Poor academic performance in offspring of survivors with childhood or adolescent central nervous system tumor in Sweden

The number of children who were born after their parents were diagnosed with central nervous system (CNS) tumor is increasing, but it remains largely unknown regarding the academic performance of these children. We aimed to investigate whether children of survivors with childhood or adolescent CNS tumor were associated with poor academic performance. Children of survivors of CNS tumor were identified by combining the nationwide Swedish Cancer Register and the Multi‐Generation Register, and those who have completed compulsory education in Sweden between 1989 and 2015 were included in our study. “Poor academic performance” was defined as a z‐score of the academic performance below the 10th percentile. Conditional logistic regression and quantile regression were used to examine the association. A total of 655 children were born after their parental diagnosis of CNS tumor and they had 1.39 times higher risk of achieving poor academic performance as compared to the matched comparisons (95% CI = 1.10‐1.76). The poor academic performance was even more pronounced in boys, among those with a paternal diagnosis of CNS tumor and those with a parental ependymoma. The observed association differed depending on preterm birth. In addition, the strength of the association declined with the increased quantiles of academic performance z‐score. Our data suggest that parental CNS tumor affects the subsequent academic achievements among children born after the parental tumor.


| INTRODUCTION
The development of medical therapies allows a greater number of patients with pediatrics diseases, including patients with central nervous system (CNS) tumor, to survive until parenting age, leading to the increase in the number of children having a parent with physical illness. 1 Available researches suggest that parental physical conditions might play a role in physical and psychological functioning for their children, and both of them might be related to academic performance in school. [1][2][3][4][5] In the Nordic countries, CNS tumor ranks as the second most common cancer among the population below 20 years of age. 6

Its
Abbreviations: CI, confidence intervals; CNS, central nervous system; OR, odds ratio. incidence rate remains increasing but the mortality rate has been slightly decreased, 6 which has led to a growing number of children born after their parents have been diagnosed with CNS tumor. The impact of parental CNS tumor on academic performance in their children might be more pronounced than other cancers, probably due to its neurocognitive late effects. 7,8 The brain is developing continuously during childhood and adolescence and is more sensitive to various tumor treatments. It is well-recognized that parental cognitive function, educational and socioeconomic status play a crucial role iǹ offering educational support. [9][10][11] Furthermore, CNS tumor and its treatments were associated with a higher risk of subsequent reproductive impairment and adverse pregnancy outcomes, such as preterm birth, 8,12,13 which were considered to be associated with long-term health issues as well as academic difficulties. [14][15][16][17][18] It is thus plausible that parental diagnosis with CNS tumor in early life could affect their children's academic performance. However, this topic remains under investigation.
In this nationwide cohort study, we aimed to investigate the association between parental diagnosis with CNS tumor in childhood or adolescence and the poor academic performance among their children and to explore whether histology-and age-specific parental CNS tumor will affect their children's academic performance. In addition, we estimated how the association varied across the distribution of academic performance in quantile regression. The study cohort included children whose mother or father had a previous diagnosis with CNS tumor below 20 years old and survived for more than 5 years. For each child in the study cohort, we randomly selected five children whose parents did not have CNS tumor to generate the comparison cohort. The comparison cohort was matched by the birth year, gender, maternal and paternal age at birth, as well as maternal and paternal highest education.
A unique individual national identification number was assigned to all the residents living in Sweden for more than 3 months.
To ensure confidentiality, the national identification number was replaced by a series of numbers and was used to link the above national registers in our study.

| Assessment of covariates
Parental age at diagnosis with CNS tumor was retrieved from the Swedish Cancer Registry, modeled as childhood survivors (<14 years old) and adolescent survivors (15-19 years old). Parental age at childbirth was collected by linking the Swedish Medical Birth Register and the Swedish Multi-Generation Register. Parental highest education was collected from the Total Population Register, which was created in 1960. It was modeled as 1 to 9 years (compulsory high school or less), 10 to 11 years (practical high school or some theoretical high school) and 12+ years (theoretical high school and/or college and/or postgraduate study) in our study. Preterm birth was defined as a live birth occurring at less than 37 full weeks of gestation, which was obtained from the Swedish Medical Birth Register.

| Assessment of outcome
The outcome was academic performance measured by the final grade achieved after completing the compulsory years of education at age the original grades to ensure comparability between the two grading systems. In our study, we defined "poor academic performance" as a z-score of the academic performance below the 10th percentile in the logistic regression model. And, it was further modeled as continuous variables in the quantile regression model by modeling every tenth quantile from the 10th to the 90th.

| Statistical analysis
A Chi-squared test was used to compare the distribution of the general characteristics between the study cohort and the matched comparisons. Conditional logistic regression was used to calculate the odds ratio (OR) and 95% confidence interval (95% CI) for the association between parental diagnosis with CNS tumor and poor academic performance in their children. The association was further stratified based on the child's gender (male and female), paternal or maternal survivors, and parental age at diagnosis (childhood tumor or adolescence tumor), type of tumor, parental highest education and the time interval between parental diagnosis and childbirth (<11, 11-20 and >20), and the analyses were done using conditional logistic regression by comparing with their matched comparisons. With consideration of preterm birth might act as a mediator for the observed association, we explored further whether the observed General characteristics among offspring of survivors with central nervous system tumor and matched comparisons association was affected by preterm birth through conditional logistic regression analysis and compared to the matched comparisons.
We also calculated the remaining direct association between parental diagnosis and academic performance by multivariate logistic regression after taking into preterm birth and other factors used for the matching.
Quantile regression, which is a distribution-free method, was further used in our study due to the skewness of z scores of academic performance. 22 Rather than the limited estimation on only poor performance in the logistic regression model, this approach allowed us to examine if the association is stable across the entire distribution of z scores (ie, to explore the possible different effects on the good or poor academic performance).
All analyses were performed using SAS version 9.3 (SAS Institute, Cary, North Carolina).

| RESULTS
As shown in Table 1 Note: "Poor academic performance" was defined as z-score of academic performance below the 10th percentile in matched group. Abbreviations: CI, confidence intervals; CNS, central nervous system; OR, odd ratio.
In showed the strongest association with poor academic performance in children.
In Figure 1, we found that preterm birth strengthened the associ- Quantile regression showed that parental diagnosis with CNS tumor was associated with poor academic performance ( Figure 2 and Table S1). The association was decreased with the increasing quantile of z-score. For example, at the 10th percentile of the z-score distribution, children of survivors were associated with a lower z-score of 0.259. The overall estimated coefficient increased to zero at the 60th percentile and then almost remained stable. A similar tendency was observed for the offspring of female survivors as well as for the offspring of male survivors, but paternal diagnosis strengthened the association across the board.

| DISCUSSION
In this nationwide population-based study, parental diagnosis with CNS tumor before the age of 20 years was associated with an increased risk of achieving poor academic performance in their children who were born after their diagnosis. The positive association was more pronounced in boys and in children with a paternal diagnosis of CNS tumor, and declined with the time interval between parental diagnosis and childbirth. However, higher parental education level was likely to weaken the observed positive association. In addition, the strength of the association varied by histological subtype of parental CNS tumor, with the strongest association noted for ependymoma. Preterm birth was suggested to be an important variable for the observed association. The association was noticeable only in the lower quantiles of z-score, suggesting that parental diagnosis was related to a higher risk of "poor" academic performance but not related to lower risk of "good" academic performance.
A Danish registered-based study found that children who experienced parental cancer in early adulthood achieved a slightly lower final grade and had a higher risk of achieving low educational attainment (relative risk ratio: 1.20; 95% CI 1.14-1.25). 4 They observed that the highest risk to get lower socioeconomic attainment was found in children experiencing parental cancer under 4 years old; this indicates the persistence of the effect of parental cancer. 4 The impact of parental cancer experience (children born before parental diagnosis) on children's academic performance was largely represented through psychological functioning. However, children born after parental diagnosis of cancer might be affected by treatment-related late effects, physically and psychologically. But until now, the impact has not been investigated.
In this population-based study, we specifically explored the asso- Furthermore, quantile regression found that the negative association between parental diagnosis and z-score of academic achievements declined with the increasing quantiles of z-score. This result was reasonable and supportive of the main finding in our study, suggesting that parental diagnosis tended to result in an increased risk of getting poor academic performance but did not make a difference in the likelihood of getting a good academic performance.
To the best of our knowledge, this is the first study to explore