Today, skin biopsies can play an important role in the diagnosis of peripheral nerve disorders and have yielded another diagnostic tool for the neurologist. One of the commonly reported neuropathologic abnormalities observed in skin biopsies is a reduction of epidermal nerve density. Analyzing the changes in the morphology and density of epidermal nerves is of immense diagnostic and prognostic value in peripheral neuropathies. These changes also provide an assessment of disease progression and of tissue responses to regenerative treatments. Combined with immunohistochemical studies, newly evolved skin biopsy and epidermal count techniques have the potential to provide significant information about the pathogenesis of many peripheral nervous system diseases. They have great potential for impacts on both research and clinical approaches to treatment. Evolution of a standardized and validated counting methodology and significant advances in procuring skin biopsies have opened up a wide spectrum of applications that make the technology easy to apply in practice. The application of this technology may lead to early detection of many common peripheral nerve diseases and an enhanced understanding of disease onset and progression. In this article we review the state of current research and clinical practice in the use of skin biopsies and epidermal nerve densities.
The skin is the largest organ in the body, covering the entire external surface of the human body. It serves as a protective barrier, facilitates in sensory perception, immunologic surveillance, and thermoregulation and controls insensible fluid loss. It constantly regenerates throughout life as its outer layers are shed and are replaced by inner layers. The thickness of the skin varies according to anatomic location, sex, and age of the individual.
The skin is densely innervated (Fig. 1A) and sensory impulses from it are transmitted to the brain by fibers of primary sensory neurons located in trigeminal and dorsal root ganglia. These neurons are a heterogeneous population and include both mechanoreceptors and nociceptors. Nociceptive neurons detect noxious thermal, mechanical, and chemical stimuli that evoke a painful sensation and can warn of impending bodily injury. The peripherally directed axons of dorsal root ganglia neurons run in peripheral nerves, terminate in skin, and innervate a variety of cutaneous structures, including sweat glands (Fig. 1B), hair follicles, Merkel cells, Meissner's corpuscles, blood vessels and the epidermis (1,2).
In the skin, the nerve bundles typically course through the dermis vertically and form horizontal subepidermal neural plexuses in the papillary dermis. From the papillary dermal plexuses these nerves travel vertically, lose their Schwann cell investment at the dermoepidermal junction, penetrate the epidermal basement membrane, ascend between the keratinocytes, and terminate in the keratin or superficial layers of the stratum spinosum as free nerve endings (Fig. 1C).
Cutaneous innervation consists mainly of unmyelinated fibers, which account for approximately 90% of all dermal nerve fibers (3). The unmyelinated axons are organized into Remak bundles, which contain Schwann cells enveloping 1 or more unmyelinated axons and the profile is enclosed by continuous basal lamina. The individual Remak bundles in the dermis typically contain 2 to 3 axons (Fig. 1D). Larger and deeper dermal nerves have a complete perineurium with a basal lamina and junctional complexes. Many smaller and more superficial nerves lack perineurium and are surrounded by collagen bundles oriented in the long axis of the nerve fibers, resembling endoneurial collagen.
Morphologically, the sensory axons within the skin are classified into myelinated (A-fibers) and unmyelinated (C-fibers) fibers. Neurophysiologically, they are categorized as A-δ and C-fibers. A-δ fibers are small, thinly myelinated axons subserving mechanoreceptors and thermal receptors that transduce and transmit pain (nociception). C-fibers are very thin unmyelinated axons that slowly transmit nociception sensation. Cool threshold detection is mediated by A-δ fibers and C-fibers and warm sensation by C-fibers (4). Slow-conducting unmyelinated (C) afferents also signal light touch (5). Axonal conduction velocity depends on both the myelin thickness and nerve fiber diameter, and unmyelinated C-fibers conduct impulses at about 1 m/s whereas myelinated A-fibers conduct at 40 to 1,000 m/s (2). Large myelinated Aα and Aβ fibers conduct vibration and light touch, and these fibers are not present in the epidermis (1).
Neurochemically, epidermal nerve endings can be broadly categorized as either peptidergic or nonpeptidergic. Peptidergic axons densely express calcitonin gene-related peptide (CGRP) along with other peptides including the neurotransmitter substance P. Nonpeptidergic afferents are identified by their binding sites for the plant lectin IB4. Using genetically encoded axonal tracers expressed from the Mrgrd locus, a subpopulation of small-diameter (presumed to be nociceptive, IB4+) nonpeptidergic epidermal axons that express the molecular marker Mrgprd (Mrgprd+) have been identified in mice. These Mrgprd+ fibers course through the stratum basalis and stratum spinosum and terminate in the stratum granulosum. In contrast, CGRP+ fibers terminate in the stratum spinosum of the epidermis. Mrgprd+ fibers do not innervate any specific cutaneous sensory structure and there is complete absence of Mrgprd+ fiber innervation in all other peripheral tissues. The central projections derived from distinct epidermal innervation zones terminate in adjacent laminae in the dorsal spinal cord (6).
Sweat glands in the dermis are highly immunoreactive to pan-axonal marker protein gene product 9.5(PGP 9.5) (Fig. 1B). The majority are vasoactive intestinal polypeptide+ and some are CGRP+ but are sparsely innervated by substance P-immunoreactive fibers (7,8). Characteristically, these fibers are cholinergic sympathetic axons with a few noradrenergic fibers. Cholinergic/noradrenergic coexpression, which is a mode of autonomic regulation, is a unique feature of the primate peripheral nervous system and has not yet been identified in other portions of the human sympathetic nervous system (9). Dermal blood vessels are innervated by both sensory and autonomic fibers (sympathetic and parasympathetic), and most of these are unmyelinated (Fig. 2A). The autonomic fibers lie significantly closer to the endothelial cell layer and smooth muscle cells compared with the sensory fibers (10).
Neurotrophic factors play an important role in the growth and maintenance of these terminal nerve endings. C-fiber nociceptors respond to trophic factors: nerve growth factor (NGF), neurotrophin 3, glial cell line-derived neurotrophic factor, and brain-derived neurotrophic factor (11,12). NGF binds to 2 classes of transmembrane protein receptors: tyrosine kinase family (high-affinity receptor) and p75 (the low-affinity and pan-neurotrophin receptor). The p75 receptor can be activated even in the absence of tyrosine kinase A. Expression of both NGF and p75 are important in the regeneration of peripheral nerves after injury (13,14). NGF and its receptor, tyrosine kinase A, are important for survival and target innervation of small diameter sensory neurons and sympathetic neurons during development (15). Expression of NGF is maintained in the adult, especially in target tissues such as skin keratinocytes, indicating a possible role in maintenance and plasticity of these neurons in adulthood. Previous experiments in rats using an NGF antiserum infusion addressed this question (16) by demonstrating that the levels of NGF in sympathetic and sensory neurons after axotomy partly regulate subsequent changes in neuropeptide expression. Yet, it remains unclear whether NGF is important for maintaining survival or innervation of sensory and sympathetic neurons in the adult nervous system.
Analysis of Skin Biopsies in Cutaneous Nerve
Large sensory and motor nerve fibers in peripheral nerve disorders are typically assessed using electrodiagnostic methods or sural nerve biopsy. However, electrodiagnostic methods are not useful in assessing unmyelinated nerve fibers simply because this population is "invisible" to nerve conduction velocity studies. Within the epidermis, the individual unmyelinated axons are separated from each other as they pass between the keratinocytes, and at the light microscopic level it is possible to identify and count axons using immunocytochemical stains. Before the advent of sensitive pan-axonal markers such as PGP 9.5 (17), silver stains were used but were too insensitive for practical use. For almost 15 years skin biopsy has been used as a reliable tool to assess the epidermal and dermal nerve fibers. The skin can be sampled at several anatomical sites simultaneously, and biopsies can be repeated serially over time (18-23).
Sampling of skin to analyze epidermal nerves is done essentially by using 2 techniques: skin blister and skin punch techniques.
Skin Blister Technique
A suction capsule is placed over the skin and the epidermis is separated from the dermis at the level of the epidermal basement membrane (24). The technique is least invasive, painless, and bloodless, but the procedure is time-consuming and the relationship of dermal nerves to epidermal nerves cannot be studied.
Skin Punch Technique
Cutaneous nerve evaluation using 3-mm skin punches (8,22,25) is easy to perform, no sutures are required, and the test leaves behind only a very small scar. The major advantages of skin biopsies are intraepidermal nerves fibers (IENFs) can be quantified, dermal innervation and morphologic features can be evaluated, and sections can be used for ultrastructural studies, polymerase chain reaction analysis, and laser pressure catapulting studies. In addition, from skin punches, epidermal sheets can be separated and subjected to protein and mRNA analysis.
Processing and Immunohistochemistry
Most laboratories use either a PLP (paraformaldehyde, lysine, and periodate) or Zamboni (2% paraformaldehyde and picric acid) fixative, as tissues fixed in formalin tend to produce a fragmented appearance of nerve fibers (22,26). PGP 9.5 is a cytoplasmic neuronal marker commonly used to investigate small caliber sensory fibers in skin biopsies, and anti-PGP 9.5 targets ubiquitin carboxyl-terminal esterase L1, an enzyme found exclusively and ubiquitously in neurons (27). Other markers are antibodies against specific cytoskeletal filaments (Fig. 2B), axonal membrane, and epitopes (28,29).
Fifty-micrometer-thick sections are cut perpendicular to the epidermis to visualize both dermis and epidermis using a freezing sliding microtome. At least 3 sections of 50 μm from each biopsy are evaluated using bright field light microscopy. Thick sections allow the visualization of cutaneous nerves, as continuous wavy bundles and as single nerves.
Quantitation of Epidermal Nerves
This involves 2 distinct steps: counting the number of IENFs under light microscope at 40× magnification, and then measuring the length of the epidermis along the upper margin of the stratum corneum. The linear IENF density is calculated and expressed as the number of fibers per millimeter of epidermal length (26,30). Two different counting rules have evolved: 1) counting only the number of nerve fibers crossing the epidermal basement membrane (Fig. 3A), and 2) counting that includes isolated nerve fragments in epidermis that do not cross the basement membrane (31,32) (Fig. 3B). We prefer to use the latter count rule as it provides the advantage of including the isolated fragments in the count and a higher IENF number with less error is likely to be obtained. This lower error becomes critically important in evaluating neuropathic sites where only a few fibers are present. Measuring the length of epidermis using both computerized image software and a microscope intraocular lens ruler has produced similar results (33). IENF density can also be derived simply by dividing the number of IENFs by 3 mm, and this result has correlated significantly with IENF density measured by image analysis (34). Very good correlation exists between the various numerical methods (35). The European Federation of Neurological Studies task force has established guidelines and rules for counting and reporting IENFs (31,32).
With strict counting rules and intensive training, a high degree of inter- and intrarater reliability has been achieved in IENF density estimation. Density variations among adjacent sections within the same site are minimal, and a high degree of correlation of IENF densities between 2 punches at the same anatomical site has been demonstrated (26,36). When the lowest 5th percentile of the normative range in the distal leg is used as a cut-off value to identify normal, the technique had a diagnostic efficiency of 88%, a positive predictive value of 75%, and a negative predictive value of 90% for the detection of small fiber sensory neuropathy (26). Although this study used formalin fixation, IENF density differences between formalin and PLP fixation have not been appreciative (37). In contrast, estimation of IENF using confocal microscopy is expensive and labor intensive. Fifty- to 100-μm-thick sections are used for confocal studies (23,38,39). As the IENF density depends on the optical sections created by the user, wide variations in results occur, and normative data have not yet been established using this technique.
IENF Density As a Diagnostic Tool
IENF density normative data are available for only a few anatomical sites: proximal thigh, distal thigh, distal leg, distal forearm, trunk and heel. Three major studies have shown that in healthy subjects, IENF density is typically higher in the proximal thigh than the distal leg with a normal proximal to distal length-dependent gradient in innervation (26,37,40) (Table, Fig. 4A, B).
In normal individuals, higher IENF densities are found in very young individuals (<20 years). The effect of gender, age, and height on distal leg innervation has been investigated by several groups. A study that focused on subjects aged 16 to 82 years (26) did not find an effect of increasing age other than subjects in the youngest decile (16-20 years) having higher densities. Other studies of healthy control subjects have reported a mild inverse, gender-adjusted relationship between IENF density and age with a decrease of 0.6 to 1.8 fibers/mm per decade (37,40). It appears that excluding the extremes of age, the effect of age on epidermal innervation is modest, if any. Similarly, there has been variation in the role that gender plays with some studies reporting no effect (26) and others finding mild increases in epidermal innervation among females (37,40). No significant relationship of race, height, or weight on IENF density has been reported (26,37). Interestingly, height, presumably as a surrogate for axonal length, is a risk factor for the development of diabetic neuropathy and has been linked to reduced rates of IENF regeneration rates (48). The epidermal nerve fiber density measure has shown high specificity and reliability and published data in the normal population could serve as a good baseline for detecting small fiber sensory neuropathy (26,40,49,50). Studies have estimated IENF in the forearm, trunk, and heel, but these data have not been consistent (22,41,51).
In patients with sensory neuropathy patients the IENF density is consistently and significantly lower than that in controls (Fig. 4C, D) (20,23,39,50,52-54). The IENF density measurements were found to be more sensitive than sensory nerve conduction studies for diagnosing small fiber sensory neuropathy (19,23,42,43,50,55).
Morphologic changes are seen in both epidermal and dermal nerves in peripheral neuropathy and sometimes in otherwise healthy individuals from local effects of trauma. Although some small-sized swellings, beading, and varicosities may be present in the leg skin biopsy of normal individuals, attenuation of fibers, large globular and fusiform-shaped swellings (Fig. 5A), dystrophic changes, and tortuous and increasingly complex branching are some of the morphologic changes that have been described in the skin biopsies of patients with neuropathy (3,18,22,42,44,56,57). Swellings were more numerous in the calf of patients with small fiber neuropathy and also tended to be more numerous in diabetic thighs (57).
The higher density of swellings in the distal leg of neuropathic patients correlates well with impaired heat-pain threshold, development of symptoms, and progression of disease (42,44). These axonal swellings are thought to presage the subsequent loss of axons; that is, large swellings serve as predictive markers of nerve fiber degeneration. In a study of 28 patients with sensory complaints of unknown etiology, large axonal swellings on the initial skin biopsy subsequently showed a decline in IENF density on repeated biopsies. This decline was not observed in patients without nerve fiber swellings or in those with smaller nerve fiber swellings (18).
Ultrastructural examination of skin biopsies from patients with neuropathy have shown that dilated cytoskeletal organelles, disintegration of axoplasmic organelles, watery axoplasm, and accumulation of abnormal mitochondria within axons and Schwann cells contribute to the formation of axonal swellings in terminal nerve endings (Fig. 5B). Pathologic changes were similar in all types of sensory neuropathic cases examined without any specific recognizable differentiating features. Thus, nerve fiber swellings are now considered to represent a predegenerative change, and the presence of larger swellings at the lower limb can be used as a predictive marker to assess the progression of neuropathy. Further studies are needed to describe a methodology to quantify the swellings.
Intraepidermal Nerve Fibers Density in Peripheral Nerve Disease
It is estimated that the global prevalence of diabetes mellitus in people of all ages will increase from 2.8% in the year 2000 to 4.4% in 2030. The total number of people with diabetes is projected to rise from 171 million in 2000 to 366 million in 2030 (58). Epidemiologic studies suggest that 45% to 60% of all diabetic patients develop neuropathy (59,60). Abnormalities on nerve conduction studies (NCS) or electromyographic changes were found in 18% to 25% of patients at the time of diabetes diagnosis (61).
Direct metabolic effects due to hyperglycemia, oxidative injury, mitochondrial dysfunction, and disruption of axonal transport by advanced glycation end products have been implicated in the pathogenesis of neuropathy (62,63).
A number of epidermal nerve quantification studies have been completed in diabetic patients and in experimental diabetic animals. IENF densities are reduced in type 2 diabetic patients compared with those in age-matched healthy controls and correlate with changes in warm sensory detection thresholds and the amplitude of the sural sensory action potential. The extent of denervation increases with the duration of type 2 diabetes (43). There is an inverse correlation between IENF density and severity of neuropathy (20,64), with progressively decreasing densities from proximal to distal sites (39). IENF density is also lower in those presenting with neuropathic pain compared with those presenting without such pain (64). Differences between people with type I versus type II diabetes have not been identified although severity of IENF loss has been linked to the severity of glucose dysmetabolism (50).
Increased incidence of impaired glucose tolerance is seen in neuropathy patients, particularly in those with pain and sensory loss and is more common in patients with apparently idiopathic polyneuropathy than in the age-matched general population (61,65). IENF density is reduced even in patients with neuropathy who have impaired glucose tolerance neuropathy with normal NCS (66). The neuropathy associated with impaired glucose tolerance is generally milder than the neuropathy associated with diabetes mellitus. Small nerve fibers are prominently affected and may be the earliest detectable sign of neuropathy in glucose dysmetabolism (45,50,67,68).
By using a human model of IENF regeneration, it has been shown that reduced rates of nerve regeneration occur in diabetic subjects without any evidence of neuropathy (29), providing a rationale to include subjects without overt neuropathy in trials of regenerative agents. Pancreas transplantation has not produced the expected major beneficial effects on small nerve fibers, probably because of the reduced regeneration capacity of patients with advanced diabetes. Severe depletion of epidermal and dermal fibers was noted in both lower limbs during the post-transplant period (69). To register nerve fiber regeneration, prolonged observation periods will probably be necessary, and IENF quantification can be used to assess the progress of polyneuropathy during post-transplant follow up.
Although it is generally believed that diabetic neuropathy is due to chronic hyperglycemia, experience from patients with insulinoma and experimental studies show that periods of hypoglycemia from overcontrol of glucose levels may also cause neuropathy in diabetic patients. The plantar nerves of diabetic eu-/hypoglycemic BB/Wor rats treated with insulin implants exhibited a distinct neuropathy, which is accompanied by mild alterations in the epidermal innervation of plantar skin and a more obviously abnormal nerve terminal pattern in plantar muscle (70).
Human Immunodeficiency Virus-Associated Sensory Neuropathies
Various types of peripheral neuropathies have known to be associated with human immunodeficiency virus (HIV) infection, and 10% to 35% of adults with AIDS are known to develop HIV-associated sensory neuropathies (HIV-SN) (71,72). Two subtypes are manifest: one predominantly associated with HIV infection (HIV-associated distal sensory neuropathy), and the other associated with specific neurotoxic antiretroviral drugs (antiretroviral toxic neuropathy). IENF density is typically reduced in a length-dependent manner in adult patients with HIV infection who have neuropathic signs independent of neuropathic symptoms (44).
Follow-up of HIV-infected medically asymptomatic, non-neuropathic individuals over a period of time has shown that a lower leg IENF density, a higher cooling threshold, and a higher heat pain threshold for minimal pain were found to be associated with a greater risk of transition to symptomatic HIV-associated distal sensory neuropathy. There is a longitudinal association between leg IENF and quantitative sensory testing, and serial small fiber evaluation in chronic neuropathies can be useful in predicting patients at risk for transition to HIV-associated distal sensory neuropathy (73,74). Decreased IENF density was associated with higher levels of neuropathic pain, higher plasma HIV RNA levels, and lower CD4 counts (54).
In subjects with advanced HIV, IENF density correlates with severity of neuropathy severity and with the level of neuropathic pain quantified by both the Gracely Pain Scale and the Visual Analogue Scale. IENF density at the distal leg also correlates with sural amplitude, but IENF at both distal leg and proximal thigh correlated negatively with toe cooling detection threshold. In contrast to the study by Polydefkis et al (54), IENF density from patients receiving highly active antiretroviral therapy was not associated with exposure to neurotoxic antiretroviral drugs, CD4 cell count, or plasma log HIV-RNA level. These findings suggest that in this group of patients with advanced HIV, the above previously established risk factors for HIV and HIV-SN are no longer determinants of the severity of HIV-SN in the current highly active antiretroviral therapy era (74).
In a prospective cohort study HIV-infected patients were examined at 2 different geographic locations to emphasize demographic differences. Reduced vibration thresholds and epidermal nerve fiber densities had the highest diagnostic efficiency of the laboratory indicators of neuropathy examined, but were found to be relatively insensitive in isolation (75). IENF density in longitudinal studies can be used to identify risk factors for neuropathy progression and to monitor epidermal nerve fiber regeneration in clinical trials of regenerative agents.
Sensory neuropathy also occurs in a significant proportion of HIV-infected pediatric patients (76,77). Twenty-eight HIV-seropositive patients aged 5.6 to 18.2 years were compared with 48 HIV-seronegative subjects aged 6.5 to 17.8 years. NCS and quantitative sensory testing did not significantly differ between patients and control subjects, but HIV-seropositive patients as a group showed a higher IENF density than control subjects. This paradoxical increase in innervation indicates age-specific differences in the expression of HIV-SN and the difficulty in establishing the presence of peripheral neuropathy in the pediatric and adolescent group (C. Luciano, unpublished observation). Regeneration studies using human models of collateral (excision model) and regenerative (capsaicin model) sprouting showed reduced rates of epidermal nerve fiber regeneration. This abnormality was more pronounced in those with evidence of baseline neuropathy but was also observed in those with normal peripheral nerve function and no neuropathy symptoms. The rate of regeneration was highly dependent upon the baseline distal thigh IENF density but was independent of CD4 cell count, HIV viral load, HIV duration, gender, race, or apolipoprotein E4 status. Regeneration was not further reduced in those subjects exposed to dideoxynucleoside medications. Swellings were not a prominent feature among subjects in that study, suggesting that reductions in regeneration may be the earliest change that occurs. These findings also imply that there is a progression from the introduction of an insult such as infection with HIV to development of a reduced regeneration rate followed by morphologic changes and finally the establishment of peripheral neuropathy. In this context, an improvement in regeneration would be the earliest step toward the improvement of nerve function in the setting of a regenerative neuropathy study (78).
To date, no agents have been shown to be effective in speeding neuronal regeneration in humans although a therapeutic agent, acetyl-l-carnitine, which is vital for mitochondrial dysfunction, has been reported to enhance epidermal innervation in HIV-SN (79). A trial of recombinant human erythropoietin is now underway because of its presumed neuroprotective properties (80,81).
Idiopathic Small Fiber Sensory Neuropathy
Patients with a clinical diagnosis of possible idiopathic small fiber neuropathy often present with painful burning feet, but on clinical examination show normal strength, proprioception, tendon reflexes, and normal electrophysiologic responses. These patients often have a reduced IENF density at the calf with normal proximal nerve fiber densities, indicating that idiopathic small fiber neuropathy is typically a length-dependent process (19,82). Skin biopsies can be useful to distinguish neuropathies from radiculopathies, which typically do not produce epidermal nerve changes because the damage occurs proximal to the dorsal root ganglion. Follow-up skin biopsies of patients with idiopathic small fiber sensory neuropathy with durations ranging from 12 to 28 months show a decrease of IENF density in the leg, indicating the progressive destruction of these fibers in the disease (42).
Small and large fibers are not evenly distributed within cutaneous nerves, and this inconsistency increases the complexity of morphometric quantification of nerve fibers in biopsies such as those of the sural nerve. One study comparing sural nerve biopsy to skin biopsy demonstrated that a reduction in IENF density was the only evidence for small fiber neuropathy in 23% of patients (55). In 1 large study skin biopsy was found to be more sensitive than a quantitative sudomotor axon reflex test or quantitative sensory testing in diagnosing small fiber neuropathy (23). In another study with painful feet of unknown cause, an excellent correlation was observed between results of the quantitative sudomotor axon reflex test, cooling abnormalities, and loss of IENFs (83). Although sural sensory nerve action potentials have served as the traditional electrophysiologic marker of distal sensory polyneuropathy, in patients with suspected distal sensory polyneuropathy who have normal NCS measuring both medial plantar sural nerve action potentials and small sensory fibers will serve as complementary tools in evaluation of distal SN (23,84,85).
Leprosy is a chronic granulomatous disorder affecting multiple peripheral nerves and skin. Cutaneous nerve fibers are affected in all types of leprosy (86, 87), resulting in anesthetic or hypoesthetic skin lesions. In a large multicentric cohort study in multibacillary patients, unmyelinated C-fibers were found to be more frequently affected than small myelinated Aδ fibers and NCS, and warm and cold threshold measurements have emerged as the most promising tools for early detection of leprosy (88). One of the major challenges in leprosy control programs is to diagnose indeterminate skin lesions, relapses, and dermatologic conditions that may mimic leprosy. Although no attempt has been made to quantify epidermal nerve fibers in leprosy skin lesions, a large study evaluated cutaneous innervation and neuropeptides using immunohistochemical methods. Neuropeptide immunoreactivity was seen in only 14% of the indeterminate leprosy specimens and was completely absent in other types of leprosy, highlighting the diagnostic significance of early disappearance of neuropeptide immunoreactivity in leprosy lesions (89). Because the 3-mm punch technique is simple to perform, nontechnical staff can be trained in the quantification technique, and IENF density of skin lesions could serve as an additional tool along with skin smear examination and NCS to increase the diagnostic efficiency of detecting neuropathy in leprosy control programs.
Chemotherapy-Induced Peripheral Neuropathy
Chemotherapeutic agents produce both acute neurotoxicity and chronic neuropathy (90, 91). A number of effective chemotherapeutic agents, including platinum compounds, taxanes, and vinca alkaloids produce length-dependent axonal injury and even worsen preexisting neuropathy (92). In an experimental rat model of paclitaxel- and vincristine-evoked painful peripheral neuropathies, a partial degeneration of the sensory innervation of the epidermis and upregulation of PGP 9.5 in epidermal Langerhans cells have been demonstrated in glabrous skin of the plantar hind paw (93). IENF density with total neuropathy score along the course of chemotherapy could help to predict the course and severity of chemotherapy-induced peripheral neuropathy.
Excess pyridoxine (vitamin B6) ingestion causes a dose-dependent peripheral neuropathy characterized by necrosis of dorsal root ganglion and degeneration of long myelinated fibers in the sciatic nerve. In experimental rats as well as in human subjects lower doses of pyridoxine also result in development of small fiber involvement as the earliest and predominant abnormality, and higher doses lead to involvement of both large and small fibers (94-96). We have observed 1 patient who developed a sensory neuropathy after consuming excessive amounts of pyridoxine. Serial skin biopsies demonstrated epidermal denervation that recovered over a period of several months (V. Choudhry, personal communication, 2007).
Skin Biopsy in Non-Length-Dependent Neuropathies
In patients with sensory ganglion degeneration, symptoms of ataxia and proprioceptive sensory loss predominate and the distribution of sensory impairment is widespread and asymmetric (97-99). A study compared biopsies from patients with ganglionopathy and axonal neuropathy with those from normal control subjects. Skin biopsies at the proximal thigh and the distal leg in patients with axonal neuropathies showed significantly lower values at the distal site of the leg, confirming the length-dependent loss of cutaneous innervation. In the patients with ganglionopathy, in contrast, the degeneration of small-diameter sensory fibers was seen equally in all locations (46). It should be noted that a severe length-dependent neuropathy will also show denervation at all sites and cannot be clearly differentiated from ganglionopathy on a neuropathologic basis alone.
Herpes zoster is caused by reactivation of latent varicella zoster virus within the sensory ganglia, resulting in destruction and degeneration of the respective spinal and peripheral axons (100). Several pathologic processes for the cause of postherpetic neuralgic pain have been suggested. IENF was examined in chronic postherpetic neuralgia-affected skin, and the loss of cutaneous innervation inversely correlated with allodynia. It is thought that surviving cutaneous primary afferent nociceptors spontaneously became active and sensitized and contributed to postherpetic neuralgic pain and allodynia (101). Some have suggested that degeneration of primary afferent neurons causes central hyperactivity (102). These studies expressed epidermal density as an average number of epidermal neurites/mm2 of skin surface area (101), a method of counting that is generally not used for other neuropathies.
Skin Biopsy in Inflammatory Demyelinating Neuropathies
Guillain-Barré Syndrome (GBS) or acute inflammatory demyelinating polyneuropathy is an acute inflammatory neuropathy, and, according to various clinical, neurophysiologic, and pathologic studies large-diameter myelinated nerves are affected. Axonal degeneration of motor and sensory nerves has been described in the demyelinating form of GBS (103). Skin biopsy from the distal leg of patients with the demyelinating form of GBS was investigated for the involvement of small sensory fibers. IENF density was reduced in 55% of the cases in comparison with age- and gender-matched controls, and the values correlated with functional disabilities. These findings suggest that GBS is not a purely large-fiber neuropathy but that small-fiber sensory and autonomic neuropathies exist in a significant proportion of patients with GBS (104).
Chronic Inflammatory Demyelinating Polyneuropathy
When IENF densities and thermal thresholds in patients with chronic inflammatory demyelinating polyneuropathy were tested, patients with chronic inflammatory demyelinating polyneuropathy showed lower IENF density than that of control subjects. The low IENF density and elevated thermal threshold were associated with autonomic symptoms. Patients with chronic inflammatory demyelinating polyneuropathy have small-fiber sensory and autonomic neuropathies in addition to the immunologic inflammation in large diameter myelinated fibers (105).
Sarcoidosis is a chronic progressive granulomatous multisystem disorder that may affect virtually any part of the nervous system, and patients can present with peripheral nerve pain and fatigue (106,107). Patients with sarcoidosis with normal nerve conduction studies had shown reduced temperature sensitivity and low IENF in the leg compared with control subjects. Small fiber sensory neuropathy occurs in sarcoidosis, and reduced IENF may be the only detectable abnormality in patients with sarcoidosis presenting with peripheral nerve pain (53).
Hepatitis C Infection
The most frequent form of hepatitis C infection (HCV)-associated neuropathy is distal sensory or sensorimotor polyneuropathy (108). Most cases occur in association with cryoglobulinemia (109) and vasculitis (110). A multicentric study in an unselected, untreated referral population with HCV infection has shown that electrophysiologic peripheral neuropathy occurs in 15.3% of patients (111). Coinfection with HIV and HCV has become increasingly prevalent and distal symmetric polyneuropathy is the most common form of peripheral neuropathy in HIV and HCV infections. However, in a cohort study of 147 HIV-infected adults, HCV was not found to be significantly associated with HIV-related sensory neuropathies (75).
Vasculitis is one of the rare causes of peripheral neuropathy, and clinical manifestations include motor weakness, reduced sensitivity, and neuropathic pain. Small myelinated nerves were shown to be more susceptible to ischemia than larger myelinated fibers in experimentally induced ischemia in Wistar rats (112). Patients with motor and sensory impairments have shown significant IENF reduction, demonstrating that the depletion and degeneration of small-diameter sensory nerves in vasculitis occur as frequently as those of large-diameter nerves (113,114).
Skin Biopsy in Autoimmune Diseases
Skin biopsies have played an important role in identifying small fiber neuropathy in chronic inflammatory autoimmune diseases.
Sensory neuropathy with prominent ataxia and ganglionitis are well-recognized forms of neuropathy associated with Sjögren syndrome. Sural biopsies show a wide range in loss of both large and small myelinated and unmyelinated axons (98,115). A study of 20 consecutive patients with Sjögren neuropathy showed that 60% of the subjects had nonpatchy symptoms and a non-length-dependent pattern of epidermal nerve fiber loss, suggesting that patients with this disorder commonly have a small-fiber sensory neuronopathy rather than a "dying-back" axonopathy (116). This hypothesis has been further confirmed in a population-based study in a small group of patients (117).
Systemic Lupus Erythematosus
The prevalence of peripheral neuropathy in systemic lupus erythematosus (SLE) varies from 5% to 27%, and it is characterized by a mild sensory or sensorimotor neuropathy (47,118,119). Pure small-diameter nerve fiber neuropathy occurs in SLE with a significant reduction in IENF density. Disease activity, disease duration, or routine hematologic, biochemical, or immunologic variables did not influence the number of IENFs in patients with SLE (36,120).
The degree of loss of small-diameter nerve fibers among patients with these chronic inflammatory autoimmune diseases varies. Sixty patients with SLE, 61 patients with primary Sjögren syndrome, and 52 patients with rheumatoid arthritis were compared with 106 healthy subjects. Densities were significantly less in patients with SLE compared with patients with rheumatoid arthritis or healthy subjects. Nerve fiber densities were also reduced in patients with primary Sjögren syndrome compared with healthy subjects. Only 8 patients (13%) with SLE, 2 patients (3%) with primary Sjögren syndrome, and 2 patients (4%) with rheumatoid arthritis have shown densities below the lower reference limit of 3.4 fibers/mm (121). Despite the lower average mean values, very few patients actually meet a diagnosis of small fiber neuropathy.
Celiac disease is an autoimmune disease of the small intestine and the gastrointestinal symptoms result from ingestion of gluten in genetically susceptible people. Neurologic complications are estimated to occur in 10% of affected patients (122), with sensory neuropathy being one of the most frequently reported neurologic abnormalities (123). Despite the association between celiac disease and small fiber neuropathy, a definitive causal relationship has not been established. Interestingly, in patients presenting with asymmetric numbness and paresthesias, low IENF densities occur at the thigh, forearm, and distal leg, suggesting a widespread involvement of small fibers (124). Detection of anti-gliadin antibodies can be useful in diagnosis (125).
This is an autosomal recessive cardioneurodegenerative disorder resulting from an inability to produce the protein frataxin. It is due to extensive trinucleotide repeat expansions in the first intron of the gene encoding frataxin, resulting in defective transcriptions and protein deficiency (126). Patients with Friedreich ataxia have been evaluated by skin biopsy and have shown significant loss of epidermal fibers, reduced autonomic innervation to sweat glands, arrector pilorum muscles, and arterioles in parallel with their clinical findings of impaired thermal sensitivity, tactile thresholds, and decreased mechanical pain detection (127).
Fabry disease is an X-linked recessive disorder caused by deficiency of α-galactosidase that results in accumulation of neutral glycosphingolipids (principally ceramide trihexoside) primarily in endothelial cells (128) of blood-brain/nerve barriers within the central or peripheral nervous systems (129). In the peripheral nervous system it produces a painful small fiber neuropathy (130-132). Patients with Fabry disease may have reduced IENF density at distal biopsy sites with relatively preserved proximal nerve fiber density (133). Twenty-five heterozygous male subjects with Fabry disease did not show an increase in epidermal innervation density after α-galactosidase A (agalsidase-α) enzyme replacement at 6 months. After an additional year of therapy, there was a significant reduction in IENF density in the patient group as a whole, perhaps attributable to the declining glomerular filtration rate and the effects of uremia on peripheral nerves. Thermal thresholds remained unchanged. This study suggests that longer treatment periods may be needed for nerve fiber regeneration or that enzyme replacement therapy alone is not sufficient to reverse the pathology of this disease (134).
Congenital Insensitivity to Pain With Anhydrosis
Congenital insensitivity to pain with anhydrosis or hereditary sensory neuropathy type IV (135) is a very rare disorder and clinical manifestations include mental retardation, congenital analgesia that results in self-mutilation, multiple scars, fractures, and anhidrosis (136). Ultrastructural and morphometric studies of sural nerves show loss of both unmyelinated and small myelinated fibers (137,138). Skin biopsy examination in a 10-year-old girl has shown extensive loss of epidermal nerve fibers and denervation of dermal adnexal structures including sweat glands and arrector pili (138).
Abnormalities in Dermal Innervation
Glabrous skin is richly innervated by myelinated fibers in the dermis and morphologically these fibers appear similar to the myelinated fibers in sciatic and sural nerves. Quantification of myelinated fibers and mechanoreceptors in glabrous skin in normal subjects has shown that the cutaneous segments of Aβ fibers undergo repeated branching and shortening of internodal length during their course from the nerve trunk (139).
Charcot-Marie-Tooth disease is caused by mutations in myelin proteins and evaluation of myelinated dermal fibers in skin has revealed pathologic abnormalities in axolemmal molecular architecture (140,141). Patients with anti-myelin-associated glycoprotein neuropathy have shown IgM deposits on dermal myelinated fibers with a greater prevalence at the distal site of extremities. These findings highlight the use of glabrous skin in identifying abnormalities in inherited demyelinating and chronic demyelinating neuropathies (142).
Many studies have shown reduced sweat gland innervation in patients with peripheral neuropathy (20,22,105,128,139,143). A few researchers have tried different methods to quantify and measure sweat gland innervation. Decreased nerve fiber length around sweat glands in diabetic patients (52), reduced innervation in patients with Ross syndrome (144,145), and familial dysautonomia (146) have been demonstrated. Larger studies with validation of the quantification methods in both healthy control subjects and in diseased conditions are needed for quantification of dermal innervation in sweat glands.
Use of Skin Biopsies in Human Regeneration Studies of Sensory Nerves
As the epidermis is highly innervated by small sensory nerves, skin biopsies provide easy accessibility to study the effect of chemical, thermal, and mechanical trauma on the regeneration and degeneration of sensory nerves. Two types of nerve injury models have evolved: an "incision model" in which, following transection of subepidermal plexes, the sprouting epidermal axons outside the incision line complete the reinnervation by 30 to 75 days. In the "excision model" removal of the incised cylinder of skin resulted in a denervated area, and no complete reinnervation of epidermis was seen even after 23 months (147). The effect of a therapeutic compound, neuroimmunophilin ligand timcodar dimesylate, on collateral nerve sprouting was measured using the excision model. Though the compound did not improve collateral sprouting, the model was consistent across treatment groups and had a low coefficient of variation (148).
Application of capsaicin, an ingredient of hot chilli pepper, to the skin produces a reproducible chemical axotomy of the epidermal nerves. Capsaicin can be administered intradermally or topically. Serial biopsies have shown that nerve regeneration after topical capsaicin is faster than after intradermal delivery (149,150). This "capsaicin model" has been used to study the degenerative (150) and regenerative properties of epidermal fibers (29) after chemical axotomy. The rate of regeneration of IENF was lower in diabetic subjects irrespective of the presence or absence of neuropathy, indicating that abnormalities in peripheral nerve function are present early in diabetes, even before signs or symptoms develop (29). In a randomized, double-blind, placebo-controlled trial, the effect of timcodar dimesylate, a putative regenerative neurophilin compound, on the regeneration of epidermal fibers after capsaicin nerve injury was investigated. There was no difference in the regeneration rate between the treatment and placebo arms, but the baseline IENF density and the height and race of the participants influenced the rate of regeneration (48). These models hold potential for use in clinical trials to study the effect of regenerative drugs and growth factors on sensory nerve fibers.
Skin biopsy analysis of nerve fibers is a technique that has opened a new window to help visualize the previously inaccessible small unmyelinated fibers. Over the last 2 decades there has been a dramatic increase in our understanding of cutaneous innervation, leading to improved diagnostic and therapeutic interventions. The European Federation of Neurological Societies Task Force has formulated guidelines for IENF quantification (32) and summarized the utility of the technique. Normative data from healthy subjects matched for age and gender and from appropriate anatomical sites should be used to interpret neuropathy. Quality control measures must be followed at all levels to ensure the optimal examination of skin biopsies. The laboratory techniques are not automated, and artifacts or errors can potentially be introduced at the stage of sectioning, staining, or interpreting. Intra- and interobserver ratings and quality control between laboratories should be periodically assessed. Skin biopsies could become a critical tool for studies of axon-Schwann cell interactions that are vital for the survival and maintenance of axons. Our data suggest that Schwann cells are lost during prolonged denervation and also in chronic peripheral neuropathies (3). In neuroregenerative trials it is vital to identify group of patients who retain Schwann cells and who are most likely to respond to regenerative drugs. Identifying sites with low or absent IENF or with more normal IENF would help in identifying patients who would progress and/or benefit from therapeutic intervention.
The authors thank Barbara Freeman and Christopher Wright for expert technical assistance.