Sensitivity and specificity of choroidal neovascularization detection in age-related macular degeneration with optical coherence tomography angiography

Abstract

Purpose : To determine the sensitivity and specificity of OCT angiography in the detection of choroidal neovascularization (CNV) in age-related macular degeneration (AMD).

Methods : Thirty-two eyes with treatment naïve CNV, 19 eyes with non-neovascular AMD, and 20 healthy eyes from age-matched controls underwent OCT angiography with a spectral OCT (Optovue, Inc.). The 3D angiogram was segmented into separate en face views of the inner retinal slab, outer retinal slab, and choroid. Masked graders reviewed the en face angiograms. Abnormal vasculature in the outer retinal slab served as candidate CNV. The sensitivity and specificity of CNV detection with OCT angiography was determined for each grader.

Results : Of 32 eyes with CNV, both graders identified 26 true positives with OCT angiography, resulting in a sensitivity of 81.3%. Four of the 6 false negatives had large subretinal hemorrhage (SRH) that prevented detection of CNV with OCT angiography (Figure 1). In these cases, some leakage was evident on fluorescein angiography (FA) and interpreted as CNV. Of the 2 other false negatives, 1 was associated with a large fibrovascular pigment epithelial detachment, and the other had a very small CNV that was difficult to discriminate from projection artifact. The sensitivity improved to 92.9% for both graders if eyes with SRH were excluded. Of 39 eyes without CNV, Grader A identified 37 true negatives for a specificity of 94.9%. One false positive occurred because of motion artifact in an eye with non-neovascular AMD. The other false positive occurred due to a small region of flow in the outer retinal slab in the center of geographic atrophy in an eye with no clinical signs of CNV. Grader B identified 38 true negatives for a specificity of 97.4%. One false positive was associated with motion artifact in a healthy control eye.

Conclusions : OCT angiography has high sensitivity and specificity in the detection of CNV. SRH is the primary cause of missed CNV detection. Since SRH is easily recognizable on clinical examination and a known association with CNV, this may not be a significant practical drawback. Motion artifact was the primary reason for false positive detection. Better software to suppress motion artifact is needed.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.

 

Figure 1: Example of false negative. CNV is not visible on OCT angiography due to SRH. Leakage is seen on FA (blue circle).