Session Item

Irradiation of healthy brain tissue can lead to anatomical and functional deficits, a phenomenon known as radiation-induced brain injury. Especially cognitive and executional impairments can lead to a marked decrease in the patient’s quality of life after radiation therapy (RT), and have been linked to morphological changes in the brain. Of particular interest have been white matter, hippocampus and cerebral cortex. Less is known about the effects of RT the subcortical grey matter (GM) structures. Atrophy of deep GM structures is associated with impaired cognitive function in patients with multiple sclerosis, Alzheimer’s disease, and dementia, suggesting these structures are involved in cognitive processes. Therefore, the relation between deep GM volume and RT dose needs to be examined, to help elucidate the cause of post-RT cognitive decline.

We selected 31 patients with high quality follow-up scans who were treated with RT for glioma (grade II-IV) in our institution. The CAT12 (Computational Anatomy Toolbox) was used for the automated preprocessing and segmentation of clinical 3T T1 scans of 1x1x1 mm³ resolution. All scans were bias-field inhomogeneity and noise corrected, and automatically segmented. The following structures were segmented and analysed: amygdala, nucleus accumbens, caudate nucleus, hippocampus, globus pallidus, putamen, and thalamus. Structures completely or partially within the GTV volume were censored from analysis. The rate of volume loss was determined with non-parametric permutation tests for volume and dose.

Significant dose-dependent volume loss was observed in all examined structures, except for caudate nucleus. Rates of volume loss vary from 0.14 to 1.33 % per Gy (corresponding to 4.2% and 39.9% per 30 Gy), and are shown for all structures in Table 1. A 3D representation of the volume loss in some of these structures is shown in Figure 1.

We have found that all subcortical deep grey matter structures, with the exception of caudate nucleus, show dose-dependent volume loss after RT.
These results challenge us to reconsider the currently used sparing strategies in radiation treatment of brain tumours. Presently, hippocampal sparing RT has been adopted in several institutions. However, sparing the dose in the hippocampus leads to higher doses in surrounding cerebral tissues, which we have shown are also susceptible to radiation-induced damage. Therefore, future research has to focus on the relation between clinical outcomes (like cognitive and motor function) and morphologic changes in the entire brain, not just selected structures. This way we can conclusively say which structures should be avoided in RT planning to prevent radiation-induced damage. Precise sparing of healthy brain is possible with novel techniques like proton therapy and VMAT. This could lead to improved cognition and quality of life in patients undergoing treatment for brain tumours.