Article Text


Unusual clustering of brain tumours in a family with NF1 and variable expression of cutaneous features


Neurofibromatosis type 1 (NF1) is one of the commonest autosomal dominant disorders in man. It is characterised by café au lait spots, peripheral neurofibromas, Lisch nodules, axillary freckling, skeletal dysplasia, and optic glioma. Symptomatic brain tumours occur in 1.5-2.2% of patients with NF1. We report here a family where seven members developed brain tumours. Of these, three have a clinical history strongly suggestive of NF1, while two do not fulfil diagnostic criteria for the disorder. A splice site mutation in exon 29 of the NF1 gene was found in these two subjects. This lesion is thought to be disease causative since it creates a frameshift and a premature termination of the neurofibromin protein. Different hypotheses to explain the unusual recurrence of brain tumours in this family, such as the nature of the mutation or cosegregation of other predisposing genes, are discussed.

  • neurofibromatosis
  • neurofibromin
  • splice site mutation
  • familial brain tumours

Statistics from

At least two distinct types of neurofibromatosis are recognised: neurofibromatosis type 1 (NF1) and neurofibromatosis type 2 (NF2). NF1 is one of the commonest autosomal dominant disorders in man with an estimated birth incidence of 1 in 2500/33001 and is characterised by the presence of café au lait spots, peripheral neurofibromas, Lisch nodules, axillary freckling, skeletal dysplasia, and optic nerve glioma. The disease shows high inter- and intrafamilial variability. The diagnostic criteria have been established by the National Institute of Health Consensus Development Conference.2 Numerous complications can occur and represent the major cause of morbidity and mortality. Symptomatic brain tumours are among these, occurring in 1.5-2.2% of patients.3 They do not tend to cluster in the same family.3-5 The gene for NF1 was mapped on the long arm of chromosome 17 in 19876 7 and cloned in 1990.8-10 Molecular diagnosis is now possible, though the presence of many different mutations makes this difficult. The most sensitive techniques detect approximately 70% of mutations.11

We report here a family where seven subjects developed brain tumours. Of these, three have a clinical history strongly suggestive of NF1, while two do not fulfil NF1 diagnostic criteria. However, a mutation in the NF1 gene was found in these two subjects.

Case report

The family pedigree is shown in fig 1.

Figure 1

Pedigree of the family. The proband is indicated by an arrow. Asterisks indicate subjects who have a splice site mutation in the NF1 gene.


The proband was referred to a general paediatrician at 3 years 6 months of age with an 18 month history of neurological symptoms consisting of left sided squint, dysarthria, ataxia, weakness, and recent onset of headache and vomiting. He had a family history of neurofibromatosis on the paternal side, his paternal aunt and cousins having had brain tumours and typical cutaneous features. Moreover, his paternal grandmother and great grandfather had died of brain tumours and his father had had a brain tumour removed when he was approximately 8 years old.

On examination he was alert. His visual acuity was 6/18 on both sides and he had mild disc pallor but no papilloedema. He had no cataract. Bilateral ptosis was present together with a right 6th nerve palsy. He was ataxic but able to walk and there were no focal deficits. On physical examination he had three small subcutaneous nodules on his buttocks and two to three hairy patches 0.5 cm in diameter on his back, but neither café au lait patches nor freckling.

CT and MRI scans showed the following: an intrinsic cord tumour causing focal widening of the mid dorsal cord at approximately the T4/T5 level; an extensive optic pathway tumour situated in the anterior end of the 3rd ventricle in the region of the optic chiasm, showing patchy enhancement after intravenous contrast, projecting upwards to the level of the foramen of Monroe on the left side and causing moderate lateral ventricular hydrocephalus; large swelling of the pons and medulla displacing the 4th ventricle backwards and extending downwards into the upper cervical cord, interpreted as a brainstem glioma; a small fleck of enhancement in the right cerebellar hemisphere and a further small nodular enhancement in the anterior part of the left cerebellar hemisphere, interpreted as metastatic or further primary tumours.

A ventriculoperitoneal shunt was inserted and he underwent chemotherapy.

At 6 years 4 months he was referred to the Oxford Neurofibromatosis Clinic. He showed one clearly visible café au lait patch on the anterior aspect of the left knee on naked eye examination and two café au lait patches on the back of the left leg visible only under UV light. There was no axillary freckling. On his back and buttocks there were four small lesions whose appearance was felt to be consistent with early neurofibromas. On his left brow there was a diffuse swelling approximately 2 × 4 cm, whose nature was unclear. His head circumference was on the 90th centile and height on the 50th centile.

Ophthalmological examination was performed later and showed three Lisch nodules.


The proband's father had a history from early childhood of nasal speech, dyskinesia of the left side of the extremities, and fainting spells. He was found to have a brain tumour and underwent surgery when he was 8 years old.

On physical examination at the age of 13 years he had nine café au lait patches in the axilla, back, and chest and a possible neurofibroma. A diagnosis of NF1 was made. On neurological examination he had anisocoria, dyskinesia of the upper and lower left extremities with abnormal choreiform and athetoid movements, and unilateral facial palsy on his left side. The diagnosis of a 3rd ventricle tumour partly removed was confirmed and he underwent radiotherapy.

At 19 years of age he still suffered from headache and abnormal movements.

When he was reassessed at 30 years of age after the diagnosis in his son, no café au lait patches could be seen but he had two subcutaneous nodules and multiple Campbell de Morgan spots. There were no Lisch nodules and the diagnosis of NF1 seemed unlikely. At the time of the son's assessment in Oxford, the parents had divorced and the father was unavailable for examination. He died the following year in a road accident. His relatives live in Israel and it has not been possible to contact them for personal assessment.


The proband's paternal aunt had an optic nerve glioma in childhood and old notes suggest she had skin changes typical of NF1. She had a brainstem glioma removed at the age of 27 and died at the age of 30 from complications related to it.


He suffered from squint, ptosis, and headache. CT and MRI scans performed when he was 6 years old showed a mass in the region of the optic chiasm in the suprasellar cistern, extending upwards into the region of the hypothalamus, consistent with an optic nerve glioma. He is clearly recorded as having café au lait patches. He underwent radiotherapy and has been followed up elsewhere.


He presented with headache at 7 years of age. His notes report that on physical examination he had multiple café au lait patches with axillary freckling and showed a slight imbalance affecting his left side. CT brain scan and MRI showed a very extensive tumour extending from the lower border of the second cervical vertebra, through the brain stem, posteriorly into the cerebellar hemispheres and up into the thalami and globi pallidi bilaterally. This was interpreted as a glioma and not biopsed. The child underwent chemotherapy and radiotherapy and has been followed up elsewhere.


Aged respectively 11 and 5, they were initially screened because of the strong family history. Neither of them has skin changes or neurological problems.


Both the proband's paternal grandmother, II.1, and great grandfather, I.1, were reported to have died of CNS tumours.

The clinical conclusion was that the pattern of tumours and clinical picture in III.1 and her children seemed classical of NF1. The proband and his father had minimal cutaneous involvement but their brain tumours fitted in better with the NF1 spectrum of tumours rather than that of NF2 and so this diagnosis was thought to be extremely unlikely clinically. What was extremely unusual about the whole family for NF1 was that every affected person had had brain tumours.

Materials and methods

DNA was extracted from peripheral blood.12 RNA was extracted from peripheral lymphocytes using RNAzol B (Biogenesis). In a systematic search for mutations in the NF1gene, PCR products from genomic DNA were screened for lesions in 21 exons of the NF1 gene using heteroduplex analysis.13 A total of 150 ng amplified DNA from either an NF1 patient and a control or from two different NF1 patients was added to 8 μl water and 0.6 μl EDTA 0.5 mol/l, heated at 94°C for three minutes, and then slowly cooled down to 37°C over one hour to allow formation of heteroduplexes. The resulting products were electrophoresed on a 0.8 × MDE (Mutation Detection Enhancement, Flowgen) gel for 19 hours at 5.5 W and visualised by silver staining.14 Direct sequencing was performed using the dideoxy chain terminator technique. cDNA synthesis was performed at 42°C for one hour in a 20 μl reaction volume containing 5 μg total cellular RNA, 100 ng random hexamers, 40 U RNA guard (Pharmacia), 1 × first strand buffer (Gibco BRL), 0.01 mol/l DTT, 0.5 mmol/l dNTPs, and 220 U of superscript II reverse transcriptase (Gibco-BRL). Hot start PCR was carried out to amplify segment 4 of the NF1 transcript containing exons 28-3811 in a 30 μl reaction volume containing 1 μl of 1 strand synthesis, 1 × PCR buffer, 0.8 mmol/l dNTPs, 7.5 pmol of each primer, 0.1% Triton X100, and 0.75 U ofTaq polymerase (Promega, storage buffer A).

The PCR products were purified using Quiaquick purification kit (Quiagen) and ligated into pGEWM-T vector (Promega). Plasmid DNA was extracted from individual transformed colonies after overnight culture and sequenced using Thermo Sequenase cycle sequencing kit (Amersham).


Analysis of genomic DNA of the proband, IV.6, showed a lesion in exon 29 (figs 2 and 3). Further characterisation by direct sequencing showed a splice site mutation (5546+1G to C). This change in the splice donor consensus sequence flanking intron 29 was also identified in the proband's father but was not present in the patient's unaffected brother and sister. The lesion was expected to result in the skipping of exon 29. RNA analysis from the patients identified two abnormal bands. Sequence analysis of cDNA clones showed two abnormal mRNAs in addition to the normal one, alternatively missing exon 29 or exons 29 and 30.

Figure 2

RT-PCR products of exons 28-38, run on a 2.0% agarose gel. Lanes 1-6 show the normal sized product (1989 bp) found in six NF1 patients, lane 7 the alternative splicing occurring in the proband, IV.6, lane 8 a negative control containing no cDNA, and lane 9 a 1 kb ladder. The two additional bands seen in lane 7 correspond to splicing out of exon 29, leading to a product of 1648 bp, and to splicing out of exon 29 + 30, leading to a product of 1449 bp.

Figure 3

Sequence of proband (IV.6) cDNA clones. Clone B sequence exhibits the correct splicing of exon 28 to exon 29. Clone A shows the loss of exon 29 with splicing of exon 28 to exon 30 and clone C the loss of exons 29 and 30 with splicing of exon 28 to exon 31.


The clinical history in one branch of this family (the proband's paternal aunt and cousins) is suggestive of NF1. All affected members are reported to have had typical skin changes. In addition, they all suffered from optic nerve or cerebral gliomas. They were not available for molecular testing. A diagnosis of NF2 in these subjects would be unusual as all cerebral tumours were gliomas and not meningiomas and there is no history of deafness in any of the family members to suggest undiagnosed vestibular schwannomas.

The cutaneous features in the proband and his father were extremely mild. Usually, café au lait spots appear in the first year of life in 82% of patients with NF1 and are present in all affected subjects at 4 years or less. All patients have more than six spots by the 10th year of age.3 The proband did not show any café au lait spots when he was first assessed (3 years 6 months) but had three of them at 6 years of age. The proband's father was clearly reported to have nine when he was 13 years old but did not have spots in his 30s. Although 93% of patients have more than six café au lait patches at the age of 30,3 their frequency is known to decrease with age and they can sometimes disappear. Neither the proband nor his father showed axillary freckling, which is present in 70% of affected subjects.3 Lisch nodules were not identified in the father but were present in the proband when he was 6 years old.

Since the proband and his father did not have typical cutaneous features of NF1, the diagnosis of NF2 was considered but the molecular evidence is against this. In fact, a splice junction mutation in intron 29 of the NF1 gene was identified both in the proband and his father. The mutation was a base change in the splice donor site of intron 29.

A number of splice site mutations in the NF1gene have been identified and reported to the International NF1 Genetic Analysis Consortium (http://www.clam/nf/nf1gene). All reported splicing abnormalities are expected to result in a truncated neurofibromin protein. In this case, splicing out of exon 29 or exons 29 and 30 results in a frameshift and premature termination in exon 30 or exon 31 respectively. This lesion is likely to be disease causative since it was also detected in the genomic DNA of the affected father but not in the unaffected sibs. Recently, Park et al 15 described alternative splicing of exon 29 or exon 30 or both. Interestingly, the skipping of exon 30 is exclusively found in brain tissue. Moreover, the nature of the mutation may also explain the variable expressivity of the disease in different family members, as described for other conditions.16

Although NF1 seems the most likely diagnosis in this family, the presence of symptomatic tumours in all affected subjects is extremely unusual. The incidence of brain tumours in NF1 has been overestimated in earlier publications owing to lack of awareness of a possible NF2 diagnosis. Gliomas, particularly of the optic nerve, are the only CNS tumours with a markedly increased frequency in NF1.3Subclinical optic nerve tumours occurred in 15% of patients in a hospital based series.17 In a population based study of 69 families, 135 affected members were identified and symptomatic brain tumours were present in 1.5-2.2%. These tumours do not tend to cluster in the same family.

The neurofibromin gene shows somatic mutations in gliomas and primitive neuroectodermal tumours, suggesting a role in tumour suppression in cells of ectodermal origin.18 The possible relevance of both these data and the nature of the mutation in our family are difficult to assess. To date, no genotype-phenotype correlation has been made for NF1 other than in children with whole gene deletions.3 19-21 It has been proposed that an association could exist in NF1 patients between the presence of optic gliomas and other tumours of the CNS, suggesting important pathophysiological differences between patients with and without gliomas.22 This might be related to the presence of different mutations.

Another possible explanation for the strong tendency to develop brain tumours in this family might be the cosegregation of another gene predisposing to brain tumours. TP53 was suggested as a candidate, but no other features were seen in the family that would be associated with germline mutations ofTP53. 23 In addition, an abnormal TP53 allele would be unlikely to cosegregate with the abnormal NF1 allele in the numerous family members reported here.

In summary, this family is unusual because there is a very high incidence of brain tumours, which have been confirmed as gliomas in four of the seven affected subjects and are present in every family member reported as having café au lait patches. A splice junction mutation in intron 29 of the NF1 gene has been identified in two family members (the proband and his father) with brain tumours and extremely mild cutaneous features of NF1. No other malignancies have occurred in the family, making it unlikely that another cancer predisposition gene is cosegregating with the NF1 phenotype or mutation.

Finally, rare cases of familial gliomas exist, outside the range of the clinical syndromes in which CNS cancer heredity is well established.24-26 It is possible that some of these could be caused by mutant NF1 alleles.


The authors would like to thank Miss J Maynard and Mr Domenico Carratta for technical assistance and the UK NF1 association for financial support.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.