Analysis of three risk stratification systems in a Saudi population

Abstract

Introduction

Coronary artery disease is the leading cause of death worldwide. Although there are a number of algorithms in use for determining the risk of coronary artery disease and thus predicting future cardiovascular events, the data available regarding their validity among the Saudi population are insufficient.

Objective

We studied the validity of three clinical score systems in predicting a high risk population defined as having excessive coronary calcification: the American College of Cardiology/American Heart Association (ACC/AHA) Pooled Cohort Risk Equation, the Framingham Risk Score, and the European Systematic Coronary Risk Evaluation.

Methods

We analyzed data from 462 patients aged ⩾40 years. High-risk features were if the Coronary Calcium Score was either >400 or in the ⩾75th percentile using Multi-Ethnic Study of Atherosclerosis (MESA) score. The scores for the three algorithms were then calculated using the participants’ clinical data.

Results

A total of 87 (18.8%) patients were positive for coronary calcification. Among them, 60 (13%) were classified as being at high risk according to the MESA score. Analyzing these patients by the ACC/AHA Pooled Cohort Risk Equation resulted in nine (15%) as being at low risk, 12 (20%) at intermediate risk, and 39 (65%) at high risk. The Framingham Risk classification resulted in 14 (23%) being at low risk, 13 (22%) at intermediate risk, and 33 (55%) at high risk. The European Systematic Coronary Risk Evaluation risk classification showed 24 (40%) at low risk, 12 (20%) at intermediate risk, and 24 (40%) at high risk, with p < 0.0001.

Conclusion

The ACC/AHA Pooled Cohort Risk Equation has superior risk ​ calibration compared to the other two risk-score algorithms in a Saudi population.

Abbreviations

ACC/AHA

the American College of Cardiology/American Heart Association

CAD

Coronary artery disease

FRS

Framingham Risk Score

CCS

Coronary Calcium Score

MDCT

multidetector computed tomography

MESA

Multi-Ethnic Study of Atherosclerosis

SCORE

the European Systematic Coronary Risk Evaluation

ASCVD

atherosclerotic cardiovascular disease

Introduction

Coronary artery disease (CAD) is the most common cause of death worldwide, representing a disease burden in both developing and underdeveloped countries [1]. In the USA, mortality due to cardiovascular diseases accounted for 34.3% of all deaths in 2006. Furthermore, it was reported that in 2010 an American experienced a coronary event approximately every 25 seconds while someone died from a coronary event approximately every minute. The total direct and indirect costs of CAD and stroke in the USA are estimated to be $503.2 billion—higher than for any other diagnostic group [2].

The best strategy for the management of CAD is early detection and control of risk factors such as diabetes mellitus, hypertension, dyslipidemia, and cigarette smoking. In addition, risk stratification for future cardiovascular events has an important role in identifying patients at higher risk.

Multiple risk score algorithms are available that can predict the 10-year risk of atherosclerotic cardiovascular diseases (ASCVD), the most commonly utilized being the Framingham Risk Score (FRS), the European Systematic Coronary Risk Evaluation (SCORE), and the American College of Cardiology/American Heart Association (ACC/AHA) Pooled Cohort Risk Equation [3], [4].

In Saudi Arabia, patients with acute coronary syndrome are almost a decade younger than those in developed countries [5], warranting early detection of those at high risk. However, to date no data testing the validity of risk stratification systems in the Saudi population have been published. The aim of this study was to analyze and compare the accuracy of these systems in identifying patients at higher risk for ASCVD.

Methods

Study settings and participants

A cross-sectional analysis of 918 patients underwent multidetector row computed tomography (MDCT) scan between January 2010 and December 2014 to evaluate the validity of three risk stratification systems in identifying patients at higher risk defined by the Coronary Calcium Score (CCS). Four hundred and sixty two patients older than 40 years were enrolled for the final analysis, while 456 patients were excluded (Fig. 1). The study protocol was approved by the Institutional and Regional Ethical Committee.

Figure 1.

Methods of patient enrollment to the analysis. ACC/AHA = American College of Cardiology/American Heart Association; MDCT = Multidetector computed tomography; FRS = Framingham Risk Score; MESA = Multi-Ethnic Study of Atherosclerosis; SCORE = European Systematic Coronary Risk Evaluation.

Patients’ characteristics

Variables such as sex, race, and ethnicity were obtained by self-reporting. Risk factors for CAD such as hypertension, diabetes mellitus, dyslipidemia, family history of premature CAD, and smoking history were obtained. Total cholesterol, low-density lipoprotein, and high-density lipoprotein were recorded before MDCT scan. In addition, systolic blood pressure, weight, height, waist and hip circumference, and waist–hip ratio were measured, and body mass index calculated.

CT protocol

Participants underwent a noncontrast, low-radiation calcium-scoring scan (dual-source, Definition Flash; Siemens Healthcare, Forchheim, Germany) with a gantry rotation time of 0.28 seconds. Imaging parameters included 2 × 128 × 3-mm slice collimation with 3-mm slices of a field of view extending from the tracheal bifurcation to the diaphragm. The scan was taken with breath-hold and a prospectively electrocardiography-triggered protocol with image acquisition initiated at 70% of the cardiac cycle. The imaging parameters were 120 kVp, 0.33-seconds temporal resolution, and 3-mm slice thickness.

CT image analysis

Images were reconstructed and analyzed using an Advantage Windows AW4 (GE Healthcare, Waukesha, WI, USA) to measure the calcium score using the Agatston method.

Identification of high-risk groups

All patients with either a CCS ⩾400 or a CCS in the ⩾75th percentile using the online sex-specific Multi-Ethnic Study of Atherosclerosis (MESA) score calculator [6] were identified as high risk.

Calculation of clinical risk scores

Patients’ clinical data and lipid profiles were used to calculate the three clinical risk scores, namely sex-specific FRS, sex-specific SCORE, and sex- and race-specific ACC/AHA Pooled Cohort Risk Equation, using age, diabetes mellitus, total cholesterol level, high-density lipoprotein level, smoking status, systolic blood pressure, and use of antihypertensive medication [4], [7], [8].

Statistical analysis

Quantitative variables were expressed as mean ± standard deviation and categorical variables were expressed as frequencies or percentages. For normally distributed continuous variables a two-sample t test was performed, and Chi-square exact tests were run for categorical variables to compare the high- and low-risk groups defined by MESA score. A value of p < 0.05 was considered statistically significant for all tests. All statistical analyses were performed using SPSS for Windows (Version 19.0; SPSS Inc., Chicago, IL, USA).

Results

Baseline characteristics

Baseline characteristics of the 462 patients are shown in Table 1. The mean age of the patients was 52 ± 8 years, with 315 (68%) of them being men. Coronary artery calcification was present in 87 (18.8%), of whom 69 (14.9%) were men and 18 (3.9%) were women.

Table 1.

Baseline characteristics.

Variables
Age (y), mean ±SD	52 ± 8
Male sex, n (%)	315 (68)
​Body mass index (kg/m2), mean ± SD	29.9 ± 4.7
Body weight (kg), mean ± SD	83.2 ± 13
Diabetes mellitus, n (%)	128 (27.5)
Hypertension, n (%)	207 (44.7)
Dyslipidemia, n (%)	24 (5)
Family history of coronary artery disease, n (%)	33 (7)
Current smoking, n (%)	80 (17.4)
Cholesterol (mmol/L), mean ± SD	4.2 ± 1.3
Low-density lipoprotein (mmol/L), mean ± SD	2.6 ± 1.1
High-density lipoprotein (mmol/L), mean ± SD	1.2 ± 0.3
Coronary calcium score	31 ± 136

SD = standard deviation.

MESA score and high-risk patients

Using the MESA score calculator, 60 (13%) patients were in the ⩾75th percentile and comprised the high-risk group. Chi-square exact test analysis showed a significantly higher prevalence of Diabetes mellitus and Hypertension in the high-risk than in the lower-risk group (Table 2).

Table 2.

Comparison between the high- and low-risk group.

Variables	High-risk group	Low-risk group	p
No. of patients	60 (13)	402 (87)
Age (y), mean ± SD	54 ± 10	51 ± 8	0.09
No. of men, n (%)	45 (75)	270 (14)	0.3
Diabetes, n (%)	32 (53)	96 (24)	<0.0001
Hypertension, n (%)	42 (70)	165 (41)	0.001
Smoking, n (%)	8 (13)	72 (18)	0.27
Dyslipidemia, n (%)	5 (8)	19 (4.7)	0.3
Family history of coronary artery disease, n (%)	8 (13)	25 (6.2)	0.1
Coronary calcium score, mean ± SD	230 ± 315	1 ± 6	<0.0001

SD = standard deviation.

Ten-year calculated risk

The expected 10-year risk for each system was stratified into three categories as low, intermediate, and high. Analysis of all patients using the ACC/AHA Pooled Cohort Risk Equation resulted in 264 (57%) classified as low risk, 60 (13%) as intermediate risk, and 138 (30%) as high risk. With FRS, 198 (42%) were classified as low, 132 (29%) as intermediate, and 132 (29%) as high risk. Using the SCORE system, 300 (65%) were classified as low, 48 (10%) as intermediate, and 114 (25%) as high risk (Fig. 2).

Figure 2.

Risk stratification of all participants. ACC/AHA = American College of Cardiology/American Heart Association; FRS = Framingham Risk Score; SCORE = European Systematic Coronary Risk Evaluation.

The three scoring systems overestimated the predicted cardiovascular risk by 130% for the ACC/AHA pooled cohort risk equation, 120% for FRS, and 90% for SCORE.

Patients with a high MESA score

Analyzing these patients using the ACC/AHA Pooled Cohort Risk Equation resulted in nine (15%) as being at low risk, 12 (20%) at intermediate risk, and 39 (65%) at high risk. The FRS risk classification showed 14 (23%) at low risk, 13 (22%) at intermediate risk, and 33 (55%) at high risk. The SCORE risk classification demonstrated 24 (40%) at low risk, 12 (20%) at intermediate risk, and 24 (40%) at high risk (p < 0.0001; Fig. 3).

Figure 3.

Risk stratification of high risk group showed that American College of Cardiology/American Heart Association (ACC/AHA) Pooled Cohort Risk Equation to be more sensitive than Framingham Risk Score (FRS) and the European Systematic Coronary Risk Evaluation (SCORE) in identifying patients at high risk.

Discussion

In the present study, we tested the utility of three different risk stratification systems in predicting patients with a high 10-year ASCVD risk. We found the ACC/AHA Pooled Cohort Risk Equation to be more sensitive than FRS and SCORE in identifying patients at high risk among the Saudi population. In addition we found that ​diabetes mellitus and hypertension were the predominant risk factors in the high risk group, defined by higher coronary calcification.

Cardiovascular diseases is a major health issue in Saudi Arabia, with approximately 5.5% of the population in the age group 30–70 years having CAD [9]. Furthermore, the Saudi Project for Assessment of Coronary Events registry has reported a higher prevalence of CAD risk factors among patients presenting with acute coronary syndrome and a higher prevalence of diabetes mellittus in comparison with individuals in developed countries [5].

Cardiovascular risk stratification is considered an important strategy that aims to classify the general population into either low-, intermediate-, or high-risk groups, which may help in identifying individuals susceptible to hard cardiovascular events and in predicting persons who might benefit from preventative measures, thus resulting in less cardiovascular mortality by decreasing the prevalence of risk factors for CAD [10], [11] and increasing the implementation of various primary prevention methods, such as the use of aspirin, lowering of blood pressure and low-density lipoprotein cholesterol in the indicated group, and reduction of body weight in cases of adiposity [12], [13].

At present there are multiple risk stratification algorithms in use. For decades FRS was the most widely used for the calculation of 10-year risk in asymptomatic individuals. However, the FRS underestimates the lifetime risk, especially in younger individuals with multiple risks and in women; moreover, it does not predict the incidence of stroke [14], [15]. This necessitated the introduction of the ACC/AHA Pooled Cohort Risk Equation to overcome these limitations. In this system ​diabetes mellitus was included as a predictor variable, and fatal and nonfatal stroke were added to the ASCVD end points [16], [17].

Many studies showed that coronary calcification detected using noncontrast MDCT can accurately predict the coronary plaque burden, with a high correlation of >0.9 between CCS and coronary plaque area [18]. In addition, CCS can provide prognostic information regarding the risk of hard cardiovascular events. With a CCS ⩾400 there is 2.5-fold increase in cardiovascular diseases risk when compared with a CCS <10. CCS is considered an independent predictor, and superior to the traditional risk factor of CAD in the assessment of subclinical CAD and future ASCVD events [6], [19], [20].

In the present study, we found that all the three risk stratification systems tend to overestimate the predicted cardiovascular risk by 90% to 130%. A recent study by DeFilippis et al. [21] investigated five different risk scores and reported findings consistent with the present study. These investigators found that the ACC/AHA Pooled Cohort Risk Equation system and the FRS tended to overestimate the risk by 86% and 53%, respectively, while the SCORE was not tested. Similar findings were reported by Muntner et al. [22], who found that the observed and predicted 5-year cardiovascular diseases events in their cohort in the REasons for Geographic and Racial Differences in Stroke and ACC/AHA Pooled Cohort Risk Equation were similar, but found an apparent overestimation of risk mainly in the group with a 10-year ASCVD risk ⩾10%.

Risk overestimation may lead clinicians to start patients on medications for primary prevention that is not indicated, which may expose them to otherwise avoidable adverse effects. However, it may negatively affect the cost-effectiveness of the primary prevention program. For these reasons, further studies with long-term follow-up of ASCVD events are needed for a better assessment of risk stratification algorithms.

The present study indicated that the ACC/AHA Pooled Cohort Risk Equation is more sensitive than other risk scores in the Saudi population, so its use in asymptomatic individuals should provide valuable information regarding future cardiac risk, and may indicate persons who might benefit from measures taken to prevent ASCVD.

Our study has certain limitations: it is a single-center study and is a cross-sectional analysis that depends mainly on the CCS as the gold standard approach to identify the high-risk group, with no follow-up of patients for any ASCVD events. In this study the sample size was relatively small and the high-risk group consisted of a small percentage of the total patient cohort.

Conclusion

Among the Saudi population, the ACC/AHA Pooled Cohort Risk Equation has a higher sensitivity than FRS and SCORE in predicting patient’s future risk driven by the MESA score. Further evidence is needed to verify the validity of the test in formulating preventive measures for ASCVD.

Disclosure: Authors have nothing to disclose with regard to commercial support.