Glucocorticoid therapy for trauma - ready for prime time?

developed hospital-acquired pneumonia by day 28 (hazard ratio (HR), 0.51; 95% confi dence interval (CI), 0.30-0.83; P = 0.007). In the modifi ed ITT analysis, 20 of 56 patients (35.7%) in the hydrocortisone group and 31 of 57 patients (54.4%) in the placebo group developed hospital-acquired pneumonia by day 28 (HR, 0.47; 95% CI, 0.25-0.86; P = 0.01). Mechanical ventilation–free days increased with hydrocortisone use by 4 days (95% CI, 2-7; P = 0.001) in the ITT analysis and 6 days (95% CI, 2-11; P < 0.001) in the modifi ed ITT analysis. Hyponatremia was observed in 7 of 76 (9.2%) in the placebo group vs. none in the hydrocortisone group (absolute diff erence, −9%; 95% CI, −16% to −3%; P = 0.01). Four of 76 patients (5.3%) in the placebo group and 6 of 73 (8.2%) in the hydrocortisone group died (absolute diff erence, 3%; 95% CI, −5% to 11%; P = 0.44).


Background
Th e role of stress-dose hydrocortisone in the management of trauma patients is currently unknown.

Methods
Objective: To test the effi cacy of hydrocortisone therapy in trauma patients. Design: Multicenter randomized, double blind, placebocontrolled study. Setting: Seven intensive care units (ICUs) in France during November 2006 to August 2009. Subjects: 150 patients with severe trauma who required ICU stay for at least 48 hours were included in the study. Intervention: Patients were randomly assigned to a continuous intravenous infusion of either hydrocortisone (200 mg/day for 5 days, followed by 100 mg on day 6 and 50 mg on day 7) or placebo. Th e treatment was stopped if patients had an appropriate adrenal response. Outcomes: Hospital-acquired pneumonia within 28 days. Secondary outcomes included the duration of mechanical ventilation, ICU length of stay, hyponatremia, and death.

Conclusions
In intubated trauma patients, the use of an intravenous stress-dose of hydrocortisone, compared with placebo, resulted in a decreased risk of hospital-acquired pneumonia.

Commentary
Severe trauma is a leading cause of death, especially in younger adults. Patients who survive the initial trauma are at high risk for infections, which often increases subsequent morbidity and mortality. Pneumonia is a common infection, occurring in 40-60% of trauma patients. Shock and resuscitation following trauma may lead to an exaggerated immune response [1]. Th e "two hit hypothesis" states that an initial less severely injured patient may eventually develop worsening of multi-organ failure as a result of reactivation of their infl ammatory response due to minor intercurrent event [2], such as an infection. Th us, preventing infection in trauma patients is important.
Th e eff ect of glucocorticoid therapy on susceptibility to infections is controversial. Several lines of evidence suggest that glucocorticoid therapy may reduce risk of infection. For example, higher circulating levels of proinfl ammatory cytokines, such as tumor necrosis factor (TNF) and interleukin (IL)-6, may increase risk of infection in animal models and human studies [3,4]. Gluco corticoids inhibit nuclear factor-kB and trans cription of genes that code for pro-infl ammatory cytokines, includ ing TNF, IL-1, IL-6, IL-8, and interferon (IFN)-γ [5,6], and may therefore prevent infection. Th e acute proinfl ammatory response following trauma is followed by a compensatory anti-infl ammatory response, where immune cells may experi ence tolerance (immuno paralysis). Although immuno paralysis may be important to dampen the deleterious eff ects of the pro-infl ammatory response, patients may be at higher risk for secondary infections during this phase. By reducing the early infl ammatory response, gluco corticoids may reduce the magnitude of the subse quent anti-infl ammatory response, thereby reduc ing risk of secondary infections. However, some studies suggest that glucocorticoid therapy could have deleter ious eff ects on infection risk. For instance, the risk of secondary infections was higher in patients who received gluco corticoid therapy in the CORTICUS [7] trial. No large randomized clinical trial has examined the effi cacy of glucocorticoids to prevent infections and the current study by Roquilly et al. remains one of the largest studies to test this hypothesis.
In this study the author enrolled 150 patients with an injury severity score of 15 and above and included patients with traumatic brain injury (managed at a level 1 trauma center) and randomized them to cortisol and placebo treated arms. Baseline cortisol was measured and patients who were cortisol-defi cient were given steroid treatment for 1 week. Th e author determined occurrence of pneumonia within 28 days of hospitali zation and found that patient who received steroid treatment had a signifi cantly lower incidence of pneumonia compared to placebo (HR, 0.51; P = 0.007).
Th e results of this study are in contrast with the results of the CRASH [8] study. Th e CRASH study was a large multicenter randomized clinical trial conducted in 10,000 patients with head injury, as evidenced by Glasgow Coma Score of 14 or less, to determine whether high dose glucocorticoids improve outcomes. However, there were important diff erences between the HYPOLYTE and CRASH studies (Table 1). In contrast to CRASH, which focused on traumatic head injury patients alone, HYPOLYTE enrolled a broad population of patients with trauma and included a smaller subset (44.96%) with head injury. Both dose and duration of glucocorticoid therapy were diff erent. CRASH used high dose steroids, which may lead to immunosuppression. HYPOLYTE used lowmoderate dose glucocorticoids, which are less likely to cause immune suppression and may be important in patients who have critical illness-related corticosteroid insuffi ciency (CIRCI). Th e CRASH study treated all patients with glucocorticoid therapy, whereas the HYPOLYTE study discontinued therapy in those who did not meet criteria for relative adrenal insuffi ciency. Duration of therapy also diff ered between these studies and shorter course of steroids can be associated with rebound infl ammation. Finally, the primary endpoint in these studies was also diff erent. For example, the primary endpoint was all-cause mortality at 2 weeks in the CRASH study, whereas the HYPOLYTE study examined eff ect of steroids on hospital-acquired pneumonia. Th e CRASH study was stopped early due to higher mortality in the steroid treated arm, whereas the current study showed no diff erence in mortality between the two treat ment groups.
Th is study by Roquilly et al. has several limitations. First, cortisol levels drawn prior to administering steroids may be infl uenced by the prior use of etomidate for intubation, (63% of patients in this study received etomidate prior to intubation), which can cause adrenal suppression for approximately 24 hours [9,10]. Th us, the high rate of adrenal insuffi ciency may be infl uenced by etomidate administration. Second, it is unclear whether steroid therapy should be targeted based on presence or absence of relative adrenal insuffi ciency. Although the corticotropin stimulation test is recommended to diagnose CIRCI, results of the CORTICUS trial showed no diff erence in the effi cacy of glucocorticoid therapy among those with and without relative adrenal insufficiency. Th ird, the investigators did not assess the longterm eff ects of steroid use, such as myopathy and critical illness neuropathy, a potential complication of glucocorticoid therapy. Finally, the study did not determine mechanisms of benefi cial eff ects of steroids. For example, whether steroids reduced risk of infection by reducing pro-infl ammatory markers or improved immune cell function is not known.

Recommendation
Although results of this study are intriguing, glucocorticoids cannot be recommended for routine use in trauma patients. Th ese results suggest a nee d to con duct a larger trial to understand whether a reduction in infection observed in this study translates into improved mortality.