Decisional answer tree analysis of exudative age-related macular degeneration treatment outcomes

Abstract

The use of intravitreal ranibizumab in exudative age-related macular degeneration (eAMD) has become commonplace. We aim to investigate the early predictors of this treatment outcome. Seventy-one treatment-naive eyes of 71 patients with eAMD of all lesion subtypes who received intravitreal ranibizumab treatment and completed 12 months of follow-up were included. All patients were loaded with three injections of ranibizumab at monthly intervals. Further injections were given if clinically indicated based on logMAR best-corrected visual acuity (BCVA) and optical coherence tomography findings. Casenotes of eligible patients were reviewed retrospectively. The main outcome measure was logMAR BCVA change at month 12. The mean number of injections given over 12 months was 5.4 ± 1.9. A total of 88.7 % of the patients achieved visual stabilisation (loss of <15 letters) and 15.0 % achieved visual improvement (gain of ≥15 letters). The mean letter change at 12 months was +0.3 letters. Regression analysis showed that baseline BCVA and letter change at month 3 predicted visual acuity outcome at month 12 (baseline BCVA: t = 6.97, p < 0.001; letter change: t = 5.84, p < 0.01) but age, gender and eAMD in the fellow eye were not predictive. Finally, a decisional answer tree model demonstrated that letter change at month 3 was a strong predictor of visual outcome at month 12 with an overall accuracy of 69 %. We found that letter change from baseline at month 3 was strongly predictive of visual outcome at month 12.