Ophthalmology (Eye & Ear Hospital) - Theses

Permanent URI for this collection

Search Results

Now showing 1 - 7 of 7
  • Item
    Thumbnail Image
    Predicting progression in age-related macular degeneration
    Goh, Kai Lyn ( 2023-08)
    Background: Predicting which individuals will develop vision-threatening complications of age-related macular degeneration (AMD) is a challenging task, as traditional models based on colour fundus photography (CFP) correctly identify less than half of those who subsequently develop late AMD at 95% specificity. Thus, there is a great need to improve risk stratification for individuals with the early stages of AMD. Aims: To examine the prognostic significance of novel pathological characteristics related to drusen phenotypes and pigmentary abnormalities in the early stages of AMD. Methods: Retinal imaging data from 280 eyes from 140 individuals with bilateral large drusen enrolled in a longitudinal observational study was evaluated. Individuals underwent multimodal imaging (MMI) and microperimetry at baseline and then 6-monthly for up to 3-years. Disease progression was primarily evaluated based on the development of MMI-defined late AMD, and secondarily based on the rate of visual sensitivity decline on microperimetry prior to late AMD development. Four retinal specialists assessed the likelihood that each eye at baseline would progress with CFP, and then with MMI, to determine if MMI improves their ability to predict late AMD development. Baseline images were assessed for the: (i) presence of cuticular drusen, and extent of (ii) hyporeflective cores within drusen (HCD), (iii) hyperpigmentary abnormalities (HPAs), and (iv) hyperreflective foci (HRF) that do not spatially correspond to HPAs [HRF(OCT+/CFP-)]. The association with progression and impact on visual sensitivity of each feature was examined, including adjustments for well-established risk factors for progression (drusen volume from optical coherence tomography, presence of pigmentary abnormalities on CFP, and age). Results: The prediction of late AMD development by retinal specialists was improved when using MMI compared to CFP. However, a basic prediction model (age, presence of pigmentary abnormalities, and drusen volume) outperformed clinicians. In this cohort, neither the presence of cuticular drusen nor extent of HCD were significantly associated with an increased rate of progression to late AMD, reduced mean visual sensitivity at baseline, or an increased rate of visual sensitivity decline, after adjusting for well-established risk factors of progression. The quantification of HPA extent did not significantly improve the prediction of late AMD development compared to HPA presence, and the addition of HRF(OCT+/CFP-) extent to HPA extent also did not improve performance. Both HPA and HRF(OCT+/CFP-) extent were independently associated with reduced sector-based visual sensitivity, with the latter also associated with a significantly faster rate of visual sensitivity decline. Conclusions: Accounting for drusen phenotypes such as cuticular drusen and HCD, or the quantity of HPAs and HRF(OCT+/CFP-), did not significantly improve the prediction of late AMD development above what could be achieved by well-established risk factors. However, a basic prediction model using these parameters – drusen volume, presence of pigmentary abnormalities and age – outperformed retinal specialists, suggesting that such a model could improve counselling and monitoring of individuals in clinical practice. Such a model could also be used to better identify an enriched cohort to improve feasibility of future interventional trials, and thus help expedite the discovery of preventative treatments in the early stages of AMD.
  • Item
    No Preview Available
    Application of artificial intelligence to analysis and progression of geographic atrophy in age-related macular degeneration
    Arslan, Janan ( 2021)
    Background: Geographic atrophy (GA) is a debilitating eye disease affecting adults 50 years and older and is the late stage of age-related macular degeneration (AMD). It is characterised by the presence of lesions in the retina, which are areas of cellular death that cause irreversible vision loss. The dark lesions are usually surrounded by bright regions of hyperfluorescence. GA is a progressive condition with no therapeutic treatments available to arrest or slow down its course. GA is highly variable in its progression, with some patients progressing faster to vision loss than others. While a plethora of publications are available on the investigation of this disease, there remains an uncertainty with regards to our understanding of the disease, its manifestation and progression. Several clinical and non-clinical features along with different statistical modelling techniques have been explored to date. However, there is a lack of consensus regarding the most appropriate model or features that are thought to be useful in a real-time clinical setting. Current available tools only annotate GA lesions using the semi-automated region-growing algorithm. In addition to its semi-automation (which still requires manual and time-consuming work from clinical graders), the current tool can only retrospectively assess progression based on all available data; the software does not have predictive capabilities. Ideally, being able to predict the disease state and progression of GA at initial consultation would be of immense clinical benefit to both the ophthalmologist and patient. Artificial intelligence (AI) has been suggested over the past decade to having the capabilities to achieve such a goal. However, the publications in the GA-AI space thus far have predominantly focused on the automation of the primary GA feature of interest, lesions, with minor focus on GA progression. The automation and detection of hyperfluorescence has been neglected altogether. Thus, there is great scope to apply AI to understand GA manifestation and progression. Purpose and Aims: This thesis describes research to automate the diagnosis of GA, as well as understand the patterns of GA progression using a combination of AI, image processing, mathematics, statistics and computer science. The aims of the thesis were to (1) automate the detection of lesions, but with greater accuracy and efficiency as compared to the current work in the literature, while simultaneously automating hyperfluorescence areas for the first time ever to offer ophthalmologists complete automation of GA features, (2) investigate patterns of GA progression by evaluating prospective regression models by applying our current understanding of the clinical and physical assumptions of GA growth, (3) understand the impact of epistemic uncertainty – a metric to measure variability due to incomplete knowledge of a disease or process – in GA modelling, (4) quantify and identify subgroups of GA lesions and hyperfluorescence areas using machine learning (ML) techniques in order to explain whether varying shapes and sizes of GA features could explain varying progression rates, (5) extract and rank relevant imaging features using ML and develop a statistical model for GA growth, and (6) present a preliminary software platform which can be extended beyond the scope of this thesis. The Methods and Results achieved for the Aims are summarised below. Methods: Data Subjects included in this thesis were AMD participants diagnosed with GA. Data were collected from macular studies from the Centre for Eye Research Australia (CERA) and from a private ophthalmology practice. Extracted data included basic demographic information, such as age and sex, and the imaging modality fundus autofluorescence (FAF) – which produces a grayscale image that enhances the detection of lesions and hyperfluorescence regions. Ground truth labels were also annotated by two graders, and the intraclass correlation coefficient (ICC) was used to measure consistency between graders. Aim 1 Lesion automation included a pipeline of image pre-processing (e.g., contrast limited adaptive histogram equalisation) and semantic segmentation using the U-Net architecture and ground truth labels. Metrics used for assessing lesion automation included the Dice similarity coefficient (DSC), specificity, sensitivity, mean absolute error (MAE), accuracy, and precision. The automation of the hyperfluorescent regions were conducted using pseudocolouring techniques. This involved the application of a colour palette to the FAF images (i.e., specifically the JET colour map), which highlighted the changing intensities in hyperfluorescent regions. Hyperfluorescence was automatically extracted by specifying the colour ranges which covered hyperfluorescent regions. For hyperfluorescence automation, no ground truth labels were available, and thus qualitative assessments were conducted. Aims 2 & 3 Patterns of GA growth – the cumulative increase of total GA surface area within the retina over time – were investigated by evaluating a range of regression models: linear, logarithmic, power, exponential, quadratic, and quadratic without linear term. The physical and clinical assumptions of growth were included as part of the analyses (i.e., growth cannot be infinite; there is finite retinal area in which the lesion growth will hit and then the growth curve should plateau). For Aim 3, the prospective models were evaluated using epistemic uncertainty, particularly model structure uncertainty. Uncertainty was quantified as U=1-r2 where r2=1-SSR/SSO and SSR/SSO is the sum of square residuals divided by the total sum of squares in the data. The metric U is the proportion of total unexplained variability not accounted for by the regression model. The smaller the U, the better the fit of a model. Aim 4 Prospective patterns and groupings of lesions and hyperfluorescent areas were elucidated using several unsupervised clustering methods. Cluster performance was measured using Silhouette coefficient (SC), Davies-Bouldin Index (DBI), and Calinski-Harabasz Index (CHI). If appropriate groups were identified, nomenclature was assigned to these groups based on the patterns seen within the cluster. Aim 5 Feature extraction and ranking was conducted using the ML algorithm XGBoost. An image-based linear mixed-effects model was designed to account for slope change based on within-subject variability and inter-eye correlation. Metrics used to assess the linear mixed-effects model included marginal and condition R2 (RM2 and RC2), Pearson’s correlation coefficient (r), root mean square error (RMSE), mean error (ME), MAE, mean absolute deviation (MAD), Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and log-likelihood. Aim 6 Aim 6 involved the design of a software that will be built upon further but is beyond the scope of this thesis. This software was built using Python and PyQT5. Results: Data The final dataset consisted of 702 images from 51 patients. The cohort consisted of 99 eyes, 49 left eyes (49.5%) and 50 right eyes (50.5%). A total of 359 images were for the left eye and 343 images were for the right eye. The cohort consisted of 38 females (74.5%) and 13 males (25.5%) with an average age of 76.7 +/- 8.9 years. Total follow-up time was 61.5 +/- 25.3 months. For manual annotation of ground truth labels, the ICC for consistency between the two graders was 0.9855 (95% CI: 0.9298, 0.9971). Aim 1 For lesion automation, the DSC was 0.9780 +/- 0.0124, sensitivity was 0.9903 +/- 0.0041, specificity was 0.7498 +/- 0.0955, MAE was 0.0376 +/- 0.0184, accuracy was 0.9774 +/- 0.0090, and precision was 0.9837 +/- 0.0116. Assessments of hyperfluorescence areas revealed three distinct regions of hyperfluorescence. They have been named as: early-stage hyperfluorescence, intermediate-stage hyperfluorescence, and late-stage hyperfluorescence. Qualitatively, both the lesion and hyperfluorescence automations produced visually accurate outcomes. The automations ran 10 times faster than the current semi-automated segmentation. Aims 2 & 3 The linear regression model was identified as most representative of peak GA growth. It had the lowest average uncertainty (U = 0.025), highest average coefficient-of-determination (R2 = 0.92), and applicability of the model was supported by a high correlation coefficient, r, with statistical significance (P = 0.01). Given the assumptions of GA growth (i.e., plateau due to limited retinal space) and the results confirming peak growth was linear, a hypothesis was made that GA growth overall followed a sigmoidal growth curve from disease onset to plateau. Using records from existing patients, the sigmoidal hypothesis was tested in case studies with sufficient longitudinal data from clinical follow-ups. The sigmoidal growth curve revealed a closer fit in these rare cases, and it is recommended that future GA progression studies include additional data to further validate the lower- and upper-tail ends of the sigmoidal function. Aim 4 GA lesions, together with early-, intermediate-, and late-stage hyperfluorescence areas were subject to cluster analysis using the method of k-Means. Meaningful clusters were identified for lesions, early- and late-stage hyperfluorescence regions, with each GA feature having k=3 clusters. For lesions, SC = 0.799, DBI = 0.180, and CHI = 4313.316. For early-stage hyperfluorescence, SC = 0.597, DBI = 0.915, and CHI = 186.989. For late-stage hyperfluorescence, SC = 0.593, DBI = 1.013, and CHI = 217.325. No meaningful clusters were identified for intermediate-stage hyperfluorescence. Aim 5 A linear mixed-effects model with 15 FAF imaging features produced average RM2 = 0.84 +/- 0.01, RC2 = 0.95 +/- 0.002, r = 0.97 +/- 0.006, RMSE = 1.44 +/- 0.06, ME = 0.09 +/- 0.15, MAE = 1.04 +/- 0.06, MAD = 3.52 +/- 0.67, AIC = 1718.41 +/- 11.08, BIC = 1798.99 +/- 11.20, and log-likelihood = -839.21 +/- 5.54. The model was also executed using the popularised square root and log(Yi+1) transformations, however, the original scaled GA total area using mm2/year produced the best results. Aim 6 A preliminary and user-friendly software was designed, which allows the end-user to select a baseline FAF image and then automatically segment lesion, early-, intermediate-, and late-stage hyperfluorescence regions from the image. Then, if the user wishes to toggle from image to image with all features being automatically updated, they simply need to select the ‘Next’ button rather than reloading a new image. A demonstration video can be found at https://youtu.be/WidWf70MKX4. Conclusions: This thesis adopted a multidisciplinary approach combining AI, image processing, mathematics, statistics and computer science to automate GA feature extraction, ranking of GA features, understanding the pattern of GA growth, and the development of a preliminary GA growth model and software. There are many clinical impacts of this work. The automation of clinical processes enables clinicians to focus on their patients and not on the repetitive and time-consuming components of diagnostic and prognostic processes. Furthermore, the automation also guarantees that both clinicians and patients save considerable time in consultations. Additionally, by understanding the underlying patterns of GA growth, we could take more decisive steps to develop appropriate interventions to arrest the progression of the disease or cure it altogether. Future development of the software could extend to multimodal imaging, such as spectral domain optical coherence tomography images or colour fundus photographs. The addition of other predictors is possible, such as genetic heritability.
  • Item
    Thumbnail Image
    Topographic rod function in intermediate age-related macular degeneration
    Rose, Rose ( 2019)
    Background Age-related macular degeneration (AMD) is a complex multifactorial disease that affects the elderly. In the early stages of disease, dark adaptation problems are more significant for patients than visual acuity impairment. For decades, studies of rod function related to dark adaptation issues suggested that rod function could be useful as a functional marker for differentiating AMD severity and monitoring disease progression. However, there remains many unanswered questions about how best to investigate rod function in the early stages of AMD. Earlier studies were not able to phenotype AMD cases to the extent we can today as they relied only on colour fundus photographs to determine AMD subgroups. We now have the opportunity to look at different AMD phenotypes using multimodal imaging techniques. These advances have added clarity to the phenotyping of AMD, as high-risk features such as reticular pseudodrusen (RPD), hyperreflective foci (HRF), and nascent geographic atrophy (nGA) can now be identified. At the same time as the advances in detecting anatomical changes were made, instruments that could measure rod function in a more thorough way were improving. Previously, instruments measured rod function at only a single retinal location, or could only detect large losses in sensitivity due to a lack of dynamic range. Recent advances have seen perimeters that are able to measure rod function at multiple locations in the one setting and also have a larger dynamic range to detect subtle rod dysfunction. I have used one of these new tools in my studies. Aims Two-color dark-adapted chromatic perimeter (DACP) is a novel device, designed and manufactured in Melbourne, Australia. It was first used in 2015 when I commenced my PhD. This perimeter measures rod function at multiple locations with sufficiently large range of stimulus intensities to detect subtle changes seen in early AMD. My thesis aimed to investigate the ability of the DACP to reliably detect and record rod sensitivities. I then utilized this new perimeter to investigate the ability of rod function to act as a robust functional biomarker that could be used to determine AMD disease severity and to monitor disease progression, and to compare rod function in AMD cases to normal control participants. I was also able to separate my AMD cohort into those with or without RPD, a high-risk AMD phenotype, to investigate the impact of this phenotype on rod function. Both static and dynamic rod functional changes were recorded at baseline visits and then again at 6- and 12-months. The results of this study will contribute to our understanding of functional loss in AMD and help clinicians and researchers when designing research protocols aiming to evaluate rod functional impairment before vision loss.   Methods This study was conducted between 2015 and 2018 at the Macular Research Unit, Centre for Eye Research Australia. During that time, three main studies were completed. An initial assessment of test-retest reliability of the DACP was followed by a cross-sectional study comparing the severity of rod dysfunction based on point wise sensitivity (PWS) and point wise sensitivity difference (PWSD) of two color sensitivities. AMD cases were divided into an AMD group with traditional drusen only and an RPD group, and rod function was evaluated at multiple ring eccentricities within 24º of the central macula. A further 12-months evaluation of static (PWSD) and dynamic rod functional changes (rod intercept time (RIT) and rod recovery rate (RRR)) within the central 12º was conducted in participants with intermediate AMD with and without RPD as well as normal control eyes. Results The DACP did not have ceiling or floor issues during the intrasession and intersession testing, with a coefficient of repeatability of ± 5 dB. Intermediate AMD (iAMD) with the presence of RPD was found to be associated with poorer rod function compared with iAMD without RPD and normal control eyes, but only within the central macular 8º. Rod dysfunction was more significant if the eyes were prebleached before undertaking the sensitivity measurements. In a longitudinal study, dynamic rod functional changes, ie. RRR, was the only parameter that changed over 12 months, and only in the iAMD without RPD group. Change was found only at the peripheral 12º ring eccentricities. Conclusion The DACP offers a great opportunity to evaluate topographic rod function when eyesight is still normal. The findings from this study support that concept of rod function being impaired in the early stages of AMD. The presence of RPD in cases of iAMD were associated with further reductions in rod function. Measuring dynamic rod function offers a more sensitive functional marker in determining severity of AMD within the central 8 degrees of the macula, but changes over time were only detected at more peripheral locations. Longer follow-up will be beneficial to determine how rod function deteriorates over time and correlates with progression to vision loss.  
  • Item
    Thumbnail Image
    Novel clinical biomarkers of disease in early stages of age-related macular degeneration
    Wu, Zhichao ( 2014)
    Background: The lack of effective disease biomarkers in the early stages of age-related macular degeneration (AMD) is a major impediment to the ability to sensitively identify individuals at risk of progression to the late stages where vision is threatened, and thus also the development and evaluation of novel interventions that aim to slow or prevent such progression. Aims: To investigate and identify potential anatomical and functional biomarkers that can be effectively used for clinical studies in the early stages of AMD. Methods: Functional markers including low luminance visual acuity (LLVA), microperimetry and multifocal electroretinography (mfERG) were examined in participants with the early stages of AMD at cross-section and longitudinally over a 12-month period. Structural measures were then developed via identification on spectral-domain optical coherence tomography (SD-OCT), and their relationship with functional changes was examined. Results: The intrasession test-retest variability of microperimetry was improved by discarding the first examination for participants who had not previously performed microperimetry before, since it was subject to a learning effect. LLVA was on average no more effective than best-corrected visual acuity (BCVA), in detecting functional changes in the early stages of AMD, with microperimetry detecting a greater extent than both acuity measures. The magnitude of measured functional deficit was also greater on microperimetry than mfERG. When examined longitudinally, microperimetry was able to detect subtle functional changes over the 12-month period in intermediate AMD eyes that was not detected by LLVA or BCVA, especially in the parafoveal region. Longitudinal changes on mfERG were insignificant for implicit time and unclear for response amplitudes. In AMD eyes, the relative intensity of the inner-segment ellipsoids (ISe) band on SD-OCT correlated with mfERG implicit time and was reduced compared to control participants. The degree of retinal pigment epithelium (RPE) band elevation and ISe band integrity on SD-OCT were found to be independent predictors of microperimetric retinal sensitivity. In a retrospective, longitudinal analysis, a unique series of features that preceded the development of atrophy on SD-OCT was described as “nascent geographic atrophy (nGA)”. These features were present in a notable percentage of intermediate AMD eyes when examined in a cross-sectional study. The microperimetric retinal sensitivity in areas of nGA were reduced on average compared to surrounding areas of drusen and/or pigmentary abnormalities, but were not consistently the worst performing point in an eye. Conclusions: These findings show that microperimetry can be a potentially useful functional biomarker in the early stages of AMD. The structural parameters determined on SD-OCT can also provide rapidly acquired, more objective biomarkers, some of which correlate with the functional measures. Unique anatomical features detected on SD-OCT that were defined as nGA, appears to be the most specific marker to predict the development of atrophy. These novel clinical biomarkers can now be exploited to better understand the disease severity and progression for eyes with the early stages of AMD, and to improve the evaluation of novel interventions for these stages.
  • Item
    Thumbnail Image
    Modeling cost effectiveness of current routine treatments for neovascular age-related macular degeneration from a healthcare payer’s perspective
    Finger, Robert Patrick ( 2013)
    Neovascular age-related macular degeneration (nvAMD) is one of the leading causes of blindness in developed countries including Australia, and current gold-standard treatment is with anti-Vascular Endothelial Growth Factor (VEGF) drugs. However, cost-effectiveness (CE) of this treatment has to date only been investigated based on phase III clinical trial data, assessing treatment outcomes in only the study eye over only two years. As effectiveness as well as healthcare use are commonly overestimated in clinical trials, and patients require ongoing treatment beyond the first two years, this approach does not reflect day to day clinical reality. Thus, first the long term effectiveness and healthcare resource use in 200 patients treated with anti-VEGF drugs were assessed with up to 5 years follow-up. Mean treatment duration was 37 (±13) months and mean number of injections were 21(±11;7 in year 1-3, 6 in year 4, 5 in year 5). Visual acuity in the treated eye improved from baseline to last follow-up (49 to 56 letters read (ltrs), p<0.001). Fourty % of patients were treated in both eyes during year 1, and almost 50% by year 4. Treatment costs were highest in the first year (A$18,296 ± 7,991), and lower for uniocular (A$16,123±6,757) than for binocular treatment ($21,487±8,610). Cost decreased to A$11,420 (62%) in year five, with a much steeper decrease in uniocular treatment (to $7,698; 48%). Secondly the importance of using visual acuity (VA) in both eyes in outcomes and utility assessments in eye health were established in a large sample of over 1300 patients and controls. Using the Vision and Quality of Life (VisQOL) multi-attribute utility instrument (MAUI), utility scores decreased significantly with deteriorating vision in both the better and worse eyes when analysed separately. When stratified by 6 health states of vision impairment in both eyes, called vision states, VisQoL utilities decreased as VA declined in the worse eye despite stable VA in the better eye, demonstrating that calculating utilities based only on better eye VA is likely to underestimate the impact of vision impairment and thus treatment alleviating the impairment, particularly when the better eye has no or little VA loss and the worse eye is moderately to severely visually impaired. Thirdly, the CE of anti-VEGF treatment for nv AMD in standard medical practice in Victoria was assessed based on the collected data. In order to demonstrate the necessity of using vision states which capture VA of both eyes rather than modeling by treated or better eye only, several Markov models (MM) were built using TreeAge software, with health transition probabilities based on our real life treatment and utility inputs as described above. Costs and rewards were discounted at 3.5%/year and final results tested in probabilistic sensitivity analyses. Based on the clinical data MMs ran for 5 years, with all treatment assumed to be with ranibizumab (RBZ), the approved drug listed on the Pharmaceutical Benefits Schedule (PBS). The CE ratio was A$15,685/Quality Adjusted Life Year (QALY) for better eye, 16,296/QALY for treated eye and 15,780/QALY for both eye models, with the both eye MM generating the highest amount of QALYs (1.80 compared to 1.74 in MM1 and 1.70 in MM2). Based on these results, all subsequent MM assessing CE were based on vision states rather than treated or better eye only. The overall CE ratio was A$17,900/QALY, under the assumption that all injections were done using RBZ. Assuming that bevacizumab (BVZ, off-label use, not listed on PBS) was used instead of RBZ, the overall CE was A$ 3,196/QALY. Comparing both results for RBZ and BVZ with published CE ratios, all modelled treatment scenarios (both agents) were found to be below commonly applied cut-offs of A$ 30,000 - 70,000/QALY. In conclusion, good long-term effectiveness of anti-VEGF treatment under real life conditions was demonstrated, all relevant costs associated with this treatment were captured, and a novel method to assess the impact of vision impairment and its translation into CE modeling was developed. Based on this, anti-VEGF treatment with both, RBZ and BVZ, was found to be cost-effective over the five year period assessed.
  • Item
    Thumbnail Image
    Assessment of environmental and genetic risk factors for age-related macular degeneration
    ADAMS, MADELEINE ( 2012)
    This thesis consists of two sections. Section one presents the results of associations found between environmental factors and age-related macular degeneration (AMD) in a prospective cohort study, Section 2 presents genetic associations and interactions found between AMD risk genes and environmental risk factors from a case-control study nested in the cohort. Section One Aim: To investigate associations between environmental factors and age-related macular degeneration (AMD) in the Melbourne Collaborative Cohort Study (MCCS). Material & Methods: Participants included 21 287 men and women, who were aged 40–69 years at baseline (1990–1994 (48 to 86 years at follow up). Participant's body compositions were measured and alcohol and nutrient intakes were estimated from a food frequency questionnaire at baseline in 1990-94. At follow up from 2003-2007, digital macula photographs of both eyes were taken and evaluated for signs of early and late AMD. Section Two Aim: Associations between genotype and AMD were assessed using a nested case-control study (2,287 cases, 2,287 controls individually matched on age, sex and country-of-origin), selected from the Melbourne Collaborative Cohort Study. Statistical interactions between selected environmental and genetic risk factors were evaluated.
  • Item
    Thumbnail Image
    Retinal associations of diabetes and vascular disease
    Jeganathan, V. Swetha ( 2009)
    Background: Diabetes mellitus and vascular diseases have a significant impact on the eye. Aim: To determine the prevalence, risk factors, and racial/ethnic differences of major eye conditions, particularly retinal conditions, associated with diabetes and vascular diseases. Scope: To date, the majority of studies have examined the association of retinal vascular calibre and diabetes in predominantly white Caucasian populations. Further elucidation of ethnic differences in effects of hyperglycaemia on early microvascular disease is relevant, particularly amongst Asians where diabetes is likely to see the largest increase in prevalence over the next decade. We therefore examined these findings from three Asian population-based studies, the Singapore Malay Eye Study (n=3280), Singapore Prospective Cohort Study and Singapore Cardiovascular Cohort Study 2 (n=3748). Results: The prevalence of diabetic retinopathy in the Singapore Malay Eye Study was 35%, and associated with longer duration of diabetes, poorer glycemic and blood pressure control. More importantly, 9.0% had vision-threatening retinopathy, and retinopathy was found in 6.0% of people without diabetes. Retinal vascular calibre changes were incriminated in diseases such as diabetes and hypertension, independent of traditional cardiovascular risk factors. Wider venular calibre was independently associated with early age-related macular degeneration. We also found a novel association between peripheral artery disease and glaucoma, stronger in persons with diabetes, independent of vascular risk factors, supporting the vascular theory of glaucoma. Implications: Subtle changes in retina, including retinal vascular calibre may be early markers of widespread microvascular changes in diabetes, resulting from chronic hyperglycaemia and other pathogenic processes. These results will have broad implications for understanding the impact of both microvascular and macrovascular complications of diabetes in the Asia Pacific region and targeting relevant therapeutic interventions.