The relation between immunohistochemical product discreteness and leukocyte manual quantification reliability

The discreteness of immunohistochemical products may interfere with detecting individual cells when obtaining manual quantification even with the application of a nuclear stain. To examine the influence of discreteness, we: (1) determined the relation between manual quantification reliability and discreteness, (2) analyzed the reliability between the manual and computer-assisted approaches, and (3) determined the relation between the reliability of the manual and computer approaches and discreteness. Sixty murine injured skeletal muscle sections were immunolabeled using the antineutrophil antibody, 7/4. Using computer-generated section images, an observer counted the number of cells on 2 different occasions (T1 and T2). Computer cell counts per section and discreteness per section (as indicated by the max pixel area per section) were determined. A moderate correlation was found for the relation between manual reliability and discreteness (P<0.01). The rankings of manual and computer counts were different (P<0.01), suggesting a lack of reliability between these approaches. The relation between the reliability of the manual and computer approaches and discreteness was moderately correlated (P<0.01). Samples with a computer count 200 or more had a greater mean max pixel area than those with a count <200 (P<0.0001). In conclusion, manual quantification may be more appropriate with samples with a leukocyte count <200, because discreteness is higher within this cell count range. Alternatively, when samples consist of 200 or more cells and discreteness is low, then the approaches that account for discreteness should be utilized.