Clinical manifestations

Presentation and course

The clinical presentations of hemispheric supratentorial low-grade gliomas and neuronal and mixed neuronal-glial tumors are related predominantly to the area of brain from which the astrocytomas arise (86; 39). The most common presentations are nonspecific and nonlocalizing headaches. Seizures are another frequent presenting symptom and are more common in lower-grade lesions than in high-grade gliomas. Any type of seizure may occur, including focal seizures with secondary generalization, complex partial seizures, and generalized seizures. Seizures may precede diagnosis by months to years. With the wider use of epilepsy surgery for patients with intractable epilepsy, an increasing number of children with longstanding seizures and apparently static or slowly progressing cortical (especially temporal and, to a lesser degree, frontal) lesions are found to harbor low-grade glial tumors or mixed neuronal and glial low-grade tumors (gangliogliomas).

Focal neurologic deficits, especially focal motor deficits, are common presenting symptoms in pediatric patients, occurring in up to 40% of children with low-grade tumors. Sensory deficits are less common in pediatrics. Rarely, a frontal lobe lesion may present with apparent ataxia, believed to be due to involvement of fronto-pontine pathways.

Children with tuberous sclerosis are at risk of developing subependymal giant cell astrocytomas. Such lesions are often intraventricular and can cause obstructive hydrocephalus (72). Children with NF1 may present as patients without NF1, but also may have lesions found on “screening” studies when they are apparently asymptomatic.

The differential diagnosis in a child with a supratentorial mass lesion is varied. Abscesses, chronic granulomatous lesions, and cerebrovascular accidents may mimic low-grade neoplasms. Other primary central nervous system tumors that relatively frequently arise in the supratentorial region in childhood include high-grade gliomas, primitive neuroectodermal tumors, ependymomas, choroid plexus carcinomas, and choroid plexus papillomas. As stated previously, low-grade astrocytomas outnumber high-grade neoplasms in childhood by a ratio of at least 2:1, and in some series by a ratio of closer to 4:1. The only way to clearly separate tumor types is by biopsy. However, for specific diagnosis, a relatively large sample of the tumor must be histologically studied. Biopsies or small resections may give misleading information due to sampling error. A variety of immunohistochemical diagnostic techniques, especially proliferation markers, are being evaluated for their specificity and sensitivity in distinguishing between higher-grade tumors and those that will act more benignly. These techniques include staining for the Ki-67 antigen, and an antibody to MIB-1, which is also thought to identify the Ki-67 antigen (22; 87). These proliferation markers have been used to predict outcome for adults and children with low-grade neoplasms. However, they have not been widely incorporated, as of yet, into classification schemas for childhood central nervous system tumors. Other tests, such as PET and SPECT, may suggest the presence of a low-grade neoplasm rather than a high-grade tumor, as low-grade tumors have a metabolism similar to that of surrounding brain.

Prognosis and complications

The overall prognosis for children with low-grade cortical gliomas and neuronal and mixed neuronal-glial tumors is relatively good compared to other forms of childhood central nervous system malignancy. The average 10-year rate of survival for those patients diagnosed before the age of 20 years is over 90% (17; 40). Some histological patterns have been associated with better outcome. Pilocytic astrocytomas, gangliogliomas, and some rarer infantile tumors (desmoplastic infantile gangliogliomas, dysembryonic neuroepithelial tumors, and possibly xanthoastrocytoma) have been associated with better outcome (114; 38; 128). Within the pediatric-type diffuse gliomas, tumors with BRAF fusions seem to have the best prognosis and lowest chance of malignant transformation (29). Postoperative regression has been noted in desmoplastic infantile gangliogliomas (124). As stated previously, children with low-grade gliomas have better prognoses than do adults for long-term survival; however, it is unclear at what age a patient can be considered to have an adult prognosis. This is likely a factor of biology, not age (122). Childhood low-grade supratentorial gliomas rarely mutate into higher-grade lesions, whereas the majority of adult tumors mutate over three to five years after diagnosis (47). However, this might not be true for those tumors with a BRAF v600E mutation (76). Also, co-deletion of CDKN2A suggested poorer prognosis and a tendency to mutate to a higher-grade glioma (19; 76). In a retrospective review of 461 patients with cortical low-grade tumors, the 15-year survival rate was over 80% for patients under 19 years of age, as compared to less than 40% for older patients (77). Other variables associated with increased survival included lack of major preoperative neurologic deficits, longer duration of symptoms prior to surgery, seizures at the presentation of illness, and lack of major postoperative neurologic deficits.

Although some histological types of low-grade gliomas may be related to different rates of survival, the separation of tumors by grading has been variably related to survival. In general, when pilocytic tumors are considered to be grade 1 lesions, survival is better for grade 1 lesions than for grade 2 lesions. On the other hand, when grade 1 and grade 2 lesions have been separated by alternative classification systems using other histological criteria, the relationship between grading and outcome is less clear (33). Hoshino has found an association between proliferative indices, such as bromodeoxyuridine labeling, and outcome; both children and adults who have lower labeling indices have a better prognosis (56; 22). MIB-1 proliferative index has been shown to predict survival in grade 2 astrocytomas (91). Pilomyxoid chiasmatic tumors have been associated with poorer prognosis (70). It is likely the most important factor in prognosis is the molecular alteration of the tumor (29; 85).

Clinical vignette

A seven-year-old boy was referred for evaluation of seizures that had developed over the past year. The seizures initially began as focal seizures involving the left arm, but, before the evaluation, generalized seizures had been noted. There were also episodes of staring.

On clinical examination, the child was alert and oriented. His pupils were equal and reactive, and his discs were flat. Extraocular movements were full, and the remainder of his cranial nerves was normal. Motor and coordination testing was completely within normal limits. Reflexes were two out of four and symmetric, and toes were down-going.

On CT, no specific abnormalities were noted. On MRI, an ill-defined hypodensity on T1-weighted images and an increased intensity on T2-weighted images were seen in the right temporal lobe. There was mild heterogeneous contrast enhancement.

Biological basis

Etiology and pathogenesis

The etiology of supratentorial low-grade gliomas neuronal and mixed neuronal-glial tumors is essentially unknown for the majority. There is a higher incidence of supratentorial low-grade tumors in children with neurofibromatosis type 1; however, this relationship is not as strong as the relationship between hypothalamic low-grade gliomas and neurofibromatosis type 1 (18).

Children with tuberous sclerosis are at risk for development of subependymal giant cell astrocytomas (72; 35). Those children with p53 germ-line mutations (Li-Fraumeni syndrome) and mismatch repair mutations have a greater likelihood of developing low-grade gliomas that transform into higher-grade tumors.

The molecular pathogenesis of pediatric low-grade gliomas and neuronal and mixed neuronal glial tumors is increasingly understood. For many years the relationship between NF1 and the development of low-grade gliomas was recognized, and more recently gangliogliomas have also been associated with NF1 (95). The majority of patients with NF1 and presumed low-grade gliomas do not undergo surgery because many lesions are indolent, being found on screening aminations and are considered “incidental” findings; they are presumed to be pilocytic astrocytomas (95). Those lesions that cause neurologic compromise are more likely to be biopsied. In older children and especially young adults, some NF1-associated low-grade gliomas have been found to harbor additional mutational changes such as CDKN2A/B, or ATRX, or both mutations and tend to act more aggressively (109). Such “anaplastic” piloid astrocytomas are difficult to manage.

Pilocytic astrocytomas are composed of elongated cells with tapering processes. They may be microcystic or have a large single cyst associated with a tumor nodule. Tumor cells in pilocytic astrocytomas tend to form compact parallel bundles, there is often a biphasic pattern where the pilocytic areas are intermixed with more loosely structured microcystic regions (114). A variant of the pilocytic astrocytomas, the pilomyxoid astrocytoma, was described in 1999 as a pediatric tumor with monomorphous piloid features. It was believed to a have poorer prognosis than the classical pilocytic astrocytoma and possibly a greater tendency to disseminate to the nervous system (69). In the 2016 WHO classification pilomyxoid astrocytomas are not considered a separate entity although molecular studies have shown that some have a different gene expression profile; there is a tendency in some cases to mature to more of a classic pilocytic astrocytoma subtype (69).

BRAF genetic alterations associated with increased signaling through the RAS/MAPK pathway occur in the majority of children with non-NF1 associated pilocytic astrocytomas (60). The most common mutation seen is the KIAA1549:BRAF gene fusion. The resultant fusion protein lacks the BRAF regulatory domain, which causes increased signaling through the MAPK pathway. The fusion underlies the majority of cerebellar, and to a lesser extent, diencephalic low-grade gliomas. It is present at a lower frequency in supratentorial low-grade gliomas. BRAFV600E mutations are seen in pilocytic astrocytomas but can also be seen in other forms of low-grade gliomas, especially diffuse astrocytic tumors (102; 60; 07; 113; 61; 30). FGRR1 mutations are the third most common molecular genetic abnormalities seen. FGFR2 mutations are seen less commonly. Multiomic neuropathology increases accuracy in diagnosis (121).

The fibrillary variant of pediatric low-grade astrocytomas is comprised of cells with predominantly hyperchromatic, oval to regular nuclei (114). This tumor type would now be considered under the grouping of pediatric-type diffuse low-grade gliomas. The cytoplasm of these cells can be fairly scant. Protoplasmic and gemistocytic astrocytomas can be seen. Glial fibrillary acidic protein expression occurs in the majority of low-grade supratentorial diffuse astrocytomas. BRAFV600E mutations occur in this subgroup and have been shown in a series to portend a poorer prognosis, as late transformation to a higher grade glioma has been seen (107; 92). Other activating mutations can be seen in diffuse pediatric low-grade gliomas such as FGFR1, PTPN11, and NTRK2 fusion genes (106). The most common alterations reported to date are rearrangements in the MYB family of transcription factors (107).

Other forms of pediatric low-grade gliomas are less common. Angiocentric gliomas can be seen in both children and young adults. MYB gene alterations are present in almost all cases with QKI being the primary fusion partner (05).

Astroblastomas are now defined as a unique entity and are histologically hallmarked by the presence of GFAP positive cells with pseudorosettes (129). They can be diagnosed in childhood and have been noted to have a variety of different molecular underpinnings including MN1 rearrangements, BRAF rearrangements, and even two cases with RELA rearrangements (52). Astroblastomas are a molecularly heterogeneous group and on methylation testing they can cluster with different tumors such as pleomorphic xanthoastrocytomas and ependymomas (87).

IDH mutant low-grade astrocytomas are relatively uncommon in pediatric-aged patients, but have been increasingly recognized (87). Initially they were thought to make up less than 5% of pediatric low-grade gliomas, but this may be an underestimation. Overall, these tumors act more benignly in the pediatric age than they do in adulthood. However, they may transform over time into higher grade tumors.

Pleomorphic xanthoastrocytoma is a complex tumor composed of a mixture of histologically variable tumor cells including fibrillary astrocytomas and giant multinucleated forms (114). The multinucleated cells typically contain lipid vacuoles. Despite this pleomorphic appearance, some pleomorphic xanthoastrocytomas tend to have a good prognosis. The most common molecular underpinning of the pleomorphic xanthoastrocytomas are V600E mutations (87; 88).

The dysembryoplastic neuroepithelial tumor also is most likely to present with seizures and is characterized by the presence of columns of oligodendroglial-like cells, ganglion cells, and astrocytes. The most common molecular genetic abnormalities in dysembryoplastic neuroepithelial tumors are FGFR1 mutations including FGFR1 activating point mutations and activating gene fusions (110).

Desmoplastic infantile astrocytomas (DIA) and desmoplastic infantile gangliogliomas tend to present early in life. They have a characteristic imaging appearance in which a contrast enhancing solid nodule occurs associated with a large cystic component (126; 10). Often the large cystic component actually lies medial to the tumor nodule. The separation between desmoplastic infantile astrocytomas and desmoplastic infantile gangliogliomas is often arbitrary and on methylation analysis these tumors tend to cluster together (87). Once again, a variety of different molecularly genetic findings can be seen including BRAF mutations, such as the BRAFV600E mutation, but also BRAFV600D mutations (45; 11).

Rosette-forming glial neuronal tumors are rare and usually occur deeper in brain, including the diencephalic region. FGFR1 mutations seem to characterize the vast majority of these tumors (118).

Diffuse leptomeningeal glioneuronal tumor (DLGNT) is an increasingly recognized tumor of childhood characterized by diffuse leptomeningeal enhancement of the brain and spine. Previously identified under different nomenclatures, including diffuse leptomeningeal oligodendroglial tumors, disseminated pilocytic astrocytomas, and anaplastic astrocytomas, this tumor is separable on methylation into different subtypes (23). The relatively distinct subtypes have somewhat different prognoses. One subtype is characterized by deletions of 1p/19q, but without mutations in IDH1 and IDH2. A second subset only has abnormality in chromosome 1p. KIAA1549:BRAF fusion was thought to be ubiquitous within this tumor subgroup, but now is recognized to only occur in a subset of tumors. Other fusions have also been noted (23; 118; 15).

In the future it is unclear whether these histologically identified different subgroups will continue to exist or whether molecular classification, primarily basing classification on the molecular subtype of the tumor rather than its histology appearance, will be the main driver underlying classification.

Epidemiology

Because of the previously stated problems in determining which tumors should be considered low-grade supratentorial gliomas and the tendency for some series to include, and others to exclude, thalamic and other diencephalic glial tumors in reviews of supratentorial gliomas and neuronal and mixed neuronal-glial tumors, the exact incidence of these tumors is hard to discern (32). The incidence of childhood astrocytomas may be increasing, especially in girls (54). Approximately 25% to 30% of all childhood primary central nervous neoplasms arise in the cerebral hemispheres. In infants, this frequency is higher, as nearly 40% of tumors arise in the cerebral cortex. Of the cerebral tumors occurring in children outside the period of infancy, 40% to 50% are low-grade astrocytomas, occurring at peak incidence between the ages of eight and 12 years. Low-grade cerebral astrocytomas outnumber higher-grade glial tumors by a 3:1 or greater ratio. The reported incidence of low-grade glial cerebral tumors is changing, in part due to the wider use of epilepsy surgery. Some children with intractable epilepsy, who are believed to harbor static lesions, are found at surgery to have low-grade glial tumors, especially gangliogliomas. In some series, gangliogliomas have made up as high as 10% of all cortical gliomas diagnosed in children less than 15 years of age. The desmoplastic infantile ganglioglioma and the dysembryonic neuroepithelial tumor are lesions diagnosed predominantly in infancy, and have been increasingly reported in the past decade; as have other mixed low-grade neuronal-glial tumors. Oligodendrogliomas tend to be diagnosed later in childhood and make up less than 5% of childhood cortical tumors.

Prevention

There are no known means to prevent the development of low-grade cortical tumors. There is an association between low-grade cortical tumors and prior cranial radiation.

Diagnostic workup

Both CT and MRI are sensitive in the detection of cortical low-grade gliomas and neuronal and mixed neuronal-glial tumors (28). Low-grade glial tumors are predominantly hypodense on CT. On MRI, they are hypointense on T1-weighted sequences and hyperintense on T2-weighted sequences. Enhancement following administration of contrast agents is variable. These tumors tend to be infiltrative, but at times, on imaging, seem relatively well marginated. This is especially true for gangliogliomas, as they tend to show little or no contrast enhancement and have high T2-weighted signal abnormalities on MRI, usually without associated cystic components. The pilomyxoid variant of pilocytic astrocytoma has a dominant solid component, often with a nonenhancing component, and seems to be more frequently disseminated (78). Pleomorphic xanthoastrocytomas classically have an enhancing solid nodule that is cortically based, with apparent meningeal involvement and a deeper nonenhancing cystic component. Fibrillary low-grade astrocytomas have a more variable neuroradiographic picture, and their presence is suggested by a limited amount of associated edema and a modest amount of mass effect, as compared to overall tumor size. Oligodendrogliomas are usually hypodense or of mixed density on MR. CT can often be helpful in the diagnosis of an oligodendroglioma because the majority of these tumors show some degree of calcification.

EEG, demonstrating focal areas of electroencephalographic slowing, can be useful in suggesting the presence of an underlying neoplasm, but is of much less value in exact diagnosis and tumor localization than is CT or MRI. Intraoperative electrocorticography is often used in patients with seizures to demonstrate the extent of the epileptogenic foci, but is not terribly useful in localizing the site of tumor. The use of electrocorticography may be helpful in surgical planning, especially as regards resection of tissue around the tumor (epileptogenic foci) to control seizures. PET and SPECT have been more widely used in the evaluation of adult cortical tumors. PET is usually performed with various agents that reflect brain metabolism; the presence of a high metabolic rate in a tumor, as compared to the metabolic activity of the surrounding brain, suggests that the tumor may be relatively malignant rather than a low-grade astrocytoma. Similarly, SPECT scanning with agents that also may reflect tumor metabolism, has been used to separate high-grade from low-grade neoplasms. Neither of these techniques has been widely used in pediatrics and, at this point, cannot be used to reliably separate high-grade from low-grade tumors. There is some evidence that pilocytic astrocytomas, although low-grade, will often show increased metabolism on both PET and SPECT. In addition, neither PET nor SPECT can separate low-grade gliomas from other low-grade neoplasms, radiation necrosis, or other processes in the brain (eg, infections, strokes) that can also mimic low-grade hemispheric astrocytomas.

Management

At present, the management of childhood cortical low-grade tumors and neuronal and mixed neuronal-glial tumors remains unsettled. Most observers suggest that, when possible, the treatment of choice is complete resection (34). After complete or near-complete resection, especially in pilocytic astrocytomas, most investigators recommend no further antitumor therapy. A complete total resection in cortical low-grade tumors results in survival rates greater than 90% after more than five years from diagnosis (58; 55; 53; 104; 128; 40). Survival of patients with subtotal resections is somewhat less favorable (40). Event-free survival is approximately 90% at five years, but closer to 50% at 10 years, and is considerably lower at 10 years in the 10% to 20% range in those who have had partial resections or biopsies (40). Gross total resections are possible in 40% to 80% of hemispheric tumors, but in a lesser number of patients with diencephalic or deep-seated tumors. The most effective surgical management of children with intractable seizures and low-grade gliomas remains unsettled, as it is unclear whether resection of the lesion plus the ictal focus results in better disease control or seizure control than total resection of the lesion alone (08; 100; 103).

The benefits of radiation therapy for children with low-grade cortical gliomas are unclear. The majority of studies have shown longer survival in children with subtotally resected tumors who have received radiation as compared to those treated with surgery alone (81; 117; 104), and is essentially unknown for neuronal and mixed neuronal-glial tumors. There have been attempts to develop prospective randomized studies for patients with subtotally resected tumors; in these studies, patients, after subtotal resection, were to be either randomized to observation alone or immediate postresection radiotherapy. However, the accruals on some of these studies were so poor that they had to be halted. Until such studies are completed, the benefits of immediate postoperative radiotherapy for children with partially resected low-grade tumors must be considered to be unproven. In general, those studies that have been completed have suggested that radiation therapy may prolong progression-free survival but that overall survival 10 and 20 years after treatment with resection followed by radiation is similar to the overall survival of patients treated with surgery alone. In a trial, five-year progression-free survival was 65% after radiotherapy and plateaued at 62% at 10 years (40). External fractionated radiotherapy in a study resulted in a 70% progression-free survival rate and a 97% overall survival rate at 10 years (93). Proton beam irradiation for cortical low-grade astrocytomas can, in some situations, spare irradiation to the temporal lobe, hippocampus, and hypothalamic-pituitary axis, resulting in theoretically less sequelae (44).

There is no evidence that whole-brain radiation therapy has any benefit over local radiation therapy in children with low-grade tumors. Whole-brain radiotherapy certainly carries a higher risk of neurocognitive sequelae than does local radiotherapy. Studies are being performed using more focused radiation therapy in children with residual disease after surgery (119). These studies have employed interstitial radiotherapy (71; 123) and, more recently, focused external beam radiation therapy. Although the early results with these techniques are encouraging, long-term disease control and safety, especially in children, remains unknown.

Chemotherapy has been used for children with low-grade gliomas over the past few years. Various chemotherapeutic regimens have been used, primarily in children with deep-seated (diencephalic tumors) not amenable to gross total resections. Drugs and drug combinations that may be of some utility include carboplatinum, cyclophosphamide, actinomycin and vincristine, carboplatinum and vincristine, VP 16, and a 5-drug regimen containing nitrosourea (99; 97; 79; 101; 37; 105; 80). The combination of carboplatinum and vincristine has been noted to result in a nearly 60% objective tumor shrinkage rate and a 90% to 95% initial tumor control rate in children with newly diagnosed, partially resected, low-grade tumors, including cortical low-grade gliomas (97; 94). In the majority of these studies, although chemotherapy may be of initial benefit, 50% of patients will progress within three years of the initiation of chemotherapy. For this reason and the concern that radiation therapy is more harmful in younger patients, chemotherapeutic regimens have been used predominantly in younger patients, especially in children less than five years of age. Studies are presently underway evaluating multiple different types of chemotherapy in supratentorial low-grade gliomas to determine long-term efficacy and safety. In a randomized, prospective study of 274 children without neurofibromatosis type 1 and with low-grade gliomas, the combination of thioguanine, procarbazine, lomustine, and vincristine (TPCV) was found to be at least equally as effective as carboplatin and vincristine. However, the relative toxicities of the regimens may result in choice of one treatment approach over the other. Younger age, greater tumor size, and thalamic involvement predicted poorer outcome (03). Children with neurofibromatosis type 1, treated only with carboplatin and vincristine, had a better prognosis (02). The addition of etoposide to carboplatin and vincristine did not improve overall or progression-free survival (41). Temozolomide has been studied, and results have been variable, with some studies suggesting a low rate of objective response and others suggesting a higher rate of tumor shrinkage (46; 65).The combination of carboplatin, vincristine, and temozolomide has also been used with good efficacy (16). The combination of cisplatin and etoposide has been used in four patients (90). Weekly vinblastine has also been shown to be active (73; 14; 75).

Avastin with or without irinotecan can also be effective, and in certain situations has been associated with improved clinical outcomes (96; 57; 62; 43).

Another approach actively utilized is molecular-targeted therapy blocking growth pathways believed active in tumors (51), such as the Ras-MAP Kinase pathway (MEK inhibitors and v600e inhibitors) and mTOR in both patients with and without neurofibromatosis type 1 (20; 21; 49; 48; 06; 98). The MEK inhibitors, in those with neurofibromatosis 1 and BRAF mutations, have likely shown the best efficacy (06; 31). In a phase 2 study of Selumetinib, a MEK-inhibitor, 25 children without NF1 with recurrent low-grade gliomas after treatment with carboplatin containing chemotherapy, nine (36%) achieved a radiographic partial response. The majority of the remaining either had less marked tumor shrinkage or stable disease and the two-year progression-free survival was 70±9%. All had tumors either with BRAF fusions or V600E mutations and responses were seen in both molecular tumor types. A separate arm of the study evaluating only those with visual pathway tumors is still being analyzed (31). These results have led to a prospective phase 3 randomized trial in children with newly diagnosed, non-totally resected, low-grade gliomas, including those of the visual pathway, comparing treatment with selumetinib to the combination of carboplatin and vincristine.

In the same phase 2 study of selumetinib for recurrent low-grade gliomas, a cohort of 25 children with NF1 and progressive low-grade gliomas were studied and an even better response rate and two-year progression-free survival was seen (96±4%), also leading to a newly opened, similar randomized prospective study comparing the same agents. Other MEK inhibitors are in active phase 2 study for progressive low-grade gliomas (13; 111).

For those children with recurrent V600E BRAF mutated low-grade gliomas and low-grade mixed neuronal-glial tumors, responses have also been seen after treatment with V600E BRAF inhibitors (66; 74), and a prospective randomized trial has recently completed accrual comparing the combination of dabrafenib, a BRAF-inhibitor and trametinib, a MEK-inhibitor to carboplatin and vincristine chemotherapy. The results in children with BRAF mutated low-grade gliomas and glioneuronal tumors demonstrated the superiority of the combination to chemotherapy in both response and progression-free survival. Patients who received dual molecular-targeted therapy also had better overall quality-of-life scores than those receiving chemotherapy (12). In still another study using a second generation BRAF-inhibitor, tovarafenib, a high-response rate was seen even in children already treated with and progressing after receiving MEK-inhibitors or first generation BRAF V600E inhibitors (67). The use of one V600E BRAF inhibitor, sorafenib, resulted in paradoxic tumor activation in BRAF-fusion low-grade gliomas, and such treatment is contraindicated (63). The toxicity profiles of BRAF-inhibitors and MEK inhibitors are different from most chemotherapies, and rash is a frequent side-effect. The long-term sequelae of such biological therapies are not well characterized. For children with symptomatic subependymal astrocytomas and tuberous sclerosis, agents that inhibit mTOR have been shown to cause significant reductions in tumor size, eliminating or delaying the need for surgery (72; 35).

The surgical management of children with recurrent or intractable seizure and known low-grade lesions remains somewhat debatable (42; 09). Removal of the lesion itself may result in excellent seizure control. However, there are some retrospective studies using historical controls that suggest that resection of the lesion plus the epileptogenic focus, as determined by electrocorticography, is more effective in long-term epilepsy control than is the removal of the lesion alone. Preliminary evidence also suggests that surgical removal, using electrocorticography, may also allow for safer, more extensive resections.

Outcomes

The location of the tumor, extent of surgical resection, and employment of radiation therapy have all been related to tumor outcome, but in a highly interrelated fashion. Bithalamic involvement predicts poor outcome (108; 03). Some studies have found that gross surgical resection is related to improved survival, especially in younger patients. Laws and colleagues suggested that a point scale, based on a personality change, the level of consciousness at the time of diagnosis, the extent of resection, tumor site, and postoperative status could predict long-term (15-year) survival (77). Using this type of point system, "good-risk" patients had greater than a 50% predicted 15-year survival as compared to 16% for those patients with "average" or "poor-risk" disease. This point system has not been widely employed by others. In an attempt to unravel which clinical factors independently affect outcome, multicentered studies have been completed. In 518 patients evaluated prospectively by the Children’s Oncology Group, gross total resection without residual disease was the predominant predictor of progression survival; on univariate analysis, juvenile pilocytic astrocytoma and gangllioma histology also were associated with better outcomes (128). Similar findings were found in a large European series (40). A European international study of 1044 children found that chiasmatic/postchiasmatic location was a predictor of poor disease control (26). In another international study of 639 children, young age, fibrillary astrocytoma, and subtotal resection were associated with less favorable disease control (120). Higher MIB-1 labeling index has been related to a poorer prognosis (89). However, it is conceivable that all of these clinical and histologic features will not be as important as molecular genetic findings BRAF-KIAA 1549 fusion abnormalities have been associated with a favorable prognosis, even in total resected pilocytic tumors (50), whereas BRAF v600E mutations associated with CDKN2A deletions and chromosome 7 gain have been related to poorer outcomes (92; 112; 76).

Long-term survivors may also have significant cognitive, social, and behavioral deficits (01). Surgical treatment of tumor-related temporal lobe epilepsy results in excellent seizure control in patients with totally removed tumors. Poorer seizure control occurs in patients with more than 10 years of epilepsy prior to removal, incomplete seizure control, or a remote electroencephalographic focus (103).

Special considerations

Anesthesia

No information is available, except for the precautions taken for patients with possible increased intracranial pressure.