Attention and executive functioning in children and adolescents treated for high-risk acute lymphoblastic leukemia: A report from the Children’s Oncology Group (COG)

Abstract

Objectives:

Survivors of childhood B-acute lymphoblastic leukemia (B-ALL) are at risk for difficulties with attention and executive functioning (EF) as a late effect of treatment. The present study aimed to identify treatment and demographic factors associated with risk for difficulties with executive functioning in youth treated for high-risk B-ALL.

Method:

Children and adolescents with B-ALL treated on Children’s Oncology Group (COG) protocol AALL0232 were randomized to high-dose or escalating-dose methotrexate (MTX), and either dexamethasone or prednisone during the induction phase. Neuropsychological functioning was evaluated via protocol AALL06N1, including performance-based and parent-report measures, for 177 participants (57% female, 81% white; Mean age at diagnosis = 8.4 years; SD=5.0) 8–24 months following treatment completion.

Results:

Mean scores for all attention and EF measures were within the average range, with no significant differences as a function of MTX delivery or steroid treatment (all ps>.05). In multivariable models, participants with U.S. public insurance exhibited significantly greater parent-reported EF difficulties than those with U.S. private or non-U.S. insurance (ps≤ .05). Additionally, participants diagnosed under 10 years of age performed significantly more poorly on measures of attention (i.e., continuous performance task, ps≤. 05) and EF (i.e., verbal fluency and tower planning task, ps≤ .05).

Conclusions:

For survivors of pediatric B-ALL, treatment-related factors were not associated with attention or EF outcomes. In contrast, outcomes varied by demographic characteristics, including age and insurance type, an indicator of economic hardship. Future research is needed to more directly assess the contribution of socioeconomic status on cognitive outcomes in survivors.

Children and adolescents treated for high-risk acute lymphoblastic leukemia (ALL) are at risk for experiencing a variety of neurocognitive late effects, with significant implications for daily living. General intelligence may be impacted to some extent, with ALL survivors having mean IQ scores that are lower than peers without cancer, but typically within normative ranges.^1–4 However, while less than a quarter of survivors exhibit intellectual deficits, a much higher proportion of ALL survivors demonstrate clinically significant difficulties with attention, processing speed, and executive dysfunction.⁵

Executive functioning (EF) is an umbrella term referring to a set of higher-order cognitive skills involving inhibitory control, organization, time management, working memory, planning, and complex decision-making. These skills support academic learning, and are also linked to the ability to self-regulate behavioral and emotional responses.⁶ Attention skills are not typically included in EF, although they are closely related. The ability to attend to information is a critical prerequisite for much of the higher-order thinking involved in EF, and executive skills also help an individual to plan and execute attentional priorities.⁷ Because attention and EF skills are involved in so many aspects of children’s home, school, and social functioning, even mild to moderate difficulties can have wide-reaching impact on development, including susceptibility to mental and behavioral health problems, establishing healthy physical habits, and occupational success.^7–9 Notably, EF skills are among the last cognitive abilities to mature, as they rely on complex neurophysiological networks that develop into the third decade of life.¹⁰ Accordingly, even in the context of relatively stable, average intellectual profiles, EF problems may be more prevalent among survivors than among same-aged peers.^5,11 For those affected, these impairments place youth at risk for difficulties over time, including reduced academic performance,^12,13 adaptive functioning,^14,15 and quality of life.^16,17

There is growing consensus about the incidence and severity of EF and attention problems in survivors of ALL. However, discrepancies in the literature suggest that outcomes are highly variable, and may be dependent on specific treatment, host, and methodological factors. Therapies that include cranial radiation increase the risk of neurocognitive dysfunction^18,19, though dysfunction is commonly described in its absence^4,20,21. Host factors including female biological sex^4,22, younger age at diagnosis^1,11,23, and lower SES^24,25 also contribute to differences in neurocognitive outcomes. Importantly, both performance-based and parent-reported EF difficulties appear to be related to changes in both brain structure and function following treatment. Kesler and colleagues²⁶ recently documented reduced global efficiency as measured by structural and functional connectomes in survivors of ALL with weak EF. A recent meta-analysis of neurocognitive outcomes in survivors of ALL correlated changes in specific brain regions with impaired EF, including reduced grey and white matter volume in cortical and subcortical regions as well as functional changes in frontal areas and the hippocampus.²⁷

This wealth of existing literature allows us to conclude that EF and attention skills are almost certainly affected to some degree in at least a portion of survivors. However, a clearer understanding of how outcomes are moderated by demographic and treatment-related factors is needed in order to predict risks for late effects in individual patients. The impact of corticosteroid treatment is particularly variable in the literature. Specifically, the use of dexamethasone may confer greater risk, given its penetrance into the central nervous system.^28,29 In a 2014 meta-analysis, Warris and colleagues³⁰ concluded that, although half of included studies reported more neuropsychological difficulties following treatment with dexamethasone versus prednisone, particularly for long-term outcomes, differences were not likely to be clinically meaningful. However, very few outcomes assessing attention and EF were available for this analysis. More recently, Phillips and colleagues found that female ALL survivors treated with dexamethasone evidenced differences in brain structure³¹ and function,³² including reduced volume of subcortical structures and thinner cortices, compared with male survivors.

In an effort to more clearly characterize the impact of different treatment strategies on EF skills, the current study presents an analysis of a sample of survivors of pediatric high-risk B-ALL. We examined the effect of several potential risk factors, including different methotrexate (MTX) and corticosteroid treatment strategies and host characteristics, on both parent-reported and performance-based neurocognitive outcomes.²⁵ In our initial reports from this sample, we found that processing speed, intellectual functioning, academic outcomes, and adaptive functioning were unrelated to MTX delivery method and corticosteroid type,^25,33 though differences were reported in relation to insurance status and child age at diagnosis. Here, we examine a subset of participants who completed more detailed testing of attention and EF.

Methods

Children and adolescents diagnosed with high risk B-ALL were enrolled on Children’s Oncology Group (COG) protocol AALL06N1, designed to evaluate the neurocognitive impact of treatment delivered on COG treatment protocol AALL0232.³⁴ Per AALL0232, eligible patients were first randomly assigned to induction therapy including either 14 days of dexamethasone (10 mg/m²/day) or 28 days of prednisone (60 mg/m²/day). In addition, participants were randomized to an interim maintenance regimen including either four courses of high-dose (HD) MTX (5 g/m² / 24 hours) with leucovorin rescue or five courses of escalating-dose MTX (100–300 mg/m² /IV push) with PEG asparaginase but no leucovorin rescue (Capizzi). An increased rate of osteonecrosis in children ≥10 years of age who had been randomized to dexamethasone during induction led to non-random assignment of children 10 and older to prednisone during induction, beginning in 2008. Younger children continued to be randomized.

Eligibility criteria for AALL06N1 included enrollment on AALL0232 in first remission; age at diagnosis 1–18 years; and a primary language of English or Spanish. Exclusion criteria included pre-existing neurodevelopmental conditions (e.g., Down syndrome), significant sensory impairments, or disease with CNS involvement.²⁵ Eight to 24 months after completion of therapy, consenting participants completed a battery of neuropsychological tasks and caregivers completed questionnaires characterizing their child’s socioemotional functioning. The original battery of performance-based tests in AALL06N1 was amended in 2011 due to slow accrual and inconsistent completion of measures. This amendment abbreviated the evaluation requirements to be consistent with the COG Standardized Neuropsychological Battery. However, outcomes in this report include attention and EF outcomes that were part of the original battery.

Measures

The Adaptive Behavior Assessment System, Second Edition (ABAS-II)³⁵, a parent-report questionnaire used to assess children’s adaptive functioning.. Results using the composite scales for this measure have been previously reported³³; here, we included results from the Self-Direction subscale, which characterizes the extent to which children can initiate and monitor necessary activities in daily life (both aspects of EF). Scores are reported as Z-scores with higher scores indicating worse functioning.

The Behavior Rating Inventory of Executive Function (BRIEF)³⁶ is a parent-completed questionnaire assessing children’s EF in daily life. It is comprised of eight clinical scales, two broader indexes and an overall score. The Behavior Regulation Index describes difficulties with both behavior and emotion regulation, including problems shifting from one activity to another, inhibiting behavioral impulses, and modulating emotional responses. The Metacognition Index includes several aspects of cognitive EF, such as organizing activities and belongings, using time efficiently, initiating and completing tasks, and interpersonal awareness. The current study utilized the Behavioral Regulation, Metacognition, and Working Memory indices, which are reported as Z-scores with higher scores reflecting poorer EF.

The Conner’s Continuous Performance Test (CPT-II)³⁷ is a 14-minute computerized assessment of inattention, impulsivity, sustained attention, and vigilance for individuals aged ≥6 years. Individuals are asked to respond by pressing the keyboard spacebar when they see target stimuli on the screen, and to refrain from pressing the spacebar when they see a non-target. The current study utilized omission, commission, hit reaction time, hit reaction time standard error, and variability scores, all reported as Z-scores, with higher scores reflecting more attention difficulties.

The Delis-Kaplan Executive Function (DKEFS)³⁸ is a patient-performance measure of EF. The Verbal Fluency Test, used in the current study for participants aged ≥8. measures aspects of verbal organization and cognitive efficiency. Children are asked to quickly name as many items as they can belonging to a certain category or starting with a specific letter. Scores are reported as Z-scores, with higher scores indicating worse performance.

The Tower of London (TOL)³⁹ is a patient-performance measure assessing multiple aspects of executive function. Individuals aged 7+ are asked to copy patterns of beads placed on a board with pegs of differing heights, using the fewest moves possible and without breaking a specific set of rules. Successful completion of items involves multiple aspects of EF, including strategic planning, working memory, and impulse control. The Total Correct score used in the current study is presented as a Z-score, with higher scores reflecting worse performance.

The Developmental Neuropsychological Assessment (NEPSY)⁴⁰is a battery of tests tapping various domains of cognitive functioning. For this study, the NEPSY Tower and Verbal Fluency tasks were used for children who were too young to complete the DKEFS (i.e., ages 5–7) and/or TOL (i.e., ages 5–6) tasks. For analysis, scaled scores (the original metric used in the NEPSY) were converted to Z-scores, with higher scores indicating worse performance.

Statistical Analysis

Descriptive statistics were calculated for site-reported demographic and clinical characteristics, including MTX delivery mode, corticosteroid type, and time off therapy. Primary analysis was performed with multiple linear regression that included biological sex; age at diagnosis (age ≥ 10 years or age < 10 years); race; ethnicity; insurance status; and time since completion of treatment as covariates. Age was dichotomized at diagnosis to be consistent with the age cutoff for classification as HR per NIH guidelines, and the non-random assignment of children ≥10 to prednisone during induction. Primary independent variables were MTX delivery method and type of corticosteroid. The primary outcomes were attention and EF indices from the above-described performance-based and parent-report measures. Impairment was defined as scores falling 1.5 SD or more above (i.e., BRIEF and CPT) or below (i.e., TOL/NEPSY Tower/DKEFS/ABAS-II) measure norms. Note that we reverse-coded TOL, NEPSY, and DKEFS scores so that scores for all outcomes could be interpreted consistently (i.e., higher scores indicating worse performance). One sample Z-tests with one-sided p-values tested whether observed impairment rates were higher than those expected in a normative population. All tests used p<0.05 to determine significance.

Results

Sample Demographics

The sample included 181 children and adolescents between the ages of 1 and 18 years at diagnosis. Participants were primarily female, White, not Hispanic/Latino, and had US Private or Military insurance (Table 1). Approximately half of the participants were below the age of 10 years at diagnosis (48.6%). Thirteen participants (7.2%) were administered cognitive assessment measures (n=6) and/or parent-report measures (n=13) in Spanish. MTX dosing regimens and corticosteroid types were also relatively evenly represented in the sample (Table 1).

Table 1.

Demographic and Clinical Characteristics

	N	%
Biological Sex
Male	80	44.2
Female	101	55.8
Race
White	147	81.2
Asian	6	03.3
Black	5	02.8
Unknown	23	12.7
Ethnicity
Hispanic or Latino	33	18.2
Not Hispanic or Latino	137	75.7
Unknown/Not Reported	11	06.1
Insurance Status
US Public	47	26.0
US Private or Military	102	56.4
Non-US	26	14.4
Unknown (Including Self-Pay)	6	03.3
Age at Diagnosis
<10 years	88	48.6
≥ 10 years	93	51.4
MTX Delivery
High-dose	86	47.5
Escalating dose	95	52.5
Corticosteroid
Dexamethasone	82	45.3
Prednisone	99	54.7

Participants were between the ages of 4 and 21 years (M age=12.5 years, SD=5.0 years) at the time of neurocognitive evaluation, and were between 8 and 24 months off therapy (M time off therapy=14.3 months, SD=3.8 months). Compared with eligible patients in AALL0232 who did not enroll in the neurocognitive study, participants in AALL06N1 were somewhat younger at diagnosis (mean age difference=1.2 years; p<.01) and slightly less likely to identify as Hispanic/Latino (19% v 22%; p=.02)²⁵.

Rates of Impairment in Attention and EF

Mean scores for performance measures and the BRIEF were uniformly within the average range for the sample. Rates of impairment on performance-based measures of attention and EF were within normal limits compared to rates in the standardization samples for most measures (p>.05). The only exception was response variability on the CPT-II, for which the impairment rate significantly exceeded expectations (impaired n=18, 15.9% vs. 10% expected, p=.02). Impairment rates on the BRIEF were more variable. Specifically, rates of impairment on the Behavioral Regulation Index were not significantly different between survivors of ALL and the standardization sample, but rates for the Working Memory and Metacognition Indices were significantly higher for survivors (n=38, 22.5%, p≤.001 and n=28, 16.9%, p<.001, respectively). In addition, impairment rates on the ABAS-II Self-Direction scale also were significantly higher than expectations (n=29, 17.4%, p≤.001).

EF as a Function of MTX Dosing and Corticosteroid Treatment

After controlling for age at diagnosis, sex, race, ethnicity, time off treatment, and type of insurance, there were no significant differences in attention or EF scores for children who received HD-MTX versus escalating-dose MTX (Table 2). Additionally, there were no significant differences in attention or EF outcomes as a function of treatment with prednisone or dexamethasone during induction (Table 2).

Table 2.

Methotrexate Delivery and Corticosteroid Treatment

		Methotrexate Delivery			Steroid
	N	HD-MTX	Escalating-Dose MTX	p-Value	Dexamethasone	Prednisone	p-Value
BRIEF Behavior Reg	154	−0.04(1.07)	−0.03(1.28)	0.95	−0.01(1.22)	−0.06(1.16)	0.98
BRIEF Working Mem	169	0.36(1.37)	0.29(1.33)	0.80	0.28(1.26)	0.36(1.42)	0.65
BRIEF Metacognition	166	0.17(1.25)	0.17(1.19)	>0.99	0.14(1.14)	0.20(1.27)	0.68
CPT-II Commissions	113	0.19(0.96)	0.24(0.88)	0.91	0.26(0.96)	0.16(0.86)	0.55
CPT-II Omissions	113	0.26(1.37)	0.26(1.31)	0.81	0.13(1.27)	0.40(1.40)	0.31
CPT-II Variability	113	0.14(1.31)	0.25(1.07)	0.28	0.19(1.20)	0.21(1.19)	0.96
CPT-II Hit Reaction Time	113	−0.20(1.05)	−0.23(1.14)	0.71	−0.32(0.95)	−0.09(1.22)	0.24
CPT-II Hit Reaction SE	99	0.04(1.36)	0.07(1.10)	0.60	−0.06(1.20)	0.17(1.26)	0.62
Tower Total Correct	107	−0.29(1.12)	0.10(1.18)	0.06	−0.08(1.16)	−0.10(1.18)	0.93
Category Fluency	123	−0.28(1.12)	−0.44(1.12)	0.25	−0.36(1.14)	−0.37(1.10)	0.96
ABAS Self Direction	167	0.03(1.22)	0.22(1.30)	0.16	0.08(1.22)	0.17(1.30)	0.27

Note: All measures presented as standardized Z-scores, with higher scores indicating worse performance. p-values refer to adjusted effects from models including all covariates (sex, race, age, ethnicity, insurance status, time off treatment). Estimates are unadjusted means. Unless otherwise noted, all comparisons were nonsignificant.

EF as a Function of Patient Demographics

When controlling for other covariates, parents of children covered by US public health insurance reported that their children experienced significantly more difficulties with working memory, behavioral regulation, metacognition, and self-direction than children covered by US private or military health insurance (all p values < .05, Table 3). There were no significant differences in parent-reported working memory, behavioral regulation, or metacognition scores based on age at diagnosis (<10 vs ≥10 years; Table 3). Finally, parents rated male survivors as having worse self-direction than females (p=.01).

Table 3.

Covariates Predicting Parent Report of Working Memory, Behavioral Regulation, and Metacognition.

Note: p-values refer to adjusted effects from models including all covariates; Estimates are unadjusted mean standardized Z-scores, with higher scores indicating worse performance. Unless otherwise noted, all comparisons were nonsignificant.

^*p < .05

Patients less than 10 years at diagnosis experienced significantly more difficulties with attention (i.e., omissions, reaction time, reaction time consistency, and reaction time variability on the CPT-II), than children 10 years and older when controlling for other covariates (p values <.05, Table 4, Figure 1). Younger children also demonstrated significantly more difficulties with EF as measured by the category fluency scores (p values <.05, Table 5, Figure 1). In contrast, there were no significant differences in performance-based measures of attention or EF based on insurance status or type of corticosteroid received (Tables 2, 4, and 5). Age by insurance status interactions and sex by steroid interactions were examined, but were not significant in any models.

Table 4.

Covariates Predicting Performance on the CPT-II

Note: p values refer to adjusted effects from models including all covariates. Estimates are unadjusted mean standardized Z-scores, with higher scores indicating worse performance. Unless otherwise noted, all comparisons were nonsignificant.

^*p <.05

Figure 1.

*p <.05

Note: Scaled Z-scores with higher scores indicating worse outcomes presented for all measures. p-values based on multivariable models controlling for all covariates (sex, race, ethnicity, insurance status, time off treatment).

Table 5.

Covariates Predicting Performance on EF Tasks

Note: p values refer to adjusted effects from models including all covariates. Estimates are unadjusted mean standardized Z-scores, with higher scores indicating worse performance. Unless otherwise noted, all comparisons were nonsignificant.

^*p < .05

Discussion

We examined attention and EF in survivors of pediatric high-risk B-ALL randomized to two different forms of MTX delivery and two different steroids during induction. Results indicated no significant differences in attention or EF based on treatment assignment. However, consistent with our earlier reports on estimated intellectual functioning and academic achievement in this cohort,^25,33 attention and EF varied as a function of demographic predictors including age at diagnosis, race, and insurance status. Specifically, performance-based outcomes measuring both attention and EF were worse for children under age 10 at time of diagnosis compared to children 10 and older. In addition, Black or Asian participants exhibited superior verbal fluency scores compared to participants whose race was unknown. Finally, parent-reported EF was significantly worse for children covered by US public insurance, which has historically been utilized as an indicator of economic hardship.

Our findings add to the literature describing outcomes associated with use of different ALL treatment strategies. Considering these results in conjunction with recent studies of steroid-associated outcomes, it is unlikely that dexamethasone contributes significantly to risk of attention and EF problems following treatment. Moreover, treatment with HD-MTX does not confer additional risk for these specific outcomes compared to escalating-dose MTX, although the impact of cumulative MTX dosing remains unclear.

Survivors in our sample were, on average, more likely to show both performance-based and parent-reported difficulties with attention and EF than their same-age peers. This is consistent with recent longitudinal and meta-analytic studies indicating that, while a majority of survivors exhibit intact functioning, a substantial minority demonstrate difficulties with attention and EF.^11,41 Understanding the implications of these task-based difficulties identified in our sample is complicated, however. Prior work has identified difficulties with response variability on the CPT as characteristic of both ALL survivors¹¹ and individuals with Attention Deficit Hyperactivity Disorder.⁴² In the latter, response variability has been associated with reduced functional connectivity⁴³ which is also documented in survivors of ALL²⁶. Response variability reflects difficulty with vigilance to a task – small lapses of attention that prevent consistent responding,⁴⁴ although the precise implications for everyday life are poorly understood.

Although there is evidence of consistent, though modest, difficulties with efficient planning and problem-solving in our sample, EF is context-dependent and increasingly thought to be moderated by situational elements including emotional salience and motivation.⁴⁵ Neither factor was measured in our study or in other studies of cancer survivors, making it difficult to determine the extent to which survivors may struggle with so-called “hot” EF, defined as problem-solving under conditions of heightened emotional salience. ^46,47 Interestingly, parents tended to rate survivors as having greater difficulties with working memory and metacognition than with behavior regulation compared to healthy peers in the standardization sample. Working memory and metacognition are important skills for complex and applied aspects of schoolwork like reading comprehension and math reasoning.^48,49 Additionally, these skills have been linked to different brain networks than inhibitory control systems;⁴⁵ thus our findings suggest that development of brain systems supporting specific aspects of EF may be differently affected. As noted above, although there is robust evidence of alterations in both brain structure and function following treatment for pediatric ALL, there is no consistent pattern of neurodevelopmental alteration of intrinsic networks.²⁷

Participants who were younger at diagnosis had worse scores on most performance-based measures, including those indicating difficulties with sustained attention and impulsivity, verbal fluency, and efficient planning and problem-solving. This pattern of findings is consistent with published literature suggesting that when children are treated for ALL prior to maturation of structural and functional brain networks that support higher-order thinking, they are more vulnerable to domain-specific difficulties in this area.^4,27 Of importance, however, we lack information about how disruption of typical educational activities during treatment may further impact development of these skills in the youngest children. Early education settings are thought to be critical for learning both practical and social problem-solving.⁵⁰ If younger children are attending school less consistently, or not at all, during phases of their treatment, it is possible they are missing out on opportunities for development of these skills.

Beyond academics, EF skills are implicated in the development of health behaviors,⁵¹ which is salient for ALL survivors who are known to be at risk for overweight and reduced physical activity.^52,53 In adults, there is evidence that EF skills have a reciprocal relationship with health behaviors, such that poor engagement in healthy activities may inhibit EF skill development, which may make it harder to implement and sustain healthy habits like diet and exercise. There is also recent evidence that physical fitness among survivors of pediatric ALL is related to neurocognitive outcomes, including EF skills.⁵² This implies that interventions aimed at improving health behavior and/or executive functions may have additive benefits across both domains for survivors.

Participants receiving public insurance in our sample had higher rates of parent-reported difficulties with EF – but not significantly lower task-based performance - than those with private insurance. This raises the question of whether social determinants of health may interact with treatment-related risk factors to impact the extent to which survivors experience difficulties in everyday life. Children with fewer economic and environmental resources, as well as greater exposure to environmental stressors, may experience greater difficulties with EF.^54,55 Thus, it may be that lower-resourced survivors of pediatric ALL have the same skill level but fewer opportunities to apply and practice their skills than peers with greater resources. Although we did not find an interaction between age at diagnosis and insurance status in our sample, it is possible that younger children experiencing ALL in the context of fewer economic and environmental opportunities may be at higher risk for EF difficulties. This possibility could be better explored in future research by using more robust indicators of socioeconomic status (e.g., parental education, family income), or a single indicator of community and family access to economic, environmental, and health resources, such as the Child Opportunity Index.⁵⁶ It will also be important to investigate how availability and quality of educational accommodations and school-family communication may mitigate adverse outcomes.

Limitations in this study include limited representation of minoritized individuals, variability in timing of the assessment (8–24 months post-diagnosis); use of a cross-sectional assessment strategy; and use of insurance status as a proxy for SES. In addition, patients enrolled on the optional, companion study, AALL06N1, represented less than 20% of eligible patients treated on AALL0232. As reported previously,²⁵ participants evaluated in AALL06N1 were younger and less ethnically diverse than the larger sample from AALL0232. Although these differences were significant, the effect sizes were rather small, making it difficult to see an obvious source of bias between the groups. Finally, the original battery for AALL06N1, including tasks analyzed here, was extensive and specialized, requiring availability of a neuropsychologist. This likely impacted both participation (at the site and individual patient levels) and data compliance, both of which should be considered additional limitations of our study. Prospective, longitudinal studies are needed to further clarify the contribution of SES and other demographics related to cognitive outcomes for survivors of pediatric cancer.

Table of Abbreviations:

ABAS-II

Adaptive Behavior Assessment System, Second Edition

ALL

Acute Lymphoblastic Leukemia

B-ALL

B-cell Acute Lymphoblastic Leukemia

BRIEF

Behavior Rating Inventory of Executive Functioning

CNS

Central nervous system

COG

Children’s Oncology Group

CPT

Continuous Performance Test

DKEFS

Delis-Kaplan Executive functioning System

EF

Executive functioning

HD-MTX

High-Dose Methotrexate

MTX

Methotrexate

NEPSY

Developmental Neuropsychological Assessment

SES

Socioeconomic status

TOL

Tower of London

US

United States

Data Sharing and Availability:

The Children’s Oncology Group makes data available in accordance with policies stipulated by the National Institutes of Health. Detailed information on COG data sharing is available at: https://childrensoncologygroup.org/data-sharing