To determine whether baseline drusen load, as measured using spectral-domain optical coherence tomography (SD OCT), is a useful predictor of development of advanced age-related macular degeneration (AMD).
Retrospective cohort study.
setting : Academic clinical practice. study population : All patients with non-neovascular AMD and no retinal pigment epithelial (RPE) atrophy at baseline who were seen between 2007 and 2012 in a single academic retina practice. A minimum of 1 year of follow-up was required. observation : Drusen load (area and volume) was assessed using automated SD OCT software algorithms. main outcome measure : RPE atrophy area, assessed using an automated SD OCT software algorithm, and the development of neovascular AMD.
Eighty-three patients met the inclusion criteria with a mean age of 80 years and a mean follow-up time of 2.8 years. Repeated-measures analysis of variance showed an association between drusen area ( P = .005) and drusen volume ( P = .001) and the development of RPE atrophy. We also found an association between drusen area ( P = .001) and drusen volume ( P = .001) and the development of neovascular AMD.
Drusen load, as measured using SD OCT, is associated with the development of RPE atrophy and neovascular AMD. SD OCT assessments of drusen load are simple and practical measurements that may be useful in stratifying the risk of developing advanced AMD. These measurements have potential applications in both routine clinical care and clinical trials.
Age-related macular degeneration (AMD) is the leading cause of vision loss in patients over 50 in the developed world, with the majority of this vision loss stemming from 2 forms of advanced AMD: geographic atrophy (GA) in advanced non-neovascular AMD; and neovascular AMD. New treatments for neovascular AMD have significantly reduced its associated vision loss, but vision loss from GA continues to remain an untreatable consequence of non-neovascular AMD.
Recent studies have looked at various risk factors for progression to GA. These studies have uncovered a number of predictive factors: age, drusen size, reticular drusen, location of drusen, area of drusen, retinal pigment epithelium (RPE) pigmentary changes, drusenoid pigment epithelial detachments, and focal hyperpigmentation. The majority of these studies have used a manual assessment of color photographs to examine these various factors, which is a time- and labor-intensive method of analyzing parameters that is not practical in daily clinical practice.
In addition to phenotypic appearance, a number of genetic markers have also been shown to be helpful in predicting GA progression, including polymorphisms in the complement factor H gene, the ABCR gene, and Toll-like receptor 3 gene, among others.
Drusen have also been investigated as a marker of progression to neovascular AMD, with varying results. In addition, a number of other markers (eg, genetics, indocyanine green fluorescence patterns, and fundus autofluoresence ) have shown promise as indicators of conversion to neovascular AMD.
Newer spectral-domain optical coherence tomography (SD OCT) technology has enabled automated measurement of drusen area and volume, as well as area of RPE atrophy. Although these automated measurements may show reduced sensitivity of drusen determination compared to color photographs and measurements may exhibit some undulation over time, they have been validated as a method of following the development of GA and its progression.
The aim of the current study was to determine whether automated SD OCT analysis of drusen load (area and volume) at baseline is predictive of progression to GA and neovascular AMD in patients with non-neovascular AMD. We wished to determine if SD OCT analysis of drusen load has potential application in routine clinical care and in clinical trials as a predictor of advanced AMD development.
We performed a retrospective chart review of all patients seen by the retina service at the University of British Columbia between December 2007 and December 2012. Ethics approval from the University of British Columbia Research Ethics Board was obtained. Eligibility criteria included the diagnosis of non-neovascular AMD in the study eye and a minimum of 1 year of follow-up. Study eyes were fellow eyes of eyes that had neovascular AMD and were being treated with anti-VEGF therapy using a treat-and-extend protocol. Eyes with other significant macular pathology or media opacity that precluded the collection of high-quality images were excluded from the study. For eligible patients, demographic information including age and sex was collected. Baseline and final Snellen visual acuity was also collected, and was converted to logMAR equivalents for statistical analysis.
Drusen Load and Retinal Pigment Epithelium Atrophy Area Measurement
Drusen load (both area and volume) and RPE atrophy area were quantified using SD OCT. Eyes were scanned using the Cirrus HD-OCT (Carl Zeiss Meditec Inc, Dublin, California, USA), with either the 200 × 200 or 512 × 128 macular cube protocols. We used OCT-derived RPE atrophy area as a surrogate for GA area. Drusen load and RPE atrophy area were measured using the Advanced RPE Analysis tool on the Cirrus HD-OCT software. The Advanced RPE Analysis tool enables measurement of drusen load using a segmentation algorithm that isolates drusen within the sub-RPE space. RPE atrophy area is determined by looking at areas within a 5 mm circle where RPE is absent or has lost integrity and is accompanied by hyper-reflectivity in the choroid; the parameter is called “sub-RPE illumination area” (mm 2 ). The analysis is performed on the 3D macular cube acquired spanning 6 × 6 mm.
All SD OCT scans from every patient visit during the follow-up period were examined. We manually checked the automated segmentation of all scans, and scans with segmentation errors were excluded from analysis.
The data set was manually truncated to a maximum of 15 visits per patient, preserving the initial 3 and final 3 visits. Only data points that showed no variability with adjacent visits (both preceding and subsequent visits) were removed to a maximum remaining data set of 15 visits per patient. Outliers, namely those data points that showed variability of >3 standard deviation (SD) away from the mean of the preceding and subsequent visits, were also removed and replaced with the mean value of the data points from the preceding and subsequent visit.
Univariate analysis was performed using t test when data were approximately normal; otherwise the Mann-Whitney U test was used. For categorical variables we used the χ 2 test. Pearson and Spearman correlation was also used for univariate comparisons.
Repeated-measures analysis of variance (RM-ANOVA) was used to model the development of RPE atrophy with respect to drusen load, using all 15 data points per patient. For all other analyses, baseline drusen load and final RPE atrophy area were determined by averaging the values of the first 3 and final 3 visits (this was done to account for the variability in the measurements).
To account for the variable follow-up time in our cohort, we used a multiple linear regression to model baseline drusen area and volume on the final measure of RPE atrophy by adjusting for covariates including age, sex, and follow-up time. Likewise, Cox regression was used to model baseline drusen load on neovascular AMD by adjusting for the same covariates.
Alpha was set at P < .05. Data were analyzed using SPSS Windows version 20 (SPSS, Chicago, Illinois, USA).
Charts of 722 patients with AMD were reviewed to identify 83 patients who met the inclusion criteria. The mean follow-up time was 2.8 years (range 1.0–5.0, SD 1.1). Fifty-nine percent (49/83) of patients were female, and patients had a mean age of 80 years (SD 8). Mean drusen area at presentation was 0.59 mm 2 (SD 1.04) and mean drusen volume was 0.03 mm 3 (SD 0.06). A total of 23 patients developed GA; the mean area of GA increased over time with average area of 0.02 mm 2 (SD 0.13) at baseline and of 0.12 mm 2 (SD 0.23) at the final visit. Sample OCT data showing segmentation and drusen measurements is demonstrated in the Figure . Mean signal strength of the SD OCT scans was 7.6 (SD 1.3).
Univariate analyses did not show any association between age and sex and the development of RPE atrophy ( Table 1 ). However, baseline drusen area ( P = .001) and drusen volume ( P = .001) were associated with the development of RPE atrophy in univariate analyses ( Table 1 ). These results were confirmed with RM-ANOVA, which did not find an association with age and sex but showed a statistically significant association with baseline drusen area ( P = .005) and drusen volume ( P = .001) and the development of RPE atrophy ( Table 1 ).
Using a multiple linear regression model, only drusen volume was found to significantly correlate with development of RPE atrophy ( P = .001).
Spearman correlation was carried out and demonstrated a trend toward correlation between visual acuity and RPE atrophy area at the final visit, although this was not statistically significant (r = 0.196, P = .076).
Only 7 eyes in our cohort developed neovascular AMD. Univariate analyses did not show any association between age, sex, drusen area, and drusen volume and the development of neovascular AMD ( Table 2 ). RM-ANOVA confirmed the lack of association with age and sex but found an association with drusen area ( P = .001) and drusen volume ( P = .001).