Association between visceral fat area and heart rate variability in high altitude migrants with OSA: the mediating effect of insulin resistance

Abstract

Objective

This study was conducted to investigate the association between visceral fat area (VFA) and heart rate variability (HRV) in high-altitude migrants with obstructive sleep apnea (OSA).

Method

We conducted a cross-sectional study comprising 152 OSA participants from 2022 to 2024. We employed multivariable linear regression to further elucidate the association between VFA and HRV. Mediation analysis was utilized to investigate the indirect effects of insulin resistance and white blood cell counts on this relationship. Sensitivity analysis assessed the robustness of the results. To explore the influence of gender on the results, we conducted gender-specific subgroup analyses and interaction tests.

Results

Multivariable regression analysis revealed that for every 20-unit increase in VFA, there were significant reductions in SDNN (-4.00, 95% CI: -5.90, -2.11), SDANN (-3.51, 95% CI: -5.43, -1.60), SDNN index (-1.35, 95% CI: -2.13, -0.56), rMSSD (-0.92, 95% CI: -1.51, -0.33), and pNN50 (-0.58, 95% CI: -0.99, -0.17). Additionally, a significant positive association was found between VFA and the low-frequency/high-frequency ratio (LF/HF ratio) (0.25, 95% CI: 0.11, 0.39). Mediation analysis indicated a significant mediating effect of HOMA-IR on the VFA-HRV association, with proportions of 20.32%, 18.76%, and 26.23% for SDNN index, rMSSD, and LF/HF ratio, respectively. The mediating effect based on white blood cell count did not reach statistical significance. There was no gender difference in the association between VFA and HRV. The sensitivity analysis indicated that the findings remained robust.

Conclusion

Our findings indicated that visceral fat serves as a significant determinant of cardiovascular health among OSA patients residing at high altitudes and may represent a viable target for intervention and preventive strategies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12944-025-02626-2.

Introduction

Obstructive sleep apnea (OSA) is characterized by recurrent upper airway collapse and hypopnea during sleep, which can lead to cyclic intermittent hypoxia, hypercapnia, and disturbed sleep structure [1]. This unique pathophysiological process has been confirmed to cause an increased risk of diabetes [2], dyslipidemia [3], and metabolic syndrome [4] through multiple mechanisms such as activating the hypothalamic-pituitary-adrenal axis, insulin resistance, oxidative stress, and inflammatory response. Furthermore, OSA is not merely a pivotal risk factor for cardiovascular diseases (CVD), but also significantly associated with poor clinical prognosis [5–8]. At high altitudes, symptoms of OSA may be further aggravated by diminished atmospheric pressure and lower oxygen concentrations [9]. A study conducted in Peru confirmed that a significantly elevated prevalence of sleep apnea compared to lowlanders, accompanied by a higher Apnea-Hypopnea Index (AHI) [10]. Therefore, this unique hypoxic pattern, characterized by the combined effects of intermittent and sustained hypoxia, might pose a threat to the cardiovascular health of OSA patients residing at high altitudes.

Heart rate variability (HRV), defined as the fluctuations in the time intervals between consecutive heartbeats, is widely acknowledged as a sensitive indicator of autonomic nervous system activity. Autonomic nervous system dysfunction is intricately associated with various cardiovascular diseases, including heart failure, sudden cardiac death, and stroke, and serves as a pivotal indicator of adverse clinical prognosis [11–15]. It has been reported that patients with OSA have increased sympathetic nerve activity, decreased parasympathetic nerve activity, and reduced HRV [16]. However, the HRV of OSA patients residing at high altitudes may be further influenced by hypobaric hypoxia. As a metabolically active endocrine organ, visceral fat could directly influence autonomic nervous regulation by secreting adipokines and pro-inflammatory mediators [17]. Numerous studies further revealed that VFA has a stronger association with cardiovascular diseases than other obesity indicators [18–20]. Previous studies have confirmed that visceral fat serves as a critical risk factor for HRV [21–23]. However, in the high-altitude hypoxic environment, the interaction between visceral fat deposition and autonomic nerve dysfunction in patients with OSA remains unclear. Therefore, we conducted a retrospective study at high altitudes to investigate the relationship between the visceral fat area (VFA) and HRV in OSA patients at high altitudes.

Method

Participants

The study was conducted from 2022 to 2024 at the Plateau Acclimation and de-acclimation and Health Promotion Research Center. Initially, Han residents with OSA who had migrated to Lhasa and resided there for a duration exceeding one year were included (N = 182). Subsequently, individuals with incomplete dynamic electrocardiogram and bio-impedance analysis were excluded from the study (N = 178). Third, participants taking medications that influence autonomic nervous function, including beta-blockers, were also excluded (N = 172). Furthermore, individuals diagnosed with diabetes mellitus, coronary heart disease, atrial fibrillation, heart failure (LVEF < 50%), and thyroid dysfunction were excluded (N = 164). Finally, subjects with incomplete covariable data were further excluded from the analysis (N = 152). Ultimately, a total of 152 participants were included in the final analysis (Figure S1). This study received approval from the Biomedical Research Ethics Committee and was conducted in accordance with the principles of the Declaration of Helsinki. As the study was retrospective, informed consent was waived.

Measurement of visceral fat area

In this study, a body composition analyzer (InBody Co. Ltd., 270) was used to measure participants’ visceral fat area. The operating principle of the analyzer was based on Bioelectrical Impedance Analysis. During the test, participants were instructed to fast and wear light clothing, removing footwear and socks. The operator placed contact electrodes on the participants’ limbs and entered basic information such as height, weight, and gender. After the measurement, the instrument reported the visceral fat area directly.

Sleep respiratory monitoring

In our study, sleep breathing monitoring was performed using the sleep breathing screening instrument (Contec, RS01). AHI is the number of apnea plus hypopnea occurrences per hour of sleep on average. The diagnostic criteria were as follows: AHI < 5 (normal), 5 ≤ AHI < 15 (mild OSA), 15 ≤ AHI < 30 (moderate OSA), and ≥ 30 events /h (severe OSA).

Dynamic electrocardiogram

All subjects were monitored by a wearable dynamic electrocardiograph (Synwing Tech., 401). Cardiac autonomic nerve function was assessed using time domain analysis and frequency domain analysis. The time domain analysis of heart rate variability mainly includes the following indicators: SDNN refers to the standard deviation of all adjacent RR intervals over a certain period under normal sinus rhythm; SDANN is the standard deviation of the mean of all sinus RR intervals within 5 min. The SDNN index is the mean of the interval standard deviation of sinus RR every 5 min over 24 h. rMSSD is the root mean square of the difference between adjacent sinus RR intervals over a given time; pNN50 represents the percentage of total sinus beats with an interval difference between adjacent RR ≥ 50 ms. The decrease in the above indicators suggests an increase in sympathetic nervous activity and a decrease in parasympathetic nervous activity. Conversely, an increase in these indicators signifies a potentiation of parasympathetic activity alongside a reduction in sympathetic activity. Frequency domain analysis is evaluated by the low-frequency/high-frequency ratio (LF/HF ratio), which serves as an indicator of the balance between sympathetic and parasympathetic nerve activity.

Blood tests

Clinical data, including age, gender, body mass index (BMI), comorbidities, and medication history, were obtained from the electronic medical records of the Lhasa Health Promotion Center. Complete blood count analysis was performed using an automated blood analyzer (Dymind, DF55). Biochemical indicators were measured using an automated biochemical analyzer (Beckman, AU5800). Plasma insulin concentration was measured using an automated chemiluminescent immunoassay analyzer (Mindray, CL6000i).

Statistical analysis

Continuous variables that adhered to a normal distribution were presented as mean (standard error), whereas continuous variables with a non-normal distribution were reported as median (interquartile ranges). Categorical variables were described by frequency and percentage. To compare normally distributed continuous variables, analysis of variance (ANOVA) was employed, while the non-parametric Kruskal-Wallis test was utilized for the analysis of non-normally distributed data. The Chi-squared test was applied to assess the frequency distribution of categorical variables We used the VIM package to explore and visualize the structure of missing values in the data. Additionally, multivariable linear regression analysis was conducted to evaluate the relationship between visceral fat area (VFA) and HRV, as measured by SDNN, SDANN, SDNN index, rMSSD, pNN50, and LF/HF ratio. The results are reported as β coefficients with 95% CIs. Since the study included six dependent variables, we implemented a Bonferroni correction to mitigate the risk of type I error. All p-values were adjusted by the Bonferroni test, and p< 0.05/6 outcomes were considered statistically significant. The covariables in our study included age, gender, physical activity (never exercise, irregular exercise, regular exercise (cumulative > 3 months/year, and more than 3 times per week, and more than 150 min per week)), smoking status (never, former, current), drinking status (never, former, current), white blood cell, pulse pressure, LDL-cholesterol. Mediation analysis was performed to investigate the mediating effect of insulin resistance and inflammation level (white blood cell counts) on the relationship between VFA and HRV. Sensitivity analysis was conducted to further adjust for the apnea-hypopnea index (AHI) and blood oxygen saturation levels. Moreover, we performed data imputation for missing values using the random forest algorithm and conducted a sensitivity analysis to assess the robustness of our results. We stratified by gender and conducted subgroup analyses and interaction tests to compare the differences between genders. Sample size estimation was performed using G*Power 3.1. We targeted a statistical power (1-β) of 80%, and the effect size (f²) was specified as 0.15. For the linear multiple regression analysis in G*Power, we designated two predictors. Ultimately, the minimum sample size calculated was 101 cases. All analyses were executed using R version 4.2.1.

Results

The baseline characteristics of 152 patients with OSA were shown in Table 1. We conducted Little’s test to examine whether the data were completely randomly missing (p = 0.186), and visualized the missing values using the VIM package (Figure S2). In this study, the mean age of participants was 38.31 ± 8.03 years, and 57.6% of them were male. By categorizing VFA into quantiles, we observed a progressive increase in the proportion of participants who smoked and consumed alcohol, corresponding to higher VFA levels. Regarding metabolic parameters, increased VFA was significantly associated with gradual elevations in diastolic blood pressure, systolic blood pressure, low-density lipoprotein, BMI, and insulin resistance. Furthermore, the severity of sleep apnea in OSA patients also worsened as VFA levels increased. Notably, HRV gradually decreased as VFA increased.

Table 1.

Baseline characteristics

	Level	Overall	Q1	Q2	Q3	p
N		152	51	50	51
Age		38.31 (8.03)	35.63 (6.90)	38.72 (8.74)	40.59 (7.72)	0.006
Gender (%)	Male	88 (57.89)	15 (29.41)	25 (50.00)	48 (94.12)	< 0.001
	Female	64 (42.11)	36 (70.59)	25 (50.00)	3 ( 5.88)
Physical activity (%)	Never	12 ( 7.89)	5 ( 9.80)	5 (10.00)	2 ( 3.92)	0.086
	Irregular	89 (58.55)	26 (50.98)	25 (50.00)	38 (74.51)
	Regular	51 (33.55)	20 (39.22)	20 (40.00)	11 (21.57)
Smoking status (%)	Never	117 (76.97)	47 (92.16)	43 (86.00)	27 (52.94)	< 0.001
	Former	12 ( 7.89)	2 ( 3.92)	2 ( 4.00)	8 (15.69)
	Current	23 (15.13)	2 ( 3.92)	5 (10.00)	16 (31.37)
Drinking status (%)	Never	76 (50.00)	32 (62.75)	30 (60.00)	14 (27.45)	0.003
	Former	9 ( 5.92)	2 ( 3.92)	2 ( 4.00)	5 ( 9.80)
	Current	67 (44.08)	17 (33.33)	18 (36.00)	32 (62.75)
BMI (kg/m2)		22.78 (2.88)	20.27 (1.64)	22.76 (2.02)	25.33 (2.31)	< 0.001
White blood cell		5.66 (1.56)	5.28 (1.63)	5.75 (1.47)	5.96 (1.53)	0.082
Systolic blood pressure (mmHg)		119.55 (17.54)	110.65 (13.97)	119.68 (17.43)	128.31 (16.63)	< 0.001
Diastolic blood pressure (mmHg)		81.66 (13.03)	74.88 (10.05)	81.32 (11.10)	88.78 (13.87)	< 0.001
LDL-cholesterol (mmol/L)		2.78 (0.75)	2.54 (0.61)	2.77 (0.68)	3.03 (0.86)	0.004
Saturation of peripheral oxygen		93.00 [91.00, 94.00]	93.00 [91.00, 95.00]	93.00 [90.25, 94.00]	93.00 [91.00, 94.50]	0.822
HOMA-IR		1.20 [0.79, 1.74]	0.86 [0.69, 1.46]	1.21 [0.76, 1.71]	1.47 [1.09, 2.29]	< 0.001
SDNN (ms)		131.19 (32.89)	141.77 (35.73)	133.43 (31.50)	118.40 (27.04)	0.001
SDANN (ms)		119.97 (32.24)	128.65 (34.32)	122.71 (33.03)	108.61 (25.99)	0.005
SDNN_Index (ms)		53.77 (13.47)	57.32 (13.43)	54.75 (12.97)	49.25 (12.98)	0.008
RMSSD (ms)		29.51 (9.77)	32.85 (9.43)	30.47 (10.08)	25.23 (8.29)	< 0.001
PNN50 (ms)		9.36 (6.73)	11.59 (7.14)	9.98 (6.94)	6.51 (5.01)	< 0.001
LF/HF ratio		3.61 (2.41)	2.51 (1.05)	3.27 (1.74)	5.04 (3.16)	< 0.001
Apnea-hypopnea index		22.10 (14.68)	19.32 (13.86)	19.96 (14.19)	26.96 (14.98)	0.013

Continuous variables: Normal distribution were presented as mean (standard error), non-normal distribution were reported as median [interquartile ranges]; Categorical variables: n (percentage); HRV, Heart rate variability; SDNN, SD of all normal to normal intervals; SDANN, SD deviation of sequential 5-min normal to normal interval; SDNN index, the mean value of the SD of all the normal to normal intervals in every 5-min interval within 24 h; rMSSD, the root mean square of successive differences; pNN50, the number of pairs of adjacent normal to normal intervals differing by > 50 ms in the entire recording divided by the total number of normal to normal intervals; LF/HF, the ratio of low frequency/high frequency; SD, standard deviation; HOMA-IR: Insulin resistance index;

The scatter plot demonstrated the association between VFA and HRV (Fig. 1). To further assess the association between VFA and HRV, we conducted a multivariable regression analysis (Table 2). In the fully adjusted model, for every 20-unit increase in VFA, SDNN, SDANN, SDNN index, rMSSD, and pNN50 decreased significantly by 4.00 (95% CI: -5.90, -2.11), 3.51 (95% CI: -5.43, -1.60), 1.35 (95% CI: -2.13, -0.56), 0.92 (95% CI: -1.51, -0.33), and 0.58 (95% CI: -0.99, -0.17). In addition, our study observed a significant positive association between VFA and the LF/HF ratio, with a 0.25 (95% CI: 0.11, 0.39) increase in LF/HF ratio for every 20-unit increase in VFA. According to the mediation analysis, there was a significant mediating effect of insulin resistance index (HOMA-IR) on the association between VFA and SDNN index, rMSSD, and LF/HF ratio. The results showed that for the SDNN index, rMSSD, and LF/HF ratio, the mediating effects proportions of HOMA-IR were 20.32%, 18.76%, and 26.23%, respectively (Table 3). However, when white blood cell counts were analyzed as a mediating variable, we did not find a statistically significant mediating effect (Table S1). In sensitivity analysis, we further adjusted for AHI and blood oxygen saturation, and the findings remained consistent, suggesting that an increase in VFA was significantly associated with a decrease in HRV (Table S2). Furthermore, after imputing the missing values, our results remained robust, demonstrating the reliability of the research conclusion (Table S3). Although the result of the pNN50 index failed to reach the statistical significance threshold after Bonferroni correction (p = 0.009 > 0.05/6), a negative association was still observed. We further conducted subgroup analysis and interaction tests on gender variables, and the results did not show significant gender differences (Table S4).

Fig. 1.

Scatter plot of visceral fat area and heart rate variability

Table 2.

The associations of visceral fat area and HRV indices

	Model 1		Model 2
	β(95%CI)	P-value	β(95%CI)	P-value
SDNN (ms)
per 20 units increment	-4.29 (-6.04, -2.55)	< 0.001*	-4.00 (-5.90, -2.11)	< 0.001*
SDANN (ms)
per 20 units increment	-3.76 (-5.51, -2.01)	< 0.001*	-3.51 (-5.43, -1.60)	< 0.001*
SDNN index (ms)
per 20 units increment	-1.53 (-2.26, -0.80)	< 0.001*	-1.35 (-2.13, -0.56)	< 0.001*
rMSSD (ms)
per 20 units increment	-1.03 (-1.57, -0.50)	< 0.001*	-0.92 (-1.51, -0.33)	0.003*
pNN50 (ms)
per 20 units increment	-0.65 (-1.03, -0.28)	< 0.001*	-0.58 (-0.99, -0.17)	0.006*
LF/HF
per 20 units increment	0.28 (0.15, 0.40)	< 0.001*	0.25 (0.11, 0.39)	< 0.001*

Model 1 adjusted for age group (20 ≤ Age < 35, 35 ≤ Age < 45, Age ≥ 45), gender;

Model 2 further adjusted for physical activity (never exercise, irregular exercise, regular exercise (cumulative > 3 months/year, and more than 3 times per week, and more than 150 min per week), smoking status (never, former, current), drinking status (never, former, current), white blood cell, pulse pressure, LDL-cholesterol; HRV, Heart rate variability; SDNN, SD of all normal to normal intervals; SDANN, SD deviation of sequential 5-min normal to normal interval; SDNN index, the mean value of the SD of all the normal to normal intervals in every 5-min interval within 24 h; rMSSD, the root mean square of successive differences; pNN50, the number of pairs of adjacent normal to normal intervals differing by > 50 ms in the entire recording divided by the total number of normal to normal intervals; LF/HF, the ratio of low frequency/high frequency; SD, standard deviation;

* Bonferroni-corrected threshold α’=0.05/6 outcomes

Table 3.

Mediation analysis of HOMA-IR in the associations of visceral fat area and HRV indices

	Indirect effect	Direct effect	Total effect	Mediation proportion
	β(95%CI)	β(95%CI)	β(95%CI)	P value
SDNN (ms)	-0.39 (-1.05, 0.16)	-3.61 (-5.61, -1.67)*	-4.00 (-5.90, -2.13)*	9.09%, 0.150
SDANN (ms)	-0.22 (-0.93, 0.43)	-3.32 (-5.26, -1.34)*	-3.54 (-5.44, -1.49)*	5.81%, 0.460
SDNN index (ms)	-0.28 (-0.62, -0.04)*	-1.06 (-1.85, -0.28)*	-1.34 (-2.10, -0.56)*	20.32%, 0.026
rMSSD (ms)	-0.18 (-0.43, -0.02)*	-0.74 (-1.34, -0.19)*	-0.92 (-1.50, -0.37)*	18.76%, 0.036
pNN50 (ms)	-0.12 (-0.28, -0.01)	-0.46 (-0.86, 0.06)	-0.59 (-0.99, -0.17)*	20.10%, 0.070
LF/HF	0.07 (0.02, 0.13)*	0.18 (0.05, 0.31)*	0.25 (0.12, 0.38)*	26.23%, 0.004

Mediation analysis adjusted for age group (20 ≤ Age < 35, 35 ≤ Age < 45, Age ≥ 45), gender, physical activity (never exercise, irregular exercise, regular exercise (cumulative > 3 months/year, and more than 3 times per week, and more than 150 min per week)), smoking status (never, former, current), drinking status (never, former, current), white blood cell, pulse pressure, LDL-cholesterol; HRV, Heart rate variability; SDNN, SD of all normal to normal intervals; SDANN, SD deviation of sequential 5-min normal to normal interval; SDNN index, the mean value of the SD of all the normal to normal intervals in every 5-min interval within 24 h; rMSSD, the root mean square of successive differences; pNN50, the number of pairs of adjacent normal to normal intervals differing by > 50 ms in the entire recording divided by the total number of normal to normal intervals; LF/HF, the ratio of low frequency/high frequency; SD, standard deviation; HOMA-IR: Insulin resistance index; * <0.05

Discussion

With increasing altitude, atmospheric pressure and inspired oxygen partial pressure gradually decrease, which can cause systemic hypoxemia. When lowlanders are acutely exposed to high altitudes, the peripheral chemoreceptors located in the carotid body are stimulated by hypoxia, sending signals to the solitary nucleus and rostral ventrolateral medulla, driving an increase in sympathetic nervous tone [24]. Previous studies confirmed that after exposure to high altitude, autonomic nervous activity is significantly impaired, with sympathetic nervous activity becoming relatively dominant [25–28]. A randomized controlled crossover study demonstrated that patients with pulmonary vascular disease showed a significant change in HRV after exposure to high altitude (2500 m), both in the awake state and nocturnal [27]. Another cross-sectional study compared the differences in HRV among miners in high-altitude and plain areas. The findings confirmed a significant decrease in SDNN (110.82 vs. 141.44, p = 0.008) and a notable increase in the LF/HF ratio (858.86 vs. 371.33, p = 0.003) among high-altitude miners, indicating an imbalance in autonomic nervous function [29]. Chen et al. investigated the relationship between HRV indices changes and acute mountain sickness (AMS) in healthy individuals after a rapid ascent to high altitude (3180 m) [28]. They confirmed that, compared to sea level, the LF/HF ratio significantly increased after high-altitude exposure. Furthermore, it was observed that during high-altitude exposure, the changes in HRV indices (SDRR, TP, LF, and HF) were significantly greater in the AMS group. In contrast, these indices showed a significant decrease in subjects without AMS. This suggested that the difference in adaptability to high altitude was significantly related to the change in HRV [28]. Nonetheless, the underlying mechanisms contributing to impaired autonomic nervous function at high altitudes remain inadequately investigated. Currently, it is speculated that this impairment may be associated with structural damage to the heart, excessive activation of inflammatory responses, and alterations in the endocrine system.

Our study, based on lowlanders with OSA at high altitudes, is the first to demonstrate a significant negative association between visceral fat and SDNN, SDANN, SDNN index, rMSSD, and pNN50. This finding, validated in both time-domain and frequency-domain measures, further underscores that visceral fat accumulation may contribute to cardiometabolic disease risk. Interestingly, our study observed a positive association between VFA and the LF/HF ratio, which was consistent with a previous study [30]. The LF/HF ratio is mainly used to evaluate the relative balance state between the sympathetic nerve and the parasympathetic nerve. An increase in this indicator usually suggests autonomic nerve dysfunction and indicates enhanced sympathetic nerve activity. A reduction in HRV is significantly associated with adverse cardiovascular events, such as coronary heart disease [31], atrial fibrillation [32], heart failure [31], and all-cause mortality [33, 34]. Janszky et al. included 251 middle-aged women hospitalized for acute coronary syndrome, aiming to evaluate the predictive power of HRV for long-term mortality in middle-aged women with coronary heart disease [33]. This study found that for every 25% reduction in the SDNN index, the risk of all-cause mortality increased significantly by 56% (HR 1.56, 95% CI 1.19–2.05). Another longitudinal study involving 14,019 participants with an average age of 54 ± 5.7 years confirmed that for every 1 standard deviation reduction in log-transformed SDNN and log-transformed rMSSD, the risk of atrial fibrillation increased by 14% and 7%, respectively [32]. The cohort study conducted by Kaze et al., which included 7,160 participants with type 2 diabetes, found that for each reduction of one standard deviation in SDNN, the risk of heart failure increased by 23% (HR 1.23, 95% CI 1.08–1.41) [35]. It is worth noting that our study focused for the first time on the HRV levels of OSA patients in high-altitude environments.

The association between visceral obesity and insulin resistance has been widely confirmed [36–41]. Our study suggests that in OSA patients, visceral fat is not only associated with insulin resistance, but may also exacerbate the risk of cardiovascular disease by affecting HRV. Abnormal metabolism of visceral fat caused by hypoxia may explain our findings. A prior animal study has demonstrated that intermittent hypoxia pattern can downregulate insulin-receptor substrate (IRS-1) mRNA and inhibit the insulin signaling pathway [42]. Furthermore, intermittent hypoxia has been shown to enhance the infiltration of pro-inflammatory adipose tissue macrophages and directly induce macrophage polarisation, which is significantly correlated with the severity of insulin resistance [42]. Gileles-Hillel et al. further confirmed that intermittent hypoxia not only promoted the increase in the number of M1 proinflammatory macrophages in visceral white adipose tissue but also led to dysregulation in the electron transport chain of mitochondria, thereby increasing the production of reactive oxygen species [43], which may exacerbate insulin resistance. Moreover, Khalyfa et al. conducted a study that evaluated the cellular heterogeneity of visceral white adipose tissue after simulated intermittent hypoxia (IH) exposure using combined single-nucleus RNA sequencing (snRNA-seq) and aggregate RNA-seq methods, and performed differential expression gene (DEGs) analyses [44]. The results indicated that intermittent hypoxia is primarily involved in pathways related to metabolic dysfunction and insulin resistance. Many previous clinical studies have confirmed the association between insulin resistance and HRV, which provides further support for our findings [45–48]. In addition to its association with insulin resistance, visceral fat may also contribute to autonomic dysfunction through various other mechanisms. Visceral adipose tissue can secrete a variety of pro-inflammatory cytokines, which contribute to autonomic dysfunction by affecting sympathetic nervous activity [49]. A recent cross-sectional study of 128 adult healthy subjects confirmed that visceral fat levels were significantly associated with a variety of inflammatory factors, including hsCRP, TNF-α, IL-6, and white blood cell count [50]. This finding has also been validated in elderly [51] and postmenopausal women [52]. In addition, visceral fat accumulation increases the secretion of leptin and resistin as well as decreases the secretion of adiponectin [53–56]. A previous animal study demonstrated that intravenous leptin infusion significantly increased sympathetic nerve activity in Sprague-Dawley rats [57]. A cohort study conducted in England indicated that elevated baseline adiponectin levels in patients with type 2 diabetes were correlated with an increased HRV, which highlights the influence of adiponectin levels on autonomic nervous function [58]. Wakabayashi et al. also confirmed a significant negative association between adiponectin levels and the LF/HF ratio in patients with type 2 diabetes (β = -0.332, P = 0.020) [59].

Our study has several limitations that warrant consideration. First, the cross-sectional, single-center design of our study introduces the potential for selection bias and limits our ability to infer causal relationships. Second, our study lacked a comparison with plain OSA patients, and could not explore whether the high-altitude environment aggravated the impairment of autonomic nervous function caused by visceral fat. Third, we did not directly measure indices of sympathetic nervous activity, such as plasma adrenaline, norepinephrine, and catecholamine levels. Evaluating these markers can provide a more comprehensive understanding of the impact of autonomic nerve activity at high altitude. Fourth, VFA was assessed using BIA, a method known to be less accurate than computed tomography or magnetic resonance imaging. It is susceptible to measurement errors. Fifth, the sample size of this study is limited. In the future, it is necessary to expand the sample size and include subjects of different ages and ethnic groups to improve the universality of the research results. Sixth, there may be uncontrolled confounding factors, such as dietary habits or genetic background, which could potentially influence the observed associations.

Despite the aforementioned limitations, this study yields findings of significant clinical relevance. First, we have substantiated the crucial role of visceral adipose tissue in cardiovascular injury under high-altitude conditions. Therefore, the assessment of VAT should be prioritized during the diagnosis and management of obstructive sleep apnea (OSA) patients residing in high-altitude regions. Furthermore, our research underscores the importance of implementing targeted interventions in OSA patients, including dietary regulation [60] and physical exercise [61], to mitigate VAT accumulation. In summary, this study provides preliminary evidence suggesting a potential association between VAT and HRV in OSA patients at high altitudes, highlighting the pivotal role of VAT in cardiovascular risk management.

Conclusion

Our findings indicated that visceral fat serves as a significant determinant of cardiovascular health among OSA patients residing at high altitudes and may represent a viable target for intervention and preventive strategies. Future research is warranted to further elucidate the intricate interplay between visceral fat and autonomic nervous dysfunction within the context of hypobaric hypoxia, as well as to investigate the underlying mechanisms involved.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

OSA

Obstructive sleep apnea

VFA

Visceral Fat Area

HRV

Heart rate variability

CVD

Cardiovascular diseases

AHI

Apnea-Hypopnea Index

HOMA-IR

Insulin resistance index

Author contributions

SSJ was the first author of the paper and made important contributions to the acquisition, analysis, and interpretation of the data and the drafting of the paper. SSJ, YXF, and YW contributed to the revision of the manuscript. YXF, HWL, YW, DDH, and XPC contributed to the data collection. YHW, XPC, and LMZ contributed to the study design. All authors contributed to the article and approved the submitted version.

Funding

This study is funded by the technology department of Tibet, the central government guides the local science and technology development fund project (No.XZ202202YD0011C), Science and Technology Program of Tibet Autonomous Region (Grant number: XZ202303ZY0004G), Science and Technology Major Project of Tibetan Autonomous Region of China (Grant number: XZ202201ZD0001G01), and Key Research and Development Program of the Department of Science and Technology of the Tibet Autonomous Region (XZ202402ZY0003).

Data availability

The data set for this study may be obtained by applying to the corresponding author for a reasonable reason.

Declarations

Competing interests

The authors declare no competing interests.