Sugar-sweetened beverage consumption and incident depression: an exploratory multi-omics analysis of candidate biological mediators Authors: Zhirong Li, Yating Miao, Ting Yu, Lina Jin, Yanan Ma, Xinyao Zhang Published: 2026 ABSTRACT BackgroundDepression is a leading cause of mental and physical disability globally, with its onset and progression influenced by a complex interplay of dietary, psychological, and biological factors. Recent research suggests a link between sugar-sweetened beverage (SSB) consumption and depression risk, although the potential biological pathways underlying this association remain poorly understood.MethodsThis study utilized data from 192,045 participants in the UK Biobank to examine the prospective association between SSB consumption and incident depression using Cox proportional hazards models. SSBs were defined as the sum of five beverage categories assessed via the Oxford WebQ 24-hour dietary recall. Directional consistency of the association was further examined across three external supporting datasets encompassing diverse populations: NHANES, YRBSS, and the Lianyungang Municipal School Health and Risk Factor Surveillance Study Dataset. We further investigated whether proteins, metabolites, inflammatory markers, and brain imaging phenotypes may serve as candidate mediators statistically consistent with mediation of the SSB-depression association.ResultsHigh SSB consumption was associated with an 18% higher risk of incident depression compared with non-consumers (HR = 1.18; 95% CI: 1.11–1.25), with consistent directional associations observed across external supporting datasets. A plasma proteomic signature comprising 229 proteins was constructed using elastic net regularization and was associated with an increased risk of incident depression. Exploratory mediation analyses identified 72 proteins, 36 metabolites, and 5 inflammatory markers as candidate mediators, with IL1RN showing the strongest protein-level candidate mediating effect (9.6%), and Unsaturation and neutrophil count showing the strongest metabolite- and inflammatory marker-level effects, respectively.ConclusionsThis study provides preliminary evidence that proteins, metabolites, and inflammatory markers may serve as candidate mediators statistically consistent with mediation of the association between SSB consumption and incident depression. These findings are exploratory and hypothesis-generating, and future experimental studies are needed to validate these candidate pathways and assess their potential as targets for dietary interventions in depression prevention.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12937-026-01352-4. Introduction Depression is a mental disorder characterized by low mood or a prolonged loss of interest in activities, and it has become a leading cause of mental and physical disability globally [1, 2]. Over the past 30 years, the incidence of depression worldwide has increased by nearly 59% [3]. Notably, the initial onset of depression is often accompanied by a high risk of recurrence, and its recurrence rate persists throughout the entire lifespan [4]. The occurrence and progression of depression result from a complex interplay of dietary, psychological, genetic, and biological factors [5–7]. Given the lack of definitive treatments and the substantial burden of the disease, identifying modifiable risk factors to prevent or delay the onset of depression represents an urgent public health priority. Dietary habits, particularly the consumption of sugar-sweetened beverages (SSB), have increased dramatically worldwide [8–10], a trend that mirrors the concurrent rise in metabolic and mental health disorders [11], and thus have attracted considerable research interest as a potentially modifiable risk factor. Evidence from prospective studies indicates that high SSB consumption is associated with an elevated risk of depression; however, existing studies are subject to several limitations. First, many studies examining the association between SSB consumption and depression have focused predominantly on single, homogeneous populations, lacking sufficient data across diverse age groups, ethnicities, and cultural backgrounds [12]. Consequently, it remains unclear whether and to what extent SSB intake influences depression risk across different demographic groups. Second, SSBs may be associated with the development of depression through alterations in key biomarkers implicated in its pathophysiology [13, 14]. However, no study to date has systematically investigated the potential mediating roles of proteins, metabolites, inflammatory markers, or brain structural changes in the relationship between SSB consumption and depression. Third, the effect of SSBs on depression may be mediated through multiple interconnected biological processes, encompassing the molecular functions, cellular components, and biological pathways of relevant proteins [15]. A deeper understanding of the biological functions of these potentially mediating proteins would help elucidate the underlying biological mechanisms linking SSB consumption to depression. Nevertheless, this aspect remains insufficiently addressed in the existing literature. Here, this study utilizes data from 192,045 participants in the UK Biobank (UKB) to explore the prospective association between SSB consumption and the risk of incident depression. Additionally, data from the National Health and Nutrition Examination Survey (NHANES), the Youth Risk Behavior Surveillance System (YRBSS), and the Lianyungang Municipal School Health and Risk Factor Surveillance Dataset (Jiangsu Province) are employed as external supporting datasets to further examine the directional consistency of the association between SSB consumption and depressive symptoms across diverse populations. To explore potential biological pathways, this research also investigates whether proteins, metabolites, inflammatory markers, and brain imaging phenotypes may serve as candidate mediators statistically consistent with mediation of this association, and further explores the biological functions of the relevant candidate mediating proteins. Results Participant characteristics We included 192,045 participants who had completed at least one dietary assessment and were free of depression at baseline. The mean age was 56.2 years (SD = 7.96), and 54.1% were women. The majority were of White ethnicity (91.1%). Over half of the participants (56.8%, 109,082 of 192,045) reported no SSB consumption, 57.2% had never smoked, and 6.0% reported lifetime abstinence from alcohol. Over a median follow-up of 12.2 years (IQR: 12.1–13.0; range 0.01–15.2; mean = 12.2), 6,589 participants (3.4%) developed incident depression. Detailed baseline characteristics are provided in Table S1. Association between SSB consumption, protein signature, and depression We examined the associations of SSB consumption and related protein signature with incident depression among participants free of depression at baseline using Cox proportional hazards regression models. After adjustment for baseline age, sex, ethnicity, body mass index, Townsend deprivation index, education, smoking status, alcohol intake, physical activity (IPAQ), PHQ-4 score, and total energy intake, participants with high SSB consumption had 18% higher risks of depression, compared with non-consumers (HR = 1.18; 95% CI: 1.11–1.25). No association was observed for low SSB consumption (HR = 0.98; 95% CI: 0.92–1.05, P = 0.619) (Fig. 1A; Table S5). Consistent directional associations were observed across external supporting datasets encompassing diverse populations with varying age ranges and geographic backgrounds, including NHANES (United States, 18–79 years), the Lianyungang Municipal Database (Jiangsu Province, China, school-aged to university students), and YRBSS (United States, ~ 12–18 years). For instance, in NHANES, each 1-SD increase in SSB consumption was associated with a 0.21-point increase in PHQ-9 score (Fig. 1B; Tables S6-S7). Baseline demographic characteristics for these cohorts are presented in Tables S2-S4. Fig. 1Associations between SSB consumption, proteomic signatures and depression. A Association between sugar-sweetened beverage (SSB) consumption and depression in the UK Biobank (UKB); grey lines denote 95% confidence intervals. B Replication of the observed association between SSB consumption and depression in independent cohorts: Lianyungang (the Lianyungang Municipal School Health and Risk Factor Surveillance Dataset, Jiangsu Province), NHANES (US National Health and Nutrition Examination Survey), and YRBSS (Youth Risk Behavior Surveillance System). *P < 0.05, **P < 0.01, ***P < 0.001. C Construction of the SSB-related proteomic signature. D, Importance of individual proteins in the modelling process estimated by elastic net. E Association between the SSB proteomic signature and depression; filled circles represent hazard ratios (HRs), with lines indicating 95% confidence intervals Associations between SSB consumption, proteomic signatures and depression. A Association between sugar-sweetened beverage (SSB) consumption and depression in the UK Biobank (UKB); grey lines denote 95% confidence intervals. B Replication of the observed association between SSB consumption and depression in independent cohorts: Lianyungang (the Lianyungang Municipal School Health and Risk Factor Surveillance Dataset, Jiangsu Province), NHANES (US National Health and Nutrition Examination Survey), and YRBSS (Youth Risk Behavior Surveillance System). *P < 0.05, **P < 0.01, ***P < 0.001. C Construction of the SSB-related proteomic signature. D, Importance of individual proteins in the modelling process estimated by elastic net. E Association between the SSB proteomic signature and depression; filled circles represent hazard ratios (HRs), with lines indicating 95% confidence intervals To characterize the molecular correlates of SSB consumption, elastic net (alpha = 0.1, L1 ratio = 0.1) was applied to construct an SSB protein signature. Among 2,923 proteins tested, 229 proteins were selected for the final signature (Fig. 1C). Of these, 112 proteins had positive weights, with MMP9, MIA, KLK13, and DNER showing the largest absolute coefficients, while 117 proteins had negative weights, with GHRL, ELN, FUT3-FUT5, and CNDP1 showing the largest absolute coefficients (Fig. 1D, Table S8). Gene Ontology (GO) analysis revealed enrichment in positive regulation of the MAPK cascade (Biological Process), external encapsulating structure and extracellular matrix (Cellular Component), and cytokine activity, glycosaminoglycan binding, sulfur compound binding, and endopeptidase activity (Molecular Function). Results of GO, Kyoto Encyclopedia of Genes and Genomes (KEGG), and protein-protein interaction (PPI) analyses are provided in Fig. S2 and Table S9-10. The SSB protein signature was examined both as a continuous variable and by quartiles. Each 1-SD increase in the signature was associated with a 14% higher risk of incident depression (HR = 1.14; 95% CI: 1.04–1.24, P < 0.001). Compared with the lowest quartile, participants in Q4 had 1.54-fold higher risks of depression (HR = 1.54; 95% CI: 1.20–1.98) (Fig. 1E, Table S5). To further evaluate the robustness of the primary association, we conducted sensitivity analyses stratifying by depression ascertainment source. The association between high SSB consumption and incident depression was most consistent and statistically significant among hospital-ascertained cases (HR = 1.19; 95% CI: 1.11–1.27; P < 0.001). Among self-reported cases, a modest association was observed for low SSB consumption (HR = 1.25; 95% CI: 1.05–1.48; P = 0.011). No significant associations were observed among primary care-ascertained cases, likely reflecting the smaller number of cases (N = 461) available from this source [16] (Fig. S1). Potential mediating role of proteins in the SSB-depression association We first observed that the SSB protein signature was statistically consistent with mediating 24.2% of the association between SSB consumption and incident depression (Fig. 2A). To identify specific candidate mediators, we performed protein-level mediation analyses across all 2,923 proteins. As shown previously, multiple proteins were associated with SSB consumption; here, we further tested their associations with incident depression. After FDR correction, 355 proteins were significantly associated with incident depression (Fig. 2B; Table S16). The intersection yielded 150 proteins that were significantly related to both SSB consumption and incident depression and were therefore included in mediation analyses. Fig. 2Mediating role of proteins in the association between SSB consumption and depression. A, Mediation effect of SSB-related proteomic signatures on the SSB-depression association. B, Volcano plot showing associations between 2,923 plasma proteins and depression. C, Proteins with significant mediating effects; the x axis represents βa (SSB-protein association) and the y axis represents βb (protein-depression association). D, Protein-protein interaction (PPI) network of mediating proteins. E, KEGG pathway enrichment analysis of mediating proteins. F, Gene Ontology (GO) enrichment analysis of mediating proteins Mediating role of proteins in the association between SSB consumption and depression. A, Mediation effect of SSB-related proteomic signatures on the SSB-depression association. B, Volcano plot showing associations between 2,923 plasma proteins and depression. C, Proteins with significant mediating effects; the x axis represents βa (SSB-protein association) and the y axis represents βb (protein-depression association). D, Protein-protein interaction (PPI) network of mediating proteins. E, KEGG pathway enrichment analysis of mediating proteins. F, Gene Ontology (GO) enrichment analysis of mediating proteins Of these, 72 proteins remained as candidate mediators after FDR correction. IL1RN exhibited the strongest candidate mediating effect, accounting for 9.6% of the SSB-depression association, followed by LGALS9 (7.5%), IGFBP4 (6.3%), and PLAUR (6.1%) (Fig. 2C, Table S11). Functional characterization of these 72 candidate mediating proteins revealed that, in PPI networks, CCL3 had the highest degree of connectivity, followed by IL6 (Fig. 2D). KEGG enrichment analysis identified cytokine-cytokine receptor interaction and PI3K-Akt signaling pathway as the top enriched pathways (Fig. 2E, Table S12). GO analysis highlighted leukocyte migration (Biological Process), extracellular matrix and external encapsulating structure (Cellular Component), and cytokine activity and cytokine receptor binding (Molecular Function) as the most enriched terms (Fig. 2F, Table S13). Potential mediating roles of additional biological variables and structural equation modeling We further conducted mediation analyses for 91 brain structural features, 170 metabolites, and 9 inflammatory markers using the same framework applied to proteins. In the brain structural analysis, we identified associations between SSB consumption and several brain structures; however, no significant associations were observed between brain structure and incident depression (Table S17). This lack of association likely reflects the limited number of incident depression cases (N = 150) following the exclusion of participants with pre-existing depression prior to the brain MRI wave, which may have resulted in insufficient statistical power. Among SSB-brain structure associations, 10 associations survived FDR correction across 48 cortical regions, with the strongest effect observed in the Temporal Fusiform Cortex (β = -0.022, 95% CI: -0.032 to -0.011, P = 5.72 × 10− 5). Among 15 subcortical structures, only the left hippocampus survived FDR correction (β = -0.018, 95% CI: -0.029 to -0.007, P = 0.002). Notably, 21 of the 28 cerebellar volumes showed significant associations with SSB consumption, with the strongest effects observed in lobules VIIIb (left), VIIb (left), and VIIIa (right) (β = -0.024, - 0.024, and - 0.023, respectively) (Fig. 3A, Table S17). Fig. 3Mediation analyses of additional biological variables and structural equation modeling (SEM). A, Associations between SSB consumption and brain structures; red indicates positive associations and blue indicates negative associations. B, C, Mediation effects of metabolites and inflammatory markers on the SSB–depression association; the x-axis represents βa (SSB-mediator association), the y-axis represents βb (mediator–depression association), and dot size reflects the proportion mediated. D, SEM depicting the pathway from SSB consumption to metabolism, inflammation, and depression; standardized coefficients and statistical significance are shown alongside each path. E, Key latent variables for metabolism and inflammation highlighted in the SEM Mediation analyses of additional biological variables and structural equation modeling (SEM). A, Associations between SSB consumption and brain structures; red indicates positive associations and blue indicates negative associations. B, C, Mediation effects of metabolites and inflammatory markers on the SSB–depression association; the x-axis represents βa (SSB-mediator association), the y-axis represents βb (mediator–depression association), and dot size reflects the proportion mediated. D, SEM depicting the pathway from SSB consumption to metabolism, inflammation, and depression; standardized coefficients and statistical significance are shown alongside each path. E, Key latent variables for metabolism and inflammation highlighted in the SEM In metabolite mediation analyses, 36 metabolites remained as candidate mediators after FDR correction (Table S18). The strongest candidate mediators were Unsaturation and XL-HDL-CE, explaining 3.8% and 1.7% of the SSB-depression association, respectively (Fig. 3B, Table S14). Of the 9 inflammatory markers examined, 5 remained as candidate mediators after FDR correction. Among these, neutrophil count accounted for the largest candidate mediation proportion (1.5%) (Fig. 3C, Table S15). In the SSB-neutrophil association, the effect estimate was β = 0.017 (95% CI = 0.013–0.022, P = 4.70 × 10− 14), while the neutrophil-depression association yielded an HR of 1.103 (95% CI = 1.077–1.129, P = 9.24 × 10− 15) (Fig. 3C; Table S19). We next integrated significant metabolites and inflammatory markers identified in mediation analyses into a structural equation model (SEM) to further examine the joint contributions of metabolic and inflammatory variables to the SSB-depression association. The SSB-metabolism-inflammation-depression pathways were statistically significant (standardized coefficients: -0.049 to 0.025; all P < 0.001) (Fig. 3D). Among metabolic variables, IDL-C was identified as the primary latent factor, followed by IDL-CE and IDL-FC. For inflammatory markers, neutrophil count, systemic immune-inflammation index (SII), and C-reactive protein (CRP) were identified as the main latent factors (Fig. 3E). Discussion In this large-scale prospective cohort study, we found that higher SSB consumption was associated with an increased risk of incident depression, with consistent directional associations observed across external supporting datasets encompassing diverse populations and heterogeneous outcome definitions. Further analyses suggested that a higher SSB protein signature was associated with an increased risk of incident depression. By integrating multimodal omics data, we identified 72 proteins, 36 metabolites, and 5 inflammatory markers as candidate mediators statistically consistent with mediation of the association between SSB consumption and incident depression, offering preliminary insights into plausible biological pathways that may be relevant to this association and warrant further investigation. Higher SSB consumption was associated with an increased risk of incident depression in the UKB. Consistent directional associations were observed across external supporting datasets encompassing diverse populations and heterogeneous outcome definitions, providing supporting evidence for the generalizability of this association across different populations and measurement approaches. Our results are consistent with those of several previous studies, collectively supporting the role of SSB consumption as a potentially modifiable risk factor for depression [12, 17–19]. To the best of our knowledge, this study is among the first to construct a plasma proteomic signature associated with SSB consumption and to examine its relationship with incident depression, extending the analysis beyond self-reported dietary behavior to the molecular level. This approach offers a more objective and quantifiable characterization of the internal biological correlates of SSB consumption, complementing self-reported dietary assessments and providing preliminary biological evidence for the observed SSB-depression association [20–22]. Our study provides preliminary evidence that proteins, metabolites, and inflammatory markers may serve as candidate mediators statistically consistent with mediation of the association between SSB consumption and incident depression. To our knowledge, this is among the first studies to systematically examine whether these pathways statistically mediate the SSB-depression association. Among the 72 candidate mediating proteins identified, IL1RN showed the strongest candidate mediating effect, potentially accounting for 9.6% of the association, followed by LGALS9. These findings highlight the potential importance of immune-metabolic dysregulation in the biological pathways that may be relevant to the association between SSB consumption to depression risk, and are consistent with the broader inflammatory hypothesis of depression. The inflammatory hypothesis of depression posits that dysregulation of immune and inflammatory processes plays an important role in the pathophysiology of depression [23–25]. Supporting this view, circulating inflammatory markers are frequently elevated in individuals with depression and have been shown to correlate positively with depressive symptom severity, particularly in clinical samples [26, 27]. The candidate mediating role of IL1RN identified in our study is biologically plausible in this context. A previous proteomic study of depression-related proteins found that IL1RN levels were significantly associated with depression, suggesting a potential role for this protein in regulating inflammatory responses relevant to depression risk [28, 29]. IL1RN encodes interleukin-1 receptor antagonist, an endogenous inhibitor of the pro-inflammatory interleukin-1 (IL-1) signaling pathway [30]. Its candidate mediating role in our analysis raises the possibility that SSB consumption may influence incident depression risk in part by modulating IL-1 family signaling and downstream neuroinflammatory pathways, though this hypothesis requires experimental validation [31]. This is further supported by proteomic studies of cerebrospinal fluid, which have identified elevated levels of inflammatory proteins in individuals with depression, suggesting that inflammatory dysregulation may influence mood regulation through neurotoxic processes [32]. The enrichment analyses provide preliminary evidence that the candidate mediating proteins are enriched in immune and inflammatory pathways, suggesting that immune dysregulation may represent a plausible biological pathway relevant to the association between SSB consumption and incident depression risk. The enrichment of leukocyte migration as the most enriched biological process is biologically plausible in this context. Peripheral immune cell migration across the blood-brain barrier is a hallmark of neuroinflammation, and dysregulation of this process has been implicated in the pathophysiology of depression [33, 34]. SSB consumption may potentially contribute to this process by promoting systemic inflammation that facilitates peripheral immune cell recruitment to the central nervous system, though this hypothesis requires experimental validation [35, 36]. In terms of molecular function, the enrichment of cytokine activity and cytokine receptor binding is consistent with the established role of cytokine signaling in depression pathophysiology. Circulating cytokines, including IL-1β, IL-6, and TNF-α, have been shown to activate microglia and astrocytes upon receptor binding, promoting neuroinflammation and dysregulation of the HPA axis [25]. These findings align with the broader cytokine hypothesis of depression, in which elevated peripheral cytokine levels contribute to central nervous system dysfunction through multiple neuroimmune pathways [37]. The enrichment of extracellular matrix and external encapsulating structure as the most enriched cellular components is also noteworthy. The extracellular matrix plays a critical role in maintaining neuronal microenvironment integrity, and alterations in ECM composition have been linked to neuronal dysfunction and mood disorders [38]. SSB-induced metabolic and inflammatory changes may potentially contribute to ECM remodelling, though the directionality and causal nature of this relationship in the context of depression remain to be established. KEGG enrichment analysis identified cytokine-cytokine receptor interaction as the most enriched pathway, consistent with the GO molecular function findings and further supporting the potential role of cytokine signaling networks in the candidate mediating proteins identified in our analysis [25, 39]. Taken together, these enrichment findings suggest that immune-metabolic dysregulation, particularly involving cytokine signaling, leukocyte migration, and extracellular matrix remodelling, may represent plausible biological pathways through which SSB consumption could influence the risk of incident depression [40]. These observations are preliminary and hypothesis-generating, and future studies integrating genomic and multi-omics data are needed to validate these candidate pathways and assess their potential as biomarkers or therapeutic targets [41]. Building on our finding that proteins may serve as candidate mediators statistically consistent with mediation of the SSB-depression association, we further examined whether metabolites and inflammatory markers show similar patterns. Among the 36 candidate mediating metabolites identified, Unsaturation and XL-HDL-CE showed the strongest candidate mediating effects, suggesting that SSB-associated lipid metabolic dysregulation may be statistically relevant to the risk of incident depression. Abnormalities in lipid metabolism, including alterations in cholesterol ester composition and lipid unsaturation, have been implicated in neuroinflammation and mood disorder pathophysiology through multiple mechanisms, including modulation of membrane fluidity, synaptic function, and neuroinflammatory signaling [42]. SSB consumption has been shown to alter circulating lipid profiles [43], and our findings raise the possibility that these lipid metabolic changes may be statistically consistent with partial mediation of the observed SSB-depression association, though the causal directionality of these relationships remains to be established in future experimental studies. Among inflammatory markers, neutrophil count showed the largest candidate mediation proportion, which is biologically plausible given that neutrophil activation can trigger the release of pro-inflammatory mediators that may contribute to neuroinflammatory processes relevant to depression [44]. Elevated neutrophil-related inflammatory indices have been associated with depressive symptoms in prior prospective studies [45]. Furthermore, SEM integrating these candidate mediators identified statistically significant SSB-metabolism-inflammation-depression pathways, with IDL-C and neutrophil count emerging as the primary latent factors for metabolic and inflammatory components respectively, collectively reflecting a state of low-grade systemic inflammation [37]. Taken together, these findings provide preliminary evidence that immune-metabolic dysregulation may be statistically associated with the SSB-depression association, consistent with the broader concept of immuno-metabolic depression [40], though these observations remain exploratory and require validation in future experimental studies. Several strengths of this study merit acknowledgement. First, to the best of our knowledge, this is among the first studies to construct a plasma proteomic signature associated with SSB consumption and examine its prospective relationship with incident depression, extending the analysis from self-reported dietary behaviour to the molecular level. Second, we conducted an exploratory multi-omics mediation analysis integrating proteins, metabolites, and inflammatory markers, providing preliminary insights into plausible biological pathways linking SSB consumption to the risk of incident depression. Several limitations should also be acknowledged. First, the mediation analyses presented in this study are exploratory in nature and should not be interpreted as evidence of causal mediation. The biomarker measurements were obtained prior to the dietary assessment, which does not satisfy the temporal assumptions required for classical mediation analysis. Combined with the observational study design, the likelihood of residual confounding, and the absence of external replication of the omics findings, all mechanistic interpretations should be treated as hypothesis-generating and require validation in future experimental and longitudinal studies. Second, the UKB cohort is predominantly of White European ancestry, which limits the generalizability of both the proteomic findings and the primary association to other ethnic groups. Third, SSB consumption was assessed by self-reported dietary recall, introducing the possibility of measurement error and recall bias. Fourth, incident depression was ascertained through linkage to electronic health records from multiple healthcare institutions, which may result in incomplete case ascertainment and variability in diagnostic practice across providers. Fifth, UKB participants tend to be healthier and of higher socioeconomic status than the general population, a phenomenon known as the “healthy volunteer” effect, which may lead to underestimation of the true SSB-depression association and limit the generalizability of these findings to broader populations. Finally, although major potential confounders were adjusted for, residual confounding from unmeasured or unknown factors cannot be excluded. Conclusion In conclusion, this large-scale prospective study demonstrates that higher SSB consumption is associated with an increased risk of incident depression, with consistent directional associations observed across diverse populations. Exploratory multi-omics mediation analyses identified proteins, metabolites, and inflammatory markers as candidate mediators, providing preliminary insights into plausible biological pathways linking SSB consumption to depression risk. These findings provide a basis for future experimental studies aimed at validating these candidate pathways and identifying potential targets for dietary interventions in depression prevention. Methods Study design and participants This study was conducted using data from the UKB, a large prospective cohort that recruited over 500,000 individuals aged 33–73 years between 2006 and 2010 from 22 assessment centers across the United Kingdom. At baseline, participants completed touchscreen questionnaires covering demographics, socioeconomic status, lifestyle, and medical history, and underwent physical examinations and laboratory assessments. Detailed descriptions of the UKB protocol have been published previously. Dietary intake was assessed via 24-h online dietary recalls administered by email during the 2011–2012 follow-up period. Participants with unreliable dietary data (Field ID: 100026) were excluded, leaving 209,806 individuals with valid dietary assessments. We further excluded 17,194 participants with depression at baseline and 567 participants lost to follow-up. The final analytic sample comprised 192,045 UKB participants. All participants provided electronic informed consent. No financial incentives were offered. This study was conducted using the UKB resource under application number 769,193. Assessment of sugar-sweetened beverages SSB consumption was assessed using the Oxford WebQ, a validated web-based 24-h dietary recall questionnaire that records the consumption of up to 206 foods and 32 beverages. Within WebQ, SSBs were defined as the sum of five beverage categories: fizzy drinks, squash, fruit smoothies, dairy smoothies, hot chocolate (Field IDs: 100170, 100180, 100220, 100230, 100550). Standard portion sizes were assigned to each beverage type (330 g for fizzy drinks, 250 g for squash, 260 g for fruit smoothies, 260 g for dairy smoothies, 143 g for hot chocolate) [46]. Total daily SSB consumption was calculated as the sum of servings across these categories. Participants who reported no SSB consumption were categorized as “none.” Among consumers, consumption was further divided into two groups based on the median and classified as “low” (below the median) or “high” (at or above the median). For participants who completed at least two dietary assessments, mean consumption across recalls was calculated to reduce within-person variability. The date of the last completed dietary recall was considered the baseline for follow-up. Definition of depression Incident depression was defined using the International Classification of Diseases, 10th Revision (ICD-10), based on the first recorded diagnosis of depressive episode (F32) or recurrent depressive disorder (F33) (Field IDs: 130894, 130896). The earliest date of diagnosis was obtained from multiple sources, including death registries, primary care records, hospital admissions, self-reported information, and other health data sources. Follow-up time was calculated as the interval between the date of the last dietary assessment (baseline) and the date of first depression diagnosis. Participants with follow-up time ≤ 0 or with missing follow-up information (Field ID: 191) were excluded. For participants who did not develop depression during follow-up, censoring occurred at the earliest of either the end of follow-up (8 July 2024) or the date of death. For sensitivity analyses, incident depression cases were additionally classified by ascertainment source into three subgroups: (1) hospital admissions data (ICD-10: F32-F33; Field ID: 41270); (2) self-reported diagnoses obtained from the baseline touchscreen questionnaire (Field IDs: 20002, 20126); and (3) primary care records (Read V2 and CTV3 codes corresponding to depressive episode or recurrent depressive disorder; Field ID: 42040). Plasma proteomics Plasma proteomic profiling was conducted in a subset of UKB participants as part of the UK Biobank Pharma Proteomics Project (UKB-PPP). Proteins were quantified using the Olink Explore 3072 platform, which employs proximity extension assay (PEA) technology combined with next-generation sequencing to provide relative protein abundances. After quality control, a total of 2,923 proteins were available for analysis (Table S20). Missing protein values were imputed using the k-nearest neighbors (kNN) algorithm with k = 5. Additional candidate mediators In addition to plasma proteomics, three categories of biomarkers were incorporated into the mediation analyses: circulating metabolites, inflammatory markers, and brain structural features. A total of 170 metabolites with absolute concentrations were included (Table S21), which were grouped into six major categories: very-low-density lipoprotein (VLDL; n = 50), high-density lipoprotein (HDL; n = 36), low-density lipoprotein (LDL; n = 29), intermediate-density lipoprotein (IDL; n = 7), total concentration traits (n = 8), and other metabolic traits (n = 40). 9 inflammatory markers were evaluated: systemic immune-inflammation index (SII = platelet count × neutrophil count / lymphocyte count), neutrophil-to-lymphocyte ratio (NLR = neutrophil count / lymphocyte count), platelet-to-lymphocyte ratio (PLR = platelet count / lymphocyte count), lymphocyte-to-monocyte ratio (LMR = lymphocyte count / monocyte count), C-reactive protein (CRP), absolute neutrophil count, platelet count, lymphocyte count, and monocyte count (Table S22). Brain structural features comprised 91 volumetric traits derived from MRI imaging, including 48 cortical volumes, 15 subcortical volumes, and 28 cerebellar volumes (Table S23). The temporal sequence of all mediators, SSB, and incident depression is summarized in Fig. S3. Covariates All multivariable models were adjusted for a set of sociodemographic and lifestyle covariates that are established determinants of both dietary behavior and depression. These included age (continuous), sex (male/female), and ethnic group (self-reported, categorized as White or others). Socioeconomic position was captured by the Townsend deprivation index (TDI; continuous). Educational level was classified into: (1) College or university degree (2), A levels/AS levels or equivalent (3), O levels/GCSEs or equivalent (4), CSEs or equivalent (5), NVQ/HND/HNC or equivalent (6), Other professional qualifications (e.g. nursing, teaching) (7), None of the above. Lifestyle factors comprised body mass index (BMI), categorized as underweight (< 18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), or obesity (≥ 30 kg/m²). Smoking status was defined as never, previous or current. Alcohol intake frequency was categorized as daily or almost daily, three to four times per week, once or twice per week, one to three times per month, special occasions only or never. Physical activity was assessed using the International Physical Activity Questionnaire (IPAQ) and classified as low, moderate, or high. Psychological distress was measured using the Patient Health Questionnaire-4 (PHQ-4; continuous), and total daily energy intake (kJ/day) was included to account for overall dietary consumption. Construction of the SSB proteomic signature An elastic net regression model was employed to construct the SSB proteomic signature. Specifically, we identified 20,135 participants with available SSB consumption and proteomic data, who were subsequently partitioned into a training set (N = 16,108) and a test set (N = 4,027). To prevent data leakage, protein expression values were standardized within the training set, and the derived parameters were applied to transform the test set accordingly. Five-fold cross-validation combined with grid search was performed on the full training set to identify optimal hyperparameters (L1 ratio = 0.1, alpha = 0.1). For stability selection, 80% of training samples were randomly subsampled across 50 iterations to ensure robust protein selection. Proteins appearing in at least 60% of iterations were retained, yielding 229 candidate proteins for subsequent modeling. The stability of the selected proteins was further assessed in the held-out test set, where 40.2% of the 229 training-selected proteins received non-zero elastic net coefficients when the model was independently re-fitted using the same hyperparameters, indicating consistent protein selection across independent samples. Finally, elastic net coefficients were estimated using the full sample. The SSB proteomic signature for each participant was computed as the weighted sum of protein expression values multiplied by their corresponding regression coefficients, yielding a continuous composite score reflecting SSB consumption-associated proteomic patterns. SSB-depression association and cross-cohort analyses Cox proportional hazards models were used to examine the association between SSB consumption and the risk of incident depression in the UKB cohort. All models were adjusted for the relevant covariates described above. To evaluate the robustness of the primary SSB-depression association with respect to depression ascertainment source, sensitivity analyses were conducted in the UKB by restricting the outcome to depression cases identified from each individual source separately (hospital admissions data only, self-report only, and primary care only). To further contextualize and corroborate these findings, supplementary analyses were conducted across three independent datasets: the U.S. NHANES, the YRBSS, and the Lianyungang Municipal School Health and Risk Factor Surveillance Dataset (Jiangsu Province). These cohorts span different age groups (from children and adolescents to older adults) and diverse geographic and cultural contexts (United Kingdom, United States, and China), strengthening the generalizability of our findings. The NHANES is a program conducted by the U.S. Centers for Disease Control and Prevention (CDC) that assesses the health and nutritional status of adults and children in the United States through interviews and physical examinations. It uses a complex, multistage, probability sampling design to generate nationally representative estimates. We included NHANES cycles that contained relevant dietary intake data, depressive symptom measures (via PHQ-9), and covariates consistent with the UKB analysis. The YRBSS, developed by the CDC, monitors health-related behaviors among high school students in the United States, including dietary patterns, mental health status, and substance use. The dataset is collected biennially through anonymous, self-administered questionnaires in school settings. We selected survey years that contained detailed information on SSB consumption and depressive symptoms to align with our primary analyses. The Lianyungang Municipal Database is a large-scale student health surveillance initiative jointly supported by the Lianyungang CDC. The program monitors common diseases and health-related behaviors among students across primary school, secondary school, and university levels. The most recent wave (2022–2024) collected data from nearly 20,000 participants. Standardized questionnaires assessed SSB consumption, mental health indicators (including depressive symptoms), and sociodemographic factors. This dataset provides a valuable lens into adolescent and young adult populations in East Asia, complementing the UKB and U.S.-based cohorts. Mediation analysis To investigate potential mediators of the association between SSB consumption and incident depression, we examined four categories of variables: 2,923 circulating proteins, 170 metabolites with absolute concentrations, 9 inflammatory markers, and 91 brain structural features. Protein values were used as provided by the UKB, while metabolites were log-transformed using log(x + 1) and then standardized. Inflammatory markers and brain structural features were standardized directly. Taking proteomics as an example, for each protein, we first assessed its association with SSB consumption using linear regression and then tested its association with incident depression using Cox regression, adjusting for the same covariates as in the main analysis. Variables that showed significant associations in both models after FDR correction were retained for formal mediation analysis. Mediation analyses were conducted using the ‘mediation’ package in R, with 1,000 bootstrap iterations to estimate indirect effects and mediation proportions. The same procedure was applied to metabolites and inflammatory markers. For brain structural features, because MRI scans in the UKB were performed during the second assessment wave, we excluded participants with an incident depression diagnosis prior to imaging to avoid reverse causality. Structural equation modeling Following the mediation analysis, metabolism and inflammation were identified as candidate mediators statistically consistent with mediation in the association between SSB consumption and incident depression. To further investigate these relationships, we employed Structural Equation Modeling (SEM) to assess the direct and indirect effects within the proposed statistical model, which included SSB consumption, metabolic processes, inflammatory responses, and incident depression. In the present study, the pathway of SSB-metabolism-inflammation-depression was prioritized for SEM analysis. Latent variables for each component were derived based on the results from the mediation analysis, where only those variables that showed significance after FDR correction were included. The SEM approach allowed us to examine both the direct effects of SSB consumption on metabolism and inflammation, as well as their subsequent effects on incident depression. The model was fit using the lavaan package in R, employing the BFGS optimization method. We assessed model fit using a range of indices, including standardized path coefficients and various fit measures, to evaluate the adequacy of the model. The results of the SEM analysis provided insights into the complex relationships between SSB consumption, metabolic pathways, inflammation, and their collective influence on incident depression. Enrichment and protein-protein interaction analysis Proteins that were identified as significant in the mediation analysis after FDR correction were subjected to pathway enrichment analysis and PPI network analysis to explore the biological characteristics of SSB-related proteins. Pathway enrichment analysis was performed using the GO and KEGG databases, focusing on biological processes, molecular functions, and cellular components. The analysis was carried out using the clusterProfiler package in R, and pathways with a corrected p-value < 0.05 were considered significantly enriched. For the PPI analysis, we constructed a protein interaction network using the STRING database (version 12.0) and visualized it using Cytoscape (version 3.10.3). The interactions were filtered to include only medium-confidence interactions (score > 0.4), and key hub proteins within the network were identified using centrality measures such as degree, betweenness, and closeness. The resulting PPI network helped identify potential protein clusters and key proteins that may mediate the relationship between SSB consumption and incident depression. Additionally, for proteins identified through elastic net, we performed the same pathway enrichment and PPI network analyses. The results of these analyses are presented in the supplementary materials. Supplementary Information Supplementary Material 1: Supplement Tables and Figures Contents: Table S1. Baseline Characteristics of the UK Biobank Participants. Table S2. Baseline Characteristics of the NHANES Participants. Table S3. Baseline Characteristics of the YRBSS Participants. Table S4. Baseline Characteristics of the Lianyungang Municipal Database* Participants. Table S5. Associations between SSB and SSB signature with incident depression in the UK Biobank participants. Table S6. Associations between SSB with depression in the NHANES and the Lianyungang Municipal Database* participants. Table S7. Associations between SSB with depression in the YRBSS participants. Table S8. Variable selection using elastic net. Table S9. Top 30 significantly enriched GO terms (Biological Process, Cellular component, Molecular function) for proteins selected by elastic net, ranked by FDR. Table S10. KEGG pathway enrichment analysis of proteins selected by elastic net. Table S11. Plasma protein mediating the association between SSB and incident depression. Table S12. KEGG pathways enriched for proteins mediating the effect of SSB on depression. Table S13. GO enrichment of candidate mediator proteins mediating the effect of SSB on depression, showing terms from Biological Process, Cellular Component, and Molecular Function domains. Table S14. Plasma metabolites mediating the association between SSB and incident depression. Table S15. Inflammatory markers mediating the association between SSB and incident depression. Table S16. Associations of sugar-sweetened beverage intake with plasma protein and of plasma protein with Depression Risk. Table S17. Associations of sugar-sweetened beverage intake with brain area and of brain area with depression risk. Table S18. Associations of sugar-sweetened beverage intake with plasma metabolites and plasma metabolites with depression risk. Table S19. Associations of sugar-sweetened beverage intake with inflammatory markers and of plasma metabolites with depression risk. Table S20, Proteins and their abbreviations analyzed in this study. Table S21, Metabolites and their abbreviations analyzed in this study. Table S22, Inflammation markers and their details analyzed in this study. Table S23, Brain structures and their abbreviations analyzed in this study. Figure S1. Sensitivity analysis of the association between SSB and depression across different case ascertainment sources. Figure S2. Functional characterization of elastic net selected protein biomarkers through integrated enrichment and network analysis. Figure S3. Longitudinal study timeline. Supplementary Material 1: Supplement Tables and Figures Contents: Table S1. Baseline Characteristics of the UK Biobank Participants. Table S2. Baseline Characteristics of the NHANES Participants. Table S3. Baseline Characteristics of the YRBSS Participants. Table S4. Baseline Characteristics of the Lianyungang Municipal Database* Participants. Table S5. Associations between SSB and SSB signature with incident depression in the UK Biobank participants. Table S6. Associations between SSB with depression in the NHANES and the Lianyungang Municipal Database* participants. Table S7. Associations between SSB with depression in the YRBSS participants. Table S8. Variable selection using elastic net. Table S9. Top 30 significantly enriched GO terms (Biological Process, Cellular component, Molecular function) for proteins selected by elastic net, ranked by FDR. Table S10. KEGG pathway enrichment analysis of proteins selected by elastic net. Table S11. Plasma protein mediating the association between SSB and incident depression. Table S12. KEGG pathways enriched for proteins mediating the effect of SSB on depression. Table S13. GO enrichment of candidate mediator proteins mediating the effect of SSB on depression, showing terms from Biological Process, Cellular Component, and Molecular Function domains. Table S14. Plasma metabolites mediating the association between SSB and incident depression. Table S15. Inflammatory markers mediating the association between SSB and incident depression. Table S16. Associations of sugar-sweetened beverage intake with plasma protein and of plasma protein with Depression Risk. Table S17. Associations of sugar-sweetened beverage intake with brain area and of brain area with depression risk. Table S18. Associations of sugar-sweetened beverage intake with plasma metabolites and plasma metabolites with depression risk. Table S19. Associations of sugar-sweetened beverage intake with inflammatory markers and of plasma metabolites with depression risk. Table S20, Proteins and their abbreviations analyzed in this study. Table S21, Metabolites and their abbreviations analyzed in this study. Table S22, Inflammation markers and their details analyzed in this study. Table S23, Brain structures and their abbreviations analyzed in this study. Figure S1. Sensitivity analysis of the association between SSB and depression across different case ascertainment sources. Figure S2. Functional characterization of elastic net selected protein biomarkers through integrated enrichment and network analysis. Figure S3. Longitudinal study timeline.