Methods to improve coronary heart disease (CHD) risk prediction include incorporation of coronary artery calcium (CAC) within risk models and considering longer time horizons such as evaluation of lifetime cardiovascular risk (LTR). We compared the accuracy of 10-year Framingham risk score (FRS), LTR, and Multi-Ethnic Study of Atherosclerosis (MESA) arterial age-adjusted 10-year risk for prediction of incident CHD events in men in the Prospective Army Coronary Calcium Project. We studied 1,633 healthy men (mean age 43 years, range 40 to 50 years, mean FRS 4.6%) with electron-beam computed tomography for CAC. Events (CHD death, myocardial infarction, acute coronary syndrome with nonelective coronary revascularization) were prospectively assessed over 5.6 ± 1.5 years. Predicted risk using 10-year FRS for CHD and cardiovascular disease, LTR, and MESA arterial age were evaluated in relation with CAC and CHD events. CAC prevalence was strongly related to LTR, increasing in a graded fashion from 10.1% to 66.7% across 8 categories of increasing LTR. On receiver operating characteristic analysis, MESA arterial age (area under curve 0.78, 95% confidence interval 0.64 to 0.93) had the largest area under the curve but similar areas under the curve were observed for 10-year risk (CHD 0.74, 0.61 to 0.86; cardiovascular disease 0.70, 0.59 to 0.82), LTR (0.68, 0.49 to 0.76), and LTR with CAC as a covariate (0.76, 0.63 to 0.89). Inclusion of family history of CHD or body mass index did not improve model accuracy. In conclusion, increasing LTR was associated with increasingly prevalent CAC in this low-risk cohort and inclusion of CAC improved the accuracy of LTR for short-term event prediction.
The additive value of coronary artery calcium (CAC) quantification to estimates of lifetime cardiovascular risk (LTR) and the comparative predictive accuracy of LTR versus 10-year Framingham risk score (FRS) and Multi-Ethnic Study of Atherosclerosis (MESA) arterial age-adjusted FRS for incident coronary heart disease (CHD) events has not been evaluated. We sought to assess the association of LTR to subclinical CAC and to compare the predictive accuracy of 10-year FRS and LTR, individually and in combination with CAC Agatston scores, to MESA arterial age-adjusted FRS for the prediction of incident CHD events in men in the Prospective Army Coronary Calcium (PACC) Project.
Methods
The rationale, design, and methods of the PACC Project have been previously described. Briefly, all active-duty Army personnel, 40 to 50 years old, and living within the National Capital Area of the Walter Reed Healthcare System were invited to undergo electron beam computed tomography for Agatston CAC quantification at the time of a periodic mandated physical examination. Participants with a history of CHD (previous myocardial infarction and/or coronary revascularization) or a history of angina pectoris according to the Rose questionnaire were ineligible. The study was conducted from October 26, 1998 to February 19, 2003, during which 2,000 unselected men and women provided written informed consent for participation. Of enrolled subjects, 1,640 were men of whom 1,639 completed the CAC scan and form the basis of this report. Women participants were not included in this analysis because of a significantly smaller sample size and low prevalence of CAC and CHD incidence. The study was approved by the local institutional review board.
A detailed medical history, including assessment for hypertension, diabetes mellitus, and current medications, was performed by study investigators. A family history of CHD was defined as a history of sudden death, myocardial infarction, or coronary revascularization in a relative before age of 55 years for men or 65 years for women. Blood pressure at rest was recorded as the average of 3 seated measurements taken 5 minutes apart using an automated sphygmomanometer. Hypertension was defined as systolic blood pressure >135 mm Hg, diastolic blood pressure >85 mm Hg, or a history of hypertension (treated or untreated). Height and weight were measured and body mass index calculated as kilograms per meter squared. Fasting total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride cholesterol, and serum glucose were measured at enrollment. Low-density lipoprotein cholesterol was quantified using a direct assay. Electron beam computed tomography was performed using an Imatron C-150 scanner (San Francisco, California) and CAC was quantified using the Agatston scoring method by a single trained physician who was blinded to clinical variables, as previously described. A scan was considered positive for CAC when the total coronary Agatston score was >0.
The 10-year FRSs for incident CHD and cardiovascular disease (CVD) were each calculated using measured risk factor variables as specified within regression equations from the Framingham Heart Study. Lifetime risk was assessed according to the method of Berry et al in which standard CV risk factors such as systolic blood pressure, diastolic blood pressure, total cholesterol, and smoking status were graded according to the following severity scale: (0) optimal (systolic blood pressure <120 mm Hg, diastolic blood pressure <80 mm Hg, total cholesterol <180 mg/dl, nonsmoker), (1) nonoptimal (systolic blood pressure 120 to 139 mm Hg, diastolic blood pressure 80 to 89 mm Hg, total cholesterol 180 to 199 mg/dl, nonsmoker), (2) increased (systolic blood pressure 140 to 159 mm Hg, diastolic blood pressure 90 to 99 mm Hg, total cholesterol 200 to 239 mg/dl, nonsmoker), or (3) severely increased (systolic blood pressure ≥160 mm Hg, diastolic blood pressure ≥100 mm Hg, total cholesterol ≥240 mg/dl, treated hypertension, treated hyperlipidemia, active smoking). Points were assigned for systolic blood pressure (0 through 3), diastolic blood pressure (0 through 3), total cholesterol (0 through 3), and smoking status (0 if negative, 3 if currently smoking) and summed for each participant to quantify LTR (score range 0 through maximum of 9). In addition, presence of ≥1 increased risk factor was considered predictive of high LTR, defined as >39% lifetime risk for incident CVD. MESA arterial age was determined by an online calculator ( http://www.mesa-nhlbi.org/Calcium/ArterialAge.aspx ) and then used to calculate an arterial age-adjusted 10-year FRS for CHD. Patients with diabetes mellitus, defined as a reported history of diabetes, treatment with diabetic medication(s), or a fasting blood glucose >126 mg/dl, were excluded from the present analysis based on the known increased lifetime risk inherent to this unique risk factor, consistent with previous studies of LTR.
The primary outcome was incident CHD events, defined as death from CHD, myocardial infarction, or hospitalized acute coronary syndrome with nonelective coronary revascularization. CV outcomes were ascertained by annual telephonic contact by clinical nurse coordinators and all events were adjudicated by 2 cardiologists unaware of risk factors or CAC results.
Continuous variables were expressed as mean ± 1 SD or median with interquartile range and evaluated using Student’s unpaired t test or Mann–Whitney U test, as appropriate. Categorical variables were presented as frequency with percentage and evaluated using chi-square or Fisher’s exact test, where indicated. Risk variables not showing a normal distribution were logarithmically transformed for use in linear models and/or hypothesis testing. To compare the prognostic value of the studied risk prediction models, individual multivariable logistic regression models were compared using clinical risk prediction for 10-year FRS (CHD and CVD), LTR, LTR plus CAC as a covariate (LTR + CAC), and the MESA arterial age-adjusted FRS, for the outcome of observed incident CHD events following CAC testing. Exploratory logistic models for CHD event prediction were developed including different risk models as covariates. Receiver operating characteristics curves were constructed for each model and compared for event prediction. Statistical significance was accepted for 2-sided p values <0.05. All calculations were performed using SPSS 17 for Windows (SPSS, Inc., Chicago, Illinois).
Results
In total 1,633 men met inclusion criteria for the present analysis and their baseline characteristics are listed in Table 1 . The study population was predominately Caucasian (72%) of mean age 43 ± 3 years. The most prevalent CV risk factor was a family history of CHD (31.7%) and 14.5% of patients had hypertension. Mean baseline 10-year FRS for CHD was 4.6 ± 2.6 with 75.7%, 19.3%, 4.4%, and 0.1% of patients identified as low (<6%), low-intermediate (6% to <10%), intermediate (10% to <20%), and high (≥20%) CHD risk, respectively. A minority of patients were taking aspirin, statins, or antihypertensive medications at baseline. On CAC testing, mean Agatston score was 20 ± 111 with 22% of patients possessing a CAC score >0.
| Variable | |
|---|---|
| Age (years) | 42.9 ± 2.8 |
| Ethnicity | |
| Caucasian | 1,173 (71.8%) |
| African-American | 291 (17.8%) |
| Other (Asian, Hispanic, other) | 169 (10.3%) |
| Family history of coronary heart disease | 518 (31.7%) |
| Hypertension | 237 (14.5%) |
| Current tobacco use | 113 (6.9%) |
| Metabolic syndrome | 108 (6.6%) |
| Systolic blood pressure (mm Hg) | 124.3 ± 12.0 |
| Diastolic blood pressure (mm Hg) | 77.6 ± 8.8 |
| Body mass index (kg/m 2 ) | 27.8 ± 3.5 |
| Total cholesterol (mg/dl) | 204.2 ± 36.1 |
| Low-density lipoprotein cholesterol (mg/dl) | 128.5 ± 31.4 |
| High-density lipoprotein cholesterol (mg/dl) | 50.4 ± 12.6 |
| Triglyceride cholesterol (mg/dl) | 129.8 ± 86.5 |
| Fasting glucose (mg/dl) | 92.7 ± 11.1 |
| Baseline medications | |
| Aspirin | 193 (11.8%) |
| Statin | 93 (5.7%) |
| Antihypertensive | 98 (6.0%) |
| 10-Year Framingham risk score (%) | |
| Coronary heart disease | 4.6 ± 2.6 |
| Cardiovascular disease | 7.3 ± 3.9 |
| Agatston coronary calcium score >0 | 366 (22.4%) |
| Mean Agatston coronary calcium score | 20 ± 111 |
Incorporation of the CAC score into the MESA arterial age calculator significantly increased the “vascular age” of participants from 42.9 ± 2.7 to 44.1 ± 11.0 years (p <0.001). There was also a significant increase in 10-year FRS for CHD when adjusted for the MESA arterial age, increasing from 4.6 ± 2.6 to 7.7 ± 8.2 (p <0.001).
LTR scores significantly varied in the study population with most patients possessing a LTR score of 3, as shown in Figure 1 . In the entire population, 62.5% of patients were defined as having a high (>39%) predicted LTR of CVD (≥1 increased risk factor), with 71.4% of these patients identified at low (<6%) short-term CHD risk using the 10-year FRS. LTR was strongly related to prevalent CAC. The prevalence of subclinical calcified coronary atherosclerosis increased in a graded fashion from 10.1% to 66.7% of subjects across 8 categories (LTR scores 0 to 7) of increasing LTR ( Figure 2 ).
