Cardiac autonomic responses to cortical electrical stimulation: a SEEG study

Recent growing neuroimaging evidence support that a set of cortical regions - the central autonomic network - is involved in autonomic control, but its functional organization remains unclear. We studied the direct autonomic cardiac effects produced by 1500 direct cortical electrical stimulations in 43 patients with epilepsy (32.8 ± 8.6 years old, 19 females) undergoing intracerebral recordings during presurgical evaluation. The time course of RR interval (RRI) reactivity and its variability were studied. Nearly half (48.6 %, n = 729) of the cortical stimulations resulted in a cardiac response, divided almost equally between bradycardia (24.47 %) and tachycardia (24.13 %), with no difference between right and left stimulations. Bradycardia was marked by an increase in parasympathetic heart control (increase in HF power and decrease in LF/HF ratio), while tachycardia was marked by a predominance in sympathetic heart control (decrease in HF power and increase in LF/HF ratio). We individualized a main network, where evoked bradycardia and tachycardia were strong, consisting of amygdala, posterior insula, frontal mesial premotor/prefrontal cortex, and anterior cingulate. Other brain regions were also involved, but to a lesser degree, with regions mostly in the limbic system and neocortex (sensory-motor/premotor and lateral temporal regions). These results highlight a close relationship between cerebral cortex and heart. Two hierarchically ordered networks were identified. A 'core' autonomic network strongly involved in cardiovascular regulation, consistent with the classical definition of CAN in functional imaging. But also a more 'widespread' autonomic network, both consistent with a major role of the cortex in continuous autonomic cardiac adjustments to high level emotional, cognitive or sensorimotor cortical activities. This study establishes for the first time a functional mapping of cardiac responses evoked by cortical electrical stimulations, and evidenced hierarchically ordered networks that extends the classical model of CAN.