Causal Parametric Mapping of Language Using Direct Electrical Stimulation During Awake Brain Surgery

Introduction: Direct Electrical Stimulation (DES) is the standard for functional language mapping during awake brain tumor surgery. Mapping is based on a binary measure of performance: regions are ‘eloquent’ if DES elicits errors in naming, or ‘safe to remove’ if it does not. Clinically, DES is assumed to have no impact on language processing on correct (negative mapping) trials. We show that response time variance on negative mapping trials contains significant latent information about language function and introduce a framework that integrates positive and negative mapping to inform DES language mapping.

Methods: Analyses included 2,498 picture-naming trials during awake surgery of left perisylvian areas in 19 patients. Response times were measured from picture onset to spoken response on correct trials. Variance in DES intensity, location, duration, and timing (relative to picture onset) was assessed. Parallel analyses spatially mapped the relation of DES parameters to variance in performance at the group level using ‘searchlight’ analyses.

Results: Higher stimulation intensity was associated with increased error rate (p < 0.0004) and slower response times (p < 0.0002). Closer proximity to errors (within or across patients) predicted slower response times on correct trials (p < 0.0001, p< 0.005). DES duration was positively related to error rate (p < 0.0001) and response times (p < 0.0001). Critically, this relation is location-dependent, with the strongest effect in speech motor cortex and anterior supramarginal gyrus. A cross-validated predictive model using multiple linear regression found that DES intensity, location, duration, and timing collectively predicted 54% of response time variance on negative mapping trials (p < 0.0001).

Conclusion : Causal Parametric Mapping provides a novel approach to integrate positive and negative mapping trials into a graded measure reflecting each region’s role in language. These findings suggest that binary mapping methods may underestimate the extent of ‘eloquent’ cortex. An ongoing multisite trial will assess whether indices from negative mapping trials explain unique variance in language outcomes 3months post-surgery.