Abstract
Objective: To use skin biopsy specimens to quantitate the cutaneous innervation density of Fabry patients who had preserved renal function.
Background: The small fiber neuropathy of Fabry disease is difficult to detect and quantitate by conventional methods. Because this neuropathy is a common characteristic of Fabry disease, quantitating changes in this parameter would be helpful in demonstrating the effectiveness of enzyme or gene replacement therapy.
Methods: Patients underwent skin biopsy at the thigh and foot. Innervation density was determined by counting free nerve endings in the epidermis. These data were compared with nerve conduction studies, and in selected patients, fiber quantitation of sural nerve biopsy specimens.
Results: The Fabry patients had normal results of nerve conduction studies and large fiber quantitation by sural nerve biopsy. However, the involvement of small cutaneous fibers in these patients was easily demonstrable and quantifiable by skin biopsy. All patients showed severe loss of intraepidermal innervation at the ankle, but fiber loss at the distal thigh was proportionately less severe.
Conclusions: The nerve damage in Fabry patients with preserved renal function involves exclusively small myelinated and unmyelinated fibers, and skin biopsy is a useful in detecting and quantitating such damage. Comparison of cutaneous innervation density with quantitation of sural nerve biopsy specimens demonstrated that skin biopsy specimens were as sensitive in detecting the presence of neuropathy as were the nerve specimens. It is speculated that analysis of cutaneous innervation may provide a useful marker of the nervous system’s response to specific therapy for Fabry disease.
Fabry disease is an X-linked recessive disorder caused by deficiency of α-galactosidase A activity.1 Hemizygotes develop deposits of neutral glycosphingolipids, principally ceramide trihexoside,2 throughout the nervous system,3-12 but the primary site of accumulation appears to be vascular endothelial cells. These vascular deposits are thought to be responsible for the clinically significant features of the disease: renal failure, stroke, and myocardial infarction.
Another important aspect of Fabry disease is a painful, small-fiber neuropathy, unique to this disease, which often brings patients to neurologic attention before the other serious manifestations appear. The symptoms of this neuropathy can occur in patients as young as 5 years of age, who characteristically have intermittent bouts of burning, aching pain in the hands and feet. Although the pain may be severe, routine physical examination fails to detect any neurologic abnormality. Moreover, in patients who have not yet developed renal insufficiency, electrophysiologic studies detect no abnormality.13 Sural nerve biopsy has been used to quantitate the nerve damage in Fabry disease,14,15 but it has several drawbacks. First, the procedure is invasive and can lead to chronic pain (e.g., neuroma formation). Second, each sural nerve can be sampled only once, rendering nerve biopsy impractical for longitudinal studies.
Interest in studying cutaneous innervation has been kindled by the ability to visualize intraepidermal axons using antibodies to a panaxonal molecule, PGP 9.5.16-18 By counting the number of free nerve endings visualized in the epidermis, cutaneous innervation density can be determined.17-19 This technique should prove useful in studying the natural history of diseases affecting the peripheral nervous system by allowing quantitation of serial biopsy specimens over time. Such longitudinal studies may also provide valuable information on the regenerative effects of specific therapies on the peripheral nervous system.
To see if analysis of skin biopsy specimens could detect and quantitate the neuropathy in Fabry disease, we recruited hemizygous patients with preserved renal function for skin biopsy evaluation. The innervation densities of these patients were compared with their electrophysiologic evaluations and, in three patients who underwent sural nerve biopsy, with their sural nerve quantitation.
Methods.
Patients.
Twenty Fabry patients (hemizygotes, ages 19 to 56 years) with preserved renal function were examined. Diagnoses were made by enzymatic analysis of cultured skin fibroblasts.20 Eight normal male volunteers (ages 22 to 43 years) underwent skin biopsy to determine normal innervation density. The volunteers were examined by a neurologist and found to have a normal peripheral nerve examination. All Fabry patients in this study showed impairment of cold perception in the distal lower extremities on examination. The neurologic examination for cold perception was done using an aluminum disk 3 centimeters in diameter at room temperature. The disk was applied to the skin on the foot first, and brought proximally until the patient could detect cold.
Inclusion criteria for this study included a creatinine clearance of ≥50 mL/min to minimize the likelihood of neuropathy caused by renal insufficiency.21 Exclusion criteria were concurrent diseases that might cause peripheral neuropathy, e.g., diabetes mellitus and AIDS. All patients enrolled in this study underwent nerve conduction studies (NCS), and 18 of the 20 patients described in this study underwent EMG before enrollment.
Approval of this protocol was granted by the Institutional Review Board of the National Institutes of Neurological Diseases and Stroke. Informed consent was obtained from each participant.
Electrophysiology.
NCS was performed using surface electrodes and standard techniques.22-24 Recordings were made from the median and peroneal motor (including F waves) and median and sural sensory nerves. Limb temperature was monitored and kept above 32 °C at all times. The anterior tibialis muscle was examined by EMG in 18 of the 20 patients before entry into the study; other muscles were tested only if an abnormality was found. Detailed results of the EMG/NCS as well as quantitative sensory testing will be described in a separate article, although preliminary results of the electrophysiology and quantitative sensory testing has been published in abstract form.13
Skin biopsy specimens.
Three-millimeter skin biopsy specimens were removed from two sites from each patient. The distal site was inferior to the lateral malleolus. The proximal site was the lateral thigh, above and lateral to the top of the patella. Lidocaine without epinephrine anesthetized the sites. The specimens were removed and fixed in 10% buffered formalin (pH 7.4) for 1 to 3 days at 4 °C, rinsed in phosphate-buffered saline, and cryoprotected in 20% glycerol/phosphate-buffered saline overnight at 4 °C. After freezing, 50-mm sections were cut on a sliding microtome (Microm 400R, Heidelberg, Germany).
Sural nerve biopsy specimens.
Sural nerves were removed from the right leg of three patients after skin biopsy specimens had been taken. The nerves were fixed in 3% glutaraldehyde/4% paraformaldehyde at 4 °C and processed for light microscopy and electron microscopy (EM).
Immunocytochemistry.
Biopsy sections were stained according to the method of McCarthy et al.17
Quantitation.
Serial 50-mm sections were examined by light microscopy under high magnification (×1,000) and the number of epidermal fibers was counted. A fiber was counted if its free end terminated beyond the nuclei of the basal cell layer. Although the epidermal sheet has numerous infoldings (papillary dermis), which might cause difficulties in distinguishing whether a nerve fiber is within the dermis, in practice this was not difficult. By focusing on individual axons throughout the thickness of the epidermis, it was easy to determine whether the fiber was in the same focal plane as the nuclei of the epidermal cells and thus within the epidermis. This was confirmed in preliminary studies by double-labeling immunohistochemistry using the polyclonal PGP 9.5 antibody and a monoclonal antibody to collagen type IV (to stain the epidermal basal lamina) and visualizing with fluorescein antirabbit and rhodamine antimouse antibodies (not shown). Furthermore, sections were evaluated by laser confocal microscopy, confirming that the stained axons were clearly within the epidermis.
The average number of free axon endings per linear millimeter of each section was determined. A minimum of four sections was evaluated per biopsy specimen, with a minimum of 4 linear millimeters of epidermis quantitated per patient. For sural nerve specimens, 20 electron micrographs were taken of each sample. These were chosen randomly by the intersection of grid bars. Small fibers (myelinated and unmyelinated) were identified by a single observer (J.W.G.), counted, and expressed as fibers per square millimeter.19 Myelinated fibers were quantitated by computer-assisted counting of plastic sections. Results of the sural nerve specimens from the Fabry patients were compared with the normal control values established by the diagnostic laboratory (J.W.G.).
Results.
Innervation of skin and its microscopic appearance. Neurologic examination of the Fabry patients showed consistent, often severe deficits in cold perception in the lower distal extremities. Most also demonstrated abnormalities in the hands, although this was often subtle. Pinprick and vibration sensation on neurologic examination, however, were normal in the Fabry patients. Control subjects showed no abnormalities in cold, pinprick, or vibration sensation.
Antiserum to PGP 9.5 was used to stain axons in the dermis and epidermis of control subjects and patients. In normal epidermis, two types of innervating fibers can be observed. Single fibers can be seen branching off subepithelial nerve bundles to enter the epidermis (figure 1a). In figure 1B, a bundle of axons can be seen entering the epithelium as a group. Axons subsequently separate from the bundle to course independently within the epidermis. We refer to this second type of innervating fiber as complex. Although several axons can often be seen in a bundle entering the epidermis, we do not know whether branching of individual axons occurs. The gross morphology of the intraepidermal fibers was similar in the Fabry patients and the control subjects, although there appeared to be fewer of the complex fibers in Fabry patients with more advanced epidermal denervation.
Figure 1. Innervation of epidermis in normal control subjects. (a) A single, simple fiber can be seen to enter into and terminate within the epidermis. (b) A subepithelial nerve (arrow) sends off a small bundle of axons, which penetrate the basal lamina of the epidermis (arrowhead). Once within the epidermis, individual axons branch off the bundle to form a candelabra pattern of innervation, which we refer to as the complex form of innervation.
Quantitation results: Normal subjects and Fabry patients.
Epidermis from the control subjects had an innervation density of 5.3 ± 0.7 endings per linear millimeter of skin sections at the distal site (not shown) and 18.5 ± 3.5 endings per linear millimeter at the proximal site (table 1). To see whether trauma caused by previous biopsies affects innervation density in adjacent skin, three control subjects underwent biopsy at the distal thigh on separate occasions over an 18-month period. Each biopsy was performed within 5 mm of the previous biopsy. No change in innervation density between biopsies was seen over that time (data not shown). Similarly, Fabry patients underwent biopsy at 6-month intervals for periods ranging from 1 to 3 years. The Fabry patients maintained their innervation density over time, similarly to the control subjects, although two patients demonstrated a rapid decline in innervation density following an increase in spontaneous pain.
Creatinine clearance (CrCl), forced expiratory flow (FEF 50%), and epidermal innervation density in patients with Fabry disease and normal control subjects
In the Fabry patients, the distal biopsy site showed essentially no remaining fibers innervating the epidermis (0 to 2.4 fibers per linear millimeter, versus control subjects, 4.7 to 6.5 fibers per linear millimeter) except in one patient. That patient, who had normal innervation density at the thigh (13.9 ± 1.5 fibers per linear millimeter) nevertheless had >50% reduction in innervation density at the distal site. At the thigh, innervation densities in the Fabry patients ranged from 0.71 ± 0.15 fibers per linear millimeter (Patient 8) to 13.9 ± 1.5 fibers per linear millimeter (Patient 3) (table 1).
Quantitative sensory testing and electrophysiology of Fabry patients with preserved renal function.
Cutaneous thermal (cold and warm) and vibratory detection thresholds were estimated from the hand and foot, using the Computer-Assisted Sensory Examination (CASE IV) system (W.R. Medical Electronics Co., Stillwater, MN) and a two-alternative forced-choice algorithm at 25 discrete steps of stimulus intensities. The results were compared with those from 22 healthy subjects of similar age (20 to 55 years; mean 34.8).
NCS and EMG studies were performed on 18 of 20 Fabry patients in this study (see also reference 13). Although six patients had mild conduction slowing at the wrist, there were no generalized conduction abnormalities. Sensory action potential amplitudes were normal in all patients.
Sural nerve biopsy specimens: Comparison of skin biopsy quantitation with sural EM quantitation.
The number and distribution of large myelinated fibers in the sural nerve specimens from two of the three Fabry patients showed no abnormalities (table 2; part of the third specimen was damaged and the myelinated axon density could not be accurately quantitated by light microscopy). In contrast, a loss of unmyelinated fibers was apparent by EM (figure 2, a and b; table 2). Groups of denervated Schwann cells were present throughout the biopsy specimens (see arrows, figure 2a; higher magnification, figure 2b). Lipid inclusions were prominent in the perineurium (arrows, figure 2c) and endothelial cells. Although large myelinated fibers were not reduced in numbers, occasional fibers did show a pathologic state at the EM level (figure 2d). As can be seen in table 2, the densities of small myelinated and unmyelinated fibers were reduced in comparison with normal control values.
Comparison of sural nerve quantitation with thigh innervation density
Figure 2. Electron micrographs of sural nerve biopsy specimens from patients with Fabry disease. (a) Field of unmyelinated fibers with numerous denervated Schwann cells. A cluster of denervated Schwann cells appears at the left (arrowheads). (b) Several denervated Schwann cells at higher magnification, with two normal unmyelinated axons remaining. The perineurium of the nerves from the Fabry patients typically has prominent lipid inclusions (c, arrows), but no lipid inclusions are present within the endoneurium (not shown). Although there is no detectable loss of large myelinated fibers in patients with Fabry disease who have preserved renal function, rare degenerating large fibers were seen (d).
Comparison of sural nerve quantitation with thigh innervation in the same patients showed that despite a similar loss of unmyelinated fibers in the sural nerves in all three patients, cutaneous innervation densities at the thigh showed distinctly different degrees of small fiber loss in each patient (table 2), with either no (Patient 3), moderate (Patient 1), or severe (Patient 6) denervation. At the level of the foot, two patients had end-stage denervation, and one (Patient 3) had >50% reduction in innervation density.
Comparison of skin innervation with other measures of disease burden.
Deterioration in kidney function is a general feature of Fabry disease. However, the patients described here were selected on the basis of relatively well preserved kidney function. This was to ensure that the neuropathy being quantitated was the result of the metabolic derangement proximately caused by the decreased α-galactosidase activity and not by secondary factors, such as uremia. Nevertheless, we were interested in comparing other quantifiable measures of disease burden with each patient’s cutaneous innervation density. Figure 3a shows a graph of cutaneous innervation density at the thigh compared with creatinine clearance, and figure 3b shows the comparison of cutaneous innervation with one aspect of pulmonary function: the forced expiratory flow at 50% exhalation (FEF 50%). We had observed that nearly all patients with Fabry disease had abnormal pulmonary function tests, with FEF 50% being the most commonly affected parameter. The graphs show that there is no correlation between innervation density at the thigh and the other measures of disease burden (correlation coefficient with creatinine clearance: 0.01; with FEF 50%: 0.10).
Figure 3. Comparison of epidermal innervation density at the thigh with (a) creatinine clearance (CrCl) and (b) forced expiratory flow (FEF 50%). There is no apparent correlation of innervation density with CrCl in Fabry patients with CrCl ≥50 mL/min or FEF 50% patients before they develop.
Discussion.
This study demonstrates several points. First, quantitation of intraepidermal fibers is a simple and sensitive measure of the peripheral nerve damage in patients with Fabry disease. Second, in patients with preserved renal function, only small fibers, myelinated and unmyelinated, are lost. This selective loss of small nerve fibers in Fabry patients before they develop renal dysfunction was first detected by quantitative sensory testing (QST),13 and our findings are in agreement. Previous studies showing widespread conduction slowing in patients with Fabry disease may have been measuring abnormalities caused by the renal damage characteristic of the latter stages of the disease and not proximately caused by decreased α-galactosidase activity. Third, the skin innervation data demonstrate a length dependence on the distribution of nerve damage. All patients demonstrated a greater proportional loss of innervation at the distal biopsy site than at the proximal biopsy site. This is well illustrated in Patient 3, who had normal innervation at the proximal biopsy site but <50% of normal innervation at the distal site. Moreover, on careful neurologic examination, all patients showed a distal loss of cold perception in the upper and lower extremities, with greater involvement of the lower extremities. Finally, even young adult patients had detectable cutaneous denervation, indicating that this method of evaluation is not restricted to older patients with advanced disease.
The intraepidermal innervation densities that we measured reflect our measurement technique. We counted only the number of epidermal nerve endings per linear millimeter, a technique already validated by previous studies,17,18 whereas other groups have additionally measured total length of intraepidermal fibers to provide another measure of involvement of intraepidermal nerve fibers in neuropathy. In comparison with these previous studies, our reliance on only the intraepidermal innervation density has the advantage of its relative simplicity and rapidity. However, our technique is likely to have a greater element of error. On the other hand, the total length of intraepidermal fibers is a function of the thickness of the epidermis. In at least some patients with neuropathy, epidermal thickness can vary (see reference 25), complicating the interpretation of total fiber length measurements. The measurement of intraepidermal fiber density is independent of epidermal thickness.
The gold standard for identifying and quantitating small fiber neuropathy is nerve biopsy. Our data suggest that analysis of cutaneous innervation may be as useful in detecting the small fiber neuropathy in Fabry patients as is sural nerve biopsy. However, skin biopsy has several advantages over nerve biopsy, particularly in the case of Fabry disease. In the patients who underwent sural biopsy, there was <50% loss of small myelinated and unmyelinated fibers in the sural nerves, and no apparent loss of large myelinated fibers. In contrast, there was an almost complete loss of intraepidermal innervation at the ankle in the same patients. Thus, the axons in the epidermis appear to be derived from fibers selectively affected in Fabry disease, whereas the bulk of the sural fibers appear unaffected in patients with preserved renal function. Skin biopsy may therefore be selectively sampling the axons most at risk in these patients.
The most important advantage of skin biopsy over sural nerve biopsy is the ability to perform multiple biopsies over time. Because the peripheral nervous system is capable of robust nerve fiber regeneration, serial biopsies could prove useful in detecting and quantitating increases in cutaneous innervation density resulting from supportive or specific therapies. Moreover, because cutaneous fibers have been shown to sprout collaterals to innervate adjacent skin that has lost its innervation,26,27 an increase in cutaneous innervation caused by collateral sprouting may be detectable after successful therapy even in patients with advanced neuropathy.
Acknowledgments
Supported in part by a NINDS grant P01 26643 to J.C.M. and J.W.G.
Acknowledgment
The authors thank Dr. James Dambrosia for statistical advice, and Devera G. Schoenberg, MS, for editorial assistance.
- Received June 18, 1998.
- Accepted December 19, 1998.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within the last 8 weeks.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.