| Sumario: | The assessment of intraepidermal nerve fiber density (IENFD) can be an objective, valid, and useful tool for the diagnosis of small-fiber neuropathy. IENFD testing involves sampling a small area of epidermis; a reduced IENFD relative to published age and sex normal values aids the diagnosis of small-fiber neuropathy. Patients often have borderline values, however, raising concerns of sampling bias, human error, and technical variations. In this issue of Neurology((R)), Engelstad et al.(1) report on efforts to refine the sampling process and the analytical protocol for IENFD assessment. The investigators obtained 3-mm skin punch biopsies from the distal leg and proximal thigh in healthy participants and patients with diabetes mellitus. Processed specimens were cut into 50-[mu]m sections, taking 10 serial 'skip sections' (sections 5, 7, 9, 11-23) for analysis. The epidermal nerve fibers/mm were counted by one technician and audited for accuracy by another. The variability of IENFD among sections was calculated for different numbers of sections counted (e.g., 4 vs 10 sections evaluated). The more sections reviewed, the lower the variability in IENFD (figure 1 of their article). While not surprising, this result is important because it suggests the number of sections to be quantified might need to be tailored for each patient, thus increasing reliability. Most laboratories currently base IENFD on measurement of 4 sections.(2) The authors provide results from several patients with 95% confidence intervals (CIs) that straddle the divide between normal and abnormal when 4 sections are counted, but whose 95% CIs narrow dramatically and fall clearly into normal or abnormal when 10 sections are counted (figure 2 of their article). These data suggest that improved accuracy results when using CIs to determine the appropriate number of sections to be analyzed, but in the examples provided, the average of 4 samples would still have yielded the correct diagnosis. It is nevertheless reasonable to assume that narrower CIs have a greater likelihood of identifying the correct diagnosis.
|
|---|