Cardiovascular diseases (CVDs) are the leading cause of morbidity and mortality globally and the underlying contributor to this burden is mainly atherosclerosis [1]. The pathophysiology of atherosclerosis is well known and progression can be slow, sometimes beginning in childhood and progressing at varying levels of pace into adulthood prior to mediating clinical, ischemic events [2].

A major risk factor for atherosclerosis is visceral obesity that may increase atherosclerosis through metabolic changes which drive the processes of inflammation and fat deposition [3, 4]. Hypertriglyceridemic waist (HTGW) phenotype is a surrogate marker of visceral obesity [5]. Since this phenotype consists of measures of hypertriglyceridemia and elevated waist circumference it can also be used as an inexpensive screening tool to identify individuals at high risk for CVDs. It has been theorized that accumulation of adipose tissue and level of triglyceride present in the blood are chiefly implicated in the risk for atherosclerosis [3]. Though the role of triglyceride levels in the causation of atherosclerosis is still subject of much debate in the scientific community [6]. Nevertheless, HTGW could possibly act as a way to identify individuals at high-risk for subclinical atherosclerosis.

Hypertriglyceridemic waist (HTGW) is primarily a marker of atherogenic metabolic triad which is constituted of hyperinsulinemia, hyperapolipoprotein B, and small LDL particles [7]. The HTGW phenotype has been shown to be a good predictor of increased visceral adiposity [8]. This phenotype has also previously been associated with atherosclerosis [9], abnormal coronary angiography [7, 10, 11], and prospective studies have found HTGW to be a predictor of coronary heart disease [12, 13].

A few studies have previously explored the cross-sectional association between HTGW and atherosclerosis [9]; however, given atherosclerosis is known to develop over time, understanding how this association might change is necessary. Therefore, the purpose of this sub-study was to assess the association between baseline hypertriglyceridemic waist and subclinical carotid atherosclerosis 5 years later using indices of IMT and plaque presence in a multi-ethnic population of men and women with no pre-existing cardiovascular diseases.

A total of 517 of 813 participants (63%) recruited at baseline completed the 5-year follow-up assessment. The main reasons for not completing the study were lost-to-follow-up in which the participant’s contact information was no longer valid and refusal to participate due to lack of interest. Completers were older (p = 0.008), had a lower BMI (p = 0.005), waist circumference (p < 0.001), less likely to have HTGW (p < 0.001), be a current smoker (p < 0.001), be Aboriginal (p < 0.001), but more likely to have a higher education (p < 0.001) and a higher income (p < 0.001) than non-completers (Additional file 1: Table S1).

Table 1 outlines the baseline characteristics of the study participants by HTGW components. Significant differences were observed for ethnicity (p < 0.001), BMI (p < 0.001), total cholesterol (p < 0.001), HDL-C (p < 0.001), systolic blood pressure (p < 0.001) and smoking status (p < 0.001) by HTGW phenotype. The prevalence of current smokers was greater in the HTGW group. Baseline IMT and total area also varied significantly by HTGW whereas plaque presence did not.

Presence of baseline HTGW phenotype was weakly, but positively correlated with 5-year IMT (r_pb = 0.11, p = 0.02), total area (r_pb = 0.13, p = 0.00), and plaque presence (2.06 (1.13, 3.74), p = 0.02). Additionally, a significant inverse association was observed between the presence of elevated TG phenotype at baseline with the 5-year total area (r_pb = − 0.11, p = 0.01) (Additional file 1: Table S2). However, this association was not significant when adjusting for sociodemographic and traditional risk factors in Table 2. When adjusting for sociodemographic factors (age, sex, ethnicity, income level, maximum education) and family history in model A, a statistically and clinically significant association was observed between the presence of HTGW phenotype with average IMT (0.08 [0.04, 0.11], p < 0.001), total area (0.20 [0.07, 0.33], p = 0.002), and plaque presence (OR = 2.17 [1.13, 4.19], p = 0.02) compared to the non-HTGW group (Table 2). Presence of elevated WC at baseline was also positively associated with 5-year IMT outcome (0.05 [0.02, 0.09], p = 0.002) compared to the non-HTGW group. Upper limits of normal for average IMT and total area are roughly 0.9 mm and 18.3 mm², respectively [16]. When adjusting for traditional risk factors of atherosclerosis (BMI, smoking status, total cholesterol, HDL-C, and systolic blood pressure systolic blood pressure) and sociodemographic factors in model B, we found no significance between elevated WC, elevated TG, and HTGW phenotypes with indices of atherosclerosis, compared to the non-HTGW group. We also investigated the change in standardized β for the variables in each linear regression model (Additional file 1: Table S3) and found that compared to model A, HTGW phenotype was not significant in model B but total cholesterol and systolic blood pressure were significant. Total cholesterol and systolic blood pressure were associated with average IMT (std. β = 0.09, p = 0.03; std. β = 0.12, p = 0.003), total area (std. β = 0.19, p < 0.001; std. β = 0.12, p = 0.01), and plaque presence (OR = 1.52, p < 0.001; OR = 1.28, p = 0.05) in model B. Total cholesterol and systolic blood pressure may be largely mediating the association between elevated WC and HTGW phenotypes with IMT, the effect of HTGW with total area, and the effect of HTGW with plaque presence.

Next, we examined whether there was a correlation between change in HTGW phenotypes with change in the indices of subclinical atherosclerosis over 5 years. The status of change in HTGW phenotypes was categorized into four different groups of (1) Without HTGW at Baseline and Follow-up (N = 376); (2) With HTGW at Baseline and Follow-up (N = 49); (3) Acquired HTGW (N = 50); (4) Reversed HTGW (N = 37). We found that all groups had a significant increase in IMT, total area, and plaque presence in proportions that were similar between groups (IMT p = 0.99; total area p = 0.35; plaque presence p = 0.06) (Table 3). The only exception was the group that Reversed HTGW. This group showed no change in the proportion of participants with plaques present (73.0%). Adjustments for sociodemographic and traditional risk factors did not change these findings (Additional file 1: Table S4 and S5).

Furthermore, the regression models A and B were run a second time for all atherosclerosis indices while controlling for the baseline reading of each index (Table 4). We observed no significant associations between any of the baseline HTGW phenotypes with any of the 5-year indices of atherosclerosis when controlling for the baseline readings of each index.

This study assessed the relationship between hypertriglyceridemic waist (HTGW) at baseline with 5-year indicators of subclinical atherosclerosis in a multi-ethnic cohort. Our findings indicate that HTGW phenotype is a statistically significant and clinically meaningful predictor of 5-year subclinical atherosclerosis measures of IMT, total area, and plaque presence independent of sociodemographic factors and family history. However, this effect appears to be mediated largely by known risk factors based on our finding that when adjusting for traditional risk factors (total cholesterol and systolic blood pressure), this association was not significant.

Our findings are congruent with the cross-sectional studies conducted in a diverse range of other asymptomatic populations including Caucasian men [17], multi-ethnic populations (Aboriginal, Chinese, European, South Asian) [9], Canadian Cree [18], patients with chronic kidney disease [19], patients living with type 2 diabetes [8], and subjects with cardiometabolic risk factors [20] that found a positive correlation between HTGW and subclinical atherosclerosis. In the total area and plaque presence models, presence of elevated WC or elevated TG alone did not correlate with higher total area or higher proportion with plaque presence. However, the presence of both of these phenotypes (i.e., HTGW phenotype) was correlated with greater total area and a higher proportion with plaque presence. This observation suggests a synergistic elevation of risk. This theory is in line with previous cross-sectional findings in the same multi-ethnic sample that found a stronger correlation between HTGW with subclinical atherosclerosis than between elevated TG and elevated WC individually with subclinical atherosclerosis [9]. Indeed, the HTGW phenotype has been shown to be associated with increased risk of cardiometabolic diseases [21‐30]. Additionally, multiple studies have reported an elevated risk of diabetes [31‐40] and cardiovascular diseases [12, 41‐46] among individuals with HTGW, highlighting the potential value of this phenotype as a marker of CVD risk factors and poor outcomes.

Change in HTGW phenotypes did not correlate with change in indices of subclinical atherosclerosis over 5 years. Hence, acquisition or reversal of HTGW cannot be used as reliable tool to track one’s risk level for atherosclerosis. Within the group that experienced a reversal in their HTGW phenotype, we observed a 5-year halt in the proportion of participants that presented with plaques. This observation indicates that lifestyle measures and reversal of HTGW may be beneficial. However, we did not observe a halt in the progression of IMT or total area in this group. This leads us to hypothesize that the mechanism for increasing IMT and total area might be different from the mechanism for increasing plaque presence which accounts for the difference in associations. This difference in underlying mechanisms would likely be independent of the variables that were adjusted for in our models as our adjustments did not affect this observation. Furthermore, baseline HTGW phenotypes were not associated with 5-year change in indices of atherosclerosis. This finding indicates that whether a patient has any of the HTGW phenotypes does not impact the progression of their atherosclerosis over 5 years.

The strength of our study lies in exploring the relationship between HTGW and subclinical atherosclerosis longitudinally. The ethnic and sociodemographic diversity in our study population allowed us to establish a significantly positive correlation between HTGW with subclinical atherosclerosis that shows independence from income level, maximum education, and ethnicity. Another strength of this study is in using multiple markers of subclinical atherosclerosis (IMT, total area, and presence of plaques) and utilization of carotid ultrasound which have allowed for a robust and direct measure of atherosclerosis, sensitive enough to detect early atherogenesis.

The original study adopted a purposeful sampling strategy which stratified participants by ethnicity and sex and ensured equal representation from each group and may not be representative of the general population. The ethnicities represented in our sample are Aboriginal, Chinese, South Asian, and European communities. Hence, the findings of this paper may not hold valid in other ethnic groups. Furthermore, our assessments did not take into account the fact that categorical assessments of elevated waist circumference may require different cut-offs by ethnicity. Our sample was also limited to individuals between the ages of 30 and 65 years. Hence, the association between HTGW and 5-year subclinical atherosclerosis outcomes in younger and older age groups might need to be further examined. Nevertheless, the age and ethnic groups represented in our sample are those that have been identified at high risk for CVD and should be targeted for screening practices. Our assessment of the association of HTGW phenotypes with the progression of atherosclerosis was limited by the fact that a significant majority of participants already demonstrated evidence of subclinical atherosclerosis at baseline. Future longitudinal studies assessing participants with no evidence of subclinical atherosclerosis at baseline may be able to provide a better assessment of the association between HTGW phenotypes with development and progression of subclinical atherosclerosis. Furthermore, it is possible that the use of more detailed measurements of IMT, including 3D variables, could have detected more subtle findings. However, the use of well validated methods of carotid IMT in this study suggest that use of other well accepted measures of carotid atherosclerotic burden could not have shown different results. As a sub-study, the present investigation consists of secondary analysis. Future studies powered for the outcome may be warranted. Age and gender-based classifications of IMT into normal versus abnormal categories [47] could provide a more clinically meaningful evaluation of participants. However, stratification of participants by age and gender would significantly reduce our statistical power to detect within and between group differences Hence, we used IMT as a continuous variable while adjusting for age and gender in our models. Due to the significant number of non-completers this study is subject to the possibility of type II error. There is also a possibility of residual confounding and random error in this study. Furthermore, while we did not adjust for medication use, only 4.6% of participants were taking a statin at the 5-year follow-up. Therefore, it is unlikely that the use of statins could have had a meaningful impact on our findings.

Our findings showed that baseline presence of HTGW phenotype in healthy individuals is a significant predictor of 5-year subclinical atherosclerosis independent of sociodemographic factors and family history. However, this effect seems to be largely mediated by known risk factors (total cholesterol and systolic blood pressure) which are often measured at the same time as the elements of HTGW phenotype (TG and WC). Hence, utility of HTGW phenotype as a clinical tool remains unproven. Our results also indicated that the change in presence or absence of HTGW phenotype does not correlate with a change in indices of subclinical atherosclerosis over 5 years. However, in the absence of more long term follow up, data on atherosclerotic burden in other areas such as coronary calcium score, and clinical end points, the utility of monitoring and tracking HTGW cannot be dismissed. Future research should examine the relationship between HTGW phenotype and indices of atherosclerosis among individuals with manifest cardiovascular disease. Another important area for future research is to examine whether there indeed is an alternative mechanism for development of plaque that is distinctively different from IMT and total area. Alternative measurements of hypertriglyceridemic waist or new variables that can be independently measured outside the clinical setting and provide predictive insight, can be highly impactful in identification of at-risk individuals for subclinical atherosclerosis in a way that is cost-effective and adoptable in high numbers.