Global cell-by-cell evaluation of endothelial viability after two methods of graft preparation in Descemet membrane endothelial keratoplasty

Abstract

Purpose: To describe a novel method of global cell viability assessment for Descemet membrane endothelial keratoplasty (DMEK) and the comparison of two contemporary methods of donor tissue preparation.

Methods: DMEK transplants were prepared using two different methods: liquid bubble separation and manual peeling (n=8 each group). Samples were incubated with Hoechst, calcein-AM and ethidium homodimer prior to mounting on a curved imaging chamber. Z-stacked fluorescence microscopy images were combined to produce an in-focus global image capable of resolving all cell nuclei. Image processing software was used to define a calcein-positive live cell area, count all cell nuclei within this area and subtract ethidium-positive dead cells to derive the total viable endothelial cell count. Corrected global cell density was calculated by dividing the number of viable cells by the graft area, which had been corrected for imaging a curved surface.

Results: Corrected global cell density was lower than the central endothelial cell density in both groups: 85.5% of the pre-preparation central endothelial cell density in the peel group and 75.8% in the bubble group. Corrected global cell density was significantly lower in the liquid bubble separation group than in the peel group (p=0.04).

Conclusions: Eye bank estimations of central endothelial cell density overestimate true cell density after graft preparation in DMEK. A peel method is less damaging and more consistent than a liquid bubble method. Cell loss correlated strongly with the degree of stromal hydration prior to bubble separation in the liquid bubble group.

Keywords: Cornea; Eye (Tissue) Banking; Treatment Surgery.

Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://www.bmj.com/company/products-services/rights-and-licensing/

Figures

Figure 1

(A) Schematic diagram of the customised Descemet membrane endothelial keratoplasty (DMEK) imaging chamber. A radius of curvature of 6.9 mm was chosen to match the standard posterior corneal curvature, allowing DMEK donor specimens to lie flat, without wrinkles, in their native orientation. (B) Photograph of customised imaging chamber inverse mounted in the microscope stage for image acquisition. (C) The true area of the graft is that of a spherical cap and not a circle. Assuming that the graft is a circle underestimates the area and artificially increases the calculated cell density. True graft areas were calculated using the average measured graft diameter and the base curvature of the imaging chamber. The predicted area for a circle from an 8.00 mm trephine is 50.27 mm2. The area of a spherical cap is calculated as 2πRh, since True area of 8.00 mm DMEK transplant=55.49 mm2 (2πRh=2π×6.9×1.28). Individual calculations for the transplants were made using average graft diameter.

Figure 2

Flow chart outlining image processing and quantification steps.

Figure 3

(A) TRITC/ethidium channel image showing manual demarcation of graft edge using the polygon selection toll in Image J (yellow outline). (B) Yellow line defined in the TRITC/ethidium image is copied and applied to the FITC/calcein channel image. The edge of calcein-positive area is defined in the same way as the graft edge (blue outline). The difference in the areas of the yellow and blue selections was defined as the trephination damage area. (C) The FITC/calcein channel image is thresholded and segmented. The area of living cells (red area) is used to create a selection mask (green lines). (D) Ultraviolet/Hoechst channel images were colour inverted and the viable graft area selection applied (green outline). Each nucleus (black dot) within the selection mask is counted. Zoomed in area (top right) showing all nuclei within the mask have been counted, as shown by the presence of a red dot in the centre of nucleus. Nuclei outside the mask are not counted (no red dot in centre of nucleus). (E) TRITC/ethidium channel image is inverted, and the living area selection is applied (green outline). Each nucleus (black dot) within the selection mask is counted. Zoomed in area (top right) showing ethidium-staining nuclei within the mask have been counted, as shown by the presence of a red dot in the centre of nucleus.

Figure 4

Graph of area covered by viable cells against the amount of fluid injected prior to bubble formation. The numbers indicate the order of each sample in the experiment and show that increased stromal hydration was not limited to the early cases.

Figure 5

(A) Photomicrograph showing areas of cell damage staining with trypan blue on the corneoscleral button prior to peeling and trephination (white arrow heads). Yellow outline shows the site of trephination, which was off-centred to avoid the site of a previous corneal incision for cataract surgery. Yellow outlines shows site of trephination. (B) Immunofluorescence image of peeled Descemet membrane endothelial keratoplasty, areas of non-viable cells within the trephination area closely match those observed on the button prior to peeling but are easier to identify using this method (white arrow heads). Site of forceps fixation is marked with a yellow arrowhead. (C) Initial stromal hydration occurs prior to a peripheral Descemet membrane fluid bubble occurring. (D) The site of stromal hydration corresponds with maximal endothelial damage (red arrow) and cannot always be avoided by off-centred trephination.