Abstract
Purpose: :
To develop a risk score from established risk factors for predicting the 5-year risk of GA in patients with bilateral drusen.
Methods: :
CAPT patients had ≥10 large drusen (>125µ) in each eye. One eye was assigned randomly to laser treatment and the other was observed. Trained graders identified GA on annual fundus photographs through 5 years. Three outcomes for GA were used: CAPT endpoint (total area >1 DA), involving the foveal center (CGA), and ≥1 area anywhere. Baseline risk factors (age, smoking status, hypertension, AREDS simple scale, and night vision symptom score) were assessed with a logistic regression model of the person-specific risk of endpoint GA. Regression coefficients were used in assigning risk points (pts) for each factor. Area under the ROC curve (AUC) and sensitivity and specificity for selected threshold values of risk score were used to assess the accuracy of prediction. The GA risk score was also applied to predict CGA and GA anywhere, and GA in untreated eyes only.
Results: :
Among 942 CAPT participants without GA at baseline, 64 (6.8%) participants developed endpoint GA, 90 (9.6%) developed CGA, and 342 (34.4%) developed GA anywhere, in either eye. The 15-pt GA risk score was developed based on age (4 pts), smoking status (1 pt), hypertension (1 pt), AREDS simple scale (5 pts) and night vision score (4 pts). The 5-year incidence of endpoint GA increased with GA risk score: 0.6% for <7 pts, 3% for 7-8 pts, 4% for 9 pts, 6% for 10 pts, 11% for 11 pts, and 15% for ≥12 pts. The GA risk score predicted development of endpoint GA moderately well with ROC AUC 0.72 (95% CI: 0.65-0.77). When a cutpoint of 9 was used to classify people as high risk for endpoint GA, sensitivity was 78% and specificity was 58% and were 69% and 68%, respectively, with a cutpoint of 10. The risk score also predicted CGA (AUC: 0.74 (0.69-0.78)) and GA anywhere (AUC: 0.67 (0.64-0.71)) moderately well. Similar results were obtained when the risk score was applied to GA in untreated eyes only.
Conclusions: :
The GA risk score was highly predictive of GA in CAPT participants. With verification in other groups, it may provide a useful tool to identify high risk GA patients for preventive intervention.
Keywords: age-related macular degeneration • clinical (human) or epidemiologic studies: risk factor assessment • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology