The role of selenium in depression: a systematic review and meta-analysis of human observational and interventional studies

The results of human studies are inconsistent regarding selenium and depressive disorders. Therefore, we aimed to conduct a systematic review and meta-analysis of observational and interventional studies and provided an overview of the role of selenium in depression. Three databases including Medline, Scopus, and Web of Science were searched on June 30, 2020 and updated on April 12, 2021. Also, we searched in electronical databases of WHO Global Index Medicus and ClinicalTrials.gov. No time or language restrictions were used for the search. A random effects model was used to pool effect sizes. In total, 20 studies were included in the systematic review, and 15 studies were included in the meta-analysis. There were no significant differences in serum selenium levels between patients with depression and healthy subjects (WMD: 2.12 mg/L; 95% CI: − 0.11, 4.36; I2 = 98.0%, P < 0.001). Also, no significant correlation was found between serum levels of selenium and depression scores (r: − 0.12; 95% CI: − 0.33, 0.08; I2 = 73.5%, P = 0.010). Nevertheless, there was a significant negative association between high selenium intake and the risk of postpartum depression (OR: 0.97; 95% CI: 0.95, 0.99; I2 = 0.0%, P = 0.507). In addition, selenium supplementation significantly reduced depressive symptoms (WMD: − 0.37; 95% CI: − 0.56, − 0.18; I2 = 0.0%, P = 0.959). Taken these results together, selenium seems to have a protective role against postpartum depression and can be considered as a beneficial adjuvant therapy in patients with depression. Further studies are necessary to draw definitive conclusions.

Data extraction and quality assessment. The following characteristics were collected from the included publications: the first author's last name, year of publication, country where the study was conducted, sample size, gender and mean or median age of participants, study design, type of depressive disorder, assessment tool of depression, adjusted covariates, main results, values of selenium intake or levels of serum or plasma selenium, serum selenium values at baseline and after-treatment, methodologies of selenium measurement. The quality assessment of observational studies was performed using the Newcastle-Ottawa Scale (NOS) 26 , and the quality of interventional studies was evaluated using the Cochrane Collaboration Risk of Bias Tool (CCRBT) 27 .
Data synthesis and analysis. To improve normal distribution, correlation coefficients between serum selenium levels and depression scores were converted to z-values using Fisher's r-to-z transformation. Subsequently, following formula was used to converted back to r-values when effect sizes were calculated: ES (z) = ½ ln [(1 + r)/(1 − r)] 28 . We converted standard errors (SE) to standard deviations (SD) using the formula SD = SE × √N. To calculate SD from 95% confidence interval, following formula was used: SD = √N × (upper limit − lower limit) ÷ 3.92. A random-effects model was used to calculate pooled effect size to compare serum selenium levels between depressive patients and healthy controls 29 . We used the random-effects model because inter-study heterogeneity was high. The random-effects model should be used for pooling heterogeneous studies 30 . Similar method was applied to compare change in depression scores between selenium supplementation and control groups. Since included clinical trials used different tools to assess depression score, pooled effect was calculated via Hedges' g 31 . Log-transformed odds ratios of depression across different categories of selenium intake were also applied to calculate overall effect sizes. Overall effect sizes were reported as odds ratio (OR), weighted mean difference (WMD) and correlation coefficient (r). I-squared (I 2 ) statistic was reported as an indicator of between-study heterogeneity. To detect the potential sources of heterogeneity, a subgroup analysis was applied when a significant between-study heterogeneity was observed. Sensitivity analysis was performed as a complementary analysis to assess robustness of results. Begg's rank correlation test and Egger's linear regression test were used to test publication bias. The potential effect of publication bias was assessed using trim-and-fill analysis. All statistical analyses were performed using Stata software (version 11.2, Stata Corporation, College Station, Texas, USA); additionally, analyses were two-tailed, and statistical significance was set at P < 0.05.

Systematic review.
Characteristics of studies eligible for the systematic review are summarized in Table 1.
Two cross-sectional studies found no significant correlation between selenium intake and depressive symptoms 32,42 , whereas in one of the studies, there was a significant inverse correlation between selenium intake and depressive symptoms in males 32 . Also, two cross-sectional studies found no significant correlation between serum selenium and depression scores 18,19 . In addition, one cross-sectional study revealed a negative association between serum selenium levels and the risk of depression 15 . Furthermore, two case-control studies reported lower levels of selenium in depressive subjects compared to healthy controls 39,40 . Moreover, three cross-sectionals 37,38,41 , two case-controls 16,43 , and two cohorts 17,33 reported a significant negative association between selenium intake and the risk of depression.
Two randomized controlled trials reported correlation coefficients. They found a significant inverse relationship between serum selenium levels and depression symptoms 21,36 . All interventional studies used selenium supplements except for one study that assessed the effect of selenium rich diet on depression symptoms 36 . The dose of selenium supplementation was varied from 100 to 200 μg [20][21][22] . A beneficial effect of selenium on depressive symptoms was reported in three studies 20,21,36 . However, one clinical trial found no significant effect of selenium on depression scores 22 . Quality assessment of studies. The results of the CCRBT showed that all included randomized controlled trials had high quality (Table 2). According to the NOS, all case-control and prospective cohort studies obtained ≤ 4 stars, i.e., low quality scores (Tables 3, 4). Similarly, the quality of all cross-sectional studies was low except for Ghimire 38 and Li 37 that respectively received good and excellent quality (Table 5).

Meta-analysis.
From 20 studies included in systematic review, five studies were not selected for meta-analysis 15,32,34,35,42 . Two studies not included to meta-analysis reported the correlation coefficient between dietary intake of selenium and depression 32,42 . We could not pool these two studies because the score of depression was derived from different depression assessment instruments. One study was not included to meta-analysis because it reported median depression score across tertiles of plasma selenium concentration 34 . A reported selenium intake across tertiles of mood thermometer 35 . Since similar report was not found in other studies, we did not include this study to meta-analysis. Another study not selected for meta-analysis reported regression coefficient 15 . Therefore, a quantitative analysis was performed on 15 studies including 45,795 participants [16][17][18][19][20][21][22]33,[36][37][38][39][40][41]43 .
Correlation between serum selenium levels and depression scores was assessed in four studies. The metaanalysis showed no significant correlation between serum levels of selenium and depression scores (r: − 0.12; 95% CI: − 0.33, 0.08) (Fig. 2). Although a significant between-study heterogeneity was found (I 2 = 73.5%, P = 0.010), we could not run a subgroup analysis due to the insufficient number of studies. Moreover, there was an evidence of significant publication bias using Egger's (P = 0.029) and Begg's (P = 0.042) tests. Notwithstanding, trim-and-fill analysis indicated that no trimming could be performed and the data remained unchanged.
Comparison of serum selenium levels between depressive patients and healthy controls was reported in two studies. As shown in Fig. 3, the pooled results revealed that there were no significant differences in serum selenium concentrations between patients with depression and healthy subjects (WMD: 2.12 mg/L; 95% CI: − 0.11, 4.36). There was a significant heterogeneity between studies (I 2 = 98.0%, P < 0.001). However, we could not run a subgroup analysis because of the insufficient number of studies. Moreover, the result did not show Association between selenium intake and the risk of depression was reported in seven studies. The pooled risk of depression in the highest compared with the lowest categories of selenium intake was 0.98 with 95% CI of 0.93 to 1.04. A significantly high heterogeneity was observed between studies (I 2 = 82.7%, P < 0.001). Therefore, we subgrouped studies based on the type of depression (postpartum or other types of depression) (Fig. 4). There was a significant association between selenium intake and the risk of postpartum depression (OR: 0.97; 95% CI: 0.95, 0.99; I 2 = 0.0%, P = 0.507). Nevertheless, no significant association was found between selenium intake   Table 4. Quality assessment of the included cohort studies according to the Newcastle-Ottawa Scale.

First author (year)
Representativeness of the exposed cohort Selection of the non-exposed cohort

Ascertainment of exposure
Outcome not present at start of study

Assessment of outcome
Follow-up long enough for outcomes to occur

Adequacy of follow-up of cohorts
Total quality score Sánchez-Villegas (2018) -* -* * -* -4 Table 5. Quality assessment of the included cross-sectional studies according to the Newcastle-Ottawa Scale. Between-subgroup heterogeneity was also high for the type of depression (P = 0.012). Moreover, we did not find any evidence of publication bias using Egger's (P = 0.65) and Begg's (P = 0.80) tests. The Effect of selenium supplementation on depression scores was examined in three studies. The effect of supplementation with selenium on depression scores is shown in Fig. 5. The meta-analysis indicated a significant reduction in depression symptoms following selenium supplementation compared with placebo (WMD: − 0.37; 95% CI: − 0.56, − 0.18). There was no significant heterogeneity between studies (I 2 = 0.0%, P = 0.959). Moreover, no evidence of publication bias was found using Egger's (P = 0.11) and Begg's (P = 0.12) tests.

Sensitivity analysis.
The sequential exclusion of each study from the pooled analysis did not significantly change the overall effect sizes except for the correlation between serum selenium concentrations and depression scores. This was significantly altered by excluding the study of Ekramzadeh et al. (r: − 0.20; 95% CI: − 0.381, − 0.031). In addition, the sequential removal of each study from the pooled analysis did not eliminate the heterogeneity except for the association between selenium intake and the relative risk of depression.

Discussion
This meta-analysis revealed that there was no significant correlation between serum selenium levels and depression scores. In addition, no significant differences were observed between depressive and healthy subjects in serum selenium concentrations. In contrast, a significant inverse association was found between selenium intake and the risk of postpartum depression. Moreover, the meta-analysis of randomized controlled trials indicated a significant reduction in depression symptoms after selenium supplementation compared with placebo. To the  www.nature.com/scientificreports/ best of our knowledge, the present study is the first systematic review and meta-analysis of human observational and interventional studies that comprehensively investigated the role of selenium in depressive disorders. Prior to this study, three systematic reviews suggested that nutrients such as selenium may be protective against postpartum depression. Nevertheless, these studies only focused on perinatal depression, not other types of depression. Moreover, they did not run a meta-analysis [44][45][46] . The findings of this meta-analysis did not show any significant correlation between serum selenium concentrations and depression scores. In contrast, one study reported that there was a significant direct association between high dietary selenium intake and mood improvement 47 . According to the previous studies, serum selenium levels could not estimate the absolute intake of selenium 48 . In fact, some factors including demographic variables and health status may influence serum selenium concentrations 15 . Among the studies included in the present meta-analysis, confounding variables were adjusted in only one study 18 . Moreover, it seems that brain function is impaired by long-term (not short-term) exposure to low serum selenium levels 49 . In spite of this  www.nature.com/scientificreports/ fact, the included studies reported no data regarding the duration of selenium deficiency. These reasons may explain the non-significant association between serum selenium and depression symptoms in our meta-analysis. Nevertheless, it is noteworthy that removal of the study by Ekramzadeh et al. significantly changed this result and brought about a significant negative correlation between serum concentrations of selenium and depression scores. Ekramzadeh et al. investigated the relationship of serum selenium with depression in hemodialysis subjects. They measured serum levels of selenium before the beginning of the hemodialysis session and adjusted multiple confounding factors, unlike other three included studies 18 .
In this study, no significant association was observed between selenium intake and the risk of depression. The included observational studies estimated selenium intake from foods as well as nutritional supplements. Therefore, it is possible that their results were confounded by the bioavailability of dietary or supplementary selenium. Cumulative evidence has proposed that selenium bioavailability is affected by the chemical form of selenium (organic or inorganic). Organic selenium is more bioavailable than inorganic selenium and also retains in tissues more 50 . Similarly, the effectiveness of inorganic supplements of selenium has been reported to be less than that of organic supplements 51 . Moreover, components such as heavy metals, fiber, lipids, dietary sulfur, and oxalate can have antagonistic effects on the bioavailability of dietary selenium 52,53 . Furthermore, selenium methionine and selenium cysteine were decreased during cooking processes 54 . Unfortunately, the included studies did not report any data on the bioaccessibility and bioavailability of selenium in diet or supplements. Future studies need to be focused on these issues.
The subgroup analysis revealed that high selenium intake was significantly associated with low risk of postpartum depression. Due to the placental transfer of selenium to the fetus, maternal serum selenium levels are reduced during pregnancy, especially in the 3 rd trimester. In addition, selenium is secreted in maternal breast milk as a component of selenoproteins. These processes increase the daily selenium requirement of pregnant and lactating women, which may result in selenium deficiency if not compensated properly 55 . It should be noted that supplementary selenium is more effective than dietary selenium in the improvement of low serum selenium levels 56 . In this meta-analysis, all studies conducted on postpartum depression considered supplementary, but not dietary, intake of selenium. This could contribute to the observed significant association between selenium intake and the risk of postpartum depression.
Interestingly, the present meta-analysis indicated that selenium supplementation significantly decreased depressive symptoms. Several mechanisms can explain this beneficial effect of selenium on depression. Selenium is known as a key regulatory factor of inflammatory and oxidative responses. Selenium deficiency can disrupt the function of multiple antioxidant enzymes such as glutathione peroxidase and thioredoxin reductases, which protect cells against oxidative damage 57,58 . Furthermore, inflammation is regarded as a part of depression pathogenesis 59 . Therefore, anti-inflammatory properties of selenium may help to improve depressive symptoms 60,61 . It is also possible that selenium affects depression symptoms through the modulation of neurotransmitter turnover as well as regulation of thyroid function 62,63 .
Several techniques have been suggested to determine serum concentration of selenium including atomic absorption spectrometry, molecular, atomic fluorescence spectrometry, inductively coupled plasma-mass spectrometry (ICP-MS) and graphite furnace atomic absorption spectrometry, flame atomic absorption, electrothermal atomic absorption spectrometry [64][65][66][67] . Atomic fluorescence spectrometry has higher sensitivity and is simpler than atomic absorption spectrometry. However, it has some detection limits 68 . Graphite furnace atomic absorption spectrometry is a selective, sensitive and easy method, however it is a single element technique 69,70 . Electrothermal atomic absorption spectrometry requires a small sample volume. This method is sophisticated and expensive 66,71 . Flame atomic absorption spectroscopy as a precise method, requires high sample preparation 72 . Serum or plasma selenium is assessed usually by two common methods of ICP-MS and atomic absorption spectrometry. ICP-MS is higher sensitivity than atomic spectrometry. It has multi-element capability, good stability and detects qualitative and quantitative trace element. However, this method is relatively expensive 73 . Atomic absorption spectrometry has a low detection limit. Although, this method is comparatively inexpensive, it is not an exclusive detection technique 72 . As a result, the different methodologies used to measure selenium in serum may be considered as one of the sources of heterogeneity. The method used to measure serum concentration of selenium in included studies are reported in Supplementary Table 1. Unfortunately, we could not evaluate the effect of this factor on the study findings. It has been proposed that serum concentration of selenium may be affected by sex and age 74 . The association between serum selenium level and gender is not clear. Some previous investigations indicated that there was no significant difference in serum selenium between males and females [75][76][77] . However, several studies reported serum concentration of selenium was related to gender 78,79 . Some studies reported that serum selenium concentration was higher in men compared with women 78,79 . In contrast, one study revealed that women had higher serum selenium in comparison with men 80 . It is possible that some factors including differences in sexual hormones, smoking and dietary habits play a role in relationship between gender to serum selenium level [80][81][82] . Also, the findings of studies regarding the effect of age on serum selenium concentration are inconsistent. According to the previous studies, no significant association between serum selenium and age was found 79,83 . However, this finding was not approved by some studies 84,85 . It seems that changes in body selenium distribution, dietary habits and hormonal status probably affect selenium concentration through different ages 82,[86][87][88] . For example, plasma estrogen is positively related to serum selenium. Therefore, change of estrogen status throughout the life cycle can influence serum selenium in women 89 . Moreover, a significant reduction in serum selenium has been reported in elderly individuals 78 . Accumulation of inflammatory factors, change in physiology conditions, inadequate intake of selenium-rich sources and inefficient absorption of dietary selenium are contributed in the relationship between the declined serum selenium level and aging [90][91][92] .
This study has several limitations. First, there were high levels of heterogeneity in all analyses except for the effect of selenium supplementation on depression scores. Second, due to the insufficient number of studies, we could not run subgroup analyses for all outcomes except for selenium intake and the relative risk of depression. www.nature.com/scientificreports/ Third, potential confounding factors were not adjusted in some of the included studies, which might affect the findings.
Strengths of the present study should also be considered. First, this study is the first meta-analysis that investigated the role of selenium in depressive disorders. Second, we conducted a comprehensive search using several databases to identify eligible studies. Third, we included both interventional and observational studies in this systematic review and meta-analysis to perform a comprehensive assessment regarding selenium and depression.

Conclusion
In conclusion, the findings of this systematic review and meta-analysis suggest that high selenium intake may have a protective role against postpartum depression. In addition, our findings support that supplementation with selenium can be effective in reducing depressive symptoms. Nevertheless, further studies are needed to draw definitive conclusions.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.