Biological Age Acceleration, Genetic Susceptibility, and Incident Glaucoma Risk

Abstract

Purpose: To evaluate the association of biological age acceleration with incident glaucoma risk and examine whether genetic predisposition modifies it.

Methods: We included 318,556 UK Biobank participants without baseline glaucoma. Biological age was calculated using the Klemera–Doubal method Biological Age (KDM-BA) and PhenoAge algorithms. Hazard ratios (HRs) and 95% confidence intervals (CIs) of the association between biological age acceleration and incident glaucoma, and their interaction with genetic risk were analyzed by Cox regression models. Mendelian randomization analyses investigated causal associations.

Results: After a median follow-up of 13.5 years, 6553 participants developed glaucoma. Biological age acceleration was associated with an increased glaucoma risk. Each 5-year increment in biological age acceleration was linked to higher glaucoma risk (KDM-BA acceleration: HR, 1.12, 95% CI, 1.07–1.16; PhenoAge acceleration, HR, 1.09, 95% CI, 1.06–1.13). Biologically older participants had a higher glaucoma risk than younger participants (KDM-BA acceleration, HR, 1.10, 95% CI, 1.05–1.16; PhenoAge acceleration, HR, 1.07, 95% CI, 1.02–1.13). Genetic risk modified these relationships (all P for interactions < 0.05). Biologically older participants with high genetic risk had the highest glaucoma risk (KDM-BA acceleration, HR, 2.33, 95% CI, 2.15–2.52; PhenoAge acceleration, HR, 2.21, 95% CI, 2.05–2.38). No causal relationships were found in the Mendelian randomization analysis.

Conclusions: Biological age acceleration was associated with an increased glaucoma risk, and this relationship was modified by genetic risk. However, no causal relationship was established, and further research is needed to investigate the nature of the association.

Glaucoma, a chronic neurodegenerative eye disease, is the most prevalent cause of permanent blindness globally.¹ Over the past 30 years, its prevalence has risen by 10.7%, especially among older adults.² With an aging global population and increasing life expectancy, the burden of visual impairment owing to glaucoma has become a substantial social concern.³ 

Advanced age is a prominent risk factor for glaucoma, encompassing various molecular and physiological changes that lead to a more fragile vascular system, connective tissue, and retinal ganglion cells.⁴ However, among individuals of the same chronological age, considerable heterogeneity in physiological functions and aging processes may result in different disease vulnerabilities.⁵ Biological age has been identified as a more accurate measure of aging rates. Various metrics have been suggested to assess biological age, such as leukocyte telomere length, epigenetic clocks, proteomic biomarkers, and clinical biomarkers.⁶ Clinical biomarkers-based biological age is cost effective while ensuring measurement aging accuracy and is linked to numerous age-related conditions.^7–9 However, there is limited evidence regarding the association between biological age acceleration determined by clinical biomarkers and incident glaucoma. Additionally, glaucoma results from a combination of pathways and may be impacted by genetic and environmental elements.¹⁰ 

Therefore, the current study aims to evaluate the associations of biological age acceleration with glaucoma and IOP. We then explore the potential interactions between age acceleration and genetic factors on glaucoma development. We further use Mendelian randomization (MR) analysis to infer causal relationships between biological age and glaucoma. 

Methods

Study Population

The UK Biobank is a community-based prospective cohort study recruiting more than 500,000 middle-aged residents from England, Scotland, and Wales. Participants completed demographic and lifestyle questionnaires and provided biological samples at baseline.¹¹ After excluding participants who were lost to follow-up, had missing data on necessary clinical biomarkers for calculating biological age acceleration, or had baseline glaucoma, the remaining 318,556 participants were included in the main analyses. Additionally, 2898 participants lacking complete genetic data were excluded, leaving 315,658 participants for genetic analyses (Supplementary Fig. S1). To explore the association between biological age acceleration and IOP, we included 73,786 participants with complete ophthalmic IOP data in cross-sectional analyses. The UK Biobank received ethical approval from the North West Multicenter Research Ethics Committee. All participants provided written informed consent. 

Assessment of Biological Age Acceleration

Klemera–Doubal method Biological Age (KDM-BA) and the PhenoAge algorithm were used to construct biological age from the clinical biomarkers, which have been validated in UK populations and shown good performance in predicting age-related health outcomes.^12,13 Briefly, forced expiratory volume in 1 second, systolic blood pressure, and seven blood biomarkers were used to compute KDM-BA; PhenoAge was computed from nine blood biomarkers. More details on included biomarkers and biological age acceleration can be found in the Supplementary Methods. We classified individuals as biologically younger if their biological age acceleration was less than or equal to 0 and as biologically older if it was greater than 0. Moreover, we further divided participants into four groups based on the biological age acceleration quartile. 

Assessment of Glaucoma and IOP

Baseline glaucoma was defined as having a data-linkage glaucoma diagnosis, a self-reported glaucoma diagnosis or receiving glaucoma therapy. We recorded incident glaucoma during follow-up by data linkage to the hospital using International Classification of Diseases, 9th and 10th editions, codes for glaucoma (Supplementary Table S1), and previous studies have provided details of the reliability and validity of data on the above registry.^14,15 Specifically, POAG, secondary glaucoma, or glaucoma suspect cases were not included in our analysis. Follow-up time was recorded from the date of enrollment until the date of death, date of withdrawal from the study, or date of the end of the follow-up, whichever came first. 

The IOP value was measured by corneal-compensated IOP at baseline, which integrates the applanation tensions inward and outward. The IOP value was taken as the mean of both eyes unless data from only one eye were provided. We excluded participants who had undergone eye surgery or eye infection within the previous month, as well as those suffering from glaucoma or receiving glaucoma therapy, owing to the ambiguity of the untreated IOP values.¹⁶ 

Polygenic Risk Scores (PRS) for Glaucoma

Genetic data on UK Biobank participants were generated using the Affymetrix UK BiLEVE Axiom Array and the UK Biobank Axiom Array genotyping platforms. Quality control and imputation methods for these platforms have been described previously.¹⁷ We constructed a PRS for glaucoma using 2673 independent single nucleotide polymorphisms (SNPs) identified in a multi-trait analysis of genome-wide association study (MTAG), which provides a more comprehensive reflection of its genetic basis.¹⁸ Based on the MTAG-derived PRS, we applied the effect sizes from the original MTAG study using a weighted sum of the SNPs, then categorized participants into high genetic risk (above the mean PRS) and low genetic risk (below the mean PRS) to ensure a more balanced and interpretable analysis. 

Assessment of Covariates

Covariates were selected based on previous studies,^14,19 including demographic data (chronological age, sex, ethnicity, and education), Townsend deprivation index, body mass index, smoking status, drinking status, healthy diet, physical activity, wearing glasses, diabetes, and hypertension. A healthy diet was defined as adherence to at least five elements of the dietary recommendations according to a previous study.²⁰ Physical activity was measured by subjective reporting and was represented by the metabolic equivalent task minutes per week.²¹ Wearing glasses was defined as wearing glasses or contact lenses. Diabetes and hypertension were defined by collecting from self-reported diagnoses or receiving drug treatment. The specific definitions of the covariates can be found in the Supplementary Methods and Supplementary Table S1. 

Statistical Analysis

We used multiple imputations by chained equations to impute missing covariate values.²² Detailed missing covariate information is presented in Supplementary Table S2. We examined Kaplan–Meier curves and scaled Schoenfeld residuals on functions of time when regarding age acceleration as a categorical or continuous variable, and found it satisfies assumptions of the proportional hazards. For continuous biological age acceleration, effect estimates were derived for each 5-year increment. Cox regression was used to evaluate the impact of biological age acceleration on incident glaucoma and hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated. General linear regression examined associations between biological age acceleration and IOP. Restricted cubic spline assessed the dose–response relationship between age acceleration and glaucoma risk or IOP. Two models were constructed: model 1 adjusted for chronological age and sex, and model 2 further included adjustment for ethnicity, education, Townsend deprivation index, body mass index, smoking status, drinking status, healthy diet, physical activity, wearing glasses, diabetes, and hypertension. In combined association analyses, biological age acceleration and genetic risk were treated as binary variables. Using low genetic risk and biologically younger as the reference, we investigated their combined effects on glaucoma and their multiplicative and additive interactions. 

We performed subgroup analyses stratified by several potential risk factors to evaluate whether they may modify the association of biological age acceleration with glaucoma, including sex, chronological age, obesity status, current smoking, current drinking, healthy diet, and high physical activity. Obesity was defined as body mass index of 30 kg/m² or greater. High physical activity was defined as achieving 600 metabolic equivalent task minutes per week or more. Four sensitivity analyses were conducted. First, we excluded those without complete covariates data. Second, we excluded those who developed glaucoma during 2 years. Third, we included those who had a complete medication history and further adjusted use of steroid drugs.²³ Fourth, we further adjusted for height and salt intake, which may modulate the risk of glaucoma.^24,25 All statistical analyses were performed using R version 4.2.2 with a two-sided P value of less than 0.05 statistically significant. 

Mendelian Randomization

Two-sample MR was used to investigate causal associations between biological age and glaucoma. The GWAS summary data for glaucoma were obtained from the FinnGen database.²⁶ KDM-BA data were obtained from the UK Biobank GWAS,²⁷ and PhenoAge SNP data from a GWAS meta-analysis.²⁸ Causal associations were estimated using the random-effects inverse-variance weighted method. Further details on the MR design are available in the Supplementary Methods. 

Results

Baseline Characteristics of Participants

The baseline characteristics of the 318,556 participants are presented in Table 1. 6553 participants developed glaucoma during a median follow-up of 13.5 years (interquartile range, 12.7–14.1 years). In general, the mean chronological age of participants was 56.34 ± 8.10 years, and 172,114 (54.0%) were female. Supplementary Figure S2 displays the distributions of biological age and chronological age for participants. KDM-BA and PhenoAge were strongly correlated with chronological age (Supplementary Fig. S3). Participants who developed glaucoma during follow-up had higher baseline biological ages than those who did not (KDM-BA, 60.50 years vs 55.32 years; PhenoAge, 55.53 years vs 50.14 years). 

Table 1.

 

Baseline Characteristics of Participants by Incident Glaucoma During Follow-up

 

Associations of Biological Age Accelerations and Risk of Glaucoma

As shown in Table 2, biological age acceleration was associated with an increased glaucoma risk. For KDM-BA acceleration, each 5-year increment was linked to a 12% increased risk of glaucoma (HR, 1.12; 95% CI, 1.07–1.16). Biologically older participants had a 10% increased glaucoma risk (HR, 1.10; 95% CI, 1.05–1.16) than biologically younger participants. The highest quartile of KDM-BA acceleration had an HR of 1.16 (95% CI, 1.08–1.26) compared with the lowest quartile. For PhenoAge acceleration, each 5-year increment increased the risk by 9% (HR, 1.09; 95% CI, 1.06-1.13). Biologically older participants had a 7% increased glaucoma risk (HR, 1.07; 95% CI, 1.02–1.13) than biologically younger participants. The HR in the highest quartile of PhenoAge acceleration was 1.19 (95% CI, 1.10–1.27). RCS curves showed a J-shaped relationship between biological age acceleration (both KDM-BA and PhenoAge) and glaucoma risk, with risk rising when biological age acceleration exceeded zero (Fig. 1). 

Table 2.

 

Association of Biological Age Accelerations With Incident Glaucoma

 

Figure 1.

 

Association between biological age acceleration and incident glaucoma. (A) Klemera–Doubal method Biological Age (KDM-BA) acceleration. (B) PhenoAge acceleration. Restricted cubic spline regression model adjusted for chronological age, sex, ethnicity, education, Townsend deprivation index, body mass index, smoking status, drinking status, healthy diet, physical activity, wearing glasses, diabetes, and hypertension. CI, confidence interval; HR, hazard ratio.

Associations of Age Accelerations and IOP

Based on baseline data on biological age acceleration and IOP with adjustment for potential confounding factors, per 5-year increment in KDM-BA acceleration was associated with an 0.02-mm Hg increase in IOP (β = 0.02). This correlation persisted as statistically significant when categorizing KDM-BA acceleration into biologically younger and biologically older (β = 0.04). Additionally, a 1.00-unit increase in KDM-BA acceleration corresponded to a 0.05-unit increase in IOP for participants in the highest quartile compared with those in the lowest quartile (β = 0.05) (Supplementary Table S3). Similarly, a J-shaped relationship between biological age acceleration and IOP was observed in Supplementary Figure S4 (P-nonlinear = 0.023). Conversely, we only observed a weak association between each 5-year increment in PhenoAge acceleration and IOP (β = -0.012), and no statistically significant nonlinear correlation was presented between PhenoAge acceleration and IOP (P-nonlinear = 0.131). 

Joint Effects and Interactions of Biological Age Acceleration and Genetic Risk

High genetic risk was associated with a 2.03-fold increased risk of glaucoma (95% CI, 1.92–2.13). A risk gradient was observed across the genetic risk quartiles, with the highest quartile showing a 2.89-fold increased glaucoma risk compared with the lowest quartile (95% CI, 2.68–3.12) (Supplementary Figure S5). As shown in Figure 2, the highest glaucoma risk was observed among biologically older and high genetic risk. Specifically, when combining KDM-BA acceleration and genetic risk, biologically older and high genetic risk had a 2.33-fold increased glaucoma risk (95% CI, 2.15–2.52) compared with biologically younger and low genetic risk. Similarly, when combining PhenoAge acceleration and genetic risk resulted in a 2.21-fold increased glaucoma risk (95% CI, 2.05–2.38). 

Figure 2.

 

The joint association of biological age acceleration and genetic risk in relation to risk of glaucoma. (A) Klemera–Doubal method Biological Age (KDM-BA) acceleration. (B) PhenoAge acceleration. The results were adjusted for chronological age, sex, ethnicity, education, Townsend deprivation index, body mass index, smoking status, drinking status, healthy diet, physical activity, wearing glasses, diabetes, and hypertension. CI, confidence interval; HR, hazard ratio.

Notably, statistically significant interactions were observed between biological age acceleration and genetic risk for glaucoma on the multiplicative scale (P for interaction < 0.05). In addition, additive effects were observed between the KDM-BA and genetic risk (Supplementary Table S4). The associations between biological age acceleration and glaucoma risk were stronger among those with lower genetic risk. For KDM-BA acceleration, biologically older was linked to a 17% increased risk (HR, 1.17; 95% CI, 1.06–1.28) among low genetic risk, whereas the increase was only 7% (HR, 1.07; 95% CI, 1.00–1.14) among those at high genetic risk. Similarly, for PhenoAge acceleration, biologically older had a 13% increased glaucoma risk among low genetic risk (HR, 1.13; 95% CI, 1.03–1.23), whereas no statistically significant association was found among those with high genetic risk (HR, 1.04; 95% CI, 0.98–1.11) (Supplementary Fig. S6). 

Subgroup and Sensitivity Analyses

The results of stratified analyses based on several demographic and behavioral factors are shown in Figure 3. For KDM-BA acceleration, the association between biologically older and glaucoma risk was stronger among those with low physical activity (P for interaction = 0.006). For PhenoAge acceleration, the association between biologically older and glaucoma risk was stronger among males (P for interaction = 0.005) and participants with obesity (P for interaction = 0.007). No other significant interactions were found (P for interaction > 0.05). We performed sensitivity analyses in examining the robustness of the association between biological age acceleration and glaucoma. No material differences were found when excluding those without complete covariates data or developed glaucoma during 2 years. When we included those with complete data on drug use records and adjusted for use of steroids, the results were similar. Adjusting for height and salt intake did not change the findings substantially (Supplementary Table S5). 

Figure 3.

 

Association of biological age acceleration with the risk of glaucoma stratified by potential risk factors. The results were adjusted for chronological age, sex, ethnicity, education, Townsend deprivation index, body mass index, smoking status, drinking status, healthy diet, physical activity, wearing glasses, diabetes, and hypertension. CI, confidence interval; HR, hazard ratio; KDM-BA, Klemera–Doubal method Biological Age.

Causal Association of Biological Age With Glaucoma

The results of the random-effects inverse-variance weighted method did not support the causal effect of biological age on glaucoma. No statistically significant causal effect of KDM-BA (odds ratio, 1.01; 95% CI, 0.96–1.07) and PhenoAge (odds ratio, 1.01; 95% CI, 0.99–1.02) was found on glaucoma. Similarly, no significant associations were found using any other MR methods (Supplementary Table S6). The MR–Egger intercept test revealed no evidence of horizontal pleiotropy. A significant degree of heterogeneity was observed in KDM-BA to glaucoma, which was assessed using a random-effects model. No significant heterogeneity was found among the PhenoAge-related SNPs and glaucoma (Supplementary Table S7). Funnel plots visualizing overall heterogeneity are presented in Supplementary Figure S7. The leave-one-out plot and scatter plot are shown in Supplementary Figures S8 and S9. 

Discussion

In this prospective study, we found that biological age acceleration was associated with an increased glaucoma risk, and this relationship was modified by genetic risk. The greatest glaucoma risk was observed in those with high genetic risk and were biologically older. Moreover, this association was pronounced in males, those with obesity, and those who engaged in low physical activity. 

Considering that individuals of comparable chronological ages have diverse physiological conditions owing to variations in the biological aging rate, biological aging algorithms can capture different dimensions of aging. Recent studies have indicated that biological age acceleration is associated with several health outcomes, such as dementia⁹ and cardiovascular diseases.²⁹ We are the first to report J-shaped associations between biological age acceleration and glaucoma, indicating that as biological age surpasses chronological age, the glaucoma risk increases dramatically. Our study identified novel associations between biological age (including KDM-BA and PhenoAge) acceleration and glaucoma, expanding the clinical significance of biological aging. 

In this study, the effects of KDM-BA acceleration on glaucoma were stronger among males. In agreement with our findings, a previous study revealed that males aged between 45 and 75 years exhibited a greater glaucoma burden than females, with the difference increasing with age.³⁰ This finding could suggest that males are more likely to develop glaucoma as they age. The gender differences in glaucoma risk may be attributed to estrogen, which protects retinal ganglion cells by enhancing proliferation, preserving the retinal nerve fiber layer, decreasing IOP, and improving ocular blood flow.³¹ Its deficiency accelerates optic nerve aging and glaucomatous damage.³² Although declining estrogen after menopause may have negative impacts on females, lifetime cumulative estrogen exposure could provide long-lasting ocular benefits. In contrast, males tend to be biologically older than females of the same chronological age,³³ leading to earlier deterioration in their vascular system, connective tissue, and retinal ganglion cells, which increases their glaucoma risk. We also reported that the associations between PhenoAge acceleration and glaucoma were more pronounced among participants with obesity and those with low activity. In line with a previous study, obesity and low physical activity were related to increased glaucoma.³⁴ Obesity is an important contributor to the acceleration of neurodegenerative processes and plays a crucial role in the development of glaucoma.³⁵ In contrast, physical activity has neuroprotective effects³⁶ and is also attributed to a protective role in decelerating the aging process.³⁷ 

Our study also identified that participants with elevated KDM-BA acceleration displayed increased IOP at baseline while elevated PhenoAge acceleration did not. The discrepancies between the two biological age measures and IOP may be due to their different biomarkers and algorithms, which quantify the aging process differently. Specifically, KDM-BA tends to reflect the cardiometabolic aspects of the aging process, while PhenoAge captures age-related immune responses.²⁷ IOP is influenced by risk factors in metabolic disorders¹⁶; however, the association between immune cell components and IOP remains poorly understood.³⁸ Additionally, PhenoAge tends to estimate a younger biological age than chronological age in UK Biobank participants,³⁹ because it reflects mortality risk and they have lower all-cause mortality rates than the general population.⁴⁰ In contrast, KDM-BA correlates more closely with chronological age.⁴¹ The nonmonotonic relationship between IOP and chronological age was observed in an ophthalmic epidemiology study in Europe, suggesting that IOP fluctuates with age, which may explain the differing associations with these two biological age measures.⁴² 

Our observational data found a potential nonlinear association between biological aging and glaucoma; however, the MR analyses provided limited evidence supporting causal associations. Previous MR analysis implicated four epigenetic clocks that represent DNA methylation-based biological age did not show consistent causal results in the relationship with glaucoma,⁴³ which ignores the existence of nonlinear causality. Standard MR is based on the assumption of linearity, but nonlinear causality cannot be excluded.⁴⁴ Additionally, low heritability estimates for the instrumental variables used for biological age phenotypes may underpowered to detect these effects in MR analysis.²⁸ There is therefore a need to investigate the relationship between biological aging and glaucoma risk thoroughly by additional larger GWAS studies. 

There were several mechanisms responsible for the increased glaucoma risk attributable to biological aging. In the process of aging, trabecular meshwork cells and retinal ganglion cells are susceptible to oxidative stress, mitochondrial dysfunction, and DNA damage, which develop and remain in a cell-senescent state.⁴ Senescent vascular endothelial cells further lead to impaired angiogenesis and reduce blood flow to the optic nerve head, causing damage to retinal ganglion cells.⁴⁵ Accumulated senescent cells are primarily responsible for causing degenerative changes and leading to a loss of retinal ganglion cells and vision. Senescent cells also cause structural changes, such as trabecular thickening and fusion, resulting in a reduced aqueous humor outflow and increasing IOP.⁴ These combined effects significantly contribute to glaucoma pathologies. 

Our finding that genetic factors were crucial in glaucoma development is consistent with previous studies.^18,46 This study is the first to examine the association between biological age acceleration and glaucoma risk, to the best of our knowledge. We found an interaction between biological age acceleration and genetic risk presented on glaucoma risk. The adverse effect of being biologically older was more pronounced among participants with low genetic risk. One possible explanation is that the cumulative effects of high genetic risk on retinal ganglion cells damage over time may overshadow the impact of biological age acceleration, making individuals with lower genetic risk more vulnerable. Biological age acceleration also did not increase dementia (a neurodegenerative disease similar to glaucoma) risk in those with high genetic risk in a previous study.⁴⁷ Nevertheless, biologically older participants with high genetic risk still had the highest risk of incident glaucoma compared with biologically younger participants with low genetic risk. This finding emphasized the potential of decelerating biological age acceleration as a preventive strategy for reducing glaucoma prevention. 

The strengths of this study include its large sample size and extended follow-up, which ensured reliable results and evaluated the impact of biological age acceleration on glaucoma over a long time. However, the present study has several limitations. First, biological age was assessed at baseline, limiting the evaluation of dynamic change of biological age on the glaucoma risk. Second, the specific glaucoma subtypes could not be confirmed owing to the limitations of available data. Identifying glaucoma subtypes with distinct pathological mechanisms may provide clearer insights into their different associations with biological aging. Third, UK Biobank participants generally have better health and greater health awareness than the general population,⁴⁰ potentially leading to a younger biological age and lower glaucoma prevalence, which may introduce healthy volunteer bias. Fourth, conducting a more detailed analysis of nonlinear causal relationships between biological age and glaucoma was limited because only summary-level statistics were accessible, which warrants further investigation. Fifth, considering the facts that residual confounding by unknown or unmeasured factors may persist and associations do not mean causality, the results should be interpreted with caution, despite the fact that we accounted for several possible confounding factors. Finally, the results of our study are predominantly derived from European populations, which may not represent other ethnicities fully. 

Conclusions

This study found that biological age acceleration was associated with increased glaucoma risk, and this relationship is modified by genetic risk. However, no causal relationship was established. Given the global population aging and the burden of glaucoma, further research is needed to confirm these findings. 

Acknowledgments

The authors thank the participants and staff in the UK Biobank and the FinnGen study for their contribution. This research has been conducted using the UK Biobank resource under application number 98679. 

Supported by the Construction of High-level University of Guangdong (G624330242), Guangdong Province Universities and Colleges Pearl River Scholar Funded Scheme (2019). 

Disclosure: W.-Q. Song, None; W.-F. Zhong, None; Z.-H. Li, None; D. Liu, None; J.-J. Ren, None; D. Shen, None; J. Gao, None; P.-L. Chen, None; J. Yang, None; X.-M. Wang, None; F.-F. You, None; C. Li, None; H. Chen, None; J.-H. Xie, None; C. Mao, None