Taking tissue samples from the placenta: an illustration of principles and strategies.

Tissue samples are removed from placentas for a variety of reasons associated with a host of investigative techniques, including chorionic villus sampling, villus explant culture, cell culture, proteomic analysis, gene expression profiling, microscopy and morphometry. Apart from the latter, especially stereological analysis, many studies provide extremely limited information on how the samples were selected. At worst, we learn little more than the placenta was sampled. Sometimes, studies provide sufficient detail to reveal flaws in sampling, e.g. the selection of placentomes based on size rather than mere presence. Occasionally, the reader is informed, without further explanation, that representative samples were taken or that samples from placentas in different study groups were taken from standard or similar sites. Such statements raise doubts about the unbiasedness of the sampling process, leave the reader in ignorance of the quality of the final sample, thwart attempts at achieving study repeatability and compromise interpretations of the validity of study outcomes. And yet study outcomes depend critically on the selection process because sampling influences study errors, notably precision (random error) and bias (systematic error). This article aims to review the basic principles and virtues of random sampling in general and the practical utilities of variants of it. For many functional and structural studies, it suffices to randomise the positions of tissue samples but, in certain structural studies, orientation must also be randomised. Therefore, sampling tools for stereological estimation of membrane surface areas, tubule lengths and layer thicknesses are mentioned. Although emphasis is accorded to the placenta, the principles apply equally well to other organs and to lower levels of organisation including the subcellular. It is hoped that this review will inform future study designs, encourage greater transparency and facilitate sampling improvements.