Neurofibromatosis 1 (NF1) is a common, autosomal dominant, neurocutaneous disease.¹ There is wide variety of disease expression in patients with NF1 and its numerous complications involve many of the body systems. The neurological manifestations may arise from tumours and malformations of the nervous system, deformities of the skull and skeleton, or pressure by neurofibromas on the peripheral nerves, spinal nerve roots, and spinal cord.² Neurofibromatous neuropathy has been reported as a rare manifestation and is characterised by a distal sensorimotor neuropathy associated with diffuse neurofibromatous change in thickened peripheral nerves.³ Until the last decade, the genetic and clinical distinction between NF1 and neurofibromatosis NF2 (NF2) was not always clearly established.⁴ In retrospect, the majority of cases of neuropathy and neurofibromatosis reported in the literature have been associated with NF2.^3,5–8 NF2 neurofibromatous neuropathy is entirely different clinically and histologically from NF1 associated neurofibromatous neuropathy.⁸

In this paper we describe eight patients with NF1 and neurofibromatous neuropathy, which is the largest group of cases so far reported. The patients all attend our multidisciplinary neurocutaneous clinic comprising 600 NF1 patients, suggesting that this complication of NF1 may not be as rare as previously supposed, although it still only affects 1.3% of patients.³ We describe the clinical and neurophysiological features in all our patients, the nerve biopsy appearances in two, and the causative NF1 mutation in a further two of the three patients who agreed to mutation testing.

METHODS

The eight patients consist of all those patients with NF1 who also had a symmetrical polyneuropathy and who had been referred to the Guy’s Hospital multidisciplinary Neurocutaneous Clinic for general assessment of neurofibromatosis 1, between 1985 and 2003.⁹ During this period 600 patients with NF1 were assessed clinically. Ethical approval was obtained for the study. General medical and neurological assessments were performed on all patients and nerve conduction studies were undertaken on all patients with symptoms or signs of a peripheral neuropathy. Magnetic resonance imaging of the spine was performed on seven patients with a 1.5 T superconducting system (Philips Gyroscan S15, Philips Medical Systems). Images were taken in the axial and coronal planes with a slice thickness of 5 mm and an interscan distance of 0.5 mm. Coronal STIR images (TR/TE 2000/25, TI 150) were performed and axial T1 images (TR/TE 700/20) were carried out before and following the administration of gadolinium meglumine triamine pentacetic acid contrast medium at 0.2 ml/kg.

Sural nerve biopsy

Two patients underwent sural nerve biopsy for the purpose of diagnosis. Portions of the nerve were processed into paraffin and also into epon for 1 μm sections stained with thionin and acridine orange and ultrathin sections for electron microscopy as previously described.¹⁰

Mutational analysis

Three patients agreed to donate blood for mutation analysis. Mutation analysis was performed using a combination of chemical cleavage of mismatch, single strand conformational analysis, and direct sequencing. DNA samples from 70 non-NF1 patients were used as controls. Haploinsufficiency was excluded at the genomic and mRNA level prior to mutation analysis with chemical cleavage of mismatch, by demonstration of heterozygosity at exon 5 (Rsa1 polymorphism) in genomic DNA and cDNA, respectively.

Laboratory investigations

The following investigations were carried out: haemoglobin, erythrocyte sedimentation rate, serum folate, vitamin B12, urea, creatinine and electrolytes, liver function, blood glucose, autoantibodies, serum immunoglobulins, protein electrophoresis, thyroid function, and treponema pallidum microhaemagglutination assay.

RESULTS

The clinical, radiological, and neurophysiological features of all the cases are reported in tables 1–3 and illustrated in figs 1 and 2. The laboratory investigations mentioned for known causes of neuropathy were normal in all patients. The neurophysiological features were those of a predominantly sensory, length dependent axonal neuropathy with abnormally small sural nerve sensory action potentials, relatively normal median nerve sensory action potentials, slightly delayed distal motor conduction and F wave latencies, and slightly slowed motor nerve conduction velocities (table 3).

Nerve biopsy

Patient 1

A sural nerve biopsy showed non-uniform pathological changes. Some fascicles displayed a marked loss of axons of all calibres, whereas in others the axonal drop out was mild and confined to a sub-perineurial zone. There was little evidence of ongoing active degeneration in any fascicle. Regions of focal perineurial disruption, associated with a hypercellular epineurium and disorganised sub-perineurium, were prominent features of fascicles that displayed the most marked axonal loss. At these sites, the sub-perineurial endoneurium was filled with bundles of collagen and numerous process bearing cells, none of which were associated with axons. On morphological criteria, these cells were identified as either chronically denervated Schwann cells¹¹ or perineurial cells, whose long cytoplasmic processes were associated with continuous or patchy basal laminae, and were studded with caveolae, or fibroblasts (fig 3). Schwann cell-ensheathed axons, both myelinated and non-myelinated, were present between the discontinuous perineurial layers, sometimes associated with reduplicated fragments of basal lamina and fibrous long spacing collagen. Perineurial cells and fibroblasts (the latter occasionally surrounding collagen pockets) filled the inter-fascicular epineurium (fig 4). Mast cells were present, often close to a blood vessel, within the epineurium and endoneurium.

Patient 3

A right partial thickness sural nerve biopsy of five fascicles showed a significant reduction of myelinated nerve fibres and an increased number of fibroblast-like cells within the fascicles. Some of the axons were thinly myelinated and associated with unmyelinated axons in structures resembling onion bulbs, and many persisting axons were small and unmyelinated.

Mutation analysis

Molecular studies did not detect a whole-gene deletion in any of the patients tested.

Patient 1

Mutation analysis detected a novel mutation in exon 23.2 of the NF1 gene. This alteration involved the deletion of cytosine at position 4071, which resulted in a premature stop codon. The deletion is predicted to generate a truncated neurofibromin of 1383 amino acids.

Patient 3

Mutation analysis revealed a substitution of leucine to proline at codon 1243 (CTG to CCG). This change results in the introduction of a non-aliphatic amino acid group into the NF1 guanosine triphosphatase related domain (GRD) of the peptide. Such a change is likely to disrupt the structure of the GRD and to be a disease causing mutation.

Patient 4

Mutation analysis failed to detect the mutation in this patient.

DISCUSSION

Of our 600 NF1 patients, eight had neurofibromatous neuropathy. None was initially referred because of symptoms relating to their neuropathy. Consequently, this condition might be under-diagnosed because sensory symptoms might be incorrectly attributed to cutaneous or subcutaneous neurofibromas. The presence of peripheral neuropathy may be revealed by a detailed examination of the nervous system.

There was a variable age of onset of neuropathic symptoms, but we noted a distinctive clinical phenotype, characterised by unusually early development of dermal and subcutaneous neurofibromas, occurring in large numbers and affecting extensive areas of the body. In six of the eight cases we detected neurofibromas arising proximally from multiple nerve roots in the spine. In our experience spinal nerve root neurofibromas are common in NF1 but do not usually cause neuropathic symptoms or signs. In our patients the neuropathic symptoms were predominantly mild and sensory. Only one patient required treatment with a combination of gabapentin and amitriptyline for neuropathic pain. The patients were followed up for between 1 and 10 years and there was no evidence of significant clinical or neurophysiological deterioration of the neurofibromatous neuropathy. This is in contradistinction to neuropathy in NF2 where there is variable disease progression. Patients 1 and 8 developed a high grade MPNST, which occurs with increased frequency in patients with NF1 and plexiform neurofibromas and often carries a poor prognosis.¹² It remains to be determined whether patients with early development of large numbers of neurofibromas and neurofibromatous neuropathy are at higher risk of malignant change.

We have identified disease causing mutations in patients 1 and 3. Molecular studies did not reveal any whole-gene deletions. A disease causing mutation was not detected in patient 4. All of our eight patients had NF1 as a new mutation and we are unable to assess the effect of familial aggregation of neurofibromatous neuropathy in NF1. In our patients neurofibromatous neuropathy was not associated with any particular genotype, which may be explained by influence from unidentified modifying genes.¹³ We (MU) have detected a mutation identical to that in patient 1 in another individual with NF1 who, however, does not have clinical manifestations of a peripheral neuropathy and has normal nerve conduction studies. Undoubtedly, a larger molecular study would be helpful in determining the mutational spectrum of NF1 neurofibromatous neuropathy.

We confirm the previous histological report of neurofibromatous neuropathy in NF1.³ Neurofibromas contain a mixture of Schwann cells (up to 80%), fibroblasts, perineurial cells, and axons^14–17 lying within a collagen-rich extracellular matrix. The perineurial disruption we have described is entirely consistent with a failure to initiate and/or maintain appropriate cell-cell signalling between Schwann cells, fibroblasts, and perineurial cells, and accords with previous findings that Schwann cells and fibroblasts are abnormal in NF1 associated neurofibromas.^18–22 The interdependence of axons and their associated Schwann cells is well established,^23,24 whereas relatively little is known about the interactions that occur between Schwann cells, fibroblasts, and perineurial cells.^25,26 However, there is now compelling evidence that Schwann cell derived desert hedgehog plays a significant role in perineurial differentiation and fascicle formation.^27,28 It is therefore of interest that a paracrine hedgehog signalling pathway has been implicated in the formation of neurofibromas in NF1.²⁹

The clinical and neurophysiological features in all our eight patients were of an indolent length dependent sensorimotor neuropathy with predominantly sensory signs. It was painless in seven out of eight patients. These features and the nerve biopsy findings indicate diffuse affection of peripheral nerve function. Neurofibromatous neuropathy is probably more common than previously thought since the symptoms were not prominent and detailed neurological examination was sometimes necessary to detect the clinical deficits. It has to be distinguished from the cumulative effects of multiple neurofibromas on spinal roots, nerve trunks, and peripheral nerves. We have evaluated all of our 600 NF1 patients for symptoms and clinical signs of neuropathy. Of these, seven (age range 29–35 years) had symptoms of tingling and numbness but the nerve conduction studies were all normal. Two of them had thickened spinal nerve roots. We are confident that clinical manifestations in our series are not more frequent than we have described, but we cannot exclude the possibility that subclinical peripheral neuropathy is more common. Our findings indicate that patients with peripheral neuropathy should be examined for evidence of peripheral neuropathy. If found, neurophysiological studies should be performed and other possible, often treatable, causes of peripheral neuropathy should be sought. In their absence a diagnosis of neurofibromatous neuropathy can then be made, often without resort to a peripheral nerve biopsy. Further research is needed to confirm that neurofibromatous neuropathy is associated with plexiform neurofibromas and an increased risk of developing MPNSTs. Further research is also needed to elucidate the contribution of the NF1 mutation and individual cell types to the diffuse neurofibromatous change in peripheral nerves which characterises neurofibromatous neuropathy.