The Therapeutic Effect of Exogenous Melatonin on Depressive Symptoms: A Systematic Review and Meta-Analysis

Background Depression-related mortality and morbidity pose growing public health burdens worldwide. Although the therapeutic effect of exogenous melatonin on depression has been investigated, findings remain inconsistent. We conducted this systematic review and meta-analysis to clarify the effectiveness of melatonin in the treatment of depression, including primary and secondary depression symptoms. Methods We searched the online databases of PubMed, EMBASE, and the Cochrane Library for original studies published up to May 2021. We used STATA 14.0 software to synthesize the results of included studies. To evaluate the effectiveness of melatonin, we calculated the standardized mean differences (SMDs) and 95% confidence intervals (CIs) of depression scores between the melatonin and placebo groups. Results Our literature search returned 754 publications, among which 19 studies with 1,178 patients (715 women, 463 men; mean age: 56.77 years) met inclusion criteria. Melatonin dosages ranged from 2 to 25 mg per day; treatment durations were between 10 days and 3.5 years. Our synthesized results showed that melatonin was not found significantly beneficial for alleviating depressive symptoms (SMD = −0.17, 95% CI = [−0.38, 0.05]). Subgroup analysis demonstrated that the decrease in depression scores measured with the Beck Depression Inventory (BDI) was significant (SMD = −0.52, 95% CI = [−0.73, −0.31]). Conclusions There is very limited evidence for effects of melatonin on depression.


INTRODUCTION
Depression, one of the most common psychiatric disorders, contributes to morbidity and mortality in adults and adolescents worldwide (1). The World Health Organization reported that the number of people suffering from depression increased by more than 18% from 2005 to 2015 (2). Depression is considered as a heterogeneous disease on account of complex pathological mechanisms and multiple etiologies (1). Currently, the efficacy for depression therapy is limited due to the need for long-term treatment and variation in symptom presentation (2). Antidepressant medication lacks uniform effectiveness. The large treatment-refractory population indicates the need to explore additional therapeutic agents, such as novel antidepressive treatments that have fewer side effects (3,4). Nevertheless, successful treatment is believed to enable a dramatic improvement in patients' overall functioning and quality of life (5).
Melatonin, also known as 5-methoxy-N-acetyltryptamine, is a pleiotropic neurohormone secreted by the pineal gland at low levels during the daytime and high levels at night (6). Disorders of melatonin secretion were found to be historically associated with depression; however, when used as a therapy for this condition, exogenous melatonin is thought to function by normalizing patients' disrupted circadian architecture via binding to two receptors (MT 1 and MT 2 ) in the suprachiasmatic nucleus and other parts of the brain (7)(8)(9)(10).
At present, studies exploring the therapeutic impact of exogenous melatonin on depression treatment have shown inconsistent results (11)(12)(13)(14)(15)(16). In 2014, a systematic review and meta-analysis revealed that exogenous melatonin had no therapeutic or prophylactic effect on depression (17). Another systematic review involving three trials reported a non-significant effect of melatonin on treating depressive episodes (18). However, five newly published randomized controlled trials (RCTs) demonstrated the beneficial effects of melatonin in depression treatment (14,(19)(20)(21)(22). In these five studies, melatonin was treated for 12 weeks, and the participants were with moderate to severe depression. While the previously published studies that did not testify the effectiveness of melatonin involved participants with mild depression symptom. Here, we report a systematic review and meta-analysis and provide updated findings on the effectiveness of melatonin on depression treatment including primary and secondary depression symptoms.

METHODS
Our study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Supplementary Table S1 and Supplementary Box 1).

Inclusion and Exclusion Criteria
Original studies that met the following criteria were included in our meta-analysis: (1) study participants were adults (aged ≥ 18 years); (2) the effect of melatonin on remission of depression was investigated among participants with depressive symptoms; (3) RCTs or randomized crossover trials were performed; (4) the study involved the diagnosis and/or measurement of depression; (5) the severity of depression was rated by a selfor clinician-administered questionnaire; (6) the mean severity of depressive symptoms was compared between melatonin and placebo groups; (7) the dose and mode of administration were reported; and (8) the study was published in English.
Exclusion criteria were as follows: (1) melatonin was not given as an intervention; (2) no scores on depressive symptoms were reported in the melatonin and placebo groups; and (3) the effect of melatonin or placebo was not reported separately from other interventions.

Search Strategies and Study Screening
We searched PubMed, EMBASE, and the Cochrane Library for trials published up to May 15, 2021.
To improve the search strategy, we used restriction options in accordance with the population, intervention, comparison, and outcomes principle: study participants were selected from an adult population; the intervention agent was melatonin; the comparison agent was placebo; and the outcome was the relevant score for depressive symptoms as measured by investigators. Specific search parameters were as follows: (1) PubMed: clinical trial, RCT, humans, adults, English language; (2) EMBASE: controlled clinical trial, RCT, clinical study, humans, 18+ years, English language; (3) the Cochrane Library: no restrictions.
Two authors (CL and DaM) independently reviewed the literature to screen original studies by reading the title, abstract, and/or full text of each. References cited in the included studies were also screened for eligibility. All inconsistencies and disagreements were resolved through discussion with a third author (ML).

Data Collection
Two authors (CL and DaM) extracted data from the selected studies, such as the study characteristics, intervention, primary outcomes, measurement instruments, and adverse events. For studies that provided the mean and standard deviation (SD) of depression scores, we recorded summary data directly; otherwise, we calculated summary data using statistical approaches. Of studies in which the 95% confidence interval (CI) was reported, the SD was calculated using the following equation: In studies in which the median and interquartile range (IQR) were reported, the mean and SD were computed as follows (23):

Statistical Analysis
The meta-analysis was conducted with STATA 14.0 software (Stata Corp, College Station, TX, USA). The standardized mean difference (SMD) and 95% CI were synthesized to evaluate the difference in depression scores between the melatonin and placebo groups. Statistical heterogeneity was examined on the basis of Cochran's Q test and the I 2 statistic. When p < 0.10 and I 2 > 50%, original studies were regarded as demonstrating heterogeneity, and a random-effect model meta-analysis was undertaken; otherwise, a fixed-effect model meta-analysis was carried out. Subgroup analyses were also performed on the basis of depression scales, melatonin dosage, and/or treatment duration. To assess the robustness of the meta-analysis, sensitivity analysis was conducted by removing each original study one at a time. Publication bias was identified through funnel plot analysis along with Egger's regression asymmetry test. In terms of the quality of each study, the risk of bias was assessed via Review Manager Version 5.4 (Cochrane Collaboration, Oxford, UK), including seven domains with each domain containing three types of risk (23).

Characteristics of Included Studies
As shown in Table 1

Effect of Melatonin on Depressive Symptoms
As shown in Figure 2 and Table 2 Figure S1 and Table 2). However, no significant results were observed at other dosages, including 2, 2.5, 3, 5, 6, and 25 mg/day. As depicted in Supplementary Figure S2 and Table 2

Effect of Melatonin on Depressive Symptoms as Measured by BDI
Eight studies reported depression severity measured using the BDI. As displayed in Supplementary Figure S3A and Table 2 Figure S3C and Table 2).
The above data indicate that the most effective strategy was to treat melatonin for 12 weeks with a daily dose of 10 mg.

Effect of Melatonin on Depressive Symptoms as Measured by HADS-D
Four original studies used the HADS-D to evaluate depressive symptoms. As illustrated in Supplementary Figure S4 and Table 2 Figure S5). No significant heterogeneity was observed across the above studies.

Effect of Melatonin on Depressive Symptoms as Measured by CES-D
We synthesized two studies that measured depressive symptoms with the CES-D. As shown in Supplementary Figure S6 and Table 2

Effect of Melatonin on Depressive Symptoms as Measured by MDI
Two original studies used the MDI to evaluate depression severity. As listed in Supplementary Figure S7 and

Effect of Melatonin on Depressive Symptoms as Measured by GDS
We included two studies that reported depression severity with the GDS. As Supplementary Figure S8 and

Effect of Melatonin on Depressive Symptoms as Measured by HDRS-21, CSDD, or ATYP
Regarding other scales, only one study reported depressive symptoms with the HDRS-21, CSDD, or ATYP. As such, no meta-analysis was conducted.

Evaluation of Risk of Bias According to the Cochrane Risk of Bias Tool
The risk of bias among included studies is presented in Figure 3.
All studies had a low risk of bias on random sequence generation (i.e., selection bias) and blinding of participants and personnel (i.e., performance bias). The random assignment method and generation of random sequences were clarified. The application of blinding was described as well. Thirteen studies involved a low risk of bias on allocation concealment (i.e., selection bias); six other studies had an unclear risk. As for detection bias induced by a lack of blinding in outcome assessment, 13 studies had a low risk of bias while six had an unclear risk. Only one study had a high risk of attrition bias that might be caused by the uneven frequencies of participants lost to follow-up between the intervention and control groups. Eleven studies had a low bias on selective reporting (i.e., reporting bias) because these studies were carried out in accordance with the registered protocol; the other eight studies did not indicate the clinical trial registry in detail.

Publication Bias and Sensitivity Analysis
A funnel plot and Egger's test were used to confirm whether the meta-analysis results were influenced by publication bias. As presented in Figure 4, Supplementary Figures S9, S10, no significant bias emerged in meta-analysis of the combination of eight scales (P Egger ′ s = 0.278), the BDI (P Egger ′ s = 0.459), and the HADS-D (P Egger ′ s = 0.219). Sensitivity analyses showed no substantial changes in the combined results when each study was removed in turn, highlighting the stability of our metaanalysis (Supplementary Figures S11-S13). Tests for the CES-D, MDI, and GDS were not carried out due to the small number of included studies.

DISCUSSION
Our systematic review and meta-analysis demonstrated that melatonin was not significantly effective for alleviating depressive symptoms. Although statistically positive results were observed at the dosages of 10 and 20 mg/day, there was very limited evidence for effects of melatonin on depression. Eight scales were involved in the measurement of depressive symptoms, among which the effect of melatonin was significant when the BDI was used. However, no significant effects were identified when depressive symptoms were measured with the HADS-D, CES-D, MDI, or GDS. Depressive disorders are characterized by psychological, behavioral, and physiological changes, affecting 17-20% of the population globally (36). Circadian rhythm disturbance is a major feature of depressive symptoms: 70% of patients with severe depression suffer from sleep disturbances (37,38). The association between depression and desynchronized circadian rhythms suggests that agents improving circadian rhythms may contribute to relieving depression. Melatonin, an endogenous hormone mainly synthesized in the pineal gland, plays a key role in synchronizing circadian rhythms. By binding to two receptors (MT 1 and MT 2 ) in the suprachiasmatic nucleus, melatonin regulates the expression of clock genes, resulting in a chronobiotic effect (26,37,39). Patients with depression have shown reduced peak values of plasma melatonin at night with a disordered rhythm phase (40). Exogenous melatonin also has an effect on synchronizing rhythms; it is thus thought to hold promise for treating depressive symptoms (38,(41)(42)(43). Furthermore, sleep disorders and biological rhythm abnormalities among individuals with depression are generally affected by sex. Women are twice as likely as men to suffer from depression due to the influence of gonadal hormones, which are involved in sleep regulation and human neurotransmitter regulation (44).
No biomarkers have yet been identified for the evaluation of depression; rather, some scales and questionnaires have been developed. The BDI is commonly used and consists of 21 items to assess cognitive, affective, behavioral, and somatic symptoms of depression (45,46). And studies found that exogenous melatonin could lead to modulation in specific physiological functions like control of human mood, behavior, cognition, and sleep regulation by activating membrane MT 1 /MT 2 melatonin receptors (47)(48)(49). In practice, the BDI is used not only to assess depression in psychiatric patients but also to screen for depressive symptoms in various patient groups (50,51), and most participants in this meta-analysis suffered secondary depression from different diseases. Moreover, the BDI provides information on a relatively wide range of levels of depression, and is typically considered a standard tool for depression measurement with high accuracy and efficiency (45,52). In terms of other scales, the HADS-D consists of seven questions and is often used with patients with moderate depression (53,54). The HADS-D is designed to evaluate the anxiety and depressive symptoms of patients without physical symptoms (e.g., sleep disorders) (55). However, this scale provides less information than others (45). By contrast, the 10-item MDI is more accurate for assessing major (i.e., moderate to severe) depression (56). The CES-D, composed of 20 items, is also used in the general population and primary healthcare institutions (57). Yet, the CES-D is considered more appropriate for evaluating general distress than depression (58). The GDS is a self-rated scale used primarily to assess depression in older adults (35). In our sample, the effect of melatonin was found statistically positive when using the BDI, but not significant when using the other four depression scales (i.e., HADS-D, CES-D, MDI, and GDS).
As a medical sleep aid, a single dose of 1-10 mg of melatonin has is considered standard, but the optimal dose for depression is not yet known (59). Based on our findings, the dose of 10 mg/day might work more efficiently. However, no significant results were observed at other dosages, including 2, 2.5, 3, 5, 6, and 25 mg/day. In addition, studies have reported that 10 mg per day is a balanced option considering that melatonin is a safe, nontoxic drug with minor adverse events compared to placebo in long-term treatment (15,16). With regard to treatment duration, no consistent program has been established for psychiatric use (60). Our subgroup analysis implied that 12-week treatment was more effective in alleviating depression, in line with the duration for melatonin receptor agonists agomelatine and ramelteon (61)(62)(63). However, no significant results were observed after 1.5, 2, 3, 4, 5, 6, 8, and 48 weeks treatment. This indicated that there is very limited evidence for effects of melatonin on depression.
No obvious side effects of melatonin were found in all original studies included in this systematic review. Some studies also showed that melatonin had no side effects in the short term and had mild side effects in the long term (59,64).

Limitations
Some limitations should be noted in our study. First, eight scales were used to measure depressive symptoms, which led to variation. Differences in these scales, as well as in participants' demographics and comorbidity, likely resulted in inconsistent outcomes between studies. Second, the number of studies included in our meta-analysis was small and might restrict the generalizability of our findings. Differences in melatonin dosage, treatment duration, and baseline treatment influenced the consolidation of our results. Finally, on the basis of studies included in meta-analysis, we cannot identify the specific type of depression on which melatonin may be effective than others.

CONCLUSIONS
In conclusion, no sufficient results were obtained to evidence the effects of melatonin on depression. Melatonin might have no potential therapeutic effect on depression.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
MN, YZ, and WZ supervised and designed this study. CL, ML, and DaM participated in the design and planning process. CL and DaM extracted and analyzed data and wrote the first draft of the manuscript. ML, TW, XZ, YH, and DaM checked the extracted and analyzed data. ML, EA, MN, YZ, and WZ revised the manuscript. All authors had access to all data in the study and are responsible for the integrity of the data and the accuracy of the analysis. All authors contributed to article and approved the final manuscript.

ACKNOWLEDGMENTS
We thank all authors of the included studies of this metaanalysis. The authors also wish to thank Dr. Jun Wen from Edith Cowan University for his insightful comments and professional proofreading to help improve the quality of this paper. Supplementary Table S1 | Checklist of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). Y, the item was reported in article; N, the item was not reported.