Childhood Neurotoxicity and Brain Resilience to Adverse Events during Adulthood

Abstract

Objective: This study used childhood cancer survivors as a novel model to study whether children who experience central nervous system (CNS) injury are at higher risk for neurocognitive impairment associated with subsequent late onset chronic health conditions (CHCs).

Methods: Adult survivors of childhood cancer (n = 2,859, ≥10 years from diagnosis, ≥18 years old) completed a comprehensive neurocognitive battery and clinical examination. Neurocognitive impairment was defined as age-adjusted z score < 10th percentile. Participants impaired on ≥3 tests had global impairment. CHCs were graded using the Common Terminology Criteria for Adverse Events v4.3 (grade 1, mild; 2, moderate; 3, severe/disabling; 4, life-threatening) and were combined into a severity/burden score by frequency and grade (none/low, medium, high, and very high). A total of 1,598 survivors received CNS-directed therapy including cranial radiation, intrathecal methotrexate, or neurosurgery. Logistic regression estimated the odds of neurocognitive impairment associated with severity/burden score and grade 2 to 4 conditions, stratified by CNS treatment.

Results: CNS-treated survivors performed worse than non-CNS-treated survivors on all neurocognitive tests and were more likely to have global neurocognitive impairment (46.9% vs 35.3%, p < 0.001). After adjusting for demographic and treatment factors, there was a dose-response association between severity/burden score and global neurocognitive impairment, but only among CNS-treated survivors (high odds ratio [OR] = 2.24, 95% confidence interval [CI] = 1.42-3.53; very high OR = 4.07, 95% CI = 2.30-7.17). Cardiovascular and pulmonary conditions were associated with processing speed, executive function, and memory impairments in CNS-treated but not non-CNS-treated survivors who were impacted by neurologic conditions.

Interpretation: Reduced cognitive/brain reserve associated with CNS-directed therapy during childhood may make survivors vulnerable to adverse cognitive effects of cardiopulmonary conditions during adulthood. ANN NEUROL 2021;89:534-545.

Figure 1:

Participant enrollment and completion flowchart

Figure 2:. Risk of Global Neurocognitive Impairment Associated with Severity/Burden Score Among A) CNS Treated Survivors and Non-CNS-Treated survivors and B) Specific CNS Exposures.

The bars and error bars below represent odds ratios and 95% confidence intervals for the risk of global neurocognitive impairment (age adjusted Z-score <10^th percentile on ≥3 tests) associated with severity/burden score. The black dotted line represents the null reference association (OR=1, none/low burden score). In both CNS- and non-CNS-treated survivors, a high or very high severity/burden score was associated with an increased risk of global neurocognitive impairment compared to those with no conditions or a low burden score. A dose response relationship between burden score and risk of global neurocognitive impairment exists among CNS-treated survivors. Those with a low severity/burden score have only grade 1 condition, those with a medium score have ≥1 grade 2 and/or 1 grade 3 condition, those with a high score have ≥ 2 grade 3 conditions or 1 grade 4 and 1 grade 3 conditions, those with a very high score have ≥ 2 grade 4 events or ≥ 2 grade 3 conditions and a grade 4 condition. Logistic regression models are adjusted for age at diagnosis, gender, race, time since diagnosis, any high-dose cytarabine and cumulative dose of high-dose methotrexate. The CNS model is further adjusted for neurosurgery, cumulative cranial radiation dose, and intrathecal methotrexate dose.

Figure 3:. Risk of Neurocognitive Impairment Associated with A)Cardiovascular or Pulmonary Conditions and B) Endocrine or Neurologic Conditions.

The bars and error bars below represent odds ratios (OR) and 95% confidence intervals for the risk of neurocognitive impairment associated with any grade 2 or higher condition. The black dotted line represents the null reference association (OR=1, no grade 2 or higher condition). Logistic regression models are adjusted for age at diagnosis, gender, race, time since diagnosis, any high-dose cytarabine and cumulative dose of high-dose methotrexate. Central nervous system (CNS) models are further adjusted for neurosurgery, cumulative cranial radiation dose, and intrathecal methotrexate dose.

Figure 4:. Risk of Neurocognitive Impairment Associated with Cerebrovascular Conditions or Peripheral Neuropathy.

The bars and error bars below represent odds ratios (OR) and 95% confidence intervals for the risk of neurocognitive impairment associated with any grade 2 or higher cerebrovascular conditions (red) or peripheral neuropathy (blue) condition among A) CNS-treated and B) non-CNS-treated survivors. Cerebrovascular conditions included cerebrovascular disease or accidents and intracranial hemorrhage. The black dotted line represents the null reference association (OR=1, no grade 2 or higher condition). Logistic regression models are adjusted for age at diagnosis, gender, race, time since diagnosis, any high-dose cytarabine and cumulative dose of high-dose methotrexate. CNS models are further adjusted for neurosurgery, cumulative cranial radiation dose, and intrathecal methotrexate dose.

Figure 5:. Structural Paths for the Mediation of Treatment Effects on Neurocognitive Function.

Final path models are presented for each outcome. Models are adjusted for sex, age at diagnosis, and time since diagnosis. Model fit indices are represented by the comparative fit index (CFI), Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA). Each model includes only significant paths and is labeled with a standardized beta. Curved lines with double arrows indicate covariances between chronic health conditions. Neurologic, pulmonary, cardiovascular, and endocrine conditions were significantly associated with one another and allowed to covary.