Background

The global prevalence of hypertension is rising. In 2019, approximately 12.78 million people worldwide had hypertension [1], leading to a large global burden of cardiovascular disease and premature death [2]. Previous studies have demonstrated that insulin resistance (IR) [3], obesity [4], and hypertension are closely interrelated.

Different measures of obesity have been defined, including the well-recognized body mass index (BMI) reflecting total body mass[BMI~(weight/height**2)], waist circumference (WC) reflecting abdominal obesity, and body fat percentage (BF%). Considerable evidence from longitudinal studies confirms that WC and BMI are significantly associated with an increased risk of hypertension in diverse populations [4, 5]. Moreover, obesity plays a role in the pathogenesis of IR [6, 7]. As summarized in a meta-analysis, obesity indicators such as WC and BMI were most commonly used for variable adjustments when analyzing the relationship between IR and hypertension [3]. Some researchers found that obesity may have a mediating effect on the association between IR and hypertension [8, 9], but others reported that obesity does not modify this relationship [10]. Therefore, the role of obesity in the relationship between IR and hypertension based on the results of existing studies remains controversial. Recent studies have suggested that BF% is significantly associated with risk of hypertension [11, 12]. It is worth noting that a Korean cohort study revealed BF% as a predictor of hypertension, even in nonobese individuals who were defined based on BMI and WC criteria [13]. However, no study has evaluated the role of BF% on the association between IR and hypertension. IR and obesity are risk factors for hypertension, that often coexist. Only a few studies have reported an interaction between obesity and IR in hypertension. A review reported that obesity, particularly if defined by WC, appears to have an important influence on the IR–hypertension relationship [14], and a study in postmenopausal women revealed a significant interaction between WC and log homeostasis model assessment-IR (HOMA-IR) on systolic blood pressure (SBP) [15]. Results from a prospective cohort study, using IR to define unhealthy metabolic statuses, found that overweight and obese participants in IR groups showed a significant and independent risk of hypertension [16]. Zhang et al. [17] have shown that adiposity in the development of hypertension is modified by IR. Therefore, obesity and IR are interconnected in various ways. The role of the interactions between different obesity indicators and IR on hypertension risk remains unclear.

Although the coexistence of IR and obesity as major risk factors for hypertension has been widely documented, Chinese rural populations may show different associations. Herein, we aimed to assess the association between IR and hypertension in a Chinese rural population and further explored the interactions between different obesity indicators (WC, BMI, and BF%) and IR on hypertension risk.

Materials and methods

Study population

This study was derived from the China Northwest Natural Population Cohort, Ningxia Project (CNC-NX), an ongoing population-based prospective cohort study. The CNC-NX protocol has been previously reported in detail [18]. Briefly, 15,802 participants (age range 35–74 years) from 45 villages in Wuzhong and Shizuishan City in the Ningxia Hui Autonomous Region of China were enrolled at baseline between March 2018 and May 2019. Demographic characteristics and anthropometric and biochemical measurements were obtained from all participants. Of these 15,802 participants, approximately 30% (5300 participants) were randomly selected comprising a representative subcohort. Among these 5300 individuals, we excluded 73 with missing data on fasting insulin levels, 117 with missing lipid parameters, 75 with missing blood pressure data, and 147 with other missing variables. Therefore, 4888 participants were included in our analysis.

General information and anthropometric measurements

Information on general characteristics (e.g., age, sex, educational status, cigarette smoking, alcohol intake, disease history, medication history) was extracted by trained research assistants from questionnaires. The body height was measured without shoes. BMI, WC, and BF% were estimated using bioelectrical impedance analysis (BIA) devices (InBody 370 system, Biospace, Korea) according to standard operating guidelines. The participants removed outer garments and stood barefoot on the BIA device, which passed small electrical currents through the body to estimate body composition. SBP and diastolic blood pressure (DBP) were measured using an OMRON automatic monitor (OMRON-7124, Omron Corporation, Japan) after the participants had rested for at least 5 min. Two consecutive readings were obtained and the average of the two readings was calculated as the blood pressure value for analysis.

Biochemical measurements

Blood samples were collected from the participants between 6:00 and 8:00 a.m. after 8–12 h of fasting. Biochemical indicators were measured using a biochemical autoanalyzer (Mindray BS-430, Shenzhen, China). Fasting insulin levels were measured using a chemiluminescence immunoassay analyzer (Mindray CL-2000i, Shenzhen, China). IR was estimated using the HOMA-IR and calculated using the following formula: HOMA-IR = fasting glucose (mmol/L) × fasting insulin (mIU/mL)/22.5 [19].

Definitions

Hypertension was defined as SBP ≥ 140 mmHg, DBP ≥ 90 mmHg, self-reported hypertension, or current use of blood pressure-lowering medication. Obesity was defined as: (1) WC ≥ 90 cm for men and ≥ 85 cm for women, (2) BMI ≥ 30 (kg/m2), and (3) BF% ≥25% in men and ≥ 35% in women [20]. IR was defined in the present study as an HOMA-IR value above the 75th percentile (> 2.59).

Statistical analysis

Participant characteristics were described according to the presence or absence of hypertension. Except for insulin levels and HOMA-IR, which are presented as the median, other continuous variables are presented as the mean ± standard deviation. Nonparametric tests or t-tests were used to compare differences between the non-hypertension and hypertension groups, where appropriate. Categorical variables such as sex, education status, cigarette smoking, and alcohol intake are presented as frequencies (%) and were compared using the chi-square test. A binary logistic regression model adjusted for related potential confounders was used to examine the association between HOMA-IR and hypertension, especially in Model 4 with added WC, BMI, and BF%, to test whether different obesity indicators affect the relationship between HOMA-IR and hypertension. Test for trends based on variables containing median values for each quartile. In addition, we assessed the additive interaction based on Hosmer et al. [21] to calculate the relative excess risk (RERI) and attributable proportion (AP) due to interaction if the 95% confidence interval (CI) did not include 0, suggesting that significant interactions exist. These parameters have been proposed as interaction measures in epidemiologic studies [30,31,32]. Recent cohort studies have demonstrated that BF% measured by BIA is independently associated with cardiovascular events [33], including hypertension [12]. Moreover, BF% is more strongly correlated with cardiovascular risk [33, 34] and hypertension [35] than with BMI or WC. Thus, whether BF% affects the relationship between IR and hypertension independent of BMI and WC needs to be confirmed. Our study found additive interactions between high HOMA-IR and obesity, defined as a high BF%, on hypertension. Interestingly, our study findings suggest that high truncal and leg BF% with high HOMA-IR has significant interactions with hypertension only in women, but not in men. The differences in the distribution of BF% between men and women might provide an explanation for this discrepancy; women had higher truncal BF% than men (36.65 vs. 28.21, P < 0.001), and leg BF% in women was higher than that in men (33.31 vs. 27.43, P < 0.001). A growing body of evidence supports that truncal adipose tissue has a significant impact on the development of IR and diabetes mellitus [36], and a study found that truncal fat-to-leg fat ratio is significantly associated with cardiovascular disease in patients with type 2 diabetes independently of BMI and WC [37]. Tillin et al. [38] also described that IR and truncal obesity account for the two-fold excess diabetes risk in Indian Asian and African Caribbean women, but not in men, in a cohort followed up for 20 years. The underlying explanation might be that truncal obesity causes an inflammatory state and leads to metabolic diseases such as hypertension and IR. However, the interactions of BF% with IR on hypertension risk have not been explored before as performed in the current study, perhaps because BF% measurements are time-consuming and expensive if using the gold standard approach of DXA. In addition, most studies tend to evaluate the relationship between total BF% and IR or metabolic disease, not considering it as a confounder to adjust or explore the interaction effects between IR and BF%. Our study may advance reasonable proposals to focus on both truncal BF% and IR in the development of hypertension. In addition to BMI, WC, and BF%, a body shape index (ABSI) and waist-to-height ratio (WHtR) serve as valuable indicators of obesity. ABSI was proposed based on a person’s WC and adjusting for their height and weight [39]. Higher ABSI have been associated with increased risks of mortality [39, 40], and hypertension [41]. Our findings indicate that elevated ABSI is a risk factor for hypertension, with a stronger association observed in males compared to females (Supplementary Table 1). However, in our study, there was no significant association between WHtR and hypertension after adjusting for traditional confounders (Supplementary Table 2). Future research should consider analyzing WHtR trends as a marker of changing central adiposity.

Our study may provide additional insights into the association between different obesity indicators and IR on hypertension risk. To our knowledge, this is the first study to consider that BF% has a confounding effect on the relationship between IR and the risk of hypertension, and we found for the first time that BF% mediates this relationship, especially in women with high truncal and leg BF% and high IR. However, the present study has certain limitations. First, as it was cross-sectional, we cannot assert causality, and the results require a longitudinal cohort study for verification. Second, BF% was measured using BIA instead of the gold standard approach of DXA. This might have introduced some bias. Nevertheless, existing research has shown a strong correlation between BIA-derived BF% and DXA in the general adult population [42]. Similarly, BIA-estimated WC shows a high correlation with direct manual measurement [43].Overall, BIA estimation of body composition is more suitable for large epidemiological studies where it can facilitate rapid screening of body composition metrics. However, BIA measures are dependent on several factors, including age, gender, ethnicity, overweight/obesity conditions, and the environment. Therefore, BIA measurements should be based on specific BIA equations tailored for different populations in the studies [44]. Our BIA devices, purchased from South Korea, are suitable for use in Asia and have specific calibration equations for the Chinese population. Third, as this study was performed in a Chinese rural population, the participants were relatively old (approximately 45% of participants were 60 years or older), and 23% had abnormal glucose tolerance; therefore, the generalizability of the study results to other ethnicities might be questionable.

Conclusions

Our findings indicate that BF% modifies the association between IR and increased risk of hypertension in women with high truncal and leg BF%, but not in men.