Classification of teased nerve fibers for multicenter clinical trials.

ABSTRACT

Teased nerve fibers are used widely in both clinical and experimental neuropathology, but anecdotal evidence indicates that even experienced readers find little agreement on categories for teased fiber classification. To develop a classification scheme that could be used and understood by both experienced and naive readers, specific criteria were developed for normal fibers and those exhibiting Wallerian degeneration, demyelination, hypomyelination, remyelination, and abnormal paranodal myelination. Twenty fibers teased from human sural nerve biopsies were selected as examples of one or more of these categories. Ten readers, including seven having no previous experience with teased fibers, were given a set of instructions and asked to score each fiber for all matching categories. These readers averaged high rates of true positive (56-85%) classifications, while average false positive (3-18%) rates were much lower. Among the three experienced readers, true positive agreements averaged between 75 and 100% across the fiber classifications. False positives were correspondingly low, ranging between 0 and 8%. These results suggest that it is possible to design an easily learned, meaningful scheme for classifying teased nerve fibers.