To assess the impact of using different sampling areas (fields of view) on the reliability of goblet cell density (GCD) estimates from conjunctival impression cytology (CIC) specimens from healthy individuals.
The CIC specimens were collected from the exposed nasal bulbar conjunctiva of 5 adult subjects (average age, 23 years) onto Biopore (Millicell) membranes and stained with Giemsa. A region from each of the specimens that contained abundant goblet cells was examined by light microscopy using a ×40 magnification objective lens, ×20 and ×10 lenses, the images were enlarged, and the goblet cells were marked and counted. The GCD values per square millimeter were calculated and then the impact of counting between 10 to many and 10 to few goblet cells assessed.
The mean GCD estimates at ×400 magnification, ×200, and ×100 were 950 ± 226, 620 ± 154 and 471 ± 158 cells per square millimeter, respectively; these values were statistically different (P<0.05). The GCD estimates could change by as much as ±31.6%, ±12.2%, and ±4.2% for differences of ±10 cells counted per image.
As a result of variability in goblet cell distribution across CIC specimens, the estimates of GCD can be expected to be different according to the sampling area used for goblet cell counts. Furthermore, the use of a small sampling area (high power field of view) is likely to result in an unacceptably large uncertainty (variability) in the GCD estimates.