Metabolomic Signatures Linking Depressive Symptoms to Atherosclerotic Cardiovascular Disease Comorbidity: Evidence from the UK Biobank

Metabolomic Signatures Linking Depressive Symptoms to Atherosclerotic Cardiovascular Disease Comorbidity: Evidence from the UK Biobank | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Metabolomic Signatures Linking Depressive Symptoms to Atherosclerotic Cardiovascular Disease Comorbidity: Evidence from the UK Biobank Angela Koloi, Kevin Dobretz, Dafne Damman, Rick Quax, Dimitris Zaridis, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6422455/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Depression and cardiovascular disease (CVD) frequently co-occur and exacerbate each other's clinical outcomes, potentially through shared immune or metabolic dysregulation. This study used tripartite network analysis to examine interactions among depressive symptoms, metabolic biomarkers, and CVD, aiming to identify mediating metabolic pathways. Using UK Biobank data (N = 35 711; 56% female; mean age 64), NMR-based plasma metabolites, PHQ-9 depressive symptoms, and ICD-10 atherosclerotic CVD were analyzed. A tripartite network (symptom-metabolite-CVD) was constructed using partial correlation analyses to identify metabolites shared between depression and CVD. These metabolites were then subsequently entered in a longitudinal mediation analysis, spanning 6 years, to quantify metabolite-mediated associations between depressive symptoms and incident CVD. Models progressively adjusted for: (1) age and sex; (2) socioeconomic status, lifestyle factors, and antidepressant use; and (3) BMI. Analyses were repeated using PHQ-9 total scores as the exposure. The initial unadjusted network analysis identified 11 shared metabolic markers. Longitudinal mediation revealed six symptom-metabolite-CVD pathways, whereas only one metabolite-mediated pathway emerged using total PHQ-9 scores. After full adjustment, four pathways remained significant: alpha-1-glycoprotein acetyls (AGP) mediated associations involving appetite changes and fatigue (both p < .001), and monounsaturated fatty acids (MUFAs) mediated associations involving appetite changes and sleep disturbances (both p < .05). AGP (an inflammatory acute-phase protein) and MUFAs link energy-related depressive symptoms (appetite changes, fatigue, sleep disturbances) to CVD. This symptom-level approach enhances precision in identifying shared biological mechanisms underlying the co-morbidity between depression and CVD, potentially informing novel avenues for tailored prevention and treatment strategies. Health sciences/Diseases/Psychiatric disorders/Depression Biological sciences/Molecular biology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Major Depressive Disorder (MDD), the most prevalent psychiatric disorder globally, frequently co-occurs with cardiovascular diseases (CVD) [ 1 , 2 ], exacerbating disease severity and mortality risk [ 3 ]. Both conditions impose a substantial individual and societal burden in part due to their chronic nature [ 4 – 7 ]. Evidence suggests that their co-morbidity may involve shared immune or metabolic mechanisms, particularly in lipid metabolism and inflammatory pathways [ 2 , 3 , 8 ]. Large-scale metabolomics studies have consistently identified a distinctive inflammatory and metabolic profile associated with depression [ 9 – 11 ], characterized by elevated alpha-1-glycoprotein acetyls (AGP), triglycerides, very-low-density lipoproteins (VLDL), and reduced levels of high-density lipoproteins (HDL). Additionally, meta-analyses and systematic reviews [ 12 – 14 ] have further confirmed elevated levels of inflammatory markers in depression, supporting an inflammatory subtype associated with increased cardiovascular risk [ 15 – 17 ]. Genomic studies indicate that genetic liability for MDD elevates atherosclerotic CVD risk, with causal effects partially mediated by shared metabolic factors [ 18 ]. Research into depression-CVD comorbidity, however, faces three significant knowledge gaps. Firstly, most studies separately examine the biological pathways underlying either depression or CVD, with limited integration of shared or mutually interactive mechanisms. Second, there is a scarcity of longitudinal research explicitly investigating whether such shared metabolic pathways indeed link depression to subsequent CVD. Thirdly, most research relies on aggregate measures of overall depression severity, potentially masking the distinct contributions of individual symptoms to cardiovascular risk. For instance, specific depressive symptoms (e.g., fatigue, sleep disturbances) may independently or interactively influence cardiovascular outcomes [ 19 ] highlighting the importance of considering symptom heterogeneity to inform targeted interventions. To address these gaps, our study employed a tripartite network analysis aiming to simultaneously examine interactions among depressive symptoms, metabolic biomarkers, and atherosclerotic cardiovascular disease (CVD, Fig. 1 A). This analytical approach allowed for the identification of specific metabolic pathways that may mediate the progression from depression to incident CVD; a hypothesis subsequently tested through longitudinal mediation analyses (Fig. 1 B). Network analysis is uniquely suited to addressing heterogeneity within depression [ 20 , 21 ], enabling the identification of symptom-level relationships within a complex system accounting for all variables. By utilizing an advanced integrative data analysis framework combining cross-sectional and longitudinal methods, this study aimed to identify biomarkers bridging depression and CVD, thereby enhancing our understanding of the mechanisms of comorbidity and potentially informing targeted intervention strategies. Methods Dataset The UK Biobank (UKB) is a large-scale population-based cohort consisting of approximately 500 000 participants aged 40 to 69 years, recruited between 2006 and 2010 from 22 assessment centres across England, Scotland, and Wales. The cohort was designed to include a diverse mix of urban and rural residents, reflecting a broad socioeconomic demographic [ 22 ]. Analyses were limited to 157 286 participants with available data on a mental health questionnaire (MHQ), which included the PHQ-9 (see below). We excluded participants diagnosed with schizophrenia or form of psychosis, personality disorders, and manic episodes associated with bipolar disorder, resulting in an analytical sample of 155 649 individuals. Nuclear magnetic resonance (NMR) metabolite data (see Methods section: Plasma Biomarker Profiling by NMR ) were available for a subset of UKB participants (N = 118 461). The final study sample comprised 35 711 participants after ensuring the availability of NMR metabolite data and further exclusion of missing data (attributes with more than 15% missing values). Assessment Patient Health Questionnaire-9 (PHQ-9) We assessed depressive symptoms using both individual items and the total PHQ-9 sum score [ 23 ]. The PHQ-9 evaluates nine symptoms on a 4-point Likert scale (0 = "not at all" to 3 = "nearly every day") (Supplementary Table 1; [ 24 ] The sum score (range: 0–27) was calculated by summing all item scores, with higher scores indicating greater depression severity. Atherosclerotic Cardiovascular Disease (ACVD) The presence of CVD at the time of depression assessment and throughout a 6-year follow-up period was assessed by retrieving the earliest date of a CVD outcome from twelve fields in the UKB dataset. Dates of CVD events were primarily retrieved from hospital inpatient records using International Classification of Diseases-10 (ICD-10) summary diagnoses (data fields 41202 (primary diagnoses), 41204 (secondary diagnoses), 41270 and 41280 (first-time diagnoses with names and dates)), from which we identified conditions characterized by atherosclerosis (see Supplementary Table 2 for the complete list of ICD-10 codes). Additionally, CVD outcomes were confirmed through death records, including date of death (field 40000) and primary and secondary causes of death (fields 40001, 40002). Four data fields directly reporting the dates of the first events for ST-elevation myocardial infraction (STEMI, field 42002), non-STEMI (NSTEMI, field 42004), ischaemic stroke (field 42008) and angina pectoris (field 131296) were combined with those extracted from inpatient records. Dates from these fields were merged to determine the earliest CVD event for each participant, relative to the time of depression assessment and the 6-year follow-up. The outcome was defined as a binary score (yes/no) indicating whether a participant developed CVD during the follow-up period. At the time of depression assessment, 2 131 participants had existing CVD cases. Excluding these, 1 326 new CVD cases were identified during the 6-year follow-up (see Table 1 ). Table 1 Descriptive statistics in UK Biobank. Mental health questionnaire assessment Characteristics subset with plasma NMR N = 35 711 Missing, % Age years (mean ± SD) 63.98 ( ± 7.74) 0 Sex (F) (%) 56.30 0 Townsend Deprivation Index (%) 0.14 average 31.10 most 12.36 least 56.55 Lifestyles Smoking status (%) 0.23 never 57.43 former 35.34 current 7.01 Alcohol consumption (%) 0.16 never 8.38 monthly or less 12.93 two to four times a month 18.66 two to three times a week 29.71 four or more times a week 30.16 Moderate Physical activity (%) 85.96 2.7 BMI (kg/m 2 ) (mean ± SD) 26.77( ± 4.56) 1.28 Medication Antidepressant use (%) 5.94 0 Health indicators PHQ-9 > = 10 (%) 66.01 0 Atherosclerotic CVD N = 2 131 0 6 years follow-up assessment Atherosclerotic CVD N = 1 326 0 Covariates Covariates descriptions of UKB are presented in Table 1 . In addition to sex (data field 31) and age (data field 21003), sociodemographic status was assessed using the Townsend Deprivation Index (TDI). The TDI (data field 22189) is based on the individual’s postcode, which leverages census data to incorporate information on employment, house and car ownership, and household crowing [ 25 ]. Smoking (data field 20116) was self-reported as “prefer not to answer”, “never”, “previous” or “current”. Frequency of drinking alcohol (data field 20414) use was self-reported as “prefer not to answer”, “never”, “four or more times a week”, “two to three times a week”, “two to four times a month”, “monthly or less”. Responses of “prefer not to answer” were treated as missing data and excluded from analysis. Moderate physical activity was assessed using self-reported data on the number of days per week participants engaged in moderate physical activity lasting 10 + minutes (data field 884). Responses were categorized into a binary variable: participants reporting 0 days per week were classified as inactive, while those reporting 1 or more days per week were classified as moderate active. Antidepressant use was considered as a binary variable (1 = participant used any of the antidepressants listed in Supplementary Table 3; 0 = participant did not use any of these medications). BMI was calculated as weight (kg) divided by the square of height (m 2 ) (data field 21001). Plasma Biomarker Profiling by NMR Baseline plasma samples from UKB participants underwent biomarker profiling by Nightingale Health Plc., using nuclear magnetic resonance (NMR) spectroscopy. Details of the Nightingale Health NMR biomarker platform have been described [ 26 ]. Biomarker data were available for a randomly selected subset of 118 461 ethylenediaminetetraacetic acid (EDTA)-treated plasma samples from UKB participants, providing a notably extensive metabolic dataset compared to earlier metabolomics studies. Each plasma sample yielded 249 metabolic measures, including 168 absolute concentrations and 81 ratio measures. For the present analyses, we chose to focus on 47 markers encompassing lipids, fatty acids (FAs), and low molecular weight metabolites. These biomarkers were chosen based on prior metabolomics research demonstrating their associations with depression [ 27 , 28 ]. To avoid redundancy, we excluded additional sub-measures and ratios related to these lipoproteins, such as lipid composition and particle concentration of lipoprotein subclasses, as well as lipid and FAs ratios. Additionally, C-reactive protein (CRP) was included in the analyses due to strong supporting evidence from meta-analyses [ 12 , 29 ] and gene expression studies [ 15 , 16 ], consistently linking elevated CRP levels to key depressive symptoms. Biomarkers, their respective groupings, and associated units are described in Supplementary Table 4. Distribution plots of raw metabolite data and CRP are presented in Supplementary Fig. 1. Pre-process Missingness Missingness on demographics and health indicators can be found in Table 1 . Based on the results of Little's MCAR test [ 30 ], the missing data was not completely at random (MCAR), hence we cannot drop these values. To handle the missing data, we implemented Multivariate Imputation (MI) using the iterative imputer of scikit-learn version 0.14.4. This approach effectively captures complex relationships between variables, such as non-linearity, by using a round-robin technique, imputing missing values, updating models, and re-imputing across multiple iterations. Outliers We applied the Isolation Forest algorithm [ 31 ] to detect outliers. This algorithm isolates anomalies by randomly selecting a feature and splitting the data between its maximum and minimum values and repeating this process. The distribution of the outliers was visualized using box plots (Supplementary Fig. 2), and the identified outliers were subsequently removed from the dataset. Normalization Data were pre-processed according to the manufacturer’s standardized protocol, previously applied in other large-scale studies [ 10 ]. Initially, any values exceeding 5 standard deviations from the mean were removed to eliminate extreme outliers. Subsequently, each data point was incremented by 1, followed by a natural log transformation to normalize the distribution. Finally, metabolites data underwent normalization to adjust all values to a range from 0–1. Analysis Plan Graphical Model The cross-sectional network included three groups: metabolites, symptoms, and CVD. We excluded intra-group connections (e.g., symptom-symptom or metabolite-metabolite) and focused solely on between-group associations (metabolite-symptom, metabolite-CVD). To construct the tripartite network, we employed partial correlation analysis to identify significant shared metabolites between individual symptoms of depression and CVD while adjusting for all other variables in the system. Partial correlation quantifies the relationship between two variables by removing the influence of all other variables. For ordinal depressive symptoms, we used Spearman partial correlation, which measures the monotonic relationship between variables after residualizing them with respect to all other variables (including other metabolites, depressive symptoms, and CVD). For binary CVD outcomes, we used point-biserial correlation, which evaluates the association between continuous metabolites and binary CVD after residualizing both with respect to all other variables. In both cases, residuals were obtained by performing regression analyses on the variables of interest with respect to all other variables, and correlations were computed between these residuals to reflect direct associations free from the influence of other variables. The resulting network represented variables as nodes and partial correlations as edges. This method assumes that the relationships among the variables are linear and that the residuals from the regression analysis are normally distributed. Sensitivity Analyses To assess the stability of the network edges, we performed bootstrap resampling with 1,000 iterations for each group. In each iteration: (i) data was resampled with replacement, (ii) partial correlations were calculated between all pairs of metabolites and symptoms, and (iii) significant edges ( p < .05) were recorded along with their correlation coefficients (either positive or negative). For each potential edge in the network, we computed: (i) the proportion of bootstrap samples in which the edge appeared (frequency), (ii) the average partial correlation coefficient across bootstrap samples (mean weight), and (iii) the standard deviation of the partial correlation coefficients (weight standard deviation). We considered edges that appeared in at least 75% of the bootstrap samples. To further assess the robustness of associations, a permutation test with 1,000 iterations was performed. For each pair, observed partial correlations were calculated using the original data. To test the null hypothesis of no association, symptom values were randomly permuted across samples while keeping metabolite values fixed [ 32 ]. Partial correlations were then recalculated for each permutation, generating a null distribution of correlation values. A permutation-based p-value ( p perm ​) was calculated as the proportion of permuted correlations with an absolute value equal to or greater than the observed correlation. Pairs with p perm ​ < .05 were considered statistically significant. Centrality assessment To determine key metabolites bridging symptoms and CVD, we calculated betweenness centrality, a widely recognized metric in network analysis [ 33 ]. This metric quantifies how often a node (in this case, a metabolite) acts as a bridge between other nodes in the network. Betweenness centrality is calculated by determining the number of shortest paths between all pairs of nodes that pass-through a given node. Metabolites with high betweenness centrality play crucial roles in mediating information flow within the network, potentially linking depression symptoms to CVD [ 34 ]. To validate the significance of central nodes, we employed a permutation-based approach. We first calculated betweenness centrality for each metabolite in the original network. We then created 1000 permuted networks by randomly shuffling depression symptoms and CVD risk data, generating a null distribution of centrality values. By comparing observed centralities to these null distributions, we computed p perm for each metabolite. Mediation Analysis We conducted a longitudinal mediation analysis to examine how individual depressive symptoms influenced CVD incidence six years later, mediated by identified bridge metabolites, as shown in Fig. 1 B. The mediation analysis is fundamentally based on regression analysis and involves a series of regression models to quantify the relationships between the independent variable (X; depressive symptoms), the mediator (M; metabolites), and the dependent variable (Y; CVD incidence). The total effect of depressive symptoms on CVD is estimated using a logistic regression model: $$\:logit\left(P\left(Y=1\right)\right)=\:{\beta\:}_{0}+{\beta\:}_{1}X+\:\epsilon\:,\:\epsilon\::error\:term$$ Next, the effect of depressive symptoms on the metabolites is estimated using a linear regression model (since the mediator is continuous): $$\:M=\:{\beta\:}_{0}+{\beta\:}_{2}X+\epsilon\:,\:\epsilon\::error\:term\:\:$$ Finally, the direct effect of depressive symptoms on CVD and the effect of the mediator on CVD are estimated using a logistic regression model: $$\:logit\left(P\left(Y=1\right)\right)=\:{\beta\:}_{0}+{\beta\:}_{3}X+{\beta\:}_{4}M+\:\epsilon\:,\:\epsilon\::error\:term$$ The indirect effect is calculated as the product of the coefficients: $$\:Indirect\:Effect=\:{{\beta\:}}_{2}+{{\beta\:}}_{4}$$ The mediation analysis incorporated three progressive models to account for potential confounders. The base model adjusted for age and gender. The second model expanded on this by additionally controlling for TDI, alcohol consumption, smoking, physical activity, and antidepressant use. The final model included all previous factors plus BMI. This stepped approach allowed us to isolate the mediating effects of metabolites on the depression-CVD relationship while systematically controlling for demographic, socioeconomic, lifestyle, and health factors. To robustly estimate indirect effects (ACME) and their 95% confidence intervals, we conducted non-parametric bootstrapping with 1000 resamples. Confidence intervals were derived using the percentile method [ 35 ], which does not assume normality of the sampling distribution, thereby enhancing the reliability of results. Statistical analysis focused on key mediation metrics to clarify the relationships between depressive symptoms, metabolic biomarkers, and CVD. The average causal mediation effect (ACME) quantified the average change in CVD risk attributed to the mediator influenced by depressive symptoms. The average direct effects (ADE) measured the average impact of depressive symptoms on CVD, independent of the mediator. The total effect combined both indirect (ACME) and direct (ADE) effects, while the proportion mediated represents the fraction of the total effect on CVD explained by each mediator (metabolites). Bootstrapping analyses were performed in Python 3.11, utilising the statsmodels library (version 0.14.4). We conducted an additional longitudinal mediation analysis to examine the effect of the PHQ-9 total score on CVD incidence, mediated through the identified bridge metabolites. This secondary analysis enabled a direct comparison analysis of mediation effects derived from individual depressive symptoms versus the total PHQ-9 score. The same covariate sets were consistently applied across analyses to facilitate direct comparability of findings. Results Descriptives After merging depression assessments with NMR biomarker data and completing preprocessing steps (including missing data imputation, outlier detection, and normalization), the refined dataset comprised 33 925 participants (mean age: 64 years; 56% female). Within this sample, 66.01% of participants reported a PHQ-9 score of 10 or above, indicating moderate to severe depressive symptoms. Supplementary Table 5 provides a detailed breakdown of depressive symptom frequency and missing data rates. Supplementary Fig. 2 depicts the outlier analysis for metabolite measurements prior to normalization. Network and Sensitivity Analyses Figure 2 presents the tripartite network constructed through partial correlation. Only stable connections among depressive symptoms, metabolites, and CVD are visualized. Stability and robustness of these associations were assessed through bootstrap resampling and permutation tests, presented in Supplementary Fig. 3 and Supplementary Fig. 4, respectively. Permutation tests confirmed 11 significant bridge associations linking depressive symptoms and CVD ( p < .05), involving acetoacetate, CRP, creatinine, AGP, glutamine, histidine, linoleic acid, monounsaturated Fatty Acids (MUFAs), omega-6 FAs, total cholines, and tyrosine. Appetite changes appeared as the most frequently associated symptom (6 times), followed by sleep problems and fatigue (4 and 3 times, respectively). These associations, presented in Fig. 3 as partial correlations, formed symptom-metabolite-CVD triplets /pathways utilized subsequently in mediation analyses. Supplementary Fig. 5 presents the betweenness centrality assessment, highlighting citrate as the most central and statistically significant metabolite after the permutation test, followed by tyrosine, 3-hydroxybutyrate, acetate, and creatinine. Mediation Analysis Exposure: Individual depressive symptoms Using identified stable metabolite pathways from the network analysis, mediation analysis quantified longitudinal associations between depressive symptoms and incident CVD. Figure 4 shows the estimated indirect effects (ε) for symptom-metabolite combinations. Six significant mediation pathways initially emerged: appetite changes to CVD via CRP, creatinine, AGP, and MUFAs; sleep disturbances via MUFAs; and fatigue via AGP. After progressive covariate adjustments, including BMI, four pathways were retained: appetite changes were associated with CVD through AGP ( p < .001) and MUFAs ( p < .05); sleep problems maintained their link to CVD via MUFAs (p < .05); and fatigue remained connected to CVD through AGP ( p < .001). Detailed results of the mediation analysis are presented in Supplementary Table 6. Exposure: PHQ-9 sum score A secondary mediation analysis using total PHQ-9 scores examined the same metabolite mediators: acetoacetate, CRP, creatinine, AGP, glutamine, histidine, linoleic acid, MUFAs, omega-6 FAs, total cholines, and tyrosine (identified as bridge). The PHQ-9 sum score's effect on CVD was significantly mediated by AGP ( p < .05) in the age and gender adjusted model, but this relationship did not persist after controlling for additional lifestyle factors and BMI. Discussion This study employed an integrative analytic approach combining cross-sectional network and longitudinal mediation analysis to identify metabolic pathways mediating the relationship between depressive symptoms and incident CVD. Contrasting symptom-level analyses with aggregate PHQ-9 score revealed greater specificity of symptom-based approaches, particularly linking energy-related depressive symptoms (appetite changes, fatigue, sleep disturbances) were linked to CVD via metabolic mediators (Fig. 3 ). In line with our hypothesis, longitudinal mediation analysis confirmed and extended these findings, revealing six significant symptom-metabolite-CVD incidence pathways. After progressive covariate adjustments, AGP and MUFAs persisted as significant mediators. The distinct mediation effects observed for individual depressive symptoms, compared to total PHQ-9 scores, further underlined the value of symptom-level analysis, and suggest that individual symptoms may align with unique biological signatures influencing cardiovascular risk. Providing our observation with a sound empirical basis is that metabolic and inflammatory pathways have been implicated in both depression and CVD [ 2 , 8 , 11 ], and specifically research linking these pathways with “atypical, energy-related” symptom (AES) profile, which includes symptoms like hyperphagia, weight gain, hypersomnia, fatigue, and leaden paralysis [ 36 ]. AGP consistently mediated associations between appetite changes, fatigue, and CVD, exemplifying the broader observation that chronic low-grade inflammation may be a critical shared pathway linking CVD risk with atypical, energy-related depressive symptoms [ 37 ]. While our analysis could not differentiate between increased or decreased appetite, previous literature indicates specifically increased appetite is associated with higher AGP levels [ 36 ]. The observed AGP associations support its potential role as a stable and predictive biomarker of future cardiovascular and metabolic risks [ 38 ], further emphasizing its clinical relevance. Similarly, MUFAs remained significant mediators linking depressive symptoms (particularly appetite changes and sleep disturbances) to later incident CVD. This finding aligns with the literature suggesting lipid metabolism and dietary fat intake play pivotal roles in the comorbidity between depression and CVD [ 39 , 40 ]. Diets rich in MUFAs, such as the Mediterranean diet, have consistently demonstrate cardiovascular and mental health benefits [ 41 , 42 ]. For example, the PREDIMED trial demonstrates that a Mediterranean diet reduce depression risk [ 42 ], which is backed by observational research [ 41 ]. These protective effects likely involve anti-inflammatory actions, improved lipid profiles, and enhanced endothelial function underscoring the importance of further research into such shared nutritional determinants [ 43 ]. Further investigations should consider bidirectional relationships, as metabolic alterations could both result from and exacerbate depressive symptoms, justifying further longitudinal studies exploring such reciprocal interactions. Moreover, future research could examine whether targeted nutritional or pharmacological interventions addressing inflammation and lipid metabolism can mitigate cardiovascular risk among depressed individuals, particularly those presenting atypical, energy-related symptoms. Despite its strengths, including the use of robust statistical methods and a large representative cohort, this study has limitations. The inability to distinguish between increased and decreased appetite and subtypes of sleep disturbances (e.g., insomnia vs. hypersomnia), the analysis may obscure important distinctions in depressive subtypes such as such as atypical versus melancholic depression. Part of the DSM atypical depression specifier increased appetite and hypersomnia may underly divergent CVD risk [ 44 ]. Future research should explore the synergistic interactions between multiple biological mediators in understanding the relationship between depressive symptoms and CVD comorbidity. By employing a joint mediation analysis, researchers can uncover novel pathways connecting simultaneously examine inflammatory markers, lipid metabolites, and neuroendocrine factors to uncover hidden pathways linking these complex health conditions. In conclusion, our findings substantiate the hypothesis of immunometabolic subtype of depression, which links low-grade inflammation and metabolic dysregulation to depression with atypical, energy-related symptoms [ 45 ]. These disruptions may, in turn, contribute to the development of CVD. AGP and MUFAs emerged as critical mediators of the relationship between depressive symptoms and incident CVD highlighting the shared metabolic pathways. These findings align and expand previous observational and gene expression studies linking AGP [ 10 , 36 – 38 ] and MUFAs [ 41 , 42 ] to both depression and CVD. Clinically, the results point to a two-pronged approach: combining symptom-level analysis for personalized CVD prevention with interventions targeting metabolic pathways to reduce cardiovascular risk in depressed individuals with atypical, energy-related symptoms. Declarations Conflict of Interest The authors have no conflict of interest to disclose. Supplementary Information Supplementary information is available at MP’s website. Acknowledgements This work was supported by the EU-project T0_AITION (grant number 848146). References Gold SM, Köhler-Forsberg O, Moss-Morris R, Mehnert A, Miranda JJ, Bullinger M, et al. Comorbid depression in medical diseases. Nat Rev Dis Primers. 2020;6:69. Berk M, Köhler-Forsberg O, Turner M, Penninx BWJH, Wrobel A, Firth J, et al. Comorbidity between major depressive disorder and physical diseases: a comprehensive review of epidemiology, mechanisms and management. World Psychiatry. 2023;22:366–387. Hare DL, Toukhsati SR, Johansson P, Jaarsma T. Depression and cardiovascular disease: a clinical review. European Heart Journal. 2014;35:1365–1372. Almeida SS, Zizzi FB, Cattaneo A, Comandini A, Di Dato G, Lubrano E, et al. Management and Treatment of Patients With Major Depressive Disorder and Chronic Diseases: A Multidisciplinary Approach. Front Psychol. 2020;11. Deaton C, Froelicher ES, Wu LH, Ho C, Shishani K, Jaarsma T. The Global Burden of Cardiovascular Disease. European Journal of Cardiovascular Nursing. 2011;10:S5–S13. Joseph P, Leong D, McKee M, Anand SS, Schwalm J-D, Teo K, et al. Reducing the Global Burden of Cardiovascular Disease, Part 1. Circulation Research. 2017;121:677–694. Liu J, Liu Y, Ma W, Tong Y, Zheng J. Temporal and spatial trend analysis of all-cause depression burden based on Global Burden of Disease (GBD) 2019 study. Sci Rep. 2024;14:12346. Penninx BWJH. Depression and cardiovascular disease: Epidemiological evidence on their linking mechanisms. Neuroscience & Biobehavioral Reviews. 2017;74:277–286. Julkunen H, Cichońska A, Tiainen M, Koskela H, Nybo K, Mäkelä V, et al. Atlas of plasma NMR biomarkers for health and disease in 118,461 individuals from the UK Biobank. Nat Commun. 2023;14:604. Bot M, Milaneschi Y, Al-Shehri T, Amin N, Garmaeva S, Onderwater GLJ, et al. Metabolomics Profile in Depression: A Pooled Analysis of 230 Metabolic Markers in 5283 Cases With Depression and 10,145 Controls. Biol Psychiatry. 2020;87:409–418. Jansen R, Milaneschi Y, Schranner D, Kastenmuller G, Arnold M, Han X, et al. The metabolome-wide signature of major depressive disorder. Mol Psychiatry. 2024:1–12. Osimo EF, Baxter LJ, Lewis G, Jones PB, Khandaker GM. Prevalence of low-grade inflammation in depression: a systematic review and meta-analysis of CRP levels. Psychological Medicine. 2019;49:1958–1970. Mac Giollabhui N, Ng TH, Ellman LM, Alloy LB. The longitudinal associations of inflammatory biomarkers and depression revisited: systematic review, meta-analysis, and meta-regression. Mol Psychiatry. 2021;26:3302–3314. Osimo EF, Pillinger T, Rodriguez IM, Khandaker GM, Pariante CM, Howes OD. Inflammatory markers in depression: A meta-analysis of mean differences and variability in 5,166 patients and 5,083 controls. Brain Behav Immun. 2020;87:901–909. Milaneschi Y, Kappelmann N, Ye Z, Lamers F, Moser S, Jones PB, et al. Association of inflammation with depression and anxiety: evidence for symptom-specificity and potential causality from UK Biobank and NESDA cohorts. Molecular Psychiatry. 2021;26:7393. Kappelmann N, Arloth J, Georgakis MK, Czamara D, Rost N, Ligthart S, et al. Dissecting the Association Between Inflammation, Metabolic Dysregulation, and Specific Depressive Symptoms: A Genetic Correlation and 2-Sample Mendelian Randomization Study. JAMA Psychiatry. 2021;78:161–170. Suneson K, Lindahl J, Chamli Hårsmar S, Söderberg G, Lindqvist D. Inflammatory Depression—Mechanisms and Non-Pharmacological Interventions. International Journal of Molecular Sciences. 2021;22:1640. Bergstedt J, Pasman JA, Ma Z, Harder A, Yao S, Parker N, et al. Distinct biological signature and modifiable risk factors underlie the comorbidity between major depressive disorder and cardiovascular disease. Nat Cardiovasc Res. 2024;3:754–769. Li H, Zheng D, Li Z, Wu Z, Feng W, Cao X, et al. Association of Depressive Symptoms With Incident Cardiovascular Diseases in Middle-Aged and Older Chinese Adults. JAMA Network Open. 2019;2:e1916591. Borsboom D, Deserno MK, Rhemtulla M, Epskamp S, Fried EI, McNally RJ, et al. Network analysis of multivariate data in psychological science. Nat Rev Methods Primers. 2021;1:1–18. Fried EI, Stockert S von, Haslbeck JMB, Lamers F, Schoevers RA, Penninx BWJH. Using network analysis to examine links between individual depressive symptoms, inflammatory markers, and covariates. Psychological Medicine. 2020;50:2682–2690. Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779. Löwe B, Unützer J, Callahan CM, Perkins AJ, Kroenke K. Monitoring depression treatment outcomes with the patient health questionnaire-9. Med Care. 2004;42:1194–1201. Davis KAS, Coleman JRI, Adams M, Allen N, Breen G, Cullen B, et al. Mental health in UK Biobank - development, implementation and results from an online questionnaire completed by 157 366 participants: a reanalysis. BJPsych Open. 2020;6:e18. Townsend P, Phillimore P, Beattie A. Health and Deprivation: Inequality and the North. London: Routledge; 2023. Soininen P, Kangas AJ, Würtz P, Suna T, Ala-Korpela M. Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Cardiovascular Epidemiology and Genetics. Circulation: Cardiovascular Genetics. 2015;8:192–206. Rydin AO, Milaneschi Y, Quax R, Li J, Bosch JA, Schoevers RA, et al. A network analysis of depressive symptoms and metabolomics. Psychological Medicine. 2023;53:7385–7394. de Kluiver H, Jansen R, Milaneschi Y, Bot M, Giltay EJ, Schoevers R, et al. Metabolomic profiles discriminating anxiety from depression. Acta Psychiatrica Scandinavica. 2021;144:178–193. Dowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK, et al. A meta-analysis of cytokines in major depression. Biol Psychiatry. 2010;67:446–457. Little RJA. A Test of Missing Completely at Random for Multivariate Data with Missing Values. Journal of the American Statistical Association. 1988;83:1198–1202. Cortes D. Isolation forests: looking beyond tree depth. 2021. Anderson MJ. Permutation tests for univariate or multivariate analysis of variance and regression. Can J Fish Aquat Sci. 2001;58:626–639. Newman MEJ. Introduction: A short introduction to networks and why we study them. In: Newman M, editor. Networks: An Introduction, Oxford University Press; 2010. p. 0. Mcglashan J, Johnstone M, Creighton D, de la Haye K, Allender S. Quantifying a Systems Map: Network Analysis of a Childhood Obesity Causal Loop Diagram. PLOS ONE. 2016;11. Preacher KJ, Hayes AF. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods. 2008;40:879–891. de Kluiver H, Jansen R, Penninx BWJH, Giltay EJ, Schoevers RA, Milaneschi Y. Metabolomics signatures of depression: the role of symptom profiles. Transl Psychiatry. 2023;13:1–10. Chiesa ST, Charakida M, Georgiopoulos G, Roberts JD, Stafford SJ, Park C, et al. Glycoprotein Acetyls: A Novel Inflammatory Biomarker of Early Cardiovascular Risk in the Young. Journal of the American Heart Association. 2022;11:e024380. Fung E, Chan EYS, Ng KH, Yu KM, Li H, Wang Y. Towards clinical application of GlycA and GlycB for early detection of inflammation associated with (pre)diabetes and cardiovascular disease: recent evidence and updates. Journal of Inflammation. 2023;20:32. Schwingshackl L, Hoffmann G. Monounsaturated fatty acids and risk of cardiovascular disease: synopsis of the evidence available from systematic reviews and meta-analyses. Nutrients. 2012;4:1989–2007. Khandaker GM, Zuber V, Rees JMB, Carvalho L, Mason AM, Foley CN, et al. Shared mechanisms between coronary heart disease and depression: findings from a large UK general population-based cohort. Mol Psychiatry. 2020;25:1477–1486. Zheng X, Chen Y, Lin S-Q, Liu T, Liu C-A, Ruan G-T, et al. The relationship between different fatty acids intake and the depressive symptoms: A population-based study. J Affect Disord. 2024;357:68–76. Sánchez-Villegas A, Martínez-González MA, Estruch R, Salas-Salvadó J, Corella D, Covas MI, et al. Mediterranean dietary pattern and depression: the PREDIMED randomized trial. BMC Medicine. 2013;11:208. Kim KH, Kim Y, Seo KW. Efficacy of monounsaturated fatty acids in reducing risk of the cardiovascular diseases, cancer, inflammation, and insulin resistance: a narrative review. Ann Clin Nutr Metab. 2023;15:2–7. Case SM, Sawhney M, Stewart JC. Atypical depression and double depression predict new-onset cardiovascular disease in U.S. adults. Depression and Anxiety. 2018;35:10–17. Penninx BWJH, Lamers F, Jansen R, Berk M, Khandaker GM, De Picker L, et al. Immuno-metabolic depression: from concept to implementation. The Lancet Regional Health - Europe. 2025;48:101166. Additional Declarations The authors have declared there is NO conflict of interest to disclose Supplementary Files SupplementalOnlineContentMP.docx Supplemental information (figures and tables) Cite Share Download PDF Status: Under Review Version 1 posted Unknown event 18 Nov, 2025 Editorial decision: Reject after peer review 06 Oct, 2025 Review # 1 received at journal 05 Aug, 2025 Reviewer # 2 agreed at journal 01 Aug, 2025 Reviewer # 1 agreed at journal 24 Jul, 2025 Reviewers invited by journal 24 Jul, 2025 Editor assigned by journal 15 Apr, 2025 Submission checks completed at journal 15 Apr, 2025 First submitted to journal 11 Apr, 2025 Unknown event 11 Apr, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6422455","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":490495715,"identity":"97a6ca19-ac95-48e5-91da-269813ef8a59","order_by":0,"name":"Angela Koloi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAqUlEQVRIiWNgGAWjYFACNgaGBwwSchBOAbFaEhgkjCEcA+K1MCQ2EK2Fv4Et8UFChUX6hhvJD5gLiNEicYDtsEHCGYncDTfSDJhnEOWwA+xtEoltQC1nDhgw8xCjRR6qJd3gzPEPxGkxOMB2DKQlweB4D5G2GB5mSwb5xXDm8Z6Cw0RpkTveZvjgQ0WdPN9h9o2PeSqI0MLAjMQ+QIyGUTAKRsEoGAVEAABHPzAZ64+56gAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0009-0001-8197-3749","institution":"University of Amsterdam","correspondingAuthor":true,"prefix":"","firstName":"Angela","middleName":"","lastName":"Koloi","suffix":""},{"id":490495716,"identity":"d9252eee-b217-40ca-b5dc-f4f96761bc6e","order_by":1,"name":"Kevin Dobretz","email":"","orcid":"https://orcid.org/0000-0001-9138-4836","institution":"Geneva University Hospitals","correspondingAuthor":false,"prefix":"","firstName":"Kevin","middleName":"","lastName":"Dobretz","suffix":""},{"id":490495717,"identity":"72675cf9-e38b-440f-b36e-7f643e05895a","order_by":2,"name":"Dafne Damman","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Dafne","middleName":"","lastName":"Damman","suffix":""},{"id":490495718,"identity":"a731a28c-f022-4bb6-b5a2-90db497397e2","order_by":3,"name":"Rick Quax","email":"","orcid":"https://orcid.org/0000-0002-0299-0074","institution":"University of Amsterdam","correspondingAuthor":false,"prefix":"","firstName":"Rick","middleName":"","lastName":"Quax","suffix":""},{"id":490495719,"identity":"43ec66c0-8f07-472f-913e-8ddb337c573b","order_by":4,"name":"Dimitris Zaridis","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Dimitris","middleName":"","lastName":"Zaridis","suffix":""},{"id":490495720,"identity":"c0b4eeb1-2dd3-48c5-955a-e750b10ef5c6","order_by":5,"name":"Costas Papaloukas","email":"","orcid":"https://orcid.org/0000-0002-6736-5536","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Costas","middleName":"","lastName":"Papaloukas","suffix":""},{"id":490495721,"identity":"f5c1c06d-cf7c-47d4-93b7-a396623cccfc","order_by":6,"name":"Dimitrios Fotiadis","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Dimitrios","middleName":"","lastName":"Fotiadis","suffix":""},{"id":490495722,"identity":"b237e7e6-52ea-4945-8c41-a161112a192d","order_by":7,"name":"Jos Bosch","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Jos","middleName":"","lastName":"Bosch","suffix":""}],"badges":[],"createdAt":"2025-04-10 18:05:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6422455/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6422455/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":87726360,"identity":"6d557b13-eccc-490c-a602-6b16f826e3c3","added_by":"auto","created_at":"2025-07-28 10:51:22","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":207655,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHypothesized Pipeline:\u003c/strong\u003e(A) Tripartite network with three layers: (1) Individual symptoms of depression, (2) Metabolites, and (3) Atherosclerotic cardiovascular disease (CVD); (B) Mediation analysis examining the indirect link between depressive symptoms and CVD through metabolites.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-6422455/v1/24fe5993a97c44c174b80854.png"},{"id":87726361,"identity":"b06afa40-2d74-4b95-88e2-4fde71a63c73","added_by":"auto","created_at":"2025-07-28 10:51:22","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2588114,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eTripartite network with three layers: (1) Individual symptoms of depression, (2) Metabolites, and (3) CVD. Edges are color-coded to represent the direction of association: red for negative and blue for positive. Nodes are coloured to represent different groupings of depressive symptoms, metabolites (e.g., amino acids, apolipoproteins),and CVD. Only significant associations (p \u0026lt; .05) are shown, based on results of the stability analysis.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-6422455/v1/681a8372e9f354564dc4c705.png"},{"id":87726373,"identity":"6851d229-4a2d-4d13-a415-8b997d410e8d","added_by":"auto","created_at":"2025-07-28 10:51:22","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1034662,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eStable bridge metabolites linking depressive symptoms and CVD after permutation tests. Only significant associations (p \u0026lt; .05) are visualized.\u003c/em\u003e Blue indicates positive partial correlations, while red represents negative correlations.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-6422455/v1/8cb2a2f80668d52ebd64692d.png"},{"id":87726364,"identity":"a6c101bb-0423-4c40-9cb8-c0aadf8a72de","added_by":"auto","created_at":"2025-07-28 10:51:22","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2108084,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eForest plot of mediation analysis showing average causal mediation effects (ACME) with 95% confidence intervals across three covariate adjustment sets. Each horizontal bar represents a symptom-metabolite-CVD pathway, with bar length indicating the confidence interval. Bar colors denote different covariate adjustment sets. Squares (blue) represent models adjusted for age and gender; circles (green) show additional adjustment for socioeconomic factors (TDI), lifestyle behaviors (alcohol, smoking, physical activity), and antidepressant use; triangles (red) include further adjustment for BMI. Gray markers indicate statistically non-significant pathways (p≥0.05), while colored markers show significant effects (β values labeled; *p\u0026lt;0.05, **p\u0026lt;0.001). The dashed vertical line at zero represents the null effect. Pathways are ordered by symptoms → metabolite → CVD relationships.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-6422455/v1/e674c096f659d451db80c0d4.png"},{"id":87726359,"identity":"f2eff7bd-ec0d-4ba0-8f6a-f5ac305873da","added_by":"auto","created_at":"2025-07-28 10:51:22","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1502288,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eForest plot of mediation analysis showing average causal mediation effects (ACME) with 95% confidence intervals across three covariate adjustment sets. Each horizontal bar represents a symptom-metabolite-CVD pathway, with bar length indicating the confidence interval. Bar colors denote different covariate adjustment sets. Squares (blue) represent models adjusted for age and gender; circles (green) show additional adjustment for socioeconomic factors (TDI), lifestyle behaviors (alcohol, smoking, physical activity), and antidepressant use; triangles (red) include further adjustment for BMI. Gray markers indicate statistically non-significant pathways (p≥0.05), while colored markers show significant effects (β values labeled; *p\u0026lt;0.05). The dashed vertical line at zero represents the null effect. Pathways are ordered by PHQ-9 sum score → metabolite → CVD relationships.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-6422455/v1/dddfa3acfa985c90c51b63e2.png"},{"id":92839569,"identity":"b0abfae6-6d20-4c73-aee2-92be49cd9de1","added_by":"auto","created_at":"2025-10-06 08:39:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":7068188,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6422455/v1/ec0729a9-0500-43df-8be3-5eea85063946.pdf"},{"id":87726370,"identity":"88acca19-6479-4b72-b6c5-89516f30c0f9","added_by":"auto","created_at":"2025-07-28 10:51:22","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":27732136,"visible":true,"origin":"","legend":"Supplemental information (figures and tables)","description":"","filename":"SupplementalOnlineContentMP.docx","url":"https://assets-eu.researchsquare.com/files/rs-6422455/v1/114c7e36f6d8e26ae96f3f3f.docx"}],"financialInterests":"The authors have declared there is \u003cb\u003eNO\u003c/b\u003e conflict of interest to disclose","formattedTitle":"Metabolomic Signatures Linking Depressive Symptoms to Atherosclerotic Cardiovascular Disease Comorbidity: Evidence from the UK Biobank","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMajor Depressive Disorder (MDD), the most prevalent psychiatric disorder globally, frequently co-occurs with cardiovascular diseases (CVD) [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e], exacerbating disease severity and mortality risk [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Both conditions impose a substantial individual and societal burden in part due to their chronic nature [\u003cspan additionalcitationids=\"CR5 CR6\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Evidence suggests that their co-morbidity may involve shared immune or metabolic mechanisms, particularly in lipid metabolism and inflammatory pathways [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eLarge-scale metabolomics studies have consistently identified a distinctive inflammatory and metabolic profile associated with depression [\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], characterized by elevated alpha-1-glycoprotein acetyls (AGP), triglycerides, very-low-density lipoproteins (VLDL), and reduced levels of high-density lipoproteins (HDL). Additionally, meta-analyses and systematic reviews [\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] have further confirmed elevated levels of inflammatory markers in depression, supporting an inflammatory subtype associated with increased cardiovascular risk [\u003cspan additionalcitationids=\"CR16\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Genomic studies indicate that genetic liability for MDD elevates atherosclerotic CVD risk, with causal effects partially mediated by shared metabolic factors [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eResearch into depression-CVD comorbidity, however, faces three significant knowledge gaps. Firstly, most studies separately examine the biological pathways underlying either depression or CVD, with limited integration of shared or mutually interactive mechanisms. Second, there is a scarcity of longitudinal research explicitly investigating whether such shared metabolic pathways indeed link depression to subsequent CVD. Thirdly, most research relies on aggregate measures of overall depression severity, potentially masking the distinct contributions of individual symptoms to cardiovascular risk. For instance, specific depressive symptoms (e.g., fatigue, sleep disturbances) may independently or interactively influence cardiovascular outcomes [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e] highlighting the importance of considering symptom heterogeneity to inform targeted interventions.\u003c/p\u003e \u003cp\u003eTo address these gaps, our study employed a tripartite network analysis aiming to simultaneously examine interactions among depressive symptoms, metabolic biomarkers, and atherosclerotic cardiovascular disease (CVD, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). This analytical approach allowed for the identification of specific metabolic pathways that may mediate the progression from depression to incident CVD; a hypothesis subsequently tested through longitudinal mediation analyses (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). Network analysis is uniquely suited to addressing heterogeneity within depression [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], enabling the identification of symptom-level relationships within a complex system accounting for all variables. By utilizing an advanced integrative data analysis framework combining cross-sectional and longitudinal methods, this study aimed to identify biomarkers bridging depression and CVD, thereby enhancing our understanding of the mechanisms of comorbidity and potentially informing targeted intervention strategies.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDataset\u003c/h2\u003e \u003cp\u003eThe UK Biobank (UKB) is a large-scale population-based cohort consisting of approximately 500 000 participants aged 40 to 69 years, recruited between 2006 and 2010 from 22 assessment centres across England, Scotland, and Wales. The cohort was designed to include a diverse mix of urban and rural residents, reflecting a broad socioeconomic demographic [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Analyses were limited to 157 286 participants with available data on a mental health questionnaire (MHQ), which included the PHQ-9 (see below). We excluded participants diagnosed with schizophrenia or form of psychosis, personality disorders, and manic episodes associated with bipolar disorder, resulting in an analytical sample of 155 649 individuals. Nuclear magnetic resonance (NMR) metabolite data (see Methods section: \u003cem\u003ePlasma Biomarker Profiling by NMR\u003c/em\u003e) were available for a subset of UKB participants (N\u0026thinsp;=\u0026thinsp;118 461). The final study sample comprised 35 711 participants after ensuring the availability of NMR metabolite data and further exclusion of missing data (attributes with more than 15% missing values).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eAssessment\u003c/h3\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003ePatient Health Questionnaire-9 (PHQ-9)\u003c/h2\u003e \u003cp\u003eWe assessed depressive symptoms using both individual items and the total PHQ-9 sum score [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. The PHQ-9 evaluates nine symptoms on a 4-point Likert scale (0 = \"not at all\" to 3 = \"nearly every day\") (Supplementary Table\u0026nbsp;1; [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] The sum score (range: 0\u0026ndash;27) was calculated by summing all item scores, with higher scores indicating greater depression severity.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eAtherosclerotic Cardiovascular Disease (ACVD)\u003c/h3\u003e\n\u003cp\u003eThe presence of CVD at the time of depression assessment and throughout a 6-year follow-up period was assessed by retrieving the earliest date of a CVD outcome from twelve fields in the UKB dataset. Dates of CVD events were primarily retrieved from hospital inpatient records using International Classification of Diseases-10 (ICD-10) summary diagnoses (data fields 41202 (primary diagnoses), 41204 (secondary diagnoses), 41270 and 41280 (first-time diagnoses with names and dates)), from which we identified conditions characterized by atherosclerosis (see Supplementary Table\u0026nbsp;2 for the complete list of ICD-10 codes). Additionally, CVD outcomes were confirmed through death records, including date of death (field 40000) and primary and secondary causes of death (fields 40001, 40002). Four data fields directly reporting the dates of the first events for ST-elevation myocardial infraction (STEMI, field 42002), non-STEMI (NSTEMI, field 42004), ischaemic stroke (field 42008) and angina pectoris (field 131296) were combined with those extracted from inpatient records.\u003c/p\u003e \u003cp\u003eDates from these fields were merged to determine the earliest CVD event for each participant, relative to the time of depression assessment and the 6-year follow-up. The outcome was defined as a binary score (yes/no) indicating whether a participant developed CVD during the follow-up period. At the time of depression assessment, 2 131 participants had existing CVD cases. Excluding these, 1 326 new CVD cases were identified during the 6-year follow-up (see Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cem\u003eDescriptive statistics in UK Biobank.\u003c/em\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003e\u003cem\u003eMental health questionnaire assessment\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCharacteristics\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003esubset with plasma NMR\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003eN\u0026thinsp;=\u003c/b\u003e\u0026thinsp;\u003cb\u003e35 711\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eMissing, %\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAge\u003c/b\u003e \u003cem\u003eyears (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e63.98 (\u003cem\u003e\u0026plusmn;\u003c/em\u003e\u0026thinsp;7.74)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSex (F)\u003c/b\u003e \u003cem\u003e(%)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e56.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eTownsend Deprivation Index\u003c/b\u003e \u003cem\u003e(%)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.14\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eaverage\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003emost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12.36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eleast\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e56.55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eLifestyles\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSmoking status\u003c/b\u003e \u003cem\u003e(%)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.23\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003enever\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e57.43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eformer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e35.34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecurrent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAlcohol consumption\u003c/b\u003e \u003cem\u003e(%)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.16\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003enever\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003emonthly or less\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12.93\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etwo to four times a month\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18.66\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003etwo to three times a week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e29.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003efour or more times a week\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eModerate Physical activity\u003c/b\u003e \u003cem\u003e(%)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e85.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBMI\u003c/b\u003e \u003cem\u003e(kg/m\u003c/em\u003e\u003csup\u003e\u003cem\u003e2\u003c/em\u003e\u003c/sup\u003e\u003cem\u003e) (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e26.77(\u003cem\u003e\u0026plusmn;\u003c/em\u003e\u0026thinsp;4.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.28\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMedication\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAntidepressant use\u003c/b\u003e \u003cem\u003e(%)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHealth indicators\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePHQ-9\u0026thinsp;\u0026gt;\u0026thinsp;=\u0026thinsp;10\u003c/b\u003e \u003cem\u003e(%)\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e66.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAtherosclerotic CVD\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN\u0026thinsp;=\u0026thinsp;2 131\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c3\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003e6 years follow-up assessment\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAtherosclerotic CVD\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN\u0026thinsp;=\u0026thinsp;1 326\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e\n\u003ch3\u003eCovariates\u003c/h3\u003e\n\u003cp\u003eCovariates descriptions of UKB are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. In addition to sex (data field 31) and age (data field 21003), sociodemographic status was assessed using the Townsend Deprivation Index (TDI). The TDI (data field 22189) is based on the individual\u0026rsquo;s postcode, which leverages census data to incorporate information on employment, house and car ownership, and household crowing [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Smoking (data field 20116) was self-reported as \u0026ldquo;prefer not to answer\u0026rdquo;, \u0026ldquo;never\u0026rdquo;, \u0026ldquo;previous\u0026rdquo; or \u0026ldquo;current\u0026rdquo;. Frequency of drinking alcohol (data field 20414) use was self-reported as \u0026ldquo;prefer not to answer\u0026rdquo;, \u0026ldquo;never\u0026rdquo;, \u0026ldquo;four or more times a week\u0026rdquo;, \u0026ldquo;two to three times a week\u0026rdquo;, \u0026ldquo;two to four times a month\u0026rdquo;, \u0026ldquo;monthly or less\u0026rdquo;. Responses of \u0026ldquo;prefer not to answer\u0026rdquo; were treated as missing data and excluded from analysis. Moderate physical activity was assessed using self-reported data on the number of days per week participants engaged in moderate physical activity lasting 10\u0026thinsp;+\u0026thinsp;minutes (data field 884). Responses were categorized into a binary variable: participants reporting 0 days per week were classified as inactive, while those reporting 1 or more days per week were classified as moderate active. Antidepressant use was considered as a binary variable (1\u0026thinsp;=\u0026thinsp;participant used any of the antidepressants listed in Supplementary Table\u0026nbsp;3; 0\u0026thinsp;=\u0026thinsp;participant did not use any of these medications). BMI was calculated as weight (kg) divided by the square of height (m\u003csup\u003e2\u003c/sup\u003e) (data field 21001).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003ePlasma Biomarker Profiling by NMR\u003c/h2\u003e \u003cp\u003eBaseline plasma samples from UKB participants underwent biomarker profiling by Nightingale Health Plc., using nuclear magnetic resonance (NMR) spectroscopy. Details of the Nightingale Health NMR biomarker platform have been described [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eBiomarker data were available for a randomly selected subset of 118 461 ethylenediaminetetraacetic acid (EDTA)-treated plasma samples from UKB participants, providing a notably extensive metabolic dataset compared to earlier metabolomics studies. Each plasma sample yielded 249 metabolic measures, including 168 absolute concentrations and 81 ratio measures. For the present analyses, we chose to focus on 47 markers encompassing lipids, fatty acids (FAs), and low molecular weight metabolites. These biomarkers were chosen based on prior metabolomics research demonstrating their associations with depression [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. To avoid redundancy, we excluded additional sub-measures and ratios related to these lipoproteins, such as lipid composition and particle concentration of lipoprotein subclasses, as well as lipid and FAs ratios.\u003c/p\u003e \u003cp\u003eAdditionally, C-reactive protein (CRP) was included in the analyses due to strong supporting evidence from meta-analyses [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] and gene expression studies [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], consistently linking elevated CRP levels to key depressive symptoms. Biomarkers, their respective groupings, and associated units are described in Supplementary Table\u0026nbsp;4. Distribution plots of raw metabolite data and CRP are presented in Supplementary Fig.\u0026nbsp;1.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003ePre-process\u003c/h3\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eMissingness\u003c/h2\u003e \u003cp\u003eMissingness on demographics and health indicators can be found in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Based on the results of Little's MCAR test [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], the missing data was not completely at random (MCAR), hence we cannot drop these values. To handle the missing data, we implemented Multivariate Imputation (MI) using the iterative imputer of scikit-learn version 0.14.4. This approach effectively captures complex relationships between variables, such as non-linearity, by using a round-robin technique, imputing missing values, updating models, and re-imputing across multiple iterations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eOutliers\u003c/h2\u003e \u003cp\u003eWe applied the Isolation Forest algorithm [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e] to detect outliers. This algorithm isolates anomalies by randomly selecting a feature and splitting the data between its maximum and minimum values and repeating this process. The distribution of the outliers was visualized using box plots (Supplementary Fig.\u0026nbsp;2), and the identified outliers were subsequently removed from the dataset.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eNormalization\u003c/h2\u003e \u003cp\u003eData were pre-processed according to the manufacturer\u0026rsquo;s standardized protocol, previously applied in other large-scale studies [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Initially, any values exceeding 5 standard deviations from the mean were removed to eliminate extreme outliers. Subsequently, each data point was incremented by 1, followed by a natural log transformation to normalize the distribution. Finally, metabolites data underwent normalization to adjust all values to a range from 0\u0026ndash;1.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eAnalysis Plan\u003c/h2\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003eGraphical Model\u003c/h2\u003e \u003cp\u003eThe cross-sectional network included three groups: metabolites, symptoms, and CVD. We excluded intra-group connections (e.g., symptom-symptom or metabolite-metabolite) and focused solely on between-group associations (metabolite-symptom, metabolite-CVD). To construct the tripartite network, we employed partial correlation analysis to identify significant shared metabolites between individual symptoms of depression and CVD while adjusting for all other variables in the system. Partial correlation quantifies the relationship between two variables by removing the influence of all other variables. For ordinal depressive symptoms, we used Spearman partial correlation, which measures the monotonic relationship between variables after residualizing them with respect to all other variables (including other metabolites, depressive symptoms, and CVD). For binary CVD outcomes, we used point-biserial correlation, which evaluates the association between continuous metabolites and binary CVD after residualizing both with respect to all other variables. In both cases, residuals were obtained by performing regression analyses on the variables of interest with respect to all other variables, and correlations were computed between these residuals to reflect direct associations free from the influence of other variables. The resulting network represented variables as nodes and partial correlations as edges. This method assumes that the relationships among the variables are linear and that the residuals from the regression analysis are normally distributed.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eSensitivity Analyses\u003c/h2\u003e \u003cp\u003eTo assess the stability of the network edges, we performed bootstrap resampling with 1,000 iterations for each group. In each iteration: (i) data was resampled with replacement, (ii) partial correlations were calculated between all pairs of metabolites and symptoms, and (iii) significant edges (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.05) were recorded along with their correlation coefficients (either positive or negative). For each potential edge in the network, we computed: (i) the proportion of bootstrap samples in which the edge appeared (frequency), (ii) the average partial correlation coefficient across bootstrap samples (mean weight), and (iii) the standard deviation of the partial correlation coefficients (weight standard deviation). We considered edges that appeared in at least 75% of the bootstrap samples.\u003c/p\u003e \u003cp\u003eTo further assess the robustness of associations, a permutation test with 1,000 iterations was performed. For each pair, observed partial correlations were calculated using the original data. To test the null hypothesis of no association, symptom values were randomly permuted across samples while keeping metabolite values fixed [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Partial correlations were then recalculated for each permutation, generating a null distribution of correlation values. A permutation-based p-value (\u003cem\u003ep\u003c/em\u003e\u003csub\u003e\u003cem\u003eperm\u003c/em\u003e\u003c/sub\u003e​) was calculated as the proportion of permuted correlations with an absolute value equal to or greater than the observed correlation. Pairs with \u003cem\u003ep\u003c/em\u003e\u003csub\u003e\u003cem\u003eperm\u003c/em\u003e\u003c/sub\u003e​ \u0026lt; .05 were considered statistically significant.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eCentrality assessment\u003c/h2\u003e \u003cp\u003eTo determine key metabolites bridging symptoms and CVD, we calculated betweenness centrality, a widely recognized metric in network analysis [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. This metric quantifies how often a node (in this case, a metabolite) acts as a bridge between other nodes in the network. Betweenness centrality is calculated by determining the number of shortest paths between all pairs of nodes that pass-through a given node. Metabolites with high betweenness centrality play crucial roles in mediating information flow within the network, potentially linking depression symptoms to CVD [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. To validate the significance of central nodes, we employed a permutation-based approach. We first calculated betweenness centrality for each metabolite in the original network. We then created 1000 permuted networks by randomly shuffling depression symptoms and CVD risk data, generating a null distribution of centrality values. By comparing observed centralities to these null distributions, we computed \u003cem\u003ep\u003c/em\u003e\u003csub\u003e\u003cem\u003eperm\u003c/em\u003e\u003c/sub\u003e for each metabolite.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eMediation Analysis\u003c/h2\u003e \u003cp\u003eWe conducted a longitudinal mediation analysis to examine how individual depressive symptoms influenced CVD incidence six years later, mediated by identified bridge metabolites, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003eB. The mediation analysis is fundamentally based on regression analysis and involves a series of regression models to quantify the relationships between the independent variable (X; depressive symptoms), the mediator (M; metabolites), and the dependent variable (Y; CVD incidence).\u003c/p\u003e \u003cp\u003eThe total effect of depressive symptoms on CVD is estimated using a logistic regression model:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:logit\\left(P\\left(Y=1\\right)\\right)=\\:{\\beta\\:}_{0}+{\\beta\\:}_{1}X+\\:\\epsilon\\:,\\:\\epsilon\\::error\\:term$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eNext, the effect of depressive symptoms on the metabolites is estimated using a linear regression model (since the mediator is continuous):\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:M=\\:{\\beta\\:}_{0}+{\\beta\\:}_{2}X+\\epsilon\\:,\\:\\epsilon\\::error\\:term\\:\\:$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eFinally, the direct effect of depressive symptoms on CVD and the effect of the mediator on CVD are estimated using a logistic regression model:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:logit\\left(P\\left(Y=1\\right)\\right)=\\:{\\beta\\:}_{0}+{\\beta\\:}_{3}X+{\\beta\\:}_{4}M+\\:\\epsilon\\:,\\:\\epsilon\\::error\\:term$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe indirect effect is calculated as the product of the coefficients:\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:Indirect\\:Effect=\\:{{\\beta\\:}}_{2}+{{\\beta\\:}}_{4}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe mediation analysis incorporated three progressive models to account for potential confounders. The base model adjusted for age and gender. The second model expanded on this by additionally controlling for TDI, alcohol consumption, smoking, physical activity, and antidepressant use. The final model included all previous factors plus BMI. This stepped approach allowed us to isolate the mediating effects of metabolites on the depression-CVD relationship while systematically controlling for demographic, socioeconomic, lifestyle, and health factors.\u003c/p\u003e \u003cp\u003eTo robustly estimate indirect effects (ACME) and their 95% confidence intervals, we conducted non-parametric bootstrapping with 1000 resamples. Confidence intervals were derived using the percentile method [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e], which does not assume normality of the sampling distribution, thereby enhancing the reliability of results. Statistical analysis focused on key mediation metrics to clarify the relationships between depressive symptoms, metabolic biomarkers, and CVD. The average causal mediation effect (ACME) quantified the average change in CVD risk attributed to the mediator influenced by depressive symptoms. The average direct effects (ADE) measured the average impact of depressive symptoms on CVD, independent of the mediator. The total effect combined both indirect (ACME) and direct (ADE) effects, while the proportion mediated represents the fraction of the total effect on CVD explained by each mediator (metabolites). Bootstrapping analyses were performed in Python 3.11, utilising the statsmodels library (version 0.14.4).\u003c/p\u003e \u003cp\u003eWe conducted an additional longitudinal mediation analysis to examine the effect of the PHQ-9 total score on CVD incidence, mediated through the identified bridge metabolites. This secondary analysis enabled a direct comparison analysis of mediation effects derived from individual depressive symptoms versus the total PHQ-9 score. The same covariate sets were consistently applied across analyses to facilitate direct comparability of findings.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eDescriptives\u003c/h2\u003e \u003cp\u003eAfter merging depression assessments with NMR biomarker data and completing preprocessing steps (including missing data imputation, outlier detection, and normalization), the refined dataset comprised 33 925 participants (mean age: 64 years; 56% female). Within this sample, 66.01% of participants reported a PHQ-9 score of 10 or above, indicating moderate to severe depressive symptoms. Supplementary Table\u0026nbsp;5 provides a detailed breakdown of depressive symptom frequency and missing data rates. Supplementary Fig.\u0026nbsp;2 depicts the outlier analysis for metabolite measurements prior to normalization.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eNetwork and Sensitivity Analyses\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents the tripartite network constructed through partial correlation. Only stable connections among depressive symptoms, metabolites, and CVD are visualized. Stability and robustness of these associations were assessed through bootstrap resampling and permutation tests, presented in Supplementary Fig.\u0026nbsp;3 and Supplementary Fig.\u0026nbsp;4, respectively.\u003c/p\u003e \u003cp\u003ePermutation tests confirmed 11 significant bridge associations linking depressive symptoms and CVD (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.05), involving acetoacetate, CRP, creatinine, AGP, glutamine, histidine, linoleic acid, monounsaturated Fatty Acids (MUFAs), omega-6 FAs, total cholines, and tyrosine. Appetite changes appeared as the most frequently associated symptom (6 times), followed by sleep problems and fatigue (4 and 3 times, respectively). These associations, presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e3\u003c/span\u003e as partial correlations, formed symptom-metabolite-CVD triplets /pathways utilized subsequently in mediation analyses. Supplementary Fig.\u0026nbsp;5 presents the betweenness centrality assessment, highlighting citrate as the most central and statistically significant metabolite after the permutation test, followed by tyrosine, 3-hydroxybutyrate, acetate, and creatinine.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eMediation Analysis\u003c/h2\u003e \u003cdiv id=\"Sec22\" class=\"Section3\"\u003e \u003ch2\u003eExposure: Individual depressive symptoms\u003c/h2\u003e \u003cp\u003eUsing identified stable metabolite pathways from the network analysis, mediation analysis quantified longitudinal associations between depressive symptoms and incident CVD. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows the estimated indirect effects (ε) for symptom-metabolite combinations. Six significant mediation pathways initially emerged: appetite changes to CVD via CRP, creatinine, AGP, and MUFAs; sleep disturbances via MUFAs; and fatigue via AGP. After progressive covariate adjustments, including BMI, four pathways were retained: appetite changes were associated with CVD through AGP (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001) and MUFAs (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.05); sleep problems maintained their link to CVD via MUFAs (p\u0026thinsp;\u0026lt;\u0026thinsp;.05); and fatigue remained connected to CVD through AGP (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001). Detailed results of the mediation analysis are presented in Supplementary Table\u0026nbsp;6.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eExposure: PHQ-9 sum score\u003c/h2\u003e \u003cp\u003eA secondary mediation analysis using total PHQ-9 scores examined the same metabolite mediators: acetoacetate, CRP, creatinine, AGP, glutamine, histidine, linoleic acid, MUFAs, omega-6 FAs, total cholines, and tyrosine (identified as bridge). The PHQ-9 sum score's effect on CVD was significantly mediated by AGP (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.05) in the age and gender adjusted model, but this relationship did not persist after controlling for additional lifestyle factors and BMI.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study employed an integrative analytic approach combining cross-sectional network and longitudinal mediation analysis to identify metabolic pathways mediating the relationship between depressive symptoms and incident CVD. Contrasting symptom-level analyses with aggregate PHQ-9 score revealed greater specificity of symptom-based approaches, particularly linking energy-related depressive symptoms (appetite changes, fatigue, sleep disturbances) were linked to CVD via metabolic mediators (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e3\u003c/span\u003e). In line with our hypothesis, longitudinal mediation analysis confirmed and extended these findings, revealing six significant symptom-metabolite-CVD incidence pathways. After progressive covariate adjustments, AGP and MUFAs persisted as significant mediators. The distinct mediation effects observed for individual depressive symptoms, compared to total PHQ-9 scores, further underlined the value of symptom-level analysis, and suggest that individual symptoms may align with unique biological signatures influencing cardiovascular risk. Providing our observation with a sound empirical basis is that metabolic and inflammatory pathways have been implicated in both depression and CVD [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], and specifically research linking these pathways with \u0026ldquo;atypical, energy-related\u0026rdquo; symptom (AES) profile, which includes symptoms like hyperphagia, weight gain, hypersomnia, fatigue, and leaden paralysis [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAGP consistently mediated associations between appetite changes, fatigue, and CVD, exemplifying the broader observation that chronic low-grade inflammation may be a critical shared pathway linking CVD risk with atypical, energy-related depressive symptoms [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. While our analysis could not differentiate between increased or decreased appetite, previous literature indicates specifically increased appetite is associated with higher AGP levels [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. The observed AGP associations support its potential role as a stable and predictive biomarker of future cardiovascular and metabolic risks [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e], further emphasizing its clinical relevance.\u003c/p\u003e \u003cp\u003eSimilarly, MUFAs remained significant mediators linking depressive symptoms (particularly appetite changes and sleep disturbances) to later incident CVD. This finding aligns with the literature suggesting lipid metabolism and dietary fat intake play pivotal roles in the comorbidity between depression and CVD [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Diets rich in MUFAs, such as the Mediterranean diet, have consistently demonstrate cardiovascular and mental health benefits [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. For example, the PREDIMED trial demonstrates that a Mediterranean diet reduce depression risk [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e], which is backed by observational research [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. These protective effects likely involve anti-inflammatory actions, improved lipid profiles, and enhanced endothelial function underscoring the importance of further research into such shared nutritional determinants [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFurther investigations should consider bidirectional relationships, as metabolic alterations could both result from and exacerbate depressive symptoms, justifying further longitudinal studies exploring such reciprocal interactions. Moreover, future research could examine whether targeted nutritional or pharmacological interventions addressing inflammation and lipid metabolism can mitigate cardiovascular risk among depressed individuals, particularly those presenting atypical, energy-related symptoms.\u003c/p\u003e \u003cp\u003eDespite its strengths, including the use of robust statistical methods and a large representative cohort, this study has limitations. The inability to distinguish between increased and decreased appetite and subtypes of sleep disturbances (e.g., insomnia vs. hypersomnia), the analysis may obscure important distinctions in depressive subtypes such as such as atypical versus melancholic depression. Part of the DSM atypical depression specifier increased appetite and hypersomnia may underly divergent CVD risk [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFuture research should explore the synergistic interactions between multiple biological mediators in understanding the relationship between depressive symptoms and CVD comorbidity. By employing a joint mediation analysis, researchers can uncover novel pathways connecting simultaneously examine inflammatory markers, lipid metabolites, and neuroendocrine factors to uncover hidden pathways linking these complex health conditions.\u003c/p\u003e \u003cp\u003eIn conclusion, our findings substantiate the hypothesis of immunometabolic subtype of depression, which links low-grade inflammation and metabolic dysregulation to depression with atypical, energy-related symptoms [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. These disruptions may, in turn, contribute to the development of CVD. AGP and MUFAs emerged as critical mediators of the relationship between depressive symptoms and incident CVD highlighting the shared metabolic pathways. These findings align and expand previous observational and gene expression studies linking AGP [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan additionalcitationids=\"CR37\" citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e] and MUFAs [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e] to both depression and CVD. Clinically, the results point to a two-pronged approach: combining symptom-level analysis for personalized CVD prevention with interventions targeting metabolic pathways to reduce cardiovascular risk in depressed individuals with atypical, energy-related symptoms.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eConflict of Interest\u003c/h2\u003e \u003cp\u003eThe authors have no conflict of interest to disclose.\u003c/p\u003e \u003c/p\u003e\u003cp\u003e \u003ch2\u003eSupplementary Information\u003c/h2\u003e \u003cp\u003eSupplementary information is available at MP\u0026rsquo;s website.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAcknowledgements\u003c/h2\u003e \u003cp\u003eThis work was supported by the EU-project T0_AITION (grant number 848146).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGold SM, K\u0026ouml;hler-Forsberg O, Moss-Morris R, Mehnert A, Miranda JJ, Bullinger M, et al. Comorbid depression in medical diseases. Nat Rev Dis Primers. 2020;6:69.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBerk M, K\u0026ouml;hler-Forsberg O, Turner M, Penninx BWJH, Wrobel A, Firth J, et al. Comorbidity between major depressive disorder and physical diseases: a comprehensive review of epidemiology, mechanisms and management. World Psychiatry. 2023;22:366\u0026ndash;387.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHare DL, Toukhsati SR, Johansson P, Jaarsma T. Depression and cardiovascular disease: a clinical review. European Heart Journal. 2014;35:1365\u0026ndash;1372.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlmeida SS, Zizzi FB, Cattaneo A, Comandini A, Di Dato G, Lubrano E, et al. Management and Treatment of Patients With Major Depressive Disorder and Chronic Diseases: A Multidisciplinary Approach. Front Psychol. 2020;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeaton C, Froelicher ES, Wu LH, Ho C, Shishani K, Jaarsma T. The Global Burden of Cardiovascular Disease. European Journal of Cardiovascular Nursing. 2011;10:S5\u0026ndash;S13.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoseph P, Leong D, McKee M, Anand SS, Schwalm J-D, Teo K, et al. Reducing the Global Burden of Cardiovascular Disease, Part 1. Circulation Research. 2017;121:677\u0026ndash;694.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu J, Liu Y, Ma W, Tong Y, Zheng J. Temporal and spatial trend analysis of all-cause depression burden based on Global Burden of Disease (GBD) 2019 study. Sci Rep. 2024;14:12346.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePenninx BWJH. Depression and cardiovascular disease: Epidemiological evidence on their linking mechanisms. Neuroscience \u0026amp; Biobehavioral Reviews. 2017;74:277\u0026ndash;286.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJulkunen H, Cichońska A, Tiainen M, Koskela H, Nybo K, M\u0026auml;kel\u0026auml; V, et al. Atlas of plasma NMR biomarkers for health and disease in 118,461 individuals from the UK Biobank. Nat Commun. 2023;14:604.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBot M, Milaneschi Y, Al-Shehri T, Amin N, Garmaeva S, Onderwater GLJ, et al. Metabolomics Profile in Depression: A Pooled Analysis of 230 Metabolic Markers in 5283 Cases With Depression and 10,145 Controls. Biol Psychiatry. 2020;87:409\u0026ndash;418.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJansen R, Milaneschi Y, Schranner D, Kastenmuller G, Arnold M, Han X, et al. The metabolome-wide signature of major depressive disorder. Mol Psychiatry. 2024:1\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOsimo EF, Baxter LJ, Lewis G, Jones PB, Khandaker GM. Prevalence of low-grade inflammation in depression: a systematic review and meta-analysis of CRP levels. Psychological Medicine. 2019;49:1958\u0026ndash;1970.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMac Giollabhui N, Ng TH, Ellman LM, Alloy LB. The longitudinal associations of inflammatory biomarkers and depression revisited: systematic review, meta-analysis, and meta-regression. Mol Psychiatry. 2021;26:3302\u0026ndash;3314.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOsimo EF, Pillinger T, Rodriguez IM, Khandaker GM, Pariante CM, Howes OD. Inflammatory markers in depression: A meta-analysis of mean differences and variability in 5,166 patients and 5,083 controls. Brain Behav Immun. 2020;87:901\u0026ndash;909.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMilaneschi Y, Kappelmann N, Ye Z, Lamers F, Moser S, Jones PB, et al. Association of inflammation with depression and anxiety: evidence for symptom-specificity and potential causality from UK Biobank and NESDA cohorts. Molecular Psychiatry. 2021;26:7393.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKappelmann N, Arloth J, Georgakis MK, Czamara D, Rost N, Ligthart S, et al. Dissecting the Association Between Inflammation, Metabolic Dysregulation, and Specific Depressive Symptoms: A Genetic Correlation and 2-Sample Mendelian Randomization Study. JAMA Psychiatry. 2021;78:161\u0026ndash;170.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuneson K, Lindahl J, Chamli H\u0026aring;rsmar S, S\u0026ouml;derberg G, Lindqvist D. Inflammatory Depression\u0026mdash;Mechanisms and Non-Pharmacological Interventions. International Journal of Molecular Sciences. 2021;22:1640.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBergstedt J, Pasman JA, Ma Z, Harder A, Yao S, Parker N, et al. Distinct biological signature and modifiable risk factors underlie the comorbidity between major depressive disorder and cardiovascular disease. Nat Cardiovasc Res. 2024;3:754\u0026ndash;769.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi H, Zheng D, Li Z, Wu Z, Feng W, Cao X, et al. Association of Depressive Symptoms With Incident Cardiovascular Diseases in Middle-Aged and Older Chinese Adults. JAMA Network Open. 2019;2:e1916591.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBorsboom D, Deserno MK, Rhemtulla M, Epskamp S, Fried EI, McNally RJ, et al. Network analysis of multivariate data in psychological science. Nat Rev Methods Primers. 2021;1:1\u0026ndash;18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFried EI, Stockert S von, Haslbeck JMB, Lamers F, Schoevers RA, Penninx BWJH. Using network analysis to examine links between individual depressive symptoms, inflammatory markers, and covariates. Psychological Medicine. 2020;50:2682\u0026ndash;2690.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL\u0026ouml;we B, Un\u0026uuml;tzer J, Callahan CM, Perkins AJ, Kroenke K. Monitoring depression treatment outcomes with the patient health questionnaire-9. Med Care. 2004;42:1194\u0026ndash;1201.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavis KAS, Coleman JRI, Adams M, Allen N, Breen G, Cullen B, et al. Mental health in UK Biobank - development, implementation and results from an online questionnaire completed by 157 366 participants: a reanalysis. BJPsych Open. 2020;6:e18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTownsend P, Phillimore P, Beattie A. Health and Deprivation: Inequality and the North. London: Routledge; 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSoininen P, Kangas AJ, W\u0026uuml;rtz P, Suna T, Ala-Korpela M. Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Cardiovascular Epidemiology and Genetics. Circulation: Cardiovascular Genetics. 2015;8:192\u0026ndash;206.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRydin AO, Milaneschi Y, Quax R, Li J, Bosch JA, Schoevers RA, et al. A network analysis of depressive symptoms and metabolomics. Psychological Medicine. 2023;53:7385\u0026ndash;7394.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Kluiver H, Jansen R, Milaneschi Y, Bot M, Giltay EJ, Schoevers R, et al. Metabolomic profiles discriminating anxiety from depression. Acta Psychiatrica Scandinavica. 2021;144:178\u0026ndash;193.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDowlati Y, Herrmann N, Swardfager W, Liu H, Sham L, Reim EK, et al. A meta-analysis of cytokines in major depression. Biol Psychiatry. 2010;67:446\u0026ndash;457.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLittle RJA. A Test of Missing Completely at Random for Multivariate Data with Missing Values. Journal of the American Statistical Association. 1988;83:1198\u0026ndash;1202.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCortes D. Isolation forests: looking beyond tree depth. 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAnderson MJ. Permutation tests for univariate or multivariate analysis of variance and regression. Can J Fish Aquat Sci. 2001;58:626\u0026ndash;639.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNewman MEJ. Introduction: A short introduction to networks and why we study them. In: Newman M, editor. Networks: An Introduction, Oxford University Press; 2010. p. 0.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcglashan J, Johnstone M, Creighton D, de la Haye K, Allender S. Quantifying a Systems Map: Network Analysis of a Childhood Obesity Causal Loop Diagram. PLOS ONE. 2016;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePreacher KJ, Hayes AF. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods. 2008;40:879\u0026ndash;891.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Kluiver H, Jansen R, Penninx BWJH, Giltay EJ, Schoevers RA, Milaneschi Y. Metabolomics signatures of depression: the role of symptom profiles. Transl Psychiatry. 2023;13:1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChiesa ST, Charakida M, Georgiopoulos G, Roberts JD, Stafford SJ, Park C, et al. Glycoprotein Acetyls: A Novel Inflammatory Biomarker of Early Cardiovascular Risk in the Young. Journal of the American Heart Association. 2022;11:e024380.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFung E, Chan EYS, Ng KH, Yu KM, Li H, Wang Y. Towards clinical application of GlycA and GlycB for early detection of inflammation associated with (pre)diabetes and cardiovascular disease: recent evidence and updates. Journal of Inflammation. 2023;20:32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchwingshackl L, Hoffmann G. Monounsaturated fatty acids and risk of cardiovascular disease: synopsis of the evidence available from systematic reviews and meta-analyses. Nutrients. 2012;4:1989\u0026ndash;2007.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhandaker GM, Zuber V, Rees JMB, Carvalho L, Mason AM, Foley CN, et al. Shared mechanisms between coronary heart disease and depression: findings from a large UK general population-based cohort. Mol Psychiatry. 2020;25:1477\u0026ndash;1486.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng X, Chen Y, Lin S-Q, Liu T, Liu C-A, Ruan G-T, et al. The relationship between different fatty acids intake and the depressive symptoms: A population-based study. J Affect Disord. 2024;357:68\u0026ndash;76.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS\u0026aacute;nchez-Villegas A, Mart\u0026iacute;nez-Gonz\u0026aacute;lez MA, Estruch R, Salas-Salvad\u0026oacute; J, Corella D, Covas MI, et al. Mediterranean dietary pattern and depression: the PREDIMED randomized trial. BMC Medicine. 2013;11:208.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim KH, Kim Y, Seo KW. Efficacy of monounsaturated fatty acids in reducing risk of the cardiovascular diseases, cancer, inflammation, and insulin resistance: a narrative review. Ann Clin Nutr Metab. 2023;15:2\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCase SM, Sawhney M, Stewart JC. Atypical depression and double depression predict new-onset cardiovascular disease in U.S. adults. Depression and Anxiety. 2018;35:10\u0026ndash;17.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePenninx BWJH, Lamers F, Jansen R, Berk M, Khandaker GM, De Picker L, et al. Immuno-metabolic depression: from concept to implementation. The Lancet Regional Health - Europe. 2025;48:101166.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"translational-psychiatry","isNatureJournal":false,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"tp","sideBox":"Learn more about [Translational Psychiatry](http://www.nature.com/tp/)","snPcode":"41398","submissionUrl":"https://mts-tp.nature.com/cgi-bin/main.plex","title":"Translational Psychiatry","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6422455/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6422455/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Depression and cardiovascular disease (CVD) frequently co-occur and exacerbate each other's clinical outcomes, potentially through shared immune or metabolic dysregulation. This study used tripartite network analysis to examine interactions among depressive symptoms, metabolic biomarkers, and CVD, aiming to identify mediating metabolic pathways. Using UK Biobank data (N = 35 711; 56% female; mean age 64), NMR-based plasma metabolites, PHQ-9 depressive symptoms, and ICD-10 atherosclerotic CVD were analyzed. A tripartite network (symptom-metabolite-CVD) was constructed using partial correlation analyses to identify metabolites shared between depression and CVD. These metabolites were then subsequently entered in a longitudinal mediation analysis, spanning 6 years, to quantify metabolite-mediated associations between depressive symptoms and incident CVD. Models progressively adjusted for: (1) age and sex; (2) socioeconomic status, lifestyle factors, and antidepressant use; and (3) BMI. Analyses were repeated using PHQ-9 total scores as the exposure. The initial unadjusted network analysis identified 11 shared metabolic markers. Longitudinal mediation revealed six symptom-metabolite-CVD pathways, whereas only one metabolite-mediated pathway emerged using total PHQ-9 scores. After full adjustment, four pathways remained significant: alpha-1-glycoprotein acetyls (AGP) mediated associations involving appetite changes and fatigue (both p \u003c .001), and monounsaturated fatty acids (MUFAs) mediated associations involving appetite changes and sleep disturbances (both p \u003c .05). AGP (an inflammatory acute-phase protein) and MUFAs link energy-related depressive symptoms (appetite changes, fatigue, sleep disturbances) to CVD. This symptom-level approach enhances precision in identifying shared biological mechanisms underlying the co-morbidity between depression and CVD, potentially informing novel avenues for tailored prevention and treatment strategies.","manuscriptTitle":"Metabolomic Signatures Linking Depressive Symptoms to Atherosclerotic Cardiovascular Disease Comorbidity: Evidence from the UK Biobank","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-28 10:51:17","doi":"10.21203/rs.3.rs-6422455/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"transferred","content":"Translational Psychiatry","date":"2025-11-19T00:20:40+00:00","index":"","fulltext":""},{"type":"decision","content":"Reject after peer review","date":"2025-10-06T08:29:23+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"This content is not available.","date":"2025-08-05T11:17:35+00:00","index":1,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2025-08-01T10:06:38+00:00","index":2,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2025-07-25T00:04:53+00:00","index":1,"fulltext":"This content is not available."},{"type":"reviewersInvited","content":"","date":"2025-07-24T18:58:49+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-04-15T09:53:46+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-04-15T09:40:11+00:00","index":"","fulltext":""},{"type":"submitted","content":"Molecular Psychiatry","date":"2025-04-11T10:05:16+00:00","index":"","fulltext":""},{"type":"checksFailed","content":"","date":"2025-04-11T09:46:23+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"translational-psychiatry","isNatureJournal":false,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"tp","sideBox":"Learn more about [Translational Psychiatry](http://www.nature.com/tp/)","snPcode":"41398","submissionUrl":"https://mts-tp.nature.com/cgi-bin/main.plex","title":"Translational Psychiatry","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8419f220-6184-4df2-9614-7323e127c93b","owner":[],"postedDate":"July 28th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":52082240,"name":"Health sciences/Diseases/Psychiatric disorders/Depression"},{"id":52082241,"name":"Biological sciences/Molecular biology"}],"tags":[],"updatedAt":"2025-11-27T15:55:58+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-28 10:51:17","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6422455","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6422455","identity":"rs-6422455","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}