This study utilized data from the National Health and Nutrition Examination Survey (NHANES), a program conducted in the United States to assess the health and nutritional status of both adults and children (https://www.cdc.gov/nchs/nhanes/). NHANES is a cross-sectional survey based on the general U.S. population. Data from NHANES 2007–2015 were analyzed, and 29,040 participants were initially selected, all of whom provided written informed consent. The focus was primarily on adults aged 20 and above. After combining the databases using unique identifiers for each participant, cases with missing data on physical examinations, alcohol consumption, smoking, and carotenoid intake (19,335 cases) were excluded. Ultimately, 7,554 participants were included in the study (Figure 1).

Biological age was assessed at three different levels to quantify the degree of biological aging, including cardiovascular biological age (BaC), liver biological age (BaL), and kidney biological age (BaK). The biomarkers related to BaC were collected, including triglycerides, high-density lipoprotein, low-density lipoprotein, fasting blood glucose, systolic blood pressure, diastolic blood pressure, and total cholesterol. The biomarkers related to BaL included gamma-glutamyl transferase, serum aspartate aminotransferase, serum alanine aminotransferase, and albumin, because of the low organ specificity, ALP elevation may also occur in bone diseases (13), placental diseases and intestinal diseases (14), it was not included in the evaluation of liver biological age. The biomarkers for BaK included serum creatinine and serum uric acid. The biological age of different organs was calculated using the biological age algorithm KDM (15, 16). The ratio of biological age to chronological age (BA ratio) was used as an indicator for evaluating biological aging. Additionally, the difference between biological age and chronological age (BA acceleration) was used as a measure of aging acceleration.

Lutein and zeaxanthin (LZ), as carotenoids in the diet, play an important role in maintaining human health. Based on the variables collected in the NHANES database, we included both dietary and supplement sources of LZ and calculated the total intake of LZ from both sources. The intake of LZ from both sources was measured over 2 days, and the average intake of the 2 days was used as the final LZ intake value.

The following variables were collected: SMQ020, SMQ040, SMQ050Q, SMQ050U. Smoking status was classified into three categories based on the information from these variables: Current smoker, Former smoker, and Never smoker. Individuals who had smoked fewer than 100 cigarettes in their lifetime were considered as “Never smokers,” while those who had stopped smoking for more than 1 month were classified as “Former smokers.”

BMI was categorized into the following groups based on BMI values: “Underweight (< 18.5),” “Normal (18.5 to < 25),” “Overweight (25 to < 30),” “Obese (30 or greater).”

Total calorie intake was obtained from the variables collected in the NHANES database, including intake data from both the first and second days. If data from the second day were available, the average intake from both days was used; otherwise, only the intake from the first day was considered.

The following variables were collected: “ALQ11,” “ALQ10,” “ALQ120Q,” “ALQ120U.” The alcohol consumption was classified into four categories: “Non-drinker,” “ < 1 drink/month,” “1–10 drinks/month,” and “10+ drinks/month.” Participants who had consumed fewer than 12 alcoholic drinks in their lifetime were categorized as “Non-drinker.”

We included additional covariates based on previous studies. Participants' demographic information, including gender, age, race (Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, and Other races), education level (less than 9th grade, 9-11th grade, high school graduate, some college/AA degree, and college graduate), and household income, were collected via questionnaires. These factors may have potential associations with biological aging or dietary carotenoid intake.

Survival data were linked to the National Death Index (NDI), and mortality follow-up data for NHANES participants from 2007 to 2015 were incorporated to construct Cox proportional hazards models to evaluate the effect of lutein intake on all-cause mortality.

The processed data were further adjusted using Propensity Score Matching (PSM) to account for potential confounders, including age, gender, race, education level, and income.

Data mining and analysis were performed using R version 4.4.0. Microarray datasets from Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) were retrieved using the R package “GEOquery” (17). Specifically, we used the dataset GSE151683 (18), which includes transcriptomic data related to high lutein and high lycopene intake. In this study, we focused on the high lutein intake group and the low lutein control group, and differential gene expression was analyzed using the “limma” package (19) to identify fold changes between groups. The differentially expressed genes were then subjected to subsequent analysis.

Biological process enrichment analysis was performed using the Gene Prioritization by Evidence Counting (GPEC) algorithm from Metascape (http://metascape.org/), with the inclusion criteria of P value < 0.01, a minimum count of 3, and an enrichment factor >1.5. Clustering of the entry nodes was conducted based on the similarity of the members. Gene set enrichment analysis (GSEA) was conducted using the R package “clusterProfiler” (20) to assess changes in biological functions following sustained high lutein intake. The background gene sets (C2: curated gene sets, C5: ontology gene sets) were provided by the Molecular Signatures Database (MSIGDB) (https://www.gsea-msigdb.org/gsea/index.jsp). Metabolic pathway enrichment was performed using the R package “IOBR” (21), integrating three algorithms (ssGSEA, PCA, and z-score) to quantify the metabolic enrichment level of each sample. Differential scoring analysis was then performed using “limma” based on t-values to evaluate metabolic differences between groups.

Immune scores for different samples were calculated using the immune algorithms (xCell, MCPcounter) from the R package “IOBR” to quantify differences in the abundance of various immune cell types in peripheral blood after sustained high lutein intake between different groups. MCP-counter and xCell algorithms quantified the abundance of immune cells between clusters based on marker gene expression. Only immune cell types showing statistically significant differences (P < 0.05) between groups were reported.

All analyses were performed using R software version 4.4.0. Non-linear relationships between lutein intake and biological age acceleration across different patterns were tested using restricted cubic splines (RCS) implemented in the “rms” R package. Continuous variables were expressed as interquartile ranges (IQR), and categorical variables were expressed as counts (n) and percentages (%). Multivariate logistic regression models and multivariate Cox regression models were used to analyze the relationship between lutein intake and biological age acceleration across different patterns, as well as the association with all-cause mortality. Lutein intake levels were categorized into quartiles (Q1, Q2, Q3, and Q4) with Q1 as the reference group, and the odds ratios (ORs) and 95% confidence intervals (95% CIs) were used to describe the relationship between lutein intake and biological aging. Further non-linear relationship tests between lutein intake and biological age acceleration were conducted using RCS with the “rms” R package. Demographic differences in carotenoid intake levels, including age, gender, race, education, smoking, alcohol intake, ratio of family income to poverty (PIR), BMI, waist circumference, total calorie intake, and different sources of LZ intake, were compared using the Kruskal-Wallis test. Interaction analysis was performed using the likelihood ratio test. Paired differences between groups were compared using two-tailed paired Student's t-test. A P value of < 0.05 was considered statistically significant for all analyses.

In this study, after applying the inclusion and exclusion criteria, we initially included 9,705 participants (Figure 1). We divided the participants into four groups (Q1–Q4) based on their total LZ intake, ranging from low to high. The Q1 group [0.32 (0.21, 0.41)] had the lowest LZ intake level, while the Q4 group [3.04 (2.19, 5.10)] had the highest. Baseline characteristics showed significant differences in the distribution of most covariates across the four groups (Supplementary Table S1). To reduce the confounding effect of covariates on the epidemiological outcome variables, we performed PSM to adjust for data distribution. After PSM adjustment, 7,554 participants were included in the final analysis (Table 1).

Adjusted baseline data indicated significant differences in both dietary and supplement sources of LZ intake across the four groups (P < 0.001). There were also significant differences in the distribution of gender (P = 0.002), age (P = 0.004), education level (P < 0.001), income (P < 0.001), total calorie intake (P < 0.001), and waist circumference (P = 0.048) across the different lutein intake groups (Table 1). In terms of education level, the highest intake group (Q4) had a larger proportion of participants with higher education (P < 0.001). Among the Q4 group, 35% had a college degree or higher, while only 3.3% had less than a 9th-grade education. The Q4 group also had the highest income level [3.30 (1.64, 5.00)]. Additionally, the total calorie intake in Q3 [2,127 (1,647, 2,749)] and Q4 [2,010 (1,588, 2,641)] was relatively higher.

To evaluate whether LZ intake influences the progression of biological aging, we calculated the biological age at various levels (cardiovascular, kidney, liver, and overall) for participants and assessed the aging trend using biological age ratios.

First, we evaluated the distribution of biological age ratios for different levels of dietary LZ intake (Supplementary Table S2). As LZ intake increased, the biological age ratios for different organ systems (cardiovascular, kidney, liver, and overall) significantly decreased (pre-PSM adjustment: P < 0.05). After adjusting for covariates using PSM, although the median values for all four biological age ratios decreased, only the overall biological age ratio showed a significant difference (P = 0.025), with the ratios for Q1 and Q4 being 0.99 (0.84, 1.16) and 0.95 (0.82, 1.13), respectively. Next, we evaluated the distribution of biological age ratios at different total LZ intake levels. As total LZ intake increased (Table 2), the biological age ratios for cardiovascular, kidney, liver, and overall aging significantly decreased (before PSM adjustment: P < 0.001 or P = 0.001). After PSM adjustment, significant differences were still observed in three biological age ratios (kidney, liver, and overall), with median values for organ and overall biological age ratios in the high LZ intake groups (Q3 and Q4) remaining below 1, indicating that LZ intake from multiple sources collectively influences the biological aging process.

To further assess the impact of LZ intake on the acceleration of biological aging in different organ systems, we calculated the biological age acceleration for four types (cardiovascular, kidney, liver, and overall) and constructed three models adjusted for different covariates (Figure 2). The results showed that total LZ intake significantly reduced the biological age acceleration in various organs (cardiovascular, kidney). Cardiovascular aging acceleration decreased by 0.89 (95% CI: 0.83–0.95), 0.90 (95% CI: 0.85–0.96), and 0.92 (95% CI: 0.87–0.99) in the three models, respectively. Kidney aging acceleration decreased by 0.91 (95% CI: 0.86–0.96), 0.90 (95% CI: 0.85–0.96), and 0.92 (95% CI: 0.88–0.97), respectively. Total LZ intake also reduced the overall aging acceleration in all three models by 0.90 (95% CI: 0.84–0.95), 0.91 (95% CI: 0.85–0.96), and 0.93 (95% CI: 0.88–0.93), respectively. After adjusting for covariates, the effect of total LZ intake on liver biological age acceleration was attenuated.

We further evaluated the differential effects of LZ intake on biological aging acceleration at different intake levels (Figure 3). In three models measuring overall biological age acceleration, compared to the Q1 group, the biological age acceleration in the Q4 group was reduced by 0.21 (95% CI: 0.03–0.26), 0.29 (95% CI: 0.10–0.84), and 0.32 (95% CI: 0.12–0.84), with significant trend tests in all three models (P for trend: 0.006; 0.021; 0.029). Additionally, we assessed the differential effects of LZ intake on the biological aging acceleration of other organs at different intake levels (Supplementary Table S3). For kidney and liver biological aging acceleration, high LZ intake (Q4) showed more significant differences in the effect on aging acceleration compared to low LZ intake (Q1) (all P < 0.05). For kidney and liver biological aging acceleration, the trend tests for the different LZ intake levels were also more significant (all P < 0.05).

To further assess the linear relationship between total LZ intake and biological age acceleration, we constructed three models of Restricted Cubic Splines (RCS) for the evaluation of biological aging acceleration across different patterns. For overall biological aging acceleration (Figure 4), in Model 1 (P overall < 0.0001), Model 2 (P overall = 0.0001), and Model 3 (P overall = 0.0005), the correlation between the biological aging acceleration beta values and the transformed total LZ intake was significant, with no significant non-linear relationship observed. Additionally, in the evaluation of biological aging acceleration in different organs (Supplementary Figure S1), after multiple adjustments in the models, significant correlations were still evident (Model 3_Cardio, P = 0.0241; Model 3_Kidney, P = 0.0162; Model 3_Liver, P = 0.0021). Although none of the models for biological aging acceleration in the three organs exhibited significant non-linear trends, in the liver biological aging acceleration assessment, we observed that continuous increases in LZ intake did not lead to a substantial reduction in the model beta values. This suggests that total LZ intake can significantly reduce the risk of biological aging acceleration within a certain range, while excessive intake may not effectively further delay biological aging acceleration.

Next, we constructed multivariate Cox regression models to assess the relationship between different levels of total LZ intake and all-cause mortality (Figure 5). The results of all three models indicated a significant trend effect for total LZ intake on all-cause mortality (P < 0.001). Compared to the Q1 group, the risk of all-cause mortality in the Q4 group was reduced by 0.55 (0.42, 0.72) times, 0.60 (0.45, 0.80) times, and 0.60 (0.45, 0.80) times in the three models, respectively. These results suggest that the level of LZ intake has a significant impact on all-cause mortality risk.

To assess the stability of the results, we performed stratified analyses by different covariates to examine the impact of gender, age, and BMI on the effectiveness of the models (Supplementary Figure S2). For overall biological aging acceleration, gender did not significantly affect the relationship between total LZ intake and biological aging acceleration across the three models (all P for interaction > 0.05). However, gender had a significant interaction effect with total LZ intake and all-cause mortality only in Model 1 (P for interaction < 0.001). In contrast, different age groups showed significant interaction effects across all models assessing biological aging acceleration (P for interaction, Model 1: 0.021; Model 2: 0.016; Model 3: 0.024). The results showed that in the older age group (≥60 years), LZ intake had a more significant impact on biological aging acceleration (P, Model 1: 0.002; Model 2: 0.009; Model 3: 0.024), and similar results were observed in the evaluation of the association between different levels of LZ intake and all-cause mortality risk (Age group ≥60 years: all P < 0.001). We also noted that in the younger age group (Age group 20–39 years), the relationship between total LZ intake and biological aging acceleration as well as all-cause mortality was not significant. Furthermore, in the evaluation of interaction effects by different BMI categories, no significant effects were found in Model 3 for biological aging acceleration and all-cause mortality.

Additionally, in the evaluation of the effects of different covariates on organ-specific biological aging acceleration (Supplementary Figure S3), we found significant interaction effects of age group on the association between total LZ intake and cardiovascular and kidney biological aging acceleration (P for interaction < 0.001). Similarly, in the older age group, total LZ intake had a more significant effect on reducing biological aging acceleration. This result suggests that the benefits of LZ intake are more pronounced in older populations with faster aging progression.

To further explore the intrinsic mechanisms by which lutein as main carotenoid mediate the aging process, we collected data from a microarray dataset of genes differentially expressed after 8 weeks of high lutein intake. After performing biological process (BP) enrichment clustering analysis (Figure 6), we found that lutein mainly modulates pathways involved in glycogen metabolic processes, cell division, chaperone-mediated protein folding, regulation of DNA metabolic processes, regulation of transferase activity, chromatin remodeling, and vascular processes in the circulatory system. Further examination of the clustering results (Supplementary Figure S4) revealed that Cluster 5 (regulation of DNA metabolic process) was primarily associated with telomere regulation, suggesting that lutein may help combat aging acceleration by maintaining telomerase activity. GSEA also indicated that lutein intake could upregulate anti-aging related biological processes (Supplementary Figure S5). Furthermore, we observed that aging resistance may also be maintained through certain post-translational modification processes (Supplementary Figure S4).

The acceleration of the aging process can lead to changes in certain metabolism-related pathways. Therefore, we further investigated the metabolic pathway changes following sustained lutein intake. The results showed (Figure 7) that certain anti-aging metabolic processes were significantly upregulated, such as Folate One Carbon Metabolism, Valine, Leucine, and Isoleucine Degradation, Nicotinamide Adenine Dinucleotide Biosynthesis, Sphingolipid Metabolism, and Primary Bile Acid Biosynthesis. Moreover, the upregulation of fatty acid degradation-related pathways further supports the lipid-lowering effect of lutein (18).

Studies have shown that immune dysregulation can also accelerate the aging process. We continued to explore the impact of sustained lutein intake on immune cell abundance. The results indicated (Supplementary Figure S6) that after 8 weeks of continuous lutein intake, the overall immune score increased compared to the control group. Notably, we observed a downregulation of Th1 cell abundance, as Th1 cells are primarily involved in the body's inflammatory response. This suggests that lutein may, to some extent, delay the aging process through inflammation resistance. The downregulation of the interleukin-mediated inflammatory response pathways further supports this finding (Supplementary Figure S5).

Related studies have shown that the aging process is often accompanied by various changes in phenotypes, including SASP, cell cycle, apoptosis resistance, alterations in cell surface marker expression, and the involvement of certain non-coding RNAs. To investigate this further, we collected various phenotypic markers related to the aging process (22–26) (Supplementary Table S4). The results indicated that after 8 weeks of continuous lutein intake, the expression of most markers in different phenotypes was downregulated (Supplementary Figure S7). The changes in various aging-related phenotypes further confirm the anti-aging efficacy of lutein.