Intraepidermal nerve fiber density in patients with painful sensory neuropathy

Peripheral neuropathies may preferentially involve small-diameter myelinated and unmyelinated nerve fibers. The clinical characteristics of such selective small-fiber involvement include insensitivity to pain and temperature, as well as spontaneous pain and hyperesthesia. ^[1] Sensory symptoms tend to begin distally in the feet, and slowly progress proximally in a length-dependent fashion, a pattern ascribed to "dying back" of affected axons. ^[2] In many instances, there are associated symptoms of autonomic insufficiency such as impotence and loss of sweating. ^[1] Identifiable causal or predisposing conditions include diabetes, ^[2] systemic amyloidosis, ^[3] HIV infection, ^[4] exposure to neurotoxic medications, ^[5] and the hereditary sensory and autonomic neuropathies. ^[6] However, most cases are idiopathic. ^[1,7]

Unless there is associated large-fiber sensory neuropathy, these patients have few objective physical signs and normal nerve conduction studies. Quantitative sensory testing (QST) may be a useful test for small sensory fiber dysfunction, ^[8,9] but it is a subjective psychophysical test; and while its utility in populations is well documented, ^[9] its use as a diagnostic test in individual patients has limitations. Currently there is no universally accepted confirmatory test available to establish the diagnosis of small-fiber sensory neuropathy or quantitate its severity objectively.

We previously reported the identification and quantification of unmyelinated intraepidermal nerve fibers (IENF) in skin obtained using a punch skin biopsy ^[10]-a technique that is easy, minimally invasive, and repeatable. The inter-observer reliability for quantification of IENF density is 96%. IENF density is significantly reduced in skin obtained from patients with idiopathic, ^[10] HIV-associated, ^[10] and diabetic ^[11] sensory neuropathies. In this study, we extended our initial observations, and correlated IENF density with other measures of sensory dysfunction in patients with painful sensory neuropathies (painful SN), including clinical findings, QST, and nerve conduction studies.

Methods.

We identified 20 patients with painful SN from the Peripheral Nerve and HIV Neurology clinics at Johns Hopkins. All patients had distally accentuated symptoms of painful dysesthesias and numbness, without significant weakness or sensory ataxia. We included patients with reduced or absent ankle reflexes, but otherwise deep tendon reflexes were normal. Clinical details and the results of electrophysiologic testing were obtained from the patients' medical records.

The control group consisted of 20 age-matched individuals randomly selected from a cohort of 98 healthy volunteers who have undergone neurologic screening, skin biopsy, and QST as part of an ongoing project to establish normative data for IENF density. None had any clinical features suggestive of neuropathy-including symptoms or signs of neuropathy, history of alcoholism, diabetes, or exposure to neurotoxic medications.

All subjects underwent skin biopsies from both the distal calf (approximately 10 cm above the lateral malleolus) and proximal lateral thigh. After local infiltration of 2% lidocaine with epinephrine, 3-mm diameter punch skin biopsy specimens were obtained using standard techniques. Specimens were fixed in 10% formalin, and free-floating 50-micro m freezing microtome sections were immunostained with the panaxonal marker PGP 9.5 (ubiquitin hydrolase, Ultraclone, UK). Individual IENFs were counted by a single technician blinded to clinical details, and expressed as a linear density (number of fibers per mm of epidermis). Technical details have been described in a previous publication. ^[10]

To correlate clinical signs of neuropathy with QST findings and IENF density, we developed a small-fiber sensory neuropathy grade (0-3), which was based on the distribution of abnormalities to pinprick or temperature sensitivity. Controls were considered grade 0. Patients with grade 1 SN had sensory abnormalities that were confined to the feet. Patients with grade 2 SN had sensory abnormalities that involved the feet and legs, but were confined to the lower extremities. Patients with grade 3 SN had sensory abnormalities that involved both the upper and lower extremities. In general, the severity of pin and thermal abnormalities at the foot was greater in the patients with more extensive sensory involvement.

The majority of both the patients with neuropathy (17) and controls (18) had QST performed using the Computer Assisted Sensory Examination (CASE IV) system. ^[9] A percentile value for sensory threshold to both vibration and cooling at the foot was generated in each case using published age-matched normative data. ^[9] A sensory threshold equal to or greater than the 95th percentile for age was considered abnormal. For further analysis, QST threshold percentiles were divided into three groups: <95 percentile (normal), 95 to 98th percentile, and >or=to99th percentile.

The results of routine nerve conduction studies and EMG were available for 12 (60%) of the patients with neuropathy.

Statistical analysis was performed using InStat Version 1.11a (GraphPAD Software, San Diego, CA). Group means were compared using the unpaired, two-tailed Mann-Whitney test. Paired continuous variables were correlated using linear regression. Chi-square analysis was used for comparing categorical variables. Continuous variables were compared between clinical severity groups by logarithmic normalization, analysis of variance (ANOVA), and subsequent sub-group analysis using the Student's t test.

Results.

The mean age of patients with neuropathy was 52 years (range 24-77 years), and the mean duration of neuropathic symptoms at the time of evaluation was 2.8 years (range 2 months-10 years). Etiologies for painful SN in this group of patients were as follows: diabetes (2), HIV infection (5), HIV infection with exposure to the neurotoxic antiretroviral agents didanosine [ddI] or stavudine [d4T] (3), and idiopathic (10). Six of the patients with painful SN were classified as clinical grade 1, eleven as grade 2, and three as grade 3 (Table 1).

Mean IENF density at the calf was 4.9 mm sup -1 in patients with neuropathy and 16.3 mm sup -1 in controls (p < 0.0001, Mann-Whitney). Mean IENF density at the thigh was 11.3 mm sup -1 in patients with neuropathy and 23.8 mm sup -1 in controls (p < 0.001, Mann-Whitney).

There was a progressive diminution in mean IENF density with increasing painful SN clinical grade (Figure 1) (calf skin: p < 0.0001, thigh skin p = 0.001; ANOVA).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Intraepidermal nerve fiber (IENF) density from calf (black circles) and thigh (white squares) skin correlated against clinical estimates of neuropathy severity. The data shown are group means with standard error bars.

Mean IENF density at the calf was significantly reduced compared with controls in patients with grade 1 (p < 0.0002), 2 (p < 0.0001), and 3 (p < 0.02) painful SN. In addition, there was a significant difference in mean calf IENF density between patients with grade 1 neuropathy and those with more severe neuropathies (grade 2: p = 0.05, grade 3: p < 0.03). Mean IENF density at the thigh was significantly reduced compared with controls only in patients with grade 2 (p < 0.004) and 3 (p = 0.001) painful SN.

QST thermal thresholds to cooling were abnormal in 82% of patients with painful SN and 17% of controls (p < 0.0002; Fisher's exact test). QST vibration thresholds were abnormal in 60% of patients with painful SN and 17% of controls (p = 0.03; Fisher's exact test). However, there was no significant correlation between QST threshold percentile and either mean IENF density at the calf (thermal: p > 0.8, vibration: p = 0.5; ANOVA), or clinical grade (thermal: p > 0.8, vibration: p = 0.1; ANOVA) for patients with painful SN.

Routine nerve conduction studies and EMG were available for 12 of the patients with painful SN. The sural sensory nerve action potentials (SNAPs) were abnormal in eight (67%)-absent in three and of low amplitude in five. In all cases with abnormal sural SNAPs, median SNAP amplitudes were normal. There was no significant correlation between sural SNAP amplitude and either IENF density (correlation coefficient r² = 0.222, p = 0.122) or clinical estimates of small fiber sensory neuropathy severity (p > 0.8, ANOVA). In all cases, the deep peroneal compound muscle action potential amplitudes and needle EMG of distal leg muscles were normal.

Discussion.

IENF density is significantly reduced in skin from patients with painful SN compared with age-matched controls. In patients with painful SN, IENF density is significantly reduced even in skin from regions proximal to the areas of clinically identifiable sensory abnormalities. Thus patients with grade 1 neuropathy, in whom sensory abnormalities are restricted to the toes and feet, have significantly reduced IENF densities at the calf compared with controls. However, patients with such distally restricted symptoms and signs have normal IENF densities at the thigh. With increasing painful SN severity and centripetal spread of sensory abnormalities, intraepidermal nerve fibers become progressively depleted from calf skin, and also (to a lesser degree) thigh skin. Patients with the most extensive painful SN have grossly reduced IENF densities at both the calf and thigh, reflecting severe small-diameter nerve fiber degeneration at both proximal and distal sites. Hence, IENF density correlates well with the clinical length-dependent features of painful SN.

Although, in this study, an abnormal QST percentile for thermal threshold to cooling at the foot correlated well with the presence of painful SN (sensitivity 82%, specificity 83%), QST thresholds did not correlate significantly either with IENF density or painful SN grade. Other investigators have reported a similar lack of correlation. ^[11] Possible explanations include: Quantitative sensory testing is performed at the foot, distal to the site used for skin biopsy, over skin that is clinically involved in all the patients with painful SN (even those with clinically mild neuropathy). Hence, QST performed at the foot may not be sensitive for discriminating between different severities of a length-dependent neuropathic process. Secondly, QST threshold to vibration or cooling may not be the most appropriate physiologic modalities to correlate with IENF density. Other investigators have found a better correlation between tests of sudomotor function than QST, with both IENF density ^[11] and clinical features, in painful SN. ^[7,12]

Two thirds of the patients with painful SN who underwent nerve conduction studies had abnormally low amplitude or absent sural SNAPs, indicating coinvolvement of both small- and large-diameter sensory nerve fibers. However, sural SNAP amplitude correlated with neither IENF density nor clinical grade of painful SN. Moreover, none of the patients in this group had extensive large sensory fiber involvement. Even patients with grade 3 painful SN had normal median nerve SNAP amplitudes, despite unequivocal clinical abnormalities of small-fiber sensory function involving the hands.

The remaining third of patients with painful SN have sensory neuropathies that are restricted to small-diameter sensory fibers. Despite prominent symptoms of pain and numbness, these patients have normal ankle reflexes and normal sural SNAP amplitudes. Skin biopsy for quantification of IENF density is particularly useful in these cases, as it may constitute the only objective abnormality.

The lack of a universally accepted objective test of small fiber sensory function in painful SN has hindered studies both of natural history and response to treatment. We have shown that diminished IENF density can be used to confirm the presence of painful SN and give an objective quantitative estimate of its severity. Hence, punch skin biopsy for IENF quantification is a useful test for both the diagnosis and staging of painful SN. Moreover, it is a simple and minimally invasive procedure that can be performed on patients in a serial fashion. It will therefore not only be a helpful technique for painful SN diagnosis in the outpatient clinic, but will also be useful for following patient progression over time. As such, it is suitable as an outcome measure in therapeutic studies for small-fiber neuropathy.