Introduction

Atherosclerosis: Evidence of origins in youth

Vascular aging begins at conception, and precursors of atherosclerosis are present as early as fetal life [1, 2]. Atherosclerosis is a chronic process that begins early in life and accelerated by adverse genetic and acquired atherosclerotic cardiovascular disease (ASCVD) risk factors such as obesity, hypertension, dyslipidemia, and insulin resistance. An autopsy study of 204 young persons, 2–39 years, showed that as the number of ante-mortem ASCVD risk factors increased, so did the severity of asymptomatic coronary and aortic atherosclerosis [3].

Table 1 provides ASCVD risk factors as updated by the American Heart Association [4••].

Table 1 High and moderate ASCVD risk factors [4••]
Full size table

How to Assess the Vasculature for CVD in Youth

Several non-invasive methods have been proposed to assess vascular health in youth, including (1) vascular structure which measures carotid intima media thickness (cIMT), (2) endothelial function which measures metabolic capability of endothelium to release nitrous oxide in response to stimuli, and (3) arterial stiffness which measures the intrinsic “elastic” properties of the arterial wall [5–7].

These measures are defined as “surrogates” as they are noninvasive measurements which are convenient to perform, have shorter imaging time, and do not require use of radioactive intravenous contrast materials. Therefore, they serve as a “proxy” measurements of other gold standard techniques used for CVD risk stratification such as angiography, cardiac MRI (magnetic resonance imaging), cardiac positron emission tomography (PET) scanning, and coronary computed tomographic (CT) angiography.

Non-invasive methods examining structural and functional vascular parameters such as arterial wall thickness, stiffness, and endothelial function are commonly used in adults for risk stratification and prediction of CVD such as myocardial infraction (MI) and stroke. Adult with elevated ASCVD risk factors (n = 2232; 63 years, 58% women) had thicker measurements of cIMT, which predicted stroke and myocardial infarction [8], and elevated baseline pulse wave velocity (PWV) was associated with a 48% in CVD risk even after adjustment for ASCVD risk factors [9]. While these techniques are available in youth studies, validating their utility is limited.

In this review, we provide rationale on how currently available surrogates can help identify the presence of subclinical atherosclerosis and affirm their role in identifying ASCVD risk.

Most measurements of vascular surrogates require an overnight fast and performed after the subject has been supine ≥ 10 min in a quiet room. In this review, we provide information from the literature within the past 5 years for the most commonly used surrogates. Surrogates are rated from those that are the most clinically relevant and useful to those that are less so.

Vascular Structure

**Carotid intima media thickness (cIMT) is the average of the end diastolic intima-media thickness of the right and left common carotid arterial far wall (10 mm proximal to the bulb) using B mode imaging and a 12-MHz linear array transducer, with automatic edge detection analysis technology. Imaging protocols are used in pediatric studies and may not align with the standard assessment guidelines published by American Heart Association (1). Sass et al. studied cIMT in 160 children (10–18 years) and found no correlation with age and sex [10].

In normal-weight children (n = 141; age 12.5 years), Torigoe [11••] reported mean cIMT 0.44 (interquartile range = 0.41–0.47).cIMT has been measured in high-risk groups such as familial hypercholesterolemia (FH) and diabetes with over 400 publication reporting results using this surrogate in the past 5 years (Table 2). cIMT increases across CVD risk factors, with the highest values reported in adolescents with type 2 diabetes (T2D), compared to those with hypertension, high triglycerides/low HDL (insulin resistance), followed by youth with obesity, being significantly higher compared to lean peers [12, 13]. In a 40-year longitudinal study of 3 to 18-year-old Finnish children (Young Finns Study: YFS; n = 1809), low HDL/LDL ratio in childhood and adulthood translated to a greater adult cIMT compared to other ratio trajectories (improving, deteriorating, consistently good) [14]. Studies of obese, T2D, and lean adolescents have shown cIMT progression (average 4.6 years, n = 226) is determined by age, baseline cIMT, glucose, and LDL-C at baseline [15].

Table 2 Relevant studies that used CMIT in disease states which are considered high to moderate risk for ASCVD in over 400 publications were published in the last 5 years
Full size table

Advantages and Disadvantages

cIMT is a safe and reproducible surrogate which can track subclinical atherosclerosis and can also be used to assess effectiveness of interventions to reverse early atherosclerosis. cIMT physiologically increases with age, growth [16–19], male gender [19, 20], pubertal maturation [20], ethnicity [17], and geography [21, 22]. Normal values of cIMT in youth differ by tracing method (point to point versus manual trace versus auto trace) at locations such as common carotid, bulb, and internal and composite cIMT. When manual tracing was used in 275 healthy lean children, cIMT measurements differed by site, the common carotid versus bulb versus internal cIMT. Combined AHA and European Pediatric Cardiology Working Group [23] published normal values but recognized that scanning and readings differed by the side (right versus left carotid or both) and site (common vs. bulb vs. internal vs. composite) which were measured (1).

Recently, semi-automated edge detection programs such as radiofrequency signal tracing quality intima media thickness (RF-QIMT) have emerged as more accurate method since they circumvent B mode image quality and the level of experience of the vascular ultrasonographer. Torkar et al. published mean CIMT using the RF-QIMT in 1137 non-obese normotensive children (mean age 12 + 3.5 years, females n = 625) using age, height, and gender-specific percentile tables, showing cIMT increased with age, height, hip circumference, BMI, and is higher in males [24].

Endothelial Function

**Flow mediated dilatation (FMD) uses a 5–12 MHz linear array transducer and automatic edge detection algorithm to obtain longitudinal end diastolic images of the brachial artery above the antecubital fossa. Images are obtained 1 min before blood pressure cuff inflation, during the 5-min inflation (50 mmHg > systolic blood pressure to induce ischemia) and 5-min post deflation (usually at 60, 90, and 120 s). FMD is calculated as the percentage increase in diameter after reactive hyperemia compared to baseline (1). FMD reaches a peak at 10 years-of-age and declines after 60 years-of-age, being affected by age, BMI, and resting blood flow [25]. Torigoe [11••] showed in normal weight children (n = 141; age 12.5 years) a change in FMD% of 7.1% (IQR 5.2–9.7).

Studies

Table 3 shows some key studies among over 100 articles published in the last 5 years of disease states which are considered high to moderate risk for CVD. In healthy adolescents, a lower apolipoprotein B/apolipoprotein a ratio was associated with lower FMD measured 21 years later, a finding which was independent of LDL-C levels at baseline [26, 27]. In youth with FH (n = 60; 7–17 years), there is lower mean FMD compared to healthy controls although there is wide variability [28]. Urbina et al. showed FMD to be significantly different between lean (n = 241) versus obese (n = 234) versus T2D (n = 195) in youth aged 10–24 years (6.97 ± 1.19 vs 5.51 ± 1.04 vs 5.28 ± 1.00: all p < 0.05) [29]. Youth with T2D (n = 60) have reduced FMD (5.2 ± 0.9 vs. 6.1 ± 1.2%/mmHg: p < 0.01) compared to those with T1D (n = 535) [30].

Table 3 Relevant studies that used flow-mediated dilatation in disease states which are considered moderate to high risk for ASCVD in over 100 articles published in the last 5 years
Full size table

Peripheral Arterial Tonometry (PAT: Non-Ultrasound Measure)

An appropriate finger cuff is placed on the index finger of each hand and inflates to sense blood volume at baseline and after ischemia (cuff inflated for 5 min), and the difference was used to derive the reactive hyperemia index (RHI). Laser flow Doppler (LFD) measures microvascular function and correlates with FMD [31].

Advantages and Disadvantages

All endothelial measurements (FMD, PAT, and LFD) require expertise and operator training as repeat measurements can vary by 5–10%, have biological variability [32, 33], and are affected by fasting versus non-fasting state, temperature, position of probe, and physiological hemodynamic changes such as menstruation in females [34, 35]. FMD when used as the primary surrogate often does not show measurement differences between the ASCVD risk and control group, and the need to correct for shear rate (mean velocity at baseline) is unclear as the associations between ASCVD risk factors, and FMD may be attributable to reduce stimulus for dilation rather than impaired response to hyperemia (9). Other investigators choose not to correct for shear rate as they state that the relationship between shear rate and FMD is weak, nonlinear, or varies between samples, and that ratio normalization should not be applied [36].

***Arterial stiffness describes the stiffness of the arterial wall which is modulated by the integrity of the extracellular matrix as well as the tone of the smooth muscle tone (endothelial function) in the vessel wall. Arterial stiffness measures the rate at which a pressure waves along a vessel segment. The American Heart Association [37] has published a recommendation of standardizing vascular research dealing with arterial stiffness.

Arterial stiffness can be measured using MRI (higher cost, requires access to a scanner), ultrasound measures (arterial distensibility of carotid, aorta and brachial vessels) or echocardiography (pulse wave velocity), or non-imaging modalities like pulse wave velocity (PWV) and augmentation index.

Carotid stiffness (ultrasound) is the baseline 2D image measurement of the artery followed by M mode cursor to measure minimum and maximum stiffness of the carotid artery. Instead of manual measurements, a radiofrequency wall tracker can be used.

Pulse wave velocity (PWV) records electrocardiographic-gated waveforms between carotid and femoral arteries (cf PWV: stiffness of central arteries) or carotid and radial arteries (cr PWV: stiffness of peripheral arteries). The distance between the two points is measured and PWV = length/ transit time in m/s, with higher PWV reflective of stiffer vessels. Tonometry uses pressure sensors (hand held, clamps, or cuffs) with applanation tonometer (Sphygmocor AtCor Medical, Australia) being the most widely used. Published reproducibility of PWV using with SphygmorCor by Bland Altman analysis is excellent [38]. Mechanotransducer-based techniques (Complior, ALAM Medical France) and pulse pen (DiaTecne, Milan Italy) are devices used widely in Europe. Volumetric, photoplethysmographic, and oscillometric methods are also used. Torigoe et al. (11) showed in normal weight children (n = 141; age 12.5 years) cf PWV is 5.2 (4.4–5.6 m/s) and cr PWV 6.8 (6.2–7.5 m/s).

Table 4 lists key studies among the over 500 PWV publications within the last 5 years. The number of ASCVD risk factors (high LDL-C) present in childhood is associated with increased PWV measured at 30–45 years [39]. When obese and lean adolescents were compared a high TG/HDL ratio reflected more atherogenic particles, and an increased ratio was associated with greater PWV in obese youth [40, 41]. Wadwa et al. reported in the SEARCH study that youth (age 10–23 years) with T2D (n = 60) have greater impairments in PWV than age-matched youth with T1D (n = 535 PWV: 6.4 ± 1.3 vs. 5.3 ± 0.8 m/s: p < 0.01) [30]. Studies in adolescents with FH have shown that LDL-C, LDL sub particle size, and number were not independent predictors of PWV, though HDL-size did appear to be a predictor, independent of ASCVD risk markers [42]. Overall, determinants of PWV progression over time include adiposity, BP, glycemic control, and LDL [43•].

Table 4 Relevant studies that used PWV as surrogate in disease states at moderate to high risk for ASCVD in over > 500 articles published in the last 5 years
Full size table

Augmentation index (AI) and central waveform are also measured. Pressure peaks created by the antegrade waveform (from the left ventricle) and retrograde (reflected from the periphery to the center) create the augmentation pressure; AI is derived as the augmentation pressure (difference in the amplitude of the two waves)/pulse pressure (difference between systolic and diastolic central arterial pressure) × 100 adjusted to a heart rate of 75 beats/min [37]. Healthy flexible vessels will return waveform slowly arriving late in the cardiac cycle while stiff vessels return the waveform during early systole augmenting central aortic pressure [44]. The augmentation index adjusted to 75 beats/min (AIx 75) in study of 141 children and adolescents (age = 12.5 years (6.7–15.2 years); 48% male, weight 43.4 kg (23.9–57) was − 5 (− 13 to 5). AIx 75 was significantly different between lean (n = 241) versus obese (n = 234) versus T2D (n = 195) aged 10–24 years (− 0.5 ± 10.8 vs 2.7 ± 11.6 vs 6.6 ± 11.3: all p < 0.05) (29). Recent follow up of adolescents with T2D (Treatment Options of Adolescents and Youth: TODAY n = 304, 34% male; duration of diabetes 8 years) reported indices of PWV and AIx (75%) worsened over the 5-year assessment interval (cf PWV 6.3 to 7.1 m/s; cr PWV 7.8 to 9 m/s; AIx 9.4 to 14.9), and these changes tracked with higher BP measurements and increasing levels of HbA1c [45].

Advantages and Disadvantages

Longitudinal cohorts such as the Cardiovascular Risk in Young Finns have demonstrated higher PWV in adulthood with clustering of ASCVD in childhood [46]. Investigators have found that arterial stiffness increases with age [47, 48] which is related to the vessel wall size and distensibility of the vessel wall, starting around age 2.3 years-of-age and plateauing around age 15 years as measured using MRI imaging [49]. Fischer et al. found sex differences in PWV both before and after puberty [48] while in contrast, Voges found none [49]. PWV also changed by race and sex in a study of 559 young adults (mean 22 years) [50]. Factors confounding the measurement of the arterial stiffness are mean arterial pressure and heart rate recorded at the time of measurements [51, 52]. Measurements should be conducted in a quiet temperature-controlled room, performed after participants have supine for 10 min, caffeine avoided for 2–4 h, and in female subjects at the same stage during their menstrual cycle. These methodological confounders should be considered when comparing indexes obtained from different vascular beds (carotid femoral; cf versus brachial; ba PWV). Selective devices (SphycomoCor versus Complior) report 5–15% differences in values related to the algorithm used. There is also a lack of agreement of what constitutes normal PWV in youth when comparing oscillometric device (single point estimate) which tend to be lower, versus results utilizing a tonometric device. Brachial ankle PWV is reported in cm/s rather than m/s, but results are much higher than carotid femoral PWV (n = 262 adolescents; 12–18 years) [53]. Tonometric devices report similar results, and some have been cross validated [1, 54] [11••, 55]. The most important confounder is the distance (surface measurements between the recorded sites) which is often performed using a tape measure (calipers being preferred) and can account for up to 30% difference in the PWV calculation. Operators performing arterial stiffness require training and should be able to demonstrate consistently reproducible results.

*CAC

Agatston et al. used helical and spiral CT measurement of microcrystalline calcium within the lipid core to calculate a volume composite, which was calculated as coronary artery calcification (CAC). This modality has not been extensively used to study early atherosclerosis in youth, due to concern for radiation exposure, except in a few studies for Kawasaki disease [56], FH [57], and diabetes [58]. There exists no correlation with other vascular surrogates, and no threshold exists about the amount calcification by age, ethnicity, and sex. CAC is valuable as a surrogate for advanced vascular lesion; thus, it has a limited role in youth.

Issues with Measuring Vascular Surrogates in Youth

To date, the literature for measuring vascular parameters has been limited by the lack of references and the lack of standardization of the equipment and protocols used across centers. Vascular surrogates have a complex relationship with age and anthropometric variables as there are physiological changes in the arterial wall during periods of rapid somatic growth such as infancy and adolescence. Torigoe et al. [11••] studied the most common used vascular studies in 292 healthy children (0–18 years) and proposed pediatric reference values using multivariate regression models adjusting for the effect weight, height, and age.

Conclusions

Measurement of vascular health is important in primary prevention, and in those with subclinical disease, a helpful guide to determine the need for more aggressive treatment in high-risk ASCVD groups. Use of vascular surrogates in longitudinal studies has independently predicted cardiovascular end points such as MI, stroke, and death. Over the last few decades, several devices and different approaches have been used which have created challenges but also provided opportunities to better delineate the utility of surrogates in children and adolescents. There exist gaps in our understanding of surrogates in children and adolescents.

  • Lack of validation of measurement methods in children. Can dedicated devices be identified which are reliable non-invasive tools? This will require validation studies in youth undergoing simultaneous invasive catherization or MRI-based imaging modalities. Alternatively, it may be reasonable to validate studies using dedicated devices that have been widely used in prospective trials such as PWV measurements by SphygmoCor.

  • Lack of normative data by age/body, size/puberty stage, sex, and race. In the last 5 years, several studies have been performed by single centers to provide age, sex, and gender-specific normative data [11••, 24]. A comparison of different cohorts demonstrates high heterogeneity beyond ethnic and geographical factors limiting the generalizability of results. The American Heart Association [59], European Pediatric Cardiology [23], and American Society of Echocardiography [17] recognize the value of surrogates to provide actionable information on CVD risk especially in pediatric populations with moderate to high-risk ASCVD risk conditions. Such organization (1) needs to analyze the currently available literature and provided updated consensus statement on the acceptable imaging protocols, devices, and how to account for the confounding variables such as age, gender, sex, and ethnic differences. To date, in the USA, there is no normative data on vascular surrogates in children based on racial and ethnic differences. Therefore, such guidelines will serve as valuable resource for investigators planning to use surrogates in prospective studies.

  • Longitudinal data in children with high-risk ASCVD in the in Bogalusa study, Young Finns, and Avon Longitudinal Study of Parents and Children (ALSPAC) have shown over the past several decades that these surrogates can predict ASCVD outcomes. CIMT and PWV have predicted hard CV events, although normative data across pediatric populations are lacking. Endothelial function testing has high variability. With consensus and better reference values, surrogates will increasingly become standardized and offer a valid and reliable tool for ASCVD risk stratification, and to assess the effect of interventions in high-risk ASCVD groups