NHANES is a continuing survey of the health and nutritional status of the United States population (10), and it is conducted by members of the Centers for Disease Control and Prevention (NCHS). Participant data are obtained via an intricate, classified, multistage probabilistic cluster sampling design, and involves 5,000 people each year. The collected information includes subject demographics, body measurements, laboratory test results, and dietary information (11). Prior to the survey, all subjects provided a written informed consent for survey participation and data usage in health-related statistical research.

Program details, data acquisition protocols, and available data are open to the public, and can be accessed at https://wwwn.cdc.gov/nchs/nhanes/. This investigation strictly followed the Strengthening the reporting of observational studies in epidemiology (STROBE) criteria for reported of observational studies in epidemiology (12). We examined 2 survey cycles spanning the years 2007–2010. Subjects were aged ≥ 40 with complete data, and among those excluded from analysis were participants with incomplete marital, family income-to-poverty ratio, education status, body mass index (BMI), smoking consumption status, alcohol consumption status, diabetes, phenotype age, SII, DII, LE8 or other information. Ultimately, we included 2,901 subjects in our analyses (Supplementary Figure 1).

Adult Treatment Panel III (ATP III) of the National Cholesterol Education Program (NCEP) (3) states that MetS is the presence of ≥3 of the following characteristics: concentric obesity (with waist circumference > 35 inches or 88 cm among women and 40 inches or 102 cm among men), insulin resistance (fasting blood glucose > 100 mg/dL or lack of diabetes treatment), hypertension (SBP 130 mmHg or DBP 85 mmHg or requirement of hypertensive medication), elevated triglycerides (TG 150), and reduced high-density lipoprotein (HDL) cholesterol (HDL 40 among men and HDL 50 among women).

The LE8 scoring system was based on the following questionnaires: 4 health behavior questions (diet, physical activity, nicotine exposure, and sleep duration) and 4 health parameters [BMI, non-HDL cholesterol, blood glucose (BG), and blood pressure (BP)]. The total score ranged between 0 and 100, and the overall LE8 score was calculated based on the average of 8 metrics. Detailed algorithms for calculating the LE8 scores for each of the metrics to NHANES data can be found in Supplementary Table 1. Following the American Heart Association definition, the LE8 scores were considered low (0–49 points), moderate (50–79 points), and high (80–100 points) (13). Dietary metrics were evaluated based on the 2015 Healthy Eating Index (HEI). Subjects were asked to recall two 24-h diets, and the data was integrated into the US Department of Agriculture (USDA) food pattern equivalence data to compute the HEI-2015 scores (14). Moreover, we utilized self-reports of the frequency and duration of intense or moderate physical activity over the past 30 days, and recorded the smoking status, sleep duration, diabetes history, medication history, BP, height, and weight during physical examination. BMI was defined as weight (kg) divided by height (m) squared. BP was recorded as the average reading of 3 consecutive measurements. Subjects were fasted for ≥8 h, prior to 5 mL blood sample collection, and the samples were dispatched to the central laboratory for complete blood count and biochemical (namely, blood lipids, BG, CRP and glycosylated hemoglobin) evaluation.

Systemic inflammation was measured using both biochemical or hematological indicators in normal blood tests or as ratios from the measurements. SII was computed as follows: SII = P × N/L, whereby, P, N, and L were the preoperative peripheral platelet, neutrophil, and lymphocyte counts, respectively (15).

DII was first introduced in 2004, its first-generation protocols and verification were presented in 2009, and an enhanced version was offered in 2014 using comparative data from 11 national food and nutrition databases from 4 separate continents. We also measured relevant inflammatory markers, namely, CRP, IL-1β, IL-4, IL-6, IL-10, and TNF-α, to generate food parameter-specific DII scores, which, were, in turn, used to generate an overall DII score for individual subjects analyzed in this study (16).

The NHANES data comes with 28 available parameters that can be used to compute DII. These include energy, protein, carbohydrate, dietary fiber, total fatty acid, total saturated fatty acid, monounsaturated fatty acids (MUFA), polyunsaturated fatty acids (PUFA), cholesterol, β-carotene, niacin, folate, magnesium, iron, zinc, selenium, caffeine, alcohol, n3 polyunsaturated fatty acid, n6 polyunsaturated fatty acid, and vitamins A, B1, B2, B6, B12, C, D, and E. An augmented DII score represents an inflammation-triggering diets, and a reduced score represents an anti-inflammatory diet (17).

We utilized the Klemera and Doubal method of calculating biological age using 8 biomarkers, namely, CRP, circulating creatinine, glycosylated hemoglobin, circulating albumin, circulating total cholesterol, circulating urea nitrogen, circulating alkaline phosphatase, and systolic BP (18).

The phenotypic age was computed using 9 age-related variables, namely, chronological age, albumin, creatinine, glucose, CRP, lymphocyte percent, average cell volume, red blood cell distribution width, alkaline phosphatase, and white blood cell count (19).

Circulating Klotho levels were assessed using ELISA (IBL International, Japan). All samples were twice assessed and the average of the two results were analyzed. Per plate, we also evaluated two quality control samples using reduced and elevated Klotho concentrations for accuracy assessment. The sample analyses were repeated if the assigned value was >2 standard deviations from the expected value. The assay sensitivity was 4.33 pg./mL, and both the intra- and inter-assay coefficient of variations were <5%. Please refer to https://wwwn.cdc.gov/Nchs/Nhanes/2007-2008/SSKL_E.htm for a detailed summary of the ELISA protocol (20).

Covariates were defined as patient age, sex, ethnicity/race, marital status, family income to poverty ratio, education status, BMI, smoking consumption status, alcohol consumption status, and physical activity. Race was classified as white, black, Mexican, other. Education status was described as <11th grade, high school graduate, college graduate or above. Family income to poverty ratio was separated into 4 categories, based on a 1.3 to >5 scale as follows: 1.3, 1.3–3, 3–5, and >5. An augmented ratio represented a higher income status. Marital status was classified as follows: married, divorced, unmarried. Alcohol consumption frequency and consumer sex was recorded, and alcohol consumption was defined as: never, former, mild, moderate, and heavy (21). Smoking status can categorized as: currently smoking, formerly smoking, or never. Never smokers were those who smoked <than 100 cigarettes in their lives, ex-smokers were those who smoked > 100 cigarettes, but no longer smoking, and current smokers were defined as smoking > 100 cigarettes, with consistency or inconsistency (22). BMI was defined as follows (23): <25, 25–30, and >30 kg/m². Physical activity (PA) can be divided into: People were asked about the types of exercise they did in the past month. These included walking, running, cycling, swimming, dancing, weightlifting, and other activities. The intensity of each activity was given as a metabolic equivalent (MET) and is in the original dataset. We multiplied the exercise frequency and its MET to get an activity score to evaluate the PA levels.

We employed the NHANES criteria for statistical analysis (oversampling, stratification, and clustering) to predict the appropriate number of U.S. adults to use in this study. In addition, we also evaluated the required statistical tests for weight adjustment. Categorical variables are presented in n (%), and inter-group differences were assessed via chi-squared test.

We also employed logistic regression models to assess relationships between inflammatory indices (SII, DII), biological aging (phenotypic age, circulating Klotho, biological age), and LE8 with MetS. To control for confounding factors, three weighted logistic regression models were used in this study: Crude model, No adjustment for any potential influence factors; Model 1, Adjusted for Sex, Age and Ethnic/race; Model 2, Adjusted for Sex, Age, Ethnic/race, Marital, Family income-to-poverty ratio Education levels and Alcohol consumption status. The acquired data are presented as weighted (OR [95% CI]). We also assessed sub-categories like patient gender, age, and ethnicity to further elucidate the aforementioned relationships within various populations. Herein, the LE8 (high, moderate and low CVH), health behavior (high, moderate and low CVH) and health factor (high, moderate and low CVH) scores were regarded as the reference group. We generated restricted cubic spline (RCS) plots to display patterns in variables of significance in logistic regression. Using the RCS plots, we determined the presence or absence of a nonlinear association between the mentioned exposure factors and MetS.

The possibility of the inflammatory index (SII, DII) and biological aging (phenotypic age, circulating Klotho, biological age) as a modulator of the LE8 and MetS association was further examined using a parallel mediation model employing a quasi-Bayesian Monte Carlo technique with 1,000 normal approximation-based simulations (R package “mediation”). Direct effect (DE) indicated the LE8-mediated regulation of MetS in the absence of mediators. Indirect effect (IE) indicated the LE8-mediated modulation of MetS using a mediator. Mediation was quantified as follows: IE divided by TE (total effect).

Various packages in R 4.2.2 (R Project for Statistical Computing), including, nhanesR (0.9.2.8), survey, compareGroups, dplyr, tidyverse, do, scatterplot3d, MASS, poLCA, finalfit, Hmisc, lattice, Formula, rms, and foreign were used to conduct the analyses. Significance was set at two-sided p-value < 0.05.

Among the 2,901 subjects, 1,153 had a MetS diagnosis. Table 1 summarizes all participant demographics. In all, we observed marked differences in patient age, household income to poverty ratio, education status, BMI, smoking status, alcohol intake, physical activity, LE8 and LE8 components (health factors and health behaviors), various scores, including HEI, PA, Sleep, BMI, Non-HDL, Glucose, and BP, as well as phenotypic age, biological age, and DII among MetS and non-MetS subjects.

Table 1 summarizes all participant demographics. Among the 2,901 participants included in this study (48.2% female), a mean age of 54.9 years, and a mean LE8 score of 70.4533 (standard error of 0.5150). 1, 153 individuals were diagnosed with MetS. Participants with MetS were older, had a higher BMI, and were married compared to those without MetS. The MetS group had lower LE8 scores and had higher Phenotypic age, Biological age, and DII.

As depicted in Supplementary Figure 2, there were also strong differences in the MetS-ATP and related components among MetS and non-MetS subjects. In particular, MetS subjects exhibited substantially enhanced hyperglycemia, abnormal TG, abnormal HDL, and obesity, relative to non-MetS, however, the hypertensive incidences were similar between MetS and non-MetS participants.

Based on our LE8 and related component univariate analyses, biological aging and inflammatory markers were strongly associated with MetS risk (Table 2). Apart from the PA score, smoke score, circulating Klotho and SII, other covariates showed marked association with MetS development (p < 0.05).

Table 3 depicts the conclusions of our multivariate analysis of LE8 and its components with MetS. Following possible confounding factor adjustment, we developed several models for evaluation of the association between LE8 and MetS. We revealed that, with a rise in the LE8 and health factor scores, there was a drastic reduction in MetS risk (P-trend < 0.05). The MetS incidence odds ratios (95% confidence intervals) among LE8 middle and high scoring cohorts were (OR = 0.5163, 95Cl 0.2818~0.9460) and (OR = 0.2027, 95Cl 0.0871~0.4714), respectively, relative to the LE8 low scoring cohort. Furthermore, relative to the low health factor scores, the MetS odds ratios (95% confidence interval) among middle and high health factor scores were (OR = 0.1737, 95%Cl 0.1080~0.2794), (OR = 0.0278, 95%Cl 0.0130~0.0591) respectively (Figure 1). Using the 8 LE8 components as continuous variables, and following confounding factors adjustment, we revealed that 4 health factor scores were inversely proportional to MetS risk, and the OR values were as follows: BMI (OR = 0.9663, 95%Cl 0.9598~0.9728), Glucose (OR = 0.9768, 95%Cl 0.9678~0.9859), BP (OR = 0.9813, 95%Cl 0.9724~0.9903), and Non-HDL (OR = 0.9843, 95%Cl 0.9768~0.9918).

Scoring the LE8 scores and health factors and health behaviors as low, medium and high, after adjusting for the covariates of Sex, Age and Ethnic/race in Model 1, the ORs of participants in the high LE8 grouping and the high health factors grouping compared to the low scoring group were (OR = 0.1955, 95% Cl 0.0903,0.4232), (OR = 0.0282, 95%Cl 0.0140,0.0567) (p < 0.05 for trend), and after further adjusting for Marital, Family income-to-poverty ratio Education levels and Alcohol consumption status, the associations in model 2 remained unchanged, with a 79.73% reduction in MetS risk for the high subgroup of LE8 compared to the low subgroup, OR = 0.2027 (95%Cl 0.0871,0.4714), and a 97.22% reduction in MetS risk for the high subgroup of health factors compared to the low subgroup, OR = 0.0278 (95%Cl 0.0130,0.0591). When the eight components of LE8 were used as continuous variables, it was observed that after adjusting for confounders, the four health factor scores were negatively associated with MetS, with OR values of Score BMI (OR = 0.9663, 95% Cl 0.9598~0.9728), Score Glucose (OR = 0.9768, 95% Cl 0.9678~0.9859), Score BP (OR = 0.9813, 95%Cl 0.9724~0.9903), and Score Non-HDL (OR = 0.9843, 95%Cl 0.9768~0.9918), whereas Health Behavior as a whole and its four components were not significantly associated with MetS (p > 0.05).

As presented in Figure 2, following confounding factor adjustment, a nonlinear association was discovered between MetS risk and several variables, including, the LE8 scores, health behavior scores, health factor scores, phenotype age, circulating Klotho, biological age, SII, DII, HEI, Physical Activity, Smoke, Sleep, BMI, Non-HDL, Glucose, and BP (P-nonlinear <0.001; estimated OR = 1).

Nonetheless, we observed no marked independent correlation between circulating Klotho, DII, HEI, physical activity, non-HDL, and glucose and MetS risk in the overall model (P-overall > 0.05).

Table 4 summarizes the subgroup analysis results from LE8 and MetS risk. Suggested that the risk of MetS prevalence significantly decreased with progressively higher LE8 and health factor group scores across gender, race, household economic status, BMI, physical activity, and alcohol intake (P-trend < 0.05). In case of women, whites, other races, household income-to-poverty ratio ≥ 5, BMI ≥ 30, and heavy drinkers, elevation in health behavior scores drastically decreased MetS risk (p < 0.05 for trend).

Furthermore, we also assessed associations between MetS and 2 inflammatory indices and 3 biological aging indicators. Based on our analysis, biological and phenotypic ages were strongly associated with MetS risk in various gender, race, family economic status, BMI, physical activity and alcohol intake populations (p < 0.05; Supplementary Tables 3A,B). Higher levels of CVH were associated with lower risk of MetS compared to low CVH (P-trend<0.05), with the strongest association being physical activity, with a high CVH OR = 0.0000 (95%Cl 0.0000,0.0000) compared to low CVH in the <600 min/month group. Further analysis of its components (health factors and health behaviors) in LE8, see Supplementary Tables 2A,B, similarly saw the same trend of negative associations between health factor scores and MetS across subgroups (P-trend < 0.05). In contrast, the negative association of health behavior scores with MetS was correlated among participants who were female, white, had a household income to poverty ratio of ≥5, BMI ≥ 30, Physical Activity ≥ 600, and heavy drinkers, with ORs for high-scoring health behaviors compared to low-scoring health behaviors of (0.4676, 95%Cl 0.2669,0.8191), (0.5399, 95%Cl 0.3285,0.8873), (0.3444,95%Cl 0.1251,0.9483), (0.3790, 95%Cl 0.2073,0.6928), (0.5829, 95%Cl 0.3279,1.0362), (0.3206, 95%Cl 0.1728, 0.5948) (P-trend < 0.05).

The association of two inflammatory indices and three biological aging indicators with MetS was also examined in subgroup analyses, suggesting that biological age and phenotypic age were positively associated with MetS among participants in different gender, race, household economic status, BMI, physical activity, and alcohol intake groups (p < 0.05; Supplementary Tables 3A,B).

We employed parallel mediation assessment to assess modulatory roles of biological aging (phenotypic age, circulating Klotho, biological age) and inflammatory index (SII, DII) in the intricate relationship between the LE8, health behavior, and health factor scores and MetS risk.

Of note, phenotypic age exhibited a strong modulatory role on the relationship between LE8, health behavior, and health factor scores and MetS risk, with mediation proportions 0.13%, p ≤ 0.001; 0.0.69%, p = 0.240; 0.08%; and p ≤ 0.001, respectively.

Biological age exhibited a strong modulatory role on the relationship between LE8 and MetS risk, with mediation proportions at 0.08%, p ≤ 0.001 (Figure 2).

A new MetS definition was presented at the International Diabetes Federation (IDF) workshop in 2005, which was distinct from the ATP III description in that it emphasized central obesity as a prerequisite (23). In 2009, IDF renewed the MetS definition yet again, reducing the importance of waist measurement from 2006. Under the new definition, waist size was only used as a preliminary screening tool, with strict thresholds set for the waist circumference, based on distinct ethnic populations (24). Given that certain articles employed the IDF-based MetS diagnosis in 2005 and 2009, sensitivity analysis was conducted to validate whether distinct diagnostic criteria impacted the conclusions of this investigation (Supplementary Tables 4–7). We revealed robust inverse relationships between LE8 and health factors and MetS even under different diagnostic criteria (Figure 3).