Speech mapping in awake high-grade glioma resection: subcortical tract proximity as a predictor of language outcomes.

Abstract

Maximizing extent of resection (EOR) for language-eloquent high-grade glioma (HGG) must be balanced against the risk of neurological deficit. Predictors of postoperative decline in language remain poorly characterized. This study aimed to evaluate intraoperative predictors of postoperative language decline, along with predictors of subsequent temporal recovery. The authors conducted a single-center study of patients undergoing awake craniotomy for the resection of language-eloquent HGG utilizing diffusion tractography, 5-aminolevulinic acid (5-ALA), and subcortical stimulation (SCS) with intraoperative language testing. Cases were reviewed between January 2017 and November 2024. Data assessing intraoperative language function, SCS parameters, EOR, along with language deficit at the 48-hour, 2-week, and 3-month follow-ups were collected. Receiver operating characteristic curves and Youden's Index were used to identify optimal subcortical stimulatory thresholds predicative of postoperative deficit in respective language domains for individual fiber tracts, including the inferior fronto-occipital fasciculus (IFOF), arcuate fasciculus (AF) and superior longitudinal fasciculus (SLF), and inferior longitudinal fasciculus (ILF). A total of 130 patients (78 male, 52 female; mean age 57.1 years) were included. Postoperatively, a new or worsened language deficit was observed in 69 patients (53.1%). This was permanent (> 3 months) in 12 cases (9.2%). Clinical deterioration during awake language testing was a significant predictor of short-term (< 3 months) decline in language domains including semantic processing (IFOF) (RR 3.47, p = 0.0002), reading (ILF) (RR 21.4, p = 0.0025), and auditory naming/repetition (AF/SLF) (RR 6.98, p = 0.001). Factors associated with permanent postoperative language decline were the presence of preoperative speech deficit (RR 2.65, p = 0.020), intraoperative deterioration in ILF-related reading function (RR 8.92, p = 0.0407), and positive SCS of multiple white matter language tracts. Individual stimulation thresholds predictive of functional decline are presented for IFOF, AF, and ILF. This study evaluated multimodal resection of language-eloquent HGGs using awake mapping, 5-ALA, and tractography. It highlights the significant risk of transient decline in language function following eloquent tumor resection, particularly if a patient has an existing language deficit. The posterior ILF with its associated reading function appears to be most sensitive to decline and shows the least propensity for functional recovery. Deficits correlated with specific white matter tract involvement, especially when multiple tracts were affected. Proposed stimulation thresholds offer a novel guide for safer resections, supporting a multimodal strategy to balance maximal tumor removal with language preservation.