Determination of corneal endothelial cell density in French eye banks: second look

Abstract

Abstract Purpose Considering the importance of having a precise, robust and especially reproducible ECD counting method, Afssaps organized from April 2008 to June 2009 a second assessment of the reliability of the routine cell count within the 18 french Eye banks. Methods The study design was similar to the first assessment driven by the laboratory ‘Biology, engineering and imaging of Corneal Graft’ in 2003 (Transplantation 2004; 78: 1299‐1302).5 test corneas (1 mm2 of flat mounted, fixed and alizarin stained human corneal endothelium) were selected and sent to the 18 Eye banks. All the usual technicians of each bank had to count the test corneas using the routine method(s) employed to assess grafts. Results 430 counts were carried out by 70 eye banks technicians, by manual and/or image analysis system. 42% (180/430) deviated by more than 10% from the expected ECD. Among them, 128 were over‐estimated (max +88%) and 52 were under‐estimated (max ‐31%). 2 banks constantly over‐estimated (in the mean +31,7% and +42,7%, no calibration and/or material problem) but the 16 other banks were in average within ±13% from expected ECDs. For manual methods, a statistically significant difference between banks was observed for the 5 test corneas, whereas no difference was observed with image analyzers. ECD obtained with the analysers were closer to expected values than with the manual methods. Compared to the 2003 study, reliability of ECD determination globally improved. Conclusion Image analysis systems prove more reliable (precise and with a lower intra and inter observer variability) than manual counting methods. This ‘second look’ of Eye banks will allow editing recommendations to improve ECD determination.