Improved correction of quantitative nuclear DNA (ploidy) measurements in tissue sections

To evaluate, using a computer model, the advantages for ploidy measurements of selecting only center-containing sections of nuclei in ultrathin, very thick or relatively thick tissue sections.

The computed corpuscle sectioning program was run on a personal computer. Its synthetic data were corrected by a variety of correction algorithms.

When only center-containing sections of nuclei were selected in ultrathin sections, spherical nuclei could be corrected perfectly, and mildly prolate ellipsoidal nuclei with a selection bias in favor of elliptical nuclear section profiles could be corrected with high fidelity. Ultrathin sections most faithfully represented the true height of the peak of highest ploidy and showed better peak discrimination than other choices of section thickness, but small sample size, wavy sections, markedly inhomogeneous intranuclear DNA distribution and oblate ellipsoidal nuclei represented significant limitations of this approach. As nuclear prolation increased, peak definition worsened, and the peak of highest ploidy was falsely shortened. Results were unaffected by errors in the estimation of section thickness when an internal diploid standard was used. The effect of variable internuclear DNA concentration in mildly or moderately prolate ellipsoidal nuclei was nil. The choice of correction algorithm was unimportant, except that the reference curve method was better able to analyze oblate ellipsoidal nuclei, wavy sections and nuclei with inhomogeneous intranuclear DNA, and provided superior insight into nuclear and section parameters. Thick and very thick sections did not require correction and, unlike ultrathin sections, were immune to markedly inhomogeneous intranuclear DNA distribution, to nonspherical nuclear shape and to focal variation in section thickness (waviness), but (in relatively thick more than in very thick sections) the height of the peak of highest ploidy was falsely shortened, often markedly, and peak definition was worse.

Choice of section thickness and selection of only center-containing nuclear sections for analysis with a bias in favor of elliptical nuclear section profiles in ultrathin sections are very important for optimal results; the choice of correction algorithm is less important.