Advertisement

HDL-C, triglycerides and carotid IMT: A meta-analysis of 21,000 patients with automated edge detection IMT measurement

      Highlights

      • We assess the cross-sectional association of CCAIMT with HDL and triglycerides.
      • We included six large observational studies that used the same CCAIMT protocol.
      • CCAIMT was associated with low HDL levels independently of LDL levels.
      • CCAIMT was not related with triglycerides after controlling for confounding factors.
      • Similar findings than those found with clinical events rather than CCAIMT.

      Abstract

      Objective

      Common carotid artery (CCA) intima-media thickness (IMT), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C), but not triglyceride levels, are markers of future cardiovascular events. The relationship between these three factors is, however, unclear.

      Methods

      We included six large observational studies that used the same harmonized, B-mode ultrasound protocol, the same software for IMT measurement by automatic edge detection on CCA in a plaque-free region, following the Mannheim consensus, and certification of all sonographers. Using the best view of the CCA, the sonographer had to confirm that the quality index was ≥0.5 on a measurement performed on 10-mm length. We used individual data meta-analysis to estimate the cross-sectional associations of lipids with CCA-IMT.

      Results

      Overall, 21,587 patients with complete information on lipids and CCA-IMT were available. Age- and sex-adjusted CCA-IMT differed by −7.8 μm (95% CI −9.1 to −6.5 μm, P < 0.001) per 1 SD higher HDL-C level. After further adjustment for other atherosclerosis risk factors, the relationship was attenuated, but remained significant (regression coefficient, −3.7 μm; P < 0.001). This was found regardless of LDL-C levels (P for heterogeneity = 0.70). After adjustment for age and sex, triglycerides were positively associated with CCA-IMT, overall and in each LDL-C subgroup, but not after further adjustments for other risk factors.

      Conclusions

      Relationships between HDL-C and triglyceride levels and CCA-IMT were consistent with that previously observed with clinical events by the Emergency Risk Collaboration group, including at low LDL-C levels. This reinforces the need to verify whether raising HDL-C levels decreases both CCA-IMT and future clinical events.

      Keywords

      1. Introduction

      The lowering of low-density lipoprotein cholesterol (LDL-C) is associated with a reduction in the incidence of cardiovascular events, including stroke. Raising levels of high-density lipoprotein cholesterol (HDL-C) and decreasing triglycerides may present an effective additional strategy to prevent complications of cardiovascular disease, although no clinical trial has yet verified this hypothesis [
      • Amarenco P.
      • Labreuche J.
      Lipid management in the prevention of stroke: review and updated meta-analysis of statins for stroke prevention.
      ]. Strong evidence exists for an epidemiological association between HDL-C and triglyceride levels and risk of cardiovascular disease [
      • Sarwar N.
      • Danesh J.
      • Eiriksdottir G.
      • et al.
      Triglycerides and the risk of coronary heart disease: 10,158 incident cases among 262,525 participants in 29 Western prospective studies.
      ,
      • Freiberg J.J.
      • Tybjaerg-Hansen A.
      • Jensen J.S.
      • Nordestgaard B.G.
      Nonfasting triglycerides and risk of ischemic stroke in the general population.
      ,
      • Gordon D.J.
      • Probstfield J.L.
      • Garrison R.J.
      • et al.
      High-density lipoprotein cholesterol and cardiovascular disease. Four prospective American studies.
      ,
      • Barter P.
      • Gotto A.M.
      • LaRosa J.C.
      • et al.
      HDL cholesterol, very low levels of LDL cholesterol, and cardiovascular events.
      ], but only HDL-C concentration independently contributed to the reduction of cardiovascular risk in a large individual meta-analysis [
      • Di Angelantonio E.
      • Sarwar N.
      • Perry P.
      Emerging Risk Factors Collaboration
      Major lipids, apolipoproteins, and risk of vascular disease.
      ].
      In addition, there is a lack of evidence of an association between HDL-C and/or triglycerides and carotid intima-media thickness (IMT) [
      • Amarenco P.
      • Labreuche J.
      • Touboul P.J.
      High-density lipoprotein-cholesterol and risk of stroke and carotid atherosclerosis: a systematic review.
      ,
      • Labreuche J.
      • Touboul P.J.
      • Amarenco P.
      Plasma triglyceride levels and risk of stroke and carotid atherosclerosis: a systematic review of the epidemiological studies.
      ], a marker of systemic atherosclerosis that is associated with well-established cardiovascular risk factors [
      • Mackinnon A.D.
      • Jerrard-Dunne P.
      • Sitzer M.
      • Buehler A.
      • von Kegler S.
      • Markus H.S.
      Rates and determinants of site-specific progression of carotid artery intima-media thickness: the carotid atherosclerosis progression study.
      ,
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      ], and with cardiovascular outcomes [
      • Lorenz M.W.
      • Markus H.S.
      • Bots M.L.
      • Rosvall M.
      • Sitzer M.
      Prediction of clinical cardiovascular events with carotid intima-media thickness: a systematic review and meta-analysis.
      ,
      • Polak J.F.
      • Pencina M.J.
      • Pencina K.M.
      • O'Donnell C.J.
      • Wolf P.A.
      • Sr D'Agostino RB.
      Carotid-wall intima-media thickness and cardiovascular events.
      ]. Common carotid IMT has also been strongly associated with the occurrence of future internal carotid artery plaques [
      • Von Sarnowski B.
      • Lüdemann J.
      • Völzke H.
      • Dörr M.
      • Kessler C.
      • Schminke U.
      Common carotid intima-media thickness and framingham risk score predict incident carotid atherosclerotic plaque formation: longitudinal results from the study of health in Pomerania.
      ,
      • Zureik M.
      • Ducimetiere P.
      • Touboul P.J.
      • et al.
      Common carotid intima-media thickness predicts occurrence of carotid atherosclerotic plaques: longitudinal results from the Aging Vascular Study (EVA) study.
      ], which are a cause of atherothrombotic stroke. If common carotid artery (CCA)-IMT is associated with lower levels of HDL-C, patients with high CCA-IMT could be a proper population to test the hypothesis that pharmacologic interventions that raise HDL-C levels reduce the occurrence of internal carotid artery plaques, and thereby may reduce the risk of stroke and other cardiovascular events. The rationale to use HDL-C-raising and triglyceride-lowering drugs to reduce carotid IMT is currently lacking [
      • Briel M.
      • Ferreira-Gonzalez I.
      • You J.J.
      • et al.
      Association between change in high density lipoprotein cholesterol and cardiovascular disease morbidity and mortality: systematic review and meta-regression analysis.
      ,
      • Singh I.M.
      • Shishehbor M.H.
      • Ansell B.J.
      High-density lipoprotein as a therapeutic target: a systematic review.
      ,
      • Labreuche J.
      • Deplanque D.
      • Touboul P.J.
      • Bruckert E.
      • Amarenco P.
      Association between change in plasma triglyceride levels and risk of stroke and carotid atherosclerosis: systematic review and meta-regression analysis.
      ]. Among ultrasonographic measurements of carotid IMT, CCA-IMT is the most reproducible and assessable measure, and is believed to be a good tool for cardiovascular risk assessment and intervention studies [
      • Touboul P.J.
      • Hennerici M.G.
      • Meairs S.
      • et al.
      Mannheim carotid intima-media thickness consensus (2004-2006). An update on behalf of the Advisory Board of the 3rd and 4th Watching the Risk Symposium, 13th and 15th European Stroke Conferences, Mannheim, Germany, 2004, and Brussels, Belgium, 2006.
      ].
      We performed a systematic collection of all databases from observational studies that have evaluated CCA-IMT according to the Mannheim consensus [
      • Touboul P.J.
      • Hennerici M.G.
      • Meairs S.
      • et al.
      Mannheim carotid intima-media thickness consensus (2004-2006). An update on behalf of the Advisory Board of the 3rd and 4th Watching the Risk Symposium, 13th and 15th European Stroke Conferences, Mannheim, Germany, 2004, and Brussels, Belgium, 2006.
      ], using the same ultrasound protocol, and the same software for IMT, by automated edge detection measurement, and performed a meta-analysis on individual data. The objective was to provide reliable estimates of cross-sectional associations of HDL-C and triglyceride levels with CCA-IMT. We also evaluated the relationship between LDL-C/HDL-C ratio and CCA-IMT since it has been previously shown that LDL-C/HDL-C ratio is a reliable predictor of cardiovascular risk [
      • McQueen M.J.
      • Hawken S.
      • Wang X.
      • et al.
      Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study.
      ]. This report is based on six large observational studies including a total of 21,587 individuals.

      2. Methods

      Data from six cross-sectional, observational ultrasound studies that evaluated CCA-IMT using the same electronic measure protocol (M'ATH™, Intelligence in Medical Technologies, France) and following the Mannheim consensus recommendations [
      • Touboul P.J.
      • Hennerici M.G.
      • Meairs S.
      • et al.
      Mannheim carotid intima-media thickness consensus (2004-2006). An update on behalf of the Advisory Board of the 3rd and 4th Watching the Risk Symposium, 13th and 15th European Stroke Conferences, Mannheim, Germany, 2004, and Brussels, Belgium, 2006.
      ] were combined. Individual study designs have been published for five of the studies: Paroi Artérielle et Risque Cardiovasculaire (PARC) [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
      ], Paroi Artérielle et Risque Cardiovasculaire in Asia Africa/Middle East and Latin America (PARC-AALA) [
      • Touboul P.J.
      • Hernandez-Hernandez R.
      • Kucukoglu S.
      Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
      ], the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      ], Vicenza Thrombophilia and Atherosclerosis (VITA) [
      • Tosetto A.
      • Prati P.
      • Baracchini C.
      • Manara R.
      • Rodeghiero F.
      Age-adjusted reference limits for carotid intima-media thickness as better indicator of vascular risk: population-based estimates from the VITA project.
      ] and Registry Evaluation MEMory in Buttrio e Remanzacco (REMEMBER) [
      • Prati P.
      • Casaroli M.
      • Bignamini A.
      • et al.
      Cognitive impairment and carotid atherosclerosis in a general Italian midlife and old population.
      ]. The remaining study, Plaque Intima-media Inflammation (PI2), was a 2-year prospective observational study looking at inflammatory markers of atherosclerosis and plaque occurrence in an asymptomatic population with increased IMT and without plaque. This study was stopped prematurely by the sponsor for economic reasons and a low inclusion rate.

      2.1 Data collection and definitions

      Demographics and clinical measurements (medical history, anthropometric measures, blood pressures, lipid and glucose concentrations) from individual patients were collected at enrollment with a standardized form specific to each study.
      Hypertension was defined as blood pressure (systolic/diastolic) ≥140/90 mmHg or current blood pressure-lowering drug treatment. Dyslipidemia was defined as LDL-C ≥160 mg/dL or current lipid-lowering drug treatment. Diabetes was defined as blood glucose ≥126 mg/dL or a current treatment for diabetes (type 1 or type 2). In PARC [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
      ], the glucose plasma level was not assessed at enrollment and subjects were classified as diabetic if the physician diagnosed a medical history of diabetes or the patient was taking an anti-diabetes drug. Personal history of coronary heart disease included a history of myocardial infarction and angina except in VITA where only history of myocardial infarction was collected [
      • Tosetto A.
      • Prati P.
      • Baracchini C.
      • Manara R.
      • Rodeghiero F.
      Age-adjusted reference limits for carotid intima-media thickness as better indicator of vascular risk: population-based estimates from the VITA project.
      ]. Family history of coronary heart disease was defined as a father or brother having fatal or nonfatal myocardial infarction before the age of 55 years, or mother or sister having fatal or nonfatal myocardial infarction before the age of 65 years. Early menopause was defined as menopause before the age of 40 years. Lipid parameters (HDL-C, triglyceride, total cholesterol levels) were obtained from blood sampling at the time of inclusion, but in none of the studies there was a centralized measurement. LDL-C was calculated using the Friedewald formula (total cholesterol–HDL-C–triglycerides/5) [
      • Friedewald W.T.
      • Levy R.I.
      • Fredrickson D.S.
      Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.
      ].

      2.2 Ultrasound protocol

      To harmonize the recordings and transfer of data, before being selected for the study (PARC [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
      ], PARC-ALAA [
      • Touboul P.J.
      • Hernandez-Hernandez R.
      • Kucukoglu S.
      Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
      ], PI2 and CARMELA [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      ]) all the centers were required to have an ultrasound B-mode device no more than 7 years old, fitted with a probe with an emission frequency of at least 5 MHz. The sonographer was required to have more than 1 year of experience in carotid sonography and agree to follow a half-day training session for the presentation of the study protocol and the acquisition and transfer of ultrasound images.

      2.2.1 Data acquisition, recording procedure, and center certification

      All centers were provided with dedicated software for image acquisition, measurements, and storage (M'Ath®-Std©, Intelligence in medical Technologies, France). Carotid scans stored on the hard disk of the computer were transferred to a core laboratory for image analysis (Intelligence in Medical Technologies, France). The certification of the center before inclusion was based on the carotid artery examinations of three subjects obtained following the procedures presented during the investigators' training session. All centers had three attempts to get certification. Less than 5% of the centers were able to obtain their certification after multiple training sessions and no center was excluded after failing three certifications.

      2.2.2 Carotid ultrasound examination

      The near and far walls of the middle parts of the right and left common carotid arteries were scanned longitudinally to assess the best angle of incidence. The lateral probe incidence was recommended as it offers the best view in the mid field. Moreover, in these situations, the image has a better definition in the near field for inter-adventitial diameter measurement. Once the best view of the CCA had been obtained, the sonographer had to confirm that the quality index was ≥0.5 on a measurement performed on a 10-mm length. A total of 150 values could be obtained automatically by edge detection of this 10-mm length; a quality index ≥0.5 meant that ≥50% of these measurements were available for IMT averaging. The quality index was displayed automatically after having drawn a line of 10 mm parallel to the far wall in the lumen of the CCA in a plaque-free region. If this index was <0.5, the physician had to acquire another image of the CCA in order to improve acquisition.

      2.3 Statistical analysis

      For the primary meta-analysis, we used an one-stage approach to assess the cross-sectional associations of HDL-C, triglyceride and LDL-C/HDL-C ratio levels with CCA-IMT, in which a single model was fitted in combined studies. Combined linear regression coefficients were calculated using a mixed linear regression model including a random effect for studies and fixed effects for lipid and other covariates [
      • Higgins J.P.
      • Whitehead A.
      • Turner R.M.
      • Omar R.Z.
      • Thompson S.G.
      Meta-analysis of continuous outcome data from individual patients.
      ]. Analyses on triglycerides were performed on loge transformed values to reduce the skewness of its distribution. To examine the linearity of the cross-sectional associations, we categorized the lipid levels into quintiles and calculated the regression coefficients using as a reference, the highest quintile for HDL-C and the lowest quintile for triglycerides and LDL-C/HDL-C ratio. Since there was no evidence for a non-linear association, regression coefficients were calculated per 1 standard deviation (SD) higher lipid levels. Regression coefficients were interpreted in terms of mean difference in CCA-IMT. All regression coefficients were adjusted for age and sex, with an additional pre-specified adjustment for geographical region in PARC-AALA [
      • Touboul P.J.
      • Hernandez-Hernandez R.
      • Kucukoglu S.
      Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
      ] and for city in CARMELA [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      ] according to specific study design. Further adjustment for pre-specified confounding factors (including body mass index, diabetes, smoking, systolic blood pressure, antihypertensive treatment, personal and family history of coronary heart disease, lipid-lowering treatment and LDL-C) were also performed. Heterogeneity across subgroups was quantified by formal interaction tests. We repeated the meta-analysis using a two-stage approach, in which individual linear regression models were fit for each study and summarized using random-effects meta-analysis [
      • van Houwelingen H.C.
      • Arends L.R.
      • Stijnen T.
      Advanced methods in meta-analysis: multivariate approach and meta-regression.
      ]. We quantified the between-study heterogeneity of the estimates by calculating the I² statistics with 95% confidence intervals (CIs) [
      • Higgins J.P.
      • Thompson S.G.
      Quantifying heterogeneity in a meta-analysis.
      ].
      Using the primary meta-analysis approach, a sensitivity analysis was performed after exclusion of PI2 (which restricted the population to patients free from a history of coronary heart disease and carotid plaques) and REMEMBER [
      • Prati P.
      • Casaroli M.
      • Bignamini A.
      • et al.
      Cognitive impairment and carotid atherosclerosis in a general Italian midlife and old population.
      ] (which restricted the population to patients >55 years). Finally, in order to explore whether the relationships of HDL-C and triglycerides with CCA-IMT were maintained even at low LDL-C levels, we repeated the main analysis in three LDL-C subgroups (≤100, 101–130, >130 mg/dL).
      Statistical testing was conducted at the two-tailed α-level of 0.05. Data were analyzed using the SAS software version 9.3 (SAS Institute, Cary, NC).

      3. Results

      Overall, 21,587 patients with complete information on HDL-C, triglycerides, and CCA-IMT measurements contributed to the present individual meta-analysis. Patient characteristics varied by study, as described in Table 1. The combined mean age at enrollment was 51 years and 49% were men. Prevalence of vascular risk factors also varied across studies; mean HDL-C ranged from 46 mg/dL to 61 mg/dL and median triglycerides from 103 mg/dL to 133 mg/dL. Combined mean CCA-IMT was 0.696 mm, ranging from 0.668 to 0.786 mm across studies.
      Table 1Characteristics from the six cross-sectional studies contributing to the current meta-analysis.
      CARMELA
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      (n = 10,415)
      PARC
      • McQueen M.J.
      • Hawken S.
      • Wang X.
      • et al.
      Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study.
      (n = 5264)
      PARC-AALA
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
      (n = 2328)
      PI2 (n = 554)REMEMBER
      • Tosetto A.
      • Prati P.
      • Baracchini C.
      • Manara R.
      • Rodeghiero F.
      Age-adjusted reference limits for carotid intima-media thickness as better indicator of vascular risk: population-based estimates from the VITA project.
      (n = 606)
      VITA
      • Touboul P.J.
      • Hernandez-Hernandez R.
      • Kucukoglu S.
      Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
      (n = 2420)
      Age, y44.7 ± 11.360.6 ± 11.649.8 ± 8.356.3 ± 7.566.8 ± 8.354.0 ± 4.5
      Male4852 (46.6)2999 (57.0)1123 (48.2)278 (50.2)286 (47.2)1073 (44.3)
      Body mass index, kg/m227.3 ± 4.825.8 ± 4.126.3 ± 4.526.9 ± 4.227.9 ± 4.225.7 ± 4.0
      Diabetes566 (5.4)657 (12.5)171 (7.4)86 (15.5)110 (18.2)108 (4.5)
      Current smokers3312 (31.8)795 (15.1)206 (8.9)126 (22.7)101 (16.7)567 (23.4)
      Antihypertensive treatment1198 (11.5)2022 (38.7)600 (25.8)266 (48.0)258 (42.6)398 (16.5)
      Systolic blood pressure, mmHg120 ± 18136 ± 16138 ± 22133 ± 16140 ± 19138 ± 18
      Diastolic blood pressure, mmHg77 ± 1180 ± 986 ± 1379 ± 1080 ± 1083 ± 9
      Lipid-lowering treatment233 (2.2)2077 (39.6)168 (7.2)226 (40.8)69 (11.4)1980 (81.8)
      Total cholesterol, mg/dL198 ± 42215 ± 42209 ± 44214 ± 39223 ± 40234 ± 38
      LDL-C, mg/dL121 ± 35134 ± 37133 ± 40129 ± 34137 ± 34161 ± 35
      HDL-C, mg/dL46 ± 1356 ± 1750 ± 1360 ± 1661 ± 1449 ± 13
      LDL-C/HDL-C ratio2.8 ± 1.02.6 ± 1.12.9 ± 1.32.3 ± 0.92.3 ± 0.73.5 ± 1.2
      Triglycerides, mg/dL132 (90–195)107 (77–150)133 (88–186)110 (81–149)108 (81–149)103 (76–147)
      Personal history of CHD432 (4.2)1450 (27.6)150 (6.4)1 (0.2)95 (15.8)27 (1.1)
      Familial history of CHD1378 (13.2)802 (15.4)201 (8.6)109 (19.7)99 (16.4)216 (8.9)
      CCA-IMT, mm0.668 ± 0.1050.751 ± 0.1230.692 ± 0.1140.786 ± 0.0900.761 ± 0.1030.668 ± 0.078
      Carotid plaques1077 (10.5)NA492 (22.8)8 (1.4)399 (65.8)12.6 (304)
      Values are mean ± SD, number (%) or median (IQR). CARMELA indicates Cardiovascular Risk Factor Multiple Evaluation in Latin America; PARC, Paroi Artérielle et Risque Cardiovasculaire; PARC-AALA, Paroi Arterielle et Risque Cardiovasculaire in Asia, Africa/Middle East and Latin America; PI2, Plaque Intima-media Inflammation; REMEMBER, Registry Evaluation Memory in Buttrio e Remanzacco; VITA, Vicenza Thrombophilia and Atherosclerosis; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; CHD, coronary heart disease; CCA-IMT, common carotid artery intima-media thickness; SD, standard deviation; IQR, interquartile range; NA, not available.

      3.1 HDL-C and CCA-IMT

      In the age- and sex-adjusted analysis, mean CCA-IMT decreased with increasing quintiles of HDL-C (Fig. 1A). After further adjustment for atherosclerosis risk factors, the mean change in CCA-IMT associated with HDL-C quintiles was attenuated (Fig. 1A). In the one-stage approach meta-analysis adjusted for age and sex, CCA-IMT was estimated to differ by −7.8 μm (95% CI −9.1 to −6.5 μm, P < 0.001) per 1 SD higher HDL-C level (Table 2). After further adjustment for other atherosclerosis risk factors, the relationship was attenuated, but remained significant with a mean CCA-IMT change of −3.7 μm (95% CI −5.1 to −2.3 μm; P < 0.001). Similar findings were found in our secondary analysis using a two-stage approach (fully adjusted mean CCA-IMT difference −3.5 μm, 95% CI −5.7 to −1.2 μm, P = 0.002). There was substantial heterogeneity between studies (I2 = 71%; 95% CI 55–81%). In a sensitivity analysis excluding the PI2 and REMEMBER studies, the fully adjusted mean CCA-IMT difference was −3.7 μm (95% CI −5.1 to −2.3 μm, P < 0.001) per 1 SD higher HDL-C level.
      Figure thumbnail gr1
      Fig. 1Adjusted mean difference in CCA-IMT across quintiles of HDL-C (A) and triglyceride (B) levels. Error bars indicate 95% CIs. The fully adjusted analyses were based on the sample of patients with non-missing covariates including age, sex, body mass index, diabetes, current smoker, antihypertensive treatment, systolic blood pressure, personal and familial history of coronary heart disease, lipid-lowering treatment and LDL-C level (n = 20,939). Regression coefficients were calculated from linear mixed model including study as random effect and by using the upper quintile for HDL-C and lower quintile for triglyceride as references.
      Table 2Adjusted mean difference of CCA-IMT per 1 SD higher HDL-C and loge triglyceride levels in individual and combined studies.
      Per 14.8 higher HDL-CPer 0.55 higher Loge TG
      β, μm95% CIPβ, μm95% CIP
      Age- and sex-adjusted analyses
      CARMELA
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      −9.5−11.4 to −7.5<0.0016.14.6–7.6<0.001
      PARC
      • McQueen M.J.
      • Hawken S.
      • Wang X.
      • et al.
      Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study.
      −5.5−8.2 to −2.7<0.0019.26.0–12.4<0.001
      PARC-ALLA
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
      −6.0−10.9 to −1.10.01513.59.1–18.0<0.001
      PI2−4.8−11.4 to 1.80.1482.8−5.1 to 10.80.490
      REMEMBER
      • Tosetto A.
      • Prati P.
      • Baracchini C.
      • Manara R.
      • Rodeghiero F.
      Age-adjusted reference limits for carotid intima-media thickness as better indicator of vascular risk: population-based estimates from the VITA project.
      −11.5−19.3 to −3.80.0042.7−5.4 to 10.80.513
      VITA
      • Touboul P.J.
      • Hernandez-Hernandez R.
      • Kucukoglu S.
      Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
      −8.5−12.0 to −4.9<0.0018.75.3–12.0<0.001
      One-stage meta-analysis−7.8−9.1 to −6.5<0.0017.36.1–8.5<0.001
      Two-stage meta-analysis−7.7−9.5 to −5.9<0.0018.05.6–10.3<0.001
      Heterogeneity: I2 (95% CI)38% (1%–61%)75% (62%–84%)
      Fully-adjusted analyses
      CARMELA
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      −6.1−8.0 to −4.1<0.0010.2−1.5 to 2.00.799
      PARC
      • McQueen M.J.
      • Hawken S.
      • Wang X.
      • et al.
      Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study.
      −1.4−4.2 to 1.40.3221.2−2.2 to 4.70.496
      PARC-ALLA
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
      0.3−4.5 to 5.10.891−0.1−5.0 to 4.80.704
      PI2−0.5−7.5 to 6.50.886−2.0−10.5 to 6.60.655
      REMEMBER
      • Tosetto A.
      • Prati P.
      • Baracchini C.
      • Manara R.
      • Rodeghiero F.
      Age-adjusted reference limits for carotid intima-media thickness as better indicator of vascular risk: population-based estimates from the VITA project.
      −8.8−16.9 to −0.70.033−6.0−15.5 to 3.70.224
      VITA
      • Touboul P.J.
      • Hernandez-Hernandez R.
      • Kucukoglu S.
      Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
      −4.4−8.1 to −0.70.0192.0−1.7 to 5.70.301
      One-stage meta-analysis−3.7−5.1 to −2.3<0.001−0.1−1.5 to 1.30.900
      Two-stage meta-analysis−3.5−5.7 to −1.20.0020.3−1.0 to 1.70.632
      Heterogeneity: I2 (95% CI)71% (55%–81%)0% (0%–22%)
      The fully adjusted analyses were based on the sample of patients with non-missing covariates including age, sex, body mass index, diabetes, current smokers, antihypertensive treatment, systolic blood pressure, personal and familial history of coronary heart disease, lipid-lowering treatment and LDL-C level (n = 20,939).
      CCA-IMT indicates common carotid artery intima-media thickness; SD, standard deviation; HDL-C, high-density lipoprotein cholesterol; TG, triglycerides; β, regression coefficient; CARMELA, Cardiovascular Risk Factor Multiple Evaluation In Latin America; PARC, Paroi Artérielle et Risque Cardiovasculaire; PARC-AALA, Paroi Arterielle et Risque Cardiovasculaire in Asia, Africa/Middle East and Latin America; PI2, Plaque Intima-media Inflammation; REMEMBER, Registry Evaluation Memory in Buttrio e Remanzacco; VITA, Vicenza Thrombophilia and Atherosclerosis.
      In a subgroup analysis, the inverse association between HDL-C and CCA-IMT was stronger in current smokers, in patients with low body mass index, and was only found in patients with no family history of coronary heart disease (Fig. 2A). The presence of carotid plaques did not modify the association of HDL-C and CCA-IMT (P for interaction = 0.737; Fig. 2A). The adjusted mean CCA-IMT difference per 1 SD higher HDL-C level was −4.6 μm (95% CI −6.2 to −2.9 μm) in carotid plaque-free patients and −3.7 μm (95% CI −8.3 to 0.9 μm) in patients with carotid plaques (Fig. 2A).
      Figure thumbnail gr2
      Fig. 2Associations of HDL-C (A) and triglyceride (B) levels with CCA-IMT according to other conventional risk factors. β indicates regression coefficient (μm); CHD, coronary artery disease; HDL-C, high-density lipoprotein cholesterol. *Excluding PI2 as personal history of CHD and carotid plaques were exclusion criteria. †Excluding PARC
      [
      • McQueen M.J.
      • Hawken S.
      • Wang X.
      • et al.
      Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study.
      ]
      as no information on carotid plaque status was available.

      3.2 Triglycerides and CCA-IMT

      As for the HDL-C analysis, there was no evidence of a non-linear relationship between triglycerides and CCA-IMT (Fig. 1B). In the one-stage as well as in the two-stage meta-analyses, triglycerides were positively associated with CCA-IMT after adjustment for age and sex, but not after further adjustment for other atherosclerosis risk factors (Fig. 1B; Table 2). There was substantial heterogeneity in the age- and sex-adjusted meta-analysis (I2 = 75%, 95% CI 62–84%), but not in fully adjusted meta-analysis (I2 = 0%). In a sensitivity analysis excluding the PI2 and REMEMBER studies, the fully adjusted mean CCA-IMT difference was 0.2 μm (95% CI −1.3 to 1.6, P = 0.825).
      The association between triglycerides and CCA-IMT was opposite in older and younger patients (Fig. 2B). In the other subgroups, the association of triglycerides and CCA-IMT was essentially null (Fig. 2B).

      3.3 LDL-C/HDL-C ratio and CCA-IMT

      There was strong evidence of linear relationship between LDL-C/HDL-C ratio and CCA-IMT, with fully adjusted mean CCA-IMT difference in one-stage meta-analysis of 5.2 μm (95% CI 3.4–7.1 μm, P < 0.001) per 1 SD higher LDL-C/HDL-C ratio (e-Fig. 1 and e-Table 1). There was a substantial heterogeneity between studies, both in age- and sex-adjusted meta-analysis (I2 = 87%; 95% CI 81–91%), and in fully-adjusted analysis (I2 = 67%; 95% CI 49–79%). In a sensitivity analysis excluding the PI2 and REMEMBER studies, the fully adjusted mean CCA-IMT difference was 5.2 μm (95% CI 3.3 to 7.0, P < 0.001). In subgroup analysis, several heterogeneities were found, with a stronger association in patients younger than 60 years, in current smokers and in patients with no family history of coronary heart disease (e-Fig. 2).

      3.4 LDL-C subgroup analysis

      As shown in Fig. 3A, the negative association between HDL-C level and CCA-IMT was found regardless of LDL-C levels. Estimated adjusted mean CCA-IMT differences per 1 SD higher HDL-C level were −2.7 μm (95% CI −5.2 to −0.2 μm) in patients with LDL-C ≤ 100 mg/dL, −4.0 μm (95% CI −6.5 to −1.6 μm) in patients with LDL-C of 101–130 mg/dL, and −3.9 μm (95% CI −6.1 to −1.8) in patients with LDL-C > 130 mg/dL (P for heterogeneity = 0.700).
      Figure thumbnail gr3
      Fig. 3Association of CCA-IMT with HDL-C (A) and triglyceride (B) levels according to LDL-C levels. Error bars indicate 95% CIs. The fully adjusted analyses were based on the sample of patients with non-missing covariates including age, sex, body mass index, diabetes, current smokers, antihypertensive treatment, systolic blood pressure, personal and familial history of coronary heart disease and lipid-lowering treatment. Regression coefficients were calculated within quintiles (quintiles calculated in overall population) by using linear mixed model including study as random effect.
      Similarly to the main analysis, triglyceride levels appeared to be related to CCA-IMT in each LDL-C subgroup only in the age- and sex-adjusted analysis (Fig. 3B). The fully adjusted mean CCA-IMT differences were 0.7 μm (95% CI −1.8 to 3.1 μm) in patients with LDL-C ≤ 100 mg/dL, −0.8 μm (95% CI −3.3 to 1.6 μm) in patients with LDL-C of 101–130 mg/dL, and 1.1 μm (95% CI −1.2 to 3.4 μm) in patients with LDL-C > 130 mg/dL.

      4. Discussion

      In the present cross-sectional meta-analysis, CCA-IMT was associated with low HDL-C levels, independently of LDL-C levels, whereas CCA-IMT was not associated with triglycerides.
      The protective effect of higher age- and sex-adjusted HDL-C levels and CCA-IMT that persisted after adjustment for other atherosclerosis risk factors, as well as the relationship between higher age- and sex-adjusted triglyceride levels and increased CCA-IMT that disappeared after further adjustment for other atherosclerosis risk factors, were very similar to what was observed in the Emerging Risk Factor Collaboration meta-analysis between HDL-C and triglyceride levels and clinical events [
      • Di Angelantonio E.
      • Sarwar N.
      • Perry P.
      Emerging Risk Factors Collaboration
      Major lipids, apolipoproteins, and risk of vascular disease.
      ]. We further found that the protective effect on CCA-IMT of higher HDL-C levels was maintained at low LDL-C levels, reinforcing the need to verify whether raising HDL-C levels decreases both CCA-IMT and future clinical events.
      CCA-IMT is a well-established marker of future development of atherosclerotic plaque at the bifurcation or the internal carotid artery origin [
      • Von Sarnowski B.
      • Lüdemann J.
      • Völzke H.
      • Dörr M.
      • Kessler C.
      • Schminke U.
      Common carotid intima-media thickness and framingham risk score predict incident carotid atherosclerotic plaque formation: longitudinal results from the study of health in Pomerania.
      ,
      • Zureik M.
      • Ducimetiere P.
      • Touboul P.J.
      • et al.
      Common carotid intima-media thickness predicts occurrence of carotid atherosclerotic plaques: longitudinal results from the Aging Vascular Study (EVA) study.
      ]. CCA-IMT is a diffuse thickening of intima and media layers of the arterial wall involving rheologic and remodeling processes, while carotid plaque is a focal disease of the intima and media layers involving local lipid infiltration, inflammatory reaction then a cascade of numerous local atherogenic events. The differential relationship between CCA-IMT and HDL-C and triglyceride levels that we observed in this meta-analysis may be due to pleiotropic effects (i.e., anti-inflammatory effect, effect on smooth muscle cells, improvement of endothelial dysfunction) of HDL-C on CCA-IMT, which may not be true for triglycerides. On the other hand, triglyceride particles, which are well established as atherogenic factors [
      • Sarwar N.
      • Danesh J.
      • Eiriksdottir G.
      • et al.
      Triglycerides and the risk of coronary heart disease: 10,158 incident cases among 262,525 participants in 29 Western prospective studies.
      ], may not be associated with IMT thickening. The lack of modification of the relationship between IMT and HDL or TG levels by the presence of carotid plaque (Fig. 2), suggests that the significant relationship we found may not be true for carotid plaque, as already shown [
      • Gardener H.
      • Della Morte D.
      • Elkind M.S.
      • Sacco R.L.
      • Rundek T.
      Lipids and carotid plaque in the Northern Manhattan Study (NOMAS).
      ].
      Our study has various strengths and limitations. Strengths include: meta-analysis on individual data based on more than 21,000 individuals; harmonization of CCA-IMT measurements by means of a software common to all studies (M'ATH™-Intelligence in Medical Technologies-France) that allowed electronic measurement with the same algorithm averaging about 150 measures per 10 mm of CCA assessed; the provided ultrasound image acquisition of CCA-IMT had a quality index >0.5 (i.e., more than 50% of the region could be successfully detected by the software, which was checked by the investigator before freezing the image) in all studies; a central adjudication in four of the studies (PARC [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
      ], PARC-AALA [
      • Touboul P.J.
      • Hernandez-Hernandez R.
      • Kucukoglu S.
      Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
      ], CARMELA [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      ], PI2; n = 18,561) in the same core lab (Intelligence in Medical Technologies™); certification of all investigators by the core lab prior to the start of the study based on carotid ultrasound acquisition of three different patients; electronic transmission of data acquisitions to the core lab and a verified quality index above 50%; and the fact that measurement of CCA-IMT followed Mannheim consensus recommendations [
      • Touboul P.J.
      • Hennerici M.G.
      • Meairs S.
      • et al.
      Mannheim carotid intima-media thickness consensus (2004-2006). An update on behalf of the Advisory Board of the 3rd and 4th Watching the Risk Symposium, 13th and 15th European Stroke Conferences, Mannheim, Germany, 2004, and Brussels, Belgium, 2006.
      ] in all studies. Limitations include that only one of the six studies (CARMELA) [
      • Touboul P.J.
      • Vicaut E.
      • Labreuche J.
      • et al.
      Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
      ] was population based, while the others were based on investigators randomly selected among primary care cardiology practice or from general community hospitals; and that the studies were cross sectional with significant heterogeneity in patients' characteristic across them. Since LDL-C levels were determined indirectly according to Friedewald equation, we couldn't exclude that LDL-C levels were underestimated [
      • Martin S.S.
      • Blaha M.J.
      • Elshazly M.B.
      • et al.
      Friedewald-estimated versus directly measured low-density lipoprotein cholesterol and treatment implications.
      ] and that a potential for confounding cannot be ruled out completely. In addition, our findings could not be generalized to the secondary cardiovascular disease prevention, since most of the patients were enrolled without history of cardiovascular event.

      5. Conclusions

      Our data strengthen the case for using CCA-IMT evolution as an endpoint for the evaluation of the yield of HDL-C-raising agents. However these data need to be confirmed by longitudinal studies including prospective IMT and plaque evaluation and clinical outcome.

      Funding sources

      Funding for this study was provided in part by SOS-ATTAQUE CEREBRALE .

      Disclosures

      We have no conflicts of interest.

      Acknowledgments

      Jenny Lloyd, PhD, provided editorial assistance in preparing the final version of the manuscript, including editing, checking content and language, and formatting, and received compensation from SOS-ATTAQUE CEREBRALE.

      References

        • Amarenco P.
        • Labreuche J.
        Lipid management in the prevention of stroke: review and updated meta-analysis of statins for stroke prevention.
        Lancet Neurol. 2009; 8: 453-463
        • Sarwar N.
        • Danesh J.
        • Eiriksdottir G.
        • et al.
        Triglycerides and the risk of coronary heart disease: 10,158 incident cases among 262,525 participants in 29 Western prospective studies.
        Circulation. 2007; 115: 450-458
        • Freiberg J.J.
        • Tybjaerg-Hansen A.
        • Jensen J.S.
        • Nordestgaard B.G.
        Nonfasting triglycerides and risk of ischemic stroke in the general population.
        J Am Med Assoc. 2008; 300: 2142-2152
        • Gordon D.J.
        • Probstfield J.L.
        • Garrison R.J.
        • et al.
        High-density lipoprotein cholesterol and cardiovascular disease. Four prospective American studies.
        Circulation. 1989; 79: 8-15
        • Barter P.
        • Gotto A.M.
        • LaRosa J.C.
        • et al.
        HDL cholesterol, very low levels of LDL cholesterol, and cardiovascular events.
        N Engl J Med. 2007; 357: 1301-1310
        • Di Angelantonio E.
        • Sarwar N.
        • Perry P.
        • Emerging Risk Factors Collaboration
        Major lipids, apolipoproteins, and risk of vascular disease.
        J Am Med Assoc. 2009; 302: 1993-2000
        • Amarenco P.
        • Labreuche J.
        • Touboul P.J.
        High-density lipoprotein-cholesterol and risk of stroke and carotid atherosclerosis: a systematic review.
        Atherosclerosis. 2008; 196: 489-496
        • Labreuche J.
        • Touboul P.J.
        • Amarenco P.
        Plasma triglyceride levels and risk of stroke and carotid atherosclerosis: a systematic review of the epidemiological studies.
        Atherosclerosis. 2009; 203: 331-345
        • Mackinnon A.D.
        • Jerrard-Dunne P.
        • Sitzer M.
        • Buehler A.
        • von Kegler S.
        • Markus H.S.
        Rates and determinants of site-specific progression of carotid artery intima-media thickness: the carotid atherosclerosis progression study.
        Stroke. 2004; 35: 2150-2154
        • Touboul P.J.
        • Vicaut E.
        • Labreuche J.
        • et al.
        Common carotid artery intima-media thickness: the Cardiovascular Risk Factor Multiple Evaluation in Latin America (CARMELA) study results.
        Cerebrovasc Dis. 2011; 31: 43-50
        • Lorenz M.W.
        • Markus H.S.
        • Bots M.L.
        • Rosvall M.
        • Sitzer M.
        Prediction of clinical cardiovascular events with carotid intima-media thickness: a systematic review and meta-analysis.
        Circulation. 2007; 115: 459-467
        • Polak J.F.
        • Pencina M.J.
        • Pencina K.M.
        • O'Donnell C.J.
        • Wolf P.A.
        • Sr D'Agostino RB.
        Carotid-wall intima-media thickness and cardiovascular events.
        N Engl J Med. 2011; 365: 213-221
        • Von Sarnowski B.
        • Lüdemann J.
        • Völzke H.
        • Dörr M.
        • Kessler C.
        • Schminke U.
        Common carotid intima-media thickness and framingham risk score predict incident carotid atherosclerotic plaque formation: longitudinal results from the study of health in Pomerania.
        Stroke. 2010; 41: 2375-2377
        • Zureik M.
        • Ducimetiere P.
        • Touboul P.J.
        • et al.
        Common carotid intima-media thickness predicts occurrence of carotid atherosclerotic plaques: longitudinal results from the Aging Vascular Study (EVA) study.
        Arterioscler Thromb Vasc Biol. 2000; 20: 1622-1629
        • Briel M.
        • Ferreira-Gonzalez I.
        • You J.J.
        • et al.
        Association between change in high density lipoprotein cholesterol and cardiovascular disease morbidity and mortality: systematic review and meta-regression analysis.
        BMJ. 2009; 338: b92
        • Singh I.M.
        • Shishehbor M.H.
        • Ansell B.J.
        High-density lipoprotein as a therapeutic target: a systematic review.
        J Am Med Assoc. 2007; 298: 786-798
        • Labreuche J.
        • Deplanque D.
        • Touboul P.J.
        • Bruckert E.
        • Amarenco P.
        Association between change in plasma triglyceride levels and risk of stroke and carotid atherosclerosis: systematic review and meta-regression analysis.
        Atherosclerosis. 2010; 212: 9-15
        • Touboul P.J.
        • Hennerici M.G.
        • Meairs S.
        • et al.
        Mannheim carotid intima-media thickness consensus (2004-2006). An update on behalf of the Advisory Board of the 3rd and 4th Watching the Risk Symposium, 13th and 15th European Stroke Conferences, Mannheim, Germany, 2004, and Brussels, Belgium, 2006.
        Cerebrovasc Dis. 2007; 23: 75-80
        • McQueen M.J.
        • Hawken S.
        • Wang X.
        • et al.
        Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study.
        Lancet. 2008; 372: 224-233
        • Touboul P.J.
        • Vicaut E.
        • Labreuche J.
        • et al.
        Design, baseline characteristics and carotid intima-media thickness reproducibility in the PARC study.
        Cerebrovasc Dis. 2005; 19: 57-63
        • Touboul P.J.
        • Hernandez-Hernandez R.
        • Kucukoglu S.
        Carotid artery intima media thickness, plaque and Framingham cardiovascular score in Asia, Africa/Middle East and Latin America: the PARC-AALA study.
        Int J Cardiovasc Imaging. 2007; 23: 557-567
        • Tosetto A.
        • Prati P.
        • Baracchini C.
        • Manara R.
        • Rodeghiero F.
        Age-adjusted reference limits for carotid intima-media thickness as better indicator of vascular risk: population-based estimates from the VITA project.
        J Thromb Haemost. 2005; 3: 1224-1230
        • Prati P.
        • Casaroli M.
        • Bignamini A.
        • et al.
        Cognitive impairment and carotid atherosclerosis in a general Italian midlife and old population.
        Neuroepidemiology. 2006; 27: 33-38
        • Friedewald W.T.
        • Levy R.I.
        • Fredrickson D.S.
        Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.
        Clin Chem. 1972; 18: 499-502
        • Higgins J.P.
        • Whitehead A.
        • Turner R.M.
        • Omar R.Z.
        • Thompson S.G.
        Meta-analysis of continuous outcome data from individual patients.
        Stat Med. 2001; 20: 2219-2241
        • van Houwelingen H.C.
        • Arends L.R.
        • Stijnen T.
        Advanced methods in meta-analysis: multivariate approach and meta-regression.
        Stat Med. 2002; 21: 589-624
        • Higgins J.P.
        • Thompson S.G.
        Quantifying heterogeneity in a meta-analysis.
        Stat Med. 2002; 21: 1539-1558
        • Gardener H.
        • Della Morte D.
        • Elkind M.S.
        • Sacco R.L.
        • Rundek T.
        Lipids and carotid plaque in the Northern Manhattan Study (NOMAS).
        BMC Cardiovasc Disord. 2009; 9: 55
        • Martin S.S.
        • Blaha M.J.
        • Elshazly M.B.
        • et al.
        Friedewald-estimated versus directly measured low-density lipoprotein cholesterol and treatment implications.
        J Am Coll Cardiol. 2013; 20: 732-739

      Linked Article