Cognitive, language, and school performance in children and young adults treated for low‐grade astrocytoma in the posterior fossa in childhood

Abstract Background Pilocytic astrocytoma is the most common brain tumour type in childhood located in the posterior fossa, and treated mainly with surgery. These tumours have low mortality, but knowledge concerning its long‐term outcome is sparse. Aim The aim of this study was to investigate whether children treated for pilocytic astrocytoma in the posterior fossa had late complications affecting cognition, language and learning. Methods This descriptive single‐centre study includes eight children and 12 adults treated as children for pilocytic astrocytoma in the posterior fossa, with a mean follow‐up time of 12.4 (range 5–19) years. Well‐established tests of intelligence, executive, language and academic function were used. Results Intelligence tests showed average results compared with norms. Five patients scored <−1 SD (70–84) and 3 low average (85–92) on full scale IQ. The patients scored average on subtests regarding executive function, except for significantly lower results in inhibition/switching (p = .004). In Rey complex figure test half of the patients scored below −1 SD. Language tests were normal except for significantly lower results in naming ability (p = .049) and in inference (p = .046). In academic tests, results were average, except for significantly lower results in reading speed (p = .024). Patients with learning difficulties performed worse in the tests. Conclusions The patients' functional outcome was favourable but, a not‐negligible part of the patients displayed neurocognitive difficulties as revealed by extensive neuro‐cognitive and academic testing. Thus, it is important to identify those in need of more thorough cognitive and pedagogic follow‐up programmes, including school interventions.


| INTRODUCTION
The yearly incidence of brain tumours in Sweden during  was 4.2/100 000 in children younger than 15 years. Survival rates have improved but vary across different tumour types. 1 There are also variations regarding medical, cognitive, and psychological long-term outcome. This may be caused by the tumour itself or by the treatment. 2 Long-term consequences are mainly reported among children treated for high-grade tumours, 3 but in an earlier study, where all children treated for brain tumours at Uppsala University Children's hospital 1995-2006 were studied, we found physical and cognitive complications also among children treated for low-grade tumours. 4 There is controversy regarding the role of tumour location. Studies have demonstrated that supratentorial tumours entail higher morbidity than infratentorial tumours. 5 In a study by Patel, 6 the supratentorial and infratentorial tumour location groups did not differ regarding intellectual functioning. However, survivors of infratentorial tumours performed worse on selected measures of intellectual functioning and on parent-reported social-emotional functioning.
The cerebellum has an important role in coordinating different neurological functions. Injuries may affect executive function, spatial cognition, personality, and language, 7-9 as well as fine- 10 and grossmotor performance. 11 A cerebellar cognitive affective syndrome (CCAS) has been described, characterised by deficits in executive function, spatial cognition, linguistic processing, and affect regulation. 7,9 The post-operative paediatric cerebellar mutism syndrome (pCMS), defined in a consensus paper by Gudrunadottir is another known complication after posterior fossa surgery. 12 It typically manifests 1-2 days after surgery and affects speech, emotions, personality, and behaviour. 13 The mutism is always transient, but recovery may be prolonged. 12 pCMS affects mainly children treated for medulloblastoma but may also occur in children treated for ependymoma or pilocytic astrocytoma. 13,14 Pilocytic astrocytoma is the most common brain tumour type in childhood located in the posterior fossa, mainly treated with surgery. 5,15 These tumours have a low mortality and the long-term functional outcome is favourable. 5 Although there are studies describing neurological, cognitive, and behavioural complications among these patients, there is still a lack of knowledge concerning the long-term outcome. [16][17][18][19][20][21][22] In one study, Traunwieser reported that all patients treated for low-grade glioma (LGG) were at risk of experiencing long-term cognitive impairments. These deficits occurred even following complete/ subtotal resection of cerebellar LGGs. 23 In earlier studies, we found a favourable clinical outcome, but some patients reported neurological complications and learning difficulties which were unmet in school. 4,24 Against this background of ambiguous results of long-term functional outcome for children treated for low-grade pilocytic astrocytoma in the posterior fossa, our aim was to investigate whether these patients had complications affecting intelligence, executive function, memory, language and academic performance and if some fulfilled the criteria for CCAS or suffered from pCMS. We also wanted to investigate if the patients had learning difficulties and if this had led to increased educational support.

| Participants
This single-center study was performed 2015-2017 at Uppsala University Children's Hospital, Sweden, a tertiary referral center for children with tumours in the central nervous system, serving six counties in mid-Sweden with a population of 1.7 million people.
Patients were retrieved from the local and the National Brain Tumour Registry. A total of 27 patients <18 years of age with a low-grade astrocytoma in the posterior fossa diagnosed and treated in childhood 1995 through 2011 were identified. At the time of this investigation, nine were still children (9-17 years) and 18 young adults (21-33 years).
All patients had at least 5 years of follow-up after the end of treatment.
This study includes the same patient group as a former study from our research group 24 except two adults who only participated in telephone interviews. Three patients did not answer several invitations, and two declined participation. Thus, 20 patients agreed to take part (12 adults and eight children; 74%). The mean age at tumour presentation was 8.3 years (SD 4.3), and the mean age at participation in the study was 20.2 (SD 7.3) years, range 9-33 years. Mean time from diagnosis to participation was 12.2 (SD 4.6) years.
All patients were treated surgically, and in 17 patients the operation was considered as a complete resection. Three patients had a remaining tumour and were re-operated shortly after the initial operation. Another three patients relapsed; one was treated with re-surgery, one with re-surgery and chemotherapy and one with gamma knife radio surgery. These three patients did not differ from the rest although the patient treated with only re-surgery reported learning difficulties and scored low average in full-scale intelligence quotient (FSIQ).

| Methods
Upon acceptance, participants were asked to come to the Folke Bernadotte Regional Rehabilitation Centre to undergo investigations performed by a multi-professional team, 2 days for adults and 3 days

| Test of visuospatial construction skills and visuospatial memory
Rey complex figure test (RCT) was used to asses visuospatial construction and memory, 30 using two parts of the test, immediate and delayed recall. Results were calculated into T-scores (M 50, SD 10) with American norms.

| Tests of language functions
Clinical evaluation of language fundamentals, fourth edition (CELF-4) 31 is an instrument for identifying language difficulties and disorders in children 5-12:11 years. Five indices are included in this study: Core language score (general language ability), receptive language, expressive language, language content, and language memory. Results were Tests of word fluency (FAS, animal and verb fluency) measure the subject's ability to generate words within a restricted time limit and within a certain category. In this study, the letter (or phonemic) fluency test called FAS (using the initial letters F, A, and S) was used 34 together with a semantic (or category) fluency test aimed at semantic categories (animals) and actions (verbs). 34

| Tests of academic performance
Diagnostic manual for analysis of reading and writing skills is a screening instrument for assessing spelling, word knowledge, reading velocity, and reading comprehension. 35 Swedish normative data exist for children aged 7-17 years.
Adler screening tools for mathematical skills is a test of mathematic performance in arithmetic for children 7-19 years. 36 All results were converted into the stanine scale where the normal distribution is divided into nine steps (M 5, SD 1.97).

| STATISTICS
Statistics were calculated using the SPSS statistical program, version 26.
One sample Wilcoxon signed rank test was used due to the small sample size and unknown distribution. Spearman's rank correlation coefficient was calculated to investigate the association of age at diagnosis with neurocognitive, language, and academic performance. The level of significance was set at p < .05. All analyses were exploratory and nominal significance levels are reported without adjustment for multiplicity.

| Test of visuospatial construction skills and visuospatial memory
All patients performed RCT, but two children had very low results on both subtests that did not yield any valid T-scores and were therefore not included when the mean scores for the sample were calculated.    In the tests of executive functioning, four had results <À2 SD and 1 < À1 SD in colour-word inhibition, and the result for inhibition/ switching was significantly below average. This indicates difficulties with inhibition and cognitive flexibility, which are parts of the executive processes responsible for purposeful goal-directed behaviour and the ability to perform novel or complex tasks. 38 Even if the mean results in RCT was in the normal range, 50%-45% of the patients had results < À1 SD in immediate and delayed recall respectively, including two children with very low results. This can indicate difficulties with visuospatial construction and memory. In addition, RCT can provide information about executive performance, for example cognitive flexibility, organisation, and working memory. 39 Thus, our findings suggest impaired executive functioning. This is in line with findings reported by Koustenis who found that children treated for both low-and high-grade tumours in the cerebellum exhibited similar patterns of impairment in executive functioning. Although deficits were larger in those treated for high-grade tumours, both groups had deficits affecting forward thinking, inhibition and mental flexibility. 40 Although BRIEF did not show any significant results in the different indices and no participant had results above the cut off for executive difficulties, 54% of the adults reported difficulties with planning/ Eight patients reported learning difficulties in the interviews.

| Tests of language function
Among these we found the lowest test results, which stress the importance of performing cognitive tests in order to identify patients with potential learning difficulties. In a study by Lönnerblad, 41 patients treated for brain tumours in childhood performed worse in Swedish, mathematics, and English compared with controls. Interestingly, there were no differences in outcome between survivors treated for high-or-low-grade tumours. This strengthens our findings of cognitive difficulties in some of the investigated patients. This pattern of function loss may be due to lesions to cerebro-cerebellar circuits following treatment of cerebellar tumours. 7 We found no correlations between age at diagnosis and the results of the intelligence tests. This is in accordance with a study by Pletschko. 22 However, we found that age at diagnosis and reading speed were significantly correlated. This suggests that young children with an immature brain are more susceptible to damage, which may lead to learning difficulties. 41 There are contradictory results in the literature concerning the association between tumour location in the cerebellum and neurocognitive outcome. 37 In a study by Khajuria,37 there was no significant association between tumour location and neurocognitive outcome, which was also reported by Beebe. 19 Due to our small sample, we could not investigate this.
Patients treated for pilocytic astrocytoma in the posterior fossa have a favourable clinical outcome. However, 40% of the participants in our study have reported considerable learning difficulties that could have been detected by cognitive and academic testing. These learning difficulties were not dealt with in school, and further academic support would have been warranted.

| CONCLUSIONS
The long-term functional outcome for children treated for low-grade astrocytoma in the posterior fossa is favourable. However, some patients have cognitive and academic difficulties which can be revealed by neuro-cognitive and academic testing, including interviews of academic performance in school. Therefore, it is imperative to identify those in need of more thorough cognitive follow-up programmes, including interventions in school. Moreover, there is a need to establish appropriate collaboration between paediatric neuro-oncology clinics and the educational system.

ACKNOWLEDGMENTS
This study was supported by grants from the Barncancerfonden, the Hedström Foundation and the Gillbergska stiftelsen. Financial support was also provided through a regional agreement on medical training