| Radiologic findings of the head and spine in neurofibromatosis 1 (NF1) in Northern Finland | ||
|---|---|---|
| Prev | Chapter 2. Review of the literature | Next |
The extent and nature of the different CNS lesions in NF1 have become better known now that computed tomography and magnetig resonace imaging have been applied to both the diagnosis and the follow-up of the patients and their relatives. The use of MRI has shown that T2 hyperintense lesions are the most common brain lesions in NF1, especially in children, and that optic gliomas are the most common tumours. Different types of intraparenchymal benign or malignant tumours and brainstem tumours have been reported to occur in NF1 (Huson et al. 1994, Bilaniuk et al. 1997), and nontumorous hydrocephalus and macrocephaly have also been shown to be features of NF1 (Huson et al. 1994, Spadaro et al. 1986, Gonzales et al. 1990, Menor et al, 1991). Medullary tumours occur, but they are more common in NF2 than in NF1. Spinal neurofibromas and other lesions are well known to be part of the NF1 phenotype (Egelhoff et al. 1992, Thakkar et al. 1999).
The most common abnormality seen on MR imaging in patients with NF1 are small multiple areas of high signal intensity in the brain on T2-weighted images (Goldstein et al. 1989, Mapstone 1992 ). These lesions have also been called unidentified bright objects (UBOs; Balestri et al. 1994), unidentified neurofibromatosis objects (UNOs; Pont and Elster 1992), focal areas of signal intensity (FASIs; Cohen et al. 1990–91), neurofibromatosis bright objects (NBOs; Mukonoweshuro et al. 1998) or regional signal hyperintensities (RSHs; Steen et al. 2001).
T2 hyperintense lesions are often bilateral and may be symmetric or asymmetric, and there is generally no evidence of a mass effect, surrounding oedema or contrast enhancement. They may appear as either singular or multiple lesions. On T1-weighted images, the lesions usually appear isointense to slightly hypointense (Duffner et al. 1989). The lesions in the globus pallidus may occasionally have a mild mass effect, and they may be bright on T1-weighted images (Aoki et al. 1989, Sevick et al. 1992, Smirniotopoulos & Murphy 1992, Ferner et al. 1993). The lesions in the globus pallidus are usually sharply defined, whereas the lesions in the cerebellum and brainstem tend to be more confluent and may involve the midbrain, pons and medulla contiguously (Goldstein et al. 1989). On CT, there are usually no corresponding abnormalities, although focal areas of hypodensity have been reported in the basal ganglia on CT (Duffner et al. 1989). Menor et al. (1992) compared the visualization of bright lesions in a prospective CT and MRI study of 41 children with NF1. Lesions were detected on MR in 22 (45%) of the patients, but on CT in only 7 patients (17%). The smallest lesion detected on CT measured 0.8 cm, and there was no correlation between the size of the lesions on MR and their visibility on CT. None of the observed lesions enhanced with contrast medium on either MRI or CT.
The occurrence of T2 hyperintense lesions has been subject to several studies since their early detection in 1989 (Duffner et al.1989, Goldstein et al.1989). In contrast to most hamartomatous lesions in NF1, they often disappear or diminish over time, the pathological changes underlying them have not yet been clarified, and their significance has also remained largely unknown.
The prevalence of T2 hyperintense lesions in children with NF1 has beed found to be high, ranging from 60% to 93% of affected children in different studies, while their frequency in adults is low (Duffner et al.1989, Pont & Elster 1992, Sevick et al. 1992, Balestri et al. 1994, Menor et al. 1998) (Table 1). They have been encountered from the early age of about one year onwards, and they have been found useful as an additional diagnostic criterion of NF1. In the study of Menor et al. (1998), MR imaging contributed to a definitive diagnosis of NF1 in 53% of suspected cases, Parazzini et al. (1995) found T2 hyperintense lesions in 100% of their 11 NF1 patients, and Griffiths et al. (1999) found them in 93% of the children examined with MRI.
T2 hyperintense lesions have been reported to be more frequent in NF1 patients with optic gliomas (Hurst et al. 1988, Di Mario et al. 1993). In the study of Aoki et al. (1989), they were found in 89% of the patients with optic glioma but in only 44% of the patients without glioma. In the study of Balestri et al. (1993), the corresponding figures were 67% and 47%. All the seven patients of Van Es et al. (1996) with T2 lesions also had optic glioma. Rosenbaum et al. (1999), on the other hand, found T2 lesions in 86% of the studied children with NF1 (n = 31), but were unable show a correlation between them and optic pathway gliomas.
As listed in Table 1 and also observed in other studies (Hurst et al. 1988, Goldstein et al. 1989, Menor et al. 1992, Di Mario et al. 1993, Ferner et al. 1993, Shu et al. 1993), T2 hyperintense lesions have been encountered at numerous locations of the brain, but most often in the basal ganglia/globus pallidus. The cerebellum and the brainstem have also been common locations for lesions. They have been found in the intracerebral optic regions in 8 to 40% of the patients (Duffner et al. 1989, van Es et al. 1996, Menor et al. 1998). Sener et al. (2000b) found hamartomas of the septum pellucidum in three out of 86 NF1 patients (3.4%). The lesions were especially detectable on fluid attenuation inversion recovery (FLAIR) and proton-weighted images. T2 hyperintense lesions have also been described in hippocampal areas (Turbidy et al. 2001).
Table 1. Occurrence, localization and follow-up of T2 hyperintense brain lesions in NF1 in previous studies.
| Study | Duffner et al. 1989 | Aoki et al. 1989 | Sevick et al. 1992 | Itoh et al. 1994 | van Es et al. 1996 | Menor et al. 1998 |
|---|---|---|---|---|---|---|
| Patients (n) | 47 | 53 | 43 | 44 | 50 | 72 |
| Age range (y) | 0.8–18 | 1–78 | 1–31 | 1–50 | 8–16 | 0.8–14 |
| Mean age | 15.2 | 10 | 16.2 | 7 | ||
| Patients with T2 lesions (n/%) | 29/62 | 32/60 | 34/79 | 34/77 | 32/64 | 56/78 |
| Sites of T2 lesions (% of patients affected) | ||||||
| basal ganglia | 33* | 3.** | 88 | 48 | 72 | |
| cerebellum | 21* | 13 | 47 | 15 | 57 | |
| cerebellar peduncles | 3* | 1.** | 36 | |||
| dentate nuclei | 26 | |||||
| brainstem | 12* | 2.** | 22 | 79 | 12 | 44 |
| midbrain | 4.** | 10 | ||||
| periaqueductal area | 6 | |||||
| internal capsules | 3* | 19 | 4 | 11 | ||
| thalamus | 3* | 5.** | 15 | |||
| optic system | 18* | 40 | 8 | |||
| supratentorial white matter | 6* | 6.** | 19 | 21 | 24 | |
| corpus callosum | 1* | 4 | ||||
| other | 4* | 11 | ||||
| MRI follow-up T2 lesions | ||||||
| patients | 18 | 13 | 19 | 13 | 4 | 45 |
| mean follow-up time (y) | 2 | 2 | 2 | 4–5 | 3 | |
| regression of T2 lesions (no of patients) | 1 | 2 | 7 | 7 | 4 | 42/40*** |
| unchanged | 16 | 10 | 3 | 2 | 27/34*** | |
| progression | 0 | 1 | 4 | 3 | 25/16*** | |
| mixed | 1 | 4 | 1 | 7/11*** | ||
| * = number of sites of lesion, ** = in an order of frequency, *** = % of the patients, basal ganglia/cerebellum and brainstem | ||||||
The development and follow-up of T2 hyperintensive lesions has been reported in several studies (those in Table 1; Mautner et al. 1995, Carella & Medicamento 1997, Griffiths et al.1999). Although the patient groups have been relatively small and the mean follow-up times only a few years, it has appeared that these lesions may increase in size and number throughout childhood, but that they often show a tendency to resolve with increasing age and have less often been seen in adults. In the studies of Sevick et al. (1992) and Itoh et al. (1994), the patients showing progression of the lesions had a mean age of 5 and 3.7 years, while the mean ages of the patients showing regression were 13 and 11.4 years, respectively. In addition, some of their patients showed lesions that remained unchanged, and some had mixed-type lesions. In the study of Mautner et al. (1995) of 20 NF1 patients aged 5–51 years, 11 patients (median age 10 years) had unchanged MR images. A reduction in the size or complete resolution of the lesions was found in 9 patients (median age 10 years).
In the study of Menor et al. (1998), the lesions showed progression in children aged under 10 years, but diminished in older patients. 18% of the 44 patients studied presented with atypical unenhanced lesions showing either edema, some mass effect or hypointensity on short TR images; two of the lesions were considered brainstem gliomas. Griffiths et al. (1999) found T2 lesions to be rare in children under 4 years of age, common at 4–10 years of age, and significantly reduced after 10 years of age. Atypical appearance of T2 hyperintense lesions was reported in two patients by Raininko et al. (2001). In an 18-year-old patient, an enhancing lesion was found bulging into the lateral ventricle, but it had almost disappeared after two years´ follow-up. The other patient was an 11-year-old boy who had a non-enhancing lesion in the left cerebellar hemisphere at a location that had been normal in MRI six years previously. The lesion remained stable one and a half years later. Morris et al. (1997) also reported spontaneous disappearance of an enhancing brain lesion in a child with NF1.
The exact nature of T2 hyperintensive lesions is not yet exactly known. This may be partly due to nonspecificity of the bright lesions: prolongation of T2 relaxation occurs with various pathologic lesions, and no specific tissue composition can be deduced on the basis of the visual finding. They have been suggested to represent hamartomas, brain heterotopias, zones of gliosis, glial scars, low-grade tumours, areas of delayed demyelinization or areas of vacuolar or spongiotic change. Delayed or abnormal brain maturation in these areas has also been suggested to cause the lesions (Braffman & Naidich 1994, Itoh et al. 1994, Di Paolo et al. 1995). In two NF1 children, Zimmerman et al. (1992) were able to demonstrate microscopically the presence of hyperplastic or dysplastic glial proliferation in the T2 hyperintense areas of the globus pallidus and midbrain peduncles.
The bright lesions have also been thought to represent infarcts secondary to vascular occlusive dysplasia, which is known to occur in NF (Crawford et al. 1988), or increased free-water content of the tissue (Mirowitz et al. 1989, Balestri et al. 1993). Normal brain tissue has also been demonstrated at biopsy of T2 high-intensity foci (Di Mario et al. 1993). Because the lesions in the basal ganglia may show a mass effect and be hyperintense even on T1 images (whereas the white matter lesions are generally isointense on T1 images), it has been suggested that the basal ganglia lesions may have a different pathogenesis (Sevick et al. 1992).
Ectopic collections of Schwann cells have been described in NF1 patients, mostly in the spinal cord, but also in the cerebral cortex and basal ganglia (Rubinstein 1986, Mirowitz et al. 1989), and hamartomatous collections of melanocytes in the brain have also been found. Because lesions manifesting as hyperintensities on T1 images are uncommon (lipids, methemoglobin, melanin), the authors suggested that the hyperintensities in NF patients may be caused by dense clustering of myelinated Schwann cells or melanocytes in the basal ganglia, while the small focal signal hyperintensities on T2 images could be caused by other tissues within the hamartomas or could result from reactive gliosis leading to an increased water content. Myelinated areas of the brain appear to have high signal intensity compared to the cortex and white matter on T1 images, and the high intensity in the case of melanocyte collections could be related to the paramagnetic properties of free radicals in melanin. Similar T1 hyperintensity has been detected in brain metastases of melanoma.
The clinical significance of hyperintense T2 lesions in NF1 patients is unknown. It has been suggested that such lesions could be associated with the increased frequency of mental retardation or specific learning and behavioural problems and seizures in the patients. The studies so far done on the correlation of the high-intensity T2 foci and the neurocognitive deficits in NF1 patients have been contradictory. In an extensive literature review by Ozonoff (1999), a summary of 71 articles from the years 1986 to1997 on studies of neuropsychological function in NF1 is presented. According to it, the presence of T2 hyperintense lesions of the brain did not show a consistent relatioship with the cognitive phenotypes of the patients. In some studies, the presence of T2 hyperintense brain lesions was considered predictive of a lower IQ, while in some reports the total number of brain lesions correlated negatively with cognitive function.
Although the T2 hyperintense lesions have, in most cases, been regarded as innocent changes, development of areas with T2 signal hyperintensity into astrocytoma has been described. Carella and Medicamento (1997) presented a 23-year-old patient with NF1, in whom an anaplastic astrocytoma developed from areas of T2 signal hyperintensity in the cerebellum. Previous follow-up examinations two and three years earlier had not shown any progression or any clue towards an unfavourable outcome. In the 32-year-old patient described by Miaux et al. (1997), glioblastoma multiforme developed from a non-enhancing hyperintense periventricular lesion during a follow-up of three years.
Hyperintense cerebral white matter lesions on T2-weighted MR images have been described in conditions other than NF1, and they have thus not been thought to be pathognomonic to NF1. They have been found in ischemia, infections, demyelinating diseases, such as multiple sclerosis (MS) and amyotropic lateral sclerosis (ALS), Wilson’s disease, dementia and Whipple’s disease, and they have also been found in increasing numbers in otherwise healthy elderly persons along with increasing age (Duprez et al. 1996, Nazer et al. 1993, Yetkin et al. 1993).
Artifacts and normal structures, such as a deep sulcus, substantia gelatinosa, a gyrus or a perivascular Virchow-Robin space, may mimic true white matter lesions. Miyazaki et al. (1991) have described deep white matter lesions mimicking T2 hyperintensive lesions also in cases of tuberous sclerosis, congenital muscular dystrophy, congenital myotonic dystrophy, autism, epilepsy and congenital heart disease.
Mirowitz et al. (1989) found areas of increased signal intensity relative to cerebral white matter on T1-weighted MR images in the basal ganglia region in seven out of 35 NF1 patients (mean age 12 years). The globus pallidus and portions of capsula interna were involved in all patients, and the posterolateral parts of the thalami were involved in five cases. In addition, one patient had T1 hyperintensities in the corpus callosum. In most patients, the hyperintensities on T2 images were much more limited than the hyperintensities on T1 images. The lesions appeared smoothly marginated, somewhat nodular and with no mass effect, edema or contrast enhancement on all sequencies.
Terada et al. (1996 ) studied 35 NF1 patients and found hyperintense T1 lesions in 8 patients aged 3 to 17 years. Seven of them showed hyperintense T2 foci in the same areas in the globus pallidus. No enhancement with gadopentate dimeglumine was noted. It was found that, in some patients, the T1 foci developed later than the T2 foci and did not regress during follow-up for up to 90 months, while the T2 foci showed regression. They suggested that the T1 lesions may be characteristic of lesions of basal ganglia only, but not of lesions in the posterior fossa or cerebral white matter.
T1 hyperintense lesions of the basal ganglia due to manganese deposition have been reported in patients who have undergone gastrointestinal surgery with perioperative parenteral nutrition (Iwase et al. 2002). T1 hyperintense lesions of the globus pallidus were described in children with portosystemic encephalopathy (Yanai et al. 1995) and also in children with chronic liver disease without obvious hepatic encephalopathy (Ballauf et al. 1994). Similar lesions are also found in patients with melanotic metastases of melanoma (Isiklar et al.1995), in herpes simplex encephalitis (Shian & Shi 1996), in acute measles encephalitis (Voudris et al. 2002) and in tuberous sclerosis (Wippold et a. 1992) as well as in Tay-Sachs disease (Mugikura et al. 1996).
CNS tumours in NF1 primarily originate from astrocytes and neurons (Aoki et al. 1989). Optic and parenchymal gliomas are the most usual tumours, and brain gliomas have been reported in 1% to 3% of patients with NF1. They are usually low-grade astrocytomas, appearing earlier than in the general population, and also more often multicentric. Gliomas of the brainstem, hypothalamus and the third ventricle are also relatively common, while diffuse gliomas of the cerebral hemispheres, cerebellum or spinal cord occur rarely (Gardeur et al. 1983, Bogdanno et al. 1988, Huson et al. 1989, Shu et al. 1993). Guillamo et al. (2002) found 127 CNS tumours in 104 NF1 patients; 48 tumours were optic gliomas, 21 brainstem tumours and 22 tumours in other locations. 21 patients (20%) had multiple tumours.They found extra-optic location, tumour diagnosis in adulthood and symptomatic tumours to be associated with shorter survival. Malignant gliomas are rare in NF, and lobar gliomas have been reported to be more infiltrative than paraventricular ones (Gardeur et al. 1983). Vinchon et al. (2000) had opereted on six NF 1 children with radiologically progressive cerebellar astrocytomas, one of them malignant. They found the tumours associated with NF 1 to have a better prognosis than the non-NF-associated ones. Ependymomas, primitive neuroectodermal tumours, meningiomas and neurofibrosarcomas also occur (Sato 1992, Braffman & Naidich 1994), and pituitary adenomas have been described (Gardeur et al. 1983). Colloid cysts (Bogdanno et al. 1988) and epidermoid cysts (Shu et al. 1993) have also been reported.
In a series of 69 Northern Finnish NF1 patients with histologic tumour verification, Pöyhönen et al. (1997b) reported five patients with malignant brain tumours. These included three children with pilocytic astrocytoma and two adults with glioblastoma multiforme and anaplastic astrocytoma of the cerebellum, respectively.
In the series of van Es et al. (1996), only one patient out of the 50 studied NF1 children had a CNS tumour (ependymoma) other than optic glioma. Menor et al. (1998) reported 8/72 children with NF1 to have a CNS tumour other than optic glioma; of these, two had been verified histologically.
Astrocytic brainstem tumours in NF patients may differ in presentation, histology and natural history from similar-appearing tumours in non-NF1 patients. The tumours in NF patients may progress slowly for an indeterminate length of time, whereas most brainstem tumours in children without NF progress rapidly (Cohen et al. 1990–91). Raffel et al. (1989) suggested that the brainstem tumours in NF1 patients might be extremely slowly growing tumours or static hamartomas, although their nature is difficult to ascertain. Neurofibromatosis-associated brainstem gliomas seemed to have a better prognosis than sporadic gliomas.
Bilaniuk et al. (1997) described 25 NF1 patients shown by MRI to have diffuse (n = 12) and focal (n = 13) brainstem tumours diagnosed at the mean age of 7.8 years (range 1.1–15.2 years.) In four patients, the tumour was found on routine screening. Medullary enlargement was most frequent (68%), followed by pontine (52%) and midbrain enlargement. In 30% of the patients both brainstem and optic pathway tumours were present. Surgery was performed on four patients and revealed fibrillary astrocytomas, one of which progressed into an anaplastic astrocytoma. Diffuse tumours in patients with NF1 appear to have a much more favourable prognosis than those in patients without NF1.
Molloy et al. (1995) presented 17 NF1 patients with brainstem tumours, located primarily in the medulla in 14 (82%) patients, in contrast to the pontine tumour location in the non-NF1 patients. The mean age of the NF1 patients at tumour diagnosis was 101 months. 5 (29%) patients had neurologic signs and symptoms suggestive of brainstem dysfunction. Contrast enhancement was seen in 7 patients (41%) on MR. 6 (33%) patients showed evidence of radiographic tumour progression, while only three showed clinical progression. In two of these, histological study revealed either fibrillary or anaplastic astrocytomas. 15 (88%) patients remained alive after a median follow-up time of 52 months. The brainstem tumours of the NF1 patients showed less aggressive behaviour than the tumours in the non-NF group. The tumours can be differentiated from non-neoplastic T2 high signal intensities because they exhibit focal or diffuse brainstem enlargement, have a mass effect, may show enhancement with gadolinium and often result in obstructive hydrocephalus.
Some periaqueductal gliomas may mimic the appearance of benign aqueductal stenosis, and it has been suggested that aqueductal stenosis in neurofibromatosis should be considered as suggestive of probable mesencephalic glioma necessitating follow-up examinations (Gardeur et al. 1983).
It is very difficult to differentiate benign, insignificant lesions from true gliomas. The presence of vasogenic edema and mass effect, enhancement with gadolinium, cavitation of the lesion, hypointensity of the lesion on T1 images and increasing size of the lesion on serial scans are suggestive of a glioma in patients with NF (Mulvihill 1990).
Faravelli et al. (1999) presented an unusual clustering of brain tumours in a family with NF1 and various cutaneous features.
Obstructive hydrocephalus associated with NF1 may be caused by infratentorial neoplasms, midline arteriovenous malformations or primary, nontumorous aqueductal stenosis (Spadaro et al. 1986, Huson et al. 1988, Aoki et al. 1989, Gonzales et al. 1990, Menor et al. 1991). The earliest cases were described in 1927 and 1940 (Horwich et al. 1983). Afifi et al. (1988b) presented a series of 289 NF patients collected over 20 years. Of them, hydrocephalus was found in 15 (5%) cases; eight were caused by intracranial tumours, two had Chiari malformation, three had aqueductal stenosis, and two were of undetermined etiology. Pou-Serradell et al. (1989) presented nontumoural hydrocephalus in 9 out of 30 NF1 patients referred to MRI because of neurological problems.
The frequency of nontumorous aqueductal stenosis in NF1 has been 2% according to Riccardi and Einchner (1986). In most cases, nontumoural aqueductal stenosis has been detected in the first or second decade of life. Afifi et al. (1988a) found a total of 28 NF1 patients with nontumoural aqueductal stenosis in the literature.
Afifi et al. (1988a) studied three cases of nontumoural aqueductal stenosis in children with NF1. In them, MRI revealed an area of high intensity in the tectal and periaqueductal regions, and the findings remained stable during follow-up for one year. On T1-weighted images, kinking of the aqueduct without surrounding hypodense lesions was found. The hyperdense areas on T2-weighted images were assumed to represent hamartomas.
The pathogenesis of aqueductal stenosis in NF1 is heterogeneous but has been suggested to be due to direct expression of the NF1 gene. Periaqueductal gliosis has been found in some cases of stenosis, and septum formation and forking of the aqueduct have also been detected (Horwich et al. 1983, Spadaro et al. 1986, Schreiber & Quade 1990). Also, polyp-like growths of ependymal granulation tissue obstructing the aqueduct have been seen (Horwich et al. 1986). In MR images, Braffman & Naidich (1994) were able to show hydrocephalus, bulbous dilatation of the third ventricle, fusion of soft tissue of the midbrain with resultant obstruction of the aqueduct and absence of the CSF flow-void sign in their patients with NF1.
Afifi et al. (1988b) presented two adult NF1 patients with ventriculomegaly and Chiari type 1 malformation. Both patients had similar anomalies of the craniocervical junction: hypoplastic occipital condyles, a short basiocciput and an invaginated dens with compression on a low-lying medulla.
It is important to differentiate between instances of hydrocephalus due to aqueductal stenosis and macrocephaly, which is common in NF1. Macrocephaly usually appears without neurological symptoms, and the patient’s head circumference remains constantly at or above the 97th centile, whereas in cases of obstructive hydrocephalus, there is an increase of the head circumference over time relative to the growth curves (Horwich et al. 1986).
Mild ventricular supratentorial dilatation has been detected commonly in NF patients (Gardeur et al. 1983, Spadaro et al. 1986). Maki et al. (1981), using CT scans on NF patients (mostly NF1), demonstrated mild dilatation of the lateral ventricles without periventricular hypodensity in 44% of the 18 children. 60% of the 21 adult patients showed slight dilatation of the lateral ventricles, and one third of them also had dilatation of the third ventricle. In some of these patients, ventricular stasis without a block was found on radioisotope cisternography. Altogether 55% of the children and 85% of the adults had abnormal findings on CT, mild ventricular dilatation being the most common finding. Callosal agesis, cavum septi pellucidi and atrophic changes in brain were also found in some patients.
Enlargement of the quadrigeminal cistern, cisterna magna or subarachnoid spaces or subarachnoid cysts may be seen in NF1 (Menor et al. 1991, Di Mario et al. 1993, Ruggieri et al. 1995). Chiari I and II type malformations have been reported in a few cases of NF1 (Bognanno et al. 1988, Menor et al. 1991, Ruggieri et al. 1995), and hypoplasia of the temporal lobe has been detected in some patients (Imana Martinez & Martinez San Millan 1993). Porencephaly (Ruggieri et al. 1995) and localized megalencephaly, polymicrogyria and pachygyria have also been found (Gardeur et al. 1983). The cerebral ischemia described in NF1 may be due to arterial dysplasia.
Enlargement of the internal auditory canal due to dural ectasia may be seen in NF1 patients (Sarwar & Swischuk 1977, Aoki et al. 1989, Shu et al. 1993, Braffman & Naidich 1994). This may cause audiologic dysfunction. On MRI scans, the seventh and eighth cranial nerves appear normal in the enlarged auditory canal and there is no enhancement with gadolinium.
Agenesis of the corpus callosum has been described in association with NF1 (Jacoby et al. 1980, Maki et al. 1981). Atlas et al. (1988a) presented total agenesis of the corpus callosum and gyral malformations on MRI examinations of identical twins with NF1. Hypoplasia and lipoma of the corpus callosum have also been reported (Shu et al. 1993). In a patient with NF1, thin corpus callosum detected on MR scans was associated with microcephaly (Di Mario et al. 1993). Ruggieri et al. (1995) found two patients with NF1 with partial agenesis of corpus callosum on MRI. A cyst of the corpus callosum was reported in a patient with NF1 by Duffner et al. (1989).
Huson and Hughes (1994) have summarized the malformative CNS lesions detected by several investigators in autopsy studies of NF1 patients. These have included abnormal architecture of the cerebral cortex, hypertrophic gliosis of the optic nerve, hyperplastic proliferative gliosis, subependymal gliofibrillary nodules, which are histologically indistinguishable from small subependymal pilocytic astrocytomas and have been associated with aqueduct stenosis in NF1, meningoencephalic cerebellar gliosis, micronodular vascular proliferation in the nervous system and the visceral organs (vascular neurofibromatosis) and hydromyelia.
Pathologic intracranial calcifications occur in 10% of NF patients (Galanski and Benz 1978), mostly in ones with NF2 with meningiomas. Laminated type of unilateral cerebellar calcification in CT examinations has been found in NF1 (Jacoby et al. 1980 ), and the calcification was thought to be due to hamartomatous lesions. Suprasellar extensions of optic gliomas in NF1 may sometimes show calcification, which may simulate craniopharyngeoma (Davidson 1966).
Arts and Van Dongen (1986) found multiple subependymal nontumorous calcifications without contrast enhancement or expansivity in three patients with NF2. On the basis of this finding and the literature referred to, they suggested that nontumorous intracranial calcifications may be a rare sign of both NF1 and NF2.
Both occlusive lesions and aneurysms of cerebral arteries have been described as associated with NF1 in occasional patients.
In the occlusive disease of cerebral arteries described by several authors (Tomsick et al. 1976, Taboada et al. 1979, Crawford et al. 1988), the angiographic findings have included occlusion of the supraclinoid internal carotid artery, the proximal anterior cerebral and middle cerebral arteries and, less commonly, the more distal branches of these arteries. Occlusion of the posterior cerebral artery has been reported, although involvement of the posterior circulation is rare. Basal teleangiectasiae producing a moyamoya appearance have been described to be associated with occlusions of the major intracranial vessels.
The occlusive disease of the cerebral arteries in NF has been found most often in children, in whom it has often been associated with cerebral infarction and has not been progressive. The occlusive lesions have been thought to be caused by intimal smooth muscle cell proliferation.
Intracranial aneurysms (Tomsick et al. 1976) and arteriovenous malformations and aneurysms of cervical arteries have been described in patients with NF1 (Deans et al. 1982, Westacott et al. 1988, Schiewinck & Piepgras 1991, Hoffman et al. 1998). These lesions usually manifest in the fifth or sixth decade of life, and they are more common in female than male patients. The vertebral arteries are most commonly affected.
The generalized mesenchymal dysplasia in NF1 may cause structural weakness of the arterial wall and predispose the patient to aneurysm formation. (Deans et al. 1982). Mechanical factors (chiropractic manipulation, etc.) have also been thought to play a role in the formation of vertebral aneurysms. The primary weakness of the arterial wall in NF1 could lead to the development of an arteriovenous fistula through rupture of a preexisting aneurysm or direct rupture of the artery into adjacent veins.