The Preliminary Study On Serum Metabolites And Metabonomics Of Patients With Major Depressive Disorder

Objectives: To investigate the changes of monoamine neurotransmittermetabolitesin major depressive disorder （MDD）patients’ serums and explore the connection betweendiscovered changes and MDD. To establish the serum metabolite profile of MDDpatients,match identified metabolites to specific metabolic pathways, screen for MDD-characteristic metabolic markers，andlay the foundation for further investigation of MDDpathogenesis and clinical diagnosticsThe study consists of two parts：Part1： A correlation study between disturbances of serum monoamineneurotransmitter and depressionMethods: HPLC--ECD was used to determine the levels of pinephrine （E）,epinephrine （NE）, erotonin hydrochloride （5-HT）,3，4-dihydroxyphenylaceticacid(DOPAC), homovanillic acid（HVA）, and5-hydroxyindole-3-acetica cid（5-HIAA）in serumsof20MDD patients （HAMD>24, experimental group, EG） and20healthyadults (contro1group,CG). The concentrations, ratios and coefficient correlations ofmonoamine neurotransmitters in EG and CG serums were compared.Results: The concentrations of NE, E and DOPAC showed significant differencesbetween EG and CG. In EG serum, NE, E and DOPAC were1.98±1.66ng/L,2.53±1.01ng/L, and92.76±46.67ng/L, respectively.In CG serum, the valueswere4.47±3.95ng/L,3.49±1.05ng/L, and196±75.78ng/L, respectively. In EG, the levels of5-HT（0.57±0.39）ng/L,5-HIAA（114.78±37.95）ng/L, and HAV（1.74±1.69）ng/L were lower than thesein CG,but not significant (P>0.05). There were obvious correlations in EG between thechanges of DOPAC and NE,5-HT and5-HIAA, HVA and5-HIAA, respectively; whereasthe same correlations were not found in CG. The ratios of5-HT/5-HIAA,5-HIAA/NE,HVA/NE, and DOPAC/HVA were significantly lower in EG than these in CG, andcorrelation coefficients of these ratios showed significant differences between EG and CG. Part2： The preliminary metabolomic study of depression patients’ serumMethods:We applied ultra-performance liquid chromatography coupled with tandemmass spectrometry（UPLC-MS/MS）and gas chromatography coupled with tandem massspectrometry（GC-MS）to analyze the serum samples of20MDD patients and20healthycontrols in order to discovermetabolic differences. Multivariate statistical analyses,includingWelch’s two-sample t-test, Principal Components Analysis（PCA）, Partial LeastSquares-Discrimination Analysis（PLS-DA）, and Random Forest were performed on thedatasets. Potential biomarkers related to MDD were discovered through these statisticalanalyses.Results:In total,345metabolites were identified from the serum samples.Unsupervised PCA analysis showedclear separation between the MDD group and thecontrol group. Supervised PLS-DA analysis also resulted in separation of the MDD groupand the control group, with a predictive power of the model equaled to44.72%(Q2=0.447).Random Forest analysis had much better model predictive power, with a predictiveaccuracy of92.5%. The top30metabolites responsible for the separation between MDDand healthy serum in Random Forest analysis were listed，which were potential biomarkersthat included amino acid metabolites, fatty acid metabolites, energy metabolites, steroidhormone metabolites, and heme metabolites.Conclusions:1. Disorders of blood monoamine neurotransmitters and their metabolites have strongconnections with MDD.2. Untargeted metabolomic analysis of MDD and healthy serum samples indicatedclear metabolic differences between the two serum types by Principal ComponentsAnalysis（PCA）, Partial Least Squares-Discrimination Analysis（PLS-DA）, and RandomForest analysis.3. Using supervised random forest analysis,30important metabolites making the mostsignificant to the separation between MDD and healthy serum were discovered (out of thetotal of345identified metabolites). The30metabolites were involved in amino acidmetabolism, fatty acid metabolism, energy metabolism, steroid metabolism, and hememetabolism. This research laid a foundation for further studies of the pathogenesis ofdepression and clinical diagnostics.