Net reclassification indices have recently become popular statistics for measuring the prediction increment of new biomarkers. We review the various types of net reclassification indices and their correct interpretations. We evaluate the advantages and disadvantages of quantifying the prediction increment with these indices. For pre-defined risk categories, we relate net reclassification indices to existing measures of the prediction increment. We also consider statistical methodology for constructing confidence intervals for net reclassification indices and evaluate the merits of hypothesis testing based on such indices. We recommend that investigators using net reclassification indices should report them separately for events (cases) and nonevents (controls). When there are two risk categories, the components of net reclassification indices are the same as the changes in the true-positive and false-positive rates. We advocate use of true- and false-positive rates and suggest it is more useful for investigators to retain the existing, descriptive terms. When there are three or more risk categories, we recommend against net reclassification indices because they do not adequately account for clinically important differences in shifts among risk categories. The category-free net reclassification index is a new descriptive device designed to avoid pre-defined risk categories. However, it suffers from many of the same problems as other measures such as the area under the receiver operating characteristic curve. In addition, the category-free index can mislead investigators by overstating the incremental value of a biomarker, even in independent validation data. When investigators want to test a null hypothesis of no prediction increment, the well-established tests for coefficients in the regression model are superior to the net reclassification index. If investigators want to use net reclassification indices, confidence intervals should be calculated using bootstrap methods rather than published variance formulas. The preferred single-number summary of the prediction increment is the improvement in net benefit.
Coronary heart disease (CHD) is the leading cause of death in the United States, yet assessing risk of its development remains challenging. The present study evaluates a new automated assay of small dense low-density lipoprotein cholesterol content (sdLDL-C) and whether sdLDL-C is a risk factor for CHD compared with LDL-C or small LDL particle concentrations derived from nuclear magnetic resonance spectroscopy.
Approach and Results
sdLDL-C was measured using a new automated enzymatic method, and small LDL concentrations were obtained by nuclear magnetic resonance in 4387 Multi-Ethnic Study of Atherosclerosis participants. Cox regression analysis estimated hazard ratios for developing CHD for 8.5 years after adjustments for age, race, sex, systolic blood pressure, hypertension medication use, high-density lipoprotein cholesterol, and triglycerides. Elevated sdLDL-C was a risk factor for CHD in normoglycemic individuals. Those in the top sdLDL-C quartile showed higher risk of incident CHD (hazard ratio, 2.41; P=0.0037) compared with those in the bottom quartile and indicated greater CHD risk than the corresponding quartile of LDL-C (hazard ratio, 1.75; P=0.019). The association of sdLDL-C with CHD risk remained significant when LDL-C (<2.57 mmol/L) was included in a multivariate model (hazard ratio, 2.37; P=0.012). Nuclear magnetic resonance–derived small LDL concentrations did not convey a significant risk of CHD. Those with impaired fasting glucose or diabetes mellitus showed higher sdLDL-C and small LDL concentrations but neither was associated with higher CHD risk in these individuals.
This new automated method for sdLDL-C identifies risk for CHD that would remain undetected using standard lipid measures, but only in normoglycemic, nondiabetic individuals.
Previous longitudinal cohort studies have suggested an association between baseline depressive symptoms and incident hypertension. We assessed this possible association using data from the Multi-ethnic Study of Atherosclerosis, a population-based prospective cohort study of 6,814 US adults from 4 different racial/ethnic groups. Baseline users of antihypertensive medications and participants lost to follow-up were excluded leaving 3914 participants. Patients with baseline depressive symptoms (n=622) were defined using a high score on the Center for Epidemiologic Studies Depression Scale (≥ 16) or the use of an antidepressant medication. Hypertension was defined as systolic blood pressure ≥ 140, diastolic blood pressure ≥90 or new use of antihypertensive medications plus physician diagnosis. Estimates were adjusted for known risk factors including: age, sex, baseline blood pressure, diabetes, and body mass index. Untreated blood pressure was estimated using an imputation approach. A total of 477 participants developed hypertension. Using relative risk regression, patients with baseline depressive symptoms did not have an increased risk of incident hypertension (Relative Risk = 1.02; 95% Confidence Interval (CI):0.99 to 1.05) although an association between tricyclic antidepressants and hypertension (Relative Risk 1.20; 95% CI:1.05 to 1.37) was observed in sub-group analysis. Depression, even after adjustment for covariates, was associated with small changes in systolic (+2.4 mmHG; 95% CI: 0.2 to 4.7) and diastolic (+0.8 mmHG; 95% CI: −0.6 to 2.3) blood pressure. Depressive symptoms may be associated with slight increases in blood pressure in this multi-ethnic cohort but it is premature to conclude much without longer studies in other populations.
Multi-Ethnic Study of Atherosclerosis; depression; hypertension; blood pressure; imputation; censored normal regression
Previous research has suggested that emerging evidence from randomized controlled trials (RCTs) is often not reflected in physician selection of medication class for first-line anti-hypertensive therapy.
To evaluate the association of RCT evidence in December 2002 from the Antihypertensive and Lipid-Lowering treatment to prevent Heart Attack Trial (ALLHAT) on use of anti-hypertensive medications over time in a multi-ethnic cohort.
The Multi-Ethnic Study of Atherosclerosis study, a prospective cohort study of 6,814 adults from 4 ethnic groups, had four separate assessments of drug use. Users of anti-hypertensive medications at baseline were excluded. We evaluated temporal changes in the medication class reported by new users of antihypertensive medications.
After the exclusion of antihypertensive drug users at baseline, 32% of new users of anti-hypertensive drugs seen at exam 2 were prescribed a diuretic. The publication of ALLHAT was associated with a subsequent increase in the proportion of new users taking diuretics at exam 3 compared with exam 2 (Relative Risk (RR):1.31; 95% Confidence Interval (CI):1.09–1.59). After the report from ALLHAT, the proportion of users of diuretics seen at exam 3 rose to 44% (starting in 2004) and 39% in exam 4 (starting in 2005). This increase in the proportion of diuretic use among new users of anti-hypertensive medications declined slightly but could still be detected at exam 4 as compared to exam 2 (RR:1.28; 95% CI:1.04–1.57).
The randomized trial evidence from the ALLHAT study was temporally associated with a moderate increase in diuretic use.
Multi-Ethnic Study of Atherosclerosis; antihypertensive medications; drug utilization; longitudinal
Cardiac magnetic resonance imaging (MRI) is a non-invasive technique used to accurately and reproducibly measure biological parameters such as left ventricular mass. However, some subjects either refuse or are unable to complete testing, and the impact of excluding these missing data from predictive models is unknown.
Multiple imputation was applied to cardiac MRI data that were previously analyzed using a complete case approach. The model variables – 10 traditional cardiovascular risk factors and five sociodemographic variables – were used as a basis for imputation. Men and women were imputed separately. The primary focus was assessing the change in the cardiovascular predictors of left ventricular geometry and systolic function.
Although 27% of participants were missing cardiac MRI data, multiple imputation returned results similar to those of a complete case analysis. These results were robust to the point of including additional variables in the imputation analysis above and beyond the model variables. The degree of variance explained by the models increased marginally but the statistical inference was altered for only two predictors out of 53 cardiovascular risk factors using multiple imputation.
The results suggest that the cardiac MRI data in the Multi-Ethnic Study of Atherosclerosis (MESA) do not substantively change when missing data are handled using multiple imputation. Future analyses of cardiac MRI data may consider the complete case approach to be adequate despite the high rate of missing data in this population.
Comparison of methods; Complete case; Magnetic resonance imaging; Multiple imputation
Previous basic and cross-sectional studies obtained conflicting results regarding the association of pathogens with coronary artery calcium (CAC). The aim of this study is to prospectively evaluate this association in a population-based cohort.
We examined 5,744 individuals aged 45-84 years at baseline (2000-02) who underwent repeated CAC assessment on average 2.4 years later (a half at visit 2 [2002-04] and the other half at visit 3 [2004-05]). CAC incidence was defined as newly detectable CAC at follow-up (475 cases of 2,942 participants). CAC progression was defined as annualized change in CAC Agatston score ≥10 units/year if baseline CAC score >0 to <100 or ≥10%/year if baseline score ≥100 (1,537 cases of 2,802 participants). Seropositivity was assessed in the entire cohort for Chlamydia pneumonia and in a random sample (n=873) for Helicobacter pylori, cytomegalovirus, herpes simplex virus, and hepatitis A virus.
Seropositivity to Chlamydia pneumoniae was not significantly associated with CAC incidence (odds ratio [OR] 1.11 [95% CI, 0.88-1.39], P=0.371) or progression (1.14 [0.96-1.36], P=0.135) even in unadjusted models. When CAC incidence and progression were combined, we observed significant association with Chlamydia pneumoniae seropositivity before adjustment (OR 1.17 [1.03-1.33], P=0.016) but not in a model adjusting for traditional risk factors (1.04 [0.90-1.19], P=0.611). The results were consistent across subgroups according to age, sex, and race/ethnicity. None of five pathogens or their accrual was associated with CAC incidence and progression in the subsample.
Our prospective study does not support the pathophysiological involvement of these pathogens in CAC development.
Coronary Calcium; Atherosclerosis; Pathogens; Infection
To evaluate the predictive value of abdominal aortic calcium (AAC) for incident cardiovascular disease (CVD) independent of coronary artery calcium (CAC).
Approach and Results
We evaluated the association of AAC with CVD in 1974 men and women aged 45 to 84 years randomly selected from the Multi-Ethnic Study of Atherosclerosis participants who had complete AAC and CAC data from computed tomographic scans. AAC and CAC were each divided into following 3 percentile categories: 0 to 50th, 51st to 75th, and 76th to 100th. During a mean of 5.5 years of follow-up, there were 50 hard coronary heart disease events, 83 hard CVD events, 30 fatal CVD events, and 105 total deaths. In multivariable-adjusted Cox models including both AAC and CAC, comparing the fourth quartile with the ≤50th percentile, AAC and CAC were each significantly and independently predictive of hard coronary heart disease and hard CVD, with hazard ratios ranging from 2.4 to 4.4. For CVD mortality, the hazard ratio was highly significant for the fourth quartile of AAC, 5.9 (P=0.01), whereas the association for the fourth quartile of CAC (hazard ratio, 2.1) was not significant. For total mortality, the fourth quartile hazard ratio for AAC was 2.7 (P=0.001), and for CAC, it was 1.9, P=0.04. Area under the receiver operating characteristic curve analyses showed improvement for both AAC and CAC separately, although improvement was greater with CAC for hard coronary heart disease and hard CVD, and greater with AAC for CVD mortality and total mortality. Sensitivity analyses defining AAC and CAC as continuous variables mirrored these results.
AAC and CAC predicted hard coronary heart disease and hard CVD events independent of one another. Only AAC was independently related to CVD mortality, and AAC showed a stronger association than CAC with total mortality.
aortic diseases; calcium; cardiovascular diseases; diagnostic imaging; epidemiology
Risk markers including coronary artery calcium (CAC), carotid intima-media thickness (CIMT), ankle-brachial Index (ABI), brachial flow-mediated dilation (FMD), high sensitivity C -reactive protein (hs-CRP) and family history (FH) of coronary heart disease (CHD) have been reported to improve on the Framingham risk score (FRS) for prediction of CHD. However, there are no direct comparisons of these markers for risk prediction in a single cohort.
We compared improvement in prediction of incident CHD/cardiovascular disease (CVD) of these 6 risk markers within intermediate risk participants (5 % < FRS < 20%) in the Multi-Ethnic Study of Atherosclerosis (MESA).
Design, Setting and Participants
Of 6814 MESA participants from 6 US field centers, 1330 were intermediate risk, without diabetes mellitus, and had complete data on all 6 markers. Recruitment spanned July 2000 to September 2002; follow-up extended through May 2011. Probability- weighted Cox proportional hazard models were used to estimate hazard ratios (HR). Area under the receiver operator characteristic curve (AUC) and net reclassification improvement (NRI) were used to compare incremental contributions of each marker when added to the FRS + race/ethnicity.
Main Outcome Measures
Incident CHD defined as MI, angina followed by revascularization, resuscitated cardiac arrest or CHD death. Incident CVD additionally included stroke or CVD death.
After median follow-up of 7.6 years (IQR 7.3 – 7.8 years), 94 CHD and 123 CVD events occurred. CAC, ABI, hs-CRP and FH were independently associated with incident CHD in multivariable analyses [HR (95%CI: 2.60(1.94-3.50), 0.79(0.66-0.95), 1.28(1.00-1.64) and 2.18(1.38-3.42) respectively]. CIMT and FMD were not associated with incident CHD in multivariable analyses [HR (95%CI) 1.17(0.95- 1.45) and 0.95(0.78 −1.14) respectively]. Although the addition of the markers individually to the FRS +race/ethnicity improved the AUC, CAC afforded the highest increment (0.623 vs. 0.784) while FMD afforded the least [0.623 vs. 0.639]. For incident CHD, the NRI with CAC was 0.659, FMD 0.024, ABI 0.036, CIMT 0.102, FH 0.160 and hs-CRP 0.079. Similar results were obtained for incident CVD.
CAC, ABI, hs-CRP and FH are independent predictors of incident CHD/CVD in intermediate risk individuals. CAC provides superior discrimination and risk reclassification compared with other risk markers.
Coronary artery calcium (CAC), measured by computed tomography (CT), has strong predictive value for incident cardiovascular disease (CVD) events. The standard CAC score is the Agatston, which is weighted upward for greater calcium density. However, some data suggest increased plaque calcium density may be protective for CVD.
To determine the independent associations of CAC volume and CAC density with incident CVD events.
Design, Setting, and Participants
Multicenter, prospective observational MESA study (Multi-Ethnic Study of Atherosclerosis), conducted at 6 US field centers of 3398 men and women from 4 race/ethnicity groups; non-Hispanic white, African American, Hispanic, and Chinese. Participants were aged 45-84 years, free of known CVD at baseline, had CAC greater than 0 on their baseline CT, and were followed up through October 2010.
Main Outcomes and Measures
Incident coronary heart disease (CHD) and all CVD events
During a median of 7.6 years of follow-up, there were 175 CHD events and an additional 90 other CVD events for a total of 265 CVD events. With both lnCAC volume and CAC density scores in the same multivariable model, the lnCAC volume score showed an independent association with incident CHD, with a hazard ratio (HR) of 1.81 (95% CI, 1.47-2.23) per standard deviation (SD = 1.6) increase, absolute risk increase 6.1 per 1000 person-years, and for CVD an HR of 1.68 (95% CI, 1.42-1.98) per SD increase, absolute risk increase 7.9 per 1000 person-years. Conversely, the CAC density score showed an independent inverse association, with an HR of 0.73 (95% CI, 0.58-0.91) per SD (SD = 0.7) increase for CHD, absolute risk decrease 5.5 per 1000 person-years, and an HR of 0.71 (95% CI, 0.60-0.85) per SD increase for CVD, absolute risk decrease 8.2 per 1000 person years. Area under the receiver operating characteristic curve analyses showed significantly improved risk prediction with the addition of the density score to a model containing the volume score for both CHD and CVD. In the intermediate CVD risk group, the area under the curve for CVD increased from 0.53 (95% CI, 0.48-0.59) to 0.59 (95% CI, 0.54-0.64), P = .02.
Conclusions and Relevance
CAC volume was positively and independently associated with CHD and CVD risk. At any level of CAC volume, CAC density was inversely and significantly associated with CHD and CVD risk. The role of CAC density should be considered when evaluating current CAC scoring systems.
To compare the association of the Framingham Risk Score (FRS) and Reynolds Risk Score (RRS) with subclinical atherosclerosis, assessed by incidence and progression of coronary artery calcium (CAC).
The comparative effectiveness of competing risk algorithms for indentifying subclinical atherosclerosis is unknown.
The Multi-Ethnic Study of Atherosclerosis (MESA) is a prospective cohort study of 6,814 participants free of baseline CVD. All participants underwent risk factor assessment, as well as baseline and follow-up CAC testing. We assessed the performance of the FRS and RRS to predict CAC incidence and progression using relative risk and robust linear regression.
The study population included 5,140 individuals (61±10 years, 47% males, mean follow-up: 3.1±1.3 years). Among 53% of subjects (n=2,729) with no baseline CAC, 18% (n=510) developed incident CAC. Both the FRS and RRS were significantly predictive of incident CAC [RR 1.40 (95% CI 1.29 – 1.52), and RR 1.41 (95% CI 1.30 – 1.54) per 5% increase in risk, respectively] and CAC progression [mean CAC score change 6.92 (95% CI 5.31 – 8.54) and 6.82 (95% CI 5.51 – 8.14) per 5% increase]. Discordance in risk category classification (< or > 10% 10-year CHD risk) occurred in 13.7%, with only the RRS consistently adding predictive value for incidence and progression of CAC. These subclinical atherosclerosis findings are supported by a CHD events analysis over 5.6±0.7 year follow-up.
Both the RRS and FRS predict onset and progression of subclinical atherosclerosis. However, the RRS may provide additional predictive information when discordance between the scoring systems exists.
coronary artery calcium progression; subclinical atherosclerosis; risk prediction; Reynolds Risk Score; Framingham Risk Score
Although higher visit-to-visit variability (VVV) of blood pressure (BP) is associated with increased cardiovascular disease risk, the physiological basis for VVV of BP is incompletely understood.
We examined the associations of aortic distensibility (assessed by magnetic resonance imaging) and artery elasticity indices (determined by radial artery pulse contour analysis) with VVV of BP in 2,640 and 4,560 participants, respectively, from the Multi-Ethnic Study of Atherosclerosis. Arterial measures were obtained at exam 1. BP readings were taken at exam 1 and at 3 follow-up visits at 18-month intervals (exams 2, 3, and 4). VVV was defined as the SD about each participant’s mean systolic BP (SBP) across visits.
The mean SDs of SBP were inversely associated with aortic distensibility: 7.7, 9.9, 10.9, and 13.2mm Hg for quartiles 4, 3, 2, and 1 of aortic distensibility, respectively (P trend < 0.001). This association remained significant after adjustment for demographics, cardiovascular risk factors, mean SBP, and antihypertensive medication use (P trend < 0.01). In a fully adjusted model, lower quartiles of large artery and small artery elasticity (LAE and SAE) indices were also associated with higher mean SD of SBP (P trend = 0.02 for LAE; P trend < 0.001 for SAE).
In this multiethnic cohort, functional alterations of central and peripheral arteries were associated with greater long-term VVV of SBP.
arteries; blood pressure; epidemiology; hypertension; vasculature.
The measurement of carotid intima-media thickness (CIMT) is a valid method to quantify levels of atherosclerosis. The present study was conducted to compare the strengths of associations between CIMT and cardiovascular risk factors in two different populations.
The Multi-Ethnic Study of Atherosclerosis (MESA) and the Heinz Nixdorf Recall Study (HNR) are two population-based prospective cohort studies of subclinical cardiovascular disease. All Caucasian subjects aged 45 to 75 years from these cohorts who were free of baseline cardiovascular disease (n = 2,820 in HNR, n = 2,270 in MESA) were combined. CIMT images were obtained using B-mode sonography at the right and left common carotid artery and measured 1 cm starting from the bulb.
In both studies, age, male sex, and systolic blood pressure showed the strongest association (P < .0001 for each) for a higher CIMT. The mean of mean far wall CIMT was slightly higher in MESA participants (0.71 vs 0.67 mm). Almost all significant variables were consistent between the two cohorts in both magnitude of association with CIMT and statistical significance, including age, sex, smoking, diabetes, cholesterol levels, and blood pressure. For example, the association with systolic blood pressure was (ΔSD = 0.011; 95% confidence interval, 0.0009 to 0.014) per mm Hg in MESA and (ΔSD = 0.010; 95% confidence interval, 0.005 to 0.021) per mm Hg in HNR. This consistency persisted throughout the traditional (Framingham) risk factors.
A comparison of the associations between traditional cardiovascular risk factors and CIMT across two culturally diverse populations showed remarkable consistency.
Carotid intima-media thickness; Subclinical atherosclerosis; Multi-Ethnic Study of Atherosclerosis; MESA; Heinz Nixdorf Recall Study; HNR
Evaluating disparities in health care is an important aspect of understanding differences in disease risk. The purpose of this study is to describe methodology for estimating such disparities, with application to a large multi-ethnic cohort study.
The Multi-Ethnic Study of Atherosclerosis (MESA) includes 6814 participants aged 45–84 years free of cardiovascular disease. Prevalence ratio (PR) regression was used to model baseline lipid lowering medication (LLM) or anti-hypertensive medication use at baseline as a function of gender, race, risk factors and estimated pre-treatment biomarker values.
Hispanics and African-Americans had lower prevalence of medication use than non-Hispanic whites, even at the same risk factor profile. This became non-significant after adjustment for socio-economic status. Although gender did not influence the prevalence of LLM use (PR=1.09, 95% CI 0.95 to 1.25), there were differences in the association of diabetes and HDL with LLM use by gender. Men were significantly less likely to be on anti-hypertensive medications than women (PR=0.86, 95% CI 0.80 to 0.92, p<0.001) and this was not explained by risk factors or socioeconomic status. Lack of health insurance strongly influenced medication use, controlling for risk factors and other markers of socio-economic status.
Disparities exist in the treatment of cholesterol and hypertension. Hispanics and African Americans had less use of LLM, men had less use of anti-hypertensives. Risk factors have differential associations with medication use depending on gender. Methods described in this paper can provide improved disparity estimation in observational cohort studies.
disparities; medication; statistical methods; statins; anti-hypertensives
Adipokines are secreted from adipose tissue, influence energy homeostasis and may contribute to the association between obesity and hypertension. Among 1,897 participants enrolled in the Multi-Ethnic Study of Atherosclerosis, we examined associations between blood pressure and leptin, tumor necrosis factor – α [TNFα], resistin and total adiponectin. The mean age and body mass index was 64.7 years and 28.1 respectively, and 50% were female. After adjustment for risk factors, a 1-standard deviation increment higher leptin level was significantly associated with higher systolic (5.0 mmHg), diastolic (1.9), mean arterial (2.8) and pulse pressures (3.6), as well as a 34% higher odds for being hypertensive (p < 0.01 for all). These associations were not materially different when the other adipokines, as well as body mass index, waist circumference or waist to hip ratio, were additionally added to the model. Notably, the associations between leptin and hypertension were stronger in men, but were not different by race/ethnic group, body mass index or smoking status. Adiponectin, resistin and TNFα were not independently associated with blood pressure or hypertension. Higher serum leptin, but not adiponectin, resistin or TNFα, is associated with higher levels of all measures of blood pressure, as well as a higher odds of hypertension, independent of risk factors, anthropometric measures and other selected adipokines.
adipokine; leptin; blood pressure; hypertension; ethnicity
Elevation in plasma activity of von Willebrand Factor (vWF) reflects endothelial dysfunction and predicts death in pulmonary arterial hypertension (PAH). Higher vWF activity is also associated with lower right ventricular (RV) ejection fraction in PAH. Little is known about the relationship between vWF and RV structure and function in adults without cardiovascular disease. In the current investigation, we included 1,976 participants with MRI assessment of RV structure and function and measurement of vWF activity from the Multi-Ethnic Study of Atherosclerosis. Multivariable linear regression was used to estimate the associations between vWF activity and measures of RV structure and function after adjusting for demographics, anthropometrics, smoking, diabetes mellitus, hypertension and the corresponding left ventricular (LV) parameter. The average vWF activity was 140.7 ± 57.2%. Elevated vWF activity was independently associated with lower RV mass, RV end-diastolic volume and RV stroke volume in models with and without adjustment for the corresponding LV parameter (all p < 0.05). There was no association observed between vWF activity and RV ejection fraction. In conclusion, higher vWF activity is associated with lower RV mass, RV end-diastolic volume and RV stroke volume. These associations are independent of common cardiovascular risk factors and LV morphologic changes.
Cardiovascular Imaging; Biomarkers; Pulmonary Hypertension; Right Ventricle
The aim of the present study was to evaluate how torsion is influenced by left ventricular (LV) remodeling associated with age, gender and hypertension in a large community-based population.
Methods and Results
Myocardial shortening and torsion were assessed by tagged cardiac magnetic resonance (CMR) in 1478 participants without clinically apparent cardiovascular disease in the Multi-Ethnic Study of Atherosclerosis (MESA). Torsion was defined as the difference between apical and basal rotation, divided by slice distance. In multivariable linear regression models, older age was associated with lower stroke volume (−3.6 ml/decade, p<0.001) and higher LV mass –to-volume ratio (0.03 g/ml/decade, p<0.001) along with lower circumferential shortening (−0.17%/decade, p<0.05). Torsion, however, was greater at older ages (0.14 °/decade, p<0.001) and in women (0.37°/cm vs. men, p<0.001). Hypertensive participants had higher LV mass and LV mass –to-volume ratio (15.5g and 0.07 g/ml, respectively, p<0.001 for both). Circumferential shortening was lower in hypertensive (−0.42%, p<0.01), whereas torsion was higher after adjustment for age and gender (0.17°/cm, p<0.05).
Older age is associated with lower LV volumes and greater relative wall thickness, and accompanied by lower circumferential myocardial shortening, whereas torsion is greater with older age. Hypertensive individuals have greater LV volumes and relative wall thickness and lower circumferential shortening. Torsion, however, is greater in hypertension independent of age and gender. Torsion may therefore represent a compensatory mechanism to maintain an adequate stroke volume and cardiac output in the face of progressively reduced LV volumes and myocardial shortening associated with hypertension and aging.
Torsion; Hypertension; Age; remodeling; Cardiac Magnetic Resonance
Systemic inflammation has been linked to the development of heart failure in population studies including MESA (Multi-Ethnic Study of Atherosclerosis) but little evidence exists regarding potential mechanism of this relationship. In this study, we used longitudinal MRI follow-up analysis to examine whether C-reactive protein (CRP) levels relate to progressive myocardial functional deterioration as a potential mechanism of incident heart failure.
Regional myocardial functional data from MESA participants who had baseline CRP measurement and also underwent tagged cardiac MRI both at baseline and at five-year follow-up were analyzed. Left ventricular (LV) midwall and mid-slice peak circumferential strain (Ecc), of which a more negative value denotes stronger regional myocardial function, was measured. Ecc change was calculated as the difference between baseline and follow-up Ecc.
During the follow-up period, participants (n=785) with elevated CRP experienced a decrease in strain, independent of age, gender and ethnicity (B=0.081; ΔEcc change per 1mg/L CRP change, 95% CI 0.036–0.126, p<0.001, Model 1), and additionally beyond systolic blood pressure, heart rate, diabetes, smoking status, body mass index, current medication and glomerular filtration rate (B=0.099, 0.052–0.145, p<0.001, Model 2). The relationship remained statistically significant after further adjustment for LV mass, coronary calcium score and interim clinical coronary events (B=0.098, 0.049–0.147, p<0.001, Model 3).
Higher CRP levels are related to progressive myocardial functional deterioration independent of subclinical atherosclerosis and clinical coronary events in asymptomatic individuals without previous history of heart disease.
inflammation; myocardial function; magnetic resonance imaging
Studying the effects of medications on endpoints in an observational setting is an important yet challenging problem due to confounding by indication. The purpose of this study is to describe methodology for estimating such effects while including prevalent medication users. These techniques are illustrated in models relating statin use to cardiovascular disease (CVD) in a large multi-ethnic cohort study.
The Multi-Ethnic Study of Atherosclerosis (MESA) includes 6814 participants aged 45-84 years free of CVD. Confounding by indication was mitigated using a two step approach: First, the untreated values of cholesterol were treated as missing data and the values imputed as a function of the observed treated value, dose and type of medication, and participant characteristics. Second, we construct a propensity-score modeling the probability of medication initiation as a function of measured covariates and estimated pre-treatment cholesterol value. The effect of statins on CVD endpoints were assessed using weighted Cox proportional hazard models using inverse probability weights based on the propensity score.
Based on a meta-analysis of randomized controlled trials (RCT) statins are associated with a reduced risk of CVD (relative risk ratio = 0.73, 95% CI: 0.70, 0.77). In an unweighted Cox model adjusting for traditional risk factors we observed little association of statins with CVD (hazard ratio (HR) = 0.97, 95% CI: 0.60, 1.59). Using weights based on a propensity model for statins that did not include the estimated pre-treatment cholesterol we observed a slight protective association (HR = 0.92, 95% CI: 0.54-1.57). Results were similar using a new-user design where prevalent users of statins are excluded (HR = 0.91, 95% CI: 0.45-1.80). Using weights based on a propensity model with estimated pre-treatment cholesterol the effects of statins (HR = 0.74, 95% CI: 0.38, 1.42) were consistent with the RCT literature.
The imputation of pre-treated cholesterol levels for participants on medication at baseline in conjunction with a propensity score yielded estimates that were consistent with the RCT literature. These techniques could be useful in any example where inclusion of participants exposed at baseline in the analysis is desirable, and reasonable estimates of pre-exposure biomarker values can be estimated.
Multiple imputation; Confounding by indication; Propensity score; Inverse probability of treatment weights; Statins
The integrated discrimination improvement (IDI) index is a popular tool for evaluating the capacity of a marker to predict a binary outcome of interest. Recent reports have proposed that the IDI is more sensitive than other metrics for identifying useful predictive markers. In this article, the authors use simulated data sets and theoretical analysis to investigate the statistical properties of the IDI. The authors consider the common situation in which a risk model is fitted to a data set with and without the new, candidate predictor(s). Results demonstrate that the published method of estimating the standard error of an IDI estimate tends to underestimate the error. The z test proposed in the literature for IDI-based testing of a new biomarker is not valid, because the null distribution of the test statistic is not standard normal, even in large samples. If a test for the incremental value of a marker is desired, the authors recommend the test based on the model. For investigators who find the IDI to be a useful measure, bootstrap methods may offer a reasonable option for inference when evaluating new predictors, as long as the added predictive capacity is large.
biological markers; bootstrap confidence interval; prediction; risk assessment; sampling distribution; sampling error; selection bias; type I error
Alcohol use has been consistently found to have a J-shaped association with coronary heart disease, with moderate drinkers exhibiting a decreased risk compared to both heavy drinkers and non-drinkers. However, studies of the association between alcohol use and subclinical coronary artery disease have conflicted.
To determine whether alcohol is associated with the presence, amount, or progression of coronary calcium over a 2- to 4-year period.
MESA is a prospective community-based cohort study of subclinical cardiovascular disease in a multi-ethnic cohort. In 2000–2002, 6814 participants free of clinical cardiovascular disease were enrolled at 6 participating centers.
There were 3766 (55.5%) current drinkers, 1635 (24.1%) former drinkers, and 1390 (20.5%) never drinkers included in the analysis. Although light to moderate alcohol consumption was associated with lower coronary heart disease risk, we found no evidence of a protective or J-shaped association of alcohol and coronary artery calcium (CAC). In fact there was evidence that heavy consumption of hard liquor was associated with greater CAC accumulation. Other alcoholic beverages were not associated with CAC prevalence, incidence or progression.
This is the first large study to evaluate the association of alcohol and coronary artery calcium in four racial/ethnic groups, and to evaluate progression of calcification. These results suggest that the cardiovascular benefits that may be derived from light to moderate alcohol consumption are not mediated through reduced CAC accumulation.
Racial/ethnic differences in the incidence and severity of heart failure (HF) are not well understood, but may be related to pre-existing variations in myocardial function.
To examine racial/ethnic differences in regional myocardial function among asymptomatic individuals free of known cardiovascular disease.
Design, setting and patients
The Multi-Ethnic Study of Atherosclerosis is a prospective, observational study of individuals without baseline cardiovascular disease, representing four major racial/ethnic groups. A total of 1099 study participants underwent cardiac MRI with tissue tagging; for each study, peak systolic strain (Ecc) and strain rate (SRs) were determined in four left ventricular (LV) regions.
Main outcome measures
Multiple linear regression was used to analyse the relationship between race/ethnicity and regional strain (Ecc and SRs) while adjusting for cardiovascular risk factors.
Compared with other racial/ethnic groups, Chinese-Americans had the greatest magnitude of Ecc in a majority of LV regions (–19.60±3.78, p<0.05); Chinese-Americans also had the greatest absolute values for SRs in all regions, reflecting higher rate of systolic contraction (–2.01±0.76, p<0.05). Conversely, African-Americans had the lowest Ecc values (–17.50±4.00, p<0.05) in the majority of wall regions while Hispanics demonstrated the lowest rate of contractility in all wall regions (–1.44±0.50, p≤0.001) in comparison with the other racial/ethnic groups. These race-based differences remained significant in the majority of LV wall regions after adjusting for multiple variables, including hypertension and LV mass.
Important race-based differences in regional LV systolic function in a large cohort of asymptomatic individuals have been demonstrated. Further research is needed to investigate the possible mechanisms related to the race/ethnicity-based variations found in this study.
Coronary artery calcium (CAC), carotid intima-media thickness, and left ventricular (LV) mass and geometry offer the potential to characterize incident cardiovascular disease (CVD) risk in clinically asymptomatic individuals. The objective of the study was to compare these cardiovascular imaging measures for their overall and sex-specific ability to predict CVD.
Methods and Results
The study sample consisted of 4965 Multi-Ethnic Study of Atherosclerosis participants (48% men; mean age, 62±10 years). They were free of CVD at baseline and were followed for a median of 5.8 years. There were 297 CVD events, including 187 coronary heart disease (CHD) events, 65 strokes, and 91 heart failure (HF) events. CAC was most strongly associated with CHD (hazard ratio [HR], 2.3 per 1 SD; 95% CI, 1.9 to 2.8) and all CVD events (HR, 1.7; 95% CI, 1.5 to 1.9). Most strongly associated with stroke were LV mass (HR, 1.3; 95% CI, 1.1 to 1.7) and LV mass/volume ratio (HR, 1.3; 95% CI, 1.1 to 1.6). LV mass showed the strongest association with HF (HR, 1.8; 95% CI, 1.6 to 2.1). There were no significant interactions for imaging measures with sex and ethnicity for any CVD outcome. Compared with traditional risk factors alone, overall risk prediction (C statistic) for future CHD, HF, and all CVD was significantly improved by adding CAC, LV mass, and CAC, respectively (all P<0.05).
There was no evidence that imaging measures differed in association with incident CVD by sex. CAC was most strongly associated with CHD and CVD; LV mass and LV concentric remodeling best predicted stroke; and LV mass best predicted HF.
imaging; cardiovascular diseases; sex
Biomarkers of inflammation and hemostasis have been associated with left ventricular (LV) mass. We studied relationships of C-reactive protein (CRP), interleukin-6 (IL6), D-dimer, soluble intercellular adhesion molecule-1 (sICAM-1), plasminogen activator inhibitor 1 (PAI-1), soluble thrombomodulin (sTM), soluble tumor necrosis factor type 1 receptor (sTNFR1), von Willebrand factor (vWF), soluble E-selectin (sE-selectin), factor VIII, fibrinogen, matrix metalloproteinase 3 (MMP3), and matrix metalloproteinase 9 (MMP9) with LV mass in an asymptomatic population. Multi-Ethnic Study of Atherosclerosis participants underwent magnetic resonance imaging to characterize LV mass; biomarkers were measured using standardized protocols (N = 763 to 4979). Adjusted models were used to associate each biomarker with LV mass while correcting for potential confounding.
LV mass was associated with many biomarkers after adjustment for demographic characteristics and traditional cardiovascular risk factors. Although the demographic and risk factor adjustments attenuated the association of CRP and IL6 with LV mass, further adjustment for weight changed regression coefficients from positive to negative for CRP and IL6 for LV mass. sTM, Factor VIII, and vWF were directly associated with LV mass in fully-adjusted models. For sTNFR1, sICAM-1, D-dimer, fibrinogen, and PAI-1, adjustment for risk factors and weight rendered associations with LV mass nonsignificant.
In this large cohort free of clinical cardiovascular disease, several hemostasis and inflammation markers were associated with LV mass. The unusual finding of a negative relationship of CRP and IL6 with LV mass only after adjustment for weight suggests that the effects of inflammation on LV mass are strongly influenced by obesity.
Left ventricle; biomarker; hemostasis; inflammation
The purpose of this study was to examine the association of both a low and a high ankle-brachial index (ABI) with incident cardiovascular events in a multi-ethnic cohort.
Abnormal ankle-brachial indices (ABIs), both low and high, are associated with elevated cardiovascular disease (CVD) risk. However, it is unknown whether this association is consistent across different ethnic groups, and whether it is independent of both newer biomarkers and other measures of subclinical atherosclerotic CVD.
6647 non-Hispanic white, African-American, Hispanic, and Chinese men and women aged 45–84 years from free-living populations in six United States field centers and free of clinical CVD at baseline had extensive measures of traditional and newer biomarker risk factors, and measures of subclinical CVD, including the ABI. Incident CVD, defined as coronary disease, stroke, or other atherosclerotic CVD death, was determined over a mean follow-up of 5.3 years.
Both a low (<1.00) and a high (≥ 1.40) ABI were associated with incident CVD events. Gender- specific and ethnic-specific analyses showed consistent results. Hazard ratios were 1.77 (p<.001) for a low and 1.85 (p=.050) for a high ABI after adjustment for both traditional and newer biomarker CVD risk factors, and the ABI significantly improved risk discrimination. Further adjustment for coronary artery calcium score, common and internal carotid intimal medial thickness, and major ECG abnormalities only modestly attenuated these hazard ratios.
In this study both a low and a high ABI were associated with elevated CVD risk in persons free of known CVD, independent of standard and novel risk factors, and independent of other measures of subclinical CVD. Further research should address the cost-effectiveness of measuring the ABI in targeted population groups.
peripheral arterial disease; ankle-brachial index; cardiovascular events; risk factors; subclinical atherosclerosis
We sought to examine the relationship between circulating interleukin-6 (IL-6) level and regional left-ventricular (LV) function among apparently healthy individuals free of cardiovascular disease.
Methods and results
Using magnetic resonance myocardial tagging, we determined peak systolic circumferential strain (Ecc) as a measure of regional systolic function in 894 asymptomatic participants in the Multi-Ethnic Study of Atherosclerosis. Ecc was analysed by harmonic phase imaging separately in the LV anterior wall, septum, lateral wall, and inferior wall. Global Ecc was calculated as the average of Ecc in all myocardial segments. We performed multivariable linear regression to evaluate the independent associations between log IL-6 and Ecc, after adjusting for demographic features, cardiovascular risk factors, and markers of subclinical atherosclerosis. The inverse relationships between IL-6 and absolute Ecc were similar in both genders. In multivariable analysis, higher IL-6 level was independently associated with reduced systolic function (less negative Ecc) in the septum [regression coefficient = 1.03 per unit higher log IL-6, 95% confidence interval (CI) 0.26–1.79, P = 0.008] and inferior wall (regression coefficient = 1.65, 95% CI 0.74–2.56, P < 0.001), but not in the anterior wall (P = 0.27) or lateral wall (P = 0.52). Overall, there was an independent inverse association between IL-6 and global Ecc (regression coefficient = 0.94, 95% CI 0.37–1.51, P = 0.001). Compared with C-reactive protein, higher IL-6 level demonstrates a stronger independent association with reduced regional systolic function.
In asymptomatic men and women without documented cardiovascular disease, there is a strong, independent, inverse relationship between IL-6 and regional LV systolic function. These findings suggest that IL-6 may underlie the pathogenetic link between inflammation, LV dysfunction and incipient heart failure. The observed variable relationships between IL-6 and systolic function across different LV regions warrant further investigations.
Heart failure; Myocardial contraction; Interleukin-6; Magnetic resonance imaging