Clinical manifestations

Presentation and course

Medulloblastoma, arising most commonly from the area of the cerebellar vermis, causes neurologic symptoms by filling the fourth ventricle and blocking cerebrospinal fluid egress or infiltrating cerebellar tissue. Approximately one third of patients have evidence of leptomeningeal spread at the time of diagnosis (16; 02). The time from onset of symptoms to diagnosis is usually 3 months or less (86). Occasionally, the tumor presents apoplectically, due to acute hemorrhage into the tumor with resultant obtundation, followed by coma.

The most common early symptoms of medulloblastoma are the nonspecific and nonlocalizing findings of increased intracranial pressure due to obstruction of the fourth ventricle. Papilledema, headaches, vomiting (especially morning vomiting), and lethargy are present in up to 90% of patients by the time of diagnosis (86). Ataxia is usually present early in the disease and results in truncal unsteadiness, which is later followed by appendicular cerebellar symptoms. Diplopia may be either secondary to pressure-related sixth nerve paresis or due to ocular motor dysfunction caused by direct infiltration of the tumor into the brainstem. Hyperreflexia secondary to ventricular dilatation is common, but corticospinal tract dysfunction is relatively infrequent. The tumor may invade the overlying meninges, and head tilt can be caused by herniation of the cerebellar tonsils or by meningeal irritation. Trochlear dysfunction with resultant head tilt is less common. In adults, medulloblastomas are more likely to arise laterally and cause seventh and eighth nerve dysfunction early in the course of illness.

As the tumor may affect extremely young children with unfused sutures, initial symptoms in children with congenital tumors may be relatively nonspecific and include anorexia, failure to thrive, and increased irritability. Macrocephaly, spread cranial sutures with an open and bulging fontanelle, and paralysis of upgaze (the "setting-sun" sign) are characteristic presentations in extremely young patients.

Despite the relatively frequent rate of leptomeningeal dissemination at the time of diagnosis, most patients with disseminated disease at the time of diagnosis do not have specific symptoms referable to leptomeningeal deposits. Occasionally, patients have back pain or radicular symptomatology due to tumor spread. Other symptoms of leptomeningeal spread, such as seizures, focal neurologic deficits, motor weakness, or spinal cord compression, are rare.

Prognosis and complications

Overall 5-year, disease-free survival for children with medulloblastoma ranges between 60% and 80% (14; 69). There is evidence that patients can be broadly separated into two prognostic subgroups (72; 27). Factors that have been found to be of prognostic importance include age at diagnosis, extent of surgical resection, local extent of tumor diagnosis, and extent of dissemination at diagnosis. Patients with so-called average-risk disease include those who are older (possibly greater than 7 years of age at time of diagnosis), those who have totally resected tumors that have not invaded the brainstem, and those who have no evidence of leptomeningeal dissemination at the time of diagnosis. Patients in the higher or less favorable risk group include those with disseminated tumor at the time of diagnosis, and possibly those who have had partially resected tumors or tumors that have invaded the brainstem at diagnosis (73; 02; 27; 01). A variety of molecular factors have been related to prognosis; however, they have been superseded by more expansive genomic analysis, including RNA expression, DNA methylation profiles, and DNA sequencing on both fresh frozen and formalin fixed tissue (28; 31; 47; 46; 79; 80). Such transcription findings can be combined with clinical parameters to create even more robust predictors of outcome (68; 37). Biological parameters such as nuclear beta-catenin staining and monosomy 6 (evidence of aberrant WNT signaling) are predictive of better outcome and elevated MYCC expression and amplification have been shown to be predictive of poorer outcome (64; 25; 67; 43; 68; 83; 105). TP53 mutation, which occurs primarily in SHH-driven and WNT-driven tumors, and connotes a poorer prognosis in SHH-driven tumors (120). Adding to the complexity is that even early childhood SHH medulloblastomas can be prognostically stratified (113), and possibly also the WNT-driven tumors (36).

A plethora of papers have been published over the past decade that have further divided medulloblastoma into multiple subgroups, often with differing prognosis (67; 46; 68; 48; 97). Even within the WNT subgroup, which as noted above is the group that has the best overall prognosis, molecularly-defined subgroups of patients have been noted in adults that carry less than a 100% survival rate (48). Four different molecular subgroups are now identified within the SHH grouping and the presence or absence of TP 53 mutation continues to be an important predictive factor, with wild type TP 53 tumors having a better prognosis. The eight subgroups, which have now been delineated predominantly by methylation profiling in the non-WNT/non-SHH pathway activated tumors have been shown to have different prognosis (97). Tumors within this subgroup that have MYC amplification and possible gain have a poorer prognosis, with less than 50% of patients surviving five years from diagnosis. In contradistinction, those tumors with specific chromosome losses, such as loss of chromosome 8 and 11, or chromosome gains (gain of chromosome 7), have an excellent prognosis. The presence of an isochromosome 17q, found predominately within the non-WNT/non-SHH activated tumors, has been associated with a poor prognosis in some studies, but not in all.

Clinical vignette

A 5-year-old boy developed unsteadiness associated with headaches and nausea 10 days before evaluation. In retrospect, the child had been complaining of mild headaches for six weeks. During the 24 hours before assessment, the child seemed sleepy and fell when he tried to walk.

On examination, the child was alert but irritable. He reported that he had a severe headache and vomited that morning on awakening. He had bilaterally reactive pupils but bilateral papilledema. Visual acuity and fields seemed to be intact. His face moved symmetrically, and the remainder of his cranial nerves was normal. When the child sat, there was marked truncal unsteadiness, and he reached with bilateral dysmetria. When he attempted to walk, he fell quickly to either side. Reflexes were diffusely increased.

Neuroimaging studies showed a large, somewhat hyperintense mass filling the fourth ventricle and causing marked hydrocephalus. The mass homogeneously enhanced. On MRI, the mass was decreased signal intensity on T1-weighted images and isointense on T2-weighted images. The mass seemed to be plastered to the back of the brain stem.

Biological basis

Etiology and pathogenesis

The etiology of medulloblastoma for the majority of children and adults remains unknown, but at least 10% and likely more can be linked to specific germ-line mutations. Inherited conditions that have been related to an increased risk of medulloblastoma include neurofibromatosis, Turcot syndrome, the nevoid basal cell carcinoma syndrome (Gorlin syndrome), and Li-Fraumeni syndrome. Identification of multiple cancer predisposition syndromes has raised concerns that many more patients with medulloblastoma may have germ-line mutations predisposing to development of medulloblastoma (119). As an example, heterozygous germline mutations in the G proteins-coupled receptor 161 (gpr161) were identified in approximately 3% of cases of SHH-activated medulloblastoma (09).

The likelihood of developing treatment-associated secondary tumors is a real issue for those with germ-line mutations, and this has led to a re-examination of treatment approaches, especially in those with SHH-driven tumors arising in patients who harbor an underlying P53 mutant germ line mutation (99; 119).

In the largest review to date, germ-line mutations possibly predisposing to tumor development were identified in 11% of 673 patients with medulloblastoma (116). Genes identified included APC, BRCA2, PALB2, PTCH1, SUFU, and TP53. Genetic predispositions varied in prevalence among medulloblastoma molecular subgroups and was highest for the sonic hedgehog subgroup, occurring in 20%; SUFU and PTCH1 accounted for the majority of predisposition in this subgroup. APC mutations were seen in the WNT subgroup.

Intra-utero irradiation has been related to an increased incidence of tumor development. Other environmental factors, such as drugs, nitroso-compounds, and electromagnetic fields, have been linked in some studies to the development of medulloblastoma, but data to support such associations are preliminary and equivocal. A history of measles immunization and SV40 exposure has been linked to development of medulloblastoma in one study (60; 87; 15).

There is extensive work to demonstrate that medulloblastoma, although made up predominantly of undifferentiated small round cells, frequently displays foci within the tumor of apparent differentiation along identifiable cellular lines. Neuronal and glial differentiation can be frequently identified utilizing immunohistochemical markers. A specific tumor marker for medulloblastoma still does not exist, although the majority of tumors will show synaptophysin expression. Histologic subvarieties of medulloblastomas have been identified, including the large cell variant and tumors with extensive nodularity, which may carry different prognosis (34; 23). In older children, the large-cell or anaplastic variant has been related to a poorer outcome (24). In younger children, especially infants, the desmoplastic or nodular histologic variant has been associated with better outcome (34).

Detailed genetic investigations have disclosed reproducible genetic abnormalities in medulloblastoma, and these findings have been critical in improving understanding of the disease and suggesting potential means of more precise, effective treatment (81; 102; 78; 37). Amplification and overexpression of MYCC and MYCN has been associated with poorer survival (31).

Based on molecular genetic findings, the four main subgroups of medulloblastoma have been identified, and molecular subclassification provides further understandings concerning cellular origin, biological mechanisms, underlying tumor development, and prognosis. These subgroups are WNT, sonic hedgehog (SHH), group 3, and group 4 (sometimes considered as one subgroup, non-WNT/non-SHH) (24; 31; 92; 110; 47; 79; 25; 67; 103; 43; 46; 68; 94; 105; 85; 120; 45; 89; 84; 100; 95; 11; 37; 40; 97). TP53 and OTX2 have been related to poorer outcome (36).

The WNT subgroup comprises approximately 10% of all medulloblastomas and is characterized by a WNT signaling gene expression signature and beta-catenin nuclear staining. The developmental origin of the WNT subset, which occurs predominantly in older children, adolescents, and young adults, is believed to be from the embryonal rhombic lip region (25; 11; 97). These tumors are usually classical medulloblastoma by histology, but they may infrequently be large cell or anaplastic variants. They are usually not disseminated at time of diagnosis. The WNT tumors most commonly have CTNNB1 mutations, and in the 5% to 10% that do not, the remaining majority has APC mutations. Such APC mutations are usually associated with a germline APC mutation, and children may have stigmata of Turcot syndrome. The WNT medulloblastomas characteristically show 6q loss (monosomy 6), but this may not present in patients older than 18 years of age with WNT tumors. This is the subtype of medulloblastoma with the best prognosis. In those less than 21 years of age, 90% or greater of patients are cured after appropriate treatment (79; 11; 66; 97).

The sonic hedgehog (SHH) variants of medulloblastoma, in total, make up approximately 25% of all medulloblastomas (25; 95; 90). Embryologically, these tumors are believed to emanate from the external granular layer of the cerebellum. Genetically, they are characterized by chromosome 9 deletions, and mutations may occur at multiple sites along the SHH pathway, including PTCH1, PTCH2, SMO, SUFU, and GLI2. Germline mutations occur in the SHH-subgroup including mutations in GPR161 in very young children and Li-Fraumeni syndrome (90; 09). Sonic hedgehog medulloblastomas tend to be bimodal and occur in infants and younger children (those less than 3 years of age), and then in adulthood. However, they may arise in children over 3 years of age, and children in the older childhood subgroup have the worst prognosis.

The most common subsets of SHH-driven medulloblastoma arise in children younger than 3 years of age and are SHH-β and SHH-γ. The SHH-β form is more frequently metastatic. The SHH-γ form, which has the best prognosis, is histologically characterized by extensive nodularity. In SHH-β, SMO mutations tend to be associated with a favorable prognosis, whereas SUFU mutations have a less favorable one. PTCH1 mutations can be seen both in subgroup SHH-β and SHH-γ.

In older children, subtype SHH-α is more likely to occur and is hallmarked by MYCN and GLI2 amplifications. SHH-α medulloblastomas have also frequently associated tp53 mutations and carry a poor prognosis, as 40% or less of children survive after treatment. The fourth SHH subtype termed SHH-δ underlies the majority of adult cases, and in this subgroup, both PTCH1 and SMO mutations can occur.

The non-WNT/non-SHH activated subtype of medulloblastoma, which includes previously named group 3 and group 4 in total, makes up the majority of cases of medulloblastoma. Group 3 medulloblastomas comprises approximately 25% of all medulloblastomas cases, whereas group 4 medulloblastomas makes up approximately 40%. The origins of non WNT/non SHH were unknown until the recognition that atypical cells that reside in the persistent rhombic lip-PeRLs act as a premalignant process and can transform into non WNT/non SHH medulloblastomas (39). There is a male predominance for non-WNT/non-SHH activated tumors. Non-WNT/non-SHH activated tumors with MYC-amplification are a well-recognized subset (11; 97). Within the non-WNT/non-SHH subgroup, there have been further subdivisions identified in addition to those with MYC amplification.

As noted previously, MYC-amplification is frequently present in group 3 medulloblastomas, and when present it connotes poor prognosis. Isochromosome (i17q) is seen in non-WNT/non-SHH medulloblastomas. Although different classification systems have suggested somewhat differing subdivisions, non-WNT/non-SHH tumors have been recognized to be comprised of eight different molecular subgroups by methylation array testing (97). These subgroups are identified by activation of various genes, including tandem duplication of SNCAIP gene with resultant activation of PRDM6 by enhancer high-jacking, MCYN amplification, and GFI1 or GFI1B overexpression. Some of the subgroups have been fairly definitively shown to have specific prognosis, for example, chromosome 11 loss has been associated with an excellent prognosis, whereas other subgroups have been shown to connote not only a poorer prognosis, but also a tendency for late relapse.

Table 2. Simplified Biological Stratification of Medulloblastoma (105; 97)

Epidemiology

Medulloblastoma occurs throughout the world, and there is no evidence that the tumor is more likely to occur in any ethnic or racial subpopulation (53). The tumor occurs in slightly higher incidence in males than females (a ratio of 1.3 to 1). However, as noted previously, some biological subgroups do have gender differences. The tumor tends to occur in the first decade of life with a bimodal distribution, peaking at 3 to 4 years of age and again later at 8 to 10 years of age. However, the lesion may occur at any time, including adulthood.

The tumor is the most common malignant primary central nervous system neoplasm occurring in childhood. It makes up approximately 20% of all childhood brain tumors and 30% to 40% of all posterior fossa childhood brain tumors. Up to 10% of children harbor germ-line mutations, which likely predispose to tumor development (116).

Prevention

There are no known means to prevent the development of medulloblastoma. Risk factors include the presence of various genetic diseases, such as neurofibromatosis, Turcot syndrome, and the nevoid basal cell carcinoma syndrome.

Differential diagnosis

The primary clinical distinction is the separation of medulloblastoma from other tumors that may arise in the posterior fossa in childhood, including cerebellar astrocytomas, brainstem gliomas, and ependymomas. Rarely, other conditions may be confused with medulloblastoma, including postviral cerebellar ataxia (postviral cerebellitis), drug ingestion, or a posterior fossa abscess. Cerebellar astrocytoma is equally as common as medulloblastoma in the posterior fossa. This tumor tends to occur at a somewhat later age, peaking between 5 and 15 years of age, and more commonly presents with lateralized cerebellar deficits early in the course of illness. The tumor then grows towards the midline and obstructs cerebrospinal fluid flow. Ependymomas may mimic medulloblastomas, but tend to have a somewhat longer, more insidious course before diagnosis. Ependymomas are also more likely to result in cranial nerve deficits. Brainstem gliomas are more likely to present with multiple cranial nerve deficits, unsteadiness, and long tract signs. Approximately one third of brainstem gliomas will cause obstructive hydrocephalus by the time of diagnosis. Neuroradiographic studies can usually separate medulloblastomas from other posterior fossa tumors, but surgery is indicated in all cases of presumed medulloblastoma, both to confirm diagnosis and for treatment.

Diagnostic workup

If a posterior fossa tumor is thought likely, initial diagnostic workup usually consists only of either a CT or MRI, performed with and without contrast enhancement (121). Both procedures are 95% to 100% reliable in demonstrating the presence of the tumor. On CT scans, medulloblastomas are usually isodense or hyperdense masses that enhance readily after contrast. Although the enhancement is usually homogeneous, heterogeneous enhancement may be present. There may be associated hemorrhage, calcifications, and, less frequently, cystic changes within the tumor. Associated hydrocephalus occurs in over 80% of cases. On MRI, medulloblastoma is usually well visualized. The tumor tends to be of decreased signal intensity on T1-weighted images and hypo- or iso-intense on T2-weighted images.

Tumor extent, especially infiltration into the brainstem and extension through the foramen magnum, is appreciated better on MRI than on CT. As in the case of CT, the post-contrast scan usually shows extensive homogeneous or, less commonly, heterogeneous enhancement. Increased tumor vascularity is often present. Results with MRI spectroscopy demonstrate that the tumor usually has biochemical changes similar to those described in patients with malignant gliomas.

In stable patients, evaluation for extent of dissemination can be useful prior to surgery. Spinal MRI, with and without gadolinium, demonstrates evidence of leptomeningeal disease in 20% to 50% of patients.

Following histological confirmation, extent of disease evaluation is indicated in all patients to help in treatment planning (72). MRI of the entire neuro-axis, preferably performed prior to surgery to avoid postoperative artifacts, is optimal to access for disease dissemination (69). Cerebrospinal fluid cytological examination is also indicated, as it provides complimentary evidence for tumor dissemination (30). The utility of bone scans and bone marrow examinations in the evaluation of extent of disease remains unproven and is now rarely done. If there are roles for such examinations, they should probably be limited to children younger than 3 years of age who are widely disseminated throughout the neuroaxis at diagnosis and may be at highest risk for systemic dissemination.

Molecular subclassification by a variety of different genetic techniques are now required for optimal medulloblastoma subclassification and disease risk stratification (80; 98; 58). Such molecular findings are increasingly used to guide therapy.

Management

At present, management for children with medulloblastoma is based on postoperative disease stratification and increasing molecular subgrouping. Surgery is the initial step in treatment, and gross total resection is possible in over 75% of patients. Surgery may be complicated by direct brain stem or cerebellar injury, post-operative infection or post-operative cerebellar mutism syndrome. The latter is the delayed onset of mutism, a few hours after surgery, and emotional liability, often associated with hypotonia, cerebellar dysfunction, and supranuclear palsies. Thought to be related to dentate thalamocortical damage, due partially to cerebellar vermian damage, it occurs in nearly 25% of children and results in permanent sequelae in approximately one half of those affected (88). Most studies report more favorable outcome after extensive resections, and patients whose tumors are only biopsied rarely survive. Although gross total resection has been related to improved outcomes, this association is attenuated by molecular subgroup affiliation (109). There is little evidence that overall survival is better after gross total resection as compared to “near-total” resection. Surgery alone is never adequate treatment.

Stratification into risk groups, following surgery, is the next critical component of management. The coupling of clinical staging information with molecular results is quickly becoming the standard for risk stratification children with medulloblastoma (31; 25; 46; 66). As noted previously, clinical staging includes evaluation for extent of disease by both imaging of the entire neuroaxis by MRI and lumbar cerebrospinal fluid cytological examination. MRIs of the spine can be difficult to interpret following surgery due to blood clots and other postoperative changes, and for these reasons, MRI of the entire neuroaxis is best performed prior to surgery. In addition, assessment of the amount of residual disease present after surgery and the presence of extensive anaplasia on microscopy are factors that are usually included in clinical risk stratification. Conventionally, patients have been separated into those with average-risk disease, which is defined as total or near total resection (less than 1.5 cm2 of residual disease following surgery), nonanaplastic histology, and no evidence of tumor dissemination. Those with high or poor-risk disease have potentially detrimental findings including greater than 1.5 cm2 residual disease, disseminated disease either seen on MRI scan or confirmed by positive cerebrospinal fluid cytological examination for tumor cells, or anaplastic histology (greater than 50% of the tumor is anaplastic). Molecular findings that would result in patients being considered to be in the higher risk disease categorization are somewhat influx, although there is consensus that non-WNT/non-SHH activated medulloblastomas with MYC amplification connotes poor risk, as well as those SHH-activated tumors, which harbor concomitant p53 mutation. Other molecular findings have not yet been routinely included in risk stratification determination.

The concept of very good risk patients, mainly those with WNT-activated tumors without tumor dissemination at the time of diagnosis and those patients with group 4 tumors with chromosome 11 loss, has been accepted (11; 76; 97). Conversely, patients with widely disseminated disease, SHH-activated tumors with concomitant p53 mutation, or MYC-amplified group 3 tumors are considered very high risk (11; 97). Such combined clinical and molecular characterizations are the backbones of ongoing studies, as well as newly developed prospective clinical trials.

The backbone of postsurgical treatment is craniospinal radiation therapy with supplemental local boost radiotherapy (72; 41; 27). Conventional treatment consists of 3600 cGy of craniospinal radiotherapy and local boost therapy up to a total posterior fossa dose of 5580 cGy. There is no clear evidence that hyperfractionated radiation therapy improves outcome (82). For children with average-risk disease, craniospinal radiotherapy may be sufficient for disease control; the best reported long-term survival rates range between 55% and 70% at five years. In addition, the doses of radiotherapy required for disease control may cause significant sequelae. In a randomized trial, a reduction of the craniospinal radiotherapy from 3600 to 2340 cGy in patients with localized disease at diagnosis was found to be ineffective, as a higher rate of leptomeningeal disease relapse and a lower rate of overall disease-free survival occurred in patients receiving the lower dose of radiotherapy (17). However, an update of this study showed the differences in survival between the two arms to have narrowed (108). Also, in a prospective randomized study performed by the International Society of Pediatric Oncology, no difference was found in survival of nondisseminated patients treated with reduced-dose radiotherapy as compared to conventional-dose therapy (06). In another International Society of Pediatric Oncology study, pre-irradiation chemotherapy with vincristine, VP16, and carboplatin resulted in better progression-free survival (but not overall survival) than treatment with 3600 cGy of craniospinal RT alone (106). Delays in the completion of radiotherapy detrimentally affect outcome (106; 05; 117).

To try to reduce sequelae secondary to the scatter of craniospinal radiation to other organs, a variety of approaches such as intensity-modulated radiotherapy and proton-bean radiotherapy have been used, with preliminary results demonstrating a reduction in radiation exposure to organs, such as the heart and liver, without a deterioration in the rate of tumor control (75; 77; 96; 22).

A prospective, randomized trial has compared two forms of adjuvant chemotherapy and reduced doses of radiotherapy (2340 cGy) in an attempt to both improve survival for children with average-risk disease and reduce sequelae. This was partially based on one study in 68 children demonstrating approximately an 80% event-free survival after treatment with 2340 cGy of craniospinal radiotherapy and adjuvant, post-radiotherapy lomustine, vincristine, and cisplatinum chemotherapy (70). The results of the randomized trial of 421 children between 3 and 21 years of age demonstrated a 5-year event-free survival of 81% (69). Another study using the same dose of radiotherapy and high-dose chemotherapy demonstrated similar outcomes (29). One pilot study demonstrated a 70% (plus or minus 20%) 5-year disease-free survival rate when even further reduced craniospinal radiotherapy was coupled with chemotherapy in 10 patients (35). In addition, lower cumulative doses of post-radiotherapy cisplatin have not been associated with poorer rates of disease control (65).

In 2021, the results of the Children’s Oncology Group study comparing 1800 cGy of craniospinal radiation to 2340 cGy of craniospinal radiation in children between 3 and 7 years of age with nondisseminated, nonanaplastic medulloblastoma with less than 1.5 cm² of residual disease after surgery was reported (63). In this trial, which randomized over 500 children, treatment with the reduced dose of craniospinal radiation (1800 cGy) was associated with inferior progression free survival and a higher rate of disseminated relapse than seen in patients treated with standard 2340 cGY of craniospinal radiation. Although this study was not statistically powered for molecular stratification, group 4 patients had the poorer survival with the reduced dose of craniospinal radiation. It should, however, be noted that within the subgroup of children between 3 and 7, the reduction of the craniospinal radiation dose was associated with improved neurocognitive outcome. In the same study, all patients between 3 and 21 years of age were randomized between whole posterior fossa and tumor only with a margin boost radiation therapy. Progression free and overall survival were found to be equivalent within the two radiation boost subgroups. An interesting finding that needs to be confirmed in a larger population is those patients with SHH-pathway activated tumors had possibly better overall progression free survival with the more restrictive radiation boost. Because germline predisposition testing was not done routinely in the study, these results are difficult to interpret, but one possibility is the lower volume of boost radiation therapy was associated with better outcomes in children with germline mutations underlying their SHH tumors because the “late relapsers” in the SHH group were not medulloblastoma relapses but rather radiation-induced secondary high-grade tumors occurring in regions in the larger radiation-boost posterior fossa fields.

In another recently completed Children’s Oncology Group study also reported in 2021, children between the ages of 3 and 21 years of age with high-risk medulloblastoma were randomized to receive either carboplatin or no carboplatin during radiation therapy on a daily basis (55). Both arms of patients in the study continued to receive vincristine during radiotherapy. Although there was not a difference in progression-free and overall survival for the study as a whole, those children with group 3 tumors (on post-HOC evaluation) receiving carboplatin during therapy had better progression-free survival than those who did not receive carboplatin. Another part of this prospective randomized study evaluated the addition of retinoic acid following completion of radiation; this randomization did not show a difference in survival or progression-free survival in those who did or did not receive retinoic acid.

For patients with high-risk disease, there is evidence that adjuvant chemotherapy is of benefit. Although responses can be seen after pre-irradiation chemotherapy, there is no evidence that such approaches improve survival (52; 50; 118; 32; 62; 107). In a Children's Cancer Group Study, patients treated with pre-irradiation 8-in-1 therapy had poorer survival rates than those receiving chemotherapy after radiation, raising the question of whether delay of radiotherapy results in poorer overall disease control (01). A study by the German Society of Pediatric Hematology-Oncology also demonstrated poorer survival if pre-radiation chemotherapy was given in average-risk patients, but there was no survival difference in high-risk patients (49).

Another approach is to utilize radiation at a higher total dose, but deliver it in a different dose fraction sequence—hyperfractionated accelerated radiation therapy. Its use has not been demonstrated to result in clear-cut survival benefits for patients (61; 44). Follow-up studies for those treated with hyperfractionated accelerated radiotherapy and thiotepa demonstrated a subgroup of children who developed white matter changes not usually seen after standard radiotherapy (112). Such changes suggest a higher likelihood of long-term neurocognitive sequelae and may be confused with tumor recurrence. Even another approach is the use of chemotherapy after initial tumor biopsy, in an attempt to shrink the tumor prior to definitive surgery (38) and make subsequent surgery more extensive and less hazardous.

For children younger than 3 years of age with non-SHH activated medulloblastoma, standard treatment carries with it an extremely high risk of significant long-term intellectual sequelae. Combinations such as MOPP (mechlorethamine, vincristine, prednisone, and procarbazine) and the 4-drug regimen of cyclophosphamide, cisplatinum, VP-16, and vincristine have shown the ability to control disease in subsets of patients following surgery without the use of radiotherapy (19). Approximately 40% of children with medulloblastoma treated with chemotherapy alone will have stable disease while receiving chemotherapy. This is especially true for those children with totally resected, nondisseminated tumors prior to the initiation of treatment. More aggressive regimens may be more effective, including those utilizing autologous stem cell rescue (33; 111; 20). Some evidence suggests that those patients with a complete response to chemotherapy may not require radiotherapy (33; 101; 111; 90). Studies have also shown that methotrexate given intraventricularly and systemically at high doses may improve survival for infants and obviate the need for radiotherapy (92). However, such an approach may cause additive neurotoxicity. Also, in relapsed patients with local disease, long-term salvage has been reported after re-treatment with high-dose chemotherapy and local radiotherapy (21). Still, other approaches include the addition of local radiotherapy to the primary tumor site following chemotherapy in those children with nondisseminated disease and the use of intrathecal mafosfamide, a pre-activated form of cyclophosphamide (04; 10). The latter two approaches have not clearly improved outcome.

Significantly better survival rates have been seen in infants and young children (arbitrarily less than 3 or 4 years of age dependent on study) with SHH-activated medulloblastoma treated with chemotherapy alone. Progression-free and overall survival rates of 75% have been obtained after treatment with high-dose chemotherapy with stem-cell rescue or with regimens that have included both intravenous methotrexate and/or intrathecal chemotherapy. Treatment with a regimen that did not include methotrexate or intrathecals was not as successful (90; 54).

Outcomes

Reported progression-free and overall survival rates for patients with medulloblastoma have risen over the past two decades. This is likely due to multiple factors including the widespread use of cisplatin-based adjuvant chemotherapy regimens and the conventional acceptance of the extensive primary site resection as a critical component of treatment. For children greater than three years of age with average-risk disease after treatment with craniospinal and local boost radiotherapy and adjuvant chemotherapy, during and after radiation therapy, expected 5-year progression-free survival rates of greater than 75% or greater have been documented (29; 69). The majority of patients surviving greater than five years from diagnosis are “cured” of their disease. However, late relapse may occur. Late failures have to be carefully separated from the development of secondary tumors. Even for those patients with poor risk disease after treatment with both craniospinal and local boost radiotherapy and chemotherapy, 5-year progression free survival rates between 50% and 65% have been reported (42).

As noted previously, there may be subgroups of patients greater than three years of age who carry an even better survival rate, such as those children with WNT-driven tumors, with survival rates which exceeds 90%. In contradistinction, children with MYC-amplified group 3 tumors and those with p53 mutated SHH tumors have a much poorer prognosis, with survival rates of 40% or less five years from diagnosis. Similarly, children with wide spread disseminated disease carry a poor prognosis.

For children less than three years of age, prognosis is not as favorable as noted in older children, either because of biological differences between tumors in older and younger patients or because of the reluctance to use craniospinal radiation therapy in the younger children. However, for those with SHH tumors without concomitant p53 mutation, treatment with chemotherapy, especially aggressive chemotherapy utilizing drugs at high doses (supplemented by bone marrow or autologous stem cell rescue) or intrathecal chemotherapy, has a relatively favorable prognosis, as 75% or greater of patients can be expected to be alive five years from diagnosis (38; 93; 13). The prognosis of infants and children less than three years of age treated with chemotherapy alone with non-SHH driven tumors is less favorable, ranging from 25% to 50% (38; 10; 13; 54).

The follow-up posttreatment for children with medulloblastoma is relatively standardized. There is consensus that posttreatment MRI scanning of the brain and spine are indicated, usually every three months for the first two to three years after treatment, and then at a less frequent interval (6 months) until five years from diagnosis or end of treatment. The need for routine cerebrospinal fluid monitoring in children without evidence of dissemination at diagnosis is less clear. “Liquid biopsy,” assessing for cell-free DNA in cerebrospinal fluid, is a promising technique; if found to be a reliable means to detect early relapse prior to MR detection, it will dramatically alter the timing and need for cerebrospinal fluid assessment (26; 56).

The sequelae of treatment for children with medulloblastoma are significant. Although surgery is usually well tolerated for the majority of the patients, with relatively few patients having significant long-term, surgically related morbidity, some patients develop a poorly understood postsurgical syndrome of cerebellar mutism, obtundation, and severe motor dysfunction (12; 18). Craniospinal radiotherapy has been related to the development of significant intellectual and endocrinological long-term deficits (71). Intellectual deficits, including declines in overall intelligence and more selective dysfunctions, such as memory deficits, speech dysfunction, visual-spatial abnormalities, and attention difficulties, are progressive and more common in younger patients. Chemotherapy may also cause sequelae. As an example, cisplatinum may cause audiologic dysfunction (65). Overall, survivors of medulloblastoma commonly have life-altering neurologic, neurocognitive, and psychosocial sequelae (03).

Late failures may occur in children with medulloblastoma, but need to be carefully separated from secondary tumors. Secondary tumors, including the development of glioblastoma multiforme and leukemia, are being increasingly reported (74; 08). The benefits of surveillance testing are still being debated, although most still recommend routine follow-up studies for at least five years after diagnosis (51; 114). Primary relapse may occur outside the nervous system, especially in patients who have received postsurgery radiotherapy without adjuvant chemotherapy (104). When relapse occurs, independent of it occurring “early” or “late” after diagnosis, medulloblastoma molecular subtypes remain stable, although with time the tumor gains other mutations, making treatment of recurrence difficult (115).