Association between capillary congestion and macular edema recurrence in chronic branch retinal vein occlusion through quantitative analysis of OCT angiography

Abstract

This study aims to quantitatively investigate the optical coherence tomographic angiography (OCTA) findings of capillary congestion and its association with macular edema (ME) recurrence in chronic branch retinal vein occlusion (BRVO). We retrospectively reviewed the medical records of 115 consecutive patients with major ischemic BRVO who reached stable macula (without ME for two consecutive visits) at baseline (the first visit within the stable period). All patients were classified into a recurrence or non-recurrence groups depending on ME recurrence. Capillary congestion of deep capillary plexuses (DCP-C) and other abnormal capillary lesions were segmented, and their areas, vascular densities, and mean retinal thicknesses (MRT) were calculated. The main outcomes were differences between the two groups and risk factors for recurrence among baseline and OCTA parameters. A total of 76 eyes were included, of which 22 (28.9%) recurred. DCP-C existed in all eyes at baseline. MRT of DCP-C (p = 0.006) was greater in the recurrence group. Greater MRT of DCP-C (OR: 1.044; p = 0.002) and more frequent intravitreal injections (OR: 1.803; p < 0.001) were associated with a higher risk of relapsing ME. DCP-C may contribute to the anatomical stability of chronic BRVO and simultaneously be the source of ME.

Figure 1

Patient selection and capillary congestion at baseline. (a) A patient presented with acute major branch retinal vein occlusion (BRVO) at the right inferotemporal arteriovenous crossing (blue arrow). (b, c) Three months after the initial visit, a wide capillary nonperfusion area (yellow dotted double arrow) is shown on fluorescein angiography (FA), confirming ischemic BRVO. FA presents collateral vessels (red arrows) with fluorescein leakage at the late phase. (d) He reached stable macula at baseline 20 months after the initial visit with seven anti-vascular endothelial growth factor intravitreal injections and sectorial photocoagulation. (e, f) At baseline, en-face optical coherence tomographic angiography corresponding to the green box area reveals the nonperfusion areas on superficial and deep capillary plexuses (DCP). (g) The central subfield macular thickness was 214 µm without any fluids at baseline. (h) The gamma was adjusted to clarify the capillary congestion of DCP (DCP-C) from the (f) image. The retinal arterioles (yellow asterisks) from the nonperfusion area supply flow to DCP-C, and the flow drains to the central vortex venules on the opposite side (red arrows). (i) Abundant flows of DCP-C were noted on the composite image of en-face DCP and the color-coded vessel density map.

Figure 2

Segmentation of capillary congestion, abnormal vascular lesion, and foveal avascular zone in each capillary plexus. Segmentation process for the same eye shown in Fig. 1. (a) Red indicates an abnormal vascular lesion of the superficial capillary plexus (SCP-A). (b) Cyan indicates capillary congestion of the SCP. (c) Orange indicates the foveal avascular zone (FAZ) of the SCP. (d) Green indicates an abnormal vascular lesion of the deep capillary plexus (DCP-A). (e) Blue indicates capillary congestion of the DCP (DCP-C). (f) The purple region is the FAZ of the DCP. Ophthalmologists saved these six sets of manual segmentation data as 600 × 600 size of two-dimensional binary matrices composed of TRUE or FALSE. Colored pixels were assigned to TRUE and colorless pixels to FALSE. (g) DCP-A was larger than SCP-A, and the transitional zone (TZ, yellow) was formed in the region of abnormal DCP under normal SCP. (h) The DCP-C image was superimposed on the TZ. The pink-colored region indicates DCP-C comprised in the TZ, and 45.0% of DCP-C was spread in the TZ. (i) The gray-scaled retinal thickness map was superimposed on the segmented DCP-C image. The mean retinal thickness of the DCP-C (blue region) was calculated to be 276.1 µm at baseline.

Figure 3

Alignment of en-face optical coherence tomography at recurrence to baseline en-face optical coherence tomographic angiography to visualize the territory of macular edema. (a–c) The patient noted in Fig. 2 showed macular edema (ME) recurrence at 6 months after baseline. Red arrows indicate capillary congestion of the deep capillary plexus (DCP-C) surrounded by ME in horizontal (a) and vertical (b) B-scan optical coherence tomography (OCT). In quantitatively analyzing the ratio of increased retinal thickness (RIRT) on a pixel-by-pixel basis, an alignment is required to correct for relative eye movement between two visits. (d) Sixteen control points were selected at the apparent bifurcation of the en-face OCT image at the recurrence. (e) Superficial capillary plexus OCT angiography (OCTA) at baseline presents selected control points at the same bifurcation. Matlab software calculated the geometric relationship between two images in a 3 × 3 affine transformation matrix through the coordinates of two sets of control points. (f) The color of the en-face OCT was designated as orange after the transformation. The OCTA at baseline was designated as blue and merged with the transformed en-face OCT image to confirm the alignment. By applying the same matrix to the macular thickness color map at recurrence, it is possible to analyze the RIRT in pixel units. (g) The RIRT of each pixel was calculated, and the territory of ME was plotted with a contour plot. Python software assigned different colors to pixels whose ratio increased by 1.1 or more (assigned colors are indicated in a color bar at the upper right corner). Then, the boundary of DCP-C was covered with blue lines. The maximum RIRT was 2.833 at the fovea; DCP-C was juxtaposed with the ME territory and two local maxima.

Figure 4

Contour plots of the territory of macular edema with boundaries of deep capillary congestion at recurrence. Patients with chronic stable branch retinal vein occlusion experienced recurrence of macular edema (ME) from baseline. (a–c) Apexes of the ratio of an increased retinal thickness (RIRT) were in the capillary nonperfusion area (NPA) of the affected side. The other local maximums were in the foveal avascular zone (FAZ). Capillary congestion of the deep capillary plexus (DCP-C) surrounded the local apexes of RIRT. (d–f) Unlike the cases depicted in (a–c), the apexes of RIRT existed in the FAZ. Some DCP-C segments were not involved in ME. (g–i) Similar to the cases depicted in (d–f), the apexes of RIRT were in the FAZ. Most DCP-Cs existed on the border of ME territories, and in other DCP-Cs (h), ME has just emerged.