Limited Research Investigating the Added Value of MRI in Predicting Future Cognitive Morbidity in Survivors of Paediatric Brain Tumours: A Call to Action for Clinical Neuroimaging Researchers =============================================================================================================================================================================================== * D. Griffiths-King * C. Delivett * A. Peet * J. Waite * J. Novak ## Abstract Survivors of pediatric brain tumour patients are at high risk of cognitive morbidity. There is clinical benefit in being able to reliably predict, at the individual patient level, whether a patient will experience these difficulties or not, the degree of impairment, and the domains affected. Whilst established risk factors exist, quantitative analysis of MRI could provide added predictive value towards this goal, above and beyond existing clinical risk models. The current systematic review aims to answer the question “Do MRI markers predict future cognitive functioning in pediatric brain tumour survivors?”. Studies of pediatric brain tumour patients which test the value of MRI variables in predicting later neuropsychological outcomes were searched up to July 2022. Only included were studies where MRI scans were acquired at an earlier timepoint and used to predict a child’s performance on cognitive tests at a later timepoint. Surprisingly few studies were identified by the search process which specifically investigated MRI measures of cerebellar and white matter damage as features in predicting cognitive outcomes. However, the important finding of this review is that the current literature is limited and those identified had small sample sizes and were rated as poor quality for the purposes of prediction. Overall, current findings are at high risk of bias and thus the quality and impact of conclusions are limited. Given the significant impact for this clinical population that predictive models would enable, the current review affirms the need for a ‘call to action’ for medical imaging researchers in pediatric neuro-oncology. Keywords * MRI * neuropsychology outcomes * prediction * Brain Tumour * children ## Introduction ### Individual Outcomes for Childhood Brain Tumour Patients Survival from cancer in childhood has seen great improvement in recent decades [1]. Consequently, there is a large and increasing population of adult survivors of childhood cancers [1, 2], with approximately 1 in 530 young adults between the ages of 20 and 39 being a survivor of childhood cancer [3]. This is especially true in pediatric brain tumours, the most common solid tumours in children (roughly 20%) [4], where survival is now estimated at around 95% for cerebellar pilocytic astrocytoma, and 60-80% for medulloblastoma [5-8]. Thus, there is an ever increasing need to focus on ensuring quality of life for the future of these children. Many children with brain tumours experience neurocognitive effects at some point in their disease course, resulting in dysfunction in domains of cognition, emotion, and behaviour. The estimated risk for children with brain tumours of having emotional, psychosocial, and attention problems are 15%, 12% and 12% respectively, according to a recent meta-analysis [9]. Even at 10-year survival, these patients still demonstrate neuropsychological and psychosocial impairment across multiple domains [10]. Performance over time demonstrates an inability to acquire new skills and cognitive abilities at the same rate as healthy peers of the same age, rather than a loss of previously acquired abilities [11]. This may explain why these difficulties are likely to persist long-term and are non-transient. The number of post-cancer life-years is much greater for pediatric rather than adult survivors, and these years include numerous important milestones such as education and interpersonal relationship development [12]. Long-term difficulties are likely to profoundly affect participation for these children, at home, school and later in the workplace, likely resulting in poorer long-term educational and employment outcomes [13, 14]. This results in persistent burden for families and healthcare systems [15]. Whilst survival must always be the utmost priority, research aimed at limiting this cognitive morbidity in this group is now needed to ensure likelihood of reaching their potential, despite their illness [16]. Whilst disease and treatment will inevitably place all pediatric brain tumour patients at some level of risk for poor cognitive outcomes, knowing individualised risk at the patient level, an estimate of the severity of difficulties they will face, and the specific domains that are likely to be impacted, will influence clinical practice. There is significant variability in outcomes at the individual patient-level, but this is currently understudied [15]. Person-centred analytical approaches across a large longitudinal sample of paediatric brain tumour patients, show distinct classes / phenotypes with unique profiles in social, cognitive, and attentional difficulties over time [17]. Similar subgroups have been identified across cross-sectional neuropsychological data in pediatric brain tumour patients [18]. Further research is needed to translate this into further individualised models of risk and resilience, for use in clinical settings. ### Clinical Benefits from Prediction of Cognitive Outcomes Prediction of individual-level neurocognitive outcomes would enable timely and tailored input from school and allied health services, promoting outcomes for these children, with limited healthcare resources being efficiently prioritised for those most at risk. It would also help healthcare professionals prepare patients for these difficulties and help reduce uncertainty over the future for families. Children who are undergoing treatment, would also clinically benefit from predictive research for cognitive morbidities. In children where treatment of their brain tumour is more difficult, adjuvant therapy may include radiotherapy, which is known to have significant impact on a child’s neurodevelopment. This is due to brain injury from the primary and secondary effects of radiotherapy, especially in paediatrics where there is specific vulnerability (e.g., due to younger children not yet having reached peak myelin maturity) [19, 20]. Whilst developments in treatment have mitigated some risk of neurocognitive toxicity (e.g. proton beam radiotherapy [21]), there is still need for clinicians to make decisions about treatment strategy to further reduce risk to a child’s neurocognitive development, using existing disease and age related risk factors [19, 20]. More accurate prediction of individual-level risk of cognitive morbidity (even across domain and severity), would enable clinicians to further adapt and personalise treatment schedules with a greater focus on risk to quality-of-life whilst maintaining efficacy in treating disease [22]. Overall, there is clinical benefit for a range of patients in knowing individualised prediction of neurocognitive outcomes. ### Predicting Cognitive Outcomes There are many established risk factors for poor long-term neuropsychological outcome that need to be understood to provide a comprehensive risk profile at the individual child level [23]. Recent neurodevelopmental models based on known risk factors have been proposed to explain outcomes for brain tumour survivors, specifically in medulloblastoma [24-26], taking into consideration the complex disease-, treatment- and host-related factors that may influence these outcomes. Many aspects can result in neurodevelopmental insults to the developing brain which may explain and underpin these neurobehavioral morbidities [27, 28] and thus are significant risk factors for these poor outcomes [23]. These range from physical factors such as treatment effects (i.e. resection and/or adjuvant therapy [19, 29]) and individual differences (i.e. age at diagnosis [29], cognitive reserve), but also psychological factors (i.e. Early Childhood Adversity, threat exposure) and environmental factors (i.e. Socioeconomic Status (SES) and social support) [27]. See [23] for a model of cognitive risk in pediatric brain tumour survivors. Essentially, neurocognitive outcomes are complex and are dependent on several interacting factors [12]. Risk-based and exposure-related guidelines and models have been developed to manage these neurocognitive late-effects of pediatric brain tumours [23, 30]. Neurobehavioral morbidities are predicted by clinical variables such as radiotherapy, chemotherapy, neurosurgery, and parental education but less-so age at diagnosis, gender, or time since diagnosis [12, 13, 19, 31-34]. The Neurological Predictor Scale (NPS) was designed as an ordinal scale to capture the cumulative effect of some of these risk factors on outcomes, and somewhat predicts IQ, adaptive functioning and processing speed and working memory, at both short- and long-term follow-up [33, 35-38]. This cumulative measure captures unique variance, above and beyond the individual predictors. ### MRI as a Novel Predictor of Outcomes MRI measures are likely to be a good proxy of the burden of brain tumours and their treatment thus, are likely to be predictive of cognitive impairment at the individual patient-level. Qualitative reporting of MRI does in fact predict outcomes, with brainstem invasion, midline location of the tumour, and tumour type predicting post-operative cerebellar mutism syndrome, a (typically) transient, neurological morbidity seen in this population [39]. Quantitative alterations to the brain’s structure and function, specifically microstructural changes to the white matter (WM) of the brain, during the developmental period, could be the common neuroanatomical substrate of these outcomes [24, 26]. See [40] and [22] for a review of MRI in pediatric brain tumours. Recent successes and interest in using MRI to predict neurodevelopmental outcomes in premature infants [41], or even decline in neurocognitive functioning in older adults [42] highlights the potential opportunities offered by MRI in this space. Therefore, magnetic resonance imaging (MRI) is likely to provide highly relevant features which provide ‘added-value’ in predicting outcomes beyond clinical risk factors alone. A key consideration, however, is the timing of the MRI used to predict these outcomes. The MRI used for prediction needs to be early enough in the disease course and non-contemporaneous from the later, outcome of interest. There is currently no consensus on the optimal timing of MRI with which to make such a prediction. This systematic review specifically investigates existing literature using MRI scans, taken at any point in the disease course, to predict non-contemporaneous, later neuropsychological outcomes in survivors of pediatric brain tumours. Whilst there is existing literature of existing established clinical predictors of cognitive late effects in this population, this review aims to assess studies using MRI as a predictive modality, with the goal of assessing whether quantitative analysis of MRI provides ‘added-value’ in these risk models. ## Method We conducted this systematic review in accordance with Preferred Reporting Items for Systematic Review and Meta-Analysis (PRIMSA) guidelines [43], an overview of which is reported in Figure 1. Initially, a limited search of the Web of Science database was undertaken in June 2022 for the purpose of refining the search terms. Due to the wide-ranging classifications of central nervous system (CNS) tumours, we only used terms pertaining to the most common paediatric histological diagnoses accounting for 85% of total incidence rates (Central Brain Tumour Registry of United States, 2014-2018 [44]). Based on our initial search, we pre-registered our review protocol through the International Prospective Register of Systematic Reviews (PROSPERO) database (registration number CRD42022343161). ![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/01/13/2024.01.12.24301212/F1.medium.gif) [Figure 1.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301212/F1) Figure 1. PRISMA flow diagram A comprehensive search of Embase, MEDLINE, PsycINFO, Scopus and Web of Science was conducted in July 2022 using the designed search, resulting in 8,632 records. Minor alterations were made to the search terms for each database to account for differing Boolean operators. Additionally, we also searched the Open Science Framework (OSF) preprints archive for relevant articles that had not otherwise appeared as published texts in our main search. We included any longitudinal study concerning patients diagnosed with a brain tumour before the age of 18, who had MRI data that clearly preceded an age-appropriate, standardised test of cognitive ability (e.g., WISC-V). Central to our main research aim, we included those studies that explicitly reported an association between future cognitive outcomes based on prior MRI. Meta-analyses and literature reviews that did not report new data were excluded, however, reference lists of relevant papers were searched for additional studies of interest. Search results were not restricted by publication date but were limited to those written in English. In addition to our pre-registered exclusion criteria, we also excluded patients with neurofibromatosis, tuberous sclerosis, or acute lymphoblastic leukaemia as these were considered significant confounds for predicting cognitive outcomes. We also excluded non-peer reviewed articles, such as conference abstracts and theses. Inclusion/exclusion criteria are further detailed in Table 1. View this table: [Table 1.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301212/T1) Table 1. Inclusion and exclusion criteria for identifying publications for the systematic review Identified records were first imported into MS Excel and duplicates removed. Following a short pilot, two independent reviewers (CD + DGK) screened the titles and abstracts of all the identified papers against the inclusion criteria. Full texts of suitable papers were subsequently retrieved and screened by both reviewers for final inclusion in the review. For completeness, we also reviewed the reference lists and citations of those papers marked for inclusion for additional studies that may have been missed. At each stage of the process, disagreements were discussed until a consensus was met. Per our pre-registration, data extraction was completed by one reviewer (DGK), whilst a second reviewer evaluated data extraction of all papers for correctness (JN). The data extraction tool was initially developed for this research protocol and was later refined based upon the findings of the search results. This was not based on an existing tool, and items were selected based on discussion within the research team. Data from each study included: (1) year of publication, (2) study aims and/or hypotheses, (3) study location (i.e., country, recruiting hospital), (4) number of patients, (5) patient characteristics (i.e., years recruited, diagnoses, treatments, cognitive outcomes, age at diagnosis/MRI/neuropsychological evaluation), and (6) statistical analyses. We had initially registered our intention to assess the validity of the included studies using the Transparent Reporting of a multivariate prediction model for Individual Prognosis or Diagnosis (TRIPOD) guidelines; however, these were deemed unsuitable given that none of the studies reported using predictive modelling in their approach. Instead, studies were reviewed (by DGK) using the Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK) checklist [45, 46], a checklist for assessing reporting quality specific to the domain of oncology. Whilst designed for marker/assay testing, the domain relevance and prognostic nature renders this a relevant tool. We considered the MRI measures as the ‘marker’ under investigation and the neuropsychological assessment as ‘endpoints’ for the purposes of the tool. ## Results After reviewing titles and abstracts, 162 records were selected (see Fig1), and 153 full-text articles were assessed for eligibility. Of those, 4 were included. Manual reference and citation checking of these selected articles (and identified literature reviews deemed to be relevant), identified no additional studies. Detailed information about the included studies can be found in Table 2. View this table: [Table 2.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301212/T2) Table 2. ### Study characteristics and reporting quality Many studies were excluded because the MRI did not precede neuropsychological assessment (for instance because of the matched timepoints of neuroimaging acquisition and test assessment), thus not defining them as ‘predictive’ studies. In a small number of cases, the text was ambiguous to the order of testing (i.e., [47-50]), but did not refer to prediction, and thus were not included. For the selected studies, sample sizes were small and ranged between n=7 and n=22. The most common tumour type across studies was medulloblastoma (n=35), then astrocytoma (n=17) with relatively few ependymoma and choroid plexus papilloma (n= 3 & 1 respectively). Age at diagnosis across the studies ranged from 2.2 years to 15.6 (based on ranges and inter quartile range (IQR)). All studies selected associative statistical approaches (i.e., correlational analyses). Using the REMARK checklist, studies were assessed against each reporting item (*Item 1-20*), and here we report items where reporting was limited across the studies (i.e. one or less studies reported the item). No studies gave a rationale for sample size (Item 10, 0/4 studies), likely due to the limited samples in each study, however it was unclear as to whether these were the entirety of eligible patients within the given timeframe (as only 1 study gave a full accounting of the flow of patients in the study, *Item 12*, 1/4 studies) In terms of “Analysis and presentation”, studies performed poorly for a number of items (*Item 15, 17, 18*, all 0/4 with no studies completing the item to be reported). Firstly, no study presented an effect size for the predictive analysis (*Item 15*, 0/4 studies). Further, included studies did not conduct analysis of added value, including the MRI marker and ‘standard prognostic variables’ which are established (*Item* 17, 0/4 studies) nor sensitivity analysis/validation. ### White Matter (WM) predictors Of the studies assessed, three utilised diffusion tensor imaging (DTI) to image white matter as a predictor of outcomes. Liguoro et al. measured the fractional anisotropy (FA) and volumetry of spinocerebellar (SC), dentorubrothalamocortical (DRTC) and corticopontocerebellar tracts (frontopontocerebellar (FPC), parieto-pontocerebellar (PPC), occipitopontocerebellar (OPC), and temporo-pontocerebellar (TPC)) [51]. Significant relationships were found between tracts relevant to cerebellar connectivity, and the Developmental Neuropsychology Assessment (NEPSY) and full-scale IQ (FSIQ) measured approximately 5 months later [51]. Specifically, FSIQ correlated significantly with spherical and planar indices of the right PPC (r=-1, p=0.017 and r=0.886, p=0.033), with increases to planar index and decreases in spherical index associated with IQ [51]. Liguoro et al. also found significant correlations between specific fibre tract characteristics and tasks measuring attention, memory, sensorimotor, social perception, and visuospatial processing domains. However, only visuospatial processing showed convergent validity with significant correlations across two different tasks measuring this same domain [51]. In this study, the bilateral PPC and SC tracts were most commonly correlated with the neuropsychology tasks [51]. Partanen et al., used MRI from the treatment period 3 months after diagnosis (including during and after treatment) to predict change in intellectual functioning over a 36 month period after diagnosis. A significant reduction in FSIQ over time was found but this was not related to diffusion measures (FA, mean diffusivity (MD), radial diffusivity (RD) and axial diffusivity (AD)) for the cortical spinal tract (CST), inferior fronto-occipital fasciculus (IFOF), inferior longitudinal fasciculus (ILF), optic radiations (OR), and uncinate fasciculus (UF) [52]. Partanen et al. did however show that declines over time in processing speed index, observed only in a subgroup of patients experiencing local therapy (i.e., focal radiation) versus cranial spinal radiation, was predicted by baseline anisotropy in left inferior fronto-occipital fascicle (IFOF), with lower FA being related to greater decline [52]. Neither patient groups showed a difference in the left IFOF for diffusions measures compared to controls. Riggs et. al. [53] utilised chronically acquired MRI (approximately 5 years post diagnosis) investigated correlations between whole brain WM volume, FA of both the left and right UF and the general memory index of the CMS (in a subset of n=10 patients, outcomes measured between 2 and 19 months after MRI). Only FA of the left UF was significantly related to memory (R=.64, p=.045), not the right uncinate fasciculus or total WM volume (as measured by structural MRI), with increased FA related to increases in memory performance. The volume of the PPC tract also positively correlated with memory performance (R=.71, p=.045) in Parten’s study [52]. Strength of correlational relationship between indices of white matter integrity and neuropsychological outcome were large (according to Cohen’s criteria) ranging from |r|=.64 - |r|=1. However, the very limited sample sizes (n=10 & n=7) from which these were drawn gives reason for concern over the interpretation of these estimates. ### Grey Matter (GM) predictors In Riggs et. al., No correlation was found between total GM volume or left hippocampus with the general memory index of the CMS, but the right hippocampal volume, measured at a chronic timepoint, showed significant positive correlation with memory outcomes 2-19 months after MRI (R=.71, p=.02) [53]. It is important to note in this study, that the right hippocampus, rather than the GM volume and left hippocampus, was significantly smaller in the patient group compared to healthy controls. ### Lesion predictors Zilli et. al. [54] used a lesion-symptom mapping approach, to investigate the overlap of lesions in children with versus without psychological impairment. The lesions investigated where tumour lesions, frontal insertion of ventriculoperitoneal drainages and ventricular volumes, as drawn on the T1w MRI. They found the greatest tumour lesion overlap and therefore greatest damage was found in median cerebellar, specifically paravermal and vermal regions. Regions of interest for the lateral ventricles also overlapped in impaired children, suggesting hydrocephalus as additional cause of future impairment. ### Meta-analysis Despite registering our intention, reviewed studies were not of sufficient quality to conduct any form of meta-analysis due to varying measurement strategies, gaps in reported descriptive variables, and low power. ## Discussion ### Review of the State of Research The current systematic review aimed to investigate existing literature using MRI to predict later, and non-contemporaneous neuropsychological outcomes in children with brain tumours. No studies reviewed here set out the rationale for and aimed to predict future outcomes using model development and validation approaches. The lack of scientific attention given to this topic is surprising given the dearth of literature advocating for such research. The papers identified and reviewed here, did in fact conduct analyses to this effect, but only due to the fortuitous nature of the selected timepoints, and intervals between the activities of MRI scanning and neuropsychological testing. Despite an extensive search strategy, evidence with which to answer the current research question was extremely limited, with the major finding being a severe lack of studies in this area. The primary result of the review must therefore be viewed as a need for further research in this very important research area, with study designs that directly tackle the need for outcome prediction in these cohorts. The reason for this brevity of research in this field is unclear. Whilst our systematic search strategy was extensive, there were also difficulties in identifying papers due to poor reporting practices. For instance, for some of the studies where the full text was assessed, the timing of MRI in relation to assessment was ambiguous or unclear [47-50, 55]. Factors affecting the number of research articles in this area could include previous focus on survival, where increasing survival rates are now placing a greater need for research on late effects. It is important to also consider that neurocognitive effects are also only one of many potential late effects experienced by this population [56, 57]. Research in this area may also be impacted by the availability of clinical data with which to carry out this research. For children with pediatric brain tumours, there is an abundance of clinical MRI, with medical imaging required for vital for tumour detection and diagnosis, surgical and radiotherapy planning, and monitoring of treatment response and recurrence of disease. But this is not necessarily echoed in access to neurocognitive assessment – testing is performed based upon clinical need or clinical trial protocol for instance, and thus this limits available retrospective datasets. This data, also comes from a heterogeneous cohort, with these children facing heterogenous brain injuries as a result of their disease and treatment. Identifying homogenous patient groups inevitably results in the smaller sample sizes seen in the current studies. Overall, these factors are liable to impact the quantity of research studies in this field. Whilst number of studies was limited, the quality of existing studies was also a significant issue. In the reporting quality assessment (using the REMARK checklist [45, 46]) identified studies did not meet important criteria for development of prognostic markers. Specifically, studies failed to conduct additional analyses necessary for this development, such as sensitivity or ‘added-value’ analyses – although this was likely due to limited sample sizes, therefore lacking statistical power necessary for these additional analyses. Studies are typically involving “retrospective, monocentric study investigating a pediatric disease with low annual incidence” [54] however future work will require larger sample sizes than those of the studies presented here. To note, the checklist comments on a number of items pertaining to model building and multivariate analyses, which were not conducted in the current studies. Overall, the findings from the reviewed research are limited – they have limited sample sizes and are rated as low quality in terms of prediction studies. Without proper validation and replication, the quality and impact of any conclusions must be viewed as limited and/or potentially spurious. However, the findings are briefly discussed here in terms of wider literature. This should be seen in the context of guiding hypothesis-driven future research and/or promoting future validation and/or replication of these findings. ### Summary of Findings of Reviewed Studies #### Cerebellar Damage Given common posterior fossa presentation in pediatric brain tumours, it is unsurprising that studies in this review a-priori selected regions of interest within cerebellar structures and related fibre projections from this region. Damage to these circuits predicted outcomes [51], with lesions to the median cerebellar regions common in cognitively impaired patients [54]. Studies of contemporaneous MRI and neuropsychology measures have found similar. Horská and colleagues [58] found a decrease in vermis volumes over a 6-month period were significantly related to radiation dose, and final volume after this period related to neuropsychological measures of motor speed. Significant recent evidence suggests that the posterior cerebellar lobes are key in maintain cognitive performance [59], and animal models suggest that intact cerebellar activity is required to enable typical developmental trajectories of cognitive abilities (in mice) [60]. Essentially, the cerebellum plays as an integral node in many distributed neural circuits that underpin multiple cognitive functions [61, 62]. Radical cerebellar resection has also been associated with extensive WM microstructure changes across the brain [63]. Overall, it is unsurprising that damage to cerebellar regions (through injury and treatment effects) may lead to multiple cognitive morbidities. #### WM damage Riggs, Bouffet [53] argued that global measures of WM may be indicative of general injury and thus correlate well with general ability, however, integrity of discrete tracts (such as the UF) may be a better predictor of specific cognitive abilities – in this case memory. Previous reviews of cross-sectional research suggests a model where disorganised WM microstructure is related to poorer cognitive abilities, especially processing speed and memory deficits [22], by indicting that this ‘damage-related impairment’ is established early, and therefore WM microstructure is a potential biomarker to predict later impairment. Both preclinical and patient studies suggest a loss of both GM and WM volume, and failure of normal WM gain in pediatric brain tumour survivors [15]. There are multiple mechanisms of WM damage; hydrocephalous having direct neurotoxic effects on periventricular WM due to decreased perfusion and oedema [64] or intragenic effect of adjunct therapy (chemo and radiotherapy) as measurable by reduced volume and alterations to microstructural properties and failure of expected WM development [49, 65]. WM damage is likely non-transient, in non-irradiated patients 15 years after diagnosis, FA measures are reduced and are associated with impaired cognitive flexibility [66]. What is apparent is that, across treatment and disease effects, the subsequent WM injury is relevant to poor outcomes, including emotion, cognition, and behaviour [13, 49, 67]. ### Recommendations for Future Research #### Limited longitudinal studies The biggest limitation of the current research field is the lack of longitudinal research answering this research question. Whilst there have been multiple studies understanding the contemporaneous neuroanatomical substrates of poor neuropsychological outcomes in pediatric brain tumours, from acutely post treatment to very long-term survivors, these have not translated into similarly large body of work understanding long-term risk (as highlighted by our findings) of cognitive morbidity. Further longitudinal research is needed to assess whether the contemporaneous neuroanatomical substrates of long-term impairment are in fact predictive in the context of longitudinal studies. These longitudinal studies would also provide an opportunity to disentangle the developmental and age-related effects on this prediction-task. For several of the measures highlighted in this review (FA/MD etc.) there are known developmental trajectories [68] which will necessarily interact with disease-related changes. There is also likely to be unique effects of brain insult, across tumour growth, and treatment related injury at different ages, resulting in varying levels of long-term impairments [69]. The field will need to rely upon longitudinal studies (with sufficient sample size/statistical power) that can sufficiently disentangle these interactions. #### Study Variables: Timing There are two important timings to consider for the field moving forward to truly provide the most clinical impact moving forward. Firstly, the time between ‘predictive’ MRI and neurocognitive assessment. Across the reported studies, this was extremely short, limited to less than a year in most cases. This is potentially less useful clinically, where predictions of outcomes further out (2years, 5years etc) may be of greater impact, especially in cases where the gap between them and their peers grow with the failure to ‘keep up’. However, given that the reported studies did not set out to develop prognostic models, this is unsurprising, and unfair to judge them so. The second important timing to consider is the time between diagnosis and ‘predictive’ MRI. In order for this predictive information to be useful for patient management and treatment planning, this interval needs to be short, especially given the urgency under which these patients need to receive potentially life-saving care. There is no consensus on the optimal timing of MR imaging to use for predictive purposes. Selection of which MRI is likely to be most predictive (in terms of reliability, accuracy etc.) will not be trivial for future research. There is **s**ignificant evidence to suggest that the treatment effects of injury are damaging to the still-developing brain and thus post treatment MRI may hold the complete picture of brain structural damage which lays the scaffold for future cognitive impairment. These scans may however come much later in the disease course and, as the timing of the MRI used for prediction increases further from the diagnostic scan, it is increasingly possible that the outcome of interest (cognitive morbidities) has already presented, and thus prediction is not useful (as the observation and/or measurement of the presenting difficulties is inherently the best predictor at that time point). There is also the rationale of aiming for earliest prediction possible, to best enable early neuropsychological intervention to support better outcomes for these children. These children undergo MRI scanning at a number of timepoints in their disease course (e.g. diagnostic imaging, pre- and post-surgical evaluation, progression monitoring etc), and so there is significant data for potential retrospective studies to investigate effect of MRI timing on prediction. Direct comparisons between models using MRI from different timepoints will be meaningful to elicit after which events (i.e., diagnosis, tumour growth, treatment etc.) damage has been sufficient to indicate future cognitive morbidities. This will be challenging due to potential differences in MRI scans between events (i.e., initial diagnostic MRI at a tertiary hospital versus post-treatment MRI at a specialist centre). However, it is important to consider timing of both MRI and neuropsychological assessment in the framework of predicting outcomes. There may be consolidation periods post-surgery that impact outcome measurement over time, and thus research also needs to be clear in the timing of outcome measurement and / or study endpoint. #### Added Value of MRI One limitation of the current state of the research literature in this field is that the added value of MRI in prediction has not been established, above and beyond existing approaches. No reviewed studies assessed existing risk factors in a multivariate analysis to test the relative contributions, and therefore added value, of early MRI in predicting future neurobehavioral morbidities. However, Partanen et. al. reported that none of the medical or treatment variables that they tested predicted change in IQ scores over time [52]. This is despite these medical variables (Neurological Predictor Scale and presence of cerebellar mutism syndrome) predicting acute/contemporaneous neuropsychology outcomes, and MRI-derived measures of baseline WM injury being significantly related with outcomes [52]. This is limited evidence to support the incremental validity of MRI as a predictor of long-term outcomes. Future studies should ensure to test for unique and additional predictive power offered by quantitative MRI variables #### Study Variables: Approach to ROIs Across the studies reviewed here two conducted analyses in regions-of-interest (ROIs) directly related to sites of brain insult in these patients [51, 52], one in ROIs related to the cognitive comorbidity under investigation [53], and only one investigating characteristics of the lesion itself [54]. This does not consider how the wider brain network may be influenced by the brain tumour, and this information may explain/predict additional variance in outcomes. For instance, in paediatric neurological disorders/syndromes, differences in brain morphometry or connectivity have been found beyond the site of pathology (i.e. paediatric epilepsy [70]) or in the absence of frank pathology (i.e. mild paediatric TBI [71], MRI-negative epilepsy [72]). Disconnectome symptom mapping, shows that non-homologous lesions to the same brain network can generate the same cognitive sequalae in terms of deficits. [73]. Many compensatory and ‘rerouting’ models of functional brain activity post injury suggest that regions beyond the focal lesion may explain some sparing of cognitive abilities (another important factor in predicting endpoint neurobehavioral morbidities). These findings all show that disparate, diffuse, and non-lesioned changes to the brain, including tissue which may be typically thought of as ‘spared’, could also explain variance in neurobehavioral morbidities. Connectivity approaches to MRI have shown utility in contemporaneous measurements of MRI and neuropsychology [74, 75]. These neurobiological effects of injury beyond the focal lesion may provide further prognostic information towards the aim of a predictive model, however, to test a greater number of regions larger sample sizes will be necessary to accurately estimate statistical models across many more ROIs. This highlights one of the key challenges for future studies in this field being data collection. ### Limitations of Review It should be noted that, despite an extensive search, no study explicitly investigated the research question of whether MRI could be used for long term prediction of neurocognitive outcomes in pediatric brain tumour patients. Described studies were reviewed here due to non-primary analyses which fulfilled inclusion criteria, and therefore it may be the case that other studies with such analyses may have been missed in the review process (for instance if these secondary analyses were not mentioned in the abstract). To address this, we erred on the side of caution in reviewing abstracts, using full-text review as a method to identify these relevant secondary analyses. ## Conclusion As early as 2008, it was proposed that to truly balance the aggressiveness of treatment for childhood CNS tumours, against the relative quality of life due to cognitive impairment, an important factor is knowing the likelihood of any one individual experiencing neurocognitive impairment [12]. This individualised risk is key in purported models of monitoring and managing of neurocognitive functioning in children with brain tumours [76]. There has also been significant work in the field of pediatric brain tumours proposing developmental cognitive neuroscience models of late effects in survivors [12, 22, 77, 78]. These models, built on contemporaneous measures of cognition and brain development, alongside cross-sectional data, are inherently limited. Knowing individualised risk of long-term cognitive morbidity ahead of time would have significant clinical impact; to inform clinical management, prioritise resources/support, and reduce uncertainty for families. Overall, there exists plenty clinical reasoning to prompt scientific enthusiasm and attention for this topic. However, despite these early calls for prediction, and models with which to guide these predictive studies, this systematic review highlights that the number of truly predictive studies (requiring a period between predictive features and long-term outcomes) is still limited. In conclusion, Given the increased number of adult survivors of childhood brain tumours, the poorer long-term cognitive, educational and employment outcomes [10, 13, 14] and the significant burden this represents to patients, families and healthcare, work now needs to be completed to integrate predictive data into these models, which will expand their explanatory value and utility to clinical practice. This will be an important next step in delivering further clinical impact for this patient group. Given the great potential that MRI provides in investigating neurobiological effects of disease and treatment at the individual-level, the plethora of multimodal imaging available in these clinical populations and finally the positive clinical benefit this could offer, there is exciting opportunities for this type of research. ## Supporting information Supplemental Materials [[supplements/301212_file03.docx]](pending:yes) ## Data Availability Data extracted from included studies is publicly available and found (in its entirety) in Table 2. Lists of reviewed studied, at each stage of the PRISMA flowchart are available from the authors upon reasonable request. ## Data Availability Data extracted from included studies is publicly available and found (in its entirety) in Table 2. Lists of reviewed studied, at each stage of the PRISMA flowchart are available from the authors upon reasonable request. ## Acknowledgments DGK is funded by a post-doctoral award from Aston University College of Health and Life Sciences, awarded to JN & DGK. * Received January 12, 2024. * Revision received January 12, 2024. * Accepted January 13, 2024. * © 2024, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/) ## References 1. 1.Miller, K.D., L. Nogueira, T. Devasia, A.B. Mariotto, K.R. Yabroff, A. Jemal, J. Kramer, and R.L. Siegel, Cancer treatment and survivorship statistics, 2022. CA Cancer J Clin, 2022. 72(5): p. 409–436. 2. 2.Phillips, S.M., L.S. Padgett, W.M. Leisenring, K.K. Stratton, K. Bishop, K.R. Krull, C.M. Alfano, T.M. Gibson, J.S. de Moor, D.B. Hartigan, G.T. Armstrong, L.L. Robison, J.H. Rowland, K.C. Oeffinger, and A.B. Mariotto, Survivors of childhood cancer in the United States: prevalence and burden of morbidity. Cancer Epidemiol Biomarkers Prev, 2015. 24(4): p. 653–63. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiY2VicCI7czo1OiJyZXNpZCI7czo4OiIyNC80LzY1MyI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzAxLzEzLzIwMjQuMDEuMTIuMjQzMDEyMTIuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 3. 3.1. A.C. Society American Cancer Society, Cancer Facts & Figures 2014, A.C. Society, Editor. 2014, American Cancer Society: Atlanta. 4. 4.Ward, E., C. DeSantis, A. Robbins, B. Kohler, and A. Jemal, Childhood and adolescent cancer statistics, 2014. CA Cancer J Clin, 2014. 64(2): p. 83–103. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3322/caac.21219&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=24488779&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000332963800003&link_type=ISI) 5. 5.Tabash, M.A., Characteristics, survival and incidence rates and trends of pilocytic astrocytoma in children in the United States; SEER-based analysis. J Neurol Sci, 2019. 400: p. 148–152. 6. 6.Ramaswamy, V. and M.D. Taylor, Medulloblastoma: From Myth to Molecular. J Clin Oncol, 2017. 35(21): p. 2355–2363. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1200/JCO.2017.72.7842&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=28640708&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 7. 7.Gatta, G., G. Zigon, R. Capocaccia, J.W. Coebergh, E. Desandes, P. Kaatsch, G. Pastore, R. Peris-Bonet, C.A. Stiller, and E.W. Group, Survival of European children and young adults with cancer diagnosed 1995-2002. Eur J Cancer, 2009. 45(6): p. 992–1005. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.ejca.2008.11.042&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19231160&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000265519700004&link_type=ISI) 8. 8.Prevention, C.f.D.C., Declines in cancer death rates among children and adolescents in the United States, 1999–2014. 2016. 9. 9.Wang, Y., A.P.Y. Liu, T.M. Lee, W.H.S. Wong, D.Y.T. Fong, L.K. Leung, M.M.K. Shing, D.T. Ku, G.C. Chan, and W.W. Tso, Neurobehavioral Impairment in Pediatric Brain Tumor Survivors: A Meta-Analysis. Cancers (Basel), 2022. 14(13). 10. 10.Maddrey, A.M., J.A. Bergeron, E.R. Lombardo, N.K. McDonald, A.F. Mulne, P.D. Barenberg, and D.C. Bowers, Neuropsychological performance and quality of life of 10 year survivors of childhood medulloblastoma. J Neurooncol, 2005. 72(3): p. 245–53. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s11060-004-3009-z&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15937648&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 11. 11.Palmer, S.L., O. Goloubeva, W.E. Reddick, J.O. Glass, A. Gajjar, L. Kun, T.E. Merchant, and R.K. Mulhern, Patterns of intellectual development among survivors of pediatric medulloblastoma: a longitudinal analysis. Journal of Clinical Oncology, 2001. 19(8): p. 2302–2308. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiamNvIjtzOjU6InJlc2lkIjtzOjk6IjE5LzgvMjMwMiI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzAxLzEzLzIwMjQuMDEuMTIuMjQzMDEyMTIuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 12. 12.Askins, M.A. and B.D. Moore, 3rd., Preventing neurocognitive late effects in childhood cancer survivors. J Child Neurol, 2008. 23(10): p. 1160–71. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/0883073808321065&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=18952582&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000260313200007&link_type=ISI) 13. 13.Lassaletta, A., E. Bouffet, D. Mabbott, and A.V. Kulkarni, Functional and neuropsychological late outcomes in posterior fossa tumors in children. Childs Nerv Syst, 2015. 31(10): p. 1877–90. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s00381-015-2829-9&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26351237&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 14. 14.Devine, K.A., S. Christen, R.L. Mulder, M.C. Brown, L.M. Ingerski, L. Mader, E.J. Potter, C. Sleurs, A.S. Viola, S. Waern, L.S. Constine, M.M. Hudson, L.C.M. Kremer, R. Skinner, G. Michel, J. Gilleland Marchak, F.S.M. Schulte, and G. International Guidelines Harmonization Group Psychological Late Effects, Recommendations for the surveillance of education and employment outcomes in survivors of childhood, adolescent, and young adult cancer: A report from the International Late Effects of Childhood Cancer Guideline Harmonization Group. Cancer, 2022. 128(13): p. 2405–2419. 15. 15.Al Dahhan, N.Z., E. Cox, B.J. Nieman, and D.J. Mabbott, Cross-translational models of late-onset cognitive sequelae and their treatment in pediatric brain tumor survivors. Neuron, 2022. 110(14): p. 2215–2241. 16. 16.Hardy, K.K., M.M. Hudson, and K.R. Krull, Life-Altering Consequences of Neurocognitive Impairment in Survivors of Pediatric Cancer. J Clin Oncol, 2021. 39(16): p. 1693–1695. 17. 17.Willard, V.W., K.S. Berlin, H.M. Conklin, and T.E. Merchant, Trajectories of psychosocial and cognitive functioning in pediatric patients with brain tumors treated with radiation therapy. Neuro Oncol, 2019. 21(5): p. 678–685. 18. 18.Sharkey, C.M., L.L. Mullins, A.H. Clawson, A. Gioia, M.A.W. Hawkins, J.M. Chaney, K.S. Walsh, and K.K. Hardy, Assessing neuropsychological phenotypes of pediatric brain tumor survivors. Psychooncology, 2021. 30(8): p. 1366–1374. 19. 19.Major, N., N.A. Patel, J. Bennett, E. Novakovic, D. Poloni, M. Abraham, N.J. Brown, J.L. Gendreau, R. Sahyouni, and J. Loya, The Current State of Radiotherapy for Pediatric Brain Tumors: An Overview of Post-Radiotherapy Neurocognitive Decline and Outcomes. J Pers Med, 2022. 12(7). 20. 20.Ajithkumar, T., S. Price, G. Horan, A. Burke, and S. Jefferies, Prevention of radiotherapy-induced neurocognitive dysfunction in survivors of paediatric brain tumours: the potential role of modern imaging and radiotherapy techniques. Lancet Oncol, 2017. 18(2): p. e91–e100. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1470-2045(17)30030-X&link_type=DOI) 21. 21.Peterson, R.K. and T.Z. King, A systemic review of pediatric neuropsychological outcomes with proton versus photon radiation therapy: A call for equity in access to treatment. J Int Neuropsychol Soc, 2022:p. 1–14. 22. 22.Wauters, M., A. Uyttebroeck, L. De Waele, C. Sleurs, and S. Jacobs, Neuroimaging Biomarkers and Neurocognitive Outcomes in Pediatric Medulloblastoma Patients: a Systematic Review. Cerebellum, 2021. 20(3): p. 462–480. 23. 23.Oyefiade, A., I. Paltin, C.R. De Luca, K.K. Hardy, D.R. Grosshans, M. Chintagumpala, D.J. Mabbott, and L.S. Kahalley, Cognitive Risk in Survivors of Pediatric Brain Tumors. J Clin Oncol, 2021. 39(16): p. 1718–1726. 24. 24.Palmer, S.L., Neurodevelopmental impact on children treated for medulloblastoma: a review and proposed conceptual model. Dev Disabil Res Rev, 2008. 14(3): p. 203–10. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/ddrr.32&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=18924159&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000262726500004&link_type=ISI) 25. 25.Wolfe, K.R., A. Madan-Swain, and R.K. Kana, Executive dysfunction in pediatric posterior fossa tumor survivors: a systematic literature review of neurocognitive deficits and interventions. Dev Neuropsychol, 2012. 37(2): p. 153–75. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1080/87565641.2011.632462&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=22339228&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 26. 26.King, T.Z., A.S. Ailion, M.E. Fox, and S.M. Hufstetler, Neurodevelopmental model of long-term outcomes of adult survivors of childhood brain tumors. Child Neuropsychol, 2019. 25(1): p. 1–21. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 27. 27.Marusak, H.A., A.S. Iadipaolo, F.W. Harper, F. Elrahal, J.W. Taub, E. Goldberg, and C.A. Rabinak, Neurodevelopmental consequences of pediatric cancer and its treatment: applying an early adversity framework to understanding cognitive, behavioral, and emotional outcomes. Neuropsychol Rev, 2018. 28(2): p. 123–175. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 28. 28.King, A.A., K. Seidel, C. Di, W.M. Leisenring, S.M. Perkins, K.R. Krull, C.A. Sklar, D.M. Green, G.T. Armstrong, L.K. Zeltzer, E. Wells, M. Stovall, N.J. Ullrich, K.C. Oeffinger, L.L. Robison, and R.J. Packer, Long-term neurologic health and psychosocial function of adult survivors of childhood medulloblastoma/PNET: a report from the Childhood Cancer Survivor Study. Neuro Oncol, 2017. 19(5): p. 689–698. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/neuonc/now242&link_type=DOI) 29. 29.de Ruiter, M.A., R. van Mourik, A.Y. Schouten-van Meeteren, M.A. Grootenhuis, and J. Oosterlaan, Neurocognitive consequences of a paediatric brain tumour and its treatment: a meta-analysis. Dev Med Child Neurol, 2013. 55(5): p. 408–17. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/dmcn.12020&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=23157447&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 30. 30.Nathan, P.C., S.K. Patel, K. Dilley, R. Goldsby, J. Harvey, C. Jacobsen, N. Kadan-Lottick, K. McKinley, A.K. Millham, I. Moore, M.F. Okcu, C.L. Woodman, P. Brouwers, F.D. Armstrong, and C. Children’s Oncology Group Long-term Follow-up Guidelines Task Force on Neurocognitive/Behavioral Complications After Childhood, Guidelines for identification of, advocacy for, and intervention in neurocognitive problems in survivors of childhood cancer: a report from the Children’s Oncology Group. Arch Pediatr Adolesc Med, 2007. 161(8): p. 798–806. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/archpedi.161.8.798&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17679663&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000248583000011&link_type=ISI) 31. 31.Krull, K.R., S. Minoshima, M. Edelmann, B. Morris, N.D. Sabin, T.M. Brinkman, G.T. Armstrong, L.L. Robison, M.M. Hudson, and B. Shulkin, Regional brain glucose metabolism and neurocognitive function in adult survivors of childhood cancer treated with cranial radiation. J Nucl Med, 2014. 55(11): p. 1805–10. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6Njoiam51bWVkIjtzOjU6InJlc2lkIjtzOjEwOiI1NS8xMS8xODA1IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxMi5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 32. 32.Carpentieri, S.C., D.P. Waber, S.L. Pomeroy, R.M. Scott, L.C. Goumnerova, M.W. Kieran, A.L. Billett, and N.J. Tarbell, Neuropsychological functioning after surgery in children treated for brain tumor. Neurosurgery, 2003. 52(6): p. 1348–56; discussion 1356-7. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1227/01.NEU.0000064804.00766.62&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12762880&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000183366600035&link_type=ISI) 33. 33.Taiwo, Z., S. Na, and T.Z. King, The Neurological Predictor Scale: A predictive tool for long-term core cognitive outcomes in survivors of childhood brain tumors. Pediatr Blood Cancer, 2017. 64(1): p. 172–179. 34. 34.Levitch, C.F., A.A. Holland, J. Bledsoe, S.Y. Kim, M. Barnett, S. Ramjan, and S.A. Sands, Comparison of neuropsychological functioning in pediatric posterior fossa tumor survivors: Medulloblastoma, low-grade astrocytoma, and healthy controls. Pediatr Blood Cancer, 2022. 69(2): p. e29491. 35. 35.Micklewright, J.L., T.Z. King, R.D. Morris, and N. Krawiecki, Quantifying pediatric neuro-oncology risk factors: development of the neurological predictor scale. J Child Neurol, 2008. 23(4): p. 455–8. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/0883073807309241&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=18182641&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000254550900018&link_type=ISI) 36. 36.Papazoglou, A., T.Z. King, R.D. Morris, and N. Krawiecki, Parent report of attention problems predicts later adaptive functioning in children with brain tumors. Child Neuropsychol, 2009. 15(1): p. 40–52. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=18608223&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 37. 37.King, T.Z. and S. Na, Cumulative neurological factors associated with long-term outcomes in adult survivors of childhood brain tumors. Child Neuropsychol, 2016. 22(6): p. 748–60. 38. 38.Kautiainen, R.J., S.D. Na, and T.Z. King, Neurological predictor scale is associated with academic achievement outcomes in long-term survivors of childhood brain tumors. J Neurooncol, 2019. 142(1): p. 193–201. 39. 39.Bae, D., V.V. Mlc, and C.E. Catsman-Berrevoets, Preoperative prediction of postoperative cerebellar mutism syndrome. Validation of existing MRI models and proposal of the new Rotterdam pCMS prediction model. Childs Nerv Syst, 2020. 36(7): p. 1471–1480. 40. 40.Kesler, S.R., C. Sleurs, B.C. McDonald, S. Deprez, E. van der Plas, and B.J. Nieman, Brain Imaging in Pediatric Cancer Survivors: Correlates of Cognitive Impairment. J Clin Oncol, 2021. 39(16): p. 1775–1785. 41. 41.Baker, S. and Y. Kandasamy, Machine learning for understanding and predicting neurodevelopmental outcomes in premature infants: a systematic review. Pediatr Res, 2022:p. 1–7. 42. 42.Li, X., X. Wang, L. Su, X. Hu, and Y. Han, Sino Longitudinal Study on Cognitive Decline (SILCODE): protocol for a Chinese longitudinal observational study to develop risk prediction models of conversion to mild cognitive impairment in individuals with subjective cognitive decline. BMJ Open, 2019. 9(7): p. e028188. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NzoiYm1qb3BlbiI7czo1OiJyZXNpZCI7czoxMToiOS83L2UwMjgxODgiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyNC8wMS8xMy8yMDI0LjAxLjEyLjI0MzAxMjEyLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 43. 43.Page, M.J., J.E. McKenzie, P.M. Bossuyt, I. Boutron, T.C. Hoffmann, C.D. Mulrow, L. Shamseer, J.M. Tetzlaff, E.A. Akl, S.E. Brennan, R. Chou, J. Glanville, J.M. Grimshaw, A. Hrobjartsson, M.M. Lalu, T. Li, E.W. Loder, E. Mayo-Wilson, S. McDonald, L.A. McGuinness, L.A. Stewart, J. Thomas, A.C. Tricco, V.A. Welch, P. Whiting, and D. Moher, The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Syst Rev, 2021. 10(1): p. 89. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s13643-021-01626-4&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=33781348&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 44. 44.Ostrom, Q.T., G. Cioffi, K. Waite, C. Kruchko, and J.S.J.N.-o. Barnholtz-Sloan, CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2014–2018. 2021. 23(Supplement_3): p. iii1–iii105. 45. 45.McShane, L.M., D.G. Altman, W. Sauerbrei, S.E. Taube, M. Gion, and G.M. Clark, Reporting recommendations for tumor marker prognostic studies (REMARK). Journal of the National Cancer Institute, 2005. 97(16): p. 1180–1184. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/jnci/dji237&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=16106022&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000232272800007&link_type=ISI) 46. 46.Sauerbrei, W., S.E. Taube, L.M. McShane, M.M. Cavenagh, and D.G. Altman, Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK): An Abridged Explanation and Elaboration. J Natl Cancer Inst, 2018. 110(8): p. 803–811. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/jnci/djy088&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29873743&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 47. 47.Aukema, E.J., M.W. Caan, N. Oudhuis, C.B. Majoie, F.M. Vos, L. Reneman, B.F. Last, M.A. Grootenhuis, and A.Y. Schouten-van Meeteren, White matter fractional anisotropy correlates with speed of processing and motor speed in young childhood cancer survivors. Int J Radiat Oncol Biol Phys, 2009. 74(3): p. 837–43. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.ijrobp.2008.08.060&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19117694&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000266754800027&link_type=ISI) 48. 48.Mulhern, R.K., S.L. Palmer, W.E. Reddick, J.O. Glass, L.E. Kun, J. Taylor, J. Langston, and A. Gajjar, Risks of young age for selected neurocognitive deficits in medulloblastoma are associated with white matter loss. J Clin Oncol, 2001. 19(2): p. 472–9. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiamNvIjtzOjU6InJlc2lkIjtzOjg6IjE5LzIvNDcyIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxMi5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 49. 49.Reddick, W.E., D.J. Taghipour, J.O. Glass, J. Ashford, X.P. Xiong, S.J. Wu, M. Bonner, R.B. Khan, and H.M. Conklin, Prognostic Factors that Increase the Risk for Reduced White Matter Volumes and Deficits in Attention and Learning for Survivors of Childhood Cancers. Pediatric Blood & Cancer, 2014. 61(6): p. 1074–1079. 50. 50.Tso, W.W.Y., E.S.K. Hui, T.M.C. Lee, A.P.Y. Liu, P. Ip, V. Vardhanabhuti, K.K.F. Cheng, D.Y.T. Fong, D.H.F. Chang, F.K.W. Ho, K.M. Yip, D.T.L. Ku, D.K.L. Cheuk, C.W. Luk, M.K. Shing, L.K. Leung, P.L. Khong, and G.C. Chan, Brain Microstructural Changes Associated With Neurocognitive Outcome in Intracranial Germ Cell Tumor Survivors. Front Oncol, 2021. 11: p. 573798. 51. 51.Liguoro, I., E. Passone, T. Zilli, M. Maieron, M.C. De Colle, M. Skrap, V. Dolcemascolo, G. Sommariva, P. Cogo, and B. Tomasino, Possible association between the integrity of cerebellar pathways and neurocognitive performance in children with posterior fossa tumors. Pediatr Blood Cancer, 2020. 67(9): p. e28538. 52. 52.Partanen, M., E. Bouffet, S. Laughlin, D. Strother, J. Hukin, J. Skocic, K. Szulc-Lerch, and D.J. Mabbott, Early changes in white matter predict intellectual outcome in children treated for posterior fossa tumors. Neuroimage Clin, 2018. 20: p. 697–704. 53. 53.Riggs, L., E. Bouffet, S. Laughlin, N. Laperriere, F. Liu, J. Skocic, N. Scantlebury, F. Wang, N.J. Schoenhoff, D. Strother, J. Hukin, C. Fryer, D. McConnell, and D.J. Mabbott, Changes to memory structures in children treated for posterior fossa tumors. J Int Neuropsychol Soc, 2014. 20(2): p. 168–80. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1017/S135561771300129X&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=24460980&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 54. 54.Zilli, T., V. Dolcemascolo, E. Passone, M. Maieron, M.C. De Colle, M. Skrap, T. Ius, I. Liguoro, M. Venchiarutti, P. Cogo, and B. Tomasino, A multimodal approach to the study of children treated for posterior fossa tumor: A review of the literature and a pilot study. Clin Neurol Neurosurg, 2021. 207: p. 106819. 55. 55.Kline, C., S. Stoller, L. Byer, D. Samuel, J.M. Lupo, M.A. Morrison, A.M. Rauschecker, P. Nedelec, W. Faig, D.B. Dubal, H.J. Fullerton, and S. Mueller, An Integrated Analysis of Clinical, Genomic, and Imaging Features Reveals Predictors of Neurocognitive Outcomes in a Longitudinal Cohort of Pediatric Cancer Survivors, Enriched with CNS Tumors (Rad ART Pro). Front Oncol, 2022. 12: p. 874317. 56. 56.Rey-Casserly, C. and T.J.C.o.i.p. Diver, Late effects of pediatric brain tumors. 2019. 31(6): p. 789–796. 57. 57.Alemany, M., R. Velasco, M. Simó, and J.J.N.-O.P. Bruna, Late effects of cancer treatment: consequences for long-term brain cancer survivors. 2021. 8(1): p. 18–30. 58. 58.Horska, A., A. Laclair, M. Mohamed, C.T. Wells, T. McNutt, K.J. Cohen, M. Wharam, E.M. Mahone, and W. Kates, Low cerebellar vermis volumes and impaired neuropsychologic performance in children treated for brain tumors and leukemia. AJNR Am J Neuroradiol, 2010. 31(8): p. 1430–7. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiYWpuciI7czo1OiJyZXNpZCI7czo5OiIzMS84LzE0MzAiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyNC8wMS8xMy8yMDI0LjAxLjEyLjI0MzAxMjEyLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 59. 59.Hoang, D.H., A. Pagnier, K. Guichardet, F. Dubois-Teklali, I. Schiff, G. Lyard, E. Cousin, and A. Krainik, Cognitive disorders in pediatric medulloblastoma: what neuroimaging has to offer. J Neurosurg Pediatr, 2014. 14(2): p. 136–44. 60. 60.Badura, A., J.L. Verpeut, J.W. Metzger, T.D. Pereira, T.J. Pisano, B. Deverett, D.E. Bakshinskaya, and S.S. Wang, Normal cognitive and social development require posterior cerebellar activity. Elife, 2018. 7. 61. 61.Clark, S.V., E.S. Semmel, H.A. Aleksonis, S.N. Steinberg, and T.Z. King, Cerebellar-Subcortical-Cortical Systems as Modulators of Cognitive Functions. Neuropsychol Rev, 2021. 31(3): p. 422–446. 62. 62.Schmahmann, J.D., The cerebellum and cognition. Neurosci Lett, 2019. 688: p. 62–75. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.neulet.2018.07.005&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29997061&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 63. 63.Gomes, C.A., K.M. Steiner, N. Ludolph, T. Spisak, T.M. Ernst, O. Mueller, S.L. Goricke, F. Labrenz, W. Ilg, N. Axmacher, and D. Timmann, Resection of cerebellar tumours causes widespread and functionally relevant white matter impairments. Hum Brain Mapp, 2021. 42(6): p. 1641–1656. 64. 64.Moberget, T., S. Andersson, T. Lundar, B.J. Due-Tonnessen, A. Heldal, T. Endestad, and L.T. Westlye, Longterm supratentorial brain structure and cognitive function following cerebellar tumour resections in childhood. Neuropsychologia, 2015. 69: p. 218–31. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.neuropsychologia.2015.02.007&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25665770&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 65. 65.Witzmann, K., F. Raschke, and E.G.C. Troost, MR Image Changes of Normal-Appearing Brain Tissue after Radiotherapy. Cancers (Basel), 2021. 13(7). 66. 66.Billiet, T., I. Elens, C. Sleurs, A. Uyttebroeck, R. D’Hooge, J. Lemiere, and S. Deprez, Brain Connectivity and Cognitive Flexibility in Nonirradiated Adult Survivors of Childhood Leukemia. J Natl Cancer Inst, 2018. 110(8): p. 905–913. 67. 67.Raghubar, K.P., E.M. Mahone, K.O. Yeates, K.M. Cecil, M. Makola, and M.D. Ris, Working memory and attention in pediatric brain tumor patients treated with and without radiation therapy. Child Neuropsychol, 2017. 23(6): p. 642–654. 68. 68.Krogsrud, S.K., A.M. Fjell, C.K. Tamnes, H. Grydeland, L. Mork, P. Due-Tonnessen, A. Bjornerud, C. Sampaio-Baptista, J. Andersson, H. Johansen-Berg, and K.B. Walhovd, Changes in white matter microstructure in the developing brain--A longitudinal diffusion tensor imaging study of children from 4 to 11years of age. Neuroimage, 2016. 124(Pt A): p. 473–486. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.neuroimage.2015.09.017&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26375208&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 69. 69.Jones, R.M. and S.S. Pattwell, Future considerations for pediatric cancer survivorship: Translational perspectives from developmental neuroscience. Dev Cogn Neurosci, 2019. 38: p. 100657. 70. 70.Boutzoukas, E.M., J. Crutcher, E. Somoza, L.N. Sepeta, X. You, W.D. Gaillard, G.L. Wallace, and M.M. Berl, Cortical thickness in childhood left focal epilepsy: Thinning beyond the seizure focus. Epilepsy Behav, 2020. 102: p. 106825. 71. 71.King, D.J., K.R. Ellis, S. Seri, and A.G. Wood, A systematic review of cross-sectional differences and longitudinal changes to the morphometry of the brain following paediatric traumatic brain injury. Neuroimage Clin, 2019. 23: p. 101844. 72. 72.Zelko, F.A., H.R. Pardoe, S.R. Blackstone, G.D. Jackson, and A.T. Berg, Regional brain volumes and cognition in childhood epilepsy: does size really matter? Epilepsy Res, 2014. 108(4): p. 692–700. 73. 73.Foulon, C., L. Cerliani, S. Kinkingnéhun, R. Levy, C. Rosso, M. Urbanski, E. Volle, and M. Thiebaut de Schotten, Advanced lesion symptom mapping analyses and implementation as BCBtoolkit. Gigascience, 2018. 7(3): p. 1–17. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/gigascience/gix120&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29300887&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom) 74. 74.Sleurs, C., J. Blommaert, D. Batalle, M. Verly, S. Sunaert, R. Peeters, J. Lemiere, A. Uyttebroeck, and S. Deprez, Cortical thinning and altered functional brain coherence in survivors of childhood sarcoma. Brain Imaging Behav, 2021. 15(2): p. 677–688. 75. 75.Sleurs, C., S. Jacobs, S.J. Counsell, D. Christiaens, J.D. Tournier, S. Sunaert, K. Van Beek, A. Uyttebroeck, S. Deprez, D. Batalle, and J. Lemiere, Brain network hubs and cognitive performance of survivors of childhood infratentorial tumors. Radiother Oncol, 2021. 161: p. 118–125. 76. 76.Jacola, L.M., M. Partanen, J. Lemiere, M.M. Hudson, and S. Thomas, Assessment and Monitoring of Neurocognitive Function in Pediatric Cancer. J Clin Oncol, 2021. 39(16): p. 1696–1704. 77. 77.Ailion, A.S., T.Z. King, L. Wang, M.E. Fox, H. Mao, R.M. Morris, and B. Crosson, Cerebellar Atrophy in Adult Survivors of Childhood Cerebellar Tumor. J Int Neuropsychol Soc, 2016. 22(5): p. 501–11. 78. 78.Palmer, S.L., J.O. Glass, Y. Li, R. Ogg, I. Qaddoumi, G.T. Armstrong, K. Wright, C. Wetmore, A. Broniscer Gajjar, and W.E. Reddick, White matter integrity is associated with cognitive processing in patients treated for a posterior fossa brain tumor. Neuro Oncol, 2012. 14(9): p. 1185–93. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/neuonc/nos154&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=22898373&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301212.atom)