There is no simple adequate sampling scheme for estimating the myelinated fibre size distribution in human peripheral nerve: a statistical ultrastructural study.

Morphometric studies of peripheral nerves (PN) usually involve some sampling of the myelinated fibres (MF). In order to scrutinize the statistical properties of the sampling processes in common use and the reliability of the resulting estimates, a quantitative analysis of human superficial peroneal nerves from 8 different normal subjects was undertaken at the ultrastructural level, both in terms of MF spatial distribution and of their size distribution. This study used sampling rates involving more than 10% of the whole myelinated fibre population observed in each nerve fascicle. However, in nearly all the fascicles evaluated, the sampling fluctuations are so high that neither the number of axons nor their diameter distribution can be assessed with enough accuracy. A systematic study of the myelinated fibres shows that the spatial distribution of their size is not uniform. This marked heterogeneity in the MF size distribution imposes measurement of large enough samples (500 or 600 MFs usually represent about one-half or two-thirds of the whole MF population) in a way to secure a reliable enough estimate of the density and size distributions. However, the practical usefulness of sampling schemes requiring more than one-half of the whole MF population in a nerve fascicle, is questionable.