Professional Documents
Culture Documents
Objectives: Reversible adrenal insufficiency has been fre- Measurements and Main Results: Toll-like receptor 4 were
quently diagnosed in critically ill patients with sepsis who have expressed in adrenal gland and primary fasciculata-reticularis cells.
either low basal cortisol levels or low cortisol responses to Plasma corticosterone response to ACTH was decreased in rats
adrenocorticotrophic hormone (ACTH) stimulation. It is generally receiving preinjection of LPS. LPS pretreatment caused a significant
accepted that a phenomenon called “endotoxin tolerance” con- decrease in corticosterone production in response to subsequent
tributes to immunosuppression during sepsis. The present study ACTH and LPS stimulation in primary fasciculata-reticularis cells.
was to investigate whether endotoxin tolerance occurs in the LPS pretreatment inhibited ACTH- and LPS-induced expression of
adrenal gland, leading to hyporesponsiveness of adrenal gland steroid metabolizing enzymes. LPS significantly decreased toll-like
during sepsis. receptor 4 and ACTH receptor expression.
Design: Controlled laboratory experiment. Conclusions: Pre-exposure to LPS resulted in hyporesponsive-
Setting: University research laboratory. ness to ACTH stimulation in rats. In vitro, LPS pretreatment
Subjects: Sprague-Dawley male rats 200 –250 g, and primary impaired corticosterone production of F/R cells in response to
isolated adrenal fasciculata-reticularis cells. Interventions: Rats ACTH and LPS, which was associated with decreased expression
received intra-arterial injection of purified lipopolysaccharide of synthetic enzymes required for corticosterone production. Our
(LPS, 0.5 mg/kg) through indwelling femoral arterial catheters, results indicate that endotoxin tolerance of adrenal gland is one
and 24 h later the adrenocortical sensitivity to exogenous ACTH of mechanisms for adrenocortical insufficiency during sepsis.
(10 ng/kg) was detected. Primary F/R cells were pretreated with (Crit Care Med 2011; 39:571–574)
LPS at 0.1–100 ng/mL or with ACTH at 0.01–10 ng/mL, and then KEY WORDS: corticosteroids; steroids; infections; sepsis; septic
challenged, in fresh media, with 1 g/mL LPS or 10 ng/mL ACTH. shock; meta-analyses
T he use of corticosteroids for evidence emerging from recent studies patients with septic shock. The contradic-
patients with infections, sep- and meta-analyses. tory findings in these two large multicen-
sis, or septic shock has been tered studies created controversy on the
controversial for decades (1). Overview on Recent Studies appropriateness of giving steroids to any
Despite the many studies performed on and Meta-Analyses septic patient or patient with septic shock
corticosteroids in septic patients, there or in contrast only to those patients with
are still many unresolved issues. These The use of high doses of steroids as septic shock not responsive to aggressive
include whether steroids are beneficial anti-inflammatory agents was common conventional therapy. Because several
for septic patients without shock, for pa- until several studies in patients with se-
new studies evaluating the use of steroids
tients with septic shock, when given early vere sepsis and septic shock demon-
in septic patients have been performed
or late, the optimal dose, given in a bolus strated no survival benefit and a potential
over the last few years, four recent meta-
or continuous infusion, the duration of for superinfections (2– 6). Subsequently,
analyses (12–15) have been published at-
therapy, whether to wean and whether low-dose steroids for patients with septic
shock, used for relative adrenal insuffi- tempting to shed light on the contro-
fludrocortisone should be added. In this versy. The additional studies and meta-
commentary, we provide some guidance ciency, were shown in several small stud-
ies to reverse shock and improve survival analyses should help elucidate which
for physicians on what appears to be rea-
(7–9). Based primarily on the Annane septic patients might benefit from corti-
sonable practice in light of conflicting
study (9), a large randomized, multicen- costeroids. Unfortunately, they also show
tered, controlled trial (RCT), the Surviv- some conflicting results (Table 1).
ing Sepsis Campaign guidelines recom- Meta-analyses are useful to improve
From the Department of Anesthesiology and Crit- the validity of consistent effects observed
ical Care Medicine (CLS, SG, YGW) and the Center for mended the use of hydrocortisone for
Clinical Quality and Safety (MB), Hadassah Hebrew patients with septic shock requiring va- throughout individual studies and to ex-
University Medical Center, Jerusalem, Israel. sopressors despite fluid replacement (10) plore potential reasons for discrepant re-
The authors have not disclosed any potential con- and steroids once again were commonly sults. As shown in Table 1, reversal of
flicts of interest. shock by steroids is consistently observed
For information regarding this article, E-mail: used for patients with septic shock. Re-
charles.sprung@ekmd.huji.ac.il cently, the Sprung et al Corticus study throughout the recent meta-analyses and
Copyright © 2011 by the Society of Critical Care (11), in contrast to the previous Annane RCTs. Meta-analyses may demonstrate an
Medicine and Lippincott Williams & Wilkins study (9), did not demonstrate improved effect not shown in individual studies be-
DOI: 10.1097/CCM.0b013e31820ab1ec survival after hydrocortisone therapy in cause they have more statistical power,
Source: First Author, Year Marik, 2008 (12) Annane, 2009 (13) Sligl, 2009 (14) Minneci, 2009 (15) Annane, 2002 (9) Sprung, 2008 (11)
Type of study Meta-analysis Meta-analysis Meta-analysis Meta-analysis Large RCT Large RCT
Total number of patients 965 2384 1876 2468 300 499
Reversal of shock by steroids Yes Yes Yes Yes Yes Yes
Heterogeneity No Yesa No No NA NA
Reduced mortality by steroids No Yes No No Yes No
Heterogeneity No Yesa No Yesa NA NA
but as shown in Table 1, conclusions on are publication bias and heterogeneity who previously received steroids for var-
the effect of corticosteroids on mortality (16). Negative studies, unpublished be- ious medical disorders and develop septic
are contradictory between meta-analyses cause of lack of interest by authors and shock should receive steroid supplemen-
or RCTs of comparable size. In addition, editors, can completely wipe out an effect tation as a result of their history (10).
in the recent meta-analyses of sepsis apparent in RCTs and meta-analyses (26). Whether to use steroids in any patient
studies (12–15), different methods were Publication bias may be present in the with severe sepsis or septic shock appears
used to include or exclude patients, dif- literature on steroids in septic shock be- to be more controversial.
ferent studies were included in their eval- cause a funnel plot analysis showed a The meta-analyses of Annnane et al
uations, and hence unsurprisingly the disproportionately high number of small (13) and Minneci et al (15) analyzed pa-
authors came up with different conclu- trials with benefit from steroids, suggest- tients with both sepsis and septic shock.
sions. ing underreporting of negative small tri- Although many of the mechanisms in-
als (15). Heterogeneity between studies volved in the disease process may be sim-
Conflicting Evidence Between can also be an important determinant of ilar, the severity of disease, morbidity,
Meta-Analyses and RCTs: A variability of results that allow definition mortality, and response to therapy may
General Problem of a patient subpopulation or treatment be very different for patients with sepsis
regimens more likely to be effective. An and those with septic shock (4, 9, 27–29).
Conflicting results between meta- analysis of heterogeneity in Table 1 sug- Interestingly, these two meta-analyses
analyses and large RCTs are not unusual gested decreased mortality by low-dose used different publications for their anal-
(16), including studies in the recent crit- steroids only in the most severely ill yses and for some issues had different
ical care literature. One meta-analysis patients. conclusions (13, 15). Minneci et al (15)
suggested that noninvasive ventilation It should be remembered that these assessed five studies (30 –34), which An-
lowers mortality in patients with cardiac observations in meta-analyses, retrospec- nane et al did not (13) and Annane et al
failure (17), whereas a subsequent large tive by nature, are exploratory research (13) evaluated one study (35) Minneci et
RCT showed no benefit (18). In head mainly allowing generation of hypotheses al did not (15). Although both analyses
trauma, although meta-analyses were in- to be confirmed or refuted by prospective found differences in mortality between
conclusive, a large RCT concluded that studies. Instability of evidence will often
steroids increase the risk of death (19). In high- and low-dose steroid treatment and
be resolved by larger trials. Uncertainty reversal of shock in steroid-treated pa-
childhood bacterial meningitis, a Co- about the role of steroids in several com-
chrane systematic review (20) showed tients, overall mortality was not different
mon conditions (head trauma [19], men- between patients treated with corticoste-
that steroids reduce severe hearing loss, a ingitis [21], and sepsis) suggests that in-
conclusion refuted by a subsequent RCT roids (both high and low doses) when
stability of evidence may be the result of compared with control patients (13, 15).
(21). Discrepancies may occur even if the the lack of funding for larger trials, espe-
meta-analysis includes studies with con- The lower mortality found in the steroid-
cially because corticosteroids are now ge- treated patients with low-dose steroids
tradictory conclusions. For instance, neric. New medications, with massive
there have been contradictory studies (12–15) included a disproportionately
funding for megatrials by the pharmaceu- high number of published small studies
on the impact of tight glucose control tical companies, appear to achieve faster
with insulin in critically ill patients; a demonstrating beneficial steroid effects
homogeneity of information. Older drugs with a potential publication bias (15).
large RCT showing an increase in mor- may paradoxically leave us with less clear
tality in the intervention group (22) was Septic shock studies arranged in chrono-
evidence in some important areas where logical order found a steroid survival ben-
not confirmed by a meta-analysis that public funding is not sufficient.
included this study (23). Conversely, a efit in the smaller, earlier studies, which
Cochrane systematic review recently was not found in later, larger studies (14).
demonstrated that ultrasound guidance Recommendations for Practice Minneci et al (15) found diversity and
during embryo transfer improves the significance for steroid effect based on
chances of pregnancies (24) despite in- Meanwhile, what is a physician to do? patient severity of disease but Annane et
clusion of a negative large RCT in the The meta-analyses recommend the use of al (13) did not find diversity and only a
meta-analysis (25). steroids for patients with vasopressor- trend for steroid effect based on severity
Major explanations for discrepancies dependent or refractory septic shock (12– of disease. In patients with less severe
between meta-analyses and large RCTs 15). There is also consensus that patients disease, steroids appeared harmful,