Patients and methods

PATIENTS

We retrospectively reviewed the clinical and radiological records of children who presented with tuberous sclerosis and met the following criteria: (1) patients referred at the Saint Vincent de Paul Hospital between 1987 and 1996; (2) who had at least a 3 year follow up; (3) who had at least two MR scans during follow up, the first one during the year after diagnosis and the second during the last 2 years of follow up.

Sixty patients (32 girls and 28 boys) met these criteria.

Among them, we selected those who exhibited, on the last MRI, a mass located near the foramen of Monro. Patients with lesions elsewhere but no lesion near the foramen of Monro were excluded. There were 24 patients selected (40%), 11 girls and 13 boys, 16 with subependymal nodules and eight with subependymal giant cell astrocytomas .

METHODS

For these 24 patients, we retrospectively reviewed the serial brain examinations, CT and MRI, in a double blind study. We separated this population into two groups:

group 1 comprised 16 patients who presented subependymal nodules near the foramen of Monro but did not develop subependymal giant cell astrocytomas; group 2 comprised eight patients who presented subependymal nodules near the foramen of Monro and developed subependymal giant cell astrocytomas during follow up.

The following definitions were given to the radiologically disclosed lesions: (1) we considered as a subependymal nodule any non-obstructive nodular lesion measuring<10 mm in diameter; (2) any mass measuring >10 mm in diameter with growth more than 5 mm demonstrated on serial radiological investigations was viewed as subependymal giant cell astrocytomas.

For each CT and MRI, we reviewed the radiological characteristics of subependymal nodules and subependymal giant cell astrocytomas: their numbers and diameters, whether they were calcified (on CT) or enhanced after gadolinium injection (on MRI), and their growth from the previous investigation. We also studied the clinical characteristics of the patients, including those related to the tumour.

Because CT has not been performed in all patients and has been repeated in only a few patients, final radiological characteristics were described on MRI. Sequences and techniques retrospectively analysed in all patients included gradient echo T1 for axial slices and spin echo T2 for coronal and sometimes axial slices.Size of the mass was retrospectively manually measured in two orthogonal planes and converted to mm using a scale. The highest value was retained as the “diameter” of the mass.

Results

In both groups combined, 20 out of the 24 children presented with subependymal nodules at the first cerebral imaging. The four patients without any detectable nodules had had CT as first investigation, before the age of 1 year (table 1). In three of them, MRI was performed before 18 months of age and did not show any nodule.

GROUP 1

In the 16 patients who did not develop subependymal giant cell astrocytomas, the subependymal nodules exhibited special characteristics (table 2). The nodules located near the foramen of Monro were single in 10 patients and multiple in the other six. They were <5 mm in 12. They were completely calcified in 13 patients, all of them measuring>5 mm. Gadolinium injection was performed in 13 patients and produced enhancement in only two .

There were no familial antecedents of tuberous sclerosis in this group. Ten patients had had infantile spasms and six partial epilepsy. Follow up ranged from 3 to 18 years (mean 6 years). At the last visit, seven patients were seizure free with antiepileptic drugs but for the four others, epilepsy was only partially controlled. Four children had psychomotor delay although they were seizure free.

GROUP 2

Of the eight patients (five girls, three boys) who developed subependymal giant cell astrocytomas, three were familial. Six had had infantile spasms and two had partial epilepsy. All the patients but one (4) had evidence of subependymal nodules on the first radiological investigation. Patient 4 was the youngest and his first MRI was performed at 2 months of age. The tumours were then first noticed between 1 and 9 years of age (mean 4 years).

Five patients developed intracranial hypertension and were therefore operated on in an emergency setting. Two others were operated on because of evidence of tumour growth on serial radiological examination with signs of obstruction of the ventricular system but in a non-acute situation. The remaining patient has not been operated on yet because the ventricular dilatation remains stable despite the tumour growth. No changes in seizure frequency could be directly correlated with tumour development. All tumours removed underwent neuropathological examination and disclosed typical aspects of subependymal giant cell astrocytomas.

For the seven patients in whom subependymal nodules were detectable on the first imaging, tumours developed at the same place as the nodules (figs 1-4). Subependymal nodules preceding the tumour exhibited common characteristics (table 3). They measured>5 mm in diameter on first radiological examination except for those in one patient. They were only partially or not at all calcified. This was significantly different from the group who did not develop tumour (p<0.001, χ² test). Enhancement was obtained in five out of the seven patients who underwent gadolinium administration compared with only two out of 13 patients who did not develop tumour (p<0.05, χ² test).

Discussion

The present study shows an incidence of 13% of subependymal giant cell astrocytomas in a paediatric population of tuberous sclerosis. Nodules located near the foramen of Monro, not calcified and enhanced after gadolinium, have a higher probability of evolving into a tumour, particularly in familial cases of tuberous sclerosis. These radiological risk factors can be detected on MRI as early as the age of 1 year. From these data, a regular imaging follow up can be advised for patients with tuberous sclerosis at risk of developing a tumour, to manage its resection in a non-acute situation before the tumour becomes obstructive.

Many recent studies based on clinical and radiological criteria point to a total population prevalence of tuberous sclerosis of between 1/10 000 and 1/30 000.27 More than 90% of the patients have subependymal nodules but 2% to 14% of them will develop brain tumours.1 7 8 19 28-30 Growth of the tumour is usually reported in adults; in the Mayo Clinic series comprising 355 patients with tuberous sclerosis, 10% of patients developed subependymal giant cell astrocytomas.24 In the present series, the first one dedicated to paediatric patients, we found that 40% have masses adjacent to the foramen of Monro and that tumour growth can also occur before the age of 10 years, with an incidence similar to that of adults (13%). However, these rates are not epidemiologically relevant as our series was limited to a single centre and did not include non-epileptic patients with tuberous sclerosis.

All the tumours adequately identified by histological criteria were subependymal giant cell astrocytomas, in adults as well as in the present paediatric series. Although they are benign in nature and grow slowly, they represent the major cause of death in patients with tuberous sclerosis.7-10 27 Tumours resulted in death in 50% of the patients after the age of 10 years in the Mayo Clinic series.7 8 Death occurs because of increased intracranial pressure after misdiagnosed tumour growth, or less commonly after intratumorous haemorrhage.23 In the present paediatric series, most patients with brain tumour presented with symptoms of raised intracranial pressure due to obstructive hydrocephalus, and this is consistent with previous reports of patients with subependymal giant cell astrocytomas. Two children were operated on in a non-acute situation because of serial growth disclosed by MRI and growing hydrocephalus. Such a procedure might decrease mortality and morbidity that may be associated with an emergency operation.20-22Using new neurosurgical techniques such as microsurgery or endoscopy, it will be possible in the near future to routinely excise a growing mass, before hydrocephalus appears.31

It is important, therefore, to identify risk factors for developing a growing tumour when following up patients with tuberous sclerosis. Clinical features do not seem to be accurate before the occurrence of intracranial hypertension. Although some reports describe subependymal giant cell astrocytomas presenting only with worsening of seizures,17 no modification of seizure frequency was noticed in our series. How seizures would relate to the presence of subependymal giant cell astrocytomas is not clear; tubers can act as seizure foci,32 but subependymal giant cell astrocytomas themselves are unlikely to provoke seizures, although hydrocephalus may exacerbate a pre-existing seizure disorder.

Sequential radioimaging follow up shows that subependymal giant cell astrocytomas correspond to growing subependymal nodules. In the single case without preexisting subependymal nodules, a nodule could have been missed because CT was performed before the age of 1 year and MRI was performed on a first generation instrument. The continuum from subependymal nodules to subependymal giant cell astrocytomas has already been strongly suggested on serial MRIs and neuropathology as the histopathological features of the two entities are identical.9 26 Moreover, some subependymal nodules do have growing potential as they may be absent on the first imaging, even when performed after 1 year of age, and appear later. subependymal nodules and subependymal giant cell astrocytomas could therefore be considered as a continuum. However, subependymal giant cell astrocytomas involve a minority of patients with tuberous sclerosis whereas subependymal nodules are almost constant features. The subependymal nodules that will turn into tumour seems to have distinct characteristics that could be identified as risk factors: (1) diameter above 5 mm, (2) uncomplete calcification, (3) enhancement after gadolinium administration. These characteristics can be identified as early as 1 year of age. Sequential cerebral radioimaging is the only means of disclosing growing tumours but the adequate frequency for follow up investigation has not been established to date in children, given the need for sedation and cost as limiting constraints. The present study shows that nodules with tumorous potentiality could be identified early in life thus helping to define a high risk population. The minimal interval between the time the mass was detected near the foramen of Monro and the time the mass grew significantly, ranged from 1 to 3 years in our series (table 3). It seems, therefore, reasonable to repeat MRI every year in “at risk” children in a preliminary prospective way.

Physiopathological mechanisms underlying tumour growth in tuberous sclerosis are unknown but it is striking that, for retinoblastomas, familial patients seem to exhibit a higher risk for tumour development.33 Normal antioncogenes may be deleted or inactivated. Loss of heterozygocity has been demonstrated in tuberous sclerosis astrocytomas, as it has in retinoblastomas.34The risk for a second mutation is indeed particularly higher in patients who have inherited the disorder, and whose cells therefore carry the first mutation.35

Conclusion

We propose the following for children with tuberous sclerosis: MRI with and without gadolinium administration should be performed at the time of diagnosis of tuberous sclerosis and at the age of 2 years. When there are nodules located near the foramen of Monro and if they measure>5 mm in diameter, are not or incompletely calcified, and are enhanced by gadolinium, MRI should be repeated every year until the age of 10 years, particularly in patients with a family history of tuberous sclerosis. Re-evaluation of this procedure should be performed periodically in a prospective study to select more appropriately the indications and date for neuroimaging. Further follow up would also determine whether subependymal nodules and subependymal giant astrocytomas can be considered as two distinct lesions with respect of their growth potential.