Abstract
Purpose: :
Serial lens examinations of the AREDS cohort provide an opportunity to minimize the misclassification error that can occur when cataract phenotype is based on a single examination.
Methods: :
Slit lamp and retroillumination lens photographs were taken in AREDS at baseline and annually starting at the second annual visit. The photographs were graded centrally using the Age–Related Eye Disease System for Classifying Cataracts. Nuclear grades (0.9 to 6.1) were assigned using a series of standard photographs. Percentage of pupillary involvement was used to assign cortical and posterior subcapsular (PSC) grades. Cutpoints were established for the presence/absence of each type of opacity (present = ≥ 4.0 for nuclear, ≥ 10 % for cortical, or ≥ 5% of central 5 mm for PSC). Two approaches utilizing the serial data were used to assign cataract phenotype. In one, an algorithm gave special weight to consistency of "present" grades at the last 3 exams. The other applied a second order regression model to the serial grades to predict the opacity grade at the final visit. Both approaches gave special consideration to the occurrence of cataract surgery. Person phenotypes were established, and results from the 2 approaches compared.
Results: :
4,628 AREDS participants aged 55–80 years at baseline with at least one natural lens were followed for an average of 9.7 years (range 0–12.9 years). Person phenotype assignments were the same for the 2 approaches in 4,564 (98.6 %) participants. Cataract phenotypes for the AREDS cohort were: no cataract = 1,459 (31.5%), nuclear cataract = 1,267 (27.4%), cortical cataract = 1,383 (29.9%), PSC = 535 (11.6%), cataract surgery/no specific opacity type = 335 (7.2%), and questionable = 417 (9.0%). The number with "pure" cataract phenotypes were: nuclear = 758, cortical = 864, and PSC = 135.
Conclusions: :
Data from serial examinations have been used in an attempt to minimize misclassification error in assigning cataract phenotypes due in part to the inherent variability of grading images over time. The large pool of AREDS participants with well–characterized cataract phenotypes and the availability of participant DNA offer unique opportunities for genetic studies of cataract.
Keywords: cataract • genetics • detection