LAUSR

OBJECTIVE Cerebral damage in frontal, parietal, and temporal brain areas and, probably more importantly, their interconnections can lead to deficits in language. However, neural plasticity and repair allow the brain to partly compensate for neural injury, mediated by both functional and structural changes. In this study, the authors sought to systematically investigate the relationship between language performance in brain tumor patients and structural perisylvian pathways (i.e., the arcuate fasciculus [AF]) using probabilistic fiber tracking on diffusion tensor imaging. The authors used a previously proposed model in which the AF is divided into anterior, long, and posterior segments. The authors hypothesized that right-handed patients with gliomas in the language-dominant (left) hemisphere would benefit from a more symmetrical or right-lateralized language pathway in terms of better preservation of language abilities. Furthermore, they investigated to what extent specific tumor characteristics, including proximity to the AF, affect language outcome in such patients. METHODS Twenty-seven right-handed patients (12 males and 15 females; mean age 52 ± 16 years) with 11 low-grade and 16 high-grade gliomas of the left hemisphere underwent 3-T diffusion-weighted MRI (30 directions) and language assessment as part of presurgical planning. For a systematic quantitative evaluation of the AF, probabilistic fiber tracking with a 2 regions of interest approach was carried out. Volumes of the 3 segments of both hemispheric AFs were evaluated by quantifying normalized and thresholded pathways. Resulting values served to generate the laterality index of the AFs. RESULTS Patients without language deficits tended to have an AF that was symmetric or lateralized to the right, whereas patients with deficits in language significantly more often demonstrated a left-lateralized posterior segment of the AF. Patients with high-grade gliomas had more severe language deficits than those with low-grade gliomas. Backward logistic regression revealed the laterality index of the posterior AF segment and tumor grade as the only independent statistically significant predictors for language deficits in this cohort. CONCLUSIONS In addition to the well-known fact that tumor entity influences behavioral outcome, the authors' findings suggest that the right homologs of structural language-associated pathways could be supportive for language function and facilitate compensation mechanisms after brain damage in functionally eloquent areas. This further indicates that knowledge about preoperative functional redistribution (identified by neurofunctional imaging) increases the chance for total or near-total resections of tumors in eloquent areas. In the future, longitudinal studies with larger groups are mandatory to overcome the methodological limitations of this cross-sectional study and to map neuroplastic changes associated with language performance and rehabilitation in brain tumor patients.