Resting-state functional magnetic resonance imaging of high altitude patients with obstructive sleep apnoea hypopnoea syndrome

The objective of the study was to observe brain function changes in Obstructive Sleep Apnoea Hypopnoea Syndrome (OSAHS) patients at high altitude. Resting-state functional magnetic resonance imaging (rs-fMRI) in patients with OSAHS was assessed using regional homogeneity (ReHo), amplitude of low frequency fluctuation (ALFF) and functional connectivity (FC). In this study, 36 male patients with OSAHS and 38 healthy male subjects were recruited from high-altitude areas, specifically, altitudes of 2,000–3,000 m. OSAHS was diagnosed by polysomnography (PSG). The blood oxygen level-dependent (BOLD) signals of OSAHS patients and healthy controls in the resting state were obtained and compared using ReHo, ALFF and FC methods. The posterior cingulate cortex (PCC) was selected as the seed region in the comparison of FC between the two groups. Compared with the healthy control group, multiple brain functions in the OSAHS patient group were different. There were correlations between the brain function values of some brain regions and demographic data. We also found that in contrast to earlier findings with individuals in plains areas, the brain function at the frontal lobe and the precuneus were higher in OSAHS patients, and the PCC showed higher FC with the left caudate, which may be due to the high-altitude hypoxic environment.


ReHo.
Compared with the healthy control group, the OSAHS patient group had higher ReHo values in the left superior frontal gyrus, right anterior cingulate, left parahippocampus, right postcentral gyrus, right hippocampus and right precuneus. Compared with the healthy control group, the OSAHS patient group had decreased ReHo values in the left cuneus and the left precuneus. A cluster was defined as a block of continuously connected voxels containing more than 22 voxels as the threshold value (Table 2 and Fig. 1).

ALFF.
Compared with the healthy control group, the OSAHS patient group had higher ALFF values in the right middle cingulate, left medial superior frontal gyrus, right anterior cingulate, right hippocampus and left parahippocampus. Compared with the healthy control group, the OSAHS patient group had decreased ALFF values in the right calcarine, right inferior occipital gyrus, right middle occipital gyrus, left calcarine and right cerebellum_7b. A cluster was defined as a block of continuously connected voxels containing more than 22 voxels as the threshold value (Table 3 and Fig. 2). Correlations between functional brain measures and AHI and MSaO2. The OSAHS patients and controls had significantly different AHI and MSaO2 levels. In the brain region where ReHo values were higher, ReHo values in the left superior frontal gyrus appear to be negatively related to AHI values (P = 0.001, r = − 0.517, Fig. 4). In another brain region where ReHo values were higher, ReHo values in the right postcentral gyrus appear to be positively related to AHI values (P = 0.020, r = 0.386, Fig. 5). In the brain region where ALFF values were decreased, ALFF values in the right middle occipital gyrus appear to be negatively related to MSaO2 levels (P = 0.040, r = − 0.344, Fig. 6).  As mentioned earlier, BOLD reflects first blood oxygenation and secondly brain functions linked to cellular metabolism. Chronic hypoxia-related changes caused by combination of OSAHS and high altitude and explored by BOLD techniques may be related to: (1) Anatomical changes due to "long time scale" modifications of the brain connectivity and of local brain functioning, caused by chronic brain oxygenation deprivation, and able to modify the basal activity of the "default mode".
(2) vascular and metabolic changes due to the "short timescale" adaptation of the vascular functions to hypoxic perturbations, like changes in blood properties (like MSaO 2 ), cerebral vasculature adaptation, decreased (or increased by compensation) local metabolism, etc.
For the first case, The Change in brain anatomy may affect the cognitive function of patients. Dosenbach et al. 16 found that the anterior prefrontal cortex, anterior insula/frontal operculum, basal ganglia and thalamus are part of the cingulo-opercular network (CON), and these regions may be involved in activation, maintenance, and monitoring of task execution. It can be inferred that if a brain region that is part of the CON network is abnormal, it may affect task execution function in OSAHS patients. Surveys conducted by Santarnecchi et al. 17 indicated that the intrinsic connections between the central anterior gyrus and central posterior gyrus were enhanced in OSAHS patients, and the authors suggested that this enhancement may be related to higher activity of muscle tissue associated with sleep. In the present study, ReHo values in the right postcentral gyrus were  www.nature.com/scientificreports/ higher, indicating that the function of this brain region was changed, which may affect the activity of related muscle tissue and cause sleep disturbance. In this study, brain activity in various regions of the frontal lobe were higher in the OSAHS patients. Thomas et al. 18 applied a 2-back working memory task to study patients with obstructive sleep-disordered breathing. Their results showed that the dorsolateral prefrontal lobe was always in a negative activation state, and the success rate and response time of patients were not significantly improved  . Correlation between brain regions and AHI. In the brain region where ReHo values were higher, ReHo values in the left superior frontal gyrus appear to be negatively related to AHI values (P = 0.001, r = − 0.517). ReHo regional homogeneity, AHI apnoea-hypopnoea index. www.nature.com/scientificreports/ Figure 5. Correlation between brain regions and AHI. In the brain region where ReHo values were higher, ReHo values in the right postcentral gyrus appear to be positively related to AHI values (P = 0.020, r = 0.386).
ReHo regional homogeneity, AHI apnoea-hypopnoea index. www.nature.com/scientificreports/ after treatment with positive airway pressure. This may explain why working memory has been associated with decreased dorsolateral prefrontal function. Therefore, it can be inferred that the higher brain activity in various regions of the frontal lobe may be related to functional abnormalities such as memory loss. Li et al. 19 suggested that the activation of the cingulate gyrus was related to the brain network that controls breathing and the function of the autonomic nervous system. Some researchers 20,21 have found in animal experiments that intermittent hypoxia and sleep fragmentation may lead to neuron loss in the hippocampus, and some researchers 22 suggested that the hippocampus in OSAHS patients may atrophy, which indicated that the changes in hippocampal brain function in this study may have been due to hypoxia and sleep disorders in patients. The occipital lobe is the visual cortical centre. Some researchers 23 have suggested that breathing disorders in sleep may promote the occurrence of normal-tension glaucoma (NTG), and the incidences of visual field defects and glaucoma are higher in OSAHS patients. One of the surprising findings of this study was that in contrast to earlier findings 6,24 , ReHo and ALFF values in the frontal lobe and the precuneus were higher in the OSAHS patients, and the PCC showed higher FC with the left caudate. In addition, we found no regions where functional connectivity with the PCC was significantly decreased. These findings deserve our attention.
Concerning the second case, mentioned field inhomogeneities, which are created by the blood vessel network in the brain, are responsible for the intra-or extra-vascular BOLD contrast 25 . BOLD signals are affected by intra-or extravascular factors. The transformation between paramagnetic deoxyhaemoglobin and diamagnetic oxyhemoglobin affects the magnetic susceptibility of blood, resulting in spatial variation in the water Larmor frequency, which causes an intravascular frequency shift (IFS), ESD is caused by the static magnetic field changes appearing outside large and small vessels. IDA is caused by the changes in blood water T2 in the intravascular compartment. EDD is caused by tissue water diffusing in the internal magnetic field gradients (MFGs), which are created by the vascular system 26 .
The Qinghai-Tibet Plateau is the highest and largest plateau in the world, and it is also known as the "Roof of the World". A series of pathophysiological changes in the human body are caused by the special geographical environment in the plateau area, such as low oxygen pressure, cold temperatures, enhanced ultraviolet radiation and large temperature differences between day and night. High altitude can damage brain blood vessels, and damaged cerebrovascular reactivity can interfere with the transportation of oxygen 27,28 .
Hypoxic hypoxia (decreased partial pressure in dioxygen) is the essence and starting link of pathophysiological changes and may be caused both by high altitude and OSAHS. IFS is related to vessel radii and orientation, so that the vasoconstriction, which is caused by hypoxia at high altitude may decrease the deoxyhaemoglobin fraction in blood vessels and may indirectly affect the BOLD signal. In our study, brain function values of the patient and healthy control groups may have all decreased, but the magnitude of the decrease in values was different, so that the brain function values in some brain regions in the OSAHS patients showed a relatively high state. IDA is related to the content of red blood cells, so erythrocytopenia caused by high altitude may also contribute to changes in brain function measures 26 .
The OSAHS patients and controls had significantly different AHI and MSaO2 levels. In the brain regions where ReHo values were higher, ReHo values in the right postcentral gyrus appeared to be positively related to AHI values, and in the brain region where ALFF values were decreased, ALFF values in the right middle occipital gyrus appeared to be negatively related to MSaO2 levels. These results may indicate that the higher the oxygen content is, the smaller the change in brain function. However, in the brain region where ReHo values were higher, ReHo values in the left superior frontal gyrus appeared to be negatively related to AHI values; the mechanism underlying this complication is currently unknown.
Further exploration of the "short" and "long" timescale determinism of the measure of hypoxic cerebral changes with BOLD may shed light on this aspect.
In conclusion, in high altitude, we found that the brain function of OSAHS patients compared to controls were changed, especially in the right middle cingulate, left medial superior frontal gyrus, right anterior cingulate, right hippocampus and left parahippocampal regions. These changes in brain function in OSAHS patients at high altitude are different from those previously found in OSAHS patients in plains areas. These factors can make patients living at high altitude more aware of the importance of treatment, especially patients at high altitude. To our knowledge, hypoxia can lead to vasoconstriction, and the main reason is the dysfunction of endothelial cells during hypoxia 29 . For patients with OSAHS at high altitude, owing to the more severe hypoxia, the treatment may be of different from that in low altitude areas. In addition to continuous positive airway pressure, cerebral blood flow may be higher by protecting vascular endothelial cells. Some measures to prevent and treat encephalopathy at high altitude may also be applicable to prevent brain changes in OSAHS patients, such as raising the head 30∘ in the supine position 30 .
This research has limitations. Owing to the lack of a reliable cognitive assessment scale, we lack complete clinical information, such as a neuropsychological evaluation. Due to limited quality of non-MRI data, we did not correlate between MRI-derived brain function and EEG or blood, nasal or oral pressure data. However, from the perspective of imaging alone, at least the findings may increase the attention of OSAHS patients at high altitude. Further studies on this basis requires also to expand the number of samples to increase the statistical power of our results.

conclusions
The changes in some brain functions in OSAHS patients at high altitude are different from those in plains areas. These may be due to the high altitude hypoxia environment and attention should be paid to monitor OSAHS patients at high altitude, owing to potentially unpredictable and more severe neurological and sleep changes Scientific RepoRtS | (2020) 10:15546 | https://doi.org/10.1038/s41598-020-72339-2 www.nature.com/scientificreports/ than in plain areas. Further studies focused on the effect of altitude in OSAHS should explore the consequences specific functional changes described in this study.

Methods
Subjects. Thirty-six males from high altitude regions (2,000-3,000 m) who were diagnosed with OSAHS at Qinghai University Affiliated Hospital were enrolled in this study. Considering that hypertension may have an impact on brain function, we collected patients with blood pressure below 130/80 mm/Hg. The patient inclusion criteria were as follows: Han Chinese individuals (different ethnic groups, such as Tibetans, have different lifestyles and may have different adaptability to anoxic environments. Therefore, they may have different brain structures and functional impairments resulting from OSAHS 31 ); newly diagnosed patients without any treatment prior to diagnosis; individuals with no serious cardiovascular or cerebrovascular diseases (e.g., severe myocardial infarction, arrhythmia, stroke); right-handed individuals. The patient exclusion criteria were as following: sleep disorder other than OSAHS; mental or neurological disease; alcohol or history of taking psychotropic substances; and magnetic resonance imaging contraindications. The control group recruited 38 healthy male volunteers who matched the OSAHS group on altitude, age, and education level. The subjects in the control group were all right-handed and Han Chinese. In the control group, subjects had no intracranial disease, physical examination results were normal as screened by professional physicians, no clinical symptoms of OSAHS, and the normal indicators reported through polysomnography (PSG) monitoring. The altitude of the city where the scanner was located is 2,295 m. All patients and healthy controls come from high altitude areas (2000-3,000 m) and are native. ReHo and ALFF analysis. Acquired images were separated and processed to obtain resting-state series.

PSG. PSG (
Data were processed by SPM8 (Statistical Parametric Mapping) software (https://www.fil.ion.ucl.ac.uk./spm8/) based on the MATLAB 2010 b (MathWorks, Natick, Massachusetts) platform. The first 10 time points in the rs-fMRI series were discarded, and we preserved the data from 140 time points in the time series. Data with head motion greater than 1.5 mm in x, y, z directions or 1.5 degrees were excluded. The data for each individual were normalized to Montreal Neurological Institute (MNI) space. Images were resampled to 3 mm 3 resolution and smoothed with a 8 mm 3 full-width at half-maximum (FWHM) Gaussian kernel. The filter (0.01 < f < 0.08 Hz) was used to reduce the effects of low frequency drift and high frequency physiological noise. The preprocessed data were de-linearly drifted and filtered using the ALFF tool in REST 1.8 software (https ://www.restf mri.net/ forum /REST_V1.8). ALFF values were obtained after the power spectrum of the signal at 0.01 to 0.08 Hz was squared. After normalization, standardized ALFF values were obtained. The Resting-State fMRI Data Analysis Toolkit (https ://www.restf mri.net) was used to obtain Kendall's coefficient concordance (KCC), which can reflect synchronous activity in the brain. ReHo values for each subject were obtained by evaluating the KCC between each voxel in the brain and the nearest 26 voxel time series around it 13 . The computing formula is In this formula, W is the KCC among the given voxels and ranged from 0 to 1. Ri refers to the sum rank of the ith time point. R = [(n + 1)K]/2 means the average value of Ri. n refers to the number of ranks; here, n = 140. K refers to the number of time series within a measured cluster; here, K = 27. FC analysis. The preprocessed data were de-linearly drifted by REST software, and bandpass filtering (0.01-0.08 Hz) was performed for each time series. Generally, the seed point correlation analysis method selects the PCC as the seed point 32 . Then, we analysed the correlations between the PCC and the whole brain voxel time series, calculated the correlation coefficient r, and converted r into a Z value conforming to the normal distribution by Fisher Z transformation. Finally, a single-sample t-test was performed on the OSAHS group and the control group using REST software, and then a two-sample t-test was performed. REST software was used to render functionally connected abnormal brain regions. All results were corrected using AlphaSim for multiple W = ( (R i ) 2 − n( R) 2 ) × 12k 2 (n 3 − n).