Comparison of Neuropsychological Functioning in Pediatric Posterior Fossa Tumor Survivors: Medulloblastoma, Low-Grade Astrocytoma, and Healthy Controls

Abstract

Background:

Neuropsychological comparison of medulloblastoma (MB) and cerebellar low-grade astrocytoma (LGA) survivors to controls can clarify treatment-related neurocognitive late effects. While both brain tumor groups undergo surgery to the posterior fossa, children with MB additionally receive craniospinal irradiation with boost and chemotherapy. This study provides an updated comparison of neuropsychological functioning in these two groups and examines effects of demographic risk factors upon outcomes.

Procedure:

Forty-two children (16 MB, 9 LGA, 17 controls) completed measures of intellectual functioning, verbal learning/memory, visual-motor integration, and fine motor functioning. The effects of age at diagnosis, time since diagnosis, gender, fatigue, and social status on neuropsychological functioning were examined.

Results:

MB survivors demonstrated the worst neurocognitive late effects, but they were less severe and extensive than in prior studies. LGA survivors’ mean scores were below normative expectations in working memory, processing speed, and fine motor functioning. In this overall sample, processing speed difficulties were independent of fine motor functioning and fatigue. Higher parental education was associated with better intellectual functioning, working memory, delayed recall, and visual-motor integration. Neuropsychological function was not associated with gender, age at diagnosis, or time since diagnosis.

Conclusion:

The results support that contemporary treatment approaches with craniospinal irradiation plus boost and chemotherapy confer the greatest risk for late effects, while surgical resection is associated with subtle but important neurocognitive difficulties. Ultimately, this study furthers our understanding of factors impacting neuropsychological function in pediatric MB and LGA survivors and contributes to empirical support for close monitoring and targeted interventions into survivorship.

Introduction

Central nervous system (CNS) tumors are the second most common form of cancer in children, accounting for 21% of all pediatric cancer diagnoses.¹ In children, two-thirds of all CNS tumors occur in the posterior fossa region. The two most common posterior fossa tumor types include medulloblastoma (40%), a malignant tumor, and cerebellar astrocytoma (20–35%), a low-grade tumor.² With improvements in medical treatments, survival rates for children with posterior fossa tumors has increased significantly, with the current five-year survival rate for medulloblastoma (MB) ranging from 60–80% and > 90% for cerebellar low-grade astrocytoma (LGA).³ Accordingly, there has been a growing number of studies on the neurocognitive late effects of posterior fossa tumor diagnosis and treatment, which is important for informing interventions to improve academic achievement⁴ and quality of life⁵, and can help inform expectations from treatment for medical providers, survivors, and parents.

Neuropsychological comparison of childhood MB and LGA survivors to controls is a useful method for examining treatment-related neurocognitive late effects.⁶ Both tumor groups undergo surgical resection of the posterior fossa tumor, but only MB requires additional multimodal therapy, including craniospinal irradiation (CSI) plus irradiation boost to the posterior fossa region and chemotherapy. Comparison of childhood MB to LGA survivors provides the opportunity to clarify neurocognitive late effects specific to CSI, posterior fossa irradiation boost, and chemotherapy, while controlling for effects of brain tumor location and surgery.

To our knowledge, seven studies have directly compared aspects of neuropsychological function in these two tumor groups, with MB survivors experiencing more pronounced neurocognitive late effects than LGA survivors.^6,7 However, these studies present limitations. Participants were recruited prior to 2005, which indicates treatment on older protocols that utilized higher dosages of CSI.^8–14 In studies with recruitment before 1991 that used CSI dosages of over 3200 cGy, the MB survivors displayed very low intellectual functioning and performed over one standard deviation below the LGA group.^8–10 In contrast, this study’s recruitment began in 2015, when MB treatment protocols utilized lower dosages of CSI (1800 or 2340 cGy), narrower volume and lower dose of boost to the tumor bed, and in some cases the use of proton beam irradiation,^15,16 which have all been demonstrated to be less neuro-toxic to white matter in the developing brain^17–19 and may be associated with fewer neurocognitive late effects.²⁰ Newer surgical approaches to posterior fossa tumor resection, such as cerebellomedullary fissure opening techniques and frameless based stereotaxy, have improved the efficacy of, and reduced complications of, surgical resection.²¹ Accordingly, the pattern and magnitude of differences between MB and LGA groups may differ for survivors treated with more recently developed protocols.

Furthermore, the number of posterior fossa tumor survivors examined in the prior comparative literature is small, as several manuscripts were published using the same samples.^11–14 Some studies also did not use a pure MB sample, thereby introducing more diversity in terms of tumor characteristics, treatment protocols, and medical complications.^13,14

Notably, of the studies utilizing independent samples and pure MB groups, most have not included a control group,^9,10 which can clarify the effects of surgery on neurocognitive outcomes for LGA survivors. Specifically, lesions to the cerebellum from surgical resection can affect neurocognitive functioning through the disruption of essential cerebello-thalamic-cerebral feed-forward and feedback circuitry.^22,23 In prior studies that compared LGA survivors to control groups, the LGA groups performed worse on measures of attention, executive functioning, and memory.^11,12,14 Overall, the current literature supports that survivors of LGA display less severe deficits than MB survivors but still perform worse than controls. This study aimed to update that literature by examining survivors who underwent newer treatment protocols, utilizing a pure MB sample, and control group.

The present study investigated the potential effects of demographic risk factors on neuropsychological functioning. While younger age at diagnosis and increased time since treatment are consistently associated with worse neurocognitive outcome in MB survivors,⁷ these factors are often not predictive of late effects in LGA survivors.²⁴ The role of gender is less clear, with some studies finding that female gender is associated with worse outcomes in pediatric brain tumor survivors^25,26, while other studies have not supported this finding.^27,28 Although fatigue is known to be associated with cognitive difficulties in adult cancer survivors,²⁹ the effect of fatigue on neuropsychological functioning has not been extensively examined in pediatric cancer survivors. For MB survivors, proxies for higher socioeconomic status (SES), such as marital status and parental education, have been associated with better attention and working memory abilities following surgical resection alone.^30,31

The current study’s primary aim was to provide an updated and comprehensive assessment of the neurocognitive functioning of these two discrete posterior fossa tumor groups compared to age matched healthy controls. We hypothesized that the MB group would perform worse on measures of neurocognitive functioning than LGA survivors, both of whom would perform worse than controls. Our secondary aim was to further understand treatment-related and demographic late effects in these two tumor groups. We hypothesized that lower age at diagnosis and increased time since diagnosis would negatively impact neurocognitive functioning in the MB group but not the LGA group. Lastly, in conjunction with our secondary aim, we performed exploratory analyses of previously mixed or under-examined risk factors (gender, fatigue, and SES) on neuropsychological functioning.

Methods

Participants.

Participants were recruited from 2015 – 2020 from four academic medical centers: Columbia University Medical Center, Children’s Medical Center Dallas, Seattle Children’s Hospital, and Memorial Sloan Kettering Cancer Center. Sample included children who successfully completed treatment for either non-metastatic (M0) posterior fossa MB or posterior fossa LGA without relapse or evidence of residual disease, as well as age-matched healthy controls. Participants were six to sixteen years of age at the time of enrollment, and tumor survivor participants were one to ten years post-completion of medical treatment for their posterior fossa brain tumor. Per study inclusion criteria, participants spoke fluent English. Institutional Review Board approval was obtained by each participating center. Tumor survivor participants were recruited by treatment teams, principal investigators, or research staff, and healthy controls were recruited by flyers and word of mouth from parents and staff. Three healthy controls were siblings of tumor survivor participants. A parent of each participating child provided informed written consent, and assent from the child was obtained.

Procedures.

Survivors of MB received standard treatment at their respective institutions, which was required to consist of gross total resection, craniospinal irradiation with conformal posterior fossa boost radiotherapy, and chemotherapy for inclusion in the study. Seven participants in the MB group received proton beam irradiation, and nine received photon beam irradiation. Survivors of LGA were treated with gross total resection only. All survivors were assessed at least one-year post-treatment by licensed psychologists.

Measures.

Participants completed reliable and valid neuropsychological measures, which provide age-adjusted normative scores used in the analyses. Supplementary Table 1 provides a full description of the measures. Intellectual functioning was assessed with the Wechsler Intelligence Scale for Children, Fifth Edition. Verbal learning and memory were assessed with the California Verbal Learning Test, Children’s Version. Visual-motor integration was assessed with the Beery-Buktenica Test of Visual-Motor Integration. Fine motor functioning was assessed with the Grooved Pegboard or Purdue Pegboard. To assess fatigue, participants completed the self-report PedsQL Multidimensional Fatigue Scale (range 0–100; higher scores indicate lower fatigue). Parents completed the Barratt Simplified Measure of Social Status as a proxy for SES (range 8–66; higher scores indicate higher social status).

Statistical Analyses.

Most neuropsychological measures were normally distributed except for verbal memory recognition and bilateral fine motor functioning. One-way ANOVAs were used to compare the demographic and medical characteristics of the groups. Then, one-way ANOVAs/Kruskal-Wallis tests were conducted to assess for differences in neuropsychological function by diagnostic group, and post-hoc Tukey Honestly Significant Difference/Nemenyi tests (depending on normality of the variable) to examine individual comparisons. Both p values and effect sizes represented by eta-squared (η²) are reported, as well as p values following adjustment using the false discovery rate.³² Further, the percent of children within each diagnostic group who scored at or below the low average range was calculated (i.e., ≤25th percentile based on the normal distribution curve). Finally, Pearson correlation coefficients or independent sample t-tests/Wilcoxon signed-rank tests were used to examine the associations between demographic risk factors (age at diagnosis, time since diagnosis, gender, fatigue, SES) and neuropsychological function. Significant correlations were controlled for in relevant analyses using ANCOVAs.

Results

The sample was comprised of 42 participants: 16 survivors of MB, 9 survivors of LGA, and 17 age-matched healthy controls. Table 1 provides a summary of the participant characteristics by diagnostic group. On average, the participants were 12.3 years of age (SD = 2.8) and the tumor groups were 4.3 years post-diagnosis (SD = 3.2). There were no significant group differences in terms of age at assessment, general fatigue, parental education, parental occupational level, handedness, or gender. The tumor groups did not differ significantly in terms of age at diagnosis or time since diagnosis.

TABLE 1.

Demographic Characteristics of Diagnostic Group presented as mean (SD) or n (%)

	Medulloblastoma (n=16)	Low-grade Astrocytoma (n=9)	Healthy Control (n=17)	p
Sex (Female)	11 (69%)	6 (67%)	8 (47%)	.40
Age at Diagnosis (Years)†	7.00 (3.44)	6.67 (3.08)	-	.81
Age at Assessment (Years)†	12.39 (2.80)	11.94 (3.28)	12.49 (2.76)	.90
Time Since Diagnosis (Months)	45.38 (35.38)	62.89 (43.48)	-	.29
Handedness (Right)	11 (69%)	8 (89%)	13 (77%)	.53
General Fatiguea	65.6 (20.5)	71.3 (15.1)	73.2 (13.6)	.44
General Social Statusb	47.4 (11.7)	49.7 (11.0)	49.6 (13.1)	.08
Parental Education Levelb	15.8 (4.2)	17.3 (3.5)	17.5 (3.3)	.40
Parental Occupation Levelb	31.6 (8.1)	32.5 (8.3)	31.9 (11.6)	.98

Note.

^ahigher score = lower fatigue

^bhigher score = higher social status.

^†Indicates analyses only in tumor group (n=25).

Comparison of Neuropsychological Function by Diagnostic Group.

Mean neuropsychological test scores by diagnostic group and mean group comparisons are listed in Table 2. For most measures, the group means reflect a stepwise pattern wherein MB survivors performed worse than LGA survivors, who in turn performed worse than controls (see Figure 1), although not all individual comparisons reached statistical significance. A one-way ANOVA revealed that the effect of diagnostic group on overall intellectual functioning was significant (F[2,39] = 5.20, p = .010, η² = 0.21). Post hoc analyses indicated that the MB group (M = 94.75, SD = 14.79) performed significantly lower than the healthy control group (M = 109.88, SD = 12.39), whereas this was not true for the LGA group (M = 101.67, SD = 13.03). Amongst the index scores that comprise overall intellectual functioning, there was a significant effect of diagnostic group on verbal reasoning (p = .003, η² = 0.26), visual spatial reasoning (p = .016, η² = 0.19), and processing speed (p = .001, η² = 0.31), and a trend-level effect for working memory (p = .06, η² = 0.14) and fluid reasoning (p = .07, η² = 0.12). These findings were in the medium-to-large effect size range and were consistent following adjustment using the false-discovery rate.

TABLE 2.

Mean Group Comparisons of Neuropsychological Function among Healthy Controls (n=17), Low-Grade Astrocytoma (n=9), and Medulloblastoma (n=16)

	HC		LGA		MB		One-way ANOVA/Kruskal-Wallis			Tukey HSD/Nemenyi Test
	M	SD	M	SD	M	SD	p	FDR	η2	HC-LGA	HC-MB	LGA-MB
Intellectual Functioninga
Full Scale IQ	109.9	12.4	101.7	13.0	94.8	14.8	.01*	.03*	0.21	.31	.007*	.44
Verbal Comprehension	110.1	12.9	104.1	12.5	93.1	13.7	.003*	.01*	0.26	.51	.002*	.12
Visual Spatial	103.8	12.7	107.4	10.4	93.6	12.5	.02*	.04*	0.19	.75	.06	.03*
Fluid Reasoning	111.7	9.8	103.6	15.2	100.6	16.7	.07	.10	0.12	.34	.07	.86
Working Memory	107.3	13.8	98.3	14.3	94.7	16.4	.06	.10	0.14	.32	.052	.82
Processing Speed	104.8	14.8	93.8	14.6	81.7	17.7	.001*	.01*	0.31	.23	.001*	.18
Verbal Memory (CVLT)b
Total Learning	0.6	0.7	0.2	0.9	−0.2	1.0	.04*	.09	0.15	.48	.03*	.56
Short-Delay Recall	0.6	0.8	0.1	0.9	0.3	0.8	.29	.34	0.06	.30	.53	.82
Long-Delay Recall	0.5	0.8	0.2	1.0	0.4	0.7	.66	.66	0.02	.63	.94	.81
Recognition	0.3	0.6	0.3	1.0	0.4	0.6	.56	.61	−0.02	.69	.66	.99
Visual Motor Integrationa	93.5	14.2	100.2	8.7	85.4	19.9	.08	.10	0.12	.57	.33	.08
Fine Motor Functioningb
Dominant	−0.5	1.7	−0.5	1.3	−1.9	1.8	.04*	.09	0.11	.94	.04*	.22
Non-Dominant	−0.6	1.0	−1.3	2.1	−4.5	4.3	.003*	.01*	0.25	.86	.003*	.06

Note. HC = Healthy Control, LGA = low-grade astrocytoma, MB = medulloblastoma, FDR = False Discovery Rate, IQ = Intellectual Quotient, CVLT = California Verbal Learning Test for Children. Superscript indicates data scale:

^aStandard Score, higher score = better performance

^bz-score, higher score = better performance.

^*Indicates statistically significant difference, p < .05.

Bold indicates a medium to large effect size (η² ≥ 0.06)

FIGURE 1.

Box and Whisker Plot of Intellectual Functioning by Diagnostic Group

There was also a significant effect of diagnostic group on immediate verbal learning of a word list over five trials (F[2,39] = 3.41, p = .043, η² = 0.15), whereas there was no effect for short- and long-delay free recall and recognition. In addition, there was a significant effect of diagnostic group on bilateral fine motor functioning (Dominant hand: H[2] = 6.12, p = .047, η² =0.11; Non-dominant hand: H[2] = 11.73, p = .003, η² =0.25). There was a trend-level effect of diagnostic group for visual motor integration (F[2,39] = 2.64, p = .08, η² = 0.12). For significant ANOVAs, post hoc individual comparisons showed that the MB group performed significantly worse than the control group but not the LGA group, except for visual spatial functioning, in which the MB group performed worse than the LGA group but not the control group.

We performed exploratory analyses of the relationship between fine motor functioning, visual motor integration, and processing speed given that the processing speed measures utilized required fine motor skills. In the whole sample, processing speed was correlated with fine motor functioning in the dominant (r = .35, p = .02) and non-dominant hand (r = .49, p = .001), as well as visual motor integration (r = .39, p = .01). When fine motor functioning and visual motor integration were controlled for together using an ANCOVA, the relationship between diagnostic group and processing speed remained significant (F[1,4] = 4.93, p = .002; diagnosis, p = .03), whereas the covariates were not significant.

Rate of Below-Average Scores on Neuropsychological Measures by Diagnostic Group.

Figure 2 displays the obtained rate of scores below the mean in each diagnostic group, defined as being ≤25th percentile based on the normal distribution curve. The MB group had high rates of scores below the mean across nearly all domains, with over 60% of scores in the below average range for processing speed, visual motor integration, and fine motor functioning. The LGA group had elevated rates of scores below the mean for working memory (33%), processing speed (44%), and fine motor functioning in the non-dominant hand (33%). The control group performed better than the population rate in most domains, except for visual motor integration (30%) and fine motor functioning in the non-dominant hand (35%).

FIGURE 2.

Rate of Below Average Scores (≤25%) on Neuropsychological Measures by Diagnostic Group

Note. FSIQ = Full Scale Intellectual Quotient, SD = short delay, LD = long delay, FMF = fine motor functioning.

Demographic Risk Factors and Neuropsychological Function.

As shown in Table 3, in the tumor groups, neither age at diagnosis nor time since diagnosis was significantly related to any neuropsychological variables following correction for multiple analyses. There was no relationship between gender and neuropsychological functioning in the entire sample. In contrast, in the entire sample, general fatigue was correlated with processing speed (r = .36, p = .02); as overall fatigue increased, processing speed decreased. Given that this was an exploratory hypothesis, the p-value was not adjusted using the false discovery rate. When fatigue was controlled for using an ANCOVA, the relationship between diagnosis and processing speed remained significant (F[1,3] = 6.88, p = .001; diagnosis, p = .004). Regarding SES, parental education was correlated with intellectual functioning (r = .37, p = .009), working memory (r = .49, p = .001), delayed recall (r = .33, p = .03), and visual-motor integration (r = .48, p = .001); as parental education increased, neuropsychological function increased. When parental education was controlled for using an ANCOVA, the relationship between intellectual functioning and diagnosis remained significant (F[1,3] = 5.36, p = .004; diagnosis, p = .03). However, the relationship between diagnosis and working memory (F[1,3] = 6.20, p = .002; diagnosis, p = .07) and visual-motor integration (F[1,3] = 5.53, p = .003; diagnosis, p = .15) was no longer significant when controlling for parental education. Parental occupational level was not significantly related to neuropsychological function.

TABLE 3.

Correlations Among Demographic Risk Factors and Neuropsychological Function in Total Sample (n=42)

	Age at Diagnosis†		Time since Diagnosis†		Gender		General Fatigue¥		Parental Education¥		Parental Occupation Level¥
	r	p	r	p	t	p	r	p	r	p	r	p
Intellectual Functioning
Full Scale IQ	−.12	.57	.27	.20	1.22	.23	.10	.53	.37	.009*	0.21	0.21
Verbal Comprehension	−.13	.53	.24	.25	1.47	.15	.04	.78	.33	.11	0.19	0.37
Visual Spatial	−.12	.58	.27	.18	0.81	.42	.07	.69	.14	.24	−0.12	0.40
Fluid Reasoning	.12	.56	.06	.76	1.20	.24	.02	.46	.24	.054	0.13	0.39
Working Memory	.06	.77	.25	.23	1.78	.08	.47	.29	.49	.001*	0.31	0.08
Processing Speed	−.17	.41	.42	.04; FDR= .75	0.24	.81	.36	.02*	.18	.20	0.15	0.52
Verbal Memory (CVLT)
Total Learning	.02	.93	.05	.82	0.01	.99	.30	.06	.20	.15	0.15	0.22
Short-Delay Recall	−.12	.56	.12	.57	−0.64	.52	.19	.24	.16	.39	0.06	0.57
Long-Delay Recall	−.06	.78	.01	.98	−1.75	.09	.04	.79	.33	.03*	0.16	0.31
Recognition	−.28	.18	.17	.42	−0.56	.82	.27	.09	.04	.45	0.14	0.49
Visual Motor Integration	−.22	.29	.32	.12	0.49	.63	.14	.38	.48	.001*	0.30	0.10
Fine Motor Functioning
Dominant	.13	.56	.04	.85	0.28	.78	.30	.06	−.12	.64	−0.09	0.27
Non-Dominant	.05	.81	.21	.32	0.13	.90	.15	.36	−.12	.09	−0.08	0.29

Note. Pearson and Spearman Rank correlations were conducted depending on normality of the variable. Independent t-tests and Wilcoxon Signed-Rank Tests were conducted for Gender. FDR = False Discovery Rate, IQ = Intellectual Quotient, CVLT = California Verbal Learning Test for Children.

^†Indicates analyses only in tumor group (n=25).

^¥P-values were not adjusted using the FDR for Fatigue and Social Status, as they were exploratory hypotheses.

^*Indicates statistically significant difference, p < .05.

Discussion

The present study found that survivors of MB demonstrated greater deficits than controls, while LGA survivors performed within normal limits but still lower than population norm expectations within the domains of working memory, processing speed, and fine motor functioning. Additionally, in the entire sample, processing speed difficulties were independent of fine motor functioning and fatigue, while better intellectual functioning, working memory, verbal memory recall, and visual-motor integration abilities were associated with higher parental education. Age at diagnosis, time since diagnosis, and gender were not associated with neuropsychological functioning. Overall, these findings support that medical treatment factors are important predictors of neurocognitive late effects in childhood posterior fossa survivors, with cranial irradiation and chemotherapy conferring greatest risk, while surgery alone can still negatively affect some essential neurocognitive domains.

As hypothesized, MB survivors displayed the most pronounced neurocognitive late effects, with significantly worse performance relative to controls on measures of intellectual functioning, verbal reasoning, processing speed, verbal learning, and bilateral fine motor functioning; non-significant trends were detected for fluid reasoning and attention/working memory. Contrasting existing research demonstrating widespread neurocognitive deficits in MB survivors,^7,30 most domains were within normal limits except for processing speed, visual-motor integration, and fine motor functioning. Regarding processing speed, deficits were independent of visual-motor integration and fine motor functioning, suggesting that slower performance on processing speed measures is not wholly attributable to fine motor deficits in MB survivors. Rather, processing speed deficits in this population likely reflect a slower rate of cognitive processing of information secondary to white matter injury from cranial irradiation and chemotherapy.^33,34 Among MB survivors, fine motor functioning was poor bilaterally but significantly impaired in the non-dominant hand, indicating greater difficulty with non-dominant (typically right) hemisphere functions. As has been seen in other studies of MB survivors,³⁵ difficulties in verbal memory were the result of problems in the initial learning of novel information rather than in the retrieval or retention of that information.

Our hypothesis regarding LGA survivors was partially supported. Most LGA survivors were functioning within the average range in many domains but overall means were generally lower than the control group. Moreover, there were elevated rates of below average scores compared to population norms in the domains of working memory, processing speed, and fine motor functioning. These findings align with the extant literature, indicating that LGA survivors may demonstrate difficulties with attention/working memory and processing speed.^24,36,37 Due to the reciprocal connections between the cerebellum and prefrontal cortex, cerebellar lesions likely disrupt access to cognitive control systems in the frontal-subcortical circuitry,³⁸ potentially leading to difficulties regulating attention and performing tasks quickly.

Although means for the MB group were lower than for the LGA group across most domains, these differences often did not reach statistical significance. A potential explanation is that MB survivors did not demonstrate the same level of neurocognitive impairment across domains as seen in prior studies due, at least in part, to use of lower CSI dosages. To illustrate, in numerous prior studies,^8–10 IQ was lower (ranging from 74 to 80) and CSI dosages were higher (> 3200 cGy) than in the present sample (mean IQ of 94.75, CSI dosages < 2340 cGy). However, samples from two of those prior studies were tested at a longer time since diagnosis (16 to 17 years)^9,10 than the present sample (4.3 years), and greater years post-treatment is typically associated with greater neurocognitive late effects. Additionally, domains clinically observed to demonstrate large differences between MB and LGA groups—such as processing speed, visual-motor integration, and fine motor functioning—did not reach statistical significance due to the sample size of the study. This is supported by trend-level differences for visual-motor integration and effect sizes in the medium to large range. The one domain in which the MB group demonstrated significantly worse performance than the LGA group was visual spatial reasoning.

Novel to the present study, fatigue was associated with slower processing speed in the overall sample, a common finding in the sleep literature³⁹ that has not been demonstrated in these populations. Although pediatric brain tumor survivors tend to have high levels of fatigue,⁴⁰ formal assessment of fatigue is not routinely included in neuropsychological assessments. Together with prior findings on the association between fatigue and lower quality of life in pediatric brain tumor survivors,⁴¹ future clinical care and research should include routine assessments of fatigue.

Higher parental education, a proxy for SES, was associated with better neurocognitive outcomes in the domains of intellectual functioning, working memory, verbal memory, and visual-motor integration. Parental education contributed to some neurocognitive outcomes above and beyond the role of diagnostic group, which is consistent with the strong relationship between SES and neurocognitive function in children without a significant medical history.⁴² Additionally, in survivors of pediatric brain tumors, SES has been found to be more predictive of neurocognitive outcomes than some demographic variables.³¹ The current findings support the emerging literature that SES is also an important demographic risk factor in pediatric brain tumor survivors.

The lack of an association between neuropsychological functioning and age at diagnosis or time since diagnosis in our sample of MB survivors is inconsistent with some previously published findings in this population.^7,43 However, such null findings are not uncommon in the MB literature,¹³ especially for survivors who had been treated with protocols aimed to minimize neurotoxicity and/or newer treatment protocols.^20,44 Consistent with our findings, most studies do not find an effect of age at diagnosis or time since diagnosis on neurocognitive outcomes in LGA survivors.^9,12,24 Also of note, our findings suggest that female gender was not a risk factor for neurocognitive late effects in posterior fossa brain tumor survivors.

Strengths of the present study included multi-site recruitment and administration of the same neuropsychological battery across geographically varied locations across the United States, as well as minimal missing data. However, there are several limitations to consider. The current sample size was small and may not have had sufficient power to detect all meaningful group differences. It is possible that with a larger sample, non-significant differences between the MB and LGA group may have reached statistical significance. In addition, due to sample size, the impact of irradiation modality (i.e., proton versus photon beam) on neurocognitive function was unable to be analyzed in the MB group. It is possible that less severe neurocognitive outcomes in the MB group relative to prior studies were due in part to neurocognitive sparing in the 7 out of 16 (44%) MB participants who received proton beam irradiation. However, the smaller sample size was a trade-off for having two discrete tumor groups in this pediatric population. By enrolling only children with non-disseminated MB and LGA in the posterior fossa region, tumor and surgical resection location could more confidently be controlled for. This aided assessment of the unique effects of different treatment factors on neurocognitive functioning. Another limitation is that data regarding radiation dosage and volume, as well as the specific chemotherapies received, were not collected. Further, incidence of medical complications (e.g., posterior fossa syndrome, hydrocephalus, shunt, hearing loss) were not collected and are risk factors for poorer neurocognitive outcomes.^4,45,46 In addition, information regarding race/ethnicity was not collected. The limited number of clinical and treatment variables collected leaves unanswered questions about the generalizability of the results. Therefore, future, prospective studies with larger samples should include collection of more treatment, medical, and demographic variables to address these limitations and to improve the generalizability of the current findings.

In conclusion, this study provides an updated and comprehensive comparison of neuropsychological functioning in survivors of the most common CNS tumors of childhood across multiple sites. It adds to the existing evidence that specific treatments for pediatric posterior fossa brain tumors are important predictors of neurocognitive late effects. While our MB survivors demonstrated worse neurocognitive late effects than LGA survivors, those late effects were less severe and extensive in our sample than in prior studies. In LGA survivors, surgical resection remains associated with mild neurocognitive difficulties. These findings underscore the critical need for serial neuropsychological assessment in both MB and LGA survivors, as even mild neurocognitive difficulties can have profoundly negative effects over time. Lastly, collecting SES data and assessing fatigue is warranted for future research. Ultimately, this study contributes to our understanding of late effects in pediatric posterior fossa tumor survivors and provides empirical support for serial monitoring of these survivors’ neurocognitive functioning and development to improve survivorship outcomes.

Abbreviations Key:

MB

Medulloblastoma

LGA

Low-grade astrocytoma

CNS

Central nervous system

CSI

Craniospinal irradiation

SES

Socioeconomic status

Data Availability Statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.