Historical Perspective and Epidemiology

The term medulloblastoma (MB) was initially proposed by Bailey and Cushing in 1925 to refer to the common, midline cerebellar tumor of childhood initially labeled as “spongioblastoma cerebelli.”¹ This group of tumors was thought to derive from a multipotent stem cell known as the “medulloblast.” Despite the fact that the medulloblast has yet to be identified, this group continues to be referred to as medulloblastoma.

The most common malignant pediatric posterior fossa tumor,² MB accounts for 20% to 25% of all pediatric brain tumors.^3–6 Its incidence is estimated at 2 to 6 cases per million children per year, with approximately 540 new cases diagnosed annually in the United States.^3,7–10 The incidence of MB appears to be declining, although this observation has been debated.^3,10 MB is also one of the most common brain tumors of infancy and was detected in 12.2% in one large retrospective review.¹¹ The median age at diagnosis, however, is approximately 6 to 9 years.^9,12–14 Although MB is typically considered a pediatric malignancy, up to 30% of cases occur in adulthood.¹⁵ In adults, the annual incidence has been estimated to be 0.5 case per million per year.¹⁶ The male-to-female ratio is approximately 2 : 1.^3,17

MB typically arises in the midline cerebellum, in the region of the mid and inferior vermis.^13,18 The tumor may grow to occupy much of the fourth ventricle, where it blocks circulation of cerebrospinal fluid (CSF) and causes hydrocephalus. MB may spread along the cerebellar peduncles and extend upward through the tentorial hiatus or downward into the cervical spinal canal. Brainstem invasion may also occur. In addition, up to 30% of patients have metastatic disease at diagnosis.¹⁹

Genetics

MB is a neoplasm resulting from dysregulation of normal developmental processes.²⁰ In all likelihood, MB originates from the granule cell precursor (GCP) cell of the developing cerebellum. However, some studies suggest that more than one cell of origin exists and that this may explain in part the histologic and molecular heterogeneity seen with MB.^21,22 Recent evidence indicates that MB may arise from either neuronal stem cells or granule cell lineage–restricted neuronal precursors.^23,24

Because MB typically occurs on the surface of the cerebellum, it was postulated that GCPs are the MB cell of origin.²¹ This theory has been supported by evidence from genetically engineered mouse models.^21,25 During normal development, GCPs migrate from the rhombic lip and pass dorsally over the surface of the cerebellar anlage to form a transient secondary germinal zone called the external granule layer. Within the outer external granule layer, cells undergo a period of rapid expansion, subsequently exit the cell cycle, and migrate inward to form the internal granule layer. Several developmental signaling pathways, including hedgehog and Wnt/wingless, are known to play a role in this process and, when disrupted, may contribute to the pathogenesis of MB.

Evidence for this theory is provided by two familial cancer syndromes: Gorlin’s syndrome and Turcot’s syndrome.²⁶ Gorlin’s syndrome is a rare, autosomal dominant disorder that predisposes patients to the development of cancer, including multiple cutaneous basal cell carcinomas. MB also develops in up to 5% of patients with Gorlin’s syndrome. Gorlin’s syndrome results from germline mutations of the PTCH1 gene, which encodes the transmembrane receptor involved in hedgehog signaling. As a result, hedgehog signaling is upregulated aberrantly, thereby contributing to tumorigenesis. The hedgehog pathway remains the best characterized signaling pathway implicated in the pathogenesis of MB. Several mouse models, engineered to exhibit increased hedgehog signaling, show an increased frequency of MB formation, even recapitulating the leptomeningeal dissemination seen in the human disease.^25,27 Mutations involving PTCH1 or downstream hedgehog signaling components (SUFU, PTCH2, SMO) are seen in up to 25% of sporadic MB cases as well.^28–30 Aberrant hedgehog signaling typically results in formation of the desmoplastic MB subtype.^31,32

Turcot’s syndrome is another rare heritable disorder caused by mutation of the adenomatous polyposis coli (APC) gene. It predisposes patients to colon cancer, as well as central nervous system tumors, including MB.^33,34 APC functions to regulate the Wnt/wingless pathway. Mutations affecting many components of the Wnt/wingless pathway (APC, AXIN, β-catenin) have been identified in up to 15% of sporadic MB cases, with activating mutations of β-catenin being most common.^33,35–38

A third familial cancer syndrome associated with MB formation is Li-Fraumeni syndrome. Li-Fraumeni syndrome results from inherited mutations of the p53 tumor suppressor gene.³⁹ Affected patients are at increased risk for the development of a wide spectrum of malignancies, including MB.⁴⁰ Nonetheless, mutations of p53 are uncommon in sporadic cases of MB.⁴¹ Additional syndromes in which MB has been observed include Rubenstein-Taybi and Aicardi’s syndrome.^26,42

Another genetic abnormality seen in MB is amplification of the myc family of oncogenes—seen in up to 10% of cases.^43,44 Microarray studies have identified unique molecular MB subgroups defined by their gene expression profiles.³¹ Molecular subgroups characterized by Wnt/wingless signaling, hedgehog signaling, and expression of neuronal differentiation genes, photoreceptor genes, or both, have been identified.^31,45,46

Specific, recurrent cytogenetic abnormalities are also seen. Common chromosomal copy number changes include gain of chromosomes 1q and 7, as well as loss of chromosomes 22, 11, 10q, and 17p.^13,43,47 Loss of 17p is observed in up to 50% of cases and may occur in the context of an isochromosome 17q.^43,47 Despite research efforts, the putative tumor suppressor gene located on chromosome 17p has not been definitively identified. Many additional genetic and epigenetic events have been implicated in MB pathogenesis; however, an exhaustive review of these events is beyond the scope of this chapter.⁴⁸

Pathology

Rorke initially proposed incorporating MB in a group of tumors that include retinoblastoma, neuroblastoma, and pineoblastoma, referred to as primitive neuroectodermal tumors (PNETs).⁴⁹ Subsequent gene expression profiling studies have demonstrated that MB is indeed a separate entity, distinct from supratentorial PNETs.⁵⁰ The World Health Organization (WHO) recently updated its classification of central nervous system tumors, with modifications in the classification of MB.¹³ According to the revised classification, all MBs are considered WHO grade IV tumors. Histologic variants include the classic, desmoplastic/nodular MB with extensive nodularity (MBEN), anaplastic, and large-cell subtypes (Fig. 201-1).¹³

FIGURE 201-1 Neuropathology of medulloblastoma. A, Hematoxylin and eosin (H&E) stain of classic medulloblastoma (200×). B, H&E stain of desmoplastic medulloblastoma (100×). C, Reticulin stain of desmoplastic medulloblastoma demonstrating reticulin-positive internodular zones and reticulin-free pale islands (100×). D, H&E stain of anaplastic medulloblastoma (400×). Note the increased nuclear-to-cytoplasmic ratio. Arrows highlight the cell-cell wrapping characteristic of this subtype. E, H&E stain of large-cell medulloblastoma demonstrating large nuclei with prominent nucleoli (400×).

On gross pathologic inspection, MB appears as a pinkish gray to purple mass, commonly arising from the medullary velum and deriving most of its blood supply from the posterior inferior cerebellar artery. Some lesions are firm, discrete masses, whereas others are soft and friable. In a minority of cases, the tumor invades the floor of the fourth ventricle. Occasionally, gross hemorrhage is seen. CSF dissemination may result in a white, “sugar-coated” appearance of the cerebellar surface.

Microscopically, classic MB appears as a dense sheet of small, basophilic cells with little cytoplasm and round to oval hyperchromatic nuclei (Fig. 201-1A). A high mitotic index may be seen. Evidence of differentiation along the neuronal or glial lineage is noted in up to 50% of cases.⁵¹ Homer-Wright rosettes are commonly present.¹³

The desmoplastic/nodular MB subtype accounts for approximately 20% of cases, although it has been shown to account for as little as 5% and as many as 57% depending on the patient population examined.⁵² Grossly, this variant is often located more laterally within the cerebellar hemisphere and may exhibit pial invasion. Microscopically, it is characterized by pale, reticulin-free nodules surrounded by reticulin-positive collagen fibers (Fig. 201-1B and C). These nodules represent regions of more advanced neuronal differentiation.

In the current edition of the WHO classification, two new MB variants were introduced: anaplastic MB (WHO grade IV) and MBEN (WHO grade IV).^13,53 Anaplastic MB is characterized by significant nuclear pleomorphism, nuclear molding, cell-cell wrapping, and a high mitotic index (Fig. 201-1D).^13,53 Similarly, the large-cell variant is characterized by large round nuclei, prominent nucleoli, nuclear molding, and abundant cytoplasm (Fig. 201-1E). The large-cell and anaplastic MB variants overlap considerably in appearance and are often considered one subgroup (LC/A) in clinical studies.^13,53 CSF dissemination at diagnosis is more common with the large-cell and anaplastic subtypes. MBEN typically occurs in patients younger than 3 years and is characterized by the presence of lobular, grape-like nodules as a result of expansion of the reticulin-free zones. This variant is closely related to desmoplastic/nodular MB. When compared with the desmoplastic/nodular variant, the reticulin-positive internodular zones are markedly reduced.

Although the results of immunohistochemical analysis of MB samples can vary, the majority display some degree of neuronal differentiation with immunopositivity for synaptophysin, neuron-specific enolase, and class III tubulin.^13,54 Expression of glial fibrillary acidic protein may occur, although its incidence varies.^54–57

Clinical Findings

Children with MB may have a variety of signs and symptoms. The most common constellation of symptoms is the triad of headache, lethargy, and vomiting.⁵⁸ Most children have a short clinical history, with symptoms being present for fewer than 1.5 months in approximately 50% of patients and fewer than 3 months in about 75%.⁵ In infants and young children, the diagnosis is suspected in the setting of irritability, loss of appetite, weight loss, and failure to thrive. In addition, they may display signs of increased intracranial pressure, including lethargy, drowsiness, vomiting, sunsetting, a full fontanelle, or an increasing head circumference. Older children may complain of headache, neck stiffness, dizziness, or diplopia. On neurological examination, patients demonstrate truncal or appendicular ataxia, dysmetria, nystagmus, or cranial nerve palsies. A head tilt, signifying descent of the cerebellar tonsils into the foramen magnum with compression of the C1 or C2 nerve roots, may also be observed. These signs and symptoms are not specific to MB, and the differential diagnosis should include other posterior fossa lesions such as astrocytoma, ependymoma, or cystic mass lesions.

Diagnostic Imaging

Because of its cellular nature and high nuclear-to-cytoplasmic ratio, MB typically appears as a homogeneous hyperdense midline mass within the posterior fossa on non–contrast-enhanced computed tomography, with variable amounts of peritumoral edema and hydrocephalus. Calcification, necrosis, cystic degeneration, and hemorrhage are seen in a minority of cases. After administration of contrast material, MB demonstrates homogeneous enhancement.

Magnetic resonance imaging (MRI) is the imaging modality of choice for the diagnosis and preoperative evaluation and staging of MB patients. The current standard of care should be to perform non–contrast-enhanced and contrast-enhanced MRI of both the brain and spine preoperatively when the presumptive diagnosis is MB.⁵⁹ In addition to preoperative studies, early postoperative MRI (within 48 hours) to assess the degree of residual tumor is important for staging MB patients and directing further therapy. When early postoperative spinal MRI is not performed, one must wait at least 2 weeks after surgery before obtaining such imaging to more accurately distinguish between true CSF dissemination and postoperative artifacts.

On MRI, MB typically appears as a hypointense to isointense mass on T1-weighted images and as an isointense to hyperintense mass on T2-weighted images.^18,60 Gadolinium enhancement is typically robust but may appear homogeneous or slightly heterogeneous.^18,60,61 MRI identifies intratumoral cysts in up to 75% of patients (Fig. 201-2).⁶¹ Metastatic spread within the spinal canal is seen as foci of enhancement on T1-weighted imaging. Diffusion-weighted images demonstrate restricted diffusion.⁶² Proton magnetic resonance spectroscopy may assist in discriminating between the various pediatric posterior fossa tumors.⁶³

FIGURE 201-2 Fourteen-year-old boy with a 2-month history of headache, nausea, and vomiting. Left, Axial Fast Spoiled Gradient-Recalled Echo pulse sequence (FSPGR) Bravo magnetic resonance imaging (MRI) showing a mass lesion in the right cerebellar hemisphere displacing the fourth ventricle. The lesion is of variable signal intensity. Middle, Axial contrast-enhanced MRI showing an enhancing lesion with regions of cystic degeneration. Right, Sagittal MRI with contrast enhancement shows the position of the posterior fossa mass lesion in reference to the fourth ventricle and tentorium.

Staging and Prognostic Factors

Careful preoperative and postoperative staging is crucial for directing therapy and providing an estimate of patient prognosis. The staging system proposed by Chang and colleagues classified patient lesions according to two parameters: tumor stage (T) and metastasis stage (M).⁶⁴ The T stage, based on preoperative tumor size and location, has not proved to be useful in estimating prognosis and directing adjuvant therapy. In contrast, the M stage has been shown to be of clinical significance.⁶⁵ The initial M staging was divided into five subcategories. M0 referred to tumors with no evidence of gross, cerebrospinal, or hematogenous metastases. M1 patients had positive microscopic CSF cytology. M2 patients had evidence of gross nodular seeding involving the cerebral or cerebellar subarachnoid space or third/lateral ventricles. The M3 stage demonstrated gross nodular seeding of the spinal subarachnoid space, and the M4 stage was assigned to patients with extraneural metastases.

Currently, average-risk patients are defined as children 3 years and older without evidence of gross or microscopic metastatic disease at diagnosis and with less than 1.5 cm² of residual tumor after surgical resection. High-risk patients include those younger than 3 years or any patient with evidence of metastatic tumor spread or significant residual tumor (>1.5 cm²) after surgery.

One of the factors most consistently associated with prognosis is age at diagnosis.^66–72 In a review of 38 patients with newly diagnosed MB who were treated between 1994 and 2003, age younger than 3 years at diagnosis was significantly associated with worse overall survival (OS) than was age older than 3 (36.3% versus 73.4%, respectively).⁷³ Poor outcomes in younger patients may be a reflection of differences in treatment patterns, such as delay or avoidance of adjuvant radiotherapy.^4,71 The fact that the MBEN subtype carries a better prognosis despite typically being diagnosed in younger patients suggests that age in and of itself may not be the causative factor in the different prognosis seen between very young and older patients.^74,75 Nonetheless, evidence suggests that stratification by age is clinically significant and that further stratification of MB patients according to age may be warranted. Sure and associates observed better outcomes in patients older than 10 years than in younger patients.⁷⁶ A recent Canadian Pediatric Brain Tumor Consortium study of adolescent MB patients found that time to relapse was linearly proportional to age at diagnosis.⁷⁷ This study suggested that adolescent patients may follow a unique clinical course and that revised risk stratification protocols may be needed for this subgroup of patients.

Absence of metastatic disease at diagnosis has been associated with improved OS and event-free survival (EFS).^65,78–82 In the Children’s Cancer Study Group CCG921 trial, M stage showed prognostic significance in children older than 3 years, with M0, M1, and M2+ patients demonstrating 5-year progression-free survival (PFS) rates of 70%, 57%, and 40%, respectively.⁶⁵ In high-risk patients, M stage also appears to predict outcome.⁸³ In a study of 115 children with high-risk MB, Verlooy and coworkers observed 5-year EFS rates of 68.8%, 58.8%, and 43.1% for patients with residual posterior fossa disease, stage M1 disease, and stage M2/3 disease, respectively.⁸³

Maximal safe surgical resection has long been one of the primary interventions for MB because it was believed that children with minimal residual tumor burden had better PFS, especially those without metastatic disease.^{3,66,70,71,76,82,84–86} The CCG921 trial showed that the 5-year PFS rate for children without metastatic disease at diagnosis and less than 1.5 cm² of residual tumor after resection was 78% versus 54% for those with 1.5 cm² or greater residual disease.⁶⁵ However, there does not appear to be any advantage in PFS from achieving gross total resection versus near-total resection (>90%) with less than 1.5 cm² of residual tumor.^58,87 Furthermore, residual tumor with brainstem involvement did not confer a worse prognosis.⁸⁷

Despite its clinical utility, a criticism of the current staging system is its inability to differentiate between average-risk patients who may benefit from more aggressive intervention to maximize survival or less aggressive treatment to minimize the deleterious consequences of therapy. The identification of additional clinical, histologic, and molecular prognostic markers may improve the ability to estimate prognosis and direct risk-adapted therapy for MB patients.

Additional clinical factors identified as prognostic markers include patient sex, as well as histologic factors such as tumor subtype, nodularity, and the degree of anaplasia. Patient sex has been associated with prognosis in a limited number of reviews.⁷⁰ Weil and colleagues observed that female patients had better outcomes after surgery, postoperative radiotherapy, and chemotherapy.⁸⁸ Tabori and associates found that adolescent girls fared better than adolescent boys.⁷⁷ With respect to MB subtype, MBEN carries the best prognosis despite affecting very young children.^74,75 The LC/A subtypes demonstrate the worst prognoses.^79,89–91 Desmoplastic MB is thought to carry a more favorable prognosis than classic MB, but this finding is not observed consistently.^89,92–98 Nodularity is seen in up to 29% of MBs, although the extent of nodularity can vary.⁹⁵ Only extensive nodularity (between 96% and 100%) is associated with better survival.⁹⁵ Anaplasia has inconsistently shown prognostic significance in patients with MB. Most large series support anaplasia as a poor prognostic marker. Reviewing 474 MB patients from six recent Pediatric Oncology Group (POG) protocols, Brown and coworkers found that long-term survival in patients with large-cell MB with anaplasia was less common than in patients with large-cell MB without anaplasia.⁹⁹ In a review of 347 patients enrolled in the International Society of Pediatric Oncology SIOP II trial, 5-year EFS was significantly shorter in patients with tumors showing severe anaplasia than in those with mild to moderate anaplasia.¹⁰⁰ In this group of patients the degree of anaplasia was the most significant histologic prognostic feature.¹⁰⁰ Similarly, Eberhart and associates observed an association between severe anaplasia and poor prognosis.¹⁰¹ Conversely, anaplasia was not identified as a significant prognostic marker in a study of 74 MB cases, including 16 with the LC/A subtype.¹⁰² In a study examining adult MB patients, no correlation between overall survival and any histologic feature, including nodularity, desmoplasia, nuclear size, nuclear polymorphism, necrosis, or endothelial proliferation, was seen.¹⁰³

In recent years, investigators have begun to correlate molecular markers with clinical outcomes. Segal and colleagues first observed an association of high TrkC expression and improved survival, with 5-year survival rates of 89% in high expressers versus 46% in low TrkC expressers.¹⁰⁴ Ray and coworkers observed a correlation between TrkC immunopositivity and good prognosis.⁸⁰ An early study by Grotzer and associates found that little to no TrkC expression conferred a 4.8-fold greater risk for death than did high TrkC expression and that TrkC was a significant predictor of both PFS and OS.¹⁰⁵ Examining archived tumor samples from the cohort of patients enrolled in the HIT ’91 trial, Grotzer and colleagues similarly found TrkC expression to be an independent prognostic factor.¹⁰⁶ In some studies, however, an association between TrkC expression and improved prognosis was not identified.¹⁰⁷

Amplification of the myc family of oncogenes (MYCC and MYCN) has been implicated in the pathogenesis of MB. In a study of 124 well-characterized primary MB samples, low c-myc expression was identified as a good prognostic marker.¹⁰⁸ Eberhart and colleagues identified c-myc mRNA expression in 31% of primary tumors, and its expression correlated with the anaplastic subtype and reduced survival.¹⁰¹ Examining a cohort of patients derived from the SIOP/United Kingdom Children’s Cancer Study Group (UKCCSG) PNET-3 trial, Lamont and coauthors identified c-myc and N-myc amplification in 6% and 8%, respectively.⁷⁹ Myc family gene amplification was also associated with the LC/A subtype in this study, with high-level amplification defining a group of patients at increased risk for poor outcomes.⁷⁹ Analysis of samples from the HIT ’91 trial similarly identified c-myc gene expression as a significant prognostic marker.¹⁰⁶ Patients with elevated TrkC and low c-myc were identified as a “good risk” subgroup, whereas those with low TrkC and high c-myc were at increased risk for poor outcomes.¹⁰⁶ As with TrkC, however, c-myc and N-myc gene expression is not identified as a significant marker across all studies.¹⁰⁷

Overexpression of the ERBB2 oncogene has also been implicated in MB. Gilbertson and coauthors reported an association between ERBB2 expression status and clinical outcomes.¹⁰⁹ Patients with tumors expressing ERBB2 in more than 50% of cells had a 10-year survival rate of just 10% versus 48% for patients expressing ERBB2 in less than 50% of tumor cells.¹⁰⁹ Tissue microarray analysis identified ERBB2 immunopositivity as an independent marker of poor prognosis.⁸⁰ Expression of ERBB2 has also been associated with the LC/A subtype.¹⁰⁷ Gajjar and coworkers observed that 100% of patients with average-risk MB and negative ERBB2 status remained alive at a median follow-up of 5.6 years as opposed to only 54% of those with average-risk MB and positive ERBB2 status.¹⁰⁷

Activated Wnt/wingless signaling has been implicated in the pathogenesis of 15% to 25% of primary MB cases, a unique molecular subgroup of patients who cannot be identified on the basis of clinical or pathologic features alone.^45,110 Along with activation of the Wnt pathway, these tumors commonly exhibit loss of one copy of chromosome 6.⁴⁵ Nuclear positivity for β-catenin (a marker of Wnt pathway activation) has been associated with improved OS and EFS regardless of MB subtype.¹¹⁰ All children with positive nuclear β-catenin staining and either the LC/A subtype or metastatic disease at diagnosis remained alive at 5 years’ follow-up.¹¹⁰

As a result of ongoing research effort, the clinical, pathologic, and molecular criteria used to assign MB patients to prognostically significant subgroups will continue to evolve. Several novel molecular markers have already been identified and should be considered for inclusion in future clinical trials for prospective validation of their prognostic utility. This will gain increasing importance as novel MB therapies against specific molecular targets are designed and implemented in clinical practice.

Treatment

The collaborative effort of neurosurgeons, neuro-oncologists, radiation oncologists, neuroradiologists, and neuropathologists is crucial for delivering appropriate and timely multimodality therapy to patients with MB. Currently, most patients are treated with a combination of maximal surgical resection, craniospinal irradiation (CSI), and chemotherapy.

Surgery

Surgical resection has historically played a central role in the treatment of children with MB and remains a crucial component of current therapy. Although initial attempts at resection were associated with significant mortality rates, current series report nearly absent surgical mortality.^12,111,112 Surgical goals include obtaining a tissue diagnosis, achieving maximal safe tumor resection, relieving critical structures from mass effect, and addressing any associated hydrocephalus.

Symptomatic hydrocephalus is present in the majority of children with posterior fossa tumors at the time of diagnosis.^113–115 Depending on the patient’s symptoms and the severity of the associated hydrocephalus, a decision must be made whether to treat the hydrocephalus beforehand or at the time of tumor resection. Options for treating the hydrocephalus at initial encounter include temporary CSF diversion via an external ventricular drain (EVD), diversion through a ventriculoperitoneal shunt (VPS), and more recently, diversion via endoscopic third ventriculostomy (ETV). The decision to resort to VPS or ETV must be tempered by the fact that only 10% to 40% of patients ultimately require permanent CSF diversion after resection of the tumor.^114–118 Factors suggesting the need for permanent postoperative shunting include more severe hydrocephalus at diagnosis, younger patient age, and larger preoperative tumor size.^{73,113,116,117} Patients with MB, as opposed to other posterior fossa tumors, also appear to be at higher risk for postoperative hydrocephalus requiring a shunt.^113,118 In the majority of cases, however, patients with symptomatic hydrocephalus may be managed with intravenous corticosteroids alone, with surgical resection performed on a semielective basis. For patients seen in extremis because of symptomatic hydrocephalus, emergency insertion of an EVD is the most common strategy used.

At surgery, patients are typically placed in the prone position. This position allows the use of a horseshoe-type head holder in very young patients, for whom the use of a pin-type head holder is not possible. Operative adjuncts that may be used during surgery include frameless stereotaxy, ultrasonography, and evoked potential or cranial nerve monitoring, or both. Surgical site preparation and draping should allow creation of an occipital bur hole for intraoperative EVD placement if required.

A standard midline posterior fossa exposure is performed, with the incision extending from slightly rostral to the inion down to the midcervical region. The cervical musculature is mobilized laterally off the occiput and cervical laminae in the midline avascular plane. A posterior fossa craniotomy is then performed. If tumor is known to extend into the cervical spinal canal, laminectomy of the upper vertebrae may be performed at this time. The dura is then opened in a Y fashion, with care taken to achieve hemostasis of the occipital and circular dural sinuses, which are commonly encountered. Suture ligation is preferred over metal clips because it is more secure and does not result in metal artifact on postoperative MRI (Fig. 201-3).

FIGURE 201-3 Left, Operative view of medulloblastoma before neurosurgical removal. The tumor is found protruding through the fourth ventricle and is displacing the cerebellar tonsils upward. Right, After complete neurosurgical resection, the floor of the fourth ventricle is inspected for residual tumor. The landmarks of the fourth ventricle can be appreciated.

Via a standard, midline approach to a fourth ventricular MB, the cerebellar tonsils are separated in the midline. Cotton pads are placed along the floor of the fourth ventricle when a clear plane is visible between it and the tumor. Similarly, pads are used to help develop the plane between tumor and normal cerebellum for as great a distance as possible before tumor removal is begun. For adequate exposure over the perimeter of the tumor, the distal portion of the vermis often needs to be divided with microneurosurgical techniques. Resection of the tumor begins from its dorsal aspect, which often requires lateral displacement of the posterior inferior cerebellar arteries. The tumor is internally debulked with suction or the Cavitron and specimens are taken for pathologic assessment and tumor banking. As the tumor is debulked, the lateral tumor-cerebellum interface continues to be developed. One then approaches the floor of the fourth ventricle and encounters the pad that was placed on it at the beginning of the procedure. Care must be taken to identify areas where tumor may invade the floor of the fourth ventricle or cerebellar peduncles and not continue with tumor resection in these regions for fear of postoperative neurological deficits. At the end of the procedure, one inspects the aqueduct of Sylvius, the lateral foramina of Luschka, and the obex to ensure complete tumor removal.

As an alternative neurosurgical approach to MB, the neurosurgeon can select the inferior telovelar approach, which provides access to tumor in the fourth ventricle without the need for a vermian incision.¹¹⁹ A cerebellar hemispheric MB is typically approached with a horizontal or vertical incision through the cerebellar cortex that is the shortest distance from the tumor. The tumor–normal cerebellum interface is then developed in the same manner as described earlier before tumor removal.

Meticulous hemostasis is achieved with judicious use of bipolar coagulation and gentle pressure and is confirmed with a Valsalva maneuver. Care is taken to ensure a watertight dural closure, followed by replacement of the bone flap and multilayer closure of the superficial tissues. If an EVD was placed at the beginning of the procedure, it is maintained in the early postoperative period, followed by gradual weaning and discontinuation. Difficulty weaning off the EVD or the development of a tense pseudomeningocele implies continued hydrocephalus and should raise the question of whether permanent CSF diversion is required. MRI is performed within 48 hours postoperatively to assess the degree of residual disease (Fig. 201-4). Significant residual tumor (>1.5 cm²) should prompt consideration of early repeat resection, unless the procedure was stopped early because of problems with hemostasis or invasion of tumor into critical structures.

FIGURE 201-4 Same patient as in Figure 201-2 after neurosurgical resection of the lesion. The lesion was a large-cell, anaplastic medulloblastoma. Left, Axial contrast-enhanced magnetic resonance imaging (MRI) showing complete neurosurgical resection of the lesion without contrast enhancement. Middle, Sagittal contrast-enhanced MRI shows a defect from surgery but no evidence of residual tumor. Right, Despite remaining neurologically well 12 months after radiation treatment and chemotherapy, axial MRI at this time shows recurrence in the posterior fossa at the site of the original surgery, as well as in the right cerebellopontine angle as evidence of metastatic disease.

Buy Membership for Neurosurgery Category to continue reading. Learn more here