were presented as counts (percentage). Baseline characteristics between nonsurvivors and survivors were compared using a Mann-Whitney U test or the chi-squared test as appropriate. Initial univariate analysis was done to assess the statistical significance of the observed difference between nonsurvival and survival groups for each parameter.
Multivariate logistic regression and Cox proportional hazards models were used to assess the association between serum DKK-1 levels and 30-day death when appropriate. Odds ratio (OR), hazard ratio (HR), and 95% confidence interval (CI) were calculated. All parameters that were revealed to be significant (p < .05) in the univariate analysis were further analyzed using multivariate analysis to identify those parameters that retained significance while accounting for all relevant variables. Differences in terms of serum DKK-1 levels were compared among multiple groups using the Kruskal-Wallis H test.
Bivariate correlations were analyzed using Spearman's rank correlation coefficient. A receiver operating characteristic curve (ROC) was constructed with serum levels of DKK-1 as prognostic variable and 30-day death as classification variable. Area under ROC (AUC) and its 95% CI were estimated. A combined logistic regression model was configured to assess the additive effect of serum DKK-1 levels combined with other variables. Two-tailed p < .05 was considered to be statistically significant. All analyses were performed utilizing the Statistical Package for the Social Sciences version 19.0 (SPSS Inc.).

| RE SULTS
During the study period, we at first assessed a total of 125 patients with sTBI based on the inclusion criteria, and further, in accordance with the exclusion criteria, we removed 31 sTBI patients because of the following reasons: (a) surgery (one cases) and trauma (two cases) with recent 4 weeks; (b) previous neurological diseases, including cerebral infarction (two cases), intracerebral hemorrhage (three cases), subarachnoid hemorrhage (two cases), brain tumors (two cases), and others (four cases); (c) specific medication usage, including antiplatelet (three cases), anticoagulant medication (two cases), and others (three cases); and (d) severe diseases in other organs, including uremia (one cases), liver cirrhosis (one cases), malignancy (two cases), and others (three cases). Eventually, 94 patients with sTBI were analyzed. Additionally, there were 94 controls composed of healthy volunteers. In terms of age and gender percentage, no significant differences existed between patients and controls.
This group of sTBI patients contained 56 males and 38 females, and their age ranged from 18 to 73 years (median, 42 years; the upper-lower quartiles, 29-53 years). Hypertension, diabetes mellitus, and dyslipidemia were found in 15, 12, and 16 patients, respectively. A total of 44 patients smoked cigarettes, and 50 patients belonged to alcohol consumers. As regards traumatic causes, most patients (45 patients, 47.9%) were traumatized by automobile/motorcycle, the second cause was fall/jump (38 patients, 40.4%), and other causes appeared in 11 patients (11.7%). Clinical severity was assessed using postresuscitation GCS scores ranging from 3 to 8, with a median value of 5 (the upper-lower quartiles, 4-7). Unreactive pupils were observed at admission among 43 patients (45.8%). All patients were admitted within 6.0 hr after trauma (range, 0.5-6.0 hr; median, 2.3 hr; the upper-lower quartiles, 2.3-3.4 hr). Abnormal cisterns occurred in 74 patients, and midline shift above 5 mm appeared in 59 patients. The other trauma-related radiological positive appearances were listed in Table 1. There was a median value of 5 at Rotterdam CT classification (range, 3-6; the upper-lower quartiles, 4-6). In total, 58 patients underwent surgery in the first 24 hr after trauma. In addition, there were acute traumatic coagulopathy in 29 patients (30.9%), progressive hemorrhagic injury in 21 patients (22.3%), acute lung injury in 25 patients (26.6%), and post-traumatic cerebral infarction in 11 patients (11.7%).
In the current study, among sTBI patients, the median value of serum DKK-1 levels was 3.7 ng/ml, its levels ranged from 0.9 to 7.9 ng/ml, and its interquartile range was from 2.4 to 5.2 ng/ml; simultaneously, serum DKK-1 levels of controls, with a median value of 1.0 ng/ml, ranged from 0.4 to 3.4 pg/ml (interquartile range, 0.8-1.8 pg/ml). Next, statistical analysis showed that serum DKK-1 levels were profoundly higher in sTBI patients than in controls (p < .001).
In this study, there were 22 patients (23.4%) who died within 1 month following head trauma. Serum DKK-1 levels of nonsurvivors were substantially higher than those of controls (median, 5.4 ng/ml; range, 1.8-7.9 ng/ml; interquartile range, 4.3-7.0 ng/ ml versus median, 3.3 ng/ml; range, 0.9-6.9 ng/ml; interquartile range, 2.1-4.7 ng/ml). However, patients were divided into two groups in accordance with the median value of serum DKK-1 levels (3.7 ng/ml). Just as displayed in Table 2, as compared to survivors, nonsurvivors tended to show a significantly higher proportion of serum DKK-1 levels above 3.7 ng/ml, unreactive pupils, abnormal cisterns, midline shift > 5 mm, acute traumatic coagulopathy, acute lung injury, progressive hemorrhagic injury, and post-traumatic cerebral infarction, as well as were likely to have markedly older age, higher serum C-reactive protein levels, higher blood glucose levels, higher blood leukocyte count, lower admission GCS scores, and higher Rotterdam CT classification (all p < .05). Afterward, all the preceding variables found significant in univariate analyses were incorporated in the binary logistic regression model, and subsequently, it was revealed that serum DKK-1 levels > 3.7 ng/ml, GCS scores, and Rotterdam CT classification retained as the three independent predictors for 30-day death, with OR values of 8.573 (95% CI, 1.386-53.020), 0.370 (95% CI, 0.189-0.723), and 2.967 (95% CI, 1.526-5.769), respectively.
In this group of sTBI patients, the mean overall survival time was 25.3 days (95% CI, 23.3-27.2 days). In Figure 1, patients with serum DKK-1 levels more than 3.7 ng/ml showed significantly TA B L E 1 The trauma-related radiological positive appearances among patients with traumatic brain injury shorter overall time than other remainders.  Figure 2). Similarly, while serum DKK-1 level was identified as a continuous variable, comparative analysis among multiple groups showed that, with increasing CT grade or decreasing GCS scores, serum DKK-1 levels were significantly raised (Table 4). Moreover, serum DKK-1 levels were dichotomized based on its median value (3.7 ng/ml) and thereby, it was found that, patients with higher CT TA B L E 3 The factors associated with 30-overall survival after traumatic brain injury Note: Data were yielded using the univariate Cox's proportional hazard analysis.

F I G U R E 2
Graph depicting intimate and inverse correlation of serum Dickkopf-1 levels with Glasgow Coma Scale (GCS) scores among traumatic brain injury patients and portraying close and positive correlation of serum Dickkopf-1 levels with Rotterdam computerized tomography classification in patients with severe traumatic brain injury. CT means computerized tomography grade or lower GCS score had a significant higher percentage of serum DKK-1 levels above 3.7 ng/ml (Table 4).

| D ISCUSS I ON
Up to date, there have been many experimental studies detecting increased expression of DKK-1 in brain tissues with acute brain injury (Cappuccio et al., 2005;Mastroiacovo et al., 2009). Also, it has been verified that DKK-1 exerted a harmful effect on acute brain injury (Cappuccio et al., 2005;Mastroiacovo et al., 2009). In an intracerebral hemorrhage model, serum level of DKK-1 did not differ between the intracerebral hemorrhage and sham groups (Li et al., 2017). Interestingly, the previous two epidemiological investigations showed that, in humans with acute ischemic stroke, DKK-1 levels in the peripheral blood were actually higher as compared with healthy controls (He et al., 2016;Seifert-Held et al., 2011). To the best of my knowledge, our study is the first one determining circulating DKK-1 levels in head trauma patients. Based on our study enrolling a total of 94 sTBI and 94 controls, it was found that serum DKK-1 levels were substantially elevated after sTBI in humans.
As regards the relationship between circulating DKK-1 levels and disease severity of acute brain injury, only the two study has been done, in which no correlation of DKK-1 levels was found with stroke severity (reflected by the National Institutes of Health Stroke Scale) in human acute ischemic stroke (He et al., 2016;Seifert-Held et al., 2011 Abbreviations: CT, computerized tomography; GCS, Glasgow Coma Scale.
TA B L E 4 Differences of serum Dickkopf-1 levels by trauma severity levels showed significant prognostic accuracy in discriminating nonsurvivors from survivors. In addition, serum DKK-1 levels did not improve the prognostic power of GCS scores, while it enhanced that of Rotterdam CT classification. In summary, serum DKK-1 might serve as a potential prognostic biomarker for short-term mortality in human head trauma.

| CON CLUS IONS
Our study provides the first evidence for a release of DKK-1 into the circulation in patients with sTBI. We also find that elevated serum DKK-1 levels at baseline are associated with severity and death at 30 days after sTBI, indicating that DKK-1 may be a potential prognostic biomarker for sTBI.

ACK N OWLED G M ENTS
The authors thank all participants for their providing blood samples.

CO N FLI C T O F I NTE R E S T S
The authors have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.