Keywords Fludrocortisone - glucocorticosteroids - hydrocortisone - mineralocorticoids - septic
shock - steroids
INTRODUCTION
Septic shock is characterized by a dysregulated host response to infection complicated
by circulatory or cellular dysfunction[1 ] and high short-term mortality.[2 ] In addition to other measures, corticosteroids are suggested for septic shock management.[1 ] Despite sound physiologic plausibility, extensive investigation into the efficacy
of exogenous steroids have yielded variable outcomes.[3 ]
Although most prior investigations have examined the impact of glucocorticosteroid
supplementation alone, it remains unclear whether concomitant mineralocorticoid administration
may be of additional benefit, as indicated by recent studies.[4 ],[5 ],[6 ]
We performed a systematic review and meta-analysis of the literature with the aim
of further elucidating this matter.
MATERIALS AND METHODS
This systematic review and meta-analysis was performed according to the Preferred
Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). MEDLINE, Scopus,
and Cochrane databases were reviewed for potentially eligible randomized controlled
trials (RCTs) and observational studies comparing the administration of glucocorticosteroids
at a dose of 50–300mg/day of hydrocortisone or equivalent in combination to fludrocortisone
(steroid combination) as an adjunct treatment in septic shock versus placebo or no
corticosteroids.
Study selection and data extraction
The algorithm used for the MEDLINE, Scopus, and Cochrane databases was as following:
“(sepsis OR septic shock OR septicemia) AND (corticosteroids OR steroids)” with the
necessary adjustments. The search was limited to human studies and adult population
(>18 years). No restrictions on language, publication date, or publication status
were present. Two investigators (LB and DF) independently reviewed all retrieved references
based on study title and abstract. Full texts were reviewed for all possibly relevant
studies, and inclusion criteria were applied. A third independent investigator (LP)
was involved as needed to reach consensus. To identify further eligible studies, manual
searches of the references list of the included studies and pertinent reviews were
performed.
Two independent reviewers (LB and DF) extracted data from the included studies using
a predefined data collection form. Discrepancies were resolved with the involvement
of a third reviewer (LP). Data for the following baseline variables were extracted:
study name, first author, year of publication, duration of trial, population enrolled,
number of participants enrolled, inclusion and exclusion criteria, definition of septic
shock, mean age of participants, gender distribution of participants, severity of
septic shock based on scoring tools available from studies (acute physiology, age,
chronic health evaluation, Sequential Organ Failure Assessment [SOFA]), performance
of a cosyntropin test, type, dose, route of administration, and duration of corticosteroids
administered.
Outcomes
The primary efficacy end point was the all-cause mortality within 30 days. Mortality
measured between 25 and 30 days following diagnosis of septic shock was considered
equivalent. Secondary end points included intermediate-term mortality (31 days to
six months following diagnosis of septic shock), long-term mortality (beyond six months
from the diagnosis of septic shock), intensive care unit (ICU) mortality, in-hospital
mortality, shock reversal within 30 days, vasopressor-free days, ventilator-free days,
duration of ICU admission, and incidence of serious adverse effects including hyperglycemia,
gastrointestinal (GI) bleeding, delirium, and secondary infection.
Risk of bias assessment
Two independent reviewers (PAB and LB) assessed the risk of bias of the included studied
using the Cochrane tool for randomized studies and the Robins-I tool for non-randomized
studies.
Data synthesis and statistical analysis
Definitions of the included outcomes were used as defined in the original studies.
A random effects model was selected a priori because the included studies had heterogeneous
study design and baseline patients’ characteristics.[7 ] Forest plots were used to illustrate the individual study findings and the random
effects meta-analysis results. The I-square statistic (I
2 ) was used to assess for heterogeneity among the studies. Dichotomous outcomes were
calculated as odds ratios (ORs) with 95% confidence intervals (CIs) for the primary
outcome and the secondary outcomes. For continuous outcomes, we calculated the mean
differences with 95% CIs. A 95% CI not containing 1 for OR or a P value < 0.05 was considered as statistically significant. Statistical analysis was
conducted with R (version 3.4.3) with RStudio (version 1.1.447 RStudio Inc., Boston,
Massachusetts).
RESULTS
Studies selection and characteristics
In total, 1215 records were screened and 18 full-text articles were assessed for eligibility.
Of these articles, only four studies, two RCTs, and two observational studies met
the inclusion criteria and were included in the qualitative and quantitative analysis.[6 ],[8 ],[9 ],[10 ] The study selection process is presented with a PRISMA flow diagram [Figure 1 ]. The characteristics of the studies are summarized in [Table 1 ].
Figure 1: The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)
flowchart
Table 1
Characteristics of the included studies
Study (year)
Type of study
N
Shock definition
Shock reversal
Treatment duration
Glucocorticosteroid/ dose
Mineralocorticoid/dose
a = (1) Systolic arterial pressure lower than 90 mm Hg for at least 1 h despite adequate
fluid replacement and more than 5 pg/Icg of body weight of dopamine or current treatment
with epinephrine or norepinephrine, (2) urinary output of less than 0.5 mL/Icg of
body weight for at least 1 h or ratio of arterial oxygen tension to the fraction of
inspired oxygen (PaO2/FiO2) of less than 280 mm Hg, (3) arterial lactate levels higher
than 2 mmol/L, (4) need for mechanical ventilation, and (5) randomization within 3
h b = (1) Sequential Organ Failure Assessment (SOFA) score of 3 or 4 for at least
two organs and at least 6 h, (2) receipt of vasopressor therapy (norepinephrine, epinephrine,
or any other vasopressor at a dose of ≥0.25 μg/kg of body weight per minute or ≥1
mg/h) for at least 6 h to maintain a systolic blood pressure of at least 90 mm Hg
or a mean blood pressure of at least 65 mm Hg, (3) randomization within 24 h of septic
shock onset c = (1) systolic blood pressure (SBP) not more than 90 mm Hg or mean arterial
pressure not more than 70 mm Hg within 1 h before the start of arginine vasopressin
infusion, (2) positive fluid balance, (3) mechanical ventilation, (4) at least two
systemic inflammatory response syndrome (SIRS) criteria (one criterion in addition
to mechanical ventilation), and (5) positive result in microbial culture or strong
clinical suspicion of infection with the initiation of antimicrobials d = No shock
definition but presence of one or more acute organ dysfunctions *Not reported, **We
only included patients to whom vasopressors were administered
Annane et al .[8 ] (2002)
RCT
299
a
Pressor withdrawal
7 days
Hydrocortisone/Bolus Intravenous, 50 mg q6 h
Fludrocortisone/Bolus Per os, 50 pg every 24 hours
Annane et al .[6 ] (2018)
RCT
1241
b
Pressor withdrawal
7 days
Hydrocortisone/Bolus IV, 50 mg every 6 hours
Fludrocortisone/Bolus PO, 50 pg every 24 hours
Bauer et al .[9 ] (2008)
Retrospective,case-control
42
c
Cessation of vasopressors > 6 h
5 days minimum
Hydrocortisone/Bolus IV, 50 mg every 6 hours
Fludrocortisone/Bolus PO, 50 pg every 24 hours
Beale et al .[10 ] (2010)**
Retrospective analysis of PRPGRESS database
8968
D
Not defined
Not defined
Equivalent or lesser potency to hydrocortisone 50 mg/6 hourly
9-alpha fludrocortisone/Bolus PO, 50 pg every 24 hours
In total, 10,550 patients were included. The greatest number of those (6747) came
from one of the observational studies.[10 ] The percentage of male patients was 68%; however, the study that contributed most
of the patients did not report their sex.[10 ] Details on baseline patient characteristics are presented in [Table 2 ]. The included studies were deemed to present low risk of bias. The detailed assessment
of risk of bias is presented in detail in [Figure 2 ].
Figure 2: The risk of bias assessment
Table 2
Patient baseline characteristics of the included studies
Annane et al .[8 ] (2002)
Annane et al .[6 ] (2018)
Bauer et al .[9 ] (2008)
Beale et al .[10 ] (2010)
Meta-analysis
Steroids
No steroids
Steroids
No steroids
Steroids
No steroids
Steroids
No steroids
Steroids
No steroids
Total
a = not reported in Beale et al .,[10 ] b = not reported in Annane et al .[6 ] (2018), c = not reported in Annane et al .[8 ] (2002), d = not reported in Bauer et al .[9 ]
N
150
149
614
627
21
21
3,051
5,917
3,836
6,714
10,550
Age
62
60
66
66
63.5
67.7
62.4
59.5
62.97
60.l4
61.17
Malea
96
104
424
427
12
11
532
542
l,074
SAPS IIa
60
57
56
56
56.8
59.2
56.79
56.27
56.53
APACHE IIbc
27.1
27.7
24.7
22.1
24.72
22.l2
23.01
SOFAc
12
11
11
10.1
10.1
8.6
l0.43
8.83
9.40
Positive culture/documented
121
126
450
441
14
10
585
577
1162
pathogena
Epinephrinea
41
31
53
58
1
3
95
92
187
Norepinephrinea
46
48
534
552
10
9
590
609
1,199
Dopaminea,b
136
137
4
4
140
141
281
Dobutamineabd
53
51
53
51
104
Phenylephrinea,b,c
8
1
8
1
9
Vasopressorb,c,d
2,794
4,366
Mechanical ventilation
150
149
567
569
21
21
2,80l
4,743
3,539
5,482
9,02l
Renal replacement therapyc
161
168
8
1
895
981
l,064
1,150
2,214
Primary outcome
Data concerning the previously defined short-term mortality were available for a total
of 1582 patients, collected from three studies (two RCTs and one observational).[6 ],[8 ],[9 ] From those, 785 were included in the treatment arm and had a short-term mortality
rate of 0.38, whereas 797 patients were included in the placebo arm with a mortality
rate of 0.43. Patients treated with the steroid combination were shown to have a lower
short-term mortality (OR, 0.78, CI, 0.64–0.96, I
2 = 0.00%; and P = 0.0199) [Figure 3 ].
Figure 3: Effect of corticosteroids versus placebo on short-term mortality
Secondary outcomes
The glucocorticoid and mineralocorticoid group had lower in-ICU mortality (OR, 0.77,
CI, 0.63–0.95, I
2 = 0.00%), with patient data available from three studies (two RCTs and one observational),[6 ],[8 ],[9 ] and improved shock reversal rate within 30 days (OR, 0.69, CI, 0.54–0.89, I
2 = 0.00%), with patient data available from two RCTs.[6 ],[8 ] No significant difference was observed between the two groups regarding the hospital
mortality (OR, 0.93, CI, 0.55–1.58, I
2 = 92.30%), data available from all included studies,[6 ],<xref>8-10</xref> GI hemorrhage (OR, 1.05, CI, 0.77–1.42, I
2 = 0.00%), with patient data available from two RCTs,[6 ],[8 ] or superinfection (OR, 0.95, CI, 0.64–1.42, I
2 = 21.38%), with patient data available from two RCTs[6 ],[8 ]
[Figure 4 ]
[Figure 5 ]
[Figure 6 ]
[Figure 7 ]
[Figure 8 ]. No data were available to pool in more than one study concerning the rest of the
investigated outcomes, namely intermediate-term, long-term mortality, vasopressor-free
days, ventilator-free days, duration of ICU admission, and incidence of serious adverse
effects, including hyperglycemia and delirium.
Figure 4: Effect of corticosteroids versus placebo on intensive care unit mortality
Figure 5: Effect of corticosteroids versus placebo on hospital mortality
Figure 6: Effect of corticosteroids versus placebo on shock reversal
Figure 7: Effect of corticosteroids versus placebo on gastrointestinal hemorrhage
Figure 8: Effect of corticosteroids versus placebo on superinfection
Meta-analysis of the RCTs only
A separate analysis of the RCTs was performed.[6 ],[8 ] The patients in the steroids arm showed lower short-term mortality (OR, 0.79, CI,
0.64–0.97, I
2 = 0.00%), decreased in-ICU mortality (OR, 0.77, CI, 0.63–0.95, I
2 = 0.00%), and decreased in-hospital mortality (OR, 0.77, CI, 0.63–0.95, I
2 = 0.00%) [Figure 9 ]
[Figure 10 ]
[Figure 11 ].
Figure 9: Effect of corticosteroids versus placebo on short-term mortality (randomized
trials only)
Figure 10: Effect of corticosteroids versus placebo on intensive care unit mortality
(randomized trials only)
Figure 11: Effect of corticosteroids versus placebo on hospital mortality (randomized
trials only)
DISCUSSION
This systematic review and meta-analysis was performed to investigate the simultaneous
use of low-dose glucocorticosteroids and mineralocorticosteroids as an adjunct treatment
in the management of septic shock. We showed that the steroid combination, including
mineralocorticosteroids, decreased the short-term mortality of patients with septic
shock. We additionally found that patients receiving both glucocorticosteroids and
mineralocorticosteroids had lower ICU mortality and improved shock reversal rate within
30 days. We found no increased rate of serious adverse events associated with the
administration of these therapies including incidence of GI hemorrhage or superinfection.
No impact on in-hospital mortality was noted with the inclusion of all studies, although
improvement was discovered when the meta-analysis pooled data only from the RCTs.
The question of using corticosteroids within the realm of sepsis and septic shock
has been attempted to be answered over a number of years. There has been a constant
interest on the subject that resulted in primary research studies as well as meta-analyses
assessing the utility of glucocorticosteroids.[11 ] We chose to focus on the concurrent use of both types of corticosteroids because
of the increasing volume of evidence that mineralocorticosteroids can be beneficial
to the management of sepsis and septic shock. Although multiple prior meta-analyses
have been performed to examine the impact of steroids for adjunctive treatment of
septic shock, to the best of our knowledge, this is the first investigation to focus
on the use of mineralocorticoids along with glucocorticoids.
Despite an element of discrepancy among the studies, including differences in their
choice of primary outcome and more importantly their definition of septic shock, we
chose the short-term mortality as our primary outcome, as it was thought to be the
most clinically relevant. All included studies showed mortality benefit but with mixed
levels of statistical significance. Our pooled result agreed with the aforementioned
trend with the added bonus of increased power. Despite its innate limitations as a
mortality measure, ICU mortality was also found to be very similarly decreased among
the treatment group.
The rational of administering glucocorticosteroids in the setting of septic shock
mainly focuses on reversing the multilevel dysfunction that it causes to the hypothalamus-pituitary-adrenal-tissue
axis. There are several parallel functioning mechanisms that would contribute to this
phenomenon, the analysis of which falls outside the scope of this study.[12 ],[13 ],[14 ],[15 ] However, incorporating all these knowledge into clinical practice has been challenging,
at least partially because there is no consensus concerning the optimal method of
assessing the axis in sepsis conditions.[16 ] As such, the effect of glucocorticosteroid administration on hemodynamic status
and cellular function of administering glucocorticosteroids in sepsis conditions has
been investigated directly with human and animal models.[17 ]
The mechanisms and the effects of mineralocorticosteroids are similar to the ones
associated with glucocorticosteroids. First of all, their action is performed via
cytoplasmic receptors that alter protein synthesis on activation,[17 ],[18 ] but also via non-genomic and more rapid in onset mechanisms involving plasma membrane
receptor activation. The nature and exact function of this latter mechanism has recently
been better investigated and not only does it provoke a vasoactive effect during sepsis
but also this effect is entirely independent of glucocorticosteroids.[18 ],[19 ] The perceived clinical manifestation of the action of those mechanisms is the restoration
of alpha-1 adrenergic activity with subsequent improvement of the measured blood pressure
in not only animal but also human models.[20 ]
The combined effect and interaction of glucocorticosteroids and mineralocorticosteroids
is less than well-understood and complex on several levels. First, the action of both
types of steroids is mediated at least partially independently, as detailed earlier.
Second, despite the fact that both natural steroids have equal affinity of the glucocorticoid
receptor,[21 ] the glucocorticosteroid analog, hydrocortisone, and the mineralocorticosteroid analog,
fludrocortisone, used in most experimental series, show significant potency difference
in activating this receptor.[22 ] Third, it is indeed documented that the enzyme responsible for the intracellular
metabolism of corticosteroids in the aldosterone sensitive tissues, 11β-hydroxysteroid
dehydrogenase type 2, is saturated by the daily dose of 200mg of hydrocortisone alone,[17 ],[23 ] and as such it could be argued that the activation of the mineralocorticosteroid
receptor (MR) could be solely performed by the remaining hydrocortisone without the
need for fludrocortisone. However, this statement might be misleading, as protein
synthesis provoked by the binding of a ligand to the receptor is further regulated
by a significant variety of additional factors, mainly coactivator and corepressor
proteins.[24 ] The function of a number of these molecules depends on the nuclear redox state,
essentially functioning as a redox status sensing mechanism.[25 ] The activation of 11β-hydroxysteroid dehydrogenase type 2 directly results in intracellular
redox status changes via the metabolism of nicotinamide adenine dinucleotide (NAD+)/NADH.[26 ] In short, the MR-ligand binding could very well produce different results with regard
to protein synthesis, with glucocorticosteroids functioning either as an activator
or an inhibitor depending on the redox state of the tissue. As expected, this effect
cannot be accurately predicted during septic shock, especially in the endothelial
as well as other blood vessel cells.[17 ],[24 ] Although difficult to exhaust the physiology behind their interaction, the net effect
of the simultaneous use of glucocorticosteroids and mineralocorticosteroids seems
to be the achievement of improved blood pressure control, decreased pressor requirements
in healthy and septic conditions in human and animal models.[4 ],[5 ]
The aforementioned mechanisms could explain the discrepancy of the results between
the adjunctive corticosteroid treatment in critically ill patients with septic shock
and activated protein C and corticosteroids for human septic shock trials,[6 ],[27 ] especially with regard to their primary outcome, with the first failing and the
latter succeeding in showing statistically significant 90-day mortality benefit between
the treatment and placebo groups. Other factors might have attributed to this discrepancy
as well. First, there were methodological differences between the two trials, with
the ADRENAL trial using an infusion of hydrocortisone instead of bolus doses used
in APROCCHSS trial, thus achieving therapeutic levels later, and second, allowing
longer time to randomization than APROCCHSS trial. These differences may have caused
adequate temporal delay to miss a potentially reversible stage of shock in the ADRENAL
trial. Finally, the baseline characteristics of the patients included in the two trials
were deemed unequal, especially concerning the severity of the shock.
At this point, it is important to mention findings of the Corticosteroids and Intensive
Insulin Therapy for Septic Shock (COIITSS) trial.[28 ] In a study designed to assess the effect of tight glycemic control versus regular
practice control as well as the effect of the combination of glucocorticosteroid and
mineralocorticosteroid versus the sole use of glucocorticosteroids, no difference
in the measured variables was revealed. Despite this result, it is thought that the
effort to answer both questions might have compromised its ability to tackle either
of them. Furthermore, the research team that conducted the study has suggested that
it might have been underpowered to demonstrate the difference between the two treatments.
Regarding our secondary end points, we are unable to reach a safe conclusion concerning
hospital mortality. Although not entirely clear, the aforementioned discrepancy could
partially be attributed to a difference in the definitions of shock (including the
significant detail of inclusion of patients before or after a level of initial fluid
resuscitation), the definition of resolution of shock, the specific definition of
said measures in its study, and the difference in practice among the study centers.
A characteristic example for the latter would be the frequency and speed of the decision
to de-escalate terminal patients from ICU, directly altering not only ICU but hospital
mortality as well. Especially concerning the assessment of the rate of shock resolution,
the documented measures differed between vasopressor-free days and vasopressor use
duration, making statistical analysis difficult without access to patient-level data.
Finally, intermediate and long-term mortality were indeed assessed as secondary end
points but the paucity of data for these end points was observed in the form of lack
of well-aligned, more long-term (e.g., 90, 180 days, or 1 year) mortality reporting
between the studies.
To the best of our knowledge, this was the first systematic review and meta-analysis
dedicated to the study of the effect of both steroids as an adjunct therapy to the
septic shock treatment. Our methodology adhered to PRISMA guidelines and our search
included all the major databases. The quality of the included studies was assessed
for biases with the appropriate assessment tools. Finally, the heterogeneity of the
included studies was low.
Our study did have certain limitations. First, the number of studies that met our
criteria was small. The clinical question of steroid use in sepsis had been debated
over a number of decades with no conclusive answer. Despite the relatively large number
of studies using glucocorticosteroids for the management of sepsis, a far smaller
number combined them with mineralocorticosteroids. As a result of the aforementioned,
it was deemed necessary to include observational studies. However, the results were
not substantially altered by the inclusion of the observational studies, even though
the main contribution to in-hospital mortality patient numbers belonged to one observational
study. Finally, one of the major questions, rate of shock resolution, was left unanswered
because of the different methods of assessing this variable. Studies tend to use and
publish either the duration of shock or pressor-free days, measures that are not interchangeable,
as explained earlier.
In conclusion, our systematic review and meta-analysis concluded that coadministration
of glucocorticosteroids and mineralocorticosteroids to patients with septic shock
results in decreased short-term mortality, decreases ICU mortality, and increases
incidence of shock resolution in 30 days compared to placebo. No definitive effect
was appreciated on hospital mortality as well as on the incidence of undesired effects,
specifically development of superinfection and GI hemorrhage. Although future investigation
into the role of adjunctive mineralocorticoids in septic shock is needed, our analysis
suggests that clinicians may have a sound basis to administer these agents along with
glucocorticosteroids to patients with septic shock.
