Associations of adiposity, atherogenic lipid phenotypes, systemic immune-inflammatory indices, and vascular aging markers with depression in US adults: NHANES 2005–2020

Abstract

The biological mechanisms linking cardiometabolic health, immune-inflammatory activation, and vascular aging with depression remain incompletely understood. We aimed to examine the associations of body mass index (BMI), atherogenic lipid phenotypes, systemic immune-inflammatory indices, and vascular aging markers with the risk of moderate-to-severe depression among US adults. We analyzed data from 17,011 participants in the National Health and Nutrition Examination Survey (NHANES) 2005–2020. Depression was defined as a Patient Health Questionnaire-9 score ≥ 10. Independent predictors included BMI, atherogenic lipid profiles (atherogenic index of plasma, high-density lipoprotein cholesterol [HDL-C], triglyceride/HDL-C, non-HDL-C, Castelli indices), immune-inflammatory indices (systemic immune-inflammation index, neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, neutrophil-to-platelet ratio), and vascular aging markers (estimated pulse wave velocity, urinary albumin-to-creatinine ratio). Weighted logistic regression and receiver operating characteristic analyses assessed associations and predictive performance. Among 17,011 participants, 8477 were women and 8534 were men. Moderate-to-severe depression prevalence was 10.9% in women and 6.2% in men. In fully adjusted models, higher BMI, atherogenic index of plasma, triglyceride/HDL-C, Castelli indices, systemic immune-inflammation index, neutrophil-to-lymphocyte ratio, neutrophil-to-platelet ratio, estimated pulse wave velocity, and urinary albumin-to-creatinine ratio were significantly associated with greater odds of depression. HDL-C showed an inverse association in women. Associations were generally stronger in women than in men, particularly for atherogenic lipid indices and inflammatory markers. Predictive models incorporating these indices improved classification of moderate-to-severe depression beyond sociodemographic and clinical factors, with area under the curve increases of 0.024 to 0.045 (all P < .05). Adiposity, dyslipidemia, systemic immune-inflammatory activation, and vascular aging were independently associated with depression. These findings support the role of cardiometabolic and inflammatory pathways in depression pathogenesis and highlight potential avenues for prevention and intervention.

1. Introduction

Depression is a leading contributor to the global burden of disease and disability, affecting more than 300 million individuals worldwide and substantially increasing morbidity and mortality.^[1,2] Despite advances in understanding its clinical heterogeneity, the biological mechanisms linking depression to systemic health remain only partially elucidated.^[3,4] Epidemiologic studies have suggested strong bidirectional associations between depression and metabolic and vascular conditions, including obesity, dyslipidemia, systemic inflammation, and cardiovascular disease (CVD).^[4-6] These observations raise the possibility that shared biological pathways may contribute to both somatic and mental health outcomes.

Previous investigations have reported consistent associations between excess adiposity and increased risk of depression.^[4,7] Dysregulated lipid metabolism, particularly elevated triglyceride-to-HDLcholesterol ratios, has also been linked with depressive symptoms.^[8,9] At the same time, evidence from clinical and population studies supports a central role of chronic low-grade inflammation and immune dysregulation in the development and persistence of depression.^[10-13] Moreover, indices of vascular aging, such as increased arterial stiffness and elevated urinary albumin excretion, have been associated with cognitive decline and mood disorders, suggesting an important contribution of vascular health to brain function.^[14,15] However, most prior studies have considered these risk domains in isolation, limiting the ability to delineate their interrelationships and relative contributions to depression.

The current study was designed to extend this literature by jointly evaluating body mass index (BMI), atherogenic lipid profiles, systemic immune-inflammatory indices, and vascular aging markers in relation to moderate-to-severe depression within a large nationally representative US cohort. Unlike prior analyses that typically focused on a single biological domain,^[8,10,14] our work integrates multiple mechanistic pathways within the same analytic framework. By comparing predictive performance across these domains and examining sex-specific patterns, the study provides novel insights into the relative roles of metabolic, inflammatory, and vascular processes in depression. This comprehensive approach offers an opportunity to refine mechanistic hypotheses regarding the obesity-inflammation-vascular aging pathway to depression and to identify potential targets for early prevention and intervention.

2. Materials and methods

2.1. Study participants and recruitment

We used cross-sectional data from the National Health and Nutrition Examination Survey (NHANES), a continuous, nationally representative survey of the noninstitutionalized US population.^[16,17] Between 2005 and 2020, a total of 85,750 participants were examined. Participants were eligible if they were aged ≥20 years, were not pregnant at the time of examination, completed the 9-item Patient Health Questionnaire (PHQ-9) assessment, and had available data on BMI, lipid measures, immune cell counts, and vascular aging markers required to construct the predefined indices. We excluded 36,769 individuals younger than 20 years of age and 795 pregnant women. Additional exclusions included 2966 participants with missing data on BMI, 25,085 with incomplete data on lipid phenotype indices (atherogenic index of plasma [AIP], high-density lipoprotein cholesterol (HDL-C), triglyceride-to-HDL-C ratio, non-HDL-C, Castelli index I, and Castelli index II), 67 with missing values for immune-inflammation indices (systemic immune-inflammation index (SII), neutrophil-to-lymphocyteratio [NLR], platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), and neutrophil-to-platelet ratio [NPR]), 1839 with missing vascular-aging indices (estimated pulse wave velocity [ePWV] and urinary albumin-to-creatinine ratio [uACR]), and 1218 without data on the PHQ-9 for depressive symptoms. The final analytic sample consisted of 17,011 adults (8477 women and 8534 men). Among women, 924 (10.9%) were classified as having moderate-to-severe depression, whereas 527 (6.2%) of men were classified as such (Fig. 1). NHANES protocols were approved by the National Center for Health Statistics Research Ethics Review Board; written informed consent was obtained from all participants. All procedures were conducted in accordance with the ethical standards of the Declaration of Helsinki.

Figure 1.

Flow chart of study participant’s selection.

2.2. Measures

2.2.1. Body mass index, atherogenic lipid profiles, immune-inflammatory indices, and vascular aging indices

Anthropometric measurements and blood pressure were obtained during standardized examinations at the Mobile Examination Center. Laboratory biomarkers (lipids, creatinine, urinary albumin and creatinine, and complete blood counts) were measured using standardized NHANES protocols.

BMI was calculated as weight in kilograms divided by the square of height in meters. Atherogenic lipid profiles included the (AIP, calculated as log[triglycerides/HDL cholesterol]), HDL cholesterol, triglyceride-to-HDL cholesterol ratio, non-HDL cholesterol,^[18] Castelli index I (total cholesterol/HDL cholesterol), and Castelli index II (LDL cholesterol/HDL cholesterol). Immune-inflammatory indices included the SII (platelet count * neutrophil count/lymphocyte count), NLR, PLR, LMR, and NPR. Vascular-aging indices included ePWV, derived from age and mean blood pressure (MBP), and uACR.

2.2.2. Depression assessment

Depressive symptoms during the prior 14 days were assessed with the PHQ-9, administered by trained interviewers during in-person visits at the Mobile Examination Center.^[19] Each item is scored from 0 to 3 (0, not at all; 3, nearly every day) and addresses loss of interest, depressed mood, sleep disturbance, low energy, appetite change, feelings of worthlessness, impaired concentration, psychomotor change, and thoughts of self-harm. Scores are summed to yield a total from 0 to 27; a threshold of ≥10 denotes clinically significant depressive symptoms.^[19] Severity categories are 1 to 4 (minimal), 5 to 9 (mild), 10 to 14 (moderate), 15 to 19 (moderately severe), and 20 to 27 (severe).^[19]

Consistent with prior epidemiologic studies, a PHQ-9 score of ≥10 was used to define clinically significant depression in the primary analyses. Severity-specific analyses were conducted as secondary and sensitivity analyses. Specifically, multinomial models were used for the 5-category PHQ-9 severity outcome, and receiver operating characteristic analyses were repeated using alternative thresholds (≥5, ≥10, ≥15, and ≥20) as sensitivity analyses.

2.2.3. Vascular-aging index

Vascular aging was indexed by ePWV, derived from age and MBP using the reference values collaboration equation: ePWV = 9.587 − 0.402 * Age + 0.004560 * Age² − 0.00002621 * Age² * MBP + 0.003176 * Age * MBP − 0.01832 * MBP, with MBP = DBP + 0.4 * (SBP − DBP; mm Hg). This surrogate closely tracks measured carotid-femoral PWV and predicts cardiovascular outcomes.^[20]

2.2.4. Sociodemographic and clinical covariates

Sociodemographic variables included age, sex, self-identified race or ethnicity, educational attainment, and poverty-to-income ratio. Clinical covariates included estimated glomerular filtration rate, diabetes mellitus status, CVD, cancer history, smoking status, alcohol consumption, and moderate-to-vigorous physical activity. Missing values for selected covariates (poverty-to-income ratio, education, diabetes, CVD, cancer, and alcohol consumption) were imputed using random forest imputation with the R package missRanger (version 2.4.0).

2.3. Data analysis

All analyses incorporated NHANES survey weights, strata, and primary sampling units to account for the complex sampling design and generate nationally representative estimates.^[17] Descriptive statistics were calculated for continuous variables (means and standard deviations) and categorical variables (%). Group differences between individuals with no or mild depression and those with moderate-to-severe depression were assessed using survey-weighted t-tests for continuous variables and Rao–Scott chi-square tests for categorical variables.

Associations of BMI, atherogenic lipid profiles, immune-inflammatory indices, and vascular aging markers with moderate-to-severe depression were estimated using survey-weighted logistic regression models. Both unadjusted and multivariable models were examined, with the latter adjusting for all sociodemographic and clinical covariates. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported. Analyses were stratified by sex to evaluate potential differences in associations between men and women.

To assess the predictive performance of individual biomarker domains, we calculated the area under the curve (AUC) for models including covariates alone and for models additionally including each biomarker domain. Differences in AUC were tested using DeLong’s method. A 2-sided P value of <.05 was considered statistically significant. All analyses were conducted using R version 4.3.0 and the survey package.

In addition to binary logistic regression, multinomial logistic regression models were fitted to examine associations across PHQ-9 depression severity categories (0–4, 5–9, 10–14, 15–19, and 20–27), using the no/minimal depression group as the reference. Model discrimination was further evaluated across alternative PHQ-9 thresholds (≥5, ≥10, ≥15, and ≥20) using AUC and DeLong’s tests.

3. Results

3.1. Population characteristics

A total of 17,011 adults were included after exclusions (Fig. 1), with 8477 women and 8534 men. The mean age was 46.3 years; 48.7% were women. Moderate-to-severe depression (PHQ-9 ≥ 10) was identified in 924 women (10.9%) and 527 men (6.2%). Table 1 summarizes weighted baseline characteristics by depression status. Compared with those without or with mild symptoms, participants with moderate-to-severe depression were younger, more often female, had lower socioeconomic status, and a higher prevalence of diabetes, CVD, and smoking.

Table 1.

Baseline characteristics of study participants in the NHANES database.

	Overall (n = 17,011)		Women (n = 8477)		Men (n = 8534)
	No/mild depression (n = 15,560)	Moderate to severe depression (n = 1451)	No/mild depression (n = 7553)	Moderate to severe depression (n = 924)	No/mild depression (n = 8007)	Moderate to severe depression (n = 527)
Age (yr)	50.3 (17.7)	50.1 (16.0)	50.2 (17.7)	50.1 (15.8)	50.3 (17.8)	50.3 (16.4)
AIP†	−0.08 (0.3)	−0.02 (0.3)***	−0.15 (0.3)	−0.06 (0.3)***	−0.02 (0.3)	0.04 (0.3)**
PIR¶¶¶	2.6 (1.6)	1.7 (1.4)***	2.6 (1.6)	1.7 (1.4)***	2.7 (1.6)	1.7 (1.3)***
BMI, kg/m2	29.1 (6.8)	31.1 (8.0)***	29.5 (7.6)	31.9 (8.5)***	28.7 (6.0)	29.5 (6.8)**
eGFR‡, mL/min/1.73 m2	94.6 (22.0)	94.2 (21.8)	95.4 (22.5)	95.1 (21.3)	93.8 (21.4)	92.6 (22.7)
HDL-C, mg/dL	54.2 (15.9)	52.7 (16.3)***	59.3 (16.4)	54.9 (15.6)***	49.4 (13.8)	48.8 (17.0)
TG/HDL-C	2.4 (1.9)	2.9 (2.4)***	2.1 (1.6)	2.6 (1.9)***	2.8 (2.2)	3.4 (3.0)***
Non-HDL-C, mg/dL	135.3 (39.8)	139.0 (43.1)**	133.8 (39.8)	138.8 (43.3)***	136.8 (39.8)	139.2 (42.8)
Castelli I§	3.7 (1.2)	3.9 (1.4)***	3.5 (1.1)	3.8 (1.2)***	4.0 (1.3)	4.2 (1.7)*
Castelli II∥	2.2 (1.0)	2.4 (1.1)***	2.0 (0.9)	2.3 (1.0)***	2.4 (1.0)	2.5 (1.3)
SII¶	514.5 (387.9)	565.3 (362.1)***	533.4 (322.7)	574.7 (356.7)***	496.7 (439.9)	548.9 (366.0)**
NLR#	2.1 (1.2)	2.2 (1.2)**	2.1 (1.1)	2.1 (1.2)	2.2 (1.2)	2.3 (1.2)**
PLR††	130.8 (51.7)	130.2 (56.2)	136.2 (51.9)	134.4 (58.3)	125.7 (51.0)	122.9 (51.5)
LMR‡‡	4.1 (1.8)	4.3 (1.9)**	4.4 (1.9)	4.5 (1.9)	3.8 (1.7)	3.8 (1.8)
NPR§§	0.02 (0.01)	0.02 (0.01)***	0.02 (0.01)	0.02 (0.01)**	0.02 (0.01)	0.02 (0.01)***
ePWV¶¶, m/s	8.3 (2.1)	8.5 (2.3)**	8.1 (2.1)	8.4 (2.3)**	8.4 (2.0)	8.6 (2.2)*
uACR‖‖	0.4 (3.1)	0.7 (3.7)**	0.4 (3.1)	0.7 (3.7)*	0.4 (3.1)	0.8 (3.8)**
Race, n (%)	–		–	**	–
Mexican American	2337 (15.0)	209 (14.4)	1118 (14.8)	141 (15.3)	1219 (15.2)	68 (12.9)
Non-Hispanic White	6600 (42.4)	592 (40.8)	3169 (42.0)	342 (37.0)	3431 (42.9)	250 (47.4)
Non-Hispanic Black	3231 (20.8)	339 (23.4)	1584 (21.0)	232 (25.1)	1647 (20.6)	107 (20.3)
Other Race	3392 (21.8)	311 (21.4)	1682 (22.2)	209 (22.6)	1710 (21.3)	102 (19.4)
Education Level¶¶¶, n (%)	–	***	–	***	–	***
College or above	8540 (54.9)	586 (40.4)	4343 (57.5)	400 (43.3)	4197 (52.4)	186 (35.3)
≤High school	7020 (45.1)	865 (59.6)	3210 (42.5)	524 (56.7)	3810 (47.6)	341 (64.7)
Diabetes Mellitus##, ¶¶¶, n (%)	–	***	–	***	–	***
No	13,442 (86.4)	1138 (78.4)	6590 (87.3)	722 (78.1)	6852 (85.6)	416 (78.9)
Yes	2118 (13.6)	313 (21.6)	963 (12.7)	202 (21.9)	1155 (14.4)	111 (21.1)
CVD†††,¶¶¶, n (%)	–	***	–	***	–	***
No	13,952 (89.7)	1157 (79.7)	6928 (91.7)	741 (80.2)	7024 (87.7)	416 (78.9)
Yes	1608 (10.3)	294 (20.3)	625 (8.3)	183 (19.8)	983 (12.3)	111 (21.1)
Cancer¶¶¶, n (%)	–		–		–
No	14,073 (90.4)	1299 (89.5)	6785 (89.8)	816 (88.3)	7288 (91.0)	483 (91.7)
Yes	1487 (9.6)	152 (10.5)	786 (10.2)	108 (11.7)	719 (9.0)	44 (8.3)
Current smoking, n (%)	–	*	–		–
No	11,646 (74.8)	1122 (77.3)	6105 (80.8)	754 (81.6)	5541 (69.2)	368 (69.8)
Yes	3914 (25.2)	329 (22.7)	1448 (19.2)	170 (18.4)	2466 (30.8)	159 (30.2)
Alcohol consumption‡‡‡,¶¶¶, n (%)	–		–		–
No	3363 (21.6)	335 (23.1)	2318 (30.7)	275 (29.8)	1045 (13.1)	60 (11.4)
Yes	12,197 (78.4)	1116 (76.9)	5235 (69.3)	649 (70.2)	6962 (86.9)	467 (88.6)
MVPA§§§, n (%)	–		–		–
No	7091 (45.6)	646 (44.5)	2955 (39.1)	391 (42.3)	4136 (51.7)	255 (48.4)
Yes	8469 (54.4)	805 (55.5)	4598 (60.9)	533 (57.7)	3871 (48.3)	272 (51.6)

Values are means (SDs) for continuous variables and percentages for categorical variables. Continuous variables were compared using survey-weighted t-tests, and categorical variables using Rao–Scott χ² tests.

AIP = atherogenic index of plasma, BMI = body mass index, CVD = cardiovascular disease, DBP = diastolic blood pressure, eGFR = estimated glomerular filtration rate, ePWV = estimated pulse wave velocity, HDL-C = high-density lipoprotein cholesterol, LDL-C = low-density lipoprotein cholesterol, LMR = lymphocyte-to-monocyte ratio, MBP = mean blood pressure, MVPA = moderate-to-vigorous physical activity, NHANES = National Health and Nutrition Examination Survey, NLR = neutrophil-to-lymphocyte ratio, non-HDL-C = non-high-density lipoprotein cholesterol, NPR = neutrophil-to-platelet ratio, PIR = poverty-to-income ratio, PLR = platelet-to-lymphocyte ratio, SBP = systolic blood pressure, SD = standard deviation, SII = systemic immune-inflammation index, TC = total cholesterol, TG = triglycerides, TG/HDL-C = triglyceride-to-high-density lipoprotein cholesterol, uACR = urinary albumin-to-creatinine ratio.

^†AIP = log₁₀(TG [mmol/L]/HDL [mmol/L]); convert to mmol/L first (TG * 0.01129; HDL-C * 0.02586), then substitute into the formula.

^‡eGFR (mL/min/1.73 m²) was calculated using the CKD-EPI 2021 creatinine equation: 142 * min (Scr/κ, 1)^α × max (Scr/κ, 1) ^−1.200 * 0.9938^age * 1.012 (if female), where Scr is IDMS-traceable serum creatinine (mg/dL); κ = 0.7 (female) or 0.9 (male); α = −0.241 (female) or −0.302 (male); age in years.

^§Castelli index I (CRI-I), calculated as total cholesterol divided by HDL cholesterol (TC/HDL-C); unitless. Higher values indicate a more atherogenic lipid profile.

^∥Castelli index II (CRI-II), calculated as LDL cholesterol divided by HDL cholesterol (LDL-C/HDL-C); unitless. Higher values indicate a more atherogenic lipid profile.

^¶Systemic immune-inflammation index (SII), calculated as platelet count * neutrophil count/lymphocyte count (all in 10⁹/L); unitless. Higher values indicate greater systemic inflammation.

^#Neutrophil-to-lymphocyte ratio (NLR), calculated as neutrophil count/ lymphocyte count (all in 10⁹/L); unitless.

^††Platelet-to-lymphocyte ratio (PLR), calculated as platelet count/lymphocyte count (all in 10⁹/L); unitless.

^‡‡Lymphocyte-to-monocyte ratio (LMR), calculated as lymphocyte count (all in 10⁹/L)/ monocyte count (all in 10⁹/L); unitless.

^§§Neutrophil-to-platelet ratio (NPR), calculated as neutrophil count/platelet count (all in 10⁹/L); unitless.

^¶¶ePWV was derived from age and mean blood pressure (MBP) using the reference values equation: ePWV = 9.587 − 0.402 * Age + 0.004560 * Age² − 0.00002621 * Age² * MBP + 0.003176 * Age * MBP − 0.01832 * MBP; MBP = DBP + 0.4 * (SBP − DBP; mm Hg).

^‖‖Urinary albumin-to-creatinine ratio (uACR), measured in a spot urine sample as milligrams of albumin per gram of creatinine (mg/g); higher values indicate greater albuminuria.

^##In this study, diabetes was defined as self-reported diabetes, the use of glucose-lowering medication, or an HbA_1c level of ≥6.5%.

^†††Cardiovascular disease (CVD): composite indicator based on NHANES self-reported doctor diagnoses – MCQ160B (ever told had congestive heart failure), MCQ160C (ever told had coronary heart disease), MCQ160D (ever told had angina/angina pectoris), MCQ160E (ever told had a heart attack), and MCQ160F (ever told had a stroke). CVD was considered present if any item was answered “Yes”.

^‡‡‡Participants were asked: “In any 1 year, have you had at least 12 drinks of any type of alcoholic beverage?” A drink was defined as 12 oz beer, 5 oz wine, or 1.5 oz liquor. Responses of “Yes” were coded as ≥12 drinks/yr; “No” otherwise; refusals/unknowns set to missing.

^§§§MVPA (moderate-to-vigorous physical activity, work domain). Defined from NHANES items PAQ605 (vigorous work activity) and PAQ620 (moderate work activity): self-reported engagement in ≥10 minutes continuously of vigorous activity (large increases in breathing/heart rate; for example, heavy lifting, digging, construction) and/or moderate activity (small increases; e.g., brisk walking, carrying light loads).

^¶¶¶Missing values for PIR (9.05%), education level (0.05%), diabetes mellitus (0.08%), CVD (0.47%), cancer (0.05%), and alcohol consumption (0.08%) in the NHANES study, were imputed using random forest imputation with the R package missRanger (version 2.4.0).

Significance level: *P < .05.

Significance level: **P < .01.

Significance level: ***P < .001.

BMI and atherogenic lipid indices, particularly AIP, TG-to-HDL-C ratio, and Castelli indices, were higher in depressed individuals, whereas HDL-C was lower. Immune-inflammatory indices such as systemic immune-inflammation index and NLR were also elevated, consistent with a proinflammatory state. Vascular aging markers, including ePWV and uACR, were increased, indicating arterial stiffness and microvascular injury. These associations were present in both sexes but were more pronounced in women (Table 1).

3.2. Correlation analysis of body mass index, atherogenic lipid profiles, immune-inflammatory indices, vascular aging markers

In survey-weighted logistic regression models adjusted for sociodemographic and clinical covariates, higher body-mass index was significantly associated with greater odds of moderate-to-severe depression. Similarly, adverse atherogenic lipid profiles, including higher AIP, elevated triglyceride-to-HDL cholesterol ratio, and higher Castelli indices, were consistently linked with an increased risk of depression. Markers of systemic inflammation showed robust associations: higher SII, NLR, and NPR were each associated with greater odds of depression, whereas a higher lymphocyte-to-monocyte ratio was inversely associated. Measures of vascular aging were also related to depression. Both increased ePWV and higher uACR were independently associated with moderate-to-severe depressive symptoms. Sex-stratified analyses revealed broadly consistent associations across men and women, although effect sizes were generally larger among women. Full results are presented in Figure 2 and Tables S1 to S3, Supplemental Digital Content, https://links.lww.com/MD/R365.

Figure 2.

Heatmaps of pairwise Spearman correlations among atherogenic lipid indices (AIP, TG/HDL-C ratio, HDL-C), vascular aging (ePWV), depressive symptom status, and clinical covariates in the total study population (A), women (B), and men (C). Only the lower triangular matrix is displayed. Tiles represent correlation coefficients (ρ), with blue indicating inverse and red indicating positive correlations. P values were derived from survey-weighted Spearman analyses, with statistical significance denoted as ***P < .001, **P < .01, and *P < .05. Continuous predictors were analyzed as numeric variables, and categorical covariates were included according to their coded levels. AIP = atherogenic index of plasma, BMI = body mass index, eGFR = estimated glomerular filtration rate, ePWV = estimated pulse wave velocity, HDL-C = high-density lipoprotein cholesterol, LDL-C = low-density lipoprotein cholesterol, LMR = lymphocyte-to-monocyte ratio, MBP = mean blood pressure, MVPA = moderate-to-vigorous physical activity, NLR = neutrophil-to-lymphocyte ratio, non-HDL-C, non-high-density lipoprotein cholesterol, NPR = neutrophil-to-platelet ratio, PIR = poverty-to-income ratio, PLR = platelet-to-lymphocyte ratio, SII = systemic immune-inflammation index, TG = triglycerides; uACR = urinary albumin-to-creatinine ratio.

3.3. Associations among body mass index, atherogenic lipids, immune-inflammatory indices, vascular aging indices, and moderate-to-severe depression

Table 2 shows the associations between each biomarker and moderate-to-severe depression in overall, sex-stratified, and multivariable-adjusted models. In the overall population, higher BMI was associated with an increased odds of depression (adjusted OR, 1.02; 95% CI: 1.01–1.03). AIP was strongly associated with depression (adjusted OR, 1.95; 95% CI: 1.51–2.52), with stronger associations observed in women (adjusted OR, 2.15; 95% CI: 1.56–2.95) than in men (adjusted OR, 1.73; 95% CI: 1.08–2.77). TG-to-HDL-C ratio and Castelli indices also demonstrated consistent associations.

Table 2.

Associations of body mass index, atherogenic lipid profiles, immune-inflammatory indices, vascular aging, and moderate-to-severe depression.

	Overall (n = 17,011)		Women (n = 8477)		Men (n = 8534)
	OR (95% CI) unadjusted	OR (95% CI) adjusted	OR (95% CI) unadjusted	OR (95% CI) adjusted	OR (95% CI) unadjusted	OR (95% CI) adjusted
Body mass index and moderate-to-severe depression (OR)†
BMI, kg/m2	1.03 (1.02–1.04)***	1.02 (1.01–1.03)***	1.04 (1.03–1.05)***	1.03 (1.02–1.04)***	1.01 (0.99–1.03)	1.01 (0.99–1.02)
Atherogenic lipid profiles and moderate-to-severe depression (OR)†
AIP	1.90 (1.51–2.38)***	1.95 (1.51–2.52)***	2.76 (2.09–3.65)***	2.15 (1.56–2.95)***	2.01 (1.28–3.14)***	1.73 (1.08–2.77)***
HDL-C, mg/dL	0.99 (0.98–0.99)**	0.99 (0.98–0.99)*	0.98 (0.97–0.98)***	0.99 (0.98–0.99)**	0.99 (0.98–1.01)	1.00 (0.99–1.01)
TG/HDL-C	1.10 (1.07–1.13)***	1.10 (1.07–1.14)***	1.16 (1.11–1.21)***	1.11 (1.06–1.16)***	1.12 (1.07–1.17)***	1.10 (1.05–1.15)***
Non-HDL-C, mg/dL	1.002 (1.001–1.003)*	1.003 (1.001–1.005)**	1.003 (1.001–1.006)*	1.004 (1.002–1.007)**	1.001 (0.998–1.004)	1.002 (0.999–1.005)
Castelli I	1.12 (1.07–1.17)***	1.13 (1.08–1.18)***	1.28 (1.18–1.38)***	1.20 (1.11–1.30)***	1.10 (1.03–1.18)**	1.08 (1.03–1.13)*
Castelli II	1.10 (1.04–1.17)**	1.12 (1.06–1.18)***	1.31 (1.18–1.46)***	1.23 (1.12–1.36)***	1.05 (0.95–1.16)	1.03 (0.94–1.13)
Immune-inflammatory indices and moderate-to-severe depression (OR)†
SII	1.004 (1.003–1.005)**	1.003 (1.002–1.004)*	1.003 (1.002–1.004)*	1.002 (1.001–1.003)*	1.001 (0.999–1.002)	1.000 (0.999–1.001)
NLR	1.04 (1.03–1.05)**	1.03 (1.02–1.04)*	1.02 (1.01–1.03)*	1.005 (1.002–1.008)*	1.10 (1.02–1.19)*	1.07 (0.98–1.17)
PLR	0.99 (0.99–1.00)	0.99 (0.99–1.00)	0.99 (0.98–1.00)	0.99 (0.98–1.00)	0.99 (0.99–1.00)	0.99 (0.99–1.00)
LMR	1.06 (1.03–1.10)***	1.04 (1.01–1.07)*	1.07 (1.02–1.11)**	1.05 (1.01–1.09)**	1.01 (0.95–1.08)	1.01 (0.95–1.07)
NPR	1.15 (1.10–1.20)***	1.12 (1.08–1.16)**	1.07 (1.03–1.11)**	1.05 (1.01–1.10)*	1.03 (0.98–1.09)	1.01 (0.96–1.08)
Vascular aging and moderate-to-severe depression (OR)†
ePWV, m/s	1.17 (1.14–1.20)***	1.09 (1.03–1.15)***	1.09 (1.04–1.14)***	1.05 (1.01–1.09)***	1.07 (1.03–1.11)**	1.03 (1.01–1.05)*
uACR	1.04 (1.03–1.05)**	1.02 (1.01–1.03)*	1.03 (1.02–1.04)**	1.02 (1.01–1.03)*	1.02 (1.00–1.04)*	1.00 (0.99–1.02)

AIP = atherogenic index of plasma, BMI = body mass index, ePWV = estimated pulse wave velocity, HDL-C = high-density lipoprotein cholesterol, LDL-C = low-density lipoprotein cholesterol, LMR = lymphocyte-to-monocyte ratio, MVPA = moderate-to-vigorous physical activity, NLR = neutrophil-to-lymphocyte ratio, non-HDL-C = non-high-density lipoprotein cholesterol, NPR = neutrophil-to-platelet ratio, PLR = platelet-to-lymphocyte ratio, SII = systemic immune-inflammation index, TG/HDL-C = triglyceride-to-high-density lipoprotein cholesterol, uACR = urinary albumin-to-creatinine ratio.

^†Survey-weighted multivariable logistic regression models adjusted for age, sex, race, education level, PIR, eGFR, diabetes mellitus, cardiovascular disease, cancer, current smoking, alcohol consumption, and MVPA.

Significance level: *P < .05.

Significance level: **P < .01.

Significance level: ***P < .001.

Among inflammatory markers, higher NLR, systemic immune-inflammation index, and NPR were each associated with greater odds of depression, particularly among women. PLR did not show consistent associations, and LMR exhibited a modest positive association. Regarding vascular aging, higher ePWV and uACR were both associated with moderate-to-severe depression, with adjusted ORs of 1.09 (95% CI: 1.03–1.15) and 1.02 (95% CI: 1.01–1.03), respectively, in the overall population. These findings indicate that metabolic, inflammatory, and vascular processes are independently associated with depression even after adjusting for confounders.

3.4. Improved classification of anthropometric, lipid, inflammatory, and vascular predictors of depression

The incremental predictive value of these biomarkers is summarized in Table 3. In the overall population, the base model (model 1), which included sociodemographic and clinical covariates, yielded an AUC of 0.703 (95% CI: 0.689–0.717). Addition of BMI (model 2) improved the AUC modestly to 0.738, with a statistically significant ΔAUC of 0.035 (P = .001). Incorporating atherogenic lipid indices (model 3) increased the AUC to 0.739 (ΔAUC = 0.036, P = .003). Adding immune-inflammatory indices (model 4) and vascular aging markers (model 5) further increased the AUCs to 0.727 and 0.737, both with significant improvements.

Table 3.

Diagnostic performance of body mass index, atherogenic lipid profiles, immune-Inflammatory indices, and vascular aging in classifying moderate-to-severe depression.

Predictors	Overall (n = 17,011)	Women (n = 8477)	Men (n = 8534)
	AUC	AUC	AUC
Body mass index
Model 1†	0.703 (0.689–0.717)	0.683 (0.665–0.701)	0.703 (0.681–0.725)
Model 2‡	0.738 (0.724–0.752)	0.721 (0.703–0.739)	0.714 (0.692–0.736)
Delong’s test	ΔAUC = 0.035, P = .001	ΔAUC = 0.038, P = .010	ΔAUC = 0.011, P = .168
Atherogenic lipid profiles
Model 1†	0.703 (0.689–0.717)	0.683 (0.665–0.701)	0.703 (0.681–0.725)
Model 3§	0.739 (0.725–0.753)	0.721 (0.703–0.739)	0.720 (0.697–0.742)
Delong’s test	ΔAUC = 0.036, P = .003	ΔAUC = 0.038, P = .010	ΔAUC = 0.017, P = .158
Immune-inflammatory indices
Model 1†	0.703 (0.689–0.717)	0.683 (0.665–0.701)	0.703 (0.681–0.725)
Model 4∥	0.727 (0.713–0.741)	0.692 (0.674–0.710)	0.730 (0.708–0.752)
Delong’s test	ΔAUC = 0.024, P = .005	ΔAUC = 0.009, P = .079	ΔAUC = 0.027, P = .018
Vascular aging
Model 1†	0.703 (0.689–0.717)	0.683 (0.665–0.701)	0.703 (0.681–0.725)
Model 5¶	0.737 (0.723–0.751)	0.697 (0.679–0.716)	0.737 (0.714–0.759)
Delong’s test	ΔAUC = 0.034, P = .013	ΔAUC = 0.014, P = .113	ΔAUC = 0.034, P = .017
All predictors
Model 1†	0.703 (0.689–0.717)	0.683 (0.665–0.701)	0.703 (0.681–0.725)
Model 6#	0.748 (0.734–0.762)	0.730 (0.712–0.748)	0.748 (0.726–0.771)
Delong’s test	ΔAUC = 0.045, P < .001	ΔAUC = 0.047, P < .001	ΔAUC = 0.046, P < .001

AIP = atherogenic index of plasma, BMI = body mass index, ePWV = estimated pulse wave velocity, HDL-C = high-density lipoprotein cholesterol, LDL-C = low-density lipoprotein cholesterol, LMR = lymphocyte-to-monocyte ratio, MVPA = moderate-to-vigorous physical activity, NLR = neutrophil-to-lymphocyte ratio, non-HDL-C = non-high-density lipoprotein cholesterol, NPR = neutrophil-to-platelet ratio, PLR = platelet-to-lymphocyte ratio, SII = systemic immune-inflammation index, TG/HDL-C = triglyceride-to-high-density lipoprotein cholesterol, uACR = urinary albumin-to-creatinine ratio.

^†Survey-weighted multivariable logistic regression models included age, sex, race, education level, PIR, eGFR, diabetes mellitus, cardiovascular disease, cancer, current smoking, alcohol consumption, and MVPA.

^‡Survey-weighted multivariable logistic regression models included model 1 and BMI.

^§Survey-weighted multivariable logistic regression models included model 1 and atherogenic lipid profiles (AIP, HDL-C, TG/HDL-C, non-HDL-C, Castelli I, Castelli II).

^∥Survey-weighted multivariable logistic regression models included model 1 and immune-inflammatory indices (SII, NLR, PLR, LMR, NPR).

^¶Survey-weighted multivariable logistic regression models included model 1 and vascular aging indices (ePWV, uACR).

^#Survey-weighted multivariable logistic regression models included model 1 and above all predictors.

Significance level: *P < .05.

Significance level: **P < .01.

Significance level: ***P < .001.

The most pronounced improvement was observed when all predictors were combined in model 6, with an AUC of 0.748 (95% CI: 0.734–0.762), corresponding to a ΔAUC of 0.045 compared with the base model (P < .001). Subgroup analyses revealed similar trends in women and men, although the magnitude of improvement was generally greater among women. These results demonstrate that metabolic, inflammatory, and vascular markers collectively enhance the ability to classify individuals with moderate-to-severe depression beyond traditional risk factors.

3.5. Sensitivity analyses using alternative PHQ-9 thresholds

To address potential heterogeneity across the full spectrum of depressive symptoms, we further examined baseline characteristics and biomarker associations across all 5 PHQ-9 severity categories (0–4, 5–9, 10–14, 15–19, and 20–27). As shown in Table S4, Supplemental Digital Content, https://links.lww.com/MD/R365, increasing depression severity was associated with progressively higher BMI, more adverse atherogenic lipid profiles, elevated immune-inflammatory indices, and greater vascular aging markers. These trends were generally more pronounced among women than men.

Using multinomial logistic regression with PHQ-9 scores of 0 to 4 as the reference category, we observed graded increases in ORs for most cardiometabolic, inflammatory, and vascular markers across increasing depression severity levels (Table S5, Supplemental Digital Content, https://links.lww.com/MD/R365). Notably, markers such as the AIP, triglyceride-to-HDL cholesterol ratio, ePWV, and uACR demonstrated a clear dose–response pattern, whereas PLR showed no consistent association. Sex-stratified analyses yielded similar patterns, with generally stronger associations observed among women, supporting the robustness of the primary findings.

To further evaluate the robustness of predictive performance, we assessed model discrimination across multiple PHQ-9 thresholds (≥5, ≥10, ≥15, and ≥20). As shown in Table S6, Supplemental Digital Content, https://links.lww.com/MD/R365, models incorporating cardiometabolic, inflammatory, and vascular biomarkers consistently improved discrimination compared with the base model across all thresholds. The largest gains in AUC were observed when all biomarker domains were combined, with consistent improvements in both women and men.

4. Discussion

In this large, nationally representative cohort of US adults, we observed consistent associations of BMI, atherogenic lipid indices, systemic immune-inflammatory markers, and vascular aging measures with moderate-to-severe depression. These results reinforce the concept that depression is not merely a psychiatric disorder but rather a systemic condition with strong metabolic, immune, and vascular components.^[2,21,22] Our findings add to previous studies by integrating multiple biological domains into a unified framework and highlighting sex-specific vulnerabilities, thus extending prior work that has often focused on isolated biomarkers.^[23-25]

Obesity and depression have been linked in numerous longitudinal and meta-analytic studies.^[26,27] Elevated BMI was independently associated with depression in our analysis, consistent with prior evidence that excess adiposity increases the risk of incident depressive episodes.^[28] Adipose tissue secretes pro-inflammatory cytokines, including tumor necrosis factor-α and interleukin-6, which activate hypothalamic–pituitary–adrenal axis dysregulation and serotonergic imbalance.^[29] Insulin resistance associated with obesity contributes to altered glucose metabolism in the brain, oxidative stress, and mitochondrial dysfunction, which can exacerbate depressive symptomatology.^[30,31] The longitudinal trajectory of obesity leading to chronic inflammation and subsequent mood disturbance has been demonstrated in prospective cohorts, suggesting a plausible causal pathway.^[13,32]

Our results also underscore the role of lipid dysregulation in depression. Prior reports have shown that low HDL cholesterol and high triglyceride-to-HDL ratios are associated with increased risk of depression.^[8,33,34] Lipids modulate neuronal membrane fluidity and receptor signaling, thereby influencing serotonin transporter binding and neurotransmission.^[35] Oxidized lipoproteins also promote monocyte activation and endothelial dysfunction, which contribute to neurovascular injury.^[36,37] The stronger associations observed among women may reflect sex-related differences in lipid metabolism, hormonal regulation, and immune function.^[38-40] For example, estrogen influences both lipid homeostasis and cytokine signaling, and its fluctuations during reproductive transitions may amplify susceptibility to depression.^[41,42]

Chronic immune activation emerged as another significant correlate of depression. Systemic inflammatory indices, such as neutrophil-to-lymphocyte and NPRs, capture shifts in immune balance that have been implicated in psychiatric illness.^[43,44] Elevated inflammatory cytokines activate the kynurenine pathway, diverting tryptophan metabolism away from serotonin synthesis and leading to the production of neurotoxic metabolites that impair neuroplasticity.^[45,46] Neuroimaging studies further demonstrate that peripheral inflammation is linked to reduced connectivity in corticolimbic circuits, providing mechanistic evidence for the immune hypothesis of depression.^[47,48] Moreover, Mendelian randomization studies indicate a bidirectional link, with genetically elevated inflammatory markers increasing depression risk.^[49,50]

Vascular aging markers, including arterial stiffness and albuminuria, were consistently associated with depression in our analysis. These findings support the vascular depression hypothesis, which proposes that cerebrovascular pathology contributes to late-life depression.^[51,52] Arterial stiffness reduces cerebral perfusion, promotes white matter hyperintensities, and disrupts the blood–brain barrier, thereby facilitating neuroinflammation.^[53,54] Albuminuria reflects systemic microvascular damage and endothelial dysfunction, both of which have been linked to depressive symptoms and cognitive decline.^[55,56] Prior prospective studies have confirmed that greater pulse wave velocity and microalbuminuria predict incident depression.^[56,57]

An integrated interpretation of our findings suggests a sequential pathway in which obesity initiates systemic inflammation and dyslipidemia, which subsequently accelerate vascular aging, ultimately increasing vulnerability to depression. Evidence from longitudinal cohorts supports such temporal ordering, as obesity predicts inflammation, inflammation predicts vascular dysfunction, and vascular injury predicts depression onset.^[58-60] This framework aligns with systems biology perspectives that view depression as a disorder emerging from the intersection of metabolic, immune, and vascular pathways.^[61,62]

Sex-specific analyses revealed that women displayed greater sensitivity to atherogenic lipid indices, whereas men showed relatively weaker associations. This observation is consistent with epidemiologic data indicating that women have nearly twice the lifetime prevalence of depression compared with men.^[38] Biological explanations include sex differences in sex hormone fluctuations, immune system reactivity, and mitochondrial function.^[63,64] For instance, estrogen exerts both neuroprotective and pro-inflammatory effects depending on the context, and its decline during perimenopause is associated with heightened vulnerability to depression.^[65] Women also display more robust humoral immune responses, which may amplify inflammatory contributions to mood disorders.^[66]

From a clinical and public health perspective, these findings have several implications. Depression risk stratification may be improved by incorporating metabolic, inflammatory, and vascular biomarkers alongside psychosocial factors.^[67,68] Interventions targeting weight, lipid levels, and inflammation may reduce depression risk. Randomized trials show that physical activity, dietary modification, and statin therapy lower systemic inflammation and improve mood, while anti-cytokine therapies demonstrate antidepressant effects in patients with high baseline inflammation.^[68,69] Vascular protective agents, such as antihypertensives, may also serve as adjunctive strategies for prevention.^[70]

Importantly, our severity-specific analyses demonstrated a graded relationship between cardiometabolic, inflammatory, and vascular aging markers and depressive symptom burden, suggesting that these biological alterations may accumulate across the continuum of depression severity rather than being confined to a dichotomous clinical threshold. This finding supports the biological plausibility of our primary results and addresses concerns regarding potential information loss from binary categorization of depression. Nevertheless, these severity-specific analyses should be interpreted as exploratory.

Despite these strengths, our study has limitations. The cross-sectional design precludes definitive causal inference, and bidirectional associations cannot be excluded, as depression itself promotes behavioral and physiological changes that worsen metabolic and vascular health.^[71] Although survey weights improve generalizability, residual confounding from unmeasured variables is possible. Future longitudinal analyses, particularly those integrating genetic, neuroimaging, and biomarker data, are needed to clarify causal mechanisms. Interventional studies will be crucial to determine whether modifying obesity, lipid abnormalities, inflammation, or vascular dysfunction can meaningfully reduce depression risk or improve treatment outcomes.^[6,37,72]

5. Conclusion

In summary, this nationally representative study demonstrates that higher BMI, atherogenic lipid phenotypes, systemic immune-inflammatory activation, and vascular aging are all independently associated with moderate-to-severe depression. These findings reinforce the view of depression as a systemic disorder with contributions from metabolic, immune, and vascular pathways. Integrating these biological domains into risk assessment frameworks may improve early detection and guide preventive strategies. Our results highlight obesity-inflammation-vascular aging as a potential mechanistic pathway to depression, and suggest that interventions aimed at weight control, lipid regulation, inflammation modulation, and vascular protection may hold promise in reducing the burden of depression at both population and individual levels.

Author contributions

Conceptualization: Hongjin An, Li Zhang, Yiran Li.

Data curation: Jing Liu, Hailan Zhu.

Methodology: Jing Liu, Hailan Zhu.

Software: Jing Liu, Hailan Zhu.

Visualization: Jing Liu, Hailan Zhu.

Writing – review & editing: Hongjin An, Li Zhang, Yiran Li, Lingmei He, Yachong Zhao.

Supplementary Material

medi-105-e47657-s001.xlsx ^{(76.4KB, xlsx)}