INTRODUCTION
Improvements in treatments have led to increased survivorship rates following pediatric brain tumors, creating a need for investigation into long-term outcomes in this population (Ostrom, 2017). Research thus far has indicated that among other challenges, pediatric brain tumor survivors experience neurological sequelae that disrupt everyday adaptive functioning (AF) skills. Practical skills (e.g., self-care and home-living skills) are found to be significantly impaired for pediatric brain tumor patients, both during acute treatment (Robinson et al., Reference Cabrera, Edelstein, Mason and Tartaglia2015) and approximately 2 years post-treatment (Hoskinson et al., Reference Hoskinson, Wolfe, Yeates, Mahone, Cecil and Ris2018). Long-term survivors of pediatric brain tumors have persisting motor and physical skills deficits, which are associated with limited social roles compared with healthy controls (Ness et al., Reference Ness, Morris, Nolan, Howell, Gilchrist, Stovall and Neglia2010). During serial examinations across a 5-year follow-up, pediatric brain tumor survivors’ scores declined significantly on the Vineland Adaptive Behavior Scales (VABS) Communication Index (Netson, Conklin, Wu, Xiong, & Merchant, Reference Netson, Conklin, Wu, Xiong and Merchant2013), indicating that long-term difficulties emerge as the demands on survivors in their communication skills, self-care, and daily life increase with age. In a heterogeneous sample of long-term survivors of pediatric brain tumors, Mean (M) = 15.8 years postdiagnosis, AF impairments were found in 29% of the sample [defined by the Scales of Independent Behavior-Revised (SIB-R) z-score ≤ −1.5] (King & Na, Reference Carvalho, Buelow, Ready and Grace2016). As survivors enter adulthood, difficulties with daily living activities and responsibilities can persist, yet differences in tumor treatment and medical complications contribute to a wide range in outcomes. Identifying the impact of multiple specific risk factors is essential to assess a heterogeneous group with variety in treatment types, medical complications, and tumor locations.
The Neurological Predictor Scale (NPS; Micklewright, King, Morris, & Krawiecki, Reference Micklewright, King, Morris and Krawiecki2008) score is a quantification of the cumulative impact of various tumor- and treatment-related neurological risk factors (i.e., radiation, chemotherapy, neurosurgery, endocrine dysfunction, hydrocephalus, and seizure medication). The NPS allows clinical researchers to capture the cumulative neurological contributors to heterogenous outcomes with an easy to calculate scale. The NPS is associated with cognitive outcomes over and above individual risk factors, suggesting that the cumulative examination of risk factors can better characterize the heterogeneity in survivors’ long-term outcomes (Kautiainen, Na, & King, Reference Kautiainen, Na and King2019; King & Na, Reference King and Na2016; Micklewright et al., Reference Micklewright, King, Morris and Krawiecki2008; Papazoglou, King, Morris, & Krawiecki, Reference Papazoglou, King, Morris and Krawiecki2008; Taiwo, Na, & King, Reference Taiwo, Na and King2017). Prior research has demonstrated that the NPS is related to core cognitive abilities, executive function, and functional outcomes in pediatric brain tumor survivors (King & Na, Reference King and Na2016; McCurdy, Rane, Daly, & Jacobson, Reference McCurdy, Rane, Daly and Jacobson2016; Na et al., Reference Na, Li, Crosson, Dotson, MacDonald, Mao and King2018).
While the NPS has been shown to predict cognitive outcomes, findings on its relation to informant-reported AF are mixed. The NPS has been shown to be predictive of AF outcomes with medium-to-large effect sizes in a heterogeneous sample of adult survivors of childhood brain tumors (King & Na, Reference Carvalho, Buelow, Ready and Grace2016). However, McCurdy et al. (Reference McCurdy, Rane, Daly and Jacobson2016) did not find that the NPS was significantly associated with caregiver-reported adaptive behavior skills in child-age survivors. The adult sample on average over 15 years postdiagnosis and the child sample on average 3 years from diagnosis are at very different points in their developmental trajectory (King et al., Reference Carvalho, Buelow, Ready and Grace2016; McCurdy et al., Reference McCurdy, Rane, Daly and Jacobson2016). Tumor and treatment-related variables have been shown to cause neurocognitive late effects that impact survivors’ functioning several years later, as survivors reach young adulthood (Ellenberg et al., Reference Ellenberg, Liu, Gioia, Yasui, Packer, Mertens and Robison2009; Hobbie et al., Reference Hobbie, Ogle, Reilly, Barakat, Lucas, Ginsberg and Deatrick2016). Young adult survivors are a crucial group to follow to understand the impact of treatment variables on brain microstructure (e.g., white matter tracts) and developing cognitive abilities (Mulhern, Merchant, Gajjar, Reddick, & Kun, Reference Mulhern, Merchant, Gajjar, Reddick and Kun2004). Therefore, mechanisms that drive the relationship between the cumulative effect of treatment variables and long-term survivors’ AF should be explored further.
Frontal system impairments such as apathy and executive dysfunction (ED) are often observed following brain tumors in both adults and children (Carroll et al., Reference Carroll, Watson, Spoudeas, Hawkins, Walker, Clare and Ring2013; Fox & King, Reference Fox and King2016; Gregg et al., Reference Gregg, Arber, Ashkan, Brazil, Bhangoo, Beaney and Yágüez2014). Apathy is a lack of motivation and goal-directed activity, and while these symptoms may occur within a depressive episode, apathy is a unique disorder (Marin, Reference Marin1996). In a cross-sectional study of 117 adult survivors of childhood posterior fossa brain tumors, 35% of survivors displayed clinically significant apathy compared with only 18% of sibling controls (Carroll et al., Reference Carroll, Watson, Spoudeas, Hawkins, Walker, Clare and Ring2013). ED includes problems regulating interference control, strategic planning, effortful and flexible organization, goal-directed preparedness to act, and social discourse (Denckla, Reference Denckla and Lyon1994). Apathy may limit goal-directed behavior but is a conceptually distinct disorder from ED, as an individual’s motivation and drive toward a goal is separate from their preparedness to act on a goal (Le Heron, Holroyd, Salamone, & Husain, Reference Le Heron, Holroyd, Salamone and Husain2019). Given that apathy and ED can be localized in the frontal areas of the brain, researchers are interested in the distinct behavioral profiles associated with frontal system dysfunction (Masterman & Cummings, Reference Masterman and Cummings1997). A singular frontal lobe syndrome is not supported by anatomical or neuropsychological research, and we may provide a better understanding of long-term survivors’ adaptive deficits by examining the unique profiles of apathy and ED (Heilman & Valenstein, Reference Hartman, Houwen, Scherder and Visscher2010; Robinson, Calamia, Gläscher, Bruss, & Tranel, Reference Robinson, Calamia, Gläscher, Bruss and Tranel2014). Another frontal system clinical syndrome is disinhibition, or difficulties with inhibitory control and managing impulsivity (Grace & Malloy, Reference Grace and Malloy2001). These three behavioral syndromes have been linked to three frontostriatalthalamic circuits: the dorsolateral prefrontal circuit with executive function, the lateral orbital prefrontal circuit with disinhibition, and the anterior cingulate circuit with apathy (Cummings, Reference Cummings1993; Malloy & Grace; Reference Malloy and Grace2005). The Frontal Systems Behavior Scale (FrSBe) is an informant behavior scale utilized to capture the three frontal syndromes of disinhibition, apathy, and ED (Grace & Malloy, Reference Grace and Malloy2001).
Gregg and colleagues utilized the FrSBe to study apathy, mood disorders, and ED in adult patients with frontal tumors, adult patients with nonfrontal tumors, and healthy siblings (Gregg et al., Reference Gregg, Arber, Ashkan, Brazil, Bhangoo, Beaney and Yágüez2014). The researchers found that 40% of frontal and nonfrontal tumor survivors at least 1 year out from diagnosis presented with clinical levels of apathy and ED. This finding indicates that despite the critical role of the frontal cortex for social and emotional functioning, a tumor in other locations may also disrupt frontal subcortical systems that impact mood, emotional, and behavioral regulation.
Informant report data from the FrSBe correlates with activities of daily living (ADL) across older adult neurological populations (Carvalho, Buelow, Ready, & Grace, Reference Carvalho, Buelow, Ready and Grace2016). Specifically, the FrSBe Apathy subscale is a significant predictor of basic ADLs after controlling for patients’ age, gender, education, and cognitive ability. The decreased motivation and goal-seeking behavior observed in apathy likely contributes to difficulties with completing basic everyday tasks for older adults with a range of neurological disorders. For adult brain tumor survivors, the relationship between frontal systems dysfunction and adaptive behavior is likely influenced by structural changes, such as a reduction in white matter integrity due to treatment and tumor complications (Brinkman et al., Reference Brinkman, Reddick, Luxton, Glass, Sabin, Srivastava and Krull2012; King et al., Reference King, Wang and Mao2015). Therefore, it is essential to consider the neurological risk of the individual survivor when working to understand the relationship between frontal systems impairment and AF.
While many cognitive abilities develop throughout earlier childhood, executive functions continue to develop into young adulthood (Luna et al., Reference Luna, Thulborn, Munoz, Merriam, Garver, Minshew and Sweeney2001). Deficits in executive function may not be as noticeable in younger children due to the structure of the classroom, classroom size, behavioral expectations, and supportive interventions (Eccles & Roeser, Reference Eccles, Roeser, R.M. and L.2009). Long-term follow-up has the advantage of assessing a survivor once they are beyond the structure of immediate interventions and school-based learning. Young adulthood requires management of responsibilities and new challenging tasks, which often require executive skills such as sequencing, inhibition, initiation, and planning. Executive function predicts long-term social engagement and communication outcomes (e.g., participation in leisure activities and social functioning) for young adult survivors of pediatric traumatic brain injury (Yeates et al., Reference Yeates, Swift, Taylor, Wade, Drotar, Stancin and Minich2004). Additionally, disruption of executive skills will impact a survivors’ motivation to learn, initiate, and sustain motor tasks. Poor inhibition leading to shorter planning time has been shown to be the underlying mechanism for the relationship between ED and motor deficits in children with intellectual disabilities (Hartman, Houwen, Scherder, & Visscher, Reference Hartman, Houwen, Scherder and Visscher2010). Frontal systems dysfunction may explain part of the impact of treatment on long-term AF due to increased demands on and expectations of survivors’ level of independence as they age. The present study aims to investigate the utility of the NPS in predicting AF outcomes via frontal systems dysfunction with the interest of developing considerations for intervention. Given known effects of frontal-subcortical system disruptions on AF, this study assessed the NPS’ relationship with AF as mediated by frontal systems dysfunction, measured by the FrSBe, in long-term pediatric brain tumor survivors.
METHODS
Participants and Procedures
Participants were 75 survivors of childhood brain tumors (41 females; M age = 23.5 years, SD age = 4.5 years). Survivors were at least 16 years old and were a minimum of 5 years past the time of diagnosis (M = 15.9 years, SD = 6.3 years). Demographics, tumor treatment, tumor type, and tumor location are listed in Table 1. The local Institutional Review Boards approved this study, and all data were obtained in compliance with the Helsinki Declaration. Survivors were recruited from three sources: (1) an advertisement in an annual newsletter from the Brain Tumor Foundation of Georgia in which survivors were encouraged to call and inquire about the study; (2) a previous longitudinal childhood brain tumor study; and (3) a large southeastern hospital system database. Participants from the initial sample were excluded from the study if they had a pervasive developmental disorder or neurofibromatosis (N = 6). Additionally, participants were excluded if English was not their first language (N = 1). Survivors’ medical information was obtained from a retrospective medical records review. Two graduate students independently gathered information on tumor type, tumor location, treatment protocol, radiation dosage, chemotherapy, surgery, surgery complications, hydrocephalus, seizure medications, and type of endocrine dysfunction. The agreement on the data extraction was 98% between the raters. The minor discrepancy was resolved before the graduate students single and double entered the participants’ medical data into a database.
Table 1. Descriptives of survivor sample (N = 75)
WASI = Wechsler Abbreviated Scale of Intelligence; SD = standard deviation.
Measures
Wechsler Abbreviated Scale of Intelligence
The Wechsler Abbreviated Scale of Intelligence (WASI) (Wechsler, Reference Wechsler1999) is comprised of four subtests to measure intelligence. The WASI was administered to participants at the time of evaluation to estimate a full-scale IQ. The mean, standard deviation, and range of IQ scores for all participants are included in Table 1.
Neurological Predictor Scale
The NPS is a measure that quantifies tumor-related treatments and neurological complications with one cumulative score (Micklewright et al., Reference Micklewright, King, Morris and Krawiecki2008) The NPS scale ranges from 0, the lowest level of neurological risk, to 11, the highest level of neurological risk. The measure items are comprised of questions on treatment factors (i.e., radiation, chemotherapy, and neurosurgery) and the presence or absence of other neurological factors (i.e., endocrine dysfunction, hydrocephalus, and seizure medications). After consent was provided, information for the NPS was gathered from retrospective medical review. A procedure for coding was developed before the study. Two graduate students trained in the procedure and scoring independently summed their scores.
Frontal Systems Behavior Scale
The FrSBe is a 46-item behavior rating scale developed to access apathy, disinhibition, and ED in adult neurological populations (Grace & Malloy, Reference Grace and Malloy2001). All three subscales were found to be reliable (α = .84−.91) during scale development (Malloy & Grace, Reference Malloy and Grace2005). Additionally, construct validity was demonstrated by findings where patients with frontal damage obtained significantly higher FrSBe scores than patients with nonfrontal damage and healthy controls (Grace, Stout, & Malloy, Reference Grace, Stout and Malloy1999). The distribution of FrSBe scores has a mean of 50 and standard deviation of 10. Therefore, Apathy, ED, and Disinhibition subscales are identified as reaching clinical levels at a score of T ≥ 65, z ≥ 1.5. The present study utilized the informant version of the FrSBe. The informant was a person who had lived with the survivor for at least a 1-year period in the past. Of the 75 informants, 63 were parents, 5 were family members who were not a parent (i.e., grandparent or sibling), 5 were significant others, and 2 were unrelated friends/roommates. The z-scores were calculated from age-based norms. Given that we have three minors included in our sample, the one 16-year-old participant and the two 17-year-old participants were scored using the 18-year-old norms. The mean scores for the survivors at the age of 18 years and above and the three survivors under 18 were not significantly different for FrSBe Apathy (p = .86), FrSBe Disinhibition (p = .57), and FrSBe ED (p = .56).
Scales of Independent Behavior- Revised
The SIB-R is a norm-referenced assessment of adaptive and maladaptive behavior in school, home, employment, and community settings (Bruininks, Reference Marin1996). The internal consistency of the SIB-R subscales ranges from .7 to .9, and the test–retest reliabilities for all subtests were above .8 (Strauss et al., Reference Strauss, Sherman and Spreen2006). The original Scales of Independent Behavior has demonstrated convergent and divergent validity, based on correlations (r = .83) with another measure of AF, the VABS (Middleton, Keene, & Brown, Reference Middleton, Keene and Brown1990). Administration of this measure yields estimates of motor, social interaction/communication, personal living, community living, and broad independent living skills. Lower scores reflect poorer AF. The SIB-R is a structured interview that was completed by the same informant who completed the FrSBe. All raw scores were converted into age-normed z-scores.
Data Analysis
All data analyses were run on IBM SPSS 25.0 (IBM, 2017). Bivariate Pearson correlations for the continuous outcome variables (i.e., NPS, FrSBe, and SIB-R scores) were conducted. We used Spearman’s rho correlation to compare the categorical variables of sex, socioeconomic status (SES), as measured by the Hollingshead scale, years of education, and age of the child at diagnosis (Hollingshead, Reference Hollingshead1970). Sex and SES were not significantly related to the outcome variables. However, years of education, the age of the child at diagnosis, and the time between treatment and examination were related to the SIB-R z-scores. Years of education, while informative in describing our sample, may be conceptualized as another outcome related to our predictor and mediator variables and would not be a meaningful covariate in its own right. Age at diagnosis is related to long-term neurocognitive functioning; however, there is not an established relationship between age at diagnosis and AF (Ashford et al., Reference Ashford, Netson, Clark, Merchant, Santana, Wu and Conklin2014; Mulhern et al., Reference Mulhern, Merchant, Gajjar, Reddick and Kun2004). AF deficits have been related to longer time since diagnosis (Ris & Noll, Reference Ris and Noll1994; Stargatt et al., Reference Stargatt, Rosenfeld, Anderson, Hassall, Maixner and Ashley2006). Therefore, age at diagnosis and time between diagnosis and assessment were utilized as covariates to understand further the potentially predictive value of these variables. Parallel mediation analyses were conducted using the PROCESS macro for SPSS (Hayes, Reference Hayes2014). PROCESS calculates indirect effects by constructing confidence intervals based on resampling of the data with a replacement bootstrapping method. Data were resampled 10,000 times to approximate the sampling distribution, which resolves the issue of a skewed distribution in smaller samples. PROCESS allows for the inclusion of covariates within the mediation model and age at diagnosis and time between diagnosis and assessment were included as a predictor variable along with the NPS. We utilized Model 4 within PROCESS for parallel mediations including covariates.
RESULTS
Tests of normality, linearity, heteroscedasticity, and independence were not violated, and the sample was normally distributed. Probability–probability (P–P) plots were utilized to visualize the distribution of scores for the predictors and outliers of interest. Outlier testing was conducted to determine the influence that NPS, FrSBe subscales, and age at diagnosis have on dependent variables. One survivor violated outlier testing on Social Communication (SC) and Personal Living (PL) Scales, based on a studentized residual lower than −2.5. The individual had acceptable scores on the other scales of the SIB-R, and their score was winsorized to reflect the closest acceptable score (SC z-score change from −5.87 to −3.8 and PL z-score change from −5.07 to −3.33). One different survivor violated outlier testing on Community Living (CL) Scale, based on a studentized residual lower than −2.5. Their score was winsorized to reflect the next lowest acceptable score (z-score change from −4.8 to −4.67). The NPS was correlated with FrSBe Apathy and ED domains but not Disinhibition. NPS was correlated with all SIB-R domains, and a correlational matrix is included in Table 2. Higher scores on the NPS, representing more complex treatment and sequelae, were correlated with lower z-scores on the SIB-R subscales. Higher FrSBe z-scores, representing greater frontal systems impairment, were correlated with lower SIB-R z-scores. Descriptive data for survivors’ performances on the NPS, FrSBe scales, and SIB-R scales are presented in Table 3.
Table 2. Correlational matrix of NPS scores, FrSBe scales z-scores, and SIB-R scales z-scores
False discovery rate critical value (q) cut-off = .05.
a Significant after controlling for false discovery rate.
Table 3. Descriptive data for NPS, FrSBe scales, and SIB-R scales
NPS scores range from 0 to 11. The z-scores are reported for FrSBe and SIB-R scales.
Parallel mediation models were conducted with the NPS predicting SIB-R scales via FrSBe Apathy and ED, with age at diagnosis and time between diagnosis and assessment as covariates. The FrSBe Disinhibition Scale was not utilized in models because it was not correlated with the NPS. The NPS was a significant predictor for each subscale within its model [Motor Skills (MS), p = .0001; SC, p = .002; PL, p = .004; CL, p = .007]. Including apathy in the overall model did not change the significance of direct or indirect pathways, but it also did not add additional variance to the model. The regression coefficients, the indirect effect, and bootstrapped confidence intervals for each mediation model are presented by SIB-R domain. Parallel mediation models, including direct effects and covariates, are represented by Figure 1a–d.
Fig. 1. (a) Parallel mediation models for motor skills. (b) Mediation model for personal living skills. (c) Mediation model for community living skills. (d) Mediation model for social communication skills.
Motor Skills
The total indirect effect of ED and Apathy on MS was significant (b = −.11; SE = .06; CI: −.24, −.01). The indirect effect of Apathy on MS was not significant (b = −.03; SE = .41; CI: −.13, .03). Additionally, the indirect effect of ED was not significant (b = −.07; SE = .53; CI: −.2, .05). Finally, age at diagnosis (p = .81) and time between diagnosis and examination (p = .57) were not significant predictors in the model.
Personal Living
The total indirect effect of ED and Apathy on PL was significant (b = −.11; SE = .05; CI: −.22, −.02). For both Apathy (CI [−.15, .006]) and ED (CI [−.16, .007]), the bootstrap confidence intervals crossed over zero, and there were no significant indirect effects. Additionally, age at diagnosis (p = .95) and time between diagnosis and examination (p = .14) were not significant predictors in the model.
Community Living
The total indirect effect of ED and Apathy on CL was significant (b = −.11; SE = .05; CI: −.22, −.01). The indirect effect of ED on CL was significant (b = −.08; SE = .05; CI: −.19, −.01), as determined by the bootstrap confidence interval not crossing over zero. The indirect effect of Apathy on CL was not significant (b = −.03; SE = .03; CI: −.1, .03). Additionally, the covariates, age at diagnosis (p = .11) and time between diagnosis and assessment (p = .27) were not significant predictors in the model.
Social Communication
The total indirect effect of ED and Apathy on SC was significant (b = −.09; SE = .05; CI: −.2, −.02). The indirect effect of ED on SC was significant (b = −.07; SE = .04; CI: −.15, −.09). The indirect effect of Apathy on SC was not significant (b = −.02; SE = .02; CI: −.09, .02). Age at diagnosis (p = .34) and time between diagnosis and assessment (p = .16) were not significant predictors in the model.
DISCUSSION
The current study examined whether the relationship between cumulative neurological risk and AF is mediated by frontal system dysfunction in a heterogeneous group of long-term pediatric brain tumor survivors. A higher NPS score, a quantification of more complex brain tumor treatment and neurological sequelae, was related to poorer long-term AF, apathy, and ED but not to disinhibition. This finding reflects the discrepancy between the covert behaviors displayed as a result of ED and apathy, such as inattention or a lack of motivation, and the overt behaviors from disinhibition, such as impulsivity and an inability to sit still. Such distinctions have been observed in tumor survivors in the literature. A study of adult (nonpediatric) tumor survivors found significant differences in informant ratings before diagnosis and at the time of neuropsychological assessment on the FrSBe Apathy and ED subscales but not for FrSBe Disinhibition (Cabrera, Edelstein, Mason, & Tartaglia, Reference Cabrera, Edelstein, Mason and Tartaglia2015), suggesting that tumors and treatments selectively impact more covert executive skills. Puhr et al. (Reference Puhr, Ruud, Anderson, Due-Tønnessen, Skarbø, Finset and Andersson2019) found that on self-report of executive function, measured by the Behavior Rating Inventory of Executive Function-Adult (BRIEF-A), pediatric brain tumor survivors had significantly more remarkable deficits when examining covert behaviors (e.g., organization, planning, and attentional control) than overt deficits (e.g., aggressive and intrusive behavior) compared with healthy controls (Puhr et al., Reference Puhr, Ruud, Anderson, Due-Tønnessen, Skarbø, Finset and Andersson2019). It is possible that for covert behaviors, post-treatment changes in survivors reflect the intensity of their cumulative tumor treatment and are observed both by survivors themselves and close friends or relatives.
The survivors who faced greater tumor and treatment sequelae also experienced more reported difficulties with AF, apathy, and ED is in line with previous work by our colleagues (King & Na, Reference King and Na2016; Taiwo et al., Reference Taiwo, Na and King2017). The NPS provides clinical utility by identifying those most at risk for cognitive and adaptive difficulties, going over and above predictions of the cognitive and psychosocial challenges from individual tumor or treatment factors (Robinson, Fraley, Pearson, Kuttesch, & Compas, Reference Robinson, Fraley, Pearson, Kuttesch and Compas2013; Turner, Rey-Casserly, Liptak, & Chordas, Reference Turner, Rey-Casserly, Liptak and Chordas2009). Furthermore, the present study found that the deleterious effects of composite treatment and neurological sequelae on adaptive function domains may occur via ED.
The effects of composite treatment and sequelae on complex elements of daily living such as SC and CL skills appear to be particularly mediated by ED; that is, individual differences in ED increased the strength of the relationship between more tumor and treatment sequelae and more difficulty with social and community-based skills. The ability to generate responses and self-monitor behavior in interpersonal interactions are required within each of these domains. Both parent reports and neuropsychological testing of executive function have been related to social skills difficulties in the areas of communication, cooperation, and self-control in brain tumor survivors (Wolfe et al., Reference Wolfe, Walsh, Reynolds, Mitchell, Reddy, Paltin and Madan-Swain2013). The present study extends these findings to suggest that these deficits in executive functions are the means by which more medically compromised survivors struggle in social situations. Additionally, it is reasonable that CL skills are also impacted, as they include an aspect of goal-directed preparedness and strategic planning. In line with this, our research group found that the relationship between group membership (survivors vs. healthy controls) and CL abilities was mediated by poor planning skills, a facet of executive function (King et al., Reference King, Smith and Ivanisevic2015). Independent daily living skills are therefore not just impacted by the severity of neurological sequelae but also dependent on the survivors’ executive function.
Despite presently observed and previously established high frequencies of apathy in pediatric brain tumor patients and survivors (Fox & King, Reference Fox and King2016; Gregg et al., Reference Gregg, Arber, Ashkan, Brazil, Bhangoo, Beaney and Yágüez2014), apathy did not add significantly to the mediation models. All of the survivors in our sample who were impaired (z-score greater than 1.5 standard deviations from the age-normed mean) on Apathy were also impaired on ED, which may partially explain why apathy did not contribute additional variance despite being significantly related to AF. Apathy has been uniquely related to pediatric brain tumor survivors’ long-term outcomes and pituitary disorder (Fox & King, Reference Fox and King2016). Therefore, apathy and ED should be conceptualized as discrete patterns of behavior composed of complementary and interacting behavior profiles. For example, the lack of initiation and motivation seen as a result of apathy is related to ED, particularly in the domain of goal-directed behavior (Levy & Dubois, Reference Levy and Dubois2006). Motivation to be part of interactions and an ability to initiate that experience are required to engage in social discourse and join community activities, the two AF deficits that were mediated by ED. Additionally, the executive skills of setting a goal, planning, and attending a community gathering may be obstructed by the presence of apathy (McPherson, Fairbanks, Tiken, Cummings, & Back-Madruga, Reference McPherson, Fairbanks, Tiken, Cummings and Back-Madruga2002). Further work should aim to disentangle the unique symptom of apathy, which may be present in other disorders or syndromes, and its association with adverse outcomes for pediatric brain tumor survivors.
SC and CL were also related to age at diagnosis, with those receiving a diagnosis at a younger age performing significantly worse on these measures. This result diverges from a recent study that did not find a relationship between age at diagnosis and AF (Ashford et al., Reference Ashford, Netson, Clark, Merchant, Santana, Wu and Conklin2014). However, since all the survivors in Ashford et al.’s sample were children closer to treatment, the adaptive behaviors that parents expect of their survivors were likely not as complex as the behaviors required for survivor adults. Adults are expected to create social circles independently, even after significant changes in school networks, neighborhoods, or childhood friendships (Tokuno, Reference Tokuno1986). While simpler social demands may be navigable by survivors during childhood, long-term survivors may demonstrate a dampened developmental trajectory resulting in later problems (McCurdy et al., Reference McCurdy, Rane, Daly and Jacobson2016). An earlier age at diagnosis and treatment may impact the developing brain in ways that are not measurable until later in development (King et al., Reference King, Wang and Mao2015). Ultimately, our study found that adding age at diagnosis and time between diagnosis and assessment as covariates did not change the overall significant mediation or indirect effects. However, these factors should continue to be included in mediation models as covariates, as they are related to long-term AF outcomes in survivors of childhood brain tumors.
Although the indirect effects of ED and apathy were not significant in the models predicting MS or PL, the total indirect effects on both models were significant. NPS alone is a strong predictor of both of these elements of daily living; however, that the models with all their components are also significant reinforces the complexities that underlie survivors’ AF. For instance, higher-order motor abilities as assessed by the SIB-R rely on planning and initiation, as do PL skills such as scheduling doctors’ appointments and performing household maintenance. It remains important that researchers and clinicians conceptualize these activities of daily living as the amalgamation of the functions they require.
This study is strengthened by the heterogeneity of the group overall, as it demonstrates possible pathways of challenges for all pediatric brain tumor survivors. However, the majority of the survivors in our sample (64%) had tumors in the posterior fossa. Recent research has found impacts to connections between the cerebellum and frontal functions via right cerebellar-left frontal white matter tracts in posterior fossa tumor survivors (Ailion, Roberts, Crosson, & King, Reference Ailion, Roberts, Crosson and King2019; Law et al., Reference Law, Smith, Greenberg, Bouffet, Taylor, Laughlin and Mabbott2017). The frontal lobes are thought to be particularly vulnerable for pediatric brain tumor survivors due to later developing white matter connections (Qiu, Kwong, Chan, Leung, & Khong, Reference Qiu, Kwong, Chan, Leung and Khong2007), and treatment factors’ influence on attention span and working memory have been shown to be mediated by cerebrocerebellar pathway microstructure in posterior fossa tumor survivors (Ailion et al., Reference Ailion, King, Roberts, Tang, Turner, Conway and Crosson2020; Law et al., Reference Law, Smith, Greenberg, Bouffet, Taylor, Laughlin and Mabbott2017). This was echoed in the present studies’ findings of greater tumor and treatment factors correlating with poorer executive functioning. Additionally, disruptions in frontal-striatal and frontal-subcortical circuits are associated with apathy (Morettu & Signori, Reference Carvalho, Buelow, Ready and Grace2016), supporting our assessment of this domain and its inclusion in the overall model.
Limitations and Future Directions
The current study demonstrates that the NPS is an effective predictor of long-term AF with performance in some domains mediated by ED. Although strong in its recruitment of heterogeneous long-term survivors at least 5 years past diagnosis, one limitation of this study is that the design is cross-sectional, and future longitudinal studies are needed to fully understand adaptive outcomes for pediatric tumor survivors. Given the small and specific population for recruitment, this study may be susceptible to recruitment bias as survivors chose to participate. For example, survivors who are higher functioning may be more capable and willing to undergo neuropsychological testing. However, survivors who are more impaired may be looking for assistance with the long-term difficulties they are experiencing, and these individuals may intentionally seek out research studies in which to participate. The challenges faced by this group may be resultant of greater tumor and treatment sequelae and therefore a higher NPS score. We conducted tests of data assumptions and plotted the distribution of our predictors and outcome variables to ensure this would not bias the study in one direction. Since we utilized reports from the same informants for both the SIB-R and FrSBe measures, this study may be susceptible to shared method variance. Future studies should utilize more current measures of AF, as the SIB-R includes items that were more relevant when the data were first collected in year 2004 (e.g., using the yellow pages to find a resource).
This study highlights the importance of focusing on developing executive functions in childhood and adolescence. Based on McCurdy et al. (Reference McCurdy, Rane, Daly and Jacobson2016) research with child survivors, the association between ED and low initiation may not significantly impact survivors’ AF until expectations increase as survivors gain greater independence and enter adulthood (McCurdy et al., Reference McCurdy, Rane, Daly and Jacobson2016). Families and providers of survivors should not wait to intervene on these emerging executive abilities that may be impacted by neurological sequalae. The unique challenges experienced by this population suggest that cognitive interventions should be centered around managing ED, addressing AF demands, and increasing independence (Tarazi, Mahone, & Zabel, Reference Tarazi, Mahone and Zabel2007). The present study suggests that management of ED may translate into improved SC and CL skills. Successful interventions may resemble a pediatric pilot cognitive remediation summer program that targeted independence and AF in survivors by providing metacognitive strategy training, skills-based remediation, cognitive behavioral therapy, and parent training (Murdaugh, King, & O’toole, Reference Murdaugh, King and O’toole2019). Given the proposed influence of executive function on AF for pediatric patients with neurological disorders, cognitive remediation programs should work to strengthen these skills in unison so the additive effects lead to optimal outcomes (Tarazi et al., Reference Tarazi, Mahone and Zabel2007). The present study adds to the field by clarifying the specificity of executive function deficits to be targeted and for which individuals, namely, those with more tumor and treatment sequelae.
This study provides further evidence of the utility of a composite measure of tumor treatment and neurological sequelae as a predictor of various late effects. The NPS provides clinical researchers of heterogenous survivor groups with a tool that characterizes neurological sequelae of brain tumor survivors and facilitates research in this less common neurological population. Targeted intervention for ED can be used in conjunction with the NPS to identify the groups most vulnerable to day-to-day challenges in functional independence and set them on a path for improved outcomes as they transition into adulthood.
ACKNOWLEDGEMENTS
This research was supported by a Research Scholar Grant from the American Cancer Society (TZK, #RSGPB-CPPB-114044) and the Georgia State University Second Century Initiative Neurogenomics Fellowship program (RJK).
CONFLICT OF INTEREST
The authors have nothing to disclose.
