Abstracts

Rationale: Cognitive impairment is a major comorbidity of frontal lobe epilepsy (FLE) and detrimentally impacts quality of life. Our knowledge of its neural underpinnings, however, is remarkably limited. Here, we employ neuropsychological tests and four cognitive functional MRI (fMRI) paradigms to probe the organization of networks subserving language and working memory (WM) in FLE. We complement voxel-based activation maps with analysis of task effects across established functional networks1, and we profile activation patterns in relation to a macroscale gradient, reflecting an organizational principle of whole-brain connectivity2. We further investigate the relationship between task-related functional profiles and cognitive scores, and assess the influence of disease-related variables on functional imaging patterns.  Methods: We analyzed 84 individuals: 58 patients with drug-resistant FLE (33/25 left/right; 50% female; 52% non-lesional, 48% with small lesion [mostly focal cortical dysplasia], median age at onset: 10 years) and 26 age- and sex-matched healthy controls. Neuropsychological tests assessed IQ, verbal reasoning, WM, executive function, expressive language, verbal and visual learning. Letter fluency and verb generation 3T-fMRI tasks probed language networks. WM-fMRI consisted of a two-back verbal paradigm and a visuo-spatial n-back task entailing two levels of difficulty. Voxel-wise single-subject activation and group comparisons were derived via SPM12. We then aligned functional activation maps to surface templates and summarized activity patterns with respect to seven established functional communities1. We also stratified our findings relative to the principal gradient of functional connectome organization, placing unimodal sensory/motor systems at one end, and high-level transmodal regions at the opposite end2. The gradient was divided into 20 consecutive parts (bins), with task-related signal quantified at each bin. Groups were compared via linear models.  Results: In FLE, impaired performance in most cognitive domains coexisted with normative premorbid IQ (p<0.05, FDR). For language (Fig.1) and verbal WM (Fig.2) tasks, FLE patients showed diffusely reduced frontoparietal cortical recruitment, encompassing dorsal attention, ventral attention, frontoparietal systems, and impaired deactivation of default-mode regions. Profiling task-related changes along the principal gradient highlighted global disorganization of neural recruitment in FLE, implicating disrupted shifts within its middle portion (perception-guided cognition) and increased involvement of brain areas at the transmodal end. For visual WM (Fig.2), enhanced frontoparietal and default-mode activity occurred in FLE during low-level task demands, whilst higher WM load resulted in defective recruitment of additional resources and lower performance. Subgroup analyses indicated no differences between left and right FLE. Multiple regressions substantiated an association between task-relevant network engagement and neurobehavioural performance and unveiled linear relations between early age at onset and disruption of frontoparietal recruitment for WM.  Conclusions: We convey the first functional cartography of language and executive dysfunction in FLE. Pronounced cognitive impairment is paralleled by global disorganization of macroscale networks, implicating both perceptually-guided and transmodal cognitive systems. Changes are independent of seizure focus lateralization but modulated by age at seizure onset. Cognitive system saturation may prematurely occur for low-level task demands, relating to poorer performance. Ref. 1) Yeo et al., J Neurophysiol 2011; 2) Margulies et al., PNAS 2016. Funding: Epilepsy Society, Wellcome Trust (Project Grant No 079474), National Institute for Health Research University College London Hospitals Biomedical Research Centre, Brain Research UK, Berkeley Fellowship (UCL, Gonville and Caius College, Cambridge), Ermenegildo Zegna Founder's Scholarship.