Radiation Therapy for Medulloblastoma

Abstract

Radiation therapy remains a cornerstone in the clinical care of patients with medulloblastoma. Nevertheless, while treatment with radiation improves disease control, it is also associated with potential late effects impacting neurologic and neurocognitive function, pituitary function as well as hearing. The development of secondary, treatment-related malignancies is an uncommon but possible late outcome. In this chapter, we review the role of radiation therapy in medulloblastoma, as well as changes in management that have resulted both from technological therapeutic advancements and enterprising cooperative group clinical trials for this disease. Moreover, with increasing utilization of molecular diagnostics both for prognosis and stratification of treatment, we also endeavor to explore opportunities to further improve the delivery of radiation therapy to patients with varied risk of disease recurrence.

1. Introduction

Radiation therapy has played an important role in the management of patients with medulloblastoma. Historically, it was well-recognized from autopsy studies that medulloblastoma exhibited a propensity to spread through the ventricular pathways of the central nervous system (CNS). Investigators therefore acknowledged that radiation treatment volumes would necessarily need to include the entire neuroaxis and meningeal surface as part of a composite care plan, with treatment delivered to all potential sites of disease in a continuous manner on a daily basis. Likewise, because the tumor had access to fluid pathways at all time points of management, radiation oncology plans were designed to treat the meningeal surface during each treatment session. To these ends, additional dosimetric and treatment delivery strategies were employed to match radiation treatment fields while minimizing potential overlap of dose in critical normal tissue, the most significant of these being the spinal cord. These strategies generated success in tumor control. However, the epidemiology of medulloblastoma is such that more than 40% of patients affected with this disease are identified before the age of 5 and approximately 70% of patients affected by the disease are identified before the age of 10, a critical period of anatomical development in which important structures in the CNS, vertebral bodies, and other vulnerable normal tissue have not yet reached full maturity. Moreover, multiple additional normal tissues received collateral exit doses from spinal treatment fields, including but not limited to the posterior pharyngeal wall, thyroid, cardiopulmonary structures, small bowel, kidneys, and soft tissue pelvic structures. The vertebral bodies and pelvis bony structures were likewise at risk for injury and titration of growth and development, again visibly affecting the most vulnerable and youngest of the patients relative to both standing and sitting height. Modern clinical protocols work to address these issues through comprehensive review of outcomes and adjustments in therapy to attempt to mitigate these potential side effects. Although titration adjustments in the posterior fossa boost volume have been adopted without concomitant compromise in disease control, subsequent attempts to titrate both volume and dose of the entire neuroaxis have been decidedly less successful. Therefore, considerable work needs to be completed through modern risk categorization to determine if there are distinct medulloblastoma patient subpopulations—that is, those with otherwise more favorable disease biology—where radiation dose volume titration may be reasonable to revisit and study through investigatory clinical trials. Conversely, it is equally important to identify potential high-risk population cohorts who may stand to benefit from augmented or intensified therapy. Interdigitation with chemotherapy brings both strength and consequence as well as strategies to make certain that patients are given the benefits of both therapies without unintended risk of injury [1, 2, 3, 4, 5, 6].

In this chapter, we review the history of radiation therapy in the treatment of medulloblastoma aligned with the development of modern technology to optimize tumor control with decreased risk. We also review the future role of radiation therapy with modern classification of pathology and biomarker expression products in consideration of additional adjustments that may be made in treatment technique and patient stratification moving forward.

2. History of radiation therapy in medulloblastoma and the development of treatment technique

In 1930, the Neurosurgeon Harvey Cushing first reported surgical resection of medulloblastoma located in the posterior fossa and identified the challenges and limitations of favorable patient outcomes due to leptomeningeal spread of disease throughout the neuroaxis, which limited the success of primary surgery. Recognizing the need to treat the entire meningeal surface, Edith Paterson and RF Farr published a manuscript from Christie hospital and the Holt Radium Institute (Manchester, England) using orthovoltage therapy to treat the entire neuroaxis. These investigators reported on 27 patients treated from 1941 to 1950 with pathology validated by a single pathologist. The authors confirmed that care was taken to include the entire CNS and skull base in the therapy field with the inferior limit of the spine field at the second sacral vertebra corresponding to thecal sac termination. The patients were treated with a single posterior field, including the entire CNS, with patients lying in the prone position. Lead was used to limit the width of the spine fields. The investigators found an advantage in tilting the spine field at 20 degrees toward the head to achieve a more uniform dose distribution through the treatment field. The anterior component of the cranium had dose augmented by supplementary fields as it was recognized that a single posterior field of radiation would be insufficient to cover intracranial contents. The patients were treated AP-PA with shielding applied to the posterior field to accommodate dose received in that location from the initial posterior field. Optimal dose was thought to be 3500 rads over 5 weeks. Results from this investigation revealed a 65% survival rate for the entire population at 5 years, with a 54% survival for children only. Observations at 3 years post-treatment included one child with neurocognitive limitations (which also pre-dated therapy) and shortening of sitting height in children treated at ages below the age of 10 [7, 8].

As linear accelerator technology became available in both Europe and North America, a more comprehensive understanding of the geometry of radiation fields became visible and cranial-spinal radiation therapy techniques matured with evolving technology. Because radiation beams generated from linear accelerators expanded at a distance in a pyramidal distribution, treatment planning strategies were developed to accommodate this geometry. Simply abutting the cranium and spine fields would result in overlap over the spinal cord, providing a greater dose than prescribed and an increased risk for permanent neurologic injury. Patients were treated in the prone or supine position with prone treatments more commonly preferred by radiation oncologists for reproducibility of a match line. Supine treatment, however, was preferred by pediatric anesthesiologists to allow airway protection in younger children who required anesthesia for daily treatment. With the cranium treated with lateral fields, the table would be rotated to create a non-divergent field edge at the inferior field border in order to carefully match the spine field to the head field. The collimator of the cranial field was angled to match the divergence angle of the spine field with a separation of 1 cm in order to create a “cold” match point over the cervical spine (Figure 1). The match point would be moved each day with three separate positions. This technique was referred to as “feathering” in order to make certain that the “cold” under-dosed region was not always in the same anatomic area. Demonstrating department and investigator expertise in this and similar techniques was a benchmark requirement for participation in clinical trials for the Children’s Oncology Group (COG). Clinical and computational expertise was needed to ensure patients were treated in a uniform manner on study in order for the results of the study to be trusted and become incorporated into daily clinical management. These early investigator treatments were designed and delivered as two-dimensional field plans; however, with the emergence of volumetric and three-dimensional radiation therapy treatment planning tools, treatment execution would influence the approach to radiation therapy field design and drive planning toward disease associated and volumetric tumor targets [9, 10, 11, 12].

Figure 1.

Match fields over the cervical spine with the angle of the collimator of the cranial fields matching the divergence angle of the spine field. Field overlap in the lumbar region was deep to the cord ensuring absence of “hot spots” within this critical structure.

One of the major challenges imposed by historical approaches to cranial-spinal radiation therapy is that the target of requisite CNS anatomy, including the entire meningeal surface, cribriform plate, skull base, and thecal sac—was covered by default, standard fields designated solely by two-dimensional planning. Often, deviations in medulloblastoma clinical protocols during this era were driven by incomplete coverage of these structures using two-dimensional geometry. Over-blocking of the orbits and skull base was generally the root cause of the study deviation and investigators could point toward recurrences of disease associated with incomplete anatomical coverage of targets. Boost strategies were also designed by clinical anatomical boundaries with the posterior fossa field design directed to the anterior clinoid, top of the tentorium, and internal occipital protuberance as targets (Figure 2). This generated increased consistency and treatment uniformity among investigators in radiation therapy field design; however, specific questions about dose titration to anatomical structures unintentionally treated including visual pathways, hippocampus, and cochlea could not be addressed through this mechanism as the dose received by normal tissue was equal to the prescription dose to the disease. As sequelae of management became more visible in outcome analysis, clinical trialists became interested in understanding quantitative volumetric dose. They investigated if conformal avoidance of normal tissue and dose volume titration of these structures could lessen toxicity [13, 14, 15].

Figure 2.

Traditional boost fields generating prescription doses to the cochlea region.

3. Volumetric-based radiation oncology treatment planning for medulloblastoma

Treatment planning for radiation therapy moved from fluoroscopy to computed tomography (CT) with the advent of three-dimensional planning and forever changed the landscape for treatment planning in this disease. Individual structures within the CNS and spinal axis could be imaged, identified contoured as individual objects, with dose calculated to the volume of normal tissue. Investigators could now evaluate the relationship of dose to the clinical target and critical normal tissue in a developmental manner. For example, the cribriform plate resides in the same axial plane as the lens in children up until the approximate age of 9 years. At this time, the frontal sinus begins to mature, and by the age of 12, the cribriform plate is generally 1.2 cm superior and rostral to the lens. Therefore, if one over-blocked the orbit, the cribriform region would also be unintentionally blocked. Volumetric planning with CT reduced the likelihood of this error. In early clinical trials using volumetric imaging for the posterior fossa component of therapy, investigators could now visualize limitations in anatomically applied radiation therapy fields. Lateral fields would unintentionally include the cochlea to high dose, posterior fields would exit into the oral cavity, and vertex fields would exit into the thyroid delivering dose that would be unknowingly remarkably high in young children as the distance to the thyroid gland in a young child is much less than it is in an adult when viewed through the lens of a vertex field. Improving radiation field placement in these situations with field volume titration could be partially optimized through this process by extracting the best of each field while minimizing the negative consequence of each field. This process also established the importance of imaging in the evaluation of medulloblastoma.

Magnetic resonance imaging (MRI) with contrast has become the standard of care in this disease for both the spine and CNS. Staging, target definition, and follow-up care are all now influenced by this standard of care imaging; the presence and absence of residual disease in the posterior fossa, presence of spinal metastasis, location of the thecal sac, and oncologic treatment outcomes are all measured by imaging. Our neuroradiology colleagues are now integral members of the multidisciplinary treatment team.

Medulloblastoma protocol A9961 remains an excellent example of the importance of imaging and peer review of imaging on the study. Often, studies will identify criteria for measuring the presence or absence of residual disease and spinal metastasis, which may/may not coincide with specific institutional criteria. During a retrospective review of imaging on protocol A9961, approximately 10% of patients enrolled into the study were actually not study-eligible due to greater than 1.5 cm² residual tumor in the posterior fossa, or presence of spinal disease otherwise not identified at pre-study entry. An additional group of patients was ineligible due to the presence of motion artifact on submitted imaging, confounding the process of tumor quantification and disease assessment by designated study imaging reviewers. For this study, 189 institutions were needed to reach the accrual objective of 421 patients, therefore, the 10% loss in patient population was an unforeseen detriment for the assessment of this study outcome. Therefore, the quality of the imaging and its interpretation is an essential component of management of a successful study and the importance of imaging for these study patients cannot be overstated. Nevertheless, the protocol experience provided insight into how best to manage quality assurance of both imaging and radiation therapy in medulloblastoma. Additional issues with protocol execution were also revelatory and generative in the following manner: the protocol was an intergroup trial between the two pediatric clinical trial groups—the Children’s Cancer Study Group (CCSG) and the Pediatric oncology Group (POG). CCSG, however, did not participate in pre-therapy review of treatment objects. As a result, there were multiple deviations on study, generally related to volume limitations at the level of the cribriform plate, skull base, and posterior fossa, thus limiting commensurate comparison of all patients as a singular cohort. This highlights one of the many challenges in clinical trial development and implementation: to create a uniform study population for the results of the trial to be trusted and translated directly into clinical practice [16, 17, 18, 19, 20].

By limiting study deviations with improvement in the rapid digital transfer of imaging and radiation therapy treatment objects to quality assurance centers, important questions in this disease could be answered with confidence that the quality and consistency of complex radiation delivery would not hinder the interpretation of the study. COG protocol ACNS0331 asked two important study questions related to radiation therapy. The central question was to evaluate whether the final phase of management had to be tailored to the anatomical configuration of the posterior fossa as historically described and performed. This study randomized the radiation posterior fossa boost to either historical anatomically defined fields or a more limited area of imaging-defined disease and the surgical cavity (involved field targeting). This change, if found valid, would, in turn, spare considerable normal tissue including the cochlea and temporal lobes from unnecessary radiation and late effects. ANCS0331 had quality assurance processes embedded in the trial to review imaging and radiation therapy treatment objects in real time to ensure the objects were of sufficient quality for study compliance and that the individual radiation therapy plans met protocol standards. The study successfully established involved field image guided targeting as the standard of care in these average risk patients for the final phase of management. The study also asked a question in patients 3–7 years of age as to whether or not the radiation dose in the first phase of craniospinal irradiation could be reduced to 18 Gy. This proved to be unsuccessful for the general population of medulloblastoma patients treated on this study. However, because of favorable outcomes for the 10% of medulloblastoma patients expressing Wingless-related integration site (WNT) biomarkers, an opportunity arose to revisit radiation dose titration and chemotherapy reduction in this distinct molecular subgroup of patients [21, 22, 23, 24].

Newer protocols exclusively treating standard risk patients with WNT expression have embraced this task. Nevertheless, this is a challenging endeavor because the otherwise apparently favorable WNT biomarker is expressed in only around 10% of the entire medulloblastoma patient population; therefore, to effectively power the study, participation from centers worldwide is needed to meet accrual objectives. Because of the limited study population, the quality of data from each patient also needs to be complete and meet study objectives, as any deviation or limitation in data acquisition would hinder the goals and objectives of the study analysis. This likewise requires that the quality assurance centers have data management processes in situ that can respond to sites in real time, generally on a same-day basis for clinical trial management. The study will acquire valuable data to see if radiation and chemotherapy dose titration can be applied to a molecularly defined and favorable population. Because of nimble and effective data acquisition and transfer tools, prospective trials on limited subsets of patients can now be achieved with trusted outcomes [21].

4. The importance of imaging and how modern MRI signals influence staging

CNS MRI is the imaging modality of choice for the diagnosis of medulloblastoma. It aids in differentiating medulloblastoma from other pediatric CNS malignancies in addition to completing staging. Medulloblastoma classically presents on MRI as a midline posterior fossa tumor. The typical presentation of the tumor is hypo- or iso-intense on T1 images, heterogeneous or hypointense on T2, and enhancing on post-contrast T1 images [25, 26]. Diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) can aid in further differentiating medulloblastoma from other CNS tumors such as ependymomas and pilocytic astrocytomas [26, 27]. Risk stratification of medulloblastoma includes histological assessment, Chang staging, age, and amount of disease remaining post-operatively. MRI determines two of four risk stratification criteria. Chang staging is categorized as M0 through M4 and historically relied on CT and craniospinal fluid analysis. However, CT alone is typically unable to distinguish M0 from M2/3 disease (M2 is intracranial seeding, M3 is seeding in subarachnoid space), a distinction that could change a patient’s risk stratification from low to high. MRI has replaced CT as its detailed diagnostics allow for the appreciation of gross seeding in the CNS [28]. In addition, post-operative brain MRI at 24–48 hours provides assessment of extent of resection and residual disease burden [27]. Greater than 1.5 cc² of residual disease is one of the components of high-risk disease [26, 29, 30]. Appropriate risk stratification is imperative to adjuvant treatment recommendations and therefore patient outcome.

Imaging has become an important adjudicator for both site and study investigators. Review of imaging in real time not only ensures correct staging but also allows for patients to be properly treated per protocol or randomized to the appropriate protocol. Because of real-time data management tools, trial investigators and radiologists worldwide have access to images on a same-day basis and can participate in management decisions in a timely manner. Investigators can in turn achieve consistency in the interpretation of both disease response and progression as well as making certain that patients assigned standard risk status by institutions are placed into the correct study. Imaging is essential to modern clinical trial management and is highly important to the quality assurance process for each trial.

5. The importance of molecular classification

In addition to clinically and imaging-defined risk stratification, molecular subtyping of medulloblastoma informs outcomes. Currently, there are four molecular subtypes of medulloblastoma: Group 3, Group 4, Sonic Hedgehog (SHH), and WNT [31, 32]. Although the determination of subtype is not currently incorporated into adjuvant treatment guidelines, there are groups actively investigating this question by integrating clinical and molecular risk-directed therapy approaches [31, 32]. Given that there can be limited accessibility to molecular testing, the interpretation of MRI for diagnosis of subtypes is also under investigation. One method to identify subtypes of medulloblastoma is based on generalized patterns of neuro-anatomic location. For example, Group 3/Group 4 tumors tend to arise in the 4th ventricle, whereas WNT tumors tend to arise within the Foramen of Luschka or adjacent cerebrospinal fluid spaces. However, it is important to remember that these correlations are not absolute. With advanced MRI sequences, further delineation of subtypes is becoming apparent based on the appearance on MRI images. Radiomics and machine learning algorithms are currently being used on presurgical MRIs to predict medulloblastoma subtypes. Zhang et al. found that T1-correlation is greater for SHH and lower for WNT. Reddy et al. found that Group 3/4 tumors showed the highest median ADC values in the enhancing sold tumor, in contrast with WNT tumors which had the lowest median ADC values. [33] MRI correlations to molecular subtypes will be further refined as protocol data matures and can be correlated to outcome imaging [28, 33, 34].

6. The importance of radiation therapy and quality assurance

The development of protocols in medulloblastoma foreshadowed the importance of imaging and quality assurance in the execution of radiation therapy in this disease. Investigators have recognized that study deviations in medulloblastoma treatment have the potential to influence individual patient outcomes. After deviations were identified in protocol A9961, study groups and investigators made efforts to build consensus in contouring objects in a standardized manner and create a uniform study group. Uniform study sets and patients with treatment executed in a protocol-compliant format become invaluable to investigators to study additional questions not anticipated at the time of trial design. Models can be developed to evaluate opportunities for improved patient care and generate metrics for toxicity evaluation during and after the study is completed [20, 22, 23, 24, 35, 36, 37, 38, 39].

Protocols also call attention to important steps forward for the management of medulloblastoma patients. Since the Paterson and Farr paper in 1953, radiation oncologists used the bony anatomy of the posterior fossa for the boost volume. This had convenience as the top of the tentorium was remarkably consistent at 7.5 cm from the top of the skull in an age-transparent fashion. Early radiation oncologists would place their thumb at the foramen magnum and extend the third finger to the top of the skull and the top of the tentorium would reside two-thirds distance from the foramen magnum. The posterior clinoid and the internal occipital protuberance would be the landmarks used to draw the “tent”. The A9961 study evaluated the application of an image-guided boost directed to the surgical cavity as a point of randomization and the study demonstrated that volume titration was non-inferior to historical anatomical coverage. This was a major step forward for radiation oncology as we could now apply image-guided targeting to this disease and limit radiation dose to unintentional targets including but not limited to the cochlea and anterior brain stem in the proximity of the basilar artery. Because this was managed in a protocol setting, our discipline is confident in the outcome of the study, and this strategy for final phase management to an image-guided target sparing additional normal tissue has now been readily applied into standard clinical practice (Figure 3) [21].

Figure 3.

Axial geometry of image guided boost using volumetric arcs generating minimal dose to the cochlea.

The A9961 study also placed emphasis on developing consensus contouring anatomy for therapy. With fluoroscopic imaging, the cribriform plate and meningeal surface are at times challenging to visualize. If one contours what is seen as the brain/CNS, the egress points of cranial nerve exit through the skull will be under-contoured, and if not recognized, the meningeal surface would be unintentionally under-treated (Figures 4 and 5). When changes in practice habits occur and are subsequently embedded into protocols, it takes time for the global radiation oncology community to adapt to changes in protocol. This was especially true in all disease areas as we transitioned to volumetric treatment planning using three-dimensional images to replace fluoroscopy. Over time, investigators have been successful in transitioning radiation therapy treatment planning processes into volumetric targeting, which is now exclusively embedded into the National Clinical Trials Network portfolio of clinical studies.

Figure 4.

Meningeal surface dose when targets are contoured through the clivus and egress points of the cranial nerves.

Figure 5.

Compromised coverage of the cranial nerves when the clivus is not fully contoured.

As technology of radiation therapy has progressively advanced, patients have benefitted from accompanying process improvements and favorable treatment outcomes. The incorporation of image guidance into modern radiation therapy care has afforded a greater degree of daily treatment reproducibility, generating confidence in decreasing expanded targets including the planning target volume of radiation therapy treatment fields, which decreases radiation dose to normal tissue. As proton therapy centers increase in number across the world, an increasing number of medulloblastoma patients are being treated with particle therapy. Protons do not have an exit dose, therefore, treating a spine field with protons eliminates unintended dose to cardiac, pulmonary, and additional normal tissue. Protons also permit decreased dose to growth segments of bony structures, allowing for less consequential impact on normal physiologic processes of growth and development. A comparison of dose to structures between protons and photons is presented in Figures 6 and 7 [40, 41, 42, 43, 44, 45].

Figure 6.

Radiation dose to the spine with photon fields. Note the dose to central lung and cardiac structures.

Figure 7.

Radiation fields to the spine with proton therapy. Note the decreased dose to lung and cardiac structures compared to the photon plan.

7. Conclusions

Radiation therapy has played and continues to play an important role in the care of patients with medulloblastoma. Clinical trials have supported how best to optimize systemic therapy in this disease and will serve to determine how radiation dose and volume titration or augmentation can be performed to further enhance cure while decreasing normal tissue late treatment effects. Current technology permits selected areas of volume titration, and as functional imaging becomes incorporated into clinical trials, selected areas in the central nervous system can be contoured and treated with dose gradients further sparing normal tissue in vulnerable patients. Novel radiation delivery techniques including proton and particle therapies could also significantly improve the therapeutic ratio. The expansion of biomarker-driven therapies, coupled with the selected use of normal tissue-sparing radiation therapy techniques, offers considerable promise in further patient treatment stratification that stands to improve outcomes for patients with this disease therapeutic ratio [46, 47, 48, 49, 50, 51, 52, 53, 54].

Acknowledgments

This work was supported in part by the NCI Grant to the Imaging and Radiation Oncology Core, CA180803.

Conflict of interest

The authors declare no conflict of interest.