Slide Slam A20
Organizing Variables Affecting fMRI Estimates of Language Dominance in Patients with Brain Tumors
Monika Polczynska1; 1University of California, Los Angeles
While functional magnetic resonance (fMRI) has become a useful and widely used method to assess language dominance before brain surgery, the technique has limitations that need to be considered when interpreting its results. Numerous variables can affect the assessment of language dominance using pre-surgical fMRI in patients with brain tumors (Nadkarni et al. 2017, Batouli et al. 2016). This work (Połczyńska 2021) organizes the variables into two broad categories of confounding and modulating factors. Confounding factors give the appearance of changed language dominance. Most confounding factors are fMRI-specific, and the confounding factors would not distort other methods examining language dominance (e.g., the Wada test). The impact of confounding factors on the evaluation of fMRI language dominance can be substantial. There are tumor-related, and fMRI analysis confounds. The tumor-related confounds include tumor characteristics (tumor location, tumor grade, volume, and the age of onset) and tumor-induced conditions (aphasia, prior neurosurgery). The fMRI analysis confounds represent technical aspects of fMRI methods that can also disrupt the assessment of language dominance (e.g., a fixed versus an individual threshold, using a single task versus a panel of tasks, a whole-hemisphere versus a region of interest approach). Modulating factors can modify language dominance without confounding it. Modulating factors are not fMRI-specific, and they can impact language dominance both in healthy individuals and neurosurgical patients. The effect of most modulating factors on fMRI language dominance is suggested to be smaller than that of confounding factors. Modulating factors include demographics (age, sex, handedness), and linguistic factors (early bilingualism, sign language, and language characteristics, e.g., tones). Three retrospective cases of brain tumors in the language-dominant left hemisphere are presented to illustrate how the confounding and modulating factors can impact pre-surgical estimates of fMRI language dominance. The patients performed three fMRI language tasks: object naming, auditory responsive naming, and verbal responsive naming. An experienced neuropsychologist created language maps using minimal pre-processing. Task-related activations were identified through the application of a Pearson’s Correlation Coefficient. An individual threshold was applied for each individual, and a conjunction of the language maps was performed. The approach has been demonstrated to be valid, systematic, and reliable (Benjamin et al. 2017). The first case was diagnosed with WHO grade IV glioblastoma in Broca’s area. Signal loss in and around the lesion from prior resection was reported. The second case had grade III frontal glioblastoma in the left hemisphere. Despite a large tumor volume, she suffered from no language deficits. She displayed left hemisphere dominance with a bilateral representation of Broca’s region. The third case was a left-handed, early Spanish-English bilingual with grade II anaplastic astrocytoma in the left parietal lobe. Spanish and English were left hemisphere dominant, with a bilateral organization of Broca’s area in both languages and the basal temporal language area in Spanish. Intraoperative language mapping was conducted in all three patients. In sum, organizing the multiple variables into the two distinct categories can help interpret the results of pre-surgical language mapping with fMRI.