Metabolomics of Major Depressive Disorder: A Systematic Review of Clinical Studies

Although the understanding of the pathophysiology of major depressive disorder (MDD) has advanced greatly, this has not been translated into improved outcomes. To date, no biomarkers have been identified for the diagnosis, prognosis, and therapeutic management of MDD. Thus, we aim to review the biomarkers that are differentially expressed in MDD. A systematic review was conducted in January 2022 in the PubMed/MEDLINE, Scopus, Embase, PsycINFO, and Gale Academic OneFile databases for clinical studies published from January 2001 onward using the following terms: “Depression” OR “Depressive disorder” AND “Metabolomic.” Multiple metabolites were found at altered levels in MDD, demonstrating the involvement of cellular signaling metabolites, components of the cell membrane, neurotransmitters, inflammatory and immunological mediators, hormone activators and precursors, and sleep controllers. Kynurenine and acylcarnitine were identified as consistent with depression and response to treatment. The most consistent evidence found was regarding kynurenine and acylcarnitine. Although the data obtained allow us to identify how metabolic pathways are affected in MDD, there is still not enough evidence to propose changes to current diagnostic and therapeutic actions. Some limitations are the heterogeneity of studies on metabolites, methods for detection, analyzed body fluids, and treatments used. The experiments contemplated in the review identified increased or reduced levels of metabolites, but not necessarily increased or reduced the activity of the associated pathways. The information acquired through metabolomic analyses does not specify whether the changes identified in the metabolites are a cause or a consequence of the pathology.


Introduction And Background
Major depressive disorder (MDD), one of the most common psychiatric conditions, has a major impact on health systems around the world, with a worldwide prevalence of 17% and an annual incidence of around 6% in the general population [1]. It is associated with an increased risk of cardiovascular disease, metabolic syndrome, obesity, stroke and increased global mortality correlated with metabolic changes [2]. However, biomarkers have not yet been identified for the diagnosis, prognosis and therapeutic management of MDD. Such biomarkers could be the metabolites of cells, tissues and body fluids, such as peptides, amino acids, saccharides, phospholipids, coenzymes and nucleotides [3]. The investigation of molecules through metabolomic analysis could assist in the discovery of biomarkers potentially related to the predisposition, development and prognosis of MDD and other mental illnesses [4].
The possible benefits to psychiatry range from a better understanding of the pathophysiology of MDD [5,6] to novel strategies for its management, such as tests that suggest the best treatment option for patients according to their metabolomic profile, or tests that are capable of monitoring the metabolic-specific response to treatment. Additionally, such tests might even improve the staging of MDD and reveal personal predispositions for the condition.

Review
The other studies compared MDD individuals with and without medication or MDD patients with healthy controls, but not all studies specified which drugs were used ( Table 2) [11,16,19,20,[26][27][28]33,39,51]. One study that used fecal samples did not find any difference in metabolomic profile after correction of statistical analyses [57]. The remaining studies did not assess the interference of a specific therapy on metabolomic profile. The variety of methodologic and metabolic profiles hinders a precise conclusion of the effects of treatment on metabolites. However, the studies suggest possible biomarkers as predictors for the treatment of depression.
Many metabolites were related to MDD in the analyzed studies, and the methods used for detection were heterogeneous, with emphasis on liquid chromatography and gas chromatography associated with mass spectrometry and nuclear magnetic resonance spectroscopy. The fluids evaluated were plasma, serum, urine and cerebrospinal fluid, and for purposes of description, the metabolites were classified according to the fluids in which they were evaluated and their molecular characteristics.

Plasma/serum
Lipids Sphingomyelin, a sphingolipid that has a structural and cellular signaling function and is abundant in nerve tissues (particularly in myelin), is a fundamental component of cell membranes, and a high hydroxysphingomyelin/sphingomyelin ratio has proven to be a predictor of a good response to antidepressant treatment [36,59]. Patients with depression treated with intravenous ketamine had a sphingomyelin serum level increase at the time-point of 230 minutes after infusion. On the third day after ketamine infusion, the sphingomyelin levels appeared to decrease [29].
Arachidonic acid is an essential fatty acid and a major component of the cell membrane. In MDD patients, arachidonic acid plasma levels appear to be lower than in healthy subjects [22,52]. Treatment with sertraline is implicated in the augmentation of arachidonic acid in MDD patients, which is associated with a reduction in depressive symptoms [13]. High levels of arachidonic acid were predictors of recurrence [54]. Additionally, one study showed lower levels of linoleic, oleic, and heptadecylic acid and cholesterol in MDD patients with early stress [16].
Additionally, valeric acid was also found to be reduced in MDD patients, which could be explained by dysregulation of the brain-gut-microbiota axis or increased N-methyl-d-aspartate (NMDA) receptor activity [46]. On the other hand, nervonic acid, a monounsaturated fatty acid important for myelin synthesis, has been elevated in patients with MDD [26] but cerebrospinal fluid nervous acid levels did not differ between MDD patients and healthy controls [53].
A higher LDL/HDL ratio was found in MDD patients along with a reduction in omega-3 fatty acids levels [47,60] and a positive correlation between the Montgomery-Åsberg Depression Rating Scale (MADRS) score and serum levels of LDL, triglycerides, cholesterol, free cholesterol, phospholipids and apolipoprotein B before treatment [50]. Furthermore, a study that compared eight weeks of treatment with fluoxetine found reduced ApoB (lipoprotein B)/ApoA1 (lipoprotein A1) [58]. Additionally, Zheng et al. compared patients untreated in their first episode of MDD with healthy controls and found higher levels of polyunsaturated fatty acids and VLDL/LDL ratio in the MDD group. Low levels of cholesterol were present in MDD patients [16].

Amines
Phosphoethanolamine is a precursor to cell membrane phospholipids and is related to stages of cell metabolism such as apoptosis, which is why it has been studied in oncology and other areas [61]. A study suggests phosphoethanolamine as a possible biomarker for depression, as lower levels were found in MDD patients compared to healthy controls. However, the small sample size limited the comparison between MDD individuals undergoing pharmacological treatment and drug-naïve MDD subjects [31].
Higher baseline levels of melatonin, a hormone produced in the pineal gland using tryptophan as a substrate and whose primary function is sleep control, correlated with better response to treatment, and in responders their levels increased more [14].

Neurotransmitters
The main inhibitory neurotransmitter in the central nervous system (CNS), gamma-aminobutyric acid (GABA), as well as dopamine, a CNS modulating neurotransmitter, showed elevated levels in depressed patients [20,22,30]. Differently, histamine, one of the main chemical mediators involved in the allergic inflammatory response, seems to be low in MDD [22]. Leucine-enkephalin, an endogenous opioid with a high affinity for the delta-opioid receptor, was also found to be reduced in MDD patients [46].

Amino Acids
Low levels of tryptophan and kynurenine have been observed in MDD [30,32,56]. Indoles (metabolites of tryptophan) are elevated in MDD patients [55]. An inverse relationship was also observed between serum kynurenine levels and the severity of MDD: the lower the levels, the greater the severity [32]. In responders to an antidepressant, the kynurenine/melatonin and 3-hydroxyquinurenine/melatonin ratios decreased and the metabolite 3-hydroxyquinurenine also contributed to distinguish responders and non-responders [10,14]. Furthermore, antidepressant treatment increased kynurenine levels [56], and increased kynurenine levels were also associated with increased Beck Depression Inventory (BDI) scores after repetitive transcranial magnetic stimulation (rTMS) in treatment-resistant depression [49].
During treatment with intravenous ketamine, there was a slight increase in kynurenine (a tryptophan metabolite necessary for the synthesis of vitamin B3) 230 minutes after treatment. There was also a small reduction in these levels on the third day after this intervention [29]. Ketamine non-responders showed higher levels of D-serine and L-serine when compared with ketamine responders and controls [18]. In treatments with ketamine and esketamine, there was a reduction in metabolites of tryptophan and tyrosine [21].
Isoleucine, an essential amino acid that has direct effects on hypothalamic regulation by increasing satiety, has been linked to MDD when at low levels [13,20,37]. Low levels of gamma-glutamyl leucine were also associated with MDD, which may suggest abnormalities in the function of glutathione, an important antioxidant [46]. Arginine, which is an important amino acid for immune function and wound healing, has its lowest levels in MDD. Additionally, arginine levels increased significantly in recovered patients [33]. Treated patients showed higher levels of L-tyrosine, N-acetylornithine, kynurenine and lower levels of Lisoleucine [56].
Hung et al. showed that, in comparison to healthy controls, patients with MDD in complete remission presented significantly lower levels of metabolites related to pyruvate metabolism, via the tricarboxylic acid (TCA) cycle, linked to the metabolism of amino acids including alanine, aspartate, glutamate, arginine and proline and the metabolites of glycine, serine, and threonine. These findings corroborate those of Diang et al., who had previously reported lower levels of leucine and higher levels of alanine, serine, and proline in patients with MDD, pointing to an important involvement of amino acids in the pathophysiology and treatment response of MDD.

Genetic Products
Inosine, a nucleoside that has been studied as a neuroprotective in pathologies such as stroke, Parkinson's disease, and multiple sclerosis, showed low levels in depressed patients. In contrast, the serum levels of xanthine and adenosine were high in MDD [23,33]. Xanthine is a purine base and many stimulants such as caffeine and theobromine are derived from it [62]. Adenosine, on the other hand, is an endogenous purine that performs some functions in the CNS such as inducing sleep and relieving anxiety symptoms.
Others 5-Hydroxyindoleacetic acid, a metabolite of serotonin, tended to increase with the treatment of MDD, and the reduction of serotonin in the body can trigger emotional instability, insomnia, anxiety, and increased appetite [37].
Acylcarnitine, a hormonal activator that has been studied in pathologies such as Alzheimer's disease, showed low levels in subjects with MDD and elevated levels after treatment [17,40]. In contrast, dopamine, normetanephrine and metanephrine (metabolites of catecholamine) have shown high levels in MDD [22].
Glycoprotein acetylation (GlycA) is a novel inflammatory marker based on protein plasma glycosylation. There was a significant association between baseline GlycA levels and depression persistence. The role of high-sensitivity c-reactive protein (hsCRP), a traditional inflammatory marker, was also investigated, as well as the role of these inflammatory markers in the progression of depressive symptoms. GlycA levels predicted depression persistence. The association was robust and significant in fully adjusted models. Moreover, GlycA was superior to hsCRP in predicting depression persistence [41]. Other glycoproteins such as α-1-acid glycoprotein 1 and leucine-rich α-2-glycoprotein were higher in the MDD group [42]. These findings corroborate the involvement of this molecule in MDD.
Compared with controls, MDD patients with full remission had significantly lower expression of succinic acid, acetic acid, and pyruvic acid [51].

Amino Acids
Elevated levels of homocysteine, a sulfhydryl amino acid formed from methionine, are related to neurological and cerebrovascular diseases. Young and middle-aged MDD patients had low levels of ltyrosine compared with healthy controls [39]. In a study with plasma samples, the results were concordant, with treated patients showing higher levels of l-tyrosine [56].

Acids
Malonate, which is an inhibitor of cellular respiration, has lower levels in MDD [12,25]. It also appeared as a differential metabolite between depressed women and men: while females with MDD presented lower levels of malonate in comparison to healthy subjects, the same was not observed for males. Of note, females with MDD who responded or remitted after antidepressant treatment presented levels of urinary malonate concentration similar to those of healthy participants [20]. Using samples of young and middle-aged MDD patients, higher levels of citric acid and oxoglutaric acid and lower levels of quinolinic acid were found in this population compared to healthy individuals [39].

Cerebrospinal fluid
Amino Acids MDD patients in remission presented differences in tryptophan and tyrosine metabolism compared to MDD patients without remission and controls. The group in remission also had higher methionine levels and higher methionine/glutathione ratios than the other MDD and control groups, suggesting the involvement of methylation pathways and oxidative stress [11].

Discussion
Metabolomics research is an area of learning that explores metabolic pathways associated with various health problems, helps in understanding the pathophysiology of diseases, including mental disorders, and enables the discovery of biomarkers. This systematic review aimed to identify metabolites that are differentially expressed in MDD. After analyzing the selected studies, we found many metabolites related to MDD diagnosis and/or treatment, acting with cell signalers, cell membrane components, neurotransmitters, inflammatory and immunological mediators, hormonal activators and precursors and sleep controllers. However, there was a wide variation in the analyzed fluids and assessment methods, possibly due to the heterogeneity of metabolites found.
As was expected, compounds from the tryptophan, tyrosine, and purine pathways were differently expressed in MDD patients in many of the reviewed studies, as metabolic factors in the kynurenine pathway are considered possible mechanisms involved in the pathophysiology of MDD. The kynurenine pathway begins with the conversion of tryptophan to kynurenine. For the first branch, kynurenine is transformed into 3hydroxyanthranilic acid and quinolinic acid, which are N-methyl-d-aspartate (NMDA) receptor agonists that exert neurotoxic effects. For the second branch, kynurenine is transformed into kynurenic acid by kynurenine aminotransferases. Kynurenic acid is an NMDA receptor antagonist, which exerts a neuroprotective effect [43]. Tryptophan is an essential amino acid necessary for the production of serotonin and melatonin [63]. The two main metabolism pathways for tryptophan are 5-hydroxytryptophan and kynurenine.
Lower baseline plasma kynurenine is significantly associated with the severity of depressive symptoms and suicidal ideation. Kynurenine biosynthesis in the brain occurs primarily in astrocytes while tryptophan catabolism occurs mainly in glial cells. Kynurenine can cross the blood-brain barrier. Peripheral kynurenine, primarily generated in the liver, is the source of ∼60% of CNS kynurenine. However, the relationship of plasma kynurenine concentration to MDD symptom severity remains unclear and further investigations are necessary. A toxic kynurenine metabolite acts as an NMDA receptor agonist, which has been linked to depressive symptoms and other psychiatric manifestations [32].
Rotroff et al. also studied this metabolic pathway through a clinical trial with ketamine or esketamine in the treatment of MDD. Esketamine was the most potent enantiomer as an NMDA receptor antagonist. Metabolic changes have been demonstrated in relation to glutamate and tryptophan metabolism. Glutamic acid levels are increased 240 minutes after ketamine exposure. Ketamine is known to block the glutamatergic NMDA receptor; thus, the possible effect of increased glutamate levels could shift glutamatergic signaling from NMDA receptor to AMPA receptor to enhance the 5HT1B receptor activity that is hypothesized to be required for antidepressant effects. Treatment with either ketamine or esketamine resulted in decreased tryptophan metabolites. It is clear that the glutamatergic system appears to contribute to the risk and severity of MDD, requiring further investigation in this regard through original studies.
Another metabolite that deserves to be highlighted in our review is acylcarnitine. Ahmed et al. identified that more severe forms of depression are associated with reductions in short-chain acylcarnitines after SSRI treatment. This finding, and the relationship of acylcarnitine levels with mitochondrial fatty acid βoxidation and branched-chain amino acid catabolism, suggests that the pathobiology of MDD may manifest, in part, through metabolomic dysfunction. Further, these findings may reflect changes in mitochondrial function or ATP production in patients with MDD. Moaddel et al. found a decrease in acylcarnitine concentrations after ketamine treatment compared to placebo.
Previously published reviews brought together studies that evaluated metabolites in order to identify reliable biomarkers for MDD. In agreement with the present systematic review, Macdonald et al. found a diversity of analyzed fluids and methodologies used. Urine, cerebrospinal fluid, plasma, and serum were also identified as analyzed fluids. Analysis techniques such as the use of gas and liquid chromatography combined with mass spectrometry, capillary electrophoresis time of flight mass spectrometry (CE-TOF-MS), NMR and liquid chromatography with electrochemical coulometric array detection (LCECA) were observed [64]. These techniques are widely used because they allow the simultaneous detection of numerous metabolites [65].
Specific biomarkers were found to be related to MDD by Macdonald et al. (2019), with glycine, alanine, citrate and formate increasing and phenylalanine, valine, aminoethanol, and hypurate is shown to be negatively regulated. Other metabolites were found to be only potentially consistent. Some metabolic pathways were found to be involved in the pathophysiology of MDD such as coenzyme Q biosynthesis, glycine-serine-threonine metabolism, tyrosine metabolism, pyrimidine metabolism, and steroid biosynthesis [44,45,56,66].
Our findings might be helpful to researchers in the field and to future research question formulations, by elucidating which metabolites seem to be associated with MDD pathophysiology, laboratory diagnosis and therapeutic approaches and which, therefore, should be investigated further. It is also important to discover specific molecular biomarkers for mental disorders, as an objective and complementary method to currently existing diagnoses, which use concepts that mostly have a subjective character. The metabolic profile can be used as a response predictor, thus assisting in making more targeted decisions [65].
The experiments considered in the review identified increased or reduced levels of metabolites, but not necessarily increased or reduced activity of the associated pathways. The information acquired through metabolic analyses does not specify whether the changes identified in the metabolites are a cause or a consequence of the pathology. It is not known, therefore, whether the affected pathways in different psychiatric conditions represent causal mechanisms of the diseases. Gadad et al. concluded that no biomarker has been translated into clinical practice for the diagnosis of depression or treatment definition. However, some recent studies have suggested the possibility of diagnostic metabolic panels for depressive disorders, such as phosphatidylserine (16:0/16:1) for MDD, with an AUC value of 0.876 [48].
Some limitations are the heterogeneity of studies on metabolites, methods for detection, analyzed body fluids, and treatments used. Not all studies had detailed assessment tools such as Hazard Ratio. Thus, it was not possible to carry out the synthesis with meta-analysis. To minimize the heterogeneity, we chose to include in this review only patients with unipolar depression, without associated health conditions, such as pregnancy and puerperium, hepatitis, and acquired immunodeficiency syndrome. However, prevalent health conditions such as diabetes mellitus, systemic arterial hypertension, obesity, and dyslipidemia were not detailed in the studies. MDD is not a phenotypically and genetically homogeneous disorder. Rather, MDD might be seen as a highly prevalent syndrome, with a wide polygenicity and present in people of different ages and ethnic backgrounds [67]. Likewise, several gene-environment interactions have been suggested as risk factors of MDD. Because of such heterogeneity, metabolite-mapping related to MDD is a challenge.
As seen in most of the studies cited in this review, MDD metabolite investigation protocols do not take into count that heterogeneity; therefore, their results might be influenced by samples with significantly different phenotypes, severity, ages or treatment approaches. The data obtained were not sufficient to distinguish changes directly related to disease or as a consequence of life habits, phenotypic characteristics and effects related to treatment. Therefore, it would be advantageous for metabolomic studies to investigate more homogeneous MDD subtypes.

Conclusions
The results presented in our review show that several metabolites are altered in MDD and change with treatment, and the most consistent evidence available relates to kynurenine and acylcarnitine. However, there is still not enough evidence to propose changes in the diagnosis or therapeutic management of MDD. In view of the limitations presented, we suggest the investigation of metabolites in MDD in more homogeneous conditions, taking into consideration both the MDD phenotype and the patient's characteristics. Considering the benefits that biomarkers can bring to the pathophysiological understanding, diagnosis and treatment of MDD, further metabolomics research is a necessity.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.