Introduction
Traumatic brain injury (TBI) is the leading cause of mortality and disability among young patients throughout the world. It is a major health and socioeconomic problem [
1,
2]. Hospital-acquired pneumonia (HAP), whose incidence ranges from 40 to 60% for severe traumatic brain-injured patients [
3], is associated with poor neurologic outcome and death [
4]. TBI induces a disturbance of the normally balanced interplay between pro-inflammatory and anti-inflammatory mechanisms leading to a greater susceptibility to infections including HAP [
5].
During the early post-traumatic period, the release of danger-associated molecular pattern by injured cells results in systemic inflammatory response syndrome (SIRS) which is characterized by increased levels of CRP and cytokines [
6‐
8]. Elevated blood concentrations of CRP upon ICU admission are correlated with an increased risk of organ failure and death [
9] and persistent systemic inflammatory response syndrome is predictive of hospital-acquired infection in trauma patients [
10]. To avoid the dramatic consequences of an overwhelming SIRS such as organ failure, an anti-inflammatory response (called "compensatory anti-inflammatory response syndrome" : CARS) is rapidly triggered by the host, including cortisol secretion by adrenal glands after afferent impulses from the site of injury [
11].
It is true that the administration of corticosteroids in a condition that is at risk of secondary infection may seem inappropriate. However, we have shown that low dose of hydrocortisone prevents the occurrence of hospital-acquired pneumonia in multiple trauma patients [
12]. The anti-inflammatory properties of corticosteroids reduce lung inflammation secondary to trauma and enhance the functions of immune cells like dendritic cells and thus could limit secondary bacterial pneumonia [
13]. The administration of corticosteroids in a context of relative immunosuppression shown after any acute condition is counter intuitive but low-dose corticosteroids may enhance immunity. Indeed, low-dose hydrocortisone improves the phagocytic abilities of neutrophils, decreases the blood concentration of anti-inflammatory cytokines such as interleukin-10, and increases the blood concentrations of interferon-γ and interleukin-12, cytokines involved in the host defense against infections [
14,
15].
We aimed to assess the predictive values of the cortisol/CRP ratio (surrogate marker of the CARS/SIRS balance) for the development of HAP in TBI patients. We also evaluated if this ratio may help to select patients who would benefit from corticosteroid therapy to prevent HAP.
Materials and methods
Study design
This study is a sub-study of the Corti-TC trial [
3] (NCT 01093261) a multicentre, randomized, double-blind, placebo-controlled trial of hydrocortisone and fludrocortisone in severe traumatic brain injury. Patients admitted in 19 French ICUs were enrolled in the Corti-TC trial from Sept. 1, 2010, to Nov 29, 2012. Prior to enrollment, written informed consent was obtained from a next-of-kin. Retrospective consent was obtained from patients when it was possible.
Patients
In the Corti-TC trial, inclusion criteria were age between 15 and 65 years, severe traumatic brain injury (Glasgow coma scale score ≤ 8 and trauma-associated lesion on brain CT scan), and enrolment within 24 h of trauma [
16,
17]. Exclusion criteria were as follows: treatment with corticosteroids in the previous 6 months, immunosuppression, pregnancy, tetraplegia, or antibiotic treatment at the time of inclusion. In this sub-study of the Corti-TC trial (
n = 330 patients), we included the 179 patients with available blood samples.
Corticosteroid therapy
For the purpose of the Corti-TC trial, patients received either hydrocortisone (200 mg per day tapered) and fludrocortisone (50 μg tablet once per day) or double placebo for 10 days. Before receiving study drug, adrenal function was assessed with a short corticotropin test. Treatment was stopped if patients had no adrenal insufficiency which was defined as basal blood cortisol concentration of less than 150 μg/L (413 nmol/L) or a maximum increase of less than 90 μg/L (248 nmol/L) in the 60 min after a short corticotropin test.
Endpoints
In the Corti-TC study, the primary outcome was hospital-acquired pneumonia on day 28 of follow-up in patients with or without corticosteroid therapy. In this sub-study, the primary outcome was also the rate of hospital-acquired pneumonia in severe traumatic brain-injured patients.
Blood samples
Blood samples were collected in the first 24 h after trauma, before any administration of the Corti-TC treatment. Sera were frozen − 80 °C upon dosage. Concentration of CRP and plasma cortisol concentrations (free and total) were investigated in sera collected in patients from the Corti-TC study with available samples.
Measurement of CRP, cortisol, and cortisolemia
All biochemical measurements of CRP, Transcortin, and total and free cortisol were performed at the laboratory of Clinical Biochemistry, Nantes University Hospital. The laboratory is licensed according to the ISO 15189 accreditation standard for clinical laboratories.
Serum CRP and total cortisol were determined in single measurement with an electrochemiluminescence immunoassay (C-Reactive Protein Gen.3 and Elecsys Cortisol II respectively) on the Cobas e602-module of the automated cobas®8000 system (Roche Diagnostics, Mannheim, Germany). The lower limits of detection (LoD) of the CRP and cortisol assays were 0.3 mg/L and 0.54 μg/L respectively. Each CRP and cortisol runs were validated by measuring two levels of quality control material prior to starting the experiment.
Samples for serum-free cortisol determination were prepared by equilibrating 500 μL of serum at 37 °C for 15 min in Centrifree 30,000 molecular weight cut-off Centrifugal filters (Merck Millipore, Tullagreen, Ireland) before centrifugation at 1500g for 30 min at 37 °C. Free cortisol was determined in the ultrafiltrate as previously describe for total cortisol.
Serum Transcortin concentrations were measured in duplicate using Human Corticosteroid Binding Globulin ELISA (BioVendor, Brno, Czech Republic) according to the manufacturer’s instructions. Each series of assays (1 ELISA plate) was validated by two internal controls. The lower limit of detection (LoD) was 0.01 μg/mL.
Care of patients with severe TBI
All care provided to patients with severe TBI followed the international guidelines in effect at the time of randomization, including respiratory management, temperature management, stress ulcer prophylaxis, nutrition, fluid therapy, glucose management, intracranial pressure monitoring, and management [
18] as done in our previous studies on the subject [
19,
20].
Hospital-acquired pneumonia (HAP) definition
Pneumonia was suspected as diagnosis when at least two of the following signs: body temperature > 38 °C; leukocytosis > 12,000/mL or leukopenia < 4000/mL; purulent pulmonary secretions; were associated with the appearance of a new infiltrate or changes in an existing infiltrate on the chest X-ray. Diagnosis was confirmed by tests on a respiratory tract sample using a quantitative culture with a predefined positive threshold of 10
4 colony-forming units per milliliter (CFU/mL) for a broncho-alveolar lavage or nonbronchoscopic sample or 10
3CFU/mL for a protected specimen brush. Respiratory samples were always obtained before starting any new antibiotic treatment. HAP was defined as pneumonia that occurred 48 h after admission [
21]. All HAP recorded in the study where early-onset pneumonia (< 7 days).
Data collection
Overall, patient characteristics, including demographics, injury severity score and abbreviated injury score, fluid infusions, vasopressors, antibiotic prophylaxis, CRP rates, plasma cortisol concentrations (free and total), surgery, infections, organ failures, length of ventilatory support, ICU hospitalization, and 28th day mortality, were recorded.
Statistical analysis
Continuous data were described as median [1st–3rd quartiles] and values were compared using the Wilcoxon-Mann-Whitney test. Categorical data were described as N (%) and values were compared using the chi-square test or the Fisher exact test.
The cut-off for distinguishing low versus high (cortisoltotal/CRP) ratio was based on logistic regression between HAP and (cortisoltotal/CRP) ratio. The value corresponding to the largest Youden index was selected as the cut-off. Variables that were associated with HAP at the 0.15 level in univariate analysis were included in a multiple logistic regression. Then, variables that were non-significant at the 0.05 level (Wald test) were removed one by one. Multiple logistic regressions were performed in the global sample, in the placebo group and in the corticosteroid group.
Discussion
In this sub-study of Corti-TC trial [
3], we aimed to establish a correlation between the ratio “inflammatory response”/CARS and the occurrence of HAP in head trauma patients. After a TBI, pro-inflammatory cytokine secretion is a physiologic process which aims to induce damaged tissues healing and anti-bacterial activity by activating both innate and adaptive immunity. In order to balance an excessive pro-inflammatory response, the CNS induces an important anti-inflammatory response leading to increased susceptibility to infections [
5]. This anti-inflammatory response is mediated by the sympathetic nervous system [
22], parasympathetic nervous system [
23], and the hypothalamic-pituitary system via glucocorticoid secretion. In critically ill patients, reduced cortisol breakdown contributes to abnormal blood cortisol levels [
24]. This phenomenon, called Critical Illness-Related Corticosteroid Insufficiency (CIRCI), corresponds to the impairment of the hypothalamic-pituitary axis (HPA) during critical illness resulting from inadequate anti-inflammatory response for the severity of a given patient [
25].
In order to correct the post-traumatic immunosuppression, many therapies have been tested in recent years. They aimed either to limit the initial SIRS (and thus the CARS) in particular by the use of low-dose glucocorticoids [
3,
12] or to restore the secretion of pro-inflammatory cytokines by the use of IFN-γ, GM-CSF [
26], or interleukin 12 [
27]. In a multicenter, randomized, double-blind, placebo-controlled trial, Torres et al. showed that among patients with severe community-acquired pneumonia and high initial inflammatory response (CRP > 150 mg/L), methylprednisolone (0.5 mg/kg/12 h) compared with placebo decreased treatment failure [
28]. In major trauma patients, CIRCI occurs frequently and is associated with uncontrolled inflammatory response, longer vasopressors infusion and poor outcomes [
29]. In a large randomized trial in multiple trauma patients, we found that hydrocortisone therapy prevented the development of hospital-acquired pneumonia by day 28 in patients with CIRCI (defined by a change in baseline cortisol at 60 min of < 9 μg/dl after ACTH (250 μg) administration) [
12]. However, in head trauma patients, we found no interaction between response to corticosteroid therapy and CIRCI status (using the same definition as previously described) [
3]. The actualized recommendations for the diagnosis of CIRCI provide that ACTH stimulation test was not superior to random cortisol for the routine diagnosis of CIRCI [
30]. Moreover, measuring plasma-free cortisol level over plasma total cortisol level was not recommended in patients with suspected CIRCI [
30]. Here, the total and free cortisol blood levels were strongly correlated independently of the inflammatory status of the patient, explaining why we choose to focus on total cortisolemia.
In our study, the correlation between total cortisol and CRP levels could be explained by the early secretion of interleukin-6 (IL-6) following trauma. Indeed, IL-6 is a pro-inflammatory cytokine known to have HPA-activating activity independent of ACTH. Thus, human IL-6 increases plasma concentrations cortisol in mice [
31] and during immune system activation, such as post-traumatic inflammation or septic shock. In head-injured children, serum IL-6 and CRP levels are elevated and correlated to the severity of head trauma [
32] and increased levels of IL-6 in the early phase of severe acute traumatic brain injury is associated with the high inflammatory response such as development of ARDS [
33].
In association with appropriate antibiotherapy, methylprednisolone administration was also associated with a faster reduction in blood IL-6 and CRP levels in the first 24 h of treatment of community-acquired pneumonia [
34]. Before initiation of glucocorticoid therapy, basal cortisol level is positively correlated with IL-6. Corticosteroids reduce the production of IL-6 and the migration of inflammatory cells into the alveolar space leading to avoid an overwhelming inflammatory response. In patients with systemic autoimmune disease, which leads to an inflammatory state, the introduction of glucocorticoid reduced the IL-6 level and contribute to the apparent suppression of the HPA axis [
35]. Among patients with traumatic brain injury, IL-6 is correlated with inflammatory states, high CRP rates, and the occurrence of HAP [
36].
A recent meta-analysis of corticosteroids in pneumonia found that hydrocortisone was not useful in this context, and only prednisone or methyl prednisolone was beneficial [
37]. However, this meta-analysis considered community-acquired pneumonia (CAP) rather than HAP or ventilator-acquired pneumonia (VAP). The micro-organisms involved in each entity are different; in CAP, they are frequently virulent and transmitted by inhaled aerosols. Moreover, respiratory physiology and immunity is severely impaired in patients suffering from HAP or VAP and admitted in ICU. Indeed, mechanical ventilation promotes a specific histological pattern of pneumonia [
38,
39]. Furthermore, comparative analysis of the host response to CAP and to HAP in patients with critical illness has revealed distinct transcriptional and plasma protein responses [
40] showing the functional alterations of the immune response in patients admitted to hospital.
Here, taken separately, pro- and anti-inflammatory biomarkers (CRP and cortisol respectively) failed to predict the development of HAP. However, patients with high cortisol
total/CRP ratio (> 3) have a higher susceptibility to develop HAP and for these patients, the introduction of low dose of corticosteroids is able to reduce this susceptibility. This effect of corticosteroid therapy is not found in patients with a ratio < 3
. There are two means by which low-dose corticosteroids could decrease the rate of secondary pneumonia. Firstly, through an anti-inflammatory effect (patients with a low cortisol/CRP ratio) that decrease the excessive inflammatory response and therefore the compensatory CARS response (immunosuppression). The initiation of corticosteroid therapy may for example reduce IL-6-dependent HPA stimulation, limiting the anti-inflammatory response and thus the prevalence of HAP in highly inflammatory patients. Secondly, corticosteroids may also directly enhance immunity (patients with a high cortisol/CRP ratio). Indeed, low-dose hydrocortisone improves the phagocytic abilities of neutrophils, decreases the blood concentration of anti-inflammatory cytokines such as interleukin-10, and increases the blood concentrations of interferon γ and interleukin-12, cytokines enhancing immunity and involved in the host defense against infections [
14]. Glucocorticoids modulate dendritic cells during and after inflammation [
13] allowing less tissue damage and therefore less sensitivity to bacterial infections. In septic shock or in viral pneumonia, glucocorticoids restore major histocompatibility complex class II expression on myeloid cells, suggesting a better antigen presentation by antigen-presenting cells during treatment [
41,
42]. We and other groups have shown that hydrocortisone enhances immunity in the context of any acute immunosuppressive condition like severe trauma or sepsis [
14,
15]. More specifically, we have demonstrated that trauma-induced immunosuppression is characterized by an interleukin-10-dependent elimination of dendritic cell by natural killer cells and that hydrocortisone improves outcome by limiting this immunosuppressive feedback loop [
15].
The main strength of this ancillary study is the data from a randomized, multicentre, double-blind, controlled trial. This is the first study discriminating head injured patients at risk for developing pneumonia by using an easy-to-use anti- and pro-inflammatory factor ratio in common practice. Another strength of our study is the safety of low-dose corticosteroid use in trauma patients. Indeed, by closely monitoring patients’ natremia and glycemia, there are no serious adverse events recorded in the two large randomized trials we have conducted on the field [
3,
12]. The use of low-dose corticosteroids therefore seems to us to be safe in patients suffering from severe TBI. However, high-dose corticosteroids are not recommended in this context because they provide serious safety issues [
43]. Some limitations must be noted; first, it is impossible to know whether the effect of hydrocortisone is due to the restoration of post-inflammation homeostasis or the correction of the initial CIRCI although the use of a combination of a pro- and anti-inflammatory factor suggests that this effect is due to the correction of the post-inflammation disorder. However, the dosage of other more specific factors, such as IL-6 or IL-10, may help to refine the diagnosis of patients at risk of post-traumatic stress disorder. Second, this ratio could be refined by adding other factors such as the trauma severity score (Glasgow score) or other objective variables (gender, age, medical history), but more patients are needed to implement such a score. Third, to be validated, this ratio must be the subject of a randomized controlled trial comparing management of HAP prevention by corticosteroid therapy in patients at risk (ratio > 3). Finally, our study also suffers from insufficient evidence on secondary outcomes (duration of mechanical ventilation, ICU length of stay…). This is probably due to the size of the samples understudy and this issue could be improved by conducting a specific prospective study to validate the ratio.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.