Corneal endothelial validation in the eye bank: differences in automated methods and repeatability

To evaluate reproducibility of endothelial cell density (ECD) measurements using the Konan Cell Check D in donor corneas by two different ophthalmologists and to compare the two automated cell count methods (center and flex-center) available in the software of this specular microscope.

ECD values were quantified in 54 donor corneas by two independent investigators using the Cell Check D (Konan Medical USA Inc) with both automated cell count methods. In the center method, at least 30 contiguous cells are marked. For the flex-center method, an area is delineated and only the cells within the designated area are counted.

The mean ECD was 2473.81±378.22 cells/mm². Good ECD intergrader reproducibility for the center (ICC=0.821) and the flex-center method (ICC=0.784) were noted. Poor reliability was observed for coefficient of variation and hexagonality (ICC≤0.265). When both methods for ECD analysis were compared, a moderate correlation for the two independent graders using the two manual (center and flex-center) methods was detected (correlation coefficient of 0.678 and 0.745 for each of the investigators). Comparison between methods yielded significantly higher ECD with the flex-center method (P=0.013). When corneas were divided by ECD, those under 2200 cells/mm² and those between 2200 and 2700 cells/mm² also had significantly higher ECD with the flex-center method (P<0.022).

ECD values are reproducible with both methods, although the flex-center method ECDs tend to be higher, particularly in cases of low ECD. Eye banks and surgeons should exercise caution in making decisions based only on small differences in ECD.