Olivier Aron, Laure Salmon, Nicolas Nguyen, Dènahin Toffa, Sami Obaid, Alexis Robin, Dang Khoa Nguyen
Across the spectrum of reported RS-from those triggered by elementary somatosensory or auditory stimuli to more elaborate triggers such as eating, music, reading or praxis-we found that favorable surgical outcomes are more consistently associated with the localization of NuS than with NsT. In RS elicited by elementary stimuli, SOZ often lies adjacent to primary sensory or motor cortices. In contrast, RS provoked by complex or cognitive tasks tend to implicate associative cortical regions, and the latency to ictal onset reflects the degree of network involvement. Notably, specific semiologic patterns-particularly the presence or absence of an aura and the involvement of dorsal versus ventral processing streams-serve as clinical markers to infer SOZ location and its relationship to functional networks.
CONTEXT: Reflex seizures (RS) are defined by their consistent provocation by specific stimuli, encompassing a broad range from elementary sensory inputs to complex cognitive tasks. While RS are often encountered in clinical practice, their surgical management remains sparsely reported and poorly systematized.
METHODS: This Seminar reviews surgical cases of RS using a network-based analytical framework, with a focus on delineating the functional anatomy of (i) networks subserving the triggers (NsT), (ii) networks underlying seizure semiology (NuS), and (iii) seizure onset zones (SOZ), aiming to identify criteria predictive of surgical feasibility.
RESULTS: Across the spectrum of reported RS-from those triggered by elementary somatosensory or auditory stimuli to more elaborate triggers such as eating, music, reading or praxis-we found that favorable surgical outcomes are more consistently associated with the localization of NuS than with NsT. In RS elicited by elementary stimuli, SOZ often lies adjacent to primary sensory or motor cortices. In contrast, RS provoked by complex or cognitive tasks tend to implicate associative cortical regions, and the latency to ictal onset reflects the degree of network involvement. Notably, specific semiologic patterns-particularly the presence or absence of an aura and the involvement of dorsal versus ventral processing streams-serve as clinical markers to infer SOZ location and its relationship to functional networks.
IMPLICATIONS: We propose a tripartite framework for evaluating surgical candidacy in RS based on (1) clinical context (etiology, age), (2) stimulus characteristics (modality and latency), and (3) seizure semiology (ventral vs. dorsal stream involvement). This model enables clinicians to estimate the overlap between epileptogenic and physiological networks and guide the individualized surgical strategies, although advanced neurophysiological mapping remains essential. Despite limited data, our findings support the role of epilepsy surgery in selected RS cases and advocate for prospective studies leveraging multimodal network analyses to optimize patient selection and surgical outcomes.