Spectral-domain OCT

All images were acquired by using a commercially available OCT device, Cirrus HD-OCT 5000 (Carl Zeiss Meditec Inc, Dublin, California, USA). This is a spectral-domain OCT with a 5 μm axial resolution, a wavelength of 840nm and takes 68,000 axial scans per seconds. One drop of artificial tear was instilled 5 minutes before the OCT examination to avoid any artifact from the ocular surface.

The Anterior Segment Cube 512x128 scanning protocol (which acquires a series of 128 horizontal scan lines, each composed of 512 A-scan creating a 3D image of the data) was used for the pattern analysis. In this case, the anterior surface of the cube (3D image) was representative of the outer corneal layers (e.g. the ocular surface) and therefore was excluded from our examination; whereas turning upside down the cube by the embedded software, the innermost layer of the cube was representative of the corneal endothelium alone or in combination with the Descemet’s membrane (DM).

Conversely, the Anterior Segment 5 Line Raster scanning protocol was used to in-depth evaluate the corneal structure and to measure the central corneal thickness.

All OCT scans were analyzed by Image J 1.50 software (National Institutes of Health, Bethesda, MD) to investigate the qualitative and quantitative features of the inner surface of the cornea. Each single corneal endothelium image was divided into 4 quadrants of the same size (supero-temporal, ST; supero-nasal, SN; infero-temporal, IT; infero-nasal, IN), and the mean and maximum reflectivity was calculated for each region.

All 3D AS-OCT images were classified in three different patterns according the signal distribution and the level of reflectivity (low, moderate, and high). We defined the apparent reflectivity based on the following grading scale: from black to blue was absence or low reflectivity, from green to yellow was moderate reflectivity, and red was high/intense reflectivity.

Consequently, pattern 1 was characterized by a homogenous, and low reflectivity (Fig 1); pattern 2 was defined by the presence of hyper-reflective orange-yellowish points scattered through the four quadrants (moderate reflectivity) (Fig 2); and pattern 3 was characterized by a mottled appearance of the entire inner surface with a variable number of hyper-reflective areas, resembling a marmoreal look-like (moderate/high reflectivity) (Fig 3). For simplicity, pattern 1 was defined as ‘homogeneous’, pattern 2 as “dotty”, and pattern 3 as “marmoreal”.

(a) slit-lamp biomicroscopy; (b) specular endothelial microscopy; (c) cross-sectional AS-OCT image confirming a normal corneal structure; (d) 3D AS-OCT image of a 512x128 cube scan showing a homogenous reflectivity of the inner corneal surface, with no hyper-reflective points/areas.

(a) slit-lamp biomicroscopy showing a beaten metal-like appearance of the corneal endothelium with the characteristic confluent guttae; (b) specular endothelial microscopy confirming the loss of endothelial cells; (c) cross-sectional AS-OCT scan displaying a faint hyper-reflectivity of the corneal endothelium; (d) 3D AS-OCT image of a 512x128 cube scan showing a large number of hyper-reflective orange-yellowish points scattered through the four quadrants of the inner corneal surface.

(a) slit-lamp biomicroscopy showing important alterations of the cornea including the presence of endothelial guttae and full-thickness edematous opacities; (b) specular endothelial microscopy confirming the advanced stage of the disease with few residual endothelial cells; (c) cross-sectional AS-OCT scan disclosing irregularities of all corneal layers with a non-homogeneous reflectivity; (d) 3D AS-OCT image of a 512x128 cube scan showing a mottled appearance of the inner corneal surface with a variable number of hyper-reflective areas, resembling a marmoreal look-like.

As described above regarding the biomicroscopic classification, two corneal specialists (MF, PEN) scored the OCT images. Any disagreement between the two corneal specialists was resolved by a third examiner (CI).