Conventional multimodal imaging and OCT-A diagnoses were established by an independent evaluation by two trained readers at the Odeon Ophthalmology Center. There was high (P < 0.05) interobserver agreement for CNV detection in both conventional multimodal and OCT-A imaging analyses. A simple κ coefficient was used to assess interimaging methodology agreement on CNV detection in AOFVD and was 0.86 (95% confidence interval: 0.61–1).
Figure 1 shows the conventional multimodal imaging (
Figs. 1A,
1B,
1C) and OCT-A (
Figs. 1D,
1E,
1F) images from a 46-year-old white male with AOFVD and a doubtful secondary CNV in the left eye. The FA image (
Fig. 1A) demonstrated in the late phase the presence of an irregularly stained central lesion with some weak leakage mainly from the nasal borders. The ICGA (
Fig. 1B) during the venous phase showed the presence of a hyperfluorescent lesion in the foveal area. SD-OCT (
Fig. 1C) showed a dome-shaped subretinal lesion in the vitelliform stage. The content of the lesion was homogenously hyper-reflective. The 30-μm–thick OCT-A C-scan (
Fig. 1D), passing through the lesion (
Figs. 1E,
1F), illustrated a well-circumscribed CNV in the subfoveal area, with the typical “spoked-wheel” shape, several large central vessels branching into smaller ones toward the periphery, and some anastomoses and loops. On the OCT-A B-scan (
Fig. 1E), an evident decorrelation signal coming from various structures was visible. Multiple hyperintense microdots in the retina, representing the superficial capillary plexus and the deep capillary plexus, were detected in the ganglion cell layer and the inner nuclear layer, respectively. A hyperintense signal coming from the choroidal vasculature (choriocapillaris [Sattler and Haller layers]) was also shown. Moreover, a pathologic hyperintense signal (due to the presence of perfused new vessels) was evident in the lesion area; some tiny hyperintense structures were visible above the reference plan (red lines) of the previously described C-scan.
Among the five cases with CNV detected by conventional multimodal imaging, four cases were also found to have CNV by using OCT-A. The sensitivity of CNV detection by OCT-A was therefore four of five (80%).
Figure 2 shows conventional multimodal imaging (
Figs. 2A,
2B,
2C) and OCT-A (
Figs. 2D,
2E,
2F) images from a 74-year-old white female with AOFVD and doubtful secondary CNV in the right eye. The FA image (
Fig. 2A) demonstrates in the late phase the presence of an irregular weak hyperfluorescence in the macular area. The ICGA (
Fig. 2B) during the venous phase shows the presence of a hypofluorescent lesion in the macular area due to the masking effect of the subretinal material accumulation, without any evidence of a neovascular network. SD-OCT (
Fig. 2C) demonstrated a dome-shaped subretinal lesion in the vitelliform stage. The content of the lesion was hyper-reflective with discrete pigmented epithelium detachment. The 30-μm–thick OCT-A C-scan (
Fig. 2D) and corresponding OCT-A B-scan (
Fig. 2E) did not show any hyperintense decorrelation signal that could be due to a CNV presence in the subfoveal area. The OCT-A C-scan showed a grayish, round lesion of relatively homogeneous signal intensity with no evidence of perfused (hyperintense) structures inside that blocked the reflected image of retinal vessels, which is normally present at the outer retina.
19
Of the 20 images without CNV on conventional multimodal imaging, all 20 were also determined to have no CNV on OCT-A. Therefore, the specificity of CNV detection by OCT-A was 20 of 20 (100%).