Ethics Statement

This study was conducted according to the Helsinki human subject doctrine and approved by The USC Ethics Committee/Institutional Review Board.

Preliminary Search Strategy

The primary objective of the current search strategy was to identify all published studies reporting the incidence of neurobehavioral, social, or affective/emotional outcomes in patients diagnosed and treated for childhood craniopharyngioma (≤18 years of age). A detailed systematic search strategy was developed which included all four authors and was conducted using the PubMed database. All search results were limited to studies published between 1990-2012 in order to preferentially select for patients treated with current multimodality regimens and screened using current neuropsychological assessments. Search criteria were based on keywords related to CP in children and neurobehavioral, social, and emotional outcomes (Figure 1) using the following search terms: “Craniopharyngioma and pediatric,” “craniopharyngioma and behavioral,” “craniopharyngioma and neurobehavioral,” “craniopharyngioma and psychiatric,” “craniopharyngioma and children and behavior,” “craniopharyngioma and social,” "craniopharyngioma and emotional," and “craniopharyngioma and depression/anxiety.” In addition, the reference lists of all relevant articles were examined for reports of additionally relevant studies. The study was conducted in accordance with the 2009 preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines for methodology of systematic reviews[10].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Flow diagram detailing the flow of information and study selection through the systematic review process.

https://doi.org/10.1371/journal.pone.0076562.g001

Inclusion and Exclusion Criteria

For inclusion in the review, studies had to: 1) Be written in the English language, 2) Involve human subjects only, 3) Include non-adult cases, (defined as age ≤18) at time of craniopharyngioma diagnosis and treatment, 4) Include a minimum number of 10 cases reported in the series (in order to maximize homogeneity of screening assessments used), and 5) Conduct formal assessment of at least one neurobehavioral/social/emotional outcome using a validated test. Cohort studies, case-control studies, and case-series including greater than 10 subjects were included, whereas case reports were excluded. Posttest data after any treatment of CP, including surgical resection and/or radiation, was sufficient for inclusion. Studies that included a cohort of individuals treated for CP as children as part of a larger study were also included if the analysis for this specific sub-cohort was reported separately. Studies were excluded if: 1) They focused on a clinical intervention, 2) Analysis did not separate data from age (≤18 versus >18 years old) at diagnosis, 3) Analysis did not separate CP from other study or tumor groups.

Implementation of search strategy and study selection process

A preliminary review of the search results was conducted. Four reviewers were assigned a portion of the search results (Figure 1), and independently screened the titles and abstracts of all pertinent results, applying the inclusion/exclusion criteria. If it was unclear from the abstract whether a study met the entry criteria, the full text was assessed independently. A secondary review was subsequently conducted in which full text articles of the studies included from the preliminary search were retrieved. Two reviewers independently assessed the eligibility of each of the full text articles based on the inclusion/exclusion criteria. Following the independent review, two reviewers jointly determined which articles would be included. A total of 4 discrepancies were noted and resolved through following review by a third or fourth independent reviewer. All excluded studies and reasons for exclusion were documented. An electronic database was developed, and two sets of two reviewers were assigned to half the studies for formal in-depth review.

Information extraction and data collection process

For included studies, information was collected regarding characteristics including year of study, study sample size, study design, modalities of craniopharyngioma treatment (i.e. surgical, radiation, or both), age at treatment, follow-up times, neuropsychological assessments used, sex of patients, and proportions of patients with hypopituitarism, diabetes insipidus, hypothalamic involvement, obesity, or hydrocephalus.

The primary outcome(s) recorded were: Prevalence of overall neuro-behavioral dysfunction, social impairment, school dysfunction, emotional/affective dysfunction, and prevalence of patients with impaired health-related quality of life. Secondary outcomes analyzed (if primary outcomes were not available) included reports of mean or median cohort scores of neurobehavioral, social, emotional, or QoL assessments, and any control group scores recorded. For calculation of combined prevalence, a weighted proportion was calculated from all studies reporting pertinent results, based on the sample size with available data.

Potential for Bias

In order to limit potential bias, inclusion and exclusion criteria were defined prior to conducting the literature search. All studies included in this systematic review were reviewed by two independent reviewers, and extracted data from the two independent reviewers was compared to ensure accuracy. Discrepancies were resolved through review and discussion with all authors until consensus was achieved.

There is a considerable amount of variation across the different studies with respect to questionnaires and techniques used to measure neuropsychological, social, school and emotional/affective disturbances. Only studies that used validated assessment tools were included. Assessments used to analyze neurobehavioral dysfunction included: Child Behavior Checklist (CBCL) total score, Youth Self Report total score, Behavior Rating Inventory of Executive Function (BRIEF), Vineland Behavior Adaptive Composite, and the Children’s Global Assessment Scale (C-GAS). Social function was assessed using: CBCL, YSR, Personality Inventory Scale for Children (PIC), Kidscreen-52, and semi-structured interviews with patients and/or parents. School performance was assessed via interviews with parents and/or teachers, and the School Achievement Test. Emotional/affective disturbances were assessed using the Beck Depression Inventory, CBCL, YSR, Personality Inventory Scale for Children, Health Utilities Index Mark 2, Short Mood and Feelings Questionnaire (SMFQ), and the Multistory Depression Inventory for Children (MDI-C). Studies administering the questionnaire directly to patients, or to parents as proxy respondents, were included. Information attainment was also quite variable, in that questionnaires administered over the phone, via mail, or in person were included.

Quality assessment

The formal quality assessment described above ensured that well-validated neurocognitive tools were used in all studies included. As the vast majority of studies had a cohort or case-control design, and therefore lacked randomization, a quality review of randomization was not performed.

Preliminary and Secondary Search Results

Of the 471 studies identified in the preliminary search, 11 met inclusion criteria for analysis of the primary outcome (prevalence of neuro-behavioral, social, or emotional impairment), and 3 additional articles reported mean values only of the case population without prevalence rates. (Table 1) Formal neuropsychiatric assessments were utilized in all included studies and included, but were not limited to, the following: CBCL, Personality Inventory Scale in Children, Pediatric Quality of Life Inventory (Persil), Beck Depression Inventory, HU-12, Kid screen 52, Multistory Depression Inventory for Children (MDI-C), World Health Organization Quality of Life (WHO-QOL), Health Utilities Index (HUI), Psychological General Well-being (PGWB), Short Mood and Feelings Questionnaire (SMFQ), Young Adult Checklist, and Child Health Questionnaire (CHQ-PF50).

Patient Demographics

A total of 502 patients were included among all 11 studies contributing to the primary outcome, with available prevalence rates specifically reported in 304 patients (Table 1)[4,11–20]. Mean follow-up times from treatment ranged from 2 to 17 years. A majority of studies were cohort studies based from single center hospitals. Three additional studies reported secondary outcomes (group means) in 208 patients of a total 451 patients[21–23]. Interventions and comorbidities of the sample populations are reported in Table 2.

Primary outcomes from the eleven included studies are reported in Table 3. Overall neurobehavioral dysfunction was reported in 51 of 90 children (57%) with available data (weighted mean=56.7%)[11,12,15–17]. Social impairment was specifically reported in 91 of 222 (41%) children treated for craniopharyngiomas (weighted mean=41.7%) [4,11,12,14,16–18,20]. Social impairment consisted primarily of behaviors such as withdrawal, internalizing, and hostility. School dysfunction was specifically reported in 48 of 136 patients (35%, weighted mean= 35.3%)[4,11,12,16–18]. Difficulty in maintaining relationships was reported in 17 of 43 patients (40%)[12,17].

One additional study reported mean values for a cohort of 27 patients treated with radiation and surgery for childhood CP[21]. Although baseline CBCL values were not abnormal for the group as a whole, the authors noted that worsened neurobehavioral outcomes were correlated with time from treatment, obesity, presence of diabetes insipidus, and presence of hydrocephalus.

Emotional/Affective Dysfunction in patients treated for Childhood Craniopharyngioma

Emotional-affective dysfunction was reported in 58 of 146 patients (40%) with available data (Table 3), and frequently included depression, anxiety, mood swings, irritability, and emotional outbursts (weighted mean=39.7%)[12–19]. Depressive symptoms were specifically reported in all 58 patients with emotional dysfunction, and anxiety in 8 of 12 patients with pertinent data [15].

Quality of Life in Survivors of Childhood Craniopharyngioma

Quality of life was screened and reported impairment as a prevalence in 95 patients, which was reduced in 49 patients (52%)[4,13,17]. QoL was most often affected due to social and body image issues rather than neurological or cognitive impairments. Although they did not report prevalence of reduced quality of life, 2 additional studies reported affected QoL in cohorts of patients with treated craniopharyngiomas, particularly in social functioning[22,23]. Although Sands et al did not report affected mean values for 22 CP survivors in overall psychosocial quality of life, low average responses were reported for “social functioning” and “parent impact of emotion”[23]. The authors cited “social problems” and “somatic problems” as the most profoundly affected clinical categories. Kendall-Taylor et al reported affected quality of life outcomes across all groups, and a correlation between worsened QoL and age of onset of CP and age of growth hormone deficiency[22].

Etiology of Social and neurobehavioral dysfunction in childhood Craniopharyngioma

The reasons for the high prevalence of social-behavioral and emotional dysfunction in treated craniopharyngioma patients are likely to be multifactorial, and remain to be completely understood. Although outside of the scope of the current systematic review, there is a high prevalence of co-morbid conditions in patients treated for CP, including hypothalamic involvement, hormonal manifestations (hypopituitarism and DI), obesity, and presence of hydrocephalus. Additionally, both neurobehavioral and social dysfunction appears to be disproportionally higher when compared to other treated benign intracranial tumors arising in the pediatric population[29]. Additional potential risk factors which merit consideration with regard to etiology of neurobehavioral disease include age at diagnosis and treatment, time since treatment, tumor size, requirement for CSF shunting, surgical and radiation history, and several others. Neurobehavioral compromise is therefore likely to arise not from one specific etiology, but from a variety of these sources, with particular mechanisms such as frontal lobe dysfunction, hormonal imbalance, and hypothalamic/diencephalic dysfunction each warranting further study. Cranial irradiation following brain tumor resection has also been noted to correlate with neurocognitive changes[30–32]. Finally, many patients and families maintain a chronic fear of tumor recurrence and anxiety regarding prior treatments[27,33].

One intriguing theory behind the impaired social interaction in children with craniopharyngiomas may relate to previously unexplored neuro-endocrinological aspects of the disease, in particular oxytocin and desmopressin deficiency. Considering the known propensity for hypothalamic involvement in children with CP, and the high incidence of permanent diabetes insipid us (loss of functional desmopressin production and secretion) correlating with hypothalamic, infundibular, and neurohypophyseal dysfunction, we postulate that oxytocin deficiency may also be a common development in these patients, and may partially contribute to the observed social dysfunction. Although no data is currently available to support this theory, recent experiments performed following inhaled administration of oxytocin have demonstrated improved potential for trust and social connectivity in children with autism spectrum disorder, conduct disorders, attachment disorders, anxiety, and depression[34,35]. The prosocial effects of oxytocin are known to play an important role in developing attachment and trust, social memory, and fear reduction[36–38]. In an experimental primate study by Smith et al, in which oxytocin or oxytocin-receptor antagonists were administered to marmoset monkeys, the authors demonstrated that oxytocin activity played a role in altering partner-directed social behavior during pair interactions[39]. The authors concluded that central oxytocin may facilitate the process of pair-bond formation and social relationships in marmoset monkeys[39]. The paucity of data regarding oxytocin levels in this population underscore the need for more research in this area; The exact role of oxytocin deficiency in patients with treated childhood CP remains to be further explored.

Risk Factors Associated with Worsened Neurobehavioral Outcome

Although the nature of our systematic review did not permit conduction of a meta-analysis with regard to identifying risk factors correlating with worsened neurobehavioral and social outcomes, primarily because of the heterogeneity of parameters studied across various studies and small numbers of patients per study, individual studies did report selected risk factors correlating with worsened social-behavioral and emotional outcomes. In a retrospective study by Sands et al examining quality of life, and using the CBCL to measure social and behavioral outcomes in children, the authors noted that factors correlating with worsened outcome included retrochiasmatic tumor location, recurrence, and additional surgery[23]. In a large study measuring quality of life in 152 adults treated for childhood CP, worsened QoL was associated with age of onset and onset of GH deficiency[22]. Another study by Dolson et al reviewed social and behavioral outcomes in 27 children treated for CP using the Achenbach CBCL score. They determined that risk factors for worsened outcomes included diabetes insipid us, CSF shunting, requirement for an Ommaya reservoir (for intermittent cyst drainage), and low peak pre-irradiation Growth Hormone levels[21]. In a study by Pedreira et al assessing health-related quality of life on 25 patients treated for childhood QoL, risk factors correlating with worsened functional neurobehavioral outcomes included hypothalamic involvement, younger age of onset, presence of hydrocephalus, and tumor recurrence. Taken together, these findings further suggest that the degree of hypothalamic involvement, endocrinological impairment, and tumor recurrence are likely to correlate with severity of social-behavioral and emotional impairment.

Limitations of Current Study

This systematic review assessed the prevalence of neurobehavioral, social, school and emotional/affective dysfunction in patients diagnosed and treated for craniopharyngioma as a child. Data from the studies included in this systematic review consistently demonstrate that there is high prevalence of neurobehavioral, social, school and emotional/affective dysfunction in patients diagnosed and treated with CP as a child. There is, however, a wide range of variability in factors analyzed, including age at diagnosis, degree of hypothalamic involvement and resection, time since treatment, and variations in treatment approach across the different studies included in the review. Furthermore, there is a high prevalence of clinical factors including diabetes insipid us and obesity in study populations analyzed, which may additionally act as confounding variables. Analyses addressing the significant impact of these variables could not be assessed due to the limited amount of data and small sample sizes.

In addition to limitations from the wide variety in patient characteristics included that may have confounded these findings, the current study is inherently limited by the heterogeneity of neuropsychological assessments utilized and the means of conducting them, thereby potentially introducing informational bias into each study. Potential for selection bias may derive from the recruitment strategy and inclusion criteria of each study (i.e. characteristics of patients with treated CP that did or did not participate in the studies), as well as during the course of follow-up. Finally, the potential for publication bias is an inherent limitation of any systematic review. These limitations highlight the need for prospective studies with larger patient samples using homogenous assessments in order to better understand the influence and potentially confounding effects of these variables.