Functional Connectivity Changes in Primary Motor Cortex Subregions of Patients With Obstructive Sleep Apnea

Abstract

Background and purpose: Obstructive sleep apnea (OSA) is linked to cognitive impairment and altered motor-related brain networks. This study examined functional connectivity (FC) changes in subregions of the primary motor cortex (M1) in patients with OSA and their association with sleep structure, cognition, and clinical features.

Methods: Sixty-five patients with OSA and 65 healthy controls (HC) participants matched in age and educational background were included. Resting-state functional MRI data were acquired for all participants using a 3T MRI system. Based on the Human Brainnetome Atlas, we analyzed FC changes of 12 subregions of M1 across the whole brain in patients with OSA. The two-sample t-tests were conducted to compare FC values between subregions of M1 and other brain regions in two groups. Partial correlation analyses examined the association between FC and clinical variables in patients with OSA. Additionally, we employed three machine learning algorithms-support vector machine (SVM), random forest (RF), and logistic regression (LR)-to distinguish patients with OSA from HC based on FC features.

Results: Compared to HC, the OSA group found that significant FC enhancements were identified in right A6cdl with the left inferior parietal lobule (IPL); left A4tl with the left inferior frontal gyrus (IFG), bilateral middle frontal gyrus (MFG), and left IPL; and left A6cvl with the right parahippocampal gyrus, bilateral MFG, left IFG, left superior temporal gyrus, and right cingulate gyrus. After Bonferroni correction, a negative correlation was observed between the FC value of A4tl (L)-IPL (L) and N2 (p < 0.05). Furthermore, SVM yielded the highest area under the receiver operating characteristic (ROC) curve (AUC) among all classifiers, indicating its superior performance in discriminating OSA patients from HC based on FC features.

Conclusion: The study demonstrates that OSA significantly impacts brain functional networks, particularly affecting motor control through altered FC in subregions of M1. These alterations correlate with upper airway dysfunction and cognitive impairments, increasing accident risks. The high-accuracy SVM classification based on FC patterns demonstrates potential as a diagnostic biomarker for OSA. Future research should explore M1 FC patterns as diagnostic markers and neuromodulation therapies.

Keywords: functional connectivity; obstructive sleep apnea; primary motor cortex; subregion; support vector machine.

FIGURE 1

Flowchart illustrating the main steps of data processing and analysis for obstructive sleep apnea patients and healthy controls. HC, healthy controls; OSA, obstructive sleep apnea; ROC, receiver operating characteristic.

FIGURE 2

The 12 primary motor cortex subregions, including bilateral A4hf, A6cdl, A4ul, A4t, A4tl, and A6cvl.

FIGURE 3

Altered FC between right A6cdl (ROI4) of primary motor cortex subregions and the whole brain in patients with OSA and HC (two‐tailed GRF correction, voxel‐level [p < 0.001], and cluster‐level [p < 0.05]). ROI, region of interest.

FIGURE 4

Altered FC between left A4tl (ROI9) of primary motor cortex subregions and the whole brain in patients with OSA and HC (two‐tailed GRF correction, voxel‐level [p < 0.001], and cluster‐level [p < 0.05]). ROI, region of interest.

FIGURE 5

Altered FC between left A6cvl (ROI11) of primary motor cortex subregions and the whole brain in patients with OSA and HC(two‐tailed GRF correction, voxel‐level [p < 0.001], and cluster‐level [p < 0.05]). ROI, region of interest.

FIGURE 6

The histogram indicates the mean value of FC between the two groups. Detailed information about these altered regions is provided in Table 3. Values are presented as the mean ± standard error. CG, cingulate gyrus; FC, functional connectivity; IFG, inferior frontal gyrus; IPL, inferior parietal lobule; L, left; MFG, middle frontal gyrus; PG, parahippocampa gyrus; R, right; STG, superior temporal gyrus.

FIGURE 7

The correlations between the altered FC values and clinical variables in abnormal brain regions among patients with OSA. In patients with OSA, there was a significant correlation between FC values with intergroup differences (patients with OSA vs. HCs) and clinical assessments. * and ** denote existing statistical difference before and after Bonferroni correction. AHI, apnea‐hypopnea index; C1‐4, cluster 1‐4; FC, functional connectivity; L, left; MoCA, Montreal Cognitive Assessment; N1/2, sleep stage 1/2; ODI, oxygen desaturation index; OSA, obstructive sleep apnea; R, right; ROI, region of interest.

FIGURE 8

ROC curves for classifying OSA and HC using static FC features with SVM, RF, and LR models. SVM showed the highest AUC (0.84), followed by LR (0.82) and RF (0.77). AUC, area under the ROC curve; LR, logistic regression; RF, random forest; ROC, receiver operating characteristic; sFC, static functional connectivity; SVM, support vector machine.