Spinal Cord

Radiation therapy can result in severe damage to the spinal cord, with functional transection of the cord at the affected level representing the most dire of such toxicities. Radiation damage takes the form of chronic progressive myelitis. The 5% incidence of radiation myelopathy probably lies between 57 Gy and 61 Gy in the absence of chemotherapy, using standard fractionation. At doses less than 45 Gy the risk of myelitis is less than 0.2%. At 73 Gy the risk of myelitis approaches 50%.^20,21 A number of reports have indicated that tolerance of the cervical spinal cord to radiation toxicity is somewhat higher than 45 Gy in adults. However, it is unclear what the cervical spinal cord tolerance is in pediatric patients.

Radiation to the spinal cord can also result in Lhermitte syndrome, which is characterized by tingling, numbness, and a sensation of electric shock. Symptoms are often present only with neck flexion. Lhermitte syndrome is typically self-limiting, presenting within the first 1 to 3 months following radiation and having an average duration of 3 to 4 months. Craniospinal radiation can result in decreased truncal, or sitting, height. This is due to decreased growth of the vertebral bodies following radiation therapy, and becomes clinically evident at doses greater than 20 Gy.

Brain

Acute reactions during radiation therapy, thought to result from disruption of the blood-brain barrier, are uncommon. However, there are reports of edema following single conventional fractions.²² Clinically, apparent acute changes are more common with hypofractionated doses, such as are used in radiosurgery.²³ Subacute reactions are more common and are thought to be due to transient demyelination.²⁴ These effects typically occur within the first few months following radiation and usually resolve within 6 to 9 months. Cranial nerve palsies have also been reported.²⁵ Severe subacute effects such as rapidly progressive ataxia are rarer and are generally associated with fractions larger than 2.0 Gy and total doses larger than 50 Gy.²⁶

The late effects of radiation are primarily due to radiation necrosis. Symptoms are related to the neuroanatomic location of necrosis (sensory, motor, speech/receptive deficits, seizures) and may, in addition, be due to increased intracranial pressure. Focal necrosis is uncommon with doses less than 60 Gy given with conventional fractionation.²⁷

Large-volume radiation therapy can result in diffuse white matter changes. Clinically, these can result in lassitude, personality change, or neurocognitive deficits. Multiple studies have examined the effect of whole-brain radiation therapy on intellect in patients treated for leukemia. Radiation-associated depression in intelligence quotient (IQ) has been noted by a number of authors.^28,29 Halberg and colleagues³⁰ compared three groups of patients. The first group received 18 Gy (1.8 Gy per daily fraction) cranial irradiation, and the second received 24 Gy cranial irradiation. The third group consisted of other oncology patients who did not receive cranial irradiation. Lower IQ scores were noted in the group receiving 24 Gy. High-dose methotrexate and female sex appear to increase risk of intellectual deficits.^31,32 The effects of cranial irradiation are also more severe the younger the child. Intellectual deficits resulting from radiation most commonly result in difficulty acquiring new knowledge, decreased processing speed, and memory deficits (most frequently short-term).³³

Studies of children irradiated for primary brain tumors have also shown intellectual deficits. Effects of radiation are more difficult to evaluate in this setting, as most of these patients also have had surgical resection. However, as with acute lymphoblastic leukemia patients, younger age appears to result in a higher rate of neurocognitive deficits. Larger fields and higher doses also appear to cause higher rates of toxicity.

Pediatric CNS tumors require a multimodality approach, including surgery, chemotherapy, and radiation. Attribution of toxicities such as neurocognitive decline to a single modality, most often radiotherapy, is not supported by the literature. Kiehna and colleagues³⁴ conducted a prospective trial of 120 children and young adults to evaluate neurocognitive aptitudes before, during, and after CRT. In this comprehensive study from St. Jude Children’s Research Hospital, investigators used the Conners continuous performance test to measure selective attention, sustained attention, impulsivity, and reaction time. During a 6- to 7-week course of CRT, children exhibited problems with cognitive processing, attention, and reaction times. Overall attention indices worsened during CRT, with significant risk factors including progression of symptoms prior to CRT, young age, and diagnoses of craniopharyngioma or low-grade astrocytoma.³⁴

After completion of CRT, serial assessments were performed at regular intervals and revealed that overall attentional capacity remained relatively stable over time, with some important exceptions. Specifically, the data documented significant worsening of inattentiveness and deterioration of global attention deficits that were particularly associated with craniopharyngiomas, optic pathway and diencephalic tumors, supratentorial tumors, subtotal resection, more than two surgeries, and the presence of an Ommaya reservoir. Analyses by tumor histology were particularly instructive, as inattentiveness steadily increased in those with incompletely resected craniopharyngiomas and optic pathway gliomas, and conversely, limited effects were observed in ependymoma patients despite their significantly younger age. Taken together, these results suggest that the tumor itself contributes significantly to such cognitive decline, and CRT is clearly not the sole or central culprit. Rather, acquired attentional disorders during and following CRT are due to a multitude of insults and are associated with tumor type, location, comorbidities, and surgical resections.

The group at St. Jude Children’s Research Hospital has also evaluated academic achievement and verbal memory performance in children after CRT. Factors including female gender, number of surgeries, and hydrocephalus, predicted worse verbal memory performance at baseline. Of key importance, however, verbal memory stabilized and improved over time, rather than declined.³⁵ In a similar study from the same group, serial assessments of academic achievement revealed that reading was the aptitude most susceptible to decline after CRT, whereas math and spelling skills remained stable.³⁶

Eye

The lens of the eye is exquisitely sensitive to radiation. Radiation-induced cataracts are caused by damage to the germinal zone at the equator of the lens. Initially, this results in a central opacity, which progresses to an opaque cortex. The threshold for radiation damage is 8 Gy in a single fraction, or 10 Gy to 15 Gy in fractionated doses. More rapid cataract formation is associated with higher doses of radiation.

Radiation-induced retinopathy appears to have a threshold of 46 Gy at conventional fractionation (1.8 to 2.0 Gy per fraction). Retinopathy is typically seen beginning from 6 months to 3 years following radiation. It is characterized by macular edema, nonperfusion, and neovascularization.

The optic nerves and chiasm are also at risk for damage from radiation. Damage to the optic nerve is characterized by a pale optic disc, abnormal papillary response, and visual deficits. Damage to the optic nerve or chiasm is potentially blinding. The threshold for damage is 50 Gy with fractions of 1.8 Gy. Every effort should be made to keep these structures below their tolerated doses.

Ear

Radiation-induced sensorineural hearing loss is dose-dependent and is more severe in younger patients. Hearing loss can result from doses greater than 40 to 50 Gy, usually developing within 6 to 12 months of treatment.³⁷ High-frequency hearing loss is seen in 25% to 50% of patients who received greater than 50 to 60 Gy to inner ear structures.^38–40 Hearing loss is typically attributed to radiation changes induced in the cochlea and vasculature. Ototoxicity related to cisplatinum chemotherapy is well documented.^41,42 Cranial irradiation before or concurrent with cisplatinum chemotherapy enhances ototoxicity.^41,43 Chronic otitis can also develop following radiation therapy because of obstruction of the eustachian canal.

Endocrine

Whole-cranial irradiation or more focal radiation that includes the hypothalamic-pituitary axis can result in neuroendocrine abnormalities. The neuroendocrine complications that can result from radiation therapy are summarized in Table 53-5. Growth hormone (GH) production appears to be the most prone to disruption by radiation therapy. GH deficiency worsens over time and may follow radiation doses as low as 12 Gy.⁴⁴ This appears to be a result of decreased GH-releasing hormone (GH-RH) in the hypothalamus, as GH deficiency is seen in patients undergoing hypothalamic radiation with pituitary sparing. Clinically, GH deficiency can result in short stature, bone loss, and metabolic abnormalities. Treatment with synthetic GH can allow children to maintain their expected growth percentile despite irradiation.

Table 53-5 Endocrine Abnormalities Resulting From Radiation

ACTH, Corticotropin; LH/FSH, luteinizing hormone/follicle-stimulating hormone; TRH/TSH, thyrotropin-releasing hormone/thyrotropin.

Other endocrine deficiencies, including decreased thyroid-stimulating hormone, corticotropin, follicle-stimulating hormone, and luteinizing hormone appear to have a higher threshold. These effects are typically seen following doses greater than 40 Gy. It is important that children treated for tumors in the region of the hypothalamic-pituitary axis be followed closely for endocrine abnormalities, so that replacement therapy can be initiated in a timely fashion.

Medulloblastoma

Medulloblastoma comprises approximately 20% of pediatric CNS tumors, although one third of medulloblastomas occur in adults.⁴⁵ Up to one third of medulloblastomas in children are associated with dissemination beyond the primary site at the time of diagnosis. Age younger than 3 years, inability to perform a gross total surgical removal of tumor, metastases, and diffuse anaplasia in the histologic specimen are associated with poor survival. Historically, standard treatment delivered 36 Gy to the brain and spine with a boost to the posterior fossa totalling 54 Gy. Significant abnormalities in intellectual development, growth, and hormonal function commonly occurred after CSI.^46,47

Today, CSI followed by chemotherapy is the standard of care for both average- and high-risk children ages 3 and older. In average-risk disease (defined as gross totally resected tumor without severe anaplasia, and no metastases), 23.4 Gy to the craniospinal axis, plus a boost to 54 Gy to the posterior fossa, followed by chemotherapy, is standard and has resulted in 5-year survival of 80% or better.⁴⁸ Areas of active investigation in average-risk disease are lowering of the craniospinal dosage to 18 Gy, conformal posterior fossa radiotherapy to the tumor volume, and intensification of chemotherapy with autologous peripheral blood stem cell support.⁴⁹ In high-risk disease (defined as subtotally resected tumor, severely anaplastic histology, or presence of metastases), 36 Gy craniospinal plus a boost at the posterior fossa to 54 Gy, followed by chemotherapy is standard, and results in 5-year survival rates of 50% to 70%. Incorporation of chemotherapy during irradiation and intensification of chemotherapy following radiation therapy are subjects of current trials. Among infants with medulloblastoma, many approaches are competing in trials, particularly chemotherapy with limited posterior fossa radiation, high-dose methotrexate with or without later CSI,⁵⁰ or very-high-dose chemotherapy without radiation therapy. Chemotherapeutics typically used in medulloblastoma include cisplatin, vincristine, cyclophosphamide, sometimes lomustine, and etoposide, or occasionally topotecan or thiotepa.

Until recently, standard of care has dictated that medulloblastoma and other lesions of the posterior fossa, be treated to the contents of the posterior fossa, with at least a 1-cm margin and the inferior border at C2. Volumes that target the entire posterior fossa deliver significant doses to the parietal, occipital, and temporal lobes. Limiting radiation doses to these regions of normal brain may reduce long-term toxicity. Hearing may be spared by reducing cochlear dose, endocrine function may be preserved by reducing pituitary-hypothalamic dose, and neurocognitive aptitude may be protected by reducing temporal lobe dose. Therefore, ongoing cooperative studies are evaluating whether the target volume for the primary site tumor boost in average-risk medulloblastoma can be reduced from whole posterior fossa to the tumor volume plus margin without compromising disease control.

Simulation

Most contemporary CSI treatment and planning approaches rely on the availability of MRI and CT data sets and 3-D capabilities of modern radiation treatment planning (3D-RTP) systems. Consequently, CT or MR simulations of CSI treatments are now more prevalent. The field setup and matching principles previously used during fluoroscopic simulation can now be easily reproduced and manipulated in the virtual space of the 3D-RTP system. For these and historical reasons, a “classic” fluoroscopic simulation is described in the following text, helping to illustrate potential pitfalls and treatment planning considerations related to the CSI. Although we describe a prone setup, the reader should be aware that a variety of supine techniques have also been reported in literature. The main historical reason for the prone setup was the ability to visualize and monitor skin gaps for junction matching and junctional shifts. Such visual monitoring is no longer required in the presence of alternative setup verification techniques such as daily portal imaging or on-board cone-beam CT. Furthermore, junction shifts can be simulated in the 3D-RTP system and automated during treatment delivery. Additionally, dose inhomogeneities observed during 3D-RTP session can now be considered and adequate compensations calculated using field-in-field techniques. This is particularly relevant when correcting for the inhomogeneities related to variations in the SSD for different portions of the spine field. Often, potential limitations of the conventional (two-dimensional [2-D]) CSI treatment can be uncovered, visualized, and corrected before commencing treatment of the patient.

The boost phase of medulloblastoma treatment also benefits from 3D-RTP and the use of highly-conformal 3D-CRT or IMRT techniques. These techniques offer a distinct advantage over conventional 2-D or early 3-D techniques in their ability to spare normal, or uninvolved, tissues both in the vicinity and at a distance from the target, thereby minimizing late treatment-related morbidity. As a general principle, the full scope of imaging and 3-D treatment planning techniques should be used to simulate, plan, and treat pediatric patients with medulloblastoma. At the heart of effective treatment planning is the thorough understanding of CSI setup, simulation, and inherent treatment limitations—these are discussed in the following text.

The vast majority of patients receiving CSI are children. Many have neurologic deficits, making it difficult for them to be motionless and cooperative for any length of time. More than any other patients, they require proper stabilization during treatment setup, and often assistive anesthesia. The field arrangement usually consists of opposed lateral brain portals and a posterior spinal axis field, all of which are matched in the region of the cervical spine. This set-up usually requires that patients be placed in the prone position. For most patients, the prone position is less comfortable than the supine, and it is therefore less likely to be consistently maintained throughout the treatment course unless the patient is positioned in a thermoplastic mold or a similar device (e.g., an Alpha Cradle). This is less problematic for cooperative adult patients than with children. Small children requiring anesthesia present a problem because the anesthesiologist requires access to the airway. These patients are usually treated in either a full-body plaster cast with an open facial area or using a customizable immobilization device that supports the forehead and the mandible (Fig. 53-6), with access for anesthesia and airway suction devices.

FIGURE 53-6 • Craniospinal immobilization and setup. A, Head support for prone immobilization with adjustable slider for controlling neck flexion and extension. This device supports the head at two contact points, the forehead and the chin. In our practice, we find it helpful to create a plaster mold of the chin to reduce daily setup uncertainty. B, This is a 31-year-old man with high-risk (disseminated) medulloblastoma immobilized in a prone position for craniospinal computed tomography (CT) simulation. In addition to the prone headrest, a thermoplastic mask is stretched over the occiput and the calvarium to fix the head in the treatment position. In this case, it was necessary to use two spine fields in addition to the cranial fields. Two spinal isocenters, one corresponding to each spinal field, were marked on the patient at midsagittal line (directly overlying the posterior spinous processes, not shown) and on the sides of the patient using lateral laser localizers after spine depth was estimated using the pre-treatment MRI scan or the CT scout image. The isocenter for the cranial field pair was similarly marked.

Craniospinal Irradiation Considerations

When treating the craniospinal axis, it is essential to be aware of potential technique limitations. Different techniques to deliver CSI have been described and variously address some of these known limitations.^51–53

Craniospinal treatment is complicated by the fact that several adjacent fields must be used. Typically, the brain is treated through lateral opposed fields, whereas the spinal axis is treated with one or two posterior spinal fields (depending on the length of the spine). Field matching between the two posterior fields is usually accomplished by calculating the separation between the two fields on the skin surface, the so-called “skin-gap” calculation. It is important to match the adjacent radiation treatment fields at the anterior edge of the spinal cord, or alternatively at the posterior edge of the vertebral column. This approach prevents relative overdose of the spinal cord, but will result in a relative cold spot. Matching the inferior border of the lateral cranial fields with the superior border of the most cephalad spinal field is more complicated, because the cephalad aspect of the posterior spine field is diverging cephalad and the caudal aspect of the brain fields is diverging caudad and posteriorly. The first step is to rotate the collimator of the lateral fields so that its inferior border is parallel to the divergence of the superior aspect of the spinal field (Fig. 53-7). This prevents the superior aspect of the spinal field from exiting into the anterior-superior aspect of the brain fields. In effect, it moves the anterior portion of the lateral brain fields cephalad and out of the path of the spine field. Second, to avoid overlap resulting from inferior divergence of the caudal border of the lateral fields into the cephalad aspect of the posterior spinal field, the foot of the couch is rotated toward the collimator so that the caudal field margins of the two fields become parallel and cross the patient’s neck in a straight line, that is, perpendicular to the patient’s sagittal axis (Fig. 53-8).

FIGURE 53-7 • This figure illustrates the collimator rotation of the lateral cranial fields to match the divergence of the posterior spinal field (or the more cephalad field if two spinal fields are used).

FIGURE 53-8 • Rotation of the couch toward the gantry is necessary to match the caudal margin of the lateral cranial fields with the cephalad margin of the posterior spinal field. The two margins must be parallel and cross the patient’s neck in a straight line perpendicular to the patient’s midsagittal plane.

Field junctions are usually moved two or three times during a course of treatment to spread out the dose inhomogeneities at the junction.⁵⁴ The field junctions are often shifted by adjusting either cranial or caudal portions of the abutting fields using asymmetric jaws.

Although the practice of turning the couch for the lateral cranial fields solves the field-matching problem, depending on the location of the cranial field central axes (or anterior-superior position of the isocenter), it may cause the most inferior aspect of the opposite temporal fossa to be underdosed or missed (Fig. 53-9). This problem is generally avoided in practice by placing the isocenter for the brain fields in the central portion of the brain in the midsagittal plane.

FIGURE 53-9 • This figure illustrates a potential complication of the craniospinal setup. A couch angle on the lateral cranial fields can cause the contralateral temporal lobe to be underdosed. A similar problem can be encountered when attempting to minimize divergence into the contralateral eye by placing the whole-brain isocenter at the level of the lateral canthus during craniospinal irradiation. Although this technique works when only treating cranial fields, the addition of the couch rotation causes the contralateral eye to be projected more cephalad and cannot be shielded without also blocking brain tissue. We generally recommend placing the isocenter in the vicinity of the pituitary fossa.

Another note of caution should be added here, because when the foot of the couch is rotated toward the collimator for treatment of each of the lateral brain fields, the treatment distance (SSD) in the caudal half of the lateral brain field is effectively decreased whereas the cephalad half is increased (Fig. 53-10). The change in SSD becomes larger as the distance from the isocenter increases. The SSD is usually shorter by 1 to 2 cm in the cervical spine, depending on the couch angle and the location of the isocenter of the lateral brain fields. The length by which the distance is decreased, which must be measured at the appropriate distance from the isocenter, can be found by measuring the distance between the sagittal laser line when the couch is set to 0 degrees versus where it is when the couch is rotated to the treatment position. It is important to recognize that this change in SSD is in the same direction for both of the lateral brain fields.

FIGURE 53-10 • The rotation of the couch to accomplish a field match results in decrease of the treatment distance in the region of the posterior fossa and the cervical spine while treatment distance increases in the cephalad portions of the brain. This can lead to dose variations, particularly to an overdose of the cervical spinal cord. For this reason, it is advisable to evaluate and monitor this region during the treatment planning and delivery.

The shorter SSD causes the dose in the cervical spine to increase. The dose here is also higher than in the midbrain because of the smaller thickness of the neck. For this reason, it is important always to calculate and record the cervical spine dose in the daily treatment record, taking into account both the reduced separation and the shorter SSD. The same issues that apply to the cervical spine also apply to the posterior fossa. This area is also moved toward the collimator when the couch is rotated, but this is of lesser consequence than in the cervical spine because it is closer to the isocenter and therefore change in SSD is smaller. The transverse separation is usually larger here than in the cervical spine, thus excessive dose is less likely. These problems can easily be overcome with the use of 3D-RTP systems.

It is difficult to treat the entire brain without having divergence into the eye on the opposite side. The problem is analogous to that described in Fig. 53-9, in which the contralateral temporal lobe may be underdosed. One solution is to set the isocenter, and thereby the central axis of the beam, at the lateral canthus of the eye, thereby eliminating the beam divergence. Although this approach may reduce eye irradiation, this technique generally cannot be used in the setting of CSI. In the setting of CSI, when the couch is rotated to eliminate the caudal beam divergence of the lateral cranial fields, the opposite eye is projected more cephalad and cannot be adequately shielded without also blocking brain tissue. A gantry angle here will not solve the problem because that motion causes both eyes to move posteriorly with respect to the beam. Although the anterior-posterior motion can be changed only by a gantry angle, the cephalad-caudad motion can be changed only by a couch rotation.

The dose along the spinal axis varies because of varying depth and SSD, which usually is most noticeable near the surface and in large fields. An isodose distribution calculated on the patient’s sagittal midline contour provides an estimate of the dose heterogeneity along the spinal axis. Obtaining a sagittal contour and the depth of the spinal cord may not always be possible using a fluoroscopic simulator, but this is usually not a problem when CT simulation is used. It is best to prescribe the dose on the central axis at a depth as determined on a CT simulator, or, if the depth can be determined from diagnostic CT or MRI images, at an average of depths measured at several points along the spinal cord. In our institutional experience, the dose varies along the spinal axis from about 80% to 95%. This variation can be greater in patients who are unable to lie prone without propping up of the chest. In such patients, the SSD, which is shorter in the thoracic spine where the spinal cord is usually shallow, can be reduced by positioning the patient flat with the arms raised above the head. Alternatively, the dose can be calculated and appropriate compensation using field-in-field technique devised when using 3D-RTP systems.

In general, an attempt should be made to keep the patient’s posterior surface horizontal, an isodose distribution should be calculated on a sagittal contour, and the dose gradient within the spinal canal should be considered when the dose prescription is formulated. All fields must be carefully matched to prevent relative overdose or underdose at junctional sites (Fig. 53-11).^54,55

FIGURE 53-11 • Craniospinal irradiation (CSI) plan. The patient, a 31-year-old computer programmer, had a 1-month history of headaches, dizziness, unsteadiness, and also nausea and vomiting on admission. Magnetic resonance imaging revealed a 3.5-cm contrast-enhancing mass centered in the cerebellar vermis, which was totally resected. Pathologic evaluation was consistent with medulloblastoma; cerebrospinal fluid cytology was positive for atypical cells. The patient underwent computed tomography (CT) simulation for CSI. A, Sagittal, coronal, and axial views of the CSI treatment dosimetry. The craniospinal junction was placed in the low cervical spine to minimize diverging dose from the posterior spinal field into the mandible and the oral cavity. A junction shift was performed every 9 Gy until the total spine dose of 36 Gy was achieved. B, A digitally-reconstructed radiograph depicting a kyphoscoliotic patient with a relatively short neck and limited ability to achieve neck extension. A small, left-sided lung block was used to block the lung parenchyma while maintaining spinal field width of 5 cm or larger. In adults, it is not necessary to treat the entirety of the vertebral body, but in children with active growth plates, it is essential to ensure uniform dosimetry across all vertebral bodies.

Ependymomas

Ependymomas occur most frequently in the posterior fossa and are treated with 54 Gy with a local field, achieving relatively good 5-year survival rates.⁵⁶ Less-than-complete surgical resection leads to a very high rate of recurrence, in spite of adjuvant radiotherapy. Most centers treat with local therapy only. There is no proven benefit from chemotherapy in children with ependymoma, although novel agents are now bing investigated at relapse as well as radiosurgery.

Radiation Therapy Guidelines

Successful treatment of localized ependymoma requires maximal safe resection followed by high-dose radiation therapy. Because doses of 54 to 59.4 Gy are required, 3-D conformal radiation techniques, including IMRT, should be used whenever possible to mitigate treatment-related toxicity. Adjuvant radiation treatment should start no later than 30 days after surgical resection, or as soon as adequate postresection healing and recovery takes places (usually no earlier than the third postoperative week). In select patients with residual disease, a referral to a tertiary medical center specialized in treating pediatric patients may be warranted for second surgery to maximally debulk any remaining tumor prior to adjuvant radiation treatment. Radiation, even at high doses, cannot compensate for inadequate resection. Both pre- and postoperative contrast-enhanced MRI and CT scans should be used whenever possible for treatment planning.

The GTV includes the tumor bed and all residual gross disease at the primary site based on the preoperative imaging. Preoperative imaging is useful in defining the tissues initially involved with disease and the postoperative or preirradiation neuroimaging identifies any residual disease and the surgical bed. In majority of cases, the GTV is a collapsed surgical bed. The CTV includes the GTV with a 1.0- to 1.5-cm anatomically-constrained margin meant to treat subclinical microscopic disease. The CTV should be limited by the bony calvarium, tentorium, and falx. A geometric expansion beyond the CTV of 0.3 to 0.5 cm generates the PTV. The choice of PTV margin depends on the expected immobilization, image registration and fusion, and daily variability in patient positioning. The PTV may extend into or beyond the bone, but usually does not extend beyond the surface of the patient.

The dose should be prescribed to the highest isodose surface that encompasses both PTV and CTV with the least inhomogeneity (Fig. 53-12). Ideally, 95% of the PTV should be covered by 95% of the prescription dose; PTV dose should be within +10% and −5% of the prescription. These guidelines aim to minimize target dose heterogeneity and consequently possible associated irreversible neurologic sequelae. Without compromising the target coverage, the treatment plan should aim to minimize dose to the spinal cord, brainstem, optic chiasm, optic nerves, auditory system (cochleae), hypothalamic-pituitary complex, and temporal lobes. In situations in which the critical structure tolerance will be exceeded, a reduction in the target volume (staged cone-down) will be required after the tolerance dose has been reached. This may require that portions of the PTV, CTV, or GTV in or near the region of the critical structure will be underdosed. For example, when the cumulative spinal cord dose has reached 50 Gy, the spinal cord should be excluded from the treatment (or excluded after 54 Gy if disease extends into foramen magnum).

FIGURE 53-12 • Ependymoma 7-field IMRT treatment plan. The patient, a 22-month-old girl with 3-month history of nausea, vomiting, and failure to thrive, was found to have a large (5.6 × 4.9 × 5.3 cm), heterogeneously enhancing mass in the posterior fossa with cystic-appearing areas and extension into the foramen magnum; the mass was subtotally resected and the pathologic examination revealed anaplastic ependymoma. The patient was treated with chemotherapy on a trial aimed to delay definitive radiation. At disease progression, definitive radiation to a dose of 59.4 Gy was initiated. The treatment plan is shown in the three cardinal views.

The recommended total dose for localized ependymoma is 54 to 59.4 Gy using conventional fractionation (1.8 Gy/fraction), given the propensity of these tumors to recur locally. Patients younger than 18 months old with gross total resection of a localized disease may be treated to a total dose of 54 Gy; all other patients, even those younger than 18 months old, with residual disease must be treated to a total dose of 59.4 Gy.

Brainstem Gliomas

Diffuse brainstem gliomas are nearly universally fatal within 3 years of diagnosis.^57,58 Children present abruptly, often with a month or less of weakness, ataxia, and cranial neuropathies, and typically an abducens paresis producing horizontal strabismus. These poor-prognosis, diffuse pontine gliomas are usually treated with radiation only. Although radiotherapy is often performed without biopsy, the influence of surgical intervention or biopsy on this disease (excluding dorsally exophytic lesions) is uncertain. Hyperfractionated radiotherapy of 1.17 Gy twice a day to a total dose of 70.2 Gy has been shown to produce similar survival to that of other treatment schemes, with an acceptable rate of morbidity.⁵⁹ Current clinical trials are testing novel molecular therapeutics, antiangiogenesis agents, and radiosensitizers in this tumor.

Attention should be paid to the exception of dorsally exophytic lesions and other focal brainstem tumors, often pilocytic astrocytomas or sometimes gangliogliomas, which can sometimes be cured by surgery alone. Radiotherapy may be effective against tumor progression or tumors that prove refractory to surgical resection.

Radiation Therapy Guidelines

Brainstem gliomas are currently treated with conventionally-fractionated radiation therapy (1.8 Gy/fraction) to doses of 54 to 55.8 Gy, after many trials demonstrated no benefit to dose escalation, hyperfractionation, or radiosensitization using conventional agents. For delineation of target volumes, MR neuroimaging is mandatory. The MRI sequence that best defines the extent of disease should be used; the T2 sequence is usually the most appropriate to define the GTV. The CTV includes the GTV with a 1.0- to 1.5-cm geometric expansion in all directions. It is, however, permissible to use asymmetric geometric margins. For example, a 1.5-cm expansion may be used in the craniocaudal direction within the brainstem and a 1.0-cm margin may be used in the anterior-posterior direction. The PTV includes the CTV with 0.3- to 0.5-cm margin, depending on the immobilization techniques and the expected daily setup variation. With the exception of dorsally exophytic lesions, in which surgical resection has a role in the management, the majority of patients are treated with radiation shortly after diagnosis.

Three-dimensional conformal treatment techniques, including IMRT, should be used in the treatment of brainstem glioma patients (Fig. 53-13). Two-dimensional techniques, with a midline prescription point, are discouraged, given their inherent limitations to spare uninvolved surrounding anatomic structures and tendency to deposit higher doses within the temporal lobes. More than 95% of the PTV should be covered by the 95% isodose surface (CTV and GTV must be within the prescription volume); the PTV dose should be within +10% and −5% of the prescription to minimize dose heterogeneity within the brainstem. Doses to the spinal cord, brainstem, optic chiasm, optic nerves, auditory system (cochleae), hypothalamic-pituitary complex, and temporal lobes should be monitored, recorded, and maintained within the accepted tolerance limits.

FIGURE 53-13 • Brainstem glioma field IMR treatment plan. The patient, a 6-year, 7-month-old girl, was in her usual state of health until 2 weeks prior to admission when her mother noticed that the patient was having difficulty walking, and occasionally bumping into objects. In addition, the patient had difficulty feeding herself and later speaking. At admission, magnetic resonance imaging (MRI) revealed an expansile circumscribed intra-axial mass centered in the pons measuring 4.0 × 3.4 × 4.2 cm, obliterating the prepontine space and compressing the fourth ventricle posteriorly. The mass had a heterogeneous low T1 and high T2 signal on MRI. Perfusion imaging demonstrated increased cerebral blood volume within the areas of increased T2 signal, consistent with tumor vascularity. The patient was placed on high-dose dexamethasone with some improvement in her symptoms, but then rapidly went on to initiate definitive radiation therapy to a dose of 55.8 Gy in 1.8 Gy daily fractions. The treatment plan is shown in the three cardinal views.

Cerebral and Cerebellar Astrocytomas

The prognosis for astrocytomas is much better for children than for adults, particularly because many of the tumors are low-grade pilocytic (WHO grade I) or diffuse astrocytomas (WHO grade II). Even the rate of diffuse WHO grade II astrocytomas that undergo anaplastic transformation is much less than that in adults.⁶⁰ Thus, the role of radiotherapy following surgery for low-grade astrocytomas is controversial, and, in practice today, is deferred following gross total resection and frequently in subtotally resected low-grade gliomas until there is further evidence of growth.⁶¹ The majority of cerebellar astrocytomas carry a good prognosis when treated with surgery alone and are often of the pilocytic type. Even with recurrent or subtotally resected cerebellar tumors, irradiation is withheld and repeat surgery usually pursued in lieu of radiotherapy.

For diencephalic or chiasmatic astrocytomas at the hypothalamus or chiasm, often minimally amenable to resection, carboplatin and vincristine chemotherapy are used in children presenting before age 10 to stabilize or reduce tumor burden and delay radiotherapy until the child is older and can better tolerate acute and long-term effects of radiation.⁶²

Germ Cell Tumors

Intracranial germ-cell tumors (GCTs) represent 3% to 11% of pediatric and 1% of adult brain tumors.⁶³ Intracranial GCTs compose two distinct histologic groupings: germinomas are the most common histologic subtype, whereas NGGCT represent less than one third of intracranial GCTs and consist of embryonal carcinoma, endodermal sinus (yolk sac) tumor, choriocarcinoma, mature and immature teratoma, and GCT of mixed cellular origin.⁵ Historically, radiation alone, frequently encompassing a large treatment volume, constituted the gold standard of treatment for intracranial GCTs.^64–66 However, the morbidity of radiation therapy in children,^67–72 particularly CSI, prompted many investigators to explore approaches that reduce the volume and dose of radiotherapy while preserving high cure rates.

The development of strategies to reduce the morbidity of radiotherapy has been hampered by an absence of randomized trials comparing treatment approaches. Several large retrospective studies have shed light on persistent controversies in the management of intracranial GCTs. Current literature supports distinct therapeutic approaches for germinomas versus NGGCT. Although many published studies group germinomas with NGGCT, their distinct prognosis points to the need to approach these clinical entities individually.

Published reports corroborate poorer prognosis for patients with NGGCT compared with those with germinomas.^65,73 Radiation alone produces 5-year survival rates as low as 30% to 40%.⁷⁴ More recent studies that demonstrate excellent response rates to chemotherapy have shifted the standard treatment of NGGCT to combined-modality therapy consisting of chemotherapy and radiation, and these approaches are now being explored in North America.⁷⁵ Many or most investigators still consider craniospinal rather than local-field radiotherapy standard in this disease.

Of the various histologic subtypes of intracranial GCTs, germinomas represent the most common and prognostically favorable subgroup. Historically, intracranial germinoma has been treated with radiation alone, producing excellent cure rates.^63,64,66 The significant long-term toxicity of radiation, particularly in children,^67–72 has prompted investigation to explore alternative treatment that minimizes the dose and volume of irradiation. However, these alternative approaches must preserve the high cure rates established with radiation alone.

Several series have reported excellent clinical outcome with preirradiation chemotherapy followed by focal irradiation.^76–81 The appeal of this approach lies in lower radiation dose and smaller radiation fields that presumably translate into reduced long-term toxicity. Radiation alone continues to be the gold standard for the treatment of intracranial germinoma. Promising approaches such as combined chemotherapy and focal, reduced-dose irradiation are being compared with radiation alone in a prospective, randomized trial.

In addition to the role of chemotherapy, a point of controversy in the treatment of localized germinoma has been the appropriate field of radiation.⁸² Current evidence substantiates omitting CSI from the treatment of localized germinoma. Spinal failure rates of less than 10% in the absence of CSI, reported in most contemporary series, do not justify routine incorporation of CSI into the treatment strategy for localized germinoma.^73,83–87 Nevertheless, the precise irradiation field appropriate for localized germinoma remains unclear. Several studies report higher recurrence rates associated with radiation targeting the tumor volume only.^73,83,88,89 We recommend whole-ventricular irradiation, followed by a boost to the primary tumor for localized germinoma when using radiation alone.⁹⁰ The literature supports a dose of greater than or equal to 45 Gy to the primary tumor for germinomas treated with radiation alone. In the context of combined chemotherapy and radiation, many investigators have tailored the dose of radiation to the tumor response observed following chemotherapy. Doses of radiation following chemotherapy range from 24 to 40 Gy.

Radiation Therapy Guidelines

NGGCTs, just like their nonseminoma analogs outside of the CNS, tend to be resistant to most forms of therapy. Many different combined-modality approaches were tested in clinical trials, but most suffered from poor event-free survival and high rates of clinical progression. A recently closed COG protocol (ACNS 0122) aimed to assess the ability of 18 weeks of neoadjuvant chemotherapy (alternating combinations of carboplatin/etoposide and ifosfamide/etoposide every 3 weeks) to eliminate measurable disease prior to radiation therapy (with or without preirradiation surgery). Patients with complete response (CR) or less than CR and second-look surgery went on to receive CSI if they experienced normalization of serum and CSF markers. On this study, patients were treated with conventionally-fractionated CSI (1.8 Gy/fraction) to a total dose of 36 Gy followed by a conformal boost radiation for additional 18 Gy, resulting in the total boost volume dose of 54 Gy.

For CSI, the CTV_csi included the entire craniospinal axis (whole-brain and spinal volumes). To create the PTV_csi, a 0.5-cm margin was added to the brain-only portion of the CTV_csi, and 1-cm lateral and inferior margins were added to the spinal portion. For boost treatments, the GTV_boost was defined as the prechemotherapy volume, including the tumor bed and any residual disease. The CTV_boost included the GTV_boost with a 1-cm anatomically-constrained margin. The PTV_boost was defined as the CTV_boost with 0.3- to 0.5-cm, institution-specific geometric margin to account for immobilization and daily setup uncertainties. To define these volumes, both contrast-enhanced and unenhanced MRI were necessary and MRI and CT registration (fusion) was mandated for conformal planning. The study permitted treatment of multiple intracranial tumors, each of which was defined as a separate GTV volume. The GTV for spinal metastases was also defined as the prechemotherapy volume, the tumor bed, and any residual disease. Spinal disease was treated with a 9 Gy boost after 36 Gy of CSI for a total dose of 45 Gy. Dose heterogeneity limits are similar to those described previously for CSI of medulloblastoma.

In contrast to NGGCTs, the treatment volumes for localized germinoma are controversial. A current COG study (ACNS 0232) in patients with pure germinoma is designed to examine whether the addition of preradiotherapy chemotherapy will permit response-based reduction in radiation volumes and doses to improve functional outcomes while preserving disease control. Because of increasing tendency to treat smaller volumes, the importance of proper staging cannot be overemphasized in the radiotherapeutic management of these patients. The diagnosis of germinoma requires a biopsy, especially in the setting of negative serum or CSF biomarkers, and even if the lesion has a characteristic radiographic appearance. Localized germinoma is treated with whole-ventricular (WV) radiation to 24 Gy followed by a 21 Gy focal boost to the primary tumor. The standard treatment of disseminated disease is CSI to 24 Gy with a 21 Gy involved-field boost, for a total primary site dose of 45 Gy. Unlike the treatment of other pediatric CNS tumors, radiation is delivered using 1.5 Gy daily fractions.

Treatment volumes for disseminated germinoma are analogous to those used for NGGCTs. The GTV is defined as contrast-enhancing tumor on the initial diagnostic MRI (before patient receives any form of therapy). In the unlikely circumstances that the gross disease is nonenhancing, a “mass-like” T2-signal abnormality corresponding to the primary or metastatic tumor should be designated as the GTV. The CTV_boost includes the GTV plus a 1.5-cm, anatomically-constrained margin, limited by the tentorium or calvarium. The PTV_boost is defined as institution-specific 0.3- to 0.5-cm expansion beyond CTV_boost that accounts for immobilization and daily setup variation. Generally, PTV_boost does not extend beyond the subcutaneous tissues of the scalp. In cases of localized germinoma, patients are treated with WV radiation before proceeding with involved-field boost. For WV irradiation, the CTV_wv must encompass the lateral, third, and fourth ventricles as outlined on co-registered (fused) CT and MRI volumes with a 1-cm geometric margin. If a shunt was placed or ventriculostomy performed at the time of diagnosis to relieve hydrocephalus, or if preradiation chemotherapy was administered and resulted in a ventricular volume decrease, the ventricular volume immediately prior to radiation (not at diagnosis) should be used for treatment planning. A 0.5-cm geometric expansion beyond the CTV_wv generates the PTV_wv.

Patients with intracranial, localized germinoma are generally simulated in a supine position with maximal comfortable neck flexion and head-only thermoplastic mask for immobilization. It is critical to ensure that the chin, forehead, and nasal bridge fit snugly inside the mask to minimize patient motion. Whenever possible, WV irradiation should be delivered with photon energies of 6 mV or more and either a four-field plan, consisting of an opposed lateral pair and anterior-to-posterior beams (angled to avoid the orbits), or IMRT (Fig. 53-14). The use of an opposed lateral pair should be avoided because of a relative increase in the temporal lobe dose compared with more conformal multibeam techniques.⁹¹ Similar principles should be followed for the involved-field boost.

FIGURE 53-14 • Germinoma treated with whole-ventricular radiation (7-field IMRT plan). The patient, a 23-year-old man with a 4-month history of headaches and then nausea, vomiting, and unsteadiness, was diagnosed with a pineal region germinoma. Gross total resection was achieved and followed by a whole-ventricular irradiation with a surgical bed boost. The treatment plan is shown in the three cardinal views.

Craniopharyngiomas

These lesions are not malignant but, because of their location in the suprasellar region, are often associated with high morbidity. The judgment and skill of the operating surgeon is therefore critical if significant endocrine and visual problems are to be avoided. There is a suggestion that subtotal resection followed by local irradiation to 54 Gy offers progression-free survival similar to that of gross total resection with a much lower rate of morbidity, but contradictory studies are published as well.⁹²

Atypical Teratoid Rhabdoid Tumor, Meningiomas, and Other Tumors

ATRT remains a notoriously aggressive tumor with propensity for metastasis, and even with chemotherapy and CSI, relapse rates are high. Embryonal tumors, such as pineoblastoma, ependymoblastoma, and supratentorial PNET, are treated similarly to medulloblastoma with chemotherapy and 36 Gy of CSI plus a boost to the tumor volume. Pineoblastomas appear to fare the best of these small-cell tumors.

Meningiomas are rare in children and can behave more aggressively than adult counterparts, although they are treated in a similar fashion, with maximal surgical resection followed by radiotherapy for subtotal resections or anaplastic (WHO grade III) histology. Hemangiopericytomas are like meningiomas, but appear to be more aggressive and are treated to a total dose up to 60 Gy.

Gangliogliomas are usually WHO grade I and rarely require radiotherapy, even when subtotally resected. Likewise, DNT and PXA are usually treated by complete surgical removal and do not require radiotherapy or chemotherapy.