Skip to main content
NCBI home page
Epigenetics logoLink to Epigenetics
. 2025 Oct 17;20(1):2566496. doi: 10.1080/15592294.2025.2566496

Identification of blood-derived DNA methylation biomarkers of glaucoma and intraocular pressure measurements in three European ancestry cohorts including the Canadian longitudinal study on aging

Xiaofan Jiang a,b,c, Omar A Mahroo a,b,c, Anthony P Khawaja b,d, Dorota Skowronska-Krawczyk e,f, David A Mackey g,h, Janey L Wiggs i, Christopher J Hammond a,c, Pirro G Hysi a,c,j,k,l,✉,†, Christopher G Bell m,n,✉,†
PMCID: PMC12536625  PMID: 41105805

ABSTRACT

Glaucoma is a major cause of blindness globally and its prevalence rises with age. This study explored systemic blood-derived DNA methylation epigenetic biomarkers for association with glaucoma and intraocular pressure (IOP). Blood-derived DNA methylation (DNAm) was analyzed in 1,201 European participants from the Canadian Longitudinal Study on Aging (CLSA; Illumina EPIC v1 array) and 843 European participants from TwinsUK (450k array). An Epigenome-Wide Association Study (EWAS) for glaucoma and IOP was conducted, adjusting for age, sex, tobacco smoking, and leukocyte cell types. DNAm-based EpiScores estimates for 108 plasma protein levels were evaluated for associations with glaucoma and IOP. Additionally, ‘biological’ age acceleration, estimated using five established DNAm ‘clocks,’ was assessed for glaucoma and IOP and replicated in The Health and Retirement Study (HRS; n = 3,453). EWAS analyses of glaucoma and IOP in individual cohorts did not identify genome-wide significant associations. However, a combined EWAS for overlapping probes in both cohorts identified two epigenome-wide significant CpGs: cg03498697 in the FRMD3 promoter (p = 6.86x10−8) and cg06044751 intronically within PALLD (p = 1.76x10−7). EpiScore analysis revealed one IOP Bonferroni-significant association with TNFRSF1B levels in the meta-analysis of both cohorts (p = 1.31x10−4). DNAm ‘clock’ analysis in the HRS identified a GrimAge-positive age acceleration associated with glaucoma (p = 0.01). This study identified significant epigenetic blood-derived biomarkers that are associated with glaucoma and IOP. These findings warrant replication in larger and more diverse populations as well as via longitudinal analysis to assess their robustness and potential predictive power.

KEYWORDS: Glaucoma, intraocular pressure, Epigenetics, DNA methylation, biomarkers, CLSA

Introduction

The etiology and mechanisms underlying glaucoma, a leading cause of irreversible blindness, have been extensively investigated through genome-wide association studies (GWAS) [1,2]. Although this approach has identified numerous genetic loci linked to glaucoma and intraocular pressure (IOP), its main risk factor and predictor, the potential role of systemic mechanisms as well as surrogate tissue biomarkers of glaucoma have not been exhaustively explored [3,4].

Epigenetics, including DNA methylation (DNAm) and post-translational histone modifications, offers a promising avenue for exploring the underlying systemic mechanisms [5]. DNAm, the most common epigenetic modification of DNA, involves covalent addition of a methyl group to cytosine, primarily in differentiated cells at CpG dinucleotides [6]. It is an important regulator of gene expression and cell differentiation [7]. Owing to the cellular-specific epigenetic signatures, DNAm profiles can be used to estimate the composition of heterogeneous samples [8]. Furthermore, peripheral blood DNAm has gained significant evidence as a robust surrogate biomarker for many common diseases [9]. Although some epigenetic studies on glaucoma have been conducted [10,11], human population DNAm epigenome-wide association studies (EWAS) for this trait remain underexplored.

Robust and replicated estimators for circulating plasma protein levels, termed EpiScores, have also been constructed from blood DNAm profiles [12]. These measures are being evaluated as tools for disease risk stratification [13]. Interestingly, these quantitative estimates are potentially better at capturing the long-term trends. For example, the EpiScore for C-Reactive Protein (CRP) is a stronger biomarker for chronic inflammation than direct serum CRP measurement [14].

Additionally, aging is a major risk factor for multiple chronic diseases [15–17], including eye disorders like glaucoma, cataracts, and macular degeneration. The growing elderly population is increasing the burden of these diseases and straining healthcare resources [18]. Consequently, there is a pressing need for preventative strategies, predictive diagnostics, and deeper understanding of the molecular pathophysiology of aging. One of the major hallmarks of aging is alterations in the epigenome, including DNAm [19,20]. The advent of high-throughput DNAm arrays has enabled the construction of DNAm ‘clocks’ that accurately predict chronological age as well as capturing aspects of ‘biological’ age [21]. Furthermore, positively accelerated biological age, calculated via these DNAm clocks, has been positively correlated with age-related morbidity and mortality risk [22,23]. Ocular hypertension has been shown to increase stress responses and age, as measured by DNAm ‘clocks’ in experimental models of eye disease [24].

High-throughput DNAm array data enable a multifaceted interrogation for disease-related biomarkers [25]. In this study, we investigated blood-derived DNAm association with glaucoma and IOP through EWAS, EpiScore estimators, and DNAm ‘clock’ analysis in the Canadian Longitudinal Research on Aging (CLSA) study [26] and TwinsUK cohort [27] as well as The Health and Retirement Study (HRS) for DNAm ‘clock’ replication [28].

Methods

Study cohorts

CLSA: This study included 1,201 CLSA participants with available DNAm data and of European ancestry as verified through principal component ancestry clustering of their genotypes [29]. Participants underwent evaluations, including in-home interviews and visits to CLSA data collection sites. Data on demographic traits, medical history, and lifestyle variables were collected, and informed consent and fasting blood samples were obtained [26]. Phenotypes: This study focused on glaucoma and IOP measurement. IOP was measured at the CLSA data collection sites using a Reichert Ocular Response Analyzer (Reichert Technologies, Depew, NY, USA). The average of corneal compensated IOP measurements (IOPcc) of the right and left eyes was used to derive participant-level IOP values. If one eye had missing IOP data, the IOP value of the other eye was used. For IOP analysis alone, a quality control filter was applied, requiring measurements to be within the range of 5–40 mmHg, and values for individuals receiving eye drops and eye operations that could have influenced the measurement of IOP were removed. Separate readings were taken for each eye, and the mean IOP was calculated. Individuals with glaucoma were defined as those who responded positively to the question ‘Has a doctor ever told you that you have glaucoma?.’ No granular information about the type of glaucoma was available, but the primary open-angle glaucoma is by far the most common form of glaucoma in Canadian populations [30].

TwinsUK

This study included 843 subjects assessed for blood-derived DNAm from the TwinsUK cohort, which forms part of the UK Adult Twin Registry. The cohort consisted primarily of adult female twins, both monozygotic and dizygotic, ranging in age from 18 to 82 years [27]. The participants underwent comprehensive phenotypic assessments, including detailed questionnaires on lifestyle and medical history, clinical examinations, and biological sample collection. Peripheral blood samples were collected from each participant for DNA extraction and DNAm analysis. Informed consent was obtained from all participants. In the TwinsUK project, glaucoma status was self-reported based on follow-up questionnaires supplemented by information obtained through communication with participants’ local opticians [31]. IOPcc measurements were obtained from Ocular Response Analyzer (ORA, Reichert®, Buffalo, NY), a non-contact air-puff tonometer as described elsewhere [32]. Two readings were taken for each eye and the mean IOP was calculated. In subjects receiving IOP-lowering medication, the imputed IOP was calculated by increasing the measured value by 30%, as per efficacy data from standard therapeutics [33]. Exclusion criteria included any form of glaucoma surgery with the potential to alter IOP (e.g., trabeculectomy, laser surgery, and cataract surgery).

DNA methylation processing and quality control

In the CLSA cohort, genomic DNA was extracted from the whole blood samples of participants, and DNA methylation data were generated using the Illumina Infinium MethylationEPIC v1 BeadChip microarray [34]. These data were processed and quality-controlled as detailed in the CLSA Data Support Document (https://www.clsa-elcv.ca/wp-content/uploads/2023/06/clsa_datasupportdoc_epigenetics_v2.0_2022nov30_final.pdf) and are available for public access upon application through the CLSA online data access process. Briefly, the extracted DNA from peripheral blood mononuclear cells (PBMCs) was bisulfite (BS)-converted, and the arrays were processed and read via Illumina iScan to generate raw DNAm iDAT files. DNAm analysis was initially performed for 865,918 probes in each sample for a total of 1,479 individuals. Quality control (QC) processing and normalization were performed using R packages lumi [35] and WateRmelon [36]. Five samples were removed because of missing sex information and four samples because of low BS conversion. Multimapping and inconsistent EPIC v1 probes were removed [37]. Additionally, via PCA, sample beta-value correlation, and control probe outlier detection, 29 samples were flagged and removed from the subsequent analysis. Probes with poor detection P-values ( > 0.01 in > 10% samples) and bead counts ( < 3 beads per signal in ≥10% samples) were eliminated [38]. For the remaining 783,136 probes in 1,446 samples, missing beta values were imputed via the imputeknn algorithm and normalized with BMIQ [39]. Technical batch effects were corrected using the ComBat function [40]. Further to the CLSA QC, chromosome X and Y probes were removed, as well as additional poor multimapping and genetically confounded probes from McCartney et al. [41]. To reduce the influence of genetic confounding, only the majority of the European population samples were included in this study. Overall, 780,845 probes and 1,201 individuals were used for the downstream analyses.

DNA methylation was assessed in the TwinsUK cohort using the Illumina Infinium HumanMethylation450 BeadChip on DNA derived from whole blood samples. The methylation protocols and QC procedures were detailed previously, although the male samples were not excluded in this case [42]. This involved the detection of poor samples and probes, removal of chromosome X and Y probes, removal of known 450k multimapping and confounded probes from Chen et al [43]. and BMIQ for probe-type bias correction [39]. This resulted in a final set of 425,159 sites and 843 European individuals.

Epigenome-wide association study (EWAS)

We conducted an epigenome-wide association study (EWAS) using multivariable linear regression to estimate associations of glaucoma and IOP separately with DNA methylation at each CpG site, adjusting for potential confounders, including age, sex, smoking status, and blood cell proportions. To account for the relatedness of individuals in the TwinsUK cohort, including monozygotic and dizygotic twin pairs, we applied linear mixed-effects models using the ‘lme4’ package in R (version 4.3.3), with both family ID and zygosity included as random effects. This approach [44] accounts for within-pair correlation and adjusts for shared genetic and environmental influences, thereby reducing the risk of inflated Type I error due to non-independence of observations. Analyses were conducted using R package ‘meffil’ (R version 4.3.3) with the ‘All’ covariates model. The EPIC v1 epigenome-wide statistical significance threshold was set as p ≤ 9x10−8 [45]. For the TwinsUK 450k array analysis, the significance threshold was p ≤ 2.4x10−7 [46]. Estimated proportions of major blood cell types (CD8+ T cells, CD4+ T cells, natural killer [NK] cells, B cells, monocytes, and granulocytes) were calculated using the Houseman reference-based deconvolution algorithm [47], implemented via the estimateCellCounts2 function in the ‘minfi’ R package (R version 4.3.3). These cell type estimates were included as covariates in EWAS analyses to adjust for differences in cell composition between samples. In the combined analysis, 387,289 probes that were common across both the EPIC and 450k arrays and passed through QC for the relevant array were utilized. The meta-analysis EWAS of both cohorts was performed using METAL (http://www.sph.umich.edu/csg/abecasis/metal/) [48] with sample-size weighting, also referred to as a fixed-effects meta-analysis, as used previously [49]. This method ensured that the results from both the EPIC and 450k arrays were appropriately aggregated, accounting for differences in cohort size while maintaining robustness in detecting significant associations at the epigenetic level.

Epigenetic plasma protein estimates (EpiScores)

EpiScores for 109 plasma protein levels that are independent of known pQTL effects, as formulated by Gadd et al. [12] and implemented in the ‘MethylDetectR’ R-package, were calculated [50]. These EpiScores were tested for association with glaucoma and IOP, including covariates of age and sex. In addition, zygosity was included as a random effect for the TwinsUK dataset. Then, a meta-analysis was conducted using METAL [48] with effect-size weighting based on the inverse of the corresponding standard errors. A Bonferroni threshold for this analysis was calculated as a total of 218 tests with 109 EpiScores for each of the two phenotypes (glaucoma and IOP), that is, 0.05/(2 × 109) = 2.3x10−4.

DNA methylation clocks

Five DNAm clocks, Horvath [51], Hannum et al [38], PhenoAge [52], GrimAge [53], and DunedinPACE [54] were evaluated. Horvath’s DNAm Age Calculator (https://dnamage.genetics.ucla.edu/) was used to estimate Horvath and GrimAge measures. The Houseman algorithm [47] incorporated within the online calculator was used to calculate the predicted proportions of six blood cell types (B cells, CD4 T cells, CD8 T cells, natural killer cells, granulocytes, and monocytes). The methylCIPHER R package [55] (https://github.com/MorganLevineLab/methylCIPHER) was used to compute the results for additional clocks. The residuals generated by regressing the measure of epigenetic age over chronological age were computed for each epigenetic clock and then processed for further regression analysis. Normally distributed continuous variables are presented as the mean ± SD. We applied multivariate regression to investigate the association between epigenetic age and disease. The linear models were fitted using the R package ‘lme4’ (R version 4.3.3). All models possessing an epigenetic age measure included covariates for age, sex, smoking status, and leukocyte count. In addition, zygosity was included as a random effect in analyses of the TwinsUK dataset.

Replication cohort (HRS)

The Health and Retirement Study (HRS) from the University of Michigan is a longitudinal study in the United States supported by the National Institute on Aging and the Social Security Administration [28]. The HRS collects extensive data on demographics, health, financial status, family dynamics, and employment. Informed consent was obtained from all participants, and the study was approved by the ethics committee. DNAm from blood-derived samples was analyzed with the Illumina EPIC v1 array, and standard quality control was performed. The HRS cohort (v1.0, release Nov 2020) calculated 13 DNAm clocks, which includes three (Horvath, Hannum et al. and GrimAge) assessed in our study, and these were accessed as a replication cohort (https://hrsdata.isr.umich.edu/data-products/epigenetic-clocks). These data included 3,453 European subjects with predicted epigenetic age and glaucoma phenotypic data. Phenotypic data were obtained from HRS ‘Physical Health questionnaires based on participants’ affirmative responses to the question: ‘Has a doctor ever treated you for glaucoma?’ No distinction was made between the different glaucoma types.

Results

Demography

We obtained DNA methylation (DNAm) array data from individuals in two cohorts, CLSA and TwinsUK. Participants in the CLSA cohort ranged in age from 45 to 86 years, with a mean age of 63 years, whereas those in the TwinsUK cohort ranged from 18 to 82 years, with a mean age of 58 years (Figure 1(a,b), Table 1). After quality control (QC), the European subset of the CLSA cohort comprised 1,201 individuals with a final set of 780,845 methylation probes (via the EPIC v1 DNAm array). Among these individuals, 55 (4.6%) had glaucoma and 1,114 (92.8%) had valid intraocular pressure (IOP) data after data cleaning (Table 1). The TwinsUK cohort comprised of 843 European subjects with 426,112 methylation sites (via the 450k DNAm array). Within this group, 27 individuals self-reported doctor-diagnosed primary open-angle glaucoma and 459 had valid IOP data (Supplementary Table S1).

Figure 1.

Figure 1.

Open in a new tab

Bidirectional Manhattan plots for epigenome-wide association study (EWAS) between DNA methylation and glaucoma in CLSA, TwinsUK, and combined meta-analysis.

The genome-wide significance threshold is 9x10−8 for CLSA (EPIC v1), 2.4x10−7 for TwinsUK, and Combined (450k). The suggested significance threshold s 1x10−5 for all analyses. Y-axis: Positive direction indicates significant DNA hypermethylation, negative direction indicates significant DNA hypomethylation with trait.

Table 1.

Demographic information for each cohort.

  CLSA TwinsUK HRS
DNA methylation data (European-only) 1201 843 3453
Age range (years) 45-86 18-82 50-100
Mean (SD) Age (years) 60.2 (10.3) 58.0 (10.2) 73.3 (8.9)
Number of glaucoma cases 55 27 215
IOP available 1114 459 -
Open in a new tab

Cohort details for the CLSA (Canadian Longitudinal Study on Aging), TwinsUK, and HRS (Health and Retirement Study). The table includes the number of participants with available DNA methylation data (European ancestry only), age range, mean (SD) age, number of glaucoma cases, and availability of intraocular pressure (IOP) measurements. The sample sizes refer to individuals with DNA methylation data and not to the full cohort. IOP data are not available for HRS.

Epigenome-wide association (EWAS) for glaucoma and IOP

An EWAS for glaucoma and IOP was first performed in the CLSA cohort. This included the covariates of age, sex, smoking status, and leukocyte cell proportion (see Methods). The QQ plots for these EWAS analyses are shown in Supplementary Figures S1 and S2. No results reached the recommended EPIC v1 epigenome-wide significance threshold (p ≤ 9x10−8) [45]. However, 14 and 5 suggestively associated (p ≤ 1x10−5) [56] CpGs with glaucoma and IOP, respectively, were identified in the CLSA (see Figure 1(a) and Supplementary Figure S3A, Table 2 and 3).

Table 2.

EWAS results for glaucoma in CLSA, TwinsUK, and combined cohorts.

CpGs CLSA (n = 1195)
 TwinsUK (n = 485)
 Meta (n = 1657)
 Gene Region
beta P value beta P value Zscore P value N Direction
cg03498697 1.56E-2 5.94E-5 2.50E-2 1.85E-4 5.40 6.86E-8* 1680 + + FRMD3 Shore
cg06044751 −7.520E-3 1.32E-4 -1.27E-2 1.98E-4 −5.22 1.76E-7* 1680 - - PALLD Gene Body
cg19523063 −2.44E-2 5.05E-7* NA NA −5.02 5.05E-7* 1195 - NA NA
cg14346847 4.28E-3 5.46E-7* NA NA 5.01 5.46E-7* 1195 + ANKRD10 Gene Body
cg03892308 −4.27E-2 7.87E-6* -3.90E-2 3.03E-2 −4.93 8.12E-7* 1680 - - PCDHGA4 Shore
cg09618626 −1.07E-2 1.67E-6* NA NA −4.79 1.67E-6* 1195 - NA NA
cg01602252 7.06E-3 2.01E-6* NA NA 4.75 2.01E-6* 1195 + NA NA
cg11472333 −6.40E-3 5.34E-6* NA NA −4.55 5.34E-6* 1195 - NA NA
cg21227848 NA NA 2.90E-2 7.32E-6* 4.48 7.32E-6* 485 + DMR CpG Island
cg11371204 −2.34E-3 8.32E-6* NA NA −4.46 8.32E-6* 1195 - SMARCA4 Gene Body
cg03963555 −2.10E-2 9.03E-6* NA NA −4.44 9.03E-6* 1195 - PI16 5’UTR
cg10495606 −4.60E-3 1.47E-4 -6.14E-3 2.13E-2 −4.44 9.06E-6* 1680 - - PRKCE Gene Body
cg08395122 −2.13E-2 4.40E-6* -8.16E-3 4.39E-1 −4.29 1.80E-5 1680 - - PCDHGA4 CpG Island
cg11126621 3.31E-3 4.27E-6* 5.07E-4 5.20E-1 4.22 2.41E-5 1680 + + YPEL5 CpG Island
cg12304937 −1.54E-2 1.10E-6* 1.10E-3 8.74E-1 −4.03 5.71E-5 1680 + - PRKAR1B Shelf
cg18223359 2.17E-3 3.28E-6* -8.44E-5 9.15E-1 3.87 1.10E-4 1680 - + COIL CpG Island
cg20277260 2.13E-3 6.78E-6* -1.75E-4 8.46E-1 3.69 2.23E-4 1680 - + ARID3B CpG Island
cg19830245 7.19E-3 1.74E-6* -2.95E-3 8.31E-2 3.10 1.93E-3 1680 - + FLCN CpG Island
cg02674186 8.57E-4 8.75E-1 -5.32E-2 5.39E-6* −2.31 2.08E-2 1680 - + NA NA
cg24291500 −2.89E-3 3.47E-1 2.89E-2 4.67E-6* 1.67 9.54E-2 1680 + - REST Shore
Open in a new tab

CpG probes are listed for results that pass the suggestive p-value (*) threshold of < 1x10−5 in any of the individual analyses. The genome-wide significance level (**) was 9x10-8 for EPIC (CLSA) and 2.4x10−7 for 450k (TwinsUK and Combined). The combined Z-score and Direction columns indicate whether the DNAm directional change was consistent in the two studies.

Table 3.

EWAS results for iop in CLSA, TwinsUK, and combined cohorts.

CpGs CLSA (n = 1198)
 TwinsUK (n = 459)
 Meta (n = 1680)
 Gene Region
beta P value beta P value Zscore P value n Direction
cg05771094 NA NA −6.00E-4 2.43E-6* −4.71 2.43E-6* 459 - NA Shore
cg21430218 −2.70E-4 1.37E-4 −4.47E-4 7.24E-3 −4.66 3.22E-6* 1657 - - SYCP3 Shore
cg00463006 −4.18E-4 3.67E-6* NA NA −4.63 3.67E-6* 1198 - RHOBTB3 Gene Body
cg23715682 5.29E-5 6.44E-4 7.26E-5 1.40E-3 4.58 4.57E-6* 1657 + + CDKL1 Shore
cg18114363 −2.14E-4 5.14E-5 −1.22E-4 3.65E-2 −4.54 5.53E-6* 1657 - - NA NA
cg24211304 −6.01E-4 6.37E-5 −4.17E-4 3.06E-2 −4.54 5.69E-6* 1657 - - NA NA
cg10611744 5.12E-4 5.75E-6* NA NA 4.54 5.75E-6* 1198 + C16orf74 Gene Body
cg15032098 6.83E-4 8.81E-6* 6.83E-4 1.62E-1 4.52 6.31E-6* 1657 + + UCHL1 Shore
cg09915124 −5.56E-4 7.63E-6* NA NA −4.48 7.63E-6* 1198 - NA NA
cg07364694 −3.30E-4 8.16E-6* NA NA −4.46 8.16E-6* 1198 - NA NA
cg21361702 1.81E-4 1.08E-3 1.97E-4 1.68E-3 4.43 9.32E-6* 1657 + + REPIN1 NA
cg23628297 −1.80E-4 3.98E-1 −2.38E-3 2.55E-6* −3.19 1.40E-3 1657 - - NA NA
cg10857441 6.07E-4 4.41E-1 6.90E-3 7.63E-6* 3.01 2.61E-3 1657 + + POLS 5’UTR
cg14392653 −2.63E-5 6.77E-1 −1.89E-3 8.25E-6* −2.70 6.93E-3 1657 - - NEK6 3’UTR
cg24834751 −2.44E-5 8.56E-1 −1.36E-3 1.35E-6* −2.70 6.99E-3 1657 - - NA NA
cg26787998 3.11E-5 7.19E-1 3.34E-4 9.29E-6* 2.64 8.31E-3 1657 + + TAPBP 3’UTR
cg15120534 −4.35E-6 9.39E-1 −3.40E-4 9.80E-6* −2.39 1.68E-2 1657 - - NA Shelf
cg01973394 8.09E-5 6.92E-1 −2.29E-3 6.44E-6* −2.04 4.16E-2 1657 - + NA NA
Open in a new tab

CpG probes are listed for results that pass the suggestive p-value (*) threshold of < 1x10−5 in any of the individual analyses. The genome-wide significance level (**) was 9x10-8 for EPIC (CLSA) and 2.4x10−7 for 450k (TwinsUK and Combined). The combined Z-score and Direction columns indicate whether the DNAm directional change was consistent in the two studies.

An EWAS was also performed in the TwinsUK cohort, with no results passing the epigenome-wide significance threshold for the 450k array of p ≤ 2.4x10−7 [46]. Suggestive results were identified for five and two CpGs for glaucoma and IOP, respectively (Figure 1(b) and Supplementary Figure S3B, Tables 2 and 3).

A meta-EWAS was performed in the subset of ~387k probes that were present in both CLSA and TwinsUK i.e., only the subset common to both the EPIC v1 and 450k arrays, using the statistical threshold for the latter (p ≤ 2.4x10−7). This meta-analysis of the two cohorts comprised 1,657 individuals and 82 glaucoma cases (see Methods). Twelve CpGs passed the suggestive threshold, with a consistent direction of DNAm change across both the datasets. Additionally, two of these CpGs were significantly associated with glaucoma (cg03498697, p = 6.86x10−8; cg06044751, p = 1.76x10−7). CpG cg03498697 resides within the downstream shore region of the 5’ CpG island of the FRMD3 (FERM Domain Containing 3) gene, which encodes a multifunctional structural protein [57]. The promoter region of this gene demonstrates selective activity across a range of tissues in EpiMap 18-state chromatin segmentation [58]. In blood tissue, this ranges from Active Promoter Flanking (TssFlnk) to Repressed Polycomb (ReprPC). In eye tissue, it possesses a Bivalent Transcription Start Site (TssBiv) signature. The second CpG, cg06044751, resides intronically within PALLD (Palladin), which encodes a cytoskeletal protein. Another interesting CpG site, cg10495606 is located within the first intron of PRKCE, (Protein kinase C epsilon), a crucial enzyme that regulates various cellular processes, including immune response, apoptosis, and neuronal signalling [59]. The CpG locus itself displays a quiescent function in both blood-derived and eye tissue chromatin segmentation data [58].

To further investigate these findings, an eFORGE analysis [55] was performed for tissue-specific regulatory enrichment on the EWAS results. In results obtained from the top associated 1,000 probes from the denser EPIC array used in the CLSA, with the entire array as background, we observed a significant enrichment (FDR q value < 0.05) in the consolidated Roadmap Epigenomics DNase-I hypersensitivity sites (DHS), including in embryonic (H9) and haematopoietic stem cells, and fetal brain (female) tissue (Figure 2 and Supplementary Table S2).

Figure 2.

Figure 2.

Open in a new tab

eFORGE tissue enrichment result for the top 1000 probes from the CLSA EWAS analysis.

Plasma protein DNAm EpiScores are associated with glaucoma and IOP

The EpiScores for 109 plasma proteins were derived from DNAm data obtained from the CLSA and TwinsUK cohorts. In the CLSA cohort, DNAm EpiScores for CCL22, L-selectin, and PAPP-A negatively correlated with glaucoma status (Table 2, Figure 3(a), nominal p < 0.05). Inconsistent with previous observations [60], the HDL-cholesterol EpiScore was positively correlated with glaucoma. Furthermore, 23 separate EpiScores demonstrated nominally significant correlations with the quantitative traits of IOP (Table 4, nominal p < 0.05). These included biomarkers related to extracellular matrix integrity (ADAMTS, p = 0.043), vitamin E transport (Afamin, p = 0.032), immune response (CD163, p = 0.017), and components of the complement system (C5a, C9, p = 0.039, and 0.032, respectively). IOP EpiScore correlations were also detected with the cell adhesion molecules VCAM-1 and L-selectin, which are involved in leukocyte endothelial transmigration. However, the association between IOP and L-selectin was positive, in contrast to the glaucoma association. EpiScores were also calculated in the TwinsUK dataset. These TwinsUK results differed from CLSA with two nominal associations for glaucoma and ten for IOP (See Table 4). A sensitivity analysis was conducted to assess the effect of IOP-lowering medication intake by excluding subjects in TwinsUK who had taken these medications and then re-running the regression analysis on IOP measurements. The results were consistent for the top nominally significant proteins. None of the proteins met the Bonferroni significance threshold (Supplementary Table S3).

Figure 3.

Figure 3.

Open in a new tab

(A) Violin plot showing the significant correlation between the EpiScore estimate for CCL22 levels and glaucoma (p = 0.03); (b) Violin plot showing the significant correlation between the EpiScore estimate for TNFRSF1B levels and IOP (p = 1.31E-4), reaching Bonferroni significance; (c) Violin plots showing range of DNAm clock ages in CLSA cohort; (d) Violin plot showing significant correlations between GrimAge age acceleration and glaucoma in HRS cohort.

Table 4.

EpiScore plasma protein DNAm estimator associations in CLSA, TwinsUK and Combined cohorts.

EpiScore measure Trait CLSA
 TwinsUK
 Meta-Analysis
Odds ratio SE P value Odds ratio SE P value Odds ratio SE P value
CCL17 Glaucoma NS NS NS −2.12 0.90 1.85E-2 −2.12 0.56 3.42E-2
CCL22 Glaucoma −7.91 2.65 3.00E-3 NS NS NS −2.16 1.97 3.11E-2
HDL.Cholesterol Glaucoma 0.09 0.05 4.00E-2 NS NS NS NS NS NS
L.selectin Glaucoma −1.29 0.62 3.90E-2 NS NS NS NS NS NS
PAPP.A Glaucoma −3.99 1.87 3.30E-2 NS NS NS NS NS NS
ADAMTS IOP 2.02 32.70 4.30E-2 NS NS NS 2.29 19.95 2.22E-2
Afamin IOP 2.15 60.40 3.20E-2 NS NS NS 2.02 30.56 4.36E-2
B2.microglobulin IOP −2.62 7.85 9.00E-3 NS NS NS −2.66 6.73 7.77E-3
CCL18 IOP −2.84 12.87 5.00E-3 −2.15 16.02 3.18E-2 −3.56 10.03 3.70E-4
CCL25_C.C IOP NS NS NS 2.12 34.20 3.44E-2 NS NS NS
CD163 IOP −2.38 10.54 1.70E-2 NS NS NS −2.29 9.82 2.19E-2
CD48_antigen IOP NS NS NS 1.97 72.36 4.89E-2 NS NS NS
CDL5 IOP −2.61 15.87 9.00E-3 NS NS NS −2.56 12.09 1.04E-2
CHIT.1 IOP −2.08 18.20 3.80E-2 NS NS NS −1.97 12.61 4.87E-2
Complement.C5a IOP −2.07 13.94 3.90E-2 NS NS NS NS NS NS
Complement.c9 IOP −2.15 19.72 3.20E-2 NS NS NS −2.18 9.37 2.89E-2
CRP IOP −2.70 19.20 7.00E-3 NS NS NS −2.40 12.11 1.65E-2
CXCL10 IOP −2.17 10.86 3.00E-2 NS NS NS −2.31 9.57 2.11E-2
EN.RAGE IOP −1.99 21.27 4.70E-2 NS NS NS NS NS NS
EZR IOP −2.41 38.51 1.60E-2 NS NS NS NS NS NS
Galectin.4 IOP NS NS NS 2.22 33.20 2.66E-2 NS NS NS
GHR IOP NS NS NS 2.37 13.99 1.80E-2 2.31 11.94 2.10E-2
GPIba IOP 2.01 13.94 4.40E-2 NS NS NS NS NS NS
HGFA IOP −2.21 20.40 2.70E-2 NS NS NS −2.70 15.98 6.95E-3
IGFBP.1 IOP 2.04 15.79 4.10E-2 NS NS NS NS NS NS
L.selectin IOP 2.02 11.25 4.40E-2 NS NS NS NS NS NS
Lymphotoxin.abeta IOP 2.65 18.18 8.00E-3 NS NS NS 2.48 11.54 1.31E-2
N.CDase IOP NS NS NS 2.05 23.19 4.06E-2 NS NS NS
NEP IOP −2.02 15.42 4.40E-2 NS NS NS NS NS NS
Osteomodulin IOP NS NS NS −2.01 13.57 4.51E-2 NS NS NS
PAPP.A IOP NS NS NS −2.59 8.90 9.77E-3 −2.69 8.60 7.17E-3
S100.A9 IOP −2.91 37.99 4.00E-3 NS NS NS NS NS NS
SIGLEC1 IOP NS NS NS −2.65 14.36 8.29E-3 −2.46 8.58 1.38E-2
Stanniocalcin.1 IOP −2.96 15.09 3.00E-3 NS NS NS −2.87 12.65 4.08E-3
TNFRSF1B IOP −3.54 15.84 4.16E-4 NS NS NS −3.82 12.67 1.31E-4*
VCAM1 IOP −2.04 21.77 4.10E-2 NS NS NS NS NS NS
Open in a new tab

*Bonferroni significant (p < 2.3x10−4). “NS” denotes ‘non significant’ (p > 0.05).

Due to the relatively small sample sizes and lack of significance after adjusting for multiple testing in the EpiScore findings from the two cohorts (CLSA and TwinsUK), we performed a meta-analysis for both cohorts. We identified two EpiScores (CCL17 and CCL22) that were nominally significantly correlated with glaucoma (Table 4 and Supplementary Table S4). Additionally, 24 EpiScore associations were nominally significant for IOP, including one protein (TNFRSF1B, p = 1.31x10−4) at the Bonferroni significance level (see Table 4, Figure 3B).

Epigenetic biological age acceleration and glaucoma

We calculated the biological age estimate for the five DNAm ‘clocks.’ These included the first-generation Horvath and Hannum et al., the second-generation PhenoAge and GrimAge, and third-generation DunedinPACE clocks. As expected, all these DNAm clocks showed a strong correlation with chronological age and across the range of biological clock estimators (Supplementary Figure S4). The calculated ranges for these DNAm clock-age estimates are shown in Figure 3C and Supplementary Table S5. Biological age acceleration was then calculated via residual, and this was assessed for its association with glaucoma (see Methods).

Nominal associations were identified with decelerated age, unexpectedly, via GrimAge and DunedinPACE in CLSA (Table 5). None of these decelerated age associations remained significant after multiple-testing correction or were supported by the IOP measurement in either the CLSA or TwinsUK cohorts (Supplementary Table S6). A similar sensitivity analysis was performed for IOP measurements in the TwinsUK cohort by excluding participants who had taken IOP-lowering medication. These results remained consistent with previous findings that showed no significant relationship (Supplementary Table S7). We then utilised the HRS database to attempt to further explore DNAm clock associations [28]. The HRS does not provide individual CpG probe data but makes available pre-calculated epigenetic aging results from three DNAm clocks (HorvathAge, HannumAge and DNAmGrimAge). In the HRS cohort, 215 of the 3,453 subjects with available clock data had glaucoma. In this larger dataset, we found that positively accelerated GrimAge (DNAmGrimAgeAccel) was significantly associated with glaucoma after adjusting for age and sex (p = 0.01) (Figure 3D).

Table 5.

Results for age acceleration results calculated via residuals for DNAm clocks and association with glaucoma.

Age Acceleration CLSA
 Twin
 HRS
Beta SE P value Beta SE P value Beta SE P value
Hannum et al. −0.002 0.002 3.02E-1 0.004 0.002 6.40E-2 0.001 0.001 4.25E-1
Horvath −0.002 0.001 7.87E-2 0.004 0.002 6.75E-2 0.002 0.001 1.24E-1
PhenoAge −0.002 0.001 1.86E-1 0.001 0.002 7.76E-1 NA NA NA
DNAmGrimAge −0.005 0.002 4.32E-2* 0.002 0.004 6.50E-1 0.003 0.001 1.11E-2*
DunedinPACE −0.159 0.081 4.88E-2* 0.018 0.112 8.72E-1 NA NA NA
Open in a new tab

*Nominally significant (p < 0.05, values in bold).

Discussion

In this study, we explored the relationship between potential blood-derived DNA methylation (DNAm) biomarkers with glaucoma and IOP in two distinct cohorts: CLSA and TwinsUK. We examined three modalities derived from DNAm data as potential biomarkers: DNAm levels assessed by EWAS, EpiScores, and DNAm clock associations.

We observed two significant genome-wide results for the combined EWAS of cg03498697 within FRMD3 and cg06044751 within PALLD However, the nominal and suggestive results are also of interest and should be investigated in future studies with additional power. The blood-derived EWAS findings can only be inferred as potential biomarkers and are not directly involved in the disease process. However, it is possible that particular changes may occur across multiple tissues, as can be the case with some age-related DNAm changes [61]. Validation in disease-relevant tissues is needed to support any pathogenic role. The FRMD3 gene is implicated in diabetic nephropathy and has been suggested to play a role in endothelial cell function and vascular integrity [62], which are also relevant to glaucoma pathophysiology, particularly in relation to vascular dysfunction and retinal ganglion cell susceptibility. Of note, the PRKCE gene has been identified as a genetic risk locus for glaucoma via GWAS and prioritized as a druggable target due to its association with Protein Kinase C inhibitors and Protein Kinase C Epsilon inhibitors [1].

EpiScore analysis of DNAm estimators of long-term plasma protein levels highlighted several blood-derived biomarkers with nominal associations with glaucoma and IOP in individual cohorts. Among these, TNFRSF1B exhibited a Bonferroni-significant association with IOP with a consistent direction in the combined analysis. The TNF pathways, including TNFRSF1B have been shown to be upregulated in animal studies of mild IOP elevation, with a potential connection with sterile inflammation and inflammaging [24].

The other nominal associations of these plasma proteins in ocular diseases remain to be further explored for their potential utility as novel biomarkers and their underlying mechanisms. The findings for ADAMTS are of particular interest, since genes in this family have been associated with glaucoma or related traits in various models and studies [63–66] and Afamin, a vitamin E binding glycoprotein [67], is enriched in the aqueous humor of glaucoma patients [68].

The ability to accurately quantify aging-related DNAm changes has opened novel avenues for investigating aging-related diseases, such as glaucoma. By assessing biological age via DNAm clocks, we identified nominal associations with age acceleration. Results from the smaller number of glaucoma cases in the CLSA defied expectations that positive age acceleration would be associated with this age-related disease. However, in the larger and more powerful HRS cohort, DNAmGrimAge showed a significant positive association with glaucoma after adjusting for age and sex, suggesting that further exploration is warranted for phenotypic DNAm clocks, such as GrimAge, or the formulation of a glaucoma-specific DNAm clock [21], as novel predictive biomarkers for glaucoma. This aligns with recent research showing that accelerated epigenetic aging is associated with faster progression of primary open-angle glaucoma [69]. Furthermore, the timing of these glaucoma-associated DNAm changes requires a detailed analysis to determine whether any of these have predictive ability. As shown for other common complex diseases, improvements in risk prediction may be possible by combining inherited genetic risk via Polygenic Risk Scores (PRS) with environmental and aging factors captured by blood-derived DNAm risk scores (MRS) [34]. Along with increasing the power of DNAm-related studies, the integration of these epigenetic data with glaucoma PRS is also worthy of future exploration.

While providing valuable insights into epigenetic markers associated with glaucoma, this study has several limitations. One limitation of this study is the inability to distinguish glaucoma subtypes. Primary open-angle glaucoma is the most common form and accounts for the vast majority of cases in British populations [70], rarer forms have distinct underlying causes and molecular profiles that may have introduced phenotypic heterogeneity in the data. This expected reduction in statistical power would have however not affected the specificity of the epigenetic associations. The sample size was relatively small and not fully representative of the diverse population of glaucoma patients worldwide. A larger cohort would provide more robust statistical power and greater ability to detect subtle epigenetic variations associated with glaucoma. Given that the cohorts were population-based studies of healthy aging, the number of glaucoma cases may have been affected by healthy volunteer bias, reducing the power of our statistical associations and the ability to fully explore the heterogeneity of the disease.

While blood-derived DNA methylation offers practical advantages for large-scale population studies and non-invasive biomarker discovery, it is unlikely to reflect tissue-specific epigenetic alterations occurring in key ocular structures such as the optic nerve, trabecular meshwork, or Schlemm’s canal, that directly implicated in glaucoma pathophysiology. Therefore, the blood-based findings in this study should be interpreted as systemic markers, potentially reflecting broader processes such as aging or inflammation, rather than direct disease mechanisms. For example, methylation profiling of trabecular meshwork and Schlemm’s canal cells has revealed differential methylation in glaucoma-associated genes and pathways related to extracellular matrix remodeling and aqueous humor outflow [71,72]. Tissue-level analyses across ocular structures have also highlighted epigenetic regulation of developmental genes relevant to glaucoma pathogenesis [73]. These findings underscore the importance of integrating ocular tissue epigenetic data to validate blood-based associations. Future longitudinal studies that track methylation changes alongside intraocular pressure (IOP) and optic nerve status within individuals may improve our understanding of glaucoma progression and enhance the predictive utility of epigenetic biomarkers.

A key challenge of EWAS using Illumina methylation arrays is the functional interpretation of individual CpG site associations. While certain CpG sites demonstrate statistically significant associations with disease phenotypes, such as the significant probe cg03498697 near FRMD3 in our study, the potential functional relevance of single CpG needs to be formulated through its epigenomic location (Promoter, Enhancer, Insulator, Genic Transcribed Region, etc). The effect on gene expression levels will depend on tissue specificity, genomic context, and possible regulatory interactions over long genomic distances. CpG methylation within Transcription Factor Binding Sites can repress or promote the binding of Transcription Factors with downstream functional sequelae [74]. Therefore, these associations are hypothesis-generating and require integrating with other types of genomic and transcriptomic data or functional validation through methods such as CRISPR-based gene activation (CRISPRa) and inhibition (CRISPRi) [75] to establish biological significance.This study provides predominately suggestive and not yet fully statistically significant evidence associating changes in DNAm, EpiScores, and epigenetic age acceleration with glaucoma and IOP. Better-powered studies in larger, more diverse populations are needed in the future to replicate the identified possible DNAm biomarkers to confirm their validity and robustness. Longitudinal studies are essential to determine whether these epigenetic markers can predict glaucoma risk and aid in early diagnosis and intervention.

Supplementary Material

Supplemental Material
KEPI_A_2566496_SM5536.zip (5.1MB, zip)

Acknowledgments

This research was made possible using the data/biospecimens collected by the Canadian Longitudinal Study on Aging (CLSA). Funding for the CLSA is provided by the Government of Canada through the Canadian Institutes of Health Research (CIHR) under grant reference LSA 94473, the Canada Foundation for Innovation, and the provinces of Newfoundland, Nova Scotia, Quebec, Ontario, Manitoba, Alberta, and British Columbia. This research has been conducted using the CLSA dataset [Baseline Comprehensive Dataset - Version 7.0, Follow-up 1 Comprehensive Dataset - Version 5.0, Vital Status, Genome-wide data version 3.0, Epigenetics Version 1.1, Baseline Retinal Scans and Follow-up 1 Retinal Scans, under Application Number [2104037]. The CLSA is led by Drs. Parminder Raina, Christina Wolfson, and Susan Kirkland. The opinions expressed in this manuscript are the author’s own and do not necessarily reflect the views of the Canadian Longitudinal Study on Aging.

This research has been conducted using the TwinsUK resource. TwinsUK is funded by the Wellcome Trust, Medical Research Council, Versus Arthritis, European Union Horizon 2020, Chronic Disease Research Foundation (CDRF), Wellcome Leap Dynamic Resilience Programme (co-funded by Temasek Trust), Zoe Ltd, the National Institute for Health and Care Research (NIHR) Clinical Research Network (CRN) and Biomedical Research Centre based at Guy’s and St Thomas’ NHS Foundation Trust in partnership with King’s College London. TwinsUK published DNA methylation 450k array data was generated by T Spector, J Bell, W Yuan. The DSK laboratory is supported in part by an unrestricted grant from Research Preventing Blindness to the University of California, Irvine, Department of Ophthalmology. HRS is supported by the National Institute on Aging (NIA U01AG009740). The genotyping was partially funded by separate awards from NIA (RC2 AG036495 and RC4 AG039029). Our genotyping was conducted by the NIH Center for Inherited Disease Research (CIDR) at Johns Hopkins University. Genotyping quality control and final preparation were performed by the Genetics Coordinating Center at University of Washington (Phases 1-3) and the University of Michigan (Phase 4).

Members of The International Glaucoma Genetics Consortium (IGGC) are listed in Supplementary Table 8.

C.G.B. and P.G.H. supervised the study and O.A.M., A.P.K., D.S.-K., D.A.M., J.L.W. and C.J.H. provided comments. X.J. and P.G.H. contributed to data acquisition. X.J. performed the primary analysis in the TwinsUK dataset, prepared the analyses pipeline and coding that was used in the rest of the analyses and conducted meta-analyses, with contributions from C.G.B., and P.G.H. X.J., C.G.B., and P.G.H. prepared the manuscript, with comments from all co-authors.

Funding Statement

This research was funded by the BrightFocus grant, [G2021011S] (PGH, CGB).

Disclosure statement

APK has acted as a paid consultant or lecturer to Abbvie, Aerie, Allergan, Google Health, Heidelberg Engineering, Novartis, Reichert, Santen, Thea and Topcon. The remaining authors declare no conflicts of interest.

Data availability statement

Data are available from the Canadian Longitudinal Study on Aging (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data. Twins UK data can be requested at http://twinsuk.ac.uk. The Health and Retirement Study is a publicly available data set available through the University of Michigan at http://hrsonline.isr.umich.edu/.

Accessibility

The authors have applied for open access the Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version.

Ethics statements

The CLSA is conducted in accordance with the Canadian Tri-Council Policy Statement and the Declaration of Helsinki. The CLSA has received ethics board approval from McGill University Health Centre (MUHC), McMaster University, University of Sherbrooke, University of Manitoba, University of Victoria, Simon Fraser University, Island Health, University of Calgary, Bruyère Continuing Care, University of British Columbia, Memorial University, Dalhousie University and its protocol is reviewed annually. The TwinsUK study had ethical approval from Guys & St Thomas’ Trust (GSTT) Ethics Committee and was conducted in accordance with the tenets of the Declaration of Helsinki. HRS study uses a public use dataset and therefore does not require additional Institutional Review Board approval. Primary data collection for the Health and Retirement Study was approved thorough the University of Michigan Institutional Review Board. Informed consent was obtained from all participants.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15592294.2025.2566496

References

  • [1].Han X, Gharahkhani P, Hamel AR, et al. Large-scale multitrait genome-wide association analyses identify hundreds of glaucoma risk loci. Nat Genet. 2023. Jul;55(7):1116–17. doi: 10.1038/s41588-023-01428-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [2].Gharahkhani P, Jorgenson E, Hysi P, et al. Genome-wide meta-analysis identifies 127 open-angle glaucoma loci with consistent effect across ancestries. Nat Commun. 2021. Feb 24;12(1):1258. doi: 10.1038/s41467-020-20851-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [3].Liu WW, Sun Y.. Epigenetics in glaucoma: a link between histone methylation and neurodegeneration. J Clin Invest. 2023. Aug 1;133(15). doi: 10.1172/JCI173784 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [4].Guan J, Li Z, Wumaier A, et al. Critical role of transcriptome-wide m6A methylation in the aqueous humor of patients with pseudoexfoliation glaucoma. Exp Eye Res. 2023. Jun;231:109473. doi: 10.1016/j.exer.2023.109473 [DOI] [PubMed] [Google Scholar]
  • [5].Bell CG. Epigenomic insights into common human disease pathology. Cell Mol Life Sci. 2024. Apr 11;81(1):178. doi: 10.1007/s00018-024-05206-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [6].Schubeler D. Function and information content of DNA methylation. Nature. 2015. Jan 15;517(7534):321–326. doi: 10.1038/nature14192 [DOI] [PubMed] [Google Scholar]
  • [7].Moore LD, Le T, Fan GP. DNA methylation and its basic function. Neuropsychopharmacol. 2013. Jan;38(1):23–38. doi: 10.1038/npp.2012.112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [8].Titus AJ, Gallimore RM, Salas LA, et al. Cell-type deconvolution from DNA methylation: a review of recent applications. Hum Mol Genet. 2017. Oct 1;26(R2):R216–R224. doi: 10.1093/hmg/ddx275 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [9].Hillary RF, McCartney DL, Smith HM, et al. Blood-based epigenome-wide analyses of 19 common disease states: a longitudinal, population-based linked cohort study of 18,413 Scottish individuals. PLOS Med. 2023. Jul;20(7):e1004247. doi: 10.1371/journal.pmed.1004247 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Chen J, Yuan XL, Zhou X, et al. Mendelian randomization implicates causal association between epigenetic age acceleration and age-related eye diseases or glaucoma endophenotypes. Clin Epigenet. 2024. Aug 14;16(1):106. doi: 10.1186/s13148-024-01723-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [11].Burdon KP, Awadalla MS, Mitchell P, et al. DNA methylation at the 9p21 glaucoma susceptibility locus is associated with normal-tension glaucoma. Ophthalmic Genet. 2018. Apr;39(2):221–227. doi: 10.1080/13816810.2017.1413659 [DOI] [PubMed] [Google Scholar]
  • [12].Gadd DA, Hillary RF, McCartney DL, et al. Epigenetic scores for the circulating proteome as tools for disease prediction. Elife. 2022. Jan 13;11. doi: 10.7554/eLife.71802 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [13].Gadd DA, Smith HM, Mullin D, et al. DNAm scores for serum GDF15 and NT-proBNP levels associate with a range of traits affecting the body and brain. MedRxiv. 2023:2023.10.18.23297200. [DOI] [PMC free article] [PubMed]
  • [14].Hillary RF, Ng HK, McCartney DL, et al. Blood-based epigenome-wide analyses of chronic low-grade inflammation across diverse population cohorts. Cell Genom. 2024. May 8;4(5):100544. doi: 10.1016/j.xgen.2024.100544 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Niccoli T, Partridge L. Ageing as a risk factor for disease. Curr Biol. 2012. Sep 11;22(17):R741–R752. doi: 10.1016/j.cub.2012.07.024 [DOI] [PubMed] [Google Scholar]
  • [16].Laitinen A, Laatikainen L, Härkänen T, et al. Prevalence of major eye diseases and causes of visual impairment in the adult Finnish population: a nationwide population-based survey. Acta Ophthalmol. 2010. Jun;88(4):463–471. doi: 10.1111/j.1755-3768.2009.01566.x [DOI] [PubMed] [Google Scholar]
  • [17].Beard JR, Officer AM, Cassels AK. The world report on ageing and Health. Gerontologist. 2016. Apr;56:S163–S166. doi: 10.1093/geront/gnw037 [DOI] [PubMed] [Google Scholar]
  • [18].Partridge L, Deelen J, Slagboom PE. Facing up to the global challenges of ageing. Nature. 2018. Sep;561(7721):45–56. doi: 10.1038/s41586-018-0457-8 [DOI] [PubMed] [Google Scholar]
  • [19].López-Otin C, Blasco MA, Partridge L, et al. Hallmarks of aging: an expanding universe. Cell. 2023. Jan 19;186(2):243–278. doi: 10.1016/j.cell.2022.11.001 [DOI] [PubMed] [Google Scholar]
  • [20].Skowronska-Krawczyk D. Hallmarks of aging: causes and consequences. Aging Biol. 2023;1(1):20230011. doi: 10.59368/agingbio.20230011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].Bell CG, Lowe R, Adams PD, et al. DNA methylation aging clocks: challenges and recommendations. Genome Biol. 2019. Nov 25;20(1):249. doi: 10.1186/s13059-019-1824-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [22].Marioni RE, Shah S, McRae AF, et al. DNA methylation age of blood predicts all-cause mortality in later life. Genome Biol. 2015. Jan 30;16(1). doi: 10.1186/s13059-015-0584-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [23].Zhang WQ, Qu J, Liu GH, et al. The ageing epigenome and its rejuvenation. Nat Rev Mol Cell Bio. 2020. Mar;21(3):137–150. doi: 10.1038/s41580-019-0204-5 [DOI] [PubMed] [Google Scholar]
  • [24].Xu Q, Rydz C, Nguyen Huu VA, et al. Stress induced aging in mouse eye. Aging Cell. 2022. Dec;21(12):e13737. doi: 10.1111/acel.13737 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [25].Christofidou P, Bell CG. The predictive power of profiling the DNA methylome in human health and disease. Epigenomics. 2025. Jun;17(9):599–610. doi: 10.1080/17501911.2025.2500907 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [26].Raina P, Wolfson C, Kirkland S, et al. Cohort profile: the Canadian Longitudinal Study on Aging (CLSA). Int J Epidemiol. 2019. Dec 1;48(6):1752–1753j [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [27].Kurushima Y, Tsai PC, Castillo-Fernandez J, et al. Epigenetic findings in periodontitis in Uk twins: a cross-sectional study. Clin Epigenet. 2019. Feb 13;11(1):27. doi: 10.1186/s13148-019-0614-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [28].Crimmins EM, Thyagarajan B, Levine ME, et al. Associations of age, sex, race/ethnicity, and education with 13 epigenetic clocks in a nationally representative U.S. Sample: the health and retirement study. J Gerontol A Biol Sci Med Sci. 2021. May 22;76(6):1117–1123. doi: 10.1093/gerona/glab016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [29].Forgetta V, Li R, Darmond-Zwaig C, et al. Cohort profile: genomic data for 26 622 individuals from the Canadian Longitudinal Study on Aging (CLSA). BMJ Open. 2022. Mar 10;12(3):e059021. doi: 10.1136/bmjopen-2021-059021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [30].Buys YM, Gaspo R, Kwok K, et al. Referral source, symptoms, and severity at diagnosis of ocular hypertension or open-angle glaucoma in various practices. Can J Ophthalmol. 2012. Jun;47(3):217–222. doi: 10.1016/j.jcjo.2012.03.031 [DOI] [PubMed] [Google Scholar]
  • [31].Nag A, Venturini C, Small KS, et al. A genome-wide association study of intra-ocular pressure suggests a novel association in the gene FAM125B in the TwinsUK cohort. Hum Mol Genet. 2014. Jun 15;23(12):3343–3348. doi: 10.1093/hmg/ddu050 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Hysi PG, Cheng CY, Springelkamp H, et al. Genome-wide analysis of multi-ancestry cohorts identifies new loci influencing intraocular pressure and susceptibility to glaucoma. Nat Genet. 2014. Oct;46(10):1126–1130. doi: 10.1038/ng.3087 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].van der Valk R, Webers CAB, Schouten JSAG, et al. Intraocular pressure–lowering effects of all commonly used glaucoma drugs: a meta-analysis of randomized clinical trials. Ophthalmology. 2005. July 01;112(7):1177–1185. doi: 10.1016/j.ophtha.2005.01.042 [DOI] [PubMed] [Google Scholar]
  • [34].Nabais MF, Gadd DA, Hannon E, et al. An overview of DNA methylation-derived trait score methods and applications. Genome Biol. 2023. Feb 16;24(1):28. doi: 10.1186/s13059-023-02855-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [35].Du P, Kibbe WA, Lin SM. Lumi: a pipeline for processing Illumina microarray. Bioinformatics. 2008. Jul 1;24(13):1547–1548. doi: 10.1093/bioinformatics/btn224 [DOI] [PubMed] [Google Scholar]
  • [36].Pidsley R, Y Wong CC, Volta M, et al. A data-driven approach to preprocessing Illumina 450K methylation array data. BMC Genomics. 2013. May 1;14(1):293. doi: 10.1186/1471-2164-14-293 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [37].Pidsley R, Zotenko E, Peters TJ, et al. Critical evaluation of the Illumina MethylationEPIC BeadChip microarray for whole-genome DNA methylation profiling. Genome Biol. 2016. Oct 7;17(1):208. doi: 10.1186/s13059-016-1066-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [38].Hannum G, Guinney J, Zhao L, et al. Genome-wide methylation profiles reveal quantitative views of human aging rates. Mol Cell. 2013. Jan 24;49(2):359–367. doi: 10.1016/j.molcel.2012.10.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [39].Teschendorff AE, Marabita F, Lechner M, et al. A beta-mixture quantile normalization method for correcting probe design bias in Illumina Infinium 450 k DNA methylation data. Bioinformatics. 2013. Jan 15;29(2):189–196. doi: 10.1093/bioinformatics/bts680 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [40].Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics. 2007. Jan;8(1):118–127. doi: 10.1093/biostatistics/kxj037 [DOI] [PubMed] [Google Scholar]
  • [41].McCartney DL, Walker RM, Morris SW, et al. Identification of polymorphic and off-target probe binding sites on the Illumina Infinium MethylationEPIC BeadChip. Genom Data. 2016. Sep;9:22–24. doi: 10.1016/j.gdata.2016.05.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [42].Bell CG, Xia Y, Yuan W, et al. Novel regional age-associated DNA methylation changes within human common disease-associated loci. Genome Biol. 2016. Sep 23;17(1):193. doi: 10.1186/s13059-016-1051-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [43].Chen YA, Lemire M, Choufani S, et al. Discovery of cross-reactive probes and polymorphic CpGs in the Illumina Infinium HumanMethylation450 microarray. Epigenetics. 2013. Feb;8(2):203–209. doi: 10.4161/epi.23470 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [44].Jarrar ZA, Al-Nosairy KO, Jiang X, et al. Temporal-to-nasal macular ganglion cell and inner plexiform layer ratios in a large adult Twin cohort: correlations with age and heritability. Invest Ophthalmol Vis Sci. 2024. Feb 1;65(2):26. doi: 10.1167/iovs.65.2.26 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [45].Mansell G, Gorrie-Stone TJ, Bao Y, et al. Guidance for DNA methylation studies: statistical insights from the Illumina epic array. BMC Genomics. 2019. May 14;20(1):366. doi: 10.1186/s12864-019-5761-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [46].Saffari A, Silver MJ, Zavattari P, et al. Estimation of a significance threshold for epigenome-wide association studies. Genet Epidemiol. 2018. Feb;42(1):20–33. doi: 10.1002/gepi.22086 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [47].Houseman EA, Accomando WP, Koestler DC, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinf. 2012. May 08;13(1):86. doi: 10.1186/1471-2105-13-86 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [48].Willer CJ, Li Y, Abecasis GR. Metal: fast and efficient meta-analysis of genomewide association scans. Bioinformatics. 2010. Sep 1;26(17):2190–2191. doi: 10.1093/bioinformatics/btq340 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [49].Reese SE, Xu CJ, den Dekker HT, et al. Epigenome-wide meta-analysis of DNA methylation and childhood asthma. J Allergy Clin Immunol. 2019. Jun;143(6):2062–2074. doi: 10.1016/j.jaci.2018.11.043 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [50].Hillary RF, Marioni RE. MethylDetectR: a software for methylation-based health profiling. Wellcome Open Res. 2020;5:283. doi: 10.12688/wellcomeopenres.16458.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [51].Horvath S. DNA methylation age of human tissues and cell types. Genome Biol. 2013;14(10). doi: 10.1186/gb-2013-14-10-r115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [52].Levine ME, Lu AT, Quach A, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging-US. 2018. Apr;10(4):573–591. doi: 10.18632/aging.101414 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [53].Lu AT, Quach A, Wilson JG, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging-US. 2019. Jan 31;11(2):303–327. doi: 10.18632/aging.101684 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [54].Belsky DW, Caspi A, Corcoran DL, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. Elife. 2022. Jan 14;11. doi: 10.7554/eLife.73420 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [55].Thrush KL, Higgins-Chen AT, Liu Z, et al. R methylCIPHER: a methylation clock investigational package for hypothesis-driven evaluation & Research. BioRxiv. 2022. Jul 13:2022.07.13.499978.
  • [56].Madden RA, McCartney DL, Walker RM, et al. Birth weight associations with DNA methylation differences in an adult population. Epigenetics. 2021. Jun-Jul;16(7):783–796. doi: 10.1080/15592294.2020.1827713 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [57].Ma J, Gong Y, Sun X, et al. Tumor suppressor FRMD3 controls mammary epithelial cell fate determination via notch signaling pathway. Sci Adv. 2024. Jul 5;10(27):eadk8958. doi: 10.1126/sciadv.adk8958 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [58].Boix CA, James BT, Park YP, et al. Regulatory genomic circuitry of human disease loci by integrative epigenomics. Nature. 2021. Feb;590(7845):300–307. doi: 10.1038/s41586-020-03145-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [59].Lim PS, Sutton CR, Rao S. Protein kinase C in the immune system: from signalling to chromatin regulation. Immunology. 2015. Dec;146(4):508–522. doi: 10.1111/imm.12510 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [60].Posch-Pertl L, Michelitsch M, Wagner G, et al. Cholesterol and glaucoma: a systematic review and meta-analysis. Acta Ophthalmol. 2022. Mar;100(2):148–158. doi: 10.1111/aos.14769 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [61].Seale K, Horvath S, Teschendorff A, et al. Making sense of the ageing methylome. Nat Rev Genet. 2022. Oct;23(10):585–605. doi: 10.1038/s41576-022-00477-6 [DOI] [PubMed] [Google Scholar]
  • [62].Buffon MP, Sortica DA, Gerchman F, et al. FRMD3 gene: its role in diabetic kidney disease. A narrative review. Diabetol Metab Syndr. 2015;7:118. doi: 10.1186/s13098-015-0114-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [63].Kuchtey J, Olson LM, Rinkoski T, et al. Mapping of the disease locus and identification of ADAMTS10 as a candidate gene in a canine model of primary open angle glaucoma. PLOS Genet. 2011. Feb;7(2):e1001306. doi: 10.1371/journal.pgen.1001306 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [64].Wu HJ, Mortlock DP, Kuchtey RW, et al. Altered ocular fibrillin microfibril composition in mice with a glaucoma-causing mutation of Adamts10. Invest Ophthalmol Vis Sci. 2021. Aug 2;62(10):26. doi: 10.1167/iovs.62.10.26 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [65].Wain LV, Verwoert GC, O’Reilly PF, et al. Genome-wide association study identifies six new loci influencing pulse pressure and mean arterial pressure. Nat Genet. 2011. Sep 11;43(10):1005–1011. doi: 10.1038/ng.922 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [66].Springelkamp H, Iglesias AI, Mishra A, et al. New insights into the genetics of primary open-angle glaucoma based on meta-analyses of intraocular pressure and optic disc characteristics. Hum Mol Genet. 2017. Jan 15;26(2):438–453. doi: 10.1093/hmg/ddw399 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [67].Jerkovic L, Voegele AF, Chwatal S, et al. Afamin is a novel human vitamin E-binding glycoprotein characterization and in vitro expression. J Proteome Res. 2005. May-Jun;4(3):889–899. doi: 10.1021/pr0500105 [DOI] [PubMed] [Google Scholar]
  • [68].Liu H, Anders F, Funke S, et al. Proteome alterations in aqueous humour of primary open angle glaucoma patients. Int J Ophthalmol. 2020;13(1):176–179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [69].Medeiros FA, Varma A, Jammal AA, et al. Accelerated epigenetic aging is associated with faster glaucoma progression: a DNA methylation study. MedRxiv. 2024. Oct 12:2024.10.12.24315371. [DOI] [PMC free article] [PubMed]
  • [70].Fu DJ, Ademisoye E, Shih V, et al. Burden of glaucoma in the United Kingdom: a multicenter analysis of United Kingdom glaucoma services. Ophthalmol Glaucoma. 2023. Jan-Feb;6(1):106–115. doi: 10.1016/j.ogla.2022.08.007 [DOI] [PubMed] [Google Scholar]
  • [71].McDonnell F, Irnaten M, Clark AF, et al. Hypoxia-induced changes in DNA methylation alter RASAL1 and TGFbeta1 expression in human trabecular meshwork cells. PLOS ONE. 2016;11(4):e0153354. doi: 10.1371/journal.pone.0153354 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [72].Cai J, Drewry MD, Perkumas K, et al. Differential DNA methylation patterns in human Schlemm’s canal endothelial cells with glaucoma. Mol Vis. 2020;26:483–493. [PMC free article] [PubMed] [Google Scholar]
  • [73].Berdasco M, Gomez A, Rubio MJ, et al. DNA methylomes reveal biological networks involved in human eye development, functions and associated disorders. Sci Rep. 2017. Sep 18;7(1):11762. doi: 10.1038/s41598-017-12084-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [74].Yin Y, Morgunova E, Jolma A, et al. Impact of cytosine methylation on DNA binding specificities of human transcription factors. Science. 2017. May 5;356(6337). doi: 10.1126/science.aaj2239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [75].Stricker SH, Koferle A, Beck S. From profiles to function in epigenomics. Nat Rev Genet. 2017. Jan;18(1):51–66. doi: 10.1038/nrg.2016.138 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
KEPI_A_2566496_SM5536.zip (5.1MB, zip)

Data Availability Statement

Data are available from the Canadian Longitudinal Study on Aging (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data. Twins UK data can be requested at http://twinsuk.ac.uk. The Health and Retirement Study is a publicly available data set available through the University of Michigan at http://hrsonline.isr.umich.edu/.

Articles from Epigenetics are provided here courtesy of Taylor & Francis

RESOURCES