Corticosteroids and Septic Shock

The use of corticosteroid therapy in patients with sepsis and septic shock has a long and controversial history. The first study suggesting that steroids could prove beneficial in this patient population was published in 1951,¹ and the results of more than 50 clinical investigations exploring this hypothesis have been published since.^{2 - 3} Early reports⁴ suggested that survival was improved; but subsequent large, randomized studies and meta-analyses failed to show a mortality benefit and have even indicated that steroid therapy may be harmful.^{2 - 3 ,5}

Despite this historical burden, several small studies have suggested recently that glucocorticoids might afford clinical benefit to rigorously characterize groups of patients with infection complicated by hypotension unresponsive to fluids and by organ dysfunction.^{6 - 7} In particular, the high incidence of acquired adrenal insufficiency, or lack of adrenal reserve, in such critically ill patients has provided a strong rationale for the prolonged administration of low-dose corticosteroids in this setting.⁸ Importantly, relatively few patients with septic shock have absolute adrenal insufficiency, with serum cortisol levels of 10 µg/dL or less.⁹ However, a lack of adrenal reserve defined by an increase in serum cortisol of 9 µg/dL or less in response to corticotropin administration appears to be common and is associated with increased mortality. In a study of 32 patients with septic shock, 13 had a diminished corticotropin response, all of whom died subsequently.¹⁰ Among the 19 patients with adequate adrenal reserve, only 6 died. In a second similar study, high mortality rates (between 67% and 82%) were identified in patients with inadequate responses to corticotropin.¹¹

In this issue of THE JOURNAL, the multicenter trial from France by Annane and colleagues¹² provides further insight into the incidence of adrenal dysfunction in sepsis and its clinical significance. Of 300 patients with septic shock randomly assigned to receive either hydrocortisone (50 mg intravenously 6 hourly) and fludrocortisone (50-µg tablet daily) or receive placebo for 7 days following a corticotropin stimulation test, 229 (approximately two thirds) were found to have inadequate adrenal reserve. In this sub group, 28-day mortality was significantly less among those who received steroids (53%) compared with placebo (63%). Importantly, no such difference was evident in patients with adequate adrenal reserve. This investigation is particularly important in demonstrating that alterations in adrenal function are common in critical illness and that steroid supplementation can improve survival in a well-defined group of patients.

What were the clinical characteristics and severity of illness of the patients included and how generally applicable are the results? First, the study population had severe septic shock, as evidenced by hypotension unresponsive both to fluid resuscitation and to low doses of the pressor agent, dopamine. Second, all patients required mechanical ventilation and had inadequate urine output, lactic acidosis, or acute lung injury at study entry, indicating significant organ dysfunction. Third, even with corticosteroid replacement, more than half the patients with inadequate adrenal reserve still died, underscoring the need for additional therapies aimed at reversing organ dysfunction and improving outcome in such severely ill patients.

What are the possible explanations for these impressive results? The most compelling appears to be linked to appropriate steroid replacement therapy in patients with inadequate adrenal reserve. The lack of benefit of corticosteroids in patients without adrenal insufficiency strongly supports this mechanism of action. Additionally, corticosteroids have important immunomodulatory effects, including decreasing the transcription of proinflammatory genes, at least in part through modifying the activation of the transcriptional regulatory factor nuclear transcription factor κB.¹³ Such anti-inflammatory properties have been postulated to underlie putative beneficial actions of glucocorticoids in critical illnesses associated with increased inflammatory processes, such as acute lung injury and septic shock.^{14 - 15} In patients with inadequate adrenal reserve, steroid therapy facilitated the withdrawal of vasopressor therapy within 28 days in 57% of the steroid group and 40% of those receiving placebo. The beneficial effects of glucocorticoids on vascular tone, leading to improved organ perfusion, might also, therefore, be relevant.

Do these results indicate that previous work demonstrating an adverse effect of steroid supplements in patients with septic shock was flawed? Probably not; the study by Annane et al¹² showed that of patients without relative adrenal sufficiency, 61% (22/36) patients in the steroid group died compared with 53% (18/34) in the placebo group. Although this difference was not statistically significant, the possibility that steroid therapy may be detrimental in patients with septic shock who maintain adequate adrenal reserve should be considered.

What further work remains to be done? First, additional investigations characterizing more precisely the patient groups likely to benefit from steroid therapy are required, and observational studies identifying the extent to which a lack of adrenal reserve is prevalent among critically ill patients would undoubtedly be helpful. Second, the extent to which the attributable benefit is linked to severity of illness, and the precise dose and composition (glucocorticoid to mineralocorticoid ratio) of any steroid supplement requires clarification. Third, interactions between such steroid supplementation and other interventions shown to improve outcome in septic shock, such as activated protein C¹⁶ ; and with specific supportive measures such as protective ventilation strategies¹⁷ in the (large) subgroup of patients with septic shock who develop acute respiratory distress syndrome, remain to be elucidated. Fourth, tight glucose control with intensive insulin therapy has been shown recently to be associated with improved survival in critically ill patients.¹⁸ The therapeutic use of corticosteroids may make attaining this level of glucose control more difficult.

How should these results change clinical practice? Certainly, the findings of Annane et al suggest that altered hypothalamic-pituitary-adrenal axis function is common in patients with septic shock and that abnormalities in adrenal reserve may occur in other critical illnesses as well. The results also provide a strong rationale to initiate corticosteroid therapy in mechanically ventilated patients with sepsis complicated by hypotension unresponsive to fluids and associated with other organ dysfunction. However, a corticotropin stimulation test, and not simply a random serum cortisol level, should be obtained in such patients before beginning combined glucocorticoid and mineralocorticoid therapy. If serum cortisol levels increase by more than 9 µg/dL in response to corticotropin, corticosteroids should not be initiated or must be discontinued, given the possibility of an adverse effect. Although indications for corticosteroids remain limited in intensive care, the study by Annane et al demonstrates that the use of corticosteroids should be considered for some critically ill septic patients with acquired adrenal function abnormalities.