Melatonin Application in Assisted Reproductive Technology: A Systematic Review and Meta-Analysis of Randomized Trials

Objective: To study whether melatonin treatment can increase clinical pregnancy rate and live birth rate in assisted reproductive technology (ART) cycles. Methods: Literature searches were conducted to retrieve randomized trials that reported the effect of melatonin treatment on ART outcomes. Databases searched included PubMed, EMBASE, Cochrane Library, Web of Science, and Google Scholar. Results: Ten studies matched the inclusion criteria. Clinical pregnancy was reported in all of the included studies and live birth was reported in three studies. Melatonin treatment significantly increased the clinical pregnancy rate [OR = 1.43 (1.11, 1.86), power = 0.98, 10 RCTs, low-quality evidence] but not the live birth rate [OR = 1.38 (0.78, 2.46), power = 0.34, 3 RCTs, low-quality evidence]. Melatonin treatment increased the number of oocyte collected [SMD = 0.34 (0.01, 0.67), 7 RCTs, low-quality evidence], the number of maturated oocyte [SMD = 0.56 (0.27, 0.85), 7 RCTs, low-quality evidence], and the number of good quality embryo [MD = 0.36 (0.18, 0.55), 3 RCTs, low-quality evidence]. Melatonin treatment significantly increased the biochemical pregnancy rate [OR = 1.65 (1.14, 2.38), 6 RCTs, low-quality evidence] and had no significant effect on the miscarriage rate [OR = 1.28 (0.65, 2.51), 5 RCTs, low-quality evidence]. Conclusion: Melatonin treatment significantly increases the clinical pregnancy rate but not live birth rate in ART cycles. Melatonin treatment also increases the number of oocyte collected, maturated oocyte, and good quality embryo. No clear evidence suggested that melatonin treatment increased the adverse events in ART cycles. The actual findings may be compromised due to the wide heterogeneity of the included IVF patients, from PCOS to low ovarian reserve.


INTRODUCTION
Infertility is defined as the inability to conceive for at least 1 year (1). Infertility is a common condition affecting 10-20% of women at the reproductive age across different countries (2,3). Assisted reproductive technology (ART) has allowed millions of infertile couples in the world to conceive successfully since 1978 (4). Although substantial advance has been made in the past decades, the chance of achieving a live birth through ART is not high (5,6). Several strategies aiming to increase the live birth rate are currently being used, such as endometrial scratching (7), assisted hatching of human embryos (8), the addition of drugs to improve successful rate (9,10), etc.
The successful rate of ART cycles can be affected by several internal as well as external factors (11,12). Oxidative stress is a state characterized by an imbalance between pro-oxidant molecules and antioxidant defenses and it plays a vital role in the pathogenesis of female infertility (13,14). Oxidative stress causes toxic effects on oocyte maturation and is considered one of the causes of poor oocyte quality (15). Increased oxidative stress can lead to reduced oocyte maturation rate and fertilization rate, which will result in a reduced possibility of full-term pregnancy (15). Melatonin is the main hormone derived from the pineal gland. Other extrapineal organs, like the gastrointestinal tract and female ovary, can also secret this hormone (16). Melatonin has been reported to regulate several physiological processes, including circadian rhythms, endoplasmic reticulum stress response, apoptosis and autophagy, and mitochondrial homeostasis (16,17). Melatonin and its metabolites also protect cells from oxidative stress by acting as a free radical scavenger (17).
In the last decade, many studies, including randomized trials, have reported the application of melatonin in ART cycles. However, the sample size is small in most of these studies and the evidence generated is of low quality. Additionally, the results in these studies are controversial and no definite conclusion has been made currently. In this systematic review, we aim to perform an in-depth overview to evaluate the effects of melatonin application in ART cycles.

Inclusion Criteria
We included randomized trial studies that investigated melatonin application in ART cycles, including in vivo treatment and in vitro application for oocyte or embryo culture. Studies not written in English were excluded. Reviews, conference abstracts, case report studies, and study protocols were also excluded.

Literature Search
Two authors (K-LH and XY) independently searched the database of PubMed, EMBASE, Cochrane Library, Web of Science, and Google Scholar from January 1978 to November 2019. The PICO search method was used to collect related literature. Patients were those who underwent ART cycles, including intrauterine insemination (IUI), in vitro fertilization (IVF) and/or Intra-Cytoplasmic Sperm Injection (ICSI) and subsequent embryo transfer (ET). Interventions included melatonin or its analog treatment with or without any adjuvant treatment. The comparators were treatments without melatonin or placebo. Outcomes included clinical pregnancy, live birth rate, oocyte and embryo quality, and miscarriage. And the study design was randomized trials. The key search terms included but not limited to "melatonin, " "assisted reproductive technology, " "in vitro fertilization, " "live birth, " "oocyte quality, " "randomized trials." The detailed search terms and methods could be seen in Supplemental Table 1.

Study Selection
Two authors (K-LH and XY) independently scrutinized all of the titles and abstracts according to the predefined inclusion criteria. The full manuscripts of the studies were obtained if the titles and abstracts were considered to be relevant for inclusion. Any disagreement between the two authors was resolved by a third review author (DZ). References of all included studies judgments to identify relevant articles not captured by the electronic searches.

Data Extraction
Two authors (K-LH and XY) independently extracted data from included trials. Any disagreements were solved by consulting another author (DZ). In cases we identified a study with multiple publications, the main trial report was used as the reference and additional details were supplemented from other papers. The data extracted from the eligible studies included the sample size, publication year, time frame, country, inclusion and exclusion criteria, diagnosis of participants, the protocol for ART, the definition of outcomes, and the data of outcomes.

Study Quality Assessment and Publication Bias
Two reviewers (K-LH and SW) independently conducted the quality assessment of the included studies. To evaluate the risk of bias, we followed the Cochrane Collaboration's criteria (version 5.1.0, Available from www.cochrane-handbook.org) for judging the risk of bias and the studies were classified as being of low, high, or unclear risk of bias. Funnel plot was used to assess the publication bias.

Statistical Analysis
The Review Manager version 5 was used to merge and analyze the extracted data. Forest plots were created for each outcome. The results were combined for meta-analysis using the Mantel/Haenszel model. A fixed-effect model was used where no statistically significant heterogeneity is present (I² < 50%). When substantial heterogeneity was observed (I² > 50%), we would address it by rechecking data and excluding studies with a high risk of bias for sensitivity analysis. If substantial heterogeneity persisted, a random-effect model was used. The discontinuous results were shown by odds ratio (OR) with a 95% confidence interval (CI). The continuous results were shown by the difference in means (MD) with 95% CI for a fixed-effect model or standard mean difference (SMD) with 95% CI for a random-effect model. Statistical significance is set at a P level of 0.05. The statistical power of the meta-analysis was conducted for the main outcomes.

Outcomes and Additional Analysis
The primary outcomes included clinical pregnancy rate and live birth rate. Secondary outcomes included oocyte retrieval number, the number of the maturated oocyte (MII), the number of the top quality embryo, biochemical pregnancy rate, miscarriage rate, and adverse event. Subgroup analysis was conducted for different protocols of melatonin application and only clinical pregnancy rate was addressed. In cases that there were multiple doses of melatonin treatment in a single study, the data of all the treatment groups were first merged and then was considered as one intervention group.

Characteristics of the Included Studies
The PRISMA flow diagram of the review process was presented in Figure 1. A total of 10 randomized trials were included for analysis (18)(19)(20)(21)(22)(23)(24)(25)(26)(27). Characteristics of the included studies were presented in Tables 1, 2. The included studies varied in publication date from 2010 to 2019. A total of 1,203 participants were included for analysis. Six hundred and forty five were allocated to the melatonin treatment group and 558 were allocated to the control group. The funnel plot showed no publication bias for the included studies (Supplemental Figure 1). The qualitative analysis of the included studies could be found in Supplemental Figure 2. Each risk of bias item presented as percentages across all included studies could be found in Supplemental Figure 3. Most studies were at high risk of bias and only two studies were of good quality (22,24). From these included studies, sample sizes varied from 30 women to 331 women. Nine were in-vivo studies and 1 was the in-vitro application of melatonin. Three studies focused on women with polycystic ovarian syndrome (PCOS); the patients in 3 studies were none special; the other 4 studies focused on women with unexplained infertility, diminished ovarian reserve, poor oocyte quality, and sleep disturbances, respectively. One study included women undergoing IUI and 9 studies focused on women undergoing in vitro fertilization (IVF) and/or Intra-Cytoplasmic Sperm Injection ICSI and subsequent embryo transfer ET. The dose of 3 mg melatonin was most commonly used in these studies. One study used 10 µmol/l in the culture medium for embryo culture (26). One study used 2/4/8 mg melatonin for intervention in three parallel groups (24). Three studies used 3 mg melatonin together with myo-inositol and folic acid (18,20,21). Other detailed characteristics of these studies could be found in Tables 1, 2.

Clinical Pregnancy per Allocated Woman
Ten studies reported the effect of melatonin on clinical pregnancy (Figure 2). Meta-analysis suggested that melatonin treatment significantly increased clinical pregnancy rate [OR = 1.43 (1.11, 1.86), P < 0.01, power = 0.98]. The studies included for metaanalysis had low heterogeneity with an I 2 value of 0%.

Live Birth per Allocated Woman
Three studies reported the effect of melatonin on live birth (Figure 3). Meta-analysis suggested that melatonin treatment did not increase the live birth rate [OR = 1.38 (0.78, 2.46), P > 0.05, power = 0.34]. The studies included for meta-analysis had low heterogeneity with an I 2 value of 0%.

Secondary Outcome
The Average Number of Oocyte Retrieved per Allocated Woman The data of the number of oocytes collected were available to be extracted and synthesized in 7 studies (Figure 4). Meta-analysis suggested that melatonin treatment significantly increased the number of oocyte collected [SMD = 0.34 (0.01, 0.67), P < 0.05, random-effect].

The Average Number of Maturated Oocyte per Allocated Woman
The data of comparison of the number of maturated oocyte were available to be extracted and synthesized in 7 studies (Figure 5). Meta-analysis suggested that melatonin treatment significantly increased the number of maturated oocyte [SMD = 0.56 (0.27, 0.85), P = 0.0001, random-effect].

The Average Number of Good Quality Embryo per Allocated Woman
The data of comparison of the number of good quality embryos were available to be extracted and synthesized in 3 studies (Figure 6). Meta-analysis suggested that melatonin treatment significantly increased the number of good quality embryo [MD = 0.36 (0.18, 0.55), P = 0.0001]. The studies included for metaanalysis had low heterogeneity with an I 2 value of 19%.

Biochemical Pregnancy per Allocated Woman
Six studies reported the effect of melatonin on biochemical pregnancy (Figure 7). Meta-analysis suggested that melatonin treatment significantly increased the biochemical pregnancy rate [OR = 1.65 (1.14, 2.38), P < 0.01]. The studies included for meta-analysis had low heterogeneity with an I 2 value of 0%.

Miscarriage per Allocated Woman
The data of the rate of miscarriage were available to be extracted and synthesized in 5 studies (Figure 8). Metaanalysis suggested that melatonin treatment significantly had no significant effect on the miscarriage rate [OR = 1.28 (0.65, 2.51), P > 0.05]. The studies included for meta-analysis had low heterogeneity with an I 2 value of 0%.

Adverse Events
One study reported that one woman treated with melatonin had a term live birth of a baby with an absent right kidney. Two patients were diagnosed with pre-eclampsia. One patient was diagnosed with placenta previa (24). Other potential adverse events, like ectopic pregnancy and ovarian hyperstimulation syndrome, were not different between the melatonin treatment group and the control group, or not reported.

Clinical Pregnancy per Allocated Woman for Melatonin vs. Control Studies
Six in-vivo studies reported the effect of melatonin vs. control on clinical pregnancy (19,(22)(23)(24)(25)27). Meta-analysis suggested that melatonin treatment significantly increased clinical pregnancy rate [OR = 1.55 (1.02, 2.35), P < 0.05]. The studies included for meta-analysis had low heterogeneity with an I 2 value of 0%.

Clinical Pregnancy per Allocated Woman for Myo-inositol Plus Melatonin vs. Myo-inositol Studies
Two in-vivo studies reported the effect of included plus melatonin vs. myo-inositol on clinical pregnancy (20,21). Meta-analysis suggested insignificant difference between the two groups [OR

DISCUSSION
This systematic review includes 10 randomized control studies for quantitative analysis and the result indicates that melatonin treatment significantly increases the clinical pregnancy rate in ART cycles, and the effect is probably mediated by increasing the quality of oocytes and embryos. Melatonin treatment has no significant effect on the live birth rate and the result needs to be confirmed by future studies with a large sample size. Melatonin has been reported to regulate several physiological processes, including circadian rhythms, endoplasmic reticulum stress response, apoptosis and autophagy, and mitochondrial homeostasis (16,17). Melatonin and its metabolites also protect cells from oxidative stress by acting as a free radical scavenger that is able to deactivate a variety of reactive oxygen species (17,28,29). Accumulating studies indicate that increased oxidative stress in the peritoneal, serum, and follicular microenvironments can result in poor oocyte quality and compromise the reproductive potential of women (30)(31)(32). It is likely that reducing the reactive oxygen species in the microenvironment can protect the oocyte and the embryo from oxidative stress. Indeed, recent studies suggest that melatonin concentration in follicular fluid is associated with oocyte maturation rate and good quality embryo rate in women undergoing ART procedures (33). Additionally, melatonin treatment in females for 3 mg per day or higher doses can significantly increase the serum and follicle concentration of melatonin (22,24,27). Therefore, it is reasonable that melatonin treatment can increase oocyte and embryo quality and subsequent pregnancy outcomes. It should be noted that melatonin concentration dynamically changes within a day. In humans, melatonin secretion begins since nightfall, reaches a peak level in the middle of the night and decreases gradually during the second half of the night (34). Therefore, it is necessary to collect the samples at a specific time in all participants.
A previous prospective, longitudinal, cohort study indicates that myo-inositol plus melatonin treatment significantly improves oocyte quality in women who failed to conceive in previous in vitro fertilization cycles due to poor oocyte quality (35). Additionally, the number of maturated oocytes and the number top-quality embryos transferred were significantly higher than the previous IVF cycle (35). In agreement, one randomized trial including women with poor oocyte quality history shows that melatonin together with myo-inositol and folic acid treatment produces more maturated oocytes and top quality embryos than myo-inositol and folic acid treatment (21). However, the clinical pregnancy rate is not significantly increased in the melatonin together with myo-inositol and folic acid treatment group (21). The negative result of the clinical pregnancy rate in this study can be attributed to the small sample size. Consistent with these studies, our metaanalysis also suggests that melatonin treatment significantly increases the number of maturated oocytes and the number of good quality embryos. A previous meta-analysis shows that melatonin treatment does not significantly increase clinical pregnancy in women undergoing ART (36). This study includes only five randomized trials and the sample size is not large. It is necessary to update the data because recent randomized studies show improved pregnancy outcomes after melatonin treatment in ART cycles. Furthermore, our study also shows that the number of oocytes retrieved is significantly increased, although the significance is not detected most randomized trials. The application of melatonin in patients with PCOS is promising according to our study, and future studies addressing the role of melatonin in these patients will be interesting.
Our study shows that the miscarriage rate is not different between the melatonin treatment group and the control group. Several obstetric complications are reported in women receiving melatonin treatment (24). Although it does not necessarily mean that melatonin treatment increases the rate of obstetric complications because only one patient for each complication is reported, future studies with a large sample size should better report the data and provide robust evidence.
The main limitation of this study is the low quality of most studies included in the meta-analysis. Additionally, a limited number of cases are included in most individual studies and the data of live birth is reported in only 3 studies. Even we combined the data of these studies, we did not found the improvement of live birth rate after the treatment of melatonin. The negative result may result from the small sample size and future studies address the effect of melatonin in ART cycles should better report this outcome and provide more robust evidence. The wide heterogeneity of the included IVF patients, from PCOS to low ovarian reserve, may compromise the actual findings in this study.

CONCLUSION
This systematic review suggests that melatonin treatment significantly increases the clinical pregnancy rate in ART cycles. Melatonin treatment also increases the number of oocyte collected, maturated oocyte, and good quality embryo. No clear evidence suggested that melatonin treatment increased the adverse events in ART cycles. Melatonin treatment has no significant effect on the live birth rate and the result needs to be confirmed by future studies with a large sample size. The actual findings may be compromised due to the wide heterogeneity of the included IVF patients, from PCOS to low ovarian reserve.

AUTHOR CONTRIBUTIONS
K-LH and XY reviewed the literature and extracted the data. K-LH and SW assessed the quality of included studies. K-LH designed the study, wrote the manuscript, and designed the figures and tables. DZ provided some key ideas for this manuscript. All authors participated in the discussion of analysis and interpretation of data in this article.