Analysis of differential shrinkage in frozen brain sections and its implications for the use of guard zones in stereology.

Increasing numbers of neuroanatomists are using stereological methods, and unbiased stereological estimation rules recommend the use of guard zones with the optical disector method to count objects of interest within a volume. Although these methods are statistically unbiased, we believe there is a need to explore sources of systematic bias (e.g., effects of tissue processing and sectioning) that may be affecting estimates of object number. Toward this end, we evaluated neuron distribution through, and tissue shrinkage in, non-embedded tissue cut on a freezing microtome. Our data show that in the x- and y-planes there are minimal changes in tissue area during tissue processing, sectioning, and staining. In the z-axis (perpendicular to the cutting surface), however, sections shrink to ∼25% of the cut thickness. This z-axis shrinkage was quite variable between sections (coefficient of variation about 10%) but stable within the same section (coefficient of variation about 3%). Lastly, individual particle densities are non-uniform through the thickness of the section when the densities should have been uniform. We advise experimenters to use a new protocol, a modified optical disector, for estimation when objects to be counted are marked such that the x-, y-, and z-coordinates are recorded through the full thickness of a section and guard zones are applied post data collection based on the characteristics of the object distribution along the z-axis. It is likely that individual experiments with different embedding materials and histological processing steps could require guard zones of varying sizes, or none at all, depending on object distribution in the z-axis.