Metabolic vulnerability, genetic susceptibility, and incident age-related eye diseases: a prospective cohort study

Abstract

Background:

Metabolic dysregulation is increasingly recognized as a systemic process contributing to chronic disease development, yet prospective evidence linking integrated metabolic vulnerability to age-related eye diseases remains limited. We investigated whether a biomarker-based metabolic vulnerability index (MVX) was associated with incident age-related ocular diseases and whether joint consideration of MVX and genetic susceptibility may help characterize relative risk patterns.

Methods:

A prospective population-based cohort of 206,311 participants from the UK Biobank was analyzed. MVX was evaluated as the primary exposure. Incident age-related macular degeneration (AMD), cataract, diabetic retinopathy (DR), and glaucoma were ascertained as outcomes. Associations were examined using Cox proportional hazards models, with hazard ratios (HRs) and 95% confidence intervals (CIs) estimated per 1–standard deviation (SD) increase in MVX. As secondary exploratory analyses, polygenic risk score (PRS) analyses were performed to explore whether metabolic vulnerability and genetic susceptibility jointly characterized relative risk patterns.

Results:

During follow-up, 4,144 participants developed AMD, 13,574 cataract, 1,483 DR, and 5,525 glaucoma. After multivariable adjustment for demographic, socioeconomic, clinical, and lifestyle factors, each 1-SD increase in MVX was associated with higher risks of incident AMD (HR = 1.07; 95% CI: 1.03–1.11), cataract (HR = 1.04; 95% CI: 1.02–1.06), and DR (HR = 1.11; 95% CI: 1.05–1.18), whereas no significant association was observed for glaucoma (HR = 1.00; 95% CI: 0.97–1.03). In joint analyses, individuals with both high genetic risk and elevated MVX exhibited the greatest risks of AMD (HR = 2.32; 95% CI: 2.01–2.67), cataract (HR = 1.62; 95% CI: 1.49–1.76), and DR (HR = 3.84; 95% CI: 2.91–5.06), compared with those with low genetic risk and low MVX.

Conclusion:

These findings suggest that MVX may be relevant to population-level patterns of risk for several age-related eye diseases. However, further studies are needed to determine whether MVX provides meaningful predictive value or clinical utility beyond conventional risk factors.

Introduction

Age-related eye diseases, including age-related macular degeneration (AMD), cataract, diabetic retinopathy (DR), and glaucoma, are among the leading causes of visual impairment and blindness worldwide (1–3). With population aging and the increasing prevalence of metabolic disorders, the burden of these conditions is projected to rise substantially, resulting in profound consequences for vision-related quality of life and public health systems (4, 5). Although multiple clinical and demographic risk factors have been identified, current approaches to risk prediction and early identification of individuals at highest risk for developing age-related ocular diseases remain limited (6–8). In particular, most existing studies have focused on isolated metabolic traits or single biomarkers, which may fail to capture the cumulative and systemic nature of metabolic vulnerability across the life course. Consequently, there is a need for integrative markers that more comprehensively characterize metabolic risk profiles and improve understanding of population-level risk patterns for age-related eye disease.

Metabolic dysregulation has been increasingly recognized as an important contributor to the development of major ocular diseases in mid to late adulthood (9). Epidemiological evidence has linked obesity, insulin resistance, hypertension, dyslipidemia, and related lifestyle factors to elevated risks of AMD, cataract, and DR, and has suggested potential contributions to glaucomatous optic neuropathy through systemic vascular and metabolic pathways (9–11). From a biological perspective, these metabolic disturbances may affect ocular tissues through shared mechanisms, including chronic low-grade inflammation, oxidative stress, and microvascular dysfunction (12, 13). Notably, metabolic risk factors tend to cluster and progress over time rather than occur in isolation, indicating that composite measures capturing overall metabolic vulnerability may more accurately reflect real-world risk than individual biomarkers alone (14, 15).

The metabolic vulnerability index (MVX) is an integrated, biomarker-based measure designed to capture systemic metabolic vulnerability across multiple biological pathways (16). This composite approach may more accurately reflect cumulative metabolic burden under real-world conditions than individual biomarkers considered in isolation. Previous studies have shown that MVX is associated with the occurrence of diseases across multiple organ systems, including cardiovascular, neurological, and respiratory conditions, and has demonstrated utility in predicting and stratifying mortality risk in community-based populations (17–20). However, prospective evidence linking MVX to the development of age-related eye diseases remains limited, and it is unclear whether MVX provides additional population-level information beyond genetic susceptibility.

Genetic susceptibility represents an important background determinant of age-related eye diseases (21, 22). Whereas Polygenic risk scores (PRS) summarize relatively stable inherited predisposition, MVX reflects systemic metabolic vulnerability that is more dynamic and potentially modifiable (16, 23). We therefore considered it informative, as a secondary objective, to explore whether these two dimensions might jointly characterize relative risk patterns for age-related eye diseases at the population level.

Accordingly, we conducted a prospective cohort study in the UK Biobank to examine whether MVX was associated with the incidence of AMD, cataract, DR, and glaucoma. Our primary objective was to evaluate the prospective association between MVX and these outcomes using Cox proportional hazards models, with MVX modeled primarily as a continuous exposure per 1–standard deviation (SD) increase. As secondary analyses, we further characterized these associations using alternative exposure parameterizations and restricted cubic splines, and explored whether metabolic vulnerability and genetic susceptibility jointly characterized relative risk patterns.

Methods

Study population

The UK Biobank is a large prospective cohort study that recruited over 500,000 participants aged 40–69 years across the United Kingdom between 2006 and 2010. Baseline assessments comprised detailed questionnaires, physical measurements, and the collection of biological samples, with ongoing follow-up through linkage to health-related records (24). Of the full cohort (n = 501,939), participants were excluded if they had: (1) missing biomarker data required for MVX calculation (n = 228,069); (2) missing covariate data (n = 64,260); (3) prevalent age-related eye diseases at baseline (n = 2,141); or (4) missing genetic data (n = 1,158). After applying these exclusion criteria, 206,311 participants were included in the primary analyses (Figure 1). The UK Biobank study received ethical approval from the North West Multicentre Research Ethics Committee.

Figure 1

Assessment of metabolic vulnerability

MVX was calculated according to a previously published, sex-specific framework based on six circulating biomarkers quantified by nuclear magnetic resonance (NMR) spectroscopy, including glycoprotein acetyls (GlycA), small HDL particle (sHDL), leucine, valine, isoleucine, and citrate. Briefly, the inflammation vulnerability index (IVX) was derived from GlycA, sHDL, and their interaction term, whereas the metabolic malnutrition index (MMX) was derived from leucine, valine, isoleucine, and citrate, including quadratic terms where applicable. MVX was subsequently calculated as a function of IVX, ln (MMX), and their interaction using sex-specific coefficients derived from the original published framework. Prior to score construction, each biomarker was winsorized at the 1st and 99th percentiles to reduce the influence of extreme values, and the resulting IVX, MMX, and MVX scores were normalized to a range of 1–100. Full sex-specific equations, biomarker definitions, and score-construction procedures are provided in the Supplementary Methods. For the primary analyses, MVX was analyzed as a continuous variable and standardized per 1–SD increase. IVX and MMX were evaluated in secondary analyses using continuous and categorical parameterizations. For categorical analyses, each index was further classified into low and high groups according to the median value (low group as the reference) and into quartiles (Q1–Q4, with Q1 as the reference).

Construction of PRS

To quantify individual-level genetic susceptibility to age-related eye diseases, disease-specific polygenic risk scores (PRS) for AMD, cataract, DR, and glaucoma were constructed using two approaches: (1) a curated weighted PRS based on established risk variants and (2) a genome-wide PRS generated using PRS-CS (continuous shrinkage). For the weighted PRS, lead single-nucleotide polymorphisms (SNPs) associated with AMD, cataract, DR, and glaucoma were extracted from previously published studies curated in the PGS Catalog, comprising 19, 647, 39, and 243 SNPs, respectively (25–27) (Tables S1–S4). SNPs were aligned to the reported effect allele, and each variant was weighted by the corresponding log-odds ratio. The weighted PRS was calculated as the sum of effect-allele dosages multiplied by their respective weights across all included SNPs. Genome-wide PRS were constructed using PRS-CS, which applies Bayesian regression with continuous shrinkage priors. This approach incorporates linkage disequilibrium (LD) information from an external European reference panel derived from the 1,000 Genomes Project to estimate posterior SNP effect sizes without reliance on arbitrary P-value thresholds (28). Genome-wide PRS were calculated by summing SNP dosages weighted by PRS-CS posterior effect estimates based on external genome-wide association study (GWAS) summary statistics for each outcome (29).

For interaction analyses, genetic risk based on each PRS was categorized into low (quintile 1), intermediate (quintiles 2–4), and high (quintile 5) risk groups, consistent with prior literature (30). All PRS values were standardized prior to inclusion in regression models.

Assessment of age-related eye diseases

Four common eye diseases with substantial age-related burden were investigated: AMD, cataract, DR, and glaucoma. The primary outcomes were incident age-related eye diseases. Incident cases were ascertained through linked hospital inpatient records from the UK Biobank Hospital Episode Statistics (HES) database using International Classification of Diseases codes (ICD-10 and ICD-9) (Table S5). For each outcome, incident disease was defined as the first recorded diagnosis occurring after the baseline assessment date. To restrict analyses to incident events, participants with a recorded diagnosis of the corresponding eye disease on or before baseline were excluded. For each outcome, the event date was defined as the date of the first qualifying diagnosis recorded in the HES data.

Covariates

Covariates were selected a priori based on previous literature and their potential associations with both MVX and age-related eye diseases. Baseline variables included chronological age (years), sex (male or female), ethnicity (white, mixed, Asian, black, or others), education level (degree-level or professional qualification vs. other levels), body mass index (BMI, kg/m²), Townsend Deprivation Index, physical activity (high, moderate, or low), sleep duration (long ≥8 h, moderate 7–8 h, or short ≤ 6 h), healthy diet (yes or no), smoking status (yes or no), and alcohol consumption (yes or no). Physical activity was categorized as high (≥3,000 MET-min/week), moderate (600–3,000 MET-min/week), or low (< 600 MET-min/week) according to total weekly metabolic equivalent task minutes (31). Definitions of smoking status, alcohol consumption, and healthy diet were based on previously published criteria; detailed variable definitions are provided in the Supplementary material (32, 33) (Table S5).

Statistical analyses

Baseline characteristics were summarized as means ± SD for continuous variables and counts (%) for categorical variables. The primary analyses evaluated the associations between continuous MVX and incident age-related eye diseases using Cox proportional hazards models, with HRs and 95% CIs estimated per 1-SD increase in MVX. As secondary analyses, MVX was additionally examined using categorical parameterizations, including median-based groups (low vs. high, with low as the reference) and quartiles (Q1–Q4, with Q1 as the reference), to assess the robustness of the findings across alternative exposure specifications. IVX and MMX were also evaluated in secondary analyses using continuous and categorical parameterizations. Restricted cubic spline functions were applied as secondary analyses to characterize potential non-linearity in the associations between MVX and each outcome, using four knots placed at the 5th, 35th, 65th, and 95th percentiles of MVX, with the median MVX value serving as the reference.

Two multivariable models were fitted. Model 1 was adjusted for age, sex, ethnicity, education level, BMI, and the Townsend Deprivation Index. Model 2 was further adjusted for physical activity, sleep duration, smoking status, alcohol consumption, and healthy diet. The proportional hazards assumption was assessed using Schoenfeld residuals.

Genetic susceptibility was evaluated using standardized PRS. Associations between PRS and each outcome were assessed using Cox models adjusted for the covariates described above. As secondary exploratory analyses, we further explored whether metabolic vulnerability and genetic susceptibility jointly characterized relative risk patterns through joint classification of MVX (low vs. high) and PRS categories (low: quintile 1; intermediate: quintiles 2–4; high: quintile 5). Additive interaction between MVX and PRS was quantified using the relative excess risk due to interaction (RERI) and the attributable proportion due to interaction (AP), with 95% CIs estimated using 1,000 bootstrap resamples (36). Subgroup analyses were also conducted as exploratory analyses, stratified by sex, age group, BMI category, smoking status, and physical activity, to explore potential heterogeneity of associations.

All statistical tests were two-sided, and P < 0.05 was considered statistically significant. All analyses were conducted within the UK Biobank Research Analysis Platform.

Results

Participants

Baseline characteristics of the included participants are presented in Table 1. The mean chronological age at baseline was 56.19 ± 8.10 years, and 99,676 participants (48.31%) were male. The majority of participants were of White ethnicity (95.4%), and 35.1% reported a degree-level or professional qualification. The mean BMI was 27.29 ± 4.67 kg/m², and the mean Townsend Deprivation Index was −1.49 ± 3.00. During follow-up, incident cases included 4,144 AMD (2.0%), 13,574 cataract (6.6%), 1,483 DR (0.7%), and 5,525 glaucoma (2.7%). Distributions of MVX, IVX, and MMX stratified by age-related eye disease status are shown in Supplementary Figures S1–S3.

MVX and incident age-related eye diseases risk

In the primary analyses, higher continuous MVX was associated with increased risks of incident AMD, cataract, and DR. Per 1-SD increase in MVX, the HRs were 1.07 (95% CI: 1.03–1.11) for AMD, 1.05 (95% CI: 1.03–1.07) for cataract, and 1.12 (95% CI: 1.06–1.19) for DR after adjustment for age, sex, ethnicity, education level, BMI, and the Townsend Deprivation Index. No association was observed between MVX and glaucoma (HR = 1.00; 95% CI: 0.97–1.03). These associations were materially unchanged after further adjustment for physical activity, sleep duration, smoking status, alcohol consumption, and healthy diet.

As secondary analyses, restricted cubic spline analyses and categorical parameterizations were used to further characterize the shape and robustness of the observed associations. The MVX–risk relationships for AMD and cataract showed broadly monotonic increases without evidence of non-linearity (P_nonlinear > 0.05), whereas a non-linear association was observed for DR (P_nonlinear = 0.008), with risk increasing more steeply at higher MVX levels. No significant association between MVX and glaucoma was observed in spline analyses (P_overall = 0.215) (Figure 2). When MVX was dichotomized at the median value (39.42), participants in the high MVX group had higher risks of AMD (HR = 1.10; 95% CI: 1.02–1.19) and cataract (HR = 1.07; 95% CI: 1.03–1.12) compared with those in the low MVX group, whereas no significant associations were observed for DR or glaucoma (Figure 3). In quartile analyses, associations generally strengthened across increasing MVX categories. Compared with Q1, participants in Q4 had higher risks of AMD (HR = 1.20; 95% CI: 1.08–1.33), cataract (HR = 1.12; 95% CI: 1.06–1.19), and DR (HR = 1.22; 95% CI: 1.03–1.44) (Table 2).

Figure 2

Figure 3

Subgroup analyses

As secondary exploratory analyses, subgroup analyses were conducted to evaluate whether the associations between MVX and incident age-related eye diseases differed across participant subgroups (Figure 4). For AMD and cataract, the associations were generally consistent across subgroups, with no evidence of effect modification by sex, age, BMI, physical activity, sleep duration, healthy diet, smoking status, or alcohol consumption (all P for interaction > 0.05). For DR, significant effect modification was observed by sex (P for interaction < 0.001), age group (P for interaction = 0.011), and BMI category (P for interaction < 0.001). The association between MVX and incident DR was stronger among women than men, stronger among participants aged < 60 years than those aged ≥60 years, and stronger among participants with BMI ≥30 kg/m² than those with BMI < 30 kg/m². For glaucoma, no statistically significant overall association with MVX was observed in the main analyses; however, heterogeneity across subgroups was suggested for physical activity (P for interaction = 0.038) and healthy diet (P for interaction = 0.022).

Figure 4

PRS and incident age-related eye diseases risk

As secondary exploratory analyses, we further examined whether genetic susceptibility, quantified using PRS, was associated with incident age-related eye diseases. Higher PRS were generally associated with increased risks of incident age-related eye diseases after adjustment for all covariates (Table 3 and Tables S6–S9). Using PRS-CS, each 1-SD increase in PRS was associated with higher risks of incident AMD (HR = 1.36; 95% CI: 1.32–1.40), DR (HR = 1.55; 95% CI: 1.47–1.63), and glaucoma (HR = 1.37; 95% CI: 1.34–1.41), whereas no significant association was observed for cataract (HR = 1.01; 95% CI: 0.99–1.02). When categorized into genetic risk groups, participants in the highest PRS category exhibited the greatest risks of AMD (HR = 2.16; 95% CI: 1.96–2.38), DR (HR = 3.51; 95% CI: 2.90–4.25), and glaucoma (HR = 2.36; 95% CI: 2.16–2.58), compared with those in the lowest risk group (Figure S4). For the weighted PRS, each 1-SD increase was associated with higher risks of incident AMD (HR = 1.36; 95% CI: 1.32–1.40), cataract (HR = 1.16; 95% CI: 1.14–1.18), DR (HR = 1.17; 95% CI: 1.12–1.23), and glaucoma (HR = 1.27; 95% CI: 1.24–1.31) (Figure S5). Consistently, participants in the highest weighted PRS category had the greatest risks of age-related eye diseases compared with those in the lowest category.

To support the subsequent exploratory joint analyses, the predictive performance of PRS-CS and the weighted PRS was compared using Harrell's C-index. PRS-CS demonstrated slightly better discrimination for AMD (0.7570 vs. 0.7567), DR (0.8000 vs. 0.7850), and glaucoma (0.7128 vs. 0.7056), whereas the weighted PRS showed superior performance for cataract (0.7391 vs. 0.7355). Accordingly, PRS-CS was selected as the primary PRS for AMD, DR, and glaucoma, and the weighted PRS was selected for cataract in the subsequent joint analyses.

Joint effects and interactions of MVX and genetic risk

As secondary exploratory analyses, we further examined whether MVX and genetic susceptibility jointly characterized relative risk patterns for age-related eye diseases. Using the primary PRS selected based on C-index performance (PRS-CS for AMD, DR, and glaucoma; weighted PRS for cataract), joint analyses indicated that genetic susceptibility was strongly associated with incident age-related eye diseases, and participants with concomitantly high genetic risk and high MVX generally exhibited the greatest risks (Table 4). Compared with the low genetic risk/low MVX group, the high genetic risk/high MVX group had the highest risks of incident AMD (HR = 2.32; 95% CI: 2.01–2.67), cataract (HR = 1.62; 95% CI: 1.49–1.76), and DR (HR = 3.84; 95% CI: 2.91–5.06). For glaucoma, genetic risk remained strongly associated with disease incidence, whereas MVX did not meaningfully further increase risk within genetic strata. Specifically, among participants with high genetic risk, the HRs for glaucoma were comparable between those with high MVX (HR = 2.28; 95% CI: 2.00–2.60) and those with low MVX (HR = 2.48; 95% CI: 2.19–2.80).

Additive interaction analyses provided evidence of synergism between MVX and genetic risk for AMD only. Among participants with high PRS, high MVX was associated with a positive additive interaction on the risk of incident AMD, as indicated by a relative excess risk due to interaction (RERI) of 0.42 (95% CI: 0.15–0.70) and an attributable proportion (AP) of 0.18 (95% CI: 0.07–0.30). No evidence of additive interaction was observed for cataract, DR, or glaucoma (Table S10).

Discussion

In this prospective cohort study, the primary finding was that higher MVX was associated with increased risks of incident AMD, cataract, and DR, but not glaucoma, after multivariable adjustment. Secondary analyses using alternative exposure parameterizations and restricted cubic splines were broadly consistent with these findings and further suggested a non-linear association between MVX and DR, with risk increasing more prominently at higher MVX levels. Additional exploratory analyses further suggested that metabolic vulnerability and genetic susceptibility may jointly characterize relative risk patterns for age-related eye diseases, particularly for AMD. Taken together, these findings support the relevance of MVX as a biomarker of metabolic vulnerability associated with several age-related eye diseases at the population level, although further work is required to determine whether it offers meaningful predictive or clinical utility beyond conventional risk factors.

Previous studies have linked adverse cardiometabolic profiles, including obesity, dysglycaemia or diabetes, hypertension, and dyslipidaemia, to increased risks of AMD, cataract, and DR, although the strength and consistency of these associations have varied across populations and study designs (9, 34–37). However, most prior work has focused on individual metabolic traits or composite clinical definitions, and prospective evidence capturing the joint burden of biomarker-based metabolic vulnerability across multiple ocular endpoints remains limited. The present findings indicate that an integrated biomarker-based index of metabolic vulnerability, as reflected by MVX, is prospectively associated with incident AMD, cataract, and DR, whereas no association was observed for glaucoma. The observed non-linear pattern for DR suggests that the relationship between metabolic vulnerability and DR risk may differ across the MVX distribution, with a more pronounced increase in risk at higher MVX levels than would be expected under a linear assumption. In addition, a growing body of PRS research has demonstrated genetic risk stratification for age-related eye diseases (38–40). By comparing two PRS construction strategies and carrying forward the better-performing score for each outcome, the present study suggests that metabolic vulnerability and genetic susceptibility may provide complementary population-level information, although the incremental predictive value of such integration remains to be established.

The observed associations between higher MVX and incident AMD, cataract, and DR are biologically plausible, as MVX captures integrated metabolic vulnerability across multiple pathways that converge on ocular aging (17). Metabolic dysregulation is closely linked to chronic low-grade inflammation, oxidative stress, endothelial dysfunction, and microvascular impairment, which may influence the choroid, retina, and other ocular tissues through impaired perfusion, altered lipid handling, and cumulative cellular injury (41, 42). These processes have been implicated in retinal pigment epithelium dysfunction and drusen-related changes in AMD, lens protein oxidation and glycation in cataract, and capillary damage with breakdown of the blood–retinal barrier in DR (43–45). The non-linear association observed for DR may reflect a threshold-like pattern, whereby metabolic vulnerability exerts a disproportionate impact once systemic metabolic burden exceeds a higher range, consistent with progressive microvascular injury and limited compensatory capacity (46). In contrast, the absence of a significant overall association with glaucoma may suggest that systemic metabolic vulnerability plays a relatively smaller role than eye-specific factors, such as intraocular pressure, anterior segment anatomy, and optic nerve susceptibility. In addition, any metabolic contribution to glaucoma risk may be heterogeneous across disease subtypes or mediated indirectly through vascular comorbidity rather than acting as a dominant determinant (47). Nevertheless, these proposed pathways remain speculative, and future experimental and mechanistic studies are warranted to clarify causal mechanisms and potential heterogeneity across disease subtypes.

The subgroup findings for DR should be interpreted as exploratory and hypothesis-generating rather than as definitive evidence of effect modification. Within this exploratory framework, the association between higher MVX and incident DR appeared stronger among women, individuals younger than 60 years, and those with obesity. One possible explanation is that sex-related differences in metabolic and inflammatory phenotypes, as well as microvascular susceptibility, may modify how systemic metabolic vulnerability is translated into retinal microvascular injury, potentially contributing to a stronger association in women (48, 49). The age-specific pattern may suggest that elevated MVX in midlife captures earlier or more aggressive metabolic dysregulation that precedes clinically apparent retinopathy, whereas among older adults the relative contribution of MVX may be attenuated by longer disease duration, treatment effects, competing risks, and survival-related selection. Similarly, obesity may amplify this association, as excess adiposity is closely linked to insulin resistance, chronic low-grade inflammation, and endothelial dysfunction, all of which may contribute to retinal microvascular damage (50, 51). However, these interpretations remain tentative and require confirmation in future studies.

Genetic susceptibility represents an important and relatively stable component of risk for age-related eye diseases, and the PRS-based analyses in the present study were intended as secondary exploratory extensions of the main MVX question rather than as the primary focus of our study. In this context, the joint analyses suggest that metabolic vulnerability and genetic susceptibility may provide complementary information at the population level, with individuals exhibiting both high MVX and high genetic risk generally showing the highest relative risks for AMD, cataract, and DR (30, 52). The observed additive interaction for AMD further raises the possibility that systemic metabolic vulnerability and inherited susceptibility may act jointly in shaping disease risk on an absolute scale. However, these findings should be interpreted cautiously. The present study was not designed as a formal prediction study, and we did not evaluate whether adding MVX, PRS, or their combination materially improves discrimination, calibration, reclassification, or clinical decision-making beyond conventional risk factors. Accordingly, these PRS-based and joint analyses should be viewed as exploratory and hypothesis-generating, rather than as evidence of immediate clinical utility for screening or prevention. Additive interaction was evaluated using RERI and AP, which are informative for public health interpretation because they assess whether the excess risk associated with combined exposures exceeds the sum of the excess risks associated with each exposure alone (53).

This study has several strengths. First, analyses were conducted in a large, well-characterized prospective cohort with long-term follow-up, enabling robust estimation of incident risks across multiple age-related eye disease endpoints. Second, MVX was derived from an integrated panel of circulating biomarkers and evaluated using complementary exposure parameterizations, including continuous measures, median-based categories, quartiles, and restricted cubic splines. This approach enhances interpretability and allows assessment of potential non-linearity in associations. Third, genetic susceptibility was assessed using two PRS construction strategies, which allowed a more structured exploratory evaluation of whether metabolic vulnerability and inherited susceptibility may jointly characterize relative risk patterns. Fourth, comprehensive covariate adjustment and exploratory subgroup analyses were applied, and additive interaction was quantified using RERI and AP with bootstrap-derived CIs. Together, these approaches provide an interpretable evaluation of association patterns and joint risk relationships.

Several limitations should be acknowledged. First, selection bias should be considered. Of the 501,939 participants in the full UK Biobank cohort, only 206,311 were included in the present analyses after exclusions, primarily because of missing biomarker data required for MVX calculation and missing covariate information. This substantial restriction may have resulted in an analytic sample that differed systematically from the overall UK Biobank population, thereby limiting generalizability and potentially introducing selection bias. In addition, UK Biobank participants are not fully representative of the general population. Therefore, the findings should be interpreted with caution and validated in other cohorts. Second, outcome ascertainment relied on linked hospital inpatient records from the UK Biobank Hospital Episode Statistics database. As a result, milder or earlier-stage cases managed in outpatient or community settings may not have been captured, which is particularly relevant for conditions such as cataract and early AMD. This may have led to underascertainment of less severe disease, introduced outcome misclassification, and potentially attenuated the observed associations. Third, MVX was assessed at baseline and does not account for longitudinal changes in metabolic vulnerability or the influence of subsequent treatment, which could result in exposure misclassification during follow-up. Fourth, although extensive covariates were included, residual confounding cannot be entirely excluded, and the subgroup and interaction analyses were exploratory and involved multiple comparisons; these findings should therefore be interpreted cautiously. Fifth, the present study was not designed as a formal prediction study. We did not evaluate whether MVX, alone or in combination with PRS, materially improves discrimination, calibration, reclassification, or clinical decision-making beyond conventional risk factors. Therefore, the findings should not be interpreted as establishing immediate clinical utility for diagnosis, screening, or prevention. Sixth, the performance and transferability of PRS may be reduced in populations outside predominantly European ancestry, underscoring the need for external validation in more diverse cohorts. Future studies should seek to replicate these findings in independent populations, evaluate time-varying or repeated measures of metabolic vulnerability, incorporate more detailed ocular phenotypes and clinical measurements, and determine whether MVX, alone or in combination with genetic susceptibility, provides meaningful predictive value beyond conventional risk factors.

Conclusions

In this prospective UK Biobank cohort, higher metabolic vulnerability, as assessed by MVX, was associated with increased risks of incident AMD, cataract, and DR, but not glaucoma. Secondary exploratory analyses further suggested that metabolic vulnerability and genetic susceptibility may jointly characterize relative risk patterns for some age-related eye diseases, particularly AMD. Together, these findings support the relevance of MVX as a biomarker of metabolic vulnerability associated with several age-related eye diseases at the population level. However, further studies are required to determine whether MVX, alone or in combination with genetic susceptibility, provides meaningful predictive value or clinical utility beyond conventional risk factors.

References

• 1.

  GBD 2019 Blindness and Vision Impairment Collaborators; Vision Loss Expert Group of the Global Burden of Disease Study. Causes of blindness and vision impairment in 2020 and trends over 30 years, and prevalence of avoidable blindness in relation to VISION 2020: the right to sight: an analysis for the global burden of disease study. Lancet Glob Health. (2021) 9:e144-e160. doi: 10.1016/S2214-109X(21)00050-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2.

  WeinbergJGaurMSwaroopATaylorA. Proteostasis in aging-associated ocular disease. Mol Aspects Med. (2022) 88:101157. doi: 10.1016/j.mam.2022.101157

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3.

  Vision Loss Expert Group of the Global Burden of Disease Study; GBD 2019 Blindness and Vision Impairment Collaborators. Global estimates on the number of people blind or visually impaired by age-related macular degeneration: a meta-analysis from 2000 to 2020. Eye. (2024) 38:2070–2082. doi: 10.1038/s41433-024-03050-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4.

  DasSAhmadZSuryawanshiAKumarA. Innate immunity dysregulation in aging eye and therapeutic interventions. Ageing Res Rev. (2022) 82:101768. doi: 10.1016/j.arr.2022.101768

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5.

  HansmanDSDuJCassonRJPeetDJ. Eye on the horizon: the metabolic landscape of the RPE in aging and disease. Prog Retin Eye Res. (2025) 104:101306. doi: 10.1016/j.preteyeres.2024.101306

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6.

  ShenRYZhangYChenLJCheungCYLiangYThamCCet al. Ocular and systemic risk factors and biomarkers for primary glaucoma: an umbrella review of systematic reviews with meta-analyses. Invest Ophthalmol Vis Sci. (2025) 66:35. doi: 10.1167/iovs.66.12.35

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7.

  JiangCMellesRBSanganiPHoffmannTJHysiPGGlymourMMet al. Association of behavioral and clinical risk factors with cataract: a two-sample mendelian randomization study. Invest Ophthalmol Vis Sci. (2023) 64:19. doi: 10.1167/iovs.64.10.19

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 8.

  CuiYCuiJXueCCMaoYJonasJBWangYXet al. Five-year incidence of age-related macular degeneration and its risk factors in adult chinese population: the tongren health care study. Transl Vis Sci Technol. (2024) 13:10. doi: 10.1167/tvst.13.12.10

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9.

  SongWChenWChiJLiuXZhuW. The role of lipid metabolism disorder in the progression and treatment of ocular vascular diseases. Surv Ophthalmol. (2026) 71:1–13. doi: 10.1016/j.survophthal.2025.09.002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10.

  RauscherFGElzeTFranckeMMartinez-Perez ME LiYWirknerKet al. Glucose tolerance and insulin resistance/sensitivity associate with retinal layer characteristics: the LIFE-Adult-Study. Diabetologia. (2024) 67:928–39. doi: 10.1007/s00125-024-06093-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11.

  KleinBEKleinRLeeKEMeuerSM. Socioeconomic and lifestyle factors and the 10-year incidence of age-related cataracts. Am J Ophthalmol. (2003) 136:506–12. doi: 10.1016/S0002-9394(03)00290-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12.

  LongHXiongYLiuHYangMLiuTGongCet al. IL-6 Exacerbates oxidative damage of RPE cells by indirectly destabilizing the mRNA of DNA repair genes. Inflammation. (2025) 48:2323–40. doi: 10.1007/s10753-024-02192-2

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13.

  Rejas-GonzálezRMontero-CalleAPastora SalvadorNCrespo CarballésMJAusín-GonzálezESánchez-NavesJet al. Unraveling the nexus of oxidative stress, ocular diseases, and small extracellular vesicles to identify novel glaucoma biomarkers through in-depth proteomics. Redox Biol. (2024) 77:103368. doi: 10.1016/j.redox.2024.103368

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14.

  AlbertiKGEckelRHGrundySMZimmetPZCleemanJIDonatoKAet al. Harmonizing the metabolic syndrome: a joint interim statement of the international diabetes federation task force on epidemiology and prevention; national heart, lung, and blood institute; american heart association; world heart federation; international atherosclerosis society; and international association for the study of obesity. Circulation. (2009) 120:1640–5. doi: 10.1161/CIRCULATIONAHA.109.192644

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15.

  HakalaJOPahkalaKJuonalaMSaloPKähönenMHutri-KähönenNet al. Cardiovascular risk factor trajectories since childhood and cognitive performance in midlife: the cardiovascular risk in young finns study. Circulation. (2021) 143:1949–61. doi: 10.1161/CIRCULATIONAHA.120.052358

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16.

  OtvosJDShalaurovaIMayHTMuhlesteinJBWilkinsJTMcGarrahRWet al. Multimarkers of metabolic malnutrition and inflammation and their association with mortality risk in cardiac catheterisation patients: a prospective, longitudinal, observational, cohort study. Lancet Healthy Longev. (2023) 4:e72–82. doi: 10.1016/S2666-7568(23)00001-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17.

  LiJManQWangYCuiMLiJXuKet al. The metabolic vulnerability index as a novel tool for mortality risk stratification in a large-scale population-based cohort. Redox Biol. (2025) 81:103585. doi: 10.1016/j.redox.2025.103585

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18.

  WicksTRWolskaAGhazalDJakimovskiDWeinstock-GuttmanBBurnhamAet al. Frailty exacerbates disability in progressive multiple sclerosis. Ann Clin Transl Neurol. (2025) 13:654–64. doi: 10.1002/acn3.70243

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19.

  JinCHuaPWangCWangBZhangY. Association between multimarkers of metabolic malnutrition and inflammation and risk of chronic obstructive pulmonary disease and lung function: a prospective cohort study of UK biobank. Nutr Metab. (2025) 22:143. doi: 10.1186/s12986-025-01042-8

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20.

  ConnersKMShearerJJJooJParkHManemannSMRemaleyATet al. The metabolic vulnerability index: a novel marker for mortality prediction in heart failure. JACC Heart Fail. (2024) 12:290–300. doi: 10.1016/j.jchf.2023.06.013

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21.

  SiggsOMHanXQassimASouzeauEKuruvillaSMarshallHNet al. Association of monogenic and polygenic risk with the prevalence of open-angle glaucoma. JAMA Ophthalmol. (2021) 139:1023–8. doi: 10.1001/jamaophthalmol.2021.2440

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22.

  KwongAZawistowskiMFritscheLGZhanXBragg-GreshamJBranhamKEet al. Whole genome sequencing of 4,787 individuals identifies gene-based rare variants in age-related macular degeneration. Hum Mol Genet. (2024) 33:374–85. doi: 10.1093/hmg/ddad189

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23.

  WandHLambertSATamburroCIacoccaMAO'SullivanJWSillariCet al. Improving reporting standards for polygenic scores in risk prediction studies. Nature. (2021) 591:211–9. doi: 10.1038/s41586-021-03243-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24.

  SudlowCGallacherJAllenNBeralVBurtonPDaneshJet al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. (2015) 12:e1001779. doi: 10.1371/journal.pmed.1001779

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 25.

  PrivéFAschardHCarmiSFolkersenLHoggartCO'ReillyPFet al. Portability of 245 polygenic scores when derived from the UK Biobank and applied to 9 ancestry groups from the same cohort. Am J Hum Genet. (2022) 109:12–23. doi: 10.1016/j.ajhg.2021.11.008

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26.

  CraigJEHanXQassimAHassallMCooke BaileyJNKinzyTGet al. Multitrait analysis of glaucoma identifies new risk loci and enables polygenic prediction of disease susceptibility and progression. Nat Genet. (2020) 52:160–6. doi: 10.1038/s41588-019-0556-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 27.

  TanigawaYQianJVenkataramanGJustesen JM LiRTibshiraniRHastieTet al. Significant sparse polygenic risk scores across 813 traits in UK Biobank. PLoS Genet. (2022) 18:e1010105. doi: 10.1371/journal.pgen.1010105

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28.

  GeTChen CY NiYFengYASmollerJW. Polygenic prediction via Bayesian regression and continuous shrinkage priors. Nat Commun. (2019) 10:1776. doi: 10.1038/s41467-019-09718-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29.

  VermaAHuffmanJERodriguezAConeryMLiuMHoYLet al. Diversity and scale: genetic architecture of 2068 traits in the VA Million Veteran Program. Science. (2024) 385:eadj1182. doi: 10.1126/science.adj1182

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30.

  JinGLvJYangMWangMZhuMWangTet al. Genetic risk, incident gastric cancer, and healthy lifestyle: a meta-analysis of genome-wide association studies and prospective cohort study. Lancet Oncol. (2020) 21:1378–86. doi: 10.1016/S1470-2045(20)30460-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31.

  GuHChenRFangTXuJZhangYBianCet al. Associations of physical activity with the risks of osteoarthritis and subtypes: a population-based cohort study of UK Biobank data. Bone Joint Res. (2025) 14:656–65. doi: 10.1302/2046-3758.147.BJR-2024-0529.R1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32.

  SaidMAVerweijNvan der HarstP. Associations of combined genetic and lifestyle risks with incident cardiovascular disease and diabetes in the UK biobank study. JAMA Cardiol. (2018) 3:693–702. doi: 10.1001/jamacardio.2018.1717

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 33.

  ZhangYBChenCPanXFGuoJLiYFrancoOHet al. Associations of healthy lifestyle and socioeconomic status with mortality and incident cardiovascular disease: two prospective cohort studies. Bmj. (2021) 373:n604. doi: 10.1136/bmj.n604

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34.

  ZhangQYTieLJWuSSLvPLHuangHWWangWQet al. Overweight, obesity, and risk of age-related macular degeneration. Invest Ophthalmol Vis Sci. (2016) 57:1276–83. doi: 10.1167/iovs.15-18637

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35.

  LeskeMCWuSYHennisAConnellAMHymanLSchachatA. Diabetes, hypertension, and central obesity as cataract risk factors in a black population. the barbados eye study Ophthalmology. (1999) 106:35–41. doi: 10.1016/S0161-6420(99)90003-9

  • CrossRef
  • Google Scholar

• 36.

  DiraniMXieJFenwickEBenarousRReesGWongTYet al. Are obesity and anthropometry risk factors for diabetic retinopathy? The diabetes management project. Invest Ophthalmol Vis Sci. (2011) 52:4416–21. doi: 10.1167/iovs.11-7208

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37.

  HanXOngJSHewittAWGharahkhaniPMacGregorS. The effects of eight serum lipid biomarkers on age-related macular degeneration risk: a Mendelian randomization study. Int J Epidemiol. (2021) 50:325–36. doi: 10.1093/ije/dyaa178

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38.

  LeNQHeWMacGregorS. Polygenic risk scores and genetically complex eye disease. Annu Rev Vis Sci. (2024) 10:403–23. doi: 10.1146/annurev-vision-102122-103958

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39.

  HuangYCLiaoWLLinHJHuangYTChangYWYangJSet al. Prediction of risk, severity, and progression of primary open-angle glaucoma in a Taiwanese population based on polygenic risk scores. Am J Ophthalmol. (2026) 281:355–62. doi: 10.1016/j.ajo.2025.09.040

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40.

  WangSHHuangYCChengCWChangYWLiaoWL. Impact of the trans-ancestry polygenic risk score on type 2 diabetes risk, onset age and progression among population in Taiwan. Am J Physiol Endocrinol Metab. (2024) 326:E547–e554. doi: 10.1152/ajpendo.00252.2023

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41.

  ZandersLArifajDWagnerJUGDimmelerS. Cellular senescence, inflammaging and cardiovascular disease. Immunol Rev. (2026) 337:e70084. doi: 10.1111/imr.70084

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42.

  SinghIPoyntenAMKrishnanAVWillcoxMDPMarkoulliM. The ocular surface in type 2 diabetes: pathophysiology and impact of anti-diabetic drugs. Prog Retin Eye Res. (2026) 110:101417. doi: 10.1016/j.preteyeres.2025.101417

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 43.

  MauryaMBoraKBlomfieldAKPavlovichMCHuangSLiuCHet al. Oxidative stress in retinal pigment epithelium degeneration: from pathogenesis to therapeutic targets in dry age-related macular degeneration. Neural Regen Res. (2023) 18:2173–81. doi: 10.4103/1673-5374.369098

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 44.

  TruscottRJ. Age-related nuclear cataract-oxidation is the key. Exp Eye Res. (2005) 80:709–25. doi: 10.1016/j.exer.2004.12.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 45.

  TangLXuGTZhangJF. Inflammation in diabetic retinopathy: possible roles in pathogenesis and potential implications for therapy. Neural Regen Res. (2023) 18:976–82. doi: 10.4103/1673-5374.355743

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46.

  DuhEJSunJKStittAW. Diabetic retinopathy: current understanding, mechanisms, and treatment strategies. JCI Insight. (2017) 2:e93751. doi: 10.1172/jci.insight.93751

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47.

  WeinrebRNAungTMedeirosFA. The pathophysiology and treatment of glaucoma: a review. JAMA. (2014) 311:1901–11. doi: 10.1001/jama.2014.3192

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48.

  MarshMLOliveiraMNVieira-PotterVJ. Adipocyte metabolism and health after the menopause: the role of exercise. Nutrients. (2023) 15:444. doi: 10.3390/nu15020444

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49.

  CamonCGarrattMCorreaSM. Exploring the effects of estrogen deficiency and aging on organismal homeostasis during menopause. Nat Aging. (2024) 4:1731–44. doi: 10.1038/s43587-024-00767-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50.

  PackerMLamCSPButlerJZannadFVaduganathanMBorlaugBA. Is type 2 diabetes a modifiable risk factor for the evolution and progression of heart failure with a preserved ejection fraction?J Am Coll Cardiol. (2025) 86:1917–31. doi: 10.1016/j.jacc.2025.07.052

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51.

  de JonghRTSerné EH RGIJde VriesGStehouwerCD. Impaired microvascular function in obesity: implications for obesity-associated microangiopathy, hypertension, and insulin resistance. Circulation. (2004) 109:2529–35. doi: 10.1161/01.CIR.0000129772.26647.6F

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52.

  WangTDuanWJiaXHuangXLiuYMengFet al. Associations of combined phenotypic ageing and genetic risk with incidence of chronic respiratory diseases in the UK Biobank: a prospective cohort study. Eur Respir J. (2024) 63:2301720. doi: 10.1183/13993003.01720-2023

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53.

  AnderssonTAlfredssonLKällbergHZdravkovicSAhlbomA. Calculating measures of biological interaction. Eur J Epidemiol. (2005) 20:575–79. doi: 10.1007/s10654-005-7835-x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar