Risk estimation for coronary heart disease (CHD) has evolved from a single risk factor approach to a multivariable approach (i.e., the Framingham Risk Score) (1- 2). This global risk score aids primary prevention efforts by identifying high-risk patients who may benefit most from drug therapy in the near term. However, this risk prediction algorithm has limitations, classifying most individuals younger than 50 years of age as low risk regardless of risk-factor burden (3- 5). In contrast, differences in risk-factor burden in younger adults can have substantial influence on the remaining lifetime risk of atherosclerotic complications, heart failure, and death (4,6- 9).

Recently, we showed that among individuals younger than 50 years of age with a low 10-year predicted risk, further stratification into 2 groups of similar size is possible: those with low short-term/high lifetime predicted risk and those with low short-term/low lifetime predicted risk (10). Furthermore, those with low short-term/high lifetime predicted risk had a greater burden and progression of subclinical atherosclerosis as measured by coronary artery calcium (CAC) and carotid intima-media thickness (IMT), suggesting that these structural changes mediate the marked differences in lifetime risk of atherosclerotic cardiovascular disease (CVD) (11). Of interest, more than three-fourths of individuals with low short-term/high lifetime risk were CAC negative (10), suggesting the presence of alternative subclinical abnormalities among those with low short-term but high lifetime predicted risk.

Left ventricular hypertrophy (LVH) is an important cardiac phenotype that is associated with adverse cardiovascular outcomes including myocardial infarction and overall mortality (6,12). In exploring potential mechanisms for these associations, we recently demonstrated that increased left ventricular (LV) mass, wall thickness, and concentricity are associated with subclinical atherosclerosis as measured by CAC (13). Due to this association with CAC (13), as well as the relationship of LVH with adverse cardiovascular outcomes (6,12), we hypothesized that adults 30 to 50 years of age with low short-term/high lifetime predicted risk of atherosclerotic CVD would have a greater prevalence of LVH and an increased abdominal aortic wall thickness (AWT), a measure of subclinical aortic atherosclerosis (14), compared with those with low short-term/low lifetime predicted risk. We further hypothesized that these associations would persist among CAC-negative individuals.

The design of the DHS and details of variable definitions were described previously (15- 16). Briefly, the DHS is a multiethnic, population-based study of residents of Dallas County that was prospectively designed to assess the presence of and risk factors for subclinical CVD. In the present study, we restricted our analyses to those who underwent cardiac magnetic resonance (CMR) and were 30 to 50 years of age at the time of their third visit in the DHS.

The details of the CMR protocol were previously described (16). To measure LV cavity and wall volume, the endocardial and epicardial borders on the slices were traced manually. To address quality control and potential errors in outliers in the DHS, a second observer rechecked the contours for 302 analyses (10.8% of the short-axis scans). This included all participants with a raw LV mass <100 g (97 individuals), all with a LV mass >2 SDs above the mean (105 individuals), and a random selection of an additional 100 individuals from the remaining participants. The interobserver difference for LV mass was 9.2 ± 5 g (5.8 ± 3.5%; n = 15), the intraobserver difference was 10.5 ± 8.6 g (7.1 ± 6.0%; n = 8), and the interscan variability was 4.9 ± 10.9 g (2.9 ± 7.5%; n = 8) (16), which was similar to what we previously reported for the measurement of LV mass by CMR at our institution (17). LV concentricity was defined as LV mass/LV end-diastolic volume. LV mass was indexed to 3 commonly used measures of body size (body surface area, height^2.7, and fat-free mass), and LVH was defined as LV mass/body surface area ≥89 g/m² (women) and ≥112 g/m² (men), LV mass/height^2.7 ≥39 g/m^2.7 (women) and ≥48 g/m^2.7 (men), and LV mass/fat-free mass ≥3.7 g/kg (men and women) (16).

AWT was measured on the infrarenal abdominal aorta (18). Six transverse slices were obtained per participant. For each slice, the adventitial and luminal boundaries were drawn and were converted to areas, assuming concentric circles, using the analytic software. The radius of each was then calculated, and the mean AWT for each slice was calculated as the difference between adventitial and luminal radii. The mean AWT per participant was calculated as the mean value of the corresponding per-slice AWT measures across the number of slices for each participant. Participants with overt aneurysms were not included in analyses. The intraclass correlation coefficient between luminal cross-sectional area measurements for the 2 observers was 0.94 and the mean interobserver difference was 4.2 ± 6.6%, as previously reported (18). We previously described the interstudy variability in AWT measurement in our laboratory (0.03 ± 0.15 mm; n = 32) (19).

Electron beam computed tomography scanning was performed using an Imatron C-150XP EBCT scanner (Imatron Inc., San Bruno, California) (20). Duplicate scans were performed within 1 to 2 min. CAC scores were recorded in Agatston units, and the average of 2 scans was used as the final CAC score. As previously described (20), there was high interscan variability in CAC scores in the range of 0 to 10. Therefore, CAC was defined as present with a mean score >10 Agatston units.

As suggested by the Adult Treatment Panel III risk assessment tool (1), we used age, sex, serum levels of total and high-density lipoprotein cholesterol, smoking, blood pressure, and treatment for hypertension to estimate the short-term risk of CHD. Low predicted short-term risk was defined as an estimated 10-year risk <10% of fatal or nonfatal CHD. Although the Adult Treatment Panel III guidelines define high short-term risk as >20% (1), we wanted to focus on individuals with truly low short-term risk and therefore were more inclusive in our definition of high short-term risk (≥10% estimated 10-year risk or the presence of diabetes mellitus) as in our previous studies (10). Participants with self-reported myocardial infarction or stroke were classified as high short-term risk (n = 42). Participants with low short-term risk were further divided into 2 categories: low lifetime predicted risk and high lifetime predicted risk of atherosclerotic CVD (Table 1) (10- 11). All participants with high lifetime predicted risk had a predicted lifetime risk of ≥39% through age 95 years and had ≥1 risk factors requiring medical treatment. This stratification resulted in 3 mutually exclusive risk groups: 1) low short-term/low lifetime predicted risk; 2) low short-term/high lifetime predicted risk; and 3) high predicted short-term risk.

Statistical analyses were performed using SAS for Windows (release 9.2; SAS Institute, Inc., Cary, North Carolina). All analyses were stratified by sex. LVH functioned as the outcome variable, and the risk group functioned as the predictor variable. Baseline characteristics, measures of cardiac phenotypes, and prevalent LVH were computed for the 3 risk groups. Continuous variables were compared using the general linear models, whereas categorical variables were compared using the chi-square test or Fisher exact test. Multiple comparisons were assessed for those variables that were statistically different across the groups via Tukey's method for continuous variables and a resampling, bootstrapping technique for categorical variables. We also created a composite end point of subclinical CVD that included the presence of LVH, detectable CAC, or AWT >75th percentile and compared it across the 3 risk groups. To assess whether associations of the risk groups with LVH and AWT were distinct from associations with CAC, sensitivity analyses were performed after excluding participants with CAC.

A total of 2,707 participants between 30 and 65 years of age underwent CMR in the DHS. Of these, 1,883 participants were 30 to 50 years of age. Information on some of the baseline characteristics was missing for 79 participants, leaving 1,804 participants in the present study cohort.

Baseline characteristics of study participants are shown in (Table 2). Almost 86% of the participants had low predicted short-term risk. Further stratification by predicted lifetime risk resulted in 2 similarly sized groups with low short-term/low lifetime predicted risk and low short-term/high lifetime predicted risk. Although participants with low short-term/high lifetime predicted risk had higher levels of traditional risk factors compared with those with low short-term/low lifetime predicted risk (Table 2), the 10-year risk for both subgroups (estimated 10-year risk of CHD events was 3% vs. 1% for men and 1% vs. 0.1% for women, respectively) was well below the threshold for treatment based on current guidelines (1).

In comparison with participants with low short-term/low lifetime predicted risk (Table 3), those with low short-term/high lifetime predicted risk had greater LV mass, LV wall thickness, concentricity, AWT, and prevalent CAC (p < 0.001 for all) but similar indexed LV end-diastolic and end-systolic volumes.

To assess the prevalence of LVH in the 3 risk groups, we analyzed indexed LVH as a dichotomous variable. Depending on the method of indexing, the age-adjusted odds of LVH were 1.6- to 3.3-fold greater in the low short-term/high lifetime predicted risk group compared with the low short-term/low lifetime predicted risk group. The findings were consistent when LVH was defined by LV mass in the top tertile for the study cohort (data not shown). Overall, we observed a similar pattern of results across sexes (Figure 1) and ethnic subgroups (Figure 2), although not all the associations reached statistical significance (likely due to the small sample size of subgroups).

To determine whether our findings were primarily due to any single risk factor, we sequentially excluded participants with one of the major risk factors (i.e., hypertension [n = 415], smoking [n = 520], hypercholesterolemia [n = 152], and diabetes mellitus [n = 141]). Although exclusion of participants with hypertension decreased the magnitude of observed associations, both men and women (respectively) with low short-term/high lifetime risk versus those with low short-term/low lifetime risk continued to have higher median values for indexed LV mass (92 g/m² vs. 88 g/m² and 72 g/m² vs. 70 g/m², respectively; p < 0.05 for both). Exclusion of participants with other major risk factors did not significantly influence our findings (data not shown).

The composite end point of LVH (irrespective of indexing method), CAC, and AWT was 1.3- to 1.9-fold more common among participants with low short-term/high lifetime predicted risk compared with the low short-term/low lifetime predicted risk group (Figure 3).

CAC was absent in >80% of the participants (n = 1,525). Of these, 1,363 (89%) had low predicted short-term risk. In this subgroup, LV mass, prevalent LVH, and AWT were greater in participants with low short-term/high lifetime predicted risk compared with those with low short-term/low lifetime predicted risk (Table 4).

In this large multiethnic population-based study, we found that a majority of adults 30 to 50 years of age were at low predicted short-term risk of CHD. However, more than one-half of these participants had high lifetime predicted risk for the development of atherosclerotic CVD. There was significantly increased LV mass and 1.6- to 3.3-fold greater prevalence of LVH (depending on the method of indexing) among participants with low short-term/high lifetime predicted risk versus those with low short-term/low lifetime predicted risk. Similarly, there was 1.3- to 1.9-fold increase in the composite end point of LVH, CAC, and AWT among participants with low short-term/high lifetime predicted risk versus those with low short-term/low lifetime predicted risk. These findings were qualitatively similar across sexes and ethnic groups. In sensitivity analysis, similar findings were observed after excluding participants with CAC.

Recently, we found that individuals with low short-term but high lifetime predicted risk of atherosclerotic CVD had an increased burden of subclinical atherosclerosis as determined by the presence of CAC, carotid IMT, and CAC progression compared with those with low short-term and low lifetime predicted risk (10). In an analysis of an independent cohort (the DHS), we confirmed that among subjects with low short-term risk, those with high lifetime predicted risk versus those with low lifetime predicted risk have an increased burden of CAC. Moreover, we extend this observation to additional subclinical CVD phenotypes, including LVH and AWT, the latter being a novel measure of subclinical atherosclerosis in a vascular bed distinct from the coronary arteries.

As seen previously (10- 11), significant differences existed in the risk-factor burden between individuals with low short-term/low lifetime predicted risk and low short-term/high lifetime predicted risk. Although these risk factors may not lead to evident CVD in the short-term, they lead to structural changes that over a lifetime increase the risk of atherosclerotic complications, heart failure, and death (6,8). LVH has been shown to be associated with subclinical atherosclerosis (13), adverse cardiovascular outcomes, and overall mortality (6,12). The association of high lifetime risk with LVH, especially of the concentric phenotype, adds validity to using predicted lifetime risk for further risk stratification in young adults.

Atherosclerosis in the peripheral arteries is associated with adverse clinical events (21). The risk factors for aortic atherosclerosis are similar to those for carotid atherosclerosis and increased aortic IMT demonstrated by ultrasonography has been associated with subclinical atherosclerosis (14,22). However, there are few magnetic resonance imaging–based data regarding the utility of AWT as a marker of subclinical atherosclerosis (23), with the majority of studies reporting data on carotid IMT, aortic plaque burden (24), and plaque prevalence (22). Nevertheless, the association of high lifetime risk with increased AWT supports our previous observations that high lifetime risk is associated with subclinical atherosclerosis (10).

As observed previously (10), CAC was absent in >80% of individuals with low short-term/high lifetime predicted risk. Importantly, we show that the association of high lifetime risk with LVH, concentric remodeling, and AWT persisted after excluding participants with CAC. Specifically, in subjects without detectable CAC, a concentric LVH phenotype, prevalent LVH, and an increased AWT were more common among participants with low short-term/high lifetime predicted risk than those with low short-term/low lifetime predicted risk. These findings indicate that the association of high lifetime risk with LV structural changes and abdominal AWT is at least, in part, independent of coronary atherosclerosis.

Lifetime risk prediction is an emerging concept in clinical medicine. In young adults, the use of short-term risk prediction may be misleading because many individuals with low short-term risk may still be at high risk for the development of CVD over their remaining life span. In these individuals, further risk stratification using predicted lifetime risk may assist in better communication of long-term risk and promote healthy lifestyle changes. By demonstrating an association of high predicted lifetime risk with LVH, a well-established marker of increased adverse cardiovascular events, and AWT, these data provide support to the notion that lifetime risk prediction provides an important and valid refinement in the classification of young individuals at low short-term risk.

Our study has several strengths. To our knowledge, this is the first study that analyzed the prevalence of LVH and AWT in the context of lifetime predicted risk of atherosclerotic CVD. Also, we included data from a large population-based sample with detailed phenotyping of participants, including CMR measurements of cardiac structure and function. This allowed a more precise determination of the associations between lifetime predicted risk and LVH. Our study also has several limitations. Due to the cross-sectional design of the study, outcomes related to LVH could not be determined. Furthermore, although the DHS is a multiethnic sample, the risk prediction algorithm that we applied was derived from the Framingham Heart Study, which enrolled exclusively white subjects. However, published reports suggest that this risk prediction model provides reliable estimates of CVD burden for both whites and nonwhites (8,25). Although the absolute differences in LV mass and AWT between the 2 groups were small, these differences must be interpreted in view of the fact that these individuals are young and apparently healthy and the magnitude of these differences would be expected to increase over time. Furthermore, coronary calcification is a later manifestation of coronary atherosclerosis, and the lack of detectable CAC does not exclude coronary atherosclerosis.

We demonstrated for the first time that participants with low short-term/high lifetime predicted risk of atherosclerotic CVD compared with those with low short-term/low lifetime predicted risk had a 1.6- to 3.3-fold greater prevalence of LVH, significantly increased LV mass of a concentric phenotype, and increased AWT. Further, these associations persisted among CAC-negative individuals. These data, coupled with our previous findings (10), suggest that an assessment of lifetime risk of CVD can aid in the identification of a subgroup of individuals 30 to 50 years of age who are at low risk over a 10-year period but nevertheless have a significant subclinical CVD burden.