This article is more than 5 years old. Information may no longer be current.
Drusen measurement algorithm challenges belief that more drusen increase risk of CNV
Study data determine fellow eye involvement is the most consistent risk factor of an eye’s conversion to wet AMD.
Click Here to Manage Email Alerts
A larger drusen area does not necessarily correlate with an increased risk of conversion to choroidal neovascularization, according to a study.
Thomas R. Friberg, MD, and colleagues developed a computerized algorithm that retrospectively assessed an eye’s relative risk of conversion to wet age-related macular degeneration, based primarily on drusen measurements obtained from two previous studies.
“Our results, taken in aggregate, challenge the belief that an eye is necessarily at a correspondingly higher risk of developing wet AMD as more and more drusen develop and as drusen area increases,” Friberg and colleagues said in a study published in Investigative Ophthalmology & Visual Science. “The risk of CNV imparted from the presence of drusen in an eye seems, in fact, to reach a maximum threshold in some cohorts. In addition, we found that where the drusen are located is also relevant.”
Developing the algorithm
Friberg led the Prophylactic Treatment of Age-related Macular Degeneration (PTAMD) trial, in which subthreshold infrared laser spots were used to induce the resorption of drusen. Because drusen were central to the PTAMD trial, Friberg determined that development of a drusen analyzer would create a new outlook on the analysis of the disease.
Developed over the course of 3 months by Friberg and colleagues in Crete, Greece, the algorithm program is complex and was validated across hundreds of fundus images, he said.
“Because of a career-long interest in macular degeneration, I thought it would be useful to develop an automated algorithm to detect pathology in the fundus and drusen in particular,” Friberg said in an interview with OSN Retina.
The Drusen Analyzer (Iridex) uses an interactive computer algorithm that was written specifically to detect drusen within a high-resolution digital image, according to the study. The algorithm ignores or rejects any confounding morphological abnormalities, such as retinal pigment epithelial atrophy or exudates.
Figure. The right eye of a patient (left) with dry AMD shows multiple soft drusen. The left eye of a different patient (right) has many more drusen and a more extensive drusen area. Nevertheless, having a greater drusen area does not necessarily impart a higher risk of conversion to neovascular AMD for the eye, according to Thomas R. Friberg, MD.
Images: Friberg TR
“Before this, most of the work done in the area was dependent upon human readers and the use of templates, a method prone to error and subjectivity and replete with imprecision,” Friberg said. “We simply wanted to quantify the analysis and make it more robust, with the goal of revisiting the relationship of drusen to pathology and risk from a fresh perspective.”
Research findings
Friberg and colleagues retrospectively studied 820 eyes enrolled in the Age-Related Eye Disease Study (AREDS) and 129 eyes enrolled in the PTAMD trial.
Based on baseline fundus images taken from two central macular regions, the image analysis algorithm determined drusen size, distribution, drusen area and hyperpigmentation.
The relative risk for an eye’s conversion to CNV was then calculated based on drusen area, the presence of one or five large drusen, hyperpigmentation and fellow eye status.
Friberg and colleagues then developed retrospective risk models based on the eye’s morphological features seen at baseline in both the AREDS and PTAMD studies.
The parameters assessed included the presence or absence of hyperpigmentation in the central 1,000- and 3,000-µm diameter regions of the macula, whether or not the fellow eye was affected by previous neovascular AMD, and the total drusen area, measured as a continuous variable, within the central 1,000- and 3,000-µm regions.
Study participants were subdivided into one of four models. Model 1 analyzed both AREDS and PTAMD participants, allowing for an interaction between drusen area and the specific study in which the patient participated; model 2 included AREDS participants only; model 3 included PTAMD participants only; and model 4 analyzed both AREDS and PTAMD participants but without the interaction in model 1.
Friberg and colleagues found a statistically significant correlation between drusen area and the particular study in which the subject was enrolled (P < .05).
In the 1,000-µm region for AREDS eyes, the odds ratios for the presence of a large druse, the presence of hyperpigmentation and fellow eye involvement were 2.60, 1.71 and 6.44, respectively. In the same region for PTAMD eyes, the odds ratios were 8.24, 1.37 and 17.56, respectively.
In the 3,000-µm region for AREDS eyes, the odds ratios for the presence of a large druse, the presence of hyperpigmentation and fellow eye involvement were 3.45, 3.40 and 4.59, respectively. In the same region for PTAMD eyes, the odds ratios were non-significant, 6.58 and 11.62, respectively.
“We have to challenge established dogma, for it can often mislead us,” Friberg said. “I no longer believe that drusen area, and for that matter, drusen volume, is such a compelling area of research.”
The strongest and most consistent risk factor across all models was fellow eye involvement, according to the study.
“Our statistical method calculated the risk for each eye alone and did not require us to express risk on a per-subject basis, as is done in other risk assessment strategies,” the study authors said. “In this way, the relative importance of each risk factor for an eye can be seen more clearly.”
Physicians must remember, however, that just because something is measurable does not mean the measurements are always relevant, Friberg said.
“Drusen area, measured in the central macula, is relevant up to a point,” he said. “But our work showed that it becomes less important after the central drusen area reaches a certain limit — in our case, an area equivalent to about 60 drusen.”
Response to criticism
“Population-based studies are useful, but as physicians, the patient in front of us is our focus. Knowing risk factors is of general but not too compelling importance,” Friberg said.
In response to criticism of the study’s data, Friberg conducted a survival analysis of the data using a frailty model to account for both eyes. He also completed a separate analysis of the study data after removing participants who had less than 200 days of follow-up.
In a subsequent issue of Investigative Ophthalmology & Visual Science, Friberg said the results of these analyses confirmed the study’s original results that drusen area, as a continuous variable, was not a significant risk factor for the development of CNV within the study’s cohort.
“I was not surprised by the initial criticism of these articles, which could be summarized by saying our analyses must have been flawed,” he said. “We, however, had been working in the field for years and were confident with our conclusions. Everyone learns from such challenges, including ourselves.” – by Ashley Biro