Biomarkers and Assessment of Subclinical Atherosclerosis for the Prediction of Cardiovascular Disease: What is the Current Evidence?
- First Online:
- Cite this article as:
- Morrissey, R.P. & Rana, J.S. Curr Cardiovasc Risk Rep (2013) 7: 108. doi:10.1007/s12170-013-0297-x
- 123 Views
Cardiovascular disease is the leading cause of death in the United States and Europe, with the majority being coronary deaths. The first presentation of cardiovascular disease, often in patients without significant traditional risk factors, is often myocardial infarction. Strategies utilizing biomarkers and assessment of subclinical atherosclerosis have been shown previously to correlate with cardiovascular disease. This article will review current evidence for these strategies, of which, measurement of coronary artery calcium has been shown to provide the greatest risk assessment, discrimination, and risk reclassification for coronary heart disease.
KeywordsCardiovascular riskFramingham risk scoreBiomarkersBrain natriuretic peptideC-reactive proteinCoronary artery calciumCarotid intima media thicknessGenome-wide association
Despite an overall decline in cardiovascular death, 33 % and 47 % of all deaths in the United States (US) and in Europe respectively, are still attributable to cardiovascular disease (CVD) [1, 2]. Coronary heart disease (CHD) is still the leading cause of death in people <65 years of age in Europe , and upward of 980,000 people in the US are expected to have a first myocardial infarction (MI) in the United States . The monetary cost of CVD in the US and Europe is 300 billion US dollars and 200 billion euros, respectively [1, 2]. In order to further stymie the threat of CVD, efforts have been made to identify earlier the individuals at risk for CHD. This review will highlight the current evidence from 2012 with respect to risk stratification utilizing biomarkers, genome-wide association studies, and assessment of subclinical atherosclerosis.
Of biomarkers recently and currently under investigation with respect to risk stratification in asymptomatic patients, homocysteine, uric acid, brain natriuretic peptide (BNP), c-reactive protein (CRP), and troponin (Tn) are the most widely studied and commercially available. Other investigational biomarkers, among others, include interleukin-6, myeloperoxidase, plasminogen activator-1, fibrinogen, and von Willebrand factor. Mirroring the current trend, we will focus on BNP, CRP and Tn.
A “multi-biomarker” score comprising three novel biomarkers (soluble ST2, growth differentiation factor-15 (GDF-15), and high-sensitivity troponin I) was added to BNP and CRP to determine prognostic value in 3428 individuals from the Framingham cohort who were followed for a mean of 11.3 years . All biomarkers correlated with death except Tn, and all correlated with heart failure (HF) except CRP; all markers correlated with major cardiovascular (CV) event (MI, unstable angina (UA), CHD death, heart failure and stroke) except CRP; and only CRP correlated with the combined endpoint of CHD events (MI, UA, CHD death). However, the hazard ratios were modest at best, with the greatest adjusted risk association being between GDF-15 and death at 1.52 (confidence interval (CI) 1.37–1.67) after adjusting for other biomarkers. For example, the association between BNP and HF was 1.29 (CI 1.10–1.52) and between CRP and CHD event was 1.21 (CI 1.02–1.43). Combining all five markers improved risk stratification, in particular for scores in the highest quartile: HR 3.2 (CI 2.18–4.7) for death, 6.25(2.63–14.82) for HF, although only 1.87 (1.28–2.73) for all major CV events.
The utility of BNP has further been shown in a series of 300 asymptomatic patients who underwent evaluation for “silent cardiac target organ damage” (myocardial ischemia, left ventricular hypertrophy, systolic dysfunction, diastolic dysfunction or left atrial enlargement) by transthoracic echocardiography, stress echocardiography and/or myocardial perfusion imaging . The area under the curve (AUC) for BNP to identify any form of silent cardiac dysfunction was 0.78 versus for high sensitivity Tn T was 0.70. The AUC for BNP plus Tn was 0.81. In contrast, the AUC was 0.61 for microalbuminuria 0.58 for eGFR, and 0.49 for uric acid.
Recent studies comparing the utility of biomarkers and subclinical atherosclerosis for risk assessment of cardiovascular disease
Mean age (SD)
Follow-up, median years
Change in ROC
Change in ROC
Change in ROC
Change in ROC
Rotterdam study [5••]
In 1286 asymptomatic individuals in the EISNER study, individual biomarkers (CRP, interleukin-6, myeloperoxidase, BNP, plasminogen activator-1) added almost nothing to the FRS alone (HR 1.1 (CI 1.06–1.14), c-statistic 0.73 (0.66–0.82) [7••]. However, a “multi-biomarker” strategy did improve upon the FRS alone with a HR of 2.1 (CI 1.1–3.8), and a c-statistic of 0.75 (0.68–0.84); however, this was not statistically significant. There was no significant reclassification of risk based on a multi biomarker strategy (NRI 0.04) (Table 1). Similarly, in the Multi-Ethnic Study of Atherosclerosis (MESA) cohort, the addition of CRP to FRS did not significantly improve risk prediction in 1330 asymptomatic non-diabetics followed for a median of 7.6 years: adjusted HR 1.28 (CI 1.00–1.64) (Table 1) [8••].
The Emerging Risk Factors Collaboration analyzed data from 52 prospective studies with a total of 246,669 individuals without a history of CVD and analyzed the addition of HDL, CRP, and fibrinogen to a risk prediction model based on age, sex, smoking status, blood pressure, history of diabetes, and total cholesterol level . While the results were statistically significant, they were not clinically meaningful with an increase in the c-statistic by 0.005 with the addition of HDL and by 0.0039 with the addition of CRP as well as HDL. The NRI for CRP was 1.52 %. Indeed, the strongest predictor of incident CVD was age with a HR of 1.9 (1.86–1.94) versus 1.2 (1.18–1.22) for CRP. Also published this year, in a meta-analysis of 62,154 statin-treated patients, non-HDL levels were associated with the greatest risk of future CVD events compared to LDL and ApoB concentrations . It is interesting to note that in data recently published from the Prospective Army Coronary Calcium (PACC) project, only age and non-HDL were independently associated with the presence of CAC (and not LDL or HDL), although the HRs were quite low, e.g., 1.16 (1.12–1.19) for non-HDL .
Thus, overall the results from recent studies for the use of biomarkers in risk stratification are far from enthusiastic.
Genome-wide Association Studies
It was anticipated that the highest-yield for identifying CVD risk would be at the genetic level, and given the “residual risk” of CHD after controlling conventional risk factors, the identification of a genetic predisposition toward CHD and CVD would seem promising. Indeed, genetic variants associated with the regulation of blood pressure were shown last year to be correlated with HTN, left ventricular wall thickness, stroke and CAD . However, polymorphisms in specific genes may only account for a very small percentage in blood pressure variance between individuals . A meta-analysis of three genome-wide association (GWA) studies—the Cardiovascular Risk in Young Finns, the Bogalusa Heart Study, and the Health 2000 Survey—analyzed the association of 24 single nucleotide polymorphisms previously identified to be associated with the development of CAD  and analyzed changes in carotid intima-media thickness and carotid elasticity  in over 4500 individuals. In both unadjusted and adjusted models, there were no significant associations between the SNPs and the measures of subclinical atherosclerosis. However, these populations were very young (Finns Study mean age 37.5 years), and not only are these measures of subclinical atherosclerosis unlikely to change significantly over 7 years in these young individuals, as will be discussed subsequently, these measurement tests are not significantly associated with CVD risk themselves. This study adds to several GWA studies that have not shown significant associations between various SNPs and CVD risk .
It may be that a better understanding of transcription/translation and proteomics, i.e., biological function related to the various SNPs may provide insight into the development CVD, as most of these SNPs currently reside in areas believed to be non-coding regions in the human genome, and may play a role in gene regulation.
Among measurement of subclinical atherosclerosis the most commonly-performed and best-studied are carotid intima-media thickness (CIMT), coronary artery calcium (CAC), and ankle-brachial index (ABI).
The Rotterdam Study (mean age 69 years) compared FRS with biomarkers, CIMT and CAC (Table 1) [5••]. CAC was associated the most with CHD risk with a HR of 6.2 (CI 3.4–11.5) as opposed to CIMT (HR 1.6, CI1.1–2.3). Additionally, CAC was also the most discriminative risk marker with a change in c-statistic of 0.05 (0.02–0.06); the only other significant marker was BNP with a change in c-stastic of 0.02 (0.01–0.04). The NRI for CAC plus FRS was 39.3 (26.8–51.7) versus 4.6 (−0.1–9.3) for CIMT; again, the next-closest marker was BNP at 33. Reclassification of risk based on CAC plus FRS was better for men (NRI of 50.9), and did not perform as well for women compared to BNP plus FRS (25.5 versus 27.8).
In the MESA population (mean age 64), in addition to biomarkers, CAC was also compared to ABI, brachial flow-mediated dilatation (BFMD), and CIMT (Table 1); the strongest association with incident CHD (MI, unstable angina, resuscitated cardiac arrest, CHD death) was with CAC with an adjusted HR of 2.6 (1.94–3.5) [8••]. CIMT and BFMDI were not significantly associated with CHD: HRs 1.17 (0.95–1.45) and 0.93 (0.74–1.16), respectively. Similar results were seen for incident CVD (addition of stroke and CVD death endpoints). CAC showed the greatest incremental improvement in discrimination based on AUC for both CHD and CVD: FRS plus CAC 0.784, FRS plus CIMT 0.652, FRS plus ABI 0.65, FRS plus BFMD 0.639 versus FRS alone 0.623. Lastly, adding CAC to FRS resulted in the most appropriate reclassification of risk for both CHD (NRI 0.659) and CVD (NRI 0.466). Other markers had no effect on reclassification. Similar results were seen in the EISNER cohort (mean age 58.6 years) (Table 1). The HR for FRS plus CAC was 1.7 (1.4–2.0) with a c-statistic of 0.84; the addition of a multiple biomarkers score to FRS plus CAC did not increase risk or improve discrimination. Unlike for biomarkers, the NRI showed significant improvement with the addition of CAC to FRS (NRI 0.35).
The limited value of CIMT in risk stratification was also shown in two meta-analyses this year. The first reviewed the association of CIMT with first-time MI or stroke over a median follow-up of 11 years. The HR was 1.09 (1.07–1.12), with a c-statistic of 0.75, and a reclassification improvement of 0.8 % . The second analyzed the association of CIMT progression based on two scans performed a median of 4 years apart with MI and stroke over a mean follow-up of 7 years yielding an adjusted HR was 0.98 (0.95–1.01) .
CAC appears to offer not only the greatest risk stratification for CVD with respect to assessment of subclincal atherosclerosis, but also over other modalities and over biomarkers over traditional risk factors such as the FRS. This improvement in risk stratification is seen for both men  and for women , and across multiple age ranges  and ethnicities . Furthermore, it has recently been shown that CAC assessment improves risk factor modification without significantly increasing downstream costs [23•]. In 2137 asymptomatic individuals who underwent risk-factor counseling, half were randomized to CAC scanning. In the patients who underwent CAC assessment, there were significant dose-response improvements in systolic and diastolic blood pressure, total cholesterol, low-density lipoprotein cholesterol, triglycerides, weight, and FRS.
Several landmark studies analyzing the association of biomarkers and measurements of subclinical atherosclerosis with CVD risk prediction were published in 2012. Based upon these, it would seem that with the possible exception of BNP, there is not much role for routine measurement of biomarkers for the purpose of risk prediction. Genetic risk assessment, while potentially still promising, is still in its infancy. CAC measurement provides significant risk prediction compared to CIMT and other modes of subclinical atherosclerosis assessment. Furthermore, biomarkers including BNP and CRP as well as combinations thereof do not provide further risk stratification or discrimination compared to CAC. CAC also provides the best reclassification of risk. Lastly, CAC measurement improves risk factor modification without significantly increasing downstream costs. With the cost of CAC scanning declining, it would seem to be the best approach to risk stratification beyond conventional risk factors.
No potential conflicts of interest relevant to this article were reported.