Effects of isoflurane anesthesia on resting-state fMRI signals and functional connectivity within primary somatosensory cortex of monkeys

Abstract

Introduction: Correlated low-frequency fluctuations of resting-state functional magnetic resonance imaging (rsfMRI) signals have been widely used for inferring intrinsic brain functional connectivity (FC). In animal studies, accurate estimate of anesthetic effects on rsfMRI signals is demanded for reliable interpretations of FC changes. We have previously shown that inter-regional FC can reliably delineate local millimeter-scale circuits within digit representations of primary somatosensory cortex (S1) subregions (areas 3a, 3b, and 1) in monkeys under isoflurane anesthesia. The goals of this study are to determine (1) the general effects of isoflurane on rsfMRI signals in the S1 circuit and (2) whether the effects are functional- and regional- dependent, by quantifying the relationships between isoflurane levels, power and inter-regional correlation coefficients in digit and face regions of distinct S1 subregions.

Methods: Functional MRI data were collected from male adult squirrel monkeys at three different isoflurane levels (1.25%, 0.875%, and 0.5%). All scans were acquired on a 9.4T magnet with a 3-cm-diameter surface transmit-receive coil centered over the S1 cortex. Power and seed-based inter-regional functional connectivity analyses were subsequently performed.

Results: As anesthesia level increased, we observed (1) diminishing amplitudes of signal fluctuations, (2) reduced power of fluctuations in the low-frequency band used for connectivity measurements, (3) decreased inter-voxel connectivity around seed regions, and (4) weakened inter-regional FC across all pairs of regions of interest (digit-to-digit). The low-frequency power measures derived from rsfMRI signals from control muscle regions, however, did not exhibit any isoflurane level-related changes. Within the isoflurane dosage range we tested, the inter-regional functional connectivity differences were still detectable, and the effects of isoflurane did not differ across region-of-interest (ROI) pairs.

Conclusion: Our data demonstrate that isoflurane induced similar dose-dependent suppressive effects on the power of rsfMRI signals and local fine-scale FC across functionally related but distinct S1 subregions.

Keywords: fMRI; hand; non‐human primates; resting state; somatosensory system.

Figure 1

Blood oxygenation level‐dependent (BOLD) resting‐state signal time series and power spectral density at three different anesthesia levels. (a) Time series of a voxel in the somatosensory region in a representative animal of the same session. The monkey was under three different isoflurane levels: 1.25%, 0.875%, and 0.5% plotted in black, blue, and red lines, respectively, with three runs at each level. (b) Power spectral density estimates with the Fast Fourier Transform for one of the time series at 1.25%, 0.875%, and 0.5% isoflurane levels

Figure 2

Relationship between the power of resting‐state fMRI blood oxygenation level‐dependent (BOLD) signals and level of anesthesia. (a) Representative fALFF maps of the somatosensory region under three different isoflurane levels. Color bar: range of fALFF values. (b) Group boxplots of the mean fALFF and ALFF in the somatosensory regions from two squirrel monkeys. Center lines of the boxplot represent the medians while the box limits present the 25th and 75th percentiles. Outliers are represented by the dots. Crosses indicate the means for 1.25% (n = 7 runs), 0.875% (n = 7), and 0.5% (n = 6) isoflurane levels, respectively. **p ≤ .005; *p ≤ .05 (nonparametric Mann–Whitney test) (c) Scatter plots of fALFF and ALFF measurements in the muscle control region in one monkey. Green dots indicate the mean values. a, anterior; l, lateral; m, medial; p, posterior

Figure 3

Effects of anesthesia level on fALFF and ALFF of different S1 subregions (areas 3a, 3b, and 1). Boxcar plots of group z‐scores of the fALFF (a–d) and ALFF (e–h) as a function of anesthesia level at specific regions of interest: area 1, area 3b, and area 3a digit regions and the area 3b face control region. Center lines of the boxplot represent the medians while the box limits present the 25th and 75th percentiles. Outliers are represented by the dots. Crosses indicate the mean for 1.25% (n = 7 runs), 0.875% (n = 7), and 0.5% (n = 6) isoflurane levels, respectively

Figure 4

Tactile stimuli evoked fMRI activation map and region of interest (ROI) seed‐based voxel‐wise correlation maps within S1 cortex. (a) Left: blood oxygenation level‐dependent (BOLD) activation map (thresholded at t > 2.7) to simultaneous 8 Hz vibrotactile stimulation of digits 1 & 3. Scale bar: range of t values. Color dots represent the overlaid microelectrode penetration sites with color indicating the receptive field of neuron on individual digits. Horizontal dotted blue lines: estimated inter‐area borders. Vertical dotted blue line: estimated hand‐face border. CS and LS: central and lateral sulci. Voxels with the strongest response (highest t values) were chosen in areas 3b and 1 as seeds for rsFC analysis indicated by squared blue boxes. Right: aligned original electrophysiology map (brain surface blood vessel map). (b–d) Effects of isoflurane levels on the voxel‐wise correlation maps (thresholded at r > .75) of three ROI seeds (area 3b: b, area 1: c, face control: d) in one representative monkey. The yellow voxels indicate the locations of seeds. Scale bars indicate the range of r values

Figure 5

Group quantification of anesthesia effects on inter‐regional functional connectivity within S1 cortex. Boxplots of correlation strengths (r values) of four region‐of‐interest (ROI)‐pairs (from left to right: area 3b digit and area 3a digit; area 1 digit and area 3a digit; area 3b digit and area 1 digit; area 1 digit and face control) at three isoflurane levels. **p ≤ .005; *p ≤ .05 (nonparametric Mann–Whitney test). The red cross represents an outlier data point. Regardless of ROI pairs, anesthesia effects (at three different levels) were significant p < .0001 (two‐way ANOVA test)