Gui-long Feng, Miao-miao Zheng, Shi-hong Yao, Yin-qi Li, Shao-jun Zhang, Wei-jing Wen, Kai Fan, Jia-li Zhang, Xiao Zhang

1 Department of Emergency, the First Hospital of Shanxi Medical University, Taiyuan 030001, China

2 Department of Emergency, Zhenjiang First People’s Hospital, Zhenjiang 212000, China

3 Department of Emergency, Yuncheng Central Hospital, Yuncheng 044500, China

4 Department of Endocrinology, the Fourth Affi liated Hospital, Zhejiang University School of Medicine,Yiwu 322000, China

KEYWORDS: Adrenocortical insuffi ciency; Risk factor; Predictor; Traumatic brain injury

INTRODUCTION

Severe traumatic brain injury (TBI) is a leading cause of death and disability.[1,2]Neuroendocrine dysfunction after TBI has received increased attention in recent years to further understand its pathophysiology.[3,4]The dysfunctions of the hypothalamic-pituitary-adrenal(HPA) axis deserve serious attention due to their impact on the recovery of neural function.[5,6]

The HPA axis is a key mediator in the protective stress-immune pathway after TBI as it regulates the secretion of glucocorticoids and other hormones.[6]Hypopituitarism and/or adrenocortical insufficiency(AI) can lead to inappropriate responses to stress;affect homeostatic balance, inflammation, sleep, and psychiatric disorders;[6]and aggravate nerve damage and brain edema.A high incidence (25%-70%) of AI after severe TBI has been reported in previous studies.[5,7,8]AI can also inf luence the hemodynamics, and adrenal crisis is a life-threatening emergency that requires immediate recognition and treatment.[9]These effects affect the patient’s outcome adversely.[5,10]

AI after brain injury is often ignored by physicians,owing to the lack of specific symptoms.Moreover,plasma adreno-cortico-tropic-hormone (ACTH) and cortisol levels are not routinely tested in clinical practice.A delayed or omitted diagnosis of AI may result in a poor outcome in patients with TBI.

This study aims to investigate the incidence of AI after brain injury.Further, it will evaluate 14 possible related impact factors to identify risk factors and independent predictive variables of AI after TBI.Based on this, a prediction model will be built to aid in the early identif ication or diagnosis of AI after TBI.

METHODS

Patients

Patients with acute TBI were prospectively enrolled in the study over a 24-month period from the Emergency Trauma Centre of the First Hospital of Shanxi Medical University.Patient consent was obtained at the time of enrolment.

The patients were enrolled according to the inclusion and exclusion criteria.Brain injury was conf irmed by at least three experts from the committee of trauma, and the severity of TBI was assessed on admission using the Glasgow Coma Scale (GCS) score.

Inclusion criteria were: confirmed to have suffered head injury; admission within the first 24 hours after injury; age over 14 years; and TBI verified by at least three neurosurgeons via computed tomography (CT) or magnetic resonance imaging (MRI).

Exclusion criteria were: endocrine diseases such as Addison’s disease, Sinmond’s disease, Sheehan’s disease,and thyroid disease; pituitary tumor or cranopharyngeal tube tumor; long-term use of hormones in the past; a history of seizures; complicated with serious infections;major torso trauma; brain death after brain trauma.

Preset impact factors

Based on previous studies and observations in clinical practice, 14 possible impact factors were preset—gender,age, GCS score on admission, vital signs (body temperature,pulse, respiratory, and mean arterial pressure [MAP]),urinary volume, serum sodium level, cerebral hernia, frontal lobe contusion, cisterna-ambiens hemorrhage, diff use axonal injury (DAI), and skull base fracture.

Data collection

Baseline demographics (gender and age) and injury-specific data were recorded, including admission GCS score, admission vital signs, clinical symptoms/manifestations indicating cerebral hernia or basilar skull fracture (such as severe headache, frequent vomiting,other symptoms of raised intracranial pressure, changes in consciousness, vital signs and pupils, pyramidal sign,raccoon eyes, cerebrospinal fluid [CSF] otorrhea or rhinorrhea, and Battle’s sign), urinary volume per day,head CT or MRI f indings (such as cerebral hernia, frontal lobe contusion, cisterna ambiens hemorrhage, DAI, and intracranial gas), and serum sodium, ACTH, and cortisol concentrations (0:00, 8:00 and 16:00).

Data grouping

Patients were categorized into the AI group when the patients’ serum ACTH concentrations were ＜1.6 pmol/L (the lowest value in the normal range) or the cortisol circadian rhythm was abnormal and the cortisol concentration was＜154 nmol/L within ten days after TBI.Other patients were enrolled into the non-AI group.

In accordance with the requirements of statistical methods, patients were regrouped considering each impact factor as a categorical variable.(1) Gender:male and female; (2) age: older group (≥60 years) and young group (＜60 years) according to the World Health Organization (WHO) definition for older person; (3)GCS score: mild group (GCS 13-15), moderate group(GCS 9-12), and severe group (GCS 3-8); (4) four vital signs—body temperature (T), pulse (P), respiratory rate(RR), and MAP: normal groups and abnormal groups;the dichotomization was based on whether the value was within the range of compensation; the compensation ranges in the normal groups were 36.0-38.4 ℃ (T),50-100 beats/minute (P), 12-24 breaths/minute (RR), and 80-110 mmHg (MAP) (1 mmHg=0.133 kPa); (5) urinary volume: normal group (400-2,500 mL) and abnormal group (＞2,500 or ＜400 mL); (6) serum sodium level:normal group (135-145 mmol/L) and abnormal group(＞145 mmol/L or ＜135 mmol/L); (7) head CT or MRI: yes or no group based on the presence or absence of injury,such as cerebral hernia, frontal lobe contusion, cisterna ambiens hemorrhage or/and DAI on CT or MRI; (8)skull base fracture: yes or no group based on the clinical symptoms/manifestations, such as raccoon eyes, CSF otorrhea or rhinorrhea, Battle’s sign, and intracranial gas.

Statistical analysis

Data were analyzed using the SPSS 22.0 statistical software.AP-value ＜0.05 was considered statistically significant.Continuous variables were presented as mean±standard deviation (SD).Categorical variables were presented as enumeration data.

Univariate analysis was performed on each preset impact factor.The Chi-square test was used to analyze these categorical variables, while Wilcoxon rank-sum test was used to analyze ranked data.AP-value ＜0.05 was the screening criterion for a variable (impact factor) to be entered into the logistic regression model.We assigned values to variables that were entered in the logistic regression equation, and multiple logistic regression analysis was performed using the SPSS.Regression coefficients were tested by Waldχ2, and aP-value ＜0.05 was considered statistically significant.The variable with aP-value ＞0.05 was removed from the model.The related independent risk factors were established, and a prediction model of AI was developed.The logistic regression equation was presented as logitP=ln (P/1-P) =β0+β1X1+β2X2+… +βmXm.Pwas the probability of secondary AI in a patient with TBI, 1-Pwas the probability of normal adrenal cortical function,β0was the constant term,βmwas the regression coeffi cient, andXmwas the risk factor.

RESULTS

A total of 108 patients with acute TBI were enrolled in the study, of whom 75 patients were males and 33 patients were females.The average patient age was 51.69±7.07 years.The main causes of injury were falls,violence, and traffi c accidents.

Incidence of AI and nine related factors

The incidence of AI was 31.5% (34/108) after TBI(Table 1).The incidence of AI according to each preset factor and the analysis results are shown in Table 1.

Univariate analysis of categorical variables (i.e.,13 preset factors excluded GCS score) was performed using Chi-square test (Table 1), meanwhile analysis of ranked data (i.e., GCS score) was performed using the Wilcoxon rank-sum test (Table 1).In nine factors (old age, abnormal MAP, abnormal urinary volume, alteration of serum sodium, cerebral hernia, frontal lobe contusion,DAI, skull base fracture, and low GCS score), thediff erences between the AI group and non-AI group were statistically signif icant (P＜0.05), indicating that the nine factors were probably related to the occurrence of AI after TBI.The incidence of AI was especially high in the moderate and severe groups based on the GCS score.

Table 1.Clinical data in the AI group and non-AI group (n, %) and results of Chi-square test (t/χ2) or Wilcoxon rank-sum test (Z) on preset related factors of AI after TBI

Meanwhile, the differences were not statistically significant in five other factors (gender, body temperature, pulse, respiratory, and cisterna-ambiens hemorrhage), indicating that the five factors probably had no eff ect on AI after TBI.

Three independent variables and a multifactor regression prediction model of AI after TBI

As shown in Table 2, the nine factors (variables)mentioned above with aP-value ＜0.05 in univariate analysis were entered into multiple logistic regression analysis.The assignment of the variables is described in Table 2.

The results of the multiple logistic regression analysis are shown in Table 3.ThePvalues of three variables (urinary volume [X4], serum sodium level[X5], DAI [X8]) were ＜0.05 in the test of regression coefficients.The three variables entered into the regression model were statistically significant, and the regression coefficients were 2.583 (β4), 2.235 (β5) and 2.269 (β8).The results indicated that the three factors were independent variables.The developed prediction model of AI after TBI was logitP=ln (P/[1-P])=-3.552+2.583X4+2.235X5+2.269X8, respectively.AI after TBI was the dependent variable asYvariable, andPwas the probability of secondary AI after TBI in a patient.

Table 2.The factors entered into multiple logistic regression and the assignment

Table 3.Logistic regression parameter estimation and testing

DISCUSSION

Appropriate responses to acute stress would promote survival by altering physiological processes and behavior,such as redistribution of energy and increasing availability of fuels.[11]Damage to the HPA axis or secondary AI after TBI would impair the protective stress response of the body,affect hemodynamics, and exacerbate central inflammation and neurodegeneration.[6,11]Moreover, cortisol deficiency is prone to cause the adrenal crisis.[9,12]Therefore, the early recognition and management of high-risk patients with AI are crucial for neural function recovery.The awareness of predictive risk factors is extremely helpful in the early recognition of AI after TBI.

In the 14 preset factors investigated in this study,nine factors were possibly related to AI after TBI in univariate analysis, and three risk factors were the independent related predictive variables in multiple logistic regression analysis.The prediction model of AI after TBI should be helpful for assessing the possibility of AI after TBI.

Gender was a preset impact factor not related to AI in this study.The incidence of AI was 33.3% in males and 27.3% in females (P＞0.05).

The incidence of AI in older patients (45.5%) was signif icantly higher than that in younger patients (25.3%;P＜0.05).Old age was proposed as a related impact factor of AI after TBI possibly because the protective stress reactions in elderly people gradually decline, and they have poor pituitary-adrenal storage.The results were consistent with those of previous reports.[13,14]Although Cohan’s study indicated that a lower age was related to the occurrence of AI,[8]the patients enrolled were diff erent from those enrolled in our study.

The results also showed that the severity of brain injury was related to AI after TBI, and the GCS score was presumed to be a related risk factor of AI.The incidence of AI increased with a decrease in the GCS score, and the incidence of AI was 56.0% in the severe group (Table 1).The results were consistent with those of previous studies.[8,13,14]Brain edema, vascular spasm,hemorrhage, or high intracranial pressure (ICP) could cause severe ischemia or damage to the hypothalamuspituitary in the severe group.The regulating function of the HPA axis is then impaired, resulting in AI.[4,15,16]

Of the vital signs analyzed, only abnormal MAP was suggested to be related to AI after TBI.The incidence of AI in the abnormal MAP group (57.1%) was signif icantly higher than that in the normal MAP group (22.5%;P＜0.05).The maintenance of cerebral blood f low (CBF)is dependent on the cerebral perfusion pressure (CPP).A lower MAP or/and a higher ICP could lead to a reduced CPP and a reduced CBF.This can cause the ischemia of hypothalamus and pituitary gland.Some other studies have also supported that the occurrence of AI after TBI is related to early brain ischemia or high cranial pressure.[4,8,16]

Abnormal alteration of urine volume and even diabetes insipidus (DI) after TBI often indicate the impairment of the hypothalamus or/and pituitary that causes AI.In our study, the incidence of AI in the abnormal urine volume group (69.6%) was signif icantly higher than that in the normal group (21.2%;P＜0.05),confirming the relationship between alteration of urine volume and AI after TBI.This alteration was presumed to be a related risk factor of AI.In addition to the selfregulation of urine volume in kidneys, the alteration also relies on a variety of neuro-humoral factors such as aldosterone, antidiuretic hormone (ADH), and atrial natriuretic peptide.The impairment of HPA could disrupt the central regulation of urine volume.[16-18]

Alteration of serum sodium level was related to AI after TBI and was confirmed to be an independent risk factor in our study.Seven patients had hypernatremia(f ive of them had DI), and 39 patients had hyponatremia(12 of them had DI).The incidence of AI in the abnormal sodium group (54.3%) was signif icantly higher than that in the normal group (14.5%;P＜0.05).Hyponatremia might be associated with an impaired HPA axis in addition to decreased sodium intake, dehydration, and diuresis.It could also be caused by abnormal secretion of aldosterone and cortisol, syndrome of inappropriate secretion of ADH (SIADH), or cerebral salt wasting syndrome.[18-20]However, hypernatremia is often caused by central DI,[21]or cerebral salt retention syndrome.Alteration in sodium level after TBI often indicates the dysfunction of these regulatory hormones.

The formation of cerebral hernia means a reduction in perfusion, extensive ischemia, and edema in the brain tissue, and subsequently causes HPA damage.In our study, the incidence of AI in the cerebral hernia group(53.8%) was significantly higher than that in the noncerebral hernia group (24.4 %;P＜0.05).We concluded that if TBI patients had brain hernia, the possibility of secondary AI would increase.

In this study, four trauma factors were analyzed, and three factors were suggested to be related to AI after TBI—frontal lobe contusion (40.0% vs.20.8%,P＜0.05),skull base fractures (50.0% vs.23.0%,P＜0.05), and DAI(57.1% vs.22.5%,P＜0.05) (Table 1).The pituitary gland is at the base of skull and close to the frontal lobe.The skull base fracture or/and the contusion at the bottom of the frontal lobe could damage the hypothalamus and pituitary or its blood supply.[14,16,22]The hypothalamohypophyseal portal system is also more prone to damage than any other brain tissue in DAI.[23]DAI was an independent risk factor in our multivariate analysis.

CONCLUSIONS

In summary, the incidence of AI increases with a decrease in the GCS score.Nine factors are possibly related to AI after TBI (age, GCS score, MAP, urinary volume, serum sodium level, brain hernia, frontal lobe contusion, DAI, and skull base fracture).Plasma ACTH and cortisol levels should be routinely tested in clinical practice if the risk factors are identified.Three risk factors (abnormal urinary volume, serum sodium level,and DAI) are the independent predictive variables.The identifications of any risk factor of the three should indicate the occurrence of AI, and early tests and treatment should be performed in clinical practice.

Funding:This work was supported by a grant from the National Clinical Specialty Construction Project of China (2013-544).

Ethical approval:This work was approved by the local Ethics Committee of the First Hospital of Shanxi Medical University.

Conf licts of interest:The authors state that there is no conf lict of interests involving the study.

Contributors:GLF proposed the study and drafted the manuscript.MMZ collected the data of patients and carried out the statistics analysis.All authors contributed to the design, collection of data, and interpretation of the study and further drafts.