Introduction

The APOA5 gene, encoding apolipoprotein (apo) A-V, is one of the major genetic determinants of circulating triglyceride levels, as demonstrated in animal models [1]. In humans, both single-gene and genome-wide association studies in different populations have confirmed the importance of APOA5 in affecting triglyceride concentrations [2, 3]. Circulating triglyceride levels, which are also affected by overweight/obesity and insulin resistance [4], are a risk factor for atherosclerosis [5] and can initiate atherogenic dyslipidemia [6].

In clinical practice, pulse wave velocity (PWV) is widely used to reflect arterial stiffness, an index of atherosclerosis [7, 8]. Brachial-ankle PWV (baPWV) measurements, which are easier to perform than carotid-femoral PWV measurements, have been used as a marker to screen for vascular damage and cardiovascular risk in the general population [7, 9]. Deterioration of arterial stiffness has been associated with hospitalization for various forms of new-onset cardiovascular disease, such as heart failure [10]. A cardiovascular outcome resulting from hypertriglyceridemia is arterial stiffness, a pathological condition associated with vascular damage [7]. An inverse relationship between plasma triglycerides and plasma apo A-V and between hepatic APOA5 expression and plasma triglyceride levels has been described in an animal-model study [1], whereas an unexpected and significant positive correlation between circulating apo A-V and triglyceride levels in men and women has been demonstrated [11, 12]. This difference might be because most human studies examining the relationship between apo A-V and triglycerides are cross-sectional.

Recent literature has shown that increasing muscle strength is a positive factor that leads to reductions in arterial stiffness in overweight/obese females [13]. Although improvements in muscle strength may affect arterial stiffness, which increases with age, longitudinal interactions between genetic and environmental factors that might influence the progression of arterial stiffness remain unknown. Thus, studies exploring the longitudinal effect of APOA5 -1131T > C on the age-related progression of arterial stiffness and circulating concentrations of triglycerides and apo A-V are necessary. In this study, we examined whether the longitudinal interaction between APOA5 -1131T > C (TT vs. C allele) and body weight (normal-weight vs. overweight) affects the age-related progression of arterial stiffness and the circulating levels of triglycerides and apo A-V of healthy subjects.

Methods

Subjects

We performed a 3-year prospective cohort study that included 800 heathy subjects (30–65 years old), who underwent triennial medical evaluations at the National Health Insurance Corporation Ilsan Hospital in Goyang, Korea during the period from January 2008 to December 2013. Of these, 503 nondiabetic and nonobese individuals were finally selected. Exclusion criteria included the following: (1) current and/or past history of cardiovascular disease, (2) diabetes mellitus (fasting glucose levels ≥ 126 mg/dL), (3) abnormal liver or renal function, (4) thyroid or pituitary disease, (5) pregnancy or lactation, and (6) regular use of any medication. Before participation, the purpose of the study was carefully explained to all participants, and their written informed consent was obtained. The study protocol was approved by the Institutional Review Board of the National Health Insurance Corporation Ilsan Hospital and Yonsei University and was carried out in accordance with the Helsinki Declaration. Sample size was determined and calculated using the ‘pwr’ package in R software v.3.4.1 (https://cran.r-project.org/). In an exploratory pilot study, the triglyceride level in the normal-weight group was 87.9 ± 48.4 mg/dL (mean ± standard deviation), which was lower than the triglyceride level in the overweight group (129.4 ± 68.7 mg/dL). The sample size was determined via a two-sample t-test calculation with an effect size of d = 0.698, a power of 0.8, and a level of significance of α = 0.05. The results indicated that a minimum of 33 subjects per group were necessary; therefore, we selected more subjects than this minimum to increase the statistical power of our analysis.

Genotyping of rs662799 (APOA5 -1131T > C)

Among the five common APOA5 SNPs that have been studied, -1131T > C and 56 C > G (S19W) are considered to be functional tag SNPs [1, 14, 15]. However, the S19W SNP was monomorphic in the population of this study. Thus, the APOA5 -1131T > C SNP was selected as the functional SNP to investigate further. Genotyping was performed via SNP-IT™ assays using single primer extension technology (SNPstream 25 K™ System, Orchid BioSciences, NJ). Detailed methods for the genotyping have been previously described [16].

Clinical and biochemical assessments

Detailed information on the clinical and biochemical assessments have been described in a previous paper [17]. Body weight, height and waist circumference were measured, and body mass index (BMI) was calculated in units of kilograms per square meter (kg/m2). Systolic and diastolic blood pressure (BP) was assessed in a supine position after a resting period. Blood samples were collected following an overnight fast of at least 12 h. Fasting triglycerides, total, and high-density lipoprotein (HDL) cholesterol, glucose, insulin, low-density lipoprotein (LDL) particle size, and apo A-V were measured as previously described [17]. baPWV was measured using an automatic waveform analyzer (model VP-1000; Nippon Colin Ltd., Komaki, Japan).

Statistical analysis

Statistical analyses were performed using SPSS version 21.0 (IBM/SPSS, Chicago, IL, USA). Hardy-Weinberg equilibrium (HWE) was tested for using PLINK version 1.07 (http://pngu.mgh.harvard.edu/purcell/plink/). Differences in clinical variables between two groups (normal-weight vs. overweight or TT vs. C allele) were tested using independent t-tests. Paired t-tests were performed to determine the differences at the 3-year follow-up from the baseline in each group. The interactions between genotype and bodyweight were tested using a two-way ANOVA. Multiple linear regression analyses were performed to identify major independent predictors of changes in baPWV and triglyceride levels. Pearson’s correlation coefficient was used to examine the relationships between variables. Heat maps were created to visualize and evaluate the relationships among metabolites and the biochemical measurements in the study population. Logarithmic transformations were performed on skewed variables. For descriptive purposes, the mean values are presented as untransformed values. The results are expressed as the means ± standard error (SE). A two-tailed P-value < 0.05 was considered statistically significant.

Results

Distribution of rs662799 (APOA5 -1131T > C) in normal-weight and overweight subjects

We divided the cohort into 2 groups: the normal-weight group (18.5 kg/m2 ≤ BMI < 25 kg/m2, n = 349) and the overweight group (25 kg/m2 ≤ BMI < 30 kg/m2, n = 154). The genotype frequencies did not deviate significantly from Hardy-Weinberg equilibrium (P > 0.05). The frequencies of the minor alleles of APOA5 -1131T > C were 0.297 and 0.315 in normal-weight and overweight individuals, respectively, which are similar to the values observed in the Korean population [18]. Among the 349 normal-weight individuals, 168 were homozygous (TT) for the T allele, 155 were heterozygous for the C allele (TC) and 26 were homozygous (CC) for the C allele of the APOA5 -1131T > C polymorphism. Among the 154 overweight individuals, 68 were homozygous (TT) for the T allele, 75 were heterozygous for the C allele (TC) and 11 were homozygous (CC) for the C allele of the APOA5 -1131T > C polymorphism. We pooled the heterozygotes (TC) and rare-allele homozygotes (CC) to increase statistical power.

Clinical characteristics and biochemical parameters at baseline and the 3-year follow-up

With regard to the APOA5 -1131T > C polymorphism, there were no significant differences between the normal-weight and overweight groups at baseline in their age and sex distributions across the genotypes (Table 1). Additionally, there were no significant differences in smoking or drinking status across the genotypes in the normal-weight and overweight groups at baseline and at the 3-year follow-up (data not shown). In the normal-weight group, there was no APOA5 -1131T > C genotype effect at baseline, but at the 3-year follow-up, a genotype-effect was found in relation to HDL cholesterol: HDL cholesterol levels were lower in normal-weight individuals with the C allele than in those with the TT genotype (Table 1).

Table 1 Biochemical parameters according to rs662799 (APOA5 -1131 T > C) genotype at baseline and 3-year follow-up in healthy individuals who were normal-weight or overweight
Full size table

The overweight group showed higher BMI values, systolic and diastolic BP, and glucose and insulin levels and lower HDL cholesterol levels at both baseline and the 3-year follow-up than the normal-weight group, irrespective of genotype. Overweight individuals with the C allele showed higher total cholesterol levels at baseline and a smaller LDL particle size at both baseline and the 3-year follow-up. After 3 years, overweight C allele carriers showed significant increases in systolic BP and a significant decrease in HDL cholesterol relative to baseline. At the 3-year follow-up in the overweight group, C allele carriers showed lower HDL cholesterol levels and a smaller LDL particle size than TT carriers (Table 1).

Interaction between the APOA5 -1131T > C genotype and body weight in relation to 3-year changes in triglycerides, apo A-V, systolic BP and baPWV

The effects of the APOA5 -1131T > C genotype on mean changes in triglycerides, apo A-V, and baPWV in the normal-weight and overweight groups at 3 years are shown in Fig. 1. In the normal-weight group, individuals with the C allele showed higher serum triglyceride levels than those with the TT genotype. The overweight group showed higher serum triglyceride levels at both baseline and the 3-year follow-up than the normal-weight group, irrespective of genotype. After 3-years, overweight C allele carriers showed significant increases in triglycerides and baPWV. At the 3-year follow-up, C allele carriers showed higher triglycerides than TT carriers in the overweight group. Furthermore, at the 3-year follow-up, overweight TT and C allele carriers showed higher baPWVs than normal-weight subjects with the TT and C alleles, respectively (Fig. 1).

Fig. 1
figure 1

Effect of the APOA5 -1131T > C genotype on changes from baseline for triglycerides, LDL particle size, apo A-V, and baPWV in the normal-weight and overweight groups at 3-year follow-up. Mean ± SE. §tested by logarithmic transformation. aP < 0.05 for the comparison between subjects with the TT genotype and subjects with the C allele in the normal-weight group at baseline. bP < 0.05 for the comparison between subjects with the TT genotype and subjects with the C allele in the normal-weight group at 3-year follow-up. cP < 0.05 for the comparison between subjects with the TT genotype and subjects with the C allele in the overweight group at baseline. dP < 0.05 for the comparison between subjects with the TT genotype and subjects with the C allele in the overweight group at 3-year follow-up. eP < 0.05 for the comparison between subjects with the TT genotype in the normal-weight and overweight groups at baseline. fP < 0.05 for the comparison between subjects with the TT genotype in the normal-weight and overweight groups at 3-year follow-up. gP < 0.05 for the comparison between subjects with the C allele in the normal-weight and overweight groups at baseline. hP < 0.05 for the comparison between subjects with the C allele in the normal-weight and overweight groups at 3-year follow-up. iP < 0.05 for the comparison of changes in values between subjects with the TT genotype and subjects with the C allele in the normal-weight group. jP < 0.05 for the comparison of changes in values between subjects with the TT genotype and subjects with the C allele in the overweight group. kP < 0.05 for the comparison of changes in values between subjects with the TT genotype in the normal-weight and overweight groups. lP < 0.05 for the comparison of changes in values between subjects with the C allele in the normal-weight and overweight groups. *P < 0.05, **P < 0.01, and ***P < 0.001 compared with levels at baseline for the appropriate group by paired t-tests

Full size image

The interaction of gene and body weight did not significantly affect changes in fasting apo A-V (Fig. 1), BP, HDL cholesterol, total cholesterol, glucose, insulin, or LDL particle size (data not shown). At the 3-year follow-up, after adjusting for age, sex, smoking and drinking, the results show a significant interaction between the APOA5 -1131T > C genotype and body weight (overweight compared with normal-weight) in relation to changes in triglycerides (P for interaction = 0.013) and baPWV (P for interaction = 0.047). In the overweight group, there was an APOA5 -1131T > C genotype effect on changes in triglycerides: subjects with the C allele had greater increases in serum triglyceride concentrations than subjects with the TT genotype. However, this genotypic effect on changes in triglycerides was not observed in normal-weight subjects. Additionally, overweight subjects with the C allele had greater increases in serum triglyceride concentration than normal-weight subjects with the C allele. Overweight subjects with the C allele also had greater increases in baPWV than normal-weight subjects with the C allele. Similarly, overweight subjects with the C allele had greater increases in systolic BP than normal-weight subjects with the C allele (Table 1); however, there was no significant effect of a gene × body weight interaction on systolic BP.

Correlation among changes in BMI, systolic and diastolic BP, triglycerides, HDL and total cholesterol, glucose, insulin, LDL particle size, apo A-V, and baPWV

Figure 2 shows correlations among changes (differences from baseline) in BMI, systolic and diastolic BP, triglycerides, HDL cholesterol, total cholesterol, glucose, insulin, LDL particle size, apo A-V, and baPWV values in the normal-weight and overweight groups according to APOA5 -1131T > C genotype. In the normal-weight group with the TT genotype, changes in triglycerides correlated positively with changes in BMI, glucose, and insulin values and negatively with changes in LDL particle size and apo A-V level (P < 0.001, Fig. 3). Changes in baPWV were positively correlated with changes in glucose level and systolic and diastolic BP in the normal-weight group with the TT genotype. In the normal-weight group with the C allele, changes in triglycerides correlated negatively with changes in HDL cholesterol level and LDL particle size and positively with changes in BMI, insulin level and baPWV (P = 0.011, Fig. 3), which was positively correlated with changes in systolic and diastolic BP in the normal-weight group with the C allele. In the overweight group with the TT genotype, changes in triglycerides were negatively correlated with changes in LDL particle size and apo A-V level (P = 0.002, Fig. 3). Changes in baPWV were positively correlated with changes in glucose level and systolic and diastolic BP in the normal-weight group with the C allele. In the overweight group with the C allele, changes in triglycerides correlated negatively with changes in HDL cholesterol and apo A-V levels (P = 0.028, Fig. 3) and LDL particle size and positively with changes in BMI, insulin level and baPWV (P = 0.004, Fig. 3), which was positively correlated with changes in systolic and diastolic BP in the overweight group with the C allele. Other correlations among the changes in clinical and biochemical parameters are shown in Fig. 2.

Fig. 2
figure 2

Correlations between changes (∆) in apo A-V and triglycerides and between changes in triglycerides and baPWV in the normal-weight (○) and overweight (*) groups according to APOA5 -1131 T > C genotype. a Correlation between changes in apo A-V and triglycerides for subjects with the TT genotype. b Correlation between changes in triglycerides and baPWV for subjects with the TT genotype. c Correlation between changes in apo A-V and triglycerides for subjects with the TC or CC genotype. d Correlation between changes in triglycerides and baPWV for subjects with the TC or CC genotype

Full size image
Fig. 3
figure 3

Correlation matrix for changes in clinical parameters in the normal-weight and overweight groups according to APOA5 -1131 T > C genotype. Correlations were obtained by deriving Pearson correlation coefficients. Red indicates a positive correlation, and blue indicates a negative correlation

Full size image

Because the regulation of changes in triglyceride levels is complex, a multiple linear regression analysis was performed that included all subjects (n = 503) to determine the independent effects of the following variables on changes in triglycerides: age, APOA5 -1131T > C genotype, and baseline triglyceride levels, as well as changes in BMI, systolic and diastolic BP and the levels of glucose, insulin, total and HDL cholesterol, and apo A-V. Changes in triglyceride in all subjects (n = 503) were affected by the APOA5 -1131T > C genotype, age, baseline triglyceride level and by changes in BMI and apo A-V level (R2 = 0.258, P < 0.001). In the normal-weight group (n = 349), changes in triglycerides were affected by baseline triglyceride levels and changes in BMI, glucose, apo A-V and systolic BP values (R2 = 0.286, P = 0.034). In the overweight group (n = 154), changes in triglycerides were affected by baseline triglyceride levels and changes in apo A-V levels (R2 = 0.209, P = 0.013).

We also performed a multiple linear regression analysis including all subjects (n = 503) to determine the independent effects of the following variables on changes in baPWV: age, APOA5 -1131T > C genotype, baseline baPWV, as well as changes in BMI and systolic and diastolic BP and in the levels of glucose, insulin, total and HDL cholesterol, and apo A-V and LDL particle size. Changes in baPWV in all subjects (n = 503) were affected by age, baseline baPWV, and changes in systolic BP and triglyceride level (R2 = 0.193, P < 0.001). In the normal-weight group (n = 349), changes in baPWV were affected by age, baseline baPWV and changes in triglyceride and apo A-V levels (R2 = 0.204, P = 0.032). In the overweight group (n = 154), changes in baPWV were affected by changes in systolic BP, LDL particle size and triglyceride level (R2 = 0.413, P = 0.002).

Discussion

Arterial stiffness increases with age, but the interaction between genetic and environmental factors that might influence its progression are unknown. This prospective study examined the longitudinal interactive effects of APOA5 -1131T > C and body weight on circulating triglycerides and the progression rate of arterial stiffness. baPWV, which reflects arterial stiffness [9, 19] was measured at baseline and re-examined within a mean follow-up period of 3 years. The main finding of this study is that the interactive effect between APOA5 -1131C variants and overweight can accelerate the age-related elevation of arterial stiffness through the regulation of circulating triglyceride in healthy subjects.

At 3-years, the overweight group showed higher baPWVs than the normal-weight group, without considering genotype, but only the overweight subjects with the C allele showed an increase in baPWV over the 3-year period. Therefore, the overweight subjects with the C allele had greater increases in baPWV than the normal-weight subjects with the C allele. This result suggests that an acceleration in the age-related enhancement of arterial stiffness can partly result from a longitudinal interaction between APOA5 -1131C variants and overweight. The existence of a strong association between the presence of metabolic syndrome, which includes elevated levels of circulating triglycerides, and arterial stiffness has been shown in many cross-sectional studies [9, 20, 21] and follow-up studies [7, 8, 22]. However, the longitudinal effect of triglyceride-related genes on arterial stiffness in the general population has not been clarified.

There was a significant and weakly positive correlation between the changes in triglycerides and in baPWV in the normal-weight group with the -1131C allele; however, a strongly positive correlation was present between the changes in triglycerides and baPWV in the overweight group with -1131C allele. This is partly due to the greater increases in triglycerides at the 3-year follow-up in the overweight C allele carriers than in those in the normal-weight group. These interactive effects between the APOA5 gene and overweight on circulating triglyceride levels and arterial stiffness support the idea that circulating triglycerides play a role as one of the major inducers of arterial stiffness in overweight/obese subjects [4, 23, 24]. Recently, Guardiola et al. [2]. have shown that APOA5 variants predispose hyperlipidemic patients to atherogenic dyslipidemia and subclinical atherosclerosis and that this predisposition is significantly stronger in subjects with BMI ≥ 25 kg/m2 than in normal-weight subjects. In this study, a multiple linear regression analysis revealed that changes in triglycerides, systolic BP, and LDL particle size were independent determinants of the longitudinal progression of arterial stiffness in the subjects in the overweight groups.

The rare APOA5 -1131C alleles were associated with an atherogenic profile characterized by high levels of small, dense LDL [2]. Atherogenic changes associated with LDL cholesterol induced by triglycerides have been found to occur at concentrations as low as 1.7 mmol/L, which is the level at which small, dense LDLs become predominant [25]. Recently, Vishnu et al [8]. found that baPWV was significantly associated with atherogenic small-sized LDL particles, independent of other cardiovascular risk factors. Takahashi et al [26]. also reported a significant association between very small LDL cholesterol and baPWV. Similarly, in the overweight group in this study, a change in LDL particle size was one of the independent determinants of a change in baPWV. At the 3-year follow-up, overweight subjects with the C allele showed smaller LDL particle sizes and lower plasma apo A-V concentrations than those with the TT genotype.

In all subjects, changes in triglycerides were independently affected by APOA5 -1131T > C genotype, age, baseline triglyceride level, and changes in BMI and apo A-V levels. There were significant negative correlations between changes in triglycerides and plasma apo A-V concentrations in overweight individuals with the C allele and in TT carriers. This is in contrast to previous observations of significant positive correlations between circulating apo A-V and triglyceride levels in men and women [11, 12]. These different results could be partly explained by difference in study design. In general, previous human studies on the relationship between apo A-V and triglyceride levels have been cross-sectional; however, the current study was longitudinal. Moreover, significant negative correlations between changes in triglycerides and apo A-V were observed for subjects with the TT genotype regardless of body weight, whereas for C allele carriers, there was only a significant negative correlation in the overweight group. Subjects in the three groups with strong negative correlations between apo A-V and triglycerides, which included all subjects except for normal-weight individuals with the C allele, showed marked changes in triglycerides and apo A-V during the 3-year study period. However, normal-weight individuals with the C allele did not exhibit substantial changes in apo A-V during the study period despite having more dramatic changes in triglycerides. We can therefore speculate that insufficient changes in apo A-V may not affect correlations between changes in apo A-V and triglycerides. Future analysis would be necessary for further consideration of this interrelationship.

In terms of limitations, our study only included normal-weight and overweight subjects without diabetes or cardiovascular diseases; therefore, the present results may not extrapolate to subjects who are obese, diabetic or have cardiovascular diseases. In addition, we did not fully demonstrate regarding the exact underlying mechanisms of the genetic effects of APOA5 -1131C allele in normal-weight group, thus, the further studies with a large sample size are needed. Despite this limitation, we found that the longitudinal interaction between APOA5 -1131C variants and overweight could accelerate the age-related increase in arterial stiffness through the regulation of circulating triglycerides in healthy subjects. Therefore, the management of hypertriglyceridemia is important for preventing the progression to advanced arterial stiffness, especially in overweight subjects with the APOA5 -1131C allele.