Microstates of the cortical brain-heart axis

Abstract

Electroencephalographic (EEG) microstates are brain states with quasi-stable scalp topography. Whether such states extend to the body level, that is, the peripheral autonomic nerves, remains unknown. We hypothesized that microstates extend at the brain-heart axis level as a functional state of the central autonomic network. Thus, we combined the EEG and heartbeat dynamics series to estimate the directional information transfer originating in the cortex targeting the sympathovagal and parasympathetic activity oscillations and vice versa for the afferent functional direction. Data were from two groups of participants: 36 healthy volunteers who were subjected to cognitive workload induced by mental arithmetic, and 26 participants who underwent physical stress induced by a cold pressure test. All participants were healthy at the time of the study. Based on statistical testing and goodness-of-fit evaluations, we demonstrated the existence of microstates of the functional brain-heart axis, with emphasis on the cerebral cortex, since the microstates are derived from EEG. Such nervous-system microstates are spatio-temporal quasi-stable states that exclusively refer to the efferent brain-to-heart direction. We demonstrated brain-heart microstates that could be associated with specific experimental conditions as well as brain-heart microstates that are non-specific to tasks.

Keywords: EEG; HRV; brain-heart; microstates.

FIGURE 1

Experimental results from the cognitive workload (CW) dataset. The left column makes explicit the direction of the brain‐heart interplay (BHI) and heart rate variability (HRV) frequency band involved; the right column reports the global explained variance (GEV), as median across‐subject ± standard deviation, associated to the backfitting operation on the microstate prototypes, which are represented in the central part of the figure.

FIGURE 2

Graphical representation of the brain‐heart interplay (BHI) microstates occurrences in the two experimental conditions of the CW dataset, for both the heart rate variability (HRV) frequency bands considered (top panel for $C_{\text{Brain}\to \mathrm{LF}}$, and bottom panel for $C_{\text{Brain}\to \mathrm{HF}}$. Each histogram's bar represents the subject‐wise sample given by the number of occurrences of given microstates (whose prototype topography is represented at the basis of the histogram bar).

FIGURE 3

Experimental results from the cold pressor test (CPT) dataset. The left column makes explicit the direction of the brain‐heart interplay (BHI) and HRV‐frequency band involved; the right column reports the global explained variance (GEV), as median across‐subject $\pm$ standard deviation, associated with the backfitting operation on the microstates prototypes, which are represented in the central part of the figure.

FIGURE 4

p‐values associated to global dissimilarity (GD) calculated for all BHI brain‐to‐LF (left panel) and brain‐to‐HF (right panel) microstates, extracted during resting state in CPT dataset (rows) and CW dataset (columns). Colored cells and bold text highlight significant GD (p‐value $<.05$).

FIGURE 5

Schematic representation of the proposed computational methodology.