NCOG-22. ELDER AGE INSTEAD OF LOW EDUCATION PREDICTS POOR LANGUAGE FUNCTION IN RECURRENT LEFT HEMISPHERIC DIFFUSE GLIOMA PATIENTS

Abstract PURPOSE Language processing is multifactorial, requiring interactions between cognitive functions and connections of cerebral cortexes and subcortical fibers in dominant hemisphere. Previously, we identified distinct patient factors and tumor characters influencing language performance in treatment-naïve diffuse glioma. In this study, we aimed to investigate predictive factors between treatment-naïve and recurrent patients with diffuse glioma. MATERIALS AND METHODS Grade 2 to 4 diffuse glioma patients who underwent awake surgery with completed preoperative language assessment were included. The Chinese version of the short-form Boston Diagnostic Aphasia Examination (sfBDAE) was used for language testing. Patient age, sex, education level, tumor grade, language pathway involvement, T1 contrast enhanced (C+), and FLAIR (T2) volume were selected as variables. Univariate and multivariate logistic regression analysis were used on language function impairment. The statistical significance was defined as a p-value less than 0.05 by SPSS20.0. RESULTS 50.7% of all awake surgeries (112 of 221) in 171 patients had repeated operations in the retrospective cohort. Comparing to treatment-naïve patients, patients with recurrent tumor had higher academic year (p=0.016), more subventricular zone invasion (p=0.03), and more subcortical language pathway involvement (p=0.035), whereas showed no difference on language performance. In language function analysis, low education was the sole predictive factor for poor language function in treatment-naïve patients. In sharp contrast with low education, elder age factor was exclusively correlated with poor language function in recurrent patients. Furthermore, times of craniotomy was also an independent predictive factor in recurrent group. CONCLUSIONS Neurocognitive reserve is considered a protective factor against brain damage. Both age and education level are critical factors for cognitive reserve. Whether the phenomenon of a distinct and mutually exclusive patient factor affecting language function between treatment-naïve and recurrent diffuse glioma patients could be explained by the model of cognitive reserve necessitate more study.