The Relationship between Haemorrhage and OCT Signs and the Likelihood of Neovascular AMD being Active during Follow-up: Data from the IVAN study

Abstract

Purpose: To quantify the ability of retinal features on colour photographs and optical coherence tomography (OCT) line scans to diagnose active neovascular age-related macular degeneration (nAMD) classified by fluorescein angiography (FA).

Methods: Image sets collected as part of the IVAN study were graded prospectively in the Network of Reading Centres UK for presence/absence of haemorrhage and intraretinal fluid cysts (IRF), neuroretinal foveal thickness (NFT) and height of subretinal fluid (SRF) at the foveal centre. FA were graded for presence/absence of leakage (reference standard for activity). Feature discrimination between presence /absence of activity was quantified by receiver operating characteristic curve (ROC) areas. The relative increase/decrease in odds of activity for each parameter was described using positive/negative likelihood ratios (LR+ / LR-) and the effect of combining data using logistic regression and ROC statistics.

Results: The prevalence of activity in FAs at baseline (92%) precluded cross- sectional analysis. Data available at 12 and 24 months were: 12 months : 478 colours, 513 OCT, 449 FA; 24 months 442 colours, 456 OCT, 436 FA. Activity was present in 41% (183/449) and 38% (168/436) of FA at 12 and 24 months. Haemorrhages and IRF were present in 53 (11%) and 186 (36%) at 12 and 35 (8%) and 159 (35%) at 24 months. Median NFT heights (mm) were 0.15 (inter-quartile range (IQR), 0.12 - 0.19) at 12 months, and 0.14 (IQR, 0.11 to 0.18) at 24 months. 14% and 15% of OCTs showed any SRF at 12 and 24 months: median SRF heights where present were 0.07 (0.05 to 0.10) at both times. LR+ and LR- were: haemorrhage 4.38 and 0.84 (12 months), 5.51 and 0.87 ( 24 months); IRF were 1.72 and 0.72 (12 months), 1.25 and 0.88 (24 months); any SRF 5.47 and 0.76 (12 months), (8.10 and 0.71 (24 months). ROC areas for NFT and SRF, or combinations of features, were consistently ≤0.7 with no cut-off providing useful additional discrimination.

Conclusions: None of the features, separately or in combination, ‘diagnoses’ FA-classified active disease well. Haemorrhage and any SRF have high specificity, so help to ‘rule in’ the presence of activity; however, the majority of FAs classified with active disease did not have these features. The routine adoption of OCT-guided treatment for nAMD needs to be carefully considered.

Keywords: 412 age-related macular degeneration • 464 clinical (human) or epidemiologic studies: risk factor assessment • 466 clinical (human) or epidemiologic studies: treatment/prevention assessment/controlled clinical trials