Resectability of White Matter Tracts in Patients With Language-Critical Gliomas

Achieving maximal safe resections is critical for patients with language-eloquent gliomas. White matter tracts (WMTs) may offer valuable insights for surgical decision making, but the resectability of specific tracts remains unclear. In this study, we aim to identify predictors of permanent postoperative language deficits (PLDs) using resections of WMTs, illuminating the functional relevance of specific white matter segments.

This retrospective cohort study analyzed 78 consecutive glioma patients undergoing resection of language-eloquent tumors. WMT reconstructions were performed for 6 language-associated tracts. Tracts were parcellated according to anatomic location. The positive predictive value (PPV-tract resections with PLDs divided by total tract resections) was calculated for each tract segment. Resected tract segments (RTS) were normalized to Montreal Neurological Institute space.

WMT resection strongly predicted PLDs (odds ratio = 16.00, P = .0115), with high negative predictive value (NPV, 97.4%), but low PPV (29.6%). RTS in patients with PLDs clustered within the white matter of the temporoparietal junction and middle temporal lobe. RTS without sequela localized more diffusely, with notable concentrations in the frontal operculum, anterior temporal lobe, and orbitofrontal region. Tract segments in the frontal lobe demonstrated near-complete resectability (frontal arcuate fasciculus [AF], frontal inferior fronto-occipital fasciculus, frontal uncinate fasciculus, frontal superior longitudinal fasciculus PPV = 0). Tract segments in the parietal and temporal lobe demonstrated significantly less resectability (parietal AF PPV = 0.4; temporal AF PPV = 1; temporal inferior fronto-occipital fasciculus PPV = 0.5; temporal inferior longitudinal fasciculus PPV = 0.5). The volume of tract resected did not significantly differ according to functional outcome (U = 84.0, P = .70). Resections of parietal and temporal tract segments robustly predicted PLDs (odds ratio = 29.4, P < .001), with a notable increase in PPV (50%) compared with the generalized model.

This study demonstrates that resecting WMTs is a robust predictor of PLDs. WMTs in the orbitofrontal region, anterior temporal lobe, and frontal operculum were resectable whereas tract resections in the parietal and temporal lobes significantly increased risk of PLDs.