Efficacy and safety of corticosteroids in immunocompetent patients with septic shock

Abstract

BACKGROUND:

The use of corticosteroids in septic shock has been studied for many decades but yielded conflicting results. We conducted a systematic review to evaluate the efficacy and the safety of corticosteroids in immunocompetent patients with septic shock.

METHODS:

Medline via PubMed, Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library, and EMBASE were searched from inception to March 2020. Two reviewers independently identified randomized controlled trials (RCTs) comparing corticosteroids with a control group for immunocompetent patients with septic shock. Data were abstracted and reported following the Cochrane Handbook for Systematic Review of Intervention and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. The efficacy outcome included mortality and shock reversal. The safety outcomes were infection, gastrointestinal bleeding, and hyperglycemia.

RESULTS:

Nine RCTs with a total of 1,298 patients were included. Compared with the control group, corticosteroid group did not lower the short-term (28 or 30 days) mortality (risk ratio [RR] 0.95, 95% confidence interval (CI) 0.85 to 1.06, inconsistency [I²]=0%, trial sequential analysis [TSA]-adjusted CI 0.83 to 1.09, moderate-certainty evidence). Corticosteroids significantly shortened the time to shock reversal compared with the control group (mean difference [MD] –21.56 hours; 95% CI –32.95 to –10.16, I²=0%; TSA-adjusted CI –33.33 to –9.78, moderate-certainty evidence). The corticosteroid treatment was associated with an increased risk of hyperglycemia but not the infection or gastrointestinal bleeding.

CONCLUSIONS:

The corticosteroid treatment is not associated with lower short- or long- term mortality compared with placebo in immunocompetent patients with septic shock. However, corticosteroids significantly shorten the time to shock reversal without increasing the risk of infection. The patient’s immune status should also be considered during clinical treatment and clinical trials in future.

INTRODUCTION

Septic shock, a life-threatening organ dysfunction caused by the dysregulated host response to infection, is characterized by severe circulatory, cellular, and metabolic abnormalities.[1] It has been regarded as a formidable clinical challenge associated with mortality 30% to 40%.[2,3] Septic shock is a common clinical syndrome, but has pronounced heterogeneity such as variable infection sites and sources, pathogen species, and host comorbidities.[4] There has been an increasing emphasis on evidence-based adjunct therapy beyond hemodynamic support and antimicrobial therapy.[5,6]

Corticosteroids have been used in the treatment of patients with septic shock for more than half a century.[7] Till now, nearly thirty randomized controlled trials (RCTs) have evaluated the efficacy of corticosteroids in these patients but yielded different results, including two well-known RCTs published in the year 2018.[8,9] Twelve systematic reviews since 2018 have been conducted to try to address the discrepancy in these previous trials by classifying the doses of steroids and the severity of shock.[10-21]

However, these studies and reviews have not yet addressed the heterogeneity of the patient population, such as the immunological state of a patient, which is another important clinical aspect and may result in significant enrollment bias.

Recently, focusing on immunocompromised patients with septic shock, we performed an observational cohort study, and found that corticosteroid therapy had adverse effects on survival, hemodynamic stability, and hospital duration in the selected population.[22] Therefore, we aim to perform a systematic review which eliminated the impact of immune status to assess the benefits and risks of corticosteroids in septic shock, and to identify the exact group of patients who may benefit from corticosteroid treatment.

METHODS

Search strategy

We systematically performed electronic search of Medline via PubMed, Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library, and EMBASE from inception to March 12, 2020. We combined MeSH and title/abstract keywords, such as “steroids”, “glucocorticoids”, “corticosteroids”, “prednisolon”, “methylprednisolon”, “prednison”, “dexamethasone”, “triamcinolon”, “fludrocortisone”, “betamethasone”, “hydrocortisone”, “sepsis”, and “shock, septic” to identify all RCTs comparing corticosteroids with a control group for immunocompetent patients with septic shock.

Study selection

Two authors independently identified the trials for inclusion based on their titles and abstracts, and evaluated the full texts of the papers.

Eligibility criteria

(1) Population. Immunocompetent adult patients with septic shock, defined based on the definition of included trials, were eligible for inclusion. Sepsis patients without circulatory failure were excluded. The immunocompetent patient was defined as the exclusion of one or more immunocompromised underlying conditions, including immunosuppression, immunodeficiency, immunosuppressive therapy, human immunodeficiency virus positive or acquired immune deficiency syndrome, advanced or end-stage neoplasm, and organ transplant recipients. (2) Intervention. All types of corticosteroids were included, regardless of the formula, dose, start time, and duration of treatment. (3) Control. The control group was allowed for the following interventions: placebo, saline, or no intervention. (4) Outcomes. The primary outcome was short-term mortality during intensive care unit (ICU) or hospital stay. The “short term” was defined as the mortality on day 28 or day 30. The secondary outcomes included mortality variables, the number of patients with shock reversal (stable hemodynamic status more than 24 hours after withdrawal of vasopressor therapy) within 28 days, and time to shock reversal. The safety outcomes included infection, gastrointestinal bleeding, and hyperglycemia. (5) Type of study. All trials included were RCTs, irrespective of language or publication status.

Data extraction and quality assessment

Characteristics of participants, study design, and outcomes for analyses were extracted following a standardized data extraction form by two reviewers independently. Two investigators independently assessed the risk of bias according to the Cochrane Handbook for Systematic Review of Intervention to assign a value of “high”, “low”, or “unclear” for each trial.

We used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology to evaluate the quality of evidence associated with each major outcome and present the results in the summary of findings (SoFs) table.

Statistical analysis

All statistical analyses were performed on Review Manager 5.3 software and trial sequential analysis (TSA) v.0.9.5.10 beta.[23] We presented results as relative risk ratio (RR) for dichotomous data and mean difference (MD) for continuous data, which were pooled using the Mantel-Haenszel (M-H) and inverse variance method, respectively. Both RR and MD were provided with 95% confidence interval (CI). Heterogeneity was assessed by the Chi-square test with significance set at a P-value of 0.05, and quantitatively by inconsistency (I²) statistics. We reported all results from a more conservative random-effect model taking into consideration clinical heterogeneity. Subgroup analyses were also performed for all outcomes based on the trial quality.

TSA

We performed TSA to assess the increased risk of random errors due to the relatively sparse data and repeated significance testing. The result was displayed on a TSA diagram with a TSA-adjusted CI and an adjusted level of statistical significance. TSA was used to appropriately reduce the risk of a wrong conclusion in a meta-analysis that did not achieve the required information size (RIS). TSA-adjusted CI was calculated by the random-effect model for diversity (D2) with 5% risk of type I error and a power of 80%. For the estimate of the RIS, we set the intervention effect of a 15% relative risk reduction (RRR), and calculated the control event incidence from the conventional meta-analysis.

RESULTS

Study characteristics

Of the 4,034 records identified in our research, full texts of 207 records were reviewed, and 27 trials initially included were assessed for patients by immune status. Ultimately, nine RCTs were included in our systematic review.[24-32] The results of the search and selection flow diagram were shown in Figure 1. The detailed descriptions of the included trials were presented in Table 1. Nine RCTs with a total of 1,298 participants were finally analyzed, comprising 667 in the corticosteroid group and 631 in the control group.[24-32]

Figure 1.

Flow diagram showing results of the search and selection of eligible studies. RCT: randomized controlled trial; study design: not RCT; population: no exclusion of immunosuppression or not septic shock; intervention: not corticosteroids.

Table 1.

Characteristics of included RCTs comparing corticosteroids versus control in immunocompetent patients with septic shock

Mortality

The short-term mortality in the corticosteroid and the control groups was 43.8% (292/667) and 45.2% (285/631), respectively. The pooled analysis revealed no statistically significant effects of corticosteroids (RR 0.95, 95% CI 0.85 to 1.06, P=0.37, I²=0%, TSA-adjusted CI 0.83 to 1.09, moderate-certainty evidence) (Figures 2 and 3, Table 2). TSA with RRR 15% produced an incidence of 45.1% and 38.3% in the control and corticosteroid groups, respectively. The cumulative Z-curves crossed the futility area, which excluded an effect size of 15% RRR or larger (Figure 3).

Figure 2.

Forest plot of all trials for short-term mortality. CI: confidence interval; M-H: Mantel-Hansen; df: degrees of freedom.

Figure 3.

Trial sequential analysis of all trials for short-term mortality. TSA: trial sequential analysis. The required information size was 1,671 patients. The incidence in the control arm of 45.1% with a relative risk reduction of 15.0% produced an incidence of 38.3% in the corticosteroid group. The TSA-adjusted 95% confidence interval for a relative risk of 0.95 was 0.83 to 1.09 and the cumulative Z-curves crossed futility area.

Table 2.

Summary of findings for all included RCTs (grading of recommendations assessment, development, and evaluation)

For the long-term mortality, the pooled estimate of RR for 1-year mortality for corticosteroids compared with control was 0.96 (95% CI 0.87 to 1.07, P=0.49, I²=0%, high-certainty evidence). Compared with placebo or the control group, corticosteroids lowered the 7-day mortality (RR 0.68, 95% CI 0.51 to 0.90, P<0.01, I²=0%, low-certainty evidence) in initial meta-analysis. However, the TSA-adjusted CI of the random-effect model was 0.39 to 1.16 without the TSA monitoring boundary being crossed, which was not statistically significant and indicated that the effect was uncertain.

Shock reversal

The conventional analysis revealed a statistically significant shortening of time to shock reversal in favor of corticosteroids (MD –21.56 hours, 95% CI –32.95 to –10.16, P<0.01, I²=0%, TSA-adjusted CI –33.33 to –9.78, moderate-certainty evidence). For shock reversal within 28 days, there was no significant difference between the corticosteroid group and the control group.

Safety outcomes

Corticosteroids likely increased the rates of hyperglycemia (RR 1.14, 95% CI 1.03 to 1.27, P=0.01, I²=0%, TSA-adjusted CI 1.00 to 1.30, moderate-certainty evidence). However, the side effects of corticosteroids on infection and gastrointestinal bleeding were not significant.

Subgroup analyses for outcomes based on trial quality

Subgroup analyses for outcomes were performed according to the risk of bias. The results did not demonstrate a beneficial effect of corticosteroids in reducing short-term mortality in the subgroup of high-quality trials (RR 0.89, 95% CI 0.74 to 1.08, P=0.24; I²=0%). For other mortality outcomes, results from trials at the low risk of bias did not substantially differ from the results of all trials.

Study quality

There were two trials classified as low risk of bias,[29,30] four trials as unclear risk of bias,[26-28,32] and three trials as high risk of bias.[24,25,31] Due to the number of studies included in each analysis less than ten, publication bias was not evaluated.

DISCUSSION

In this meta-analysis of nine RCTs with 1,298 patients with septic shock, we found no benefits of corticosteroids on either short-term mortality or long-term mortality. Our pooled analysis revealed that the administration of corticosteroids resulted in shorter time to shock reversal compared with the control group.

To the best of our knowledge, this is the first systematic review or meta-analysis to assess the efficacy and safety of corticosteroids in patients with septic shock based on the patients’ immune status. Previous reviews mainly enrolled patients with sepsis or septic shock and performed subgroup analyses based on the trial quality, the doses and regimens of corticosteroids, and the severity of diseases.[10-15] However, there was no differentiation or discussion of the immunological status of patients.

There is a consensus on the definitions for the immunocompetent state and the immunocompromised status: the former is usually defined as the exclusion of the latter. There were some variations in the definition of “immunocompromised” in each of the aforementioned studies.

The mechanism of corticosteroids in septic shock may be its ability to down-regulate the pro-inflammatory response.[33] However, the balance between the immune enhancement and suppression is highly dependent on the immune activation of the host as well as the dose and duration of corticosteroid therapy.[33,34] Immunocompetent patients may exhibit a profound hyper-inflammatory response followed by the cascade of events in the early stage of the disease, when the application of corticosteroids for control of systemic inflammatory response syndrome may be beneficial. The theory might partially account for the findings that corticosteroids significantly reduced the time to shock reversal. While signs of compensated anti-inflammatory response syndrome may predominate in the whole stages of immunocompromised patients, the assignment of corticosteroids might strengthen the immunosuppression resulting in the accelerated deterioration of septic shock.[34,35] Although corticosteroids did not reduce the short-term and long-term mortalities in immunocompetent patients with septic shock, it was helpful in shock reversal without increasing the risk of infection. Given the findings, the administration of corticosteroids could be considered in immunocompetent patients suffering from septic shock to achieve hemodynamic stability.[36]

Our study has several limitations. Firstly, the systematic review was not registered in the International Prospective Register of Systematic Reviews (PROSPERO), and no protocol has been published. Secondly, we tried to contact the authors of included trials to gather data on immunocompetent persons, but many trials were excluded because of the lack of detailed data on the patients’ immune status. Thirdly, there were many “unclear” ratings for risk of bias assessments, although we attempted to contact trial authors to clarify these ambiguities.

CONCLUSIONS

Corticosteroid therapy is not associated with the lower short- or long-term mortalities compared with placebo in immunocompetent patients with septic shock. However, corticosteroids significantly shorten the time to shock reversal without increasing the risk of infection. The patient’s immune status should also be considered during clinical treatment and clinical trials in future.