Piffalls in diagnostic molecular pathology--significance of sampling error.

Today, molecular diagnostic tests are widely used in clinical medicine with polymerase chain reaction (PCR)-based techniques being of particular interest. In tissue specimens, however, false-positive and false-negative results can be obtained if pathomorphological and processing aspects are not considered. We therefore studied the impact of tissue sampling in three widely used diagnostic tests: (1) assessment of clonality in B-cell non-Hodgkin's lymphoma, (2) analysis of microsatellite instability (MSI) in colorectal neoplasia, and (3) demonstration of mycobacterium tuberculosis. Tissue sections of routinely formalin-fixed and paraffin-embedded diagnostic specimens were taken, and the significance of sampling was systematically investigated using laser microdissection or processing of the entire section. PCR analyses were done according to standard protocols. False-positive pseudo-monoclonality was obtained in small gastrointestinal biopsies and in laser microdissected lymph follicles of non-neoplastic tonsils. False negativity (pseudo-stability) could be demonstrated in a case with hereditary rectal adenoma if whole tissue sections were taken without microdissection of the most dysplastic subareas. Demonstration of mycobacteria failed in tissue sections of a formalin-fixed lymph node that was positive after complete digestion or in fresh frozen material of the same patient. In diagnostic molecular pathology, sampling error is an important source of false-positive and false-negative results, particularly if disease- and tissue-specific morphological features, such as sample size, type of fixation, and intralesional heterogeneity, are ignored.