Genetic risk of glaucoma is associated with vascular and retinal nerve fibre wedge defects

To evaluate the association between localised vascular and retinal nerve fibre layer (RNFL) loss and genetic risk for glaucoma and cardiovascular disease using polygenic risk scores (PRS).


| I N T RODUC T ION
Glaucoma refers to a group of optic neuropathies for which the aetiology is complex.While elevated intraocular pressure (IOP) has been recognised as causing mechanical damage to the optic nerve head, it is not present in all glaucoma patients and does not necessarily cause glaucomatous progression (Casson et al., 2012;Weinreb et al., 2014).Alternatively, the vascular theory is a non-IOP-driven explanation for glaucomatous loss as a consequence of microvascular dysfunction and insufficient ocular blood flow (Flammer, 1994).In support of this, large cohort population studies have identified cardiovascular risk factors with increased risk of glaucoma (Bonomi et al., 2000;Tielsch et al., 1995).
Optical coherence tomography angiography (OCTA) enables non-invasive, high-resolution visualisation of the retinal vasculature, particularly at the inner retina.OCTA has become a focus of glaucoma research, and vessel density parameters have been shown to be altered in glaucoma, yet there have been mixed reports of diagnostic ability in comparison to the routinely used structural OCT measurements, including retinal nerve fibre layer (RNFL) thickness and progression (Hou et al., 2019;Rabiolo et al., 2021;Rao et al., 2017;Triolo et al., 2017;Yarmohammadi et al., 2016).Recently, the focus has extended to exploring focal patterns of vascular loss in ocular diseases, including foveal avascular zone enlargement and irregularity (Choi et al., 2017;Kwon et al., 2018), and localised wedge defects (LeTran et al., 2022;Saks et al., 2022).We have previously reported wedge-shaped vascular defects at a prevalence of approximately 16% in suspect and early primary openangle glaucoma (POAG) eyes, with strong concordance to localised RNFL wedge defects (Saks et al., 2022).
It is unclear why some individuals develop localised well-defined vascular wedge defects while others do not, and whether one feature is secondary to the other.It is also not known whether the presence of these defects in early disease indicates a likelihood for progression to manifest glaucoma.To explore the link with known glaucoma-related genes, we investigated subjects for the presence of wedge defects with a pre-established polygenic risk score (PRS) for glaucoma (Craig et al., 2020).Moreover, to determine whether the presence of wedge defects indicated a greater risk of systemic vascular disease, as opposed to being glaucoma-specific, we investigated their presence with a PRS for cardiovascular disease (Inouye et al., 2018).Finally, we assessed rates of structural progression and RNFL wedge defects in those with and without vascular changes, and assessed the inter-relationship with optic disc haemorrhages, as these are an established risk factor for glaucoma progression (Leske et al., 2007).As per the author's knowledge, this is the first instance of investigation of vascular patterns of loss in both early manifest glaucoma and glaucoma suspects, with a polygenic risk assessment.

| M ET HOD S
This study was approved by the Macquarie University Human Research Ethics Committee and Southern Adelaide Clinical Research Ethics Committee.Written informed consent was obtained from all participants, and all study procedures adhered to the tenets of the Declaration of Helsinki.

| Participants
This study includes 858 eyes from 455 individuals enrolled in the Progression Risk Of Glaucoma: RElevant SNPs (single nucleotide polymorphisms) of Significant Association (PROGRESSA) study.PROGRESSA is an ongoing study of glaucoma suspects and early manifest glaucoma participants with up to 10 years follow up data.All participants were aged 18 years and older.Participants were diagnosed as glaucoma suspects (normal visual field with optic disc suggestive of glaucoma as determined using the Disc Damage Likelihood Scale (DDLS) (Bayer et al., 2002) with/without glaucoma family history) or ocular hypertensive (DDLS grade ≤1 and IOP ≥22 mmHg on two or more occasions), or early manifest glaucoma (abnormal disc with Stage 1 visual field defect; Mills et al., 2006).
Individuals were included in this study if they had valid genetic data and usable (within the quality threshold and no prominent artefact obscuring identification of wedge defects) OCTA imaging.Within the 455 individuals, 52 eyes were excluded due to lack of OCTA or poor-quality OCTA.
All individuals underwent full clinical and ophthalmic evaluation at baseline and at subsequent visits, as per previous report (Saks et al., 2022); data were included from the visit where the vascular wedge defect was found.The presence of disc haemorrhage (DH) refers to any previous or current visit.Family history of glaucoma, and history of hypertension, diabetes, heart attack or cerebrovascular event (stroke or transient ischemic attack) were self-reported, and positive record refers to at any visit.Myopia was indirectly measured using refractive error, expressed as spherical equivalent (SE).High myopes were not excluded (11 eyes had SE ≤ −6D).Eyes that had received previous trabeculectomy/trabeculotomy surgery were not excluded (17 eyes).

| Vascular phenotyping
This study focuses on the presence or absence of focal vascular wedge defects.The presence of vascular wedge defects was confirmed by two masked investigators (Saks and Schulz) by scrutiny of macula and optic nerve head OCTA scans, with an interobserver agreement of 96%.Where disagreement occurred, the investigators reviewed the cases together to reach consensus.Wedge defects were identified as being a darkened region, extending out from the optic nerve head towards the macula in an arcuate manner on the Spectralis OCTA macula (Heidelberg Engineering, Germany) and Angio-Plex optic nerve head and macula (Carl Zeiss Meditec Pty Ltd) scans (Saks et al., 2022).An example of an individual with vascular and RNFL wedges can be seen in Figure 1.Wedge defects were documented as present or absent but were not quantified in terms of extent or area (Saks et al., 2022) for this study due to the fact that two different OCTA devices (see below) had been used at different sites, introducing inter-device variability (Li et al., 2018).All visits were retrospectively analysed to determine whether a vascular wedge defect arose throughout observation.
OCTA scans were included from both AngioPlex and Spectralis systems.Scans were exported at the superficial vascular complex level (inner limiting membrane to inner plexiform layer).AngioPlex optic disc and macula 6 × 6 mm square scans of signal strength ≥7 were included.Spectralis macula scans of 8.7 × 4.4 mm were exported with projection artefact removal turned on, contrast auto set (1:4) and quality score of ≥25.
Scans from both systems were inspected for obvious artefacts, such as large floaters and prominent scan lines, which impacted visualisation, and were excluded if the errors were within the designated analysis area.

| Structural phenotyping
Defects were also investigated in the RNFL using the Cirrus OCT (Carl Zeiss, Meditec) Optic Disc Cube 200 × 200 protocol with a 3.46 mm diameter centred around the optic disc.Scans without obvious segmentation errors or artefacts within the analysis area and with signal strength ≥7 were included.We investigated in a subset of 90 subjects a qualitative assessment for the presence of RNFL wedge-shaped defects but found a semiquantitative approach to be more reliable (see discussion).RNFL wedge defects were therefore defined as having either 16 or more red (significant) pixels along the temporal superior or inferior RNFL bundle trajectories on the RNFL Deviation Map and/or significantly flagged superotemporal or inferotemporal RNFL clock hours thicknesses whereby superotemporal was defined by hours 10 and 11, and inferotemporal by hours 6 and 7 (right eye), and as hours 1 and 2, and 4 and 5, respectively for the left eye.Sectoral thickness loss within the temporal superior and inferior regions was also explored using the RNFL clock hours thickness chart.22 eyes had RNFL data excluded due to poor scan quality, no OCT acquired on the date of the OCTA, or OCT acquired with different systems (Spectralis, Heidelberg) which could not be included in the thickness quantification for the reliability of data (Saks et al., 2020).OCT data were included from the same visit as the OCTA.
Global RNFL progression rates up till the date of the included OCTA visit were obtained from the Cirrus OCT Guided Progression Analysis tool (CarlZeiss, 2016) in eyes which had at a minimum of five scans from at least 2 years follow-up, and from the Spectralis RNFL 12° circle scan automated progression (n = 836).Eyes with less than the required number of visits were excluded from the analysis.Our previous research established that while RNFL thickness was not comparable between the two OCT devices, the rates of RNFL progression were comparable (Saks et al. 2020).

| Polygenic risk scores
Two PRS derivations were used in this study: a multitrait analysis for glaucoma PRS (Craig et al., 2020) and a cardiovascular PRS (Inouye et al., 2018;Marshall et al., 2022).Briefly, the multi-trait glaucoma PRS was developed from a glaucoma genome-wide association study (7947 glaucoma cases, 119 318 controls) combined with vertical cup-to-disc ratio (VCDR) data from 67 040 United Kingdom Biobank and 23 899 International Glaucoma Genetics Consortium participants, as well as IOP data from 103 914 United Kingdom Biobank and 29 578 International Glaucoma Genetics Consortium participants.It was then validated using cohorts of early and advanced glaucoma and the Blue Mountains Eye Study (Craig et al., 2020).The cardiovascular disease PRS was developed from a meta-analysis of 1.7 million genetic variants for coronary artery disease with validation in 22 242 disease cases and 460 387 non-disease cases from the United Kingdom Biobank (Inouye et al., 2018).Using this score, a per-allele weighted PRS for cardiovascular disease was then calculated for the PROGRESSA cohort, replicated in the Australia and New Zealand Registry of Advanced Glaucoma Study cohort and normalised using a control cohort from the QSkin Sun and Health Study (Marshall et al., 2022).Genotyping for the current study cohort was performed using Human-CoreExome arrays (Illumina) as previously described (MacGregor et al., 2018).

| Statistical analysis
Statistical analysis was completed using the Statistical Package for Social Sciences (SPSS; Windows, version 28).All analyses were two-tailed with significance at alpha = 0.05.Descriptive statistics were calculated as mean and standard deviation.Shapiro-Wilk tests revealed non-normal distribution for all continuous variables, as such non-parametric tests were used.Demographic characteristics were analysed by Mann-Whitney U tests (continuous variables) and Chi-square tests (categorical).Correlations were bivariate with Spearman's rho reported.Linear, unstructured, generalised estimating equations (GEE) models were used for analysis of main variables.Within the model, the use of multiple eyes from participants was controlled for as a within-subjects variable, and the model was consistently adjusted for predictors including sex (factor) and age (covariate).The significance of each parameter was evaluated by main effects.PRS values were input as normalised z scores.

| R E SU LT S
Of the 858 eyes included in this study, 495 were diagnosed as glaucoma suspects, 291 had manifest POAG, and 72 were ocular hypertensive.There were 187 eyes found to have vascular wedge defects (22%), which were identified in each diagnostic group, i.e., 42 suspect eyes, 144 manifest POAG but only one ocular hypertensive.The glaucoma PRS (z score) was significantly higher in individuals with manifest glaucoma compared to glaucoma suspects, which included the ocular hypertensives (p = 0.027, mean (standard deviation) = 0.56 (0.95) and 0.41 (0.93), respectively).There was no significant difference in the cardiovascular PRS between manifest and suspect glaucoma eyes.Individuals in this study were predominantly Caucasian (n = 429), with the remaining ethnicities including Asian (n = 5), Middle Eastern (n = 1), Hispanic (n = 1), mixed ethnicity (n = 4), and 15 individuals did not have ethnicity recorded.The relevant demographic and ocular characteristics of these participants are summarised in Table 1.
Eyes with vascular wedge defects, compared to eyes without vascular defects, were statistically more likely to be older, with lower IOP, higher VCDR, lower central corneal thickness (CCT), worse visual acuity, more myopic, with worse mean deviation, greater history of DH and were more likely to be treated for glaucoma.All of b Data unavailable for 2 eyes with vascular wedge defects and 10 eyes without.
these are with known risks or features of glaucoma, except for the lower IOP.There was no difference in the gender of the groups, or in the proportion of family history of glaucoma, or systemic vascular conditions (hypertension, diabetes, heart attack or cerebrovascular event).As expected, there was a strong association between vascular wedge presence and RNFL wedge presence (r = 0.602 p < 0.001), with 162 eyes showing both vascular and RNFL wedges in similar distribution.Table 2 demonstrates whether the presence of RNFL wedge defects is associated with similar demographic and ocular characteristics to vascular wedges.
Similar to eyes with vascular wedge defects, eyes with RNFL wedge defects were statistically more likely to be older, with lower IOP, higher VCDR, lower CCT, worse visual acuity, worse mean deviation, greater history of DH and were more likely to be treated for glaucoma.They were also more likely to be male and have hypertension.There was no difference between those with and without RNFL wedge defects in spherical equivalent, the proportion of family history of glaucoma, or diabetes, heart attack or cerebrovascular event.RNFL wedge defects were identified across all stages of glaucoma progression including 94 suspect glaucoma eyes, 188 manifest and 6 ocular hypertensive eyes.
As shown in Table 3, vascular wedge presence was significantly associated with the glaucoma PRS (p < 0.001) but not the cardiovascular PRS (p = 0.828).Similarly, RNFL wedge presence was associated with the glaucoma PRS (p = 0.020) but not the cardiovascular PRS (p = 0460).
T A B L E 3 Association between wedge defect presence and polygenic risk for glaucoma and cardiovascular disease.Vascular wedge were predominantly in the inferotemporal region (88 eyes, 47%) with an additional 55 eyes (29%) showing wedge defects in both inferotemporal and superotemporal regions, and the remaining 44 in the superotemporal region (24%).

Characteristics
Figure 2 shows an example of multiple wedge defect presentation, as seen in an individual with early manifest glaucoma.Figure 3 demonstrates that eyes with one vascular wedge (n = 133) were more likely to have a significantly higher glaucoma PRS than eyes with no vascular wedge defects (CI: 0.38, 0.50; p = 0.020).Furthermore, eyes with two or more vascular wedges (n = 54) were even more likely to have a higher glaucoma PRS score than eyes with no vascular wedge defects (95% CI: 0.37, 0.50; p = 0.005).However, there was no statistically significant difference in the PRS in eyes with one vascular wedge compared to eyes with two or more vascular wedges (CI: 0.53, 0.80; p = 0.222).To determine whether there was a difference the risk of glaucoma in individuals with only one eye showing a vascular wedge or with both eyes showing vascular wedges, individuals with only one eye included (regardless of vascular wedge status) were excluded from the following subanalysis (67 eyes excluded).As seen in Figure 4, within the 397 individuals remaining, individuals with vascular wedge defects (in one or both eyes) had significantly higher PRS score than individuals with no vascular wedge defects.The presence of vascular wedges in both eyes was more statistically significant (95% CI: 0.36, 0.50; p = 0.006) than the presence of vascular wedge defects in only one eye (95% CI: 0.37, 0.51; p = 0.047), compared to no vascular wedges.
Cross-sectionally, there was a significant difference in sectoral thicknesses of the peripapillary RNFL (pRNFL) between eyes with and without vascular wedge defects whereby eyes with vascular wedge defects had approximately 31 μm thinner superotemporal pRNFL and approximately 43 μm thinner pRNFL inferotemporal sector thickness (both p < 0.001).
Eyes with vascular wedge defects were also associated with faster rates of global RNFL progression (average progression = −0.91 ± 0.89 μm/year) than eyes without vascular wedge defects (average progression = −0.45± 0.69 μm/year), even after controlling for age and gender of individuals (p ≤ 0.001).Similarly, eyes with RNFL wedge defects were associated with faster global RNFL progression (average progression = −0.81± 0.90 μm/year) than eyes without RNFL wedge defects (average progression = −0.43 ± 0.64 μm/year), which persisted after controlling for age and gender (p ≤ 0.001).In a model looking at global RNFL progression, which included both wedge defect types as well as age and sex, vascular wedge defects and RNFL wedge defects both had a significant association with RNFL progression (p ≤ 0.001 and p = 0.008, respectively).

| DI SC US SION
This study identified that a higher genetic risk for glaucoma as defined by the PRS was associated with a greater likelihood of vascular wedge defects being present.The study also showed that multiple vascular wedges per eye (Figure 3) and both eyes with vascular wedge defects (Figure 4) was associated with a further increased PRS for POAG.Furthermore, although there was a very strong concordance between vascular wedge defects and RNFL wedge defects (162 eyes had both types in similar distribution), some participants demonstrated vascular change without RNFL defects (21 eyes) and vice versa (126 eyes).Vascular wedge defects compared to RNFL wedge defects were associated with similar, yet faster, rates of glaucoma progression based on global RNFL loss.While vascular wedges and RNFL wedges are highly inter-related, and both are features of glaucoma, their pathophysiology is still in question.The lack of association with systemic cardiovascular PRS suggests that the vascular dropout could be a locally driven process.It could still be that vascular dropout is secondary to the RNFL loss, but these findings may support the existence of a vascular glaucoma phenotype in some cases, while others manifest RNFL loss with no change in vasculature.
The glaucoma PRS has been shown to aid in identifying individuals who are likely to develop progressive glaucoma (Craig et al., 2020), and this study indicates that those individuals may be more likely to present with a vascular phenotype.Identifying this predisposition early may influence treatment and clinical decisions and demonstrates a potential advantage of incorporating OCTA monitoring into clinical care.
Both vascular wedges and RNFL wedge defects were more associated with older age, lower IOP, higher VCDR, lower CCT, lower visual acuity, worse mean deviation and receiving glaucoma treatment.Although this may suggest that vascular wedge defects are a feature of progressed glaucoma, they also appeared in suspect eyes, as well as early manifest glaucoma eyes, often prior to any major functional damage (eyes with vascular wedge mean deviation on average = −3.61dB, eyes with RNFL wedge mean deviation on average = −3.04dB).Localised RNFL wedge defects are commonly reported as a feature of normal tension glaucoma (Jonas Schiro, Suh et al., 2010), which may explain why both vascular and RNFL wedge defects were associated with lower IOP than eyes without localised defects.However, it is also possible that individuals with defects had lower IOP as a consequence of receiving more aggressive IOP-lowering treatment.This is likely since many patients are on treatment, so those with more clinically apparent disc damage were receiving tighter IOP control.
Vascular wedge defects are strongly associated with RNFL wedge defects, yet they are not always co-located.The only demographic or ophthalmic features which differed between eyes with and without RNFL wedge defects as compared to eyes with and without vascular wedge defects were hypertension, gender, and spherical equivalent.Localised RNFL defects have been previously associated with hypertension (Jung et al., 2018;Xu et al., 2013) and cerebrovascular events (Jung et al., 2018;Wang et al., 2014).We did not find an association with a history of cerebrovascular event with either defect, potentially due to the small percentage of individuals who had experienced this (4% of total eyes), and the reliance on self-reported systemic health data.However, we did find that hypertension was more common in eyes with RNFL wedge defects but was not statistically different in vascular wedge defects.Consistent with our study, LeTran et al. also found no associations between systemic vascular disease and vascular wedge defects (Le-Tran et al., 2022).One might have expected these defects should relate to systemic vascular disease, but this is not supported by the data.Although it is unclear why there may be a gender bias for RNFL wedge defects, a large Korean population-based study also found a greater propensity for RNFL defects in men as compared to women (Na et al., 2017).Further research is needed to understand this difference and why it exists in RNFL defects but not vascular defects.Spherical equivalent was significantly lower (more myopic) in eyes with vascular wedge defects compared to those without but was not specifically related to eyes with/without RNFL wedge defects, even though myopia is associated with glaucomatous loss.Although the above may suggest vascular wedges are associated with myopia, the spherical equivalent for eyes with and without vascular wedge defects was predominantly within the emmetropic range.This study did not include a large number of myopic eyes (77% eyes SE ≥ −1D, 9% eyes SE ≤ 3 indicating moderate or high myopia), suggesting that although statistically significant, in this study myopia may not be a clinically relevant factor in wedge development.
History of DH was determined to be more likely in eyes with both vascular wedge defects and RNFL wedge defects than in eyes without either.DH is strongly linked with localised RNFL wedge defects (Jonas & Schiro, 1994) and has been shown to predict the location of future RNFL defects in ocular hypertension (Airaksinen et al., 1981).A strong relationship has also been reported between DH history and vascular wedge defects (LeTran et al., 2022;Saks et al., 2022).Since DH is an established factor in glaucoma progression (Kim & Park, 2017), it is unsurprising that DH strongly correlated with genetic risk for glaucoma.However, perhaps surprisingly, DH did not correlate with the cardiovascular PRS.This supports the notion that DH may be related to focal vascular abnormalities such as microinfarctions or small branch vein occlusions, or due to disruption of supporting lamina connective tissue around the optic disc (Drance, 1989;Quigley et al., 1981).The inclusion of DH in the multiple regression model of vascular wedge defects and the glaucoma PRS, suggested that DH had an effect on this relationship, yet the presence of vascular wedge defect was a feature of being at highrisk for glaucoma independent of DH.However, the inclusion in the multiple regression model with RNFL defects reduced the interaction between RNFL defects and PRS to approaching significance.It should be noted that DH may have occurred between visits and so may be under-reported.It appears as though vascular wedge defects may be directly or indirectly, via DH, related to a localised ischemic event; however, further research into the relationship between DH and localised vessel density loss may help explain the vascular wedge defect phenotype.
RNFL defects were associated with the glaucoma PRS, which is unsurprising given that thinning of the RNFL is an early feature of glaucoma progression (Airaksinen et al., 1984;Medeiros et al., 2012;Miki et al., 2014).The glaucoma PRS has been associated with the rate of thinning within the most affected RNFL quadrant in early glaucoma (Craig et al., 2020).In a recent study of healthy individuals from the United Kingdom Biobank, there was minimal overlap between genes that influence inner retinal (RNFL and ganglion cell inner plexiform layer) thickness and those which have been associated with glaucoma (Currant et al., 2021).Furthermore, IOP was shown to have a genetic causal relationship with glaucoma; however, inner retinal thickness was not.Since inner retinal thickness was a cross-sectional measure, it may be possible that change in inner retinal thickness has a stronger relationship with glaucoma genetic risk.It may also be possible that a peripapillary measure of RNFL would have been more sensitive to glaucomatous change than the macular outcome used in the study of Currant et al.This is supported by the current study in that RNFL wedge defects, which are more prominent at the peripapillary region, were associated with genetic risk of glaucoma.
This study considered the presence or absence of vascular and RNFL wedge defects; however, it did not examine the progression of these defects over time.Localised RNFL defects have been shown to most commonly progress towards the macula and deepen (Suh et al., 2010).Future studies should investigate the progression of both vascular and RNFL wedge defects to determine whether they progress concurrently or in a heterogenous manner to each other.The question still remains whether the vascular dropout is secondary to reduced demand, or the primary factor in ganglion cell dropout.
RNFL wedge defects were identified by a predominantly automated process involving comparison with normative values within the OCT software in order to minimise potential subjectivity and human bias, yet since there was no automated software process available for vascular images at the time of analysis, vascular wedge defects identified by expert review of face images.Although identifying RNFL wedge defects by this same en-face review may have been a more appropriate comparison to reduce potential noise from these different techniques, this would have been a less objective measure thus introducing additional biases.The similarity of RNFL wedge identification using a subjective approach and the automated approach was assessed, and found to be 93% similar, in a sub-study of 90, 10% of, cases.Therefore, given the quantitative approach was more repeatable and less subjective, this methodology was chosen.It is possible that one technique may be more accurate at identifying wedge presence, consequently, RNFL and vascular wedge comparison should be interpreted with clinical caution as image acquisition, observation and analysis differences may impact these variables rendering biases.
While there is inter-device variability that limits direct comparison of structural thickness, vessel densities, and quantification of defects between devices, the rates of RNFL progression (Saks et al., 2020), and vessel visibility (Li et al., 2018) have been shown to be comparable between the Spectralis and Zeiss OCT and OCTA devices.However, this limitation restricted the vascular wedge defect analysis to presence or absence of defect, and the RNFL wedge defect analysis to one device -the Cirrus OCT.It is possible that by being inconsistent in the devices included in each analysis, this introduced variability of unknown weighting.Future studies should limit their data collection to one device in order to minimise this risk.This would also enable exploration of the characteristics of wedge defects such as area and vessel density within vascular defects, and the nature of their progression, to determine if there are any associations with the risk of glaucoma.
Genetic risk of cardiovascular disease has been proposed to be more associated with a predominantly macular ganglion cell layer/inner plexiform layer pattern of glaucomatous loss as compared to predominantly pRNFL or an equivalent combination of the two (Marshall et al., 2022).This could explain why eyes with wedge defects, both vascular and RNFL, did not show an association with increased risk of cardiovascular disease (Table 3).
This study is limited in that it relies on PRS scores that were validated for European ancestry.Although this matches most of our cohort, we have not excluded individuals based on ethnicity in attempt to maximise generalisability of results, yet this may influence results.Validation in other ethnicities is required (Cooke Bailey et al., 2023).
The current study showed that individuals with localised vascular and RNFL wedge defects, in particularly multiple vascular wedge defects per eye, have a higher polygenic risk score for glaucoma, and may be at higher risk for glaucoma progression.The pathophysiological cause for the vascular dropout remains to be determined.Individuals with higher genetic risk of glaucoma based on the PRS were more likely to have retinal vascular defects, as well as structural glaucomatous loss, but this did not relate to systemic cardiovascular risk.The study demonstrates that in at least some of the glaucoma population there may be a more vascular phenotype.

F
Example of inferotemporal vascular and retinal nerve fibre layer wedge defect in an individual with manifest glaucoma.(a) AngioPlex optic nerve head and (b) macula scans of vascular wedge.(c) Optic disc cube scan showing significant (red pixel) loss in the inferotemporal region, and borderline significant superotemporal loss; reflected in the clock hour chart (d).Scans acquired at Flinders Medical Centre, Adelaide.

F
Example of Spectralis optical coherence tomography (OCT) and angiography (OCTA) scans of an individual with early manifest glaucoma, showing two vascular wedge defects within the same eye, one superotemporal and the other inferior/inferotemporal.Clockwise from left: optic nerve head scan OCTA scan, OCT posterior pole deviation map of the RNFL and macula angiography scan.Images acquired at the Macquarie Ophthalmology Clinic, Macquarie University Hospital, Sydney.F I G U R E 3 Association between glaucoma PRS (represented by z score) and number of vascular wedges per eye.

F
Association between glaucoma PRS (represented by z score) and number of eyes with vascular wedge per individual.
AC K NOW L E DGM E N T S Open access publishing facilitated by Macquarie University, as part of the Wiley -Macquarie University agreement via the Council of Australian University Librarians.F U N DI NG I N FOR M AT ION Macquarie University, Grant/Award Numbers: 2018221, 20191448; National Health and Medical Research Council, Grant/Award Numbers: 1048037, 1150144, 1154543.ORC I D Danit G. Saks https://orcid.org/0000-0001-5488-1884R E F E R E NC E S

With vascular wedge defects (n = 187 eyes) Without vascular wedge defects (n = 671 eyes) p (with v without vascular wedge)
Demographic and ocular characteristics of eyes with and without vascular wedge defects.
T A B L E 1Note: Demographic and ocular characteristics of eyes with and without vascular wedge defects.Data are mean (standard deviation) unless otherwise specified.Continuous variable p values calculated by Mann-Whitney U tests, nominal values by Chi-square tests.* indicates significant p value.Abbreviations: BCVA, best corrected visual acuity; CCT, central corneal thickness; DH, disc haemorrhage; IOP, intraocular pressure; PRS, polygenic risk score (z score); VCDR, vertical cup/disc ratio.a Data unavailable for 3 eyes with vascular wedge defects and 26 eyes without.

With RNFL wedge defects (n = 288 eyes) Without RNFL wedge defects (n = 548 eyes) p (with v without RNFL wedge)
Demographic and ocular characteristics of eyes with and without RNFL wedge defects.
T A B L E 2 Note: Demographic and ocular characteristics of eyes with and without RNFL wedge defects.Data are mean (standard deviation) unless otherwise specified.Continuous variable p values calculated by Mann-Whitney U tests, nominal values by Chi-square tests.*indicates significant p value.Abbreviations: BCVA, best corrected visual acuity; CCT, central corneal thickness; DH, disc haemorrhage; IOP, intraocular pressure; PRS, polygenic risk score (z score); VCDR vertical cup/disc ratio.a Data unavailable for 5 eyes with RNFL wedge defects and 7 without.