Subclinical sensory neuropathy in late-onset restless legs syndrome

Patients and methods.

The results of the above pilot study (data not shown) suggested that as many as 45% of patients with RLS might have subclinical sensory neuropathies; this prompted us to pursue a larger study. Twenty-four sequential patients with RLS from the Baltimore area who presented for treatment at the Johns Hopkins Sleep Disorders Center were asked to participate in this study, which was approved by the Institutional Review Board. All 24 patients gave consent; two missed their initial visit and were dropped from the study. No patients who were offered participation refused. A sleep disorder specialist evaluated the remaining 22 and all fulfilled the International Restless Legs Syndrome Study Group25 diagnostic criteria for RLS. Patients were not selected in any fashion and were felt to be representative of patients in tertiary RLS clinics. Those patients who had previously been treated for RLS responded to conventional treatments. All patients experienced bilateral symptoms and had RLS as their primary neurologic complaint. Patients 19 and 22 had a history of lower back pain or radiculopathy. Patient 15 had evidence of mild large-fiber neuropathy by previous sural nerve biopsy. None of the patients had comorbid conditions such as diabetes, vitamin B₁₂ deficiency, uremia, or alcohol abuse.

A board-certified sleep disorder specialist reviewed the clinical chart for each patient to determine age, gender, family history of RLS, age at onset of RLS symptoms, and the years required for RLS to progress to daily symptoms. This chart review is performed periodically for all patients in the RLS clinic and was not specifically a part of this study or completed with knowledge of the neuropathy test results. Family history was considered positive for RLS if the patient reported any first-degree relative who also had RLS. The total number of years with RLS was determined by the difference between age at onset and current age and the years with daily symptoms was determined by subtracting the years to progress to daily symptoms from the years with RLS. Consistent with prior studies,3 patients were divided into late-onset (symptoms occurring after age 45) and early-onset groups (symptoms appearing before age 45). The age at onset determined from this chart review has been shown to correlate well with a patient’s independent self-report on a questionnaire (r = 0.98), with complete agreement between the two methods for determining early versus late onset.3

The clinical severity of RLS was graded on the Johns Hopkins RLS Severity Scale: 1 for mild, 2 for moderate, or 3 for severe, based on the reported time of usual symptom onset. Mild RLS was defined as usual symptom onset at bedtime; moderate as after 6 PM but before bedtime; and severe as before 6 PM. This clinical scale has been used as a standard measure in prior studies,26,27⇓ and has an interrater agreement of 92% for trained clinical raters.26

The protocol for all subjects consisted of a RLS questionnaire, a modified Neuropathy Symptoms and Change (NSC) questionnaire,28 proximal thigh and distal calf skin biopsies, bilateral sural and deep peroneal nerve conduction studies, quantitative sensory threshold testing for cooling and vibration, and neurologic examination with Neurologic Impairment Score (NIS) determination (a quantitative measure of abnormalities on neurologic examination).28 The NSC (a validated questionnaire of neuropathy symptoms) was altered to include questions focusing only on the lower extremities without any emphasis on symptom change.29 The NIS (originally called the Neurologic Disability Score) is a validated instrument developed to express overall neurologic impairment from peripheral neuropathy as a single score. Punch skin biopsies (3 mm) were performed under local anesthetic at the distal leg and the proximal thigh, as in other studies.30,31⇓

The protocol for skin biopsy and IENF quantification has been described elsewhere.31 Briefly, all biopsies were read blinded to the clinical status, in a consensus conference (M.P., J.C.M.) as consistent with SSFL or normal. SSFL was defined as IENF loss at the distal site relative to the proximal site. Skin biopsies from the patients with RLS were mixed with control samples and a blinded technician who determined the IENF densities. The proximal thigh:distal leg IENF ratio was compared with published normative data31 and used as the final measure of SSFL. This ratio is meaningful only if the IENF density at the proximal site is within normal limits, as occurred in all of our subjects. Subjects with a ratio above the 95th percentile were considered to have SSFL. Patient 19 had a ratio between the 80th and 95th percentiles and was placed in the SSFL group. Figure 1 depicts the typical proximal–distal changes seen in patients with and without SSFL. Nerve conduction studies were performed using standard techniques.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Magnification ×195 before reduction. (A) Proximal thigh skin biopsy from Patient 9 showing normal intraepidermal nerve fiber density. (B) Distal leg biopsy from same patient. (C) Proximal thigh skin biopsy from Patient 16. The nerve fibers are segmented and varicose (consistent with predegenerative changes) although the intraepidermal nerve fiber density was within normal limits. (D) Patient 16’s distal leg biopsy reveals marked loss of epidermal and dermal nerve fibers.

Results.

Neuropathy occurrence.

Eight patients (36%) were found to have neuropathy. Three (Patients 2, 3, and 22) had mild involvement of large sensory fibers as determined by decreased sural amplitudes but normal densities of epidermal fibers; two (Patients 16 and 17) had mixed large-fiber neuropathy and SSFL reflected in abnormal electrodiagnostic studies and reduced epidermal fiber densities. The most pronounced abnormality in these patients was the SSFL. Three patients (1, 18, and 19) had pure SSFL, bringing the total number of patients with SSFL to five (23%). The only criterion for the diagnosis of SSFL was an abnormal proximal thigh:distal leg IENF ratio. The remaining 14 patients had no evidence of large- or small-fiber neuropathy. Figure 2 depicts the results of the skin biopsy readings from the 22 patients with RLS.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Box and whisker plot demonstrates intraepidermal nerve fiber (IENF) densities in patients with restless legs syndrome, without and with small sensory fiber loss (SSFL). The SSFL group shows a marked drop off of intraepidermal nerve fiber density at the distal leg compared with the proximal thigh. The line bisecting the box represents the mean; the top and bottom of the box represent the 25% and 75% CI; and the bars represent the 95% CI. Dashed lines = proximal thigh; dots = distal leg.

At least nine patients had normal IENF densities, but we identified morphologic changes of uncertain significance such as nerve fiber swellings, increased branching, or tortuosity. These morphologic changes appeared to be overrepresented among those in the late-onset group (Patients 20, 21, and 22) and in those with abnormal sural amplitudes (Patients 3 and 4). Although we believe these morphologic changes were abnormal, they were difficult to quantify and these patients were not classified as having SSFL. Figure 3 depicts an IENF from a patient with RLS with prominent swellings. Electron micrographs through swellings from the same patient suggested degeneration (not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. A skin biopsy section taken from the distal leg of a patient with restless legs syndrome. The image depicts an epidermal nerve fiber with swellings. Magnification ×450 before reduction.

Relation of neuropathies to other subject characteristics.

Compared with those with normal biopsies, patients with SSFL showed no difference (p < 0.05) in any of the other variables examined; i.e., gender, age, years with RLS, years with daily symptoms of RLS, or percentage with positive family history (see table 1). Two of these relationships were marginally significant. Compared with those with RLS without SSFL, patients with RLS and SSFL had fewer years with RLS (12.8 ± 4.0 versus 30.4 ± 4.6; Student’s t-test = 2.0; p = 0.06) and a less common family history of RLS (20% versus 65% had first-degree relatives with RLS; χ² = 3.11; p = 0.078). Similarly, compared with those with normal sural nerve amplitudes, patients with large-fiber neuropathy showed no relationship with any of these subject variables except for a marginal difference indicating a less common family history of RLS (20% versus 65%; χ² = 3.11; p = 0.08).

Discussion.

An association between RLS and neuropathy has been suggested since the original description by Ekbom.1 Most studies pursuing an association between the two have focused on large fiber involvement. An increased incidence of RLS among neuropathic patients has been reported32-34⇓⇓ as has evidence of neuropathy among patients with RLS.35 Studies addressing RLS prevalence among patients with neuropathy required abnormal electromyographic and nerve conduction velocity results as evidence of peripheral neuropathy, thereby excluding patients with “pure” small-fiber neuropathy.32 We suspect that the incidence would have been higher had such patients been included. Another study reporting large-fiber neuropathy associated with RLS noted that several patients had prominent burning and shooting pains in the feet8—symptoms that are seen in neuropathies with prominent small fiber involvement. Finally, the most consistent abnormality in a series of primary patients with RLS found to have neuropathy was abnormal warm quantitative sensory thresholds, seen in seven of eight patients, suggesting dysfunction of C and Aδ fiber nerves.4 Thus, whereas most studies to date have focused on large fiber dysfunction, there is indirect evidence of small fiber involvement.

One might argue that the SSFL-associated skin biopsy changes are the product of RLS-associated leg trauma. We believe that this is not the case. The fact that patients with early-onset RLS had fewer skin biopsy changes despite longer durations of RLS argues that skin biopsy changes were not due to trauma resulting from RLS-associated leg movements. Unfortunately, we did not directly measure leg rubbing or trauma in our patients.

The high prevalence of abnormal nerve fiber morphologies—such as nerve fiber swellings among patients with late-onset despite normal IENF density—may suggest that a greater percentage of patients might ultimately develop SSFL. Swellings are typically seen at proximal sites in patients with distal SSFL; they are not present among normal control subjects.36 Importantly, they are not seen in models of epidermal fiber regeneration in healthy control subjects after axotomy (unpublished observations: M.P., J.W.G., J.C.M.) and are seen in nerve fibers degenerating as a result of capsaicin application (unpublished observations: M.P., J.W.G., J.C.M.). These observations led us to postulate that the morphologic changes are predegenerative. Longitudinal data suggest that many patients with morphologically abnormal IENF go on to develop SSFL (unpublished observations: M.P., J.W.G., J.C.M.). Furthermore, the appearance of IENF on electron microscopy (EM) images suggests degeneration. It may be that these degenerating fibers act as irritable receptors giving rise to pain and dysesthesias. Although these morphologic changes are easy to recognize,36 they are difficult to quantify. Efforts at establishing normative data for the skin biopsy technique31 have focused on fiber density irrespective of morphology. It may be appropriate to calculate a density of morphologically abnormal fibers or, conversely, count only those fibers that are morphologically normal.

Patients with RLS can be stratified based on the age at onset. Patients found to have large-fiber neuropathy by electromyographic and nerve conduction velocity testing also had a later age at onset and more rapid progression of their RLS symptoms when compared with patients with RLS without neuropathy.35 These results may indicate that many patients with large-fiber neuropathy are also likely to have a component of small-fiber neuropathy.37 Indeed, in our study, 40% of patients with large-fiber neuropathy also had SSFL. The higher prevalence of large-fiber neuropathy in other studies35 may have resulted from the inclusion of patients with radiculopathy or abnormal results on electromyography, which was not performed in our study. Unfortunately, our small sample size makes comparisons between the large- and small-fiber neuropathy groups difficult.

Our finding that patients with RLS and neuropathy tend not to have family histories of RLS (p = 0.08) is consistent with another study that reported a positive family history in 23 of 25 patients with idiopathic RLS (92%) versus only two of 15 (13%) with neuropathic RLS.35 These values are consistent with our results: 65% of patients with RLS without neuropathy had a positive family history compared with 20% of RLS subjects with neuropathy. Notably, none of the eight primary patients with RLS found to have abnormal sural nerve biopsies had a family history of RLS.4

Our observation that the patients with SSFL had higher NIS scores than their counterparts with normal skin biopsies is consistent with these patients having neuropathy and describing their RLS symptoms as painful. The differences between the groups were small, and suggest that physical examination may not be sensitive enough to separate the groups reliably. Whereas other studies have not included similar objective measurements, it has been generally observed that the neuropathy associated with RLS is often subtle.35 Skin biopsy results offer an objective extension of the physical examination in such cases.

The difference in NSC scores between the skin biopsy groups was small, although patients with SSFL did report painful symptoms more frequently (p = 0.001). This implies that SSFL in patients with RLS may be difficult to identify by history alone and again underscores that the neuropathy associated with RLS is subtle. Other authors have reported similar observations. In a series of families with hereditary amyloidosis, RLS anticipated any evidence of neuropathy by as many as 30 months,9 and other groups commented that the “majority of their RLS patients with neuropathy lacked clinical evidence of neuropathy.”35 This observation is consistent with neuropathy in RLS being largely subclinical and suggests that RLS may be the only indicator of SSFL—especially if the RLS occurs later in life and is rapidly progressive. Perhaps RLS is a form of neuropathic pain in these patients. That is, whereas some SSFL patients experience burning, numbness, or electric-like jabs, others may experience only restless legs. This alteration in sensory perception may be due to rewiring in the dorsal horn that has been reported in animal models of C and A fiber injury.38,39⇓ It may be that the modulation of this nociceptive stimulation is disturbed in the brain and triggers RLS symptoms even though patients are unaware of the nociceptive stimuli.

We have shown that a fraction of patients with RLS have a subclinical small-fiber neuropathy and suggest that small-fiber neuropathy may manifest as RLS. This subgroup tends to have a later age at disease onset, commonly reports pain with RLS, and generally has no family history of RLS. Therefore, it may be appropriate to perform skin biopsies in patients with late-onset RLS or who report pain as a feature of their RLS. Our findings may explain why diseases such as diabetes, renal failure, amyloidosis, and cryoglobulinemia are RLS risk factors and may offer insight into novel treatment strategies.