Over the past 10 years there have been significant advances in the appreciation and understanding of the neurocognitive outcomes in pediatric brain tumor survivors. The importance of functional outcomes in pediatric survivors is exemplified by the inclusion of a chapter on neurocognitive considerations in a primarily medical text. The survival rate of children and adolescents with various types of brain tumors has improved markedly,1,2 and 5 to 40% of survivors exhibit functionally disruptive neurocognitive impairments,3 making such considerations imperative relative to quality of life and long-term functional outcomes. In fact, these outcomes have informed and even guided treatment and the development of novel treatment protocols. This chapter reviews research on primary and secondary factors in neurocognitive outcomes over the past decade, and discusses current treatment approaches and future research directions.
Neurocognitive Late Effects
The emergence of neurocognitive difficulties over time in pediatric cancer survivors is referred to as late effects. 4 This term describes the often increasing functional disruptions observed with regard to cognitive processes in survivors with diverse disease and treatment histories. The course of these emerging impairments is unique in that there is a protracted trajectory compared with other acquired neurologic insults (e.g., traumatic brain injury, stroke) in pediatric patients, which often results in immediate neurocognitive changes and often recovery.5 The pattern of decline appears to be age-dependent, such that younger patients show a more immediate decline with attenuation over time, whereas older patients demonstrate a more protracted decline, with evaluable deficits not present until 18 to 24 months posttreatment.6,7
Research has focused on documenting the presence and role of specific neuropsychological impairments on functioning in addition to the documented changes in global intellect. Processing speed, executive functions (including attention), and memory have been shown to be the most vulnerable neurocognitive domains.6–13 The disruption of each of these functions negatively affects learning,14 particularly in children with concurrent seizures requiring medication, and those children with third ventricle or cerebellar tumors.15
Some recent studies have highlighted the lack of baseline data in studies of neurocognitive late effects, citing possible pre-morbid deficits, which may be related to the primary disease.16 It is challenging to establish a “true” baseline level of functioning in these children, as treatment is expedited from the moment of diagnosis. In the majority of cases, either no baseline is established, or the closest estimate of premorbid functioning is gathered after chemotherapy has been initiated and before radiotherapy.
One of the primary neuropathological mechanism associated with these functional deficits appears to relate to secondary vascular insufficiency, with particular effects on cortical white matter in various brain regions.17–20 In patients treated with radiation, approximately half show progressive white matter changes over time,21 which appear to be dose dependent.22 Extreme cases of white matter disease are observed in some patients treated with radiation therapy, manifesting in leukoencephalopathy or cerebral atrophy. Animal models have demonstrated demyelinating processes and necrosis in rats treated with various doses of radiation that are evident as early as 1 month postradiation and intensifying for up to 6 months. Neural stem cell and progenitor cells were specifically affected, which may be related to diminished brain growth and subsequent dysfunction.23 Functionally, white matter volume loss secondary to radiation and chemotherapy has been demonstrated to be related to global intellect.24
Factors Related to Outcome
Maureen Dennis25 proposed a threshold model for children with medical conditions related to functional outcomes that provides a useful framework for understanding the complex interplay of risk factors that relate to outcome in these children ( Fig. 48.1 ). The primary considerations relate to biological risk, development, time, and reserve (cognitive, social, physical). In general, pediatric brain tumors equate to higher biological risk, primarily related to treatment approaches more than the tumor itself (with the exception of supratentorial tumors). In addition, secondary effects of the tumor and associated treatments, including hydrocephalus and seizures (and antiepileptic drugs [AEDs]), further increase the biological risk when present. Development (age at onset), time since onset, and treatment factors enter into the equation, as does a child′s premorbid level of functioning.
Several known primary and secondary factors have been demonstrated to have impact on neurocognitive outcomes in children with tumors of the central nervous system. Primary factors include treatment dynamics (surgery, chemotherapy, radiotherapy) and disease or tumor characteristics. Secondary factors shown to relate to outcome include individual patient characteristics such as age and gender, time that has elapsed since treatment, and neurologic sequelae (e.g., hydrocephalus, seizures).
Surgical resection of tumors is generally not discussed as one of the primary factors associated with outcomes. However, Carpentieri and colleagues26 documented disruptions in verbal memory, visuospatial skills, and motor functions in a group of children treated with surgery exclusively. But this study is limited by the lack of consideration of presurgical functional levels. A postsurgical complication that has been documented in 25 to 33% of patients undergoing resection of medulloblastoma is the cerebellar mutism syndrome (CMS).27 Beyond the transient features of this syndrome (e.g., mutism, irritability/agitation, ataxia, hypotonia), neurocognitive deficits persist, including executive dysfunction, particularly related to attention, cognitive and behavioral flexibility, strategic planning, and initiation.27 Recent research has demonstrated the presence of abnormalities on imaging that are associated with CMS.28 Specifically, children who developed CMS following surgery for medulloblastoma were more likely to exhibit brainstem and cerebellar medullary angle involvement preoperatively. Although not obvious immediately, there was evidence 1 year postoperatively of significantly more atrophy of the total cerebellum, vermis, and brainstem compared with children without CMS. Consequently, the CMS group exhibited greater long-term neurocognitive deficits as well. There was a difference of 15 IQ points between the groups on average, and 60% of the CMS group fell in the impaired range, compared with only 14% in the non-CMS group. These findings hint at a pathological mechanism involving disruption of the dentatothalamocortical outflow pathway, which may relate to the more chronic functional impairments in CMS, namely the neurocognitive and behavioral sequelae. This is a reasonable hypothesis given the recent understanding of the relationship of the cerebellum to higher order cognitive functions, including executive functions, and emotional regulation.29
Assessing the role of chemotherapy in pediatric brain tumor patients is difficult because chemotherapy is often part of a multiagent treatment regimen and because of disease complications, such as hydrocephalus and seizures. Nonetheless, the Head Start II study suggested that young children with malignant brain tumors treated with chemotherapy and autologous hematopoietic cell transplantation (AuHCT) following resection were spared some of the neurocognitive consequences associated with early radiation therapy, at least within the first 2 years following treatment.30 Similarly, children with medulloblastoma treated with chemotherapy alone were shown to function significantly higher than children who received radiation, but still significantly worse than normal controls.31
One study attempting to evaluate the effects of chemotherapy regimens in patients with medulloblastoma found significant neurocognitive susceptibility in patients treated with intrathecal methotrexate (MTX), compared with those treated with intravenous MTX or controls,32 as they performed significantly worse on all neuropsychological measures. However, another small study of children with chiasmatic-hypothalamic tumors reported no significant change in IQ over time.33
The timing of chemotherapy, specifically MTX, has also been shown to relate to outcome, and the effect is pronounced in girls and in the younger cohort. Specifically, the greatest functional impairments are observed in children who received concurrent MTX and radiation therapy versus those who received MTX prior to radiation.34,35 Further, intrathecal MTX is more commonly associated with the development of leukoencephalopathy, which also has impact on neurocognitive functioning.36
Changes in radiation therapy (RT) approaches have been applied over the past decade attempting to minimize the long-term sequelae associated with conventional RT. Reduced doses and volume considerations (focal or conformal approaches) have been the focus of these studies.
Recent research has focused on reduced-dose radiation therapy (RRT), primarily in craniospinal and whole-brain radiotherapy, to find the lowest curative dose37,38 while minimizing functional and quality-of-life disruptions in pediatric survivors. There is a clear dose-dependent relationship with cognitive functions in this population.39–42 Conventional craniospinal radiation dosing (55 Gy) results in a high level of cognitive morbidity, and significant deterioration in global intellect over time, in some cases in excess of 25 IQ points.43–45 Twenty-three to 50 percent of survivors demonstrate IQs less than 80 (≥ 1.5 SD below the mean) at follow-up.46,47 Additional areas of neurocognitive impairment have also been demonstrated to be related to higher radiation doses,48 including processing speed,40,42 attention, verbal reasoning and memory,42 visual perception, language, and academic functioning.41,49 The left temporal lobe is particularly susceptible to greater neurocognitive declines when receiving higher dose volumes to that region,50,51 and higher percentages of total brain and supratentorial regions receiving > 45 Gy have been shown to be related to poorer intellectual outcomes.51
Less radiation-related neurotoxicity has been demonstrated in RRT studies, primarily related to global intellectual outcomes. However, even reduced doses carry some morbidity, and average IQ decrements of four points per year have been documented.52 Comparisons made between craniospinal radiation doses on functional outcomes found greater neurotoxicity at 36 Gy than at 23.4 or 18 Gy. Patients treated with 36 Gy perform an average of 8 to 13 IQ points lower than those receiving 23.4 Gy.37,39 Douw et al′s53 study comparing low-grade glioma patients with or without low-dose radiation therapy revealed significantly poorer executive functioning and processing speed in the radiated group, whereas core memory and motor functions did not appear to be affected in this sample. Neuroanatomically, acceleration of white matter volume deterioration has also been linked to doses of 36 Gy even compared with 23.4 Gy.19
Beyond differences in RRT, comparisons among craniospinal RT (CSRT), whole brain, and focal radiation reveal differential outcomes in intelligence and other neurocognitive domains, primarily with regard to severity rather than presence or absence. In contrast to the 25-point decline in IQ documented in the groups of patients requiring CSRT/whole-brain doses, an average 6- to 11-point decline in IQ has been demonstrated in partial brain radiation.54 Jalali and colleagues50 reported a > 10% diminishment in IQ in a third of patients in their study 2 years after focal radiation (the patients’ diagnosis was primarily craniopharyngioma).50 Compared with children receiving focal posterior fossa radiation only, craniospinal doses of 36 and 24 Gy resulted in 21- and 13-point differences in IQ, respectively.39 Other neurocognitive domains are also impacted by focal approaches, including processing speed, attention, auditory and visual learning/memory, academic skills, adaptive functions, and behavior.54–56 This again highlights the idea that while the severity of neurocognitive impairments may be diminished with focal radiation, there remain a notable number of neurocognitive sequelae over time ( Table 48.1 ).
Several factors mediate the relationship between radiation and neurocognitive outcomes. Age has been shown to account for more variance than radiation, and the factors of age and radiation combined represent an additive effect on outcome.50 These effects are particularly salient for those patients diagnosed and treated at younger ages,41 and it has been suggested that long-term IQ can be predicted by baseline intellect, radiation dose, and age at time of radiotherapy.37 There is currently no conclusive evidence that reduced doses result in better outcomes in older children.57,58
One of the more exciting developments in radiotherapy is the increasing knowledge and availability of proton beam radiation therapy as opposed to traditional photon radiotherapy.59,60 The potential benefits of proton beam therapy include increased accuracy and the sparing of healthy tissue during treatment.61 This is an important advance, as the inadvertent involvement of healthy tissue even in the most advanced conformal techniques has been shown to relate to the emergence of neurocognitive and functional deficits.62,63 Further, this new technology may allow for radiation therapy in the youngest patients, potentially increasing survival rates, while limiting the disruption in neuro-cognitive development. Although the potential benefits of proton beam radiation on neurocognitive outcomes has yet to be fully demonstrated, it is expected that minimizing exposure to healthy tissue may positively impact outcomes.
Molecular genetics is becoming an increasingly important focus of research in the area of long-term functional outcomes in children treated with radiation. There is some evidence that specific genetic polymorphisms (GSTM1 and GSTT1) are related to poorer neurologic and neurocognitive outcomes in medulloblastoma survivors.64 Patients with the most significant postradiation sequelae have been shown to have more than four single nucleotide polymorphisms (SNPs) in candidate genes.65 Further, Svensson and colleagues66 documented crude differences in gene expression profiles between patients with and patients without severe adverse effects from radiotherapy over time.