Comparison of endothelial cell density determined by eye banks and by a centralized reading center in the Cornea Preservation Time Study (CPTS)

Abstract

Purpose : To assess the agreement between eye banks (EB) and a reading center (RC) for preoperative endothelial cell density (ECD).

Methods : The CPTS was a randomized clinical trial involving 1330 donor corneas from 23 eye banks to study the effects of preservation time on outcomes after Descemet stripping automated endothelial keratoplasty(DSAEK). EB technicians were trained on the CPTS imaging protocol, image quality and ECD determination. Specular microscopy images were taken either immediately following lamellar dissection by EB technicians (After Dissection Group) or just prior to shipping for corneas to be dissected by surgeons (Before Shipping Group). The EB determined the central ECD on 1 image using their standard image analysis method (center method, variable frame, or fixed frame). The EB submitted 3 images to the RC. Each of two independent RC readers determined ECD (variable frame method) on the 3 images and the 3 measurements were averaged. If the readers differed by ≥ 5%, a third reader adjudicated to determine final ECD.

Results : Among 1289 corneas with both EB- and RC-determined ECD, the mean (standard deviation) ECD was 2773 cells/mm² (300) by the EB and 2758 cells/mm² (395) by the RC. Bland-Altman analysis (Figure 1) showed that the mean difference between EB and RC ECD was 14 cells with 95% limits of agreement of (-664 and 692) [EB mean > RC mean; p <0.001]. Among the 955 corneas in the After Dissection Group, the mean difference between EB and RC ECD was -3 cells with 95% limits of agreement of (-713 and 706) [EB mean ≈ RC mean]. Among the 334 corneas in the Before Shipping Group, the mean difference between EB and RC ECD was 64 cells with 95% limits of agreement of (-505 and 633)(Figure 2). The variability between EB and RC ECD was greater in the After Dissection Group (p=0.02).

Conclusions : The mean pre-operative difference of ECD determined by EB and RC in CPTS was statistically significant but not clinically relevant. EB-determined ECD after dissection was similar to RC-determined ECD but EB-determined ECD before shipping was not statistically similar to RC-determined ECD. The higher variability in the After Dissection Group may be related to increased difficulty in imaging post dissection.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.