Stress-Dose Corticosteroid Therapy for Sepsis
from Pharmacotherapy
Clinical Studies
Historical Perspective
The association of inflammation with sepsis has long been established. Corticosteroids were first administered to treat sepsis as early as 1940.[89] Until a decade ago, common practice was to administer short courses (< 24 hrs) of extremely high doses (supraphysiologic) of corticosteroids in an effort to halt the systemic inflammatory response syndrome. Results of two meta-analyses that included several randomized, double-blind studies conducted from 1966-1993 showed no survival benefit when supraphysiologic doses of corticosteroids were administered for short-term treatment of sepsis or ARDS.[90, 91] Moreover, higher infection rates were associated with corticosteroids. The authors of these two meta-analyses recommended that high-dose corticosteroids be abandoned for the treatment of sepsis and ARDS.[90, 91] Of interest, several of these studies found that despite no long-term survival benefit, therapy with corticosteroids was associated with early shock reversal or blood pressure elevation.[92-95] With the recognition of adrenal insufficiency as a physiologic component of sepsis and ALI-ARDS, the focus of recent research has involved adrenocortical replacement dosage regimens administered for several days to weeks, with doses corresponding to the stress level of the disease state (physiologic or stress dose).Sepsis
The administration of stress-dose regimens of exogenous corticosteroids should reduce the systemic inflammatory process associated with the host defense response to sepsis. Results of a randomized, placebo-controlled study that enrolled 24 patients with hyperdynamic septic shock who were vasopressor dependent demonstrated that serum concentrations of the proinflammatory cytokines IL-6 and IL-8 were significantly reduced within 24 hours of starting stress-dose hydrocortisone.[96] Survivors had the greatest reductions of either cytokine. Compared with placebo, IL-6 concentrations were significantly lower after 5 days of hydrocortisone, but IL-8 concentrations were not statistically significantly different. In this study, hydrocortisone was administered as a 100-mg bolus, followed by 0.18 mg/kg/hour for at least 6 days, and gradually discontinued in daily 24-mg increments once the infection was treated successfully. The investigators did not define successful treatment. The inciting microbes and the specific antimicrobials that were administered were not reported, but most infections were polymicrobial (54%) from abdominal sources (46%), nosocomial pneumonia (29%), or community acquired pneumonia (25%). A cohort study by the same authors showed that serum concentrations of the late inflammatory mediators phospholipase A2 and neutrophil elastase were reduced within 48 hours of starting stress-dose hydrocortisone.[97] Hydrocortisone was administered as a 100-mg bolus followed by 10 mg/hour, and tapered by the attending physician after shock reversal. The investigators did not define shock reversal. Of note, the serum concentration of the antiinflammatory cytokine IL-10 was not altered by stress-dose hydrocortisone. Elevated heart rate and body temperature, two manifestations of systemic inflammatory response syndrome, and serum cortisol concentration returned to normal during hydrocortisone administration but trended upward after hydrocortisone was discontinued. This observational study was designed so that the results from one center that administered hydrocortisone were compared with those from several control centers that did not administer hydrocortisone. Consequently, only 12 (21%) of 57 patients with severe sepsis or septic shock received hydrocortisone. The study protocol did not influence therapy, and the investigators did not specify other treatment differences between centers or possible population differences, so consistency of therapy among centers is questionable. No information was provided regarding infection site, inciting microbe, or antimicrobial therapy. These two studies provide evidence that stress-dose administration of hydrocortisone reduces the systemic inflammatory response associated with severe sepsis and septic shock. However, the data are limited by poor study design and small patient populations that were not well described.
Patients with adrenal insufficiency display a rightward shift in the dose-response curve to norepinephrine.[98] Moreover, vasopressor response to norepinephrine is correlated positively with the degree of corticotropin-induced cortisol production.[98, 99] Through upregulation of adrenergic receptors, stress-dose corticosteroids should enhance vasopressor activity. Phenylephrine responsiveness was restored 1 hour after the intravenous administration of a single dose of hydrocortisone 50 mg in 12 patients with septic shock who had impaired adrenocortical function.[100] A similar response was obtained in 12 age-matched control subjects. Numerous case series have shown rapid shock reversal and/or discontinuation of vasopressor therapy after starting stress-dose hydrocortisone in patients with refractory vasopressor-dependent septic shock (Table 1).[63, 69, 101-108] Of note, three of these reports found no difference in adrenocortical function between patients who responded to hydrocortisone and those who did not.[105-107] One report showed that despite similar overall hemodynamic response rates for shock reversal, patients with relative adrenal insufficiency were able to discontinue vasopressor therapy significantly faster after hydrocortisone was begun than were patients with normal adrenocortical function (median of 3 days vs 5 days, respectively, p<0.01).[108] However, mortality rates were similar (45% vs 44%, respectively). Although limited, these reports provide initial evidence that the administration of stress-dose corticosteroids may sensitize patients with septic shock to the hemodynamic effects of vasopressor therapy. Whether the presence of adrenal insufficiency enhances response is questionable, as overall hemodynamic response rates and mortality rates were similar regardless of adrenal function.
Several placebo-controlled studies evaluating stress-dose corticosteroids are described in this article according to mortality results. Results of one case-control study[97] and three randomized, double-blind studies[96, 109, 110] found that hydrocortisone therapy did not alter mortality in patients with severe sepsis or vasopressor dependency (Table 2). The case-control study[97] already was described in this review. Although mortality rates were not statistically significantly different, only 3 (25%) of the 12 patients treated with hydrocortisone died while in the hospital compared with 25 (56%) of the 45 control patients. Of note, all 12 patients who received hydrocortisone were in shock at baseline compared with only 27 (60%) of the 45 control patients. Therefore, patients receiving corticosteroids were sicker than the control group, but mortality rate was half that of the control group. One of the randomized, double-blind studies[109] was published in 1963, so the applicability of the methodology and results to modern day medicine are questionable. In this study, 329 patients with life-threatening infection were randomly assigned to placebo or hydrocortisone 300 mg, followed by a daily taper of 50 mg for a total of 6 days of treatment. Limited patient demographic data were provided, but 25.8% of subjects had meningitis, 22.4% had pneumonia, and 12.8% had bacteremia. Antibiotic and vasopressor therapies were not altered by the study protocol. Overall mortality rates were 34.7% in the hydrocortisone group and 22.6% in the placebo group.
Another previously described study[96] found that despite no differences in mortality rates between the hydrocortisone and placebo groups, median daily scores of sepsis-related organ failure assessment decreased significantly after 3 days of hydrocortisone (14 on day 0 to 11 on day 3, p<0.05) but remained constant in the placebo group.[96] Mean baseline acute physiology and chronic health evaluation (APACHE) II scores were 25 ± 1 for the hydrocortisone group and 27 ± 1 for the control group. Therefore, organ function improved only in the group of patients who received hydrocortisone, despite similar baseline levels of illness severity. The investigators did not report duration of ICU stays; therefore, whether accelerated reversal of organ dysfunction resulted in shorter ICU stays is not known.
A third randomized, double-blind study[110] of 40 patients with hyperdynamic septic shock showed no mortality difference and similar overall hemodynamic response rates despite faster vasopressor therapy discontinuation in the group receiving hydrocortisone compared with that in the placebo group (median 2 vs 7 days, respectively, p=0.005). Respiratory function tended to improve quicker with hydrocortisone administration (p=0.09), resulting in substantially fewer days of mechanical ventilation (median 18 vs 38 days, p=0.19) and shorter ICU stays (median 27 vs 44 days, p=0.27). Hydrocortisone was administered as a 100-mg bolus, followed by 0.18 mg/kg/hour for as long as vasopressor therapy was required, then 0.08 mg/kg/hour for 6 days, and gradually discontinued in daily 24-mg increments. Patient demographics were similar, with mean APACHE II scores of 26 ± 1 and 27 ± 1 for the hydrocortisone group and placebo group, respectively. Hemodynamic parameters were similar at the time of enrollment, with the mean duration of shock before recruitment being 20.7 ± 3.67 and 26.5 ± 4.39 hours for the hydrocortisone group and placebo group, respectively. Thirty patients (75%) had ARDS at the time of enrollment. The most frequent infections were abdominal in origin (40%), nosocomial pneumonia (32.5%), and community acquired pneumonia (17.5%), and the most common microbes were gram positive (52.5%), polymicrobial (27.5%), and gram negative (15%). This study was conducted at a single center in Germany. Many of their routine ICU practices are not standard in North America, including selective decontamination of the gastrointestinal tract, antithrombin and immunoglobulin therapies for severe sepsis, and dopexamine or dobutamine therapy to enhance cardiac output. Of interest, a retrospective analysis of this study found hydrocortisone therapy, compared with standard therapy, reduced the frequency of posttraumatic stress disorder in patients surviving sepsis (18.5% vs 59.3%, respectively; p=0.01).[115]
A fourth randomized, double-blind study[111] of patients with septic shock who required vasopressor therapy for more than 72 hours did not report mortality data. However, similar to the previous study, the results of this study found that hydrocortisone therapy was associated with accelerated discontinuation of vasopressor therapy.[111] Sixteen (70%) of 23 patients receiving hydrocortisone were vasopressor independent on day 3 of therapy compared with only 7 (33%) of 21 patients receiving placebo (p=0.001). The mean duration of vasopressor therapy was 72 hours in the hydrocortisone group compared with 122 hours in the placebo group (p<0.005). Since this study has been reported only as an abstract, it is difficult to ascertain the clinical relevance of these results.
In all cases, the studies that reported no mortality benefit with stress-dose hydrocortisone found accelerated resolution of shock or vasopressor discontinuation. Patients were not stratified according to adrenal function in the above-mentioned studies.
Three randomized, double-blind, placebo-controlled studies found that mortality was significantly reduced with stress-dose corticosteroid therapy (Table 2).[112-114] The applicability of the results of one of these studies is questionable as dexamethasone therapy was investigated for severe typhoid infection, an extremely rare disease.[112] In this study, 38 patients with typhoid-induced mental status changes were randomly assigned to placebo or dexamethasone 3 mg/kg followed by 1 mg/kg every 6 hours for eight doses. Limited patient demographic data were provided, but 6 (30%) of 20 dexamethasone-treated patients had shock before enrollment compared with 5 (28%) of 18 patients receiving placebo. All patients had the organism isolated from bone marrow, and 80% had bacteremia. Chloramphenicol was provided to all patients. Overall mortality rates were 10% in the dexamethasone group and 56% in the placebo group (p=0.003). Adrenal function was not assessed in this study but was assessed in the two remaining studies.[113, 114]
The largest randomized study of stress-dose corticosteroid therapy is published only as an abstract.[113] In this study, 299 patients with septic shock were enrolled, including 229 (76.6%) with adrenal insufficiency based on an increase of serum cortisol of less than 9 µg/dl after corticotropin 0.25 mg. All patients required vasopressor therapy for less than 8 hours before enrollment. Hydrocortisone was administered as 50 mg every 5 hours for 7 days. Concurrently, fludrocortisone 50 µg/day was administered enterally to prevent hydrocortisone-induced hyperkalemia. The relative risk reduction in mortality with hydrocortisone therapy compared with placebo was 0.67 (95% confidence interval [CI] 0.474-0.946) in patients with adrenal insufficiency. The relative risk reduction was similar in patients without adrenal insufficiency but was not statistically significant (0.696, 95% CI 0.349-1.387). Whether the statistical insignificance is due to too few patients to show a difference or to a lack of clinical benefit in patients with normal adrenal function is indeterminate.
In another study, 41 patients with septic shock who required more than 48 hours of vasopressor therapy were randomly assigned to placebo or hydrocortisone 100 mg every 8 hours for 5 days and tapered by 50% increments every 3 days over 6 days.[114] Only 12 (29%) of the 41 patients with shock had adrenal insufficiency, defined as an increase of serum cortisol concentration of less than 6 µg/dl after corticotropin 0.25 mg. However, only 4 (19%) of 21 patients receiving hydrocortisone had adrenal insufficiency compared with 8 (42%) of 19 patients receiving placebo (p=0.093). Demographic data were similar between the groups, with 63% of patients being enrolled in the surgical ICU and 32% of patients experiencing at least two organ failures. The most frequent infections were abdominal (39%) and pulmonary (37%) in origin, and the most common microbes were gram negative (44%), gram positive (17%), and polymicrobial (12%). Mean elapsed time from the onset of shock to administration of the first dose was 6 ± 5 days and 5 ± 3 days for the hydrocortisone group and placebo group, respectively. Hemodynamic response rates and mortality rates within groups were similar between patients with and those without adrenocortical dysfunction but consistently favored hydrocortisone therapy. Seven-day shock reversal, defined as a stable systolic blood pressure greater than 90 mm Hg for at least 24 hours without volume expansion or vasopressor therapy, was achieved in 15 (68%) hydrocortisone-treated patients compared with 4 (21%) control patients (p=0.007). In the hydrocortisone group, hemodynamic response rates were 75% and 67% for patients with and those without adrenal insufficiency, respectively. However, rates of shock reversal for the two placebo subgroups were 25% and 18%, respectively (p<0.05 vs subgroups receiving hydrocortisone). Mortality rates at day 28 were 32% in the hydrocortisone group compared with 63% in the placebo group (p=0.045). In summary, stress-dose therapy with hydrocortisone in patients with septic shock who require vasopressor treatment is associated with accelerated hemodynamic stability and reduced mortality. When to begin stress-dose corticosteroids and whether patients with and those without adrenal insufficiency respond differently require further investigation.
In general, patients were excluded from studies if they were younger than 18 years; were immunocompromised; received corticosteroids before developing ALI-ARDS; or had active pancreatitis, terminal illness, hemorrhagic shock, recent gastrointestinal hemorrhage, or extensive burns. None of the case series reported adverse events, but a few adverse events were attributable to corticosteroid therapy in the comparative studies. The combined rates of occurrence of gastrointestinal hemorrhage, reported in four studies, were 5.3% (12 of 227) for corticosteroid therapy and 2.3% (5 of 216) for placebo.[109, 110, 112, 114] The combined rates of occurrence of secondary infections, reported in four studies, were 15.1% (35 of 232) for corticosteroid therapy and 13.9% (30 of 216) for placebo.[109, 110, 112, 114] One study reported no laboratory aberrations,[114] whereas another reported statistically significant hypernatremia during the first 7 days of hydrocortisone therapy.[110] Serum sodium concentrations exceeded 155 mEq/L in 6 (30%) of 20 patients receiving hydrocortisone therapy compared with only 1 (5%) of 20 patients receiving placebo. Sodium concentrations returned to normal values after hydrocortisone was discontinued, as per protocol. The same study found serum alanine aminotransferase concentrations were statistically significantly elevated from baseline after 14 days of hydrocortisone therapy. Serum glucose and urea nitrogen concentrations tended to be higher with hydrocortisone therapy. What clinical significance the biochemical aberrations have is indeterminate, but one must question how blinding was maintained.ALI or ARDS
Similar to sepsis, the administration of stress-dose regimens of exogenous corticosteroids should reduce the systemic inflammatory process associated with ALI-ARDS. The administration of methylprednisolone 200 mg followed by 2-3 mg/kg/day for a mean of 15 ± 9 days after the onset of ARDS significantly reduced the serum concentrations of the proinflammatory cytokines IL-1 and IL-6 by days 5 and 7 of therapy, respectively.[116] Similar to sepsis, survivors had the greatest reduction. Alveolar concentrations of IL-1, IL-6, and TNF- were significantly reduced by day 14 of methylprednisolone therapy. The median APACHE II score was 23, and all nine patients in this cohort study had sepsis-induced ALI-ARDS. Serum and alveolar concentrations of collagen markers were reduced during 7 days of stress-dose methylprednisolone.[117] Similar to sepsis, serum and alveolar concentrations of the antiinflammatory cytokine IL-4 were not altered by stress-dose methylprednisolone.
The combined mortality rate in five case series of patients with ARDS treated with stress-dose methylprednisolone is 20.6% (14 of 68 patients), considerably lower than the 34-60% reported for standard therapy for ARDS (Table 3).[118-122] Similarly, two of three case-control analyses showed substantial, although statistically insignificant, reductions in mortality with corticosteroid treatment (Table 4).[123-125] The first was a retrospective analysis of 31 cases of ALI-ARDS, 29 with ARDS, in a medical or surgical ICU.[123] Sixteen of these patients received methylprednisolone 120 mg/day until improvement of unresolving ALI. The investigators did not define improvement, but methylprednisolone was started after a mean of 9.7 ± 0.7 days in the ICU. Demographic data and other therapeutic interventions were similar. Mean admission APACHE II scores were 14.4 ± 1.6 and 14.8 ± 1.6 for the methylprednisolone group and control group, respectively. The cause of ALI-ARDS was pneumonia for all patients, with pneumococcus representing 58% of the microbes. Oxygenation, represented by the PaO2:FiO2 ratio, improved significantly after 3 days of methylprednisolone but did not change in the control group (60 ± 12.9 vs -6.0 ± 7.6, respectively, p=0.004). However, duration of mechanical ventilation (20.5 ± 2.3 vs 20.1 ± 2.7 days, respectively), length of ICU stay (20.9 ± 2.2 vs 20.8 ± 2.8 days, respectively), and 30-day mortality rates (19% vs 20%, respectively) did not differ between groups. The second case-control analysis was also retrospective in design and involved 31 patients with infection-induced ARDS (17 due to sepsis and 14 due to pneumonia).[124] The median duration of ARDS was 15 days. Thirteen patients received methylprednisolone 100-250 mg/day for 1-3 days, which was tapered over 5-44 days. Demographic data were similar, with median APACHE II scores of 17 and 19 for the methylprednisolone group and control group, respectively. Oxygenation did not change significantly after 48 hours of methylprednisolone but decreased in the control group (median change of 5 vs -26, respectively, p=0.039). After 7 days, oxygenation improved by 23% in the methylprednisolone group but decreased by 11.1% in the control group. Lung injury score did not change after 7 days of methylprednisolone therapy but increased 11.1% in the control group. Of greatest interest is that only 5 (38%) methylprednisolone-treated patients died in the ICU compared with 12 (67%) control patients (p=0.117).
The third case-control analysis retrospectively evaluated 18 patients with life-threatening varicella pneumonia from 1980-1996.[125] Six patients received hydrocortisone 200 mg every 6 hours for 48 hours, the majority during more recent years of the evaluation. Demographic data were similar between groups, except the baseline PaO2:FiO2 ratio was significantly lower in the hydrocortisone group (median 86.5 vs 129.5, p=0.045). Follow-up oxygenation parameters were not reported, so the effect of hydrocortisone on oxygenation is indeterminate. The group receiving hydrocortisone had a significantly shorter ICU stay (median 5.5 vs 12 days, p=0.008) and hospital stay (median 10 vs 20 days, p<0.006). Mortality rates were not significantly different, but no patient receiving hydrocortisone died compared with four patients (33%) in the control group. Unfortunately, it is difficult to attribute clinical differences between groups to corticosteroid therapy because patient data collection occurred over 16 years, with most patients who received hydrocortisone being evaluated recently. Therefore, differences between groups may be due to changes in other therapies or diagnostic approaches that occurred over the time period.
After enrollment of 24 patients, the only randomized, double-blind study was terminated by an external safety review board because methylprednisolone therapy for ARDS was associated with significantly lower ICU and hospital mortality rates compared with placebo.[126] Intravenous methylprednisolone was started on day 7 of ARDS as a 2-mg/kg bolus, then 0.5 mg/kg/day every 6 hours for 2 weeks or extubation, then tapered over 3 weeks. Demographic data were similar between groups, with mean APACHE III scores of 58 ± 14 and 55 ± 16 for the methylprednisolone group and placebo group, receptively. Twelve patients (50%) had undergone surgery within 2 days of enrollment, and 17 (71%) had sepsis-induced ARDS with 13 (54%) experiencing shock. The ICU mortality rates were 0% and 62% for the methylprednisolone group and placebo group, receptively (p=0.002). Two (12%) of 16 methylprednisolone-treated patients died during their hospital stay compared with 5 (62%) of 8 patients receiving placebo (p=0.03). Neither death in the methylprednisolone group was from unresolving ARDS, whereas all five deaths in the placebo group were attributed to ARDS. Oxygenation improved 34.8%, and lung injury score decreased 33.3% by day 5 of methylprednisolone therapy compared with no change in either parameter for the placebo group. As a result, the median duration of ventilation was 11.5 days with methylprednisolone therapy compared with 23 days in the placebo group (p=0.001). Of note, serum concentrations of the proinflammatory cytokines IL-1, IL-6, and TNF- were significantly reduced after 3 days of methylprednisolone therapy and 5 days of placebo.[78] Although adrenocortical function was not assessed, serum corticotropin and cortisol concentrations were reduced significantly after 3 days of methylprednisolone therapy but did not change with placebo, suggesting that exogenous administration of methylprednisolone resets the HPA axis.
These results are encouraging; however, there are several concerns with this study. Owing to early study termination, only 24 (24%) of 99 patients were enrolled. Of the eight patients receiving placebo, four eventually received corticosteroids because the protocol provided a blind crossover to the other treatment arm if a patient did not improve after 10 days. The methodology included weekly bronchoscopies for surveillance of new pulmonary infections, a procedure that likely would not be conducted at most institutions. When patient characteristics are compared at onset of ARDS to study entry (day 7 of ARDS), the group that received methylprednisolone appeared to be improving to a greater extent than the placebo group, even before treatment was started. For example, the PaO2:FiO2 ratio increased 46.4% in the methylprednisolone group versus only 14.6% in the placebo group during the 7 days before starting the respective treatment arms.
Definitive recommendations cannot be made regarding corticosteroid therapy for ALI-ARDS. In most of the studies, the patients had infection-induced ALI-ARDS. A course of stress-dose methylprednisolone should be considered only in cases of refractory sepsis-induced ARDS when impending mortality is likely. A multicenter study is under way, with enrollment of 125 of 180 subjects, and should provide definitive information.[127]
In general, patients were excluded from ALI-ARDS studies if they were younger than 18 years; were immunocompromised; had received corticosteroids before developing ALI-ARDS; or had a terminal illness, recent gastrointestinal hemorrhage, fungal infection, uncontrolled diabetes mellitus, or extensive burns. Three of the five case series reported adverse events.[120-122] The combined rates of occurrence of gastrointestinal hemorrhage and secondary infections were 10% (4 of 42) and 28% (12 of 42), respectively. Two of the three case-control studies reported adverse events.[123, 124] No cases of gastrointestinal hemorrhage were reported in 29 patients receiving methylprednisolone and 33 patients receiving placebo. The combined rates of occurrence of secondary infections were 31% (9 of 29) in the corticosteroid group and 15% (5 of 33) in the placebo group. Similarly, the infection rate/day of mechanical ventilation was nearly double for methylprednisolone therapy compared with placebo in the randomized, double blind study (relative risk 1.8, 95% CI 0.86-3.76).[126] Six (38%) of 16 patients treated with methylprednisolone developed new ventilator-associated pneumonia compared with only 1 (12%) of 8 patients receiving placebo. Unfortunately, early termination of the study may have prevented sufficient enrollment to detect a statistically significant difference. The occurrence of hyperglycemia was similar between groups.
This is a part of article Stress-Dose Corticosteroid Therapy for Sepsis Taken from "Generic Chloromycetin (Chloramphenicol) Information" Information Blog
A Lincolnshire woman who lost her newborn twins to a bacterial infection is hoping to raise awareness of the condition which killed them.
