P06.03.A SEMI-SPONTANEOUS SPEECH WITH ORAL PICTURE DESCRIPTION IN GLIOBLASTOMA PATIENTS BEFORE AND AFTER AWAKE TUMOUR RESECTION

Abstract BACKGROUND Spontaneous speech analysis is a reliable method to detect language problems in neurological populations. Deviations in spontaneous speech of glioma patients consist of incomplete sentences, shorter mean length of utterance (MLU), repetitions, self-corrections and phonological paraphasias. So far, in glioblastoma patients spontaneous speech analyses are scarce. As spontaneous speech analysis is time consuming, an oral picture description task (semi-spontaneous speech) is an alternative which is proven to be sensitive in stroke patients. MATERIAL AND METHODS We analyzed a cohort of 28 patients (part of the ongoing randomized controlled SAFE-trial) with a glioblastoma in left hemispheric eloquent areas who underwent awake surgery. An oral picture description task (Comprehensive Aphasia Test) was administered preoperatively and three months postoperatively and analyzed at conceptual level, productivity, grammatical organization and language errors. Performance was compared to normative data of healthy population. RESULTS In the pre- and postoperative semi-spontaneous speech, most common errors were interjections, incomplete sentences and repetitions. Compared to healthy population, pre- and postoperatively significant (p < 0.05) fewer adequate content words, a shorter MLU, fewer compound sentences and more language errors were produced; fewer causal relations were produced preoperatively. Between pre- and postoperative performance, most variables did not significantly change (p > 0.05) except for a higher percentage of repetitions and more initiation problems postoperatively. Preoperatively, patients with a frontal tumor expressed significantly more causal relations than patients with a temporal tumor (p = 0.025). CONCLUSION An oral picture description task appeared to be sensitive to detect language disturbances in glioblastoma patients. In accordance with earlier research, most errors concerned incomplete sentences. Compared to healthy population, glioblastoma patients are less efficient in transferring content information, using shorter, less syntactically complex sentences with more language errors. The characteristics of spontaneous speech in glioblastoma resemble the performance of lower graded glioma patients albeit more severely impaired. Tumor location in the frontal lobe has a beneficial effect on the expression of causal relations compared to tumor location in the temporal lobe. Oral picture description is less time-consuming than the traditional interviewing setting serving as an efficient clinical tool in standard language examination. As most variables did not significantly change postoperatively, it seems that spontaneous speech in this patient group was spared by awake resection. An RCT is ongoing to prove the benefit of awake surgery in glioblastoma in a larger cohort in relation to survival and quality of life.