Association of pulse wave velocity and intima‐media thickness with cardiovascular risk factors in young adults

Abstract Pulse wave velocity (PWV), a measure of arterial stiffness, and intima‐media thickening (IMT), a measure of early atherosclerosis, are intermediate markers of cardiovascular disease which are predictive of cardiovascular events. Traditionally, both were thought to result from accumulative exposure to traditional cardiovascular risk factors. However, their association with risk factors in young adults in low‐income settings is unknown. We sought to investigate the association between PWV and IMT with traditional cardiovascular risk factors in the Andhra Pradesh Children and Parents Study cohort from Southern India. Male and female adults (N = 1440) aged between 20 and 24 years underwent measures of PWV and IMT. Exposure variables included smoking, body mass index (BMI), mean arterial pressure (MAP), glucose, homeostatic model assessment of insulin resistance (HOMA‐IR), total cholesterol, high‐density lipoprotein cholesterol (HDL‐cholesterol), and triglycerides. Association between outcome and exposure variables was assessed using linear regression analysis. Average values for PWV and IMT were 5.9 ± 0.6 m/s and 0.5 ± 0.1 mm. In univariable analysis, PWV associated with MAP, BMI, smoking, total cholesterol, glucose, and HOMA‐IR and IMT associated with MAP, BMI, tobacco use, and HDL‐cholesterol. In multivariable analysis, PWV remained strongly positively associated with MAP increasing by 0.5 m/s (P < .001) for a 10 mm Hg increase in MAP (R 2 = .37). In contrast, IMT negatively associated with HDL‐cholesterol (β = −.10; P = .012, R 2 = .02). There was weak evidence that PWV and IMT positively associated with BMI. In young adults from Southern India, PWV positively associated with blood pressure and IMT negatively associated with HDL‐cholesterol. This suggests separate etiologies for atherosclerosis and arterial stiffening in young adults.


| 175
CECELJA Et AL. mortality rates have continued to rise in low-income countries. 3 In particular, India has some of the highest cardiovascular mortality rates in the world reaching an estimate of 349 deaths per 100 000 in men and 265 per 100 000 in women. 3 This rise in cardiovascular mortality is closely associated with demographic shifts, epidemiological transition, and urbanization leading to changing behavior and dietary patterns.
Cardiovascular disease is a chronic disease which develops as a result of exposure to risk factors from childhood and throughout the life course. 4  Traditionally, both IMT and PWV were thought to result from accumulative exposure to cardiovascular risk factors over the life course, explaining their association with cardiovascular morbidity and mortality. 7 While an association between IMT and cardiovascular risk factors is well established, 8 observational data suggest that PWV is primarily influenced by increased blood pressure (BP) and diabetes 9 but not other cardiovascular risk factors. However, these findings are based on data from high-income countries and across a wide range of age groups. A little is known about the association between IMT and PWV to traditional cardiovascular risk factors in low-and middle-income countries and at the early stages of vascular pathologies. To date, most of the studies looking at the association between PWV and traditional risk factors in young adults have been conducted in high-income countries. [10][11][12][13][14][15][16][17][18] Two studies have been conducted in South Africa 19 and Brazil 20 but with a relatively small sample size of <220 participants.
There are currently no studies looking at the association between both PWV and IMT to traditional risk factors in young individuals in low-or middle-income countries, including South-Asian communities.
The aim of the present study was to investigate the cross-sectional association between PWV and IMT with traditional cardiovascular risk factors in male and female young adults aged between 20 and 24 years in the Andhra Pradesh Children and Parents Study (APCAPS) cohort conducted in Southern India.

| Cohort profile
Andhra Pradesh Children and Parents Study is built upon the Hyderabad Nutrition Trial (HNT), Southern India, conducted between 1987 and 1990. APCAPS is a prospective cohort study which aimed to follow-up the children that took part in the HNT trial as previously described and is outlined in Figure 1. 21

| Outcome variables
Pulse wave velocity was measured using the Vicorder device (Skidmore Medical Limited). 21 Carotid and femoral artery pressure waveforms were recorded simultaneously by placing BP cuffs around the neck (30 mm wide cuff) and upper thigh (100 mm wide cuff) with the subjects in the supine position ( Figure 2). The cuffs were inflated to 60 mm Hg, and pressure waveforms were recorded for 3 seconds using a volume displacement method. The foot of the pressure waveform was identified using a cross-correlation algorithm centered at the peak of the second derivative of pressure.
The difference in time between pulse arrival at the carotid artery in comparison with the femoral artery was taken as the transit time.
The difference in distance between the two sites was measured using a tape measure from the upper edge of the femoral cuff (distance 2, Figure 2) to the sternal notch minus the distance between the lower edge of the carotid cuff to the sternal notch (distance 1, Figure 2). PWV was calculated by dividing the distance by the transit time in meters per second (m/s). PWV was measured three times, and the average was used for all further analysis.
For measurement of IMT, carotid artery was visualized using B-mode ultrasonography (Ethiroli Tiny-16a, Surabhi Biomedical Instrumentation India). IMT was measured as the distance between the first and second bright lines which represent the lumen-intima interface and media-adventitia interface, respectively. This technique has been shown to relate closely to measurements made using pathological evaluations 22 with a mean difference of 0.02-0.14 mm between repeat measures. 23 IMT was measured offline using an image analysis package (Carotid Plaque Texture Analysis Software, Copyright LifeQ Ltd). Two measurements of IMT were made by separate operators and averaged, and the mean value of IMT was used for further analysis.
Previous studies have reported good reproducibility of IMT measurement using ultrasound with an intra-and inter-observer variability of 5.4%-5.8% and 10.5%, respectively. 24

| Blood pressure and heart rate
Blood pressure was measured three times, and an average was used for analysis. Measurements were made using an oscillometric device Omron HEM 7300, which has been validated for accuracy according to the British Hypertension Society guidelines. MAP was calculated as [(2 × DBP) + systolic]/3. Heart rate (HR) was measured in triplicate using Omron M5-I (Omron, Matsusaka Co.), and an average of the three measurements was used for analysis. HR was normally distributed.

| Smoking
Information on whether a participant had smoked, chewed, or snuffed tobacco on a regular basis (at least weekly) was gathered as part of a questionnaire. Each participant was categorized according to the status of smoking, chewing, or snuffing tobacco as: current (used in the last 6 months), former (ceased use >6 months ago), or never smoked, chewed, or snuffed tobacco.

| Blood lipids
Total cholesterol was measured using an enzymatic cholesterol oxidase method, HDLcholesterol was measured using an homogeneous assay, and triglycerides were measured using enzymatic colorimetric, GPO-PAP method (Instrument: Synchron CX9, Reagent source: BeckmanCoulter). LDL-cholesterol was not used in analysis as it is derived from total cholesterol and HDL-cholesterol and would introduce multicollinearity in the data analysis (observed correlation between LDL and total cholesterol was r = .94).

| Body mass index
For calculation of participant's BMI, weight (kg) measurements were made in duplicate using a digital weighing machine (Model PS16, Beurer).
Height (mm) was also measured in duplicate with the participant standing using a portable Seca Leicester height measure, Chasmors. BMI was then calculated using the formula weight (kg)/height 2 (m 2 ), where weight and height were averages of the duplicate measures.

| Glucose and insulin
Glucose was measured by an enzymatic method using glucose oxidase/peroxidase-4-aminophenazone-phenol method. Insulin was

| Alcohol consumption
Information on alcohol consumption was gathered as part of a questionnaire. Alcohol consumption was split according to intake of local spirits, branded spirits, wine, and beer and frequency of consumption recorded. For each type of alcohol frequency of present consumption was classified according to the following categories: daily/ most days, weekends only, 1-2 times a month, special occasions, and never. For further analysis, the alcohol type with the highest frequency of drinking was used.

| Data analysis
Data analysis was performed using Stata version 14. A small number of individuals had a history of stroke (n = 1), coronary heart disease (n = 3) and taking medication for diabetes (n = 2), these participants were removed from further analysis giving a total sample size of 1440 participants (Table 1) Univariable linear regression analysis was used to assess the association of PWV and IMT (outcome variables) to cardiovascular risk factor (exposure variable) adjusting for village clustering using robust standard errors using the cluster function within Stata, age, and sex. The number of individuals included in analysis, beta coefficient, 95% confidence interval, P-values (obtained using the Wald test), and coefficient of determination (R 2 ) are presented in the results. Because the final multivariable model included only individuals without any missing data, univariable analysis was repeated only in individuals with complete data to ensure findings were consistent (Table S1 and S2). Linear regression assumes that residuals are normally distributed, and the validity of this assumption was checked using a normal quantile plot. The assumption that there is constant variance among dependent and independent variables was checked by plotting standardized residuals against the fitted values. In this analysis, it is assumed that the

| Participant characteristics
The mean age of all participants was 20.9 ± 1.6 years and 60% of participants were male (Table 1). Average values for IMT and PWV were 0.5 ± 0.1 mm and 5.9 ± 0.6 m/s, respectively, with weak evidence that PWV was slightly higher in men compared to women (mean PWV 5.9 vs 5.8 m/s, P = .09). Average BMI was 19.5 ± 2.8 kg/m 2 . Most participants had never used tobacco (86%), 13% were current smokers and 1% were former smokers with a higher prevalence of current male smokers compared to women (15% vs 10%). Forty-nine percent of participants had never had alcohol and most individual that did drink did so 1-2 times a month (19%) or on special occasions (16%). Fourteen percent drank on

| Univariable analysis
After adjusting for age, sex, and village clusters, there was strong evidence of a positive linear association between PWV and MAP, BMI, total serum cholesterol, triglycerides, glucose, and HOMA-IR (Table 2). There was no association with HDL-cholesterol. There was strong evidence of a negative association between PWV and frequency of drinking and weak evidence of an association with smoking status (P = .057).
There was evidence of a positive association between IMT and tobacco use (P = .017), frequency of drinking (P = .03), BMI (P = .065) and a negative association with HDL-cholesterol. There was no evidence of an association between IMT and MAP, total cholesterol, triglyceride, and glucose.

| Multivariable analysis
The The results remained unchanged in sensitivity analysis that excluded outliers (Table S3).  Table 4).
The beta coefficient was the same as that reported in univariable analysis suggesting that addition of other risk factors did not confound the association between IMT and HDL-cholesterol. There was weak evidence of an independent positive association between IMT and BMI where IMT increased by 0.002 mm (95% CI: 0.00, 0.005; P = .098) for a one unit increase in BMI ( Table 4). The beta coefficient was the same as observed in univariable analysis making confounding unlikely. There was no evidence that IMT was associated with total cholesterol, tobacco use or triglyceride levels within the model. The second model additional included alcohol use. There was no evidence of an association between alcohol and IMT but the non-significant beta coefficients between smoking and IMT was re- Recalculated MAP using only the last 2 blood pressure measurements did not change the association with intima-media thickness.
The results remained unchanged in sensitivity analysis that excluded outliers and when analysis was limited to participants with IMT under 0.90 mm (Table S4 and S5, respectively).

| D ISCUSS I ON
The present study investigated the association between PWV and IMT to cardiovascular risk factors in young adults from a transitioning rural community in Southern India taking part in the APCAPS study. The first major finding is that PWV strongly positively associated with MAP, accounting for 37% of the variability in PWV, but not to other traditional cardiovascular risk factors. There was a modest positive association between PWV and BMI but not with other risk factors including serum cholesterol levels, lipoproteins, triglycerides, tobacco use or impaired glycemia. The second key finding in the present study is that there was a strong negative association between IMT and HDL-cholesterol, accounting for 2% of the variability in IMT, but not with other modifiable cardiovascular risk factors.
The average PWV in the present cohort was 5.9 ± 0.6 m/s which is comparable to that observed previously for the same age group in studies conducted in high-income countries. [10][11][12][13][14][15][16][17][18] Our findings of an association between PWV and BP are also consistent with previous studies in young adults in high-income countries [10][11][12][13][14][15][16][17][18] and in one study conducted in Brazil. 20 This suggests that the process of arterial stiffening in young individuals may be comparable between low-and high-income countries. High BP may directly increase arterial stiffness by transferring stress from more compliant elastin fibers to stiffer collagen fibers at higher pressure. 26 Alternatively and/or concomitantly, increased large artery stiffness can causally increase BP, through reduced buffering of the pulse pressure increase following left ventricular contraction. 27 Due to the cross-sectional nature of the present study, the direction of the relationship between PWV and BP could not be ascertained. However, a recent longitudinal study conducted in over 4000 Chinese adults showed elevated PWV to precede development of isolated systolic hypertension after adjustment for baseline BP. 28 The present study also found a modest positive association between PWV and BMI. This association has been previously observed in cross-sectional [11][12][13][14] and longitudinal studies. 29 However, PWV measurements may be biased when distance measurements are measured over body surface area as any abdominal obesity may overestimate the distance or vice versa.
In support of this, a recent study found a positive association between BMI when distance was measured using surface measurement. However, this association disappeared once aortic distance was measured using magnetic resonance imaging (MRI). 30 Lack of association between PWV and measures of impaired glycemia in multivariable analysis is surprising as previous studies in adults have reported an association between PWV and diabetes. 9 However, in these studies the reported association was relatively weak and accounted for 5% of the variability in PWV which may not be detectable in this young cohort of healthy individuals with low levels of glucose. 9 In support of this, cross-sectional analysis of young adults with type I diabetes or those classified as pre-diabetic did not have elevated PWV compared to healthy individuals. 13,14 The lack of association between PWV and other traditional cardiovascular risk factors including smoking and measures of dyslipidemia is consistent with findings from cross-sectional 9 and longitudinal studies 31 conducted in adults. Taken together this suggests that the process of arterial stiffening is separate from the pathological process of atherosclerosis and is consistent with a lack of correlation between measures of atherosclerosis and stiffening measured along the same vascular region. 32,33 The second key finding in the present study is that there was a strong negative association between IMT and HDL-cholesterol, although this accounted for a small proportion of the variability in IMT (2%). This is consistent with some studies from high-income countries in young adults, 34,35 but not all. 17,36,37 A meta-analysis that included >21 000 adults reported a negative association between HDL-cholesterol and IMT, after adjustment for other atherosclerosis risk factors. 38 Furthermore this association was TA B L E 2 Univariate linear regression analysis of the association of PWV and IMT (per 10 mm) to cardiovascular risk factors adjusted for age, sex, and village level clustering (analysis is limited to participants with complete data with each line representing separate models)  A lack of association between triglyceride and IMT observed in the present study is consistent with the findings from previous studies in the same age group from high-income countries. 17,34,35,37,40 In addition, the present study found no association between IMT and either BMI or tobacco use. In the present study, the range of BMI was 16.7-22.9. Previous studies from high-income countries where BMI was associated with IMT had larger BMI range with a mean of 26.8 in the Muscatine offspring study, 36 24.7 in the ARYA study 17 and 28.0 In the Bogalusa Heart study. 35 This suggests the absence of an association between BMI and IMT may be explained by a lack of exposure to obesity in this cohort. Consistent with previous studies, tobacco use did not associate with IMT in the present study. A lack of association between IMT and smoking may be due to the relatively recent onset of smoking. 36 There was no association between IMT with BP and impaired glycemia. High IMT may results from development of fatty streaks within the intima of the blood vessel wall as part of the atherosclerotic process. In addition, vascular smooth cell hypertrophy within the media may also explain increased intima-media thickness. Smooth muscle cell hypertrophy is primarily caused by elevated BP 41 and could explain the observed association between IMT and BP which has previously been reported. 17,34,36,37 In the present study, a lack of association between IMT and BP suggests that variation in IMT may primarily be due to intimal changes as part of the atherosclerotic process rather than vascular remodeling due to raised BP. The present study found no association between measures of impaired glycemia and IMT. It is well established that individuals with diabetes have a higher risk of cardiovascular events compared to non-diabetic individuals. However, the association between measures of impaired glycemia in non-diabetic patients to cardiovascular events is inconsistent 42 and in line with the findings of the present study.

| Strengths and limitations
The present study has several methodological strengths which should be noted. Firstly, to the best of our knowledge this is the first study to investigating the association between PWV and IMT to tra- individuals (85%) drinking only 1-2 times a month (of these 49% were completely abstinent).

| CON CLUS ION
In the present cross-sectional study including male and female young adults from Southern India we found strong evidence of an (a) positive association between PWV and BP which accounted for 37% of the variability in PWV and (b) negative association between IMT and HDL-cholesterol which accounted for 2% of the variability in IMT.

CO N FLI C T O F I NTE R E S T
None.

AUTH O R CO NTR I B UTI O N S
MC contributed to design of work, analysis and interpretation of data, and drafting of manuscript. RS contributed to design of work and acquisition of data. BK contributed to design of work and