Cross-sectional study of the association between triglyceride glucose-body mass index and obstructive sleep apnea risk

Abstract

BACKGROUND

The triglyceride glucose-body mass index (TyG-BMI) is a novel indicator of insulin resistance (IR). Obstructive sleep apnea (OSA) is a prevalent disorder characterized by recurrent complete or partial collapse of the pharyngeal airway during sleep; however, the relationship between these two conditions remains unexplored. We hypothesized that a higher TyG-BMI is associated with the occurrence of OSA.

AIM

To assess the association between TyG-BMI and OSA in adults in the United States.

METHODS

A cross-sectional study was conducted utilizing data from the National Health and Nutrition Examination Surveys spanning from 2005-2008 to 2015-2018. TyG-BMI was calculated as Ln [triglyceride (mg/dL) × fasting blood glucose (mg/dL)/2] × BMI. Restricted cubic splines were used to analyze the risk of TyG-BMI and OSA occurrence. To identify potential nonlinear relationships, we combined Cox proportional hazard regression with smooth curve fitting. We also conducted sensitivity and subgroup analyses to verify the robustness of our findings.

RESULTS

We included 16794 participants in the final analysis. Multivariate regression analysis showed that participants with a higher TyG-BMI had a higher OSA incidence. After adjusting for all covariates, TyG-BMI was positively correlated with the prevalence of OSA (odds ratio: 1.28; 95% confidence interval: 1.17, 1.40; P < 0.001); no significant nonlinear relationship was observed. Subgroup analysis showed no strong correlation between TyG-BMI and OSA in patients with diabetes. The correlation between TyG-BMI and OSA was influenced by age, sex, smoking status, marital status, hypertensive stratification, and obesity; these subgroups played a moderating role between TyG-BMI and OSA. Even after adjusting for all covariates, there was a positive association between TYG-BMI and OSA prevalence.

CONCLUSION

A higher TyG-BMI index is linked to higher chances of developing OSA. As TyG-BMI is an indicator of IR, managing IR may help reduce the risk of OSA.

Key Words: Obstructive sleep apnea; Triglyceride glucose-body mass index; Insulin resistance; Cross-sectional; National Health and Nutrition Examination Surveys

Core Tip: This study confirmed a positive linear exponential risk relationship between triglyceride glucose-body mass index (TyG-BMI) and obstructive sleep apnea (OSA) in adults in the United States, underscoring the potential utility of TyG-BMI as an evaluation and monitoring tool for insulin resistance (IR) in patients with OSA. The results suggest that assessing IR through the TyG-BMI index may help in the clinical evaluation of patients with OSA, independently of other metabolic risk factors.

INTRODUCTION

Obstructive sleep apnea (OSA) is a chronic sleep disorder characterized by recurrent complete or partial collapse of the pharyngeal airway during sleep. This results in the cessation or significant reduction in airflow, known as apnea. This condition induces chronic intermittent hypoxia and disruption of normal sleep structure and rhythm[1]. OSA can lead to daytime sleepiness, which affects patient safety and is associated with cerebrovascular complications and cognitive dysfunction, particularly in severe cases[2]. It is estimated that approximately one-seventh of the global adult population, equivalent to one billion individuals, suffer from OSA[3]. Despite its prevalence, diagnosis and treatment rates remain low, representing an increasing concern worldwide due to its substantial economic burden[4].

The relationship between OSA and type 2 diabetes mellitus (T2DM) has recently garnered attention. It has been shown that the severity of sleep apnea correlates with an increased risk of T2DM, and that they mutually affect each other[5]. The incidence of T2DM among patients with OSA ranges from 15% to 30%. Insulin resistance (IR) is a key factor linking OSA and T2DM. Numerous cross-sectional, observational, and large-population studies have shown independent correlations among OSA, IR, and T2DM[6-8]. Patients with OSA have a higher likelihood of developing IR, which can exacerbate blood sugar control issues in patients with diabetes[9,10]. Obesity is a major risk factor for OSA, with at least 30% of patients with obesity suffering from it, and 60% of patients with OSA being obese[11]. Both conditions are closely related and can lead to increased sympathetic nervous system activity, oxidative stress, and chronic low-grade inflammation, ultimately leading to metabolic dysfunctions such as visceral white adipose tissue IR[12]. Research has shown a significant correlation between relief in IR and changes in the visceral adiposity index, with body mass index (BMI) considered a predictive indicator of IR among patients with OSA[13]. Certain data indicate that a 6-unit increase in BMI correlates with a 4-fold increase in the risk of OSA[14]. Therefore, the relationship between IR and OSA has emerged as a prominent area of research.

The triglyceride glucose (TyG) index, first proposed by Simental-Mendia, is well correlated with traditional IR indicators such as the hyperinsulin-normal glucose clamp and homeostasis model assessment of IR[15]. The TyG-BMI, which is based on the TyG index, was officially proposed in 2016. Including BMI as an obesity indicator has improved the predictive ability of this model. Er et al[16] found that among various biomarkers, TyG-BMI has the strongest association with the homeostasis model assessment of IR. They also suggested that it has a greater diagnostic value than the TyG index. Recently, TyG-BMI gained approval for prediabetes assessment, marking a new area of research[15]. However, the relationship between TyG-BMI and OSA remains unclear.

In this study, we analyzed the association between TyG-BMI and OSA using large population data from the National Health and Nutrition Examination Survey (NHANES). We hypothesized that a higher TyG-BMI is associated with the occurrence of OSA. Our findings may provide insights into the underlying pathophysiological mechanisms of the syndrome.

MATERIALS AND METHODS

Study design and population

This study utilized data from NHANES 2005-2018. This program was designed to investigate the health and nutritional status of American adult individuals through interviews, examinations, and laboratory assessments. The program encompasses four discontinuous cycles conducted between the 2005-2008 and 2015-2018 periods. The data comprised 33096 patients with OSA and 72257 patients with available TyG-BMI data. From this pool, 16794 eligible participants were selected for further analyses. The study procedure is illustrated in detail in Figure 1. The exclusion criteria comprised the following: (1) Missing data of diagnostic processes for OSA; and (2) Missing data of covariates. The National Center for Health Statistics Research Ethics Review Board granted the approval for the conduction of NHANES, and written informed consent was obtained from each participant. This study was reviewed and approved by the Ethics Committee of Shenzhen Bao'an Chinese Medicine Hospital Ethics Committee.

Figure 1 Flow chart of sample selection from National Health and Nutrition Examination Surveys 2005-2018. NHANES: National Health and Nutrition Examination Surveys; OSA: Obstructive sleep apnea.

Outcome and exposure factors

The primary exposure factor examined in this study was TyG-BMI. The BMI, TyG index, and TyG-BMI index were calculated as follows: BMI = body weight (kg)/[height (m)]²; TyG = Ln [fasting triglycerides (mg/dL) × fasting plasma glucose (mg/dL)/2][17]; and TyG-BMI = TyG index × BMI (kg/m²), respectively[16]. The main outcome of interest was the occurrence of OSA.

Covariates

Extra covariates were retrieved from each NHANES cycle. Continuous variables included age and poverty income ratio, while categorical variables included sex, age, ethnicity, educational level, marital status, smoking status, and drinking status. Several potential confounding variables were adjusted based on the findings of previous studies. Specifically, age was categorized into the following ranges: 18-39 years, 40-59 years, and ≥ 60 years. The poverty income ratio (%) was categorized as follows: ≤ 1.3, > 1.3, ≤ 3.5, and > 3.5. Ethnicity was categorized as Mexican American, other Hispanic, non-Hispanic white, black, or other. The participants were also categorized according to their BMI (kg/m²) as follows: < 25, 25-29.9, and ≥ 30. The categories corresponding to educational level were “less than high school”, “high school”, or “beyond high school”, and those corresponding to marital status were “married” and “other”. Self-reported medical conditions included hypertension, diabetes, smoking, and alcohol consumption.

Statistical analyses

All statistical analyses were conducted in accordance with the United States Centers for Disease Control and Prevention guidelines, using appropriate NHANES sampling weights and considering a complex multistage cluster survey design in the analysis. Categorical variables are expressed as percentages, and continuous variables as mean ± standard deviation. We used a weighted Student's t-test (for continuous variables) and a weighted χ² test (for categorical variables) to assess differences in the presence or absence of OSA. Multiple logistic regression was used to analyze the relationship between TyG-BMI and OSA according to three different models: In model 1, there were no adjustment covariates. Model 2 was adjusted for age, sex, and ethnicity. Model 3 was adjusted for sex, age, ethnicity, poverty-to-income ratio, BMI, education, marital status, smoking, alcohol consumption, diabetes, hypertension, and other factors. After adjusting for covariates, restricted cubic splines were used to visualize the dose-response relationship between TyG-BMI and OSA. Subgroup analysis was also performed, stratified by sex, age, marital status, alcohol consumption, smoking status, BMI, and T2DM, and stratified multiple regression analysis was applied. In addition, we used a log-likelihood test model to examine the heterogeneity of associations between subgroups. A stepwise coefficient test regression method was used to verify the mediating effect of different subgroup variables on the relationship between TyG-BMI and OSA. P < 0.05 was considered statistically significant. All analyses were performed using R version 3.4.3 (http://www.Rproject.org, the R Foundation). All raw data can refer to the Supplementary Table 1.

RESULTS

Participant characteristics

Table 1 presents the baseline characteristics of participants categorized by TyG-BMI. Supplementary Table 2 provides baseline data. A total of 16794 United States adults met the study criteria. Participants were categorized into two groups based on the presence or absence of OSA. Among all participants, OSA was observed in 4.740% of the women and 7.657% of the men. The median TyG-BMI for patients with OSA was 277.49 (101.08), while that for patients without OSA was 243.63 (84.52). The differences for each variable analyzed were statistically significant between the groups in all cases.

Table 1 Characteristics of participants (by category of triglyceride glucose-body mass index in the National Health and Nutrition Examination Surveys).

Characteristic	OSA status		P value
	With OSA	Without OSA
Participants (n)	2082	14712
TyG-BMI, median (IQR)	277.49 (101.08)	243.63 (84.52)	< 0.001
Age, years (%)			< 0.001
18-39	2.864	31.916
40-59	5.198	27.516
≥ 60	4.335	28.171
Sex (%)			< 0.001
Female	4.740	45.606
Male	7.657	41.998
Ethnicity (%)			< 0.001
Mexican American	1.780	14.255
Other Hispanic	1.316	8.086
Non-Hispanic white	5.526	38.097
Non-Hispanic black	2.602	18.888
Other ethnicities	1.173	8.277
Poverty income ratio (%)			< 0.001
≤ 1.3	3.710	24.533
>1.3 and ≤ 3.5	4.829	34.381
> 3.5	3.859	28.689
BMI (kg/m2) (%)			< 0.001
< 25	1.947	25.646
25-29.9	3.513	29.826
≥ 30	6.937	32.131
Education level (%)			< 0.001
Less than high school	2.936	19.530
High school	3.090	20.478
More than high school	6.371	44.594
Marital status (%)			< 0.001
Married	7.711	48.785
Others	4.686	38.817
Hypertension (%)			< 0.001
Yes	5.913	28.582
No	6.484	59.021
Diabetes (%)			< 0.001
Yes	2.525	10.569
No	9.873	77.033
Smoking (%)			< 0.001
Yes	6.711	39.026
No	5.687	48.577
Alcohol consumption (%)			< 0.001
Yes	9.241	63.785
No	3.156	23.818

Body mass index was categorised as < 25, 25-29.9, and ≥ 30 kg/m², which corresponded to normal weight, overweight, and obesity, respectively. BMI: Body mass index; IQR: Interquartile range; NHANES: National Health and Nutrition Examination Surveys; OSA: Obstructive sleep apnea; TyG-BMI: Triglyceride glucose-body mass index.

Regression analysis

The TyG-BMI quartile values were as follows: < 208 for Q1, (208-247) for Q2, (247-295) for Q3, and ≥ 295 for Q4. Table 2 shows that in the unadjusted odds ratio (OR) [95% confidence interval (CI) = 1.51 (1.45, 1.58)], minimally adjusted [1.52 (1.46, 1.59)], and fully adjusted models [1.28 (1.17, 1.40)], TyG-BMI was positively associated with the risk of OSA occurrence. Regression analysis revealed that participants with higher TyG-BMI values had a higher likelihood of OSA occurrence. Upon dividing TyG-BMI into quartiles, participants in the first quartile had a higher risk of OSA occurrence before adjustment (Q1: 3.58; P < 0.001). Across all three models, a significant relationship was observed between the TyG-BMI of the fourth quartile and the risk of OSA (all P < 0.05). The risk of OSA in participants with a lower TyG-BMI in the fourth quartile [Q4: 1.94 (1.46, 2.56)] was higher than that in participants in the lowest quartile (Q1; P < 0.05).

Table 2 Association between triglyceride glucose-body mass index and obstructive sleep apnea.

Exposure	Model 1		Model 2		Model 3
	OR (95%CI)	P value	OR (95%CI)	P value	OR (95%CI)	P value
TyG-BMI	1.31 (1.21, 1.41)	< 0.001	1.29 (1.18, 1.40)	< 0.001	1.10 (1.01, 1.21)	0.033
TyG-BMI quartile
Q1, < 8.19	Reference		Reference		Reference
Q2, ≥ 8.19, < 8.59	1.85 (1.40, 2.45)	< 0.001	1.57 (1.18, 2.10)	0.002	1.33 (0.99, 1.78)	0.059
Q3, ≥ 8.59, < 9.03	2.37 (1.81, 3.11)	< 0.001	2.02 (1.52, 2.67)	< 0.001	1.51 (1.13, 2.02)	0.005
Q4, ≥ 9.03	2.48 (1.89, 3.25)	< 0.001	2.26 (1.7, 3.00)	< 0.001	1.40 (1.04, 1.89)	0.029

Model 1 not adjusted for any covariate. Model 2 adjusted for age, sex, and ethnicity. Model 3 Adjusted for sex, age, ethnicity, poverty income ratio, education, marital status, smoking, alcohol use, diabetes, and hypertension. Body mass index was categorized as < 25, 25-29.9, and ≥ 30 kg/m², which corresponds to normal weight, overweight, and obesity, respectively. CI: Confidence interval; OR: Odds ratio; OSA: Obstructive sleep apnea; TyG-BMI: Triglyceride glucose-body mass index.

Subgroup analysis

Subgroup analysis was used to evaluate the stability of the association between TyG-BMI and OSA prevalence in different states. We tested the interactions among BMI, age, sex, diabetes, high blood pressure, smoking, and alcohol consumption. The results showed that only BMI, age, sex, diabetes, hypertension, and marital status had statistically significant associations (all interactions P < 0.001). As shown in Figure 2, there was no significant correlation between TyG-BMI and OSA in patients with a BMI of 25-29.9 kg/m² or with diabetes (all P > 0.05). However, there was a significant positive correlation between TyG-BMI and OSA in patients that did not smoke or had hypertension. These was verified for both sexes and regardless of marital status (all P < 0.05). Subsequently, we conducted a mediating analysis considering those subgroups with a statistically significant interaction, and found that BMI accounted for 0.03% of the mediating effect of TyG-BMI on OSA, age accounted for 4.885%, sex accounted for 7.34%, and diabetes status accounted for 1.52%. Hypertension accounted for 8.12% of the mediating effect, and “other” as marital status accounted for 3.82%, as shown in Table 3.

Figure 2 Forest map of association between triglyceride glucose-body mass index and obstructive sleep apnea.

Table 3 Subgroup and meditation analysis.

Subgroups	OR (95%CI)	P value	P for interaction	Mediation %
BMI			< 0.001	0.03
< 25	1.30 (1.01-1.65)	0.046
25-29.9	1.06 (0.92-1.21)	0.448
≥ 30	1.46 (1.28-1.67)	< 0.001
Age			< 0.001	4.885
18-39	1.31 (1.10,1.56)	0.003
40-59	1.29 (1.16-1.44)	< 0.001
≥ 60	1.18 (1.02-1.37)	0.029
Sex			< 0.001	7.34
Female	1.39 (1.20-1.60)	< 0.001
Male	1.20 (1.07-1.34)	0.001
Marital status			< 0.001	3.82
Others	1.40 (1.22-1.61)	< 0.001
Married	1.19 (1.06-1.33)	0.002
Diabetes			0.01	1.52
No	1.30 (1.18-1.43)	< 0.001
Yes	1.16 (0.93-1.43)	0.182
Smoking			0.06
No	1.39 (1.22-1.58)	< 0.001
Yes	1.19 (1.05-1.34)	0.005
Hypertension			< 0.001	8.12
No	1.25 (1.11-1.41)	< 0.001
Yes	1.26 (1.10-1.44)	< 0.001

The results of subgroup analysis were adjusted for all covariates considered in model 3 except effect modifier. Body mass index was categorized as < 25, 25-29.9, and ≥ 30 kg/m², which corresponded to normal or low weight, overweight, and obesity, respectively. BMI: Body mass index; CI: Confidence interval; OR: Odds ratio.

Nonlinear analysis

After adjusting for all covariates, there was a positive correlation between TyG-BMI and OSA prevalence (Figure 3), with no clear evidence of a nonlinear relationship.

Figure 3 Restricted cubic splines for the association of triglyceride glucose-body mass index and obstructive sleep apnea.

DISCUSSION

To the best of our knowledge, this is the first study to assess the correlation between the TyG-BMI index and the risk of developing OSA by analyzing large population data from the NHANES.

OSA is a sleep-related disorder that induces hypoxia and is currently recognized as a risk factor for various conditions, including cardiovascular disease, cerebrovascular disease, dementia, and T2DM[18-20]. IR has been identified as the pathogenic mechanism linking OSA to its related diseases[21-23]. The precise mechanism by which OSA induces IR remains unclear; however, chronic intermittent hypoxia, fragmented sleep, and sympathetic hyperactivity are believed to be the primary contributing factors[24].

Chronic intermittent hypoxemia is the primary pathophysiological pathway through which OSA influences IR[25]. Short-term intermittent nocturnal hypoxia has been observed to increase IR[26], whereas long-term exposure to hypoxia leads to sustained IR increase, indicating that glucose metabolism disorders render individuals more susceptible to IR due to intermittent hypoxia[27]. Related studies suggest that IR may result from mitochondrial oxidative stress[28], induction of chronic elevation of hypoxia inducible factor 1 alpha in serum, disruption of pancreatic b-cell inhibition of glucose transport[29], interference with a single gut microbiome extracellular vesicle pathway, and disruption of adipocyte homeostasis[30] and other pathways, leading to the occurrence of IR. Additionally, research has shown that complete sleep deprivation reduces glucose tolerance[31]. A previous study involving 1000 male patients with OSA found that individuals with an average sleep duration of less than 5 hours were two times more likely to develop diabetes compared to those who slept for 7-8 hours[32]. This indicates a significant association between the quality and duration of sleep and IR. Similarly, another study provided evidence that intermittent hypoxia induces a proinflammatory phenotype in adipose tissue, which may be a key step in the development of OSA and IR, and also provided a potential mechanism by which IR could affect the occurrence of OSA[21].

Furthermore, activation of the sympathetic nervous system has been observed in patients with OSA. Research indicates that continuous low-oxygen levels and re-oxygenation during respiratory pauses can affect the sensitivity of peripheral chemoreceptors in these patients[33], resulting in cortical thickening, autonomic neuroplasticity, and increased activity of the sympathetic nervous system[34]. Consequently, patients with OSA often exhibit enhanced sympathetic nervous system responses[35], which can lead to vasoconstriction, decreased glucose intake, and elevated blood sugar levels.

In addition, metabolic disorders and obesity should be considered as confounding factors[36]. Bikov et al[37] adjusted for the correlation between TyG and BMI in OSA; however, even in non-obese individuals, a correlation was observed between BMI and the TyG index. This emphasizes the need to adjust for BMI when investigating the association between OSA and metabolic factors. The TyG-BMI index is associated with the BMI and TyG indices and is relevant in both obese and non-obese individuals. This may explain the uncertainty in the effect of BMI on TyG and OSA in the subgroup analysis. Relevant studies have shown that obesity plays a role in the pathophysiological pathways of OSA and IR. Studies have shown that in OSA patients, the pharyngeal airway structure collapses, and obesity structurally increases the tissue pressure around the pharyngeal airway, thereby narrowing the airway and causes chronic intermittent hypoxia[38]. Intermittent hypoxia exacerbates metabolic dysfunction in obesity by promoting IR[39]. Several meta-analyses of adult and pediatric studies[40,41] have found that a short sleep duration increases the risk of obesity, possibly due to lower leptin levels, higher ghrelin levels, and increased hunger and appetite[42]. Leptin is a hormone produced by adipose tissue in the brain that stimulates the breakdown of white fat[43]. Decreased leptin levels result in white fat accumulation, which leads to obesity.

Our study had several strengths. First, this is the first study to examine the association between TyG-BMI and the likelihood of developing OSA. Our findings revealed a positive correlation between TyG-BMI and the incidence of OSA, suggesting that TyG-BMI can serve as an evaluation and monitoring indicator of IR in patients with OSA. This study provides novel and practical insight into the diagnosis and treatment of OSA, including appropriate weight control, especially for individuals with visceral obesity. The findings of our study indicate that assessing IR, particularly through the TyG-BMI index, could be beneficial to patients with OSA. As a new indicator related to TyG, TyG-BMI combined with BMI, a measurement indicator related to human obesity, may have a potential impact on diseases related to human long-term health caused by obesity, compared with TyG. In a clinical setting, the TyG-BMI index may be more informative because it has the potential to reflect metabolic and anthropometric factors, which can be critical in people with varying degrees of obesity. Studies have also confirmed that TyG-BMI has better accuracy in metabolic related diseases such as arteriosclerosis, metabolic syndrome and steatosis[44,45]. We speculate that the TyG-BMI index may provide nuanced insights into metabolic health that the TyG index alone cannot capture. This index could be incorporated into routine clinical assessment of patients with OSA, even in the absence of any other metabolic risk factors. Second, our study was based on a large-scale analysis of the NHANES population, which includes a diverse and sizable sample of 16794 United States individuals from multiple regions and belonging to different ethnicities. However, our study also had certain limitations. The primary limitation concerns the causality between TyG-BMI and the risk of OSA occurrence. Due to the cross-sectional study design, information on OSA was obtained through surveys, which could introduce bias in the data. Some participants may have hesitated to respond to relevant questions for various reasons, resulting in unanswered questions, whereas others may have intentionally omitted certain information when completing the questionnaire. Despite our efforts to adjust for potential confounding factors, there may still be factors that were not accounted for.

CONCLUSION

Using a nationally representative population, this study confirms a positive linear risk relationship between TyG-BMI and OSA in adults in the United States. A higher TyG-BMI index was linked to a higher chance of developing OSA. As TyG-BMI is an indicator of IR, managing IR may help reduce the risk of OSA. TyG-BMI can be used as an indicator of IR, preferentially in patients with certain risk factors (obesity, a family history of diabetes, or other metabolic syndromes), and may be combined with other screening tools in a screening program for OSA. This study provides a novel strategy to improve the screening, diagnosis, and management of OSA, although further prospective studies are needed to validate the findings.