Prediction of visual function from automatically quantified optical coherence tomography biomarkers in patients with geographic atrophy using machine learning.

Geographic atrophy (GA) is a vision-threatening manifestation of age-related macular degeneration (AMD), one of the leading causes of blindness globally. Objective, rapid, reliable, and scalable quantification of GA from optical coherence tomography (OCT) retinal scans is necessary for disease monitoring, prognostic research, and clinical endpoints for therapy development. Such automatically quantified biomarkers on OCT are likely to further elucidate structure-function correlation in GA and thus the pathophysiological mechanisms of disease development and progression. In this work, we aimed to predict visual function with machine-learning applied to automatically acquired quantitative imaging biomarkers in GA. A post-hoc analysis of data from a clinical trial and routine clinical care was conducted. A deep-learning automated segmentation model was applied on OCT scans from 476 eyes (325 patients) with GA. A separate machine learning prediction model (Random Forest) used the resultant quantitative OCT (qOCT) biomarkers to predict cross-sectional visual acuity under standard (VA) and low luminance (LLVA). The primary outcome was regression coefficient (r2) and mean absolute error (MAE) for cross-sectional VA and LLVA in Early Treatment Diabetic Retinopathy Study (ETDRS) letters. OCT parameters were predictive of VA (r2 0.40 MAE 11.7 ETDRS letters) and LLVA (r2 0.25 MAE 12.1). Normalised random forest feature importance, as a measure of the predictive value of the three constituent features of GA; retinal pigment epithelium (RPE)-loss, photoreceptor degeneration (PDR), hypertransmission and their locations, was reported both on voxel-level heatmaps and ETDRS-grid subfields. The foveal region (46.5%) and RPE-loss (31.1%) had greatest predictive importance for VA. For LLVA, however, non-foveal regions (74.5%) and PDR (38.9%) were most important. In conclusion, automated qOCT biomarkers demonstrate predictive significance for VA and LLVA in GA. LLVA is itself predictive of GA progression, implying that the predictive qOCT biomarkers provided by our model are also prognostic.

United Kingdom National Ophthalmology Database Study: Diabetic Retinopathy; Report 1: prevalence of centre-involving diabetic macular oedema and other grades of maculopathy and retinopathy in hospital eye services.

AIMS: To report estimates of the prevalence of diabetic retinopathy (DR) and maculopathy grades for a large cohort of patients managed by the UK hospital eye service (HES). METHODS: Anonymised data were extracted from 30 UK NHS hospital trusts using a single ophthalmic electronic medical record (EMR) for the period from April 2000 to November 2010 to create the National Ophthalmology Database (NOD). From 2007, the EMR facilitated capture of a nationally agreed-upon standardised data set (DR Structured Assessment) relating to the presence or absence of clinical signs of DR and maculopathy. An algorithm in the software automatically calculated the Early Treatment of Diabetic Retinopathy Study grades of retinopathy and maculopathy. RESULTS: Between 2007 and 2010, 307,538 patients had data on the NOD, with 76,127 (24.8%) patients having been recorded as having diabetes. The proportion of patients with diabetes who had a structured assessment increased from 50.7% (2007) to 86.8% (2010). In each NHS year, 12.6-20.6% of eyes with structured assessments had no DR; 59.6-67.3% had non-proliferative DR; and 18.3-20.9% had active or regressed proliferative DR. Clinically significant macular oedema was present in 15.8-18.1% of eyes, and in 8.7-10.0% of eyes, this involved the central macula. CONCLUSION: This study provides contemporary estimates of the prevalence of retinopathy and maculopathy grades in a large cohort of patients with diabetes managed by the UK HES. Centre-involving diabetic macular oedema, potentially amenable to anti-VEGF therapy, is present in the eyes of almost 10% of these patients. This information is useful for clinicians, health-care economists, and commissioners involved in planning and delivering diabetic eye services.

Trends in rates of retinal surgery in England from 1968 to 2004: studies of hospital statistics.

AIM: To describe trends over time and geographical variation in rates of vitreo-retinal surgery in England from 1968 to 2004. METHODS: Routinely collected hospital statistics were analysed for England, using the Hospital In-patient Enquiry and Hospital Episode Statistics from 1968 to 2004, and for the Oxford National Health Service Region using the Oxford Record Linkage Study from 1963 to 2004. RESULTS: Annual admission rates in England for surgery were about 5 episodes per 100 000 population in the 1960s, rising gradually to about 10 per 100 000 in the early 1990s, and then more sharply to 30 by 2004. The Oxford Record Linkage Study shows that multiple admissions per person were rare, which confirms that the observed increase represents a real increase in the number of people treated. Annual rates in England for buckle procedures declined to about 6 episodes per 100 000 population in 2004, whereas vitrectomy surgery rose to about 26 episodes per 100 000 population in 2004. CONCLUSION: Vitreo-retinal surgery has developed over the last 40 years in England, rapidly so over the past 15 years. Vitrectomy surgery has become much more common in England and buckle surgery has shown a steady decline. We demonstrate statistically significant geographical variation in the current annual rate of surgery between local authorities in England.

Trends in rates of trabeculectomy in England.

AIMS: Rates of glaucoma surgery have declined in North America and continental Europe in recent years. The aim of this study was to examine trends over time and regional variation in rates of trabeculectomy in England. METHODS: The hospital in-patient enquiry (HIPE), hospital episode statistics (HES), and the Oxford record linkage study (ORLS) were analysed for annual trabeculectomy admissions between 1976 and 2004. RESULTS: Annual rates of admission for trabeculectomy rose 10-fold from 1976 to 1995: from 3.7 (95% confidence intervals 3.5-3.9) admissions per 100,000 population in 1976 to a peak of 38.7 (38.1-39.3) in 1995. Admission rates then declined sharply and have begun to reach a plateau at around 10.6 (10.3-10.9) in 2004. The highest surgical rates during the period 1997-2004 were found in the 80- to 84-year-old age group. Geographical analysis showed wide variation across local authority areas in annual rates of trabeculectomy, from 4 (2.3-5.2) to 33 (29.0-36.5) people per 100,000 population in 1998-2004. The rate of surgery by local authority showed little or no association with the level of social deprivation in each area. CONCLUSIONS: The rate of patients undergoing trabeculectomy increased substantially over 20 years. This was followed by a profound reduction in rates of trabeculectomy from 1995, which coincides with the introduction of new topical medications to reduce intraocular pressure. Wide regional variation in rates of trabulectomy was found, but there was no evidence of reduced access to glaucoma surgery in deprived areas.