Purpose
There are multiple classification systems for age-related macular degeneration (AMD), mostly based on patients from clinical trials. We validated two AMD classification systems using 5-year incident late AMD data from the Blue Mountains Eye Study (BMES) cohort, and compared the results with published estimates from the Age-Related Eye Diseases Study (AREDS) clinical trial.
Methods
Participants of the BMES were aged 40-99 years at baseline, n=2134; participants of the AREDS clinical trial were aged 55-80 years, n=3640. AMD lesions in the BMES were graded from stereoscopic fundus photographs and classified according to:1)AREDS Simplified Severity Scale, and 2)Basic Clinical Classification Scale. 5-year late AMD incidence rates (defined as the presence of geographic atrophy or choroidal neovascularization in either eye of subjects without late AMD in any eye at baseline) were presented by categories of both classifications and compared with published estimates from the AREDS population.
Results
The Simplified Severity Scale calculates a risk score based on the presence of large drusen and pigment abnormalities in both eyes.(Table 1) The Basic Clinical Classification Scale categorizes these lesions as Intermediate AMD regardless of bilaterality. There were 32 patients with incident late AMD in either eye in BMES, and 316 in AREDS. The Simplified Scale classified similar proportions of participants who developed incident late AMD in both the BMES and AREDS samples for levels 1, 2, and 4, with the exception of level 3 (Fig 1A). The Basic Clinical Classification Scale categorized consistently higher late AMD incidence rates across all risk levels in the BMES compared to AREDS (Fig 1B).
Conclusions
The AREDS Simplified Scale classifies late AMD risk levels similarly when applied to population-based and clinical trial samples. The Basic Clinical Classification resulted in higher incidence rates in all categories when applied to population-based compared to clinical trial samples. The choice of classification system may need to take into account the study population as different incidence rates may be reported.