AWAKE CRANIOTOMY FOR SPEECH ELOQUENT GLIOBLASTOMA: UNDERSTANDING PREDICTORS OF POST-OPERATIVE SPEECH RECOVERY

Abstract AIMS Maximising extent of resection (EOR) with speech-eloquent Glioblastoma must be balanced against risk of neu- rological deficit. Predictors of post-operative decline in speech remain poorly characterised, along with predic- tors of temporal recovery. METHODS We conducted a single-centre study of patients undergoing awake craniotomy, utilising diffusion tractogra- phy, 5-ALA and subcortical stimulation with intraoperative language testing for primary resection of speech- eloquent IDH-Wildtype Glioblastoma. Cases were reviewed between January 2017 and November 2024. Data was collected assessing intraoperative speech changes, subcortical stimulation parameters, EOR, speech deficit at 48 hours, 2 weeks, and 3 months follow-up. RESULTS 130 patients (78M:52F, mean age 57.1 years) were included. Post-operatively, new or progressive dysphasia was observed in 69 patients (48.5%). This was permanent (>3 months) in just 16 cases (12.3%). Clinical dete- rioration during awake speech testing was a significant predictor of short-term (<3 months) decline in speech domains including auditory naming/repetition (RR of 6.98, p=0.001), semantic errors (RR 3.474, p=0.0002), and reading/visual naming function (RR 21.4, p=0.0025). Key predictors of permanent post-operative speech deficit were the presence of pre-operative deficit RR 2.65 (p=0.020), intraoperative deterioration in reading or visual naming function (RR 8.92, p=0.0407) and positive subcortical stimulation of multiple white matter speech tracts (for 2 tracts RR 10.2, p<0.0001; for 3 tracts RR 13.6, p<0.0001). Using receiver operating curves and Youden’s In- dex, subcortical stimulatory thresholds predicative of post-operative deficit in respective speech domains were calculated for individual fibre tracts including the inferior fronto-occipital fasciculus (IFOF), arcuate fasciculus (AF) and inferior longitudinal fasciculus (ILF) CONCLUSION There is significant risk of transient dysphasia following speech-eloquent tumour resection, particularly if a pa- tient has an existing speech deficit. The ILF and associated visual naming/reading function are most sensitive to intraoperative damage and show least propensity for functional recovery. There are higher risks of permanent speech deficit if multiple tracts are positively stimulated intraoperatively.