Tissue-related heterogeneity is a well-known pattern that is always detectable in cancer tissues at a more or less high degree. Sometimes this variability is of very low level and does not affect molecular analysis, but sometimes it can give highly contradictory results depending on the analyzed area. It is possible to recognize at least three types of this kind of heterogeneity. The first one is related to tissue complexity. Tumor tissues do not only contain cancer cells but a variable degree of fibrosis that is characteristic of the tumor type as a desmoplastic reaction. It could also be related to the size of the tumor with central necrosis substituted by a dense fibrosis due to insufficient neo-angiogenesis and related hypoxia. In most tumors there are also cells with normal genome, such as reactive inflammatory cells that are part of the carcinogenetic process, or residues of normal tissues involved in the invasion process. It is easy to realize how the presence of one or more of these cancer tissue components, especially if they are quantitatively relevant, can completely alter the results of molecular analysis, giving false positive or negative results. This is usually the only tissue heterogeneity taken today into account during tissue micro-dissection, when histological examination is performed and mechanical or laser dissection is suggested. In many tumors such as breast and colon cancer, it is common to find normal tissue together with the neoplasia in histological specimens and this mixture is often accepted in the sample analyzed for diagnostic molecular signatures such as Mammaprint, Oncotype DX, PAM50. It has been shown that the presence of normal tissue can modify, even if in a limited quantity, the category of risk to a less aggressive than the one detected in the pure breast tumor tissue [3].

A more complicated type of heterogeneity is the one related to different histological patterns in the same tumor, which sometimes could also represent different levels of molecular differentiation. At the moment this phenomenon is still badly defined and deserves more attention by pathologists and molecular biologists. For example, we could expect to find differences when analyzing differentiated and anaplastic areas of the same tumor that can often be found in human cancers. We need to better define those characteristics not only in a general way but also for specific types of tumors, by evaluating the importance of this factor in the reproducibility of molecular analysis in diagnostics and clinical research.

For quite a long time we have been aware that there are different functional areas in a tumor such as borders and the center of an invasive neoplasia that display different gene expression patterns [4]. Usually the central part of the tumor is more affected by hypoxia and cellularity is low with a higher fibrotic component, whereas the border can be crowded with cells with activated proteolytic enzymes and a more frequent interaction with reactive cells. The border itself can give better information on the aggressiveness of the tumor. When the tumor is large and the position of the analyzed tissues is unknown, this can heavily affect molecular diagnostics and research analyses.

Of course there are differences in tissue heterogeneity among tumour types and a more complete analysis is necessary to establish specific characteristics. However it is possible to consider some general rules that could help a higher level of standardization like those reported in Box 2.

Very short time between cutting micro-dissected area sections and the extraction of nucleic acids (especially for RNA) should pass, otherwise this could be another pre-analytical source of variability due to possible further degradation of nucleic acids by contamination of environment nucleases.