UCSF

Introduction - Radiation therapy (RT) received by brain tumor patients after tumor resection can affect brain connectivity in various degrees. Structural connectivity changes in brain matter can be modeled by diffusion MRI and tractography between areas of the brain that act as nodes of the network. We hypothesize that graph theory measures, including global network descriptors and local node characteristics, may change with the application of radiation therapy and evolve over time.

Methods - Six patients with brain tumor resections were imaged either before or immediately following RT and then again six months later. Diffusion data were registered with a brain atlas (AAL), and post-gad T1- and T2-weighted FLAIR images were clinically marked for tumor-affected tissues. Networks were constructed using 90 AAL regions as nodes and connecting streamlines as edges, weighted by the average fractional anisotropy (FA) within those streamlines. Four global and four local network measures were examined. Tumor-affected node sets were defined in each patient and examined for changes, then compared against contralateral nodes and all non-affected nodes. Obtained network measures were correlated with cognitive data.

Results - Network average clustering coefficient increased significantly between scans (p=0.0313). Local clustering, local efficiency, and node strength showed increasing trends in selected node sets. Node strength increased significantly with T1 and T2 image-based definitions of tumor-affected nodes six months post-RT in a single patient. Correlation was detected between cognitive scores and betweenness centrality in select tumor-affected nodes.

Conclusions - Several connectivity measures appeared to be altered in patients after radiation therapy in a span of six months, including increases in network average clustering. Small sample size limited the level of confidence of observed changes in local measures and correlation to cognitive scores but suggests connectivity changes in tumor-affected regions and betweenness centrality in relation to cognitive scores as points of interests in future studies.