Unresponsiveness induced by sevoflurane and propofol is associated with reduced basal forebrain cholinergic nuclei functional connectivity in humans, a pilot exploratory study

Abstract

Studies suggest the cholinergic system is involved in anesthesia-induced unconsciousness, hence unresponsiveness. A significant source of cholinergic innervation comes from basal forebrain cholinergic nuclei (BFCN), with bi-directional connections between anterior BFCN and the default mode network (DMN). Since DMN functional connectivity (FC) is consistently reduced during anesthesia-induced unresponsiveness in humans, we hypothesized that BFCN-FC during anesthesia-induced unresponsiveness is reduced and particularly, anterior BFCN-FC reductions might be related to DMN-FC reductions. Resting-state fMRI (rs-fMRI) signal correlations (i.e., a proxy for FC) were calculated. FC seeds were anterior and posterior BFCN and the DMN. Rs-fMRI data come from healthy male controls during wakefulness and anesthesia with sevoflurane (n = 15) (at fixed concentrations: 2 and 3 vol%) and propofol titrated to the endpoint of clinical unresponsiveness (n = 12), respectively. FC state differences were tested via paired t-tests; FC changes for anterior BFCN and DMN were associated via correlation analysis. We found reduced anterior and posterior BFCN-FC with sevoflurane and propofol compared to wakefulness. The correlation between reduced DMN-FC-and anterior BFCN-FC reductions was r = 0.57 (p = 0.01) for sevoflurane 3 vol%, r = 0.34 (p = 0.11) for sevoflurane 2 vol% and r = 0.47 (p = 0.06) for propofol. In summary, in this exploratory pilot study, we demonstrated reduced BFCN-FC and a potential correlation between reduced anterior BFCN-FC and DMN-FC during sevoflurane and propofol anesthesia. This suggests DMN changes as a potential factor of anterior BFCN-FC reductions during anesthesia-induced unresponsiveness and BFCN-FC reduction as a potential sign of such state.

1. Introduction

The cholinergic system is a widespread neuromodulatory system modulating sleep, arousal, and cognition (Ananth et al., 2023; Ballinger et al., 2016; Picciotto et al., 2012). Evidence from multiple scales – from micro-/mesoscopic studies in animals to macroscopic studies in humans – indicates that the cholinergic system is involved in the modulation of consciousness and behavior, including anesthesia-induced unresponsiveness. For instance in rodents, intra-thalamic microinjection of nicotine, which modulates cholinergic receptors, in the midline thalamus of anesthetized animals induces wake-like behavior despite continuous administration of sevoflurane (Alkire et al., 2007); chemogenetic stimulation of cholinergic neurons in the basal forebrain promotes both arousal and prefrontal acetylcholine concentration increase (Dean et al., 2022); or cholinergic prefrontal cortical stimulation with carbachol, a mixed cholinergic agonist, results in wake-like behavior despite continuous administration of sevoflurane (Pal et al., 2018). In humans, pharmacological experiments show that administration of physostigmine, a central anticholinesterase inhibitor, reverses the effects of propofol-induced unconsciousness and is prevented by the pre-treatment with scopolamine, a central non-selective muscarinic antagonist (Meuret et al., 2000). Results are less clear with sevoflurane. For example, during mild sedation with sevoflurane (sevoflurane 0.6 vol%), physostigmine did not impact recovery in one study at all (Paraskeva et al., 2002); however, Plourde and colleagues showed that the administration of physostigmine induced wake-like behavior in five out of eight participants, which was associated with the recovery of auditory-evoked potentials, but not to the bispectral index of electroencephalography (EEG) recordings (Plourde et al., 2003).

Cholinergic innervation of the brain originates mainly from two sources, namely from a widespread system of intermingled cholinergic interneurons across, for example, the cortex or striatum (Kljakic et al., 2017; Obermayer et al., 2019), and from distinct neuromodulatory cholinergic nuclei projecting in a largely topographic manner to most parts of the brain (Mesulam et al., 1983; Zaborszky et al., 2015). Concerning the latter projecting system, source nuclei are found in the pontine regions, namely the laterodorsal and pedunculopontine tegmental nuclei (Jones et al., 1987), and in the basal forebrain i.e., the basal forebrain cholinergic nuclei (BFCN) with four neuronal clusters Ch1-Ch4 (Mesulam et al., 1983; Zaborszky et al., 2015). From these, the Ch4 group largely corresponds to the nucleus basalis of Meynert (constituting mainly the posterior BFCN), which provides the major source of brain cholinergic innervation, projecting to large parts of the cortex (Mesulam et al., 1983), while the nuclei Ch1-3 refer to the neurons of the medial septum (Ch1), the vertical (Ch2) and horizontal (Ch3) limb of the diagonal band of Broca (constituting mainly the anterior BFCN) (Mesulam et al., 1983), predominantly innervating the hippocampus and centro-medial cortical regions (i.e., limbic cortex) (Mesulam et al., 1983; Zaborszky et al., 2015; Fritz et al., 2019). Concerning control of or input to BFCN, top-down projections from cortex and subcortex to the BFCN have been described (Souza et al., 2022), including monosynaptic input from prefrontal cortices as well as thalamic midline nuclei (Ananth et al., 2023; Do et al., 2016; Gielow et al., 2017; Hu et al., 2016; Mesulam et al., 1984; Zaborszky et al., 1991).

Functional neuroimaging studies (i.e., mainly functional magnetic resonance imaging (MRI)) have demonstrated that functional connectivity (FC) – i.e., correlation of functional MRI signal time courses - of the cholinergic basal forebrain nuclei (BFCN) during wakefulness mimic cholinergic projections from the BFCN to the cortex, suggesting fMRI-FC as a proxy for BFCN connectivity (Fritz et al., 2019; Markello et al., 2018; Yuan et al., 2019). In particular, FC from anterior and posterior subdivisions of the BFCN towards the cortex is distinct, being largely consistent with topographic BFCN projections i.e., the anterior BFCN fluctuations correlate with those of the anterior cingulate and orbitofrontal cortices, while the posterior BFCN links with, for example, associative cortices. Furthermore, it has also been shown that FC patterns of the anterior BFCN include medial-prefrontal cortical areas, which are associated with functional brain networks – i.e., stable and distinctive fMRI signal correlation patterns –, especially the default mode network (DMN), covering medial cingulate and lateral temporoparietal cortices (Mesulam et al., 1983; Luiten et al., 1987). The disruption of DMN-FC has been related to states of altered consciousness (Vanhaudenhuyse et al., 2010; Demertzi et al., 2014), including anesthesia-induced unresponsiveness with sevoflurane and propofol (Palanca et al., 2017; Boveroux et al., 2010; Guldenmund et al., 2013; Golkowski et al., 2019), making DMN-FC decrease one of the most stable signs of anesthesia-induced unresponsiveness. Such consistent DMN-FC decreases during anesthesia-induced unresponsiveness and controlling cortical projections into BFCN suggest that potential anterior BFCN changes might depend on DMN-FC decreases.

Thus, linking these findings concerning cholinergic effects during anesthesia-induced unresponsiveness and BFCN's prominent role in controlling cholinergic effects across the brain, we focused on BFCN functional connectivity (BFCN-FC) and its role during such state. Furthermore, we focused on BFCN-FC with the DMN due to both the extensive connectivity between BFCN and regions of the DMN and the DMN-FC's sensitivity for anesthesia. We hypothesized that (i) FC of both anterior and posterior BFCN is reduced during states of induced unresponsiveness and that (ii) this reduction, primarily from anterior BFCN-FC, is related to the established reduction of DMN-FC during anesthesia-induced unresponsiveness. Using previously recorded data from healthy participants exposed to either propofol or sevoflurane (Ranft et al., 2016; Jordan et al., 2013), we investigated, in a pilot exploratory way, both anterior and posterior BFCN-FC changes during anesthesia-induced unresponsiveness. In separate analyses, we implemented a seed-based FC approach with the anterior and posterior BFCN as seeds. We tested the effects of sevoflurane- (i.e., 2 and 3 vol%) and propofol-induced unconsciousness on BFCN-FC and its association with DMN-FC reduction.

2. Methods

The analyzed datasets were derived from previously published simultaneous electroencephalogram (EEG)-fMRI studies in healthy male adults performed at the Technical University of Munich, Germany, investigating sevoflurane (Ranft et al., 2016) and propofol (Jordan et al., 2013) effects on brain activity, respectively. Furthermore, fMRI data from both studies have been recently re-analyzed to investigate anesthesia effects on functional connectivity dynamics across several brain networks (Golkowski et al., 2019). Both studies were in line with the Declaration of Helsinki and approved by the Ethical Committee of the Medical School of the Technical University of Munich (Sevoflurane study, approval number 5602/12; Propofol study, approval number 2301-09). Study participants were given detailed information about the methods and potential risks and gave their written informed consent before the experiments. A detailed description of the participants and study protocol is also reported in their original publications (Ranft et al., 2016; Jordan et al., 2013).

2.1. Participants and anesthesia

2.1.1. Sevoflurane study

In brief, twenty healthy adult males (20-36y, mean 26.0y) were recruited for the study. Combined EEG-fMRI data were acquired in a repeated measures design, collecting three measurement time points for each subject, namely (in chronologic order in terms of data acquisition): (i) wakefulness pre-anesthesia, (ii) sevoflurane 3 vol%, and (iii) sevoflurane 2 vol%. Incomplete data sets (N = 4) were excluded due to missing data attributed to technical problems during scanning, one extra exclusion was done due to corrupted data after preprocessing (see below). 15 participants provided complete data sets, i.e., 350 vol of resting-state fMRI for each state.

Sevoflurane was administered in oxygen via a tight-fitting facemask using an MRI-compatible anesthesia machine (Fabius Tiro, Dräger, Germany). Sevoflurane as well as oxygen and carbon dioxide were measured by a cardiorespiratory monitor (Datex AS/3, General Electric, USA); standard monitoring according to the American Society of Anesthesiologists was performed. An end-tidal sevoflurane concentration of 0.4% was administered for 5 min, then increased in a stepwise fashion by 0.2 vol% until the participant was unresponsive, which was defined as loss of responsiveness corresponding to a Ramsey Scale of 6. The Ramsay Sedation Scale is a clinical metric that classifies the awareness level into six categories, from agitation (Score 1) to unresponsiveness (Score 6) (Ramsay et al., 1974). After loss of responsiveness, sevoflurane concentration was increased to reach an end-tidal concentration of 3% at which a laryngeal mask was inserted (i-gel, Intersurgical, United Kingdom). Data from steady levels with 3 vol% (state ‘sevoflurane 3 vol%‘) and 2 vol% (state ‘sevoflurane 2 vol%‘) were recorded for 10 min each.

2.1.2. Propofol study

In brief, 15 adult healthy males (21-32y, mean 25.8y) were recruited for the study; analogous to the sevoflurane study, combined EEG-fMRI data were acquired. For each subject, two states were acquired, namely “wakefulness pre-anesthesia” (i.e., wakefulness before the intervention) and “propofol-induced anesthesia”. 12 participants provided complete data sets, i.e., 300 vol of resting-state fMRI for each state. Data from three participants were excluded due to excessive head movement during the recordings (see below). Propofol was induced by propofol titration using a target-controlled infusion (TCI) pump (Open TCI; Space infusion system; Braun Medical, Melsungen, Germany) to obtain constant effect-site concentrations (i.e., start at 1.2 μg/ml, increasing steps in 0.4 μg/ml) up to the loss of responsiveness corresponding to a Ramsay Scale of 5–6. When a steady state was reached (propofol concentrations were maintained for 10 min to ensure equilibrium), recordings of “propofol-induced anesthesia” were recorded for 10 min.

2.2. Neuroimaging data acquisition and preprocessing

Data acquisition for both experiments, sevoflurane and propofol, were performed on a 3T whole-body MRI scanner (Achieva Quasar Dual 3.0T 16CH, The Netherlands) with an 8-channel, phased-array head coil. For estimating BOLD fluctuations, a T2∗-weighted gradient echo-planar imaging sequence with the following parameters was implemented: echo time = 30 ms, repetition time = 2000 ms, flip angle = 75°, field of view = 220 × 220 mm, matrix size = 72 x 72, 32 axial slices, acquisition order “interleaved, odd first”, slice thickness = 3 mm with an interslice gap of 1 mm. We used a magnetization-prepared rapid acquisition with gradient echoes (MPRAGE) sequence to assess high-resolution T1-weighted anatomical images (voxel size = 1 mm isotropic) were acquired before the rs-fMRI scanning session.

FMRI data preprocessing was performed using the Data Processing Assistant for Resting-State fMRI (DPARSF, http://rfmri.org/DPARSF) (Chao-Gan et al., 2010) from the toolbox for Data Processing & Analysis of Brain Imaging (DPABI, htttp://rfmri.org/DPABI) (Yan et al., 2016) based on the software package Statistical Parametric Mapping (SPM12, htttp://www.fil.ion.ucl.ac.uk/spm) implemented in MATLAB v18b. Before computing canonical preprocessing steps, two physiological and motion noise correction procedures were applied to the raw fMRI data to account both for physiology- (e.g. heartbeat) and motion-induced artifacts (e.g. vessel movements, particularly in the brainstem): 1) Physiological noise regression was based on Physiologic Estimation by Temporal Independent Component Analysis (PESTICA, http://www.nitrc.org/projects/pestica/), which implements a temporal independent component analysis (ICA) to estimate time courses related to cardiac and respiratory fluctuations (Beall et al., 2007). These monitored pulse and respiration cycles were used for retrospective image correction and voxel-wise physiological noise correction (Glover et al., 2000). 2) Slice Oriented Motion Correction (SLOMOCO, http://www.nitrc.org/projects/pestica, version 4) was used to regress slice-wise rigid body motion parameters according to a second order-voxel and a slice-specific motion regression model (Beall et al., 2014). After performing these correction procedures, 15 out of the 16 subjects from the sevoflurane cohort passed a visual control of the functional images. The functional images of the excluded subject included systematic band-like artifacts across slices, preventing downstream analyses. Further preprocessing steps included discarding the first eight volumes, re-alignment, slice-time correction, co-registration, head motion correction following the Friston 24-parameter model, wavelet-based despiking to avoid sudden involuntary movement artifacts especially during anesthetized states, cerebral spinal fluid and white matter regression, normalization to the MNI152 3 mm space, linear detrending, bandpass filtering (0.01–0.1 Hz) and smoothing using a Gaussian kernel of 6 mm full width at half maximum (FWHM). All participants’ head motion in the sevoflurane data set remained under the threshold for exclusion, defined as a mean framewise displacement >0.2 mm (Power et al., 2014). Three subjects were discarded from the propofol data set due to excessive motion. Motion during sedation states is excepted, especially considering that during neither. The sevoflurane and propofol study experiments did not use any co-adjuvant drug to avoid involuntary movement as it is used in clinical practice.

2.3. FC outcomes and statistical analysis

2.3.1. Estimation of BFCN and DMN functional connectivity maps

Seed-based functional connectivity analyses for both anterior and posterior BFCN, as well as DMN for each participant and session were calculated separately using the Data Processing Assistant for Resting-State fMRI (DPARSF, http://rfmri.org/DPARSF) (Chao-Gan et al., 2010): Seeds for anterior and posterior BFCN were derived from Fritz et al., 2019) (Fig. 1, Fig. 2 A); the DMN seed was defined as the collection of 34 spherical (i.e., 5 mm diameter) regions-of-interest (ROIs), which were created based on Dosenbach and colleagues (Dosenbach et al., 2010) and available at the DPARSF toolbox (Fig. 3 A). To validate the topographical overlap of our results with the DMN and to define ROIs for the downstream analyses, we used a canonical DMN template derived from Yeo and colleagues (Yeo et al., 2011) (cf., e.g., blue mask in Fig. 1 A1 and B1).

Fig. 1.

Anesthesia-induced unresponsiveness is associated with a reduced anterior BFCN-FC. A. Anterior BFCN mask for seed-based FC analysis. The mask is derived from (Fritz et al., 2019). B. Sevoflurane study. B1. Anterior BFCN-FC during wakefulness pre-anesthesia. One-sample t-test of anterior BFCN-FC during wakefulness. B2. Anterior BFCN-FC reduction during sevoflurane 3 vol%-induced unresponsiveness. Paired t-test anterior BFCN-FC t-contrast wakefulness > sevoflurane 3 vol%. B3. Anterior BFCN-FC reduction during sevoflurane 2 vol%-induced unresponsiveness. Paired t-test anterior BFCN-FC t-contrast wakefulness > sevoflurane 2 vol%, for figure rendering purposes only the height p-threshold was p < 0.01, FWE cluster-level corrected. C. Propofol study. C1. Anterior BFCN-FC during wakefulness pre-anesthesia. One-sample t-test of anterior BFCN-FC during wakefulness. C2. Anterior BFCN-FC reduction during propofol-induced unresponsiveness. Paired t-test anterior BFCN-FC t-contrast wakefulness > propofol-induced unresponsiveness.

Fig. 2.

Anesthesia-induced unresponsiveness is associated with a reduced posterior BFCN-FC. A. Posterior BFCN mask for seed-based FC analysis. The mask is derived from (Fritz et al., 2019). B. Sevoflurane study. B1. Posterior BFCN-FC during wakefulness pre-anesthesia. One-sample t-test of posterior BFCN-FC during wakefulness. B2. Posterior BFCN-FC reduction during sevoflurane 3 vol%-induced unresponsiveness. Paired t-test posterior BFCN-FC t-contrast wakefulness > sevoflurane 3 vol%. B3. Posterior BFCN-FC reduction during sevoflurane 2 vol%-induced unresponsiveness. Paired t-test posterior BFCN-FC t-contrast wakefulness > sevoflurane 2 vol%, for figure rendering purposes only the height p-threshold was p < 0.01, FWE cluster-level corrected. C. Propofol study. C1. Posterior BFCN-FC during wakefulness pre-anesthesia. One-sample t-test of posterior BFCN-FC during wakefulness. C2. Posterior BFCN-FC reduction during propofol-induced unresponsiveness. Paired t-test posterior BFCN-FC t-contrast wakefulness > propofol-induced unresponsiveness.

Fig. 3.

Anesthesia-induced unresponsiveness is associated with a reduced within-DMN-FC. A. DMN mask for seed-based FC analysis. The 5-mm sphere-DMN-ROIs were created based on DMN coordinates from (Dosenbach et al., 2010). B. Sevoflurane study. B1. DMN-FC during wakefulness pre-anesthesia. One-sample t-test of DMN connectivity during wakefulness. B2. Within-DMN-FC reduction during sevoflurane 3 vol%-induced unresponsiveness. Paired t-test DMN-FC t-contrast wakefulness > sevoflurane 3 vol%. B3. Within-DMN-FC reduction during sevoflurane 2 vol%-induced unresponsiveness. Paired t-test DMN-FC t-contrast wakefulness > sevoflurane 2 vol%. C. Propofol study. C1. DMN-FC during wakefulness pre-anesthesia. One-sample t-test of DMN-FC during wakefulness. C2. Within-DMN-FC reduction during propofol-induced unresponsiveness. Paired t-test DMN-FC t-contrast wakefulness > propofol-induced unresponsiveness.

To derive the FC maps for the different ROIs (i.e., anterior/posterior BNFC as well as DMN), the mean averaged time course across all voxels within the respective seed ROI mask was correlated with the time course of each voxel within the brain following a voxel-wise bivariate Pearson correlation analysis. For each seed ROI, this procedure generated a subject-wise correlation coefficient map per session (i.e., wakefulness pre-anesthesia and the different anesthesia-induced unresponsiveness-states) per anesthetic (i.e., sevoflurane and propofol-induced unresponsiveness). The correlation maps were normalized using a Fisher r-to-z-transformation to enable across-voxel and across-subject comparisons. As a result, each subject produced one z-FC map per ROI per session per anesthesia data set. Analyses of differences were done by extracting z correlation values from the respective maps.

2.3.2. Statistical comparisons

For statistical analysis, we used the Statistical Parametric Mapping (SPM12, htttp://www.fil.ion.ucl.ac.uk/spm) toolkit implemented in MATLAB v.18b. We report condition differences using a threshold of p < 0.05 corrected for the family-wise error rate (FWER, using a voxel-wise height threshold p < 0.005), to account for multiple comparisons. We chose this thresholding scheme based on current best-practice guidelines (Lieberman et al., 2009; Meteyard et al., 2020) and previous reports (Ben Simon et al., 2018; Halai et al., 2014; Iannaccone et al., 2015; Poldrack et al., 2017).

2.3.2.1. BFCN-FC during anesthesia-induced unresponsiveness

Our first research question was to test whether BFCN-FC during anesthesia-induced unresponsiveness anesthesia-induced unresponsiveness is reduced. To address this question, we first checked the general topography of BFCN-FC during wakefulness (pre-anesthesia) using a one-sample t-test, in which we probed all subjects’ BFCN-FC maps against zero, as described above. To study the relationship of these FC patterns with the DMN, we overlaid the resulting statistical BFCN-FC maps to a previously published canonical DMN template from Yeo and colleagues (cf. Fig. 1 B1 and C1). Second, we tested the different anesthesia-induced unresponsiveness states (i.e., 2 and 3 vol% sevoflurane and propofol-induced unresponsiveness) against wakefulness using paired t-tests. Again, since our main interest was brain regions associated with the DMN, we checked the overlap of resulting statistical maps with the canonical DMN template (Fig. 1 B1-B3, C1-C2).

2.3.2.2. Are BFCN-FC changes associated with DMN-FC changes during anesthesia-induced unresponsiveness?

Our second research question was whether anterior BFCN-FC reductions were associated with DMN-FC reductions during anesthesia-induced unresponsiveness. To do so, we first analyzed – as described above for BFCN-FC - the general topography of DMN-FC within our sample using one-sample t-tests for wakefulness state (Fig. 3 B1 and C1); then we compared DMN-FC of the different anesthesia-induced unresponsiveness states against wakefulness using paired t-tests (Fig. 3). Second, concerning the link between BFCN-FC changes and DMN-FC changes during anesthesia-induced unresponsiveness, we applied correlation analysis to test for the association of DMN-FC changes (ΔDMN-FC) and anterior BFCN-FC changes (Δ anterior BFCN-FC) during anesthesia-induced unresponsiveness, as follows (cf. also Fig. 4): To assess Δ anterior BFCN-FC, we extracted z-correlation values from z-FC maps per subject from those voxels for which BFCN-FC significantly decreased during anesthesia-induced unresponsiveness at the group level (cf. right column in Fig. 1), however only considering those voxels which overlapped with the DMN (cf. blue mask in Fig. 1). We calculated the difference of wakefulness and anesthesia-induced unresponsiveness of anterior BFCN (i.e., Δ anterior BFCN-FC) by subtracting the z-correlation coefficients for each subject both during wakefulness and anesthesia-induced unresponsiveness (e.g., wakefulness - sevoflurane 3 vol%). ΔDMN-FC was assessed in the same way. This left us with two values for each of the 15 subjects, which we correlated and tested against 0 using a one-tailed t-test (Fig. 4).

Fig. 4.

Correlation analyses anterior BFCN. A. Sevoflurane study. A1.$\Delta$Wakefulness pre-anesthesia vs. sevoflurane 3 vol%.$\Delta$ Anterior BFCN-anterior DMN-FC reduction explained 33% of the $\Delta$ within-DMN-FC reduction during sevoflurane 3 vol% (r = 0.57, R² = 0.33, p = 0.01, one-tailed correlations were bootstrapped (n = 1000) 95% CI: [0.35; 0.79]). These results were tested at an alpha level of p_FWE < 0.05. A2.$\Delta$Wakefulness pre-anesthesia vs. sevoflurane 2 vol%.$\Delta$ Anterior BFCN-anterior DMN-FC reduction explained 11% of the $\Delta$ within-DMN-FC reduction during sevoflurane 2 vol% (r = 0.34, R² = 0.11, p = 0.11, one-tailed correlations were bootstrapped (n = 1000) 95% CI: [-0.33; 0.74]). These results were tested at an alpha level of p_FWE < 0.05. B. Propofol study.$\Delta$Wakefulness pre-anesthesia vs. propofol.$\Delta$ Anterior BFCN-anterior DMN-FC reduction explained 22% of the $\Delta$ within-DMN-FC reduction under propofol-induced unresponsiveness (r = 0.47, R² = 0.22, p = 0.06, one-tailed correlations were bootstrapped (n = 1000) 95% CI: [0.05; 0.86]). These results were tested at an alpha level of p_FWE < 0.05.

3. Results

3.1. Anesthesia-induced unresponsiveness is associated with reduced BFCN-FC

Our first research question addressed BFCN-FC reductions during anesthesia-induced unresponsiveness; mainly, we wanted to test whether potential BFCN FC reductions occur within the DMN i.e., overlap with the DMN. Fig. 1 summarizes the results of our analysis. We found that during wakefulness, anterior BFCN-FC covers mainly the anterior prefrontal cortex (Fig. 1 B1 and Table 1), overlapping especially with anterior parts of the DMN (cf. overlapping regions of the statistical map in hot colors and the DMN-map in blue in Fig. 1 B1 and C1). We made this observation both for the subject samples of the sevoflurane (Fig. 1 B1 and Table 1) and the propofol study (Fig. 1 C1 and Table 2), indicating the consistency of BFCN-FC. Critically, we found reductions of anterior BFCN-FC for all states of unresponsiveness induced by sevoflurane 2 and 3 vol% and propofol (hot color maps in Fig. 1 B2, B3, Table 1, Table 2), suggesting reduced anterior BFCN-FC during anesthesia-induced unresponsiveness. We also found a significant overlap between the regions exhibiting reduced anterior BFCN-FC and the DMN (blue maps in Fig. 1).

Table 1.

BFCN-FC during wakefulness and reduction during anesthesia-induced unresponsiveness.

Regions	Hemisphere	MNI Coordinates [x;y;z]	Z value	Cluster size	P value
Sevoflurane
Anterior BFCN-FC
1)One sample t-test
Subcallosal cortex	R	[3; 6;-6]	7.75	4629	<0.001
Subcallosal cortex	L	[-6; 6;-9]	7.17
Accumbens	R	[15; 18;-6]	6.88
2) Paired t-test awake pre-anesthesia > sevoflurane 3 vol%
Accumbens	R	[15; 18;-6]	6.22	1403	<0.001
Subcallosal cortex	L	[0; 15;-3]	5.77
Caudate	L	[-15; 21;-3]	5.29
3) Paired t-test awake pre-anesthesia > sevoflurane 2 vol%
Frontal orbital cortex	L	[-15; 36;-18]	3.83	323	<0.001
Brainstem	L	[0;-21;-15]	3.62
Subcallosal cortex	R	[3; 30;-15]	3.50
Posterior BFCN-FC
1) One sample t-test
Accumbens	L	[-21; 3;-9]	7.65	4986	<0.001
Putamen	R	[27;-6;-9]	7.24
Putamen	L	[-27;-6;-9]	7.18
Anterior cingulate cortex	L	[-6; 39; 3]	5.28	676	<0.001
Anterior cingulate cortex	R	[3; 24; 24]	4.50
Supplementary motor cortex	L	[-6; 6;54]	4.43
Cerebellum	L	[-9;-78;-42]	5.18	375	<0.001
Cerebellum	L	[-18;-69;-21]	4.50	195	0.001
2) Paired t-test awake pre-anesthesia > sevoflurane 3 vol%
Caudate	L	[-6; 9;3]	5.54	1511	<0.001
Amygdala	L	[-27;-12;-15]	5.29
Putamen	R	[27; 9;6]	5.28
Anterior cingulate cortex	L	[-3; 36; 6]	4.08	83	0.044
3) Paired t-test awake pre-anesthesia > sevoflurane 2 vol%
Cerebellum	R	[9;-81;-42]	4.29	140	0.004
Cerebellum	L	[-18;-78;-45]	4.08
Putamen	L	[-30;-15;-9]	4.25	295	<0.001
Parahippocampus	L	[-21;-27;-18]	4.23
Thalamus	L	[-3;-3;-6]	3.98
Frontal orbital cortex	R	[36; 30;-3]	4.03	220	<0.001
Insula	R	[33; 18; 3]	3.74
Frontal orbital cortex	L	[-33; 30; 3]	3.74	108	0.017
Insula	L	[-36; 15;-6]	3.37

One sample t-test and paired t-tests, significance threshold was set to p ≤ 0.05. We corrected for inflation of false positive rates associated with multiple testing using family wise error (FWE) cluster-level extend correction with a voxel-wise height threshold of p < 0.005. ∗Paired t-test anterior BFCN wakefulness > sevoflurane 2 vol%, voxels-wise height threshold p < 0.01 FWE corrected.

Table 2.

BFCN-FC during wakefulness and reduction during anesthesia-induced unresponsiveness.

Regions	Hemisphere	MNI Coordinates [x;y;z]	Z value	Cluster size	P value
Propofol
Anterior BFCN-FC
1) One sample t-test
Subcallosal cortex	L	[-3; 6;-9]	6.76	3478	<0.001
Caudate	L	[-12; 18;-3]	6.18
Supplementary motor cortex	L	[-3; 9;54]	4.52	411	<0.001
Superior frontal gyrus	L	[-24; 3;57]	4.32
Anterior cingulate cortex	L	[-6; 9;30]	4.29
Precuneus	L	[0;-57; 6]	4.03	88	0.019
Precuneus	R	[3;-54; 18]	3.37
Precental gyrus	L	[-51;-3; 42]	4.01	92	0.015
Middle frontal gyrus	L	[-45;-6; 36]	3.46
Occipital fusiform gyrus	L	[-33;-63;-21]	3.47	84	0.024
Cerebellum	L	[-30;-54;-27]	3.46
2) Paired t-test awake pre-anesthesia > propofol
Anterior cingulate cortex	R	[6; 36;-6]	4.78	288	<0.001
Paracingulate gyrus	R	[12; 48;-3]	4.06
Anterior cingulate cortex	L	[-6; 36;-6]	3.77
Posterior BFCN-FC
1) One sample t-test
Parahippocampus	L	[-18; 3;-18]	6.52	4639	<0.001
Putamen	L	[-21; 0;-9]	6.12
Amygdala	R	[27; 0;-12]	6.03
Supplementary motor cortex	L	[-3; 0;57]	5.36	293	<0.001
Cerebellum	L	[-21;-72;-51]	4.75	685	<0.001
Cerebellum	R	[15;-78;-51]	4.34
2) Paired t-test awake pre-anesthesia > propofol
Inferior frontal gyrus	R	[42; 39; 3]	3.92	104	0.008
Occipital fusiform gyrus	L	[-27;-72;-6]	3.77	90	0.018
Supracalcarine cortex	L	[-24;-60;-15]	3.72

One sample t-test and paired t-tests, significance threshold was set to p ≤ 0.05. We corrected for inflation of false positive rates associated with multiple testing using family wise error (FWE) cluster-level extend correction with a voxel-wise height threshold of p < 0.005. ∗Paired t-test anterior BFCN wakefulness > propofol, voxels-wise height threshold p < 0.01 FWE corrected.

For the posterior BFCN, we found – for the sevoflurane study - that during wakefulness, FC covers mid-cingulate regions (Fig. 2 B1 and Table 1), with only minimal overlap with the DMN in most anterior parts of the dorsal cingulate cortex. For the propofol study, while FC during wakefulness covers broadly the same regions as for the sevoflurane study (Fig. 2 C1 and Table 2), there was no overlap with the DMN (Fig. 2 C1). Reductions of posterior BFCN-FC were primarily observed in states of responsiveness induced by sevoflurane 2 and 3 vol% (Fig. 2 B2, B3 and Table 1) and only minimal during propofol-induced unresponsiveness (Fig. 2 C2 and Table 2); all of these were outside of DMN boundaries (Fig. 2 B2, B3 and C2).

3.2. Reduction of DMN-FC during anesthesia-induced unresponsiveness is associated with reduced anterior BFCN-FC

Our second research question addressed whether DMN-FC reductions during anesthesia-induced unresponsiveness are linked to anterior BFCN-FC reductions. We verified the commonly-observed DMN-FC reduction (Tables S1 and S2) during anesthesia-induced unresponsiveness also in our sample as follows: first, we confirmed substantial overlaps of DMN-FC in both our study populations and a canonical DMN mask, indicating robust DMN-FC mapping (cf. overlapping regions of the statistical map in hot colors and the DMN-map in blue in Fig. 3 B1 and C1). Secondly, we found DMN-FC reductions for all three states of responsiveness, namely induced by sevoflurane 2 vol%, 3 vol%, and propofol (hot color maps in Fig. 3 B2, B3, C2, Tables S1 and S2). Finally, we assessed associations of BFCN-FC reductions within the DMN to within-DMN-FC reductions using a correlation analysis: For the sevoflurane 3 vol% state, we found a significant positive association (R² = 0.33, p = 0.01), i.e., the more anterior BFCN-FC was reduced during sevoflurane 3 vol%-induced unresponsiveness, the stronger was the within-DMN-FC-reduction (Fig. 4 A1). For sevoflurane 2 vol% the results were inconclusive: R² = 0.11 (p = 0.11) (Fig. 4 A2); as well as for propofol-induced unresponsiveness R² = 0.22 (p = 0.06) (Fig. 4 B).

To test for the specificity of the association between ΔDMN-FC and Δ anterior BFCN-FC during anesthesia-induced unresponsiveness for the anterior BFCN, we performed the same analysis but replaced the anterior BFCN with the posterior BFCN. For the posterior BFCN, the association between BFCN-FC reductions within the DMN and within-DMN-FC was not significant for sevoflurane 3 vol%-induced unresponsiveness (R² = 0.01, p = 0.35). For propofol data, we did not find any overlapping brain regions between reductions of posterior BFCN-FC and within-DMN-FC (Fig. S1). This result suggests a specific association between anterior BFCN changes and DMN-FC changes.

4. Discussion

To study both changes of BFCN-FC during anesthesia-induced unresponsiveness and their associations to DMN-FC alterations, we re-analyzed fMRI data from awake healthy participants and during the administration of sevoflurane at different concentrations (2 and 3 vol%) or propofol. The disruption of DMN-FC has been related to states of altered consciousness (Vanhaudenhuyse et al., 2010; Demertzi et al., 2014), including anesthesia-induced unresponsiveness with sevoflurane and propofol (Palanca et al., 2017; Boveroux et al., 2010; Guldenmund et al., 2013; Golkowski et al., 2019), making DMN-FC decrease one of the most stable signs of anesthesia-induced unresponsiveness. In this exploratory pilot study, we found reduced anterior and posterior BFCN-FC during sevoflurane- and propofol-induced unresponsiveness. Furthermore, within-DMN-FC-reductions were associated with anterior BFCN-FC reductions for sevoflurane 3 vol% R² = 0.33 (p = 0.01), for sevoflurane 2 vol% and propofol was inconclusive (sevoflurane 2 vol%: R² = 0.11, p = 0.11, Fig. 4 A2; propofol: R² = 0.22, p = 0.06). To the best of our knowledge for the first time, we demonstrate, firstly, reduced BFCN-FC during states of propofol and sevoflurane-induced anesthesia; secondly, our findings hint at a potential link between decreased anterior BFCN-FC and decreased DMN-FC for the same states. In summary, the first result suggests decreased BFCN-FC during anesthesia-induced unresponsiveness. In contrast, the second suggests that the association between changes in DMN-FC and anterior BFCN-FC might reflect DMN-mediated top-down effects of anesthesia on BFCN connectivity.

4.1. Reduced BFCN-FC as a marker of anesthesia-induced unresponsiveness

In the present study, we demonstrated reduced FC of the basal forebrain cholinergic nuclei during propofol and sevoflurane anesthesia, the latter at different doses (Fig. 1). Our finding aligns with several studies demonstrating BOLD-FC changes, particularly in midline cortices, due to alterations in the cholinergic transmission (Peeters et al., 2020; Shah et al., 2015, 2016). First, it has been reported that direct manipulation of the cholinergic system in rodents results in BOLD-FC changes in large-scale functional networks including the DMN (Peeters et al., 2020; Shah et al., 2015, 2016). Similarly in humans, pharmacological manipulation of the cholinergic transmission was shown to affect BOLD-FC, including the DMN and the hippocampal network (Blautzik et al., 2016; Goveas et al., 2011; Griffanti et al., 2016; Klaassens et al., 2019; Sole-Padulles et al., 2009; Zaidel et al., 2012), whereby cerebral blood flow in regions associated with the medial cholinergic pathways seem particularly affected by such action (Li et al., 2012). Complementary, evidence from Alzheimer's Disease – a condition particularly affecting the cholinergic neurons (Mesulam et al., 1988; Muir, 1997; Schliebs et al., 2011) – suggests that damage to these very systems reduces BOLD-FC in large-scale functional networks (Damoiseaux et al., 2012; Hafkemeijer et al., 2017). Further supportive, a recent study investigating a computational model concluded – by manipulating the activity of cholinergic neurons – that modulation of the cholinergic system suppresses DMN-FC (Sanda et al., 2022). Therefore, we propose that the observed consistent BFCN-FC reduction likely reflects changes in cholinergic transmission.

We demonstrated reduced BFCN-FC for diverse anesthetic agents (propofol and sevoflurane) and different concentrations (sevoflurane at 2 and 3 vol%). Due to this consistency of decreased BFCN-FC across sevoflurane and propofol and different concentrations of sevoflurane, we conclude that BFCN-FC reduction accompanies anesthesia-induced unresponsiveness. However, the generalization of these findings might be problematic due to two main points. First, the nature of anesthetic agents might influence results: sevoflurane and propofol are GABAergic anesthetics, which implies that findings might not be the same with agents acting through other mechanisms (e.g., ketamine, dexmedetomidine, etomidate, thiopental, nitrous oxide, isoflurane, desflurane, among others). Second, unresponsiveness is not equal to unconsciousness: General anesthesia is a complex process and can elicit different experiences in each subject due to the individual nature of pharmacodynamical processes (Bonhomme et al., 2019). For instance, the goal of surgical anesthesia is ‘unconsciousness’ (i.e., a complete absence of a subjective experience); further states have been described, for instance, unresponsive anesthetized individuals may have conscious experience (e.g., dreams), or maybe unresponsive but perceive the surroundings (i.e., disconnected unconsciousness) (Bonhomme et al., 2019; Sanders et al., 2012). These premises suggest that unresponsiveness and unconsciousness are at least partly dissociable processes (Sanders et al., 2012). In our experiment, ‘unconsciousness’ was defined based on the lack of a behavioral response, which was strictly technically merely a demonstration of ‘unresponsiveness’. However, due to the complexity and the high inter-subject variability of experiences, the term anesthesia induced-unconsciousness has been classically used interchangeably with anesthesia-induced unresponsiveness throughout the published literature.

4.2. Explaining reduced BFCN-FC during anesthesia-induced unresponsiveness – correlated BFCN-FC and DMN-FC decreases

In the following sections, we will discuss potential non-exclusive mechanisms contributing to anesthesia-induced unresponsiveness-related BFCN-FC decreases. We will relate these mechanisms with the correlation of BFCN-FC decreases with DMN-FC decreases during anesthesia-induced unresponsiveness.

4.2.1. Top-down model

Anesthesia-induced unresponsiveness, for example, induced by propofol and sevoflurane anesthetics, is associated with aberrant cortical coherence, e.g., increased EEG-based coherence among frontal cortices at the alpha frequency range (7–12Hz) (Akeju et al., 2014; Cimenser et al., 2011; Purdon et al., 2013, 2015) or with BOLD-FC changes among large-scale brain networks, most consistently in the DMN (Palanca et al., 2015, 2017; Boveroux et al., 2010; Ranft et al., 2016; Jordan et al., 2013; Martuzzi et al., 2010). More specifically, several studies have demonstrated that propofol and sevoflurane at different concentrations alter static FC (i.e., the measure we have used in the current study) (Palanca et al., 2015, 2017; Boveroux et al., 2010; Ranft et al., 2016; Jordan et al., 2013; Martuzzi et al., 2010) as well as dynamic FC measures in the DMN (Golkowski et al., 2019; Hudetz et al., 2015; Barttfeld et al., 2015). This evidence suggests a direct effect of anesthesia on cortico-cortical coherence, compatible with a ‘top-down’ or ‘cortical network-level’ mechanism explaining subcortical changes induced by cortical changes during anesthesia-induced unresponsiveness (Cavanna et al., 2018; Dehaene et al., 2011; Koch et al., 2016; Tononi, 2012). Anatomically, the BFCN do not only project topographically to DMN areas, such as cingulate cortex and precuneus (Mesulam et al., 1983; Zaborszky et al., 2015; Bigl et al., 1982; Bloem et al., 2014; Jones et al., 1976; Pearson et al., 1983), mirrored by BOLD-FC in humans (Fritz et al., 2019; Markello et al., 2018; Yuan et al., 2019), but also receive direct projections from these areas for cortical control (Ananth et al., 2023; Do et al., 2016; Gielow et al., 2017; Hu et al., 2016; Mesulam et al., 1984; Zaborszky et al., 1991). Therefore, anesthesia-induced cortical effects might be reflected by BFCN-FC changes following the ‘top-down’ model paradigm. This model is supported by our finding of an association between DMN-FC decreases during anesthesia-induced unresponsiveness and BFCN decreases. In summary, BFCN-FC decreases might be induced by DMN-FC decreases during this state.

4.2.2. Bottom-up model

Besides exclusive cortical effects, anesthesia-induced effects on brainstem neuromodulatory nuclei, such as ventral tegmental area (VTA), raphe nuclei, and locus coeruleus, might also influence BFCN-FC. These nuclei are part of the brainstem arousal pathways which modulate arousal states by projecting directly to the cortex (Starzl et al., 1951; Parvizi et al., 2001; Edlow et al., 2012), thalamus, such as intralaminar thalamic and further upper brainstem nuclei (Parvizi et al., 2001; Van der Werf et al., 2002; Steriade et al., 1982), and BFCN (Parvizi et al., 2001; Jones et al., 1985; Jones, 2004). This basic idea can be traced back to classical experiments, demonstrating the general dependence of forebrain activity on the neuromodulatory brainstem input (Moruzzi et al., 1949; Bremer, 1935). Therefore, BFCN-FC changes might depend on three distinct pathways: (i) A direct bottom-up influence of brainstem neuromodulation on BFCN. For example, suppose anesthesia affected the VTA neurons whose axons project directly into the cholinergic basal forebrain neurons Field (Gaykema and Zaborszky, 1996). In that case, this might affect cholinergic transmission and, hence, BFCN-FC directly. (ii) An indirect brainstem influence on BFCN mediated by the cortex (see top-down model). For example, if (as above) anesthesia affected the VTA neurons whose axons project directly into cholinergic basal forebrain neurons (Gaykema et al., 1996), this might also affect VTA neurons projecting directly to prefrontal cortices (Morales et al., 2017; Oades et al., 1987) which further project to the BFCN, such that BFCN cholinergic transmission and hence BFCN-FC would be affected indirectly via this route. (iii) An indirect brainstem influence mediated by the thalamus. For example, suppose (again as above) anesthesia affected the VTA neurons whose axons project directly into the cholinergic basal forebrain neurons Field (Gaykema and Zaborszky, 1996). In that case, this might also affect those VTA neurons that project directly to the midline thalamic nuclei and from there further to the cholinergic BFCN, such as the paraventricular thalamus (Hu et al., 2016). This route would affect cholinergic transmission and, hence, BFCN-FC indirectly by the brainstem via the thalamus.

In summary, the effects of anesthesia on the arousal system might be operating through three distinct pathways, including direct bottom-up brainstem neuromodulation on BFCN, indirect top-down brainstem neuromodulation on BFCN mediated by the cortex, and indirect brainstem BFCN neuromodulation via the thalamus. All these pathways might – if neuromodulation of BFCN is affected – also influence BFCN-FC.

4.2.3. Direct effects on BFCN-FC

Beyond these potential top-down/bottom-up mechanisms, other neurobiological effects might directly contribute to the observed reduced BFCN-FC during anesthesia-induced unresponsiveness. Provided FC is a measure calculated from the BOLD signal sensitive to changes in the blood oxygenation level; both neural changes and/or changes in the vascular-hemodynamic system are potential candidates. We will briefly discuss potential mechanisms of action for those two systems in the following.

On a neuronal level, sevoflurane increases GABAergic activity while decreasing NMDA activity, affecting the neuronal transmission (Garcia et al., 2010; Petrenko et al., 2014). Propofol, in turn, directly influences the GABAergic activity (Tang et al., 2018). For example, propofol affects vesicle transport in distal axons in rat hippocampal cells (Frank et al., 2022). Based on these findings, reduced BFCN-FC might reflect the direct neuronal effects of sevoflurane and propofol. One should note that such effects are not specific to cholinergic neurons.

Due to effects on the vascular-hemodynamic system, both sevoflurane and propofol might impact BOLD-based FC. Sevoflurane potentially acts via increasing arterial stiffness, which ultimately decreases the systemic mean arterial pressure (Juhasz et al., 2019). Cerebral autoregulation and co-variation with CO2 seem unaffected (Juhasz et al., 2019; Kitaguchi et al., 1993). Similarly, propofol is reported to evoke systemic hemodynamic changes eventually decreasing cardiac output (Bilotta et al., 2001), which might be mediated by decreases in myocardial blood flow and oxygen consumption (Stephan et al., 1986); however, effects on cerebral blood flow are not well studied and are contradictory (Mikkelsen et al., 2016).

4.2.4. Clinical implications

Our results suggest that changes in functional connectivity patterns of BFCN could be linked to behavioral loss of responsiveness/consciousness in humans. Although the generality of these results needs to be investigated (i.e., analysis with further anesthetic drugs and/or pathological causes of loss of consciousness such as trauma brain injury), BFCN-FC could be a potential candidate for a neuroimaging marker of consciousness.

The interest in finding neuroimaging markers of disorders of consciousness including anesthesia-induced unconsciousness, has increased over the past decades, especially considering the availability of non-invasive neuroimaging methods such as fMRI (Calhoun et al., 2021; Zhang et al., 2020; Sanz et al., 2021). Recent publications have explored this topic; for instance, Hahn and colleagues included data from human deep sleep and propofol sedation in monkeys and demonstrated that large-scale synchronization of resting-state fMRI signals reduced during both unconscious states independent, suggesting this measure as a marker of state of consciousness (Hahn et al., 2021). Also, Luppi and colleagues including data from anesthetized volunteers with propofol and patients with disorders of consciousness showed that unconscious conditions were characterized by a reduction in temporal and spatial functional diversity (Luppi et al., 2019). Yet a neuroimaging marker linked to neurochemical modulation for human consciousness has not been thoroughly investigated. A recent publication (Spindler et al., 2021) showed interesting results related to the dopaminergic system, namely dopaminergic ventral tegmental area (VTA) FC reductions during propofol sedation and patients with disorders of consciousness. Notably, studies targeting the cholinergic system as mentioned above still need to be included.

5. Limitations

When interpreting our results, both general limitations regarding the reliability of the BOLD signal, as well as specific limitations regarding the current study population, must be considered.

Regarding general limitations, it first needs to be emphasized that the BOLD signal is only an indirect measurement of the neuronal signal: Neuronal activity evokes molecular and, ultimately, hemodynamic changes, which are co-observed with increased oxygenated blood flow (Ogawa et al., 1990; Buxton, 2013; Heeger et al., 2002). Experiments in primates have shown that BOLD responses correlate with neuronal-derived signals, such as local field potentials, an electrophysiological signal mainly reflecting neuronal information's input and local processing (Logothetis et al., 2001, 2004). Diverse experiments studying neuronal activity and concurrent hemo- and vasodynamic measurements have supported this association and have further validated its use for neuroimaging studies (Du et al., 2014; Goense et al., 2008, 2012; Schwalm et al., 2017; Shmuel et al., 2008), yet it is not possible to establish the origin of the neurons contributing to the signal; therefore statements that our findings describe only acetylcholine transmission or cholinergic neuronal activity have to be critically evaluated.

Furthermore, the spatiotemporal resolution of fMRI experiments limits the interpretability of our results. To precisely map biological processes, both high temporal and spatial resolution are desired. However, provided hemodynamics is the source of the BOLD contrast, spatial resolution is limited by microvascular rather than neuronal changes (Petridou et al., 2019). Temporal resolution is limited by a similar effect, namely the hemodynamic response time, whereby the BOLD response lags neuronal activity and peaks around 5–6 s after its onset (Buckner et al., 1996). Moreover, BFCN seeds are noted to have a relatively low signal-to-noise ratio due to their anatomical position. Attempts to optimize spatial and temporal resolution in fMRI protocols typically reduce the signal-to-noise ratio when aiming for whole-brain coverage, as we did in the present study (Seidel et al., 2020). Nevertheless, for 3T BOLD imaging, even high spatial resolution was reported to convey biologically accurate information (Iranpour et al., 2015). Based on that, our finding of reduced BFCN-FC during anesthesia-induced unresponsiveness is a physiologically valid deduction.

Besides these general limitations regarding the interpretability of BOLD imaging, additional specific caveats regarding our sample must be considered: Notice that we only studied healthy male subjects. This is problematic for two reasons: (1) The BOLD response in males is different as compared to females (Levin et al., 1998) which might explain both altered FC (David et al., 2018) and FC dynamics (Murray et al., 2021) between sexes. (2) Cholinergic activity of BFCN might be susceptible to hormone profiles (Kelley et al., 2014) and thus producing sex-specific BFCN-FC. Hence, an only-male sample reduces generalizability of our findings and, most likely, decreases also some critical elements of variability, which may have improved their ability to detect condition-based differences.

Moreover, the investigated study population was relatively small, such that our results must replicated in future studies of larger sample sizes. However, provided that despite such potential lack of power we could demonstrate reduced BFCN-FC during anesthesia-induced unresponsiveness for all investigated drugs and doses, this might even strengthen our results since it suggests a larger effect size (Serdar et al., 2021). It might furthermore follow that the relationship between reduced BFCN-FC and reduced DMN-FC is a smaller effect, provided we could only demonstrate a significant association for the sevoflurane 3 vol% condition, but only at-trend associations for sevoflurane 2%vol and propofol. However, it remains subject to future studies to disentangle a drug-/dose dependency of the relationship between reduced BFCN-FC and DMN-FC during anesthesia-induced unresponsiveness from a small effect size. Finally, due to the small sample size, more specific statistical models, such as meditation analyses, could not be performed. Such analyses could help establish a better relationship between the variables.

6. Conclusion

BFCN-FC is reduced during sevoflurane- and propofol-induced anesthesia, with a potential link between reduced anterior BFCN-FC and reduced DMN-FC. These findings suggest changes in the DMN as a potential factor of anterior BFCN-FC reductions during anesthesia-induced unresponsiveness and BFCN-FC reduction as a potential sign of such a state.

Data availability statement

Time courses from ROIs are available under request. Please get in touch with the corresponding author, Juliana Zimmermann (juliana.zimmermann@tum.de).

Funding statement

The experiments were funded by the Departments of Anesthesiology and Intensive Care, Neurology, and Neuroradiology of the Klinikum rechts der Isar of the Technical University of Munich. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflict of interest disclosure

Andreas Ranft received a research grant from the German Research Foundation (Deutsche Forschungsgemeinschaft), which is in no way related to the submitted work. The grant is for the randomized multicentre trial ‘ACT in Stroke’ (registered DRKS00023679 at: https://www.germanctr.de/drks_web)., additionally he has won a grant by ESAIC for a clinical observational study called ARCTIC-I (NCT04522856) also not related to the submitted work. ESAIC provides networking and data management support and reimburses travel expenses for its congresses. Marlene Tahedl received a Walter-Benjamin Postdoc Stipend from the Deutsche Forschungsgemeinschaft (DFG, TA, 1902/1-1) for a project not related to the present one. The remaining authors do not have any conflict of interest to declare. The remaining authors do not have any conflict of interest, either financial or non-financial competing interests, to declare.

CRediT authorship contribution statement

Juliana Zimmermann: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. Rachel Nuttall: Data curation, Formal analysis, Investigation, Writing – review & editing. Daniel Golkowski: Conceptualization, Data curation, Investigation, Methodology, Writing – review & editing. Gerhard Schneider: Conceptualization, Data curation, Methodology, Supervision, Writing – review & editing, Resources. Andreas Ranft: Conceptualization, Data curation, Investigation, Supervision, Writing – review & editing. Rüdiger Ilg: Conceptualization, Data curation, Investigation, Methodology, Writing – review & editing. Afra Wohlschlaeger: Conceptualization, Formal analysis, Methodology, Supervision, Writing – review & editing. Christian Sorg: Conceptualization, Methodology, Resources, Supervision, Writing – original draft, Writing – review & editing. Marlene Tahedl: Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the supplementary data to this article: