Assessment of the reliability of human corneal endothelial cell-density estimates using a noncontact specular microscope.

PURPOSE: We sought to determine the variance in endothelial cell density (ECD) estimates for human corneal endothelia.
METHODS: Noncontact specular micrographs were obtained from white subjects without any history of contact lens wear, or major eye disease or surgery; subjects were within four age groups (children, young adults, older adults, senior citizens). The endothelial image was scanned, and the areas from > or =75 cells measured from an overlay by planimetry. The cell-area values were used to calculate the ECD repeatedly so that the intra- and intersubject variation in an average ECD estimate could be made by using different numbers of cells (5, 10, 15, etc.).
RESULTS: An average ECD of 3,519 cells/mm2 (range, 2,598-5,312 cells/mm2) was obtained of counts of 75 cells/ endothelium from individuals aged 6-83 years. Average ECD estimates in each age group were 4,124, 3,457, 3,360, and 3,113 cells/mm2, respectively. Analysis of intersubject variance revealed that ECD estimates would be expected to be no better than +/-10% if only 25 cells were measured per endothelium, but approach +/-2% if 75 cells are measured.
CONCLUSION: In assessing the corneal endothelium by noncontact specular microscopy, cell count should be given, and this should be > or =75/ endothelium for an expected variance to be at a level close to that recommended for monitoring age-, stress-, or surgery-related changes.