Small fibre neuropathy (SFN) is a frequent cause of neuropathic pain in patients with a “burning feet” syndrome.1 2 Numerous disorders have been associated with SFN, including diabetes mellitus or impaired glucose handling, nutritional deficiencies, end stage renal failure, cryoglobulinaemia, coeliac disease, Sjögren syndrome, paraneoplastic neuropathy and infection with HIV.1^–7 A large proportion of cases are eventually classified as idiopathic after extensive diagnostic evaluation has excluded these and other rarer conditions.8^–11 SFN usually manifests as a length dependent process with the toes and feet initially affected followed by progression in a distal to proximal gradient; early involvement of the hands or proximal regions of the limbs has been reported but is distinctly unusual.10 12 13 We have encountered patients with neuropathic pain and pure or predominantly small fibre sensory loss involving proximal regions of the limbs, face and trunk, either sparing the acral extremities or with simultaneous involvement of distal and proximal areas in a non-length dependent pattern. This pattern suggests that the pathology may be localised to small fibre neurons in the dorsal root ganglia10 12 13 and is similar to what had been described clinically and pathologically in patients with Fabry disease.14 We report the clinical, laboratory, electrophysiological and skin biopsy findings in a cohort of patients with features suggestive of a non-length dependent small fibre neuropathy or ganglionopathy (SFG).

METHODS

Patient selection

Case files were reviewed from four tertiary neuromuscular centres selecting patients referred for evaluation for a non-length dependent pain or burning skin syndrome of undiagnosed cause with any of the following complaints: (1) neuropathic pain or impaired small fibre sensation involving the proximal regions of the limbs, face or trunk; (2) onset of neuropathic symptoms in the face, scalp, mouth, tongue, trunk, hands, arms or proximal legs before the feet; or (3) pain that began in the proximal and distal regions of the limbs simultaneously (eg, “total body burning”). Patients were excluded if symptoms or findings were confined to dermatomal or named cutaneous sensory nerve territories. Those with any motor findings were also excluded; patients with mild paresthesias, subjectively reduced light touch sensation in affected regions and slight vibration loss in the toes were included but those with predominantly large fibre abnormalities (with or without associated pain)—for example, impaired joint position, pseudoathetosis of the limbs, sensory ataxia or Romberg sign—were excluded. Furthermore, patients with an established cause for a ganglionopathy (eg, paraneoplastic ganglionopathy with anti-Hu antibodies), pre-existing Sjögren syndrome with a painful, mixed large and small fibre ataxic neuropathy or ganglionopathy, exposure to cisplatinum, hereditary sensory neuropathy, Fabry disease and vitamin B6 intoxication were also excluded. Lastly, patients who had a skin biopsy for analysis of the intraepidermal nerve fibre density (IEND) showing a length dependent pattern (ie, abnormal calf IEND and normal thigh IEND) were also excluded.

Diagnostic evaluation

Laboratory studies

The following laboratory tests were obtained: complete blood count (23 patients), renal (19), liver (21) and thyroid (21) function tests, lipid screen (14), C reactive protein (9), erythrocyte sedimentation rate (23), fasting blood glucose (19), glycosylated haemoglobin (13), oral glucose tolerance test (10), antinuclear antibody (ANA, 21), encapsulated nuclear antibody (11), anti-SSA and SSB antibodies (19), antineutrophil cytoplasmic antibody (12), serum immunofixation (19), urine immunoelectrophoresis (13), cryoglobulins (11), angiotensin converting enzyme (15), hepatitis B and C titres (11), Lyme titre (16), HIV titre (9), vitamin B12 and folate levels (23), tissue transglutaminase and antigliadin antibodies (15), and anti-Hu antibody (13).

Electrophysiological assessment

Nerve conduction studies were performed in all patients at the time of the initial evaluation according to standardised methods previously described.15 Motor conduction studies were performed using supramaximal percutaneous stimulation with surface electrode recordings. Sensory nerve action potentials (SNAP) were recorded using ring (median and ulnar) or bar (sural) electrodes, and skin temperature was maintained at 32°C. The median, ulnar, peroneal and tibial motor nerves, and median, ulnar, radial and sural sensory nerves were sampled in most patients.

Sympathetic skin responses in the hands and feet (SSR, five patients) and heart rate interval variability (HRIV) following deep breathing and Valsalva manoeuvre (five patients) were performed according to previously described methods.16 SSRs were classified as present or absent following auditory and tactile stimulation. HRIV was assessed by determining an inspiratory:expiratory ratio, and a value below 1.10 was considered abnormal and indicative of vagal nerve dysfunction.16

Pathological studies

Four patients had a sural nerve biopsy. Nerve tissue was fixed at resting length. One segment was fixed in 4% glutaraldehyde in 0.1 M phosphate buffer, post-fixed in osmium tetroxide and embedded in epoxy resin. This specimen was sectioned at 1 μm and stained with toluidine blue. A second segment of nerve was fixed in 10% neutral buffered formalin, embedded in paraffin, cut at 8 μm, and stained with haematoxylin and eosin, and with Congo red. Electron microscopy and quantitative morphometry were not routinely performed.

Seventeen patients had a skin biopsy from the calf and thigh analysed in two laboratories according to standardised procedures and using identical methodology.17^–19 Briefly, epidermal innervation was assessed using 3 mm punch skin biopsies taken from the distal leg (10 cm proximal to the lateral malleolus) and from the proximal lateral thigh. The skin biopsies were fixed, cryoprotected, sectioned and immunostained with polyclonal antibodies to the panaxonal marker, protein gene product 9.5.18 Blinded observers assessed the linear density/mm of the IEND from three to six 50 μm thick sections selected at random from each biopsy specimen, using previously published techniques.18 19 IEND was considered abnormal if below the fifth percentile based on previously derived normative data; a density at or below 5.0/mm for the calf and 8.0/mm for the proximal thigh was considered abnormal.17 20 In patients with low normal IEND, the biopsy was also classified as abnormal if morphological abnormalities were observed (eg, prominent axon swellings, frequent nerve fibre fragmentation, dystrophic fibres or excessive fibre “beading” or inflammatory cell infiltrates).20 21 In two patients, the skin biopsy was performed at other institutions and the IEND was not determined; results were described qualitatively as normal, or slightly, moderately or significantly reduced. Five patients also had a skin biopsy of the medial forearm, and an IEND value below 4.0/mm was considered abnormal.22

Other diagnostic studies

Minor salivary gland biopsy was obtained in 14 patients and lumbar puncture was performed in six. Several patients also had cervical spine MRI studies, chest radiography, chest and abdominal CT scans, skeletal survey and fat pad biopsy.

RESULTS

Illustrative case (patient No 1)

A 45-year-old man related 10 years of slowly progressive, symmetric pain and sensory loss beginning simultaneously in the feet and hands and spreading to the legs, arms and scalp. The pain was characterised as severe, constant, burning, aching and throbbing. He complained of numbness in the feet and finger pads that spread proximally to the thighs and hands over years. There were frequent, spontaneous sweating fits but no other autonomic symptoms. Examination showed normal strength, muscle bulk, deep tendon reflexes, gait and coordination testing, joint position and an absent Romberg sign. Vibration sensation was mildly reduced at the toes, with impaired pinprick and light touch over the mid-region of the forehead sparing the nose and lips, reduced over the scalp posteriorly and absent from the fingertips extending to the elbows and from the toes to the mid-thigh level. The ANA titre was 1:160, and an IgG kappa monoclonal protein was detected. Skeletal bone survey, abdominal fat pad and salivary gland biopsies were normal. Electrodiagnostic studies showed normal motor potentials and late responses, no denervation and reduced SNAP amplitudes in the arms and legs. Sural nerve biopsy showed moderate T cell (positive for CD3 staining) and B cell (positive for CD20 staining) epineurial, perivascular lymphocytes with reduced axonal density and epineurial fibrosis. Electron microscopy showed loss of C fibres and Schwann cell stacks. Skin biopsy showed similar degrees of reduced IEND at the calf, thigh and forearm. There was no improvement following treatment with corticosteroids, intravenous immune globulin or ciclosporin. Sequential and combination trials of gabapentin, topiramate, desipramine, nortriptyline, duloxetine, mexiletine and lamotrigine were ineffective. The pain was only moderately controlled with high doses of narcotics.

Clinical features

The clinical features of the patients are summarised in table 1. Twenty-three patients were identified, 12 men and 11 women, with an average age of 50 years (range 19–65). The average duration of symptoms at the time of the initial evaluation was 2.5 years (range 3 months to 14 years). Patients described their symptoms using terms such as “sunburn sensation”, “burning prickling”, “frostbite pain” and “my body feels like it is burning at the stake”. The pain was characterised more specifically as burning (n = 22), prickling (n = 13), shooting (n = 13) or allodynic (n = 11) when patients wore gloves, socks or clothing. Onset was over days in eight, evolved over months in eight and progressed slowly in seven; in the latter group, the pain intensity continued to worsen over years. In those with stable symptoms, the average time to reach a nadir was 22 months (range 1 day to 9 years (one patient had continuous slow progression)). Onset was in the hands or face before the legs in 10; the face (n = 12), scalp (n = 10), tongue (n = 6), trunk (n = 15) and hands and feet (n = 21) were involved at the time of maximal symptoms (fig 1). Four observed impaired temperature perception in affected regions. Approximately half of the patients also complained of numbness and paresthesias in the hands and feet. Autonomic symptoms were uncommon; four had sicca complex, three noted excessive sweating “fits”, two had orthostatic dizziness and one each had constipation and erectile dysfunction.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Body diagrams representing the regional distribution of neuropathic pain at the peak of symptoms in 23 patients with small fibre ganglionopathy. The number to the left of each figure corresponds to the patient number in tables 1 and 2.

A viral syndrome preceded the onset of symptoms in one patient; in another, the condition began abruptly several weeks after an influenza vaccination complicated by serum sickness (widespread arthralgias, fever and rash for several weeks). Another observed recurrent and transient worsening of pain after upper respiratory infections. One had type I diabetes mellitus, another had autoimmune thyroid disease and one had hepatitis C infection and rheumatoid arthritis and was taking prednisone at 30 mg/day at the onset of neuropathic symptoms. One patient had coeliac disease confirmed by distal duodenal biopsy and was on a gluten-free diet for 3 years prior to the onset of neuropathic symptoms. Another patient had a remote history of chronic lymphocytic leukaemia and was treated with cycles of cyclophosphamide, vincristine and prednisone followed by remission. There were no neurotoxic complications during or immediately after chemotherapy, and the neuropathic symptoms developed 2 years later.

No patient had limb weakness, impaired joint position, gait ataxia or a Romberg sign. Vibration sensation was reduced or absent at the metatarsophalangeal joint in 10 and abnormal at the distal interphalangeal joint in three. There was reduced pinprick sensation over the face in 10, scalp and anterior chest wall in an escutcheon pattern in eight each, and diminished or absent in the fingers and hands (n = 12) and feet (n = 18). There was also patchy loss of pinprick sensation over the proximal limbs in eight. Deep tendon reflexes were normal except for an absent Achilles reflex in four patients and absent patella and Achilles reflexes in one.

Diagnostic evaluation

Pertinent diagnostic studies are summarised in table 2. Abnormal glucose metabolism was detected by an oral glucose tolerance test in five patients (2 h glucose >140 mg/dl in three, fasting glucose >99 mg/dl in two). One patient had elevated tissue transglutaminase antibodies suggestive of coeliac disease but a jejunal biopsy was negative. Immunofixation demonstrated an IgG kappa monoclonal protein in one patient, and further diagnostic studies excluded a malignant plasma cell disorder. One patient had a positive HLA B27 test with transient arthralgias at the onset of the neuropathic symptoms but no diagnostic evidence of ankylosing spondylitis. Other laboratory studies were normal or negative. No malignancy was detected in those who had chest radiography, chest and abdominal CT, skeletal survey and fat pad biopsy. Cervical spine MRI in three patients was normal. Minor salivary gland biopsy demonstrated pathological features diagnostic of Sjögren syndrome in three patients (one with sicca symptoms, none with anti-SSA or SSB antibodies).

Motor nerve conduction studies were normal in all patients. The median SNAP was normal in 16 of 20 (80%) patients, ulnar in 12 of 19 (63%), radial in 10 of 11 (91%) and sural in 18 of 23 (74%). In six patients (26%), the upper limb sensory potential amplitudes were reduced or absent while the sural SNAP was normal.

SSRs and HRIV studies were abnormal in two patients each. CSF was acellular and the protein level was normal in four and slightly elevated in two patients (58 and 69 mg/dl).

The sural nerve biopsy specimen was normal in two patients. One specimen showed mild loss of myelinated axons and small clusters of epineurial perivascular CD20+ lymphocytes without vessel wall invasion, as previously noted (patient No 1). The other showed reduced axon density with excessive axon clusters and a small focus of perivascular lymphocytes around a single vessel.

Skin biopsy showed reduced IEND in the thigh equal to or more prominent than the calf in 15 patients (table 2). A pattern of thigh IEND abnormal/calf IEND normal was found in six; both sites had an abnormal IEND in nine patients. Two of six patients had an IEND of the forearm just above the lower limit of normal, but the density was lower than the calf in both instances. In another patient, only a calf biopsy was obtained and showed a reduced IEND. Morphological abnormalities were also common: seven had prominent axonal swellings in the dermis and epidermis, two had excessive fibre fragmentation and two others had perivascular inflammatory cell infiltrates. In summary, in the 16 patients who had calf and thigh skin biopsies, all demonstrated a non-length dependent pattern of IEND loss.

Treatment response and follow-up

The average duration of follow-up was 2 years (range 2 months to 7 years). Any treatment response needs to be interpreted cautiously because therapy was administered at the discretion of the treating physician and was unblinded. Seven patients were treated with prednisone at daily dosages between 20 and 60 mg for at least 1 month, and two others received 1 g/day of intravenous methylprednisolone, one for 3 days and the other for 5 days. Neuropathic pain resolved in only one patient (patient No 13) after taking prednisone at 20 mg/day, followed by a recurrence of symptoms after tapering the dose to 15 mg/day. Five of these patients also received intravenous immunoglobulin (IVIg 2 gm/kg). One had an approximately 30% reduction in pain after 3 monthly infusions, and another had substantial improvement (resolution of pain in the hands and arms, minimal pain in the feet) after monthly infusions over 3 years. Three others had no response.

Various combinations of anticonvulsants and antidepressants were administered to 22 patients with moderate pain reduction in 13; in 14, a combination of three or more agents was necessary (table 2). Eight also required narcotics for satisfactory pain control. Despite aggressive symptomatic management, 10 had chronic and disabling neuropathic pain at the time of their last visit. None had spontaneous improvement.

DISCUSSION

We have described a clinical pattern of neuropathic pain involving the proximal regions of the limbs, tongue, face, scalp and trunk in a non-length dependent manner suggestive of a ganglionopathy with preferential involvement of small fibre neurons. The onset can be acute, subacute or chronic, and with two exceptions (a viral syndrome and an influenza vaccination), there were no apparent precipitating factors. Dysaesthetic, burning pain of the involved body region was the most common symptom. Many patients likened the symptoms to a sunburn sensation; painful prickling and electrical shooting pains were also common, and almost half had allodynia. Complaints of mild numbness and paresthesias in the hands and feet were common, but there were minimal or no findings of large fibre impairment, and none developed sensory ataxia or a gait disturbance. Autonomic symptoms were also uncommon, but a systematic autonomic evaluation was not performed and we cannot exclude autonomic nervous system involvement within the spectrum of SFG, similar to what has been described in patients with idiopathic SFN.23 Our experience suggests that this is a disabling illness due almost exclusively to the severity of chronic and widespread neuropathic pain. The condition progressed slowly, had a limited response to immune therapies in a small number of treated patients and often was refractory to aggressive symptomatic pain management. Other investigators have also described a similar clinical pattern in a smaller number of patients.9 11 12

SFG contrasts with the classic large fibre ganglionopathies associated with anti-Hu antibodies or Sjögren syndrome, with these conditions characterised by pseudoathetosis, marked imbalance and gait disturbance caused by proprioceptive impairment, although neuropathic pain also may be prominent.11 13 Indeed, paraneoplastic neuropathy has become an increasingly recognised cause of a pure or predominantly small fibre neuropathy or ganglionopathy,7 yet the anti-Hu antibody was negative in those who were tested and none of our patients had clinical or radiographic features of a malignancy despite extensive evaluation. The large fibre sensory neuronopathies are presumably related to an immune mediated attack of dorsal root ganglion cells, including those subserving large fibre modalities.24 We speculate that in our patients the disorder may represent relatively selective dysfunction of the dorsal ganglia cells subserving small nerve fibres. The evidence favouring this localisation is: (1) bilateral distribution with early and prominent involvement of proximal regions, especially the face and tongue, or the hands before or simultaneously with the legs and feet in a non-length dependent distribution; (2) relative sparing of large nerve fibres; (3) skin biopsy specimens corroborating a non-length dependent pattern of IEND loss in proximal sites of the thigh or forearm, sparing the calf or simultaneous involvement of proximal and distal sites (see below); and (4) a slowly progressive course, perhaps suggesting loss of small dorsal root ganglia (DRG) neurons. The distribution can be symmetrical or asymmetrical, similar to the conditions associated with large fibre ganglionopathy. We recognise this hypothesis can only be confirmed by pathological examination of the DRG, and thus our observations remain preliminary as none of our patients had a DRG biopsy. Lastly, the features suggestive of a SFG can be distinguished clinically from the common variety of length dependent, distal SFN, typical of painful diabetic neuropathy, HIV associated neuropathy, systemic amyloidosis and idiopathic forms, and from rare cases of sensory mononeuritis multiplex or sensory perineuritis due to the invariable focal or multifocal nature of the latter conditions affecting discrete cutaneous sensory nerve territories.25

Laboratory testing showed a potential metabolic or immune process in almost half of our cases, notably diabetes mellitus or impaired glucose tolerance, Sjögren syndrome, monoclonal gammopathy, coeliac disease and hepatitis C infection. Others have also described a similar non-length dependent small fibre neuronopathy in patients with Sjögren syndrome.6 11 In accordance with our findings, this suggests that Sjögren syndrome may be associated with a classical large fibre ganglionopathy, SFG, or a mixed pattern.6 11 24 Similarly, coeliac disease has been recognised as a cause of non-length dependent small fibre neuropathy but was confirmed in only one of our 15 patients tested (this case was not previously reported).5 Although less than half of our patients were tested for cryoglobulinaemia, the disorder could be adequately excluded in the remainder because of the lack of purpura or other clinical features and normal ANA and complement levels in those who were tested.4 We were surprised by the high frequency of patients with abnormal glucose handling (24%), and postulate that in some cases SFG may reflect a rare neuropathic phenotype associated with diabetes mellitus or impaired glucose tolerance. Indeed, animal studies have shown a marked reduction in DRG neurons in rats with streptozotocin induced diabetes mellitus, yet there was a disproportionate loss of large rather than small DRG neurons.26 27 This phenotype should be distinguished from other recognised forms of proximal or multifocal neuropathy in diabetic patients, such as truncal and radiculoplexus neuropathy related to a T cell mediated vasculitis, with prominent motor involvement and ability to localise the disorder to one or more nerve roots, plexus or peripheral nerves.28 Many of our cases were considered idiopathic after extensive evaluation, but this conclusion should be tempered by the retrospective nature of our data analysis and varying extent of laboratory testing from different centres.

Consistent with the putative site of the disease, routine electrodiagnostic studies were usually normal.1–3 8 The amplitude of the sensory nerve potentials was reduced or absent in some cases, indicating associated large fibre dysfunction. In 26% of our patients, the median or ulnar SNAP amplitude was reduced or absent while the sural potential was normal, supporting a non-length dependent pattern in such cases.

The limited number of sural nerve biopsy specimens in our group precludes any definitive conclusions with respect to an inflammatory, immune mediated or other angiopathic mechanism, and others have also demonstrated the limited utility of nerve biopsies from patients with SFN.19 29 However, two of our patients had modest inflammatory infiltrates, suggestive of an immune disorder, and similar pathological features have been reported rarely in patients previously considered to have “idiopathic” SFN.30 31 Two of our patients apparently were responsive to corticosteroids or IVIg; our limited experience and another report suggest that in some patients, SFG may be an inflammatory condition.31 In contrast with the inflammatory infiltrates observed in the DRG of patients with Sjögren’s syndrome associated with large fibre ganglionopathy,24 we are uncertain of any relationship between distal sensory nerve inflammation and pathology at the level of the DRG in our cases.

It should be emphasised that because of the peculiar distribution of symptoms and paucity of neurological findings, many of our patients were previously considered to be narcotic seeking or have a somatisation disorder or another psychiatric illness as an explanation for their symptoms. Indeed, patients with prominent burning symptoms isolated to the mouth and tongue, a condition referred to as “burning mouth syndrome”, have had their condition attributed to psychological causes. However, recent studies have confirmed reduced epithelial nerve fibres in the tongue, likely reflecting a trigeminal neuropathy involving small fibres or perhaps a restricted form of a small fibre ganglionopathy.32 In our patients, skin biopsy was especially helpful by providing objective evidence of small nerve fibre involvement and demonstrating a non-length dependent pattern. A virtually identical IEND pattern has also been observed in patients with large fibre sensory ganglionopathy and hyperalgesic symptoms and were distinguished from those with various axonal polyneuropathies who had reduced IEND in the characteristic distal to proximal gradient.13 19

It was our impression that this neuropathic small fibre pain syndrome was largely refractory to symptomatic medications. Although those with distal SFN often experience satisfactory pain control following treatment with one or two non-narcotic analgesics,33 many of our patients required three or more such medications along with numerous trials of moderate to high dose narcotic agents and remain with inadequate pain control. The long term prognosis of patients with SFG is unknown because of our limited follow-up, but thus far none have experienced spontaneous recovery.