Three databases were systematically searched for related studies (PubMed, Scopus, and Web of Science). Two reviewers (NDE and AS) searched the databases for records that were published from January 1, 1970, until September 9, 2022, using “melatonin”, “testicular histopathology”, and their related keywords. Thereafter manual citation searching was done based on reference lists of the included studies. The search strategy was not restricted based on the language and is provided in Supplementary material 1 .

To begin, the electronic removal of duplicate records was carried out. Subsequently, two reviewers (NDE and AS) independently utilized the Rayyan online tool for managing systematic reviews to screen the records by titles and abstracts (22). Then, full-text screening of the relevant records was conducted to meet the eligibility criteria. Disagreements were resolved through discussions. Studies that satisfied the following requirements were included: (1) controlled animal studies, (2) at least one intervention group receiving a melatonin regimen, (3) rodents exposed to metabolic disorders (such as diabetes, obesity, dyslipidemia, and hypo- and hyperthyroidism) resulting in oxidative stress to testicular tissue, (4) reported hallmarks of testicular tissue (such as histopathologic, biochemical, and sperm analyses), and (5) at least one control group with comparable stress.

Exclusion criteria consisted of the following: (1) non-rodent animals exposed to other types of stresses (such as physical, chemotherapy, toxins, and heavy metals), (2) in-vitro and ex-vivo studies, (3) combination therapy using melatonin with other drugs, (4) failed to report relevant outcomes, (5) healthy controls without metabolic disorders-induced stress, and (6) treatment involving derivatives of melatonin. Additionally, reviews, letters, human trials, and non-peer-reviewed publications were excluded from the review.

Two independent reviewers (NDE and FE) extracted the data into Excel spreadsheets. Controversies were resolved by the involvement of a third author (AS). One reviewer (NDE) cleaned the data to become suitable for analyses. The data were extracted using the following variables: (1) study characteristics (first author and publication year), (2) rodent characteristics (age, species, type of stress, and sample size), (3) melatonin (dosing, duration of treatment, timing, and route of administration), (4) tissue, plasma, and histopathological indices, and (5) time of outcome assessment.

The risk of bias in the studies was evaluated by utilization of the SYRCLE tool for animal intervention studies (23). Each article was independently reviewed by two reviewers (NDE and FE), who classified them as high, low, or unclear for each bias domain. In case of disagreements, a consensus was reached by the involvement of a third reviewer (AS).

Meta-analyses were done by a single reviewer (AS) using Stata MP Version 16 (StataCorp, College Station, TX, USA). Two (NDE and MA) independent reviewers supervised the meta-analyses. Under the random effect model (DerSimonian-Laird), Standardized Mean Difference (SMD, Cohen’s d) was utilized as the effect measure and a less than 0.05 p-value was considered statistically significant. The residual between-study heterogeneity was examined using I² and p-value. An I² value of less than 25% was considered low heterogeneity, between 25% and 50% was considered moderate heterogeneity, and greater than 50% was considered substantial heterogeneity (24). For missing data, reviewers contacted the authors via available emails and waited at least one month to receive any responses. If missing data were crucial, the study was removed from the analysis. In the case of common control arms, intervention arms were combined using Cochrane’s formula to avoid overcalculations (25). When minimum, first quartile, median, third quartile, and maximum were the only available statistics, mean and standard deviation was estimated utilizing methods outlined in prior research (26, 27).

To explore the sources of heterogeneity, subgroup analyses were done. Subgroup analyses were conducted if at least two separate study arms were available for each subgroup. Also, for outcomes with more than ten observations, publication bias was evaluated using funnel plots and Egger’s regression test.

Figure 1 summarizes the search results. The systematic database searching resulted in records 10,039 (PubMed: 1,375, Scopus: 4,826, and Web of Science: 3,838) and manual citation searching had three additional results. Using automatic duplicate detection techniques, 1,016 records were removed initially. 9,026 records were screened based on their titles and abstracts which resulted in 46 potentially eligible studies, all of which were successfully retrieved for full-texts. 22 articles were excluded at this stage due to ineligible outcome (n= 2), population (n= 19), and intervention (n= 1). Finally, a total of 24 articles were included in the systematic review and meta-analysis.

The included studies were published between 2005 and 2022 and all of them were in English. The rodents included in the studies comprised mice (n=3) (28–30) and rats (n=21) (17, 18, 31–49). Across the included studies, rodents were subjected to induction of diabetes (n=11) (18, 28, 31, 34, 35, 38–41, 46, 48), obesity (n=8) (17, 30, 32, 33, 36, 37, 42, 49), hyperthyroidism (n=3) (43, 44, 47) and hypothyroidism (n=1) (44), hyperlipidemia (n=1) (29), and food deprivation (n=1) (45). To induce diabetes, streptozotocin (18, 28, 34, 35, 38–40, 46, 48) and alloxan (31, 41) were used. To induce obesity, high-fat diet (17, 29, 30, 37, 42, 49), high-fructose diet (36), and leptin (32, 33) were used. To induce hyper- and hypothyroidism, levothyroxine (43, 44, 47) and carbimazole (44) were employed, respectively. To induce hyperlipidemia and food deprivation, high-fat diet (17, 29, 30, 37, 49) and one-third of the normal daily food ration (45) were used, respectively. Melatonin was administered orally (n=14) (17, 18, 28–30, 32–34, 36, 39, 41, 44, 46, 49) and intraperitoneally (n=10) (31, 35, 37, 38, 40, 42, 43, 45, 47, 48). All of the studies have administered melatonin after the induction of injuries. Table 1 presents the study characteristics. The detailed characteristics of the injury induction and melatonin therapy are presented in Supplementary Material 2 .

31 outcomes were pooled from the included studies. These outcomes included histopathologic characteristics (abnormal sperm morphology, apoptotic cells, apoptotic index, Johnsen’s testicular biopsy score, seminiferous epithelial height, tubular diameter, tubular basement membrane thickness, sperm count, and motility), weights (absolute epididymis, testis, and body weight, body weight gain, epididymal adipose tissue weight, and relative testis to body weight), hormonal characteristics (androgen receptor expression, serum FSH, LH, testosterone levels), markers of oxidative stress (tissue and serum GPx, andMDA activity, tissue CAT, GSH, and SOD activity), and exploratory outcomes (serum HDL, LDL, total cholesterol, triglyceride, and blood glucose level). The pooled effect sizes are summarized in Figure 2 .

For each domain of the SYRCLE tool for assessment of the risk of bias in animal studies, studies scored 1 if were low risk. The studies’ overall scores were between 2 and 4. Studies suffered from unclear risk of bias in selection, performance, and detection bias. On the other hand, studies were mostly low risk for bias in attrition and reporting domains. Detailed results of the risk of bias assessment are presented in Figure 8 and Supplementary material 3 .

Except for outcomes with a few available studies, in which it is expected that pooled effect size shows low robustness, the leave-one-out meta-analysis demonstrates that the overall effect sizes closely align with the vertical line representing the overall effect size. Additionally, the confidence interval lines of these effect sizes intersect with the vertical red line encompassing all the studies, indicating that no single study has a significant impact on the results of our meta-analysis. The results of the leave-one-out meta-analyses are presented in Supplementary material 4 .

Using funnel plots and Egger’s regression test, publication bias was assessed and the results are provided in Supplementary material 5 . Four outcomes were eligible for assessment: absolute testis and body weights, serum testosterone level, and tissue MDA activity, of which, the two formers were asymmetrically scattered on the funnel plots, suggesting the possibility of the small-study effect. Although, due to the presence of high heterogeneity, interpretations should be made with caution.

To explore the heterogeneity, subgroup analyses were done. The results are shown in Table 2 . 17 studies were eligible for subgroup analyses based on the mechanism of injury induction, melatonin cumulative dose, route of melatonin administration, type of injury, and rodent species. For each outcome, studies were categorized based on the melatonin cumulative dose pertaining to the median value of cumulative doses in that outcome.

By conducting subgroup analyses, statistically significant between-group differences were observed in the following outcomes, suggesting possible sources of heterogeneity: serum FSH, testosterone, and LH levels, body weight, sperm count, serum and tissue MDA activity, tissue GSH, SOD, and blood sugar levels.

Awad et al. reported that the seminal level of melatonin is lower than its serum concentration in males with fertile normozoospermia, oligo asthenozoospermia, and non-obstructive azoospermia (50). They observed that there is a significant and positive correlation between the serum level and seminal levels of melatonin (50). Furthermore, all infertile groups had lower serum and seminal levels of melatonin compared with fertile normozoospermic men, and there were strong correlations between sperm motility and serum and seminal levels of melatonin (50). Consistently, a Chinese cross-sectional study with 970 outpatients from a reproductive medicine center reported that men with poor sleep quality have lower total motility, progressive motility, concentration, total count, and normal sperm morphology (51). Interestingly, a recent clinical trial reported that compared with the placebo group, the use of melatonin for 3 months after sub-inguinal varicocelectomy significantly improved several parameters of semen analysis, including sperm concentration, motility, and normal morphology, showing the protective effects of melatonin (52).

Metabolic diseases, such as diabetes, can distort the germinal layers, degenerate seminiferous tubules and interstitial tissues, decrease the diameter of seminiferous tubules, and reduce germinal epithelium height (53–55). Based on our findings, melatonin improved histopathological tissue score and increased seminiferous epithelial height and tubular diameter. Efficient spermatogenesis relies on the complex interaction between Sertoli cells and germ cells in seminiferous tubules (56); therefore, by preserving the normal structure of seminiferous tubules in metabolic diseases, melatonin can potentially provide the prerequisites for spermatogenesis.

Most of the previous studies reported that melatonin activates the HPG axis, and increases the serum level of gonadotropin-releasing hormone (GnRH), LH, FSH, and testosterone (17); however, some studies reported the opposite. For instance, Oliveira et al. reported that melatonin downregulates neurokinin B and kisspeptin, two stimulators of GnRH release, in male diabetic rats (57, 58). They claimed that pinealectomy can upregulate GnRh and potentiate the HPG axis in adult rats (59). On the other hand, melatonin is a metabolic regulator of Leydig cells and protects them against harmful stimuli such as metabolic stress (28, 60). Melatonin can prevent apoptosis of Leydig cells and promote the expression of testosterone synthesis-related genes such as steroidogenic acute regulatory protein (StAR), transcription factor GATA-4 (Gata4), and steroidogenic factor 1 (SF1) in Leydig cells (61); thus, melatonin can restore testosterone production by Leydig cells in stress conditions. The pooled data from all studies revealed that treatment with melatonin did not significantly alter the serum levels of testosterone, LH, and FSH in rodents with metabolic diseases, but subgroup group analysis revealed that the cumulative dose of melatonin and the type of metabolic disease should be considered when interpreting the effect of melatonin on sex hormones. Herein, higher cumulative doses (>210 mg/kg) of melatonin markedly increased the serum levels of LH and FSH in rodents with metabolic diseases, while lower doses of melatonin were less effective. Similarly, melatonin more effectively increased the serum levels of testosterone in diabetic rodents compared with obese rodents.

Previously, several studies have revealed that metabolic diseases such as obesity, diabetes, dyslipidemia, hypothyroidism, and hyperthyroidism can lead to excessive oxidative stress in the testes, while melatonin can vigorously potentiate endogenous anti-oxidant defense by upregulating SOD, GSH, and GPx (30, 37, 49, 52, 62, 63). Melatonin can increase the testicular expression of nuclear factor erythroid 2–related factor 2 (Nrf2), which is a main transcription factor for many anti-oxidant genes and strengthens the anti-oxidant defense (64). Our meta-analysis showed that melatonin can markedly decrease the serum and tissue levels of MDA and significantly upregulate GSH; however, results were non-significant for other anti-oxidants such as SOD, CAT, and GPx. However, subgroup analysis revealed that melatonin could significantly upregulate them in some subgroups of rodents. For example, melatonin significantly increased SOD activity in alloxan-treated rodents compared with high-fat diet-fed rodents. In addition to oxidative stress, melatonin significantly mitigated the inflammatory response and suppressed the expression of inflammatory cytokines, such as tumor necrosis factor α (TNF-α) and interleukin 1β (IL1β), in the testicular tissue of high-fat diet rats (42, 62).

Moreover, melatonin can regulate the metabolism of Sertoli cells (65). Herein, it has been shown that melatonin can increase glucose uptake of Sertoli cells by upregulating glucose transporter 1 (GLUT1) and simultaneously enhance their lactate production ability (65). As lactate is a source of energy for germ cells and possesses anti-apoptotic properties, it can immensely improve male infertility (65). Besides, it was observed that the high glycemic status in diabetic or hyperlipidemic mice can induce severe endoplasmic reticulum (ER) stress and apoptosis in Sertoli cells, while treatment with melatonin reverses these alterations (28, 29). Similarly, using a mice model of palmitic acid-induced lipotoxicity, Xu et al. unveiled that melatonin attenuates ER stress, mitochondrial dysfunction, oxidative stress, and apoptosis in spermatogonial stem cells and enhances sirtuin 1 (SIRT1)-mediated p53 deacetylation, which allows G2/M transition and allows cell cycle progression (52). Metabolic perturbation suppressed AMPK/SIRT1 in the testicular tissue of diabetic mice and high-fat diet-fed mice, while treatment with melatonin markedly enhanced the AMP-activated protein kinase (AMPK)/SIRT1 pathway in their testis (30, 64). Consistently, SIRT1 knockdown abolished the protective effects of melatonin on the detrimental effects of high glucose concentration on Leydig cells (64). Previously, it has been demonstrated that SIRT1 is a master regulator of metabolism and pivotal for spermatogenesis and spermatozoa differentiation (66). Mice with SIRT1 deletion had smaller testis, reduced number of spermatozoa, higher percentages of abnormal spermatozoa, and decreased fertility (66). In addition, melatonin treatment was shown to ameliorate impaired autophagy, mitophagy, and mitochondrial biogenesis in Leydig cells exposed to high glucose concentrations (64).

Our meta-analysis demonstrated that melatonin not only improves dyslipidemia-induced inflammatory response in the testicular tissue but also significantly reduces the serum levels of total cholesterol and LDL. Additionally, our meta-analysis revealed that melatonin partly increases HDL levels and lowers triglyceride levels; however, its effect on the serum level of HDL and triglyceride was not statistically significant.

To the best of our knowledge, this is the first systematic review and meta-analysis on the protective effects of melatonin against metabolic disorder-induced injuries to rodents’ testicular tissue. Robust pieces of evidence are provided which may facilitate future research. Although, conclusions should be made with caution as this is a rodent study and the results may not be reproducible in larger animals and human models. Although exploring the sources of heterogeneity is one of the added values of this review, in many cases our efforts failed with high residual heterogeneity. The remaining heterogeneity, which could be due to severe methodological discrepancy among the included studies, may alter the interpretation of the results. Furthermore, when interpreting the results, one must pay attention to the robustness of our results in each outcome and the fact that most of the studies had a high risk of bias. Also, conclusions regarding the effectiveness should consider the fact that none of the included studies have aimed to investigate the adverse events and parallel outcomes during melatonin therapy.