Structural and functional characterization of nerve fibres in polyneuropathy and healthy subjects

2.3.1 Skin biopsy and staining

One skin biopsy from the distal leg (10 cm above the lateral malleolus) was taken from each participant under sterile conditions, using a 3-mm disposable biopsy punch (Miltex, York, PA, USA). Immediately after obtaining, the biopsy was fixed in 2% paraformaldehyde-lysine-periodate for 18-24h, followed by an overnight cryoprotection in 20% glycerol and 0.08 M Sorenson’s PO₄ buffer.

Each biopsy was cut into 50 μm cryostat sections perpendicular to the skin surface (Microm Cryostat HM 500 OM, Zeiss, Germany), and three sections were sampled systematically and immunostained, using the free floating protocol as described elsewhere using rabbit anti-human PGP 9.5 (1:1000; AbD Serotec, Dusseldorf, Germany) as a primary antibody and horseradish peroxidase-marked goat anti-rabbit as secondary antibody (1:200; DakoCytomation, Glostrup, Denmark) [23, 37].

2.3.2 Analysis

All measurements from skin biopsies were performed using a light microscope (Olympus BX51 microscope) connected to an Olympus DP70 camera (Olympus, Tokyo, Japan), a Heidenhain ND 281 encoder (Heidenhain, Schaumburg, IL, USA), and a Prior Proscan II motorized stage (Prior Scientific Inc., Rockland, MA, USA). Length measurement of each section, length measurement of swellings, NFLD estimations as well as control of the hardware was performed using the newCAST software (Visiopharm, Hoersholm, Denmark).

The epidermal length of three sections was measured, and small fibres crossing the basal membrane to the epidermis were counted under a normal light microscope (60 × oil immersion lens, Olympus UPlanSApo, NA = 1.35) to obtain IENFD, see Fig. 1, right. All sections were counted twice with 4 weeks apart by the same investigator (PRK), who was blinded to the origin of the biopsies during both counts.

Fig. 1

Left: estimation of dNFLD in NewCAST. The green moving line intersecting the dermal nerve fibre represents the intersection between the focal plane and the virtual plane. Nerve fibres that intersect the green moving line when they are in focus get a mark. The sampling box is 90 μπι × 80 μπι Right: example of an IENF with axonal swelling larger than 1.5 μπι. The sections were immunostained for PGP 9.5 and the images were taken with a 60× objective.

2.3.3 Global spatial sampling

The method is described in detail elsewhere [23]. Briefly, to get an unbiased estimation of NFLD, there must be an equal probability for every single fibre in the sample to be hit by a test plane. In global spatial sampling, software generated, randomly orientated and parallel isotropic virtual planes within a virtual 3D box are superimposed over the region of interest on the computer screen, having an equal probability of intersecting the nerve fibres. Every nerve fibre in focus that intersects with the virtual planes is counted, and the length is estimated by the equation below, see Fig. 1, left. In the next field of view, the orientation of the box is changed at random.

Global nerve fibre length density, Ly (nerve fibre/area of interest) is estimated by:

where Σ Q (nervefiber) is the sum of marked nerve fibres in the area of interest (epidermis or dermis), p(box) is the number of corners in one sampling box, avg a(plane) is the average of the sum of areas of isotropic oriented planes in one sampling box, and Σ P(area) is the number of sampling box corners hitting the region of interest. The sampling box volume divided by the plane distance is a (plane). The sampling box height was set to 15 μm, the counting box area was set to 7200 μm², the x- and y-sampling steps were 127 μm × 113 μm, and the plane separation distance was 25 μm.

2.4 Statistics

t-test was used to compare continuous variables between DSP patients and controls. Non-normally distributed data were compared using Mann-Whitney test, and presented as medians with interquartile range (IQR) Associations between structural and functional parameters were determined using linear regression (Sigmaplot version 12.0, Systat Software, Inc., Chicago, IL, USA). We calculated the coefficient of variation (CV), which is defined as the ratio of the standard deviation to the mean. Based on previous experience, we determined that groups of at least 15 subjects would be required for examining the correlation between structural and functional parameters.

4.1 Morphological changes

There is currently no gold standard for diagnosing small fibre involvement in DSP. The diagnosis is typically based on patient’s history, pain distribution and characteristics, and neurological examination including sensory assessment of thermal and mechanical thresholds, as determined in our study by a neurologist (TSJ). While skin biopsies are useful, the literature reports that up to 12% of patients with small fibre neuropathy diagnosed by history and neurologist examination, have normal IENFD [5]. It is therefore important to improve the ability to diagnose these patients based on objective parameters. Among the 17 subjects diagnosed by the above clinical criteria, 13 had abnormalities in all skin biopsy parameters (adjusted for age for IENFD), and additional three subjects were in the lower end of normal values for IENFD, eNFLD, dNFLD and had higher axonal swelling ratios, indicating that both epidermal and dermal nerve fibres undergo degeneration in DSP with small fibre symptomatology [24,38]. One patient had normal skin biopsy results. These results are consistent with previous findings, where we demonstrated a reduced eNFLD and dNFLD in small fibre neuropathy patients compared with healthy controls [23,24].

Nerve conduction evaluation was available from 13/17 patients, which was a part of their clinical evaluation. Seven of the 13 patients had normal nerve conduction studies (NCS), while the remaining six patients showed signs of large fibre involvement (mild-moderate sensorimotor axonal polyneuropathy). The patients with normal NCS had higher IENFD values compared to the patients with large fibre involvement (4.23 vs. 2.74, respectively), a finding that confirms the results from a previous study [41]. These findings suggest that detectable damage to the larger nerve fibres occurs first when there is pronounced damage in the nociceptors.

Comparing morphological changes such as IENFD between studies can be difficult. Immunohistological stainings are difficult to standardize and staining protocols and counting rules may not be identical between studies, all variations that are likely to cause variations between laboratories.

2.3.2 Correlations between structure and function

Several studies have demonstrated a correlation between IENFD, dNFLD and/or eNFLD; one showed that a combination of IENFD and dermal NFLD increases diagnostic specificity and sensitivity in diagnosing small fibre neuropathy [1,23,38]. In the present study, a higher diagnostic sensitivity and specificity were obtained by combining the determined values of both IENFD and epidermal NFLD, thus allowing a better differentiation between DSP and healthy controls. Combining a functional measure (the WSI) and a structural measure (IENFD) improved the sensitivity from 82.4% to 94.2% while the specificity remained unchanged at 84.2%. As shown by Nebuchennykh and colleagues, the choice of reference values is important when considering the diagnostic value of IENFD [42]. The diagnostic sensitivity and specificity of IENFD in a cohort of patients with both pure small fibre damage and also large nerve fibre involvement and different etiologies changed substantially based on different reference values, with ROC analysis showing the best sensitivity (0.77) but Z-score and 5th percentile analysis yielded the best specificity (0.98 and 0.95, respectively).

In healthy controls, the structural skin biopsy parameters correlated with each other better than in DSP patients; it was only in healthy controls that IENFD showed a significant correlation with both epidermal and dermal fibre lengths. These are important findings, as we currently do not know whether pathology that involves nerve fibre density loss is reflected in a similar change in nerve fibre length in the dermis and the epidermis. These may represent different, perhaps independent, measures of small fibre pathology mechanisms.

In DSP patients, HPT correlated with IENFD, consistent with the notion that degeneration of C-fibres (resulting in low IENFD) impairs the ability to detect noxious heat and thus results in increased HPT. In controls, HPT correlated better with eNFLD, further suggesting that epidermal fibre length and density may represent somewhat different small fibre pathology parameters. Surprisingly, other QST parameters such as cold detection and cold pain thresholds did not correlate well with morphological small fibre parameters. This may indicate that the correlation between morphology and function is more direct in heat-sensing C-fibres than in cold-sensing Aδ-fibres. In addition, this phenomenon might be explained by the fact that degeneration and regeneration occur in concert among neuropathy and healthy control subjects. The skin is by nature constantly being replaced and epidermal nerves have a strong capacity to regenerate [39]. One possibility is that hypersensitive regenerating sprouts may counteract sensory loss that would be expected to occur in nerve fibre loss. Alternatively, there may be other structures within the skin that contribute to functional measurement but are not reflected by IENFD, NFLD or axonal swellings. An additional possibility is that while morphological small fibres measures are entirely objective, the assessments of threshold or supra-threshold patient-reported parameters inherently involve central processing and modulation components. Further investigation is needed to elucidate whether psychological, cognitive and other patient-related factors may interfere with peripheral stimulation processing in QST.

Clinically we determined an index for warm sensibility termed WSI, a sensory parameter derived from thermal threshold measurement. It represents the range in which non-noxious heat is perceived and can be used for early diagnosis of small fibre neuropathy [30]. Nine of our 17 DSP patients showed WSI = 0, i.e., complete inability to perceive non-noxious heat, compared to none of the healthy controls. Given that the combination of IENFD and WSI increased the diagnostic sensitivity up to 94.1%, WSI, similarly to HPT, may be another small fibre (predominantly C-fibre specific) measure that better correlates with small fibre loss in DSP.

Topically applied capsaicin activates the TRPV1 receptors located on the nerve endings of C-fibres and causes pain and a local vasodilation response. The response to topical capsaicin was different between DSP patients and controls, with 8 DSP subjects not experiencing any pain from capsaicin application, as opposed to only one such control. Two DSP patients (vs. 6 controls) experienced moderate to severe pain from capsaicin application. It is important to note that the local vasodilation response (flux area) correlated positively with the intensity of capsaicin pain in controls, suggesting an association between the local and centrally mediated effects of small fibre TRPV1 activation by capsaicin. Such correlation, however, was not observed in DSP patients, nor did the morphological small fibre parameters correlate with capsaicin pain response in DSP. The reason for this is not known. While TRPV1 expression in C-fibres may be relatively homogenous in a healthy population, its expression in neuropathy may be substantially altered [40], so that TRPV1-mediated responses may not correlate with small fibre morphology in DSP. One possibility is that hypersensitivity in remaining nerve fibres, exhibit an exaggerated response to TRPV1 stimulation [33]. However, as long as the contribution of degenerating and regenerating fibres to a functional outcome is unknown it is hypothetical what the result of a reduction in nerve fibre count may cause in terms of functional change. A double immunohistological staining of PGP 9.5 and an antibody against the TRPV1 channel would further clarify this, but a reliable antibody against TRPV1 in human skin is not commercially available. In addition, since the pain response, as opposed to the local vasodilation response, is centrally mediated, the ongoing pain in DSP patients may affect the intensity of pain experience following capsaicin application [34]. The fact that MDT correlated with local vasodilation response but not pain response to capsaicin, supports this notion.

This study presents a rigorous comparison between functional and multiple morphological nerve parameters from skin biopsies in patients with DSP and healthy controls, including parameters such as eNFDL and dNFLD that have not been previously evaluated in this context. Cut-off values or normative reference values are needed for the length densities. While all DSP patients had signs of small fibre involvement, one limitation of the study is that the group is heterogeneous in terms of DSP aetiology and large nerve fibre involvement; therefore these findings need to be confirmed in further, larger studies. However, such heterogeneity is common in neuropathy clinics, as only a fraction of patients present with pure small fibre neuropathy. Another disadvantage of the study is the difference in gender ratios between patients (24% female) and controls (63% female) and that the mean age of the patients is 10 years higher compared to controls, but normative IENFD reference values are higher in females compared to males and nerve fibre density declines with age [6]. Supporting the generalizability of our results is the fact that the thermal sensory thresholds observed in this study are in concordance with previously reported results from larger studies [1,7].

This study has several important findings. We observed a differential pattern of correlation between structural and functional parameters in control subjects compared to DSP patients, suggesting a more direct relationship in healthy controls. Furthermore, our findings suggest that combining IENFD with measurement of epidermal nerve fibre length improves the diagnostic sensitivity and specificity of distal symmetric polyneuropathies with small fibre involvement.