Author + information
- Received April 24, 2013
- Revision received July 10, 2013
- Accepted July 30, 2013
- Published online November 19, 2013.
- Mohamed B. Elshazly, MD∗,†∗ (, )
- Seth S. Martin, MD†,
- Michael J. Blaha, MD, MPH†,
- Parag H. Joshi, MD†,
- Peter P. Toth, MD, PhD‡,§,
- John W. McEvoy, MB BCh†,
- Mohammed A. Al-Hijji, MD†,
- Krishnaji R. Kulkarni, PhD‖,
- Peter O. Kwiterovich, MD†,
- Roger S. Blumenthal, MD† and
- Steven R. Jones, MD†
- ∗Department of Cardiovascular Medicine, Cleveland Clinic, Cleveland, Ohio
- †Johns Hopkins Ciccarone Center for the Prevention of Heart Disease, Baltimore, Maryland
- ‡Department of Preventive Cardiology, CGH Medical Center, Sterling, Illinois
- §University of Illinois College of Medicine, Peoria, Illinois
- ‖Atherotech Diagnostics Lab, Birmingham, Alabama
- ↵∗Reprint requests and correspondence:
Dr. Mohamed B. Elshazly, Department of Cardiovascular Medicine, Cleveland Clinic, 9500 Euclid Avenue, J3-4, Cleveland, Ohio 44195.
Objectives This study sought to examine patient-level discordance between population percentiles of non–high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C).
Background Non–HDL-C is an alternative to LDL-C for risk stratification and lipid-lowering therapy. The justification for the present guideline-based non–HDL-C cutpoints of 30 mg/dl higher than the LDL-C cutpoints remains largely untested.
Methods We assigned population percentiles to non–HDL-C and Friedewald-estimated LDL-C values of 1,310,440 U.S. adults with triglyceride levels <400 mg/dl who underwent lipid testing by vertical spin density gradient ultracentrifugation (Atherotech, Birmingham, Alabama) from 2009 to 2011.
Results LDL-C cutpoints of 70, 100, 130, 160, and 190 mg/dl were in the same population percentiles as non–HDL-C values of 93, 125, 157, 190, and 223 mg/dl, respectively. Non–HDL-C values reclassified a significant proportion of patients within or to a higher treatment category compared with Friedewald LDL-C values, especially at LDL-C levels in the treatment range of high-risk patients and at triglyceride levels ≥150 mg/dl. Of patients with LDL-C levels <70 mg/dl, 15% had a non–HDL-C level ≥100 mg/dl (guideline-based cutpoint) and 25% had a non–HDL-C level ≥93 mg/dl (percentile-based cutpoint); if triglyceride levels were 150 to 199 mg/dl concurrently, these values were 22% and 50%, respectively.
Conclusions There is significant patient-level discordance between non–HDL-C and LDL-C percentiles at lower LDL-C and higher triglyceride levels, which has implications for the treatment of high-risk patients. Current non–HDL-C cutpoints for high-risk patients may need to be lowered to match percentiles of LDL-C cutpoints. Relatively small absolute reductions in non–HDL-C cutpoints result in substantial reclassification of patients to higher treatment categories with potential implications for risk assessment and treatment. (The Very Large Database of Lipids [VLDL-2]; NCT01698489)
Cardiovascular disease (CVD) is the leading cause of death in the developed world, accounting for 33% of deaths in the United States and 47% of deaths in Europe (1,2). For the past 25 years, low-density lipoprotein cholesterol (LDL-C) has been the primary lipid parameter for risk stratification and goal-directed therapy. However, an epidemic of obesity and metabolic syndrome has evolved over the past few decades, mostly due to changes in diet and lifestyle. Approximately 1 of 3 U.S. adults currently has metabolic syndrome (1). As a result, we have witnessed an increasing prevalence of elevated triglyceride-rich remnant lipoproteins, characteristic of insulin resistance. These lipoproteins include very-low-density lipoproteins (VLDL) and their remnants, intermediate-density lipoproteins, and chylomicron remnant particles, whose contribution to atherogenic risk is accounted for by non–HDL-C not LDL-C.
Current guidelines recommend using non–HDL-C as a secondary treatment target in patients with triglyceride levels ≥200 mg/dl (3–5), setting non–HDL-C goals 30 mg/dl higher than respective LDL-C goals. However, some reports have suggested using non–HDL-C goals at the same population percentiles as the respective LDL-C goals (6,7).
In this report, we examine patient-level discordance between non–HDL-C and LDL-C percentiles at different LDL-C and triglyceride strata and implications for risk assessment and treatment.
We examined consecutive lipid profiles from 1,310,440 U.S. adults ≥18 years of age with triglyceride levels <400 mg/dl who underwent direct ultracentrifugation of cholesterol by the vertical auto profile (VAP; Atherotech Diagnostics Lab, Birmingham, Alabama) from 2009 to 2011 (8,9). “Consecutive” indicates that we only included the first available lipid profile for each patient. Consistent with routine clinical practice, LDL-C levels were estimated using the Friedewald formula, thus excluding patients with triglyceride levels ≥400 mg/dl (9,10).
VAP lipid measurement
The VAP is an inverted rate zonal, single vertical spin, density-gradient ultracentrifugation technique that directly measures cholesterol concentrations of the 5 lipoprotein classes (LDL-C, VLDL cholesterol, intermediate-density lipoprotein cholesterol, HDL-C, and lipoprotein[a]) and their subclasses. Triglyceride levels were directly measured using the Abbott Architect C8000 system (Abbott Park, Illinois) (8,9). The accuracy of VAP lipid parameters was cross-validated with reference standards as previously described (9).
Raw individual patient data were extracted at Atherotech, cleaned of duplicate samples, and then de-identified and transferred to the senior investigator. The master database, the Very Large Database of Lipids (VLDL), is maintained at The Johns Hopkins Hospital (Baltimore, Maryland) and registered at clinicaltrials.gov (NCT01698489). The Johns Hopkins Institutional Review Board declared the study exempt.
Friedewald-estimated LDL-C levels were calculated as [total cholesterol – HDL-C – triglycerides/5]. Non–HDL-C levels were calculated as [total cholesterol – HDL-C]. We assigned population percentiles to LDL-C and non–HDL-C levels and then determined the percentiles corresponding to LDL-C cutpoints in current guidelines (70, 100, 130, 160, and 190 mg/dl) (3–5).
Reclassification was defined as present when the non–HDL-C level reclassified a patient within or to a higher (upward) or lower (downward) treatment category compared with the Friedewald LDL-C level. The analysis was performed using guideline-based non–HDL-C cutpoints, defined as 30 mg/dl higher than LDL-C cutpoints, and percentile-based cutpoints, defined as those at equivalent percentiles to LDL-C cutpoints. We focused on upward reclassification because current guidelines recommend using non–HDL-C level only as a secondary treatment target after the LDL-C target is reached; thus, downward reclassification becomes irrelevant.
Statistical analyses and logarithmically scaled pseudocolor density plots were generated in R Version 2.15.1 (Vienna, Austria) and Stata Version 11.0 (College Station, Texas).
Patients were 59 ± 15 years of age (mean ± SD), 52% were women, and lipid parameter distributions were nearly superimposable with recent lipid data from the 2007 to 2008 National Health and Nutrition Examination Survey (11), as previously described by our group (9) (Online Fig. 1).
LDL-C cutpoints of 70, 100, 130, 160, and 190 mg/dl were at the same population percentiles as non–HDL-C values of 93, 125, 157, 190, and 223 mg/dl, respectively (Table 1).
Non–HDL-C and LDL-C percentile discordance
We visually assessed discordance between LDL-C and non–HDL-C percentiles and found greater discordance at lower LDL-C and higher triglyceride levels (Fig. 1). Similarly, the absolute difference between non–HDL-C and LDL-C percentiles was more pronounced with greater in-group variation at lower LDL-C and higher triglyceride levels (Online Fig. 2).
Reclassification of treatment category by non–HDL-C level
When using conventional non–HDL-C cutpoints, non–HDL-C levels reclassified 10.5% (n = 137,744) of the study group upward and 22.3% (n = 291,499) downward. Using percentile-based cutpoints, 14.2% (n = 186,106) were reclassified upward and 13.7% (n = 178,860) downward (Fig. 2).
Upward reclassification occurred more frequently at lower LDL-C and higher triglyceride levels (for additional discussion, see Reclassification Analysis in the Online Appendix). Of patients with an LDL-C level <70 mg/dl, 15% had a non–HDL-C level ≥100 mg/dl (the guideline-based cutpoint) and 25% had a non–HDL-C level ≥93 mg/dl (the percentile-based cutpoint); these values were 22% and 50%, respectively, if triglyceride levels were concurrently 150 to 199 mg/dl (Fig. 3A). Similarly, of patients with LDL-C levels between 70 and 99 mg/dl, 12% had a non–HDL-C level ≥130 mg/dl and 17% had a non–HDL-C level ≥125 mg/dl; these values were 17% and 35%, respectively, if triglyceride levels were concurrently 150 to 199 mg/dl (Fig. 3B).
Our study highlights the magnitude of patient-level discordance between LDL-C and non–HDL-C percentiles. They are most discordant when accuracy is most crucial, at low LDL-C and high triglyceride levels. Therefore, conventional non–HDL-C cutpoints for high-risk patients may need to be lowered to match percentiles of LDL-C cutpoints.
Non–HDL-C: a better marker for CVD risk assessment and treatment
The National Cholesterol Education Program Adult Treatment Panel III guidelines state that “In most persons with triglycerides <200 mg/dl, adding VLDL-C to LDL-C would be expected to provide little additional power to predict CVD” (3); this is a disputable statement given the increasing prevalence of obesity, diabetes mellitus, and metabolic syndrome. Non–HDL-C represents the aggregate cholesterol content of apolipoprotein B–containing atherogenic lipoproteins, including LDL, VLDL, intermediate-density lipoprotein, remnants, and lipoprotein(a); in principle, this is a broader, more inclusive measure of atherogenic risk. Recent evidence suggests that non–HDL-C is superior for risk prediction and might be a more effective target for lipid-lowering therapy, particularly in high-risk patients (12–15). A meta-analysis of 233,455 patients showed that non–HDL-C is a more potent marker of CVD risk than LDL-C (16). Calculating the number of clinical events prevented by a high-risk treatment regimen in those in the >70th percentile of the U.S. adult population, Sniderman et al. (16) suggested that a non–HDL-C based strategy may prevent 300,000 more events than an LDL-C strategy over a 10-year period.
In addition, measurement of the non–HDL-C level has no additional cost or inconvenience because it is easily calculated from the standard lipid profile without the need for prior fasting. Moreover, the adoption of non–HDL-C across all levels of triglycerides would substantially simplify implementation of clinical guidelines.
Potential implications for guideline development
Guideline-based non–HDL-C cutpoints are based on the assumption that a normal VLDL cholesterol level exists when triglyceride levels are <150 mg/dl, which is <30 mg/dl as estimated by the Friedewald formula (3). More recent evidence suggests that a biologically optimal fasting triglyceride level is <100 mg/dl (17); thus, a normal VLDL cholesterol level is likely closer to 20 mg/dl, also suggesting that non–HDL-C cutpoints should be 20 mg/dl higher than LDL-C cutpoints.
Studying patients with acute coronary syndromes, Ballantyne et al. suggested that the current non–HDL-C goal should be lowered by 8 to 10 mg/dl to match LDL-C and apolipoprotein B treatment goals in the very-high-risk category (18). Other reports have recommended lowering non–HDL-C cutpoints to match percentiles of LDL-C cutpoints (6,7). In our study, the non–HDL-C values with percentile equivalence to LDL-C cutpoints of 100 and 70 mg/dl were 125 and 93 mg/dl, respectively. Therefore, non–HDL-C cutpoints may need to be lowered by 5 mg/dl and 7 mg/dl for the high-risk and very-high-risk categories, respectively. This leads to substantial upward reclassification of patients, particularly at concurrent high triglyceride levels. For example, of patients with LDL-C levels <70 mg/dl and concurrent triglyceride levels of 150 to 199 mg/dl, more than twice as many patients were reclassified upward when the non–HDL-C cutpoint was lowered from 100 to 93 mg/dl (Fig. 3A).
Our study also showed that the triglyceride threshold in the current guidelines of ≥200 mg/dl for using non–HDL-C as a secondary treatment target may need to be lowered given that considerable upward reclassification occurs also at triglyceride levels of 150 to 199 mg/dl (Fig. 2).
We have limited clinical and demographic data regarding the full risk factor profile of our population. Therefore, reclassification analyses are inferred on the basis of the lipid profile only and we cannot determine its impact on clinical outcomes.
The nearly superimposable age, sex, and lipid distributions between the samples from the VLDL study and the National Health and Nutrition Examination Survey suggest that our study comprises a reasonable population of patients engaged in prevention and treatment of atherosclerosis, not a special population that underwent VAP testing. We do not know the percentage of patients who were taking a statin. Some samples in our study may have been acquired in a nonfasting state, but this is not uncommon in routine practice.
Despite focusing on upward reclassification in accordance with current guidelines, there was considerable downward reclassification in patients with triglyceride levels <150 mg/dl (Fig. 2). The significance of downward reclassification remains unclear in the literature and current guidelines, and whether these patients should be treated to LDL-C versus non–HDL-C goal needs further scrutiny.
Our study of 1.3 million patients builds on prior evidence that there is significant patient-level discordance between percentiles of LDL-C and non–HDL-C, particularly when accuracy is most crucial, at lower LDL-C and higher triglyceride levels. Therefore, lowering conventional non–HDL-C cutpoints for high-risk patients to match percentiles of LDL-C cutpoints as well as wider adoption of non–HDL-C in clinical practice may potentially improve secondary prevention outcomes and residual risk assessment and treatment.
For reclassification analysis information and supplemental figures, please see the online version of this article.
Atherotech provided the investigators with de-identified data generated from commercial lipid analyses and did not provide payments for the research or manuscript writing and did not participate in data analysis or influence the conclusions. This study was initiated by the investigators and did not receive any specific funding. Drs. Martin, McEvoy, and Joshi are supported by the Pollin Fellowship in Preventive Cardiology. Dr. Martin is supported by the Marie-Josée and Henry R. Kravis endowed fellowship. Dr. Toth is a consultant for Atherotech Diagnostics Lab, Amgen, Genzyme, Kowa, Boehringer Ingelheim, Liposcience, and Merck; and is a member of the speaker's bureau for AstraZeneca, Amarin, GlaxoSmithKline, Kowa, Merck & Co., and Genzyme. Dr. Kulkarni is a research director at Atherotech Diagnostics Lab; owns uncashable stocks for Atherotech Diagnostics Labs; and receives royalties from the University of Alabama at Birmingham. Dr. Kwiterovich is a consultant for Merck & Co.; and has received research grants from Amarin, GlaxoSmithKline, and Pfizer. Dr. Jones is a member of the scientific advisory board and has received research grant support from Atherotech Diagnostics; and has unexercised stock options for and is on the scientific advisory board of LabRoots/BioConference Live. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cardiovascular disease
- high-density lipoprotein cholesterol
- low-density lipoprotein cholesterol
- vertical auto profile
- very-low-density lipoprotein cholesterol
- Received April 24, 2013.
- Revision received July 10, 2013.
- Accepted July 30, 2013.
- American College of Cardiology Foundation
- Roger V.L.,
- Go A.S.,
- Lloyd-Jones D.M.,
- et al.
- Nichols M, Townsend N, Luengo-Fernandez R, et al. 2012 European Cardiovascular Disease Statistics. European Heart Network, Brussels, European Society of Cardiology, Sophia Antipolis.
- National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III)
- Developed with the special contribution of: European Association for Cardiovascular Prevention & Rehabilitation, Authors/Task Force Members,
- Reiner Z.,
- et al.
- Contois J.H.,
- McConnell J.P.,
- Sethi A.A.,
- et al.
- Martin S.S.,
- Blaha M.J.,
- Elshazly M.B.,
- et al.
- Friedewald W.T.,
- Levy R.I.,
- Fredrickson D.S.
- Bittner V.,
- Hardison R.,
- Kelsey S.F.,
- et al.
- Varbo A.,
- Benn M.,
- Tybjærg-Hansen A.,
- Jørgensen A.B.,
- Frikke-Schmidt R.,
- Nordestgaard B.G.
- Sniderman A.D.,
- Williams K.,
- Contois J.H.,
- et al.
- Miller M.,
- Stone N.J.,
- Ballantyne C.,
- et al.
- Ballantyne C.M.,
- Pitt B.,
- Loscalzo J.,
- Cain V.A.,
- Raichlen J.S.