Purpose

Acute lymphoblastic leukemia (ALL) represents the largest diagnostic group of survivors of childhood cancer^1,2. Even when treated with chemotherapy alone, these survivors remain at high risk for neurocognitive impairment, particularly in attention and working memory^3,4. Younger age at diagnosis, female gender, and treatment intensity are known risk factors for cognitive deficits in ALL survivors^5,6. In this study, we used functional magnetic resonance imaging (fMRI) in a large cohort of survivors of ALL treated with a contemporary chemotherapy-only protocol to investigate associations among clinical risk factors, neurocognitive performance and brain activity.

Methods

This study was approved by the IRB and written informed consent was obtained from all participants and/or their parents. Eligible survivors included patients enrolled in the St. Jude Total XVI study, a chemotherapy-only protocol (no cranial radiation) lasting 2.5 years, and with no preexisting neurologic morbidity. Functional imaging was obtained for 92 survivors of ALL (48.4% females; age at exam=13.6± 3.7 years) at the end of therapy. Working memory performance was evaluated based on measures of attention and working memory completed as part of a neurocognitive evaluation in the clinic. Imaging was performed on 3T scanners (Trio or Skyra, Siemens, Malvern, PA) using a single-shot T2*-weighted EPI pulse sequence (TR = 2.06 s, TE = 30 ms, FOV = 192 mm, matrix = 64 x 64, slice thickness = 3.5 mm). Survivors completed Verbal and Object N-back tasks administered at three working memory loads (0, 1 and 2-back). Image data analysis was performed using Statistical Parametric Mapping software (SPM12, Wellcome Institute of Neurology, London, UK). Functional images from each subject were realigned to correct for inter-scan head motion, normalized to the Montreal Neurological Institute brain template and smoothed with a Gaussian kernel of 6-mm full width at half-maximum. The smoothed and normalized images were resampled to 2-mm isotropic resolution. Statistical contrasts were calculated to identify working memory-related networks. Contrast images from each survivor were then used as variables in second-level, random-effect analyses to identify patterns of brain activation. Clinical variables and neurocognitive scores were entered as covariates to identify areas correlated with brain activity. Group differences were tested with a two-sample t-test. Cluster-level analysis was conducted with threshold p<0.001 (uncorrected for multiple comparisons) and minimum cluster size of 5 voxels. The threshold for significance was p<0.05, FWE corrected for multiple comparisons at the voxel-level. Descriptive statistics were calculated using R software (http://cran.r-project.org/) for demographic characteristics and task performance during clinic neurocognitive testing (raw scores and standardized scores) and fMRI. Correlations were tested using Pearson’s test.

Results

N-back performance during fMRI is summarized in Table 1. The N-back task activated an extensive fronto-parietal network including bilateral insula and anterior cingulate cortex (ACC) (Figure 1, 2-back > 0-back). Brain activation in some regions was correlated with neurocognitive performance during the N-back task are shown in Figure 2 (2-back > 0-back). Decreased brain activity in the dorsolateral prefrontal cortex (DLPFC) and the ACC was significantly associated with lower performance on behavioral measures of decision speed and increased omission errors. Increased activity in areas including the right parietal and left posterior cingulum was significantly correlated with lower working memory performance measured in the clinic. Brain activation in some areas during the N-back task was also associated with demographic factors (Figure 3, 2-back > 0-back). Females had greater activation than males, and activation was lower in children who were younger at ALL diagnosis.

Discussion

ALL survivors with poor working memory test performance showed decreased brain activation in areas including the right ACC and the right DLPFC suggesting compromised function in those areas. In contrast, activity increased in right parietal cortex and left posterior cingulum in patients with lower working memory performance, suggestion compensatory function in those areas; however this mechanism seems to be failing. Decreased brain activation was correlated with younger age in areas of the working memory brain network. The increased brain activity in females compared to males could be due to the effect of sex hormones⁷ or other factors like treatment.

Conclusion

Working memory-related brain activation was associated with important clinical risk factors and with neurocognitive performance. The overall pattern of behavioral and imaging results might be explained by altered or compensatory engagement of neural systems to try to maintain task performance despite neurocognitive impairment. Analysis of fMRI responses by medical and demographic factors may help to clarify the pathophysiology of neurocognitive deficits often seen in ALL survivors. Thus, fMRI may help in selecting patients for remedial intervention based on neural phenotypes.

Acknowledgements

We thank Patrick Hanby for his help with this project.