Synopsis

Keywords: Tumors, Radiotherapy, DTI, MR spectroscopy

Cranial radiotherapy (CRT) is effective; however, survivors of childhood cancer often suffer from cognitive deficits. Two adult patient groups that received different CRT exposures as children were included, as well as healthy volunteers. The MR-examination comprised of anatomical imaging, DTI and MR spectroscopy of the hippocampus. Differences were found between the group that received highest CRT dose and the control group in hippocampal volume, tNAA/tCr-ratio of hippocampus and several DTI-measures in e.g., corpus callosum. Correlations was found between DTI-measures (MD/RD/AD) in some of the studied white matter tracts and several of the used neuropsychological tests, e.g., BVMT-R Sum and Coding.

INTRODUCTION

Modern cranial radiotherapy (CRT), together with improvements in other treatment modalities has resulted in an increased population of long-term survivors of childhood cancer. However, for tumours in the brain or head-and-neck (H&N) region, the CRT causes radiation exposure also to the healthy brain tissue, which can result in cognitive deficits¹. Late effects have been studied histologically, showing e.g. vascular abnormality, demyelination and white matter necrosis², however, the pathogenesis of these damages is still poorly understood.
Neuroimaging using magnetic resonance (MR) methods can provide information on tissue morphology, physiology and metabolism and have thereby a great potential to provide imaging biomarkers for in vivo brain applications regarding late effects due to CRT.
This study aimed to evaluate very late effects of CRT on neuroimaging biomarkers in adult survivors of malignant childhood posterior fossa tumours, mainly medulloblastoma (MB), and nasopharyngeal tumours (NP), representing two groups with different RT exposures to different parts of the brain, and to compare the neuroimaging results of these survivors to healthy controls (C). A second aim was to study if alterations in neuroimaging biomarkers correlated with reduced cognitive function.

METHODS

Adult childhood cancer survivors treated for two different types of tumours were studied: childhood posterior fossa tumours, mostly medulloblastoma (n=10), nasophayngeal tumours (n=6), as well as a group of age and sex-matched healthy volunteers (n=11). Characteristic data regarding the subjects and their treatment are summarized in table 1. The study was approved by the local ethical committee (ref numbers: 721-2015 with amendment 2021-03732 and 1067-16 with amendment T1068-17).
All MR measurements were performed on a Philips Gyroscan Achieva 3T release 5.1.7 to 5.7.1 (Philips Medical systems, Eindhoven, the Netherlands) using the 32 channel headcoil. The MR protocol comprised of a 3D T1-weighted turbo field echo sequence, DTI (32 directions TE/TR 86/4075 ms, voxel size 2³mm³) and MRS separately in the left and right anterior hippocampus (PRESS, TE/TR 40/2000 ms, VOI size= 10*15*30 mm³).
Hippocampus volumes and intracranial volumes were obtained from FreeSurfer segmentations³. DTI parameters (fractional anisotropy (FA), mean diffusivity (MD), radial diffusivity (RD) and axial diffusivity (AD)) were estimated with FSL⁴ and evaluated in the following tracts generated by TractSeg⁵: cingulum, fornix , superior longitudinal fascicle part I, II and III, uncinate fascicle and corpus callosum (CC; both full tract and all seven subdivisions). MRS data was evaluated with LCModel⁶ using in house simulated basis sets⁷. A preliminary evaluation of the MRS data has been presented at ISMRM 2020⁸.
The evaluation of the neurocognitive function of the cohort has been published⁹. The tests included in the evaluation were: Rey Auditory Verbal Learning Test (RAVLT, measuring verbal learning (RAVLT-Sum) and memory (RAVLT-Ret)), Brief Visuospatial Memory Test-Revised (BVMT-R, measuring visuo-spatial learning (BVMT-R sum) and memory (BVMT-R ret)), Trail Making Test (TMT A and B, measuring visual attention, process speed and mental flexibility(B)), D-KEFS Color-Word Interference Test ((CWIT) 1 to 4, measuring verbal speed and interference, 4 also adds a measure of cognitive flexibility) and WAIS-IV Coding (Coding, measuring mental speed, sustained attention, and visuo-motor co-ordination).
Mann-Whitney U-test was used to test for group differences. Given the exploratory focus of this study, no correction for multiple correlations was performed. Statistical significance was set to p≤0.05 for evaluation of hippocampal volume and MRS, and p≤0.005 for the DTI-measures (FA/MD/RA/AD). The correlation analysis was performed using linear regression and R² is given by the square of the Pearson correlation coefficient.

RESULTS & DISCUSSION

Hippocampal volume was lower in MB compared to NP and controls, figure 1, as expected¹⁰. MRS showed lower mean tNAA/tCr ratio in both hippocampi compared to controls and NP. There was also a not statistically significant tendency for a lower Glu/tCr ratio for MB as compared to NP and controls (fig 2). Lowered tNAA indicates long-term neuronal damage and dysfunction¹¹ while lowered Glu could be caused by neuronal and glial damage and dysfunction¹².
Eight out of 60 possible combinations of evaluated white matter tracts and DTI-measures (FA, MD, AD & RD) differed between MB and controls (fig 3). The only significant difference between MB and NP was found in the CC tract for MD, AD, and RD.
Correlations between neurocognitive measures and DTI measures were primarily found between each of the tests: BVMT-R Sum (visual learning), RAVLT Sum (verbal learning) and Coding (processing speed), and the CC and its subdivisions (fig 4). The R²-coefficient was moderate (R²<0.4), but a general pattern could be seen where MB separated from the other two groups in both neurocognitive performance and DTI measures.

CONCLUSION

This study found differences more than 20 years after CRT between the group that received highest CRT dose and the control group regarding hippocampal volume, tNAA/tCr-ratio in the hippocampus and in several DTI-measures, e.g in corpus callosum. A correlation between e.g. lower mean diffusivity in the studied white matter tracts and several of the neuropsychological tests, e.g. BVMT-R Sum and Coding, was also found.
These results indicate associations between the MR-derived neuroimaging biomarkers and impairments in psychomotor speed, learning and memory following radiation to the brain during childhood. Thus, MRI and MRS could add valuable information to increase the understanding of the pathogenesis of radiation-induced brain damage.