Associations Between High Triglycerides and Arterial Stiffness in a Population-Based Sample: Results from the Kardiovize 2030 Study

Background The term arterial stiffness (ArSt) describes the reduced capability of an artery to expand and contract in response to pressure changes and is recognized as an independent predictor of cardiovascular diseases (CVD). The evidence relating ArSt and triglycerides (TG) shows contradictory results. The aim of this paper is to assess the association between high TG and ArSt, using the cardioankle vascular index (CAVI). Methods Subjects aged between 25 and 64 years from a random population-based sample were evaluated between 2013–2016. Data from questionnaires, blood pressure, anthropometric measures and blood samples were collected to identify the metabolic syndrome (MetS). CAVI was measured using VaSera VS-1500N devise (Fukuda Denshi Co., Ltd., Japan). Subjects with history of CVD or chronic renal disease were excluded. Results 1934 participants, 44.7% males, were included. Median age was 48 (Interquartile Range [IQR] 19) years, TG levels were 1.05 (0.793) mmol/L, and CAVI 7.24 (1.43) points. Prevalence of high CAVI was 10.0% (14.5% in males and 6.4% in females; p < 0.001). Correlation between TG and CAVI was 0.136 (p < 0.001). High CAVI values were more prevalent among participants with MetS, high blood pressure, dysglycemia, abdominal obesity, and high total cholesterol. Using binary regression analysis, high TG were associated with high CAVI, even after adjustment for other MetS components, age, gender, smoking status and cholesterol (OR = 1.630, 95% CI = 1.061–2.505, p = 0.026). Conclusion TG levels were correlated with ArSt, measured as CAVI. High TG was associated with high CAVI independent of multiple cardiometabolic risk factors.


Aim
Aim of this study is to evaluate the association between high TG and ArSt, measured by CAVI, adjusting for MetS components and other traditional risk factors, in a random population-based sample of European adults.

Methods:
Design and Population: The study design, sampling, and implementation were described previously [26]. In brief, Kardiovize Brno 2030 is a prospective population-based study with a random strati ed by sex and gender sample of 25 to 64 year-old residents of the city of Brno [26]. The recruitment and core baseline examinations were completed in 2014. Follow-up will be carried out regularly at 5-year intervals, the rst one is being conducted and expected to in 2020, It is envisioned that the study will run until 2030 [26].
Brno is the second largest city in the Czech Republic (after Prague). As of 1 January 2013, Brno had a general population of 373,327 residents [27]. The study aimed to enroll 1% of the adult population of Brno randomly selected and strati ed by sex and age. Eligibility criteria included permanent residence in Brno, and registration (required by the law) with any of the ve state-run health insurance companies operating in the Czech Republic [26].

Sampling and Recruitment:
Survey sampling was done in January 2013 with the technical assistance from the largest (state-run) health insurance company using the registries of all health insurance companies, except one that declined to cooperate (thus excluding 8.9% of the population). A random strati ed sample of 3300 persons adjusted for a response rate of 64.4% was drawn from the registries. The health insurance companies mailed invitation letters with a description of the study goals while ensuring the con dentiality of personal information. The invitation letters were mailed in January 2013, with two reminder mailings. Following the same procedure, another random sample of 3077 was selected in April 2014; the study target of 1% of the adult urban population was met on 19 December 2014. Based on the two samplings with a total of 6,377 randomly selected invitees, the overall response rate was 33.9%. No information on non-respondents was available due to con dentiality restrictions. Despite the relatively low response rate, the number of enrolled participants was large enough to ensure optimal representativeness of sociodemographic strata in the sample [26]. A total of 2160 individuals were enrolled in the study and 226 records were excluded from the present analysis due to either missing information on ArSt or MetS components, or presence of self-reported history of CVD, de ned as stroke or chronic ischaemic disease, or chronic renal disease. Figure 1 summarizes recruitment of the participants and further exclusion of non-eligible records.

Data Collection:
The baseline health assessment face-to-face health interview, and comprehensive questionnaire was performed by trained nurses and physicians at the International Clinical Research Center of the St Anne's University Hospital in Brno, who also entered the collected data into the web-based research electronic data capture (REDCap) database. The questionaire included demographics, socioeconomic status (age, gender, education, household income, and occupation), cardiovascular risk behaviors (smoking status, nutrition, alcohol consumption, and physical activity), history (family and personal, medications, and hospitalizations), mental health (depression, stress level) and a food frequency questionnaire. CAVI was measured using VaSera VS-1500N device (Fukuda Denshi Co., Ltd., Japan). Laboratory analyses were performed with 12-hour fasting blood samples. TG levels were assessed by a well-established enzymatic calometric method (Roche Diagnostics GmbH, Germany), using a Modular SWA P800 analyzer (Roche, Co. KG, Germany) and manual tape measurement of waist, hip, and neck circumference [26].
De nition of Variables: MetS was de ned as simultaneous presence of 3 or more of the following components: elevated TG level ≥ 1.7 mmol/l, or treatment with brates or nicotinic acid; low HDL level (< 1 mmol/l in men and < 1.3 mmol/l in women), or treatment with brates or nicot,inic acid; previously diagnosed diabetes Results:
CAVI is de ned as stiffness derived from the β coe cient, obtained from the Bramwel-Hill equation: Where PWV is pulse wave velocity, ρ is blood density, Ps and Pd are systolic and diastolic BP values in mmHg [4]. High CAVI group was de ned as those subjects with ≥ 9 points and normal as those with CAVI < 9, based on the presence of advance arteriosclerosis [4][5][6][28][29][30][31][32]. Levels of carbon monoxide in the expired breath < 10 ppm were compatible with the de nition of non-smoker.

Statistical analysis
Analyses were performed using the SPSS software (SPSS, version 23.0, Armonk, NY: IBM Corp.).
Kolmogorov-Smirnov test was conducted in order to assess the normality of the continuous variables. Variables were non-normally distributed and presented as median (interquartile range), their differences were evaluated using Mann-Whitney U test. Correlation Spearman analysis between TG and CAVI was assessed. Proportions were presented as percentage and differences were determined by χ 2 test. Univariate analysis was used to assess risk factors related with CAVI as a binary outcome (High ≥ 9 or normal 9) and was presented as OR and 95% CI. A multivariate analysis was done using CAVI as a binary outcome and TG as a dichotomic variable (high ≥ 1.7 mmol/l and normal 1.7 mmol/l), and adjusted by multiple confounders (See Directed Acyclic Graph in Supplementary les). In model 1 by age and gender, and in subsequent models by high waist circumference, elevated fasting glucose, systolic and diastolic BP levels, HDL values, smoking status, and total cholesterol. In multicollinearity test with CAVI as outcome, the values of the variance in ation factors in theses models were 4.0. Statistical signi cance was considered p < 0.05. Data are presented as median (Interquartile range) and differences were assessed using Mann-Whitney U test.

Associations between CAVI and Triglycerides
The prevalence of high CAVI was 10.0%, higher in men than women, 14.5% and 6.4% (p < 0.001), respectively (  High CAVI (%) OR 95% CI p Three models were created to assess the association between high TG and high CAVI (Fig. 2). Discussion: To the best of our knowledge, this may be the rst analysis speci cally designed to address the question if high TG are associated with ArSt, where previous reports show contradictory results. Using a randomnly selected large population-based sample, high TG (≥ 1.7 mmol/l) increased the odds of having high CAVI (≥ 9) by 63%, independed of multiple confounding variables as age, gender, MetS components, total cholesterol, and smoking habits. The prevalence of high CAVI was 10.0% and was associated with male gender, higher age, high BP, dysglycemia, abdominal obesity, and total cholesterol, but not related to smoking and low HDL-c.
On the contrary, in 18 countries from Europe [19], assessing 2224 subjects, aged 40 and older, PWV was higher in subjects with MetS compared with those without (9.57 ± 0.06 vs 8.65 ± 0.10; p < 0.001), but CAVI was similar in those two groups (8.34 ± 0.03 vs 8.29 ± 0.04; p = 0.40). In the multivariate analysis, PWV was positively correlated with age, BP, glucose and HDL-c, but not with waist circumference and TG; CAVI was positively correlated with age, gender, BP, glucose, but not with TG and HDL, and negatively correlated with waist circumference. Authors don't provide a clear explanation with contradictory results relating waist circumference and CAVI [19]. In a prospective evaluation of 2106 middle aged subjects with MetS from Lithuania [17]. high CAVI values at the baseline was related with higher risk for CVD events after around four years. At the baseline, high CAVI values were related to worse cardiometabolic pro le, but not with TG value (p = 0.891) [17]. In Korea, in 1144 adults, older than 18 years from Gyeonggi [12], assessing the association between MetS and CAVI reported that CAVI was independently related with age, sex, diastolic BP, and uric acid, but not with waist circumference, plasma glucose, HDL-c, and TG [12], In two Chinese population studies [13,20], TG were signi cantly correlated with CAVI, but this association disappeared after multiple adjustments.
Discrepancies between results of the studies might be partially explained by the differences in the population sample sizes, inclusion criteria for the subject's recruitment, way of ArSt quanti cation or the variety of adjustment variables. Studies, performed on large population samples tended to observe positive association between ArSt and TG, however, some of them used PWV as an ArSt marker [8,9,24]. Also, studies that failed to nd an association between TG and ArSt, were often conducted on the population samples with MetS [17,19] or diabetes [18], meanwhile studies that are indicating positive relationship included mostly healthy subjects [9,22,25]. More prospective studies are needed, in order to clarify the risk of elevated TG and it's effect on the arterial wall state.
The whole spectrum of possible underlying pathophysiological mechanisms of the in uence of lipid pro le on ArSt has not been well established yet. However, abnormal lipid pro le simultaneously in uences several pathways -development of atherosclerotic plaques, oxidative stress, in ammation enhancement, endothelial dysfunction and low availability of nitric oxide [33]. From the point of view of atherosclerosis and CVD, there are four main mechanisms which can indirectly increase CVD risk. First, hydrolysis of postprandial chylomicrons or endogenously formed VLDL leads to further formation of cholesterol-rich remnants, which can enter the subendothelial space through the scavenger receptors and promote formation of the foam-cells [23]. Second, higher Apolipoprotein (Apo) CIII might also have an impact on the metabolism of TGs, through inhibition of TG hydrolysis and increased formation of dense, oxidation-prone low density lipoprotein particles [9,34]. Liver fat mass was also directly associated with the amount of secreted very low density lipoprotein [34]. Third, high TG might disrupt the mechanism of reverse cholesterol transport [34]. Fourth, in vitro analysis indicates that high TG might also stimulate expression of endothelial mediators, such as endothelin-1, promoting endothelial dysfunction [23].
The main limitation of the present report is that the cross-sectional design doesn't allow to establish causality between TG and ArSt. Independent association between TG and CAVI as continuous variables was not reported, because the assumptions of linear regression analysis were not met. The future prospective results of this study will allow us to examine the predictive value of lipid pro les on ArSt. Binary regression models associating high TG ≥ 1.7 mmol/l and high CAVI ≥ 9. * adjusted by age and gender. ** adjusted by age, gender, high waist circumference ≥80 cm in females and ≥94 cm in males, elevated fasting glucose ≥ 5,6 mmol/l or treatment, systolic and diastolic BP levels, HDL values. *** as model 2, further adjusted by smoking status (4 categories) and TC levels. Abbreviations: BP -blood pressure, CAVI -cardio-ankle vascular index, CI -con dence interval, HDL -HDL cholesterol, OR -Odds ratio, TC -total cholesterol, TG -triglycerides

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.