Image Quality and Accuracy of Eye Bank Measurement of Donor Cornea Endothelial Cell Density in the Cornea Donor Study

Abstract

Abstract: : Purpose: The Cornea Donor Study (CDS) is a prospective, double-blind study designed to evaluate whether or not donor age is associated with long-term (5-year) graft survival. The Specular Microscopy Ancillary Study (SMAS) is examining long-term endothelial cell loss as it relates to donor age in this same population. Quality and accuracy of cell counts were determined from the SMAS donor endothelial images that were submitted to the Specular Microscopy Reading Center (SMRC) at CWRU/UHC from 22 eye banks. Methods: A certified technician at the SMRC assessed the quality of the eye bank donor images (analyzable from fair to excellent vs. unanalyzable or poor) and determined the central endothelial cell density utilizing a variable frame analysis technique. This density was compared to the eye-bank-determined cell density, which was obtained by each eye bank utilizing their standard specular microscope and cell counting methods. Results: 1101 patients were enrolled in CDS between January 2000 and July 2002. 675 (61%) donor images were analyzed by the SMRC. Image quality was categorized by the SMRC as excellent (8%), good (62%), fair (27%), and poor (3%). Of the 652 analyzable images, 68% of the eye-bank-determined cell densities were within 10% of the SMRC-determined cell densities, whereas 25% were greater than 10% higher and 7% were greater than 10% lower. Conclusions: The majority of the eye bank images submitted to the SMRC were analyzable and within 10% of the SMRC count. However, despite steps to validate eye bank cell counting methods prior to study entry, 3% of the images were not analyzable and 32% of the images were more than 10% different than the SMRC count. These findings suggest that standardization of quality and cell counting criteria for specular microscopy in the eye banking community should be considered.

Keywords: cornea: endothelium • transplantation • clinical (human) or epidemiologic studies: tre