The impact of selenium administration on severe sepsis or septic shock: a meta-analysis of randomized controlled trials

Introduction The efficacy of selenium administration to treat severe sepsis or septic shock remains controversial. We conduct a systematic review and meta-analysis to explore the impact of selenium administration on severe sepsis or septic shock. Methods We search PubMed, EMbase, Web of science, EBSCO, and Cochrane library databases through May 2020 for randomized controlled trials (RCTs) assessing the effect of selenium administration on severe sepsis or septic shock. Meta-analysis is performed using the random-effect model. Results Five RCTs involving 1482 patients are included in the meta-analysis. Overall, compared with control group in septic patients, selenium administration is not associated with reduced 28-day mortality (RR=0.93; 95% CI=0.73 to 1.19; P=0.58), but results in substantially decreased all-cause mortality (RR=0.78; 95% CI=0.63 to 0.98; P=0.03) and length of hospital stay (MD=-3.09; 95% CI=-5.68 to -0.50; P=0.02). Conclusion Selenium administration results in notable decrease in all-cause mortality and length of hospital stay, but shows no substantial influence on the 28-day mortality, length of ICU stay, duration of vasopressor therapy, the incidence of acute renal failure, adverse events, and serious adverse events for septic patients.


Introduction
Sepsis and septic shock remain a major health care problem, and the incidence of severe sepsis has increased by 1.5 % ever year 1-3 . Severe sepsis has emerged as the leading cause of hospitalization and accounts for 2 % of all hospital admission, 59 % of the patients in intensive care unit (ICU) 4, 5 . The mortality of severe sepsis and septic shock still remains unacceptably high despite of the treatment development 6-8 . failure and death 12 . Decreased plasma selenium levels are found to be associated with excess mortality 13 . Intravenous administration of selenium is reported to restore the activity of glutathione peroxidase, attenuates oxidative stress, and may improve survival in septic patients [14][15][16] . In contrast, a cocktail of antioxidants and vitamins containing 800 ug of selenium do not decrease 28-day mortality in patients receiving mechanical ventilation 17 . This increase in the concentration of selenium is associated with the reduction in the oxidative stress and sequential organ failure assessment (SOFA) score 15 .
The use of selenium has not been well established in septic patients. Recently, several studies on the topic have been published, and the results have been conflicting [18][19][20] . With accumulating evidence, we therefore perform a systematic review and meta-analysis of RCTs to investigate the efficacy of selenium administration to treat septic patients.

Materials and methods
Ethical approval and patient consent are not required because this is a systematic review and meta-analysis of previously published studies. The systematic review and meta-analysis are conducted and reported in adherence to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 21 .

Search strategy and study selection
Two investigators have independently searched the following databases (inception to May 2020): PubMed, EMbase, Web of science, EBSCO, and Cochrane library databases. The electronic search strategy is conducted using the following keywords: s/span>elenite or selenium, and sepsis or septic. We also check the reference lists of the screened full-text studies to identify other potentially eligible trials. The inclusive selection criteria are as follows: (i) population: patients diagnosed with severe sepsis, septic shock; (ii) intervention: selenium administration; (iii) comparison: placebo; (iv) study design: RCT.

Data extraction and outcome measures
We have extracted the following information: author, number of patients, age, body mass index, Acute Physiologic and Chronic Health Evaluation II (APACHE II) scores etc. The data have been extracted independently by two investigators, and discrepancies are resolved by consensus. We also contact the corresponding author to obtain the data when necessary. No simplifications and assumptions are made. The primary outcomes are 28-day mortality and all-cause mortality. Secondary outcomes include length of ICU stay, length of hospital stay, duration of vasopressor therapy, the incidence of acute renal failure, dverse events, and serious adverse events.

Assessment for risk of bias
We used the risk of bias tool to assess the quality of individual studies in accordance with the Cochrane Handbook for Systematic Reviews of Interventions 22 , and the sources of bias included selection bias, performance bias, attrition bias, detection bias, reporting bias, and other potential sources of bias. The bias was evaluated and rated: high, low, and unclear 23 . Two investigators independently searched articles, extracted data, and assessed the quality of included studies. Any discrepancy was solved by consensus.

Statistical analysis
We estimate the mean difference (MD) with 95% confidence interval (CI) for continuous outcomes (length of ICU stay, length of hospital stay, and duration of vasopressor therapy) and the RR with 95% CIs for dichotomous outcomes (8-days mortality and all-cause mortality, the incidence of acute renal failure, adverse events, and serious adverse events). A random-effects model is used regardless of heterogeneity. Heterogeneity is reported using the I2 statistic, and I2 > 50% indicate significant heterogeneity 24 . Whenever significant heterogeneity is present, we search for potential sources of heterogeneity via omitting one study in turn for the meta-analysis or performing subgroup analysis. Publication bias is not evaluated because of the limited number (<10) of included studies. All statistical analyses are performed using Review Manager Version 5.3 (The Cochrane Collaboration, Software Update, Oxford, UK).

Literature search, study characteristics and quality assessment
A detailed flowchart of the search and selection results is shown in Figure 1. 596 publications are searched after the initial search of databases. 161 duplicates and 426 papers are excluded after checking the titles/abstracts. 4 studies are removed because of the study design25-28 and five RCTs that meet our inclusion criteria are finally included in the meta-analysis 16,[18][19][20]29 .

Assessment of risk of bias
Risk of bias analysis (Figure 2) showed that one study had high risk of bias because it was a single-blind RCT 19,29 , and another trial had high risk of bias due to the non-blindness 18 . Primary outcome: 28-day mortality and all-cause mortality These two outcome data are analyzed with the random-effects model, and the pooled estimate of the four included RCTs suggested that compared to con-trol group in septic patients, selenium administration shows no substantial influence on the 28-day mortality (RR=0.93; 95% CI=0.73 to 1.19; P=0.58), with low heterogeneity among the studies (I2=44%, heterogeneity P=0.15) (Figure 3). In contrast, selenium administration is found to significantly reduce all-cause mortality (RR=0.78; 95% CI=0.63 to 0.98; P=0.03) after pooling the results of three included RCTs, with low heterogeneity among the studies (I2=8%, heterogeneity P=0.34) (Figure 4).

Sensitivity analysis
Low heterogeneity is observed among the included studies for the 28-day mortality and all-cause mortality. Thus, we do not perform sensitivity analysis by omitting one study in each turn to detect the source of heterogeneity.

Discussion
Our meta-analysis suggests that selenium administration is associated with significantly reduced all-cause mortality and length of hospital stay, but has no remarkable influence on 28-day mortality, length of intensive care unit (ICU) stay, duration of vasopressor therapy, and the incidence of acute renal failure in septic patients.
An overwhelming inflammation occurs in the initial phase of sepsis, during which migration of the neutrophils into the inflamed tissue releases free radicals that result in capillary congestion due to the damage of the endothelium and epithelium tissue, and respiratory fail-ure due to leukocyte and macrophage infiltration into the site of inflammation of the respiratory system 14,30 . Therapies counteracting the inflammation and oxidative stress remain attractive and challenging. Selenium is located on the catalytic center of most selenoenzymes, serves as an essential trace element for the biosynthesis and function of about 25 known selenocysteine containing selenoproteins 31 . One of the best-known redox systems is glutathione complex consisting of the selenium-dependent peroxidases and the thioredoxin reductases 32,33 . Selenium has been reported to have an important role as anti-inflammatory agent by tightly regulating the expression of proinflammatory genes in immune cells 34,35 .
Some multiple-center trials confirm the efficacy of high-dose sodium selenite supplementation in patients with severe sepsis and septic shock to reduce of 28-day mortality 16,36,37 . However, other clinical studies are not consistent with this results. The 28-day mortality is not decreased after selenium administration in septic patients and in critically ill patients 15,20 . One clinical study demonstrates that administration of selenium leads to a slight decrease in the duration of mechanical ventilation and length of ICU stay 38 . In this meta-analysis, the mortality and length of hospital stay was confirmed to be reduced after selenium intervention in septic patients.
Patients are resuscitated with selenium initially, and end up with fewer recurrent pneumonia and sepsis, but they are weaned quicker due to cardioprotective properties of selenium against reperfusion and reoxygenation. Thus, selenium may increase sepsis-related cardiomyopathy and other complications 38 . However, there is no increase in adverse events, and serious adverse events after selenium administration for septic patients based on the results of our meta-analysis.
The signaling pathways in inflammation and oxidative stress in vivo are quite complex during the sepsis, and there is always a cross talk between inflammation, oxidative stress and coagulation 39,40 . Regarding the effect of selenium on oxidative stress in septic patient, selenium administration is associated with a significant increase in the plasma activity glutathione peroxidase from day 3 to 14, but has no effect on the average plasma levels of inflammatory cytokines IL-6, IL-8 and IL-10 14, 41 . The production and release of sepsis mediators is a complex network rather than as a cascade. Although one of the substances responsible for the initial phase is blocked, other mediators will likely maintain the septic response 42,43 . In 165 patients requiring mechanical ventilation for sepsis or septic shock, a study formula consists of high level of omega-3 fatty acids and γ-linolenic oil, less omega-6 fatty acids, and high doses of the antioxidant vitamins E and C and selenium, and patients receiving this study formula result in significant decrease in mortality and organ dysfunction, as well as the improvement in oxygenation, ventilator-free days and ICU-free days 44 .
This meta-analysis has several potential limitations that should be taken into account. Firstly, our analysis is based on only five RCTs and three of them have a relatively small sample size (n<100). Overestimation of the treatment effect is more likely in smaller trials compared with larger samples. Next, the methods and duration of selenium administration are different in included RCTs, and may have an influence on the pooling results. Finally, some unpublished and missing data may lead bias to the pooled effect.

Conclusion
Selenium administration may have the important ability to reduce mortality and length of hospital stay in patients with sepsis or septic shock, and selenium should be recommended to be administrated in clinical work with caution.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Conflict of interest
None.

Funding
Not applicable.