Reviews

Review of Publication: Gibbons, C.H., Illigens, B.M.W, Wang, N., R. Freeman (2010) Quantification of Sudomotor Innervation: a Comparison of 3 Methods. Muscle Nerve, 42, pp. 112 – 119.

A decrease in the quantity of sweat gland innervation might be an early indicator of distal small fiber neuropathy. Neuropathy, for instance in diabetes, can cause pain, impaired sensation, ulceration, and infection that may result in a need for amputation. Unbiased stereology is the best way to quantify morphometric parameters in biological systems that are observed under the microscope; specifically, the spaceballs probe can be used to estimate the total length of fibers that innervate sweat glands.

Different methods of quantifying nerve fibers in human skin biopsies are compared to unbiased stereology in Quantification of Sudomotor Innervation: a Comparison of 3 Methods (Gibbons et al., 2010). For unbiased stereology the cycloids for Sv probe was used instead of Spaceballs that had not been invented yet. Biopsies from the legs of thirty-six diabetic and seventy-two control humans were sectioned at fifty microns, and four sections from the middle of each biopsy were randomly selected (Gibbons, et al., 2010). A biopsy, by definition, is not collected using systematic random sampling; the sampling data is based on a small random part of the region, not the whole skin. However, systematic random sampling is more efficient than random sampling and could have been used on the biopsy itself. Sections were stained with the peripheral nerve stain, PGP 9.5. Three separate methods of quantifying sweat gland innervation and another assay that quantifies intra-epidermal nerve fiber density were performed on the sections from these biopsies. Three additional biopsies from healthy subjects were also taken for colocalization of PGP 9.5 and tyrosine hydroxylase (TH marks sweat gland neuroendocrine cells) for unbiased stereology, that is cycloids for Sv (also see, Gibbons, et al., 2012). Sections were examined with a 20x objective and an unsharp mask filter was used.

 

INTRA-EPIDERMAL NERVE FIBER DENSITY

Intersections of epidermal nerve fibers with the basement membrane were counted; if the fiber goes between epidermis and dermis, that intersection is counted (Gibbons et al., 2010). The number of these intersections are divided by the length of the basement membrane, that is the length of the border between the dermis and the epidermis. Then this intersections per length ratio was divided by the area of the sweat gland. This assay uses the basement membrane as the planar-probe so it will only give information about the amount of fibers at the dermal-epidermal border and will miss any fibers that are parallel to this membrane. It’s true that the length per volume is proportional to the intersections per area, but using this procedure, the area should be of the basement membrane, not of the sweat gland, and the intersections per area should be multiplied by a factor of two (see ‘length’ for more details on the applicable formula; length per volume is equal to twice the number of intersections counted per area). Nevertheless, this is the assay the authors are using as a benchmark to compare to the following techniques used to quantify sweat gland fiber density.

 

SWEAT GLAND NERVE FIBER DENSITY

Sweat Gland Nerve Fiber Density: semi-quantitative

Every whole sweat gland in the chosen sections was examined (Gibbons et al., 2010). Human experts picked a rating from 0-4 with 0 as no sweat gland innervation and 4 as the most.

Sweat Gland Nerve Fiber Density: automated

Instead of examining every whole sweat gland, microscope fields were selected with the aid of a computer. The paper (Gibbons et al., 2010) does not mention how the fields are selected, but systematic random would be the best way. A threshold is set to determine if each pixel has sweat gland nerve fibers present or not. The percent innervation is the number of positive pixels divided by all pixels on any part of a sweat gland that is in the field.

Sweat Gland Nerve Fiber Density: manual

Every whole sweat gland in the chosen sections was included. A grid of ten micron diameter circles, spaced at 50 microns in X and 25 microns in Y, was super-imposed on the image (Gibbons et al., 2010). The circles are small enough so that only one nerve can usually intercept them resulting in two intersections. Intersections of sweat gland nerve fibers with the circles are counted, and the results expressed as number of circles intercepted/number of circles. This technique will not estimate the length and it will miss intersections that are perpendicular to the plane of sectioning.

Sweat Gland Nerve Fiber Density: unbiased stereology

Unbiased stereology was used as a ‘gold standard’ to compare with the other methods. The cycloids for Lv probe was used on complete sweat glands (Gibbons et al., 2010, also see, Gibbons et al., 2012 and Stocks, et al., 1996). Just six years later however, the paper introducing spherical probes for length, or spaceballs, was published (Mouton et al., 2002). Spaceballs is far easier to use than Cycloids for Lv, because there is no requirement for vertical sections, the probe itself is isotropic!

RESULTS

The intra-epidermal nerve fiber density was lower in diabetic than in non-diabetic patients, and in diabetic subjects correlated with neuropathy.

For sweat gland fibers, when the manual (circle assay) and the automated (automatic segmentation by setting a threshold) were compared with unbiased stereology (cycloids for Sv) there was a good correlation. All three methods showed lower sweat gland never fiber density in diabetic than in non-diabetic subjects, and the lack of fibers correlated with the amount of neuropathy. The manual method had tighter data; the authors speculate it is because with the automatic method the thresholding can be arbitrary and the thickness of the fibers, not just the length is included in the analysis. The manual and the automated method for looking at sweat gland fibers correlated with the intra-epidermal nerve fiber density, but the semi-quantitative method for scoring sweat gland fibers (subjective score assigned by judges) did not. Reliability scores that compared performance among observers were also worse for the semi-quantitative method than for the manual or automated method.

DISCUSSION

This paper is useful because it shows the semi-quantitative method does not work, especially when used with a diabetic group with less severe neuropathy. The authors, however, also say the automated method and manual method are just as good as unbiased stereology, and that the automated method is faster but the manual method takes length into account only (not thickness) and doesn’t involve the vagaries of thresholding. Despite this they trumpet the automatic method. Unbiased stereology for length estimation is much less cumbersome now since spaceballs has been invented and neither cycloids for Lv, requiring vertical sections, nor image plane as probe, requiring isotropic sections, has to be used. With Spaceballs, the tissue can be oriented any way the researcher wants, that is, preferentially and the researcher has the advantage of not relying on thresholding (automated method) and of using proven and accepted stereological formulas, which is not the case for the manual method.

RECOMMENDATIONS

If the sections are thick, use spaceballs to estimate length. If the sections are thin, you will have to use isotropic sections and you can use the image plane as the probe.

 

Gibbons, C.H., Illigens, B.M.W, Wang, N., R. Freeman (2010) Quantification of Sudomotor Innervation: a Comparison of 3 Methods. Muscle Nerve, 42, pp. 112 – 119.

Gibbons, C.H., Illigens, B.M.W., and N. Wang, et al. (2012) Quantification of Sweat Gland Innervation: A Clinical-Pathological Correlation, Neurology, 72, p. 1479-1486.

Mouton PR, Gokhale AM, Ward NL, West MJ. (2002) Stereological length estimation using spherical probes. J Microsc., 206, pp 54-64.

Stocks, E.A., McArthur, J.C., and P.R. Mouton (1996) An Unbiased Method for Estimation of Total Epidermal Nerve Fibre Length, J. Neurocytology, 25, pp. 637 – 644.