SD OCT Features and Interobserver Agreement in Myopic Choroidal Neovascularization

Abstract

Purpose: : To evaluate interobserver agreement in SD OCT scan assessment of myopic patients with active choroidal neovascularization (CNV).

Methods: : 22 myopic eyes (refraction error >-6D or axial length >26 mm) with active CNV were studied with fluorescein angiography (FA) and OCT, performed on the same day. In FA localization of the CNV was evaluated in agreement by two expert examiners (11 subfoveal, 5 iuxta and 6 extrafoveal). For each patient a single 3D OCT scan with the best image quality was chosen to be examined independently by 4 different observers. The examiners measured manually central foveal thickness (CFT) and assessed the presence/absence of the following aspects, considered as characteristic signs of the CNV: hyperreflective tissue at the level of retinal pigment epithelium (RPE) and its localization respect to the fovea (extra or subfoveal), subretinal fluid, intraretinal fluid, intraretinal cysts and serous pigment epithelium detachment (PED).We computed intraclass correlation coefficient (ICC)/95% limits of agreement (95%LA) of manual measurement of CFT (automatic measure was the same for each scan) as +/-1.96 times the within-subject standard deviation. Agreement among examiners about presence/absence of typical features was calculated with k statistics.

Results: : 95%LA for CFT measurement was +-83 micron and ICC was 0.87. On average, manual measurements underestimated retina thickness by -59 micron compared to automatic measurement (288 vs. 229 micron). In all patient the CNV was identified in SD OCT as hyperreflective tissue at the level of RPE. CNVs were subfoveal in 5-7 cases using OCT, depending on the observer. Good agreement was achieved regarding CNV localization (k= 0,79) and the presence of cysts (k= 0,87). Agreement was poor regarding detection of intraretinal fluid (k= 0.34) and subretinal fluid (k=0,48). The k value for PED was nil due to disagreement about a very rare feature (only one examiner found it in two patients).

Conclusions: : SD OCT can help to investigate the features of active myopic CNV. There is agreement among observers in recognizing CNV location and intraretinal cysts, but less so for intraretinal and subretinal fluid. Compared to CNV localization with FA, in OCT a smaller number of subfoveal CNVs were recognized. Furthermore, in myopic CNV there is a high variability among observers in manual measurement of CFT.

Keywords: myopia • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • choroid: neovascularization