Quantification of sweat gland innervation: a clinical-pathologic correlation

Abstract

Objective: To evaluate a novel method to quantify the density of nerve fibers innervating sweat glands in healthy control and diabetic subjects, to compare the results to an unbiased stereologic technique, and to identify the relationship to standardized physical examination and patient-reported symptom scores.

Methods: Thirty diabetic and 64 healthy subjects had skin biopsies performed at the distal leg and distal and proximal thigh. Nerve fibers innervating sweat glands, stained with PGP 9.5, were imaged by light microscopy. Sweat gland nerve fiber density (SGNFD) was quantified by manual morphometry. As a gold standard, three additional subjects had biopsies analyzed by confocal microscopy using unbiased stereologic quantification. Severity of neuropathy was measured by standardized instruments including the Neuropathy Impairment Score in the Lower Limb (NIS-LL) while symptoms were measured by the Michigan Neuropathy Screening Instrument.

Results: Manual morphometry increased with unbiased stereology (r = 0.93, p < 0.01). Diabetic subjects had reduced SGNFD compared to controls at the distal leg (p < 0.001), distal thigh (p < 0.01), and proximal thigh (p < 0.05). The SGNFD at the distal leg of diabetic subjects decreased as the NIS-LL worsened (r = -0.89, p < 0.001) and was concordant with symptoms of reduced sweat production (p < 0.01).

Conclusions: We describe a novel method to quantify the density of nerve fibers innervating sweat glands. The technique differentiates groups of patients with mild diabetic neuropathy from healthy control subjects and correlates with both physical examination scores and symptoms relevant to sudomotor dysfunction. This method provides a reliable structural measure of sweat gland innervation that complements the investigation of small fiber neuropathies.

Figure 1 Skin biopsy analysis of sweat glands imaged with confocal microscopy A 3-mm punch skin biopsy was exhaustively sectioned and stained with PGP 9.5 (a pan-axonal marker) and tyrosine hydroxylase (a stain of sweat gland neuroendocrine tissue). Four illustrated sample tissue sections are shown through three sweat glands in the biopsy (A). Representative confocal images through three different sweat glands in the same biopsy are shown (B–I); sections B and I are taken at 10x magnification, sections C–H are at 20× magnification. The confocal images of the sweat glands (B–I) reveal the nerve fibers stained by the pan-axonal marker PGP 9.5 (red) innervating the sweat gland tubules stained by tyrosine hydroxylase (green). Samples from three different sweat glands (B, C–H, I) are shown to highlight sweat glands within the same biopsy. Z-stack confocal images of the same sweat gland are shown to highlight the sweat gland nerve fiber density seen within the same sweat gland in different tissue sections (200 μm apart) and in different planes of view (8 μm apart: C–E, F–H). Scale bar = 100 μm.

Figure 2 The quantification of sudomotor nerve fibers The unbiased stereologic sampling of sweat glands (A) involves serially sectioning the punch biopsy in a random plane perpendicular to the tissue surface (B), studying a single tissue section (C, D), digitally identifying the nerve fibers (E), and orienting a test grid of cycloids over the sweat glands and counting the intersections between cycloids and nerve fibers in a three-dimensional stepping pattern (F). The estimated length (L_E) of nerve fibers within the sweat gland was calculated from the ratio of tested area to cycloid test length (T_A/T_L), multiplied by the total cycloid intercepts counted (ΣI), divided by the magnification during counting (M), divided by the percent of the sweat gland sampled (SG). SG = (sampled sections/total sections) × (sampled area/total area) × (sampled depth of field/total depth of field). Thus the estimated length of nerve fibers within a given area is expressed by the following equation: L_E = (T_A/T_L × ΣI)/(M × SG). In this study the T_A/T_L = 3.05 cm, M = 661×, magnification and SG = 40% (the biopsies had 50% percent of the tissue sections imaged, 100% of the area sampled, and 80% of the depth of field sampled). The manual quantification of sweat gland nerve fiber density by light microscopy involves taking a digital image of the sweat gland at 20× magnification, shown in G. The same image, taken out of focus, is shown in H, with the selected area of interest highlighted in green. In I, the background staining from the out of focus image (H) is removed from the baseline image (G), providing a composite image (I). The nerve fibers in the composite image (I) have a grid placed over it (J); any circle partially or wholly contained within the AOI is eligible for counting. Nerve fibers that intercept the grid are counted manually (K); nerve fibers that touch but do not enter the circle are not counted. Scale bar = 50 μm.

Figure 3 Sweat gland nerve fiber density (SGNFD) by location in control and diabetic subjects The SGNFD are shown by site at the distal leg, distal thigh, and proximal thigh. Diabetic subjects are denoted by the clear box plots, healthy controls by the gray box plots. The box plots demonstrate the median value, with first and third quartiles outlined by the box, 10th and 90th percentiles by the whisker lines, and individual outliers shown as solid dots. ***p < 0.001, **p < 0.01, *p < 0.05.

Figure 4 Relationships between SGNFD, IENFD, NIS-LL and unbiased stereology (A) The relationship between sweat gland nerve fiber density (SGNFD) and the Neuropathy Impairment Score in the Lower Limb (NIS-LL). Each dot represents individual subjects with diabetes. The NIS-LL score is plotted against the SGNFD at the distal leg. The unbroken black line shows the regression (r = −0.89, p < 0.001). (B) The relationship between SGNFD and an unbiased stereologic estimate of nerve fiber length. Confocal images with an extended depth of field (20 μm). The results were compared against a gold standard unbiased stereologic estimate of nerve fiber length (expressed as length of nerve fibers in a cubic millimeter of sweat gland tissue) as seen in figure 2. Each dot represents individual sweat glands in the three sample biopsies. The unbroken black line shows the regression (r = 0.93, p < 0.01). (C) The relationship between intraepidermal nerve fiber density (IENFD) and the NIS-LL. Each dot represents individual subjects with diabetes. The NIS-LL is plotted against the IENFD at the distal leg. The unbroken black line shows the regression (r = −0.82, p < 0.001).

Figure 5 Thickened sudomotor nerve fibers in a subject with diabetes A sweat gland from the distal thigh of a control subject (A) and a subject with diabetes who has an Neuropathy Impairment Score in the Lower Limb of 4 (B). Nerve fibers in the diabetic subject have reduced density and appear thickened (B) compared to the control subject (A). Scale bar = 50 μm.