Abstract

Aims—To examine the basic assumptions made during DNA ploidy analysis of histological sections with an aim to eliminate methodological errors that have lead to conflicting results with this technique.

Methods—A rat liver imprint and histological sections together with sections and whole nuclei cytospins of human breast tumour biopsy specimens were stained with azure A Schiff's reagent and used to investigate the effects of computer imaging, histological section thickness, nuclear volume, and shape corrections.

Results—The rat liver imprint demonstrated a linear relation between mean nuclear transmittance and nuclear area for each of the three (2C, 4C, 8C) hepatocyte clusters. This finding was used to produce similar, proportional integrated optical density (IOD) measurements from rat liver sections by selecting only nuclei that were sectioned through their centres, as assumed by the mathematics for volume correction. The limitations of computer imaging necessitated an edge (glare) correction for each nucleus so that nuclei of different sizes could be analysed. Shape correction was required to analyse nuclei of different morphologies. Normal human lymphocytes, squamous epithelium and fibroblasts were all measured with similar IODs. DNA ploidy values obtained from 7 μm human breast tumour sections (using lymphocytes as controls) correlated well with those obtained from the whole nuclei cytospins from the same tissue blocks.

Conclusions—With an improved understanding of the theoretical and technical aspects of ploidy analysis of tissue sections, reproducible and consistent results are possible. These results can be integrated into routine histopathology investigations alongside immunohistochemistry and molecular diagnostic techniques.